{"step": 1, "loss": 10.946487426757812, "grad_norm": 20.437564849853516, "learning_rate": 0.0, "elbo": 10.94649, "mask_frac": 0.5002, "mem_alloc_gib": 48.7, "mem_res_gib": 49.0, "epoch": 5.864414731409805e-05}
{"step": 1, "loss": 10.946487426757812, "grad_norm": 20.437564849853516, "learning_rate": 0.0, "elbo": 10.94649, "mask_frac": 0.5002, "mem_alloc_gib": 48.7, "mem_res_gib": 49.0, "epoch": 5.864414731409805e-05}
{"step": 1, "loss": 10.946487426757812, "grad_norm": 20.437564849853516, "learning_rate": 0.0, "elbo": 10.94649, "mask_frac": 0.5002, "mem_alloc_gib": 48.7, "mem_res_gib": 49.0, "epoch": 5.864414731409805e-05}
{"step": 5, "loss": 10.923234939575195, "grad_norm": 20.506141662597656, "learning_rate": 4.0000000000000003e-07, "elbo": 10.92324, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.00029322073657049027}
{"step": 10, "loss": 10.783003234863282, "grad_norm": 17.39012336730957, "learning_rate": 9e-07, "elbo": 10.783, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0005864414731409805}
{"step": 15, "loss": 10.523235321044922, "grad_norm": 12.049789428710938, "learning_rate": 1.4000000000000001e-06, "elbo": 10.52324, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0008796622097114708}
{"step": 20, "loss": 10.256251525878906, "grad_norm": 8.420187950134277, "learning_rate": 1.9e-06, "elbo": 10.25625, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.001172882946281961}
{"step": 25, "loss": 10.034284973144532, "grad_norm": 5.925222873687744, "learning_rate": 2.4000000000000003e-06, "elbo": 10.03428, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0014661036828524513}
{"step": 30, "loss": 9.851235961914062, "grad_norm": 4.106457233428955, "learning_rate": 2.9e-06, "elbo": 9.85124, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0017593244194229415}
{"step": 35, "loss": 9.732817077636719, "grad_norm": 2.8747718334198, "learning_rate": 3.4000000000000005e-06, "elbo": 9.73282, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.002052545155993432}
{"step": 40, "loss": 9.645738220214843, "grad_norm": 2.3522257804870605, "learning_rate": 3.9e-06, "elbo": 9.64574, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.002345765892563922}
{"step": 45, "loss": 9.607598114013673, "grad_norm": 2.142099380493164, "learning_rate": 4.4e-06, "elbo": 9.6076, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0026389866291344124}
{"step": 50, "loss": 9.579229736328125, "grad_norm": 2.1324408054351807, "learning_rate": 4.9000000000000005e-06, "elbo": 9.57923, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0029322073657049026}
{"step": 55, "loss": 9.558194732666015, "grad_norm": 2.029500722885132, "learning_rate": 5.4e-06, "elbo": 9.55819, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.003225428102275393}
{"step": 60, "loss": 9.531572723388672, "grad_norm": 1.9798697233200073, "learning_rate": 5.9e-06, "elbo": 9.53157, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.003518648838845883}
{"step": 65, "loss": 9.503349304199219, "grad_norm": 1.9889276027679443, "learning_rate": 6.4000000000000006e-06, "elbo": 9.50335, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0038118695754163733}
{"step": 70, "loss": 9.477739715576172, "grad_norm": 1.952657699584961, "learning_rate": 6.900000000000001e-06, "elbo": 9.47774, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.004105090311986864}
{"step": 75, "loss": 9.432261657714843, "grad_norm": 1.9988417625427246, "learning_rate": 7.4e-06, "elbo": 9.43226, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.004398311048557354}
{"step": 80, "loss": 9.404764556884766, "grad_norm": 1.9713267087936401, "learning_rate": 7.9e-06, "elbo": 9.40476, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.004691531785127844}
{"step": 85, "loss": 9.371302032470703, "grad_norm": 1.9357669353485107, "learning_rate": 8.400000000000001e-06, "elbo": 9.3713, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.004984752521698334}
{"step": 90, "loss": 9.327629089355469, "grad_norm": 1.962589979171753, "learning_rate": 8.9e-06, "elbo": 9.32763, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.005277973258268825}
{"step": 95, "loss": 9.285244750976563, "grad_norm": 1.8981937170028687, "learning_rate": 9.4e-06, "elbo": 9.28524, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.005571193994839315}
{"step": 100, "loss": 9.236039733886718, "grad_norm": 1.9614810943603516, "learning_rate": 9.900000000000002e-06, "elbo": 9.23604, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.005864414731409805}
{"step": 105, "loss": 9.192499542236328, "grad_norm": 1.931610107421875, "learning_rate": 1.04e-05, "elbo": 9.1925, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.006157635467980296}
{"step": 110, "loss": 9.167371368408203, "grad_norm": 1.8791650533676147, "learning_rate": 1.09e-05, "elbo": 9.16737, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.006450856204550786}
{"step": 115, "loss": 9.119976043701172, "grad_norm": 1.9685463905334473, "learning_rate": 1.1400000000000001e-05, "elbo": 9.11998, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.006744076941121276}
{"step": 120, "loss": 9.063107299804688, "grad_norm": 1.8967530727386475, "learning_rate": 1.19e-05, "elbo": 9.06311, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.007037297677691766}
{"step": 125, "loss": 9.008465576171876, "grad_norm": 1.8288639783859253, "learning_rate": 1.24e-05, "elbo": 9.00847, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.007330518414262257}
{"step": 130, "loss": 8.96900634765625, "grad_norm": 1.874037742614746, "learning_rate": 1.29e-05, "elbo": 8.96901, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0076237391508327465}
{"step": 135, "loss": 8.912767028808593, "grad_norm": 1.8908793926239014, "learning_rate": 1.3400000000000002e-05, "elbo": 8.91277, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.007916959887403237}
{"step": 140, "loss": 8.863271331787109, "grad_norm": 1.7356853485107422, "learning_rate": 1.3900000000000002e-05, "elbo": 8.86327, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.008210180623973728}
{"step": 145, "loss": 8.797885131835937, "grad_norm": 1.724465250968933, "learning_rate": 1.44e-05, "elbo": 8.79789, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.008503401360544218}
{"step": 150, "loss": 8.7541259765625, "grad_norm": 1.7602684497833252, "learning_rate": 1.49e-05, "elbo": 8.75413, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.008796622097114707}
{"step": 155, "loss": 8.701918029785157, "grad_norm": 1.9564487934112549, "learning_rate": 1.54e-05, "elbo": 8.70192, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.009089842833685198}
{"step": 155, "loss": 8.701918029785157, "grad_norm": 1.9564487934112549, "learning_rate": 1.54e-05, "elbo": 8.70192, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.009089842833685198}
{"step": 160, "loss": 8.629994964599609, "grad_norm": 1.6649842262268066, "learning_rate": 1.59e-05, "elbo": 8.62999, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.009383063570255689}
{"step": 165, "loss": 8.580753326416016, "grad_norm": 1.6218488216400146, "learning_rate": 1.6400000000000002e-05, "elbo": 8.58075, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.00967628430682618}
{"step": 170, "loss": 8.529246520996093, "grad_norm": 1.6853879690170288, "learning_rate": 1.69e-05, "elbo": 8.52925, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.009969505043396668}
{"step": 175, "loss": 8.474040985107422, "grad_norm": 1.7303943634033203, "learning_rate": 1.74e-05, "elbo": 8.47404, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.010262725779967159}
{"step": 180, "loss": 8.401912689208984, "grad_norm": 1.924438714981079, "learning_rate": 1.79e-05, "elbo": 8.40191, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.01055594651653765}
{"step": 185, "loss": 8.358472442626953, "grad_norm": 1.6992523670196533, "learning_rate": 1.84e-05, "elbo": 8.35847, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.01084916725310814}
{"step": 190, "loss": 8.301174926757813, "grad_norm": 1.5179553031921387, "learning_rate": 1.8900000000000002e-05, "elbo": 8.30117, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.01114238798967863}
{"step": 195, "loss": 8.254473876953124, "grad_norm": 1.4154189825057983, "learning_rate": 1.94e-05, "elbo": 8.25447, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.01143560872624912}
{"step": 200, "loss": 8.19441909790039, "grad_norm": 1.6819829940795898, "learning_rate": 1.9900000000000003e-05, "elbo": 8.19442, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.01172882946281961}
{"step": 205, "loss": 8.146116638183594, "grad_norm": 1.3216034173965454, "learning_rate": 2.04e-05, "elbo": 8.14612, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.012022050199390101}
{"step": 210, "loss": 8.092884826660157, "grad_norm": 1.2454771995544434, "learning_rate": 2.09e-05, "elbo": 8.09288, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.012315270935960592}
{"step": 215, "loss": 8.046769714355468, "grad_norm": 1.2777314186096191, "learning_rate": 2.1400000000000002e-05, "elbo": 8.04677, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.01260849167253108}
{"step": 220, "loss": 7.982801818847657, "grad_norm": 1.39699387550354, "learning_rate": 2.19e-05, "elbo": 7.9828, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.012901712409101571}
{"step": 225, "loss": 7.952233123779297, "grad_norm": 1.2393653392791748, "learning_rate": 2.2400000000000002e-05, "elbo": 7.95223, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.013194933145672062}
{"step": 230, "loss": 7.909907531738281, "grad_norm": 1.0895955562591553, "learning_rate": 2.29e-05, "elbo": 7.90991, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.013488153882242553}
{"step": 235, "loss": 7.870196533203125, "grad_norm": 1.25977623462677, "learning_rate": 2.3400000000000003e-05, "elbo": 7.8702, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.013781374618813043}
{"step": 240, "loss": 7.816111755371094, "grad_norm": 1.0836719274520874, "learning_rate": 2.39e-05, "elbo": 7.81611, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.014074595355383532}
{"step": 245, "loss": 7.789069366455078, "grad_norm": 0.9781352281570435, "learning_rate": 2.44e-05, "elbo": 7.78907, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.014367816091954023}
{"step": 250, "loss": 7.768223571777344, "grad_norm": 0.8694442510604858, "learning_rate": 2.4900000000000002e-05, "elbo": 7.76822, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.014661036828524513}
{"step": 255, "loss": 7.7146453857421875, "grad_norm": 1.0528538227081299, "learning_rate": 2.54e-05, "elbo": 7.71465, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.014954257565095004}
{"step": 260, "loss": 7.702301025390625, "grad_norm": 1.2861250638961792, "learning_rate": 2.5900000000000003e-05, "elbo": 7.7023, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.015247478301665493}
{"step": 265, "loss": 7.661091613769531, "grad_norm": 1.0670582056045532, "learning_rate": 2.64e-05, "elbo": 7.66109, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.015540699038235984}
{"step": 270, "loss": 7.645149230957031, "grad_norm": 0.8333109021186829, "learning_rate": 2.6900000000000003e-05, "elbo": 7.64515, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.015833919774806474}
{"step": 275, "loss": 7.625498962402344, "grad_norm": 0.6192725300788879, "learning_rate": 2.7400000000000002e-05, "elbo": 7.6255, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.016127140511376963}
{"step": 280, "loss": 7.612419891357422, "grad_norm": 1.829878807067871, "learning_rate": 2.7900000000000004e-05, "elbo": 7.61242, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.016420361247947456}
{"step": 285, "loss": 7.581439208984375, "grad_norm": 0.7163926362991333, "learning_rate": 2.84e-05, "elbo": 7.58144, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.016713581984517945}
{"step": 290, "loss": 7.5772041320800785, "grad_norm": 0.7965758442878723, "learning_rate": 2.8899999999999998e-05, "elbo": 7.5772, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.017006802721088437}
{"step": 295, "loss": 7.553953552246094, "grad_norm": 1.3220690488815308, "learning_rate": 2.94e-05, "elbo": 7.55395, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.017300023457658926}
{"step": 300, "loss": 7.554412841796875, "grad_norm": 1.821884036064148, "learning_rate": 2.9900000000000002e-05, "elbo": 7.55441, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.017593244194229415}
{"step": 305, "loss": 7.548698425292969, "grad_norm": 1.389422059059143, "learning_rate": 3.04e-05, "elbo": 7.5487, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.017886464930799907}
{"step": 305, "loss": 7.548698425292969, "grad_norm": 1.389422059059143, "learning_rate": 3.04e-05, "elbo": 7.5487, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.017886464930799907}
{"step": 310, "loss": 7.521994018554688, "grad_norm": 0.7039105892181396, "learning_rate": 3.09e-05, "elbo": 7.52199, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.018179685667370396}
{"step": 315, "loss": 7.518759918212891, "grad_norm": 0.8242306113243103, "learning_rate": 3.1400000000000004e-05, "elbo": 7.51876, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.01847290640394089}
{"step": 320, "loss": 7.503944396972656, "grad_norm": 1.5265400409698486, "learning_rate": 3.19e-05, "elbo": 7.50394, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.018766127140511377}
{"step": 325, "loss": 7.511822509765625, "grad_norm": 1.928954839706421, "learning_rate": 3.24e-05, "elbo": 7.51182, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.019059347877081866}
{"step": 330, "loss": 7.492367553710937, "grad_norm": 1.064961314201355, "learning_rate": 3.29e-05, "elbo": 7.49237, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.01935256861365236}
{"step": 335, "loss": 7.4856689453125, "grad_norm": 0.8707479238510132, "learning_rate": 3.3400000000000005e-05, "elbo": 7.48567, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.019645789350222848}
{"step": 340, "loss": 7.486344909667968, "grad_norm": 1.747108817100525, "learning_rate": 3.3900000000000004e-05, "elbo": 7.48634, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.019939010086793336}
{"step": 345, "loss": 7.4907470703125, "grad_norm": 1.3938325643539429, "learning_rate": 3.4399999999999996e-05, "elbo": 7.49075, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.02023223082336383}
{"step": 350, "loss": 7.475447082519532, "grad_norm": 1.7938463687896729, "learning_rate": 3.49e-05, "elbo": 7.47545, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.020525451559934318}
{"step": 355, "loss": 7.466240692138672, "grad_norm": 2.624174118041992, "learning_rate": 3.54e-05, "elbo": 7.46624, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.02081867229650481}
{"step": 360, "loss": 7.483716583251953, "grad_norm": 2.119776964187622, "learning_rate": 3.59e-05, "elbo": 7.48372, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0211118930330753}
{"step": 365, "loss": 7.451760864257812, "grad_norm": 1.4433372020721436, "learning_rate": 3.6400000000000004e-05, "elbo": 7.45176, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.021405113769645788}
{"step": 370, "loss": 7.459950256347656, "grad_norm": 2.836400032043457, "learning_rate": 3.69e-05, "elbo": 7.45995, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.02169833450621628}
{"step": 375, "loss": 7.444071960449219, "grad_norm": 1.1292959451675415, "learning_rate": 3.74e-05, "elbo": 7.44407, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.02199155524278677}
{"step": 380, "loss": 7.447001647949219, "grad_norm": 3.148174285888672, "learning_rate": 3.79e-05, "elbo": 7.447, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.02228477597935726}
{"step": 385, "loss": 7.447611236572266, "grad_norm": 2.300147771835327, "learning_rate": 3.8400000000000005e-05, "elbo": 7.44761, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.02257799671592775}
{"step": 390, "loss": 7.441314697265625, "grad_norm": 2.5889089107513428, "learning_rate": 3.8900000000000004e-05, "elbo": 7.44131, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.02287121745249824}
{"step": 395, "loss": 7.433176422119141, "grad_norm": 0.9860729575157166, "learning_rate": 3.94e-05, "elbo": 7.43318, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.023164438189068732}
{"step": 400, "loss": 7.424050903320312, "grad_norm": 2.9547488689422607, "learning_rate": 3.99e-05, "elbo": 7.42405, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.02345765892563922}
{"step": 405, "loss": 7.4241691589355465, "grad_norm": 3.0620534420013428, "learning_rate": 4.0400000000000006e-05, "elbo": 7.42417, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.023750879662209713}
{"step": 410, "loss": 7.4091033935546875, "grad_norm": 2.2797815799713135, "learning_rate": 4.09e-05, "elbo": 7.4091, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.024044100398780202}
{"step": 415, "loss": 7.395671844482422, "grad_norm": 3.0201029777526855, "learning_rate": 4.14e-05, "elbo": 7.39567, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.02433732113535069}
{"step": 420, "loss": 7.38836669921875, "grad_norm": 1.9919530153274536, "learning_rate": 4.19e-05, "elbo": 7.38837, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.024630541871921183}
{"step": 425, "loss": 7.3760986328125, "grad_norm": 4.565190315246582, "learning_rate": 4.24e-05, "elbo": 7.3761, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.024923762608491672}
{"step": 430, "loss": 7.372272491455078, "grad_norm": 2.962057113647461, "learning_rate": 4.29e-05, "elbo": 7.37227, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.02521698334506216}
{"step": 435, "loss": 7.361956787109375, "grad_norm": 4.349313735961914, "learning_rate": 4.3400000000000005e-05, "elbo": 7.36196, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.025510204081632654}
{"step": 440, "loss": 7.342765808105469, "grad_norm": 2.2033684253692627, "learning_rate": 4.39e-05, "elbo": 7.34277, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.025803424818203143}
{"step": 445, "loss": 7.347283172607422, "grad_norm": 3.0059759616851807, "learning_rate": 4.44e-05, "elbo": 7.34728, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.026096645554773635}
{"step": 450, "loss": 7.321890258789063, "grad_norm": 2.889418125152588, "learning_rate": 4.49e-05, "elbo": 7.32189, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.026389866291344124}
{"step": 455, "loss": 7.307731628417969, "grad_norm": 2.8768844604492188, "learning_rate": 4.5400000000000006e-05, "elbo": 7.30773, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.026683087027914613}
{"step": 460, "loss": 7.286898803710938, "grad_norm": 2.2775256633758545, "learning_rate": 4.5900000000000004e-05, "elbo": 7.2869, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.026976307764485105}
{"step": 465, "loss": 7.289517974853515, "grad_norm": 2.2915914058685303, "learning_rate": 4.64e-05, "elbo": 7.28952, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.027269528501055594}
{"step": 470, "loss": 7.2838493347167965, "grad_norm": 3.0328457355499268, "learning_rate": 4.69e-05, "elbo": 7.28385, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.027562749237626086}
{"step": 475, "loss": 7.281768035888672, "grad_norm": 3.52927827835083, "learning_rate": 4.74e-05, "elbo": 7.28177, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.027855969974196575}
{"step": 480, "loss": 7.2668304443359375, "grad_norm": 3.0424609184265137, "learning_rate": 4.79e-05, "elbo": 7.26683, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.028149190710767064}
{"step": 485, "loss": 7.247576904296875, "grad_norm": 2.1521077156066895, "learning_rate": 4.8400000000000004e-05, "elbo": 7.24758, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.028442411447337557}
{"step": 490, "loss": 7.256070709228515, "grad_norm": 1.8632839918136597, "learning_rate": 4.89e-05, "elbo": 7.25607, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.028735632183908046}
{"step": 495, "loss": 7.254112243652344, "grad_norm": 4.264181137084961, "learning_rate": 4.94e-05, "elbo": 7.25411, "mask_frac": 0.4996, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.029028852920478538}
{"step": 500, "loss": 7.241679382324219, "grad_norm": 2.254246711730957, "learning_rate": 4.99e-05, "elbo": 7.24168, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.029322073657049027}
{"step": 505, "loss": 7.244352722167969, "grad_norm": 2.9336307048797607, "learning_rate": 5.0400000000000005e-05, "elbo": 7.24435, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.029615294393619516}
{"step": 510, "loss": 7.235144805908203, "grad_norm": 3.265122890472412, "learning_rate": 5.0900000000000004e-05, "elbo": 7.23514, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.029908515130190008}
{"step": 515, "loss": 7.2103118896484375, "grad_norm": 3.4265356063842773, "learning_rate": 5.14e-05, "elbo": 7.21031, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.030201735866760497}
{"step": 520, "loss": 7.196208190917969, "grad_norm": 3.702160358428955, "learning_rate": 5.19e-05, "elbo": 7.19621, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.030494956603330986}
{"step": 525, "loss": 7.198480224609375, "grad_norm": 2.689183473587036, "learning_rate": 5.2400000000000007e-05, "elbo": 7.19848, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03078817733990148}
{"step": 530, "loss": 7.166403961181641, "grad_norm": 3.7374446392059326, "learning_rate": 5.2900000000000005e-05, "elbo": 7.1664, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.031081398076471967}
{"step": 535, "loss": 7.1668037414550785, "grad_norm": 5.413199424743652, "learning_rate": 5.3400000000000004e-05, "elbo": 7.1668, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03137461881304246}
{"step": 540, "loss": 7.159618377685547, "grad_norm": 3.530359983444214, "learning_rate": 5.390000000000001e-05, "elbo": 7.15962, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03166783954961295}
{"step": 545, "loss": 7.143070983886719, "grad_norm": 2.479814052581787, "learning_rate": 5.440000000000001e-05, "elbo": 7.14307, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03196106028618344}
{"step": 550, "loss": 7.11619873046875, "grad_norm": 3.4709489345550537, "learning_rate": 5.4900000000000006e-05, "elbo": 7.1162, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.032254281022753926}
{"step": 555, "loss": 7.113214111328125, "grad_norm": 2.1275341510772705, "learning_rate": 5.5400000000000005e-05, "elbo": 7.11321, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03254750175932442}
{"step": 560, "loss": 7.093794250488282, "grad_norm": 3.065830945968628, "learning_rate": 5.590000000000001e-05, "elbo": 7.09379, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03284072249589491}
{"step": 565, "loss": 7.067970275878906, "grad_norm": 3.569607734680176, "learning_rate": 5.6399999999999995e-05, "elbo": 7.06797, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0331339432324654}
{"step": 570, "loss": 7.071418762207031, "grad_norm": 2.9970924854278564, "learning_rate": 5.69e-05, "elbo": 7.07142, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03342716396903589}
{"step": 575, "loss": 7.071140289306641, "grad_norm": 2.5760252475738525, "learning_rate": 5.74e-05, "elbo": 7.07114, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03372038470560638}
{"step": 580, "loss": 7.052984619140625, "grad_norm": 2.720843553543091, "learning_rate": 5.79e-05, "elbo": 7.05299, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.034013605442176874}
{"step": 585, "loss": 7.043546295166015, "grad_norm": 3.9123990535736084, "learning_rate": 5.8399999999999997e-05, "elbo": 7.04355, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03430682617874736}
{"step": 590, "loss": 7.008705139160156, "grad_norm": 2.740968704223633, "learning_rate": 5.89e-05, "elbo": 7.0087, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03460004691531785}
{"step": 595, "loss": 7.022434997558594, "grad_norm": 4.471524238586426, "learning_rate": 5.94e-05, "elbo": 7.02243, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03489326765188834}
{"step": 600, "loss": 6.996028137207031, "grad_norm": 2.940654993057251, "learning_rate": 5.99e-05, "elbo": 6.99603, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03518648838845883}
{"step": 605, "loss": 7.007231140136719, "grad_norm": 3.643775463104248, "learning_rate": 6.04e-05, "elbo": 7.00723, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.035479709125029325}
{"step": 610, "loss": 6.978254699707032, "grad_norm": 2.607917308807373, "learning_rate": 6.09e-05, "elbo": 6.97826, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.035772929861599814}
{"step": 615, "loss": 6.995501708984375, "grad_norm": 2.7716476917266846, "learning_rate": 6.14e-05, "elbo": 6.9955, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0360661505981703}
{"step": 620, "loss": 6.969767761230469, "grad_norm": 2.8595166206359863, "learning_rate": 6.19e-05, "elbo": 6.96977, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03635937133474079}
{"step": 625, "loss": 6.960858917236328, "grad_norm": 4.127187252044678, "learning_rate": 6.24e-05, "elbo": 6.96086, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03665259207131128}
{"step": 630, "loss": 6.932295227050782, "grad_norm": 3.709109306335449, "learning_rate": 6.29e-05, "elbo": 6.9323, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03694581280788178}
{"step": 635, "loss": 6.916038513183594, "grad_norm": 4.469090461730957, "learning_rate": 6.340000000000001e-05, "elbo": 6.91604, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.037239033544452266}
{"step": 640, "loss": 6.898457336425781, "grad_norm": 4.6927337646484375, "learning_rate": 6.390000000000001e-05, "elbo": 6.89846, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.037532254281022755}
{"step": 645, "loss": 6.884527587890625, "grad_norm": 4.280451774597168, "learning_rate": 6.440000000000001e-05, "elbo": 6.88453, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.037825475017593244}
{"step": 650, "loss": 6.8901123046875, "grad_norm": 4.005171298980713, "learning_rate": 6.49e-05, "elbo": 6.89011, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03811869575416373}
{"step": 655, "loss": 6.857077789306641, "grad_norm": 2.7763144969940186, "learning_rate": 6.54e-05, "elbo": 6.85708, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03841191649073422}
{"step": 660, "loss": 6.845001983642578, "grad_norm": 4.512447834014893, "learning_rate": 6.59e-05, "elbo": 6.845, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03870513722730472}
{"step": 665, "loss": 6.8405029296875, "grad_norm": 3.479581356048584, "learning_rate": 6.64e-05, "elbo": 6.8405, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.038998357963875206}
{"step": 670, "loss": 6.817413330078125, "grad_norm": 3.5989205837249756, "learning_rate": 6.690000000000001e-05, "elbo": 6.81741, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.039291578700445695}
{"step": 675, "loss": 6.800352478027344, "grad_norm": 4.976712703704834, "learning_rate": 6.740000000000001e-05, "elbo": 6.80035, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.039584799437016184}
{"step": 680, "loss": 6.807061767578125, "grad_norm": 5.046385288238525, "learning_rate": 6.790000000000001e-05, "elbo": 6.80706, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.03987802017358667}
{"step": 685, "loss": 6.772757720947266, "grad_norm": 3.139475107192993, "learning_rate": 6.840000000000001e-05, "elbo": 6.77276, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04017124091015717}
{"step": 690, "loss": 6.791271209716797, "grad_norm": 3.9592981338500977, "learning_rate": 6.89e-05, "elbo": 6.79127, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04046446164672766}
{"step": 695, "loss": 6.767657470703125, "grad_norm": 3.7842934131622314, "learning_rate": 6.939999999999999e-05, "elbo": 6.76766, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04075768238329815}
{"step": 700, "loss": 6.7245735168457035, "grad_norm": 3.44053316116333, "learning_rate": 6.99e-05, "elbo": 6.72457, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.041050903119868636}
{"step": 705, "loss": 6.735823059082032, "grad_norm": 4.712395191192627, "learning_rate": 7.04e-05, "elbo": 6.73582, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.041344123856439124}
{"step": 710, "loss": 6.706550598144531, "grad_norm": 3.3229892253875732, "learning_rate": 7.09e-05, "elbo": 6.70655, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04163734459300962}
{"step": 715, "loss": 6.691983795166015, "grad_norm": 4.697400093078613, "learning_rate": 7.14e-05, "elbo": 6.69198, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04193056532958011}
{"step": 720, "loss": 6.659974670410156, "grad_norm": 5.49943208694458, "learning_rate": 7.19e-05, "elbo": 6.65997, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0422237860661506}
{"step": 725, "loss": 6.644309997558594, "grad_norm": 4.790216445922852, "learning_rate": 7.24e-05, "elbo": 6.64431, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04251700680272109}
{"step": 730, "loss": 6.612455749511719, "grad_norm": 5.583310127258301, "learning_rate": 7.29e-05, "elbo": 6.61246, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.042810227539291576}
{"step": 735, "loss": 6.632963562011719, "grad_norm": 4.4725470542907715, "learning_rate": 7.340000000000001e-05, "elbo": 6.63296, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04310344827586207}
{"step": 740, "loss": 6.616123962402344, "grad_norm": 4.113293647766113, "learning_rate": 7.390000000000001e-05, "elbo": 6.61612, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04339666901243256}
{"step": 745, "loss": 6.578456115722656, "grad_norm": 2.625943422317505, "learning_rate": 7.44e-05, "elbo": 6.57846, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04368988974900305}
{"step": 750, "loss": 6.571685028076172, "grad_norm": 5.679938793182373, "learning_rate": 7.49e-05, "elbo": 6.57168, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04398311048557354}
{"step": 755, "loss": 6.546337890625, "grad_norm": 2.780986785888672, "learning_rate": 7.54e-05, "elbo": 6.54634, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04427633122214403}
{"step": 760, "loss": 6.5499717712402346, "grad_norm": 2.8950886726379395, "learning_rate": 7.59e-05, "elbo": 6.54997, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04456955195871452}
{"step": 765, "loss": 6.530178833007812, "grad_norm": 4.685689926147461, "learning_rate": 7.64e-05, "elbo": 6.53018, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04486277269528501}
{"step": 770, "loss": 6.5244293212890625, "grad_norm": 3.9769508838653564, "learning_rate": 7.69e-05, "elbo": 6.52443, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0451559934318555}
{"step": 775, "loss": 6.504503631591797, "grad_norm": 4.13289213180542, "learning_rate": 7.740000000000001e-05, "elbo": 6.5045, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04544921416842599}
{"step": 780, "loss": 6.5063224792480465, "grad_norm": 3.359389305114746, "learning_rate": 7.790000000000001e-05, "elbo": 6.50632, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04574243490499648}
{"step": 785, "loss": 6.477024841308594, "grad_norm": 3.5012896060943604, "learning_rate": 7.840000000000001e-05, "elbo": 6.47702, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.046035655641566975}
{"step": 785, "loss": 6.477024841308594, "grad_norm": 3.5012896060943604, "learning_rate": 7.840000000000001e-05, "elbo": 6.47702, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.046035655641566975}
{"step": 790, "loss": 6.45162353515625, "grad_norm": 3.622696876525879, "learning_rate": 7.890000000000001e-05, "elbo": 6.45162, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.046328876378137464}
{"step": 795, "loss": 6.453890991210938, "grad_norm": 3.8139381408691406, "learning_rate": 7.94e-05, "elbo": 6.45389, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04662209711470795}
{"step": 800, "loss": 6.432835388183594, "grad_norm": 2.978405714035034, "learning_rate": 7.99e-05, "elbo": 6.43284, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04691531785127844}
{"step": 805, "loss": 6.430967712402344, "grad_norm": 4.445767402648926, "learning_rate": 8.04e-05, "elbo": 6.43097, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04720853858784893}
{"step": 810, "loss": 6.410301208496094, "grad_norm": 3.759596824645996, "learning_rate": 8.090000000000001e-05, "elbo": 6.4103, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.047501759324419426}
{"step": 815, "loss": 6.4132331848144535, "grad_norm": 4.166411399841309, "learning_rate": 8.14e-05, "elbo": 6.41323, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.047794980060989915}
{"step": 820, "loss": 6.397861480712891, "grad_norm": 3.9450364112854004, "learning_rate": 8.19e-05, "elbo": 6.39786, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.048088200797560404}
{"step": 825, "loss": 6.397912979125977, "grad_norm": 4.591938495635986, "learning_rate": 8.24e-05, "elbo": 6.39791, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04838142153413089}
{"step": 830, "loss": 6.367859649658203, "grad_norm": 1.9852029085159302, "learning_rate": 8.29e-05, "elbo": 6.36786, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04867464227070138}
{"step": 835, "loss": 6.347188568115234, "grad_norm": 4.583708763122559, "learning_rate": 8.34e-05, "elbo": 6.34719, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04896786300727187}
{"step": 840, "loss": 6.363375854492188, "grad_norm": 3.350879192352295, "learning_rate": 8.39e-05, "elbo": 6.36338, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.04926108374384237}
{"step": 845, "loss": 6.337858581542969, "grad_norm": 3.4169673919677734, "learning_rate": 8.44e-05, "elbo": 6.33786, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.049554304480412856}
{"step": 850, "loss": 6.322226715087891, "grad_norm": 4.406144142150879, "learning_rate": 8.49e-05, "elbo": 6.32223, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.049847525216983345}
{"step": 855, "loss": 6.306781005859375, "grad_norm": 2.217327117919922, "learning_rate": 8.54e-05, "elbo": 6.30678, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.050140745953553834}
{"step": 860, "loss": 6.316825103759766, "grad_norm": 3.6831271648406982, "learning_rate": 8.59e-05, "elbo": 6.31682, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05043396669012432}
{"step": 865, "loss": 6.3178871154785154, "grad_norm": 4.041606903076172, "learning_rate": 8.64e-05, "elbo": 6.31789, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05072718742669482}
{"step": 870, "loss": 6.293086242675781, "grad_norm": 4.199977874755859, "learning_rate": 8.69e-05, "elbo": 6.29309, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05102040816326531}
{"step": 875, "loss": 6.271713638305664, "grad_norm": 3.3730390071868896, "learning_rate": 8.740000000000001e-05, "elbo": 6.27171, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.051313628899835796}
{"step": 880, "loss": 6.260800170898437, "grad_norm": 2.7815043926239014, "learning_rate": 8.790000000000001e-05, "elbo": 6.2608, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.051606849636406285}
{"step": 885, "loss": 6.273345184326172, "grad_norm": 3.021195650100708, "learning_rate": 8.840000000000001e-05, "elbo": 6.27335, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.051900070372976774}
{"step": 890, "loss": 6.254629516601563, "grad_norm": 2.712017297744751, "learning_rate": 8.89e-05, "elbo": 6.25463, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05219329110954727}
{"step": 895, "loss": 6.248159790039063, "grad_norm": 2.7783875465393066, "learning_rate": 8.94e-05, "elbo": 6.24816, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05248651184611776}
{"step": 900, "loss": 6.231017303466797, "grad_norm": 3.697244167327881, "learning_rate": 8.99e-05, "elbo": 6.23102, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05277973258268825}
{"step": 905, "loss": 6.205392074584961, "grad_norm": 2.5942554473876953, "learning_rate": 9.04e-05, "elbo": 6.20539, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05307295331925874}
{"step": 910, "loss": 6.198539733886719, "grad_norm": 2.405848503112793, "learning_rate": 9.090000000000001e-05, "elbo": 6.19854, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.053366174055829226}
{"step": 915, "loss": 6.189400482177734, "grad_norm": 3.8281590938568115, "learning_rate": 9.140000000000001e-05, "elbo": 6.1894, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05365939479239972}
{"step": 920, "loss": 6.172320556640625, "grad_norm": 3.502168655395508, "learning_rate": 9.190000000000001e-05, "elbo": 6.17232, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05395261552897021}
{"step": 925, "loss": 6.177680206298828, "grad_norm": 2.530919313430786, "learning_rate": 9.240000000000001e-05, "elbo": 6.17768, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0542458362655407}
{"step": 930, "loss": 6.160501861572266, "grad_norm": 2.315645694732666, "learning_rate": 9.290000000000001e-05, "elbo": 6.1605, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05453905700211119}
{"step": 935, "loss": 6.157517623901367, "grad_norm": 2.872652769088745, "learning_rate": 9.340000000000001e-05, "elbo": 6.15752, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05483227773868168}
{"step": 940, "loss": 6.143212890625, "grad_norm": 3.218564748764038, "learning_rate": 9.39e-05, "elbo": 6.14321, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05512549847525217}
{"step": 945, "loss": 6.1409656524658205, "grad_norm": 4.614645004272461, "learning_rate": 9.44e-05, "elbo": 6.14097, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05541871921182266}
{"step": 950, "loss": 6.154094696044922, "grad_norm": 2.6353952884674072, "learning_rate": 9.49e-05, "elbo": 6.1541, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05571193994839315}
{"step": 955, "loss": 6.127119827270508, "grad_norm": 2.812032461166382, "learning_rate": 9.54e-05, "elbo": 6.12712, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05600516068496364}
{"step": 960, "loss": 6.115871810913086, "grad_norm": 3.6853115558624268, "learning_rate": 9.59e-05, "elbo": 6.11587, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05629838142153413}
{"step": 965, "loss": 6.116252517700195, "grad_norm": 3.32552433013916, "learning_rate": 9.64e-05, "elbo": 6.11625, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.056591602158104624}
{"step": 970, "loss": 6.0971321105957035, "grad_norm": 3.293760061264038, "learning_rate": 9.69e-05, "elbo": 6.09713, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05688482289467511}
{"step": 975, "loss": 6.093246841430664, "grad_norm": 3.8014540672302246, "learning_rate": 9.74e-05, "elbo": 6.09325, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0571780436312456}
{"step": 980, "loss": 6.079878234863282, "grad_norm": 2.846625566482544, "learning_rate": 9.790000000000001e-05, "elbo": 6.07988, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05747126436781609}
{"step": 985, "loss": 6.066719436645508, "grad_norm": 2.3690359592437744, "learning_rate": 9.84e-05, "elbo": 6.06672, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05776448510438658}
{"step": 990, "loss": 6.069704818725586, "grad_norm": 1.9344216585159302, "learning_rate": 9.89e-05, "elbo": 6.0697, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.058057705840957076}
{"step": 995, "loss": 6.043212890625, "grad_norm": 2.3303351402282715, "learning_rate": 9.94e-05, "elbo": 6.04321, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.058350926577527565}
{"step": 1000, "loss": 6.044330596923828, "grad_norm": 2.281099557876587, "learning_rate": 9.99e-05, "elbo": 6.04433, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.058644147314098054}
{"step": 1005, "loss": 6.026134490966797, "grad_norm": 3.086951732635498, "learning_rate": 0.0001, "elbo": 6.02613, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05893736805066854}
{"step": 1010, "loss": 6.022621917724609, "grad_norm": 2.462397575378418, "learning_rate": 0.0001, "elbo": 6.02262, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05923058878723903}
{"step": 1015, "loss": 6.014914321899414, "grad_norm": 2.4440317153930664, "learning_rate": 0.0001, "elbo": 6.01491, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.05952380952380952}
{"step": 1020, "loss": 6.007720947265625, "grad_norm": 1.942592740058899, "learning_rate": 0.0001, "elbo": 6.00772, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.059817030260380016}
{"step": 1025, "loss": 6.00762710571289, "grad_norm": 2.0504560470581055, "learning_rate": 0.0001, "elbo": 6.00763, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.060110250996950505}
{"step": 1030, "loss": 6.008885192871094, "grad_norm": 2.7140982151031494, "learning_rate": 0.0001, "elbo": 6.00889, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.060403471733520994}
{"step": 1035, "loss": 5.987040710449219, "grad_norm": 3.2940542697906494, "learning_rate": 0.0001, "elbo": 5.98704, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06069669247009148}
{"step": 1040, "loss": 6.009857177734375, "grad_norm": 2.9750428199768066, "learning_rate": 0.0001, "elbo": 6.00986, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06098991320666197}
{"step": 1045, "loss": 5.987844085693359, "grad_norm": 2.0283069610595703, "learning_rate": 0.0001, "elbo": 5.98784, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06128313394323247}
{"step": 1050, "loss": 5.950113677978516, "grad_norm": 2.4798202514648438, "learning_rate": 0.0001, "elbo": 5.95011, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06157635467980296}
{"step": 1055, "loss": 5.973909759521485, "grad_norm": 2.4190680980682373, "learning_rate": 0.0001, "elbo": 5.97391, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.061869575416373446}
{"step": 1060, "loss": 5.973874282836914, "grad_norm": 2.2425975799560547, "learning_rate": 0.0001, "elbo": 5.97387, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.062162796152943935}
{"step": 1065, "loss": 5.943381118774414, "grad_norm": 2.5147926807403564, "learning_rate": 0.0001, "elbo": 5.94338, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.062456016889514424}
{"step": 1070, "loss": 5.948942184448242, "grad_norm": 2.0704739093780518, "learning_rate": 0.0001, "elbo": 5.94894, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06274923762608492}
{"step": 1075, "loss": 5.929697418212891, "grad_norm": 2.134849786758423, "learning_rate": 0.0001, "elbo": 5.9297, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0630424583626554}
{"step": 1080, "loss": 5.930939865112305, "grad_norm": 2.974104166030884, "learning_rate": 0.0001, "elbo": 5.93094, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0633356790992259}
{"step": 1085, "loss": 5.910593414306641, "grad_norm": 2.2162322998046875, "learning_rate": 0.0001, "elbo": 5.91059, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0636288998357964}
{"step": 1090, "loss": 5.9234062194824215, "grad_norm": 2.5652689933776855, "learning_rate": 0.0001, "elbo": 5.92341, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06392212057236688}
{"step": 1095, "loss": 5.913240051269531, "grad_norm": 2.4011197090148926, "learning_rate": 0.0001, "elbo": 5.91324, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06421534130893737}
{"step": 1100, "loss": 5.911714172363281, "grad_norm": 2.147601366043091, "learning_rate": 0.0001, "elbo": 5.91171, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06450856204550785}
{"step": 1105, "loss": 5.898237609863282, "grad_norm": 1.7167505025863647, "learning_rate": 0.0001, "elbo": 5.89824, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06480178278207835}
{"step": 1110, "loss": 5.880857467651367, "grad_norm": 2.1247501373291016, "learning_rate": 0.0001, "elbo": 5.88086, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06509500351864884}
{"step": 1115, "loss": 5.886954879760742, "grad_norm": 2.5366973876953125, "learning_rate": 0.0001, "elbo": 5.88695, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06538822425521933}
{"step": 1120, "loss": 5.893357849121093, "grad_norm": 2.492136240005493, "learning_rate": 0.0001, "elbo": 5.89336, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06568144499178982}
{"step": 1125, "loss": 5.86734619140625, "grad_norm": 1.9950774908065796, "learning_rate": 0.0001, "elbo": 5.86735, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0659746657283603}
{"step": 1130, "loss": 5.878724670410156, "grad_norm": 2.2143657207489014, "learning_rate": 0.0001, "elbo": 5.87872, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0662678864649308}
{"step": 1135, "loss": 5.895437240600586, "grad_norm": 3.0356600284576416, "learning_rate": 0.0001, "elbo": 5.89544, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0665611072015013}
{"step": 1140, "loss": 5.876200485229492, "grad_norm": 1.9428576231002808, "learning_rate": 0.0001, "elbo": 5.8762, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06685432793807178}
{"step": 1145, "loss": 5.848583984375, "grad_norm": 1.93916654586792, "learning_rate": 0.0001, "elbo": 5.84858, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06714754867464227}
{"step": 1150, "loss": 5.844795989990234, "grad_norm": 1.7285397052764893, "learning_rate": 0.0001, "elbo": 5.8448, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06744076941121276}
{"step": 1155, "loss": 5.826005554199218, "grad_norm": 2.0484938621520996, "learning_rate": 0.0001, "elbo": 5.82601, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06773399014778325}
{"step": 1160, "loss": 5.840854263305664, "grad_norm": 1.719922423362732, "learning_rate": 0.0001, "elbo": 5.84085, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06802721088435375}
{"step": 1165, "loss": 5.820372772216797, "grad_norm": 1.859072208404541, "learning_rate": 0.0001, "elbo": 5.82037, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06832043162092423}
{"step": 1165, "loss": 5.820372772216797, "grad_norm": 1.859072208404541, "learning_rate": 0.0001, "elbo": 5.82037, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06832043162092423}
{"step": 1170, "loss": 5.802307891845703, "grad_norm": 2.6593103408813477, "learning_rate": 0.0001, "elbo": 5.80231, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06861365235749473}
{"step": 1175, "loss": 5.817132949829102, "grad_norm": 2.339778423309326, "learning_rate": 0.0001, "elbo": 5.81713, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06890687309406521}
{"step": 1180, "loss": 5.8191478729248045, "grad_norm": 2.190136194229126, "learning_rate": 0.0001, "elbo": 5.81915, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0692000938306357}
{"step": 1185, "loss": 5.805575561523438, "grad_norm": 2.577291488647461, "learning_rate": 0.0001, "elbo": 5.80558, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0694933145672062}
{"step": 1190, "loss": 5.791278076171875, "grad_norm": 1.5683233737945557, "learning_rate": 0.0001, "elbo": 5.79128, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.06978653530377668}
{"step": 1195, "loss": 5.8001564025878904, "grad_norm": 1.8295139074325562, "learning_rate": 0.0001, "elbo": 5.80016, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07007975604034718}
{"step": 1200, "loss": 5.804066848754883, "grad_norm": 1.444475769996643, "learning_rate": 0.0001, "elbo": 5.80407, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07037297677691766}
{"step": 1205, "loss": 5.7888648986816404, "grad_norm": 1.9755678176879883, "learning_rate": 0.0001, "elbo": 5.78887, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07066619751348815}
{"step": 1210, "loss": 5.773960876464844, "grad_norm": 1.7804861068725586, "learning_rate": 0.0001, "elbo": 5.77396, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07095941825005865}
{"step": 1215, "loss": 5.767090606689453, "grad_norm": 2.122293710708618, "learning_rate": 0.0001, "elbo": 5.76709, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07125263898662913}
{"step": 1220, "loss": 5.770270538330078, "grad_norm": 1.7629472017288208, "learning_rate": 0.0001, "elbo": 5.77027, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07154585972319963}
{"step": 1225, "loss": 5.774205017089844, "grad_norm": 1.420184850692749, "learning_rate": 0.0001, "elbo": 5.7742, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07183908045977011}
{"step": 1230, "loss": 5.742815399169922, "grad_norm": 1.6309596300125122, "learning_rate": 0.0001, "elbo": 5.74281, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0721323011963406}
{"step": 1235, "loss": 5.749902725219727, "grad_norm": 1.6280606985092163, "learning_rate": 0.0001, "elbo": 5.7499, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0724255219329111}
{"step": 1240, "loss": 5.737998962402344, "grad_norm": 1.802838683128357, "learning_rate": 0.0001, "elbo": 5.738, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07271874266948158}
{"step": 1245, "loss": 5.7311546325683596, "grad_norm": 1.406936764717102, "learning_rate": 0.0001, "elbo": 5.73115, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07301196340605208}
{"step": 1250, "loss": 5.731855010986328, "grad_norm": 2.010554790496826, "learning_rate": 0.0001, "elbo": 5.73186, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07330518414262256}
{"step": 1255, "loss": 5.717388916015625, "grad_norm": 1.6959350109100342, "learning_rate": 0.0001, "elbo": 5.71739, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07359840487919306}
{"step": 1260, "loss": 5.728883361816406, "grad_norm": 1.544614315032959, "learning_rate": 0.0001, "elbo": 5.72888, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07389162561576355}
{"step": 1265, "loss": 5.713534164428711, "grad_norm": 1.9600435495376587, "learning_rate": 0.0001, "elbo": 5.71353, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07418484635233404}
{"step": 1270, "loss": 5.70401611328125, "grad_norm": 1.730939507484436, "learning_rate": 0.0001, "elbo": 5.70402, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07447806708890453}
{"step": 1275, "loss": 5.714927673339844, "grad_norm": 1.7026596069335938, "learning_rate": 0.0001, "elbo": 5.71493, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07477128782547501}
{"step": 1280, "loss": 5.71948356628418, "grad_norm": 1.6782236099243164, "learning_rate": 0.0001, "elbo": 5.71948, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07506450856204551}
{"step": 1285, "loss": 5.689502716064453, "grad_norm": 1.7985626459121704, "learning_rate": 0.0001, "elbo": 5.6895, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07535772929861599}
{"step": 1290, "loss": 5.675146102905273, "grad_norm": 1.5942983627319336, "learning_rate": 0.0001, "elbo": 5.67515, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07565095003518649}
{"step": 1295, "loss": 5.699903106689453, "grad_norm": 1.9295586347579956, "learning_rate": 0.0001, "elbo": 5.6999, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07594417077175698}
{"step": 1300, "loss": 5.693026733398438, "grad_norm": 1.5669374465942383, "learning_rate": 0.0001, "elbo": 5.69303, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07623739150832747}
{"step": 1305, "loss": 5.6700904846191404, "grad_norm": 1.7472915649414062, "learning_rate": 0.0001, "elbo": 5.67009, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07653061224489796}
{"step": 1310, "loss": 5.678384399414062, "grad_norm": 1.6671180725097656, "learning_rate": 0.0001, "elbo": 5.67838, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07682383298146844}
{"step": 1315, "loss": 5.6681465148925785, "grad_norm": 1.7454572916030884, "learning_rate": 0.0001, "elbo": 5.66815, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07711705371803894}
{"step": 1320, "loss": 5.661833190917969, "grad_norm": 2.114462375640869, "learning_rate": 0.0001, "elbo": 5.66183, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07741027445460943}
{"step": 1325, "loss": 5.665628051757812, "grad_norm": 2.223008394241333, "learning_rate": 0.0001, "elbo": 5.66563, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07770349519117992}
{"step": 1330, "loss": 5.669906997680664, "grad_norm": 1.462861180305481, "learning_rate": 0.0001, "elbo": 5.66991, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07799671592775041}
{"step": 1330, "loss": 5.669906997680664, "grad_norm": 1.462861180305481, "learning_rate": 0.0001, "elbo": 5.66991, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07799671592775041}
{"step": 1335, "loss": 5.630372619628906, "grad_norm": 1.6007416248321533, "learning_rate": 0.0001, "elbo": 5.63037, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0782899366643209}
{"step": 1340, "loss": 5.626885223388672, "grad_norm": 2.2866623401641846, "learning_rate": 0.0001, "elbo": 5.62689, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07858315740089139}
{"step": 1345, "loss": 5.649534606933594, "grad_norm": 1.7298041582107544, "learning_rate": 0.0001, "elbo": 5.64953, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07887637813746189}
{"step": 1350, "loss": 5.638214111328125, "grad_norm": 1.4581819772720337, "learning_rate": 0.0001, "elbo": 5.63821, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07916959887403237}
{"step": 1355, "loss": 5.629811859130859, "grad_norm": 2.096001148223877, "learning_rate": 0.0001, "elbo": 5.62981, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07946281961060286}
{"step": 1360, "loss": 5.634543609619141, "grad_norm": 1.706337571144104, "learning_rate": 0.0001, "elbo": 5.63454, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.07975604034717335}
{"step": 1365, "loss": 5.610284042358399, "grad_norm": 1.6616500616073608, "learning_rate": 0.0001, "elbo": 5.61028, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08004926108374384}
{"step": 1370, "loss": 5.60257568359375, "grad_norm": 1.5824193954467773, "learning_rate": 0.0001, "elbo": 5.60258, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08034248182031434}
{"step": 1375, "loss": 5.60522346496582, "grad_norm": 1.5745972394943237, "learning_rate": 0.0001, "elbo": 5.60522, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08063570255688482}
{"step": 1380, "loss": 5.615194702148438, "grad_norm": 1.8913947343826294, "learning_rate": 0.0001, "elbo": 5.61519, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08092892329345532}
{"step": 1385, "loss": 5.6042121887207035, "grad_norm": 2.2969539165496826, "learning_rate": 0.0001, "elbo": 5.60421, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0812221440300258}
{"step": 1390, "loss": 5.5910076141357425, "grad_norm": 1.7786465883255005, "learning_rate": 0.0001, "elbo": 5.59101, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0815153647665963}
{"step": 1395, "loss": 5.603875732421875, "grad_norm": 2.040985584259033, "learning_rate": 0.0001, "elbo": 5.60388, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08180858550316679}
{"step": 1400, "loss": 5.59454116821289, "grad_norm": 1.6066259145736694, "learning_rate": 0.0001, "elbo": 5.59454, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08210180623973727}
{"step": 1405, "loss": 5.583332443237305, "grad_norm": 1.5324504375457764, "learning_rate": 0.0001, "elbo": 5.58333, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08239502697630777}
{"step": 1410, "loss": 5.576168823242187, "grad_norm": 1.8420995473861694, "learning_rate": 0.0001, "elbo": 5.57617, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08268824771287825}
{"step": 1415, "loss": 5.576212310791016, "grad_norm": 1.2285863161087036, "learning_rate": 0.0001, "elbo": 5.57621, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08298146844944874}
{"step": 1415, "loss": 5.576212310791016, "grad_norm": 1.2285863161087036, "learning_rate": 0.0001, "elbo": 5.57621, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08298146844944874}
{"step": 1420, "loss": 5.579942321777343, "grad_norm": 1.6992775201797485, "learning_rate": 0.0001, "elbo": 5.57994, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08327468918601924}
{"step": 1425, "loss": 5.576903915405273, "grad_norm": 1.647340178489685, "learning_rate": 0.0001, "elbo": 5.5769, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08356790992258972}
{"step": 1430, "loss": 5.569424819946289, "grad_norm": 1.8890306949615479, "learning_rate": 0.0001, "elbo": 5.56942, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08386113065916022}
{"step": 1435, "loss": 5.559458160400391, "grad_norm": 1.4669487476348877, "learning_rate": 0.0001, "elbo": 5.55946, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0841543513957307}
{"step": 1440, "loss": 5.574175262451172, "grad_norm": 1.7951385974884033, "learning_rate": 0.0001, "elbo": 5.57418, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0844475721323012}
{"step": 1445, "loss": 5.54768180847168, "grad_norm": 1.4425801038742065, "learning_rate": 0.0001, "elbo": 5.54768, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08474079286887169}
{"step": 1450, "loss": 5.565180969238281, "grad_norm": 1.9815301895141602, "learning_rate": 0.0001, "elbo": 5.56518, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08503401360544217}
{"step": 1455, "loss": 5.56678352355957, "grad_norm": 1.6441653966903687, "learning_rate": 0.0001, "elbo": 5.56678, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08532723434201267}
{"step": 1460, "loss": 5.537227630615234, "grad_norm": 1.3454171419143677, "learning_rate": 0.0001, "elbo": 5.53723, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08562045507858315}
{"step": 1465, "loss": 5.528540802001953, "grad_norm": 1.7872103452682495, "learning_rate": 0.0001, "elbo": 5.52854, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08591367581515365}
{"step": 1470, "loss": 5.537783432006836, "grad_norm": 1.8874022960662842, "learning_rate": 0.0001, "elbo": 5.53778, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08620689655172414}
{"step": 1475, "loss": 5.505364227294922, "grad_norm": 1.4587854146957397, "learning_rate": 0.0001, "elbo": 5.50536, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08650011728829463}
{"step": 1480, "loss": 5.507867431640625, "grad_norm": 1.5645438432693481, "learning_rate": 0.0001, "elbo": 5.50787, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08679333802486512}
{"step": 1485, "loss": 5.492387390136718, "grad_norm": 1.494585633277893, "learning_rate": 0.0001, "elbo": 5.49239, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0870865587614356}
{"step": 1490, "loss": 5.519903945922851, "grad_norm": 2.190727710723877, "learning_rate": 0.0001, "elbo": 5.5199, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0873797794980061}
{"step": 1495, "loss": 5.504063415527344, "grad_norm": 1.4244486093521118, "learning_rate": 0.0001, "elbo": 5.50406, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0876730002345766}
{"step": 1500, "loss": 5.5203399658203125, "grad_norm": 1.5145708322525024, "learning_rate": 0.0001, "elbo": 5.52034, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08796622097114708}
{"step": 1505, "loss": 5.479319000244141, "grad_norm": 1.3973063230514526, "learning_rate": 0.0001, "elbo": 5.47932, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08825944170771757}
{"step": 1510, "loss": 5.488341903686523, "grad_norm": 1.3362886905670166, "learning_rate": 0.0001, "elbo": 5.48834, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08855266244428805}
{"step": 1515, "loss": 5.484632110595703, "grad_norm": 1.640687346458435, "learning_rate": 0.0001, "elbo": 5.48463, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08884588318085855}
{"step": 1520, "loss": 5.493888854980469, "grad_norm": 1.2850618362426758, "learning_rate": 0.0001, "elbo": 5.49389, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08913910391742905}
{"step": 1525, "loss": 5.482449340820312, "grad_norm": 1.3766169548034668, "learning_rate": 0.0001, "elbo": 5.48245, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08943232465399953}
{"step": 1530, "loss": 5.479377365112304, "grad_norm": 1.452475905418396, "learning_rate": 0.0001, "elbo": 5.47938, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.08972554539057002}
{"step": 1535, "loss": 5.502588272094727, "grad_norm": 1.726760983467102, "learning_rate": 0.0001, "elbo": 5.50259, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0900187661271405}
{"step": 1540, "loss": 5.464424896240234, "grad_norm": 1.7164591550827026, "learning_rate": 0.0001, "elbo": 5.46442, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.090311986863711}
{"step": 1545, "loss": 5.457987213134766, "grad_norm": 1.5679124593734741, "learning_rate": 0.0001, "elbo": 5.45799, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0906052076002815}
{"step": 1550, "loss": 5.470961380004883, "grad_norm": 1.7424564361572266, "learning_rate": 0.0001, "elbo": 5.47096, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09089842833685198}
{"step": 1555, "loss": 5.473760604858398, "grad_norm": 1.3766571283340454, "learning_rate": 0.0001, "elbo": 5.47376, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09119164907342248}
{"step": 1560, "loss": 5.4583282470703125, "grad_norm": 1.5145065784454346, "learning_rate": 0.0001, "elbo": 5.45833, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09148486980999296}
{"step": 1565, "loss": 5.452304840087891, "grad_norm": 1.4856325387954712, "learning_rate": 0.0001, "elbo": 5.4523, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09177809054656345}
{"step": 1570, "loss": 5.446224975585937, "grad_norm": 1.3909505605697632, "learning_rate": 0.0001, "elbo": 5.44623, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09207131128313395}
{"step": 1575, "loss": 5.436179351806641, "grad_norm": 1.8742233514785767, "learning_rate": 0.0001, "elbo": 5.43618, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09236453201970443}
{"step": 1575, "loss": 5.436179351806641, "grad_norm": 1.8742233514785767, "learning_rate": 0.0001, "elbo": 5.43618, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09236453201970443}
{"step": 1580, "loss": 5.452613067626953, "grad_norm": 1.5151764154434204, "learning_rate": 0.0001, "elbo": 5.45261, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09265775275627493}
{"step": 1580, "loss": 5.452613067626953, "grad_norm": 1.5151764154434204, "learning_rate": 0.0001, "elbo": 5.45261, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09265775275627493}
{"step": 1585, "loss": 5.419060897827149, "grad_norm": 1.364055871963501, "learning_rate": 0.0001, "elbo": 5.41906, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09295097349284541}
{"step": 1590, "loss": 5.432427978515625, "grad_norm": 1.564875841140747, "learning_rate": 0.0001, "elbo": 5.43243, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0932441942294159}
{"step": 1595, "loss": 5.431855773925781, "grad_norm": 1.3497470617294312, "learning_rate": 0.0001, "elbo": 5.43186, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0935374149659864}
{"step": 1600, "loss": 5.421163177490234, "grad_norm": 1.501900315284729, "learning_rate": 0.0001, "elbo": 5.42116, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09383063570255688}
{"step": 1605, "loss": 5.422922134399414, "grad_norm": 1.6023987531661987, "learning_rate": 0.0001, "elbo": 5.42292, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09412385643912738}
{"step": 1610, "loss": 5.4299263000488285, "grad_norm": 1.4562946557998657, "learning_rate": 0.0001, "elbo": 5.42993, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09441707717569786}
{"step": 1615, "loss": 5.412625885009765, "grad_norm": 1.5180517435073853, "learning_rate": 0.0001, "elbo": 5.41263, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09471029791226836}
{"step": 1620, "loss": 5.394305419921875, "grad_norm": 1.5216145515441895, "learning_rate": 0.0001, "elbo": 5.39431, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09500351864883885}
{"step": 1625, "loss": 5.413130187988282, "grad_norm": 1.4576340913772583, "learning_rate": 0.0001, "elbo": 5.41313, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09529673938540933}
{"step": 1630, "loss": 5.420223617553711, "grad_norm": 1.3236796855926514, "learning_rate": 0.0001, "elbo": 5.42022, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09558996012197983}
{"step": 1635, "loss": 5.399630355834961, "grad_norm": 1.4310894012451172, "learning_rate": 0.0001, "elbo": 5.39963, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09588318085855031}
{"step": 1640, "loss": 5.388528823852539, "grad_norm": 1.4203943014144897, "learning_rate": 0.0001, "elbo": 5.38853, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09617640159512081}
{"step": 1645, "loss": 5.3824913024902346, "grad_norm": 1.6006873846054077, "learning_rate": 0.0001, "elbo": 5.38249, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09646962233169129}
{"step": 1650, "loss": 5.388770294189453, "grad_norm": 1.8362679481506348, "learning_rate": 0.0001, "elbo": 5.38877, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09676284306826179}
{"step": 1655, "loss": 5.397698211669922, "grad_norm": 1.4640394449234009, "learning_rate": 0.0001, "elbo": 5.3977, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09705606380483228}
{"step": 1660, "loss": 5.381443023681641, "grad_norm": 1.3293490409851074, "learning_rate": 0.0001, "elbo": 5.38144, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09734928454140276}
{"step": 1665, "loss": 5.353569793701172, "grad_norm": 1.135095238685608, "learning_rate": 0.0001, "elbo": 5.35357, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09764250527797326}
{"step": 1670, "loss": 5.37060546875, "grad_norm": 1.337819218635559, "learning_rate": 0.0001, "elbo": 5.37061, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09793572601454374}
{"step": 1675, "loss": 5.381435394287109, "grad_norm": 1.862113118171692, "learning_rate": 0.0001, "elbo": 5.38144, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09822894675111424}
{"step": 1680, "loss": 5.365283966064453, "grad_norm": 1.5412505865097046, "learning_rate": 0.0001, "elbo": 5.36528, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09852216748768473}
{"step": 1685, "loss": 5.367770385742188, "grad_norm": 1.7769739627838135, "learning_rate": 0.0001, "elbo": 5.36777, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09881538822425522}
{"step": 1690, "loss": 5.350542449951172, "grad_norm": 1.5375072956085205, "learning_rate": 0.0001, "elbo": 5.35054, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09910860896082571}
{"step": 1695, "loss": 5.364511871337891, "grad_norm": 1.93940007686615, "learning_rate": 0.0001, "elbo": 5.36451, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.0994018296973962}
{"step": 1700, "loss": 5.340047454833984, "grad_norm": 1.2794506549835205, "learning_rate": 0.0001, "elbo": 5.34005, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09969505043396669}
{"step": 1705, "loss": 5.343333053588867, "grad_norm": 1.5117679834365845, "learning_rate": 0.0001, "elbo": 5.34333, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.09998827117053719}
{"step": 1710, "loss": 5.335862731933593, "grad_norm": 1.2715171575546265, "learning_rate": 0.0001, "elbo": 5.33586, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10028149190710767}
{"step": 1715, "loss": 5.3597969055175785, "grad_norm": 1.3627973794937134, "learning_rate": 0.0001, "elbo": 5.3598, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10057471264367816}
{"step": 1720, "loss": 5.338835906982422, "grad_norm": 1.2183690071105957, "learning_rate": 0.0001, "elbo": 5.33884, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10086793338024864}
{"step": 1725, "loss": 5.348732757568359, "grad_norm": 1.4427785873413086, "learning_rate": 0.0001, "elbo": 5.34873, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10116115411681914}
{"step": 1730, "loss": 5.329243469238281, "grad_norm": 1.6210435628890991, "learning_rate": 0.0001, "elbo": 5.32924, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10145437485338964}
{"step": 1735, "loss": 5.325817489624024, "grad_norm": 1.5221832990646362, "learning_rate": 0.0001, "elbo": 5.32582, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10174759558996012}
{"step": 1740, "loss": 5.353779602050781, "grad_norm": 1.2803337574005127, "learning_rate": 0.0001, "elbo": 5.35378, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10204081632653061}
{"step": 1745, "loss": 5.325409698486328, "grad_norm": 1.8872041702270508, "learning_rate": 0.0001, "elbo": 5.32541, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1023340370631011}
{"step": 1750, "loss": 5.318965148925781, "grad_norm": 1.417452335357666, "learning_rate": 0.0001, "elbo": 5.31897, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10262725779967159}
{"step": 1750, "loss": 5.318965148925781, "grad_norm": 1.417452335357666, "learning_rate": 0.0001, "elbo": 5.31897, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10262725779967159}
{"step": 1755, "loss": 5.305430221557617, "grad_norm": 1.3939166069030762, "learning_rate": 0.0001, "elbo": 5.30543, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10292047853624209}
{"step": 1760, "loss": 5.313982391357422, "grad_norm": 1.3330093622207642, "learning_rate": 0.0001, "elbo": 5.31398, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10321369927281257}
{"step": 1765, "loss": 5.322059249877929, "grad_norm": 1.2459667921066284, "learning_rate": 0.0001, "elbo": 5.32206, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10350692000938307}
{"step": 1770, "loss": 5.317482757568359, "grad_norm": 1.5584805011749268, "learning_rate": 0.0001, "elbo": 5.31748, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10380014074595355}
{"step": 1775, "loss": 5.318650817871093, "grad_norm": 1.7871376276016235, "learning_rate": 0.0001, "elbo": 5.31865, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10409336148252404}
{"step": 1780, "loss": 5.313266754150391, "grad_norm": 1.5689653158187866, "learning_rate": 0.0001, "elbo": 5.31327, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10438658221909454}
{"step": 1785, "loss": 5.300495529174805, "grad_norm": 1.3403207063674927, "learning_rate": 0.0001, "elbo": 5.3005, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10467980295566502}
{"step": 1790, "loss": 5.302235412597656, "grad_norm": 1.1976300477981567, "learning_rate": 0.0001, "elbo": 5.30224, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10497302369223552}
{"step": 1795, "loss": 5.327320861816406, "grad_norm": 1.3607947826385498, "learning_rate": 0.0001, "elbo": 5.32732, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.105266244428806}
{"step": 1800, "loss": 5.2925865173339846, "grad_norm": 1.4325939416885376, "learning_rate": 0.0001, "elbo": 5.29259, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1055594651653765}
{"step": 1805, "loss": 5.2926483154296875, "grad_norm": 1.1246578693389893, "learning_rate": 0.0001, "elbo": 5.29265, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10585268590194699}
{"step": 1810, "loss": 5.28243408203125, "grad_norm": 1.3129100799560547, "learning_rate": 0.0001, "elbo": 5.28243, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10614590663851747}
{"step": 1815, "loss": 5.279279708862305, "grad_norm": 1.1856602430343628, "learning_rate": 0.0001, "elbo": 5.27928, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10643912737508797}
{"step": 1820, "loss": 5.267265319824219, "grad_norm": 1.1722218990325928, "learning_rate": 0.0001, "elbo": 5.26727, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10673234811165845}
{"step": 1825, "loss": 5.274128723144531, "grad_norm": 1.4293568134307861, "learning_rate": 0.0001, "elbo": 5.27413, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10702556884822895}
{"step": 1830, "loss": 5.288044738769531, "grad_norm": 1.4941948652267456, "learning_rate": 0.0001, "elbo": 5.28804, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10731878958479944}
{"step": 1835, "loss": 5.270060348510742, "grad_norm": 1.7999039888381958, "learning_rate": 0.0001, "elbo": 5.27006, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10761201032136992}
{"step": 1840, "loss": 5.2976829528808596, "grad_norm": 1.306249976158142, "learning_rate": 0.0001, "elbo": 5.29768, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10790523105794042}
{"step": 1845, "loss": 5.274390792846679, "grad_norm": 1.8439046144485474, "learning_rate": 0.0001, "elbo": 5.27439, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1081984517945109}
{"step": 1850, "loss": 5.27305908203125, "grad_norm": 1.4159977436065674, "learning_rate": 0.0001, "elbo": 5.27306, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1084916725310814}
{"step": 1855, "loss": 5.275656127929688, "grad_norm": 1.3459175825119019, "learning_rate": 0.0001, "elbo": 5.27566, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1087848932676519}
{"step": 1860, "loss": 5.254342651367187, "grad_norm": 1.446482539176941, "learning_rate": 0.0001, "elbo": 5.25434, "mask_frac": 0.5014, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10907811400422238}
{"step": 1865, "loss": 5.26490592956543, "grad_norm": 1.2811980247497559, "learning_rate": 0.0001, "elbo": 5.26491, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10937133474079287}
{"step": 1870, "loss": 5.241669082641602, "grad_norm": 1.221207618713379, "learning_rate": 0.0001, "elbo": 5.24167, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10966455547736335}
{"step": 1875, "loss": 5.25934944152832, "grad_norm": 1.436643362045288, "learning_rate": 0.0001, "elbo": 5.25935, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.10995777621393385}
{"step": 1880, "loss": 5.236965942382812, "grad_norm": 1.32108473777771, "learning_rate": 0.0001, "elbo": 5.23697, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11025099695050435}
{"step": 1885, "loss": 5.2150421142578125, "grad_norm": 1.3972129821777344, "learning_rate": 0.0001, "elbo": 5.21504, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11054421768707483}
{"step": 1890, "loss": 5.241666793823242, "grad_norm": 1.2577441930770874, "learning_rate": 0.0001, "elbo": 5.24167, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11083743842364532}
{"step": 1895, "loss": 5.234326171875, "grad_norm": 1.5223262310028076, "learning_rate": 0.0001, "elbo": 5.23433, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1111306591602158}
{"step": 1900, "loss": 5.232312774658203, "grad_norm": 1.3298407793045044, "learning_rate": 0.0001, "elbo": 5.23231, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1114238798967863}
{"step": 1905, "loss": 5.2341064453125, "grad_norm": 1.3431038856506348, "learning_rate": 0.0001, "elbo": 5.23411, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1117171006333568}
{"step": 1910, "loss": 5.2129768371582035, "grad_norm": 1.2888758182525635, "learning_rate": 0.0001, "elbo": 5.21298, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11201032136992728}
{"step": 1915, "loss": 5.230998229980469, "grad_norm": 1.2108722925186157, "learning_rate": 0.0001, "elbo": 5.231, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11230354210649778}
{"step": 1920, "loss": 5.201280975341797, "grad_norm": 1.404004454612732, "learning_rate": 0.0001, "elbo": 5.20128, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11259676284306826}
{"step": 1925, "loss": 5.212445831298828, "grad_norm": 1.3694994449615479, "learning_rate": 0.0001, "elbo": 5.21245, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11288998357963875}
{"step": 1930, "loss": 5.240765380859375, "grad_norm": 1.3233774900436401, "learning_rate": 0.0001, "elbo": 5.24077, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11318320431620925}
{"step": 1935, "loss": 5.2197826385498045, "grad_norm": 1.4529075622558594, "learning_rate": 0.0001, "elbo": 5.21978, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11347642505277973}
{"step": 1940, "loss": 5.221485137939453, "grad_norm": 1.4335482120513916, "learning_rate": 0.0001, "elbo": 5.22148, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11376964578935023}
{"step": 1945, "loss": 5.211058807373047, "grad_norm": 1.4585540294647217, "learning_rate": 0.0001, "elbo": 5.21106, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11406286652592071}
{"step": 1950, "loss": 5.225931549072266, "grad_norm": 1.35467529296875, "learning_rate": 0.0001, "elbo": 5.22593, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1143560872624912}
{"step": 1955, "loss": 5.211449432373047, "grad_norm": 1.1999605894088745, "learning_rate": 0.0001, "elbo": 5.21145, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1146493079990617}
{"step": 1960, "loss": 5.183591842651367, "grad_norm": 1.1860431432724, "learning_rate": 0.0001, "elbo": 5.18359, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11494252873563218}
{"step": 1965, "loss": 5.194832611083984, "grad_norm": 1.576404333114624, "learning_rate": 0.0001, "elbo": 5.19483, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11523574947220268}
{"step": 1970, "loss": 5.1906993865966795, "grad_norm": 1.4075236320495605, "learning_rate": 0.0001, "elbo": 5.1907, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11552897020877316}
{"step": 1975, "loss": 5.190128326416016, "grad_norm": 1.4433348178863525, "learning_rate": 0.0001, "elbo": 5.19013, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11582219094534366}
{"step": 1980, "loss": 5.188206481933594, "grad_norm": 1.3434247970581055, "learning_rate": 0.0001, "elbo": 5.18821, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11611541168191415}
{"step": 1985, "loss": 5.178695297241211, "grad_norm": 1.4360648393630981, "learning_rate": 0.0001, "elbo": 5.1787, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11640863241848463}
{"step": 1990, "loss": 5.179191589355469, "grad_norm": 1.2653499841690063, "learning_rate": 0.0001, "elbo": 5.17919, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11670185315505513}
{"step": 1995, "loss": 5.174922943115234, "grad_norm": 1.4997786283493042, "learning_rate": 0.0001, "elbo": 5.17492, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11699507389162561}
{"step": 2000, "loss": 5.190386962890625, "grad_norm": 1.2694076299667358, "learning_rate": 0.0001, "elbo": 5.19039, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11728829462819611}
{"step": 2005, "loss": 5.170973205566407, "grad_norm": 1.395290493965149, "learning_rate": 0.0001, "elbo": 5.17097, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11758151536476659}
{"step": 2010, "loss": 5.170632171630859, "grad_norm": 1.3492636680603027, "learning_rate": 0.0001, "elbo": 5.17063, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11787473610133709}
{"step": 2015, "loss": 5.17076301574707, "grad_norm": 1.205538272857666, "learning_rate": 0.0001, "elbo": 5.17076, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11816795683790758}
{"step": 2020, "loss": 5.160328674316406, "grad_norm": 1.3822582960128784, "learning_rate": 0.0001, "elbo": 5.16033, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11846117757447806}
{"step": 2025, "loss": 5.1764671325683596, "grad_norm": 1.2532063722610474, "learning_rate": 0.0001, "elbo": 5.17647, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11875439831104856}
{"step": 2030, "loss": 5.17938232421875, "grad_norm": 1.224662184715271, "learning_rate": 0.0001, "elbo": 5.17938, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11904761904761904}
{"step": 2035, "loss": 5.168976593017578, "grad_norm": 1.313152551651001, "learning_rate": 0.0001, "elbo": 5.16898, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11934083978418954}
{"step": 2035, "loss": 5.168976593017578, "grad_norm": 1.313152551651001, "learning_rate": 0.0001, "elbo": 5.16898, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11934083978418954}
{"step": 2040, "loss": 5.171247482299805, "grad_norm": 1.2219425439834595, "learning_rate": 0.0001, "elbo": 5.17125, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11963406052076003}
{"step": 2040, "loss": 5.171247482299805, "grad_norm": 1.2219425439834595, "learning_rate": 0.0001, "elbo": 5.17125, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11963406052076003}
{"step": 2045, "loss": 5.16657600402832, "grad_norm": 1.3818120956420898, "learning_rate": 0.0001, "elbo": 5.16658, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.11992728125733051}
{"step": 2050, "loss": 5.142396545410156, "grad_norm": 1.3166205883026123, "learning_rate": 0.0001, "elbo": 5.1424, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12022050199390101}
{"step": 2055, "loss": 5.154984664916992, "grad_norm": 1.2023041248321533, "learning_rate": 0.0001, "elbo": 5.15498, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12051372273047149}
{"step": 2060, "loss": 5.154532623291016, "grad_norm": 1.2671480178833008, "learning_rate": 0.0001, "elbo": 5.15453, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12080694346704199}
{"step": 2065, "loss": 5.122574234008789, "grad_norm": 1.3774948120117188, "learning_rate": 0.0001, "elbo": 5.12257, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12110016420361248}
{"step": 2070, "loss": 5.140113830566406, "grad_norm": 1.221950650215149, "learning_rate": 0.0001, "elbo": 5.14011, "mask_frac": 0.5013, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12139338494018297}
{"step": 2075, "loss": 5.155398941040039, "grad_norm": 1.427563190460205, "learning_rate": 0.0001, "elbo": 5.1554, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12168660567675346}
{"step": 2080, "loss": 5.148586273193359, "grad_norm": 1.1820176839828491, "learning_rate": 0.0001, "elbo": 5.14859, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12197982641332394}
{"step": 2085, "loss": 5.1139476776123045, "grad_norm": 1.3777456283569336, "learning_rate": 0.0001, "elbo": 5.11395, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12227304714989444}
{"step": 2090, "loss": 5.110556793212891, "grad_norm": 1.5195015668869019, "learning_rate": 0.0001, "elbo": 5.11056, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12256626788646494}
{"step": 2095, "loss": 5.118608856201172, "grad_norm": 1.5616059303283691, "learning_rate": 0.0001, "elbo": 5.11861, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12285948862303542}
{"step": 2100, "loss": 5.135029602050781, "grad_norm": 1.4357327222824097, "learning_rate": 0.0001, "elbo": 5.13503, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12315270935960591}
{"step": 2105, "loss": 5.128987121582031, "grad_norm": 1.297671914100647, "learning_rate": 0.0001, "elbo": 5.12899, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1234459300961764}
{"step": 2110, "loss": 5.113686370849609, "grad_norm": 1.2133255004882812, "learning_rate": 0.0001, "elbo": 5.11369, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12373915083274689}
{"step": 2115, "loss": 5.122412872314453, "grad_norm": 1.319990634918213, "learning_rate": 0.0001, "elbo": 5.12241, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12403237156931739}
{"step": 2120, "loss": 5.117286682128906, "grad_norm": 1.3570058345794678, "learning_rate": 0.0001, "elbo": 5.11729, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12432559230588787}
{"step": 2125, "loss": 5.121057510375977, "grad_norm": 1.0855655670166016, "learning_rate": 0.0001, "elbo": 5.12106, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12461881304245837}
{"step": 2130, "loss": 5.110958862304687, "grad_norm": 1.2882845401763916, "learning_rate": 0.0001, "elbo": 5.11096, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12491203377902885}
{"step": 2135, "loss": 5.093230056762695, "grad_norm": 1.4657214879989624, "learning_rate": 0.0001, "elbo": 5.09323, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12520525451559936}
{"step": 2140, "loss": 5.1172538757324215, "grad_norm": 1.1005067825317383, "learning_rate": 0.0001, "elbo": 5.11725, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12549847525216984}
{"step": 2145, "loss": 5.085456085205078, "grad_norm": 1.0501444339752197, "learning_rate": 0.0001, "elbo": 5.08546, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12579169598874032}
{"step": 2150, "loss": 5.096225738525391, "grad_norm": 1.2004131078720093, "learning_rate": 0.0001, "elbo": 5.09623, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1260849167253108}
{"step": 2155, "loss": 5.10453987121582, "grad_norm": 1.3168096542358398, "learning_rate": 0.0001, "elbo": 5.10454, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1263781374618813}
{"step": 2160, "loss": 5.0851490020751955, "grad_norm": 1.175308108329773, "learning_rate": 0.0001, "elbo": 5.08515, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1266713581984518}
{"step": 2165, "loss": 5.110736465454101, "grad_norm": 1.300927996635437, "learning_rate": 0.0001, "elbo": 5.11074, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12696457893502228}
{"step": 2170, "loss": 5.077146911621094, "grad_norm": 1.2038425207138062, "learning_rate": 0.0001, "elbo": 5.07715, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1272577996715928}
{"step": 2175, "loss": 5.078502655029297, "grad_norm": 1.2979011535644531, "learning_rate": 0.0001, "elbo": 5.0785, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12755102040816327}
{"step": 2180, "loss": 5.0799713134765625, "grad_norm": 1.4091987609863281, "learning_rate": 0.0001, "elbo": 5.07997, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12784424114473375}
{"step": 2185, "loss": 5.094830322265625, "grad_norm": 1.0764024257659912, "learning_rate": 0.0001, "elbo": 5.09483, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12813746188130423}
{"step": 2190, "loss": 5.084860229492188, "grad_norm": 1.483386516571045, "learning_rate": 0.0001, "elbo": 5.08486, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12843068261787474}
{"step": 2195, "loss": 5.087973022460938, "grad_norm": 1.2934627532958984, "learning_rate": 0.0001, "elbo": 5.08797, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12872390335444522}
{"step": 2200, "loss": 5.085176086425781, "grad_norm": 1.1440306901931763, "learning_rate": 0.0001, "elbo": 5.08518, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1290171240910157}
{"step": 2205, "loss": 5.066912078857422, "grad_norm": 1.3717726469039917, "learning_rate": 0.0001, "elbo": 5.06691, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12931034482758622}
{"step": 2210, "loss": 5.079823303222656, "grad_norm": 1.4958916902542114, "learning_rate": 0.0001, "elbo": 5.07982, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1296035655641567}
{"step": 2215, "loss": 5.080884552001953, "grad_norm": 1.358647346496582, "learning_rate": 0.0001, "elbo": 5.08088, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.12989678630072718}
{"step": 2220, "loss": 5.073492813110351, "grad_norm": 1.229124665260315, "learning_rate": 0.0001, "elbo": 5.07349, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1301900070372977}
{"step": 2225, "loss": 5.069731140136719, "grad_norm": 1.285810947418213, "learning_rate": 0.0001, "elbo": 5.06973, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13048322777386817}
{"step": 2230, "loss": 5.05253791809082, "grad_norm": 1.1453484296798706, "learning_rate": 0.0001, "elbo": 5.05254, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13077644851043865}
{"step": 2235, "loss": 5.051285171508789, "grad_norm": 1.2648749351501465, "learning_rate": 0.0001, "elbo": 5.05129, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13106966924700914}
{"step": 2240, "loss": 5.064391708374023, "grad_norm": 1.241528034210205, "learning_rate": 0.0001, "elbo": 5.06439, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13136288998357964}
{"step": 2245, "loss": 5.053488159179688, "grad_norm": 1.2159843444824219, "learning_rate": 0.0001, "elbo": 5.05349, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13165611072015013}
{"step": 2250, "loss": 5.049527740478515, "grad_norm": 1.138782024383545, "learning_rate": 0.0001, "elbo": 5.04953, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1319493314567206}
{"step": 2255, "loss": 5.044044494628906, "grad_norm": 1.4827903509140015, "learning_rate": 0.0001, "elbo": 5.04404, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13224255219329112}
{"step": 2260, "loss": 5.0715576171875, "grad_norm": 1.2213575839996338, "learning_rate": 0.0001, "elbo": 5.07156, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1325357729298616}
{"step": 2265, "loss": 5.050323486328125, "grad_norm": 1.4049601554870605, "learning_rate": 0.0001, "elbo": 5.05032, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13282899366643208}
{"step": 2270, "loss": 5.062247848510742, "grad_norm": 1.5380785465240479, "learning_rate": 0.0001, "elbo": 5.06225, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1331222144030026}
{"step": 2275, "loss": 5.046965026855469, "grad_norm": 1.1310198307037354, "learning_rate": 0.0001, "elbo": 5.04697, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13341543513957307}
{"step": 2280, "loss": 5.023358535766602, "grad_norm": 1.2052842378616333, "learning_rate": 0.0001, "elbo": 5.02336, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13370865587614356}
{"step": 2285, "loss": 5.030619812011719, "grad_norm": 1.260602355003357, "learning_rate": 0.0001, "elbo": 5.03062, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13400187661271404}
{"step": 2290, "loss": 5.059983825683593, "grad_norm": 1.1514854431152344, "learning_rate": 0.0001, "elbo": 5.05998, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13429509734928455}
{"step": 2295, "loss": 5.019143295288086, "grad_norm": 1.1415995359420776, "learning_rate": 0.0001, "elbo": 5.01914, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13458831808585503}
{"step": 2300, "loss": 5.035683441162109, "grad_norm": 1.2993639707565308, "learning_rate": 0.0001, "elbo": 5.03568, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1348815388224255}
{"step": 2305, "loss": 5.042318725585938, "grad_norm": 1.2868804931640625, "learning_rate": 0.0001, "elbo": 5.04232, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13517475955899602}
{"step": 2310, "loss": 5.033543395996094, "grad_norm": 1.423563003540039, "learning_rate": 0.0001, "elbo": 5.03354, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1354679802955665}
{"step": 2315, "loss": 5.03515396118164, "grad_norm": 1.0961861610412598, "learning_rate": 0.0001, "elbo": 5.03515, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13576120103213699}
{"step": 2320, "loss": 5.0335548400878904, "grad_norm": 1.3319162130355835, "learning_rate": 0.0001, "elbo": 5.03355, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1360544217687075}
{"step": 2325, "loss": 5.015369415283203, "grad_norm": 1.105364203453064, "learning_rate": 0.0001, "elbo": 5.01537, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13634764250527798}
{"step": 2330, "loss": 5.017719268798828, "grad_norm": 1.1850358247756958, "learning_rate": 0.0001, "elbo": 5.01772, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13664086324184846}
{"step": 2335, "loss": 5.026498794555664, "grad_norm": 1.2118704319000244, "learning_rate": 0.0001, "elbo": 5.0265, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13693408397841894}
{"step": 2340, "loss": 5.008815765380859, "grad_norm": 1.1812556982040405, "learning_rate": 0.0001, "elbo": 5.00882, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13722730471498945}
{"step": 2345, "loss": 5.025453186035156, "grad_norm": 1.2004190683364868, "learning_rate": 0.0001, "elbo": 5.02545, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13752052545155993}
{"step": 2350, "loss": 5.022541809082031, "grad_norm": 1.2529833316802979, "learning_rate": 0.0001, "elbo": 5.02254, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13781374618813041}
{"step": 2355, "loss": 5.012379455566406, "grad_norm": 1.275384545326233, "learning_rate": 0.0001, "elbo": 5.01238, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13810696692470092}
{"step": 2360, "loss": 5.014447784423828, "grad_norm": 1.0568578243255615, "learning_rate": 0.0001, "elbo": 5.01445, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1384001876612714}
{"step": 2365, "loss": 5.002682495117187, "grad_norm": 1.073764443397522, "learning_rate": 0.0001, "elbo": 5.00268, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1386934083978419}
{"step": 2370, "loss": 4.997465515136719, "grad_norm": 1.1284290552139282, "learning_rate": 0.0001, "elbo": 4.99747, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1389866291344124}
{"step": 2375, "loss": 5.0111328125, "grad_norm": 1.3380863666534424, "learning_rate": 0.0001, "elbo": 5.01113, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13927984987098288}
{"step": 2380, "loss": 5.008679962158203, "grad_norm": 1.2506552934646606, "learning_rate": 0.0001, "elbo": 5.00868, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13957307060755336}
{"step": 2385, "loss": 4.999848175048828, "grad_norm": 1.186933159828186, "learning_rate": 0.0001, "elbo": 4.99985, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.13986629134412384}
{"step": 2390, "loss": 5.000840759277343, "grad_norm": 1.2346783876419067, "learning_rate": 0.0001, "elbo": 5.00084, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14015951208069435}
{"step": 2395, "loss": 4.9825904846191404, "grad_norm": 1.0770323276519775, "learning_rate": 0.0001, "elbo": 4.98259, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14045273281726484}
{"step": 2400, "loss": 4.982586669921875, "grad_norm": 1.1978256702423096, "learning_rate": 0.0001, "elbo": 4.98259, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14074595355383532}
{"step": 2405, "loss": 4.984527587890625, "grad_norm": 1.1744637489318848, "learning_rate": 0.0001, "elbo": 4.98453, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14103917429040583}
{"step": 2410, "loss": 4.988859558105469, "grad_norm": 1.2328593730926514, "learning_rate": 0.0001, "elbo": 4.98886, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1413323950269763}
{"step": 2415, "loss": 4.971556854248047, "grad_norm": 1.0847572088241577, "learning_rate": 0.0001, "elbo": 4.97156, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1416256157635468}
{"step": 2420, "loss": 4.975756072998047, "grad_norm": 1.2964564561843872, "learning_rate": 0.0001, "elbo": 4.97576, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1419188365001173}
{"step": 2425, "loss": 4.967469787597656, "grad_norm": 1.3052468299865723, "learning_rate": 0.0001, "elbo": 4.96747, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14221205723668778}
{"step": 2430, "loss": 4.984537124633789, "grad_norm": 1.0621610879898071, "learning_rate": 0.0001, "elbo": 4.98454, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14250527797325827}
{"step": 2435, "loss": 4.981452178955078, "grad_norm": 1.15813148021698, "learning_rate": 0.0001, "elbo": 4.98145, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14279849870982875}
{"step": 2440, "loss": 4.969306945800781, "grad_norm": 1.676693081855774, "learning_rate": 0.0001, "elbo": 4.96931, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14309171944639926}
{"step": 2445, "loss": 4.968242645263672, "grad_norm": 1.247310996055603, "learning_rate": 0.0001, "elbo": 4.96824, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14338494018296974}
{"step": 2450, "loss": 4.990269470214844, "grad_norm": 1.2080484628677368, "learning_rate": 0.0001, "elbo": 4.99027, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14367816091954022}
{"step": 2455, "loss": 4.970929718017578, "grad_norm": 1.0419373512268066, "learning_rate": 0.0001, "elbo": 4.97093, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14397138165611073}
{"step": 2460, "loss": 4.969306945800781, "grad_norm": 1.3381948471069336, "learning_rate": 0.0001, "elbo": 4.96931, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1442646023926812}
{"step": 2465, "loss": 4.967673873901367, "grad_norm": 1.1503807306289673, "learning_rate": 0.0001, "elbo": 4.96767, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1445578231292517}
{"step": 2470, "loss": 4.968066024780273, "grad_norm": 1.2689331769943237, "learning_rate": 0.0001, "elbo": 4.96807, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1448510438658222}
{"step": 2475, "loss": 4.975127029418945, "grad_norm": 1.4164390563964844, "learning_rate": 0.0001, "elbo": 4.97513, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1451442646023927}
{"step": 2480, "loss": 4.965142440795899, "grad_norm": 1.0901768207550049, "learning_rate": 0.0001, "elbo": 4.96514, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14543748533896317}
{"step": 2485, "loss": 4.954307556152344, "grad_norm": 1.209619402885437, "learning_rate": 0.0001, "elbo": 4.95431, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14573070607553365}
{"step": 2490, "loss": 4.951070404052734, "grad_norm": 1.2348352670669556, "learning_rate": 0.0001, "elbo": 4.95107, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14602392681210416}
{"step": 2495, "loss": 4.9522651672363285, "grad_norm": 0.9345862865447998, "learning_rate": 0.0001, "elbo": 4.95226, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14631714754867464}
{"step": 2500, "loss": 4.955776977539062, "grad_norm": 1.2732627391815186, "learning_rate": 0.0001, "elbo": 4.95578, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14661036828524512}
{"step": 2500, "loss": 4.955776977539062, "grad_norm": 1.2732627391815186, "learning_rate": 0.0001, "elbo": 4.95578, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14661036828524512}
{"step": 2505, "loss": 4.946250152587891, "grad_norm": 1.107619047164917, "learning_rate": 0.0001, "elbo": 4.94625, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14690358902181563}
{"step": 2510, "loss": 4.950692749023437, "grad_norm": 1.2930948734283447, "learning_rate": 0.0001, "elbo": 4.95069, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14719680975838612}
{"step": 2515, "loss": 4.956946945190429, "grad_norm": 0.9993621110916138, "learning_rate": 0.0001, "elbo": 4.95695, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1474900304949566}
{"step": 2520, "loss": 4.95125732421875, "grad_norm": 1.0418599843978882, "learning_rate": 0.0001, "elbo": 4.95126, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1477832512315271}
{"step": 2525, "loss": 4.949594497680664, "grad_norm": 1.1393146514892578, "learning_rate": 0.0001, "elbo": 4.94959, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1480764719680976}
{"step": 2530, "loss": 4.950519943237305, "grad_norm": 1.1257656812667847, "learning_rate": 0.0001, "elbo": 4.95052, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14836969270466807}
{"step": 2535, "loss": 4.950689697265625, "grad_norm": 1.1070815324783325, "learning_rate": 0.0001, "elbo": 4.95069, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14866291344123855}
{"step": 2540, "loss": 4.936194610595703, "grad_norm": 1.3556898832321167, "learning_rate": 0.0001, "elbo": 4.93619, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14895613417780906}
{"step": 2545, "loss": 4.957874298095703, "grad_norm": 1.1888344287872314, "learning_rate": 0.0001, "elbo": 4.95787, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14924935491437955}
{"step": 2550, "loss": 4.933316040039062, "grad_norm": 1.1684889793395996, "learning_rate": 0.0001, "elbo": 4.93332, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14954257565095003}
{"step": 2555, "loss": 4.930287170410156, "grad_norm": 1.3958206176757812, "learning_rate": 0.0001, "elbo": 4.93029, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.14983579638752054}
{"step": 2560, "loss": 4.947838592529297, "grad_norm": 1.3189572095870972, "learning_rate": 0.0001, "elbo": 4.94784, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15012901712409102}
{"step": 2565, "loss": 4.924427032470703, "grad_norm": 1.2659852504730225, "learning_rate": 0.0001, "elbo": 4.92443, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1504222378606615}
{"step": 2570, "loss": 4.915815734863282, "grad_norm": 1.0379664897918701, "learning_rate": 0.0001, "elbo": 4.91582, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15071545859723198}
{"step": 2575, "loss": 4.922835922241211, "grad_norm": 1.276001214981079, "learning_rate": 0.0001, "elbo": 4.92284, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1510086793338025}
{"step": 2580, "loss": 4.939435958862305, "grad_norm": 1.2578487396240234, "learning_rate": 0.0001, "elbo": 4.93944, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15130190007037297}
{"step": 2585, "loss": 4.933565139770508, "grad_norm": 1.1793807744979858, "learning_rate": 0.0001, "elbo": 4.93357, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15159512080694346}
{"step": 2590, "loss": 4.910115051269531, "grad_norm": 1.0781517028808594, "learning_rate": 0.0001, "elbo": 4.91012, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15188834154351397}
{"step": 2595, "loss": 4.91551628112793, "grad_norm": 1.3800164461135864, "learning_rate": 0.0001, "elbo": 4.91552, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15218156228008445}
{"step": 2600, "loss": 4.910502624511719, "grad_norm": 1.094908595085144, "learning_rate": 0.0001, "elbo": 4.9105, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15247478301665493}
{"step": 2605, "loss": 4.923577499389649, "grad_norm": 1.1287739276885986, "learning_rate": 0.0001, "elbo": 4.92358, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15276800375322544}
{"step": 2610, "loss": 4.915113067626953, "grad_norm": 1.2297152280807495, "learning_rate": 0.0001, "elbo": 4.91511, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15306122448979592}
{"step": 2615, "loss": 4.916178894042969, "grad_norm": 1.2681185007095337, "learning_rate": 0.0001, "elbo": 4.91618, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1533544452263664}
{"step": 2620, "loss": 4.911692047119141, "grad_norm": 1.2444602251052856, "learning_rate": 0.0001, "elbo": 4.91169, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15364766596293689}
{"step": 2625, "loss": 4.907736968994141, "grad_norm": 1.1832340955734253, "learning_rate": 0.0001, "elbo": 4.90774, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1539408866995074}
{"step": 2630, "loss": 4.907033920288086, "grad_norm": 1.1341485977172852, "learning_rate": 0.0001, "elbo": 4.90703, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15423410743607788}
{"step": 2635, "loss": 4.900775146484375, "grad_norm": 1.0724241733551025, "learning_rate": 0.0001, "elbo": 4.90078, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15452732817264836}
{"step": 2640, "loss": 4.906300354003906, "grad_norm": 1.1955527067184448, "learning_rate": 0.0001, "elbo": 4.9063, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15482054890921887}
{"step": 2645, "loss": 4.908354949951172, "grad_norm": 1.2135568857192993, "learning_rate": 0.0001, "elbo": 4.90835, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15511376964578935}
{"step": 2650, "loss": 4.897692489624023, "grad_norm": 1.1266285181045532, "learning_rate": 0.0001, "elbo": 4.89769, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15540699038235983}
{"step": 2655, "loss": 4.897811508178711, "grad_norm": 1.5971287488937378, "learning_rate": 0.0001, "elbo": 4.89781, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15570021111893034}
{"step": 2660, "loss": 4.908198547363281, "grad_norm": 1.0924921035766602, "learning_rate": 0.0001, "elbo": 4.9082, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15599343185550082}
{"step": 2665, "loss": 4.901294708251953, "grad_norm": 1.2126833200454712, "learning_rate": 0.0001, "elbo": 4.90129, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1562866525920713}
{"step": 2670, "loss": 4.893850708007813, "grad_norm": 1.0670262575149536, "learning_rate": 0.0001, "elbo": 4.89385, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1565798733286418}
{"step": 2675, "loss": 4.8958290100097654, "grad_norm": 1.173545002937317, "learning_rate": 0.0001, "elbo": 4.89583, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1568730940652123}
{"step": 2680, "loss": 4.896595764160156, "grad_norm": 1.3569122552871704, "learning_rate": 0.0001, "elbo": 4.8966, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15716631480178278}
{"step": 2685, "loss": 4.897843551635742, "grad_norm": 1.155124545097351, "learning_rate": 0.0001, "elbo": 4.89784, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15745953553835326}
{"step": 2690, "loss": 4.881786346435547, "grad_norm": 1.028867483139038, "learning_rate": 0.0001, "elbo": 4.88179, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15775275627492377}
{"step": 2695, "loss": 4.879304504394531, "grad_norm": 1.1822794675827026, "learning_rate": 0.0001, "elbo": 4.8793, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15804597701149425}
{"step": 2700, "loss": 4.879425430297852, "grad_norm": 1.0641969442367554, "learning_rate": 0.0001, "elbo": 4.87943, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15833919774806474}
{"step": 2705, "loss": 4.865487289428711, "grad_norm": 1.1978462934494019, "learning_rate": 0.0001, "elbo": 4.86549, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15863241848463525}
{"step": 2710, "loss": 4.898887634277344, "grad_norm": 1.0020930767059326, "learning_rate": 0.0001, "elbo": 4.89889, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.15892563922120573}
{"step": 2715, "loss": 4.886761474609375, "grad_norm": 1.118491768836975, "learning_rate": 0.0001, "elbo": 4.88676, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1592188599577762}
{"step": 2720, "loss": 4.872737884521484, "grad_norm": 1.3399560451507568, "learning_rate": 0.0001, "elbo": 4.87274, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1595120806943467}
{"step": 2725, "loss": 4.868120956420898, "grad_norm": 1.158599853515625, "learning_rate": 0.0001, "elbo": 4.86812, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1598053014309172}
{"step": 2730, "loss": 4.884989929199219, "grad_norm": 1.2657432556152344, "learning_rate": 0.0001, "elbo": 4.88499, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16009852216748768}
{"step": 2735, "loss": 4.870775604248047, "grad_norm": 1.0840551853179932, "learning_rate": 0.0001, "elbo": 4.87078, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16039174290405817}
{"step": 2740, "loss": 4.855854034423828, "grad_norm": 1.0166219472885132, "learning_rate": 0.0001, "elbo": 4.85585, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16068496364062868}
{"step": 2745, "loss": 4.878772735595703, "grad_norm": 1.2082725763320923, "learning_rate": 0.0001, "elbo": 4.87877, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16097818437719916}
{"step": 2750, "loss": 4.886483764648437, "grad_norm": 1.4823936223983765, "learning_rate": 0.0001, "elbo": 4.88648, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16127140511376964}
{"step": 2755, "loss": 4.864241027832032, "grad_norm": 1.1302073001861572, "learning_rate": 0.0001, "elbo": 4.86424, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16156462585034015}
{"step": 2760, "loss": 4.868212890625, "grad_norm": 1.2545490264892578, "learning_rate": 0.0001, "elbo": 4.86821, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16185784658691063}
{"step": 2765, "loss": 4.860008239746094, "grad_norm": 1.31172776222229, "learning_rate": 0.0001, "elbo": 4.86001, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1621510673234811}
{"step": 2770, "loss": 4.878634643554688, "grad_norm": 1.0809155702590942, "learning_rate": 0.0001, "elbo": 4.87863, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1624442880600516}
{"step": 2775, "loss": 4.86565933227539, "grad_norm": 0.946161687374115, "learning_rate": 0.0001, "elbo": 4.86566, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1627375087966221}
{"step": 2780, "loss": 4.850774765014648, "grad_norm": 1.0365064144134521, "learning_rate": 0.0001, "elbo": 4.85077, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1630307295331926}
{"step": 2785, "loss": 4.85241470336914, "grad_norm": 1.0710768699645996, "learning_rate": 0.0001, "elbo": 4.85241, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16332395026976307}
{"step": 2785, "loss": 4.85241470336914, "grad_norm": 1.0710768699645996, "learning_rate": 0.0001, "elbo": 4.85241, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16332395026976307}
{"step": 2790, "loss": 4.875817108154297, "grad_norm": 1.308167815208435, "learning_rate": 0.0001, "elbo": 4.87582, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16361717100633358}
{"step": 2795, "loss": 4.856101226806641, "grad_norm": 1.2713323831558228, "learning_rate": 0.0001, "elbo": 4.8561, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16391039174290406}
{"step": 2800, "loss": 4.845724105834961, "grad_norm": 1.3220491409301758, "learning_rate": 0.0001, "elbo": 4.84572, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16420361247947454}
{"step": 2805, "loss": 4.855319976806641, "grad_norm": 1.0731185674667358, "learning_rate": 0.0001, "elbo": 4.85532, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16449683321604505}
{"step": 2810, "loss": 4.880852127075196, "grad_norm": 1.341562271118164, "learning_rate": 0.0001, "elbo": 4.88085, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16479005395261553}
{"step": 2815, "loss": 4.8388313293457035, "grad_norm": 1.0174893140792847, "learning_rate": 0.0001, "elbo": 4.83883, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16508327468918602}
{"step": 2820, "loss": 4.858737182617188, "grad_norm": 1.2198082208633423, "learning_rate": 0.0001, "elbo": 4.85874, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1653764954257565}
{"step": 2825, "loss": 4.846211242675781, "grad_norm": 1.0729680061340332, "learning_rate": 0.0001, "elbo": 4.84621, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.165669716162327}
{"step": 2830, "loss": 4.8429206848144535, "grad_norm": 1.1401150226593018, "learning_rate": 0.0001, "elbo": 4.84292, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1659629368988975}
{"step": 2835, "loss": 4.834847259521484, "grad_norm": 1.1297109127044678, "learning_rate": 0.0001, "elbo": 4.83485, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16625615763546797}
{"step": 2840, "loss": 4.840232849121094, "grad_norm": 1.0992070436477661, "learning_rate": 0.0001, "elbo": 4.84023, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16654937837203848}
{"step": 2845, "loss": 4.846598815917969, "grad_norm": 1.1670790910720825, "learning_rate": 0.0001, "elbo": 4.8466, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16684259910860896}
{"step": 2850, "loss": 4.8339988708496096, "grad_norm": 1.385730504989624, "learning_rate": 0.0001, "elbo": 4.834, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16713581984517945}
{"step": 2855, "loss": 4.836075973510742, "grad_norm": 0.9765023589134216, "learning_rate": 0.0001, "elbo": 4.83608, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16742904058174995}
{"step": 2860, "loss": 4.829730987548828, "grad_norm": 1.1566113233566284, "learning_rate": 0.0001, "elbo": 4.82973, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16772226131832044}
{"step": 2865, "loss": 4.83391227722168, "grad_norm": 1.5146013498306274, "learning_rate": 0.0001, "elbo": 4.83391, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16801548205489092}
{"step": 2870, "loss": 4.8457588195800785, "grad_norm": 1.0719228982925415, "learning_rate": 0.0001, "elbo": 4.84576, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1683087027914614}
{"step": 2875, "loss": 4.830405426025391, "grad_norm": 1.0268259048461914, "learning_rate": 0.0001, "elbo": 4.83041, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1686019235280319}
{"step": 2880, "loss": 4.840502548217773, "grad_norm": 1.0060865879058838, "learning_rate": 0.0001, "elbo": 4.8405, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1688951442646024}
{"step": 2885, "loss": 4.818592071533203, "grad_norm": 1.4532195329666138, "learning_rate": 0.0001, "elbo": 4.81859, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16918836500117287}
{"step": 2885, "loss": 4.818592071533203, "grad_norm": 1.4532195329666138, "learning_rate": 0.0001, "elbo": 4.81859, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16918836500117287}
{"step": 2890, "loss": 4.829833984375, "grad_norm": 1.1657230854034424, "learning_rate": 0.0001, "elbo": 4.82983, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16948158573774338}
{"step": 2895, "loss": 4.82995376586914, "grad_norm": 0.9914016127586365, "learning_rate": 0.0001, "elbo": 4.82995, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.16977480647431387}
{"step": 2900, "loss": 4.822637939453125, "grad_norm": 1.1337940692901611, "learning_rate": 0.0001, "elbo": 4.82264, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17006802721088435}
{"step": 2900, "loss": 4.822637939453125, "grad_norm": 1.1337940692901611, "learning_rate": 0.0001, "elbo": 4.82264, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17006802721088435}
{"step": 2905, "loss": 4.821896362304687, "grad_norm": 1.0365145206451416, "learning_rate": 0.0001, "elbo": 4.8219, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17036124794745483}
{"step": 2910, "loss": 4.808308410644531, "grad_norm": 1.0400265455245972, "learning_rate": 0.0001, "elbo": 4.80831, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17065446868402534}
{"step": 2915, "loss": 4.837464904785156, "grad_norm": 1.0283048152923584, "learning_rate": 0.0001, "elbo": 4.83746, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17094768942059582}
{"step": 2920, "loss": 4.811058807373047, "grad_norm": 1.0408469438552856, "learning_rate": 0.0001, "elbo": 4.81106, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1712409101571663}
{"step": 2925, "loss": 4.830168914794922, "grad_norm": 1.0444285869598389, "learning_rate": 0.0001, "elbo": 4.83017, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1715341308937368}
{"step": 2930, "loss": 4.831594085693359, "grad_norm": 1.2135305404663086, "learning_rate": 0.0001, "elbo": 4.83159, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1718273516303073}
{"step": 2935, "loss": 4.820682525634766, "grad_norm": 1.1563098430633545, "learning_rate": 0.0001, "elbo": 4.82068, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17212057236687778}
{"step": 2940, "loss": 4.820388412475586, "grad_norm": 1.19757080078125, "learning_rate": 0.0001, "elbo": 4.82039, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1724137931034483}
{"step": 2945, "loss": 4.80458984375, "grad_norm": 1.1196624040603638, "learning_rate": 0.0001, "elbo": 4.80459, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17270701384001877}
{"step": 2950, "loss": 4.807529449462891, "grad_norm": 1.115408182144165, "learning_rate": 0.0001, "elbo": 4.80753, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17300023457658925}
{"step": 2955, "loss": 4.806457138061523, "grad_norm": 1.2555952072143555, "learning_rate": 0.0001, "elbo": 4.80646, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17329345531315973}
{"step": 2960, "loss": 4.801681137084961, "grad_norm": 1.1700494289398193, "learning_rate": 0.0001, "elbo": 4.80168, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17358667604973024}
{"step": 2965, "loss": 4.802988433837891, "grad_norm": 1.070810079574585, "learning_rate": 0.0001, "elbo": 4.80299, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17387989678630072}
{"step": 2970, "loss": 4.820628356933594, "grad_norm": 1.0713202953338623, "learning_rate": 0.0001, "elbo": 4.82063, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1741731175228712}
{"step": 2975, "loss": 4.789900588989258, "grad_norm": 1.1148804426193237, "learning_rate": 0.0001, "elbo": 4.7899, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17446633825944172}
{"step": 2980, "loss": 4.8062389373779295, "grad_norm": 1.0810902118682861, "learning_rate": 0.0001, "elbo": 4.80624, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1747595589960122}
{"step": 2985, "loss": 4.776848983764649, "grad_norm": 1.1833219528198242, "learning_rate": 0.0001, "elbo": 4.77685, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17505277973258268}
{"step": 2990, "loss": 4.795020294189453, "grad_norm": 1.098770260810852, "learning_rate": 0.0001, "elbo": 4.79502, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1753460004691532}
{"step": 2995, "loss": 4.796526336669922, "grad_norm": 1.2392569780349731, "learning_rate": 0.0001, "elbo": 4.79653, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17563922120572367}
{"step": 3000, "loss": 4.799547576904297, "grad_norm": 1.0804343223571777, "learning_rate": 0.0001, "elbo": 4.79955, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17593244194229415}
{"step": 3005, "loss": 4.7776641845703125, "grad_norm": 1.866988182067871, "learning_rate": 0.0001, "elbo": 4.77766, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17622566267886464}
{"step": 3010, "loss": 4.793832397460937, "grad_norm": 0.9084110260009766, "learning_rate": 0.0001, "elbo": 4.79383, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17651888341543515}
{"step": 3015, "loss": 4.800395965576172, "grad_norm": 1.2740622758865356, "learning_rate": 0.0001, "elbo": 4.8004, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17681210415200563}
{"step": 3020, "loss": 4.783384704589844, "grad_norm": 1.2191276550292969, "learning_rate": 0.0001, "elbo": 4.78338, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1771053248885761}
{"step": 3020, "loss": 4.783384704589844, "grad_norm": 1.2191276550292969, "learning_rate": 0.0001, "elbo": 4.78338, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1771053248885761}
{"step": 3025, "loss": 4.787153244018555, "grad_norm": 1.2654963731765747, "learning_rate": 0.0001, "elbo": 4.78715, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17739854562514662}
{"step": 3030, "loss": 4.780703735351563, "grad_norm": 0.998590350151062, "learning_rate": 0.0001, "elbo": 4.7807, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1776917663617171}
{"step": 3035, "loss": 4.816294097900391, "grad_norm": 1.2437981367111206, "learning_rate": 0.0001, "elbo": 4.81629, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17798498709828758}
{"step": 3035, "loss": 4.816294097900391, "grad_norm": 1.2437981367111206, "learning_rate": 0.0001, "elbo": 4.81629, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17798498709828758}
{"step": 3040, "loss": 4.789542007446289, "grad_norm": 1.0491021871566772, "learning_rate": 0.0001, "elbo": 4.78954, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1782782078348581}
{"step": 3045, "loss": 4.774539947509766, "grad_norm": 1.3455857038497925, "learning_rate": 0.0001, "elbo": 4.77454, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17857142857142858}
{"step": 3050, "loss": 4.80179443359375, "grad_norm": 1.1477673053741455, "learning_rate": 0.0001, "elbo": 4.80179, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17886464930799906}
{"step": 3050, "loss": 4.80179443359375, "grad_norm": 1.1477673053741455, "learning_rate": 0.0001, "elbo": 4.80179, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17886464930799906}
{"step": 3055, "loss": 4.784046936035156, "grad_norm": 1.2272746562957764, "learning_rate": 0.0001, "elbo": 4.78405, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17915787004456954}
{"step": 3060, "loss": 4.775359344482422, "grad_norm": 1.0693153142929077, "learning_rate": 0.0001, "elbo": 4.77536, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17945109078114005}
{"step": 3060, "loss": 4.775359344482422, "grad_norm": 1.0693153142929077, "learning_rate": 0.0001, "elbo": 4.77536, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17945109078114005}
{"step": 3065, "loss": 4.785107040405274, "grad_norm": 1.1694680452346802, "learning_rate": 0.0001, "elbo": 4.78511, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.17974431151771053}
{"step": 3070, "loss": 4.791721343994141, "grad_norm": 1.1039454936981201, "learning_rate": 0.0001, "elbo": 4.79172, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.180037532254281}
{"step": 3075, "loss": 4.785713195800781, "grad_norm": 1.123449444770813, "learning_rate": 0.0001, "elbo": 4.78571, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18033075299085152}
{"step": 3080, "loss": 4.780200958251953, "grad_norm": 1.0112824440002441, "learning_rate": 0.0001, "elbo": 4.7802, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.180623973727422}
{"step": 3085, "loss": 4.776071166992187, "grad_norm": 0.9709073305130005, "learning_rate": 0.0001, "elbo": 4.77607, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1809171944639925}
{"step": 3090, "loss": 4.758792877197266, "grad_norm": 1.3179305791854858, "learning_rate": 0.0001, "elbo": 4.75879, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.181210415200563}
{"step": 3095, "loss": 4.763344573974609, "grad_norm": 0.9550114274024963, "learning_rate": 0.0001, "elbo": 4.76334, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18150363593713348}
{"step": 3100, "loss": 4.7711639404296875, "grad_norm": 1.0035005807876587, "learning_rate": 0.0001, "elbo": 4.77116, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18179685667370396}
{"step": 3100, "loss": 4.7711639404296875, "grad_norm": 1.0035005807876587, "learning_rate": 0.0001, "elbo": 4.77116, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18179685667370396}
{"step": 3105, "loss": 4.757177734375, "grad_norm": 0.9746994972229004, "learning_rate": 0.0001, "elbo": 4.75718, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18209007741027444}
{"step": 3110, "loss": 4.779315185546875, "grad_norm": 1.0707992315292358, "learning_rate": 0.0001, "elbo": 4.77932, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18238329814684495}
{"step": 3115, "loss": 4.7734027862548825, "grad_norm": 0.9664779901504517, "learning_rate": 0.0001, "elbo": 4.7734, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18267651888341543}
{"step": 3120, "loss": 4.750485610961914, "grad_norm": 1.0378581285476685, "learning_rate": 0.0001, "elbo": 4.75049, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18296973961998592}
{"step": 3125, "loss": 4.7631996154785154, "grad_norm": 1.038610816001892, "learning_rate": 0.0001, "elbo": 4.7632, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18326296035655643}
{"step": 3130, "loss": 4.770164871215821, "grad_norm": 1.0449837446212769, "learning_rate": 0.0001, "elbo": 4.77017, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1835561810931269}
{"step": 3135, "loss": 4.771099853515625, "grad_norm": 1.0861661434173584, "learning_rate": 0.0001, "elbo": 4.7711, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1838494018296974}
{"step": 3140, "loss": 4.768097686767578, "grad_norm": 1.0772497653961182, "learning_rate": 0.0001, "elbo": 4.7681, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1841426225662679}
{"step": 3145, "loss": 4.757160949707031, "grad_norm": 1.28603196144104, "learning_rate": 0.0001, "elbo": 4.75716, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18443584330283838}
{"step": 3150, "loss": 4.753977203369141, "grad_norm": 1.004940152168274, "learning_rate": 0.0001, "elbo": 4.75398, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18472906403940886}
{"step": 3155, "loss": 4.755023193359375, "grad_norm": 1.3889011144638062, "learning_rate": 0.0001, "elbo": 4.75502, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18502228477597935}
{"step": 3160, "loss": 4.738777160644531, "grad_norm": 1.0956299304962158, "learning_rate": 0.0001, "elbo": 4.73878, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18531550551254986}
{"step": 3165, "loss": 4.751914978027344, "grad_norm": 0.9200780391693115, "learning_rate": 0.0001, "elbo": 4.75191, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18560872624912034}
{"step": 3170, "loss": 4.760741806030273, "grad_norm": 1.5215798616409302, "learning_rate": 0.0001, "elbo": 4.76074, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18590194698569082}
{"step": 3175, "loss": 4.75057373046875, "grad_norm": 1.204877257347107, "learning_rate": 0.0001, "elbo": 4.75057, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18619516772226133}
{"step": 3180, "loss": 4.749491500854492, "grad_norm": 0.9833300709724426, "learning_rate": 0.0001, "elbo": 4.74949, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1864883884588318}
{"step": 3185, "loss": 4.748503112792969, "grad_norm": 1.1588292121887207, "learning_rate": 0.0001, "elbo": 4.7485, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1867816091954023}
{"step": 3190, "loss": 4.75704345703125, "grad_norm": 1.1344873905181885, "learning_rate": 0.0001, "elbo": 4.75704, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1870748299319728}
{"step": 3195, "loss": 4.745890426635742, "grad_norm": 1.1307369470596313, "learning_rate": 0.0001, "elbo": 4.74589, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18736805066854328}
{"step": 3200, "loss": 4.747586822509765, "grad_norm": 1.0496577024459839, "learning_rate": 0.0001, "elbo": 4.74759, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18766127140511377}
{"step": 3205, "loss": 4.750003051757813, "grad_norm": 1.0320700407028198, "learning_rate": 0.0001, "elbo": 4.75, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18795449214168425}
{"step": 3210, "loss": 4.7387840270996096, "grad_norm": 1.1047991514205933, "learning_rate": 0.0001, "elbo": 4.73878, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18824771287825476}
{"step": 3215, "loss": 4.7359062194824215, "grad_norm": 1.0854582786560059, "learning_rate": 0.0001, "elbo": 4.73591, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18854093361482524}
{"step": 3220, "loss": 4.742343139648438, "grad_norm": 1.1006108522415161, "learning_rate": 0.0001, "elbo": 4.74234, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18883415435139572}
{"step": 3225, "loss": 4.730160140991211, "grad_norm": 1.107024073600769, "learning_rate": 0.0001, "elbo": 4.73016, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.18912737508796623}
{"step": 3230, "loss": 4.744716644287109, "grad_norm": 0.9255626201629639, "learning_rate": 0.0001, "elbo": 4.74472, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1894205958245367}
{"step": 3235, "loss": 4.729971313476563, "grad_norm": 1.0751652717590332, "learning_rate": 0.0001, "elbo": 4.72997, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1897138165611072}
{"step": 3240, "loss": 4.737179565429687, "grad_norm": 1.0743999481201172, "learning_rate": 0.0001, "elbo": 4.73718, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1900070372976777}
{"step": 3245, "loss": 4.725858688354492, "grad_norm": 0.9552168250083923, "learning_rate": 0.0001, "elbo": 4.72586, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1903002580342482}
{"step": 3250, "loss": 4.74178466796875, "grad_norm": 0.9786060452461243, "learning_rate": 0.0001, "elbo": 4.74178, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19059347877081867}
{"step": 3255, "loss": 4.7310234069824215, "grad_norm": 1.3312194347381592, "learning_rate": 0.0001, "elbo": 4.73102, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19088669950738915}
{"step": 3260, "loss": 4.752585601806641, "grad_norm": 1.3735796213150024, "learning_rate": 0.0001, "elbo": 4.75259, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19117992024395966}
{"step": 3265, "loss": 4.72659912109375, "grad_norm": 1.3246800899505615, "learning_rate": 0.0001, "elbo": 4.7266, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19147314098053014}
{"step": 3270, "loss": 4.727349090576172, "grad_norm": 1.0253764390945435, "learning_rate": 0.0001, "elbo": 4.72735, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19176636171710063}
{"step": 3275, "loss": 4.718534851074219, "grad_norm": 1.023289442062378, "learning_rate": 0.0001, "elbo": 4.71853, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19205958245367113}
{"step": 3280, "loss": 4.725822448730469, "grad_norm": 1.185764193534851, "learning_rate": 0.0001, "elbo": 4.72582, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19235280319024162}
{"step": 3285, "loss": 4.712643432617187, "grad_norm": 1.1233512163162231, "learning_rate": 0.0001, "elbo": 4.71264, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1926460239268121}
{"step": 3290, "loss": 4.701814270019531, "grad_norm": 1.254276156425476, "learning_rate": 0.0001, "elbo": 4.70181, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19293924466338258}
{"step": 3295, "loss": 4.725931167602539, "grad_norm": 0.9805157780647278, "learning_rate": 0.0001, "elbo": 4.72593, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1932324653999531}
{"step": 3300, "loss": 4.711648559570312, "grad_norm": 1.2625809907913208, "learning_rate": 0.0001, "elbo": 4.71165, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19352568613652357}
{"step": 3305, "loss": 4.721068954467773, "grad_norm": 0.9605720043182373, "learning_rate": 0.0001, "elbo": 4.72107, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19381890687309405}
{"step": 3310, "loss": 4.719356536865234, "grad_norm": 1.2771544456481934, "learning_rate": 0.0001, "elbo": 4.71936, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19411212760966456}
{"step": 3315, "loss": 4.729865264892578, "grad_norm": 1.1424880027770996, "learning_rate": 0.0001, "elbo": 4.72987, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19440534834623505}
{"step": 3320, "loss": 4.7182365417480465, "grad_norm": 0.9566795229911804, "learning_rate": 0.0001, "elbo": 4.71824, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19469856908280553}
{"step": 3325, "loss": 4.705816268920898, "grad_norm": 1.0288524627685547, "learning_rate": 0.0001, "elbo": 4.70582, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19499178981937604}
{"step": 3330, "loss": 4.7113700866699215, "grad_norm": 1.0582540035247803, "learning_rate": 0.0001, "elbo": 4.71137, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19528501055594652}
{"step": 3335, "loss": 4.725521469116211, "grad_norm": 0.970737099647522, "learning_rate": 0.0001, "elbo": 4.72552, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.195578231292517}
{"step": 3340, "loss": 4.712046432495117, "grad_norm": 1.0574036836624146, "learning_rate": 0.0001, "elbo": 4.71205, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19587145202908748}
{"step": 3345, "loss": 4.709675598144531, "grad_norm": 1.1173391342163086, "learning_rate": 0.0001, "elbo": 4.70968, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.196164672765658}
{"step": 3350, "loss": 4.700960922241211, "grad_norm": 1.1921111345291138, "learning_rate": 0.0001, "elbo": 4.70096, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19645789350222848}
{"step": 3355, "loss": 4.708597564697266, "grad_norm": 1.0306634902954102, "learning_rate": 0.0001, "elbo": 4.7086, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19675111423879896}
{"step": 3360, "loss": 4.687208557128907, "grad_norm": 0.9286825060844421, "learning_rate": 0.0001, "elbo": 4.68721, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19704433497536947}
{"step": 3365, "loss": 4.719990921020508, "grad_norm": 1.0224913358688354, "learning_rate": 0.0001, "elbo": 4.71999, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19733755571193995}
{"step": 3370, "loss": 4.706217956542969, "grad_norm": 0.9699579477310181, "learning_rate": 0.0001, "elbo": 4.70622, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19763077644851043}
{"step": 3375, "loss": 4.700669860839843, "grad_norm": 0.9877455830574036, "learning_rate": 0.0001, "elbo": 4.70067, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19792399718508094}
{"step": 3380, "loss": 4.706613540649414, "grad_norm": 1.165692687034607, "learning_rate": 0.0001, "elbo": 4.70661, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19821721792165142}
{"step": 3385, "loss": 4.702599716186524, "grad_norm": 1.0300146341323853, "learning_rate": 0.0001, "elbo": 4.7026, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1985104386582219}
{"step": 3390, "loss": 4.691514587402343, "grad_norm": 0.9872795939445496, "learning_rate": 0.0001, "elbo": 4.69151, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1988036593947924}
{"step": 3395, "loss": 4.710415649414062, "grad_norm": 0.9709939360618591, "learning_rate": 0.0001, "elbo": 4.71042, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.1990968801313629}
{"step": 3400, "loss": 4.686265182495117, "grad_norm": 0.9013175368309021, "learning_rate": 0.0001, "elbo": 4.68627, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19939010086793338}
{"step": 3405, "loss": 4.700743103027344, "grad_norm": 0.9269633889198303, "learning_rate": 0.0001, "elbo": 4.70074, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19968332160450386}
{"step": 3410, "loss": 4.697847747802735, "grad_norm": 1.1789731979370117, "learning_rate": 0.0001, "elbo": 4.69785, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.19997654234107437}
{"step": 3415, "loss": 4.692118072509766, "grad_norm": 1.2296066284179688, "learning_rate": 0.0001, "elbo": 4.69212, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20026976307764485}
{"step": 3420, "loss": 4.705442428588867, "grad_norm": 0.9072972536087036, "learning_rate": 0.0001, "elbo": 4.70544, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20056298381421533}
{"step": 3425, "loss": 4.691160583496094, "grad_norm": 1.0695031881332397, "learning_rate": 0.0001, "elbo": 4.69116, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20085620455078584}
{"step": 3430, "loss": 4.689171981811524, "grad_norm": 0.9816654324531555, "learning_rate": 0.0001, "elbo": 4.68917, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20114942528735633}
{"step": 3435, "loss": 4.678354644775391, "grad_norm": 1.1135928630828857, "learning_rate": 0.0001, "elbo": 4.67835, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2014426460239268}
{"step": 3440, "loss": 4.686548233032227, "grad_norm": 1.1837340593338013, "learning_rate": 0.0001, "elbo": 4.68655, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2017358667604973}
{"step": 3445, "loss": 4.6941680908203125, "grad_norm": 1.1597728729248047, "learning_rate": 0.0001, "elbo": 4.69417, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2020290874970678}
{"step": 3450, "loss": 4.679721832275391, "grad_norm": 1.3434422016143799, "learning_rate": 0.0001, "elbo": 4.67972, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20232230823363828}
{"step": 3455, "loss": 4.701976013183594, "grad_norm": 1.0241369009017944, "learning_rate": 0.0001, "elbo": 4.70198, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20261552897020876}
{"step": 3460, "loss": 4.701605224609375, "grad_norm": 0.9547503590583801, "learning_rate": 0.0001, "elbo": 4.70161, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20290874970677927}
{"step": 3465, "loss": 4.698007202148437, "grad_norm": 1.2440705299377441, "learning_rate": 0.0001, "elbo": 4.69801, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20320197044334976}
{"step": 3470, "loss": 4.687039184570312, "grad_norm": 1.2498340606689453, "learning_rate": 0.0001, "elbo": 4.68704, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20349519117992024}
{"step": 3475, "loss": 4.690407562255859, "grad_norm": 1.0885580778121948, "learning_rate": 0.0001, "elbo": 4.69041, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20378841191649075}
{"step": 3480, "loss": 4.684662246704102, "grad_norm": 0.9939964413642883, "learning_rate": 0.0001, "elbo": 4.68466, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20408163265306123}
{"step": 3485, "loss": 4.672370147705078, "grad_norm": 1.0173863172531128, "learning_rate": 0.0001, "elbo": 4.67237, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2043748533896317}
{"step": 3485, "loss": 4.672370147705078, "grad_norm": 1.0173863172531128, "learning_rate": 0.0001, "elbo": 4.67237, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2043748533896317}
{"step": 3490, "loss": 4.674398803710938, "grad_norm": 0.8261673450469971, "learning_rate": 0.0001, "elbo": 4.6744, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2046680741262022}
{"step": 3495, "loss": 4.667684936523438, "grad_norm": 0.9310750365257263, "learning_rate": 0.0001, "elbo": 4.66769, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2049612948627727}
{"step": 3500, "loss": 4.655297088623047, "grad_norm": 1.1105234622955322, "learning_rate": 0.0001, "elbo": 4.6553, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20525451559934318}
{"step": 3505, "loss": 4.666008377075196, "grad_norm": 1.115145206451416, "learning_rate": 0.0001, "elbo": 4.66601, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20554773633591367}
{"step": 3510, "loss": 4.651618194580078, "grad_norm": 0.9763636589050293, "learning_rate": 0.0001, "elbo": 4.65162, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20584095707248418}
{"step": 3515, "loss": 4.66680793762207, "grad_norm": 1.097957968711853, "learning_rate": 0.0001, "elbo": 4.66681, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20613417780905466}
{"step": 3520, "loss": 4.667825698852539, "grad_norm": 0.8817471861839294, "learning_rate": 0.0001, "elbo": 4.66783, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20642739854562514}
{"step": 3525, "loss": 4.66149673461914, "grad_norm": 1.1709861755371094, "learning_rate": 0.0001, "elbo": 4.6615, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20672061928219565}
{"step": 3530, "loss": 4.677172088623047, "grad_norm": 1.107667088508606, "learning_rate": 0.0001, "elbo": 4.67717, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20701384001876613}
{"step": 3535, "loss": 4.668038940429687, "grad_norm": 0.9511038661003113, "learning_rate": 0.0001, "elbo": 4.66804, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20730706075533661}
{"step": 3540, "loss": 4.668875885009766, "grad_norm": 1.0645321607589722, "learning_rate": 0.0001, "elbo": 4.66888, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2076002814919071}
{"step": 3545, "loss": 4.6621955871582035, "grad_norm": 1.0168205499649048, "learning_rate": 0.0001, "elbo": 4.6622, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2078935022284776}
{"step": 3550, "loss": 4.678887939453125, "grad_norm": 1.0659303665161133, "learning_rate": 0.0001, "elbo": 4.67889, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2081867229650481}
{"step": 3555, "loss": 4.671516036987304, "grad_norm": 0.9285050630569458, "learning_rate": 0.0001, "elbo": 4.67152, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20847994370161857}
{"step": 3560, "loss": 4.655829620361328, "grad_norm": 0.9365557432174683, "learning_rate": 0.0001, "elbo": 4.65583, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20877316443818908}
{"step": 3565, "loss": 4.670267486572266, "grad_norm": 0.979785144329071, "learning_rate": 0.0001, "elbo": 4.67027, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20906638517475956}
{"step": 3570, "loss": 4.658868026733399, "grad_norm": 1.1113948822021484, "learning_rate": 0.0001, "elbo": 4.65887, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20935960591133004}
{"step": 3575, "loss": 4.643156433105469, "grad_norm": 1.0643421411514282, "learning_rate": 0.0001, "elbo": 4.64316, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20965282664790055}
{"step": 3580, "loss": 4.654785919189453, "grad_norm": 1.1856878995895386, "learning_rate": 0.0001, "elbo": 4.65479, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.20994604738447104}
{"step": 3585, "loss": 4.645536804199219, "grad_norm": 0.9561641812324524, "learning_rate": 0.0001, "elbo": 4.64554, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21023926812104152}
{"step": 3590, "loss": 4.654410552978516, "grad_norm": 1.2915490865707397, "learning_rate": 0.0001, "elbo": 4.65441, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.210532488857612}
{"step": 3595, "loss": 4.655246734619141, "grad_norm": 1.0364491939544678, "learning_rate": 0.0001, "elbo": 4.65525, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2108257095941825}
{"step": 3600, "loss": 4.645408630371094, "grad_norm": 1.0586224794387817, "learning_rate": 0.0001, "elbo": 4.64541, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.211118930330753}
{"step": 3605, "loss": 4.657096862792969, "grad_norm": 1.0410475730895996, "learning_rate": 0.0001, "elbo": 4.6571, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21141215106732347}
{"step": 3610, "loss": 4.630951690673828, "grad_norm": 1.083535075187683, "learning_rate": 0.0001, "elbo": 4.63095, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21170537180389398}
{"step": 3615, "loss": 4.655402374267578, "grad_norm": 0.9411219358444214, "learning_rate": 0.0001, "elbo": 4.6554, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21199859254046446}
{"step": 3620, "loss": 4.660095977783203, "grad_norm": 0.9702185392379761, "learning_rate": 0.0001, "elbo": 4.6601, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21229181327703495}
{"step": 3625, "loss": 4.6267951965332035, "grad_norm": 1.1552276611328125, "learning_rate": 0.0001, "elbo": 4.62679, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21258503401360543}
{"step": 3625, "loss": 4.6267951965332035, "grad_norm": 1.1552276611328125, "learning_rate": 0.0001, "elbo": 4.62679, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21258503401360543}
{"step": 3630, "loss": 4.649006652832031, "grad_norm": 0.9947168827056885, "learning_rate": 0.0001, "elbo": 4.64901, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21287825475017594}
{"step": 3635, "loss": 4.630628204345703, "grad_norm": 0.9741429686546326, "learning_rate": 0.0001, "elbo": 4.63063, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21317147548674642}
{"step": 3640, "loss": 4.630393981933594, "grad_norm": 1.063621997833252, "learning_rate": 0.0001, "elbo": 4.63039, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2134646962233169}
{"step": 3645, "loss": 4.639750671386719, "grad_norm": 0.9661415219306946, "learning_rate": 0.0001, "elbo": 4.63975, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2137579169598874}
{"step": 3650, "loss": 4.6230613708496096, "grad_norm": 1.1307483911514282, "learning_rate": 0.0001, "elbo": 4.62306, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2140511376964579}
{"step": 3655, "loss": 4.628629302978515, "grad_norm": 1.030551552772522, "learning_rate": 0.0001, "elbo": 4.62863, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21434435843302838}
{"step": 3660, "loss": 4.6359516143798825, "grad_norm": 0.926780641078949, "learning_rate": 0.0001, "elbo": 4.63595, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21463757916959889}
{"step": 3665, "loss": 4.637616348266602, "grad_norm": 1.2290709018707275, "learning_rate": 0.0001, "elbo": 4.63762, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21493079990616937}
{"step": 3670, "loss": 4.638419342041016, "grad_norm": 1.0420267581939697, "learning_rate": 0.0001, "elbo": 4.63842, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21522402064273985}
{"step": 3675, "loss": 4.640008163452149, "grad_norm": 1.0166023969650269, "learning_rate": 0.0001, "elbo": 4.64001, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21551724137931033}
{"step": 3680, "loss": 4.639599609375, "grad_norm": 0.986816942691803, "learning_rate": 0.0001, "elbo": 4.6396, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21581046211588084}
{"step": 3685, "loss": 4.629119110107422, "grad_norm": 1.0298353433609009, "learning_rate": 0.0001, "elbo": 4.62912, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21610368285245132}
{"step": 3690, "loss": 4.6302040100097654, "grad_norm": 1.0921720266342163, "learning_rate": 0.0001, "elbo": 4.6302, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2163969035890218}
{"step": 3695, "loss": 4.631623840332031, "grad_norm": 1.0604344606399536, "learning_rate": 0.0001, "elbo": 4.63162, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21669012432559231}
{"step": 3700, "loss": 4.636288833618164, "grad_norm": 0.9599320888519287, "learning_rate": 0.0001, "elbo": 4.63629, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2169833450621628}
{"step": 3705, "loss": 4.615207672119141, "grad_norm": 0.9689627885818481, "learning_rate": 0.0001, "elbo": 4.61521, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21727656579873328}
{"step": 3705, "loss": 4.615207672119141, "grad_norm": 0.9689627885818481, "learning_rate": 0.0001, "elbo": 4.61521, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21727656579873328}
{"step": 3710, "loss": 4.634861755371094, "grad_norm": 1.020667552947998, "learning_rate": 0.0001, "elbo": 4.63486, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2175697865353038}
{"step": 3715, "loss": 4.629246520996094, "grad_norm": 0.9578857421875, "learning_rate": 0.0001, "elbo": 4.62925, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21786300727187427}
{"step": 3720, "loss": 4.627053070068359, "grad_norm": 1.0518708229064941, "learning_rate": 0.0001, "elbo": 4.62705, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21815622800844475}
{"step": 3725, "loss": 4.622793579101563, "grad_norm": 1.096949815750122, "learning_rate": 0.0001, "elbo": 4.62279, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21844944874501523}
{"step": 3730, "loss": 4.634007263183594, "grad_norm": 1.379791498184204, "learning_rate": 0.0001, "elbo": 4.63401, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21874266948158574}
{"step": 3735, "loss": 4.649462890625, "grad_norm": 1.014580249786377, "learning_rate": 0.0001, "elbo": 4.64946, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21903589021815623}
{"step": 3740, "loss": 4.61693000793457, "grad_norm": 1.0019851922988892, "learning_rate": 0.0001, "elbo": 4.61693, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2193291109547267}
{"step": 3745, "loss": 4.645677185058593, "grad_norm": 1.370888590812683, "learning_rate": 0.0001, "elbo": 4.64568, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.21962233169129722}
{"step": 3750, "loss": 4.624431228637695, "grad_norm": 1.10905921459198, "learning_rate": 0.0001, "elbo": 4.62443, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2199155524278677}
{"step": 3755, "loss": 4.623565292358398, "grad_norm": 1.394844889640808, "learning_rate": 0.0001, "elbo": 4.62357, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22020877316443818}
{"step": 3760, "loss": 4.605828857421875, "grad_norm": 1.3617160320281982, "learning_rate": 0.0001, "elbo": 4.60583, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2205019939010087}
{"step": 3765, "loss": 4.631684112548828, "grad_norm": 0.9923453330993652, "learning_rate": 0.0001, "elbo": 4.63168, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22079521463757917}
{"step": 3770, "loss": 4.648683547973633, "grad_norm": 0.9778392314910889, "learning_rate": 0.0001, "elbo": 4.64868, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22108843537414966}
{"step": 3775, "loss": 4.601787567138672, "grad_norm": 1.0486966371536255, "learning_rate": 0.0001, "elbo": 4.60179, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22138165611072014}
{"step": 3780, "loss": 4.622042846679688, "grad_norm": 0.9715142250061035, "learning_rate": 0.0001, "elbo": 4.62204, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22167487684729065}
{"step": 3785, "loss": 4.626981353759765, "grad_norm": 1.2919914722442627, "learning_rate": 0.0001, "elbo": 4.62698, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22196809758386113}
{"step": 3790, "loss": 4.622826766967774, "grad_norm": 0.9367780685424805, "learning_rate": 0.0001, "elbo": 4.62283, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2222613183204316}
{"step": 3795, "loss": 4.602470779418946, "grad_norm": 0.9722598195075989, "learning_rate": 0.0001, "elbo": 4.60247, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22255453905700212}
{"step": 3800, "loss": 4.5955963134765625, "grad_norm": 0.9968383312225342, "learning_rate": 0.0001, "elbo": 4.5956, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2228477597935726}
{"step": 3800, "loss": 4.5955963134765625, "grad_norm": 0.9968383312225342, "learning_rate": 0.0001, "elbo": 4.5956, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2228477597935726}
{"step": 3805, "loss": 4.621110534667968, "grad_norm": 0.9398154020309448, "learning_rate": 0.0001, "elbo": 4.62111, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22314098053014308}
{"step": 3805, "loss": 4.621110534667968, "grad_norm": 0.9398154020309448, "learning_rate": 0.0001, "elbo": 4.62111, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22314098053014308}
{"step": 3810, "loss": 4.604676818847656, "grad_norm": 0.9373259544372559, "learning_rate": 0.0001, "elbo": 4.60468, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2234342012667136}
{"step": 3815, "loss": 4.62536392211914, "grad_norm": 1.146803617477417, "learning_rate": 0.0001, "elbo": 4.62536, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22372742200328408}
{"step": 3820, "loss": 4.6065116882324215, "grad_norm": 0.8806237578392029, "learning_rate": 0.0001, "elbo": 4.60651, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22402064273985456}
{"step": 3825, "loss": 4.605603790283203, "grad_norm": 0.9802260398864746, "learning_rate": 0.0001, "elbo": 4.6056, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22431386347642504}
{"step": 3830, "loss": 4.616727447509765, "grad_norm": 0.9212145805358887, "learning_rate": 0.0001, "elbo": 4.61673, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22460708421299555}
{"step": 3835, "loss": 4.6105499267578125, "grad_norm": 1.0373966693878174, "learning_rate": 0.0001, "elbo": 4.61055, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22490030494956603}
{"step": 3840, "loss": 4.611476135253906, "grad_norm": 1.462336540222168, "learning_rate": 0.0001, "elbo": 4.61148, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22519352568613651}
{"step": 3845, "loss": 4.613965606689453, "grad_norm": 1.0027092695236206, "learning_rate": 0.0001, "elbo": 4.61397, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22548674642270702}
{"step": 3850, "loss": 4.603685760498047, "grad_norm": 0.9981302618980408, "learning_rate": 0.0001, "elbo": 4.60369, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2257799671592775}
{"step": 3855, "loss": 4.609204483032227, "grad_norm": 0.9347818493843079, "learning_rate": 0.0001, "elbo": 4.6092, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.226073187895848}
{"step": 3860, "loss": 4.607998275756836, "grad_norm": 0.9620808959007263, "learning_rate": 0.0001, "elbo": 4.608, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2263664086324185}
{"step": 3865, "loss": 4.608002853393555, "grad_norm": 1.0093094110488892, "learning_rate": 0.0001, "elbo": 4.608, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22665962936898898}
{"step": 3870, "loss": 4.607094192504883, "grad_norm": 1.0811822414398193, "learning_rate": 0.0001, "elbo": 4.60709, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22695285010555946}
{"step": 3875, "loss": 4.6106712341308596, "grad_norm": 0.9146448373794556, "learning_rate": 0.0001, "elbo": 4.61067, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22724607084212994}
{"step": 3880, "loss": 4.602963256835937, "grad_norm": 0.9222270846366882, "learning_rate": 0.0001, "elbo": 4.60296, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22753929157870045}
{"step": 3885, "loss": 4.599362564086914, "grad_norm": 1.029880166053772, "learning_rate": 0.0001, "elbo": 4.59936, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22783251231527094}
{"step": 3890, "loss": 4.60969352722168, "grad_norm": 0.9130651950836182, "learning_rate": 0.0001, "elbo": 4.60969, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22812573305184142}
{"step": 3895, "loss": 4.585784530639648, "grad_norm": 1.0560530424118042, "learning_rate": 0.0001, "elbo": 4.58578, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22841895378841193}
{"step": 3900, "loss": 4.592781066894531, "grad_norm": 1.0169928073883057, "learning_rate": 0.0001, "elbo": 4.59278, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2287121745249824}
{"step": 3905, "loss": 4.595635986328125, "grad_norm": 0.9809831976890564, "learning_rate": 0.0001, "elbo": 4.59564, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2290053952615529}
{"step": 3910, "loss": 4.580006408691406, "grad_norm": 0.9599353671073914, "learning_rate": 0.0001, "elbo": 4.58001, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2292986159981234}
{"step": 3915, "loss": 4.5902854919433596, "grad_norm": 1.041661262512207, "learning_rate": 0.0001, "elbo": 4.59029, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22959183673469388}
{"step": 3915, "loss": 4.5902854919433596, "grad_norm": 1.041661262512207, "learning_rate": 0.0001, "elbo": 4.59029, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22959183673469388}
{"step": 3920, "loss": 4.604524612426758, "grad_norm": 1.0563480854034424, "learning_rate": 0.0001, "elbo": 4.60452, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.22988505747126436}
{"step": 3925, "loss": 4.588922119140625, "grad_norm": 1.0124366283416748, "learning_rate": 0.0001, "elbo": 4.58892, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23017827820783485}
{"step": 3930, "loss": 4.613782119750977, "grad_norm": 1.0282936096191406, "learning_rate": 0.0001, "elbo": 4.61378, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23047149894440536}
{"step": 3930, "loss": 4.613782119750977, "grad_norm": 1.0282936096191406, "learning_rate": 0.0001, "elbo": 4.61378, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23047149894440536}
{"step": 3935, "loss": 4.5815895080566404, "grad_norm": 0.9263477325439453, "learning_rate": 0.0001, "elbo": 4.58159, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23076471968097584}
{"step": 3940, "loss": 4.57556037902832, "grad_norm": 1.0421236753463745, "learning_rate": 0.0001, "elbo": 4.57556, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23105794041754632}
{"step": 3945, "loss": 4.57928237915039, "grad_norm": 0.9477463364601135, "learning_rate": 0.0001, "elbo": 4.57928, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23135116115411683}
{"step": 3950, "loss": 4.58848876953125, "grad_norm": 1.0079647302627563, "learning_rate": 0.0001, "elbo": 4.58849, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2316443818906873}
{"step": 3955, "loss": 4.598503112792969, "grad_norm": 0.9686462879180908, "learning_rate": 0.0001, "elbo": 4.5985, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2319376026272578}
{"step": 3960, "loss": 4.581238555908203, "grad_norm": 1.2036471366882324, "learning_rate": 0.0001, "elbo": 4.58124, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2322308233638283}
{"step": 3965, "loss": 4.591516494750977, "grad_norm": 0.996313214302063, "learning_rate": 0.0001, "elbo": 4.59152, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23252404410039879}
{"step": 3970, "loss": 4.577881622314453, "grad_norm": 0.9826664924621582, "learning_rate": 0.0001, "elbo": 4.57788, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23281726483696927}
{"step": 3975, "loss": 4.579068756103515, "grad_norm": 0.944316029548645, "learning_rate": 0.0001, "elbo": 4.57907, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23311048557353975}
{"step": 3980, "loss": 4.575828552246094, "grad_norm": 0.9373348951339722, "learning_rate": 0.0001, "elbo": 4.57583, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23340370631011026}
{"step": 3985, "loss": 4.5856681823730465, "grad_norm": 0.9492429494857788, "learning_rate": 0.0001, "elbo": 4.58567, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23369692704668074}
{"step": 3990, "loss": 4.568650817871093, "grad_norm": 1.0232404470443726, "learning_rate": 0.0001, "elbo": 4.56865, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23399014778325122}
{"step": 3995, "loss": 4.576231384277344, "grad_norm": 1.0723519325256348, "learning_rate": 0.0001, "elbo": 4.57623, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23428336851982173}
{"step": 4000, "loss": 4.58082275390625, "grad_norm": 0.9735103845596313, "learning_rate": 0.0001, "elbo": 4.58082, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23457658925639222}
{"step": 4005, "loss": 4.580538940429688, "grad_norm": 1.1285839080810547, "learning_rate": 0.0001, "elbo": 4.58054, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2348698099929627}
{"step": 4010, "loss": 4.576457214355469, "grad_norm": 1.0044163465499878, "learning_rate": 0.0001, "elbo": 4.57646, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23516303072953318}
{"step": 4015, "loss": 4.562297821044922, "grad_norm": 1.072094202041626, "learning_rate": 0.0001, "elbo": 4.5623, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2354562514661037}
{"step": 4020, "loss": 4.574095153808594, "grad_norm": 1.1880617141723633, "learning_rate": 0.0001, "elbo": 4.5741, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23574947220267417}
{"step": 4025, "loss": 4.561313629150391, "grad_norm": 0.935147225856781, "learning_rate": 0.0001, "elbo": 4.56131, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23604269293924465}
{"step": 4030, "loss": 4.562753677368164, "grad_norm": 0.9556302428245544, "learning_rate": 0.0001, "elbo": 4.56275, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23633591367581516}
{"step": 4035, "loss": 4.567867660522461, "grad_norm": 1.0846935510635376, "learning_rate": 0.0001, "elbo": 4.56787, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23662913441238564}
{"step": 4040, "loss": 4.572234725952148, "grad_norm": 1.0313342809677124, "learning_rate": 0.0001, "elbo": 4.57223, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23692235514895613}
{"step": 4045, "loss": 4.546224975585938, "grad_norm": 1.2939554452896118, "learning_rate": 0.0001, "elbo": 4.54622, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23721557588552664}
{"step": 4050, "loss": 4.578832626342773, "grad_norm": 0.9457508325576782, "learning_rate": 0.0001, "elbo": 4.57883, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23750879662209712}
{"step": 4055, "loss": 4.5787109375, "grad_norm": 0.9959350824356079, "learning_rate": 0.0001, "elbo": 4.57871, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2378020173586676}
{"step": 4060, "loss": 4.5693916320800785, "grad_norm": 0.9803762435913086, "learning_rate": 0.0001, "elbo": 4.56939, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23809523809523808}
{"step": 4065, "loss": 4.562069320678711, "grad_norm": 0.939852237701416, "learning_rate": 0.0001, "elbo": 4.56207, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2383884588318086}
{"step": 4070, "loss": 4.560548400878906, "grad_norm": 1.0024913549423218, "learning_rate": 0.0001, "elbo": 4.56055, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23868167956837907}
{"step": 4075, "loss": 4.556158447265625, "grad_norm": 0.9355496764183044, "learning_rate": 0.0001, "elbo": 4.55616, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23897490030494956}
{"step": 4080, "loss": 4.576498413085938, "grad_norm": 0.9399572014808655, "learning_rate": 0.0001, "elbo": 4.5765, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23926812104152007}
{"step": 4085, "loss": 4.574707412719727, "grad_norm": 0.9213935732841492, "learning_rate": 0.0001, "elbo": 4.57471, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23956134177809055}
{"step": 4090, "loss": 4.545302200317383, "grad_norm": 1.1070681810379028, "learning_rate": 0.0001, "elbo": 4.5453, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.23985456251466103}
{"step": 4095, "loss": 4.551914978027344, "grad_norm": 1.033837914466858, "learning_rate": 0.0001, "elbo": 4.55191, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24014778325123154}
{"step": 4100, "loss": 4.562852478027343, "grad_norm": 0.8743427395820618, "learning_rate": 0.0001, "elbo": 4.56285, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24044100398780202}
{"step": 4105, "loss": 4.574506378173828, "grad_norm": 0.9183300733566284, "learning_rate": 0.0001, "elbo": 4.57451, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2407342247243725}
{"step": 4105, "loss": 4.574506378173828, "grad_norm": 0.9183300733566284, "learning_rate": 0.0001, "elbo": 4.57451, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2407342247243725}
{"step": 4110, "loss": 4.55590591430664, "grad_norm": 0.9276518821716309, "learning_rate": 0.0001, "elbo": 4.55591, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24102744546094299}
{"step": 4110, "loss": 4.55590591430664, "grad_norm": 0.9276518821716309, "learning_rate": 0.0001, "elbo": 4.55591, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24102744546094299}
{"step": 4115, "loss": 4.575655364990235, "grad_norm": 1.0129362344741821, "learning_rate": 0.0001, "elbo": 4.57566, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2413206661975135}
{"step": 4120, "loss": 4.537330627441406, "grad_norm": 0.9243236184120178, "learning_rate": 0.0001, "elbo": 4.53733, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24161388693408398}
{"step": 4125, "loss": 4.550699615478516, "grad_norm": 0.9201699495315552, "learning_rate": 0.0001, "elbo": 4.5507, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24190710767065446}
{"step": 4130, "loss": 4.543686294555664, "grad_norm": 0.896750271320343, "learning_rate": 0.0001, "elbo": 4.54369, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24220032840722497}
{"step": 4130, "loss": 4.543686294555664, "grad_norm": 0.896750271320343, "learning_rate": 0.0001, "elbo": 4.54369, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24220032840722497}
{"step": 4135, "loss": 4.543855285644531, "grad_norm": 1.2472628355026245, "learning_rate": 0.0001, "elbo": 4.54386, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24249354914379545}
{"step": 4140, "loss": 4.525697708129883, "grad_norm": 0.9935011267662048, "learning_rate": 0.0001, "elbo": 4.5257, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24278676988036593}
{"step": 4145, "loss": 4.5452827453613285, "grad_norm": 0.9817330241203308, "learning_rate": 0.0001, "elbo": 4.54528, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24307999061693644}
{"step": 4150, "loss": 4.541548538208008, "grad_norm": 0.9930524230003357, "learning_rate": 0.0001, "elbo": 4.54155, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24337321135350692}
{"step": 4155, "loss": 4.5358928680419925, "grad_norm": 1.3096745014190674, "learning_rate": 0.0001, "elbo": 4.53589, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2436664320900774}
{"step": 4160, "loss": 4.557326126098633, "grad_norm": 1.0159766674041748, "learning_rate": 0.0001, "elbo": 4.55733, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2439596528266479}
{"step": 4165, "loss": 4.560203933715821, "grad_norm": 1.2685375213623047, "learning_rate": 0.0001, "elbo": 4.5602, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2442528735632184}
{"step": 4170, "loss": 4.544101333618164, "grad_norm": 0.897757887840271, "learning_rate": 0.0001, "elbo": 4.5441, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24454609429978888}
{"step": 4175, "loss": 4.548933792114258, "grad_norm": 0.9469509124755859, "learning_rate": 0.0001, "elbo": 4.54893, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24483931503635936}
{"step": 4180, "loss": 4.538737869262695, "grad_norm": 0.9416372776031494, "learning_rate": 0.0001, "elbo": 4.53874, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24513253577292987}
{"step": 4185, "loss": 4.526321029663086, "grad_norm": 0.9114320278167725, "learning_rate": 0.0001, "elbo": 4.52632, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24542575650950035}
{"step": 4190, "loss": 4.543342971801758, "grad_norm": 0.9542516469955444, "learning_rate": 0.0001, "elbo": 4.54334, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24571897724607084}
{"step": 4195, "loss": 4.531101226806641, "grad_norm": 0.9227150082588196, "learning_rate": 0.0001, "elbo": 4.5311, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24601219798264135}
{"step": 4200, "loss": 4.536427688598633, "grad_norm": 1.0179245471954346, "learning_rate": 0.0001, "elbo": 4.53643, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24630541871921183}
{"step": 4205, "loss": 4.5364940643310545, "grad_norm": 0.9806710481643677, "learning_rate": 0.0001, "elbo": 4.53649, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2465986394557823}
{"step": 4210, "loss": 4.5416511535644535, "grad_norm": 0.9003105163574219, "learning_rate": 0.0001, "elbo": 4.54165, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2468918601923528}
{"step": 4215, "loss": 4.548404693603516, "grad_norm": 0.928198516368866, "learning_rate": 0.0001, "elbo": 4.5484, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2471850809289233}
{"step": 4220, "loss": 4.537823867797852, "grad_norm": 0.9182760715484619, "learning_rate": 0.0001, "elbo": 4.53782, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24747830166549378}
{"step": 4225, "loss": 4.544781494140625, "grad_norm": 0.9477521181106567, "learning_rate": 0.0001, "elbo": 4.54478, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24777152240206426}
{"step": 4230, "loss": 4.541860198974609, "grad_norm": 1.2380120754241943, "learning_rate": 0.0001, "elbo": 4.54186, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24806474313863477}
{"step": 4235, "loss": 4.544217681884765, "grad_norm": 0.879469096660614, "learning_rate": 0.0001, "elbo": 4.54422, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24835796387520526}
{"step": 4240, "loss": 4.52969970703125, "grad_norm": 1.0526340007781982, "learning_rate": 0.0001, "elbo": 4.5297, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24865118461177574}
{"step": 4245, "loss": 4.528092575073242, "grad_norm": 0.9123741388320923, "learning_rate": 0.0001, "elbo": 4.52809, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24894440534834625}
{"step": 4250, "loss": 4.540900421142578, "grad_norm": 1.0186280012130737, "learning_rate": 0.0001, "elbo": 4.5409, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.24923762608491673}
{"step": 4255, "loss": 4.534482955932617, "grad_norm": 0.8771601319313049, "learning_rate": 0.0001, "elbo": 4.53448, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2495308468214872}
{"step": 4260, "loss": 4.529645538330078, "grad_norm": 0.8966130614280701, "learning_rate": 0.0001, "elbo": 4.52965, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2498240675580577}
{"step": 4265, "loss": 4.522235870361328, "grad_norm": 0.9531166553497314, "learning_rate": 0.0001, "elbo": 4.52224, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2501172882946282}
{"step": 4270, "loss": 4.534079742431641, "grad_norm": 0.9951554536819458, "learning_rate": 0.0001, "elbo": 4.53408, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2504105090311987}
{"step": 4275, "loss": 4.51696548461914, "grad_norm": 0.9745646119117737, "learning_rate": 0.0001, "elbo": 4.51697, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.25070372976776917}
{"step": 4280, "loss": 4.527442550659179, "grad_norm": 1.0054497718811035, "learning_rate": 0.0001, "elbo": 4.52744, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2509969505043397}
{"step": 4285, "loss": 4.525704956054687, "grad_norm": 1.0803275108337402, "learning_rate": 0.0001, "elbo": 4.52571, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.25129017124091013}
{"step": 4290, "loss": 4.515298461914062, "grad_norm": 0.8988998532295227, "learning_rate": 0.0001, "elbo": 4.5153, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.25158339197748064}
{"step": 4295, "loss": 4.51439094543457, "grad_norm": 1.2376646995544434, "learning_rate": 0.0001, "elbo": 4.51439, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.25187661271405115}
{"step": 4300, "loss": 4.52940559387207, "grad_norm": 0.9461541771888733, "learning_rate": 0.0001, "elbo": 4.52941, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2521698334506216}
{"step": 4305, "loss": 4.526014709472657, "grad_norm": 1.0156939029693604, "learning_rate": 0.0001, "elbo": 4.52602, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2524630541871921}
{"step": 4310, "loss": 4.529000854492187, "grad_norm": 0.8238926529884338, "learning_rate": 0.0001, "elbo": 4.529, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2527562749237626}
{"step": 4315, "loss": 4.531956481933594, "grad_norm": 1.184974193572998, "learning_rate": 0.0001, "elbo": 4.53196, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2530494956603331}
{"step": 4320, "loss": 4.526504135131836, "grad_norm": 0.9998613595962524, "learning_rate": 0.0001, "elbo": 4.5265, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2533427163969036}
{"step": 4325, "loss": 4.505126190185547, "grad_norm": 0.9531392455101013, "learning_rate": 0.0001, "elbo": 4.50513, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2536359371334741}
{"step": 4330, "loss": 4.5112464904785154, "grad_norm": 0.9178735613822937, "learning_rate": 0.0001, "elbo": 4.51125, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.25392915787004455}
{"step": 4335, "loss": 4.516979217529297, "grad_norm": 1.0667015314102173, "learning_rate": 0.0001, "elbo": 4.51698, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.25422237860661506}
{"step": 4340, "loss": 4.517490386962891, "grad_norm": 1.2546290159225464, "learning_rate": 0.0001, "elbo": 4.51749, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2545155993431856}
{"step": 4345, "loss": 4.531879425048828, "grad_norm": 0.9775177240371704, "learning_rate": 0.0001, "elbo": 4.53188, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.254808820079756}
{"step": 4350, "loss": 4.52246322631836, "grad_norm": 1.0240215063095093, "learning_rate": 0.0001, "elbo": 4.52246, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.25510204081632654}
{"step": 4355, "loss": 4.526968383789063, "grad_norm": 0.90456622838974, "learning_rate": 0.0001, "elbo": 4.52697, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.25539526155289705}
{"step": 4360, "loss": 4.504724502563477, "grad_norm": 0.9467153549194336, "learning_rate": 0.0001, "elbo": 4.50472, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2556884822894675}
{"step": 4365, "loss": 4.5361480712890625, "grad_norm": 1.001144528388977, "learning_rate": 0.0001, "elbo": 4.53615, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.255981703026038}
{"step": 4370, "loss": 4.519499969482422, "grad_norm": 1.0047497749328613, "learning_rate": 0.0001, "elbo": 4.5195, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.25627492376260846}
{"step": 4375, "loss": 4.531842041015625, "grad_norm": 0.9443824291229248, "learning_rate": 0.0001, "elbo": 4.53184, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.256568144499179}
{"step": 4380, "loss": 4.497110366821289, "grad_norm": 0.943471372127533, "learning_rate": 0.0001, "elbo": 4.49711, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2568613652357495}
{"step": 4385, "loss": 4.525429534912109, "grad_norm": 1.0726088285446167, "learning_rate": 0.0001, "elbo": 4.52543, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.25715458597231994}
{"step": 4390, "loss": 4.5047260284423825, "grad_norm": 0.9379811882972717, "learning_rate": 0.0001, "elbo": 4.50473, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.25744780670889045}
{"step": 4395, "loss": 4.507879257202148, "grad_norm": 0.8866432309150696, "learning_rate": 0.0001, "elbo": 4.50788, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.25774102744546096}
{"step": 4400, "loss": 4.518882751464844, "grad_norm": 1.0190635919570923, "learning_rate": 0.0001, "elbo": 4.51888, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2580342481820314}
{"step": 4405, "loss": 4.50185546875, "grad_norm": 1.025127649307251, "learning_rate": 0.0001, "elbo": 4.50186, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2583274689186019}
{"step": 4410, "loss": 4.513029098510742, "grad_norm": 0.9292516112327576, "learning_rate": 0.0001, "elbo": 4.51303, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.25862068965517243}
{"step": 4415, "loss": 4.50964584350586, "grad_norm": 0.9209949970245361, "learning_rate": 0.0001, "elbo": 4.50965, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2589139103917429}
{"step": 4420, "loss": 4.501356506347657, "grad_norm": 1.1011154651641846, "learning_rate": 0.0001, "elbo": 4.50136, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2592071311283134}
{"step": 4425, "loss": 4.501789855957031, "grad_norm": 0.9657236933708191, "learning_rate": 0.0001, "elbo": 4.50179, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2595003518648839}
{"step": 4430, "loss": 4.504133605957032, "grad_norm": 0.8483617305755615, "learning_rate": 0.0001, "elbo": 4.50413, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.25979357260145436}
{"step": 4435, "loss": 4.5006366729736325, "grad_norm": 0.9641113877296448, "learning_rate": 0.0001, "elbo": 4.50064, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.26008679333802487}
{"step": 4440, "loss": 4.49627914428711, "grad_norm": 0.8469149470329285, "learning_rate": 0.0001, "elbo": 4.49628, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2603800140745954}
{"step": 4445, "loss": 4.485132217407227, "grad_norm": 0.9847970604896545, "learning_rate": 0.0001, "elbo": 4.48513, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.26067323481116583}
{"step": 4450, "loss": 4.510214996337891, "grad_norm": 1.12317955493927, "learning_rate": 0.0001, "elbo": 4.51022, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.26096645554773634}
{"step": 4455, "loss": 4.509906768798828, "grad_norm": 0.944618284702301, "learning_rate": 0.0001, "elbo": 4.50991, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.26125967628430685}
{"step": 4460, "loss": 4.510390090942383, "grad_norm": 0.9794960021972656, "learning_rate": 0.0001, "elbo": 4.51039, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2615528970208773}
{"step": 4465, "loss": 4.500073623657227, "grad_norm": 0.8955798149108887, "learning_rate": 0.0001, "elbo": 4.50007, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2618461177574478}
{"step": 4470, "loss": 4.504786682128906, "grad_norm": 1.1034594774246216, "learning_rate": 0.0001, "elbo": 4.50479, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.26213933849401827}
{"step": 4475, "loss": 4.494761276245117, "grad_norm": 1.1145286560058594, "learning_rate": 0.0001, "elbo": 4.49476, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2624325592305888}
{"step": 4480, "loss": 4.494661712646485, "grad_norm": 0.9492529630661011, "learning_rate": 0.0001, "elbo": 4.49466, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2627257799671593}
{"step": 4485, "loss": 4.495373916625977, "grad_norm": 0.8298590779304504, "learning_rate": 0.0001, "elbo": 4.49537, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.26301900070372974}
{"step": 4490, "loss": 4.498955535888672, "grad_norm": 1.355624794960022, "learning_rate": 0.0001, "elbo": 4.49896, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.26331222144030025}
{"step": 4495, "loss": 4.481423568725586, "grad_norm": 1.3005778789520264, "learning_rate": 0.0001, "elbo": 4.48142, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.26360544217687076}
{"step": 4500, "loss": 4.499174499511719, "grad_norm": 0.9881749749183655, "learning_rate": 0.0001, "elbo": 4.49917, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2638986629134412}
{"step": 4505, "loss": 4.485126495361328, "grad_norm": 0.9094829559326172, "learning_rate": 0.0001, "elbo": 4.48513, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2641918836500117}
{"step": 4510, "loss": 4.496585845947266, "grad_norm": 0.896825909614563, "learning_rate": 0.0001, "elbo": 4.49659, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.26448510438658224}
{"step": 4515, "loss": 4.47790756225586, "grad_norm": 1.0457857847213745, "learning_rate": 0.0001, "elbo": 4.47791, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2647783251231527}
{"step": 4520, "loss": 4.490329360961914, "grad_norm": 0.9135520458221436, "learning_rate": 0.0001, "elbo": 4.49033, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2650715458597232}
{"step": 4525, "loss": 4.495619201660157, "grad_norm": 1.3562604188919067, "learning_rate": 0.0001, "elbo": 4.49562, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2653647665962937}
{"step": 4530, "loss": 4.4910888671875, "grad_norm": 0.9943324327468872, "learning_rate": 0.0001, "elbo": 4.49109, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.26565798733286416}
{"step": 4535, "loss": 4.489614868164063, "grad_norm": 0.980583667755127, "learning_rate": 0.0001, "elbo": 4.48961, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2659512080694347}
{"step": 4540, "loss": 4.505668258666992, "grad_norm": 1.2425472736358643, "learning_rate": 0.0001, "elbo": 4.50567, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2662444288060052}
{"step": 4545, "loss": 4.4877464294433596, "grad_norm": 1.0396316051483154, "learning_rate": 0.0001, "elbo": 4.48775, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.26653764954257564}
{"step": 4550, "loss": 4.500979232788086, "grad_norm": 0.9450427293777466, "learning_rate": 0.0001, "elbo": 4.50098, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.26683087027914615}
{"step": 4555, "loss": 4.486631774902344, "grad_norm": 0.9551349878311157, "learning_rate": 0.0001, "elbo": 4.48663, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.26712409101571666}
{"step": 4560, "loss": 4.487152099609375, "grad_norm": 1.1295498609542847, "learning_rate": 0.0001, "elbo": 4.48715, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2674173117522871}
{"step": 4565, "loss": 4.481184387207032, "grad_norm": 1.183786153793335, "learning_rate": 0.0001, "elbo": 4.48118, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2677105324888576}
{"step": 4570, "loss": 4.495964431762696, "grad_norm": 0.8904838562011719, "learning_rate": 0.0001, "elbo": 4.49596, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2680037532254281}
{"step": 4575, "loss": 4.472069168090821, "grad_norm": 1.1408612728118896, "learning_rate": 0.0001, "elbo": 4.47207, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2682969739619986}
{"step": 4580, "loss": 4.486399841308594, "grad_norm": 0.9176010489463806, "learning_rate": 0.0001, "elbo": 4.4864, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2685901946985691}
{"step": 4585, "loss": 4.486558532714843, "grad_norm": 0.8815539479255676, "learning_rate": 0.0001, "elbo": 4.48656, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.26888341543513955}
{"step": 4590, "loss": 4.4971565246582035, "grad_norm": 1.0701302289962769, "learning_rate": 0.0001, "elbo": 4.49716, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.26917663617171006}
{"step": 4595, "loss": 4.487739562988281, "grad_norm": 1.0460734367370605, "learning_rate": 0.0001, "elbo": 4.48774, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.26946985690828057}
{"step": 4600, "loss": 4.489653778076172, "grad_norm": 1.15713369846344, "learning_rate": 0.0001, "elbo": 4.48965, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.269763077644851}
{"step": 4605, "loss": 4.494769287109375, "grad_norm": 0.9417960047721863, "learning_rate": 0.0001, "elbo": 4.49477, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.27005629838142153}
{"step": 4610, "loss": 4.488881683349609, "grad_norm": 0.9675475358963013, "learning_rate": 0.0001, "elbo": 4.48888, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.27034951911799204}
{"step": 4615, "loss": 4.449818420410156, "grad_norm": 1.0021240711212158, "learning_rate": 0.0001, "elbo": 4.44982, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2706427398545625}
{"step": 4620, "loss": 4.483446884155273, "grad_norm": 1.0564061403274536, "learning_rate": 0.0001, "elbo": 4.48345, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.270935960591133}
{"step": 4625, "loss": 4.475310134887695, "grad_norm": 1.0275146961212158, "learning_rate": 0.0001, "elbo": 4.47531, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2712291813277035}
{"step": 4630, "loss": 4.473977279663086, "grad_norm": 0.9344931244850159, "learning_rate": 0.0001, "elbo": 4.47398, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.27152240206427397}
{"step": 4635, "loss": 4.477703857421875, "grad_norm": 1.090520977973938, "learning_rate": 0.0001, "elbo": 4.4777, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2718156228008445}
{"step": 4640, "loss": 4.472550964355468, "grad_norm": 1.0299803018569946, "learning_rate": 0.0001, "elbo": 4.47255, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.272108843537415}
{"step": 4645, "loss": 4.486374282836914, "grad_norm": 0.9410706162452698, "learning_rate": 0.0001, "elbo": 4.48637, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.27240206427398544}
{"step": 4650, "loss": 4.463907623291016, "grad_norm": 0.9408074021339417, "learning_rate": 0.0001, "elbo": 4.46391, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.27269528501055595}
{"step": 4655, "loss": 4.454412841796875, "grad_norm": 0.97746741771698, "learning_rate": 0.0001, "elbo": 4.45441, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.27298850574712646}
{"step": 4660, "loss": 4.468953323364258, "grad_norm": 0.8482887744903564, "learning_rate": 0.0001, "elbo": 4.46895, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2732817264836969}
{"step": 4665, "loss": 4.455562591552734, "grad_norm": 0.8759891390800476, "learning_rate": 0.0001, "elbo": 4.45556, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.27357494722026743}
{"step": 4670, "loss": 4.476309204101563, "grad_norm": 1.2570101022720337, "learning_rate": 0.0001, "elbo": 4.47631, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2738681679568379}
{"step": 4675, "loss": 4.480067443847656, "grad_norm": 1.0441752672195435, "learning_rate": 0.0001, "elbo": 4.48007, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2741613886934084}
{"step": 4680, "loss": 4.462603759765625, "grad_norm": 0.9563014507293701, "learning_rate": 0.0001, "elbo": 4.4626, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2744546094299789}
{"step": 4685, "loss": 4.448892593383789, "grad_norm": 1.1716262102127075, "learning_rate": 0.0001, "elbo": 4.44889, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.27474783016654936}
{"step": 4690, "loss": 4.453960418701172, "grad_norm": 0.9308747053146362, "learning_rate": 0.0001, "elbo": 4.45396, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.27504105090311987}
{"step": 4695, "loss": 4.471257019042969, "grad_norm": 0.9080168008804321, "learning_rate": 0.0001, "elbo": 4.47126, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2753342716396904}
{"step": 4700, "loss": 4.464822006225586, "grad_norm": 0.872204065322876, "learning_rate": 0.0001, "elbo": 4.46482, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.27562749237626083}
{"step": 4705, "loss": 4.468863677978516, "grad_norm": 0.933599054813385, "learning_rate": 0.0001, "elbo": 4.46886, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.27592071311283134}
{"step": 4710, "loss": 4.464975738525391, "grad_norm": 0.9313304424285889, "learning_rate": 0.0001, "elbo": 4.46498, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.27621393384940185}
{"step": 4715, "loss": 4.473119735717773, "grad_norm": 0.9159789085388184, "learning_rate": 0.0001, "elbo": 4.47312, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2765071545859723}
{"step": 4720, "loss": 4.459022521972656, "grad_norm": 0.8775367736816406, "learning_rate": 0.0001, "elbo": 4.45902, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2768003753225428}
{"step": 4725, "loss": 4.4605659484863285, "grad_norm": 0.954160749912262, "learning_rate": 0.0001, "elbo": 4.46057, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2770935960591133}
{"step": 4730, "loss": 4.47855224609375, "grad_norm": 0.9247435927391052, "learning_rate": 0.0001, "elbo": 4.47855, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2773868167956838}
{"step": 4735, "loss": 4.4489387512207035, "grad_norm": 0.9209519028663635, "learning_rate": 0.0001, "elbo": 4.44894, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2776800375322543}
{"step": 4740, "loss": 4.449122619628906, "grad_norm": 1.003352403640747, "learning_rate": 0.0001, "elbo": 4.44912, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2779732582688248}
{"step": 4745, "loss": 4.452780151367188, "grad_norm": 1.4788076877593994, "learning_rate": 0.0001, "elbo": 4.45278, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.27826647900539525}
{"step": 4745, "loss": 4.452780151367188, "grad_norm": 1.4788076877593994, "learning_rate": 0.0001, "elbo": 4.45278, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.27826647900539525}
{"step": 4750, "loss": 4.4671977996826175, "grad_norm": 0.9566009044647217, "learning_rate": 0.0001, "elbo": 4.4672, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.27855969974196576}
{"step": 4755, "loss": 4.46614990234375, "grad_norm": 0.933016300201416, "learning_rate": 0.0001, "elbo": 4.46615, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2788529204785362}
{"step": 4760, "loss": 4.453419494628906, "grad_norm": 1.1140177249908447, "learning_rate": 0.0001, "elbo": 4.45342, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2791461412151067}
{"step": 4765, "loss": 4.454498291015625, "grad_norm": 0.9480468034744263, "learning_rate": 0.0001, "elbo": 4.4545, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.27943936195167723}
{"step": 4770, "loss": 4.452849578857422, "grad_norm": 0.8359913229942322, "learning_rate": 0.0001, "elbo": 4.45285, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2797325826882477}
{"step": 4775, "loss": 4.473102188110351, "grad_norm": 1.0654592514038086, "learning_rate": 0.0001, "elbo": 4.4731, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2800258034248182}
{"step": 4780, "loss": 4.448442459106445, "grad_norm": 0.9045840501785278, "learning_rate": 0.0001, "elbo": 4.44844, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2803190241613887}
{"step": 4785, "loss": 4.466700744628906, "grad_norm": 0.88127201795578, "learning_rate": 0.0001, "elbo": 4.4667, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28061224489795916}
{"step": 4790, "loss": 4.471813201904297, "grad_norm": 0.976530909538269, "learning_rate": 0.0001, "elbo": 4.47181, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28090546563452967}
{"step": 4795, "loss": 4.449519348144531, "grad_norm": 1.013723373413086, "learning_rate": 0.0001, "elbo": 4.44952, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2811986863711002}
{"step": 4800, "loss": 4.4646553039550785, "grad_norm": 0.8165955543518066, "learning_rate": 0.0001, "elbo": 4.46466, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28149190710767064}
{"step": 4805, "loss": 4.449859619140625, "grad_norm": 0.9589222073554993, "learning_rate": 0.0001, "elbo": 4.44986, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28178512784424115}
{"step": 4810, "loss": 4.472366714477539, "grad_norm": 1.2357739210128784, "learning_rate": 0.0001, "elbo": 4.47237, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28207834858081166}
{"step": 4815, "loss": 4.451877593994141, "grad_norm": 1.1573810577392578, "learning_rate": 0.0001, "elbo": 4.45188, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2823715693173821}
{"step": 4820, "loss": 4.445896148681641, "grad_norm": 1.009934663772583, "learning_rate": 0.0001, "elbo": 4.4459, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2826647900539526}
{"step": 4825, "loss": 4.456396484375, "grad_norm": 0.9681873917579651, "learning_rate": 0.0001, "elbo": 4.4564, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28295801079052313}
{"step": 4830, "loss": 4.445028686523438, "grad_norm": 0.9737779498100281, "learning_rate": 0.0001, "elbo": 4.44503, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2832512315270936}
{"step": 4835, "loss": 4.4559272766113285, "grad_norm": 0.8796208500862122, "learning_rate": 0.0001, "elbo": 4.45593, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2835444522636641}
{"step": 4840, "loss": 4.462104034423828, "grad_norm": 0.9699661731719971, "learning_rate": 0.0001, "elbo": 4.4621, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2838376730002346}
{"step": 4845, "loss": 4.436653900146484, "grad_norm": 1.032260775566101, "learning_rate": 0.0001, "elbo": 4.43665, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28413089373680506}
{"step": 4850, "loss": 4.450336456298828, "grad_norm": 1.3413910865783691, "learning_rate": 0.0001, "elbo": 4.45034, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28442411447337557}
{"step": 4855, "loss": 4.450486373901367, "grad_norm": 0.9960321187973022, "learning_rate": 0.0001, "elbo": 4.45049, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.284717335209946}
{"step": 4860, "loss": 4.458564758300781, "grad_norm": 0.8703786730766296, "learning_rate": 0.0001, "elbo": 4.45856, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28501055594651653}
{"step": 4865, "loss": 4.434770202636718, "grad_norm": 0.920576274394989, "learning_rate": 0.0001, "elbo": 4.43477, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28530377668308704}
{"step": 4870, "loss": 4.443575668334961, "grad_norm": 0.895121693611145, "learning_rate": 0.0001, "elbo": 4.44358, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2855969974196575}
{"step": 4875, "loss": 4.453520202636719, "grad_norm": 0.9272449612617493, "learning_rate": 0.0001, "elbo": 4.45352, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.285890218156228}
{"step": 4880, "loss": 4.42967529296875, "grad_norm": 0.9973222613334656, "learning_rate": 0.0001, "elbo": 4.42968, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2861834388927985}
{"step": 4885, "loss": 4.449581146240234, "grad_norm": 0.964108943939209, "learning_rate": 0.0001, "elbo": 4.44958, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28647665962936897}
{"step": 4890, "loss": 4.456886291503906, "grad_norm": 0.8470417857170105, "learning_rate": 0.0001, "elbo": 4.45689, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2867698803659395}
{"step": 4895, "loss": 4.433383941650391, "grad_norm": 0.7920133471488953, "learning_rate": 0.0001, "elbo": 4.43338, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28706310110251}
{"step": 4900, "loss": 4.4344940185546875, "grad_norm": 0.9457386136054993, "learning_rate": 0.0001, "elbo": 4.43449, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28735632183908044}
{"step": 4900, "loss": 4.4344940185546875, "grad_norm": 0.9457386136054993, "learning_rate": 0.0001, "elbo": 4.43449, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28735632183908044}
{"step": 4905, "loss": 4.428269958496093, "grad_norm": 1.0038992166519165, "learning_rate": 0.0001, "elbo": 4.42827, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28764954257565095}
{"step": 4910, "loss": 4.4533546447753904, "grad_norm": 1.0337226390838623, "learning_rate": 0.0001, "elbo": 4.45335, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28794276331222146}
{"step": 4915, "loss": 4.421560668945313, "grad_norm": 0.8360466361045837, "learning_rate": 0.0001, "elbo": 4.42156, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2882359840487919}
{"step": 4920, "loss": 4.450141143798828, "grad_norm": 0.8332361578941345, "learning_rate": 0.0001, "elbo": 4.45014, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2885292047853624}
{"step": 4925, "loss": 4.426342010498047, "grad_norm": 0.9973518252372742, "learning_rate": 0.0001, "elbo": 4.42634, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28882242552193294}
{"step": 4930, "loss": 4.436605834960938, "grad_norm": 1.0428515672683716, "learning_rate": 0.0001, "elbo": 4.43661, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2891156462585034}
{"step": 4935, "loss": 4.434675216674805, "grad_norm": 0.9714522957801819, "learning_rate": 0.0001, "elbo": 4.43468, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2894088669950739}
{"step": 4940, "loss": 4.437723159790039, "grad_norm": 0.9254006147384644, "learning_rate": 0.0001, "elbo": 4.43772, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2897020877316444}
{"step": 4945, "loss": 4.43240966796875, "grad_norm": 1.070839524269104, "learning_rate": 0.0001, "elbo": 4.43241, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.28999530846821486}
{"step": 4950, "loss": 4.439204406738281, "grad_norm": 0.8781822919845581, "learning_rate": 0.0001, "elbo": 4.4392, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2902885292047854}
{"step": 4955, "loss": 4.440623474121094, "grad_norm": 0.8733159303665161, "learning_rate": 0.0001, "elbo": 4.44062, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2905817499413558}
{"step": 4960, "loss": 4.421413421630859, "grad_norm": 0.906715989112854, "learning_rate": 0.0001, "elbo": 4.42141, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.29087497067792634}
{"step": 4965, "loss": 4.425850296020508, "grad_norm": 0.8744743466377258, "learning_rate": 0.0001, "elbo": 4.42585, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.29116819141449685}
{"step": 4965, "loss": 4.425850296020508, "grad_norm": 0.8744743466377258, "learning_rate": 0.0001, "elbo": 4.42585, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.29116819141449685}
{"step": 4970, "loss": 4.439791870117188, "grad_norm": 1.088161826133728, "learning_rate": 0.0001, "elbo": 4.43979, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2914614121510673}
{"step": 4975, "loss": 4.4250732421875, "grad_norm": 0.9426880478858948, "learning_rate": 0.0001, "elbo": 4.42507, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2917546328876378}
{"step": 4980, "loss": 4.434076309204102, "grad_norm": 0.8735845685005188, "learning_rate": 0.0001, "elbo": 4.43408, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2920478536242083}
{"step": 4985, "loss": 4.413339996337891, "grad_norm": 1.085872769355774, "learning_rate": 0.0001, "elbo": 4.41334, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2923410743607788}
{"step": 4990, "loss": 4.427012252807617, "grad_norm": 0.9272135496139526, "learning_rate": 0.0001, "elbo": 4.42701, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2926342950973493}
{"step": 4995, "loss": 4.4302112579345705, "grad_norm": 1.0114262104034424, "learning_rate": 0.0001, "elbo": 4.43021, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2929275158339198}
{"step": 5000, "loss": 4.422574615478515, "grad_norm": 0.8960594534873962, "learning_rate": 0.0001, "elbo": 4.42257, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.29322073657049025}
{"step": 5005, "loss": 4.428549957275391, "grad_norm": 0.9702987670898438, "learning_rate": 0.0001, "elbo": 4.42855, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.29351395730706076}
{"step": 5010, "loss": 4.430634307861328, "grad_norm": 0.8874711394309998, "learning_rate": 0.0001, "elbo": 4.43063, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.29380717804363127}
{"step": 5010, "loss": 4.430634307861328, "grad_norm": 0.8874711394309998, "learning_rate": 0.0001, "elbo": 4.43063, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.29380717804363127}
{"step": 5015, "loss": 4.427690887451172, "grad_norm": 0.8691120743751526, "learning_rate": 0.0001, "elbo": 4.42769, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2941003987802017}
{"step": 5020, "loss": 4.438381958007812, "grad_norm": 0.9339315295219421, "learning_rate": 0.0001, "elbo": 4.43838, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.29439361951677223}
{"step": 5025, "loss": 4.431410217285157, "grad_norm": 0.8502923250198364, "learning_rate": 0.0001, "elbo": 4.43141, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.29468684025334274}
{"step": 5030, "loss": 4.428723907470703, "grad_norm": 0.8675495386123657, "learning_rate": 0.0001, "elbo": 4.42872, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2949800609899132}
{"step": 5035, "loss": 4.431962585449218, "grad_norm": 0.9055271744728088, "learning_rate": 0.0001, "elbo": 4.43196, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2952732817264837}
{"step": 5040, "loss": 4.422634887695312, "grad_norm": 1.0076851844787598, "learning_rate": 0.0001, "elbo": 4.42264, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2955665024630542}
{"step": 5045, "loss": 4.40551986694336, "grad_norm": 0.8449249267578125, "learning_rate": 0.0001, "elbo": 4.40552, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.29585972319962467}
{"step": 5050, "loss": 4.42484130859375, "grad_norm": 0.9297534227371216, "learning_rate": 0.0001, "elbo": 4.42484, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2961529439361952}
{"step": 5055, "loss": 4.436268615722656, "grad_norm": 0.9173672795295715, "learning_rate": 0.0001, "elbo": 4.43627, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.29644616467276563}
{"step": 5060, "loss": 4.426990127563476, "grad_norm": 0.9369417428970337, "learning_rate": 0.0001, "elbo": 4.42699, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.29673938540933614}
{"step": 5065, "loss": 4.412259674072265, "grad_norm": 0.8590895533561707, "learning_rate": 0.0001, "elbo": 4.41226, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.29703260614590665}
{"step": 5070, "loss": 4.418246459960938, "grad_norm": 1.000563383102417, "learning_rate": 0.0001, "elbo": 4.41825, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2973258268824771}
{"step": 5075, "loss": 4.411395645141601, "grad_norm": 0.9308990240097046, "learning_rate": 0.0001, "elbo": 4.4114, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2976190476190476}
{"step": 5080, "loss": 4.4161632537841795, "grad_norm": 0.9467957615852356, "learning_rate": 0.0001, "elbo": 4.41616, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2979122683556181}
{"step": 5085, "loss": 4.402112579345703, "grad_norm": 0.9349578619003296, "learning_rate": 0.0001, "elbo": 4.40211, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2982054890921886}
{"step": 5090, "loss": 4.392820739746094, "grad_norm": 0.9292928576469421, "learning_rate": 0.0001, "elbo": 4.39282, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2984987098287591}
{"step": 5095, "loss": 4.404512786865235, "grad_norm": 0.8526135087013245, "learning_rate": 0.0001, "elbo": 4.40451, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2987919305653296}
{"step": 5100, "loss": 4.401319885253907, "grad_norm": 1.0350043773651123, "learning_rate": 0.0001, "elbo": 4.40132, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.29908515130190005}
{"step": 5105, "loss": 4.402824401855469, "grad_norm": 0.9166789054870605, "learning_rate": 0.0001, "elbo": 4.40282, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.29937837203847056}
{"step": 5110, "loss": 4.41046028137207, "grad_norm": 0.8692349195480347, "learning_rate": 0.0001, "elbo": 4.41046, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2996715927750411}
{"step": 5115, "loss": 4.417889022827149, "grad_norm": 1.0129705667495728, "learning_rate": 0.0001, "elbo": 4.41789, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.2999648135116115}
{"step": 5120, "loss": 4.409715270996093, "grad_norm": 1.0532625913619995, "learning_rate": 0.0001, "elbo": 4.40972, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30025803424818204}
{"step": 5125, "loss": 4.413873291015625, "grad_norm": 1.0345733165740967, "learning_rate": 0.0001, "elbo": 4.41387, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30055125498475255}
{"step": 5130, "loss": 4.416123962402343, "grad_norm": 0.8907435536384583, "learning_rate": 0.0001, "elbo": 4.41612, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.300844475721323}
{"step": 5135, "loss": 4.4146263122558596, "grad_norm": 0.8548767566680908, "learning_rate": 0.0001, "elbo": 4.41463, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3011376964578935}
{"step": 5140, "loss": 4.406088256835938, "grad_norm": 0.9202634692192078, "learning_rate": 0.0001, "elbo": 4.40609, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30143091719446397}
{"step": 5145, "loss": 4.400989532470703, "grad_norm": 1.007171392440796, "learning_rate": 0.0001, "elbo": 4.40099, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3017241379310345}
{"step": 5150, "loss": 4.399717712402344, "grad_norm": 0.9256280660629272, "learning_rate": 0.0001, "elbo": 4.39972, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.302017358667605}
{"step": 5155, "loss": 4.404558563232422, "grad_norm": 0.8718470335006714, "learning_rate": 0.0001, "elbo": 4.40456, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30231057940417544}
{"step": 5160, "loss": 4.4047996520996096, "grad_norm": 0.888069748878479, "learning_rate": 0.0001, "elbo": 4.4048, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30260380014074595}
{"step": 5165, "loss": 4.397315979003906, "grad_norm": 0.9560965895652771, "learning_rate": 0.0001, "elbo": 4.39732, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30289702087731646}
{"step": 5165, "loss": 4.397315979003906, "grad_norm": 0.9560965895652771, "learning_rate": 0.0001, "elbo": 4.39732, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30289702087731646}
{"step": 5170, "loss": 4.4014122009277346, "grad_norm": 0.8686624765396118, "learning_rate": 0.0001, "elbo": 4.40141, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3031902416138869}
{"step": 5170, "loss": 4.4014122009277346, "grad_norm": 0.8686624765396118, "learning_rate": 0.0001, "elbo": 4.40141, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3031902416138869}
{"step": 5175, "loss": 4.380520629882812, "grad_norm": 0.8921920657157898, "learning_rate": 0.0001, "elbo": 4.38052, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3034834623504574}
{"step": 5180, "loss": 4.389118957519531, "grad_norm": 0.8462498188018799, "learning_rate": 0.0001, "elbo": 4.38912, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30377668308702793}
{"step": 5185, "loss": 4.400332641601563, "grad_norm": 1.0841615200042725, "learning_rate": 0.0001, "elbo": 4.40033, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3040699038235984}
{"step": 5190, "loss": 4.402812194824219, "grad_norm": 1.09038507938385, "learning_rate": 0.0001, "elbo": 4.40281, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3043631245601689}
{"step": 5195, "loss": 4.4058837890625, "grad_norm": 0.9650253653526306, "learning_rate": 0.0001, "elbo": 4.40588, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3046563452967394}
{"step": 5200, "loss": 4.415076828002929, "grad_norm": 1.5665408372879028, "learning_rate": 0.0001, "elbo": 4.41508, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30494956603330986}
{"step": 5205, "loss": 4.388642120361328, "grad_norm": 0.8797205686569214, "learning_rate": 0.0001, "elbo": 4.38864, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30524278676988037}
{"step": 5210, "loss": 4.3989715576171875, "grad_norm": 0.8925445675849915, "learning_rate": 0.0001, "elbo": 4.39897, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3055360075064509}
{"step": 5215, "loss": 4.39404067993164, "grad_norm": 1.2971657514572144, "learning_rate": 0.0001, "elbo": 4.39404, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30582922824302133}
{"step": 5220, "loss": 4.399996185302735, "grad_norm": 1.0102696418762207, "learning_rate": 0.0001, "elbo": 4.4, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30612244897959184}
{"step": 5225, "loss": 4.394789123535157, "grad_norm": 0.8843942284584045, "learning_rate": 0.0001, "elbo": 4.39479, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30641566971616235}
{"step": 5230, "loss": 4.405742645263672, "grad_norm": 0.9230180978775024, "learning_rate": 0.0001, "elbo": 4.40574, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3067088904527328}
{"step": 5235, "loss": 4.3987476348876955, "grad_norm": 0.8771287202835083, "learning_rate": 0.0001, "elbo": 4.39875, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3070021111893033}
{"step": 5240, "loss": 4.388455963134765, "grad_norm": 0.907324492931366, "learning_rate": 0.0001, "elbo": 4.38846, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30729533192587377}
{"step": 5245, "loss": 4.378498840332031, "grad_norm": 0.9243960976600647, "learning_rate": 0.0001, "elbo": 4.3785, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3075885526624443}
{"step": 5250, "loss": 4.389712142944336, "grad_norm": 0.8032333254814148, "learning_rate": 0.0001, "elbo": 4.38971, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3078817733990148}
{"step": 5255, "loss": 4.391575622558594, "grad_norm": 1.0400197505950928, "learning_rate": 0.0001, "elbo": 4.39158, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30817499413558525}
{"step": 5260, "loss": 4.394805526733398, "grad_norm": 0.9385921955108643, "learning_rate": 0.0001, "elbo": 4.39481, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30846821487215575}
{"step": 5265, "loss": 4.3982421875, "grad_norm": 0.9183088541030884, "learning_rate": 0.0001, "elbo": 4.39824, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30876143560872626}
{"step": 5270, "loss": 4.393403625488281, "grad_norm": 0.9436502456665039, "learning_rate": 0.0001, "elbo": 4.3934, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3090546563452967}
{"step": 5275, "loss": 4.387872314453125, "grad_norm": 0.9166738986968994, "learning_rate": 0.0001, "elbo": 4.38787, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30934787708186723}
{"step": 5280, "loss": 4.389641189575196, "grad_norm": 0.98926842212677, "learning_rate": 0.0001, "elbo": 4.38964, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30964109781843774}
{"step": 5280, "loss": 4.389641189575196, "grad_norm": 0.98926842212677, "learning_rate": 0.0001, "elbo": 4.38964, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.30964109781843774}
{"step": 5285, "loss": 4.371294403076172, "grad_norm": 0.9476403594017029, "learning_rate": 0.0001, "elbo": 4.37129, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3099343185550082}
{"step": 5290, "loss": 4.386578750610352, "grad_norm": 1.224324345588684, "learning_rate": 0.0001, "elbo": 4.38658, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3102275392915787}
{"step": 5295, "loss": 4.389901351928711, "grad_norm": 1.0064677000045776, "learning_rate": 0.0001, "elbo": 4.3899, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3105207600281492}
{"step": 5300, "loss": 4.378177261352539, "grad_norm": 0.8690823316574097, "learning_rate": 0.0001, "elbo": 4.37818, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.31081398076471967}
{"step": 5305, "loss": 4.391236877441406, "grad_norm": 0.8781641125679016, "learning_rate": 0.0001, "elbo": 4.39124, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3111072015012902}
{"step": 5310, "loss": 4.398254776000977, "grad_norm": 1.0180190801620483, "learning_rate": 0.0001, "elbo": 4.39825, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3114004222378607}
{"step": 5315, "loss": 4.376791381835938, "grad_norm": 0.9442299604415894, "learning_rate": 0.0001, "elbo": 4.37679, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.31169364297443114}
{"step": 5320, "loss": 4.39398193359375, "grad_norm": 0.9662157297134399, "learning_rate": 0.0001, "elbo": 4.39398, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.31198686371100165}
{"step": 5325, "loss": 4.407920837402344, "grad_norm": 0.9579958915710449, "learning_rate": 0.0001, "elbo": 4.40792, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.31228008444757216}
{"step": 5330, "loss": 4.418427276611328, "grad_norm": 1.0345944166183472, "learning_rate": 0.0001, "elbo": 4.41843, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3125733051841426}
{"step": 5335, "loss": 4.378160858154297, "grad_norm": 0.8343212008476257, "learning_rate": 0.0001, "elbo": 4.37816, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3128665259207131}
{"step": 5340, "loss": 4.3801319122314455, "grad_norm": 0.8622703552246094, "learning_rate": 0.0001, "elbo": 4.38013, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3131597466572836}
{"step": 5345, "loss": 4.368029403686523, "grad_norm": 1.1737589836120605, "learning_rate": 0.0001, "elbo": 4.36803, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3134529673938541}
{"step": 5350, "loss": 4.378059387207031, "grad_norm": 0.8920882344245911, "learning_rate": 0.0001, "elbo": 4.37806, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3137461881304246}
{"step": 5355, "loss": 4.3701019287109375, "grad_norm": 0.8935408592224121, "learning_rate": 0.0001, "elbo": 4.3701, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.31403940886699505}
{"step": 5360, "loss": 4.3647407531738285, "grad_norm": 0.978373110294342, "learning_rate": 0.0001, "elbo": 4.36474, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.31433262960356556}
{"step": 5360, "loss": 4.3647407531738285, "grad_norm": 0.978373110294342, "learning_rate": 0.0001, "elbo": 4.36474, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.31433262960356556}
{"step": 5365, "loss": 4.386107635498047, "grad_norm": 0.8745178580284119, "learning_rate": 0.0001, "elbo": 4.38611, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.31462585034013607}
{"step": 5370, "loss": 4.374513244628906, "grad_norm": 0.942106306552887, "learning_rate": 0.0001, "elbo": 4.37451, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3149190710767065}
{"step": 5375, "loss": 4.389498138427735, "grad_norm": 0.9006301164627075, "learning_rate": 0.0001, "elbo": 4.3895, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.31521229181327703}
{"step": 5380, "loss": 4.381805801391602, "grad_norm": 0.9233251810073853, "learning_rate": 0.0001, "elbo": 4.38181, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.31550551254984754}
{"step": 5385, "loss": 4.37243537902832, "grad_norm": 1.0098687410354614, "learning_rate": 0.0001, "elbo": 4.37244, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.315798733286418}
{"step": 5390, "loss": 4.3724311828613285, "grad_norm": 1.0296989679336548, "learning_rate": 0.0001, "elbo": 4.37243, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3160919540229885}
{"step": 5395, "loss": 4.3940788269042965, "grad_norm": 0.980127215385437, "learning_rate": 0.0001, "elbo": 4.39408, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.316385174759559}
{"step": 5400, "loss": 4.386929321289062, "grad_norm": 0.9401847720146179, "learning_rate": 0.0001, "elbo": 4.38693, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3166783954961295}
{"step": 5405, "loss": 4.375178527832031, "grad_norm": 0.9971193671226501, "learning_rate": 0.0001, "elbo": 4.37518, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3169716162327}
{"step": 5410, "loss": 4.385352325439453, "grad_norm": 0.8796749114990234, "learning_rate": 0.0001, "elbo": 4.38535, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3172648369692705}
{"step": 5415, "loss": 4.3868560791015625, "grad_norm": 0.880456268787384, "learning_rate": 0.0001, "elbo": 4.38686, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.31755805770584095}
{"step": 5420, "loss": 4.3938850402832035, "grad_norm": 1.1306966543197632, "learning_rate": 0.0001, "elbo": 4.39388, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.31785127844241146}
{"step": 5425, "loss": 4.36277961730957, "grad_norm": 1.0715217590332031, "learning_rate": 0.0001, "elbo": 4.36278, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3181444991789819}
{"step": 5430, "loss": 4.365885925292969, "grad_norm": 0.9239332675933838, "learning_rate": 0.0001, "elbo": 4.36589, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3184377199155524}
{"step": 5435, "loss": 4.353737258911133, "grad_norm": 1.0807466506958008, "learning_rate": 0.0001, "elbo": 4.35374, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.31873094065212293}
{"step": 5440, "loss": 4.371803283691406, "grad_norm": 0.9974966049194336, "learning_rate": 0.0001, "elbo": 4.3718, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3190241613886934}
{"step": 5445, "loss": 4.380418395996093, "grad_norm": 1.0106163024902344, "learning_rate": 0.0001, "elbo": 4.38042, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3193173821252639}
{"step": 5450, "loss": 4.373342132568359, "grad_norm": 1.1738054752349854, "learning_rate": 0.0001, "elbo": 4.37334, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3196106028618344}
{"step": 5455, "loss": 4.37288818359375, "grad_norm": 0.8392961025238037, "learning_rate": 0.0001, "elbo": 4.37289, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.31990382359840486}
{"step": 5460, "loss": 4.383453369140625, "grad_norm": 1.4445981979370117, "learning_rate": 0.0001, "elbo": 4.38345, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.32019704433497537}
{"step": 5465, "loss": 4.364987182617187, "grad_norm": 0.9954307675361633, "learning_rate": 0.0001, "elbo": 4.36499, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3204902650715459}
{"step": 5470, "loss": 4.3753211975097654, "grad_norm": 0.8724679350852966, "learning_rate": 0.0001, "elbo": 4.37532, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.32078348580811633}
{"step": 5475, "loss": 4.3728782653808596, "grad_norm": 0.9100583791732788, "learning_rate": 0.0001, "elbo": 4.37288, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.32107670654468684}
{"step": 5480, "loss": 4.356813430786133, "grad_norm": 0.9884727001190186, "learning_rate": 0.0001, "elbo": 4.35681, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.32136992728125735}
{"step": 5485, "loss": 4.355451202392578, "grad_norm": 0.8528525233268738, "learning_rate": 0.0001, "elbo": 4.35545, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3216631480178278}
{"step": 5490, "loss": 4.365425872802734, "grad_norm": 0.9612634181976318, "learning_rate": 0.0001, "elbo": 4.36543, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3219563687543983}
{"step": 5495, "loss": 4.359165191650391, "grad_norm": 0.8772268295288086, "learning_rate": 0.0001, "elbo": 4.35917, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3222495894909688}
{"step": 5500, "loss": 4.354281234741211, "grad_norm": 0.9621481895446777, "learning_rate": 0.0001, "elbo": 4.35428, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3225428102275393}
{"step": 5505, "loss": 4.3604881286621096, "grad_norm": 1.001129150390625, "learning_rate": 0.0001, "elbo": 4.36049, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3228360309641098}
{"step": 5510, "loss": 4.3779254913330075, "grad_norm": 1.0035622119903564, "learning_rate": 0.0001, "elbo": 4.37793, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3231292517006803}
{"step": 5515, "loss": 4.385901641845703, "grad_norm": 0.9608007669448853, "learning_rate": 0.0001, "elbo": 4.3859, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.32342247243725075}
{"step": 5520, "loss": 4.362286758422852, "grad_norm": 0.9309964776039124, "learning_rate": 0.0001, "elbo": 4.36229, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.32371569317382126}
{"step": 5525, "loss": 4.389976119995117, "grad_norm": 0.8964813947677612, "learning_rate": 0.0001, "elbo": 4.38998, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3240089139103917}
{"step": 5530, "loss": 4.368750762939453, "grad_norm": 1.0196552276611328, "learning_rate": 0.0001, "elbo": 4.36875, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3243021346469622}
{"step": 5535, "loss": 4.368207931518555, "grad_norm": 0.9680610299110413, "learning_rate": 0.0001, "elbo": 4.36821, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.32459535538353274}
{"step": 5540, "loss": 4.385219573974609, "grad_norm": 1.1552526950836182, "learning_rate": 0.0001, "elbo": 4.38522, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3248885761201032}
{"step": 5545, "loss": 4.367339324951172, "grad_norm": 0.9009314179420471, "learning_rate": 0.0001, "elbo": 4.36734, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3251817968566737}
{"step": 5550, "loss": 4.354313659667969, "grad_norm": 0.8921080827713013, "learning_rate": 0.0001, "elbo": 4.35431, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3254750175932442}
{"step": 5555, "loss": 4.370719909667969, "grad_norm": 1.0370354652404785, "learning_rate": 0.0001, "elbo": 4.37072, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.32576823832981466}
{"step": 5560, "loss": 4.376369476318359, "grad_norm": 0.9723401665687561, "learning_rate": 0.0001, "elbo": 4.37637, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3260614590663852}
{"step": 5565, "loss": 4.377158355712891, "grad_norm": 0.9685486555099487, "learning_rate": 0.0001, "elbo": 4.37716, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3263546798029557}
{"step": 5570, "loss": 4.362525558471679, "grad_norm": 0.8889334797859192, "learning_rate": 0.0001, "elbo": 4.36253, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.32664790053952614}
{"step": 5575, "loss": 4.362755584716797, "grad_norm": 1.0737717151641846, "learning_rate": 0.0001, "elbo": 4.36276, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.32694112127609665}
{"step": 5580, "loss": 4.375237655639649, "grad_norm": 0.8618814945220947, "learning_rate": 0.0001, "elbo": 4.37524, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.32723434201266716}
{"step": 5585, "loss": 4.363181304931641, "grad_norm": 1.047988772392273, "learning_rate": 0.0001, "elbo": 4.36318, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3275275627492376}
{"step": 5590, "loss": 4.346797943115234, "grad_norm": 0.8337668180465698, "learning_rate": 0.0001, "elbo": 4.3468, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3278207834858081}
{"step": 5595, "loss": 4.3469280242919925, "grad_norm": 0.8829246163368225, "learning_rate": 0.0001, "elbo": 4.34693, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.32811400422237863}
{"step": 5600, "loss": 4.351654815673828, "grad_norm": 0.9384025931358337, "learning_rate": 0.0001, "elbo": 4.35165, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3284072249589491}
{"step": 5605, "loss": 4.353910064697265, "grad_norm": 0.9658961892127991, "learning_rate": 0.0001, "elbo": 4.35391, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3287004456955196}
{"step": 5610, "loss": 4.368572235107422, "grad_norm": 0.9274353384971619, "learning_rate": 0.0001, "elbo": 4.36857, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3289936664320901}
{"step": 5615, "loss": 4.3502960205078125, "grad_norm": 0.8862936496734619, "learning_rate": 0.0001, "elbo": 4.3503, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.32928688716866056}
{"step": 5620, "loss": 4.363735961914062, "grad_norm": 0.924312949180603, "learning_rate": 0.0001, "elbo": 4.36374, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.32958010790523107}
{"step": 5625, "loss": 4.364941024780274, "grad_norm": 0.790682852268219, "learning_rate": 0.0001, "elbo": 4.36494, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3298733286418015}
{"step": 5630, "loss": 4.3519237518310545, "grad_norm": 0.939403235912323, "learning_rate": 0.0001, "elbo": 4.35192, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.33016654937837203}
{"step": 5635, "loss": 4.339649200439453, "grad_norm": 0.9166695475578308, "learning_rate": 0.0001, "elbo": 4.33965, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.33045977011494254}
{"step": 5640, "loss": 4.357931900024414, "grad_norm": 0.8676535487174988, "learning_rate": 0.0001, "elbo": 4.35793, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.330752990851513}
{"step": 5645, "loss": 4.350003051757812, "grad_norm": 1.0463283061981201, "learning_rate": 0.0001, "elbo": 4.35, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3310462115880835}
{"step": 5650, "loss": 4.358033752441406, "grad_norm": 0.8235794305801392, "learning_rate": 0.0001, "elbo": 4.35803, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.331339432324654}
{"step": 5655, "loss": 4.366607666015625, "grad_norm": 0.8210904598236084, "learning_rate": 0.0001, "elbo": 4.36661, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.33163265306122447}
{"step": 5660, "loss": 4.344841766357422, "grad_norm": 0.9661358594894409, "learning_rate": 0.0001, "elbo": 4.34484, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.331925873797795}
{"step": 5665, "loss": 4.350035858154297, "grad_norm": 0.9125949740409851, "learning_rate": 0.0001, "elbo": 4.35004, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3322190945343655}
{"step": 5665, "loss": 4.350035858154297, "grad_norm": 0.9125949740409851, "learning_rate": 0.0001, "elbo": 4.35004, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3322190945343655}
{"step": 5670, "loss": 4.373481750488281, "grad_norm": 0.9688924551010132, "learning_rate": 0.0001, "elbo": 4.37348, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.33251231527093594}
{"step": 5670, "loss": 4.373481750488281, "grad_norm": 0.9688924551010132, "learning_rate": 0.0001, "elbo": 4.37348, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.33251231527093594}
{"step": 5675, "loss": 4.348561477661133, "grad_norm": 0.8438032865524292, "learning_rate": 0.0001, "elbo": 4.34856, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.33280553600750645}
{"step": 5680, "loss": 4.356617736816406, "grad_norm": 0.9078202843666077, "learning_rate": 0.0001, "elbo": 4.35662, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.33309875674407696}
{"step": 5685, "loss": 4.3524658203125, "grad_norm": 0.9667680263519287, "learning_rate": 0.0001, "elbo": 4.35247, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3333919774806474}
{"step": 5690, "loss": 4.348857116699219, "grad_norm": 0.9346603155136108, "learning_rate": 0.0001, "elbo": 4.34886, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3336851982172179}
{"step": 5695, "loss": 4.334992218017578, "grad_norm": 0.9228404760360718, "learning_rate": 0.0001, "elbo": 4.33499, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.33397841895378844}
{"step": 5700, "loss": 4.346273803710938, "grad_norm": 0.8395447731018066, "learning_rate": 0.0001, "elbo": 4.34627, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3342716396903589}
{"step": 5705, "loss": 4.3387596130371096, "grad_norm": 0.8757764101028442, "learning_rate": 0.0001, "elbo": 4.33876, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3345648604269294}
{"step": 5710, "loss": 4.335503387451172, "grad_norm": 0.8456771373748779, "learning_rate": 0.0001, "elbo": 4.3355, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3348580811634999}
{"step": 5715, "loss": 4.334674072265625, "grad_norm": 0.9407454133033752, "learning_rate": 0.0001, "elbo": 4.33467, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.33515130190007036}
{"step": 5720, "loss": 4.361714172363281, "grad_norm": 0.9572494626045227, "learning_rate": 0.0001, "elbo": 4.36171, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3354445226366409}
{"step": 5725, "loss": 4.357909774780273, "grad_norm": 0.8795923590660095, "learning_rate": 0.0001, "elbo": 4.35791, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.33573774337321133}
{"step": 5730, "loss": 4.333259201049804, "grad_norm": 0.8686926960945129, "learning_rate": 0.0001, "elbo": 4.33326, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.33603096410978184}
{"step": 5735, "loss": 4.335382461547852, "grad_norm": 1.1042221784591675, "learning_rate": 0.0001, "elbo": 4.33538, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.33632418484635235}
{"step": 5740, "loss": 4.342060852050781, "grad_norm": 0.8828523755073547, "learning_rate": 0.0001, "elbo": 4.34206, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3366174055829228}
{"step": 5745, "loss": 4.356159210205078, "grad_norm": 0.9206140041351318, "learning_rate": 0.0001, "elbo": 4.35616, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3369106263194933}
{"step": 5750, "loss": 4.341014099121094, "grad_norm": 1.0260564088821411, "learning_rate": 0.0001, "elbo": 4.34101, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3372038470560638}
{"step": 5755, "loss": 4.350099182128906, "grad_norm": 1.0626106262207031, "learning_rate": 0.0001, "elbo": 4.3501, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3374970677926343}
{"step": 5760, "loss": 4.326971817016601, "grad_norm": 0.953473687171936, "learning_rate": 0.0001, "elbo": 4.32697, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3377902885292048}
{"step": 5765, "loss": 4.370954132080078, "grad_norm": 0.9224706292152405, "learning_rate": 0.0001, "elbo": 4.37095, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3380835092657753}
{"step": 5770, "loss": 4.348226165771484, "grad_norm": 0.9409202933311462, "learning_rate": 0.0001, "elbo": 4.34823, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.33837673000234575}
{"step": 5775, "loss": 4.347708892822266, "grad_norm": 0.9274499416351318, "learning_rate": 0.0001, "elbo": 4.34771, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.33866995073891626}
{"step": 5780, "loss": 4.340666580200195, "grad_norm": 0.8518363237380981, "learning_rate": 0.0001, "elbo": 4.34067, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.33896317147548677}
{"step": 5785, "loss": 4.334851837158203, "grad_norm": 0.8241580724716187, "learning_rate": 0.0001, "elbo": 4.33485, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3392563922120572}
{"step": 5790, "loss": 4.327935028076172, "grad_norm": 0.9948921799659729, "learning_rate": 0.0001, "elbo": 4.32794, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.33954961294862773}
{"step": 5795, "loss": 4.328939437866211, "grad_norm": 0.8965212106704712, "learning_rate": 0.0001, "elbo": 4.32894, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.33984283368519824}
{"step": 5800, "loss": 4.333981323242187, "grad_norm": 0.9189440608024597, "learning_rate": 0.0001, "elbo": 4.33398, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3401360544217687}
{"step": 5805, "loss": 4.341570663452148, "grad_norm": 0.8552932143211365, "learning_rate": 0.0001, "elbo": 4.34157, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3404292751583392}
{"step": 5810, "loss": 4.3477928161621096, "grad_norm": 0.9306387901306152, "learning_rate": 0.0001, "elbo": 4.34779, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.34072249589490966}
{"step": 5815, "loss": 4.347163391113281, "grad_norm": 0.8833119869232178, "learning_rate": 0.0001, "elbo": 4.34716, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.34101571663148017}
{"step": 5820, "loss": 4.346043395996094, "grad_norm": 0.880719780921936, "learning_rate": 0.0001, "elbo": 4.34604, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3413089373680507}
{"step": 5825, "loss": 4.333560943603516, "grad_norm": 0.8876187205314636, "learning_rate": 0.0001, "elbo": 4.33356, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.34160215810462113}
{"step": 5830, "loss": 4.334189224243164, "grad_norm": 0.9714201092720032, "learning_rate": 0.0001, "elbo": 4.33419, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.34189537884119164}
{"step": 5835, "loss": 4.344195556640625, "grad_norm": 0.890422523021698, "learning_rate": 0.0001, "elbo": 4.3442, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.34218859957776215}
{"step": 5835, "loss": 4.344195556640625, "grad_norm": 0.890422523021698, "learning_rate": 0.0001, "elbo": 4.3442, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.34218859957776215}
{"step": 5840, "loss": 4.323964309692383, "grad_norm": 0.9689633250236511, "learning_rate": 0.0001, "elbo": 4.32396, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3424818203143326}
{"step": 5845, "loss": 4.322160339355468, "grad_norm": 0.8550013899803162, "learning_rate": 0.0001, "elbo": 4.32216, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3427750410509031}
{"step": 5850, "loss": 4.326084899902344, "grad_norm": 0.8656849265098572, "learning_rate": 0.0001, "elbo": 4.32608, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3430682617874736}
{"step": 5855, "loss": 4.342846298217774, "grad_norm": 0.8732386231422424, "learning_rate": 0.0001, "elbo": 4.34285, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3433614825240441}
{"step": 5860, "loss": 4.310198974609375, "grad_norm": 0.8797048330307007, "learning_rate": 0.0001, "elbo": 4.3102, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3436547032606146}
{"step": 5865, "loss": 4.320690155029297, "grad_norm": 0.8065237402915955, "learning_rate": 0.0001, "elbo": 4.32069, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3439479239971851}
{"step": 5870, "loss": 4.335799789428711, "grad_norm": 0.9958323240280151, "learning_rate": 0.0001, "elbo": 4.3358, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.34424114473375556}
{"step": 5875, "loss": 4.322121429443359, "grad_norm": 1.236547589302063, "learning_rate": 0.0001, "elbo": 4.32212, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.34453436547032606}
{"step": 5880, "loss": 4.341778182983399, "grad_norm": 0.9051525592803955, "learning_rate": 0.0001, "elbo": 4.34178, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3448275862068966}
{"step": 5885, "loss": 4.33911247253418, "grad_norm": 0.9274961352348328, "learning_rate": 0.0001, "elbo": 4.33911, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.34512080694346703}
{"step": 5890, "loss": 4.329252624511719, "grad_norm": 1.0971282720565796, "learning_rate": 0.0001, "elbo": 4.32925, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.34541402768003754}
{"step": 5895, "loss": 4.3207237243652346, "grad_norm": 0.8487752676010132, "learning_rate": 0.0001, "elbo": 4.32072, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.34570724841660805}
{"step": 5900, "loss": 4.312459945678711, "grad_norm": 0.8817336559295654, "learning_rate": 0.0001, "elbo": 4.31246, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3460004691531785}
{"step": 5905, "loss": 4.326115417480469, "grad_norm": 1.225534200668335, "learning_rate": 0.0001, "elbo": 4.32612, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.346293689889749}
{"step": 5910, "loss": 4.315342712402344, "grad_norm": 0.9577856659889221, "learning_rate": 0.0001, "elbo": 4.31534, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.34658691062631947}
{"step": 5915, "loss": 4.3262886047363285, "grad_norm": 0.8631490468978882, "learning_rate": 0.0001, "elbo": 4.32629, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.34688013136289}
{"step": 5920, "loss": 4.323106384277343, "grad_norm": 0.8369814157485962, "learning_rate": 0.0001, "elbo": 4.32311, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3471733520994605}
{"step": 5925, "loss": 4.320539855957032, "grad_norm": 0.9914078116416931, "learning_rate": 0.0001, "elbo": 4.32054, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.34746657283603094}
{"step": 5930, "loss": 4.321520233154297, "grad_norm": 1.0575957298278809, "learning_rate": 0.0001, "elbo": 4.32152, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.34775979357260145}
{"step": 5935, "loss": 4.3225563049316404, "grad_norm": 0.9019129276275635, "learning_rate": 0.0001, "elbo": 4.32256, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.34805301430917196}
{"step": 5940, "loss": 4.331454849243164, "grad_norm": 0.8817641139030457, "learning_rate": 0.0001, "elbo": 4.33145, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3483462350457424}
{"step": 5945, "loss": 4.324812698364258, "grad_norm": 0.8133848309516907, "learning_rate": 0.0001, "elbo": 4.32481, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3486394557823129}
{"step": 5950, "loss": 4.308796691894531, "grad_norm": 0.8277784585952759, "learning_rate": 0.0001, "elbo": 4.3088, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.34893267651888343}
{"step": 5955, "loss": 4.314208984375, "grad_norm": 0.8734506368637085, "learning_rate": 0.0001, "elbo": 4.31421, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3492258972554539}
{"step": 5960, "loss": 4.325275421142578, "grad_norm": 0.9420191049575806, "learning_rate": 0.0001, "elbo": 4.32528, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3495191179920244}
{"step": 5965, "loss": 4.304213333129883, "grad_norm": 0.8100919723510742, "learning_rate": 0.0001, "elbo": 4.30421, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3498123387285949}
{"step": 5970, "loss": 4.327786636352539, "grad_norm": 0.8994249105453491, "learning_rate": 0.0001, "elbo": 4.32779, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.35010555946516536}
{"step": 5975, "loss": 4.321086120605469, "grad_norm": 1.103353500366211, "learning_rate": 0.0001, "elbo": 4.32109, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.35039878020173587}
{"step": 5980, "loss": 4.328816604614258, "grad_norm": 0.9245221614837646, "learning_rate": 0.0001, "elbo": 4.32882, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3506920009383064}
{"step": 5985, "loss": 4.317247772216797, "grad_norm": 0.8546449542045593, "learning_rate": 0.0001, "elbo": 4.31725, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.35098522167487683}
{"step": 5990, "loss": 4.321479797363281, "grad_norm": 0.9320574402809143, "learning_rate": 0.0001, "elbo": 4.32148, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.35127844241144734}
{"step": 5995, "loss": 4.314209365844727, "grad_norm": 0.9041662216186523, "learning_rate": 0.0001, "elbo": 4.31421, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.35157166314801785}
{"step": 6000, "loss": 4.3198799133300785, "grad_norm": 0.8734085559844971, "learning_rate": 0.0001, "elbo": 4.31988, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3518648838845883}
{"step": 6005, "loss": 4.332905578613281, "grad_norm": 0.9195017218589783, "learning_rate": 0.0001, "elbo": 4.33291, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3521581046211588}
{"step": 6010, "loss": 4.322621917724609, "grad_norm": 0.9307588934898376, "learning_rate": 0.0001, "elbo": 4.32262, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3524513253577293}
{"step": 6015, "loss": 4.318489837646484, "grad_norm": 0.8304792642593384, "learning_rate": 0.0001, "elbo": 4.31849, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3527445460942998}
{"step": 6020, "loss": 4.3318641662597654, "grad_norm": 0.9451255798339844, "learning_rate": 0.0001, "elbo": 4.33186, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3530377668308703}
{"step": 6025, "loss": 4.318289184570313, "grad_norm": 0.9911568760871887, "learning_rate": 0.0001, "elbo": 4.31829, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.35333098756744075}
{"step": 6030, "loss": 4.313410949707031, "grad_norm": 0.823268473148346, "learning_rate": 0.0001, "elbo": 4.31341, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.35362420830401126}
{"step": 6035, "loss": 4.306256103515625, "grad_norm": 0.8117833137512207, "learning_rate": 0.0001, "elbo": 4.30626, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.35391742904058177}
{"step": 6040, "loss": 4.299287414550781, "grad_norm": 0.8462064266204834, "learning_rate": 0.0001, "elbo": 4.29929, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3542106497771522}
{"step": 6045, "loss": 4.30966796875, "grad_norm": 1.4566173553466797, "learning_rate": 0.0001, "elbo": 4.30967, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.35450387051372273}
{"step": 6050, "loss": 4.317223358154297, "grad_norm": 1.0511102676391602, "learning_rate": 0.0001, "elbo": 4.31722, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.35479709125029324}
{"step": 6055, "loss": 4.308264923095703, "grad_norm": 0.833744466304779, "learning_rate": 0.0001, "elbo": 4.30826, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3550903119868637}
{"step": 6060, "loss": 4.313686370849609, "grad_norm": 0.825183629989624, "learning_rate": 0.0001, "elbo": 4.31369, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3553835327234342}
{"step": 6065, "loss": 4.310315322875977, "grad_norm": 1.2380883693695068, "learning_rate": 0.0001, "elbo": 4.31032, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3556767534600047}
{"step": 6070, "loss": 4.3039390563964846, "grad_norm": 0.9496084451675415, "learning_rate": 0.0001, "elbo": 4.30394, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.35596997419657517}
{"step": 6075, "loss": 4.311515045166016, "grad_norm": 0.8360791802406311, "learning_rate": 0.0001, "elbo": 4.31152, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3562631949331457}
{"step": 6080, "loss": 4.3151100158691404, "grad_norm": 0.8124875426292419, "learning_rate": 0.0001, "elbo": 4.31511, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3565564156697162}
{"step": 6085, "loss": 4.304819107055664, "grad_norm": 0.8626351952552795, "learning_rate": 0.0001, "elbo": 4.30482, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.35684963640628664}
{"step": 6090, "loss": 4.302798461914063, "grad_norm": 0.8086729645729065, "learning_rate": 0.0001, "elbo": 4.3028, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.35714285714285715}
{"step": 6095, "loss": 4.308356475830078, "grad_norm": 0.8511576652526855, "learning_rate": 0.0001, "elbo": 4.30836, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.35743607787942766}
{"step": 6100, "loss": 4.292833709716797, "grad_norm": 1.008961796760559, "learning_rate": 0.0001, "elbo": 4.29283, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3577292986159981}
{"step": 6105, "loss": 4.31072883605957, "grad_norm": 0.9049668312072754, "learning_rate": 0.0001, "elbo": 4.31073, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3580225193525686}
{"step": 6110, "loss": 4.31384162902832, "grad_norm": 0.8427786231040955, "learning_rate": 0.0001, "elbo": 4.31384, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3583157400891391}
{"step": 6115, "loss": 4.301853942871094, "grad_norm": 1.008734107017517, "learning_rate": 0.0001, "elbo": 4.30185, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3586089608257096}
{"step": 6120, "loss": 4.29986686706543, "grad_norm": 0.8508639335632324, "learning_rate": 0.0001, "elbo": 4.29987, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3589021815622801}
{"step": 6125, "loss": 4.29742202758789, "grad_norm": 0.8606888055801392, "learning_rate": 0.0001, "elbo": 4.29742, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.35919540229885055}
{"step": 6130, "loss": 4.310703277587891, "grad_norm": 1.0063892602920532, "learning_rate": 0.0001, "elbo": 4.3107, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.35948862303542106}
{"step": 6135, "loss": 4.290863037109375, "grad_norm": 0.9480385780334473, "learning_rate": 0.0001, "elbo": 4.29086, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.35978184377199157}
{"step": 6140, "loss": 4.321770477294922, "grad_norm": 0.974643349647522, "learning_rate": 0.0001, "elbo": 4.32177, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.360075064508562}
{"step": 6145, "loss": 4.3200531005859375, "grad_norm": 1.295767068862915, "learning_rate": 0.0001, "elbo": 4.32005, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.36036828524513254}
{"step": 6150, "loss": 4.3105815887451175, "grad_norm": 0.9312281608581543, "learning_rate": 0.0001, "elbo": 4.31058, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.36066150598170305}
{"step": 6155, "loss": 4.303075790405273, "grad_norm": 0.9253210425376892, "learning_rate": 0.0001, "elbo": 4.30308, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3609547267182735}
{"step": 6160, "loss": 4.297353363037109, "grad_norm": 1.10887610912323, "learning_rate": 0.0001, "elbo": 4.29735, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.361247947454844}
{"step": 6165, "loss": 4.295002746582031, "grad_norm": 0.8345480561256409, "learning_rate": 0.0001, "elbo": 4.295, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3615411681914145}
{"step": 6170, "loss": 4.298644256591797, "grad_norm": 0.7867035269737244, "learning_rate": 0.0001, "elbo": 4.29864, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.361834388927985}
{"step": 6175, "loss": 4.313805389404297, "grad_norm": 0.9219383001327515, "learning_rate": 0.0001, "elbo": 4.31381, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3621276096645555}
{"step": 6180, "loss": 4.291008377075196, "grad_norm": 0.9785344004631042, "learning_rate": 0.0001, "elbo": 4.29101, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.362420830401126}
{"step": 6185, "loss": 4.287350845336914, "grad_norm": 0.9022655487060547, "learning_rate": 0.0001, "elbo": 4.28735, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.36271405113769645}
{"step": 6190, "loss": 4.293446350097656, "grad_norm": 1.0503774881362915, "learning_rate": 0.0001, "elbo": 4.29345, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.36300727187426696}
{"step": 6195, "loss": 4.313095474243164, "grad_norm": 0.7772877812385559, "learning_rate": 0.0001, "elbo": 4.3131, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3633004926108374}
{"step": 6195, "loss": 4.313095474243164, "grad_norm": 0.7772877812385559, "learning_rate": 0.0001, "elbo": 4.3131, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3633004926108374}
{"step": 6200, "loss": 4.29273681640625, "grad_norm": 0.9076658487319946, "learning_rate": 0.0001, "elbo": 4.29274, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3635937133474079}
{"step": 6205, "loss": 4.30046272277832, "grad_norm": 0.8482488989830017, "learning_rate": 0.0001, "elbo": 4.30046, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.36388693408397843}
{"step": 6210, "loss": 4.291542053222656, "grad_norm": 0.9409318566322327, "learning_rate": 0.0001, "elbo": 4.29154, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3641801548205489}
{"step": 6215, "loss": 4.290142822265625, "grad_norm": 0.8203776478767395, "learning_rate": 0.0001, "elbo": 4.29014, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3644733755571194}
{"step": 6220, "loss": 4.299540710449219, "grad_norm": 0.8573862910270691, "learning_rate": 0.0001, "elbo": 4.29954, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3647665962936899}
{"step": 6225, "loss": 4.3084968566894535, "grad_norm": 0.9116848111152649, "learning_rate": 0.0001, "elbo": 4.3085, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.36505981703026036}
{"step": 6230, "loss": 4.287424468994141, "grad_norm": 0.8536362051963806, "learning_rate": 0.0001, "elbo": 4.28742, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.36535303776683087}
{"step": 6235, "loss": 4.282019805908203, "grad_norm": 0.9590116143226624, "learning_rate": 0.0001, "elbo": 4.28202, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3656462585034014}
{"step": 6240, "loss": 4.309794998168945, "grad_norm": 0.8404724597930908, "learning_rate": 0.0001, "elbo": 4.3098, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.36593947923997183}
{"step": 6245, "loss": 4.288531112670898, "grad_norm": 0.9168916940689087, "learning_rate": 0.0001, "elbo": 4.28853, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.36623269997654234}
{"step": 6250, "loss": 4.292399215698242, "grad_norm": 1.0496281385421753, "learning_rate": 0.0001, "elbo": 4.2924, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.36652592071311285}
{"step": 6255, "loss": 4.291933059692383, "grad_norm": 0.8754566311836243, "learning_rate": 0.0001, "elbo": 4.29193, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3668191414496833}
{"step": 6260, "loss": 4.287898254394531, "grad_norm": 0.8059777617454529, "learning_rate": 0.0001, "elbo": 4.2879, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3671123621862538}
{"step": 6265, "loss": 4.2985481262207035, "grad_norm": 0.816800057888031, "learning_rate": 0.0001, "elbo": 4.29855, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3674055829228243}
{"step": 6270, "loss": 4.291597366333008, "grad_norm": 0.8802399635314941, "learning_rate": 0.0001, "elbo": 4.2916, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3676988036593948}
{"step": 6275, "loss": 4.283325958251953, "grad_norm": 0.9302679300308228, "learning_rate": 0.0001, "elbo": 4.28333, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3679920243959653}
{"step": 6280, "loss": 4.301357269287109, "grad_norm": 1.2148367166519165, "learning_rate": 0.0001, "elbo": 4.30136, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3682852451325358}
{"step": 6285, "loss": 4.3000938415527346, "grad_norm": 0.8696272373199463, "learning_rate": 0.0001, "elbo": 4.30009, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.36857846586910625}
{"step": 6290, "loss": 4.298143768310547, "grad_norm": 0.9147844910621643, "learning_rate": 0.0001, "elbo": 4.29814, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.36887168660567676}
{"step": 6295, "loss": 4.301969146728515, "grad_norm": 0.8592822551727295, "learning_rate": 0.0001, "elbo": 4.30197, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3691649073422472}
{"step": 6300, "loss": 4.300576019287109, "grad_norm": 0.8304126262664795, "learning_rate": 0.0001, "elbo": 4.30058, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3694581280788177}
{"step": 6305, "loss": 4.291870880126953, "grad_norm": 0.8885568380355835, "learning_rate": 0.0001, "elbo": 4.29187, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.36975134881538824}
{"step": 6310, "loss": 4.312479400634766, "grad_norm": 1.113840103149414, "learning_rate": 0.0001, "elbo": 4.31248, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3700445695519587}
{"step": 6315, "loss": 4.292678451538086, "grad_norm": 0.8640848994255066, "learning_rate": 0.0001, "elbo": 4.29268, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3703377902885292}
{"step": 6320, "loss": 4.297267913818359, "grad_norm": 0.7984692454338074, "learning_rate": 0.0001, "elbo": 4.29727, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3706310110250997}
{"step": 6325, "loss": 4.280239105224609, "grad_norm": 0.86545729637146, "learning_rate": 0.0001, "elbo": 4.28024, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.37092423176167016}
{"step": 6330, "loss": 4.281786346435547, "grad_norm": 1.0021425485610962, "learning_rate": 0.0001, "elbo": 4.28179, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3712174524982407}
{"step": 6335, "loss": 4.29915771484375, "grad_norm": 0.8711860775947571, "learning_rate": 0.0001, "elbo": 4.29916, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3715106732348112}
{"step": 6340, "loss": 4.283679962158203, "grad_norm": 0.9416987299919128, "learning_rate": 0.0001, "elbo": 4.28368, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.37180389397138164}
{"step": 6345, "loss": 4.277528762817383, "grad_norm": 0.8645340204238892, "learning_rate": 0.0001, "elbo": 4.27753, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.37209711470795215}
{"step": 6350, "loss": 4.298361206054688, "grad_norm": 0.8784101009368896, "learning_rate": 0.0001, "elbo": 4.29836, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.37239033544452266}
{"step": 6355, "loss": 4.288435745239258, "grad_norm": 0.8826510906219482, "learning_rate": 0.0001, "elbo": 4.28844, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3726835561810931}
{"step": 6360, "loss": 4.291518020629883, "grad_norm": 0.8570727705955505, "learning_rate": 0.0001, "elbo": 4.29152, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3729767769176636}
{"step": 6365, "loss": 4.275915145874023, "grad_norm": 0.84145587682724, "learning_rate": 0.0001, "elbo": 4.27592, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.37326999765423413}
{"step": 6370, "loss": 4.2860057830810545, "grad_norm": 0.8674665689468384, "learning_rate": 0.0001, "elbo": 4.28601, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3735632183908046}
{"step": 6375, "loss": 4.290818405151367, "grad_norm": 0.8623324632644653, "learning_rate": 0.0001, "elbo": 4.29082, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3738564391273751}
{"step": 6380, "loss": 4.278475952148438, "grad_norm": 0.7525646686553955, "learning_rate": 0.0001, "elbo": 4.27848, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3741496598639456}
{"step": 6385, "loss": 4.268251037597656, "grad_norm": 0.9212021231651306, "learning_rate": 0.0001, "elbo": 4.26825, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.37444288060051606}
{"step": 6390, "loss": 4.294443511962891, "grad_norm": 0.9132865071296692, "learning_rate": 0.0001, "elbo": 4.29444, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.37473610133708657}
{"step": 6395, "loss": 4.281442260742187, "grad_norm": 0.8489581942558289, "learning_rate": 0.0001, "elbo": 4.28144, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.375029322073657}
{"step": 6395, "loss": 4.281442260742187, "grad_norm": 0.8489581942558289, "learning_rate": 0.0001, "elbo": 4.28144, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.375029322073657}
{"step": 6400, "loss": 4.3003894805908205, "grad_norm": 0.8985033631324768, "learning_rate": 0.0001, "elbo": 4.30039, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.37532254281022753}
{"step": 6405, "loss": 4.288031768798828, "grad_norm": 0.8064484000205994, "learning_rate": 0.0001, "elbo": 4.28803, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.37561576354679804}
{"step": 6410, "loss": 4.270404815673828, "grad_norm": 0.8798331022262573, "learning_rate": 0.0001, "elbo": 4.27041, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3759089842833685}
{"step": 6415, "loss": 4.276762390136719, "grad_norm": 0.8461848497390747, "learning_rate": 0.0001, "elbo": 4.27676, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.376202205019939}
{"step": 6420, "loss": 4.276455688476562, "grad_norm": 0.9364176392555237, "learning_rate": 0.0001, "elbo": 4.27646, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3764954257565095}
{"step": 6425, "loss": 4.25711669921875, "grad_norm": 0.891755223274231, "learning_rate": 0.0001, "elbo": 4.25712, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.37678864649307997}
{"step": 6430, "loss": 4.275737762451172, "grad_norm": 0.9935751557350159, "learning_rate": 0.0001, "elbo": 4.27574, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3770818672296505}
{"step": 6435, "loss": 4.258276748657226, "grad_norm": 0.9406741857528687, "learning_rate": 0.0001, "elbo": 4.25828, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.377375087966221}
{"step": 6440, "loss": 4.264933013916016, "grad_norm": 0.9699382781982422, "learning_rate": 0.0001, "elbo": 4.26493, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.37766830870279144}
{"step": 6445, "loss": 4.268426895141602, "grad_norm": 0.8494886755943298, "learning_rate": 0.0001, "elbo": 4.26843, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.37796152943936195}
{"step": 6450, "loss": 4.2618358612060545, "grad_norm": 0.9147095084190369, "learning_rate": 0.0001, "elbo": 4.26184, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.37825475017593246}
{"step": 6455, "loss": 4.272365951538086, "grad_norm": 0.873744785785675, "learning_rate": 0.0001, "elbo": 4.27237, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3785479709125029}
{"step": 6460, "loss": 4.278227615356445, "grad_norm": 0.8729603886604309, "learning_rate": 0.0001, "elbo": 4.27823, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3788411916490734}
{"step": 6465, "loss": 4.259767532348633, "grad_norm": 1.0217896699905396, "learning_rate": 0.0001, "elbo": 4.25977, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.37913441238564394}
{"step": 6470, "loss": 4.267244338989258, "grad_norm": 0.8696024417877197, "learning_rate": 0.0001, "elbo": 4.26724, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3794276331222144}
{"step": 6475, "loss": 4.270637512207031, "grad_norm": 0.8262410759925842, "learning_rate": 0.0001, "elbo": 4.27064, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3797208538587849}
{"step": 6480, "loss": 4.261436462402344, "grad_norm": 0.8520125150680542, "learning_rate": 0.0001, "elbo": 4.26144, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3800140745953554}
{"step": 6480, "loss": 4.261436462402344, "grad_norm": 0.8520125150680542, "learning_rate": 0.0001, "elbo": 4.26144, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3800140745953554}
{"step": 6485, "loss": 4.275006103515625, "grad_norm": 0.8595845103263855, "learning_rate": 0.0001, "elbo": 4.27501, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.38030729533192587}
{"step": 6490, "loss": 4.2554786682128904, "grad_norm": 1.023453950881958, "learning_rate": 0.0001, "elbo": 4.25548, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3806005160684964}
{"step": 6490, "loss": 4.2554786682128904, "grad_norm": 1.023453950881958, "learning_rate": 0.0001, "elbo": 4.25548, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3806005160684964}
{"step": 6495, "loss": 4.278805541992187, "grad_norm": 1.036184549331665, "learning_rate": 0.0001, "elbo": 4.27881, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.38089373680506683}
{"step": 6500, "loss": 4.2631683349609375, "grad_norm": 1.1983736753463745, "learning_rate": 0.0001, "elbo": 4.26317, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.38118695754163734}
{"step": 6505, "loss": 4.279051208496094, "grad_norm": 0.9547120928764343, "learning_rate": 0.0001, "elbo": 4.27905, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.38148017827820785}
{"step": 6510, "loss": 4.277320098876953, "grad_norm": 1.179972767829895, "learning_rate": 0.0001, "elbo": 4.27732, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3817733990147783}
{"step": 6515, "loss": 4.271208572387695, "grad_norm": 0.834514319896698, "learning_rate": 0.0001, "elbo": 4.27121, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3820666197513488}
{"step": 6520, "loss": 4.269678115844727, "grad_norm": 0.9244512319564819, "learning_rate": 0.0001, "elbo": 4.26968, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3823598404879193}
{"step": 6525, "loss": 4.269001007080078, "grad_norm": 0.9346728324890137, "learning_rate": 0.0001, "elbo": 4.269, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3826530612244898}
{"step": 6530, "loss": 4.262934875488281, "grad_norm": 0.7610816955566406, "learning_rate": 0.0001, "elbo": 4.26294, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3829462819610603}
{"step": 6535, "loss": 4.274139404296875, "grad_norm": 1.0278397798538208, "learning_rate": 0.0001, "elbo": 4.27414, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3832395026976308}
{"step": 6540, "loss": 4.275493621826172, "grad_norm": 0.8394950032234192, "learning_rate": 0.0001, "elbo": 4.27549, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.38353272343420125}
{"step": 6545, "loss": 4.267266464233399, "grad_norm": 0.8772136569023132, "learning_rate": 0.0001, "elbo": 4.26727, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.38382594417077176}
{"step": 6550, "loss": 4.274371337890625, "grad_norm": 0.796963632106781, "learning_rate": 0.0001, "elbo": 4.27437, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.38411916490734227}
{"step": 6555, "loss": 4.245309066772461, "grad_norm": 0.9739511013031006, "learning_rate": 0.0001, "elbo": 4.24531, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3844123856439127}
{"step": 6560, "loss": 4.249686813354492, "grad_norm": 1.0650893449783325, "learning_rate": 0.0001, "elbo": 4.24969, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.38470560638048323}
{"step": 6565, "loss": 4.26055679321289, "grad_norm": 0.881364643573761, "learning_rate": 0.0001, "elbo": 4.26056, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.38499882711705374}
{"step": 6570, "loss": 4.250393676757812, "grad_norm": 0.842028796672821, "learning_rate": 0.0001, "elbo": 4.25039, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3852920478536242}
{"step": 6575, "loss": 4.289974975585937, "grad_norm": 0.9464851021766663, "learning_rate": 0.0001, "elbo": 4.28997, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3855852685901947}
{"step": 6580, "loss": 4.257446670532227, "grad_norm": 0.8217102289199829, "learning_rate": 0.0001, "elbo": 4.25745, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.38587848932676516}
{"step": 6585, "loss": 4.265399169921875, "grad_norm": 0.8280614018440247, "learning_rate": 0.0001, "elbo": 4.2654, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.38617171006333567}
{"step": 6590, "loss": 4.270102691650391, "grad_norm": 0.8530092835426331, "learning_rate": 0.0001, "elbo": 4.2701, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3864649307999062}
{"step": 6595, "loss": 4.253696441650391, "grad_norm": 0.7996034026145935, "learning_rate": 0.0001, "elbo": 4.2537, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.38675815153647664}
{"step": 6600, "loss": 4.269049072265625, "grad_norm": 1.1055560111999512, "learning_rate": 0.0001, "elbo": 4.26905, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.38705137227304715}
{"step": 6605, "loss": 4.259454727172852, "grad_norm": 0.9624217748641968, "learning_rate": 0.0001, "elbo": 4.25945, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.38734459300961765}
{"step": 6610, "loss": 4.268157958984375, "grad_norm": 1.0691369771957397, "learning_rate": 0.0001, "elbo": 4.26816, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3876378137461881}
{"step": 6615, "loss": 4.237999343872071, "grad_norm": 0.9525002241134644, "learning_rate": 0.0001, "elbo": 4.238, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3879310344827586}
{"step": 6620, "loss": 4.272549438476562, "grad_norm": 0.9594753384590149, "learning_rate": 0.0001, "elbo": 4.27255, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.38822425521932913}
{"step": 6625, "loss": 4.260515213012695, "grad_norm": 0.9455481171607971, "learning_rate": 0.0001, "elbo": 4.26052, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3885174759558996}
{"step": 6625, "loss": 4.260515213012695, "grad_norm": 0.9455481171607971, "learning_rate": 0.0001, "elbo": 4.26052, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3885174759558996}
{"step": 6630, "loss": 4.271750259399414, "grad_norm": 1.1619198322296143, "learning_rate": 0.0001, "elbo": 4.27175, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3888106966924701}
{"step": 6635, "loss": 4.2627006530761715, "grad_norm": 0.8937236666679382, "learning_rate": 0.0001, "elbo": 4.2627, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3891039174290406}
{"step": 6640, "loss": 4.2504230499267575, "grad_norm": 0.8076204061508179, "learning_rate": 0.0001, "elbo": 4.25042, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.38939713816561106}
{"step": 6645, "loss": 4.23974609375, "grad_norm": 0.96971195936203, "learning_rate": 0.0001, "elbo": 4.23975, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.38969035890218157}
{"step": 6650, "loss": 4.261549377441407, "grad_norm": 0.9643658399581909, "learning_rate": 0.0001, "elbo": 4.26155, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3899835796387521}
{"step": 6655, "loss": 4.24125862121582, "grad_norm": 0.8429256677627563, "learning_rate": 0.0001, "elbo": 4.24126, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39027680037532253}
{"step": 6660, "loss": 4.2542778015136715, "grad_norm": 0.9617050290107727, "learning_rate": 0.0001, "elbo": 4.25428, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39057002111189304}
{"step": 6665, "loss": 4.252489852905273, "grad_norm": 1.0221030712127686, "learning_rate": 0.0001, "elbo": 4.25249, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39086324184846355}
{"step": 6670, "loss": 4.246722412109375, "grad_norm": 0.831798255443573, "learning_rate": 0.0001, "elbo": 4.24672, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.391156462585034}
{"step": 6675, "loss": 4.272980499267578, "grad_norm": 0.7822600603103638, "learning_rate": 0.0001, "elbo": 4.27298, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3914496833216045}
{"step": 6680, "loss": 4.268906784057617, "grad_norm": 0.822212278842926, "learning_rate": 0.0001, "elbo": 4.26891, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39174290405817497}
{"step": 6685, "loss": 4.252684020996094, "grad_norm": 0.8077529668807983, "learning_rate": 0.0001, "elbo": 4.25268, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3920361247947455}
{"step": 6690, "loss": 4.239646911621094, "grad_norm": 0.8562494516372681, "learning_rate": 0.0001, "elbo": 4.23965, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.392329345531316}
{"step": 6695, "loss": 4.256363296508789, "grad_norm": 0.9477724432945251, "learning_rate": 0.0001, "elbo": 4.25636, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39262256626788644}
{"step": 6700, "loss": 4.2510520935058596, "grad_norm": 0.9418426156044006, "learning_rate": 0.0001, "elbo": 4.25105, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39291578700445695}
{"step": 6705, "loss": 4.252907562255859, "grad_norm": 0.8129106163978577, "learning_rate": 0.0001, "elbo": 4.25291, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39320900774102746}
{"step": 6710, "loss": 4.256035995483399, "grad_norm": 0.7977810502052307, "learning_rate": 0.0001, "elbo": 4.25604, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3935022284775979}
{"step": 6710, "loss": 4.256035995483399, "grad_norm": 0.7977810502052307, "learning_rate": 0.0001, "elbo": 4.25604, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3935022284775979}
{"step": 6715, "loss": 4.260253143310547, "grad_norm": 0.896023154258728, "learning_rate": 0.0001, "elbo": 4.26025, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3937954492141684}
{"step": 6720, "loss": 4.243950271606446, "grad_norm": 0.8537954688072205, "learning_rate": 0.0001, "elbo": 4.24395, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39408866995073893}
{"step": 6725, "loss": 4.2478080749511715, "grad_norm": 0.8903520107269287, "learning_rate": 0.0001, "elbo": 4.24781, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3943818906873094}
{"step": 6730, "loss": 4.2444805145263675, "grad_norm": 0.8541366457939148, "learning_rate": 0.0001, "elbo": 4.24448, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3946751114238799}
{"step": 6735, "loss": 4.2350303649902346, "grad_norm": 0.9660822749137878, "learning_rate": 0.0001, "elbo": 4.23503, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3949683321604504}
{"step": 6740, "loss": 4.259606170654297, "grad_norm": 0.8254265189170837, "learning_rate": 0.0001, "elbo": 4.25961, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39526155289702086}
{"step": 6745, "loss": 4.260099792480469, "grad_norm": 0.7943558692932129, "learning_rate": 0.0001, "elbo": 4.2601, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39555477363359137}
{"step": 6750, "loss": 4.255237579345703, "grad_norm": 0.8418757319450378, "learning_rate": 0.0001, "elbo": 4.25524, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3958479943701619}
{"step": 6755, "loss": 4.260415649414062, "grad_norm": 0.9244757890701294, "learning_rate": 0.0001, "elbo": 4.26042, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39614121510673234}
{"step": 6760, "loss": 4.2433349609375, "grad_norm": 1.082889199256897, "learning_rate": 0.0001, "elbo": 4.24334, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39643443584330285}
{"step": 6760, "loss": 4.2433349609375, "grad_norm": 1.082889199256897, "learning_rate": 0.0001, "elbo": 4.24334, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39643443584330285}
{"step": 6765, "loss": 4.249413681030274, "grad_norm": 0.8586344122886658, "learning_rate": 0.0001, "elbo": 4.24941, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39672765657987336}
{"step": 6770, "loss": 4.258990097045898, "grad_norm": 0.896302342414856, "learning_rate": 0.0001, "elbo": 4.25899, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3970208773164438}
{"step": 6775, "loss": 4.248018646240235, "grad_norm": 0.8492780923843384, "learning_rate": 0.0001, "elbo": 4.24802, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3973140980530143}
{"step": 6780, "loss": 4.2457733154296875, "grad_norm": 0.9144601821899414, "learning_rate": 0.0001, "elbo": 4.24577, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3976073187895848}
{"step": 6785, "loss": 4.243315887451172, "grad_norm": 0.8416526317596436, "learning_rate": 0.0001, "elbo": 4.24332, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3979005395261553}
{"step": 6790, "loss": 4.2374732971191404, "grad_norm": 0.8426079750061035, "learning_rate": 0.0001, "elbo": 4.23747, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3981937602627258}
{"step": 6795, "loss": 4.250822448730469, "grad_norm": 0.9093570113182068, "learning_rate": 0.0001, "elbo": 4.25082, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39848698099929625}
{"step": 6800, "loss": 4.251371765136719, "grad_norm": 0.839450478553772, "learning_rate": 0.0001, "elbo": 4.25137, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39878020173586676}
{"step": 6805, "loss": 4.246613311767578, "grad_norm": 0.8271737694740295, "learning_rate": 0.0001, "elbo": 4.24661, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39907342247243727}
{"step": 6810, "loss": 4.240675354003907, "grad_norm": 0.8624260425567627, "learning_rate": 0.0001, "elbo": 4.24068, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.3993666432090077}
{"step": 6815, "loss": 4.240090179443359, "grad_norm": 0.8409119248390198, "learning_rate": 0.0001, "elbo": 4.24009, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39965986394557823}
{"step": 6815, "loss": 4.240090179443359, "grad_norm": 0.8409119248390198, "learning_rate": 0.0001, "elbo": 4.24009, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39965986394557823}
{"step": 6820, "loss": 4.243825531005859, "grad_norm": 0.9605604410171509, "learning_rate": 0.0001, "elbo": 4.24383, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39995308468214874}
{"step": 6820, "loss": 4.243825531005859, "grad_norm": 0.9605604410171509, "learning_rate": 0.0001, "elbo": 4.24383, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.39995308468214874}
{"step": 6825, "loss": 4.2548774719238285, "grad_norm": 0.9362788796424866, "learning_rate": 0.0001, "elbo": 4.25488, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4002463054187192}
{"step": 6830, "loss": 4.246916580200195, "grad_norm": 0.8334392309188843, "learning_rate": 0.0001, "elbo": 4.24692, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4005395261552897}
{"step": 6835, "loss": 4.258654022216797, "grad_norm": 0.8525261878967285, "learning_rate": 0.0001, "elbo": 4.25865, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4008327468918602}
{"step": 6840, "loss": 4.256681823730469, "grad_norm": 0.864489734172821, "learning_rate": 0.0001, "elbo": 4.25668, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.40112596762843067}
{"step": 6845, "loss": 4.258134841918945, "grad_norm": 0.9731190800666809, "learning_rate": 0.0001, "elbo": 4.25813, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4014191883650012}
{"step": 6850, "loss": 4.250762176513672, "grad_norm": 0.8299580216407776, "learning_rate": 0.0001, "elbo": 4.25076, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4017124091015717}
{"step": 6855, "loss": 4.256723785400391, "grad_norm": 0.8044449687004089, "learning_rate": 0.0001, "elbo": 4.25672, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.40200562983814214}
{"step": 6860, "loss": 4.238737487792969, "grad_norm": 0.8394749164581299, "learning_rate": 0.0001, "elbo": 4.23874, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.40229885057471265}
{"step": 6865, "loss": 4.235372924804688, "grad_norm": 1.0531924962997437, "learning_rate": 0.0001, "elbo": 4.23537, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4025920713112831}
{"step": 6870, "loss": 4.24548110961914, "grad_norm": 0.8227145671844482, "learning_rate": 0.0001, "elbo": 4.24548, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4028852920478536}
{"step": 6875, "loss": 4.2401580810546875, "grad_norm": 0.8654863238334656, "learning_rate": 0.0001, "elbo": 4.24016, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4031785127844241}
{"step": 6880, "loss": 4.2533611297607425, "grad_norm": 0.858418345451355, "learning_rate": 0.0001, "elbo": 4.25336, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4034717335209946}
{"step": 6885, "loss": 4.2603199005126955, "grad_norm": 0.9632779359817505, "learning_rate": 0.0001, "elbo": 4.26032, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4037649542575651}
{"step": 6890, "loss": 4.236962890625, "grad_norm": 0.8356364369392395, "learning_rate": 0.0001, "elbo": 4.23696, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4040581749941356}
{"step": 6895, "loss": 4.243572235107422, "grad_norm": 0.8715144395828247, "learning_rate": 0.0001, "elbo": 4.24357, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.40435139573070605}
{"step": 6900, "loss": 4.242497253417969, "grad_norm": 0.852342963218689, "learning_rate": 0.0001, "elbo": 4.2425, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.40464461646727656}
{"step": 6905, "loss": 4.255740737915039, "grad_norm": 0.9466582536697388, "learning_rate": 0.0001, "elbo": 4.25574, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4049378372038471}
{"step": 6910, "loss": 4.242324447631836, "grad_norm": 0.8291256427764893, "learning_rate": 0.0001, "elbo": 4.24232, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4052310579404175}
{"step": 6915, "loss": 4.259708404541016, "grad_norm": 0.9461563229560852, "learning_rate": 0.0001, "elbo": 4.25971, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.40552427867698804}
{"step": 6920, "loss": 4.257173156738281, "grad_norm": 1.26850163936615, "learning_rate": 0.0001, "elbo": 4.25717, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.40581749941355855}
{"step": 6925, "loss": 4.230835342407227, "grad_norm": 0.8974547386169434, "learning_rate": 0.0001, "elbo": 4.23084, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.406110720150129}
{"step": 6930, "loss": 4.240644073486328, "grad_norm": 0.8021663427352905, "learning_rate": 0.0001, "elbo": 4.24064, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4064039408866995}
{"step": 6935, "loss": 4.241759490966797, "grad_norm": 0.9110661745071411, "learning_rate": 0.0001, "elbo": 4.24176, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.40669716162327}
{"step": 6940, "loss": 4.234934616088867, "grad_norm": 0.8687753081321716, "learning_rate": 0.0001, "elbo": 4.23493, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4069903823598405}
{"step": 6945, "loss": 4.240195083618164, "grad_norm": 0.8239843249320984, "learning_rate": 0.0001, "elbo": 4.2402, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.407283603096411}
{"step": 6950, "loss": 4.218852233886719, "grad_norm": 0.7863125801086426, "learning_rate": 0.0001, "elbo": 4.21885, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4075768238329815}
{"step": 6955, "loss": 4.240988159179688, "grad_norm": 0.7923346161842346, "learning_rate": 0.0001, "elbo": 4.24099, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.40787004456955195}
{"step": 6960, "loss": 4.2397712707519535, "grad_norm": 1.0051426887512207, "learning_rate": 0.0001, "elbo": 4.23977, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.40816326530612246}
{"step": 6965, "loss": 4.2446342468261715, "grad_norm": 0.9472318291664124, "learning_rate": 0.0001, "elbo": 4.24463, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4084564860426929}
{"step": 6970, "loss": 4.245270919799805, "grad_norm": 0.9452993273735046, "learning_rate": 0.0001, "elbo": 4.24527, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4087497067792634}
{"step": 6970, "loss": 4.245270919799805, "grad_norm": 0.9452993273735046, "learning_rate": 0.0001, "elbo": 4.24527, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4087497067792634}
{"step": 6975, "loss": 4.231680679321289, "grad_norm": 0.8978990912437439, "learning_rate": 0.0001, "elbo": 4.23168, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.40904292751583393}
{"step": 6980, "loss": 4.2480323791503904, "grad_norm": 0.8185480237007141, "learning_rate": 0.0001, "elbo": 4.24803, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4093361482524044}
{"step": 6985, "loss": 4.2184711456298825, "grad_norm": 0.8749507069587708, "learning_rate": 0.0001, "elbo": 4.21847, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4096293689889749}
{"step": 6985, "loss": 4.2184711456298825, "grad_norm": 0.8749507069587708, "learning_rate": 0.0001, "elbo": 4.21847, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4096293689889749}
{"step": 6990, "loss": 4.250006866455078, "grad_norm": 0.878084123134613, "learning_rate": 0.0001, "elbo": 4.25001, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4099225897255454}
{"step": 6990, "loss": 4.250006866455078, "grad_norm": 0.878084123134613, "learning_rate": 0.0001, "elbo": 4.25001, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4099225897255454}
{"step": 6995, "loss": 4.239102554321289, "grad_norm": 0.8527531027793884, "learning_rate": 0.0001, "elbo": 4.2391, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.41021581046211586}
{"step": 7000, "loss": 4.2343299865722654, "grad_norm": 1.4518340826034546, "learning_rate": 0.0001, "elbo": 4.23433, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.41050903119868637}
{"step": 7005, "loss": 4.239434814453125, "grad_norm": 1.0424916744232178, "learning_rate": 0.0001, "elbo": 4.23943, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4108022519352569}
{"step": 7010, "loss": 4.23516845703125, "grad_norm": 0.8273687958717346, "learning_rate": 0.0001, "elbo": 4.23517, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.41109547267182733}
{"step": 7015, "loss": 4.221843719482422, "grad_norm": 0.7947181463241577, "learning_rate": 0.0001, "elbo": 4.22184, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.41138869340839784}
{"step": 7020, "loss": 4.222918701171875, "grad_norm": 0.8982750177383423, "learning_rate": 0.0001, "elbo": 4.22292, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.41168191414496835}
{"step": 7025, "loss": 4.228099060058594, "grad_norm": 0.9513468742370605, "learning_rate": 0.0001, "elbo": 4.2281, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4119751348815388}
{"step": 7030, "loss": 4.240668869018554, "grad_norm": 0.834047257900238, "learning_rate": 0.0001, "elbo": 4.24067, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4122683556181093}
{"step": 7035, "loss": 4.2315727233886715, "grad_norm": 0.8924645781517029, "learning_rate": 0.0001, "elbo": 4.23157, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4125615763546798}
{"step": 7040, "loss": 4.230722427368164, "grad_norm": 0.9597987532615662, "learning_rate": 0.0001, "elbo": 4.23072, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4128547970912503}
{"step": 7045, "loss": 4.2356719970703125, "grad_norm": 0.822295069694519, "learning_rate": 0.0001, "elbo": 4.23567, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4131480178278208}
{"step": 7050, "loss": 4.239433288574219, "grad_norm": 0.8390294313430786, "learning_rate": 0.0001, "elbo": 4.23943, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4134412385643913}
{"step": 7055, "loss": 4.245598220825196, "grad_norm": 0.850812554359436, "learning_rate": 0.0001, "elbo": 4.2456, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.41373445930096175}
{"step": 7060, "loss": 4.239042663574219, "grad_norm": 0.7881056666374207, "learning_rate": 0.0001, "elbo": 4.23904, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.41402768003753226}
{"step": 7065, "loss": 4.232490539550781, "grad_norm": 0.9149858951568604, "learning_rate": 0.0001, "elbo": 4.23249, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4143209007741027}
{"step": 7070, "loss": 4.222879028320312, "grad_norm": 1.097380518913269, "learning_rate": 0.0001, "elbo": 4.22288, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.41461412151067323}
{"step": 7075, "loss": 4.225741958618164, "grad_norm": 0.859915554523468, "learning_rate": 0.0001, "elbo": 4.22574, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.41490734224724374}
{"step": 7080, "loss": 4.232703399658203, "grad_norm": 0.8642914891242981, "learning_rate": 0.0001, "elbo": 4.2327, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4152005629838142}
{"step": 7085, "loss": 4.230914306640625, "grad_norm": 0.8969066143035889, "learning_rate": 0.0001, "elbo": 4.23091, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4154937837203847}
{"step": 7090, "loss": 4.222644805908203, "grad_norm": 0.7944468855857849, "learning_rate": 0.0001, "elbo": 4.22264, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4157870044569552}
{"step": 7095, "loss": 4.226636123657227, "grad_norm": 1.1993260383605957, "learning_rate": 0.0001, "elbo": 4.22664, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.41608022519352567}
{"step": 7100, "loss": 4.2200569152832035, "grad_norm": 0.8070636987686157, "learning_rate": 0.0001, "elbo": 4.22006, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4163734459300962}
{"step": 7105, "loss": 4.241344451904297, "grad_norm": 0.7948938608169556, "learning_rate": 0.0001, "elbo": 4.24134, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4166666666666667}
{"step": 7110, "loss": 4.233437728881836, "grad_norm": 0.7282071709632874, "learning_rate": 0.0001, "elbo": 4.23344, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.41695988740323714}
{"step": 7115, "loss": 4.228651428222657, "grad_norm": 0.959193766117096, "learning_rate": 0.0001, "elbo": 4.22865, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.41725310813980765}
{"step": 7120, "loss": 4.2254585266113285, "grad_norm": 0.8159191608428955, "learning_rate": 0.0001, "elbo": 4.22546, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.41754632887637816}
{"step": 7125, "loss": 4.230293273925781, "grad_norm": 0.8369490504264832, "learning_rate": 0.0001, "elbo": 4.23029, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4178395496129486}
{"step": 7130, "loss": 4.214530181884766, "grad_norm": 0.7881821990013123, "learning_rate": 0.0001, "elbo": 4.21453, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4181327703495191}
{"step": 7135, "loss": 4.222506713867188, "grad_norm": 0.8409315347671509, "learning_rate": 0.0001, "elbo": 4.22251, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.41842599108608963}
{"step": 7140, "loss": 4.2037700653076175, "grad_norm": 0.7511770129203796, "learning_rate": 0.0001, "elbo": 4.20377, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4187192118226601}
{"step": 7145, "loss": 4.2249908447265625, "grad_norm": 0.8538911938667297, "learning_rate": 0.0001, "elbo": 4.22499, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4190124325592306}
{"step": 7150, "loss": 4.225154113769531, "grad_norm": 0.797991156578064, "learning_rate": 0.0001, "elbo": 4.22515, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4193056532958011}
{"step": 7155, "loss": 4.2213188171386715, "grad_norm": 0.9087567925453186, "learning_rate": 0.0001, "elbo": 4.22132, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.41959887403237156}
{"step": 7160, "loss": 4.2101997375488285, "grad_norm": 0.8998656868934631, "learning_rate": 0.0001, "elbo": 4.2102, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.41989209476894207}
{"step": 7165, "loss": 4.2176513671875, "grad_norm": 0.969825029373169, "learning_rate": 0.0001, "elbo": 4.21765, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4201853155055125}
{"step": 7170, "loss": 4.213871002197266, "grad_norm": 0.8472341299057007, "learning_rate": 0.0001, "elbo": 4.21387, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.42047853624208303}
{"step": 7175, "loss": 4.2223976135253904, "grad_norm": 0.8550902605056763, "learning_rate": 0.0001, "elbo": 4.2224, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.42077175697865354}
{"step": 7180, "loss": 4.217184448242188, "grad_norm": 0.8423354029655457, "learning_rate": 0.0001, "elbo": 4.21718, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.421064977715224}
{"step": 7185, "loss": 4.216533660888672, "grad_norm": 0.8279585242271423, "learning_rate": 0.0001, "elbo": 4.21653, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4213581984517945}
{"step": 7190, "loss": 4.224011993408203, "grad_norm": 0.8393847346305847, "learning_rate": 0.0001, "elbo": 4.22401, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.421651419188365}
{"step": 7195, "loss": 4.2179115295410154, "grad_norm": 0.8032644391059875, "learning_rate": 0.0001, "elbo": 4.21791, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.42194463992493547}
{"step": 7200, "loss": 4.22698745727539, "grad_norm": 0.8644391894340515, "learning_rate": 0.0001, "elbo": 4.22699, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.422237860661506}
{"step": 7205, "loss": 4.208614349365234, "grad_norm": 0.8342469930648804, "learning_rate": 0.0001, "elbo": 4.20861, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4225310813980765}
{"step": 7210, "loss": 4.187124633789063, "grad_norm": 0.8676016330718994, "learning_rate": 0.0001, "elbo": 4.18712, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.42282430213464695}
{"step": 7215, "loss": 4.207663345336914, "grad_norm": 0.8089556097984314, "learning_rate": 0.0001, "elbo": 4.20766, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.42311752287121746}
{"step": 7220, "loss": 4.212125015258789, "grad_norm": 0.9493081569671631, "learning_rate": 0.0001, "elbo": 4.21213, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.42341074360778796}
{"step": 7225, "loss": 4.202633285522461, "grad_norm": 0.8256261944770813, "learning_rate": 0.0001, "elbo": 4.20263, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4237039643443584}
{"step": 7230, "loss": 4.211189270019531, "grad_norm": 1.1562429666519165, "learning_rate": 0.0001, "elbo": 4.21119, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.42399718508092893}
{"step": 7235, "loss": 4.219256210327148, "grad_norm": 0.880875825881958, "learning_rate": 0.0001, "elbo": 4.21926, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.42429040581749944}
{"step": 7240, "loss": 4.217108535766601, "grad_norm": 1.0921841859817505, "learning_rate": 0.0001, "elbo": 4.21711, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4245836265540699}
{"step": 7245, "loss": 4.222117614746094, "grad_norm": 1.0358667373657227, "learning_rate": 0.0001, "elbo": 4.22212, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4248768472906404}
{"step": 7250, "loss": 4.206439208984375, "grad_norm": 0.8197948336601257, "learning_rate": 0.0001, "elbo": 4.20644, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.42517006802721086}
{"step": 7255, "loss": 4.202196884155273, "grad_norm": 0.8273170590400696, "learning_rate": 0.0001, "elbo": 4.2022, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.42546328876378137}
{"step": 7260, "loss": 4.218968200683594, "grad_norm": 0.8297354578971863, "learning_rate": 0.0001, "elbo": 4.21897, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4257565095003519}
{"step": 7265, "loss": 4.204374694824219, "grad_norm": 0.757856011390686, "learning_rate": 0.0001, "elbo": 4.20437, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.42604973023692233}
{"step": 7270, "loss": 4.213939666748047, "grad_norm": 0.8845436573028564, "learning_rate": 0.0001, "elbo": 4.21394, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.42634295097349284}
{"step": 7275, "loss": 4.221925735473633, "grad_norm": 0.8564050793647766, "learning_rate": 0.0001, "elbo": 4.22193, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.42663617171006335}
{"step": 7280, "loss": 4.210443496704102, "grad_norm": 0.9164201021194458, "learning_rate": 0.0001, "elbo": 4.21044, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4269293924466338}
{"step": 7285, "loss": 4.211296081542969, "grad_norm": 0.8386427760124207, "learning_rate": 0.0001, "elbo": 4.2113, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4272226131832043}
{"step": 7290, "loss": 4.216798782348633, "grad_norm": 0.7820401787757874, "learning_rate": 0.0001, "elbo": 4.2168, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4275158339197748}
{"step": 7295, "loss": 4.190030288696289, "grad_norm": 0.8309757709503174, "learning_rate": 0.0001, "elbo": 4.19003, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4278090546563453}
{"step": 7300, "loss": 4.218330383300781, "grad_norm": 0.8829749226570129, "learning_rate": 0.0001, "elbo": 4.21833, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4281022753929158}
{"step": 7300, "loss": 4.218330383300781, "grad_norm": 0.8829749226570129, "learning_rate": 0.0001, "elbo": 4.21833, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4281022753929158}
{"step": 7305, "loss": 4.2121734619140625, "grad_norm": 0.9094589948654175, "learning_rate": 0.0001, "elbo": 4.21217, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4283954961294863}
{"step": 7310, "loss": 4.216693878173828, "grad_norm": 0.9260194301605225, "learning_rate": 0.0001, "elbo": 4.21669, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.42868871686605675}
{"step": 7315, "loss": 4.217887115478516, "grad_norm": 0.8944380879402161, "learning_rate": 0.0001, "elbo": 4.21789, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.42898193760262726}
{"step": 7320, "loss": 4.219515609741211, "grad_norm": 0.8574396967887878, "learning_rate": 0.0001, "elbo": 4.21952, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.42927515833919777}
{"step": 7325, "loss": 4.207582092285156, "grad_norm": 0.8291810750961304, "learning_rate": 0.0001, "elbo": 4.20758, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4295683790757682}
{"step": 7330, "loss": 4.193492126464844, "grad_norm": 0.8392199277877808, "learning_rate": 0.0001, "elbo": 4.19349, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.42986159981233873}
{"step": 7335, "loss": 4.223683929443359, "grad_norm": 0.7978636622428894, "learning_rate": 0.0001, "elbo": 4.22368, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43015482054890924}
{"step": 7340, "loss": 4.199251937866211, "grad_norm": 0.7954457998275757, "learning_rate": 0.0001, "elbo": 4.19925, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4304480412854797}
{"step": 7345, "loss": 4.213938522338867, "grad_norm": 1.2324354648590088, "learning_rate": 0.0001, "elbo": 4.21394, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4307412620220502}
{"step": 7350, "loss": 4.210829162597657, "grad_norm": 0.7409840226173401, "learning_rate": 0.0001, "elbo": 4.21083, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43103448275862066}
{"step": 7355, "loss": 4.226516342163086, "grad_norm": 0.8655832409858704, "learning_rate": 0.0001, "elbo": 4.22652, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4313277034951912}
{"step": 7360, "loss": 4.206990432739258, "grad_norm": 0.8249728083610535, "learning_rate": 0.0001, "elbo": 4.20699, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4316209242317617}
{"step": 7365, "loss": 4.204444122314453, "grad_norm": 0.8818356394767761, "learning_rate": 0.0001, "elbo": 4.20444, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43191414496833214}
{"step": 7370, "loss": 4.186018371582032, "grad_norm": 0.8167677521705627, "learning_rate": 0.0001, "elbo": 4.18602, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43220736570490265}
{"step": 7370, "loss": 4.186018371582032, "grad_norm": 0.8167677521705627, "learning_rate": 0.0001, "elbo": 4.18602, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43220736570490265}
{"step": 7375, "loss": 4.2164253234863285, "grad_norm": 0.8384881615638733, "learning_rate": 0.0001, "elbo": 4.21643, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43250058644147316}
{"step": 7380, "loss": 4.1964569091796875, "grad_norm": 0.9827966094017029, "learning_rate": 0.0001, "elbo": 4.19646, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4327938071780436}
{"step": 7385, "loss": 4.184292221069336, "grad_norm": 0.8000251054763794, "learning_rate": 0.0001, "elbo": 4.18429, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4330870279146141}
{"step": 7390, "loss": 4.198336791992188, "grad_norm": 0.8047207593917847, "learning_rate": 0.0001, "elbo": 4.19834, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43338024865118463}
{"step": 7395, "loss": 4.2247051239013675, "grad_norm": 0.8409757018089294, "learning_rate": 0.0001, "elbo": 4.22471, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4336734693877551}
{"step": 7400, "loss": 4.219678497314453, "grad_norm": 0.8585604429244995, "learning_rate": 0.0001, "elbo": 4.21968, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4339666901243256}
{"step": 7405, "loss": 4.206784439086914, "grad_norm": 0.8758118152618408, "learning_rate": 0.0001, "elbo": 4.20678, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4342599108608961}
{"step": 7410, "loss": 4.193030548095703, "grad_norm": 0.8848378658294678, "learning_rate": 0.0001, "elbo": 4.19303, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43455313159746656}
{"step": 7415, "loss": 4.1983787536621096, "grad_norm": 0.8341296315193176, "learning_rate": 0.0001, "elbo": 4.19838, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43484635233403707}
{"step": 7420, "loss": 4.208625411987304, "grad_norm": 0.9119594693183899, "learning_rate": 0.0001, "elbo": 4.20863, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4351395730706076}
{"step": 7425, "loss": 4.1998748779296875, "grad_norm": 1.1006830930709839, "learning_rate": 0.0001, "elbo": 4.19987, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43543279380717803}
{"step": 7430, "loss": 4.183567810058594, "grad_norm": 0.830979585647583, "learning_rate": 0.0001, "elbo": 4.18357, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43572601454374854}
{"step": 7435, "loss": 4.200656890869141, "grad_norm": 0.8007174730300903, "learning_rate": 0.0001, "elbo": 4.20066, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43601923528031905}
{"step": 7440, "loss": 4.200690460205078, "grad_norm": 0.898400068283081, "learning_rate": 0.0001, "elbo": 4.20069, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4363124560168895}
{"step": 7445, "loss": 4.209716033935547, "grad_norm": 0.8928118348121643, "learning_rate": 0.0001, "elbo": 4.20972, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43660567675346}
{"step": 7450, "loss": 4.194747543334961, "grad_norm": 0.8927621245384216, "learning_rate": 0.0001, "elbo": 4.19475, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43689889749003047}
{"step": 7455, "loss": 4.203757476806641, "grad_norm": 0.8308244943618774, "learning_rate": 0.0001, "elbo": 4.20376, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.437192118226601}
{"step": 7460, "loss": 4.202238082885742, "grad_norm": 0.8254621028900146, "learning_rate": 0.0001, "elbo": 4.20224, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4374853389631715}
{"step": 7465, "loss": 4.208320236206054, "grad_norm": 0.8201037049293518, "learning_rate": 0.0001, "elbo": 4.20832, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43777855969974194}
{"step": 7470, "loss": 4.196811676025391, "grad_norm": 0.8840219974517822, "learning_rate": 0.0001, "elbo": 4.19681, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43807178043631245}
{"step": 7475, "loss": 4.203755569458008, "grad_norm": 0.818651020526886, "learning_rate": 0.0001, "elbo": 4.20376, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43836500117288296}
{"step": 7480, "loss": 4.219181823730469, "grad_norm": 0.8093113899230957, "learning_rate": 0.0001, "elbo": 4.21918, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4386582219094534}
{"step": 7480, "loss": 4.219181823730469, "grad_norm": 0.8093113899230957, "learning_rate": 0.0001, "elbo": 4.21918, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4386582219094534}
{"step": 7485, "loss": 4.203413009643555, "grad_norm": 0.8642663359642029, "learning_rate": 0.0001, "elbo": 4.20341, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4389514426460239}
{"step": 7490, "loss": 4.20709114074707, "grad_norm": 0.7777994871139526, "learning_rate": 0.0001, "elbo": 4.20709, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.43924466338259444}
{"step": 7495, "loss": 4.19367790222168, "grad_norm": 0.9939634203910828, "learning_rate": 0.0001, "elbo": 4.19368, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4395378841191649}
{"step": 7500, "loss": 4.201727294921875, "grad_norm": 0.8911154866218567, "learning_rate": 0.0001, "elbo": 4.20173, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4398311048557354}
{"step": 7505, "loss": 4.200843048095703, "grad_norm": 0.8023245334625244, "learning_rate": 0.0001, "elbo": 4.20084, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4401243255923059}
{"step": 7510, "loss": 4.18765869140625, "grad_norm": 0.8188135027885437, "learning_rate": 0.0001, "elbo": 4.18766, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.44041754632887636}
{"step": 7515, "loss": 4.204521179199219, "grad_norm": 0.9164714813232422, "learning_rate": 0.0001, "elbo": 4.20452, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4407107670654469}
{"step": 7520, "loss": 4.198455429077148, "grad_norm": 0.9115272164344788, "learning_rate": 0.0001, "elbo": 4.19846, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4410039878020174}
{"step": 7525, "loss": 4.2175453186035154, "grad_norm": 0.8952917456626892, "learning_rate": 0.0001, "elbo": 4.21755, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.44129720853858784}
{"step": 7530, "loss": 4.197547912597656, "grad_norm": 0.9467998147010803, "learning_rate": 0.0001, "elbo": 4.19755, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.44159042927515835}
{"step": 7535, "loss": 4.177705383300781, "grad_norm": 0.8656415939331055, "learning_rate": 0.0001, "elbo": 4.17771, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.44188365001172886}
{"step": 7540, "loss": 4.190095520019531, "grad_norm": 0.7828074097633362, "learning_rate": 0.0001, "elbo": 4.1901, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4421768707482993}
{"step": 7545, "loss": 4.207987213134766, "grad_norm": 0.8292473554611206, "learning_rate": 0.0001, "elbo": 4.20799, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4424700914848698}
{"step": 7545, "loss": 4.207987213134766, "grad_norm": 0.8292473554611206, "learning_rate": 0.0001, "elbo": 4.20799, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4424700914848698}
{"step": 7550, "loss": 4.192159271240234, "grad_norm": 0.8223305940628052, "learning_rate": 0.0001, "elbo": 4.19216, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4427633122214403}
{"step": 7555, "loss": 4.196892166137696, "grad_norm": 0.9493463635444641, "learning_rate": 0.0001, "elbo": 4.19689, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4430565329580108}
{"step": 7560, "loss": 4.181873321533203, "grad_norm": 1.108014702796936, "learning_rate": 0.0001, "elbo": 4.18187, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4433497536945813}
{"step": 7565, "loss": 4.215614318847656, "grad_norm": 0.8598500490188599, "learning_rate": 0.0001, "elbo": 4.21561, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.44364297443115175}
{"step": 7570, "loss": 4.197008514404297, "grad_norm": 0.8231614232063293, "learning_rate": 0.0001, "elbo": 4.19701, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.44393619516772226}
{"step": 7575, "loss": 4.194722747802734, "grad_norm": 1.0023353099822998, "learning_rate": 0.0001, "elbo": 4.19472, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.44422941590429277}
{"step": 7580, "loss": 4.193488693237304, "grad_norm": 0.902878999710083, "learning_rate": 0.0001, "elbo": 4.19349, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4445226366408632}
{"step": 7585, "loss": 4.19695930480957, "grad_norm": 0.9317202568054199, "learning_rate": 0.0001, "elbo": 4.19696, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.44481585737743373}
{"step": 7590, "loss": 4.1898658752441404, "grad_norm": 1.409597396850586, "learning_rate": 0.0001, "elbo": 4.18987, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.44510907811400424}
{"step": 7595, "loss": 4.194254684448242, "grad_norm": 0.8179757595062256, "learning_rate": 0.0001, "elbo": 4.19425, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4454022988505747}
{"step": 7600, "loss": 4.195890426635742, "grad_norm": 0.8739061951637268, "learning_rate": 0.0001, "elbo": 4.19589, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4456955195871452}
{"step": 7605, "loss": 4.198229217529297, "grad_norm": 0.8642066717147827, "learning_rate": 0.0001, "elbo": 4.19823, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4459887403237157}
{"step": 7610, "loss": 4.181244659423828, "grad_norm": 0.8555581569671631, "learning_rate": 0.0001, "elbo": 4.18124, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.44628196106028617}
{"step": 7615, "loss": 4.191977691650391, "grad_norm": 0.912563145160675, "learning_rate": 0.0001, "elbo": 4.19198, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4465751817968567}
{"step": 7620, "loss": 4.188880157470703, "grad_norm": 0.9647259712219238, "learning_rate": 0.0001, "elbo": 4.18888, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4468684025334272}
{"step": 7625, "loss": 4.2054290771484375, "grad_norm": 0.760282576084137, "learning_rate": 0.0001, "elbo": 4.20543, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.44716162326999764}
{"step": 7630, "loss": 4.196053314208984, "grad_norm": 0.7683029174804688, "learning_rate": 0.0001, "elbo": 4.19605, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.44745484400656815}
{"step": 7635, "loss": 4.206560516357422, "grad_norm": 0.8393563032150269, "learning_rate": 0.0001, "elbo": 4.20656, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4477480647431386}
{"step": 7640, "loss": 4.188703155517578, "grad_norm": 0.9162663221359253, "learning_rate": 0.0001, "elbo": 4.1887, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4480412854797091}
{"step": 7645, "loss": 4.1785133361816404, "grad_norm": 0.9772797226905823, "learning_rate": 0.0001, "elbo": 4.17851, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4483345062162796}
{"step": 7650, "loss": 4.196142578125, "grad_norm": 0.8013942241668701, "learning_rate": 0.0001, "elbo": 4.19614, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4486277269528501}
{"step": 7655, "loss": 4.1926124572753904, "grad_norm": 0.8173372149467468, "learning_rate": 0.0001, "elbo": 4.19261, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4489209476894206}
{"step": 7660, "loss": 4.190497970581054, "grad_norm": 0.807421863079071, "learning_rate": 0.0001, "elbo": 4.1905, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4492141684259911}
{"step": 7665, "loss": 4.1878204345703125, "grad_norm": 0.833602249622345, "learning_rate": 0.0001, "elbo": 4.18782, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.44950738916256155}
{"step": 7670, "loss": 4.196340179443359, "grad_norm": 0.8833894729614258, "learning_rate": 0.0001, "elbo": 4.19634, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.44980060989913206}
{"step": 7675, "loss": 4.183230209350586, "grad_norm": 0.798677384853363, "learning_rate": 0.0001, "elbo": 4.18323, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4500938306357026}
{"step": 7680, "loss": 4.1898338317871096, "grad_norm": 0.8616429567337036, "learning_rate": 0.0001, "elbo": 4.18983, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.45038705137227303}
{"step": 7685, "loss": 4.167446899414062, "grad_norm": 0.8016529679298401, "learning_rate": 0.0001, "elbo": 4.16745, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.45068027210884354}
{"step": 7690, "loss": 4.188301849365234, "grad_norm": 0.7752119898796082, "learning_rate": 0.0001, "elbo": 4.1883, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.45097349284541405}
{"step": 7695, "loss": 4.192750930786133, "grad_norm": 0.8899313807487488, "learning_rate": 0.0001, "elbo": 4.19275, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4512667135819845}
{"step": 7700, "loss": 4.1988975524902346, "grad_norm": 1.3856818675994873, "learning_rate": 0.0001, "elbo": 4.1989, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.451559934318555}
{"step": 7705, "loss": 4.185064697265625, "grad_norm": 0.7967100739479065, "learning_rate": 0.0001, "elbo": 4.18506, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4518531550551255}
{"step": 7710, "loss": 4.179136276245117, "grad_norm": 0.8071812987327576, "learning_rate": 0.0001, "elbo": 4.17914, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.452146375791696}
{"step": 7715, "loss": 4.179199600219727, "grad_norm": 1.0546562671661377, "learning_rate": 0.0001, "elbo": 4.1792, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4524395965282665}
{"step": 7720, "loss": 4.192221832275391, "grad_norm": 0.7912710309028625, "learning_rate": 0.0001, "elbo": 4.19222, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.452732817264837}
{"step": 7725, "loss": 4.185321807861328, "grad_norm": 1.0014954805374146, "learning_rate": 0.0001, "elbo": 4.18532, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.45302603800140745}
{"step": 7730, "loss": 4.208395767211914, "grad_norm": 0.9133712649345398, "learning_rate": 0.0001, "elbo": 4.2084, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.45331925873797796}
{"step": 7735, "loss": 4.173847961425781, "grad_norm": 0.7971651554107666, "learning_rate": 0.0001, "elbo": 4.17385, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4536124794745484}
{"step": 7740, "loss": 4.177717590332032, "grad_norm": 0.8470864295959473, "learning_rate": 0.0001, "elbo": 4.17772, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4539057002111189}
{"step": 7745, "loss": 4.180551528930664, "grad_norm": 0.8429899215698242, "learning_rate": 0.0001, "elbo": 4.18055, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.45419892094768943}
{"step": 7750, "loss": 4.187915420532226, "grad_norm": 0.8134303092956543, "learning_rate": 0.0001, "elbo": 4.18792, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4544921416842599}
{"step": 7755, "loss": 4.1751056671142575, "grad_norm": 0.9448834657669067, "learning_rate": 0.0001, "elbo": 4.17511, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4547853624208304}
{"step": 7760, "loss": 4.193013381958008, "grad_norm": 0.8742293119430542, "learning_rate": 0.0001, "elbo": 4.19301, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4550785831574009}
{"step": 7765, "loss": 4.1869548797607425, "grad_norm": 0.9248284101486206, "learning_rate": 0.0001, "elbo": 4.18696, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.45537180389397136}
{"step": 7770, "loss": 4.1635009765625, "grad_norm": 0.8368368148803711, "learning_rate": 0.0001, "elbo": 4.1635, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.45566502463054187}
{"step": 7775, "loss": 4.164447021484375, "grad_norm": 0.8982976675033569, "learning_rate": 0.0001, "elbo": 4.16445, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4559582453671124}
{"step": 7780, "loss": 4.186299133300781, "grad_norm": 0.9325571656227112, "learning_rate": 0.0001, "elbo": 4.1863, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.45625146610368283}
{"step": 7785, "loss": 4.181914138793945, "grad_norm": 1.1466243267059326, "learning_rate": 0.0001, "elbo": 4.18191, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.45654468684025334}
{"step": 7790, "loss": 4.181330871582031, "grad_norm": 0.8220081329345703, "learning_rate": 0.0001, "elbo": 4.18133, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.45683790757682385}
{"step": 7795, "loss": 4.166526031494141, "grad_norm": 0.8379613161087036, "learning_rate": 0.0001, "elbo": 4.16653, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4571311283133943}
{"step": 7800, "loss": 4.169757080078125, "grad_norm": 1.0629230737686157, "learning_rate": 0.0001, "elbo": 4.16976, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4574243490499648}
{"step": 7805, "loss": 4.187806701660156, "grad_norm": 0.9440805315971375, "learning_rate": 0.0001, "elbo": 4.18781, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4577175697865353}
{"step": 7810, "loss": 4.165150451660156, "grad_norm": 0.870791494846344, "learning_rate": 0.0001, "elbo": 4.16515, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4580107905231058}
{"step": 7815, "loss": 4.179643249511718, "grad_norm": 0.9150164127349854, "learning_rate": 0.0001, "elbo": 4.17964, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4583040112596763}
{"step": 7820, "loss": 4.178868103027344, "grad_norm": 0.8644953966140747, "learning_rate": 0.0001, "elbo": 4.17887, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4585972319962468}
{"step": 7825, "loss": 4.210498809814453, "grad_norm": 0.7773127555847168, "learning_rate": 0.0001, "elbo": 4.2105, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.45889045273281726}
{"step": 7830, "loss": 4.1911766052246096, "grad_norm": 0.8074390292167664, "learning_rate": 0.0001, "elbo": 4.19118, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.45918367346938777}
{"step": 7835, "loss": 4.160223770141601, "grad_norm": 0.7844667434692383, "learning_rate": 0.0001, "elbo": 4.16022, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4594768942059582}
{"step": 7840, "loss": 4.173008728027344, "grad_norm": 0.8502594828605652, "learning_rate": 0.0001, "elbo": 4.17301, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.45977011494252873}
{"step": 7845, "loss": 4.158856582641602, "grad_norm": 0.80558180809021, "learning_rate": 0.0001, "elbo": 4.15886, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.46006333567909924}
{"step": 7850, "loss": 4.185580444335938, "grad_norm": 0.7984511256217957, "learning_rate": 0.0001, "elbo": 4.18558, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4603565564156697}
{"step": 7855, "loss": 4.170903015136719, "grad_norm": 0.8934657573699951, "learning_rate": 0.0001, "elbo": 4.1709, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4606497771522402}
{"step": 7860, "loss": 4.187773895263672, "grad_norm": 1.0442930459976196, "learning_rate": 0.0001, "elbo": 4.18777, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4609429978888107}
{"step": 7865, "loss": 4.187646102905274, "grad_norm": 0.813366711139679, "learning_rate": 0.0001, "elbo": 4.18765, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.46123621862538117}
{"step": 7870, "loss": 4.172489166259766, "grad_norm": 0.8161780834197998, "learning_rate": 0.0001, "elbo": 4.17249, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4615294393619517}
{"step": 7875, "loss": 4.193240737915039, "grad_norm": 0.8498789668083191, "learning_rate": 0.0001, "elbo": 4.19324, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4618226600985222}
{"step": 7880, "loss": 4.1710765838623045, "grad_norm": 0.7875169515609741, "learning_rate": 0.0001, "elbo": 4.17108, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.46211588083509264}
{"step": 7885, "loss": 4.177079772949218, "grad_norm": 0.7710455656051636, "learning_rate": 0.0001, "elbo": 4.17708, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.46240910157166315}
{"step": 7890, "loss": 4.173828125, "grad_norm": 0.8878782391548157, "learning_rate": 0.0001, "elbo": 4.17383, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.46270232230823366}
{"step": 7895, "loss": 4.178538513183594, "grad_norm": 0.9066809415817261, "learning_rate": 0.0001, "elbo": 4.17854, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4629955430448041}
{"step": 7900, "loss": 4.158937072753906, "grad_norm": 0.8097683191299438, "learning_rate": 0.0001, "elbo": 4.15894, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4632887637813746}
{"step": 7905, "loss": 4.172878265380859, "grad_norm": 0.8503831028938293, "learning_rate": 0.0001, "elbo": 4.17288, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.46358198451794513}
{"step": 7910, "loss": 4.188429641723633, "grad_norm": 0.8847071528434753, "learning_rate": 0.0001, "elbo": 4.18843, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4638752052545156}
{"step": 7915, "loss": 4.180047607421875, "grad_norm": 0.8713324069976807, "learning_rate": 0.0001, "elbo": 4.18005, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4641684259910861}
{"step": 7920, "loss": 4.1907085418701175, "grad_norm": 0.8235464096069336, "learning_rate": 0.0001, "elbo": 4.19071, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4644616467276566}
{"step": 7925, "loss": 4.1697845458984375, "grad_norm": 0.9204462766647339, "learning_rate": 0.0001, "elbo": 4.16978, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.46475486746422706}
{"step": 7930, "loss": 4.158796310424805, "grad_norm": 0.8341113328933716, "learning_rate": 0.0001, "elbo": 4.1588, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.46504808820079757}
{"step": 7935, "loss": 4.182761383056641, "grad_norm": 0.8695325255393982, "learning_rate": 0.0001, "elbo": 4.18276, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.465341308937368}
{"step": 7940, "loss": 4.174540710449219, "grad_norm": 0.8537183403968811, "learning_rate": 0.0001, "elbo": 4.17454, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.46563452967393854}
{"step": 7945, "loss": 4.175474548339844, "grad_norm": 0.8613653779029846, "learning_rate": 0.0001, "elbo": 4.17547, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.46592775041050905}
{"step": 7950, "loss": 4.1741996765136715, "grad_norm": 0.8309460282325745, "learning_rate": 0.0001, "elbo": 4.1742, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4662209711470795}
{"step": 7955, "loss": 4.153443145751953, "grad_norm": 0.8253329992294312, "learning_rate": 0.0001, "elbo": 4.15344, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.46651419188365}
{"step": 7960, "loss": 4.15815544128418, "grad_norm": 0.8058229088783264, "learning_rate": 0.0001, "elbo": 4.15816, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4668074126202205}
{"step": 7965, "loss": 4.166358184814453, "grad_norm": 0.8766012787818909, "learning_rate": 0.0001, "elbo": 4.16636, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.467100633356791}
{"step": 7970, "loss": 4.161231994628906, "grad_norm": 0.8603659868240356, "learning_rate": 0.0001, "elbo": 4.16123, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4673938540933615}
{"step": 7975, "loss": 4.155414581298828, "grad_norm": 0.8921834826469421, "learning_rate": 0.0001, "elbo": 4.15541, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.467687074829932}
{"step": 7980, "loss": 4.167262268066406, "grad_norm": 0.9754250049591064, "learning_rate": 0.0001, "elbo": 4.16726, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.46798029556650245}
{"step": 7985, "loss": 4.172721099853516, "grad_norm": 0.8841999173164368, "learning_rate": 0.0001, "elbo": 4.17272, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.46827351630307296}
{"step": 7985, "loss": 4.172721099853516, "grad_norm": 0.8841999173164368, "learning_rate": 0.0001, "elbo": 4.17272, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.46827351630307296}
{"step": 7990, "loss": 4.163205337524414, "grad_norm": 0.7628805041313171, "learning_rate": 0.0001, "elbo": 4.16321, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.46856673703964347}
{"step": 7995, "loss": 4.163490295410156, "grad_norm": 1.0230939388275146, "learning_rate": 0.0001, "elbo": 4.16349, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4688599577762139}
{"step": 8000, "loss": 4.1560005187988285, "grad_norm": 0.8339589238166809, "learning_rate": 0.0001, "elbo": 4.156, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.46915317851278443}
{"step": 8005, "loss": 4.164443588256836, "grad_norm": 0.8704351782798767, "learning_rate": 0.0001, "elbo": 4.16444, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.46944639924935494}
{"step": 8010, "loss": 4.1706687927246096, "grad_norm": 0.7910287976264954, "learning_rate": 0.0001, "elbo": 4.17067, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4697396199859254}
{"step": 8015, "loss": 4.173469543457031, "grad_norm": 0.7835119962692261, "learning_rate": 0.0001, "elbo": 4.17347, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4700328407224959}
{"step": 8020, "loss": 4.174532699584961, "grad_norm": 0.9240177273750305, "learning_rate": 0.0001, "elbo": 4.17453, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.47032606145906636}
{"step": 8025, "loss": 4.148712921142578, "grad_norm": 1.056235909461975, "learning_rate": 0.0001, "elbo": 4.14871, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.47061928219563687}
{"step": 8030, "loss": 4.165680313110352, "grad_norm": 1.3511842489242554, "learning_rate": 0.0001, "elbo": 4.16568, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4709125029322074}
{"step": 8035, "loss": 4.149631500244141, "grad_norm": 0.8218174576759338, "learning_rate": 0.0001, "elbo": 4.14963, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.47120572366877783}
{"step": 8040, "loss": 4.171733856201172, "grad_norm": 0.7723366022109985, "learning_rate": 0.0001, "elbo": 4.17173, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.47149894440534834}
{"step": 8045, "loss": 4.170047378540039, "grad_norm": 0.785686194896698, "learning_rate": 0.0001, "elbo": 4.17005, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.47179216514191885}
{"step": 8050, "loss": 4.16680908203125, "grad_norm": 0.9115771055221558, "learning_rate": 0.0001, "elbo": 4.16681, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4720853858784893}
{"step": 8055, "loss": 4.182100296020508, "grad_norm": 0.904439389705658, "learning_rate": 0.0001, "elbo": 4.1821, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4723786066150598}
{"step": 8060, "loss": 4.164114379882813, "grad_norm": 1.1587984561920166, "learning_rate": 0.0001, "elbo": 4.16411, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4726718273516303}
{"step": 8065, "loss": 4.16998062133789, "grad_norm": 0.8036185503005981, "learning_rate": 0.0001, "elbo": 4.16998, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4729650480882008}
{"step": 8070, "loss": 4.169711303710938, "grad_norm": 0.8845058083534241, "learning_rate": 0.0001, "elbo": 4.16971, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4732582688247713}
{"step": 8075, "loss": 4.158287048339844, "grad_norm": 0.8755517601966858, "learning_rate": 0.0001, "elbo": 4.15829, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4735514895613418}
{"step": 8080, "loss": 4.181374359130859, "grad_norm": 0.829632043838501, "learning_rate": 0.0001, "elbo": 4.18137, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.47384471029791225}
{"step": 8085, "loss": 4.167231750488281, "grad_norm": 0.8538321256637573, "learning_rate": 0.0001, "elbo": 4.16723, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.47413793103448276}
{"step": 8090, "loss": 4.1479637145996096, "grad_norm": 0.8016154170036316, "learning_rate": 0.0001, "elbo": 4.14796, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.47443115177105327}
{"step": 8095, "loss": 4.174275970458984, "grad_norm": 0.8349810242652893, "learning_rate": 0.0001, "elbo": 4.17428, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4747243725076237}
{"step": 8100, "loss": 4.163806533813476, "grad_norm": 0.8671524524688721, "learning_rate": 0.0001, "elbo": 4.16381, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.47501759324419424}
{"step": 8105, "loss": 4.1594291687011715, "grad_norm": 0.7847620248794556, "learning_rate": 0.0001, "elbo": 4.15943, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.47531081398076475}
{"step": 8110, "loss": 4.1684013366699215, "grad_norm": 1.3091696500778198, "learning_rate": 0.0001, "elbo": 4.1684, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4756040347173352}
{"step": 8110, "loss": 4.1684013366699215, "grad_norm": 1.3091696500778198, "learning_rate": 0.0001, "elbo": 4.1684, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4756040347173352}
{"step": 8115, "loss": 4.18038330078125, "grad_norm": 0.7806649208068848, "learning_rate": 0.0001, "elbo": 4.18038, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4758972554539057}
{"step": 8120, "loss": 4.1626121520996096, "grad_norm": 0.8732619285583496, "learning_rate": 0.0001, "elbo": 4.16261, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.47619047619047616}
{"step": 8125, "loss": 4.155002593994141, "grad_norm": 0.7129701972007751, "learning_rate": 0.0001, "elbo": 4.155, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4764836969270467}
{"step": 8130, "loss": 4.175196838378906, "grad_norm": 0.8093693256378174, "learning_rate": 0.0001, "elbo": 4.1752, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4767769176636172}
{"step": 8135, "loss": 4.151451110839844, "grad_norm": 0.8100855350494385, "learning_rate": 0.0001, "elbo": 4.15145, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.47707013840018764}
{"step": 8140, "loss": 4.149877548217773, "grad_norm": 0.813477098941803, "learning_rate": 0.0001, "elbo": 4.14988, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.47736335913675815}
{"step": 8145, "loss": 4.173564910888672, "grad_norm": 1.0255593061447144, "learning_rate": 0.0001, "elbo": 4.17356, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.47765657987332866}
{"step": 8150, "loss": 4.152350997924804, "grad_norm": 0.7829867601394653, "learning_rate": 0.0001, "elbo": 4.15235, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4779498006098991}
{"step": 8155, "loss": 4.161847305297852, "grad_norm": 0.8379412889480591, "learning_rate": 0.0001, "elbo": 4.16185, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4782430213464696}
{"step": 8160, "loss": 4.16876449584961, "grad_norm": 0.7862647175788879, "learning_rate": 0.0001, "elbo": 4.16876, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.47853624208304013}
{"step": 8165, "loss": 4.16171875, "grad_norm": 0.8988814353942871, "learning_rate": 0.0001, "elbo": 4.16172, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4788294628196106}
{"step": 8170, "loss": 4.153520202636718, "grad_norm": 0.798761248588562, "learning_rate": 0.0001, "elbo": 4.15352, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4791226835561811}
{"step": 8175, "loss": 4.153607177734375, "grad_norm": 0.8671048879623413, "learning_rate": 0.0001, "elbo": 4.15361, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4794159042927516}
{"step": 8180, "loss": 4.146827697753906, "grad_norm": 0.8717813491821289, "learning_rate": 0.0001, "elbo": 4.14683, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.47970912502932206}
{"step": 8185, "loss": 4.188514709472656, "grad_norm": 0.7913803458213806, "learning_rate": 0.0001, "elbo": 4.18851, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.48000234576589257}
{"step": 8190, "loss": 4.145201873779297, "grad_norm": 0.9066973924636841, "learning_rate": 0.0001, "elbo": 4.1452, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4802955665024631}
{"step": 8195, "loss": 4.140983581542969, "grad_norm": 0.793569028377533, "learning_rate": 0.0001, "elbo": 4.14098, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.48058878723903353}
{"step": 8200, "loss": 4.1534065246582035, "grad_norm": 0.7593680024147034, "learning_rate": 0.0001, "elbo": 4.15341, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.48088200797560404}
{"step": 8205, "loss": 4.162026977539062, "grad_norm": 0.7898593544960022, "learning_rate": 0.0001, "elbo": 4.16203, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.48117522871217455}
{"step": 8210, "loss": 4.163462829589844, "grad_norm": 0.79458087682724, "learning_rate": 0.0001, "elbo": 4.16346, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.481468449448745}
{"step": 8215, "loss": 4.169795989990234, "grad_norm": 0.8614377975463867, "learning_rate": 0.0001, "elbo": 4.1698, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4817616701853155}
{"step": 8220, "loss": 4.1550132751464846, "grad_norm": 0.8965746760368347, "learning_rate": 0.0001, "elbo": 4.15501, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.48205489092188597}
{"step": 8225, "loss": 4.187446975708008, "grad_norm": 1.2399029731750488, "learning_rate": 0.0001, "elbo": 4.18745, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4823481116584565}
{"step": 8230, "loss": 4.162495422363281, "grad_norm": 0.7702012658119202, "learning_rate": 0.0001, "elbo": 4.1625, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.482641332395027}
{"step": 8235, "loss": 4.15987548828125, "grad_norm": 0.7575708627700806, "learning_rate": 0.0001, "elbo": 4.15988, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.48293455313159744}
{"step": 8240, "loss": 4.162060928344727, "grad_norm": 0.8849235773086548, "learning_rate": 0.0001, "elbo": 4.16206, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.48322777386816795}
{"step": 8245, "loss": 4.173390960693359, "grad_norm": 0.8593355417251587, "learning_rate": 0.0001, "elbo": 4.17339, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.48352099460473846}
{"step": 8250, "loss": 4.155214309692383, "grad_norm": 0.9535435438156128, "learning_rate": 0.0001, "elbo": 4.15521, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4838142153413089}
{"step": 8255, "loss": 4.146313095092774, "grad_norm": 0.9057966470718384, "learning_rate": 0.0001, "elbo": 4.14631, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4841074360778794}
{"step": 8260, "loss": 4.143760681152344, "grad_norm": 0.8587320446968079, "learning_rate": 0.0001, "elbo": 4.14376, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.48440065681444994}
{"step": 8265, "loss": 4.156658935546875, "grad_norm": 0.8215851187705994, "learning_rate": 0.0001, "elbo": 4.15666, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4846938775510204}
{"step": 8270, "loss": 4.157478332519531, "grad_norm": 1.006791114807129, "learning_rate": 0.0001, "elbo": 4.15748, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4849870982875909}
{"step": 8275, "loss": 4.155798721313476, "grad_norm": 0.8081496357917786, "learning_rate": 0.0001, "elbo": 4.1558, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4852803190241614}
{"step": 8280, "loss": 4.1442817687988285, "grad_norm": 1.2389253377914429, "learning_rate": 0.0001, "elbo": 4.14428, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.48557353976073186}
{"step": 8285, "loss": 4.146079635620117, "grad_norm": 0.8725124597549438, "learning_rate": 0.0001, "elbo": 4.14608, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4858667604973024}
{"step": 8290, "loss": 4.147039413452148, "grad_norm": 0.8579606413841248, "learning_rate": 0.0001, "elbo": 4.14704, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4861599812338729}
{"step": 8295, "loss": 4.165841674804687, "grad_norm": 0.8555323481559753, "learning_rate": 0.0001, "elbo": 4.16584, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.48645320197044334}
{"step": 8300, "loss": 4.150317001342773, "grad_norm": 0.8028645515441895, "learning_rate": 0.0001, "elbo": 4.15032, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.48674642270701385}
{"step": 8305, "loss": 4.161406707763672, "grad_norm": 0.786956250667572, "learning_rate": 0.0001, "elbo": 4.16141, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.48703964344358436}
{"step": 8310, "loss": 4.147460174560547, "grad_norm": 0.9350705742835999, "learning_rate": 0.0001, "elbo": 4.14746, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4873328641801548}
{"step": 8315, "loss": 4.149784088134766, "grad_norm": 0.7965216040611267, "learning_rate": 0.0001, "elbo": 4.14978, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4876260849167253}
{"step": 8320, "loss": 4.147763061523437, "grad_norm": 0.9176917672157288, "learning_rate": 0.0001, "elbo": 4.14776, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4879193056532958}
{"step": 8325, "loss": 4.1591033935546875, "grad_norm": 0.8141194581985474, "learning_rate": 0.0001, "elbo": 4.1591, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4882125263898663}
{"step": 8330, "loss": 4.150977325439453, "grad_norm": 0.8661426901817322, "learning_rate": 0.0001, "elbo": 4.15098, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4885057471264368}
{"step": 8335, "loss": 4.153521347045898, "grad_norm": 0.7912882566452026, "learning_rate": 0.0001, "elbo": 4.15352, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.48879896786300725}
{"step": 8340, "loss": 4.16158447265625, "grad_norm": 1.0851585865020752, "learning_rate": 0.0001, "elbo": 4.16158, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.48909218859957776}
{"step": 8345, "loss": 4.15918197631836, "grad_norm": 0.9290952682495117, "learning_rate": 0.0001, "elbo": 4.15918, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.48938540933614827}
{"step": 8350, "loss": 4.168410873413086, "grad_norm": 0.7862175703048706, "learning_rate": 0.0001, "elbo": 4.16841, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4896786300727187}
{"step": 8355, "loss": 4.152832412719727, "grad_norm": 0.9608004093170166, "learning_rate": 0.0001, "elbo": 4.15283, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.48997185080928923}
{"step": 8360, "loss": 4.145835113525391, "grad_norm": 0.7606785893440247, "learning_rate": 0.0001, "elbo": 4.14584, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.49026507154585974}
{"step": 8365, "loss": 4.148811721801758, "grad_norm": 1.010252833366394, "learning_rate": 0.0001, "elbo": 4.14881, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4905582922824302}
{"step": 8370, "loss": 4.17222900390625, "grad_norm": 0.8811052441596985, "learning_rate": 0.0001, "elbo": 4.17223, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4908515130190007}
{"step": 8375, "loss": 4.170277404785156, "grad_norm": 0.8615368604660034, "learning_rate": 0.0001, "elbo": 4.17028, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4911447337555712}
{"step": 8380, "loss": 4.152915954589844, "grad_norm": 1.6590734720230103, "learning_rate": 0.0001, "elbo": 4.15292, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.49143795449214167}
{"step": 8385, "loss": 4.145099639892578, "grad_norm": 0.8198543787002563, "learning_rate": 0.0001, "elbo": 4.1451, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4917311752287122}
{"step": 8390, "loss": 4.130083084106445, "grad_norm": 0.8208073973655701, "learning_rate": 0.0001, "elbo": 4.13008, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4920243959652827}
{"step": 8395, "loss": 4.144093704223633, "grad_norm": 0.8472787141799927, "learning_rate": 0.0001, "elbo": 4.14409, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.49231761670185314}
{"step": 8400, "loss": 4.130008316040039, "grad_norm": 0.7894567847251892, "learning_rate": 0.0001, "elbo": 4.13001, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.49261083743842365}
{"step": 8405, "loss": 4.12957534790039, "grad_norm": 0.7975947856903076, "learning_rate": 0.0001, "elbo": 4.12958, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4929040581749941}
{"step": 8410, "loss": 4.1560710906982425, "grad_norm": 0.7430965304374695, "learning_rate": 0.0001, "elbo": 4.15607, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4931972789115646}
{"step": 8415, "loss": 4.134183120727539, "grad_norm": 0.7951921224594116, "learning_rate": 0.0001, "elbo": 4.13418, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.49349049964813513}
{"step": 8420, "loss": 4.138215637207031, "grad_norm": 0.7876370549201965, "learning_rate": 0.0001, "elbo": 4.13822, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4937837203847056}
{"step": 8425, "loss": 4.137961578369141, "grad_norm": 1.187164306640625, "learning_rate": 0.0001, "elbo": 4.13796, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4940769411212761}
{"step": 8430, "loss": 4.128718566894531, "grad_norm": 0.8155369162559509, "learning_rate": 0.0001, "elbo": 4.12872, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4943701618578466}
{"step": 8435, "loss": 4.141825103759766, "grad_norm": 0.7712507843971252, "learning_rate": 0.0001, "elbo": 4.14183, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.49466338259441706}
{"step": 8440, "loss": 4.152875137329102, "grad_norm": 0.7866531610488892, "learning_rate": 0.0001, "elbo": 4.15288, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.49495660333098757}
{"step": 8445, "loss": 4.1473228454589846, "grad_norm": 0.810888946056366, "learning_rate": 0.0001, "elbo": 4.14732, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4952498240675581}
{"step": 8450, "loss": 4.131866455078125, "grad_norm": 0.8316232562065125, "learning_rate": 0.0001, "elbo": 4.13187, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.49554304480412853}
{"step": 8455, "loss": 4.140480041503906, "grad_norm": 1.1445610523223877, "learning_rate": 0.0001, "elbo": 4.14048, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.49583626554069904}
{"step": 8460, "loss": 4.134228897094727, "grad_norm": 0.7862867712974548, "learning_rate": 0.0001, "elbo": 4.13423, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.49612948627726955}
{"step": 8465, "loss": 4.139548492431641, "grad_norm": 0.8072391748428345, "learning_rate": 0.0001, "elbo": 4.13955, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.49642270701384}
{"step": 8470, "loss": 4.129290008544922, "grad_norm": 0.9582456946372986, "learning_rate": 0.0001, "elbo": 4.12929, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4967159277504105}
{"step": 8475, "loss": 4.131985473632812, "grad_norm": 0.8768434524536133, "learning_rate": 0.0001, "elbo": 4.13199, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.497009148486981}
{"step": 8480, "loss": 4.146512603759765, "grad_norm": 0.7655365467071533, "learning_rate": 0.0001, "elbo": 4.14651, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4973023692235515}
{"step": 8485, "loss": 4.139485549926758, "grad_norm": 0.8435392379760742, "learning_rate": 0.0001, "elbo": 4.13949, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.497595589960122}
{"step": 8490, "loss": 4.153841400146485, "grad_norm": 0.7831606864929199, "learning_rate": 0.0001, "elbo": 4.15384, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4978888106966925}
{"step": 8495, "loss": 4.145341110229492, "grad_norm": 0.8456084132194519, "learning_rate": 0.0001, "elbo": 4.14534, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.49818203143326295}
{"step": 8500, "loss": 4.150953674316407, "grad_norm": 0.8242064118385315, "learning_rate": 0.0001, "elbo": 4.15095, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.49847525216983346}
{"step": 8505, "loss": 4.155810165405273, "grad_norm": 0.8226311802864075, "learning_rate": 0.0001, "elbo": 4.15581, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4987684729064039}
{"step": 8510, "loss": 4.14598274230957, "grad_norm": 0.7933971285820007, "learning_rate": 0.0001, "elbo": 4.14598, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4990616936429744}
{"step": 8515, "loss": 4.136613845825195, "grad_norm": 0.7901386618614197, "learning_rate": 0.0001, "elbo": 4.13661, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.49935491437954493}
{"step": 8520, "loss": 4.143810272216797, "grad_norm": 0.7913089394569397, "learning_rate": 0.0001, "elbo": 4.14381, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4996481351161154}
{"step": 8525, "loss": 4.140924072265625, "grad_norm": 1.0214507579803467, "learning_rate": 0.0001, "elbo": 4.14092, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.4999413558526859}
{"step": 8530, "loss": 4.128712463378906, "grad_norm": 0.8059308528900146, "learning_rate": 0.0001, "elbo": 4.12871, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5002345765892564}
{"step": 8535, "loss": 4.150112152099609, "grad_norm": 0.7477666735649109, "learning_rate": 0.0001, "elbo": 4.15011, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5005277973258269}
{"step": 8540, "loss": 4.143193054199219, "grad_norm": 0.7941258549690247, "learning_rate": 0.0001, "elbo": 4.14319, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5008210180623974}
{"step": 8545, "loss": 4.1319580078125, "grad_norm": 0.7322304248809814, "learning_rate": 0.0001, "elbo": 4.13196, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5011142387989679}
{"step": 8550, "loss": 4.137186431884766, "grad_norm": 0.7410470843315125, "learning_rate": 0.0001, "elbo": 4.13719, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5014074595355383}
{"step": 8555, "loss": 4.153128051757813, "grad_norm": 0.8329765200614929, "learning_rate": 0.0001, "elbo": 4.15313, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5017006802721088}
{"step": 8560, "loss": 4.136379241943359, "grad_norm": 0.7092733383178711, "learning_rate": 0.0001, "elbo": 4.13638, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5019939010086794}
{"step": 8565, "loss": 4.119370651245117, "grad_norm": 0.9017776250839233, "learning_rate": 0.0001, "elbo": 4.11937, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5022871217452498}
{"step": 8570, "loss": 4.142498016357422, "grad_norm": 0.8154722452163696, "learning_rate": 0.0001, "elbo": 4.1425, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5025803424818203}
{"step": 8575, "loss": 4.129923629760742, "grad_norm": 0.7901061177253723, "learning_rate": 0.0001, "elbo": 4.12992, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5028735632183908}
{"step": 8580, "loss": 4.1365619659423825, "grad_norm": 0.7802621126174927, "learning_rate": 0.0001, "elbo": 4.13656, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5031667839549613}
{"step": 8585, "loss": 4.117621612548828, "grad_norm": 0.781836986541748, "learning_rate": 0.0001, "elbo": 4.11762, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5034600046915317}
{"step": 8590, "loss": 4.149868774414062, "grad_norm": 0.7973801493644714, "learning_rate": 0.0001, "elbo": 4.14987, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5037532254281023}
{"step": 8595, "loss": 4.136037445068359, "grad_norm": 0.7655191421508789, "learning_rate": 0.0001, "elbo": 4.13604, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5040464461646728}
{"step": 8600, "loss": 4.139761734008789, "grad_norm": 0.7958011627197266, "learning_rate": 0.0001, "elbo": 4.13976, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5043396669012432}
{"step": 8605, "loss": 4.1371604919433596, "grad_norm": 0.7885409593582153, "learning_rate": 0.0001, "elbo": 4.13716, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5046328876378138}
{"step": 8610, "loss": 4.152616119384765, "grad_norm": 0.8495621681213379, "learning_rate": 0.0001, "elbo": 4.15262, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5049261083743842}
{"step": 8615, "loss": 4.138817596435547, "grad_norm": 0.8328672647476196, "learning_rate": 0.0001, "elbo": 4.13882, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5052193291109547}
{"step": 8620, "loss": 4.13799819946289, "grad_norm": 0.7692972421646118, "learning_rate": 0.0001, "elbo": 4.138, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5055125498475252}
{"step": 8625, "loss": 4.143169403076172, "grad_norm": 0.8198220133781433, "learning_rate": 0.0001, "elbo": 4.14317, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5058057705840957}
{"step": 8630, "loss": 4.129398345947266, "grad_norm": 1.1787115335464478, "learning_rate": 0.0001, "elbo": 4.1294, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5060989913206662}
{"step": 8635, "loss": 4.123130798339844, "grad_norm": 0.7825298309326172, "learning_rate": 0.0001, "elbo": 4.12313, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5063922120572367}
{"step": 8640, "loss": 4.129697036743164, "grad_norm": 0.7944924831390381, "learning_rate": 0.0001, "elbo": 4.1297, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5066854327938072}
{"step": 8645, "loss": 4.127973556518555, "grad_norm": 0.807621955871582, "learning_rate": 0.0001, "elbo": 4.12797, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5069786535303776}
{"step": 8650, "loss": 4.138015365600586, "grad_norm": 0.7196875214576721, "learning_rate": 0.0001, "elbo": 4.13802, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5072718742669482}
{"step": 8655, "loss": 4.1408027648925785, "grad_norm": 0.7824382185935974, "learning_rate": 0.0001, "elbo": 4.1408, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5075650950035187}
{"step": 8660, "loss": 4.1166637420654295, "grad_norm": 0.8375548124313354, "learning_rate": 0.0001, "elbo": 4.11666, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5078583157400891}
{"step": 8665, "loss": 4.127230072021485, "grad_norm": 0.7953649759292603, "learning_rate": 0.0001, "elbo": 4.12723, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5081515364766597}
{"step": 8670, "loss": 4.1098274230957035, "grad_norm": 0.8030808568000793, "learning_rate": 0.0001, "elbo": 4.10983, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5084447572132301}
{"step": 8675, "loss": 4.144309616088867, "grad_norm": 0.8455341458320618, "learning_rate": 0.0001, "elbo": 4.14431, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5087379779498006}
{"step": 8680, "loss": 4.129547500610352, "grad_norm": 0.8448601365089417, "learning_rate": 0.0001, "elbo": 4.12955, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5090311986863711}
{"step": 8685, "loss": 4.131552124023438, "grad_norm": 0.8896735310554504, "learning_rate": 0.0001, "elbo": 4.13155, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5093244194229416}
{"step": 8690, "loss": 4.128755950927735, "grad_norm": 0.8901966214179993, "learning_rate": 0.0001, "elbo": 4.12876, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.509617640159512}
{"step": 8695, "loss": 4.132412338256836, "grad_norm": 1.074439287185669, "learning_rate": 0.0001, "elbo": 4.13241, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5099108608960826}
{"step": 8700, "loss": 4.147787475585938, "grad_norm": 0.9668256640434265, "learning_rate": 0.0001, "elbo": 4.14779, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5102040816326531}
{"step": 8705, "loss": 4.122073364257813, "grad_norm": 0.7999292612075806, "learning_rate": 0.0001, "elbo": 4.12207, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5104973023692235}
{"step": 8710, "loss": 4.1214141845703125, "grad_norm": 1.0494494438171387, "learning_rate": 0.0001, "elbo": 4.12141, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5107905231057941}
{"step": 8715, "loss": 4.120834732055664, "grad_norm": 1.2387974262237549, "learning_rate": 0.0001, "elbo": 4.12083, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5110837438423645}
{"step": 8720, "loss": 4.147681427001953, "grad_norm": 0.7430554032325745, "learning_rate": 0.0001, "elbo": 4.14768, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.511376964578935}
{"step": 8725, "loss": 4.115468978881836, "grad_norm": 0.9192441701889038, "learning_rate": 0.0001, "elbo": 4.11547, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5116701853155056}
{"step": 8730, "loss": 4.142421722412109, "grad_norm": 0.7700181603431702, "learning_rate": 0.0001, "elbo": 4.14242, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.511963406052076}
{"step": 8735, "loss": 4.1206298828125, "grad_norm": 0.8342505693435669, "learning_rate": 0.0001, "elbo": 4.12063, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5122566267886465}
{"step": 8740, "loss": 4.128981018066407, "grad_norm": 1.1129529476165771, "learning_rate": 0.0001, "elbo": 4.12898, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5125498475252169}
{"step": 8745, "loss": 4.135094451904297, "grad_norm": 1.2472339868545532, "learning_rate": 0.0001, "elbo": 4.1351, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5128430682617875}
{"step": 8750, "loss": 4.129027557373047, "grad_norm": 0.8393332958221436, "learning_rate": 0.0001, "elbo": 4.12903, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.513136288998358}
{"step": 8755, "loss": 4.123611068725586, "grad_norm": 0.8263057470321655, "learning_rate": 0.0001, "elbo": 4.12361, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5134295097349284}
{"step": 8760, "loss": 4.115231323242187, "grad_norm": 0.7988448143005371, "learning_rate": 0.0001, "elbo": 4.11523, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.513722730471499}
{"step": 8765, "loss": 4.111352157592774, "grad_norm": 0.7202323079109192, "learning_rate": 0.0001, "elbo": 4.11135, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5140159512080694}
{"step": 8770, "loss": 4.127300262451172, "grad_norm": 0.8507987856864929, "learning_rate": 0.0001, "elbo": 4.1273, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5143091719446399}
{"step": 8775, "loss": 4.117613220214844, "grad_norm": 0.8162272572517395, "learning_rate": 0.0001, "elbo": 4.11761, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5146023926812104}
{"step": 8780, "loss": 4.141753387451172, "grad_norm": 0.9376168251037598, "learning_rate": 0.0001, "elbo": 4.14175, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5148956134177809}
{"step": 8785, "loss": 4.114214324951172, "grad_norm": 0.7487699389457703, "learning_rate": 0.0001, "elbo": 4.11421, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5151888341543513}
{"step": 8790, "loss": 4.12005615234375, "grad_norm": 0.7672486901283264, "learning_rate": 0.0001, "elbo": 4.12006, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5154820548909219}
{"step": 8795, "loss": 4.1147613525390625, "grad_norm": 0.892281174659729, "learning_rate": 0.0001, "elbo": 4.11476, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5157752756274924}
{"step": 8800, "loss": 4.1027473449707035, "grad_norm": 0.7652032375335693, "learning_rate": 0.0001, "elbo": 4.10275, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5160684963640628}
{"step": 8805, "loss": 4.133616256713867, "grad_norm": 0.8648849129676819, "learning_rate": 0.0001, "elbo": 4.13362, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5163617171006334}
{"step": 8810, "loss": 4.124132537841797, "grad_norm": 0.7785856127738953, "learning_rate": 0.0001, "elbo": 4.12413, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5166549378372038}
{"step": 8815, "loss": 4.139149475097656, "grad_norm": 0.7262364029884338, "learning_rate": 0.0001, "elbo": 4.13915, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5169481585737743}
{"step": 8820, "loss": 4.11767349243164, "grad_norm": 0.78727126121521, "learning_rate": 0.0001, "elbo": 4.11767, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5172413793103449}
{"step": 8825, "loss": 4.125479125976563, "grad_norm": 0.835338294506073, "learning_rate": 0.0001, "elbo": 4.12548, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5175346000469153}
{"step": 8830, "loss": 4.107513809204102, "grad_norm": 2.5861239433288574, "learning_rate": 0.0001, "elbo": 4.10751, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5178278207834858}
{"step": 8835, "loss": 4.116310119628906, "grad_norm": 0.7953509092330933, "learning_rate": 0.0001, "elbo": 4.11631, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5181210415200563}
{"step": 8840, "loss": 4.119696807861328, "grad_norm": 0.8533490896224976, "learning_rate": 0.0001, "elbo": 4.1197, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5184142622566268}
{"step": 8845, "loss": 4.100836181640625, "grad_norm": 1.069806694984436, "learning_rate": 0.0001, "elbo": 4.10084, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5187074829931972}
{"step": 8850, "loss": 4.135231018066406, "grad_norm": 0.8454214930534363, "learning_rate": 0.0001, "elbo": 4.13523, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5190007037297678}
{"step": 8855, "loss": 4.130347442626953, "grad_norm": 0.8505374193191528, "learning_rate": 0.0001, "elbo": 4.13035, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5192939244663383}
{"step": 8860, "loss": 4.117597198486328, "grad_norm": 0.7755979299545288, "learning_rate": 0.0001, "elbo": 4.1176, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5195871452029087}
{"step": 8865, "loss": 4.128369903564453, "grad_norm": 1.1009490489959717, "learning_rate": 0.0001, "elbo": 4.12837, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5198803659394793}
{"step": 8870, "loss": 4.1020050048828125, "grad_norm": 0.8204982876777649, "learning_rate": 0.0001, "elbo": 4.10201, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5201735866760497}
{"step": 8875, "loss": 4.112681198120117, "grad_norm": 0.7790603637695312, "learning_rate": 0.0001, "elbo": 4.11268, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5204668074126202}
{"step": 8880, "loss": 4.14307861328125, "grad_norm": 0.7447506189346313, "learning_rate": 0.0001, "elbo": 4.14308, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5207600281491908}
{"step": 8885, "loss": 4.132874298095703, "grad_norm": 0.7884635329246521, "learning_rate": 0.0001, "elbo": 4.13287, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5210532488857612}
{"step": 8890, "loss": 4.140370941162109, "grad_norm": 0.7520080804824829, "learning_rate": 0.0001, "elbo": 4.14037, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5213464696223317}
{"step": 8895, "loss": 4.1357879638671875, "grad_norm": 0.7649786472320557, "learning_rate": 0.0001, "elbo": 4.13579, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5216396903589022}
{"step": 8900, "loss": 4.133990859985351, "grad_norm": 0.8524554967880249, "learning_rate": 0.0001, "elbo": 4.13399, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5219329110954727}
{"step": 8905, "loss": 4.1232868194580075, "grad_norm": 0.9039467573165894, "learning_rate": 0.0001, "elbo": 4.12329, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5222261318320431}
{"step": 8905, "loss": 4.1232868194580075, "grad_norm": 0.9039467573165894, "learning_rate": 0.0001, "elbo": 4.12329, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5222261318320431}
{"step": 8910, "loss": 4.122122192382813, "grad_norm": 0.7886723875999451, "learning_rate": 0.0001, "elbo": 4.12212, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5225193525686137}
{"step": 8915, "loss": 4.116820526123047, "grad_norm": 0.7642598152160645, "learning_rate": 0.0001, "elbo": 4.11682, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5228125733051842}
{"step": 8920, "loss": 4.117059707641602, "grad_norm": 0.9289022088050842, "learning_rate": 0.0001, "elbo": 4.11706, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5231057940417546}
{"step": 8920, "loss": 4.117059707641602, "grad_norm": 0.9289022088050842, "learning_rate": 0.0001, "elbo": 4.11706, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5231057940417546}
{"step": 8925, "loss": 4.127414321899414, "grad_norm": 0.8340423107147217, "learning_rate": 0.0001, "elbo": 4.12741, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5233990147783252}
{"step": 8930, "loss": 4.110250473022461, "grad_norm": 0.7755618095397949, "learning_rate": 0.0001, "elbo": 4.11025, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5236922355148956}
{"step": 8935, "loss": 4.094575881958008, "grad_norm": 0.77837735414505, "learning_rate": 0.0001, "elbo": 4.09458, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5239854562514661}
{"step": 8940, "loss": 4.126515197753906, "grad_norm": 0.9247434735298157, "learning_rate": 0.0001, "elbo": 4.12652, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5242786769880365}
{"step": 8945, "loss": 4.116573333740234, "grad_norm": 0.8371924757957458, "learning_rate": 0.0001, "elbo": 4.11657, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5245718977246071}
{"step": 8950, "loss": 4.117104339599609, "grad_norm": 0.884459912776947, "learning_rate": 0.0001, "elbo": 4.1171, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5248651184611776}
{"step": 8955, "loss": 4.130873489379883, "grad_norm": 0.8006719946861267, "learning_rate": 0.0001, "elbo": 4.13087, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.525158339197748}
{"step": 8960, "loss": 4.122833633422852, "grad_norm": 0.7760027050971985, "learning_rate": 0.0001, "elbo": 4.12283, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5254515599343186}
{"step": 8965, "loss": 4.123465728759766, "grad_norm": 0.8454068899154663, "learning_rate": 0.0001, "elbo": 4.12347, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.525744780670889}
{"step": 8970, "loss": 4.105521011352539, "grad_norm": 0.842519223690033, "learning_rate": 0.0001, "elbo": 4.10552, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5260380014074595}
{"step": 8975, "loss": 4.107923889160157, "grad_norm": 0.9480006098747253, "learning_rate": 0.0001, "elbo": 4.10792, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.52633122214403}
{"step": 8980, "loss": 4.095783996582031, "grad_norm": 0.7558596730232239, "learning_rate": 0.0001, "elbo": 4.09578, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5266244428806005}
{"step": 8985, "loss": 4.130862045288086, "grad_norm": 0.834254264831543, "learning_rate": 0.0001, "elbo": 4.13086, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.526917663617171}
{"step": 8990, "loss": 4.1018321990966795, "grad_norm": 0.7670955657958984, "learning_rate": 0.0001, "elbo": 4.10183, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5272108843537415}
{"step": 8995, "loss": 4.125206375122071, "grad_norm": 0.7963238954544067, "learning_rate": 0.0001, "elbo": 4.12521, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.527504105090312}
{"step": 9000, "loss": 4.104961013793945, "grad_norm": 0.8140357136726379, "learning_rate": 0.0001, "elbo": 4.10496, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5277973258268824}
{"step": 9005, "loss": 4.109303665161133, "grad_norm": 0.897360622882843, "learning_rate": 0.0001, "elbo": 4.1093, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.528090546563453}
{"step": 9010, "loss": 4.1096343994140625, "grad_norm": 0.843919575214386, "learning_rate": 0.0001, "elbo": 4.10963, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5283837673000235}
{"step": 9015, "loss": 4.110011291503906, "grad_norm": 0.9944485425949097, "learning_rate": 0.0001, "elbo": 4.11001, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5286769880365939}
{"step": 9020, "loss": 4.123478698730469, "grad_norm": 0.7343248128890991, "learning_rate": 0.0001, "elbo": 4.12348, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5289702087731645}
{"step": 9025, "loss": 4.100984191894531, "grad_norm": 0.8701388835906982, "learning_rate": 0.0001, "elbo": 4.10098, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5292634295097349}
{"step": 9030, "loss": 4.119255828857422, "grad_norm": 0.7932597398757935, "learning_rate": 0.0001, "elbo": 4.11926, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5295566502463054}
{"step": 9035, "loss": 4.099277114868164, "grad_norm": 0.7695010304450989, "learning_rate": 0.0001, "elbo": 4.09928, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.529849870982876}
{"step": 9040, "loss": 4.106448745727539, "grad_norm": 0.8049638867378235, "learning_rate": 0.0001, "elbo": 4.10645, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5301430917194464}
{"step": 9045, "loss": 4.118513107299805, "grad_norm": 0.8126670122146606, "learning_rate": 0.0001, "elbo": 4.11851, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5304363124560169}
{"step": 9050, "loss": 4.1083831787109375, "grad_norm": 0.7365300059318542, "learning_rate": 0.0001, "elbo": 4.10838, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5307295331925874}
{"step": 9055, "loss": 4.116896057128907, "grad_norm": 0.8044511079788208, "learning_rate": 0.0001, "elbo": 4.1169, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5310227539291579}
{"step": 9060, "loss": 4.128984832763672, "grad_norm": 0.9518481492996216, "learning_rate": 0.0001, "elbo": 4.12898, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5313159746657283}
{"step": 9065, "loss": 4.106513595581054, "grad_norm": 1.0016294717788696, "learning_rate": 0.0001, "elbo": 4.10651, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5316091954022989}
{"step": 9070, "loss": 4.117923355102539, "grad_norm": 0.7379984855651855, "learning_rate": 0.0001, "elbo": 4.11792, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5319024161388693}
{"step": 9075, "loss": 4.099014663696289, "grad_norm": 0.8552210927009583, "learning_rate": 0.0001, "elbo": 4.09901, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5321956368754398}
{"step": 9080, "loss": 4.105742263793945, "grad_norm": 0.7705785036087036, "learning_rate": 0.0001, "elbo": 4.10574, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5324888576120104}
{"step": 9085, "loss": 4.124180603027344, "grad_norm": 0.8223763108253479, "learning_rate": 0.0001, "elbo": 4.12418, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5327820783485808}
{"step": 9090, "loss": 4.12115249633789, "grad_norm": 0.8090901374816895, "learning_rate": 0.0001, "elbo": 4.12115, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5330752990851513}
{"step": 9095, "loss": 4.101555633544922, "grad_norm": 0.8967928886413574, "learning_rate": 0.0001, "elbo": 4.10156, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5333685198217218}
{"step": 9100, "loss": 4.102666854858398, "grad_norm": 0.7787806391716003, "learning_rate": 0.0001, "elbo": 4.10267, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5336617405582923}
{"step": 9105, "loss": 4.107820129394531, "grad_norm": 0.8071550130844116, "learning_rate": 0.0001, "elbo": 4.10782, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5339549612948628}
{"step": 9110, "loss": 4.123517608642578, "grad_norm": 0.7973411083221436, "learning_rate": 0.0001, "elbo": 4.12352, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5342481820314333}
{"step": 9115, "loss": 4.1352996826171875, "grad_norm": 0.7604669332504272, "learning_rate": 0.0001, "elbo": 4.1353, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5345414027680038}
{"step": 9120, "loss": 4.124125671386719, "grad_norm": 0.7226369380950928, "learning_rate": 0.0001, "elbo": 4.12413, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5348346235045742}
{"step": 9125, "loss": 4.094917297363281, "grad_norm": 0.7794358730316162, "learning_rate": 0.0001, "elbo": 4.09492, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5351278442411447}
{"step": 9130, "loss": 4.121977233886719, "grad_norm": 0.8692811727523804, "learning_rate": 0.0001, "elbo": 4.12198, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5354210649777152}
{"step": 9135, "loss": 4.097240829467774, "grad_norm": 0.8937742710113525, "learning_rate": 0.0001, "elbo": 4.09724, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5357142857142857}
{"step": 9140, "loss": 4.116383361816406, "grad_norm": 0.8567917346954346, "learning_rate": 0.0001, "elbo": 4.11638, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5360075064508562}
{"step": 9145, "loss": 4.111013031005859, "grad_norm": 0.7691600918769836, "learning_rate": 0.0001, "elbo": 4.11101, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5363007271874267}
{"step": 9150, "loss": 4.128033065795899, "grad_norm": 0.7763527631759644, "learning_rate": 0.0001, "elbo": 4.12803, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5365939479239972}
{"step": 9155, "loss": 4.098055267333985, "grad_norm": 0.8388184905052185, "learning_rate": 0.0001, "elbo": 4.09806, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5368871686605676}
{"step": 9160, "loss": 4.112093353271485, "grad_norm": 0.8569207191467285, "learning_rate": 0.0001, "elbo": 4.11209, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5371803893971382}
{"step": 9165, "loss": 4.104413604736328, "grad_norm": 0.8105647563934326, "learning_rate": 0.0001, "elbo": 4.10441, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5374736101337086}
{"step": 9170, "loss": 4.113620758056641, "grad_norm": 1.1254597902297974, "learning_rate": 0.0001, "elbo": 4.11362, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5377668308702791}
{"step": 9175, "loss": 4.112947463989258, "grad_norm": 0.8300163149833679, "learning_rate": 0.0001, "elbo": 4.11295, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5380600516068497}
{"step": 9175, "loss": 4.112947463989258, "grad_norm": 0.8300163149833679, "learning_rate": 0.0001, "elbo": 4.11295, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5380600516068497}
{"step": 9180, "loss": 4.112221527099609, "grad_norm": 0.8166015148162842, "learning_rate": 0.0001, "elbo": 4.11222, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5383532723434201}
{"step": 9185, "loss": 4.105339813232422, "grad_norm": 0.8726441264152527, "learning_rate": 0.0001, "elbo": 4.10534, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5386464930799906}
{"step": 9190, "loss": 4.104389953613281, "grad_norm": 0.8221547603607178, "learning_rate": 0.0001, "elbo": 4.10439, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5389397138165611}
{"step": 9195, "loss": 4.1003364562988285, "grad_norm": 0.7808411717414856, "learning_rate": 0.0001, "elbo": 4.10034, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5392329345531316}
{"step": 9200, "loss": 4.097819137573242, "grad_norm": 0.8968193531036377, "learning_rate": 0.0001, "elbo": 4.09782, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.539526155289702}
{"step": 9205, "loss": 4.102664947509766, "grad_norm": 0.7758597135543823, "learning_rate": 0.0001, "elbo": 4.10266, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5398193760262726}
{"step": 9210, "loss": 4.117410659790039, "grad_norm": 0.7880005240440369, "learning_rate": 0.0001, "elbo": 4.11741, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5401125967628431}
{"step": 9215, "loss": 4.1008140563964846, "grad_norm": 0.7998254299163818, "learning_rate": 0.0001, "elbo": 4.10081, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5404058174994135}
{"step": 9220, "loss": 4.109349060058594, "grad_norm": 0.7948133945465088, "learning_rate": 0.0001, "elbo": 4.10935, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5406990382359841}
{"step": 9225, "loss": 4.09310302734375, "grad_norm": 0.8312661647796631, "learning_rate": 0.0001, "elbo": 4.0931, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5409922589725545}
{"step": 9230, "loss": 4.1048583984375, "grad_norm": 0.7661923766136169, "learning_rate": 0.0001, "elbo": 4.10486, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.541285479709125}
{"step": 9235, "loss": 4.114118957519532, "grad_norm": 0.786271870136261, "learning_rate": 0.0001, "elbo": 4.11412, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5415787004456956}
{"step": 9240, "loss": 4.123975372314453, "grad_norm": 0.983856737613678, "learning_rate": 0.0001, "elbo": 4.12398, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.541871921182266}
{"step": 9245, "loss": 4.105426025390625, "grad_norm": 1.1580966711044312, "learning_rate": 0.0001, "elbo": 4.10543, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5421651419188365}
{"step": 9250, "loss": 4.101992797851563, "grad_norm": 0.866817831993103, "learning_rate": 0.0001, "elbo": 4.10199, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.542458362655407}
{"step": 9255, "loss": 4.104186248779297, "grad_norm": 0.7798430323600769, "learning_rate": 0.0001, "elbo": 4.10419, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5427515833919775}
{"step": 9260, "loss": 4.109206771850586, "grad_norm": 0.7302224636077881, "learning_rate": 0.0001, "elbo": 4.10921, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5430448041285479}
{"step": 9265, "loss": 4.106218338012695, "grad_norm": 0.737060010433197, "learning_rate": 0.0001, "elbo": 4.10622, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5433380248651185}
{"step": 9265, "loss": 4.106218338012695, "grad_norm": 0.737060010433197, "learning_rate": 0.0001, "elbo": 4.10622, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5433380248651185}
{"step": 9270, "loss": 4.133702087402344, "grad_norm": 1.0246366262435913, "learning_rate": 0.0001, "elbo": 4.1337, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.543631245601689}
{"step": 9275, "loss": 4.0981597900390625, "grad_norm": 0.8356115221977234, "learning_rate": 0.0001, "elbo": 4.09816, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5439244663382594}
{"step": 9280, "loss": 4.096174240112305, "grad_norm": 0.7385044097900391, "learning_rate": 0.0001, "elbo": 4.09617, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.54421768707483}
{"step": 9285, "loss": 4.101705932617188, "grad_norm": 0.8640216588973999, "learning_rate": 0.0001, "elbo": 4.10171, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5445109078114004}
{"step": 9290, "loss": 4.096998596191407, "grad_norm": 0.8237974047660828, "learning_rate": 0.0001, "elbo": 4.097, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5448041285479709}
{"step": 9295, "loss": 4.106158065795898, "grad_norm": 0.856690526008606, "learning_rate": 0.0001, "elbo": 4.10616, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5450973492845415}
{"step": 9300, "loss": 4.104347229003906, "grad_norm": 0.759966254234314, "learning_rate": 0.0001, "elbo": 4.10435, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5453905700211119}
{"step": 9305, "loss": 4.089200592041015, "grad_norm": 0.8148581385612488, "learning_rate": 0.0001, "elbo": 4.0892, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5456837907576824}
{"step": 9310, "loss": 4.106600189208985, "grad_norm": 0.7833085060119629, "learning_rate": 0.0001, "elbo": 4.1066, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5459770114942529}
{"step": 9315, "loss": 4.101327514648437, "grad_norm": 0.8891838192939758, "learning_rate": 0.0001, "elbo": 4.10133, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5462702322308234}
{"step": 9320, "loss": 4.0941413879394535, "grad_norm": 0.8738412857055664, "learning_rate": 0.0001, "elbo": 4.09414, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5465634529673938}
{"step": 9325, "loss": 4.0980583190917965, "grad_norm": 1.1457328796386719, "learning_rate": 0.0001, "elbo": 4.09806, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5468566737039643}
{"step": 9330, "loss": 4.105833435058594, "grad_norm": 0.8302566409111023, "learning_rate": 0.0001, "elbo": 4.10583, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5471498944405349}
{"step": 9335, "loss": 4.103914642333985, "grad_norm": 1.1692113876342773, "learning_rate": 0.0001, "elbo": 4.10391, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5474431151771053}
{"step": 9340, "loss": 4.0909576416015625, "grad_norm": 0.8336611986160278, "learning_rate": 0.0001, "elbo": 4.09096, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5477363359136758}
{"step": 9345, "loss": 4.08787956237793, "grad_norm": 0.7383843064308167, "learning_rate": 0.0001, "elbo": 4.08788, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5480295566502463}
{"step": 9350, "loss": 4.088631439208984, "grad_norm": 0.7845789194107056, "learning_rate": 0.0001, "elbo": 4.08863, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5483227773868168}
{"step": 9355, "loss": 4.113653564453125, "grad_norm": 0.7201595306396484, "learning_rate": 0.0001, "elbo": 4.11365, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5486159981233872}
{"step": 9360, "loss": 4.090837478637695, "grad_norm": 0.7925769090652466, "learning_rate": 0.0001, "elbo": 4.09084, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5489092188599578}
{"step": 9365, "loss": 4.090701675415039, "grad_norm": 0.8948714733123779, "learning_rate": 0.0001, "elbo": 4.0907, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5492024395965283}
{"step": 9370, "loss": 4.094524383544922, "grad_norm": 0.8541393876075745, "learning_rate": 0.0001, "elbo": 4.09452, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5494956603330987}
{"step": 9375, "loss": 4.079046630859375, "grad_norm": 0.8350661993026733, "learning_rate": 0.0001, "elbo": 4.07905, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5497888810696693}
{"step": 9380, "loss": 4.1004638671875, "grad_norm": 0.7850583791732788, "learning_rate": 0.0001, "elbo": 4.10046, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5500821018062397}
{"step": 9385, "loss": 4.09113883972168, "grad_norm": 0.7381396889686584, "learning_rate": 0.0001, "elbo": 4.09114, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5503753225428102}
{"step": 9390, "loss": 4.119182586669922, "grad_norm": 0.7974188327789307, "learning_rate": 0.0001, "elbo": 4.11918, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5506685432793808}
{"step": 9395, "loss": 4.1015464782714846, "grad_norm": 1.066970944404602, "learning_rate": 0.0001, "elbo": 4.10155, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5509617640159512}
{"step": 9400, "loss": 4.0990455627441404, "grad_norm": 0.7807862758636475, "learning_rate": 0.0001, "elbo": 4.09905, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5512549847525217}
{"step": 9405, "loss": 4.090166854858398, "grad_norm": 0.795857846736908, "learning_rate": 0.0001, "elbo": 4.09017, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5515482054890922}
{"step": 9410, "loss": 4.085909271240235, "grad_norm": 0.767871618270874, "learning_rate": 0.0001, "elbo": 4.08591, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5518414262256627}
{"step": 9415, "loss": 4.104988861083984, "grad_norm": 0.760766863822937, "learning_rate": 0.0001, "elbo": 4.10499, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5521346469622331}
{"step": 9420, "loss": 4.1073356628417965, "grad_norm": 0.7576218247413635, "learning_rate": 0.0001, "elbo": 4.10734, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5524278676988037}
{"step": 9425, "loss": 4.104351806640625, "grad_norm": 0.7691395282745361, "learning_rate": 0.0001, "elbo": 4.10435, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5527210884353742}
{"step": 9430, "loss": 4.089314270019531, "grad_norm": 0.770104169845581, "learning_rate": 0.0001, "elbo": 4.08931, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5530143091719446}
{"step": 9435, "loss": 4.0931556701660154, "grad_norm": 1.0791205167770386, "learning_rate": 0.0001, "elbo": 4.09316, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5533075299085152}
{"step": 9440, "loss": 4.087838745117187, "grad_norm": 0.9545953869819641, "learning_rate": 0.0001, "elbo": 4.08784, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5536007506450856}
{"step": 9445, "loss": 4.0878852844238285, "grad_norm": 0.7809723019599915, "learning_rate": 0.0001, "elbo": 4.08788, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5538939713816561}
{"step": 9450, "loss": 4.076327896118164, "grad_norm": 0.7997875213623047, "learning_rate": 0.0001, "elbo": 4.07633, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5541871921182266}
{"step": 9455, "loss": 4.084158325195313, "grad_norm": 0.7939555048942566, "learning_rate": 0.0001, "elbo": 4.08416, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5544804128547971}
{"step": 9460, "loss": 4.077918243408203, "grad_norm": 1.1152151823043823, "learning_rate": 0.0001, "elbo": 4.07792, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5547736335913676}
{"step": 9465, "loss": 4.086883544921875, "grad_norm": 0.7803972959518433, "learning_rate": 0.0001, "elbo": 4.08688, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5550668543279381}
{"step": 9470, "loss": 4.099790954589844, "grad_norm": 0.832543134689331, "learning_rate": 0.0001, "elbo": 4.09979, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5553600750645086}
{"step": 9475, "loss": 4.0892585754394535, "grad_norm": 0.9050208330154419, "learning_rate": 0.0001, "elbo": 4.08926, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.555653295801079}
{"step": 9480, "loss": 4.086615753173828, "grad_norm": 1.079833984375, "learning_rate": 0.0001, "elbo": 4.08662, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5559465165376496}
{"step": 9485, "loss": 4.079948425292969, "grad_norm": 0.7555973529815674, "learning_rate": 0.0001, "elbo": 4.07995, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.55623973727422}
{"step": 9490, "loss": 4.0951789855957035, "grad_norm": 0.7383574843406677, "learning_rate": 0.0001, "elbo": 4.09518, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5565329580107905}
{"step": 9495, "loss": 4.094287872314453, "grad_norm": 0.7541618943214417, "learning_rate": 0.0001, "elbo": 4.09429, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5568261787473611}
{"step": 9500, "loss": 4.079123687744141, "grad_norm": 0.8107120990753174, "learning_rate": 0.0001, "elbo": 4.07912, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5571193994839315}
{"step": 9505, "loss": 4.079604721069336, "grad_norm": 0.7904237508773804, "learning_rate": 0.0001, "elbo": 4.0796, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.557412620220502}
{"step": 9505, "loss": 4.079604721069336, "grad_norm": 0.7904237508773804, "learning_rate": 0.0001, "elbo": 4.0796, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.557412620220502}
{"step": 9510, "loss": 4.095171356201172, "grad_norm": 0.7776443958282471, "learning_rate": 0.0001, "elbo": 4.09517, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5577058409570724}
{"step": 9515, "loss": 4.099558258056641, "grad_norm": 0.7778709530830383, "learning_rate": 0.0001, "elbo": 4.09956, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.557999061693643}
{"step": 9520, "loss": 4.093737030029297, "grad_norm": 1.0221656560897827, "learning_rate": 0.0001, "elbo": 4.09374, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5582922824302134}
{"step": 9525, "loss": 4.092362976074218, "grad_norm": 0.7838019728660583, "learning_rate": 0.0001, "elbo": 4.09236, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5585855031667839}
{"step": 9530, "loss": 4.103594589233398, "grad_norm": 0.7989736795425415, "learning_rate": 0.0001, "elbo": 4.10359, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5588787239033545}
{"step": 9535, "loss": 4.085788726806641, "grad_norm": 0.7663798332214355, "learning_rate": 0.0001, "elbo": 4.08579, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5591719446399249}
{"step": 9540, "loss": 4.105133056640625, "grad_norm": 0.7604300379753113, "learning_rate": 0.0001, "elbo": 4.10513, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5594651653764954}
{"step": 9545, "loss": 4.08824462890625, "grad_norm": 0.8060073852539062, "learning_rate": 0.0001, "elbo": 4.08824, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5597583861130659}
{"step": 9550, "loss": 4.090131378173828, "grad_norm": 0.7054162621498108, "learning_rate": 0.0001, "elbo": 4.09013, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5600516068496364}
{"step": 9555, "loss": 4.092613983154297, "grad_norm": 0.7958646416664124, "learning_rate": 0.0001, "elbo": 4.09261, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5603448275862069}
{"step": 9560, "loss": 4.090444946289063, "grad_norm": 0.781687319278717, "learning_rate": 0.0001, "elbo": 4.09044, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5606380483227774}
{"step": 9565, "loss": 4.08486328125, "grad_norm": 0.7390739321708679, "learning_rate": 0.0001, "elbo": 4.08486, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5609312690593479}
{"step": 9570, "loss": 4.095033264160156, "grad_norm": 0.8250897526741028, "learning_rate": 0.0001, "elbo": 4.09503, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5612244897959183}
{"step": 9575, "loss": 4.087022018432617, "grad_norm": 0.823697566986084, "learning_rate": 0.0001, "elbo": 4.08702, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5615177105324889}
{"step": 9580, "loss": 4.099806213378907, "grad_norm": 0.8028882741928101, "learning_rate": 0.0001, "elbo": 4.09981, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5618109312690593}
{"step": 9585, "loss": 4.1081398010253904, "grad_norm": 0.8426643013954163, "learning_rate": 0.0001, "elbo": 4.10814, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5621041520056298}
{"step": 9590, "loss": 4.089661407470703, "grad_norm": 0.9949518442153931, "learning_rate": 0.0001, "elbo": 4.08966, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5623973727422004}
{"step": 9595, "loss": 4.0952606201171875, "grad_norm": 0.7103604674339294, "learning_rate": 0.0001, "elbo": 4.09526, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5626905934787708}
{"step": 9600, "loss": 4.093377685546875, "grad_norm": 0.7520995140075684, "learning_rate": 0.0001, "elbo": 4.09338, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5629838142153413}
{"step": 9605, "loss": 4.091032409667969, "grad_norm": 0.8413224220275879, "learning_rate": 0.0001, "elbo": 4.09103, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5632770349519118}
{"step": 9610, "loss": 4.077729797363281, "grad_norm": 0.7634052634239197, "learning_rate": 0.0001, "elbo": 4.07773, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5635702556884823}
{"step": 9615, "loss": 4.070901489257812, "grad_norm": 0.7348796725273132, "learning_rate": 0.0001, "elbo": 4.0709, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5638634764250527}
{"step": 9620, "loss": 4.0869606018066404, "grad_norm": 0.821052074432373, "learning_rate": 0.0001, "elbo": 4.08696, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5641566971616233}
{"step": 9625, "loss": 4.087129211425781, "grad_norm": 0.8191456198692322, "learning_rate": 0.0001, "elbo": 4.08713, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5644499178981938}
{"step": 9625, "loss": 4.087129211425781, "grad_norm": 0.8191456198692322, "learning_rate": 0.0001, "elbo": 4.08713, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5644499178981938}
{"step": 9630, "loss": 4.082928466796875, "grad_norm": 0.864578127861023, "learning_rate": 0.0001, "elbo": 4.08293, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5647431386347642}
{"step": 9635, "loss": 4.099586105346679, "grad_norm": 0.8836309909820557, "learning_rate": 0.0001, "elbo": 4.09959, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5650363593713348}
{"step": 9640, "loss": 4.071971130371094, "grad_norm": 0.7431219816207886, "learning_rate": 0.0001, "elbo": 4.07197, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5653295801079052}
{"step": 9645, "loss": 4.066053771972657, "grad_norm": 0.8465880751609802, "learning_rate": 0.0001, "elbo": 4.06605, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5656228008444757}
{"step": 9650, "loss": 4.092847442626953, "grad_norm": 0.7245523929595947, "learning_rate": 0.0001, "elbo": 4.09285, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5659160215810463}
{"step": 9655, "loss": 4.084667205810547, "grad_norm": 0.8178381323814392, "learning_rate": 0.0001, "elbo": 4.08467, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5662092423176167}
{"step": 9660, "loss": 4.07962646484375, "grad_norm": 0.7763703465461731, "learning_rate": 0.0001, "elbo": 4.07963, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5665024630541872}
{"step": 9665, "loss": 4.086495590209961, "grad_norm": 0.7095606327056885, "learning_rate": 0.0001, "elbo": 4.0865, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5667956837907577}
{"step": 9670, "loss": 4.061275482177734, "grad_norm": 0.7107833027839661, "learning_rate": 0.0001, "elbo": 4.06128, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5670889045273282}
{"step": 9675, "loss": 4.097374725341797, "grad_norm": 0.7614464163780212, "learning_rate": 0.0001, "elbo": 4.09738, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5673821252638986}
{"step": 9680, "loss": 4.082648468017578, "grad_norm": 0.7080006003379822, "learning_rate": 0.0001, "elbo": 4.08265, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5676753460004692}
{"step": 9685, "loss": 4.094915771484375, "grad_norm": 0.8652099370956421, "learning_rate": 0.0001, "elbo": 4.09492, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5679685667370397}
{"step": 9690, "loss": 4.069500350952149, "grad_norm": 0.7944786548614502, "learning_rate": 0.0001, "elbo": 4.0695, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5682617874736101}
{"step": 9695, "loss": 4.068344497680664, "grad_norm": 1.2298930883407593, "learning_rate": 0.0001, "elbo": 4.06834, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5685550082101807}
{"step": 9700, "loss": 4.095315551757812, "grad_norm": 0.7898703217506409, "learning_rate": 0.0001, "elbo": 4.09532, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5688482289467511}
{"step": 9705, "loss": 4.0788219451904295, "grad_norm": 0.8141958713531494, "learning_rate": 0.0001, "elbo": 4.07882, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5691414496833216}
{"step": 9710, "loss": 4.0847312927246096, "grad_norm": 0.9117618799209595, "learning_rate": 0.0001, "elbo": 4.08473, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.569434670419892}
{"step": 9715, "loss": 4.085336303710937, "grad_norm": 0.8448883891105652, "learning_rate": 0.0001, "elbo": 4.08534, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5697278911564626}
{"step": 9720, "loss": 4.084850311279297, "grad_norm": 0.7401039004325867, "learning_rate": 0.0001, "elbo": 4.08485, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5700211118930331}
{"step": 9720, "loss": 4.084850311279297, "grad_norm": 0.7401039004325867, "learning_rate": 0.0001, "elbo": 4.08485, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5700211118930331}
{"step": 9725, "loss": 4.071549606323242, "grad_norm": 0.7555489540100098, "learning_rate": 0.0001, "elbo": 4.07155, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5703143326296035}
{"step": 9730, "loss": 4.091943740844727, "grad_norm": 0.76727294921875, "learning_rate": 0.0001, "elbo": 4.09194, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5706075533661741}
{"step": 9735, "loss": 4.075977325439453, "grad_norm": 0.7797159552574158, "learning_rate": 0.0001, "elbo": 4.07598, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5709007741027445}
{"step": 9740, "loss": 4.102748107910156, "grad_norm": 1.231652855873108, "learning_rate": 0.0001, "elbo": 4.10275, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.571193994839315}
{"step": 9745, "loss": 4.063438415527344, "grad_norm": 0.773464024066925, "learning_rate": 0.0001, "elbo": 4.06344, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5714872155758856}
{"step": 9750, "loss": 4.092020416259766, "grad_norm": 0.8456698656082153, "learning_rate": 0.0001, "elbo": 4.09202, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.571780436312456}
{"step": 9755, "loss": 4.078538513183593, "grad_norm": 0.797070324420929, "learning_rate": 0.0001, "elbo": 4.07854, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5720736570490265}
{"step": 9760, "loss": 4.091987991333008, "grad_norm": 0.7341258525848389, "learning_rate": 0.0001, "elbo": 4.09199, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.572366877785597}
{"step": 9765, "loss": 4.07495346069336, "grad_norm": 0.7350015044212341, "learning_rate": 0.0001, "elbo": 4.07495, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5726600985221675}
{"step": 9770, "loss": 4.0669097900390625, "grad_norm": 0.7382321357727051, "learning_rate": 0.0001, "elbo": 4.06691, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5729533192587379}
{"step": 9775, "loss": 4.088314056396484, "grad_norm": 0.9271866083145142, "learning_rate": 0.0001, "elbo": 4.08831, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5732465399953085}
{"step": 9780, "loss": 4.080732345581055, "grad_norm": 0.7294750809669495, "learning_rate": 0.0001, "elbo": 4.08073, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.573539760731879}
{"step": 9785, "loss": 4.069842910766601, "grad_norm": 0.8806512355804443, "learning_rate": 0.0001, "elbo": 4.06984, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5738329814684494}
{"step": 9785, "loss": 4.069842910766601, "grad_norm": 0.8806512355804443, "learning_rate": 0.0001, "elbo": 4.06984, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5738329814684494}
{"step": 9790, "loss": 4.081464004516602, "grad_norm": 0.8154478669166565, "learning_rate": 0.0001, "elbo": 4.08146, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.57412620220502}
{"step": 9795, "loss": 4.076124572753907, "grad_norm": 0.6945603489875793, "learning_rate": 0.0001, "elbo": 4.07612, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5744194229415904}
{"step": 9800, "loss": 4.1107734680175785, "grad_norm": 0.7832258939743042, "learning_rate": 0.0001, "elbo": 4.11077, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5747126436781609}
{"step": 9805, "loss": 4.073836898803711, "grad_norm": 0.8133929371833801, "learning_rate": 0.0001, "elbo": 4.07384, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5750058644147314}
{"step": 9810, "loss": 4.086754608154297, "grad_norm": 0.783818781375885, "learning_rate": 0.0001, "elbo": 4.08675, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5752990851513019}
{"step": 9815, "loss": 4.087649536132813, "grad_norm": 0.7386660575866699, "learning_rate": 0.0001, "elbo": 4.08765, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5755923058878724}
{"step": 9820, "loss": 4.070503997802734, "grad_norm": 0.793114185333252, "learning_rate": 0.0001, "elbo": 4.0705, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5758855266244429}
{"step": 9825, "loss": 4.07348861694336, "grad_norm": 0.756540060043335, "learning_rate": 0.0001, "elbo": 4.07349, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5761787473610134}
{"step": 9825, "loss": 4.07348861694336, "grad_norm": 0.756540060043335, "learning_rate": 0.0001, "elbo": 4.07349, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5761787473610134}
{"step": 9830, "loss": 4.069200897216797, "grad_norm": 1.0783472061157227, "learning_rate": 0.0001, "elbo": 4.0692, "mask_frac": 0.5013, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5764719680975838}
{"step": 9835, "loss": 4.077165222167968, "grad_norm": 0.8145726919174194, "learning_rate": 0.0001, "elbo": 4.07717, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5767651888341544}
{"step": 9840, "loss": 4.080276870727539, "grad_norm": 0.8409931063652039, "learning_rate": 0.0001, "elbo": 4.08028, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5770584095707249}
{"step": 9845, "loss": 4.065052032470703, "grad_norm": 0.7175452709197998, "learning_rate": 0.0001, "elbo": 4.06505, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5773516303072953}
{"step": 9850, "loss": 4.065058898925781, "grad_norm": 0.7371772527694702, "learning_rate": 0.0001, "elbo": 4.06506, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5776448510438659}
{"step": 9855, "loss": 4.058531951904297, "grad_norm": 0.7963671684265137, "learning_rate": 0.0001, "elbo": 4.05853, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5779380717804363}
{"step": 9860, "loss": 4.0814769744873045, "grad_norm": 0.8861827254295349, "learning_rate": 0.0001, "elbo": 4.08148, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5782312925170068}
{"step": 9865, "loss": 4.0590354919433596, "grad_norm": 0.7575666904449463, "learning_rate": 0.0001, "elbo": 4.05904, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5785245132535773}
{"step": 9870, "loss": 4.074767303466797, "grad_norm": 0.7690837383270264, "learning_rate": 0.0001, "elbo": 4.07477, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5788177339901478}
{"step": 9875, "loss": 4.075712585449219, "grad_norm": 0.837044894695282, "learning_rate": 0.0001, "elbo": 4.07571, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5791109547267183}
{"step": 9880, "loss": 4.072248077392578, "grad_norm": 0.8718556761741638, "learning_rate": 0.0001, "elbo": 4.07225, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5794041754632888}
{"step": 9885, "loss": 4.0773872375488285, "grad_norm": 0.8988029956817627, "learning_rate": 0.0001, "elbo": 4.07739, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5796973961998593}
{"step": 9890, "loss": 4.071860504150391, "grad_norm": 1.0338810682296753, "learning_rate": 0.0001, "elbo": 4.07186, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5799906169364297}
{"step": 9895, "loss": 4.08099250793457, "grad_norm": 0.7502069473266602, "learning_rate": 0.0001, "elbo": 4.08099, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5802838376730002}
{"step": 9900, "loss": 4.073158264160156, "grad_norm": 0.8589254021644592, "learning_rate": 0.0001, "elbo": 4.07316, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5805770584095707}
{"step": 9905, "loss": 4.062622833251953, "grad_norm": 0.7418677806854248, "learning_rate": 0.0001, "elbo": 4.06262, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5808702791461412}
{"step": 9910, "loss": 4.072069931030273, "grad_norm": 1.0283863544464111, "learning_rate": 0.0001, "elbo": 4.07207, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5811634998827117}
{"step": 9915, "loss": 4.074941253662109, "grad_norm": 0.9032464623451233, "learning_rate": 0.0001, "elbo": 4.07494, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5814567206192822}
{"step": 9920, "loss": 4.057003784179687, "grad_norm": 0.7533677816390991, "learning_rate": 0.0001, "elbo": 4.057, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5817499413558527}
{"step": 9925, "loss": 4.066205978393555, "grad_norm": 0.7731001377105713, "learning_rate": 0.0001, "elbo": 4.06621, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5820431620924231}
{"step": 9930, "loss": 4.083917236328125, "grad_norm": 0.7325822710990906, "learning_rate": 0.0001, "elbo": 4.08392, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5823363828289937}
{"step": 9935, "loss": 4.092123794555664, "grad_norm": 0.7460812330245972, "learning_rate": 0.0001, "elbo": 4.09212, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5826296035655641}
{"step": 9940, "loss": 4.061941528320313, "grad_norm": 0.7840985059738159, "learning_rate": 0.0001, "elbo": 4.06194, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5829228243021346}
{"step": 9945, "loss": 4.076819610595703, "grad_norm": 0.7407606244087219, "learning_rate": 0.0001, "elbo": 4.07682, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5832160450387052}
{"step": 9950, "loss": 4.063754272460938, "grad_norm": 0.7150115966796875, "learning_rate": 0.0001, "elbo": 4.06375, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5835092657752756}
{"step": 9955, "loss": 4.082320022583008, "grad_norm": 0.7404932975769043, "learning_rate": 0.0001, "elbo": 4.08232, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5838024865118461}
{"step": 9960, "loss": 4.081979370117187, "grad_norm": 0.8010625839233398, "learning_rate": 0.0001, "elbo": 4.08198, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5840957072484166}
{"step": 9965, "loss": 4.0765235900878904, "grad_norm": 0.7274087071418762, "learning_rate": 0.0001, "elbo": 4.07652, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5843889279849871}
{"step": 9970, "loss": 4.071776580810547, "grad_norm": 0.7683598399162292, "learning_rate": 0.0001, "elbo": 4.07178, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5846821487215575}
{"step": 9975, "loss": 4.087747955322266, "grad_norm": 0.7447288036346436, "learning_rate": 0.0001, "elbo": 4.08775, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5849753694581281}
{"step": 9980, "loss": 4.053016662597656, "grad_norm": 0.7447044849395752, "learning_rate": 0.0001, "elbo": 4.05302, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5852685901946986}
{"step": 9985, "loss": 4.081682968139648, "grad_norm": 0.7887129187583923, "learning_rate": 0.0001, "elbo": 4.08168, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.585561810931269}
{"step": 9990, "loss": 4.081962585449219, "grad_norm": 0.7485388517379761, "learning_rate": 0.0001, "elbo": 4.08196, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5858550316678396}
{"step": 9995, "loss": 4.071574401855469, "grad_norm": 0.7367774248123169, "learning_rate": 0.0001, "elbo": 4.07157, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.58614825240441}
{"step": 10000, "loss": 4.0720367431640625, "grad_norm": 0.8239917159080505, "learning_rate": 0.0001, "elbo": 4.07204, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5864414731409805}
{"step": 10005, "loss": 4.062506103515625, "grad_norm": 0.7715042233467102, "learning_rate": 0.0001, "elbo": 4.06251, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5867346938775511}
{"step": 10010, "loss": 4.076930236816406, "grad_norm": 0.7877938151359558, "learning_rate": 0.0001, "elbo": 4.07693, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5870279146141215}
{"step": 10015, "loss": 4.084507751464844, "grad_norm": 0.8278393745422363, "learning_rate": 0.0001, "elbo": 4.08451, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.587321135350692}
{"step": 10020, "loss": 4.060517501831055, "grad_norm": 0.9581581354141235, "learning_rate": 0.0001, "elbo": 4.06052, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5876143560872625}
{"step": 10020, "loss": 4.060517501831055, "grad_norm": 0.9581581354141235, "learning_rate": 0.0001, "elbo": 4.06052, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5876143560872625}
{"step": 10025, "loss": 4.0602867126464846, "grad_norm": 0.8830257058143616, "learning_rate": 0.0001, "elbo": 4.06029, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.587907576823833}
{"step": 10030, "loss": 4.077664566040039, "grad_norm": 0.773468554019928, "learning_rate": 0.0001, "elbo": 4.07766, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5882007975604034}
{"step": 10030, "loss": 4.077664566040039, "grad_norm": 0.773468554019928, "learning_rate": 0.0001, "elbo": 4.07766, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5882007975604034}
{"step": 10035, "loss": 4.0620574951171875, "grad_norm": 0.7723097801208496, "learning_rate": 0.0001, "elbo": 4.06206, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.588494018296974}
{"step": 10040, "loss": 4.075764465332031, "grad_norm": 0.8374214172363281, "learning_rate": 0.0001, "elbo": 4.07576, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5887872390335445}
{"step": 10045, "loss": 4.065089416503906, "grad_norm": 0.7929126620292664, "learning_rate": 0.0001, "elbo": 4.06509, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5890804597701149}
{"step": 10050, "loss": 4.069870376586914, "grad_norm": 0.8028703331947327, "learning_rate": 0.0001, "elbo": 4.06987, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5893736805066855}
{"step": 10055, "loss": 4.06951675415039, "grad_norm": 0.8062699437141418, "learning_rate": 0.0001, "elbo": 4.06952, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5896669012432559}
{"step": 10060, "loss": 4.064991760253906, "grad_norm": 0.751223623752594, "learning_rate": 0.0001, "elbo": 4.06499, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5899601219798264}
{"step": 10065, "loss": 4.090282440185547, "grad_norm": 0.7503570318222046, "learning_rate": 0.0001, "elbo": 4.09028, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.590253342716397}
{"step": 10070, "loss": 4.076855087280274, "grad_norm": 0.7678165435791016, "learning_rate": 0.0001, "elbo": 4.07686, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5905465634529674}
{"step": 10075, "loss": 4.071107482910156, "grad_norm": 0.8932929635047913, "learning_rate": 0.0001, "elbo": 4.07111, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5908397841895379}
{"step": 10080, "loss": 4.076911544799804, "grad_norm": 0.8054429292678833, "learning_rate": 0.0001, "elbo": 4.07691, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5911330049261084}
{"step": 10085, "loss": 4.066293334960937, "grad_norm": 0.7410283088684082, "learning_rate": 0.0001, "elbo": 4.06629, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5914262256626789}
{"step": 10085, "loss": 4.066293334960937, "grad_norm": 0.7410283088684082, "learning_rate": 0.0001, "elbo": 4.06629, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5914262256626789}
{"step": 10090, "loss": 4.080894470214844, "grad_norm": 0.8492499589920044, "learning_rate": 0.0001, "elbo": 4.08089, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5917194463992493}
{"step": 10095, "loss": 4.063566589355469, "grad_norm": 1.2416452169418335, "learning_rate": 0.0001, "elbo": 4.06357, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5920126671358198}
{"step": 10100, "loss": 4.076744842529297, "grad_norm": 0.7764478325843811, "learning_rate": 0.0001, "elbo": 4.07674, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5923058878723904}
{"step": 10105, "loss": 4.061030578613281, "grad_norm": 0.7770340442657471, "learning_rate": 0.0001, "elbo": 4.06103, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5925991086089608}
{"step": 10110, "loss": 4.073935699462891, "grad_norm": 0.7485775351524353, "learning_rate": 0.0001, "elbo": 4.07394, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5928923293455313}
{"step": 10115, "loss": 4.069566345214843, "grad_norm": 0.9367078542709351, "learning_rate": 0.0001, "elbo": 4.06957, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5931855500821018}
{"step": 10120, "loss": 4.064736175537109, "grad_norm": 0.7885707020759583, "learning_rate": 0.0001, "elbo": 4.06474, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5934787708186723}
{"step": 10125, "loss": 4.068714904785156, "grad_norm": 0.7091343998908997, "learning_rate": 0.0001, "elbo": 4.06871, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5937719915552427}
{"step": 10130, "loss": 4.048285293579101, "grad_norm": 0.6780269742012024, "learning_rate": 0.0001, "elbo": 4.04829, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5940652122918133}
{"step": 10135, "loss": 4.071144104003906, "grad_norm": 0.7024767398834229, "learning_rate": 0.0001, "elbo": 4.07114, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5943584330283838}
{"step": 10140, "loss": 4.061550903320312, "grad_norm": 0.823898196220398, "learning_rate": 0.0001, "elbo": 4.06155, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5946516537649542}
{"step": 10145, "loss": 4.05437126159668, "grad_norm": 0.8617291450500488, "learning_rate": 0.0001, "elbo": 4.05437, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5949448745015248}
{"step": 10150, "loss": 4.056629180908203, "grad_norm": 0.7533734440803528, "learning_rate": 0.0001, "elbo": 4.05663, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5952380952380952}
{"step": 10155, "loss": 4.069499206542969, "grad_norm": 0.8062280416488647, "learning_rate": 0.0001, "elbo": 4.0695, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5955313159746657}
{"step": 10160, "loss": 4.06151123046875, "grad_norm": 0.7669251561164856, "learning_rate": 0.0001, "elbo": 4.06151, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5958245367112363}
{"step": 10165, "loss": 4.070487213134766, "grad_norm": 0.8364665508270264, "learning_rate": 0.0001, "elbo": 4.07049, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5961177574478067}
{"step": 10170, "loss": 4.078385925292968, "grad_norm": 0.8398986458778381, "learning_rate": 0.0001, "elbo": 4.07839, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5964109781843772}
{"step": 10175, "loss": 4.060748291015625, "grad_norm": 0.7289391756057739, "learning_rate": 0.0001, "elbo": 4.06075, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5967041989209477}
{"step": 10180, "loss": 4.0514472961425785, "grad_norm": 0.7807574272155762, "learning_rate": 0.0001, "elbo": 4.05145, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5969974196575182}
{"step": 10185, "loss": 4.056633758544922, "grad_norm": 0.7414342761039734, "learning_rate": 0.0001, "elbo": 4.05663, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5972906403940886}
{"step": 10190, "loss": 4.049014282226563, "grad_norm": 0.8151629567146301, "learning_rate": 0.0001, "elbo": 4.04901, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5975838611306592}
{"step": 10195, "loss": 4.084279632568359, "grad_norm": 0.8527905344963074, "learning_rate": 0.0001, "elbo": 4.08428, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5978770818672297}
{"step": 10200, "loss": 4.078264617919922, "grad_norm": 0.7874825596809387, "learning_rate": 0.0001, "elbo": 4.07826, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5981703026038001}
{"step": 10205, "loss": 4.044626617431641, "grad_norm": 0.8378686308860779, "learning_rate": 0.0001, "elbo": 4.04463, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5984635233403707}
{"step": 10210, "loss": 4.082772827148437, "grad_norm": 0.7479903101921082, "learning_rate": 0.0001, "elbo": 4.08277, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5987567440769411}
{"step": 10210, "loss": 4.082772827148437, "grad_norm": 0.7479903101921082, "learning_rate": 0.0001, "elbo": 4.08277, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5987567440769411}
{"step": 10215, "loss": 4.041719436645508, "grad_norm": 0.8045471906661987, "learning_rate": 0.0001, "elbo": 4.04172, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5990499648135116}
{"step": 10220, "loss": 4.05225830078125, "grad_norm": 0.8644149899482727, "learning_rate": 0.0001, "elbo": 4.05226, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5993431855500821}
{"step": 10225, "loss": 4.061677169799805, "grad_norm": 0.749817967414856, "learning_rate": 0.0001, "elbo": 4.06168, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.5996364062866526}
{"step": 10230, "loss": 4.063823699951172, "grad_norm": 0.7366182804107666, "learning_rate": 0.0001, "elbo": 4.06382, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.599929627023223}
{"step": 10235, "loss": 4.056197738647461, "grad_norm": 0.7723664045333862, "learning_rate": 0.0001, "elbo": 4.0562, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6002228477597936}
{"step": 10240, "loss": 4.052301025390625, "grad_norm": 0.7501629590988159, "learning_rate": 0.0001, "elbo": 4.0523, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6005160684963641}
{"step": 10245, "loss": 4.059630966186523, "grad_norm": 0.8329192996025085, "learning_rate": 0.0001, "elbo": 4.05963, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6008092892329345}
{"step": 10250, "loss": 4.07214469909668, "grad_norm": 0.7856448292732239, "learning_rate": 0.0001, "elbo": 4.07214, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6011025099695051}
{"step": 10255, "loss": 4.066184616088867, "grad_norm": 0.8691504001617432, "learning_rate": 0.0001, "elbo": 4.06618, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6013957307060755}
{"step": 10260, "loss": 4.0643455505371096, "grad_norm": 0.8082539439201355, "learning_rate": 0.0001, "elbo": 4.06435, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.601688951442646}
{"step": 10265, "loss": 4.083202362060547, "grad_norm": 0.7854197025299072, "learning_rate": 0.0001, "elbo": 4.0832, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6019821721792166}
{"step": 10270, "loss": 4.0605110168457035, "grad_norm": 0.7582296133041382, "learning_rate": 0.0001, "elbo": 4.06051, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.602275392915787}
{"step": 10275, "loss": 4.058817291259766, "grad_norm": 0.8182489275932312, "learning_rate": 0.0001, "elbo": 4.05882, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6025686136523575}
{"step": 10280, "loss": 4.070046615600586, "grad_norm": 1.0626459121704102, "learning_rate": 0.0001, "elbo": 4.07005, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6028618343889279}
{"step": 10285, "loss": 4.067112731933594, "grad_norm": 0.8124667406082153, "learning_rate": 0.0001, "elbo": 4.06711, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6031550551254985}
{"step": 10290, "loss": 4.068159866333008, "grad_norm": 0.7757775783538818, "learning_rate": 0.0001, "elbo": 4.06816, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.603448275862069}
{"step": 10295, "loss": 4.052891540527344, "grad_norm": 0.7462929487228394, "learning_rate": 0.0001, "elbo": 4.05289, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6037414965986394}
{"step": 10300, "loss": 4.0493427276611325, "grad_norm": 0.9552906155586243, "learning_rate": 0.0001, "elbo": 4.04934, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.60403471733521}
{"step": 10305, "loss": 4.041222381591797, "grad_norm": 0.8273400068283081, "learning_rate": 0.0001, "elbo": 4.04122, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6043279380717804}
{"step": 10310, "loss": 4.050606918334961, "grad_norm": 0.7950786352157593, "learning_rate": 0.0001, "elbo": 4.05061, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6046211588083509}
{"step": 10315, "loss": 4.0529624938964846, "grad_norm": 0.714389443397522, "learning_rate": 0.0001, "elbo": 4.05296, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6049143795449214}
{"step": 10320, "loss": 4.0586692810058596, "grad_norm": 0.7368884086608887, "learning_rate": 0.0001, "elbo": 4.05867, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6052076002814919}
{"step": 10325, "loss": 4.054436492919922, "grad_norm": 0.7669080495834351, "learning_rate": 0.0001, "elbo": 4.05444, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6055008210180624}
{"step": 10330, "loss": 4.048781204223633, "grad_norm": 0.7859706878662109, "learning_rate": 0.0001, "elbo": 4.04878, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6057940417546329}
{"step": 10330, "loss": 4.048781204223633, "grad_norm": 0.7859706878662109, "learning_rate": 0.0001, "elbo": 4.04878, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6057940417546329}
{"step": 10335, "loss": 4.064472198486328, "grad_norm": 0.7708370685577393, "learning_rate": 0.0001, "elbo": 4.06447, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6060872624912034}
{"step": 10340, "loss": 4.057465362548828, "grad_norm": 0.6485885977745056, "learning_rate": 0.0001, "elbo": 4.05747, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6063804832277738}
{"step": 10345, "loss": 4.04845085144043, "grad_norm": 0.8371037244796753, "learning_rate": 0.0001, "elbo": 4.04845, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6066737039643444}
{"step": 10350, "loss": 4.041520690917968, "grad_norm": 0.774988055229187, "learning_rate": 0.0001, "elbo": 4.04152, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6069669247009148}
{"step": 10355, "loss": 4.0634925842285154, "grad_norm": 0.7922829389572144, "learning_rate": 0.0001, "elbo": 4.06349, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6072601454374853}
{"step": 10360, "loss": 4.065969848632813, "grad_norm": 0.7939398288726807, "learning_rate": 0.0001, "elbo": 4.06597, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6075533661740559}
{"step": 10365, "loss": 4.057954788208008, "grad_norm": 0.837628185749054, "learning_rate": 0.0001, "elbo": 4.05795, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6078465869106263}
{"step": 10370, "loss": 4.046866607666016, "grad_norm": 0.697968065738678, "learning_rate": 0.0001, "elbo": 4.04687, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6081398076471968}
{"step": 10375, "loss": 4.052522277832031, "grad_norm": 0.7367140650749207, "learning_rate": 0.0001, "elbo": 4.05252, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6084330283837673}
{"step": 10380, "loss": 4.0369415283203125, "grad_norm": 0.7784507274627686, "learning_rate": 0.0001, "elbo": 4.03694, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6087262491203378}
{"step": 10385, "loss": 4.053067016601562, "grad_norm": 0.7653777003288269, "learning_rate": 0.0001, "elbo": 4.05307, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6090194698569082}
{"step": 10390, "loss": 4.040096282958984, "grad_norm": 0.7792329788208008, "learning_rate": 0.0001, "elbo": 4.0401, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6093126905934788}
{"step": 10395, "loss": 4.0473167419433596, "grad_norm": 0.7838742733001709, "learning_rate": 0.0001, "elbo": 4.04732, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6096059113300493}
{"step": 10400, "loss": 4.0579780578613285, "grad_norm": 0.7500662803649902, "learning_rate": 0.0001, "elbo": 4.05798, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6098991320666197}
{"step": 10405, "loss": 4.059164428710938, "grad_norm": 0.8107898831367493, "learning_rate": 0.0001, "elbo": 4.05916, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6101923528031903}
{"step": 10410, "loss": 4.0620380401611325, "grad_norm": 0.7817366719245911, "learning_rate": 0.0001, "elbo": 4.06204, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6104855735397607}
{"step": 10415, "loss": 4.0530242919921875, "grad_norm": 0.8614340424537659, "learning_rate": 0.0001, "elbo": 4.05302, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6107787942763312}
{"step": 10420, "loss": 4.062590789794922, "grad_norm": 0.944159209728241, "learning_rate": 0.0001, "elbo": 4.06259, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6110720150129018}
{"step": 10425, "loss": 4.061310577392578, "grad_norm": 0.7156311273574829, "learning_rate": 0.0001, "elbo": 4.06131, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6113652357494722}
{"step": 10430, "loss": 4.053585052490234, "grad_norm": 0.8243806958198547, "learning_rate": 0.0001, "elbo": 4.05359, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6116584564860427}
{"step": 10435, "loss": 4.043406677246094, "grad_norm": 0.7099548578262329, "learning_rate": 0.0001, "elbo": 4.04341, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6119516772226132}
{"step": 10440, "loss": 4.062213134765625, "grad_norm": 0.7756883502006531, "learning_rate": 0.0001, "elbo": 4.06221, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6122448979591837}
{"step": 10445, "loss": 4.073605728149414, "grad_norm": 0.7359620928764343, "learning_rate": 0.0001, "elbo": 4.07361, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6125381186957541}
{"step": 10450, "loss": 4.0524242401123045, "grad_norm": 0.7117235660552979, "learning_rate": 0.0001, "elbo": 4.05242, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6128313394323247}
{"step": 10455, "loss": 4.0248973846435545, "grad_norm": 0.7877048850059509, "learning_rate": 0.0001, "elbo": 4.0249, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6131245601688952}
{"step": 10460, "loss": 4.045462799072266, "grad_norm": 0.734789252281189, "learning_rate": 0.0001, "elbo": 4.04546, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6134177809054656}
{"step": 10465, "loss": 4.050471496582031, "grad_norm": 0.8093528747558594, "learning_rate": 0.0001, "elbo": 4.05047, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6137110016420362}
{"step": 10470, "loss": 4.030262756347656, "grad_norm": 0.8769282698631287, "learning_rate": 0.0001, "elbo": 4.03026, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6140042223786066}
{"step": 10475, "loss": 4.0643562316894535, "grad_norm": 0.7088821530342102, "learning_rate": 0.0001, "elbo": 4.06436, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6142974431151771}
{"step": 10480, "loss": 4.055810546875, "grad_norm": 0.9078906178474426, "learning_rate": 0.0001, "elbo": 4.05581, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6145906638517475}
{"step": 10485, "loss": 4.065283584594726, "grad_norm": 0.833906888961792, "learning_rate": 0.0001, "elbo": 4.06528, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6148838845883181}
{"step": 10490, "loss": 4.061245346069336, "grad_norm": 0.9055723547935486, "learning_rate": 0.0001, "elbo": 4.06125, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6151771053248886}
{"step": 10495, "loss": 4.0596778869628904, "grad_norm": 0.7567042112350464, "learning_rate": 0.0001, "elbo": 4.05968, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.615470326061459}
{"step": 10500, "loss": 4.0355072021484375, "grad_norm": 0.8108766674995422, "learning_rate": 0.0001, "elbo": 4.03551, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6157635467980296}
{"step": 10505, "loss": 4.054216384887695, "grad_norm": 1.0911539793014526, "learning_rate": 0.0001, "elbo": 4.05422, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6160567675346}
{"step": 10510, "loss": 4.032448577880859, "grad_norm": 0.7038781046867371, "learning_rate": 0.0001, "elbo": 4.03245, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6163499882711705}
{"step": 10515, "loss": 4.047151184082031, "grad_norm": 0.7587671279907227, "learning_rate": 0.0001, "elbo": 4.04715, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.616643209007741}
{"step": 10520, "loss": 4.067290878295898, "grad_norm": 0.6947284936904907, "learning_rate": 0.0001, "elbo": 4.06729, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6169364297443115}
{"step": 10525, "loss": 4.0548828125, "grad_norm": 0.7990607619285583, "learning_rate": 0.0001, "elbo": 4.05488, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.617229650480882}
{"step": 10530, "loss": 4.05779914855957, "grad_norm": 0.761326253414154, "learning_rate": 0.0001, "elbo": 4.0578, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6175228712174525}
{"step": 10535, "loss": 4.047387313842774, "grad_norm": 0.9290179014205933, "learning_rate": 0.0001, "elbo": 4.04739, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.617816091954023}
{"step": 10540, "loss": 4.059609985351562, "grad_norm": 0.7708966135978699, "learning_rate": 0.0001, "elbo": 4.05961, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6181093126905934}
{"step": 10545, "loss": 4.042425537109375, "grad_norm": 0.7818138599395752, "learning_rate": 0.0001, "elbo": 4.04243, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.618402533427164}
{"step": 10550, "loss": 4.057269287109375, "grad_norm": 0.7649484276771545, "learning_rate": 0.0001, "elbo": 4.05727, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6186957541637345}
{"step": 10555, "loss": 4.042432022094727, "grad_norm": 0.7869093418121338, "learning_rate": 0.0001, "elbo": 4.04243, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6189889749003049}
{"step": 10560, "loss": 4.051918411254883, "grad_norm": 0.7329980731010437, "learning_rate": 0.0001, "elbo": 4.05192, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6192821956368755}
{"step": 10565, "loss": 4.0499919891357425, "grad_norm": 0.9132257699966431, "learning_rate": 0.0001, "elbo": 4.04999, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6195754163734459}
{"step": 10570, "loss": 4.046018981933594, "grad_norm": 1.0296545028686523, "learning_rate": 0.0001, "elbo": 4.04602, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6198686371100164}
{"step": 10575, "loss": 4.046282196044922, "grad_norm": 0.784378170967102, "learning_rate": 0.0001, "elbo": 4.04628, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.620161857846587}
{"step": 10580, "loss": 4.047173309326172, "grad_norm": 0.8031247854232788, "learning_rate": 0.0001, "elbo": 4.04717, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6204550785831574}
{"step": 10585, "loss": 4.048261642456055, "grad_norm": 0.7881636619567871, "learning_rate": 0.0001, "elbo": 4.04826, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6207482993197279}
{"step": 10590, "loss": 4.065301132202149, "grad_norm": 0.7216522097587585, "learning_rate": 0.0001, "elbo": 4.0653, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6210415200562984}
{"step": 10595, "loss": 4.058834838867187, "grad_norm": 0.826532781124115, "learning_rate": 0.0001, "elbo": 4.05883, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6213347407928689}
{"step": 10600, "loss": 4.043629455566406, "grad_norm": 0.7541577219963074, "learning_rate": 0.0001, "elbo": 4.04363, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6216279615294393}
{"step": 10605, "loss": 4.031443786621094, "grad_norm": 0.7598041296005249, "learning_rate": 0.0001, "elbo": 4.03144, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6219211822660099}
{"step": 10605, "loss": 4.031443786621094, "grad_norm": 0.7598041296005249, "learning_rate": 0.0001, "elbo": 4.03144, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6219211822660099}
{"step": 10610, "loss": 4.070674896240234, "grad_norm": 0.7552370429039001, "learning_rate": 0.0001, "elbo": 4.07068, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6222144030025804}
{"step": 10615, "loss": 4.050088500976562, "grad_norm": 0.8031512498855591, "learning_rate": 0.0001, "elbo": 4.05009, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6225076237391508}
{"step": 10620, "loss": 4.049044418334961, "grad_norm": 0.7635412812232971, "learning_rate": 0.0001, "elbo": 4.04904, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6228008444757214}
{"step": 10625, "loss": 4.0416259765625, "grad_norm": 0.7693352699279785, "learning_rate": 0.0001, "elbo": 4.04163, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6230940652122918}
{"step": 10630, "loss": 4.057600402832032, "grad_norm": 0.7573575973510742, "learning_rate": 0.0001, "elbo": 4.0576, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6233872859488623}
{"step": 10635, "loss": 4.045446395874023, "grad_norm": 0.800489604473114, "learning_rate": 0.0001, "elbo": 4.04545, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6236805066854328}
{"step": 10640, "loss": 4.058164978027344, "grad_norm": 0.6980737447738647, "learning_rate": 0.0001, "elbo": 4.05816, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6239737274220033}
{"step": 10645, "loss": 4.045326995849609, "grad_norm": 0.7176931500434875, "learning_rate": 0.0001, "elbo": 4.04533, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6242669481585738}
{"step": 10650, "loss": 4.033374786376953, "grad_norm": 0.7016207575798035, "learning_rate": 0.0001, "elbo": 4.03337, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6245601688951443}
{"step": 10655, "loss": 4.046566390991211, "grad_norm": 0.7588128447532654, "learning_rate": 0.0001, "elbo": 4.04657, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6248533896317148}
{"step": 10660, "loss": 4.046212005615234, "grad_norm": 0.8551952838897705, "learning_rate": 0.0001, "elbo": 4.04621, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6251466103682852}
{"step": 10665, "loss": 4.036945724487305, "grad_norm": 0.8716395497322083, "learning_rate": 0.0001, "elbo": 4.03695, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6254398311048557}
{"step": 10670, "loss": 4.050093078613282, "grad_norm": 0.8063547015190125, "learning_rate": 0.0001, "elbo": 4.05009, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6257330518414262}
{"step": 10675, "loss": 4.049794006347656, "grad_norm": 0.7436598539352417, "learning_rate": 0.0001, "elbo": 4.04979, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6260262725779967}
{"step": 10680, "loss": 4.042892074584961, "grad_norm": 0.9150645136833191, "learning_rate": 0.0001, "elbo": 4.04289, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6263194933145672}
{"step": 10685, "loss": 4.051051330566406, "grad_norm": 0.721540093421936, "learning_rate": 0.0001, "elbo": 4.05105, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6266127140511377}
{"step": 10690, "loss": 4.0315101623535154, "grad_norm": 0.8326950669288635, "learning_rate": 0.0001, "elbo": 4.03151, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6269059347877082}
{"step": 10695, "loss": 4.065592575073242, "grad_norm": 0.7755587697029114, "learning_rate": 0.0001, "elbo": 4.06559, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6271991555242786}
{"step": 10700, "loss": 4.040889739990234, "grad_norm": 0.7421045303344727, "learning_rate": 0.0001, "elbo": 4.04089, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6274923762608492}
{"step": 10705, "loss": 4.056832885742187, "grad_norm": 0.753165066242218, "learning_rate": 0.0001, "elbo": 4.05683, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6277855969974196}
{"step": 10710, "loss": 4.052988052368164, "grad_norm": 0.7266404032707214, "learning_rate": 0.0001, "elbo": 4.05299, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6280788177339901}
{"step": 10715, "loss": 4.051217651367187, "grad_norm": 0.708084762096405, "learning_rate": 0.0001, "elbo": 4.05122, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6283720384705607}
{"step": 10720, "loss": 4.049945449829101, "grad_norm": 0.6933819055557251, "learning_rate": 0.0001, "elbo": 4.04995, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6286652592071311}
{"step": 10725, "loss": 4.040671539306641, "grad_norm": 0.796379804611206, "learning_rate": 0.0001, "elbo": 4.04067, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6289584799437016}
{"step": 10730, "loss": 4.049106216430664, "grad_norm": 0.7216471433639526, "learning_rate": 0.0001, "elbo": 4.04911, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6292517006802721}
{"step": 10735, "loss": 4.029452896118164, "grad_norm": 0.7384823560714722, "learning_rate": 0.0001, "elbo": 4.02945, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6295449214168426}
{"step": 10740, "loss": 4.047211456298828, "grad_norm": 0.7458154559135437, "learning_rate": 0.0001, "elbo": 4.04721, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.629838142153413}
{"step": 10745, "loss": 4.040067291259765, "grad_norm": 1.0097393989562988, "learning_rate": 0.0001, "elbo": 4.04007, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6301313628899836}
{"step": 10750, "loss": 4.0251930236816404, "grad_norm": 0.9157501459121704, "learning_rate": 0.0001, "elbo": 4.02519, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6304245836265541}
{"step": 10755, "loss": 4.032859802246094, "grad_norm": 0.856356143951416, "learning_rate": 0.0001, "elbo": 4.03286, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6307178043631245}
{"step": 10760, "loss": 4.05705680847168, "grad_norm": 1.0142096281051636, "learning_rate": 0.0001, "elbo": 4.05706, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6310110250996951}
{"step": 10765, "loss": 4.057625579833984, "grad_norm": 0.7753083109855652, "learning_rate": 0.0001, "elbo": 4.05763, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6313042458362655}
{"step": 10770, "loss": 4.0265144348144535, "grad_norm": 0.7746351361274719, "learning_rate": 0.0001, "elbo": 4.02651, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.631597466572836}
{"step": 10775, "loss": 4.050021743774414, "grad_norm": 0.8372845649719238, "learning_rate": 0.0001, "elbo": 4.05002, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6318906873094066}
{"step": 10780, "loss": 4.054494476318359, "grad_norm": 0.7536221146583557, "learning_rate": 0.0001, "elbo": 4.05449, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.632183908045977}
{"step": 10780, "loss": 4.054494476318359, "grad_norm": 0.7536221146583557, "learning_rate": 0.0001, "elbo": 4.05449, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.632183908045977}
{"step": 10785, "loss": 4.058065795898438, "grad_norm": 0.805348813533783, "learning_rate": 0.0001, "elbo": 4.05807, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6324771287825475}
{"step": 10790, "loss": 4.042734527587891, "grad_norm": 0.7331140637397766, "learning_rate": 0.0001, "elbo": 4.04273, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.632770349519118}
{"step": 10795, "loss": 4.023499298095703, "grad_norm": 0.7465380430221558, "learning_rate": 0.0001, "elbo": 4.0235, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6330635702556885}
{"step": 10800, "loss": 4.040307235717774, "grad_norm": 0.714354395866394, "learning_rate": 0.0001, "elbo": 4.04031, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.633356790992259}
{"step": 10800, "loss": 4.040307235717774, "grad_norm": 0.714354395866394, "learning_rate": 0.0001, "elbo": 4.04031, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.633356790992259}
{"step": 10805, "loss": 4.043906021118164, "grad_norm": 1.2842025756835938, "learning_rate": 0.0001, "elbo": 4.04391, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6336500117288295}
{"step": 10810, "loss": 4.053613662719727, "grad_norm": 0.7467771768569946, "learning_rate": 0.0001, "elbo": 4.05361, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6339432324654}
{"step": 10810, "loss": 4.053613662719727, "grad_norm": 0.7467771768569946, "learning_rate": 0.0001, "elbo": 4.05361, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6339432324654}
{"step": 10815, "loss": 4.066761016845703, "grad_norm": 0.7515745162963867, "learning_rate": 0.0001, "elbo": 4.06676, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6342364532019704}
{"step": 10820, "loss": 4.048279190063477, "grad_norm": 0.6980555653572083, "learning_rate": 0.0001, "elbo": 4.04828, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.634529673938541}
{"step": 10825, "loss": 4.036416625976562, "grad_norm": 0.7481781244277954, "learning_rate": 0.0001, "elbo": 4.03642, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6348228946751114}
{"step": 10830, "loss": 4.0163108825683596, "grad_norm": 0.8230010867118835, "learning_rate": 0.0001, "elbo": 4.01631, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6351161154116819}
{"step": 10835, "loss": 4.0274311065673825, "grad_norm": 0.7946995496749878, "learning_rate": 0.0001, "elbo": 4.02743, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6354093361482525}
{"step": 10840, "loss": 4.039156341552735, "grad_norm": 0.7630744576454163, "learning_rate": 0.0001, "elbo": 4.03916, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6357025568848229}
{"step": 10845, "loss": 4.022434997558594, "grad_norm": 0.7190455198287964, "learning_rate": 0.0001, "elbo": 4.02243, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6359957776213934}
{"step": 10850, "loss": 4.053525161743164, "grad_norm": 0.808498203754425, "learning_rate": 0.0001, "elbo": 4.05352, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6362889983579638}
{"step": 10855, "loss": 4.020359802246094, "grad_norm": 0.7626382112503052, "learning_rate": 0.0001, "elbo": 4.02036, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6365822190945344}
{"step": 10860, "loss": 4.032855224609375, "grad_norm": 0.8112645745277405, "learning_rate": 0.0001, "elbo": 4.03286, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6368754398311048}
{"step": 10865, "loss": 4.04193115234375, "grad_norm": 0.7454619407653809, "learning_rate": 0.0001, "elbo": 4.04193, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6371686605676753}
{"step": 10870, "loss": 4.040616607666015, "grad_norm": 0.7531948685646057, "learning_rate": 0.0001, "elbo": 4.04062, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6374618813042459}
{"step": 10875, "loss": 4.042254638671875, "grad_norm": 0.7338870167732239, "learning_rate": 0.0001, "elbo": 4.04225, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6377551020408163}
{"step": 10880, "loss": 4.047869873046875, "grad_norm": 1.1861501932144165, "learning_rate": 0.0001, "elbo": 4.04787, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6380483227773868}
{"step": 10885, "loss": 4.047967147827149, "grad_norm": 0.785855770111084, "learning_rate": 0.0001, "elbo": 4.04797, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6383415435139573}
{"step": 10890, "loss": 4.041717529296875, "grad_norm": 0.8050014972686768, "learning_rate": 0.0001, "elbo": 4.04172, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6386347642505278}
{"step": 10895, "loss": 4.037147521972656, "grad_norm": 0.7397003769874573, "learning_rate": 0.0001, "elbo": 4.03715, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6389279849870982}
{"step": 10895, "loss": 4.037147521972656, "grad_norm": 0.7397003769874573, "learning_rate": 0.0001, "elbo": 4.03715, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6389279849870982}
{"step": 10900, "loss": 4.02830924987793, "grad_norm": 0.7149421572685242, "learning_rate": 0.0001, "elbo": 4.02831, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6392212057236688}
{"step": 10905, "loss": 4.0372467041015625, "grad_norm": 0.7595585584640503, "learning_rate": 0.0001, "elbo": 4.03725, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6395144264602393}
{"step": 10910, "loss": 4.05279769897461, "grad_norm": 1.377605676651001, "learning_rate": 0.0001, "elbo": 4.0528, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6398076471968097}
{"step": 10915, "loss": 4.0374900817871096, "grad_norm": 0.794684648513794, "learning_rate": 0.0001, "elbo": 4.03749, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6401008679333803}
{"step": 10920, "loss": 4.033985137939453, "grad_norm": 0.9815566539764404, "learning_rate": 0.0001, "elbo": 4.03399, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6403940886699507}
{"step": 10925, "loss": 4.041982269287109, "grad_norm": 0.7252653241157532, "learning_rate": 0.0001, "elbo": 4.04198, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6406873094065212}
{"step": 10930, "loss": 4.018853759765625, "grad_norm": 0.7064533829689026, "learning_rate": 0.0001, "elbo": 4.01885, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6409805301430918}
{"step": 10935, "loss": 4.041499328613281, "grad_norm": 0.7318860292434692, "learning_rate": 0.0001, "elbo": 4.0415, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6412737508796622}
{"step": 10940, "loss": 4.034981155395508, "grad_norm": 0.7275093793869019, "learning_rate": 0.0001, "elbo": 4.03498, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6415669716162327}
{"step": 10945, "loss": 4.031140899658203, "grad_norm": 0.7981676459312439, "learning_rate": 0.0001, "elbo": 4.03114, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6418601923528032}
{"step": 10950, "loss": 4.059568786621094, "grad_norm": 1.3253546953201294, "learning_rate": 0.0001, "elbo": 4.05957, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6421534130893737}
{"step": 10955, "loss": 4.029187393188477, "grad_norm": 1.090364933013916, "learning_rate": 0.0001, "elbo": 4.02919, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6424466338259441}
{"step": 10960, "loss": 4.039703369140625, "grad_norm": 0.7297698855400085, "learning_rate": 0.0001, "elbo": 4.0397, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6427398545625147}
{"step": 10965, "loss": 4.037687683105469, "grad_norm": 0.7409418821334839, "learning_rate": 0.0001, "elbo": 4.03769, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6430330752990852}
{"step": 10970, "loss": 4.0299430847167965, "grad_norm": 0.7656422257423401, "learning_rate": 0.0001, "elbo": 4.02994, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6433262960356556}
{"step": 10975, "loss": 4.048551940917969, "grad_norm": 0.7977684736251831, "learning_rate": 0.0001, "elbo": 4.04855, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6436195167722262}
{"step": 10980, "loss": 4.023291015625, "grad_norm": 0.7893739342689514, "learning_rate": 0.0001, "elbo": 4.02329, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6439127375087966}
{"step": 10985, "loss": 4.025139617919922, "grad_norm": 0.9043422937393188, "learning_rate": 0.0001, "elbo": 4.02514, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6442059582453671}
{"step": 10990, "loss": 4.045413970947266, "grad_norm": 0.7313483357429504, "learning_rate": 0.0001, "elbo": 4.04541, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6444991789819376}
{"step": 10995, "loss": 4.034091949462891, "grad_norm": 0.7819533348083496, "learning_rate": 0.0001, "elbo": 4.03409, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6447923997185081}
{"step": 11000, "loss": 4.031857681274414, "grad_norm": 0.7162859439849854, "learning_rate": 0.0001, "elbo": 4.03186, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6450856204550786}
{"step": 11005, "loss": 4.022055053710938, "grad_norm": 0.7860352993011475, "learning_rate": 0.0001, "elbo": 4.02206, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6453788411916491}
{"step": 11010, "loss": 4.027892684936523, "grad_norm": 0.7436181306838989, "learning_rate": 0.0001, "elbo": 4.02789, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6456720619282196}
{"step": 11015, "loss": 4.049080657958984, "grad_norm": 0.8057268261909485, "learning_rate": 0.0001, "elbo": 4.04908, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.64596528266479}
{"step": 11020, "loss": 4.038119125366211, "grad_norm": 0.7488049268722534, "learning_rate": 0.0001, "elbo": 4.03812, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6462585034013606}
{"step": 11025, "loss": 4.0135032653808596, "grad_norm": 0.823341429233551, "learning_rate": 0.0001, "elbo": 4.0135, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.646551724137931}
{"step": 11030, "loss": 4.035621643066406, "grad_norm": 0.8010722994804382, "learning_rate": 0.0001, "elbo": 4.03562, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6468449448745015}
{"step": 11035, "loss": 4.035046005249024, "grad_norm": 0.8168622255325317, "learning_rate": 0.0001, "elbo": 4.03505, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6471381656110721}
{"step": 11040, "loss": 4.0283447265625, "grad_norm": 0.8119094371795654, "learning_rate": 0.0001, "elbo": 4.02834, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6474313863476425}
{"step": 11045, "loss": 4.0306846618652346, "grad_norm": 0.8628028035163879, "learning_rate": 0.0001, "elbo": 4.03068, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.647724607084213}
{"step": 11050, "loss": 4.039915084838867, "grad_norm": 0.8677850961685181, "learning_rate": 0.0001, "elbo": 4.03992, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6480178278207834}
{"step": 11055, "loss": 4.033511352539063, "grad_norm": 0.6979336738586426, "learning_rate": 0.0001, "elbo": 4.03351, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.648311048557354}
{"step": 11060, "loss": 4.029236602783203, "grad_norm": 0.7677962183952332, "learning_rate": 0.0001, "elbo": 4.02924, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6486042692939245}
{"step": 11065, "loss": 4.011492156982422, "grad_norm": 0.7288206815719604, "learning_rate": 0.0001, "elbo": 4.01149, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6488974900304949}
{"step": 11070, "loss": 4.027717590332031, "grad_norm": 0.7065606117248535, "learning_rate": 0.0001, "elbo": 4.02772, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6491907107670655}
{"step": 11075, "loss": 4.032051467895508, "grad_norm": 0.8531278967857361, "learning_rate": 0.0001, "elbo": 4.03205, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6494839315036359}
{"step": 11080, "loss": 4.035982131958008, "grad_norm": 0.7644762396812439, "learning_rate": 0.0001, "elbo": 4.03598, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6497771522402064}
{"step": 11085, "loss": 4.030474090576172, "grad_norm": 0.7883253693580627, "learning_rate": 0.0001, "elbo": 4.03047, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.650070372976777}
{"step": 11090, "loss": 4.0247955322265625, "grad_norm": 0.7670316696166992, "learning_rate": 0.0001, "elbo": 4.0248, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6503635937133474}
{"step": 11095, "loss": 4.018141555786133, "grad_norm": 0.7092275023460388, "learning_rate": 0.0001, "elbo": 4.01814, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6506568144499179}
{"step": 11100, "loss": 4.016705322265625, "grad_norm": 0.7769389152526855, "learning_rate": 0.0001, "elbo": 4.01671, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6509500351864884}
{"step": 11105, "loss": 4.021514129638672, "grad_norm": 0.767438530921936, "learning_rate": 0.0001, "elbo": 4.02151, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6512432559230589}
{"step": 11110, "loss": 4.012667083740235, "grad_norm": 0.7412839531898499, "learning_rate": 0.0001, "elbo": 4.01267, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6515364766596293}
{"step": 11115, "loss": 4.035308837890625, "grad_norm": 0.9917327165603638, "learning_rate": 0.0001, "elbo": 4.03531, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6518296973961999}
{"step": 11115, "loss": 4.035308837890625, "grad_norm": 0.9917327165603638, "learning_rate": 0.0001, "elbo": 4.03531, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6518296973961999}
{"step": 11120, "loss": 4.0259864807128904, "grad_norm": 0.7648653388023376, "learning_rate": 0.0001, "elbo": 4.02599, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6521229181327703}
{"step": 11125, "loss": 4.027069854736328, "grad_norm": 0.7398228049278259, "learning_rate": 0.0001, "elbo": 4.02707, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6524161388693408}
{"step": 11130, "loss": 4.0147346496582035, "grad_norm": 0.7585936188697815, "learning_rate": 0.0001, "elbo": 4.01473, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6527093596059114}
{"step": 11130, "loss": 4.0147346496582035, "grad_norm": 0.7585936188697815, "learning_rate": 0.0001, "elbo": 4.01473, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6527093596059114}
{"step": 11135, "loss": 4.031697845458984, "grad_norm": 0.7576830387115479, "learning_rate": 0.0001, "elbo": 4.0317, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6530025803424818}
{"step": 11140, "loss": 4.030796813964844, "grad_norm": 0.763455867767334, "learning_rate": 0.0001, "elbo": 4.0308, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6532958010790523}
{"step": 11145, "loss": 4.0392616271972654, "grad_norm": 0.8431694507598877, "learning_rate": 0.0001, "elbo": 4.03926, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6535890218156228}
{"step": 11150, "loss": 4.027151107788086, "grad_norm": 0.7748448252677917, "learning_rate": 0.0001, "elbo": 4.02715, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6538822425521933}
{"step": 11155, "loss": 4.030059051513672, "grad_norm": 0.7334899306297302, "learning_rate": 0.0001, "elbo": 4.03006, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6541754632887637}
{"step": 11160, "loss": 4.030315399169922, "grad_norm": 0.7183801531791687, "learning_rate": 0.0001, "elbo": 4.03032, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6544686840253343}
{"step": 11165, "loss": 4.029745101928711, "grad_norm": 0.734869122505188, "learning_rate": 0.0001, "elbo": 4.02975, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6547619047619048}
{"step": 11170, "loss": 4.010466766357422, "grad_norm": 0.7069803476333618, "learning_rate": 0.0001, "elbo": 4.01047, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6550551254984752}
{"step": 11175, "loss": 4.030167770385742, "grad_norm": 0.7537464499473572, "learning_rate": 0.0001, "elbo": 4.03017, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6553483462350458}
{"step": 11180, "loss": 4.0245414733886715, "grad_norm": 0.8116324543952942, "learning_rate": 0.0001, "elbo": 4.02454, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6556415669716162}
{"step": 11185, "loss": 4.042720794677734, "grad_norm": 0.7614364624023438, "learning_rate": 0.0001, "elbo": 4.04272, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6559347877081867}
{"step": 11190, "loss": 4.033164596557617, "grad_norm": 1.0421291589736938, "learning_rate": 0.0001, "elbo": 4.03316, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6562280084447573}
{"step": 11195, "loss": 4.032585906982422, "grad_norm": 0.7432817220687866, "learning_rate": 0.0001, "elbo": 4.03259, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6565212291813277}
{"step": 11200, "loss": 4.029493713378907, "grad_norm": 0.7371865510940552, "learning_rate": 0.0001, "elbo": 4.02949, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6568144499178982}
{"step": 11205, "loss": 4.035148620605469, "grad_norm": 0.7657822966575623, "learning_rate": 0.0001, "elbo": 4.03515, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6571076706544687}
{"step": 11210, "loss": 4.019342041015625, "grad_norm": 0.8255417943000793, "learning_rate": 0.0001, "elbo": 4.01934, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6574008913910392}
{"step": 11215, "loss": 4.006324005126953, "grad_norm": 0.801296055316925, "learning_rate": 0.0001, "elbo": 4.00632, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6576941121276096}
{"step": 11220, "loss": 4.011666107177734, "grad_norm": 0.6949084401130676, "learning_rate": 0.0001, "elbo": 4.01167, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6579873328641802}
{"step": 11225, "loss": 4.025778198242188, "grad_norm": 0.73626309633255, "learning_rate": 0.0001, "elbo": 4.02578, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6582805536007507}
{"step": 11230, "loss": 4.039898681640625, "grad_norm": 0.7277762293815613, "learning_rate": 0.0001, "elbo": 4.0399, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6585737743373211}
{"step": 11235, "loss": 4.036561584472656, "grad_norm": 0.8426260352134705, "learning_rate": 0.0001, "elbo": 4.03656, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6588669950738916}
{"step": 11240, "loss": 4.018735122680664, "grad_norm": 0.7592558264732361, "learning_rate": 0.0001, "elbo": 4.01874, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6591602158104621}
{"step": 11245, "loss": 4.022704315185547, "grad_norm": 0.7350326180458069, "learning_rate": 0.0001, "elbo": 4.0227, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6594534365470326}
{"step": 11250, "loss": 4.015306854248047, "grad_norm": 0.8059170246124268, "learning_rate": 0.0001, "elbo": 4.01531, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.659746657283603}
{"step": 11255, "loss": 4.002803802490234, "grad_norm": 0.8314979076385498, "learning_rate": 0.0001, "elbo": 4.0028, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6600398780201736}
{"step": 11260, "loss": 4.0181121826171875, "grad_norm": 0.7972023487091064, "learning_rate": 0.0001, "elbo": 4.01811, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6603330987567441}
{"step": 11265, "loss": 4.0231281280517575, "grad_norm": 0.7470526695251465, "learning_rate": 0.0001, "elbo": 4.02313, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6606263194933145}
{"step": 11270, "loss": 4.027113342285157, "grad_norm": 0.8684269785881042, "learning_rate": 0.0001, "elbo": 4.02711, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6609195402298851}
{"step": 11275, "loss": 4.016916656494141, "grad_norm": 0.8057727813720703, "learning_rate": 0.0001, "elbo": 4.01692, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6612127609664555}
{"step": 11280, "loss": 4.024062728881836, "grad_norm": 0.7581280469894409, "learning_rate": 0.0001, "elbo": 4.02406, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.661505981703026}
{"step": 11285, "loss": 4.021664428710937, "grad_norm": 0.7242745161056519, "learning_rate": 0.0001, "elbo": 4.02166, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6617992024395966}
{"step": 11290, "loss": 4.013324356079101, "grad_norm": 0.7799978256225586, "learning_rate": 0.0001, "elbo": 4.01332, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.662092423176167}
{"step": 11295, "loss": 4.018746185302734, "grad_norm": 0.7177415490150452, "learning_rate": 0.0001, "elbo": 4.01875, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6623856439127375}
{"step": 11300, "loss": 4.029177474975586, "grad_norm": 1.2318310737609863, "learning_rate": 0.0001, "elbo": 4.02918, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.662678864649308}
{"step": 11305, "loss": 4.009996795654297, "grad_norm": 0.6931954026222229, "learning_rate": 0.0001, "elbo": 4.01, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6629720853858785}
{"step": 11310, "loss": 4.01947021484375, "grad_norm": 0.748286783695221, "learning_rate": 0.0001, "elbo": 4.01947, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6632653061224489}
{"step": 11315, "loss": 4.027650451660156, "grad_norm": 0.7121704816818237, "learning_rate": 0.0001, "elbo": 4.02765, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6635585268590195}
{"step": 11320, "loss": 4.020469665527344, "grad_norm": 0.7167524099349976, "learning_rate": 0.0001, "elbo": 4.02047, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.66385174759559}
{"step": 11325, "loss": 4.00806999206543, "grad_norm": 0.8171713948249817, "learning_rate": 0.0001, "elbo": 4.00807, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6641449683321604}
{"step": 11330, "loss": 4.0172889709472654, "grad_norm": 0.7435358762741089, "learning_rate": 0.0001, "elbo": 4.01729, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.664438189068731}
{"step": 11335, "loss": 4.0322105407714846, "grad_norm": 0.8230147361755371, "learning_rate": 0.0001, "elbo": 4.03221, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6647314098053014}
{"step": 11340, "loss": 4.017615509033203, "grad_norm": 1.4027459621429443, "learning_rate": 0.0001, "elbo": 4.01762, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6650246305418719}
{"step": 11345, "loss": 4.018539810180664, "grad_norm": 0.7249190211296082, "learning_rate": 0.0001, "elbo": 4.01854, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6653178512784425}
{"step": 11350, "loss": 4.025764465332031, "grad_norm": 0.7174099683761597, "learning_rate": 0.0001, "elbo": 4.02576, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6656110720150129}
{"step": 11355, "loss": 4.0120891571044925, "grad_norm": 0.7913573980331421, "learning_rate": 0.0001, "elbo": 4.01209, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6659042927515834}
{"step": 11360, "loss": 4.015022277832031, "grad_norm": 0.9902146458625793, "learning_rate": 0.0001, "elbo": 4.01502, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6661975134881539}
{"step": 11365, "loss": 4.0005542755126955, "grad_norm": 0.7434998750686646, "learning_rate": 0.0001, "elbo": 4.00055, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6664907342247244}
{"step": 11370, "loss": 4.012704467773437, "grad_norm": 0.7342117428779602, "learning_rate": 0.0001, "elbo": 4.0127, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6667839549612948}
{"step": 11375, "loss": 4.01515007019043, "grad_norm": 0.730926513671875, "learning_rate": 0.0001, "elbo": 4.01515, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6670771756978654}
{"step": 11380, "loss": 4.010109329223633, "grad_norm": 0.7639469504356384, "learning_rate": 0.0001, "elbo": 4.01011, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6673703964344359}
{"step": 11385, "loss": 4.005024719238281, "grad_norm": 0.7779965400695801, "learning_rate": 0.0001, "elbo": 4.00502, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6676636171710063}
{"step": 11390, "loss": 4.0181396484375, "grad_norm": 0.7787409424781799, "learning_rate": 0.0001, "elbo": 4.01814, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6679568379075769}
{"step": 11395, "loss": 4.011409759521484, "grad_norm": 0.7257749438285828, "learning_rate": 0.0001, "elbo": 4.01141, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6682500586441473}
{"step": 11400, "loss": 4.013528060913086, "grad_norm": 0.6918138265609741, "learning_rate": 0.0001, "elbo": 4.01353, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6685432793807178}
{"step": 11405, "loss": 4.032317352294922, "grad_norm": 0.728631854057312, "learning_rate": 0.0001, "elbo": 4.03232, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6688365001172883}
{"step": 11410, "loss": 4.016032409667969, "grad_norm": 0.7723419666290283, "learning_rate": 0.0001, "elbo": 4.01603, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6691297208538588}
{"step": 11415, "loss": 4.010477828979492, "grad_norm": 0.6974599361419678, "learning_rate": 0.0001, "elbo": 4.01048, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6694229415904293}
{"step": 11420, "loss": 4.032657623291016, "grad_norm": 0.7359170317649841, "learning_rate": 0.0001, "elbo": 4.03266, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6697161623269998}
{"step": 11425, "loss": 4.021942138671875, "grad_norm": 0.7992058992385864, "learning_rate": 0.0001, "elbo": 4.02194, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6700093830635703}
{"step": 11430, "loss": 4.015771865844727, "grad_norm": 0.7158420085906982, "learning_rate": 0.0001, "elbo": 4.01577, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6703026038001407}
{"step": 11435, "loss": 4.020046997070312, "grad_norm": 0.7706803679466248, "learning_rate": 0.0001, "elbo": 4.02005, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6705958245367112}
{"step": 11440, "loss": 4.011557769775391, "grad_norm": 0.7450464963912964, "learning_rate": 0.0001, "elbo": 4.01156, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6708890452732817}
{"step": 11445, "loss": 4.025648498535157, "grad_norm": 0.7191128730773926, "learning_rate": 0.0001, "elbo": 4.02565, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6711822660098522}
{"step": 11450, "loss": 4.0014190673828125, "grad_norm": 0.7705096006393433, "learning_rate": 0.0001, "elbo": 4.00142, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6714754867464227}
{"step": 11455, "loss": 4.017964172363281, "grad_norm": 0.7419270277023315, "learning_rate": 0.0001, "elbo": 4.01796, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6717687074829932}
{"step": 11460, "loss": 3.9951751708984373, "grad_norm": 0.7943211793899536, "learning_rate": 0.0001, "elbo": 3.99517, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6720619282195637}
{"step": 11465, "loss": 4.00767936706543, "grad_norm": 0.7959721088409424, "learning_rate": 0.0001, "elbo": 4.00768, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6723551489561341}
{"step": 11470, "loss": 4.010863876342773, "grad_norm": 0.7144691348075867, "learning_rate": 0.0001, "elbo": 4.01086, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6726483696927047}
{"step": 11470, "loss": 4.010863876342773, "grad_norm": 0.7144691348075867, "learning_rate": 0.0001, "elbo": 4.01086, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6726483696927047}
{"step": 11475, "loss": 3.9988471984863283, "grad_norm": 0.8049424886703491, "learning_rate": 0.0001, "elbo": 3.99885, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6729415904292751}
{"step": 11480, "loss": 4.021044921875, "grad_norm": 0.7689248919487, "learning_rate": 0.0001, "elbo": 4.02104, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6732348111658456}
{"step": 11485, "loss": 4.021091079711914, "grad_norm": 0.7251812219619751, "learning_rate": 0.0001, "elbo": 4.02109, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6735280319024162}
{"step": 11490, "loss": 4.012229156494141, "grad_norm": 0.7398205995559692, "learning_rate": 0.0001, "elbo": 4.01223, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6738212526389866}
{"step": 11495, "loss": 3.997056579589844, "grad_norm": 0.7299215793609619, "learning_rate": 0.0001, "elbo": 3.99706, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6741144733755571}
{"step": 11500, "loss": 4.021523284912109, "grad_norm": 0.8892385959625244, "learning_rate": 0.0001, "elbo": 4.02152, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6744076941121276}
{"step": 11505, "loss": 4.023222351074219, "grad_norm": 0.8505001068115234, "learning_rate": 0.0001, "elbo": 4.02322, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6747009148486981}
{"step": 11510, "loss": 4.007262802124023, "grad_norm": 0.784885823726654, "learning_rate": 0.0001, "elbo": 4.00726, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6749941355852686}
{"step": 11515, "loss": 4.002808380126953, "grad_norm": 0.8585939407348633, "learning_rate": 0.0001, "elbo": 4.00281, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6752873563218391}
{"step": 11520, "loss": 4.019986724853515, "grad_norm": 0.8044946193695068, "learning_rate": 0.0001, "elbo": 4.01999, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6755805770584096}
{"step": 11525, "loss": 3.9974315643310545, "grad_norm": 0.7421023845672607, "learning_rate": 0.0001, "elbo": 3.99743, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.67587379779498}
{"step": 11530, "loss": 4.009490966796875, "grad_norm": 0.726246178150177, "learning_rate": 0.0001, "elbo": 4.00949, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6761670185315506}
{"step": 11535, "loss": 4.000762176513672, "grad_norm": 0.7266200184822083, "learning_rate": 0.0001, "elbo": 4.00076, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.676460239268121}
{"step": 11540, "loss": 4.008961486816406, "grad_norm": 0.702704668045044, "learning_rate": 0.0001, "elbo": 4.00896, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6767534600046915}
{"step": 11545, "loss": 4.024118423461914, "grad_norm": 0.7536278963088989, "learning_rate": 0.0001, "elbo": 4.02412, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6770466807412621}
{"step": 11550, "loss": 4.01201171875, "grad_norm": 0.7127920985221863, "learning_rate": 0.0001, "elbo": 4.01201, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6773399014778325}
{"step": 11555, "loss": 4.0176643371582035, "grad_norm": 0.7948833107948303, "learning_rate": 0.0001, "elbo": 4.01766, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.677633122214403}
{"step": 11560, "loss": 4.003678894042968, "grad_norm": 0.9407204389572144, "learning_rate": 0.0001, "elbo": 4.00368, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6779263429509735}
{"step": 11565, "loss": 4.0295158386230465, "grad_norm": 0.8320264220237732, "learning_rate": 0.0001, "elbo": 4.02952, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.678219563687544}
{"step": 11570, "loss": 4.036924362182617, "grad_norm": 0.7901050448417664, "learning_rate": 0.0001, "elbo": 4.03692, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6785127844241144}
{"step": 11575, "loss": 4.022565460205078, "grad_norm": 1.0477725267410278, "learning_rate": 0.0001, "elbo": 4.02257, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.678806005160685}
{"step": 11580, "loss": 4.011981582641601, "grad_norm": 0.7384265661239624, "learning_rate": 0.0001, "elbo": 4.01198, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6790992258972555}
{"step": 11585, "loss": 4.013262557983398, "grad_norm": 0.7422664165496826, "learning_rate": 0.0001, "elbo": 4.01326, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6793924466338259}
{"step": 11590, "loss": 4.008049774169922, "grad_norm": 0.779403805732727, "learning_rate": 0.0001, "elbo": 4.00805, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6796856673703965}
{"step": 11595, "loss": 4.011224365234375, "grad_norm": 0.7422956824302673, "learning_rate": 0.0001, "elbo": 4.01122, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6799788881069669}
{"step": 11600, "loss": 4.011375045776367, "grad_norm": 0.7834144234657288, "learning_rate": 0.0001, "elbo": 4.01137, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6802721088435374}
{"step": 11605, "loss": 4.006570434570312, "grad_norm": 0.7552828788757324, "learning_rate": 0.0001, "elbo": 4.00657, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.680565329580108}
{"step": 11610, "loss": 4.0073505401611325, "grad_norm": 0.7255575656890869, "learning_rate": 0.0001, "elbo": 4.00735, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6808585503166784}
{"step": 11615, "loss": 3.995939254760742, "grad_norm": 0.7924529314041138, "learning_rate": 0.0001, "elbo": 3.99594, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6811517710532489}
{"step": 11620, "loss": 3.997890090942383, "grad_norm": 0.7093684077262878, "learning_rate": 0.0001, "elbo": 3.99789, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6814449917898193}
{"step": 11625, "loss": 4.0090984344482425, "grad_norm": 1.1310017108917236, "learning_rate": 0.0001, "elbo": 4.0091, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6817382125263899}
{"step": 11630, "loss": 4.007569122314453, "grad_norm": 0.7569341659545898, "learning_rate": 0.0001, "elbo": 4.00757, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6820314332629603}
{"step": 11635, "loss": 3.9950180053710938, "grad_norm": 0.8153905868530273, "learning_rate": 0.0001, "elbo": 3.99502, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6823246539995308}
{"step": 11640, "loss": 4.01573486328125, "grad_norm": 1.213346242904663, "learning_rate": 0.0001, "elbo": 4.01574, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6826178747361014}
{"step": 11645, "loss": 3.9914840698242187, "grad_norm": 0.7682644128799438, "learning_rate": 0.0001, "elbo": 3.99148, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6829110954726718}
{"step": 11650, "loss": 4.022254180908203, "grad_norm": 0.719088613986969, "learning_rate": 0.0001, "elbo": 4.02225, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6832043162092423}
{"step": 11655, "loss": 4.004545211791992, "grad_norm": 0.8043907284736633, "learning_rate": 0.0001, "elbo": 4.00455, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6834975369458128}
{"step": 11660, "loss": 4.000525665283203, "grad_norm": 0.7621895670890808, "learning_rate": 0.0001, "elbo": 4.00053, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6837907576823833}
{"step": 11665, "loss": 4.012653732299805, "grad_norm": 0.7817133665084839, "learning_rate": 0.0001, "elbo": 4.01265, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6840839784189537}
{"step": 11670, "loss": 4.006115341186524, "grad_norm": 0.7468187808990479, "learning_rate": 0.0001, "elbo": 4.00612, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6843771991555243}
{"step": 11675, "loss": 4.009358215332031, "grad_norm": 0.7911964058876038, "learning_rate": 0.0001, "elbo": 4.00936, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6846704198920948}
{"step": 11680, "loss": 4.023969268798828, "grad_norm": 0.8086869120597839, "learning_rate": 0.0001, "elbo": 4.02397, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6849636406286652}
{"step": 11685, "loss": 4.012225341796875, "grad_norm": 0.7349275350570679, "learning_rate": 0.0001, "elbo": 4.01223, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6852568613652358}
{"step": 11685, "loss": 4.012225341796875, "grad_norm": 0.7349275350570679, "learning_rate": 0.0001, "elbo": 4.01223, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6852568613652358}
{"step": 11690, "loss": 4.013861083984375, "grad_norm": 0.7159821391105652, "learning_rate": 0.0001, "elbo": 4.01386, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6855500821018062}
{"step": 11695, "loss": 4.004839324951172, "grad_norm": 0.7232310175895691, "learning_rate": 0.0001, "elbo": 4.00484, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6858433028383767}
{"step": 11700, "loss": 4.019096755981446, "grad_norm": 0.7503079771995544, "learning_rate": 0.0001, "elbo": 4.0191, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6861365235749473}
{"step": 11705, "loss": 4.024192810058594, "grad_norm": 0.7662746906280518, "learning_rate": 0.0001, "elbo": 4.02419, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6864297443115177}
{"step": 11710, "loss": 3.986815643310547, "grad_norm": 0.6975728869438171, "learning_rate": 0.0001, "elbo": 3.98682, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6867229650480882}
{"step": 11715, "loss": 4.0065162658691404, "grad_norm": 0.8296488523483276, "learning_rate": 0.0001, "elbo": 4.00652, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6870161857846587}
{"step": 11720, "loss": 4.011986160278321, "grad_norm": 1.0531790256500244, "learning_rate": 0.0001, "elbo": 4.01199, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6873094065212292}
{"step": 11725, "loss": 4.032420349121094, "grad_norm": 0.9433901309967041, "learning_rate": 0.0001, "elbo": 4.03242, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6876026272577996}
{"step": 11730, "loss": 3.99877815246582, "grad_norm": 0.7918660044670105, "learning_rate": 0.0001, "elbo": 3.99878, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6878958479943702}
{"step": 11735, "loss": 4.019363021850586, "grad_norm": 0.7267846465110779, "learning_rate": 0.0001, "elbo": 4.01936, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6881890687309407}
{"step": 11740, "loss": 4.004806900024414, "grad_norm": 0.8015903234481812, "learning_rate": 0.0001, "elbo": 4.00481, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6884822894675111}
{"step": 11745, "loss": 4.012731552124023, "grad_norm": 0.8032624125480652, "learning_rate": 0.0001, "elbo": 4.01273, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6887755102040817}
{"step": 11750, "loss": 4.000193786621094, "grad_norm": 0.7264854907989502, "learning_rate": 0.0001, "elbo": 4.00019, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6890687309406521}
{"step": 11755, "loss": 4.0080615997314455, "grad_norm": 0.7343717217445374, "learning_rate": 0.0001, "elbo": 4.00806, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6893619516772226}
{"step": 11760, "loss": 4.02823486328125, "grad_norm": 0.8015009164810181, "learning_rate": 0.0001, "elbo": 4.02823, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6896551724137931}
{"step": 11765, "loss": 4.007646942138672, "grad_norm": 0.7912763953208923, "learning_rate": 0.0001, "elbo": 4.00765, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6899483931503636}
{"step": 11770, "loss": 4.006626892089844, "grad_norm": 0.7653283476829529, "learning_rate": 0.0001, "elbo": 4.00663, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6902416138869341}
{"step": 11775, "loss": 4.0000755310058596, "grad_norm": 0.8244530558586121, "learning_rate": 0.0001, "elbo": 4.00008, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6905348346235046}
{"step": 11780, "loss": 3.9902000427246094, "grad_norm": 0.8453087210655212, "learning_rate": 0.0001, "elbo": 3.9902, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6908280553600751}
{"step": 11785, "loss": 3.9980712890625, "grad_norm": 0.796097993850708, "learning_rate": 0.0001, "elbo": 3.99807, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6911212760966455}
{"step": 11790, "loss": 3.992802047729492, "grad_norm": 0.7688470482826233, "learning_rate": 0.0001, "elbo": 3.9928, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6914144968332161}
{"step": 11795, "loss": 4.009260559082032, "grad_norm": 0.8024004697799683, "learning_rate": 0.0001, "elbo": 4.00926, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6917077175697866}
{"step": 11800, "loss": 4.002081298828125, "grad_norm": 0.6758188009262085, "learning_rate": 0.0001, "elbo": 4.00208, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.692000938306357}
{"step": 11805, "loss": 4.005170822143555, "grad_norm": 0.7586377263069153, "learning_rate": 0.0001, "elbo": 4.00517, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6922941590429276}
{"step": 11810, "loss": 4.010879135131836, "grad_norm": 0.8597705364227295, "learning_rate": 0.0001, "elbo": 4.01088, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.692587379779498}
{"step": 11815, "loss": 4.003482818603516, "grad_norm": 0.7183590531349182, "learning_rate": 0.0001, "elbo": 4.00348, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6928806005160685}
{"step": 11820, "loss": 4.0001075744628904, "grad_norm": 0.7634389400482178, "learning_rate": 0.0001, "elbo": 4.00011, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6931738212526389}
{"step": 11825, "loss": 4.0015922546386715, "grad_norm": 0.78467857837677, "learning_rate": 0.0001, "elbo": 4.00159, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6934670419892095}
{"step": 11830, "loss": 4.0018150329589846, "grad_norm": 0.7134576439857483, "learning_rate": 0.0001, "elbo": 4.00181, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.69376026272578}
{"step": 11835, "loss": 4.0023853302001955, "grad_norm": 0.7079212069511414, "learning_rate": 0.0001, "elbo": 4.00239, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6940534834623504}
{"step": 11840, "loss": 4.006180953979492, "grad_norm": 0.714805543422699, "learning_rate": 0.0001, "elbo": 4.00618, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.694346704198921}
{"step": 11845, "loss": 3.99365234375, "grad_norm": 1.00508713722229, "learning_rate": 0.0001, "elbo": 3.99365, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6946399249354914}
{"step": 11850, "loss": 4.001591491699219, "grad_norm": 0.7699795961380005, "learning_rate": 0.0001, "elbo": 4.00159, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6949331456720619}
{"step": 11855, "loss": 4.009324645996093, "grad_norm": 0.7647586464881897, "learning_rate": 0.0001, "elbo": 4.00932, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6952263664086324}
{"step": 11860, "loss": 3.9902313232421873, "grad_norm": 0.6717805862426758, "learning_rate": 0.0001, "elbo": 3.99023, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6955195871452029}
{"step": 11865, "loss": 3.9871589660644533, "grad_norm": 1.0721561908721924, "learning_rate": 0.0001, "elbo": 3.98716, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6958128078817734}
{"step": 11870, "loss": 4.014125061035156, "grad_norm": 0.780966579914093, "learning_rate": 0.0001, "elbo": 4.01412, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6961060286183439}
{"step": 11875, "loss": 3.9847179412841798, "grad_norm": 0.7320898771286011, "learning_rate": 0.0001, "elbo": 3.98472, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6963992493549144}
{"step": 11880, "loss": 4.016844177246094, "grad_norm": 0.7206257581710815, "learning_rate": 0.0001, "elbo": 4.01684, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6966924700914848}
{"step": 11885, "loss": 4.0007568359375, "grad_norm": 0.7630663514137268, "learning_rate": 0.0001, "elbo": 4.00076, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6969856908280554}
{"step": 11890, "loss": 4.015929412841797, "grad_norm": 0.83298259973526, "learning_rate": 0.0001, "elbo": 4.01593, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6972789115646258}
{"step": 11895, "loss": 4.004234313964844, "grad_norm": 0.7510067820549011, "learning_rate": 0.0001, "elbo": 4.00423, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6975721323011963}
{"step": 11900, "loss": 4.0116119384765625, "grad_norm": 1.096529245376587, "learning_rate": 0.0001, "elbo": 4.01161, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6978653530377669}
{"step": 11905, "loss": 4.000381851196289, "grad_norm": 0.7378886938095093, "learning_rate": 0.0001, "elbo": 4.00038, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6981585737743373}
{"step": 11905, "loss": 4.000381851196289, "grad_norm": 0.7378886938095093, "learning_rate": 0.0001, "elbo": 4.00038, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6981585737743373}
{"step": 11910, "loss": 3.995453643798828, "grad_norm": 0.7855200171470642, "learning_rate": 0.0001, "elbo": 3.99545, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6984517945109078}
{"step": 11915, "loss": 3.998879241943359, "grad_norm": 0.7214322686195374, "learning_rate": 0.0001, "elbo": 3.99888, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6987450152474783}
{"step": 11920, "loss": 4.01751480102539, "grad_norm": 0.868827223777771, "learning_rate": 0.0001, "elbo": 4.01751, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6990382359840488}
{"step": 11925, "loss": 3.997508239746094, "grad_norm": 0.7570980787277222, "learning_rate": 0.0001, "elbo": 3.99751, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6993314567206192}
{"step": 11930, "loss": 4.013258743286133, "grad_norm": 0.7447478175163269, "learning_rate": 0.0001, "elbo": 4.01326, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6996246774571898}
{"step": 11935, "loss": 4.0195869445800785, "grad_norm": 0.8138691782951355, "learning_rate": 0.0001, "elbo": 4.01959, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.6999178981937603}
{"step": 11940, "loss": 4.0070068359375, "grad_norm": 0.8141955733299255, "learning_rate": 0.0001, "elbo": 4.00701, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7002111189303307}
{"step": 11945, "loss": 3.9937423706054687, "grad_norm": 0.7453293204307556, "learning_rate": 0.0001, "elbo": 3.99374, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7005043396669013}
{"step": 11950, "loss": 4.005550384521484, "grad_norm": 0.6658647656440735, "learning_rate": 0.0001, "elbo": 4.00555, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7007975604034717}
{"step": 11955, "loss": 3.96637077331543, "grad_norm": 0.8135953545570374, "learning_rate": 0.0001, "elbo": 3.96637, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7010907811400422}
{"step": 11960, "loss": 4.000434875488281, "grad_norm": 0.7326380014419556, "learning_rate": 0.0001, "elbo": 4.00043, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7013840018766128}
{"step": 11965, "loss": 4.007796859741211, "grad_norm": 0.7536133527755737, "learning_rate": 0.0001, "elbo": 4.0078, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7016772226131832}
{"step": 11970, "loss": 3.998055267333984, "grad_norm": 0.7514755725860596, "learning_rate": 0.0001, "elbo": 3.99806, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7019704433497537}
{"step": 11975, "loss": 4.000758361816406, "grad_norm": 0.7722024917602539, "learning_rate": 0.0001, "elbo": 4.00076, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7022636640863242}
{"step": 11980, "loss": 4.001861190795898, "grad_norm": 0.7621858716011047, "learning_rate": 0.0001, "elbo": 4.00186, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7025568848228947}
{"step": 11985, "loss": 4.004915618896485, "grad_norm": 0.7568869590759277, "learning_rate": 0.0001, "elbo": 4.00492, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7028501055594651}
{"step": 11990, "loss": 4.002358245849609, "grad_norm": 0.9135347008705139, "learning_rate": 0.0001, "elbo": 4.00236, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7031433262960357}
{"step": 11995, "loss": 4.00371208190918, "grad_norm": 0.992131769657135, "learning_rate": 0.0001, "elbo": 4.00371, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7034365470326062}
{"step": 12000, "loss": 4.004167938232422, "grad_norm": 0.7888074517250061, "learning_rate": 0.0001, "elbo": 4.00417, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7037297677691766}
{"step": 12005, "loss": 4.014087677001953, "grad_norm": 0.6898695826530457, "learning_rate": 0.0001, "elbo": 4.01409, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7040229885057471}
{"step": 12010, "loss": 4.01307487487793, "grad_norm": 0.929095447063446, "learning_rate": 0.0001, "elbo": 4.01308, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7043162092423176}
{"step": 12015, "loss": 4.00231704711914, "grad_norm": 0.7575517296791077, "learning_rate": 0.0001, "elbo": 4.00232, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7046094299788881}
{"step": 12020, "loss": 4.002159500122071, "grad_norm": 0.7148591876029968, "learning_rate": 0.0001, "elbo": 4.00216, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7049026507154585}
{"step": 12020, "loss": 4.002159500122071, "grad_norm": 0.7148591876029968, "learning_rate": 0.0001, "elbo": 4.00216, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7049026507154585}
{"step": 12025, "loss": 3.9928932189941406, "grad_norm": 0.7661805748939514, "learning_rate": 0.0001, "elbo": 3.99289, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7051958714520291}
{"step": 12030, "loss": 3.9966400146484373, "grad_norm": 0.7397993803024292, "learning_rate": 0.0001, "elbo": 3.99664, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7054890921885996}
{"step": 12035, "loss": 3.9914798736572266, "grad_norm": 0.7104146480560303, "learning_rate": 0.0001, "elbo": 3.99148, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.70578231292517}
{"step": 12040, "loss": 3.9914501190185545, "grad_norm": 0.7174729108810425, "learning_rate": 0.0001, "elbo": 3.99145, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7060755336617406}
{"step": 12045, "loss": 3.9763736724853516, "grad_norm": 0.793826162815094, "learning_rate": 0.0001, "elbo": 3.97637, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.706368754398311}
{"step": 12050, "loss": 4.000254440307617, "grad_norm": 1.1197230815887451, "learning_rate": 0.0001, "elbo": 4.00025, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7066619751348815}
{"step": 12055, "loss": 3.9856197357177736, "grad_norm": 0.7736952900886536, "learning_rate": 0.0001, "elbo": 3.98562, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7069551958714521}
{"step": 12055, "loss": 3.9856197357177736, "grad_norm": 0.7736952900886536, "learning_rate": 0.0001, "elbo": 3.98562, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7069551958714521}
{"step": 12060, "loss": 4.008695220947265, "grad_norm": 0.7251931428909302, "learning_rate": 0.0001, "elbo": 4.0087, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7072484166080225}
{"step": 12065, "loss": 3.9884510040283203, "grad_norm": 0.8655736446380615, "learning_rate": 0.0001, "elbo": 3.98845, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.707541637344593}
{"step": 12070, "loss": 3.992867279052734, "grad_norm": 0.6793051362037659, "learning_rate": 0.0001, "elbo": 3.99287, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7078348580811635}
{"step": 12075, "loss": 3.9918121337890624, "grad_norm": 0.6759766340255737, "learning_rate": 0.0001, "elbo": 3.99181, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.708128078817734}
{"step": 12080, "loss": 3.9903194427490236, "grad_norm": 0.746059000492096, "learning_rate": 0.0001, "elbo": 3.99032, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7084212995543044}
{"step": 12085, "loss": 4.006793975830078, "grad_norm": 0.7256945967674255, "learning_rate": 0.0001, "elbo": 4.00679, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.708714520290875}
{"step": 12090, "loss": 3.9820751190185546, "grad_norm": 0.737888514995575, "learning_rate": 0.0001, "elbo": 3.98208, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7090077410274455}
{"step": 12095, "loss": 3.980946350097656, "grad_norm": 1.2560755014419556, "learning_rate": 0.0001, "elbo": 3.98095, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7093009617640159}
{"step": 12100, "loss": 3.988111877441406, "grad_norm": 0.7402591109275818, "learning_rate": 0.0001, "elbo": 3.98811, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7095941825005865}
{"step": 12105, "loss": 4.000995635986328, "grad_norm": 0.7626619935035706, "learning_rate": 0.0001, "elbo": 4.001, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7098874032371569}
{"step": 12110, "loss": 4.010363388061523, "grad_norm": 0.7146110534667969, "learning_rate": 0.0001, "elbo": 4.01036, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7101806239737274}
{"step": 12115, "loss": 3.998773956298828, "grad_norm": 0.8105245232582092, "learning_rate": 0.0001, "elbo": 3.99877, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.710473844710298}
{"step": 12120, "loss": 4.015367889404297, "grad_norm": 0.7248037457466125, "learning_rate": 0.0001, "elbo": 4.01537, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7107670654468684}
{"step": 12125, "loss": 3.967760467529297, "grad_norm": 0.6916241645812988, "learning_rate": 0.0001, "elbo": 3.96776, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7110602861834389}
{"step": 12130, "loss": 3.9959487915039062, "grad_norm": 0.6959226131439209, "learning_rate": 0.0001, "elbo": 3.99595, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7113535069200094}
{"step": 12135, "loss": 3.9838539123535157, "grad_norm": 0.781885027885437, "learning_rate": 0.0001, "elbo": 3.98385, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7116467276565799}
{"step": 12140, "loss": 3.9890159606933593, "grad_norm": 1.0599391460418701, "learning_rate": 0.0001, "elbo": 3.98902, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7119399483931503}
{"step": 12145, "loss": 3.9808013916015623, "grad_norm": 0.7700481414794922, "learning_rate": 0.0001, "elbo": 3.9808, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7122331691297209}
{"step": 12150, "loss": 3.9830657958984377, "grad_norm": 0.7211422324180603, "learning_rate": 0.0001, "elbo": 3.98307, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7125263898662914}
{"step": 12155, "loss": 3.994052505493164, "grad_norm": 0.7307606339454651, "learning_rate": 0.0001, "elbo": 3.99405, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7128196106028618}
{"step": 12160, "loss": 3.989154052734375, "grad_norm": 0.821872889995575, "learning_rate": 0.0001, "elbo": 3.98915, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7131128313394324}
{"step": 12165, "loss": 3.9928821563720702, "grad_norm": 0.7321733236312866, "learning_rate": 0.0001, "elbo": 3.99288, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7134060520760028}
{"step": 12170, "loss": 3.972666931152344, "grad_norm": 0.7732943296432495, "learning_rate": 0.0001, "elbo": 3.97267, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7136992728125733}
{"step": 12175, "loss": 3.9919456481933593, "grad_norm": 0.9804067611694336, "learning_rate": 0.0001, "elbo": 3.99195, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7139924935491438}
{"step": 12180, "loss": 3.9829055786132814, "grad_norm": 0.7097530364990234, "learning_rate": 0.0001, "elbo": 3.98291, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7142857142857143}
{"step": 12185, "loss": 3.9892250061035157, "grad_norm": 0.6890974640846252, "learning_rate": 0.0001, "elbo": 3.98922, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7145789350222848}
{"step": 12190, "loss": 3.991257095336914, "grad_norm": 0.747166633605957, "learning_rate": 0.0001, "elbo": 3.99126, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7148721557588553}
{"step": 12195, "loss": 3.9993911743164063, "grad_norm": 0.7051324844360352, "learning_rate": 0.0001, "elbo": 3.99939, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7151653764954258}
{"step": 12200, "loss": 3.999209976196289, "grad_norm": 0.8986974954605103, "learning_rate": 0.0001, "elbo": 3.99921, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7154585972319962}
{"step": 12205, "loss": 3.9727188110351563, "grad_norm": 0.8163567781448364, "learning_rate": 0.0001, "elbo": 3.97272, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7157518179685667}
{"step": 12210, "loss": 3.992687225341797, "grad_norm": 0.6979306936264038, "learning_rate": 0.0001, "elbo": 3.99269, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7160450387051372}
{"step": 12215, "loss": 3.9781333923339846, "grad_norm": 0.9128879308700562, "learning_rate": 0.0001, "elbo": 3.97813, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7163382594417077}
{"step": 12220, "loss": 3.9840938568115236, "grad_norm": 0.7624239921569824, "learning_rate": 0.0001, "elbo": 3.98409, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7166314801782782}
{"step": 12225, "loss": 3.994648742675781, "grad_norm": 0.7324520349502563, "learning_rate": 0.0001, "elbo": 3.99465, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7169247009148487}
{"step": 12230, "loss": 3.995928192138672, "grad_norm": 0.7593554258346558, "learning_rate": 0.0001, "elbo": 3.99593, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7172179216514192}
{"step": 12235, "loss": 3.981639099121094, "grad_norm": 0.7501670718193054, "learning_rate": 0.0001, "elbo": 3.98164, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7175111423879896}
{"step": 12240, "loss": 3.9873188018798826, "grad_norm": 0.7004395127296448, "learning_rate": 0.0001, "elbo": 3.98732, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7178043631245602}
{"step": 12245, "loss": 3.992720031738281, "grad_norm": 0.70015949010849, "learning_rate": 0.0001, "elbo": 3.99272, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7180975838611307}
{"step": 12250, "loss": 4.004860305786133, "grad_norm": 0.7283539175987244, "learning_rate": 0.0001, "elbo": 4.00486, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7183908045977011}
{"step": 12255, "loss": 3.980371856689453, "grad_norm": 0.7363201975822449, "learning_rate": 0.0001, "elbo": 3.98037, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7186840253342717}
{"step": 12260, "loss": 3.989376449584961, "grad_norm": 0.7286770939826965, "learning_rate": 0.0001, "elbo": 3.98938, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7189772460708421}
{"step": 12265, "loss": 3.985258865356445, "grad_norm": 0.8240026831626892, "learning_rate": 0.0001, "elbo": 3.98526, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7192704668074126}
{"step": 12270, "loss": 3.971230316162109, "grad_norm": 0.6952580809593201, "learning_rate": 0.0001, "elbo": 3.97123, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7195636875439831}
{"step": 12275, "loss": 3.976381301879883, "grad_norm": 0.7465401291847229, "learning_rate": 0.0001, "elbo": 3.97638, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7198569082805536}
{"step": 12280, "loss": 3.9918041229248047, "grad_norm": 0.700549304485321, "learning_rate": 0.0001, "elbo": 3.9918, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.720150129017124}
{"step": 12285, "loss": 3.9781944274902346, "grad_norm": 0.7429269552230835, "learning_rate": 0.0001, "elbo": 3.97819, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7204433497536946}
{"step": 12290, "loss": 3.977276611328125, "grad_norm": 0.7083912491798401, "learning_rate": 0.0001, "elbo": 3.97728, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7207365704902651}
{"step": 12295, "loss": 3.9817352294921875, "grad_norm": 0.8341633081436157, "learning_rate": 0.0001, "elbo": 3.98174, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7210297912268355}
{"step": 12300, "loss": 3.9917060852050783, "grad_norm": 0.7288970947265625, "learning_rate": 0.0001, "elbo": 3.99171, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7213230119634061}
{"step": 12305, "loss": 3.9793170928955077, "grad_norm": 0.7605525255203247, "learning_rate": 0.0001, "elbo": 3.97932, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7216162326999765}
{"step": 12305, "loss": 3.9793170928955077, "grad_norm": 0.7605525255203247, "learning_rate": 0.0001, "elbo": 3.97932, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7216162326999765}
{"step": 12310, "loss": 4.00775375366211, "grad_norm": 0.7392657995223999, "learning_rate": 0.0001, "elbo": 4.00775, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.721909453436547}
{"step": 12315, "loss": 3.991216278076172, "grad_norm": 0.7862195372581482, "learning_rate": 0.0001, "elbo": 3.99122, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7222026741731176}
{"step": 12320, "loss": 3.9868137359619142, "grad_norm": 0.9333333969116211, "learning_rate": 0.0001, "elbo": 3.98681, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.722495894909688}
{"step": 12325, "loss": 3.985002136230469, "grad_norm": 0.7733875513076782, "learning_rate": 0.0001, "elbo": 3.985, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7227891156462585}
{"step": 12330, "loss": 4.002764892578125, "grad_norm": 0.7652291655540466, "learning_rate": 0.0001, "elbo": 4.00276, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.723082336382829}
{"step": 12335, "loss": 4.012721252441406, "grad_norm": 0.7149873375892639, "learning_rate": 0.0001, "elbo": 4.01272, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7233755571193995}
{"step": 12340, "loss": 3.994196319580078, "grad_norm": 0.7638578414916992, "learning_rate": 0.0001, "elbo": 3.9942, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.72366877785597}
{"step": 12345, "loss": 3.994734191894531, "grad_norm": 0.7671326994895935, "learning_rate": 0.0001, "elbo": 3.99473, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7239619985925405}
{"step": 12350, "loss": 3.982587432861328, "grad_norm": 0.7200393676757812, "learning_rate": 0.0001, "elbo": 3.98259, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.724255219329111}
{"step": 12355, "loss": 3.9781524658203127, "grad_norm": 0.8522334098815918, "learning_rate": 0.0001, "elbo": 3.97815, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7245484400656814}
{"step": 12360, "loss": 3.984864044189453, "grad_norm": 0.7682991623878479, "learning_rate": 0.0001, "elbo": 3.98486, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.724841660802252}
{"step": 12365, "loss": 3.985243225097656, "grad_norm": 0.8185008764266968, "learning_rate": 0.0001, "elbo": 3.98524, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7251348815388224}
{"step": 12370, "loss": 3.9824676513671875, "grad_norm": 0.6991338729858398, "learning_rate": 0.0001, "elbo": 3.98247, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7254281022753929}
{"step": 12375, "loss": 3.987453079223633, "grad_norm": 0.6786594986915588, "learning_rate": 0.0001, "elbo": 3.98745, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7257213230119635}
{"step": 12380, "loss": 3.977046585083008, "grad_norm": 0.9430490732192993, "learning_rate": 0.0001, "elbo": 3.97705, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7260145437485339}
{"step": 12385, "loss": 3.9842689514160154, "grad_norm": 0.7352896928787231, "learning_rate": 0.0001, "elbo": 3.98427, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7263077644851044}
{"step": 12390, "loss": 3.9780418395996096, "grad_norm": 0.7758809924125671, "learning_rate": 0.0001, "elbo": 3.97804, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7266009852216748}
{"step": 12395, "loss": 3.97452392578125, "grad_norm": 0.7338746786117554, "learning_rate": 0.0001, "elbo": 3.97452, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7268942059582454}
{"step": 12400, "loss": 3.989791488647461, "grad_norm": 0.7599043846130371, "learning_rate": 0.0001, "elbo": 3.98979, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7271874266948158}
{"step": 12405, "loss": 3.991820526123047, "grad_norm": 0.6807523965835571, "learning_rate": 0.0001, "elbo": 3.99182, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7274806474313863}
{"step": 12410, "loss": 3.9789276123046875, "grad_norm": 0.7055266499519348, "learning_rate": 0.0001, "elbo": 3.97893, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7277738681679569}
{"step": 12415, "loss": 3.9861141204833985, "grad_norm": 0.7718476057052612, "learning_rate": 0.0001, "elbo": 3.98611, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7280670889045273}
{"step": 12420, "loss": 3.979386901855469, "grad_norm": 0.7402657270431519, "learning_rate": 0.0001, "elbo": 3.97939, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7283603096410978}
{"step": 12425, "loss": 3.9852035522460936, "grad_norm": 0.7419004440307617, "learning_rate": 0.0001, "elbo": 3.9852, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7286535303776683}
{"step": 12430, "loss": 3.9741775512695314, "grad_norm": 0.728891134262085, "learning_rate": 0.0001, "elbo": 3.97418, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7289467511142388}
{"step": 12435, "loss": 3.9770587921142577, "grad_norm": 0.8214954137802124, "learning_rate": 0.0001, "elbo": 3.97706, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7292399718508092}
{"step": 12440, "loss": 3.9800071716308594, "grad_norm": 0.7208756804466248, "learning_rate": 0.0001, "elbo": 3.98001, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7295331925873798}
{"step": 12445, "loss": 3.981891632080078, "grad_norm": 0.6735103130340576, "learning_rate": 0.0001, "elbo": 3.98189, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7298264133239503}
{"step": 12450, "loss": 3.971296691894531, "grad_norm": 0.7639864087104797, "learning_rate": 0.0001, "elbo": 3.9713, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7301196340605207}
{"step": 12455, "loss": 3.9751022338867186, "grad_norm": 0.9438875317573547, "learning_rate": 0.0001, "elbo": 3.9751, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7304128547970913}
{"step": 12460, "loss": 3.981858825683594, "grad_norm": 0.8128398060798645, "learning_rate": 0.0001, "elbo": 3.98186, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7307060755336617}
{"step": 12465, "loss": 3.9819561004638673, "grad_norm": 1.1555449962615967, "learning_rate": 0.0001, "elbo": 3.98196, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7309992962702322}
{"step": 12470, "loss": 3.9907569885253906, "grad_norm": 0.7221661806106567, "learning_rate": 0.0001, "elbo": 3.99076, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7312925170068028}
{"step": 12475, "loss": 3.978138732910156, "grad_norm": 0.7696574926376343, "learning_rate": 0.0001, "elbo": 3.97814, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7315857377433732}
{"step": 12480, "loss": 3.9715644836425783, "grad_norm": 0.7952724099159241, "learning_rate": 0.0001, "elbo": 3.97156, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7318789584799437}
{"step": 12485, "loss": 3.9953857421875, "grad_norm": 0.7298588156700134, "learning_rate": 0.0001, "elbo": 3.99539, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7321721792165142}
{"step": 12485, "loss": 3.9953857421875, "grad_norm": 0.7298588156700134, "learning_rate": 0.0001, "elbo": 3.99539, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7321721792165142}
{"step": 12490, "loss": 3.986371612548828, "grad_norm": 0.7473030686378479, "learning_rate": 0.0001, "elbo": 3.98637, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7324653999530847}
{"step": 12495, "loss": 3.976116180419922, "grad_norm": 0.7152776122093201, "learning_rate": 0.0001, "elbo": 3.97612, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7327586206896551}
{"step": 12500, "loss": 3.991254425048828, "grad_norm": 0.650210440158844, "learning_rate": 0.0001, "elbo": 3.99125, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7330518414262257}
{"step": 12505, "loss": 3.9821006774902346, "grad_norm": 0.8958038091659546, "learning_rate": 0.0001, "elbo": 3.9821, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7333450621627962}
{"step": 12510, "loss": 3.964899444580078, "grad_norm": 0.8062170147895813, "learning_rate": 0.0001, "elbo": 3.9649, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7336382828993666}
{"step": 12515, "loss": 3.9768856048583983, "grad_norm": 0.7953070998191833, "learning_rate": 0.0001, "elbo": 3.97689, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7339315036359372}
{"step": 12515, "loss": 3.9768856048583983, "grad_norm": 0.7953070998191833, "learning_rate": 0.0001, "elbo": 3.97689, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7339315036359372}
{"step": 12520, "loss": 3.9723316192626954, "grad_norm": 0.7219707369804382, "learning_rate": 0.0001, "elbo": 3.97233, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7342247243725076}
{"step": 12525, "loss": 3.9689685821533205, "grad_norm": 0.8035550117492676, "learning_rate": 0.0001, "elbo": 3.96897, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7345179451090781}
{"step": 12530, "loss": 3.979051971435547, "grad_norm": 1.0427204370498657, "learning_rate": 0.0001, "elbo": 3.97905, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7348111658456487}
{"step": 12535, "loss": 3.995196533203125, "grad_norm": 0.6824062466621399, "learning_rate": 0.0001, "elbo": 3.9952, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7351043865822191}
{"step": 12540, "loss": 3.9741954803466797, "grad_norm": 0.791115403175354, "learning_rate": 0.0001, "elbo": 3.9742, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7353976073187896}
{"step": 12545, "loss": 3.984108734130859, "grad_norm": 0.7633599042892456, "learning_rate": 0.0001, "elbo": 3.98411, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7356908280553601}
{"step": 12550, "loss": 3.974922943115234, "grad_norm": 0.7268097996711731, "learning_rate": 0.0001, "elbo": 3.97492, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7359840487919306}
{"step": 12555, "loss": 3.964244842529297, "grad_norm": 0.731468677520752, "learning_rate": 0.0001, "elbo": 3.96424, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.736277269528501}
{"step": 12560, "loss": 3.987176513671875, "grad_norm": 0.7314063906669617, "learning_rate": 0.0001, "elbo": 3.98718, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7365704902650716}
{"step": 12560, "loss": 3.987176513671875, "grad_norm": 0.7314063906669617, "learning_rate": 0.0001, "elbo": 3.98718, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7365704902650716}
{"step": 12565, "loss": 3.973424530029297, "grad_norm": 0.7295058965682983, "learning_rate": 0.0001, "elbo": 3.97342, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.736863711001642}
{"step": 12570, "loss": 3.984149932861328, "grad_norm": 0.800707221031189, "learning_rate": 0.0001, "elbo": 3.98415, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7371569317382125}
{"step": 12575, "loss": 3.9609180450439454, "grad_norm": 0.7643341422080994, "learning_rate": 0.0001, "elbo": 3.96092, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7374501524747831}
{"step": 12580, "loss": 3.995503616333008, "grad_norm": 0.8212182521820068, "learning_rate": 0.0001, "elbo": 3.9955, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7377433732113535}
{"step": 12585, "loss": 3.9741729736328124, "grad_norm": 0.7487315535545349, "learning_rate": 0.0001, "elbo": 3.97417, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.738036593947924}
{"step": 12585, "loss": 3.9741729736328124, "grad_norm": 0.7487315535545349, "learning_rate": 0.0001, "elbo": 3.97417, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.738036593947924}
{"step": 12590, "loss": 3.960711669921875, "grad_norm": 0.6979207992553711, "learning_rate": 0.0001, "elbo": 3.96071, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7383298146844944}
{"step": 12595, "loss": 3.975006103515625, "grad_norm": 0.7071486115455627, "learning_rate": 0.0001, "elbo": 3.97501, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.738623035421065}
{"step": 12600, "loss": 3.970417785644531, "grad_norm": 0.7435627579689026, "learning_rate": 0.0001, "elbo": 3.97042, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7389162561576355}
{"step": 12605, "loss": 3.9686664581298827, "grad_norm": 0.9337610006332397, "learning_rate": 0.0001, "elbo": 3.96867, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7392094768942059}
{"step": 12610, "loss": 3.9956512451171875, "grad_norm": 0.6951873898506165, "learning_rate": 0.0001, "elbo": 3.99565, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7395026976307765}
{"step": 12615, "loss": 3.972193145751953, "grad_norm": 0.8368632197380066, "learning_rate": 0.0001, "elbo": 3.97219, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7397959183673469}
{"step": 12620, "loss": 3.972075653076172, "grad_norm": 0.7966761589050293, "learning_rate": 0.0001, "elbo": 3.97208, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7400891391039174}
{"step": 12625, "loss": 3.9862350463867187, "grad_norm": 0.7181451916694641, "learning_rate": 0.0001, "elbo": 3.98624, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.740382359840488}
{"step": 12630, "loss": 3.9781448364257814, "grad_norm": 0.763146162033081, "learning_rate": 0.0001, "elbo": 3.97814, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7406755805770584}
{"step": 12635, "loss": 3.984114074707031, "grad_norm": 1.2805954217910767, "learning_rate": 0.0001, "elbo": 3.98411, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7409688013136289}
{"step": 12640, "loss": 3.9816715240478517, "grad_norm": 0.8256291747093201, "learning_rate": 0.0001, "elbo": 3.98167, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7412620220501994}
{"step": 12645, "loss": 3.9899940490722656, "grad_norm": 1.0024696588516235, "learning_rate": 0.0001, "elbo": 3.98999, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7415552427867699}
{"step": 12650, "loss": 3.9660114288330077, "grad_norm": 0.8024489879608154, "learning_rate": 0.0001, "elbo": 3.96601, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7418484635233403}
{"step": 12655, "loss": 3.9816856384277344, "grad_norm": 0.7163108587265015, "learning_rate": 0.0001, "elbo": 3.98169, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7421416842599109}
{"step": 12660, "loss": 3.9672161102294923, "grad_norm": 0.7583478093147278, "learning_rate": 0.0001, "elbo": 3.96722, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7424349049964813}
{"step": 12660, "loss": 3.9672161102294923, "grad_norm": 0.7583478093147278, "learning_rate": 0.0001, "elbo": 3.96722, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7424349049964813}
{"step": 12665, "loss": 3.957010269165039, "grad_norm": 0.8274374008178711, "learning_rate": 0.0001, "elbo": 3.95701, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7427281257330518}
{"step": 12670, "loss": 3.983864974975586, "grad_norm": 0.7438846826553345, "learning_rate": 0.0001, "elbo": 3.98387, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7430213464696224}
{"step": 12675, "loss": 3.9803398132324217, "grad_norm": 0.8989521861076355, "learning_rate": 0.0001, "elbo": 3.98034, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7433145672061928}
{"step": 12680, "loss": 3.975167083740234, "grad_norm": 0.7471759915351868, "learning_rate": 0.0001, "elbo": 3.97517, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7436077879427633}
{"step": 12685, "loss": 3.9758575439453123, "grad_norm": 0.6819429993629456, "learning_rate": 0.0001, "elbo": 3.97586, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7439010086793338}
{"step": 12690, "loss": 3.9833450317382812, "grad_norm": 0.7000651359558105, "learning_rate": 0.0001, "elbo": 3.98334, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7441942294159043}
{"step": 12695, "loss": 3.976262664794922, "grad_norm": 0.780905544757843, "learning_rate": 0.0001, "elbo": 3.97626, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7444874501524748}
{"step": 12700, "loss": 3.977892303466797, "grad_norm": 0.7795465588569641, "learning_rate": 0.0001, "elbo": 3.97789, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7447806708890453}
{"step": 12700, "loss": 3.977892303466797, "grad_norm": 0.7795465588569641, "learning_rate": 0.0001, "elbo": 3.97789, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7447806708890453}
{"step": 12705, "loss": 3.9637130737304687, "grad_norm": 0.7197383046150208, "learning_rate": 0.0001, "elbo": 3.96371, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7450738916256158}
{"step": 12705, "loss": 3.9637130737304687, "grad_norm": 0.7197383046150208, "learning_rate": 0.0001, "elbo": 3.96371, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7450738916256158}
{"step": 12710, "loss": 3.970328521728516, "grad_norm": 0.770021378993988, "learning_rate": 0.0001, "elbo": 3.97033, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7453671123621862}
{"step": 12715, "loss": 3.973583984375, "grad_norm": 0.7258472442626953, "learning_rate": 0.0001, "elbo": 3.97358, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7456603330987568}
{"step": 12720, "loss": 3.9751327514648436, "grad_norm": 0.7830584645271301, "learning_rate": 0.0001, "elbo": 3.97513, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7459535538353272}
{"step": 12725, "loss": 3.958943176269531, "grad_norm": 0.7509618997573853, "learning_rate": 0.0001, "elbo": 3.95894, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7462467745718977}
{"step": 12730, "loss": 3.9768692016601563, "grad_norm": 0.8420830965042114, "learning_rate": 0.0001, "elbo": 3.97687, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7465399953084683}
{"step": 12735, "loss": 3.9824695587158203, "grad_norm": 0.7466697692871094, "learning_rate": 0.0001, "elbo": 3.98247, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7468332160450387}
{"step": 12740, "loss": 3.975784683227539, "grad_norm": 0.7381055355072021, "learning_rate": 0.0001, "elbo": 3.97578, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7471264367816092}
{"step": 12745, "loss": 3.995395278930664, "grad_norm": 0.8834232687950134, "learning_rate": 0.0001, "elbo": 3.9954, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7474196575181797}
{"step": 12750, "loss": 3.96954460144043, "grad_norm": 0.7028665542602539, "learning_rate": 0.0001, "elbo": 3.96954, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7477128782547502}
{"step": 12755, "loss": 3.963910675048828, "grad_norm": 0.8245861530303955, "learning_rate": 0.0001, "elbo": 3.96391, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7480060989913206}
{"step": 12760, "loss": 3.969490814208984, "grad_norm": 0.7787524461746216, "learning_rate": 0.0001, "elbo": 3.96949, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7482993197278912}
{"step": 12765, "loss": 3.9678173065185547, "grad_norm": 0.7280220985412598, "learning_rate": 0.0001, "elbo": 3.96782, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7485925404644617}
{"step": 12770, "loss": 3.9688568115234375, "grad_norm": 0.7086752653121948, "learning_rate": 0.0001, "elbo": 3.96886, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7488857612010321}
{"step": 12775, "loss": 3.9749603271484375, "grad_norm": 0.685617983341217, "learning_rate": 0.0001, "elbo": 3.97496, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7491789819376026}
{"step": 12780, "loss": 3.982700729370117, "grad_norm": 0.7772814631462097, "learning_rate": 0.0001, "elbo": 3.9827, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7494722026741731}
{"step": 12785, "loss": 3.9743167877197267, "grad_norm": 0.711525559425354, "learning_rate": 0.0001, "elbo": 3.97432, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7497654234107436}
{"step": 12790, "loss": 3.996367263793945, "grad_norm": 0.7741665840148926, "learning_rate": 0.0001, "elbo": 3.99637, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.750058644147314}
{"step": 12795, "loss": 3.944602203369141, "grad_norm": 0.740139901638031, "learning_rate": 0.0001, "elbo": 3.9446, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7503518648838846}
{"step": 12800, "loss": 3.9768184661865233, "grad_norm": 0.7263373732566833, "learning_rate": 0.0001, "elbo": 3.97682, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7506450856204551}
{"step": 12805, "loss": 3.9719261169433593, "grad_norm": 0.7167874574661255, "learning_rate": 0.0001, "elbo": 3.97193, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7509383063570255}
{"step": 12810, "loss": 3.9769737243652346, "grad_norm": 0.7150084376335144, "learning_rate": 0.0001, "elbo": 3.97697, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7512315270935961}
{"step": 12815, "loss": 3.9835609436035155, "grad_norm": 0.6907564997673035, "learning_rate": 0.0001, "elbo": 3.98356, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7515247478301665}
{"step": 12820, "loss": 3.95780029296875, "grad_norm": 0.7319846153259277, "learning_rate": 0.0001, "elbo": 3.9578, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.751817968566737}
{"step": 12825, "loss": 3.9798885345458985, "grad_norm": 0.895020604133606, "learning_rate": 0.0001, "elbo": 3.97989, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7521111893033076}
{"step": 12830, "loss": 3.97593994140625, "grad_norm": 0.7283346652984619, "learning_rate": 0.0001, "elbo": 3.97594, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.752404410039878}
{"step": 12835, "loss": 3.97520751953125, "grad_norm": 0.7316550016403198, "learning_rate": 0.0001, "elbo": 3.97521, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7526976307764485}
{"step": 12840, "loss": 3.9546375274658203, "grad_norm": 0.7705691456794739, "learning_rate": 0.0001, "elbo": 3.95464, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.752990851513019}
{"step": 12845, "loss": 3.974772644042969, "grad_norm": 0.8255910873413086, "learning_rate": 0.0001, "elbo": 3.97477, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7532840722495895}
{"step": 12850, "loss": 3.9650619506835936, "grad_norm": 0.7982603311538696, "learning_rate": 0.0001, "elbo": 3.96506, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7535772929861599}
{"step": 12855, "loss": 3.970409393310547, "grad_norm": 0.7732062935829163, "learning_rate": 0.0001, "elbo": 3.97041, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7538705137227305}
{"step": 12860, "loss": 3.9644329071044924, "grad_norm": 0.7120054364204407, "learning_rate": 0.0001, "elbo": 3.96443, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.754163734459301}
{"step": 12865, "loss": 3.963753509521484, "grad_norm": 0.8250868916511536, "learning_rate": 0.0001, "elbo": 3.96375, "mask_frac": 0.5014, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7544569551958714}
{"step": 12870, "loss": 3.9671234130859374, "grad_norm": 0.7859631776809692, "learning_rate": 0.0001, "elbo": 3.96712, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.754750175932442}
{"step": 12875, "loss": 3.9563323974609377, "grad_norm": 0.7288811206817627, "learning_rate": 0.0001, "elbo": 3.95633, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7550433966690124}
{"step": 12880, "loss": 3.9772533416748046, "grad_norm": 0.7116135358810425, "learning_rate": 0.0001, "elbo": 3.97725, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7553366174055829}
{"step": 12885, "loss": 3.955687713623047, "grad_norm": 0.7943840622901917, "learning_rate": 0.0001, "elbo": 3.95569, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7556298381421535}
{"step": 12890, "loss": 3.9500076293945314, "grad_norm": 0.7217401266098022, "learning_rate": 0.0001, "elbo": 3.95001, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7559230588787239}
{"step": 12895, "loss": 3.9618125915527345, "grad_norm": 0.7100901007652283, "learning_rate": 0.0001, "elbo": 3.96181, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7562162796152944}
{"step": 12900, "loss": 3.968400573730469, "grad_norm": 0.7506474256515503, "learning_rate": 0.0001, "elbo": 3.9684, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7565095003518649}
{"step": 12905, "loss": 3.958087921142578, "grad_norm": 0.7722354531288147, "learning_rate": 0.0001, "elbo": 3.95809, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7568027210884354}
{"step": 12910, "loss": 3.9748886108398436, "grad_norm": 0.6789022088050842, "learning_rate": 0.0001, "elbo": 3.97489, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7570959418250058}
{"step": 12910, "loss": 3.9748886108398436, "grad_norm": 0.6789022088050842, "learning_rate": 0.0001, "elbo": 3.97489, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7570959418250058}
{"step": 12915, "loss": 3.952869415283203, "grad_norm": 0.7149224281311035, "learning_rate": 0.0001, "elbo": 3.95287, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7573891625615764}
{"step": 12920, "loss": 3.967711639404297, "grad_norm": 0.7425854206085205, "learning_rate": 0.0001, "elbo": 3.96771, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7576823832981469}
{"step": 12925, "loss": 3.9629364013671875, "grad_norm": 0.9373985528945923, "learning_rate": 0.0001, "elbo": 3.96294, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7579756040347173}
{"step": 12930, "loss": 3.9678253173828124, "grad_norm": 0.6931662559509277, "learning_rate": 0.0001, "elbo": 3.96783, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7582688247712879}
{"step": 12935, "loss": 3.9577991485595705, "grad_norm": 0.7599284052848816, "learning_rate": 0.0001, "elbo": 3.9578, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7585620455078583}
{"step": 12940, "loss": 3.9620513916015625, "grad_norm": 0.7112835645675659, "learning_rate": 0.0001, "elbo": 3.96205, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7588552662444288}
{"step": 12945, "loss": 3.972704315185547, "grad_norm": 0.7348747253417969, "learning_rate": 0.0001, "elbo": 3.9727, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7591484869809993}
{"step": 12950, "loss": 3.9549945831298827, "grad_norm": 0.7202005386352539, "learning_rate": 0.0001, "elbo": 3.95499, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7594417077175698}
{"step": 12955, "loss": 3.9647380828857424, "grad_norm": 0.7421839833259583, "learning_rate": 0.0001, "elbo": 3.96474, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7597349284541403}
{"step": 12960, "loss": 3.971051025390625, "grad_norm": 1.1098902225494385, "learning_rate": 0.0001, "elbo": 3.97105, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7600281491907108}
{"step": 12965, "loss": 3.9610107421875, "grad_norm": 0.7068226933479309, "learning_rate": 0.0001, "elbo": 3.96101, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7603213699272813}
{"step": 12970, "loss": 3.968560791015625, "grad_norm": 0.7537121176719666, "learning_rate": 0.0001, "elbo": 3.96856, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7606145906638517}
{"step": 12975, "loss": 3.924440383911133, "grad_norm": 0.7117179036140442, "learning_rate": 0.0001, "elbo": Infinity, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7609078114004222}
{"step": 12980, "loss": 3.9693492889404296, "grad_norm": 0.831415057182312, "learning_rate": 0.0001, "elbo": 3.96935, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7612010321369928}
{"step": 12985, "loss": 3.959238052368164, "grad_norm": 0.7619336843490601, "learning_rate": 0.0001, "elbo": 3.95924, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7614942528735632}
{"step": 12990, "loss": 3.995644760131836, "grad_norm": 0.7234070897102356, "learning_rate": 0.0001, "elbo": 3.99564, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7617874736101337}
{"step": 12990, "loss": 3.995644760131836, "grad_norm": 0.7234070897102356, "learning_rate": 0.0001, "elbo": 3.99564, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7617874736101337}
{"step": 12995, "loss": 3.9692344665527344, "grad_norm": 0.6852831244468689, "learning_rate": 0.0001, "elbo": 3.96923, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7620806943467042}
{"step": 12995, "loss": 3.9692344665527344, "grad_norm": 0.6852831244468689, "learning_rate": 0.0001, "elbo": 3.96923, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7620806943467042}
{"step": 13000, "loss": 3.9857242584228514, "grad_norm": 0.7691049575805664, "learning_rate": 0.0001, "elbo": 3.98572, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7623739150832747}
{"step": 13005, "loss": 3.9724586486816404, "grad_norm": 0.8029722571372986, "learning_rate": 0.0001, "elbo": 3.97246, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7626671358198451}
{"step": 13005, "loss": 3.9724586486816404, "grad_norm": 0.8029722571372986, "learning_rate": 0.0001, "elbo": 3.97246, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7626671358198451}
{"step": 13010, "loss": 3.9568614959716797, "grad_norm": 0.6621502637863159, "learning_rate": 0.0001, "elbo": 3.95686, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7629603565564157}
{"step": 13015, "loss": 3.9511383056640623, "grad_norm": 0.9340648055076599, "learning_rate": 0.0001, "elbo": 3.95114, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7632535772929862}
{"step": 13020, "loss": 3.9595005035400392, "grad_norm": 0.7489987015724182, "learning_rate": 0.0001, "elbo": 3.9595, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7635467980295566}
{"step": 13025, "loss": 3.9710906982421874, "grad_norm": 0.8018252849578857, "learning_rate": 0.0001, "elbo": 3.97109, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7638400187661272}
{"step": 13030, "loss": 3.9713790893554686, "grad_norm": 0.779914379119873, "learning_rate": 0.0001, "elbo": 3.97138, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7641332395026976}
{"step": 13035, "loss": 3.9680824279785156, "grad_norm": 0.6998621821403503, "learning_rate": 0.0001, "elbo": 3.96808, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7644264602392681}
{"step": 13040, "loss": 3.9586471557617187, "grad_norm": 0.7215809226036072, "learning_rate": 0.0001, "elbo": 3.95865, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7647196809758386}
{"step": 13045, "loss": 3.9639961242675783, "grad_norm": 0.8099877834320068, "learning_rate": 0.0001, "elbo": 3.964, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7650129017124091}
{"step": 13050, "loss": 3.9713836669921876, "grad_norm": 0.7465173006057739, "learning_rate": 0.0001, "elbo": 3.97138, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7653061224489796}
{"step": 13055, "loss": 3.9621387481689454, "grad_norm": 1.056357502937317, "learning_rate": 0.0001, "elbo": 3.96214, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7655993431855501}
{"step": 13060, "loss": 3.9674556732177733, "grad_norm": 0.7176072001457214, "learning_rate": 0.0001, "elbo": 3.96746, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7658925639221206}
{"step": 13065, "loss": 3.960645294189453, "grad_norm": 0.7490888237953186, "learning_rate": 0.0001, "elbo": 3.96065, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.766185784658691}
{"step": 13070, "loss": 3.976102828979492, "grad_norm": 0.7814699411392212, "learning_rate": 0.0001, "elbo": 3.9761, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7664790053952616}
{"step": 13075, "loss": 3.9645370483398437, "grad_norm": 0.7616294026374817, "learning_rate": 0.0001, "elbo": 3.96454, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.766772226131832}
{"step": 13080, "loss": 3.9597068786621095, "grad_norm": 0.6760324239730835, "learning_rate": 0.0001, "elbo": 3.95971, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7670654468684025}
{"step": 13085, "loss": 3.967833709716797, "grad_norm": 1.0494576692581177, "learning_rate": 0.0001, "elbo": 3.96783, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7673586676049731}
{"step": 13090, "loss": 3.963422393798828, "grad_norm": 0.7287184000015259, "learning_rate": 0.0001, "elbo": 3.96342, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7676518883415435}
{"step": 13095, "loss": 3.9836040496826173, "grad_norm": 0.741455614566803, "learning_rate": 0.0001, "elbo": 3.9836, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.767945109078114}
{"step": 13100, "loss": 3.989872360229492, "grad_norm": 0.772224485874176, "learning_rate": 0.0001, "elbo": 3.98987, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7682383298146845}
{"step": 13105, "loss": 3.9652473449707033, "grad_norm": 0.7540927529335022, "learning_rate": 0.0001, "elbo": 3.96525, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.768531550551255}
{"step": 13110, "loss": 3.9828567504882812, "grad_norm": 0.7446997165679932, "learning_rate": 0.0001, "elbo": 3.98286, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7688247712878254}
{"step": 13115, "loss": 3.9668437957763674, "grad_norm": 0.7571038007736206, "learning_rate": 0.0001, "elbo": 3.96684, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.769117992024396}
{"step": 13120, "loss": 3.9795631408691405, "grad_norm": 0.7639612555503845, "learning_rate": 0.0001, "elbo": 3.97956, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7694112127609665}
{"step": 13125, "loss": 3.9785526275634764, "grad_norm": 0.7658854722976685, "learning_rate": 0.0001, "elbo": 3.97855, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7697044334975369}
{"step": 13130, "loss": 3.9586700439453124, "grad_norm": 0.9800888895988464, "learning_rate": 0.0001, "elbo": 3.95867, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7699976542341075}
{"step": 13135, "loss": 3.9407413482666014, "grad_norm": 1.117918610572815, "learning_rate": 0.0001, "elbo": 3.94074, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7702908749706779}
{"step": 13140, "loss": 3.9682632446289063, "grad_norm": 0.8743329644203186, "learning_rate": 0.0001, "elbo": 3.96826, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7705840957072484}
{"step": 13145, "loss": 3.969366455078125, "grad_norm": 0.7817656993865967, "learning_rate": 0.0001, "elbo": 3.96937, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.770877316443819}
{"step": 13150, "loss": 3.974268341064453, "grad_norm": 0.7478367686271667, "learning_rate": 0.0001, "elbo": 3.97427, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7711705371803894}
{"step": 13155, "loss": 3.970046615600586, "grad_norm": 0.7730793952941895, "learning_rate": 0.0001, "elbo": 3.97005, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7714637579169599}
{"step": 13160, "loss": 3.9571453094482423, "grad_norm": 0.6961960196495056, "learning_rate": 0.0001, "elbo": 3.95715, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7717569786535303}
{"step": 13165, "loss": 3.960816192626953, "grad_norm": 0.6906630396842957, "learning_rate": 0.0001, "elbo": 3.96082, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7720501993901009}
{"step": 13170, "loss": 3.958168792724609, "grad_norm": 0.6753984093666077, "learning_rate": 0.0001, "elbo": 3.95817, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7723434201266713}
{"step": 13175, "loss": 3.964990234375, "grad_norm": 0.7582046985626221, "learning_rate": 0.0001, "elbo": 3.96499, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7726366408632418}
{"step": 13180, "loss": 3.9585479736328124, "grad_norm": 0.7902534008026123, "learning_rate": 0.0001, "elbo": 3.95855, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7729298615998124}
{"step": 13185, "loss": 3.9777191162109373, "grad_norm": 0.7079014182090759, "learning_rate": 0.0001, "elbo": 3.97772, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7732230823363828}
{"step": 13190, "loss": 3.960662841796875, "grad_norm": 0.7745235562324524, "learning_rate": 0.0001, "elbo": 3.96066, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7735163030729533}
{"step": 13195, "loss": 3.9605026245117188, "grad_norm": 0.8091423511505127, "learning_rate": 0.0001, "elbo": 3.9605, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7738095238095238}
{"step": 13200, "loss": 3.96845703125, "grad_norm": 0.7447068095207214, "learning_rate": 0.0001, "elbo": 3.96846, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7741027445460943}
{"step": 13205, "loss": 3.963877868652344, "grad_norm": 0.7039284706115723, "learning_rate": 0.0001, "elbo": 3.96388, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7743959652826647}
{"step": 13210, "loss": 3.9511741638183593, "grad_norm": 0.7222027778625488, "learning_rate": 0.0001, "elbo": 3.95117, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7746891860192353}
{"step": 13215, "loss": 3.9608577728271483, "grad_norm": 0.7483682632446289, "learning_rate": 0.0001, "elbo": 3.96086, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7749824067558058}
{"step": 13220, "loss": 3.960759735107422, "grad_norm": 0.7021785378456116, "learning_rate": 0.0001, "elbo": 3.96076, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7752756274923762}
{"step": 13225, "loss": 3.964081573486328, "grad_norm": 1.2362772226333618, "learning_rate": 0.0001, "elbo": 3.96408, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7755688482289468}
{"step": 13230, "loss": 3.9657413482666017, "grad_norm": 0.7252511978149414, "learning_rate": 0.0001, "elbo": 3.96574, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7758620689655172}
{"step": 13235, "loss": 3.958363723754883, "grad_norm": 0.7261228561401367, "learning_rate": 0.0001, "elbo": 3.95836, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7761552897020877}
{"step": 13240, "loss": 3.957305908203125, "grad_norm": 0.7566533088684082, "learning_rate": 0.0001, "elbo": 3.95731, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7764485104386583}
{"step": 13245, "loss": 3.957854461669922, "grad_norm": 0.6874170899391174, "learning_rate": 0.0001, "elbo": 3.95785, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7767417311752287}
{"step": 13250, "loss": 3.9478363037109374, "grad_norm": 0.7839134335517883, "learning_rate": 0.0001, "elbo": 3.94784, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7770349519117992}
{"step": 13255, "loss": 3.96168212890625, "grad_norm": 0.7092355489730835, "learning_rate": 0.0001, "elbo": 3.96168, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7773281726483697}
{"step": 13260, "loss": 3.956051254272461, "grad_norm": 0.8459446430206299, "learning_rate": 0.0001, "elbo": 3.95605, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7776213933849402}
{"step": 13265, "loss": 3.957886505126953, "grad_norm": 0.718984842300415, "learning_rate": 0.0001, "elbo": 3.95789, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7779146141215106}
{"step": 13270, "loss": 3.9412982940673826, "grad_norm": 0.8335031270980835, "learning_rate": 0.0001, "elbo": 3.9413, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7782078348580812}
{"step": 13275, "loss": 3.9509998321533204, "grad_norm": 0.7063925862312317, "learning_rate": 0.0001, "elbo": 3.951, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7785010555946517}
{"step": 13280, "loss": 3.9472930908203123, "grad_norm": 0.7773343920707703, "learning_rate": 0.0001, "elbo": 3.94729, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7787942763312221}
{"step": 13285, "loss": 3.966197967529297, "grad_norm": 0.7687731981277466, "learning_rate": 0.0001, "elbo": 3.9662, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7790874970677927}
{"step": 13290, "loss": 3.9702728271484373, "grad_norm": 0.7513619065284729, "learning_rate": 0.0001, "elbo": 3.97027, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7793807178043631}
{"step": 13295, "loss": 3.9678245544433595, "grad_norm": 0.720496654510498, "learning_rate": 0.0001, "elbo": 3.96782, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7796739385409336}
{"step": 13300, "loss": 3.9558425903320313, "grad_norm": 0.7210285067558289, "learning_rate": 0.0001, "elbo": 3.95584, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7799671592775042}
{"step": 13305, "loss": 3.9472381591796877, "grad_norm": 0.6896692514419556, "learning_rate": 0.0001, "elbo": 3.94724, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7802603800140746}
{"step": 13310, "loss": 3.954611968994141, "grad_norm": 0.6804596781730652, "learning_rate": 0.0001, "elbo": 3.95461, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7805536007506451}
{"step": 13315, "loss": 3.9635520935058595, "grad_norm": 0.7339571118354797, "learning_rate": 0.0001, "elbo": 3.96355, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7808468214872156}
{"step": 13320, "loss": 3.9591144561767577, "grad_norm": 0.6957741975784302, "learning_rate": 0.0001, "elbo": 3.95911, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7811400422237861}
{"step": 13325, "loss": 3.9586250305175783, "grad_norm": 0.7099934816360474, "learning_rate": 0.0001, "elbo": 3.95863, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7814332629603565}
{"step": 13330, "loss": 3.953496551513672, "grad_norm": 0.7616279125213623, "learning_rate": 0.0001, "elbo": 3.9535, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7817264836969271}
{"step": 13335, "loss": 3.9571128845214845, "grad_norm": 0.703370213508606, "learning_rate": 0.0001, "elbo": 3.95711, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7820197044334976}
{"step": 13340, "loss": 3.965504455566406, "grad_norm": 0.7153110504150391, "learning_rate": 0.0001, "elbo": 3.9655, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.782312925170068}
{"step": 13345, "loss": 3.9573715209960936, "grad_norm": 0.7227593660354614, "learning_rate": 0.0001, "elbo": 3.95737, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7826061459066386}
{"step": 13350, "loss": 3.9489593505859375, "grad_norm": 0.7303069233894348, "learning_rate": 0.0001, "elbo": 3.94896, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.782899366643209}
{"step": 13355, "loss": 3.951967239379883, "grad_norm": 1.0606023073196411, "learning_rate": 0.0001, "elbo": 3.95197, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7831925873797795}
{"step": 13360, "loss": 3.9603477478027345, "grad_norm": 0.7330567836761475, "learning_rate": 0.0001, "elbo": 3.96035, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7834858081163499}
{"step": 13365, "loss": 3.9564342498779297, "grad_norm": 0.8215835690498352, "learning_rate": 0.0001, "elbo": 3.95643, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7837790288529205}
{"step": 13370, "loss": 3.9557521820068358, "grad_norm": 0.750627338886261, "learning_rate": 0.0001, "elbo": 3.95575, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.784072249589491}
{"step": 13375, "loss": 3.9609241485595703, "grad_norm": 0.7555514574050903, "learning_rate": 0.0001, "elbo": 3.96092, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7843654703260614}
{"step": 13380, "loss": 3.9496086120605467, "grad_norm": 0.7091836333274841, "learning_rate": 0.0001, "elbo": 3.94961, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.784658691062632}
{"step": 13385, "loss": 3.9505138397216797, "grad_norm": 0.7226826548576355, "learning_rate": 0.0001, "elbo": 3.95051, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7849519117992024}
{"step": 13390, "loss": 3.947652816772461, "grad_norm": 0.6798979640007019, "learning_rate": 0.0001, "elbo": 3.94765, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7852451325357729}
{"step": 13395, "loss": 3.9587322235107423, "grad_norm": 0.7070361971855164, "learning_rate": 0.0001, "elbo": 3.95873, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7855383532723434}
{"step": 13400, "loss": 3.956651306152344, "grad_norm": 0.6956268548965454, "learning_rate": 0.0001, "elbo": 3.95665, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7858315740089139}
{"step": 13405, "loss": 3.948578643798828, "grad_norm": 0.7258760929107666, "learning_rate": 0.0001, "elbo": 3.94858, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7861247947454844}
{"step": 13410, "loss": 3.9518112182617187, "grad_norm": 1.05775785446167, "learning_rate": 0.0001, "elbo": 3.95181, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7864180154820549}
{"step": 13415, "loss": 3.9646717071533204, "grad_norm": 0.954519510269165, "learning_rate": 0.0001, "elbo": 3.96467, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7867112362186254}
{"step": 13420, "loss": 3.9627750396728514, "grad_norm": 0.7180382013320923, "learning_rate": 0.0001, "elbo": 3.96277, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7870044569551958}
{"step": 13425, "loss": 3.964055633544922, "grad_norm": 0.7049603462219238, "learning_rate": 0.0001, "elbo": 3.96406, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7872976776917664}
{"step": 13430, "loss": 3.955321502685547, "grad_norm": 0.7122533917427063, "learning_rate": 0.0001, "elbo": 3.95532, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7875908984283368}
{"step": 13435, "loss": 3.953515625, "grad_norm": 0.8392705321311951, "learning_rate": 0.0001, "elbo": 3.95352, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7878841191649073}
{"step": 13440, "loss": 3.947966384887695, "grad_norm": 0.7360411882400513, "learning_rate": 0.0001, "elbo": 3.94797, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7881773399014779}
{"step": 13445, "loss": 3.945137786865234, "grad_norm": 0.7367808818817139, "learning_rate": 0.0001, "elbo": 3.94514, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7884705606380483}
{"step": 13450, "loss": 3.944132995605469, "grad_norm": 0.7518209218978882, "learning_rate": 0.0001, "elbo": 3.94413, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7887637813746188}
{"step": 13455, "loss": 3.9823844909667967, "grad_norm": 0.9990326166152954, "learning_rate": 0.0001, "elbo": 3.98238, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7890570021111893}
{"step": 13460, "loss": 3.979108428955078, "grad_norm": 0.7322399616241455, "learning_rate": 0.0001, "elbo": 3.97911, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7893502228477598}
{"step": 13465, "loss": 3.967193603515625, "grad_norm": 0.7223077416419983, "learning_rate": 0.0001, "elbo": 3.96719, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7896434435843303}
{"step": 13470, "loss": 3.950548553466797, "grad_norm": 0.8250237703323364, "learning_rate": 0.0001, "elbo": 3.95055, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7899366643209008}
{"step": 13475, "loss": 3.9412776947021486, "grad_norm": 0.7944087386131287, "learning_rate": 0.0001, "elbo": 3.94128, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7902298850574713}
{"step": 13480, "loss": 3.9648208618164062, "grad_norm": 0.7147793173789978, "learning_rate": 0.0001, "elbo": 3.96482, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7905231057940417}
{"step": 13485, "loss": 3.9441177368164064, "grad_norm": 0.7135763764381409, "learning_rate": 0.0001, "elbo": 3.94412, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7908163265306123}
{"step": 13490, "loss": 3.956621551513672, "grad_norm": 0.7421659827232361, "learning_rate": 0.0001, "elbo": 3.95662, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7911095472671827}
{"step": 13495, "loss": 3.942006301879883, "grad_norm": 0.7136695384979248, "learning_rate": 0.0001, "elbo": 3.94201, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7914027680037532}
{"step": 13500, "loss": 3.970653533935547, "grad_norm": 0.7182351350784302, "learning_rate": 0.0001, "elbo": 3.97065, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7916959887403238}
{"step": 13505, "loss": 3.950139617919922, "grad_norm": 0.7432692646980286, "learning_rate": 0.0001, "elbo": 3.95014, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7919892094768942}
{"step": 13510, "loss": 3.9485206604003906, "grad_norm": 0.7681848406791687, "learning_rate": 0.0001, "elbo": 3.94852, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7922824302134647}
{"step": 13515, "loss": 3.9451080322265626, "grad_norm": 0.6806999444961548, "learning_rate": 0.0001, "elbo": 3.94511, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7925756509500352}
{"step": 13520, "loss": 3.9528781890869142, "grad_norm": 0.691135585308075, "learning_rate": 0.0001, "elbo": 3.95288, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7928688716866057}
{"step": 13525, "loss": 3.954218292236328, "grad_norm": 0.731604278087616, "learning_rate": 0.0001, "elbo": 3.95422, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7931620924231761}
{"step": 13530, "loss": 3.9486351013183594, "grad_norm": 0.6909167766571045, "learning_rate": 0.0001, "elbo": 3.94863, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7934553131597467}
{"step": 13535, "loss": 3.942401885986328, "grad_norm": 0.7455182671546936, "learning_rate": 0.0001, "elbo": 3.9424, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7937485338963172}
{"step": 13540, "loss": 3.945587158203125, "grad_norm": 0.6895268559455872, "learning_rate": 0.0001, "elbo": 3.94559, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7940417546328876}
{"step": 13545, "loss": 3.948428726196289, "grad_norm": 0.7162792086601257, "learning_rate": 0.0001, "elbo": 3.94843, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7943349753694581}
{"step": 13550, "loss": 3.955326461791992, "grad_norm": 0.6871119737625122, "learning_rate": 0.0001, "elbo": 3.95533, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7946281961060286}
{"step": 13555, "loss": 3.944024658203125, "grad_norm": 0.6907575726509094, "learning_rate": 0.0001, "elbo": 3.94402, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7949214168425991}
{"step": 13560, "loss": 3.958443450927734, "grad_norm": 0.6774381995201111, "learning_rate": 0.0001, "elbo": 3.95844, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7952146375791695}
{"step": 13565, "loss": 3.9593173980712892, "grad_norm": 0.6846628189086914, "learning_rate": 0.0001, "elbo": 3.95932, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7955078583157401}
{"step": 13570, "loss": 3.9469654083251955, "grad_norm": 0.8424345850944519, "learning_rate": 0.0001, "elbo": 3.94697, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7958010790523106}
{"step": 13575, "loss": 3.9272247314453126, "grad_norm": 0.6821775436401367, "learning_rate": 0.0001, "elbo": 3.92723, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.796094299788881}
{"step": 13580, "loss": 3.933309555053711, "grad_norm": 0.7103540301322937, "learning_rate": 0.0001, "elbo": 3.93331, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7963875205254516}
{"step": 13585, "loss": 3.9265941619873046, "grad_norm": 0.8792213201522827, "learning_rate": 0.0001, "elbo": 3.92659, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.796680741262022}
{"step": 13590, "loss": 3.95787353515625, "grad_norm": 0.691135048866272, "learning_rate": 0.0001, "elbo": 3.95787, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7969739619985925}
{"step": 13595, "loss": 3.943872833251953, "grad_norm": 0.6748220920562744, "learning_rate": 0.0001, "elbo": 3.94387, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7972671827351631}
{"step": 13600, "loss": 3.951702880859375, "grad_norm": 0.6726579070091248, "learning_rate": 0.0001, "elbo": 3.9517, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7975604034717335}
{"step": 13605, "loss": 3.954478073120117, "grad_norm": 0.7509181499481201, "learning_rate": 0.0001, "elbo": 3.95448, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.797853624208304}
{"step": 13610, "loss": 3.9567420959472654, "grad_norm": 0.7892853617668152, "learning_rate": 0.0001, "elbo": 3.95674, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7981468449448745}
{"step": 13615, "loss": 3.9377010345458983, "grad_norm": 0.7376644015312195, "learning_rate": 0.0001, "elbo": 3.9377, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.798440065681445}
{"step": 13620, "loss": 3.96363525390625, "grad_norm": 0.7297573685646057, "learning_rate": 0.0001, "elbo": 3.96364, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7987332864180154}
{"step": 13625, "loss": 3.932318115234375, "grad_norm": 0.713489294052124, "learning_rate": 0.0001, "elbo": 3.93232, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.799026507154586}
{"step": 13630, "loss": 3.9686450958251953, "grad_norm": 1.2017136812210083, "learning_rate": 0.0001, "elbo": 3.96865, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7993197278911565}
{"step": 13635, "loss": 3.948334884643555, "grad_norm": 0.8180814385414124, "learning_rate": 0.0001, "elbo": 3.94833, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7996129486277269}
{"step": 13640, "loss": 3.9402839660644533, "grad_norm": 0.7914928197860718, "learning_rate": 0.0001, "elbo": 3.94028, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.7999061693642975}
{"step": 13645, "loss": 3.9457305908203124, "grad_norm": 0.8496752381324768, "learning_rate": 0.0001, "elbo": 3.94573, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8001993901008679}
{"step": 13650, "loss": 3.9500747680664063, "grad_norm": 0.9097112417221069, "learning_rate": 0.0001, "elbo": 3.95008, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8004926108374384}
{"step": 13655, "loss": 3.950562286376953, "grad_norm": 0.7735891938209534, "learning_rate": 0.0001, "elbo": 3.95056, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.800785831574009}
{"step": 13660, "loss": 3.949684906005859, "grad_norm": 0.8111099600791931, "learning_rate": 0.0001, "elbo": 3.94968, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8010790523105794}
{"step": 13665, "loss": 3.9646209716796874, "grad_norm": 0.7821014523506165, "learning_rate": 0.0001, "elbo": 3.96462, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8013722730471499}
{"step": 13670, "loss": 3.950963592529297, "grad_norm": 0.7485494017601013, "learning_rate": 0.0001, "elbo": 3.95096, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8016654937837204}
{"step": 13675, "loss": 3.9571609497070312, "grad_norm": 0.7441568374633789, "learning_rate": 0.0001, "elbo": 3.95716, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8019587145202909}
{"step": 13680, "loss": 3.9521156311035157, "grad_norm": 0.8023244142532349, "learning_rate": 0.0001, "elbo": 3.95212, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8022519352568613}
{"step": 13685, "loss": 3.9430496215820314, "grad_norm": 0.7300165891647339, "learning_rate": 0.0001, "elbo": 3.94305, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8025451559934319}
{"step": 13690, "loss": 3.954570007324219, "grad_norm": 0.668544352054596, "learning_rate": 0.0001, "elbo": 3.95457, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8028383767300024}
{"step": 13695, "loss": 3.955445098876953, "grad_norm": 0.7221682667732239, "learning_rate": 0.0001, "elbo": 3.95544, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8031315974665728}
{"step": 13695, "loss": 3.955445098876953, "grad_norm": 0.7221682667732239, "learning_rate": 0.0001, "elbo": 3.95544, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8031315974665728}
{"step": 13700, "loss": 3.944749450683594, "grad_norm": 0.7444431781768799, "learning_rate": 0.0001, "elbo": 3.94475, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8034248182031434}
{"step": 13705, "loss": 3.945604705810547, "grad_norm": 0.7657429575920105, "learning_rate": 0.0001, "elbo": 3.9456, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8037180389397138}
{"step": 13710, "loss": 3.9650588989257813, "grad_norm": 0.7259165048599243, "learning_rate": 0.0001, "elbo": 3.96506, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8040112596762843}
{"step": 13710, "loss": 3.9650588989257813, "grad_norm": 0.7259165048599243, "learning_rate": 0.0001, "elbo": 3.96506, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8040112596762843}
{"step": 13715, "loss": 3.9487770080566404, "grad_norm": 0.7137026190757751, "learning_rate": 0.0001, "elbo": 3.94878, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8043044804128548}
{"step": 13715, "loss": 3.9487770080566404, "grad_norm": 0.7137026190757751, "learning_rate": 0.0001, "elbo": 3.94878, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8043044804128548}
{"step": 13720, "loss": 3.9399185180664062, "grad_norm": 0.68772953748703, "learning_rate": 0.0001, "elbo": 3.93992, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8045977011494253}
{"step": 13725, "loss": 3.950389862060547, "grad_norm": 0.681594967842102, "learning_rate": 0.0001, "elbo": 3.95039, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8048909218859958}
{"step": 13730, "loss": 3.9463871002197264, "grad_norm": 0.679650604724884, "learning_rate": 0.0001, "elbo": 3.94639, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8051841426225662}
{"step": 13735, "loss": 3.9426044464111327, "grad_norm": 0.6571642160415649, "learning_rate": 0.0001, "elbo": 3.9426, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8054773633591368}
{"step": 13740, "loss": 3.943439483642578, "grad_norm": 0.6947667598724365, "learning_rate": 0.0001, "elbo": 3.94344, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8057705840957072}
{"step": 13745, "loss": 3.944780731201172, "grad_norm": 0.7462324500083923, "learning_rate": 0.0001, "elbo": 3.94478, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8060638048322777}
{"step": 13750, "loss": 3.93076171875, "grad_norm": 0.7023314833641052, "learning_rate": 0.0001, "elbo": 3.93076, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8063570255688483}
{"step": 13755, "loss": 3.9569000244140624, "grad_norm": 0.7310894727706909, "learning_rate": 0.0001, "elbo": 3.9569, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8066502463054187}
{"step": 13760, "loss": 3.9536739349365235, "grad_norm": 0.715226948261261, "learning_rate": 0.0001, "elbo": 3.95367, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8069434670419892}
{"step": 13765, "loss": 3.949013519287109, "grad_norm": 0.7710257172584534, "learning_rate": 0.0001, "elbo": 3.94901, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8072366877785597}
{"step": 13770, "loss": 3.950933074951172, "grad_norm": 0.7428561449050903, "learning_rate": 0.0001, "elbo": 3.95093, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8075299085151302}
{"step": 13775, "loss": 3.9413246154785155, "grad_norm": 0.7708972692489624, "learning_rate": 0.0001, "elbo": 3.94132, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8078231292517006}
{"step": 13780, "loss": 3.9523719787597655, "grad_norm": 0.7585227489471436, "learning_rate": 0.0001, "elbo": 3.95237, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8081163499882712}
{"step": 13785, "loss": 3.955879974365234, "grad_norm": 0.7338880300521851, "learning_rate": 0.0001, "elbo": 3.95588, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8084095707248417}
{"step": 13790, "loss": 3.932796859741211, "grad_norm": 0.6980993151664734, "learning_rate": 0.0001, "elbo": 3.9328, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8087027914614121}
{"step": 13795, "loss": 3.946173095703125, "grad_norm": 0.7333190441131592, "learning_rate": 0.0001, "elbo": 3.94617, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8089960121979827}
{"step": 13800, "loss": 3.932748031616211, "grad_norm": 0.68276047706604, "learning_rate": 0.0001, "elbo": 3.93275, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8092892329345531}
{"step": 13805, "loss": 3.9366878509521483, "grad_norm": 0.7165934443473816, "learning_rate": 0.0001, "elbo": 3.93669, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8095824536711236}
{"step": 13810, "loss": 3.9395378112792967, "grad_norm": 0.7154456973075867, "learning_rate": 0.0001, "elbo": 3.93954, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8098756744076941}
{"step": 13815, "loss": 3.9473907470703127, "grad_norm": 0.7056788802146912, "learning_rate": 0.0001, "elbo": 3.94739, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8101688951442646}
{"step": 13820, "loss": 3.938878631591797, "grad_norm": 0.6909164190292358, "learning_rate": 0.0001, "elbo": 3.93888, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.810462115880835}
{"step": 13825, "loss": 3.931488800048828, "grad_norm": 0.6855966448783875, "learning_rate": 0.0001, "elbo": 3.93149, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8107553366174056}
{"step": 13830, "loss": 3.9373722076416016, "grad_norm": 0.7662737965583801, "learning_rate": 0.0001, "elbo": 3.93737, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8110485573539761}
{"step": 13835, "loss": 3.941677474975586, "grad_norm": 0.6847068071365356, "learning_rate": 0.0001, "elbo": 3.94168, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8113417780905465}
{"step": 13840, "loss": 3.945759582519531, "grad_norm": 0.6354483366012573, "learning_rate": 0.0001, "elbo": 3.94576, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8116349988271171}
{"step": 13845, "loss": 3.9420139312744142, "grad_norm": 1.424946665763855, "learning_rate": 0.0001, "elbo": 3.94201, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8119282195636875}
{"step": 13850, "loss": 3.9416324615478517, "grad_norm": 0.6793634295463562, "learning_rate": 0.0001, "elbo": 3.94163, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.812221440300258}
{"step": 13850, "loss": 3.9416324615478517, "grad_norm": 0.6793634295463562, "learning_rate": 0.0001, "elbo": 3.94163, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.812221440300258}
{"step": 13855, "loss": 3.9395477294921877, "grad_norm": 0.6828522682189941, "learning_rate": 0.0001, "elbo": 3.93955, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8125146610368286}
{"step": 13860, "loss": 3.951259994506836, "grad_norm": 0.7290768027305603, "learning_rate": 0.0001, "elbo": 3.95126, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.812807881773399}
{"step": 13865, "loss": 3.9337493896484377, "grad_norm": 0.7109860181808472, "learning_rate": 0.0001, "elbo": 3.93375, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8131011025099695}
{"step": 13870, "loss": 3.9444557189941407, "grad_norm": 0.7068204283714294, "learning_rate": 0.0001, "elbo": 3.94446, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.81339432324654}
{"step": 13875, "loss": 3.943987274169922, "grad_norm": 0.715007483959198, "learning_rate": 0.0001, "elbo": 3.94399, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8136875439831105}
{"step": 13880, "loss": 3.9402957916259767, "grad_norm": 0.6676851511001587, "learning_rate": 0.0001, "elbo": 3.9403, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.813980764719681}
{"step": 13885, "loss": 3.953002166748047, "grad_norm": 0.8066519498825073, "learning_rate": 0.0001, "elbo": 3.953, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8142739854562515}
{"step": 13890, "loss": 3.9449813842773436, "grad_norm": 0.7595664262771606, "learning_rate": 0.0001, "elbo": 3.94498, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.814567206192822}
{"step": 13895, "loss": 3.939029312133789, "grad_norm": 0.7806624174118042, "learning_rate": 0.0001, "elbo": 3.93903, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8148604269293924}
{"step": 13900, "loss": 3.936115264892578, "grad_norm": 0.723326563835144, "learning_rate": 0.0001, "elbo": 3.93612, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.815153647665963}
{"step": 13905, "loss": 3.925257110595703, "grad_norm": 0.8040821552276611, "learning_rate": 0.0001, "elbo": 3.92526, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8154468684025334}
{"step": 13910, "loss": 3.9514816284179686, "grad_norm": 0.7168756723403931, "learning_rate": 0.0001, "elbo": 3.95148, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8157400891391039}
{"step": 13915, "loss": 3.92958984375, "grad_norm": 0.6817865967750549, "learning_rate": 0.0001, "elbo": 3.92959, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8160333098756745}
{"step": 13920, "loss": 3.9451560974121094, "grad_norm": 0.8470290899276733, "learning_rate": 0.0001, "elbo": 3.94516, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8163265306122449}
{"step": 13925, "loss": 3.9449462890625, "grad_norm": 0.8096350431442261, "learning_rate": 0.0001, "elbo": 3.94495, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8166197513488154}
{"step": 13930, "loss": 3.950931167602539, "grad_norm": 0.7063377499580383, "learning_rate": 0.0001, "elbo": 3.95093, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8169129720853858}
{"step": 13935, "loss": 3.9646034240722656, "grad_norm": 0.8429863452911377, "learning_rate": 0.0001, "elbo": 3.9646, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8172061928219564}
{"step": 13940, "loss": 3.944673538208008, "grad_norm": 0.6609563231468201, "learning_rate": 0.0001, "elbo": 3.94467, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8174994135585268}
{"step": 13945, "loss": 3.9641864776611326, "grad_norm": 0.6787313222885132, "learning_rate": 0.0001, "elbo": 3.96419, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8177926342950973}
{"step": 13950, "loss": 3.947205352783203, "grad_norm": 0.6854975819587708, "learning_rate": 0.0001, "elbo": 3.94721, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8180858550316679}
{"step": 13955, "loss": 3.9369476318359373, "grad_norm": 0.7109336256980896, "learning_rate": 0.0001, "elbo": 3.93695, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8183790757682383}
{"step": 13955, "loss": 3.9369476318359373, "grad_norm": 0.7109336256980896, "learning_rate": 0.0001, "elbo": 3.93695, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8183790757682383}
{"step": 13960, "loss": 3.9261417388916016, "grad_norm": 0.7191494703292847, "learning_rate": 0.0001, "elbo": 3.92614, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8186722965048088}
{"step": 13965, "loss": 3.9389698028564455, "grad_norm": 0.7248720526695251, "learning_rate": 0.0001, "elbo": 3.93897, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8189655172413793}
{"step": 13970, "loss": 3.92689208984375, "grad_norm": 0.7354082465171814, "learning_rate": 0.0001, "elbo": 3.92689, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8192587379779498}
{"step": 13975, "loss": 3.9342571258544923, "grad_norm": 0.6922556757926941, "learning_rate": 0.0001, "elbo": 3.93426, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8195519587145202}
{"step": 13980, "loss": 3.9298942565917967, "grad_norm": 0.718302845954895, "learning_rate": 0.0001, "elbo": 3.92989, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8198451794510908}
{"step": 13985, "loss": 3.9414093017578127, "grad_norm": 0.6603794693946838, "learning_rate": 0.0001, "elbo": 3.94141, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8201384001876613}
{"step": 13990, "loss": 3.9426952362060548, "grad_norm": 0.6627614498138428, "learning_rate": 0.0001, "elbo": 3.9427, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8204316209242317}
{"step": 13995, "loss": 3.9440277099609373, "grad_norm": 0.7325000166893005, "learning_rate": 0.0001, "elbo": 3.94403, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8207248416608023}
{"step": 14000, "loss": 3.938903045654297, "grad_norm": 0.861457884311676, "learning_rate": 0.0001, "elbo": 3.9389, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8210180623973727}
{"step": 14005, "loss": 3.9271171569824217, "grad_norm": 0.6783907413482666, "learning_rate": 0.0001, "elbo": 3.92712, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8213112831339432}
{"step": 14010, "loss": 3.93011474609375, "grad_norm": 0.7179871797561646, "learning_rate": 0.0001, "elbo": 3.93011, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8216045038705138}
{"step": 14015, "loss": 3.9497718811035156, "grad_norm": 0.9703930020332336, "learning_rate": 0.0001, "elbo": 3.94977, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8218977246070842}
{"step": 14020, "loss": 3.9447021484375, "grad_norm": 0.7751291394233704, "learning_rate": 0.0001, "elbo": 3.9447, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8221909453436547}
{"step": 14025, "loss": 3.931407928466797, "grad_norm": 0.7116492986679077, "learning_rate": 0.0001, "elbo": 3.93141, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8224841660802252}
{"step": 14030, "loss": 3.9590503692626955, "grad_norm": 0.7252249717712402, "learning_rate": 0.0001, "elbo": 3.95905, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8227773868167957}
{"step": 14035, "loss": 3.935166931152344, "grad_norm": 0.6696187257766724, "learning_rate": 0.0001, "elbo": 3.93517, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8230706075533661}
{"step": 14040, "loss": 3.9374454498291014, "grad_norm": 0.6966036558151245, "learning_rate": 0.0001, "elbo": 3.93745, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8233638282899367}
{"step": 14045, "loss": 3.9337615966796875, "grad_norm": 0.7113581895828247, "learning_rate": 0.0001, "elbo": 3.93376, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8236570490265072}
{"step": 14050, "loss": 3.9385608673095702, "grad_norm": 0.701473593711853, "learning_rate": 0.0001, "elbo": 3.93856, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8239502697630776}
{"step": 14055, "loss": 3.937216949462891, "grad_norm": 0.7254927754402161, "learning_rate": 0.0001, "elbo": 3.93722, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8242434904996482}
{"step": 14060, "loss": 3.9314083099365233, "grad_norm": 0.6849808692932129, "learning_rate": 0.0001, "elbo": 3.93141, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8245367112362186}
{"step": 14065, "loss": 3.925859832763672, "grad_norm": 0.6893289685249329, "learning_rate": 0.0001, "elbo": 3.92586, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8248299319727891}
{"step": 14070, "loss": 3.944163513183594, "grad_norm": 0.7798888683319092, "learning_rate": 0.0001, "elbo": 3.94416, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8251231527093597}
{"step": 14075, "loss": 3.9253379821777346, "grad_norm": 0.6936769485473633, "learning_rate": 0.0001, "elbo": 3.92534, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8254163734459301}
{"step": 14080, "loss": 3.934539794921875, "grad_norm": 0.7591990232467651, "learning_rate": 0.0001, "elbo": 3.93454, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8257095941825006}
{"step": 14085, "loss": 3.912598419189453, "grad_norm": 0.7229399085044861, "learning_rate": 0.0001, "elbo": 3.9126, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8260028149190711}
{"step": 14090, "loss": 3.9386878967285157, "grad_norm": 0.6959803104400635, "learning_rate": 0.0001, "elbo": 3.93869, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8262960356556416}
{"step": 14095, "loss": 3.931482696533203, "grad_norm": 0.9752503037452698, "learning_rate": 0.0001, "elbo": 3.93148, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.826589256392212}
{"step": 14100, "loss": 3.9412696838378904, "grad_norm": 0.80025714635849, "learning_rate": 0.0001, "elbo": 3.94127, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8268824771287826}
{"step": 14105, "loss": 3.9449554443359376, "grad_norm": 0.890360414981842, "learning_rate": 0.0001, "elbo": 3.94496, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.827175697865353}
{"step": 14110, "loss": 3.9632911682128906, "grad_norm": 0.7770546078681946, "learning_rate": 0.0001, "elbo": 3.96329, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8274689186019235}
{"step": 14115, "loss": 3.9485809326171877, "grad_norm": 0.7116581797599792, "learning_rate": 0.0001, "elbo": 3.94858, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.827762139338494}
{"step": 14120, "loss": 3.9277523040771483, "grad_norm": 0.7823325395584106, "learning_rate": 0.0001, "elbo": 3.92775, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8280553600750645}
{"step": 14125, "loss": 3.9506622314453126, "grad_norm": 0.714630663394928, "learning_rate": 0.0001, "elbo": 3.95066, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.828348580811635}
{"step": 14130, "loss": 3.929123306274414, "grad_norm": 0.7347106337547302, "learning_rate": 0.0001, "elbo": 3.92912, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8286418015482054}
{"step": 14135, "loss": 3.928898239135742, "grad_norm": 0.8138493299484253, "learning_rate": 0.0001, "elbo": 3.9289, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.828935022284776}
{"step": 14140, "loss": 3.9332809448242188, "grad_norm": 0.6861354112625122, "learning_rate": 0.0001, "elbo": 3.93328, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8292282430213465}
{"step": 14145, "loss": 3.9394500732421873, "grad_norm": 0.7155947685241699, "learning_rate": 0.0001, "elbo": 3.93945, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8295214637579169}
{"step": 14150, "loss": 3.9368324279785156, "grad_norm": 0.7356805205345154, "learning_rate": 0.0001, "elbo": 3.93683, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8298146844944875}
{"step": 14155, "loss": 3.9435630798339845, "grad_norm": 0.724087119102478, "learning_rate": 0.0001, "elbo": 3.94356, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8301079052310579}
{"step": 14160, "loss": 3.9530044555664063, "grad_norm": 0.7894868850708008, "learning_rate": 0.0001, "elbo": 3.953, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8304011259676284}
{"step": 14165, "loss": 3.9426719665527346, "grad_norm": 0.8679986596107483, "learning_rate": 0.0001, "elbo": 3.94267, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.830694346704199}
{"step": 14170, "loss": 3.922415924072266, "grad_norm": 0.7732303738594055, "learning_rate": 0.0001, "elbo": 3.92242, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8309875674407694}
{"step": 14175, "loss": 3.923126220703125, "grad_norm": 0.7538703083992004, "learning_rate": 0.0001, "elbo": 3.92313, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8312807881773399}
{"step": 14180, "loss": 3.9263259887695314, "grad_norm": 0.8334113955497742, "learning_rate": 0.0001, "elbo": 3.92633, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8315740089139104}
{"step": 14185, "loss": 3.9360145568847655, "grad_norm": 0.7147442102432251, "learning_rate": 0.0001, "elbo": 3.93601, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8318672296504809}
{"step": 14190, "loss": 3.9324409484863283, "grad_norm": 0.7306475639343262, "learning_rate": 0.0001, "elbo": 3.93244, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8321604503870513}
{"step": 14195, "loss": 3.9432952880859373, "grad_norm": 0.7115693688392639, "learning_rate": 0.0001, "elbo": 3.9433, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8324536711236219}
{"step": 14200, "loss": 3.9409271240234376, "grad_norm": 0.7026626467704773, "learning_rate": 0.0001, "elbo": 3.94093, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8327468918601924}
{"step": 14200, "loss": 3.9409271240234376, "grad_norm": 0.7026626467704773, "learning_rate": 0.0001, "elbo": 3.94093, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8327468918601924}
{"step": 14205, "loss": 3.921876907348633, "grad_norm": 0.7196137309074402, "learning_rate": 0.0001, "elbo": 3.92188, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8330401125967628}
{"step": 14210, "loss": 3.9276664733886717, "grad_norm": 0.7026599049568176, "learning_rate": 0.0001, "elbo": 3.92767, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8333333333333334}
{"step": 14215, "loss": 3.9261722564697266, "grad_norm": 0.6829244494438171, "learning_rate": 0.0001, "elbo": 3.92617, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8336265540699038}
{"step": 14220, "loss": 3.9238739013671875, "grad_norm": 0.9573344588279724, "learning_rate": 0.0001, "elbo": 3.92387, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8339197748064743}
{"step": 14225, "loss": 3.9364688873291014, "grad_norm": 0.8043171763420105, "learning_rate": 0.0001, "elbo": 3.93647, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8342129955430448}
{"step": 14230, "loss": 3.9454696655273436, "grad_norm": 0.686107337474823, "learning_rate": 0.0001, "elbo": 3.94547, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8345062162796153}
{"step": 14235, "loss": 3.926136016845703, "grad_norm": 0.763691782951355, "learning_rate": 0.0001, "elbo": 3.92614, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8347994370161858}
{"step": 14240, "loss": 3.934160614013672, "grad_norm": 0.8026384115219116, "learning_rate": 0.0001, "elbo": 3.93416, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8350926577527563}
{"step": 14245, "loss": 3.955760955810547, "grad_norm": 0.8357568979263306, "learning_rate": 0.0001, "elbo": 3.95576, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8353858784893268}
{"step": 14250, "loss": 3.9335922241210937, "grad_norm": 0.7520979046821594, "learning_rate": 0.0001, "elbo": 3.93359, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8356790992258972}
{"step": 14255, "loss": 3.914642333984375, "grad_norm": 0.8168923854827881, "learning_rate": 0.0001, "elbo": 3.91464, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8359723199624678}
{"step": 14255, "loss": 3.914642333984375, "grad_norm": 0.8168923854827881, "learning_rate": 0.0001, "elbo": 3.91464, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8359723199624678}
{"step": 14260, "loss": 3.9385147094726562, "grad_norm": 0.74410080909729, "learning_rate": 0.0001, "elbo": 3.93851, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8362655406990382}
{"step": 14265, "loss": 3.9439250946044924, "grad_norm": 0.7440984845161438, "learning_rate": 0.0001, "elbo": 3.94392, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8365587614356087}
{"step": 14270, "loss": 3.929448699951172, "grad_norm": 0.7028571367263794, "learning_rate": 0.0001, "elbo": 3.92945, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8368519821721793}
{"step": 14275, "loss": 3.945881652832031, "grad_norm": 0.9730269908905029, "learning_rate": 0.0001, "elbo": 3.94588, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8371452029087497}
{"step": 14280, "loss": 3.926007843017578, "grad_norm": 0.6616150736808777, "learning_rate": 0.0001, "elbo": 3.92601, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8374384236453202}
{"step": 14285, "loss": 3.9359481811523436, "grad_norm": 0.6765521168708801, "learning_rate": 0.0001, "elbo": 3.93595, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8377316443818907}
{"step": 14290, "loss": 3.9410701751708985, "grad_norm": 1.2718687057495117, "learning_rate": 0.0001, "elbo": 3.94107, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8380248651184612}
{"step": 14295, "loss": 3.9224334716796876, "grad_norm": 0.7012840509414673, "learning_rate": 0.0001, "elbo": 3.92243, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8383180858550316}
{"step": 14300, "loss": 3.930928039550781, "grad_norm": 0.7019351720809937, "learning_rate": 0.0001, "elbo": 3.93093, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8386113065916022}
{"step": 14305, "loss": 3.9324600219726564, "grad_norm": 0.7471646666526794, "learning_rate": 0.0001, "elbo": 3.93246, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8389045273281727}
{"step": 14310, "loss": 3.933167266845703, "grad_norm": 0.7132432460784912, "learning_rate": 0.0001, "elbo": 3.93317, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8391977480647431}
{"step": 14315, "loss": 3.938751983642578, "grad_norm": 0.9991176724433899, "learning_rate": 0.0001, "elbo": 3.93875, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8394909688013136}
{"step": 14320, "loss": 3.9445217132568358, "grad_norm": 0.7663125395774841, "learning_rate": 0.0001, "elbo": 3.94452, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8397841895378841}
{"step": 14325, "loss": 3.942278289794922, "grad_norm": 0.8070974946022034, "learning_rate": 0.0001, "elbo": 3.94228, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8400774102744546}
{"step": 14330, "loss": 3.9274875640869142, "grad_norm": 0.7409180998802185, "learning_rate": 0.0001, "elbo": 3.92749, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.840370631011025}
{"step": 14335, "loss": 3.9288261413574217, "grad_norm": 0.6681528687477112, "learning_rate": 0.0001, "elbo": 3.92883, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8406638517475956}
{"step": 14340, "loss": 3.9311569213867186, "grad_norm": 0.7108678817749023, "learning_rate": 0.0001, "elbo": 3.93116, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8409570724841661}
{"step": 14345, "loss": 3.9412216186523437, "grad_norm": 0.7150884866714478, "learning_rate": 0.0001, "elbo": 3.94122, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8412502932207365}
{"step": 14350, "loss": 3.9323211669921876, "grad_norm": 0.7098731398582458, "learning_rate": 0.0001, "elbo": 3.93232, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8415435139573071}
{"step": 14355, "loss": 3.9532539367675783, "grad_norm": 0.7244904041290283, "learning_rate": 0.0001, "elbo": 3.95325, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8418367346938775}
{"step": 14360, "loss": 3.922138977050781, "grad_norm": 0.7059333324432373, "learning_rate": 0.0001, "elbo": 3.92214, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.842129955430448}
{"step": 14365, "loss": 3.9427486419677735, "grad_norm": 0.8547398447990417, "learning_rate": 0.0001, "elbo": 3.94275, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8424231761670186}
{"step": 14370, "loss": 3.9391151428222657, "grad_norm": 0.827252984046936, "learning_rate": 0.0001, "elbo": 3.93911, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.842716396903589}
{"step": 14375, "loss": 3.954650115966797, "grad_norm": 0.885047435760498, "learning_rate": 0.0001, "elbo": 3.95465, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8430096176401595}
{"step": 14380, "loss": 3.928871917724609, "grad_norm": 0.7513026595115662, "learning_rate": 0.0001, "elbo": 3.92887, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.84330283837673}
{"step": 14385, "loss": 3.9356903076171874, "grad_norm": 0.7432697415351868, "learning_rate": 0.0001, "elbo": 3.93569, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8435960591133005}
{"step": 14390, "loss": 3.945698547363281, "grad_norm": 0.7113543748855591, "learning_rate": 0.0001, "elbo": 3.9457, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8438892798498709}
{"step": 14395, "loss": 3.9167076110839845, "grad_norm": 0.6966306567192078, "learning_rate": 0.0001, "elbo": 3.91671, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8441825005864415}
{"step": 14400, "loss": 3.9331733703613283, "grad_norm": 0.9154729843139648, "learning_rate": 0.0001, "elbo": 3.93317, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.844475721323012}
{"step": 14405, "loss": 3.938142776489258, "grad_norm": 0.6800400018692017, "learning_rate": 0.0001, "elbo": 3.93814, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8447689420595824}
{"step": 14410, "loss": 3.9389225006103517, "grad_norm": 0.8955810070037842, "learning_rate": 0.0001, "elbo": 3.93892, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.845062162796153}
{"step": 14415, "loss": 3.9381168365478514, "grad_norm": 0.6949982643127441, "learning_rate": 0.0001, "elbo": 3.93812, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8453553835327234}
{"step": 14420, "loss": 3.9364578247070314, "grad_norm": 0.877103328704834, "learning_rate": 0.0001, "elbo": 3.93646, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8456486042692939}
{"step": 14425, "loss": 3.927288818359375, "grad_norm": 0.8039405345916748, "learning_rate": 0.0001, "elbo": 3.92729, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8459418250058645}
{"step": 14430, "loss": 3.926125335693359, "grad_norm": 0.6896088719367981, "learning_rate": 0.0001, "elbo": 3.92613, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8462350457424349}
{"step": 14435, "loss": 3.92663688659668, "grad_norm": 0.6756700873374939, "learning_rate": 0.0001, "elbo": 3.92664, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8465282664790054}
{"step": 14440, "loss": 3.9262325286865236, "grad_norm": 0.6463046073913574, "learning_rate": 0.0001, "elbo": 3.92623, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8468214872155759}
{"step": 14445, "loss": 3.923831558227539, "grad_norm": 0.6850543022155762, "learning_rate": 0.0001, "elbo": 3.92383, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8471147079521464}
{"step": 14450, "loss": 3.9256366729736327, "grad_norm": 0.721243143081665, "learning_rate": 0.0001, "elbo": 3.92564, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8474079286887168}
{"step": 14455, "loss": 3.9278587341308593, "grad_norm": 0.7514412999153137, "learning_rate": 0.0001, "elbo": 3.92786, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8477011494252874}
{"step": 14460, "loss": 3.935334014892578, "grad_norm": 0.7000937461853027, "learning_rate": 0.0001, "elbo": 3.93533, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8479943701618579}
{"step": 14465, "loss": 3.946572113037109, "grad_norm": 0.7137024998664856, "learning_rate": 0.0001, "elbo": 3.94657, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8482875908984283}
{"step": 14470, "loss": 3.9303253173828123, "grad_norm": 0.6945671439170837, "learning_rate": 0.0001, "elbo": 3.93033, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8485808116349989}
{"step": 14475, "loss": 3.9306983947753906, "grad_norm": 0.689466118812561, "learning_rate": 0.0001, "elbo": 3.9307, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8488740323715693}
{"step": 14480, "loss": 3.927642059326172, "grad_norm": 0.7446926236152649, "learning_rate": 0.0001, "elbo": 3.92764, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8491672531081398}
{"step": 14485, "loss": 3.926888275146484, "grad_norm": 0.7452509999275208, "learning_rate": 0.0001, "elbo": 3.92689, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8494604738447104}
{"step": 14490, "loss": 3.919423294067383, "grad_norm": 0.7600051164627075, "learning_rate": 0.0001, "elbo": 3.91942, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8497536945812808}
{"step": 14495, "loss": 3.9411582946777344, "grad_norm": 0.7287554740905762, "learning_rate": 0.0001, "elbo": 3.94116, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8500469153178513}
{"step": 14500, "loss": 3.9236530303955077, "grad_norm": 0.7281007170677185, "learning_rate": 0.0001, "elbo": 3.92365, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8503401360544217}
{"step": 14505, "loss": 3.9379302978515627, "grad_norm": 0.6954300403594971, "learning_rate": 0.0001, "elbo": 3.93793, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8506333567909923}
{"step": 14510, "loss": 3.9146705627441407, "grad_norm": 0.6700117588043213, "learning_rate": 0.0001, "elbo": 3.91467, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8509265775275627}
{"step": 14515, "loss": 3.927687072753906, "grad_norm": 0.7360660433769226, "learning_rate": 0.0001, "elbo": 3.92769, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8512197982641332}
{"step": 14515, "loss": 3.927687072753906, "grad_norm": 0.7360660433769226, "learning_rate": 0.0001, "elbo": 3.92769, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8512197982641332}
{"step": 14520, "loss": 3.9239768981933594, "grad_norm": 0.7111667990684509, "learning_rate": 0.0001, "elbo": 3.92398, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8515130190007038}
{"step": 14525, "loss": 3.9298690795898437, "grad_norm": 0.7211822867393494, "learning_rate": 0.0001, "elbo": 3.92987, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8518062397372742}
{"step": 14530, "loss": 3.9294620513916017, "grad_norm": 0.742752730846405, "learning_rate": 0.0001, "elbo": 3.92946, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8520994604738447}
{"step": 14535, "loss": 3.917020797729492, "grad_norm": 0.8385570049285889, "learning_rate": 0.0001, "elbo": 3.91702, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8523926812104152}
{"step": 14540, "loss": 3.913410186767578, "grad_norm": 0.7227262854576111, "learning_rate": 0.0001, "elbo": 3.91341, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8526859019469857}
{"step": 14545, "loss": 3.9282791137695314, "grad_norm": 0.6844958066940308, "learning_rate": 0.0001, "elbo": 3.92828, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8529791226835561}
{"step": 14550, "loss": 3.9362106323242188, "grad_norm": 0.7657031416893005, "learning_rate": 0.0001, "elbo": 3.93621, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8532723434201267}
{"step": 14555, "loss": 3.9234169006347654, "grad_norm": 0.6464800238609314, "learning_rate": 0.0001, "elbo": 3.92342, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8535655641566972}
{"step": 14560, "loss": 3.910265350341797, "grad_norm": 0.8853412866592407, "learning_rate": 0.0001, "elbo": 3.91027, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8538587848932676}
{"step": 14565, "loss": 3.925889587402344, "grad_norm": 0.7971265316009521, "learning_rate": 0.0001, "elbo": 3.92589, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8541520056298382}
{"step": 14570, "loss": 3.9189090728759766, "grad_norm": 0.781083345413208, "learning_rate": 0.0001, "elbo": 3.91891, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8544452263664086}
{"step": 14575, "loss": 3.9204193115234376, "grad_norm": 0.7012718319892883, "learning_rate": 0.0001, "elbo": 3.92042, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8547384471029791}
{"step": 14580, "loss": 3.930613327026367, "grad_norm": 0.8175078630447388, "learning_rate": 0.0001, "elbo": 3.93061, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8550316678395496}
{"step": 14585, "loss": 3.919879913330078, "grad_norm": 0.6944919228553772, "learning_rate": 0.0001, "elbo": 3.91988, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8553248885761201}
{"step": 14590, "loss": 3.926082992553711, "grad_norm": 0.6759992241859436, "learning_rate": 0.0001, "elbo": 3.92608, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8556181093126906}
{"step": 14595, "loss": 3.9215766906738283, "grad_norm": 0.7818063497543335, "learning_rate": 0.0001, "elbo": 3.92158, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8559113300492611}
{"step": 14600, "loss": 3.9121639251708986, "grad_norm": 0.7259697318077087, "learning_rate": 0.0001, "elbo": 3.91216, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8562045507858316}
{"step": 14605, "loss": 3.9170135498046874, "grad_norm": 0.7319144010543823, "learning_rate": 0.0001, "elbo": 3.91701, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.856497771522402}
{"step": 14610, "loss": 3.915262222290039, "grad_norm": 0.7235801219940186, "learning_rate": 0.0001, "elbo": 3.91526, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8567909922589726}
{"step": 14615, "loss": 3.911993408203125, "grad_norm": 0.9289193153381348, "learning_rate": 0.0001, "elbo": 3.91199, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.857084212995543}
{"step": 14620, "loss": 3.923605728149414, "grad_norm": 0.7548310160636902, "learning_rate": 0.0001, "elbo": 3.92361, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8573774337321135}
{"step": 14625, "loss": 3.941880798339844, "grad_norm": 1.1674338579177856, "learning_rate": 0.0001, "elbo": 3.94188, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8576706544686841}
{"step": 14630, "loss": 3.919296646118164, "grad_norm": 0.7130007147789001, "learning_rate": 0.0001, "elbo": 3.9193, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8579638752052545}
{"step": 14635, "loss": 3.921971893310547, "grad_norm": 0.7519428730010986, "learning_rate": 0.0001, "elbo": 3.92197, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.858257095941825}
{"step": 14640, "loss": 3.9346508026123046, "grad_norm": 0.6934232115745544, "learning_rate": 0.0001, "elbo": 3.93465, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8585503166783955}
{"step": 14645, "loss": 3.9161571502685546, "grad_norm": 1.1238429546356201, "learning_rate": 0.0001, "elbo": 3.91616, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.858843537414966}
{"step": 14650, "loss": 3.925212860107422, "grad_norm": 0.7404669523239136, "learning_rate": 0.0001, "elbo": 3.92521, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8591367581515365}
{"step": 14655, "loss": 3.9196407318115236, "grad_norm": 0.7563150525093079, "learning_rate": 0.0001, "elbo": 3.91964, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.859429978888107}
{"step": 14660, "loss": 3.91126708984375, "grad_norm": 0.7106794714927673, "learning_rate": 0.0001, "elbo": 3.91127, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8597231996246775}
{"step": 14665, "loss": 3.916573333740234, "grad_norm": 0.6831326484680176, "learning_rate": 0.0001, "elbo": 3.91657, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8600164203612479}
{"step": 14670, "loss": 3.923389434814453, "grad_norm": 0.7890481352806091, "learning_rate": 0.0001, "elbo": 3.92339, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8603096410978185}
{"step": 14675, "loss": 3.9313488006591797, "grad_norm": 0.7210034132003784, "learning_rate": 0.0001, "elbo": 3.93135, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8606028618343889}
{"step": 14680, "loss": 3.9174537658691406, "grad_norm": 0.6876875162124634, "learning_rate": 0.0001, "elbo": 3.91745, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8608960825709594}
{"step": 14685, "loss": 3.913616180419922, "grad_norm": 0.6565881967544556, "learning_rate": 0.0001, "elbo": 3.91362, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.86118930330753}
{"step": 14690, "loss": 3.9368141174316404, "grad_norm": 0.6885596513748169, "learning_rate": 0.0001, "elbo": 3.93681, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8614825240441004}
{"step": 14695, "loss": 3.926563262939453, "grad_norm": 0.6677625775337219, "learning_rate": 0.0001, "elbo": 3.92656, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8617757447806709}
{"step": 14700, "loss": 3.9222976684570314, "grad_norm": 0.7105025053024292, "learning_rate": 0.0001, "elbo": 3.9223, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8620689655172413}
{"step": 14705, "loss": 3.9264755249023438, "grad_norm": 0.7344515323638916, "learning_rate": 0.0001, "elbo": 3.92648, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8623621862538119}
{"step": 14710, "loss": 3.921696090698242, "grad_norm": 0.9055591225624084, "learning_rate": 0.0001, "elbo": 3.9217, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8626554069903823}
{"step": 14715, "loss": 3.917500305175781, "grad_norm": 0.7112310528755188, "learning_rate": 0.0001, "elbo": 3.9175, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8629486277269528}
{"step": 14720, "loss": 3.909756088256836, "grad_norm": 0.9058635830879211, "learning_rate": 0.0001, "elbo": 3.90976, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8632418484635234}
{"step": 14725, "loss": 3.9371360778808593, "grad_norm": 0.6792476773262024, "learning_rate": 0.0001, "elbo": 3.93714, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8635350692000938}
{"step": 14730, "loss": 3.924676513671875, "grad_norm": 0.7133789658546448, "learning_rate": 0.0001, "elbo": 3.92468, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8638282899366643}
{"step": 14735, "loss": 3.9186016082763673, "grad_norm": 0.6606675982475281, "learning_rate": 0.0001, "elbo": 3.9186, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8641215106732348}
{"step": 14740, "loss": 3.9156494140625, "grad_norm": 0.6718937754631042, "learning_rate": 0.0001, "elbo": 3.91565, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8644147314098053}
{"step": 14745, "loss": 3.9082103729248048, "grad_norm": 0.7207398414611816, "learning_rate": 0.0001, "elbo": 3.90821, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8647079521463757}
{"step": 14750, "loss": 3.916643524169922, "grad_norm": 0.7212634682655334, "learning_rate": 0.0001, "elbo": 3.91664, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8650011728829463}
{"step": 14755, "loss": 3.9289230346679687, "grad_norm": 0.7047523856163025, "learning_rate": 0.0001, "elbo": 3.92892, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8652943936195168}
{"step": 14760, "loss": 3.9144638061523436, "grad_norm": 0.7163375616073608, "learning_rate": 0.0001, "elbo": 3.91446, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8655876143560872}
{"step": 14765, "loss": 3.929247283935547, "grad_norm": 0.6856943368911743, "learning_rate": 0.0001, "elbo": 3.92925, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8658808350926578}
{"step": 14770, "loss": 3.9279949188232424, "grad_norm": 0.872696042060852, "learning_rate": 0.0001, "elbo": 3.92799, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8661740558292282}
{"step": 14775, "loss": 3.9137989044189454, "grad_norm": 0.744635283946991, "learning_rate": 0.0001, "elbo": 3.9138, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8664672765657987}
{"step": 14780, "loss": 3.899460220336914, "grad_norm": 0.706207275390625, "learning_rate": 0.0001, "elbo": 3.89946, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8667604973023693}
{"step": 14785, "loss": 3.915183258056641, "grad_norm": 0.7048513889312744, "learning_rate": 0.0001, "elbo": 3.91518, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8670537180389397}
{"step": 14785, "loss": 3.915183258056641, "grad_norm": 0.7048513889312744, "learning_rate": 0.0001, "elbo": 3.91518, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8670537180389397}
{"step": 14790, "loss": 3.9186847686767576, "grad_norm": 0.6848391890525818, "learning_rate": 0.0001, "elbo": 3.91868, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8673469387755102}
{"step": 14795, "loss": 3.9211585998535154, "grad_norm": 0.7946265339851379, "learning_rate": 0.0001, "elbo": 3.92116, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8676401595120807}
{"step": 14800, "loss": 3.9341880798339846, "grad_norm": 0.6975570321083069, "learning_rate": 0.0001, "elbo": 3.93419, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8679333802486512}
{"step": 14805, "loss": 3.9249290466308593, "grad_norm": 0.6711987853050232, "learning_rate": 0.0001, "elbo": 3.92493, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8682266009852216}
{"step": 14810, "loss": 3.911749267578125, "grad_norm": 0.7643118500709534, "learning_rate": 0.0001, "elbo": 3.91175, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8685198217217922}
{"step": 14810, "loss": 3.911749267578125, "grad_norm": 0.7643118500709534, "learning_rate": 0.0001, "elbo": 3.91175, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8685198217217922}
{"step": 14815, "loss": 3.934825134277344, "grad_norm": 0.7042977809906006, "learning_rate": 0.0001, "elbo": 3.93482, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8688130424583627}
{"step": 14820, "loss": 3.914368438720703, "grad_norm": 0.7389717102050781, "learning_rate": 0.0001, "elbo": 3.91437, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8691062631949331}
{"step": 14820, "loss": 3.914368438720703, "grad_norm": 0.7389717102050781, "learning_rate": 0.0001, "elbo": 3.91437, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8691062631949331}
{"step": 14825, "loss": 3.9207401275634766, "grad_norm": 0.8011306524276733, "learning_rate": 0.0001, "elbo": 3.92074, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8693994839315037}
{"step": 14830, "loss": 3.9236686706542967, "grad_norm": 0.6843167543411255, "learning_rate": 0.0001, "elbo": 3.92367, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8696927046680741}
{"step": 14830, "loss": 3.9236686706542967, "grad_norm": 0.6843167543411255, "learning_rate": 0.0001, "elbo": 3.92367, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8696927046680741}
{"step": 14835, "loss": 3.9164073944091795, "grad_norm": 0.7264983654022217, "learning_rate": 0.0001, "elbo": 3.91641, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8699859254046446}
{"step": 14840, "loss": 3.9242012023925783, "grad_norm": 0.6826650500297546, "learning_rate": 0.0001, "elbo": 3.9242, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8702791461412152}
{"step": 14845, "loss": 3.9050926208496093, "grad_norm": 0.6870185136795044, "learning_rate": 0.0001, "elbo": 3.90509, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8705723668777856}
{"step": 14850, "loss": 3.915557861328125, "grad_norm": 0.7786411046981812, "learning_rate": 0.0001, "elbo": 3.91556, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8708655876143561}
{"step": 14855, "loss": 3.9148834228515623, "grad_norm": 0.7659633755683899, "learning_rate": 0.0001, "elbo": 3.91488, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8711588083509266}
{"step": 14860, "loss": 3.9218936920166017, "grad_norm": 0.731305718421936, "learning_rate": 0.0001, "elbo": 3.92189, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8714520290874971}
{"step": 14865, "loss": 3.9060592651367188, "grad_norm": 0.6726815700531006, "learning_rate": 0.0001, "elbo": 3.90606, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8717452498240675}
{"step": 14870, "loss": 3.915724182128906, "grad_norm": 0.6966370940208435, "learning_rate": 0.0001, "elbo": 3.91572, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8720384705606381}
{"step": 14875, "loss": 3.9109134674072266, "grad_norm": 0.8701578378677368, "learning_rate": 0.0001, "elbo": 3.91091, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8723316912972086}
{"step": 14880, "loss": 3.915663146972656, "grad_norm": 0.7306233644485474, "learning_rate": 0.0001, "elbo": 3.91566, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.872624912033779}
{"step": 14885, "loss": 3.9297603607177733, "grad_norm": 0.8523918986320496, "learning_rate": 0.0001, "elbo": 3.92976, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8729181327703495}
{"step": 14890, "loss": 3.912791061401367, "grad_norm": 1.187381625175476, "learning_rate": 0.0001, "elbo": 3.91279, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.87321135350692}
{"step": 14895, "loss": 3.9116680145263674, "grad_norm": 0.8057488203048706, "learning_rate": 0.0001, "elbo": 3.91167, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8735045742434905}
{"step": 14900, "loss": 3.91253547668457, "grad_norm": 0.7136684656143188, "learning_rate": 0.0001, "elbo": 3.91254, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8737977949800609}
{"step": 14905, "loss": 3.9278030395507812, "grad_norm": 0.6658750176429749, "learning_rate": 0.0001, "elbo": 3.9278, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8740910157166315}
{"step": 14910, "loss": 3.9172893524169923, "grad_norm": 0.7300441265106201, "learning_rate": 0.0001, "elbo": 3.91729, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.874384236453202}
{"step": 14915, "loss": 3.9251754760742186, "grad_norm": 0.8334526419639587, "learning_rate": 0.0001, "elbo": 3.92518, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8746774571897724}
{"step": 14920, "loss": 3.9231971740722655, "grad_norm": 0.6687991619110107, "learning_rate": 0.0001, "elbo": 3.9232, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.874970677926343}
{"step": 14925, "loss": 3.913364028930664, "grad_norm": 0.6903349161148071, "learning_rate": 0.0001, "elbo": 3.91336, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8752638986629134}
{"step": 14930, "loss": 3.926646423339844, "grad_norm": 0.7575923800468445, "learning_rate": 0.0001, "elbo": 3.92665, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8755571193994839}
{"step": 14935, "loss": 3.9153976440429688, "grad_norm": 0.7087998986244202, "learning_rate": 0.0001, "elbo": 3.9154, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8758503401360545}
{"step": 14940, "loss": 3.93048095703125, "grad_norm": 0.7916889786720276, "learning_rate": 0.0001, "elbo": 3.93048, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8761435608726249}
{"step": 14945, "loss": 3.9128692626953123, "grad_norm": 0.7743945717811584, "learning_rate": 0.0001, "elbo": 3.91287, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8764367816091954}
{"step": 14950, "loss": 3.9207107543945314, "grad_norm": 0.6745837926864624, "learning_rate": 0.0001, "elbo": 3.92071, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8767300023457659}
{"step": 14955, "loss": 3.9120269775390626, "grad_norm": 0.6857266426086426, "learning_rate": 0.0001, "elbo": 3.91203, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8770232230823364}
{"step": 14960, "loss": 3.928556442260742, "grad_norm": 0.7522261738777161, "learning_rate": 0.0001, "elbo": 3.92856, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8773164438189068}
{"step": 14965, "loss": 3.9067642211914064, "grad_norm": 0.7100756764411926, "learning_rate": 0.0001, "elbo": 3.90676, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8776096645554774}
{"step": 14970, "loss": 3.9093650817871093, "grad_norm": 0.746027410030365, "learning_rate": 0.0001, "elbo": 3.90937, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8779028852920479}
{"step": 14975, "loss": 3.914051818847656, "grad_norm": 0.7999956607818604, "learning_rate": 0.0001, "elbo": 3.91405, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8781961060286183}
{"step": 14980, "loss": 3.9279571533203126, "grad_norm": 0.735468327999115, "learning_rate": 0.0001, "elbo": 3.92796, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8784893267651889}
{"step": 14985, "loss": 3.9042129516601562, "grad_norm": 0.6986055374145508, "learning_rate": 0.0001, "elbo": 3.90421, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8787825475017593}
{"step": 14990, "loss": 3.933256912231445, "grad_norm": 0.6907299160957336, "learning_rate": 0.0001, "elbo": 3.93326, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8790757682383298}
{"step": 14995, "loss": 3.9111366271972656, "grad_norm": 0.7811946272850037, "learning_rate": 0.0001, "elbo": 3.91114, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8793689889749003}
{"step": 15000, "loss": 3.928599166870117, "grad_norm": 0.7411274313926697, "learning_rate": 0.0001, "elbo": 3.9286, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8796622097114708}
{"step": 15005, "loss": 3.922490692138672, "grad_norm": 0.8079403042793274, "learning_rate": 0.0001, "elbo": 3.92249, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8799554304480413}
{"step": 15005, "loss": 3.922490692138672, "grad_norm": 0.8079403042793274, "learning_rate": 0.0001, "elbo": 3.92249, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8799554304480413}
{"step": 15010, "loss": 3.919014739990234, "grad_norm": 0.8194673657417297, "learning_rate": 0.0001, "elbo": 3.91901, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8802486511846118}
{"step": 15010, "loss": 3.919014739990234, "grad_norm": 0.8194673657417297, "learning_rate": 0.0001, "elbo": 3.91901, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8802486511846118}
{"step": 15015, "loss": 3.937120819091797, "grad_norm": 0.8765370845794678, "learning_rate": 0.0001, "elbo": 3.93712, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8805418719211823}
{"step": 15020, "loss": 3.9035903930664064, "grad_norm": 0.8254178762435913, "learning_rate": 0.0001, "elbo": 3.90359, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8808350926577527}
{"step": 15020, "loss": 3.9035903930664064, "grad_norm": 0.8254178762435913, "learning_rate": 0.0001, "elbo": 3.90359, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8808350926577527}
{"step": 15025, "loss": 3.9169918060302735, "grad_norm": 0.7417984008789062, "learning_rate": 0.0001, "elbo": 3.91699, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8811283133943233}
{"step": 15030, "loss": 3.909515380859375, "grad_norm": 0.6987138986587524, "learning_rate": 0.0001, "elbo": 3.90952, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8814215341308937}
{"step": 15035, "loss": 3.9259880065917967, "grad_norm": 0.9304761290550232, "learning_rate": 0.0001, "elbo": 3.92599, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8817147548674642}
{"step": 15040, "loss": 3.9214248657226562, "grad_norm": 0.993668794631958, "learning_rate": 0.0001, "elbo": 3.92142, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8820079756040348}
{"step": 15045, "loss": 3.929117965698242, "grad_norm": 0.6919536590576172, "learning_rate": 0.0001, "elbo": 3.92912, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8823011963406052}
{"step": 15050, "loss": 3.9165138244628905, "grad_norm": 0.7054188251495361, "learning_rate": 0.0001, "elbo": 3.91651, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8825944170771757}
{"step": 15055, "loss": 3.922446441650391, "grad_norm": 0.7529137134552002, "learning_rate": 0.0001, "elbo": 3.92245, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8828876378137462}
{"step": 15060, "loss": 3.922684097290039, "grad_norm": 0.6734769940376282, "learning_rate": 0.0001, "elbo": 3.92268, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8831808585503167}
{"step": 15065, "loss": 3.899771499633789, "grad_norm": 0.7554317712783813, "learning_rate": 0.0001, "elbo": 3.89977, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8834740792868871}
{"step": 15070, "loss": 3.9068923950195313, "grad_norm": 0.7031293511390686, "learning_rate": 0.0001, "elbo": 3.90689, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8837673000234577}
{"step": 15075, "loss": 3.9241870880126952, "grad_norm": 0.683794379234314, "learning_rate": 0.0001, "elbo": 3.92419, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8840605207600282}
{"step": 15080, "loss": 3.9097946166992186, "grad_norm": 0.696912407875061, "learning_rate": 0.0001, "elbo": 3.90979, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8843537414965986}
{"step": 15085, "loss": 3.926347351074219, "grad_norm": 0.702730655670166, "learning_rate": 0.0001, "elbo": 3.92635, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8846469622331691}
{"step": 15090, "loss": 3.906980514526367, "grad_norm": 0.6818354725837708, "learning_rate": 0.0001, "elbo": 3.90698, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8849401829697396}
{"step": 15095, "loss": 3.9127666473388674, "grad_norm": 0.7667765021324158, "learning_rate": 0.0001, "elbo": 3.91277, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8852334037063101}
{"step": 15100, "loss": 3.9127296447753905, "grad_norm": 0.6841612458229065, "learning_rate": 0.0001, "elbo": 3.91273, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8855266244428805}
{"step": 15105, "loss": 3.918866729736328, "grad_norm": 0.6962152123451233, "learning_rate": 0.0001, "elbo": 3.91887, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8858198451794511}
{"step": 15110, "loss": 3.913145065307617, "grad_norm": 0.7278990149497986, "learning_rate": 0.0001, "elbo": 3.91315, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8861130659160216}
{"step": 15115, "loss": 3.9124046325683595, "grad_norm": 0.6994120478630066, "learning_rate": 0.0001, "elbo": 3.9124, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.886406286652592}
{"step": 15120, "loss": 3.920305633544922, "grad_norm": 0.7092472314834595, "learning_rate": 0.0001, "elbo": 3.92031, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8866995073891626}
{"step": 15125, "loss": 3.930080795288086, "grad_norm": 0.7022956013679504, "learning_rate": 0.0001, "elbo": 3.93008, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.886992728125733}
{"step": 15130, "loss": 3.901886749267578, "grad_norm": 0.7403676509857178, "learning_rate": 0.0001, "elbo": 3.90189, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8872859488623035}
{"step": 15135, "loss": 3.9059627532958983, "grad_norm": 0.6732778549194336, "learning_rate": 0.0001, "elbo": 3.90596, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8875791695988741}
{"step": 15140, "loss": 3.924494171142578, "grad_norm": 0.7144407629966736, "learning_rate": 0.0001, "elbo": 3.92449, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8878723903354445}
{"step": 15145, "loss": 3.9153076171875, "grad_norm": 0.6976706385612488, "learning_rate": 0.0001, "elbo": 3.91531, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.888165611072015}
{"step": 15150, "loss": 3.9217926025390626, "grad_norm": 0.6906530857086182, "learning_rate": 0.0001, "elbo": 3.92179, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8884588318085855}
{"step": 15155, "loss": 3.9072322845458984, "grad_norm": 0.7910943031311035, "learning_rate": 0.0001, "elbo": 3.90723, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.888752052545156}
{"step": 15160, "loss": 3.9075523376464845, "grad_norm": 0.7569636702537537, "learning_rate": 0.0001, "elbo": 3.90755, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8890452732817264}
{"step": 15165, "loss": 3.912963104248047, "grad_norm": 0.6770871877670288, "learning_rate": 0.0001, "elbo": 3.91296, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.889338494018297}
{"step": 15170, "loss": 3.907057189941406, "grad_norm": 0.7041548490524292, "learning_rate": 0.0001, "elbo": 3.90706, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8896317147548675}
{"step": 15175, "loss": 3.896067810058594, "grad_norm": 0.6976936459541321, "learning_rate": 0.0001, "elbo": 3.89607, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8899249354914379}
{"step": 15180, "loss": 3.9177947998046876, "grad_norm": 0.7546864151954651, "learning_rate": 0.0001, "elbo": 3.91779, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8902181562280085}
{"step": 15185, "loss": 3.927130126953125, "grad_norm": 0.679918110370636, "learning_rate": 0.0001, "elbo": 3.92713, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8905113769645789}
{"step": 15190, "loss": 3.91895751953125, "grad_norm": 0.7441970705986023, "learning_rate": 0.0001, "elbo": 3.91896, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8908045977011494}
{"step": 15195, "loss": 3.911337661743164, "grad_norm": 0.7869828939437866, "learning_rate": 0.0001, "elbo": 3.91134, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.89109781843772}
{"step": 15200, "loss": 3.910730743408203, "grad_norm": 0.7124872803688049, "learning_rate": 0.0001, "elbo": 3.91073, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8913910391742904}
{"step": 15205, "loss": 3.8911231994628905, "grad_norm": 0.8425612449645996, "learning_rate": 0.0001, "elbo": 3.89112, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8916842599108609}
{"step": 15210, "loss": 3.9062423706054688, "grad_norm": 0.7703009843826294, "learning_rate": 0.0001, "elbo": 3.90624, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8919774806474314}
{"step": 15215, "loss": 3.9051467895507814, "grad_norm": 0.775618314743042, "learning_rate": 0.0001, "elbo": 3.90515, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8922707013840019}
{"step": 15220, "loss": 3.9195602416992186, "grad_norm": 0.7218764424324036, "learning_rate": 0.0001, "elbo": 3.91956, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8925639221205723}
{"step": 15225, "loss": 3.903802490234375, "grad_norm": 0.685939371585846, "learning_rate": 0.0001, "elbo": 3.9038, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8928571428571429}
{"step": 15230, "loss": 3.910740280151367, "grad_norm": 0.7449996471405029, "learning_rate": 0.0001, "elbo": 3.91074, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8931503635937134}
{"step": 15235, "loss": 3.909479522705078, "grad_norm": 0.7353118658065796, "learning_rate": 0.0001, "elbo": 3.90948, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8934435843302838}
{"step": 15240, "loss": 3.91153564453125, "grad_norm": 0.6845181584358215, "learning_rate": 0.0001, "elbo": 3.91154, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8937368050668544}
{"step": 15245, "loss": 3.8934558868408202, "grad_norm": 0.9655044674873352, "learning_rate": 0.0001, "elbo": 3.89346, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8940300258034248}
{"step": 15250, "loss": 3.922101593017578, "grad_norm": 0.8210145235061646, "learning_rate": 0.0001, "elbo": 3.9221, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8943232465399953}
{"step": 15255, "loss": 3.896891403198242, "grad_norm": 0.6935737729072571, "learning_rate": 0.0001, "elbo": 3.89689, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8946164672765659}
{"step": 15260, "loss": 3.907257843017578, "grad_norm": 0.8746846318244934, "learning_rate": 0.0001, "elbo": 3.90726, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8949096880131363}
{"step": 15265, "loss": 3.911110687255859, "grad_norm": 0.9064338207244873, "learning_rate": 0.0001, "elbo": 3.91111, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8952029087497068}
{"step": 15270, "loss": 3.915818786621094, "grad_norm": 0.996780514717102, "learning_rate": 0.0001, "elbo": 3.91582, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8954961294862772}
{"step": 15275, "loss": 3.9268871307373048, "grad_norm": 0.694388210773468, "learning_rate": 0.0001, "elbo": 3.92689, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8957893502228478}
{"step": 15280, "loss": 3.9272697448730467, "grad_norm": 0.694218635559082, "learning_rate": 0.0001, "elbo": 3.92727, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8960825709594182}
{"step": 15285, "loss": 3.906427001953125, "grad_norm": 0.6777786016464233, "learning_rate": 0.0001, "elbo": 3.90643, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8963757916959887}
{"step": 15290, "loss": 3.9110626220703124, "grad_norm": 0.6947025060653687, "learning_rate": 0.0001, "elbo": 3.91106, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8966690124325593}
{"step": 15295, "loss": 3.901258087158203, "grad_norm": 0.7919085621833801, "learning_rate": 0.0001, "elbo": 3.90126, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8969622331691297}
{"step": 15300, "loss": 3.9071754455566405, "grad_norm": 0.7007303237915039, "learning_rate": 0.0001, "elbo": 3.90718, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8972554539057002}
{"step": 15305, "loss": 3.897726058959961, "grad_norm": 0.7235147953033447, "learning_rate": 0.0001, "elbo": 3.89773, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8975486746422707}
{"step": 15310, "loss": 3.9121883392333983, "grad_norm": 0.797180712223053, "learning_rate": 0.0001, "elbo": 3.91219, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8978418953788412}
{"step": 15315, "loss": 3.912248229980469, "grad_norm": 0.6672468185424805, "learning_rate": 0.0001, "elbo": 3.91225, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8981351161154116}
{"step": 15320, "loss": 3.8930286407470702, "grad_norm": 0.9407083988189697, "learning_rate": 0.0001, "elbo": 3.89303, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8984283368519822}
{"step": 15325, "loss": 3.8976814270019533, "grad_norm": 0.6941900849342346, "learning_rate": 0.0001, "elbo": 3.89768, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8987215575885527}
{"step": 15330, "loss": 3.9149734497070314, "grad_norm": 0.7341879606246948, "learning_rate": 0.0001, "elbo": 3.91497, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8990147783251231}
{"step": 15335, "loss": 3.908844757080078, "grad_norm": 0.7083783745765686, "learning_rate": 0.0001, "elbo": 3.90884, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8993079990616937}
{"step": 15340, "loss": 3.8991729736328127, "grad_norm": 0.6577613949775696, "learning_rate": 0.0001, "elbo": 3.89917, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8996012197982641}
{"step": 15345, "loss": 3.895852279663086, "grad_norm": 0.763042688369751, "learning_rate": 0.0001, "elbo": 3.89585, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.8998944405348346}
{"step": 15350, "loss": 3.9060211181640625, "grad_norm": 0.75171959400177, "learning_rate": 0.0001, "elbo": 3.90602, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9001876612714051}
{"step": 15355, "loss": 3.9045562744140625, "grad_norm": 0.7323753833770752, "learning_rate": 0.0001, "elbo": 3.90456, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9004808820079756}
{"step": 15360, "loss": 3.8965469360351563, "grad_norm": 0.6811570525169373, "learning_rate": 0.0001, "elbo": 3.89655, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9007741027445461}
{"step": 15365, "loss": 3.8976455688476563, "grad_norm": 0.752902626991272, "learning_rate": 0.0001, "elbo": 3.89765, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9010673234811166}
{"step": 15370, "loss": 3.911885452270508, "grad_norm": 0.799361526966095, "learning_rate": 0.0001, "elbo": 3.91189, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9013605442176871}
{"step": 15375, "loss": 3.9041770935058593, "grad_norm": 0.6851804256439209, "learning_rate": 0.0001, "elbo": 3.90418, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9016537649542575}
{"step": 15380, "loss": 3.8947959899902345, "grad_norm": 0.6810711026191711, "learning_rate": 0.0001, "elbo": 3.8948, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9019469856908281}
{"step": 15385, "loss": 3.900373077392578, "grad_norm": 0.7591833472251892, "learning_rate": 0.0001, "elbo": 3.90037, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9022402064273986}
{"step": 15390, "loss": 3.8980579376220703, "grad_norm": 0.7280128002166748, "learning_rate": 0.0001, "elbo": 3.89806, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.902533427163969}
{"step": 15395, "loss": 3.9105628967285155, "grad_norm": 0.7481987476348877, "learning_rate": 0.0001, "elbo": 3.91056, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9028266479005396}
{"step": 15395, "loss": 3.9105628967285155, "grad_norm": 0.7481987476348877, "learning_rate": 0.0001, "elbo": 3.91056, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9028266479005396}
{"step": 15400, "loss": 3.906882476806641, "grad_norm": 0.7360532283782959, "learning_rate": 0.0001, "elbo": 3.90688, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.90311986863711}
{"step": 15405, "loss": 3.905160903930664, "grad_norm": 0.7322742342948914, "learning_rate": 0.0001, "elbo": 3.90516, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9034130893736805}
{"step": 15410, "loss": 3.908112716674805, "grad_norm": 0.7087521553039551, "learning_rate": 0.0001, "elbo": 3.90811, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.903706310110251}
{"step": 15415, "loss": 3.900070571899414, "grad_norm": 0.7300769686698914, "learning_rate": 0.0001, "elbo": 3.90007, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9039995308468215}
{"step": 15420, "loss": 3.903850555419922, "grad_norm": 0.7247332334518433, "learning_rate": 0.0001, "elbo": 3.90385, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.904292751583392}
{"step": 15425, "loss": 3.9075279235839844, "grad_norm": 0.7442634105682373, "learning_rate": 0.0001, "elbo": 3.90753, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9045859723199625}
{"step": 15430, "loss": 3.910232162475586, "grad_norm": 0.6908485889434814, "learning_rate": 0.0001, "elbo": 3.91023, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.904879193056533}
{"step": 15435, "loss": 3.915835952758789, "grad_norm": 0.7352603077888489, "learning_rate": 0.0001, "elbo": 3.91584, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9051724137931034}
{"step": 15440, "loss": 3.9089035034179687, "grad_norm": 0.6798110604286194, "learning_rate": 0.0001, "elbo": 3.9089, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.905465634529674}
{"step": 15445, "loss": 3.903089141845703, "grad_norm": 0.7227689623832703, "learning_rate": 0.0001, "elbo": 3.90309, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9057588552662444}
{"step": 15450, "loss": 3.8973983764648437, "grad_norm": 0.6894391179084778, "learning_rate": 0.0001, "elbo": 3.8974, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9060520760028149}
{"step": 15455, "loss": 3.906077194213867, "grad_norm": 0.6906275749206543, "learning_rate": 0.0001, "elbo": 3.90608, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9063452967393855}
{"step": 15460, "loss": 3.9158889770507814, "grad_norm": 0.6927534341812134, "learning_rate": 0.0001, "elbo": 3.91589, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9066385174759559}
{"step": 15465, "loss": 3.916712188720703, "grad_norm": 0.6568702459335327, "learning_rate": 0.0001, "elbo": 3.91671, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9069317382125264}
{"step": 15470, "loss": 3.8975963592529297, "grad_norm": 0.658316969871521, "learning_rate": 0.0001, "elbo": 3.8976, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9072249589490968}
{"step": 15475, "loss": 3.898193359375, "grad_norm": 0.6778470277786255, "learning_rate": 0.0001, "elbo": 3.89819, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9075181796856674}
{"step": 15480, "loss": 3.9021671295166014, "grad_norm": 0.7124010920524597, "learning_rate": 0.0001, "elbo": 3.90217, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9078114004222378}
{"step": 15485, "loss": 3.8988059997558593, "grad_norm": 0.7068120837211609, "learning_rate": 0.0001, "elbo": 3.89881, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9081046211588083}
{"step": 15490, "loss": 3.8964393615722654, "grad_norm": 0.7930758595466614, "learning_rate": 0.0001, "elbo": 3.89644, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9083978418953789}
{"step": 15495, "loss": 3.915013885498047, "grad_norm": 0.7818783521652222, "learning_rate": 0.0001, "elbo": 3.91501, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9086910626319493}
{"step": 15500, "loss": 3.8976696014404295, "grad_norm": 0.6749697327613831, "learning_rate": 0.0001, "elbo": 3.89767, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9089842833685198}
{"step": 15505, "loss": 3.888209915161133, "grad_norm": 0.6885761618614197, "learning_rate": 0.0001, "elbo": 3.88821, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9092775041050903}
{"step": 15510, "loss": 3.923566436767578, "grad_norm": 0.996014416217804, "learning_rate": 0.0001, "elbo": 3.92357, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9095707248416608}
{"step": 15515, "loss": 3.8962799072265626, "grad_norm": 0.726693332195282, "learning_rate": 0.0001, "elbo": 3.89628, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9098639455782312}
{"step": 15520, "loss": 3.904368591308594, "grad_norm": 0.9664900898933411, "learning_rate": 0.0001, "elbo": 3.90437, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9101571663148018}
{"step": 15525, "loss": 3.9220817565917967, "grad_norm": 0.6795126795768738, "learning_rate": 0.0001, "elbo": 3.92208, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9104503870513723}
{"step": 15530, "loss": 3.9104461669921875, "grad_norm": 1.0292867422103882, "learning_rate": 0.0001, "elbo": 3.91045, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9107436077879427}
{"step": 15535, "loss": 3.9146713256835937, "grad_norm": 0.7070148587226868, "learning_rate": 0.0001, "elbo": 3.91467, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9110368285245133}
{"step": 15540, "loss": 3.8929466247558593, "grad_norm": 0.7102672457695007, "learning_rate": 0.0001, "elbo": 3.89295, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9113300492610837}
{"step": 15545, "loss": 3.884902572631836, "grad_norm": 0.8191619515419006, "learning_rate": 0.0001, "elbo": 3.8849, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9116232699976542}
{"step": 15550, "loss": 3.890738677978516, "grad_norm": 0.6955388784408569, "learning_rate": 0.0001, "elbo": 3.89074, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9119164907342248}
{"step": 15555, "loss": 3.9013427734375, "grad_norm": 0.8892243504524231, "learning_rate": 0.0001, "elbo": 3.90134, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9122097114707952}
{"step": 15555, "loss": 3.9013427734375, "grad_norm": 0.8892243504524231, "learning_rate": 0.0001, "elbo": 3.90134, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9122097114707952}
{"step": 15560, "loss": 3.909830856323242, "grad_norm": 0.6773487329483032, "learning_rate": 0.0001, "elbo": 3.90983, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9125029322073657}
{"step": 15565, "loss": 3.8945423126220704, "grad_norm": 0.7642573118209839, "learning_rate": 0.0001, "elbo": 3.89454, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9127961529439362}
{"step": 15570, "loss": 3.9054088592529297, "grad_norm": 0.6864359974861145, "learning_rate": 0.0001, "elbo": 3.90541, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9130893736805067}
{"step": 15575, "loss": 3.90989990234375, "grad_norm": 0.676336407661438, "learning_rate": 0.0001, "elbo": 3.9099, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9133825944170771}
{"step": 15580, "loss": 3.897599792480469, "grad_norm": 0.6961869597434998, "learning_rate": 0.0001, "elbo": 3.8976, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9136758151536477}
{"step": 15585, "loss": 3.9121490478515626, "grad_norm": 0.6721507906913757, "learning_rate": 0.0001, "elbo": 3.91215, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9139690358902182}
{"step": 15590, "loss": 3.892852783203125, "grad_norm": 0.7317367196083069, "learning_rate": 0.0001, "elbo": 3.89285, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9142622566267886}
{"step": 15595, "loss": 3.8975894927978514, "grad_norm": 0.7103834748268127, "learning_rate": 0.0001, "elbo": 3.89759, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9145554773633592}
{"step": 15595, "loss": 3.8975894927978514, "grad_norm": 0.7103834748268127, "learning_rate": 0.0001, "elbo": 3.89759, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9145554773633592}
{"step": 15600, "loss": 3.898659515380859, "grad_norm": 0.6906558871269226, "learning_rate": 0.0001, "elbo": 3.89866, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9148486980999296}
{"step": 15605, "loss": 3.8926773071289062, "grad_norm": 0.7139092087745667, "learning_rate": 0.0001, "elbo": 3.89268, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9151419188365001}
{"step": 15610, "loss": 3.8969207763671876, "grad_norm": 0.6797270774841309, "learning_rate": 0.0001, "elbo": 3.89692, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9154351395730707}
{"step": 15615, "loss": 3.884546661376953, "grad_norm": 0.684014618396759, "learning_rate": 0.0001, "elbo": 3.88455, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9157283603096411}
{"step": 15620, "loss": 3.905614471435547, "grad_norm": 0.9417849779129028, "learning_rate": 0.0001, "elbo": 3.90561, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9160215810462116}
{"step": 15625, "loss": 3.897621917724609, "grad_norm": 0.7395232319831848, "learning_rate": 0.0001, "elbo": 3.89762, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9163148017827821}
{"step": 15630, "loss": 3.89306640625, "grad_norm": 0.7026258707046509, "learning_rate": 0.0001, "elbo": 3.89307, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9166080225193526}
{"step": 15630, "loss": 3.89306640625, "grad_norm": 0.7026258707046509, "learning_rate": 0.0001, "elbo": 3.89307, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9166080225193526}
{"step": 15635, "loss": 3.8915740966796877, "grad_norm": 0.7715721130371094, "learning_rate": 0.0001, "elbo": 3.89157, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.916901243255923}
{"step": 15640, "loss": 3.8926364898681642, "grad_norm": 0.6602727174758911, "learning_rate": 0.0001, "elbo": 3.89264, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9171944639924936}
{"step": 15645, "loss": 3.8996627807617186, "grad_norm": 0.6938442587852478, "learning_rate": 0.0001, "elbo": 3.89966, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9174876847290641}
{"step": 15650, "loss": 3.912372589111328, "grad_norm": 0.7075089812278748, "learning_rate": 0.0001, "elbo": 3.91237, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9177809054656345}
{"step": 15655, "loss": 3.895093536376953, "grad_norm": 0.7496673464775085, "learning_rate": 0.0001, "elbo": 3.89509, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.918074126202205}
{"step": 15660, "loss": 3.915854263305664, "grad_norm": 0.710696280002594, "learning_rate": 0.0001, "elbo": 3.91585, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9183673469387755}
{"step": 15665, "loss": 3.8922260284423826, "grad_norm": 0.6809701323509216, "learning_rate": 0.0001, "elbo": 3.89223, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.918660567675346}
{"step": 15670, "loss": 3.8961448669433594, "grad_norm": 0.7528361678123474, "learning_rate": 0.0001, "elbo": 3.89614, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9189537884119164}
{"step": 15675, "loss": 3.8862632751464843, "grad_norm": 0.6792828440666199, "learning_rate": 0.0001, "elbo": 3.88626, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.919247009148487}
{"step": 15680, "loss": 3.8947853088378905, "grad_norm": 1.0376726388931274, "learning_rate": 0.0001, "elbo": 3.89479, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9195402298850575}
{"step": 15685, "loss": 3.894377517700195, "grad_norm": 0.7382336258888245, "learning_rate": 0.0001, "elbo": 3.89438, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9198334506216279}
{"step": 15690, "loss": 3.886853790283203, "grad_norm": 0.6760978698730469, "learning_rate": 0.0001, "elbo": 3.88685, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9201266713581985}
{"step": 15695, "loss": 3.9026222229003906, "grad_norm": 0.6523123979568481, "learning_rate": 0.0001, "elbo": 3.90262, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9204198920947689}
{"step": 15700, "loss": 3.907830810546875, "grad_norm": 0.8088993430137634, "learning_rate": 0.0001, "elbo": 3.90783, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9207131128313394}
{"step": 15705, "loss": 3.9118865966796874, "grad_norm": 0.7429893612861633, "learning_rate": 0.0001, "elbo": 3.91189, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.92100633356791}
{"step": 15710, "loss": 3.9022361755371096, "grad_norm": 1.2017822265625, "learning_rate": 0.0001, "elbo": 3.90224, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9212995543044804}
{"step": 15715, "loss": 3.893535614013672, "grad_norm": 0.6226740479469299, "learning_rate": 0.0001, "elbo": 3.89354, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9215927750410509}
{"step": 15720, "loss": 3.9004440307617188, "grad_norm": 0.9280105829238892, "learning_rate": 0.0001, "elbo": 3.90044, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9218859957776214}
{"step": 15725, "loss": 3.89046630859375, "grad_norm": 0.7431617975234985, "learning_rate": 0.0001, "elbo": 3.89047, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9221792165141919}
{"step": 15730, "loss": 3.902296447753906, "grad_norm": 0.6875286102294922, "learning_rate": 0.0001, "elbo": 3.9023, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9224724372507623}
{"step": 15735, "loss": 3.90106201171875, "grad_norm": 0.772994875907898, "learning_rate": 0.0001, "elbo": 3.90106, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9227656579873329}
{"step": 15740, "loss": 3.888875198364258, "grad_norm": 0.855135440826416, "learning_rate": 0.0001, "elbo": 3.88888, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9230588787239034}
{"step": 15745, "loss": 3.896800994873047, "grad_norm": 0.7196357846260071, "learning_rate": 0.0001, "elbo": 3.8968, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9233520994604738}
{"step": 15750, "loss": 3.890019989013672, "grad_norm": 0.8729459643363953, "learning_rate": 0.0001, "elbo": 3.89002, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9236453201970444}
{"step": 15755, "loss": 3.8987403869628907, "grad_norm": 0.6927309036254883, "learning_rate": 0.0001, "elbo": 3.89874, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9239385409336148}
{"step": 15760, "loss": 3.895668792724609, "grad_norm": 0.6862214207649231, "learning_rate": 0.0001, "elbo": 3.89567, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9242317616701853}
{"step": 15765, "loss": 3.9033626556396483, "grad_norm": 0.8316755890846252, "learning_rate": 0.0001, "elbo": 3.90336, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9245249824067558}
{"step": 15770, "loss": 3.9033660888671875, "grad_norm": 0.7000769376754761, "learning_rate": 0.0001, "elbo": 3.90337, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9248182031433263}
{"step": 15775, "loss": 3.9060592651367188, "grad_norm": 0.6686025857925415, "learning_rate": 0.0001, "elbo": 3.90606, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9251114238798968}
{"step": 15780, "loss": 3.9020721435546877, "grad_norm": 0.758908748626709, "learning_rate": 0.0001, "elbo": 3.90207, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9254046446164673}
{"step": 15785, "loss": 3.8750038146972656, "grad_norm": 0.6918037533760071, "learning_rate": 0.0001, "elbo": 3.875, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9256978653530378}
{"step": 15790, "loss": 3.9179569244384767, "grad_norm": 0.9411673545837402, "learning_rate": 0.0001, "elbo": 3.91796, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9259910860896082}
{"step": 15795, "loss": 3.9047782897949217, "grad_norm": 0.694246768951416, "learning_rate": 0.0001, "elbo": 3.90478, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9262843068261788}
{"step": 15800, "loss": 3.9148490905761717, "grad_norm": 0.680550217628479, "learning_rate": 0.0001, "elbo": 3.91485, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9265775275627492}
{"step": 15805, "loss": 3.918955612182617, "grad_norm": 0.7346400022506714, "learning_rate": 0.0001, "elbo": 3.91896, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9268707482993197}
{"step": 15810, "loss": 3.911287307739258, "grad_norm": 0.7379500865936279, "learning_rate": 0.0001, "elbo": 3.91129, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9271639690358903}
{"step": 15815, "loss": 3.8800117492675783, "grad_norm": 0.7594764828681946, "learning_rate": 0.0001, "elbo": 3.88001, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9274571897724607}
{"step": 15820, "loss": 3.8981117248535155, "grad_norm": 0.650003969669342, "learning_rate": 0.0001, "elbo": 3.89811, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9277504105090312}
{"step": 15825, "loss": 3.903668212890625, "grad_norm": 0.7277937531471252, "learning_rate": 0.0001, "elbo": 3.90367, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9280436312456017}
{"step": 15830, "loss": 3.89674072265625, "grad_norm": 0.7021604180335999, "learning_rate": 0.0001, "elbo": 3.89674, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9283368519821722}
{"step": 15830, "loss": 3.89674072265625, "grad_norm": 0.7021604180335999, "learning_rate": 0.0001, "elbo": 3.89674, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9283368519821722}
{"step": 15835, "loss": 3.8891292572021485, "grad_norm": 0.6757365465164185, "learning_rate": 0.0001, "elbo": 3.88913, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9286300727187426}
{"step": 15840, "loss": 3.888662338256836, "grad_norm": 0.6845607161521912, "learning_rate": 0.0001, "elbo": 3.88866, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9289232934553132}
{"step": 15845, "loss": 3.8833126068115233, "grad_norm": 0.6743137836456299, "learning_rate": 0.0001, "elbo": 3.88331, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9292165141918837}
{"step": 15850, "loss": 3.89830322265625, "grad_norm": 0.68475741147995, "learning_rate": 0.0001, "elbo": 3.8983, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9295097349284541}
{"step": 15855, "loss": 3.9021236419677736, "grad_norm": 0.8292171359062195, "learning_rate": 0.0001, "elbo": 3.90212, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9298029556650246}
{"step": 15860, "loss": 3.906245803833008, "grad_norm": 0.7609280347824097, "learning_rate": 0.0001, "elbo": 3.90625, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9300961764015951}
{"step": 15865, "loss": 3.8937004089355467, "grad_norm": 0.6421497464179993, "learning_rate": 0.0001, "elbo": 3.8937, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9303893971381656}
{"step": 15870, "loss": 3.8969161987304686, "grad_norm": 0.8590867519378662, "learning_rate": 0.0001, "elbo": 3.89692, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.930682617874736}
{"step": 15875, "loss": 3.920584869384766, "grad_norm": 0.6992043852806091, "learning_rate": 0.0001, "elbo": 3.92058, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9309758386113066}
{"step": 15880, "loss": 3.888108825683594, "grad_norm": 0.7097393274307251, "learning_rate": 0.0001, "elbo": 3.88811, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9312690593478771}
{"step": 15885, "loss": 3.889202117919922, "grad_norm": 0.6629558801651001, "learning_rate": 0.0001, "elbo": 3.8892, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9315622800844475}
{"step": 15890, "loss": 3.90662841796875, "grad_norm": 0.685600757598877, "learning_rate": 0.0001, "elbo": 3.90663, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9318555008210181}
{"step": 15895, "loss": 3.906574249267578, "grad_norm": 0.7836308479309082, "learning_rate": 0.0001, "elbo": 3.90657, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9321487215575885}
{"step": 15895, "loss": 3.906574249267578, "grad_norm": 0.7836308479309082, "learning_rate": 0.0001, "elbo": 3.90657, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9321487215575885}
{"step": 15900, "loss": 3.901593017578125, "grad_norm": 0.6710041761398315, "learning_rate": 0.0001, "elbo": 3.90159, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.932441942294159}
{"step": 15905, "loss": 3.8866653442382812, "grad_norm": 0.9413506388664246, "learning_rate": 0.0001, "elbo": 3.88667, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9327351630307296}
{"step": 15910, "loss": 3.9035701751708984, "grad_norm": 0.6869386434555054, "learning_rate": 0.0001, "elbo": 3.90357, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9330283837673}
{"step": 15915, "loss": 3.886566925048828, "grad_norm": 0.739858090877533, "learning_rate": 0.0001, "elbo": 3.88657, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9333216045038705}
{"step": 15920, "loss": 3.907440948486328, "grad_norm": 0.710367739200592, "learning_rate": 0.0001, "elbo": 3.90744, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.933614825240441}
{"step": 15925, "loss": 3.8778472900390626, "grad_norm": 0.6811809539794922, "learning_rate": 0.0001, "elbo": 3.87785, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9339080459770115}
{"step": 15930, "loss": 3.8941680908203127, "grad_norm": 0.711327850818634, "learning_rate": 0.0001, "elbo": 3.89417, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.934201266713582}
{"step": 15935, "loss": 3.909911346435547, "grad_norm": 0.7059581875801086, "learning_rate": 0.0001, "elbo": 3.90991, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9344944874501525}
{"step": 15940, "loss": 3.8868370056152344, "grad_norm": 0.6905975341796875, "learning_rate": 0.0001, "elbo": 3.88684, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.934787708186723}
{"step": 15945, "loss": 3.8890071868896485, "grad_norm": 0.6687213182449341, "learning_rate": 0.0001, "elbo": 3.88901, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9350809289232934}
{"step": 15950, "loss": 3.8939491271972657, "grad_norm": 0.7706610560417175, "learning_rate": 0.0001, "elbo": 3.89395, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.935374149659864}
{"step": 15955, "loss": 3.8734661102294923, "grad_norm": 0.7111058831214905, "learning_rate": 0.0001, "elbo": 3.87347, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9356673703964344}
{"step": 15955, "loss": 3.8734661102294923, "grad_norm": 0.7111058831214905, "learning_rate": 0.0001, "elbo": 3.87347, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9356673703964344}
{"step": 15960, "loss": 3.8949493408203124, "grad_norm": 0.718766987323761, "learning_rate": 0.0001, "elbo": 3.89495, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9359605911330049}
{"step": 15965, "loss": 3.909700393676758, "grad_norm": 0.8458817601203918, "learning_rate": 0.0001, "elbo": 3.9097, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9362538118695755}
{"step": 15970, "loss": 3.886463165283203, "grad_norm": 0.7635656595230103, "learning_rate": 0.0001, "elbo": 3.88646, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9365470326061459}
{"step": 15975, "loss": 3.8899772644042967, "grad_norm": 0.7476522922515869, "learning_rate": 0.0001, "elbo": 3.88998, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9368402533427164}
{"step": 15980, "loss": 3.873850631713867, "grad_norm": 0.734954297542572, "learning_rate": 0.0001, "elbo": 3.87385, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9371334740792869}
{"step": 15985, "loss": 3.8879619598388673, "grad_norm": 0.7301540374755859, "learning_rate": 0.0001, "elbo": 3.88796, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9374266948158574}
{"step": 15990, "loss": 3.869606781005859, "grad_norm": 0.6659875512123108, "learning_rate": 0.0001, "elbo": 3.86961, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9377199155524278}
{"step": 15995, "loss": 3.8923622131347657, "grad_norm": 0.7606770992279053, "learning_rate": 0.0001, "elbo": 3.89236, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9380131362889984}
{"step": 16000, "loss": 3.8883987426757813, "grad_norm": 0.6839836835861206, "learning_rate": 0.0001, "elbo": 3.8884, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9383063570255689}
{"step": 16005, "loss": 3.893121337890625, "grad_norm": 0.7065641283988953, "learning_rate": 0.0001, "elbo": 3.89312, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9385995777621393}
{"step": 16010, "loss": 3.883824920654297, "grad_norm": 0.6937219500541687, "learning_rate": 0.0001, "elbo": 3.88382, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9388927984987099}
{"step": 16015, "loss": 3.897869110107422, "grad_norm": 0.6945167183876038, "learning_rate": 0.0001, "elbo": 3.89787, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9391860192352803}
{"step": 16020, "loss": 3.9004257202148436, "grad_norm": 0.713674008846283, "learning_rate": 0.0001, "elbo": 3.90043, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9394792399718508}
{"step": 16025, "loss": 3.8916873931884766, "grad_norm": 0.7334564328193665, "learning_rate": 0.0001, "elbo": 3.89169, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9397724607084214}
{"step": 16030, "loss": 3.892719268798828, "grad_norm": 0.7780683040618896, "learning_rate": 0.0001, "elbo": 3.89272, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9400656814449918}
{"step": 16035, "loss": 3.8927574157714844, "grad_norm": 0.8128870129585266, "learning_rate": 0.0001, "elbo": 3.89276, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9403589021815623}
{"step": 16040, "loss": 3.897845458984375, "grad_norm": 0.6889303922653198, "learning_rate": 0.0001, "elbo": 3.89785, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9406521229181327}
{"step": 16045, "loss": 3.8935317993164062, "grad_norm": 0.7045227289199829, "learning_rate": 0.0001, "elbo": 3.89353, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9409453436547033}
{"step": 16050, "loss": 3.897513580322266, "grad_norm": 0.7108080983161926, "learning_rate": 0.0001, "elbo": 3.89751, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9412385643912737}
{"step": 16055, "loss": 3.8962657928466795, "grad_norm": 0.678314208984375, "learning_rate": 0.0001, "elbo": 3.89627, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9415317851278442}
{"step": 16060, "loss": 3.894816589355469, "grad_norm": 0.7509080767631531, "learning_rate": 0.0001, "elbo": 3.89482, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9418250058644148}
{"step": 16065, "loss": 3.886277770996094, "grad_norm": 0.6795443892478943, "learning_rate": 0.0001, "elbo": 3.88628, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9421182266009852}
{"step": 16070, "loss": 3.882884216308594, "grad_norm": 0.6740003228187561, "learning_rate": 0.0001, "elbo": 3.88288, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9424114473375557}
{"step": 16075, "loss": 3.8829689025878906, "grad_norm": 0.7542081475257874, "learning_rate": 0.0001, "elbo": 3.88297, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9427046680741262}
{"step": 16080, "loss": 3.889379119873047, "grad_norm": 0.6930205821990967, "learning_rate": 0.0001, "elbo": 3.88938, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9429978888106967}
{"step": 16085, "loss": 3.881819152832031, "grad_norm": 0.7020488381385803, "learning_rate": 0.0001, "elbo": 3.88182, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9432911095472671}
{"step": 16090, "loss": 3.8761383056640626, "grad_norm": 0.7818021178245544, "learning_rate": 0.0001, "elbo": 3.87614, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9435843302838377}
{"step": 16095, "loss": 3.891239547729492, "grad_norm": 0.7017569541931152, "learning_rate": 0.0001, "elbo": 3.89124, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9438775510204082}
{"step": 16100, "loss": 3.9051578521728514, "grad_norm": 0.7443188428878784, "learning_rate": 0.0001, "elbo": 3.90516, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9441707717569786}
{"step": 16105, "loss": 3.874803161621094, "grad_norm": 1.1955891847610474, "learning_rate": 0.0001, "elbo": 3.8748, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9444639924935492}
{"step": 16110, "loss": 3.896209716796875, "grad_norm": 0.7454782724380493, "learning_rate": 0.0001, "elbo": 3.89621, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9447572132301196}
{"step": 16115, "loss": 3.905345153808594, "grad_norm": 0.7126341462135315, "learning_rate": 0.0001, "elbo": 3.90535, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9450504339666901}
{"step": 16120, "loss": 3.9019561767578126, "grad_norm": 0.6673683524131775, "learning_rate": 0.0001, "elbo": 3.90196, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9453436547032606}
{"step": 16125, "loss": 3.886187744140625, "grad_norm": 0.7479344606399536, "learning_rate": 0.0001, "elbo": 3.88619, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9456368754398311}
{"step": 16130, "loss": 3.8927879333496094, "grad_norm": 0.7771457433700562, "learning_rate": 0.0001, "elbo": 3.89279, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9459300961764016}
{"step": 16135, "loss": 3.8894527435302733, "grad_norm": 0.7200183868408203, "learning_rate": 0.0001, "elbo": 3.88945, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9462233169129721}
{"step": 16140, "loss": 3.896520233154297, "grad_norm": 0.7249922752380371, "learning_rate": 0.0001, "elbo": 3.89652, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9465165376495426}
{"step": 16145, "loss": 3.8845989227294924, "grad_norm": 0.7004166841506958, "learning_rate": 0.0001, "elbo": 3.8846, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.946809758386113}
{"step": 16150, "loss": 3.868189239501953, "grad_norm": 0.7127329707145691, "learning_rate": 0.0001, "elbo": 3.86819, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9471029791226836}
{"step": 16150, "loss": 3.868189239501953, "grad_norm": 0.7127329707145691, "learning_rate": 0.0001, "elbo": 3.86819, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9471029791226836}
{"step": 16155, "loss": 3.8924327850341798, "grad_norm": 0.7205867767333984, "learning_rate": 0.0001, "elbo": 3.89243, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.947396199859254}
{"step": 16160, "loss": 3.8798667907714846, "grad_norm": 0.712585985660553, "learning_rate": 0.0001, "elbo": 3.87987, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9476894205958245}
{"step": 16165, "loss": 3.8819976806640626, "grad_norm": 0.7851698994636536, "learning_rate": 0.0001, "elbo": 3.882, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9479826413323951}
{"step": 16170, "loss": 3.8871864318847655, "grad_norm": 0.7137745022773743, "learning_rate": 0.0001, "elbo": 3.88719, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9482758620689655}
{"step": 16175, "loss": 3.8777976989746095, "grad_norm": 0.6962295770645142, "learning_rate": 0.0001, "elbo": 3.8778, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.948569082805536}
{"step": 16180, "loss": 3.9000221252441407, "grad_norm": 0.7492682337760925, "learning_rate": 0.0001, "elbo": 3.90002, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9488623035421065}
{"step": 16185, "loss": 3.8924537658691407, "grad_norm": 0.6803422570228577, "learning_rate": 0.0001, "elbo": 3.89245, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.949155524278677}
{"step": 16190, "loss": 3.8689292907714843, "grad_norm": 0.7251715660095215, "learning_rate": 0.0001, "elbo": 3.86893, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9494487450152475}
{"step": 16195, "loss": 3.8827606201171876, "grad_norm": 0.7289137244224548, "learning_rate": 0.0001, "elbo": 3.88276, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.949741965751818}
{"step": 16195, "loss": 3.8827606201171876, "grad_norm": 0.7289137244224548, "learning_rate": 0.0001, "elbo": 3.88276, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.949741965751818}
{"step": 16200, "loss": 3.886322784423828, "grad_norm": 0.6929587721824646, "learning_rate": 0.0001, "elbo": 3.88632, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9500351864883885}
{"step": 16205, "loss": 3.901589202880859, "grad_norm": 0.6610324382781982, "learning_rate": 0.0001, "elbo": 3.90159, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9503284072249589}
{"step": 16210, "loss": 3.8646469116210938, "grad_norm": 0.6864447593688965, "learning_rate": 0.0001, "elbo": 3.86465, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9506216279615295}
{"step": 16215, "loss": 3.893016815185547, "grad_norm": 0.6817606091499329, "learning_rate": 0.0001, "elbo": 3.89302, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9509148486981}
{"step": 16220, "loss": 3.890796661376953, "grad_norm": 0.7253110408782959, "learning_rate": 0.0001, "elbo": 3.8908, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9512080694346704}
{"step": 16225, "loss": 3.8855575561523437, "grad_norm": 1.0725609064102173, "learning_rate": 0.0001, "elbo": 3.88556, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.951501290171241}
{"step": 16230, "loss": 3.889665222167969, "grad_norm": 0.6810306310653687, "learning_rate": 0.0001, "elbo": 3.88967, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9517945109078114}
{"step": 16235, "loss": 3.8851158142089846, "grad_norm": 0.9343734979629517, "learning_rate": 0.0001, "elbo": 3.88512, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9520877316443819}
{"step": 16240, "loss": 3.868952178955078, "grad_norm": 0.7115954160690308, "learning_rate": 0.0001, "elbo": 3.86895, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9523809523809523}
{"step": 16245, "loss": 3.8809974670410154, "grad_norm": 0.7334715723991394, "learning_rate": 0.0001, "elbo": 3.881, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9526741731175229}
{"step": 16250, "loss": 3.885911560058594, "grad_norm": 0.7284114360809326, "learning_rate": 0.0001, "elbo": 3.88591, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9529673938540933}
{"step": 16255, "loss": 3.882571792602539, "grad_norm": 0.6690505743026733, "learning_rate": 0.0001, "elbo": 3.88257, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9532606145906638}
{"step": 16260, "loss": 3.886113739013672, "grad_norm": 0.7395874261856079, "learning_rate": 0.0001, "elbo": 3.88611, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9535538353272344}
{"step": 16265, "loss": 3.8968887329101562, "grad_norm": 0.819202721118927, "learning_rate": 0.0001, "elbo": 3.89689, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9538470560638048}
{"step": 16270, "loss": 3.8896854400634764, "grad_norm": 0.756309986114502, "learning_rate": 0.0001, "elbo": 3.88969, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9541402768003753}
{"step": 16275, "loss": 3.8858909606933594, "grad_norm": 1.0180482864379883, "learning_rate": 0.0001, "elbo": 3.88589, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9544334975369458}
{"step": 16280, "loss": 3.8754302978515627, "grad_norm": 0.6870152354240417, "learning_rate": 0.0001, "elbo": 3.87543, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9547267182735163}
{"step": 16285, "loss": 3.8906295776367186, "grad_norm": 0.6945451498031616, "learning_rate": 0.0001, "elbo": 3.89063, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9550199390100867}
{"step": 16290, "loss": 3.89876708984375, "grad_norm": 0.6648361682891846, "learning_rate": 0.0001, "elbo": 3.89877, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9553131597466573}
{"step": 16295, "loss": 3.893057632446289, "grad_norm": 0.7288633584976196, "learning_rate": 0.0001, "elbo": 3.89306, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9556063804832278}
{"step": 16300, "loss": 3.8746269226074217, "grad_norm": 0.6733601093292236, "learning_rate": 0.0001, "elbo": 3.87463, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9558996012197982}
{"step": 16305, "loss": 3.873614501953125, "grad_norm": 0.785114586353302, "learning_rate": 0.0001, "elbo": 3.87361, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9561928219563688}
{"step": 16310, "loss": 3.8890151977539062, "grad_norm": 0.6844495534896851, "learning_rate": 0.0001, "elbo": 3.88902, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9564860426929392}
{"step": 16315, "loss": 3.881807327270508, "grad_norm": 0.7106145024299622, "learning_rate": 0.0001, "elbo": 3.88181, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9567792634295097}
{"step": 16320, "loss": 3.884420394897461, "grad_norm": 0.7070325016975403, "learning_rate": 0.0001, "elbo": 3.88442, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9570724841660803}
{"step": 16325, "loss": 3.890613555908203, "grad_norm": 0.8704900145530701, "learning_rate": 0.0001, "elbo": 3.89061, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9573657049026507}
{"step": 16330, "loss": 3.884496307373047, "grad_norm": 0.8240184783935547, "learning_rate": 0.0001, "elbo": 3.8845, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9576589256392212}
{"step": 16335, "loss": 3.88221435546875, "grad_norm": 0.7112240791320801, "learning_rate": 0.0001, "elbo": 3.88221, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9579521463757917}
{"step": 16340, "loss": 3.889996337890625, "grad_norm": 0.779047966003418, "learning_rate": 0.0001, "elbo": 3.89, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9582453671123622}
{"step": 16345, "loss": 3.8886829376220704, "grad_norm": 0.6739089488983154, "learning_rate": 0.0001, "elbo": 3.88868, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9585385878489326}
{"step": 16350, "loss": 3.8819618225097656, "grad_norm": 0.7179830074310303, "learning_rate": 0.0001, "elbo": 3.88196, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9588318085855032}
{"step": 16355, "loss": 3.8859153747558595, "grad_norm": 0.738000214099884, "learning_rate": 0.0001, "elbo": 3.88592, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9591250293220737}
{"step": 16360, "loss": 3.8831756591796873, "grad_norm": 1.0227996110916138, "learning_rate": 0.0001, "elbo": 3.88318, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9594182500586441}
{"step": 16365, "loss": 3.881216049194336, "grad_norm": 0.6872427463531494, "learning_rate": 0.0001, "elbo": 3.88122, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9597114707952147}
{"step": 16370, "loss": 3.8802410125732423, "grad_norm": 0.6929242014884949, "learning_rate": 0.0001, "elbo": 3.88024, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9600046915317851}
{"step": 16375, "loss": 3.871976852416992, "grad_norm": 0.7140938639640808, "learning_rate": 0.0001, "elbo": 3.87198, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9602979122683556}
{"step": 16380, "loss": 3.888325500488281, "grad_norm": 0.7028904557228088, "learning_rate": 0.0001, "elbo": 3.88833, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9605911330049262}
{"step": 16385, "loss": 3.8800724029541014, "grad_norm": 0.7062919735908508, "learning_rate": 0.0001, "elbo": 3.88007, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9608843537414966}
{"step": 16390, "loss": 3.8910728454589845, "grad_norm": 0.6938381195068359, "learning_rate": 0.0001, "elbo": 3.89107, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9611775744780671}
{"step": 16395, "loss": 3.8797962188720705, "grad_norm": 0.7508905529975891, "learning_rate": 0.0001, "elbo": 3.8798, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9614707952146376}
{"step": 16400, "loss": 3.884748840332031, "grad_norm": 0.7053667306900024, "learning_rate": 0.0001, "elbo": 3.88475, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9617640159512081}
{"step": 16405, "loss": 3.8815944671630858, "grad_norm": 0.7137973308563232, "learning_rate": 0.0001, "elbo": 3.88159, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9620572366877785}
{"step": 16410, "loss": 3.879841613769531, "grad_norm": 0.6968556642532349, "learning_rate": 0.0001, "elbo": 3.87984, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9623504574243491}
{"step": 16415, "loss": 3.8920440673828125, "grad_norm": 0.7915109992027283, "learning_rate": 0.0001, "elbo": 3.89204, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9626436781609196}
{"step": 16420, "loss": 3.886058044433594, "grad_norm": 0.6635036468505859, "learning_rate": 0.0001, "elbo": 3.88606, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.96293689889749}
{"step": 16425, "loss": 3.882461166381836, "grad_norm": 0.7041868567466736, "learning_rate": 0.0001, "elbo": 3.88246, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9632301196340605}
{"step": 16430, "loss": 3.8933963775634766, "grad_norm": 0.9282665848731995, "learning_rate": 0.0001, "elbo": 3.8934, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.963523340370631}
{"step": 16435, "loss": 3.875677490234375, "grad_norm": 0.7118464708328247, "learning_rate": 0.0001, "elbo": 3.87568, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9638165611072015}
{"step": 16440, "loss": 3.8743057250976562, "grad_norm": 0.7311320304870605, "learning_rate": 0.0001, "elbo": 3.87431, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9641097818437719}
{"step": 16445, "loss": 3.880172348022461, "grad_norm": 0.8189799785614014, "learning_rate": 0.0001, "elbo": 3.88017, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9644030025803425}
{"step": 16450, "loss": 3.888154983520508, "grad_norm": 0.7732260823249817, "learning_rate": 0.0001, "elbo": 3.88815, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.964696223316913}
{"step": 16455, "loss": 3.8770431518554687, "grad_norm": 0.6668978929519653, "learning_rate": 0.0001, "elbo": 3.87704, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9649894440534834}
{"step": 16460, "loss": 3.8878105163574217, "grad_norm": 0.7368302941322327, "learning_rate": 0.0001, "elbo": 3.88781, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.965282664790054}
{"step": 16465, "loss": 3.8943252563476562, "grad_norm": 0.683544397354126, "learning_rate": 0.0001, "elbo": 3.89433, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9655758855266244}
{"step": 16465, "loss": 3.8943252563476562, "grad_norm": 0.683544397354126, "learning_rate": 0.0001, "elbo": 3.89433, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9655758855266244}
{"step": 16470, "loss": 3.887370300292969, "grad_norm": 0.6840529441833496, "learning_rate": 0.0001, "elbo": 3.88737, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9658691062631949}
{"step": 16475, "loss": 3.879043960571289, "grad_norm": 0.679566502571106, "learning_rate": 0.0001, "elbo": 3.87904, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9661623269997655}
{"step": 16480, "loss": 3.874920654296875, "grad_norm": 0.7266801595687866, "learning_rate": 0.0001, "elbo": 3.87492, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9664555477363359}
{"step": 16485, "loss": 3.880710983276367, "grad_norm": 0.6978594660758972, "learning_rate": 0.0001, "elbo": 3.88071, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9667487684729064}
{"step": 16490, "loss": 3.8749534606933596, "grad_norm": 0.6782578825950623, "learning_rate": 0.0001, "elbo": 3.87495, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9670419892094769}
{"step": 16490, "loss": 3.8749534606933596, "grad_norm": 0.6782578825950623, "learning_rate": 0.0001, "elbo": 3.87495, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9670419892094769}
{"step": 16495, "loss": 3.8891944885253906, "grad_norm": 0.7595153450965881, "learning_rate": 0.0001, "elbo": 3.88919, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9673352099460474}
{"step": 16500, "loss": 3.8836090087890627, "grad_norm": 0.767765998840332, "learning_rate": 0.0001, "elbo": 3.88361, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9676284306826178}
{"step": 16505, "loss": 3.8859806060791016, "grad_norm": 0.666750431060791, "learning_rate": 0.0001, "elbo": 3.88598, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9679216514191884}
{"step": 16510, "loss": 3.889856719970703, "grad_norm": 0.6648244857788086, "learning_rate": 0.0001, "elbo": 3.88986, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9682148721557589}
{"step": 16515, "loss": 3.8829067230224608, "grad_norm": 0.7915830612182617, "learning_rate": 0.0001, "elbo": 3.88291, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9685080928923293}
{"step": 16520, "loss": 3.871294403076172, "grad_norm": 0.6335331201553345, "learning_rate": 0.0001, "elbo": 3.87129, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9688013136288999}
{"step": 16525, "loss": 3.8874488830566407, "grad_norm": 0.6837727427482605, "learning_rate": 0.0001, "elbo": 3.88745, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9690945343654703}
{"step": 16530, "loss": 3.8947887420654297, "grad_norm": 0.9099586009979248, "learning_rate": 0.0001, "elbo": 3.89479, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9693877551020408}
{"step": 16535, "loss": 3.89929084777832, "grad_norm": 0.7282750606536865, "learning_rate": 0.0001, "elbo": 3.89929, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9696809758386113}
{"step": 16540, "loss": 3.8823932647705077, "grad_norm": 0.7189205884933472, "learning_rate": 0.0001, "elbo": 3.88239, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9699741965751818}
{"step": 16545, "loss": 3.919757080078125, "grad_norm": 0.7308251857757568, "learning_rate": 0.0001, "elbo": 3.91976, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9702674173117523}
{"step": 16550, "loss": 3.8841041564941405, "grad_norm": 0.7054210305213928, "learning_rate": 0.0001, "elbo": 3.8841, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9705606380483228}
{"step": 16555, "loss": 3.9020042419433594, "grad_norm": 1.0075502395629883, "learning_rate": 0.0001, "elbo": 3.902, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9708538587848933}
{"step": 16560, "loss": 3.8950599670410155, "grad_norm": 0.654367983341217, "learning_rate": 0.0001, "elbo": 3.89506, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9711470795214637}
{"step": 16565, "loss": 3.880986785888672, "grad_norm": 0.7330950498580933, "learning_rate": 0.0001, "elbo": 3.88099, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9714403002580343}
{"step": 16570, "loss": 3.89400634765625, "grad_norm": 0.6850776076316833, "learning_rate": 0.0001, "elbo": 3.89401, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9717335209946047}
{"step": 16575, "loss": 3.88880615234375, "grad_norm": 0.702141284942627, "learning_rate": 0.0001, "elbo": 3.88881, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9720267417311752}
{"step": 16580, "loss": 3.896891784667969, "grad_norm": 0.7803791761398315, "learning_rate": 0.0001, "elbo": 3.89689, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9723199624677458}
{"step": 16580, "loss": 3.896891784667969, "grad_norm": 0.7803791761398315, "learning_rate": 0.0001, "elbo": 3.89689, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9723199624677458}
{"step": 16585, "loss": 3.8886768341064455, "grad_norm": 0.9228174686431885, "learning_rate": 0.0001, "elbo": 3.88868, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9726131832043162}
{"step": 16590, "loss": 3.894117736816406, "grad_norm": 1.0928887128829956, "learning_rate": 0.0001, "elbo": 3.89412, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9729064039408867}
{"step": 16595, "loss": 3.8987022399902345, "grad_norm": 0.6899202466011047, "learning_rate": 0.0001, "elbo": 3.8987, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9731996246774572}
{"step": 16600, "loss": 3.884015655517578, "grad_norm": 1.015430212020874, "learning_rate": 0.0001, "elbo": 3.88402, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9734928454140277}
{"step": 16605, "loss": 3.8884521484375, "grad_norm": 0.6842033863067627, "learning_rate": 0.0001, "elbo": 3.88845, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9737860661505982}
{"step": 16610, "loss": 3.8895217895507814, "grad_norm": 0.7187766432762146, "learning_rate": 0.0001, "elbo": 3.88952, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9740792868871687}
{"step": 16615, "loss": 3.8777545928955077, "grad_norm": 0.7370337247848511, "learning_rate": 0.0001, "elbo": 3.87775, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9743725076237392}
{"step": 16620, "loss": 3.8976470947265627, "grad_norm": 0.7586446404457092, "learning_rate": 0.0001, "elbo": 3.89765, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9746657283603096}
{"step": 16625, "loss": 3.866602325439453, "grad_norm": 0.7857327461242676, "learning_rate": 0.0001, "elbo": 3.8666, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9749589490968801}
{"step": 16630, "loss": 3.8752212524414062, "grad_norm": 0.7946004867553711, "learning_rate": 0.0001, "elbo": 3.87522, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9752521698334506}
{"step": 16635, "loss": 3.880982208251953, "grad_norm": 0.7439355254173279, "learning_rate": 0.0001, "elbo": 3.88098, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9755453905700211}
{"step": 16640, "loss": 3.8830982208251954, "grad_norm": 0.7260632514953613, "learning_rate": 0.0001, "elbo": 3.8831, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9758386113065916}
{"step": 16645, "loss": 3.861891174316406, "grad_norm": 0.6877557635307312, "learning_rate": 0.0001, "elbo": 3.86189, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9761318320431621}
{"step": 16650, "loss": 3.8568267822265625, "grad_norm": 0.8802042007446289, "learning_rate": 0.0001, "elbo": 3.85683, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9764250527797326}
{"step": 16655, "loss": 3.8759490966796877, "grad_norm": 0.6791860461235046, "learning_rate": 0.0001, "elbo": 3.87595, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.976718273516303}
{"step": 16660, "loss": 3.8766048431396483, "grad_norm": 0.6908900737762451, "learning_rate": 0.0001, "elbo": 3.8766, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9770114942528736}
{"step": 16665, "loss": 3.889849090576172, "grad_norm": 0.6798255443572998, "learning_rate": 0.0001, "elbo": 3.88985, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.977304714989444}
{"step": 16670, "loss": 3.8682430267333983, "grad_norm": 0.9029837846755981, "learning_rate": 0.0001, "elbo": 3.86824, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9775979357260145}
{"step": 16675, "loss": 3.8841922760009764, "grad_norm": 0.8018358945846558, "learning_rate": 0.0001, "elbo": 3.88419, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9778911564625851}
{"step": 16680, "loss": 3.8853092193603516, "grad_norm": 0.7072798609733582, "learning_rate": 0.0001, "elbo": 3.88531, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9781843771991555}
{"step": 16685, "loss": 3.8617740631103517, "grad_norm": 0.6726468205451965, "learning_rate": 0.0001, "elbo": 3.86177, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.978477597935726}
{"step": 16690, "loss": 3.864955520629883, "grad_norm": 0.7230769395828247, "learning_rate": 0.0001, "elbo": 3.86496, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9787708186722965}
{"step": 16695, "loss": 3.8726264953613283, "grad_norm": 0.7951024174690247, "learning_rate": 0.0001, "elbo": 3.87263, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.979064039408867}
{"step": 16700, "loss": 3.871335983276367, "grad_norm": 0.7327281832695007, "learning_rate": 0.0001, "elbo": 3.87134, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9793572601454374}
{"step": 16705, "loss": 3.8873252868652344, "grad_norm": 0.6938645839691162, "learning_rate": 0.0001, "elbo": 3.88733, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.979650480882008}
{"step": 16710, "loss": 3.8760845184326174, "grad_norm": 0.8334684371948242, "learning_rate": 0.0001, "elbo": 3.87608, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9799437016185785}
{"step": 16715, "loss": 3.892121124267578, "grad_norm": 0.6996515989303589, "learning_rate": 0.0001, "elbo": 3.89212, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9802369223551489}
{"step": 16720, "loss": 3.8583702087402343, "grad_norm": 0.7086140513420105, "learning_rate": 0.0001, "elbo": 3.85837, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9805301430917195}
{"step": 16725, "loss": 3.8713645935058594, "grad_norm": 0.6927879452705383, "learning_rate": 0.0001, "elbo": 3.87136, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9808233638282899}
{"step": 16730, "loss": 3.8739501953125, "grad_norm": 0.7265781760215759, "learning_rate": 0.0001, "elbo": 3.87395, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9811165845648604}
{"step": 16735, "loss": 3.8684764862060548, "grad_norm": 0.8301185369491577, "learning_rate": 0.0001, "elbo": 3.86848, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.981409805301431}
{"step": 16740, "loss": 3.880695343017578, "grad_norm": 1.2942700386047363, "learning_rate": 0.0001, "elbo": 3.8807, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9817030260380014}
{"step": 16745, "loss": 3.885763931274414, "grad_norm": 0.7151215672492981, "learning_rate": 0.0001, "elbo": 3.88576, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9819962467745719}
{"step": 16750, "loss": 3.8726360321044924, "grad_norm": 0.7518812417984009, "learning_rate": 0.0001, "elbo": 3.87264, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9822894675111424}
{"step": 16755, "loss": 3.886189270019531, "grad_norm": 0.7475162744522095, "learning_rate": 0.0001, "elbo": 3.88619, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9825826882477129}
{"step": 16760, "loss": 3.8661102294921874, "grad_norm": 0.6881226301193237, "learning_rate": 0.0001, "elbo": 3.86611, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9828759089842833}
{"step": 16765, "loss": 3.8811416625976562, "grad_norm": 0.7233870625495911, "learning_rate": 0.0001, "elbo": 3.88114, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9831691297208539}
{"step": 16770, "loss": 3.879352569580078, "grad_norm": 0.6763051748275757, "learning_rate": 0.0001, "elbo": 3.87935, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9834623504574244}
{"step": 16775, "loss": 3.882190704345703, "grad_norm": 0.7776250839233398, "learning_rate": 0.0001, "elbo": 3.88219, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9837555711939948}
{"step": 16780, "loss": 3.888467788696289, "grad_norm": 0.7966330051422119, "learning_rate": 0.0001, "elbo": 3.88847, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9840487919305654}
{"step": 16785, "loss": 3.887799835205078, "grad_norm": 0.7260128855705261, "learning_rate": 0.0001, "elbo": 3.8878, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9843420126671358}
{"step": 16790, "loss": 3.863104248046875, "grad_norm": 0.828284502029419, "learning_rate": 0.0001, "elbo": 3.8631, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9846352334037063}
{"step": 16795, "loss": 3.886128616333008, "grad_norm": 0.8764094710350037, "learning_rate": 0.0001, "elbo": 3.88613, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9849284541402769}
{"step": 16800, "loss": 3.8811534881591796, "grad_norm": 0.6601842641830444, "learning_rate": 0.0001, "elbo": 3.88115, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9852216748768473}
{"step": 16805, "loss": 3.870465087890625, "grad_norm": 1.1623280048370361, "learning_rate": 0.0001, "elbo": 3.87047, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9855148956134178}
{"step": 16810, "loss": 3.880748748779297, "grad_norm": 0.7233186960220337, "learning_rate": 0.0001, "elbo": 3.88075, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9858081163499882}
{"step": 16810, "loss": 3.880748748779297, "grad_norm": 0.7233186960220337, "learning_rate": 0.0001, "elbo": 3.88075, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9858081163499882}
{"step": 16815, "loss": 3.873790740966797, "grad_norm": 1.0760828256607056, "learning_rate": 0.0001, "elbo": 3.87379, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9861013370865588}
{"step": 16820, "loss": 3.890186309814453, "grad_norm": 0.6564065217971802, "learning_rate": 0.0001, "elbo": 3.89019, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9863945578231292}
{"step": 16825, "loss": 3.8696010589599608, "grad_norm": 0.7181804180145264, "learning_rate": 0.0001, "elbo": 3.8696, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9866877785596997}
{"step": 16830, "loss": 3.867911529541016, "grad_norm": 0.7166882157325745, "learning_rate": 0.0001, "elbo": 3.86791, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9869809992962703}
{"step": 16835, "loss": 3.8863311767578126, "grad_norm": 0.7957325577735901, "learning_rate": 0.0001, "elbo": 3.88633, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9872742200328407}
{"step": 16840, "loss": 3.861053466796875, "grad_norm": 0.6951032280921936, "learning_rate": 0.0001, "elbo": 3.86105, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9875674407694112}
{"step": 16845, "loss": 3.8745704650878907, "grad_norm": 0.6833360195159912, "learning_rate": 0.0001, "elbo": 3.87457, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9878606615059817}
{"step": 16850, "loss": 3.8739524841308595, "grad_norm": 0.77222740650177, "learning_rate": 0.0001, "elbo": 3.87395, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9881538822425522}
{"step": 16855, "loss": 3.865995407104492, "grad_norm": 0.7576783895492554, "learning_rate": 0.0001, "elbo": 3.866, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9884471029791226}
{"step": 16855, "loss": 3.865995407104492, "grad_norm": 0.7576783895492554, "learning_rate": 0.0001, "elbo": 3.866, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9884471029791226}
{"step": 16860, "loss": 3.873464584350586, "grad_norm": 0.6664952635765076, "learning_rate": 0.0001, "elbo": 3.87346, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9887403237156932}
{"step": 16865, "loss": 3.8953285217285156, "grad_norm": 0.6912926435470581, "learning_rate": 0.0001, "elbo": 3.89533, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9890335444522637}
{"step": 16870, "loss": 3.878491973876953, "grad_norm": 0.7320753335952759, "learning_rate": 0.0001, "elbo": 3.87849, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9893267651888341}
{"step": 16875, "loss": 3.851564407348633, "grad_norm": 0.6782935857772827, "learning_rate": 0.0001, "elbo": 3.85156, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9896199859254047}
{"step": 16880, "loss": 3.879402923583984, "grad_norm": 1.077169418334961, "learning_rate": 0.0001, "elbo": 3.8794, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9899132066619751}
{"step": 16885, "loss": 3.858129119873047, "grad_norm": 0.6565809845924377, "learning_rate": 0.0001, "elbo": 3.85813, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9902064273985456}
{"step": 16890, "loss": 3.866010284423828, "grad_norm": 0.9335492849349976, "learning_rate": 0.0001, "elbo": 3.86601, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9904996481351162}
{"step": 16895, "loss": 3.875202178955078, "grad_norm": 0.7026296257972717, "learning_rate": 0.0001, "elbo": 3.8752, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9907928688716866}
{"step": 16900, "loss": 3.8833148956298826, "grad_norm": 0.7538704872131348, "learning_rate": 0.0001, "elbo": 3.88331, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9910860896082571}
{"step": 16905, "loss": 3.8724639892578123, "grad_norm": 0.7087455987930298, "learning_rate": 0.0001, "elbo": 3.87246, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9913793103448276}
{"step": 16910, "loss": 3.867003631591797, "grad_norm": 0.9164911508560181, "learning_rate": 0.0001, "elbo": 3.867, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9916725310813981}
{"step": 16915, "loss": 3.864344024658203, "grad_norm": 0.706290066242218, "learning_rate": 0.0001, "elbo": 3.86434, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9919657518179685}
{"step": 16920, "loss": 3.8688697814941406, "grad_norm": 0.7157207131385803, "learning_rate": 0.0001, "elbo": 3.86887, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9922589725545391}
{"step": 16925, "loss": 3.8885326385498047, "grad_norm": 0.9373825788497925, "learning_rate": 0.0001, "elbo": 3.88853, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9925521932911096}
{"step": 16930, "loss": 3.8866226196289064, "grad_norm": 0.7423227429389954, "learning_rate": 0.0001, "elbo": 3.88662, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.99284541402768}
{"step": 16935, "loss": 3.884494400024414, "grad_norm": 0.6811620593070984, "learning_rate": 0.0001, "elbo": 3.88449, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9931386347642506}
{"step": 16940, "loss": 3.8868972778320314, "grad_norm": 0.8022823333740234, "learning_rate": 0.0001, "elbo": 3.8869, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.993431855500821}
{"step": 16945, "loss": 3.8731151580810548, "grad_norm": 0.7877117991447449, "learning_rate": 0.0001, "elbo": 3.87312, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9937250762373915}
{"step": 16950, "loss": 3.878226089477539, "grad_norm": 0.7572935819625854, "learning_rate": 0.0001, "elbo": 3.87823, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.994018296973962}
{"step": 16955, "loss": 3.8755977630615233, "grad_norm": 0.7096073031425476, "learning_rate": 0.0001, "elbo": 3.8756, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9943115177105325}
{"step": 16960, "loss": 3.8650463104248045, "grad_norm": 0.8287988305091858, "learning_rate": 0.0001, "elbo": 3.86505, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.994604738447103}
{"step": 16965, "loss": 3.867938232421875, "grad_norm": 0.7443661689758301, "learning_rate": 0.0001, "elbo": 3.86794, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9948979591836735}
{"step": 16970, "loss": 3.8674957275390627, "grad_norm": 0.695080578327179, "learning_rate": 0.0001, "elbo": 3.8675, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.995191179920244}
{"step": 16975, "loss": 3.8751861572265627, "grad_norm": 0.7597139477729797, "learning_rate": 0.0001, "elbo": 3.87519, "mask_frac": 0.4995, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9954844006568144}
{"step": 16980, "loss": 3.856332778930664, "grad_norm": 0.7400152087211609, "learning_rate": 0.0001, "elbo": 3.85633, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.995777621393385}
{"step": 16985, "loss": 3.8677642822265623, "grad_norm": 0.663264811038971, "learning_rate": 0.0001, "elbo": 3.86776, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9960708421299554}
{"step": 16990, "loss": 3.860371398925781, "grad_norm": 1.0829037427902222, "learning_rate": 0.0001, "elbo": 3.86037, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9963640628665259}
{"step": 16995, "loss": 3.8812210083007814, "grad_norm": 0.9287906885147095, "learning_rate": 0.0001, "elbo": 3.88122, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9966572836030964}
{"step": 17000, "loss": 3.855055236816406, "grad_norm": 0.7766493558883667, "learning_rate": 0.0001, "elbo": 3.85506, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9969505043396669}
{"step": 17005, "loss": 3.852498245239258, "grad_norm": 0.8348205089569092, "learning_rate": 0.0001, "elbo": 3.8525, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9972437250762374}
{"step": 17010, "loss": 3.8761852264404295, "grad_norm": 0.6794862747192383, "learning_rate": 0.0001, "elbo": 3.87619, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9975369458128078}
{"step": 17015, "loss": 3.869489288330078, "grad_norm": 0.6685135364532471, "learning_rate": 0.0001, "elbo": 3.86949, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9978301665493784}
{"step": 17020, "loss": 3.8917510986328123, "grad_norm": 0.6721555590629578, "learning_rate": 0.0001, "elbo": 3.89175, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9981233872859488}
{"step": 17025, "loss": 3.869812774658203, "grad_norm": 0.8909497261047363, "learning_rate": 0.0001, "elbo": 3.86981, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9984166080225193}
{"step": 17025, "loss": 3.869812774658203, "grad_norm": 0.8909497261047363, "learning_rate": 0.0001, "elbo": 3.86981, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9984166080225193}
{"step": 17030, "loss": 3.8675731658935546, "grad_norm": 0.6914973855018616, "learning_rate": 0.0001, "elbo": 3.86757, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9987098287590899}
{"step": 17035, "loss": 3.875799560546875, "grad_norm": 0.7599710822105408, "learning_rate": 0.0001, "elbo": 3.8758, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9990030494956603}
{"step": 17040, "loss": 3.881475830078125, "grad_norm": 0.6701194643974304, "learning_rate": 0.0001, "elbo": 3.88148, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9992962702322308}
{"step": 17045, "loss": 3.88155517578125, "grad_norm": 0.7319712042808533, "learning_rate": 0.0001, "elbo": 3.88156, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9995894909688013}
{"step": 17050, "loss": 3.8749645233154295, "grad_norm": 0.867757260799408, "learning_rate": 0.0001, "elbo": 3.87496, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 0.9998827117053718}
{"step": 17055, "loss": 3.858633804321289, "grad_norm": 0.7125595211982727, "learning_rate": 0.0001, "elbo": 3.85863, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0001759324419424}
{"step": 17060, "loss": 3.880869674682617, "grad_norm": 0.7153647541999817, "learning_rate": 0.0001, "elbo": 3.88087, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0004691531785128}
{"step": 17065, "loss": 3.8684268951416017, "grad_norm": 0.66734379529953, "learning_rate": 0.0001, "elbo": 3.86843, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0007623739150833}
{"step": 17070, "loss": 3.8753700256347656, "grad_norm": 0.8318424820899963, "learning_rate": 0.0001, "elbo": 3.87537, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0010555946516537}
{"step": 17075, "loss": 3.865727996826172, "grad_norm": 0.775486171245575, "learning_rate": 0.0001, "elbo": 3.86573, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0013488153882242}
{"step": 17080, "loss": 3.8612045288085937, "grad_norm": 0.6748167276382446, "learning_rate": 0.0001, "elbo": 3.8612, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0016420361247949}
{"step": 17085, "loss": 3.8751899719238283, "grad_norm": 0.6861332654953003, "learning_rate": 0.0001, "elbo": 3.87519, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0019352568613653}
{"step": 17090, "loss": 3.8683380126953124, "grad_norm": 0.6831969022750854, "learning_rate": 0.0001, "elbo": 3.86834, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0022284775979358}
{"step": 17095, "loss": 3.874189758300781, "grad_norm": 0.788720965385437, "learning_rate": 0.0001, "elbo": 3.87419, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0025216983345062}
{"step": 17100, "loss": 3.871307373046875, "grad_norm": 0.7076668739318848, "learning_rate": 0.0001, "elbo": 3.87131, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0028149190710767}
{"step": 17105, "loss": 3.873462677001953, "grad_norm": 0.7843385934829712, "learning_rate": 0.0001, "elbo": 3.87346, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0031081398076471}
{"step": 17110, "loss": 3.883470153808594, "grad_norm": 0.8101687431335449, "learning_rate": 0.0001, "elbo": 3.88347, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0034013605442176}
{"step": 17115, "loss": 3.862591552734375, "grad_norm": 0.6912382245063782, "learning_rate": 0.0001, "elbo": 3.86259, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0036945812807883}
{"step": 17120, "loss": 3.8576801300048826, "grad_norm": 0.7162878513336182, "learning_rate": 0.0001, "elbo": 3.85768, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0039878020173587}
{"step": 17125, "loss": 3.879508209228516, "grad_norm": 0.7279406189918518, "learning_rate": 0.0001, "elbo": 3.87951, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0042810227539292}
{"step": 17130, "loss": 3.884316253662109, "grad_norm": 0.6712517142295837, "learning_rate": 0.0001, "elbo": 3.88432, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0045742434904996}
{"step": 17135, "loss": 3.86707763671875, "grad_norm": 0.6815882325172424, "learning_rate": 0.0001, "elbo": 3.86708, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.00486746422707}
{"step": 17140, "loss": 3.8718074798583983, "grad_norm": 0.6999824047088623, "learning_rate": 0.0001, "elbo": 3.87181, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0051606849636405}
{"step": 17140, "loss": 3.8718074798583983, "grad_norm": 0.6999824047088623, "learning_rate": 0.0001, "elbo": 3.87181, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0051606849636405}
{"step": 17145, "loss": 3.8723297119140625, "grad_norm": 0.7075410485267639, "learning_rate": 0.0001, "elbo": 3.87233, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0054539057002112}
{"step": 17150, "loss": 3.8724212646484375, "grad_norm": 0.6865888237953186, "learning_rate": 0.0001, "elbo": 3.87242, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0057471264367817}
{"step": 17155, "loss": 3.8561824798583983, "grad_norm": 0.7379284501075745, "learning_rate": 0.0001, "elbo": 3.85618, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0060403471733521}
{"step": 17160, "loss": 3.882172393798828, "grad_norm": 0.6366373300552368, "learning_rate": 0.0001, "elbo": 3.88217, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0063335679099226}
{"step": 17165, "loss": 3.8592609405517577, "grad_norm": 0.8182238340377808, "learning_rate": 0.0001, "elbo": 3.85926, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.006626788646493}
{"step": 17170, "loss": 3.860330581665039, "grad_norm": 0.771808922290802, "learning_rate": 0.0001, "elbo": 3.86033, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0069200093830635}
{"step": 17175, "loss": 3.862744903564453, "grad_norm": 0.649150013923645, "learning_rate": 0.0001, "elbo": 3.86274, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0072132301196342}
{"step": 17180, "loss": 3.858625793457031, "grad_norm": 0.6850878000259399, "learning_rate": 0.0001, "elbo": 3.85863, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0075064508562046}
{"step": 17185, "loss": 3.86610107421875, "grad_norm": 0.7143263220787048, "learning_rate": 0.0001, "elbo": 3.8661, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.007799671592775}
{"step": 17190, "loss": 3.870104217529297, "grad_norm": 0.7614879608154297, "learning_rate": 0.0001, "elbo": 3.8701, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0080928923293455}
{"step": 17195, "loss": 3.8724552154541017, "grad_norm": 0.739745557308197, "learning_rate": 0.0001, "elbo": 3.87246, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.008386113065916}
{"step": 17200, "loss": 3.883946990966797, "grad_norm": 0.7635936737060547, "learning_rate": 0.0001, "elbo": 3.88395, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0086793338024864}
{"step": 17205, "loss": 3.854741668701172, "grad_norm": 0.6665550470352173, "learning_rate": 0.0001, "elbo": 3.85474, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.008972554539057}
{"step": 17210, "loss": 3.8581573486328127, "grad_norm": 0.7347705960273743, "learning_rate": 0.0001, "elbo": 3.85816, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0092657752756276}
{"step": 17215, "loss": 3.8578025817871096, "grad_norm": 0.6631500124931335, "learning_rate": 0.0001, "elbo": 3.8578, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.009558996012198}
{"step": 17215, "loss": 3.8578025817871096, "grad_norm": 0.6631500124931335, "learning_rate": 0.0001, "elbo": 3.8578, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.009558996012198}
{"step": 17220, "loss": 3.8801918029785156, "grad_norm": 0.673985481262207, "learning_rate": 0.0001, "elbo": 3.88019, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0098522167487685}
{"step": 17225, "loss": 3.8545955657958983, "grad_norm": 0.8577448725700378, "learning_rate": 0.0001, "elbo": 3.8546, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.010145437485339}
{"step": 17230, "loss": 3.866667938232422, "grad_norm": 0.8372216820716858, "learning_rate": 0.0001, "elbo": 3.86667, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0104386582219094}
{"step": 17230, "loss": 3.866667938232422, "grad_norm": 0.8372216820716858, "learning_rate": 0.0001, "elbo": 3.86667, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0104386582219094}
{"step": 17235, "loss": 3.8660343170166014, "grad_norm": 0.651336669921875, "learning_rate": 0.0001, "elbo": 3.86603, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.01073187895848}
{"step": 17240, "loss": 3.864479827880859, "grad_norm": 0.7270358800888062, "learning_rate": 0.0001, "elbo": 3.86448, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0110250996950505}
{"step": 17245, "loss": 3.8592819213867187, "grad_norm": 0.6991639137268066, "learning_rate": 0.0001, "elbo": 3.85928, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.011318320431621}
{"step": 17250, "loss": 3.8798999786376953, "grad_norm": 0.7785236835479736, "learning_rate": 0.0001, "elbo": 3.8799, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0116115411681914}
{"step": 17255, "loss": 3.8743568420410157, "grad_norm": 1.0438495874404907, "learning_rate": 0.0001, "elbo": 3.87436, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0119047619047619}
{"step": 17260, "loss": 3.86497802734375, "grad_norm": 0.6843767166137695, "learning_rate": 0.0001, "elbo": 3.86498, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0121979826413323}
{"step": 17265, "loss": 3.8578155517578123, "grad_norm": 0.7857062220573425, "learning_rate": 0.0001, "elbo": 3.85782, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.012491203377903}
{"step": 17270, "loss": 3.8746837615966796, "grad_norm": 0.7207686305046082, "learning_rate": 0.0001, "elbo": 3.87468, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0127844241144734}
{"step": 17275, "loss": 3.8596595764160155, "grad_norm": 0.7617525458335876, "learning_rate": 0.0001, "elbo": 3.85966, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.013077644851044}
{"step": 17280, "loss": 3.8520233154296877, "grad_norm": 0.671340823173523, "learning_rate": 0.0001, "elbo": 3.85202, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0133708655876144}
{"step": 17285, "loss": 3.8663238525390624, "grad_norm": 0.6527512669563293, "learning_rate": 0.0001, "elbo": 3.86632, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0136640863241848}
{"step": 17290, "loss": 3.844316864013672, "grad_norm": 0.6346415281295776, "learning_rate": 0.0001, "elbo": 3.84432, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0139573070607553}
{"step": 17295, "loss": 3.859157943725586, "grad_norm": 0.6561976075172424, "learning_rate": 0.0001, "elbo": 3.85916, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0142505277973257}
{"step": 17300, "loss": 3.8570953369140626, "grad_norm": 0.7063185572624207, "learning_rate": 0.0001, "elbo": 3.8571, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0145437485338964}
{"step": 17305, "loss": 3.881987762451172, "grad_norm": 0.6832805871963501, "learning_rate": 0.0001, "elbo": 3.88199, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0148369692704668}
{"step": 17310, "loss": 3.8588851928710937, "grad_norm": 0.7810049653053284, "learning_rate": 0.0001, "elbo": 3.85889, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0151301900070373}
{"step": 17315, "loss": 3.8591079711914062, "grad_norm": 0.7165976166725159, "learning_rate": 0.0001, "elbo": 3.85911, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0154234107436078}
{"step": 17320, "loss": 3.8623863220214845, "grad_norm": 0.7003754377365112, "learning_rate": 0.0001, "elbo": 3.86239, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0157166314801782}
{"step": 17325, "loss": 3.8560081481933595, "grad_norm": 0.6968351602554321, "learning_rate": 0.0001, "elbo": 3.85601, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0160098522167487}
{"step": 17330, "loss": 3.8772544860839844, "grad_norm": 0.690159261226654, "learning_rate": 0.0001, "elbo": 3.87725, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0163030729533193}
{"step": 17335, "loss": 3.863240432739258, "grad_norm": 0.662601113319397, "learning_rate": 0.0001, "elbo": 3.86324, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0165962936898898}
{"step": 17340, "loss": 3.848061370849609, "grad_norm": 0.7275111079216003, "learning_rate": 0.0001, "elbo": 3.84806, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0168895144264603}
{"step": 17345, "loss": 3.8663398742675783, "grad_norm": 0.7882553935050964, "learning_rate": 0.0001, "elbo": 3.86634, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0171827351630307}
{"step": 17350, "loss": 3.8721351623535156, "grad_norm": 0.7489743232727051, "learning_rate": 0.0001, "elbo": 3.87214, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0174759558996012}
{"step": 17355, "loss": 3.877017593383789, "grad_norm": 0.6858470439910889, "learning_rate": 0.0001, "elbo": 3.87702, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0177691766361716}
{"step": 17360, "loss": 3.8560195922851563, "grad_norm": 0.6993444561958313, "learning_rate": 0.0001, "elbo": 3.85602, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0180623973727423}
{"step": 17365, "loss": 3.8694271087646483, "grad_norm": 1.1609190702438354, "learning_rate": 0.0001, "elbo": 3.86943, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0183556181093127}
{"step": 17370, "loss": 3.848033905029297, "grad_norm": 0.687882661819458, "learning_rate": 0.0001, "elbo": 3.84803, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0186488388458832}
{"step": 17375, "loss": 3.8669830322265626, "grad_norm": 0.7062722444534302, "learning_rate": 0.0001, "elbo": 3.86698, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0189420595824537}
{"step": 17380, "loss": 3.8729785919189452, "grad_norm": 0.7944954037666321, "learning_rate": 0.0001, "elbo": 3.87298, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.019235280319024}
{"step": 17385, "loss": 3.8703567504882814, "grad_norm": 0.6575788259506226, "learning_rate": 0.0001, "elbo": 3.87036, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0195285010555946}
{"step": 17390, "loss": 3.8639858245849608, "grad_norm": 0.7326087355613708, "learning_rate": 0.0001, "elbo": 3.86399, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0198217217921652}
{"step": 17395, "loss": 3.8718894958496093, "grad_norm": 0.7043692469596863, "learning_rate": 0.0001, "elbo": 3.87189, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0201149425287357}
{"step": 17400, "loss": 3.851274871826172, "grad_norm": 0.7912223935127258, "learning_rate": 0.0001, "elbo": 3.85127, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0204081632653061}
{"step": 17405, "loss": 3.8511199951171875, "grad_norm": 0.7076098918914795, "learning_rate": 0.0001, "elbo": 3.85112, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0207013840018766}
{"step": 17410, "loss": 3.8665164947509765, "grad_norm": 0.6770432591438293, "learning_rate": 0.0001, "elbo": 3.86652, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.020994604738447}
{"step": 17415, "loss": 3.8509147644042967, "grad_norm": 0.6749943494796753, "learning_rate": 0.0001, "elbo": 3.85091, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0212878254750175}
{"step": 17420, "loss": 3.863816833496094, "grad_norm": 0.6353086829185486, "learning_rate": 0.0001, "elbo": 3.86382, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0215810462115882}
{"step": 17425, "loss": 3.8816757202148438, "grad_norm": 0.7436847686767578, "learning_rate": 0.0001, "elbo": 3.88168, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0218742669481586}
{"step": 17430, "loss": 3.8602638244628906, "grad_norm": 0.7295305132865906, "learning_rate": 0.0001, "elbo": 3.86026, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.022167487684729}
{"step": 17435, "loss": 3.858267974853516, "grad_norm": 0.6526258587837219, "learning_rate": 0.0001, "elbo": 3.85827, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0224607084212995}
{"step": 17440, "loss": 3.871165466308594, "grad_norm": 0.6900450587272644, "learning_rate": 0.0001, "elbo": 3.87117, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.02275392915787}
{"step": 17445, "loss": 3.8547534942626953, "grad_norm": 0.652077853679657, "learning_rate": 0.0001, "elbo": 3.85475, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0230471498944405}
{"step": 17450, "loss": 3.846436309814453, "grad_norm": 0.674498975276947, "learning_rate": 0.0001, "elbo": 3.84644, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0233403706310111}
{"step": 17455, "loss": 3.839712142944336, "grad_norm": 0.7084980607032776, "learning_rate": 0.0001, "elbo": 3.83971, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0236335913675816}
{"step": 17460, "loss": 3.868642044067383, "grad_norm": 0.7305524349212646, "learning_rate": 0.0001, "elbo": 3.86864, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.023926812104152}
{"step": 17465, "loss": 3.8604225158691405, "grad_norm": 0.7289910912513733, "learning_rate": 0.0001, "elbo": 3.86042, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0242200328407225}
{"step": 17470, "loss": 3.8571800231933593, "grad_norm": 0.6986787915229797, "learning_rate": 0.0001, "elbo": 3.85718, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.024513253577293}
{"step": 17475, "loss": 3.860870361328125, "grad_norm": 0.7033311724662781, "learning_rate": 0.0001, "elbo": 3.86087, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0248064743138634}
{"step": 17480, "loss": 3.860670471191406, "grad_norm": 0.6446758508682251, "learning_rate": 0.0001, "elbo": 3.86067, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.025099695050434}
{"step": 17485, "loss": 3.8684036254882814, "grad_norm": 0.6807863116264343, "learning_rate": 0.0001, "elbo": 3.8684, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0253929157870045}
{"step": 17490, "loss": 3.8649520874023438, "grad_norm": 0.680932879447937, "learning_rate": 0.0001, "elbo": 3.86495, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.025686136523575}
{"step": 17495, "loss": 3.85238037109375, "grad_norm": 0.6548881530761719, "learning_rate": 0.0001, "elbo": 3.85238, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0259793572601454}
{"step": 17500, "loss": 3.857216644287109, "grad_norm": 0.6970011591911316, "learning_rate": 0.0001, "elbo": 3.85722, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.026272577996716}
{"step": 17505, "loss": 3.857148361206055, "grad_norm": 0.7109392285346985, "learning_rate": 0.0001, "elbo": 3.85715, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0265657987332863}
{"step": 17510, "loss": 3.8577117919921875, "grad_norm": 0.6650981903076172, "learning_rate": 0.0001, "elbo": 3.85771, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0268590194698568}
{"step": 17515, "loss": 3.8544681549072264, "grad_norm": 0.7186310887336731, "learning_rate": 0.0001, "elbo": 3.85447, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0271522402064275}
{"step": 17520, "loss": 3.8436084747314454, "grad_norm": 0.7964049577713013, "learning_rate": 0.0001, "elbo": 3.84361, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.027445460942998}
{"step": 17525, "loss": 3.8434539794921876, "grad_norm": 0.7141880989074707, "learning_rate": 0.0001, "elbo": 3.84345, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0277386816795684}
{"step": 17530, "loss": 3.8669113159179687, "grad_norm": 0.6773205399513245, "learning_rate": 0.0001, "elbo": 3.86691, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0280319024161388}
{"step": 17535, "loss": 3.861041259765625, "grad_norm": 0.7172529697418213, "learning_rate": 0.0001, "elbo": 3.86104, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0283251231527093}
{"step": 17540, "loss": 3.8563262939453127, "grad_norm": 0.6869024038314819, "learning_rate": 0.0001, "elbo": 3.85633, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0286183438892798}
{"step": 17545, "loss": 3.8600181579589843, "grad_norm": 0.7211355566978455, "learning_rate": 0.0001, "elbo": 3.86002, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0289115646258504}
{"step": 17550, "loss": 3.846570587158203, "grad_norm": 0.748443067073822, "learning_rate": 0.0001, "elbo": 3.84657, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0292047853624209}
{"step": 17555, "loss": 3.8527252197265627, "grad_norm": 0.9212894439697266, "learning_rate": 0.0001, "elbo": 3.85273, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0294980060989913}
{"step": 17560, "loss": 3.8442169189453126, "grad_norm": 0.6785900592803955, "learning_rate": 0.0001, "elbo": 3.84422, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0297912268355618}
{"step": 17565, "loss": 3.863615036010742, "grad_norm": 0.7863137722015381, "learning_rate": 0.0001, "elbo": 3.86361, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0300844475721322}
{"step": 17570, "loss": 3.8514225006103517, "grad_norm": 0.6664476990699768, "learning_rate": 0.0001, "elbo": 3.85142, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0303776683087027}
{"step": 17575, "loss": 3.867780303955078, "grad_norm": 0.6946207284927368, "learning_rate": 0.0001, "elbo": 3.86778, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0306708890452734}
{"step": 17580, "loss": 3.8692276000976564, "grad_norm": 0.8317500352859497, "learning_rate": 0.0001, "elbo": 3.86923, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0309641097818438}
{"step": 17585, "loss": 3.86810188293457, "grad_norm": 0.8525254130363464, "learning_rate": 0.0001, "elbo": 3.8681, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0312573305184143}
{"step": 17590, "loss": 3.8518238067626953, "grad_norm": 0.6873730421066284, "learning_rate": 0.0001, "elbo": 3.85182, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0315505512549847}
{"step": 17595, "loss": 3.851502227783203, "grad_norm": 0.7866717576980591, "learning_rate": 0.0001, "elbo": 3.8515, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0318437719915552}
{"step": 17600, "loss": 3.8552936553955077, "grad_norm": 0.7603195905685425, "learning_rate": 0.0001, "elbo": 3.85529, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0321369927281256}
{"step": 17605, "loss": 3.8664520263671873, "grad_norm": 0.66978919506073, "learning_rate": 0.0001, "elbo": 3.86645, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0324302134646963}
{"step": 17610, "loss": 3.8682491302490236, "grad_norm": 0.7297962307929993, "learning_rate": 0.0001, "elbo": 3.86825, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0327234342012668}
{"step": 17615, "loss": 3.880297088623047, "grad_norm": 0.7035202383995056, "learning_rate": 0.0001, "elbo": 3.8803, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0330166549378372}
{"step": 17620, "loss": 3.8613739013671875, "grad_norm": 0.8026608824729919, "learning_rate": 0.0001, "elbo": 3.86137, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0333098756744077}
{"step": 17625, "loss": 3.8592288970947264, "grad_norm": 0.7564869523048401, "learning_rate": 0.0001, "elbo": 3.85923, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0336030964109781}
{"step": 17630, "loss": 3.8783554077148437, "grad_norm": 0.6922031044960022, "learning_rate": 0.0001, "elbo": 3.87836, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0338963171475486}
{"step": 17635, "loss": 3.876761627197266, "grad_norm": 0.6939301490783691, "learning_rate": 0.0001, "elbo": 3.87676, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0341895378841193}
{"step": 17640, "loss": 3.8620182037353517, "grad_norm": 0.7020203471183777, "learning_rate": 0.0001, "elbo": 3.86202, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0344827586206897}
{"step": 17645, "loss": 3.8789424896240234, "grad_norm": 0.7495115399360657, "learning_rate": 0.0001, "elbo": 3.87894, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0347759793572602}
{"step": 17650, "loss": 3.8584598541259765, "grad_norm": 0.7580205202102661, "learning_rate": 0.0001, "elbo": 3.85846, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0350692000938306}
{"step": 17655, "loss": 3.8598594665527344, "grad_norm": 0.6857370138168335, "learning_rate": 0.0001, "elbo": 3.85986, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.035362420830401}
{"step": 17660, "loss": 3.8625831604003906, "grad_norm": 0.6807411313056946, "learning_rate": 0.0001, "elbo": 3.86258, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0356556415669715}
{"step": 17665, "loss": 3.8461788177490233, "grad_norm": 0.7116892337799072, "learning_rate": 0.0001, "elbo": 3.84618, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0359488623035422}
{"step": 17670, "loss": 3.879405975341797, "grad_norm": 0.9774143695831299, "learning_rate": 0.0001, "elbo": 3.87941, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0362420830401127}
{"step": 17675, "loss": 3.8653274536132813, "grad_norm": 0.7108854055404663, "learning_rate": 0.0001, "elbo": 3.86533, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0365353037766831}
{"step": 17680, "loss": 3.862000274658203, "grad_norm": 0.7516773343086243, "learning_rate": 0.0001, "elbo": 3.862, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0368285245132536}
{"step": 17685, "loss": 3.856171798706055, "grad_norm": 0.949004590511322, "learning_rate": 0.0001, "elbo": 3.85617, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.037121745249824}
{"step": 17690, "loss": 3.8676528930664062, "grad_norm": 0.7834521532058716, "learning_rate": 0.0001, "elbo": 3.86765, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0374149659863945}
{"step": 17695, "loss": 3.8591522216796874, "grad_norm": 0.7558152079582214, "learning_rate": 0.0001, "elbo": 3.85915, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.037708186722965}
{"step": 17700, "loss": 3.8624031066894533, "grad_norm": 0.7374983429908752, "learning_rate": 0.0001, "elbo": 3.8624, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0380014074595356}
{"step": 17705, "loss": 3.869578552246094, "grad_norm": 0.7016156315803528, "learning_rate": 0.0001, "elbo": 3.86958, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.038294628196106}
{"step": 17710, "loss": 3.8524860382080077, "grad_norm": 0.6665434837341309, "learning_rate": 0.0001, "elbo": 3.85249, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0385878489326765}
{"step": 17715, "loss": 3.8616207122802733, "grad_norm": 0.7309837937355042, "learning_rate": 0.0001, "elbo": 3.86162, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.038881069669247}
{"step": 17720, "loss": 3.8583454132080077, "grad_norm": 0.6835058331489563, "learning_rate": 0.0001, "elbo": 3.85835, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0391742904058174}
{"step": 17725, "loss": 3.8493721008300783, "grad_norm": 0.6834561228752136, "learning_rate": 0.0001, "elbo": 3.84937, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.039467511142388}
{"step": 17730, "loss": 3.8480911254882812, "grad_norm": 0.7188996076583862, "learning_rate": 0.0001, "elbo": 3.84809, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0397607318789586}
{"step": 17735, "loss": 3.8369102478027344, "grad_norm": 0.8179214596748352, "learning_rate": 0.0001, "elbo": 3.83691, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.040053952615529}
{"step": 17740, "loss": 3.8650489807128907, "grad_norm": 0.7853872776031494, "learning_rate": 0.0001, "elbo": 3.86505, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0403471733520995}
{"step": 17745, "loss": 3.83876953125, "grad_norm": 0.7734552025794983, "learning_rate": 0.0001, "elbo": 3.83877, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.04064039408867}
{"step": 17750, "loss": 3.8765838623046873, "grad_norm": 0.7503547668457031, "learning_rate": 0.0001, "elbo": 3.87658, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0409336148252404}
{"step": 17755, "loss": 3.8591346740722656, "grad_norm": 1.1855332851409912, "learning_rate": 0.0001, "elbo": 3.85913, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0412268355618108}
{"step": 17760, "loss": 3.8587112426757812, "grad_norm": 1.4952812194824219, "learning_rate": 0.0001, "elbo": 3.85871, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0415200562983815}
{"step": 17765, "loss": 3.8483722686767576, "grad_norm": 0.7147632837295532, "learning_rate": 0.0001, "elbo": 3.84837, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.041813277034952}
{"step": 17770, "loss": 3.8644786834716798, "grad_norm": 0.7654973864555359, "learning_rate": 0.0001, "elbo": 3.86448, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0421064977715224}
{"step": 17775, "loss": 3.8611968994140624, "grad_norm": 0.7346923351287842, "learning_rate": 0.0001, "elbo": 3.8612, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0423997185080929}
{"step": 17780, "loss": 3.8429054260253905, "grad_norm": 0.6446844935417175, "learning_rate": 0.0001, "elbo": 3.84291, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0426929392446633}
{"step": 17785, "loss": 3.8430713653564452, "grad_norm": 0.6239022016525269, "learning_rate": 0.0001, "elbo": 3.84307, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0429861599812338}
{"step": 17790, "loss": 3.8508235931396486, "grad_norm": 0.6818839311599731, "learning_rate": 0.0001, "elbo": 3.85082, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0432793807178045}
{"step": 17795, "loss": 3.8436378479003905, "grad_norm": 0.796263575553894, "learning_rate": 0.0001, "elbo": 3.84364, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.043572601454375}
{"step": 17800, "loss": 3.852083206176758, "grad_norm": 0.749106764793396, "learning_rate": 0.0001, "elbo": 3.85208, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0438658221909454}
{"step": 17805, "loss": 3.8616455078125, "grad_norm": 0.6908612251281738, "learning_rate": 0.0001, "elbo": 3.86165, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0441590429275158}
{"step": 17810, "loss": 3.877155303955078, "grad_norm": 0.7517645955085754, "learning_rate": 0.0001, "elbo": 3.87716, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0444522636640863}
{"step": 17810, "loss": 3.877155303955078, "grad_norm": 0.7517645955085754, "learning_rate": 0.0001, "elbo": 3.87716, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0444522636640863}
{"step": 17815, "loss": 3.866054916381836, "grad_norm": 0.6587589979171753, "learning_rate": 0.0001, "elbo": 3.86605, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0447454844006567}
{"step": 17820, "loss": 3.865517425537109, "grad_norm": 0.8793593645095825, "learning_rate": 0.0001, "elbo": 3.86552, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0450387051372274}
{"step": 17825, "loss": 3.843785858154297, "grad_norm": 0.7243281006813049, "learning_rate": 0.0001, "elbo": 3.84379, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0453319258737979}
{"step": 17830, "loss": 3.856572723388672, "grad_norm": 0.7720599174499512, "learning_rate": 0.0001, "elbo": 3.85657, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0456251466103683}
{"step": 17835, "loss": 3.8587966918945313, "grad_norm": 0.7226929664611816, "learning_rate": 0.0001, "elbo": 3.8588, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0459183673469388}
{"step": 17840, "loss": 3.8541633605957033, "grad_norm": 0.6598261594772339, "learning_rate": 0.0001, "elbo": 3.85416, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0462115880835092}
{"step": 17845, "loss": 3.851510238647461, "grad_norm": 0.8921381235122681, "learning_rate": 0.0001, "elbo": 3.85151, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0465048088200797}
{"step": 17850, "loss": 3.8343795776367187, "grad_norm": 0.6579152345657349, "learning_rate": 0.0001, "elbo": 3.83438, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0467980295566504}
{"step": 17850, "loss": 3.8343795776367187, "grad_norm": 0.6579152345657349, "learning_rate": 0.0001, "elbo": 3.83438, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0467980295566504}
{"step": 17855, "loss": 3.853978729248047, "grad_norm": 0.7319185137748718, "learning_rate": 0.0001, "elbo": 3.85398, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0470912502932208}
{"step": 17860, "loss": 3.871298980712891, "grad_norm": 0.6890305280685425, "learning_rate": 0.0001, "elbo": 3.8713, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0473844710297913}
{"step": 17865, "loss": 3.8464736938476562, "grad_norm": 0.7583089470863342, "learning_rate": 0.0001, "elbo": 3.84647, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0476776917663617}
{"step": 17870, "loss": 3.852067565917969, "grad_norm": 0.6490498781204224, "learning_rate": 0.0001, "elbo": 3.85207, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0479709125029322}
{"step": 17875, "loss": 3.869309997558594, "grad_norm": 0.6836749315261841, "learning_rate": 0.0001, "elbo": 3.86931, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0482641332395026}
{"step": 17880, "loss": 3.845040130615234, "grad_norm": 1.0002268552780151, "learning_rate": 0.0001, "elbo": 3.84504, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.048557353976073}
{"step": 17885, "loss": 3.8592601776123048, "grad_norm": 0.7044047117233276, "learning_rate": 0.0001, "elbo": 3.85926, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0488505747126438}
{"step": 17890, "loss": 3.8591575622558594, "grad_norm": 0.6970529556274414, "learning_rate": 0.0001, "elbo": 3.85916, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0491437954492142}
{"step": 17895, "loss": 3.8468582153320314, "grad_norm": 0.6962904930114746, "learning_rate": 0.0001, "elbo": 3.84686, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0494370161857847}
{"step": 17900, "loss": 3.840789031982422, "grad_norm": 0.707023024559021, "learning_rate": 0.0001, "elbo": 3.84079, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0497302369223551}
{"step": 17905, "loss": 3.861795425415039, "grad_norm": 0.6974694728851318, "learning_rate": 0.0001, "elbo": 3.8618, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0500234576589256}
{"step": 17910, "loss": 3.8622711181640623, "grad_norm": 0.8049870729446411, "learning_rate": 0.0001, "elbo": 3.86227, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.050316678395496}
{"step": 17915, "loss": 3.8644256591796875, "grad_norm": 0.7114613652229309, "learning_rate": 0.0001, "elbo": 3.86443, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0506098991320667}
{"step": 17920, "loss": 3.870606231689453, "grad_norm": 0.7924872636795044, "learning_rate": 0.0001, "elbo": 3.87061, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0509031198686372}
{"step": 17925, "loss": 3.8485931396484374, "grad_norm": 0.7358497977256775, "learning_rate": 0.0001, "elbo": 3.84859, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0511963406052076}
{"step": 17930, "loss": 3.8560260772705077, "grad_norm": 0.6614987254142761, "learning_rate": 0.0001, "elbo": 3.85603, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.051489561341778}
{"step": 17935, "loss": 3.8512916564941406, "grad_norm": 0.7174339890480042, "learning_rate": 0.0001, "elbo": 3.85129, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0517827820783485}
{"step": 17940, "loss": 3.8562850952148438, "grad_norm": 0.6508845090866089, "learning_rate": 0.0001, "elbo": 3.85629, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.052076002814919}
{"step": 17940, "loss": 3.8562850952148438, "grad_norm": 0.6508845090866089, "learning_rate": 0.0001, "elbo": 3.85629, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.052076002814919}
{"step": 17945, "loss": 3.850755310058594, "grad_norm": 0.7463220953941345, "learning_rate": 0.0001, "elbo": 3.85076, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0523692235514897}
{"step": 17950, "loss": 3.857494354248047, "grad_norm": 0.6560964584350586, "learning_rate": 0.0001, "elbo": 3.85749, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.05266244428806}
{"step": 17950, "loss": 3.857494354248047, "grad_norm": 0.6560964584350586, "learning_rate": 0.0001, "elbo": 3.85749, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.05266244428806}
{"step": 17955, "loss": 3.8435691833496093, "grad_norm": 0.6867692470550537, "learning_rate": 0.0001, "elbo": 3.84357, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0529556650246306}
{"step": 17960, "loss": 3.8482494354248047, "grad_norm": 0.6489271521568298, "learning_rate": 0.0001, "elbo": 3.84825, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.053248885761201}
{"step": 17965, "loss": 3.860744094848633, "grad_norm": 0.7048158645629883, "learning_rate": 0.0001, "elbo": 3.86074, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0535421064977715}
{"step": 17970, "loss": 3.869371032714844, "grad_norm": 0.6241839528083801, "learning_rate": 0.0001, "elbo": 3.86937, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.053835327234342}
{"step": 17970, "loss": 3.869371032714844, "grad_norm": 0.6241839528083801, "learning_rate": 0.0001, "elbo": 3.86937, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.053835327234342}
{"step": 17975, "loss": 3.8595607757568358, "grad_norm": 0.683987557888031, "learning_rate": 0.0001, "elbo": 3.85956, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0541285479709126}
{"step": 17980, "loss": 3.847509765625, "grad_norm": 0.6954723000526428, "learning_rate": 0.0001, "elbo": 3.84751, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.054421768707483}
{"step": 17980, "loss": 3.847509765625, "grad_norm": 0.6954723000526428, "learning_rate": 0.0001, "elbo": 3.84751, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.054421768707483}
{"step": 17985, "loss": 3.838950347900391, "grad_norm": 0.6975278258323669, "learning_rate": 0.0001, "elbo": 3.83895, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0547149894440535}
{"step": 17990, "loss": 3.8579654693603516, "grad_norm": 0.7568650245666504, "learning_rate": 0.0001, "elbo": 3.85797, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.055008210180624}
{"step": 17995, "loss": 3.8647666931152345, "grad_norm": 0.6919245719909668, "learning_rate": 0.0001, "elbo": 3.86477, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0553014309171944}
{"step": 18000, "loss": 3.8600082397460938, "grad_norm": 0.6790667176246643, "learning_rate": 0.0001, "elbo": 3.86001, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0555946516537649}
{"step": 18000, "loss": 3.8600082397460938, "grad_norm": 0.6790667176246643, "learning_rate": 0.0001, "elbo": 3.86001, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0555946516537649}
{"step": 18005, "loss": 3.8521453857421877, "grad_norm": 0.6999588012695312, "learning_rate": 0.0001, "elbo": 3.85215, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0558878723903355}
{"step": 18010, "loss": 3.8636142730712892, "grad_norm": 0.6981610655784607, "learning_rate": 0.0001, "elbo": 3.86361, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.056181093126906}
{"step": 18015, "loss": 3.8436927795410156, "grad_norm": 0.7137483954429626, "learning_rate": 0.0001, "elbo": 3.84369, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0564743138634765}
{"step": 18015, "loss": 3.8436927795410156, "grad_norm": 0.7137483954429626, "learning_rate": 0.0001, "elbo": 3.84369, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0564743138634765}
{"step": 18020, "loss": 3.8452117919921873, "grad_norm": 0.6135759353637695, "learning_rate": 0.0001, "elbo": 3.84521, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.056767534600047}
{"step": 18025, "loss": 3.8555469512939453, "grad_norm": 0.717767059803009, "learning_rate": 0.0001, "elbo": 3.85555, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0570607553366174}
{"step": 18030, "loss": 3.855348587036133, "grad_norm": 0.7589754462242126, "learning_rate": 0.0001, "elbo": 3.85535, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0573539760731878}
{"step": 18035, "loss": 3.8459415435791016, "grad_norm": 0.7503893375396729, "learning_rate": 0.0001, "elbo": 3.84594, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0576471968097585}
{"step": 18040, "loss": 3.8534847259521485, "grad_norm": 0.6829363703727722, "learning_rate": 0.0001, "elbo": 3.85348, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.057940417546329}
{"step": 18045, "loss": 3.8252029418945312, "grad_norm": 0.7013847231864929, "learning_rate": 0.0001, "elbo": 3.8252, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0582336382828994}
{"step": 18050, "loss": 3.8646171569824217, "grad_norm": 0.8551681637763977, "learning_rate": 0.0001, "elbo": 3.86462, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0585268590194699}
{"step": 18055, "loss": 3.8462345123291017, "grad_norm": 0.6756444573402405, "learning_rate": 0.0001, "elbo": 3.84623, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0588200797560403}
{"step": 18060, "loss": 3.850469207763672, "grad_norm": 0.6685364842414856, "learning_rate": 0.0001, "elbo": 3.85047, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0591133004926108}
{"step": 18065, "loss": 3.8537673950195312, "grad_norm": 0.8136768341064453, "learning_rate": 0.0001, "elbo": 3.85377, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0594065212291812}
{"step": 18070, "loss": 3.8591026306152343, "grad_norm": 0.6975540518760681, "learning_rate": 0.0001, "elbo": 3.8591, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.059699741965752}
{"step": 18075, "loss": 3.856243133544922, "grad_norm": 0.7636975646018982, "learning_rate": 0.0001, "elbo": 3.85624, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0599929627023223}
{"step": 18080, "loss": 3.8616065979003906, "grad_norm": 0.6866768598556519, "learning_rate": 0.0001, "elbo": 3.86161, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0602861834388928}
{"step": 18085, "loss": 3.8606449127197267, "grad_norm": 0.8601483702659607, "learning_rate": 0.0001, "elbo": 3.86064, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0605794041754633}
{"step": 18090, "loss": 3.872292327880859, "grad_norm": 0.6827394962310791, "learning_rate": 0.0001, "elbo": 3.87229, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0608726249120337}
{"step": 18095, "loss": 3.8595046997070312, "grad_norm": 0.7757266163825989, "learning_rate": 0.0001, "elbo": 3.8595, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0611658456486042}
{"step": 18100, "loss": 3.8511592864990236, "grad_norm": 0.959538996219635, "learning_rate": 0.0001, "elbo": 3.85116, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0614590663851748}
{"step": 18105, "loss": 3.8591808319091796, "grad_norm": 0.6958470940589905, "learning_rate": 0.0001, "elbo": 3.85918, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0617522871217453}
{"step": 18110, "loss": 3.875214767456055, "grad_norm": 0.6936483383178711, "learning_rate": 0.0001, "elbo": 3.87521, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0620455078583158}
{"step": 18115, "loss": 3.8359001159667967, "grad_norm": 0.727155327796936, "learning_rate": 0.0001, "elbo": 3.8359, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0623387285948862}
{"step": 18120, "loss": 3.839041900634766, "grad_norm": 0.8100605010986328, "learning_rate": 0.0001, "elbo": 3.83904, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0626319493314567}
{"step": 18125, "loss": 3.8354114532470702, "grad_norm": 0.6635639071464539, "learning_rate": 0.0001, "elbo": 3.83541, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0629251700680271}
{"step": 18130, "loss": 3.845020294189453, "grad_norm": 0.7629842758178711, "learning_rate": 0.0001, "elbo": 3.84502, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0632183908045978}
{"step": 18135, "loss": 3.849284362792969, "grad_norm": 0.6333636045455933, "learning_rate": 0.0001, "elbo": 3.84928, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0635116115411682}
{"step": 18140, "loss": 3.847187042236328, "grad_norm": 0.7231437563896179, "learning_rate": 0.0001, "elbo": 3.84719, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0638048322777387}
{"step": 18145, "loss": 3.845562744140625, "grad_norm": 0.7314935326576233, "learning_rate": 0.0001, "elbo": 3.84556, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0640980530143092}
{"step": 18150, "loss": 3.8556449890136717, "grad_norm": 0.76400226354599, "learning_rate": 0.0001, "elbo": 3.85564, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0643912737508796}
{"step": 18155, "loss": 3.8379478454589844, "grad_norm": 0.7366521954536438, "learning_rate": 0.0001, "elbo": 3.83795, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.06468449448745}
{"step": 18160, "loss": 3.836429214477539, "grad_norm": 0.6847171783447266, "learning_rate": 0.0001, "elbo": 3.83643, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0649777152240207}
{"step": 18165, "loss": 3.8361980438232424, "grad_norm": 0.7186843752861023, "learning_rate": 0.0001, "elbo": 3.8362, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0652709359605912}
{"step": 18170, "loss": 3.847669982910156, "grad_norm": 0.8144303560256958, "learning_rate": 0.0001, "elbo": 3.84767, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0655641566971616}
{"step": 18175, "loss": 3.8449451446533205, "grad_norm": 0.663692057132721, "learning_rate": 0.0001, "elbo": 3.84495, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.065857377433732}
{"step": 18180, "loss": 3.8487693786621096, "grad_norm": 0.6854488849639893, "learning_rate": 0.0001, "elbo": 3.84877, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0661505981703026}
{"step": 18185, "loss": 3.8480743408203124, "grad_norm": 0.7130178809165955, "learning_rate": 0.0001, "elbo": 3.84807, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.066443818906873}
{"step": 18190, "loss": 3.857183837890625, "grad_norm": 0.7403102517127991, "learning_rate": 0.0001, "elbo": 3.85718, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0667370396434437}
{"step": 18195, "loss": 3.8640182495117186, "grad_norm": 1.104701280593872, "learning_rate": 0.0001, "elbo": 3.86402, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0670302603800141}
{"step": 18200, "loss": 3.8397590637207033, "grad_norm": 0.799941897392273, "learning_rate": 0.0001, "elbo": 3.83976, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0673234811165846}
{"step": 18205, "loss": 3.836684799194336, "grad_norm": 0.7637914419174194, "learning_rate": 0.0001, "elbo": 3.83668, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.067616701853155}
{"step": 18210, "loss": 3.829656982421875, "grad_norm": 0.7222722172737122, "learning_rate": 0.0001, "elbo": 3.82966, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0679099225897255}
{"step": 18215, "loss": 3.862915802001953, "grad_norm": 0.7076325416564941, "learning_rate": 0.0001, "elbo": 3.86292, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.068203143326296}
{"step": 18220, "loss": 3.847367858886719, "grad_norm": 0.6726224422454834, "learning_rate": 0.0001, "elbo": 3.84737, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0684963640628666}
{"step": 18225, "loss": 3.8664817810058594, "grad_norm": 0.6982733607292175, "learning_rate": 0.0001, "elbo": 3.86648, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.068789584799437}
{"step": 18230, "loss": 3.8496337890625, "grad_norm": 0.7114553451538086, "learning_rate": 0.0001, "elbo": 3.84963, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0690828055360075}
{"step": 18235, "loss": 3.863606262207031, "grad_norm": 0.6548779606819153, "learning_rate": 0.0001, "elbo": 3.86361, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.069376026272578}
{"step": 18240, "loss": 3.8512317657470705, "grad_norm": 0.7587085366249084, "learning_rate": 0.0001, "elbo": 3.85123, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0696692470091484}
{"step": 18245, "loss": 3.840159225463867, "grad_norm": 0.6384826302528381, "learning_rate": 0.0001, "elbo": 3.84016, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.069962467745719}
{"step": 18250, "loss": 3.8555179595947267, "grad_norm": 0.6819256544113159, "learning_rate": 0.0001, "elbo": 3.85552, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0702556884822894}
{"step": 18255, "loss": 3.849361801147461, "grad_norm": 0.6432185769081116, "learning_rate": 0.0001, "elbo": 3.84936, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.07054890921886}
{"step": 18260, "loss": 3.8352149963378905, "grad_norm": 0.9525871276855469, "learning_rate": 0.0001, "elbo": 3.83522, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0708421299554305}
{"step": 18265, "loss": 3.8497695922851562, "grad_norm": 0.754859447479248, "learning_rate": 0.0001, "elbo": 3.84977, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.071135350692001}
{"step": 18270, "loss": 3.8544895172119142, "grad_norm": 0.6841384172439575, "learning_rate": 0.0001, "elbo": 3.85449, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0714285714285714}
{"step": 18275, "loss": 3.8547740936279298, "grad_norm": 0.6750783920288086, "learning_rate": 0.0001, "elbo": 3.85477, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0717217921651419}
{"step": 18280, "loss": 3.839114761352539, "grad_norm": 0.7606878876686096, "learning_rate": 0.0001, "elbo": 3.83911, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0720150129017123}
{"step": 18285, "loss": 3.8498779296875, "grad_norm": 0.6801553964614868, "learning_rate": 0.0001, "elbo": 3.84988, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.072308233638283}
{"step": 18290, "loss": 3.8549476623535157, "grad_norm": 0.6977383494377136, "learning_rate": 0.0001, "elbo": 3.85495, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0726014543748534}
{"step": 18295, "loss": 3.855574035644531, "grad_norm": 0.7693918347358704, "learning_rate": 0.0001, "elbo": 3.85557, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.072894675111424}
{"step": 18300, "loss": 3.855152130126953, "grad_norm": 0.6458277106285095, "learning_rate": 0.0001, "elbo": 3.85515, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0731878958479943}
{"step": 18305, "loss": 3.85386962890625, "grad_norm": 0.7177906036376953, "learning_rate": 0.0001, "elbo": 3.85387, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0734811165845648}
{"step": 18310, "loss": 3.846851348876953, "grad_norm": 0.7288267016410828, "learning_rate": 0.0001, "elbo": 3.84685, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0737743373211353}
{"step": 18315, "loss": 3.8491744995117188, "grad_norm": 0.7764604687690735, "learning_rate": 0.0001, "elbo": 3.84917, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.074067558057706}
{"step": 18320, "loss": 3.8367237091064452, "grad_norm": 0.7836275696754456, "learning_rate": 0.0001, "elbo": 3.83672, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0743607787942764}
{"step": 18325, "loss": 3.8531471252441407, "grad_norm": 0.6813885569572449, "learning_rate": 0.0001, "elbo": 3.85315, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0746539995308468}
{"step": 18330, "loss": 3.848988342285156, "grad_norm": 0.6778871417045593, "learning_rate": 0.0001, "elbo": 3.84899, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0749472202674173}
{"step": 18335, "loss": 3.843972396850586, "grad_norm": 0.7236977219581604, "learning_rate": 0.0001, "elbo": 3.84397, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0752404410039877}
{"step": 18340, "loss": 3.8293216705322264, "grad_norm": 0.7010890245437622, "learning_rate": 0.0001, "elbo": 3.82932, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0755336617405582}
{"step": 18345, "loss": 3.8479896545410157, "grad_norm": 0.6898422241210938, "learning_rate": 0.0001, "elbo": 3.84799, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0758268824771289}
{"step": 18350, "loss": 3.838540267944336, "grad_norm": 0.7835252285003662, "learning_rate": 0.0001, "elbo": 3.83854, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0761201032136993}
{"step": 18355, "loss": 3.8380577087402346, "grad_norm": 0.6790088415145874, "learning_rate": 0.0001, "elbo": 3.83806, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0764133239502698}
{"step": 18360, "loss": 3.8583534240722654, "grad_norm": 0.6990436911582947, "learning_rate": 0.0001, "elbo": 3.85835, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0767065446868402}
{"step": 18365, "loss": 3.866512680053711, "grad_norm": 0.6786105632781982, "learning_rate": 0.0001, "elbo": 3.86651, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0769997654234107}
{"step": 18370, "loss": 3.846910095214844, "grad_norm": 0.8738764524459839, "learning_rate": 0.0001, "elbo": 3.84691, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0772929861599811}
{"step": 18375, "loss": 3.8290264129638674, "grad_norm": 0.7486185431480408, "learning_rate": 0.0001, "elbo": 3.82903, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0775862068965518}
{"step": 18375, "loss": 3.8290264129638674, "grad_norm": 0.7486185431480408, "learning_rate": 0.0001, "elbo": 3.82903, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0775862068965518}
{"step": 18380, "loss": 3.8520877838134764, "grad_norm": 0.7013760805130005, "learning_rate": 0.0001, "elbo": 3.85209, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0778794276331223}
{"step": 18385, "loss": 3.8490066528320312, "grad_norm": 0.7150335311889648, "learning_rate": 0.0001, "elbo": 3.84901, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0781726483696927}
{"step": 18390, "loss": 3.842140960693359, "grad_norm": 0.6830870509147644, "learning_rate": 0.0001, "elbo": 3.84214, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0784658691062632}
{"step": 18395, "loss": 3.856962966918945, "grad_norm": 0.6611642241477966, "learning_rate": 0.0001, "elbo": 3.85696, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0787590898428336}
{"step": 18400, "loss": 3.8333587646484375, "grad_norm": 0.6979936957359314, "learning_rate": 0.0001, "elbo": 3.83336, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.079052310579404}
{"step": 18405, "loss": 3.846569061279297, "grad_norm": 0.7416452169418335, "learning_rate": 0.0001, "elbo": 3.84657, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0793455313159748}
{"step": 18410, "loss": 3.8299041748046876, "grad_norm": 0.7762389779090881, "learning_rate": 0.0001, "elbo": 3.8299, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0796387520525452}
{"step": 18415, "loss": 3.8505088806152346, "grad_norm": 0.7765761613845825, "learning_rate": 0.0001, "elbo": 3.85051, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0799319727891157}
{"step": 18420, "loss": 3.8295135498046875, "grad_norm": 0.6924477219581604, "learning_rate": 0.0001, "elbo": 3.82951, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0802251935256861}
{"step": 18425, "loss": 3.8389320373535156, "grad_norm": 0.6641021966934204, "learning_rate": 0.0001, "elbo": 3.83893, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0805184142622566}
{"step": 18430, "loss": 3.827248764038086, "grad_norm": 0.8698906302452087, "learning_rate": 0.0001, "elbo": 3.82725, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.080811634998827}
{"step": 18435, "loss": 3.8405860900878905, "grad_norm": 0.7257249355316162, "learning_rate": 0.0001, "elbo": 3.84059, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0811048557353975}
{"step": 18440, "loss": 3.832185745239258, "grad_norm": 0.7258006930351257, "learning_rate": 0.0001, "elbo": 3.83219, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0813980764719682}
{"step": 18445, "loss": 3.849221420288086, "grad_norm": 0.6732743978500366, "learning_rate": 0.0001, "elbo": 3.84922, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0816912972085386}
{"step": 18450, "loss": 3.854718780517578, "grad_norm": 0.7254971265792847, "learning_rate": 0.0001, "elbo": 3.85472, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.081984517945109}
{"step": 18455, "loss": 3.839049530029297, "grad_norm": 0.6943891048431396, "learning_rate": 0.0001, "elbo": 3.83905, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0822777386816795}
{"step": 18460, "loss": 3.833912658691406, "grad_norm": 0.6629189848899841, "learning_rate": 0.0001, "elbo": 3.83391, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.08257095941825}
{"step": 18465, "loss": 3.8544342041015627, "grad_norm": 0.7031924724578857, "learning_rate": 0.0001, "elbo": 3.85443, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0828641801548207}
{"step": 18470, "loss": 3.838161087036133, "grad_norm": 0.7171556949615479, "learning_rate": 0.0001, "elbo": 3.83816, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0831574008913911}
{"step": 18475, "loss": 3.859088897705078, "grad_norm": 0.8105496764183044, "learning_rate": 0.0001, "elbo": 3.85909, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0834506216279616}
{"step": 18480, "loss": 3.8534996032714846, "grad_norm": 0.9864534735679626, "learning_rate": 0.0001, "elbo": 3.8535, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.083743842364532}
{"step": 18485, "loss": 3.8539947509765624, "grad_norm": 0.8391327857971191, "learning_rate": 0.0001, "elbo": 3.85399, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0840370631011025}
{"step": 18490, "loss": 3.8395233154296875, "grad_norm": 0.672330379486084, "learning_rate": 0.0001, "elbo": 3.83952, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.084330283837673}
{"step": 18495, "loss": 3.840986633300781, "grad_norm": 0.6818567514419556, "learning_rate": 0.0001, "elbo": 3.84099, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0846235045742434}
{"step": 18500, "loss": 3.8374462127685547, "grad_norm": 0.8819236159324646, "learning_rate": 0.0001, "elbo": 3.83745, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.084916725310814}
{"step": 18505, "loss": 3.838215637207031, "grad_norm": 0.6657614707946777, "learning_rate": 0.0001, "elbo": 3.83822, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0852099460473845}
{"step": 18510, "loss": 3.830341339111328, "grad_norm": 0.84047931432724, "learning_rate": 0.0001, "elbo": 3.83034, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.085503166783955}
{"step": 18515, "loss": 3.8477523803710936, "grad_norm": 0.8341084122657776, "learning_rate": 0.0001, "elbo": 3.84775, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0857963875205254}
{"step": 18520, "loss": 3.850005340576172, "grad_norm": 0.7265958786010742, "learning_rate": 0.0001, "elbo": 3.85001, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0860896082570959}
{"step": 18525, "loss": 3.8606426239013674, "grad_norm": 0.7171376347541809, "learning_rate": 0.0001, "elbo": 3.86064, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0863828289936663}
{"step": 18530, "loss": 3.833261489868164, "grad_norm": 0.6916630864143372, "learning_rate": 0.0001, "elbo": 3.83326, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.086676049730237}
{"step": 18535, "loss": 3.834077835083008, "grad_norm": 0.6839744448661804, "learning_rate": 0.0001, "elbo": 3.83408, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0869692704668075}
{"step": 18540, "loss": 3.8457481384277346, "grad_norm": 0.6962758302688599, "learning_rate": 0.0001, "elbo": 3.84575, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.087262491203378}
{"step": 18545, "loss": 3.8490814208984374, "grad_norm": 0.7543414235115051, "learning_rate": 0.0001, "elbo": 3.84908, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0875557119399484}
{"step": 18550, "loss": 3.842190170288086, "grad_norm": 0.7080679535865784, "learning_rate": 0.0001, "elbo": 3.84219, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0878489326765188}
{"step": 18555, "loss": 3.859640121459961, "grad_norm": 0.7885720133781433, "learning_rate": 0.0001, "elbo": 3.85964, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0881421534130893}
{"step": 18560, "loss": 3.8415870666503906, "grad_norm": 0.8057392835617065, "learning_rate": 0.0001, "elbo": 3.84159, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.08843537414966}
{"step": 18565, "loss": 3.849870300292969, "grad_norm": 0.6785752177238464, "learning_rate": 0.0001, "elbo": 3.84987, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0887285948862304}
{"step": 18570, "loss": 3.852671813964844, "grad_norm": 0.7985442280769348, "learning_rate": 0.0001, "elbo": 3.85267, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0890218156228009}
{"step": 18575, "loss": 3.8426551818847656, "grad_norm": 0.7146263122558594, "learning_rate": 0.0001, "elbo": 3.84266, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0893150363593713}
{"step": 18580, "loss": 3.841933822631836, "grad_norm": 0.7685074806213379, "learning_rate": 0.0001, "elbo": 3.84193, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0896082570959418}
{"step": 18585, "loss": 3.8499080657958986, "grad_norm": 0.7205924391746521, "learning_rate": 0.0001, "elbo": 3.84991, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0899014778325122}
{"step": 18590, "loss": 3.83292236328125, "grad_norm": 0.6857749223709106, "learning_rate": 0.0001, "elbo": 3.83292, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.090194698569083}
{"step": 18595, "loss": 3.832386016845703, "grad_norm": 0.6847994327545166, "learning_rate": 0.0001, "elbo": 3.83239, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0904879193056534}
{"step": 18600, "loss": 3.8238983154296875, "grad_norm": 0.8104994893074036, "learning_rate": 0.0001, "elbo": 3.8239, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0907811400422238}
{"step": 18605, "loss": 3.852342224121094, "grad_norm": 0.712402880191803, "learning_rate": 0.0001, "elbo": 3.85234, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0910743607787943}
{"step": 18610, "loss": 3.8383689880371095, "grad_norm": 0.6921129822731018, "learning_rate": 0.0001, "elbo": 3.83837, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0913675815153647}
{"step": 18615, "loss": 3.8452423095703123, "grad_norm": 0.6506438851356506, "learning_rate": 0.0001, "elbo": 3.84524, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0916608022519352}
{"step": 18620, "loss": 3.8441898345947267, "grad_norm": 0.7111833691596985, "learning_rate": 0.0001, "elbo": 3.84419, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0919540229885056}
{"step": 18625, "loss": 3.826638412475586, "grad_norm": 0.6992493271827698, "learning_rate": 0.0001, "elbo": 3.82664, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0922472437250763}
{"step": 18630, "loss": 3.851995849609375, "grad_norm": 0.7266695499420166, "learning_rate": 0.0001, "elbo": 3.852, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0925404644616468}
{"step": 18635, "loss": 3.849675750732422, "grad_norm": 0.6670733094215393, "learning_rate": 0.0001, "elbo": 3.84968, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0928336851982172}
{"step": 18640, "loss": 3.846347427368164, "grad_norm": 0.7544905543327332, "learning_rate": 0.0001, "elbo": 3.84635, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0931269059347877}
{"step": 18645, "loss": 3.8460098266601563, "grad_norm": 0.6825729012489319, "learning_rate": 0.0001, "elbo": 3.84601, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0934201266713581}
{"step": 18650, "loss": 3.8304153442382813, "grad_norm": 0.6737632155418396, "learning_rate": 0.0001, "elbo": 3.83042, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0937133474079288}
{"step": 18655, "loss": 3.8452991485595702, "grad_norm": 0.897280752658844, "learning_rate": 0.0001, "elbo": 3.8453, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0940065681444993}
{"step": 18660, "loss": 3.8483543395996094, "grad_norm": 0.6647124886512756, "learning_rate": 0.0001, "elbo": 3.84835, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0942997888810697}
{"step": 18665, "loss": 3.8340614318847654, "grad_norm": 0.6822597980499268, "learning_rate": 0.0001, "elbo": 3.83406, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0945930096176402}
{"step": 18670, "loss": 3.8346431732177733, "grad_norm": 0.7196139097213745, "learning_rate": 0.0001, "elbo": 3.83464, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0948862303542106}
{"step": 18675, "loss": 3.842361831665039, "grad_norm": 0.7490147948265076, "learning_rate": 0.0001, "elbo": 3.84236, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.095179451090781}
{"step": 18680, "loss": 3.8314044952392576, "grad_norm": 0.9034222364425659, "learning_rate": 0.0001, "elbo": 3.8314, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0954726718273515}
{"step": 18685, "loss": 3.8222164154052733, "grad_norm": 0.6887336373329163, "learning_rate": 0.0001, "elbo": 3.82222, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0957658925639222}
{"step": 18690, "loss": 3.839338684082031, "grad_norm": 0.693056046962738, "learning_rate": 0.0001, "elbo": 3.83934, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0960591133004927}
{"step": 18695, "loss": 3.8287940979003907, "grad_norm": 0.7091038227081299, "learning_rate": 0.0001, "elbo": 3.82879, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0963523340370631}
{"step": 18700, "loss": 3.823801040649414, "grad_norm": 0.7003267407417297, "learning_rate": 0.0001, "elbo": 3.8238, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0966455547736336}
{"step": 18705, "loss": 3.835127258300781, "grad_norm": 0.7537055015563965, "learning_rate": 0.0001, "elbo": 3.83513, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.096938775510204}
{"step": 18710, "loss": 3.809782028198242, "grad_norm": 0.6866472363471985, "learning_rate": 0.0001, "elbo": 3.80978, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0972319962467745}
{"step": 18715, "loss": 3.8480255126953127, "grad_norm": 0.6669470071792603, "learning_rate": 0.0001, "elbo": 3.84803, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0975252169833452}
{"step": 18720, "loss": 3.8375923156738283, "grad_norm": 0.6830764412879944, "learning_rate": 0.0001, "elbo": 3.83759, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0978184377199156}
{"step": 18725, "loss": 3.8145462036132813, "grad_norm": 0.7023007869720459, "learning_rate": 0.0001, "elbo": 3.81455, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.098111658456486}
{"step": 18730, "loss": 3.8469989776611326, "grad_norm": 0.7356046438217163, "learning_rate": 0.0001, "elbo": 3.847, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0984048791930565}
{"step": 18735, "loss": 3.8400943756103514, "grad_norm": 0.704594612121582, "learning_rate": 0.0001, "elbo": 3.84009, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.098698099929627}
{"step": 18740, "loss": 3.8491649627685547, "grad_norm": 1.1494497060775757, "learning_rate": 0.0001, "elbo": 3.84917, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0989913206661974}
{"step": 18745, "loss": 3.8507205963134767, "grad_norm": 0.8568732738494873, "learning_rate": 0.0001, "elbo": 3.85072, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.099284541402768}
{"step": 18750, "loss": 3.8453907012939452, "grad_norm": 0.716827929019928, "learning_rate": 0.0001, "elbo": 3.84539, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.0995777621393386}
{"step": 18755, "loss": 3.84478759765625, "grad_norm": 0.8248330950737, "learning_rate": 0.0001, "elbo": 3.84479, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.099870982875909}
{"step": 18760, "loss": 3.8475406646728514, "grad_norm": 0.6459624767303467, "learning_rate": 0.0001, "elbo": 3.84754, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1001642036124795}
{"step": 18765, "loss": 3.8458179473876952, "grad_norm": 0.8436006903648376, "learning_rate": 0.0001, "elbo": 3.84582, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.10045742434905}
{"step": 18770, "loss": 3.84771842956543, "grad_norm": 0.648455023765564, "learning_rate": 0.0001, "elbo": 3.84772, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1007506450856204}
{"step": 18775, "loss": 3.8460044860839844, "grad_norm": 0.6354849338531494, "learning_rate": 0.0001, "elbo": 3.846, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.101043865822191}
{"step": 18780, "loss": 3.859166717529297, "grad_norm": 0.6971591114997864, "learning_rate": 0.0001, "elbo": 3.85917, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1013370865587615}
{"step": 18785, "loss": 3.825700378417969, "grad_norm": 0.7032753229141235, "learning_rate": 0.0001, "elbo": 3.8257, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.101630307295332}
{"step": 18790, "loss": 3.833129119873047, "grad_norm": 0.667855978012085, "learning_rate": 0.0001, "elbo": 3.83313, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1019235280319024}
{"step": 18795, "loss": 3.837051010131836, "grad_norm": 0.753889262676239, "learning_rate": 0.0001, "elbo": 3.83705, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1022167487684729}
{"step": 18800, "loss": 3.8365463256835937, "grad_norm": 1.2120850086212158, "learning_rate": 0.0001, "elbo": 3.83655, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1025099695050433}
{"step": 18805, "loss": 3.82626953125, "grad_norm": 0.7293087244033813, "learning_rate": 0.0001, "elbo": 3.82627, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1028031902416138}
{"step": 18810, "loss": 3.83704833984375, "grad_norm": 0.8501163125038147, "learning_rate": 0.0001, "elbo": 3.83705, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1030964109781844}
{"step": 18815, "loss": 3.8386131286621095, "grad_norm": 0.7388918399810791, "learning_rate": 0.0001, "elbo": 3.83861, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.103389631714755}
{"step": 18820, "loss": 3.82779541015625, "grad_norm": 0.6787765026092529, "learning_rate": 0.0001, "elbo": 3.8278, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1036828524513254}
{"step": 18825, "loss": 3.8304954528808595, "grad_norm": 0.675089418888092, "learning_rate": 0.0001, "elbo": 3.8305, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1039760731878958}
{"step": 18830, "loss": 3.841924285888672, "grad_norm": 0.7223823666572571, "learning_rate": 0.0001, "elbo": 3.84192, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1042692939244663}
{"step": 18835, "loss": 3.8368370056152346, "grad_norm": 0.69916832447052, "learning_rate": 0.0001, "elbo": 3.83684, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.104562514661037}
{"step": 18840, "loss": 3.8353981018066405, "grad_norm": 0.7314755916595459, "learning_rate": 0.0001, "elbo": 3.8354, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1048557353976074}
{"step": 18845, "loss": 3.843212890625, "grad_norm": 0.7304558753967285, "learning_rate": 0.0001, "elbo": 3.84321, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1051489561341779}
{"step": 18850, "loss": 3.8592147827148438, "grad_norm": 0.7021799087524414, "learning_rate": 0.0001, "elbo": 3.85921, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1054421768707483}
{"step": 18855, "loss": 3.823064422607422, "grad_norm": 0.7145794034004211, "learning_rate": 0.0001, "elbo": 3.82306, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1057353976073188}
{"step": 18860, "loss": 3.8508506774902345, "grad_norm": 0.6938574314117432, "learning_rate": 0.0001, "elbo": 3.85085, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1060286183438892}
{"step": 18865, "loss": 3.834788513183594, "grad_norm": 0.7266599535942078, "learning_rate": 0.0001, "elbo": 3.83479, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1063218390804597}
{"step": 18870, "loss": 3.833175277709961, "grad_norm": 1.1430078744888306, "learning_rate": 0.0001, "elbo": 3.83318, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1066150598170303}
{"step": 18875, "loss": 3.836609649658203, "grad_norm": 0.7096647024154663, "learning_rate": 0.0001, "elbo": 3.83661, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1069082805536008}
{"step": 18880, "loss": 3.8442462921142577, "grad_norm": 0.6855764389038086, "learning_rate": 0.0001, "elbo": 3.84425, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1072015012901713}
{"step": 18885, "loss": 3.8318855285644533, "grad_norm": 0.705344021320343, "learning_rate": 0.0001, "elbo": 3.83189, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1074947220267417}
{"step": 18890, "loss": 3.848341369628906, "grad_norm": 0.6938698887825012, "learning_rate": 0.0001, "elbo": 3.84834, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1077879427633122}
{"step": 18895, "loss": 3.8315406799316407, "grad_norm": 0.6928212642669678, "learning_rate": 0.0001, "elbo": 3.83154, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1080811634998826}
{"step": 18900, "loss": 3.838389587402344, "grad_norm": 0.6738962531089783, "learning_rate": 0.0001, "elbo": 3.83839, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1083743842364533}
{"step": 18905, "loss": 3.831182098388672, "grad_norm": 0.6925765872001648, "learning_rate": 0.0001, "elbo": 3.83118, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1086676049730237}
{"step": 18910, "loss": 3.838495635986328, "grad_norm": 0.7196712493896484, "learning_rate": 0.0001, "elbo": 3.8385, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1089608257095942}
{"step": 18915, "loss": 3.8561050415039064, "grad_norm": 0.6578316688537598, "learning_rate": 0.0001, "elbo": 3.8561, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1092540464461647}
{"step": 18920, "loss": 3.83276481628418, "grad_norm": 0.6883190870285034, "learning_rate": 0.0001, "elbo": 3.83276, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.109547267182735}
{"step": 18925, "loss": 3.835715103149414, "grad_norm": 0.7218235731124878, "learning_rate": 0.0001, "elbo": 3.83572, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1098404879193056}
{"step": 18930, "loss": 3.8275428771972657, "grad_norm": 0.7476358413696289, "learning_rate": 0.0001, "elbo": 3.82754, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1101337086558762}
{"step": 18935, "loss": 3.8478790283203126, "grad_norm": 0.6560564637184143, "learning_rate": 0.0001, "elbo": 3.84788, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1104269293924467}
{"step": 18940, "loss": 3.8189437866210936, "grad_norm": 0.7686960697174072, "learning_rate": 0.0001, "elbo": 3.81894, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1107201501290171}
{"step": 18945, "loss": 3.8423835754394533, "grad_norm": 0.7916505932807922, "learning_rate": 0.0001, "elbo": 3.84238, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1110133708655876}
{"step": 18950, "loss": 3.8440673828125, "grad_norm": 0.8420120477676392, "learning_rate": 0.0001, "elbo": 3.84407, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.111306591602158}
{"step": 18955, "loss": 3.851271057128906, "grad_norm": 0.7109711766242981, "learning_rate": 0.0001, "elbo": 3.85127, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1115998123387285}
{"step": 18960, "loss": 3.846333312988281, "grad_norm": 0.6912906765937805, "learning_rate": 0.0001, "elbo": 3.84633, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1118930330752992}
{"step": 18965, "loss": 3.8410240173339845, "grad_norm": 0.70748370885849, "learning_rate": 0.0001, "elbo": 3.84102, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1121862538118696}
{"step": 18965, "loss": 3.8410240173339845, "grad_norm": 0.70748370885849, "learning_rate": 0.0001, "elbo": 3.84102, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1121862538118696}
{"step": 18970, "loss": 3.8405483245849608, "grad_norm": 0.675610363483429, "learning_rate": 0.0001, "elbo": 3.84055, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.11247947454844}
{"step": 18975, "loss": 3.8451988220214846, "grad_norm": 0.7048308849334717, "learning_rate": 0.0001, "elbo": 3.8452, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1127726952850105}
{"step": 18980, "loss": 3.826559829711914, "grad_norm": 0.7006146311759949, "learning_rate": 0.0001, "elbo": 3.82656, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.113065916021581}
{"step": 18985, "loss": 3.83367805480957, "grad_norm": 0.6830469965934753, "learning_rate": 0.0001, "elbo": 3.83368, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1133591367581515}
{"step": 18990, "loss": 3.834971237182617, "grad_norm": 0.818397045135498, "learning_rate": 0.0001, "elbo": 3.83497, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.113652357494722}
{"step": 18995, "loss": 3.8472305297851563, "grad_norm": 0.7343406081199646, "learning_rate": 0.0001, "elbo": 3.84723, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1139455782312926}
{"step": 19000, "loss": 3.827623748779297, "grad_norm": 0.6916991472244263, "learning_rate": 0.0001, "elbo": 3.82762, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.114238798967863}
{"step": 19005, "loss": 3.8308006286621095, "grad_norm": 0.7216508388519287, "learning_rate": 0.0001, "elbo": 3.8308, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1145320197044335}
{"step": 19010, "loss": 3.8334396362304686, "grad_norm": 0.7746641635894775, "learning_rate": 0.0001, "elbo": 3.83344, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.114825240441004}
{"step": 19015, "loss": 3.8150894165039064, "grad_norm": 0.676920473575592, "learning_rate": 0.0001, "elbo": 3.81509, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1151184611775744}
{"step": 19020, "loss": 3.828631591796875, "grad_norm": 0.7363309860229492, "learning_rate": 0.0001, "elbo": 3.82863, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.115411681914145}
{"step": 19025, "loss": 3.8370033264160157, "grad_norm": 0.9229705333709717, "learning_rate": 0.0001, "elbo": 3.837, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1157049026507155}
{"step": 19030, "loss": 3.8387451171875, "grad_norm": 0.6999080777168274, "learning_rate": 0.0001, "elbo": 3.83875, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.115998123387286}
{"step": 19035, "loss": 3.835590362548828, "grad_norm": 0.7354297041893005, "learning_rate": 0.0001, "elbo": 3.83559, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1162913441238564}
{"step": 19040, "loss": 3.8451683044433596, "grad_norm": 0.7052031755447388, "learning_rate": 0.0001, "elbo": 3.84517, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.116584564860427}
{"step": 19045, "loss": 3.833284378051758, "grad_norm": 0.7170555591583252, "learning_rate": 0.0001, "elbo": 3.83328, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1168777855969974}
{"step": 19050, "loss": 3.8232295989990233, "grad_norm": 0.6873125433921814, "learning_rate": 0.0001, "elbo": 3.82323, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1171710063335678}
{"step": 19055, "loss": 3.8467552185058596, "grad_norm": 0.7864633798599243, "learning_rate": 0.0001, "elbo": 3.84676, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1174642270701385}
{"step": 19060, "loss": 3.822254943847656, "grad_norm": 0.9082793593406677, "learning_rate": 0.0001, "elbo": 3.82226, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.117757447806709}
{"step": 19065, "loss": 3.842889404296875, "grad_norm": 0.758451521396637, "learning_rate": 0.0001, "elbo": 3.84289, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1180506685432794}
{"step": 19070, "loss": 3.833278274536133, "grad_norm": 0.8541268110275269, "learning_rate": 0.0001, "elbo": 3.83328, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1183438892798498}
{"step": 19075, "loss": 3.840367889404297, "grad_norm": 0.6596163511276245, "learning_rate": 0.0001, "elbo": 3.84037, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1186371100164203}
{"step": 19080, "loss": 3.8317909240722656, "grad_norm": 0.6751783490180969, "learning_rate": 0.0001, "elbo": 3.83179, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1189303307529908}
{"step": 19085, "loss": 3.837080383300781, "grad_norm": 0.7005864977836609, "learning_rate": 0.0001, "elbo": 3.83708, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1192235514895614}
{"step": 19090, "loss": 3.8228240966796876, "grad_norm": 0.7238659858703613, "learning_rate": 0.0001, "elbo": 3.82282, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1195167722261319}
{"step": 19095, "loss": 3.8349609375, "grad_norm": 0.7011301517486572, "learning_rate": 0.0001, "elbo": 3.83496, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1198099929627023}
{"step": 19100, "loss": 3.834206390380859, "grad_norm": 0.7299273014068604, "learning_rate": 0.0001, "elbo": 3.83421, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1201032136992728}
{"step": 19105, "loss": 3.825643539428711, "grad_norm": 0.6349929571151733, "learning_rate": 0.0001, "elbo": 3.82564, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1203964344358432}
{"step": 19110, "loss": 3.8358787536621093, "grad_norm": 0.7322165966033936, "learning_rate": 0.0001, "elbo": 3.83588, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1206896551724137}
{"step": 19115, "loss": 3.8486663818359377, "grad_norm": 0.644069254398346, "learning_rate": 0.0001, "elbo": 3.84867, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1209828759089844}
{"step": 19120, "loss": 3.822528839111328, "grad_norm": 0.6395003795623779, "learning_rate": 0.0001, "elbo": 3.82253, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1212760966455548}
{"step": 19125, "loss": 3.833849334716797, "grad_norm": 0.7334453463554382, "learning_rate": 0.0001, "elbo": 3.83385, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1215693173821253}
{"step": 19125, "loss": 3.833849334716797, "grad_norm": 0.7334453463554382, "learning_rate": 0.0001, "elbo": 3.83385, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1215693173821253}
{"step": 19130, "loss": 3.8324893951416015, "grad_norm": 0.7190205454826355, "learning_rate": 0.0001, "elbo": 3.83249, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1218625381186957}
{"step": 19135, "loss": 3.84766845703125, "grad_norm": 0.6899622678756714, "learning_rate": 0.0001, "elbo": 3.84767, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1221557588552662}
{"step": 19140, "loss": 3.842823028564453, "grad_norm": 0.6702426075935364, "learning_rate": 0.0001, "elbo": 3.84282, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1224489795918366}
{"step": 19145, "loss": 3.8376808166503906, "grad_norm": 0.8364137411117554, "learning_rate": 0.0001, "elbo": 3.83768, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1227422003284073}
{"step": 19150, "loss": 3.853820037841797, "grad_norm": 0.9789928793907166, "learning_rate": 0.0001, "elbo": 3.85382, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1230354210649778}
{"step": 19155, "loss": 3.8375186920166016, "grad_norm": 0.6647722125053406, "learning_rate": 0.0001, "elbo": 3.83752, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1233286418015482}
{"step": 19160, "loss": 3.852131652832031, "grad_norm": 0.8556646704673767, "learning_rate": 0.0001, "elbo": 3.85213, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1236218625381187}
{"step": 19165, "loss": 3.8234153747558595, "grad_norm": 0.6547866463661194, "learning_rate": 0.0001, "elbo": 3.82342, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1239150832746891}
{"step": 19170, "loss": 3.835802459716797, "grad_norm": 0.7250369191169739, "learning_rate": 0.0001, "elbo": 3.8358, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1242083040112596}
{"step": 19175, "loss": 3.830463409423828, "grad_norm": 0.6631318926811218, "learning_rate": 0.0001, "elbo": 3.83046, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.12450152474783}
{"step": 19180, "loss": 3.8371646881103514, "grad_norm": 0.6869914531707764, "learning_rate": 0.0001, "elbo": 3.83716, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1247947454844007}
{"step": 19185, "loss": 3.8132877349853516, "grad_norm": 0.7400699853897095, "learning_rate": 0.0001, "elbo": 3.81329, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1250879662209712}
{"step": 19190, "loss": 3.8287464141845704, "grad_norm": 0.69834965467453, "learning_rate": 0.0001, "elbo": 3.82875, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1253811869575416}
{"step": 19195, "loss": 3.8184886932373048, "grad_norm": 0.7611233592033386, "learning_rate": 0.0001, "elbo": 3.81849, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.125674407694112}
{"step": 19200, "loss": 3.835657501220703, "grad_norm": 0.8232344388961792, "learning_rate": 0.0001, "elbo": 3.83566, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1259676284306825}
{"step": 19205, "loss": 3.831922912597656, "grad_norm": 0.7040378451347351, "learning_rate": 0.0001, "elbo": 3.83192, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1262608491672532}
{"step": 19210, "loss": 3.825379180908203, "grad_norm": 0.7027768492698669, "learning_rate": 0.0001, "elbo": 3.82538, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1265540699038237}
{"step": 19215, "loss": 3.843667984008789, "grad_norm": 0.7161503434181213, "learning_rate": 0.0001, "elbo": 3.84367, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1268472906403941}
{"step": 19220, "loss": 3.8246734619140623, "grad_norm": 0.7305541038513184, "learning_rate": 0.0001, "elbo": 3.82467, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1271405113769646}
{"step": 19225, "loss": 3.844287872314453, "grad_norm": 0.7128732204437256, "learning_rate": 0.0001, "elbo": 3.84429, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.127433732113535}
{"step": 19230, "loss": 3.829418182373047, "grad_norm": 0.6955892443656921, "learning_rate": 0.0001, "elbo": 3.82942, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1277269528501055}
{"step": 19235, "loss": 3.835704040527344, "grad_norm": 0.7595785856246948, "learning_rate": 0.0001, "elbo": 3.8357, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.128020173586676}
{"step": 19240, "loss": 3.8434219360351562, "grad_norm": 0.7110361456871033, "learning_rate": 0.0001, "elbo": 3.84342, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1283133943232466}
{"step": 19245, "loss": 3.826000213623047, "grad_norm": 0.808059573173523, "learning_rate": 0.0001, "elbo": 3.826, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.128606615059817}
{"step": 19250, "loss": 3.834613037109375, "grad_norm": 0.7815374732017517, "learning_rate": 0.0001, "elbo": 3.83461, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1288998357963875}
{"step": 19255, "loss": 3.831372833251953, "grad_norm": 0.7514182925224304, "learning_rate": 0.0001, "elbo": 3.83137, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.129193056532958}
{"step": 19255, "loss": 3.831372833251953, "grad_norm": 0.7514182925224304, "learning_rate": 0.0001, "elbo": 3.83137, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.129193056532958}
{"step": 19260, "loss": 3.832304000854492, "grad_norm": 0.8024806976318359, "learning_rate": 0.0001, "elbo": 3.8323, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1294862772695284}
{"step": 19265, "loss": 3.836502456665039, "grad_norm": 0.7537384629249573, "learning_rate": 0.0001, "elbo": 3.8365, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1297794980060991}
{"step": 19270, "loss": 3.821453857421875, "grad_norm": 0.7505369782447815, "learning_rate": 0.0001, "elbo": 3.82145, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1300727187426696}
{"step": 19275, "loss": 3.829020309448242, "grad_norm": 0.7304468154907227, "learning_rate": 0.0001, "elbo": 3.82902, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.13036593947924}
{"step": 19280, "loss": 3.8187114715576174, "grad_norm": 0.676576554775238, "learning_rate": 0.0001, "elbo": 3.81871, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1306591602158105}
{"step": 19285, "loss": 3.8315574645996096, "grad_norm": 0.6875870227813721, "learning_rate": 0.0001, "elbo": 3.83156, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.130952380952381}
{"step": 19290, "loss": 3.8323314666748045, "grad_norm": 0.6895232796669006, "learning_rate": 0.0001, "elbo": 3.83233, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1312456016889514}
{"step": 19295, "loss": 3.8231861114501955, "grad_norm": 0.7548500299453735, "learning_rate": 0.0001, "elbo": 3.82319, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1315388224255218}
{"step": 19300, "loss": 3.8321155548095702, "grad_norm": 0.898054838180542, "learning_rate": 0.0001, "elbo": 3.83212, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1318320431620925}
{"step": 19305, "loss": 3.8166603088378905, "grad_norm": 0.6791539192199707, "learning_rate": 0.0001, "elbo": 3.81666, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.132125263898663}
{"step": 19310, "loss": 3.8490520477294923, "grad_norm": 0.8045182228088379, "learning_rate": 0.0001, "elbo": 3.84905, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1324184846352334}
{"step": 19315, "loss": 3.8297348022460938, "grad_norm": 0.6522226333618164, "learning_rate": 0.0001, "elbo": 3.82974, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1327117053718039}
{"step": 19320, "loss": 3.814768981933594, "grad_norm": 0.7196148633956909, "learning_rate": 0.0001, "elbo": 3.81477, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1330049261083743}
{"step": 19320, "loss": 3.814768981933594, "grad_norm": 0.7196148633956909, "learning_rate": 0.0001, "elbo": 3.81477, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1330049261083743}
{"step": 19325, "loss": 3.8431915283203124, "grad_norm": 0.7110851407051086, "learning_rate": 0.0001, "elbo": 3.84319, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1332981468449448}
{"step": 19330, "loss": 3.8358715057373045, "grad_norm": 0.7061552405357361, "learning_rate": 0.0001, "elbo": 3.83587, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1335913675815155}
{"step": 19330, "loss": 3.8358715057373045, "grad_norm": 0.7061552405357361, "learning_rate": 0.0001, "elbo": 3.83587, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1335913675815155}
{"step": 19335, "loss": 3.8380363464355467, "grad_norm": 0.6776752471923828, "learning_rate": 0.0001, "elbo": 3.83804, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.133884588318086}
{"step": 19340, "loss": 3.8178146362304686, "grad_norm": 0.8463252782821655, "learning_rate": 0.0001, "elbo": 3.81781, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1341778090546564}
{"step": 19345, "loss": 3.834586334228516, "grad_norm": 0.7923606634140015, "learning_rate": 0.0001, "elbo": 3.83459, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1344710297912268}
{"step": 19350, "loss": 3.8319061279296873, "grad_norm": 0.6953635215759277, "learning_rate": 0.0001, "elbo": 3.83191, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1347642505277973}
{"step": 19355, "loss": 3.8190166473388674, "grad_norm": 0.7051296234130859, "learning_rate": 0.0001, "elbo": 3.81902, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1350574712643677}
{"step": 19360, "loss": 3.829182434082031, "grad_norm": 0.6852736473083496, "learning_rate": 0.0001, "elbo": 3.82918, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1353506920009382}
{"step": 19365, "loss": 3.834666442871094, "grad_norm": 0.6461421251296997, "learning_rate": 0.0001, "elbo": 3.83467, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1356439127375089}
{"step": 19370, "loss": 3.8097763061523438, "grad_norm": 0.6963369846343994, "learning_rate": 0.0001, "elbo": 3.80978, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1359371334740793}
{"step": 19375, "loss": 3.827586364746094, "grad_norm": 0.8192070126533508, "learning_rate": 0.0001, "elbo": 3.82759, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1362303542106498}
{"step": 19380, "loss": 3.8262474060058596, "grad_norm": 0.7217029929161072, "learning_rate": 0.0001, "elbo": 3.82625, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1365235749472202}
{"step": 19385, "loss": 3.8274459838867188, "grad_norm": 0.6894985437393188, "learning_rate": 0.0001, "elbo": 3.82745, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1368167956837907}
{"step": 19390, "loss": 3.8327667236328127, "grad_norm": 0.7223625779151917, "learning_rate": 0.0001, "elbo": 3.83277, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1371100164203614}
{"step": 19395, "loss": 3.829037094116211, "grad_norm": 0.6736862063407898, "learning_rate": 0.0001, "elbo": 3.82904, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1374032371569318}
{"step": 19400, "loss": 3.8256294250488283, "grad_norm": 0.7498918175697327, "learning_rate": 0.0001, "elbo": 3.82563, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1376964578935023}
{"step": 19405, "loss": 3.842659759521484, "grad_norm": 0.7438465356826782, "learning_rate": 0.0001, "elbo": 3.84266, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1379896786300727}
{"step": 19410, "loss": 3.8289459228515623, "grad_norm": 0.7405805587768555, "learning_rate": 0.0001, "elbo": 3.82895, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1382828993666432}
{"step": 19415, "loss": 3.816761779785156, "grad_norm": 0.6757308840751648, "learning_rate": 0.0001, "elbo": 3.81676, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1385761201032136}
{"step": 19420, "loss": 3.8189765930175783, "grad_norm": 0.6830000281333923, "learning_rate": 0.0001, "elbo": 3.81898, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.138869340839784}
{"step": 19425, "loss": 3.808042526245117, "grad_norm": 0.7501340508460999, "learning_rate": 0.0001, "elbo": 3.80804, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1391625615763548}
{"step": 19430, "loss": 3.828318786621094, "grad_norm": 0.6746347546577454, "learning_rate": 0.0001, "elbo": 3.82832, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1394557823129252}
{"step": 19435, "loss": 3.818614959716797, "grad_norm": 0.7020341157913208, "learning_rate": 0.0001, "elbo": 3.81862, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1397490030494957}
{"step": 19440, "loss": 3.8086048126220704, "grad_norm": 0.7337288856506348, "learning_rate": 0.0001, "elbo": 3.8086, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1400422237860661}
{"step": 19445, "loss": 3.821340560913086, "grad_norm": 0.7316522598266602, "learning_rate": 0.0001, "elbo": 3.82134, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1403354445226366}
{"step": 19450, "loss": 3.824609375, "grad_norm": 0.658837616443634, "learning_rate": 0.0001, "elbo": 3.82461, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1406286652592073}
{"step": 19455, "loss": 3.827433776855469, "grad_norm": 0.6611922979354858, "learning_rate": 0.0001, "elbo": 3.82743, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1409218859957777}
{"step": 19460, "loss": 3.8275257110595704, "grad_norm": 0.6784418225288391, "learning_rate": 0.0001, "elbo": 3.82753, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1412151067323482}
{"step": 19465, "loss": 3.8369293212890625, "grad_norm": 0.6729750037193298, "learning_rate": 0.0001, "elbo": 3.83693, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1415083274689186}
{"step": 19470, "loss": 3.8196258544921875, "grad_norm": 0.7128534317016602, "learning_rate": 0.0001, "elbo": 3.81963, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.141801548205489}
{"step": 19475, "loss": 3.8342185974121095, "grad_norm": 0.6617215275764465, "learning_rate": 0.0001, "elbo": 3.83422, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1420947689420595}
{"step": 19480, "loss": 3.829376220703125, "grad_norm": 1.107736587524414, "learning_rate": 0.0001, "elbo": 3.82938, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.14238798967863}
{"step": 19485, "loss": 3.8224979400634767, "grad_norm": 0.7144310474395752, "learning_rate": 0.0001, "elbo": 3.8225, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1426812104152007}
{"step": 19490, "loss": 3.8262264251708986, "grad_norm": 0.8592774271965027, "learning_rate": 0.0001, "elbo": 3.82623, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.142974431151771}
{"step": 19495, "loss": 3.8217323303222654, "grad_norm": 0.668068528175354, "learning_rate": 0.0001, "elbo": 3.82173, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1432676518883416}
{"step": 19500, "loss": 3.8306697845458983, "grad_norm": 1.1383073329925537, "learning_rate": 0.0001, "elbo": 3.83067, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.143560872624912}
{"step": 19505, "loss": 3.820512390136719, "grad_norm": 0.7005389332771301, "learning_rate": 0.0001, "elbo": 3.82051, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1438540933614825}
{"step": 19510, "loss": 3.8337310791015624, "grad_norm": 0.7319592833518982, "learning_rate": 0.0001, "elbo": 3.83373, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.144147314098053}
{"step": 19515, "loss": 3.826990509033203, "grad_norm": 0.7078090906143188, "learning_rate": 0.0001, "elbo": 3.82699, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1444405348346236}
{"step": 19520, "loss": 3.821340560913086, "grad_norm": 0.7123118042945862, "learning_rate": 0.0001, "elbo": 3.82134, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.144733755571194}
{"step": 19525, "loss": 3.833308792114258, "grad_norm": 0.6458553671836853, "learning_rate": 0.0001, "elbo": 3.83331, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1450269763077645}
{"step": 19530, "loss": 3.8182903289794923, "grad_norm": 0.6791799664497375, "learning_rate": 0.0001, "elbo": 3.81829, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.145320197044335}
{"step": 19535, "loss": 3.833417510986328, "grad_norm": 0.7101403474807739, "learning_rate": 0.0001, "elbo": 3.83342, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1456134177809054}
{"step": 19540, "loss": 3.8281585693359377, "grad_norm": 0.6759793162345886, "learning_rate": 0.0001, "elbo": 3.82816, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1459066385174759}
{"step": 19545, "loss": 3.822694778442383, "grad_norm": 0.6416570544242859, "learning_rate": 0.0001, "elbo": 3.82269, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1461998592540463}
{"step": 19550, "loss": 3.8189937591552736, "grad_norm": 0.692954421043396, "learning_rate": 0.0001, "elbo": 3.81899, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.146493079990617}
{"step": 19555, "loss": 3.840887451171875, "grad_norm": 0.662156879901886, "learning_rate": 0.0001, "elbo": 3.84089, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1467863007271875}
{"step": 19560, "loss": 3.8192184448242186, "grad_norm": 0.6386316418647766, "learning_rate": 0.0001, "elbo": 3.81922, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.147079521463758}
{"step": 19565, "loss": 3.837789535522461, "grad_norm": 0.715241551399231, "learning_rate": 0.0001, "elbo": 3.83779, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1473727422003284}
{"step": 19570, "loss": 3.810716247558594, "grad_norm": 0.6899678707122803, "learning_rate": 0.0001, "elbo": 3.81072, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1476659629368988}
{"step": 19575, "loss": 3.8245773315429688, "grad_norm": 0.6709190011024475, "learning_rate": 0.0001, "elbo": 3.82458, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1479591836734695}
{"step": 19580, "loss": 3.815079116821289, "grad_norm": 0.7279835343360901, "learning_rate": 0.0001, "elbo": 3.81508, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.14825240441004}
{"step": 19585, "loss": 3.8313453674316404, "grad_norm": 0.6816909909248352, "learning_rate": 0.0001, "elbo": 3.83135, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1485456251466104}
{"step": 19590, "loss": 3.8213409423828124, "grad_norm": 0.734442949295044, "learning_rate": 0.0001, "elbo": 3.82134, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1488388458831809}
{"step": 19595, "loss": 3.825426483154297, "grad_norm": 0.673926055431366, "learning_rate": 0.0001, "elbo": 3.82543, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1491320666197513}
{"step": 19600, "loss": 3.825628662109375, "grad_norm": 0.6692957878112793, "learning_rate": 0.0001, "elbo": 3.82563, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1494252873563218}
{"step": 19605, "loss": 3.8369186401367186, "grad_norm": 1.1846837997436523, "learning_rate": 0.0001, "elbo": 3.83692, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1497185080928922}
{"step": 19610, "loss": 3.831891632080078, "grad_norm": 0.6891946792602539, "learning_rate": 0.0001, "elbo": 3.83189, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.150011728829463}
{"step": 19615, "loss": 3.8467391967773437, "grad_norm": 0.6812756657600403, "learning_rate": 0.0001, "elbo": 3.84674, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1503049495660334}
{"step": 19620, "loss": 3.843374252319336, "grad_norm": 0.6740656495094299, "learning_rate": 0.0001, "elbo": 3.84337, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1505981703026038}
{"step": 19620, "loss": 3.843374252319336, "grad_norm": 0.6740656495094299, "learning_rate": 0.0001, "elbo": 3.84337, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1505981703026038}
{"step": 19625, "loss": 3.834112548828125, "grad_norm": 0.6946061849594116, "learning_rate": 0.0001, "elbo": 3.83411, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1508913910391743}
{"step": 19630, "loss": 3.839605712890625, "grad_norm": 0.733407199382782, "learning_rate": 0.0001, "elbo": 3.83961, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1511846117757447}
{"step": 19635, "loss": 3.8412395477294923, "grad_norm": 0.7097974419593811, "learning_rate": 0.0001, "elbo": 3.84124, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1514778325123154}
{"step": 19640, "loss": 3.816656494140625, "grad_norm": 0.7309871315956116, "learning_rate": 0.0001, "elbo": 3.81666, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1517710532488858}
{"step": 19645, "loss": 3.8055599212646483, "grad_norm": 0.7161581516265869, "learning_rate": 0.0001, "elbo": 3.80556, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1520642739854563}
{"step": 19650, "loss": 3.8272918701171874, "grad_norm": 0.739627480506897, "learning_rate": 0.0001, "elbo": 3.82729, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1523574947220268}
{"step": 19655, "loss": 3.814256286621094, "grad_norm": 0.6899858117103577, "learning_rate": 0.0001, "elbo": 3.81426, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1526507154585972}
{"step": 19660, "loss": 3.8287822723388674, "grad_norm": 0.7359541058540344, "learning_rate": 0.0001, "elbo": 3.82878, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1529439361951677}
{"step": 19665, "loss": 3.829844665527344, "grad_norm": 0.6764825582504272, "learning_rate": 0.0001, "elbo": 3.82984, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1532371569317381}
{"step": 19670, "loss": 3.813872146606445, "grad_norm": 0.669668436050415, "learning_rate": 0.0001, "elbo": 3.81387, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1535303776683088}
{"step": 19675, "loss": 3.8482666015625, "grad_norm": 0.6827437877655029, "learning_rate": 0.0001, "elbo": 3.84827, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1538235984048792}
{"step": 19680, "loss": 3.8159080505371095, "grad_norm": 0.7020842432975769, "learning_rate": 0.0001, "elbo": 3.81591, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1541168191414497}
{"step": 19685, "loss": 3.8206085205078124, "grad_norm": 0.6928631067276001, "learning_rate": 0.0001, "elbo": 3.82061, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1544100398780202}
{"step": 19690, "loss": 3.825811767578125, "grad_norm": 0.7584378719329834, "learning_rate": 0.0001, "elbo": 3.82581, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1547032606145906}
{"step": 19695, "loss": 3.8312713623046877, "grad_norm": 0.6716524958610535, "learning_rate": 0.0001, "elbo": 3.83127, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.154996481351161}
{"step": 19700, "loss": 3.828485870361328, "grad_norm": 0.7422614097595215, "learning_rate": 0.0001, "elbo": 3.82849, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1552897020877317}
{"step": 19705, "loss": 3.8180885314941406, "grad_norm": 0.6975295543670654, "learning_rate": 0.0001, "elbo": 3.81809, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1555829228243022}
{"step": 19710, "loss": 3.825823974609375, "grad_norm": 1.0005375146865845, "learning_rate": 0.0001, "elbo": 3.82582, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1558761435608726}
{"step": 19715, "loss": 3.845772933959961, "grad_norm": 0.6668197512626648, "learning_rate": 0.0001, "elbo": 3.84577, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.156169364297443}
{"step": 19720, "loss": 3.8275077819824217, "grad_norm": 0.757121205329895, "learning_rate": 0.0001, "elbo": 3.82751, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1564625850340136}
{"step": 19725, "loss": 3.8142169952392577, "grad_norm": 0.8762376308441162, "learning_rate": 0.0001, "elbo": 3.81422, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.156755805770584}
{"step": 19725, "loss": 3.8142169952392577, "grad_norm": 0.8762376308441162, "learning_rate": 0.0001, "elbo": 3.81422, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.156755805770584}
{"step": 19730, "loss": 3.823078155517578, "grad_norm": 0.7345219254493713, "learning_rate": 0.0001, "elbo": 3.82308, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1570490265071545}
{"step": 19735, "loss": 3.8211734771728514, "grad_norm": 0.6558542847633362, "learning_rate": 0.0001, "elbo": 3.82117, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1573422472437251}
{"step": 19740, "loss": 3.835828399658203, "grad_norm": 0.7874396443367004, "learning_rate": 0.0001, "elbo": 3.83583, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1576354679802956}
{"step": 19745, "loss": 3.8404518127441407, "grad_norm": 0.7131308913230896, "learning_rate": 0.0001, "elbo": 3.84045, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.157928688716866}
{"step": 19745, "loss": 3.8404518127441407, "grad_norm": 0.7131308913230896, "learning_rate": 0.0001, "elbo": 3.84045, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.157928688716866}
{"step": 19750, "loss": 3.830327606201172, "grad_norm": 0.6788878440856934, "learning_rate": 0.0001, "elbo": 3.83033, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1582219094534365}
{"step": 19755, "loss": 3.809652328491211, "grad_norm": 0.7169061899185181, "learning_rate": 0.0001, "elbo": 3.80965, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.158515130190007}
{"step": 19760, "loss": 3.8154720306396483, "grad_norm": 1.2040983438491821, "learning_rate": 0.0001, "elbo": 3.81547, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1588083509265776}
{"step": 19765, "loss": 3.8248992919921876, "grad_norm": 0.7246360182762146, "learning_rate": 0.0001, "elbo": 3.8249, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.159101571663148}
{"step": 19770, "loss": 3.8157623291015623, "grad_norm": 0.7417458295822144, "learning_rate": 0.0001, "elbo": 3.81576, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1593947923997185}
{"step": 19775, "loss": 3.818334197998047, "grad_norm": 0.9842625260353088, "learning_rate": 0.0001, "elbo": 3.81833, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.159688013136289}
{"step": 19780, "loss": 3.83459587097168, "grad_norm": 0.8422461152076721, "learning_rate": 0.0001, "elbo": 3.8346, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1599812338728595}
{"step": 19785, "loss": 3.8125282287597657, "grad_norm": 0.7445712089538574, "learning_rate": 0.0001, "elbo": 3.81253, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.16027445460943}
{"step": 19790, "loss": 3.8094043731689453, "grad_norm": 0.8131932616233826, "learning_rate": 0.0001, "elbo": 3.8094, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1605676753460004}
{"step": 19795, "loss": 3.8323738098144533, "grad_norm": 0.7059559226036072, "learning_rate": 0.0001, "elbo": 3.83237, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.160860896082571}
{"step": 19800, "loss": 3.826487350463867, "grad_norm": 0.7110192775726318, "learning_rate": 0.0001, "elbo": 3.82649, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1611541168191415}
{"step": 19805, "loss": 3.8289539337158205, "grad_norm": 0.7333371639251709, "learning_rate": 0.0001, "elbo": 3.82895, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.161447337555712}
{"step": 19810, "loss": 3.8269889831542967, "grad_norm": 0.6836463212966919, "learning_rate": 0.0001, "elbo": 3.82699, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1617405582922824}
{"step": 19815, "loss": 3.8090118408203124, "grad_norm": 0.746478259563446, "learning_rate": 0.0001, "elbo": 3.80901, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1620337790288529}
{"step": 19820, "loss": 3.8278213500976563, "grad_norm": 0.6515550017356873, "learning_rate": 0.0001, "elbo": 3.82782, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1623269997654235}
{"step": 19825, "loss": 3.816668701171875, "grad_norm": 0.6824467778205872, "learning_rate": 0.0001, "elbo": 3.81667, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.162620220501994}
{"step": 19830, "loss": 3.8224361419677733, "grad_norm": 0.9688439965248108, "learning_rate": 0.0001, "elbo": 3.82244, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1629134412385644}
{"step": 19835, "loss": 3.8388195037841797, "grad_norm": 0.6745648980140686, "learning_rate": 0.0001, "elbo": 3.83882, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.163206661975135}
{"step": 19840, "loss": 3.8081809997558596, "grad_norm": 0.7180274724960327, "learning_rate": 0.0001, "elbo": 3.80818, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1634998827117053}
{"step": 19845, "loss": 3.8263870239257813, "grad_norm": 0.6794613599777222, "learning_rate": 0.0001, "elbo": 3.82639, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1637931034482758}
{"step": 19850, "loss": 3.8267574310302734, "grad_norm": 0.676720142364502, "learning_rate": 0.0001, "elbo": 3.82676, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1640863241848463}
{"step": 19855, "loss": 3.8233005523681642, "grad_norm": 0.9651049375534058, "learning_rate": 0.0001, "elbo": 3.8233, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.164379544921417}
{"step": 19860, "loss": 3.8262027740478515, "grad_norm": 0.7077064514160156, "learning_rate": 0.0001, "elbo": 3.8262, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1646727656579874}
{"step": 19865, "loss": 3.841221237182617, "grad_norm": 0.7043499946594238, "learning_rate": 0.0001, "elbo": 3.84122, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1649659863945578}
{"step": 19870, "loss": 3.824844741821289, "grad_norm": 0.7152588963508606, "learning_rate": 0.0001, "elbo": 3.82484, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1652592071311283}
{"step": 19875, "loss": 3.812693786621094, "grad_norm": 0.708278477191925, "learning_rate": 0.0001, "elbo": 3.81269, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1655524278676987}
{"step": 19880, "loss": 3.8123805999755858, "grad_norm": 0.6663833856582642, "learning_rate": 0.0001, "elbo": 3.81238, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1658456486042692}
{"step": 19885, "loss": 3.8045185089111326, "grad_norm": 0.6665385365486145, "learning_rate": 0.0001, "elbo": 3.80452, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1661388693408399}
{"step": 19890, "loss": 3.8249958038330076, "grad_norm": 1.1496007442474365, "learning_rate": 0.0001, "elbo": 3.825, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1664320900774103}
{"step": 19895, "loss": 3.8323429107666014, "grad_norm": 0.7550389766693115, "learning_rate": 0.0001, "elbo": 3.83234, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1667253108139808}
{"step": 19900, "loss": 3.8154067993164062, "grad_norm": 0.7313151359558105, "learning_rate": 0.0001, "elbo": 3.81541, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1670185315505512}
{"step": 19905, "loss": 3.8152435302734373, "grad_norm": 0.693727433681488, "learning_rate": 0.0001, "elbo": 3.81524, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1673117522871217}
{"step": 19910, "loss": 3.8314414978027345, "grad_norm": 0.6494564414024353, "learning_rate": 0.0001, "elbo": 3.83144, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1676049730236921}
{"step": 19915, "loss": 3.8145347595214845, "grad_norm": 0.7097335457801819, "learning_rate": 0.0001, "elbo": 3.81453, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1678981937602626}
{"step": 19920, "loss": 3.8182167053222655, "grad_norm": 0.7172138690948486, "learning_rate": 0.0001, "elbo": 3.81822, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1681914144968333}
{"step": 19920, "loss": 3.8182167053222655, "grad_norm": 0.7172138690948486, "learning_rate": 0.0001, "elbo": 3.81822, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1681914144968333}
{"step": 19925, "loss": 3.83001823425293, "grad_norm": 0.6590734720230103, "learning_rate": 0.0001, "elbo": 3.83002, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1684846352334037}
{"step": 19930, "loss": 3.825205612182617, "grad_norm": 0.662402331829071, "learning_rate": 0.0001, "elbo": 3.82521, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1687778559699742}
{"step": 19935, "loss": 3.8177764892578123, "grad_norm": 0.678795337677002, "learning_rate": 0.0001, "elbo": 3.81778, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1690710767065446}
{"step": 19940, "loss": 3.818798065185547, "grad_norm": 0.6999260783195496, "learning_rate": 0.0001, "elbo": 3.8188, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.169364297443115}
{"step": 19945, "loss": 3.816553497314453, "grad_norm": 0.6804099082946777, "learning_rate": 0.0001, "elbo": 3.81655, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1696575181796858}
{"step": 19950, "loss": 3.8073333740234374, "grad_norm": 0.67717444896698, "learning_rate": 0.0001, "elbo": 3.80733, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1699507389162562}
{"step": 19955, "loss": 3.8210678100585938, "grad_norm": 0.6813020706176758, "learning_rate": 0.0001, "elbo": 3.82107, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1702439596528267}
{"step": 19960, "loss": 3.8268299102783203, "grad_norm": 0.6901892423629761, "learning_rate": 0.0001, "elbo": 3.82683, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1705371803893971}
{"step": 19965, "loss": 3.7908443450927733, "grad_norm": 0.6773470640182495, "learning_rate": 0.0001, "elbo": 3.79084, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1708304011259676}
{"step": 19970, "loss": 3.8120346069335938, "grad_norm": 0.6382854580879211, "learning_rate": 0.0001, "elbo": 3.81203, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.171123621862538}
{"step": 19975, "loss": 3.826312255859375, "grad_norm": 0.7170923948287964, "learning_rate": 0.0001, "elbo": 3.82631, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1714168425991085}
{"step": 19980, "loss": 3.8369598388671875, "grad_norm": 0.7590786218643188, "learning_rate": 0.0001, "elbo": 3.83696, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1717100633356792}
{"step": 19985, "loss": 3.8337005615234374, "grad_norm": 0.6863396763801575, "learning_rate": 0.0001, "elbo": 3.8337, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1720032840722496}
{"step": 19990, "loss": 3.8428165435791017, "grad_norm": 0.6714357733726501, "learning_rate": 0.0001, "elbo": 3.84282, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.17229650480882}
{"step": 19995, "loss": 3.8158027648925783, "grad_norm": 0.6816105246543884, "learning_rate": 0.0001, "elbo": 3.8158, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1725897255453905}
{"step": 19995, "loss": 3.8158027648925783, "grad_norm": 0.6816105246543884, "learning_rate": 0.0001, "elbo": 3.8158, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1725897255453905}
{"step": 20000, "loss": 3.837377166748047, "grad_norm": 0.6371399164199829, "learning_rate": 0.0001, "elbo": 3.83738, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.172882946281961}
{"step": 20005, "loss": 3.8200302124023438, "grad_norm": 0.69435715675354, "learning_rate": 0.0001, "elbo": 3.82003, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1731761670185317}
{"step": 20010, "loss": 3.826732635498047, "grad_norm": 0.6531537175178528, "learning_rate": 0.0001, "elbo": 3.82673, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1734693877551021}
{"step": 20015, "loss": 3.821258544921875, "grad_norm": 0.6661407947540283, "learning_rate": 0.0001, "elbo": 3.82126, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1737626084916726}
{"step": 20020, "loss": 3.8379100799560546, "grad_norm": 0.7720630764961243, "learning_rate": 0.0001, "elbo": 3.83791, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.174055829228243}
{"step": 20025, "loss": 3.8209457397460938, "grad_norm": 0.6472845077514648, "learning_rate": 0.0001, "elbo": 3.82095, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1743490499648135}
{"step": 20030, "loss": 3.7886512756347654, "grad_norm": 0.6646210551261902, "learning_rate": 0.0001, "elbo": 3.78865, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.174642270701384}
{"step": 20035, "loss": 3.83472900390625, "grad_norm": 0.6831712126731873, "learning_rate": 0.0001, "elbo": 3.83473, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1749354914379544}
{"step": 20040, "loss": 3.8233238220214845, "grad_norm": 0.6258209347724915, "learning_rate": 0.0001, "elbo": 3.82332, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.175228712174525}
{"step": 20045, "loss": 3.8157417297363283, "grad_norm": 0.7063307762145996, "learning_rate": 0.0001, "elbo": 3.81574, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1755219329110955}
{"step": 20050, "loss": 3.829778289794922, "grad_norm": 0.7077937722206116, "learning_rate": 0.0001, "elbo": 3.82978, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.175815153647666}
{"step": 20055, "loss": 3.80542106628418, "grad_norm": 0.7777536511421204, "learning_rate": 0.0001, "elbo": 3.80542, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1761083743842364}
{"step": 20060, "loss": 3.8198017120361327, "grad_norm": 0.7981405854225159, "learning_rate": 0.0001, "elbo": 3.8198, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1764015951208069}
{"step": 20065, "loss": 3.811702346801758, "grad_norm": 0.6692805886268616, "learning_rate": 0.0001, "elbo": 3.8117, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1766948158573773}
{"step": 20070, "loss": 3.8311416625976564, "grad_norm": 0.7440009117126465, "learning_rate": 0.0001, "elbo": 3.83114, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.176988036593948}
{"step": 20075, "loss": 3.8278324127197267, "grad_norm": 0.708858072757721, "learning_rate": 0.0001, "elbo": 3.82783, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1772812573305185}
{"step": 20080, "loss": 3.816456604003906, "grad_norm": 0.7206482291221619, "learning_rate": 0.0001, "elbo": 3.81646, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.177574478067089}
{"step": 20085, "loss": 3.8247215270996096, "grad_norm": 0.7492083311080933, "learning_rate": 0.0001, "elbo": 3.82472, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1778676988036594}
{"step": 20090, "loss": 3.80994873046875, "grad_norm": 0.6322075724601746, "learning_rate": 0.0001, "elbo": 3.80995, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1781609195402298}
{"step": 20095, "loss": 3.818069839477539, "grad_norm": 0.7153269648551941, "learning_rate": 0.0001, "elbo": 3.81807, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1784541402768003}
{"step": 20100, "loss": 3.8210010528564453, "grad_norm": 0.7535156011581421, "learning_rate": 0.0001, "elbo": 3.821, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1787473610133707}
{"step": 20105, "loss": 3.825695037841797, "grad_norm": 0.7056199312210083, "learning_rate": 0.0001, "elbo": 3.82569, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1790405817499414}
{"step": 20110, "loss": 3.810498046875, "grad_norm": 0.7166956663131714, "learning_rate": 0.0001, "elbo": 3.8105, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1793338024865119}
{"step": 20115, "loss": 3.8148780822753907, "grad_norm": 0.874658465385437, "learning_rate": 0.0001, "elbo": 3.81488, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1796270232230823}
{"step": 20120, "loss": 3.822991943359375, "grad_norm": 0.706204354763031, "learning_rate": 0.0001, "elbo": 3.82299, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1799202439596528}
{"step": 20125, "loss": 3.8303531646728515, "grad_norm": 0.6833511590957642, "learning_rate": 0.0001, "elbo": 3.83035, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1802134646962232}
{"step": 20130, "loss": 3.8204910278320314, "grad_norm": 0.6714025139808655, "learning_rate": 0.0001, "elbo": 3.82049, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.180506685432794}
{"step": 20135, "loss": 3.819753646850586, "grad_norm": 0.7632348537445068, "learning_rate": 0.0001, "elbo": 3.81975, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1807999061693644}
{"step": 20140, "loss": 3.8091205596923827, "grad_norm": 0.6998710632324219, "learning_rate": 0.0001, "elbo": 3.80912, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1810931269059348}
{"step": 20145, "loss": 3.8342559814453123, "grad_norm": 0.6760766506195068, "learning_rate": 0.0001, "elbo": 3.83426, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1813863476425053}
{"step": 20150, "loss": 3.8137115478515624, "grad_norm": 0.7095061540603638, "learning_rate": 0.0001, "elbo": 3.81371, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1816795683790757}
{"step": 20155, "loss": 3.7969295501708986, "grad_norm": 0.6586267352104187, "learning_rate": 0.0001, "elbo": 3.79693, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1819727891156462}
{"step": 20160, "loss": 3.8124130249023436, "grad_norm": 0.6554911732673645, "learning_rate": 0.0001, "elbo": 3.81241, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1822660098522166}
{"step": 20165, "loss": 3.8206390380859374, "grad_norm": 0.7528125047683716, "learning_rate": 0.0001, "elbo": 3.82064, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1825592305887873}
{"step": 20170, "loss": 3.8160469055175783, "grad_norm": 0.6824726462364197, "learning_rate": 0.0001, "elbo": 3.81605, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1828524513253578}
{"step": 20175, "loss": 3.7995094299316405, "grad_norm": 0.7133107781410217, "learning_rate": 0.0001, "elbo": 3.79951, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1831456720619282}
{"step": 20180, "loss": 3.8267276763916014, "grad_norm": 0.7837735414505005, "learning_rate": 0.0001, "elbo": 3.82673, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1834388927984987}
{"step": 20185, "loss": 3.822138214111328, "grad_norm": 0.7432383894920349, "learning_rate": 0.0001, "elbo": 3.82214, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1837321135350691}
{"step": 20190, "loss": 3.8052818298339846, "grad_norm": 0.6637492775917053, "learning_rate": 0.0001, "elbo": 3.80528, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1840253342716398}
{"step": 20195, "loss": 3.820755386352539, "grad_norm": 0.7237536907196045, "learning_rate": 0.0001, "elbo": 3.82076, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1843185550082103}
{"step": 20200, "loss": 3.8068252563476563, "grad_norm": 0.6943970918655396, "learning_rate": 0.0001, "elbo": 3.80683, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1846117757447807}
{"step": 20205, "loss": 3.8201278686523437, "grad_norm": 0.6846520900726318, "learning_rate": 0.0001, "elbo": 3.82013, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1849049964813512}
{"step": 20210, "loss": 3.814820098876953, "grad_norm": 0.7699898481369019, "learning_rate": 0.0001, "elbo": 3.81482, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1851982172179216}
{"step": 20215, "loss": 3.8229393005371093, "grad_norm": 0.7168008089065552, "learning_rate": 0.0001, "elbo": 3.82294, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.185491437954492}
{"step": 20220, "loss": 3.806021881103516, "grad_norm": 0.6950657367706299, "learning_rate": 0.0001, "elbo": 3.80602, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1857846586910625}
{"step": 20225, "loss": 3.824560546875, "grad_norm": 0.7055069208145142, "learning_rate": 0.0001, "elbo": 3.82456, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1860778794276332}
{"step": 20230, "loss": 3.796460723876953, "grad_norm": 0.6524198651313782, "learning_rate": 0.0001, "elbo": 3.79646, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1863711001642037}
{"step": 20235, "loss": 3.8066570281982424, "grad_norm": 0.7241226434707642, "learning_rate": 0.0001, "elbo": 3.80666, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1866643209007741}
{"step": 20240, "loss": 3.8058006286621096, "grad_norm": 0.8476890921592712, "learning_rate": 0.0001, "elbo": 3.8058, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1869575416373446}
{"step": 20245, "loss": 3.8324203491210938, "grad_norm": 0.6902894973754883, "learning_rate": 0.0001, "elbo": 3.83242, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.187250762373915}
{"step": 20250, "loss": 3.8272056579589844, "grad_norm": 1.1842583417892456, "learning_rate": 0.0001, "elbo": 3.82721, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1875439831104855}
{"step": 20255, "loss": 3.8152538299560548, "grad_norm": 0.7210407853126526, "learning_rate": 0.0001, "elbo": 3.81525, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1878372038470562}
{"step": 20260, "loss": 3.816092681884766, "grad_norm": 0.6682044267654419, "learning_rate": 0.0001, "elbo": 3.81609, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1881304245836266}
{"step": 20265, "loss": 3.799554443359375, "grad_norm": 0.6467839479446411, "learning_rate": 0.0001, "elbo": 3.79955, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.188423645320197}
{"step": 20270, "loss": 3.812821960449219, "grad_norm": 0.6523164510726929, "learning_rate": 0.0001, "elbo": 3.81282, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1887168660567675}
{"step": 20275, "loss": 3.831534576416016, "grad_norm": 0.7527645230293274, "learning_rate": 0.0001, "elbo": 3.83153, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.189010086793338}
{"step": 20280, "loss": 3.838456726074219, "grad_norm": 0.7172297835350037, "learning_rate": 0.0001, "elbo": 3.83846, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1893033075299084}
{"step": 20285, "loss": 3.811454010009766, "grad_norm": 1.0492383241653442, "learning_rate": 0.0001, "elbo": 3.81145, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1895965282664789}
{"step": 20290, "loss": 3.815203857421875, "grad_norm": 0.657020628452301, "learning_rate": 0.0001, "elbo": 3.8152, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1898897490030496}
{"step": 20295, "loss": 3.797132110595703, "grad_norm": 0.666881263256073, "learning_rate": 0.0001, "elbo": 3.79713, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.19018296973962}
{"step": 20300, "loss": 3.8194332122802734, "grad_norm": 0.8420718908309937, "learning_rate": 0.0001, "elbo": 3.81943, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1904761904761905}
{"step": 20305, "loss": 3.8213672637939453, "grad_norm": 0.7148869037628174, "learning_rate": 0.0001, "elbo": 3.82137, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.190769411212761}
{"step": 20310, "loss": 3.8273223876953124, "grad_norm": 0.9277396202087402, "learning_rate": 0.0001, "elbo": 3.82732, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1910626319493314}
{"step": 20315, "loss": 3.8192325592041017, "grad_norm": 0.7495941519737244, "learning_rate": 0.0001, "elbo": 3.81923, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.191355852685902}
{"step": 20320, "loss": 3.8160327911376952, "grad_norm": 0.6572062969207764, "learning_rate": 0.0001, "elbo": 3.81603, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1916490734224725}
{"step": 20325, "loss": 3.800079345703125, "grad_norm": 0.680426836013794, "learning_rate": 0.0001, "elbo": 3.80008, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.191942294159043}
{"step": 20330, "loss": 3.80902214050293, "grad_norm": 0.7367345690727234, "learning_rate": 0.0001, "elbo": 3.80902, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1922355148956134}
{"step": 20335, "loss": 3.800732421875, "grad_norm": 0.6756629347801208, "learning_rate": 0.0001, "elbo": 3.80073, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1925287356321839}
{"step": 20340, "loss": 3.827309799194336, "grad_norm": 0.6837174296379089, "learning_rate": 0.0001, "elbo": 3.82731, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1928219563687543}
{"step": 20345, "loss": 3.821440505981445, "grad_norm": 0.785214900970459, "learning_rate": 0.0001, "elbo": 3.82144, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1931151771053248}
{"step": 20350, "loss": 3.81019287109375, "grad_norm": 0.7143091559410095, "learning_rate": 0.0001, "elbo": 3.81019, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1934083978418955}
{"step": 20355, "loss": 3.8150344848632813, "grad_norm": 0.9768143892288208, "learning_rate": 0.0001, "elbo": 3.81503, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.193701618578466}
{"step": 20360, "loss": 3.8377120971679686, "grad_norm": 0.8192777633666992, "learning_rate": 0.0001, "elbo": 3.83771, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1939948393150364}
{"step": 20365, "loss": 3.820417022705078, "grad_norm": 0.7648798227310181, "learning_rate": 0.0001, "elbo": 3.82042, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1942880600516068}
{"step": 20370, "loss": 3.8224632263183596, "grad_norm": 0.6980049014091492, "learning_rate": 0.0001, "elbo": 3.82246, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1945812807881773}
{"step": 20375, "loss": 3.7937694549560548, "grad_norm": 0.8241353034973145, "learning_rate": 0.0001, "elbo": 3.79377, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.194874501524748}
{"step": 20380, "loss": 3.82562255859375, "grad_norm": 0.6753247380256653, "learning_rate": 0.0001, "elbo": 3.82562, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1951677222613184}
{"step": 20385, "loss": 3.8287811279296875, "grad_norm": 1.1764543056488037, "learning_rate": 0.0001, "elbo": 3.82878, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1954609429978889}
{"step": 20390, "loss": 3.8167465209960936, "grad_norm": 0.7509709596633911, "learning_rate": 0.0001, "elbo": 3.81675, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1957541637344593}
{"step": 20395, "loss": 3.806382751464844, "grad_norm": 0.6815057992935181, "learning_rate": 0.0001, "elbo": 3.80638, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1960473844710298}
{"step": 20400, "loss": 3.8112483978271485, "grad_norm": 0.7190744280815125, "learning_rate": 0.0001, "elbo": 3.81125, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1963406052076002}
{"step": 20405, "loss": 3.800921630859375, "grad_norm": 1.0083214044570923, "learning_rate": 0.0001, "elbo": 3.80092, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1966338259441707}
{"step": 20410, "loss": 3.817491149902344, "grad_norm": 0.9405953288078308, "learning_rate": 0.0001, "elbo": 3.81749, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1969270466807413}
{"step": 20415, "loss": 3.837152862548828, "grad_norm": 1.2731788158416748, "learning_rate": 0.0001, "elbo": 3.83715, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1972202674173118}
{"step": 20420, "loss": 3.8137245178222656, "grad_norm": 0.6743910312652588, "learning_rate": 0.0001, "elbo": 3.81372, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1975134881538823}
{"step": 20425, "loss": 3.8037925720214845, "grad_norm": 0.720910370349884, "learning_rate": 0.0001, "elbo": 3.80379, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1978067088904527}
{"step": 20425, "loss": 3.8037925720214845, "grad_norm": 0.720910370349884, "learning_rate": 0.0001, "elbo": 3.80379, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1978067088904527}
{"step": 20430, "loss": 3.817509078979492, "grad_norm": 0.8947585821151733, "learning_rate": 0.0001, "elbo": 3.81751, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1980999296270232}
{"step": 20435, "loss": 3.8245170593261717, "grad_norm": 0.6705731749534607, "learning_rate": 0.0001, "elbo": 3.82452, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1983931503635936}
{"step": 20440, "loss": 3.8134841918945312, "grad_norm": 0.8386380672454834, "learning_rate": 0.0001, "elbo": 3.81348, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1986863711001643}
{"step": 20445, "loss": 3.80501708984375, "grad_norm": 0.6385821104049683, "learning_rate": 0.0001, "elbo": 3.80502, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1989795918367347}
{"step": 20450, "loss": 3.8018814086914063, "grad_norm": 0.8202309608459473, "learning_rate": 0.0001, "elbo": 3.80188, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1992728125733052}
{"step": 20450, "loss": 3.8018814086914063, "grad_norm": 0.8202309608459473, "learning_rate": 0.0001, "elbo": 3.80188, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1992728125733052}
{"step": 20455, "loss": 3.8123619079589846, "grad_norm": 0.6856337785720825, "learning_rate": 0.0001, "elbo": 3.81236, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.1995660333098757}
{"step": 20460, "loss": 3.822165298461914, "grad_norm": 0.6813353896141052, "learning_rate": 0.0001, "elbo": 3.82217, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.199859254046446}
{"step": 20465, "loss": 3.807442474365234, "grad_norm": 0.7967454791069031, "learning_rate": 0.0001, "elbo": 3.80744, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2001524747830166}
{"step": 20470, "loss": 3.8079532623291015, "grad_norm": 0.8058314919471741, "learning_rate": 0.0001, "elbo": 3.80795, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2004456955195872}
{"step": 20475, "loss": 3.801482009887695, "grad_norm": 0.744789719581604, "learning_rate": 0.0001, "elbo": 3.80148, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2007389162561577}
{"step": 20480, "loss": 3.8189975738525392, "grad_norm": 0.7025384306907654, "learning_rate": 0.0001, "elbo": 3.819, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2010321369927281}
{"step": 20485, "loss": 3.816994476318359, "grad_norm": 0.787461519241333, "learning_rate": 0.0001, "elbo": 3.81699, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2013253577292986}
{"step": 20490, "loss": 3.807119369506836, "grad_norm": 0.7508420348167419, "learning_rate": 0.0001, "elbo": 3.80712, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.201618578465869}
{"step": 20495, "loss": 3.811146545410156, "grad_norm": 0.7293830513954163, "learning_rate": 0.0001, "elbo": 3.81115, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2019117992024395}
{"step": 20500, "loss": 3.8239208221435548, "grad_norm": 0.6778295636177063, "learning_rate": 0.0001, "elbo": 3.82392, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2022050199390102}
{"step": 20505, "loss": 3.819204330444336, "grad_norm": 0.7802640199661255, "learning_rate": 0.0001, "elbo": 3.8192, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2024982406755806}
{"step": 20510, "loss": 3.8065616607666017, "grad_norm": 0.8902842998504639, "learning_rate": 0.0001, "elbo": 3.80656, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.202791461412151}
{"step": 20515, "loss": 3.7951854705810546, "grad_norm": 0.6615850329399109, "learning_rate": 0.0001, "elbo": 3.79519, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2030846821487216}
{"step": 20520, "loss": 3.8192005157470703, "grad_norm": 0.7030910849571228, "learning_rate": 0.0001, "elbo": 3.8192, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.203377902885292}
{"step": 20525, "loss": 3.81107177734375, "grad_norm": 0.8047128915786743, "learning_rate": 0.0001, "elbo": 3.81107, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2036711236218625}
{"step": 20530, "loss": 3.807537841796875, "grad_norm": 0.7246868014335632, "learning_rate": 0.0001, "elbo": 3.80754, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.203964344358433}
{"step": 20535, "loss": 3.8093517303466795, "grad_norm": 0.653984546661377, "learning_rate": 0.0001, "elbo": 3.80935, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2042575650950036}
{"step": 20540, "loss": 3.814579391479492, "grad_norm": 0.6585443615913391, "learning_rate": 0.0001, "elbo": 3.81458, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.204550785831574}
{"step": 20545, "loss": 3.813080596923828, "grad_norm": 0.6531610488891602, "learning_rate": 0.0001, "elbo": 3.81308, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2048440065681445}
{"step": 20550, "loss": 3.827516555786133, "grad_norm": 0.6738696098327637, "learning_rate": 0.0001, "elbo": 3.82752, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.205137227304715}
{"step": 20555, "loss": 3.8140796661376952, "grad_norm": 0.7328583598136902, "learning_rate": 0.0001, "elbo": 3.81408, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2054304480412854}
{"step": 20560, "loss": 3.80697021484375, "grad_norm": 0.7847872972488403, "learning_rate": 0.0001, "elbo": 3.80697, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.205723668777856}
{"step": 20565, "loss": 3.818486785888672, "grad_norm": 0.6955916881561279, "learning_rate": 0.0001, "elbo": 3.81849, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2060168895144265}
{"step": 20570, "loss": 3.824828338623047, "grad_norm": 0.6639728546142578, "learning_rate": 0.0001, "elbo": 3.82483, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.206310110250997}
{"step": 20575, "loss": 3.8011764526367187, "grad_norm": 0.6930263042449951, "learning_rate": 0.0001, "elbo": 3.80118, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2066033309875674}
{"step": 20580, "loss": 3.8464942932128907, "grad_norm": 0.6870307326316833, "learning_rate": 0.0001, "elbo": 3.84649, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.206896551724138}
{"step": 20585, "loss": 3.812255096435547, "grad_norm": 0.6797831654548645, "learning_rate": 0.0001, "elbo": 3.81226, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2071897724607084}
{"step": 20590, "loss": 3.811728668212891, "grad_norm": 0.711570143699646, "learning_rate": 0.0001, "elbo": 3.81173, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2074829931972788}
{"step": 20595, "loss": 3.789683532714844, "grad_norm": 0.7719312906265259, "learning_rate": 0.0001, "elbo": 3.78968, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2077762139338495}
{"step": 20600, "loss": 3.8133590698242186, "grad_norm": 0.6916393637657166, "learning_rate": 0.0001, "elbo": 3.81336, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.20806943467042}
{"step": 20605, "loss": 3.8200794219970704, "grad_norm": 0.6438207626342773, "learning_rate": 0.0001, "elbo": 3.82008, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2083626554069904}
{"step": 20610, "loss": 3.810026931762695, "grad_norm": 0.6489377021789551, "learning_rate": 0.0001, "elbo": 3.81003, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2086558761435608}
{"step": 20615, "loss": 3.806536865234375, "grad_norm": 0.6497002840042114, "learning_rate": 0.0001, "elbo": 3.80654, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2089490968801313}
{"step": 20620, "loss": 3.810802459716797, "grad_norm": 0.8274475336074829, "learning_rate": 0.0001, "elbo": 3.8108, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2092423176167018}
{"step": 20625, "loss": 3.8118911743164063, "grad_norm": 0.7331326007843018, "learning_rate": 0.0001, "elbo": 3.81189, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2095355383532724}
{"step": 20630, "loss": 3.800759124755859, "grad_norm": 0.6640715599060059, "learning_rate": 0.0001, "elbo": 3.80076, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2098287590898429}
{"step": 20635, "loss": 3.8113868713378904, "grad_norm": 0.8110660314559937, "learning_rate": 0.0001, "elbo": 3.81139, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2101219798264133}
{"step": 20640, "loss": 3.7998947143554687, "grad_norm": 0.6661732792854309, "learning_rate": 0.0001, "elbo": 3.79989, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2104152005629838}
{"step": 20645, "loss": 3.8004989624023438, "grad_norm": 0.7323822975158691, "learning_rate": 0.0001, "elbo": 3.8005, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2107084212995542}
{"step": 20650, "loss": 3.8201961517333984, "grad_norm": 0.8765579462051392, "learning_rate": 0.0001, "elbo": 3.8202, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2110016420361247}
{"step": 20655, "loss": 3.818247985839844, "grad_norm": 0.6649433970451355, "learning_rate": 0.0001, "elbo": 3.81825, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2112948627726954}
{"step": 20660, "loss": 3.803422546386719, "grad_norm": 0.7038288116455078, "learning_rate": 0.0001, "elbo": 3.80342, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2115880835092658}
{"step": 20665, "loss": 3.8166267395019533, "grad_norm": 0.9931538105010986, "learning_rate": 0.0001, "elbo": 3.81663, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2118813042458363}
{"step": 20670, "loss": 3.781551742553711, "grad_norm": 0.6738998889923096, "learning_rate": 0.0001, "elbo": 3.78155, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2121745249824067}
{"step": 20675, "loss": 3.8054458618164064, "grad_norm": 0.6452013254165649, "learning_rate": 0.0001, "elbo": 3.80545, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2124677457189772}
{"step": 20680, "loss": 3.8109970092773438, "grad_norm": 0.6638656258583069, "learning_rate": 0.0001, "elbo": 3.811, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2127609664555477}
{"step": 20685, "loss": 3.8031490325927733, "grad_norm": 0.6755728721618652, "learning_rate": 0.0001, "elbo": 3.80315, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2130541871921183}
{"step": 20690, "loss": 3.811655807495117, "grad_norm": 0.9477487802505493, "learning_rate": 0.0001, "elbo": 3.81166, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2133474079286888}
{"step": 20695, "loss": 3.8204586029052736, "grad_norm": 0.6763441562652588, "learning_rate": 0.0001, "elbo": 3.82046, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2136406286652592}
{"step": 20700, "loss": 3.803268051147461, "grad_norm": 0.6553472280502319, "learning_rate": 0.0001, "elbo": 3.80327, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2139338494018297}
{"step": 20705, "loss": 3.8215904235839844, "grad_norm": 0.727354109287262, "learning_rate": 0.0001, "elbo": 3.82159, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2142270701384001}
{"step": 20710, "loss": 3.8052932739257814, "grad_norm": 0.7261942028999329, "learning_rate": 0.0001, "elbo": 3.80529, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2145202908749706}
{"step": 20710, "loss": 3.8052932739257814, "grad_norm": 0.7261942028999329, "learning_rate": 0.0001, "elbo": 3.80529, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2145202908749706}
{"step": 20715, "loss": 3.8072776794433594, "grad_norm": 0.7852824926376343, "learning_rate": 0.0001, "elbo": 3.80728, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.214813511611541}
{"step": 20720, "loss": 3.804930877685547, "grad_norm": 0.7644269466400146, "learning_rate": 0.0001, "elbo": 3.80493, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2151067323481117}
{"step": 20725, "loss": 3.8059173583984376, "grad_norm": 0.8119410872459412, "learning_rate": 0.0001, "elbo": 3.80592, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2153999530846822}
{"step": 20725, "loss": 3.8059173583984376, "grad_norm": 0.8119410872459412, "learning_rate": 0.0001, "elbo": 3.80592, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2153999530846822}
{"step": 20730, "loss": 3.7975849151611327, "grad_norm": 0.7492141723632812, "learning_rate": 0.0001, "elbo": 3.79758, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2156931738212526}
{"step": 20735, "loss": 3.7977489471435546, "grad_norm": 0.6709492206573486, "learning_rate": 0.0001, "elbo": 3.79775, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.215986394557823}
{"step": 20740, "loss": 3.809783935546875, "grad_norm": 0.6736410856246948, "learning_rate": 0.0001, "elbo": 3.80978, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2162796152943935}
{"step": 20745, "loss": 3.809622573852539, "grad_norm": 0.6600658297538757, "learning_rate": 0.0001, "elbo": 3.80962, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2165728360309642}
{"step": 20750, "loss": 3.8142955780029295, "grad_norm": 0.6719239950180054, "learning_rate": 0.0001, "elbo": 3.8143, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2168660567675347}
{"step": 20755, "loss": 3.8073040008544923, "grad_norm": 0.6756604909896851, "learning_rate": 0.0001, "elbo": 3.8073, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2171592775041051}
{"step": 20760, "loss": 3.8258094787597656, "grad_norm": 0.7528799176216125, "learning_rate": 0.0001, "elbo": 3.82581, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2174524982406756}
{"step": 20765, "loss": 3.8009101867675783, "grad_norm": 0.7315093874931335, "learning_rate": 0.0001, "elbo": 3.80091, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.217745718977246}
{"step": 20770, "loss": 3.8004104614257814, "grad_norm": 0.7507948279380798, "learning_rate": 0.0001, "elbo": 3.80041, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2180389397138165}
{"step": 20775, "loss": 3.7919864654541016, "grad_norm": 0.7445353865623474, "learning_rate": 0.0001, "elbo": 3.79199, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.218332160450387}
{"step": 20780, "loss": 3.8008331298828124, "grad_norm": 0.656684398651123, "learning_rate": 0.0001, "elbo": 3.80083, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2186253811869576}
{"step": 20785, "loss": 3.8175727844238283, "grad_norm": 0.784250020980835, "learning_rate": 0.0001, "elbo": 3.81757, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.218918601923528}
{"step": 20790, "loss": 3.808024597167969, "grad_norm": 0.7041321396827698, "learning_rate": 0.0001, "elbo": 3.80802, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2192118226600985}
{"step": 20795, "loss": 3.788595962524414, "grad_norm": 0.8182581067085266, "learning_rate": 0.0001, "elbo": 3.7886, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.219505043396669}
{"step": 20800, "loss": 3.804188537597656, "grad_norm": 0.7611117362976074, "learning_rate": 0.0001, "elbo": 3.80419, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2197982641332394}
{"step": 20805, "loss": 3.8017127990722654, "grad_norm": 0.674429714679718, "learning_rate": 0.0001, "elbo": 3.80171, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2200914848698101}
{"step": 20810, "loss": 3.793354797363281, "grad_norm": 0.6535811424255371, "learning_rate": 0.0001, "elbo": 3.79335, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2203847056063806}
{"step": 20815, "loss": 3.804536056518555, "grad_norm": 0.6341484189033508, "learning_rate": 0.0001, "elbo": 3.80454, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.220677926342951}
{"step": 20820, "loss": 3.8003379821777346, "grad_norm": 0.6653988361358643, "learning_rate": 0.0001, "elbo": 3.80034, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2209711470795215}
{"step": 20825, "loss": 3.798760986328125, "grad_norm": 0.8360251188278198, "learning_rate": 0.0001, "elbo": 3.79876, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.221264367816092}
{"step": 20830, "loss": 3.788890075683594, "grad_norm": 0.6929168105125427, "learning_rate": 0.0001, "elbo": 3.78889, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2215575885526624}
{"step": 20835, "loss": 3.804271697998047, "grad_norm": 0.6406546235084534, "learning_rate": 0.0001, "elbo": 3.80427, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2218508092892328}
{"step": 20840, "loss": 3.7996150970458986, "grad_norm": 0.774924635887146, "learning_rate": 0.0001, "elbo": 3.79962, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2221440300258035}
{"step": 20845, "loss": 3.82220458984375, "grad_norm": 0.6496642827987671, "learning_rate": 0.0001, "elbo": 3.8222, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.222437250762374}
{"step": 20850, "loss": 3.8120162963867186, "grad_norm": 0.6433900594711304, "learning_rate": 0.0001, "elbo": 3.81202, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2227304714989444}
{"step": 20855, "loss": 3.817588043212891, "grad_norm": 0.6607900261878967, "learning_rate": 0.0001, "elbo": 3.81759, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2230236922355149}
{"step": 20860, "loss": 3.811939239501953, "grad_norm": 0.7530423402786255, "learning_rate": 0.0001, "elbo": 3.81194, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2233169129720853}
{"step": 20865, "loss": 3.801746368408203, "grad_norm": 0.7055870294570923, "learning_rate": 0.0001, "elbo": 3.80175, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2236101337086558}
{"step": 20870, "loss": 3.799735260009766, "grad_norm": 0.7064525485038757, "learning_rate": 0.0001, "elbo": 3.79974, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2239033544452265}
{"step": 20875, "loss": 3.802996826171875, "grad_norm": 0.7200853824615479, "learning_rate": 0.0001, "elbo": 3.803, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.224196575181797}
{"step": 20880, "loss": 3.828321075439453, "grad_norm": 0.7425521612167358, "learning_rate": 0.0001, "elbo": 3.82832, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2244897959183674}
{"step": 20885, "loss": 3.791857147216797, "grad_norm": 0.7130522131919861, "learning_rate": 0.0001, "elbo": 3.79186, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2247830166549378}
{"step": 20890, "loss": 3.813404846191406, "grad_norm": 0.7015361785888672, "learning_rate": 0.0001, "elbo": 3.8134, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2250762373915083}
{"step": 20895, "loss": 3.8017288208007813, "grad_norm": 0.6932840943336487, "learning_rate": 0.0001, "elbo": 3.80173, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2253694581280787}
{"step": 20900, "loss": 3.805754852294922, "grad_norm": 0.9306864142417908, "learning_rate": 0.0001, "elbo": 3.80576, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2256626788646492}
{"step": 20905, "loss": 3.8059444427490234, "grad_norm": 0.6722354292869568, "learning_rate": 0.0001, "elbo": 3.80594, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2259558996012199}
{"step": 20910, "loss": 3.7936439514160156, "grad_norm": 0.6786664128303528, "learning_rate": 0.0001, "elbo": 3.79364, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2262491203377903}
{"step": 20915, "loss": 3.7981895446777343, "grad_norm": 0.6788913607597351, "learning_rate": 0.0001, "elbo": 3.79819, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2265423410743608}
{"step": 20920, "loss": 3.784328079223633, "grad_norm": 0.9333161115646362, "learning_rate": 0.0001, "elbo": 3.78433, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2268355618109312}
{"step": 20920, "loss": 3.784328079223633, "grad_norm": 0.9333161115646362, "learning_rate": 0.0001, "elbo": 3.78433, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2268355618109312}
{"step": 20925, "loss": 3.81531982421875, "grad_norm": 0.7097135186195374, "learning_rate": 0.0001, "elbo": 3.81532, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2271287825475017}
{"step": 20925, "loss": 3.81531982421875, "grad_norm": 0.7097135186195374, "learning_rate": 0.0001, "elbo": 3.81532, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2271287825475017}
{"step": 20930, "loss": 3.8068473815917967, "grad_norm": 1.1873154640197754, "learning_rate": 0.0001, "elbo": 3.80685, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2274220032840724}
{"step": 20935, "loss": 3.802770233154297, "grad_norm": 1.0029923915863037, "learning_rate": 0.0001, "elbo": 3.80277, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2277152240206428}
{"step": 20940, "loss": 3.8163825988769533, "grad_norm": 0.6467872858047485, "learning_rate": 0.0001, "elbo": 3.81638, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2280084447572133}
{"step": 20945, "loss": 3.8154132843017576, "grad_norm": 0.6857920289039612, "learning_rate": 0.0001, "elbo": 3.81541, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2283016654937837}
{"step": 20950, "loss": 3.821161651611328, "grad_norm": 0.6446499824523926, "learning_rate": 0.0001, "elbo": 3.82116, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2285948862303542}
{"step": 20955, "loss": 3.790516662597656, "grad_norm": 0.7236267924308777, "learning_rate": 0.0001, "elbo": 3.79052, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2288881069669246}
{"step": 20960, "loss": 3.801303482055664, "grad_norm": 0.9072749018669128, "learning_rate": 0.0001, "elbo": 3.8013, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.229181327703495}
{"step": 20965, "loss": 3.8045692443847656, "grad_norm": 0.9913542866706848, "learning_rate": 0.0001, "elbo": 3.80457, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2294745484400658}
{"step": 20970, "loss": 3.798237609863281, "grad_norm": 0.7344192266464233, "learning_rate": 0.0001, "elbo": 3.79824, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2297677691766362}
{"step": 20975, "loss": 3.8110420227050783, "grad_norm": 0.7372507452964783, "learning_rate": 0.0001, "elbo": 3.81104, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2300609899132067}
{"step": 20980, "loss": 3.8178558349609375, "grad_norm": 0.721350908279419, "learning_rate": 0.0001, "elbo": 3.81786, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2303542106497771}
{"step": 20985, "loss": 3.819225311279297, "grad_norm": 0.8375622630119324, "learning_rate": 0.0001, "elbo": 3.81923, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2306474313863476}
{"step": 20990, "loss": 3.820606994628906, "grad_norm": 0.7596713304519653, "learning_rate": 0.0001, "elbo": 3.82061, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2309406521229183}
{"step": 20995, "loss": 3.801064300537109, "grad_norm": 0.7236812710762024, "learning_rate": 0.0001, "elbo": 3.80106, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2312338728594887}
{"step": 21000, "loss": 3.794972229003906, "grad_norm": 0.7177913188934326, "learning_rate": 0.0001, "elbo": 3.79497, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2315270935960592}
{"step": 21005, "loss": 3.8034263610839845, "grad_norm": 0.7979158759117126, "learning_rate": 0.0001, "elbo": 3.80343, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2318203143326296}
{"step": 21005, "loss": 3.8034263610839845, "grad_norm": 0.7979158759117126, "learning_rate": 0.0001, "elbo": 3.80343, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2318203143326296}
{"step": 21010, "loss": 3.7846797943115233, "grad_norm": 0.6896952390670776, "learning_rate": 0.0001, "elbo": 3.78468, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2321135350692}
{"step": 21010, "loss": 3.7846797943115233, "grad_norm": 0.6896952390670776, "learning_rate": 0.0001, "elbo": 3.78468, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2321135350692}
{"step": 21015, "loss": 3.8114551544189452, "grad_norm": 0.751222550868988, "learning_rate": 0.0001, "elbo": 3.81146, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2324067558057705}
{"step": 21020, "loss": 3.795737457275391, "grad_norm": 0.789115309715271, "learning_rate": 0.0001, "elbo": 3.79574, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.232699976542341}
{"step": 21025, "loss": 3.8130264282226562, "grad_norm": 0.668198823928833, "learning_rate": 0.0001, "elbo": 3.81303, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2329931972789117}
{"step": 21030, "loss": 3.802587127685547, "grad_norm": 0.7222834825515747, "learning_rate": 0.0001, "elbo": 3.80259, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.233286418015482}
{"step": 21035, "loss": 3.7987060546875, "grad_norm": 0.6624637842178345, "learning_rate": 0.0001, "elbo": 3.79871, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2335796387520526}
{"step": 21040, "loss": 3.810141754150391, "grad_norm": 0.6726903319358826, "learning_rate": 0.0001, "elbo": 3.81014, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.233872859488623}
{"step": 21045, "loss": 3.790570831298828, "grad_norm": 0.6625449657440186, "learning_rate": 0.0001, "elbo": 3.79057, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2341660802251935}
{"step": 21045, "loss": 3.790570831298828, "grad_norm": 0.6625449657440186, "learning_rate": 0.0001, "elbo": 3.79057, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2341660802251935}
{"step": 21050, "loss": 3.793671417236328, "grad_norm": 0.6816113591194153, "learning_rate": 0.0001, "elbo": 3.79367, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.234459300961764}
{"step": 21055, "loss": 3.803240966796875, "grad_norm": 0.6952277421951294, "learning_rate": 0.0001, "elbo": 3.80324, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2347525216983346}
{"step": 21060, "loss": 3.806720733642578, "grad_norm": 0.7593227028846741, "learning_rate": 0.0001, "elbo": 3.80672, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.235045742434905}
{"step": 21065, "loss": 3.8035133361816404, "grad_norm": 0.7103803157806396, "learning_rate": 0.0001, "elbo": 3.80351, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2353389631714755}
{"step": 21070, "loss": 3.7994064331054687, "grad_norm": 0.6773800253868103, "learning_rate": 0.0001, "elbo": 3.79941, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.235632183908046}
{"step": 21075, "loss": 3.8125038146972656, "grad_norm": 0.7011436820030212, "learning_rate": 0.0001, "elbo": 3.8125, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2359254046446164}
{"step": 21080, "loss": 3.7842761993408205, "grad_norm": 0.9477906823158264, "learning_rate": 0.0001, "elbo": 3.78428, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2362186253811869}
{"step": 21085, "loss": 3.8093711853027346, "grad_norm": 0.7310551404953003, "learning_rate": 0.0001, "elbo": 3.80937, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2365118461177573}
{"step": 21090, "loss": 3.8121017456054687, "grad_norm": 0.7345371246337891, "learning_rate": 0.0001, "elbo": 3.8121, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.236805066854328}
{"step": 21095, "loss": 3.805999755859375, "grad_norm": 0.7017391920089722, "learning_rate": 0.0001, "elbo": 3.806, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2370982875908985}
{"step": 21100, "loss": 3.813549041748047, "grad_norm": 0.6639387607574463, "learning_rate": 0.0001, "elbo": 3.81355, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.237391508327469}
{"step": 21105, "loss": 3.7947418212890627, "grad_norm": 0.6957993507385254, "learning_rate": 0.0001, "elbo": 3.79474, "mask_frac": 0.5013, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2376847290640394}
{"step": 21110, "loss": 3.8011043548583983, "grad_norm": 0.6818423867225647, "learning_rate": 0.0001, "elbo": 3.8011, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2379779498006098}
{"step": 21115, "loss": 3.7803611755371094, "grad_norm": 0.6887681484222412, "learning_rate": 0.0001, "elbo": 3.78036, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2382711705371805}
{"step": 21120, "loss": 3.794164276123047, "grad_norm": 0.7306749224662781, "learning_rate": 0.0001, "elbo": 3.79416, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.238564391273751}
{"step": 21125, "loss": 3.8098602294921875, "grad_norm": 0.7645215392112732, "learning_rate": 0.0001, "elbo": 3.80986, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2388576120103214}
{"step": 21130, "loss": 3.7942710876464845, "grad_norm": 0.7948590517044067, "learning_rate": 0.0001, "elbo": 3.79427, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2391508327468919}
{"step": 21130, "loss": 3.7942710876464845, "grad_norm": 0.7948590517044067, "learning_rate": 0.0001, "elbo": 3.79427, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2391508327468919}
{"step": 21135, "loss": 3.776140594482422, "grad_norm": 0.6876015067100525, "learning_rate": 0.0001, "elbo": 3.77614, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2394440534834623}
{"step": 21140, "loss": 3.8035545349121094, "grad_norm": 0.6639440059661865, "learning_rate": 0.0001, "elbo": 3.80355, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2397372742200328}
{"step": 21140, "loss": 3.8035545349121094, "grad_norm": 0.6639440059661865, "learning_rate": 0.0001, "elbo": 3.80355, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2397372742200328}
{"step": 21145, "loss": 3.8088134765625, "grad_norm": 0.6860443353652954, "learning_rate": 0.0001, "elbo": 3.80881, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2400304949566032}
{"step": 21150, "loss": 3.812916564941406, "grad_norm": 0.6849285364151001, "learning_rate": 0.0001, "elbo": 3.81292, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.240323715693174}
{"step": 21155, "loss": 3.7952083587646483, "grad_norm": 0.6745406985282898, "learning_rate": 0.0001, "elbo": 3.79521, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2406169364297444}
{"step": 21160, "loss": 3.8032875061035156, "grad_norm": 0.6916137337684631, "learning_rate": 0.0001, "elbo": 3.80329, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2409101571663148}
{"step": 21165, "loss": 3.810369873046875, "grad_norm": 0.6668186783790588, "learning_rate": 0.0001, "elbo": 3.81037, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2412033779028853}
{"step": 21170, "loss": 3.804981231689453, "grad_norm": 0.7298088073730469, "learning_rate": 0.0001, "elbo": 3.80498, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2414965986394557}
{"step": 21175, "loss": 3.804834747314453, "grad_norm": 0.8371623158454895, "learning_rate": 0.0001, "elbo": 3.80483, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2417898193760264}
{"step": 21180, "loss": 3.7946971893310546, "grad_norm": 0.7535154819488525, "learning_rate": 0.0001, "elbo": 3.7947, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2420830401125968}
{"step": 21185, "loss": 3.795895004272461, "grad_norm": 0.8290486335754395, "learning_rate": 0.0001, "elbo": 3.7959, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2423762608491673}
{"step": 21190, "loss": 3.784502410888672, "grad_norm": 0.7152810096740723, "learning_rate": 0.0001, "elbo": 3.7845, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2426694815857378}
{"step": 21195, "loss": 3.7890045166015627, "grad_norm": 0.7924755215644836, "learning_rate": 0.0001, "elbo": 3.789, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2429627023223082}
{"step": 21200, "loss": 3.790868377685547, "grad_norm": 0.8325777649879456, "learning_rate": 0.0001, "elbo": 3.79087, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2432559230588787}
{"step": 21205, "loss": 3.802662658691406, "grad_norm": 0.7191636562347412, "learning_rate": 0.0001, "elbo": 3.80266, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2435491437954491}
{"step": 21205, "loss": 3.802662658691406, "grad_norm": 0.7191636562347412, "learning_rate": 0.0001, "elbo": 3.80266, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2435491437954491}
{"step": 21210, "loss": 3.791978454589844, "grad_norm": 0.7298662662506104, "learning_rate": 0.0001, "elbo": 3.79198, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2438423645320198}
{"step": 21215, "loss": 3.8003494262695314, "grad_norm": 0.6449458003044128, "learning_rate": 0.0001, "elbo": 3.80035, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2441355852685902}
{"step": 21220, "loss": 3.8070919036865236, "grad_norm": 0.6707053780555725, "learning_rate": 0.0001, "elbo": 3.80709, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2444288060051607}
{"step": 21225, "loss": 3.7967430114746095, "grad_norm": 0.6962562799453735, "learning_rate": 0.0001, "elbo": 3.79674, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2447220267417312}
{"step": 21230, "loss": 3.8063663482666015, "grad_norm": 0.8061076998710632, "learning_rate": 0.0001, "elbo": 3.80637, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2450152474783016}
{"step": 21235, "loss": 3.804170608520508, "grad_norm": 0.8240434527397156, "learning_rate": 0.0001, "elbo": 3.80417, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.245308468214872}
{"step": 21240, "loss": 3.8076019287109375, "grad_norm": 0.7538439631462097, "learning_rate": 0.0001, "elbo": 3.8076, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2456016889514427}
{"step": 21245, "loss": 3.790688705444336, "grad_norm": 0.9559108018875122, "learning_rate": 0.0001, "elbo": 3.79069, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2458949096880132}
{"step": 21250, "loss": 3.825496292114258, "grad_norm": 0.7685986757278442, "learning_rate": 0.0001, "elbo": 3.8255, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2461881304245837}
{"step": 21255, "loss": 3.807598114013672, "grad_norm": 0.6996116042137146, "learning_rate": 0.0001, "elbo": 3.8076, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.246481351161154}
{"step": 21260, "loss": 3.807508850097656, "grad_norm": 0.6682822108268738, "learning_rate": 0.0001, "elbo": 3.80751, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2467745718977246}
{"step": 21265, "loss": 3.8143661499023436, "grad_norm": 0.6419785022735596, "learning_rate": 0.0001, "elbo": 3.81437, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.247067792634295}
{"step": 21270, "loss": 3.801894760131836, "grad_norm": 0.732556939125061, "learning_rate": 0.0001, "elbo": 3.80189, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2473610133708655}
{"step": 21275, "loss": 3.8055694580078123, "grad_norm": 0.751646876335144, "learning_rate": 0.0001, "elbo": 3.80557, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2476542341074361}
{"step": 21280, "loss": 3.7900184631347655, "grad_norm": 0.6906797885894775, "learning_rate": 0.0001, "elbo": 3.79002, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2479474548440066}
{"step": 21285, "loss": 3.7966556549072266, "grad_norm": 0.7478295564651489, "learning_rate": 0.0001, "elbo": 3.79666, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.248240675580577}
{"step": 21290, "loss": 3.8085891723632814, "grad_norm": 0.7023538947105408, "learning_rate": 0.0001, "elbo": 3.80859, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2485338963171475}
{"step": 21295, "loss": 3.8004501342773436, "grad_norm": 0.7409341931343079, "learning_rate": 0.0001, "elbo": 3.80045, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.248827117053718}
{"step": 21300, "loss": 3.791717529296875, "grad_norm": 0.6488752961158752, "learning_rate": 0.0001, "elbo": 3.79172, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2491203377902886}
{"step": 21305, "loss": 3.7884857177734377, "grad_norm": 0.7153981924057007, "learning_rate": 0.0001, "elbo": 3.78849, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.249413558526859}
{"step": 21310, "loss": 3.7832595825195314, "grad_norm": 0.6369677782058716, "learning_rate": 0.0001, "elbo": 3.78326, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2497067792634295}
{"step": 21315, "loss": 3.7923263549804687, "grad_norm": 0.7099170684814453, "learning_rate": 0.0001, "elbo": 3.79233, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.25}
{"step": 21320, "loss": 3.7912319183349608, "grad_norm": 0.750217854976654, "learning_rate": 0.0001, "elbo": 3.79123, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2502932207365705}
{"step": 21325, "loss": 3.7891471862792967, "grad_norm": 0.687503457069397, "learning_rate": 0.0001, "elbo": 3.78915, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.250586441473141}
{"step": 21330, "loss": 3.8052322387695314, "grad_norm": 0.6663997769355774, "learning_rate": 0.0001, "elbo": 3.80523, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2508796622097114}
{"step": 21335, "loss": 3.7970447540283203, "grad_norm": 0.690096378326416, "learning_rate": 0.0001, "elbo": 3.79704, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.251172882946282}
{"step": 21335, "loss": 3.7970447540283203, "grad_norm": 0.690096378326416, "learning_rate": 0.0001, "elbo": 3.79704, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.251172882946282}
{"step": 21340, "loss": 3.806182098388672, "grad_norm": 0.7755553126335144, "learning_rate": 0.0001, "elbo": 3.80618, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2514661036828525}
{"step": 21345, "loss": 3.7875701904296877, "grad_norm": 1.043959140777588, "learning_rate": 0.0001, "elbo": 3.78757, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.251759324419423}
{"step": 21350, "loss": 3.7898399353027346, "grad_norm": 0.6466929316520691, "learning_rate": 0.0001, "elbo": 3.78984, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2520525451559934}
{"step": 21355, "loss": 3.807952880859375, "grad_norm": 0.6729710102081299, "learning_rate": 0.0001, "elbo": 3.80795, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2523457658925639}
{"step": 21360, "loss": 3.7935714721679688, "grad_norm": 0.6520002484321594, "learning_rate": 0.0001, "elbo": 3.79357, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2526389866291345}
{"step": 21365, "loss": 3.795449066162109, "grad_norm": 0.6404517889022827, "learning_rate": 0.0001, "elbo": 3.79545, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.252932207365705}
{"step": 21370, "loss": 3.798101806640625, "grad_norm": 0.6749706864356995, "learning_rate": 0.0001, "elbo": 3.7981, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2532254281022754}
{"step": 21375, "loss": 3.8069854736328126, "grad_norm": 0.6745633482933044, "learning_rate": 0.0001, "elbo": 3.80699, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.253518648838846}
{"step": 21380, "loss": 3.7939430236816407, "grad_norm": 0.7046282291412354, "learning_rate": 0.0001, "elbo": 3.79394, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2538118695754163}
{"step": 21385, "loss": 3.806412124633789, "grad_norm": 0.6918503642082214, "learning_rate": 0.0001, "elbo": 3.80641, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2541050903119868}
{"step": 21390, "loss": 3.79359130859375, "grad_norm": 0.8190698027610779, "learning_rate": 0.0001, "elbo": 3.79359, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2543983110485573}
{"step": 21395, "loss": 3.8043182373046873, "grad_norm": 0.7024285793304443, "learning_rate": 0.0001, "elbo": 3.80432, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2546915317851277}
{"step": 21400, "loss": 3.812256622314453, "grad_norm": 0.7286098599433899, "learning_rate": 0.0001, "elbo": 3.81226, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2549847525216984}
{"step": 21405, "loss": 3.808643341064453, "grad_norm": 0.6856688857078552, "learning_rate": 0.0001, "elbo": 3.80864, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2552779732582688}
{"step": 21410, "loss": 3.787418746948242, "grad_norm": 0.7206289172172546, "learning_rate": 0.0001, "elbo": 3.78742, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2555711939948393}
{"step": 21415, "loss": 3.7960693359375, "grad_norm": 0.7299590110778809, "learning_rate": 0.0001, "elbo": 3.79607, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2558644147314098}
{"step": 21420, "loss": 3.785688018798828, "grad_norm": 0.6702115535736084, "learning_rate": 0.0001, "elbo": 3.78569, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2561576354679804}
{"step": 21425, "loss": 3.8082244873046873, "grad_norm": 0.6642692685127258, "learning_rate": 0.0001, "elbo": 3.80822, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2564508562045509}
{"step": 21430, "loss": 3.8001968383789064, "grad_norm": 0.6682193279266357, "learning_rate": 0.0001, "elbo": 3.8002, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2567440769411213}
{"step": 21435, "loss": 3.7878395080566407, "grad_norm": 0.7058055996894836, "learning_rate": 0.0001, "elbo": 3.78784, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2570372976776918}
{"step": 21440, "loss": 3.79298095703125, "grad_norm": 0.6521108746528625, "learning_rate": 0.0001, "elbo": 3.79298, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2573305184142622}
{"step": 21445, "loss": 3.804317855834961, "grad_norm": 0.6711742281913757, "learning_rate": 0.0001, "elbo": 3.80432, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2576237391508327}
{"step": 21450, "loss": 3.7880714416503904, "grad_norm": 0.7610171437263489, "learning_rate": 0.0001, "elbo": 3.78807, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2579169598874032}
{"step": 21455, "loss": 3.793409729003906, "grad_norm": 0.7146145105361938, "learning_rate": 0.0001, "elbo": 3.79341, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2582101806239736}
{"step": 21460, "loss": 3.799639892578125, "grad_norm": 0.7444040179252625, "learning_rate": 0.0001, "elbo": 3.79964, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2585034013605443}
{"step": 21460, "loss": 3.799639892578125, "grad_norm": 0.7444040179252625, "learning_rate": 0.0001, "elbo": 3.79964, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2585034013605443}
{"step": 21465, "loss": 3.7884735107421874, "grad_norm": 0.8318735361099243, "learning_rate": 0.0001, "elbo": 3.78847, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2587966220971147}
{"step": 21470, "loss": 3.7958724975585936, "grad_norm": 0.7052065134048462, "learning_rate": 0.0001, "elbo": 3.79587, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2590898428336852}
{"step": 21475, "loss": 3.7939395904541016, "grad_norm": 0.6983616948127747, "learning_rate": 0.0001, "elbo": 3.79394, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2593830635702556}
{"step": 21480, "loss": 3.8031097412109376, "grad_norm": 0.6844002604484558, "learning_rate": 0.0001, "elbo": 3.80311, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.259676284306826}
{"step": 21485, "loss": 3.7894798278808595, "grad_norm": 0.6569631695747375, "learning_rate": 0.0001, "elbo": 3.78948, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2599695050433968}
{"step": 21490, "loss": 3.802983856201172, "grad_norm": 0.8405230045318604, "learning_rate": 0.0001, "elbo": 3.80298, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2602627257799672}
{"step": 21495, "loss": 3.78507080078125, "grad_norm": 0.7835397124290466, "learning_rate": 0.0001, "elbo": 3.78507, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2605559465165377}
{"step": 21500, "loss": 3.801227569580078, "grad_norm": 0.6448983550071716, "learning_rate": 0.0001, "elbo": 3.80123, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2608491672531081}
{"step": 21505, "loss": 3.8147724151611326, "grad_norm": 0.725936770439148, "learning_rate": 0.0001, "elbo": 3.81477, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2611423879896786}
{"step": 21510, "loss": 3.7913387298583983, "grad_norm": 0.7757270336151123, "learning_rate": 0.0001, "elbo": 3.79134, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.261435608726249}
{"step": 21515, "loss": 3.8056888580322266, "grad_norm": 0.6836444735527039, "learning_rate": 0.0001, "elbo": 3.80569, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2617288294628195}
{"step": 21520, "loss": 3.787985992431641, "grad_norm": 0.7889834642410278, "learning_rate": 0.0001, "elbo": 3.78799, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2620220501993902}
{"step": 21525, "loss": 3.7959938049316406, "grad_norm": 0.68204265832901, "learning_rate": 0.0001, "elbo": 3.79599, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2623152709359606}
{"step": 21530, "loss": 3.803386688232422, "grad_norm": 0.7158997654914856, "learning_rate": 0.0001, "elbo": 3.80339, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.262608491672531}
{"step": 21535, "loss": 3.809439849853516, "grad_norm": 0.647070050239563, "learning_rate": 0.0001, "elbo": 3.80944, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2629017124091015}
{"step": 21540, "loss": 3.7971397399902345, "grad_norm": 0.6685821413993835, "learning_rate": 0.0001, "elbo": 3.79714, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.263194933145672}
{"step": 21545, "loss": 3.785005569458008, "grad_norm": 0.7526203989982605, "learning_rate": 0.0001, "elbo": 3.78501, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2634881538822427}
{"step": 21550, "loss": 3.798758697509766, "grad_norm": 0.7084251046180725, "learning_rate": 0.0001, "elbo": 3.79876, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2637813746188131}
{"step": 21555, "loss": 3.8089481353759767, "grad_norm": 0.655408501625061, "learning_rate": 0.0001, "elbo": 3.80895, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2640745953553836}
{"step": 21560, "loss": 3.800139236450195, "grad_norm": 0.6967282295227051, "learning_rate": 0.0001, "elbo": 3.80014, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.264367816091954}
{"step": 21565, "loss": 3.792570877075195, "grad_norm": 0.7980999946594238, "learning_rate": 0.0001, "elbo": 3.79257, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2646610368285245}
{"step": 21570, "loss": 3.8109954833984374, "grad_norm": 0.7136632800102234, "learning_rate": 0.0001, "elbo": 3.811, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.264954257565095}
{"step": 21575, "loss": 3.796405792236328, "grad_norm": 0.793113112449646, "learning_rate": 0.0001, "elbo": 3.79641, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2652474783016654}
{"step": 21580, "loss": 3.7959861755371094, "grad_norm": 0.6639652252197266, "learning_rate": 0.0001, "elbo": 3.79599, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2655406990382359}
{"step": 21585, "loss": 3.7920047760009767, "grad_norm": 0.7087907791137695, "learning_rate": 0.0001, "elbo": 3.792, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2658339197748065}
{"step": 21590, "loss": 3.8098793029785156, "grad_norm": 1.0776581764221191, "learning_rate": 0.0001, "elbo": 3.80988, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.266127140511377}
{"step": 21595, "loss": 3.8022022247314453, "grad_norm": 0.8593712449073792, "learning_rate": 0.0001, "elbo": 3.8022, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2664203612479474}
{"step": 21595, "loss": 3.8022022247314453, "grad_norm": 0.8593712449073792, "learning_rate": 0.0001, "elbo": 3.8022, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2664203612479474}
{"step": 21600, "loss": 3.819924163818359, "grad_norm": 0.6896107792854309, "learning_rate": 0.0001, "elbo": 3.81992, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.266713581984518}
{"step": 21600, "loss": 3.819924163818359, "grad_norm": 0.6896107792854309, "learning_rate": 0.0001, "elbo": 3.81992, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.266713581984518}
{"step": 21605, "loss": 3.797646331787109, "grad_norm": 0.718369722366333, "learning_rate": 0.0001, "elbo": 3.79765, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2670068027210886}
{"step": 21605, "loss": 3.797646331787109, "grad_norm": 0.718369722366333, "learning_rate": 0.0001, "elbo": 3.79765, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2670068027210886}
{"step": 21610, "loss": 3.8182579040527345, "grad_norm": 0.857761561870575, "learning_rate": 0.0001, "elbo": 3.81826, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.267300023457659}
{"step": 21615, "loss": 3.7960670471191404, "grad_norm": 0.6557609438896179, "learning_rate": 0.0001, "elbo": 3.79607, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2675932441942295}
{"step": 21620, "loss": 3.8156524658203126, "grad_norm": 0.7588874697685242, "learning_rate": 0.0001, "elbo": 3.81565, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2678864649308}
{"step": 21625, "loss": 3.799603271484375, "grad_norm": 0.6419047117233276, "learning_rate": 0.0001, "elbo": 3.7996, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2681796856673704}
{"step": 21630, "loss": 3.800740432739258, "grad_norm": 0.6843309998512268, "learning_rate": 0.0001, "elbo": 3.80074, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2684729064039408}
{"step": 21635, "loss": 3.798719787597656, "grad_norm": 0.6885502934455872, "learning_rate": 0.0001, "elbo": 3.79872, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2687661271405113}
{"step": 21640, "loss": 3.7893291473388673, "grad_norm": 0.7050020098686218, "learning_rate": 0.0001, "elbo": 3.78933, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2690593478770817}
{"step": 21645, "loss": 3.7899333953857424, "grad_norm": 0.7022948265075684, "learning_rate": 0.0001, "elbo": 3.78993, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2693525686136524}
{"step": 21650, "loss": 3.78900146484375, "grad_norm": 0.6472295522689819, "learning_rate": 0.0001, "elbo": 3.789, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2696457893502229}
{"step": 21655, "loss": 3.7983814239501954, "grad_norm": 0.6594354510307312, "learning_rate": 0.0001, "elbo": 3.79838, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2699390100867933}
{"step": 21660, "loss": 3.8023719787597656, "grad_norm": 0.6488633155822754, "learning_rate": 0.0001, "elbo": 3.80237, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2702322308233638}
{"step": 21665, "loss": 3.7883274078369142, "grad_norm": 0.691643238067627, "learning_rate": 0.0001, "elbo": 3.78833, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2705254515599342}
{"step": 21670, "loss": 3.7898223876953123, "grad_norm": 0.6391605734825134, "learning_rate": 0.0001, "elbo": 3.78982, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.270818672296505}
{"step": 21675, "loss": 3.793152618408203, "grad_norm": 0.9673283100128174, "learning_rate": 0.0001, "elbo": 3.79315, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2711118930330754}
{"step": 21680, "loss": 3.809531402587891, "grad_norm": 0.6534813046455383, "learning_rate": 0.0001, "elbo": 3.80953, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2714051137696458}
{"step": 21685, "loss": 3.8131065368652344, "grad_norm": 0.713424801826477, "learning_rate": 0.0001, "elbo": 3.81311, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2716983345062163}
{"step": 21690, "loss": 3.783568572998047, "grad_norm": 0.7605955600738525, "learning_rate": 0.0001, "elbo": 3.78357, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2719915552427867}
{"step": 21695, "loss": 3.7968605041503904, "grad_norm": 0.8559890389442444, "learning_rate": 0.0001, "elbo": 3.79686, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2722847759793572}
{"step": 21700, "loss": 3.7790733337402345, "grad_norm": 0.8919321894645691, "learning_rate": 0.0001, "elbo": 3.77907, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2725779967159276}
{"step": 21705, "loss": 3.7984592437744142, "grad_norm": 0.7921954393386841, "learning_rate": 0.0001, "elbo": 3.79846, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2728712174524983}
{"step": 21710, "loss": 3.792948913574219, "grad_norm": 0.6989854574203491, "learning_rate": 0.0001, "elbo": 3.79295, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2731644381890688}
{"step": 21715, "loss": 3.8122196197509766, "grad_norm": 0.6879485845565796, "learning_rate": 0.0001, "elbo": 3.81222, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2734576589256392}
{"step": 21720, "loss": 3.7901031494140627, "grad_norm": 0.7145424485206604, "learning_rate": 0.0001, "elbo": 3.7901, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2737508796622097}
{"step": 21725, "loss": 3.78817138671875, "grad_norm": 0.7202436327934265, "learning_rate": 0.0001, "elbo": 3.78817, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2740441003987801}
{"step": 21725, "loss": 3.78817138671875, "grad_norm": 0.7202436327934265, "learning_rate": 0.0001, "elbo": 3.78817, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2740441003987801}
{"step": 21730, "loss": 3.799089813232422, "grad_norm": 0.7125071883201599, "learning_rate": 0.0001, "elbo": 3.79909, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2743373211353508}
{"step": 21730, "loss": 3.799089813232422, "grad_norm": 0.7125071883201599, "learning_rate": 0.0001, "elbo": 3.79909, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2743373211353508}
{"step": 21735, "loss": 3.788574981689453, "grad_norm": 0.675261914730072, "learning_rate": 0.0001, "elbo": 3.78858, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2746305418719213}
{"step": 21735, "loss": 3.788574981689453, "grad_norm": 0.675261914730072, "learning_rate": 0.0001, "elbo": 3.78858, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2746305418719213}
{"step": 21740, "loss": 3.7966598510742187, "grad_norm": 0.694251298904419, "learning_rate": 0.0001, "elbo": 3.79666, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2749237626084917}
{"step": 21740, "loss": 3.7966598510742187, "grad_norm": 0.694251298904419, "learning_rate": 0.0001, "elbo": 3.79666, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2749237626084917}
{"step": 21745, "loss": 3.7848209381103515, "grad_norm": 0.6860617995262146, "learning_rate": 0.0001, "elbo": 3.78482, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2752169833450622}
{"step": 21750, "loss": 3.795474624633789, "grad_norm": 0.9493275284767151, "learning_rate": 0.0001, "elbo": 3.79547, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2755102040816326}
{"step": 21755, "loss": 3.817388916015625, "grad_norm": 1.0605484247207642, "learning_rate": 0.0001, "elbo": 3.81739, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.275803424818203}
{"step": 21760, "loss": 3.7941383361816405, "grad_norm": 0.661435067653656, "learning_rate": 0.0001, "elbo": 3.79414, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2760966455547735}
{"step": 21765, "loss": 3.788460540771484, "grad_norm": 0.7255029082298279, "learning_rate": 0.0001, "elbo": 3.78846, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.276389866291344}
{"step": 21770, "loss": 3.7828514099121096, "grad_norm": 0.6841530203819275, "learning_rate": 0.0001, "elbo": 3.78285, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2766830870279147}
{"step": 21775, "loss": 3.793547439575195, "grad_norm": 0.7218494415283203, "learning_rate": 0.0001, "elbo": 3.79355, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2769763077644851}
{"step": 21780, "loss": 3.7979415893554687, "grad_norm": 0.640049934387207, "learning_rate": 0.0001, "elbo": 3.79794, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2772695285010556}
{"step": 21785, "loss": 3.7673038482666015, "grad_norm": 0.9558704495429993, "learning_rate": 0.0001, "elbo": 3.7673, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.277562749237626}
{"step": 21790, "loss": 3.796449661254883, "grad_norm": 0.7390813231468201, "learning_rate": 0.0001, "elbo": 3.79645, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2778559699741967}
{"step": 21795, "loss": 3.776663970947266, "grad_norm": 0.7199804186820984, "learning_rate": 0.0001, "elbo": 3.77666, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2781491907107672}
{"step": 21800, "loss": 3.797150421142578, "grad_norm": 0.680300235748291, "learning_rate": 0.0001, "elbo": 3.79715, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2784424114473376}
{"step": 21805, "loss": 3.778166961669922, "grad_norm": 0.7058936953544617, "learning_rate": 0.0001, "elbo": 3.77817, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.278735632183908}
{"step": 21810, "loss": 3.7933586120605467, "grad_norm": 0.7363489270210266, "learning_rate": 0.0001, "elbo": 3.79336, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2790288529204785}
{"step": 21815, "loss": 3.793448638916016, "grad_norm": 1.0480135679244995, "learning_rate": 0.0001, "elbo": 3.79345, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.279322073657049}
{"step": 21820, "loss": 3.78991813659668, "grad_norm": 0.6821839213371277, "learning_rate": 0.0001, "elbo": 3.78992, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2796152943936194}
{"step": 21825, "loss": 3.785970687866211, "grad_norm": 0.7922815084457397, "learning_rate": 0.0001, "elbo": 3.78597, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2799085151301899}
{"step": 21830, "loss": 3.7857799530029297, "grad_norm": 0.6495627760887146, "learning_rate": 0.0001, "elbo": 3.78578, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2802017358667606}
{"step": 21835, "loss": 3.794002151489258, "grad_norm": 0.6378208994865417, "learning_rate": 0.0001, "elbo": 3.794, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.280494956603331}
{"step": 21840, "loss": 3.8008026123046874, "grad_norm": 0.6289438009262085, "learning_rate": 0.0001, "elbo": 3.8008, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2807881773399015}
{"step": 21845, "loss": 3.8045814514160154, "grad_norm": 0.8085058331489563, "learning_rate": 0.0001, "elbo": 3.80458, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.281081398076472}
{"step": 21845, "loss": 3.8045814514160154, "grad_norm": 0.8085058331489563, "learning_rate": 0.0001, "elbo": 3.80458, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.281081398076472}
{"step": 21850, "loss": 3.8015846252441405, "grad_norm": 0.648171067237854, "learning_rate": 0.0001, "elbo": 3.80159, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2813746188130424}
{"step": 21855, "loss": 3.793826675415039, "grad_norm": 0.6558742523193359, "learning_rate": 0.0001, "elbo": 3.79383, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.281667839549613}
{"step": 21860, "loss": 3.8055484771728514, "grad_norm": 0.7962969541549683, "learning_rate": 0.0001, "elbo": 3.80555, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2819610602861835}
{"step": 21865, "loss": 3.788809967041016, "grad_norm": 0.7494901418685913, "learning_rate": 0.0001, "elbo": 3.78881, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.282254281022754}
{"step": 21870, "loss": 3.796986770629883, "grad_norm": 0.6782879829406738, "learning_rate": 0.0001, "elbo": 3.79699, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2825475017593244}
{"step": 21875, "loss": 3.7962921142578123, "grad_norm": 0.8581185936927795, "learning_rate": 0.0001, "elbo": 3.79629, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2828407224958949}
{"step": 21880, "loss": 3.7804290771484377, "grad_norm": 0.7459527254104614, "learning_rate": 0.0001, "elbo": 3.78043, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2831339432324653}
{"step": 21885, "loss": 3.779541015625, "grad_norm": 0.6634523868560791, "learning_rate": 0.0001, "elbo": 3.77954, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2834271639690358}
{"step": 21885, "loss": 3.779541015625, "grad_norm": 0.6634523868560791, "learning_rate": 0.0001, "elbo": 3.77954, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2834271639690358}
{"step": 21890, "loss": 3.8021919250488283, "grad_norm": 0.6725056767463684, "learning_rate": 0.0001, "elbo": 3.80219, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2837203847056065}
{"step": 21890, "loss": 3.8021919250488283, "grad_norm": 0.6725056767463684, "learning_rate": 0.0001, "elbo": 3.80219, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2837203847056065}
{"step": 21895, "loss": 3.787375640869141, "grad_norm": 0.6709846258163452, "learning_rate": 0.0001, "elbo": 3.78738, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.284013605442177}
{"step": 21900, "loss": 3.7956165313720702, "grad_norm": 0.6636154055595398, "learning_rate": 0.0001, "elbo": 3.79562, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2843068261787474}
{"step": 21905, "loss": 3.796268844604492, "grad_norm": 0.8202769160270691, "learning_rate": 0.0001, "elbo": 3.79627, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2846000469153178}
{"step": 21910, "loss": 3.809223175048828, "grad_norm": 0.725736141204834, "learning_rate": 0.0001, "elbo": 3.80922, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2848932676518883}
{"step": 21915, "loss": 3.790391540527344, "grad_norm": 0.6631428599357605, "learning_rate": 0.0001, "elbo": 3.79039, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.285186488388459}
{"step": 21915, "loss": 3.790391540527344, "grad_norm": 0.6631428599357605, "learning_rate": 0.0001, "elbo": 3.79039, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.285186488388459}
{"step": 21920, "loss": 3.7997188568115234, "grad_norm": 0.7224612236022949, "learning_rate": 0.0001, "elbo": 3.79972, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2854797091250294}
{"step": 21925, "loss": 3.800006103515625, "grad_norm": 0.7201870679855347, "learning_rate": 0.0001, "elbo": 3.80001, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2857729298615999}
{"step": 21930, "loss": 3.7847976684570312, "grad_norm": 0.8035606741905212, "learning_rate": 0.0001, "elbo": 3.7848, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2860661505981703}
{"step": 21935, "loss": 3.799199676513672, "grad_norm": 0.6551066637039185, "learning_rate": 0.0001, "elbo": 3.7992, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2863593713347408}
{"step": 21940, "loss": 3.7972373962402344, "grad_norm": 0.721494197845459, "learning_rate": 0.0001, "elbo": 3.79724, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2866525920713112}
{"step": 21940, "loss": 3.7972373962402344, "grad_norm": 0.721494197845459, "learning_rate": 0.0001, "elbo": 3.79724, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2866525920713112}
{"step": 21945, "loss": 3.788616943359375, "grad_norm": 0.7700353264808655, "learning_rate": 0.0001, "elbo": 3.78862, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2869458128078817}
{"step": 21950, "loss": 3.7852272033691405, "grad_norm": 0.7203270792961121, "learning_rate": 0.0001, "elbo": 3.78523, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2872390335444521}
{"step": 21955, "loss": 3.791524124145508, "grad_norm": 0.9628920555114746, "learning_rate": 0.0001, "elbo": 3.79152, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2875322542810228}
{"step": 21960, "loss": 3.798823928833008, "grad_norm": 0.7438051700592041, "learning_rate": 0.0001, "elbo": 3.79882, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2878254750175933}
{"step": 21965, "loss": 3.7893901824951173, "grad_norm": 0.7428892850875854, "learning_rate": 0.0001, "elbo": 3.78939, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2881186957541637}
{"step": 21970, "loss": 3.7847023010253906, "grad_norm": 0.8095384240150452, "learning_rate": 0.0001, "elbo": 3.7847, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2884119164907342}
{"step": 21975, "loss": 3.8107704162597655, "grad_norm": 0.7097660899162292, "learning_rate": 0.0001, "elbo": 3.81077, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2887051372273048}
{"step": 21980, "loss": 3.7943603515625, "grad_norm": 0.6338699460029602, "learning_rate": 0.0001, "elbo": 3.79436, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2889983579638753}
{"step": 21985, "loss": 3.8024742126464846, "grad_norm": 0.7191172242164612, "learning_rate": 0.0001, "elbo": 3.80247, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2892915787004458}
{"step": 21990, "loss": 3.784351348876953, "grad_norm": 0.8939349055290222, "learning_rate": 0.0001, "elbo": 3.78435, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2895847994370162}
{"step": 21995, "loss": 3.7896087646484373, "grad_norm": 0.6597484350204468, "learning_rate": 0.0001, "elbo": 3.78961, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2898780201735867}
{"step": 21995, "loss": 3.7896087646484373, "grad_norm": 0.6597484350204468, "learning_rate": 0.0001, "elbo": 3.78961, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2898780201735867}
{"step": 22000, "loss": 3.7964195251464843, "grad_norm": 0.7139239311218262, "learning_rate": 0.0001, "elbo": 3.79642, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2901712409101571}
{"step": 22005, "loss": 3.8016983032226563, "grad_norm": 0.717111349105835, "learning_rate": 0.0001, "elbo": 3.8017, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2904644616467276}
{"step": 22010, "loss": 3.7865135192871096, "grad_norm": 0.7497700452804565, "learning_rate": 0.0001, "elbo": 3.78651, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.290757682383298}
{"step": 22015, "loss": 3.794866180419922, "grad_norm": 0.7328106760978699, "learning_rate": 0.0001, "elbo": 3.79487, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2910509031198687}
{"step": 22020, "loss": 3.7886512756347654, "grad_norm": 0.681499719619751, "learning_rate": 0.0001, "elbo": 3.78865, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2913441238564392}
{"step": 22025, "loss": 3.8019329071044923, "grad_norm": 0.6858850717544556, "learning_rate": 0.0001, "elbo": 3.80193, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2916373445930096}
{"step": 22030, "loss": 3.781668853759766, "grad_norm": 0.6620535254478455, "learning_rate": 0.0001, "elbo": 3.78167, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.29193056532958}
{"step": 22035, "loss": 3.7857879638671874, "grad_norm": 0.6915335059165955, "learning_rate": 0.0001, "elbo": 3.78579, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2922237860661507}
{"step": 22040, "loss": 3.7702091217041014, "grad_norm": 0.6889866590499878, "learning_rate": 0.0001, "elbo": 3.77021, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2925170068027212}
{"step": 22045, "loss": 3.800017547607422, "grad_norm": 0.6706944108009338, "learning_rate": 0.0001, "elbo": 3.80002, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2928102275392916}
{"step": 22050, "loss": 3.7842529296875, "grad_norm": 0.7093422412872314, "learning_rate": 0.0001, "elbo": 3.78425, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.293103448275862}
{"step": 22055, "loss": 3.790840911865234, "grad_norm": 0.8056914806365967, "learning_rate": 0.0001, "elbo": 3.79084, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2933966690124326}
{"step": 22055, "loss": 3.790840911865234, "grad_norm": 0.8056914806365967, "learning_rate": 0.0001, "elbo": 3.79084, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2933966690124326}
{"step": 22060, "loss": 3.7919776916503904, "grad_norm": 0.6919814348220825, "learning_rate": 0.0001, "elbo": 3.79198, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.293689889749003}
{"step": 22065, "loss": 3.7816932678222654, "grad_norm": 0.7943840622901917, "learning_rate": 0.0001, "elbo": 3.78169, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2939831104855735}
{"step": 22070, "loss": 3.794268035888672, "grad_norm": 0.7315868735313416, "learning_rate": 0.0001, "elbo": 3.79427, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.294276331222144}
{"step": 22075, "loss": 3.798115921020508, "grad_norm": 0.8238484263420105, "learning_rate": 0.0001, "elbo": 3.79812, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2945695519587146}
{"step": 22075, "loss": 3.798115921020508, "grad_norm": 0.8238484263420105, "learning_rate": 0.0001, "elbo": 3.79812, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2945695519587146}
{"step": 22080, "loss": 3.7992012023925783, "grad_norm": 0.7126870155334473, "learning_rate": 0.0001, "elbo": 3.7992, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.294862772695285}
{"step": 22085, "loss": 3.7979049682617188, "grad_norm": 0.7466196417808533, "learning_rate": 0.0001, "elbo": 3.7979, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2951559934318555}
{"step": 22090, "loss": 3.7889034271240236, "grad_norm": 0.7976369857788086, "learning_rate": 0.0001, "elbo": 3.7889, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.295449214168426}
{"step": 22095, "loss": 3.7952789306640624, "grad_norm": 0.8336199522018433, "learning_rate": 0.0001, "elbo": 3.79528, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2957424349049964}
{"step": 22095, "loss": 3.7952789306640624, "grad_norm": 0.8336199522018433, "learning_rate": 0.0001, "elbo": 3.79528, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2957424349049964}
{"step": 22100, "loss": 3.8018867492675783, "grad_norm": 0.7205060124397278, "learning_rate": 0.0001, "elbo": 3.80189, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.296035655641567}
{"step": 22105, "loss": 3.799285125732422, "grad_norm": 0.7057621479034424, "learning_rate": 0.0001, "elbo": 3.79929, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2963288763781375}
{"step": 22110, "loss": 3.784791946411133, "grad_norm": 0.6852083802223206, "learning_rate": 0.0001, "elbo": 3.78479, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.296622097114708}
{"step": 22115, "loss": 3.8031429290771483, "grad_norm": 0.7105758190155029, "learning_rate": 0.0001, "elbo": 3.80314, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2969153178512784}
{"step": 22120, "loss": 3.7809528350830077, "grad_norm": 0.6592363715171814, "learning_rate": 0.0001, "elbo": 3.78095, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.297208538587849}
{"step": 22125, "loss": 3.792158508300781, "grad_norm": 0.6450772881507874, "learning_rate": 0.0001, "elbo": 3.79216, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2975017593244194}
{"step": 22130, "loss": 3.800399398803711, "grad_norm": 0.7131363153457642, "learning_rate": 0.0001, "elbo": 3.8004, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2977949800609898}
{"step": 22135, "loss": 3.7729507446289063, "grad_norm": 0.8163507580757141, "learning_rate": 0.0001, "elbo": 3.77295, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2980882007975603}
{"step": 22140, "loss": 3.7812976837158203, "grad_norm": 0.8254705667495728, "learning_rate": 0.0001, "elbo": 3.7813, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.298381421534131}
{"step": 22145, "loss": 3.7794113159179688, "grad_norm": 0.7015059590339661, "learning_rate": 0.0001, "elbo": 3.77941, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2986746422707014}
{"step": 22150, "loss": 3.7795490264892577, "grad_norm": 0.7523183226585388, "learning_rate": 0.0001, "elbo": 3.77955, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2989678630072719}
{"step": 22155, "loss": 3.7889640808105467, "grad_norm": 0.6631152629852295, "learning_rate": 0.0001, "elbo": 3.78896, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2992610837438423}
{"step": 22160, "loss": 3.7954551696777346, "grad_norm": 0.742602527141571, "learning_rate": 0.0001, "elbo": 3.79545, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.299554304480413}
{"step": 22165, "loss": 3.791921615600586, "grad_norm": 0.6685926914215088, "learning_rate": 0.0001, "elbo": 3.79192, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.2998475252169834}
{"step": 22170, "loss": 3.7919033050537108, "grad_norm": 0.7324373126029968, "learning_rate": 0.0001, "elbo": 3.7919, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.300140745953554}
{"step": 22175, "loss": 3.789813232421875, "grad_norm": 0.691849410533905, "learning_rate": 0.0001, "elbo": 3.78981, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3004339666901243}
{"step": 22180, "loss": 3.7774654388427735, "grad_norm": 0.7572858333587646, "learning_rate": 0.0001, "elbo": 3.77747, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3007271874266948}
{"step": 22185, "loss": 3.7854019165039063, "grad_norm": 0.7295211553573608, "learning_rate": 0.0001, "elbo": 3.7854, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3010204081632653}
{"step": 22190, "loss": 3.7814258575439452, "grad_norm": 0.6478199362754822, "learning_rate": 0.0001, "elbo": 3.78143, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3013136288998357}
{"step": 22195, "loss": 3.7827178955078127, "grad_norm": 0.6146789789199829, "learning_rate": 0.0001, "elbo": 3.78272, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3016068496364062}
{"step": 22195, "loss": 3.7827178955078127, "grad_norm": 0.6146789789199829, "learning_rate": 0.0001, "elbo": 3.78272, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3016068496364062}
{"step": 22200, "loss": 3.7883758544921875, "grad_norm": 0.8297249674797058, "learning_rate": 0.0001, "elbo": 3.78838, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3019000703729768}
{"step": 22205, "loss": 3.7829864501953123, "grad_norm": 0.7191992402076721, "learning_rate": 0.0001, "elbo": 3.78299, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3021932911095473}
{"step": 22210, "loss": 3.7926376342773436, "grad_norm": 0.7044093608856201, "learning_rate": 0.0001, "elbo": 3.79264, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3024865118461177}
{"step": 22215, "loss": 3.80020751953125, "grad_norm": 0.6712290048599243, "learning_rate": 0.0001, "elbo": 3.80021, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3027797325826882}
{"step": 22215, "loss": 3.80020751953125, "grad_norm": 0.6712290048599243, "learning_rate": 0.0001, "elbo": 3.80021, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3027797325826882}
{"step": 22220, "loss": 3.7926872253417967, "grad_norm": 0.7018858194351196, "learning_rate": 0.0001, "elbo": 3.79269, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3030729533192589}
{"step": 22225, "loss": 3.7937881469726564, "grad_norm": 0.7188083529472351, "learning_rate": 0.0001, "elbo": 3.79379, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3033661740558293}
{"step": 22230, "loss": 3.7755218505859376, "grad_norm": 0.6972231864929199, "learning_rate": 0.0001, "elbo": 3.77552, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3036593947923998}
{"step": 22230, "loss": 3.7755218505859376, "grad_norm": 0.6972231864929199, "learning_rate": 0.0001, "elbo": 3.77552, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3036593947923998}
{"step": 22235, "loss": 3.8086395263671875, "grad_norm": 0.6734710335731506, "learning_rate": 0.0001, "elbo": 3.80864, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3039526155289702}
{"step": 22240, "loss": 3.787347412109375, "grad_norm": 0.6689147353172302, "learning_rate": 0.0001, "elbo": 3.78735, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3042458362655407}
{"step": 22245, "loss": 3.7907310485839845, "grad_norm": 0.8069818615913391, "learning_rate": 0.0001, "elbo": 3.79073, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3045390570021111}
{"step": 22250, "loss": 3.803620147705078, "grad_norm": 0.7500664591789246, "learning_rate": 0.0001, "elbo": 3.80362, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3048322777386816}
{"step": 22255, "loss": 3.7879989624023436, "grad_norm": 0.6755149364471436, "learning_rate": 0.0001, "elbo": 3.788, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.305125498475252}
{"step": 22260, "loss": 3.7772117614746095, "grad_norm": 0.7285175919532776, "learning_rate": 0.0001, "elbo": 3.77721, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3054187192118227}
{"step": 22265, "loss": 3.778804397583008, "grad_norm": 0.6821852922439575, "learning_rate": 0.0001, "elbo": 3.7788, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3057119399483932}
{"step": 22270, "loss": 3.797166442871094, "grad_norm": 0.6437917947769165, "learning_rate": 0.0001, "elbo": 3.79717, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3060051606849636}
{"step": 22275, "loss": 3.7987762451171876, "grad_norm": 0.7787958383560181, "learning_rate": 0.0001, "elbo": 3.79878, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.306298381421534}
{"step": 22280, "loss": 3.7984947204589843, "grad_norm": 1.4005413055419922, "learning_rate": 0.0001, "elbo": 3.79849, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3065916021581045}
{"step": 22285, "loss": 3.796466064453125, "grad_norm": 0.7812937498092651, "learning_rate": 0.0001, "elbo": 3.79647, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3068848228946752}
{"step": 22290, "loss": 3.7835243225097654, "grad_norm": 0.6726865768432617, "learning_rate": 0.0001, "elbo": 3.78352, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3071780436312457}
{"step": 22295, "loss": 3.788250732421875, "grad_norm": 0.8985199332237244, "learning_rate": 0.0001, "elbo": 3.78825, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3074712643678161}
{"step": 22300, "loss": 3.7873775482177736, "grad_norm": 0.6908599734306335, "learning_rate": 0.0001, "elbo": 3.78738, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3077644851043866}
{"step": 22305, "loss": 3.818544387817383, "grad_norm": 0.7969584465026855, "learning_rate": 0.0001, "elbo": 3.81854, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.308057705840957}
{"step": 22310, "loss": 3.783949279785156, "grad_norm": 0.6531299948692322, "learning_rate": 0.0001, "elbo": 3.78395, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3083509265775275}
{"step": 22315, "loss": 3.7831043243408202, "grad_norm": 0.6631463170051575, "learning_rate": 0.0001, "elbo": 3.7831, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.308644147314098}
{"step": 22320, "loss": 3.7894798278808595, "grad_norm": 1.028457522392273, "learning_rate": 0.0001, "elbo": 3.78948, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3089373680506686}
{"step": 22325, "loss": 3.783507537841797, "grad_norm": 0.6864944696426392, "learning_rate": 0.0001, "elbo": 3.78351, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.309230588787239}
{"step": 22330, "loss": 3.776673126220703, "grad_norm": 0.7168240547180176, "learning_rate": 0.0001, "elbo": 3.77667, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3095238095238095}
{"step": 22335, "loss": 3.8009090423583984, "grad_norm": 0.6767182946205139, "learning_rate": 0.0001, "elbo": 3.80091, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.30981703026038}
{"step": 22340, "loss": 3.792183685302734, "grad_norm": 0.6948320269584656, "learning_rate": 0.0001, "elbo": 3.79218, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3101102509969504}
{"step": 22345, "loss": 3.7764949798583984, "grad_norm": 0.698972225189209, "learning_rate": 0.0001, "elbo": 3.77649, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3104034717335211}
{"step": 22350, "loss": 3.7763877868652345, "grad_norm": 0.6582363843917847, "learning_rate": 0.0001, "elbo": 3.77639, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3106966924700916}
{"step": 22355, "loss": 3.772014617919922, "grad_norm": 0.6904993653297424, "learning_rate": 0.0001, "elbo": 3.77201, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.310989913206662}
{"step": 22355, "loss": 3.772014617919922, "grad_norm": 0.6904993653297424, "learning_rate": 0.0001, "elbo": 3.77201, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.310989913206662}
{"step": 22360, "loss": 3.789270782470703, "grad_norm": 1.1234256029129028, "learning_rate": 0.0001, "elbo": 3.78927, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3112831339432325}
{"step": 22365, "loss": 3.776639938354492, "grad_norm": 0.7416329979896545, "learning_rate": 0.0001, "elbo": 3.77664, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.311576354679803}
{"step": 22370, "loss": 3.77796630859375, "grad_norm": 0.725562334060669, "learning_rate": 0.0001, "elbo": 3.77797, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3118695754163734}
{"step": 22375, "loss": 3.780600357055664, "grad_norm": 0.6793258190155029, "learning_rate": 0.0001, "elbo": 3.7806, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3121627961529438}
{"step": 22380, "loss": 3.778482437133789, "grad_norm": 0.7278582453727722, "learning_rate": 0.0001, "elbo": 3.77848, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3124560168895143}
{"step": 22385, "loss": 3.7949092864990233, "grad_norm": 0.6827019453048706, "learning_rate": 0.0001, "elbo": 3.79491, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.312749237626085}
{"step": 22390, "loss": 3.767071533203125, "grad_norm": 0.6515797972679138, "learning_rate": 0.0001, "elbo": 3.76707, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3130424583626554}
{"step": 22395, "loss": 3.7591926574707033, "grad_norm": 0.7006931304931641, "learning_rate": 0.0001, "elbo": 3.75919, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3133356790992259}
{"step": 22400, "loss": 3.809511184692383, "grad_norm": 0.6736069917678833, "learning_rate": 0.0001, "elbo": 3.80951, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3136288998357963}
{"step": 22405, "loss": 3.7592864990234376, "grad_norm": 0.66091388463974, "learning_rate": 0.0001, "elbo": 3.75929, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.313922120572367}
{"step": 22410, "loss": 3.7729496002197265, "grad_norm": 0.7256795167922974, "learning_rate": 0.0001, "elbo": 3.77295, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3142153413089375}
{"step": 22415, "loss": 3.7842376708984373, "grad_norm": 0.6682985424995422, "learning_rate": 0.0001, "elbo": 3.78424, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.314508562045508}
{"step": 22420, "loss": 3.771940231323242, "grad_norm": 0.6906456351280212, "learning_rate": 0.0001, "elbo": 3.77194, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3148017827820784}
{"step": 22425, "loss": 3.777667999267578, "grad_norm": 0.7097682356834412, "learning_rate": 0.0001, "elbo": 3.77767, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3150950035186488}
{"step": 22430, "loss": 3.787541961669922, "grad_norm": 0.7538115978240967, "learning_rate": 0.0001, "elbo": 3.78754, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3153882242552193}
{"step": 22435, "loss": 3.7880264282226563, "grad_norm": 0.6761847138404846, "learning_rate": 0.0001, "elbo": 3.78803, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3156814449917897}
{"step": 22440, "loss": 3.7774463653564454, "grad_norm": 0.688752293586731, "learning_rate": 0.0001, "elbo": 3.77745, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3159746657283602}
{"step": 22445, "loss": 3.7944427490234376, "grad_norm": 0.6791921257972717, "learning_rate": 0.0001, "elbo": 3.79444, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3162678864649309}
{"step": 22450, "loss": 3.776153564453125, "grad_norm": 0.7098541855812073, "learning_rate": 0.0001, "elbo": 3.77615, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3165611072015013}
{"step": 22455, "loss": 3.7999187469482423, "grad_norm": 0.7063648104667664, "learning_rate": 0.0001, "elbo": 3.79992, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3168543279380718}
{"step": 22460, "loss": 3.7725204467773437, "grad_norm": 0.6628684997558594, "learning_rate": 0.0001, "elbo": 3.77252, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3171475486746422}
{"step": 22465, "loss": 3.7788742065429686, "grad_norm": 0.6533799767494202, "learning_rate": 0.0001, "elbo": 3.77887, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3174407694112127}
{"step": 22470, "loss": 3.7885948181152345, "grad_norm": 0.7098302245140076, "learning_rate": 0.0001, "elbo": 3.78859, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3177339901477834}
{"step": 22475, "loss": 3.77942008972168, "grad_norm": 0.683466911315918, "learning_rate": 0.0001, "elbo": 3.77942, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3180272108843538}
{"step": 22480, "loss": 3.7665756225585936, "grad_norm": 0.7416226863861084, "learning_rate": 0.0001, "elbo": 3.76658, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3183204316209243}
{"step": 22485, "loss": 3.7769401550292967, "grad_norm": 0.6842721700668335, "learning_rate": 0.0001, "elbo": 3.77694, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3186136523574947}
{"step": 22490, "loss": 3.7713367462158205, "grad_norm": 0.7460396885871887, "learning_rate": 0.0001, "elbo": 3.77134, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3189068730940652}
{"step": 22495, "loss": 3.7753273010253907, "grad_norm": 0.6632728576660156, "learning_rate": 0.0001, "elbo": 3.77533, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3192000938306356}
{"step": 22500, "loss": 3.7872772216796875, "grad_norm": 0.6988458633422852, "learning_rate": 0.0001, "elbo": 3.78728, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.319493314567206}
{"step": 22505, "loss": 3.7689300537109376, "grad_norm": 0.7037648558616638, "learning_rate": 0.0001, "elbo": 3.76893, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3197865353037768}
{"step": 22510, "loss": 3.781512451171875, "grad_norm": 0.7219316959381104, "learning_rate": 0.0001, "elbo": 3.78151, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3200797560403472}
{"step": 22515, "loss": 3.7656982421875, "grad_norm": 0.799224317073822, "learning_rate": 0.0001, "elbo": 3.7657, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3203729767769177}
{"step": 22520, "loss": 3.779737091064453, "grad_norm": 0.7132251262664795, "learning_rate": 0.0001, "elbo": 3.77974, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3206661975134881}
{"step": 22525, "loss": 3.7937973022460936, "grad_norm": 0.7033119797706604, "learning_rate": 0.0001, "elbo": 3.7938, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3209594182500586}
{"step": 22530, "loss": 3.7761180877685545, "grad_norm": 0.7236183881759644, "learning_rate": 0.0001, "elbo": 3.77612, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3212526389866293}
{"step": 22535, "loss": 3.785077667236328, "grad_norm": 0.8729533553123474, "learning_rate": 0.0001, "elbo": 3.78508, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3215458597231997}
{"step": 22540, "loss": 3.7782161712646483, "grad_norm": 0.8142313957214355, "learning_rate": 0.0001, "elbo": 3.77822, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3218390804597702}
{"step": 22545, "loss": 3.7975818634033205, "grad_norm": 0.6466678977012634, "learning_rate": 0.0001, "elbo": 3.79758, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3221323011963406}
{"step": 22550, "loss": 3.782012939453125, "grad_norm": 0.712091863155365, "learning_rate": 0.0001, "elbo": 3.78201, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.322425521932911}
{"step": 22555, "loss": 3.8016468048095704, "grad_norm": 0.6693078279495239, "learning_rate": 0.0001, "elbo": 3.80165, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3227187426694815}
{"step": 22560, "loss": 3.7622337341308594, "grad_norm": 0.6776800155639648, "learning_rate": 0.0001, "elbo": 3.76223, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.323011963406052}
{"step": 22560, "loss": 3.7622337341308594, "grad_norm": 0.6776800155639648, "learning_rate": 0.0001, "elbo": 3.76223, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.323011963406052}
{"step": 22565, "loss": 3.779044342041016, "grad_norm": 0.754413902759552, "learning_rate": 0.0001, "elbo": 3.77904, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3233051841426224}
{"step": 22570, "loss": 3.777741241455078, "grad_norm": 0.6835095882415771, "learning_rate": 0.0001, "elbo": 3.77774, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3235984048791931}
{"step": 22575, "loss": 3.7832481384277346, "grad_norm": 0.6659643054008484, "learning_rate": 0.0001, "elbo": 3.78325, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3238916256157636}
{"step": 22575, "loss": 3.7832481384277346, "grad_norm": 0.6659643054008484, "learning_rate": 0.0001, "elbo": 3.78325, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3238916256157636}
{"step": 22580, "loss": 3.77794189453125, "grad_norm": 0.8024023771286011, "learning_rate": 0.0001, "elbo": 3.77794, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.324184846352334}
{"step": 22585, "loss": 3.777001953125, "grad_norm": 0.6676190495491028, "learning_rate": 0.0001, "elbo": 3.777, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3244780670889045}
{"step": 22590, "loss": 3.798345947265625, "grad_norm": 0.7280828356742859, "learning_rate": 0.0001, "elbo": 3.79835, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3247712878254752}
{"step": 22595, "loss": 3.7953346252441404, "grad_norm": 0.6602184772491455, "learning_rate": 0.0001, "elbo": 3.79533, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3250645085620456}
{"step": 22600, "loss": 3.7879493713378904, "grad_norm": 0.7775778770446777, "learning_rate": 0.0001, "elbo": 3.78795, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.325357729298616}
{"step": 22605, "loss": 3.7799930572509766, "grad_norm": 0.7825228571891785, "learning_rate": 0.0001, "elbo": 3.77999, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3256509500351865}
{"step": 22610, "loss": 3.7922210693359375, "grad_norm": 0.6373856067657471, "learning_rate": 0.0001, "elbo": 3.79222, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.325944170771757}
{"step": 22610, "loss": 3.7922210693359375, "grad_norm": 0.6373856067657471, "learning_rate": 0.0001, "elbo": 3.79222, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.325944170771757}
{"step": 22615, "loss": 3.7710548400878907, "grad_norm": 0.739658534526825, "learning_rate": 0.0001, "elbo": 3.77106, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3262373915083274}
{"step": 22620, "loss": 3.779586410522461, "grad_norm": 0.6648533344268799, "learning_rate": 0.0001, "elbo": 3.77959, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3265306122448979}
{"step": 22620, "loss": 3.779586410522461, "grad_norm": 0.6648533344268799, "learning_rate": 0.0001, "elbo": 3.77959, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3265306122448979}
{"step": 22625, "loss": 3.7685214996337892, "grad_norm": 0.7445201873779297, "learning_rate": 0.0001, "elbo": 3.76852, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3268238329814683}
{"step": 22630, "loss": 3.786962890625, "grad_norm": 0.6627815961837769, "learning_rate": 0.0001, "elbo": 3.78696, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.327117053718039}
{"step": 22635, "loss": 3.777388000488281, "grad_norm": 0.7098604440689087, "learning_rate": 0.0001, "elbo": 3.77739, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3274102744546095}
{"step": 22640, "loss": 3.773539352416992, "grad_norm": 0.6681236624717712, "learning_rate": 0.0001, "elbo": 3.77354, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.32770349519118}
{"step": 22645, "loss": 3.7735740661621096, "grad_norm": 0.7687448263168335, "learning_rate": 0.0001, "elbo": 3.77357, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3279967159277504}
{"step": 22650, "loss": 3.7684165954589846, "grad_norm": 0.7273839712142944, "learning_rate": 0.0001, "elbo": 3.76842, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3282899366643208}
{"step": 22655, "loss": 3.7857124328613283, "grad_norm": 0.697815477848053, "learning_rate": 0.0001, "elbo": 3.78571, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3285831574008915}
{"step": 22660, "loss": 3.779247283935547, "grad_norm": 0.7033748626708984, "learning_rate": 0.0001, "elbo": 3.77925, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.328876378137462}
{"step": 22660, "loss": 3.779247283935547, "grad_norm": 0.7033748626708984, "learning_rate": 0.0001, "elbo": 3.77925, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.328876378137462}
{"step": 22665, "loss": 3.799453353881836, "grad_norm": 0.662294328212738, "learning_rate": 0.0001, "elbo": 3.79945, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3291695988740324}
{"step": 22670, "loss": 3.7906444549560545, "grad_norm": 0.6797420382499695, "learning_rate": 0.0001, "elbo": 3.79064, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3294628196106029}
{"step": 22675, "loss": 3.779249572753906, "grad_norm": 0.8666646480560303, "learning_rate": 0.0001, "elbo": 3.77925, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3297560403471733}
{"step": 22680, "loss": 3.7808738708496095, "grad_norm": 0.6833432912826538, "learning_rate": 0.0001, "elbo": 3.78087, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3300492610837438}
{"step": 22685, "loss": 3.78419189453125, "grad_norm": 0.6954001784324646, "learning_rate": 0.0001, "elbo": 3.78419, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3303424818203142}
{"step": 22690, "loss": 3.7879356384277343, "grad_norm": 0.6686286330223083, "learning_rate": 0.0001, "elbo": 3.78794, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.330635702556885}
{"step": 22695, "loss": 3.7728538513183594, "grad_norm": 0.6959764361381531, "learning_rate": 0.0001, "elbo": 3.77285, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3309289232934554}
{"step": 22700, "loss": 3.782865524291992, "grad_norm": 0.6713268160820007, "learning_rate": 0.0001, "elbo": 3.78287, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3312221440300258}
{"step": 22705, "loss": 3.7876045227050783, "grad_norm": 0.6354326009750366, "learning_rate": 0.0001, "elbo": 3.7876, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3315153647665963}
{"step": 22710, "loss": 3.7739284515380858, "grad_norm": 0.6691488027572632, "learning_rate": 0.0001, "elbo": 3.77393, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3318085855031667}
{"step": 22715, "loss": 3.764446258544922, "grad_norm": 0.6900315880775452, "learning_rate": 0.0001, "elbo": 3.76445, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3321018062397374}
{"step": 22720, "loss": 3.7742137908935547, "grad_norm": 0.7024598121643066, "learning_rate": 0.0001, "elbo": 3.77421, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3323950269763079}
{"step": 22725, "loss": 3.78619499206543, "grad_norm": 0.645836353302002, "learning_rate": 0.0001, "elbo": 3.7862, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3326882477128783}
{"step": 22730, "loss": 3.7874359130859374, "grad_norm": 0.7468175888061523, "learning_rate": 0.0001, "elbo": 3.78744, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3329814684494488}
{"step": 22735, "loss": 3.7813385009765623, "grad_norm": 0.6552481651306152, "learning_rate": 0.0001, "elbo": 3.78134, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3332746891860192}
{"step": 22740, "loss": 3.771175003051758, "grad_norm": 0.6985365748405457, "learning_rate": 0.0001, "elbo": 3.77117, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3335679099225897}
{"step": 22740, "loss": 3.771175003051758, "grad_norm": 0.6985365748405457, "learning_rate": 0.0001, "elbo": 3.77117, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3335679099225897}
{"step": 22745, "loss": 3.7840972900390626, "grad_norm": 0.8905091881752014, "learning_rate": 0.0001, "elbo": 3.7841, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3338611306591601}
{"step": 22750, "loss": 3.7779163360595702, "grad_norm": 0.6889037489891052, "learning_rate": 0.0001, "elbo": 3.77792, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3341543513957306}
{"step": 22755, "loss": 3.7778343200683593, "grad_norm": 0.7153809070587158, "learning_rate": 0.0001, "elbo": 3.77783, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3344475721323013}
{"step": 22760, "loss": 3.79285888671875, "grad_norm": 0.736682653427124, "learning_rate": 0.0001, "elbo": 3.79286, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3347407928688717}
{"step": 22765, "loss": 3.7598648071289062, "grad_norm": 0.7543807029724121, "learning_rate": 0.0001, "elbo": 3.75986, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3350340136054422}
{"step": 22770, "loss": 3.7790248870849608, "grad_norm": 0.6540776491165161, "learning_rate": 0.0001, "elbo": 3.77902, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3353272343420126}
{"step": 22775, "loss": 3.793730926513672, "grad_norm": 0.6309193968772888, "learning_rate": 0.0001, "elbo": 3.79373, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3356204550785833}
{"step": 22780, "loss": 3.7750320434570312, "grad_norm": 0.7594279646873474, "learning_rate": 0.0001, "elbo": 3.77503, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3359136758151537}
{"step": 22785, "loss": 3.7763946533203123, "grad_norm": 0.6762232780456543, "learning_rate": 0.0001, "elbo": 3.77639, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3362068965517242}
{"step": 22790, "loss": 3.784127044677734, "grad_norm": 0.9309712648391724, "learning_rate": 0.0001, "elbo": 3.78413, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3365001172882947}
{"step": 22795, "loss": 3.7916046142578126, "grad_norm": 0.8096725940704346, "learning_rate": 0.0001, "elbo": 3.7916, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.336793338024865}
{"step": 22800, "loss": 3.7774940490722657, "grad_norm": 0.7491523623466492, "learning_rate": 0.0001, "elbo": 3.77749, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3370865587614356}
{"step": 22805, "loss": 3.7812240600585936, "grad_norm": 0.728827714920044, "learning_rate": 0.0001, "elbo": 3.78122, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.337379779498006}
{"step": 22810, "loss": 3.779147720336914, "grad_norm": 0.7565467953681946, "learning_rate": 0.0001, "elbo": 3.77915, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3376730002345765}
{"step": 22815, "loss": 3.7807140350341797, "grad_norm": 0.6582791209220886, "learning_rate": 0.0001, "elbo": 3.78071, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3379662209711471}
{"step": 22820, "loss": 3.773789978027344, "grad_norm": 0.6689926981925964, "learning_rate": 0.0001, "elbo": 3.77379, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3382594417077176}
{"step": 22825, "loss": 3.779042053222656, "grad_norm": 0.69450843334198, "learning_rate": 0.0001, "elbo": 3.77904, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.338552662444288}
{"step": 22830, "loss": 3.769338607788086, "grad_norm": 0.6886058449745178, "learning_rate": 0.0001, "elbo": 3.76934, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3388458831808585}
{"step": 22835, "loss": 3.782582092285156, "grad_norm": 0.682080090045929, "learning_rate": 0.0001, "elbo": 3.78258, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.339139103917429}
{"step": 22840, "loss": 3.7803131103515626, "grad_norm": 0.7714529037475586, "learning_rate": 0.0001, "elbo": 3.78031, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3394323246539996}
{"step": 22845, "loss": 3.7814987182617186, "grad_norm": 0.6670547127723694, "learning_rate": 0.0001, "elbo": 3.7815, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.33972554539057}
{"step": 22850, "loss": 3.773788070678711, "grad_norm": 0.6839338541030884, "learning_rate": 0.0001, "elbo": 3.77379, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3400187661271405}
{"step": 22855, "loss": 3.7794116973876952, "grad_norm": 0.6350308656692505, "learning_rate": 0.0001, "elbo": 3.77941, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.340311986863711}
{"step": 22860, "loss": 3.7862720489501953, "grad_norm": 0.6835668683052063, "learning_rate": 0.0001, "elbo": 3.78627, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3406052076002815}
{"step": 22865, "loss": 3.7652313232421877, "grad_norm": 0.6999266743659973, "learning_rate": 0.0001, "elbo": 3.76523, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.340898428336852}
{"step": 22870, "loss": 3.7944183349609375, "grad_norm": 0.692973256111145, "learning_rate": 0.0001, "elbo": 3.79442, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3411916490734224}
{"step": 22875, "loss": 3.7894126892089846, "grad_norm": 0.640805184841156, "learning_rate": 0.0001, "elbo": 3.78941, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.341484869809993}
{"step": 22875, "loss": 3.7894126892089846, "grad_norm": 0.640805184841156, "learning_rate": 0.0001, "elbo": 3.78941, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.341484869809993}
{"step": 22880, "loss": 3.781808853149414, "grad_norm": 0.7119837999343872, "learning_rate": 0.0001, "elbo": 3.78181, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3417780905465635}
{"step": 22885, "loss": 3.7962879180908202, "grad_norm": 0.6845125555992126, "learning_rate": 0.0001, "elbo": 3.79629, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.342071311283134}
{"step": 22890, "loss": 3.765740966796875, "grad_norm": 0.6688316464424133, "learning_rate": 0.0001, "elbo": 3.76574, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3423645320197044}
{"step": 22895, "loss": 3.7762237548828126, "grad_norm": 0.6812463402748108, "learning_rate": 0.0001, "elbo": 3.77622, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3426577527562749}
{"step": 22900, "loss": 3.7921722412109373, "grad_norm": 0.7150474190711975, "learning_rate": 0.0001, "elbo": 3.79217, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3429509734928455}
{"step": 22905, "loss": 3.7900985717773437, "grad_norm": 0.7985884547233582, "learning_rate": 0.0001, "elbo": 3.7901, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.343244194229416}
{"step": 22910, "loss": 3.7867977142333986, "grad_norm": 0.7868633270263672, "learning_rate": 0.0001, "elbo": 3.7868, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3435374149659864}
{"step": 22915, "loss": 3.7685306549072264, "grad_norm": 0.6884799599647522, "learning_rate": 0.0001, "elbo": 3.76853, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.343830635702557}
{"step": 22920, "loss": 3.775309371948242, "grad_norm": 0.7074761986732483, "learning_rate": 0.0001, "elbo": 3.77531, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3441238564391274}
{"step": 22925, "loss": 3.7857292175292967, "grad_norm": 0.6492458581924438, "learning_rate": 0.0001, "elbo": 3.78573, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3444170771756978}
{"step": 22930, "loss": 3.7754531860351563, "grad_norm": 0.9659838080406189, "learning_rate": 0.0001, "elbo": 3.77545, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3447102979122683}
{"step": 22935, "loss": 3.7844093322753904, "grad_norm": 0.8500522375106812, "learning_rate": 0.0001, "elbo": 3.78441, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3450035186488387}
{"step": 22940, "loss": 3.8066619873046874, "grad_norm": 0.6888828277587891, "learning_rate": 0.0001, "elbo": 3.80666, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3452967393854094}
{"step": 22945, "loss": 3.785840606689453, "grad_norm": 0.6684624552726746, "learning_rate": 0.0001, "elbo": 3.78584, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3455899601219798}
{"step": 22950, "loss": 3.779603958129883, "grad_norm": 0.6868674755096436, "learning_rate": 0.0001, "elbo": 3.7796, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3458831808585503}
{"step": 22955, "loss": 3.7705516815185547, "grad_norm": 0.7090305685997009, "learning_rate": 0.0001, "elbo": 3.77055, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3461764015951208}
{"step": 22960, "loss": 3.7727687835693358, "grad_norm": 0.6634979844093323, "learning_rate": 0.0001, "elbo": 3.77277, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3464696223316914}
{"step": 22965, "loss": 3.7876007080078127, "grad_norm": 0.68726646900177, "learning_rate": 0.0001, "elbo": 3.7876, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3467628430682619}
{"step": 22970, "loss": 3.786878204345703, "grad_norm": 0.7008379101753235, "learning_rate": 0.0001, "elbo": 3.78688, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3470560638048323}
{"step": 22975, "loss": 3.7883926391601563, "grad_norm": 0.6526645421981812, "learning_rate": 0.0001, "elbo": 3.78839, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3473492845414028}
{"step": 22980, "loss": 3.776056671142578, "grad_norm": 0.6875152587890625, "learning_rate": 0.0001, "elbo": 3.77606, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3476425052779732}
{"step": 22985, "loss": 3.7966850280761717, "grad_norm": 0.6395068168640137, "learning_rate": 0.0001, "elbo": 3.79669, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3479357260145437}
{"step": 22990, "loss": 3.761588287353516, "grad_norm": 0.8040180802345276, "learning_rate": 0.0001, "elbo": 3.76159, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3482289467511142}
{"step": 22995, "loss": 3.776417922973633, "grad_norm": 0.7363704442977905, "learning_rate": 0.0001, "elbo": 3.77642, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3485221674876846}
{"step": 23000, "loss": 3.7777000427246095, "grad_norm": 0.7038154602050781, "learning_rate": 0.0001, "elbo": 3.7777, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3488153882242553}
{"step": 23005, "loss": 3.783216857910156, "grad_norm": 0.7014189958572388, "learning_rate": 0.0001, "elbo": 3.78322, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3491086089608257}
{"step": 23010, "loss": 3.7779693603515625, "grad_norm": 0.7555910348892212, "learning_rate": 0.0001, "elbo": 3.77797, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3494018296973962}
{"step": 23010, "loss": 3.7779693603515625, "grad_norm": 0.7555910348892212, "learning_rate": 0.0001, "elbo": 3.77797, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3494018296973962}
{"step": 23015, "loss": 3.7775337219238283, "grad_norm": 0.8282755613327026, "learning_rate": 0.0001, "elbo": 3.77753, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3496950504339666}
{"step": 23020, "loss": 3.7853572845458983, "grad_norm": 0.6917625665664673, "learning_rate": 0.0001, "elbo": 3.78536, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.349988271170537}
{"step": 23025, "loss": 3.7861618041992187, "grad_norm": 0.6305220723152161, "learning_rate": 0.0001, "elbo": 3.78616, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3502814919071078}
{"step": 23030, "loss": 3.765837478637695, "grad_norm": 0.6447082161903381, "learning_rate": 0.0001, "elbo": 3.76584, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3505747126436782}
{"step": 23035, "loss": 3.786992645263672, "grad_norm": 0.7552415728569031, "learning_rate": 0.0001, "elbo": 3.78699, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3508679333802487}
{"step": 23040, "loss": 3.8058547973632812, "grad_norm": 1.0553534030914307, "learning_rate": 0.0001, "elbo": 3.80585, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3511611541168191}
{"step": 23045, "loss": 3.772313690185547, "grad_norm": 0.7700760960578918, "learning_rate": 0.0001, "elbo": 3.77231, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3514543748533896}
{"step": 23050, "loss": 3.7771942138671877, "grad_norm": 0.8020511269569397, "learning_rate": 0.0001, "elbo": 3.77719, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.35174759558996}
{"step": 23055, "loss": 3.782674026489258, "grad_norm": 0.7043883800506592, "learning_rate": 0.0001, "elbo": 3.78267, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3520408163265305}
{"step": 23060, "loss": 3.775757598876953, "grad_norm": 0.6659495234489441, "learning_rate": 0.0001, "elbo": 3.77576, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3523340370631012}
{"step": 23065, "loss": 3.7673484802246096, "grad_norm": 0.6470363140106201, "learning_rate": 0.0001, "elbo": 3.76735, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3526272577996716}
{"step": 23070, "loss": 3.7755977630615236, "grad_norm": 0.6494737267494202, "learning_rate": 0.0001, "elbo": 3.7756, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.352920478536242}
{"step": 23075, "loss": 3.774959945678711, "grad_norm": 0.7109453082084656, "learning_rate": 0.0001, "elbo": 3.77496, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3532136992728125}
{"step": 23080, "loss": 3.7816001892089846, "grad_norm": 0.713295578956604, "learning_rate": 0.0001, "elbo": 3.7816, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.353506920009383}
{"step": 23085, "loss": 3.7839195251464846, "grad_norm": 0.6734545826911926, "learning_rate": 0.0001, "elbo": 3.78392, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3538001407459537}
{"step": 23090, "loss": 3.7703712463378904, "grad_norm": 0.7268093824386597, "learning_rate": 0.0001, "elbo": 3.77037, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3540933614825241}
{"step": 23095, "loss": 3.778662109375, "grad_norm": 0.7056458592414856, "learning_rate": 0.0001, "elbo": 3.77866, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3543865822190946}
{"step": 23100, "loss": 3.78362922668457, "grad_norm": 0.7749548554420471, "learning_rate": 0.0001, "elbo": 3.78363, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.354679802955665}
{"step": 23105, "loss": 3.7816207885742186, "grad_norm": 0.891358494758606, "learning_rate": 0.0001, "elbo": 3.78162, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3549730236922355}
{"step": 23110, "loss": 3.779998016357422, "grad_norm": 0.6848501563072205, "learning_rate": 0.0001, "elbo": 3.78, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.355266244428806}
{"step": 23115, "loss": 3.78773193359375, "grad_norm": 0.6959173083305359, "learning_rate": 0.0001, "elbo": 3.78773, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3555594651653764}
{"step": 23120, "loss": 3.7787090301513673, "grad_norm": 0.7132036089897156, "learning_rate": 0.0001, "elbo": 3.77871, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3558526859019469}
{"step": 23125, "loss": 3.7841781616210937, "grad_norm": 0.684995710849762, "learning_rate": 0.0001, "elbo": 3.78418, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3561459066385175}
{"step": 23130, "loss": 3.776628875732422, "grad_norm": 0.6737823486328125, "learning_rate": 0.0001, "elbo": 3.77663, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.356439127375088}
{"step": 23135, "loss": 3.7546012878417967, "grad_norm": 0.8036414980888367, "learning_rate": 0.0001, "elbo": 3.7546, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3567323481116584}
{"step": 23135, "loss": 3.7546012878417967, "grad_norm": 0.8036414980888367, "learning_rate": 0.0001, "elbo": 3.7546, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3567323481116584}
{"step": 23140, "loss": 3.766240692138672, "grad_norm": 0.7177730202674866, "learning_rate": 0.0001, "elbo": 3.76624, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.357025568848229}
{"step": 23145, "loss": 3.7592193603515627, "grad_norm": 0.7366659045219421, "learning_rate": 0.0001, "elbo": 3.75922, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3573187895847996}
{"step": 23150, "loss": 3.779180145263672, "grad_norm": 0.6656755805015564, "learning_rate": 0.0001, "elbo": 3.77918, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.35761201032137}
{"step": 23155, "loss": 3.7830955505371096, "grad_norm": 0.7523155212402344, "learning_rate": 0.0001, "elbo": 3.7831, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3579052310579405}
{"step": 23160, "loss": 3.7771011352539063, "grad_norm": 0.732193648815155, "learning_rate": 0.0001, "elbo": 3.7771, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.358198451794511}
{"step": 23165, "loss": 3.7841236114501955, "grad_norm": 0.6636463403701782, "learning_rate": 0.0001, "elbo": 3.78412, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3584916725310814}
{"step": 23170, "loss": 3.7796096801757812, "grad_norm": 0.8648072481155396, "learning_rate": 0.0001, "elbo": 3.77961, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3587848932676518}
{"step": 23175, "loss": 3.76922492980957, "grad_norm": 0.7375538349151611, "learning_rate": 0.0001, "elbo": 3.76922, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3590781140042223}
{"step": 23180, "loss": 3.767049026489258, "grad_norm": 0.6764370799064636, "learning_rate": 0.0001, "elbo": 3.76705, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3593713347407927}
{"step": 23185, "loss": 3.775321197509766, "grad_norm": 0.6585603356361389, "learning_rate": 0.0001, "elbo": 3.77532, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3596645554773634}
{"step": 23190, "loss": 3.766824722290039, "grad_norm": 0.9364517331123352, "learning_rate": 0.0001, "elbo": 3.76682, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3599577762139339}
{"step": 23195, "loss": 3.781138229370117, "grad_norm": 0.6969500184059143, "learning_rate": 0.0001, "elbo": 3.78114, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3602509969505043}
{"step": 23200, "loss": 3.7746959686279298, "grad_norm": 0.6820330023765564, "learning_rate": 0.0001, "elbo": 3.7747, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3605442176870748}
{"step": 23205, "loss": 3.7843978881835936, "grad_norm": 0.7129091620445251, "learning_rate": 0.0001, "elbo": 3.7844, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3608374384236452}
{"step": 23210, "loss": 3.771722412109375, "grad_norm": 0.6809176206588745, "learning_rate": 0.0001, "elbo": 3.77172, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.361130659160216}
{"step": 23215, "loss": 3.771531677246094, "grad_norm": 0.8797019720077515, "learning_rate": 0.0001, "elbo": 3.77153, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3614238798967864}
{"step": 23220, "loss": 3.791817855834961, "grad_norm": 0.6965803503990173, "learning_rate": 0.0001, "elbo": 3.79182, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3617171006333568}
{"step": 23225, "loss": 3.7779197692871094, "grad_norm": 0.733869731426239, "learning_rate": 0.0001, "elbo": 3.77792, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3620103213699273}
{"step": 23230, "loss": 3.7610374450683595, "grad_norm": 0.8114908337593079, "learning_rate": 0.0001, "elbo": 3.76104, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3623035421064977}
{"step": 23235, "loss": 3.768614959716797, "grad_norm": 0.7393900752067566, "learning_rate": 0.0001, "elbo": 3.76861, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3625967628430682}
{"step": 23240, "loss": 3.7877780914306642, "grad_norm": 0.6877612471580505, "learning_rate": 0.0001, "elbo": 3.78778, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3628899835796386}
{"step": 23245, "loss": 3.7838489532470705, "grad_norm": 1.027004361152649, "learning_rate": 0.0001, "elbo": 3.78385, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3631832043162093}
{"step": 23250, "loss": 3.791756439208984, "grad_norm": 0.726600170135498, "learning_rate": 0.0001, "elbo": 3.79176, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3634764250527798}
{"step": 23255, "loss": 3.7840286254882813, "grad_norm": 0.6675902009010315, "learning_rate": 0.0001, "elbo": 3.78403, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3637696457893502}
{"step": 23255, "loss": 3.7840286254882813, "grad_norm": 0.6675902009010315, "learning_rate": 0.0001, "elbo": 3.78403, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3637696457893502}
{"step": 23260, "loss": 3.7779830932617187, "grad_norm": 0.7309455275535583, "learning_rate": 0.0001, "elbo": 3.77798, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3640628665259207}
{"step": 23265, "loss": 3.785757064819336, "grad_norm": 0.6843768954277039, "learning_rate": 0.0001, "elbo": 3.78576, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3643560872624911}
{"step": 23270, "loss": 3.7841278076171876, "grad_norm": 0.670619785785675, "learning_rate": 0.0001, "elbo": 3.78413, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3646493079990618}
{"step": 23275, "loss": 3.7679542541503905, "grad_norm": 0.6904554963111877, "learning_rate": 0.0001, "elbo": 3.76795, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3649425287356323}
{"step": 23280, "loss": 3.766450881958008, "grad_norm": 0.7385360598564148, "learning_rate": 0.0001, "elbo": 3.76645, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3652357494722027}
{"step": 23285, "loss": 3.7743499755859373, "grad_norm": 0.7172656059265137, "learning_rate": 0.0001, "elbo": 3.77435, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3655289702087732}
{"step": 23290, "loss": 3.7691665649414063, "grad_norm": 0.6506245732307434, "learning_rate": 0.0001, "elbo": 3.76917, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3658221909453436}
{"step": 23295, "loss": 3.774403381347656, "grad_norm": 0.6699445247650146, "learning_rate": 0.0001, "elbo": 3.7744, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.366115411681914}
{"step": 23295, "loss": 3.774403381347656, "grad_norm": 0.6699445247650146, "learning_rate": 0.0001, "elbo": 3.7744, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.366115411681914}
{"step": 23300, "loss": 3.782850646972656, "grad_norm": 0.6662129163742065, "learning_rate": 0.0001, "elbo": 3.78285, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3664086324184845}
{"step": 23300, "loss": 3.782850646972656, "grad_norm": 0.6662129163742065, "learning_rate": 0.0001, "elbo": 3.78285, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3664086324184845}
{"step": 23305, "loss": 3.751158905029297, "grad_norm": 0.7243636250495911, "learning_rate": 0.0001, "elbo": 3.75116, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.366701853155055}
{"step": 23310, "loss": 3.764032745361328, "grad_norm": 0.7076050639152527, "learning_rate": 0.0001, "elbo": 3.76403, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3669950738916257}
{"step": 23315, "loss": 3.78555908203125, "grad_norm": 0.6628338694572449, "learning_rate": 0.0001, "elbo": 3.78556, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3672882946281961}
{"step": 23320, "loss": 3.7648868560791016, "grad_norm": 0.6519412994384766, "learning_rate": 0.0001, "elbo": 3.76489, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3675815153647666}
{"step": 23325, "loss": 3.7677337646484377, "grad_norm": 0.6941992044448853, "learning_rate": 0.0001, "elbo": 3.76773, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.367874736101337}
{"step": 23330, "loss": 3.7743133544921874, "grad_norm": 0.8355052471160889, "learning_rate": 0.0001, "elbo": 3.77431, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3681679568379077}
{"step": 23335, "loss": 3.7791786193847656, "grad_norm": 0.6907729506492615, "learning_rate": 0.0001, "elbo": 3.77918, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3684611775744782}
{"step": 23340, "loss": 3.790167236328125, "grad_norm": 0.7811532616615295, "learning_rate": 0.0001, "elbo": 3.79017, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3687543983110486}
{"step": 23345, "loss": 3.773224639892578, "grad_norm": 0.6502687335014343, "learning_rate": 0.0001, "elbo": 3.77322, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.369047619047619}
{"step": 23350, "loss": 3.773086929321289, "grad_norm": 0.6607136726379395, "learning_rate": 0.0001, "elbo": 3.77309, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3693408397841895}
{"step": 23355, "loss": 3.7755027770996095, "grad_norm": 0.7000938057899475, "learning_rate": 0.0001, "elbo": 3.7755, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.36963406052076}
{"step": 23360, "loss": 3.7720382690429686, "grad_norm": 0.7039998173713684, "learning_rate": 0.0001, "elbo": 3.77204, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3699272812573304}
{"step": 23365, "loss": 3.7987457275390626, "grad_norm": 0.7834692001342773, "learning_rate": 0.0001, "elbo": 3.79875, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3702205019939009}
{"step": 23370, "loss": 3.7760650634765627, "grad_norm": 0.6768327355384827, "learning_rate": 0.0001, "elbo": 3.77607, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3705137227304716}
{"step": 23375, "loss": 3.8006565093994142, "grad_norm": 0.6560941338539124, "learning_rate": 0.0001, "elbo": 3.80066, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.370806943467042}
{"step": 23380, "loss": 3.756201171875, "grad_norm": 0.9107746481895447, "learning_rate": 0.0001, "elbo": 3.7562, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3711001642036125}
{"step": 23385, "loss": 3.7748252868652346, "grad_norm": 0.7526695132255554, "learning_rate": 0.0001, "elbo": 3.77483, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.371393384940183}
{"step": 23390, "loss": 3.7653488159179687, "grad_norm": 0.8109628558158875, "learning_rate": 0.0001, "elbo": 3.76535, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3716866056767534}
{"step": 23395, "loss": 3.7737098693847657, "grad_norm": 0.8100466132164001, "learning_rate": 0.0001, "elbo": 3.77371, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.371979826413324}
{"step": 23400, "loss": 3.7654769897460936, "grad_norm": 0.7389854192733765, "learning_rate": 0.0001, "elbo": 3.76548, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3722730471498945}
{"step": 23405, "loss": 3.7632400512695314, "grad_norm": 0.642431914806366, "learning_rate": 0.0001, "elbo": 3.76324, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.372566267886465}
{"step": 23410, "loss": 3.7837265014648436, "grad_norm": 0.6997803449630737, "learning_rate": 0.0001, "elbo": 3.78373, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3728594886230354}
{"step": 23415, "loss": 3.7748344421386717, "grad_norm": 0.6843104362487793, "learning_rate": 0.0001, "elbo": 3.77483, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3731527093596059}
{"step": 23420, "loss": 3.7756114959716798, "grad_norm": 0.6765033006668091, "learning_rate": 0.0001, "elbo": 3.77561, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3734459300961763}
{"step": 23425, "loss": 3.747334289550781, "grad_norm": 0.6696180105209351, "learning_rate": 0.0001, "elbo": 3.74733, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3737391508327468}
{"step": 23430, "loss": 3.7800983428955077, "grad_norm": 1.0516002178192139, "learning_rate": 0.0001, "elbo": 3.7801, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3740323715693175}
{"step": 23435, "loss": 3.7596263885498047, "grad_norm": 0.6709269881248474, "learning_rate": 0.0001, "elbo": 3.75963, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.374325592305888}
{"step": 23435, "loss": 3.7596263885498047, "grad_norm": 0.6709269881248474, "learning_rate": 0.0001, "elbo": 3.75963, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.374325592305888}
{"step": 23440, "loss": 3.7840717315673826, "grad_norm": 0.6769269704818726, "learning_rate": 0.0001, "elbo": 3.78407, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3746188130424584}
{"step": 23445, "loss": 3.7939193725585936, "grad_norm": 0.6715276837348938, "learning_rate": 0.0001, "elbo": 3.79392, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3749120337790288}
{"step": 23445, "loss": 3.7939193725585936, "grad_norm": 0.6715276837348938, "learning_rate": 0.0001, "elbo": 3.79392, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3749120337790288}
{"step": 23450, "loss": 3.7701839447021483, "grad_norm": 0.8720026612281799, "learning_rate": 0.0001, "elbo": 3.77018, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3752052545155993}
{"step": 23455, "loss": 3.777456283569336, "grad_norm": 0.7018992900848389, "learning_rate": 0.0001, "elbo": 3.77746, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.37549847525217}
{"step": 23460, "loss": 3.7653934478759767, "grad_norm": 0.6660494804382324, "learning_rate": 0.0001, "elbo": 3.76539, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3757916959887404}
{"step": 23465, "loss": 3.776197814941406, "grad_norm": 0.6572046875953674, "learning_rate": 0.0001, "elbo": 3.7762, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3760849167253109}
{"step": 23470, "loss": 3.779367446899414, "grad_norm": 0.846930742263794, "learning_rate": 0.0001, "elbo": 3.77937, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3763781374618813}
{"step": 23475, "loss": 3.7763484954833983, "grad_norm": 0.8022632002830505, "learning_rate": 0.0001, "elbo": 3.77635, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3766713581984518}
{"step": 23480, "loss": 3.7768211364746094, "grad_norm": 0.6981464624404907, "learning_rate": 0.0001, "elbo": 3.77682, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3769645789350222}
{"step": 23485, "loss": 3.7686004638671875, "grad_norm": 0.6933873295783997, "learning_rate": 0.0001, "elbo": 3.7686, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3772577996715927}
{"step": 23490, "loss": 3.761937713623047, "grad_norm": 0.7112948894500732, "learning_rate": 0.0001, "elbo": 3.76194, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3775510204081631}
{"step": 23495, "loss": 3.7695823669433595, "grad_norm": 0.6481938362121582, "learning_rate": 0.0001, "elbo": 3.76958, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3778442411447338}
{"step": 23500, "loss": 3.7614410400390623, "grad_norm": 0.6615718603134155, "learning_rate": 0.0001, "elbo": 3.76144, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3781374618813043}
{"step": 23505, "loss": 3.7688667297363283, "grad_norm": 0.6640438437461853, "learning_rate": 0.0001, "elbo": 3.76887, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3784306826178747}
{"step": 23510, "loss": 3.7732517242431642, "grad_norm": 0.7056922316551208, "learning_rate": 0.0001, "elbo": 3.77325, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3787239033544452}
{"step": 23515, "loss": 3.767913818359375, "grad_norm": 0.8332071900367737, "learning_rate": 0.0001, "elbo": 3.76791, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3790171240910158}
{"step": 23520, "loss": 3.7809349060058595, "grad_norm": 0.7881593704223633, "learning_rate": 0.0001, "elbo": 3.78093, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3793103448275863}
{"step": 23520, "loss": 3.7809349060058595, "grad_norm": 0.7881593704223633, "learning_rate": 0.0001, "elbo": 3.78093, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3793103448275863}
{"step": 23525, "loss": 3.777831268310547, "grad_norm": 0.653412401676178, "learning_rate": 0.0001, "elbo": 3.77783, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3796035655641568}
{"step": 23530, "loss": 3.7612281799316407, "grad_norm": 0.7056159973144531, "learning_rate": 0.0001, "elbo": 3.76123, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3798967863007272}
{"step": 23535, "loss": 3.7682056427001953, "grad_norm": 0.7150720953941345, "learning_rate": 0.0001, "elbo": 3.76821, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3801900070372977}
{"step": 23540, "loss": 3.765595626831055, "grad_norm": 0.808369517326355, "learning_rate": 0.0001, "elbo": 3.7656, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3804832277738681}
{"step": 23545, "loss": 3.774995803833008, "grad_norm": 0.6942086815834045, "learning_rate": 0.0001, "elbo": 3.775, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3807764485104386}
{"step": 23550, "loss": 3.7779144287109374, "grad_norm": 0.7110505700111389, "learning_rate": 0.0001, "elbo": 3.77791, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.381069669247009}
{"step": 23550, "loss": 3.7779144287109374, "grad_norm": 0.7110505700111389, "learning_rate": 0.0001, "elbo": 3.77791, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.381069669247009}
{"step": 23555, "loss": 3.761981201171875, "grad_norm": 0.645569920539856, "learning_rate": 0.0001, "elbo": 3.76198, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3813628899835797}
{"step": 23560, "loss": 3.7822593688964843, "grad_norm": 0.739741861820221, "learning_rate": 0.0001, "elbo": 3.78226, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3816561107201502}
{"step": 23565, "loss": 3.772547149658203, "grad_norm": 0.6993306875228882, "learning_rate": 0.0001, "elbo": 3.77255, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3819493314567206}
{"step": 23570, "loss": 3.7655242919921874, "grad_norm": 0.7433454394340515, "learning_rate": 0.0001, "elbo": 3.76552, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.382242552193291}
{"step": 23575, "loss": 3.7686885833740233, "grad_norm": 0.9157864451408386, "learning_rate": 0.0001, "elbo": 3.76869, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3825357729298615}
{"step": 23580, "loss": 3.7707546234130858, "grad_norm": 0.673716127872467, "learning_rate": 0.0001, "elbo": 3.77075, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3828289936664322}
{"step": 23585, "loss": 3.7856494903564455, "grad_norm": 0.7827038764953613, "learning_rate": 0.0001, "elbo": 3.78565, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3831222144030026}
{"step": 23585, "loss": 3.7856494903564455, "grad_norm": 0.7827038764953613, "learning_rate": 0.0001, "elbo": 3.78565, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3831222144030026}
{"step": 23590, "loss": 3.769671630859375, "grad_norm": 0.7345212697982788, "learning_rate": 0.0001, "elbo": 3.76967, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.383415435139573}
{"step": 23595, "loss": 3.7684562683105467, "grad_norm": 0.6516481041908264, "learning_rate": 0.0001, "elbo": 3.76846, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3837086558761436}
{"step": 23600, "loss": 3.770245361328125, "grad_norm": 0.7541012763977051, "learning_rate": 0.0001, "elbo": 3.77025, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.384001876612714}
{"step": 23605, "loss": 3.772691345214844, "grad_norm": 0.7392411231994629, "learning_rate": 0.0001, "elbo": 3.77269, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3842950973492845}
{"step": 23610, "loss": 3.7580520629882814, "grad_norm": 0.6964028477668762, "learning_rate": 0.0001, "elbo": 3.75805, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.384588318085855}
{"step": 23615, "loss": 3.7735969543457033, "grad_norm": 0.8623406887054443, "learning_rate": 0.0001, "elbo": 3.7736, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3848815388224256}
{"step": 23620, "loss": 3.7763275146484374, "grad_norm": 0.7222048044204712, "learning_rate": 0.0001, "elbo": 3.77633, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.385174759558996}
{"step": 23625, "loss": 3.7648689270019533, "grad_norm": 0.706678569316864, "learning_rate": 0.0001, "elbo": 3.76487, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3854679802955665}
{"step": 23630, "loss": 3.7660430908203124, "grad_norm": 0.6955813765525818, "learning_rate": 0.0001, "elbo": 3.76604, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.385761201032137}
{"step": 23635, "loss": 3.7749858856201173, "grad_norm": 0.6129450798034668, "learning_rate": 0.0001, "elbo": 3.77499, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3860544217687074}
{"step": 23640, "loss": 3.765192413330078, "grad_norm": 0.8134040236473083, "learning_rate": 0.0001, "elbo": 3.76519, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.386347642505278}
{"step": 23645, "loss": 3.761328125, "grad_norm": 0.6463022828102112, "learning_rate": 0.0001, "elbo": 3.76133, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3866408632418485}
{"step": 23650, "loss": 3.767045593261719, "grad_norm": 0.7109518647193909, "learning_rate": 0.0001, "elbo": 3.76705, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.386934083978419}
{"step": 23655, "loss": 3.768485260009766, "grad_norm": 0.6663285493850708, "learning_rate": 0.0001, "elbo": 3.76849, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3872273047149895}
{"step": 23660, "loss": 3.778624725341797, "grad_norm": 0.7084425687789917, "learning_rate": 0.0001, "elbo": 3.77862, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.38752052545156}
{"step": 23665, "loss": 3.775572967529297, "grad_norm": 0.7622859477996826, "learning_rate": 0.0001, "elbo": 3.77557, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3878137461881304}
{"step": 23670, "loss": 3.789872741699219, "grad_norm": 0.6780816912651062, "learning_rate": 0.0001, "elbo": 3.78987, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3881069669247008}
{"step": 23675, "loss": 3.76849365234375, "grad_norm": 0.7009735107421875, "learning_rate": 0.0001, "elbo": 3.76849, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3884001876612713}
{"step": 23680, "loss": 3.7757102966308596, "grad_norm": 0.6655061841011047, "learning_rate": 0.0001, "elbo": 3.77571, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.388693408397842}
{"step": 23685, "loss": 3.763567352294922, "grad_norm": 0.7033923268318176, "learning_rate": 0.0001, "elbo": 3.76357, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3889866291344124}
{"step": 23690, "loss": 3.7585613250732424, "grad_norm": 0.7141706347465515, "learning_rate": 0.0001, "elbo": 3.75856, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3892798498709829}
{"step": 23695, "loss": 3.759539794921875, "grad_norm": 0.6786854863166809, "learning_rate": 0.0001, "elbo": 3.75954, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3895730706075533}
{"step": 23700, "loss": 3.765336608886719, "grad_norm": 0.6649319529533386, "learning_rate": 0.0001, "elbo": 3.76534, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.389866291344124}
{"step": 23705, "loss": 3.7740379333496095, "grad_norm": 0.6399087905883789, "learning_rate": 0.0001, "elbo": 3.77404, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3901595120806944}
{"step": 23705, "loss": 3.7740379333496095, "grad_norm": 0.6399087905883789, "learning_rate": 0.0001, "elbo": 3.77404, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3901595120806944}
{"step": 23710, "loss": 3.768537902832031, "grad_norm": 0.6617467403411865, "learning_rate": 0.0001, "elbo": 3.76854, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.390452732817265}
{"step": 23715, "loss": 3.765196990966797, "grad_norm": 0.6911541223526001, "learning_rate": 0.0001, "elbo": 3.7652, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3907459535538353}
{"step": 23720, "loss": 3.7789161682128904, "grad_norm": 1.1730226278305054, "learning_rate": 0.0001, "elbo": 3.77892, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3910391742904058}
{"step": 23720, "loss": 3.7789161682128904, "grad_norm": 1.1730226278305054, "learning_rate": 0.0001, "elbo": 3.77892, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3910391742904058}
{"step": 23720, "loss": 3.7789161682128904, "grad_norm": 1.1730226278305054, "learning_rate": 0.0001, "elbo": 3.77892, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3910391742904058}
{"step": 23725, "loss": 3.788216400146484, "grad_norm": 0.7097222805023193, "learning_rate": 0.0001, "elbo": 3.78822, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3913323950269763}
{"step": 23730, "loss": 3.7670654296875, "grad_norm": 0.7564741969108582, "learning_rate": 0.0001, "elbo": 3.76707, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3916256157635467}
{"step": 23735, "loss": 3.778363800048828, "grad_norm": 0.7013550996780396, "learning_rate": 0.0001, "elbo": 3.77836, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3919188365001172}
{"step": 23735, "loss": 3.778363800048828, "grad_norm": 0.7013550996780396, "learning_rate": 0.0001, "elbo": 3.77836, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3919188365001172}
{"step": 23740, "loss": 3.7819801330566407, "grad_norm": 0.6918035745620728, "learning_rate": 0.0001, "elbo": 3.78198, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3922120572366878}
{"step": 23745, "loss": 3.7627696990966797, "grad_norm": 0.7033894658088684, "learning_rate": 0.0001, "elbo": 3.76277, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3925052779732583}
{"step": 23745, "loss": 3.7627696990966797, "grad_norm": 0.7033894658088684, "learning_rate": 0.0001, "elbo": 3.76277, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3925052779732583}
{"step": 23750, "loss": 3.766178512573242, "grad_norm": 0.7010065317153931, "learning_rate": 0.0001, "elbo": 3.76618, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3927984987098287}
{"step": 23755, "loss": 3.7495567321777346, "grad_norm": 0.6432821154594421, "learning_rate": 0.0001, "elbo": 3.74956, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3930917194463992}
{"step": 23755, "loss": 3.7495567321777346, "grad_norm": 0.6432821154594421, "learning_rate": 0.0001, "elbo": 3.74956, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3930917194463992}
{"step": 23760, "loss": 3.766143798828125, "grad_norm": 0.6324228048324585, "learning_rate": 0.0001, "elbo": 3.76614, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3933849401829699}
{"step": 23765, "loss": 3.7628250122070312, "grad_norm": 0.766012966632843, "learning_rate": 0.0001, "elbo": 3.76283, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3936781609195403}
{"step": 23770, "loss": 3.784607696533203, "grad_norm": 0.6678920388221741, "learning_rate": 0.0001, "elbo": 3.78461, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3939713816561108}
{"step": 23775, "loss": 3.7711997985839845, "grad_norm": 1.0694414377212524, "learning_rate": 0.0001, "elbo": 3.7712, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3942646023926812}
{"step": 23780, "loss": 3.7751060485839845, "grad_norm": 0.6869244575500488, "learning_rate": 0.0001, "elbo": 3.77511, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3945578231292517}
{"step": 23785, "loss": 3.755357360839844, "grad_norm": 0.6597208380699158, "learning_rate": 0.0001, "elbo": 3.75536, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3948510438658221}
{"step": 23790, "loss": 3.7690826416015626, "grad_norm": 0.6814973950386047, "learning_rate": 0.0001, "elbo": 3.76908, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3951442646023926}
{"step": 23795, "loss": 3.7775978088378905, "grad_norm": 0.6453066468238831, "learning_rate": 0.0001, "elbo": 3.7776, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.395437485338963}
{"step": 23800, "loss": 3.76240234375, "grad_norm": 0.6956188082695007, "learning_rate": 0.0001, "elbo": 3.7624, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3957307060755337}
{"step": 23805, "loss": 3.7613994598388674, "grad_norm": 0.87325119972229, "learning_rate": 0.0001, "elbo": 3.7614, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3960239268121042}
{"step": 23810, "loss": 3.7590030670166015, "grad_norm": 0.7765126824378967, "learning_rate": 0.0001, "elbo": 3.759, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3963171475486746}
{"step": 23815, "loss": 3.760334014892578, "grad_norm": 0.7463105916976929, "learning_rate": 0.0001, "elbo": 3.76033, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.396610368285245}
{"step": 23820, "loss": 3.7626922607421873, "grad_norm": 0.6938560009002686, "learning_rate": 0.0001, "elbo": 3.76269, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3969035890218156}
{"step": 23825, "loss": 3.761056137084961, "grad_norm": 0.7698524594306946, "learning_rate": 0.0001, "elbo": 3.76106, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3971968097583862}
{"step": 23830, "loss": 3.7594230651855467, "grad_norm": 0.797607958316803, "learning_rate": 0.0001, "elbo": 3.75942, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3974900304949567}
{"step": 23835, "loss": 3.771295166015625, "grad_norm": 0.669685959815979, "learning_rate": 0.0001, "elbo": 3.7713, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3977832512315271}
{"step": 23840, "loss": 3.7644798278808596, "grad_norm": 0.6688350439071655, "learning_rate": 0.0001, "elbo": 3.76448, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3980764719680976}
{"step": 23845, "loss": 3.770193099975586, "grad_norm": 0.6469337344169617, "learning_rate": 0.0001, "elbo": 3.77019, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.398369692704668}
{"step": 23850, "loss": 3.7620677947998047, "grad_norm": 0.7222287058830261, "learning_rate": 0.0001, "elbo": 3.76207, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3986629134412385}
{"step": 23855, "loss": 3.771403503417969, "grad_norm": 0.6798232197761536, "learning_rate": 0.0001, "elbo": 3.7714, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.398956134177809}
{"step": 23855, "loss": 3.771403503417969, "grad_norm": 0.6798232197761536, "learning_rate": 0.0001, "elbo": 3.7714, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.398956134177809}
{"step": 23860, "loss": 3.7756988525390627, "grad_norm": 0.7417498230934143, "learning_rate": 0.0001, "elbo": 3.7757, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3992493549143794}
{"step": 23865, "loss": 3.7818859100341795, "grad_norm": 0.6640921831130981, "learning_rate": 0.0001, "elbo": 3.78189, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.39954257565095}
{"step": 23865, "loss": 3.7818859100341795, "grad_norm": 0.6640921831130981, "learning_rate": 0.0001, "elbo": 3.78189, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.39954257565095}
{"step": 23870, "loss": 3.763368606567383, "grad_norm": 0.7178309559822083, "learning_rate": 0.0001, "elbo": 3.76337, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.3998357963875205}
{"step": 23875, "loss": 3.7577491760253907, "grad_norm": 0.6549047231674194, "learning_rate": 0.0001, "elbo": 3.75775, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.400129017124091}
{"step": 23880, "loss": 3.7638614654541014, "grad_norm": 0.8562744855880737, "learning_rate": 0.0001, "elbo": 3.76386, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4004222378606614}
{"step": 23885, "loss": 3.7507129669189454, "grad_norm": 0.690756618976593, "learning_rate": 0.0001, "elbo": 3.75071, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4007154585972321}
{"step": 23890, "loss": 3.7759071350097657, "grad_norm": 0.678473174571991, "learning_rate": 0.0001, "elbo": 3.77591, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4010086793338026}
{"step": 23895, "loss": 3.7636135101318358, "grad_norm": 0.6832998991012573, "learning_rate": 0.0001, "elbo": 3.76361, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.401301900070373}
{"step": 23900, "loss": 3.773471450805664, "grad_norm": 0.7634071111679077, "learning_rate": 0.0001, "elbo": 3.77347, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4015951208069435}
{"step": 23905, "loss": 3.763644790649414, "grad_norm": 0.6684463024139404, "learning_rate": 0.0001, "elbo": 3.76364, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.401888341543514}
{"step": 23910, "loss": 3.7663238525390623, "grad_norm": 0.6543275117874146, "learning_rate": 0.0001, "elbo": 3.76632, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4021815622800844}
{"step": 23915, "loss": 3.7624698638916017, "grad_norm": 1.5047210454940796, "learning_rate": 0.0001, "elbo": 3.76247, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4024747830166548}
{"step": 23920, "loss": 3.7707260131835936, "grad_norm": 0.7437689304351807, "learning_rate": 0.0001, "elbo": 3.77073, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4027680037532253}
{"step": 23925, "loss": 3.7634254455566407, "grad_norm": 0.6833481192588806, "learning_rate": 0.0001, "elbo": 3.76343, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.403061224489796}
{"step": 23930, "loss": 3.785589599609375, "grad_norm": 0.678112804889679, "learning_rate": 0.0001, "elbo": 3.78559, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4033544452263664}
{"step": 23935, "loss": 3.76080436706543, "grad_norm": 0.6582898497581482, "learning_rate": 0.0001, "elbo": 3.7608, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4036476659629369}
{"step": 23940, "loss": 3.782915496826172, "grad_norm": 0.6463654637336731, "learning_rate": 0.0001, "elbo": 3.78292, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4039408866995073}
{"step": 23945, "loss": 3.7620113372802733, "grad_norm": 0.6952259540557861, "learning_rate": 0.0001, "elbo": 3.76201, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.404234107436078}
{"step": 23950, "loss": 3.762287902832031, "grad_norm": 0.7275832295417786, "learning_rate": 0.0001, "elbo": 3.76229, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4045273281726485}
{"step": 23955, "loss": 3.77054443359375, "grad_norm": 0.6784959435462952, "learning_rate": 0.0001, "elbo": 3.77054, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.404820548909219}
{"step": 23960, "loss": 3.7477760314941406, "grad_norm": 0.6701957583427429, "learning_rate": 0.0001, "elbo": 3.74778, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4051137696457894}
{"step": 23965, "loss": 3.755829620361328, "grad_norm": 0.6922703981399536, "learning_rate": 0.0001, "elbo": 3.75583, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4054069903823598}
{"step": 23970, "loss": 3.772142791748047, "grad_norm": 0.7707605361938477, "learning_rate": 0.0001, "elbo": 3.77214, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4057002111189303}
{"step": 23975, "loss": 3.756020355224609, "grad_norm": 0.7230364084243774, "learning_rate": 0.0001, "elbo": 3.75602, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4059934318555007}
{"step": 23980, "loss": 3.7695938110351563, "grad_norm": 0.6946829557418823, "learning_rate": 0.0001, "elbo": 3.76959, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4062866525920712}
{"step": 23985, "loss": 3.767980194091797, "grad_norm": 0.7235070466995239, "learning_rate": 0.0001, "elbo": 3.76798, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4065798733286419}
{"step": 23990, "loss": 3.7754615783691405, "grad_norm": 0.6936708688735962, "learning_rate": 0.0001, "elbo": 3.77546, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4068730940652123}
{"step": 23995, "loss": 3.7436939239501954, "grad_norm": 0.664944052696228, "learning_rate": 0.0001, "elbo": 3.74369, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4071663148017828}
{"step": 24000, "loss": 3.7548728942871095, "grad_norm": 0.9387383460998535, "learning_rate": 0.0001, "elbo": 3.75487, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4074595355383532}
{"step": 24005, "loss": 3.7488311767578124, "grad_norm": 0.7738155126571655, "learning_rate": 0.0001, "elbo": 3.74883, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4077527562749237}
{"step": 24005, "loss": 3.7488311767578124, "grad_norm": 0.7738155126571655, "learning_rate": 0.0001, "elbo": 3.74883, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4077527562749237}
{"step": 24010, "loss": 3.750689697265625, "grad_norm": 0.7136529684066772, "learning_rate": 0.0001, "elbo": 3.75069, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4080459770114944}
{"step": 24015, "loss": 3.7601383209228514, "grad_norm": 0.6930390000343323, "learning_rate": 0.0001, "elbo": 3.76014, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4083391977480648}
{"step": 24015, "loss": 3.7601383209228514, "grad_norm": 0.6930390000343323, "learning_rate": 0.0001, "elbo": 3.76014, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4083391977480648}
{"step": 24020, "loss": 3.752410125732422, "grad_norm": 0.8121591806411743, "learning_rate": 0.0001, "elbo": 3.75241, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4086324184846353}
{"step": 24025, "loss": 3.7850826263427733, "grad_norm": 0.7438451051712036, "learning_rate": 0.0001, "elbo": 3.78508, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4089256392212057}
{"step": 24030, "loss": 3.770539474487305, "grad_norm": 0.671072781085968, "learning_rate": 0.0001, "elbo": 3.77054, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4092188599577762}
{"step": 24035, "loss": 3.7564060211181642, "grad_norm": 0.6734411716461182, "learning_rate": 0.0001, "elbo": 3.75641, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4095120806943466}
{"step": 24040, "loss": 3.7617332458496096, "grad_norm": 0.6853545904159546, "learning_rate": 0.0001, "elbo": 3.76173, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.409805301430917}
{"step": 24045, "loss": 3.7655879974365236, "grad_norm": 0.728520929813385, "learning_rate": 0.0001, "elbo": 3.76559, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4100985221674878}
{"step": 24050, "loss": 3.7625621795654296, "grad_norm": 0.6551342010498047, "learning_rate": 0.0001, "elbo": 3.76256, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4103917429040582}
{"step": 24055, "loss": 3.7769432067871094, "grad_norm": 0.7161423563957214, "learning_rate": 0.0001, "elbo": 3.77694, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4106849636406287}
{"step": 24060, "loss": 3.780244064331055, "grad_norm": 0.6611420512199402, "learning_rate": 0.0001, "elbo": 3.78024, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4109781843771991}
{"step": 24065, "loss": 3.780457305908203, "grad_norm": 0.7570111751556396, "learning_rate": 0.0001, "elbo": 3.78046, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4112714051137696}
{"step": 24070, "loss": 3.7558265686035157, "grad_norm": 0.7490153312683105, "learning_rate": 0.0001, "elbo": 3.75583, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4115646258503403}
{"step": 24075, "loss": 3.757643127441406, "grad_norm": 0.6582263708114624, "learning_rate": 0.0001, "elbo": 3.75764, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4118578465869107}
{"step": 24080, "loss": 3.769501495361328, "grad_norm": 0.7058218121528625, "learning_rate": 0.0001, "elbo": 3.7695, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4121510673234812}
{"step": 24085, "loss": 3.792423629760742, "grad_norm": 0.7083640098571777, "learning_rate": 0.0001, "elbo": 3.79242, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4124442880600516}
{"step": 24090, "loss": 3.759246063232422, "grad_norm": 0.8010603785514832, "learning_rate": 0.0001, "elbo": 3.75925, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.412737508796622}
{"step": 24095, "loss": 3.7578948974609374, "grad_norm": 0.744845986366272, "learning_rate": 0.0001, "elbo": 3.75789, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4130307295331925}
{"step": 24100, "loss": 3.743224334716797, "grad_norm": 0.657608687877655, "learning_rate": 0.0001, "elbo": 3.74322, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.413323950269763}
{"step": 24105, "loss": 3.759906768798828, "grad_norm": 0.6522454023361206, "learning_rate": 0.0001, "elbo": 3.75991, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4136171710063334}
{"step": 24110, "loss": 3.756562423706055, "grad_norm": 0.7157472968101501, "learning_rate": 0.0001, "elbo": 3.75656, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4139103917429041}
{"step": 24115, "loss": 3.759490966796875, "grad_norm": 0.6901038885116577, "learning_rate": 0.0001, "elbo": 3.75949, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4142036124794746}
{"step": 24120, "loss": 3.761954116821289, "grad_norm": 0.7077538371086121, "learning_rate": 0.0001, "elbo": 3.76195, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.414496833216045}
{"step": 24125, "loss": 3.7766033172607423, "grad_norm": 0.646013617515564, "learning_rate": 0.0001, "elbo": 3.7766, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4147900539526155}
{"step": 24130, "loss": 3.767101287841797, "grad_norm": 0.6616225838661194, "learning_rate": 0.0001, "elbo": 3.7671, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4150832746891862}
{"step": 24135, "loss": 3.771402359008789, "grad_norm": 0.7072157263755798, "learning_rate": 0.0001, "elbo": 3.7714, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4153764954257566}
{"step": 24140, "loss": 3.7754241943359377, "grad_norm": 0.8122022747993469, "learning_rate": 0.0001, "elbo": 3.77542, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.415669716162327}
{"step": 24145, "loss": 3.767289733886719, "grad_norm": 0.7009832859039307, "learning_rate": 0.0001, "elbo": 3.76729, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4159629368988975}
{"step": 24150, "loss": 3.7564754486083984, "grad_norm": 0.698983371257782, "learning_rate": 0.0001, "elbo": 3.75648, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.416256157635468}
{"step": 24155, "loss": 3.7721858978271485, "grad_norm": 0.7350764274597168, "learning_rate": 0.0001, "elbo": 3.77219, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4165493783720384}
{"step": 24160, "loss": 3.7603294372558596, "grad_norm": 0.7037254571914673, "learning_rate": 0.0001, "elbo": 3.76033, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4168425991086089}
{"step": 24165, "loss": 3.7669891357421874, "grad_norm": 0.7367058992385864, "learning_rate": 0.0001, "elbo": 3.76699, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4171358198451793}
{"step": 24170, "loss": 3.7670989990234376, "grad_norm": 0.6684870719909668, "learning_rate": 0.0001, "elbo": 3.7671, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.41742904058175}
{"step": 24175, "loss": 3.773980712890625, "grad_norm": 0.8747321963310242, "learning_rate": 0.0001, "elbo": 3.77398, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4177222613183205}
{"step": 24180, "loss": 3.756284713745117, "grad_norm": 0.6495774388313293, "learning_rate": 0.0001, "elbo": 3.75628, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.418015482054891}
{"step": 24185, "loss": 3.7596115112304687, "grad_norm": 0.6913845539093018, "learning_rate": 0.0001, "elbo": 3.75961, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4183087027914614}
{"step": 24190, "loss": 3.7486461639404296, "grad_norm": 0.6828686594963074, "learning_rate": 0.0001, "elbo": 3.74865, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4186019235280318}
{"step": 24190, "loss": 3.7486461639404296, "grad_norm": 0.6828686594963074, "learning_rate": 0.0001, "elbo": 3.74865, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4186019235280318}
{"step": 24195, "loss": 3.761180114746094, "grad_norm": 0.687215268611908, "learning_rate": 0.0001, "elbo": 3.76118, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4188951442646025}
{"step": 24200, "loss": 3.750334930419922, "grad_norm": 0.7186853289604187, "learning_rate": 0.0001, "elbo": 3.75034, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.419188365001173}
{"step": 24205, "loss": 3.7752819061279297, "grad_norm": 0.6955227851867676, "learning_rate": 0.0001, "elbo": 3.77528, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4194815857377434}
{"step": 24210, "loss": 3.771187591552734, "grad_norm": 0.6739572882652283, "learning_rate": 0.0001, "elbo": 3.77119, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4197748064743139}
{"step": 24215, "loss": 3.7631778717041016, "grad_norm": 0.732633113861084, "learning_rate": 0.0001, "elbo": 3.76318, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4200680272108843}
{"step": 24220, "loss": 3.7523902893066405, "grad_norm": 0.6856163144111633, "learning_rate": 0.0001, "elbo": 3.75239, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4203612479474548}
{"step": 24225, "loss": 3.752766418457031, "grad_norm": 0.6578918099403381, "learning_rate": 0.0001, "elbo": 3.75277, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4206544686840252}
{"step": 24230, "loss": 3.7486602783203127, "grad_norm": 0.7399937510490417, "learning_rate": 0.0001, "elbo": 3.74866, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.420947689420596}
{"step": 24235, "loss": 3.7585189819335936, "grad_norm": 0.8328801989555359, "learning_rate": 0.0001, "elbo": 3.75852, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4212409101571664}
{"step": 24240, "loss": 3.773542022705078, "grad_norm": 0.7182836532592773, "learning_rate": 0.0001, "elbo": 3.77354, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4215341308937368}
{"step": 24245, "loss": 3.7672504425048827, "grad_norm": 0.6425759196281433, "learning_rate": 0.0001, "elbo": 3.76725, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4218273516303073}
{"step": 24245, "loss": 3.7672504425048827, "grad_norm": 0.6425759196281433, "learning_rate": 0.0001, "elbo": 3.76725, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4218273516303073}
{"step": 24250, "loss": 3.757781982421875, "grad_norm": 0.733088493347168, "learning_rate": 0.0001, "elbo": 3.75778, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4221205723668777}
{"step": 24255, "loss": 3.7597305297851564, "grad_norm": 0.9697349071502686, "learning_rate": 0.0001, "elbo": 3.75973, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4224137931034484}
{"step": 24260, "loss": 3.7595306396484376, "grad_norm": 0.8512077927589417, "learning_rate": 0.0001, "elbo": 3.75953, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4227070138400189}
{"step": 24265, "loss": 3.787689971923828, "grad_norm": 0.6470168232917786, "learning_rate": 0.0001, "elbo": 3.78769, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4230002345765893}
{"step": 24270, "loss": 3.763072967529297, "grad_norm": 0.6707324385643005, "learning_rate": 0.0001, "elbo": 3.76307, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4232934553131598}
{"step": 24275, "loss": 3.758380889892578, "grad_norm": 0.6326653361320496, "learning_rate": 0.0001, "elbo": 3.75838, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4235866760497302}
{"step": 24280, "loss": 3.7644439697265626, "grad_norm": 0.6675505042076111, "learning_rate": 0.0001, "elbo": 3.76444, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4238798967863007}
{"step": 24285, "loss": 3.753409576416016, "grad_norm": 0.7303983569145203, "learning_rate": 0.0001, "elbo": 3.75341, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4241731175228711}
{"step": 24290, "loss": 3.7619579315185545, "grad_norm": 0.725982129573822, "learning_rate": 0.0001, "elbo": 3.76196, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4244663382594416}
{"step": 24295, "loss": 3.7545440673828123, "grad_norm": 0.6550226807594299, "learning_rate": 0.0001, "elbo": 3.75454, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4247595589960123}
{"step": 24300, "loss": 3.7665359497070314, "grad_norm": 0.6464139819145203, "learning_rate": 0.0001, "elbo": 3.76654, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4250527797325827}
{"step": 24305, "loss": 3.7674888610839843, "grad_norm": 0.6505206823348999, "learning_rate": 0.0001, "elbo": 3.76749, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4253460004691532}
{"step": 24310, "loss": 3.7568290710449217, "grad_norm": 0.7212045788764954, "learning_rate": 0.0001, "elbo": 3.75683, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4256392212057236}
{"step": 24315, "loss": 3.758452606201172, "grad_norm": 0.7434183955192566, "learning_rate": 0.0001, "elbo": 3.75845, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4259324419422943}
{"step": 24320, "loss": 3.7754859924316406, "grad_norm": 0.6742915511131287, "learning_rate": 0.0001, "elbo": 3.77549, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4262256626788647}
{"step": 24320, "loss": 3.7754859924316406, "grad_norm": 0.6742915511131287, "learning_rate": 0.0001, "elbo": 3.77549, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4262256626788647}
{"step": 24325, "loss": 3.769265365600586, "grad_norm": 0.6691710352897644, "learning_rate": 0.0001, "elbo": 3.76927, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4265188834154352}
{"step": 24330, "loss": 3.764522933959961, "grad_norm": 0.6682817935943604, "learning_rate": 0.0001, "elbo": 3.76452, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4268121041520057}
{"step": 24335, "loss": 3.746417236328125, "grad_norm": 0.8711259961128235, "learning_rate": 0.0001, "elbo": 3.74642, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.427105324888576}
{"step": 24340, "loss": 3.758575439453125, "grad_norm": 0.6501928567886353, "learning_rate": 0.0001, "elbo": 3.75858, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4273985456251466}
{"step": 24345, "loss": 3.73651123046875, "grad_norm": 0.8051568865776062, "learning_rate": 0.0001, "elbo": 3.73651, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.427691766361717}
{"step": 24350, "loss": 3.769771194458008, "grad_norm": 0.6976504325866699, "learning_rate": 0.0001, "elbo": 3.76977, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4279849870982875}
{"step": 24355, "loss": 3.7778282165527344, "grad_norm": 0.8207399249076843, "learning_rate": 0.0001, "elbo": 3.77783, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4282782078348581}
{"step": 24360, "loss": 3.7584197998046873, "grad_norm": 0.7069298624992371, "learning_rate": 0.0001, "elbo": 3.75842, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4285714285714286}
{"step": 24365, "loss": 3.7566410064697267, "grad_norm": 0.9738066792488098, "learning_rate": 0.0001, "elbo": 3.75664, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.428864649307999}
{"step": 24370, "loss": 3.761549377441406, "grad_norm": 0.6983405351638794, "learning_rate": 0.0001, "elbo": 3.76155, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4291578700445695}
{"step": 24375, "loss": 3.75423583984375, "grad_norm": 0.6259301900863647, "learning_rate": 0.0001, "elbo": 3.75424, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.42945109078114}
{"step": 24380, "loss": 3.7543590545654295, "grad_norm": 0.6389057040214539, "learning_rate": 0.0001, "elbo": 3.75436, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4297443115177106}
{"step": 24385, "loss": 3.763865661621094, "grad_norm": 0.9995409846305847, "learning_rate": 0.0001, "elbo": 3.76387, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.430037532254281}
{"step": 24390, "loss": 3.757607269287109, "grad_norm": 0.6831715106964111, "learning_rate": 0.0001, "elbo": 3.75761, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4303307529908516}
{"step": 24395, "loss": 3.7413158416748047, "grad_norm": 0.8096024394035339, "learning_rate": 0.0001, "elbo": 3.74132, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.430623973727422}
{"step": 24395, "loss": 3.7413158416748047, "grad_norm": 0.8096024394035339, "learning_rate": 0.0001, "elbo": 3.74132, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.430623973727422}
{"step": 24400, "loss": 3.748193359375, "grad_norm": 0.7364398241043091, "learning_rate": 0.0001, "elbo": 3.74819, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4309171944639925}
{"step": 24405, "loss": 3.748228073120117, "grad_norm": 0.6953120827674866, "learning_rate": 0.0001, "elbo": 3.74823, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.431210415200563}
{"step": 24410, "loss": 3.7620559692382813, "grad_norm": 0.6550186276435852, "learning_rate": 0.0001, "elbo": 3.76206, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4315036359371334}
{"step": 24415, "loss": 3.7725101470947267, "grad_norm": 0.7223426699638367, "learning_rate": 0.0001, "elbo": 3.77251, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.431796856673704}
{"step": 24420, "loss": 3.7472808837890623, "grad_norm": 0.628535270690918, "learning_rate": 0.0001, "elbo": 3.74728, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4320900774102745}
{"step": 24425, "loss": 3.7569427490234375, "grad_norm": 0.6394367814064026, "learning_rate": 0.0001, "elbo": 3.75694, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.432383298146845}
{"step": 24430, "loss": 3.7501247406005858, "grad_norm": 0.6381027698516846, "learning_rate": 0.0001, "elbo": 3.75012, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4326765188834154}
{"step": 24435, "loss": 3.7605770111083983, "grad_norm": 0.7759434580802917, "learning_rate": 0.0001, "elbo": 3.76058, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4329697396199859}
{"step": 24440, "loss": 3.7647426605224608, "grad_norm": 0.7155930995941162, "learning_rate": 0.0001, "elbo": 3.76474, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4332629603565565}
{"step": 24445, "loss": 3.7744190216064455, "grad_norm": 0.6645195484161377, "learning_rate": 0.0001, "elbo": 3.77442, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.433556181093127}
{"step": 24450, "loss": 3.7544212341308594, "grad_norm": 0.6754199266433716, "learning_rate": 0.0001, "elbo": 3.75442, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4338494018296974}
{"step": 24455, "loss": 3.76214599609375, "grad_norm": 0.7234907150268555, "learning_rate": 0.0001, "elbo": 3.76215, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.434142622566268}
{"step": 24460, "loss": 3.7645858764648437, "grad_norm": 0.69666987657547, "learning_rate": 0.0001, "elbo": 3.76459, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4344358433028384}
{"step": 24465, "loss": 3.7609573364257813, "grad_norm": 0.7179363369941711, "learning_rate": 0.0001, "elbo": 3.76096, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4347290640394088}
{"step": 24470, "loss": 3.763225555419922, "grad_norm": 0.8095749020576477, "learning_rate": 0.0001, "elbo": 3.76323, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4350222847759793}
{"step": 24475, "loss": 3.762546157836914, "grad_norm": 0.6894350051879883, "learning_rate": 0.0001, "elbo": 3.76255, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4353155055125497}
{"step": 24480, "loss": 3.763503646850586, "grad_norm": 0.8964736461639404, "learning_rate": 0.0001, "elbo": 3.7635, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4356087262491204}
{"step": 24485, "loss": 3.7503437042236327, "grad_norm": 0.8035765290260315, "learning_rate": 0.0001, "elbo": 3.75034, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4359019469856908}
{"step": 24490, "loss": 3.7584457397460938, "grad_norm": 0.6974519491195679, "learning_rate": 0.0001, "elbo": 3.75845, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4361951677222613}
{"step": 24495, "loss": 3.766490173339844, "grad_norm": 0.6625728011131287, "learning_rate": 0.0001, "elbo": 3.76649, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4364883884588318}
{"step": 24500, "loss": 3.7638790130615236, "grad_norm": 0.6955775618553162, "learning_rate": 0.0001, "elbo": 3.76388, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4367816091954024}
{"step": 24505, "loss": 3.7557044982910157, "grad_norm": 0.6810153722763062, "learning_rate": 0.0001, "elbo": 3.7557, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4370748299319729}
{"step": 24510, "loss": 3.7631458282470702, "grad_norm": 0.6691704988479614, "learning_rate": 0.0001, "elbo": 3.76315, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4373680506685433}
{"step": 24515, "loss": 3.7648574829101564, "grad_norm": 0.6484535336494446, "learning_rate": 0.0001, "elbo": 3.76486, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4376612714051138}
{"step": 24520, "loss": 3.7600433349609377, "grad_norm": 0.6691407561302185, "learning_rate": 0.0001, "elbo": 3.76004, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4379544921416842}
{"step": 24520, "loss": 3.7600433349609377, "grad_norm": 0.6691407561302185, "learning_rate": 0.0001, "elbo": 3.76004, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4379544921416842}
{"step": 24525, "loss": 3.7525871276855467, "grad_norm": 0.6557299494743347, "learning_rate": 0.0001, "elbo": 3.75259, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4382477128782547}
{"step": 24530, "loss": 3.768543243408203, "grad_norm": 0.6960517168045044, "learning_rate": 0.0001, "elbo": 3.76854, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4385409336148252}
{"step": 24535, "loss": 3.7587852478027344, "grad_norm": 0.779039740562439, "learning_rate": 0.0001, "elbo": 3.75879, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4388341543513956}
{"step": 24540, "loss": 3.743580627441406, "grad_norm": 0.706633985042572, "learning_rate": 0.0001, "elbo": 3.74358, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4391273750879663}
{"step": 24545, "loss": 3.7603660583496095, "grad_norm": 0.6951955556869507, "learning_rate": 0.0001, "elbo": 3.76037, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4394205958245367}
{"step": 24550, "loss": 3.760783386230469, "grad_norm": 0.6829828023910522, "learning_rate": 0.0001, "elbo": 3.76078, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4397138165611072}
{"step": 24555, "loss": 3.7368656158447267, "grad_norm": 0.7839524149894714, "learning_rate": 0.0001, "elbo": 3.73687, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4400070372976777}
{"step": 24560, "loss": 3.771733856201172, "grad_norm": 0.7038133144378662, "learning_rate": 0.0001, "elbo": 3.77173, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.440300258034248}
{"step": 24565, "loss": 3.7716575622558595, "grad_norm": 0.8059592843055725, "learning_rate": 0.0001, "elbo": 3.77166, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4405934787708188}
{"step": 24570, "loss": 3.755813217163086, "grad_norm": 0.6646985411643982, "learning_rate": 0.0001, "elbo": 3.75581, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4408866995073892}
{"step": 24575, "loss": 3.7675018310546875, "grad_norm": 0.6696890592575073, "learning_rate": 0.0001, "elbo": 3.7675, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4411799202439597}
{"step": 24580, "loss": 3.766070556640625, "grad_norm": 0.6760337948799133, "learning_rate": 0.0001, "elbo": 3.76607, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4414731409805301}
{"step": 24585, "loss": 3.7657310485839846, "grad_norm": 0.687985897064209, "learning_rate": 0.0001, "elbo": 3.76573, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4417663617171006}
{"step": 24590, "loss": 3.7291824340820314, "grad_norm": 0.7373222708702087, "learning_rate": 0.0001, "elbo": 3.72918, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.442059582453671}
{"step": 24595, "loss": 3.761079025268555, "grad_norm": 0.6361115574836731, "learning_rate": 0.0001, "elbo": 3.76108, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4423528031902415}
{"step": 24595, "loss": 3.761079025268555, "grad_norm": 0.6361115574836731, "learning_rate": 0.0001, "elbo": 3.76108, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4423528031902415}
{"step": 24600, "loss": 3.7692073822021483, "grad_norm": 0.75147545337677, "learning_rate": 0.0001, "elbo": 3.76921, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4426460239268122}
{"step": 24600, "loss": 3.7692073822021483, "grad_norm": 0.75147545337677, "learning_rate": 0.0001, "elbo": 3.76921, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4426460239268122}
{"step": 24600, "loss": 3.7692073822021483, "grad_norm": 0.75147545337677, "learning_rate": 0.0001, "elbo": 3.76921, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4426460239268122}
{"step": 24605, "loss": 3.753697967529297, "grad_norm": 0.740960955619812, "learning_rate": 0.0001, "elbo": 3.7537, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4429392446633826}
{"step": 24605, "loss": 3.753697967529297, "grad_norm": 0.740960955619812, "learning_rate": 0.0001, "elbo": 3.7537, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4429392446633826}
{"step": 24610, "loss": 3.760577392578125, "grad_norm": 0.6758154034614563, "learning_rate": 0.0001, "elbo": 3.76058, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.443232465399953}
{"step": 24615, "loss": 3.755471038818359, "grad_norm": 0.6998826861381531, "learning_rate": 0.0001, "elbo": 3.75547, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4435256861365235}
{"step": 24620, "loss": 3.7601512908935546, "grad_norm": 0.6951311230659485, "learning_rate": 0.0001, "elbo": 3.76015, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.443818906873094}
{"step": 24620, "loss": 3.7601512908935546, "grad_norm": 0.6951311230659485, "learning_rate": 0.0001, "elbo": 3.76015, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.443818906873094}
{"step": 24625, "loss": 3.7448760986328127, "grad_norm": 0.6528748273849487, "learning_rate": 0.0001, "elbo": 3.74488, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4441121276096647}
{"step": 24625, "loss": 3.7448760986328127, "grad_norm": 0.6528748273849487, "learning_rate": 0.0001, "elbo": 3.74488, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4441121276096647}
{"step": 24630, "loss": 3.757666015625, "grad_norm": 0.667327344417572, "learning_rate": 0.0001, "elbo": 3.75767, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4444053483462351}
{"step": 24635, "loss": 3.7551040649414062, "grad_norm": 0.6671695113182068, "learning_rate": 0.0001, "elbo": 3.7551, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4446985690828056}
{"step": 24640, "loss": 3.749436950683594, "grad_norm": 0.6967797875404358, "learning_rate": 0.0001, "elbo": 3.74944, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.444991789819376}
{"step": 24645, "loss": 3.7536994934082033, "grad_norm": 0.7055971026420593, "learning_rate": 0.0001, "elbo": 3.7537, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4452850105559465}
{"step": 24650, "loss": 3.76180419921875, "grad_norm": 0.6763295531272888, "learning_rate": 0.0001, "elbo": 3.7618, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.445578231292517}
{"step": 24655, "loss": 3.7542449951171877, "grad_norm": 0.6515673995018005, "learning_rate": 0.0001, "elbo": 3.75425, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4458714520290874}
{"step": 24660, "loss": 3.7496395111083984, "grad_norm": 0.7827648520469666, "learning_rate": 0.0001, "elbo": 3.74964, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4461646727656579}
{"step": 24665, "loss": 3.7557716369628906, "grad_norm": 0.8847650289535522, "learning_rate": 0.0001, "elbo": 3.75577, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4464578935022285}
{"step": 24670, "loss": 3.757892608642578, "grad_norm": 0.6846646070480347, "learning_rate": 0.0001, "elbo": 3.75789, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.446751114238799}
{"step": 24675, "loss": 3.7659023284912108, "grad_norm": 0.6836503148078918, "learning_rate": 0.0001, "elbo": 3.7659, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4470443349753694}
{"step": 24680, "loss": 3.748044967651367, "grad_norm": 0.6841686964035034, "learning_rate": 0.0001, "elbo": 3.74804, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.44733755571194}
{"step": 24685, "loss": 3.7574520111083984, "grad_norm": 0.6805392503738403, "learning_rate": 0.0001, "elbo": 3.75745, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4476307764485106}
{"step": 24690, "loss": 3.749298858642578, "grad_norm": 0.9205349087715149, "learning_rate": 0.0001, "elbo": 3.7493, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.447923997185081}
{"step": 24695, "loss": 3.7529441833496096, "grad_norm": 0.9816445112228394, "learning_rate": 0.0001, "elbo": 3.75294, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4482172179216515}
{"step": 24700, "loss": 3.771836853027344, "grad_norm": 0.6936816573143005, "learning_rate": 0.0001, "elbo": 3.77184, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.448510438658222}
{"step": 24705, "loss": 3.777701568603516, "grad_norm": 0.6917330026626587, "learning_rate": 0.0001, "elbo": 3.7777, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4488036593947924}
{"step": 24710, "loss": 3.758628082275391, "grad_norm": 0.6793819665908813, "learning_rate": 0.0001, "elbo": 3.75863, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4490968801313628}
{"step": 24715, "loss": 3.762332534790039, "grad_norm": 0.7576231360435486, "learning_rate": 0.0001, "elbo": 3.76233, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4493901008679333}
{"step": 24720, "loss": 3.7540908813476563, "grad_norm": 1.0432623624801636, "learning_rate": 0.0001, "elbo": 3.75409, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4496833216045037}
{"step": 24725, "loss": 3.774958038330078, "grad_norm": 0.6997877955436707, "learning_rate": 0.0001, "elbo": 3.77496, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4499765423410744}
{"step": 24730, "loss": 3.757543182373047, "grad_norm": 1.0047497749328613, "learning_rate": 0.0001, "elbo": 3.75754, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4502697630776449}
{"step": 24735, "loss": 3.7685688018798826, "grad_norm": 0.7497759461402893, "learning_rate": 0.0001, "elbo": 3.76857, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4505629838142153}
{"step": 24740, "loss": 3.746733856201172, "grad_norm": 0.681555449962616, "learning_rate": 0.0001, "elbo": 3.74673, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4508562045507858}
{"step": 24745, "loss": 3.752382278442383, "grad_norm": 0.6991375684738159, "learning_rate": 0.0001, "elbo": 3.75238, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4511494252873562}
{"step": 24750, "loss": 3.7594158172607424, "grad_norm": 0.9788763523101807, "learning_rate": 0.0001, "elbo": 3.75942, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.451442646023927}
{"step": 24755, "loss": 3.7634082794189454, "grad_norm": 0.7187557816505432, "learning_rate": 0.0001, "elbo": 3.76341, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4517358667604974}
{"step": 24760, "loss": 3.771222686767578, "grad_norm": 0.70695960521698, "learning_rate": 0.0001, "elbo": 3.77122, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4520290874970678}
{"step": 24765, "loss": 3.763929748535156, "grad_norm": 0.9899299144744873, "learning_rate": 0.0001, "elbo": 3.76393, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4523223082336383}
{"step": 24770, "loss": 3.7545135498046873, "grad_norm": 0.6620861291885376, "learning_rate": 0.0001, "elbo": 3.75451, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4526155289702087}
{"step": 24775, "loss": 3.754804992675781, "grad_norm": 0.6750221252441406, "learning_rate": 0.0001, "elbo": 3.7548, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4529087497067792}
{"step": 24780, "loss": 3.770394134521484, "grad_norm": 0.677754282951355, "learning_rate": 0.0001, "elbo": 3.77039, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4532019704433496}
{"step": 24785, "loss": 3.775224304199219, "grad_norm": 0.693191409111023, "learning_rate": 0.0001, "elbo": 3.77522, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4534951911799203}
{"step": 24790, "loss": 3.7510635375976564, "grad_norm": 0.6895623207092285, "learning_rate": 0.0001, "elbo": 3.75106, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4537884119164908}
{"step": 24795, "loss": 3.7506858825683596, "grad_norm": 0.6864549517631531, "learning_rate": 0.0001, "elbo": 3.75069, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4540816326530612}
{"step": 24800, "loss": 3.7643806457519533, "grad_norm": 0.9653089046478271, "learning_rate": 0.0001, "elbo": 3.76438, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4543748533896317}
{"step": 24805, "loss": 3.742924118041992, "grad_norm": 0.6702534556388855, "learning_rate": 0.0001, "elbo": 3.74292, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4546680741262021}
{"step": 24810, "loss": 3.7370464324951174, "grad_norm": 0.6824065446853638, "learning_rate": 0.0001, "elbo": 3.73705, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4549612948627728}
{"step": 24815, "loss": 3.75035400390625, "grad_norm": 0.6865177750587463, "learning_rate": 0.0001, "elbo": 3.75035, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4552545155993433}
{"step": 24820, "loss": 3.7570655822753904, "grad_norm": 0.67246013879776, "learning_rate": 0.0001, "elbo": 3.75707, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4555477363359137}
{"step": 24825, "loss": 3.74755973815918, "grad_norm": 0.6449642777442932, "learning_rate": 0.0001, "elbo": 3.74756, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4558409570724842}
{"step": 24830, "loss": 3.7411788940429687, "grad_norm": 0.6607762575149536, "learning_rate": 0.0001, "elbo": 3.74118, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4561341778090546}
{"step": 24835, "loss": 3.737477111816406, "grad_norm": 0.6834136247634888, "learning_rate": 0.0001, "elbo": 3.73748, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.456427398545625}
{"step": 24840, "loss": 3.769367218017578, "grad_norm": 0.6988770961761475, "learning_rate": 0.0001, "elbo": 3.76937, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4567206192821955}
{"step": 24845, "loss": 3.758504867553711, "grad_norm": 0.689492404460907, "learning_rate": 0.0001, "elbo": 3.7585, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.457013840018766}
{"step": 24850, "loss": 3.7601619720458985, "grad_norm": 0.6643403768539429, "learning_rate": 0.0001, "elbo": 3.76016, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4573070607553367}
{"step": 24855, "loss": 3.75147705078125, "grad_norm": 0.7309492826461792, "learning_rate": 0.0001, "elbo": 3.75148, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4576002814919071}
{"step": 24860, "loss": 3.753173065185547, "grad_norm": 0.6804452538490295, "learning_rate": 0.0001, "elbo": 3.75317, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4578935022284776}
{"step": 24865, "loss": 3.7544815063476564, "grad_norm": 0.6564541459083557, "learning_rate": 0.0001, "elbo": 3.75448, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.458186722965048}
{"step": 24870, "loss": 3.75804443359375, "grad_norm": 0.7067644596099854, "learning_rate": 0.0001, "elbo": 3.75804, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4584799437016187}
{"step": 24875, "loss": 3.747922897338867, "grad_norm": 0.7031201124191284, "learning_rate": 0.0001, "elbo": 3.74792, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4587731644381892}
{"step": 24880, "loss": 3.7626747131347655, "grad_norm": 0.77690190076828, "learning_rate": 0.0001, "elbo": 3.76267, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4590663851747596}
{"step": 24885, "loss": 3.7609703063964846, "grad_norm": 0.6757919788360596, "learning_rate": 0.0001, "elbo": 3.76097, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.45935960591133}
{"step": 24890, "loss": 3.757398223876953, "grad_norm": 0.6681348085403442, "learning_rate": 0.0001, "elbo": 3.7574, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4596528266479005}
{"step": 24895, "loss": 3.771037292480469, "grad_norm": 0.6777541041374207, "learning_rate": 0.0001, "elbo": 3.77104, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.459946047384471}
{"step": 24900, "loss": 3.7554630279541015, "grad_norm": 0.9414144158363342, "learning_rate": 0.0001, "elbo": 3.75546, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4602392681210414}
{"step": 24905, "loss": 3.7605709075927733, "grad_norm": 0.6653396487236023, "learning_rate": 0.0001, "elbo": 3.76057, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4605324888576119}
{"step": 24910, "loss": 3.768907165527344, "grad_norm": 0.6781905889511108, "learning_rate": 0.0001, "elbo": 3.76891, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4608257095941826}
{"step": 24915, "loss": 3.7513011932373046, "grad_norm": 0.6862561106681824, "learning_rate": 0.0001, "elbo": 3.7513, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.461118930330753}
{"step": 24920, "loss": 3.7627124786376953, "grad_norm": 0.7438274025917053, "learning_rate": 0.0001, "elbo": 3.76271, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4614121510673235}
{"step": 24925, "loss": 3.7429176330566407, "grad_norm": 0.685939371585846, "learning_rate": 0.0001, "elbo": 3.74292, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.461705371803894}
{"step": 24930, "loss": 3.762583923339844, "grad_norm": 0.691926121711731, "learning_rate": 0.0001, "elbo": 3.76258, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4619985925404644}
{"step": 24935, "loss": 3.7740882873535155, "grad_norm": 0.7310011386871338, "learning_rate": 0.0001, "elbo": 3.77409, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.462291813277035}
{"step": 24940, "loss": 3.760137176513672, "grad_norm": 0.8961648941040039, "learning_rate": 0.0001, "elbo": 3.76014, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4625850340136055}
{"step": 24945, "loss": 3.7527976989746095, "grad_norm": 0.6321403384208679, "learning_rate": 0.0001, "elbo": 3.7528, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.462878254750176}
{"step": 24950, "loss": 3.7534149169921873, "grad_norm": 0.6586858630180359, "learning_rate": 0.0001, "elbo": 3.75342, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4631714754867464}
{"step": 24955, "loss": 3.770436477661133, "grad_norm": 0.996891438961029, "learning_rate": 0.0001, "elbo": 3.77044, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4634646962233169}
{"step": 24960, "loss": 3.773846435546875, "grad_norm": 0.6834222674369812, "learning_rate": 0.0001, "elbo": 3.77385, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4637579169598873}
{"step": 24965, "loss": 3.7522960662841798, "grad_norm": 0.740446150302887, "learning_rate": 0.0001, "elbo": 3.7523, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4640511376964578}
{"step": 24970, "loss": 3.7590965270996093, "grad_norm": 0.6577743291854858, "learning_rate": 0.0001, "elbo": 3.7591, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4643443584330285}
{"step": 24975, "loss": 3.739301300048828, "grad_norm": 0.6626601219177246, "learning_rate": 0.0001, "elbo": 3.7393, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.464637579169599}
{"step": 24975, "loss": 3.739301300048828, "grad_norm": 0.6626601219177246, "learning_rate": 0.0001, "elbo": 3.7393, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.464637579169599}
{"step": 24980, "loss": 3.7393035888671875, "grad_norm": 0.6695449948310852, "learning_rate": 0.0001, "elbo": 3.7393, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4649307999061694}
{"step": 24985, "loss": 3.7684600830078123, "grad_norm": 0.732334554195404, "learning_rate": 0.0001, "elbo": 3.76846, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4652240206427398}
{"step": 24990, "loss": 3.7448745727539063, "grad_norm": 0.6878163814544678, "learning_rate": 0.0001, "elbo": 3.74487, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4655172413793103}
{"step": 24995, "loss": 3.746796417236328, "grad_norm": 0.6518622636795044, "learning_rate": 0.0001, "elbo": 3.7468, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.465810462115881}
{"step": 25000, "loss": 3.756602478027344, "grad_norm": 0.8232829570770264, "learning_rate": 0.0001, "elbo": 3.7566, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4661036828524514}
{"step": 25005, "loss": 3.7599205017089843, "grad_norm": 0.7010256052017212, "learning_rate": 0.0001, "elbo": 3.75992, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4663969035890219}
{"step": 25005, "loss": 3.7599205017089843, "grad_norm": 0.7010256052017212, "learning_rate": 0.0001, "elbo": 3.75992, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4663969035890219}
{"step": 25010, "loss": 3.760614776611328, "grad_norm": 0.7026626467704773, "learning_rate": 0.0001, "elbo": 3.76061, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4666901243255923}
{"step": 25015, "loss": 3.7453216552734374, "grad_norm": 0.7283320426940918, "learning_rate": 0.0001, "elbo": 3.74532, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4669833450621628}
{"step": 25020, "loss": 3.751361846923828, "grad_norm": 0.7469046115875244, "learning_rate": 0.0001, "elbo": 3.75136, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4672765657987332}
{"step": 25020, "loss": 3.751361846923828, "grad_norm": 0.7469046115875244, "learning_rate": 0.0001, "elbo": 3.75136, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4672765657987332}
{"step": 25025, "loss": 3.738506317138672, "grad_norm": 0.7194031476974487, "learning_rate": 0.0001, "elbo": 3.73851, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4675697865353037}
{"step": 25030, "loss": 3.7650146484375, "grad_norm": 0.9560595750808716, "learning_rate": 0.0001, "elbo": 3.76501, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4678630072718741}
{"step": 25035, "loss": 3.7466739654541015, "grad_norm": 0.8811624050140381, "learning_rate": 0.0001, "elbo": 3.74667, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4681562280084448}
{"step": 25040, "loss": 3.744599151611328, "grad_norm": 0.690831184387207, "learning_rate": 0.0001, "elbo": 3.7446, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4684494487450153}
{"step": 25045, "loss": 3.773408889770508, "grad_norm": 0.6660670638084412, "learning_rate": 0.0001, "elbo": 3.77341, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4687426694815857}
{"step": 25050, "loss": 3.75145263671875, "grad_norm": 0.7676746249198914, "learning_rate": 0.0001, "elbo": 3.75145, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4690358902181562}
{"step": 25050, "loss": 3.75145263671875, "grad_norm": 0.7676746249198914, "learning_rate": 0.0001, "elbo": 3.75145, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4690358902181562}
{"step": 25055, "loss": 3.757130432128906, "grad_norm": 0.6446887254714966, "learning_rate": 0.0001, "elbo": 3.75713, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4693291109547268}
{"step": 25060, "loss": 3.763406753540039, "grad_norm": 0.6564815044403076, "learning_rate": 0.0001, "elbo": 3.76341, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4696223316912973}
{"step": 25065, "loss": 3.766231918334961, "grad_norm": 0.6788294315338135, "learning_rate": 0.0001, "elbo": 3.76623, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4699155524278678}
{"step": 25070, "loss": 3.7500289916992187, "grad_norm": 0.8247717022895813, "learning_rate": 0.0001, "elbo": 3.75003, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4702087731644382}
{"step": 25075, "loss": 3.7483917236328126, "grad_norm": 0.709873616695404, "learning_rate": 0.0001, "elbo": 3.74839, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4705019939010087}
{"step": 25080, "loss": 3.7535282135009767, "grad_norm": 0.6530672907829285, "learning_rate": 0.0001, "elbo": 3.75353, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4707952146375791}
{"step": 25085, "loss": 3.754623031616211, "grad_norm": 0.6848165988922119, "learning_rate": 0.0001, "elbo": 3.75462, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4710884353741496}
{"step": 25090, "loss": 3.772100830078125, "grad_norm": 1.0253829956054688, "learning_rate": 0.0001, "elbo": 3.7721, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.47138165611072}
{"step": 25095, "loss": 3.7592506408691406, "grad_norm": 0.6794127225875854, "learning_rate": 0.0001, "elbo": 3.75925, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4716748768472907}
{"step": 25100, "loss": 3.7586883544921874, "grad_norm": 0.6606594920158386, "learning_rate": 0.0001, "elbo": 3.75869, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4719680975838612}
{"step": 25105, "loss": 3.746332550048828, "grad_norm": 0.7048637866973877, "learning_rate": 0.0001, "elbo": 3.74633, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4722613183204316}
{"step": 25110, "loss": 3.7678802490234373, "grad_norm": 0.7477406859397888, "learning_rate": 0.0001, "elbo": 3.76788, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.472554539057002}
{"step": 25115, "loss": 3.762132263183594, "grad_norm": 0.6620482206344604, "learning_rate": 0.0001, "elbo": 3.76213, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4728477597935725}
{"step": 25120, "loss": 3.7625335693359374, "grad_norm": 0.6674295663833618, "learning_rate": 0.0001, "elbo": 3.76253, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4731409805301432}
{"step": 25125, "loss": 3.7489036560058593, "grad_norm": 0.6725733876228333, "learning_rate": 0.0001, "elbo": 3.7489, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4734342012667137}
{"step": 25130, "loss": 3.7766136169433593, "grad_norm": 0.6800426840782166, "learning_rate": 0.0001, "elbo": 3.77661, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.473727422003284}
{"step": 25135, "loss": 3.7518341064453127, "grad_norm": 0.6736018061637878, "learning_rate": 0.0001, "elbo": 3.75183, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4740206427398546}
{"step": 25140, "loss": 3.745365524291992, "grad_norm": 0.6090741157531738, "learning_rate": 0.0001, "elbo": 3.74537, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.474313863476425}
{"step": 25145, "loss": 3.7358360290527344, "grad_norm": 0.6400014758110046, "learning_rate": 0.0001, "elbo": 3.73584, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4746070842129955}
{"step": 25150, "loss": 3.7651573181152345, "grad_norm": 0.7190584540367126, "learning_rate": 0.0001, "elbo": 3.76516, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.474900304949566}
{"step": 25155, "loss": 3.7649551391601563, "grad_norm": 0.6953154802322388, "learning_rate": 0.0001, "elbo": 3.76496, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4751935256861366}
{"step": 25160, "loss": 3.7524925231933595, "grad_norm": 0.7638779878616333, "learning_rate": 0.0001, "elbo": 3.75249, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.475486746422707}
{"step": 25165, "loss": 3.7388565063476564, "grad_norm": 0.621667742729187, "learning_rate": 0.0001, "elbo": 3.73886, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4757799671592775}
{"step": 25170, "loss": 3.7603378295898438, "grad_norm": 0.6870266199111938, "learning_rate": 0.0001, "elbo": 3.76034, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.476073187895848}
{"step": 25175, "loss": 3.7569103240966797, "grad_norm": 0.7028802037239075, "learning_rate": 0.0001, "elbo": 3.75691, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4763664086324184}
{"step": 25180, "loss": 3.7458080291748046, "grad_norm": 0.750564455986023, "learning_rate": 0.0001, "elbo": 3.74581, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.476659629368989}
{"step": 25185, "loss": 3.763050079345703, "grad_norm": 0.8354625105857849, "learning_rate": 0.0001, "elbo": 3.76305, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4769528501055595}
{"step": 25190, "loss": 3.7473518371582033, "grad_norm": 0.7291595935821533, "learning_rate": 0.0001, "elbo": 3.74735, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.47724607084213}
{"step": 25195, "loss": 3.749591827392578, "grad_norm": 0.8310646414756775, "learning_rate": 0.0001, "elbo": 3.74959, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4775392915787005}
{"step": 25200, "loss": 3.7624141693115236, "grad_norm": 0.7410266995429993, "learning_rate": 0.0001, "elbo": 3.76241, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.477832512315271}
{"step": 25205, "loss": 3.7685760498046874, "grad_norm": 0.7340987324714661, "learning_rate": 0.0001, "elbo": 3.76858, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4781257330518414}
{"step": 25210, "loss": 3.74476318359375, "grad_norm": 0.8418309092521667, "learning_rate": 0.0001, "elbo": 3.74476, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4784189537884118}
{"step": 25215, "loss": 3.745258331298828, "grad_norm": 0.6991091370582581, "learning_rate": 0.0001, "elbo": 3.74526, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4787121745249823}
{"step": 25220, "loss": 3.740056610107422, "grad_norm": 0.6734365820884705, "learning_rate": 0.0001, "elbo": 3.74006, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.479005395261553}
{"step": 25225, "loss": 3.757963180541992, "grad_norm": 0.6546961069107056, "learning_rate": 0.0001, "elbo": 3.75796, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4792986159981234}
{"step": 25230, "loss": 3.7459388732910157, "grad_norm": 0.6949992179870605, "learning_rate": 0.0001, "elbo": 3.74594, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4795918367346939}
{"step": 25235, "loss": 3.7507247924804688, "grad_norm": 0.644026517868042, "learning_rate": 0.0001, "elbo": 3.75072, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4798850574712643}
{"step": 25240, "loss": 3.7262126922607424, "grad_norm": 0.7981948256492615, "learning_rate": 0.0001, "elbo": 3.72621, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.480178278207835}
{"step": 25240, "loss": 3.7262126922607424, "grad_norm": 0.7981948256492615, "learning_rate": 0.0001, "elbo": 3.72621, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.480178278207835}
{"step": 25245, "loss": 3.747019958496094, "grad_norm": 0.8384989500045776, "learning_rate": 0.0001, "elbo": 3.74702, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4804714989444054}
{"step": 25250, "loss": 3.7442062377929686, "grad_norm": 0.7361197471618652, "learning_rate": 0.0001, "elbo": 3.74421, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.480764719680976}
{"step": 25255, "loss": 3.749472808837891, "grad_norm": 0.6830379366874695, "learning_rate": 0.0001, "elbo": 3.74947, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4810579404175463}
{"step": 25260, "loss": 3.740902328491211, "grad_norm": 0.6794997453689575, "learning_rate": 0.0001, "elbo": 3.7409, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4813511611541168}
{"step": 25265, "loss": 3.759658432006836, "grad_norm": 0.6936408877372742, "learning_rate": 0.0001, "elbo": 3.75966, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4816443818906873}
{"step": 25270, "loss": 3.7490711212158203, "grad_norm": 0.68132084608078, "learning_rate": 0.0001, "elbo": 3.74907, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4819376026272577}
{"step": 25275, "loss": 3.7357433319091795, "grad_norm": 0.894951581954956, "learning_rate": 0.0001, "elbo": 3.73574, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4822308233638282}
{"step": 25280, "loss": 3.724089813232422, "grad_norm": 0.8201588988304138, "learning_rate": 0.0001, "elbo": 3.72409, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4825240441003988}
{"step": 25285, "loss": 3.756395721435547, "grad_norm": 0.9109219312667847, "learning_rate": 0.0001, "elbo": 3.7564, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4828172648369693}
{"step": 25290, "loss": 3.7576770782470703, "grad_norm": 0.6960496306419373, "learning_rate": 0.0001, "elbo": 3.75768, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4831104855735397}
{"step": 25295, "loss": 3.7346961975097654, "grad_norm": 0.6600835919380188, "learning_rate": 0.0001, "elbo": 3.7347, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4834037063101102}
{"step": 25300, "loss": 3.731791305541992, "grad_norm": 0.718420684337616, "learning_rate": 0.0001, "elbo": 3.73179, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4836969270466807}
{"step": 25300, "loss": 3.731791305541992, "grad_norm": 0.718420684337616, "learning_rate": 0.0001, "elbo": 3.73179, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4836969270466807}
{"step": 25305, "loss": 3.75078125, "grad_norm": 0.7441601753234863, "learning_rate": 0.0001, "elbo": 3.75078, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4839901477832513}
{"step": 25310, "loss": 3.747466278076172, "grad_norm": 0.7510390877723694, "learning_rate": 0.0001, "elbo": 3.74747, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4842833685198218}
{"step": 25315, "loss": 3.7399925231933593, "grad_norm": 0.6947606205940247, "learning_rate": 0.0001, "elbo": 3.73999, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4845765892563922}
{"step": 25320, "loss": 3.7449493408203125, "grad_norm": 0.8547303080558777, "learning_rate": 0.0001, "elbo": 3.74495, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4848698099929627}
{"step": 25325, "loss": 3.739674377441406, "grad_norm": 0.6393847465515137, "learning_rate": 0.0001, "elbo": 3.73967, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4851630307295332}
{"step": 25330, "loss": 3.7575973510742187, "grad_norm": 0.671400785446167, "learning_rate": 0.0001, "elbo": 3.7576, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4854562514661036}
{"step": 25335, "loss": 3.765956497192383, "grad_norm": 0.6843252182006836, "learning_rate": 0.0001, "elbo": 3.76596, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.485749472202674}
{"step": 25340, "loss": 3.7374168395996095, "grad_norm": 0.6779338717460632, "learning_rate": 0.0001, "elbo": 3.73742, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4860426929392447}
{"step": 25345, "loss": 3.758580780029297, "grad_norm": 0.6784990429878235, "learning_rate": 0.0001, "elbo": 3.75858, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4863359136758152}
{"step": 25350, "loss": 3.751111602783203, "grad_norm": 0.9276162981987, "learning_rate": 0.0001, "elbo": 3.75111, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4866291344123856}
{"step": 25355, "loss": 3.75654296875, "grad_norm": 0.6654103398323059, "learning_rate": 0.0001, "elbo": 3.75654, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.486922355148956}
{"step": 25360, "loss": 3.7762725830078123, "grad_norm": 0.7011407017707825, "learning_rate": 0.0001, "elbo": 3.77627, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4872155758855266}
{"step": 25365, "loss": 3.75164794921875, "grad_norm": 0.7222169637680054, "learning_rate": 0.0001, "elbo": 3.75165, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4875087966220972}
{"step": 25370, "loss": 3.7619781494140625, "grad_norm": 0.8839724659919739, "learning_rate": 0.0001, "elbo": 3.76198, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4878020173586677}
{"step": 25375, "loss": 3.753972625732422, "grad_norm": 0.7188451290130615, "learning_rate": 0.0001, "elbo": 3.75397, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4880952380952381}
{"step": 25380, "loss": 3.727385711669922, "grad_norm": 0.8968011736869812, "learning_rate": 0.0001, "elbo": 3.72739, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4883884588318086}
{"step": 25385, "loss": 3.743352508544922, "grad_norm": 0.7008575201034546, "learning_rate": 0.0001, "elbo": 3.74335, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.488681679568379}
{"step": 25390, "loss": 3.7456085205078127, "grad_norm": 0.7420023083686829, "learning_rate": 0.0001, "elbo": 3.74561, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4889749003049495}
{"step": 25395, "loss": 3.7454925537109376, "grad_norm": 0.7332407832145691, "learning_rate": 0.0001, "elbo": 3.74549, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.48926812104152}
{"step": 25400, "loss": 3.7480186462402343, "grad_norm": 0.6542883515357971, "learning_rate": 0.0001, "elbo": 3.74802, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4895613417780904}
{"step": 25405, "loss": 3.7575782775878905, "grad_norm": 0.7086864709854126, "learning_rate": 0.0001, "elbo": 3.75758, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.489854562514661}
{"step": 25410, "loss": 3.737533950805664, "grad_norm": 0.730038046836853, "learning_rate": 0.0001, "elbo": 3.73753, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4901477832512315}
{"step": 25415, "loss": 3.769308853149414, "grad_norm": 0.7319859862327576, "learning_rate": 0.0001, "elbo": 3.76931, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.490441003987802}
{"step": 25420, "loss": 3.7439552307128907, "grad_norm": 0.763042688369751, "learning_rate": 0.0001, "elbo": 3.74396, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4907342247243724}
{"step": 25425, "loss": 3.7422836303710936, "grad_norm": 1.0318901538848877, "learning_rate": 0.0001, "elbo": 3.74228, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4910274454609431}
{"step": 25430, "loss": 3.7479400634765625, "grad_norm": 0.7237851619720459, "learning_rate": 0.0001, "elbo": 3.74794, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4913206661975136}
{"step": 25435, "loss": 3.7406051635742186, "grad_norm": 0.7021276950836182, "learning_rate": 0.0001, "elbo": 3.74061, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.491613886934084}
{"step": 25440, "loss": 3.756475830078125, "grad_norm": 0.7769206166267395, "learning_rate": 0.0001, "elbo": 3.75648, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4919071076706545}
{"step": 25445, "loss": 3.746862030029297, "grad_norm": 0.6993013620376587, "learning_rate": 0.0001, "elbo": 3.74686, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.492200328407225}
{"step": 25450, "loss": 3.7607776641845705, "grad_norm": 0.6630792021751404, "learning_rate": 0.0001, "elbo": 3.76078, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4924935491437954}
{"step": 25455, "loss": 3.7348453521728517, "grad_norm": 0.7194610834121704, "learning_rate": 0.0001, "elbo": 3.73485, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4927867698803658}
{"step": 25460, "loss": 3.7522537231445314, "grad_norm": 0.690520167350769, "learning_rate": 0.0001, "elbo": 3.75225, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4930799906169363}
{"step": 25460, "loss": 3.7522537231445314, "grad_norm": 0.690520167350769, "learning_rate": 0.0001, "elbo": 3.75225, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4930799906169363}
{"step": 25465, "loss": 3.7460968017578127, "grad_norm": 0.6595670580863953, "learning_rate": 0.0001, "elbo": 3.7461, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.493373211353507}
{"step": 25470, "loss": 3.7555999755859375, "grad_norm": 1.0467146635055542, "learning_rate": 0.0001, "elbo": 3.7556, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4936664320900774}
{"step": 25475, "loss": 3.7411590576171876, "grad_norm": 0.6624665260314941, "learning_rate": 0.0001, "elbo": 3.74116, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4939596528266479}
{"step": 25480, "loss": 3.7494449615478516, "grad_norm": 0.7101860046386719, "learning_rate": 0.0001, "elbo": 3.74944, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4942528735632183}
{"step": 25485, "loss": 3.738557052612305, "grad_norm": 0.7030308842658997, "learning_rate": 0.0001, "elbo": 3.73856, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.494546094299789}
{"step": 25490, "loss": 3.7581077575683595, "grad_norm": 0.709832489490509, "learning_rate": 0.0001, "elbo": 3.75811, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4948393150363595}
{"step": 25495, "loss": 3.753609466552734, "grad_norm": 0.6398698687553406, "learning_rate": 0.0001, "elbo": 3.75361, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.49513253577293}
{"step": 25500, "loss": 3.747385025024414, "grad_norm": 0.6970269680023193, "learning_rate": 0.0001, "elbo": 3.74739, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4954257565095004}
{"step": 25505, "loss": 3.746080017089844, "grad_norm": 0.7117724418640137, "learning_rate": 0.0001, "elbo": 3.74608, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4957189772460708}
{"step": 25510, "loss": 3.750527572631836, "grad_norm": 0.7536153197288513, "learning_rate": 0.0001, "elbo": 3.75053, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4960121979826413}
{"step": 25515, "loss": 3.74218635559082, "grad_norm": 0.6545992493629456, "learning_rate": 0.0001, "elbo": 3.74219, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4963054187192117}
{"step": 25520, "loss": 3.7609901428222656, "grad_norm": 0.71285080909729, "learning_rate": 0.0001, "elbo": 3.76099, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4965986394557822}
{"step": 25525, "loss": 3.756979751586914, "grad_norm": 0.7036832571029663, "learning_rate": 0.0001, "elbo": 3.75698, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4968918601923529}
{"step": 25530, "loss": 3.7366622924804687, "grad_norm": 0.7251954078674316, "learning_rate": 0.0001, "elbo": 3.73666, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4971850809289233}
{"step": 25535, "loss": 3.7462738037109373, "grad_norm": 0.7599373459815979, "learning_rate": 0.0001, "elbo": 3.74627, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4974783016654938}
{"step": 25540, "loss": 3.733489227294922, "grad_norm": 0.7354329228401184, "learning_rate": 0.0001, "elbo": 3.73349, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4977715224020642}
{"step": 25545, "loss": 3.7533084869384767, "grad_norm": 0.6840072274208069, "learning_rate": 0.0001, "elbo": 3.75331, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4980647431386347}
{"step": 25550, "loss": 3.739287567138672, "grad_norm": 0.73175448179245, "learning_rate": 0.0001, "elbo": 3.73929, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4983579638752054}
{"step": 25555, "loss": 3.7312038421630858, "grad_norm": 0.6754133701324463, "learning_rate": 0.0001, "elbo": 3.7312, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4986511846117758}
{"step": 25555, "loss": 3.7312038421630858, "grad_norm": 0.6754133701324463, "learning_rate": 0.0001, "elbo": 3.7312, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4986511846117758}
{"step": 25560, "loss": 3.7467994689941406, "grad_norm": 0.7055177092552185, "learning_rate": 0.0001, "elbo": 3.7468, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4989444053483463}
{"step": 25565, "loss": 3.759382629394531, "grad_norm": 0.6893249750137329, "learning_rate": 0.0001, "elbo": 3.75938, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4992376260849167}
{"step": 25570, "loss": 3.7415672302246095, "grad_norm": 0.6749013662338257, "learning_rate": 0.0001, "elbo": 3.74157, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4995308468214872}
{"step": 25575, "loss": 3.7275089263916015, "grad_norm": 0.6893430352210999, "learning_rate": 0.0001, "elbo": 3.72751, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.4998240675580576}
{"step": 25580, "loss": 3.7292686462402345, "grad_norm": 0.650484025478363, "learning_rate": 0.0001, "elbo": 3.72927, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.500117288294628}
{"step": 25585, "loss": 3.755113220214844, "grad_norm": 0.7309051752090454, "learning_rate": 0.0001, "elbo": 3.75511, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5004105090311985}
{"step": 25590, "loss": 3.747377395629883, "grad_norm": 0.6850510239601135, "learning_rate": 0.0001, "elbo": 3.74738, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5007037297677692}
{"step": 25595, "loss": 3.7381072998046876, "grad_norm": 0.6868495941162109, "learning_rate": 0.0001, "elbo": 3.73811, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5009969505043397}
{"step": 25600, "loss": 3.7382888793945312, "grad_norm": 0.6719641089439392, "learning_rate": 0.0001, "elbo": 3.73829, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5012901712409101}
{"step": 25605, "loss": 3.7595794677734373, "grad_norm": 0.6781283617019653, "learning_rate": 0.0001, "elbo": 3.75958, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5015833919774808}
{"step": 25610, "loss": 3.732393264770508, "grad_norm": 0.7240915298461914, "learning_rate": 0.0001, "elbo": 3.73239, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5018766127140513}
{"step": 25615, "loss": 3.755181884765625, "grad_norm": 0.7205783724784851, "learning_rate": 0.0001, "elbo": 3.75518, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5021698334506217}
{"step": 25620, "loss": 3.7655612945556642, "grad_norm": 0.670922577381134, "learning_rate": 0.0001, "elbo": 3.76556, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5024630541871922}
{"step": 25625, "loss": 3.749201202392578, "grad_norm": 0.7347316145896912, "learning_rate": 0.0001, "elbo": 3.7492, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5027562749237626}
{"step": 25630, "loss": 3.7548927307128905, "grad_norm": 0.7627964615821838, "learning_rate": 0.0001, "elbo": 3.75489, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.503049495660333}
{"step": 25635, "loss": 3.7449104309082033, "grad_norm": 0.7432393431663513, "learning_rate": 0.0001, "elbo": 3.74491, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5033427163969035}
{"step": 25640, "loss": 3.734588623046875, "grad_norm": 0.6359463334083557, "learning_rate": 0.0001, "elbo": 3.73459, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.503635937133474}
{"step": 25645, "loss": 3.7428184509277345, "grad_norm": 0.8234833478927612, "learning_rate": 0.0001, "elbo": 3.74282, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5039291578700444}
{"step": 25650, "loss": 3.739118957519531, "grad_norm": 0.9302634000778198, "learning_rate": 0.0001, "elbo": 3.73912, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.504222378606615}
{"step": 25650, "loss": 3.739118957519531, "grad_norm": 0.9302634000778198, "learning_rate": 0.0001, "elbo": 3.73912, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.504222378606615}
{"step": 25655, "loss": 3.7577484130859373, "grad_norm": 0.7936463952064514, "learning_rate": 0.0001, "elbo": 3.75775, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5045155993431856}
{"step": 25660, "loss": 3.743552398681641, "grad_norm": 0.6965091824531555, "learning_rate": 0.0001, "elbo": 3.74355, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.504808820079756}
{"step": 25660, "loss": 3.743552398681641, "grad_norm": 0.6965091824531555, "learning_rate": 0.0001, "elbo": 3.74355, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.504808820079756}
{"step": 25665, "loss": 3.7575942993164064, "grad_norm": 0.843110978603363, "learning_rate": 0.0001, "elbo": 3.75759, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5051020408163265}
{"step": 25670, "loss": 3.743783187866211, "grad_norm": 0.7069820761680603, "learning_rate": 0.0001, "elbo": 3.74378, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5053952615528972}
{"step": 25675, "loss": 3.7352996826171876, "grad_norm": 0.6410043239593506, "learning_rate": 0.0001, "elbo": 3.7353, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5056884822894676}
{"step": 25680, "loss": 3.7477054595947266, "grad_norm": 0.658551812171936, "learning_rate": 0.0001, "elbo": 3.74771, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.505981703026038}
{"step": 25685, "loss": 3.7437957763671874, "grad_norm": 0.640475869178772, "learning_rate": 0.0001, "elbo": 3.7438, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5062749237626085}
{"step": 25685, "loss": 3.7437957763671874, "grad_norm": 0.640475869178772, "learning_rate": 0.0001, "elbo": 3.7438, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5062749237626085}
{"step": 25690, "loss": 3.7455650329589845, "grad_norm": 0.7373499274253845, "learning_rate": 0.0001, "elbo": 3.74556, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.506568144499179}
{"step": 25695, "loss": 3.750069427490234, "grad_norm": 0.6946751475334167, "learning_rate": 0.0001, "elbo": 3.75007, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5068613652357494}
{"step": 25695, "loss": 3.750069427490234, "grad_norm": 0.6946751475334167, "learning_rate": 0.0001, "elbo": 3.75007, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5068613652357494}
{"step": 25700, "loss": 3.757125473022461, "grad_norm": 0.7303241491317749, "learning_rate": 0.0001, "elbo": 3.75713, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5071545859723199}
{"step": 25705, "loss": 3.7430419921875, "grad_norm": 0.6805646419525146, "learning_rate": 0.0001, "elbo": 3.74304, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5074478067088903}
{"step": 25710, "loss": 3.7358211517333983, "grad_norm": 0.7027652859687805, "learning_rate": 0.0001, "elbo": 3.73582, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5077410274454608}
{"step": 25710, "loss": 3.7358211517333983, "grad_norm": 0.7027652859687805, "learning_rate": 0.0001, "elbo": 3.73582, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5077410274454608}
{"step": 25715, "loss": 3.7425968170166017, "grad_norm": 0.7087153196334839, "learning_rate": 0.0001, "elbo": 3.7426, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5080342481820315}
{"step": 25720, "loss": 3.741217041015625, "grad_norm": 0.632064700126648, "learning_rate": 0.0001, "elbo": 3.74122, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.508327468918602}
{"step": 25725, "loss": 3.738733673095703, "grad_norm": 0.7074775695800781, "learning_rate": 0.0001, "elbo": 3.73873, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5086206896551724}
{"step": 25730, "loss": 3.755884552001953, "grad_norm": 0.729170024394989, "learning_rate": 0.0001, "elbo": 3.75588, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.508913910391743}
{"step": 25735, "loss": 3.762664031982422, "grad_norm": 0.7419243454933167, "learning_rate": 0.0001, "elbo": 3.76266, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5092071311283135}
{"step": 25740, "loss": 3.734056854248047, "grad_norm": 0.7721187472343445, "learning_rate": 0.0001, "elbo": 3.73406, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.509500351864884}
{"step": 25745, "loss": 3.736655426025391, "grad_norm": 0.678045392036438, "learning_rate": 0.0001, "elbo": 3.73666, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5097935726014544}
{"step": 25750, "loss": 3.76148681640625, "grad_norm": 0.7545916438102722, "learning_rate": 0.0001, "elbo": 3.76149, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5100867933380249}
{"step": 25755, "loss": 3.731753921508789, "grad_norm": 0.9788036346435547, "learning_rate": 0.0001, "elbo": 3.73175, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5103800140745953}
{"step": 25760, "loss": 3.723191833496094, "grad_norm": 0.7126745581626892, "learning_rate": 0.0001, "elbo": 3.72319, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5106732348111658}
{"step": 25765, "loss": 3.733307647705078, "grad_norm": 0.7109012007713318, "learning_rate": 0.0001, "elbo": 3.73331, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5109664555477362}
{"step": 25770, "loss": 3.7433258056640626, "grad_norm": 0.7319587469100952, "learning_rate": 0.0001, "elbo": 3.74333, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5112596762843067}
{"step": 25775, "loss": 3.743083190917969, "grad_norm": 0.6906573176383972, "learning_rate": 0.0001, "elbo": 3.74308, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5115528970208774}
{"step": 25780, "loss": 3.7396804809570314, "grad_norm": 0.7215000987052917, "learning_rate": 0.0001, "elbo": 3.73968, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5118461177574478}
{"step": 25785, "loss": 3.7448966979980467, "grad_norm": 0.6361858248710632, "learning_rate": 0.0001, "elbo": 3.7449, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5121393384940183}
{"step": 25790, "loss": 3.7533119201660154, "grad_norm": 0.7139182090759277, "learning_rate": 0.0001, "elbo": 3.75331, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.512432559230589}
{"step": 25795, "loss": 3.754735565185547, "grad_norm": 0.6884286999702454, "learning_rate": 0.0001, "elbo": 3.75474, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5127257799671594}
{"step": 25800, "loss": 3.7415332794189453, "grad_norm": 0.6935855746269226, "learning_rate": 0.0001, "elbo": 3.74153, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5130190007037299}
{"step": 25805, "loss": 3.7447883605957033, "grad_norm": 0.7495409846305847, "learning_rate": 0.0001, "elbo": 3.74479, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5133122214403003}
{"step": 25810, "loss": 3.7545444488525392, "grad_norm": 0.6647888422012329, "learning_rate": 0.0001, "elbo": 3.75454, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5136054421768708}
{"step": 25810, "loss": 3.7545444488525392, "grad_norm": 0.6647888422012329, "learning_rate": 0.0001, "elbo": 3.75454, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5136054421768708}
{"step": 25815, "loss": 3.750244903564453, "grad_norm": 0.6651427745819092, "learning_rate": 0.0001, "elbo": 3.75024, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5138986629134412}
{"step": 25820, "loss": 3.743763732910156, "grad_norm": 0.7007606029510498, "learning_rate": 0.0001, "elbo": 3.74376, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5141918836500117}
{"step": 25825, "loss": 3.738128662109375, "grad_norm": 0.6918225288391113, "learning_rate": 0.0001, "elbo": 3.73813, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5144851043865821}
{"step": 25830, "loss": 3.7490177154541016, "grad_norm": 0.749739944934845, "learning_rate": 0.0001, "elbo": 3.74902, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5147783251231526}
{"step": 25835, "loss": 3.756780242919922, "grad_norm": 0.7277609705924988, "learning_rate": 0.0001, "elbo": 3.75678, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.515071545859723}
{"step": 25840, "loss": 3.7265380859375, "grad_norm": 0.7705311179161072, "learning_rate": 0.0001, "elbo": 3.72654, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5153647665962937}
{"step": 25845, "loss": 3.743726348876953, "grad_norm": 0.7259334921836853, "learning_rate": 0.0001, "elbo": 3.74373, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5156579873328642}
{"step": 25850, "loss": 3.751512908935547, "grad_norm": 0.7967578768730164, "learning_rate": 0.0001, "elbo": 3.75151, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5159512080694346}
{"step": 25855, "loss": 3.7222431182861326, "grad_norm": 0.6466807723045349, "learning_rate": 0.0001, "elbo": 3.72224, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5162444288060053}
{"step": 25860, "loss": 3.7380905151367188, "grad_norm": 0.670215904712677, "learning_rate": 0.0001, "elbo": 3.73809, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5165376495425757}
{"step": 25865, "loss": 3.7473823547363283, "grad_norm": 0.6742104887962341, "learning_rate": 0.0001, "elbo": 3.74738, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5168308702791462}
{"step": 25870, "loss": 3.73936767578125, "grad_norm": 0.6388891935348511, "learning_rate": 0.0001, "elbo": 3.73937, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5171240910157167}
{"step": 25875, "loss": 3.742550277709961, "grad_norm": 0.6907373666763306, "learning_rate": 0.0001, "elbo": 3.74255, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5174173117522871}
{"step": 25880, "loss": 3.753071594238281, "grad_norm": 0.6697871685028076, "learning_rate": 0.0001, "elbo": 3.75307, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5177105324888576}
{"step": 25880, "loss": 3.753071594238281, "grad_norm": 0.6697871685028076, "learning_rate": 0.0001, "elbo": 3.75307, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5177105324888576}
{"step": 25885, "loss": 3.736455535888672, "grad_norm": 0.6642221212387085, "learning_rate": 0.0001, "elbo": 3.73646, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.518003753225428}
{"step": 25890, "loss": 3.7552631378173826, "grad_norm": 0.6810404658317566, "learning_rate": 0.0001, "elbo": 3.75526, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5182969739619985}
{"step": 25895, "loss": 3.731927490234375, "grad_norm": 0.6893365979194641, "learning_rate": 0.0001, "elbo": 3.73193, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.518590194698569}
{"step": 25900, "loss": 3.7308467864990233, "grad_norm": 0.8483825922012329, "learning_rate": 0.0001, "elbo": 3.73085, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5188834154351396}
{"step": 25905, "loss": 3.736380386352539, "grad_norm": 0.7196304202079773, "learning_rate": 0.0001, "elbo": 3.73638, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.51917663617171}
{"step": 25910, "loss": 3.7409725189208984, "grad_norm": 0.7929384112358093, "learning_rate": 0.0001, "elbo": 3.74097, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5194698569082805}
{"step": 25915, "loss": 3.7393753051757814, "grad_norm": 0.6777364015579224, "learning_rate": 0.0001, "elbo": 3.73938, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5197630776448512}
{"step": 25920, "loss": 3.741780090332031, "grad_norm": 0.6722480654716492, "learning_rate": 0.0001, "elbo": 3.74178, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5200562983814216}
{"step": 25925, "loss": 3.749737548828125, "grad_norm": 0.6538909673690796, "learning_rate": 0.0001, "elbo": 3.74974, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.520349519117992}
{"step": 25930, "loss": 3.7509296417236326, "grad_norm": 0.8030946254730225, "learning_rate": 0.0001, "elbo": 3.75093, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5206427398545626}
{"step": 25935, "loss": 3.756006622314453, "grad_norm": 0.6837438344955444, "learning_rate": 0.0001, "elbo": 3.75601, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.520935960591133}
{"step": 25940, "loss": 3.752043914794922, "grad_norm": 0.6804722547531128, "learning_rate": 0.0001, "elbo": 3.75204, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5212291813277035}
{"step": 25945, "loss": 3.7454364776611326, "grad_norm": 0.6998934149742126, "learning_rate": 0.0001, "elbo": 3.74544, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.521522402064274}
{"step": 25950, "loss": 3.7138763427734376, "grad_norm": 0.7542348504066467, "learning_rate": 0.0001, "elbo": 3.71388, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5218156228008444}
{"step": 25955, "loss": 3.7352474212646483, "grad_norm": 0.6643219590187073, "learning_rate": 0.0001, "elbo": 3.73525, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5221088435374148}
{"step": 25960, "loss": 3.7567111968994142, "grad_norm": 0.9427398443222046, "learning_rate": 0.0001, "elbo": 3.75671, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5224020642739855}
{"step": 25960, "loss": 3.7567111968994142, "grad_norm": 0.9427398443222046, "learning_rate": 0.0001, "elbo": 3.75671, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5224020642739855}
{"step": 25965, "loss": 3.73839111328125, "grad_norm": 0.653842568397522, "learning_rate": 0.0001, "elbo": 3.73839, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.522695285010556}
{"step": 25970, "loss": 3.7389446258544923, "grad_norm": 0.7690805792808533, "learning_rate": 0.0001, "elbo": 3.73894, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5229885057471264}
{"step": 25975, "loss": 3.7487274169921876, "grad_norm": 0.6524555087089539, "learning_rate": 0.0001, "elbo": 3.74873, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.523281726483697}
{"step": 25980, "loss": 3.7436897277832033, "grad_norm": 0.7926869988441467, "learning_rate": 0.0001, "elbo": 3.74369, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5235749472202675}
{"step": 25985, "loss": 3.7365020751953124, "grad_norm": 0.7913358807563782, "learning_rate": 0.0001, "elbo": 3.7365, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.523868167956838}
{"step": 25990, "loss": 3.740771484375, "grad_norm": 0.6677711009979248, "learning_rate": 0.0001, "elbo": 3.74077, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5241613886934084}
{"step": 25995, "loss": 3.741925811767578, "grad_norm": 0.9883923530578613, "learning_rate": 0.0001, "elbo": 3.74193, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.524454609429979}
{"step": 26000, "loss": 3.7616443634033203, "grad_norm": 0.715424656867981, "learning_rate": 0.0001, "elbo": 3.76164, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5247478301665494}
{"step": 26005, "loss": 3.744149398803711, "grad_norm": 0.6974906325340271, "learning_rate": 0.0001, "elbo": 3.74415, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5250410509031198}
{"step": 26010, "loss": 3.746187210083008, "grad_norm": 0.6827139854431152, "learning_rate": 0.0001, "elbo": 3.74619, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5253342716396903}
{"step": 26010, "loss": 3.746187210083008, "grad_norm": 0.6827139854431152, "learning_rate": 0.0001, "elbo": 3.74619, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5253342716396903}
{"step": 26015, "loss": 3.7469268798828126, "grad_norm": 0.7224517464637756, "learning_rate": 0.0001, "elbo": 3.74693, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5256274923762607}
{"step": 26020, "loss": 3.7469505310058593, "grad_norm": 0.8019222617149353, "learning_rate": 0.0001, "elbo": 3.74695, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5259207131128312}
{"step": 26025, "loss": 3.7311683654785157, "grad_norm": 0.6780900955200195, "learning_rate": 0.0001, "elbo": 3.73117, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5262139338494018}
{"step": 26030, "loss": 3.7377944946289063, "grad_norm": 0.7162080407142639, "learning_rate": 0.0001, "elbo": 3.73779, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5265071545859723}
{"step": 26035, "loss": 3.738714599609375, "grad_norm": 0.7219184041023254, "learning_rate": 0.0001, "elbo": 3.73871, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5268003753225428}
{"step": 26040, "loss": 3.747090148925781, "grad_norm": 0.6575696468353271, "learning_rate": 0.0001, "elbo": 3.74709, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5270935960591134}
{"step": 26045, "loss": 3.7276878356933594, "grad_norm": 0.6877971291542053, "learning_rate": 0.0001, "elbo": 3.72769, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5273868167956839}
{"step": 26050, "loss": 3.7306121826171874, "grad_norm": 0.6788108944892883, "learning_rate": 0.0001, "elbo": 3.73061, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5276800375322543}
{"step": 26055, "loss": 3.7293022155761717, "grad_norm": 0.6716960668563843, "learning_rate": 0.0001, "elbo": 3.7293, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5279732582688248}
{"step": 26060, "loss": 3.7325992584228516, "grad_norm": 0.6804040670394897, "learning_rate": 0.0001, "elbo": 3.7326, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5282664790053953}
{"step": 26065, "loss": 3.744648742675781, "grad_norm": 0.9016129970550537, "learning_rate": 0.0001, "elbo": 3.74465, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5285596997419657}
{"step": 26070, "loss": 3.741754150390625, "grad_norm": 0.6834542751312256, "learning_rate": 0.0001, "elbo": 3.74175, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5288529204785362}
{"step": 26075, "loss": 3.745847702026367, "grad_norm": 0.6495770812034607, "learning_rate": 0.0001, "elbo": 3.74585, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5291461412151066}
{"step": 26075, "loss": 3.745847702026367, "grad_norm": 0.6495770812034607, "learning_rate": 0.0001, "elbo": 3.74585, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5291461412151066}
{"step": 26080, "loss": 3.7449272155761717, "grad_norm": 0.6484925746917725, "learning_rate": 0.0001, "elbo": 3.74493, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.529439361951677}
{"step": 26085, "loss": 3.754264068603516, "grad_norm": 0.6823809742927551, "learning_rate": 0.0001, "elbo": 3.75426, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5297325826882477}
{"step": 26090, "loss": 3.731951904296875, "grad_norm": 0.7808970808982849, "learning_rate": 0.0001, "elbo": 3.73195, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5300258034248182}
{"step": 26095, "loss": 3.751740264892578, "grad_norm": 0.7008469700813293, "learning_rate": 0.0001, "elbo": 3.75174, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5303190241613887}
{"step": 26100, "loss": 3.7445587158203124, "grad_norm": 0.827159583568573, "learning_rate": 0.0001, "elbo": 3.74456, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5306122448979593}
{"step": 26105, "loss": 3.7676788330078126, "grad_norm": 0.6814751625061035, "learning_rate": 0.0001, "elbo": 3.76768, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5309054656345298}
{"step": 26110, "loss": 3.752022552490234, "grad_norm": 0.6444721221923828, "learning_rate": 0.0001, "elbo": 3.75202, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5311986863711002}
{"step": 26115, "loss": 3.7399341583251955, "grad_norm": 0.7292069792747498, "learning_rate": 0.0001, "elbo": 3.73993, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5314919071076707}
{"step": 26120, "loss": 3.733136749267578, "grad_norm": 0.6909555792808533, "learning_rate": 0.0001, "elbo": 3.73314, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5317851278442411}
{"step": 26125, "loss": 3.7442581176757814, "grad_norm": 0.6608158946037292, "learning_rate": 0.0001, "elbo": 3.74426, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5320783485808116}
{"step": 26130, "loss": 3.7319320678710937, "grad_norm": 0.7265251278877258, "learning_rate": 0.0001, "elbo": 3.73193, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.532371569317382}
{"step": 26135, "loss": 3.7477882385253904, "grad_norm": 0.7154196500778198, "learning_rate": 0.0001, "elbo": 3.74779, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5326647900539525}
{"step": 26135, "loss": 3.7477882385253904, "grad_norm": 0.7154196500778198, "learning_rate": 0.0001, "elbo": 3.74779, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5326647900539525}
{"step": 26140, "loss": 3.74456901550293, "grad_norm": 0.7087721228599548, "learning_rate": 0.0001, "elbo": 3.74457, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.532958010790523}
{"step": 26145, "loss": 3.7333724975585936, "grad_norm": 0.7125359773635864, "learning_rate": 0.0001, "elbo": 3.73337, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5332512315270936}
{"step": 26150, "loss": 3.742070770263672, "grad_norm": 0.8078606128692627, "learning_rate": 0.0001, "elbo": 3.74207, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.533544452263664}
{"step": 26155, "loss": 3.746504211425781, "grad_norm": 0.6816914677619934, "learning_rate": 0.0001, "elbo": 3.7465, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5338376730002345}
{"step": 26160, "loss": 3.7531059265136717, "grad_norm": 0.6573189496994019, "learning_rate": 0.0001, "elbo": 3.75311, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5341308937368052}
{"step": 26165, "loss": 3.7283058166503906, "grad_norm": 0.6766502857208252, "learning_rate": 0.0001, "elbo": 3.72831, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5344241144733757}
{"step": 26170, "loss": 3.7314517974853514, "grad_norm": 0.645163893699646, "learning_rate": 0.0001, "elbo": 3.73145, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5347173352099461}
{"step": 26175, "loss": 3.735729217529297, "grad_norm": 0.6912923455238342, "learning_rate": 0.0001, "elbo": 3.73573, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5350105559465166}
{"step": 26180, "loss": 3.7461692810058596, "grad_norm": 0.8146487474441528, "learning_rate": 0.0001, "elbo": 3.74617, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.535303776683087}
{"step": 26185, "loss": 3.7493827819824217, "grad_norm": 0.7419186234474182, "learning_rate": 0.0001, "elbo": 3.74938, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5355969974196575}
{"step": 26190, "loss": 3.7538436889648437, "grad_norm": 0.6813489198684692, "learning_rate": 0.0001, "elbo": 3.75384, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.535890218156228}
{"step": 26195, "loss": 3.7392204284667967, "grad_norm": 0.7146604657173157, "learning_rate": 0.0001, "elbo": 3.73922, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5361834388927984}
{"step": 26200, "loss": 3.739742660522461, "grad_norm": 0.6899449825286865, "learning_rate": 0.0001, "elbo": 3.73974, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5364766596293689}
{"step": 26205, "loss": 3.736797332763672, "grad_norm": 0.7294418215751648, "learning_rate": 0.0001, "elbo": 3.7368, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5367698803659395}
{"step": 26210, "loss": 3.719915008544922, "grad_norm": 0.70606529712677, "learning_rate": 0.0001, "elbo": 3.71992, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.53706310110251}
{"step": 26215, "loss": 3.725525665283203, "grad_norm": 0.6723562479019165, "learning_rate": 0.0001, "elbo": 3.72553, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5373563218390804}
{"step": 26220, "loss": 3.747216796875, "grad_norm": 0.6960316300392151, "learning_rate": 0.0001, "elbo": 3.74722, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.537649542575651}
{"step": 26225, "loss": 3.736048126220703, "grad_norm": 0.7074357271194458, "learning_rate": 0.0001, "elbo": 3.73605, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5379427633122216}
{"step": 26230, "loss": 3.742959976196289, "grad_norm": 0.67205411195755, "learning_rate": 0.0001, "elbo": 3.74296, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.538235984048792}
{"step": 26235, "loss": 3.737345886230469, "grad_norm": 0.7036055326461792, "learning_rate": 0.0001, "elbo": 3.73735, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5385292047853625}
{"step": 26240, "loss": 3.7453819274902345, "grad_norm": 0.9603764414787292, "learning_rate": 0.0001, "elbo": 3.74538, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.538822425521933}
{"step": 26245, "loss": 3.741619110107422, "grad_norm": 0.7152365446090698, "learning_rate": 0.0001, "elbo": 3.74162, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5391156462585034}
{"step": 26250, "loss": 3.761399841308594, "grad_norm": 0.6907104849815369, "learning_rate": 0.0001, "elbo": 3.7614, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5394088669950738}
{"step": 26255, "loss": 3.760240936279297, "grad_norm": 0.6888627409934998, "learning_rate": 0.0001, "elbo": 3.76024, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5397020877316443}
{"step": 26260, "loss": 3.7352848052978516, "grad_norm": 0.6648256778717041, "learning_rate": 0.0001, "elbo": 3.73528, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5399953084682148}
{"step": 26265, "loss": 3.729634094238281, "grad_norm": 0.6589856743812561, "learning_rate": 0.0001, "elbo": 3.72963, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5402885292047852}
{"step": 26270, "loss": 3.7341892242431642, "grad_norm": 0.6412859559059143, "learning_rate": 0.0001, "elbo": 3.73419, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5405817499413559}
{"step": 26275, "loss": 3.7345516204833986, "grad_norm": 0.6526793241500854, "learning_rate": 0.0001, "elbo": 3.73455, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5408749706779263}
{"step": 26280, "loss": 3.736104965209961, "grad_norm": 0.6557773947715759, "learning_rate": 0.0001, "elbo": 3.73611, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5411681914144968}
{"step": 26285, "loss": 3.7480201721191406, "grad_norm": 1.072562575340271, "learning_rate": 0.0001, "elbo": 3.74802, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5414614121510675}
{"step": 26290, "loss": 3.731425094604492, "grad_norm": 0.7398903369903564, "learning_rate": 0.0001, "elbo": 3.73143, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.541754632887638}
{"step": 26295, "loss": 3.74884033203125, "grad_norm": 0.6352856159210205, "learning_rate": 0.0001, "elbo": 3.74884, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5420478536242084}
{"step": 26300, "loss": 3.736275482177734, "grad_norm": 0.6249490976333618, "learning_rate": 0.0001, "elbo": 3.73628, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5423410743607788}
{"step": 26305, "loss": 3.7408592224121096, "grad_norm": 0.7578490972518921, "learning_rate": 0.0001, "elbo": 3.74086, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5426342950973493}
{"step": 26310, "loss": 3.7570594787597655, "grad_norm": 0.7200087904930115, "learning_rate": 0.0001, "elbo": 3.75706, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5429275158339197}
{"step": 26315, "loss": 3.744557189941406, "grad_norm": 0.6572738289833069, "learning_rate": 0.0001, "elbo": 3.74456, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5432207365704902}
{"step": 26320, "loss": 3.7272029876708985, "grad_norm": 0.7132430076599121, "learning_rate": 0.0001, "elbo": 3.7272, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5435139573070606}
{"step": 26325, "loss": 3.7396644592285155, "grad_norm": 0.7104228138923645, "learning_rate": 0.0001, "elbo": 3.73966, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.543807178043631}
{"step": 26330, "loss": 3.740245819091797, "grad_norm": 0.6539510488510132, "learning_rate": 0.0001, "elbo": 3.74025, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5441003987802018}
{"step": 26335, "loss": 3.74011344909668, "grad_norm": 0.7203420996665955, "learning_rate": 0.0001, "elbo": 3.74011, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5443936195167722}
{"step": 26340, "loss": 3.731302261352539, "grad_norm": 0.7330775856971741, "learning_rate": 0.0001, "elbo": 3.7313, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5446868402533427}
{"step": 26345, "loss": 3.736855697631836, "grad_norm": 0.8752711415290833, "learning_rate": 0.0001, "elbo": 3.73686, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5449800609899134}
{"step": 26350, "loss": 3.7267269134521483, "grad_norm": 0.6532132625579834, "learning_rate": 0.0001, "elbo": 3.72673, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5452732817264838}
{"step": 26355, "loss": 3.7395469665527346, "grad_norm": 0.6991098523139954, "learning_rate": 0.0001, "elbo": 3.73955, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5455665024630543}
{"step": 26360, "loss": 3.737257385253906, "grad_norm": 0.6842107772827148, "learning_rate": 0.0001, "elbo": 3.73726, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5458597231996247}
{"step": 26365, "loss": 3.727082443237305, "grad_norm": 0.6894490718841553, "learning_rate": 0.0001, "elbo": 3.72708, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5461529439361952}
{"step": 26370, "loss": 3.7253299713134767, "grad_norm": 0.6723558902740479, "learning_rate": 0.0001, "elbo": 3.72533, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5464461646727656}
{"step": 26375, "loss": 3.751769256591797, "grad_norm": 0.7461676597595215, "learning_rate": 0.0001, "elbo": 3.75177, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.546739385409336}
{"step": 26380, "loss": 3.7412200927734376, "grad_norm": 0.6968275904655457, "learning_rate": 0.0001, "elbo": 3.74122, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5470326061459065}
{"step": 26385, "loss": 3.7210182189941405, "grad_norm": 0.669556200504303, "learning_rate": 0.0001, "elbo": 3.72102, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.547325826882477}
{"step": 26390, "loss": 3.732052230834961, "grad_norm": 0.6513778567314148, "learning_rate": 0.0001, "elbo": 3.73205, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5476190476190477}
{"step": 26395, "loss": 3.7414398193359375, "grad_norm": 0.8292707800865173, "learning_rate": 0.0001, "elbo": 3.74144, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5479122683556181}
{"step": 26400, "loss": 3.743784713745117, "grad_norm": 0.6074082255363464, "learning_rate": 0.0001, "elbo": 3.74378, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5482054890921886}
{"step": 26405, "loss": 3.7534717559814452, "grad_norm": 0.6678458452224731, "learning_rate": 0.0001, "elbo": 3.75347, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.548498709828759}
{"step": 26410, "loss": 3.723037338256836, "grad_norm": 0.7546380162239075, "learning_rate": 0.0001, "elbo": 3.72304, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5487919305653297}
{"step": 26415, "loss": 3.753163146972656, "grad_norm": 1.1139686107635498, "learning_rate": 0.0001, "elbo": 3.75316, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5490851513019002}
{"step": 26420, "loss": 3.756336212158203, "grad_norm": 0.6850605010986328, "learning_rate": 0.0001, "elbo": 3.75634, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5493783720384706}
{"step": 26425, "loss": 3.7429309844970704, "grad_norm": 0.6444339752197266, "learning_rate": 0.0001, "elbo": 3.74293, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.549671592775041}
{"step": 26430, "loss": 3.7344009399414064, "grad_norm": 0.6827951669692993, "learning_rate": 0.0001, "elbo": 3.7344, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5499648135116115}
{"step": 26435, "loss": 3.7315742492675783, "grad_norm": 0.7960867881774902, "learning_rate": 0.0001, "elbo": 3.73157, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.550258034248182}
{"step": 26440, "loss": 3.723654937744141, "grad_norm": 0.6830714344978333, "learning_rate": 0.0001, "elbo": 3.72366, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5505512549847524}
{"step": 26445, "loss": 3.736347198486328, "grad_norm": 0.684709906578064, "learning_rate": 0.0001, "elbo": 3.73635, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.550844475721323}
{"step": 26450, "loss": 3.729840087890625, "grad_norm": 0.8340582847595215, "learning_rate": 0.0001, "elbo": 3.72984, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5511376964578933}
{"step": 26455, "loss": 3.7274497985839843, "grad_norm": 0.6634598970413208, "learning_rate": 0.0001, "elbo": 3.72745, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.551430917194464}
{"step": 26460, "loss": 3.7396484375, "grad_norm": 0.6786848306655884, "learning_rate": 0.0001, "elbo": 3.73965, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5517241379310345}
{"step": 26465, "loss": 3.7461471557617188, "grad_norm": 0.6377208828926086, "learning_rate": 0.0001, "elbo": 3.74615, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.552017358667605}
{"step": 26470, "loss": 3.7430057525634766, "grad_norm": 0.7454090118408203, "learning_rate": 0.0001, "elbo": 3.74301, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5523105794041756}
{"step": 26475, "loss": 3.7349578857421877, "grad_norm": 0.661458432674408, "learning_rate": 0.0001, "elbo": 3.73496, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.552603800140746}
{"step": 26480, "loss": 3.7593307495117188, "grad_norm": 0.8422238230705261, "learning_rate": 0.0001, "elbo": 3.75933, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5528970208773165}
{"step": 26485, "loss": 3.7292274475097655, "grad_norm": 0.6505508422851562, "learning_rate": 0.0001, "elbo": 3.72923, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.553190241613887}
{"step": 26490, "loss": 3.7285743713378907, "grad_norm": 0.7142215967178345, "learning_rate": 0.0001, "elbo": 3.72857, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5534834623504574}
{"step": 26495, "loss": 3.7329654693603516, "grad_norm": 0.7080323696136475, "learning_rate": 0.0001, "elbo": 3.73297, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5537766830870279}
{"step": 26500, "loss": 3.7361236572265626, "grad_norm": 0.6738479137420654, "learning_rate": 0.0001, "elbo": 3.73612, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5540699038235983}
{"step": 26505, "loss": 3.7229759216308596, "grad_norm": 0.6822200417518616, "learning_rate": 0.0001, "elbo": 3.72298, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5543631245601688}
{"step": 26510, "loss": 3.736281967163086, "grad_norm": 0.9487142562866211, "learning_rate": 0.0001, "elbo": 3.73628, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5546563452967392}
{"step": 26515, "loss": 3.725343704223633, "grad_norm": 0.7335153222084045, "learning_rate": 0.0001, "elbo": 3.72534, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.55494956603331}
{"step": 26520, "loss": 3.7343826293945312, "grad_norm": 0.7251141667366028, "learning_rate": 0.0001, "elbo": 3.73438, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5552427867698804}
{"step": 26525, "loss": 3.7316261291503907, "grad_norm": 0.8942455649375916, "learning_rate": 0.0001, "elbo": 3.73163, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5555360075064508}
{"step": 26530, "loss": 3.741956329345703, "grad_norm": 0.6879755258560181, "learning_rate": 0.0001, "elbo": 3.74196, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5558292282430215}
{"step": 26535, "loss": 3.757014846801758, "grad_norm": 0.7033980488777161, "learning_rate": 0.0001, "elbo": 3.75702, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.556122448979592}
{"step": 26540, "loss": 3.753313446044922, "grad_norm": 0.7173032164573669, "learning_rate": 0.0001, "elbo": 3.75331, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5564156697161624}
{"step": 26545, "loss": 3.718633270263672, "grad_norm": 0.7996748685836792, "learning_rate": 0.0001, "elbo": 3.71863, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5567088904527329}
{"step": 26550, "loss": 3.745819091796875, "grad_norm": 0.6606098413467407, "learning_rate": 0.0001, "elbo": 3.74582, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5570021111893033}
{"step": 26555, "loss": 3.7268985748291015, "grad_norm": 0.6541708111763, "learning_rate": 0.0001, "elbo": 3.7269, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5572953319258738}
{"step": 26560, "loss": 3.754103088378906, "grad_norm": 0.6560136079788208, "learning_rate": 0.0001, "elbo": 3.7541, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5575885526624442}
{"step": 26565, "loss": 3.7405242919921875, "grad_norm": 0.6664515733718872, "learning_rate": 0.0001, "elbo": 3.74052, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5578817733990147}
{"step": 26570, "loss": 3.7334922790527343, "grad_norm": 0.663369357585907, "learning_rate": 0.0001, "elbo": 3.73349, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5581749941355851}
{"step": 26575, "loss": 3.740999221801758, "grad_norm": 0.7324579358100891, "learning_rate": 0.0001, "elbo": 3.741, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5584682148721558}
{"step": 26580, "loss": 3.7518138885498047, "grad_norm": 0.7319678068161011, "learning_rate": 0.0001, "elbo": 3.75181, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5587614356087263}
{"step": 26580, "loss": 3.7518138885498047, "grad_norm": 0.7319678068161011, "learning_rate": 0.0001, "elbo": 3.75181, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5587614356087263}
{"step": 26585, "loss": 3.729863739013672, "grad_norm": 0.6941600441932678, "learning_rate": 0.0001, "elbo": 3.72986, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5590546563452967}
{"step": 26590, "loss": 3.726612854003906, "grad_norm": 0.701824426651001, "learning_rate": 0.0001, "elbo": 3.72661, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5593478770818674}
{"step": 26595, "loss": 3.730978012084961, "grad_norm": 0.6637766361236572, "learning_rate": 0.0001, "elbo": 3.73098, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5596410978184378}
{"step": 26600, "loss": 3.7374473571777345, "grad_norm": 0.6817801594734192, "learning_rate": 0.0001, "elbo": 3.73745, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5599343185550083}
{"step": 26605, "loss": 3.722992706298828, "grad_norm": 0.6870080828666687, "learning_rate": 0.0001, "elbo": 3.72299, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5602275392915788}
{"step": 26610, "loss": 3.713678741455078, "grad_norm": 0.7229723334312439, "learning_rate": 0.0001, "elbo": 3.71368, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5605207600281492}
{"step": 26615, "loss": 3.767599868774414, "grad_norm": 0.7693787813186646, "learning_rate": 0.0001, "elbo": 3.7676, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5608139807647197}
{"step": 26620, "loss": 3.720582962036133, "grad_norm": 0.6851159334182739, "learning_rate": 0.0001, "elbo": 3.72058, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5611072015012901}
{"step": 26625, "loss": 3.7238533020019533, "grad_norm": 0.6829482913017273, "learning_rate": 0.0001, "elbo": 3.72385, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5614004222378606}
{"step": 26630, "loss": 3.734427642822266, "grad_norm": 0.7229098081588745, "learning_rate": 0.0001, "elbo": 3.73443, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.561693642974431}
{"step": 26635, "loss": 3.7519927978515626, "grad_norm": 1.2317312955856323, "learning_rate": 0.0001, "elbo": 3.75199, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5619868637110015}
{"step": 26640, "loss": 3.7313003540039062, "grad_norm": 0.6978418231010437, "learning_rate": 0.0001, "elbo": 3.7313, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5622800844475722}
{"step": 26645, "loss": 3.723186492919922, "grad_norm": 0.7083508968353271, "learning_rate": 0.0001, "elbo": 3.72319, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5625733051841426}
{"step": 26650, "loss": 3.727185821533203, "grad_norm": 0.9401165843009949, "learning_rate": 0.0001, "elbo": 3.72719, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.562866525920713}
{"step": 26655, "loss": 3.741079330444336, "grad_norm": 0.6904851794242859, "learning_rate": 0.0001, "elbo": 3.74108, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5631597466572837}
{"step": 26660, "loss": 3.728789520263672, "grad_norm": 0.805681586265564, "learning_rate": 0.0001, "elbo": 3.72879, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5634529673938542}
{"step": 26665, "loss": 3.7402507781982424, "grad_norm": 0.6976442933082581, "learning_rate": 0.0001, "elbo": 3.74025, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5637461881304247}
{"step": 26670, "loss": 3.746000289916992, "grad_norm": 1.3349955081939697, "learning_rate": 0.0001, "elbo": 3.746, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.564039408866995}
{"step": 26675, "loss": 3.730707550048828, "grad_norm": 0.706407368183136, "learning_rate": 0.0001, "elbo": 3.73071, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5643326296035656}
{"step": 26680, "loss": 3.743608093261719, "grad_norm": 0.6905243396759033, "learning_rate": 0.0001, "elbo": 3.74361, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.564625850340136}
{"step": 26685, "loss": 3.7451229095458984, "grad_norm": 0.6508228778839111, "learning_rate": 0.0001, "elbo": 3.74512, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5649190710767065}
{"step": 26690, "loss": 3.7476234436035156, "grad_norm": 0.6741530895233154, "learning_rate": 0.0001, "elbo": 3.74762, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.565212291813277}
{"step": 26695, "loss": 3.747451400756836, "grad_norm": 0.6802815198898315, "learning_rate": 0.0001, "elbo": 3.74745, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5655055125498474}
{"step": 26700, "loss": 3.732721710205078, "grad_norm": 0.7190822958946228, "learning_rate": 0.0001, "elbo": 3.73272, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.565798733286418}
{"step": 26705, "loss": 3.7314208984375, "grad_norm": 0.6910989880561829, "learning_rate": 0.0001, "elbo": 3.73142, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5660919540229885}
{"step": 26710, "loss": 3.7483409881591796, "grad_norm": 0.7191604375839233, "learning_rate": 0.0001, "elbo": 3.74834, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.566385174759559}
{"step": 26715, "loss": 3.7329833984375, "grad_norm": 0.6890040040016174, "learning_rate": 0.0001, "elbo": 3.73298, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5666783954961296}
{"step": 26720, "loss": 3.736105728149414, "grad_norm": 0.6861095428466797, "learning_rate": 0.0001, "elbo": 3.73611, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5669716162327}
{"step": 26725, "loss": 3.756185531616211, "grad_norm": 0.6683018803596497, "learning_rate": 0.0001, "elbo": 3.75619, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5672648369692705}
{"step": 26730, "loss": 3.740094757080078, "grad_norm": 0.7567577362060547, "learning_rate": 0.0001, "elbo": 3.7401, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.567558057705841}
{"step": 26735, "loss": 3.736647033691406, "grad_norm": 0.8881714940071106, "learning_rate": 0.0001, "elbo": 3.73665, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5678512784424115}
{"step": 26740, "loss": 3.734619903564453, "grad_norm": 0.6702668070793152, "learning_rate": 0.0001, "elbo": 3.73462, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.568144499178982}
{"step": 26745, "loss": 3.710488510131836, "grad_norm": 0.6675503253936768, "learning_rate": 0.0001, "elbo": 3.71049, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5684377199155524}
{"step": 26750, "loss": 3.742262268066406, "grad_norm": 0.7123814225196838, "learning_rate": 0.0001, "elbo": 3.74226, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5687309406521228}
{"step": 26755, "loss": 3.732588195800781, "grad_norm": 0.6766638159751892, "learning_rate": 0.0001, "elbo": 3.73259, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5690241613886933}
{"step": 26760, "loss": 3.7047470092773436, "grad_norm": 0.6636619567871094, "learning_rate": 0.0001, "elbo": 3.70475, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.569317382125264}
{"step": 26765, "loss": 3.7509414672851564, "grad_norm": 0.6525682806968689, "learning_rate": 0.0001, "elbo": 3.75094, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5696106028618344}
{"step": 26770, "loss": 3.7353530883789063, "grad_norm": 0.6861343383789062, "learning_rate": 0.0001, "elbo": 3.73535, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5699038235984049}
{"step": 26775, "loss": 3.7358169555664062, "grad_norm": 0.8724676370620728, "learning_rate": 0.0001, "elbo": 3.73582, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5701970443349755}
{"step": 26780, "loss": 3.727772521972656, "grad_norm": 0.6578548550605774, "learning_rate": 0.0001, "elbo": 3.72777, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.570490265071546}
{"step": 26785, "loss": 3.7355667114257813, "grad_norm": 0.6735323071479797, "learning_rate": 0.0001, "elbo": 3.73557, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5707834858081164}
{"step": 26790, "loss": 3.726542663574219, "grad_norm": 0.6773928999900818, "learning_rate": 0.0001, "elbo": 3.72654, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.571076706544687}
{"step": 26795, "loss": 3.7477081298828123, "grad_norm": 0.6805819869041443, "learning_rate": 0.0001, "elbo": 3.74771, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5713699272812574}
{"step": 26800, "loss": 3.7277687072753904, "grad_norm": 0.9455089569091797, "learning_rate": 0.0001, "elbo": 3.72777, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5716631480178278}
{"step": 26805, "loss": 3.739000701904297, "grad_norm": 0.6472092270851135, "learning_rate": 0.0001, "elbo": 3.739, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5719563687543983}
{"step": 26810, "loss": 3.7258621215820313, "grad_norm": 0.7067530155181885, "learning_rate": 0.0001, "elbo": 3.72586, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5722495894909687}
{"step": 26810, "loss": 3.7258621215820313, "grad_norm": 0.7067530155181885, "learning_rate": 0.0001, "elbo": 3.72586, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5722495894909687}
{"step": 26815, "loss": 3.7318962097167967, "grad_norm": 0.6465001106262207, "learning_rate": 0.0001, "elbo": 3.7319, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5725428102275392}
{"step": 26820, "loss": 3.7177005767822267, "grad_norm": 0.6769406795501709, "learning_rate": 0.0001, "elbo": 3.7177, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5728360309641096}
{"step": 26825, "loss": 3.749350738525391, "grad_norm": 0.7007476687431335, "learning_rate": 0.0001, "elbo": 3.74935, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5731292517006803}
{"step": 26830, "loss": 3.725383758544922, "grad_norm": 0.6747730374336243, "learning_rate": 0.0001, "elbo": 3.72538, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5734224724372508}
{"step": 26835, "loss": 3.746438217163086, "grad_norm": 0.6909791827201843, "learning_rate": 0.0001, "elbo": 3.74644, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5737156931738212}
{"step": 26840, "loss": 3.7319374084472656, "grad_norm": 0.6850407719612122, "learning_rate": 0.0001, "elbo": 3.73194, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5740089139103919}
{"step": 26845, "loss": 3.7240386962890626, "grad_norm": 0.6455448865890503, "learning_rate": 0.0001, "elbo": 3.72404, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5743021346469623}
{"step": 26845, "loss": 3.7240386962890626, "grad_norm": 0.6455448865890503, "learning_rate": 0.0001, "elbo": 3.72404, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5743021346469623}
{"step": 26850, "loss": 3.727001953125, "grad_norm": 0.6974610686302185, "learning_rate": 0.0001, "elbo": 3.727, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5745953553835328}
{"step": 26855, "loss": 3.7333915710449217, "grad_norm": 0.6993145942687988, "learning_rate": 0.0001, "elbo": 3.73339, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5748885761201032}
{"step": 26860, "loss": 3.7410186767578124, "grad_norm": 0.816336452960968, "learning_rate": 0.0001, "elbo": 3.74102, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5751817968566737}
{"step": 26865, "loss": 3.7311126708984377, "grad_norm": 0.7698284387588501, "learning_rate": 0.0001, "elbo": 3.73111, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5754750175932442}
{"step": 26870, "loss": 3.7318382263183594, "grad_norm": 0.8534447550773621, "learning_rate": 0.0001, "elbo": 3.73184, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5757682383298146}
{"step": 26875, "loss": 3.7310394287109374, "grad_norm": 0.7703247666358948, "learning_rate": 0.0001, "elbo": 3.73104, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.576061459066385}
{"step": 26880, "loss": 3.7242774963378906, "grad_norm": 0.6901226043701172, "learning_rate": 0.0001, "elbo": 3.72428, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5763546798029555}
{"step": 26885, "loss": 3.7508670806884767, "grad_norm": 0.9419646263122559, "learning_rate": 0.0001, "elbo": 3.75087, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5766479005395262}
{"step": 26890, "loss": 3.7423374176025392, "grad_norm": 0.692729115486145, "learning_rate": 0.0001, "elbo": 3.74234, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5769411212760966}
{"step": 26895, "loss": 3.738149642944336, "grad_norm": 0.7025671601295471, "learning_rate": 0.0001, "elbo": 3.73815, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.577234342012667}
{"step": 26900, "loss": 3.7122882843017577, "grad_norm": 0.649589478969574, "learning_rate": 0.0001, "elbo": 3.71229, "mask_frac": 0.4996, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5775275627492378}
{"step": 26905, "loss": 3.7219261169433593, "grad_norm": 0.6828599572181702, "learning_rate": 0.0001, "elbo": 3.72193, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5778207834858082}
{"step": 26910, "loss": 3.740951156616211, "grad_norm": 0.6531699299812317, "learning_rate": 0.0001, "elbo": 3.74095, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5781140042223787}
{"step": 26915, "loss": 3.728345489501953, "grad_norm": 0.6507318019866943, "learning_rate": 0.0001, "elbo": 3.72835, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5784072249589491}
{"step": 26920, "loss": 3.719715881347656, "grad_norm": 0.6851858496665955, "learning_rate": 0.0001, "elbo": 3.71972, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5787004456955196}
{"step": 26925, "loss": 3.7276981353759764, "grad_norm": 0.7544836401939392, "learning_rate": 0.0001, "elbo": 3.7277, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.57899366643209}
{"step": 26930, "loss": 3.720755386352539, "grad_norm": 0.7098874449729919, "learning_rate": 0.0001, "elbo": 3.72076, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5792868871686605}
{"step": 26935, "loss": 3.7403980255126954, "grad_norm": 1.313576102256775, "learning_rate": 0.0001, "elbo": 3.7404, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.579580107905231}
{"step": 26940, "loss": 3.746952438354492, "grad_norm": 0.7395880222320557, "learning_rate": 0.0001, "elbo": 3.74695, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5798733286418014}
{"step": 26945, "loss": 3.7268600463867188, "grad_norm": 0.6976245641708374, "learning_rate": 0.0001, "elbo": 3.72686, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.580166549378372}
{"step": 26950, "loss": 3.731795883178711, "grad_norm": 0.6873641610145569, "learning_rate": 0.0001, "elbo": 3.7318, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5804597701149425}
{"step": 26955, "loss": 3.7429878234863283, "grad_norm": 0.6900029182434082, "learning_rate": 0.0001, "elbo": 3.74299, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.580752990851513}
{"step": 26960, "loss": 3.7213905334472654, "grad_norm": 0.7466303110122681, "learning_rate": 0.0001, "elbo": 3.72139, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5810462115880837}
{"step": 26965, "loss": 3.730446624755859, "grad_norm": 0.7307167053222656, "learning_rate": 0.0001, "elbo": 3.73045, "mask_frac": 0.5013, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5813394323246541}
{"step": 26970, "loss": 3.7527935028076174, "grad_norm": 0.6834270358085632, "learning_rate": 0.0001, "elbo": 3.75279, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5816326530612246}
{"step": 26975, "loss": 3.7447071075439453, "grad_norm": 0.675305962562561, "learning_rate": 0.0001, "elbo": 3.74471, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.581925873797795}
{"step": 26980, "loss": 3.735169219970703, "grad_norm": 0.6738702058792114, "learning_rate": 0.0001, "elbo": 3.73517, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5822190945343655}
{"step": 26985, "loss": 3.7336299896240233, "grad_norm": 0.6728871464729309, "learning_rate": 0.0001, "elbo": 3.73363, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.582512315270936}
{"step": 26990, "loss": 3.7360275268554686, "grad_norm": 0.6888622045516968, "learning_rate": 0.0001, "elbo": 3.73603, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5828055360075064}
{"step": 26995, "loss": 3.7188365936279295, "grad_norm": 0.6672109365463257, "learning_rate": 0.0001, "elbo": 3.71884, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5830987567440769}
{"step": 27000, "loss": 3.7463348388671873, "grad_norm": 1.258297324180603, "learning_rate": 0.0001, "elbo": 3.74633, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5833919774806473}
{"step": 27005, "loss": 3.741904067993164, "grad_norm": 0.7390699982643127, "learning_rate": 0.0001, "elbo": 3.7419, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5836851982172178}
{"step": 27005, "loss": 3.741904067993164, "grad_norm": 0.7390699982643127, "learning_rate": 0.0001, "elbo": 3.7419, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5836851982172178}
{"step": 27005, "loss": 3.741904067993164, "grad_norm": 0.7390699982643127, "learning_rate": 0.0001, "elbo": 3.7419, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5836851982172178}
{"step": 27010, "loss": 3.729315185546875, "grad_norm": 0.6007156372070312, "learning_rate": 0.0001, "elbo": 3.72931, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5839784189537884}
{"step": 27010, "loss": 3.729315185546875, "grad_norm": 0.6007156372070312, "learning_rate": 0.0001, "elbo": 3.72931, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5839784189537884}
{"step": 27015, "loss": 3.734030914306641, "grad_norm": 0.8133919835090637, "learning_rate": 0.0001, "elbo": 3.73403, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.584271639690359}
{"step": 27020, "loss": 3.7284805297851564, "grad_norm": 0.6651749610900879, "learning_rate": 0.0001, "elbo": 3.72848, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5845648604269293}
{"step": 27025, "loss": 3.7302314758300783, "grad_norm": 0.7042424082756042, "learning_rate": 0.0001, "elbo": 3.73023, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5848580811635}
{"step": 27030, "loss": 3.7284263610839843, "grad_norm": 0.6457875967025757, "learning_rate": 0.0001, "elbo": 3.72843, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5851513019000705}
{"step": 27035, "loss": 3.728011703491211, "grad_norm": 0.6978619694709778, "learning_rate": 0.0001, "elbo": 3.72801, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.585444522636641}
{"step": 27040, "loss": 3.7402957916259765, "grad_norm": 0.8068510293960571, "learning_rate": 0.0001, "elbo": 3.7403, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5857377433732114}
{"step": 27045, "loss": 3.7346866607666014, "grad_norm": 0.7073001861572266, "learning_rate": 0.0001, "elbo": 3.73469, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5860309641097818}
{"step": 27050, "loss": 3.721873092651367, "grad_norm": 0.7318355441093445, "learning_rate": 0.0001, "elbo": 3.72187, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5863241848463523}
{"step": 27055, "loss": 3.7220169067382813, "grad_norm": 0.6555929780006409, "learning_rate": 0.0001, "elbo": 3.72202, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5866174055829227}
{"step": 27060, "loss": 3.7426742553710937, "grad_norm": 0.6935165524482727, "learning_rate": 0.0001, "elbo": 3.74267, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5869106263194932}
{"step": 27065, "loss": 3.7455413818359373, "grad_norm": 0.6814263463020325, "learning_rate": 0.0001, "elbo": 3.74554, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5872038470560637}
{"step": 27070, "loss": 3.729491424560547, "grad_norm": 0.6724979877471924, "learning_rate": 0.0001, "elbo": 3.72949, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5874970677926343}
{"step": 27075, "loss": 3.727988433837891, "grad_norm": 0.6514295935630798, "learning_rate": 0.0001, "elbo": 3.72799, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5877902885292048}
{"step": 27080, "loss": 3.726671600341797, "grad_norm": 0.6486175656318665, "learning_rate": 0.0001, "elbo": 3.72667, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5880835092657752}
{"step": 27085, "loss": 3.7286338806152344, "grad_norm": 0.7511657476425171, "learning_rate": 0.0001, "elbo": 3.72863, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.588376730002346}
{"step": 27090, "loss": 3.7197391510009767, "grad_norm": 0.717538058757782, "learning_rate": 0.0001, "elbo": 3.71974, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5886699507389164}
{"step": 27095, "loss": 3.732950973510742, "grad_norm": 0.8855931162834167, "learning_rate": 0.0001, "elbo": 3.73295, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5889631714754868}
{"step": 27100, "loss": 3.7463844299316404, "grad_norm": 0.6509588360786438, "learning_rate": 0.0001, "elbo": 3.74638, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5892563922120573}
{"step": 27105, "loss": 3.7408821105957033, "grad_norm": 0.7174865007400513, "learning_rate": 0.0001, "elbo": 3.74088, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5895496129486277}
{"step": 27110, "loss": 3.738298034667969, "grad_norm": 0.6722555756568909, "learning_rate": 0.0001, "elbo": 3.7383, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5898428336851982}
{"step": 27115, "loss": 3.729262924194336, "grad_norm": 0.739069938659668, "learning_rate": 0.0001, "elbo": 3.72926, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5901360544217686}
{"step": 27120, "loss": 3.732984924316406, "grad_norm": 0.78404700756073, "learning_rate": 0.0001, "elbo": 3.73298, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.590429275158339}
{"step": 27125, "loss": 3.7225738525390626, "grad_norm": 0.6837067008018494, "learning_rate": 0.0001, "elbo": 3.72257, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5907224958949095}
{"step": 27130, "loss": 3.7247222900390624, "grad_norm": 0.6816504597663879, "learning_rate": 0.0001, "elbo": 3.72472, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5910157166314802}
{"step": 27135, "loss": 3.7386028289794924, "grad_norm": 0.662838876247406, "learning_rate": 0.0001, "elbo": 3.7386, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5913089373680507}
{"step": 27140, "loss": 3.7090930938720703, "grad_norm": 0.671430230140686, "learning_rate": 0.0001, "elbo": 3.70909, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5916021581046211}
{"step": 27145, "loss": 3.723215103149414, "grad_norm": 0.677844226360321, "learning_rate": 0.0001, "elbo": 3.72322, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5918953788411918}
{"step": 27150, "loss": 3.7200695037841798, "grad_norm": 0.6967824697494507, "learning_rate": 0.0001, "elbo": 3.72007, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5921885995777623}
{"step": 27155, "loss": 3.7421596527099608, "grad_norm": 0.6995875239372253, "learning_rate": 0.0001, "elbo": 3.74216, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5924818203143327}
{"step": 27160, "loss": 3.7188865661621096, "grad_norm": 0.8001232743263245, "learning_rate": 0.0001, "elbo": 3.71889, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5927750410509032}
{"step": 27165, "loss": 3.7212677001953125, "grad_norm": 0.6744524240493774, "learning_rate": 0.0001, "elbo": 3.72127, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5930682617874736}
{"step": 27170, "loss": 3.7322105407714843, "grad_norm": 0.9764469861984253, "learning_rate": 0.0001, "elbo": 3.73221, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.593361482524044}
{"step": 27175, "loss": 3.7341590881347657, "grad_norm": 0.7864521145820618, "learning_rate": 0.0001, "elbo": 3.73416, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5936547032606145}
{"step": 27180, "loss": 3.7370655059814455, "grad_norm": 0.7431182861328125, "learning_rate": 0.0001, "elbo": 3.73707, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.593947923997185}
{"step": 27185, "loss": 3.7299427032470702, "grad_norm": 0.6908766031265259, "learning_rate": 0.0001, "elbo": 3.72994, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5942411447337554}
{"step": 27190, "loss": 3.7467910766601564, "grad_norm": 0.6504005193710327, "learning_rate": 0.0001, "elbo": 3.74679, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.594534365470326}
{"step": 27195, "loss": 3.722895050048828, "grad_norm": 0.702477753162384, "learning_rate": 0.0001, "elbo": 3.7229, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5948275862068966}
{"step": 27200, "loss": 3.734880065917969, "grad_norm": 0.8696697950363159, "learning_rate": 0.0001, "elbo": 3.73488, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.595120806943467}
{"step": 27205, "loss": 3.7324363708496096, "grad_norm": 0.6706423163414001, "learning_rate": 0.0001, "elbo": 3.73244, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5954140276800375}
{"step": 27210, "loss": 3.7273441314697267, "grad_norm": 0.6600042581558228, "learning_rate": 0.0001, "elbo": 3.72734, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5957072484166082}
{"step": 27215, "loss": 3.727347564697266, "grad_norm": 0.6827043890953064, "learning_rate": 0.0001, "elbo": 3.72735, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5960004691531786}
{"step": 27220, "loss": 3.7391510009765625, "grad_norm": 0.7480504512786865, "learning_rate": 0.0001, "elbo": 3.73915, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.596293689889749}
{"step": 27225, "loss": 3.72509651184082, "grad_norm": 0.7420713901519775, "learning_rate": 0.0001, "elbo": 3.7251, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5965869106263195}
{"step": 27230, "loss": 3.7187545776367186, "grad_norm": 0.7371690273284912, "learning_rate": 0.0001, "elbo": 3.71875, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.59688013136289}
{"step": 27235, "loss": 3.7150115966796875, "grad_norm": 0.6727964282035828, "learning_rate": 0.0001, "elbo": 3.71501, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5971733520994604}
{"step": 27240, "loss": 3.7303386688232423, "grad_norm": 0.7247251272201538, "learning_rate": 0.0001, "elbo": 3.73034, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5974665728360309}
{"step": 27245, "loss": 3.7198711395263673, "grad_norm": 0.823205828666687, "learning_rate": 0.0001, "elbo": 3.71987, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5977597935726013}
{"step": 27250, "loss": 3.757395935058594, "grad_norm": 0.7116758823394775, "learning_rate": 0.0001, "elbo": 3.7574, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5980530143091718}
{"step": 27255, "loss": 3.749630355834961, "grad_norm": 0.713722288608551, "learning_rate": 0.0001, "elbo": 3.74963, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5983462350457425}
{"step": 27260, "loss": 3.7493431091308596, "grad_norm": 0.6687597632408142, "learning_rate": 0.0001, "elbo": 3.74934, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.598639455782313}
{"step": 27265, "loss": 3.724563980102539, "grad_norm": 0.6853957772254944, "learning_rate": 0.0001, "elbo": 3.72456, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5989326765188834}
{"step": 27270, "loss": 3.709136962890625, "grad_norm": 0.7138869166374207, "learning_rate": 0.0001, "elbo": 3.70914, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.599225897255454}
{"step": 27275, "loss": 3.737937164306641, "grad_norm": 0.7166622281074524, "learning_rate": 0.0001, "elbo": 3.73794, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.5995191179920245}
{"step": 27280, "loss": 3.7301673889160156, "grad_norm": 0.7153247594833374, "learning_rate": 0.0001, "elbo": 3.73017, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.599812338728595}
{"step": 27285, "loss": 3.724042510986328, "grad_norm": 0.7044864296913147, "learning_rate": 0.0001, "elbo": 3.72404, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6001055594651654}
{"step": 27290, "loss": 3.7312084197998048, "grad_norm": 0.6955354809761047, "learning_rate": 0.0001, "elbo": 3.73121, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6003987802017359}
{"step": 27295, "loss": 3.738506317138672, "grad_norm": 1.0247725248336792, "learning_rate": 0.0001, "elbo": 3.73851, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6006920009383063}
{"step": 27300, "loss": 3.733489227294922, "grad_norm": 0.6906407475471497, "learning_rate": 0.0001, "elbo": 3.73349, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6009852216748768}
{"step": 27305, "loss": 3.7223915100097655, "grad_norm": 0.6612165570259094, "learning_rate": 0.0001, "elbo": 3.72239, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6012784424114472}
{"step": 27310, "loss": 3.733620834350586, "grad_norm": 0.7060428857803345, "learning_rate": 0.0001, "elbo": 3.73362, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6015716631480177}
{"step": 27315, "loss": 3.7251087188720704, "grad_norm": 0.7304823994636536, "learning_rate": 0.0001, "elbo": 3.72511, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6018648838845884}
{"step": 27320, "loss": 3.7301864624023438, "grad_norm": 0.793417751789093, "learning_rate": 0.0001, "elbo": 3.73019, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6021581046211588}
{"step": 27325, "loss": 3.73583869934082, "grad_norm": 0.6769035458564758, "learning_rate": 0.0001, "elbo": 3.73584, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6024513253577293}
{"step": 27330, "loss": 3.714007568359375, "grad_norm": 0.6776753664016724, "learning_rate": 0.0001, "elbo": 3.71401, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6027445460943}
{"step": 27335, "loss": 3.738965606689453, "grad_norm": 0.7085759043693542, "learning_rate": 0.0001, "elbo": 3.73897, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6030377668308704}
{"step": 27340, "loss": 3.733721923828125, "grad_norm": 0.7368704080581665, "learning_rate": 0.0001, "elbo": 3.73372, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6033309875674409}
{"step": 27345, "loss": 3.712330627441406, "grad_norm": 0.7571136355400085, "learning_rate": 0.0001, "elbo": 3.71233, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6036242083040113}
{"step": 27350, "loss": 3.7264060974121094, "grad_norm": 1.0412232875823975, "learning_rate": 0.0001, "elbo": 3.72641, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6039174290405818}
{"step": 27355, "loss": 3.724721908569336, "grad_norm": 0.7030982375144958, "learning_rate": 0.0001, "elbo": 3.72472, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6042106497771522}
{"step": 27360, "loss": 3.7348907470703123, "grad_norm": 0.7036335468292236, "learning_rate": 0.0001, "elbo": 3.73489, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6045038705137227}
{"step": 27365, "loss": 3.7415882110595704, "grad_norm": 0.7212620377540588, "learning_rate": 0.0001, "elbo": 3.74159, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6047970912502931}
{"step": 27370, "loss": 3.724842071533203, "grad_norm": 0.6936198472976685, "learning_rate": 0.0001, "elbo": 3.72484, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6050903119868636}
{"step": 27375, "loss": 3.722997283935547, "grad_norm": 0.7421792149543762, "learning_rate": 0.0001, "elbo": 3.723, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.605383532723434}
{"step": 27380, "loss": 3.726835250854492, "grad_norm": 0.709520697593689, "learning_rate": 0.0001, "elbo": 3.72684, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6056767534600047}
{"step": 27385, "loss": 3.7022369384765623, "grad_norm": 0.677801787853241, "learning_rate": 0.0001, "elbo": 3.70224, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6059699741965752}
{"step": 27390, "loss": 3.7156219482421875, "grad_norm": 0.8019800186157227, "learning_rate": 0.0001, "elbo": 3.71562, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6062631949331456}
{"step": 27390, "loss": 3.7156219482421875, "grad_norm": 0.8019800186157227, "learning_rate": 0.0001, "elbo": 3.71562, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6062631949331456}
{"step": 27395, "loss": 3.7450229644775392, "grad_norm": 0.6535747051239014, "learning_rate": 0.0001, "elbo": 3.74502, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6065564156697163}
{"step": 27400, "loss": 3.7289688110351564, "grad_norm": 0.7001617550849915, "learning_rate": 0.0001, "elbo": 3.72897, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6068496364062868}
{"step": 27405, "loss": 3.731636810302734, "grad_norm": 0.7232893705368042, "learning_rate": 0.0001, "elbo": 3.73164, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6071428571428572}
{"step": 27405, "loss": 3.731636810302734, "grad_norm": 0.7232893705368042, "learning_rate": 0.0001, "elbo": 3.73164, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6071428571428572}
{"step": 27410, "loss": 3.7384902954101564, "grad_norm": 0.6987461447715759, "learning_rate": 0.0001, "elbo": 3.73849, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6074360778794277}
{"step": 27415, "loss": 3.7478492736816404, "grad_norm": 0.711710512638092, "learning_rate": 0.0001, "elbo": 3.74785, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6077292986159981}
{"step": 27420, "loss": 3.72765998840332, "grad_norm": 0.6673513054847717, "learning_rate": 0.0001, "elbo": 3.72766, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6080225193525686}
{"step": 27425, "loss": 3.7377883911132814, "grad_norm": 0.6995775103569031, "learning_rate": 0.0001, "elbo": 3.73779, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.608315740089139}
{"step": 27430, "loss": 3.712992858886719, "grad_norm": 0.7039170861244202, "learning_rate": 0.0001, "elbo": 3.71299, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6086089608257095}
{"step": 27435, "loss": 3.7282478332519533, "grad_norm": 0.6724250316619873, "learning_rate": 0.0001, "elbo": 3.72825, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.60890218156228}
{"step": 27440, "loss": 3.718297576904297, "grad_norm": 0.6345998048782349, "learning_rate": 0.0001, "elbo": 3.7183, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6091954022988506}
{"step": 27445, "loss": 3.724648284912109, "grad_norm": 0.7349030375480652, "learning_rate": 0.0001, "elbo": 3.72465, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.609488623035421}
{"step": 27450, "loss": 3.70943603515625, "grad_norm": 1.069222092628479, "learning_rate": 0.0001, "elbo": 3.70944, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6097818437719915}
{"step": 27455, "loss": 3.723061370849609, "grad_norm": 0.6975119709968567, "learning_rate": 0.0001, "elbo": 3.72306, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6100750645085622}
{"step": 27460, "loss": 3.71630859375, "grad_norm": 0.7480889558792114, "learning_rate": 0.0001, "elbo": 3.71631, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6103682852451326}
{"step": 27465, "loss": 3.7339176177978515, "grad_norm": 0.716585099697113, "learning_rate": 0.0001, "elbo": 3.73392, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.610661505981703}
{"step": 27470, "loss": 3.719343566894531, "grad_norm": 0.7830585241317749, "learning_rate": 0.0001, "elbo": 3.71934, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6109547267182736}
{"step": 27475, "loss": 3.72275390625, "grad_norm": 0.6607941389083862, "learning_rate": 0.0001, "elbo": 3.72275, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.611247947454844}
{"step": 27480, "loss": 3.733658218383789, "grad_norm": 0.6307964324951172, "learning_rate": 0.0001, "elbo": 3.73366, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6115411681914145}
{"step": 27485, "loss": 3.7200279235839844, "grad_norm": 0.824495255947113, "learning_rate": 0.0001, "elbo": 3.72003, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.611834388927985}
{"step": 27490, "loss": 3.71502685546875, "grad_norm": 0.8989217281341553, "learning_rate": 0.0001, "elbo": 3.71503, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6121276096645554}
{"step": 27495, "loss": 3.7249385833740236, "grad_norm": 0.7372245192527771, "learning_rate": 0.0001, "elbo": 3.72494, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6124208304011258}
{"step": 27500, "loss": 3.739482879638672, "grad_norm": 0.7567040324211121, "learning_rate": 0.0001, "elbo": 3.73948, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6127140511376965}
{"step": 27505, "loss": 3.7306365966796875, "grad_norm": 0.6988673806190491, "learning_rate": 0.0001, "elbo": 3.73064, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.613007271874267}
{"step": 27510, "loss": 3.727096176147461, "grad_norm": 0.7390287518501282, "learning_rate": 0.0001, "elbo": 3.7271, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6133004926108374}
{"step": 27515, "loss": 3.739598846435547, "grad_norm": 0.7378764748573303, "learning_rate": 0.0001, "elbo": 3.7396, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.613593713347408}
{"step": 27520, "loss": 3.7335830688476563, "grad_norm": 0.7039217352867126, "learning_rate": 0.0001, "elbo": 3.73358, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6138869340839785}
{"step": 27525, "loss": 3.7245719909667967, "grad_norm": 0.6906446814537048, "learning_rate": 0.0001, "elbo": 3.72457, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.614180154820549}
{"step": 27530, "loss": 3.720288848876953, "grad_norm": 0.6600560545921326, "learning_rate": 0.0001, "elbo": 3.72029, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6144733755571195}
{"step": 27535, "loss": 3.723500061035156, "grad_norm": 0.6416023373603821, "learning_rate": 0.0001, "elbo": 3.7235, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.61476659629369}
{"step": 27540, "loss": 3.723554992675781, "grad_norm": 0.6723536849021912, "learning_rate": 0.0001, "elbo": 3.72356, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6150598170302604}
{"step": 27545, "loss": 3.7383514404296876, "grad_norm": 0.7180091738700867, "learning_rate": 0.0001, "elbo": 3.73835, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6153530377668308}
{"step": 27550, "loss": 3.726582336425781, "grad_norm": 0.8727318048477173, "learning_rate": 0.0001, "elbo": 3.72658, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6156462585034013}
{"step": 27555, "loss": 3.6879959106445312, "grad_norm": 0.6973344087600708, "learning_rate": 0.0001, "elbo": Infinity, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6159394792399717}
{"step": 27560, "loss": 3.7146041870117186, "grad_norm": 0.6920155882835388, "learning_rate": 0.0001, "elbo": 3.7146, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6162326999765422}
{"step": 27565, "loss": 3.726224899291992, "grad_norm": 0.7239014506340027, "learning_rate": 0.0001, "elbo": 3.72623, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6165259207131129}
{"step": 27570, "loss": 3.7275848388671875, "grad_norm": 0.67856365442276, "learning_rate": 0.0001, "elbo": 3.72758, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6168191414496833}
{"step": 27575, "loss": 3.6904857635498045, "grad_norm": 0.6957922577857971, "learning_rate": 0.0001, "elbo": 3.69049, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6171123621862538}
{"step": 27580, "loss": 3.7124984741210936, "grad_norm": 0.7472424507141113, "learning_rate": 0.0001, "elbo": 3.7125, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6174055829228244}
{"step": 27585, "loss": 3.7417423248291017, "grad_norm": 0.7007704973220825, "learning_rate": 0.0001, "elbo": 3.74174, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.617698803659395}
{"step": 27590, "loss": 3.714279556274414, "grad_norm": 0.7486200928688049, "learning_rate": 0.0001, "elbo": 3.71428, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6179920243959653}
{"step": 27595, "loss": 3.7207889556884766, "grad_norm": 0.6681628823280334, "learning_rate": 0.0001, "elbo": 3.72079, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6182852451325358}
{"step": 27600, "loss": 3.7018089294433594, "grad_norm": 0.6847791075706482, "learning_rate": 0.0001, "elbo": 3.70181, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6185784658691063}
{"step": 27605, "loss": 3.7240352630615234, "grad_norm": 0.7829521894454956, "learning_rate": 0.0001, "elbo": 3.72404, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6188716866056767}
{"step": 27610, "loss": 3.72076301574707, "grad_norm": 0.7015843987464905, "learning_rate": 0.0001, "elbo": 3.72076, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6191649073422472}
{"step": 27615, "loss": 3.7254875183105467, "grad_norm": 0.9987555742263794, "learning_rate": 0.0001, "elbo": 3.72549, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6194581280788176}
{"step": 27620, "loss": 3.7340377807617187, "grad_norm": 0.6620383262634277, "learning_rate": 0.0001, "elbo": 3.73404, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.619751348815388}
{"step": 27625, "loss": 3.7223567962646484, "grad_norm": 0.6723323464393616, "learning_rate": 0.0001, "elbo": 3.72236, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6200445695519587}
{"step": 27630, "loss": 3.7264808654785155, "grad_norm": 0.7005978226661682, "learning_rate": 0.0001, "elbo": 3.72648, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6203377902885292}
{"step": 27635, "loss": 3.73267822265625, "grad_norm": 0.7493147253990173, "learning_rate": 0.0001, "elbo": 3.73268, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6206310110250997}
{"step": 27640, "loss": 3.7323844909667967, "grad_norm": 0.6620679497718811, "learning_rate": 0.0001, "elbo": 3.73238, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6209242317616703}
{"step": 27645, "loss": 3.7333454132080077, "grad_norm": 0.6503031253814697, "learning_rate": 0.0001, "elbo": 3.73335, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6212174524982408}
{"step": 27650, "loss": 3.7178916931152344, "grad_norm": 0.8829317092895508, "learning_rate": 0.0001, "elbo": 3.71789, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6215106732348112}
{"step": 27655, "loss": 3.7382747650146486, "grad_norm": 0.7862801551818848, "learning_rate": 0.0001, "elbo": 3.73827, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6218038939713817}
{"step": 27655, "loss": 3.7382747650146486, "grad_norm": 0.7862801551818848, "learning_rate": 0.0001, "elbo": 3.73827, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6218038939713817}
{"step": 27660, "loss": 3.73338623046875, "grad_norm": 0.7784960269927979, "learning_rate": 0.0001, "elbo": 3.73339, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6220971147079521}
{"step": 27665, "loss": 3.7433345794677733, "grad_norm": 1.2165096998214722, "learning_rate": 0.0001, "elbo": 3.74333, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6223903354445226}
{"step": 27670, "loss": 3.732073974609375, "grad_norm": 0.6810969710350037, "learning_rate": 0.0001, "elbo": 3.73207, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.622683556181093}
{"step": 27675, "loss": 3.742498016357422, "grad_norm": 0.6654646992683411, "learning_rate": 0.0001, "elbo": 3.7425, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6229767769176635}
{"step": 27680, "loss": 3.719279479980469, "grad_norm": 0.6775587201118469, "learning_rate": 0.0001, "elbo": 3.71928, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.623269997654234}
{"step": 27685, "loss": 3.706704330444336, "grad_norm": 0.7164517045021057, "learning_rate": 0.0001, "elbo": 3.7067, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6235632183908046}
{"step": 27690, "loss": 3.7283401489257812, "grad_norm": 0.645602822303772, "learning_rate": 0.0001, "elbo": 3.72834, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.623856439127375}
{"step": 27695, "loss": 3.7230682373046875, "grad_norm": 0.6643003821372986, "learning_rate": 0.0001, "elbo": 3.72307, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6241496598639455}
{"step": 27700, "loss": 3.726723480224609, "grad_norm": 0.6895679831504822, "learning_rate": 0.0001, "elbo": 3.72672, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6244428806005162}
{"step": 27705, "loss": 3.7395339965820313, "grad_norm": 0.7926459312438965, "learning_rate": 0.0001, "elbo": 3.73953, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6247361013370867}
{"step": 27710, "loss": 3.7219776153564452, "grad_norm": 1.1445541381835938, "learning_rate": 0.0001, "elbo": 3.72198, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6250293220736571}
{"step": 27715, "loss": 3.7176773071289064, "grad_norm": 0.7062707543373108, "learning_rate": 0.0001, "elbo": 3.71768, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6253225428102276}
{"step": 27720, "loss": 3.711670684814453, "grad_norm": 0.611431360244751, "learning_rate": 0.0001, "elbo": 3.71167, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.625615763546798}
{"step": 27725, "loss": 3.732339859008789, "grad_norm": 0.7354261875152588, "learning_rate": 0.0001, "elbo": 3.73234, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6259089842833685}
{"step": 27725, "loss": 3.732339859008789, "grad_norm": 0.7354261875152588, "learning_rate": 0.0001, "elbo": 3.73234, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6259089842833685}
{"step": 27730, "loss": 3.734915924072266, "grad_norm": 0.6769219040870667, "learning_rate": 0.0001, "elbo": 3.73492, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.626202205019939}
{"step": 27735, "loss": 3.717304229736328, "grad_norm": 0.6856417059898376, "learning_rate": 0.0001, "elbo": 3.7173, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6264954257565094}
{"step": 27740, "loss": 3.7314353942871095, "grad_norm": 0.6781442165374756, "learning_rate": 0.0001, "elbo": 3.73144, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6267886464930799}
{"step": 27745, "loss": 3.738935852050781, "grad_norm": 0.670728325843811, "learning_rate": 0.0001, "elbo": 3.73894, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6270818672296505}
{"step": 27750, "loss": 3.709162139892578, "grad_norm": 0.6797165870666504, "learning_rate": 0.0001, "elbo": 3.70916, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.627375087966221}
{"step": 27755, "loss": 3.7078018188476562, "grad_norm": 0.9777290225028992, "learning_rate": 0.0001, "elbo": 3.7078, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6276683087027914}
{"step": 27760, "loss": 3.7277099609375, "grad_norm": 0.7014021873474121, "learning_rate": 0.0001, "elbo": 3.72771, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.627961529439362}
{"step": 27765, "loss": 3.73323974609375, "grad_norm": 0.7569140791893005, "learning_rate": 0.0001, "elbo": 3.73324, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6282547501759326}
{"step": 27770, "loss": 3.71899299621582, "grad_norm": 0.7233777642250061, "learning_rate": 0.0001, "elbo": 3.71899, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.628547970912503}
{"step": 27775, "loss": 3.737590789794922, "grad_norm": 0.7413623332977295, "learning_rate": 0.0001, "elbo": 3.73759, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6288411916490735}
{"step": 27780, "loss": 3.7464935302734377, "grad_norm": 0.6866488456726074, "learning_rate": 0.0001, "elbo": 3.74649, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.629134412385644}
{"step": 27785, "loss": 3.7168373107910155, "grad_norm": 0.8327416777610779, "learning_rate": 0.0001, "elbo": 3.71684, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6294276331222144}
{"step": 27790, "loss": 3.7286331176757814, "grad_norm": 0.6936286687850952, "learning_rate": 0.0001, "elbo": 3.72863, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6297208538587848}
{"step": 27795, "loss": 3.7224571228027346, "grad_norm": 0.7362797856330872, "learning_rate": 0.0001, "elbo": 3.72246, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6300140745953553}
{"step": 27800, "loss": 3.728814697265625, "grad_norm": 0.7444758415222168, "learning_rate": 0.0001, "elbo": 3.72881, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6303072953319258}
{"step": 27805, "loss": 3.7178619384765623, "grad_norm": 0.6869563460350037, "learning_rate": 0.0001, "elbo": 3.71786, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6306005160684962}
{"step": 27810, "loss": 3.70709228515625, "grad_norm": 0.655046284198761, "learning_rate": 0.0001, "elbo": 3.70709, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6308937368050669}
{"step": 27815, "loss": 3.7240142822265625, "grad_norm": 0.6767092943191528, "learning_rate": 0.0001, "elbo": 3.72401, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6311869575416373}
{"step": 27820, "loss": 3.703211212158203, "grad_norm": 0.6601428985595703, "learning_rate": 0.0001, "elbo": 3.70321, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6314801782782078}
{"step": 27825, "loss": 3.7456298828125, "grad_norm": 0.7024436593055725, "learning_rate": 0.0001, "elbo": 3.74563, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6317733990147785}
{"step": 27830, "loss": 3.718551254272461, "grad_norm": 0.6475926637649536, "learning_rate": 0.0001, "elbo": 3.71855, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.632066619751349}
{"step": 27835, "loss": 3.7317649841308596, "grad_norm": 0.6507245898246765, "learning_rate": 0.0001, "elbo": 3.73176, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6323598404879194}
{"step": 27840, "loss": 3.718131256103516, "grad_norm": 0.6507453918457031, "learning_rate": 0.0001, "elbo": 3.71813, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6326530612244898}
{"step": 27845, "loss": 3.7205196380615235, "grad_norm": 0.7213929295539856, "learning_rate": 0.0001, "elbo": 3.72052, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6329462819610603}
{"step": 27850, "loss": 3.710882568359375, "grad_norm": 0.8176310658454895, "learning_rate": 0.0001, "elbo": 3.71088, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6332395026976307}
{"step": 27855, "loss": 3.710086441040039, "grad_norm": 0.8027513027191162, "learning_rate": 0.0001, "elbo": 3.71009, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6335327234342012}
{"step": 27860, "loss": 3.727825164794922, "grad_norm": 0.6678034067153931, "learning_rate": 0.0001, "elbo": 3.72783, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6338259441707716}
{"step": 27865, "loss": 3.7254043579101563, "grad_norm": 0.7126519083976746, "learning_rate": 0.0001, "elbo": 3.7254, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.634119164907342}
{"step": 27870, "loss": 3.7295085906982424, "grad_norm": 1.2225432395935059, "learning_rate": 0.0001, "elbo": 3.72951, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6344123856439128}
{"step": 27875, "loss": 3.714586639404297, "grad_norm": 0.7072376012802124, "learning_rate": 0.0001, "elbo": 3.71459, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6347056063804832}
{"step": 27880, "loss": 3.722899627685547, "grad_norm": 0.7195444107055664, "learning_rate": 0.0001, "elbo": 3.7229, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6349988271170537}
{"step": 27885, "loss": 3.7241806030273437, "grad_norm": 0.6975420713424683, "learning_rate": 0.0001, "elbo": 3.72418, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6352920478536244}
{"step": 27890, "loss": 3.72869873046875, "grad_norm": 0.7344136238098145, "learning_rate": 0.0001, "elbo": 3.7287, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6355852685901948}
{"step": 27895, "loss": 3.7100975036621096, "grad_norm": 0.7465073466300964, "learning_rate": 0.0001, "elbo": 3.7101, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6358784893267653}
{"step": 27895, "loss": 3.7100975036621096, "grad_norm": 0.7465073466300964, "learning_rate": 0.0001, "elbo": 3.7101, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6358784893267653}
{"step": 27900, "loss": 3.7052001953125, "grad_norm": 0.8901219964027405, "learning_rate": 0.0001, "elbo": 3.7052, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6361717100633357}
{"step": 27905, "loss": 3.7078506469726564, "grad_norm": 0.730937123298645, "learning_rate": 0.0001, "elbo": 3.70785, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6364649307999062}
{"step": 27910, "loss": 3.7201793670654295, "grad_norm": 0.637787401676178, "learning_rate": 0.0001, "elbo": 3.72018, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6367581515364766}
{"step": 27915, "loss": 3.7263420104980467, "grad_norm": 0.7516193985939026, "learning_rate": 0.0001, "elbo": 3.72634, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.637051372273047}
{"step": 27920, "loss": 3.739337158203125, "grad_norm": 0.716699481010437, "learning_rate": 0.0001, "elbo": 3.73934, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6373445930096175}
{"step": 27925, "loss": 3.717431640625, "grad_norm": 0.7348700165748596, "learning_rate": 0.0001, "elbo": 3.71743, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.637637813746188}
{"step": 27925, "loss": 3.717431640625, "grad_norm": 0.7348700165748596, "learning_rate": 0.0001, "elbo": 3.71743, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.637637813746188}
{"step": 27930, "loss": 3.7163925170898438, "grad_norm": 0.6980991959571838, "learning_rate": 0.0001, "elbo": 3.71639, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6379310344827587}
{"step": 27935, "loss": 3.711739730834961, "grad_norm": 0.734307050704956, "learning_rate": 0.0001, "elbo": 3.71174, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6382242552193291}
{"step": 27940, "loss": 3.734172821044922, "grad_norm": 0.6520305871963501, "learning_rate": 0.0001, "elbo": 3.73417, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6385174759558996}
{"step": 27945, "loss": 3.737749481201172, "grad_norm": 0.7961820960044861, "learning_rate": 0.0001, "elbo": 3.73775, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.63881069669247}
{"step": 27950, "loss": 3.7227134704589844, "grad_norm": 0.6523128151893616, "learning_rate": 0.0001, "elbo": 3.72271, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6391039174290407}
{"step": 27955, "loss": 3.7164386749267577, "grad_norm": 0.6462462544441223, "learning_rate": 0.0001, "elbo": 3.71644, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6393971381656112}
{"step": 27960, "loss": 3.7147972106933596, "grad_norm": 0.709744930267334, "learning_rate": 0.0001, "elbo": 3.7148, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6396903589021816}
{"step": 27965, "loss": 3.7139328002929686, "grad_norm": 0.6939998865127563, "learning_rate": 0.0001, "elbo": 3.71393, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.639983579638752}
{"step": 27970, "loss": 3.7116348266601564, "grad_norm": 0.6394999027252197, "learning_rate": 0.0001, "elbo": 3.71163, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6402768003753225}
{"step": 27975, "loss": 3.7209888458251954, "grad_norm": 0.7097294926643372, "learning_rate": 0.0001, "elbo": 3.72099, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.640570021111893}
{"step": 27980, "loss": 3.7136615753173827, "grad_norm": 0.6856293678283691, "learning_rate": 0.0001, "elbo": 3.71366, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6408632418484634}
{"step": 27985, "loss": 3.729070281982422, "grad_norm": 0.6989932656288147, "learning_rate": 0.0001, "elbo": 3.72907, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.641156462585034}
{"step": 27990, "loss": 3.7179527282714844, "grad_norm": 0.6674928069114685, "learning_rate": 0.0001, "elbo": 3.71795, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6414496833216043}
{"step": 27995, "loss": 3.7175323486328127, "grad_norm": 0.7187627553939819, "learning_rate": 0.0001, "elbo": 3.71753, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.641742904058175}
{"step": 28000, "loss": 3.718035125732422, "grad_norm": 0.6668481826782227, "learning_rate": 0.0001, "elbo": 3.71803, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6420361247947455}
{"step": 28005, "loss": 3.7287864685058594, "grad_norm": 0.699153482913971, "learning_rate": 0.0001, "elbo": 3.72879, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.642329345531316}
{"step": 28010, "loss": 3.712199020385742, "grad_norm": 0.6603887677192688, "learning_rate": 0.0001, "elbo": 3.7122, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6426225662678866}
{"step": 28015, "loss": 3.722555923461914, "grad_norm": 0.6449149250984192, "learning_rate": 0.0001, "elbo": 3.72256, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.642915787004457}
{"step": 28020, "loss": 3.721297836303711, "grad_norm": 0.6673051118850708, "learning_rate": 0.0001, "elbo": 3.7213, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6432090077410275}
{"step": 28025, "loss": 3.725650405883789, "grad_norm": 0.6594521403312683, "learning_rate": 0.0001, "elbo": 3.72565, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.643502228477598}
{"step": 28030, "loss": 3.721321868896484, "grad_norm": 0.672525942325592, "learning_rate": 0.0001, "elbo": 3.72132, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6437954492141684}
{"step": 28035, "loss": 3.7275135040283205, "grad_norm": 0.6922520995140076, "learning_rate": 0.0001, "elbo": 3.72751, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6440886699507389}
{"step": 28035, "loss": 3.7275135040283205, "grad_norm": 0.6922520995140076, "learning_rate": 0.0001, "elbo": 3.72751, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6440886699507389}
{"step": 28040, "loss": 3.716930389404297, "grad_norm": 0.7335612773895264, "learning_rate": 0.0001, "elbo": 3.71693, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6443818906873093}
{"step": 28045, "loss": 3.713053894042969, "grad_norm": 0.6873694658279419, "learning_rate": 0.0001, "elbo": 3.71305, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6446751114238798}
{"step": 28050, "loss": 3.711678314208984, "grad_norm": 0.6897056698799133, "learning_rate": 0.0001, "elbo": 3.71168, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6449683321604502}
{"step": 28055, "loss": 3.711989974975586, "grad_norm": 0.683297872543335, "learning_rate": 0.0001, "elbo": 3.71199, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.645261552897021}
{"step": 28060, "loss": 3.723419189453125, "grad_norm": 0.6724120378494263, "learning_rate": 0.0001, "elbo": 3.72342, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6455547736335914}
{"step": 28065, "loss": 3.720789337158203, "grad_norm": 0.6708964705467224, "learning_rate": 0.0001, "elbo": 3.72079, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6458479943701618}
{"step": 28070, "loss": 3.7216297149658204, "grad_norm": 0.6494807004928589, "learning_rate": 0.0001, "elbo": 3.72163, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6461412151067325}
{"step": 28075, "loss": 3.719966506958008, "grad_norm": 0.7169273495674133, "learning_rate": 0.0001, "elbo": 3.71997, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.646434435843303}
{"step": 28080, "loss": 3.6996490478515627, "grad_norm": 0.8281698822975159, "learning_rate": 0.0001, "elbo": 3.69965, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6467276565798734}
{"step": 28085, "loss": 3.713611602783203, "grad_norm": 0.8970959782600403, "learning_rate": 0.0001, "elbo": 3.71361, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6470208773164439}
{"step": 28090, "loss": 3.7013729095458983, "grad_norm": 1.060110330581665, "learning_rate": 0.0001, "elbo": 3.70137, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6473140980530143}
{"step": 28095, "loss": 3.7347373962402344, "grad_norm": 0.6824603080749512, "learning_rate": 0.0001, "elbo": 3.73474, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6476073187895848}
{"step": 28100, "loss": 3.7149070739746093, "grad_norm": 0.6912063360214233, "learning_rate": 0.0001, "elbo": 3.71491, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6479005395261552}
{"step": 28105, "loss": 3.7309219360351564, "grad_norm": 0.6885021924972534, "learning_rate": 0.0001, "elbo": 3.73092, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6481937602627257}
{"step": 28105, "loss": 3.7309219360351564, "grad_norm": 0.6885021924972534, "learning_rate": 0.0001, "elbo": 3.73092, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6481937602627257}
{"step": 28110, "loss": 3.7132110595703125, "grad_norm": 0.7240129709243774, "learning_rate": 0.0001, "elbo": 3.71321, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6484869809992961}
{"step": 28115, "loss": 3.7132972717285155, "grad_norm": 0.7364266514778137, "learning_rate": 0.0001, "elbo": 3.7133, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6487802017358668}
{"step": 28120, "loss": 3.732749176025391, "grad_norm": 0.851667582988739, "learning_rate": 0.0001, "elbo": 3.73275, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6490734224724373}
{"step": 28125, "loss": 3.7331401824951174, "grad_norm": 0.6705368757247925, "learning_rate": 0.0001, "elbo": 3.73314, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6493666432090077}
{"step": 28130, "loss": 3.7504478454589845, "grad_norm": 1.1986576318740845, "learning_rate": 0.0001, "elbo": 3.75045, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6496598639455784}
{"step": 28135, "loss": 3.7294342041015627, "grad_norm": 0.6679972410202026, "learning_rate": 0.0001, "elbo": 3.72943, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6499530846821489}
{"step": 28135, "loss": 3.7294342041015627, "grad_norm": 0.6679972410202026, "learning_rate": 0.0001, "elbo": 3.72943, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6499530846821489}
{"step": 28140, "loss": 3.7254600524902344, "grad_norm": 0.7012891173362732, "learning_rate": 0.0001, "elbo": 3.72546, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6502463054187193}
{"step": 28145, "loss": 3.7113616943359373, "grad_norm": 0.7064059376716614, "learning_rate": 0.0001, "elbo": 3.71136, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6505395261552898}
{"step": 28150, "loss": 3.7178489685058596, "grad_norm": 0.649890661239624, "learning_rate": 0.0001, "elbo": 3.71785, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6508327468918602}
{"step": 28150, "loss": 3.7178489685058596, "grad_norm": 0.649890661239624, "learning_rate": 0.0001, "elbo": 3.71785, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6508327468918602}
{"step": 28155, "loss": 3.7212982177734375, "grad_norm": 0.7917396426200867, "learning_rate": 0.0001, "elbo": 3.7213, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6511259676284307}
{"step": 28160, "loss": 3.706175994873047, "grad_norm": 0.675479531288147, "learning_rate": 0.0001, "elbo": 3.70618, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6514191883650011}
{"step": 28165, "loss": 3.721471405029297, "grad_norm": 0.7005313634872437, "learning_rate": 0.0001, "elbo": 3.72147, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6517124091015716}
{"step": 28170, "loss": 3.719401550292969, "grad_norm": 0.7733185291290283, "learning_rate": 0.0001, "elbo": 3.7194, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.652005629838142}
{"step": 28175, "loss": 3.7128746032714846, "grad_norm": 0.6973536014556885, "learning_rate": 0.0001, "elbo": 3.71287, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6522988505747125}
{"step": 28180, "loss": 3.7129150390625, "grad_norm": 0.7260884642601013, "learning_rate": 0.0001, "elbo": 3.71291, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6525920713112832}
{"step": 28185, "loss": 3.713687515258789, "grad_norm": 0.7674293518066406, "learning_rate": 0.0001, "elbo": 3.71369, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6528852920478536}
{"step": 28190, "loss": 3.721841812133789, "grad_norm": 0.7057193517684937, "learning_rate": 0.0001, "elbo": 3.72184, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.653178512784424}
{"step": 28195, "loss": 3.7069107055664063, "grad_norm": 0.6523973941802979, "learning_rate": 0.0001, "elbo": 3.70691, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6534717335209947}
{"step": 28200, "loss": 3.7126541137695312, "grad_norm": 0.6927397847175598, "learning_rate": 0.0001, "elbo": 3.71265, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6537649542575652}
{"step": 28205, "loss": 3.719731903076172, "grad_norm": 0.8536617755889893, "learning_rate": 0.0001, "elbo": 3.71973, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6540581749941357}
{"step": 28210, "loss": 3.7093658447265625, "grad_norm": 0.646193265914917, "learning_rate": 0.0001, "elbo": 3.70937, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.654351395730706}
{"step": 28215, "loss": 3.723603057861328, "grad_norm": 0.6576439738273621, "learning_rate": 0.0001, "elbo": 3.7236, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6546446164672766}
{"step": 28220, "loss": 3.7234657287597654, "grad_norm": 0.7193841338157654, "learning_rate": 0.0001, "elbo": 3.72347, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.654937837203847}
{"step": 28225, "loss": 3.709063720703125, "grad_norm": 0.858340859413147, "learning_rate": 0.0001, "elbo": 3.70906, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6552310579404175}
{"step": 28230, "loss": 3.725399398803711, "grad_norm": 0.8073087334632874, "learning_rate": 0.0001, "elbo": 3.7254, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.655524278676988}
{"step": 28235, "loss": 3.7187217712402343, "grad_norm": 0.6626763939857483, "learning_rate": 0.0001, "elbo": 3.71872, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6558174994135584}
{"step": 28240, "loss": 3.7157974243164062, "grad_norm": 0.676856279373169, "learning_rate": 0.0001, "elbo": 3.7158, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.656110720150129}
{"step": 28245, "loss": 3.727657699584961, "grad_norm": 0.7835683226585388, "learning_rate": 0.0001, "elbo": 3.72766, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6564039408866995}
{"step": 28250, "loss": 3.732605743408203, "grad_norm": 0.7061372995376587, "learning_rate": 0.0001, "elbo": 3.73261, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.65669716162327}
{"step": 28255, "loss": 3.7081745147705076, "grad_norm": 0.736245334148407, "learning_rate": 0.0001, "elbo": 3.70817, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6569903823598406}
{"step": 28260, "loss": 3.738140869140625, "grad_norm": 0.6426313519477844, "learning_rate": 0.0001, "elbo": 3.73814, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.657283603096411}
{"step": 28265, "loss": 3.7088546752929688, "grad_norm": 0.652678906917572, "learning_rate": 0.0001, "elbo": 3.70885, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6575768238329815}
{"step": 28270, "loss": 3.7162025451660154, "grad_norm": 0.7147685885429382, "learning_rate": 0.0001, "elbo": 3.7162, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.657870044569552}
{"step": 28275, "loss": 3.7142162322998047, "grad_norm": 0.7102339863777161, "learning_rate": 0.0001, "elbo": 3.71422, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6581632653061225}
{"step": 28280, "loss": 3.7240509033203124, "grad_norm": 0.8850192427635193, "learning_rate": 0.0001, "elbo": 3.72405, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.658456486042693}
{"step": 28285, "loss": 3.7190834045410157, "grad_norm": 0.6760454773902893, "learning_rate": 0.0001, "elbo": 3.71908, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6587497067792634}
{"step": 28290, "loss": 3.730510711669922, "grad_norm": 0.6575550436973572, "learning_rate": 0.0001, "elbo": 3.73051, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6590429275158338}
{"step": 28295, "loss": 3.726945495605469, "grad_norm": 0.6723503470420837, "learning_rate": 0.0001, "elbo": 3.72695, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6593361482524043}
{"step": 28300, "loss": 3.7290115356445312, "grad_norm": 0.7040747404098511, "learning_rate": 0.0001, "elbo": 3.72901, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.659629368988975}
{"step": 28300, "loss": 3.7290115356445312, "grad_norm": 0.7040747404098511, "learning_rate": 0.0001, "elbo": 3.72901, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.659629368988975}
{"step": 28305, "loss": 3.7261024475097657, "grad_norm": 0.7025541663169861, "learning_rate": 0.0001, "elbo": 3.7261, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6599225897255454}
{"step": 28310, "loss": 3.7150497436523438, "grad_norm": 0.6850268244743347, "learning_rate": 0.0001, "elbo": 3.71505, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6602158104621159}
{"step": 28315, "loss": 3.719071960449219, "grad_norm": 0.7297781109809875, "learning_rate": 0.0001, "elbo": 3.71907, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6605090311986865}
{"step": 28320, "loss": 3.701232147216797, "grad_norm": 0.6808542609214783, "learning_rate": 0.0001, "elbo": 3.70123, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.660802251935257}
{"step": 28325, "loss": 3.720253753662109, "grad_norm": 0.6740188598632812, "learning_rate": 0.0001, "elbo": 3.72025, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6610954726718274}
{"step": 28330, "loss": 3.7232776641845704, "grad_norm": 0.6893044710159302, "learning_rate": 0.0001, "elbo": 3.72328, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.661388693408398}
{"step": 28330, "loss": 3.7232776641845704, "grad_norm": 0.6893044710159302, "learning_rate": 0.0001, "elbo": 3.72328, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.661388693408398}
{"step": 28335, "loss": 3.729258728027344, "grad_norm": 0.7500540614128113, "learning_rate": 0.0001, "elbo": 3.72926, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6616819141449684}
{"step": 28340, "loss": 3.713764190673828, "grad_norm": 0.690940260887146, "learning_rate": 0.0001, "elbo": 3.71376, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6619751348815388}
{"step": 28345, "loss": 3.7365989685058594, "grad_norm": 0.6917783617973328, "learning_rate": 0.0001, "elbo": 3.7366, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6622683556181093}
{"step": 28350, "loss": 3.721462631225586, "grad_norm": 0.8597869873046875, "learning_rate": 0.0001, "elbo": 3.72146, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6625615763546797}
{"step": 28355, "loss": 3.7299007415771483, "grad_norm": 1.1322237253189087, "learning_rate": 0.0001, "elbo": 3.7299, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6628547970912502}
{"step": 28360, "loss": 3.712504577636719, "grad_norm": 0.7572932839393616, "learning_rate": 0.0001, "elbo": 3.7125, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6631480178278206}
{"step": 28365, "loss": 3.7171215057373046, "grad_norm": 0.70353102684021, "learning_rate": 0.0001, "elbo": 3.71712, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6634412385643913}
{"step": 28370, "loss": 3.7281593322753905, "grad_norm": 0.6625508666038513, "learning_rate": 0.0001, "elbo": 3.72816, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6637344593009618}
{"step": 28375, "loss": 3.7336944580078124, "grad_norm": 0.7565659284591675, "learning_rate": 0.0001, "elbo": 3.73369, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6640276800375322}
{"step": 28380, "loss": 3.721430206298828, "grad_norm": 0.6500460505485535, "learning_rate": 0.0001, "elbo": 3.72143, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6643209007741029}
{"step": 28385, "loss": 3.71258430480957, "grad_norm": 0.7001456022262573, "learning_rate": 0.0001, "elbo": 3.71258, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6646141215106733}
{"step": 28390, "loss": 3.716958999633789, "grad_norm": 1.038726806640625, "learning_rate": 0.0001, "elbo": 3.71696, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6649073422472438}
{"step": 28395, "loss": 3.726390075683594, "grad_norm": 0.6871064901351929, "learning_rate": 0.0001, "elbo": 3.72639, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6652005629838142}
{"step": 28400, "loss": 3.719084548950195, "grad_norm": 0.784277081489563, "learning_rate": 0.0001, "elbo": 3.71908, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6654937837203847}
{"step": 28405, "loss": 3.7409431457519533, "grad_norm": 0.6631690263748169, "learning_rate": 0.0001, "elbo": 3.74094, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6657870044569552}
{"step": 28410, "loss": 3.7245086669921874, "grad_norm": 0.6853365302085876, "learning_rate": 0.0001, "elbo": 3.72451, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6660802251935256}
{"step": 28415, "loss": 3.7243572235107423, "grad_norm": 0.6814175844192505, "learning_rate": 0.0001, "elbo": 3.72436, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.666373445930096}
{"step": 28420, "loss": 3.7127880096435546, "grad_norm": 0.6811280846595764, "learning_rate": 0.0001, "elbo": 3.71279, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6666666666666665}
{"step": 28425, "loss": 3.7146324157714843, "grad_norm": 0.6287779808044434, "learning_rate": 0.0001, "elbo": 3.71463, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6669598874032372}
{"step": 28425, "loss": 3.7146324157714843, "grad_norm": 0.6287779808044434, "learning_rate": 0.0001, "elbo": 3.71463, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6669598874032372}
{"step": 28430, "loss": 3.7218589782714844, "grad_norm": 0.8177543878555298, "learning_rate": 0.0001, "elbo": 3.72186, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6672531081398076}
{"step": 28435, "loss": 3.7105857849121096, "grad_norm": 0.6966089606285095, "learning_rate": 0.0001, "elbo": 3.71059, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.667546328876378}
{"step": 28440, "loss": 3.7244285583496093, "grad_norm": 0.6867913007736206, "learning_rate": 0.0001, "elbo": 3.72443, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6678395496129488}
{"step": 28445, "loss": 3.7193126678466797, "grad_norm": 0.6911938786506653, "learning_rate": 0.0001, "elbo": 3.71931, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6681327703495192}
{"step": 28450, "loss": 3.712656021118164, "grad_norm": 0.6792138814926147, "learning_rate": 0.0001, "elbo": 3.71266, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6684259910860897}
{"step": 28455, "loss": 3.7110572814941407, "grad_norm": 0.6601086854934692, "learning_rate": 0.0001, "elbo": 3.71106, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6687192118226601}
{"step": 28460, "loss": 3.710816192626953, "grad_norm": 0.6996287703514099, "learning_rate": 0.0001, "elbo": 3.71082, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6690124325592306}
{"step": 28465, "loss": 3.7152126312255858, "grad_norm": 0.6486495733261108, "learning_rate": 0.0001, "elbo": 3.71521, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.669305653295801}
{"step": 28470, "loss": 3.712438201904297, "grad_norm": 0.6582595705986023, "learning_rate": 0.0001, "elbo": 3.71244, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6695988740323715}
{"step": 28475, "loss": 3.6995819091796873, "grad_norm": 0.7142784595489502, "learning_rate": 0.0001, "elbo": 3.69958, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.669892094768942}
{"step": 28480, "loss": 3.7204833984375, "grad_norm": 0.8973924517631531, "learning_rate": 0.0001, "elbo": 3.72048, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6701853155055124}
{"step": 28485, "loss": 3.716314697265625, "grad_norm": 0.7177401185035706, "learning_rate": 0.0001, "elbo": 3.71631, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.670478536242083}
{"step": 28490, "loss": 3.7257049560546873, "grad_norm": 0.671062707901001, "learning_rate": 0.0001, "elbo": 3.72571, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6707717569786535}
{"step": 28495, "loss": 3.7153388977050783, "grad_norm": 0.6782640218734741, "learning_rate": 0.0001, "elbo": 3.71534, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.671064977715224}
{"step": 28500, "loss": 3.723349761962891, "grad_norm": 0.7399455904960632, "learning_rate": 0.0001, "elbo": 3.72335, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6713581984517947}
{"step": 28505, "loss": 3.7268836975097654, "grad_norm": 0.7460818290710449, "learning_rate": 0.0001, "elbo": 3.72688, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6716514191883651}
{"step": 28510, "loss": 3.7283416748046876, "grad_norm": 0.7273733019828796, "learning_rate": 0.0001, "elbo": 3.72834, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6719446399249356}
{"step": 28515, "loss": 3.712881851196289, "grad_norm": 1.0687336921691895, "learning_rate": 0.0001, "elbo": 3.71288, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.672237860661506}
{"step": 28520, "loss": 3.727337646484375, "grad_norm": 0.7030099034309387, "learning_rate": 0.0001, "elbo": 3.72734, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6725310813980765}
{"step": 28525, "loss": 3.708260726928711, "grad_norm": 0.6715447902679443, "learning_rate": 0.0001, "elbo": 3.70826, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.672824302134647}
{"step": 28530, "loss": 3.7264331817626952, "grad_norm": 0.9935474991798401, "learning_rate": 0.0001, "elbo": 3.72643, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6731175228712174}
{"step": 28535, "loss": 3.722626876831055, "grad_norm": 0.620182991027832, "learning_rate": 0.0001, "elbo": 3.72263, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6734107436077879}
{"step": 28540, "loss": 3.7223583221435548, "grad_norm": 0.6487663984298706, "learning_rate": 0.0001, "elbo": 3.72236, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6737039643443583}
{"step": 28545, "loss": 3.729230499267578, "grad_norm": 0.7365342974662781, "learning_rate": 0.0001, "elbo": 3.72923, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6739971850809288}
{"step": 28550, "loss": 3.726426696777344, "grad_norm": 0.7081475257873535, "learning_rate": 0.0001, "elbo": 3.72643, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6742904058174994}
{"step": 28555, "loss": 3.7326446533203126, "grad_norm": 0.9045817255973816, "learning_rate": 0.0001, "elbo": 3.73264, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.67458362655407}
{"step": 28555, "loss": 3.7326446533203126, "grad_norm": 0.9045817255973816, "learning_rate": 0.0001, "elbo": 3.73264, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.67458362655407}
{"step": 28560, "loss": 3.722180938720703, "grad_norm": 0.6487771272659302, "learning_rate": 0.0001, "elbo": 3.72218, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6748768472906403}
{"step": 28565, "loss": 3.7114124298095703, "grad_norm": 0.6844850182533264, "learning_rate": 0.0001, "elbo": 3.71141, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.675170068027211}
{"step": 28570, "loss": 3.723377990722656, "grad_norm": 0.6919759511947632, "learning_rate": 0.0001, "elbo": 3.72338, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6754632887637815}
{"step": 28575, "loss": 3.7188358306884766, "grad_norm": 0.6903670430183411, "learning_rate": 0.0001, "elbo": 3.71884, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.675756509500352}
{"step": 28580, "loss": 3.7191295623779297, "grad_norm": 0.7288510203361511, "learning_rate": 0.0001, "elbo": 3.71913, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6760497302369224}
{"step": 28585, "loss": 3.7003883361816405, "grad_norm": 0.7237471342086792, "learning_rate": 0.0001, "elbo": 3.70039, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6763429509734928}
{"step": 28590, "loss": 3.7081512451171874, "grad_norm": 0.795434296131134, "learning_rate": 0.0001, "elbo": 3.70815, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6766361717100633}
{"step": 28595, "loss": 3.710683059692383, "grad_norm": 0.675687313079834, "learning_rate": 0.0001, "elbo": 3.71068, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6769293924466337}
{"step": 28600, "loss": 3.691167449951172, "grad_norm": 0.6756069660186768, "learning_rate": 0.0001, "elbo": 3.69117, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6772226131832042}
{"step": 28605, "loss": 3.715036392211914, "grad_norm": 1.1776442527770996, "learning_rate": 0.0001, "elbo": 3.71504, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6775158339197747}
{"step": 28610, "loss": 3.720708465576172, "grad_norm": 0.6791386008262634, "learning_rate": 0.0001, "elbo": 3.72071, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6778090546563453}
{"step": 28615, "loss": 3.7301612854003907, "grad_norm": 0.6663962006568909, "learning_rate": 0.0001, "elbo": 3.73016, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6781022753929158}
{"step": 28620, "loss": 3.700722503662109, "grad_norm": 0.6448207497596741, "learning_rate": 0.0001, "elbo": 3.70072, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6783954961294862}
{"step": 28625, "loss": 3.704193115234375, "grad_norm": 0.8081250190734863, "learning_rate": 0.0001, "elbo": 3.70419, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.678688716866057}
{"step": 28630, "loss": 3.714141845703125, "grad_norm": 0.7194804549217224, "learning_rate": 0.0001, "elbo": 3.71414, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6789819376026274}
{"step": 28635, "loss": 3.701211929321289, "grad_norm": 0.6239035725593567, "learning_rate": 0.0001, "elbo": 3.70121, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6792751583391978}
{"step": 28640, "loss": 3.7163291931152345, "grad_norm": 0.925189197063446, "learning_rate": 0.0001, "elbo": 3.71633, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6795683790757683}
{"step": 28645, "loss": 3.7062957763671873, "grad_norm": 0.7779224514961243, "learning_rate": 0.0001, "elbo": 3.7063, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6798615998123387}
{"step": 28650, "loss": 3.7072715759277344, "grad_norm": 0.6348909139633179, "learning_rate": 0.0001, "elbo": 3.70727, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6801548205489092}
{"step": 28655, "loss": 3.7341983795166014, "grad_norm": 0.6575038433074951, "learning_rate": 0.0001, "elbo": 3.7342, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6804480412854796}
{"step": 28660, "loss": 3.7104873657226562, "grad_norm": 1.1045564413070679, "learning_rate": 0.0001, "elbo": 3.71049, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.68074126202205}
{"step": 28665, "loss": 3.704591751098633, "grad_norm": 0.6829663515090942, "learning_rate": 0.0001, "elbo": 3.70459, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6810344827586206}
{"step": 28670, "loss": 3.694457244873047, "grad_norm": 0.6715373992919922, "learning_rate": 0.0001, "elbo": 3.69446, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6813277034951912}
{"step": 28675, "loss": 3.740007781982422, "grad_norm": 0.6758495569229126, "learning_rate": 0.0001, "elbo": 3.74001, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6816209242317617}
{"step": 28680, "loss": 3.7248573303222656, "grad_norm": 0.6839058995246887, "learning_rate": 0.0001, "elbo": 3.72486, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6819141449683321}
{"step": 28685, "loss": 3.711212158203125, "grad_norm": 0.713364839553833, "learning_rate": 0.0001, "elbo": 3.71121, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6822073657049028}
{"step": 28690, "loss": 3.720951461791992, "grad_norm": 0.6626119017601013, "learning_rate": 0.0001, "elbo": 3.72095, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6825005864414733}
{"step": 28695, "loss": 3.7109703063964843, "grad_norm": 0.6558605432510376, "learning_rate": 0.0001, "elbo": 3.71097, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6827938071780437}
{"step": 28700, "loss": 3.7104888916015626, "grad_norm": 0.7348535656929016, "learning_rate": 0.0001, "elbo": 3.71049, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6830870279146142}
{"step": 28705, "loss": 3.700579833984375, "grad_norm": 0.785695493221283, "learning_rate": 0.0001, "elbo": 3.70058, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6833802486511846}
{"step": 28710, "loss": 3.7248363494873047, "grad_norm": 0.6846938133239746, "learning_rate": 0.0001, "elbo": 3.72484, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.683673469387755}
{"step": 28715, "loss": 3.7139678955078126, "grad_norm": 0.6897574663162231, "learning_rate": 0.0001, "elbo": 3.71397, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6839666901243255}
{"step": 28720, "loss": 3.717301940917969, "grad_norm": 0.6949324011802673, "learning_rate": 0.0001, "elbo": 3.7173, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.684259910860896}
{"step": 28725, "loss": 3.7155189514160156, "grad_norm": 0.7398678064346313, "learning_rate": 0.0001, "elbo": 3.71552, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6845531315974664}
{"step": 28730, "loss": 3.7021324157714846, "grad_norm": 0.8852208852767944, "learning_rate": 0.0001, "elbo": 3.70213, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.684846352334037}
{"step": 28735, "loss": 3.719085693359375, "grad_norm": 0.687626838684082, "learning_rate": 0.0001, "elbo": 3.71909, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6851395730706076}
{"step": 28740, "loss": 3.7210262298583983, "grad_norm": 0.7240329384803772, "learning_rate": 0.0001, "elbo": 3.72103, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.685432793807178}
{"step": 28745, "loss": 3.716037368774414, "grad_norm": 0.7025120854377747, "learning_rate": 0.0001, "elbo": 3.71604, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6857260145437485}
{"step": 28750, "loss": 3.710493469238281, "grad_norm": 0.6087895035743713, "learning_rate": 0.0001, "elbo": 3.71049, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6860192352803192}
{"step": 28755, "loss": 3.723153305053711, "grad_norm": 0.859032928943634, "learning_rate": 0.0001, "elbo": 3.72315, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6863124560168896}
{"step": 28760, "loss": 3.711996841430664, "grad_norm": 0.6903567314147949, "learning_rate": 0.0001, "elbo": 3.712, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.68660567675346}
{"step": 28765, "loss": 3.7136604309082033, "grad_norm": 0.7235659956932068, "learning_rate": 0.0001, "elbo": 3.71366, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6868988974900305}
{"step": 28770, "loss": 3.7147998809814453, "grad_norm": 0.7342115044593811, "learning_rate": 0.0001, "elbo": 3.7148, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.687192118226601}
{"step": 28775, "loss": 3.724468994140625, "grad_norm": 0.6883366703987122, "learning_rate": 0.0001, "elbo": 3.72447, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6874853389631714}
{"step": 28780, "loss": 3.7220222473144533, "grad_norm": 0.6911198496818542, "learning_rate": 0.0001, "elbo": 3.72202, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6877785596997419}
{"step": 28785, "loss": 3.7174896240234374, "grad_norm": 0.6715203523635864, "learning_rate": 0.0001, "elbo": 3.71749, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6880717804363123}
{"step": 28790, "loss": 3.7045692443847655, "grad_norm": 0.8912349343299866, "learning_rate": 0.0001, "elbo": 3.70457, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6883650011728828}
{"step": 28795, "loss": 3.7013504028320314, "grad_norm": 0.7915816903114319, "learning_rate": 0.0001, "elbo": 3.70135, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6886582219094535}
{"step": 28800, "loss": 3.7270175933837892, "grad_norm": 0.6581052541732788, "learning_rate": 0.0001, "elbo": 3.72702, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.688951442646024}
{"step": 28805, "loss": 3.7199913024902345, "grad_norm": 0.6348521709442139, "learning_rate": 0.0001, "elbo": 3.71999, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6892446633825944}
{"step": 28810, "loss": 3.7135154724121096, "grad_norm": 0.7657962441444397, "learning_rate": 0.0001, "elbo": 3.71352, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.689537884119165}
{"step": 28815, "loss": 3.728596496582031, "grad_norm": 0.6819223165512085, "learning_rate": 0.0001, "elbo": 3.7286, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6898311048557355}
{"step": 28820, "loss": 3.7248531341552735, "grad_norm": 0.7182564735412598, "learning_rate": 0.0001, "elbo": 3.72485, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.690124325592306}
{"step": 28825, "loss": 3.7108863830566405, "grad_norm": 0.6332383751869202, "learning_rate": 0.0001, "elbo": 3.71089, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6904175463288764}
{"step": 28830, "loss": 3.708736038208008, "grad_norm": 0.6923403739929199, "learning_rate": 0.0001, "elbo": 3.70874, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6907107670654469}
{"step": 28835, "loss": 3.720688247680664, "grad_norm": 0.746069073677063, "learning_rate": 0.0001, "elbo": 3.72069, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6910039878020173}
{"step": 28840, "loss": 3.7235755920410156, "grad_norm": 0.7246693968772888, "learning_rate": 0.0001, "elbo": 3.72358, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6912972085385878}
{"step": 28845, "loss": 3.7325916290283203, "grad_norm": 0.6446825265884399, "learning_rate": 0.0001, "elbo": 3.73259, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6915904292751582}
{"step": 28850, "loss": 3.7075851440429686, "grad_norm": 0.6272248029708862, "learning_rate": 0.0001, "elbo": 3.70759, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6918836500117287}
{"step": 28855, "loss": 3.7202468872070313, "grad_norm": 0.6604312658309937, "learning_rate": 0.0001, "elbo": 3.72025, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6921768707482994}
{"step": 28860, "loss": 3.707135009765625, "grad_norm": 0.6510761976242065, "learning_rate": 0.0001, "elbo": 3.70713, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6924700914848698}
{"step": 28865, "loss": 3.7126312255859375, "grad_norm": 0.6686522960662842, "learning_rate": 0.0001, "elbo": 3.71263, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6927633122214403}
{"step": 28870, "loss": 3.7116249084472654, "grad_norm": 0.8001309037208557, "learning_rate": 0.0001, "elbo": 3.71162, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.693056532958011}
{"step": 28875, "loss": 3.7006546020507813, "grad_norm": 0.762028157711029, "learning_rate": 0.0001, "elbo": 3.70065, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6933497536945814}
{"step": 28880, "loss": 3.7121185302734374, "grad_norm": 0.6486324071884155, "learning_rate": 0.0001, "elbo": 3.71212, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6936429744311519}
{"step": 28885, "loss": 3.704965591430664, "grad_norm": 0.6217193007469177, "learning_rate": 0.0001, "elbo": 3.70497, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6939361951677223}
{"step": 28890, "loss": 3.718889617919922, "grad_norm": 0.6596394777297974, "learning_rate": 0.0001, "elbo": 3.71889, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6942294159042928}
{"step": 28895, "loss": 3.72308235168457, "grad_norm": 0.7174496054649353, "learning_rate": 0.0001, "elbo": 3.72308, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6945226366408632}
{"step": 28900, "loss": 3.699695587158203, "grad_norm": 0.6207365989685059, "learning_rate": 0.0001, "elbo": 3.6997, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6948158573774337}
{"step": 28905, "loss": 3.7087417602539063, "grad_norm": 0.6935524940490723, "learning_rate": 0.0001, "elbo": 3.70874, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6951090781140041}
{"step": 28910, "loss": 3.713296890258789, "grad_norm": 0.6914047002792358, "learning_rate": 0.0001, "elbo": 3.7133, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6954022988505746}
{"step": 28915, "loss": 3.702175521850586, "grad_norm": 0.7000947594642639, "learning_rate": 0.0001, "elbo": 3.70218, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.695695519587145}
{"step": 28920, "loss": 3.703420639038086, "grad_norm": 0.9081677794456482, "learning_rate": 0.0001, "elbo": 3.70342, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6959887403237157}
{"step": 28925, "loss": 3.7089820861816407, "grad_norm": 0.6526609063148499, "learning_rate": 0.0001, "elbo": 3.70898, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6962819610602862}
{"step": 28930, "loss": 3.7218727111816405, "grad_norm": 0.707528293132782, "learning_rate": 0.0001, "elbo": 3.72187, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6965751817968566}
{"step": 28935, "loss": 3.7247238159179688, "grad_norm": 0.690118134021759, "learning_rate": 0.0001, "elbo": 3.72472, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6968684025334273}
{"step": 28940, "loss": 3.708258056640625, "grad_norm": 0.7298130393028259, "learning_rate": 0.0001, "elbo": 3.70826, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6971616232699978}
{"step": 28945, "loss": 3.706850051879883, "grad_norm": 0.7672458291053772, "learning_rate": 0.0001, "elbo": 3.70685, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6974548440065682}
{"step": 28950, "loss": 3.722148895263672, "grad_norm": 0.6487270593643188, "learning_rate": 0.0001, "elbo": 3.72215, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6977480647431387}
{"step": 28955, "loss": 3.710359573364258, "grad_norm": 0.6749352216720581, "learning_rate": 0.0001, "elbo": 3.71036, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6980412854797091}
{"step": 28960, "loss": 3.720500946044922, "grad_norm": 0.8388882279396057, "learning_rate": 0.0001, "elbo": 3.7205, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6983345062162796}
{"step": 28965, "loss": 3.7062522888183596, "grad_norm": 0.7047092318534851, "learning_rate": 0.0001, "elbo": 3.70625, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.69862772695285}
{"step": 28970, "loss": 3.7125308990478514, "grad_norm": 0.8077977895736694, "learning_rate": 0.0001, "elbo": 3.71253, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6989209476894205}
{"step": 28975, "loss": 3.705893707275391, "grad_norm": 0.6943880319595337, "learning_rate": 0.0001, "elbo": 3.70589, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.699214168425991}
{"step": 28980, "loss": 3.705525207519531, "grad_norm": 0.7397740483283997, "learning_rate": 0.0001, "elbo": 3.70553, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.6995073891625616}
{"step": 28985, "loss": 3.721630859375, "grad_norm": 0.7226142883300781, "learning_rate": 0.0001, "elbo": 3.72163, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.699800609899132}
{"step": 28990, "loss": 3.702527618408203, "grad_norm": 0.805862545967102, "learning_rate": 0.0001, "elbo": 3.70253, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7000938306357025}
{"step": 28995, "loss": 3.717687225341797, "grad_norm": 0.6798940300941467, "learning_rate": 0.0001, "elbo": 3.71769, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7003870513722732}
{"step": 29000, "loss": 3.7271900177001953, "grad_norm": 0.6604819893836975, "learning_rate": 0.0001, "elbo": 3.72719, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7006802721088436}
{"step": 29005, "loss": 3.7243927001953123, "grad_norm": 1.218514084815979, "learning_rate": 0.0001, "elbo": 3.72439, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.700973492845414}
{"step": 29010, "loss": 3.716259765625, "grad_norm": 0.6994102001190186, "learning_rate": 0.0001, "elbo": 3.71626, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7012667135819846}
{"step": 29015, "loss": 3.7211151123046875, "grad_norm": 0.6683446168899536, "learning_rate": 0.0001, "elbo": 3.72112, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.701559934318555}
{"step": 29020, "loss": 3.7124229431152345, "grad_norm": 0.6632385849952698, "learning_rate": 0.0001, "elbo": 3.71242, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7018531550551255}
{"step": 29025, "loss": 3.7112030029296874, "grad_norm": 0.95188969373703, "learning_rate": 0.0001, "elbo": 3.7112, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.702146375791696}
{"step": 29030, "loss": 3.7170753479003906, "grad_norm": 0.642632782459259, "learning_rate": 0.0001, "elbo": 3.71708, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7024395965282664}
{"step": 29035, "loss": 3.7122398376464845, "grad_norm": 0.8471327424049377, "learning_rate": 0.0001, "elbo": 3.71224, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7027328172648368}
{"step": 29040, "loss": 3.699956512451172, "grad_norm": 0.6932574510574341, "learning_rate": 0.0001, "elbo": 3.69996, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7030260380014075}
{"step": 29045, "loss": 3.692920684814453, "grad_norm": 0.6768075227737427, "learning_rate": 0.0001, "elbo": 3.69292, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.703319258737978}
{"step": 29050, "loss": 3.705316162109375, "grad_norm": 0.7366780042648315, "learning_rate": 0.0001, "elbo": 3.70532, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7036124794745484}
{"step": 29055, "loss": 3.711739349365234, "grad_norm": 0.7180767059326172, "learning_rate": 0.0001, "elbo": 3.71174, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.703905700211119}
{"step": 29060, "loss": 3.7009742736816404, "grad_norm": 0.715117871761322, "learning_rate": 0.0001, "elbo": 3.70097, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7041989209476895}
{"step": 29065, "loss": 3.7072063446044923, "grad_norm": 0.6727598905563354, "learning_rate": 0.0001, "elbo": 3.70721, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.70449214168426}
{"step": 29070, "loss": 3.7152999877929687, "grad_norm": 1.1232643127441406, "learning_rate": 0.0001, "elbo": 3.7153, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7047853624208305}
{"step": 29075, "loss": 3.720972442626953, "grad_norm": 0.684424877166748, "learning_rate": 0.0001, "elbo": 3.72097, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.705078583157401}
{"step": 29080, "loss": 3.699610137939453, "grad_norm": 0.6872892379760742, "learning_rate": 0.0001, "elbo": 3.69961, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7053718038939714}
{"step": 29085, "loss": 3.705732727050781, "grad_norm": 0.6532527208328247, "learning_rate": 0.0001, "elbo": 3.70573, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7056650246305418}
{"step": 29090, "loss": 3.730523681640625, "grad_norm": 0.8409836888313293, "learning_rate": 0.0001, "elbo": 3.73052, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7059582453671123}
{"step": 29095, "loss": 3.701060485839844, "grad_norm": 0.9118713736534119, "learning_rate": 0.0001, "elbo": 3.70106, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7062514661036827}
{"step": 29100, "loss": 3.707957458496094, "grad_norm": 0.9120551943778992, "learning_rate": 0.0001, "elbo": 3.70796, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7065446868402532}
{"step": 29105, "loss": 3.7241825103759765, "grad_norm": 0.7425019145011902, "learning_rate": 0.0001, "elbo": 3.72418, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7068379075768239}
{"step": 29110, "loss": 3.710865783691406, "grad_norm": 0.7567579746246338, "learning_rate": 0.0001, "elbo": 3.71087, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7071311283133943}
{"step": 29115, "loss": 3.6935195922851562, "grad_norm": 0.6655123233795166, "learning_rate": 0.0001, "elbo": 3.69352, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7074243490499648}
{"step": 29120, "loss": 3.7146839141845702, "grad_norm": 0.6355998516082764, "learning_rate": 0.0001, "elbo": 3.71468, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7077175697865354}
{"step": 29125, "loss": 3.7169342041015625, "grad_norm": 0.6396893858909607, "learning_rate": 0.0001, "elbo": 3.71693, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.708010790523106}
{"step": 29130, "loss": 3.7067699432373047, "grad_norm": 0.6734766960144043, "learning_rate": 0.0001, "elbo": 3.70677, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7083040112596763}
{"step": 29135, "loss": 3.7223434448242188, "grad_norm": 0.7675030827522278, "learning_rate": 0.0001, "elbo": 3.72234, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7085972319962468}
{"step": 29140, "loss": 3.7230155944824217, "grad_norm": 0.7726607322692871, "learning_rate": 0.0001, "elbo": 3.72302, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7088904527328173}
{"step": 29145, "loss": 3.7207462310791017, "grad_norm": 0.6816572546958923, "learning_rate": 0.0001, "elbo": 3.72075, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7091836734693877}
{"step": 29150, "loss": 3.6917457580566406, "grad_norm": 0.7847980260848999, "learning_rate": 0.0001, "elbo": 3.69175, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7094768942059582}
{"step": 29155, "loss": 3.690566635131836, "grad_norm": 0.663285493850708, "learning_rate": 0.0001, "elbo": 3.69057, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7097701149425286}
{"step": 29160, "loss": 3.7170379638671873, "grad_norm": 0.8426905274391174, "learning_rate": 0.0001, "elbo": 3.71704, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.710063335679099}
{"step": 29165, "loss": 3.7221546173095703, "grad_norm": 0.9644400477409363, "learning_rate": 0.0001, "elbo": 3.72215, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7103565564156697}
{"step": 29170, "loss": 3.712847900390625, "grad_norm": 0.704364001750946, "learning_rate": 0.0001, "elbo": 3.71285, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7106497771522402}
{"step": 29170, "loss": 3.712847900390625, "grad_norm": 0.704364001750946, "learning_rate": 0.0001, "elbo": 3.71285, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7106497771522402}
{"step": 29175, "loss": 3.7048553466796874, "grad_norm": 0.6407312750816345, "learning_rate": 0.0001, "elbo": 3.70486, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7109429978888107}
{"step": 29180, "loss": 3.7204261779785157, "grad_norm": 0.8164461255073547, "learning_rate": 0.0001, "elbo": 3.72043, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7112362186253813}
{"step": 29185, "loss": 3.726274108886719, "grad_norm": 0.8479381799697876, "learning_rate": 0.0001, "elbo": 3.72627, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7115294393619518}
{"step": 29190, "loss": 3.714748764038086, "grad_norm": 0.7079842686653137, "learning_rate": 0.0001, "elbo": 3.71475, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7118226600985222}
{"step": 29195, "loss": 3.7083419799804687, "grad_norm": 0.7807198166847229, "learning_rate": 0.0001, "elbo": 3.70834, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7121158808350927}
{"step": 29200, "loss": 3.7245254516601562, "grad_norm": 1.097277283668518, "learning_rate": 0.0001, "elbo": 3.72453, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7124091015716632}
{"step": 29205, "loss": 3.712687683105469, "grad_norm": 0.7530267238616943, "learning_rate": 0.0001, "elbo": 3.71269, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7127023223082336}
{"step": 29210, "loss": 3.7347740173339843, "grad_norm": 0.6729636788368225, "learning_rate": 0.0001, "elbo": 3.73477, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.712995543044804}
{"step": 29215, "loss": 3.71093635559082, "grad_norm": 0.6478157043457031, "learning_rate": 0.0001, "elbo": 3.71094, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7132887637813745}
{"step": 29220, "loss": 3.6965072631835936, "grad_norm": 0.7465466856956482, "learning_rate": 0.0001, "elbo": 3.69651, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.713581984517945}
{"step": 29225, "loss": 3.6960479736328127, "grad_norm": 0.9375573992729187, "learning_rate": 0.0001, "elbo": 3.69605, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7138752052545156}
{"step": 29225, "loss": 3.6960479736328127, "grad_norm": 0.9375573992729187, "learning_rate": 0.0001, "elbo": 3.69605, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7138752052545156}
{"step": 29230, "loss": 3.711908721923828, "grad_norm": 0.7587292790412903, "learning_rate": 0.0001, "elbo": 3.71191, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.714168425991086}
{"step": 29235, "loss": 3.7018150329589843, "grad_norm": 0.9554378390312195, "learning_rate": 0.0001, "elbo": 3.70181, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7144616467276566}
{"step": 29240, "loss": 3.7027923583984377, "grad_norm": 0.7833576798439026, "learning_rate": 0.0001, "elbo": 3.70279, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7147548674642272}
{"step": 29245, "loss": 3.6995277404785156, "grad_norm": 0.6680439114570618, "learning_rate": 0.0001, "elbo": 3.69953, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7150480882007977}
{"step": 29250, "loss": 3.701255035400391, "grad_norm": 0.6797744035720825, "learning_rate": 0.0001, "elbo": 3.70125, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7153413089373681}
{"step": 29255, "loss": 3.7216949462890625, "grad_norm": 0.7220682501792908, "learning_rate": 0.0001, "elbo": 3.72169, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7156345296739386}
{"step": 29260, "loss": 3.7225570678710938, "grad_norm": 0.6919059753417969, "learning_rate": 0.0001, "elbo": 3.72256, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.715927750410509}
{"step": 29265, "loss": 3.711394500732422, "grad_norm": 0.6415604948997498, "learning_rate": 0.0001, "elbo": 3.71139, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7162209711470795}
{"step": 29270, "loss": 3.7132553100585937, "grad_norm": 0.6898462176322937, "learning_rate": 0.0001, "elbo": 3.71326, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.71651419188365}
{"step": 29275, "loss": 3.7217506408691405, "grad_norm": 0.6534713506698608, "learning_rate": 0.0001, "elbo": 3.72175, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7168074126202204}
{"step": 29280, "loss": 3.6977439880371095, "grad_norm": 0.767484724521637, "learning_rate": 0.0001, "elbo": 3.69774, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7171006333567909}
{"step": 29285, "loss": 3.69354133605957, "grad_norm": 0.7332847118377686, "learning_rate": 0.0001, "elbo": 3.69354, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7173938540933613}
{"step": 29290, "loss": 3.7120635986328123, "grad_norm": 0.7004024982452393, "learning_rate": 0.0001, "elbo": 3.71206, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.717687074829932}
{"step": 29295, "loss": 3.7084575653076173, "grad_norm": 0.7006908059120178, "learning_rate": 0.0001, "elbo": 3.70846, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7179802955665024}
{"step": 29300, "loss": 3.720293426513672, "grad_norm": 0.6659448146820068, "learning_rate": 0.0001, "elbo": 3.72029, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.718273516303073}
{"step": 29305, "loss": 3.707584762573242, "grad_norm": 0.7494657039642334, "learning_rate": 0.0001, "elbo": 3.70758, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7185667370396436}
{"step": 29310, "loss": 3.6955322265625, "grad_norm": 0.6770943403244019, "learning_rate": 0.0001, "elbo": 3.69553, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.718859957776214}
{"step": 29315, "loss": 3.724030303955078, "grad_norm": 0.628221333026886, "learning_rate": 0.0001, "elbo": 3.72403, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7191531785127845}
{"step": 29320, "loss": 3.7309188842773438, "grad_norm": 0.6600320339202881, "learning_rate": 0.0001, "elbo": 3.73092, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.719446399249355}
{"step": 29325, "loss": 3.713224411010742, "grad_norm": 0.6368796229362488, "learning_rate": 0.0001, "elbo": 3.71322, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7197396199859254}
{"step": 29330, "loss": 3.7177520751953126, "grad_norm": 0.874468207359314, "learning_rate": 0.0001, "elbo": 3.71775, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7200328407224958}
{"step": 29335, "loss": 3.7065528869628905, "grad_norm": 0.6944902539253235, "learning_rate": 0.0001, "elbo": 3.70655, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7203260614590663}
{"step": 29340, "loss": 3.7120124816894533, "grad_norm": 0.8778019547462463, "learning_rate": 0.0001, "elbo": 3.71201, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7206192821956368}
{"step": 29345, "loss": 3.713245391845703, "grad_norm": 0.6594681739807129, "learning_rate": 0.0001, "elbo": 3.71325, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7209125029322072}
{"step": 29350, "loss": 3.7094093322753907, "grad_norm": 0.7011878490447998, "learning_rate": 0.0001, "elbo": 3.70941, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7212057236687779}
{"step": 29355, "loss": 3.7102012634277344, "grad_norm": 0.7036991119384766, "learning_rate": 0.0001, "elbo": 3.7102, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7214989444053483}
{"step": 29360, "loss": 3.69312744140625, "grad_norm": 0.7283453345298767, "learning_rate": 0.0001, "elbo": 3.69313, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7217921651419188}
{"step": 29365, "loss": 3.7072223663330077, "grad_norm": 0.6764866709709167, "learning_rate": 0.0001, "elbo": 3.70722, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7220853858784895}
{"step": 29370, "loss": 3.693816375732422, "grad_norm": 0.7157332301139832, "learning_rate": 0.0001, "elbo": 3.69382, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.72237860661506}
{"step": 29375, "loss": 3.6979019165039064, "grad_norm": 0.688799262046814, "learning_rate": 0.0001, "elbo": 3.6979, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7226718273516304}
{"step": 29380, "loss": 3.7145317077636717, "grad_norm": 0.6490650177001953, "learning_rate": 0.0001, "elbo": 3.71453, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7229650480882008}
{"step": 29385, "loss": 3.710835266113281, "grad_norm": 0.6973670721054077, "learning_rate": 0.0001, "elbo": 3.71084, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7232582688247713}
{"step": 29390, "loss": 3.713714599609375, "grad_norm": 0.6456860303878784, "learning_rate": 0.0001, "elbo": 3.71371, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7235514895613417}
{"step": 29395, "loss": 3.712239074707031, "grad_norm": 0.6806445121765137, "learning_rate": 0.0001, "elbo": 3.71224, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7238447102979122}
{"step": 29400, "loss": 3.694445037841797, "grad_norm": 0.6759020090103149, "learning_rate": 0.0001, "elbo": 3.69445, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7241379310344827}
{"step": 29405, "loss": 3.719921875, "grad_norm": 0.6758623123168945, "learning_rate": 0.0001, "elbo": 3.71992, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.724431151771053}
{"step": 29410, "loss": 3.714403533935547, "grad_norm": 0.6398929357528687, "learning_rate": 0.0001, "elbo": 3.7144, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7247243725076238}
{"step": 29415, "loss": 3.6943058013916015, "grad_norm": 0.7590557932853699, "learning_rate": 0.0001, "elbo": 3.69431, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7250175932441942}
{"step": 29420, "loss": 3.698260498046875, "grad_norm": 0.7233080267906189, "learning_rate": 0.0001, "elbo": 3.69826, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7253108139807647}
{"step": 29425, "loss": 3.6883663177490233, "grad_norm": 0.7293588519096375, "learning_rate": 0.0001, "elbo": 3.68837, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7256040347173354}
{"step": 29430, "loss": 3.718027877807617, "grad_norm": 0.6482865214347839, "learning_rate": 0.0001, "elbo": 3.71803, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7258972554539058}
{"step": 29435, "loss": 3.7115039825439453, "grad_norm": 0.6825225949287415, "learning_rate": 0.0001, "elbo": 3.7115, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7261904761904763}
{"step": 29440, "loss": 3.7192398071289063, "grad_norm": 1.2508233785629272, "learning_rate": 0.0001, "elbo": 3.71924, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7264836969270467}
{"step": 29445, "loss": 3.708626556396484, "grad_norm": 0.6936032176017761, "learning_rate": 0.0001, "elbo": 3.70863, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7267769176636172}
{"step": 29450, "loss": 3.7201766967773438, "grad_norm": 0.683194100856781, "learning_rate": 0.0001, "elbo": 3.72018, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7270701384001876}
{"step": 29455, "loss": 3.706324005126953, "grad_norm": 0.9023334980010986, "learning_rate": 0.0001, "elbo": 3.70632, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.727363359136758}
{"step": 29460, "loss": 3.707044219970703, "grad_norm": 0.7334514260292053, "learning_rate": 0.0001, "elbo": 3.70704, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7276565798733285}
{"step": 29465, "loss": 3.684642791748047, "grad_norm": 0.6965337991714478, "learning_rate": 0.0001, "elbo": 3.68464, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.727949800609899}
{"step": 29470, "loss": 3.6941883087158205, "grad_norm": 0.7380409240722656, "learning_rate": 0.0001, "elbo": 3.69419, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7282430213464697}
{"step": 29470, "loss": 3.6941883087158205, "grad_norm": 0.7380409240722656, "learning_rate": 0.0001, "elbo": 3.69419, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7282430213464697}
{"step": 29475, "loss": 3.71510009765625, "grad_norm": 0.7058398723602295, "learning_rate": 0.0001, "elbo": 3.7151, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7285362420830401}
{"step": 29480, "loss": 3.7191864013671876, "grad_norm": 0.6703943014144897, "learning_rate": 0.0001, "elbo": 3.71919, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7288294628196106}
{"step": 29485, "loss": 3.694171905517578, "grad_norm": 0.6937369704246521, "learning_rate": 0.0001, "elbo": 3.69417, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.729122683556181}
{"step": 29490, "loss": 3.7034835815429688, "grad_norm": 0.6595439910888672, "learning_rate": 0.0001, "elbo": 3.70348, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7294159042927517}
{"step": 29490, "loss": 3.7034835815429688, "grad_norm": 0.6595439910888672, "learning_rate": 0.0001, "elbo": 3.70348, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7294159042927517}
{"step": 29495, "loss": 3.7192733764648436, "grad_norm": 0.7110040783882141, "learning_rate": 0.0001, "elbo": 3.71927, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7297091250293222}
{"step": 29495, "loss": 3.7192733764648436, "grad_norm": 0.7110040783882141, "learning_rate": 0.0001, "elbo": 3.71927, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7297091250293222}
{"step": 29500, "loss": 3.7020488739013673, "grad_norm": 0.6806484460830688, "learning_rate": 0.0001, "elbo": 3.70205, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7300023457658926}
{"step": 29500, "loss": 3.7020488739013673, "grad_norm": 0.6806484460830688, "learning_rate": 0.0001, "elbo": 3.70205, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7300023457658926}
{"step": 29505, "loss": 3.712890625, "grad_norm": 0.6548235416412354, "learning_rate": 0.0001, "elbo": 3.71289, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.730295566502463}
{"step": 29505, "loss": 3.712890625, "grad_norm": 0.6548235416412354, "learning_rate": 0.0001, "elbo": 3.71289, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.730295566502463}
{"step": 29510, "loss": 3.708919143676758, "grad_norm": 0.6940318942070007, "learning_rate": 0.0001, "elbo": 3.70892, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7305887872390335}
{"step": 29515, "loss": 3.720233917236328, "grad_norm": 0.6789488196372986, "learning_rate": 0.0001, "elbo": 3.72023, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.730882007975604}
{"step": 29520, "loss": 3.7045974731445312, "grad_norm": 0.6800869703292847, "learning_rate": 0.0001, "elbo": 3.7046, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7311752287121744}
{"step": 29525, "loss": 3.7115715026855467, "grad_norm": 0.6906415224075317, "learning_rate": 0.0001, "elbo": 3.71157, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.731468449448745}
{"step": 29530, "loss": 3.6960990905761717, "grad_norm": 0.6664748787879944, "learning_rate": 0.0001, "elbo": 3.6961, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7317616701853153}
{"step": 29535, "loss": 3.6903343200683594, "grad_norm": 0.6528016924858093, "learning_rate": 0.0001, "elbo": 3.69033, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.732054890921886}
{"step": 29540, "loss": 3.7163970947265623, "grad_norm": 0.7757557034492493, "learning_rate": 0.0001, "elbo": 3.7164, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7323481116584565}
{"step": 29545, "loss": 3.715326690673828, "grad_norm": 0.726629376411438, "learning_rate": 0.0001, "elbo": 3.71533, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.732641332395027}
{"step": 29550, "loss": 3.7081283569335937, "grad_norm": 0.7462615966796875, "learning_rate": 0.0001, "elbo": 3.70813, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7329345531315976}
{"step": 29555, "loss": 3.704676055908203, "grad_norm": 0.7589481472969055, "learning_rate": 0.0001, "elbo": 3.70468, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.733227773868168}
{"step": 29560, "loss": 3.7072032928466796, "grad_norm": 0.6430266499519348, "learning_rate": 0.0001, "elbo": 3.7072, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7335209946047385}
{"step": 29565, "loss": 3.7000885009765625, "grad_norm": 0.6495271325111389, "learning_rate": 0.0001, "elbo": 3.70009, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.733814215341309}
{"step": 29570, "loss": 3.713273620605469, "grad_norm": 0.7907682657241821, "learning_rate": 0.0001, "elbo": 3.71327, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7341074360778794}
{"step": 29575, "loss": 3.7124069213867186, "grad_norm": 0.6820892095565796, "learning_rate": 0.0001, "elbo": 3.71241, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7344006568144499}
{"step": 29580, "loss": 3.725524139404297, "grad_norm": 0.6786142587661743, "learning_rate": 0.0001, "elbo": 3.72552, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7346938775510203}
{"step": 29585, "loss": 3.700257110595703, "grad_norm": 0.9465976357460022, "learning_rate": 0.0001, "elbo": 3.70026, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7349870982875908}
{"step": 29590, "loss": 3.7108158111572265, "grad_norm": 0.6943913102149963, "learning_rate": 0.0001, "elbo": 3.71082, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7352803190241612}
{"step": 29595, "loss": 3.699633026123047, "grad_norm": 0.6691994667053223, "learning_rate": 0.0001, "elbo": 3.69963, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.735573539760732}
{"step": 29600, "loss": 3.7046878814697264, "grad_norm": 0.655375599861145, "learning_rate": 0.0001, "elbo": 3.70469, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7358667604973024}
{"step": 29605, "loss": 3.726025390625, "grad_norm": 0.6590607166290283, "learning_rate": 0.0001, "elbo": 3.72603, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7361599812338728}
{"step": 29610, "loss": 3.715959930419922, "grad_norm": 0.6950054168701172, "learning_rate": 0.0001, "elbo": 3.71596, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7364532019704435}
{"step": 29615, "loss": 3.702892303466797, "grad_norm": 0.7462263703346252, "learning_rate": 0.0001, "elbo": 3.70289, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.736746422707014}
{"step": 29615, "loss": 3.702892303466797, "grad_norm": 0.7462263703346252, "learning_rate": 0.0001, "elbo": 3.70289, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.736746422707014}
{"step": 29620, "loss": 3.725795364379883, "grad_norm": 0.685566246509552, "learning_rate": 0.0001, "elbo": 3.7258, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7370396434435844}
{"step": 29625, "loss": 3.6994117736816405, "grad_norm": 0.6448817849159241, "learning_rate": 0.0001, "elbo": 3.69941, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7373328641801549}
{"step": 29630, "loss": 3.7023181915283203, "grad_norm": 0.650945782661438, "learning_rate": 0.0001, "elbo": 3.70232, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7376260849167253}
{"step": 29630, "loss": 3.7023181915283203, "grad_norm": 0.650945782661438, "learning_rate": 0.0001, "elbo": 3.70232, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7376260849167253}
{"step": 29635, "loss": 3.714703369140625, "grad_norm": 0.6706317663192749, "learning_rate": 0.0001, "elbo": 3.7147, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7379193056532958}
{"step": 29640, "loss": 3.7031875610351563, "grad_norm": 0.6670733690261841, "learning_rate": 0.0001, "elbo": 3.70319, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7382125263898662}
{"step": 29645, "loss": 3.694569396972656, "grad_norm": 0.6605997681617737, "learning_rate": 0.0001, "elbo": 3.69457, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7385057471264367}
{"step": 29650, "loss": 3.7096580505371093, "grad_norm": 0.6867520213127136, "learning_rate": 0.0001, "elbo": 3.70966, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7387989678630071}
{"step": 29655, "loss": 3.7225975036621093, "grad_norm": 0.9472178220748901, "learning_rate": 0.0001, "elbo": 3.7226, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7390921885995778}
{"step": 29660, "loss": 3.699860763549805, "grad_norm": 0.695716142654419, "learning_rate": 0.0001, "elbo": 3.69986, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7393854093361483}
{"step": 29665, "loss": 3.7098056793212892, "grad_norm": 0.6475688815116882, "learning_rate": 0.0001, "elbo": 3.70981, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7396786300727187}
{"step": 29670, "loss": 3.690586471557617, "grad_norm": 0.6321907639503479, "learning_rate": 0.0001, "elbo": 3.69059, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7399718508092892}
{"step": 29675, "loss": 3.7081588745117187, "grad_norm": 0.7000467777252197, "learning_rate": 0.0001, "elbo": 3.70816, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7402650715458599}
{"step": 29680, "loss": 3.698476791381836, "grad_norm": 0.6691187024116516, "learning_rate": 0.0001, "elbo": 3.69848, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7405582922824303}
{"step": 29685, "loss": 3.702991485595703, "grad_norm": 0.7548509836196899, "learning_rate": 0.0001, "elbo": 3.70299, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7408515130190008}
{"step": 29690, "loss": 3.712239456176758, "grad_norm": 0.6761579513549805, "learning_rate": 0.0001, "elbo": 3.71224, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7411447337555712}
{"step": 29695, "loss": 3.69946403503418, "grad_norm": 0.8358791470527649, "learning_rate": 0.0001, "elbo": 3.69946, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7414379544921417}
{"step": 29700, "loss": 3.702773666381836, "grad_norm": 0.8525394201278687, "learning_rate": 0.0001, "elbo": 3.70277, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7417311752287121}
{"step": 29705, "loss": 3.7038475036621095, "grad_norm": 0.7197300791740417, "learning_rate": 0.0001, "elbo": 3.70385, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7420243959652826}
{"step": 29710, "loss": 3.720875549316406, "grad_norm": 0.7561571598052979, "learning_rate": 0.0001, "elbo": 3.72088, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.742317616701853}
{"step": 29715, "loss": 3.7099067687988283, "grad_norm": 0.7831122279167175, "learning_rate": 0.0001, "elbo": 3.70991, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7426108374384235}
{"step": 29720, "loss": 3.7011070251464844, "grad_norm": 0.7197749614715576, "learning_rate": 0.0001, "elbo": 3.70111, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7429040581749942}
{"step": 29725, "loss": 3.7055015563964844, "grad_norm": 0.6795253157615662, "learning_rate": 0.0001, "elbo": 3.7055, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7431972789115646}
{"step": 29730, "loss": 3.713936614990234, "grad_norm": 0.6252351999282837, "learning_rate": 0.0001, "elbo": 3.71394, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.743490499648135}
{"step": 29735, "loss": 3.6886695861816405, "grad_norm": 0.6649053692817688, "learning_rate": 0.0001, "elbo": 3.68867, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7437837203847057}
{"step": 29740, "loss": 3.6937301635742186, "grad_norm": 0.6228283643722534, "learning_rate": 0.0001, "elbo": 3.69373, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7440769411212762}
{"step": 29745, "loss": 3.708063507080078, "grad_norm": 0.6546422839164734, "learning_rate": 0.0001, "elbo": 3.70806, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7443701618578467}
{"step": 29750, "loss": 3.7020523071289064, "grad_norm": 0.7042027115821838, "learning_rate": 0.0001, "elbo": 3.70205, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7446633825944171}
{"step": 29755, "loss": 3.7226829528808594, "grad_norm": 0.6943888664245605, "learning_rate": 0.0001, "elbo": 3.72268, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7449566033309876}
{"step": 29760, "loss": 3.7210720062255858, "grad_norm": 0.7104317545890808, "learning_rate": 0.0001, "elbo": 3.72107, "mask_frac": 0.5014, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.745249824067558}
{"step": 29765, "loss": 3.7073265075683595, "grad_norm": 0.73462975025177, "learning_rate": 0.0001, "elbo": 3.70733, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7455430448041285}
{"step": 29770, "loss": 3.7030193328857424, "grad_norm": 0.6709955334663391, "learning_rate": 0.0001, "elbo": 3.70302, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.745836265540699}
{"step": 29775, "loss": 3.7113128662109376, "grad_norm": 0.6911951303482056, "learning_rate": 0.0001, "elbo": 3.71131, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7461294862772694}
{"step": 29780, "loss": 3.721710205078125, "grad_norm": 0.8707585334777832, "learning_rate": 0.0001, "elbo": 3.72171, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.74642270701384}
{"step": 29785, "loss": 3.688676452636719, "grad_norm": 0.6954772472381592, "learning_rate": 0.0001, "elbo": 3.68868, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7467159277504105}
{"step": 29785, "loss": 3.688676452636719, "grad_norm": 0.6954772472381592, "learning_rate": 0.0001, "elbo": 3.68868, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7467159277504105}
{"step": 29790, "loss": 3.695952606201172, "grad_norm": 0.6721779704093933, "learning_rate": 0.0001, "elbo": 3.69595, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.747009148486981}
{"step": 29795, "loss": 3.6953571319580076, "grad_norm": 0.8822612762451172, "learning_rate": 0.0001, "elbo": 3.69536, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7473023692235516}
{"step": 29800, "loss": 3.699767303466797, "grad_norm": 0.7461770176887512, "learning_rate": 0.0001, "elbo": 3.69977, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.747595589960122}
{"step": 29805, "loss": 3.7103538513183594, "grad_norm": 0.6706157922744751, "learning_rate": 0.0001, "elbo": 3.71035, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7478888106966926}
{"step": 29810, "loss": 3.69920654296875, "grad_norm": 0.7194100022315979, "learning_rate": 0.0001, "elbo": 3.69921, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.748182031433263}
{"step": 29815, "loss": 3.7147262573242186, "grad_norm": 0.7074007987976074, "learning_rate": 0.0001, "elbo": 3.71473, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7484752521698335}
{"step": 29820, "loss": 3.7143096923828125, "grad_norm": 0.9014319777488708, "learning_rate": 0.0001, "elbo": 3.71431, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.748768472906404}
{"step": 29825, "loss": 3.708931732177734, "grad_norm": 0.746239960193634, "learning_rate": 0.0001, "elbo": 3.70893, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7490616936429744}
{"step": 29830, "loss": 3.715534210205078, "grad_norm": 0.6885539293289185, "learning_rate": 0.0001, "elbo": 3.71553, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7493549143795448}
{"step": 29835, "loss": 3.715221405029297, "grad_norm": 0.9974888563156128, "learning_rate": 0.0001, "elbo": 3.71522, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7496481351161153}
{"step": 29840, "loss": 3.7111801147460937, "grad_norm": 0.6540918350219727, "learning_rate": 0.0001, "elbo": 3.71118, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.749941355852686}
{"step": 29845, "loss": 3.7004390716552735, "grad_norm": 0.7062862515449524, "learning_rate": 0.0001, "elbo": 3.70044, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7502345765892564}
{"step": 29850, "loss": 3.7062679290771485, "grad_norm": 0.7115222811698914, "learning_rate": 0.0001, "elbo": 3.70627, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7505277973258269}
{"step": 29855, "loss": 3.7221580505371095, "grad_norm": 0.6679966449737549, "learning_rate": 0.0001, "elbo": 3.72216, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7508210180623975}
{"step": 29860, "loss": 3.7179023742675783, "grad_norm": 0.7280963659286499, "learning_rate": 0.0001, "elbo": 3.7179, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.751114238798968}
{"step": 29865, "loss": 3.6980003356933593, "grad_norm": 0.7438339591026306, "learning_rate": 0.0001, "elbo": 3.698, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7514074595355384}
{"step": 29870, "loss": 3.705426025390625, "grad_norm": 0.6761045455932617, "learning_rate": 0.0001, "elbo": 3.70543, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.751700680272109}
{"step": 29875, "loss": 3.7052257537841795, "grad_norm": 0.7437434196472168, "learning_rate": 0.0001, "elbo": 3.70523, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7519939010086794}
{"step": 29880, "loss": 3.710340881347656, "grad_norm": 0.8010134100914001, "learning_rate": 0.0001, "elbo": 3.71034, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7522871217452498}
{"step": 29885, "loss": 3.6873451232910157, "grad_norm": 0.71517413854599, "learning_rate": 0.0001, "elbo": 3.68734, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7525803424818203}
{"step": 29890, "loss": 3.7178321838378907, "grad_norm": 0.8571534156799316, "learning_rate": 0.0001, "elbo": 3.71783, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7528735632183907}
{"step": 29895, "loss": 3.7108673095703124, "grad_norm": 0.6682106852531433, "learning_rate": 0.0001, "elbo": 3.71087, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7531667839549612}
{"step": 29900, "loss": 3.703333282470703, "grad_norm": 0.7312139868736267, "learning_rate": 0.0001, "elbo": 3.70333, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7534600046915316}
{"step": 29905, "loss": 3.70911865234375, "grad_norm": 0.6173835396766663, "learning_rate": 0.0001, "elbo": 3.70912, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7537532254281023}
{"step": 29910, "loss": 3.7207122802734376, "grad_norm": 0.6715274453163147, "learning_rate": 0.0001, "elbo": 3.72071, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7540464461646728}
{"step": 29915, "loss": 3.7041954040527343, "grad_norm": 0.6858676671981812, "learning_rate": 0.0001, "elbo": 3.7042, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7543396669012432}
{"step": 29920, "loss": 3.6971145629882813, "grad_norm": 0.7154350280761719, "learning_rate": 0.0001, "elbo": 3.69711, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7546328876378139}
{"step": 29925, "loss": 3.7102081298828127, "grad_norm": 0.6813707947731018, "learning_rate": 0.0001, "elbo": 3.71021, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7549261083743843}
{"step": 29930, "loss": 3.7091964721679687, "grad_norm": 0.8984549641609192, "learning_rate": 0.0001, "elbo": 3.7092, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7552193291109548}
{"step": 29935, "loss": 3.706949996948242, "grad_norm": 0.7019350528717041, "learning_rate": 0.0001, "elbo": 3.70695, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7555125498475252}
{"step": 29940, "loss": 3.7015430450439455, "grad_norm": 0.8191888928413391, "learning_rate": 0.0001, "elbo": 3.70154, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7558057705840957}
{"step": 29945, "loss": 3.7159408569335937, "grad_norm": 0.6709583401679993, "learning_rate": 0.0001, "elbo": 3.71594, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7560989913206662}
{"step": 29950, "loss": 3.705316925048828, "grad_norm": 0.6828041076660156, "learning_rate": 0.0001, "elbo": 3.70532, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7563922120572366}
{"step": 29955, "loss": 3.7054161071777343, "grad_norm": 0.7030977606773376, "learning_rate": 0.0001, "elbo": 3.70542, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.756685432793807}
{"step": 29960, "loss": 3.715443420410156, "grad_norm": 0.7306244373321533, "learning_rate": 0.0001, "elbo": 3.71544, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7569786535303775}
{"step": 29965, "loss": 3.7080894470214845, "grad_norm": 0.6574468016624451, "learning_rate": 0.0001, "elbo": 3.70809, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7572718742669482}
{"step": 29970, "loss": 3.7062530517578125, "grad_norm": 0.6643947958946228, "learning_rate": 0.0001, "elbo": 3.70625, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7575650950035187}
{"step": 29975, "loss": 3.697479248046875, "grad_norm": 0.6782091856002808, "learning_rate": 0.0001, "elbo": 3.69748, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.757858315740089}
{"step": 29980, "loss": 3.712696075439453, "grad_norm": 0.6214607357978821, "learning_rate": 0.0001, "elbo": 3.7127, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7581515364766598}
{"step": 29985, "loss": 3.7059310913085937, "grad_norm": 0.7688885927200317, "learning_rate": 0.0001, "elbo": 3.70593, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7584447572132302}
{"step": 29990, "loss": 3.699257659912109, "grad_norm": 0.693929135799408, "learning_rate": 0.0001, "elbo": 3.69926, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7587379779498007}
{"step": 29995, "loss": 3.7091983795166015, "grad_norm": 1.038529872894287, "learning_rate": 0.0001, "elbo": 3.7092, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7590311986863711}
{"step": 30000, "loss": 3.694110870361328, "grad_norm": 0.6613322496414185, "learning_rate": 0.0001, "elbo": 3.69411, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7593244194229416}
{"step": 30005, "loss": 3.720628356933594, "grad_norm": 0.793726921081543, "learning_rate": 0.0001, "elbo": 3.72063, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.759617640159512}
{"step": 30010, "loss": 3.692247009277344, "grad_norm": 0.6848849058151245, "learning_rate": 0.0001, "elbo": 3.69225, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7599108608960825}
{"step": 30015, "loss": 3.707773971557617, "grad_norm": 0.806820809841156, "learning_rate": 0.0001, "elbo": 3.70777, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.760204081632653}
{"step": 30020, "loss": 3.7268321990966795, "grad_norm": 0.6290608048439026, "learning_rate": 0.0001, "elbo": 3.72683, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7604973023692234}
{"step": 30025, "loss": 3.7025596618652346, "grad_norm": 0.699842095375061, "learning_rate": 0.0001, "elbo": 3.70256, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.760790523105794}
{"step": 30030, "loss": 3.695179748535156, "grad_norm": 0.6943364143371582, "learning_rate": 0.0001, "elbo": 3.69518, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7610837438423645}
{"step": 30035, "loss": 3.709478759765625, "grad_norm": 1.0565952062606812, "learning_rate": 0.0001, "elbo": 3.70948, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.761376964578935}
{"step": 30040, "loss": 3.6884159088134765, "grad_norm": 0.6589259505271912, "learning_rate": 0.0001, "elbo": 3.68842, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7616701853155057}
{"step": 30045, "loss": 3.713509368896484, "grad_norm": 0.8410758972167969, "learning_rate": 0.0001, "elbo": 3.71351, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7619634060520761}
{"step": 30050, "loss": 3.692057800292969, "grad_norm": 0.7044877409934998, "learning_rate": 0.0001, "elbo": 3.69206, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7622566267886466}
{"step": 30055, "loss": 3.706251525878906, "grad_norm": 0.6951673626899719, "learning_rate": 0.0001, "elbo": 3.70625, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.762549847525217}
{"step": 30060, "loss": 3.7008438110351562, "grad_norm": 0.6914904117584229, "learning_rate": 0.0001, "elbo": 3.70084, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7628430682617875}
{"step": 30060, "loss": 3.7008438110351562, "grad_norm": 0.6914904117584229, "learning_rate": 0.0001, "elbo": 3.70084, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7628430682617875}
{"step": 30065, "loss": 3.7047073364257814, "grad_norm": 0.7624372243881226, "learning_rate": 0.0001, "elbo": 3.70471, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.763136288998358}
{"step": 30065, "loss": 3.7047073364257814, "grad_norm": 0.7624372243881226, "learning_rate": 0.0001, "elbo": 3.70471, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.763136288998358}
{"step": 30070, "loss": 3.715686798095703, "grad_norm": 0.7642698287963867, "learning_rate": 0.0001, "elbo": 3.71569, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7634295097349284}
{"step": 30075, "loss": 3.696277618408203, "grad_norm": 0.6603689193725586, "learning_rate": 0.0001, "elbo": 3.69628, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7637227304714989}
{"step": 30080, "loss": 3.706349563598633, "grad_norm": 0.6810228228569031, "learning_rate": 0.0001, "elbo": 3.70635, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7640159512080693}
{"step": 30085, "loss": 3.695151519775391, "grad_norm": 0.6724550724029541, "learning_rate": 0.0001, "elbo": 3.69515, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7643091719446398}
{"step": 30090, "loss": 3.7099258422851564, "grad_norm": 0.9733878970146179, "learning_rate": 0.0001, "elbo": 3.70993, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7646023926812104}
{"step": 30095, "loss": 3.710260772705078, "grad_norm": 0.7633898258209229, "learning_rate": 0.0001, "elbo": 3.71026, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.764895613417781}
{"step": 30100, "loss": 3.714161682128906, "grad_norm": 0.6842955350875854, "learning_rate": 0.0001, "elbo": 3.71416, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7651888341543513}
{"step": 30105, "loss": 3.716790771484375, "grad_norm": 1.0797364711761475, "learning_rate": 0.0001, "elbo": 3.71679, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.765482054890922}
{"step": 30110, "loss": 3.702117156982422, "grad_norm": 0.6962234973907471, "learning_rate": 0.0001, "elbo": 3.70212, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7657752756274925}
{"step": 30115, "loss": 3.7029930114746095, "grad_norm": 0.7614786624908447, "learning_rate": 0.0001, "elbo": 3.70299, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.766068496364063}
{"step": 30120, "loss": 3.713883972167969, "grad_norm": 0.7463974356651306, "learning_rate": 0.0001, "elbo": 3.71388, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7663617171006334}
{"step": 30125, "loss": 3.6863174438476562, "grad_norm": 0.723240315914154, "learning_rate": 0.0001, "elbo": 3.68632, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7666549378372038}
{"step": 30130, "loss": 3.6963504791259765, "grad_norm": 0.6915236711502075, "learning_rate": 0.0001, "elbo": 3.69635, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7669481585737743}
{"step": 30135, "loss": 3.6900390625, "grad_norm": 0.6844344139099121, "learning_rate": 0.0001, "elbo": 3.69004, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7672413793103448}
{"step": 30140, "loss": 3.7038368225097655, "grad_norm": 0.6740748286247253, "learning_rate": 0.0001, "elbo": 3.70384, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7675346000469152}
{"step": 30145, "loss": 3.695903778076172, "grad_norm": 0.6964672207832336, "learning_rate": 0.0001, "elbo": 3.6959, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7678278207834857}
{"step": 30150, "loss": 3.7029273986816404, "grad_norm": 0.7777390480041504, "learning_rate": 0.0001, "elbo": 3.70293, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7681210415200563}
{"step": 30155, "loss": 3.7102005004882814, "grad_norm": 0.6930555105209351, "learning_rate": 0.0001, "elbo": 3.7102, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7684142622566268}
{"step": 30160, "loss": 3.684613800048828, "grad_norm": 0.7106949687004089, "learning_rate": 0.0001, "elbo": 3.68461, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7687074829931972}
{"step": 30165, "loss": 3.7152923583984374, "grad_norm": 0.6476503610610962, "learning_rate": 0.0001, "elbo": 3.71529, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.769000703729768}
{"step": 30170, "loss": 3.721792984008789, "grad_norm": 0.6665030717849731, "learning_rate": 0.0001, "elbo": 3.72179, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7692939244663384}
{"step": 30175, "loss": 3.707879638671875, "grad_norm": 0.7356258034706116, "learning_rate": 0.0001, "elbo": 3.70788, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7695871452029088}
{"step": 30180, "loss": 3.702543258666992, "grad_norm": 0.7753907442092896, "learning_rate": 0.0001, "elbo": 3.70254, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7698803659394793}
{"step": 30180, "loss": 3.702543258666992, "grad_norm": 0.7753907442092896, "learning_rate": 0.0001, "elbo": 3.70254, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7698803659394793}
{"step": 30185, "loss": 3.7226028442382812, "grad_norm": 0.7971630692481995, "learning_rate": 0.0001, "elbo": 3.7226, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7701735866760497}
{"step": 30190, "loss": 3.701533889770508, "grad_norm": 0.6721546053886414, "learning_rate": 0.0001, "elbo": 3.70153, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7704668074126202}
{"step": 30195, "loss": 3.694118118286133, "grad_norm": 0.6647551655769348, "learning_rate": 0.0001, "elbo": 3.69412, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7707600281491906}
{"step": 30200, "loss": 3.70019645690918, "grad_norm": 0.6871697902679443, "learning_rate": 0.0001, "elbo": 3.7002, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.771053248885761}
{"step": 30205, "loss": 3.696352005004883, "grad_norm": 0.7037275433540344, "learning_rate": 0.0001, "elbo": 3.69635, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7713464696223316}
{"step": 30210, "loss": 3.704051208496094, "grad_norm": 0.6701352000236511, "learning_rate": 0.0001, "elbo": 3.70405, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7716396903589022}
{"step": 30215, "loss": 3.694676971435547, "grad_norm": 0.6644106507301331, "learning_rate": 0.0001, "elbo": 3.69468, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7719329110954727}
{"step": 30220, "loss": 3.7142784118652346, "grad_norm": 0.7084068655967712, "learning_rate": 0.0001, "elbo": 3.71428, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7722261318320431}
{"step": 30225, "loss": 3.7150989532470704, "grad_norm": 0.6747832298278809, "learning_rate": 0.0001, "elbo": 3.7151, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7725193525686138}
{"step": 30230, "loss": 3.6925376892089843, "grad_norm": 0.727177619934082, "learning_rate": 0.0001, "elbo": 3.69254, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7728125733051843}
{"step": 30235, "loss": 3.6940441131591797, "grad_norm": 0.6440286040306091, "learning_rate": 0.0001, "elbo": 3.69404, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7731057940417547}
{"step": 30240, "loss": 3.713173675537109, "grad_norm": 0.6956771612167358, "learning_rate": 0.0001, "elbo": 3.71317, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7733990147783252}
{"step": 30245, "loss": 3.702759552001953, "grad_norm": 0.6933280229568481, "learning_rate": 0.0001, "elbo": 3.70276, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7736922355148956}
{"step": 30250, "loss": 3.697935104370117, "grad_norm": 0.679202139377594, "learning_rate": 0.0001, "elbo": 3.69794, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.773985456251466}
{"step": 30255, "loss": 3.7142707824707033, "grad_norm": 0.6501011848449707, "learning_rate": 0.0001, "elbo": 3.71427, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7742786769880365}
{"step": 30260, "loss": 3.698159027099609, "grad_norm": 1.1619813442230225, "learning_rate": 0.0001, "elbo": 3.69816, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.774571897724607}
{"step": 30265, "loss": 3.7027870178222657, "grad_norm": 0.6984394192695618, "learning_rate": 0.0001, "elbo": 3.70279, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7748651184611774}
{"step": 30270, "loss": 3.7023033142089843, "grad_norm": 0.6988206505775452, "learning_rate": 0.0001, "elbo": 3.7023, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.775158339197748}
{"step": 30275, "loss": 3.694785308837891, "grad_norm": 0.6417120099067688, "learning_rate": 0.0001, "elbo": 3.69479, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7754515599343186}
{"step": 30280, "loss": 3.700566864013672, "grad_norm": 0.638412594795227, "learning_rate": 0.0001, "elbo": 3.70057, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.775744780670889}
{"step": 30285, "loss": 3.7041839599609374, "grad_norm": 0.6986324191093445, "learning_rate": 0.0001, "elbo": 3.70418, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7760380014074595}
{"step": 30290, "loss": 3.691863250732422, "grad_norm": 0.6547642350196838, "learning_rate": 0.0001, "elbo": 3.69186, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7763312221440302}
{"step": 30295, "loss": 3.700274658203125, "grad_norm": 0.6800197958946228, "learning_rate": 0.0001, "elbo": 3.70027, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7766244428806006}
{"step": 30300, "loss": 3.7021751403808594, "grad_norm": 0.803945779800415, "learning_rate": 0.0001, "elbo": 3.70217, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.776917663617171}
{"step": 30300, "loss": 3.7021751403808594, "grad_norm": 0.803945779800415, "learning_rate": 0.0001, "elbo": 3.70217, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.776917663617171}
{"step": 30305, "loss": 3.7114810943603516, "grad_norm": 0.6365105509757996, "learning_rate": 0.0001, "elbo": 3.71148, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7772108843537415}
{"step": 30310, "loss": 3.6897018432617186, "grad_norm": 0.6242495179176331, "learning_rate": 0.0001, "elbo": 3.6897, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.777504105090312}
{"step": 30315, "loss": 3.7090511322021484, "grad_norm": 0.6693689227104187, "learning_rate": 0.0001, "elbo": 3.70905, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7777973258268824}
{"step": 30320, "loss": 3.7032001495361326, "grad_norm": 0.6721495985984802, "learning_rate": 0.0001, "elbo": 3.7032, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.778090546563453}
{"step": 30325, "loss": 3.7031715393066404, "grad_norm": 0.8306316137313843, "learning_rate": 0.0001, "elbo": 3.70317, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7783837673000233}
{"step": 30330, "loss": 3.702658462524414, "grad_norm": 0.669884443283081, "learning_rate": 0.0001, "elbo": 3.70266, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7786769880365938}
{"step": 30335, "loss": 3.7036468505859377, "grad_norm": 0.8372103571891785, "learning_rate": 0.0001, "elbo": 3.70365, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7789702087731645}
{"step": 30340, "loss": 3.7032142639160157, "grad_norm": 0.7212467789649963, "learning_rate": 0.0001, "elbo": 3.70321, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.779263429509735}
{"step": 30345, "loss": 3.7058929443359374, "grad_norm": 0.6834508776664734, "learning_rate": 0.0001, "elbo": 3.70589, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7795566502463054}
{"step": 30350, "loss": 3.6881622314453124, "grad_norm": 0.6799111366271973, "learning_rate": 0.0001, "elbo": 3.68816, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.779849870982876}
{"step": 30355, "loss": 3.701094055175781, "grad_norm": 0.7238749861717224, "learning_rate": 0.0001, "elbo": 3.70109, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7801430917194465}
{"step": 30360, "loss": 3.6982837677001954, "grad_norm": 0.6778954267501831, "learning_rate": 0.0001, "elbo": 3.69828, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.780436312456017}
{"step": 30365, "loss": 3.710932159423828, "grad_norm": 0.6454607248306274, "learning_rate": 0.0001, "elbo": 3.71093, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7807295331925874}
{"step": 30370, "loss": 3.692501449584961, "grad_norm": 0.7171974778175354, "learning_rate": 0.0001, "elbo": 3.6925, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7810227539291579}
{"step": 30375, "loss": 3.7041091918945312, "grad_norm": 0.658538818359375, "learning_rate": 0.0001, "elbo": 3.70411, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7813159746657283}
{"step": 30380, "loss": 3.69517822265625, "grad_norm": 0.6796607375144958, "learning_rate": 0.0001, "elbo": 3.69518, "mask_frac": 0.4996, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7816091954022988}
{"step": 30385, "loss": 3.6932884216308595, "grad_norm": 0.6695793867111206, "learning_rate": 0.0001, "elbo": 3.69329, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7819024161388692}
{"step": 30390, "loss": 3.7138427734375, "grad_norm": 0.7396389245986938, "learning_rate": 0.0001, "elbo": 3.71384, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7821956368754397}
{"step": 30395, "loss": 3.7110218048095702, "grad_norm": 0.7052623629570007, "learning_rate": 0.0001, "elbo": 3.71102, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7824888576120104}
{"step": 30400, "loss": 3.692504119873047, "grad_norm": 0.658013105392456, "learning_rate": 0.0001, "elbo": 3.6925, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7827820783485808}
{"step": 30405, "loss": 3.678516387939453, "grad_norm": 0.6757392287254333, "learning_rate": 0.0001, "elbo": 3.67852, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7830752990851513}
{"step": 30410, "loss": 3.695081329345703, "grad_norm": 0.7332587242126465, "learning_rate": 0.0001, "elbo": 3.69508, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.783368519821722}
{"step": 30415, "loss": 3.7078369140625, "grad_norm": 0.6878718733787537, "learning_rate": 0.0001, "elbo": 3.70784, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7836617405582924}
{"step": 30420, "loss": 3.706634521484375, "grad_norm": 0.8636969923973083, "learning_rate": 0.0001, "elbo": 3.70663, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7839549612948629}
{"step": 30425, "loss": 3.6894264221191406, "grad_norm": 0.9341470003128052, "learning_rate": 0.0001, "elbo": 3.68943, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7842481820314333}
{"step": 30430, "loss": 3.7004154205322264, "grad_norm": 0.8230943083763123, "learning_rate": 0.0001, "elbo": 3.70042, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7845414027680038}
{"step": 30435, "loss": 3.7023593902587892, "grad_norm": 0.7114782333374023, "learning_rate": 0.0001, "elbo": 3.70236, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7848346235045742}
{"step": 30440, "loss": 3.698686218261719, "grad_norm": 0.7321991324424744, "learning_rate": 0.0001, "elbo": 3.69869, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7851278442411447}
{"step": 30445, "loss": 3.7065608978271483, "grad_norm": 0.6371778249740601, "learning_rate": 0.0001, "elbo": 3.70656, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7854210649777151}
{"step": 30450, "loss": 3.7130714416503907, "grad_norm": 0.6918458342552185, "learning_rate": 0.0001, "elbo": 3.71307, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7857142857142856}
{"step": 30455, "loss": 3.710806655883789, "grad_norm": 0.7219094038009644, "learning_rate": 0.0001, "elbo": 3.71081, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.786007506450856}
{"step": 30460, "loss": 3.688323211669922, "grad_norm": 0.7017109394073486, "learning_rate": 0.0001, "elbo": 3.68832, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7863007271874267}
{"step": 30465, "loss": 3.6960380554199217, "grad_norm": 0.6530110836029053, "learning_rate": 0.0001, "elbo": 3.69604, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7865939479239972}
{"step": 30470, "loss": 3.6955551147460937, "grad_norm": 0.659635066986084, "learning_rate": 0.0001, "elbo": 3.69556, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7868871686605676}
{"step": 30475, "loss": 3.6864471435546875, "grad_norm": 0.7267823815345764, "learning_rate": 0.0001, "elbo": 3.68645, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7871803893971383}
{"step": 30480, "loss": 3.6874130249023436, "grad_norm": 0.7286988496780396, "learning_rate": 0.0001, "elbo": 3.68741, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7874736101337088}
{"step": 30485, "loss": 3.7004127502441406, "grad_norm": 0.671864926815033, "learning_rate": 0.0001, "elbo": 3.70041, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7877668308702792}
{"step": 30490, "loss": 3.678211975097656, "grad_norm": 1.0751467943191528, "learning_rate": 0.0001, "elbo": 3.67821, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7880600516068497}
{"step": 30490, "loss": 3.678211975097656, "grad_norm": 1.0751467943191528, "learning_rate": 0.0001, "elbo": 3.67821, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7880600516068497}
{"step": 30495, "loss": 3.69473876953125, "grad_norm": 0.6757672429084778, "learning_rate": 0.0001, "elbo": 3.69474, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7883532723434201}
{"step": 30500, "loss": 3.6982391357421873, "grad_norm": 0.8664048314094543, "learning_rate": 0.0001, "elbo": 3.69824, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7886464930799906}
{"step": 30505, "loss": 3.6900596618652344, "grad_norm": 0.7101549506187439, "learning_rate": 0.0001, "elbo": 3.69006, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.788939713816561}
{"step": 30510, "loss": 3.712533950805664, "grad_norm": 0.6845918893814087, "learning_rate": 0.0001, "elbo": 3.71253, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7892329345531315}
{"step": 30515, "loss": 3.7074424743652346, "grad_norm": 0.6754611730575562, "learning_rate": 0.0001, "elbo": 3.70744, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.789526155289702}
{"step": 30520, "loss": 3.699155807495117, "grad_norm": 0.645140528678894, "learning_rate": 0.0001, "elbo": 3.69916, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7898193760262726}
{"step": 30525, "loss": 3.7156787872314454, "grad_norm": 0.6821008324623108, "learning_rate": 0.0001, "elbo": 3.71568, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.790112596762843}
{"step": 30530, "loss": 3.694152069091797, "grad_norm": 0.7040185332298279, "learning_rate": 0.0001, "elbo": 3.69415, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7904058174994135}
{"step": 30535, "loss": 3.6902278900146483, "grad_norm": 0.6532986164093018, "learning_rate": 0.0001, "elbo": 3.69023, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7906990382359842}
{"step": 30540, "loss": 3.697321319580078, "grad_norm": 0.6740646958351135, "learning_rate": 0.0001, "elbo": 3.69732, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7909922589725547}
{"step": 30545, "loss": 3.697463607788086, "grad_norm": 0.762570321559906, "learning_rate": 0.0001, "elbo": 3.69746, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.791285479709125}
{"step": 30550, "loss": 3.6795879364013673, "grad_norm": 0.6692354083061218, "learning_rate": 0.0001, "elbo": 3.67959, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7915787004456956}
{"step": 30555, "loss": 3.690190887451172, "grad_norm": 0.6783385276794434, "learning_rate": 0.0001, "elbo": 3.69019, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.791871921182266}
{"step": 30560, "loss": 3.7032142639160157, "grad_norm": 0.6966014504432678, "learning_rate": 0.0001, "elbo": 3.70321, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7921651419188365}
{"step": 30565, "loss": 3.7014442443847657, "grad_norm": 0.7111934423446655, "learning_rate": 0.0001, "elbo": 3.70144, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.792458362655407}
{"step": 30570, "loss": 3.713095855712891, "grad_norm": 0.6872096061706543, "learning_rate": 0.0001, "elbo": 3.7131, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7927515833919774}
{"step": 30575, "loss": 3.7120391845703127, "grad_norm": 0.776663064956665, "learning_rate": 0.0001, "elbo": 3.71204, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7930448041285478}
{"step": 30580, "loss": 3.70123291015625, "grad_norm": 0.622157633304596, "learning_rate": 0.0001, "elbo": 3.70123, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7933380248651185}
{"step": 30585, "loss": 3.6777076721191406, "grad_norm": 0.6792261004447937, "learning_rate": 0.0001, "elbo": 3.67771, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.793631245601689}
{"step": 30590, "loss": 3.6844539642333984, "grad_norm": 0.6966149210929871, "learning_rate": 0.0001, "elbo": 3.68445, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7939244663382594}
{"step": 30595, "loss": 3.706584930419922, "grad_norm": 0.8147809505462646, "learning_rate": 0.0001, "elbo": 3.70658, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.79421768707483}
{"step": 30600, "loss": 3.6991928100585936, "grad_norm": 0.7067638039588928, "learning_rate": 0.0001, "elbo": 3.69919, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7945109078114005}
{"step": 30605, "loss": 3.698114776611328, "grad_norm": 0.6783767938613892, "learning_rate": 0.0001, "elbo": 3.69811, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.794804128547971}
{"step": 30610, "loss": 3.6958675384521484, "grad_norm": 0.9600824117660522, "learning_rate": 0.0001, "elbo": 3.69587, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7950973492845415}
{"step": 30615, "loss": 3.7088199615478517, "grad_norm": 0.8064724206924438, "learning_rate": 0.0001, "elbo": 3.70882, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.795390570021112}
{"step": 30615, "loss": 3.7088199615478517, "grad_norm": 0.8064724206924438, "learning_rate": 0.0001, "elbo": 3.70882, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.795390570021112}
{"step": 30620, "loss": 3.7045997619628905, "grad_norm": 0.7106447815895081, "learning_rate": 0.0001, "elbo": 3.7046, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7956837907576824}
{"step": 30625, "loss": 3.6972774505615233, "grad_norm": 0.6632860898971558, "learning_rate": 0.0001, "elbo": 3.69728, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7959770114942528}
{"step": 30630, "loss": 3.6966217041015623, "grad_norm": 0.6234037280082703, "learning_rate": 0.0001, "elbo": 3.69662, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7962702322308233}
{"step": 30635, "loss": 3.7093616485595704, "grad_norm": 0.7175577282905579, "learning_rate": 0.0001, "elbo": 3.70936, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7965634529673937}
{"step": 30640, "loss": 3.719279479980469, "grad_norm": 0.7564655542373657, "learning_rate": 0.0001, "elbo": 3.71928, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7968566737039642}
{"step": 30645, "loss": 3.712264633178711, "grad_norm": 0.9201046824455261, "learning_rate": 0.0001, "elbo": 3.71226, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7971498944405349}
{"step": 30650, "loss": 3.7021812438964843, "grad_norm": 0.9822097420692444, "learning_rate": 0.0001, "elbo": 3.70218, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7974431151771053}
{"step": 30655, "loss": 3.6832794189453124, "grad_norm": 0.6614000797271729, "learning_rate": 0.0001, "elbo": 3.68328, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7977363359136758}
{"step": 30660, "loss": 3.704485321044922, "grad_norm": 0.7564448714256287, "learning_rate": 0.0001, "elbo": 3.70449, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7980295566502464}
{"step": 30665, "loss": 3.687908935546875, "grad_norm": 0.6414250135421753, "learning_rate": 0.0001, "elbo": 3.68791, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.798322777386817}
{"step": 30670, "loss": 3.7037918090820314, "grad_norm": 0.6705465912818909, "learning_rate": 0.0001, "elbo": 3.70379, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7986159981233873}
{"step": 30675, "loss": 3.6975990295410157, "grad_norm": 0.6679965257644653, "learning_rate": 0.0001, "elbo": 3.6976, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7989092188599578}
{"step": 30680, "loss": 3.6981784820556642, "grad_norm": 0.6627581715583801, "learning_rate": 0.0001, "elbo": 3.69818, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7992024395965283}
{"step": 30685, "loss": 3.6900238037109374, "grad_norm": 0.9108510613441467, "learning_rate": 0.0001, "elbo": 3.69002, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7994956603330987}
{"step": 30690, "loss": 3.6846168518066404, "grad_norm": 0.7008099555969238, "learning_rate": 0.0001, "elbo": 3.68462, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7997888810696692}
{"step": 30690, "loss": 3.6846168518066404, "grad_norm": 0.7008099555969238, "learning_rate": 0.0001, "elbo": 3.68462, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.7997888810696692}
{"step": 30695, "loss": 3.6993202209472655, "grad_norm": 0.6789785027503967, "learning_rate": 0.0001, "elbo": 3.69932, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8000821018062396}
{"step": 30700, "loss": 3.6959129333496095, "grad_norm": 0.7050220370292664, "learning_rate": 0.0001, "elbo": 3.69591, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.80037532254281}
{"step": 30705, "loss": 3.7008651733398437, "grad_norm": 0.6397982239723206, "learning_rate": 0.0001, "elbo": 3.70087, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8006685432793808}
{"step": 30710, "loss": 3.683515930175781, "grad_norm": 0.7443875670433044, "learning_rate": 0.0001, "elbo": 3.68352, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8009617640159512}
{"step": 30715, "loss": 3.7003360748291017, "grad_norm": 0.7192009687423706, "learning_rate": 0.0001, "elbo": 3.70034, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8012549847525217}
{"step": 30720, "loss": 3.7049304962158205, "grad_norm": 0.8014876842498779, "learning_rate": 0.0001, "elbo": 3.70493, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8015482054890923}
{"step": 30720, "loss": 3.7049304962158205, "grad_norm": 0.8014876842498779, "learning_rate": 0.0001, "elbo": 3.70493, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8015482054890923}
{"step": 30725, "loss": 3.695219802856445, "grad_norm": 0.6631678938865662, "learning_rate": 0.0001, "elbo": 3.69522, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8018414262256628}
{"step": 30730, "loss": 3.688892364501953, "grad_norm": 0.6818812489509583, "learning_rate": 0.0001, "elbo": 3.68889, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8021346469622332}
{"step": 30735, "loss": 3.7030288696289064, "grad_norm": 0.7203631401062012, "learning_rate": 0.0001, "elbo": 3.70303, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8024278676988037}
{"step": 30740, "loss": 3.7012710571289062, "grad_norm": 0.7013317346572876, "learning_rate": 0.0001, "elbo": 3.70127, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8027210884353742}
{"step": 30745, "loss": 3.7006370544433596, "grad_norm": 0.6948254704475403, "learning_rate": 0.0001, "elbo": 3.70064, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8030143091719446}
{"step": 30750, "loss": 3.6992706298828124, "grad_norm": 0.9049659371376038, "learning_rate": 0.0001, "elbo": 3.69927, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.803307529908515}
{"step": 30755, "loss": 3.6825828552246094, "grad_norm": 0.6672189831733704, "learning_rate": 0.0001, "elbo": 3.68258, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8036007506450855}
{"step": 30760, "loss": 3.6787796020507812, "grad_norm": 0.6585872769355774, "learning_rate": 0.0001, "elbo": 3.67878, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.803893971381656}
{"step": 30765, "loss": 3.6966064453125, "grad_norm": 0.6753554940223694, "learning_rate": 0.0001, "elbo": 3.69661, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8041871921182266}
{"step": 30770, "loss": 3.698749542236328, "grad_norm": 0.710782527923584, "learning_rate": 0.0001, "elbo": 3.69875, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.804480412854797}
{"step": 30775, "loss": 3.6981056213378904, "grad_norm": 0.655164361000061, "learning_rate": 0.0001, "elbo": 3.69811, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8047736335913676}
{"step": 30780, "loss": 3.7064422607421874, "grad_norm": 0.6795175075531006, "learning_rate": 0.0001, "elbo": 3.70644, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8050668543279382}
{"step": 30785, "loss": 3.704926300048828, "grad_norm": 0.700028657913208, "learning_rate": 0.0001, "elbo": 3.70493, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8053600750645087}
{"step": 30790, "loss": 3.7117218017578124, "grad_norm": 0.7107140421867371, "learning_rate": 0.0001, "elbo": 3.71172, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8056532958010791}
{"step": 30795, "loss": 3.709619903564453, "grad_norm": 0.6778162717819214, "learning_rate": 0.0001, "elbo": 3.70962, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8059465165376496}
{"step": 30800, "loss": 3.7057849884033205, "grad_norm": 0.6671990156173706, "learning_rate": 0.0001, "elbo": 3.70578, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.80623973727422}
{"step": 30805, "loss": 3.6915771484375, "grad_norm": 0.7192084193229675, "learning_rate": 0.0001, "elbo": 3.69158, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8065329580107905}
{"step": 30810, "loss": 3.6883506774902344, "grad_norm": 0.7226770520210266, "learning_rate": 0.0001, "elbo": 3.68835, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.806826178747361}
{"step": 30815, "loss": 3.6732719421386717, "grad_norm": 0.6675696969032288, "learning_rate": 0.0001, "elbo": 3.67327, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8071193994839314}
{"step": 30820, "loss": 3.697201156616211, "grad_norm": 0.7659946084022522, "learning_rate": 0.0001, "elbo": 3.6972, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8074126202205019}
{"step": 30825, "loss": 3.7194549560546877, "grad_norm": 0.6916540861129761, "learning_rate": 0.0001, "elbo": 3.71945, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8077058409570723}
{"step": 30830, "loss": 3.7032569885253905, "grad_norm": 0.7087271809577942, "learning_rate": 0.0001, "elbo": 3.70326, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.807999061693643}
{"step": 30835, "loss": 3.703827667236328, "grad_norm": 0.6783905029296875, "learning_rate": 0.0001, "elbo": 3.70383, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8082922824302134}
{"step": 30840, "loss": 3.697775650024414, "grad_norm": 0.7046568393707275, "learning_rate": 0.0001, "elbo": 3.69778, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.808585503166784}
{"step": 30845, "loss": 3.711724853515625, "grad_norm": 0.6859952211380005, "learning_rate": 0.0001, "elbo": 3.71172, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8088787239033546}
{"step": 30850, "loss": 3.7007408142089844, "grad_norm": 0.6880136728286743, "learning_rate": 0.0001, "elbo": 3.70074, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.809171944639925}
{"step": 30855, "loss": 3.6930519104003907, "grad_norm": 0.7107914090156555, "learning_rate": 0.0001, "elbo": 3.69305, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8094651653764955}
{"step": 30860, "loss": 3.685071563720703, "grad_norm": 0.6685307621955872, "learning_rate": 0.0001, "elbo": 3.68507, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.809758386113066}
{"step": 30865, "loss": 3.6876319885253905, "grad_norm": 0.6970500946044922, "learning_rate": 0.0001, "elbo": 3.68763, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8100516068496364}
{"step": 30870, "loss": 3.6985923767089846, "grad_norm": 0.8297634124755859, "learning_rate": 0.0001, "elbo": 3.69859, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8103448275862069}
{"step": 30875, "loss": 3.707825469970703, "grad_norm": 0.7884812951087952, "learning_rate": 0.0001, "elbo": 3.70783, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8106380483227773}
{"step": 30880, "loss": 3.707256317138672, "grad_norm": 0.7730969190597534, "learning_rate": 0.0001, "elbo": 3.70726, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8109312690593478}
{"step": 30885, "loss": 3.6987709045410155, "grad_norm": 0.7739081978797913, "learning_rate": 0.0001, "elbo": 3.69877, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8112244897959182}
{"step": 30890, "loss": 3.701410675048828, "grad_norm": 0.6301332116127014, "learning_rate": 0.0001, "elbo": 3.70141, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.811517710532489}
{"step": 30895, "loss": 3.6946693420410157, "grad_norm": 0.8336200714111328, "learning_rate": 0.0001, "elbo": 3.69467, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8118109312690593}
{"step": 30900, "loss": 3.6907680511474608, "grad_norm": 0.7172161936759949, "learning_rate": 0.0001, "elbo": 3.69077, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8121041520056298}
{"step": 30905, "loss": 3.694806671142578, "grad_norm": 0.8557367920875549, "learning_rate": 0.0001, "elbo": 3.69481, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8123973727422005}
{"step": 30910, "loss": 3.6853458404541017, "grad_norm": 0.797501802444458, "learning_rate": 0.0001, "elbo": 3.68535, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.812690593478771}
{"step": 30915, "loss": 3.691498947143555, "grad_norm": 0.8325551748275757, "learning_rate": 0.0001, "elbo": 3.6915, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8129838142153414}
{"step": 30920, "loss": 3.6786598205566405, "grad_norm": 0.6798485517501831, "learning_rate": 0.0001, "elbo": 3.67866, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8132770349519118}
{"step": 30925, "loss": 3.701888656616211, "grad_norm": 0.6676410436630249, "learning_rate": 0.0001, "elbo": 3.70189, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8135702556884823}
{"step": 30930, "loss": 3.702576446533203, "grad_norm": 0.8174073100090027, "learning_rate": 0.0001, "elbo": 3.70258, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8138634764250527}
{"step": 30935, "loss": 3.6781150817871096, "grad_norm": 0.6816110610961914, "learning_rate": 0.0001, "elbo": 3.67812, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8141566971616232}
{"step": 30940, "loss": 3.706884002685547, "grad_norm": 0.6806924939155579, "learning_rate": 0.0001, "elbo": 3.70688, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8144499178981937}
{"step": 30945, "loss": 3.691201019287109, "grad_norm": 0.7017722725868225, "learning_rate": 0.0001, "elbo": 3.6912, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.814743138634764}
{"step": 30950, "loss": 3.692401885986328, "grad_norm": 0.6956131458282471, "learning_rate": 0.0001, "elbo": 3.6924, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8150363593713348}
{"step": 30955, "loss": 3.6978885650634767, "grad_norm": 0.6638498306274414, "learning_rate": 0.0001, "elbo": 3.69789, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8153295801079052}
{"step": 30955, "loss": 3.6978885650634767, "grad_norm": 0.6638498306274414, "learning_rate": 0.0001, "elbo": 3.69789, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8153295801079052}
{"step": 30960, "loss": 3.684722137451172, "grad_norm": 1.0122690200805664, "learning_rate": 0.0001, "elbo": 3.68472, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8156228008444757}
{"step": 30965, "loss": 3.708769989013672, "grad_norm": 0.640707790851593, "learning_rate": 0.0001, "elbo": 3.70877, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8159160215810464}
{"step": 30970, "loss": 3.6977134704589845, "grad_norm": 0.6873819231987, "learning_rate": 0.0001, "elbo": 3.69771, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8162092423176168}
{"step": 30975, "loss": 3.698779296875, "grad_norm": 0.7407453656196594, "learning_rate": 0.0001, "elbo": 3.69878, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8165024630541873}
{"step": 30980, "loss": 3.6920005798339846, "grad_norm": 0.7015358209609985, "learning_rate": 0.0001, "elbo": 3.692, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8167956837907577}
{"step": 30985, "loss": 3.6904869079589844, "grad_norm": 0.725844144821167, "learning_rate": 0.0001, "elbo": 3.69049, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8170889045273282}
{"step": 30990, "loss": 3.7037590026855467, "grad_norm": 0.6510134935379028, "learning_rate": 0.0001, "elbo": 3.70376, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8173821252638986}
{"step": 30995, "loss": 3.6894420623779296, "grad_norm": 0.9592129588127136, "learning_rate": 0.0001, "elbo": 3.68944, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.817675346000469}
{"step": 31000, "loss": 3.6800804138183594, "grad_norm": 0.7698830366134644, "learning_rate": 0.0001, "elbo": 3.68008, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8179685667370395}
{"step": 31005, "loss": 3.690809631347656, "grad_norm": 0.6617564558982849, "learning_rate": 0.0001, "elbo": 3.69081, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.81826178747361}
{"step": 31010, "loss": 3.6948638916015626, "grad_norm": 0.66000896692276, "learning_rate": 0.0001, "elbo": 3.69486, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8185550082101807}
{"step": 31015, "loss": 3.70716552734375, "grad_norm": 0.6558870077133179, "learning_rate": 0.0001, "elbo": 3.70717, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8188482289467511}
{"step": 31020, "loss": 3.681143951416016, "grad_norm": 0.8048325777053833, "learning_rate": 0.0001, "elbo": 3.68114, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8191414496833216}
{"step": 31025, "loss": 3.698983001708984, "grad_norm": 0.703641951084137, "learning_rate": 0.0001, "elbo": 3.69898, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.819434670419892}
{"step": 31030, "loss": 3.6808841705322264, "grad_norm": 0.6625635623931885, "learning_rate": 0.0001, "elbo": 3.68088, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8197278911564627}
{"step": 31035, "loss": 3.6909000396728517, "grad_norm": 0.73390793800354, "learning_rate": 0.0001, "elbo": 3.6909, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8200211118930332}
{"step": 31040, "loss": 3.678173828125, "grad_norm": 0.7092825770378113, "learning_rate": 0.0001, "elbo": 3.67817, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8203143326296036}
{"step": 31045, "loss": 3.6881973266601564, "grad_norm": 0.6685908436775208, "learning_rate": 0.0001, "elbo": 3.6882, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.820607553366174}
{"step": 31050, "loss": 3.6884296417236326, "grad_norm": 0.6610405445098877, "learning_rate": 0.0001, "elbo": 3.68843, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8209007741027445}
{"step": 31055, "loss": 3.692047882080078, "grad_norm": 0.6984922289848328, "learning_rate": 0.0001, "elbo": 3.69205, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.821193994839315}
{"step": 31060, "loss": 3.699421691894531, "grad_norm": 0.7237630486488342, "learning_rate": 0.0001, "elbo": 3.69942, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8214872155758854}
{"step": 31065, "loss": 3.685131072998047, "grad_norm": 0.9593052268028259, "learning_rate": 0.0001, "elbo": 3.68513, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.821780436312456}
{"step": 31070, "loss": 3.6933860778808594, "grad_norm": 0.7150156497955322, "learning_rate": 0.0001, "elbo": 3.69339, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8220736570490264}
{"step": 31075, "loss": 3.704335403442383, "grad_norm": 0.7507290840148926, "learning_rate": 0.0001, "elbo": 3.70434, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.822366877785597}
{"step": 31080, "loss": 3.701131820678711, "grad_norm": 0.7630584836006165, "learning_rate": 0.0001, "elbo": 3.70113, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8226600985221675}
{"step": 31085, "loss": 3.6979949951171873, "grad_norm": 0.6800640225410461, "learning_rate": 0.0001, "elbo": 3.69799, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.822953319258738}
{"step": 31090, "loss": 3.694010925292969, "grad_norm": 0.7317838072776794, "learning_rate": 0.0001, "elbo": 3.69401, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8232465399953086}
{"step": 31095, "loss": 3.6951530456542967, "grad_norm": 0.6798862218856812, "learning_rate": 0.0001, "elbo": 3.69515, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.823539760731879}
{"step": 31100, "loss": 3.6962024688720705, "grad_norm": 0.6664024591445923, "learning_rate": 0.0001, "elbo": 3.6962, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8238329814684495}
{"step": 31105, "loss": 3.682625961303711, "grad_norm": 0.7489175796508789, "learning_rate": 0.0001, "elbo": 3.68263, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.82412620220502}
{"step": 31110, "loss": 3.692124938964844, "grad_norm": 0.6533073782920837, "learning_rate": 0.0001, "elbo": 3.69213, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8244194229415904}
{"step": 31115, "loss": 3.686446762084961, "grad_norm": 0.8281426429748535, "learning_rate": 0.0001, "elbo": 3.68645, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8247126436781609}
{"step": 31120, "loss": 3.687795639038086, "grad_norm": 0.8304729461669922, "learning_rate": 0.0001, "elbo": 3.6878, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8250058644147313}
{"step": 31120, "loss": 3.687795639038086, "grad_norm": 0.8304729461669922, "learning_rate": 0.0001, "elbo": 3.6878, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8250058644147313}
{"step": 31125, "loss": 3.685558319091797, "grad_norm": 0.6790962219238281, "learning_rate": 0.0001, "elbo": 3.68556, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8252990851513018}
{"step": 31130, "loss": 3.7022384643554687, "grad_norm": 0.6574535965919495, "learning_rate": 0.0001, "elbo": 3.70224, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8255923058878722}
{"step": 31130, "loss": 3.7022384643554687, "grad_norm": 0.6574535965919495, "learning_rate": 0.0001, "elbo": 3.70224, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8255923058878722}
{"step": 31135, "loss": 3.69178466796875, "grad_norm": 0.7818347811698914, "learning_rate": 0.0001, "elbo": 3.69178, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.825885526624443}
{"step": 31140, "loss": 3.691661071777344, "grad_norm": 0.7730271220207214, "learning_rate": 0.0001, "elbo": 3.69166, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8261787473610134}
{"step": 31145, "loss": 3.6945217132568358, "grad_norm": 1.2319756746292114, "learning_rate": 0.0001, "elbo": 3.69452, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8264719680975838}
{"step": 31150, "loss": 3.697534942626953, "grad_norm": 0.6603922247886658, "learning_rate": 0.0001, "elbo": 3.69753, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8267651888341545}
{"step": 31155, "loss": 3.6838401794433593, "grad_norm": 0.7468929886817932, "learning_rate": 0.0001, "elbo": 3.68384, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.827058409570725}
{"step": 31160, "loss": 3.6891128540039064, "grad_norm": 0.8013157248497009, "learning_rate": 0.0001, "elbo": 3.68911, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8273516303072954}
{"step": 31160, "loss": 3.6891128540039064, "grad_norm": 0.8013157248497009, "learning_rate": 0.0001, "elbo": 3.68911, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8273516303072954}
{"step": 31165, "loss": 3.684599685668945, "grad_norm": 0.7876059412956238, "learning_rate": 0.0001, "elbo": 3.6846, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8276448510438659}
{"step": 31170, "loss": 3.6976943969726563, "grad_norm": 0.6923801302909851, "learning_rate": 0.0001, "elbo": 3.69769, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8279380717804363}
{"step": 31175, "loss": 3.7056247711181642, "grad_norm": 1.0383065938949585, "learning_rate": 0.0001, "elbo": 3.70562, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8282312925170068}
{"step": 31180, "loss": 3.6885749816894533, "grad_norm": 0.7079224586486816, "learning_rate": 0.0001, "elbo": 3.68857, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8285245132535772}
{"step": 31185, "loss": 3.7002967834472655, "grad_norm": 0.636330783367157, "learning_rate": 0.0001, "elbo": 3.7003, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8288177339901477}
{"step": 31190, "loss": 3.6920074462890624, "grad_norm": 0.8347453474998474, "learning_rate": 0.0001, "elbo": 3.69201, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8291109547267181}
{"step": 31195, "loss": 3.6817581176757814, "grad_norm": 0.6670769453048706, "learning_rate": 0.0001, "elbo": 3.68176, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8294041754632888}
{"step": 31200, "loss": 3.6996246337890626, "grad_norm": 0.756197988986969, "learning_rate": 0.0001, "elbo": 3.69962, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8296973961998593}
{"step": 31205, "loss": 3.7027122497558596, "grad_norm": 0.8501636385917664, "learning_rate": 0.0001, "elbo": 3.70271, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8299906169364297}
{"step": 31210, "loss": 3.6906410217285157, "grad_norm": 0.7078023552894592, "learning_rate": 0.0001, "elbo": 3.69064, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8302838376730002}
{"step": 31215, "loss": 3.6863311767578124, "grad_norm": 0.7756954431533813, "learning_rate": 0.0001, "elbo": 3.68633, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8305770584095709}
{"step": 31220, "loss": 3.7087547302246096, "grad_norm": 0.6571941375732422, "learning_rate": 0.0001, "elbo": 3.70875, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8308702791461413}
{"step": 31225, "loss": 3.70257453918457, "grad_norm": 0.6974053978919983, "learning_rate": 0.0001, "elbo": 3.70257, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8311634998827118}
{"step": 31230, "loss": 3.706634521484375, "grad_norm": 0.6758177876472473, "learning_rate": 0.0001, "elbo": 3.70663, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8314567206192822}
{"step": 31235, "loss": 3.6795814514160154, "grad_norm": 0.6470054984092712, "learning_rate": 0.0001, "elbo": 3.67958, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8317499413558527}
{"step": 31240, "loss": 3.7088966369628906, "grad_norm": 0.6742562055587769, "learning_rate": 0.0001, "elbo": 3.7089, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8320431620924231}
{"step": 31245, "loss": 3.682831573486328, "grad_norm": 0.6396317481994629, "learning_rate": 0.0001, "elbo": 3.68283, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8323363828289936}
{"step": 31250, "loss": 3.6948062896728517, "grad_norm": 0.6368213295936584, "learning_rate": 0.0001, "elbo": 3.69481, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.832629603565564}
{"step": 31255, "loss": 3.6793670654296875, "grad_norm": 0.7044090628623962, "learning_rate": 0.0001, "elbo": 3.67937, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8329228243021345}
{"step": 31260, "loss": 3.6997604370117188, "grad_norm": 0.8323614597320557, "learning_rate": 0.0001, "elbo": 3.69976, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8332160450387052}
{"step": 31265, "loss": 3.709729766845703, "grad_norm": 0.6930716633796692, "learning_rate": 0.0001, "elbo": 3.70973, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8335092657752756}
{"step": 31270, "loss": 3.6855064392089845, "grad_norm": 0.794977605342865, "learning_rate": 0.0001, "elbo": 3.68551, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.833802486511846}
{"step": 31275, "loss": 3.698068618774414, "grad_norm": 0.6738014817237854, "learning_rate": 0.0001, "elbo": 3.69807, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8340957072484168}
{"step": 31280, "loss": 3.7017669677734375, "grad_norm": 0.6686674952507019, "learning_rate": 0.0001, "elbo": 3.70177, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8343889279849872}
{"step": 31285, "loss": 3.701646423339844, "grad_norm": 0.7129490971565247, "learning_rate": 0.0001, "elbo": 3.70165, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8346821487215577}
{"step": 31290, "loss": 3.678658676147461, "grad_norm": 0.676479697227478, "learning_rate": 0.0001, "elbo": 3.67866, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8349753694581281}
{"step": 31295, "loss": 3.677715301513672, "grad_norm": 0.6545194983482361, "learning_rate": 0.0001, "elbo": 3.67772, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8352685901946986}
{"step": 31300, "loss": 3.6936031341552735, "grad_norm": 0.8264610171318054, "learning_rate": 0.0001, "elbo": 3.6936, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.835561810931269}
{"step": 31305, "loss": 3.6882972717285156, "grad_norm": 0.7243392467498779, "learning_rate": 0.0001, "elbo": 3.6883, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8358550316678395}
{"step": 31310, "loss": 3.6961647033691407, "grad_norm": 0.9886229634284973, "learning_rate": 0.0001, "elbo": 3.69616, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.83614825240441}
{"step": 31315, "loss": 3.707386016845703, "grad_norm": 0.6718140840530396, "learning_rate": 0.0001, "elbo": 3.70739, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8364414731409804}
{"step": 31320, "loss": 3.6927864074707033, "grad_norm": 0.8751794099807739, "learning_rate": 0.0001, "elbo": 3.69279, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.836734693877551}
{"step": 31325, "loss": 3.6878013610839844, "grad_norm": 0.6931636333465576, "learning_rate": 0.0001, "elbo": 3.6878, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8370279146141215}
{"step": 31330, "loss": 3.69288330078125, "grad_norm": 0.6786860227584839, "learning_rate": 0.0001, "elbo": 3.69288, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.837321135350692}
{"step": 31335, "loss": 3.686742401123047, "grad_norm": 0.6831398606300354, "learning_rate": 0.0001, "elbo": 3.68674, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8376143560872626}
{"step": 31340, "loss": 3.681690979003906, "grad_norm": 0.6971137523651123, "learning_rate": 0.0001, "elbo": 3.68169, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.837907576823833}
{"step": 31345, "loss": 3.683274841308594, "grad_norm": 0.7103592157363892, "learning_rate": 0.0001, "elbo": 3.68327, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8382007975604036}
{"step": 31350, "loss": 3.6797554016113283, "grad_norm": 0.6786021590232849, "learning_rate": 0.0001, "elbo": 3.67976, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.838494018296974}
{"step": 31355, "loss": 3.6837669372558595, "grad_norm": 0.7105507254600525, "learning_rate": 0.0001, "elbo": 3.68377, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8387872390335445}
{"step": 31360, "loss": 3.699779510498047, "grad_norm": 1.0302505493164062, "learning_rate": 0.0001, "elbo": 3.69978, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.839080459770115}
{"step": 31365, "loss": 3.681507873535156, "grad_norm": 0.668302059173584, "learning_rate": 0.0001, "elbo": 3.68151, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8393736805066854}
{"step": 31365, "loss": 3.681507873535156, "grad_norm": 0.668302059173584, "learning_rate": 0.0001, "elbo": 3.68151, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8393736805066854}
{"step": 31370, "loss": 3.688719940185547, "grad_norm": 0.6600122451782227, "learning_rate": 0.0001, "elbo": 3.68872, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8396669012432558}
{"step": 31375, "loss": 3.6779022216796875, "grad_norm": 0.6998059749603271, "learning_rate": 0.0001, "elbo": 3.6779, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8399601219798263}
{"step": 31375, "loss": 3.6779022216796875, "grad_norm": 0.6998059749603271, "learning_rate": 0.0001, "elbo": 3.6779, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8399601219798263}
{"step": 31380, "loss": 3.6865116119384767, "grad_norm": 0.626422643661499, "learning_rate": 0.0001, "elbo": 3.68651, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.840253342716397}
{"step": 31380, "loss": 3.6865116119384767, "grad_norm": 0.626422643661499, "learning_rate": 0.0001, "elbo": 3.68651, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.840253342716397}
{"step": 31385, "loss": 3.6915878295898437, "grad_norm": 0.705308735370636, "learning_rate": 0.0001, "elbo": 3.69159, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8405465634529674}
{"step": 31390, "loss": 3.6843379974365233, "grad_norm": 0.6601565480232239, "learning_rate": 0.0001, "elbo": 3.68434, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8408397841895379}
{"step": 31395, "loss": 3.680536651611328, "grad_norm": 0.8105910420417786, "learning_rate": 0.0001, "elbo": 3.68054, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8411330049261085}
{"step": 31400, "loss": 3.68411865234375, "grad_norm": 0.7472572326660156, "learning_rate": 0.0001, "elbo": 3.68412, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.841426225662679}
{"step": 31405, "loss": 3.6845813751220704, "grad_norm": 0.6698651909828186, "learning_rate": 0.0001, "elbo": 3.68458, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8417194463992494}
{"step": 31410, "loss": 3.689125823974609, "grad_norm": 0.6549022197723389, "learning_rate": 0.0001, "elbo": 3.68913, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.84201266713582}
{"step": 31415, "loss": 3.693878936767578, "grad_norm": 0.6523111462593079, "learning_rate": 0.0001, "elbo": 3.69388, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8423058878723904}
{"step": 31420, "loss": 3.688197708129883, "grad_norm": 0.6935461759567261, "learning_rate": 0.0001, "elbo": 3.6882, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8425991086089608}
{"step": 31425, "loss": 3.6919708251953125, "grad_norm": 0.6749153733253479, "learning_rate": 0.0001, "elbo": 3.69197, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8428923293455313}
{"step": 31425, "loss": 3.6919708251953125, "grad_norm": 0.6749153733253479, "learning_rate": 0.0001, "elbo": 3.69197, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8428923293455313}
{"step": 31430, "loss": 3.6792343139648436, "grad_norm": 0.8590372204780579, "learning_rate": 0.0001, "elbo": 3.67923, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8431855500821017}
{"step": 31435, "loss": 3.703314208984375, "grad_norm": 0.6503637433052063, "learning_rate": 0.0001, "elbo": 3.70331, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8434787708186722}
{"step": 31440, "loss": 3.691019058227539, "grad_norm": 0.6569933891296387, "learning_rate": 0.0001, "elbo": 3.69102, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8437719915552426}
{"step": 31445, "loss": 3.700791931152344, "grad_norm": 0.7522424459457397, "learning_rate": 0.0001, "elbo": 3.70079, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8440652122918133}
{"step": 31450, "loss": 3.6925182342529297, "grad_norm": 0.6887251138687134, "learning_rate": 0.0001, "elbo": 3.69252, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8443584330283838}
{"step": 31455, "loss": 3.6835563659667967, "grad_norm": 0.8356119394302368, "learning_rate": 0.0001, "elbo": 3.68356, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8446516537649542}
{"step": 31460, "loss": 3.6887886047363283, "grad_norm": 0.6956956386566162, "learning_rate": 0.0001, "elbo": 3.68879, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.844944874501525}
{"step": 31465, "loss": 3.699940490722656, "grad_norm": 0.7291445136070251, "learning_rate": 0.0001, "elbo": 3.69994, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8452380952380953}
{"step": 31470, "loss": 3.701795959472656, "grad_norm": 0.687446653842926, "learning_rate": 0.0001, "elbo": 3.7018, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8455313159746658}
{"step": 31475, "loss": 3.685491943359375, "grad_norm": 0.6868295073509216, "learning_rate": 0.0001, "elbo": 3.68549, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8458245367112363}
{"step": 31480, "loss": 3.695355987548828, "grad_norm": 0.6688696146011353, "learning_rate": 0.0001, "elbo": 3.69536, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8461177574478067}
{"step": 31485, "loss": 3.6808319091796875, "grad_norm": 1.3385497331619263, "learning_rate": 0.0001, "elbo": 3.68083, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8464109781843772}
{"step": 31490, "loss": 3.6969451904296875, "grad_norm": 0.6453248262405396, "learning_rate": 0.0001, "elbo": 3.69695, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8467041989209476}
{"step": 31495, "loss": 3.69909553527832, "grad_norm": 0.6448096036911011, "learning_rate": 0.0001, "elbo": 3.6991, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.846997419657518}
{"step": 31500, "loss": 3.689505386352539, "grad_norm": 0.7114843130111694, "learning_rate": 0.0001, "elbo": 3.68951, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8472906403940885}
{"step": 31505, "loss": 3.701830291748047, "grad_norm": 0.7175171971321106, "learning_rate": 0.0001, "elbo": 3.70183, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8475838611306592}
{"step": 31510, "loss": 3.692029571533203, "grad_norm": 0.801159679889679, "learning_rate": 0.0001, "elbo": 3.69203, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8478770818672297}
{"step": 31515, "loss": 3.6968460083007812, "grad_norm": 0.7012524604797363, "learning_rate": 0.0001, "elbo": 3.69685, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8481703026038}
{"step": 31520, "loss": 3.6841156005859377, "grad_norm": 0.746539294719696, "learning_rate": 0.0001, "elbo": 3.68412, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8484635233403708}
{"step": 31525, "loss": 3.7026878356933595, "grad_norm": 0.6970893144607544, "learning_rate": 0.0001, "elbo": 3.70269, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8487567440769412}
{"step": 31530, "loss": 3.700330352783203, "grad_norm": 0.6755762100219727, "learning_rate": 0.0001, "elbo": 3.70033, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8490499648135117}
{"step": 31535, "loss": 3.693194580078125, "grad_norm": 0.6449295878410339, "learning_rate": 0.0001, "elbo": 3.69319, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8493431855500821}
{"step": 31540, "loss": 3.693922424316406, "grad_norm": 0.6853922605514526, "learning_rate": 0.0001, "elbo": 3.69392, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8496364062866526}
{"step": 31545, "loss": 3.706523132324219, "grad_norm": 0.6547037959098816, "learning_rate": 0.0001, "elbo": 3.70652, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.849929627023223}
{"step": 31550, "loss": 3.677562713623047, "grad_norm": 0.6528257727622986, "learning_rate": 0.0001, "elbo": 3.67756, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8502228477597935}
{"step": 31555, "loss": 3.6878952026367187, "grad_norm": 0.7007462382316589, "learning_rate": 0.0001, "elbo": 3.6879, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.850516068496364}
{"step": 31560, "loss": 3.6966327667236327, "grad_norm": 0.6736124753952026, "learning_rate": 0.0001, "elbo": 3.69663, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8508092892329344}
{"step": 31565, "loss": 3.7062244415283203, "grad_norm": 0.7240550518035889, "learning_rate": 0.0001, "elbo": 3.70622, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.851102509969505}
{"step": 31570, "loss": 3.6916572570800783, "grad_norm": 0.6910965442657471, "learning_rate": 0.0001, "elbo": 3.69166, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8513957307060755}
{"step": 31575, "loss": 3.699738311767578, "grad_norm": 0.7099609375, "learning_rate": 0.0001, "elbo": 3.69974, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.851688951442646}
{"step": 31580, "loss": 3.692474365234375, "grad_norm": 0.7250389456748962, "learning_rate": 0.0001, "elbo": 3.69247, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8519821721792167}
{"step": 31585, "loss": 3.6983619689941407, "grad_norm": 0.7214223742485046, "learning_rate": 0.0001, "elbo": 3.69836, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8522753929157871}
{"step": 31590, "loss": 3.687698745727539, "grad_norm": 0.6536020636558533, "learning_rate": 0.0001, "elbo": 3.6877, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8525686136523576}
{"step": 31590, "loss": 3.687698745727539, "grad_norm": 0.6536020636558533, "learning_rate": 0.0001, "elbo": 3.6877, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8525686136523576}
{"step": 31595, "loss": 3.6868526458740236, "grad_norm": 0.8312976360321045, "learning_rate": 0.0001, "elbo": 3.68685, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.852861834388928}
{"step": 31600, "loss": 3.704359436035156, "grad_norm": 0.7958881258964539, "learning_rate": 0.0001, "elbo": 3.70436, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8531550551254985}
{"step": 31605, "loss": 3.6810317993164063, "grad_norm": 0.6846121549606323, "learning_rate": 0.0001, "elbo": 3.68103, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.853448275862069}
{"step": 31610, "loss": 3.6983070373535156, "grad_norm": 0.7252102494239807, "learning_rate": 0.0001, "elbo": 3.69831, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8537414965986394}
{"step": 31615, "loss": 3.7004871368408203, "grad_norm": 0.7008658647537231, "learning_rate": 0.0001, "elbo": 3.70049, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8540347173352099}
{"step": 31620, "loss": 3.6942203521728514, "grad_norm": 0.7692431211471558, "learning_rate": 0.0001, "elbo": 3.69422, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8543279380717803}
{"step": 31625, "loss": 3.7042816162109373, "grad_norm": 0.6811144948005676, "learning_rate": 0.0001, "elbo": 3.70428, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8546211588083508}
{"step": 31630, "loss": 3.6820758819580077, "grad_norm": 0.6788560152053833, "learning_rate": 0.0001, "elbo": 3.68208, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8549143795449214}
{"step": 31635, "loss": 3.6896034240722657, "grad_norm": 0.6888316869735718, "learning_rate": 0.0001, "elbo": 3.6896, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.855207600281492}
{"step": 31640, "loss": 3.69617919921875, "grad_norm": 0.8836999535560608, "learning_rate": 0.0001, "elbo": 3.69618, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8555008210180624}
{"step": 31645, "loss": 3.678369140625, "grad_norm": 0.6758325099945068, "learning_rate": 0.0001, "elbo": 3.67837, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.855794041754633}
{"step": 31650, "loss": 3.715981292724609, "grad_norm": 0.7430837154388428, "learning_rate": 0.0001, "elbo": 3.71598, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8560872624912035}
{"step": 31655, "loss": 3.6875038146972656, "grad_norm": 0.6944015026092529, "learning_rate": 0.0001, "elbo": 3.6875, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.856380483227774}
{"step": 31660, "loss": 3.6867019653320314, "grad_norm": 0.7450085282325745, "learning_rate": 0.0001, "elbo": 3.6867, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8566737039643444}
{"step": 31665, "loss": 3.67998046875, "grad_norm": 0.6875435709953308, "learning_rate": 0.0001, "elbo": 3.67998, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8569669247009148}
{"step": 31670, "loss": 3.6973102569580076, "grad_norm": 0.7046023607254028, "learning_rate": 0.0001, "elbo": 3.69731, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8572601454374853}
{"step": 31675, "loss": 3.6872650146484376, "grad_norm": 0.6545157432556152, "learning_rate": 0.0001, "elbo": 3.68727, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8575533661740558}
{"step": 31680, "loss": 3.689888763427734, "grad_norm": 0.688408374786377, "learning_rate": 0.0001, "elbo": 3.68989, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8578465869106262}
{"step": 31685, "loss": 3.7025741577148437, "grad_norm": 0.6714021563529968, "learning_rate": 0.0001, "elbo": 3.70257, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8581398076471967}
{"step": 31690, "loss": 3.693616485595703, "grad_norm": 0.7810441255569458, "learning_rate": 0.0001, "elbo": 3.69362, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8584330283837673}
{"step": 31695, "loss": 3.6954235076904296, "grad_norm": 0.7976071238517761, "learning_rate": 0.0001, "elbo": 3.69542, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8587262491203378}
{"step": 31700, "loss": 3.688649368286133, "grad_norm": 0.6955488920211792, "learning_rate": 0.0001, "elbo": 3.68865, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8590194698569082}
{"step": 31705, "loss": 3.692399597167969, "grad_norm": 0.6699473857879639, "learning_rate": 0.0001, "elbo": 3.6924, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.859312690593479}
{"step": 31710, "loss": 3.68273811340332, "grad_norm": 0.6606197953224182, "learning_rate": 0.0001, "elbo": 3.68274, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8596059113300494}
{"step": 31715, "loss": 3.6913528442382812, "grad_norm": 0.6758285164833069, "learning_rate": 0.0001, "elbo": 3.69135, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8598991320666198}
{"step": 31720, "loss": 3.6722766876220705, "grad_norm": 0.7837507724761963, "learning_rate": 0.0001, "elbo": 3.67228, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8601923528031903}
{"step": 31725, "loss": 3.6890167236328124, "grad_norm": 0.6237053275108337, "learning_rate": 0.0001, "elbo": 3.68902, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8604855735397607}
{"step": 31730, "loss": 3.6851474761962892, "grad_norm": 0.7131959795951843, "learning_rate": 0.0001, "elbo": 3.68515, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8607787942763312}
{"step": 31735, "loss": 3.686100387573242, "grad_norm": 0.6733719110488892, "learning_rate": 0.0001, "elbo": 3.6861, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8610720150129016}
{"step": 31740, "loss": 3.6870986938476564, "grad_norm": 0.6981197595596313, "learning_rate": 0.0001, "elbo": 3.6871, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.861365235749472}
{"step": 31745, "loss": 3.6812389373779295, "grad_norm": 0.6372933387756348, "learning_rate": 0.0001, "elbo": 3.68124, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8616584564860426}
{"step": 31750, "loss": 3.6989036560058595, "grad_norm": 0.7202262878417969, "learning_rate": 0.0001, "elbo": 3.6989, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8619516772226132}
{"step": 31755, "loss": 3.6798439025878906, "grad_norm": 0.6614653468132019, "learning_rate": 0.0001, "elbo": 3.67984, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8622448979591837}
{"step": 31760, "loss": 3.6934303283691405, "grad_norm": 0.6963610649108887, "learning_rate": 0.0001, "elbo": 3.69343, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8625381186957541}
{"step": 31765, "loss": 3.701601028442383, "grad_norm": 0.7594153881072998, "learning_rate": 0.0001, "elbo": 3.7016, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8628313394323248}
{"step": 31770, "loss": 3.6994991302490234, "grad_norm": 0.7335891723632812, "learning_rate": 0.0001, "elbo": 3.6995, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8631245601688953}
{"step": 31775, "loss": 3.6724227905273437, "grad_norm": 0.7330710887908936, "learning_rate": 0.0001, "elbo": 3.67242, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8634177809054657}
{"step": 31780, "loss": 3.6838512420654297, "grad_norm": 0.9100380539894104, "learning_rate": 0.0001, "elbo": 3.68385, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8637110016420362}
{"step": 31785, "loss": 3.7187850952148436, "grad_norm": 1.0032243728637695, "learning_rate": 0.0001, "elbo": 3.71879, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8640042223786066}
{"step": 31790, "loss": 3.6907142639160155, "grad_norm": 0.6941338777542114, "learning_rate": 0.0001, "elbo": 3.69071, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.864297443115177}
{"step": 31795, "loss": 3.6968925476074217, "grad_norm": 0.6880545020103455, "learning_rate": 0.0001, "elbo": 3.69689, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8645906638517475}
{"step": 31800, "loss": 3.678848648071289, "grad_norm": 0.6596537232398987, "learning_rate": 0.0001, "elbo": 3.67885, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.864883884588318}
{"step": 31805, "loss": 3.687329864501953, "grad_norm": 0.7384575009346008, "learning_rate": 0.0001, "elbo": 3.68733, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8651771053248885}
{"step": 31810, "loss": 3.68074951171875, "grad_norm": 1.001582384109497, "learning_rate": 0.0001, "elbo": 3.68075, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.865470326061459}
{"step": 31815, "loss": 3.683475112915039, "grad_norm": 0.6837384700775146, "learning_rate": 0.0001, "elbo": 3.68348, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8657635467980296}
{"step": 31820, "loss": 3.6958274841308594, "grad_norm": 0.7253831028938293, "learning_rate": 0.0001, "elbo": 3.69583, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8660567675346}
{"step": 31825, "loss": 3.6975460052490234, "grad_norm": 0.6719316244125366, "learning_rate": 0.0001, "elbo": 3.69755, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8663499882711705}
{"step": 31830, "loss": 3.696246337890625, "grad_norm": 0.7103719711303711, "learning_rate": 0.0001, "elbo": 3.69625, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8666432090077412}
{"step": 31835, "loss": 3.699810791015625, "grad_norm": 0.6762964725494385, "learning_rate": 0.0001, "elbo": 3.69981, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8669364297443116}
{"step": 31840, "loss": 3.699473571777344, "grad_norm": 0.6608568429946899, "learning_rate": 0.0001, "elbo": 3.69947, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.867229650480882}
{"step": 31845, "loss": 3.684354782104492, "grad_norm": 0.6779990792274475, "learning_rate": 0.0001, "elbo": 3.68435, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8675228712174525}
{"step": 31850, "loss": 3.6975631713867188, "grad_norm": 0.7048454284667969, "learning_rate": 0.0001, "elbo": 3.69756, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.867816091954023}
{"step": 31855, "loss": 3.6979713439941406, "grad_norm": 0.7495022416114807, "learning_rate": 0.0001, "elbo": 3.69797, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8681093126905934}
{"step": 31860, "loss": 3.6756629943847656, "grad_norm": 0.6727875471115112, "learning_rate": 0.0001, "elbo": 3.67566, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.868402533427164}
{"step": 31865, "loss": 3.691209411621094, "grad_norm": 0.7186588644981384, "learning_rate": 0.0001, "elbo": 3.69121, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8686957541637343}
{"step": 31870, "loss": 3.6697113037109377, "grad_norm": 0.6768386363983154, "learning_rate": 0.0001, "elbo": 3.66971, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8689889749003048}
{"step": 31875, "loss": 3.7058174133300783, "grad_norm": 0.6755174994468689, "learning_rate": 0.0001, "elbo": 3.70582, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8692821956368755}
{"step": 31880, "loss": 3.6751129150390627, "grad_norm": 1.0485799312591553, "learning_rate": 0.0001, "elbo": 3.67511, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.869575416373446}
{"step": 31885, "loss": 3.6795555114746095, "grad_norm": 0.7770252227783203, "learning_rate": 0.0001, "elbo": 3.67956, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8698686371100164}
{"step": 31890, "loss": 3.6967800140380858, "grad_norm": 0.6705238223075867, "learning_rate": 0.0001, "elbo": 3.69678, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.870161857846587}
{"step": 31895, "loss": 3.693593215942383, "grad_norm": 0.6656087636947632, "learning_rate": 0.0001, "elbo": 3.69359, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8704550785831575}
{"step": 31900, "loss": 3.69656982421875, "grad_norm": 0.6928821802139282, "learning_rate": 0.0001, "elbo": 3.69657, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.870748299319728}
{"step": 31905, "loss": 3.6888591766357424, "grad_norm": 0.757413923740387, "learning_rate": 0.0001, "elbo": 3.68886, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8710415200562984}
{"step": 31910, "loss": 3.691455078125, "grad_norm": 0.9235156178474426, "learning_rate": 0.0001, "elbo": 3.69145, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8713347407928689}
{"step": 31915, "loss": 3.6911582946777344, "grad_norm": 0.6960132122039795, "learning_rate": 0.0001, "elbo": 3.69116, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8716279615294393}
{"step": 31920, "loss": 3.687416839599609, "grad_norm": 1.0113146305084229, "learning_rate": 0.0001, "elbo": 3.68742, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8719211822660098}
{"step": 31925, "loss": 3.6893104553222655, "grad_norm": 1.0251295566558838, "learning_rate": 0.0001, "elbo": 3.68931, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8722144030025802}
{"step": 31930, "loss": 3.692210388183594, "grad_norm": 0.7311880588531494, "learning_rate": 0.0001, "elbo": 3.69221, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8725076237391507}
{"step": 31935, "loss": 3.6800975799560547, "grad_norm": 0.6466522812843323, "learning_rate": 0.0001, "elbo": 3.6801, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8728008444757214}
{"step": 31940, "loss": 3.6812507629394533, "grad_norm": 0.6639318466186523, "learning_rate": 0.0001, "elbo": 3.68125, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8730940652122918}
{"step": 31945, "loss": 3.6869693756103517, "grad_norm": 0.6607695817947388, "learning_rate": 0.0001, "elbo": 3.68697, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8733872859488623}
{"step": 31950, "loss": 3.7006404876708983, "grad_norm": 0.653731644153595, "learning_rate": 0.0001, "elbo": 3.70064, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.873680506685433}
{"step": 31955, "loss": 3.673936462402344, "grad_norm": 0.7048368453979492, "learning_rate": 0.0001, "elbo": 3.67394, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8739737274220034}
{"step": 31960, "loss": 3.689897155761719, "grad_norm": 0.6795819997787476, "learning_rate": 0.0001, "elbo": 3.6899, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8742669481585739}
{"step": 31965, "loss": 3.69195442199707, "grad_norm": 0.6628897190093994, "learning_rate": 0.0001, "elbo": 3.69195, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8745601688951443}
{"step": 31970, "loss": 3.677826690673828, "grad_norm": 0.6810483932495117, "learning_rate": 0.0001, "elbo": 3.67783, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8748533896317148}
{"step": 31975, "loss": 3.6900650024414063, "grad_norm": 0.6673094034194946, "learning_rate": 0.0001, "elbo": 3.69006, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8751466103682852}
{"step": 31980, "loss": 3.6853538513183595, "grad_norm": 0.7417028546333313, "learning_rate": 0.0001, "elbo": 3.68535, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8754398311048557}
{"step": 31985, "loss": 3.6898338317871096, "grad_norm": 0.7158295512199402, "learning_rate": 0.0001, "elbo": 3.68983, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8757330518414261}
{"step": 31990, "loss": 3.6808792114257813, "grad_norm": 0.8420235514640808, "learning_rate": 0.0001, "elbo": 3.68088, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8760262725779966}
{"step": 31995, "loss": 3.6972885131835938, "grad_norm": 0.7041975259780884, "learning_rate": 0.0001, "elbo": 3.69729, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.876319493314567}
{"step": 32000, "loss": 3.695513916015625, "grad_norm": 0.7197346091270447, "learning_rate": 0.0001, "elbo": 3.69551, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8766127140511377}
{"step": 32005, "loss": 3.683658218383789, "grad_norm": 0.6784564852714539, "learning_rate": 0.0001, "elbo": 3.68366, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8769059347877082}
{"step": 32010, "loss": 3.6680671691894533, "grad_norm": 0.69577956199646, "learning_rate": 0.0001, "elbo": 3.66807, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8771991555242786}
{"step": 32015, "loss": 3.6744483947753905, "grad_norm": 0.668953001499176, "learning_rate": 0.0001, "elbo": 3.67445, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8774923762608493}
{"step": 32020, "loss": 3.6764808654785157, "grad_norm": 0.7015529274940491, "learning_rate": 0.0001, "elbo": 3.67648, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8777855969974198}
{"step": 32025, "loss": 3.691828155517578, "grad_norm": 0.6709591746330261, "learning_rate": 0.0001, "elbo": 3.69183, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8780788177339902}
{"step": 32030, "loss": 3.6880226135253906, "grad_norm": 0.7095426321029663, "learning_rate": 0.0001, "elbo": 3.68802, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8783720384705607}
{"step": 32035, "loss": 3.692604827880859, "grad_norm": 0.6747917532920837, "learning_rate": 0.0001, "elbo": 3.6926, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8786652592071311}
{"step": 32040, "loss": 3.6876502990722657, "grad_norm": 0.6321196556091309, "learning_rate": 0.0001, "elbo": 3.68765, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8789584799437016}
{"step": 32045, "loss": 3.687249755859375, "grad_norm": 0.6989695429801941, "learning_rate": 0.0001, "elbo": 3.68725, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.879251700680272}
{"step": 32050, "loss": 3.6791748046875, "grad_norm": 0.7456755042076111, "learning_rate": 0.0001, "elbo": 3.67918, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8795449214168425}
{"step": 32055, "loss": 3.6694522857666017, "grad_norm": 0.8030752539634705, "learning_rate": 0.0001, "elbo": 3.66945, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.879838142153413}
{"step": 32060, "loss": 3.6789451599121095, "grad_norm": 0.6467189788818359, "learning_rate": 0.0001, "elbo": 3.67895, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8801313628899836}
{"step": 32065, "loss": 3.6951847076416016, "grad_norm": 0.6495165824890137, "learning_rate": 0.0001, "elbo": 3.69518, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.880424583626554}
{"step": 32070, "loss": 3.6713836669921873, "grad_norm": 0.6518495082855225, "learning_rate": 0.0001, "elbo": 3.67138, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8807178043631245}
{"step": 32075, "loss": 3.6840633392333983, "grad_norm": 0.7041192054748535, "learning_rate": 0.0001, "elbo": 3.68406, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8810110250996952}
{"step": 32075, "loss": 3.6840633392333983, "grad_norm": 0.7041192054748535, "learning_rate": 0.0001, "elbo": 3.68406, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8810110250996952}
{"step": 32080, "loss": 3.702825164794922, "grad_norm": 0.6641871333122253, "learning_rate": 0.0001, "elbo": 3.70283, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8813042458362657}
{"step": 32085, "loss": 3.6751083374023437, "grad_norm": 0.7557423710823059, "learning_rate": 0.0001, "elbo": 3.67511, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.881597466572836}
{"step": 32090, "loss": 3.6839447021484375, "grad_norm": 0.7188184261322021, "learning_rate": 0.0001, "elbo": 3.68394, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8818906873094066}
{"step": 32095, "loss": 3.6790340423583983, "grad_norm": 0.6811396479606628, "learning_rate": 0.0001, "elbo": 3.67903, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.882183908045977}
{"step": 32095, "loss": 3.6790340423583983, "grad_norm": 0.6811396479606628, "learning_rate": 0.0001, "elbo": 3.67903, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.882183908045977}
{"step": 32100, "loss": 3.667920684814453, "grad_norm": 0.6679922342300415, "learning_rate": 0.0001, "elbo": 3.66792, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8824771287825475}
{"step": 32105, "loss": 3.678577423095703, "grad_norm": 0.7461054921150208, "learning_rate": 0.0001, "elbo": 3.67858, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.882770349519118}
{"step": 32110, "loss": 3.673394775390625, "grad_norm": 0.6810847520828247, "learning_rate": 0.0001, "elbo": 3.67339, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8830635702556884}
{"step": 32115, "loss": 3.699420928955078, "grad_norm": 0.6348441243171692, "learning_rate": 0.0001, "elbo": 3.69942, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8833567909922588}
{"step": 32120, "loss": 3.6878013610839844, "grad_norm": 0.6868067383766174, "learning_rate": 0.0001, "elbo": 3.6878, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8836500117288295}
{"step": 32125, "loss": 3.686880111694336, "grad_norm": 0.6710139513015747, "learning_rate": 0.0001, "elbo": 3.68688, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8839432324654}
{"step": 32130, "loss": 3.6853179931640625, "grad_norm": 1.0233471393585205, "learning_rate": 0.0001, "elbo": 3.68532, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8842364532019704}
{"step": 32135, "loss": 3.6902606964111326, "grad_norm": 0.6874721646308899, "learning_rate": 0.0001, "elbo": 3.69026, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.884529673938541}
{"step": 32140, "loss": 3.685655975341797, "grad_norm": 0.7096312642097473, "learning_rate": 0.0001, "elbo": 3.68566, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8848228946751115}
{"step": 32145, "loss": 3.6774200439453124, "grad_norm": 0.655449628829956, "learning_rate": 0.0001, "elbo": 3.67742, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.885116115411682}
{"step": 32150, "loss": 3.695499038696289, "grad_norm": 0.7762380242347717, "learning_rate": 0.0001, "elbo": 3.6955, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8854093361482525}
{"step": 32155, "loss": 3.695358657836914, "grad_norm": 0.7371513843536377, "learning_rate": 0.0001, "elbo": 3.69536, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.885702556884823}
{"step": 32160, "loss": 3.675324249267578, "grad_norm": 0.675728976726532, "learning_rate": 0.0001, "elbo": 3.67532, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8859957776213934}
{"step": 32165, "loss": 3.6859336853027345, "grad_norm": 0.6900577545166016, "learning_rate": 0.0001, "elbo": 3.68593, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8862889983579638}
{"step": 32170, "loss": 3.6719234466552733, "grad_norm": 0.6497055888175964, "learning_rate": 0.0001, "elbo": 3.67192, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8865822190945343}
{"step": 32170, "loss": 3.6719234466552733, "grad_norm": 0.6497055888175964, "learning_rate": 0.0001, "elbo": 3.67192, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8865822190945343}
{"step": 32175, "loss": 3.6859642028808595, "grad_norm": 0.6703240275382996, "learning_rate": 0.0001, "elbo": 3.68596, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8868754398311047}
{"step": 32180, "loss": 3.6825935363769533, "grad_norm": 0.6597579121589661, "learning_rate": 0.0001, "elbo": 3.68259, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8871686605676752}
{"step": 32185, "loss": 3.684264373779297, "grad_norm": 0.7024396061897278, "learning_rate": 0.0001, "elbo": 3.68426, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8874618813042459}
{"step": 32190, "loss": 3.6891361236572267, "grad_norm": 0.6596884727478027, "learning_rate": 0.0001, "elbo": 3.68914, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8877551020408163}
{"step": 32195, "loss": 3.702106475830078, "grad_norm": 0.7133128643035889, "learning_rate": 0.0001, "elbo": 3.70211, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8880483227773868}
{"step": 32200, "loss": 3.6891410827636717, "grad_norm": 0.6869097948074341, "learning_rate": 0.0001, "elbo": 3.68914, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8883415435139574}
{"step": 32205, "loss": 3.688526916503906, "grad_norm": 0.7116197347640991, "learning_rate": 0.0001, "elbo": 3.68853, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.888634764250528}
{"step": 32210, "loss": 3.6882362365722656, "grad_norm": 0.8969101905822754, "learning_rate": 0.0001, "elbo": 3.68824, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8889279849870984}
{"step": 32215, "loss": 3.6806907653808594, "grad_norm": 0.6488211750984192, "learning_rate": 0.0001, "elbo": 3.68069, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8892212057236688}
{"step": 32220, "loss": 3.6879257202148437, "grad_norm": 0.6561161279678345, "learning_rate": 0.0001, "elbo": 3.68793, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8895144264602393}
{"step": 32225, "loss": 3.6983909606933594, "grad_norm": 0.7218600511550903, "learning_rate": 0.0001, "elbo": 3.69839, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8898076471968097}
{"step": 32225, "loss": 3.6983909606933594, "grad_norm": 0.7218600511550903, "learning_rate": 0.0001, "elbo": 3.69839, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8898076471968097}
{"step": 32230, "loss": 3.675736999511719, "grad_norm": 0.6573998332023621, "learning_rate": 0.0001, "elbo": 3.67574, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8901008679333802}
{"step": 32230, "loss": 3.675736999511719, "grad_norm": 0.6573998332023621, "learning_rate": 0.0001, "elbo": 3.67574, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8901008679333802}
{"step": 32235, "loss": 3.677870178222656, "grad_norm": 0.72135990858078, "learning_rate": 0.0001, "elbo": 3.67787, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8903940886699506}
{"step": 32240, "loss": 3.687041091918945, "grad_norm": 0.6840125918388367, "learning_rate": 0.0001, "elbo": 3.68704, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.890687309406521}
{"step": 32245, "loss": 3.681927490234375, "grad_norm": 1.4782869815826416, "learning_rate": 0.0001, "elbo": 3.68193, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8909805301430918}
{"step": 32245, "loss": 3.681927490234375, "grad_norm": 1.4782869815826416, "learning_rate": 0.0001, "elbo": 3.68193, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8909805301430918}
{"step": 32250, "loss": 3.6959407806396483, "grad_norm": 0.7164073586463928, "learning_rate": 0.0001, "elbo": 3.69594, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8912737508796622}
{"step": 32255, "loss": 3.682146453857422, "grad_norm": 0.9341564774513245, "learning_rate": 0.0001, "elbo": 3.68215, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8915669716162327}
{"step": 32260, "loss": 3.6776046752929688, "grad_norm": 0.7407835125923157, "learning_rate": 0.0001, "elbo": 3.6776, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8918601923528033}
{"step": 32260, "loss": 3.6776046752929688, "grad_norm": 0.7407835125923157, "learning_rate": 0.0001, "elbo": 3.6776, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8918601923528033}
{"step": 32265, "loss": 3.6902442932128907, "grad_norm": 0.6838314533233643, "learning_rate": 0.0001, "elbo": 3.69024, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8921534130893738}
{"step": 32270, "loss": 3.6796627044677734, "grad_norm": 1.2265170812606812, "learning_rate": 0.0001, "elbo": 3.67966, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8924466338259442}
{"step": 32275, "loss": 3.673171615600586, "grad_norm": 0.7915748953819275, "learning_rate": 0.0001, "elbo": 3.67317, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8927398545625147}
{"step": 32280, "loss": 3.705240249633789, "grad_norm": 0.6697495579719543, "learning_rate": 0.0001, "elbo": 3.70524, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8930330752990852}
{"step": 32285, "loss": 3.672161102294922, "grad_norm": 0.6616902947425842, "learning_rate": 0.0001, "elbo": 3.67216, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8933262960356556}
{"step": 32290, "loss": 3.685801315307617, "grad_norm": 0.6505992412567139, "learning_rate": 0.0001, "elbo": 3.6858, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.893619516772226}
{"step": 32295, "loss": 3.6727859497070314, "grad_norm": 0.7797828316688538, "learning_rate": 0.0001, "elbo": 3.67279, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8939127375087965}
{"step": 32300, "loss": 3.666523742675781, "grad_norm": 0.7001346349716187, "learning_rate": 0.0001, "elbo": 3.66652, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.894205958245367}
{"step": 32305, "loss": 3.6902923583984375, "grad_norm": 1.061946988105774, "learning_rate": 0.0001, "elbo": 3.69029, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8944991789819376}
{"step": 32310, "loss": 3.6888404846191407, "grad_norm": 0.9380984902381897, "learning_rate": 0.0001, "elbo": 3.68884, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.894792399718508}
{"step": 32315, "loss": 3.7006153106689452, "grad_norm": 0.7057780027389526, "learning_rate": 0.0001, "elbo": 3.70062, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8950856204550786}
{"step": 32320, "loss": 3.6899890899658203, "grad_norm": 0.6968879699707031, "learning_rate": 0.0001, "elbo": 3.68999, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8953788411916492}
{"step": 32325, "loss": 3.6856971740722657, "grad_norm": 0.8023983240127563, "learning_rate": 0.0001, "elbo": 3.6857, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8956720619282197}
{"step": 32330, "loss": 3.6869056701660154, "grad_norm": 0.6775975823402405, "learning_rate": 0.0001, "elbo": 3.68691, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8959652826647901}
{"step": 32335, "loss": 3.693888854980469, "grad_norm": 0.8025798201560974, "learning_rate": 0.0001, "elbo": 3.69389, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8962585034013606}
{"step": 32340, "loss": 3.679828643798828, "grad_norm": 0.7029505968093872, "learning_rate": 0.0001, "elbo": 3.67983, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.896551724137931}
{"step": 32345, "loss": 3.6817726135253905, "grad_norm": 0.6905443072319031, "learning_rate": 0.0001, "elbo": 3.68177, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8968449448745015}
{"step": 32350, "loss": 3.682665252685547, "grad_norm": 0.7465608716011047, "learning_rate": 0.0001, "elbo": 3.68267, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.897138165611072}
{"step": 32355, "loss": 3.6784347534179687, "grad_norm": 0.6805530786514282, "learning_rate": 0.0001, "elbo": 3.67843, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8974313863476424}
{"step": 32360, "loss": 3.668177032470703, "grad_norm": 0.8250078558921814, "learning_rate": 0.0001, "elbo": 3.66818, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8977246070842129}
{"step": 32365, "loss": 3.6986469268798827, "grad_norm": 1.0475081205368042, "learning_rate": 0.0001, "elbo": 3.69865, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8980178278207833}
{"step": 32370, "loss": 3.690187454223633, "grad_norm": 0.7229022979736328, "learning_rate": 0.0001, "elbo": 3.69019, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.898311048557354}
{"step": 32375, "loss": 3.7046714782714845, "grad_norm": 0.6529688835144043, "learning_rate": 0.0001, "elbo": 3.70467, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8986042692939245}
{"step": 32380, "loss": 3.677326965332031, "grad_norm": 0.7050229907035828, "learning_rate": 0.0001, "elbo": 3.67733, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.898897490030495}
{"step": 32385, "loss": 3.6905548095703127, "grad_norm": 0.7662063241004944, "learning_rate": 0.0001, "elbo": 3.69055, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8991907107670656}
{"step": 32390, "loss": 3.675606918334961, "grad_norm": 0.6854925155639648, "learning_rate": 0.0001, "elbo": 3.67561, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.899483931503636}
{"step": 32395, "loss": 3.677197265625, "grad_norm": 0.7316964864730835, "learning_rate": 0.0001, "elbo": 3.6772, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.8997771522402065}
{"step": 32400, "loss": 3.691203308105469, "grad_norm": 0.6848078966140747, "learning_rate": 0.0001, "elbo": 3.6912, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.900070372976777}
{"step": 32405, "loss": 3.6867050170898437, "grad_norm": 0.8022966980934143, "learning_rate": 0.0001, "elbo": 3.68671, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9003635937133474}
{"step": 32410, "loss": 3.6774757385253904, "grad_norm": 0.6937098503112793, "learning_rate": 0.0001, "elbo": 3.67748, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9006568144499179}
{"step": 32415, "loss": 3.691514587402344, "grad_norm": 0.685036301612854, "learning_rate": 0.0001, "elbo": 3.69151, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9009500351864883}
{"step": 32420, "loss": 3.68432502746582, "grad_norm": 0.6611775755882263, "learning_rate": 0.0001, "elbo": 3.68432, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9012432559230588}
{"step": 32425, "loss": 3.685951995849609, "grad_norm": 0.6580711603164673, "learning_rate": 0.0001, "elbo": 3.68595, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9015364766596292}
{"step": 32430, "loss": 3.688536834716797, "grad_norm": 0.7416119575500488, "learning_rate": 0.0001, "elbo": 3.68854, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9018296973962}
{"step": 32435, "loss": 3.667893981933594, "grad_norm": 0.6624107956886292, "learning_rate": 0.0001, "elbo": 3.66789, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9021229181327703}
{"step": 32440, "loss": 3.693421173095703, "grad_norm": 0.7057851552963257, "learning_rate": 0.0001, "elbo": 3.69342, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9024161388693408}
{"step": 32445, "loss": 3.681378936767578, "grad_norm": 0.7538774609565735, "learning_rate": 0.0001, "elbo": 3.68138, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9027093596059115}
{"step": 32450, "loss": 3.673401641845703, "grad_norm": 0.6978302001953125, "learning_rate": 0.0001, "elbo": 3.6734, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.903002580342482}
{"step": 32455, "loss": 3.670759582519531, "grad_norm": 0.6396006345748901, "learning_rate": 0.0001, "elbo": 3.67076, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9032958010790524}
{"step": 32460, "loss": 3.6965091705322264, "grad_norm": 0.7441954016685486, "learning_rate": 0.0001, "elbo": 3.69651, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9035890218156228}
{"step": 32465, "loss": 3.6847877502441406, "grad_norm": 0.7024065256118774, "learning_rate": 0.0001, "elbo": 3.68479, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9038822425521933}
{"step": 32470, "loss": 3.684090423583984, "grad_norm": 0.6455911993980408, "learning_rate": 0.0001, "elbo": 3.68409, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9041754632887637}
{"step": 32475, "loss": 3.699040985107422, "grad_norm": 0.6924193501472473, "learning_rate": 0.0001, "elbo": 3.69904, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9044686840253342}
{"step": 32480, "loss": 3.684116744995117, "grad_norm": 0.6828393340110779, "learning_rate": 0.0001, "elbo": 3.68412, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9047619047619047}
{"step": 32485, "loss": 3.684051513671875, "grad_norm": 0.749018132686615, "learning_rate": 0.0001, "elbo": 3.68405, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.905055125498475}
{"step": 32490, "loss": 3.6886768341064453, "grad_norm": 0.8203020095825195, "learning_rate": 0.0001, "elbo": 3.68868, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9053483462350458}
{"step": 32490, "loss": 3.6886768341064453, "grad_norm": 0.8203020095825195, "learning_rate": 0.0001, "elbo": 3.68868, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9053483462350458}
{"step": 32495, "loss": 3.6845630645751952, "grad_norm": 0.8447836637496948, "learning_rate": 0.0001, "elbo": 3.68456, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9056415669716162}
{"step": 32500, "loss": 3.6810874938964844, "grad_norm": 0.7558081150054932, "learning_rate": 0.0001, "elbo": 3.68109, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9059347877081867}
{"step": 32505, "loss": 3.6932693481445313, "grad_norm": 0.7347850799560547, "learning_rate": 0.0001, "elbo": 3.69327, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9062280084447574}
{"step": 32510, "loss": 3.689303970336914, "grad_norm": 0.6942797303199768, "learning_rate": 0.0001, "elbo": 3.6893, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9065212291813278}
{"step": 32515, "loss": 3.6799552917480467, "grad_norm": 0.6767096519470215, "learning_rate": 0.0001, "elbo": 3.67996, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9068144499178983}
{"step": 32520, "loss": 3.6932701110839843, "grad_norm": 0.6678875684738159, "learning_rate": 0.0001, "elbo": 3.69327, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9071076706544687}
{"step": 32525, "loss": 3.688035583496094, "grad_norm": 0.6422039866447449, "learning_rate": 0.0001, "elbo": 3.68804, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9074008913910392}
{"step": 32530, "loss": 3.670941162109375, "grad_norm": 0.6778724193572998, "learning_rate": 0.0001, "elbo": 3.67094, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9076941121276096}
{"step": 32535, "loss": 3.6830459594726563, "grad_norm": 0.7226493954658508, "learning_rate": 0.0001, "elbo": 3.68305, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.90798733286418}
{"step": 32540, "loss": 3.6815032958984375, "grad_norm": 0.6511216759681702, "learning_rate": 0.0001, "elbo": 3.6815, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9082805536007506}
{"step": 32545, "loss": 3.687216567993164, "grad_norm": 0.7439978718757629, "learning_rate": 0.0001, "elbo": 3.68722, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.908573774337321}
{"step": 32550, "loss": 3.6814506530761717, "grad_norm": 0.6531837582588196, "learning_rate": 0.0001, "elbo": 3.68145, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9088669950738915}
{"step": 32555, "loss": 3.682697296142578, "grad_norm": 0.6971700191497803, "learning_rate": 0.0001, "elbo": 3.6827, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9091602158104621}
{"step": 32560, "loss": 3.694294738769531, "grad_norm": 0.7173938155174255, "learning_rate": 0.0001, "elbo": 3.69429, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9094534365470326}
{"step": 32565, "loss": 3.664556884765625, "grad_norm": 0.6544646620750427, "learning_rate": 0.0001, "elbo": 3.66456, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.909746657283603}
{"step": 32570, "loss": 3.6873779296875, "grad_norm": 0.8562985062599182, "learning_rate": 0.0001, "elbo": 3.68738, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9100398780201737}
{"step": 32575, "loss": 3.679779815673828, "grad_norm": 0.6716787219047546, "learning_rate": 0.0001, "elbo": 3.67978, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9103330987567442}
{"step": 32580, "loss": 3.6809925079345702, "grad_norm": 0.7192299962043762, "learning_rate": 0.0001, "elbo": 3.68099, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9106263194933146}
{"step": 32585, "loss": 3.6689125061035157, "grad_norm": 0.711829662322998, "learning_rate": 0.0001, "elbo": 3.66891, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.910919540229885}
{"step": 32590, "loss": 3.701909637451172, "grad_norm": 0.6525172591209412, "learning_rate": 0.0001, "elbo": 3.70191, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9112127609664555}
{"step": 32595, "loss": 3.671311950683594, "grad_norm": 0.6728836894035339, "learning_rate": 0.0001, "elbo": 3.67131, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.911505981703026}
{"step": 32600, "loss": 3.6770347595214843, "grad_norm": 0.7258356213569641, "learning_rate": 0.0001, "elbo": 3.67703, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9117992024395964}
{"step": 32605, "loss": 3.700043869018555, "grad_norm": 0.7676746845245361, "learning_rate": 0.0001, "elbo": 3.70004, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.912092423176167}
{"step": 32610, "loss": 3.6796600341796877, "grad_norm": 0.6421129703521729, "learning_rate": 0.0001, "elbo": 3.67966, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9123856439127374}
{"step": 32615, "loss": 3.6890674591064454, "grad_norm": 0.6828446388244629, "learning_rate": 0.0001, "elbo": 3.68907, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.912678864649308}
{"step": 32620, "loss": 3.6680072784423827, "grad_norm": 0.6962683796882629, "learning_rate": 0.0001, "elbo": 3.66801, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9129720853858785}
{"step": 32625, "loss": 3.6722129821777343, "grad_norm": 0.710420548915863, "learning_rate": 0.0001, "elbo": 3.67221, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.913265306122449}
{"step": 32630, "loss": 3.685121536254883, "grad_norm": 0.7206467390060425, "learning_rate": 0.0001, "elbo": 3.68512, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9135585268590196}
{"step": 32635, "loss": 3.6737194061279297, "grad_norm": 0.75121009349823, "learning_rate": 0.0001, "elbo": 3.67372, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.91385174759559}
{"step": 32640, "loss": 3.6714691162109374, "grad_norm": 0.6914449334144592, "learning_rate": 0.0001, "elbo": 3.67147, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9141449683321605}
{"step": 32645, "loss": 3.6883888244628906, "grad_norm": 0.6568549275398254, "learning_rate": 0.0001, "elbo": 3.68839, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.914438189068731}
{"step": 32650, "loss": 3.6974258422851562, "grad_norm": 0.65706866979599, "learning_rate": 0.0001, "elbo": 3.69743, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9147314098053014}
{"step": 32655, "loss": 3.700115966796875, "grad_norm": 0.7162982821464539, "learning_rate": 0.0001, "elbo": 3.70012, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9150246305418719}
{"step": 32660, "loss": 3.688763427734375, "grad_norm": 0.6373910903930664, "learning_rate": 0.0001, "elbo": 3.68876, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9153178512784423}
{"step": 32665, "loss": 3.677703857421875, "grad_norm": 0.6666772961616516, "learning_rate": 0.0001, "elbo": 3.6777, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9156110720150128}
{"step": 32670, "loss": 3.6815261840820312, "grad_norm": 0.6494107842445374, "learning_rate": 0.0001, "elbo": 3.68153, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9159042927515832}
{"step": 32675, "loss": 3.676758575439453, "grad_norm": 0.716414749622345, "learning_rate": 0.0001, "elbo": 3.67676, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.916197513488154}
{"step": 32680, "loss": 3.6811897277832033, "grad_norm": 0.8446125388145447, "learning_rate": 0.0001, "elbo": 3.68119, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9164907342247244}
{"step": 32685, "loss": 3.6824501037597654, "grad_norm": 0.664107084274292, "learning_rate": 0.0001, "elbo": 3.68245, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9167839549612948}
{"step": 32690, "loss": 3.6903671264648437, "grad_norm": 0.6901662349700928, "learning_rate": 0.0001, "elbo": 3.69037, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9170771756978655}
{"step": 32695, "loss": 3.698334503173828, "grad_norm": 0.6880364418029785, "learning_rate": 0.0001, "elbo": 3.69833, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.917370396434436}
{"step": 32700, "loss": 3.68992919921875, "grad_norm": 0.7458416223526001, "learning_rate": 0.0001, "elbo": 3.68993, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9176636171710064}
{"step": 32705, "loss": 3.6830375671386717, "grad_norm": 0.6992241144180298, "learning_rate": 0.0001, "elbo": 3.68304, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9179568379075769}
{"step": 32710, "loss": 3.6705650329589843, "grad_norm": 0.7620919942855835, "learning_rate": 0.0001, "elbo": 3.67057, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9182500586441473}
{"step": 32715, "loss": 3.667570877075195, "grad_norm": 0.6717955470085144, "learning_rate": 0.0001, "elbo": 3.66757, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9185432793807178}
{"step": 32720, "loss": 3.686220169067383, "grad_norm": 0.6704885363578796, "learning_rate": 0.0001, "elbo": 3.68622, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9188365001172882}
{"step": 32725, "loss": 3.6758087158203123, "grad_norm": 0.6800435781478882, "learning_rate": 0.0001, "elbo": 3.67581, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9191297208538587}
{"step": 32725, "loss": 3.6758087158203123, "grad_norm": 0.6800435781478882, "learning_rate": 0.0001, "elbo": 3.67581, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9191297208538587}
{"step": 32730, "loss": 3.6768077850341796, "grad_norm": 0.7376020550727844, "learning_rate": 0.0001, "elbo": 3.67681, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9194229415904291}
{"step": 32730, "loss": 3.6768077850341796, "grad_norm": 0.7376020550727844, "learning_rate": 0.0001, "elbo": 3.67681, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9194229415904291}
{"step": 32735, "loss": 3.689606475830078, "grad_norm": 0.7205204963684082, "learning_rate": 0.0001, "elbo": 3.68961, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9197161623269998}
{"step": 32740, "loss": 3.686473846435547, "grad_norm": 0.7747229337692261, "learning_rate": 0.0001, "elbo": 3.68647, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9200093830635703}
{"step": 32745, "loss": 3.6986358642578123, "grad_norm": 0.7709062099456787, "learning_rate": 0.0001, "elbo": 3.69864, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9203026038001407}
{"step": 32750, "loss": 3.675060272216797, "grad_norm": 0.7777467966079712, "learning_rate": 0.0001, "elbo": 3.67506, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9205958245367112}
{"step": 32755, "loss": 3.677991485595703, "grad_norm": 0.6600674986839294, "learning_rate": 0.0001, "elbo": 3.67799, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9208890452732819}
{"step": 32760, "loss": 3.6873443603515623, "grad_norm": 0.6950809359550476, "learning_rate": 0.0001, "elbo": 3.68734, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9211822660098523}
{"step": 32765, "loss": 3.683641815185547, "grad_norm": 0.6806111335754395, "learning_rate": 0.0001, "elbo": 3.68364, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9214754867464228}
{"step": 32770, "loss": 3.6853736877441405, "grad_norm": 0.7106856107711792, "learning_rate": 0.0001, "elbo": 3.68537, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9217687074829932}
{"step": 32775, "loss": 3.67752685546875, "grad_norm": 0.6611842513084412, "learning_rate": 0.0001, "elbo": 3.67753, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9220619282195637}
{"step": 32780, "loss": 3.6851837158203127, "grad_norm": 0.7468295097351074, "learning_rate": 0.0001, "elbo": 3.68518, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9223551489561341}
{"step": 32785, "loss": 3.670826721191406, "grad_norm": 0.6723650097846985, "learning_rate": 0.0001, "elbo": 3.67083, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9226483696927046}
{"step": 32790, "loss": 3.6779937744140625, "grad_norm": 0.747785210609436, "learning_rate": 0.0001, "elbo": 3.67799, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.922941590429275}
{"step": 32795, "loss": 3.675019454956055, "grad_norm": 0.6928221583366394, "learning_rate": 0.0001, "elbo": 3.67502, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9232348111658455}
{"step": 32800, "loss": 3.673572540283203, "grad_norm": 0.6783943176269531, "learning_rate": 0.0001, "elbo": 3.67357, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9235280319024162}
{"step": 32800, "loss": 3.673572540283203, "grad_norm": 0.6783943176269531, "learning_rate": 0.0001, "elbo": 3.67357, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9235280319024162}
{"step": 32805, "loss": 3.6840248107910156, "grad_norm": 0.6988323926925659, "learning_rate": 0.0001, "elbo": 3.68402, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9238212526389866}
{"step": 32810, "loss": 3.693072509765625, "grad_norm": 0.7229409217834473, "learning_rate": 0.0001, "elbo": 3.69307, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.924114473375557}
{"step": 32815, "loss": 3.684667205810547, "grad_norm": 0.6850411295890808, "learning_rate": 0.0001, "elbo": 3.68467, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9244076941121278}
{"step": 32820, "loss": 3.681381607055664, "grad_norm": 0.7007459998130798, "learning_rate": 0.0001, "elbo": 3.68138, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9247009148486982}
{"step": 32825, "loss": 3.6928802490234376, "grad_norm": 0.7006580233573914, "learning_rate": 0.0001, "elbo": 3.69288, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9249941355852687}
{"step": 32825, "loss": 3.6928802490234376, "grad_norm": 0.7006580233573914, "learning_rate": 0.0001, "elbo": 3.69288, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9249941355852687}
{"step": 32830, "loss": 3.6921886444091796, "grad_norm": 1.125211477279663, "learning_rate": 0.0001, "elbo": 3.69219, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9252873563218391}
{"step": 32835, "loss": 3.674510192871094, "grad_norm": 0.6874749660491943, "learning_rate": 0.0001, "elbo": 3.67451, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9255805770584096}
{"step": 32840, "loss": 3.680755615234375, "grad_norm": 0.7653980255126953, "learning_rate": 0.0001, "elbo": 3.68076, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.92587379779498}
{"step": 32845, "loss": 3.683573913574219, "grad_norm": 0.7047487497329712, "learning_rate": 0.0001, "elbo": 3.68357, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9261670185315505}
{"step": 32850, "loss": 3.683188629150391, "grad_norm": 0.73146653175354, "learning_rate": 0.0001, "elbo": 3.68319, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.926460239268121}
{"step": 32855, "loss": 3.676237106323242, "grad_norm": 0.7424218654632568, "learning_rate": 0.0001, "elbo": 3.67624, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9267534600046914}
{"step": 32860, "loss": 3.684772491455078, "grad_norm": 0.7132139801979065, "learning_rate": 0.0001, "elbo": 3.68477, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.927046680741262}
{"step": 32865, "loss": 3.6727508544921874, "grad_norm": 0.639495313167572, "learning_rate": 0.0001, "elbo": 3.67275, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9273399014778325}
{"step": 32865, "loss": 3.6727508544921874, "grad_norm": 0.639495313167572, "learning_rate": 0.0001, "elbo": 3.67275, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9273399014778325}
{"step": 32870, "loss": 3.6727367401123048, "grad_norm": 0.6505774855613708, "learning_rate": 0.0001, "elbo": 3.67274, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.927633122214403}
{"step": 32875, "loss": 3.6767951965332033, "grad_norm": 0.6885858178138733, "learning_rate": 0.0001, "elbo": 3.6768, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9279263429509736}
{"step": 32880, "loss": 3.6762123107910156, "grad_norm": 0.648596465587616, "learning_rate": 0.0001, "elbo": 3.67621, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.928219563687544}
{"step": 32885, "loss": 3.6721122741699217, "grad_norm": 0.6910949945449829, "learning_rate": 0.0001, "elbo": 3.67211, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9285127844241146}
{"step": 32890, "loss": 3.668013000488281, "grad_norm": 0.6577825546264648, "learning_rate": 0.0001, "elbo": 3.66801, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.928806005160685}
{"step": 32895, "loss": 3.6802101135253906, "grad_norm": 0.6872907876968384, "learning_rate": 0.0001, "elbo": 3.68021, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9290992258972555}
{"step": 32900, "loss": 3.68890495300293, "grad_norm": 0.7059410810470581, "learning_rate": 0.0001, "elbo": 3.68891, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.929392446633826}
{"step": 32905, "loss": 3.6904327392578127, "grad_norm": 0.690963089466095, "learning_rate": 0.0001, "elbo": 3.69043, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9296856673703964}
{"step": 32910, "loss": 3.6757625579833983, "grad_norm": 0.6864739656448364, "learning_rate": 0.0001, "elbo": 3.67576, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9299788881069668}
{"step": 32915, "loss": 3.671040725708008, "grad_norm": 0.6742397546768188, "learning_rate": 0.0001, "elbo": 3.67104, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9302721088435373}
{"step": 32915, "loss": 3.671040725708008, "grad_norm": 0.6742397546768188, "learning_rate": 0.0001, "elbo": 3.67104, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9302721088435373}
{"step": 32920, "loss": 3.6701499938964846, "grad_norm": 0.7008321285247803, "learning_rate": 0.0001, "elbo": 3.67015, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.930565329580108}
{"step": 32925, "loss": 3.6877426147460937, "grad_norm": 0.6679773330688477, "learning_rate": 0.0001, "elbo": 3.68774, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9308585503166784}
{"step": 32930, "loss": 3.6804340362548826, "grad_norm": 0.7140520811080933, "learning_rate": 0.0001, "elbo": 3.68043, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9311517710532489}
{"step": 32935, "loss": 3.6890727996826174, "grad_norm": 0.6793986558914185, "learning_rate": 0.0001, "elbo": 3.68907, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9314449917898193}
{"step": 32940, "loss": 3.680817794799805, "grad_norm": 0.6867206692695618, "learning_rate": 0.0001, "elbo": 3.68082, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.93173821252639}
{"step": 32945, "loss": 3.6841968536376952, "grad_norm": 0.699036717414856, "learning_rate": 0.0001, "elbo": 3.6842, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9320314332629605}
{"step": 32950, "loss": 3.661941909790039, "grad_norm": 0.6679801940917969, "learning_rate": 0.0001, "elbo": 3.66194, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.932324653999531}
{"step": 32955, "loss": 3.666341781616211, "grad_norm": 0.7279940843582153, "learning_rate": 0.0001, "elbo": 3.66634, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9326178747361014}
{"step": 32960, "loss": 3.6686164855957033, "grad_norm": 0.7195355892181396, "learning_rate": 0.0001, "elbo": 3.66862, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9329110954726718}
{"step": 32965, "loss": 3.677655029296875, "grad_norm": 0.6628729104995728, "learning_rate": 0.0001, "elbo": 3.67765, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9332043162092423}
{"step": 32970, "loss": 3.673321533203125, "grad_norm": 0.669638991355896, "learning_rate": 0.0001, "elbo": 3.67332, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9334975369458127}
{"step": 32975, "loss": 3.6752044677734377, "grad_norm": 0.6946051120758057, "learning_rate": 0.0001, "elbo": 3.6752, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9337907576823832}
{"step": 32980, "loss": 3.6781494140625, "grad_norm": 0.705566942691803, "learning_rate": 0.0001, "elbo": 3.67815, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9340839784189536}
{"step": 32985, "loss": 3.6748947143554687, "grad_norm": 0.7139321565628052, "learning_rate": 0.0001, "elbo": 3.67489, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9343771991555243}
{"step": 32990, "loss": 3.688751983642578, "grad_norm": 0.7152113914489746, "learning_rate": 0.0001, "elbo": 3.68875, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9346704198920948}
{"step": 32995, "loss": 3.682392883300781, "grad_norm": 1.1532995700836182, "learning_rate": 0.0001, "elbo": 3.68239, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9349636406286652}
{"step": 33000, "loss": 3.682115936279297, "grad_norm": 0.6664125323295593, "learning_rate": 0.0001, "elbo": 3.68212, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.935256861365236}
{"step": 33005, "loss": 3.6774593353271485, "grad_norm": 0.7052487134933472, "learning_rate": 0.0001, "elbo": 3.67746, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9355500821018063}
{"step": 33010, "loss": 3.6722286224365233, "grad_norm": 0.6650271415710449, "learning_rate": 0.0001, "elbo": 3.67223, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9358433028383768}
{"step": 33010, "loss": 3.6722286224365233, "grad_norm": 0.6650271415710449, "learning_rate": 0.0001, "elbo": 3.67223, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9358433028383768}
{"step": 33015, "loss": 3.683946228027344, "grad_norm": 0.6962343454360962, "learning_rate": 0.0001, "elbo": 3.68395, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9361365235749473}
{"step": 33020, "loss": 3.6743682861328124, "grad_norm": 0.6967703104019165, "learning_rate": 0.0001, "elbo": 3.67437, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9364297443115177}
{"step": 33025, "loss": 3.6847476959228516, "grad_norm": 0.8247206211090088, "learning_rate": 0.0001, "elbo": 3.68475, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9367229650480882}
{"step": 33030, "loss": 3.6756343841552734, "grad_norm": 0.7016980051994324, "learning_rate": 0.0001, "elbo": 3.67563, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9370161857846586}
{"step": 33035, "loss": 3.668534851074219, "grad_norm": 0.6686939001083374, "learning_rate": 0.0001, "elbo": 3.66854, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.937309406521229}
{"step": 33040, "loss": 3.687810516357422, "grad_norm": 0.6765949726104736, "learning_rate": 0.0001, "elbo": 3.68781, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9376026272577995}
{"step": 33045, "loss": 3.671308898925781, "grad_norm": 0.7243484258651733, "learning_rate": 0.0001, "elbo": 3.67131, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9378958479943702}
{"step": 33050, "loss": 3.683917236328125, "grad_norm": 0.6801052093505859, "learning_rate": 0.0001, "elbo": 3.68392, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9381890687309407}
{"step": 33055, "loss": 3.6875232696533202, "grad_norm": 0.6482564210891724, "learning_rate": 0.0001, "elbo": 3.68752, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.938482289467511}
{"step": 33060, "loss": 3.684299850463867, "grad_norm": 0.6577437520027161, "learning_rate": 0.0001, "elbo": 3.6843, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9387755102040818}
{"step": 33065, "loss": 3.6923789978027344, "grad_norm": 0.7783685326576233, "learning_rate": 0.0001, "elbo": 3.69238, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9390687309406522}
{"step": 33070, "loss": 3.6751480102539062, "grad_norm": 0.7373371720314026, "learning_rate": 0.0001, "elbo": 3.67515, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9393619516772227}
{"step": 33075, "loss": 3.6768417358398438, "grad_norm": 0.7204836010932922, "learning_rate": 0.0001, "elbo": 3.67684, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9396551724137931}
{"step": 33080, "loss": 3.6713539123535157, "grad_norm": 0.7557433843612671, "learning_rate": 0.0001, "elbo": 3.67135, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9399483931503636}
{"step": 33085, "loss": 3.692781448364258, "grad_norm": 0.7031204104423523, "learning_rate": 0.0001, "elbo": 3.69278, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.940241613886934}
{"step": 33090, "loss": 3.6832912445068358, "grad_norm": 0.6642540693283081, "learning_rate": 0.0001, "elbo": 3.68329, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9405348346235045}
{"step": 33095, "loss": 3.666047286987305, "grad_norm": 0.7143093943595886, "learning_rate": 0.0001, "elbo": 3.66605, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.940828055360075}
{"step": 33100, "loss": 3.6867721557617186, "grad_norm": 0.7995412945747375, "learning_rate": 0.0001, "elbo": 3.68677, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9411212760966454}
{"step": 33105, "loss": 3.6823963165283202, "grad_norm": 0.676844596862793, "learning_rate": 0.0001, "elbo": 3.6824, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.941414496833216}
{"step": 33110, "loss": 3.688766860961914, "grad_norm": 0.7944626212120056, "learning_rate": 0.0001, "elbo": 3.68877, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9417077175697866}
{"step": 33115, "loss": 3.6712799072265625, "grad_norm": 0.7414699792861938, "learning_rate": 0.0001, "elbo": 3.67128, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.942000938306357}
{"step": 33120, "loss": 3.673333740234375, "grad_norm": 0.6687224507331848, "learning_rate": 0.0001, "elbo": 3.67333, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9422941590429277}
{"step": 33125, "loss": 3.684461212158203, "grad_norm": 0.6780551075935364, "learning_rate": 0.0001, "elbo": 3.68446, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9425873797794981}
{"step": 33125, "loss": 3.684461212158203, "grad_norm": 0.6780551075935364, "learning_rate": 0.0001, "elbo": 3.68446, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9425873797794981}
{"step": 33130, "loss": 3.6731185913085938, "grad_norm": 0.7043036222457886, "learning_rate": 0.0001, "elbo": 3.67312, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9428806005160686}
{"step": 33130, "loss": 3.6731185913085938, "grad_norm": 0.7043036222457886, "learning_rate": 0.0001, "elbo": 3.67312, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9428806005160686}
{"step": 33135, "loss": 3.672467041015625, "grad_norm": 0.6744744181632996, "learning_rate": 0.0001, "elbo": 3.67247, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.943173821252639}
{"step": 33140, "loss": 3.695831298828125, "grad_norm": 1.182124376296997, "learning_rate": 0.0001, "elbo": 3.69583, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9434670419892095}
{"step": 33140, "loss": 3.695831298828125, "grad_norm": 1.182124376296997, "learning_rate": 0.0001, "elbo": 3.69583, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9434670419892095}
{"step": 33145, "loss": 3.684061813354492, "grad_norm": 1.016150951385498, "learning_rate": 0.0001, "elbo": 3.68406, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.94376026272578}
{"step": 33145, "loss": 3.684061813354492, "grad_norm": 1.016150951385498, "learning_rate": 0.0001, "elbo": 3.68406, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.94376026272578}
{"step": 33150, "loss": 3.6869670867919924, "grad_norm": 0.6638813614845276, "learning_rate": 0.0001, "elbo": 3.68697, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9440534834623504}
{"step": 33150, "loss": 3.6869670867919924, "grad_norm": 0.6638813614845276, "learning_rate": 0.0001, "elbo": 3.68697, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9440534834623504}
{"step": 33155, "loss": 3.6667716979980467, "grad_norm": 0.692557692527771, "learning_rate": 0.0001, "elbo": 3.66677, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9443467041989209}
{"step": 33160, "loss": 3.6791912078857423, "grad_norm": 0.7403072714805603, "learning_rate": 0.0001, "elbo": 3.67919, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9446399249354913}
{"step": 33165, "loss": 3.678233337402344, "grad_norm": 0.6809210777282715, "learning_rate": 0.0001, "elbo": 3.67823, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9449331456720618}
{"step": 33170, "loss": 3.6819198608398436, "grad_norm": 0.7072117328643799, "learning_rate": 0.0001, "elbo": 3.68192, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9452263664086324}
{"step": 33175, "loss": 3.677043151855469, "grad_norm": 0.6801334023475647, "learning_rate": 0.0001, "elbo": 3.67704, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.945519587145203}
{"step": 33180, "loss": 3.667963409423828, "grad_norm": 0.6583510637283325, "learning_rate": 0.0001, "elbo": 3.66796, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9458128078817734}
{"step": 33180, "loss": 3.667963409423828, "grad_norm": 0.6583510637283325, "learning_rate": 0.0001, "elbo": 3.66796, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9458128078817734}
{"step": 33185, "loss": 3.694048309326172, "grad_norm": 0.6769023537635803, "learning_rate": 0.0001, "elbo": 3.69405, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.946106028618344}
{"step": 33190, "loss": 3.674546813964844, "grad_norm": 0.6952552199363708, "learning_rate": 0.0001, "elbo": 3.67455, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9463992493549145}
{"step": 33190, "loss": 3.674546813964844, "grad_norm": 0.6952552199363708, "learning_rate": 0.0001, "elbo": 3.67455, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9463992493549145}
{"step": 33195, "loss": 3.6813758850097655, "grad_norm": 0.7823296785354614, "learning_rate": 0.0001, "elbo": 3.68138, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.946692470091485}
{"step": 33200, "loss": 3.671297073364258, "grad_norm": 0.7178097367286682, "learning_rate": 0.0001, "elbo": 3.6713, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9469856908280554}
{"step": 33205, "loss": 3.6795913696289064, "grad_norm": 0.6529417037963867, "learning_rate": 0.0001, "elbo": 3.67959, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9472789115646258}
{"step": 33210, "loss": 3.6673419952392576, "grad_norm": 0.6427898406982422, "learning_rate": 0.0001, "elbo": 3.66734, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9475721323011963}
{"step": 33215, "loss": 3.684803771972656, "grad_norm": 0.7342950701713562, "learning_rate": 0.0001, "elbo": 3.6848, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9478653530377668}
{"step": 33220, "loss": 3.67037353515625, "grad_norm": 0.6582282781600952, "learning_rate": 0.0001, "elbo": 3.67037, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9481585737743372}
{"step": 33225, "loss": 3.6833576202392577, "grad_norm": 0.7282240986824036, "learning_rate": 0.0001, "elbo": 3.68336, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9484517945109077}
{"step": 33230, "loss": 3.680724334716797, "grad_norm": 0.6913998126983643, "learning_rate": 0.0001, "elbo": 3.68072, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9487450152474783}
{"step": 33235, "loss": 3.674866485595703, "grad_norm": 0.6805766224861145, "learning_rate": 0.0001, "elbo": 3.67487, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9490382359840488}
{"step": 33240, "loss": 3.6905860900878906, "grad_norm": 0.6742064356803894, "learning_rate": 0.0001, "elbo": 3.69059, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9493314567206192}
{"step": 33245, "loss": 3.6698028564453127, "grad_norm": 0.6556422114372253, "learning_rate": 0.0001, "elbo": 3.6698, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.94962467745719}
{"step": 33250, "loss": 3.657522964477539, "grad_norm": 0.7204962372779846, "learning_rate": 0.0001, "elbo": 3.65752, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9499178981937604}
{"step": 33255, "loss": 3.685175323486328, "grad_norm": 0.6770569682121277, "learning_rate": 0.0001, "elbo": 3.68517, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9502111189303308}
{"step": 33260, "loss": 3.671861267089844, "grad_norm": 0.6721147298812866, "learning_rate": 0.0001, "elbo": 3.67186, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9505043396669013}
{"step": 33265, "loss": 3.6742523193359373, "grad_norm": 0.7472786903381348, "learning_rate": 0.0001, "elbo": 3.67425, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9507975604034717}
{"step": 33270, "loss": 3.6872276306152343, "grad_norm": 0.6999160647392273, "learning_rate": 0.0001, "elbo": 3.68723, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9510907811400422}
{"step": 33275, "loss": 3.6517494201660154, "grad_norm": 0.6699225306510925, "learning_rate": 0.0001, "elbo": 3.65175, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9513840018766127}
{"step": 33280, "loss": 3.6744274139404296, "grad_norm": 0.6868268251419067, "learning_rate": 0.0001, "elbo": 3.67443, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.951677222613183}
{"step": 33285, "loss": 3.6875492095947267, "grad_norm": 0.6851426959037781, "learning_rate": 0.0001, "elbo": 3.68755, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9519704433497536}
{"step": 33290, "loss": 3.6667388916015624, "grad_norm": 0.6804162263870239, "learning_rate": 0.0001, "elbo": 3.66674, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9522636640863242}
{"step": 33295, "loss": 3.6647274017333986, "grad_norm": 0.6595794558525085, "learning_rate": 0.0001, "elbo": 3.66473, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9525568848228947}
{"step": 33300, "loss": 3.664128875732422, "grad_norm": 0.6993559002876282, "learning_rate": 0.0001, "elbo": 3.66413, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9528501055594651}
{"step": 33305, "loss": 3.6775138854980467, "grad_norm": 0.897383987903595, "learning_rate": 0.0001, "elbo": 3.67751, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9531433262960358}
{"step": 33310, "loss": 3.6604377746582033, "grad_norm": 0.7158296704292297, "learning_rate": 0.0001, "elbo": 3.66044, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9534365470326063}
{"step": 33310, "loss": 3.6604377746582033, "grad_norm": 0.7158296704292297, "learning_rate": 0.0001, "elbo": 3.66044, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9534365470326063}
{"step": 33315, "loss": 3.687693405151367, "grad_norm": 0.6687071919441223, "learning_rate": 0.0001, "elbo": 3.68769, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9537297677691767}
{"step": 33315, "loss": 3.687693405151367, "grad_norm": 0.6687071919441223, "learning_rate": 0.0001, "elbo": 3.68769, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9537297677691767}
{"step": 33320, "loss": 3.6884227752685548, "grad_norm": 0.6723653078079224, "learning_rate": 0.0001, "elbo": 3.68842, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9540229885057472}
{"step": 33325, "loss": 3.6693752288818358, "grad_norm": 0.7140800952911377, "learning_rate": 0.0001, "elbo": 3.66938, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9543162092423176}
{"step": 33330, "loss": 3.67908935546875, "grad_norm": 0.7794111371040344, "learning_rate": 0.0001, "elbo": 3.67909, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.954609429978888}
{"step": 33335, "loss": 3.6790390014648438, "grad_norm": 0.6695323586463928, "learning_rate": 0.0001, "elbo": 3.67904, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9549026507154585}
{"step": 33340, "loss": 3.663465118408203, "grad_norm": 0.6949347257614136, "learning_rate": 0.0001, "elbo": 3.66347, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.955195871452029}
{"step": 33345, "loss": 3.6693447113037108, "grad_norm": 0.6315310001373291, "learning_rate": 0.0001, "elbo": 3.66934, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9554890921885995}
{"step": 33350, "loss": 3.6877059936523438, "grad_norm": 0.7217144966125488, "learning_rate": 0.0001, "elbo": 3.68771, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.95578231292517}
{"step": 33355, "loss": 3.6785850524902344, "grad_norm": 1.102860927581787, "learning_rate": 0.0001, "elbo": 3.67858, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9560755336617406}
{"step": 33360, "loss": 3.6765384674072266, "grad_norm": 0.6816049814224243, "learning_rate": 0.0001, "elbo": 3.67654, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.956368754398311}
{"step": 33365, "loss": 3.665282440185547, "grad_norm": 0.7684990763664246, "learning_rate": 0.0001, "elbo": 3.66528, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9566619751348815}
{"step": 33370, "loss": 3.680833435058594, "grad_norm": 0.6680199503898621, "learning_rate": 0.0001, "elbo": 3.68083, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9569551958714522}
{"step": 33375, "loss": 3.6618408203125, "grad_norm": 0.7662025094032288, "learning_rate": 0.0001, "elbo": 3.66184, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9572484166080226}
{"step": 33380, "loss": 3.68978271484375, "grad_norm": 0.7315869331359863, "learning_rate": 0.0001, "elbo": 3.68978, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.957541637344593}
{"step": 33385, "loss": 3.6700172424316406, "grad_norm": 0.6591295003890991, "learning_rate": 0.0001, "elbo": 3.67002, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9578348580811635}
{"step": 33390, "loss": 3.675202560424805, "grad_norm": 0.6928391456604004, "learning_rate": 0.0001, "elbo": 3.6752, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.958128078817734}
{"step": 33395, "loss": 3.6856853485107424, "grad_norm": 0.712981104850769, "learning_rate": 0.0001, "elbo": 3.68569, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9584212995543044}
{"step": 33400, "loss": 3.670201873779297, "grad_norm": 0.6652625203132629, "learning_rate": 0.0001, "elbo": 3.6702, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.958714520290875}
{"step": 33405, "loss": 3.669439697265625, "grad_norm": 0.6515827178955078, "learning_rate": 0.0001, "elbo": 3.66944, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9590077410274453}
{"step": 33410, "loss": 3.6804527282714843, "grad_norm": 0.6891077160835266, "learning_rate": 0.0001, "elbo": 3.68045, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9593009617640158}
{"step": 33415, "loss": 3.682704544067383, "grad_norm": 0.7372838854789734, "learning_rate": 0.0001, "elbo": 3.6827, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9595941825005865}
{"step": 33420, "loss": 3.6821178436279296, "grad_norm": 0.7033668756484985, "learning_rate": 0.0001, "elbo": 3.68212, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.959887403237157}
{"step": 33425, "loss": 3.686370849609375, "grad_norm": 0.909856379032135, "learning_rate": 0.0001, "elbo": 3.68637, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9601806239737274}
{"step": 33430, "loss": 3.680766296386719, "grad_norm": 0.7009492516517639, "learning_rate": 0.0001, "elbo": 3.68077, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.960473844710298}
{"step": 33435, "loss": 3.6861927032470705, "grad_norm": 0.644358217716217, "learning_rate": 0.0001, "elbo": 3.68619, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9607670654468685}
{"step": 33440, "loss": 3.667573165893555, "grad_norm": 0.7766111493110657, "learning_rate": 0.0001, "elbo": 3.66757, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.961060286183439}
{"step": 33445, "loss": 3.6995277404785156, "grad_norm": 0.7129954695701599, "learning_rate": 0.0001, "elbo": 3.69953, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9613535069200094}
{"step": 33450, "loss": 3.689920425415039, "grad_norm": 0.6931774020195007, "learning_rate": 0.0001, "elbo": 3.68992, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9616467276565799}
{"step": 33455, "loss": 3.673456573486328, "grad_norm": 0.6989889144897461, "learning_rate": 0.0001, "elbo": 3.67346, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9619399483931503}
{"step": 33460, "loss": 3.681069564819336, "grad_norm": 0.6497710943222046, "learning_rate": 0.0001, "elbo": 3.68107, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9622331691297208}
{"step": 33465, "loss": 3.661896514892578, "grad_norm": 0.6675018668174744, "learning_rate": 0.0001, "elbo": 3.6619, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9625263898662912}
{"step": 33470, "loss": 3.675769805908203, "grad_norm": 0.7023600935935974, "learning_rate": 0.0001, "elbo": 3.67577, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9628196106028617}
{"step": 33475, "loss": 3.671070861816406, "grad_norm": 0.69661945104599, "learning_rate": 0.0001, "elbo": 3.67107, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9631128313394324}
{"step": 33480, "loss": 3.678949737548828, "grad_norm": 0.6808742880821228, "learning_rate": 0.0001, "elbo": 3.67895, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9634060520760028}
{"step": 33485, "loss": 3.6754837036132812, "grad_norm": 0.8119854927062988, "learning_rate": 0.0001, "elbo": 3.67548, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9636992728125733}
{"step": 33490, "loss": 3.6688121795654296, "grad_norm": 0.634732186794281, "learning_rate": 0.0001, "elbo": 3.66881, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.963992493549144}
{"step": 33495, "loss": 3.6694549560546874, "grad_norm": 0.6576728224754333, "learning_rate": 0.0001, "elbo": 3.66945, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9642857142857144}
{"step": 33500, "loss": 3.6790325164794924, "grad_norm": 0.6940693855285645, "learning_rate": 0.0001, "elbo": 3.67903, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9645789350222849}
{"step": 33505, "loss": 3.689081573486328, "grad_norm": 0.6621814370155334, "learning_rate": 0.0001, "elbo": 3.68908, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9648721557588553}
{"step": 33510, "loss": 3.6854007720947264, "grad_norm": 0.920531690120697, "learning_rate": 0.0001, "elbo": 3.6854, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9651653764954258}
{"step": 33515, "loss": 3.670549011230469, "grad_norm": 0.7347320914268494, "learning_rate": 0.0001, "elbo": 3.67055, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9654585972319962}
{"step": 33520, "loss": 3.665108489990234, "grad_norm": 0.6605965495109558, "learning_rate": 0.0001, "elbo": 3.66511, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9657518179685667}
{"step": 33525, "loss": 3.6818222045898437, "grad_norm": 0.6857520937919617, "learning_rate": 0.0001, "elbo": 3.68182, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9660450387051371}
{"step": 33530, "loss": 3.677825164794922, "grad_norm": 0.6809183359146118, "learning_rate": 0.0001, "elbo": 3.67783, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9663382594417076}
{"step": 33535, "loss": 3.6798614501953124, "grad_norm": 0.6800457239151001, "learning_rate": 0.0001, "elbo": 3.67986, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.966631480178278}
{"step": 33535, "loss": 3.6798614501953124, "grad_norm": 0.6800457239151001, "learning_rate": 0.0001, "elbo": 3.67986, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.966631480178278}
{"step": 33540, "loss": 3.678801727294922, "grad_norm": 0.7327957153320312, "learning_rate": 0.0001, "elbo": 3.6788, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9669247009148487}
{"step": 33545, "loss": 3.6607078552246093, "grad_norm": 0.6881885528564453, "learning_rate": 0.0001, "elbo": 3.66071, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9672179216514192}
{"step": 33550, "loss": 3.6584571838378905, "grad_norm": 0.6574678421020508, "learning_rate": 0.0001, "elbo": 3.65846, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9675111423879896}
{"step": 33555, "loss": 3.6822620391845704, "grad_norm": 0.6905481219291687, "learning_rate": 0.0001, "elbo": 3.68226, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9678043631245603}
{"step": 33560, "loss": 3.6714599609375, "grad_norm": 1.0111721754074097, "learning_rate": 0.0001, "elbo": 3.67146, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9680975838611308}
{"step": 33565, "loss": 3.66949462890625, "grad_norm": 0.7101216912269592, "learning_rate": 0.0001, "elbo": 3.66949, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9683908045977012}
{"step": 33570, "loss": 3.6798492431640626, "grad_norm": 0.7833957076072693, "learning_rate": 0.0001, "elbo": 3.67985, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9686840253342717}
{"step": 33575, "loss": 3.6805046081542967, "grad_norm": 0.7124394774436951, "learning_rate": 0.0001, "elbo": 3.6805, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9689772460708421}
{"step": 33580, "loss": 3.658675765991211, "grad_norm": 0.8618826866149902, "learning_rate": 0.0001, "elbo": 3.65868, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9692704668074126}
{"step": 33585, "loss": 3.6930191040039064, "grad_norm": 0.7317866683006287, "learning_rate": 0.0001, "elbo": 3.69302, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.969563687543983}
{"step": 33590, "loss": 3.677866744995117, "grad_norm": 0.9342548847198486, "learning_rate": 0.0001, "elbo": 3.67787, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9698569082805535}
{"step": 33595, "loss": 3.6695648193359376, "grad_norm": 0.8152351975440979, "learning_rate": 0.0001, "elbo": 3.66956, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.970150129017124}
{"step": 33600, "loss": 3.697400665283203, "grad_norm": 0.727834939956665, "learning_rate": 0.0001, "elbo": 3.6974, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9704433497536946}
{"step": 33605, "loss": 3.6808563232421876, "grad_norm": 0.6622186303138733, "learning_rate": 0.0001, "elbo": 3.68086, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.970736570490265}
{"step": 33610, "loss": 3.6768466949462892, "grad_norm": 0.6749950647354126, "learning_rate": 0.0001, "elbo": 3.67685, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9710297912268355}
{"step": 33615, "loss": 3.6718460083007813, "grad_norm": 0.7875882387161255, "learning_rate": 0.0001, "elbo": 3.67185, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9713230119634062}
{"step": 33620, "loss": 3.6736488342285156, "grad_norm": 0.7629523873329163, "learning_rate": 0.0001, "elbo": 3.67365, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9716162326999767}
{"step": 33625, "loss": 3.6677528381347657, "grad_norm": 1.0532513856887817, "learning_rate": 0.0001, "elbo": 3.66775, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.971909453436547}
{"step": 33630, "loss": 3.6663414001464845, "grad_norm": 0.7346323132514954, "learning_rate": 0.0001, "elbo": 3.66634, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9722026741731176}
{"step": 33635, "loss": 3.6872940063476562, "grad_norm": 0.7724911570549011, "learning_rate": 0.0001, "elbo": 3.68729, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.972495894909688}
{"step": 33640, "loss": 3.688885498046875, "grad_norm": 0.7030640840530396, "learning_rate": 0.0001, "elbo": 3.68889, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9727891156462585}
{"step": 33645, "loss": 3.690172576904297, "grad_norm": 0.6887113451957703, "learning_rate": 0.0001, "elbo": 3.69017, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.973082336382829}
{"step": 33650, "loss": 3.6633514404296874, "grad_norm": 0.6827857494354248, "learning_rate": 0.0001, "elbo": 3.66335, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9733755571193994}
{"step": 33655, "loss": 3.659387969970703, "grad_norm": 0.6933181881904602, "learning_rate": 0.0001, "elbo": 3.65939, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9736687778559698}
{"step": 33660, "loss": 3.666458511352539, "grad_norm": 0.662970244884491, "learning_rate": 0.0001, "elbo": 3.66646, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9739619985925405}
{"step": 33665, "loss": 3.679302978515625, "grad_norm": 0.6847151517868042, "learning_rate": 0.0001, "elbo": 3.6793, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.974255219329111}
{"step": 33670, "loss": 3.6726287841796874, "grad_norm": 0.688289999961853, "learning_rate": 0.0001, "elbo": 3.67263, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9745484400656814}
{"step": 33675, "loss": 3.670152282714844, "grad_norm": 0.6999989151954651, "learning_rate": 0.0001, "elbo": 3.67015, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.974841660802252}
{"step": 33680, "loss": 3.6791828155517576, "grad_norm": 0.655415415763855, "learning_rate": 0.0001, "elbo": 3.67918, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9751348815388226}
{"step": 33685, "loss": 3.6715763092041014, "grad_norm": 0.6956544518470764, "learning_rate": 0.0001, "elbo": 3.67158, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.975428102275393}
{"step": 33690, "loss": 3.6707130432128907, "grad_norm": 0.6695858836174011, "learning_rate": 0.0001, "elbo": 3.67071, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9757213230119635}
{"step": 33695, "loss": 3.7009292602539063, "grad_norm": 0.6406168341636658, "learning_rate": 0.0001, "elbo": 3.70093, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.976014543748534}
{"step": 33700, "loss": 3.6704612731933595, "grad_norm": 0.8380477428436279, "learning_rate": 0.0001, "elbo": 3.67046, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9763077644851044}
{"step": 33705, "loss": 3.663491058349609, "grad_norm": 0.6553831696510315, "learning_rate": 0.0001, "elbo": 3.66349, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9766009852216748}
{"step": 33710, "loss": 3.6693866729736326, "grad_norm": 0.7822697162628174, "learning_rate": 0.0001, "elbo": 3.66939, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9768942059582453}
{"step": 33715, "loss": 3.6539291381835937, "grad_norm": 0.7579296827316284, "learning_rate": 0.0001, "elbo": 3.65393, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9771874266948157}
{"step": 33720, "loss": 3.6798614501953124, "grad_norm": 0.6380520462989807, "learning_rate": 0.0001, "elbo": 3.67986, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9774806474313862}
{"step": 33725, "loss": 3.669800567626953, "grad_norm": 0.7133125066757202, "learning_rate": 0.0001, "elbo": 3.6698, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9777738681679569}
{"step": 33730, "loss": 3.675870895385742, "grad_norm": 0.7141843438148499, "learning_rate": 0.0001, "elbo": 3.67587, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9780670889045273}
{"step": 33735, "loss": 3.6579521179199217, "grad_norm": 0.6990371346473694, "learning_rate": 0.0001, "elbo": 3.65795, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9783603096410978}
{"step": 33740, "loss": 3.6747390747070314, "grad_norm": 0.6856667399406433, "learning_rate": 0.0001, "elbo": 3.67474, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9786535303776684}
{"step": 33745, "loss": 3.6798858642578125, "grad_norm": 0.7409234642982483, "learning_rate": 0.0001, "elbo": 3.67989, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.978946751114239}
{"step": 33750, "loss": 3.689421463012695, "grad_norm": 0.6679820418357849, "learning_rate": 0.0001, "elbo": 3.68942, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9792399718508094}
{"step": 33755, "loss": 3.6844432830810545, "grad_norm": 0.695955753326416, "learning_rate": 0.0001, "elbo": 3.68444, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9795331925873798}
{"step": 33760, "loss": 3.6679977416992187, "grad_norm": 0.7653222680091858, "learning_rate": 0.0001, "elbo": 3.668, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9798264133239503}
{"step": 33765, "loss": 3.6829410552978517, "grad_norm": 0.7189032435417175, "learning_rate": 0.0001, "elbo": 3.68294, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9801196340605207}
{"step": 33770, "loss": 3.668140411376953, "grad_norm": 0.693757176399231, "learning_rate": 0.0001, "elbo": 3.66814, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9804128547970912}
{"step": 33770, "loss": 3.668140411376953, "grad_norm": 0.693757176399231, "learning_rate": 0.0001, "elbo": 3.66814, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9804128547970912}
{"step": 33775, "loss": 3.670155334472656, "grad_norm": 0.697728157043457, "learning_rate": 0.0001, "elbo": 3.67016, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9807060755336616}
{"step": 33780, "loss": 3.6574119567871093, "grad_norm": 0.8106796741485596, "learning_rate": 0.0001, "elbo": 3.65741, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.980999296270232}
{"step": 33785, "loss": 3.662907028198242, "grad_norm": 0.7381415367126465, "learning_rate": 0.0001, "elbo": 3.66291, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9812925170068028}
{"step": 33790, "loss": 3.683580780029297, "grad_norm": 0.6934007406234741, "learning_rate": 0.0001, "elbo": 3.68358, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9815857377433732}
{"step": 33795, "loss": 3.667885589599609, "grad_norm": 0.6598266363143921, "learning_rate": 0.0001, "elbo": 3.66789, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9818789584799437}
{"step": 33800, "loss": 3.6628097534179687, "grad_norm": 0.696026086807251, "learning_rate": 0.0001, "elbo": 3.66281, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9821721792165143}
{"step": 33800, "loss": 3.6628097534179687, "grad_norm": 0.696026086807251, "learning_rate": 0.0001, "elbo": 3.66281, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9821721792165143}
{"step": 33805, "loss": 3.6668792724609376, "grad_norm": 0.763028621673584, "learning_rate": 0.0001, "elbo": 3.66688, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9824653999530848}
{"step": 33810, "loss": 3.659931182861328, "grad_norm": 0.7141203880310059, "learning_rate": 0.0001, "elbo": 3.65993, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9827586206896552}
{"step": 33815, "loss": 3.680968475341797, "grad_norm": 0.6851702928543091, "learning_rate": 0.0001, "elbo": 3.68097, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9830518414262257}
{"step": 33820, "loss": 3.7032989501953124, "grad_norm": 0.6249521374702454, "learning_rate": 0.0001, "elbo": 3.7033, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9833450621627962}
{"step": 33825, "loss": 3.6586841583251952, "grad_norm": 0.6862894892692566, "learning_rate": 0.0001, "elbo": 3.65868, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9836382828993666}
{"step": 33830, "loss": 3.6733497619628905, "grad_norm": 0.6985527276992798, "learning_rate": 0.0001, "elbo": 3.67335, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.983931503635937}
{"step": 33835, "loss": 3.666925811767578, "grad_norm": 0.6807604432106018, "learning_rate": 0.0001, "elbo": 3.66693, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9842247243725075}
{"step": 33840, "loss": 3.6714637756347654, "grad_norm": 0.6670225858688354, "learning_rate": 0.0001, "elbo": 3.67146, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.984517945109078}
{"step": 33845, "loss": 3.6814598083496093, "grad_norm": 0.7185796499252319, "learning_rate": 0.0001, "elbo": 3.68146, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9848111658456487}
{"step": 33850, "loss": 3.675084686279297, "grad_norm": 0.6941696405410767, "learning_rate": 0.0001, "elbo": 3.67508, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.985104386582219}
{"step": 33855, "loss": 3.6512992858886717, "grad_norm": 0.6514418125152588, "learning_rate": 0.0001, "elbo": 3.6513, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9853976073187896}
{"step": 33860, "loss": 3.66795654296875, "grad_norm": 0.8621746301651001, "learning_rate": 0.0001, "elbo": 3.66796, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9856908280553602}
{"step": 33865, "loss": 3.679423522949219, "grad_norm": 0.6629466414451599, "learning_rate": 0.0001, "elbo": 3.67942, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9859840487919307}
{"step": 33865, "loss": 3.679423522949219, "grad_norm": 0.6629466414451599, "learning_rate": 0.0001, "elbo": 3.67942, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9859840487919307}
{"step": 33870, "loss": 3.6524242401123046, "grad_norm": 0.7533007264137268, "learning_rate": 0.0001, "elbo": 3.65242, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9862772695285011}
{"step": 33875, "loss": 3.6756317138671877, "grad_norm": 0.7538706660270691, "learning_rate": 0.0001, "elbo": 3.67563, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9865704902650716}
{"step": 33880, "loss": 3.688872528076172, "grad_norm": 1.0306838750839233, "learning_rate": 0.0001, "elbo": 3.68887, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.986863711001642}
{"step": 33885, "loss": 3.6851127624511717, "grad_norm": 0.6608123779296875, "learning_rate": 0.0001, "elbo": 3.68511, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9871569317382125}
{"step": 33890, "loss": 3.6750789642333985, "grad_norm": 0.6865563988685608, "learning_rate": 0.0001, "elbo": 3.67508, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.987450152474783}
{"step": 33895, "loss": 3.681135559082031, "grad_norm": 0.6401357650756836, "learning_rate": 0.0001, "elbo": 3.68114, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9877433732113534}
{"step": 33900, "loss": 3.6924156188964843, "grad_norm": 0.6456862092018127, "learning_rate": 0.0001, "elbo": 3.69242, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9880365939479239}
{"step": 33905, "loss": 3.6941112518310546, "grad_norm": 0.7343336343765259, "learning_rate": 0.0001, "elbo": 3.69411, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9883298146844943}
{"step": 33910, "loss": 3.676165008544922, "grad_norm": 0.6735513806343079, "learning_rate": 0.0001, "elbo": 3.67617, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.988623035421065}
{"step": 33915, "loss": 3.6755386352539063, "grad_norm": 0.6904743909835815, "learning_rate": 0.0001, "elbo": 3.67554, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9889162561576355}
{"step": 33920, "loss": 3.6790325164794924, "grad_norm": 0.7603839039802551, "learning_rate": 0.0001, "elbo": 3.67903, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.989209476894206}
{"step": 33920, "loss": 3.6790325164794924, "grad_norm": 0.7603839039802551, "learning_rate": 0.0001, "elbo": 3.67903, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.989209476894206}
{"step": 33925, "loss": 3.679518127441406, "grad_norm": 0.6705762147903442, "learning_rate": 0.0001, "elbo": 3.67952, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9895026976307766}
{"step": 33930, "loss": 3.6631736755371094, "grad_norm": 0.6745433807373047, "learning_rate": 0.0001, "elbo": 3.66317, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.989795918367347}
{"step": 33935, "loss": 3.665213775634766, "grad_norm": 0.6479117274284363, "learning_rate": 0.0001, "elbo": 3.66521, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9900891391039175}
{"step": 33940, "loss": 3.665077972412109, "grad_norm": 0.6534920334815979, "learning_rate": 0.0001, "elbo": 3.66508, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.990382359840488}
{"step": 33945, "loss": 3.6551498413085937, "grad_norm": 0.8218843340873718, "learning_rate": 0.0001, "elbo": 3.65515, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9906755805770584}
{"step": 33950, "loss": 3.660687255859375, "grad_norm": 1.0651588439941406, "learning_rate": 0.0001, "elbo": 3.66069, "mask_frac": 0.4996, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9909688013136289}
{"step": 33955, "loss": 3.6836036682128905, "grad_norm": 0.6448226571083069, "learning_rate": 0.0001, "elbo": 3.6836, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9912620220501993}
{"step": 33960, "loss": 3.6711669921875, "grad_norm": 0.7344462871551514, "learning_rate": 0.0001, "elbo": 3.67117, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9915552427867698}
{"step": 33965, "loss": 3.6709999084472655, "grad_norm": 0.6978361010551453, "learning_rate": 0.0001, "elbo": 3.671, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9918484635233402}
{"step": 33970, "loss": 3.679309844970703, "grad_norm": 0.6763094067573547, "learning_rate": 0.0001, "elbo": 3.67931, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.992141684259911}
{"step": 33975, "loss": 3.6668113708496093, "grad_norm": 0.6697335243225098, "learning_rate": 0.0001, "elbo": 3.66681, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9924349049964813}
{"step": 33980, "loss": 3.701995086669922, "grad_norm": 0.7297865152359009, "learning_rate": 0.0001, "elbo": 3.702, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9927281257330518}
{"step": 33985, "loss": 3.653158950805664, "grad_norm": 0.6835659146308899, "learning_rate": 0.0001, "elbo": 3.65316, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9930213464696225}
{"step": 33990, "loss": 3.6855674743652345, "grad_norm": 0.9359193444252014, "learning_rate": 0.0001, "elbo": 3.68557, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.993314567206193}
{"step": 33995, "loss": 3.684083938598633, "grad_norm": 0.932573139667511, "learning_rate": 0.0001, "elbo": 3.68408, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9936077879427634}
{"step": 34000, "loss": 3.673181915283203, "grad_norm": 0.6879629492759705, "learning_rate": 0.0001, "elbo": 3.67318, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9939010086793338}
{"step": 34005, "loss": 3.6548797607421877, "grad_norm": 0.7471395134925842, "learning_rate": 0.0001, "elbo": 3.65488, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9941942294159043}
{"step": 34010, "loss": 3.677896499633789, "grad_norm": 0.6909501552581787, "learning_rate": 0.0001, "elbo": 3.6779, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9944874501524748}
{"step": 34015, "loss": 3.6704032897949217, "grad_norm": 0.6889094710350037, "learning_rate": 0.0001, "elbo": 3.6704, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9947806708890452}
{"step": 34020, "loss": 3.669094467163086, "grad_norm": 0.746920108795166, "learning_rate": 0.0001, "elbo": 3.66909, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9950738916256157}
{"step": 34025, "loss": 3.6780979156494142, "grad_norm": 0.6994937658309937, "learning_rate": 0.0001, "elbo": 3.6781, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9953671123621861}
{"step": 34030, "loss": 3.657484436035156, "grad_norm": 0.7216588854789734, "learning_rate": 0.0001, "elbo": 3.65748, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9956603330987568}
{"step": 34035, "loss": 3.6699623107910155, "grad_norm": 0.671667218208313, "learning_rate": 0.0001, "elbo": 3.66996, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9959535538353272}
{"step": 34040, "loss": 3.6772510528564455, "grad_norm": 0.6977344155311584, "learning_rate": 0.0001, "elbo": 3.67725, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9962467745718977}
{"step": 34045, "loss": 3.6726612091064452, "grad_norm": 0.723617434501648, "learning_rate": 0.0001, "elbo": 3.67266, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9965399953084684}
{"step": 34050, "loss": 3.6858009338378905, "grad_norm": 0.7426633238792419, "learning_rate": 0.0001, "elbo": 3.6858, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9968332160450388}
{"step": 34055, "loss": 3.6613204956054686, "grad_norm": 0.7911838889122009, "learning_rate": 0.0001, "elbo": 3.66132, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9971264367816093}
{"step": 34060, "loss": 3.670276641845703, "grad_norm": 0.7123444676399231, "learning_rate": 0.0001, "elbo": 3.67028, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9974196575181797}
{"step": 34065, "loss": 3.682534027099609, "grad_norm": 0.6646880507469177, "learning_rate": 0.0001, "elbo": 3.68253, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9977128782547502}
{"step": 34070, "loss": 3.6947181701660154, "grad_norm": 1.2011626958847046, "learning_rate": 0.0001, "elbo": 3.69472, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9980060989913206}
{"step": 34075, "loss": 3.6729007720947267, "grad_norm": 0.7310740351676941, "learning_rate": 0.0001, "elbo": 3.6729, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.998299319727891}
{"step": 34080, "loss": 3.6638328552246096, "grad_norm": 0.6368361115455627, "learning_rate": 0.0001, "elbo": 3.66383, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9985925404644616}
{"step": 34085, "loss": 3.674470901489258, "grad_norm": 0.6713269352912903, "learning_rate": 0.0001, "elbo": 3.67447, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.998885761201032}
{"step": 34090, "loss": 3.680611801147461, "grad_norm": 0.6678178310394287, "learning_rate": 0.0001, "elbo": 3.68061, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9991789819376025}
{"step": 34095, "loss": 3.6798564910888674, "grad_norm": 0.6945571303367615, "learning_rate": 0.0001, "elbo": 3.67986, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9994722026741731}
{"step": 34100, "loss": 3.6920806884765627, "grad_norm": 0.8375648856163025, "learning_rate": 0.0001, "elbo": 3.69208, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 1.9997654234107436}
{"step": 34105, "loss": 3.6759078979492186, "grad_norm": 0.6503887176513672, "learning_rate": 0.0001, "elbo": 3.67591, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0000586441473143}
{"step": 34110, "loss": 3.6689994812011717, "grad_norm": 0.6496407985687256, "learning_rate": 0.0001, "elbo": 3.669, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0003518648838847}
{"step": 34115, "loss": 3.661215972900391, "grad_norm": 0.652935802936554, "learning_rate": 0.0001, "elbo": 3.66122, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.000645085620455}
{"step": 34120, "loss": 3.6778938293457033, "grad_norm": 0.6221675872802734, "learning_rate": 0.0001, "elbo": 3.67789, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0009383063570256}
{"step": 34125, "loss": 3.6750965118408203, "grad_norm": 0.6886290311813354, "learning_rate": 0.0001, "elbo": 3.6751, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.001231527093596}
{"step": 34130, "loss": 3.6607692718505858, "grad_norm": 0.6591792702674866, "learning_rate": 0.0001, "elbo": 3.66077, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0015247478301665}
{"step": 34135, "loss": 3.6729602813720703, "grad_norm": 0.7013826370239258, "learning_rate": 0.0001, "elbo": 3.67296, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.001817968566737}
{"step": 34140, "loss": 3.6525089263916017, "grad_norm": 0.6548539400100708, "learning_rate": 0.0001, "elbo": 3.65251, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0021111893033074}
{"step": 34145, "loss": 3.6764541625976563, "grad_norm": 0.8481391072273254, "learning_rate": 0.0001, "elbo": 3.67645, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.002404410039878}
{"step": 34150, "loss": 3.6436431884765623, "grad_norm": 0.6690880656242371, "learning_rate": 0.0001, "elbo": 3.64364, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0026976307764484}
{"step": 34155, "loss": 3.6623275756835936, "grad_norm": 0.7059962749481201, "learning_rate": 0.0001, "elbo": 3.66233, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.002990851513019}
{"step": 34160, "loss": 3.656934356689453, "grad_norm": 0.8890252709388733, "learning_rate": 0.0001, "elbo": 3.65693, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0032840722495897}
{"step": 34165, "loss": 3.660591888427734, "grad_norm": 0.737156867980957, "learning_rate": 0.0001, "elbo": 3.66059, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.00357729298616}
{"step": 34170, "loss": 3.662976837158203, "grad_norm": 0.8031145334243774, "learning_rate": 0.0001, "elbo": 3.66298, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0038705137227306}
{"step": 34175, "loss": 3.6763278961181642, "grad_norm": 0.6582943797111511, "learning_rate": 0.0001, "elbo": 3.67633, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.004163734459301}
{"step": 34180, "loss": 3.6741329193115235, "grad_norm": 0.6353946924209595, "learning_rate": 0.0001, "elbo": 3.67413, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0044569551958715}
{"step": 34185, "loss": 3.678688812255859, "grad_norm": 0.6704179644584656, "learning_rate": 0.0001, "elbo": 3.67869, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.004750175932442}
{"step": 34190, "loss": 3.65728874206543, "grad_norm": 1.171303629875183, "learning_rate": 0.0001, "elbo": 3.65729, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0050433966690124}
{"step": 34195, "loss": 3.679425811767578, "grad_norm": 0.6633267998695374, "learning_rate": 0.0001, "elbo": 3.67943, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.005336617405583}
{"step": 34200, "loss": 3.659349060058594, "grad_norm": 0.6765723824501038, "learning_rate": 0.0001, "elbo": 3.65935, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0056298381421533}
{"step": 34205, "loss": 3.6721256256103514, "grad_norm": 0.703441858291626, "learning_rate": 0.0001, "elbo": 3.67213, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.005923058878724}
{"step": 34210, "loss": 3.667826843261719, "grad_norm": 0.7186678051948547, "learning_rate": 0.0001, "elbo": 3.66783, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0062162796152943}
{"step": 34210, "loss": 3.667826843261719, "grad_norm": 0.7186678051948547, "learning_rate": 0.0001, "elbo": 3.66783, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0062162796152943}
{"step": 34215, "loss": 3.66864013671875, "grad_norm": 0.8157921433448792, "learning_rate": 0.0001, "elbo": 3.66864, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0065095003518647}
{"step": 34220, "loss": 3.67044677734375, "grad_norm": 0.6983188986778259, "learning_rate": 0.0001, "elbo": 3.67045, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.006802721088435}
{"step": 34225, "loss": 3.662000274658203, "grad_norm": 0.6694417595863342, "learning_rate": 0.0001, "elbo": 3.662, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.007095941825006}
{"step": 34230, "loss": 3.6697128295898436, "grad_norm": 0.6975647807121277, "learning_rate": 0.0001, "elbo": 3.66971, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0073891625615765}
{"step": 34235, "loss": 3.64593620300293, "grad_norm": 0.7558826208114624, "learning_rate": 0.0001, "elbo": 3.64594, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.007682383298147}
{"step": 34235, "loss": 3.64593620300293, "grad_norm": 0.7558826208114624, "learning_rate": 0.0001, "elbo": 3.64594, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.007682383298147}
{"step": 34240, "loss": 3.6741104125976562, "grad_norm": 0.8522248268127441, "learning_rate": 0.0001, "elbo": 3.67411, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0079756040347174}
{"step": 34245, "loss": 3.678749847412109, "grad_norm": 0.8681185245513916, "learning_rate": 0.0001, "elbo": 3.67875, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.008268824771288}
{"step": 34250, "loss": 3.6812149047851563, "grad_norm": 0.6675321459770203, "learning_rate": 0.0001, "elbo": 3.68121, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0085620455078583}
{"step": 34255, "loss": 3.6705001831054687, "grad_norm": 0.737468957901001, "learning_rate": 0.0001, "elbo": 3.6705, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.008855266244429}
{"step": 34260, "loss": 3.6687362670898436, "grad_norm": 0.7128269672393799, "learning_rate": 0.0001, "elbo": 3.66874, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0091484869809992}
{"step": 34265, "loss": 3.6558197021484373, "grad_norm": 0.7386688590049744, "learning_rate": 0.0001, "elbo": 3.65582, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0094417077175697}
{"step": 34270, "loss": 3.6658592224121094, "grad_norm": 0.6813642978668213, "learning_rate": 0.0001, "elbo": 3.66586, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.00973492845414}
{"step": 34275, "loss": 3.667779541015625, "grad_norm": 0.6869343519210815, "learning_rate": 0.0001, "elbo": 3.66778, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0100281491907106}
{"step": 34280, "loss": 3.6591506958007813, "grad_norm": 0.6983076333999634, "learning_rate": 0.0001, "elbo": 3.65915, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.010321369927281}
{"step": 34285, "loss": 3.6493694305419924, "grad_norm": 0.6855893731117249, "learning_rate": 0.0001, "elbo": 3.64937, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.010614590663852}
{"step": 34290, "loss": 3.6708984375, "grad_norm": 0.6754461526870728, "learning_rate": 0.0001, "elbo": 3.6709, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0109078114004224}
{"step": 34295, "loss": 3.6645843505859377, "grad_norm": 0.6600556969642639, "learning_rate": 0.0001, "elbo": 3.66458, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.011201032136993}
{"step": 34300, "loss": 3.6757659912109375, "grad_norm": 0.7937663793563843, "learning_rate": 0.0001, "elbo": 3.67577, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0114942528735633}
{"step": 34305, "loss": 3.6576427459716796, "grad_norm": 0.8156101703643799, "learning_rate": 0.0001, "elbo": 3.65764, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0117874736101338}
{"step": 34310, "loss": 3.6610084533691407, "grad_norm": 0.7133549451828003, "learning_rate": 0.0001, "elbo": 3.66101, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0120806943467042}
{"step": 34315, "loss": 3.6683143615722655, "grad_norm": 0.8963392376899719, "learning_rate": 0.0001, "elbo": 3.66831, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0123739150832747}
{"step": 34320, "loss": 3.6736915588378904, "grad_norm": 0.729405403137207, "learning_rate": 0.0001, "elbo": 3.67369, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.012667135819845}
{"step": 34325, "loss": 3.69573974609375, "grad_norm": 0.6608788371086121, "learning_rate": 0.0001, "elbo": 3.69574, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0129603565564156}
{"step": 34330, "loss": 3.6938419342041016, "grad_norm": 0.7366861701011658, "learning_rate": 0.0001, "elbo": 3.69384, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.013253577292986}
{"step": 34335, "loss": 3.66590576171875, "grad_norm": 0.6863507032394409, "learning_rate": 0.0001, "elbo": 3.66591, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0135467980295565}
{"step": 34340, "loss": 3.6598201751708985, "grad_norm": 0.6584275960922241, "learning_rate": 0.0001, "elbo": 3.65982, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.013840018766127}
{"step": 34345, "loss": 3.6558998107910154, "grad_norm": 0.6830412149429321, "learning_rate": 0.0001, "elbo": 3.6559, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.014133239502698}
{"step": 34350, "loss": 3.6615478515625, "grad_norm": 0.7311811447143555, "learning_rate": 0.0001, "elbo": 3.66155, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0144264602392683}
{"step": 34355, "loss": 3.67291259765625, "grad_norm": 0.7003733515739441, "learning_rate": 0.0001, "elbo": 3.67291, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0147196809758388}
{"step": 34360, "loss": 3.658014678955078, "grad_norm": 0.6903300881385803, "learning_rate": 0.0001, "elbo": 3.65801, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.015012901712409}
{"step": 34365, "loss": 3.684761810302734, "grad_norm": 0.6898698210716248, "learning_rate": 0.0001, "elbo": 3.68476, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0153061224489797}
{"step": 34370, "loss": 3.677089309692383, "grad_norm": 0.6668850779533386, "learning_rate": 0.0001, "elbo": 3.67709, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.01559934318555}
{"step": 34375, "loss": 3.670798110961914, "grad_norm": 0.6959595084190369, "learning_rate": 0.0001, "elbo": 3.6708, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0158925639221206}
{"step": 34380, "loss": 3.6734352111816406, "grad_norm": 0.6851236820220947, "learning_rate": 0.0001, "elbo": 3.67344, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.016185784658691}
{"step": 34385, "loss": 3.6742527008056642, "grad_norm": 0.684097170829773, "learning_rate": 0.0001, "elbo": 3.67425, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0164790053952615}
{"step": 34390, "loss": 3.670091247558594, "grad_norm": 0.6995691657066345, "learning_rate": 0.0001, "elbo": 3.67009, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.016772226131832}
{"step": 34395, "loss": 3.6786758422851564, "grad_norm": 0.7623721361160278, "learning_rate": 0.0001, "elbo": 3.67868, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0170654468684024}
{"step": 34400, "loss": 3.665607452392578, "grad_norm": 0.6776127815246582, "learning_rate": 0.0001, "elbo": 3.66561, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.017358667604973}
{"step": 34405, "loss": 3.6543601989746093, "grad_norm": 0.696395218372345, "learning_rate": 0.0001, "elbo": 3.65436, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0176518883415433}
{"step": 34410, "loss": 3.6692962646484375, "grad_norm": 0.6837601661682129, "learning_rate": 0.0001, "elbo": 3.6693, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.017945109078114}
{"step": 34415, "loss": 3.6589096069335936, "grad_norm": 0.9591047167778015, "learning_rate": 0.0001, "elbo": 3.65891, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0182383298146847}
{"step": 34420, "loss": 3.6694847106933595, "grad_norm": 0.6781316995620728, "learning_rate": 0.0001, "elbo": 3.66948, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.018531550551255}
{"step": 34425, "loss": 3.6710201263427735, "grad_norm": 0.7773715257644653, "learning_rate": 0.0001, "elbo": 3.67102, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0188247712878256}
{"step": 34430, "loss": 3.6681682586669924, "grad_norm": 0.6530091762542725, "learning_rate": 0.0001, "elbo": 3.66817, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.019117992024396}
{"step": 34435, "loss": 3.6837837219238283, "grad_norm": 0.6834563612937927, "learning_rate": 0.0001, "elbo": 3.68378, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0194112127609665}
{"step": 34440, "loss": 3.676224136352539, "grad_norm": 0.7463983297348022, "learning_rate": 0.0001, "elbo": 3.67622, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.019704433497537}
{"step": 34445, "loss": 3.6741966247558593, "grad_norm": 0.6622551083564758, "learning_rate": 0.0001, "elbo": 3.6742, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0199976542341074}
{"step": 34450, "loss": 3.679574203491211, "grad_norm": 0.7168809175491333, "learning_rate": 0.0001, "elbo": 3.67957, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.020290874970678}
{"step": 34455, "loss": 3.6493461608886717, "grad_norm": 0.7257720828056335, "learning_rate": 0.0001, "elbo": 3.64935, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0205840957072483}
{"step": 34460, "loss": 3.6657783508300783, "grad_norm": 0.7130683064460754, "learning_rate": 0.0001, "elbo": 3.66578, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0208773164438187}
{"step": 34460, "loss": 3.6657783508300783, "grad_norm": 0.7130683064460754, "learning_rate": 0.0001, "elbo": 3.66578, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0208773164438187}
{"step": 34465, "loss": 3.676985168457031, "grad_norm": 0.7034868001937866, "learning_rate": 0.0001, "elbo": 3.67699, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.021170537180389}
{"step": 34470, "loss": 3.6772697448730467, "grad_norm": 0.7762100696563721, "learning_rate": 0.0001, "elbo": 3.67727, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.02146375791696}
{"step": 34475, "loss": 3.665258026123047, "grad_norm": 0.6825404763221741, "learning_rate": 0.0001, "elbo": 3.66526, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0217569786535305}
{"step": 34480, "loss": 3.647856903076172, "grad_norm": 0.7345763444900513, "learning_rate": 0.0001, "elbo": 3.64786, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.022050199390101}
{"step": 34485, "loss": 3.660234832763672, "grad_norm": 0.7213042378425598, "learning_rate": 0.0001, "elbo": 3.66023, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0223434201266715}
{"step": 34490, "loss": 3.661488723754883, "grad_norm": 0.6675902009010315, "learning_rate": 0.0001, "elbo": 3.66149, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.022636640863242}
{"step": 34495, "loss": 3.6675365447998045, "grad_norm": 0.6926888227462769, "learning_rate": 0.0001, "elbo": 3.66754, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0229298615998124}
{"step": 34500, "loss": 3.6448623657226564, "grad_norm": 0.677000105381012, "learning_rate": 0.0001, "elbo": 3.64486, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.023223082336383}
{"step": 34505, "loss": 3.6526409149169923, "grad_norm": 0.7026450037956238, "learning_rate": 0.0001, "elbo": 3.65264, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0235163030729533}
{"step": 34510, "loss": 3.670589828491211, "grad_norm": 0.7107177376747131, "learning_rate": 0.0001, "elbo": 3.67059, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0238095238095237}
{"step": 34515, "loss": 3.677614212036133, "grad_norm": 1.0464500188827515, "learning_rate": 0.0001, "elbo": 3.67761, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.024102744546094}
{"step": 34520, "loss": 3.6501842498779298, "grad_norm": 0.6880454421043396, "learning_rate": 0.0001, "elbo": 3.65018, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0243959652826646}
{"step": 34525, "loss": 3.664409637451172, "grad_norm": 0.6681550741195679, "learning_rate": 0.0001, "elbo": 3.66441, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.024689186019235}
{"step": 34530, "loss": 3.676248550415039, "grad_norm": 0.6816170811653137, "learning_rate": 0.0001, "elbo": 3.67625, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.024982406755806}
{"step": 34535, "loss": 3.6734710693359376, "grad_norm": 0.8814325332641602, "learning_rate": 0.0001, "elbo": 3.67347, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0252756274923764}
{"step": 34540, "loss": 3.658560562133789, "grad_norm": 0.6982959508895874, "learning_rate": 0.0001, "elbo": 3.65856, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.025568848228947}
{"step": 34545, "loss": 3.6690834045410154, "grad_norm": 0.7037914991378784, "learning_rate": 0.0001, "elbo": 3.66908, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0258620689655173}
{"step": 34550, "loss": 3.6812713623046873, "grad_norm": 0.8308877348899841, "learning_rate": 0.0001, "elbo": 3.68127, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.026155289702088}
{"step": 34555, "loss": 3.6703636169433596, "grad_norm": 0.8869604468345642, "learning_rate": 0.0001, "elbo": 3.67036, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0264485104386583}
{"step": 34560, "loss": 3.669664764404297, "grad_norm": 0.8640767335891724, "learning_rate": 0.0001, "elbo": 3.66966, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0267417311752287}
{"step": 34565, "loss": 3.6749031066894533, "grad_norm": 0.678217351436615, "learning_rate": 0.0001, "elbo": 3.6749, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.027034951911799}
{"step": 34570, "loss": 3.6810394287109376, "grad_norm": 0.6666107773780823, "learning_rate": 0.0001, "elbo": 3.68104, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0273281726483696}
{"step": 34575, "loss": 3.667816925048828, "grad_norm": 0.6956406235694885, "learning_rate": 0.0001, "elbo": 3.66782, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.02762139338494}
{"step": 34580, "loss": 3.672076416015625, "grad_norm": 0.7578544616699219, "learning_rate": 0.0001, "elbo": 3.67208, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0279146141215105}
{"step": 34585, "loss": 3.6767333984375, "grad_norm": 0.6754438281059265, "learning_rate": 0.0001, "elbo": 3.67673, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.028207834858081}
{"step": 34590, "loss": 3.6755348205566407, "grad_norm": 0.7049353718757629, "learning_rate": 0.0001, "elbo": 3.67553, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0285010555946514}
{"step": 34595, "loss": 3.662453842163086, "grad_norm": 0.7131839394569397, "learning_rate": 0.0001, "elbo": 3.66245, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0287942763312223}
{"step": 34600, "loss": 3.676984405517578, "grad_norm": 0.6525964736938477, "learning_rate": 0.0001, "elbo": 3.67698, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.029087497067793}
{"step": 34605, "loss": 3.6596843719482424, "grad_norm": 0.6675156950950623, "learning_rate": 0.0001, "elbo": 3.65968, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0293807178043632}
{"step": 34610, "loss": 3.685260009765625, "grad_norm": 0.71651691198349, "learning_rate": 0.0001, "elbo": 3.68526, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0296739385409337}
{"step": 34615, "loss": 3.6672122955322264, "grad_norm": 0.7567634582519531, "learning_rate": 0.0001, "elbo": 3.66721, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.029967159277504}
{"step": 34620, "loss": 3.6766437530517577, "grad_norm": 0.6793002486228943, "learning_rate": 0.0001, "elbo": 3.67664, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0302603800140746}
{"step": 34625, "loss": 3.683152770996094, "grad_norm": 0.6651618480682373, "learning_rate": 0.0001, "elbo": 3.68315, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.030553600750645}
{"step": 34630, "loss": 3.6750022888183596, "grad_norm": 0.6829110383987427, "learning_rate": 0.0001, "elbo": 3.675, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0308468214872155}
{"step": 34635, "loss": 3.665296173095703, "grad_norm": 0.6890882253646851, "learning_rate": 0.0001, "elbo": 3.6653, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.031140042223786}
{"step": 34640, "loss": 3.6613700866699217, "grad_norm": 0.6532098650932312, "learning_rate": 0.0001, "elbo": 3.66137, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0314332629603564}
{"step": 34645, "loss": 3.6705047607421877, "grad_norm": 0.882678210735321, "learning_rate": 0.0001, "elbo": 3.6705, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.031726483696927}
{"step": 34650, "loss": 3.6612548828125, "grad_norm": 0.7525099515914917, "learning_rate": 0.0001, "elbo": 3.66126, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0320197044334973}
{"step": 34655, "loss": 3.6634662628173826, "grad_norm": 0.6784490942955017, "learning_rate": 0.0001, "elbo": 3.66347, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0323129251700682}
{"step": 34660, "loss": 3.6806350708007813, "grad_norm": 0.6603683233261108, "learning_rate": 0.0001, "elbo": 3.68064, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0326061459066387}
{"step": 34665, "loss": 3.663219451904297, "grad_norm": 0.6748993992805481, "learning_rate": 0.0001, "elbo": 3.66322, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.032899366643209}
{"step": 34670, "loss": 3.675342559814453, "grad_norm": 0.6607034206390381, "learning_rate": 0.0001, "elbo": 3.67534, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0331925873797796}
{"step": 34675, "loss": 3.674399566650391, "grad_norm": 0.8278089165687561, "learning_rate": 0.0001, "elbo": 3.6744, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.03348580811635}
{"step": 34680, "loss": 3.6510772705078125, "grad_norm": 0.6551205515861511, "learning_rate": 0.0001, "elbo": 3.65108, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0337790288529205}
{"step": 34685, "loss": 3.660675811767578, "grad_norm": 0.6789461970329285, "learning_rate": 0.0001, "elbo": 3.66068, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.034072249589491}
{"step": 34690, "loss": 3.662321853637695, "grad_norm": 0.7586995959281921, "learning_rate": 0.0001, "elbo": 3.66232, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0343654703260614}
{"step": 34695, "loss": 3.6697216033935547, "grad_norm": 0.6721013784408569, "learning_rate": 0.0001, "elbo": 3.66972, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.034658691062632}
{"step": 34700, "loss": 3.659537506103516, "grad_norm": 0.7134163975715637, "learning_rate": 0.0001, "elbo": 3.65954, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0349519117992023}
{"step": 34705, "loss": 3.671002197265625, "grad_norm": 0.7682563066482544, "learning_rate": 0.0001, "elbo": 3.671, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0352451325357728}
{"step": 34710, "loss": 3.6496864318847657, "grad_norm": 0.6661455631256104, "learning_rate": 0.0001, "elbo": 3.64969, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0355383532723432}
{"step": 34715, "loss": 3.663319396972656, "grad_norm": 0.6509372591972351, "learning_rate": 0.0001, "elbo": 3.66332, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.035831574008914}
{"step": 34720, "loss": 3.653746795654297, "grad_norm": 0.678314208984375, "learning_rate": 0.0001, "elbo": 3.65375, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0361247947454846}
{"step": 34725, "loss": 3.6760597229003906, "grad_norm": 0.6825223565101624, "learning_rate": 0.0001, "elbo": 3.67606, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.036418015482055}
{"step": 34730, "loss": 3.6739051818847654, "grad_norm": 0.7530955672264099, "learning_rate": 0.0001, "elbo": 3.67391, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0367112362186255}
{"step": 34735, "loss": 3.6542213439941404, "grad_norm": 0.708796501159668, "learning_rate": 0.0001, "elbo": 3.65422, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.037004456955196}
{"step": 34740, "loss": 3.6611248016357423, "grad_norm": 0.7940590977668762, "learning_rate": 0.0001, "elbo": 3.66112, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0372976776917664}
{"step": 34745, "loss": 3.6696990966796874, "grad_norm": 0.7859578132629395, "learning_rate": 0.0001, "elbo": 3.6697, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.037590898428337}
{"step": 34750, "loss": 3.652239990234375, "grad_norm": 0.7055610418319702, "learning_rate": 0.0001, "elbo": 3.65224, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0378841191649073}
{"step": 34755, "loss": 3.6595497131347656, "grad_norm": 0.6543535590171814, "learning_rate": 0.0001, "elbo": 3.65955, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0381773399014778}
{"step": 34760, "loss": 3.6608715057373047, "grad_norm": 0.7111244201660156, "learning_rate": 0.0001, "elbo": 3.66087, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.038470560638048}
{"step": 34765, "loss": 3.6643848419189453, "grad_norm": 0.7177697420120239, "learning_rate": 0.0001, "elbo": 3.66438, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0387637813746187}
{"step": 34770, "loss": 3.6664356231689452, "grad_norm": 0.653491735458374, "learning_rate": 0.0001, "elbo": 3.66644, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.039057002111189}
{"step": 34775, "loss": 3.6671268463134767, "grad_norm": 0.7050420045852661, "learning_rate": 0.0001, "elbo": 3.66713, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0393502228477596}
{"step": 34780, "loss": 3.646499252319336, "grad_norm": 0.6679331660270691, "learning_rate": 0.0001, "elbo": 3.6465, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0396434435843305}
{"step": 34785, "loss": 3.657490921020508, "grad_norm": 0.7061704397201538, "learning_rate": 0.0001, "elbo": 3.65749, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.039936664320901}
{"step": 34790, "loss": 3.651598358154297, "grad_norm": 0.6815985441207886, "learning_rate": 0.0001, "elbo": 3.6516, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0402298850574714}
{"step": 34795, "loss": 3.664725494384766, "grad_norm": 0.7395395636558533, "learning_rate": 0.0001, "elbo": 3.66473, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.040523105794042}
{"step": 34800, "loss": 3.6741455078125, "grad_norm": 0.6518201231956482, "learning_rate": 0.0001, "elbo": 3.67415, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0408163265306123}
{"step": 34805, "loss": 3.6704357147216795, "grad_norm": 0.6610407829284668, "learning_rate": 0.0001, "elbo": 3.67044, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0411095472671827}
{"step": 34810, "loss": 3.6786293029785155, "grad_norm": 0.702301025390625, "learning_rate": 0.0001, "elbo": 3.67863, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.041402768003753}
{"step": 34815, "loss": 3.674097442626953, "grad_norm": 0.6500766277313232, "learning_rate": 0.0001, "elbo": 3.6741, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0416959887403237}
{"step": 34820, "loss": 3.66019287109375, "grad_norm": 0.7789403796195984, "learning_rate": 0.0001, "elbo": 3.66019, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.041989209476894}
{"step": 34825, "loss": 3.674565887451172, "grad_norm": 0.6614245772361755, "learning_rate": 0.0001, "elbo": 3.67457, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0422824302134646}
{"step": 34830, "loss": 3.663482666015625, "grad_norm": 0.7117091417312622, "learning_rate": 0.0001, "elbo": 3.66348, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.042575650950035}
{"step": 34835, "loss": 3.665548324584961, "grad_norm": 0.6862500905990601, "learning_rate": 0.0001, "elbo": 3.66555, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0428688716866055}
{"step": 34840, "loss": 3.672687530517578, "grad_norm": 0.7737500071525574, "learning_rate": 0.0001, "elbo": 3.67269, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0431620924231764}
{"step": 34845, "loss": 3.6412139892578126, "grad_norm": 0.6539156436920166, "learning_rate": 0.0001, "elbo": 3.64121, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.043455313159747}
{"step": 34850, "loss": 3.6735591888427734, "grad_norm": 0.6918598413467407, "learning_rate": 0.0001, "elbo": 3.67356, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0437485338963173}
{"step": 34855, "loss": 3.6870471954345705, "grad_norm": 1.370589256286621, "learning_rate": 0.0001, "elbo": 3.68705, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0440417546328877}
{"step": 34860, "loss": 3.6603466033935548, "grad_norm": 0.6991685628890991, "learning_rate": 0.0001, "elbo": 3.66035, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.044334975369458}
{"step": 34865, "loss": 3.6614044189453123, "grad_norm": 0.6559602618217468, "learning_rate": 0.0001, "elbo": 3.6614, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0446281961060286}
{"step": 34870, "loss": 3.650191879272461, "grad_norm": 0.7012059688568115, "learning_rate": 0.0001, "elbo": 3.65019, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.044921416842599}
{"step": 34875, "loss": 3.6797393798828124, "grad_norm": 0.6686696410179138, "learning_rate": 0.0001, "elbo": 3.67974, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0452146375791695}
{"step": 34880, "loss": 3.659189987182617, "grad_norm": 0.7832643389701843, "learning_rate": 0.0001, "elbo": 3.65919, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.04550785831574}
{"step": 34885, "loss": 3.663745880126953, "grad_norm": 0.6798517107963562, "learning_rate": 0.0001, "elbo": 3.66375, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0458010790523105}
{"step": 34890, "loss": 3.6753875732421877, "grad_norm": 0.694196343421936, "learning_rate": 0.0001, "elbo": 3.67539, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.046094299788881}
{"step": 34895, "loss": 3.6641395568847654, "grad_norm": 0.8150795698165894, "learning_rate": 0.0001, "elbo": 3.66414, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0463875205254514}
{"step": 34900, "loss": 3.6634124755859374, "grad_norm": 0.7027676105499268, "learning_rate": 0.0001, "elbo": 3.66341, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0466807412620223}
{"step": 34905, "loss": 3.6807086944580076, "grad_norm": 0.6964035630226135, "learning_rate": 0.0001, "elbo": 3.68071, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0469739619985927}
{"step": 34910, "loss": 3.675384521484375, "grad_norm": 0.7100387811660767, "learning_rate": 0.0001, "elbo": 3.67538, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.047267182735163}
{"step": 34910, "loss": 3.675384521484375, "grad_norm": 0.7100387811660767, "learning_rate": 0.0001, "elbo": 3.67538, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.047267182735163}
{"step": 34915, "loss": 3.6561958312988283, "grad_norm": 0.6619502305984497, "learning_rate": 0.0001, "elbo": 3.6562, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0475604034717336}
{"step": 34920, "loss": 3.663216781616211, "grad_norm": 0.8033750057220459, "learning_rate": 0.0001, "elbo": 3.66322, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.047853624208304}
{"step": 34920, "loss": 3.663216781616211, "grad_norm": 0.8033750057220459, "learning_rate": 0.0001, "elbo": 3.66322, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.047853624208304}
{"step": 34925, "loss": 3.6602714538574217, "grad_norm": 0.6808481812477112, "learning_rate": 0.0001, "elbo": 3.66027, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0481468449448745}
{"step": 34930, "loss": 3.653636169433594, "grad_norm": 0.6831513047218323, "learning_rate": 0.0001, "elbo": 3.65364, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.048440065681445}
{"step": 34935, "loss": 3.645191955566406, "grad_norm": 0.6730350852012634, "learning_rate": 0.0001, "elbo": 3.64519, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0487332864180154}
{"step": 34940, "loss": 3.6565040588378905, "grad_norm": 0.6857802271842957, "learning_rate": 0.0001, "elbo": 3.6565, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.049026507154586}
{"step": 34945, "loss": 3.658345413208008, "grad_norm": 0.6903568506240845, "learning_rate": 0.0001, "elbo": 3.65835, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0493197278911564}
{"step": 34950, "loss": 3.661985397338867, "grad_norm": 0.6459689736366272, "learning_rate": 0.0001, "elbo": 3.66199, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.049612948627727}
{"step": 34955, "loss": 3.6775772094726564, "grad_norm": 0.7212004065513611, "learning_rate": 0.0001, "elbo": 3.67758, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0499061693642973}
{"step": 34960, "loss": 3.6603729248046877, "grad_norm": 0.7050953507423401, "learning_rate": 0.0001, "elbo": 3.66037, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.050199390100868}
{"step": 34960, "loss": 3.6603729248046877, "grad_norm": 0.7050953507423401, "learning_rate": 0.0001, "elbo": 3.66037, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.050199390100868}
{"step": 34965, "loss": 3.639640045166016, "grad_norm": 0.681505024433136, "learning_rate": 0.0001, "elbo": 3.63964, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0504926108374386}
{"step": 34970, "loss": 3.655256652832031, "grad_norm": 0.7476879954338074, "learning_rate": 0.0001, "elbo": 3.65526, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.050785831574009}
{"step": 34970, "loss": 3.655256652832031, "grad_norm": 0.7476879954338074, "learning_rate": 0.0001, "elbo": 3.65526, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.050785831574009}
{"step": 34975, "loss": 3.658731460571289, "grad_norm": 0.6987446546554565, "learning_rate": 0.0001, "elbo": 3.65873, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0510790523105795}
{"step": 34980, "loss": 3.656328582763672, "grad_norm": 0.6901078820228577, "learning_rate": 0.0001, "elbo": 3.65633, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.05137227304715}
{"step": 34985, "loss": 3.6677577972412108, "grad_norm": 0.6603887677192688, "learning_rate": 0.0001, "elbo": 3.66776, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0516654937837204}
{"step": 34990, "loss": 3.6600791931152346, "grad_norm": 0.7154841423034668, "learning_rate": 0.0001, "elbo": 3.66008, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.051958714520291}
{"step": 34995, "loss": 3.6600715637207033, "grad_norm": 0.6606299877166748, "learning_rate": 0.0001, "elbo": 3.66007, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0522519352568613}
{"step": 35000, "loss": 3.652777099609375, "grad_norm": 0.7219039797782898, "learning_rate": 0.0001, "elbo": 3.65278, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.052545155993432}
{"step": 35005, "loss": 3.664063262939453, "grad_norm": 0.7200224995613098, "learning_rate": 0.0001, "elbo": 3.66406, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0528383767300022}
{"step": 35010, "loss": 3.637278747558594, "grad_norm": 0.6318458914756775, "learning_rate": 0.0001, "elbo": 3.63728, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0531315974665727}
{"step": 35015, "loss": 3.6606842041015626, "grad_norm": 0.6707693338394165, "learning_rate": 0.0001, "elbo": 3.66068, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.053424818203143}
{"step": 35020, "loss": 3.664653778076172, "grad_norm": 0.7051633596420288, "learning_rate": 0.0001, "elbo": 3.66465, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0537180389397136}
{"step": 35025, "loss": 3.6531105041503906, "grad_norm": 0.6962789297103882, "learning_rate": 0.0001, "elbo": 3.65311, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0540112596762845}
{"step": 35030, "loss": 3.6711769104003906, "grad_norm": 0.7737769484519958, "learning_rate": 0.0001, "elbo": 3.67118, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.054304480412855}
{"step": 35035, "loss": 3.648151397705078, "grad_norm": 0.6435344219207764, "learning_rate": 0.0001, "elbo": 3.64815, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0545977011494254}
{"step": 35040, "loss": 3.6487548828125, "grad_norm": 0.8414318561553955, "learning_rate": 0.0001, "elbo": 3.64876, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.054890921885996}
{"step": 35045, "loss": 3.6669788360595703, "grad_norm": 0.6542972326278687, "learning_rate": 0.0001, "elbo": 3.66698, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0551841426225663}
{"step": 35050, "loss": 3.6543594360351563, "grad_norm": 0.7405425310134888, "learning_rate": 0.0001, "elbo": 3.65436, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0554773633591368}
{"step": 35055, "loss": 3.6605911254882812, "grad_norm": 0.7008076310157776, "learning_rate": 0.0001, "elbo": 3.66059, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0557705840957072}
{"step": 35060, "loss": 3.6634254455566406, "grad_norm": 0.7106281518936157, "learning_rate": 0.0001, "elbo": 3.66343, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0560638048322777}
{"step": 35060, "loss": 3.6634254455566406, "grad_norm": 0.7106281518936157, "learning_rate": 0.0001, "elbo": 3.66343, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0560638048322777}
{"step": 35065, "loss": 3.651433563232422, "grad_norm": 0.7375267148017883, "learning_rate": 0.0001, "elbo": 3.65143, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.056357025568848}
{"step": 35070, "loss": 3.6512592315673826, "grad_norm": 0.6824416518211365, "learning_rate": 0.0001, "elbo": 3.65126, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0566502463054186}
{"step": 35075, "loss": 3.659290313720703, "grad_norm": 0.6776857972145081, "learning_rate": 0.0001, "elbo": 3.65929, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.056943467041989}
{"step": 35080, "loss": 3.6600128173828126, "grad_norm": 0.7790823578834534, "learning_rate": 0.0001, "elbo": 3.66001, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0572366877785595}
{"step": 35085, "loss": 3.6550193786621095, "grad_norm": 0.7092888355255127, "learning_rate": 0.0001, "elbo": 3.65502, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0575299085151304}
{"step": 35090, "loss": 3.6537410736083986, "grad_norm": 0.6611467003822327, "learning_rate": 0.0001, "elbo": 3.65374, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.057823129251701}
{"step": 35095, "loss": 3.662770080566406, "grad_norm": 0.6934843063354492, "learning_rate": 0.0001, "elbo": 3.66277, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0581163499882713}
{"step": 35100, "loss": 3.649575424194336, "grad_norm": 0.6972343325614929, "learning_rate": 0.0001, "elbo": 3.64958, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0584095707248418}
{"step": 35105, "loss": 3.645504379272461, "grad_norm": 0.6679956912994385, "learning_rate": 0.0001, "elbo": 3.6455, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.058702791461412}
{"step": 35110, "loss": 3.6567562103271483, "grad_norm": 0.6966813206672668, "learning_rate": 0.0001, "elbo": 3.65676, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0589960121979827}
{"step": 35115, "loss": 3.6658203125, "grad_norm": 0.7221479415893555, "learning_rate": 0.0001, "elbo": 3.66582, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.059289232934553}
{"step": 35120, "loss": 3.66619987487793, "grad_norm": 0.7184253334999084, "learning_rate": 0.0001, "elbo": 3.6662, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0595824536711236}
{"step": 35125, "loss": 3.6582626342773437, "grad_norm": 0.6991737484931946, "learning_rate": 0.0001, "elbo": 3.65826, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.059875674407694}
{"step": 35130, "loss": 3.6637714385986326, "grad_norm": 0.727806031703949, "learning_rate": 0.0001, "elbo": 3.66377, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0601688951442645}
{"step": 35135, "loss": 3.6643280029296874, "grad_norm": 0.729875922203064, "learning_rate": 0.0001, "elbo": 3.66433, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.060462115880835}
{"step": 35140, "loss": 3.6630088806152346, "grad_norm": 0.654555082321167, "learning_rate": 0.0001, "elbo": 3.66301, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0607553366174054}
{"step": 35145, "loss": 3.667549896240234, "grad_norm": 0.6949908137321472, "learning_rate": 0.0001, "elbo": 3.66755, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.061048557353976}
{"step": 35150, "loss": 3.6526721954345702, "grad_norm": 0.6440150737762451, "learning_rate": 0.0001, "elbo": 3.65267, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0613417780905468}
{"step": 35155, "loss": 3.6708518981933596, "grad_norm": 1.0060725212097168, "learning_rate": 0.0001, "elbo": 3.67085, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.061634998827117}
{"step": 35160, "loss": 3.658286285400391, "grad_norm": 0.7543525695800781, "learning_rate": 0.0001, "elbo": 3.65829, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0619282195636877}
{"step": 35165, "loss": 3.6654922485351564, "grad_norm": 0.6791113018989563, "learning_rate": 0.0001, "elbo": 3.66549, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.062221440300258}
{"step": 35170, "loss": 3.6673904418945313, "grad_norm": 0.6660560369491577, "learning_rate": 0.0001, "elbo": 3.66739, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0625146610368286}
{"step": 35175, "loss": 3.6704429626464843, "grad_norm": 0.6772493124008179, "learning_rate": 0.0001, "elbo": 3.67044, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.062807881773399}
{"step": 35180, "loss": 3.6533077239990233, "grad_norm": 0.7252959609031677, "learning_rate": 0.0001, "elbo": 3.65331, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0631011025099695}
{"step": 35185, "loss": 3.673029327392578, "grad_norm": 0.6904544234275818, "learning_rate": 0.0001, "elbo": 3.67303, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.06339432324654}
{"step": 35190, "loss": 3.659026336669922, "grad_norm": 0.6741441488265991, "learning_rate": 0.0001, "elbo": 3.65903, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0636875439831104}
{"step": 35195, "loss": 3.674264907836914, "grad_norm": 0.6953937411308289, "learning_rate": 0.0001, "elbo": 3.67426, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.063980764719681}
{"step": 35200, "loss": 3.6624549865722655, "grad_norm": 0.6520687341690063, "learning_rate": 0.0001, "elbo": 3.66246, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0642739854562513}
{"step": 35205, "loss": 3.6663379669189453, "grad_norm": 0.6628921627998352, "learning_rate": 0.0001, "elbo": 3.66634, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0645672061928217}
{"step": 35210, "loss": 3.6619773864746095, "grad_norm": 0.8011850714683533, "learning_rate": 0.0001, "elbo": 3.66198, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0648604269293926}
{"step": 35215, "loss": 3.652422332763672, "grad_norm": 0.6757350564002991, "learning_rate": 0.0001, "elbo": 3.65242, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.065153647665963}
{"step": 35220, "loss": 3.656389617919922, "grad_norm": 0.6514265537261963, "learning_rate": 0.0001, "elbo": 3.65639, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0654468684025336}
{"step": 35225, "loss": 3.6581035614013673, "grad_norm": 0.6630225777626038, "learning_rate": 0.0001, "elbo": 3.6581, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.065740089139104}
{"step": 35230, "loss": 3.664690399169922, "grad_norm": 0.6932056546211243, "learning_rate": 0.0001, "elbo": 3.66469, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0660333098756745}
{"step": 35235, "loss": 3.6502151489257812, "grad_norm": 0.9710474610328674, "learning_rate": 0.0001, "elbo": 3.65022, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.066326530612245}
{"step": 35240, "loss": 3.6691795349121095, "grad_norm": 0.7097776532173157, "learning_rate": 0.0001, "elbo": 3.66918, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0666197513488154}
{"step": 35240, "loss": 3.6691795349121095, "grad_norm": 0.7097776532173157, "learning_rate": 0.0001, "elbo": 3.66918, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0666197513488154}
{"step": 35245, "loss": 3.6809162139892577, "grad_norm": 0.6930609941482544, "learning_rate": 0.0001, "elbo": 3.68092, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.066912972085386}
{"step": 35250, "loss": 3.6679931640625, "grad_norm": 0.9194359183311462, "learning_rate": 0.0001, "elbo": 3.66799, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0672061928219563}
{"step": 35255, "loss": 3.659424591064453, "grad_norm": 0.8462197780609131, "learning_rate": 0.0001, "elbo": 3.65942, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0674994135585267}
{"step": 35260, "loss": 3.658052444458008, "grad_norm": 0.7051244974136353, "learning_rate": 0.0001, "elbo": 3.65805, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.067792634295097}
{"step": 35265, "loss": 3.6529083251953125, "grad_norm": 0.6930456757545471, "learning_rate": 0.0001, "elbo": 3.65291, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0680858550316676}
{"step": 35270, "loss": 3.6502334594726564, "grad_norm": 0.6880565285682678, "learning_rate": 0.0001, "elbo": 3.65023, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0683790757682385}
{"step": 35275, "loss": 3.6597213745117188, "grad_norm": 0.7113927006721497, "learning_rate": 0.0001, "elbo": 3.65972, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.068672296504809}
{"step": 35280, "loss": 3.65142822265625, "grad_norm": 0.6496366262435913, "learning_rate": 0.0001, "elbo": 3.65143, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0689655172413794}
{"step": 35285, "loss": 3.655488967895508, "grad_norm": 0.8785625100135803, "learning_rate": 0.0001, "elbo": 3.65549, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.06925873797795}
{"step": 35290, "loss": 3.6714504241943358, "grad_norm": 0.6494368314743042, "learning_rate": 0.0001, "elbo": 3.67145, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0695519587145204}
{"step": 35295, "loss": 3.652703857421875, "grad_norm": 0.6344048380851746, "learning_rate": 0.0001, "elbo": 3.6527, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.069845179451091}
{"step": 35300, "loss": 3.6498008728027345, "grad_norm": 0.7398649454116821, "learning_rate": 0.0001, "elbo": 3.6498, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0701384001876613}
{"step": 35305, "loss": 3.6555137634277344, "grad_norm": 0.6921533346176147, "learning_rate": 0.0001, "elbo": 3.65551, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0704316209242317}
{"step": 35310, "loss": 3.650531005859375, "grad_norm": 0.6490051746368408, "learning_rate": 0.0001, "elbo": 3.65053, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.070724841660802}
{"step": 35315, "loss": 3.668553924560547, "grad_norm": 0.7050113677978516, "learning_rate": 0.0001, "elbo": 3.66855, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0710180623973726}
{"step": 35320, "loss": 3.6709030151367186, "grad_norm": 0.6741743683815002, "learning_rate": 0.0001, "elbo": 3.6709, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.071311283133943}
{"step": 35325, "loss": 3.664284896850586, "grad_norm": 0.7205178141593933, "learning_rate": 0.0001, "elbo": 3.66428, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0716045038705135}
{"step": 35330, "loss": 3.6423526763916017, "grad_norm": 0.7888270020484924, "learning_rate": 0.0001, "elbo": 3.64235, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0718977246070844}
{"step": 35335, "loss": 3.653095245361328, "grad_norm": 0.7130043506622314, "learning_rate": 0.0001, "elbo": 3.6531, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.072190945343655}
{"step": 35340, "loss": 3.6498592376708983, "grad_norm": 0.7016117572784424, "learning_rate": 0.0001, "elbo": 3.64986, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0724841660802253}
{"step": 35345, "loss": 3.6573081970214845, "grad_norm": 0.715627908706665, "learning_rate": 0.0001, "elbo": 3.65731, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.072777386816796}
{"step": 35350, "loss": 3.670096588134766, "grad_norm": 0.6807551383972168, "learning_rate": 0.0001, "elbo": 3.6701, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0730706075533663}
{"step": 35355, "loss": 3.682252883911133, "grad_norm": 0.7310425043106079, "learning_rate": 0.0001, "elbo": 3.68225, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0733638282899367}
{"step": 35360, "loss": 3.685655212402344, "grad_norm": 0.7801418304443359, "learning_rate": 0.0001, "elbo": 3.68566, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.073657049026507}
{"step": 35365, "loss": 3.6689048767089845, "grad_norm": 0.6604978442192078, "learning_rate": 0.0001, "elbo": 3.6689, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0739502697630776}
{"step": 35370, "loss": 3.6533058166503904, "grad_norm": 0.820389986038208, "learning_rate": 0.0001, "elbo": 3.65331, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.074243490499648}
{"step": 35375, "loss": 3.663824462890625, "grad_norm": 0.6660683155059814, "learning_rate": 0.0001, "elbo": 3.66382, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0745367112362185}
{"step": 35380, "loss": 3.6479698181152345, "grad_norm": 0.8967122435569763, "learning_rate": 0.0001, "elbo": 3.64797, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.074829931972789}
{"step": 35385, "loss": 3.6681869506835936, "grad_norm": 0.7282879948616028, "learning_rate": 0.0001, "elbo": 3.66819, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0751231527093594}
{"step": 35390, "loss": 3.6613277435302733, "grad_norm": 0.9086153507232666, "learning_rate": 0.0001, "elbo": 3.66133, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.07541637344593}
{"step": 35395, "loss": 3.6764080047607424, "grad_norm": 0.6656813025474548, "learning_rate": 0.0001, "elbo": 3.67641, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.075709594182501}
{"step": 35400, "loss": 3.6762451171875, "grad_norm": 0.6718469262123108, "learning_rate": 0.0001, "elbo": 3.67625, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0760028149190712}
{"step": 35405, "loss": 3.6771923065185548, "grad_norm": 0.7064347863197327, "learning_rate": 0.0001, "elbo": 3.67719, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0762960356556417}
{"step": 35410, "loss": 3.665080261230469, "grad_norm": 2.070650339126587, "learning_rate": 0.0001, "elbo": 3.66508, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.076589256392212}
{"step": 35415, "loss": 3.675409698486328, "grad_norm": 0.6697618961334229, "learning_rate": 0.0001, "elbo": 3.67541, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0768824771287826}
{"step": 35420, "loss": 3.6643608093261717, "grad_norm": 0.7361375689506531, "learning_rate": 0.0001, "elbo": 3.66436, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.077175697865353}
{"step": 35425, "loss": 3.6585201263427733, "grad_norm": 0.626900315284729, "learning_rate": 0.0001, "elbo": 3.65852, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0774689186019235}
{"step": 35430, "loss": 3.645948028564453, "grad_norm": 0.6793298721313477, "learning_rate": 0.0001, "elbo": 3.64595, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.077762139338494}
{"step": 35435, "loss": 3.6593917846679687, "grad_norm": 0.7109951972961426, "learning_rate": 0.0001, "elbo": 3.65939, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0780553600750644}
{"step": 35435, "loss": 3.6593917846679687, "grad_norm": 0.7109951972961426, "learning_rate": 0.0001, "elbo": 3.65939, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0780553600750644}
{"step": 35440, "loss": 3.682811737060547, "grad_norm": 0.6696482300758362, "learning_rate": 0.0001, "elbo": 3.68281, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.078348580811635}
{"step": 35445, "loss": 3.641016387939453, "grad_norm": 0.633562445640564, "learning_rate": 0.0001, "elbo": 3.64102, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0786418015482053}
{"step": 35450, "loss": 3.661666488647461, "grad_norm": 0.6859925985336304, "learning_rate": 0.0001, "elbo": 3.66167, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.078935022284776}
{"step": 35455, "loss": 3.6743648529052733, "grad_norm": 0.6925618648529053, "learning_rate": 0.0001, "elbo": 3.67436, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0792282430213467}
{"step": 35460, "loss": 3.6622604370117187, "grad_norm": 0.7042040228843689, "learning_rate": 0.0001, "elbo": 3.66226, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.079521463757917}
{"step": 35465, "loss": 3.6714508056640627, "grad_norm": 0.6926333904266357, "learning_rate": 0.0001, "elbo": 3.67145, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0798146844944876}
{"step": 35470, "loss": 3.6714187622070313, "grad_norm": 0.7721909880638123, "learning_rate": 0.0001, "elbo": 3.67142, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.080107905231058}
{"step": 35475, "loss": 3.6564208984375, "grad_norm": 0.6702327132225037, "learning_rate": 0.0001, "elbo": 3.65642, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0804011259676285}
{"step": 35480, "loss": 3.6746707916259767, "grad_norm": 0.6761677265167236, "learning_rate": 0.0001, "elbo": 3.67467, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.080694346704199}
{"step": 35485, "loss": 3.6345863342285156, "grad_norm": 0.801354706287384, "learning_rate": 0.0001, "elbo": 3.63459, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0809875674407694}
{"step": 35490, "loss": 3.661431884765625, "grad_norm": 0.6847831606864929, "learning_rate": 0.0001, "elbo": 3.66143, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.08128078817734}
{"step": 35495, "loss": 3.631936264038086, "grad_norm": 0.7388131022453308, "learning_rate": 0.0001, "elbo": 3.63194, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0815740089139103}
{"step": 35500, "loss": 3.657711410522461, "grad_norm": 0.6613762378692627, "learning_rate": 0.0001, "elbo": 3.65771, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0818672296504808}
{"step": 35505, "loss": 3.678881072998047, "grad_norm": 0.6982714533805847, "learning_rate": 0.0001, "elbo": 3.67888, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.082160450387051}
{"step": 35510, "loss": 3.6737392425537108, "grad_norm": 0.6619595289230347, "learning_rate": 0.0001, "elbo": 3.67374, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0824536711236217}
{"step": 35515, "loss": 3.663555908203125, "grad_norm": 0.6814638376235962, "learning_rate": 0.0001, "elbo": 3.66356, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.082746891860192}
{"step": 35520, "loss": 3.6733802795410155, "grad_norm": 0.6441161632537842, "learning_rate": 0.0001, "elbo": 3.67338, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.083040112596763}
{"step": 35525, "loss": 3.6751144409179686, "grad_norm": 0.8649734258651733, "learning_rate": 0.0001, "elbo": 3.67511, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0833333333333335}
{"step": 35530, "loss": 3.6522796630859373, "grad_norm": 0.742602527141571, "learning_rate": 0.0001, "elbo": 3.65228, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.083626554069904}
{"step": 35535, "loss": 3.6571529388427733, "grad_norm": 0.7728100419044495, "learning_rate": 0.0001, "elbo": 3.65715, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0839197748064744}
{"step": 35540, "loss": 3.6720863342285157, "grad_norm": 0.7509654760360718, "learning_rate": 0.0001, "elbo": 3.67209, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.084212995543045}
{"step": 35545, "loss": 3.655847930908203, "grad_norm": 0.6848572492599487, "learning_rate": 0.0001, "elbo": 3.65585, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0845062162796153}
{"step": 35550, "loss": 3.6578369140625, "grad_norm": 0.8973808288574219, "learning_rate": 0.0001, "elbo": 3.65784, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0847994370161858}
{"step": 35555, "loss": 3.6752105712890626, "grad_norm": 0.8314359188079834, "learning_rate": 0.0001, "elbo": 3.67521, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.085092657752756}
{"step": 35560, "loss": 3.6767288208007813, "grad_norm": 1.1761480569839478, "learning_rate": 0.0001, "elbo": 3.67673, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0853858784893267}
{"step": 35565, "loss": 3.6600685119628906, "grad_norm": 0.6770240664482117, "learning_rate": 0.0001, "elbo": 3.66007, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.085679099225897}
{"step": 35570, "loss": 3.673406982421875, "grad_norm": 0.6968743205070496, "learning_rate": 0.0001, "elbo": 3.67341, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0859723199624676}
{"step": 35575, "loss": 3.644316864013672, "grad_norm": 0.6559906005859375, "learning_rate": 0.0001, "elbo": 3.64432, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.086265540699038}
{"step": 35580, "loss": 3.6492305755615235, "grad_norm": 0.6866732239723206, "learning_rate": 0.0001, "elbo": 3.64923, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.086558761435609}
{"step": 35585, "loss": 3.6713516235351564, "grad_norm": 0.7157118916511536, "learning_rate": 0.0001, "elbo": 3.67135, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0868519821721794}
{"step": 35590, "loss": 3.6478530883789064, "grad_norm": 0.834187924861908, "learning_rate": 0.0001, "elbo": 3.64785, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.08714520290875}
{"step": 35595, "loss": 3.6507057189941405, "grad_norm": 0.6511818170547485, "learning_rate": 0.0001, "elbo": 3.65071, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0874384236453203}
{"step": 35600, "loss": 3.6700721740722657, "grad_norm": 0.6594821214675903, "learning_rate": 0.0001, "elbo": 3.67007, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0877316443818907}
{"step": 35605, "loss": 3.6616512298583985, "grad_norm": 0.8960957527160645, "learning_rate": 0.0001, "elbo": 3.66165, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.088024865118461}
{"step": 35610, "loss": 3.6658660888671877, "grad_norm": 1.1131105422973633, "learning_rate": 0.0001, "elbo": 3.66587, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0883180858550316}
{"step": 35615, "loss": 3.656005096435547, "grad_norm": 0.6963775157928467, "learning_rate": 0.0001, "elbo": 3.65601, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.088611306591602}
{"step": 35620, "loss": 3.673173522949219, "grad_norm": 0.6586289405822754, "learning_rate": 0.0001, "elbo": 3.67317, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0889045273281726}
{"step": 35625, "loss": 3.6432743072509766, "grad_norm": 0.7189998030662537, "learning_rate": 0.0001, "elbo": 3.64327, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.089197748064743}
{"step": 35630, "loss": 3.643598175048828, "grad_norm": 0.7397767901420593, "learning_rate": 0.0001, "elbo": 3.6436, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0894909688013135}
{"step": 35635, "loss": 3.6422637939453124, "grad_norm": 0.7335779666900635, "learning_rate": 0.0001, "elbo": 3.64226, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.089784189537884}
{"step": 35640, "loss": 3.658197784423828, "grad_norm": 0.6837426424026489, "learning_rate": 0.0001, "elbo": 3.6582, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.090077410274455}
{"step": 35645, "loss": 3.64428825378418, "grad_norm": 0.6741697192192078, "learning_rate": 0.0001, "elbo": 3.64429, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0903706310110253}
{"step": 35645, "loss": 3.64428825378418, "grad_norm": 0.6741697192192078, "learning_rate": 0.0001, "elbo": 3.64429, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0903706310110253}
{"step": 35650, "loss": 3.641606903076172, "grad_norm": 0.7853777408599854, "learning_rate": 0.0001, "elbo": 3.64161, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0906638517475957}
{"step": 35655, "loss": 3.6408931732177736, "grad_norm": 0.667189359664917, "learning_rate": 0.0001, "elbo": 3.64089, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.090957072484166}
{"step": 35660, "loss": 3.6548080444335938, "grad_norm": 0.6910833120346069, "learning_rate": 0.0001, "elbo": 3.65481, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0912502932207366}
{"step": 35665, "loss": 3.659557342529297, "grad_norm": 0.6905349493026733, "learning_rate": 0.0001, "elbo": 3.65956, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.091543513957307}
{"step": 35670, "loss": 3.6477161407470704, "grad_norm": 0.7548152804374695, "learning_rate": 0.0001, "elbo": 3.64772, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0918367346938775}
{"step": 35675, "loss": 3.643602752685547, "grad_norm": 0.6619182825088501, "learning_rate": 0.0001, "elbo": 3.6436, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.092129955430448}
{"step": 35680, "loss": 3.654688262939453, "grad_norm": 0.7182955145835876, "learning_rate": 0.0001, "elbo": 3.65469, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0924231761670185}
{"step": 35685, "loss": 3.6636627197265623, "grad_norm": 0.6877469420433044, "learning_rate": 0.0001, "elbo": 3.66366, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.092716396903589}
{"step": 35690, "loss": 3.6634849548339843, "grad_norm": 0.7180483937263489, "learning_rate": 0.0001, "elbo": 3.66349, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0930096176401594}
{"step": 35695, "loss": 3.6450401306152345, "grad_norm": 0.6949020624160767, "learning_rate": 0.0001, "elbo": 3.64504, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.09330283837673}
{"step": 35700, "loss": 3.664447021484375, "grad_norm": 0.7395868301391602, "learning_rate": 0.0001, "elbo": 3.66445, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0935960591133007}
{"step": 35705, "loss": 3.6797405242919923, "grad_norm": 0.6925127506256104, "learning_rate": 0.0001, "elbo": 3.67974, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.093889279849871}
{"step": 35710, "loss": 3.6686447143554686, "grad_norm": 0.7624085545539856, "learning_rate": 0.0001, "elbo": 3.66864, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0941825005864416}
{"step": 35715, "loss": 3.651612091064453, "grad_norm": 0.9149665832519531, "learning_rate": 0.0001, "elbo": 3.65161, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.094475721323012}
{"step": 35720, "loss": 3.680217742919922, "grad_norm": 0.6781027913093567, "learning_rate": 0.0001, "elbo": 3.68022, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0947689420595825}
{"step": 35725, "loss": 3.6591278076171876, "grad_norm": 0.8711135387420654, "learning_rate": 0.0001, "elbo": 3.65913, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.095062162796153}
{"step": 35730, "loss": 3.648695373535156, "grad_norm": 0.6743937730789185, "learning_rate": 0.0001, "elbo": 3.6487, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0953553835327234}
{"step": 35735, "loss": 3.658478927612305, "grad_norm": 0.6802875399589539, "learning_rate": 0.0001, "elbo": 3.65848, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.095648604269294}
{"step": 35740, "loss": 3.6558464050292967, "grad_norm": 0.6365288496017456, "learning_rate": 0.0001, "elbo": 3.65585, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0959418250058643}
{"step": 35745, "loss": 3.6551918029785155, "grad_norm": 0.6686568856239319, "learning_rate": 0.0001, "elbo": 3.65519, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.096235045742435}
{"step": 35750, "loss": 3.650374984741211, "grad_norm": 0.7277258634567261, "learning_rate": 0.0001, "elbo": 3.65037, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0965282664790053}
{"step": 35755, "loss": 3.6647953033447265, "grad_norm": 0.6805939674377441, "learning_rate": 0.0001, "elbo": 3.6648, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0968214872155757}
{"step": 35760, "loss": 3.675332260131836, "grad_norm": 0.6997017860412598, "learning_rate": 0.0001, "elbo": 3.67533, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.097114707952146}
{"step": 35765, "loss": 3.6715419769287108, "grad_norm": 0.7291285395622253, "learning_rate": 0.0001, "elbo": 3.67154, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.097407928688717}
{"step": 35770, "loss": 3.668080139160156, "grad_norm": 0.684264063835144, "learning_rate": 0.0001, "elbo": 3.66808, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0977011494252875}
{"step": 35770, "loss": 3.668080139160156, "grad_norm": 0.684264063835144, "learning_rate": 0.0001, "elbo": 3.66808, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0977011494252875}
{"step": 35775, "loss": 3.6617752075195313, "grad_norm": 0.6896786689758301, "learning_rate": 0.0001, "elbo": 3.66178, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.097994370161858}
{"step": 35780, "loss": 3.667683410644531, "grad_norm": 0.6656774878501892, "learning_rate": 0.0001, "elbo": 3.66768, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0982875908984284}
{"step": 35785, "loss": 3.6688827514648437, "grad_norm": 0.7116312980651855, "learning_rate": 0.0001, "elbo": 3.66888, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.098580811634999}
{"step": 35790, "loss": 3.6437759399414062, "grad_norm": 0.638626217842102, "learning_rate": 0.0001, "elbo": 3.64378, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0988740323715693}
{"step": 35795, "loss": 3.665857696533203, "grad_norm": 0.6893758177757263, "learning_rate": 0.0001, "elbo": 3.66586, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.09916725310814}
{"step": 35800, "loss": 3.656085205078125, "grad_norm": 0.660578191280365, "learning_rate": 0.0001, "elbo": 3.65608, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0994604738447102}
{"step": 35805, "loss": 3.6682273864746096, "grad_norm": 0.654731035232544, "learning_rate": 0.0001, "elbo": 3.66823, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.0997536945812807}
{"step": 35810, "loss": 3.656366729736328, "grad_norm": 0.6450804471969604, "learning_rate": 0.0001, "elbo": 3.65637, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.100046915317851}
{"step": 35815, "loss": 3.6546531677246095, "grad_norm": 0.7167596220970154, "learning_rate": 0.0001, "elbo": 3.65465, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1003401360544216}
{"step": 35820, "loss": 3.6457115173339845, "grad_norm": 0.743667721748352, "learning_rate": 0.0001, "elbo": 3.64571, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.100633356790992}
{"step": 35825, "loss": 3.6358211517333983, "grad_norm": 0.7447308301925659, "learning_rate": 0.0001, "elbo": 3.63582, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.100926577527563}
{"step": 35830, "loss": 3.659880828857422, "grad_norm": 0.7366108298301697, "learning_rate": 0.0001, "elbo": 3.65988, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1012197982641334}
{"step": 35835, "loss": 3.6473995208740235, "grad_norm": 0.7084254026412964, "learning_rate": 0.0001, "elbo": 3.6474, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.101513019000704}
{"step": 35840, "loss": 3.6478195190429688, "grad_norm": 1.3243904113769531, "learning_rate": 0.0001, "elbo": 3.64782, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1018062397372743}
{"step": 35845, "loss": 3.6583656311035155, "grad_norm": 0.8325717449188232, "learning_rate": 0.0001, "elbo": 3.65837, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1020994604738448}
{"step": 35850, "loss": 3.6706111907958983, "grad_norm": 0.7150986790657043, "learning_rate": 0.0001, "elbo": 3.67061, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1023926812104152}
{"step": 35855, "loss": 3.6607337951660157, "grad_norm": 0.6539914011955261, "learning_rate": 0.0001, "elbo": 3.66073, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1026859019469857}
{"step": 35860, "loss": 3.670125961303711, "grad_norm": 0.6753908395767212, "learning_rate": 0.0001, "elbo": 3.67013, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.102979122683556}
{"step": 35865, "loss": 3.6744056701660157, "grad_norm": 0.670038104057312, "learning_rate": 0.0001, "elbo": 3.67441, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1032723434201266}
{"step": 35870, "loss": 3.661068344116211, "grad_norm": 0.6665202975273132, "learning_rate": 0.0001, "elbo": 3.66107, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.103565564156697}
{"step": 35875, "loss": 3.6387706756591798, "grad_norm": 0.6671241521835327, "learning_rate": 0.0001, "elbo": 3.63877, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1038587848932675}
{"step": 35880, "loss": 3.6553436279296876, "grad_norm": 0.7200987339019775, "learning_rate": 0.0001, "elbo": 3.65534, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.104152005629838}
{"step": 35885, "loss": 3.6670181274414064, "grad_norm": 0.6986860036849976, "learning_rate": 0.0001, "elbo": 3.66702, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1044452263664084}
{"step": 35890, "loss": 3.651849365234375, "grad_norm": 0.6945321559906006, "learning_rate": 0.0001, "elbo": 3.65185, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1047384471029793}
{"step": 35895, "loss": 3.6642478942871093, "grad_norm": 0.6725345849990845, "learning_rate": 0.0001, "elbo": 3.66425, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1050316678395498}
{"step": 35900, "loss": 3.6399688720703125, "grad_norm": 1.0081160068511963, "learning_rate": 0.0001, "elbo": 3.63997, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.10532488857612}
{"step": 35905, "loss": 3.632080841064453, "grad_norm": 0.6790771484375, "learning_rate": 0.0001, "elbo": 3.63208, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1056181093126907}
{"step": 35910, "loss": 3.6727195739746095, "grad_norm": 0.7832374572753906, "learning_rate": 0.0001, "elbo": 3.67272, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.105911330049261}
{"step": 35910, "loss": 3.6727195739746095, "grad_norm": 0.7832374572753906, "learning_rate": 0.0001, "elbo": 3.67272, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.105911330049261}
{"step": 35915, "loss": 3.664255142211914, "grad_norm": 0.6701896786689758, "learning_rate": 0.0001, "elbo": 3.66425, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1062045507858316}
{"step": 35920, "loss": 3.6593841552734374, "grad_norm": 0.7877994775772095, "learning_rate": 0.0001, "elbo": 3.65938, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.106497771522402}
{"step": 35925, "loss": 3.677529525756836, "grad_norm": 0.6801364421844482, "learning_rate": 0.0001, "elbo": 3.67753, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1067909922589725}
{"step": 35930, "loss": 3.6532806396484374, "grad_norm": 0.8003167510032654, "learning_rate": 0.0001, "elbo": 3.65328, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.107084212995543}
{"step": 35935, "loss": 3.658518981933594, "grad_norm": 0.7328919768333435, "learning_rate": 0.0001, "elbo": 3.65852, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1073774337321134}
{"step": 35940, "loss": 3.668618392944336, "grad_norm": 0.6823623180389404, "learning_rate": 0.0001, "elbo": 3.66862, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.107670654468684}
{"step": 35945, "loss": 3.667026138305664, "grad_norm": 0.7995436787605286, "learning_rate": 0.0001, "elbo": 3.66703, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1079638752052547}
{"step": 35950, "loss": 3.683565139770508, "grad_norm": 0.8268135190010071, "learning_rate": 0.0001, "elbo": 3.68357, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.108257095941825}
{"step": 35955, "loss": 3.651959228515625, "grad_norm": 0.7587167620658875, "learning_rate": 0.0001, "elbo": 3.65196, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1085503166783957}
{"step": 35960, "loss": 3.650030517578125, "grad_norm": 0.6991404294967651, "learning_rate": 0.0001, "elbo": 3.65003, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.108843537414966}
{"step": 35965, "loss": 3.6721427917480467, "grad_norm": 0.8099517226219177, "learning_rate": 0.0001, "elbo": 3.67214, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1091367581515366}
{"step": 35970, "loss": 3.654539108276367, "grad_norm": 0.963527500629425, "learning_rate": 0.0001, "elbo": 3.65454, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.109429978888107}
{"step": 35975, "loss": 3.6477760314941405, "grad_norm": 0.6962860226631165, "learning_rate": 0.0001, "elbo": 3.64778, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1097231996246775}
{"step": 35980, "loss": 3.662493133544922, "grad_norm": 0.7123446464538574, "learning_rate": 0.0001, "elbo": 3.66249, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.110016420361248}
{"step": 35985, "loss": 3.6409675598144533, "grad_norm": 0.7564511299133301, "learning_rate": 0.0001, "elbo": 3.64097, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1103096410978184}
{"step": 35985, "loss": 3.6409675598144533, "grad_norm": 0.7564511299133301, "learning_rate": 0.0001, "elbo": 3.64097, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1103096410978184}
{"step": 35990, "loss": 3.6545635223388673, "grad_norm": 0.6848195791244507, "learning_rate": 0.0001, "elbo": 3.65456, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.110602861834389}
{"step": 35995, "loss": 3.6542640686035157, "grad_norm": 0.6445196270942688, "learning_rate": 0.0001, "elbo": 3.65426, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1108960825709593}
{"step": 36000, "loss": 3.6504669189453125, "grad_norm": 0.7252827286720276, "learning_rate": 0.0001, "elbo": 3.65047, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1111893033075297}
{"step": 36005, "loss": 3.6511154174804688, "grad_norm": 0.8028665781021118, "learning_rate": 0.0001, "elbo": 3.65112, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1114825240441}
{"step": 36010, "loss": 3.6404327392578124, "grad_norm": 0.6727950572967529, "learning_rate": 0.0001, "elbo": 3.64043, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.111775744780671}
{"step": 36015, "loss": 3.6404003143310546, "grad_norm": 0.6785428524017334, "learning_rate": 0.0001, "elbo": 3.6404, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1120689655172415}
{"step": 36020, "loss": 3.6505619049072267, "grad_norm": 0.8557805418968201, "learning_rate": 0.0001, "elbo": 3.65056, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.112362186253812}
{"step": 36025, "loss": 3.666995620727539, "grad_norm": 0.6701478958129883, "learning_rate": 0.0001, "elbo": 3.667, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1126554069903825}
{"step": 36025, "loss": 3.666995620727539, "grad_norm": 0.6701478958129883, "learning_rate": 0.0001, "elbo": 3.667, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1126554069903825}
{"step": 36030, "loss": 3.6554000854492186, "grad_norm": 0.698979377746582, "learning_rate": 0.0001, "elbo": 3.6554, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.112948627726953}
{"step": 36035, "loss": 3.646073913574219, "grad_norm": 0.7373290061950684, "learning_rate": 0.0001, "elbo": 3.64607, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1132418484635234}
{"step": 36035, "loss": 3.646073913574219, "grad_norm": 0.7373290061950684, "learning_rate": 0.0001, "elbo": 3.64607, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1132418484635234}
{"step": 36040, "loss": 3.644957733154297, "grad_norm": 0.8190994262695312, "learning_rate": 0.0001, "elbo": 3.64496, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.113535069200094}
{"step": 36045, "loss": 3.658371353149414, "grad_norm": 0.7448775172233582, "learning_rate": 0.0001, "elbo": 3.65837, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1138282899366643}
{"step": 36050, "loss": 3.6629108428955077, "grad_norm": 0.964061975479126, "learning_rate": 0.0001, "elbo": 3.66291, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1141215106732347}
{"step": 36055, "loss": 3.653182601928711, "grad_norm": 0.6949799060821533, "learning_rate": 0.0001, "elbo": 3.65318, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.114414731409805}
{"step": 36060, "loss": 3.6622314453125, "grad_norm": 0.7117806673049927, "learning_rate": 0.0001, "elbo": 3.66223, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1147079521463756}
{"step": 36065, "loss": 3.647506332397461, "grad_norm": 0.6614134907722473, "learning_rate": 0.0001, "elbo": 3.64751, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.115001172882946}
{"step": 36065, "loss": 3.647506332397461, "grad_norm": 0.6614134907722473, "learning_rate": 0.0001, "elbo": 3.64751, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.115001172882946}
{"step": 36070, "loss": 3.676575469970703, "grad_norm": 0.6837060451507568, "learning_rate": 0.0001, "elbo": 3.67658, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.115294393619517}
{"step": 36075, "loss": 3.6397560119628904, "grad_norm": 0.6922591328620911, "learning_rate": 0.0001, "elbo": 3.63976, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1155876143560874}
{"step": 36080, "loss": 3.6477119445800783, "grad_norm": 0.9992245435714722, "learning_rate": 0.0001, "elbo": 3.64771, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.115880835092658}
{"step": 36085, "loss": 3.652421569824219, "grad_norm": 0.6552689671516418, "learning_rate": 0.0001, "elbo": 3.65242, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1161740558292284}
{"step": 36090, "loss": 3.660125732421875, "grad_norm": 0.6611486077308655, "learning_rate": 0.0001, "elbo": 3.66013, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.116467276565799}
{"step": 36095, "loss": 3.6599960327148438, "grad_norm": 0.6579351425170898, "learning_rate": 0.0001, "elbo": 3.66, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1167604973023693}
{"step": 36100, "loss": 3.677245330810547, "grad_norm": 0.6414170861244202, "learning_rate": 0.0001, "elbo": 3.67725, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1170537180389397}
{"step": 36105, "loss": 3.6569778442382814, "grad_norm": 0.7006705403327942, "learning_rate": 0.0001, "elbo": 3.65698, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.11734693877551}
{"step": 36110, "loss": 3.6515838623046877, "grad_norm": 0.6664230823516846, "learning_rate": 0.0001, "elbo": 3.65158, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1176401595120806}
{"step": 36115, "loss": 3.660592269897461, "grad_norm": 0.6564701199531555, "learning_rate": 0.0001, "elbo": 3.66059, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.117933380248651}
{"step": 36120, "loss": 3.6330490112304688, "grad_norm": 0.754988431930542, "learning_rate": 0.0001, "elbo": 3.63305, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1182266009852215}
{"step": 36120, "loss": 3.6330490112304688, "grad_norm": 0.754988431930542, "learning_rate": 0.0001, "elbo": 3.63305, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1182266009852215}
{"step": 36125, "loss": 3.6331851959228514, "grad_norm": 0.637052595615387, "learning_rate": 0.0001, "elbo": 3.63319, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.118519821721792}
{"step": 36130, "loss": 3.6583175659179688, "grad_norm": 0.6786864995956421, "learning_rate": 0.0001, "elbo": 3.65832, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1188130424583624}
{"step": 36135, "loss": 3.6640663146972656, "grad_norm": 0.719424307346344, "learning_rate": 0.0001, "elbo": 3.66407, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1191062631949333}
{"step": 36140, "loss": 3.665264129638672, "grad_norm": 0.7497460246086121, "learning_rate": 0.0001, "elbo": 3.66526, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.119399483931504}
{"step": 36145, "loss": 3.6596771240234376, "grad_norm": 0.7115507125854492, "learning_rate": 0.0001, "elbo": 3.65968, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1196927046680742}
{"step": 36150, "loss": 3.662775421142578, "grad_norm": 0.8842142820358276, "learning_rate": 0.0001, "elbo": 3.66278, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1199859254046447}
{"step": 36155, "loss": 3.6589874267578124, "grad_norm": 0.6359195709228516, "learning_rate": 0.0001, "elbo": 3.65899, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.120279146141215}
{"step": 36160, "loss": 3.6576019287109376, "grad_norm": 0.641337513923645, "learning_rate": 0.0001, "elbo": 3.6576, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1205723668777856}
{"step": 36165, "loss": 3.659495544433594, "grad_norm": 0.6892362833023071, "learning_rate": 0.0001, "elbo": 3.6595, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.120865587614356}
{"step": 36170, "loss": 3.653211975097656, "grad_norm": 0.7635718584060669, "learning_rate": 0.0001, "elbo": 3.65321, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1211588083509265}
{"step": 36175, "loss": 3.645901107788086, "grad_norm": 0.6876832246780396, "learning_rate": 0.0001, "elbo": 3.6459, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.121452029087497}
{"step": 36180, "loss": 3.6390754699707033, "grad_norm": 0.653903067111969, "learning_rate": 0.0001, "elbo": 3.63908, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1217452498240674}
{"step": 36185, "loss": 3.6535903930664064, "grad_norm": 0.6760789155960083, "learning_rate": 0.0001, "elbo": 3.65359, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.122038470560638}
{"step": 36185, "loss": 3.6535903930664064, "grad_norm": 0.6760789155960083, "learning_rate": 0.0001, "elbo": 3.65359, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.122038470560638}
{"step": 36190, "loss": 3.656601333618164, "grad_norm": 0.6665823459625244, "learning_rate": 0.0001, "elbo": 3.6566, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1223316912972083}
{"step": 36195, "loss": 3.6584457397460937, "grad_norm": 0.7526004314422607, "learning_rate": 0.0001, "elbo": 3.65845, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1226249120337792}
{"step": 36200, "loss": 3.659297561645508, "grad_norm": 0.7042129635810852, "learning_rate": 0.0001, "elbo": 3.6593, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1229181327703497}
{"step": 36205, "loss": 3.6744384765625, "grad_norm": 1.0371174812316895, "learning_rate": 0.0001, "elbo": 3.67444, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.12321135350692}
{"step": 36210, "loss": 3.6553382873535156, "grad_norm": 0.6670002937316895, "learning_rate": 0.0001, "elbo": 3.65534, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1235045742434906}
{"step": 36215, "loss": 3.655619812011719, "grad_norm": 0.6741934418678284, "learning_rate": 0.0001, "elbo": 3.65562, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.123797794980061}
{"step": 36220, "loss": 3.6656803131103515, "grad_norm": 0.6971190571784973, "learning_rate": 0.0001, "elbo": 3.66568, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1240910157166315}
{"step": 36225, "loss": 3.65921630859375, "grad_norm": 0.6974456310272217, "learning_rate": 0.0001, "elbo": 3.65922, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.124384236453202}
{"step": 36230, "loss": 3.6691463470458983, "grad_norm": 0.6819683313369751, "learning_rate": 0.0001, "elbo": 3.66915, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1246774571897724}
{"step": 36235, "loss": 3.6528358459472656, "grad_norm": 0.6392311453819275, "learning_rate": 0.0001, "elbo": 3.65284, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.124970677926343}
{"step": 36240, "loss": 3.652749252319336, "grad_norm": 0.6608593463897705, "learning_rate": 0.0001, "elbo": 3.65275, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1252638986629133}
{"step": 36245, "loss": 3.645489501953125, "grad_norm": 0.6586004495620728, "learning_rate": 0.0001, "elbo": 3.64549, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1255571193994838}
{"step": 36250, "loss": 3.6432594299316405, "grad_norm": 0.8495762944221497, "learning_rate": 0.0001, "elbo": 3.64326, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1258503401360542}
{"step": 36255, "loss": 3.668354797363281, "grad_norm": 0.7632632255554199, "learning_rate": 0.0001, "elbo": 3.66836, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1261435608726247}
{"step": 36260, "loss": 3.6657150268554686, "grad_norm": 0.6788262724876404, "learning_rate": 0.0001, "elbo": 3.66572, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1264367816091956}
{"step": 36265, "loss": 3.6658729553222655, "grad_norm": 0.6679102778434753, "learning_rate": 0.0001, "elbo": 3.66587, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.126730002345766}
{"step": 36270, "loss": 3.6449920654296877, "grad_norm": 0.8990054726600647, "learning_rate": 0.0001, "elbo": 3.64499, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1270232230823365}
{"step": 36275, "loss": 3.6460376739501954, "grad_norm": 0.6662490367889404, "learning_rate": 0.0001, "elbo": 3.64604, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.127316443818907}
{"step": 36280, "loss": 3.6376380920410156, "grad_norm": 0.6658576726913452, "learning_rate": 0.0001, "elbo": 3.63764, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1276096645554774}
{"step": 36280, "loss": 3.6376380920410156, "grad_norm": 0.6658576726913452, "learning_rate": 0.0001, "elbo": 3.63764, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1276096645554774}
{"step": 36285, "loss": 3.6521724700927733, "grad_norm": 0.6684186458587646, "learning_rate": 0.0001, "elbo": 3.65217, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.127902885292048}
{"step": 36290, "loss": 3.6673946380615234, "grad_norm": 0.6687647700309753, "learning_rate": 0.0001, "elbo": 3.66739, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1281961060286183}
{"step": 36295, "loss": 3.6595687866210938, "grad_norm": 0.6391799449920654, "learning_rate": 0.0001, "elbo": 3.65957, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1284893267651888}
{"step": 36300, "loss": 3.6484813690185547, "grad_norm": 0.6580250263214111, "learning_rate": 0.0001, "elbo": 3.64848, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.128782547501759}
{"step": 36300, "loss": 3.6484813690185547, "grad_norm": 0.6580250263214111, "learning_rate": 0.0001, "elbo": 3.64848, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.128782547501759}
{"step": 36305, "loss": 3.661833953857422, "grad_norm": 0.7039161920547485, "learning_rate": 0.0001, "elbo": 3.66183, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1290757682383297}
{"step": 36310, "loss": 3.652571105957031, "grad_norm": 0.6777838468551636, "learning_rate": 0.0001, "elbo": 3.65257, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1293689889749}
{"step": 36315, "loss": 3.6561790466308595, "grad_norm": 0.6731910109519958, "learning_rate": 0.0001, "elbo": 3.65618, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.129662209711471}
{"step": 36320, "loss": 3.6413917541503906, "grad_norm": 0.6687458753585815, "learning_rate": 0.0001, "elbo": 3.64139, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1299554304480415}
{"step": 36325, "loss": 3.6555202484130858, "grad_norm": 0.6890708208084106, "learning_rate": 0.0001, "elbo": 3.65552, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.130248651184612}
{"step": 36325, "loss": 3.6555202484130858, "grad_norm": 0.6890708208084106, "learning_rate": 0.0001, "elbo": 3.65552, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.130248651184612}
{"step": 36330, "loss": 3.661695861816406, "grad_norm": 0.7526251077651978, "learning_rate": 0.0001, "elbo": 3.6617, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1305418719211824}
{"step": 36335, "loss": 3.658954620361328, "grad_norm": 0.7247405052185059, "learning_rate": 0.0001, "elbo": 3.65895, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.130835092657753}
{"step": 36340, "loss": 3.649580383300781, "grad_norm": 1.0025889873504639, "learning_rate": 0.0001, "elbo": 3.64958, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1311283133943233}
{"step": 36345, "loss": 3.6540878295898436, "grad_norm": 0.679740846157074, "learning_rate": 0.0001, "elbo": 3.65409, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1314215341308937}
{"step": 36350, "loss": 3.6641387939453125, "grad_norm": 0.6890644431114197, "learning_rate": 0.0001, "elbo": 3.66414, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.131714754867464}
{"step": 36355, "loss": 3.6605804443359373, "grad_norm": 0.7026991248130798, "learning_rate": 0.0001, "elbo": 3.66058, "mask_frac": 0.5013, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1320079756040347}
{"step": 36360, "loss": 3.6595420837402344, "grad_norm": 0.6862618327140808, "learning_rate": 0.0001, "elbo": 3.65954, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.132301196340605}
{"step": 36365, "loss": 3.651270294189453, "grad_norm": 0.7074316740036011, "learning_rate": 0.0001, "elbo": 3.65127, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1325944170771756}
{"step": 36370, "loss": 3.663192367553711, "grad_norm": 0.7018143534660339, "learning_rate": 0.0001, "elbo": 3.66319, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.132887637813746}
{"step": 36375, "loss": 3.669209671020508, "grad_norm": 0.721393883228302, "learning_rate": 0.0001, "elbo": 3.66921, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1331808585503165}
{"step": 36380, "loss": 3.651480865478516, "grad_norm": 1.003200650215149, "learning_rate": 0.0001, "elbo": 3.65148, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1334740792868874}
{"step": 36385, "loss": 3.6615928649902343, "grad_norm": 0.6976430416107178, "learning_rate": 0.0001, "elbo": 3.66159, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.133767300023458}
{"step": 36390, "loss": 3.642608642578125, "grad_norm": 0.7735399007797241, "learning_rate": 0.0001, "elbo": 3.64261, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1340605207600283}
{"step": 36395, "loss": 3.6664455413818358, "grad_norm": 0.676507830619812, "learning_rate": 0.0001, "elbo": 3.66645, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1343537414965987}
{"step": 36400, "loss": 3.652377700805664, "grad_norm": 0.7012555599212646, "learning_rate": 0.0001, "elbo": 3.65238, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.134646962233169}
{"step": 36405, "loss": 3.6576793670654295, "grad_norm": 0.6679376363754272, "learning_rate": 0.0001, "elbo": 3.65768, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1349401829697396}
{"step": 36410, "loss": 3.64599609375, "grad_norm": 0.6706766486167908, "learning_rate": 0.0001, "elbo": 3.646, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.13523340370631}
{"step": 36415, "loss": 3.6517452239990233, "grad_norm": 0.8328291773796082, "learning_rate": 0.0001, "elbo": 3.65175, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1355266244428805}
{"step": 36420, "loss": 3.646554946899414, "grad_norm": 0.709977924823761, "learning_rate": 0.0001, "elbo": 3.64655, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.135819845179451}
{"step": 36425, "loss": 3.6591865539550783, "grad_norm": 0.7050387859344482, "learning_rate": 0.0001, "elbo": 3.65919, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1361130659160215}
{"step": 36430, "loss": 3.6695571899414063, "grad_norm": 0.6694044470787048, "learning_rate": 0.0001, "elbo": 3.66956, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.136406286652592}
{"step": 36435, "loss": 3.6430458068847655, "grad_norm": 0.8819066882133484, "learning_rate": 0.0001, "elbo": 3.64305, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1366995073891624}
{"step": 36440, "loss": 3.654159927368164, "grad_norm": 0.6799148917198181, "learning_rate": 0.0001, "elbo": 3.65416, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1369927281257333}
{"step": 36445, "loss": 3.6481697082519533, "grad_norm": 0.6631718277931213, "learning_rate": 0.0001, "elbo": 3.64817, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1372859488623037}
{"step": 36450, "loss": 3.652466583251953, "grad_norm": 0.6464780569076538, "learning_rate": 0.0001, "elbo": 3.65247, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.137579169598874}
{"step": 36455, "loss": 3.6462135314941406, "grad_norm": 0.6852753758430481, "learning_rate": 0.0001, "elbo": 3.64621, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1378723903354446}
{"step": 36460, "loss": 3.6596885681152345, "grad_norm": 0.6745194792747498, "learning_rate": 0.0001, "elbo": 3.65969, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.138165611072015}
{"step": 36465, "loss": 3.6500526428222657, "grad_norm": 0.6825609803199768, "learning_rate": 0.0001, "elbo": 3.65005, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1384588318085855}
{"step": 36470, "loss": 3.643454742431641, "grad_norm": 0.6664520502090454, "learning_rate": 0.0001, "elbo": 3.64345, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.138752052545156}
{"step": 36475, "loss": 3.6473052978515623, "grad_norm": 0.8683560490608215, "learning_rate": 0.0001, "elbo": 3.64731, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1390452732817264}
{"step": 36480, "loss": 3.64464111328125, "grad_norm": 0.6848044395446777, "learning_rate": 0.0001, "elbo": 3.64464, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.139338494018297}
{"step": 36485, "loss": 3.65675048828125, "grad_norm": 0.653268039226532, "learning_rate": 0.0001, "elbo": 3.65675, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1396317147548674}
{"step": 36490, "loss": 3.6691917419433593, "grad_norm": 0.7057844400405884, "learning_rate": 0.0001, "elbo": 3.66919, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.139924935491438}
{"step": 36495, "loss": 3.6455947875976564, "grad_norm": 0.6812968254089355, "learning_rate": 0.0001, "elbo": 3.64559, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1402181562280083}
{"step": 36500, "loss": 3.6491355895996094, "grad_norm": 0.6856244206428528, "learning_rate": 0.0001, "elbo": 3.64914, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1405113769645787}
{"step": 36505, "loss": 3.642946243286133, "grad_norm": 0.7412734031677246, "learning_rate": 0.0001, "elbo": 3.64295, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1408045977011496}
{"step": 36510, "loss": 3.6549331665039064, "grad_norm": 0.7544237375259399, "learning_rate": 0.0001, "elbo": 3.65493, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.14109781843772}
{"step": 36515, "loss": 3.6458900451660154, "grad_norm": 0.6628013253211975, "learning_rate": 0.0001, "elbo": 3.64589, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1413910391742905}
{"step": 36520, "loss": 3.6524078369140627, "grad_norm": 0.6673009991645813, "learning_rate": 0.0001, "elbo": 3.65241, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.141684259910861}
{"step": 36525, "loss": 3.662828063964844, "grad_norm": 0.7455532550811768, "learning_rate": 0.0001, "elbo": 3.66283, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1419774806474314}
{"step": 36530, "loss": 3.6487380981445314, "grad_norm": 0.791221022605896, "learning_rate": 0.0001, "elbo": 3.64874, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.142270701384002}
{"step": 36535, "loss": 3.635769271850586, "grad_norm": 0.6577586531639099, "learning_rate": 0.0001, "elbo": 3.63577, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1425639221205723}
{"step": 36540, "loss": 3.658014678955078, "grad_norm": 0.6773925423622131, "learning_rate": 0.0001, "elbo": 3.65801, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.142857142857143}
{"step": 36545, "loss": 3.6473918914794923, "grad_norm": 0.7903779745101929, "learning_rate": 0.0001, "elbo": 3.64739, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1431503635937132}
{"step": 36550, "loss": 3.642484664916992, "grad_norm": 0.6580793857574463, "learning_rate": 0.0001, "elbo": 3.64248, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1434435843302837}
{"step": 36555, "loss": 3.6471527099609373, "grad_norm": 0.6596413850784302, "learning_rate": 0.0001, "elbo": 3.64715, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.143736805066854}
{"step": 36560, "loss": 3.642913055419922, "grad_norm": 0.6806385517120361, "learning_rate": 0.0001, "elbo": 3.64291, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1440300258034246}
{"step": 36565, "loss": 3.6622764587402346, "grad_norm": 0.6697333455085754, "learning_rate": 0.0001, "elbo": 3.66228, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1443232465399955}
{"step": 36570, "loss": 3.6632034301757814, "grad_norm": 0.7433515787124634, "learning_rate": 0.0001, "elbo": 3.6632, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.144616467276566}
{"step": 36575, "loss": 3.6689586639404297, "grad_norm": 0.6951451301574707, "learning_rate": 0.0001, "elbo": 3.66896, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1449096880131364}
{"step": 36580, "loss": 3.64166145324707, "grad_norm": 0.6769804358482361, "learning_rate": 0.0001, "elbo": 3.64166, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.145202908749707}
{"step": 36585, "loss": 3.649652862548828, "grad_norm": 0.6913779377937317, "learning_rate": 0.0001, "elbo": 3.64965, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1454961294862773}
{"step": 36590, "loss": 3.648305892944336, "grad_norm": 0.7611830234527588, "learning_rate": 0.0001, "elbo": 3.64831, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.145789350222848}
{"step": 36595, "loss": 3.6636032104492187, "grad_norm": 0.6562556624412537, "learning_rate": 0.0001, "elbo": 3.6636, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1460825709594182}
{"step": 36600, "loss": 3.654244613647461, "grad_norm": 0.7846698760986328, "learning_rate": 0.0001, "elbo": 3.65424, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1463757916959887}
{"step": 36605, "loss": 3.652405548095703, "grad_norm": 0.8362709283828735, "learning_rate": 0.0001, "elbo": 3.65241, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.146669012432559}
{"step": 36610, "loss": 3.6375705718994142, "grad_norm": 0.7037200331687927, "learning_rate": 0.0001, "elbo": 3.63757, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1469622331691296}
{"step": 36610, "loss": 3.6375705718994142, "grad_norm": 0.7037200331687927, "learning_rate": 0.0001, "elbo": 3.63757, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1469622331691296}
{"step": 36615, "loss": 3.658523941040039, "grad_norm": 0.6904363036155701, "learning_rate": 0.0001, "elbo": 3.65852, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1472554539057}
{"step": 36620, "loss": 3.649169158935547, "grad_norm": 0.7013913989067078, "learning_rate": 0.0001, "elbo": 3.64917, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1475486746422705}
{"step": 36625, "loss": 3.6634193420410157, "grad_norm": 0.8934878706932068, "learning_rate": 0.0001, "elbo": 3.66342, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.147841895378841}
{"step": 36630, "loss": 3.6667236328125, "grad_norm": 0.7418912053108215, "learning_rate": 0.0001, "elbo": 3.66672, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.148135116115412}
{"step": 36635, "loss": 3.656752014160156, "grad_norm": 0.7613415718078613, "learning_rate": 0.0001, "elbo": 3.65675, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1484283368519823}
{"step": 36640, "loss": 3.6453060150146483, "grad_norm": 0.7382286787033081, "learning_rate": 0.0001, "elbo": 3.64531, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1487215575885528}
{"step": 36645, "loss": 3.663063812255859, "grad_norm": 0.7426290512084961, "learning_rate": 0.0001, "elbo": 3.66306, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.149014778325123}
{"step": 36650, "loss": 3.658056640625, "grad_norm": 0.6214014291763306, "learning_rate": 0.0001, "elbo": 3.65806, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1493079990616937}
{"step": 36655, "loss": 3.660517120361328, "grad_norm": 0.7421731352806091, "learning_rate": 0.0001, "elbo": 3.66052, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.149601219798264}
{"step": 36660, "loss": 3.662500762939453, "grad_norm": 0.6598551869392395, "learning_rate": 0.0001, "elbo": 3.6625, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1498944405348346}
{"step": 36665, "loss": 3.657456970214844, "grad_norm": 0.9028761386871338, "learning_rate": 0.0001, "elbo": 3.65746, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.150187661271405}
{"step": 36670, "loss": 3.6678092956542967, "grad_norm": 0.9199414253234863, "learning_rate": 0.0001, "elbo": 3.66781, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1504808820079755}
{"step": 36675, "loss": 3.670641326904297, "grad_norm": 0.6696994304656982, "learning_rate": 0.0001, "elbo": 3.67064, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.150774102744546}
{"step": 36680, "loss": 3.6479476928710937, "grad_norm": 0.6697005033493042, "learning_rate": 0.0001, "elbo": 3.64795, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1510673234811164}
{"step": 36685, "loss": 3.6669944763183593, "grad_norm": 0.6933204531669617, "learning_rate": 0.0001, "elbo": 3.66699, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1513605442176873}
{"step": 36690, "loss": 3.6588043212890624, "grad_norm": 0.6964490413665771, "learning_rate": 0.0001, "elbo": 3.6588, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1516537649542578}
{"step": 36695, "loss": 3.661241912841797, "grad_norm": 0.6418117880821228, "learning_rate": 0.0001, "elbo": 3.66124, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.151946985690828}
{"step": 36700, "loss": 3.6602020263671875, "grad_norm": 0.7177948355674744, "learning_rate": 0.0001, "elbo": 3.6602, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1522402064273987}
{"step": 36705, "loss": 3.646976089477539, "grad_norm": 0.6918685436248779, "learning_rate": 0.0001, "elbo": 3.64698, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.152533427163969}
{"step": 36710, "loss": 3.6446464538574217, "grad_norm": 0.7567028403282166, "learning_rate": 0.0001, "elbo": 3.64465, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1528266479005396}
{"step": 36715, "loss": 3.646195983886719, "grad_norm": 0.7940809726715088, "learning_rate": 0.0001, "elbo": 3.6462, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.15311986863711}
{"step": 36720, "loss": 3.659906005859375, "grad_norm": 0.6753031015396118, "learning_rate": 0.0001, "elbo": 3.65991, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1534130893736805}
{"step": 36725, "loss": 3.6510181427001953, "grad_norm": 0.6764538288116455, "learning_rate": 0.0001, "elbo": 3.65102, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.153706310110251}
{"step": 36730, "loss": 3.646027374267578, "grad_norm": 0.6750544905662537, "learning_rate": 0.0001, "elbo": 3.64603, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1539995308468214}
{"step": 36735, "loss": 3.6514007568359377, "grad_norm": 0.7732831835746765, "learning_rate": 0.0001, "elbo": 3.6514, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.154292751583392}
{"step": 36740, "loss": 3.646175765991211, "grad_norm": 0.7728806138038635, "learning_rate": 0.0001, "elbo": 3.64618, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1545859723199623}
{"step": 36745, "loss": 3.646565246582031, "grad_norm": 0.6902053952217102, "learning_rate": 0.0001, "elbo": 3.64657, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1548791930565327}
{"step": 36750, "loss": 3.645172119140625, "grad_norm": 1.0208736658096313, "learning_rate": 0.0001, "elbo": 3.64517, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1551724137931036}
{"step": 36755, "loss": 3.660413360595703, "grad_norm": 0.6635101437568665, "learning_rate": 0.0001, "elbo": 3.66041, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.155465634529674}
{"step": 36760, "loss": 3.647906494140625, "grad_norm": 0.6801097393035889, "learning_rate": 0.0001, "elbo": 3.64791, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1557588552662446}
{"step": 36765, "loss": 3.6718914031982424, "grad_norm": 0.665195643901825, "learning_rate": 0.0001, "elbo": 3.67189, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.156052076002815}
{"step": 36770, "loss": 3.649457550048828, "grad_norm": 0.6641085147857666, "learning_rate": 0.0001, "elbo": 3.64946, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1563452967393855}
{"step": 36775, "loss": 3.6419418334960936, "grad_norm": 0.67481929063797, "learning_rate": 0.0001, "elbo": 3.64194, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.156638517475956}
{"step": 36780, "loss": 3.664411926269531, "grad_norm": 0.6670027375221252, "learning_rate": 0.0001, "elbo": 3.66441, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1569317382125264}
{"step": 36785, "loss": 3.653462600708008, "grad_norm": 1.120234727859497, "learning_rate": 0.0001, "elbo": 3.65346, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.157224958949097}
{"step": 36790, "loss": 3.657643508911133, "grad_norm": 0.6633575558662415, "learning_rate": 0.0001, "elbo": 3.65764, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1575181796856673}
{"step": 36795, "loss": 3.645835113525391, "grad_norm": 0.6880178451538086, "learning_rate": 0.0001, "elbo": 3.64583, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1578114004222377}
{"step": 36800, "loss": 3.64929084777832, "grad_norm": 0.7928032279014587, "learning_rate": 0.0001, "elbo": 3.64929, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.158104621158808}
{"step": 36805, "loss": 3.662694549560547, "grad_norm": 0.7314031720161438, "learning_rate": 0.0001, "elbo": 3.66269, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1583978418953786}
{"step": 36810, "loss": 3.6325721740722656, "grad_norm": 0.7536969184875488, "learning_rate": 0.0001, "elbo": 3.63257, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1586910626319495}
{"step": 36810, "loss": 3.6325721740722656, "grad_norm": 0.7536969184875488, "learning_rate": 0.0001, "elbo": 3.63257, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1586910626319495}
{"step": 36815, "loss": 3.648650360107422, "grad_norm": 0.679889440536499, "learning_rate": 0.0001, "elbo": 3.64865, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.15898428336852}
{"step": 36820, "loss": 3.656688690185547, "grad_norm": 0.6539593935012817, "learning_rate": 0.0001, "elbo": 3.65669, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1592775041050905}
{"step": 36825, "loss": 3.6600582122802736, "grad_norm": 0.7607574462890625, "learning_rate": 0.0001, "elbo": 3.66006, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.159570724841661}
{"step": 36830, "loss": 3.6501079559326173, "grad_norm": 0.7112681865692139, "learning_rate": 0.0001, "elbo": 3.65011, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1598639455782314}
{"step": 36835, "loss": 3.66839485168457, "grad_norm": 0.7601578831672668, "learning_rate": 0.0001, "elbo": 3.66839, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.160157166314802}
{"step": 36840, "loss": 3.644735336303711, "grad_norm": 0.7391015887260437, "learning_rate": 0.0001, "elbo": 3.64474, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1604503870513723}
{"step": 36845, "loss": 3.653124237060547, "grad_norm": 0.7303416728973389, "learning_rate": 0.0001, "elbo": 3.65312, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1607436077879427}
{"step": 36850, "loss": 3.646088409423828, "grad_norm": 0.6402844190597534, "learning_rate": 0.0001, "elbo": 3.64609, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.161036828524513}
{"step": 36855, "loss": 3.64449462890625, "grad_norm": 0.7221013307571411, "learning_rate": 0.0001, "elbo": 3.64449, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1613300492610836}
{"step": 36860, "loss": 3.6395938873291014, "grad_norm": 0.7045238018035889, "learning_rate": 0.0001, "elbo": 3.63959, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.161623269997654}
{"step": 36865, "loss": 3.6343925476074217, "grad_norm": 0.6830393671989441, "learning_rate": 0.0001, "elbo": 3.63439, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1619164907342245}
{"step": 36870, "loss": 3.644117736816406, "grad_norm": 0.7139775156974792, "learning_rate": 0.0001, "elbo": 3.64412, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.162209711470795}
{"step": 36875, "loss": 3.65101318359375, "grad_norm": 0.6812112927436829, "learning_rate": 0.0001, "elbo": 3.65101, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.162502932207366}
{"step": 36880, "loss": 3.648090362548828, "grad_norm": 0.6721770763397217, "learning_rate": 0.0001, "elbo": 3.64809, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1627961529439363}
{"step": 36885, "loss": 3.6433597564697267, "grad_norm": 0.6328352093696594, "learning_rate": 0.0001, "elbo": 3.64336, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.163089373680507}
{"step": 36890, "loss": 3.6417747497558595, "grad_norm": 0.672149658203125, "learning_rate": 0.0001, "elbo": 3.64177, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1633825944170773}
{"step": 36895, "loss": 3.6560821533203125, "grad_norm": 1.0039018392562866, "learning_rate": 0.0001, "elbo": 3.65608, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1636758151536477}
{"step": 36900, "loss": 3.6497161865234373, "grad_norm": 0.6659524440765381, "learning_rate": 0.0001, "elbo": 3.64972, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.163969035890218}
{"step": 36905, "loss": 3.6455062866210937, "grad_norm": 0.669147253036499, "learning_rate": 0.0001, "elbo": 3.64551, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1642622566267886}
{"step": 36905, "loss": 3.6455062866210937, "grad_norm": 0.669147253036499, "learning_rate": 0.0001, "elbo": 3.64551, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1642622566267886}
{"step": 36910, "loss": 3.636421966552734, "grad_norm": 0.7427883744239807, "learning_rate": 0.0001, "elbo": 3.63642, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.164555477363359}
{"step": 36915, "loss": 3.644911193847656, "grad_norm": 0.6525830626487732, "learning_rate": 0.0001, "elbo": 3.64491, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1648486980999295}
{"step": 36920, "loss": 3.638558578491211, "grad_norm": 0.6728376746177673, "learning_rate": 0.0001, "elbo": 3.63856, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1651419188365}
{"step": 36925, "loss": 3.6583541870117187, "grad_norm": 0.6741546392440796, "learning_rate": 0.0001, "elbo": 3.65835, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1654351395730704}
{"step": 36930, "loss": 3.6548507690429686, "grad_norm": 0.6530831456184387, "learning_rate": 0.0001, "elbo": 3.65485, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1657283603096413}
{"step": 36935, "loss": 3.6633495330810546, "grad_norm": 0.6824098825454712, "learning_rate": 0.0001, "elbo": 3.66335, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.166021581046212}
{"step": 36940, "loss": 3.6454841613769533, "grad_norm": 0.7030236124992371, "learning_rate": 0.0001, "elbo": 3.64548, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1663148017827822}
{"step": 36945, "loss": 3.667331314086914, "grad_norm": 0.6804777383804321, "learning_rate": 0.0001, "elbo": 3.66733, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1666080225193527}
{"step": 36950, "loss": 3.6557586669921873, "grad_norm": 0.6714417338371277, "learning_rate": 0.0001, "elbo": 3.65576, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.166901243255923}
{"step": 36955, "loss": 3.6413379669189454, "grad_norm": 0.6447746753692627, "learning_rate": 0.0001, "elbo": 3.64134, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1671944639924936}
{"step": 36960, "loss": 3.65574951171875, "grad_norm": 0.7110750675201416, "learning_rate": 0.0001, "elbo": 3.65575, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.167487684729064}
{"step": 36965, "loss": 3.650926208496094, "grad_norm": 0.6762901544570923, "learning_rate": 0.0001, "elbo": 3.65093, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1677809054656345}
{"step": 36970, "loss": 3.6467826843261717, "grad_norm": 0.6958580613136292, "learning_rate": 0.0001, "elbo": 3.64678, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.168074126202205}
{"step": 36975, "loss": 3.6397426605224608, "grad_norm": 0.6569961309432983, "learning_rate": 0.0001, "elbo": 3.63974, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1683673469387754}
{"step": 36980, "loss": 3.6526092529296874, "grad_norm": 0.6839905381202698, "learning_rate": 0.0001, "elbo": 3.65261, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.168660567675346}
{"step": 36985, "loss": 3.6238128662109377, "grad_norm": 0.657081663608551, "learning_rate": 0.0001, "elbo": 3.62381, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1689537884119163}
{"step": 36990, "loss": 3.6488811492919924, "grad_norm": 0.737963080406189, "learning_rate": 0.0001, "elbo": 3.64888, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.169247009148487}
{"step": 36995, "loss": 3.64307861328125, "grad_norm": 0.6951786875724792, "learning_rate": 0.0001, "elbo": 3.64308, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1695402298850572}
{"step": 37000, "loss": 3.6574462890625, "grad_norm": 0.7333166599273682, "learning_rate": 0.0001, "elbo": 3.65745, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.169833450621628}
{"step": 37005, "loss": 3.6565799713134766, "grad_norm": 0.7169680595397949, "learning_rate": 0.0001, "elbo": 3.65658, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1701266713581986}
{"step": 37010, "loss": 3.654523468017578, "grad_norm": 0.6988860368728638, "learning_rate": 0.0001, "elbo": 3.65452, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.170419892094769}
{"step": 37015, "loss": 3.6538246154785154, "grad_norm": 0.7093992829322815, "learning_rate": 0.0001, "elbo": 3.65382, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1707131128313395}
{"step": 37020, "loss": 3.6650436401367186, "grad_norm": 0.7721234560012817, "learning_rate": 0.0001, "elbo": 3.66504, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.17100633356791}
{"step": 37025, "loss": 3.650836181640625, "grad_norm": 0.6918313503265381, "learning_rate": 0.0001, "elbo": 3.65084, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1712995543044804}
{"step": 37030, "loss": 3.6637218475341795, "grad_norm": 0.8930584192276001, "learning_rate": 0.0001, "elbo": 3.66372, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.171592775041051}
{"step": 37035, "loss": 3.6502716064453127, "grad_norm": 0.8029144406318665, "learning_rate": 0.0001, "elbo": 3.65027, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1718859957776213}
{"step": 37040, "loss": 3.647044372558594, "grad_norm": 0.6367961168289185, "learning_rate": 0.0001, "elbo": 3.64704, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1721792165141918}
{"step": 37045, "loss": 3.6506729125976562, "grad_norm": 0.6385698318481445, "learning_rate": 0.0001, "elbo": 3.65067, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1724724372507622}
{"step": 37050, "loss": 3.644750213623047, "grad_norm": 0.6799655556678772, "learning_rate": 0.0001, "elbo": 3.64475, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1727656579873327}
{"step": 37055, "loss": 3.645224761962891, "grad_norm": 0.6808303594589233, "learning_rate": 0.0001, "elbo": 3.64522, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1730588787239036}
{"step": 37060, "loss": 3.6419822692871096, "grad_norm": 0.6674588322639465, "learning_rate": 0.0001, "elbo": 3.64198, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.173352099460474}
{"step": 37065, "loss": 3.6373062133789062, "grad_norm": 0.6852421760559082, "learning_rate": 0.0001, "elbo": 3.63731, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1736453201970445}
{"step": 37070, "loss": 3.6628044128417967, "grad_norm": 0.6865336894989014, "learning_rate": 0.0001, "elbo": 3.6628, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.173938540933615}
{"step": 37070, "loss": 3.6628044128417967, "grad_norm": 0.6865336894989014, "learning_rate": 0.0001, "elbo": 3.6628, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.173938540933615}
{"step": 37075, "loss": 3.6635913848876953, "grad_norm": 0.6588326692581177, "learning_rate": 0.0001, "elbo": 3.66359, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1742317616701854}
{"step": 37080, "loss": 3.657359313964844, "grad_norm": 0.6832467317581177, "learning_rate": 0.0001, "elbo": 3.65736, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.174524982406756}
{"step": 37085, "loss": 3.6394622802734373, "grad_norm": 0.7188810110092163, "learning_rate": 0.0001, "elbo": 3.63946, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1748182031433263}
{"step": 37090, "loss": 3.6374568939208984, "grad_norm": 0.6220534443855286, "learning_rate": 0.0001, "elbo": 3.63746, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1751114238798968}
{"step": 37095, "loss": 3.6599613189697267, "grad_norm": 0.6601409912109375, "learning_rate": 0.0001, "elbo": 3.65996, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.175404644616467}
{"step": 37100, "loss": 3.6496200561523438, "grad_norm": 0.6662940979003906, "learning_rate": 0.0001, "elbo": 3.64962, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1756978653530377}
{"step": 37105, "loss": 3.6537250518798827, "grad_norm": 0.6873441338539124, "learning_rate": 0.0001, "elbo": 3.65372, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.175991086089608}
{"step": 37110, "loss": 3.657927703857422, "grad_norm": 0.6885823011398315, "learning_rate": 0.0001, "elbo": 3.65793, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1762843068261786}
{"step": 37115, "loss": 3.642394256591797, "grad_norm": 0.7414289712905884, "learning_rate": 0.0001, "elbo": 3.64239, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.176577527562749}
{"step": 37120, "loss": 3.649378204345703, "grad_norm": 0.6993603110313416, "learning_rate": 0.0001, "elbo": 3.64938, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.17687074829932}
{"step": 37125, "loss": 3.6563480377197264, "grad_norm": 0.679429829120636, "learning_rate": 0.0001, "elbo": 3.65635, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1771639690358904}
{"step": 37130, "loss": 3.6479721069335938, "grad_norm": 0.6969032287597656, "learning_rate": 0.0001, "elbo": 3.64797, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.177457189772461}
{"step": 37135, "loss": 3.6614585876464845, "grad_norm": 1.0767433643341064, "learning_rate": 0.0001, "elbo": 3.66146, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1777504105090313}
{"step": 37140, "loss": 3.6584434509277344, "grad_norm": 0.7384013533592224, "learning_rate": 0.0001, "elbo": 3.65844, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1780436312456017}
{"step": 37145, "loss": 3.655223846435547, "grad_norm": 0.7072152495384216, "learning_rate": 0.0001, "elbo": 3.65522, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.178336851982172}
{"step": 37150, "loss": 3.6571414947509764, "grad_norm": 0.6963638067245483, "learning_rate": 0.0001, "elbo": 3.65714, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1786300727187426}
{"step": 37155, "loss": 3.6492950439453127, "grad_norm": 0.6840300559997559, "learning_rate": 0.0001, "elbo": 3.6493, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.178923293455313}
{"step": 37160, "loss": 3.6472709655761717, "grad_norm": 0.6548690795898438, "learning_rate": 0.0001, "elbo": 3.64727, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1792165141918836}
{"step": 37165, "loss": 3.6485214233398438, "grad_norm": 0.6564031839370728, "learning_rate": 0.0001, "elbo": 3.64852, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.179509734928454}
{"step": 37170, "loss": 3.6438419342041017, "grad_norm": 0.8043068051338196, "learning_rate": 0.0001, "elbo": 3.64384, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1798029556650245}
{"step": 37175, "loss": 3.6457443237304688, "grad_norm": 0.6635717749595642, "learning_rate": 0.0001, "elbo": 3.64574, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.180096176401595}
{"step": 37180, "loss": 3.6544673919677733, "grad_norm": 0.7992638945579529, "learning_rate": 0.0001, "elbo": 3.65447, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.180389397138166}
{"step": 37185, "loss": 3.6521934509277343, "grad_norm": 0.6758524179458618, "learning_rate": 0.0001, "elbo": 3.65219, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1806826178747363}
{"step": 37190, "loss": 3.6386703491210937, "grad_norm": 0.8677120208740234, "learning_rate": 0.0001, "elbo": 3.63867, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1809758386113067}
{"step": 37195, "loss": 3.653466796875, "grad_norm": 0.6780650019645691, "learning_rate": 0.0001, "elbo": 3.65347, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.181269059347877}
{"step": 37200, "loss": 3.6308135986328125, "grad_norm": 0.80738365650177, "learning_rate": 0.0001, "elbo": 3.63081, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1815622800844476}
{"step": 37205, "loss": 3.643931579589844, "grad_norm": 0.6819169521331787, "learning_rate": 0.0001, "elbo": 3.64393, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.181855500821018}
{"step": 37210, "loss": 3.6476181030273436, "grad_norm": 0.6484878659248352, "learning_rate": 0.0001, "elbo": 3.64762, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1821487215575885}
{"step": 37215, "loss": 3.6378746032714844, "grad_norm": 0.7456185221672058, "learning_rate": 0.0001, "elbo": 3.63787, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.182441942294159}
{"step": 37220, "loss": 3.634354019165039, "grad_norm": 0.6950825452804565, "learning_rate": 0.0001, "elbo": 3.63435, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1827351630307295}
{"step": 37225, "loss": 3.6496753692626953, "grad_norm": 0.7115485668182373, "learning_rate": 0.0001, "elbo": 3.64968, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1830283837673}
{"step": 37230, "loss": 3.6651870727539064, "grad_norm": 0.7166863679885864, "learning_rate": 0.0001, "elbo": 3.66519, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1833216045038704}
{"step": 37235, "loss": 3.6563694000244142, "grad_norm": 0.7033436894416809, "learning_rate": 0.0001, "elbo": 3.65637, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.183614825240441}
{"step": 37240, "loss": 3.6643848419189453, "grad_norm": 0.7122830152511597, "learning_rate": 0.0001, "elbo": 3.66438, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1839080459770113}
{"step": 37245, "loss": 3.6455657958984373, "grad_norm": 0.77200847864151, "learning_rate": 0.0001, "elbo": 3.64557, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.184201266713582}
{"step": 37250, "loss": 3.6575763702392576, "grad_norm": 0.7196489572525024, "learning_rate": 0.0001, "elbo": 3.65758, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1844944874501526}
{"step": 37255, "loss": 3.6542110443115234, "grad_norm": 0.6754660606384277, "learning_rate": 0.0001, "elbo": 3.65421, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.184787708186723}
{"step": 37255, "loss": 3.6542110443115234, "grad_norm": 0.6754660606384277, "learning_rate": 0.0001, "elbo": 3.65421, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.184787708186723}
{"step": 37260, "loss": 3.6585418701171877, "grad_norm": 0.6951035857200623, "learning_rate": 0.0001, "elbo": 3.65854, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1850809289232935}
{"step": 37265, "loss": 3.6521209716796874, "grad_norm": 0.6816287040710449, "learning_rate": 0.0001, "elbo": 3.65212, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.185374149659864}
{"step": 37270, "loss": 3.637279510498047, "grad_norm": 0.7321051359176636, "learning_rate": 0.0001, "elbo": 3.63728, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1856673703964344}
{"step": 37275, "loss": 3.655710220336914, "grad_norm": 0.6431258916854858, "learning_rate": 0.0001, "elbo": 3.65571, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.185960591133005}
{"step": 37280, "loss": 3.6515117645263673, "grad_norm": 0.8534963726997375, "learning_rate": 0.0001, "elbo": 3.65151, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1862538118695753}
{"step": 37285, "loss": 3.6493450164794923, "grad_norm": 0.6349447965621948, "learning_rate": 0.0001, "elbo": 3.64934, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.186547032606146}
{"step": 37290, "loss": 3.6573196411132813, "grad_norm": 0.7358932495117188, "learning_rate": 0.0001, "elbo": 3.65732, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1868402533427163}
{"step": 37295, "loss": 3.652286911010742, "grad_norm": 0.6677968502044678, "learning_rate": 0.0001, "elbo": 3.65229, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1871334740792867}
{"step": 37300, "loss": 3.6438522338867188, "grad_norm": 0.6979267001152039, "learning_rate": 0.0001, "elbo": 3.64385, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1874266948158576}
{"step": 37305, "loss": 3.644072723388672, "grad_norm": 0.6847851872444153, "learning_rate": 0.0001, "elbo": 3.64407, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.187719915552428}
{"step": 37310, "loss": 3.6313079833984374, "grad_norm": 0.700270414352417, "learning_rate": 0.0001, "elbo": 3.63131, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1880131362889985}
{"step": 37315, "loss": 3.6553001403808594, "grad_norm": 0.6956773996353149, "learning_rate": 0.0001, "elbo": 3.6553, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.188306357025569}
{"step": 37320, "loss": 3.664999771118164, "grad_norm": 0.7559155225753784, "learning_rate": 0.0001, "elbo": 3.665, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1885995777621394}
{"step": 37325, "loss": 3.662739562988281, "grad_norm": 0.8141834735870361, "learning_rate": 0.0001, "elbo": 3.66274, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.18889279849871}
{"step": 37330, "loss": 3.6497352600097654, "grad_norm": 0.6969340443611145, "learning_rate": 0.0001, "elbo": 3.64974, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1891860192352803}
{"step": 37335, "loss": 3.638877105712891, "grad_norm": 0.6742483377456665, "learning_rate": 0.0001, "elbo": 3.63888, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.189479239971851}
{"step": 37340, "loss": 3.6557083129882812, "grad_norm": 0.717944324016571, "learning_rate": 0.0001, "elbo": 3.65571, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1897724607084212}
{"step": 37345, "loss": 3.6397430419921877, "grad_norm": 0.7125677466392517, "learning_rate": 0.0001, "elbo": 3.63974, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1900656814449917}
{"step": 37350, "loss": 3.653858184814453, "grad_norm": 0.7152726054191589, "learning_rate": 0.0001, "elbo": 3.65386, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.190358902181562}
{"step": 37355, "loss": 3.646886444091797, "grad_norm": 0.7071767449378967, "learning_rate": 0.0001, "elbo": 3.64689, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1906521229181326}
{"step": 37355, "loss": 3.646886444091797, "grad_norm": 0.7071767449378967, "learning_rate": 0.0001, "elbo": 3.64689, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1906521229181326}
{"step": 37360, "loss": 3.6557285308837892, "grad_norm": 0.6775672435760498, "learning_rate": 0.0001, "elbo": 3.65573, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.190945343654703}
{"step": 37365, "loss": 3.6628517150878905, "grad_norm": 1.2369937896728516, "learning_rate": 0.0001, "elbo": 3.66285, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.191238564391274}
{"step": 37370, "loss": 3.64891357421875, "grad_norm": 1.0910462141036987, "learning_rate": 0.0001, "elbo": 3.64891, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1915317851278444}
{"step": 37375, "loss": 3.6485622406005858, "grad_norm": 0.7065521478652954, "learning_rate": 0.0001, "elbo": 3.64856, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.191825005864415}
{"step": 37380, "loss": 3.6332225799560547, "grad_norm": 0.7408915162086487, "learning_rate": 0.0001, "elbo": 3.63322, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1921182266009853}
{"step": 37385, "loss": 3.6478836059570314, "grad_norm": 0.7008287906646729, "learning_rate": 0.0001, "elbo": 3.64788, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1924114473375558}
{"step": 37390, "loss": 3.656678009033203, "grad_norm": 0.7296186089515686, "learning_rate": 0.0001, "elbo": 3.65668, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1927046680741262}
{"step": 37395, "loss": 3.643049621582031, "grad_norm": 0.8226335644721985, "learning_rate": 0.0001, "elbo": 3.64305, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1929978888106967}
{"step": 37400, "loss": 3.664192962646484, "grad_norm": 0.6776247620582581, "learning_rate": 0.0001, "elbo": 3.66419, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.193291109547267}
{"step": 37405, "loss": 3.633684539794922, "grad_norm": 0.6827128529548645, "learning_rate": 0.0001, "elbo": 3.63368, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1935843302838376}
{"step": 37410, "loss": 3.6495944976806642, "grad_norm": 0.6658087968826294, "learning_rate": 0.0001, "elbo": 3.64959, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.193877551020408}
{"step": 37415, "loss": 3.66175537109375, "grad_norm": 0.68763267993927, "learning_rate": 0.0001, "elbo": 3.66176, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1941707717569785}
{"step": 37420, "loss": 3.6472248077392577, "grad_norm": 1.0735583305358887, "learning_rate": 0.0001, "elbo": 3.64722, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.194463992493549}
{"step": 37425, "loss": 3.653764343261719, "grad_norm": 0.6992087364196777, "learning_rate": 0.0001, "elbo": 3.65376, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.19475721323012}
{"step": 37430, "loss": 3.66339225769043, "grad_norm": 0.6844651103019714, "learning_rate": 0.0001, "elbo": 3.66339, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1950504339666903}
{"step": 37435, "loss": 3.6548065185546874, "grad_norm": 0.6521947979927063, "learning_rate": 0.0001, "elbo": 3.65481, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1953436547032608}
{"step": 37440, "loss": 3.6525192260742188, "grad_norm": 0.6693265438079834, "learning_rate": 0.0001, "elbo": 3.65252, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.195636875439831}
{"step": 37445, "loss": 3.6518489837646486, "grad_norm": 0.7069966197013855, "learning_rate": 0.0001, "elbo": 3.65185, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1959300961764017}
{"step": 37450, "loss": 3.650110626220703, "grad_norm": 0.8944294452667236, "learning_rate": 0.0001, "elbo": 3.65011, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.196223316912972}
{"step": 37455, "loss": 3.643807220458984, "grad_norm": 0.7487895488739014, "learning_rate": 0.0001, "elbo": 3.64381, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1965165376495426}
{"step": 37460, "loss": 3.6465408325195314, "grad_norm": 0.7227680087089539, "learning_rate": 0.0001, "elbo": 3.64654, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.196809758386113}
{"step": 37465, "loss": 3.6379444122314455, "grad_norm": 0.7389345765113831, "learning_rate": 0.0001, "elbo": 3.63794, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1971029791226835}
{"step": 37470, "loss": 3.6556739807128906, "grad_norm": 0.6731668710708618, "learning_rate": 0.0001, "elbo": 3.65567, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.197396199859254}
{"step": 37475, "loss": 3.6586685180664062, "grad_norm": 0.6618505120277405, "learning_rate": 0.0001, "elbo": 3.65867, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1976894205958244}
{"step": 37480, "loss": 3.6332466125488283, "grad_norm": 0.7376230359077454, "learning_rate": 0.0001, "elbo": 3.63325, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.197982641332395}
{"step": 37485, "loss": 3.661063385009766, "grad_norm": 0.6568466424942017, "learning_rate": 0.0001, "elbo": 3.66106, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1982758620689653}
{"step": 37490, "loss": 3.6475025177001954, "grad_norm": 0.66353839635849, "learning_rate": 0.0001, "elbo": 3.6475, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.198569082805536}
{"step": 37495, "loss": 3.63575439453125, "grad_norm": 0.6677858829498291, "learning_rate": 0.0001, "elbo": 3.63575, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1988623035421067}
{"step": 37500, "loss": 3.6456275939941407, "grad_norm": 0.669478178024292, "learning_rate": 0.0001, "elbo": 3.64563, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.199155524278677}
{"step": 37505, "loss": 3.6549041748046873, "grad_norm": 0.7256956100463867, "learning_rate": 0.0001, "elbo": 3.6549, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.1994487450152476}
{"step": 37510, "loss": 3.64453125, "grad_norm": 0.6991504430770874, "learning_rate": 0.0001, "elbo": 3.64453, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.199741965751818}
{"step": 37515, "loss": 3.6407752990722657, "grad_norm": 0.6566015481948853, "learning_rate": 0.0001, "elbo": 3.64078, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2000351864883885}
{"step": 37520, "loss": 3.635700988769531, "grad_norm": 1.0479164123535156, "learning_rate": 0.0001, "elbo": 3.6357, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.200328407224959}
{"step": 37525, "loss": 3.6453277587890627, "grad_norm": 0.6721809506416321, "learning_rate": 0.0001, "elbo": 3.64533, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2006216279615294}
{"step": 37530, "loss": 3.6647430419921876, "grad_norm": 0.6667368412017822, "learning_rate": 0.0001, "elbo": 3.66474, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2009148486981}
{"step": 37535, "loss": 3.645050811767578, "grad_norm": 0.7057582139968872, "learning_rate": 0.0001, "elbo": 3.64505, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2012080694346703}
{"step": 37540, "loss": 3.6338897705078126, "grad_norm": 0.7349385619163513, "learning_rate": 0.0001, "elbo": 3.63389, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2015012901712407}
{"step": 37545, "loss": 3.648028564453125, "grad_norm": 0.6764199137687683, "learning_rate": 0.0001, "elbo": 3.64803, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.201794510907811}
{"step": 37550, "loss": 3.653255844116211, "grad_norm": 0.6469177603721619, "learning_rate": 0.0001, "elbo": 3.65326, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.202087731644382}
{"step": 37555, "loss": 3.6563652038574217, "grad_norm": 0.6769598126411438, "learning_rate": 0.0001, "elbo": 3.65637, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2023809523809526}
{"step": 37560, "loss": 3.64945068359375, "grad_norm": 0.6962593793869019, "learning_rate": 0.0001, "elbo": 3.64945, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.202674173117523}
{"step": 37565, "loss": 3.650995635986328, "grad_norm": 0.9162687659263611, "learning_rate": 0.0001, "elbo": 3.651, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2029673938540935}
{"step": 37570, "loss": 3.6492706298828126, "grad_norm": 0.679828941822052, "learning_rate": 0.0001, "elbo": 3.64927, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.203260614590664}
{"step": 37575, "loss": 3.648651123046875, "grad_norm": 0.6362376809120178, "learning_rate": 0.0001, "elbo": 3.64865, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2035538353272344}
{"step": 37580, "loss": 3.636016082763672, "grad_norm": 0.6817545294761658, "learning_rate": 0.0001, "elbo": 3.63602, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.203847056063805}
{"step": 37585, "loss": 3.6317176818847656, "grad_norm": 0.6895297169685364, "learning_rate": 0.0001, "elbo": 3.63172, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2041402768003753}
{"step": 37590, "loss": 3.639967346191406, "grad_norm": 0.754683256149292, "learning_rate": 0.0001, "elbo": 3.63997, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2044334975369457}
{"step": 37595, "loss": 3.6379127502441406, "grad_norm": 0.7921438813209534, "learning_rate": 0.0001, "elbo": 3.63791, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.204726718273516}
{"step": 37600, "loss": 3.6364593505859375, "grad_norm": 0.6508549451828003, "learning_rate": 0.0001, "elbo": 3.63646, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2050199390100866}
{"step": 37605, "loss": 3.6452198028564453, "grad_norm": 0.6726084351539612, "learning_rate": 0.0001, "elbo": 3.64522, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.205313159746657}
{"step": 37610, "loss": 3.6503013610839843, "grad_norm": 0.6622437834739685, "learning_rate": 0.0001, "elbo": 3.6503, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2056063804832275}
{"step": 37615, "loss": 3.6303550720214846, "grad_norm": 0.6532472968101501, "learning_rate": 0.0001, "elbo": 3.63036, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2058996012197984}
{"step": 37620, "loss": 3.6391849517822266, "grad_norm": 0.6894251704216003, "learning_rate": 0.0001, "elbo": 3.63918, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.206192821956369}
{"step": 37625, "loss": 3.642945098876953, "grad_norm": 0.6818212270736694, "learning_rate": 0.0001, "elbo": 3.64294, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2064860426929394}
{"step": 37625, "loss": 3.642945098876953, "grad_norm": 0.6818212270736694, "learning_rate": 0.0001, "elbo": 3.64294, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2064860426929394}
{"step": 37630, "loss": 3.6479824066162108, "grad_norm": 0.7197815179824829, "learning_rate": 0.0001, "elbo": 3.64798, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.20677926342951}
{"step": 37635, "loss": 3.6534549713134767, "grad_norm": 0.67743980884552, "learning_rate": 0.0001, "elbo": 3.65345, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2070724841660803}
{"step": 37640, "loss": 3.6293617248535157, "grad_norm": 0.8159220218658447, "learning_rate": 0.0001, "elbo": 3.62936, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2073657049026507}
{"step": 37645, "loss": 3.6398109436035155, "grad_norm": 0.676569402217865, "learning_rate": 0.0001, "elbo": 3.63981, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.207658925639221}
{"step": 37650, "loss": 3.6384723663330076, "grad_norm": 0.7510376572608948, "learning_rate": 0.0001, "elbo": 3.63847, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2079521463757916}
{"step": 37655, "loss": 3.640198516845703, "grad_norm": 0.6553201079368591, "learning_rate": 0.0001, "elbo": 3.6402, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.208245367112362}
{"step": 37660, "loss": 3.6593696594238283, "grad_norm": 0.8768066167831421, "learning_rate": 0.0001, "elbo": 3.65937, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2085385878489325}
{"step": 37665, "loss": 3.6452232360839845, "grad_norm": 0.6994614005088806, "learning_rate": 0.0001, "elbo": 3.64522, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.208831808585503}
{"step": 37670, "loss": 3.64892463684082, "grad_norm": 0.7535783648490906, "learning_rate": 0.0001, "elbo": 3.64892, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.209125029322074}
{"step": 37675, "loss": 3.6399803161621094, "grad_norm": 0.6601086854934692, "learning_rate": 0.0001, "elbo": 3.63998, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2094182500586443}
{"step": 37680, "loss": 3.6449050903320312, "grad_norm": 0.6825560927391052, "learning_rate": 0.0001, "elbo": 3.6449, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.209711470795215}
{"step": 37685, "loss": 3.649454116821289, "grad_norm": 0.6968007683753967, "learning_rate": 0.0001, "elbo": 3.64945, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2100046915317852}
{"step": 37690, "loss": 3.6561683654785155, "grad_norm": 0.7117652297019958, "learning_rate": 0.0001, "elbo": 3.65617, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2102979122683557}
{"step": 37695, "loss": 3.6557632446289063, "grad_norm": 0.7137584090232849, "learning_rate": 0.0001, "elbo": 3.65576, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.210591133004926}
{"step": 37700, "loss": 3.645915985107422, "grad_norm": 0.9808259010314941, "learning_rate": 0.0001, "elbo": 3.64592, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2108843537414966}
{"step": 37705, "loss": 3.649688720703125, "grad_norm": 0.661124050617218, "learning_rate": 0.0001, "elbo": 3.64969, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.211177574478067}
{"step": 37705, "loss": 3.649688720703125, "grad_norm": 0.661124050617218, "learning_rate": 0.0001, "elbo": 3.64969, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.211177574478067}
{"step": 37710, "loss": 3.627552032470703, "grad_norm": 0.6805030107498169, "learning_rate": 0.0001, "elbo": 3.62755, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2114707952146375}
{"step": 37715, "loss": 3.653054046630859, "grad_norm": 0.6601669788360596, "learning_rate": 0.0001, "elbo": 3.65305, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.211764015951208}
{"step": 37720, "loss": 3.6504920959472655, "grad_norm": 0.669002890586853, "learning_rate": 0.0001, "elbo": 3.65049, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2120572366877784}
{"step": 37725, "loss": 3.6664398193359373, "grad_norm": 0.6848054528236389, "learning_rate": 0.0001, "elbo": 3.66644, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.212350457424349}
{"step": 37730, "loss": 3.6320098876953124, "grad_norm": 0.827226459980011, "learning_rate": 0.0001, "elbo": 3.63201, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2126436781609193}
{"step": 37735, "loss": 3.648692321777344, "grad_norm": 0.8097834587097168, "learning_rate": 0.0001, "elbo": 3.64869, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2129368988974902}
{"step": 37740, "loss": 3.640595245361328, "grad_norm": 0.7156245112419128, "learning_rate": 0.0001, "elbo": 3.6406, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2132301196340607}
{"step": 37745, "loss": 3.645252227783203, "grad_norm": 0.6558582186698914, "learning_rate": 0.0001, "elbo": 3.64525, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.213523340370631}
{"step": 37750, "loss": 3.629209518432617, "grad_norm": 0.6337959170341492, "learning_rate": 0.0001, "elbo": 3.62921, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2138165611072016}
{"step": 37755, "loss": 3.6471019744873048, "grad_norm": 0.6823871731758118, "learning_rate": 0.0001, "elbo": 3.6471, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.214109781843772}
{"step": 37760, "loss": 3.660810089111328, "grad_norm": 0.8048664331436157, "learning_rate": 0.0001, "elbo": 3.66081, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2144030025803425}
{"step": 37765, "loss": 3.6540435791015624, "grad_norm": 0.6730836629867554, "learning_rate": 0.0001, "elbo": 3.65404, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.214696223316913}
{"step": 37770, "loss": 3.640924072265625, "grad_norm": 0.6981021165847778, "learning_rate": 0.0001, "elbo": 3.64092, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2149894440534834}
{"step": 37775, "loss": 3.636439895629883, "grad_norm": 0.9489254951477051, "learning_rate": 0.0001, "elbo": 3.63644, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.215282664790054}
{"step": 37780, "loss": 3.649962615966797, "grad_norm": 0.8192511200904846, "learning_rate": 0.0001, "elbo": 3.64996, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2155758855266243}
{"step": 37785, "loss": 3.6611557006835938, "grad_norm": 0.6644916534423828, "learning_rate": 0.0001, "elbo": 3.66116, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2158691062631948}
{"step": 37790, "loss": 3.6578899383544923, "grad_norm": 0.730640172958374, "learning_rate": 0.0001, "elbo": 3.65789, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2161623269997652}
{"step": 37795, "loss": 3.6580848693847656, "grad_norm": 0.9803938865661621, "learning_rate": 0.0001, "elbo": 3.65808, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.216455547736336}
{"step": 37800, "loss": 3.6521331787109377, "grad_norm": 1.0260950326919556, "learning_rate": 0.0001, "elbo": 3.65213, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2167487684729066}
{"step": 37805, "loss": 3.648814392089844, "grad_norm": 1.234226107597351, "learning_rate": 0.0001, "elbo": 3.64881, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.217041989209477}
{"step": 37810, "loss": 3.6443149566650392, "grad_norm": 0.7702385783195496, "learning_rate": 0.0001, "elbo": 3.64431, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2173352099460475}
{"step": 37815, "loss": 3.6555248260498048, "grad_norm": 1.1308575868606567, "learning_rate": 0.0001, "elbo": 3.65552, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.217628430682618}
{"step": 37820, "loss": 3.636679840087891, "grad_norm": 0.6773462295532227, "learning_rate": 0.0001, "elbo": 3.63668, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2179216514191884}
{"step": 37825, "loss": 3.661244201660156, "grad_norm": 0.9234467148780823, "learning_rate": 0.0001, "elbo": 3.66124, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.218214872155759}
{"step": 37830, "loss": 3.6606212615966798, "grad_norm": 0.6949385404586792, "learning_rate": 0.0001, "elbo": 3.66062, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2185080928923293}
{"step": 37835, "loss": 3.651934814453125, "grad_norm": 0.6466611623764038, "learning_rate": 0.0001, "elbo": 3.65193, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2188013136288998}
{"step": 37840, "loss": 3.635845184326172, "grad_norm": 0.8216529488563538, "learning_rate": 0.0001, "elbo": 3.63585, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.21909453436547}
{"step": 37845, "loss": 3.6389598846435547, "grad_norm": 0.8763930201530457, "learning_rate": 0.0001, "elbo": 3.63896, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2193877551020407}
{"step": 37850, "loss": 3.6400238037109376, "grad_norm": 0.6817775368690491, "learning_rate": 0.0001, "elbo": 3.64002, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.219680975838611}
{"step": 37855, "loss": 3.658028411865234, "grad_norm": 1.140641212463379, "learning_rate": 0.0001, "elbo": 3.65803, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2199741965751816}
{"step": 37860, "loss": 3.629803466796875, "grad_norm": 0.6599587798118591, "learning_rate": 0.0001, "elbo": 3.6298, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2202674173117525}
{"step": 37865, "loss": 3.6471584320068358, "grad_norm": 0.6669871807098389, "learning_rate": 0.0001, "elbo": 3.64716, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.220560638048323}
{"step": 37870, "loss": 3.6447895050048826, "grad_norm": 0.7008251547813416, "learning_rate": 0.0001, "elbo": 3.64479, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2208538587848934}
{"step": 37875, "loss": 3.647939682006836, "grad_norm": 0.6447592377662659, "learning_rate": 0.0001, "elbo": 3.64794, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.221147079521464}
{"step": 37880, "loss": 3.6436790466308593, "grad_norm": 0.828143835067749, "learning_rate": 0.0001, "elbo": 3.64368, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2214403002580343}
{"step": 37885, "loss": 3.6472122192382814, "grad_norm": 0.6677626967430115, "learning_rate": 0.0001, "elbo": 3.64721, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2217335209946047}
{"step": 37890, "loss": 3.635150146484375, "grad_norm": 0.6127424836158752, "learning_rate": 0.0001, "elbo": 3.63515, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.222026741731175}
{"step": 37895, "loss": 3.6284278869628905, "grad_norm": 0.7030892968177795, "learning_rate": 0.0001, "elbo": 3.62843, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2223199624677457}
{"step": 37900, "loss": 3.639936828613281, "grad_norm": 0.6542993783950806, "learning_rate": 0.0001, "elbo": 3.63994, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.222613183204316}
{"step": 37905, "loss": 3.6414325714111326, "grad_norm": 0.8166285753250122, "learning_rate": 0.0001, "elbo": 3.64143, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2229064039408866}
{"step": 37910, "loss": 3.655290222167969, "grad_norm": 0.6834231615066528, "learning_rate": 0.0001, "elbo": 3.65529, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.223199624677457}
{"step": 37915, "loss": 3.650401306152344, "grad_norm": 0.705629289150238, "learning_rate": 0.0001, "elbo": 3.6504, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2234928454140275}
{"step": 37920, "loss": 3.648634338378906, "grad_norm": 0.6676238775253296, "learning_rate": 0.0001, "elbo": 3.64863, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2237860661505984}
{"step": 37925, "loss": 3.642069625854492, "grad_norm": 0.8384914398193359, "learning_rate": 0.0001, "elbo": 3.64207, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.224079286887169}
{"step": 37930, "loss": 3.6430225372314453, "grad_norm": 0.7610584497451782, "learning_rate": 0.0001, "elbo": 3.64302, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2243725076237393}
{"step": 37935, "loss": 3.655714416503906, "grad_norm": 0.6718111038208008, "learning_rate": 0.0001, "elbo": 3.65571, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2246657283603097}
{"step": 37940, "loss": 3.6506011962890623, "grad_norm": 0.6434668898582458, "learning_rate": 0.0001, "elbo": 3.6506, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.22495894909688}
{"step": 37945, "loss": 3.6369991302490234, "grad_norm": 0.7026395797729492, "learning_rate": 0.0001, "elbo": 3.637, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2252521698334506}
{"step": 37950, "loss": 3.6359481811523438, "grad_norm": 0.7747503519058228, "learning_rate": 0.0001, "elbo": 3.63595, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.225545390570021}
{"step": 37955, "loss": 3.646004486083984, "grad_norm": 0.6746135950088501, "learning_rate": 0.0001, "elbo": 3.646, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2258386113065916}
{"step": 37955, "loss": 3.646004486083984, "grad_norm": 0.6746135950088501, "learning_rate": 0.0001, "elbo": 3.646, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2258386113065916}
{"step": 37960, "loss": 3.6529701232910154, "grad_norm": 0.6979548931121826, "learning_rate": 0.0001, "elbo": 3.65297, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.226131832043162}
{"step": 37960, "loss": 3.6529701232910154, "grad_norm": 0.6979548931121826, "learning_rate": 0.0001, "elbo": 3.65297, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.226131832043162}
{"step": 37965, "loss": 3.643950271606445, "grad_norm": 0.7630691528320312, "learning_rate": 0.0001, "elbo": 3.64395, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2264250527797325}
{"step": 37970, "loss": 3.650046157836914, "grad_norm": 0.659161388874054, "learning_rate": 0.0001, "elbo": 3.65005, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.226718273516303}
{"step": 37975, "loss": 3.658333587646484, "grad_norm": 0.6808583736419678, "learning_rate": 0.0001, "elbo": 3.65833, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2270114942528734}
{"step": 37980, "loss": 3.6529460906982423, "grad_norm": 1.088668704032898, "learning_rate": 0.0001, "elbo": 3.65295, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.227304714989444}
{"step": 37985, "loss": 3.6424026489257812, "grad_norm": 0.8326000571250916, "learning_rate": 0.0001, "elbo": 3.6424, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2275979357260147}
{"step": 37985, "loss": 3.6424026489257812, "grad_norm": 0.8326000571250916, "learning_rate": 0.0001, "elbo": 3.6424, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2275979357260147}
{"step": 37990, "loss": 3.66689453125, "grad_norm": 0.6763960123062134, "learning_rate": 0.0001, "elbo": 3.66689, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.227891156462585}
{"step": 37995, "loss": 3.6466259002685546, "grad_norm": 0.7340639233589172, "learning_rate": 0.0001, "elbo": 3.64663, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2281843771991556}
{"step": 38000, "loss": 3.653122329711914, "grad_norm": 0.6882063746452332, "learning_rate": 0.0001, "elbo": 3.65312, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.228477597935726}
{"step": 38005, "loss": 3.64505615234375, "grad_norm": 0.6828936338424683, "learning_rate": 0.0001, "elbo": 3.64506, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2287708186722965}
{"step": 38010, "loss": 3.6506622314453123, "grad_norm": 0.6324201822280884, "learning_rate": 0.0001, "elbo": 3.65066, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.229064039408867}
{"step": 38015, "loss": 3.6539730072021483, "grad_norm": 0.6569666862487793, "learning_rate": 0.0001, "elbo": 3.65397, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2293572601454374}
{"step": 38020, "loss": 3.6416961669921877, "grad_norm": 0.7163485884666443, "learning_rate": 0.0001, "elbo": 3.6417, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.229650480882008}
{"step": 38025, "loss": 3.6478908538818358, "grad_norm": 0.6972734332084656, "learning_rate": 0.0001, "elbo": 3.64789, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2299437016185784}
{"step": 38030, "loss": 3.6435447692871095, "grad_norm": 0.6727633476257324, "learning_rate": 0.0001, "elbo": 3.64354, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.230236922355149}
{"step": 38035, "loss": 3.6462745666503906, "grad_norm": 0.6617079973220825, "learning_rate": 0.0001, "elbo": 3.64627, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2305301430917193}
{"step": 38040, "loss": 3.6395843505859373, "grad_norm": 0.7025079727172852, "learning_rate": 0.0001, "elbo": 3.63958, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.23082336382829}
{"step": 38045, "loss": 3.651946258544922, "grad_norm": 0.7496087551116943, "learning_rate": 0.0001, "elbo": 3.65195, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2311165845648606}
{"step": 38050, "loss": 3.6546363830566406, "grad_norm": 0.7251245379447937, "learning_rate": 0.0001, "elbo": 3.65464, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.231409805301431}
{"step": 38055, "loss": 3.633881378173828, "grad_norm": 0.70925372838974, "learning_rate": 0.0001, "elbo": 3.63388, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2317030260380015}
{"step": 38060, "loss": 3.6296329498291016, "grad_norm": 0.7651588320732117, "learning_rate": 0.0001, "elbo": 3.62963, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.231996246774572}
{"step": 38065, "loss": 3.6416767120361326, "grad_norm": 0.679157555103302, "learning_rate": 0.0001, "elbo": 3.64168, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2322894675111424}
{"step": 38065, "loss": 3.6416767120361326, "grad_norm": 0.679157555103302, "learning_rate": 0.0001, "elbo": 3.64168, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2322894675111424}
{"step": 38070, "loss": 3.631658172607422, "grad_norm": 0.6604881286621094, "learning_rate": 0.0001, "elbo": 3.63166, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.232582688247713}
{"step": 38075, "loss": 3.641560745239258, "grad_norm": 0.6982747912406921, "learning_rate": 0.0001, "elbo": 3.64156, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2328759089842833}
{"step": 38080, "loss": 3.6273040771484375, "grad_norm": 0.7613029479980469, "learning_rate": 0.0001, "elbo": 3.6273, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.233169129720854}
{"step": 38085, "loss": 3.6555854797363283, "grad_norm": 0.77012038230896, "learning_rate": 0.0001, "elbo": 3.65559, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2334623504574243}
{"step": 38090, "loss": 3.6428314208984376, "grad_norm": 0.6785778403282166, "learning_rate": 0.0001, "elbo": 3.64283, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2337555711939947}
{"step": 38095, "loss": 3.6611953735351563, "grad_norm": 0.7747856974601746, "learning_rate": 0.0001, "elbo": 3.6612, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.234048791930565}
{"step": 38100, "loss": 3.6422935485839845, "grad_norm": 0.7757206559181213, "learning_rate": 0.0001, "elbo": 3.64229, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2343420126671356}
{"step": 38105, "loss": 3.6474029541015627, "grad_norm": 0.9520810842514038, "learning_rate": 0.0001, "elbo": 3.6474, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2346352334037065}
{"step": 38110, "loss": 3.6548130035400392, "grad_norm": 0.7289318442344666, "learning_rate": 0.0001, "elbo": 3.65481, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.234928454140277}
{"step": 38115, "loss": 3.621298980712891, "grad_norm": 0.6615511178970337, "learning_rate": 0.0001, "elbo": 3.6213, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2352216748768474}
{"step": 38120, "loss": 3.636408233642578, "grad_norm": 0.7331124544143677, "learning_rate": 0.0001, "elbo": 3.63641, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.235514895613418}
{"step": 38125, "loss": 3.6530921936035154, "grad_norm": 0.7083452939987183, "learning_rate": 0.0001, "elbo": 3.65309, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2358081163499883}
{"step": 38130, "loss": 3.6520664215087892, "grad_norm": 0.7808719277381897, "learning_rate": 0.0001, "elbo": 3.65207, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.236101337086559}
{"step": 38135, "loss": 3.6360637664794924, "grad_norm": 0.7240208387374878, "learning_rate": 0.0001, "elbo": 3.63606, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2363945578231292}
{"step": 38140, "loss": 3.646163558959961, "grad_norm": 0.6938412189483643, "learning_rate": 0.0001, "elbo": 3.64616, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2366877785596997}
{"step": 38140, "loss": 3.646163558959961, "grad_norm": 0.6938412189483643, "learning_rate": 0.0001, "elbo": 3.64616, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2366877785596997}
{"step": 38145, "loss": 3.6466461181640626, "grad_norm": 0.6553600430488586, "learning_rate": 0.0001, "elbo": 3.64665, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.23698099929627}
{"step": 38150, "loss": 3.6388259887695313, "grad_norm": 0.7152518630027771, "learning_rate": 0.0001, "elbo": 3.63883, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2372742200328406}
{"step": 38155, "loss": 3.6470077514648436, "grad_norm": 0.6180185675621033, "learning_rate": 0.0001, "elbo": 3.64701, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.237567440769411}
{"step": 38160, "loss": 3.6564918518066407, "grad_norm": 0.685491144657135, "learning_rate": 0.0001, "elbo": 3.65649, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2378606615059815}
{"step": 38165, "loss": 3.6466930389404295, "grad_norm": 0.7108525037765503, "learning_rate": 0.0001, "elbo": 3.64669, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2381538822425524}
{"step": 38170, "loss": 3.6484329223632814, "grad_norm": 0.692322313785553, "learning_rate": 0.0001, "elbo": 3.64843, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.238447102979123}
{"step": 38170, "loss": 3.6484329223632814, "grad_norm": 0.692322313785553, "learning_rate": 0.0001, "elbo": 3.64843, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.238447102979123}
{"step": 38175, "loss": 3.625578689575195, "grad_norm": 0.6518888473510742, "learning_rate": 0.0001, "elbo": 3.62558, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2387403237156933}
{"step": 38180, "loss": 3.6390220642089846, "grad_norm": 0.7089886665344238, "learning_rate": 0.0001, "elbo": 3.63902, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2390335444522638}
{"step": 38185, "loss": 3.650809478759766, "grad_norm": 0.7992385029792786, "learning_rate": 0.0001, "elbo": 3.65081, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2393267651888342}
{"step": 38190, "loss": 3.6519920349121096, "grad_norm": 0.6862541437149048, "learning_rate": 0.0001, "elbo": 3.65199, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2396199859254047}
{"step": 38195, "loss": 3.632212829589844, "grad_norm": 0.7930009961128235, "learning_rate": 0.0001, "elbo": 3.63221, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.239913206661975}
{"step": 38200, "loss": 3.640654373168945, "grad_norm": 0.6462077498435974, "learning_rate": 0.0001, "elbo": 3.64065, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2402064273985456}
{"step": 38205, "loss": 3.651516342163086, "grad_norm": 0.7035899758338928, "learning_rate": 0.0001, "elbo": 3.65152, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.240499648135116}
{"step": 38210, "loss": 3.651325225830078, "grad_norm": 0.7171152234077454, "learning_rate": 0.0001, "elbo": 3.65133, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2407928688716865}
{"step": 38215, "loss": 3.648676300048828, "grad_norm": 0.7375327348709106, "learning_rate": 0.0001, "elbo": 3.64868, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.241086089608257}
{"step": 38220, "loss": 3.6354122161865234, "grad_norm": 0.682572603225708, "learning_rate": 0.0001, "elbo": 3.63541, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2413793103448274}
{"step": 38225, "loss": 3.6374393463134767, "grad_norm": 0.6859633922576904, "learning_rate": 0.0001, "elbo": 3.63744, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.241672531081398}
{"step": 38230, "loss": 3.643305206298828, "grad_norm": 0.697231113910675, "learning_rate": 0.0001, "elbo": 3.64331, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2419657518179688}
{"step": 38235, "loss": 3.6311416625976562, "grad_norm": 0.6930292248725891, "learning_rate": 0.0001, "elbo": 3.63114, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.242258972554539}
{"step": 38235, "loss": 3.6311416625976562, "grad_norm": 0.6930292248725891, "learning_rate": 0.0001, "elbo": 3.63114, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.242258972554539}
{"step": 38240, "loss": 3.6588172912597656, "grad_norm": 0.6523206233978271, "learning_rate": 0.0001, "elbo": 3.65882, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2425521932911097}
{"step": 38245, "loss": 3.6480567932128904, "grad_norm": 0.6557092070579529, "learning_rate": 0.0001, "elbo": 3.64806, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.24284541402768}
{"step": 38250, "loss": 3.643098831176758, "grad_norm": 0.6692729592323303, "learning_rate": 0.0001, "elbo": 3.6431, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2431386347642506}
{"step": 38255, "loss": 3.6317626953125, "grad_norm": 0.6991935968399048, "learning_rate": 0.0001, "elbo": 3.63176, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.243431855500821}
{"step": 38260, "loss": 3.648722839355469, "grad_norm": 0.6381174921989441, "learning_rate": 0.0001, "elbo": 3.64872, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2437250762373915}
{"step": 38265, "loss": 3.6447471618652343, "grad_norm": 0.6850966215133667, "learning_rate": 0.0001, "elbo": 3.64475, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.244018296973962}
{"step": 38270, "loss": 3.6454933166503904, "grad_norm": 0.6619354486465454, "learning_rate": 0.0001, "elbo": 3.64549, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2443115177105324}
{"step": 38275, "loss": 3.6476905822753904, "grad_norm": 0.7121937870979309, "learning_rate": 0.0001, "elbo": 3.64769, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.244604738447103}
{"step": 38280, "loss": 3.632729339599609, "grad_norm": 0.6745389699935913, "learning_rate": 0.0001, "elbo": 3.63273, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2448979591836733}
{"step": 38285, "loss": 3.6309242248535156, "grad_norm": 0.6421409845352173, "learning_rate": 0.0001, "elbo": 3.63092, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.245191179920244}
{"step": 38290, "loss": 3.6486465454101564, "grad_norm": 0.6804013848304749, "learning_rate": 0.0001, "elbo": 3.64865, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2454844006568146}
{"step": 38295, "loss": 3.6257843017578124, "grad_norm": 0.6837617754936218, "learning_rate": 0.0001, "elbo": 3.62578, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.245777621393385}
{"step": 38300, "loss": 3.6450592041015626, "grad_norm": 0.6959997415542603, "learning_rate": 0.0001, "elbo": 3.64506, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2460708421299556}
{"step": 38305, "loss": 3.6484893798828124, "grad_norm": 0.6808958649635315, "learning_rate": 0.0001, "elbo": 3.64849, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.246364062866526}
{"step": 38310, "loss": 3.6183616638183596, "grad_norm": 0.6957902312278748, "learning_rate": 0.0001, "elbo": 3.61836, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2466572836030965}
{"step": 38315, "loss": 3.6402339935302734, "grad_norm": 0.9168188571929932, "learning_rate": 0.0001, "elbo": 3.64023, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.246950504339667}
{"step": 38320, "loss": 3.652094268798828, "grad_norm": 0.7257848978042603, "learning_rate": 0.0001, "elbo": 3.65209, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2472437250762374}
{"step": 38325, "loss": 3.651007080078125, "grad_norm": 0.7968154549598694, "learning_rate": 0.0001, "elbo": 3.65101, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.247536945812808}
{"step": 38330, "loss": 3.635057830810547, "grad_norm": 0.6718509793281555, "learning_rate": 0.0001, "elbo": 3.63506, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2478301665493783}
{"step": 38335, "loss": 3.6482269287109377, "grad_norm": 0.6929446458816528, "learning_rate": 0.0001, "elbo": 3.64823, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2481233872859487}
{"step": 38340, "loss": 3.6286041259765627, "grad_norm": 0.7047758102416992, "learning_rate": 0.0001, "elbo": 3.6286, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.248416608022519}
{"step": 38345, "loss": 3.654236602783203, "grad_norm": 0.6704776287078857, "learning_rate": 0.0001, "elbo": 3.65424, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2487098287590896}
{"step": 38350, "loss": 3.625196838378906, "grad_norm": 0.6759542226791382, "learning_rate": 0.0001, "elbo": 3.6252, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.24900304949566}
{"step": 38355, "loss": 3.641509246826172, "grad_norm": 0.7189021706581116, "learning_rate": 0.0001, "elbo": 3.64151, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.249296270232231}
{"step": 38360, "loss": 3.6488834381103517, "grad_norm": 0.6968679428100586, "learning_rate": 0.0001, "elbo": 3.64888, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2495894909688015}
{"step": 38365, "loss": 3.630517578125, "grad_norm": 0.7711222767829895, "learning_rate": 0.0001, "elbo": 3.63052, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.249882711705372}
{"step": 38370, "loss": 3.6377525329589844, "grad_norm": 0.6823127865791321, "learning_rate": 0.0001, "elbo": 3.63775, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2501759324419424}
{"step": 38375, "loss": 3.654230499267578, "grad_norm": 0.8124001026153564, "learning_rate": 0.0001, "elbo": 3.65423, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.250469153178513}
{"step": 38380, "loss": 3.628379058837891, "grad_norm": 1.0165724754333496, "learning_rate": 0.0001, "elbo": 3.62838, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2507623739150833}
{"step": 38385, "loss": 3.6419441223144533, "grad_norm": 0.8568947315216064, "learning_rate": 0.0001, "elbo": 3.64194, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2510555946516537}
{"step": 38390, "loss": 3.641063690185547, "grad_norm": 1.3697762489318848, "learning_rate": 0.0001, "elbo": 3.64106, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.251348815388224}
{"step": 38395, "loss": 3.644971466064453, "grad_norm": 0.73967444896698, "learning_rate": 0.0001, "elbo": 3.64497, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2516420361247946}
{"step": 38400, "loss": 3.645928955078125, "grad_norm": 0.6931009292602539, "learning_rate": 0.0001, "elbo": 3.64593, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.251935256861365}
{"step": 38405, "loss": 3.644883728027344, "grad_norm": 1.029780387878418, "learning_rate": 0.0001, "elbo": 3.64488, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2522284775979355}
{"step": 38410, "loss": 3.6508216857910156, "grad_norm": 0.699586808681488, "learning_rate": 0.0001, "elbo": 3.65082, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2525216983345064}
{"step": 38415, "loss": 3.649120330810547, "grad_norm": 0.6848483681678772, "learning_rate": 0.0001, "elbo": 3.64912, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.252814919071077}
{"step": 38420, "loss": 3.6327930450439454, "grad_norm": 0.6788321137428284, "learning_rate": 0.0001, "elbo": 3.63279, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2531081398076473}
{"step": 38425, "loss": 3.6245635986328124, "grad_norm": 0.7393759489059448, "learning_rate": 0.0001, "elbo": 3.62456, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.253401360544218}
{"step": 38430, "loss": 3.6467056274414062, "grad_norm": 0.6784553527832031, "learning_rate": 0.0001, "elbo": 3.64671, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2536945812807883}
{"step": 38435, "loss": 3.656271743774414, "grad_norm": 0.7231199145317078, "learning_rate": 0.0001, "elbo": 3.65627, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2539878020173587}
{"step": 38440, "loss": 3.620624542236328, "grad_norm": 0.8074706792831421, "learning_rate": 0.0001, "elbo": 3.62062, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.254281022753929}
{"step": 38445, "loss": 3.6497337341308596, "grad_norm": 0.6451359391212463, "learning_rate": 0.0001, "elbo": 3.64973, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2545742434904996}
{"step": 38450, "loss": 3.647045135498047, "grad_norm": 0.7064551711082458, "learning_rate": 0.0001, "elbo": 3.64705, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.25486746422707}
{"step": 38455, "loss": 3.641764831542969, "grad_norm": 0.671137273311615, "learning_rate": 0.0001, "elbo": 3.64176, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2551606849636405}
{"step": 38460, "loss": 3.6332523345947267, "grad_norm": 0.6687265634536743, "learning_rate": 0.0001, "elbo": 3.63325, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.255453905700211}
{"step": 38465, "loss": 3.6415802001953126, "grad_norm": 0.7110096216201782, "learning_rate": 0.0001, "elbo": 3.64158, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2557471264367814}
{"step": 38470, "loss": 3.6410758972167967, "grad_norm": 0.6870352029800415, "learning_rate": 0.0001, "elbo": 3.64108, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.256040347173352}
{"step": 38475, "loss": 3.633170318603516, "grad_norm": 0.6763571500778198, "learning_rate": 0.0001, "elbo": 3.63317, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2563335679099223}
{"step": 38480, "loss": 3.623088836669922, "grad_norm": 0.6775046586990356, "learning_rate": 0.0001, "elbo": 3.62309, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2566267886464932}
{"step": 38480, "loss": 3.623088836669922, "grad_norm": 0.6775046586990356, "learning_rate": 0.0001, "elbo": 3.62309, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2566267886464932}
{"step": 38485, "loss": 3.6387901306152344, "grad_norm": 0.7081205248832703, "learning_rate": 0.0001, "elbo": 3.63879, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2569200093830637}
{"step": 38490, "loss": 3.650913619995117, "grad_norm": 0.8189945816993713, "learning_rate": 0.0001, "elbo": 3.65091, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.257213230119634}
{"step": 38495, "loss": 3.652124786376953, "grad_norm": 0.6618097424507141, "learning_rate": 0.0001, "elbo": 3.65212, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2575064508562046}
{"step": 38500, "loss": 3.640904998779297, "grad_norm": 0.6600099205970764, "learning_rate": 0.0001, "elbo": 3.6409, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.257799671592775}
{"step": 38500, "loss": 3.640904998779297, "grad_norm": 0.6600099205970764, "learning_rate": 0.0001, "elbo": 3.6409, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.257799671592775}
{"step": 38505, "loss": 3.631726837158203, "grad_norm": 0.6780943274497986, "learning_rate": 0.0001, "elbo": 3.63173, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2580928923293455}
{"step": 38510, "loss": 3.656856918334961, "grad_norm": 0.6384804844856262, "learning_rate": 0.0001, "elbo": 3.65686, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.258386113065916}
{"step": 38515, "loss": 3.645222473144531, "grad_norm": 0.7863315343856812, "learning_rate": 0.0001, "elbo": 3.64522, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2586793338024864}
{"step": 38515, "loss": 3.645222473144531, "grad_norm": 0.7863315343856812, "learning_rate": 0.0001, "elbo": 3.64522, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2586793338024864}
{"step": 38520, "loss": 3.6410282135009764, "grad_norm": 0.7139403820037842, "learning_rate": 0.0001, "elbo": 3.64103, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.258972554539057}
{"step": 38525, "loss": 3.6644912719726563, "grad_norm": 0.7113676071166992, "learning_rate": 0.0001, "elbo": 3.66449, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2592657752756273}
{"step": 38530, "loss": 3.651417541503906, "grad_norm": 0.7320267558097839, "learning_rate": 0.0001, "elbo": 3.65142, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2595589960121982}
{"step": 38535, "loss": 3.648395538330078, "grad_norm": 0.6813434958457947, "learning_rate": 0.0001, "elbo": 3.6484, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2598522167487687}
{"step": 38540, "loss": 3.630614471435547, "grad_norm": 0.7060171365737915, "learning_rate": 0.0001, "elbo": 3.63061, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.260145437485339}
{"step": 38545, "loss": 3.633300018310547, "grad_norm": 0.6979672908782959, "learning_rate": 0.0001, "elbo": 3.6333, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2604386582219096}
{"step": 38550, "loss": 3.642783355712891, "grad_norm": 0.8427101373672485, "learning_rate": 0.0001, "elbo": 3.64278, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.26073187895848}
{"step": 38555, "loss": 3.6453926086425783, "grad_norm": 0.7025339603424072, "learning_rate": 0.0001, "elbo": 3.64539, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2610250996950505}
{"step": 38560, "loss": 3.6486751556396486, "grad_norm": 0.6639135479927063, "learning_rate": 0.0001, "elbo": 3.64868, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.261318320431621}
{"step": 38565, "loss": 3.6426666259765623, "grad_norm": 0.6713992357254028, "learning_rate": 0.0001, "elbo": 3.64267, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2616115411681914}
{"step": 38570, "loss": 3.6291595458984376, "grad_norm": 0.6784399151802063, "learning_rate": 0.0001, "elbo": 3.62916, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.261904761904762}
{"step": 38575, "loss": 3.635315704345703, "grad_norm": 0.6932504773139954, "learning_rate": 0.0001, "elbo": 3.63532, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2621979826413323}
{"step": 38580, "loss": 3.6526073455810546, "grad_norm": 0.7479699850082397, "learning_rate": 0.0001, "elbo": 3.65261, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2624912033779028}
{"step": 38585, "loss": 3.6341522216796873, "grad_norm": 0.6981755495071411, "learning_rate": 0.0001, "elbo": 3.63415, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2627844241144732}
{"step": 38590, "loss": 3.6203765869140625, "grad_norm": 0.7173205018043518, "learning_rate": 0.0001, "elbo": 3.62038, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2630776448510437}
{"step": 38590, "loss": 3.6203765869140625, "grad_norm": 0.7173205018043518, "learning_rate": 0.0001, "elbo": 3.62038, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2630776448510437}
{"step": 38595, "loss": 3.6477287292480467, "grad_norm": 0.8384847640991211, "learning_rate": 0.0001, "elbo": 3.64773, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.263370865587614}
{"step": 38600, "loss": 3.648625946044922, "grad_norm": 0.7257163524627686, "learning_rate": 0.0001, "elbo": 3.64863, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.263664086324185}
{"step": 38605, "loss": 3.6410423278808595, "grad_norm": 0.7576172351837158, "learning_rate": 0.0001, "elbo": 3.64104, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2639573070607555}
{"step": 38610, "loss": 3.6585426330566406, "grad_norm": 0.6645233631134033, "learning_rate": 0.0001, "elbo": 3.65854, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.264250527797326}
{"step": 38615, "loss": 3.644992446899414, "grad_norm": 0.6517903208732605, "learning_rate": 0.0001, "elbo": 3.64499, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2645437485338964}
{"step": 38620, "loss": 3.6446304321289062, "grad_norm": 0.7563153505325317, "learning_rate": 0.0001, "elbo": 3.64463, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.264836969270467}
{"step": 38625, "loss": 3.6451168060302734, "grad_norm": 0.7395092248916626, "learning_rate": 0.0001, "elbo": 3.64512, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2651301900070373}
{"step": 38630, "loss": 3.6422401428222657, "grad_norm": 0.669366717338562, "learning_rate": 0.0001, "elbo": 3.64224, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2654234107436078}
{"step": 38635, "loss": 3.653565216064453, "grad_norm": 0.7589920163154602, "learning_rate": 0.0001, "elbo": 3.65357, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.265716631480178}
{"step": 38640, "loss": 3.6392799377441407, "grad_norm": 0.7130433917045593, "learning_rate": 0.0001, "elbo": 3.63928, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2660098522167487}
{"step": 38645, "loss": 3.644219970703125, "grad_norm": 0.7296401858329773, "learning_rate": 0.0001, "elbo": 3.64422, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.266303072953319}
{"step": 38650, "loss": 3.6614151000976562, "grad_norm": 0.6976832747459412, "learning_rate": 0.0001, "elbo": 3.66142, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2665962936898896}
{"step": 38655, "loss": 3.6684749603271483, "grad_norm": 0.6994658708572388, "learning_rate": 0.0001, "elbo": 3.66847, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2668895144264605}
{"step": 38660, "loss": 3.6348480224609374, "grad_norm": 0.7298465371131897, "learning_rate": 0.0001, "elbo": 3.63485, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.267182735163031}
{"step": 38665, "loss": 3.6407394409179688, "grad_norm": 0.68519127368927, "learning_rate": 0.0001, "elbo": 3.64074, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2674759558996014}
{"step": 38670, "loss": 3.6421768188476564, "grad_norm": 0.6763445734977722, "learning_rate": 0.0001, "elbo": 3.64218, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.267769176636172}
{"step": 38675, "loss": 3.6281005859375, "grad_norm": 0.7054844498634338, "learning_rate": 0.0001, "elbo": 3.6281, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2680623973727423}
{"step": 38680, "loss": 3.640283203125, "grad_norm": 0.7424402832984924, "learning_rate": 0.0001, "elbo": 3.64028, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2683556181093127}
{"step": 38685, "loss": 3.633414459228516, "grad_norm": 0.6511250138282776, "learning_rate": 0.0001, "elbo": 3.63341, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.268648838845883}
{"step": 38690, "loss": 3.640604019165039, "grad_norm": 0.8468896746635437, "learning_rate": 0.0001, "elbo": 3.6406, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2689420595824537}
{"step": 38695, "loss": 3.6431625366210936, "grad_norm": 0.6722709536552429, "learning_rate": 0.0001, "elbo": 3.64316, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.269235280319024}
{"step": 38700, "loss": 3.6476314544677733, "grad_norm": 0.6471337676048279, "learning_rate": 0.0001, "elbo": 3.64763, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2695285010555946}
{"step": 38705, "loss": 3.625482177734375, "grad_norm": 0.6539000868797302, "learning_rate": 0.0001, "elbo": 3.62548, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.269821721792165}
{"step": 38710, "loss": 3.651340103149414, "grad_norm": 0.6534446477890015, "learning_rate": 0.0001, "elbo": 3.65134, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2701149425287355}
{"step": 38715, "loss": 3.653797912597656, "grad_norm": 0.7143740057945251, "learning_rate": 0.0001, "elbo": 3.6538, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.270408163265306}
{"step": 38720, "loss": 3.646075439453125, "grad_norm": 0.7055420279502869, "learning_rate": 0.0001, "elbo": 3.64608, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2707013840018764}
{"step": 38725, "loss": 3.638015365600586, "grad_norm": 0.7278444170951843, "learning_rate": 0.0001, "elbo": 3.63802, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2709946047384473}
{"step": 38730, "loss": 3.613933563232422, "grad_norm": 0.664236307144165, "learning_rate": 0.0001, "elbo": 3.61393, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2712878254750177}
{"step": 38735, "loss": 3.646731948852539, "grad_norm": 1.0295354127883911, "learning_rate": 0.0001, "elbo": 3.64673, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.271581046211588}
{"step": 38735, "loss": 3.646731948852539, "grad_norm": 1.0295354127883911, "learning_rate": 0.0001, "elbo": 3.64673, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.271581046211588}
{"step": 38740, "loss": 3.6384819030761717, "grad_norm": 0.6628722548484802, "learning_rate": 0.0001, "elbo": 3.63848, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2718742669481586}
{"step": 38745, "loss": 3.6428321838378905, "grad_norm": 0.6856558918952942, "learning_rate": 0.0001, "elbo": 3.64283, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.272167487684729}
{"step": 38750, "loss": 3.6409591674804687, "grad_norm": 0.7590873837471008, "learning_rate": 0.0001, "elbo": 3.64096, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2724607084212995}
{"step": 38755, "loss": 3.6407402038574217, "grad_norm": 0.8280096650123596, "learning_rate": 0.0001, "elbo": 3.64074, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.27275392915787}
{"step": 38760, "loss": 3.64803466796875, "grad_norm": 0.652582585811615, "learning_rate": 0.0001, "elbo": 3.64803, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2730471498944405}
{"step": 38765, "loss": 3.6420242309570314, "grad_norm": 0.773637056350708, "learning_rate": 0.0001, "elbo": 3.64202, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.273340370631011}
{"step": 38770, "loss": 3.6375694274902344, "grad_norm": 0.6804066300392151, "learning_rate": 0.0001, "elbo": 3.63757, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2736335913675814}
{"step": 38775, "loss": 3.656163787841797, "grad_norm": 0.7117475271224976, "learning_rate": 0.0001, "elbo": 3.65616, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.273926812104152}
{"step": 38780, "loss": 3.6354122161865234, "grad_norm": 0.7098143100738525, "learning_rate": 0.0001, "elbo": 3.63541, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2742200328407227}
{"step": 38785, "loss": 3.6219772338867187, "grad_norm": 0.6846170425415039, "learning_rate": 0.0001, "elbo": 3.62198, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.274513253577293}
{"step": 38790, "loss": 3.6352767944335938, "grad_norm": 0.6541948318481445, "learning_rate": 0.0001, "elbo": 3.63528, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2748064743138636}
{"step": 38795, "loss": 3.6476058959960938, "grad_norm": 0.6947044134140015, "learning_rate": 0.0001, "elbo": 3.64761, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.275099695050434}
{"step": 38800, "loss": 3.6521347045898436, "grad_norm": 0.6608049273490906, "learning_rate": 0.0001, "elbo": 3.65213, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2753929157870045}
{"step": 38805, "loss": 3.660747528076172, "grad_norm": 0.6946980357170105, "learning_rate": 0.0001, "elbo": 3.66075, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.275686136523575}
{"step": 38810, "loss": 3.638569641113281, "grad_norm": 0.7004355192184448, "learning_rate": 0.0001, "elbo": 3.63857, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2759793572601454}
{"step": 38815, "loss": 3.6349861145019533, "grad_norm": 0.6583449840545654, "learning_rate": 0.0001, "elbo": 3.63499, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.276272577996716}
{"step": 38820, "loss": 3.6463733673095704, "grad_norm": 0.7289075255393982, "learning_rate": 0.0001, "elbo": 3.64637, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2765657987332863}
{"step": 38825, "loss": 3.6385787963867187, "grad_norm": 0.6522570252418518, "learning_rate": 0.0001, "elbo": 3.63858, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.276859019469857}
{"step": 38830, "loss": 3.650164031982422, "grad_norm": 0.6733936071395874, "learning_rate": 0.0001, "elbo": 3.65016, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2771522402064273}
{"step": 38835, "loss": 3.626358413696289, "grad_norm": 0.7241461873054504, "learning_rate": 0.0001, "elbo": 3.62636, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2774454609429977}
{"step": 38840, "loss": 3.64188232421875, "grad_norm": 0.6719501614570618, "learning_rate": 0.0001, "elbo": 3.64188, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.277738681679568}
{"step": 38845, "loss": 3.650102996826172, "grad_norm": 0.671616792678833, "learning_rate": 0.0001, "elbo": 3.6501, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2780319024161386}
{"step": 38850, "loss": 3.630118560791016, "grad_norm": 0.6972079873085022, "learning_rate": 0.0001, "elbo": 3.63012, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2783251231527095}
{"step": 38855, "loss": 3.638445281982422, "grad_norm": 0.7448320984840393, "learning_rate": 0.0001, "elbo": 3.63845, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.27861834388928}
{"step": 38860, "loss": 3.64227294921875, "grad_norm": 0.6916285157203674, "learning_rate": 0.0001, "elbo": 3.64227, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2789115646258504}
{"step": 38865, "loss": 3.6337249755859373, "grad_norm": 0.719002366065979, "learning_rate": 0.0001, "elbo": 3.63373, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.279204785362421}
{"step": 38870, "loss": 3.621685791015625, "grad_norm": 0.6896326541900635, "learning_rate": 0.0001, "elbo": 3.62169, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2794980060989913}
{"step": 38875, "loss": 3.638510513305664, "grad_norm": 0.7024896144866943, "learning_rate": 0.0001, "elbo": 3.63851, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.279791226835562}
{"step": 38880, "loss": 3.6492984771728514, "grad_norm": 0.6631591320037842, "learning_rate": 0.0001, "elbo": 3.6493, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2800844475721322}
{"step": 38885, "loss": 3.6292137145996093, "grad_norm": 0.6942243576049805, "learning_rate": 0.0001, "elbo": 3.62921, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2803776683087027}
{"step": 38890, "loss": 3.6483665466308595, "grad_norm": 0.6300802230834961, "learning_rate": 0.0001, "elbo": 3.64837, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.280670889045273}
{"step": 38895, "loss": 3.6477203369140625, "grad_norm": 0.6862647533416748, "learning_rate": 0.0001, "elbo": 3.64772, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2809641097818436}
{"step": 38900, "loss": 3.635700225830078, "grad_norm": 1.0765924453735352, "learning_rate": 0.0001, "elbo": 3.6357, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2812573305184145}
{"step": 38905, "loss": 3.640431213378906, "grad_norm": 0.7148788571357727, "learning_rate": 0.0001, "elbo": 3.64043, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.281550551254985}
{"step": 38905, "loss": 3.640431213378906, "grad_norm": 0.7148788571357727, "learning_rate": 0.0001, "elbo": 3.64043, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.281550551254985}
{"step": 38910, "loss": 3.6416866302490236, "grad_norm": 0.6625014543533325, "learning_rate": 0.0001, "elbo": 3.64169, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2818437719915554}
{"step": 38915, "loss": 3.6382850646972655, "grad_norm": 0.7051976323127747, "learning_rate": 0.0001, "elbo": 3.63829, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.282136992728126}
{"step": 38920, "loss": 3.6334251403808593, "grad_norm": 0.65525883436203, "learning_rate": 0.0001, "elbo": 3.63342, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2824302134646963}
{"step": 38920, "loss": 3.6334251403808593, "grad_norm": 0.65525883436203, "learning_rate": 0.0001, "elbo": 3.63342, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2824302134646963}
{"step": 38925, "loss": 3.6400077819824217, "grad_norm": 0.6862729787826538, "learning_rate": 0.0001, "elbo": 3.64001, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2827234342012668}
{"step": 38930, "loss": 3.6431743621826174, "grad_norm": 0.67789226770401, "learning_rate": 0.0001, "elbo": 3.64317, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2830166549378372}
{"step": 38935, "loss": 3.634257507324219, "grad_norm": 0.6456884145736694, "learning_rate": 0.0001, "elbo": 3.63426, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2833098756744077}
{"step": 38940, "loss": 3.643840026855469, "grad_norm": 0.6932279467582703, "learning_rate": 0.0001, "elbo": 3.64384, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.283603096410978}
{"step": 38945, "loss": 3.624268341064453, "grad_norm": 0.993968665599823, "learning_rate": 0.0001, "elbo": 3.62427, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2838963171475486}
{"step": 38950, "loss": 3.6255752563476564, "grad_norm": 0.6995770335197449, "learning_rate": 0.0001, "elbo": 3.62557, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.284189537884119}
{"step": 38955, "loss": 3.643733596801758, "grad_norm": 0.6734707951545715, "learning_rate": 0.0001, "elbo": 3.64373, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2844827586206895}
{"step": 38960, "loss": 3.6416988372802734, "grad_norm": 0.662887454032898, "learning_rate": 0.0001, "elbo": 3.6417, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.28477597935726}
{"step": 38965, "loss": 3.6447685241699217, "grad_norm": 0.6791989207267761, "learning_rate": 0.0001, "elbo": 3.64477, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2850692000938304}
{"step": 38970, "loss": 3.6436500549316406, "grad_norm": 1.0421838760375977, "learning_rate": 0.0001, "elbo": 3.64365, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2853624208304013}
{"step": 38975, "loss": 3.6385990142822267, "grad_norm": 0.6851391196250916, "learning_rate": 0.0001, "elbo": 3.6386, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2856556415669718}
{"step": 38980, "loss": 3.651068115234375, "grad_norm": 0.6750171780586243, "learning_rate": 0.0001, "elbo": 3.65107, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.285948862303542}
{"step": 38980, "loss": 3.651068115234375, "grad_norm": 0.6750171780586243, "learning_rate": 0.0001, "elbo": 3.65107, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.285948862303542}
{"step": 38985, "loss": 3.638497543334961, "grad_norm": 0.6755670309066772, "learning_rate": 0.0001, "elbo": 3.6385, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2862420830401127}
{"step": 38990, "loss": 3.6494674682617188, "grad_norm": 0.9796240925788879, "learning_rate": 0.0001, "elbo": 3.64947, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.286535303776683}
{"step": 38995, "loss": 3.627284622192383, "grad_norm": 0.6472504138946533, "learning_rate": 0.0001, "elbo": 3.62728, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2868285245132536}
{"step": 39000, "loss": 3.6366931915283205, "grad_norm": 0.7465075254440308, "learning_rate": 0.0001, "elbo": 3.63669, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.287121745249824}
{"step": 39005, "loss": 3.6440780639648436, "grad_norm": 0.6441525220870972, "learning_rate": 0.0001, "elbo": 3.64408, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2874149659863945}
{"step": 39010, "loss": 3.6467830657958986, "grad_norm": 0.6889684200286865, "learning_rate": 0.0001, "elbo": 3.64678, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.287708186722965}
{"step": 39015, "loss": 3.638790512084961, "grad_norm": 0.7201344966888428, "learning_rate": 0.0001, "elbo": 3.63879, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2880014074595354}
{"step": 39020, "loss": 3.650745391845703, "grad_norm": 0.7362922430038452, "learning_rate": 0.0001, "elbo": 3.65075, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.288294628196106}
{"step": 39025, "loss": 3.6231616973876952, "grad_norm": 0.774191677570343, "learning_rate": 0.0001, "elbo": 3.62316, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2885878489326767}
{"step": 39030, "loss": 3.6085174560546873, "grad_norm": 0.6964209675788879, "learning_rate": 0.0001, "elbo": 3.60852, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.288881069669247}
{"step": 39030, "loss": 3.6085174560546873, "grad_norm": 0.6964209675788879, "learning_rate": 0.0001, "elbo": 3.60852, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.288881069669247}
{"step": 39035, "loss": 3.6379467010498048, "grad_norm": 0.6768935322761536, "learning_rate": 0.0001, "elbo": 3.63795, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2891742904058177}
{"step": 39035, "loss": 3.6379467010498048, "grad_norm": 0.6768935322761536, "learning_rate": 0.0001, "elbo": 3.63795, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2891742904058177}
{"step": 39040, "loss": 3.625935363769531, "grad_norm": 0.6693280339241028, "learning_rate": 0.0001, "elbo": 3.62594, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.289467511142388}
{"step": 39045, "loss": 3.642020034790039, "grad_norm": 0.7261494398117065, "learning_rate": 0.0001, "elbo": 3.64202, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2897607318789586}
{"step": 39045, "loss": 3.642020034790039, "grad_norm": 0.7261494398117065, "learning_rate": 0.0001, "elbo": 3.64202, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2897607318789586}
{"step": 39050, "loss": 3.6434036254882813, "grad_norm": 0.6936963796615601, "learning_rate": 0.0001, "elbo": 3.6434, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.290053952615529}
{"step": 39055, "loss": 3.639508056640625, "grad_norm": 0.6869695782661438, "learning_rate": 0.0001, "elbo": 3.63951, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2903471733520995}
{"step": 39060, "loss": 3.6471519470214844, "grad_norm": 0.6533999443054199, "learning_rate": 0.0001, "elbo": 3.64715, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.29064039408867}
{"step": 39065, "loss": 3.646038055419922, "grad_norm": 0.6897897124290466, "learning_rate": 0.0001, "elbo": 3.64604, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2909336148252404}
{"step": 39070, "loss": 3.6169059753417967, "grad_norm": 0.6668579578399658, "learning_rate": 0.0001, "elbo": 3.61691, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.291226835561811}
{"step": 39075, "loss": 3.6438308715820313, "grad_norm": 0.643014132976532, "learning_rate": 0.0001, "elbo": 3.64383, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2915200562983813}
{"step": 39080, "loss": 3.6348785400390624, "grad_norm": 0.7038586735725403, "learning_rate": 0.0001, "elbo": 3.63488, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2918132770349517}
{"step": 39085, "loss": 3.6302440643310545, "grad_norm": 0.685054361820221, "learning_rate": 0.0001, "elbo": 3.63024, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.292106497771522}
{"step": 39090, "loss": 3.6307273864746095, "grad_norm": 0.7332201600074768, "learning_rate": 0.0001, "elbo": 3.63073, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2923997185080927}
{"step": 39095, "loss": 3.6433635711669923, "grad_norm": 0.7184932231903076, "learning_rate": 0.0001, "elbo": 3.64336, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2926929392446636}
{"step": 39100, "loss": 3.631905746459961, "grad_norm": 0.7376804947853088, "learning_rate": 0.0001, "elbo": 3.63191, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.292986159981234}
{"step": 39105, "loss": 3.644434356689453, "grad_norm": 0.733039915561676, "learning_rate": 0.0001, "elbo": 3.64443, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2932793807178045}
{"step": 39110, "loss": 3.627366638183594, "grad_norm": 0.6964480876922607, "learning_rate": 0.0001, "elbo": 3.62737, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.293572601454375}
{"step": 39115, "loss": 3.6402130126953125, "grad_norm": 0.6914637088775635, "learning_rate": 0.0001, "elbo": 3.64021, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2938658221909454}
{"step": 39120, "loss": 3.6272693634033204, "grad_norm": 0.6901491284370422, "learning_rate": 0.0001, "elbo": 3.62727, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.294159042927516}
{"step": 39125, "loss": 3.631661224365234, "grad_norm": 0.6547749042510986, "learning_rate": 0.0001, "elbo": 3.63166, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2944522636640863}
{"step": 39130, "loss": 3.612415313720703, "grad_norm": 0.6933634877204895, "learning_rate": 0.0001, "elbo": 3.61242, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2947454844006567}
{"step": 39135, "loss": 3.634008026123047, "grad_norm": 0.6909404993057251, "learning_rate": 0.0001, "elbo": 3.63401, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.295038705137227}
{"step": 39140, "loss": 3.635801315307617, "grad_norm": 0.8009695410728455, "learning_rate": 0.0001, "elbo": 3.6358, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2953319258737976}
{"step": 39145, "loss": 3.6278915405273438, "grad_norm": 0.6412617564201355, "learning_rate": 0.0001, "elbo": 3.62789, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.295625146610368}
{"step": 39150, "loss": 3.624724578857422, "grad_norm": 0.678522527217865, "learning_rate": 0.0001, "elbo": 3.62472, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.295918367346939}
{"step": 39155, "loss": 3.626556396484375, "grad_norm": 0.7035751938819885, "learning_rate": 0.0001, "elbo": 3.62656, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2962115880835094}
{"step": 39160, "loss": 3.6434078216552734, "grad_norm": 0.6953189969062805, "learning_rate": 0.0001, "elbo": 3.64341, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.29650480882008}
{"step": 39165, "loss": 3.6392532348632813, "grad_norm": 0.6415892243385315, "learning_rate": 0.0001, "elbo": 3.63925, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2967980295566504}
{"step": 39170, "loss": 3.64219970703125, "grad_norm": 0.7236288785934448, "learning_rate": 0.0001, "elbo": 3.6422, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.297091250293221}
{"step": 39175, "loss": 3.6456714630126954, "grad_norm": 1.150887131690979, "learning_rate": 0.0001, "elbo": 3.64567, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2973844710297913}
{"step": 39175, "loss": 3.6456714630126954, "grad_norm": 1.150887131690979, "learning_rate": 0.0001, "elbo": 3.64567, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2973844710297913}
{"step": 39180, "loss": 3.633966827392578, "grad_norm": 0.7024731040000916, "learning_rate": 0.0001, "elbo": 3.63397, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2976776917663617}
{"step": 39185, "loss": 3.6490962982177733, "grad_norm": 0.673828661441803, "learning_rate": 0.0001, "elbo": 3.6491, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.297970912502932}
{"step": 39190, "loss": 3.657231903076172, "grad_norm": 0.6514843106269836, "learning_rate": 0.0001, "elbo": 3.65723, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2982641332395026}
{"step": 39195, "loss": 3.6329635620117187, "grad_norm": 0.6628583073616028, "learning_rate": 0.0001, "elbo": 3.63296, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.298557353976073}
{"step": 39200, "loss": 3.6448032379150392, "grad_norm": 0.6955218315124512, "learning_rate": 0.0001, "elbo": 3.6448, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2988505747126435}
{"step": 39205, "loss": 3.6432144165039064, "grad_norm": 0.6944300532341003, "learning_rate": 0.0001, "elbo": 3.64321, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.299143795449214}
{"step": 39210, "loss": 3.6274620056152345, "grad_norm": 0.697024405002594, "learning_rate": 0.0001, "elbo": 3.62746, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2994370161857844}
{"step": 39215, "loss": 3.6369644165039063, "grad_norm": 0.9102060198783875, "learning_rate": 0.0001, "elbo": 3.63696, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.2997302369223553}
{"step": 39220, "loss": 3.6400634765625, "grad_norm": 0.6782727241516113, "learning_rate": 0.0001, "elbo": 3.64006, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.300023457658926}
{"step": 39225, "loss": 3.6380348205566406, "grad_norm": 0.7805277109146118, "learning_rate": 0.0001, "elbo": 3.63803, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3003166783954963}
{"step": 39230, "loss": 3.6413337707519533, "grad_norm": 0.7065607309341431, "learning_rate": 0.0001, "elbo": 3.64133, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3006098991320667}
{"step": 39230, "loss": 3.6413337707519533, "grad_norm": 0.7065607309341431, "learning_rate": 0.0001, "elbo": 3.64133, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3006098991320667}
{"step": 39235, "loss": 3.635043716430664, "grad_norm": 0.6523930430412292, "learning_rate": 0.0001, "elbo": 3.63504, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.300903119868637}
{"step": 39240, "loss": 3.630655288696289, "grad_norm": 0.6718608736991882, "learning_rate": 0.0001, "elbo": 3.63066, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3011963406052076}
{"step": 39245, "loss": 3.627999496459961, "grad_norm": 0.7078289985656738, "learning_rate": 0.0001, "elbo": 3.628, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.301489561341778}
{"step": 39250, "loss": 3.6459850311279296, "grad_norm": 0.8252879977226257, "learning_rate": 0.0001, "elbo": 3.64598, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3017827820783485}
{"step": 39255, "loss": 3.642664337158203, "grad_norm": 0.6935837864875793, "learning_rate": 0.0001, "elbo": 3.64266, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.302076002814919}
{"step": 39260, "loss": 3.62524299621582, "grad_norm": 0.6942365169525146, "learning_rate": 0.0001, "elbo": 3.62524, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3023692235514894}
{"step": 39265, "loss": 3.634011077880859, "grad_norm": 0.7756218910217285, "learning_rate": 0.0001, "elbo": 3.63401, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.30266244428806}
{"step": 39270, "loss": 3.637567138671875, "grad_norm": 0.6514482498168945, "learning_rate": 0.0001, "elbo": 3.63757, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.302955665024631}
{"step": 39275, "loss": 3.636325454711914, "grad_norm": 0.7115737199783325, "learning_rate": 0.0001, "elbo": 3.63633, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3032488857612012}
{"step": 39280, "loss": 3.6244083404541017, "grad_norm": 0.7716721892356873, "learning_rate": 0.0001, "elbo": 3.62441, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3035421064977717}
{"step": 39285, "loss": 3.6393669128417967, "grad_norm": 0.6688731908798218, "learning_rate": 0.0001, "elbo": 3.63937, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.303835327234342}
{"step": 39290, "loss": 3.627402114868164, "grad_norm": 0.6599923372268677, "learning_rate": 0.0001, "elbo": 3.6274, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3041285479709126}
{"step": 39295, "loss": 3.6315940856933593, "grad_norm": 0.6591759920120239, "learning_rate": 0.0001, "elbo": 3.63159, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.304421768707483}
{"step": 39300, "loss": 3.6200786590576173, "grad_norm": 0.7001422047615051, "learning_rate": 0.0001, "elbo": 3.62008, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3047149894440535}
{"step": 39305, "loss": 3.6214805603027345, "grad_norm": 0.751551628112793, "learning_rate": 0.0001, "elbo": 3.62148, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.305008210180624}
{"step": 39310, "loss": 3.6273677825927733, "grad_norm": 0.806551992893219, "learning_rate": 0.0001, "elbo": 3.62737, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3053014309171944}
{"step": 39315, "loss": 3.6430519104003904, "grad_norm": 0.8165416121482849, "learning_rate": 0.0001, "elbo": 3.64305, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.305594651653765}
{"step": 39320, "loss": 3.634468841552734, "grad_norm": 0.8027538657188416, "learning_rate": 0.0001, "elbo": 3.63447, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3058878723903353}
{"step": 39325, "loss": 3.6376136779785155, "grad_norm": 0.7155187129974365, "learning_rate": 0.0001, "elbo": 3.63761, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.306181093126906}
{"step": 39330, "loss": 3.647154998779297, "grad_norm": 0.6886371374130249, "learning_rate": 0.0001, "elbo": 3.64715, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3064743138634762}
{"step": 39335, "loss": 3.617512512207031, "grad_norm": 0.7243740558624268, "learning_rate": 0.0001, "elbo": 3.61751, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3067675346000467}
{"step": 39340, "loss": 3.6297943115234377, "grad_norm": 0.8863019347190857, "learning_rate": 0.0001, "elbo": 3.62979, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3070607553366176}
{"step": 39345, "loss": 3.6487075805664064, "grad_norm": 0.6719314455986023, "learning_rate": 0.0001, "elbo": 3.64871, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.307353976073188}
{"step": 39350, "loss": 3.6410789489746094, "grad_norm": 0.9555439352989197, "learning_rate": 0.0001, "elbo": 3.64108, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3076471968097585}
{"step": 39355, "loss": 3.6530502319335936, "grad_norm": 0.659350574016571, "learning_rate": 0.0001, "elbo": 3.65305, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.307940417546329}
{"step": 39360, "loss": 3.618284225463867, "grad_norm": 0.6747015714645386, "learning_rate": 0.0001, "elbo": 3.61828, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3082336382828994}
{"step": 39365, "loss": 3.6394882202148438, "grad_norm": 0.675926923751831, "learning_rate": 0.0001, "elbo": 3.63949, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.30852685901947}
{"step": 39370, "loss": 3.6349514007568358, "grad_norm": 0.7034628987312317, "learning_rate": 0.0001, "elbo": 3.63495, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3088200797560403}
{"step": 39375, "loss": 3.618294525146484, "grad_norm": 0.6565158367156982, "learning_rate": 0.0001, "elbo": 3.61829, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3091133004926108}
{"step": 39380, "loss": 3.6264827728271483, "grad_norm": 0.6647475957870483, "learning_rate": 0.0001, "elbo": 3.62648, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.309406521229181}
{"step": 39385, "loss": 3.6367813110351563, "grad_norm": 0.7456197738647461, "learning_rate": 0.0001, "elbo": 3.63678, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3096997419657517}
{"step": 39390, "loss": 3.621746826171875, "grad_norm": 0.6737833023071289, "learning_rate": 0.0001, "elbo": 3.62175, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.309992962702322}
{"step": 39395, "loss": 3.644936752319336, "grad_norm": 0.6703240275382996, "learning_rate": 0.0001, "elbo": 3.64494, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.310286183438893}
{"step": 39400, "loss": 3.6364479064941406, "grad_norm": 0.6853830218315125, "learning_rate": 0.0001, "elbo": 3.63645, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3105794041754635}
{"step": 39405, "loss": 3.642504119873047, "grad_norm": 0.8025846481323242, "learning_rate": 0.0001, "elbo": 3.6425, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.310872624912034}
{"step": 39410, "loss": 3.642414855957031, "grad_norm": 0.842620849609375, "learning_rate": 0.0001, "elbo": 3.64241, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3111658456486044}
{"step": 39410, "loss": 3.642414855957031, "grad_norm": 0.842620849609375, "learning_rate": 0.0001, "elbo": 3.64241, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3111658456486044}
{"step": 39415, "loss": 3.646432113647461, "grad_norm": 0.6936414837837219, "learning_rate": 0.0001, "elbo": 3.64643, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.311459066385175}
{"step": 39420, "loss": 3.621549606323242, "grad_norm": 0.6852685809135437, "learning_rate": 0.0001, "elbo": 3.62155, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3117522871217453}
{"step": 39425, "loss": 3.643807220458984, "grad_norm": 0.712023913860321, "learning_rate": 0.0001, "elbo": 3.64381, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3120455078583158}
{"step": 39430, "loss": 3.63453369140625, "grad_norm": 0.6919813752174377, "learning_rate": 0.0001, "elbo": 3.63453, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.312338728594886}
{"step": 39435, "loss": 3.652359390258789, "grad_norm": 0.7278636693954468, "learning_rate": 0.0001, "elbo": 3.65236, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3126319493314567}
{"step": 39440, "loss": 3.6180755615234377, "grad_norm": 0.6450958847999573, "learning_rate": 0.0001, "elbo": 3.61808, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.312925170068027}
{"step": 39445, "loss": 3.6281143188476563, "grad_norm": 0.712863564491272, "learning_rate": 0.0001, "elbo": 3.62811, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3132183908045976}
{"step": 39450, "loss": 3.6406497955322266, "grad_norm": 1.0329800844192505, "learning_rate": 0.0001, "elbo": 3.64065, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.313511611541168}
{"step": 39455, "loss": 3.6541259765625, "grad_norm": 0.6893318295478821, "learning_rate": 0.0001, "elbo": 3.65413, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3138048322777385}
{"step": 39460, "loss": 3.6383705139160156, "grad_norm": 0.7153228521347046, "learning_rate": 0.0001, "elbo": 3.63837, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.314098053014309}
{"step": 39465, "loss": 3.6330718994140625, "grad_norm": 0.6965687870979309, "learning_rate": 0.0001, "elbo": 3.63307, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.31439127375088}
{"step": 39470, "loss": 3.6396743774414064, "grad_norm": 0.6536224484443665, "learning_rate": 0.0001, "elbo": 3.63967, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3146844944874503}
{"step": 39475, "loss": 3.640070343017578, "grad_norm": 0.7764460444450378, "learning_rate": 0.0001, "elbo": 3.64007, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3149777152240207}
{"step": 39480, "loss": 3.624971389770508, "grad_norm": 0.7046188116073608, "learning_rate": 0.0001, "elbo": 3.62497, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.315270935960591}
{"step": 39485, "loss": 3.6331668853759767, "grad_norm": 0.6897081136703491, "learning_rate": 0.0001, "elbo": 3.63317, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3155641566971616}
{"step": 39490, "loss": 3.631147003173828, "grad_norm": 0.6834513545036316, "learning_rate": 0.0001, "elbo": 3.63115, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.315857377433732}
{"step": 39495, "loss": 3.6381805419921873, "grad_norm": 0.6908619999885559, "learning_rate": 0.0001, "elbo": 3.63818, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3161505981703026}
{"step": 39500, "loss": 3.64078369140625, "grad_norm": 0.7003623247146606, "learning_rate": 0.0001, "elbo": 3.64078, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.316443818906873}
{"step": 39505, "loss": 3.6357627868652345, "grad_norm": 0.6952092051506042, "learning_rate": 0.0001, "elbo": 3.63576, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3167370396434435}
{"step": 39510, "loss": 3.650054168701172, "grad_norm": 0.6931474208831787, "learning_rate": 0.0001, "elbo": 3.65005, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.317030260380014}
{"step": 39515, "loss": 3.634918975830078, "grad_norm": 0.6808788180351257, "learning_rate": 0.0001, "elbo": 3.63492, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3173234811165844}
{"step": 39520, "loss": 3.6347320556640623, "grad_norm": 0.6754412055015564, "learning_rate": 0.0001, "elbo": 3.63473, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3176167018531553}
{"step": 39525, "loss": 3.6370872497558593, "grad_norm": 0.6734795570373535, "learning_rate": 0.0001, "elbo": 3.63709, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3179099225897257}
{"step": 39530, "loss": 3.63170166015625, "grad_norm": 0.7596023678779602, "learning_rate": 0.0001, "elbo": 3.6317, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.318203143326296}
{"step": 39535, "loss": 3.6434562683105467, "grad_norm": 0.6926929354667664, "learning_rate": 0.0001, "elbo": 3.64346, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3184963640628666}
{"step": 39540, "loss": 3.648712158203125, "grad_norm": 0.717940092086792, "learning_rate": 0.0001, "elbo": 3.64871, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.318789584799437}
{"step": 39545, "loss": 3.641021728515625, "grad_norm": 0.7352522015571594, "learning_rate": 0.0001, "elbo": 3.64102, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3190828055360075}
{"step": 39550, "loss": 3.626412200927734, "grad_norm": 0.7101620435714722, "learning_rate": 0.0001, "elbo": 3.62641, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.319376026272578}
{"step": 39555, "loss": 3.636463928222656, "grad_norm": 0.7231735587120056, "learning_rate": 0.0001, "elbo": 3.63646, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3196692470091484}
{"step": 39560, "loss": 3.6379558563232424, "grad_norm": 0.6640243530273438, "learning_rate": 0.0001, "elbo": 3.63796, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.319962467745719}
{"step": 39565, "loss": 3.6321556091308596, "grad_norm": 0.7273954749107361, "learning_rate": 0.0001, "elbo": 3.63216, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3202556884822894}
{"step": 39570, "loss": 3.6331504821777343, "grad_norm": 0.6963448524475098, "learning_rate": 0.0001, "elbo": 3.63315, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.32054890921886}
{"step": 39575, "loss": 3.627610778808594, "grad_norm": 0.7571191787719727, "learning_rate": 0.0001, "elbo": 3.62761, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3208421299554303}
{"step": 39580, "loss": 3.6428184509277344, "grad_norm": 0.7601151466369629, "learning_rate": 0.0001, "elbo": 3.64282, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3211353506920007}
{"step": 39585, "loss": 3.64927978515625, "grad_norm": 0.7613993287086487, "learning_rate": 0.0001, "elbo": 3.64928, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3214285714285716}
{"step": 39590, "loss": 3.648975372314453, "grad_norm": 0.7624118328094482, "learning_rate": 0.0001, "elbo": 3.64898, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.321721792165142}
{"step": 39595, "loss": 3.6373668670654298, "grad_norm": 0.9691981077194214, "learning_rate": 0.0001, "elbo": 3.63737, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3220150129017125}
{"step": 39600, "loss": 3.6439666748046875, "grad_norm": 0.7252712845802307, "learning_rate": 0.0001, "elbo": 3.64397, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.322308233638283}
{"step": 39605, "loss": 3.6340305328369142, "grad_norm": 0.7163905501365662, "learning_rate": 0.0001, "elbo": 3.63403, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3226014543748534}
{"step": 39610, "loss": 3.6412399291992186, "grad_norm": 0.7573534846305847, "learning_rate": 0.0001, "elbo": 3.64124, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.322894675111424}
{"step": 39615, "loss": 3.6330841064453123, "grad_norm": 0.6835241913795471, "learning_rate": 0.0001, "elbo": 3.63308, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3231878958479943}
{"step": 39620, "loss": 3.6340538024902345, "grad_norm": 0.6586499214172363, "learning_rate": 0.0001, "elbo": 3.63405, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.323481116584565}
{"step": 39625, "loss": 3.635003662109375, "grad_norm": 0.7068586349487305, "learning_rate": 0.0001, "elbo": 3.635, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3237743373211353}
{"step": 39630, "loss": 3.6414756774902344, "grad_norm": 0.6844520568847656, "learning_rate": 0.0001, "elbo": 3.64148, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3240675580577057}
{"step": 39635, "loss": 3.642041778564453, "grad_norm": 0.6969234943389893, "learning_rate": 0.0001, "elbo": 3.64204, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.324360778794276}
{"step": 39640, "loss": 3.633837890625, "grad_norm": 0.7564398050308228, "learning_rate": 0.0001, "elbo": 3.63384, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.324653999530847}
{"step": 39645, "loss": 3.6397159576416014, "grad_norm": 0.770185112953186, "learning_rate": 0.0001, "elbo": 3.63972, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3249472202674175}
{"step": 39650, "loss": 3.637844467163086, "grad_norm": 0.6653826236724854, "learning_rate": 0.0001, "elbo": 3.63784, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.325240441003988}
{"step": 39655, "loss": 3.6492851257324217, "grad_norm": 0.8361127972602844, "learning_rate": 0.0001, "elbo": 3.64928, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3255336617405584}
{"step": 39660, "loss": 3.646785354614258, "grad_norm": 0.8166940212249756, "learning_rate": 0.0001, "elbo": 3.64679, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.325826882477129}
{"step": 39665, "loss": 3.643147277832031, "grad_norm": 0.891666829586029, "learning_rate": 0.0001, "elbo": 3.64315, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3261201032136993}
{"step": 39670, "loss": 3.630846405029297, "grad_norm": 0.7340095043182373, "learning_rate": 0.0001, "elbo": 3.63085, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.32641332395027}
{"step": 39675, "loss": 3.6498641967773438, "grad_norm": 0.6979914903640747, "learning_rate": 0.0001, "elbo": 3.64986, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3267065446868402}
{"step": 39680, "loss": 3.645524597167969, "grad_norm": 0.6912139058113098, "learning_rate": 0.0001, "elbo": 3.64552, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3269997654234107}
{"step": 39685, "loss": 3.632917022705078, "grad_norm": 0.6700429320335388, "learning_rate": 0.0001, "elbo": 3.63292, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.327292986159981}
{"step": 39690, "loss": 3.6292259216308596, "grad_norm": 1.0119916200637817, "learning_rate": 0.0001, "elbo": 3.62923, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3275862068965516}
{"step": 39695, "loss": 3.6305503845214844, "grad_norm": 0.7585326433181763, "learning_rate": 0.0001, "elbo": 3.63055, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.327879427633122}
{"step": 39700, "loss": 3.6515682220458983, "grad_norm": 1.1353520154953003, "learning_rate": 0.0001, "elbo": 3.65157, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3281726483696925}
{"step": 39705, "loss": 3.6272552490234373, "grad_norm": 0.6964601278305054, "learning_rate": 0.0001, "elbo": 3.62726, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.328465869106263}
{"step": 39710, "loss": 3.6180187225341798, "grad_norm": 0.6857202053070068, "learning_rate": 0.0001, "elbo": 3.61802, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.328759089842834}
{"step": 39715, "loss": 3.6409835815429688, "grad_norm": 0.7224999666213989, "learning_rate": 0.0001, "elbo": 3.64098, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3290523105794043}
{"step": 39720, "loss": 3.6337718963623047, "grad_norm": 0.7807427644729614, "learning_rate": 0.0001, "elbo": 3.63377, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3293455313159748}
{"step": 39725, "loss": 3.629902648925781, "grad_norm": 0.6568261981010437, "learning_rate": 0.0001, "elbo": 3.6299, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3296387520525452}
{"step": 39730, "loss": 3.610106658935547, "grad_norm": 0.6356906890869141, "learning_rate": 0.0001, "elbo": 3.61011, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3299319727891157}
{"step": 39735, "loss": 3.628099822998047, "grad_norm": 0.7151744365692139, "learning_rate": 0.0001, "elbo": 3.6281, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.330225193525686}
{"step": 39740, "loss": 3.6304649353027343, "grad_norm": 0.6622821092605591, "learning_rate": 0.0001, "elbo": 3.63046, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3305184142622566}
{"step": 39745, "loss": 3.629689025878906, "grad_norm": 0.6890275478363037, "learning_rate": 0.0001, "elbo": 3.62969, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.330811634998827}
{"step": 39750, "loss": 3.6354904174804688, "grad_norm": 0.6607742309570312, "learning_rate": 0.0001, "elbo": 3.63549, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3311048557353975}
{"step": 39750, "loss": 3.6354904174804688, "grad_norm": 0.6607742309570312, "learning_rate": 0.0001, "elbo": 3.63549, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3311048557353975}
{"step": 39755, "loss": 3.6392753601074217, "grad_norm": 0.7282476425170898, "learning_rate": 0.0001, "elbo": 3.63928, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.331398076471968}
{"step": 39760, "loss": 3.6256595611572267, "grad_norm": 0.6948894262313843, "learning_rate": 0.0001, "elbo": 3.62566, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3316912972085384}
{"step": 39765, "loss": 3.625202178955078, "grad_norm": 0.6659373641014099, "learning_rate": 0.0001, "elbo": 3.6252, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3319845179451093}
{"step": 39770, "loss": 3.6451446533203127, "grad_norm": 0.7152413129806519, "learning_rate": 0.0001, "elbo": 3.64514, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3322777386816798}
{"step": 39775, "loss": 3.6267223358154297, "grad_norm": 0.7621399164199829, "learning_rate": 0.0001, "elbo": 3.62672, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.33257095941825}
{"step": 39780, "loss": 3.6435333251953126, "grad_norm": 0.699026882648468, "learning_rate": 0.0001, "elbo": 3.64353, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3328641801548207}
{"step": 39785, "loss": 3.6298442840576173, "grad_norm": 0.7864303588867188, "learning_rate": 0.0001, "elbo": 3.62984, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.333157400891391}
{"step": 39790, "loss": 3.6321762084960936, "grad_norm": 0.6828659176826477, "learning_rate": 0.0001, "elbo": 3.63218, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3334506216279616}
{"step": 39795, "loss": 3.641084671020508, "grad_norm": 0.7173222899436951, "learning_rate": 0.0001, "elbo": 3.64108, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.333743842364532}
{"step": 39800, "loss": 3.626250076293945, "grad_norm": 0.7381888031959534, "learning_rate": 0.0001, "elbo": 3.62625, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3340370631011025}
{"step": 39805, "loss": 3.6323619842529298, "grad_norm": 0.6575166583061218, "learning_rate": 0.0001, "elbo": 3.63236, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.334330283837673}
{"step": 39810, "loss": 3.650613784790039, "grad_norm": 0.6936313509941101, "learning_rate": 0.0001, "elbo": 3.65061, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3346235045742434}
{"step": 39815, "loss": 3.646769714355469, "grad_norm": 0.6479740738868713, "learning_rate": 0.0001, "elbo": 3.64677, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.334916725310814}
{"step": 39820, "loss": 3.623320770263672, "grad_norm": 0.663979709148407, "learning_rate": 0.0001, "elbo": 3.62332, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3352099460473843}
{"step": 39825, "loss": 3.643941116333008, "grad_norm": 0.66708904504776, "learning_rate": 0.0001, "elbo": 3.64394, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3355031667839548}
{"step": 39830, "loss": 3.624981689453125, "grad_norm": 0.6932842135429382, "learning_rate": 0.0001, "elbo": 3.62498, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.335796387520525}
{"step": 39830, "loss": 3.624981689453125, "grad_norm": 0.6932842135429382, "learning_rate": 0.0001, "elbo": 3.62498, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.335796387520525}
{"step": 39835, "loss": 3.633184051513672, "grad_norm": 0.7208045721054077, "learning_rate": 0.0001, "elbo": 3.63318, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.336089608257096}
{"step": 39840, "loss": 3.6351951599121093, "grad_norm": 0.7142987251281738, "learning_rate": 0.0001, "elbo": 3.6352, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3363828289936666}
{"step": 39845, "loss": 3.6514514923095702, "grad_norm": 0.7226829528808594, "learning_rate": 0.0001, "elbo": 3.65145, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.336676049730237}
{"step": 39850, "loss": 3.626180648803711, "grad_norm": 0.8062322735786438, "learning_rate": 0.0001, "elbo": 3.62618, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3369692704668075}
{"step": 39855, "loss": 3.6335613250732424, "grad_norm": 0.7430508136749268, "learning_rate": 0.0001, "elbo": 3.63356, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.337262491203378}
{"step": 39860, "loss": 3.6470497131347654, "grad_norm": 0.6990044713020325, "learning_rate": 0.0001, "elbo": 3.64705, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3375557119399484}
{"step": 39865, "loss": 3.634607696533203, "grad_norm": 0.7046490907669067, "learning_rate": 0.0001, "elbo": 3.63461, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.337848932676519}
{"step": 39870, "loss": 3.637246322631836, "grad_norm": 0.7084149122238159, "learning_rate": 0.0001, "elbo": 3.63725, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3381421534130893}
{"step": 39875, "loss": 3.6350364685058594, "grad_norm": 0.6985307931900024, "learning_rate": 0.0001, "elbo": 3.63504, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3384353741496597}
{"step": 39880, "loss": 3.6464027404785155, "grad_norm": 0.7237865924835205, "learning_rate": 0.0001, "elbo": 3.6464, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.33872859488623}
{"step": 39885, "loss": 3.6567367553710937, "grad_norm": 0.6875254511833191, "learning_rate": 0.0001, "elbo": 3.65674, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.339021815622801}
{"step": 39890, "loss": 3.647417449951172, "grad_norm": 0.6840646266937256, "learning_rate": 0.0001, "elbo": 3.64742, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3393150363593715}
{"step": 39895, "loss": 3.638447570800781, "grad_norm": 0.6959375739097595, "learning_rate": 0.0001, "elbo": 3.63845, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.339608257095942}
{"step": 39900, "loss": 3.631905364990234, "grad_norm": 0.675434947013855, "learning_rate": 0.0001, "elbo": 3.63191, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3399014778325125}
{"step": 39905, "loss": 3.6318058013916015, "grad_norm": 0.6954779028892517, "learning_rate": 0.0001, "elbo": 3.63181, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.340194698569083}
{"step": 39910, "loss": 3.633229064941406, "grad_norm": 0.6844778060913086, "learning_rate": 0.0001, "elbo": 3.63323, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3404879193056534}
{"step": 39915, "loss": 3.636182403564453, "grad_norm": 1.0722789764404297, "learning_rate": 0.0001, "elbo": 3.63618, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.340781140042224}
{"step": 39920, "loss": 3.633466339111328, "grad_norm": 0.6460221409797668, "learning_rate": 0.0001, "elbo": 3.63347, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3410743607787943}
{"step": 39925, "loss": 3.640228271484375, "grad_norm": 0.7059808969497681, "learning_rate": 0.0001, "elbo": 3.64023, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3413675815153647}
{"step": 39930, "loss": 3.623930740356445, "grad_norm": 0.7191687822341919, "learning_rate": 0.0001, "elbo": 3.62393, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.341660802251935}
{"step": 39935, "loss": 3.628134918212891, "grad_norm": 0.6744785308837891, "learning_rate": 0.0001, "elbo": 3.62813, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3419540229885056}
{"step": 39940, "loss": 3.640765380859375, "grad_norm": 1.031476378440857, "learning_rate": 0.0001, "elbo": 3.64077, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.342247243725076}
{"step": 39945, "loss": 3.6284805297851563, "grad_norm": 0.7984800934791565, "learning_rate": 0.0001, "elbo": 3.62848, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3425404644616465}
{"step": 39950, "loss": 3.625071716308594, "grad_norm": 0.7169412970542908, "learning_rate": 0.0001, "elbo": 3.62507, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.342833685198217}
{"step": 39955, "loss": 3.637042236328125, "grad_norm": 0.7619525194168091, "learning_rate": 0.0001, "elbo": 3.63704, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.343126905934788}
{"step": 39955, "loss": 3.637042236328125, "grad_norm": 0.7619525194168091, "learning_rate": 0.0001, "elbo": 3.63704, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.343126905934788}
{"step": 39960, "loss": 3.624125671386719, "grad_norm": 0.6662412285804749, "learning_rate": 0.0001, "elbo": 3.62413, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3434201266713584}
{"step": 39965, "loss": 3.628766632080078, "grad_norm": 0.6560746431350708, "learning_rate": 0.0001, "elbo": 3.62877, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.343713347407929}
{"step": 39970, "loss": 3.6333160400390625, "grad_norm": 0.6483631134033203, "learning_rate": 0.0001, "elbo": 3.63332, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3440065681444993}
{"step": 39975, "loss": 3.633250427246094, "grad_norm": 0.667319655418396, "learning_rate": 0.0001, "elbo": 3.63325, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3442997888810697}
{"step": 39980, "loss": 3.6412105560302734, "grad_norm": 0.6603891253471375, "learning_rate": 0.0001, "elbo": 3.64121, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.34459300961764}
{"step": 39985, "loss": 3.649822998046875, "grad_norm": 0.6927773356437683, "learning_rate": 0.0001, "elbo": 3.64982, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3448862303542106}
{"step": 39990, "loss": 3.632102203369141, "grad_norm": 0.74070143699646, "learning_rate": 0.0001, "elbo": 3.6321, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.345179451090781}
{"step": 39995, "loss": 3.617470932006836, "grad_norm": 0.8552504777908325, "learning_rate": 0.0001, "elbo": 3.61747, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3454726718273515}
{"step": 40000, "loss": 3.649018096923828, "grad_norm": 0.7246752381324768, "learning_rate": 0.0001, "elbo": 3.64902, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.345765892563922}
{"step": 40005, "loss": 3.632244873046875, "grad_norm": 0.7029978036880493, "learning_rate": 0.0001, "elbo": 3.63224, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3460591133004924}
{"step": 40010, "loss": 3.6291053771972654, "grad_norm": 0.6663131713867188, "learning_rate": 0.0001, "elbo": 3.62911, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3463523340370633}
{"step": 40015, "loss": 3.635173797607422, "grad_norm": 0.693812370300293, "learning_rate": 0.0001, "elbo": 3.63517, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.346645554773634}
{"step": 40020, "loss": 3.6224510192871096, "grad_norm": 0.7002763748168945, "learning_rate": 0.0001, "elbo": 3.62245, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3469387755102042}
{"step": 40025, "loss": 3.6347122192382812, "grad_norm": 0.7202595472335815, "learning_rate": 0.0001, "elbo": 3.63471, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3472319962467747}
{"step": 40030, "loss": 3.6323280334472656, "grad_norm": 0.6632527709007263, "learning_rate": 0.0001, "elbo": 3.63233, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.347525216983345}
{"step": 40035, "loss": 3.643670654296875, "grad_norm": 0.7335888147354126, "learning_rate": 0.0001, "elbo": 3.64367, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3478184377199156}
{"step": 40040, "loss": 3.6145584106445314, "grad_norm": 0.7428646683692932, "learning_rate": 0.0001, "elbo": 3.61456, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.348111658456486}
{"step": 40045, "loss": 3.6507965087890626, "grad_norm": 0.7422505021095276, "learning_rate": 0.0001, "elbo": 3.6508, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3484048791930565}
{"step": 40050, "loss": 3.6156875610351564, "grad_norm": 0.6510241031646729, "learning_rate": 0.0001, "elbo": 3.61569, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.348698099929627}
{"step": 40055, "loss": 3.6205490112304686, "grad_norm": 0.7102658152580261, "learning_rate": 0.0001, "elbo": 3.62055, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3489913206661974}
{"step": 40060, "loss": 3.6313648223876953, "grad_norm": 0.6701914668083191, "learning_rate": 0.0001, "elbo": 3.63136, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.349284541402768}
{"step": 40065, "loss": 3.6155364990234373, "grad_norm": 0.6872149705886841, "learning_rate": 0.0001, "elbo": 3.61554, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3495777621393383}
{"step": 40070, "loss": 3.6431915283203127, "grad_norm": 0.842875599861145, "learning_rate": 0.0001, "elbo": 3.64319, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.349870982875909}
{"step": 40075, "loss": 3.6412193298339846, "grad_norm": 0.83118736743927, "learning_rate": 0.0001, "elbo": 3.64122, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3501642036124792}
{"step": 40080, "loss": 3.6471878051757813, "grad_norm": 0.681274950504303, "learning_rate": 0.0001, "elbo": 3.64719, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.35045742434905}
{"step": 40085, "loss": 3.634718322753906, "grad_norm": 0.7338393926620483, "learning_rate": 0.0001, "elbo": 3.63472, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3507506450856206}
{"step": 40090, "loss": 3.6336139678955077, "grad_norm": 0.7176348567008972, "learning_rate": 0.0001, "elbo": 3.63361, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.351043865822191}
{"step": 40095, "loss": 3.629973602294922, "grad_norm": 0.6989772915840149, "learning_rate": 0.0001, "elbo": 3.62997, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3513370865587615}
{"step": 40100, "loss": 3.6449459075927733, "grad_norm": 0.7815394401550293, "learning_rate": 0.0001, "elbo": 3.64495, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.351630307295332}
{"step": 40105, "loss": 3.6221382141113283, "grad_norm": 0.7924343347549438, "learning_rate": 0.0001, "elbo": 3.62214, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3519235280319024}
{"step": 40110, "loss": 3.636974334716797, "grad_norm": 0.833709716796875, "learning_rate": 0.0001, "elbo": 3.63697, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.352216748768473}
{"step": 40115, "loss": 3.6471534729003907, "grad_norm": 0.669739842414856, "learning_rate": 0.0001, "elbo": 3.64715, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3525099695050433}
{"step": 40120, "loss": 3.6303855895996096, "grad_norm": 0.6263625025749207, "learning_rate": 0.0001, "elbo": 3.63039, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3528031902416138}
{"step": 40125, "loss": 3.629252624511719, "grad_norm": 0.743323802947998, "learning_rate": 0.0001, "elbo": 3.62925, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3530964109781842}
{"step": 40130, "loss": 3.627389907836914, "grad_norm": 0.6617125272750854, "learning_rate": 0.0001, "elbo": 3.62739, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3533896317147547}
{"step": 40135, "loss": 3.617889404296875, "grad_norm": 0.8936969041824341, "learning_rate": 0.0001, "elbo": 3.61789, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3536828524513256}
{"step": 40140, "loss": 3.6337661743164062, "grad_norm": 0.6974944472312927, "learning_rate": 0.0001, "elbo": 3.63377, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.353976073187896}
{"step": 40145, "loss": 3.620819091796875, "grad_norm": 0.6639214158058167, "learning_rate": 0.0001, "elbo": 3.62082, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3542692939244665}
{"step": 40150, "loss": 3.6219688415527345, "grad_norm": 0.7320469617843628, "learning_rate": 0.0001, "elbo": 3.62197, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.354562514661037}
{"step": 40155, "loss": 3.6248374938964845, "grad_norm": 0.7138955593109131, "learning_rate": 0.0001, "elbo": 3.62484, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3548557353976074}
{"step": 40160, "loss": 3.6511207580566407, "grad_norm": 0.743700385093689, "learning_rate": 0.0001, "elbo": 3.65112, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.355148956134178}
{"step": 40165, "loss": 3.635784912109375, "grad_norm": 0.7079133987426758, "learning_rate": 0.0001, "elbo": 3.63579, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3554421768707483}
{"step": 40170, "loss": 3.642289733886719, "grad_norm": 0.6665183305740356, "learning_rate": 0.0001, "elbo": 3.64229, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3557353976073188}
{"step": 40175, "loss": 3.6259140014648437, "grad_norm": 0.6729800701141357, "learning_rate": 0.0001, "elbo": 3.62591, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.356028618343889}
{"step": 40175, "loss": 3.6259140014648437, "grad_norm": 0.6729800701141357, "learning_rate": 0.0001, "elbo": 3.62591, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.356028618343889}
{"step": 40180, "loss": 3.6323997497558596, "grad_norm": 0.6928234696388245, "learning_rate": 0.0001, "elbo": 3.6324, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3563218390804597}
{"step": 40185, "loss": 3.6213455200195312, "grad_norm": 0.7443490028381348, "learning_rate": 0.0001, "elbo": 3.62135, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.35661505981703}
{"step": 40190, "loss": 3.6280357360839846, "grad_norm": 0.6925142407417297, "learning_rate": 0.0001, "elbo": 3.62804, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3569082805536006}
{"step": 40195, "loss": 3.615656280517578, "grad_norm": 0.6791322827339172, "learning_rate": 0.0001, "elbo": 3.61566, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.357201501290171}
{"step": 40200, "loss": 3.629832458496094, "grad_norm": 0.7296241521835327, "learning_rate": 0.0001, "elbo": 3.62983, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3574947220267415}
{"step": 40205, "loss": 3.619293212890625, "grad_norm": 0.662411630153656, "learning_rate": 0.0001, "elbo": 3.61929, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3577879427633124}
{"step": 40210, "loss": 3.620677947998047, "grad_norm": 0.667287290096283, "learning_rate": 0.0001, "elbo": 3.62068, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.358081163499883}
{"step": 40215, "loss": 3.646645736694336, "grad_norm": 0.690320611000061, "learning_rate": 0.0001, "elbo": 3.64665, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3583743842364533}
{"step": 40220, "loss": 3.6266510009765627, "grad_norm": 0.6206033825874329, "learning_rate": 0.0001, "elbo": 3.62665, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3586676049730237}
{"step": 40220, "loss": 3.6266510009765627, "grad_norm": 0.6206033825874329, "learning_rate": 0.0001, "elbo": 3.62665, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3586676049730237}
{"step": 40225, "loss": 3.623249053955078, "grad_norm": 0.66035395860672, "learning_rate": 0.0001, "elbo": 3.62325, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.358960825709594}
{"step": 40230, "loss": 3.6251277923583984, "grad_norm": 0.7053821086883545, "learning_rate": 0.0001, "elbo": 3.62513, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3592540464461647}
{"step": 40235, "loss": 3.629924774169922, "grad_norm": 0.6701173186302185, "learning_rate": 0.0001, "elbo": 3.62992, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.359547267182735}
{"step": 40240, "loss": 3.624235916137695, "grad_norm": 0.7252713441848755, "learning_rate": 0.0001, "elbo": 3.62424, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3598404879193056}
{"step": 40245, "loss": 3.642176055908203, "grad_norm": 0.7066728472709656, "learning_rate": 0.0001, "elbo": 3.64218, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.360133708655876}
{"step": 40250, "loss": 3.640614318847656, "grad_norm": 0.6922311782836914, "learning_rate": 0.0001, "elbo": 3.64061, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3604269293924465}
{"step": 40255, "loss": 3.6226181030273437, "grad_norm": 0.6375051736831665, "learning_rate": 0.0001, "elbo": 3.62262, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3607201501290174}
{"step": 40260, "loss": 3.6158958435058595, "grad_norm": 0.6704348921775818, "learning_rate": 0.0001, "elbo": 3.6159, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.361013370865588}
{"step": 40265, "loss": 3.6178009033203127, "grad_norm": 0.7114441394805908, "learning_rate": 0.0001, "elbo": 3.6178, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3613065916021583}
{"step": 40270, "loss": 3.6388790130615236, "grad_norm": 0.6788265109062195, "learning_rate": 0.0001, "elbo": 3.63888, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3615998123387287}
{"step": 40275, "loss": 3.642721939086914, "grad_norm": 0.7067190408706665, "learning_rate": 0.0001, "elbo": 3.64272, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.361893033075299}
{"step": 40280, "loss": 3.623036575317383, "grad_norm": 0.7116571664810181, "learning_rate": 0.0001, "elbo": 3.62304, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3621862538118696}
{"step": 40285, "loss": 3.63209228515625, "grad_norm": 1.0135159492492676, "learning_rate": 0.0001, "elbo": 3.63209, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.36247947454844}
{"step": 40290, "loss": 3.6528594970703123, "grad_norm": 0.6704163551330566, "learning_rate": 0.0001, "elbo": 3.65286, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3627726952850105}
{"step": 40295, "loss": 3.62529296875, "grad_norm": 0.7390742301940918, "learning_rate": 0.0001, "elbo": 3.62529, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.363065916021581}
{"step": 40295, "loss": 3.62529296875, "grad_norm": 0.7390742301940918, "learning_rate": 0.0001, "elbo": 3.62529, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.363065916021581}
{"step": 40300, "loss": 3.6379520416259767, "grad_norm": 0.7013715505599976, "learning_rate": 0.0001, "elbo": 3.63795, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3633591367581515}
{"step": 40305, "loss": 3.628571701049805, "grad_norm": 0.6954867243766785, "learning_rate": 0.0001, "elbo": 3.62857, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.363652357494722}
{"step": 40310, "loss": 3.623351287841797, "grad_norm": 0.703147292137146, "learning_rate": 0.0001, "elbo": 3.62335, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3639455782312924}
{"step": 40315, "loss": 3.6243019104003906, "grad_norm": 0.6920633316040039, "learning_rate": 0.0001, "elbo": 3.6243, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.364238798967863}
{"step": 40320, "loss": 3.636035919189453, "grad_norm": 0.760471522808075, "learning_rate": 0.0001, "elbo": 3.63604, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3645320197044333}
{"step": 40325, "loss": 3.635450744628906, "grad_norm": 0.7090578675270081, "learning_rate": 0.0001, "elbo": 3.63545, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.364825240441004}
{"step": 40330, "loss": 3.6194305419921875, "grad_norm": 0.7995516657829285, "learning_rate": 0.0001, "elbo": 3.61943, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3651184611775746}
{"step": 40335, "loss": 3.643296813964844, "grad_norm": 0.7526041865348816, "learning_rate": 0.0001, "elbo": 3.6433, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.365411681914145}
{"step": 40340, "loss": 3.6363857269287108, "grad_norm": 0.6725625395774841, "learning_rate": 0.0001, "elbo": 3.63639, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3657049026507155}
{"step": 40345, "loss": 3.6341739654541017, "grad_norm": 0.6563227772712708, "learning_rate": 0.0001, "elbo": 3.63417, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.365998123387286}
{"step": 40350, "loss": 3.611023712158203, "grad_norm": 0.6656145453453064, "learning_rate": 0.0001, "elbo": 3.61102, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3662913441238564}
{"step": 40355, "loss": 3.633839797973633, "grad_norm": 0.716920793056488, "learning_rate": 0.0001, "elbo": 3.63384, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.366584564860427}
{"step": 40360, "loss": 3.6386489868164062, "grad_norm": 0.6705524325370789, "learning_rate": 0.0001, "elbo": 3.63865, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3668777855969974}
{"step": 40365, "loss": 3.6363079071044924, "grad_norm": 0.7048747539520264, "learning_rate": 0.0001, "elbo": 3.63631, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.367171006333568}
{"step": 40370, "loss": 3.653010940551758, "grad_norm": 0.7204186320304871, "learning_rate": 0.0001, "elbo": 3.65301, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3674642270701383}
{"step": 40375, "loss": 3.6432151794433594, "grad_norm": 0.6892821788787842, "learning_rate": 0.0001, "elbo": 3.64322, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3677574478067087}
{"step": 40380, "loss": 3.612017822265625, "grad_norm": 0.6818006634712219, "learning_rate": 0.0001, "elbo": 3.61202, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3680506685432796}
{"step": 40385, "loss": 3.623444366455078, "grad_norm": 0.7224134802818298, "learning_rate": 0.0001, "elbo": 3.62344, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.36834388927985}
{"step": 40390, "loss": 3.6251869201660156, "grad_norm": 0.680871307849884, "learning_rate": 0.0001, "elbo": 3.62519, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3686371100164205}
{"step": 40395, "loss": 3.6232051849365234, "grad_norm": 0.6632728576660156, "learning_rate": 0.0001, "elbo": 3.62321, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.368930330752991}
{"step": 40400, "loss": 3.6230495452880858, "grad_norm": 0.6583216190338135, "learning_rate": 0.0001, "elbo": 3.62305, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3692235514895614}
{"step": 40405, "loss": 3.6237510681152343, "grad_norm": 0.664534330368042, "learning_rate": 0.0001, "elbo": 3.62375, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.369516772226132}
{"step": 40410, "loss": 3.647906494140625, "grad_norm": 0.7150294780731201, "learning_rate": 0.0001, "elbo": 3.64791, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3698099929627023}
{"step": 40415, "loss": 3.626033401489258, "grad_norm": 0.8007879257202148, "learning_rate": 0.0001, "elbo": 3.62603, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.370103213699273}
{"step": 40420, "loss": 3.6369190216064453, "grad_norm": 0.6772209405899048, "learning_rate": 0.0001, "elbo": 3.63692, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3703964344358432}
{"step": 40425, "loss": 3.632490539550781, "grad_norm": 0.7726386189460754, "learning_rate": 0.0001, "elbo": 3.63249, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3706896551724137}
{"step": 40430, "loss": 3.6324249267578126, "grad_norm": 0.9854562878608704, "learning_rate": 0.0001, "elbo": 3.63242, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.370982875908984}
{"step": 40430, "loss": 3.6324249267578126, "grad_norm": 0.9854562878608704, "learning_rate": 0.0001, "elbo": 3.63242, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.370982875908984}
{"step": 40435, "loss": 3.6396549224853514, "grad_norm": 0.7279393076896667, "learning_rate": 0.0001, "elbo": 3.63965, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3712760966455546}
{"step": 40440, "loss": 3.6176963806152345, "grad_norm": 0.7212347984313965, "learning_rate": 0.0001, "elbo": 3.6177, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.371569317382125}
{"step": 40440, "loss": 3.6176963806152345, "grad_norm": 0.7212347984313965, "learning_rate": 0.0001, "elbo": 3.6177, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.371569317382125}
{"step": 40445, "loss": 3.626053237915039, "grad_norm": 0.6581929326057434, "learning_rate": 0.0001, "elbo": 3.62605, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3718625381186955}
{"step": 40450, "loss": 3.6367202758789063, "grad_norm": 0.666285514831543, "learning_rate": 0.0001, "elbo": 3.63672, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3721557588552664}
{"step": 40455, "loss": 3.6292118072509765, "grad_norm": 0.657407820224762, "learning_rate": 0.0001, "elbo": 3.62921, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.372448979591837}
{"step": 40460, "loss": 3.652996826171875, "grad_norm": 0.7398113012313843, "learning_rate": 0.0001, "elbo": 3.653, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3727422003284073}
{"step": 40465, "loss": 3.626595687866211, "grad_norm": 0.7086600661277771, "learning_rate": 0.0001, "elbo": 3.6266, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.373035421064978}
{"step": 40470, "loss": 3.638062286376953, "grad_norm": 0.6982368230819702, "learning_rate": 0.0001, "elbo": 3.63806, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3733286418015482}
{"step": 40475, "loss": 3.6187820434570312, "grad_norm": 0.7340183854103088, "learning_rate": 0.0001, "elbo": 3.61878, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3736218625381187}
{"step": 40480, "loss": 3.6294342041015626, "grad_norm": 0.7286990880966187, "learning_rate": 0.0001, "elbo": 3.62943, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.373915083274689}
{"step": 40485, "loss": 3.6342704772949217, "grad_norm": 0.7120077610015869, "learning_rate": 0.0001, "elbo": 3.63427, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3742083040112596}
{"step": 40490, "loss": 3.6219036102294924, "grad_norm": 0.7399321794509888, "learning_rate": 0.0001, "elbo": 3.6219, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.37450152474783}
{"step": 40495, "loss": 3.634015655517578, "grad_norm": 0.7035177946090698, "learning_rate": 0.0001, "elbo": 3.63402, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3747947454844005}
{"step": 40500, "loss": 3.633961868286133, "grad_norm": 0.6800934076309204, "learning_rate": 0.0001, "elbo": 3.63396, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.375087966220971}
{"step": 40505, "loss": 3.628571701049805, "grad_norm": 0.7995970845222473, "learning_rate": 0.0001, "elbo": 3.62857, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.375381186957542}
{"step": 40505, "loss": 3.628571701049805, "grad_norm": 0.7995970845222473, "learning_rate": 0.0001, "elbo": 3.62857, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.375381186957542}
{"step": 40510, "loss": 3.6339271545410154, "grad_norm": 0.7855843901634216, "learning_rate": 0.0001, "elbo": 3.63393, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3756744076941123}
{"step": 40515, "loss": 3.641238784790039, "grad_norm": 0.9711165428161621, "learning_rate": 0.0001, "elbo": 3.64124, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3759676284306828}
{"step": 40520, "loss": 3.6419387817382813, "grad_norm": 0.724013090133667, "learning_rate": 0.0001, "elbo": 3.64194, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.376260849167253}
{"step": 40525, "loss": 3.6355621337890627, "grad_norm": 0.6793277859687805, "learning_rate": 0.0001, "elbo": 3.63556, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3765540699038237}
{"step": 40530, "loss": 3.63375244140625, "grad_norm": 0.7061428427696228, "learning_rate": 0.0001, "elbo": 3.63375, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.376847290640394}
{"step": 40535, "loss": 3.616240692138672, "grad_norm": 0.7222282290458679, "learning_rate": 0.0001, "elbo": 3.61624, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3771405113769646}
{"step": 40540, "loss": 3.6244743347167967, "grad_norm": 0.6779678463935852, "learning_rate": 0.0001, "elbo": 3.62447, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.377433732113535}
{"step": 40545, "loss": 3.617443084716797, "grad_norm": 0.8044537901878357, "learning_rate": 0.0001, "elbo": 3.61744, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3777269528501055}
{"step": 40550, "loss": 3.6339580535888674, "grad_norm": 0.6544830799102783, "learning_rate": 0.0001, "elbo": 3.63396, "mask_frac": 0.5013, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.378020173586676}
{"step": 40555, "loss": 3.6306293487548826, "grad_norm": 0.6828769445419312, "learning_rate": 0.0001, "elbo": 3.63063, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3783133943232464}
{"step": 40560, "loss": 3.6269031524658204, "grad_norm": 0.7174549698829651, "learning_rate": 0.0001, "elbo": 3.6269, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.378606615059817}
{"step": 40565, "loss": 3.629780960083008, "grad_norm": 0.7417218089103699, "learning_rate": 0.0001, "elbo": 3.62978, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3788998357963873}
{"step": 40570, "loss": 3.633417510986328, "grad_norm": 0.7682441473007202, "learning_rate": 0.0001, "elbo": 3.63342, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3791930565329578}
{"step": 40575, "loss": 3.6252132415771485, "grad_norm": 0.6853464245796204, "learning_rate": 0.0001, "elbo": 3.62521, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3794862772695287}
{"step": 40580, "loss": 3.642395782470703, "grad_norm": 0.7496591210365295, "learning_rate": 0.0001, "elbo": 3.6424, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.379779498006099}
{"step": 40585, "loss": 3.617496109008789, "grad_norm": 0.6549032330513, "learning_rate": 0.0001, "elbo": 3.6175, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3800727187426696}
{"step": 40590, "loss": 3.6288658142089845, "grad_norm": 0.6530946493148804, "learning_rate": 0.0001, "elbo": 3.62887, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.38036593947924}
{"step": 40595, "loss": 3.62545166015625, "grad_norm": 0.6872393488883972, "learning_rate": 0.0001, "elbo": 3.62545, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3806591602158105}
{"step": 40600, "loss": 3.621541213989258, "grad_norm": 0.6881400346755981, "learning_rate": 0.0001, "elbo": 3.62154, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.380952380952381}
{"step": 40605, "loss": 3.632432556152344, "grad_norm": 0.7782930731773376, "learning_rate": 0.0001, "elbo": 3.63243, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3812456016889514}
{"step": 40610, "loss": 3.6217971801757813, "grad_norm": 0.9823417663574219, "learning_rate": 0.0001, "elbo": 3.6218, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.381538822425522}
{"step": 40615, "loss": 3.6196266174316407, "grad_norm": 0.6330044865608215, "learning_rate": 0.0001, "elbo": 3.61963, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3818320431620923}
{"step": 40620, "loss": 3.6449665069580077, "grad_norm": 1.3717416524887085, "learning_rate": 0.0001, "elbo": 3.64497, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3821252638986627}
{"step": 40620, "loss": 3.6449665069580077, "grad_norm": 1.3717416524887085, "learning_rate": 0.0001, "elbo": 3.64497, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3821252638986627}
{"step": 40625, "loss": 3.6332443237304686, "grad_norm": 0.7417389750480652, "learning_rate": 0.0001, "elbo": 3.63324, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3824184846352336}
{"step": 40630, "loss": 3.633519744873047, "grad_norm": 0.8723412752151489, "learning_rate": 0.0001, "elbo": 3.63352, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.382711705371804}
{"step": 40635, "loss": 3.6189117431640625, "grad_norm": 0.7712802886962891, "learning_rate": 0.0001, "elbo": 3.61891, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3830049261083746}
{"step": 40640, "loss": 3.6217628479003907, "grad_norm": 0.7206634879112244, "learning_rate": 0.0001, "elbo": 3.62176, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.383298146844945}
{"step": 40645, "loss": 3.6380569458007814, "grad_norm": 0.773391604423523, "learning_rate": 0.0001, "elbo": 3.63806, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3835913675815155}
{"step": 40650, "loss": 3.6307914733886717, "grad_norm": 0.6646013259887695, "learning_rate": 0.0001, "elbo": 3.63079, "mask_frac": 0.5013, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.383884588318086}
{"step": 40655, "loss": 3.6227035522460938, "grad_norm": 0.64443039894104, "learning_rate": 0.0001, "elbo": 3.6227, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3841778090546564}
{"step": 40660, "loss": 3.6144542694091797, "grad_norm": 0.649175226688385, "learning_rate": 0.0001, "elbo": 3.61445, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.384471029791227}
{"step": 40665, "loss": 3.6251708984375, "grad_norm": 0.6563376188278198, "learning_rate": 0.0001, "elbo": 3.62517, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3847642505277973}
{"step": 40665, "loss": 3.6251708984375, "grad_norm": 0.6563376188278198, "learning_rate": 0.0001, "elbo": 3.62517, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3847642505277973}
{"step": 40670, "loss": 3.626282501220703, "grad_norm": 0.72288978099823, "learning_rate": 0.0001, "elbo": 3.62628, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3850574712643677}
{"step": 40675, "loss": 3.6338272094726562, "grad_norm": 0.7601113319396973, "learning_rate": 0.0001, "elbo": 3.63383, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.385350692000938}
{"step": 40680, "loss": 3.627981185913086, "grad_norm": 0.6437958478927612, "learning_rate": 0.0001, "elbo": 3.62798, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3856439127375086}
{"step": 40680, "loss": 3.627981185913086, "grad_norm": 0.6437958478927612, "learning_rate": 0.0001, "elbo": 3.62798, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3856439127375086}
{"step": 40685, "loss": 3.630941390991211, "grad_norm": 0.6464532017707825, "learning_rate": 0.0001, "elbo": 3.63094, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.385937133474079}
{"step": 40690, "loss": 3.6298770904541016, "grad_norm": 0.6581670641899109, "learning_rate": 0.0001, "elbo": 3.62988, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3862303542106496}
{"step": 40695, "loss": 3.618453598022461, "grad_norm": 0.7440532445907593, "learning_rate": 0.0001, "elbo": 3.61845, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3865235749472204}
{"step": 40700, "loss": 3.6265579223632813, "grad_norm": 0.8622713088989258, "learning_rate": 0.0001, "elbo": 3.62656, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.386816795683791}
{"step": 40705, "loss": 3.6255859375, "grad_norm": 0.7075884342193604, "learning_rate": 0.0001, "elbo": 3.62559, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3871100164203614}
{"step": 40710, "loss": 3.635264587402344, "grad_norm": 0.7368003726005554, "learning_rate": 0.0001, "elbo": 3.63526, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.387403237156932}
{"step": 40715, "loss": 3.635687255859375, "grad_norm": 0.6952731013298035, "learning_rate": 0.0001, "elbo": 3.63569, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3876964578935023}
{"step": 40720, "loss": 3.606021499633789, "grad_norm": 0.6662871241569519, "learning_rate": 0.0001, "elbo": 3.60602, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3879896786300727}
{"step": 40725, "loss": 3.6152748107910155, "grad_norm": 0.7573838233947754, "learning_rate": 0.0001, "elbo": 3.61527, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.388282899366643}
{"step": 40730, "loss": 3.630127716064453, "grad_norm": 0.7164965867996216, "learning_rate": 0.0001, "elbo": 3.63013, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3885761201032136}
{"step": 40735, "loss": 3.6279144287109375, "grad_norm": 0.6889925003051758, "learning_rate": 0.0001, "elbo": 3.62791, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.388869340839784}
{"step": 40740, "loss": 3.635606384277344, "grad_norm": 0.7260518670082092, "learning_rate": 0.0001, "elbo": 3.63561, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3891625615763545}
{"step": 40745, "loss": 3.6242889404296874, "grad_norm": 0.7085651755332947, "learning_rate": 0.0001, "elbo": 3.62429, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.389455782312925}
{"step": 40745, "loss": 3.6242889404296874, "grad_norm": 0.7085651755332947, "learning_rate": 0.0001, "elbo": 3.62429, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.389455782312925}
{"step": 40750, "loss": 3.618259811401367, "grad_norm": 0.6931264400482178, "learning_rate": 0.0001, "elbo": 3.61826, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.389749003049496}
{"step": 40755, "loss": 3.6265838623046873, "grad_norm": 0.8194186687469482, "learning_rate": 0.0001, "elbo": 3.62658, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3900422237860663}
{"step": 40760, "loss": 3.62872428894043, "grad_norm": 0.654127836227417, "learning_rate": 0.0001, "elbo": 3.62872, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.390335444522637}
{"step": 40765, "loss": 3.621874237060547, "grad_norm": 0.6418418288230896, "learning_rate": 0.0001, "elbo": 3.62187, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3906286652592073}
{"step": 40765, "loss": 3.621874237060547, "grad_norm": 0.6418418288230896, "learning_rate": 0.0001, "elbo": 3.62187, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3906286652592073}
{"step": 40770, "loss": 3.6092395782470703, "grad_norm": 0.6595944166183472, "learning_rate": 0.0001, "elbo": 3.60924, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3909218859957777}
{"step": 40775, "loss": 3.6352787017822266, "grad_norm": 0.6991097331047058, "learning_rate": 0.0001, "elbo": 3.63528, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.391215106732348}
{"step": 40780, "loss": 3.6439445495605467, "grad_norm": 1.0065637826919556, "learning_rate": 0.0001, "elbo": 3.64394, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3915083274689186}
{"step": 40785, "loss": 3.621012878417969, "grad_norm": 0.7002291083335876, "learning_rate": 0.0001, "elbo": 3.62101, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.391801548205489}
{"step": 40790, "loss": 3.6390220642089846, "grad_norm": 0.6686602830886841, "learning_rate": 0.0001, "elbo": 3.63902, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3920947689420595}
{"step": 40795, "loss": 3.6371837615966798, "grad_norm": 0.6673561334609985, "learning_rate": 0.0001, "elbo": 3.63718, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.39238798967863}
{"step": 40800, "loss": 3.6356651306152346, "grad_norm": 0.6600506901741028, "learning_rate": 0.0001, "elbo": 3.63566, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3926812104152004}
{"step": 40805, "loss": 3.624281311035156, "grad_norm": 0.7002191543579102, "learning_rate": 0.0001, "elbo": 3.62428, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.392974431151771}
{"step": 40810, "loss": 3.6327423095703124, "grad_norm": 0.6418527364730835, "learning_rate": 0.0001, "elbo": 3.63274, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3932676518883413}
{"step": 40815, "loss": 3.619576263427734, "grad_norm": 0.7156990766525269, "learning_rate": 0.0001, "elbo": 3.61958, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.393560872624912}
{"step": 40815, "loss": 3.619576263427734, "grad_norm": 0.7156990766525269, "learning_rate": 0.0001, "elbo": 3.61958, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.393560872624912}
{"step": 40820, "loss": 3.6272186279296874, "grad_norm": 0.9236752390861511, "learning_rate": 0.0001, "elbo": 3.62722, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3938540933614827}
{"step": 40825, "loss": 3.6114696502685546, "grad_norm": 0.7300692796707153, "learning_rate": 0.0001, "elbo": 3.61147, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.394147314098053}
{"step": 40825, "loss": 3.6114696502685546, "grad_norm": 0.7300692796707153, "learning_rate": 0.0001, "elbo": 3.61147, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.394147314098053}
{"step": 40830, "loss": 3.629041290283203, "grad_norm": 0.6829663515090942, "learning_rate": 0.0001, "elbo": 3.62904, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3944405348346236}
{"step": 40830, "loss": 3.629041290283203, "grad_norm": 0.6829663515090942, "learning_rate": 0.0001, "elbo": 3.62904, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3944405348346236}
{"step": 40835, "loss": 3.619488525390625, "grad_norm": 0.6952897310256958, "learning_rate": 0.0001, "elbo": 3.61949, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.394733755571194}
{"step": 40840, "loss": 3.621626281738281, "grad_norm": 0.6937105059623718, "learning_rate": 0.0001, "elbo": 3.62163, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3950269763077645}
{"step": 40845, "loss": 3.6316680908203125, "grad_norm": 0.6794688701629639, "learning_rate": 0.0001, "elbo": 3.63167, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.395320197044335}
{"step": 40850, "loss": 3.624947357177734, "grad_norm": 0.7169348001480103, "learning_rate": 0.0001, "elbo": 3.62495, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3956134177809054}
{"step": 40855, "loss": 3.6178573608398437, "grad_norm": 0.6816346049308777, "learning_rate": 0.0001, "elbo": 3.61786, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.395906638517476}
{"step": 40855, "loss": 3.6178573608398437, "grad_norm": 0.6816346049308777, "learning_rate": 0.0001, "elbo": 3.61786, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.395906638517476}
{"step": 40860, "loss": 3.624536895751953, "grad_norm": 0.6931760311126709, "learning_rate": 0.0001, "elbo": 3.62454, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3961998592540463}
{"step": 40865, "loss": 3.636404800415039, "grad_norm": 0.6981156468391418, "learning_rate": 0.0001, "elbo": 3.6364, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.396493079990617}
{"step": 40870, "loss": 3.612689971923828, "grad_norm": 0.6763444542884827, "learning_rate": 0.0001, "elbo": 3.61269, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3967863007271872}
{"step": 40875, "loss": 3.626042938232422, "grad_norm": 0.6698905825614929, "learning_rate": 0.0001, "elbo": 3.62604, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.397079521463758}
{"step": 40880, "loss": 3.6124000549316406, "grad_norm": 0.6888208985328674, "learning_rate": 0.0001, "elbo": 3.6124, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3973727422003286}
{"step": 40885, "loss": 3.6336578369140624, "grad_norm": 0.6619782447814941, "learning_rate": 0.0001, "elbo": 3.63366, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.397665962936899}
{"step": 40890, "loss": 3.6316986083984375, "grad_norm": 1.002898931503296, "learning_rate": 0.0001, "elbo": 3.6317, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3979591836734695}
{"step": 40895, "loss": 3.637839508056641, "grad_norm": 0.7232201099395752, "learning_rate": 0.0001, "elbo": 3.63784, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.39825240441004}
{"step": 40900, "loss": 3.635123443603516, "grad_norm": 0.6934031248092651, "learning_rate": 0.0001, "elbo": 3.63512, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3985456251466104}
{"step": 40905, "loss": 3.6150211334228515, "grad_norm": 0.7595177292823792, "learning_rate": 0.0001, "elbo": 3.61502, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.398838845883181}
{"step": 40910, "loss": 3.6421581268310548, "grad_norm": 0.6501432657241821, "learning_rate": 0.0001, "elbo": 3.64216, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3991320666197513}
{"step": 40915, "loss": 3.6168060302734375, "grad_norm": 0.6433065533638, "learning_rate": 0.0001, "elbo": 3.61681, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.3994252873563218}
{"step": 40920, "loss": 3.618227005004883, "grad_norm": 0.6633970737457275, "learning_rate": 0.0001, "elbo": 3.61823, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.399718508092892}
{"step": 40925, "loss": 3.6354297637939452, "grad_norm": 0.7784897685050964, "learning_rate": 0.0001, "elbo": 3.63543, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4000117288294627}
{"step": 40930, "loss": 3.626507568359375, "grad_norm": 0.7257338762283325, "learning_rate": 0.0001, "elbo": 3.62651, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.400304949566033}
{"step": 40935, "loss": 3.623457336425781, "grad_norm": 0.7418063879013062, "learning_rate": 0.0001, "elbo": 3.62346, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4005981703026036}
{"step": 40940, "loss": 3.628226470947266, "grad_norm": 0.7038906216621399, "learning_rate": 0.0001, "elbo": 3.62823, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4008913910391745}
{"step": 40945, "loss": 3.614873504638672, "grad_norm": 0.9297890067100525, "learning_rate": 0.0001, "elbo": 3.61487, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.401184611775745}
{"step": 40950, "loss": 3.6292770385742186, "grad_norm": 0.6413506269454956, "learning_rate": 0.0001, "elbo": 3.62928, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4014778325123154}
{"step": 40955, "loss": 3.6421852111816406, "grad_norm": 0.6718027591705322, "learning_rate": 0.0001, "elbo": 3.64219, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.401771053248886}
{"step": 40960, "loss": 3.6267631530761717, "grad_norm": 0.6730870604515076, "learning_rate": 0.0001, "elbo": 3.62676, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4020642739854563}
{"step": 40965, "loss": 3.635328674316406, "grad_norm": 0.758236825466156, "learning_rate": 0.0001, "elbo": 3.63533, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4023574947220268}
{"step": 40970, "loss": 3.6140472412109377, "grad_norm": 0.8178397417068481, "learning_rate": 0.0001, "elbo": 3.61405, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.402650715458597}
{"step": 40975, "loss": 3.6300464630126954, "grad_norm": 1.0902460813522339, "learning_rate": 0.0001, "elbo": 3.63005, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4029439361951677}
{"step": 40980, "loss": 3.620514678955078, "grad_norm": 0.7275047302246094, "learning_rate": 0.0001, "elbo": 3.62051, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.403237156931738}
{"step": 40985, "loss": 3.6257083892822264, "grad_norm": 0.6470356583595276, "learning_rate": 0.0001, "elbo": 3.62571, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4035303776683086}
{"step": 40990, "loss": 3.624165344238281, "grad_norm": 0.6720106601715088, "learning_rate": 0.0001, "elbo": 3.62417, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.403823598404879}
{"step": 40995, "loss": 3.6168670654296875, "grad_norm": 0.6814967393875122, "learning_rate": 0.0001, "elbo": 3.61687, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.40411681914145}
{"step": 41000, "loss": 3.6421249389648436, "grad_norm": 0.9477030634880066, "learning_rate": 0.0001, "elbo": 3.64212, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4044100398780204}
{"step": 41000, "loss": 3.6421249389648436, "grad_norm": 0.9477030634880066, "learning_rate": 0.0001, "elbo": 3.64212, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4044100398780204}
{"step": 41005, "loss": 3.6164615631103514, "grad_norm": 0.6497567296028137, "learning_rate": 0.0001, "elbo": 3.61646, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.404703260614591}
{"step": 41010, "loss": 3.625873565673828, "grad_norm": 0.6781630516052246, "learning_rate": 0.0001, "elbo": 3.62587, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4049964813511613}
{"step": 41015, "loss": 3.619648742675781, "grad_norm": 0.6827596426010132, "learning_rate": 0.0001, "elbo": 3.61965, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4052897020877317}
{"step": 41020, "loss": 3.622979736328125, "grad_norm": 0.6769475340843201, "learning_rate": 0.0001, "elbo": 3.62298, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.405582922824302}
{"step": 41025, "loss": 3.6326393127441405, "grad_norm": 0.7153015732765198, "learning_rate": 0.0001, "elbo": 3.63264, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4058761435608726}
{"step": 41025, "loss": 3.6326393127441405, "grad_norm": 0.7153015732765198, "learning_rate": 0.0001, "elbo": 3.63264, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4058761435608726}
{"step": 41030, "loss": 3.6338539123535156, "grad_norm": 0.6865702271461487, "learning_rate": 0.0001, "elbo": 3.63385, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.406169364297443}
{"step": 41035, "loss": 3.6094459533691405, "grad_norm": 0.6623611450195312, "learning_rate": 0.0001, "elbo": 3.60945, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4064625850340136}
{"step": 41040, "loss": 3.599664306640625, "grad_norm": 0.7314391732215881, "learning_rate": 0.0001, "elbo": 3.59966, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.406755805770584}
{"step": 41045, "loss": 3.636463165283203, "grad_norm": 0.9547824859619141, "learning_rate": 0.0001, "elbo": 3.63646, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4070490265071545}
{"step": 41050, "loss": 3.622113800048828, "grad_norm": 0.7038784027099609, "learning_rate": 0.0001, "elbo": 3.62211, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.407342247243725}
{"step": 41055, "loss": 3.610465240478516, "grad_norm": 0.7024166584014893, "learning_rate": 0.0001, "elbo": 3.61047, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4076354679802954}
{"step": 41060, "loss": 3.6228389739990234, "grad_norm": 0.7153693437576294, "learning_rate": 0.0001, "elbo": 3.62284, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.407928688716866}
{"step": 41065, "loss": 3.6364791870117186, "grad_norm": 0.7160780429840088, "learning_rate": 0.0001, "elbo": 3.63648, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4082219094534367}
{"step": 41065, "loss": 3.6364791870117186, "grad_norm": 0.7160780429840088, "learning_rate": 0.0001, "elbo": 3.63648, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4082219094534367}
{"step": 41070, "loss": 3.6254718780517576, "grad_norm": 0.7081663012504578, "learning_rate": 0.0001, "elbo": 3.62547, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.408515130190007}
{"step": 41075, "loss": 3.6352439880371095, "grad_norm": 0.66325843334198, "learning_rate": 0.0001, "elbo": 3.63524, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4088083509265776}
{"step": 41080, "loss": 3.6281402587890623, "grad_norm": 0.7305742502212524, "learning_rate": 0.0001, "elbo": 3.62814, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.409101571663148}
{"step": 41085, "loss": 3.642713928222656, "grad_norm": 0.8355255722999573, "learning_rate": 0.0001, "elbo": 3.64271, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4093947923997185}
{"step": 41090, "loss": 3.632590484619141, "grad_norm": 0.6835603713989258, "learning_rate": 0.0001, "elbo": 3.63259, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.409688013136289}
{"step": 41095, "loss": 3.6410194396972657, "grad_norm": 0.7387020587921143, "learning_rate": 0.0001, "elbo": 3.64102, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4099812338728595}
{"step": 41100, "loss": 3.625926971435547, "grad_norm": 0.6839507818222046, "learning_rate": 0.0001, "elbo": 3.62593, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.41027445460943}
{"step": 41105, "loss": 3.626512145996094, "grad_norm": 0.6527733206748962, "learning_rate": 0.0001, "elbo": 3.62651, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4105676753460004}
{"step": 41110, "loss": 3.63065185546875, "grad_norm": 0.7146207690238953, "learning_rate": 0.0001, "elbo": 3.63065, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.410860896082571}
{"step": 41115, "loss": 3.6212413787841795, "grad_norm": 0.7377517223358154, "learning_rate": 0.0001, "elbo": 3.62124, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4111541168191413}
{"step": 41120, "loss": 3.65050048828125, "grad_norm": 0.7428113222122192, "learning_rate": 0.0001, "elbo": 3.6505, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.411447337555712}
{"step": 41125, "loss": 3.623036193847656, "grad_norm": 0.7899875044822693, "learning_rate": 0.0001, "elbo": 3.62304, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4117405582922826}
{"step": 41130, "loss": 3.622859573364258, "grad_norm": 0.8657848834991455, "learning_rate": 0.0001, "elbo": 3.62286, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.412033779028853}
{"step": 41135, "loss": 3.6189239501953123, "grad_norm": 1.0678508281707764, "learning_rate": 0.0001, "elbo": 3.61892, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4123269997654235}
{"step": 41140, "loss": 3.6314697265625, "grad_norm": 0.7454761266708374, "learning_rate": 0.0001, "elbo": 3.63147, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.412620220501994}
{"step": 41145, "loss": 3.6380455017089846, "grad_norm": 0.6553971171379089, "learning_rate": 0.0001, "elbo": 3.63805, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4129134412385644}
{"step": 41150, "loss": 3.637347412109375, "grad_norm": 0.7517694234848022, "learning_rate": 0.0001, "elbo": 3.63735, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.413206661975135}
{"step": 41155, "loss": 3.6368179321289062, "grad_norm": 0.7747761011123657, "learning_rate": 0.0001, "elbo": 3.63682, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4134998827117053}
{"step": 41160, "loss": 3.6246246337890624, "grad_norm": 0.8151354789733887, "learning_rate": 0.0001, "elbo": 3.62462, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.413793103448276}
{"step": 41165, "loss": 3.616175079345703, "grad_norm": 0.6821975111961365, "learning_rate": 0.0001, "elbo": 3.61617, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4140863241848463}
{"step": 41170, "loss": 3.6133750915527343, "grad_norm": 0.686520516872406, "learning_rate": 0.0001, "elbo": 3.61337, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4143795449214167}
{"step": 41175, "loss": 3.6378810882568358, "grad_norm": 0.6891809105873108, "learning_rate": 0.0001, "elbo": 3.63788, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.414672765657987}
{"step": 41180, "loss": 3.6268272399902344, "grad_norm": 0.6547942161560059, "learning_rate": 0.0001, "elbo": 3.62683, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4149659863945576}
{"step": 41185, "loss": 3.617158889770508, "grad_norm": 0.6317828297615051, "learning_rate": 0.0001, "elbo": 3.61716, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.415259207131128}
{"step": 41190, "loss": 3.619877243041992, "grad_norm": 0.6452807784080505, "learning_rate": 0.0001, "elbo": 3.61988, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.415552427867699}
{"step": 41195, "loss": 3.6349864959716798, "grad_norm": 1.1656948328018188, "learning_rate": 0.0001, "elbo": 3.63499, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4158456486042694}
{"step": 41200, "loss": 3.6220470428466798, "grad_norm": 0.7038933038711548, "learning_rate": 0.0001, "elbo": 3.62205, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.41613886934084}
{"step": 41205, "loss": 3.6143890380859376, "grad_norm": 0.7026626467704773, "learning_rate": 0.0001, "elbo": 3.61439, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4164320900774103}
{"step": 41210, "loss": 3.6302444458007814, "grad_norm": 0.7498909831047058, "learning_rate": 0.0001, "elbo": 3.63024, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.416725310813981}
{"step": 41215, "loss": 3.642745590209961, "grad_norm": 0.6378501057624817, "learning_rate": 0.0001, "elbo": 3.64275, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4170185315505512}
{"step": 41220, "loss": 3.6252197265625, "grad_norm": 0.6526596546173096, "learning_rate": 0.0001, "elbo": 3.62522, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4173117522871217}
{"step": 41225, "loss": 3.6313514709472656, "grad_norm": 0.9050835371017456, "learning_rate": 0.0001, "elbo": 3.63135, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.417604973023692}
{"step": 41230, "loss": 3.6361026763916016, "grad_norm": 0.7523871660232544, "learning_rate": 0.0001, "elbo": 3.6361, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4178981937602626}
{"step": 41235, "loss": 3.6310760498046877, "grad_norm": 0.7139204740524292, "learning_rate": 0.0001, "elbo": 3.63108, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.418191414496833}
{"step": 41240, "loss": 3.626320648193359, "grad_norm": 0.6875177621841431, "learning_rate": 0.0001, "elbo": 3.62632, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4184846352334035}
{"step": 41245, "loss": 3.6269157409667967, "grad_norm": 0.7056213021278381, "learning_rate": 0.0001, "elbo": 3.62692, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4187778559699744}
{"step": 41250, "loss": 3.6193363189697267, "grad_norm": 0.7018019556999207, "learning_rate": 0.0001, "elbo": 3.61934, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.419071076706545}
{"step": 41255, "loss": 3.6162357330322266, "grad_norm": 0.6689363121986389, "learning_rate": 0.0001, "elbo": 3.61624, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4193642974431153}
{"step": 41255, "loss": 3.6162357330322266, "grad_norm": 0.6689363121986389, "learning_rate": 0.0001, "elbo": 3.61624, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4193642974431153}
{"step": 41260, "loss": 3.6238327026367188, "grad_norm": 0.6845401525497437, "learning_rate": 0.0001, "elbo": 3.62383, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4196575181796858}
{"step": 41265, "loss": 3.622736358642578, "grad_norm": 0.6470653414726257, "learning_rate": 0.0001, "elbo": 3.62274, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4199507389162562}
{"step": 41265, "loss": 3.622736358642578, "grad_norm": 0.6470653414726257, "learning_rate": 0.0001, "elbo": 3.62274, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4199507389162562}
{"step": 41270, "loss": 3.6216461181640627, "grad_norm": 0.6746450662612915, "learning_rate": 0.0001, "elbo": 3.62165, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4202439596528267}
{"step": 41275, "loss": 3.611572265625, "grad_norm": 0.6775394678115845, "learning_rate": 0.0001, "elbo": 3.61157, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.420537180389397}
{"step": 41275, "loss": 3.611572265625, "grad_norm": 0.6775394678115845, "learning_rate": 0.0001, "elbo": 3.61157, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.420537180389397}
{"step": 41280, "loss": 3.632233428955078, "grad_norm": 0.7072725892066956, "learning_rate": 0.0001, "elbo": 3.63223, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4208304011259676}
{"step": 41285, "loss": 3.6284584045410155, "grad_norm": 0.7173387408256531, "learning_rate": 0.0001, "elbo": 3.62846, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.421123621862538}
{"step": 41290, "loss": 3.6271934509277344, "grad_norm": 0.6927419304847717, "learning_rate": 0.0001, "elbo": 3.62719, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4214168425991085}
{"step": 41295, "loss": 3.6216960906982423, "grad_norm": 0.6951467990875244, "learning_rate": 0.0001, "elbo": 3.6217, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.421710063335679}
{"step": 41300, "loss": 3.622711944580078, "grad_norm": 0.7179681658744812, "learning_rate": 0.0001, "elbo": 3.62271, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4220032840722494}
{"step": 41305, "loss": 3.6175941467285155, "grad_norm": 0.6447277069091797, "learning_rate": 0.0001, "elbo": 3.61759, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.42229650480882}
{"step": 41310, "loss": 3.637562561035156, "grad_norm": 0.679801344871521, "learning_rate": 0.0001, "elbo": 3.63756, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4225897255453908}
{"step": 41315, "loss": 3.624551773071289, "grad_norm": 0.7432870268821716, "learning_rate": 0.0001, "elbo": 3.62455, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.422882946281961}
{"step": 41320, "loss": 3.6451461791992186, "grad_norm": 0.6665114164352417, "learning_rate": 0.0001, "elbo": 3.64515, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4231761670185317}
{"step": 41325, "loss": 3.6125259399414062, "grad_norm": 0.7608271241188049, "learning_rate": 0.0001, "elbo": 3.61253, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.423469387755102}
{"step": 41330, "loss": 3.6359935760498048, "grad_norm": 0.7867071032524109, "learning_rate": 0.0001, "elbo": 3.63599, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4237626084916726}
{"step": 41335, "loss": 3.6208145141601564, "grad_norm": 0.6357605457305908, "learning_rate": 0.0001, "elbo": 3.62081, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.424055829228243}
{"step": 41340, "loss": 3.620069122314453, "grad_norm": 0.6931310892105103, "learning_rate": 0.0001, "elbo": 3.62007, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4243490499648135}
{"step": 41345, "loss": 3.63258056640625, "grad_norm": 0.6832059025764465, "learning_rate": 0.0001, "elbo": 3.63258, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.424642270701384}
{"step": 41350, "loss": 3.629925537109375, "grad_norm": 0.7472088932991028, "learning_rate": 0.0001, "elbo": 3.62993, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4249354914379544}
{"step": 41355, "loss": 3.6220371246337892, "grad_norm": 0.7827274203300476, "learning_rate": 0.0001, "elbo": 3.62204, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.425228712174525}
{"step": 41360, "loss": 3.6313995361328124, "grad_norm": 0.7229512333869934, "learning_rate": 0.0001, "elbo": 3.6314, "mask_frac": 0.5013, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4255219329110953}
{"step": 41365, "loss": 3.627090072631836, "grad_norm": 0.7202163934707642, "learning_rate": 0.0001, "elbo": 3.62709, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.425815153647666}
{"step": 41370, "loss": 3.6374221801757813, "grad_norm": 0.7629281282424927, "learning_rate": 0.0001, "elbo": 3.63742, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4261083743842367}
{"step": 41375, "loss": 3.624958038330078, "grad_norm": 0.8333016633987427, "learning_rate": 0.0001, "elbo": 3.62496, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.426401595120807}
{"step": 41380, "loss": 3.617933654785156, "grad_norm": 0.6452106237411499, "learning_rate": 0.0001, "elbo": 3.61793, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4266948158573776}
{"step": 41380, "loss": 3.617933654785156, "grad_norm": 0.6452106237411499, "learning_rate": 0.0001, "elbo": 3.61793, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4266948158573776}
{"step": 41385, "loss": 3.6233169555664064, "grad_norm": 0.6569870710372925, "learning_rate": 0.0001, "elbo": 3.62332, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.426988036593948}
{"step": 41390, "loss": 3.609699249267578, "grad_norm": 0.643577516078949, "learning_rate": 0.0001, "elbo": 3.6097, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4272812573305185}
{"step": 41395, "loss": 3.6224037170410157, "grad_norm": 0.9240525364875793, "learning_rate": 0.0001, "elbo": 3.6224, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.427574478067089}
{"step": 41400, "loss": 3.6403526306152343, "grad_norm": 0.723659336566925, "learning_rate": 0.0001, "elbo": 3.64035, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4278676988036594}
{"step": 41405, "loss": 3.616409683227539, "grad_norm": 1.2157005071640015, "learning_rate": 0.0001, "elbo": 3.61641, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.42816091954023}
{"step": 41410, "loss": 3.629549026489258, "grad_norm": 0.7486332058906555, "learning_rate": 0.0001, "elbo": 3.62955, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4284541402768003}
{"step": 41415, "loss": 3.6452873229980467, "grad_norm": 0.654333233833313, "learning_rate": 0.0001, "elbo": 3.64529, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4287473610133707}
{"step": 41420, "loss": 3.619202423095703, "grad_norm": 0.7454355359077454, "learning_rate": 0.0001, "elbo": 3.6192, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.429040581749941}
{"step": 41425, "loss": 3.627008819580078, "grad_norm": 0.6616091728210449, "learning_rate": 0.0001, "elbo": 3.62701, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4293338024865117}
{"step": 41430, "loss": 3.6118484497070313, "grad_norm": 0.6856338381767273, "learning_rate": 0.0001, "elbo": 3.61185, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.429627023223082}
{"step": 41435, "loss": 3.6342906951904297, "grad_norm": 0.7207825779914856, "learning_rate": 0.0001, "elbo": 3.63429, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.429920243959653}
{"step": 41440, "loss": 3.6314430236816406, "grad_norm": 0.742638349533081, "learning_rate": 0.0001, "elbo": 3.63144, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4302134646962235}
{"step": 41445, "loss": 3.626875305175781, "grad_norm": 1.1492754220962524, "learning_rate": 0.0001, "elbo": 3.62688, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.430506685432794}
{"step": 41450, "loss": 3.6294326782226562, "grad_norm": 0.6668179035186768, "learning_rate": 0.0001, "elbo": 3.62943, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4307999061693644}
{"step": 41455, "loss": 3.6226287841796876, "grad_norm": 0.6972980499267578, "learning_rate": 0.0001, "elbo": 3.62263, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.431093126905935}
{"step": 41460, "loss": 3.643504333496094, "grad_norm": 0.913151204586029, "learning_rate": 0.0001, "elbo": 3.6435, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4313863476425053}
{"step": 41465, "loss": 3.6310462951660156, "grad_norm": 0.7103506326675415, "learning_rate": 0.0001, "elbo": 3.63105, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4316795683790757}
{"step": 41470, "loss": 3.6422069549560545, "grad_norm": 0.7027323246002197, "learning_rate": 0.0001, "elbo": 3.64221, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.431972789115646}
{"step": 41475, "loss": 3.6238639831542967, "grad_norm": 0.712175726890564, "learning_rate": 0.0001, "elbo": 3.62386, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4322660098522166}
{"step": 41480, "loss": 3.6262794494628907, "grad_norm": 0.6980806589126587, "learning_rate": 0.0001, "elbo": 3.62628, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.432559230588787}
{"step": 41485, "loss": 3.6231632232666016, "grad_norm": 0.7235622406005859, "learning_rate": 0.0001, "elbo": 3.62316, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4328524513253575}
{"step": 41490, "loss": 3.6234703063964844, "grad_norm": 0.7590695023536682, "learning_rate": 0.0001, "elbo": 3.62347, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4331456720619284}
{"step": 41495, "loss": 3.632315826416016, "grad_norm": 0.6366739869117737, "learning_rate": 0.0001, "elbo": 3.63232, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.433438892798499}
{"step": 41500, "loss": 3.6088516235351564, "grad_norm": 1.048111081123352, "learning_rate": 0.0001, "elbo": 3.60885, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4337321135350694}
{"step": 41505, "loss": 3.6184661865234373, "grad_norm": 0.7022697329521179, "learning_rate": 0.0001, "elbo": 3.61847, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.43402533427164}
{"step": 41510, "loss": 3.6185958862304686, "grad_norm": 0.6618070602416992, "learning_rate": 0.0001, "elbo": 3.6186, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4343185550082103}
{"step": 41510, "loss": 3.6185958862304686, "grad_norm": 0.6618070602416992, "learning_rate": 0.0001, "elbo": 3.6186, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4343185550082103}
{"step": 41515, "loss": 3.6386898040771483, "grad_norm": 0.6638540029525757, "learning_rate": 0.0001, "elbo": 3.63869, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4346117757447807}
{"step": 41520, "loss": 3.6375091552734373, "grad_norm": 0.7041616439819336, "learning_rate": 0.0001, "elbo": 3.63751, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.434904996481351}
{"step": 41525, "loss": 3.620920181274414, "grad_norm": 0.6857937574386597, "learning_rate": 0.0001, "elbo": 3.62092, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4351982172179216}
{"step": 41530, "loss": 3.6209793090820312, "grad_norm": 0.6478568315505981, "learning_rate": 0.0001, "elbo": 3.62098, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.435491437954492}
{"step": 41535, "loss": 3.623028564453125, "grad_norm": 0.7461704611778259, "learning_rate": 0.0001, "elbo": 3.62303, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4357846586910625}
{"step": 41540, "loss": 3.6374748229980467, "grad_norm": 0.6748342514038086, "learning_rate": 0.0001, "elbo": 3.63748, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.436077879427633}
{"step": 41545, "loss": 3.633050537109375, "grad_norm": 0.8305267095565796, "learning_rate": 0.0001, "elbo": 3.63305, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4363711001642034}
{"step": 41550, "loss": 3.6266242980957033, "grad_norm": 0.7595491409301758, "learning_rate": 0.0001, "elbo": 3.62662, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.436664320900774}
{"step": 41555, "loss": 3.6353965759277345, "grad_norm": 0.6884252429008484, "learning_rate": 0.0001, "elbo": 3.6354, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4369575416373443}
{"step": 41560, "loss": 3.625690460205078, "grad_norm": 0.7018266916275024, "learning_rate": 0.0001, "elbo": 3.62569, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4372507623739152}
{"step": 41565, "loss": 3.614430236816406, "grad_norm": 0.7234361171722412, "learning_rate": 0.0001, "elbo": 3.61443, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4375439831104857}
{"step": 41570, "loss": 3.6297035217285156, "grad_norm": 0.7389582991600037, "learning_rate": 0.0001, "elbo": 3.6297, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.437837203847056}
{"step": 41575, "loss": 3.6397354125976564, "grad_norm": 0.7266446352005005, "learning_rate": 0.0001, "elbo": 3.63974, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4381304245836266}
{"step": 41580, "loss": 3.623592758178711, "grad_norm": 0.7591334581375122, "learning_rate": 0.0001, "elbo": 3.62359, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.438423645320197}
{"step": 41585, "loss": 3.625269317626953, "grad_norm": 0.7530891299247742, "learning_rate": 0.0001, "elbo": 3.62527, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4387168660567675}
{"step": 41590, "loss": 3.624755859375, "grad_norm": 0.9943190813064575, "learning_rate": 0.0001, "elbo": 3.62476, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.439010086793338}
{"step": 41595, "loss": 3.643474578857422, "grad_norm": 0.7985978126525879, "learning_rate": 0.0001, "elbo": 3.64347, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4393033075299084}
{"step": 41600, "loss": 3.619589996337891, "grad_norm": 0.6538503766059875, "learning_rate": 0.0001, "elbo": 3.61959, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.439596528266479}
{"step": 41605, "loss": 3.6402122497558596, "grad_norm": 0.7479512095451355, "learning_rate": 0.0001, "elbo": 3.64021, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4398897490030493}
{"step": 41610, "loss": 3.6033699035644533, "grad_norm": 0.6920838952064514, "learning_rate": 0.0001, "elbo": 3.60337, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4401829697396202}
{"step": 41615, "loss": 3.627593994140625, "grad_norm": 0.6335557699203491, "learning_rate": 0.0001, "elbo": 3.62759, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4404761904761907}
{"step": 41620, "loss": 3.621672821044922, "grad_norm": 0.6587380766868591, "learning_rate": 0.0001, "elbo": 3.62167, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.440769411212761}
{"step": 41625, "loss": 3.631118392944336, "grad_norm": 0.8305714130401611, "learning_rate": 0.0001, "elbo": 3.63112, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4410626319493316}
{"step": 41630, "loss": 3.6320335388183596, "grad_norm": 0.6816043853759766, "learning_rate": 0.0001, "elbo": 3.63203, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.441355852685902}
{"step": 41635, "loss": 3.626637649536133, "grad_norm": 0.7152193188667297, "learning_rate": 0.0001, "elbo": 3.62664, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4416490734224725}
{"step": 41640, "loss": 3.6306175231933593, "grad_norm": 0.6877554059028625, "learning_rate": 0.0001, "elbo": 3.63062, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.441942294159043}
{"step": 41645, "loss": 3.6331939697265625, "grad_norm": 0.6727271676063538, "learning_rate": 0.0001, "elbo": 3.63319, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4422355148956134}
{"step": 41650, "loss": 3.6201171875, "grad_norm": 0.6876158118247986, "learning_rate": 0.0001, "elbo": 3.62012, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.442528735632184}
{"step": 41655, "loss": 3.633351135253906, "grad_norm": 0.6798315048217773, "learning_rate": 0.0001, "elbo": 3.63335, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4428219563687543}
{"step": 41660, "loss": 3.612122344970703, "grad_norm": 1.01943039894104, "learning_rate": 0.0001, "elbo": 3.61212, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4431151771053248}
{"step": 41665, "loss": 3.6282730102539062, "grad_norm": 0.7089085578918457, "learning_rate": 0.0001, "elbo": 3.62827, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4434083978418952}
{"step": 41670, "loss": 3.627462387084961, "grad_norm": 0.6596913933753967, "learning_rate": 0.0001, "elbo": 3.62746, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4437016185784657}
{"step": 41675, "loss": 3.634634780883789, "grad_norm": 0.6901237368583679, "learning_rate": 0.0001, "elbo": 3.63463, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.443994839315036}
{"step": 41680, "loss": 3.6204303741455077, "grad_norm": 0.7756457924842834, "learning_rate": 0.0001, "elbo": 3.62043, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.444288060051607}
{"step": 41685, "loss": 3.6216812133789062, "grad_norm": 0.6567729115486145, "learning_rate": 0.0001, "elbo": 3.62168, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4445812807881775}
{"step": 41690, "loss": 3.6270820617675783, "grad_norm": 0.6755946278572083, "learning_rate": 0.0001, "elbo": 3.62708, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.444874501524748}
{"step": 41695, "loss": 3.639516830444336, "grad_norm": 0.7025787830352783, "learning_rate": 0.0001, "elbo": 3.63952, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4451677222613184}
{"step": 41700, "loss": 3.6117687225341797, "grad_norm": 0.757576048374176, "learning_rate": 0.0001, "elbo": 3.61177, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.445460942997889}
{"step": 41700, "loss": 3.6117687225341797, "grad_norm": 0.757576048374176, "learning_rate": 0.0001, "elbo": 3.61177, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.445460942997889}
{"step": 41705, "loss": 3.6181480407714846, "grad_norm": 0.7180948257446289, "learning_rate": 0.0001, "elbo": 3.61815, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4457541637344593}
{"step": 41710, "loss": 3.6182559967041015, "grad_norm": 0.6839293241500854, "learning_rate": 0.0001, "elbo": 3.61826, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4460473844710298}
{"step": 41715, "loss": 3.626247024536133, "grad_norm": 0.6645135283470154, "learning_rate": 0.0001, "elbo": 3.62625, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4463406052076}
{"step": 41720, "loss": 3.6442405700683596, "grad_norm": 0.7324302196502686, "learning_rate": 0.0001, "elbo": 3.64424, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4466338259441707}
{"step": 41725, "loss": 3.619895172119141, "grad_norm": 0.843683660030365, "learning_rate": 0.0001, "elbo": 3.61989, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.446927046680741}
{"step": 41730, "loss": 3.6057586669921875, "grad_norm": 0.7465780973434448, "learning_rate": 0.0001, "elbo": 3.60576, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4472202674173116}
{"step": 41735, "loss": 3.6147193908691406, "grad_norm": 0.9690870642662048, "learning_rate": 0.0001, "elbo": 3.61472, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4475134881538825}
{"step": 41740, "loss": 3.621367645263672, "grad_norm": 0.749233603477478, "learning_rate": 0.0001, "elbo": 3.62137, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.447806708890453}
{"step": 41745, "loss": 3.6281639099121095, "grad_norm": 0.7980407476425171, "learning_rate": 0.0001, "elbo": 3.62816, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4480999296270234}
{"step": 41750, "loss": 3.60953369140625, "grad_norm": 0.6883112788200378, "learning_rate": 0.0001, "elbo": 3.60953, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.448393150363594}
{"step": 41755, "loss": 3.6304088592529298, "grad_norm": 0.7008383274078369, "learning_rate": 0.0001, "elbo": 3.63041, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4486863711001643}
{"step": 41760, "loss": 3.6311408996582033, "grad_norm": 0.9835392236709595, "learning_rate": 0.0001, "elbo": 3.63114, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4489795918367347}
{"step": 41765, "loss": 3.623065185546875, "grad_norm": 0.7613714337348938, "learning_rate": 0.0001, "elbo": 3.62307, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.449272812573305}
{"step": 41770, "loss": 3.627196502685547, "grad_norm": 0.713351845741272, "learning_rate": 0.0001, "elbo": 3.6272, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4495660333098757}
{"step": 41775, "loss": 3.6313461303710937, "grad_norm": 1.053891658782959, "learning_rate": 0.0001, "elbo": 3.63135, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.449859254046446}
{"step": 41780, "loss": 3.623576354980469, "grad_norm": 0.6574445366859436, "learning_rate": 0.0001, "elbo": 3.62358, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4501524747830166}
{"step": 41785, "loss": 3.636850357055664, "grad_norm": 0.6497429013252258, "learning_rate": 0.0001, "elbo": 3.63685, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.450445695519587}
{"step": 41790, "loss": 3.6196388244628905, "grad_norm": 0.6842155456542969, "learning_rate": 0.0001, "elbo": 3.61964, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4507389162561575}
{"step": 41790, "loss": 3.6196388244628905, "grad_norm": 0.6842155456542969, "learning_rate": 0.0001, "elbo": 3.61964, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4507389162561575}
{"step": 41795, "loss": 3.642600250244141, "grad_norm": 1.024343490600586, "learning_rate": 0.0001, "elbo": 3.6426, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.451032136992728}
{"step": 41800, "loss": 3.635794830322266, "grad_norm": 0.6977836489677429, "learning_rate": 0.0001, "elbo": 3.63579, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4513253577292984}
{"step": 41805, "loss": 3.6294483184814452, "grad_norm": 0.6836103796958923, "learning_rate": 0.0001, "elbo": 3.62945, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4516185784658693}
{"step": 41810, "loss": 3.6276779174804688, "grad_norm": 0.7396932244300842, "learning_rate": 0.0001, "elbo": 3.62768, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4519117992024397}
{"step": 41815, "loss": 3.6142311096191406, "grad_norm": 0.7766192555427551, "learning_rate": 0.0001, "elbo": 3.61423, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.45220501993901}
{"step": 41820, "loss": 3.6218551635742187, "grad_norm": 0.6946060657501221, "learning_rate": 0.0001, "elbo": 3.62186, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4524982406755806}
{"step": 41825, "loss": 3.629599761962891, "grad_norm": 0.794439971446991, "learning_rate": 0.0001, "elbo": 3.6296, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.452791461412151}
{"step": 41830, "loss": 3.6071002960205076, "grad_norm": 0.7165767550468445, "learning_rate": 0.0001, "elbo": 3.6071, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4530846821487216}
{"step": 41835, "loss": 3.6171875, "grad_norm": 0.685246467590332, "learning_rate": 0.0001, "elbo": 3.61719, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.453377902885292}
{"step": 41840, "loss": 3.60095100402832, "grad_norm": 0.7107846736907959, "learning_rate": 0.0001, "elbo": 3.60095, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4536711236218625}
{"step": 41845, "loss": 3.615401840209961, "grad_norm": 0.6962090134620667, "learning_rate": 0.0001, "elbo": 3.6154, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.453964344358433}
{"step": 41850, "loss": 3.6092239379882813, "grad_norm": 0.6952138543128967, "learning_rate": 0.0001, "elbo": 3.60922, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4542575650950034}
{"step": 41855, "loss": 3.61230583190918, "grad_norm": 0.7595515847206116, "learning_rate": 0.0001, "elbo": 3.61231, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.454550785831574}
{"step": 41860, "loss": 3.630086135864258, "grad_norm": 0.9936478734016418, "learning_rate": 0.0001, "elbo": 3.63009, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4548440065681447}
{"step": 41865, "loss": 3.620151901245117, "grad_norm": 1.0245780944824219, "learning_rate": 0.0001, "elbo": 3.62015, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.455137227304715}
{"step": 41870, "loss": 3.6191688537597657, "grad_norm": 0.9129311442375183, "learning_rate": 0.0001, "elbo": 3.61917, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4554304480412856}
{"step": 41875, "loss": 3.6220733642578127, "grad_norm": 0.7144832015037537, "learning_rate": 0.0001, "elbo": 3.62207, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.455723668777856}
{"step": 41880, "loss": 3.636988067626953, "grad_norm": 1.2709230184555054, "learning_rate": 0.0001, "elbo": 3.63699, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4560168895144265}
{"step": 41885, "loss": 3.610714340209961, "grad_norm": 0.6706746220588684, "learning_rate": 0.0001, "elbo": 3.61071, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.456310110250997}
{"step": 41890, "loss": 3.624102020263672, "grad_norm": 0.6514196395874023, "learning_rate": 0.0001, "elbo": 3.6241, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4566033309875674}
{"step": 41895, "loss": 3.609981918334961, "grad_norm": 0.701943576335907, "learning_rate": 0.0001, "elbo": 3.60998, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.456896551724138}
{"step": 41900, "loss": 3.625292205810547, "grad_norm": 0.6732972264289856, "learning_rate": 0.0001, "elbo": 3.62529, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4571897724607084}
{"step": 41905, "loss": 3.61396484375, "grad_norm": 0.8936247825622559, "learning_rate": 0.0001, "elbo": 3.61396, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.457482993197279}
{"step": 41910, "loss": 3.6251258850097656, "grad_norm": 0.7150644063949585, "learning_rate": 0.0001, "elbo": 3.62513, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4577762139338493}
{"step": 41910, "loss": 3.6251258850097656, "grad_norm": 0.7150644063949585, "learning_rate": 0.0001, "elbo": 3.62513, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4577762139338493}
{"step": 41915, "loss": 3.625279998779297, "grad_norm": 0.6833831667900085, "learning_rate": 0.0001, "elbo": 3.62528, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4580694346704197}
{"step": 41920, "loss": 3.609796905517578, "grad_norm": 0.7087821364402771, "learning_rate": 0.0001, "elbo": 3.6098, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.45836265540699}
{"step": 41925, "loss": 3.63297119140625, "grad_norm": 0.8400713801383972, "learning_rate": 0.0001, "elbo": 3.63297, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4586558761435606}
{"step": 41930, "loss": 3.6171146392822267, "grad_norm": 0.6699248552322388, "learning_rate": 0.0001, "elbo": 3.61711, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4589490968801315}
{"step": 41935, "loss": 3.6522315979003905, "grad_norm": 0.7903884649276733, "learning_rate": 0.0001, "elbo": 3.65223, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.459242317616702}
{"step": 41935, "loss": 3.6522315979003905, "grad_norm": 0.7903884649276733, "learning_rate": 0.0001, "elbo": 3.65223, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.459242317616702}
{"step": 41940, "loss": 3.614035797119141, "grad_norm": 0.6265981793403625, "learning_rate": 0.0001, "elbo": 3.61404, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4595355383532724}
{"step": 41945, "loss": 3.6133785247802734, "grad_norm": 0.7742424011230469, "learning_rate": 0.0001, "elbo": 3.61338, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.459828759089843}
{"step": 41945, "loss": 3.6133785247802734, "grad_norm": 0.7742424011230469, "learning_rate": 0.0001, "elbo": 3.61338, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.459828759089843}
{"step": 41950, "loss": 3.6108688354492187, "grad_norm": 0.6886347532272339, "learning_rate": 0.0001, "elbo": 3.61087, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4601219798264133}
{"step": 41950, "loss": 3.6108688354492187, "grad_norm": 0.6886347532272339, "learning_rate": 0.0001, "elbo": 3.61087, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4601219798264133}
{"step": 41955, "loss": 3.6112991333007813, "grad_norm": 0.6272887587547302, "learning_rate": 0.0001, "elbo": 3.6113, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.460415200562984}
{"step": 41960, "loss": 3.6449352264404298, "grad_norm": 0.6573960185050964, "learning_rate": 0.0001, "elbo": 3.64494, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4607084212995542}
{"step": 41965, "loss": 3.603715515136719, "grad_norm": 0.7200084924697876, "learning_rate": 0.0001, "elbo": 3.60372, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4610016420361247}
{"step": 41970, "loss": 3.6163105010986327, "grad_norm": 0.7008146047592163, "learning_rate": 0.0001, "elbo": 3.61631, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.461294862772695}
{"step": 41975, "loss": 3.62225227355957, "grad_norm": 0.8036834001541138, "learning_rate": 0.0001, "elbo": 3.62225, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4615880835092656}
{"step": 41980, "loss": 3.6237747192382814, "grad_norm": 0.6509895920753479, "learning_rate": 0.0001, "elbo": 3.62377, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4618813042458365}
{"step": 41985, "loss": 3.6112594604492188, "grad_norm": 0.686570942401886, "learning_rate": 0.0001, "elbo": 3.61126, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.462174524982407}
{"step": 41985, "loss": 3.6112594604492188, "grad_norm": 0.686570942401886, "learning_rate": 0.0001, "elbo": 3.61126, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.462174524982407}
{"step": 41990, "loss": 3.605498504638672, "grad_norm": 0.6541866064071655, "learning_rate": 0.0001, "elbo": 3.6055, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4624677457189774}
{"step": 41990, "loss": 3.605498504638672, "grad_norm": 0.6541866064071655, "learning_rate": 0.0001, "elbo": 3.6055, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4624677457189774}
{"step": 41995, "loss": 3.618208312988281, "grad_norm": 0.885535478591919, "learning_rate": 0.0001, "elbo": 3.61821, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.462760966455548}
{"step": 41995, "loss": 3.618208312988281, "grad_norm": 0.885535478591919, "learning_rate": 0.0001, "elbo": 3.61821, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.462760966455548}
{"step": 42000, "loss": 3.616832733154297, "grad_norm": 0.667877197265625, "learning_rate": 0.0001, "elbo": 3.61683, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4630541871921183}
{"step": 42000, "loss": 3.616832733154297, "grad_norm": 0.667877197265625, "learning_rate": 0.0001, "elbo": 3.61683, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4630541871921183}
{"step": 42005, "loss": 3.642200469970703, "grad_norm": 0.7711327075958252, "learning_rate": 0.0001, "elbo": 3.6422, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.463347407928689}
{"step": 42010, "loss": 3.6121490478515623, "grad_norm": 0.6943643093109131, "learning_rate": 0.0001, "elbo": 3.61215, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4636406286652592}
{"step": 42015, "loss": 3.6248729705810545, "grad_norm": 0.6937175393104553, "learning_rate": 0.0001, "elbo": 3.62487, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4639338494018297}
{"step": 42020, "loss": 3.6314838409423826, "grad_norm": 0.7837851643562317, "learning_rate": 0.0001, "elbo": 3.63148, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4642270701384}
{"step": 42025, "loss": 3.6227210998535155, "grad_norm": 0.7820327877998352, "learning_rate": 0.0001, "elbo": 3.62272, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4645202908749706}
{"step": 42030, "loss": 3.6251346588134767, "grad_norm": 0.6971881985664368, "learning_rate": 0.0001, "elbo": 3.62513, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.464813511611541}
{"step": 42035, "loss": 3.6212791442871093, "grad_norm": 0.7150837779045105, "learning_rate": 0.0001, "elbo": 3.62128, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4651067323481115}
{"step": 42040, "loss": 3.6390304565429688, "grad_norm": 0.7347205281257629, "learning_rate": 0.0001, "elbo": 3.63903, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.465399953084682}
{"step": 42045, "loss": 3.624859619140625, "grad_norm": 0.6725242733955383, "learning_rate": 0.0001, "elbo": 3.62486, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4656931738212524}
{"step": 42050, "loss": 3.6125347137451174, "grad_norm": 0.766190767288208, "learning_rate": 0.0001, "elbo": 3.61253, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4659863945578233}
{"step": 42055, "loss": 3.625390625, "grad_norm": 0.6897005438804626, "learning_rate": 0.0001, "elbo": 3.62539, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4662796152943938}
{"step": 42060, "loss": 3.6171173095703124, "grad_norm": 0.7093706727027893, "learning_rate": 0.0001, "elbo": 3.61712, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.466572836030964}
{"step": 42065, "loss": 3.6303482055664062, "grad_norm": 0.6860272288322449, "learning_rate": 0.0001, "elbo": 3.63035, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4668660567675347}
{"step": 42070, "loss": 3.6348304748535156, "grad_norm": 0.678311824798584, "learning_rate": 0.0001, "elbo": 3.63483, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.467159277504105}
{"step": 42075, "loss": 3.627043914794922, "grad_norm": 0.7363306879997253, "learning_rate": 0.0001, "elbo": 3.62704, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4674524982406756}
{"step": 42080, "loss": 3.6085365295410154, "grad_norm": 0.7550978064537048, "learning_rate": 0.0001, "elbo": 3.60854, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.467745718977246}
{"step": 42085, "loss": 3.6225242614746094, "grad_norm": 0.7196797132492065, "learning_rate": 0.0001, "elbo": 3.62252, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4680389397138165}
{"step": 42090, "loss": 3.6206806182861326, "grad_norm": 0.7804222106933594, "learning_rate": 0.0001, "elbo": 3.62068, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.468332160450387}
{"step": 42095, "loss": 3.62972412109375, "grad_norm": 0.7666471004486084, "learning_rate": 0.0001, "elbo": 3.62972, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4686253811869574}
{"step": 42100, "loss": 3.6265548706054687, "grad_norm": 0.6655909419059753, "learning_rate": 0.0001, "elbo": 3.62655, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.468918601923528}
{"step": 42105, "loss": 3.620381164550781, "grad_norm": 0.6872232556343079, "learning_rate": 0.0001, "elbo": 3.62038, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4692118226600988}
{"step": 42110, "loss": 3.628157043457031, "grad_norm": 0.7035824656486511, "learning_rate": 0.0001, "elbo": 3.62816, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.469505043396669}
{"step": 42115, "loss": 3.6354904174804688, "grad_norm": 0.735990583896637, "learning_rate": 0.0001, "elbo": 3.63549, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4697982641332397}
{"step": 42120, "loss": 3.620524597167969, "grad_norm": 0.713803231716156, "learning_rate": 0.0001, "elbo": 3.62052, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.47009148486981}
{"step": 42125, "loss": 3.6228912353515623, "grad_norm": 0.6712144017219543, "learning_rate": 0.0001, "elbo": 3.62289, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4703847056063806}
{"step": 42130, "loss": 3.6360137939453123, "grad_norm": 0.7386956810951233, "learning_rate": 0.0001, "elbo": 3.63601, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.470677926342951}
{"step": 42135, "loss": 3.62220458984375, "grad_norm": 0.7009434103965759, "learning_rate": 0.0001, "elbo": 3.6222, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4709711470795215}
{"step": 42140, "loss": 3.6220413208007813, "grad_norm": 0.6375516057014465, "learning_rate": 0.0001, "elbo": 3.62204, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.471264367816092}
{"step": 42145, "loss": 3.6162017822265624, "grad_norm": 0.7442579865455627, "learning_rate": 0.0001, "elbo": 3.6162, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4715575885526624}
{"step": 42150, "loss": 3.6327423095703124, "grad_norm": 0.6757152080535889, "learning_rate": 0.0001, "elbo": 3.63274, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.471850809289233}
{"step": 42155, "loss": 3.6217014312744142, "grad_norm": 0.6398109793663025, "learning_rate": 0.0001, "elbo": 3.6217, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4721440300258033}
{"step": 42160, "loss": 3.6220466613769533, "grad_norm": 0.6604061722755432, "learning_rate": 0.0001, "elbo": 3.62205, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4724372507623738}
{"step": 42165, "loss": 3.6063037872314454, "grad_norm": 0.6775334477424622, "learning_rate": 0.0001, "elbo": 3.6063, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.472730471498944}
{"step": 42170, "loss": 3.6300937652587892, "grad_norm": 0.709552526473999, "learning_rate": 0.0001, "elbo": 3.63009, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4730236922355147}
{"step": 42175, "loss": 3.6212184906005858, "grad_norm": 0.6873471140861511, "learning_rate": 0.0001, "elbo": 3.62122, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4733169129720856}
{"step": 42180, "loss": 3.621477508544922, "grad_norm": 0.7128477096557617, "learning_rate": 0.0001, "elbo": 3.62148, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.473610133708656}
{"step": 42185, "loss": 3.6228076934814455, "grad_norm": 0.6922131180763245, "learning_rate": 0.0001, "elbo": 3.62281, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4739033544452265}
{"step": 42190, "loss": 3.60296516418457, "grad_norm": 0.7979673743247986, "learning_rate": 0.0001, "elbo": 3.60297, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.474196575181797}
{"step": 42190, "loss": 3.60296516418457, "grad_norm": 0.7979673743247986, "learning_rate": 0.0001, "elbo": 3.60297, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.474196575181797}
{"step": 42195, "loss": 3.6051376342773436, "grad_norm": 0.7035157084465027, "learning_rate": 0.0001, "elbo": 3.60514, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4744897959183674}
{"step": 42200, "loss": 3.6130912780761717, "grad_norm": 0.7075778245925903, "learning_rate": 0.0001, "elbo": 3.61309, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.474783016654938}
{"step": 42205, "loss": 3.616289520263672, "grad_norm": 0.6573383212089539, "learning_rate": 0.0001, "elbo": 3.61629, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4750762373915083}
{"step": 42210, "loss": 3.617737579345703, "grad_norm": 0.6668033599853516, "learning_rate": 0.0001, "elbo": 3.61774, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4753694581280787}
{"step": 42215, "loss": 3.6223876953125, "grad_norm": 0.6873071193695068, "learning_rate": 0.0001, "elbo": 3.62239, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.475662678864649}
{"step": 42220, "loss": 3.6344749450683596, "grad_norm": 0.68746018409729, "learning_rate": 0.0001, "elbo": 3.63448, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4759558996012196}
{"step": 42225, "loss": 3.631151580810547, "grad_norm": 0.648249626159668, "learning_rate": 0.0001, "elbo": 3.63115, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.47624912033779}
{"step": 42230, "loss": 3.6075836181640626, "grad_norm": 0.7215589880943298, "learning_rate": 0.0001, "elbo": 3.60758, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.476542341074361}
{"step": 42235, "loss": 3.6191253662109375, "grad_norm": 0.6597057580947876, "learning_rate": 0.0001, "elbo": 3.61913, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4768355618109315}
{"step": 42240, "loss": 3.619081497192383, "grad_norm": 0.6511002779006958, "learning_rate": 0.0001, "elbo": 3.61908, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.477128782547502}
{"step": 42245, "loss": 3.610382080078125, "grad_norm": 1.1275427341461182, "learning_rate": 0.0001, "elbo": 3.61038, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4774220032840724}
{"step": 42250, "loss": 3.6146854400634765, "grad_norm": 0.7015336155891418, "learning_rate": 0.0001, "elbo": 3.61469, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.477715224020643}
{"step": 42255, "loss": 3.6081748962402345, "grad_norm": 0.6824316382408142, "learning_rate": 0.0001, "elbo": 3.60818, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4780084447572133}
{"step": 42260, "loss": 3.6304294586181642, "grad_norm": 1.0052458047866821, "learning_rate": 0.0001, "elbo": 3.63043, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4783016654937837}
{"step": 42265, "loss": 3.6288902282714846, "grad_norm": 0.7306851744651794, "learning_rate": 0.0001, "elbo": 3.62889, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.478594886230354}
{"step": 42270, "loss": 3.6234588623046875, "grad_norm": 0.6708774566650391, "learning_rate": 0.0001, "elbo": 3.62346, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4788881069669246}
{"step": 42275, "loss": 3.6452354431152343, "grad_norm": 0.6580170392990112, "learning_rate": 0.0001, "elbo": 3.64524, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.479181327703495}
{"step": 42280, "loss": 3.6143760681152344, "grad_norm": 0.7151080369949341, "learning_rate": 0.0001, "elbo": 3.61438, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4794745484400655}
{"step": 42285, "loss": 3.621104431152344, "grad_norm": 0.713408887386322, "learning_rate": 0.0001, "elbo": 3.6211, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.479767769176636}
{"step": 42290, "loss": 3.6149131774902346, "grad_norm": 0.687950849533081, "learning_rate": 0.0001, "elbo": 3.61491, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4800609899132064}
{"step": 42295, "loss": 3.615175247192383, "grad_norm": 0.726987898349762, "learning_rate": 0.0001, "elbo": 3.61518, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.480354210649777}
{"step": 42300, "loss": 3.6203071594238283, "grad_norm": 0.7025930881500244, "learning_rate": 0.0001, "elbo": 3.62031, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.480647431386348}
{"step": 42305, "loss": 3.6198944091796874, "grad_norm": 0.6787365078926086, "learning_rate": 0.0001, "elbo": 3.61989, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4809406521229183}
{"step": 42310, "loss": 3.6214614868164063, "grad_norm": 0.7438610792160034, "learning_rate": 0.0001, "elbo": 3.62146, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4812338728594887}
{"step": 42315, "loss": 3.6200523376464844, "grad_norm": 0.9301832914352417, "learning_rate": 0.0001, "elbo": 3.62005, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.481527093596059}
{"step": 42320, "loss": 3.6182350158691405, "grad_norm": 0.6579771637916565, "learning_rate": 0.0001, "elbo": 3.61823, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4818203143326296}
{"step": 42325, "loss": 3.6069591522216795, "grad_norm": 0.6148269176483154, "learning_rate": 0.0001, "elbo": 3.60696, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4821135350692}
{"step": 42330, "loss": 3.5941116333007814, "grad_norm": 0.6655243635177612, "learning_rate": 0.0001, "elbo": 3.59411, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4824067558057705}
{"step": 42335, "loss": 3.62979736328125, "grad_norm": 0.7028188109397888, "learning_rate": 0.0001, "elbo": 3.6298, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.482699976542341}
{"step": 42340, "loss": 3.6012557983398437, "grad_norm": 0.6507661938667297, "learning_rate": 0.0001, "elbo": 3.60126, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4829931972789114}
{"step": 42345, "loss": 3.625440979003906, "grad_norm": 0.6455940008163452, "learning_rate": 0.0001, "elbo": 3.62544, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.483286418015482}
{"step": 42350, "loss": 3.6235210418701174, "grad_norm": 0.640645444393158, "learning_rate": 0.0001, "elbo": 3.62352, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.483579638752053}
{"step": 42355, "loss": 3.6233734130859374, "grad_norm": 0.7019432187080383, "learning_rate": 0.0001, "elbo": 3.62337, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4838728594886232}
{"step": 42360, "loss": 3.6150390625, "grad_norm": 0.7313152551651001, "learning_rate": 0.0001, "elbo": 3.61504, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4841660802251937}
{"step": 42365, "loss": 3.624988555908203, "grad_norm": 0.7020329236984253, "learning_rate": 0.0001, "elbo": 3.62499, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.484459300961764}
{"step": 42370, "loss": 3.6263137817382813, "grad_norm": 0.7253475189208984, "learning_rate": 0.0001, "elbo": 3.62631, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4847525216983346}
{"step": 42375, "loss": 3.6277565002441405, "grad_norm": 0.6742516756057739, "learning_rate": 0.0001, "elbo": 3.62776, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.485045742434905}
{"step": 42380, "loss": 3.618233489990234, "grad_norm": 0.7318021655082703, "learning_rate": 0.0001, "elbo": 3.61823, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4853389631714755}
{"step": 42380, "loss": 3.618233489990234, "grad_norm": 0.7318021655082703, "learning_rate": 0.0001, "elbo": 3.61823, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4853389631714755}
{"step": 42385, "loss": 3.6036014556884766, "grad_norm": 0.7479944229125977, "learning_rate": 0.0001, "elbo": 3.6036, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.485632183908046}
{"step": 42390, "loss": 3.6083610534667967, "grad_norm": 0.6817395687103271, "learning_rate": 0.0001, "elbo": 3.60836, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4859254046446164}
{"step": 42395, "loss": 3.618182373046875, "grad_norm": 0.7121917605400085, "learning_rate": 0.0001, "elbo": 3.61818, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.486218625381187}
{"step": 42400, "loss": 3.63110237121582, "grad_norm": 0.663652241230011, "learning_rate": 0.0001, "elbo": 3.6311, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4865118461177573}
{"step": 42405, "loss": 3.6309494018554687, "grad_norm": 0.7701805233955383, "learning_rate": 0.0001, "elbo": 3.63095, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.486805066854328}
{"step": 42410, "loss": 3.598444366455078, "grad_norm": 0.6747052073478699, "learning_rate": 0.0001, "elbo": 3.59844, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4870982875908982}
{"step": 42415, "loss": 3.626536560058594, "grad_norm": 0.802259087562561, "learning_rate": 0.0001, "elbo": 3.62654, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4873915083274687}
{"step": 42415, "loss": 3.626536560058594, "grad_norm": 0.802259087562561, "learning_rate": 0.0001, "elbo": 3.62654, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4873915083274687}
{"step": 42420, "loss": 3.62664794921875, "grad_norm": 0.6852491497993469, "learning_rate": 0.0001, "elbo": 3.62665, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4876847290640396}
{"step": 42425, "loss": 3.609263610839844, "grad_norm": 0.8852055668830872, "learning_rate": 0.0001, "elbo": 3.60926, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.48797794980061}
{"step": 42430, "loss": 3.6293479919433596, "grad_norm": 0.6812976002693176, "learning_rate": 0.0001, "elbo": 3.62935, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4882711705371805}
{"step": 42435, "loss": 3.5934120178222657, "grad_norm": 0.7487099766731262, "learning_rate": 0.0001, "elbo": 3.59341, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.488564391273751}
{"step": 42440, "loss": 3.618155670166016, "grad_norm": 0.7268282771110535, "learning_rate": 0.0001, "elbo": 3.61816, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4888576120103214}
{"step": 42445, "loss": 3.619264602661133, "grad_norm": 0.6919053196907043, "learning_rate": 0.0001, "elbo": 3.61926, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.489150832746892}
{"step": 42450, "loss": 3.614129638671875, "grad_norm": 0.6888521313667297, "learning_rate": 0.0001, "elbo": 3.61413, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4894440534834623}
{"step": 42455, "loss": 3.622611236572266, "grad_norm": 0.6404711008071899, "learning_rate": 0.0001, "elbo": 3.62261, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4897372742200328}
{"step": 42460, "loss": 3.627494049072266, "grad_norm": 0.6758397221565247, "learning_rate": 0.0001, "elbo": 3.62749, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4900304949566032}
{"step": 42465, "loss": 3.6077117919921875, "grad_norm": 0.6990126967430115, "learning_rate": 0.0001, "elbo": 3.60771, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4903237156931737}
{"step": 42465, "loss": 3.6077117919921875, "grad_norm": 0.6990126967430115, "learning_rate": 0.0001, "elbo": 3.60771, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4903237156931737}
{"step": 42470, "loss": 3.6370277404785156, "grad_norm": 0.8442471623420715, "learning_rate": 0.0001, "elbo": 3.63703, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.490616936429744}
{"step": 42475, "loss": 3.62303466796875, "grad_norm": 1.297064185142517, "learning_rate": 0.0001, "elbo": 3.62303, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.490910157166315}
{"step": 42480, "loss": 3.6102561950683594, "grad_norm": 0.8040929436683655, "learning_rate": 0.0001, "elbo": 3.61026, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4912033779028855}
{"step": 42485, "loss": 3.623153305053711, "grad_norm": 0.6871998310089111, "learning_rate": 0.0001, "elbo": 3.62315, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.491496598639456}
{"step": 42490, "loss": 3.624555969238281, "grad_norm": 0.6604203581809998, "learning_rate": 0.0001, "elbo": 3.62456, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4917898193760264}
{"step": 42490, "loss": 3.624555969238281, "grad_norm": 0.6604203581809998, "learning_rate": 0.0001, "elbo": 3.62456, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4917898193760264}
{"step": 42495, "loss": 3.616110992431641, "grad_norm": 0.7408738732337952, "learning_rate": 0.0001, "elbo": 3.61611, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.492083040112597}
{"step": 42500, "loss": 3.6239593505859373, "grad_norm": 0.6914641857147217, "learning_rate": 0.0001, "elbo": 3.62396, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4923762608491673}
{"step": 42505, "loss": 3.625191879272461, "grad_norm": 0.6939945220947266, "learning_rate": 0.0001, "elbo": 3.62519, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4926694815857378}
{"step": 42510, "loss": 3.5976524353027344, "grad_norm": 0.7454419136047363, "learning_rate": 0.0001, "elbo": 3.59765, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.492962702322308}
{"step": 42515, "loss": 3.6169662475585938, "grad_norm": 0.7323576807975769, "learning_rate": 0.0001, "elbo": 3.61697, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4932559230588787}
{"step": 42520, "loss": 3.6063297271728514, "grad_norm": 0.7727869749069214, "learning_rate": 0.0001, "elbo": 3.60633, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.493549143795449}
{"step": 42525, "loss": 3.6264556884765624, "grad_norm": 0.7044947743415833, "learning_rate": 0.0001, "elbo": 3.62646, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4938423645320196}
{"step": 42530, "loss": 3.6186195373535157, "grad_norm": 0.7867666482925415, "learning_rate": 0.0001, "elbo": 3.61862, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.49413558526859}
{"step": 42535, "loss": 3.6258499145507814, "grad_norm": 0.7071660757064819, "learning_rate": 0.0001, "elbo": 3.62585, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4944288060051605}
{"step": 42540, "loss": 3.6429187774658205, "grad_norm": 0.7089381814002991, "learning_rate": 0.0001, "elbo": 3.64292, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.494722026741731}
{"step": 42540, "loss": 3.6429187774658205, "grad_norm": 0.7089381814002991, "learning_rate": 0.0001, "elbo": 3.64292, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.494722026741731}
{"step": 42545, "loss": 3.6061073303222657, "grad_norm": 0.7333744764328003, "learning_rate": 0.0001, "elbo": 3.60611, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.495015247478302}
{"step": 42550, "loss": 3.6214778900146483, "grad_norm": 0.7016448974609375, "learning_rate": 0.0001, "elbo": 3.62148, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4953084682148723}
{"step": 42555, "loss": 3.61497802734375, "grad_norm": 0.6708899140357971, "learning_rate": 0.0001, "elbo": 3.61498, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4956016889514427}
{"step": 42560, "loss": 3.618681716918945, "grad_norm": 1.2216616868972778, "learning_rate": 0.0001, "elbo": 3.61868, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.495894909688013}
{"step": 42565, "loss": 3.6097991943359373, "grad_norm": 0.724958062171936, "learning_rate": 0.0001, "elbo": 3.6098, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4961881304245837}
{"step": 42570, "loss": 3.625994873046875, "grad_norm": 0.7084938287734985, "learning_rate": 0.0001, "elbo": 3.626, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.496481351161154}
{"step": 42575, "loss": 3.6240142822265624, "grad_norm": 0.7018274664878845, "learning_rate": 0.0001, "elbo": 3.62401, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4967745718977246}
{"step": 42580, "loss": 3.6193687438964846, "grad_norm": 0.6935844421386719, "learning_rate": 0.0001, "elbo": 3.61937, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.497067792634295}
{"step": 42585, "loss": 3.6112537384033203, "grad_norm": 0.6725033521652222, "learning_rate": 0.0001, "elbo": 3.61125, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4973610133708655}
{"step": 42590, "loss": 3.6155067443847657, "grad_norm": 0.9618467688560486, "learning_rate": 0.0001, "elbo": 3.61551, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.497654234107436}
{"step": 42595, "loss": 3.6357799530029298, "grad_norm": 0.698932945728302, "learning_rate": 0.0001, "elbo": 3.63578, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4979474548440064}
{"step": 42600, "loss": 3.627271270751953, "grad_norm": 0.6991897225379944, "learning_rate": 0.0001, "elbo": 3.62727, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4982406755805773}
{"step": 42605, "loss": 3.631582260131836, "grad_norm": 0.6720477938652039, "learning_rate": 0.0001, "elbo": 3.63158, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4985338963171477}
{"step": 42610, "loss": 3.612930679321289, "grad_norm": 0.6725927591323853, "learning_rate": 0.0001, "elbo": 3.61293, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.498827117053718}
{"step": 42615, "loss": 3.640180969238281, "grad_norm": 0.6630305051803589, "learning_rate": 0.0001, "elbo": 3.64018, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4991203377902886}
{"step": 42620, "loss": 3.6224945068359373, "grad_norm": 0.7009799480438232, "learning_rate": 0.0001, "elbo": 3.62249, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.499413558526859}
{"step": 42625, "loss": 3.6139347076416017, "grad_norm": 0.8531689047813416, "learning_rate": 0.0001, "elbo": 3.61393, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.4997067792634295}
{"step": 42630, "loss": 3.6243865966796873, "grad_norm": 0.6692758202552795, "learning_rate": 0.0001, "elbo": 3.62439, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5}
{"step": 42635, "loss": 3.6117515563964844, "grad_norm": 0.7088223695755005, "learning_rate": 0.0001, "elbo": 3.61175, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5002932207365705}
{"step": 42640, "loss": 3.6095115661621096, "grad_norm": 0.7310193181037903, "learning_rate": 0.0001, "elbo": 3.60951, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.500586441473141}
{"step": 42645, "loss": 3.6303367614746094, "grad_norm": 0.7069794535636902, "learning_rate": 0.0001, "elbo": 3.63034, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5008796622097114}
{"step": 42650, "loss": 3.6007125854492186, "grad_norm": 0.7024447321891785, "learning_rate": 0.0001, "elbo": 3.60071, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.501172882946282}
{"step": 42655, "loss": 3.6147109985351564, "grad_norm": 0.6865729093551636, "learning_rate": 0.0001, "elbo": 3.61471, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5014661036828523}
{"step": 42660, "loss": 3.6151969909667967, "grad_norm": 0.6950265765190125, "learning_rate": 0.0001, "elbo": 3.6152, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5017593244194227}
{"step": 42665, "loss": 3.596875, "grad_norm": 0.680846095085144, "learning_rate": 0.0001, "elbo": 3.59688, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.502052545155993}
{"step": 42670, "loss": 3.615604019165039, "grad_norm": 0.9308932423591614, "learning_rate": 0.0001, "elbo": 3.6156, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.502345765892564}
{"step": 42675, "loss": 3.6261104583740233, "grad_norm": 0.7844762802124023, "learning_rate": 0.0001, "elbo": 3.62611, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5026389866291345}
{"step": 42680, "loss": 3.6410274505615234, "grad_norm": 0.7301039099693298, "learning_rate": 0.0001, "elbo": 3.64103, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.502932207365705}
{"step": 42685, "loss": 3.628926467895508, "grad_norm": 0.904155433177948, "learning_rate": 0.0001, "elbo": 3.62893, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5032254281022754}
{"step": 42690, "loss": 3.614552307128906, "grad_norm": 0.7218322157859802, "learning_rate": 0.0001, "elbo": 3.61455, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.503518648838846}
{"step": 42690, "loss": 3.614552307128906, "grad_norm": 0.7218322157859802, "learning_rate": 0.0001, "elbo": 3.61455, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.503518648838846}
{"step": 42695, "loss": 3.6062713623046876, "grad_norm": 0.7158499360084534, "learning_rate": 0.0001, "elbo": 3.60627, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5038118695754163}
{"step": 42700, "loss": 3.613625335693359, "grad_norm": 0.7342904210090637, "learning_rate": 0.0001, "elbo": 3.61363, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.504105090311987}
{"step": 42705, "loss": 3.6231704711914063, "grad_norm": 0.6628641486167908, "learning_rate": 0.0001, "elbo": 3.62317, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5043983110485573}
{"step": 42710, "loss": 3.6127113342285155, "grad_norm": 0.7132890820503235, "learning_rate": 0.0001, "elbo": 3.61271, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5046915317851277}
{"step": 42715, "loss": 3.610103225708008, "grad_norm": 0.6677153706550598, "learning_rate": 0.0001, "elbo": 3.6101, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5049847525216986}
{"step": 42720, "loss": 3.6234058380126952, "grad_norm": 0.6776543259620667, "learning_rate": 0.0001, "elbo": 3.62341, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.505277973258269}
{"step": 42725, "loss": 3.6113422393798826, "grad_norm": 0.7123213410377502, "learning_rate": 0.0001, "elbo": 3.61134, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5055711939948395}
{"step": 42730, "loss": 3.620017242431641, "grad_norm": 0.6695423126220703, "learning_rate": 0.0001, "elbo": 3.62002, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.50586441473141}
{"step": 42735, "loss": 3.612833023071289, "grad_norm": 0.6870044469833374, "learning_rate": 0.0001, "elbo": 3.61283, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5061576354679804}
{"step": 42740, "loss": 3.613141632080078, "grad_norm": 0.7511904239654541, "learning_rate": 0.0001, "elbo": 3.61314, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.506450856204551}
{"step": 42745, "loss": 3.6070343017578126, "grad_norm": 0.6788884401321411, "learning_rate": 0.0001, "elbo": 3.60703, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5067440769411213}
{"step": 42750, "loss": 3.628437042236328, "grad_norm": 0.6628586053848267, "learning_rate": 0.0001, "elbo": 3.62844, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.507037297677692}
{"step": 42755, "loss": 3.6151096343994142, "grad_norm": 0.7296895384788513, "learning_rate": 0.0001, "elbo": 3.61511, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5073305184142622}
{"step": 42760, "loss": 3.619429016113281, "grad_norm": 0.7444674372673035, "learning_rate": 0.0001, "elbo": 3.61943, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5076237391508327}
{"step": 42765, "loss": 3.6177955627441407, "grad_norm": 0.90706866979599, "learning_rate": 0.0001, "elbo": 3.6178, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.507916959887403}
{"step": 42770, "loss": 3.609892654418945, "grad_norm": 0.7544078826904297, "learning_rate": 0.0001, "elbo": 3.60989, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5082101806239736}
{"step": 42775, "loss": 3.6180004119873046, "grad_norm": 0.6578720211982727, "learning_rate": 0.0001, "elbo": 3.618, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.508503401360544}
{"step": 42780, "loss": 3.6086402893066407, "grad_norm": 0.7007302045822144, "learning_rate": 0.0001, "elbo": 3.60864, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5087966220971145}
{"step": 42785, "loss": 3.6173980712890623, "grad_norm": 0.7657049298286438, "learning_rate": 0.0001, "elbo": 3.6174, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.509089842833685}
{"step": 42790, "loss": 3.6124534606933594, "grad_norm": 0.6530637145042419, "learning_rate": 0.0001, "elbo": 3.61245, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5093830635702554}
{"step": 42795, "loss": 3.6180374145507814, "grad_norm": 0.7004495859146118, "learning_rate": 0.0001, "elbo": 3.61804, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5096762843068263}
{"step": 42800, "loss": 3.604911041259766, "grad_norm": 0.7270635962486267, "learning_rate": 0.0001, "elbo": 3.60491, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5099695050433968}
{"step": 42800, "loss": 3.604911041259766, "grad_norm": 0.7270635962486267, "learning_rate": 0.0001, "elbo": 3.60491, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5099695050433968}
{"step": 42805, "loss": 3.623332977294922, "grad_norm": 0.7160264253616333, "learning_rate": 0.0001, "elbo": 3.62333, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5102627257799672}
{"step": 42810, "loss": 3.608418273925781, "grad_norm": 1.0034769773483276, "learning_rate": 0.0001, "elbo": 3.60842, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5105559465165377}
{"step": 42815, "loss": 3.605635070800781, "grad_norm": 0.694446861743927, "learning_rate": 0.0001, "elbo": 3.60564, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.510849167253108}
{"step": 42820, "loss": 3.605712127685547, "grad_norm": 1.1533015966415405, "learning_rate": 0.0001, "elbo": 3.60571, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5111423879896786}
{"step": 42825, "loss": 3.634092330932617, "grad_norm": 0.7468109726905823, "learning_rate": 0.0001, "elbo": 3.63409, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.511435608726249}
{"step": 42830, "loss": 3.6123504638671875, "grad_norm": 0.6863083839416504, "learning_rate": 0.0001, "elbo": 3.61235, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5117288294628195}
{"step": 42835, "loss": 3.617992401123047, "grad_norm": 0.7487635612487793, "learning_rate": 0.0001, "elbo": 3.61799, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.51202205019939}
{"step": 42840, "loss": 3.609745407104492, "grad_norm": 0.8369210362434387, "learning_rate": 0.0001, "elbo": 3.60975, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.512315270935961}
{"step": 42845, "loss": 3.6254100799560547, "grad_norm": 0.7021980285644531, "learning_rate": 0.0001, "elbo": 3.62541, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5126084916725313}
{"step": 42850, "loss": 3.6301170349121095, "grad_norm": 0.6874083876609802, "learning_rate": 0.0001, "elbo": 3.63012, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5129017124091018}
{"step": 42855, "loss": 3.611970901489258, "grad_norm": 0.6852298974990845, "learning_rate": 0.0001, "elbo": 3.61197, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.513194933145672}
{"step": 42860, "loss": 3.613373565673828, "grad_norm": 0.671266496181488, "learning_rate": 0.0001, "elbo": 3.61337, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5134881538822427}
{"step": 42865, "loss": 3.624903106689453, "grad_norm": 0.970312774181366, "learning_rate": 0.0001, "elbo": 3.6249, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.513781374618813}
{"step": 42870, "loss": 3.606090545654297, "grad_norm": 0.6763694882392883, "learning_rate": 0.0001, "elbo": 3.60609, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5140745953553836}
{"step": 42875, "loss": 3.6158016204833983, "grad_norm": 0.7389115691184998, "learning_rate": 0.0001, "elbo": 3.6158, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.514367816091954}
{"step": 42880, "loss": 3.6152591705322266, "grad_norm": 0.7391152381896973, "learning_rate": 0.0001, "elbo": 3.61526, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5146610368285245}
{"step": 42885, "loss": 3.6173751831054686, "grad_norm": 0.7551981210708618, "learning_rate": 0.0001, "elbo": 3.61738, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.514954257565095}
{"step": 42890, "loss": 3.6146514892578123, "grad_norm": 0.72063148021698, "learning_rate": 0.0001, "elbo": 3.61465, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5152474783016654}
{"step": 42895, "loss": 3.614881896972656, "grad_norm": 0.6847013235092163, "learning_rate": 0.0001, "elbo": 3.61488, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.515540699038236}
{"step": 42900, "loss": 3.6323692321777346, "grad_norm": 0.6312847137451172, "learning_rate": 0.0001, "elbo": 3.63237, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5158339197748063}
{"step": 42905, "loss": 3.6273433685302736, "grad_norm": 0.6380859613418579, "learning_rate": 0.0001, "elbo": 3.62734, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5161271405113768}
{"step": 42910, "loss": 3.602366638183594, "grad_norm": 0.6788354516029358, "learning_rate": 0.0001, "elbo": 3.60237, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.516420361247947}
{"step": 42915, "loss": 3.6239227294921874, "grad_norm": 0.6717735528945923, "learning_rate": 0.0001, "elbo": 3.62392, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.516713581984518}
{"step": 42920, "loss": 3.6185733795166017, "grad_norm": 0.6359084844589233, "learning_rate": 0.0001, "elbo": 3.61857, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5170068027210886}
{"step": 42925, "loss": 3.632315826416016, "grad_norm": 0.7045968174934387, "learning_rate": 0.0001, "elbo": 3.63232, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.517300023457659}
{"step": 42930, "loss": 3.616751480102539, "grad_norm": 0.6755878925323486, "learning_rate": 0.0001, "elbo": 3.61675, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5175932441942295}
{"step": 42935, "loss": 3.6041633605957033, "grad_norm": 0.7470546364784241, "learning_rate": 0.0001, "elbo": 3.60416, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5178864649308}
{"step": 42940, "loss": 3.62252197265625, "grad_norm": 0.6379334926605225, "learning_rate": 0.0001, "elbo": 3.62252, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5181796856673704}
{"step": 42945, "loss": 3.6212322235107424, "grad_norm": 0.6774534583091736, "learning_rate": 0.0001, "elbo": 3.62123, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.518472906403941}
{"step": 42950, "loss": 3.6265548706054687, "grad_norm": 0.9156293272972107, "learning_rate": 0.0001, "elbo": 3.62655, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5187661271405113}
{"step": 42950, "loss": 3.6265548706054687, "grad_norm": 0.9156293272972107, "learning_rate": 0.0001, "elbo": 3.62655, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5187661271405113}
{"step": 42955, "loss": 3.618619918823242, "grad_norm": 0.7281374335289001, "learning_rate": 0.0001, "elbo": 3.61862, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5190593478770817}
{"step": 42960, "loss": 3.633927917480469, "grad_norm": 0.9610598683357239, "learning_rate": 0.0001, "elbo": 3.63393, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.519352568613652}
{"step": 42965, "loss": 3.6115188598632812, "grad_norm": 0.69696444272995, "learning_rate": 0.0001, "elbo": 3.61152, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.519645789350223}
{"step": 42970, "loss": 3.6257667541503906, "grad_norm": 0.8437926769256592, "learning_rate": 0.0001, "elbo": 3.62577, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5199390100867936}
{"step": 42975, "loss": 3.6322467803955076, "grad_norm": 0.6686816215515137, "learning_rate": 0.0001, "elbo": 3.63225, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.520232230823364}
{"step": 42980, "loss": 3.6123199462890625, "grad_norm": 0.6681950688362122, "learning_rate": 0.0001, "elbo": 3.61232, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5205254515599345}
{"step": 42985, "loss": 3.6145668029785156, "grad_norm": 0.6520650386810303, "learning_rate": 0.0001, "elbo": 3.61457, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.520818672296505}
{"step": 42990, "loss": 3.627385711669922, "grad_norm": 0.8514606952667236, "learning_rate": 0.0001, "elbo": 3.62739, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5211118930330754}
{"step": 42995, "loss": 3.618760681152344, "grad_norm": 0.684162437915802, "learning_rate": 0.0001, "elbo": 3.61876, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.521405113769646}
{"step": 43000, "loss": 3.6162345886230467, "grad_norm": 1.1320124864578247, "learning_rate": 0.0001, "elbo": 3.61623, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5216983345062163}
{"step": 43000, "loss": 3.6162345886230467, "grad_norm": 1.1320124864578247, "learning_rate": 0.0001, "elbo": 3.61623, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5216983345062163}
{"step": 43005, "loss": 3.608929443359375, "grad_norm": 0.7041214108467102, "learning_rate": 0.0001, "elbo": 3.60893, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5219915552427867}
{"step": 43010, "loss": 3.606606292724609, "grad_norm": 0.6876093745231628, "learning_rate": 0.0001, "elbo": 3.60661, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.522284775979357}
{"step": 43010, "loss": 3.606606292724609, "grad_norm": 0.6876093745231628, "learning_rate": 0.0001, "elbo": 3.60661, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.522284775979357}
{"step": 43015, "loss": 3.609600067138672, "grad_norm": 0.7346994280815125, "learning_rate": 0.0001, "elbo": 3.6096, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5225779967159276}
{"step": 43020, "loss": 3.6152320861816407, "grad_norm": 0.6892678141593933, "learning_rate": 0.0001, "elbo": 3.61523, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.522871217452498}
{"step": 43025, "loss": 3.6309673309326174, "grad_norm": 0.6482776403427124, "learning_rate": 0.0001, "elbo": 3.63097, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5231644381890685}
{"step": 43030, "loss": 3.6132450103759766, "grad_norm": 0.6533750891685486, "learning_rate": 0.0001, "elbo": 3.61325, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.523457658925639}
{"step": 43035, "loss": 3.6150978088378904, "grad_norm": 0.6859584450721741, "learning_rate": 0.0001, "elbo": 3.6151, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5237508796622095}
{"step": 43040, "loss": 3.6136741638183594, "grad_norm": 0.6783329248428345, "learning_rate": 0.0001, "elbo": 3.61367, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5240441003987804}
{"step": 43045, "loss": 3.6148853302001953, "grad_norm": 0.6376718282699585, "learning_rate": 0.0001, "elbo": 3.61489, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.524337321135351}
{"step": 43050, "loss": 3.6114166259765623, "grad_norm": 0.6404396295547485, "learning_rate": 0.0001, "elbo": 3.61142, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5246305418719213}
{"step": 43055, "loss": 3.6165496826171877, "grad_norm": 1.1067334413528442, "learning_rate": 0.0001, "elbo": 3.61655, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5249237626084917}
{"step": 43060, "loss": 3.640569305419922, "grad_norm": 0.6790252923965454, "learning_rate": 0.0001, "elbo": 3.64057, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.525216983345062}
{"step": 43065, "loss": 3.6225730895996096, "grad_norm": 0.7225849628448486, "learning_rate": 0.0001, "elbo": 3.62257, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5255102040816326}
{"step": 43070, "loss": 3.632522201538086, "grad_norm": 0.6985569596290588, "learning_rate": 0.0001, "elbo": 3.63252, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.525803424818203}
{"step": 43075, "loss": 3.6334556579589843, "grad_norm": 0.7070730924606323, "learning_rate": 0.0001, "elbo": 3.63346, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5260966455547735}
{"step": 43080, "loss": 3.620303726196289, "grad_norm": 0.748195230960846, "learning_rate": 0.0001, "elbo": 3.6203, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.526389866291344}
{"step": 43085, "loss": 3.6239559173583986, "grad_norm": 0.6963062286376953, "learning_rate": 0.0001, "elbo": 3.62396, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.526683087027915}
{"step": 43090, "loss": 3.6059104919433596, "grad_norm": 0.6712691187858582, "learning_rate": 0.0001, "elbo": 3.60591, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5269763077644853}
{"step": 43095, "loss": 3.6026687622070312, "grad_norm": 0.6760116815567017, "learning_rate": 0.0001, "elbo": 3.60267, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.527269528501056}
{"step": 43100, "loss": 3.6012733459472654, "grad_norm": 0.7242763042449951, "learning_rate": 0.0001, "elbo": 3.60127, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5275627492376262}
{"step": 43105, "loss": 3.6277305603027346, "grad_norm": 0.6649040579795837, "learning_rate": 0.0001, "elbo": 3.62773, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5278559699741967}
{"step": 43110, "loss": 3.6104164123535156, "grad_norm": 0.8149797916412354, "learning_rate": 0.0001, "elbo": 3.61042, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.528149190710767}
{"step": 43115, "loss": 3.6168285369873048, "grad_norm": 0.6486098766326904, "learning_rate": 0.0001, "elbo": 3.61683, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5284424114473376}
{"step": 43120, "loss": 3.6090797424316405, "grad_norm": 0.7336783409118652, "learning_rate": 0.0001, "elbo": 3.60908, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.528735632183908}
{"step": 43125, "loss": 3.60228271484375, "grad_norm": 0.6869011521339417, "learning_rate": 0.0001, "elbo": 3.60228, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5290288529204785}
{"step": 43130, "loss": 3.6230209350585936, "grad_norm": 0.6951670050621033, "learning_rate": 0.0001, "elbo": 3.62302, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.529322073657049}
{"step": 43135, "loss": 3.6114166259765623, "grad_norm": 0.6701415777206421, "learning_rate": 0.0001, "elbo": 3.61142, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5296152943936194}
{"step": 43140, "loss": 3.6033538818359374, "grad_norm": 0.6809455156326294, "learning_rate": 0.0001, "elbo": 3.60335, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.52990851513019}
{"step": 43145, "loss": 3.6111175537109377, "grad_norm": 1.1778725385665894, "learning_rate": 0.0001, "elbo": 3.61112, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5302017358667603}
{"step": 43150, "loss": 3.6263004302978517, "grad_norm": 0.6989762187004089, "learning_rate": 0.0001, "elbo": 3.6263, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.530494956603331}
{"step": 43155, "loss": 3.6020397186279296, "grad_norm": 0.6595660448074341, "learning_rate": 0.0001, "elbo": 3.60204, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5307881773399012}
{"step": 43160, "loss": 3.6012649536132812, "grad_norm": 0.7251705527305603, "learning_rate": 0.0001, "elbo": 3.60127, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5310813980764717}
{"step": 43165, "loss": 3.624289703369141, "grad_norm": 0.6941347718238831, "learning_rate": 0.0001, "elbo": 3.62429, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5313746188130426}
{"step": 43170, "loss": 3.6226375579833983, "grad_norm": 0.6657572388648987, "learning_rate": 0.0001, "elbo": 3.62264, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.531667839549613}
{"step": 43175, "loss": 3.606060028076172, "grad_norm": 0.6769406795501709, "learning_rate": 0.0001, "elbo": 3.60606, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5319610602861835}
{"step": 43180, "loss": 3.616592025756836, "grad_norm": 0.6882560849189758, "learning_rate": 0.0001, "elbo": 3.61659, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.532254281022754}
{"step": 43185, "loss": 3.607906723022461, "grad_norm": 0.7909504175186157, "learning_rate": 0.0001, "elbo": 3.60791, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5325475017593244}
{"step": 43190, "loss": 3.613470458984375, "grad_norm": 0.6263465881347656, "learning_rate": 0.0001, "elbo": 3.61347, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.532840722495895}
{"step": 43190, "loss": 3.613470458984375, "grad_norm": 0.6263465881347656, "learning_rate": 0.0001, "elbo": 3.61347, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.532840722495895}
{"step": 43195, "loss": 3.635567855834961, "grad_norm": 0.6571148633956909, "learning_rate": 0.0001, "elbo": 3.63557, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5331339432324653}
{"step": 43200, "loss": 3.6236061096191405, "grad_norm": 0.7411925792694092, "learning_rate": 0.0001, "elbo": 3.62361, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.533427163969036}
{"step": 43205, "loss": 3.6251312255859376, "grad_norm": 0.7174962162971497, "learning_rate": 0.0001, "elbo": 3.62513, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5337203847056062}
{"step": 43210, "loss": 3.6223846435546876, "grad_norm": 0.7231742739677429, "learning_rate": 0.0001, "elbo": 3.62238, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.534013605442177}
{"step": 43215, "loss": 3.620381164550781, "grad_norm": 0.7175509333610535, "learning_rate": 0.0001, "elbo": 3.62038, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5343068261787476}
{"step": 43220, "loss": 3.6108634948730467, "grad_norm": 0.7403987646102905, "learning_rate": 0.0001, "elbo": 3.61086, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.534600046915318}
{"step": 43225, "loss": 3.604324722290039, "grad_norm": 0.7170482277870178, "learning_rate": 0.0001, "elbo": 3.60432, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5348932676518885}
{"step": 43230, "loss": 3.6075210571289062, "grad_norm": 0.6498978734016418, "learning_rate": 0.0001, "elbo": 3.60752, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.535186488388459}
{"step": 43235, "loss": 3.607193374633789, "grad_norm": 0.9537961483001709, "learning_rate": 0.0001, "elbo": 3.60719, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5354797091250294}
{"step": 43240, "loss": 3.6167102813720704, "grad_norm": 0.7201230525970459, "learning_rate": 0.0001, "elbo": 3.61671, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5357729298616}
{"step": 43245, "loss": 3.6186920166015626, "grad_norm": 0.6992285251617432, "learning_rate": 0.0001, "elbo": 3.61869, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5360661505981703}
{"step": 43250, "loss": 3.6147796630859377, "grad_norm": 0.808357298374176, "learning_rate": 0.0001, "elbo": 3.61478, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5363593713347408}
{"step": 43255, "loss": 3.6188423156738283, "grad_norm": 0.6588794589042664, "learning_rate": 0.0001, "elbo": 3.61884, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.536652592071311}
{"step": 43260, "loss": 3.603168487548828, "grad_norm": 0.6722248196601868, "learning_rate": 0.0001, "elbo": 3.60317, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5369458128078817}
{"step": 43265, "loss": 3.6104095458984373, "grad_norm": 0.7846760153770447, "learning_rate": 0.0001, "elbo": 3.61041, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.537239033544452}
{"step": 43270, "loss": 3.613085174560547, "grad_norm": 0.6759865283966064, "learning_rate": 0.0001, "elbo": 3.61309, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5375322542810226}
{"step": 43275, "loss": 3.6085960388183596, "grad_norm": 0.8148627281188965, "learning_rate": 0.0001, "elbo": 3.6086, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.537825475017593}
{"step": 43280, "loss": 3.6261341094970705, "grad_norm": 0.7035237550735474, "learning_rate": 0.0001, "elbo": 3.62613, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5381186957541635}
{"step": 43285, "loss": 3.6087791442871096, "grad_norm": 0.7456414699554443, "learning_rate": 0.0001, "elbo": 3.60878, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5384119164907344}
{"step": 43285, "loss": 3.6087791442871096, "grad_norm": 0.7456414699554443, "learning_rate": 0.0001, "elbo": 3.60878, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5384119164907344}
{"step": 43290, "loss": 3.6209922790527345, "grad_norm": 0.6722545027732849, "learning_rate": 0.0001, "elbo": 3.62099, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.538705137227305}
{"step": 43295, "loss": 3.622196578979492, "grad_norm": 0.696056067943573, "learning_rate": 0.0001, "elbo": 3.6222, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5389983579638753}
{"step": 43300, "loss": 3.617262268066406, "grad_norm": 0.7073826193809509, "learning_rate": 0.0001, "elbo": 3.61726, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5392915787004458}
{"step": 43305, "loss": 3.61529541015625, "grad_norm": 0.6796283721923828, "learning_rate": 0.0001, "elbo": 3.6153, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.539584799437016}
{"step": 43310, "loss": 3.629909133911133, "grad_norm": 0.7418596744537354, "learning_rate": 0.0001, "elbo": 3.62991, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5398780201735867}
{"step": 43315, "loss": 3.6061363220214844, "grad_norm": 0.7113375067710876, "learning_rate": 0.0001, "elbo": 3.60614, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.540171240910157}
{"step": 43320, "loss": 3.609984588623047, "grad_norm": 0.7823342084884644, "learning_rate": 0.0001, "elbo": 3.60998, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5404644616467276}
{"step": 43325, "loss": 3.6299057006835938, "grad_norm": 0.6823688745498657, "learning_rate": 0.0001, "elbo": 3.62991, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.540757682383298}
{"step": 43330, "loss": 3.6074790954589844, "grad_norm": 0.750707745552063, "learning_rate": 0.0001, "elbo": 3.60748, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5410509031198685}
{"step": 43335, "loss": 3.618760681152344, "grad_norm": 0.6706207394599915, "learning_rate": 0.0001, "elbo": 3.61876, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5413441238564394}
{"step": 43340, "loss": 3.611977386474609, "grad_norm": 0.730363667011261, "learning_rate": 0.0001, "elbo": 3.61198, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.54163734459301}
{"step": 43345, "loss": 3.615496826171875, "grad_norm": 0.7050435543060303, "learning_rate": 0.0001, "elbo": 3.6155, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5419305653295803}
{"step": 43350, "loss": 3.612260437011719, "grad_norm": 0.6894546747207642, "learning_rate": 0.0001, "elbo": 3.61226, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5422237860661507}
{"step": 43355, "loss": 3.6169921875, "grad_norm": 0.709010124206543, "learning_rate": 0.0001, "elbo": 3.61699, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.542517006802721}
{"step": 43360, "loss": 3.6041095733642576, "grad_norm": 0.6959294080734253, "learning_rate": 0.0001, "elbo": 3.60411, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5428102275392916}
{"step": 43365, "loss": 3.6058067321777343, "grad_norm": 0.6670434474945068, "learning_rate": 0.0001, "elbo": 3.60581, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.543103448275862}
{"step": 43370, "loss": 3.6061573028564453, "grad_norm": 0.7241079211235046, "learning_rate": 0.0001, "elbo": 3.60616, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5433966690124326}
{"step": 43375, "loss": 3.6246688842773436, "grad_norm": 0.7169743180274963, "learning_rate": 0.0001, "elbo": 3.62467, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.543689889749003}
{"step": 43380, "loss": 3.6137359619140623, "grad_norm": 0.6570428609848022, "learning_rate": 0.0001, "elbo": 3.61374, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5439831104855735}
{"step": 43385, "loss": 3.607331085205078, "grad_norm": 0.6916572451591492, "learning_rate": 0.0001, "elbo": 3.60733, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.544276331222144}
{"step": 43390, "loss": 3.6081085205078125, "grad_norm": 0.6820425987243652, "learning_rate": 0.0001, "elbo": 3.60811, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5445695519587144}
{"step": 43395, "loss": 3.601839065551758, "grad_norm": 0.7525967359542847, "learning_rate": 0.0001, "elbo": 3.60184, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.544862772695285}
{"step": 43400, "loss": 3.604620361328125, "grad_norm": 0.6883350014686584, "learning_rate": 0.0001, "elbo": 3.60462, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5451559934318553}
{"step": 43405, "loss": 3.60955924987793, "grad_norm": 0.6733520030975342, "learning_rate": 0.0001, "elbo": 3.60956, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5454492141684257}
{"step": 43410, "loss": 3.5960060119628907, "grad_norm": 0.9782169461250305, "learning_rate": 0.0001, "elbo": 3.59601, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5457424349049966}
{"step": 43415, "loss": 3.614350509643555, "grad_norm": 0.6727661490440369, "learning_rate": 0.0001, "elbo": 3.61435, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.546035655641567}
{"step": 43420, "loss": 3.6170417785644533, "grad_norm": 0.6830471158027649, "learning_rate": 0.0001, "elbo": 3.61704, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5463288763781375}
{"step": 43425, "loss": 3.6243576049804687, "grad_norm": 0.7168879508972168, "learning_rate": 0.0001, "elbo": 3.62436, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.546622097114708}
{"step": 43430, "loss": 3.608969497680664, "grad_norm": 0.693510115146637, "learning_rate": 0.0001, "elbo": 3.60897, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5469153178512784}
{"step": 43430, "loss": 3.608969497680664, "grad_norm": 0.693510115146637, "learning_rate": 0.0001, "elbo": 3.60897, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5469153178512784}
{"step": 43435, "loss": 3.6052478790283202, "grad_norm": 0.7601497173309326, "learning_rate": 0.0001, "elbo": 3.60525, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.547208538587849}
{"step": 43440, "loss": 3.613225555419922, "grad_norm": 0.6678068041801453, "learning_rate": 0.0001, "elbo": 3.61323, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5475017593244194}
{"step": 43445, "loss": 3.606732177734375, "grad_norm": 0.8792417049407959, "learning_rate": 0.0001, "elbo": 3.60673, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.54779498006099}
{"step": 43450, "loss": 3.6155105590820313, "grad_norm": 0.648851215839386, "learning_rate": 0.0001, "elbo": 3.61551, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5480882007975603}
{"step": 43455, "loss": 3.606781005859375, "grad_norm": 0.6962850093841553, "learning_rate": 0.0001, "elbo": 3.60678, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.548381421534131}
{"step": 43460, "loss": 3.610584259033203, "grad_norm": 0.7276992797851562, "learning_rate": 0.0001, "elbo": 3.61058, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5486746422707016}
{"step": 43465, "loss": 3.6198516845703126, "grad_norm": 0.6636008024215698, "learning_rate": 0.0001, "elbo": 3.61985, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.548967863007272}
{"step": 43470, "loss": 3.617348861694336, "grad_norm": 0.7151145339012146, "learning_rate": 0.0001, "elbo": 3.61735, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5492610837438425}
{"step": 43475, "loss": 3.6068046569824217, "grad_norm": 0.6832660436630249, "learning_rate": 0.0001, "elbo": 3.6068, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.549554304480413}
{"step": 43480, "loss": 3.620417022705078, "grad_norm": 0.6409467458724976, "learning_rate": 0.0001, "elbo": 3.62042, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5498475252169834}
{"step": 43485, "loss": 3.6132553100585936, "grad_norm": 0.6818941831588745, "learning_rate": 0.0001, "elbo": 3.61326, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.550140745953554}
{"step": 43490, "loss": 3.6079456329345705, "grad_norm": 0.891880214214325, "learning_rate": 0.0001, "elbo": 3.60795, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5504339666901243}
{"step": 43495, "loss": 3.6083969116210937, "grad_norm": 1.2027242183685303, "learning_rate": 0.0001, "elbo": 3.6084, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.550727187426695}
{"step": 43500, "loss": 3.621596908569336, "grad_norm": 0.7128930687904358, "learning_rate": 0.0001, "elbo": 3.6216, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5510204081632653}
{"step": 43505, "loss": 3.625292205810547, "grad_norm": 0.7293376922607422, "learning_rate": 0.0001, "elbo": 3.62529, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5513136288998357}
{"step": 43510, "loss": 3.62265625, "grad_norm": 0.6609324216842651, "learning_rate": 0.0001, "elbo": 3.62266, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.551606849636406}
{"step": 43515, "loss": 3.605968475341797, "grad_norm": 0.6569852828979492, "learning_rate": 0.0001, "elbo": 3.60597, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5519000703729766}
{"step": 43520, "loss": 3.6238304138183595, "grad_norm": 1.468833565711975, "learning_rate": 0.0001, "elbo": 3.62383, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.552193291109547}
{"step": 43525, "loss": 3.6233295440673827, "grad_norm": 0.6959572434425354, "learning_rate": 0.0001, "elbo": 3.62333, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5524865118461175}
{"step": 43530, "loss": 3.6093070983886717, "grad_norm": 0.7408541440963745, "learning_rate": 0.0001, "elbo": 3.60931, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.552779732582688}
{"step": 43535, "loss": 3.612918472290039, "grad_norm": 0.7195591330528259, "learning_rate": 0.0001, "elbo": 3.61292, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.553072953319259}
{"step": 43540, "loss": 3.609937286376953, "grad_norm": 0.7904923558235168, "learning_rate": 0.0001, "elbo": 3.60994, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5533661740558293}
{"step": 43545, "loss": 3.617533493041992, "grad_norm": 0.7079459428787231, "learning_rate": 0.0001, "elbo": 3.61753, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5536593947924}
{"step": 43550, "loss": 3.6175384521484375, "grad_norm": 0.7261770963668823, "learning_rate": 0.0001, "elbo": 3.61754, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5539526155289702}
{"step": 43555, "loss": 3.611894226074219, "grad_norm": 0.6933816075325012, "learning_rate": 0.0001, "elbo": 3.61189, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5542458362655407}
{"step": 43560, "loss": 3.6140731811523437, "grad_norm": 0.7692622542381287, "learning_rate": 0.0001, "elbo": 3.61407, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.554539057002111}
{"step": 43565, "loss": 3.610606384277344, "grad_norm": 0.8089641332626343, "learning_rate": 0.0001, "elbo": 3.61061, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5548322777386816}
{"step": 43570, "loss": 3.6063301086425783, "grad_norm": 0.7419312596321106, "learning_rate": 0.0001, "elbo": 3.60633, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.555125498475252}
{"step": 43575, "loss": 3.6264854431152345, "grad_norm": 0.7226255536079407, "learning_rate": 0.0001, "elbo": 3.62649, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5554187192118225}
{"step": 43580, "loss": 3.621010589599609, "grad_norm": 0.7850936651229858, "learning_rate": 0.0001, "elbo": 3.62101, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5557119399483934}
{"step": 43585, "loss": 3.6112335205078123, "grad_norm": 0.6803401708602905, "learning_rate": 0.0001, "elbo": 3.61123, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.556005160684964}
{"step": 43590, "loss": 3.6125125885009766, "grad_norm": 0.6967080235481262, "learning_rate": 0.0001, "elbo": 3.61251, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5562983814215343}
{"step": 43595, "loss": 3.627435302734375, "grad_norm": 0.8165475130081177, "learning_rate": 0.0001, "elbo": 3.62744, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5565916021581048}
{"step": 43600, "loss": 3.605809783935547, "grad_norm": 0.6827433705329895, "learning_rate": 0.0001, "elbo": 3.60581, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5568848228946752}
{"step": 43605, "loss": 3.6179908752441405, "grad_norm": 0.6870400905609131, "learning_rate": 0.0001, "elbo": 3.61799, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5571780436312457}
{"step": 43610, "loss": 3.6226768493652344, "grad_norm": 0.7427238821983337, "learning_rate": 0.0001, "elbo": 3.62268, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.557471264367816}
{"step": 43615, "loss": 3.61497802734375, "grad_norm": 0.6949872374534607, "learning_rate": 0.0001, "elbo": 3.61498, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5577644851043866}
{"step": 43620, "loss": 3.608437347412109, "grad_norm": 0.6654537916183472, "learning_rate": 0.0001, "elbo": 3.60844, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.558057705840957}
{"step": 43625, "loss": 3.595382308959961, "grad_norm": 0.674022376537323, "learning_rate": 0.0001, "elbo": 3.59538, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5583509265775275}
{"step": 43630, "loss": 3.6169166564941406, "grad_norm": 0.6444253921508789, "learning_rate": 0.0001, "elbo": 3.61692, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.558644147314098}
{"step": 43635, "loss": 3.6253658294677735, "grad_norm": 0.6968990564346313, "learning_rate": 0.0001, "elbo": 3.62537, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5589373680506684}
{"step": 43640, "loss": 3.610601806640625, "grad_norm": 0.7393463850021362, "learning_rate": 0.0001, "elbo": 3.6106, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.559230588787239}
{"step": 43645, "loss": 3.613266372680664, "grad_norm": 0.6704378724098206, "learning_rate": 0.0001, "elbo": 3.61327, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5595238095238093}
{"step": 43650, "loss": 3.6112327575683594, "grad_norm": 0.7314422130584717, "learning_rate": 0.0001, "elbo": 3.61123, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5598170302603798}
{"step": 43655, "loss": 3.6246849060058595, "grad_norm": 0.6640627980232239, "learning_rate": 0.0001, "elbo": 3.62469, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5601102509969507}
{"step": 43660, "loss": 3.599462127685547, "grad_norm": 0.652440071105957, "learning_rate": 0.0001, "elbo": 3.59946, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.560403471733521}
{"step": 43665, "loss": 3.6256752014160156, "grad_norm": 0.722486674785614, "learning_rate": 0.0001, "elbo": 3.62567, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5606966924700916}
{"step": 43670, "loss": 3.5906166076660155, "grad_norm": 0.723465085029602, "learning_rate": 0.0001, "elbo": 3.59062, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.560989913206662}
{"step": 43675, "loss": 3.6095291137695313, "grad_norm": 0.6852865219116211, "learning_rate": 0.0001, "elbo": 3.60953, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5612831339432325}
{"step": 43680, "loss": 3.6091075897216798, "grad_norm": 0.6709256172180176, "learning_rate": 0.0001, "elbo": 3.60911, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.561576354679803}
{"step": 43685, "loss": 3.598142623901367, "grad_norm": 0.6814642548561096, "learning_rate": 0.0001, "elbo": 3.59814, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5618695754163734}
{"step": 43685, "loss": 3.598142623901367, "grad_norm": 0.6814642548561096, "learning_rate": 0.0001, "elbo": 3.59814, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5618695754163734}
{"step": 43690, "loss": 3.607218933105469, "grad_norm": 0.7162464261054993, "learning_rate": 0.0001, "elbo": 3.60722, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.562162796152944}
{"step": 43690, "loss": 3.607218933105469, "grad_norm": 0.7162464261054993, "learning_rate": 0.0001, "elbo": 3.60722, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.562162796152944}
{"step": 43695, "loss": 3.6150238037109377, "grad_norm": 0.7011691927909851, "learning_rate": 0.0001, "elbo": 3.61502, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5624560168895143}
{"step": 43700, "loss": 3.616288757324219, "grad_norm": 0.8490133285522461, "learning_rate": 0.0001, "elbo": 3.61629, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5627492376260848}
{"step": 43705, "loss": 3.620169448852539, "grad_norm": 0.7138192057609558, "learning_rate": 0.0001, "elbo": 3.62017, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5630424583626557}
{"step": 43710, "loss": 3.626103973388672, "grad_norm": 0.8084316253662109, "learning_rate": 0.0001, "elbo": 3.6261, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.563335679099226}
{"step": 43710, "loss": 3.626103973388672, "grad_norm": 0.8084316253662109, "learning_rate": 0.0001, "elbo": 3.6261, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.563335679099226}
{"step": 43715, "loss": 3.616120529174805, "grad_norm": 0.6505059599876404, "learning_rate": 0.0001, "elbo": 3.61612, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5636288998357966}
{"step": 43720, "loss": 3.6025978088378907, "grad_norm": 0.7292006611824036, "learning_rate": 0.0001, "elbo": 3.6026, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.563922120572367}
{"step": 43725, "loss": 3.6067176818847657, "grad_norm": 0.7112288475036621, "learning_rate": 0.0001, "elbo": 3.60672, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5642153413089375}
{"step": 43730, "loss": 3.603012466430664, "grad_norm": 0.7025930881500244, "learning_rate": 0.0001, "elbo": 3.60301, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.564508562045508}
{"step": 43735, "loss": 3.6162178039550783, "grad_norm": 0.6718329191207886, "learning_rate": 0.0001, "elbo": 3.61622, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5648017827820784}
{"step": 43740, "loss": 3.6052925109863283, "grad_norm": 0.6820933222770691, "learning_rate": 0.0001, "elbo": 3.60529, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.565095003518649}
{"step": 43745, "loss": 3.615904998779297, "grad_norm": 0.7915660738945007, "learning_rate": 0.0001, "elbo": 3.6159, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5653882242552193}
{"step": 43750, "loss": 3.6074729919433595, "grad_norm": 0.7142928838729858, "learning_rate": 0.0001, "elbo": 3.60747, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5656814449917897}
{"step": 43755, "loss": 3.6107994079589845, "grad_norm": 0.6599190831184387, "learning_rate": 0.0001, "elbo": 3.6108, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.56597466572836}
{"step": 43760, "loss": 3.6156131744384767, "grad_norm": 0.9092174768447876, "learning_rate": 0.0001, "elbo": 3.61561, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5662678864649306}
{"step": 43765, "loss": 3.6114635467529297, "grad_norm": 0.6785511374473572, "learning_rate": 0.0001, "elbo": 3.61146, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.566561107201501}
{"step": 43770, "loss": 3.616031265258789, "grad_norm": 0.6459888219833374, "learning_rate": 0.0001, "elbo": 3.61603, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5668543279380716}
{"step": 43775, "loss": 3.615830993652344, "grad_norm": 0.7194380760192871, "learning_rate": 0.0001, "elbo": 3.61583, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.567147548674642}
{"step": 43780, "loss": 3.6223243713378905, "grad_norm": 0.6770534515380859, "learning_rate": 0.0001, "elbo": 3.62232, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.567440769411213}
{"step": 43785, "loss": 3.6018482208251954, "grad_norm": 0.7273313999176025, "learning_rate": 0.0001, "elbo": 3.60185, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5677339901477834}
{"step": 43790, "loss": 3.607379150390625, "grad_norm": 0.693215012550354, "learning_rate": 0.0001, "elbo": 3.60738, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.568027210884354}
{"step": 43795, "loss": 3.61505126953125, "grad_norm": 1.021926760673523, "learning_rate": 0.0001, "elbo": 3.61505, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5683204316209243}
{"step": 43800, "loss": 3.6144077301025392, "grad_norm": 0.6640030741691589, "learning_rate": 0.0001, "elbo": 3.61441, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5686136523574947}
{"step": 43805, "loss": 3.6022003173828123, "grad_norm": 0.6562496423721313, "learning_rate": 0.0001, "elbo": 3.6022, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.568906873094065}
{"step": 43810, "loss": 3.6158767700195313, "grad_norm": 0.715693473815918, "learning_rate": 0.0001, "elbo": 3.61588, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5692000938306356}
{"step": 43815, "loss": 3.604852294921875, "grad_norm": 0.6466091275215149, "learning_rate": 0.0001, "elbo": 3.60485, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.569493314567206}
{"step": 43820, "loss": 3.6161739349365236, "grad_norm": 0.7070316672325134, "learning_rate": 0.0001, "elbo": 3.61617, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5697865353037765}
{"step": 43825, "loss": 3.610609436035156, "grad_norm": 0.667015016078949, "learning_rate": 0.0001, "elbo": 3.61061, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5700797560403474}
{"step": 43830, "loss": 3.6108421325683593, "grad_norm": 0.6719779372215271, "learning_rate": 0.0001, "elbo": 3.61084, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.570372976776918}
{"step": 43835, "loss": 3.6011390686035156, "grad_norm": 0.689812958240509, "learning_rate": 0.0001, "elbo": 3.60114, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5706661975134883}
{"step": 43840, "loss": 3.6047012329101564, "grad_norm": 0.6528840065002441, "learning_rate": 0.0001, "elbo": 3.6047, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.570959418250059}
{"step": 43845, "loss": 3.602490997314453, "grad_norm": 0.7150605320930481, "learning_rate": 0.0001, "elbo": 3.60249, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5712526389866293}
{"step": 43850, "loss": 3.6223121643066407, "grad_norm": 0.6766937971115112, "learning_rate": 0.0001, "elbo": 3.62231, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5715458597231997}
{"step": 43855, "loss": 3.6121452331542967, "grad_norm": 0.7606209516525269, "learning_rate": 0.0001, "elbo": 3.61215, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.57183908045977}
{"step": 43860, "loss": 3.617623138427734, "grad_norm": 0.7268351316452026, "learning_rate": 0.0001, "elbo": 3.61762, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5721323011963406}
{"step": 43865, "loss": 3.60433349609375, "grad_norm": 0.6352977156639099, "learning_rate": 0.0001, "elbo": 3.60433, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.572425521932911}
{"step": 43870, "loss": 3.602558898925781, "grad_norm": 0.6709262728691101, "learning_rate": 0.0001, "elbo": 3.60256, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5727187426694815}
{"step": 43875, "loss": 3.6164230346679687, "grad_norm": 0.763174295425415, "learning_rate": 0.0001, "elbo": 3.61642, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.573011963406052}
{"step": 43880, "loss": 3.619293975830078, "grad_norm": 0.6502229571342468, "learning_rate": 0.0001, "elbo": 3.61929, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5733051841426224}
{"step": 43885, "loss": 3.5866714477539063, "grad_norm": 0.6912708878517151, "learning_rate": 0.0001, "elbo": 3.58667, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.573598404879193}
{"step": 43890, "loss": 3.617416000366211, "grad_norm": 0.6501936316490173, "learning_rate": 0.0001, "elbo": 3.61742, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5738916256157633}
{"step": 43895, "loss": 3.5989952087402344, "grad_norm": 0.8216702938079834, "learning_rate": 0.0001, "elbo": 3.599, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.574184846352334}
{"step": 43900, "loss": 3.61773681640625, "grad_norm": 0.7042893171310425, "learning_rate": 0.0001, "elbo": 3.61774, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5744780670889043}
{"step": 43905, "loss": 3.6034217834472657, "grad_norm": 0.7334530353546143, "learning_rate": 0.0001, "elbo": 3.60342, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.574771287825475}
{"step": 43910, "loss": 3.6030982971191405, "grad_norm": 0.7198043465614319, "learning_rate": 0.0001, "elbo": 3.6031, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5750645085620456}
{"step": 43915, "loss": 3.6150306701660155, "grad_norm": 0.7376478314399719, "learning_rate": 0.0001, "elbo": 3.61503, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.575357729298616}
{"step": 43920, "loss": 3.6244476318359373, "grad_norm": 0.7610045671463013, "learning_rate": 0.0001, "elbo": 3.62445, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5756509500351865}
{"step": 43925, "loss": 3.6088687896728517, "grad_norm": 0.7214381694793701, "learning_rate": 0.0001, "elbo": 3.60887, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.575944170771757}
{"step": 43930, "loss": 3.6013778686523437, "grad_norm": 0.7394117116928101, "learning_rate": 0.0001, "elbo": 3.60138, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5762373915083274}
{"step": 43935, "loss": 3.6220767974853514, "grad_norm": 0.7627154588699341, "learning_rate": 0.0001, "elbo": 3.62208, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.576530612244898}
{"step": 43940, "loss": 3.6153446197509767, "grad_norm": 0.7310627698898315, "learning_rate": 0.0001, "elbo": 3.61534, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5768238329814683}
{"step": 43945, "loss": 3.6095260620117187, "grad_norm": 0.7193444967269897, "learning_rate": 0.0001, "elbo": 3.60953, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.577117053718039}
{"step": 43950, "loss": 3.618184280395508, "grad_norm": 1.008012294769287, "learning_rate": 0.0001, "elbo": 3.61818, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5774102744546097}
{"step": 43955, "loss": 3.6157718658447267, "grad_norm": 0.6587028503417969, "learning_rate": 0.0001, "elbo": 3.61577, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.57770349519118}
{"step": 43960, "loss": 3.6243934631347656, "grad_norm": 0.6841651201248169, "learning_rate": 0.0001, "elbo": 3.62439, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5779967159277506}
{"step": 43965, "loss": 3.604922866821289, "grad_norm": 0.7667160630226135, "learning_rate": 0.0001, "elbo": 3.60492, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.578289936664321}
{"step": 43970, "loss": 3.616259002685547, "grad_norm": 0.7036290764808655, "learning_rate": 0.0001, "elbo": 3.61626, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5785831574008915}
{"step": 43975, "loss": 3.5994903564453127, "grad_norm": 0.6819214820861816, "learning_rate": 0.0001, "elbo": 3.59949, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.578876378137462}
{"step": 43980, "loss": 3.5975387573242186, "grad_norm": 0.7994405031204224, "learning_rate": 0.0001, "elbo": 3.59754, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5791695988740324}
{"step": 43985, "loss": 3.6103794097900392, "grad_norm": 0.7019323706626892, "learning_rate": 0.0001, "elbo": 3.61038, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.579462819610603}
{"step": 43990, "loss": 3.6199188232421875, "grad_norm": 0.7212421298027039, "learning_rate": 0.0001, "elbo": 3.61992, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5797560403471733}
{"step": 43995, "loss": 3.604888153076172, "grad_norm": 0.6447035074234009, "learning_rate": 0.0001, "elbo": 3.60489, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5800492610837438}
{"step": 44000, "loss": 3.6110477447509766, "grad_norm": 0.7102403044700623, "learning_rate": 0.0001, "elbo": 3.61105, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5803424818203142}
{"step": 44005, "loss": 3.6217727661132812, "grad_norm": 0.7099698185920715, "learning_rate": 0.0001, "elbo": 3.62177, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5806357025568847}
{"step": 44010, "loss": 3.593633270263672, "grad_norm": 0.752251923084259, "learning_rate": 0.0001, "elbo": 3.59363, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.580928923293455}
{"step": 44015, "loss": 3.6200927734375, "grad_norm": 0.6847530007362366, "learning_rate": 0.0001, "elbo": 3.62009, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5812221440300256}
{"step": 44020, "loss": 3.6107711791992188, "grad_norm": 0.7456072568893433, "learning_rate": 0.0001, "elbo": 3.61077, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.581515364766596}
{"step": 44025, "loss": 3.6200721740722654, "grad_norm": 0.726749837398529, "learning_rate": 0.0001, "elbo": 3.62007, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.581808585503167}
{"step": 44030, "loss": 3.617474365234375, "grad_norm": 0.7267966866493225, "learning_rate": 0.0001, "elbo": 3.61747, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5821018062397374}
{"step": 44035, "loss": 3.611308288574219, "grad_norm": 1.0446854829788208, "learning_rate": 0.0001, "elbo": 3.61131, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.582395026976308}
{"step": 44040, "loss": 3.610749053955078, "grad_norm": 0.7181500792503357, "learning_rate": 0.0001, "elbo": 3.61075, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5826882477128783}
{"step": 44045, "loss": 3.619165802001953, "grad_norm": 0.6945469379425049, "learning_rate": 0.0001, "elbo": 3.61917, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5829814684494488}
{"step": 44045, "loss": 3.619165802001953, "grad_norm": 0.6945469379425049, "learning_rate": 0.0001, "elbo": 3.61917, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5829814684494488}
{"step": 44050, "loss": 3.6111419677734373, "grad_norm": 0.7091934680938721, "learning_rate": 0.0001, "elbo": 3.61114, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.583274689186019}
{"step": 44055, "loss": 3.596566009521484, "grad_norm": 0.6706783771514893, "learning_rate": 0.0001, "elbo": 3.59657, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5835679099225897}
{"step": 44060, "loss": 3.6100479125976563, "grad_norm": 0.6634994149208069, "learning_rate": 0.0001, "elbo": 3.61005, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.58386113065916}
{"step": 44065, "loss": 3.6159557342529296, "grad_norm": 0.6612393856048584, "learning_rate": 0.0001, "elbo": 3.61596, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5841543513957306}
{"step": 44070, "loss": 3.6033802032470703, "grad_norm": 0.8001354336738586, "learning_rate": 0.0001, "elbo": 3.60338, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5844475721323015}
{"step": 44070, "loss": 3.6033802032470703, "grad_norm": 0.8001354336738586, "learning_rate": 0.0001, "elbo": 3.60338, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5844475721323015}
{"step": 44075, "loss": 3.6237869262695312, "grad_norm": 0.6894035339355469, "learning_rate": 0.0001, "elbo": 3.62379, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.584740792868872}
{"step": 44080, "loss": 3.599457550048828, "grad_norm": 0.7100386023521423, "learning_rate": 0.0001, "elbo": 3.59946, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5850340136054424}
{"step": 44085, "loss": 3.612420654296875, "grad_norm": 0.7290034294128418, "learning_rate": 0.0001, "elbo": 3.61242, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.585327234342013}
{"step": 44090, "loss": 3.610179138183594, "grad_norm": 0.7544072270393372, "learning_rate": 0.0001, "elbo": 3.61018, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5856204550785833}
{"step": 44095, "loss": 3.6038436889648438, "grad_norm": 0.7931712865829468, "learning_rate": 0.0001, "elbo": 3.60384, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5859136758151537}
{"step": 44100, "loss": 3.6001487731933595, "grad_norm": 0.6754352450370789, "learning_rate": 0.0001, "elbo": 3.60015, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.586206896551724}
{"step": 44105, "loss": 3.6295875549316405, "grad_norm": 0.7473129034042358, "learning_rate": 0.0001, "elbo": 3.62959, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5865001172882947}
{"step": 44110, "loss": 3.600023651123047, "grad_norm": 0.6854532957077026, "learning_rate": 0.0001, "elbo": 3.60002, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.586793338024865}
{"step": 44115, "loss": 3.6217586517333986, "grad_norm": 0.7561337351799011, "learning_rate": 0.0001, "elbo": 3.62176, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5870865587614356}
{"step": 44120, "loss": 3.6204246520996093, "grad_norm": 0.8383820056915283, "learning_rate": 0.0001, "elbo": 3.62042, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.587379779498006}
{"step": 44125, "loss": 3.6001495361328124, "grad_norm": 0.675329327583313, "learning_rate": 0.0001, "elbo": 3.60015, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5876730002345765}
{"step": 44130, "loss": 3.610403823852539, "grad_norm": 0.7611393928527832, "learning_rate": 0.0001, "elbo": 3.6104, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.587966220971147}
{"step": 44135, "loss": 3.6115936279296874, "grad_norm": 0.8314067125320435, "learning_rate": 0.0001, "elbo": 3.61159, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5882594417077174}
{"step": 44140, "loss": 3.591543197631836, "grad_norm": 0.8761052489280701, "learning_rate": 0.0001, "elbo": 3.59154, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.588552662444288}
{"step": 44145, "loss": 3.610418701171875, "grad_norm": 0.7415770292282104, "learning_rate": 0.0001, "elbo": 3.61042, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5888458831808583}
{"step": 44150, "loss": 3.6278968811035157, "grad_norm": 0.6575100421905518, "learning_rate": 0.0001, "elbo": 3.6279, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.589139103917429}
{"step": 44150, "loss": 3.6278968811035157, "grad_norm": 0.6575100421905518, "learning_rate": 0.0001, "elbo": 3.6279, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.589139103917429}
{"step": 44155, "loss": 3.593151092529297, "grad_norm": 0.6531134843826294, "learning_rate": 0.0001, "elbo": 3.59315, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5894323246539996}
{"step": 44160, "loss": 3.6216289520263674, "grad_norm": 0.7112041115760803, "learning_rate": 0.0001, "elbo": 3.62163, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.58972554539057}
{"step": 44165, "loss": 3.6221290588378907, "grad_norm": 0.6804618835449219, "learning_rate": 0.0001, "elbo": 3.62213, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5900187661271405}
{"step": 44170, "loss": 3.612574005126953, "grad_norm": 0.6750351786613464, "learning_rate": 0.0001, "elbo": 3.61257, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.590311986863711}
{"step": 44175, "loss": 3.6280067443847654, "grad_norm": 0.8622156977653503, "learning_rate": 0.0001, "elbo": 3.62801, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5906052076002815}
{"step": 44180, "loss": 3.6079128265380858, "grad_norm": 0.6952463984489441, "learning_rate": 0.0001, "elbo": 3.60791, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.590898428336852}
{"step": 44185, "loss": 3.610545349121094, "grad_norm": 0.6871792078018188, "learning_rate": 0.0001, "elbo": 3.61055, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5911916490734224}
{"step": 44190, "loss": 3.5942916870117188, "grad_norm": 0.6929821372032166, "learning_rate": 0.0001, "elbo": 3.59429, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.591484869809993}
{"step": 44195, "loss": 3.598053741455078, "grad_norm": 0.655958354473114, "learning_rate": 0.0001, "elbo": 3.59805, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5917780905465637}
{"step": 44200, "loss": 3.6056278228759764, "grad_norm": 0.9941576719284058, "learning_rate": 0.0001, "elbo": 3.60563, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.592071311283134}
{"step": 44205, "loss": 3.595853805541992, "grad_norm": 0.6455947756767273, "learning_rate": 0.0001, "elbo": 3.59585, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5923645320197046}
{"step": 44210, "loss": 3.597137451171875, "grad_norm": 0.7015332579612732, "learning_rate": 0.0001, "elbo": 3.59714, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.592657752756275}
{"step": 44215, "loss": 3.6164661407470704, "grad_norm": 0.7650613784790039, "learning_rate": 0.0001, "elbo": 3.61647, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5929509734928455}
{"step": 44220, "loss": 3.6097999572753907, "grad_norm": 0.6421565413475037, "learning_rate": 0.0001, "elbo": 3.6098, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.593244194229416}
{"step": 44225, "loss": 3.626120376586914, "grad_norm": 0.7215990424156189, "learning_rate": 0.0001, "elbo": 3.62612, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5935374149659864}
{"step": 44230, "loss": 3.6013240814208984, "grad_norm": 0.6586055159568787, "learning_rate": 0.0001, "elbo": 3.60132, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.593830635702557}
{"step": 44235, "loss": 3.607133483886719, "grad_norm": 0.6524456143379211, "learning_rate": 0.0001, "elbo": 3.60713, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5941238564391274}
{"step": 44240, "loss": 3.60697021484375, "grad_norm": 0.698154091835022, "learning_rate": 0.0001, "elbo": 3.60697, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.594417077175698}
{"step": 44245, "loss": 3.6289257049560546, "grad_norm": 0.6462186574935913, "learning_rate": 0.0001, "elbo": 3.62893, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5947102979122683}
{"step": 44250, "loss": 3.6006568908691405, "grad_norm": 0.6970375776290894, "learning_rate": 0.0001, "elbo": 3.60066, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5950035186488387}
{"step": 44255, "loss": 3.5921119689941405, "grad_norm": 0.6574156284332275, "learning_rate": 0.0001, "elbo": 3.59211, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.595296739385409}
{"step": 44260, "loss": 3.623268890380859, "grad_norm": 0.6646251678466797, "learning_rate": 0.0001, "elbo": 3.62327, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5955899601219796}
{"step": 44265, "loss": 3.5860992431640626, "grad_norm": 0.7804410457611084, "learning_rate": 0.0001, "elbo": 3.5861, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.59588318085855}
{"step": 44270, "loss": 3.616699981689453, "grad_norm": 0.6387103796005249, "learning_rate": 0.0001, "elbo": 3.6167, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5961764015951205}
{"step": 44275, "loss": 3.615129089355469, "grad_norm": 0.6538716554641724, "learning_rate": 0.0001, "elbo": 3.61513, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5964696223316914}
{"step": 44280, "loss": 3.620769500732422, "grad_norm": 0.720931351184845, "learning_rate": 0.0001, "elbo": 3.62077, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.596762843068262}
{"step": 44285, "loss": 3.6018661499023437, "grad_norm": 0.6751096248626709, "learning_rate": 0.0001, "elbo": 3.60187, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5970560638048323}
{"step": 44290, "loss": 3.6303260803222654, "grad_norm": 0.7254676222801208, "learning_rate": 0.0001, "elbo": 3.63033, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.597349284541403}
{"step": 44295, "loss": 3.5971515655517576, "grad_norm": 0.6889790892601013, "learning_rate": 0.0001, "elbo": 3.59715, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5976425052779732}
{"step": 44300, "loss": 3.6163875579833986, "grad_norm": 0.7057093977928162, "learning_rate": 0.0001, "elbo": 3.61639, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5979357260145437}
{"step": 44305, "loss": 3.620679473876953, "grad_norm": 0.6662892699241638, "learning_rate": 0.0001, "elbo": 3.62068, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.598228946751114}
{"step": 44310, "loss": 3.612201690673828, "grad_norm": 0.6702008843421936, "learning_rate": 0.0001, "elbo": 3.6122, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5985221674876846}
{"step": 44315, "loss": 3.608362579345703, "grad_norm": 0.6636449694633484, "learning_rate": 0.0001, "elbo": 3.60836, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.598815388224255}
{"step": 44320, "loss": 3.6157936096191405, "grad_norm": 0.7340224385261536, "learning_rate": 0.0001, "elbo": 3.61579, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.599108608960826}
{"step": 44320, "loss": 3.6157936096191405, "grad_norm": 0.7340224385261536, "learning_rate": 0.0001, "elbo": 3.61579, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.599108608960826}
{"step": 44320, "loss": 3.6157936096191405, "grad_norm": 0.7340224385261536, "learning_rate": 0.0001, "elbo": 3.61579, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.599108608960826}
{"step": 44325, "loss": 3.6169456481933593, "grad_norm": 0.6933377385139465, "learning_rate": 0.0001, "elbo": 3.61695, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5994018296973964}
{"step": 44330, "loss": 3.588910675048828, "grad_norm": 0.7350319623947144, "learning_rate": 0.0001, "elbo": 3.58891, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.599695050433967}
{"step": 44335, "loss": 3.6114906311035155, "grad_norm": 0.7162929773330688, "learning_rate": 0.0001, "elbo": 3.61149, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.5999882711705373}
{"step": 44340, "loss": 3.614166259765625, "grad_norm": 0.703569769859314, "learning_rate": 0.0001, "elbo": 3.61417, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.600281491907108}
{"step": 44345, "loss": 3.6172164916992187, "grad_norm": 0.945905864238739, "learning_rate": 0.0001, "elbo": 3.61722, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6005747126436782}
{"step": 44345, "loss": 3.6172164916992187, "grad_norm": 0.945905864238739, "learning_rate": 0.0001, "elbo": 3.61722, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6005747126436782}
{"step": 44350, "loss": 3.6267452239990234, "grad_norm": 0.681812047958374, "learning_rate": 0.0001, "elbo": 3.62675, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6008679333802487}
{"step": 44355, "loss": 3.5963672637939452, "grad_norm": 0.6428623199462891, "learning_rate": 0.0001, "elbo": 3.59637, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.601161154116819}
{"step": 44360, "loss": 3.5926910400390626, "grad_norm": 0.6769838333129883, "learning_rate": 0.0001, "elbo": 3.59269, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6014543748533896}
{"step": 44365, "loss": 3.6104827880859376, "grad_norm": 0.7459681630134583, "learning_rate": 0.0001, "elbo": 3.61048, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.60174759558996}
{"step": 44370, "loss": 3.6212093353271486, "grad_norm": 0.6823952198028564, "learning_rate": 0.0001, "elbo": 3.62121, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6020408163265305}
{"step": 44375, "loss": 3.6233139038085938, "grad_norm": 0.723298966884613, "learning_rate": 0.0001, "elbo": 3.62331, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.602334037063101}
{"step": 44380, "loss": 3.6150840759277343, "grad_norm": 0.6745441555976868, "learning_rate": 0.0001, "elbo": 3.61508, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6026272577996714}
{"step": 44385, "loss": 3.60648193359375, "grad_norm": 0.7864968776702881, "learning_rate": 0.0001, "elbo": 3.60648, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.602920478536242}
{"step": 44390, "loss": 3.604756164550781, "grad_norm": 0.650607168674469, "learning_rate": 0.0001, "elbo": 3.60476, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6032136992728123}
{"step": 44395, "loss": 3.6163848876953124, "grad_norm": 0.6892912983894348, "learning_rate": 0.0001, "elbo": 3.61638, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.603506920009383}
{"step": 44400, "loss": 3.621228790283203, "grad_norm": 0.7129437327384949, "learning_rate": 0.0001, "elbo": 3.62123, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6038001407459537}
{"step": 44405, "loss": 3.618922805786133, "grad_norm": 0.6888516545295715, "learning_rate": 0.0001, "elbo": 3.61892, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.604093361482524}
{"step": 44410, "loss": 3.61230583190918, "grad_norm": 0.7042694091796875, "learning_rate": 0.0001, "elbo": 3.61231, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6043865822190946}
{"step": 44415, "loss": 3.6121715545654296, "grad_norm": 0.8525443077087402, "learning_rate": 0.0001, "elbo": 3.61217, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.604679802955665}
{"step": 44420, "loss": 3.6077171325683595, "grad_norm": 0.7159590721130371, "learning_rate": 0.0001, "elbo": 3.60772, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6049730236922355}
{"step": 44425, "loss": 3.6213096618652343, "grad_norm": 0.7232871651649475, "learning_rate": 0.0001, "elbo": 3.62131, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.605266244428806}
{"step": 44430, "loss": 3.5983356475830077, "grad_norm": 0.7103886604309082, "learning_rate": 0.0001, "elbo": 3.59834, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6055594651653764}
{"step": 44430, "loss": 3.5983356475830077, "grad_norm": 0.7103886604309082, "learning_rate": 0.0001, "elbo": 3.59834, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6055594651653764}
{"step": 44435, "loss": 3.615876007080078, "grad_norm": 0.7226691842079163, "learning_rate": 0.0001, "elbo": 3.61588, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.605852685901947}
{"step": 44440, "loss": 3.5990291595458985, "grad_norm": 0.6648800373077393, "learning_rate": 0.0001, "elbo": 3.59903, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6061459066385178}
{"step": 44445, "loss": 3.6164268493652343, "grad_norm": 0.6712974905967712, "learning_rate": 0.0001, "elbo": 3.61643, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.606439127375088}
{"step": 44450, "loss": 3.611683654785156, "grad_norm": 0.6806085705757141, "learning_rate": 0.0001, "elbo": 3.61168, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6067323481116587}
{"step": 44455, "loss": 3.628783416748047, "grad_norm": 0.9034536480903625, "learning_rate": 0.0001, "elbo": 3.62878, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.607025568848229}
{"step": 44460, "loss": 3.606279754638672, "grad_norm": 0.6817594766616821, "learning_rate": 0.0001, "elbo": 3.60628, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6073187895847996}
{"step": 44465, "loss": 3.608132171630859, "grad_norm": 0.9269545674324036, "learning_rate": 0.0001, "elbo": 3.60813, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.60761201032137}
{"step": 44470, "loss": 3.6143653869628904, "grad_norm": 0.6540157794952393, "learning_rate": 0.0001, "elbo": 3.61437, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6079052310579405}
{"step": 44475, "loss": 3.6161712646484374, "grad_norm": 0.6540685892105103, "learning_rate": 0.0001, "elbo": 3.61617, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.608198451794511}
{"step": 44480, "loss": 3.6235389709472656, "grad_norm": 0.6967843770980835, "learning_rate": 0.0001, "elbo": 3.62354, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6084916725310814}
{"step": 44485, "loss": 3.6144622802734374, "grad_norm": 0.7453532218933105, "learning_rate": 0.0001, "elbo": 3.61446, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.608784893267652}
{"step": 44490, "loss": 3.6252552032470704, "grad_norm": 0.732890784740448, "learning_rate": 0.0001, "elbo": 3.62526, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6090781140042223}
{"step": 44495, "loss": 3.600067901611328, "grad_norm": 0.6495212316513062, "learning_rate": 0.0001, "elbo": 3.60007, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6093713347407927}
{"step": 44500, "loss": 3.5960460662841798, "grad_norm": 0.6967467069625854, "learning_rate": 0.0001, "elbo": 3.59605, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.609664555477363}
{"step": 44505, "loss": 3.6216644287109374, "grad_norm": 0.682012140750885, "learning_rate": 0.0001, "elbo": 3.62166, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6099577762139337}
{"step": 44510, "loss": 3.603714370727539, "grad_norm": 0.6900676488876343, "learning_rate": 0.0001, "elbo": 3.60371, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.610250996950504}
{"step": 44515, "loss": 3.5902732849121093, "grad_norm": 0.7408657073974609, "learning_rate": 0.0001, "elbo": 3.59027, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6105442176870746}
{"step": 44520, "loss": 3.6040382385253906, "grad_norm": 0.7062262892723083, "learning_rate": 0.0001, "elbo": 3.60404, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6108374384236455}
{"step": 44525, "loss": 3.5998085021972654, "grad_norm": 0.6891714334487915, "learning_rate": 0.0001, "elbo": 3.59981, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.611130659160216}
{"step": 44530, "loss": 3.625248336791992, "grad_norm": 1.3430657386779785, "learning_rate": 0.0001, "elbo": 3.62525, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6114238798967864}
{"step": 44535, "loss": 3.606803131103516, "grad_norm": 0.7073857188224792, "learning_rate": 0.0001, "elbo": 3.6068, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.611717100633357}
{"step": 44540, "loss": 3.6016262054443358, "grad_norm": 0.7756453156471252, "learning_rate": 0.0001, "elbo": 3.60163, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6120103213699273}
{"step": 44545, "loss": 3.600128936767578, "grad_norm": 0.7142720818519592, "learning_rate": 0.0001, "elbo": 3.60013, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6123035421064977}
{"step": 44550, "loss": 3.6043548583984375, "grad_norm": 0.67323237657547, "learning_rate": 0.0001, "elbo": 3.60435, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.612596762843068}
{"step": 44555, "loss": 3.625653076171875, "grad_norm": 0.6626057028770447, "learning_rate": 0.0001, "elbo": 3.62565, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6128899835796386}
{"step": 44560, "loss": 3.6115455627441406, "grad_norm": 0.6659541130065918, "learning_rate": 0.0001, "elbo": 3.61155, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.613183204316209}
{"step": 44565, "loss": 3.5917690277099608, "grad_norm": 0.7916408777236938, "learning_rate": 0.0001, "elbo": 3.59177, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.61347642505278}
{"step": 44570, "loss": 3.612745666503906, "grad_norm": 0.7087569832801819, "learning_rate": 0.0001, "elbo": 3.61275, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6137696457893504}
{"step": 44575, "loss": 3.61224250793457, "grad_norm": 0.6657957434654236, "learning_rate": 0.0001, "elbo": 3.61224, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.614062866525921}
{"step": 44580, "loss": 3.60052490234375, "grad_norm": 1.0248790979385376, "learning_rate": 0.0001, "elbo": 3.60053, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6143560872624914}
{"step": 44585, "loss": 3.6244903564453126, "grad_norm": 0.6792204976081848, "learning_rate": 0.0001, "elbo": 3.62449, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.614649307999062}
{"step": 44590, "loss": 3.5808792114257812, "grad_norm": 0.6464689373970032, "learning_rate": 0.0001, "elbo": 3.58088, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6149425287356323}
{"step": 44595, "loss": 3.6149887084960937, "grad_norm": 0.6366428732872009, "learning_rate": 0.0001, "elbo": 3.61499, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6152357494722027}
{"step": 44600, "loss": 3.605303192138672, "grad_norm": 0.6993203163146973, "learning_rate": 0.0001, "elbo": 3.6053, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.615528970208773}
{"step": 44605, "loss": 3.60611572265625, "grad_norm": 0.7037259936332703, "learning_rate": 0.0001, "elbo": 3.60612, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6158221909453436}
{"step": 44610, "loss": 3.6129714965820314, "grad_norm": 0.7930845022201538, "learning_rate": 0.0001, "elbo": 3.61297, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.616115411681914}
{"step": 44615, "loss": 3.603338623046875, "grad_norm": 0.6479370594024658, "learning_rate": 0.0001, "elbo": 3.60334, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6164086324184845}
{"step": 44620, "loss": 3.5982303619384766, "grad_norm": 0.6704499125480652, "learning_rate": 0.0001, "elbo": 3.59823, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.616701853155055}
{"step": 44625, "loss": 3.5886886596679686, "grad_norm": 0.6923515796661377, "learning_rate": 0.0001, "elbo": 3.58869, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6169950738916254}
{"step": 44630, "loss": 3.5860450744628904, "grad_norm": 0.7679826617240906, "learning_rate": 0.0001, "elbo": 3.58605, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.617288294628196}
{"step": 44635, "loss": 3.6058956146240235, "grad_norm": 0.7685940265655518, "learning_rate": 0.0001, "elbo": 3.6059, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6175815153647664}
{"step": 44640, "loss": 3.6013336181640625, "grad_norm": 0.7740747332572937, "learning_rate": 0.0001, "elbo": 3.60133, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6178747361013373}
{"step": 44645, "loss": 3.6124744415283203, "grad_norm": 0.7675145268440247, "learning_rate": 0.0001, "elbo": 3.61247, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6181679568379077}
{"step": 44650, "loss": 3.604725646972656, "grad_norm": 0.7652575373649597, "learning_rate": 0.0001, "elbo": 3.60473, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.618461177574478}
{"step": 44655, "loss": 3.61802978515625, "grad_norm": 0.7448182106018066, "learning_rate": 0.0001, "elbo": 3.61803, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6187543983110486}
{"step": 44660, "loss": 3.6038581848144533, "grad_norm": 0.8312509059906006, "learning_rate": 0.0001, "elbo": 3.60386, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.619047619047619}
{"step": 44665, "loss": 3.615081787109375, "grad_norm": 0.6638615727424622, "learning_rate": 0.0001, "elbo": 3.61508, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6193408397841895}
{"step": 44670, "loss": 3.6220951080322266, "grad_norm": 0.7240074276924133, "learning_rate": 0.0001, "elbo": 3.62209, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.61963406052076}
{"step": 44675, "loss": 3.6066741943359375, "grad_norm": 0.7392721772193909, "learning_rate": 0.0001, "elbo": 3.60667, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6199272812573304}
{"step": 44680, "loss": 3.614604949951172, "grad_norm": 0.691826343536377, "learning_rate": 0.0001, "elbo": 3.61461, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.620220501993901}
{"step": 44685, "loss": 3.6206298828125, "grad_norm": 0.6859536170959473, "learning_rate": 0.0001, "elbo": 3.62063, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6205137227304713}
{"step": 44685, "loss": 3.6206298828125, "grad_norm": 0.6859536170959473, "learning_rate": 0.0001, "elbo": 3.62063, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6205137227304713}
{"step": 44690, "loss": 3.610503387451172, "grad_norm": 0.6985312700271606, "learning_rate": 0.0001, "elbo": 3.6105, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6208069434670422}
{"step": 44695, "loss": 3.618472671508789, "grad_norm": 0.6741527318954468, "learning_rate": 0.0001, "elbo": 3.61847, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6211001642036127}
{"step": 44700, "loss": 3.608880615234375, "grad_norm": 0.7297027707099915, "learning_rate": 0.0001, "elbo": 3.60888, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.621393384940183}
{"step": 44705, "loss": 3.6175987243652346, "grad_norm": 0.7694353461265564, "learning_rate": 0.0001, "elbo": 3.6176, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6216866056767536}
{"step": 44710, "loss": 3.6021026611328124, "grad_norm": 0.6459797024726868, "learning_rate": 0.0001, "elbo": 3.6021, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.621979826413324}
{"step": 44715, "loss": 3.6018745422363283, "grad_norm": 0.6769719123840332, "learning_rate": 0.0001, "elbo": 3.60187, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6222730471498945}
{"step": 44720, "loss": 3.6024608612060547, "grad_norm": 0.8043074011802673, "learning_rate": 0.0001, "elbo": 3.60246, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.622566267886465}
{"step": 44725, "loss": 3.610021209716797, "grad_norm": 0.6778925061225891, "learning_rate": 0.0001, "elbo": 3.61002, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6228594886230354}
{"step": 44730, "loss": 3.6151359558105467, "grad_norm": 0.8717228770256042, "learning_rate": 0.0001, "elbo": 3.61514, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.623152709359606}
{"step": 44735, "loss": 3.6026870727539064, "grad_norm": 0.6734400987625122, "learning_rate": 0.0001, "elbo": 3.60269, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6234459300961763}
{"step": 44740, "loss": 3.6176712036132814, "grad_norm": 0.724320113658905, "learning_rate": 0.0001, "elbo": 3.61767, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.623739150832747}
{"step": 44745, "loss": 3.5912681579589845, "grad_norm": 0.6908040046691895, "learning_rate": 0.0001, "elbo": 3.59127, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6240323715693172}
{"step": 44750, "loss": 3.6005062103271483, "grad_norm": 0.982627809047699, "learning_rate": 0.0001, "elbo": 3.60051, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6243255923058877}
{"step": 44755, "loss": 3.6165061950683595, "grad_norm": 0.7512783408164978, "learning_rate": 0.0001, "elbo": 3.61651, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.624618813042458}
{"step": 44760, "loss": 3.6176586151123047, "grad_norm": 0.7469750642776489, "learning_rate": 0.0001, "elbo": 3.61766, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6249120337790286}
{"step": 44765, "loss": 3.5940841674804687, "grad_norm": 0.8088833093643188, "learning_rate": 0.0001, "elbo": 3.59408, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6252052545155995}
{"step": 44770, "loss": 3.6188865661621095, "grad_norm": 0.7906985878944397, "learning_rate": 0.0001, "elbo": 3.61889, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.62549847525217}
{"step": 44775, "loss": 3.6085586547851562, "grad_norm": 0.8428199887275696, "learning_rate": 0.0001, "elbo": 3.60856, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6257916959887404}
{"step": 44780, "loss": 3.5893592834472656, "grad_norm": 0.7206010818481445, "learning_rate": 0.0001, "elbo": 3.58936, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.626084916725311}
{"step": 44785, "loss": 3.607789993286133, "grad_norm": 0.732965350151062, "learning_rate": 0.0001, "elbo": 3.60779, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6263781374618813}
{"step": 44790, "loss": 3.619919204711914, "grad_norm": 0.6779178380966187, "learning_rate": 0.0001, "elbo": 3.61992, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6266713581984518}
{"step": 44795, "loss": 3.6053211212158205, "grad_norm": 0.6598983407020569, "learning_rate": 0.0001, "elbo": 3.60532, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.626964578935022}
{"step": 44800, "loss": 3.6122406005859373, "grad_norm": 0.9699524641036987, "learning_rate": 0.0001, "elbo": 3.61224, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6272577996715927}
{"step": 44805, "loss": 3.6114501953125, "grad_norm": 0.7044594287872314, "learning_rate": 0.0001, "elbo": 3.61145, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.627551020408163}
{"step": 44810, "loss": 3.5966793060302735, "grad_norm": 0.716731071472168, "learning_rate": 0.0001, "elbo": 3.59668, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.627844241144734}
{"step": 44815, "loss": 3.597136306762695, "grad_norm": 0.6623731851577759, "learning_rate": 0.0001, "elbo": 3.59714, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6281374618813045}
{"step": 44820, "loss": 3.611695098876953, "grad_norm": 0.7217210531234741, "learning_rate": 0.0001, "elbo": 3.61169, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.628430682617875}
{"step": 44825, "loss": 3.613123321533203, "grad_norm": 0.9300522208213806, "learning_rate": 0.0001, "elbo": 3.61312, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6287239033544454}
{"step": 44830, "loss": 3.6237552642822264, "grad_norm": 1.1690195798873901, "learning_rate": 0.0001, "elbo": 3.62376, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.629017124091016}
{"step": 44835, "loss": 3.6067161560058594, "grad_norm": 0.7135194540023804, "learning_rate": 0.0001, "elbo": 3.60672, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6293103448275863}
{"step": 44840, "loss": 3.6072654724121094, "grad_norm": 0.6544973254203796, "learning_rate": 0.0001, "elbo": 3.60727, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6296035655641568}
{"step": 44840, "loss": 3.6072654724121094, "grad_norm": 0.6544973254203796, "learning_rate": 0.0001, "elbo": 3.60727, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6296035655641568}
{"step": 44845, "loss": 3.608206558227539, "grad_norm": 0.6413483619689941, "learning_rate": 0.0001, "elbo": 3.60821, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.629896786300727}
{"step": 44850, "loss": 3.6103111267089845, "grad_norm": 0.661287784576416, "learning_rate": 0.0001, "elbo": 3.61031, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6301900070372977}
{"step": 44855, "loss": 3.616546630859375, "grad_norm": 0.6684132218360901, "learning_rate": 0.0001, "elbo": 3.61655, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.630483227773868}
{"step": 44860, "loss": 3.6061439514160156, "grad_norm": 0.6638808250427246, "learning_rate": 0.0001, "elbo": 3.60614, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6307764485104386}
{"step": 44865, "loss": 3.6082107543945314, "grad_norm": 0.6541885137557983, "learning_rate": 0.0001, "elbo": 3.60821, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.631069669247009}
{"step": 44870, "loss": 3.594987487792969, "grad_norm": 0.9811181426048279, "learning_rate": 0.0001, "elbo": 3.59499, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6313628899835795}
{"step": 44875, "loss": 3.6208518981933593, "grad_norm": 1.0543560981750488, "learning_rate": 0.0001, "elbo": 3.62085, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.63165611072015}
{"step": 44880, "loss": 3.596147155761719, "grad_norm": 0.6522163152694702, "learning_rate": 0.0001, "elbo": 3.59615, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6319493314567204}
{"step": 44885, "loss": 3.5942306518554688, "grad_norm": 0.6710959672927856, "learning_rate": 0.0001, "elbo": 3.59423, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.632242552193291}
{"step": 44890, "loss": 3.618288040161133, "grad_norm": 0.7130109071731567, "learning_rate": 0.0001, "elbo": 3.61829, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6325357729298617}
{"step": 44895, "loss": 3.6017997741699217, "grad_norm": 0.6442502737045288, "learning_rate": 0.0001, "elbo": 3.6018, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.632828993666432}
{"step": 44900, "loss": 3.6031719207763673, "grad_norm": 0.666006863117218, "learning_rate": 0.0001, "elbo": 3.60317, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6331222144030026}
{"step": 44905, "loss": 3.6074790954589844, "grad_norm": 0.6818377375602722, "learning_rate": 0.0001, "elbo": 3.60748, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.633415435139573}
{"step": 44910, "loss": 3.5943019866943358, "grad_norm": 0.6959215402603149, "learning_rate": 0.0001, "elbo": 3.5943, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6337086558761436}
{"step": 44915, "loss": 3.61168212890625, "grad_norm": 1.1245452165603638, "learning_rate": 0.0001, "elbo": 3.61168, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.634001876612714}
{"step": 44920, "loss": 3.6042877197265626, "grad_norm": 0.7946423888206482, "learning_rate": 0.0001, "elbo": 3.60429, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6342950973492845}
{"step": 44925, "loss": 3.62716064453125, "grad_norm": 0.6979209184646606, "learning_rate": 0.0001, "elbo": 3.62716, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.634588318085855}
{"step": 44930, "loss": 3.611563873291016, "grad_norm": 0.7160153985023499, "learning_rate": 0.0001, "elbo": 3.61156, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6348815388224254}
{"step": 44935, "loss": 3.599412536621094, "grad_norm": 0.7560552954673767, "learning_rate": 0.0001, "elbo": 3.59941, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6351747595589963}
{"step": 44940, "loss": 3.6101318359375, "grad_norm": 0.8421555757522583, "learning_rate": 0.0001, "elbo": 3.61013, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6354679802955667}
{"step": 44945, "loss": 3.5971611022949217, "grad_norm": 0.6827279329299927, "learning_rate": 0.0001, "elbo": 3.59716, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.635761201032137}
{"step": 44950, "loss": 3.623554992675781, "grad_norm": 0.6931662559509277, "learning_rate": 0.0001, "elbo": 3.62355, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6360544217687076}
{"step": 44955, "loss": 3.603997802734375, "grad_norm": 0.7606465816497803, "learning_rate": 0.0001, "elbo": 3.604, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.636347642505278}
{"step": 44960, "loss": 3.604904556274414, "grad_norm": 0.7470930814743042, "learning_rate": 0.0001, "elbo": 3.6049, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6366408632418485}
{"step": 44965, "loss": 3.6136184692382813, "grad_norm": 0.6733896136283875, "learning_rate": 0.0001, "elbo": 3.61362, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.636934083978419}
{"step": 44970, "loss": 3.6028919219970703, "grad_norm": 0.6616378426551819, "learning_rate": 0.0001, "elbo": 3.60289, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6372273047149895}
{"step": 44975, "loss": 3.6124034881591798, "grad_norm": 0.6536853909492493, "learning_rate": 0.0001, "elbo": 3.6124, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.63752052545156}
{"step": 44980, "loss": 3.600840759277344, "grad_norm": 0.712608277797699, "learning_rate": 0.0001, "elbo": 3.60084, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6378137461881304}
{"step": 44980, "loss": 3.600840759277344, "grad_norm": 0.712608277797699, "learning_rate": 0.0001, "elbo": 3.60084, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6378137461881304}
{"step": 44985, "loss": 3.604644012451172, "grad_norm": 0.8182690143585205, "learning_rate": 0.0001, "elbo": 3.60464, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.638106966924701}
{"step": 44990, "loss": 3.6134872436523438, "grad_norm": 0.7002381682395935, "learning_rate": 0.0001, "elbo": 3.61349, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6384001876612713}
{"step": 44995, "loss": 3.602674865722656, "grad_norm": 0.6715835928916931, "learning_rate": 0.0001, "elbo": 3.60267, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6386934083978417}
{"step": 44995, "loss": 3.602674865722656, "grad_norm": 0.6715835928916931, "learning_rate": 0.0001, "elbo": 3.60267, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6386934083978417}
{"step": 45000, "loss": 3.6289138793945312, "grad_norm": 0.6724594831466675, "learning_rate": 0.0001, "elbo": 3.62891, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.638986629134412}
{"step": 45005, "loss": 3.614129638671875, "grad_norm": 0.6505236625671387, "learning_rate": 0.0001, "elbo": 3.61413, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6392798498709826}
{"step": 45010, "loss": 3.6072982788085937, "grad_norm": 0.7217661738395691, "learning_rate": 0.0001, "elbo": 3.6073, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6395730706075535}
{"step": 45015, "loss": 3.6075424194335937, "grad_norm": 0.7944783568382263, "learning_rate": 0.0001, "elbo": 3.60754, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.639866291344124}
{"step": 45020, "loss": 3.591107940673828, "grad_norm": 0.6842184662818909, "learning_rate": 0.0001, "elbo": 3.59111, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6401595120806944}
{"step": 45025, "loss": 3.6122055053710938, "grad_norm": 0.6998947262763977, "learning_rate": 0.0001, "elbo": 3.61221, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.640452732817265}
{"step": 45030, "loss": 3.583913040161133, "grad_norm": 0.7022619843482971, "learning_rate": 0.0001, "elbo": 3.58391, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6407459535538353}
{"step": 45035, "loss": 3.5959266662597655, "grad_norm": 0.7545560002326965, "learning_rate": 0.0001, "elbo": 3.59593, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.641039174290406}
{"step": 45040, "loss": 3.5877403259277343, "grad_norm": 0.7000869512557983, "learning_rate": 0.0001, "elbo": 3.58774, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6413323950269763}
{"step": 45045, "loss": 3.604092025756836, "grad_norm": 0.664497971534729, "learning_rate": 0.0001, "elbo": 3.60409, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6416256157635467}
{"step": 45050, "loss": 3.6104244232177733, "grad_norm": 0.6422967314720154, "learning_rate": 0.0001, "elbo": 3.61042, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.641918836500117}
{"step": 45055, "loss": 3.6173885345458983, "grad_norm": 0.9817887544631958, "learning_rate": 0.0001, "elbo": 3.61739, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6422120572366876}
{"step": 45060, "loss": 3.6146896362304686, "grad_norm": 0.6840200424194336, "learning_rate": 0.0001, "elbo": 3.61469, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6425052779732585}
{"step": 45065, "loss": 3.6077972412109376, "grad_norm": 0.7149689197540283, "learning_rate": 0.0001, "elbo": 3.6078, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.642798498709829}
{"step": 45070, "loss": 3.623470687866211, "grad_norm": 0.7498405575752258, "learning_rate": 0.0001, "elbo": 3.62347, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6430917194463994}
{"step": 45075, "loss": 3.6049530029296877, "grad_norm": 0.7255528569221497, "learning_rate": 0.0001, "elbo": 3.60495, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.64338494018297}
{"step": 45080, "loss": 3.6112632751464844, "grad_norm": 0.7019152045249939, "learning_rate": 0.0001, "elbo": 3.61126, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6436781609195403}
{"step": 45085, "loss": 3.6214439392089846, "grad_norm": 0.6833881735801697, "learning_rate": 0.0001, "elbo": 3.62144, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.643971381656111}
{"step": 45090, "loss": 3.6074928283691405, "grad_norm": 0.6574239730834961, "learning_rate": 0.0001, "elbo": 3.60749, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6442646023926812}
{"step": 45095, "loss": 3.6204692840576174, "grad_norm": 0.701617956161499, "learning_rate": 0.0001, "elbo": 3.62047, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6445578231292517}
{"step": 45100, "loss": 3.5909061431884766, "grad_norm": 0.6975942254066467, "learning_rate": 0.0001, "elbo": 3.59091, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.644851043865822}
{"step": 45105, "loss": 3.6001243591308594, "grad_norm": 0.7323009967803955, "learning_rate": 0.0001, "elbo": 3.60012, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6451442646023926}
{"step": 45105, "loss": 3.6001243591308594, "grad_norm": 0.7323009967803955, "learning_rate": 0.0001, "elbo": 3.60012, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6451442646023926}
{"step": 45110, "loss": 3.6154624938964846, "grad_norm": 0.7699158787727356, "learning_rate": 0.0001, "elbo": 3.61546, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.645437485338963}
{"step": 45110, "loss": 3.6154624938964846, "grad_norm": 0.7699158787727356, "learning_rate": 0.0001, "elbo": 3.61546, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.645437485338963}
{"step": 45115, "loss": 3.5996837615966797, "grad_norm": 1.2096341848373413, "learning_rate": 0.0001, "elbo": 3.59968, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6457307060755335}
{"step": 45120, "loss": 3.6015296936035157, "grad_norm": 0.6510022878646851, "learning_rate": 0.0001, "elbo": 3.60153, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.646023926812104}
{"step": 45125, "loss": 3.5978588104248046, "grad_norm": 0.6967028975486755, "learning_rate": 0.0001, "elbo": 3.59786, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6463171475486744}
{"step": 45130, "loss": 3.614177703857422, "grad_norm": 0.6913596391677856, "learning_rate": 0.0001, "elbo": 3.61418, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.646610368285245}
{"step": 45135, "loss": 3.6098094940185548, "grad_norm": 0.681111216545105, "learning_rate": 0.0001, "elbo": 3.60981, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6469035890218158}
{"step": 45140, "loss": 3.6168617248535155, "grad_norm": 0.6665874123573303, "learning_rate": 0.0001, "elbo": 3.61686, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6471968097583862}
{"step": 45145, "loss": 3.5945976257324217, "grad_norm": 0.6747183799743652, "learning_rate": 0.0001, "elbo": 3.5946, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6474900304949567}
{"step": 45150, "loss": 3.6261756896972654, "grad_norm": 0.6742839217185974, "learning_rate": 0.0001, "elbo": 3.62618, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.647783251231527}
{"step": 45155, "loss": 3.6099777221679688, "grad_norm": 0.6566476821899414, "learning_rate": 0.0001, "elbo": 3.60998, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6480764719680976}
{"step": 45160, "loss": 3.621946334838867, "grad_norm": 0.9225967526435852, "learning_rate": 0.0001, "elbo": 3.62195, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.648369692704668}
{"step": 45165, "loss": 3.5937732696533202, "grad_norm": 0.6526134610176086, "learning_rate": 0.0001, "elbo": 3.59377, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6486629134412385}
{"step": 45170, "loss": 3.6233123779296874, "grad_norm": 0.7915804386138916, "learning_rate": 0.0001, "elbo": 3.62331, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.648956134177809}
{"step": 45175, "loss": 3.6070182800292967, "grad_norm": 0.7697131037712097, "learning_rate": 0.0001, "elbo": 3.60702, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6492493549143794}
{"step": 45180, "loss": 3.616289520263672, "grad_norm": 0.7245194315910339, "learning_rate": 0.0001, "elbo": 3.61629, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6495425756509503}
{"step": 45185, "loss": 3.586016845703125, "grad_norm": 0.6844515204429626, "learning_rate": 0.0001, "elbo": 3.58602, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6498357963875208}
{"step": 45190, "loss": 3.6183486938476563, "grad_norm": 1.0810673236846924, "learning_rate": 0.0001, "elbo": 3.61835, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.650129017124091}
{"step": 45195, "loss": 3.590464401245117, "grad_norm": 0.6640384793281555, "learning_rate": 0.0001, "elbo": 3.59046, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6504222378606617}
{"step": 45200, "loss": 3.6073780059814453, "grad_norm": 0.7048245072364807, "learning_rate": 0.0001, "elbo": 3.60738, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.650715458597232}
{"step": 45205, "loss": 3.6087806701660154, "grad_norm": 0.6657029390335083, "learning_rate": 0.0001, "elbo": 3.60878, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6510086793338026}
{"step": 45210, "loss": 3.617115783691406, "grad_norm": 0.7036006450653076, "learning_rate": 0.0001, "elbo": 3.61712, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.651301900070373}
{"step": 45215, "loss": 3.5917404174804686, "grad_norm": 0.7023493647575378, "learning_rate": 0.0001, "elbo": 3.59174, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6515951208069435}
{"step": 45220, "loss": 3.616349792480469, "grad_norm": 1.0211597681045532, "learning_rate": 0.0001, "elbo": 3.61635, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.651888341543514}
{"step": 45225, "loss": 3.607676696777344, "grad_norm": 0.9633239507675171, "learning_rate": 0.0001, "elbo": 3.60768, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6521815622800844}
{"step": 45230, "loss": 3.605188751220703, "grad_norm": 0.7658601999282837, "learning_rate": 0.0001, "elbo": 3.60519, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.652474783016655}
{"step": 45235, "loss": 3.605768585205078, "grad_norm": 0.7142235040664673, "learning_rate": 0.0001, "elbo": 3.60577, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6527680037532253}
{"step": 45240, "loss": 3.607832336425781, "grad_norm": 0.7024215459823608, "learning_rate": 0.0001, "elbo": 3.60783, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6530612244897958}
{"step": 45245, "loss": 3.6188591003417967, "grad_norm": 0.8266618847846985, "learning_rate": 0.0001, "elbo": 3.61886, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.653354445226366}
{"step": 45250, "loss": 3.606614685058594, "grad_norm": 0.6509185433387756, "learning_rate": 0.0001, "elbo": 3.60661, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6536476659629367}
{"step": 45255, "loss": 3.599315643310547, "grad_norm": 0.7091192007064819, "learning_rate": 0.0001, "elbo": 3.59932, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.653940886699507}
{"step": 45260, "loss": 3.617023468017578, "grad_norm": 0.6735550165176392, "learning_rate": 0.0001, "elbo": 3.61702, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.654234107436078}
{"step": 45265, "loss": 3.6090431213378906, "grad_norm": 0.6727468371391296, "learning_rate": 0.0001, "elbo": 3.60904, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6545273281726485}
{"step": 45265, "loss": 3.6090431213378906, "grad_norm": 0.6727468371391296, "learning_rate": 0.0001, "elbo": 3.60904, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6545273281726485}
{"step": 45270, "loss": 3.606526565551758, "grad_norm": 0.7607765197753906, "learning_rate": 0.0001, "elbo": 3.60653, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.654820548909219}
{"step": 45275, "loss": 3.6121524810791015, "grad_norm": 0.6848108768463135, "learning_rate": 0.0001, "elbo": 3.61215, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6551137696457894}
{"step": 45280, "loss": 3.594729995727539, "grad_norm": 0.7579708099365234, "learning_rate": 0.0001, "elbo": 3.59473, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.65540699038236}
{"step": 45285, "loss": 3.6199924468994142, "grad_norm": 0.7230032682418823, "learning_rate": 0.0001, "elbo": 3.61999, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6557002111189303}
{"step": 45290, "loss": 3.604584503173828, "grad_norm": 0.6396931409835815, "learning_rate": 0.0001, "elbo": 3.60458, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6559934318555007}
{"step": 45295, "loss": 3.6031539916992186, "grad_norm": 0.7265301942825317, "learning_rate": 0.0001, "elbo": 3.60315, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.656286652592071}
{"step": 45295, "loss": 3.6031539916992186, "grad_norm": 0.7265301942825317, "learning_rate": 0.0001, "elbo": 3.60315, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.656286652592071}
{"step": 45300, "loss": 3.592246246337891, "grad_norm": 0.706092119216919, "learning_rate": 0.0001, "elbo": 3.59225, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6565798733286416}
{"step": 45305, "loss": 3.6134872436523438, "grad_norm": 0.6853159070014954, "learning_rate": 0.0001, "elbo": 3.61349, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6568730940652125}
{"step": 45310, "loss": 3.589679718017578, "grad_norm": 0.6853129267692566, "learning_rate": 0.0001, "elbo": 3.58968, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.657166314801783}
{"step": 45315, "loss": 3.6021648406982423, "grad_norm": 0.7165477871894836, "learning_rate": 0.0001, "elbo": 3.60216, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6574595355383535}
{"step": 45320, "loss": 3.619477081298828, "grad_norm": 0.6818864345550537, "learning_rate": 0.0001, "elbo": 3.61948, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.657752756274924}
{"step": 45325, "loss": 3.611349105834961, "grad_norm": 0.7428968548774719, "learning_rate": 0.0001, "elbo": 3.61135, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6580459770114944}
{"step": 45330, "loss": 3.6026588439941407, "grad_norm": 0.7172117829322815, "learning_rate": 0.0001, "elbo": 3.60266, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.658339197748065}
{"step": 45335, "loss": 3.614437484741211, "grad_norm": 0.6775642037391663, "learning_rate": 0.0001, "elbo": 3.61444, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6586324184846353}
{"step": 45340, "loss": 3.5943538665771486, "grad_norm": 0.698329746723175, "learning_rate": 0.0001, "elbo": 3.59435, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6589256392212057}
{"step": 45345, "loss": 3.6125152587890623, "grad_norm": 0.7347172498703003, "learning_rate": 0.0001, "elbo": 3.61252, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.659218859957776}
{"step": 45350, "loss": 3.594426727294922, "grad_norm": 0.7382082343101501, "learning_rate": 0.0001, "elbo": 3.59443, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6595120806943466}
{"step": 45355, "loss": 3.6171291351318358, "grad_norm": 0.6872194409370422, "learning_rate": 0.0001, "elbo": 3.61713, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.659805301430917}
{"step": 45360, "loss": 3.5976787567138673, "grad_norm": 0.7209364175796509, "learning_rate": 0.0001, "elbo": 3.59768, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6600985221674875}
{"step": 45365, "loss": 3.617540740966797, "grad_norm": 0.7157998085021973, "learning_rate": 0.0001, "elbo": 3.61754, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.660391742904058}
{"step": 45370, "loss": 3.6011283874511717, "grad_norm": 0.7134227156639099, "learning_rate": 0.0001, "elbo": 3.60113, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6606849636406285}
{"step": 45375, "loss": 3.6129714965820314, "grad_norm": 0.7528843879699707, "learning_rate": 0.0001, "elbo": 3.61297, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.660978184377199}
{"step": 45380, "loss": 3.5990196228027345, "grad_norm": 0.7009139060974121, "learning_rate": 0.0001, "elbo": 3.59902, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.66127140511377}
{"step": 45385, "loss": 3.6161483764648437, "grad_norm": 0.7242875099182129, "learning_rate": 0.0001, "elbo": 3.61615, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6615646258503403}
{"step": 45390, "loss": 3.59935302734375, "grad_norm": 0.6786954998970032, "learning_rate": 0.0001, "elbo": 3.59935, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6618578465869107}
{"step": 45395, "loss": 3.606135940551758, "grad_norm": 0.6679940819740295, "learning_rate": 0.0001, "elbo": 3.60614, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.662151067323481}
{"step": 45400, "loss": 3.5990974426269533, "grad_norm": 0.7105554938316345, "learning_rate": 0.0001, "elbo": 3.5991, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6624442880600516}
{"step": 45405, "loss": 3.617871856689453, "grad_norm": 0.7403272390365601, "learning_rate": 0.0001, "elbo": 3.61787, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.662737508796622}
{"step": 45410, "loss": 3.607202911376953, "grad_norm": 0.7580627202987671, "learning_rate": 0.0001, "elbo": 3.6072, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6630307295331925}
{"step": 45415, "loss": 3.586811065673828, "grad_norm": 0.7256088852882385, "learning_rate": 0.0001, "elbo": 3.58681, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.663323950269763}
{"step": 45420, "loss": 3.6001968383789062, "grad_norm": 0.6789301037788391, "learning_rate": 0.0001, "elbo": 3.6002, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6636171710063334}
{"step": 45425, "loss": 3.592355728149414, "grad_norm": 0.7462928295135498, "learning_rate": 0.0001, "elbo": 3.59236, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.663910391742904}
{"step": 45430, "loss": 3.596266174316406, "grad_norm": 0.6636849045753479, "learning_rate": 0.0001, "elbo": 3.59627, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.664203612479475}
{"step": 45435, "loss": 3.616144561767578, "grad_norm": 0.6521018147468567, "learning_rate": 0.0001, "elbo": 3.61614, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6644968332160452}
{"step": 45440, "loss": 3.5958045959472655, "grad_norm": 0.6896435618400574, "learning_rate": 0.0001, "elbo": 3.5958, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6647900539526157}
{"step": 45445, "loss": 3.589068603515625, "grad_norm": 0.6642091870307922, "learning_rate": 0.0001, "elbo": 3.58907, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.665083274689186}
{"step": 45450, "loss": 3.6058628082275392, "grad_norm": 0.7416906952857971, "learning_rate": 0.0001, "elbo": 3.60586, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6653764954257566}
{"step": 45455, "loss": 3.5975723266601562, "grad_norm": 1.0323108434677124, "learning_rate": 0.0001, "elbo": 3.59757, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.665669716162327}
{"step": 45460, "loss": 3.6000129699707033, "grad_norm": 0.7869197130203247, "learning_rate": 0.0001, "elbo": 3.60001, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6659629368988975}
{"step": 45465, "loss": 3.6098628997802735, "grad_norm": 0.6968251466751099, "learning_rate": 0.0001, "elbo": 3.60986, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.666256157635468}
{"step": 45470, "loss": 3.5974960327148438, "grad_norm": 0.7098475694656372, "learning_rate": 0.0001, "elbo": 3.5975, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6665493783720384}
{"step": 45475, "loss": 3.585038757324219, "grad_norm": 0.6824154853820801, "learning_rate": 0.0001, "elbo": 3.58504, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.666842599108609}
{"step": 45480, "loss": 3.6049514770507813, "grad_norm": 0.8973789215087891, "learning_rate": 0.0001, "elbo": 3.60495, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6671358198451793}
{"step": 45485, "loss": 3.604658508300781, "grad_norm": 1.2628475427627563, "learning_rate": 0.0001, "elbo": 3.60466, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.66742904058175}
{"step": 45490, "loss": 3.6104057312011717, "grad_norm": 0.6507744789123535, "learning_rate": 0.0001, "elbo": 3.61041, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6677222613183202}
{"step": 45495, "loss": 3.593849945068359, "grad_norm": 0.6522022485733032, "learning_rate": 0.0001, "elbo": 3.59385, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6680154820548907}
{"step": 45500, "loss": 3.604714202880859, "grad_norm": 0.842038631439209, "learning_rate": 0.0001, "elbo": 3.60471, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.668308702791461}
{"step": 45505, "loss": 3.5985198974609376, "grad_norm": 0.7703525424003601, "learning_rate": 0.0001, "elbo": 3.59852, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.668601923528032}
{"step": 45510, "loss": 3.5819992065429687, "grad_norm": 0.7436431646347046, "learning_rate": 0.0001, "elbo": 3.582, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6688951442646025}
{"step": 45515, "loss": 3.6079662322998045, "grad_norm": 0.6567676067352295, "learning_rate": 0.0001, "elbo": 3.60797, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.669188365001173}
{"step": 45520, "loss": 3.6037582397460937, "grad_norm": 0.6795552968978882, "learning_rate": 0.0001, "elbo": 3.60376, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6694815857377434}
{"step": 45525, "loss": 3.6017990112304688, "grad_norm": 0.8378040194511414, "learning_rate": 0.0001, "elbo": 3.6018, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.669774806474314}
{"step": 45530, "loss": 3.621847152709961, "grad_norm": 0.6918944120407104, "learning_rate": 0.0001, "elbo": 3.62185, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6700680272108843}
{"step": 45535, "loss": 3.6090755462646484, "grad_norm": 0.6635600328445435, "learning_rate": 0.0001, "elbo": 3.60908, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6703612479474548}
{"step": 45540, "loss": 3.5854026794433596, "grad_norm": 0.6569610238075256, "learning_rate": 0.0001, "elbo": 3.5854, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6706544686840252}
{"step": 45545, "loss": 3.595087432861328, "grad_norm": 0.6977202892303467, "learning_rate": 0.0001, "elbo": 3.59509, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6709476894205957}
{"step": 45550, "loss": 3.5924957275390623, "grad_norm": 0.8074080348014832, "learning_rate": 0.0001, "elbo": 3.5925, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6712409101571666}
{"step": 45555, "loss": 3.602731704711914, "grad_norm": 0.7140382528305054, "learning_rate": 0.0001, "elbo": 3.60273, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.671534130893737}
{"step": 45560, "loss": 3.617144012451172, "grad_norm": 0.7068063616752625, "learning_rate": 0.0001, "elbo": 3.61714, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6718273516303075}
{"step": 45565, "loss": 3.600995635986328, "grad_norm": 0.7160325050354004, "learning_rate": 0.0001, "elbo": 3.601, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.672120572366878}
{"step": 45565, "loss": 3.600995635986328, "grad_norm": 0.7160325050354004, "learning_rate": 0.0001, "elbo": 3.601, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.672120572366878}
{"step": 45570, "loss": 3.598550033569336, "grad_norm": 0.6881916522979736, "learning_rate": 0.0001, "elbo": 3.59855, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6724137931034484}
{"step": 45575, "loss": 3.5894054412841796, "grad_norm": 0.6975864768028259, "learning_rate": 0.0001, "elbo": 3.58941, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.672707013840019}
{"step": 45580, "loss": 3.6032459259033205, "grad_norm": 0.664553165435791, "learning_rate": 0.0001, "elbo": 3.60325, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6730002345765893}
{"step": 45585, "loss": 3.611250305175781, "grad_norm": 0.739383339881897, "learning_rate": 0.0001, "elbo": 3.61125, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6732934553131598}
{"step": 45590, "loss": 3.6015560150146486, "grad_norm": 0.6948650479316711, "learning_rate": 0.0001, "elbo": 3.60156, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.67358667604973}
{"step": 45595, "loss": 3.6034351348876954, "grad_norm": 0.6911669373512268, "learning_rate": 0.0001, "elbo": 3.60344, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6738798967863007}
{"step": 45600, "loss": 3.6129859924316405, "grad_norm": 0.7538306713104248, "learning_rate": 0.0001, "elbo": 3.61299, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.674173117522871}
{"step": 45605, "loss": 3.595874786376953, "grad_norm": 0.7260775566101074, "learning_rate": 0.0001, "elbo": 3.59588, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6744663382594416}
{"step": 45610, "loss": 3.6002159118652344, "grad_norm": 0.724595844745636, "learning_rate": 0.0001, "elbo": 3.60022, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.674759558996012}
{"step": 45615, "loss": 3.6140079498291016, "grad_norm": 0.6864385008811951, "learning_rate": 0.0001, "elbo": 3.61401, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6750527797325825}
{"step": 45620, "loss": 3.614907455444336, "grad_norm": 0.8297668099403381, "learning_rate": 0.0001, "elbo": 3.61491, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.675346000469153}
{"step": 45625, "loss": 3.5942893981933595, "grad_norm": 0.7009674310684204, "learning_rate": 0.0001, "elbo": 3.59429, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6756392212057234}
{"step": 45630, "loss": 3.580083465576172, "grad_norm": 0.6621953248977661, "learning_rate": 0.0001, "elbo": 3.58008, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6759324419422943}
{"step": 45635, "loss": 3.606121826171875, "grad_norm": 0.6648790836334229, "learning_rate": 0.0001, "elbo": 3.60612, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6762256626788647}
{"step": 45640, "loss": 3.607220458984375, "grad_norm": 0.7040189504623413, "learning_rate": 0.0001, "elbo": 3.60722, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.676518883415435}
{"step": 45645, "loss": 3.609252166748047, "grad_norm": 0.8968525528907776, "learning_rate": 0.0001, "elbo": 3.60925, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6768121041520057}
{"step": 45650, "loss": 3.599026107788086, "grad_norm": 0.6947530508041382, "learning_rate": 0.0001, "elbo": 3.59903, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.677105324888576}
{"step": 45655, "loss": 3.6078460693359373, "grad_norm": 0.7273105382919312, "learning_rate": 0.0001, "elbo": 3.60785, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6773985456251466}
{"step": 45660, "loss": 3.5862495422363283, "grad_norm": 0.6834023594856262, "learning_rate": 0.0001, "elbo": 3.58625, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.677691766361717}
{"step": 45665, "loss": 3.5918704986572267, "grad_norm": 0.6742890477180481, "learning_rate": 0.0001, "elbo": 3.59187, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6779849870982875}
{"step": 45670, "loss": 3.607832336425781, "grad_norm": 0.7273544073104858, "learning_rate": 0.0001, "elbo": 3.60783, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.678278207834858}
{"step": 45675, "loss": 3.619397735595703, "grad_norm": 0.698253333568573, "learning_rate": 0.0001, "elbo": 3.6194, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.678571428571429}
{"step": 45680, "loss": 3.6133365631103516, "grad_norm": 0.7039133310317993, "learning_rate": 0.0001, "elbo": 3.61334, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6788646493079993}
{"step": 45685, "loss": 3.5993637084960937, "grad_norm": 0.663338840007782, "learning_rate": 0.0001, "elbo": 3.59936, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6791578700445697}
{"step": 45690, "loss": 3.6154781341552735, "grad_norm": 0.795958399772644, "learning_rate": 0.0001, "elbo": 3.61548, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.67945109078114}
{"step": 45695, "loss": 3.6099166870117188, "grad_norm": 0.7150638103485107, "learning_rate": 0.0001, "elbo": 3.60992, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6797443115177106}
{"step": 45700, "loss": 3.5977630615234375, "grad_norm": 0.6865943074226379, "learning_rate": 0.0001, "elbo": 3.59776, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.680037532254281}
{"step": 45705, "loss": 3.5936927795410156, "grad_norm": 0.6670202016830444, "learning_rate": 0.0001, "elbo": 3.59369, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6803307529908516}
{"step": 45710, "loss": 3.5896286010742187, "grad_norm": 0.8144505620002747, "learning_rate": 0.0001, "elbo": 3.58963, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.680623973727422}
{"step": 45715, "loss": 3.6000247955322267, "grad_norm": 0.7456961274147034, "learning_rate": 0.0001, "elbo": 3.60002, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6809171944639925}
{"step": 45720, "loss": 3.611517333984375, "grad_norm": 0.6801279783248901, "learning_rate": 0.0001, "elbo": 3.61152, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.681210415200563}
{"step": 45725, "loss": 3.617768096923828, "grad_norm": 0.7022068500518799, "learning_rate": 0.0001, "elbo": 3.61777, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6815036359371334}
{"step": 45730, "loss": 3.6005653381347655, "grad_norm": 0.7181046009063721, "learning_rate": 0.0001, "elbo": 3.60057, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.681796856673704}
{"step": 45735, "loss": 3.6218421936035154, "grad_norm": 0.7470592260360718, "learning_rate": 0.0001, "elbo": 3.62184, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6820900774102743}
{"step": 45740, "loss": 3.6104804992675783, "grad_norm": 0.7168502807617188, "learning_rate": 0.0001, "elbo": 3.61048, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6823832981468447}
{"step": 45745, "loss": 3.605029296875, "grad_norm": 0.6566230058670044, "learning_rate": 0.0001, "elbo": 3.60503, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.682676518883415}
{"step": 45750, "loss": 3.5891700744628907, "grad_norm": 0.869500994682312, "learning_rate": 0.0001, "elbo": 3.58917, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.682969739619986}
{"step": 45755, "loss": 3.6152172088623047, "grad_norm": 0.6700111031532288, "learning_rate": 0.0001, "elbo": 3.61522, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6832629603565565}
{"step": 45760, "loss": 3.596965789794922, "grad_norm": 0.6740659475326538, "learning_rate": 0.0001, "elbo": 3.59697, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.683556181093127}
{"step": 45765, "loss": 3.61047477722168, "grad_norm": 0.6935853958129883, "learning_rate": 0.0001, "elbo": 3.61047, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6838494018296974}
{"step": 45770, "loss": 3.602799987792969, "grad_norm": 0.7874895334243774, "learning_rate": 0.0001, "elbo": 3.6028, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.684142622566268}
{"step": 45775, "loss": 3.6117084503173826, "grad_norm": 0.6950761079788208, "learning_rate": 0.0001, "elbo": 3.61171, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6844358433028384}
{"step": 45780, "loss": 3.5873207092285155, "grad_norm": 0.6609780192375183, "learning_rate": 0.0001, "elbo": 3.58732, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.684729064039409}
{"step": 45785, "loss": 3.582017517089844, "grad_norm": 0.6737627387046814, "learning_rate": 0.0001, "elbo": 3.58202, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6850222847759793}
{"step": 45790, "loss": 3.6103416442871095, "grad_norm": 0.7250356078147888, "learning_rate": 0.0001, "elbo": 3.61034, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6853155055125497}
{"step": 45795, "loss": 3.601213073730469, "grad_norm": 0.689928412437439, "learning_rate": 0.0001, "elbo": 3.60121, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6856087262491206}
{"step": 45800, "loss": 3.6089454650878907, "grad_norm": 0.6870595812797546, "learning_rate": 0.0001, "elbo": 3.60895, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.685901946985691}
{"step": 45805, "loss": 3.596657562255859, "grad_norm": 0.6820058226585388, "learning_rate": 0.0001, "elbo": 3.59666, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6861951677222615}
{"step": 45810, "loss": 3.594622039794922, "grad_norm": 0.735848605632782, "learning_rate": 0.0001, "elbo": 3.59462, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.686488388458832}
{"step": 45815, "loss": 3.6091564178466795, "grad_norm": 0.7466672658920288, "learning_rate": 0.0001, "elbo": 3.60916, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6867816091954024}
{"step": 45815, "loss": 3.6091564178466795, "grad_norm": 0.7466672658920288, "learning_rate": 0.0001, "elbo": 3.60916, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6867816091954024}
{"step": 45820, "loss": 3.603232192993164, "grad_norm": 0.6660481691360474, "learning_rate": 0.0001, "elbo": 3.60323, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.687074829931973}
{"step": 45825, "loss": 3.6000579833984374, "grad_norm": 0.7536413669586182, "learning_rate": 0.0001, "elbo": 3.60006, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6873680506685433}
{"step": 45830, "loss": 3.5881019592285157, "grad_norm": 0.7574118971824646, "learning_rate": 0.0001, "elbo": 3.5881, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.687661271405114}
{"step": 45835, "loss": 3.59881477355957, "grad_norm": 0.7332965731620789, "learning_rate": 0.0001, "elbo": 3.59881, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6879544921416842}
{"step": 45840, "loss": 3.604920196533203, "grad_norm": 0.7477477192878723, "learning_rate": 0.0001, "elbo": 3.60492, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6882477128782547}
{"step": 45845, "loss": 3.6083934783935545, "grad_norm": 0.6587011814117432, "learning_rate": 0.0001, "elbo": 3.60839, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.688540933614825}
{"step": 45850, "loss": 3.615337371826172, "grad_norm": 0.6563683748245239, "learning_rate": 0.0001, "elbo": 3.61534, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6888341543513956}
{"step": 45855, "loss": 3.6167842864990236, "grad_norm": 0.7365993857383728, "learning_rate": 0.0001, "elbo": 3.61678, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.689127375087966}
{"step": 45860, "loss": 3.6150707244873046, "grad_norm": 0.6189395785331726, "learning_rate": 0.0001, "elbo": 3.61507, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6894205958245365}
{"step": 45865, "loss": 3.6194190979003906, "grad_norm": 0.7066006064414978, "learning_rate": 0.0001, "elbo": 3.61942, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.689713816561107}
{"step": 45870, "loss": 3.6190685272216796, "grad_norm": 0.7167927622795105, "learning_rate": 0.0001, "elbo": 3.61907, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6900070372976774}
{"step": 45875, "loss": 3.6076568603515624, "grad_norm": 0.7114911079406738, "learning_rate": 0.0001, "elbo": 3.60766, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6903002580342483}
{"step": 45880, "loss": 3.608401870727539, "grad_norm": 0.6585972905158997, "learning_rate": 0.0001, "elbo": 3.6084, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.690593478770819}
{"step": 45885, "loss": 3.609114074707031, "grad_norm": 0.7495714426040649, "learning_rate": 0.0001, "elbo": 3.60911, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6908866995073892}
{"step": 45890, "loss": 3.6055767059326174, "grad_norm": 0.7302316427230835, "learning_rate": 0.0001, "elbo": 3.60558, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6911799202439597}
{"step": 45895, "loss": 3.600831985473633, "grad_norm": 0.7384791970252991, "learning_rate": 0.0001, "elbo": 3.60083, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.69147314098053}
{"step": 45900, "loss": 3.6086280822753904, "grad_norm": 0.875866174697876, "learning_rate": 0.0001, "elbo": 3.60863, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6917663617171006}
{"step": 45905, "loss": 3.601126861572266, "grad_norm": 0.7048518061637878, "learning_rate": 0.0001, "elbo": 3.60113, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.692059582453671}
{"step": 45910, "loss": 3.609157943725586, "grad_norm": 0.7650464773178101, "learning_rate": 0.0001, "elbo": 3.60916, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6923528031902415}
{"step": 45915, "loss": 3.606223297119141, "grad_norm": 0.6752452254295349, "learning_rate": 0.0001, "elbo": 3.60622, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.692646023926812}
{"step": 45920, "loss": 3.594403076171875, "grad_norm": 0.8254571557044983, "learning_rate": 0.0001, "elbo": 3.5944, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.692939244663383}
{"step": 45925, "loss": 3.6139595031738283, "grad_norm": 0.6525031328201294, "learning_rate": 0.0001, "elbo": 3.61396, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6932324653999533}
{"step": 45930, "loss": 3.591796875, "grad_norm": 0.7119786739349365, "learning_rate": 0.0001, "elbo": 3.5918, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6935256861365238}
{"step": 45935, "loss": 3.611728286743164, "grad_norm": 0.8118964433670044, "learning_rate": 0.0001, "elbo": 3.61173, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.693818906873094}
{"step": 45940, "loss": 3.598108673095703, "grad_norm": 0.8233667612075806, "learning_rate": 0.0001, "elbo": 3.59811, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6941121276096647}
{"step": 45945, "loss": 3.5791706085205077, "grad_norm": 0.664730429649353, "learning_rate": 0.0001, "elbo": 3.57917, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.694405348346235}
{"step": 45950, "loss": 3.5955665588378904, "grad_norm": 0.6438469886779785, "learning_rate": 0.0001, "elbo": 3.59557, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6946985690828056}
{"step": 45955, "loss": 3.6017379760742188, "grad_norm": 0.6993231773376465, "learning_rate": 0.0001, "elbo": 3.60174, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.694991789819376}
{"step": 45960, "loss": 3.6012222290039064, "grad_norm": 0.6559988856315613, "learning_rate": 0.0001, "elbo": 3.60122, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6952850105559465}
{"step": 45965, "loss": 3.5928253173828124, "grad_norm": 1.1135332584381104, "learning_rate": 0.0001, "elbo": 3.59283, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.695578231292517}
{"step": 45970, "loss": 3.587115478515625, "grad_norm": 0.6969377398490906, "learning_rate": 0.0001, "elbo": 3.58712, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6958714520290874}
{"step": 45975, "loss": 3.5952072143554688, "grad_norm": 0.9602000713348389, "learning_rate": 0.0001, "elbo": 3.59521, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.696164672765658}
{"step": 45980, "loss": 3.6025054931640623, "grad_norm": 0.7060055732727051, "learning_rate": 0.0001, "elbo": 3.60251, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6964578935022283}
{"step": 45985, "loss": 3.60082893371582, "grad_norm": 0.8873772025108337, "learning_rate": 0.0001, "elbo": 3.60083, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6967511142387988}
{"step": 45990, "loss": 3.6036041259765623, "grad_norm": 0.7393750548362732, "learning_rate": 0.0001, "elbo": 3.6036, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.697044334975369}
{"step": 45995, "loss": 3.582249069213867, "grad_norm": 0.6591447591781616, "learning_rate": 0.0001, "elbo": 3.58225, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6973375557119397}
{"step": 46000, "loss": 3.599794387817383, "grad_norm": 0.6530516743659973, "learning_rate": 0.0001, "elbo": 3.59979, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6976307764485106}
{"step": 46000, "loss": 3.599794387817383, "grad_norm": 0.6530516743659973, "learning_rate": 0.0001, "elbo": 3.59979, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6976307764485106}
{"step": 46005, "loss": 3.5893714904785154, "grad_norm": 0.7229706645011902, "learning_rate": 0.0001, "elbo": 3.58937, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.697923997185081}
{"step": 46010, "loss": 3.6055152893066404, "grad_norm": 0.7021849751472473, "learning_rate": 0.0001, "elbo": 3.60552, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6982172179216515}
{"step": 46015, "loss": 3.614866256713867, "grad_norm": 0.7078180909156799, "learning_rate": 0.0001, "elbo": 3.61487, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.698510438658222}
{"step": 46020, "loss": 3.587215042114258, "grad_norm": 0.6874021291732788, "learning_rate": 0.0001, "elbo": 3.58721, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6988036593947924}
{"step": 46025, "loss": 3.598357391357422, "grad_norm": 0.6513274312019348, "learning_rate": 0.0001, "elbo": 3.59836, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.699096880131363}
{"step": 46030, "loss": 3.6103267669677734, "grad_norm": 0.6917299628257751, "learning_rate": 0.0001, "elbo": 3.61033, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6993901008679333}
{"step": 46035, "loss": 3.594556427001953, "grad_norm": 0.8785936832427979, "learning_rate": 0.0001, "elbo": 3.59456, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.6996833216045037}
{"step": 46040, "loss": 3.6012649536132812, "grad_norm": 0.706139326095581, "learning_rate": 0.0001, "elbo": 3.60126, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.699976542341074}
{"step": 46045, "loss": 3.6125736236572266, "grad_norm": 0.6633727550506592, "learning_rate": 0.0001, "elbo": 3.61257, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.700269763077645}
{"step": 46050, "loss": 3.6064056396484374, "grad_norm": 0.7343655228614807, "learning_rate": 0.0001, "elbo": 3.60641, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7005629838142156}
{"step": 46055, "loss": 3.6040782928466797, "grad_norm": 0.7148258686065674, "learning_rate": 0.0001, "elbo": 3.60408, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.700856204550786}
{"step": 46060, "loss": 3.587939453125, "grad_norm": 0.7161760926246643, "learning_rate": 0.0001, "elbo": 3.58794, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7011494252873565}
{"step": 46065, "loss": 3.6055789947509767, "grad_norm": 0.7066383957862854, "learning_rate": 0.0001, "elbo": 3.60558, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.701442646023927}
{"step": 46070, "loss": 3.6085464477539064, "grad_norm": 0.7453451156616211, "learning_rate": 0.0001, "elbo": 3.60855, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7017358667604974}
{"step": 46075, "loss": 3.6158233642578126, "grad_norm": 0.7670539617538452, "learning_rate": 0.0001, "elbo": 3.61582, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.702029087497068}
{"step": 46080, "loss": 3.6007415771484377, "grad_norm": 0.7099767923355103, "learning_rate": 0.0001, "elbo": 3.60074, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7023223082336383}
{"step": 46085, "loss": 3.6124298095703127, "grad_norm": 0.8306111693382263, "learning_rate": 0.0001, "elbo": 3.61243, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7026155289702087}
{"step": 46090, "loss": 3.622259521484375, "grad_norm": 0.8806421160697937, "learning_rate": 0.0001, "elbo": 3.62226, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.702908749706779}
{"step": 46095, "loss": 3.6188087463378906, "grad_norm": 0.6924373507499695, "learning_rate": 0.0001, "elbo": 3.61881, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7032019704433496}
{"step": 46100, "loss": 3.6107551574707033, "grad_norm": 0.6815215945243835, "learning_rate": 0.0001, "elbo": 3.61076, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.70349519117992}
{"step": 46105, "loss": 3.5972930908203127, "grad_norm": 0.6856396198272705, "learning_rate": 0.0001, "elbo": 3.59729, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7037884119164906}
{"step": 46110, "loss": 3.5880455017089843, "grad_norm": 0.6820045113563538, "learning_rate": 0.0001, "elbo": 3.58805, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.704081632653061}
{"step": 46115, "loss": 3.610343170166016, "grad_norm": 0.6908042430877686, "learning_rate": 0.0001, "elbo": 3.61034, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7043748533896315}
{"step": 46120, "loss": 3.5955970764160154, "grad_norm": 0.6931967735290527, "learning_rate": 0.0001, "elbo": 3.5956, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7046680741262024}
{"step": 46125, "loss": 3.5909683227539064, "grad_norm": 0.7016478180885315, "learning_rate": 0.0001, "elbo": 3.59097, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.704961294862773}
{"step": 46130, "loss": 3.59954948425293, "grad_norm": 0.6731183528900146, "learning_rate": 0.0001, "elbo": 3.59955, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7052545155993433}
{"step": 46135, "loss": 3.598530960083008, "grad_norm": 0.7097012400627136, "learning_rate": 0.0001, "elbo": 3.59853, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7055477363359137}
{"step": 46140, "loss": 3.597119903564453, "grad_norm": 0.667020857334137, "learning_rate": 0.0001, "elbo": 3.59712, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.705840957072484}
{"step": 46145, "loss": 3.58660888671875, "grad_norm": 0.6799477338790894, "learning_rate": 0.0001, "elbo": 3.58661, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7061341778090546}
{"step": 46150, "loss": 3.6071407318115236, "grad_norm": 0.7382012605667114, "learning_rate": 0.0001, "elbo": 3.60714, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.706427398545625}
{"step": 46155, "loss": 3.5926971435546875, "grad_norm": 0.7281908392906189, "learning_rate": 0.0001, "elbo": 3.5927, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7067206192821955}
{"step": 46160, "loss": 3.6051620483398437, "grad_norm": 0.7050403952598572, "learning_rate": 0.0001, "elbo": 3.60516, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.707013840018766}
{"step": 46165, "loss": 3.6008033752441406, "grad_norm": 0.7096222043037415, "learning_rate": 0.0001, "elbo": 3.6008, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.707307060755337}
{"step": 46170, "loss": 3.5979774475097654, "grad_norm": 0.8144180178642273, "learning_rate": 0.0001, "elbo": 3.59798, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7076002814919073}
{"step": 46175, "loss": 3.6045150756835938, "grad_norm": 0.717898964881897, "learning_rate": 0.0001, "elbo": 3.60452, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.707893502228478}
{"step": 46180, "loss": 3.606568145751953, "grad_norm": 0.7252187132835388, "learning_rate": 0.0001, "elbo": 3.60657, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7081867229650483}
{"step": 46185, "loss": 3.6117767333984374, "grad_norm": 0.6676563620567322, "learning_rate": 0.0001, "elbo": 3.61178, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7084799437016187}
{"step": 46190, "loss": 3.614691162109375, "grad_norm": 0.7539612054824829, "learning_rate": 0.0001, "elbo": 3.61469, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.708773164438189}
{"step": 46195, "loss": 3.607146453857422, "grad_norm": 0.6559796929359436, "learning_rate": 0.0001, "elbo": 3.60715, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7090663851747596}
{"step": 46200, "loss": 3.600164031982422, "grad_norm": 0.7055538296699524, "learning_rate": 0.0001, "elbo": 3.60016, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.70935960591133}
{"step": 46205, "loss": 3.594931793212891, "grad_norm": 0.6825099587440491, "learning_rate": 0.0001, "elbo": 3.59493, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7096528266479005}
{"step": 46210, "loss": 3.6048828125, "grad_norm": 0.7100792527198792, "learning_rate": 0.0001, "elbo": 3.60488, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.709946047384471}
{"step": 46215, "loss": 3.6164409637451174, "grad_norm": 0.6453670263290405, "learning_rate": 0.0001, "elbo": 3.61644, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7102392681210414}
{"step": 46220, "loss": 3.5994796752929688, "grad_norm": 0.7026731967926025, "learning_rate": 0.0001, "elbo": 3.59948, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.710532488857612}
{"step": 46225, "loss": 3.599748229980469, "grad_norm": 0.7055178880691528, "learning_rate": 0.0001, "elbo": 3.59975, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7108257095941823}
{"step": 46230, "loss": 3.5990150451660154, "grad_norm": 0.6799114346504211, "learning_rate": 0.0001, "elbo": 3.59902, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.711118930330753}
{"step": 46235, "loss": 3.609251022338867, "grad_norm": 0.6475741863250732, "learning_rate": 0.0001, "elbo": 3.60925, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7114121510673233}
{"step": 46240, "loss": 3.6131092071533204, "grad_norm": 0.7393915057182312, "learning_rate": 0.0001, "elbo": 3.61311, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7117053718038937}
{"step": 46245, "loss": 3.606251907348633, "grad_norm": 0.7219039797782898, "learning_rate": 0.0001, "elbo": 3.60625, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7119985925404646}
{"step": 46250, "loss": 3.5898136138916015, "grad_norm": 0.7160125970840454, "learning_rate": 0.0001, "elbo": 3.58981, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.712291813277035}
{"step": 46255, "loss": 3.5929931640625, "grad_norm": 0.7115850448608398, "learning_rate": 0.0001, "elbo": 3.59299, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7125850340136055}
{"step": 46260, "loss": 3.603883743286133, "grad_norm": 0.680247962474823, "learning_rate": 0.0001, "elbo": 3.60388, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.712878254750176}
{"step": 46265, "loss": 3.610995864868164, "grad_norm": 0.6943657994270325, "learning_rate": 0.0001, "elbo": 3.611, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7131714754867464}
{"step": 46270, "loss": 3.5989002227783202, "grad_norm": 0.7312673926353455, "learning_rate": 0.0001, "elbo": 3.5989, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.713464696223317}
{"step": 46275, "loss": 3.6009361267089846, "grad_norm": 0.7000435590744019, "learning_rate": 0.0001, "elbo": 3.60094, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7137579169598873}
{"step": 46280, "loss": 3.607373809814453, "grad_norm": 0.8853266835212708, "learning_rate": 0.0001, "elbo": 3.60737, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.714051137696458}
{"step": 46285, "loss": 3.591200256347656, "grad_norm": 0.6874273419380188, "learning_rate": 0.0001, "elbo": 3.5912, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7143443584330282}
{"step": 46290, "loss": 3.607051467895508, "grad_norm": 0.7071090340614319, "learning_rate": 0.0001, "elbo": 3.60705, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.714637579169599}
{"step": 46295, "loss": 3.6037216186523438, "grad_norm": 0.7946973443031311, "learning_rate": 0.0001, "elbo": 3.60372, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7149307999061696}
{"step": 46300, "loss": 3.6236827850341795, "grad_norm": 0.7591429948806763, "learning_rate": 0.0001, "elbo": 3.62368, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.71522402064274}
{"step": 46305, "loss": 3.5874122619628905, "grad_norm": 0.6834484934806824, "learning_rate": 0.0001, "elbo": 3.58741, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7155172413793105}
{"step": 46310, "loss": 3.583964157104492, "grad_norm": 0.6650127172470093, "learning_rate": 0.0001, "elbo": 3.58396, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.715810462115881}
{"step": 46315, "loss": 3.6097999572753907, "grad_norm": 0.7022577524185181, "learning_rate": 0.0001, "elbo": 3.6098, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7161036828524514}
{"step": 46320, "loss": 3.597260665893555, "grad_norm": 0.6554110646247864, "learning_rate": 0.0001, "elbo": 3.59726, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.716396903589022}
{"step": 46325, "loss": 3.61123046875, "grad_norm": 0.6762685775756836, "learning_rate": 0.0001, "elbo": 3.61123, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7166901243255923}
{"step": 46330, "loss": 3.5862213134765626, "grad_norm": 0.814477801322937, "learning_rate": 0.0001, "elbo": 3.58622, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7169833450621628}
{"step": 46330, "loss": 3.5862213134765626, "grad_norm": 0.814477801322937, "learning_rate": 0.0001, "elbo": 3.58622, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7169833450621628}
{"step": 46335, "loss": 3.594432067871094, "grad_norm": 1.1873832941055298, "learning_rate": 0.0001, "elbo": 3.59443, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7172765657987332}
{"step": 46340, "loss": 3.6194664001464845, "grad_norm": 0.7236112952232361, "learning_rate": 0.0001, "elbo": 3.61947, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7175697865353037}
{"step": 46345, "loss": 3.608338165283203, "grad_norm": 0.6909724473953247, "learning_rate": 0.0001, "elbo": 3.60834, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.717863007271874}
{"step": 46350, "loss": 3.5990257263183594, "grad_norm": 0.6597527265548706, "learning_rate": 0.0001, "elbo": 3.59903, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7181562280084446}
{"step": 46355, "loss": 3.6045822143554687, "grad_norm": 0.6517741084098816, "learning_rate": 0.0001, "elbo": 3.60458, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.718449448745015}
{"step": 46360, "loss": 3.6115158081054686, "grad_norm": 0.7048502564430237, "learning_rate": 0.0001, "elbo": 3.61152, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7187426694815855}
{"step": 46365, "loss": 3.617018127441406, "grad_norm": 0.8428558707237244, "learning_rate": 0.0001, "elbo": 3.61702, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7190358902181564}
{"step": 46370, "loss": 3.5985481262207033, "grad_norm": 0.6754729747772217, "learning_rate": 0.0001, "elbo": 3.59855, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.719329110954727}
{"step": 46375, "loss": 3.593511199951172, "grad_norm": 0.7653846144676208, "learning_rate": 0.0001, "elbo": 3.59351, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7196223316912973}
{"step": 46380, "loss": 3.6037967681884764, "grad_norm": 0.6792172193527222, "learning_rate": 0.0001, "elbo": 3.6038, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7199155524278678}
{"step": 46385, "loss": 3.5929370880126954, "grad_norm": 0.8696059584617615, "learning_rate": 0.0001, "elbo": 3.59294, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.720208773164438}
{"step": 46390, "loss": 3.602861785888672, "grad_norm": 0.66970294713974, "learning_rate": 0.0001, "elbo": 3.60286, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7205019939010087}
{"step": 46395, "loss": 3.602164077758789, "grad_norm": 0.7400630712509155, "learning_rate": 0.0001, "elbo": 3.60216, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.720795214637579}
{"step": 46395, "loss": 3.602164077758789, "grad_norm": 0.7400630712509155, "learning_rate": 0.0001, "elbo": 3.60216, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.720795214637579}
{"step": 46400, "loss": 3.6018081665039063, "grad_norm": 0.7102314233779907, "learning_rate": 0.0001, "elbo": 3.60181, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7210884353741496}
{"step": 46405, "loss": 3.58857421875, "grad_norm": 0.6941843628883362, "learning_rate": 0.0001, "elbo": 3.58857, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.72138165611072}
{"step": 46410, "loss": 3.622022247314453, "grad_norm": 0.7911859154701233, "learning_rate": 0.0001, "elbo": 3.62202, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7216748768472905}
{"step": 46415, "loss": 3.593017578125, "grad_norm": 0.6967381834983826, "learning_rate": 0.0001, "elbo": 3.59302, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7219680975838614}
{"step": 46420, "loss": 3.597503662109375, "grad_norm": 0.658717155456543, "learning_rate": 0.0001, "elbo": 3.5975, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.722261318320432}
{"step": 46425, "loss": 3.609100341796875, "grad_norm": 0.781198263168335, "learning_rate": 0.0001, "elbo": 3.6091, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7225545390570023}
{"step": 46430, "loss": 3.5901840209960936, "grad_norm": 0.6574494242668152, "learning_rate": 0.0001, "elbo": 3.59018, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7228477597935727}
{"step": 46430, "loss": 3.5901840209960936, "grad_norm": 0.6574494242668152, "learning_rate": 0.0001, "elbo": 3.59018, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7228477597935727}
{"step": 46435, "loss": 3.598430633544922, "grad_norm": 0.7394867539405823, "learning_rate": 0.0001, "elbo": 3.59843, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.723140980530143}
{"step": 46440, "loss": 3.58974609375, "grad_norm": 0.7169751524925232, "learning_rate": 0.0001, "elbo": 3.58975, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7234342012667137}
{"step": 46445, "loss": 3.5990169525146483, "grad_norm": 0.826431155204773, "learning_rate": 0.0001, "elbo": 3.59902, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.723727422003284}
{"step": 46450, "loss": 3.6027667999267576, "grad_norm": 0.8853509426116943, "learning_rate": 0.0001, "elbo": 3.60277, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7240206427398546}
{"step": 46455, "loss": 3.611520004272461, "grad_norm": 0.7325605750083923, "learning_rate": 0.0001, "elbo": 3.61152, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.724313863476425}
{"step": 46460, "loss": 3.599689483642578, "grad_norm": 0.7840884923934937, "learning_rate": 0.0001, "elbo": 3.59969, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7246070842129955}
{"step": 46465, "loss": 3.605449676513672, "grad_norm": 0.7343615889549255, "learning_rate": 0.0001, "elbo": 3.60545, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.724900304949566}
{"step": 46470, "loss": 3.604268264770508, "grad_norm": 0.6961719393730164, "learning_rate": 0.0001, "elbo": 3.60427, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7251935256861364}
{"step": 46475, "loss": 3.614958953857422, "grad_norm": 0.7019839286804199, "learning_rate": 0.0001, "elbo": 3.61496, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.725486746422707}
{"step": 46480, "loss": 3.583262252807617, "grad_norm": 0.6463907957077026, "learning_rate": 0.0001, "elbo": 3.58326, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7257799671592773}
{"step": 46485, "loss": 3.6037498474121095, "grad_norm": 0.726292610168457, "learning_rate": 0.0001, "elbo": 3.60375, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7260731878958477}
{"step": 46490, "loss": 3.5919113159179688, "grad_norm": 1.0628931522369385, "learning_rate": 0.0001, "elbo": 3.59191, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7263664086324186}
{"step": 46495, "loss": 3.592441177368164, "grad_norm": 0.7127349972724915, "learning_rate": 0.0001, "elbo": 3.59244, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.726659629368989}
{"step": 46500, "loss": 3.6066543579101564, "grad_norm": 0.7050206661224365, "learning_rate": 0.0001, "elbo": 3.60665, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7269528501055595}
{"step": 46505, "loss": 3.603728485107422, "grad_norm": 0.6396292448043823, "learning_rate": 0.0001, "elbo": 3.60373, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.72724607084213}
{"step": 46510, "loss": 3.59442138671875, "grad_norm": 0.7860787510871887, "learning_rate": 0.0001, "elbo": 3.59442, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7275392915787005}
{"step": 46515, "loss": 3.5999908447265625, "grad_norm": 0.7171264886856079, "learning_rate": 0.0001, "elbo": 3.59999, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.727832512315271}
{"step": 46520, "loss": 3.5929187774658202, "grad_norm": 0.6686328053474426, "learning_rate": 0.0001, "elbo": 3.59292, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7281257330518414}
{"step": 46525, "loss": 3.591283416748047, "grad_norm": 0.7142970561981201, "learning_rate": 0.0001, "elbo": 3.59128, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.728418953788412}
{"step": 46530, "loss": 3.599126434326172, "grad_norm": 0.7227944731712341, "learning_rate": 0.0001, "elbo": 3.59913, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7287121745249823}
{"step": 46535, "loss": 3.580610656738281, "grad_norm": 0.7136657238006592, "learning_rate": 0.0001, "elbo": 3.58061, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.729005395261553}
{"step": 46540, "loss": 3.6046211242675783, "grad_norm": 0.75652015209198, "learning_rate": 0.0001, "elbo": 3.60462, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7292986159981236}
{"step": 46545, "loss": 3.606781768798828, "grad_norm": 0.7479944825172424, "learning_rate": 0.0001, "elbo": 3.60678, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.729591836734694}
{"step": 46550, "loss": 3.6031524658203127, "grad_norm": 0.7021617889404297, "learning_rate": 0.0001, "elbo": 3.60315, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7298850574712645}
{"step": 46555, "loss": 3.6131019592285156, "grad_norm": 0.7112194895744324, "learning_rate": 0.0001, "elbo": 3.6131, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.730178278207835}
{"step": 46560, "loss": 3.603157806396484, "grad_norm": 0.6853644251823425, "learning_rate": 0.0001, "elbo": 3.60316, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7304714989444054}
{"step": 46560, "loss": 3.603157806396484, "grad_norm": 0.6853644251823425, "learning_rate": 0.0001, "elbo": 3.60316, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7304714989444054}
{"step": 46565, "loss": 3.6075584411621096, "grad_norm": 0.775910496711731, "learning_rate": 0.0001, "elbo": 3.60756, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.730764719680976}
{"step": 46570, "loss": 3.582346725463867, "grad_norm": 0.6685827970504761, "learning_rate": 0.0001, "elbo": 3.58235, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7310579404175463}
{"step": 46575, "loss": 3.6121978759765625, "grad_norm": 0.6953409314155579, "learning_rate": 0.0001, "elbo": 3.6122, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.731351161154117}
{"step": 46580, "loss": 3.6024505615234377, "grad_norm": 0.6929413676261902, "learning_rate": 0.0001, "elbo": 3.60245, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7316443818906873}
{"step": 46585, "loss": 3.592724609375, "grad_norm": 0.7956385016441345, "learning_rate": 0.0001, "elbo": 3.59272, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7319376026272577}
{"step": 46590, "loss": 3.6050830841064454, "grad_norm": 0.7279513478279114, "learning_rate": 0.0001, "elbo": 3.60508, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.732230823363828}
{"step": 46595, "loss": 3.5797916412353517, "grad_norm": 0.7053464651107788, "learning_rate": 0.0001, "elbo": 3.57979, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7325240441003986}
{"step": 46600, "loss": 3.5929615020751955, "grad_norm": 0.8468228578567505, "learning_rate": 0.0001, "elbo": 3.59296, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.732817264836969}
{"step": 46605, "loss": 3.595682144165039, "grad_norm": 0.7006768584251404, "learning_rate": 0.0001, "elbo": 3.59568, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7331104855735395}
{"step": 46610, "loss": 3.6122413635253907, "grad_norm": 0.6794605255126953, "learning_rate": 0.0001, "elbo": 3.61224, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.73340370631011}
{"step": 46615, "loss": 3.596421813964844, "grad_norm": 0.6529300212860107, "learning_rate": 0.0001, "elbo": 3.59642, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.733696927046681}
{"step": 46620, "loss": 3.5924407958984377, "grad_norm": 0.7469359636306763, "learning_rate": 0.0001, "elbo": 3.59244, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7339901477832513}
{"step": 46625, "loss": 3.6137908935546874, "grad_norm": 0.6635382771492004, "learning_rate": 0.0001, "elbo": 3.61379, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.734283368519822}
{"step": 46630, "loss": 3.5971839904785154, "grad_norm": 0.701616644859314, "learning_rate": 0.0001, "elbo": 3.59718, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7345765892563922}
{"step": 46635, "loss": 3.6116092681884764, "grad_norm": 0.8979125022888184, "learning_rate": 0.0001, "elbo": 3.61161, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7348698099929627}
{"step": 46640, "loss": 3.5989795684814454, "grad_norm": 0.7435702681541443, "learning_rate": 0.0001, "elbo": 3.59898, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.735163030729533}
{"step": 46645, "loss": 3.6164085388183596, "grad_norm": 0.713707447052002, "learning_rate": 0.0001, "elbo": 3.61641, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7354562514661036}
{"step": 46650, "loss": 3.5959346771240233, "grad_norm": 0.7355327010154724, "learning_rate": 0.0001, "elbo": 3.59593, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.735749472202674}
{"step": 46650, "loss": 3.5959346771240233, "grad_norm": 0.7355327010154724, "learning_rate": 0.0001, "elbo": 3.59593, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.735749472202674}
{"step": 46655, "loss": 3.596600341796875, "grad_norm": 0.7154297828674316, "learning_rate": 0.0001, "elbo": 3.5966, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7360426929392445}
{"step": 46660, "loss": 3.604514312744141, "grad_norm": 0.7662770748138428, "learning_rate": 0.0001, "elbo": 3.60451, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7363359136758154}
{"step": 46665, "loss": 3.591436767578125, "grad_norm": 0.7582852840423584, "learning_rate": 0.0001, "elbo": 3.59144, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.736629134412386}
{"step": 46670, "loss": 3.5789093017578124, "grad_norm": 0.7076119184494019, "learning_rate": 0.0001, "elbo": 3.57891, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7369223551489563}
{"step": 46675, "loss": 3.5936729431152346, "grad_norm": 0.6943289637565613, "learning_rate": 0.0001, "elbo": 3.59367, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7372155758855268}
{"step": 46675, "loss": 3.5936729431152346, "grad_norm": 0.6943289637565613, "learning_rate": 0.0001, "elbo": 3.59367, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7372155758855268}
{"step": 46680, "loss": 3.6134544372558595, "grad_norm": 0.6836736798286438, "learning_rate": 0.0001, "elbo": 3.61345, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7375087966220972}
{"step": 46685, "loss": 3.604436492919922, "grad_norm": 0.7006353735923767, "learning_rate": 0.0001, "elbo": 3.60444, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7378020173586677}
{"step": 46690, "loss": 3.597705078125, "grad_norm": 0.7454566359519958, "learning_rate": 0.0001, "elbo": 3.59771, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.738095238095238}
{"step": 46695, "loss": 3.5940017700195312, "grad_norm": 0.655724823474884, "learning_rate": 0.0001, "elbo": 3.594, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7383884588318086}
{"step": 46700, "loss": 3.6111488342285156, "grad_norm": 0.674662709236145, "learning_rate": 0.0001, "elbo": 3.61115, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.738681679568379}
{"step": 46705, "loss": 3.6045440673828124, "grad_norm": 0.6930751800537109, "learning_rate": 0.0001, "elbo": 3.60454, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7389749003049495}
{"step": 46710, "loss": 3.614083099365234, "grad_norm": 0.6968738436698914, "learning_rate": 0.0001, "elbo": 3.61408, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.73926812104152}
{"step": 46715, "loss": 3.591330718994141, "grad_norm": 0.7596203684806824, "learning_rate": 0.0001, "elbo": 3.59133, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7395613417780904}
{"step": 46720, "loss": 3.595915985107422, "grad_norm": 0.7089394927024841, "learning_rate": 0.0001, "elbo": 3.59592, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.739854562514661}
{"step": 46725, "loss": 3.603978729248047, "grad_norm": 0.7164849042892456, "learning_rate": 0.0001, "elbo": 3.60398, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7401477832512313}
{"step": 46730, "loss": 3.595556640625, "grad_norm": 0.7011505365371704, "learning_rate": 0.0001, "elbo": 3.59556, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7404410039878018}
{"step": 46735, "loss": 3.602404022216797, "grad_norm": 0.7105777263641357, "learning_rate": 0.0001, "elbo": 3.6024, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7407342247243727}
{"step": 46740, "loss": 3.5984725952148438, "grad_norm": 0.6825636625289917, "learning_rate": 0.0001, "elbo": 3.59847, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.741027445460943}
{"step": 46745, "loss": 3.588822937011719, "grad_norm": 0.6714990139007568, "learning_rate": 0.0001, "elbo": 3.58882, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7413206661975136}
{"step": 46750, "loss": 3.6040573120117188, "grad_norm": 0.8363419771194458, "learning_rate": 0.0001, "elbo": 3.60406, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.741613886934084}
{"step": 46755, "loss": 3.595104217529297, "grad_norm": 0.6622320413589478, "learning_rate": 0.0001, "elbo": 3.5951, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7419071076706545}
{"step": 46760, "loss": 3.6072906494140624, "grad_norm": 0.8199848532676697, "learning_rate": 0.0001, "elbo": 3.60729, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.742200328407225}
{"step": 46765, "loss": 3.5940177917480467, "grad_norm": 0.7187357544898987, "learning_rate": 0.0001, "elbo": 3.59402, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7424935491437954}
{"step": 46770, "loss": 3.5985481262207033, "grad_norm": 0.6711981892585754, "learning_rate": 0.0001, "elbo": 3.59855, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.742786769880366}
{"step": 46775, "loss": 3.6020347595214846, "grad_norm": 1.181938648223877, "learning_rate": 0.0001, "elbo": 3.60203, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7430799906169363}
{"step": 46780, "loss": 3.6002117156982423, "grad_norm": 0.6941904425621033, "learning_rate": 0.0001, "elbo": 3.60021, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7433732113535068}
{"step": 46785, "loss": 3.6097381591796873, "grad_norm": 0.7809247970581055, "learning_rate": 0.0001, "elbo": 3.60974, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7436664320900777}
{"step": 46790, "loss": 3.599040985107422, "grad_norm": 0.7215259671211243, "learning_rate": 0.0001, "elbo": 3.59904, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.743959652826648}
{"step": 46795, "loss": 3.608623504638672, "grad_norm": 0.7163151502609253, "learning_rate": 0.0001, "elbo": 3.60862, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7442528735632186}
{"step": 46795, "loss": 3.608623504638672, "grad_norm": 0.7163151502609253, "learning_rate": 0.0001, "elbo": 3.60862, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7442528735632186}
{"step": 46800, "loss": 3.596164321899414, "grad_norm": 0.7263824343681335, "learning_rate": 0.0001, "elbo": 3.59616, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.744546094299789}
{"step": 46805, "loss": 3.606752395629883, "grad_norm": 0.6349268555641174, "learning_rate": 0.0001, "elbo": 3.60675, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7448393150363595}
{"step": 46810, "loss": 3.592823791503906, "grad_norm": 0.8124234676361084, "learning_rate": 0.0001, "elbo": 3.59282, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.74513253577293}
{"step": 46815, "loss": 3.594049835205078, "grad_norm": 0.6365666389465332, "learning_rate": 0.0001, "elbo": 3.59405, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7454257565095004}
{"step": 46820, "loss": 3.5739009857177733, "grad_norm": 0.6486015319824219, "learning_rate": 0.0001, "elbo": 3.5739, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.745718977246071}
{"step": 46825, "loss": 3.5876556396484376, "grad_norm": 0.7190340757369995, "learning_rate": 0.0001, "elbo": 3.58766, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7460121979826413}
{"step": 46830, "loss": 3.5923824310302734, "grad_norm": 0.6785462498664856, "learning_rate": 0.0001, "elbo": 3.59238, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7463054187192117}
{"step": 46835, "loss": 3.5861976623535154, "grad_norm": 0.6950916647911072, "learning_rate": 0.0001, "elbo": 3.5862, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.746598639455782}
{"step": 46840, "loss": 3.5914943695068358, "grad_norm": 0.7253237366676331, "learning_rate": 0.0001, "elbo": 3.59149, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7468918601923527}
{"step": 46845, "loss": 3.5870903015136717, "grad_norm": 0.7039176225662231, "learning_rate": 0.0001, "elbo": 3.58709, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.747185080928923}
{"step": 46850, "loss": 3.606690216064453, "grad_norm": 0.6852813959121704, "learning_rate": 0.0001, "elbo": 3.60669, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7474783016654936}
{"step": 46855, "loss": 3.5884368896484373, "grad_norm": 0.6348191499710083, "learning_rate": 0.0001, "elbo": 3.58844, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.747771522402064}
{"step": 46860, "loss": 3.5936294555664063, "grad_norm": 0.6595820188522339, "learning_rate": 0.0001, "elbo": 3.59363, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.748064743138635}
{"step": 46865, "loss": 3.596030426025391, "grad_norm": 0.900616466999054, "learning_rate": 0.0001, "elbo": 3.59603, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7483579638752054}
{"step": 46870, "loss": 3.6018821716308596, "grad_norm": 0.8217254877090454, "learning_rate": 0.0001, "elbo": 3.60188, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.748651184611776}
{"step": 46875, "loss": 3.614776611328125, "grad_norm": 0.6922268271446228, "learning_rate": 0.0001, "elbo": 3.61478, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7489444053483463}
{"step": 46880, "loss": 3.597647476196289, "grad_norm": 0.9322451949119568, "learning_rate": 0.0001, "elbo": 3.59765, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7492376260849167}
{"step": 46885, "loss": 3.5940860748291015, "grad_norm": 0.7062184810638428, "learning_rate": 0.0001, "elbo": 3.59409, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.749530846821487}
{"step": 46890, "loss": 3.591602325439453, "grad_norm": 0.687130331993103, "learning_rate": 0.0001, "elbo": 3.5916, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7498240675580576}
{"step": 46895, "loss": 3.6041950225830077, "grad_norm": 0.6648105382919312, "learning_rate": 0.0001, "elbo": 3.6042, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.750117288294628}
{"step": 46900, "loss": 3.601778030395508, "grad_norm": 0.6617412567138672, "learning_rate": 0.0001, "elbo": 3.60178, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7504105090311985}
{"step": 46905, "loss": 3.5954315185546877, "grad_norm": 0.8694229125976562, "learning_rate": 0.0001, "elbo": 3.59543, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7507037297677694}
{"step": 46910, "loss": 3.5804985046386717, "grad_norm": 0.7289667129516602, "learning_rate": 0.0001, "elbo": 3.5805, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.75099695050434}
{"step": 46915, "loss": 3.5978713989257813, "grad_norm": 0.6461082696914673, "learning_rate": 0.0001, "elbo": 3.59787, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7512901712409104}
{"step": 46920, "loss": 3.603655242919922, "grad_norm": 1.0630043745040894, "learning_rate": 0.0001, "elbo": 3.60366, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.751583391977481}
{"step": 46925, "loss": 3.600302505493164, "grad_norm": 0.70017409324646, "learning_rate": 0.0001, "elbo": 3.6003, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7518766127140513}
{"step": 46930, "loss": 3.583036422729492, "grad_norm": 0.6683512330055237, "learning_rate": 0.0001, "elbo": 3.58304, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7521698334506217}
{"step": 46935, "loss": 3.6036445617675783, "grad_norm": 0.6944993734359741, "learning_rate": 0.0001, "elbo": 3.60364, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.752463054187192}
{"step": 46940, "loss": 3.602603530883789, "grad_norm": 0.6722891330718994, "learning_rate": 0.0001, "elbo": 3.6026, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7527562749237626}
{"step": 46945, "loss": 3.593305206298828, "grad_norm": 0.6550052165985107, "learning_rate": 0.0001, "elbo": 3.59331, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.753049495660333}
{"step": 46950, "loss": 3.576082229614258, "grad_norm": 0.6869375109672546, "learning_rate": 0.0001, "elbo": 3.57608, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7533427163969035}
{"step": 46955, "loss": 3.5941230773925783, "grad_norm": 0.736946702003479, "learning_rate": 0.0001, "elbo": 3.59412, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.753635937133474}
{"step": 46960, "loss": 3.605582427978516, "grad_norm": 0.685901403427124, "learning_rate": 0.0001, "elbo": 3.60558, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7539291578700444}
{"step": 46965, "loss": 3.6063827514648437, "grad_norm": 0.7539173364639282, "learning_rate": 0.0001, "elbo": 3.60638, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.754222378606615}
{"step": 46970, "loss": 3.609271240234375, "grad_norm": 0.9109398722648621, "learning_rate": 0.0001, "elbo": 3.60927, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7545155993431854}
{"step": 46975, "loss": 3.6192176818847654, "grad_norm": 0.7314390540122986, "learning_rate": 0.0001, "elbo": 3.61922, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.754808820079756}
{"step": 46980, "loss": 3.5864242553710937, "grad_norm": 0.8749054670333862, "learning_rate": 0.0001, "elbo": 3.58642, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7551020408163263}
{"step": 46985, "loss": 3.5941741943359373, "grad_norm": 0.7268603444099426, "learning_rate": 0.0001, "elbo": 3.59417, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.755395261552897}
{"step": 46990, "loss": 3.6140769958496093, "grad_norm": 0.7525059580802917, "learning_rate": 0.0001, "elbo": 3.61408, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7556884822894676}
{"step": 46995, "loss": 3.6043045043945314, "grad_norm": 0.7833892703056335, "learning_rate": 0.0001, "elbo": 3.6043, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.755981703026038}
{"step": 47000, "loss": 3.589734649658203, "grad_norm": 0.6896824836730957, "learning_rate": 0.0001, "elbo": 3.58973, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7562749237626085}
{"step": 47005, "loss": 3.613686752319336, "grad_norm": 0.7061502933502197, "learning_rate": 0.0001, "elbo": 3.61369, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.756568144499179}
{"step": 47010, "loss": 3.612462615966797, "grad_norm": 0.6867008209228516, "learning_rate": 0.0001, "elbo": 3.61246, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7568613652357494}
{"step": 47015, "loss": 3.5999114990234373, "grad_norm": 0.6805146932601929, "learning_rate": 0.0001, "elbo": 3.59991, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.75715458597232}
{"step": 47020, "loss": 3.6044273376464844, "grad_norm": 0.6624357104301453, "learning_rate": 0.0001, "elbo": 3.60443, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7574478067088903}
{"step": 47025, "loss": 3.6079883575439453, "grad_norm": 0.6769567131996155, "learning_rate": 0.0001, "elbo": 3.60799, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.757741027445461}
{"step": 47030, "loss": 3.5938232421875, "grad_norm": 0.6496956944465637, "learning_rate": 0.0001, "elbo": 3.59382, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7580342481820317}
{"step": 47035, "loss": 3.59617919921875, "grad_norm": 0.7542905807495117, "learning_rate": 0.0001, "elbo": 3.59618, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.758327468918602}
{"step": 47040, "loss": 3.6169639587402345, "grad_norm": 0.7081378698348999, "learning_rate": 0.0001, "elbo": 3.61696, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7586206896551726}
{"step": 47045, "loss": 3.6001007080078127, "grad_norm": 0.6593085527420044, "learning_rate": 0.0001, "elbo": 3.6001, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.758913910391743}
{"step": 47050, "loss": 3.584442138671875, "grad_norm": 0.7270887494087219, "learning_rate": 0.0001, "elbo": 3.58444, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7592071311283135}
{"step": 47055, "loss": 3.604819488525391, "grad_norm": 0.7902035117149353, "learning_rate": 0.0001, "elbo": 3.60482, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.759500351864884}
{"step": 47060, "loss": 3.597980499267578, "grad_norm": 0.7772225737571716, "learning_rate": 0.0001, "elbo": 3.59798, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7597935726014544}
{"step": 47065, "loss": 3.590260314941406, "grad_norm": 0.8405971527099609, "learning_rate": 0.0001, "elbo": 3.59026, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.760086793338025}
{"step": 47070, "loss": 3.6118484497070313, "grad_norm": 0.6886228322982788, "learning_rate": 0.0001, "elbo": 3.61185, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7603800140745953}
{"step": 47075, "loss": 3.6161544799804686, "grad_norm": 0.7669087052345276, "learning_rate": 0.0001, "elbo": 3.61615, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7606732348111658}
{"step": 47080, "loss": 3.6008464813232424, "grad_norm": 0.7523797154426575, "learning_rate": 0.0001, "elbo": 3.60085, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7609664555477362}
{"step": 47085, "loss": 3.6005897521972656, "grad_norm": 0.699153482913971, "learning_rate": 0.0001, "elbo": 3.60059, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7612596762843067}
{"step": 47090, "loss": 3.5804569244384767, "grad_norm": 0.7022877335548401, "learning_rate": 0.0001, "elbo": 3.58046, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.761552897020877}
{"step": 47095, "loss": 3.6044776916503904, "grad_norm": 0.7674222588539124, "learning_rate": 0.0001, "elbo": 3.60448, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7618461177574476}
{"step": 47100, "loss": 3.6101112365722656, "grad_norm": 0.7386491298675537, "learning_rate": 0.0001, "elbo": 3.61011, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.762139338494018}
{"step": 47105, "loss": 3.600095748901367, "grad_norm": 0.7136309146881104, "learning_rate": 0.0001, "elbo": 3.6001, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.762432559230589}
{"step": 47110, "loss": 3.604043960571289, "grad_norm": 0.6661200523376465, "learning_rate": 0.0001, "elbo": 3.60404, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7627257799671594}
{"step": 47115, "loss": 3.593389892578125, "grad_norm": 0.936603844165802, "learning_rate": 0.0001, "elbo": 3.59339, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.76301900070373}
{"step": 47120, "loss": 3.5853950500488283, "grad_norm": 0.7094579339027405, "learning_rate": 0.0001, "elbo": 3.58539, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7633122214403003}
{"step": 47125, "loss": 3.6015556335449217, "grad_norm": 0.7228677868843079, "learning_rate": 0.0001, "elbo": 3.60156, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7636054421768708}
{"step": 47130, "loss": 3.620830535888672, "grad_norm": 0.6921676397323608, "learning_rate": 0.0001, "elbo": 3.62083, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.763898662913441}
{"step": 47135, "loss": 3.600275421142578, "grad_norm": 0.7034268379211426, "learning_rate": 0.0001, "elbo": 3.60028, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7641918836500117}
{"step": 47140, "loss": 3.6174442291259767, "grad_norm": 0.6868014335632324, "learning_rate": 0.0001, "elbo": 3.61744, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.764485104386582}
{"step": 47145, "loss": 3.589686965942383, "grad_norm": 0.7648047208786011, "learning_rate": 0.0001, "elbo": 3.58969, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7647783251231526}
{"step": 47150, "loss": 3.5978343963623045, "grad_norm": 0.8034935593605042, "learning_rate": 0.0001, "elbo": 3.59783, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.765071545859723}
{"step": 47155, "loss": 3.5909942626953124, "grad_norm": 0.7100746035575867, "learning_rate": 0.0001, "elbo": 3.59099, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.765364766596294}
{"step": 47160, "loss": 3.588213348388672, "grad_norm": 0.8207279443740845, "learning_rate": 0.0001, "elbo": 3.58821, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7656579873328644}
{"step": 47165, "loss": 3.5929306030273436, "grad_norm": 0.7143569588661194, "learning_rate": 0.0001, "elbo": 3.59293, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.765951208069435}
{"step": 47170, "loss": 3.5992977142333986, "grad_norm": 0.7221043109893799, "learning_rate": 0.0001, "elbo": 3.5993, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7662444288060053}
{"step": 47170, "loss": 3.5992977142333986, "grad_norm": 0.7221043109893799, "learning_rate": 0.0001, "elbo": 3.5993, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7662444288060053}
{"step": 47175, "loss": 3.6019805908203124, "grad_norm": 0.7153292894363403, "learning_rate": 0.0001, "elbo": 3.60198, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7665376495425757}
{"step": 47180, "loss": 3.5990089416503905, "grad_norm": 0.737492561340332, "learning_rate": 0.0001, "elbo": 3.59901, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.766830870279146}
{"step": 47185, "loss": 3.5835704803466797, "grad_norm": 0.6705780625343323, "learning_rate": 0.0001, "elbo": 3.58357, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7671240910157167}
{"step": 47190, "loss": 3.5926887512207033, "grad_norm": 0.6462015509605408, "learning_rate": 0.0001, "elbo": 3.59269, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.767417311752287}
{"step": 47195, "loss": 3.5934844970703126, "grad_norm": 0.6794047951698303, "learning_rate": 0.0001, "elbo": 3.59348, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7677105324888576}
{"step": 47200, "loss": 3.5988189697265627, "grad_norm": 0.6755030751228333, "learning_rate": 0.0001, "elbo": 3.59882, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.768003753225428}
{"step": 47200, "loss": 3.5988189697265627, "grad_norm": 0.6755030751228333, "learning_rate": 0.0001, "elbo": 3.59882, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.768003753225428}
{"step": 47205, "loss": 3.5993141174316405, "grad_norm": 0.6922029256820679, "learning_rate": 0.0001, "elbo": 3.59931, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7682969739619985}
{"step": 47210, "loss": 3.605877685546875, "grad_norm": 0.8294531106948853, "learning_rate": 0.0001, "elbo": 3.60588, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.768590194698569}
{"step": 47215, "loss": 3.573388671875, "grad_norm": 0.6867364048957825, "learning_rate": 0.0001, "elbo": 3.57339, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7688834154351394}
{"step": 47220, "loss": 3.589949035644531, "grad_norm": 0.678324818611145, "learning_rate": 0.0001, "elbo": 3.58995, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.76917663617171}
{"step": 47225, "loss": 3.5979019165039063, "grad_norm": 0.7140765190124512, "learning_rate": 0.0001, "elbo": 3.5979, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7694698569082803}
{"step": 47230, "loss": 3.6046695709228516, "grad_norm": 0.6556147933006287, "learning_rate": 0.0001, "elbo": 3.60467, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.769763077644851}
{"step": 47235, "loss": 3.5836585998535155, "grad_norm": 0.7233611345291138, "learning_rate": 0.0001, "elbo": 3.58366, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7700562983814216}
{"step": 47240, "loss": 3.6112098693847656, "grad_norm": 0.7007848024368286, "learning_rate": 0.0001, "elbo": 3.61121, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.770349519117992}
{"step": 47245, "loss": 3.5808670043945314, "grad_norm": 0.763177216053009, "learning_rate": 0.0001, "elbo": 3.58087, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7706427398545626}
{"step": 47250, "loss": 3.6037002563476563, "grad_norm": 0.8166947364807129, "learning_rate": 0.0001, "elbo": 3.6037, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.770935960591133}
{"step": 47255, "loss": 3.598524475097656, "grad_norm": 0.6823955178260803, "learning_rate": 0.0001, "elbo": 3.59852, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7712291813277035}
{"step": 47260, "loss": 3.612222671508789, "grad_norm": 0.8218519687652588, "learning_rate": 0.0001, "elbo": 3.61222, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.771522402064274}
{"step": 47265, "loss": 3.5895412445068358, "grad_norm": 0.6798659563064575, "learning_rate": 0.0001, "elbo": 3.58954, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7718156228008444}
{"step": 47270, "loss": 3.5990741729736326, "grad_norm": 0.7387179732322693, "learning_rate": 0.0001, "elbo": 3.59907, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.772108843537415}
{"step": 47275, "loss": 3.5883697509765624, "grad_norm": 0.7190521359443665, "learning_rate": 0.0001, "elbo": 3.58837, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7724020642739857}
{"step": 47280, "loss": 3.5748580932617187, "grad_norm": 0.634895920753479, "learning_rate": 0.0001, "elbo": 3.57486, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.772695285010556}
{"step": 47285, "loss": 3.6105087280273436, "grad_norm": 0.7300340533256531, "learning_rate": 0.0001, "elbo": 3.61051, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7729885057471266}
{"step": 47290, "loss": 3.5868980407714846, "grad_norm": 0.6831327080726624, "learning_rate": 0.0001, "elbo": 3.5869, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.773281726483697}
{"step": 47295, "loss": 3.605260467529297, "grad_norm": 0.7439638376235962, "learning_rate": 0.0001, "elbo": 3.60526, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7735749472202675}
{"step": 47300, "loss": 3.5908302307128905, "grad_norm": 0.7533876895904541, "learning_rate": 0.0001, "elbo": 3.59083, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.773868167956838}
{"step": 47305, "loss": 3.6043880462646483, "grad_norm": 0.7193492650985718, "learning_rate": 0.0001, "elbo": 3.60439, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7741613886934084}
{"step": 47310, "loss": 3.5829936981201174, "grad_norm": 0.6834198832511902, "learning_rate": 0.0001, "elbo": 3.58299, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.774454609429979}
{"step": 47315, "loss": 3.587465286254883, "grad_norm": 0.6937268376350403, "learning_rate": 0.0001, "elbo": 3.58747, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7747478301665494}
{"step": 47320, "loss": 3.5822769165039063, "grad_norm": 0.6954057812690735, "learning_rate": 0.0001, "elbo": 3.58228, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.77504105090312}
{"step": 47325, "loss": 3.5867252349853516, "grad_norm": 0.694414496421814, "learning_rate": 0.0001, "elbo": 3.58673, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7753342716396903}
{"step": 47330, "loss": 3.5835544586181642, "grad_norm": 0.8420559167861938, "learning_rate": 0.0001, "elbo": 3.58355, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7756274923762607}
{"step": 47335, "loss": 3.588491439819336, "grad_norm": 0.6680840253829956, "learning_rate": 0.0001, "elbo": 3.58849, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.775920713112831}
{"step": 47340, "loss": 3.581869125366211, "grad_norm": 0.7121816873550415, "learning_rate": 0.0001, "elbo": 3.58187, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7762139338494016}
{"step": 47345, "loss": 3.597029113769531, "grad_norm": 0.7353522777557373, "learning_rate": 0.0001, "elbo": 3.59703, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.776507154585972}
{"step": 47350, "loss": 3.5850360870361326, "grad_norm": 0.8755040764808655, "learning_rate": 0.0001, "elbo": 3.58504, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7768003753225425}
{"step": 47355, "loss": 3.612096405029297, "grad_norm": 0.7056068181991577, "learning_rate": 0.0001, "elbo": 3.6121, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7770935960591134}
{"step": 47360, "loss": 3.6058837890625, "grad_norm": 0.6880292892456055, "learning_rate": 0.0001, "elbo": 3.60588, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.777386816795684}
{"step": 47365, "loss": 3.5935211181640625, "grad_norm": 0.6476045846939087, "learning_rate": 0.0001, "elbo": 3.59352, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7776800375322543}
{"step": 47370, "loss": 3.589451217651367, "grad_norm": 0.7081976532936096, "learning_rate": 0.0001, "elbo": 3.58945, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.777973258268825}
{"step": 47375, "loss": 3.6133071899414064, "grad_norm": 0.6818972826004028, "learning_rate": 0.0001, "elbo": 3.61331, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7782664790053953}
{"step": 47380, "loss": 3.5913368225097657, "grad_norm": 0.6537179946899414, "learning_rate": 0.0001, "elbo": 3.59134, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7785596997419657}
{"step": 47385, "loss": 3.5962539672851563, "grad_norm": 0.686272144317627, "learning_rate": 0.0001, "elbo": 3.59625, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.778852920478536}
{"step": 47390, "loss": 3.600518798828125, "grad_norm": 0.7493917942047119, "learning_rate": 0.0001, "elbo": 3.60052, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7791461412151066}
{"step": 47395, "loss": 3.601187896728516, "grad_norm": 0.8339998722076416, "learning_rate": 0.0001, "elbo": 3.60119, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.779439361951677}
{"step": 47400, "loss": 3.597942352294922, "grad_norm": 0.6501866579055786, "learning_rate": 0.0001, "elbo": 3.59794, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.779732582688248}
{"step": 47405, "loss": 3.5915145874023438, "grad_norm": 0.6799745559692383, "learning_rate": 0.0001, "elbo": 3.59151, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7800258034248184}
{"step": 47410, "loss": 3.5969684600830076, "grad_norm": 0.672035276889801, "learning_rate": 0.0001, "elbo": 3.59697, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.780319024161389}
{"step": 47415, "loss": 3.6100837707519533, "grad_norm": 0.7907328009605408, "learning_rate": 0.0001, "elbo": 3.61008, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7806122448979593}
{"step": 47420, "loss": 3.597800445556641, "grad_norm": 0.6668440103530884, "learning_rate": 0.0001, "elbo": 3.5978, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.78090546563453}
{"step": 47425, "loss": 3.5992340087890624, "grad_norm": 0.7029322385787964, "learning_rate": 0.0001, "elbo": 3.59923, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7811986863711002}
{"step": 47430, "loss": 3.5928558349609374, "grad_norm": 0.7082005739212036, "learning_rate": 0.0001, "elbo": 3.59286, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7814919071076707}
{"step": 47435, "loss": 3.59158935546875, "grad_norm": 0.6415417194366455, "learning_rate": 0.0001, "elbo": 3.59159, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.781785127844241}
{"step": 47440, "loss": 3.5988838195800783, "grad_norm": 0.704156219959259, "learning_rate": 0.0001, "elbo": 3.59888, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7820783485808116}
{"step": 47445, "loss": 3.5861671447753904, "grad_norm": 0.7344411611557007, "learning_rate": 0.0001, "elbo": 3.58617, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.782371569317382}
{"step": 47450, "loss": 3.582783508300781, "grad_norm": 0.7142704129219055, "learning_rate": 0.0001, "elbo": 3.58278, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7826647900539525}
{"step": 47455, "loss": 3.6021324157714845, "grad_norm": 0.6727951169013977, "learning_rate": 0.0001, "elbo": 3.60213, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.782958010790523}
{"step": 47460, "loss": 3.59212646484375, "grad_norm": 0.6632062792778015, "learning_rate": 0.0001, "elbo": 3.59213, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7832512315270934}
{"step": 47465, "loss": 3.592833709716797, "grad_norm": 0.6861891150474548, "learning_rate": 0.0001, "elbo": 3.59283, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.783544452263664}
{"step": 47470, "loss": 3.590222930908203, "grad_norm": 0.6959906220436096, "learning_rate": 0.0001, "elbo": 3.59022, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7838376730002343}
{"step": 47475, "loss": 3.5949481964111327, "grad_norm": 0.6641646027565002, "learning_rate": 0.0001, "elbo": 3.59495, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7841308937368052}
{"step": 47480, "loss": 3.608175277709961, "grad_norm": 0.69944828748703, "learning_rate": 0.0001, "elbo": 3.60818, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7844241144733757}
{"step": 47485, "loss": 3.5801437377929686, "grad_norm": 0.7161989808082581, "learning_rate": 0.0001, "elbo": 3.58014, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.784717335209946}
{"step": 47490, "loss": 3.6082000732421875, "grad_norm": 0.717950165271759, "learning_rate": 0.0001, "elbo": 3.6082, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7850105559465166}
{"step": 47495, "loss": 3.5932022094726563, "grad_norm": 0.6976622939109802, "learning_rate": 0.0001, "elbo": 3.5932, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.785303776683087}
{"step": 47500, "loss": 3.5866928100585938, "grad_norm": 0.6848353147506714, "learning_rate": 0.0001, "elbo": 3.58669, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7855969974196575}
{"step": 47505, "loss": 3.5852664947509765, "grad_norm": 0.827163577079773, "learning_rate": 0.0001, "elbo": 3.58527, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.785890218156228}
{"step": 47510, "loss": 3.5865894317626954, "grad_norm": 0.6652823686599731, "learning_rate": 0.0001, "elbo": 3.58659, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7861834388927984}
{"step": 47515, "loss": 3.583548736572266, "grad_norm": 0.6627635359764099, "learning_rate": 0.0001, "elbo": 3.58355, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.786476659629369}
{"step": 47520, "loss": 3.615817642211914, "grad_norm": 0.715573787689209, "learning_rate": 0.0001, "elbo": 3.61582, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7867698803659398}
{"step": 47525, "loss": 3.601371002197266, "grad_norm": 0.7205118536949158, "learning_rate": 0.0001, "elbo": 3.60137, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.78706310110251}
{"step": 47530, "loss": 3.5976806640625, "grad_norm": 0.6755281686782837, "learning_rate": 0.0001, "elbo": 3.59768, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7873563218390807}
{"step": 47535, "loss": 3.6114402770996095, "grad_norm": 0.6839695572853088, "learning_rate": 0.0001, "elbo": 3.61144, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.787649542575651}
{"step": 47540, "loss": 3.5852497100830076, "grad_norm": 0.6739481687545776, "learning_rate": 0.0001, "elbo": 3.58525, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7879427633122216}
{"step": 47545, "loss": 3.6017242431640626, "grad_norm": 0.6673702597618103, "learning_rate": 0.0001, "elbo": 3.60172, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.788235984048792}
{"step": 47550, "loss": 3.586664581298828, "grad_norm": 0.6949877738952637, "learning_rate": 0.0001, "elbo": 3.58666, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7885292047853625}
{"step": 47555, "loss": 3.5952110290527344, "grad_norm": 0.9005464315414429, "learning_rate": 0.0001, "elbo": 3.59521, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.788822425521933}
{"step": 47560, "loss": 3.580230712890625, "grad_norm": 0.7039784789085388, "learning_rate": 0.0001, "elbo": 3.58023, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7891156462585034}
{"step": 47565, "loss": 3.5993087768554686, "grad_norm": 0.9277472496032715, "learning_rate": 0.0001, "elbo": 3.59931, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.789408866995074}
{"step": 47570, "loss": 3.584492492675781, "grad_norm": 0.6847471594810486, "learning_rate": 0.0001, "elbo": 3.58449, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7897020877316443}
{"step": 47575, "loss": 3.6117713928222654, "grad_norm": 0.7566883563995361, "learning_rate": 0.0001, "elbo": 3.61177, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7899953084682148}
{"step": 47580, "loss": 3.593830871582031, "grad_norm": 0.6652248501777649, "learning_rate": 0.0001, "elbo": 3.59383, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.790288529204785}
{"step": 47585, "loss": 3.5956954956054688, "grad_norm": 0.6722770929336548, "learning_rate": 0.0001, "elbo": 3.5957, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7905817499413557}
{"step": 47590, "loss": 3.5835418701171875, "grad_norm": 0.6912604570388794, "learning_rate": 0.0001, "elbo": 3.58354, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.790874970677926}
{"step": 47595, "loss": 3.6023796081542967, "grad_norm": 0.8003152012825012, "learning_rate": 0.0001, "elbo": 3.60238, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7911681914144966}
{"step": 47600, "loss": 3.585647964477539, "grad_norm": 0.6794154047966003, "learning_rate": 0.0001, "elbo": 3.58565, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7914614121510675}
{"step": 47605, "loss": 3.6085975646972654, "grad_norm": 0.6817049980163574, "learning_rate": 0.0001, "elbo": 3.6086, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.791754632887638}
{"step": 47610, "loss": 3.6003780364990234, "grad_norm": 0.6574364900588989, "learning_rate": 0.0001, "elbo": 3.60038, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7920478536242084}
{"step": 47615, "loss": 3.5928497314453125, "grad_norm": 0.7386645674705505, "learning_rate": 0.0001, "elbo": 3.59285, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.792341074360779}
{"step": 47620, "loss": 3.585861587524414, "grad_norm": 0.7014338970184326, "learning_rate": 0.0001, "elbo": 3.58586, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7926342950973493}
{"step": 47625, "loss": 3.6018051147460937, "grad_norm": 1.0661473274230957, "learning_rate": 0.0001, "elbo": 3.6018, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7929275158339197}
{"step": 47630, "loss": 3.5884521484375, "grad_norm": 0.7078308463096619, "learning_rate": 0.0001, "elbo": 3.58845, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.79322073657049}
{"step": 47635, "loss": 3.604030227661133, "grad_norm": 0.6845096945762634, "learning_rate": 0.0001, "elbo": 3.60403, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7935139573070606}
{"step": 47640, "loss": 3.5860221862792967, "grad_norm": 0.6757418513298035, "learning_rate": 0.0001, "elbo": 3.58602, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.793807178043631}
{"step": 47645, "loss": 3.580706787109375, "grad_norm": 0.7083747386932373, "learning_rate": 0.0001, "elbo": 3.58071, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.794100398780202}
{"step": 47650, "loss": 3.581931304931641, "grad_norm": 0.6921861171722412, "learning_rate": 0.0001, "elbo": 3.58193, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7943936195167725}
{"step": 47655, "loss": 3.593332290649414, "grad_norm": 0.6705560684204102, "learning_rate": 0.0001, "elbo": 3.59333, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.794686840253343}
{"step": 47660, "loss": 3.5929847717285157, "grad_norm": 0.7375988364219666, "learning_rate": 0.0001, "elbo": 3.59299, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7949800609899134}
{"step": 47665, "loss": 3.5996444702148436, "grad_norm": 0.6941717267036438, "learning_rate": 0.0001, "elbo": 3.59964, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.795273281726484}
{"step": 47670, "loss": 3.5973232269287108, "grad_norm": 0.7653666734695435, "learning_rate": 0.0001, "elbo": 3.59732, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7955665024630543}
{"step": 47675, "loss": 3.608420562744141, "grad_norm": 0.6784508228302002, "learning_rate": 0.0001, "elbo": 3.60842, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7958597231996247}
{"step": 47675, "loss": 3.608420562744141, "grad_norm": 0.6784508228302002, "learning_rate": 0.0001, "elbo": 3.60842, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7958597231996247}
{"step": 47680, "loss": 3.591565704345703, "grad_norm": 0.6814920902252197, "learning_rate": 0.0001, "elbo": 3.59157, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.796152943936195}
{"step": 47685, "loss": 3.5865943908691404, "grad_norm": 0.6853728294372559, "learning_rate": 0.0001, "elbo": 3.58659, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7964461646727656}
{"step": 47690, "loss": 3.6045303344726562, "grad_norm": 0.7059223651885986, "learning_rate": 0.0001, "elbo": 3.60453, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.796739385409336}
{"step": 47695, "loss": 3.590497589111328, "grad_norm": 0.8093512654304504, "learning_rate": 0.0001, "elbo": 3.5905, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7970326061459065}
{"step": 47700, "loss": 3.6039947509765624, "grad_norm": 0.8028951287269592, "learning_rate": 0.0001, "elbo": 3.60399, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.797325826882477}
{"step": 47705, "loss": 3.594377899169922, "grad_norm": 0.6796607971191406, "learning_rate": 0.0001, "elbo": 3.59438, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7976190476190474}
{"step": 47710, "loss": 3.5913902282714845, "grad_norm": 0.6788617372512817, "learning_rate": 0.0001, "elbo": 3.59139, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.797912268355618}
{"step": 47715, "loss": 3.5931243896484375, "grad_norm": 0.6668785810470581, "learning_rate": 0.0001, "elbo": 3.59312, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7982054890921884}
{"step": 47720, "loss": 3.598250961303711, "grad_norm": 0.6804541945457458, "learning_rate": 0.0001, "elbo": 3.59825, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.798498709828759}
{"step": 47725, "loss": 3.586420440673828, "grad_norm": 0.6715087890625, "learning_rate": 0.0001, "elbo": 3.58642, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7987919305653297}
{"step": 47730, "loss": 3.5890384674072267, "grad_norm": 0.7833905220031738, "learning_rate": 0.0001, "elbo": 3.58904, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7990851513019}
{"step": 47735, "loss": 3.6052364349365233, "grad_norm": 0.6869921684265137, "learning_rate": 0.0001, "elbo": 3.60524, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7993783720384706}
{"step": 47740, "loss": 3.5954029083251955, "grad_norm": 0.6351822018623352, "learning_rate": 0.0001, "elbo": 3.5954, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.799671592775041}
{"step": 47745, "loss": 3.585553741455078, "grad_norm": 0.7978439331054688, "learning_rate": 0.0001, "elbo": 3.58555, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.7999648135116115}
{"step": 47750, "loss": 3.5820777893066404, "grad_norm": 0.6911022067070007, "learning_rate": 0.0001, "elbo": 3.58208, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.800258034248182}
{"step": 47755, "loss": 3.6024608612060547, "grad_norm": 0.6912309527397156, "learning_rate": 0.0001, "elbo": 3.60246, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8005512549847524}
{"step": 47760, "loss": 3.5923698425292967, "grad_norm": 0.7510629296302795, "learning_rate": 0.0001, "elbo": 3.59237, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.800844475721323}
{"step": 47765, "loss": 3.5915542602539063, "grad_norm": 0.7041082382202148, "learning_rate": 0.0001, "elbo": 3.59155, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8011376964578933}
{"step": 47770, "loss": 3.578883743286133, "grad_norm": 0.7926814556121826, "learning_rate": 0.0001, "elbo": 3.57888, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8014309171944642}
{"step": 47775, "loss": 3.6058185577392576, "grad_norm": 0.7751047611236572, "learning_rate": 0.0001, "elbo": 3.60582, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8017241379310347}
{"step": 47780, "loss": 3.590899658203125, "grad_norm": 0.7121114730834961, "learning_rate": 0.0001, "elbo": 3.5909, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.802017358667605}
{"step": 47785, "loss": 3.592732620239258, "grad_norm": 0.7231191992759705, "learning_rate": 0.0001, "elbo": 3.59273, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8023105794041756}
{"step": 47790, "loss": 3.5903759002685547, "grad_norm": 0.7215102910995483, "learning_rate": 0.0001, "elbo": 3.59038, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.802603800140746}
{"step": 47795, "loss": 3.5939453125, "grad_norm": 0.7967034578323364, "learning_rate": 0.0001, "elbo": 3.59395, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8028970208773165}
{"step": 47800, "loss": 3.595450592041016, "grad_norm": 0.7131870985031128, "learning_rate": 0.0001, "elbo": 3.59545, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.803190241613887}
{"step": 47805, "loss": 3.596873474121094, "grad_norm": 0.6848935484886169, "learning_rate": 0.0001, "elbo": 3.59687, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8034834623504574}
{"step": 47810, "loss": 3.600080871582031, "grad_norm": 0.7631101608276367, "learning_rate": 0.0001, "elbo": 3.60008, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.803776683087028}
{"step": 47815, "loss": 3.597216033935547, "grad_norm": 1.3399486541748047, "learning_rate": 0.0001, "elbo": 3.59722, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8040699038235983}
{"step": 47820, "loss": 3.589464569091797, "grad_norm": 0.6773530840873718, "learning_rate": 0.0001, "elbo": 3.58946, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.804363124560169}
{"step": 47825, "loss": 3.57041015625, "grad_norm": 0.6753265261650085, "learning_rate": 0.0001, "elbo": 3.57041, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8046563452967392}
{"step": 47830, "loss": 3.6055683135986327, "grad_norm": 0.6708587408065796, "learning_rate": 0.0001, "elbo": 3.60557, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8049495660333097}
{"step": 47835, "loss": 3.5977527618408205, "grad_norm": 0.6703479886054993, "learning_rate": 0.0001, "elbo": 3.59775, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.80524278676988}
{"step": 47840, "loss": 3.576618194580078, "grad_norm": 0.6611201763153076, "learning_rate": 0.0001, "elbo": 3.57662, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8055360075064506}
{"step": 47845, "loss": 3.6060924530029297, "grad_norm": 0.6893573999404907, "learning_rate": 0.0001, "elbo": 3.60609, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8058292282430215}
{"step": 47850, "loss": 3.5867095947265626, "grad_norm": 0.6515403985977173, "learning_rate": 0.0001, "elbo": 3.58671, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.806122448979592}
{"step": 47855, "loss": 3.5951133728027345, "grad_norm": 0.6982861757278442, "learning_rate": 0.0001, "elbo": 3.59511, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8064156697161624}
{"step": 47860, "loss": 3.59124755859375, "grad_norm": 0.7274712324142456, "learning_rate": 0.0001, "elbo": 3.59125, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.806708890452733}
{"step": 47865, "loss": 3.5804542541503905, "grad_norm": 0.7207044363021851, "learning_rate": 0.0001, "elbo": 3.58045, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8070021111893033}
{"step": 47870, "loss": 3.596904754638672, "grad_norm": 0.6923382878303528, "learning_rate": 0.0001, "elbo": 3.59691, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8072953319258738}
{"step": 47875, "loss": 3.5913162231445312, "grad_norm": 0.7131889462471008, "learning_rate": 0.0001, "elbo": 3.59132, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8075885526624442}
{"step": 47880, "loss": 3.5930557250976562, "grad_norm": 0.7262840867042542, "learning_rate": 0.0001, "elbo": 3.59306, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8078817733990147}
{"step": 47885, "loss": 3.5826683044433594, "grad_norm": 0.6826227903366089, "learning_rate": 0.0001, "elbo": 3.58267, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.808174994135585}
{"step": 47890, "loss": 3.591147232055664, "grad_norm": 0.6522448658943176, "learning_rate": 0.0001, "elbo": 3.59115, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.808468214872156}
{"step": 47895, "loss": 3.591683197021484, "grad_norm": 0.7577225565910339, "learning_rate": 0.0001, "elbo": 3.59168, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8087614356087265}
{"step": 47900, "loss": 3.5902435302734377, "grad_norm": 0.6859487295150757, "learning_rate": 0.0001, "elbo": 3.59024, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.809054656345297}
{"step": 47905, "loss": 3.597126770019531, "grad_norm": 0.7214903235435486, "learning_rate": 0.0001, "elbo": 3.59713, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8093478770818674}
{"step": 47910, "loss": 3.5951358795166017, "grad_norm": 0.6536147594451904, "learning_rate": 0.0001, "elbo": 3.59514, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.809641097818438}
{"step": 47915, "loss": 3.582898712158203, "grad_norm": 0.6871041059494019, "learning_rate": 0.0001, "elbo": 3.5829, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8099343185550083}
{"step": 47920, "loss": 3.608643341064453, "grad_norm": 0.7464771866798401, "learning_rate": 0.0001, "elbo": 3.60864, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8102275392915788}
{"step": 47925, "loss": 3.5704708099365234, "grad_norm": 0.7115820050239563, "learning_rate": 0.0001, "elbo": 3.57047, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.810520760028149}
{"step": 47930, "loss": 3.5937103271484374, "grad_norm": 0.6513554453849792, "learning_rate": 0.0001, "elbo": 3.59371, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8108139807647197}
{"step": 47935, "loss": 3.5972354888916014, "grad_norm": 0.7158162593841553, "learning_rate": 0.0001, "elbo": 3.59724, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.81110720150129}
{"step": 47940, "loss": 3.598745346069336, "grad_norm": 0.6603230237960815, "learning_rate": 0.0001, "elbo": 3.59875, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8114004222378606}
{"step": 47945, "loss": 3.592357635498047, "grad_norm": 0.7198366522789001, "learning_rate": 0.0001, "elbo": 3.59236, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.811693642974431}
{"step": 47950, "loss": 3.5816497802734375, "grad_norm": 0.6623525619506836, "learning_rate": 0.0001, "elbo": 3.58165, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8119868637110015}
{"step": 47955, "loss": 3.6086280822753904, "grad_norm": 1.2700929641723633, "learning_rate": 0.0001, "elbo": 3.60863, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.812280084447572}
{"step": 47960, "loss": 3.5928730010986327, "grad_norm": 0.6734810471534729, "learning_rate": 0.0001, "elbo": 3.59287, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8125733051841424}
{"step": 47965, "loss": 3.580893325805664, "grad_norm": 0.7502511739730835, "learning_rate": 0.0001, "elbo": 3.58089, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.812866525920713}
{"step": 47970, "loss": 3.5960617065429688, "grad_norm": 0.7255811095237732, "learning_rate": 0.0001, "elbo": 3.59606, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8131597466572837}
{"step": 47975, "loss": 3.583327865600586, "grad_norm": 1.0269540548324585, "learning_rate": 0.0001, "elbo": 3.58333, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.813452967393854}
{"step": 47980, "loss": 3.591012954711914, "grad_norm": 0.7753385305404663, "learning_rate": 0.0001, "elbo": 3.59101, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8137461881304247}
{"step": 47985, "loss": 3.607779693603516, "grad_norm": 0.7309095859527588, "learning_rate": 0.0001, "elbo": 3.60778, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.814039408866995}
{"step": 47990, "loss": 3.594635772705078, "grad_norm": 0.6658946871757507, "learning_rate": 0.0001, "elbo": 3.59464, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8143326296035656}
{"step": 47995, "loss": 3.602610778808594, "grad_norm": 0.7176581025123596, "learning_rate": 0.0001, "elbo": 3.60261, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.814625850340136}
{"step": 48000, "loss": 3.5931358337402344, "grad_norm": 0.6754315495491028, "learning_rate": 0.0001, "elbo": 3.59314, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8149190710767065}
{"step": 48005, "loss": 3.593647003173828, "grad_norm": 0.6571306586265564, "learning_rate": 0.0001, "elbo": 3.59365, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.815212291813277}
{"step": 48010, "loss": 3.594216156005859, "grad_norm": 0.6992830038070679, "learning_rate": 0.0001, "elbo": 3.59422, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8155055125498474}
{"step": 48015, "loss": 3.5972969055175783, "grad_norm": 1.2512997388839722, "learning_rate": 0.0001, "elbo": 3.5973, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8157987332864183}
{"step": 48020, "loss": 3.5953746795654298, "grad_norm": 1.100947618484497, "learning_rate": 0.0001, "elbo": 3.59537, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8160919540229887}
{"step": 48025, "loss": 3.603737640380859, "grad_norm": 0.7370672225952148, "learning_rate": 0.0001, "elbo": 3.60374, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.816385174759559}
{"step": 48025, "loss": 3.603737640380859, "grad_norm": 0.7370672225952148, "learning_rate": 0.0001, "elbo": 3.60374, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.816385174759559}
{"step": 48030, "loss": 3.5897621154785155, "grad_norm": 0.6733763813972473, "learning_rate": 0.0001, "elbo": 3.58976, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8166783954961296}
{"step": 48035, "loss": 3.6110458374023438, "grad_norm": 0.8148212432861328, "learning_rate": 0.0001, "elbo": 3.61105, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8169716162327}
{"step": 48040, "loss": 3.587358093261719, "grad_norm": 0.6579931974411011, "learning_rate": 0.0001, "elbo": 3.58736, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8172648369692705}
{"step": 48045, "loss": 3.597209930419922, "grad_norm": 0.6549070477485657, "learning_rate": 0.0001, "elbo": 3.59721, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.817558057705841}
{"step": 48045, "loss": 3.597209930419922, "grad_norm": 0.6549070477485657, "learning_rate": 0.0001, "elbo": 3.59721, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.817558057705841}
{"step": 48050, "loss": 3.6041332244873048, "grad_norm": 0.7110847234725952, "learning_rate": 0.0001, "elbo": 3.60413, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8178512784424115}
{"step": 48055, "loss": 3.5954872131347657, "grad_norm": 0.6746610999107361, "learning_rate": 0.0001, "elbo": 3.59549, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.818144499178982}
{"step": 48060, "loss": 3.5957603454589844, "grad_norm": 0.7285844087600708, "learning_rate": 0.0001, "elbo": 3.59576, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8184377199155524}
{"step": 48065, "loss": 3.5850093841552733, "grad_norm": 0.712459146976471, "learning_rate": 0.0001, "elbo": 3.58501, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.818730940652123}
{"step": 48070, "loss": 3.6054050445556642, "grad_norm": 0.7251251339912415, "learning_rate": 0.0001, "elbo": 3.60541, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8190241613886933}
{"step": 48075, "loss": 3.595236587524414, "grad_norm": 0.7061951160430908, "learning_rate": 0.0001, "elbo": 3.59524, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8193173821252637}
{"step": 48080, "loss": 3.5753402709960938, "grad_norm": 0.7425006628036499, "learning_rate": 0.0001, "elbo": 3.57534, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.819610602861834}
{"step": 48085, "loss": 3.6011608123779295, "grad_norm": 1.0588842630386353, "learning_rate": 0.0001, "elbo": 3.60116, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8199038235984046}
{"step": 48090, "loss": 3.5901832580566406, "grad_norm": 0.6827689409255981, "learning_rate": 0.0001, "elbo": 3.59018, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8201970443349755}
{"step": 48095, "loss": 3.5899253845214845, "grad_norm": 0.7373790144920349, "learning_rate": 0.0001, "elbo": 3.58993, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.820490265071546}
{"step": 48100, "loss": 3.571329116821289, "grad_norm": 0.7090916633605957, "learning_rate": 0.0001, "elbo": 3.57133, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8207834858081164}
{"step": 48105, "loss": 3.6031444549560545, "grad_norm": 0.6958142518997192, "learning_rate": 0.0001, "elbo": 3.60314, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.821076706544687}
{"step": 48110, "loss": 3.595730209350586, "grad_norm": 0.6595765352249146, "learning_rate": 0.0001, "elbo": 3.59573, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8213699272812574}
{"step": 48115, "loss": 3.583617401123047, "grad_norm": 0.7114038467407227, "learning_rate": 0.0001, "elbo": 3.58362, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.821663148017828}
{"step": 48120, "loss": 3.585091781616211, "grad_norm": 0.6659472584724426, "learning_rate": 0.0001, "elbo": 3.58509, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8219563687543983}
{"step": 48125, "loss": 3.5886474609375, "grad_norm": 0.781948447227478, "learning_rate": 0.0001, "elbo": 3.58865, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8222495894909687}
{"step": 48130, "loss": 3.5843536376953127, "grad_norm": 0.6651259660720825, "learning_rate": 0.0001, "elbo": 3.58435, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.822542810227539}
{"step": 48135, "loss": 3.604589080810547, "grad_norm": 0.6737591624259949, "learning_rate": 0.0001, "elbo": 3.60459, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8228360309641096}
{"step": 48140, "loss": 3.587209701538086, "grad_norm": 0.6494495868682861, "learning_rate": 0.0001, "elbo": 3.58721, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8231292517006805}
{"step": 48145, "loss": 3.5762535095214845, "grad_norm": 0.8167617917060852, "learning_rate": 0.0001, "elbo": 3.57625, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.823422472437251}
{"step": 48150, "loss": 3.5791652679443358, "grad_norm": 0.6630973219871521, "learning_rate": 0.0001, "elbo": 3.57917, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8237156931738214}
{"step": 48155, "loss": 3.586030960083008, "grad_norm": 0.6878135204315186, "learning_rate": 0.0001, "elbo": 3.58603, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.824008913910392}
{"step": 48160, "loss": 3.592127227783203, "grad_norm": 0.7089864015579224, "learning_rate": 0.0001, "elbo": 3.59213, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8243021346469623}
{"step": 48165, "loss": 3.593993377685547, "grad_norm": 0.833082914352417, "learning_rate": 0.0001, "elbo": 3.59399, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.824595355383533}
{"step": 48170, "loss": 3.5707469940185548, "grad_norm": 1.0433320999145508, "learning_rate": 0.0001, "elbo": 3.57075, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8248885761201032}
{"step": 48175, "loss": 3.582216262817383, "grad_norm": 0.7049599885940552, "learning_rate": 0.0001, "elbo": 3.58222, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8251817968566737}
{"step": 48180, "loss": 3.592778778076172, "grad_norm": 0.6515222787857056, "learning_rate": 0.0001, "elbo": 3.59278, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.825475017593244}
{"step": 48185, "loss": 3.5956817626953126, "grad_norm": 0.6876358389854431, "learning_rate": 0.0001, "elbo": 3.59568, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8257682383298146}
{"step": 48185, "loss": 3.5956817626953126, "grad_norm": 0.6876358389854431, "learning_rate": 0.0001, "elbo": 3.59568, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8257682383298146}
{"step": 48190, "loss": 3.5903221130371095, "grad_norm": 0.7443892359733582, "learning_rate": 0.0001, "elbo": 3.59032, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.826061459066385}
{"step": 48195, "loss": 3.587598419189453, "grad_norm": 0.7573990821838379, "learning_rate": 0.0001, "elbo": 3.5876, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8263546798029555}
{"step": 48200, "loss": 3.605045700073242, "grad_norm": 0.7947487831115723, "learning_rate": 0.0001, "elbo": 3.60505, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.826647900539526}
{"step": 48205, "loss": 3.5972137451171875, "grad_norm": 0.6777601838111877, "learning_rate": 0.0001, "elbo": 3.59721, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8269411212760964}
{"step": 48210, "loss": 3.6025245666503904, "grad_norm": 0.7287163734436035, "learning_rate": 0.0001, "elbo": 3.60252, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.827234342012667}
{"step": 48215, "loss": 3.5820034027099608, "grad_norm": 0.6317217350006104, "learning_rate": 0.0001, "elbo": 3.582, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8275275627492378}
{"step": 48220, "loss": 3.600476837158203, "grad_norm": 0.6501202583312988, "learning_rate": 0.0001, "elbo": 3.60048, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8278207834858082}
{"step": 48225, "loss": 3.596971893310547, "grad_norm": 0.7483025789260864, "learning_rate": 0.0001, "elbo": 3.59697, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8281140042223787}
{"step": 48230, "loss": 3.590591812133789, "grad_norm": 0.9656161069869995, "learning_rate": 0.0001, "elbo": 3.59059, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.828407224958949}
{"step": 48235, "loss": 3.5914745330810547, "grad_norm": 0.7332764863967896, "learning_rate": 0.0001, "elbo": 3.59147, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8287004456955196}
{"step": 48240, "loss": 3.5784629821777343, "grad_norm": 0.672554075717926, "learning_rate": 0.0001, "elbo": 3.57846, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.82899366643209}
{"step": 48245, "loss": 3.6020553588867186, "grad_norm": 0.7718132138252258, "learning_rate": 0.0001, "elbo": 3.60206, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8292868871686605}
{"step": 48250, "loss": 3.5818042755126953, "grad_norm": 0.7395727038383484, "learning_rate": 0.0001, "elbo": 3.5818, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.829580107905231}
{"step": 48255, "loss": 3.601213836669922, "grad_norm": 0.7317253947257996, "learning_rate": 0.0001, "elbo": 3.60121, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8298733286418014}
{"step": 48260, "loss": 3.573194885253906, "grad_norm": 0.682072639465332, "learning_rate": 0.0001, "elbo": 3.5732, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8301665493783723}
{"step": 48265, "loss": 3.576974105834961, "grad_norm": 0.6864575147628784, "learning_rate": 0.0001, "elbo": 3.57697, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8304597701149428}
{"step": 48270, "loss": 3.580431365966797, "grad_norm": 0.6727749705314636, "learning_rate": 0.0001, "elbo": 3.58043, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.830752990851513}
{"step": 48275, "loss": 3.590250778198242, "grad_norm": 0.705597460269928, "learning_rate": 0.0001, "elbo": 3.59025, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8310462115880837}
{"step": 48280, "loss": 3.589580535888672, "grad_norm": 0.7031518220901489, "learning_rate": 0.0001, "elbo": 3.58958, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.831339432324654}
{"step": 48280, "loss": 3.589580535888672, "grad_norm": 0.7031518220901489, "learning_rate": 0.0001, "elbo": 3.58958, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.831339432324654}
{"step": 48285, "loss": 3.5748264312744142, "grad_norm": 0.6816630363464355, "learning_rate": 0.0001, "elbo": 3.57483, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8316326530612246}
{"step": 48290, "loss": 3.579215240478516, "grad_norm": 0.7008827328681946, "learning_rate": 0.0001, "elbo": 3.57922, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.831925873797795}
{"step": 48295, "loss": 3.6120460510253904, "grad_norm": 0.7019271850585938, "learning_rate": 0.0001, "elbo": 3.61205, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8322190945343655}
{"step": 48300, "loss": 3.588861083984375, "grad_norm": 0.786975622177124, "learning_rate": 0.0001, "elbo": 3.58886, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.832512315270936}
{"step": 48305, "loss": 3.602434539794922, "grad_norm": 0.7336483001708984, "learning_rate": 0.0001, "elbo": 3.60243, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8328055360075064}
{"step": 48310, "loss": 3.6018463134765626, "grad_norm": 0.7450301051139832, "learning_rate": 0.0001, "elbo": 3.60185, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.833098756744077}
{"step": 48315, "loss": 3.5910091400146484, "grad_norm": 0.6689745187759399, "learning_rate": 0.0001, "elbo": 3.59101, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8333919774806473}
{"step": 48320, "loss": 3.5950027465820313, "grad_norm": 0.6998768448829651, "learning_rate": 0.0001, "elbo": 3.595, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8336851982172178}
{"step": 48325, "loss": 3.5821464538574217, "grad_norm": 0.8378732204437256, "learning_rate": 0.0001, "elbo": 3.58215, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.833978418953788}
{"step": 48330, "loss": 3.5841495513916017, "grad_norm": 0.949922502040863, "learning_rate": 0.0001, "elbo": 3.58415, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8342716396903587}
{"step": 48335, "loss": 3.585112380981445, "grad_norm": 0.6957424879074097, "learning_rate": 0.0001, "elbo": 3.58511, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.834564860426929}
{"step": 48340, "loss": 3.595311737060547, "grad_norm": 0.6490244269371033, "learning_rate": 0.0001, "elbo": 3.59531, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8348580811635}
{"step": 48345, "loss": 3.5995994567871095, "grad_norm": 0.7407334446907043, "learning_rate": 0.0001, "elbo": 3.5996, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8351513019000705}
{"step": 48350, "loss": 3.5902694702148437, "grad_norm": 0.6783775687217712, "learning_rate": 0.0001, "elbo": 3.59027, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.835444522636641}
{"step": 48355, "loss": 3.5889667510986327, "grad_norm": 0.7709188461303711, "learning_rate": 0.0001, "elbo": 3.58897, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8357377433732114}
{"step": 48360, "loss": 3.6052703857421875, "grad_norm": 0.6315714120864868, "learning_rate": 0.0001, "elbo": 3.60527, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.836030964109782}
{"step": 48365, "loss": 3.609510803222656, "grad_norm": 0.6548410654067993, "learning_rate": 0.0001, "elbo": 3.60951, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8363241848463523}
{"step": 48365, "loss": 3.609510803222656, "grad_norm": 0.6548410654067993, "learning_rate": 0.0001, "elbo": 3.60951, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8363241848463523}
{"step": 48370, "loss": 3.5880306243896483, "grad_norm": 0.7440081238746643, "learning_rate": 0.0001, "elbo": 3.58803, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8366174055829227}
{"step": 48375, "loss": 3.587882232666016, "grad_norm": 0.6797226071357727, "learning_rate": 0.0001, "elbo": 3.58788, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.836910626319493}
{"step": 48380, "loss": 3.600804901123047, "grad_norm": 0.7069880366325378, "learning_rate": 0.0001, "elbo": 3.6008, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8372038470560637}
{"step": 48385, "loss": 3.5834880828857423, "grad_norm": 0.682422935962677, "learning_rate": 0.0001, "elbo": 3.58349, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8374970677926346}
{"step": 48390, "loss": 3.5863616943359373, "grad_norm": 0.7289732098579407, "learning_rate": 0.0001, "elbo": 3.58636, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.837790288529205}
{"step": 48395, "loss": 3.5880157470703127, "grad_norm": 0.6900163292884827, "learning_rate": 0.0001, "elbo": 3.58802, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8380835092657755}
{"step": 48400, "loss": 3.593230438232422, "grad_norm": 0.7083355188369751, "learning_rate": 0.0001, "elbo": 3.59323, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.838376730002346}
{"step": 48405, "loss": 3.5872093200683595, "grad_norm": 0.6976649761199951, "learning_rate": 0.0001, "elbo": 3.58721, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8386699507389164}
{"step": 48410, "loss": 3.604533386230469, "grad_norm": 0.7055498361587524, "learning_rate": 0.0001, "elbo": 3.60453, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.838963171475487}
{"step": 48415, "loss": 3.5719627380371093, "grad_norm": 0.6799541711807251, "learning_rate": 0.0001, "elbo": 3.57196, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8392563922120573}
{"step": 48420, "loss": 3.5960716247558593, "grad_norm": 0.6853131055831909, "learning_rate": 0.0001, "elbo": 3.59607, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8395496129486277}
{"step": 48425, "loss": 3.600341796875, "grad_norm": 0.7066283822059631, "learning_rate": 0.0001, "elbo": 3.60034, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.839842833685198}
{"step": 48430, "loss": 3.5869895935058596, "grad_norm": 0.7119491696357727, "learning_rate": 0.0001, "elbo": 3.58699, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8401360544217686}
{"step": 48435, "loss": 3.6096458435058594, "grad_norm": 0.63228839635849, "learning_rate": 0.0001, "elbo": 3.60965, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.840429275158339}
{"step": 48440, "loss": 3.5779090881347657, "grad_norm": 0.6458005905151367, "learning_rate": 0.0001, "elbo": 3.57791, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8407224958949095}
{"step": 48445, "loss": 3.589146041870117, "grad_norm": 0.7539469599723816, "learning_rate": 0.0001, "elbo": 3.58915, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.84101571663148}
{"step": 48445, "loss": 3.589146041870117, "grad_norm": 0.7539469599723816, "learning_rate": 0.0001, "elbo": 3.58915, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.84101571663148}
{"step": 48450, "loss": 3.602014923095703, "grad_norm": 0.6867700219154358, "learning_rate": 0.0001, "elbo": 3.60202, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8413089373680505}
{"step": 48455, "loss": 3.579485321044922, "grad_norm": 0.8115402460098267, "learning_rate": 0.0001, "elbo": 3.57949, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.841602158104621}
{"step": 48460, "loss": 3.580892562866211, "grad_norm": 0.704116702079773, "learning_rate": 0.0001, "elbo": 3.58089, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.841895378841192}
{"step": 48465, "loss": 3.5860454559326174, "grad_norm": 0.7114552855491638, "learning_rate": 0.0001, "elbo": 3.58605, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8421885995777623}
{"step": 48470, "loss": 3.591852569580078, "grad_norm": 0.7298815846443176, "learning_rate": 0.0001, "elbo": 3.59185, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8424818203143327}
{"step": 48475, "loss": 3.6059715270996096, "grad_norm": 0.754263699054718, "learning_rate": 0.0001, "elbo": 3.60597, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.842775041050903}
{"step": 48480, "loss": 3.5876190185546877, "grad_norm": 0.7338816523551941, "learning_rate": 0.0001, "elbo": 3.58762, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8430682617874736}
{"step": 48485, "loss": 3.5965591430664063, "grad_norm": 0.7314026355743408, "learning_rate": 0.0001, "elbo": 3.59656, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.843361482524044}
{"step": 48490, "loss": 3.6019027709960936, "grad_norm": 0.6903732419013977, "learning_rate": 0.0001, "elbo": 3.6019, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8436547032606145}
{"step": 48495, "loss": 3.5982078552246093, "grad_norm": 0.7365694046020508, "learning_rate": 0.0001, "elbo": 3.59821, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.843947923997185}
{"step": 48500, "loss": 3.5651424407958983, "grad_norm": 0.7194095849990845, "learning_rate": 0.0001, "elbo": 3.56514, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8442411447337554}
{"step": 48505, "loss": 3.60101432800293, "grad_norm": 0.6797987818717957, "learning_rate": 0.0001, "elbo": 3.60101, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.844534365470326}
{"step": 48510, "loss": 3.580117034912109, "grad_norm": 0.7541669607162476, "learning_rate": 0.0001, "elbo": 3.58012, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.844827586206897}
{"step": 48515, "loss": 3.5707855224609375, "grad_norm": 0.6748899221420288, "learning_rate": 0.0001, "elbo": 3.57079, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8451208069434673}
{"step": 48520, "loss": 3.58416862487793, "grad_norm": 0.7528004050254822, "learning_rate": 0.0001, "elbo": 3.58417, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8454140276800377}
{"step": 48525, "loss": 3.584705352783203, "grad_norm": 0.6667292714118958, "learning_rate": 0.0001, "elbo": 3.58471, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.845707248416608}
{"step": 48530, "loss": 3.5948814392089843, "grad_norm": 0.6892182230949402, "learning_rate": 0.0001, "elbo": 3.59488, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8460004691531786}
{"step": 48535, "loss": 3.593410110473633, "grad_norm": 0.7223817706108093, "learning_rate": 0.0001, "elbo": 3.59341, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.846293689889749}
{"step": 48540, "loss": 3.5972747802734375, "grad_norm": 0.7054776549339294, "learning_rate": 0.0001, "elbo": 3.59727, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8465869106263195}
{"step": 48545, "loss": 3.603045654296875, "grad_norm": 0.7182217836380005, "learning_rate": 0.0001, "elbo": 3.60305, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.84688013136289}
{"step": 48550, "loss": 3.595716094970703, "grad_norm": 0.6723522543907166, "learning_rate": 0.0001, "elbo": 3.59572, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8471733520994604}
{"step": 48555, "loss": 3.5914752960205076, "grad_norm": 0.7014936208724976, "learning_rate": 0.0001, "elbo": 3.59148, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.847466572836031}
{"step": 48560, "loss": 3.584605026245117, "grad_norm": 0.6895297765731812, "learning_rate": 0.0001, "elbo": 3.58461, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8477597935726013}
{"step": 48565, "loss": 3.5942367553710937, "grad_norm": 0.667567789554596, "learning_rate": 0.0001, "elbo": 3.59424, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.848053014309172}
{"step": 48570, "loss": 3.5934181213378906, "grad_norm": 0.7122998237609863, "learning_rate": 0.0001, "elbo": 3.59342, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8483462350457422}
{"step": 48575, "loss": 3.5921958923339843, "grad_norm": 0.691985011100769, "learning_rate": 0.0001, "elbo": 3.5922, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8486394557823127}
{"step": 48580, "loss": 3.5793746948242187, "grad_norm": 0.6738830208778381, "learning_rate": 0.0001, "elbo": 3.57937, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.848932676518883}
{"step": 48585, "loss": 3.594499206542969, "grad_norm": 0.7044280171394348, "learning_rate": 0.0001, "elbo": 3.5945, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.849225897255454}
{"step": 48590, "loss": 3.583544921875, "grad_norm": 0.6598756909370422, "learning_rate": 0.0001, "elbo": 3.58355, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8495191179920245}
{"step": 48595, "loss": 3.5722930908203123, "grad_norm": 0.7689363956451416, "learning_rate": 0.0001, "elbo": 3.57229, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.849812338728595}
{"step": 48600, "loss": 3.600069046020508, "grad_norm": 0.7381520867347717, "learning_rate": 0.0001, "elbo": 3.60007, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8501055594651654}
{"step": 48605, "loss": 3.587795639038086, "grad_norm": 0.7566017508506775, "learning_rate": 0.0001, "elbo": 3.5878, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.850398780201736}
{"step": 48605, "loss": 3.587795639038086, "grad_norm": 0.7566017508506775, "learning_rate": 0.0001, "elbo": 3.5878, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.850398780201736}
{"step": 48610, "loss": 3.6012619018554686, "grad_norm": 0.7456879019737244, "learning_rate": 0.0001, "elbo": 3.60126, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8506920009383063}
{"step": 48615, "loss": 3.5624931335449217, "grad_norm": 0.8104375004768372, "learning_rate": 0.0001, "elbo": 3.56249, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.850985221674877}
{"step": 48620, "loss": 3.589060974121094, "grad_norm": 0.6685087084770203, "learning_rate": 0.0001, "elbo": 3.58906, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8512784424114472}
{"step": 48620, "loss": 3.589060974121094, "grad_norm": 0.6685087084770203, "learning_rate": 0.0001, "elbo": 3.58906, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8512784424114472}
{"step": 48625, "loss": 3.5863494873046875, "grad_norm": 0.6798651218414307, "learning_rate": 0.0001, "elbo": 3.58635, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8515716631480177}
{"step": 48630, "loss": 3.5967533111572267, "grad_norm": 0.7874587774276733, "learning_rate": 0.0001, "elbo": 3.59675, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8518648838845886}
{"step": 48635, "loss": 3.569712829589844, "grad_norm": 0.6618152260780334, "learning_rate": 0.0001, "elbo": 3.56971, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.852158104621159}
{"step": 48640, "loss": 3.5779380798339844, "grad_norm": 0.7045753598213196, "learning_rate": 0.0001, "elbo": 3.57794, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8524513253577295}
{"step": 48645, "loss": 3.5783756256103514, "grad_norm": 0.7451196908950806, "learning_rate": 0.0001, "elbo": 3.57838, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8527445460943}
{"step": 48650, "loss": 3.5925960540771484, "grad_norm": 0.6734514832496643, "learning_rate": 0.0001, "elbo": 3.5926, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8530377668308704}
{"step": 48650, "loss": 3.5925960540771484, "grad_norm": 0.6734514832496643, "learning_rate": 0.0001, "elbo": 3.5926, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8530377668308704}
{"step": 48655, "loss": 3.601871109008789, "grad_norm": 0.6717541813850403, "learning_rate": 0.0001, "elbo": 3.60187, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.853330987567441}
{"step": 48660, "loss": 3.5927616119384767, "grad_norm": 0.6845998764038086, "learning_rate": 0.0001, "elbo": 3.59276, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8536242083040113}
{"step": 48665, "loss": 3.601106643676758, "grad_norm": 0.7057159543037415, "learning_rate": 0.0001, "elbo": 3.60111, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8539174290405818}
{"step": 48670, "loss": 3.5904586791992186, "grad_norm": 0.7764465808868408, "learning_rate": 0.0001, "elbo": 3.59046, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.854210649777152}
{"step": 48675, "loss": 3.580636215209961, "grad_norm": 0.7227346301078796, "learning_rate": 0.0001, "elbo": 3.58064, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8545038705137227}
{"step": 48680, "loss": 3.597705841064453, "grad_norm": 0.6629343628883362, "learning_rate": 0.0001, "elbo": 3.59771, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.854797091250293}
{"step": 48685, "loss": 3.595040130615234, "grad_norm": 0.6724332571029663, "learning_rate": 0.0001, "elbo": 3.59504, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8550903119868636}
{"step": 48690, "loss": 3.5775192260742186, "grad_norm": 0.6645985841751099, "learning_rate": 0.0001, "elbo": 3.57752, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.855383532723434}
{"step": 48695, "loss": 3.6013946533203125, "grad_norm": 0.8439652323722839, "learning_rate": 0.0001, "elbo": 3.60139, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8556767534600045}
{"step": 48695, "loss": 3.6013946533203125, "grad_norm": 0.8439652323722839, "learning_rate": 0.0001, "elbo": 3.60139, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8556767534600045}
{"step": 48700, "loss": 3.5881393432617186, "grad_norm": 0.6439017653465271, "learning_rate": 0.0001, "elbo": 3.58814, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.855969974196575}
{"step": 48705, "loss": 3.590744400024414, "grad_norm": 0.6619369387626648, "learning_rate": 0.0001, "elbo": 3.59074, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8562631949331454}
{"step": 48705, "loss": 3.590744400024414, "grad_norm": 0.6619369387626648, "learning_rate": 0.0001, "elbo": 3.59074, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8562631949331454}
{"step": 48710, "loss": 3.581876754760742, "grad_norm": 0.9662215113639832, "learning_rate": 0.0001, "elbo": 3.58188, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8565564156697163}
{"step": 48715, "loss": 3.597910690307617, "grad_norm": 0.6975094676017761, "learning_rate": 0.0001, "elbo": 3.59791, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8568496364062868}
{"step": 48720, "loss": 3.5941497802734377, "grad_norm": 0.8115341067314148, "learning_rate": 0.0001, "elbo": 3.59415, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.857142857142857}
{"step": 48725, "loss": 3.5895336151123045, "grad_norm": 1.1485283374786377, "learning_rate": 0.0001, "elbo": 3.58953, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8574360778794277}
{"step": 48730, "loss": 3.577406311035156, "grad_norm": 0.8471612930297852, "learning_rate": 0.0001, "elbo": 3.57741, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.857729298615998}
{"step": 48730, "loss": 3.577406311035156, "grad_norm": 0.8471612930297852, "learning_rate": 0.0001, "elbo": 3.57741, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.857729298615998}
{"step": 48735, "loss": 3.599971389770508, "grad_norm": 0.7655885815620422, "learning_rate": 0.0001, "elbo": 3.59997, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8580225193525686}
{"step": 48740, "loss": 3.5776824951171875, "grad_norm": 0.7056583166122437, "learning_rate": 0.0001, "elbo": 3.57768, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.858315740089139}
{"step": 48745, "loss": 3.5962753295898438, "grad_norm": 0.7226617932319641, "learning_rate": 0.0001, "elbo": 3.59628, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8586089608257095}
{"step": 48750, "loss": 3.6049549102783205, "grad_norm": 0.6519696116447449, "learning_rate": 0.0001, "elbo": 3.60495, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.85890218156228}
{"step": 48755, "loss": 3.607114791870117, "grad_norm": 0.6772785782814026, "learning_rate": 0.0001, "elbo": 3.60711, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.859195402298851}
{"step": 48760, "loss": 3.592102813720703, "grad_norm": 0.7861838340759277, "learning_rate": 0.0001, "elbo": 3.5921, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8594886230354213}
{"step": 48765, "loss": 3.6082221984863283, "grad_norm": 0.6725485324859619, "learning_rate": 0.0001, "elbo": 3.60822, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8597818437719917}
{"step": 48770, "loss": 3.586394500732422, "grad_norm": 0.693876326084137, "learning_rate": 0.0001, "elbo": 3.58639, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.860075064508562}
{"step": 48775, "loss": 3.589617919921875, "grad_norm": 0.7776564955711365, "learning_rate": 0.0001, "elbo": 3.58962, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8603682852451326}
{"step": 48780, "loss": 3.592156982421875, "grad_norm": 0.7856389880180359, "learning_rate": 0.0001, "elbo": 3.59216, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.860661505981703}
{"step": 48785, "loss": 3.590370941162109, "grad_norm": 0.8905066251754761, "learning_rate": 0.0001, "elbo": 3.59037, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8609547267182736}
{"step": 48790, "loss": 3.5952831268310548, "grad_norm": 0.6602928042411804, "learning_rate": 0.0001, "elbo": 3.59528, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.861247947454844}
{"step": 48795, "loss": 3.594539260864258, "grad_norm": 0.7153046727180481, "learning_rate": 0.0001, "elbo": 3.59454, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8615411681914145}
{"step": 48795, "loss": 3.594539260864258, "grad_norm": 0.7153046727180481, "learning_rate": 0.0001, "elbo": 3.59454, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8615411681914145}
{"step": 48800, "loss": 3.578034210205078, "grad_norm": 0.8395607471466064, "learning_rate": 0.0001, "elbo": 3.57803, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.861834388927985}
{"step": 48805, "loss": 3.5776699066162108, "grad_norm": 0.8674225807189941, "learning_rate": 0.0001, "elbo": 3.57767, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8621276096645554}
{"step": 48810, "loss": 3.588436508178711, "grad_norm": 0.7451673150062561, "learning_rate": 0.0001, "elbo": 3.58844, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.862420830401126}
{"step": 48810, "loss": 3.588436508178711, "grad_norm": 0.7451673150062561, "learning_rate": 0.0001, "elbo": 3.58844, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.862420830401126}
{"step": 48815, "loss": 3.593381881713867, "grad_norm": 0.7206103801727295, "learning_rate": 0.0001, "elbo": 3.59338, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8627140511376963}
{"step": 48820, "loss": 3.5836818695068358, "grad_norm": 0.6613101363182068, "learning_rate": 0.0001, "elbo": 3.58368, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8630072718742667}
{"step": 48825, "loss": 3.6047695159912108, "grad_norm": 0.7180674076080322, "learning_rate": 0.0001, "elbo": 3.60477, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.863300492610837}
{"step": 48830, "loss": 3.5879142761230467, "grad_norm": 0.7117136120796204, "learning_rate": 0.0001, "elbo": 3.58791, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.863593713347408}
{"step": 48830, "loss": 3.5879142761230467, "grad_norm": 0.7117136120796204, "learning_rate": 0.0001, "elbo": 3.58791, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.863593713347408}
{"step": 48835, "loss": 3.5854393005371095, "grad_norm": 0.6742342114448547, "learning_rate": 0.0001, "elbo": 3.58544, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8638869340839785}
{"step": 48840, "loss": 3.594277572631836, "grad_norm": 0.6834538578987122, "learning_rate": 0.0001, "elbo": 3.59428, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.864180154820549}
{"step": 48845, "loss": 3.580774688720703, "grad_norm": 0.8042707443237305, "learning_rate": 0.0001, "elbo": 3.58077, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8644733755571195}
{"step": 48845, "loss": 3.580774688720703, "grad_norm": 0.8042707443237305, "learning_rate": 0.0001, "elbo": 3.58077, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8644733755571195}
{"step": 48850, "loss": 3.5856643676757813, "grad_norm": 0.7409772276878357, "learning_rate": 0.0001, "elbo": 3.58566, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.86476659629369}
{"step": 48855, "loss": 3.5828388214111326, "grad_norm": 0.7286362648010254, "learning_rate": 0.0001, "elbo": 3.58284, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8650598170302604}
{"step": 48860, "loss": 3.5920284271240233, "grad_norm": 0.7023428678512573, "learning_rate": 0.0001, "elbo": 3.59203, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.865353037766831}
{"step": 48865, "loss": 3.6041168212890624, "grad_norm": 0.7259570956230164, "learning_rate": 0.0001, "elbo": 3.60412, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8656462585034013}
{"step": 48870, "loss": 3.578032684326172, "grad_norm": 0.7264137268066406, "learning_rate": 0.0001, "elbo": 3.57803, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8659394792399717}
{"step": 48875, "loss": 3.5891845703125, "grad_norm": 0.7968385219573975, "learning_rate": 0.0001, "elbo": 3.58918, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.866232699976542}
{"step": 48880, "loss": 3.587359619140625, "grad_norm": 0.6844469308853149, "learning_rate": 0.0001, "elbo": 3.58736, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.866525920713113}
{"step": 48885, "loss": 3.578134536743164, "grad_norm": 0.6809307336807251, "learning_rate": 0.0001, "elbo": 3.57813, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8668191414496835}
{"step": 48890, "loss": 3.587718963623047, "grad_norm": 0.7456210851669312, "learning_rate": 0.0001, "elbo": 3.58772, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.867112362186254}
{"step": 48890, "loss": 3.587718963623047, "grad_norm": 0.7456210851669312, "learning_rate": 0.0001, "elbo": 3.58772, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.867112362186254}
{"step": 48895, "loss": 3.598328399658203, "grad_norm": 0.6946874260902405, "learning_rate": 0.0001, "elbo": 3.59833, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8674055829228244}
{"step": 48900, "loss": 3.5745655059814454, "grad_norm": 0.761196494102478, "learning_rate": 0.0001, "elbo": 3.57457, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.867698803659395}
{"step": 48905, "loss": 3.5845497131347654, "grad_norm": 0.7769109606742859, "learning_rate": 0.0001, "elbo": 3.58455, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8679920243959653}
{"step": 48910, "loss": 3.5865009307861326, "grad_norm": 0.6636574268341064, "learning_rate": 0.0001, "elbo": 3.5865, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.868285245132536}
{"step": 48915, "loss": 3.584941101074219, "grad_norm": 0.710665225982666, "learning_rate": 0.0001, "elbo": 3.58494, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8685784658691063}
{"step": 48920, "loss": 3.588914489746094, "grad_norm": 0.6713356971740723, "learning_rate": 0.0001, "elbo": 3.58891, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8688716866056767}
{"step": 48925, "loss": 3.587451171875, "grad_norm": 0.7066166400909424, "learning_rate": 0.0001, "elbo": 3.58745, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.869164907342247}
{"step": 48930, "loss": 3.598846435546875, "grad_norm": 0.7184794545173645, "learning_rate": 0.0001, "elbo": 3.59885, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8694581280788176}
{"step": 48935, "loss": 3.590793991088867, "grad_norm": 0.7315756678581238, "learning_rate": 0.0001, "elbo": 3.59079, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.869751348815388}
{"step": 48940, "loss": 3.6054412841796877, "grad_norm": 0.672163188457489, "learning_rate": 0.0001, "elbo": 3.60544, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8700445695519585}
{"step": 48945, "loss": 3.6089996337890624, "grad_norm": 0.7284401655197144, "learning_rate": 0.0001, "elbo": 3.609, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.870337790288529}
{"step": 48950, "loss": 3.593903350830078, "grad_norm": 0.8464678525924683, "learning_rate": 0.0001, "elbo": 3.5939, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8706310110250994}
{"step": 48955, "loss": 3.6104236602783204, "grad_norm": 0.7243134379386902, "learning_rate": 0.0001, "elbo": 3.61042, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8709242317616703}
{"step": 48960, "loss": 3.5918888092041015, "grad_norm": 0.6949045658111572, "learning_rate": 0.0001, "elbo": 3.59189, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.871217452498241}
{"step": 48965, "loss": 3.5856456756591797, "grad_norm": 0.6613202095031738, "learning_rate": 0.0001, "elbo": 3.58565, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8715106732348112}
{"step": 48970, "loss": 3.5950340270996093, "grad_norm": 0.684049129486084, "learning_rate": 0.0001, "elbo": 3.59503, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8718038939713817}
{"step": 48975, "loss": 3.607282257080078, "grad_norm": 0.7387725710868835, "learning_rate": 0.0001, "elbo": 3.60728, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.872097114707952}
{"step": 48980, "loss": 3.5777442932128904, "grad_norm": 0.6608692407608032, "learning_rate": 0.0001, "elbo": 3.57774, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8723903354445226}
{"step": 48985, "loss": 3.5826431274414063, "grad_norm": 1.1535495519638062, "learning_rate": 0.0001, "elbo": 3.58264, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.872683556181093}
{"step": 48990, "loss": 3.5800430297851564, "grad_norm": 0.7418989539146423, "learning_rate": 0.0001, "elbo": 3.58004, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8729767769176635}
{"step": 48995, "loss": 3.589350128173828, "grad_norm": 0.7214261889457703, "learning_rate": 0.0001, "elbo": 3.58935, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.873269997654234}
{"step": 49000, "loss": 3.5761482238769533, "grad_norm": 0.6924867033958435, "learning_rate": 0.0001, "elbo": 3.57615, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.873563218390805}
{"step": 49005, "loss": 3.583260345458984, "grad_norm": 0.6700777411460876, "learning_rate": 0.0001, "elbo": 3.58326, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8738564391273753}
{"step": 49010, "loss": 3.601084899902344, "grad_norm": 0.6436354517936707, "learning_rate": 0.0001, "elbo": 3.60109, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8741496598639458}
{"step": 49010, "loss": 3.601084899902344, "grad_norm": 0.6436354517936707, "learning_rate": 0.0001, "elbo": 3.60109, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8741496598639458}
{"step": 49015, "loss": 3.6003646850585938, "grad_norm": 0.7267317771911621, "learning_rate": 0.0001, "elbo": 3.60036, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8744428806005162}
{"step": 49020, "loss": 3.591569519042969, "grad_norm": 0.7440914511680603, "learning_rate": 0.0001, "elbo": 3.59157, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8747361013370867}
{"step": 49020, "loss": 3.591569519042969, "grad_norm": 0.7440914511680603, "learning_rate": 0.0001, "elbo": 3.59157, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8747361013370867}
{"step": 49025, "loss": 3.580083465576172, "grad_norm": 0.6671454310417175, "learning_rate": 0.0001, "elbo": 3.58008, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.875029322073657}
{"step": 49030, "loss": 3.588827896118164, "grad_norm": 0.7148454189300537, "learning_rate": 0.0001, "elbo": 3.58883, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8753225428102276}
{"step": 49035, "loss": 3.587702178955078, "grad_norm": 0.6582583785057068, "learning_rate": 0.0001, "elbo": 3.5877, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.875615763546798}
{"step": 49040, "loss": 3.5720726013183595, "grad_norm": 0.744140625, "learning_rate": 0.0001, "elbo": 3.57207, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8759089842833685}
{"step": 49045, "loss": 3.5700424194335936, "grad_norm": 1.0250592231750488, "learning_rate": 0.0001, "elbo": 3.57004, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.876202205019939}
{"step": 49045, "loss": 3.5700424194335936, "grad_norm": 1.0250592231750488, "learning_rate": 0.0001, "elbo": 3.57004, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.876202205019939}
{"step": 49050, "loss": 3.5913249969482424, "grad_norm": 0.6983153820037842, "learning_rate": 0.0001, "elbo": 3.59132, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8764954257565094}
{"step": 49055, "loss": 3.5998023986816405, "grad_norm": 0.690240204334259, "learning_rate": 0.0001, "elbo": 3.5998, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.87678864649308}
{"step": 49060, "loss": 3.5934967041015624, "grad_norm": 0.7798309326171875, "learning_rate": 0.0001, "elbo": 3.5935, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8770818672296503}
{"step": 49065, "loss": 3.596030426025391, "grad_norm": 0.7307971715927124, "learning_rate": 0.0001, "elbo": 3.59603, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8773750879662208}
{"step": 49070, "loss": 3.594568634033203, "grad_norm": 0.7396032214164734, "learning_rate": 0.0001, "elbo": 3.59457, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8776683087027912}
{"step": 49075, "loss": 3.5777618408203127, "grad_norm": 0.7466796040534973, "learning_rate": 0.0001, "elbo": 3.57776, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8779615294393617}
{"step": 49080, "loss": 3.5959407806396486, "grad_norm": 0.7132869958877563, "learning_rate": 0.0001, "elbo": 3.59594, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8782547501759326}
{"step": 49085, "loss": 3.600910186767578, "grad_norm": 0.6884931921958923, "learning_rate": 0.0001, "elbo": 3.60091, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.878547970912503}
{"step": 49090, "loss": 3.598642349243164, "grad_norm": 0.8172630667686462, "learning_rate": 0.0001, "elbo": 3.59864, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8788411916490735}
{"step": 49095, "loss": 3.5976043701171876, "grad_norm": 0.6840195655822754, "learning_rate": 0.0001, "elbo": 3.5976, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.879134412385644}
{"step": 49100, "loss": 3.583940124511719, "grad_norm": 0.7025712132453918, "learning_rate": 0.0001, "elbo": 3.58394, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8794276331222144}
{"step": 49105, "loss": 3.5984146118164064, "grad_norm": 0.6707256436347961, "learning_rate": 0.0001, "elbo": 3.59841, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.879720853858785}
{"step": 49110, "loss": 3.583966827392578, "grad_norm": 0.855480968952179, "learning_rate": 0.0001, "elbo": 3.58397, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8800140745953553}
{"step": 49115, "loss": 3.584208297729492, "grad_norm": 0.702684223651886, "learning_rate": 0.0001, "elbo": 3.58421, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8803072953319258}
{"step": 49120, "loss": 3.5872207641601563, "grad_norm": 0.6874967813491821, "learning_rate": 0.0001, "elbo": 3.58722, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.880600516068496}
{"step": 49125, "loss": 3.5883811950683593, "grad_norm": 0.6898295879364014, "learning_rate": 0.0001, "elbo": 3.58838, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.880893736805067}
{"step": 49130, "loss": 3.571156311035156, "grad_norm": 0.6639970541000366, "learning_rate": 0.0001, "elbo": 3.57116, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8811869575416376}
{"step": 49135, "loss": 3.5957363128662108, "grad_norm": 0.8940324783325195, "learning_rate": 0.0001, "elbo": 3.59574, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.881480178278208}
{"step": 49140, "loss": 3.5981979370117188, "grad_norm": 0.9690030813217163, "learning_rate": 0.0001, "elbo": 3.5982, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8817733990147785}
{"step": 49145, "loss": 3.5891815185546876, "grad_norm": 1.0490049123764038, "learning_rate": 0.0001, "elbo": 3.58918, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.882066619751349}
{"step": 49150, "loss": 3.5858558654785155, "grad_norm": 0.6906324625015259, "learning_rate": 0.0001, "elbo": 3.58586, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8823598404879194}
{"step": 49155, "loss": 3.596237564086914, "grad_norm": 0.6774455904960632, "learning_rate": 0.0001, "elbo": 3.59624, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.88265306122449}
{"step": 49160, "loss": 3.579950714111328, "grad_norm": 0.6865398287773132, "learning_rate": 0.0001, "elbo": 3.57995, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8829462819610603}
{"step": 49165, "loss": 3.5840782165527343, "grad_norm": 0.6400513052940369, "learning_rate": 0.0001, "elbo": 3.58408, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8832395026976307}
{"step": 49170, "loss": 3.5945533752441405, "grad_norm": 0.6914107203483582, "learning_rate": 0.0001, "elbo": 3.59455, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.883532723434201}
{"step": 49175, "loss": 3.589713287353516, "grad_norm": 0.7148774266242981, "learning_rate": 0.0001, "elbo": 3.58971, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8838259441707716}
{"step": 49180, "loss": 3.571536636352539, "grad_norm": 0.6625553965568542, "learning_rate": 0.0001, "elbo": 3.57154, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.884119164907342}
{"step": 49185, "loss": 3.5724002838134767, "grad_norm": 0.6621999144554138, "learning_rate": 0.0001, "elbo": 3.5724, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8844123856439126}
{"step": 49190, "loss": 3.5886585235595705, "grad_norm": 0.6880343556404114, "learning_rate": 0.0001, "elbo": 3.58866, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.884705606380483}
{"step": 49195, "loss": 3.5762374877929686, "grad_norm": 0.6939440369606018, "learning_rate": 0.0001, "elbo": 3.57624, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8849988271170535}
{"step": 49200, "loss": 3.5981689453125, "grad_norm": 0.747749388217926, "learning_rate": 0.0001, "elbo": 3.59817, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8852920478536244}
{"step": 49205, "loss": 3.5907363891601562, "grad_norm": 0.8173255920410156, "learning_rate": 0.0001, "elbo": 3.59074, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.885585268590195}
{"step": 49210, "loss": 3.580149841308594, "grad_norm": 0.8644765615463257, "learning_rate": 0.0001, "elbo": 3.58015, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8858784893267653}
{"step": 49215, "loss": 3.5859092712402343, "grad_norm": 0.8072679042816162, "learning_rate": 0.0001, "elbo": 3.58591, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8861717100633357}
{"step": 49220, "loss": 3.589385986328125, "grad_norm": 0.7047495245933533, "learning_rate": 0.0001, "elbo": 3.58939, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.886464930799906}
{"step": 49225, "loss": 3.5956668853759766, "grad_norm": 0.8312768340110779, "learning_rate": 0.0001, "elbo": 3.59567, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8867581515364766}
{"step": 49230, "loss": 3.5698005676269533, "grad_norm": 0.7166750431060791, "learning_rate": 0.0001, "elbo": 3.5698, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.887051372273047}
{"step": 49235, "loss": 3.582907867431641, "grad_norm": 1.0002834796905518, "learning_rate": 0.0001, "elbo": 3.58291, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8873445930096175}
{"step": 49240, "loss": 3.589128112792969, "grad_norm": 0.6801019310951233, "learning_rate": 0.0001, "elbo": 3.58913, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.887637813746188}
{"step": 49245, "loss": 3.5911796569824217, "grad_norm": 0.6836490631103516, "learning_rate": 0.0001, "elbo": 3.59118, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.887931034482759}
{"step": 49250, "loss": 3.587380599975586, "grad_norm": 0.7120967507362366, "learning_rate": 0.0001, "elbo": 3.58738, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8882242552193294}
{"step": 49255, "loss": 3.5948726654052736, "grad_norm": 1.2607122659683228, "learning_rate": 0.0001, "elbo": 3.59487, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8885174759559}
{"step": 49260, "loss": 3.584065246582031, "grad_norm": 0.9063579440116882, "learning_rate": 0.0001, "elbo": 3.58407, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8888106966924703}
{"step": 49265, "loss": 3.5988525390625, "grad_norm": 0.6714107990264893, "learning_rate": 0.0001, "elbo": 3.59885, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8891039174290407}
{"step": 49270, "loss": 3.598277282714844, "grad_norm": 0.6647911071777344, "learning_rate": 0.0001, "elbo": 3.59828, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.889397138165611}
{"step": 49275, "loss": 3.593065643310547, "grad_norm": 0.7567979693412781, "learning_rate": 0.0001, "elbo": 3.59307, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8896903589021816}
{"step": 49280, "loss": 3.5774364471435547, "grad_norm": 0.8227294087409973, "learning_rate": 0.0001, "elbo": 3.57744, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.889983579638752}
{"step": 49285, "loss": 3.5843204498291015, "grad_norm": 0.715208113193512, "learning_rate": 0.0001, "elbo": 3.58432, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8902768003753225}
{"step": 49290, "loss": 3.585227203369141, "grad_norm": 0.7001368999481201, "learning_rate": 0.0001, "elbo": 3.58523, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.890570021111893}
{"step": 49295, "loss": 3.586931610107422, "grad_norm": 0.7038474082946777, "learning_rate": 0.0001, "elbo": 3.58693, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8908632418484634}
{"step": 49300, "loss": 3.598387908935547, "grad_norm": 0.6871969103813171, "learning_rate": 0.0001, "elbo": 3.59839, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.891156462585034}
{"step": 49305, "loss": 3.5837471008300783, "grad_norm": 0.8760956525802612, "learning_rate": 0.0001, "elbo": 3.58375, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8914496833216043}
{"step": 49310, "loss": 3.611700439453125, "grad_norm": 0.7214385271072388, "learning_rate": 0.0001, "elbo": 3.6117, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.891742904058175}
{"step": 49315, "loss": 3.5935073852539063, "grad_norm": 0.6837555766105652, "learning_rate": 0.0001, "elbo": 3.59351, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8920361247947453}
{"step": 49320, "loss": 3.6016956329345704, "grad_norm": 0.7391979098320007, "learning_rate": 0.0001, "elbo": 3.6017, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8923293455313157}
{"step": 49325, "loss": 3.6005382537841797, "grad_norm": 0.7132082581520081, "learning_rate": 0.0001, "elbo": 3.60054, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8926225662678866}
{"step": 49330, "loss": 3.6079391479492187, "grad_norm": 0.6971452832221985, "learning_rate": 0.0001, "elbo": 3.60794, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.892915787004457}
{"step": 49335, "loss": 3.585352325439453, "grad_norm": 0.7691048383712769, "learning_rate": 0.0001, "elbo": 3.58535, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8932090077410275}
{"step": 49340, "loss": 3.5863597869873045, "grad_norm": 0.7180370688438416, "learning_rate": 0.0001, "elbo": 3.58636, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.893502228477598}
{"step": 49345, "loss": 3.5778602600097655, "grad_norm": 0.7147272825241089, "learning_rate": 0.0001, "elbo": 3.57786, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8937954492141684}
{"step": 49350, "loss": 3.585162353515625, "grad_norm": 0.7056956887245178, "learning_rate": 0.0001, "elbo": 3.58516, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.894088669950739}
{"step": 49355, "loss": 3.588727569580078, "grad_norm": 0.6913123726844788, "learning_rate": 0.0001, "elbo": 3.58873, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8943818906873093}
{"step": 49360, "loss": 3.582875061035156, "grad_norm": 0.8486095070838928, "learning_rate": 0.0001, "elbo": 3.58288, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.89467511142388}
{"step": 49365, "loss": 3.5923961639404296, "grad_norm": 0.7002107501029968, "learning_rate": 0.0001, "elbo": 3.5924, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8949683321604502}
{"step": 49370, "loss": 3.583156204223633, "grad_norm": 0.700397789478302, "learning_rate": 0.0001, "elbo": 3.58316, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.895261552897021}
{"step": 49375, "loss": 3.5938385009765623, "grad_norm": 0.8938983082771301, "learning_rate": 0.0001, "elbo": 3.59384, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8955547736335916}
{"step": 49380, "loss": 3.5721435546875, "grad_norm": 0.6782960891723633, "learning_rate": 0.0001, "elbo": 3.57214, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.895847994370162}
{"step": 49385, "loss": 3.5793197631835936, "grad_norm": 0.7397666573524475, "learning_rate": 0.0001, "elbo": 3.57932, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8961412151067325}
{"step": 49390, "loss": 3.5974655151367188, "grad_norm": 0.6895726919174194, "learning_rate": 0.0001, "elbo": 3.59747, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.896434435843303}
{"step": 49395, "loss": 3.585643768310547, "grad_norm": 0.7943494319915771, "learning_rate": 0.0001, "elbo": 3.58564, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8967276565798734}
{"step": 49400, "loss": 3.580707550048828, "grad_norm": 0.7003666162490845, "learning_rate": 0.0001, "elbo": 3.58071, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.897020877316444}
{"step": 49405, "loss": 3.5961971282958984, "grad_norm": 0.7015663981437683, "learning_rate": 0.0001, "elbo": 3.5962, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8973140980530143}
{"step": 49410, "loss": 3.5829586029052733, "grad_norm": 0.7168616056442261, "learning_rate": 0.0001, "elbo": 3.58296, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8976073187895848}
{"step": 49415, "loss": 3.589308166503906, "grad_norm": 0.7758499979972839, "learning_rate": 0.0001, "elbo": 3.58931, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8979005395261552}
{"step": 49420, "loss": 3.6016571044921877, "grad_norm": 0.7960436940193176, "learning_rate": 0.0001, "elbo": 3.60166, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8981937602627257}
{"step": 49425, "loss": 3.5961688995361327, "grad_norm": 0.6669138073921204, "learning_rate": 0.0001, "elbo": 3.59617, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.898486980999296}
{"step": 49430, "loss": 3.590374755859375, "grad_norm": 0.7018357515335083, "learning_rate": 0.0001, "elbo": 3.59037, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8987802017358666}
{"step": 49435, "loss": 3.596828079223633, "grad_norm": 0.7053924202919006, "learning_rate": 0.0001, "elbo": 3.59683, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.899073422472437}
{"step": 49440, "loss": 3.574614715576172, "grad_norm": 0.7183050513267517, "learning_rate": 0.0001, "elbo": 3.57462, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8993666432090075}
{"step": 49445, "loss": 3.598074722290039, "grad_norm": 0.7127779126167297, "learning_rate": 0.0001, "elbo": 3.59807, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.8996598639455784}
{"step": 49450, "loss": 3.584192657470703, "grad_norm": 0.6985769867897034, "learning_rate": 0.0001, "elbo": 3.58419, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.899953084682149}
{"step": 49455, "loss": 3.607738494873047, "grad_norm": 0.8047619462013245, "learning_rate": 0.0001, "elbo": 3.60774, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9002463054187193}
{"step": 49460, "loss": 3.5915447235107423, "grad_norm": 0.7527608275413513, "learning_rate": 0.0001, "elbo": 3.59154, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9005395261552898}
{"step": 49460, "loss": 3.5915447235107423, "grad_norm": 0.7527608275413513, "learning_rate": 0.0001, "elbo": 3.59154, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9005395261552898}
{"step": 49465, "loss": 3.589788055419922, "grad_norm": 0.6794106960296631, "learning_rate": 0.0001, "elbo": 3.58979, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.90083274689186}
{"step": 49470, "loss": 3.570670318603516, "grad_norm": 0.6938847303390503, "learning_rate": 0.0001, "elbo": 3.57067, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9011259676284307}
{"step": 49475, "loss": 3.5986671447753906, "grad_norm": 0.6688828468322754, "learning_rate": 0.0001, "elbo": 3.59867, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.901419188365001}
{"step": 49480, "loss": 3.5814254760742186, "grad_norm": 0.7734557390213013, "learning_rate": 0.0001, "elbo": 3.58143, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9017124091015716}
{"step": 49485, "loss": 3.5823486328125, "grad_norm": 0.7900524735450745, "learning_rate": 0.0001, "elbo": 3.58235, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.902005629838142}
{"step": 49490, "loss": 3.60220947265625, "grad_norm": 0.7134917974472046, "learning_rate": 0.0001, "elbo": 3.60221, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9022988505747125}
{"step": 49490, "loss": 3.60220947265625, "grad_norm": 0.7134917974472046, "learning_rate": 0.0001, "elbo": 3.60221, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9022988505747125}
{"step": 49495, "loss": 3.574076461791992, "grad_norm": 0.6975343227386475, "learning_rate": 0.0001, "elbo": 3.57408, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9025920713112834}
{"step": 49500, "loss": 3.6023017883300783, "grad_norm": 0.6825235486030579, "learning_rate": 0.0001, "elbo": 3.6023, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.902885292047854}
{"step": 49505, "loss": 3.585844802856445, "grad_norm": 0.6927343010902405, "learning_rate": 0.0001, "elbo": 3.58584, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9031785127844243}
{"step": 49505, "loss": 3.585844802856445, "grad_norm": 0.6927343010902405, "learning_rate": 0.0001, "elbo": 3.58584, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9031785127844243}
{"step": 49510, "loss": 3.5965919494628906, "grad_norm": 0.7158903479576111, "learning_rate": 0.0001, "elbo": 3.59659, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9034717335209947}
{"step": 49515, "loss": 3.597673034667969, "grad_norm": 0.7375440001487732, "learning_rate": 0.0001, "elbo": 3.59767, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.903764954257565}
{"step": 49520, "loss": 3.568475341796875, "grad_norm": 0.6800523996353149, "learning_rate": 0.0001, "elbo": 3.56848, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9040581749941357}
{"step": 49525, "loss": 3.5805171966552733, "grad_norm": 0.7895728349685669, "learning_rate": 0.0001, "elbo": 3.58052, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.904351395730706}
{"step": 49530, "loss": 3.582989501953125, "grad_norm": 0.7201588749885559, "learning_rate": 0.0001, "elbo": 3.58299, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9046446164672766}
{"step": 49535, "loss": 3.57403450012207, "grad_norm": 0.6856899857521057, "learning_rate": 0.0001, "elbo": 3.57403, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.904937837203847}
{"step": 49540, "loss": 3.600864028930664, "grad_norm": 0.7538641095161438, "learning_rate": 0.0001, "elbo": 3.60086, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9052310579404175}
{"step": 49545, "loss": 3.58399658203125, "grad_norm": 0.7292651534080505, "learning_rate": 0.0001, "elbo": 3.584, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.905524278676988}
{"step": 49550, "loss": 3.593781280517578, "grad_norm": 0.6916899681091309, "learning_rate": 0.0001, "elbo": 3.59378, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9058174994135584}
{"step": 49555, "loss": 3.58099250793457, "grad_norm": 0.6829319596290588, "learning_rate": 0.0001, "elbo": 3.58099, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.906110720150129}
{"step": 49560, "loss": 3.5947586059570313, "grad_norm": 0.7302505970001221, "learning_rate": 0.0001, "elbo": 3.59476, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9064039408866993}
{"step": 49565, "loss": 3.589760589599609, "grad_norm": 0.6886103749275208, "learning_rate": 0.0001, "elbo": 3.58976, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9066971616232697}
{"step": 49570, "loss": 3.5801536560058596, "grad_norm": 0.7145403027534485, "learning_rate": 0.0001, "elbo": 3.58015, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9069903823598406}
{"step": 49570, "loss": 3.5801536560058596, "grad_norm": 0.7145403027534485, "learning_rate": 0.0001, "elbo": 3.58015, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9069903823598406}
{"step": 49575, "loss": 3.5879783630371094, "grad_norm": 0.6939472556114197, "learning_rate": 0.0001, "elbo": 3.58798, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.907283603096411}
{"step": 49580, "loss": 3.5879798889160157, "grad_norm": 0.7420389652252197, "learning_rate": 0.0001, "elbo": 3.58798, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9075768238329815}
{"step": 49585, "loss": 3.573897933959961, "grad_norm": 0.7525015473365784, "learning_rate": 0.0001, "elbo": 3.5739, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.907870044569552}
{"step": 49590, "loss": 3.57977180480957, "grad_norm": 0.6856593489646912, "learning_rate": 0.0001, "elbo": 3.57977, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9081632653061225}
{"step": 49595, "loss": 3.5873878479003904, "grad_norm": 0.66629558801651, "learning_rate": 0.0001, "elbo": 3.58739, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.908456486042693}
{"step": 49600, "loss": 3.581105041503906, "grad_norm": 0.7113918662071228, "learning_rate": 0.0001, "elbo": 3.58111, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9087497067792634}
{"step": 49605, "loss": 3.605716323852539, "grad_norm": 0.7274097204208374, "learning_rate": 0.0001, "elbo": 3.60572, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.909042927515834}
{"step": 49610, "loss": 3.596274185180664, "grad_norm": 0.7140865921974182, "learning_rate": 0.0001, "elbo": 3.59627, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9093361482524043}
{"step": 49615, "loss": 3.588610076904297, "grad_norm": 0.7294743657112122, "learning_rate": 0.0001, "elbo": 3.58861, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.909629368988975}
{"step": 49620, "loss": 3.580397033691406, "grad_norm": 0.664242684841156, "learning_rate": 0.0001, "elbo": 3.5804, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9099225897255456}
{"step": 49625, "loss": 3.603059005737305, "grad_norm": 0.6767380833625793, "learning_rate": 0.0001, "elbo": 3.60306, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.910215810462116}
{"step": 49630, "loss": 3.5874019622802735, "grad_norm": 0.7461788654327393, "learning_rate": 0.0001, "elbo": 3.5874, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9105090311986865}
{"step": 49635, "loss": 3.581861877441406, "grad_norm": 1.331372618675232, "learning_rate": 0.0001, "elbo": 3.58186, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.910802251935257}
{"step": 49640, "loss": 3.602039337158203, "grad_norm": 0.6569902896881104, "learning_rate": 0.0001, "elbo": 3.60204, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9110954726718274}
{"step": 49645, "loss": 3.5893665313720704, "grad_norm": 2.2858049869537354, "learning_rate": 0.0001, "elbo": 3.58937, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.911388693408398}
{"step": 49650, "loss": 3.602756118774414, "grad_norm": 0.6588453054428101, "learning_rate": 0.0001, "elbo": 3.60276, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9116819141449684}
{"step": 49655, "loss": 3.5691246032714843, "grad_norm": 0.6906048655509949, "learning_rate": 0.0001, "elbo": 3.56912, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.911975134881539}
{"step": 49660, "loss": 3.5910980224609377, "grad_norm": 0.6631866693496704, "learning_rate": 0.0001, "elbo": 3.5911, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9122683556181093}
{"step": 49665, "loss": 3.587493133544922, "grad_norm": 0.7780762910842896, "learning_rate": 0.0001, "elbo": 3.58749, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9125615763546797}
{"step": 49670, "loss": 3.5886772155761717, "grad_norm": 0.6525934934616089, "learning_rate": 0.0001, "elbo": 3.58868, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.91285479709125}
{"step": 49675, "loss": 3.5743877410888674, "grad_norm": 0.6808844208717346, "learning_rate": 0.0001, "elbo": 3.57439, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9131480178278206}
{"step": 49680, "loss": 3.5935283660888673, "grad_norm": 0.6798952221870422, "learning_rate": 0.0001, "elbo": 3.59353, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.913441238564391}
{"step": 49685, "loss": 3.5912036895751953, "grad_norm": 0.7533872127532959, "learning_rate": 0.0001, "elbo": 3.5912, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9137344593009615}
{"step": 49690, "loss": 3.599706268310547, "grad_norm": 0.7453209757804871, "learning_rate": 0.0001, "elbo": 3.59971, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.914027680037532}
{"step": 49695, "loss": 3.576811599731445, "grad_norm": 0.6867853403091431, "learning_rate": 0.0001, "elbo": 3.57681, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.914320900774103}
{"step": 49700, "loss": 3.584381103515625, "grad_norm": 0.7496123313903809, "learning_rate": 0.0001, "elbo": 3.58438, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9146141215106733}
{"step": 49705, "loss": 3.5869400024414064, "grad_norm": 0.6954942941665649, "learning_rate": 0.0001, "elbo": 3.58694, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.914907342247244}
{"step": 49710, "loss": 3.5921791076660154, "grad_norm": 0.7043737769126892, "learning_rate": 0.0001, "elbo": 3.59218, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9152005629838142}
{"step": 49715, "loss": 3.5841796875, "grad_norm": 0.657497763633728, "learning_rate": 0.0001, "elbo": 3.58418, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9154937837203847}
{"step": 49715, "loss": 3.5841796875, "grad_norm": 0.657497763633728, "learning_rate": 0.0001, "elbo": 3.58418, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9154937837203847}
{"step": 49720, "loss": 3.58369140625, "grad_norm": 0.6816378235816956, "learning_rate": 0.0001, "elbo": 3.58369, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.915787004456955}
{"step": 49725, "loss": 3.576548767089844, "grad_norm": 0.6827966570854187, "learning_rate": 0.0001, "elbo": 3.57655, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9160802251935256}
{"step": 49730, "loss": 3.587015151977539, "grad_norm": 0.7336581349372864, "learning_rate": 0.0001, "elbo": 3.58702, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.916373445930096}
{"step": 49735, "loss": 3.590248870849609, "grad_norm": 0.6542733907699585, "learning_rate": 0.0001, "elbo": 3.59025, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9166666666666665}
{"step": 49740, "loss": 3.5988090515136717, "grad_norm": 0.7433124780654907, "learning_rate": 0.0001, "elbo": 3.59881, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9169598874032374}
{"step": 49745, "loss": 3.5803810119628907, "grad_norm": 0.6893187165260315, "learning_rate": 0.0001, "elbo": 3.58038, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.917253108139808}
{"step": 49745, "loss": 3.5803810119628907, "grad_norm": 0.6893187165260315, "learning_rate": 0.0001, "elbo": 3.58038, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.917253108139808}
{"step": 49750, "loss": 3.5816791534423826, "grad_norm": 0.740805447101593, "learning_rate": 0.0001, "elbo": 3.58168, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9175463288763783}
{"step": 49755, "loss": 3.5893276214599608, "grad_norm": 0.9159746766090393, "learning_rate": 0.0001, "elbo": 3.58933, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.917839549612949}
{"step": 49760, "loss": 3.568918228149414, "grad_norm": 0.6663850545883179, "learning_rate": 0.0001, "elbo": 3.56892, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9181327703495192}
{"step": 49765, "loss": 3.596186065673828, "grad_norm": 0.678736686706543, "learning_rate": 0.0001, "elbo": 3.59619, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9184259910860897}
{"step": 49770, "loss": 3.5692447662353515, "grad_norm": 0.6911392211914062, "learning_rate": 0.0001, "elbo": 3.56924, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.91871921182266}
{"step": 49775, "loss": 3.569336700439453, "grad_norm": 0.7095723152160645, "learning_rate": 0.0001, "elbo": 3.56934, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9190124325592306}
{"step": 49780, "loss": 3.588764190673828, "grad_norm": 0.6542023420333862, "learning_rate": 0.0001, "elbo": 3.58876, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.919305653295801}
{"step": 49785, "loss": 3.5829498291015627, "grad_norm": 0.6791253685951233, "learning_rate": 0.0001, "elbo": 3.58295, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9195988740323715}
{"step": 49790, "loss": 3.592772674560547, "grad_norm": 0.6929115056991577, "learning_rate": 0.0001, "elbo": 3.59277, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.919892094768942}
{"step": 49795, "loss": 3.599051666259766, "grad_norm": 0.6868211030960083, "learning_rate": 0.0001, "elbo": 3.59905, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9201853155055124}
{"step": 49800, "loss": 3.5933074951171875, "grad_norm": 0.747421383857727, "learning_rate": 0.0001, "elbo": 3.59331, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.920478536242083}
{"step": 49805, "loss": 3.5731956481933596, "grad_norm": 0.7522946000099182, "learning_rate": 0.0001, "elbo": 3.5732, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9207717569786533}
{"step": 49810, "loss": 3.5851211547851562, "grad_norm": 0.7176806330680847, "learning_rate": 0.0001, "elbo": 3.58512, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9210649777152238}
{"step": 49815, "loss": 3.570515441894531, "grad_norm": 0.7292149662971497, "learning_rate": 0.0001, "elbo": 3.57052, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9213581984517947}
{"step": 49820, "loss": 3.5794906616210938, "grad_norm": 0.6636384129524231, "learning_rate": 0.0001, "elbo": 3.57949, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.921651419188365}
{"step": 49825, "loss": 3.5786361694335938, "grad_norm": 0.760677695274353, "learning_rate": 0.0001, "elbo": 3.57864, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9219446399249356}
{"step": 49830, "loss": 3.579241180419922, "grad_norm": 0.6896827220916748, "learning_rate": 0.0001, "elbo": 3.57924, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.922237860661506}
{"step": 49835, "loss": 3.600149917602539, "grad_norm": 0.7070871591567993, "learning_rate": 0.0001, "elbo": 3.60015, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9225310813980765}
{"step": 49840, "loss": 3.5924686431884765, "grad_norm": 0.7221634387969971, "learning_rate": 0.0001, "elbo": 3.59247, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.922824302134647}
{"step": 49845, "loss": 3.5848243713378904, "grad_norm": 0.6906956434249878, "learning_rate": 0.0001, "elbo": 3.58482, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9231175228712174}
{"step": 49850, "loss": 3.5797092437744142, "grad_norm": 0.6844262480735779, "learning_rate": 0.0001, "elbo": 3.57971, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.923410743607788}
{"step": 49855, "loss": 3.5770164489746095, "grad_norm": 0.6374250650405884, "learning_rate": 0.0001, "elbo": 3.57702, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9237039643443583}
{"step": 49860, "loss": 3.5789215087890627, "grad_norm": 0.6900373697280884, "learning_rate": 0.0001, "elbo": 3.57892, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9239971850809288}
{"step": 49865, "loss": 3.5673797607421873, "grad_norm": 0.751578152179718, "learning_rate": 0.0001, "elbo": 3.56738, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9242904058174997}
{"step": 49870, "loss": 3.575028991699219, "grad_norm": 0.745410680770874, "learning_rate": 0.0001, "elbo": 3.57503, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.92458362655407}
{"step": 49875, "loss": 3.584406280517578, "grad_norm": 0.6866325736045837, "learning_rate": 0.0001, "elbo": 3.58441, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9248768472906406}
{"step": 49880, "loss": 3.5560287475585937, "grad_norm": 0.6784018874168396, "learning_rate": 0.0001, "elbo": 3.55603, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.925170068027211}
{"step": 49885, "loss": 3.578050231933594, "grad_norm": 0.6487820148468018, "learning_rate": 0.0001, "elbo": 3.57805, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9254632887637815}
{"step": 49890, "loss": 3.5804519653320312, "grad_norm": 0.7876332402229309, "learning_rate": 0.0001, "elbo": 3.58045, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.925756509500352}
{"step": 49895, "loss": 3.5861785888671873, "grad_norm": 0.7560317516326904, "learning_rate": 0.0001, "elbo": 3.58618, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9260497302369224}
{"step": 49900, "loss": 3.565989685058594, "grad_norm": 0.7160230278968811, "learning_rate": 0.0001, "elbo": 3.56599, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.926342950973493}
{"step": 49905, "loss": 3.5671546936035154, "grad_norm": 0.6557715535163879, "learning_rate": 0.0001, "elbo": 3.56715, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9266361717100633}
{"step": 49910, "loss": 3.583618927001953, "grad_norm": 0.7179473042488098, "learning_rate": 0.0001, "elbo": 3.58362, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9269293924466337}
{"step": 49915, "loss": 3.5704002380371094, "grad_norm": 0.6527343988418579, "learning_rate": 0.0001, "elbo": 3.5704, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.927222613183204}
{"step": 49915, "loss": 3.5704002380371094, "grad_norm": 0.6527343988418579, "learning_rate": 0.0001, "elbo": 3.5704, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.927222613183204}
{"step": 49920, "loss": 3.589163589477539, "grad_norm": 1.315139889717102, "learning_rate": 0.0001, "elbo": 3.58916, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9275158339197747}
{"step": 49925, "loss": 3.5862335205078124, "grad_norm": 0.655911922454834, "learning_rate": 0.0001, "elbo": 3.58623, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.927809054656345}
{"step": 49930, "loss": 3.575365447998047, "grad_norm": 0.7067993879318237, "learning_rate": 0.0001, "elbo": 3.57537, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9281022753929156}
{"step": 49935, "loss": 3.5995437622070314, "grad_norm": 0.8246315717697144, "learning_rate": 0.0001, "elbo": 3.59954, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.928395496129486}
{"step": 49940, "loss": 3.603184127807617, "grad_norm": 0.8566279411315918, "learning_rate": 0.0001, "elbo": 3.60318, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.928688716866057}
{"step": 49945, "loss": 3.5754714965820313, "grad_norm": 0.6751553416252136, "learning_rate": 0.0001, "elbo": 3.57547, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9289819376026274}
{"step": 49950, "loss": 3.5731185913085937, "grad_norm": 0.8818421363830566, "learning_rate": 0.0001, "elbo": 3.57312, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.929275158339198}
{"step": 49955, "loss": 3.589333343505859, "grad_norm": 0.704856812953949, "learning_rate": 0.0001, "elbo": 3.58933, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9295683790757683}
{"step": 49960, "loss": 3.589387130737305, "grad_norm": 0.6850396990776062, "learning_rate": 0.0001, "elbo": 3.58939, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9298615998123387}
{"step": 49965, "loss": 3.5841819763183596, "grad_norm": 0.8873084187507629, "learning_rate": 0.0001, "elbo": 3.58418, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.930154820548909}
{"step": 49970, "loss": 3.5990028381347656, "grad_norm": 0.7380296587944031, "learning_rate": 0.0001, "elbo": 3.599, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9304480412854796}
{"step": 49975, "loss": 3.588251495361328, "grad_norm": 0.648920476436615, "learning_rate": 0.0001, "elbo": 3.58825, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.93074126202205}
{"step": 49975, "loss": 3.588251495361328, "grad_norm": 0.648920476436615, "learning_rate": 0.0001, "elbo": 3.58825, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.93074126202205}
{"step": 49980, "loss": 3.5869964599609374, "grad_norm": 0.755450963973999, "learning_rate": 0.0001, "elbo": 3.587, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9310344827586206}
{"step": 49980, "loss": 3.5869964599609374, "grad_norm": 0.755450963973999, "learning_rate": 0.0001, "elbo": 3.587, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9310344827586206}
{"step": 49985, "loss": 3.592346954345703, "grad_norm": 1.064133882522583, "learning_rate": 0.0001, "elbo": 3.59235, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9313277034951915}
{"step": 49990, "loss": 3.585245895385742, "grad_norm": 0.7000899910926819, "learning_rate": 0.0001, "elbo": 3.58525, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.931620924231762}
{"step": 49995, "loss": 3.5677955627441404, "grad_norm": 0.8010244965553284, "learning_rate": 0.0001, "elbo": 3.5678, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9319141449683324}
{"step": 50000, "loss": 3.5934207916259764, "grad_norm": 0.680688738822937, "learning_rate": 0.0001, "elbo": 3.59342, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.932207365704903}
{"step": 50005, "loss": 3.568254852294922, "grad_norm": 0.702635645866394, "learning_rate": 0.0001, "elbo": 3.56826, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9325005864414733}
{"step": 50010, "loss": 3.5888465881347655, "grad_norm": 0.6849567294120789, "learning_rate": 0.0001, "elbo": 3.58885, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9327938071780437}
{"step": 50015, "loss": 3.600190353393555, "grad_norm": 0.8053436279296875, "learning_rate": 0.0001, "elbo": 3.60019, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.933087027914614}
{"step": 50020, "loss": 3.572833251953125, "grad_norm": 0.7238377332687378, "learning_rate": 0.0001, "elbo": 3.57283, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9333802486511846}
{"step": 50025, "loss": 3.57437744140625, "grad_norm": 0.6542567014694214, "learning_rate": 0.0001, "elbo": 3.57438, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.933673469387755}
{"step": 50030, "loss": 3.5909584045410154, "grad_norm": 0.8828826546669006, "learning_rate": 0.0001, "elbo": 3.59096, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9339666901243255}
{"step": 50035, "loss": 3.5983131408691404, "grad_norm": 0.7114978432655334, "learning_rate": 0.0001, "elbo": 3.59831, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.934259910860896}
{"step": 50040, "loss": 3.5935184478759767, "grad_norm": 0.6975641846656799, "learning_rate": 0.0001, "elbo": 3.59352, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9345531315974664}
{"step": 50040, "loss": 3.5935184478759767, "grad_norm": 0.6975641846656799, "learning_rate": 0.0001, "elbo": 3.59352, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9345531315974664}
{"step": 50045, "loss": 3.5732681274414064, "grad_norm": 0.7002948522567749, "learning_rate": 0.0001, "elbo": 3.57327, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.934846352334037}
{"step": 50050, "loss": 3.5751670837402343, "grad_norm": 0.6641857028007507, "learning_rate": 0.0001, "elbo": 3.57517, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9351395730706074}
{"step": 50055, "loss": 3.5831764221191404, "grad_norm": 0.8029687404632568, "learning_rate": 0.0001, "elbo": 3.58318, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.935432793807178}
{"step": 50060, "loss": 3.5643638610839843, "grad_norm": 0.6765531897544861, "learning_rate": 0.0001, "elbo": 3.56436, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9357260145437483}
{"step": 50065, "loss": 3.578084182739258, "grad_norm": 0.7601343393325806, "learning_rate": 0.0001, "elbo": 3.57808, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.936019235280319}
{"step": 50070, "loss": 3.5786155700683593, "grad_norm": 0.7350319027900696, "learning_rate": 0.0001, "elbo": 3.57862, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9363124560168896}
{"step": 50075, "loss": 3.5935935974121094, "grad_norm": 0.6667455434799194, "learning_rate": 0.0001, "elbo": 3.59359, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.93660567675346}
{"step": 50080, "loss": 3.5935291290283202, "grad_norm": 0.7218686938285828, "learning_rate": 0.0001, "elbo": 3.59353, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9368988974900305}
{"step": 50085, "loss": 3.585678482055664, "grad_norm": 0.7353179454803467, "learning_rate": 0.0001, "elbo": 3.58568, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.937192118226601}
{"step": 50090, "loss": 3.5904266357421877, "grad_norm": 0.7722300887107849, "learning_rate": 0.0001, "elbo": 3.59043, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9374853389631714}
{"step": 50095, "loss": 3.5784812927246095, "grad_norm": 0.7267804145812988, "learning_rate": 0.0001, "elbo": 3.57848, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.937778559699742}
{"step": 50100, "loss": 3.5912425994873045, "grad_norm": 0.7294025421142578, "learning_rate": 0.0001, "elbo": 3.59124, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9380717804363123}
{"step": 50105, "loss": 3.591798782348633, "grad_norm": 0.7464616298675537, "learning_rate": 0.0001, "elbo": 3.5918, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.938365001172883}
{"step": 50110, "loss": 3.585308074951172, "grad_norm": 0.6944562196731567, "learning_rate": 0.0001, "elbo": 3.58531, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9386582219094537}
{"step": 50115, "loss": 3.589405059814453, "grad_norm": 0.652025043964386, "learning_rate": 0.0001, "elbo": 3.58941, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.938951442646024}
{"step": 50120, "loss": 3.570922088623047, "grad_norm": 0.6739916801452637, "learning_rate": 0.0001, "elbo": 3.57092, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9392446633825946}
{"step": 50125, "loss": 3.5613903045654296, "grad_norm": 0.6843323111534119, "learning_rate": 0.0001, "elbo": 3.56139, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.939537884119165}
{"step": 50130, "loss": 3.578899383544922, "grad_norm": 0.7117111682891846, "learning_rate": 0.0001, "elbo": 3.5789, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9398311048557355}
{"step": 50135, "loss": 3.5774253845214843, "grad_norm": 0.7726682424545288, "learning_rate": 0.0001, "elbo": 3.57743, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.940124325592306}
{"step": 50140, "loss": 3.578082275390625, "grad_norm": 0.6495299339294434, "learning_rate": 0.0001, "elbo": 3.57808, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9404175463288764}
{"step": 50145, "loss": 3.6033370971679686, "grad_norm": 0.6596904397010803, "learning_rate": 0.0001, "elbo": 3.60334, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.940710767065447}
{"step": 50150, "loss": 3.5742733001708986, "grad_norm": 0.682072103023529, "learning_rate": 0.0001, "elbo": 3.57427, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9410039878020173}
{"step": 50150, "loss": 3.5742733001708986, "grad_norm": 0.682072103023529, "learning_rate": 0.0001, "elbo": 3.57427, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9410039878020173}
{"step": 50155, "loss": 3.580152893066406, "grad_norm": 0.7177481055259705, "learning_rate": 0.0001, "elbo": 3.58015, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.941297208538588}
{"step": 50160, "loss": 3.5765762329101562, "grad_norm": 0.7185811400413513, "learning_rate": 0.0001, "elbo": 3.57658, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9415904292751582}
{"step": 50165, "loss": 3.5745079040527346, "grad_norm": 0.726078987121582, "learning_rate": 0.0001, "elbo": 3.57451, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9418836500117287}
{"step": 50170, "loss": 3.5834335327148437, "grad_norm": 0.6581079959869385, "learning_rate": 0.0001, "elbo": 3.58343, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.942176870748299}
{"step": 50170, "loss": 3.5834335327148437, "grad_norm": 0.6581079959869385, "learning_rate": 0.0001, "elbo": 3.58343, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.942176870748299}
{"step": 50175, "loss": 3.578790283203125, "grad_norm": 0.678937554359436, "learning_rate": 0.0001, "elbo": 3.57879, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9424700914848696}
{"step": 50180, "loss": 3.5831443786621096, "grad_norm": 0.6801151037216187, "learning_rate": 0.0001, "elbo": 3.58314, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.94276331222144}
{"step": 50185, "loss": 3.5782413482666016, "grad_norm": 0.745627224445343, "learning_rate": 0.0001, "elbo": 3.57824, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.943056532958011}
{"step": 50190, "loss": 3.586741638183594, "grad_norm": 0.6858232617378235, "learning_rate": 0.0001, "elbo": 3.58674, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9433497536945814}
{"step": 50195, "loss": 3.5783939361572266, "grad_norm": 0.7163298726081848, "learning_rate": 0.0001, "elbo": 3.57839, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.943642974431152}
{"step": 50200, "loss": 3.5751651763916015, "grad_norm": 0.6602131724357605, "learning_rate": 0.0001, "elbo": 3.57517, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9439361951677223}
{"step": 50205, "loss": 3.571624755859375, "grad_norm": 0.6642425060272217, "learning_rate": 0.0001, "elbo": 3.57162, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9442294159042928}
{"step": 50210, "loss": 3.5997573852539064, "grad_norm": 0.651613175868988, "learning_rate": 0.0001, "elbo": 3.59976, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9445226366408632}
{"step": 50215, "loss": 3.582772445678711, "grad_norm": 0.7263335585594177, "learning_rate": 0.0001, "elbo": 3.58277, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9448158573774337}
{"step": 50220, "loss": 3.5811370849609374, "grad_norm": 0.8501487970352173, "learning_rate": 0.0001, "elbo": 3.58114, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.945109078114004}
{"step": 50225, "loss": 3.5945144653320313, "grad_norm": 0.6749985218048096, "learning_rate": 0.0001, "elbo": 3.59451, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9454022988505746}
{"step": 50230, "loss": 3.602276611328125, "grad_norm": 0.7037060856819153, "learning_rate": 0.0001, "elbo": 3.60228, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.945695519587145}
{"step": 50235, "loss": 3.5817909240722656, "grad_norm": 0.7032801508903503, "learning_rate": 0.0001, "elbo": 3.58179, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.945988740323716}
{"step": 50240, "loss": 3.5966022491455076, "grad_norm": 0.670244574546814, "learning_rate": 0.0001, "elbo": 3.5966, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9462819610602864}
{"step": 50245, "loss": 3.577281951904297, "grad_norm": 0.7054886221885681, "learning_rate": 0.0001, "elbo": 3.57728, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.946575181796857}
{"step": 50250, "loss": 3.583591079711914, "grad_norm": 0.9367983341217041, "learning_rate": 0.0001, "elbo": 3.58359, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9468684025334273}
{"step": 50255, "loss": 3.5745567321777343, "grad_norm": 0.7143924832344055, "learning_rate": 0.0001, "elbo": 3.57456, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9471616232699978}
{"step": 50260, "loss": 3.577750396728516, "grad_norm": 0.7107589840888977, "learning_rate": 0.0001, "elbo": 3.57775, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.947454844006568}
{"step": 50265, "loss": 3.5876121520996094, "grad_norm": 0.7095639109611511, "learning_rate": 0.0001, "elbo": 3.58761, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9477480647431387}
{"step": 50270, "loss": 3.5889244079589844, "grad_norm": 0.691180944442749, "learning_rate": 0.0001, "elbo": 3.58892, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.948041285479709}
{"step": 50275, "loss": 3.586019515991211, "grad_norm": 1.0354325771331787, "learning_rate": 0.0001, "elbo": 3.58602, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9483345062162796}
{"step": 50280, "loss": 3.57910270690918, "grad_norm": 0.671675980091095, "learning_rate": 0.0001, "elbo": 3.5791, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.94862772695285}
{"step": 50285, "loss": 3.5922836303710937, "grad_norm": 0.6810869574546814, "learning_rate": 0.0001, "elbo": 3.59228, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9489209476894205}
{"step": 50290, "loss": 3.5963916778564453, "grad_norm": 0.6474206447601318, "learning_rate": 0.0001, "elbo": 3.59639, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.949214168425991}
{"step": 50295, "loss": 3.5798583984375, "grad_norm": 0.6679586172103882, "learning_rate": 0.0001, "elbo": 3.57986, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9495073891625614}
{"step": 50300, "loss": 3.571818161010742, "grad_norm": 0.7009127736091614, "learning_rate": 0.0001, "elbo": 3.57182, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.949800609899132}
{"step": 50305, "loss": 3.5791259765625, "grad_norm": 0.6617889404296875, "learning_rate": 0.0001, "elbo": 3.57913, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9500938306357023}
{"step": 50310, "loss": 3.570591354370117, "grad_norm": 0.7085382342338562, "learning_rate": 0.0001, "elbo": 3.57059, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.950387051372273}
{"step": 50315, "loss": 3.5903160095214846, "grad_norm": 0.7633902430534363, "learning_rate": 0.0001, "elbo": 3.59032, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9506802721088436}
{"step": 50320, "loss": 3.5789020538330076, "grad_norm": 0.7612540125846863, "learning_rate": 0.0001, "elbo": 3.5789, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.950973492845414}
{"step": 50325, "loss": 3.584485626220703, "grad_norm": 0.7262130975723267, "learning_rate": 0.0001, "elbo": 3.58449, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9512667135819846}
{"step": 50330, "loss": 3.590415191650391, "grad_norm": 0.7618200778961182, "learning_rate": 0.0001, "elbo": 3.59042, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.951559934318555}
{"step": 50335, "loss": 3.5769630432128907, "grad_norm": 0.7385280132293701, "learning_rate": 0.0001, "elbo": 3.57696, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9518531550551255}
{"step": 50340, "loss": 3.5825096130371095, "grad_norm": 0.8051142692565918, "learning_rate": 0.0001, "elbo": 3.58251, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.952146375791696}
{"step": 50345, "loss": 3.5773162841796875, "grad_norm": 0.6953350305557251, "learning_rate": 0.0001, "elbo": 3.57732, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9524395965282664}
{"step": 50350, "loss": 3.579730987548828, "grad_norm": 0.7133830189704895, "learning_rate": 0.0001, "elbo": 3.57973, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.952732817264837}
{"step": 50355, "loss": 3.59561767578125, "grad_norm": 0.7025686502456665, "learning_rate": 0.0001, "elbo": 3.59562, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9530260380014077}
{"step": 50360, "loss": 3.5852951049804687, "grad_norm": 0.6397512555122375, "learning_rate": 0.0001, "elbo": 3.5853, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.953319258737978}
{"step": 50365, "loss": 3.615798568725586, "grad_norm": 0.6989755034446716, "learning_rate": 0.0001, "elbo": 3.6158, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9536124794745486}
{"step": 50370, "loss": 3.5928157806396483, "grad_norm": 1.1766468286514282, "learning_rate": 0.0001, "elbo": 3.59282, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.953905700211119}
{"step": 50375, "loss": 3.5819976806640623, "grad_norm": 0.676311194896698, "learning_rate": 0.0001, "elbo": 3.582, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9541989209476895}
{"step": 50380, "loss": 3.5790897369384767, "grad_norm": 0.8504121899604797, "learning_rate": 0.0001, "elbo": 3.57909, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.95449214168426}
{"step": 50385, "loss": 3.566310501098633, "grad_norm": 0.7259839177131653, "learning_rate": 0.0001, "elbo": 3.56631, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9547853624208305}
{"step": 50390, "loss": 3.5942054748535157, "grad_norm": 0.6444743871688843, "learning_rate": 0.0001, "elbo": 3.59421, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.955078583157401}
{"step": 50395, "loss": 3.6011539459228517, "grad_norm": 0.7258883118629456, "learning_rate": 0.0001, "elbo": 3.60115, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9553718038939714}
{"step": 50400, "loss": 3.5859756469726562, "grad_norm": 1.0021629333496094, "learning_rate": 0.0001, "elbo": 3.58598, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.955665024630542}
{"step": 50405, "loss": 3.5845718383789062, "grad_norm": 0.7366008162498474, "learning_rate": 0.0001, "elbo": 3.58457, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9559582453671123}
{"step": 50410, "loss": 3.5954002380371093, "grad_norm": 0.6691750884056091, "learning_rate": 0.0001, "elbo": 3.5954, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9562514661036827}
{"step": 50415, "loss": 3.5814056396484375, "grad_norm": 0.7440071105957031, "learning_rate": 0.0001, "elbo": 3.58141, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.956544686840253}
{"step": 50420, "loss": 3.580452728271484, "grad_norm": 0.7080726623535156, "learning_rate": 0.0001, "elbo": 3.58045, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9568379075768236}
{"step": 50425, "loss": 3.5945175170898436, "grad_norm": 0.7249830365180969, "learning_rate": 0.0001, "elbo": 3.59452, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.957131128313394}
{"step": 50430, "loss": 3.595048522949219, "grad_norm": 0.6957542300224304, "learning_rate": 0.0001, "elbo": 3.59505, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9574243490499645}
{"step": 50435, "loss": 3.562870407104492, "grad_norm": 0.8072821497917175, "learning_rate": 0.0001, "elbo": 3.56287, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9577175697865354}
{"step": 50440, "loss": 3.5882308959960936, "grad_norm": 0.7716082334518433, "learning_rate": 0.0001, "elbo": 3.58823, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.958010790523106}
{"step": 50445, "loss": 3.589621734619141, "grad_norm": 0.7131969332695007, "learning_rate": 0.0001, "elbo": 3.58962, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9583040112596763}
{"step": 50445, "loss": 3.589621734619141, "grad_norm": 0.7131969332695007, "learning_rate": 0.0001, "elbo": 3.58962, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9583040112596763}
{"step": 50450, "loss": 3.577406311035156, "grad_norm": 1.1591072082519531, "learning_rate": 0.0001, "elbo": 3.57741, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.958597231996247}
{"step": 50455, "loss": 3.586151885986328, "grad_norm": 0.7227977514266968, "learning_rate": 0.0001, "elbo": 3.58615, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9588904527328173}
{"step": 50460, "loss": 3.573649215698242, "grad_norm": 0.6868823170661926, "learning_rate": 0.0001, "elbo": 3.57365, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9591836734693877}
{"step": 50465, "loss": 3.574419403076172, "grad_norm": 0.6654892563819885, "learning_rate": 0.0001, "elbo": 3.57442, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.959476894205958}
{"step": 50470, "loss": 3.6003646850585938, "grad_norm": 0.7771966457366943, "learning_rate": 0.0001, "elbo": 3.60036, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9597701149425286}
{"step": 50475, "loss": 3.5809974670410156, "grad_norm": 0.716752290725708, "learning_rate": 0.0001, "elbo": 3.581, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.960063335679099}
{"step": 50480, "loss": 3.5967498779296876, "grad_norm": 0.6648720502853394, "learning_rate": 0.0001, "elbo": 3.59675, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.96035655641567}
{"step": 50485, "loss": 3.5826244354248047, "grad_norm": 0.7164426445960999, "learning_rate": 0.0001, "elbo": 3.58262, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9606497771522404}
{"step": 50490, "loss": 3.5731163024902344, "grad_norm": 0.7432205080986023, "learning_rate": 0.0001, "elbo": 3.57312, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.960942997888811}
{"step": 50495, "loss": 3.5808231353759767, "grad_norm": 0.7306228876113892, "learning_rate": 0.0001, "elbo": 3.58082, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9612362186253813}
{"step": 50500, "loss": 3.5928627014160157, "grad_norm": 0.6587244868278503, "learning_rate": 0.0001, "elbo": 3.59286, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.961529439361952}
{"step": 50505, "loss": 3.5797714233398437, "grad_norm": 0.6693379878997803, "learning_rate": 0.0001, "elbo": 3.57977, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9618226600985222}
{"step": 50510, "loss": 3.574964904785156, "grad_norm": 0.7363816499710083, "learning_rate": 0.0001, "elbo": 3.57497, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9621158808350927}
{"step": 50515, "loss": 3.6019439697265625, "grad_norm": 0.6982817053794861, "learning_rate": 0.0001, "elbo": 3.60194, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.962409101571663}
{"step": 50520, "loss": 3.6006092071533202, "grad_norm": 0.7409434914588928, "learning_rate": 0.0001, "elbo": 3.60061, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9627023223082336}
{"step": 50525, "loss": 3.583898162841797, "grad_norm": 0.717077910900116, "learning_rate": 0.0001, "elbo": 3.5839, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.962995543044804}
{"step": 50530, "loss": 3.5920570373535154, "grad_norm": 0.6890257000923157, "learning_rate": 0.0001, "elbo": 3.59206, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9632887637813745}
{"step": 50530, "loss": 3.5920570373535154, "grad_norm": 0.6890257000923157, "learning_rate": 0.0001, "elbo": 3.59206, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9632887637813745}
{"step": 50535, "loss": 3.5767753601074217, "grad_norm": 0.7116011381149292, "learning_rate": 0.0001, "elbo": 3.57678, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.963581984517945}
{"step": 50540, "loss": 3.586679458618164, "grad_norm": 0.7298220992088318, "learning_rate": 0.0001, "elbo": 3.58668, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9638752052545154}
{"step": 50545, "loss": 3.578451156616211, "grad_norm": 0.708637535572052, "learning_rate": 0.0001, "elbo": 3.57845, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.964168425991086}
{"step": 50550, "loss": 3.5733741760253905, "grad_norm": 0.6686422824859619, "learning_rate": 0.0001, "elbo": 3.57337, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9644616467276563}
{"step": 50555, "loss": 3.5934688568115236, "grad_norm": 0.8002776503562927, "learning_rate": 0.0001, "elbo": 3.59347, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9647548674642272}
{"step": 50560, "loss": 3.5730560302734373, "grad_norm": 0.7159708738327026, "learning_rate": 0.0001, "elbo": 3.57306, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9650480882007977}
{"step": 50565, "loss": 3.578553009033203, "grad_norm": 0.6779720783233643, "learning_rate": 0.0001, "elbo": 3.57855, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.965341308937368}
{"step": 50570, "loss": 3.5688705444335938, "grad_norm": 0.6690540313720703, "learning_rate": 0.0001, "elbo": 3.56887, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9656345296739386}
{"step": 50575, "loss": 3.5766761779785154, "grad_norm": 0.6471887826919556, "learning_rate": 0.0001, "elbo": 3.57668, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.965927750410509}
{"step": 50580, "loss": 3.5809024810791015, "grad_norm": 0.7428528070449829, "learning_rate": 0.0001, "elbo": 3.5809, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9662209711470795}
{"step": 50585, "loss": 3.5935829162597654, "grad_norm": 0.6632081270217896, "learning_rate": 0.0001, "elbo": 3.59358, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.96651419188365}
{"step": 50590, "loss": 3.580026626586914, "grad_norm": 0.766786515712738, "learning_rate": 0.0001, "elbo": 3.58003, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9668074126202204}
{"step": 50595, "loss": 3.578662109375, "grad_norm": 0.6729409694671631, "learning_rate": 0.0001, "elbo": 3.57866, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.967100633356791}
{"step": 50600, "loss": 3.582631301879883, "grad_norm": 0.6862097382545471, "learning_rate": 0.0001, "elbo": 3.58263, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9673938540933613}
{"step": 50605, "loss": 3.593230056762695, "grad_norm": 0.7660003304481506, "learning_rate": 0.0001, "elbo": 3.59323, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.967687074829932}
{"step": 50610, "loss": 3.5658851623535157, "grad_norm": 0.696276068687439, "learning_rate": 0.0001, "elbo": 3.56588, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9679802955665027}
{"step": 50615, "loss": 3.5899738311767577, "grad_norm": 0.7057716846466064, "learning_rate": 0.0001, "elbo": 3.58997, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.968273516303073}
{"step": 50615, "loss": 3.5899738311767577, "grad_norm": 0.7057716846466064, "learning_rate": 0.0001, "elbo": 3.58997, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.968273516303073}
{"step": 50620, "loss": 3.5695682525634767, "grad_norm": 0.6784772276878357, "learning_rate": 0.0001, "elbo": 3.56957, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9685667370396436}
{"step": 50625, "loss": 3.568784713745117, "grad_norm": 0.9510340094566345, "learning_rate": 0.0001, "elbo": 3.56878, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.968859957776214}
{"step": 50630, "loss": 3.5885719299316405, "grad_norm": 0.7375329732894897, "learning_rate": 0.0001, "elbo": 3.58857, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9691531785127845}
{"step": 50635, "loss": 3.579197311401367, "grad_norm": 0.750825047492981, "learning_rate": 0.0001, "elbo": 3.5792, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.969446399249355}
{"step": 50640, "loss": 3.5869209289550783, "grad_norm": 0.6791161894798279, "learning_rate": 0.0001, "elbo": 3.58692, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9697396199859254}
{"step": 50645, "loss": 3.580483627319336, "grad_norm": 0.6602906584739685, "learning_rate": 0.0001, "elbo": 3.58048, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.970032840722496}
{"step": 50650, "loss": 3.5762405395507812, "grad_norm": 0.7506871223449707, "learning_rate": 0.0001, "elbo": 3.57624, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9703260614590663}
{"step": 50650, "loss": 3.5762405395507812, "grad_norm": 0.7506871223449707, "learning_rate": 0.0001, "elbo": 3.57624, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9703260614590663}
{"step": 50655, "loss": 3.594411087036133, "grad_norm": 0.6795596480369568, "learning_rate": 0.0001, "elbo": 3.59441, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9706192821956368}
{"step": 50660, "loss": 3.589551544189453, "grad_norm": 0.6672900319099426, "learning_rate": 0.0001, "elbo": 3.58955, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.970912502932207}
{"step": 50665, "loss": 3.5973430633544923, "grad_norm": 0.694924533367157, "learning_rate": 0.0001, "elbo": 3.59734, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9712057236687777}
{"step": 50670, "loss": 3.577579879760742, "grad_norm": 0.6473236083984375, "learning_rate": 0.0001, "elbo": 3.57758, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.971498944405348}
{"step": 50675, "loss": 3.5702445983886717, "grad_norm": 0.6674010157585144, "learning_rate": 0.0001, "elbo": 3.57024, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9717921651419186}
{"step": 50680, "loss": 3.584154510498047, "grad_norm": 0.6783868670463562, "learning_rate": 0.0001, "elbo": 3.58415, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9720853858784895}
{"step": 50685, "loss": 3.5736164093017577, "grad_norm": 0.7414990067481995, "learning_rate": 0.0001, "elbo": 3.57362, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.97237860661506}
{"step": 50690, "loss": 3.5814895629882812, "grad_norm": 0.7096320986747742, "learning_rate": 0.0001, "elbo": 3.58149, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9726718273516304}
{"step": 50695, "loss": 3.5891265869140625, "grad_norm": 0.678188145160675, "learning_rate": 0.0001, "elbo": 3.58913, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.972965048088201}
{"step": 50700, "loss": 3.5708824157714845, "grad_norm": 0.8168210387229919, "learning_rate": 0.0001, "elbo": 3.57088, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9732582688247713}
{"step": 50705, "loss": 3.5829368591308595, "grad_norm": 0.644110381603241, "learning_rate": 0.0001, "elbo": 3.58294, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9735514895613417}
{"step": 50710, "loss": 3.5967445373535156, "grad_norm": 0.6976369023323059, "learning_rate": 0.0001, "elbo": 3.59674, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.973844710297912}
{"step": 50715, "loss": 3.576470947265625, "grad_norm": 0.6531811356544495, "learning_rate": 0.0001, "elbo": 3.57647, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9741379310344827}
{"step": 50720, "loss": 3.578866195678711, "grad_norm": 0.7286016345024109, "learning_rate": 0.0001, "elbo": 3.57887, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.974431151771053}
{"step": 50725, "loss": 3.5765647888183594, "grad_norm": 0.6980116963386536, "learning_rate": 0.0001, "elbo": 3.57656, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.974724372507624}
{"step": 50730, "loss": 3.5746864318847655, "grad_norm": 0.9261210560798645, "learning_rate": 0.0001, "elbo": 3.57469, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9750175932441945}
{"step": 50735, "loss": 3.576318359375, "grad_norm": 0.7677637934684753, "learning_rate": 0.0001, "elbo": 3.57632, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.975310813980765}
{"step": 50740, "loss": 3.585390090942383, "grad_norm": 0.7899040579795837, "learning_rate": 0.0001, "elbo": 3.58539, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9756040347173354}
{"step": 50745, "loss": 3.593190002441406, "grad_norm": 0.7559524178504944, "learning_rate": 0.0001, "elbo": 3.59319, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.975897255453906}
{"step": 50750, "loss": 3.572317886352539, "grad_norm": 0.7093290090560913, "learning_rate": 0.0001, "elbo": 3.57232, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9761904761904763}
{"step": 50755, "loss": 3.574069595336914, "grad_norm": 0.7301366329193115, "learning_rate": 0.0001, "elbo": 3.57407, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9764836969270467}
{"step": 50760, "loss": 3.589862823486328, "grad_norm": 0.7632830739021301, "learning_rate": 0.0001, "elbo": 3.58986, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.976776917663617}
{"step": 50765, "loss": 3.5953277587890624, "grad_norm": 0.6687573194503784, "learning_rate": 0.0001, "elbo": 3.59533, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9770701384001876}
{"step": 50770, "loss": 3.576377105712891, "grad_norm": 0.8459429144859314, "learning_rate": 0.0001, "elbo": 3.57638, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.977363359136758}
{"step": 50775, "loss": 3.5849205017089845, "grad_norm": 0.6929914355278015, "learning_rate": 0.0001, "elbo": 3.58492, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9776565798733285}
{"step": 50775, "loss": 3.5849205017089845, "grad_norm": 0.6929914355278015, "learning_rate": 0.0001, "elbo": 3.58492, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9776565798733285}
{"step": 50780, "loss": 3.5658470153808595, "grad_norm": 0.7139355540275574, "learning_rate": 0.0001, "elbo": 3.56585, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.977949800609899}
{"step": 50785, "loss": 3.5830333709716795, "grad_norm": 0.7415827512741089, "learning_rate": 0.0001, "elbo": 3.58303, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9782430213464695}
{"step": 50790, "loss": 3.57860107421875, "grad_norm": 1.1751042604446411, "learning_rate": 0.0001, "elbo": 3.5786, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.97853624208304}
{"step": 50795, "loss": 3.5864501953125, "grad_norm": 0.7521055936813354, "learning_rate": 0.0001, "elbo": 3.58645, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9788294628196104}
{"step": 50800, "loss": 3.5711429595947264, "grad_norm": 0.8567811250686646, "learning_rate": 0.0001, "elbo": 3.57114, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.979122683556181}
{"step": 50805, "loss": 3.5717243194580077, "grad_norm": 0.704863429069519, "learning_rate": 0.0001, "elbo": 3.57172, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9794159042927517}
{"step": 50810, "loss": 3.5732994079589844, "grad_norm": 0.7042779326438904, "learning_rate": 0.0001, "elbo": 3.5733, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.979709125029322}
{"step": 50815, "loss": 3.6020477294921873, "grad_norm": 0.7218732833862305, "learning_rate": 0.0001, "elbo": 3.60205, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9800023457658926}
{"step": 50820, "loss": 3.586114501953125, "grad_norm": 0.8485289812088013, "learning_rate": 0.0001, "elbo": 3.58611, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.980295566502463}
{"step": 50825, "loss": 3.601457214355469, "grad_norm": 0.808820366859436, "learning_rate": 0.0001, "elbo": 3.60146, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9805887872390335}
{"step": 50830, "loss": 3.575940704345703, "grad_norm": 0.6961830854415894, "learning_rate": 0.0001, "elbo": 3.57594, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.980882007975604}
{"step": 50835, "loss": 3.5943576812744142, "grad_norm": 0.6814413666725159, "learning_rate": 0.0001, "elbo": 3.59436, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9811752287121744}
{"step": 50840, "loss": 3.5788528442382814, "grad_norm": 0.8163164258003235, "learning_rate": 0.0001, "elbo": 3.57885, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.981468449448745}
{"step": 50845, "loss": 3.6040191650390625, "grad_norm": 0.6937527656555176, "learning_rate": 0.0001, "elbo": 3.60402, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9817616701853153}
{"step": 50850, "loss": 3.5815383911132814, "grad_norm": 0.7019112706184387, "learning_rate": 0.0001, "elbo": 3.58154, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9820548909218862}
{"step": 50855, "loss": 3.575886535644531, "grad_norm": 0.6984802484512329, "learning_rate": 0.0001, "elbo": 3.57589, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9823481116584567}
{"step": 50860, "loss": 3.593595504760742, "grad_norm": 0.6799893379211426, "learning_rate": 0.0001, "elbo": 3.5936, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.982641332395027}
{"step": 50865, "loss": 3.5700454711914062, "grad_norm": 0.6906850934028625, "learning_rate": 0.0001, "elbo": 3.57005, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9829345531315976}
{"step": 50870, "loss": 3.5843448638916016, "grad_norm": 0.6951169967651367, "learning_rate": 0.0001, "elbo": 3.58434, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.983227773868168}
{"step": 50875, "loss": 3.574013900756836, "grad_norm": 0.7302346229553223, "learning_rate": 0.0001, "elbo": 3.57401, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9835209946047385}
{"step": 50880, "loss": 3.5913902282714845, "grad_norm": 0.7281321287155151, "learning_rate": 0.0001, "elbo": 3.59139, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.983814215341309}
{"step": 50885, "loss": 3.5748504638671874, "grad_norm": 0.6745734214782715, "learning_rate": 0.0001, "elbo": 3.57485, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9841074360778794}
{"step": 50890, "loss": 3.5780078887939455, "grad_norm": 0.7003770470619202, "learning_rate": 0.0001, "elbo": 3.57801, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.98440065681445}
{"step": 50895, "loss": 3.569217300415039, "grad_norm": 0.686286985874176, "learning_rate": 0.0001, "elbo": 3.56922, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9846938775510203}
{"step": 50900, "loss": 3.576458740234375, "grad_norm": 0.7100096344947815, "learning_rate": 0.0001, "elbo": 3.57646, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.984987098287591}
{"step": 50905, "loss": 3.5815711975097657, "grad_norm": 0.6466203927993774, "learning_rate": 0.0001, "elbo": 3.58157, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9852803190241612}
{"step": 50910, "loss": 3.5930557250976562, "grad_norm": 0.6970205903053284, "learning_rate": 0.0001, "elbo": 3.59306, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9855735397607317}
{"step": 50915, "loss": 3.569028854370117, "grad_norm": 1.02729070186615, "learning_rate": 0.0001, "elbo": 3.56903, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.985866760497302}
{"step": 50920, "loss": 3.5738059997558596, "grad_norm": 0.6825739741325378, "learning_rate": 0.0001, "elbo": 3.57381, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9861599812338726}
{"step": 50925, "loss": 3.604608154296875, "grad_norm": 0.8091954588890076, "learning_rate": 0.0001, "elbo": 3.60461, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9864532019704435}
{"step": 50930, "loss": 3.59307861328125, "grad_norm": 0.6851446628570557, "learning_rate": 0.0001, "elbo": 3.59308, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.986746422707014}
{"step": 50935, "loss": 3.5716339111328126, "grad_norm": 0.6703897714614868, "learning_rate": 0.0001, "elbo": 3.57163, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9870396434435844}
{"step": 50940, "loss": 3.5933170318603516, "grad_norm": 0.6769340634346008, "learning_rate": 0.0001, "elbo": 3.59332, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.987332864180155}
{"step": 50945, "loss": 3.592144012451172, "grad_norm": 0.6797960996627808, "learning_rate": 0.0001, "elbo": 3.59214, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9876260849167253}
{"step": 50950, "loss": 3.5862442016601563, "grad_norm": 0.6975454688072205, "learning_rate": 0.0001, "elbo": 3.58624, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9879193056532958}
{"step": 50955, "loss": 3.5703216552734376, "grad_norm": 0.691979706287384, "learning_rate": 0.0001, "elbo": 3.57032, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9882125263898662}
{"step": 50960, "loss": 3.581493377685547, "grad_norm": 0.6835945844650269, "learning_rate": 0.0001, "elbo": 3.58149, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9885057471264367}
{"step": 50965, "loss": 3.5647937774658205, "grad_norm": 0.7134639024734497, "learning_rate": 0.0001, "elbo": 3.56479, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.988798967863007}
{"step": 50970, "loss": 3.5728126525878907, "grad_norm": 0.7181068062782288, "learning_rate": 0.0001, "elbo": 3.57281, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.989092188599578}
{"step": 50975, "loss": 3.6004966735839843, "grad_norm": 0.6876875162124634, "learning_rate": 0.0001, "elbo": 3.6005, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9893854093361485}
{"step": 50980, "loss": 3.5711509704589846, "grad_norm": 0.6625213027000427, "learning_rate": 0.0001, "elbo": 3.57115, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.989678630072719}
{"step": 50985, "loss": 3.564698028564453, "grad_norm": 0.6914871335029602, "learning_rate": 0.0001, "elbo": 3.5647, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9899718508092894}
{"step": 50990, "loss": 3.584621047973633, "grad_norm": 0.8734860420227051, "learning_rate": 0.0001, "elbo": 3.58462, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.99026507154586}
{"step": 50995, "loss": 3.5590667724609375, "grad_norm": 0.6918940544128418, "learning_rate": 0.0001, "elbo": 3.55907, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9905582922824303}
{"step": 50995, "loss": 3.5590667724609375, "grad_norm": 0.6918940544128418, "learning_rate": 0.0001, "elbo": 3.55907, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9905582922824303}
{"step": 51000, "loss": 3.5739364624023438, "grad_norm": 0.7156712412834167, "learning_rate": 0.0001, "elbo": 3.57394, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9908515130190008}
{"step": 51000, "loss": 3.5739364624023438, "grad_norm": 0.7156712412834167, "learning_rate": 0.0001, "elbo": 3.57394, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9908515130190008}
{"step": 51005, "loss": 3.570302963256836, "grad_norm": 0.7532511949539185, "learning_rate": 0.0001, "elbo": 3.5703, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.991144733755571}
{"step": 51010, "loss": 3.5886154174804688, "grad_norm": 0.7661387324333191, "learning_rate": 0.0001, "elbo": 3.58862, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9914379544921417}
{"step": 51015, "loss": 3.591424560546875, "grad_norm": 0.7793665528297424, "learning_rate": 0.0001, "elbo": 3.59142, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.991731175228712}
{"step": 51020, "loss": 3.587044525146484, "grad_norm": 0.7028252482414246, "learning_rate": 0.0001, "elbo": 3.58704, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9920243959652826}
{"step": 51025, "loss": 3.575724411010742, "grad_norm": 0.7486331462860107, "learning_rate": 0.0001, "elbo": 3.57572, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.992317616701853}
{"step": 51025, "loss": 3.575724411010742, "grad_norm": 0.7486331462860107, "learning_rate": 0.0001, "elbo": 3.57572, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.992317616701853}
{"step": 51030, "loss": 3.5748424530029297, "grad_norm": 0.8157073855400085, "learning_rate": 0.0001, "elbo": 3.57484, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9926108374384235}
{"step": 51030, "loss": 3.5748424530029297, "grad_norm": 0.8157073855400085, "learning_rate": 0.0001, "elbo": 3.57484, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9926108374384235}
{"step": 51035, "loss": 3.582320785522461, "grad_norm": 0.6915847063064575, "learning_rate": 0.0001, "elbo": 3.58232, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.992904058174994}
{"step": 51040, "loss": 3.572296905517578, "grad_norm": 0.6690250039100647, "learning_rate": 0.0001, "elbo": 3.5723, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9931972789115644}
{"step": 51045, "loss": 3.5867958068847656, "grad_norm": 0.6572727560997009, "learning_rate": 0.0001, "elbo": 3.5868, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.993490499648135}
{"step": 51050, "loss": 3.594538116455078, "grad_norm": 0.7104662656784058, "learning_rate": 0.0001, "elbo": 3.59454, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9937837203847057}
{"step": 51055, "loss": 3.579593276977539, "grad_norm": 0.6772936582565308, "learning_rate": 0.0001, "elbo": 3.57959, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.994076941121276}
{"step": 51060, "loss": 3.574147033691406, "grad_norm": 0.7482087016105652, "learning_rate": 0.0001, "elbo": 3.57415, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9943701618578467}
{"step": 51065, "loss": 3.5850753784179688, "grad_norm": 0.6907663345336914, "learning_rate": 0.0001, "elbo": 3.58508, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.994663382594417}
{"step": 51070, "loss": 3.5866989135742187, "grad_norm": 0.6988111138343811, "learning_rate": 0.0001, "elbo": 3.5867, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9949566033309876}
{"step": 51075, "loss": 3.5755287170410157, "grad_norm": 0.6926203966140747, "learning_rate": 0.0001, "elbo": 3.57553, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.995249824067558}
{"step": 51080, "loss": 3.5926368713378904, "grad_norm": 0.6652479767799377, "learning_rate": 0.0001, "elbo": 3.59264, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9955430448041285}
{"step": 51085, "loss": 3.576914978027344, "grad_norm": 0.701216459274292, "learning_rate": 0.0001, "elbo": 3.57691, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.995836265540699}
{"step": 51090, "loss": 3.5861061096191404, "grad_norm": 0.9720337390899658, "learning_rate": 0.0001, "elbo": 3.58611, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9961294862772694}
{"step": 51095, "loss": 3.5644954681396483, "grad_norm": 0.7278298139572144, "learning_rate": 0.0001, "elbo": 3.5645, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9964227070138403}
{"step": 51100, "loss": 3.581584930419922, "grad_norm": 0.7327967882156372, "learning_rate": 0.0001, "elbo": 3.58159, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9967159277504107}
{"step": 51105, "loss": 3.5844955444335938, "grad_norm": 0.6904515624046326, "learning_rate": 0.0001, "elbo": 3.5845, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.997009148486981}
{"step": 51105, "loss": 3.5844955444335938, "grad_norm": 0.6904515624046326, "learning_rate": 0.0001, "elbo": 3.5845, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.997009148486981}
{"step": 51110, "loss": 3.572393035888672, "grad_norm": 0.7081155776977539, "learning_rate": 0.0001, "elbo": 3.57239, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9973023692235516}
{"step": 51115, "loss": 3.5920150756835936, "grad_norm": 0.7209555506706238, "learning_rate": 0.0001, "elbo": 3.59201, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.997595589960122}
{"step": 51120, "loss": 3.580748748779297, "grad_norm": 0.6828993558883667, "learning_rate": 0.0001, "elbo": 3.58075, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9978888106966926}
{"step": 51125, "loss": 3.555329132080078, "grad_norm": 0.8970452547073364, "learning_rate": 0.0001, "elbo": 3.55533, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.998182031433263}
{"step": 51130, "loss": 3.584633636474609, "grad_norm": 0.7208479046821594, "learning_rate": 0.0001, "elbo": 3.58463, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9984752521698335}
{"step": 51135, "loss": 3.5733875274658202, "grad_norm": 0.6855226159095764, "learning_rate": 0.0001, "elbo": 3.57339, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.998768472906404}
{"step": 51140, "loss": 3.5794662475585937, "grad_norm": 0.675638735294342, "learning_rate": 0.0001, "elbo": 3.57947, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9990616936429744}
{"step": 51145, "loss": 3.5911533355712892, "grad_norm": 0.713175356388092, "learning_rate": 0.0001, "elbo": 3.59115, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.999354914379545}
{"step": 51150, "loss": 3.576673126220703, "grad_norm": 0.6769310832023621, "learning_rate": 0.0001, "elbo": 3.57667, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9996481351161153}
{"step": 51155, "loss": 3.587666320800781, "grad_norm": 0.7467366456985474, "learning_rate": 0.0001, "elbo": 3.58767, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 2.9999413558526857}
{"step": 51160, "loss": 3.5691551208496093, "grad_norm": 0.6764064431190491, "learning_rate": 0.0001, "elbo": 3.56915, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.000234576589256}
{"step": 51165, "loss": 3.5766365051269533, "grad_norm": 0.8263337016105652, "learning_rate": 0.0001, "elbo": 3.57664, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.000527797325827}
{"step": 51170, "loss": 3.577442932128906, "grad_norm": 1.2481805086135864, "learning_rate": 0.0001, "elbo": 3.57744, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0008210180623975}
{"step": 51175, "loss": 3.5871650695800783, "grad_norm": 0.6929582953453064, "learning_rate": 0.0001, "elbo": 3.58717, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.001114238798968}
{"step": 51180, "loss": 3.579486846923828, "grad_norm": 0.966399610042572, "learning_rate": 0.0001, "elbo": 3.57949, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0014074595355384}
{"step": 51185, "loss": 3.578053665161133, "grad_norm": 0.6545642614364624, "learning_rate": 0.0001, "elbo": 3.57805, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.001700680272109}
{"step": 51190, "loss": 3.594587707519531, "grad_norm": 0.7839416265487671, "learning_rate": 0.0001, "elbo": 3.59459, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0019939010086794}
{"step": 51195, "loss": 3.59288330078125, "grad_norm": 0.6447132229804993, "learning_rate": 0.0001, "elbo": 3.59288, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.00228712174525}
{"step": 51200, "loss": 3.586931610107422, "grad_norm": 0.7717435956001282, "learning_rate": 0.0001, "elbo": 3.58693, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0025803424818203}
{"step": 51205, "loss": 3.5806236267089844, "grad_norm": 0.6747391819953918, "learning_rate": 0.0001, "elbo": 3.58062, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0028735632183907}
{"step": 51210, "loss": 3.5721778869628906, "grad_norm": 0.6761011481285095, "learning_rate": 0.0001, "elbo": 3.57218, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.003166783954961}
{"step": 51215, "loss": 3.5778690338134767, "grad_norm": 0.6977910995483398, "learning_rate": 0.0001, "elbo": 3.57787, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0034600046915316}
{"step": 51215, "loss": 3.5778690338134767, "grad_norm": 0.6977910995483398, "learning_rate": 0.0001, "elbo": 3.57787, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0034600046915316}
{"step": 51220, "loss": 3.590143585205078, "grad_norm": 0.6863137483596802, "learning_rate": 0.0001, "elbo": 3.59014, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.003753225428102}
{"step": 51225, "loss": 3.5862918853759767, "grad_norm": 0.6917771697044373, "learning_rate": 0.0001, "elbo": 3.58629, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.004046446164673}
{"step": 51230, "loss": 3.5865875244140626, "grad_norm": 0.6784014105796814, "learning_rate": 0.0001, "elbo": 3.58659, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0043396669012434}
{"step": 51235, "loss": 3.563581085205078, "grad_norm": 0.7826542258262634, "learning_rate": 0.0001, "elbo": 3.56358, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.004632887637814}
{"step": 51240, "loss": 3.582355499267578, "grad_norm": 0.8071961998939514, "learning_rate": 0.0001, "elbo": 3.58236, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0049261083743843}
{"step": 51245, "loss": 3.5830104827880858, "grad_norm": 0.7121264338493347, "learning_rate": 0.0001, "elbo": 3.58301, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.005219329110955}
{"step": 51250, "loss": 3.573487091064453, "grad_norm": 0.6897547245025635, "learning_rate": 0.0001, "elbo": 3.57349, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0055125498475252}
{"step": 51255, "loss": 3.573560333251953, "grad_norm": 0.7422645092010498, "learning_rate": 0.0001, "elbo": 3.57356, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0058057705840957}
{"step": 51260, "loss": 3.588043212890625, "grad_norm": 0.7190595865249634, "learning_rate": 0.0001, "elbo": 3.58804, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.006098991320666}
{"step": 51265, "loss": 3.5759323120117186, "grad_norm": 1.0305980443954468, "learning_rate": 0.0001, "elbo": 3.57593, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0063922120572366}
{"step": 51270, "loss": 3.5944259643554686, "grad_norm": 0.7385962009429932, "learning_rate": 0.0001, "elbo": 3.59443, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.006685432793807}
{"step": 51275, "loss": 3.5914077758789062, "grad_norm": 0.7074025273323059, "learning_rate": 0.0001, "elbo": 3.59141, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0069786535303775}
{"step": 51280, "loss": 3.569773483276367, "grad_norm": 0.8768141269683838, "learning_rate": 0.0001, "elbo": 3.56977, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.007271874266948}
{"step": 51285, "loss": 3.592207336425781, "grad_norm": 0.7445461750030518, "learning_rate": 0.0001, "elbo": 3.59221, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0075650950035184}
{"step": 51290, "loss": 3.57294921875, "grad_norm": 0.7229432463645935, "learning_rate": 0.0001, "elbo": 3.57295, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0078583157400893}
{"step": 51295, "loss": 3.5741004943847656, "grad_norm": 0.8097068071365356, "learning_rate": 0.0001, "elbo": 3.5741, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.00815153647666}
{"step": 51300, "loss": 3.571056365966797, "grad_norm": 0.700538158416748, "learning_rate": 0.0001, "elbo": 3.57106, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0084447572132302}
{"step": 51305, "loss": 3.5677024841308596, "grad_norm": 0.6750959157943726, "learning_rate": 0.0001, "elbo": 3.5677, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0087379779498007}
{"step": 51310, "loss": 3.585178756713867, "grad_norm": 0.7088888883590698, "learning_rate": 0.0001, "elbo": 3.58518, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.009031198686371}
{"step": 51315, "loss": 3.5809696197509764, "grad_norm": 0.8955128192901611, "learning_rate": 0.0001, "elbo": 3.58097, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0093244194229416}
{"step": 51315, "loss": 3.5809696197509764, "grad_norm": 0.8955128192901611, "learning_rate": 0.0001, "elbo": 3.58097, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0093244194229416}
{"step": 51320, "loss": 3.576799011230469, "grad_norm": 0.6389263272285461, "learning_rate": 0.0001, "elbo": 3.5768, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.009617640159512}
{"step": 51325, "loss": 3.5641300201416017, "grad_norm": 0.6838555335998535, "learning_rate": 0.0001, "elbo": 3.56413, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0099108608960825}
{"step": 51330, "loss": 3.568498992919922, "grad_norm": 0.7004485130310059, "learning_rate": 0.0001, "elbo": 3.5685, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.010204081632653}
{"step": 51335, "loss": 3.576114273071289, "grad_norm": 0.6790097951889038, "learning_rate": 0.0001, "elbo": 3.57611, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0104973023692234}
{"step": 51340, "loss": 3.5788299560546877, "grad_norm": 0.7023919820785522, "learning_rate": 0.0001, "elbo": 3.57883, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.010790523105794}
{"step": 51345, "loss": 3.5636665344238283, "grad_norm": 0.6901507377624512, "learning_rate": 0.0001, "elbo": 3.56367, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0110837438423643}
{"step": 51350, "loss": 3.5934776306152343, "grad_norm": 0.6675535440444946, "learning_rate": 0.0001, "elbo": 3.59348, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0113769645789352}
{"step": 51355, "loss": 3.5760040283203125, "grad_norm": 0.6760959029197693, "learning_rate": 0.0001, "elbo": 3.576, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0116701853155057}
{"step": 51360, "loss": 3.5917530059814453, "grad_norm": 0.7655439376831055, "learning_rate": 0.0001, "elbo": 3.59175, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.011963406052076}
{"step": 51365, "loss": 3.5877403259277343, "grad_norm": 0.8122981190681458, "learning_rate": 0.0001, "elbo": 3.58774, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0122566267886466}
{"step": 51370, "loss": 3.5859100341796877, "grad_norm": 0.6720807552337646, "learning_rate": 0.0001, "elbo": 3.58591, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.012549847525217}
{"step": 51375, "loss": 3.5748046875, "grad_norm": 0.6723123788833618, "learning_rate": 0.0001, "elbo": 3.5748, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0128430682617875}
{"step": 51380, "loss": 3.5597023010253905, "grad_norm": 0.7197936177253723, "learning_rate": 0.0001, "elbo": 3.5597, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.013136288998358}
{"step": 51385, "loss": 3.586608123779297, "grad_norm": 0.7151612639427185, "learning_rate": 0.0001, "elbo": 3.58661, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0134295097349284}
{"step": 51390, "loss": 3.5859756469726562, "grad_norm": 0.7605574131011963, "learning_rate": 0.0001, "elbo": 3.58598, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.013722730471499}
{"step": 51395, "loss": 3.5854412078857423, "grad_norm": 0.9692539572715759, "learning_rate": 0.0001, "elbo": 3.58544, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0140159512080693}
{"step": 51400, "loss": 3.5794925689697266, "grad_norm": 0.971645176410675, "learning_rate": 0.0001, "elbo": 3.57949, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0143091719446398}
{"step": 51405, "loss": 3.586290740966797, "grad_norm": 0.707221508026123, "learning_rate": 0.0001, "elbo": 3.58629, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.01460239268121}
{"step": 51410, "loss": 3.587510681152344, "grad_norm": 0.6988608241081238, "learning_rate": 0.0001, "elbo": 3.58751, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.014895613417781}
{"step": 51415, "loss": 3.57347412109375, "grad_norm": 0.7088586091995239, "learning_rate": 0.0001, "elbo": 3.57347, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0151888341543516}
{"step": 51420, "loss": 3.5710411071777344, "grad_norm": 0.7000327706336975, "learning_rate": 0.0001, "elbo": 3.57104, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.015482054890922}
{"step": 51425, "loss": 3.588539886474609, "grad_norm": 0.9682056903839111, "learning_rate": 0.0001, "elbo": 3.58854, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0157752756274925}
{"step": 51430, "loss": 3.570234680175781, "grad_norm": 0.6975557804107666, "learning_rate": 0.0001, "elbo": 3.57023, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.016068496364063}
{"step": 51435, "loss": 3.576524353027344, "grad_norm": 0.7919737696647644, "learning_rate": 0.0001, "elbo": 3.57652, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0163617171006334}
{"step": 51440, "loss": 3.570032501220703, "grad_norm": 0.7050603032112122, "learning_rate": 0.0001, "elbo": 3.57003, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.016654937837204}
{"step": 51445, "loss": 3.5716106414794924, "grad_norm": 0.6514738202095032, "learning_rate": 0.0001, "elbo": 3.57161, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0169481585737743}
{"step": 51450, "loss": 3.5805187225341797, "grad_norm": 0.6585482954978943, "learning_rate": 0.0001, "elbo": 3.58052, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0172413793103448}
{"step": 51455, "loss": 3.5667373657226564, "grad_norm": 0.6720467805862427, "learning_rate": 0.0001, "elbo": 3.56674, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.017534600046915}
{"step": 51460, "loss": 3.599599838256836, "grad_norm": 0.6835516095161438, "learning_rate": 0.0001, "elbo": 3.5996, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0178278207834857}
{"step": 51465, "loss": 3.5761157989501955, "grad_norm": 0.6911314725875854, "learning_rate": 0.0001, "elbo": 3.57612, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.018121041520056}
{"step": 51470, "loss": 3.5653038024902344, "grad_norm": 0.6990976333618164, "learning_rate": 0.0001, "elbo": 3.5653, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.018414262256627}
{"step": 51475, "loss": 3.5670055389404296, "grad_norm": 0.7721582651138306, "learning_rate": 0.0001, "elbo": 3.56701, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0187074829931975}
{"step": 51480, "loss": 3.5788639068603514, "grad_norm": 0.7395666837692261, "learning_rate": 0.0001, "elbo": 3.57886, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.019000703729768}
{"step": 51485, "loss": 3.5665950775146484, "grad_norm": 0.6765830516815186, "learning_rate": 0.0001, "elbo": 3.56659, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0192939244663384}
{"step": 51490, "loss": 3.578095626831055, "grad_norm": 0.7837682366371155, "learning_rate": 0.0001, "elbo": 3.5781, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.019587145202909}
{"step": 51490, "loss": 3.578095626831055, "grad_norm": 0.7837682366371155, "learning_rate": 0.0001, "elbo": 3.5781, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.019587145202909}
{"step": 51495, "loss": 3.5835376739501954, "grad_norm": 0.7062662839889526, "learning_rate": 0.0001, "elbo": 3.58354, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0198803659394793}
{"step": 51500, "loss": 3.58763427734375, "grad_norm": 0.7085075974464417, "learning_rate": 0.0001, "elbo": 3.58763, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0201735866760497}
{"step": 51500, "loss": 3.58763427734375, "grad_norm": 0.7085075974464417, "learning_rate": 0.0001, "elbo": 3.58763, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0201735866760497}
{"step": 51505, "loss": 3.5662628173828126, "grad_norm": 0.6832759380340576, "learning_rate": 0.0001, "elbo": 3.56626, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.02046680741262}
{"step": 51510, "loss": 3.5688617706298826, "grad_norm": 0.707728922367096, "learning_rate": 0.0001, "elbo": 3.56886, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0207600281491906}
{"step": 51515, "loss": 3.569620132446289, "grad_norm": 0.6802681088447571, "learning_rate": 0.0001, "elbo": 3.56962, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.021053248885761}
{"step": 51520, "loss": 3.586957550048828, "grad_norm": 0.7144879698753357, "learning_rate": 0.0001, "elbo": 3.58696, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0213464696223316}
{"step": 51525, "loss": 3.5793006896972654, "grad_norm": 0.7422140836715698, "learning_rate": 0.0001, "elbo": 3.5793, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.021639690358902}
{"step": 51530, "loss": 3.5792922973632812, "grad_norm": 0.8239867687225342, "learning_rate": 0.0001, "elbo": 3.57929, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0219329110954725}
{"step": 51530, "loss": 3.5792922973632812, "grad_norm": 0.8239867687225342, "learning_rate": 0.0001, "elbo": 3.57929, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0219329110954725}
{"step": 51535, "loss": 3.585388946533203, "grad_norm": 1.024007797241211, "learning_rate": 0.0001, "elbo": 3.58539, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0222261318320434}
{"step": 51540, "loss": 3.566555404663086, "grad_norm": 0.6889724135398865, "learning_rate": 0.0001, "elbo": 3.56656, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.022519352568614}
{"step": 51545, "loss": 3.56378173828125, "grad_norm": 0.6721901893615723, "learning_rate": 0.0001, "elbo": 3.56378, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0228125733051843}
{"step": 51550, "loss": 3.572800064086914, "grad_norm": 0.7575973272323608, "learning_rate": 0.0001, "elbo": 3.5728, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0231057940417547}
{"step": 51550, "loss": 3.572800064086914, "grad_norm": 0.7575973272323608, "learning_rate": 0.0001, "elbo": 3.5728, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0231057940417547}
{"step": 51555, "loss": 3.5829185485839843, "grad_norm": 0.6525410413742065, "learning_rate": 0.0001, "elbo": 3.58292, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.023399014778325}
{"step": 51555, "loss": 3.5829185485839843, "grad_norm": 0.6525410413742065, "learning_rate": 0.0001, "elbo": 3.58292, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.023399014778325}
{"step": 51560, "loss": 3.593739318847656, "grad_norm": 0.6910570859909058, "learning_rate": 0.0001, "elbo": 3.59374, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0236922355148956}
{"step": 51565, "loss": 3.5813613891601563, "grad_norm": 0.7327620387077332, "learning_rate": 0.0001, "elbo": 3.58136, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.023985456251466}
{"step": 51570, "loss": 3.563104248046875, "grad_norm": 0.7145695686340332, "learning_rate": 0.0001, "elbo": 3.5631, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0242786769880365}
{"step": 51570, "loss": 3.563104248046875, "grad_norm": 0.7145695686340332, "learning_rate": 0.0001, "elbo": 3.5631, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0242786769880365}
{"step": 51575, "loss": 3.5773151397705076, "grad_norm": 0.6854385137557983, "learning_rate": 0.0001, "elbo": 3.57732, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.024571897724607}
{"step": 51580, "loss": 3.5856422424316405, "grad_norm": 0.6955007910728455, "learning_rate": 0.0001, "elbo": 3.58564, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0248651184611774}
{"step": 51585, "loss": 3.5761302947998046, "grad_norm": 0.6900120973587036, "learning_rate": 0.0001, "elbo": 3.57613, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.025158339197748}
{"step": 51590, "loss": 3.5870452880859376, "grad_norm": 0.7161766290664673, "learning_rate": 0.0001, "elbo": 3.58705, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0254515599343184}
{"step": 51595, "loss": 3.559768295288086, "grad_norm": 0.7012873888015747, "learning_rate": 0.0001, "elbo": 3.55977, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0257447806708893}
{"step": 51600, "loss": 3.5758399963378906, "grad_norm": 0.696540355682373, "learning_rate": 0.0001, "elbo": 3.57584, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0260380014074597}
{"step": 51605, "loss": 3.5902862548828125, "grad_norm": 0.9469594955444336, "learning_rate": 0.0001, "elbo": 3.59029, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.02633122214403}
{"step": 51610, "loss": 3.549787139892578, "grad_norm": 0.7097951173782349, "learning_rate": 0.0001, "elbo": 3.54979, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0266244428806006}
{"step": 51615, "loss": 3.569867706298828, "grad_norm": 0.676031231880188, "learning_rate": 0.0001, "elbo": 3.56987, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.026917663617171}
{"step": 51620, "loss": 3.5643989562988283, "grad_norm": 0.685954749584198, "learning_rate": 0.0001, "elbo": 3.5644, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0272108843537415}
{"step": 51625, "loss": 3.5761566162109375, "grad_norm": 0.6571409702301025, "learning_rate": 0.0001, "elbo": 3.57616, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.027504105090312}
{"step": 51625, "loss": 3.5761566162109375, "grad_norm": 0.6571409702301025, "learning_rate": 0.0001, "elbo": 3.57616, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.027504105090312}
{"step": 51630, "loss": 3.579973602294922, "grad_norm": 0.6599920988082886, "learning_rate": 0.0001, "elbo": 3.57997, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0277973258268824}
{"step": 51635, "loss": 3.5693893432617188, "grad_norm": 0.6883617043495178, "learning_rate": 0.0001, "elbo": 3.56939, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.028090546563453}
{"step": 51640, "loss": 3.5622413635253904, "grad_norm": 0.7790882587432861, "learning_rate": 0.0001, "elbo": 3.56224, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0283837673000233}
{"step": 51640, "loss": 3.5622413635253904, "grad_norm": 0.7790882587432861, "learning_rate": 0.0001, "elbo": 3.56224, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0283837673000233}
{"step": 51645, "loss": 3.5858535766601562, "grad_norm": 0.6867838501930237, "learning_rate": 0.0001, "elbo": 3.58585, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.028676988036594}
{"step": 51650, "loss": 3.5654884338378907, "grad_norm": 0.6717319488525391, "learning_rate": 0.0001, "elbo": 3.56549, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0289702087731643}
{"step": 51655, "loss": 3.5645683288574217, "grad_norm": 0.7446204423904419, "learning_rate": 0.0001, "elbo": 3.56457, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0292634295097347}
{"step": 51660, "loss": 3.5780296325683594, "grad_norm": 0.6956149935722351, "learning_rate": 0.0001, "elbo": 3.57803, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0295566502463056}
{"step": 51665, "loss": 3.5802860260009766, "grad_norm": 0.7637989521026611, "learning_rate": 0.0001, "elbo": 3.58029, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.029849870982876}
{"step": 51670, "loss": 3.578155517578125, "grad_norm": 0.6959391236305237, "learning_rate": 0.0001, "elbo": 3.57816, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0301430917194465}
{"step": 51675, "loss": 3.5675708770751955, "grad_norm": 0.7869208455085754, "learning_rate": 0.0001, "elbo": 3.56757, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.030436312456017}
{"step": 51680, "loss": 3.5796669006347654, "grad_norm": 0.7130367755889893, "learning_rate": 0.0001, "elbo": 3.57967, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0307295331925874}
{"step": 51685, "loss": 3.5948997497558595, "grad_norm": 0.6732436418533325, "learning_rate": 0.0001, "elbo": 3.5949, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.031022753929158}
{"step": 51690, "loss": 3.5887176513671877, "grad_norm": 0.6830990314483643, "learning_rate": 0.0001, "elbo": 3.58872, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0313159746657283}
{"step": 51695, "loss": 3.574761962890625, "grad_norm": 0.6999237537384033, "learning_rate": 0.0001, "elbo": 3.57476, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.031609195402299}
{"step": 51700, "loss": 3.58216552734375, "grad_norm": 0.7096747756004333, "learning_rate": 0.0001, "elbo": 3.58217, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0319024161388692}
{"step": 51705, "loss": 3.5790462493896484, "grad_norm": 0.6609540581703186, "learning_rate": 0.0001, "elbo": 3.57905, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0321956368754397}
{"step": 51710, "loss": 3.5694965362548827, "grad_norm": 0.7203763127326965, "learning_rate": 0.0001, "elbo": 3.5695, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.03248885761201}
{"step": 51715, "loss": 3.577867889404297, "grad_norm": 0.667980432510376, "learning_rate": 0.0001, "elbo": 3.57787, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0327820783485806}
{"step": 51720, "loss": 3.572828674316406, "grad_norm": 0.7865721583366394, "learning_rate": 0.0001, "elbo": 3.57283, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0330752990851515}
{"step": 51725, "loss": 3.588700866699219, "grad_norm": 0.7867954969406128, "learning_rate": 0.0001, "elbo": 3.5887, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.033368519821722}
{"step": 51730, "loss": 3.590554046630859, "grad_norm": 0.707239031791687, "learning_rate": 0.0001, "elbo": 3.59055, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0336617405582924}
{"step": 51735, "loss": 3.5945156097412108, "grad_norm": 0.6930124759674072, "learning_rate": 0.0001, "elbo": 3.59452, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.033954961294863}
{"step": 51740, "loss": 3.5638492584228514, "grad_norm": 0.7649549841880798, "learning_rate": 0.0001, "elbo": 3.56385, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0342481820314333}
{"step": 51745, "loss": 3.563854217529297, "grad_norm": 0.7242857217788696, "learning_rate": 0.0001, "elbo": 3.56385, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0345414027680038}
{"step": 51750, "loss": 3.5707740783691406, "grad_norm": 0.7440909743309021, "learning_rate": 0.0001, "elbo": 3.57077, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0348346235045742}
{"step": 51755, "loss": 3.573365020751953, "grad_norm": 0.7103097438812256, "learning_rate": 0.0001, "elbo": 3.57336, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0351278442411447}
{"step": 51760, "loss": 3.5754840850830076, "grad_norm": 0.7323938012123108, "learning_rate": 0.0001, "elbo": 3.57548, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.035421064977715}
{"step": 51765, "loss": 3.5684841156005858, "grad_norm": 0.69107985496521, "learning_rate": 0.0001, "elbo": 3.56848, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0357142857142856}
{"step": 51770, "loss": 3.5521774291992188, "grad_norm": 0.751160204410553, "learning_rate": 0.0001, "elbo": 3.55218, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.036007506450856}
{"step": 51775, "loss": 3.5789459228515623, "grad_norm": 0.7403438687324524, "learning_rate": 0.0001, "elbo": 3.57895, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0363007271874265}
{"step": 51780, "loss": 3.5800010681152346, "grad_norm": 0.744055449962616, "learning_rate": 0.0001, "elbo": 3.58, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0365939479239974}
{"step": 51785, "loss": 3.5918205261230467, "grad_norm": 0.6961032152175903, "learning_rate": 0.0001, "elbo": 3.59182, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.036887168660568}
{"step": 51790, "loss": 3.579100799560547, "grad_norm": 0.7375923991203308, "learning_rate": 0.0001, "elbo": 3.5791, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0371803893971383}
{"step": 51795, "loss": 3.583446502685547, "grad_norm": 0.7676299810409546, "learning_rate": 0.0001, "elbo": 3.58345, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0374736101337088}
{"step": 51800, "loss": 3.569356155395508, "grad_norm": 0.6610884070396423, "learning_rate": 0.0001, "elbo": 3.56936, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.037766830870279}
{"step": 51805, "loss": 3.5832481384277344, "grad_norm": 0.6614997982978821, "learning_rate": 0.0001, "elbo": 3.58325, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0380600516068497}
{"step": 51810, "loss": 3.569286346435547, "grad_norm": 0.6704238653182983, "learning_rate": 0.0001, "elbo": 3.56929, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.03835327234342}
{"step": 51815, "loss": 3.590597152709961, "grad_norm": 0.697035014629364, "learning_rate": 0.0001, "elbo": 3.5906, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0386464930799906}
{"step": 51820, "loss": 3.5805702209472656, "grad_norm": 0.7072928547859192, "learning_rate": 0.0001, "elbo": 3.58057, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.038939713816561}
{"step": 51825, "loss": 3.568524169921875, "grad_norm": 0.6966469883918762, "learning_rate": 0.0001, "elbo": 3.56852, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0392329345531315}
{"step": 51830, "loss": 3.554410934448242, "grad_norm": 0.7037851214408875, "learning_rate": 0.0001, "elbo": 3.55441, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.039526155289702}
{"step": 51835, "loss": 3.5826171875, "grad_norm": 0.8175843954086304, "learning_rate": 0.0001, "elbo": 3.58262, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0398193760262724}
{"step": 51840, "loss": 3.56864013671875, "grad_norm": 0.691085934638977, "learning_rate": 0.0001, "elbo": 3.56864, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0401125967628433}
{"step": 51845, "loss": 3.593064880371094, "grad_norm": 0.719099223613739, "learning_rate": 0.0001, "elbo": 3.59307, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0404058174994137}
{"step": 51850, "loss": 3.570001220703125, "grad_norm": 0.8798097372055054, "learning_rate": 0.0001, "elbo": 3.57, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.040699038235984}
{"step": 51855, "loss": 3.5857147216796874, "grad_norm": 0.7092792987823486, "learning_rate": 0.0001, "elbo": 3.58571, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0409922589725547}
{"step": 51860, "loss": 3.577280807495117, "grad_norm": 0.8345757126808167, "learning_rate": 0.0001, "elbo": 3.57728, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.041285479709125}
{"step": 51865, "loss": 3.5671836853027346, "grad_norm": 0.6877565383911133, "learning_rate": 0.0001, "elbo": 3.56718, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0415787004456956}
{"step": 51870, "loss": 3.5694580078125, "grad_norm": 0.6737750172615051, "learning_rate": 0.0001, "elbo": 3.56946, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.041871921182266}
{"step": 51875, "loss": 3.5771110534667967, "grad_norm": 1.065415859222412, "learning_rate": 0.0001, "elbo": 3.57711, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0421651419188365}
{"step": 51880, "loss": 3.5754661560058594, "grad_norm": 0.7357929944992065, "learning_rate": 0.0001, "elbo": 3.57547, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.042458362655407}
{"step": 51885, "loss": 3.5875785827636717, "grad_norm": 0.7185169458389282, "learning_rate": 0.0001, "elbo": 3.58758, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0427515833919774}
{"step": 51890, "loss": 3.574848937988281, "grad_norm": 0.6358203291893005, "learning_rate": 0.0001, "elbo": 3.57485, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.043044804128548}
{"step": 51895, "loss": 3.5710426330566407, "grad_norm": 0.7186561822891235, "learning_rate": 0.0001, "elbo": 3.57104, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0433380248651183}
{"step": 51900, "loss": 3.571575927734375, "grad_norm": 0.8619903922080994, "learning_rate": 0.0001, "elbo": 3.57158, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0436312456016887}
{"step": 51905, "loss": 3.57237548828125, "grad_norm": 0.7083345055580139, "learning_rate": 0.0001, "elbo": 3.57238, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0439244663382596}
{"step": 51910, "loss": 3.5674381256103516, "grad_norm": 0.6600943207740784, "learning_rate": 0.0001, "elbo": 3.56744, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.04421768707483}
{"step": 51915, "loss": 3.5762596130371094, "grad_norm": 0.699028491973877, "learning_rate": 0.0001, "elbo": 3.57626, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0445109078114005}
{"step": 51920, "loss": 3.582978439331055, "grad_norm": 0.7037166357040405, "learning_rate": 0.0001, "elbo": 3.58298, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.044804128547971}
{"step": 51925, "loss": 3.572102737426758, "grad_norm": 0.670564591884613, "learning_rate": 0.0001, "elbo": 3.5721, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0450973492845415}
{"step": 51930, "loss": 3.5822422027587892, "grad_norm": 0.681023359298706, "learning_rate": 0.0001, "elbo": 3.58224, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.045390570021112}
{"step": 51935, "loss": 3.5739479064941406, "grad_norm": 0.6679970026016235, "learning_rate": 0.0001, "elbo": 3.57395, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0456837907576824}
{"step": 51940, "loss": 3.5628841400146483, "grad_norm": 0.6915574669837952, "learning_rate": 0.0001, "elbo": 3.56288, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.045977011494253}
{"step": 51945, "loss": 3.571025848388672, "grad_norm": 0.7063184380531311, "learning_rate": 0.0001, "elbo": 3.57103, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0462702322308233}
{"step": 51950, "loss": 3.5803871154785156, "grad_norm": 0.670742928981781, "learning_rate": 0.0001, "elbo": 3.58039, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0465634529673937}
{"step": 51955, "loss": 3.574919891357422, "grad_norm": 0.712516725063324, "learning_rate": 0.0001, "elbo": 3.57492, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.046856673703964}
{"step": 51960, "loss": 3.5900238037109373, "grad_norm": 0.6731613874435425, "learning_rate": 0.0001, "elbo": 3.59002, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0471498944405346}
{"step": 51965, "loss": 3.578342056274414, "grad_norm": 0.7094711661338806, "learning_rate": 0.0001, "elbo": 3.57834, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0474431151771055}
{"step": 51970, "loss": 3.570337677001953, "grad_norm": 0.6754910945892334, "learning_rate": 0.0001, "elbo": 3.57034, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.047736335913676}
{"step": 51975, "loss": 3.5814090728759767, "grad_norm": 0.6559943556785583, "learning_rate": 0.0001, "elbo": 3.58141, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0480295566502464}
{"step": 51980, "loss": 3.5692848205566405, "grad_norm": 0.6924837827682495, "learning_rate": 0.0001, "elbo": 3.56928, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.048322777386817}
{"step": 51985, "loss": 3.5587879180908204, "grad_norm": 0.6744740605354309, "learning_rate": 0.0001, "elbo": 3.55879, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0486159981233873}
{"step": 51990, "loss": 3.5686874389648438, "grad_norm": 0.6506098508834839, "learning_rate": 0.0001, "elbo": 3.56869, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.048909218859958}
{"step": 51995, "loss": 3.570851135253906, "grad_norm": 0.6821582317352295, "learning_rate": 0.0001, "elbo": 3.57085, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0492024395965283}
{"step": 52000, "loss": 3.5763389587402346, "grad_norm": 0.7472098469734192, "learning_rate": 0.0001, "elbo": 3.57634, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0494956603330987}
{"step": 52005, "loss": 3.5666885375976562, "grad_norm": 0.6827342510223389, "learning_rate": 0.0001, "elbo": 3.56669, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.049788881069669}
{"step": 52005, "loss": 3.5666885375976562, "grad_norm": 0.6827342510223389, "learning_rate": 0.0001, "elbo": 3.56669, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.049788881069669}
{"step": 52010, "loss": 3.5767059326171875, "grad_norm": 0.6462720632553101, "learning_rate": 0.0001, "elbo": 3.57671, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0500821018062396}
{"step": 52015, "loss": 3.568006134033203, "grad_norm": 0.7362462878227234, "learning_rate": 0.0001, "elbo": 3.56801, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.05037532254281}
{"step": 52020, "loss": 3.567957305908203, "grad_norm": 0.6471887826919556, "learning_rate": 0.0001, "elbo": 3.56796, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0506685432793805}
{"step": 52025, "loss": 3.5868724822998046, "grad_norm": 0.6528435349464417, "learning_rate": 0.0001, "elbo": 3.58687, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.050961764015951}
{"step": 52030, "loss": 3.565557098388672, "grad_norm": 0.8032686710357666, "learning_rate": 0.0001, "elbo": 3.56556, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.051254984752522}
{"step": 52035, "loss": 3.569997787475586, "grad_norm": 0.7196664214134216, "learning_rate": 0.0001, "elbo": 3.57, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0515482054890923}
{"step": 52040, "loss": 3.5634700775146486, "grad_norm": 0.6898525953292847, "learning_rate": 0.0001, "elbo": 3.56347, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.051841426225663}
{"step": 52045, "loss": 3.590937042236328, "grad_norm": 0.6827253103256226, "learning_rate": 0.0001, "elbo": 3.59094, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0521346469622332}
{"step": 52050, "loss": 3.574498748779297, "grad_norm": 0.6983755230903625, "learning_rate": 0.0001, "elbo": 3.5745, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0524278676988037}
{"step": 52055, "loss": 3.5866424560546877, "grad_norm": 0.6587258577346802, "learning_rate": 0.0001, "elbo": 3.58664, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.052721088435374}
{"step": 52060, "loss": 3.5641502380371093, "grad_norm": 0.7346255779266357, "learning_rate": 0.0001, "elbo": 3.56415, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0530143091719446}
{"step": 52065, "loss": 3.5794116973876955, "grad_norm": 0.6893553137779236, "learning_rate": 0.0001, "elbo": 3.57941, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.053307529908515}
{"step": 52070, "loss": 3.580066680908203, "grad_norm": 0.6758376359939575, "learning_rate": 0.0001, "elbo": 3.58007, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0536007506450855}
{"step": 52075, "loss": 3.574176025390625, "grad_norm": 0.688237190246582, "learning_rate": 0.0001, "elbo": 3.57418, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.053893971381656}
{"step": 52080, "loss": 3.5738910675048827, "grad_norm": 0.7868034243583679, "learning_rate": 0.0001, "elbo": 3.57389, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0541871921182264}
{"step": 52085, "loss": 3.568452835083008, "grad_norm": 0.6908236742019653, "learning_rate": 0.0001, "elbo": 3.56845, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.054480412854797}
{"step": 52090, "loss": 3.5768768310546877, "grad_norm": 0.6698064804077148, "learning_rate": 0.0001, "elbo": 3.57688, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0547736335913678}
{"step": 52095, "loss": 3.562437057495117, "grad_norm": 0.6947265267372131, "learning_rate": 0.0001, "elbo": 3.56244, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0550668543279382}
{"step": 52100, "loss": 3.588938903808594, "grad_norm": 0.7308595776557922, "learning_rate": 0.0001, "elbo": 3.58894, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0553600750645087}
{"step": 52105, "loss": 3.572370147705078, "grad_norm": 0.6648061871528625, "learning_rate": 0.0001, "elbo": 3.57237, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.055653295801079}
{"step": 52110, "loss": 3.5865432739257814, "grad_norm": 0.697493314743042, "learning_rate": 0.0001, "elbo": 3.58654, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0559465165376496}
{"step": 52115, "loss": 3.578391265869141, "grad_norm": 0.7100874781608582, "learning_rate": 0.0001, "elbo": 3.57839, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.05623973727422}
{"step": 52120, "loss": 3.58282470703125, "grad_norm": 0.7025790214538574, "learning_rate": 0.0001, "elbo": 3.58282, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0565329580107905}
{"step": 52125, "loss": 3.5882823944091795, "grad_norm": 0.6709315776824951, "learning_rate": 0.0001, "elbo": 3.58828, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.056826178747361}
{"step": 52130, "loss": 3.578996276855469, "grad_norm": 0.6889882683753967, "learning_rate": 0.0001, "elbo": 3.579, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0571193994839314}
{"step": 52135, "loss": 3.587407684326172, "grad_norm": 0.683637261390686, "learning_rate": 0.0001, "elbo": 3.58741, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.057412620220502}
{"step": 52140, "loss": 3.581950378417969, "grad_norm": 0.6862596869468689, "learning_rate": 0.0001, "elbo": 3.58195, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0577058409570723}
{"step": 52145, "loss": 3.5645378112792967, "grad_norm": 0.6531350612640381, "learning_rate": 0.0001, "elbo": 3.56454, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0579990616936428}
{"step": 52150, "loss": 3.5772899627685546, "grad_norm": 0.7024397850036621, "learning_rate": 0.0001, "elbo": 3.57729, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0582922824302137}
{"step": 52155, "loss": 3.5566680908203123, "grad_norm": 0.7361093163490295, "learning_rate": 0.0001, "elbo": 3.55667, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.058585503166784}
{"step": 52160, "loss": 3.579209899902344, "grad_norm": 0.6856067180633545, "learning_rate": 0.0001, "elbo": 3.57921, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0588787239033546}
{"step": 52165, "loss": 3.5618846893310545, "grad_norm": 0.7452328205108643, "learning_rate": 0.0001, "elbo": 3.56188, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.059171944639925}
{"step": 52165, "loss": 3.5618846893310545, "grad_norm": 0.7452328205108643, "learning_rate": 0.0001, "elbo": 3.56188, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.059171944639925}
{"step": 52170, "loss": 3.5544246673583983, "grad_norm": 0.683994710445404, "learning_rate": 0.0001, "elbo": 3.55442, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0594651653764955}
{"step": 52175, "loss": 3.5573463439941406, "grad_norm": 0.6968408823013306, "learning_rate": 0.0001, "elbo": 3.55735, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.059758386113066}
{"step": 52175, "loss": 3.5573463439941406, "grad_norm": 0.6968408823013306, "learning_rate": 0.0001, "elbo": 3.55735, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.059758386113066}
{"step": 52180, "loss": 3.5653125762939455, "grad_norm": 0.6447661519050598, "learning_rate": 0.0001, "elbo": 3.56531, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0600516068496364}
{"step": 52185, "loss": 3.564093017578125, "grad_norm": 0.6880864500999451, "learning_rate": 0.0001, "elbo": 3.56409, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.060344827586207}
{"step": 52190, "loss": 3.571333312988281, "grad_norm": 0.6494358777999878, "learning_rate": 0.0001, "elbo": 3.57133, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0606380483227773}
{"step": 52195, "loss": 3.578205871582031, "grad_norm": 0.6724786758422852, "learning_rate": 0.0001, "elbo": 3.57821, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0609312690593478}
{"step": 52200, "loss": 3.5706287384033204, "grad_norm": 0.6916372179985046, "learning_rate": 0.0001, "elbo": 3.57063, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.061224489795918}
{"step": 52205, "loss": 3.5589500427246095, "grad_norm": 0.6941346526145935, "learning_rate": 0.0001, "elbo": 3.55895, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0615177105324887}
{"step": 52210, "loss": 3.5779571533203125, "grad_norm": 0.6622902750968933, "learning_rate": 0.0001, "elbo": 3.57796, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0618109312690596}
{"step": 52215, "loss": 3.5628688812255858, "grad_norm": 0.7059273719787598, "learning_rate": 0.0001, "elbo": 3.56287, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.06210415200563}
{"step": 52220, "loss": 3.570771026611328, "grad_norm": 0.7104936242103577, "learning_rate": 0.0001, "elbo": 3.57077, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0623973727422005}
{"step": 52225, "loss": 3.566513442993164, "grad_norm": 0.679249107837677, "learning_rate": 0.0001, "elbo": 3.56651, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.062690593478771}
{"step": 52230, "loss": 3.5585693359375, "grad_norm": 0.6684194803237915, "learning_rate": 0.0001, "elbo": 3.55857, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0629838142153414}
{"step": 52235, "loss": 3.5681640625, "grad_norm": 0.7502149939537048, "learning_rate": 0.0001, "elbo": 3.56816, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.063277034951912}
{"step": 52240, "loss": 3.5835498809814452, "grad_norm": 0.7056401371955872, "learning_rate": 0.0001, "elbo": 3.58355, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0635702556884823}
{"step": 52245, "loss": 3.5763523101806642, "grad_norm": 0.6664377450942993, "learning_rate": 0.0001, "elbo": 3.57635, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0638634764250527}
{"step": 52250, "loss": 3.566287612915039, "grad_norm": 0.6879344582557678, "learning_rate": 0.0001, "elbo": 3.56629, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.064156697161623}
{"step": 52255, "loss": 3.576582336425781, "grad_norm": 0.6632815003395081, "learning_rate": 0.0001, "elbo": 3.57658, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0644499178981937}
{"step": 52260, "loss": 3.57220458984375, "grad_norm": 0.672486424446106, "learning_rate": 0.0001, "elbo": 3.5722, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.064743138634764}
{"step": 52265, "loss": 3.590345764160156, "grad_norm": 0.6869048476219177, "learning_rate": 0.0001, "elbo": 3.59035, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0650363593713346}
{"step": 52270, "loss": 3.5830978393554687, "grad_norm": 0.7046788334846497, "learning_rate": 0.0001, "elbo": 3.5831, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.065329580107905}
{"step": 52275, "loss": 3.584396743774414, "grad_norm": 0.7005864381790161, "learning_rate": 0.0001, "elbo": 3.5844, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.065622800844476}
{"step": 52275, "loss": 3.584396743774414, "grad_norm": 0.7005864381790161, "learning_rate": 0.0001, "elbo": 3.5844, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.065622800844476}
{"step": 52280, "loss": 3.573835754394531, "grad_norm": 0.8473257422447205, "learning_rate": 0.0001, "elbo": 3.57384, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0659160215810464}
{"step": 52285, "loss": 3.5696983337402344, "grad_norm": 0.7505322098731995, "learning_rate": 0.0001, "elbo": 3.5697, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.066209242317617}
{"step": 52290, "loss": 3.591455841064453, "grad_norm": 0.6663299798965454, "learning_rate": 0.0001, "elbo": 3.59146, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0665024630541873}
{"step": 52295, "loss": 3.5550575256347656, "grad_norm": 0.8048795461654663, "learning_rate": 0.0001, "elbo": 3.55506, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0667956837907577}
{"step": 52300, "loss": 3.57469482421875, "grad_norm": 0.70893394947052, "learning_rate": 0.0001, "elbo": 3.57469, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.067088904527328}
{"step": 52305, "loss": 3.565167236328125, "grad_norm": 0.9119921922683716, "learning_rate": 0.0001, "elbo": 3.56517, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0673821252638986}
{"step": 52310, "loss": 3.5808868408203125, "grad_norm": 0.7323352694511414, "learning_rate": 0.0001, "elbo": 3.58089, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.067675346000469}
{"step": 52315, "loss": 3.580370330810547, "grad_norm": 0.6850178837776184, "learning_rate": 0.0001, "elbo": 3.58037, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0679685667370395}
{"step": 52320, "loss": 3.5715377807617186, "grad_norm": 0.6624252796173096, "learning_rate": 0.0001, "elbo": 3.57154, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.06826178747361}
{"step": 52325, "loss": 3.583782196044922, "grad_norm": 0.671142578125, "learning_rate": 0.0001, "elbo": 3.58378, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0685550082101805}
{"step": 52330, "loss": 3.579240417480469, "grad_norm": 0.6481896638870239, "learning_rate": 0.0001, "elbo": 3.57924, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.068848228946751}
{"step": 52330, "loss": 3.579240417480469, "grad_norm": 0.6481896638870239, "learning_rate": 0.0001, "elbo": 3.57924, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.068848228946751}
{"step": 52335, "loss": 3.5634971618652345, "grad_norm": 0.6900141835212708, "learning_rate": 0.0001, "elbo": 3.5635, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.069141449683322}
{"step": 52340, "loss": 3.567343902587891, "grad_norm": 0.9506726861000061, "learning_rate": 0.0001, "elbo": 3.56734, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0694346704198923}
{"step": 52345, "loss": 3.5750335693359374, "grad_norm": 0.7128437757492065, "learning_rate": 0.0001, "elbo": 3.57503, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0697278911564627}
{"step": 52350, "loss": 3.5883930206298826, "grad_norm": 0.6728968620300293, "learning_rate": 0.0001, "elbo": 3.58839, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.070021111893033}
{"step": 52355, "loss": 3.569243621826172, "grad_norm": 0.7226905822753906, "learning_rate": 0.0001, "elbo": 3.56924, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0703143326296036}
{"step": 52360, "loss": 3.58076171875, "grad_norm": 0.7113794684410095, "learning_rate": 0.0001, "elbo": 3.58076, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.070607553366174}
{"step": 52365, "loss": 3.591619873046875, "grad_norm": 0.6400229930877686, "learning_rate": 0.0001, "elbo": 3.59162, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0709007741027445}
{"step": 52370, "loss": 3.5763580322265627, "grad_norm": 0.7105106115341187, "learning_rate": 0.0001, "elbo": 3.57636, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.071193994839315}
{"step": 52375, "loss": 3.5743255615234375, "grad_norm": 0.6730446219444275, "learning_rate": 0.0001, "elbo": 3.57433, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0714872155758854}
{"step": 52380, "loss": 3.574229431152344, "grad_norm": 1.0415207147598267, "learning_rate": 0.0001, "elbo": 3.57423, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.071780436312456}
{"step": 52385, "loss": 3.555708312988281, "grad_norm": 0.667384684085846, "learning_rate": 0.0001, "elbo": 3.55571, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0720736570490264}
{"step": 52390, "loss": 3.574688720703125, "grad_norm": 0.7754373550415039, "learning_rate": 0.0001, "elbo": 3.57469, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.072366877785597}
{"step": 52395, "loss": 3.564710998535156, "grad_norm": 0.7279490828514099, "learning_rate": 0.0001, "elbo": 3.56471, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0726600985221673}
{"step": 52400, "loss": 3.5702178955078123, "grad_norm": 0.7394050359725952, "learning_rate": 0.0001, "elbo": 3.57022, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.072953319258738}
{"step": 52400, "loss": 3.5702178955078123, "grad_norm": 0.7394050359725952, "learning_rate": 0.0001, "elbo": 3.57022, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.072953319258738}
{"step": 52405, "loss": 3.5654701232910155, "grad_norm": 0.7506295442581177, "learning_rate": 0.0001, "elbo": 3.56547, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0732465399953086}
{"step": 52405, "loss": 3.5654701232910155, "grad_norm": 0.7506295442581177, "learning_rate": 0.0001, "elbo": 3.56547, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0732465399953086}
{"step": 52410, "loss": 3.5835227966308594, "grad_norm": 0.7046142220497131, "learning_rate": 0.0001, "elbo": 3.58352, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.073539760731879}
{"step": 52415, "loss": 3.5789764404296873, "grad_norm": 0.7800638675689697, "learning_rate": 0.0001, "elbo": 3.57898, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0738329814684495}
{"step": 52420, "loss": 3.578275680541992, "grad_norm": 0.7373795509338379, "learning_rate": 0.0001, "elbo": 3.57828, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.07412620220502}
{"step": 52425, "loss": 3.5807563781738283, "grad_norm": 0.7586402297019958, "learning_rate": 0.0001, "elbo": 3.58076, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0744194229415904}
{"step": 52430, "loss": 3.5643402099609376, "grad_norm": 0.6725637316703796, "learning_rate": 0.0001, "elbo": 3.56434, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.074712643678161}
{"step": 52435, "loss": 3.5638874053955076, "grad_norm": 0.6776469349861145, "learning_rate": 0.0001, "elbo": 3.56389, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0750058644147313}
{"step": 52440, "loss": 3.5793788909912108, "grad_norm": 0.7846952676773071, "learning_rate": 0.0001, "elbo": 3.57938, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.075299085151302}
{"step": 52445, "loss": 3.586071014404297, "grad_norm": 0.6821258664131165, "learning_rate": 0.0001, "elbo": 3.58607, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0755923058878722}
{"step": 52450, "loss": 3.5840797424316406, "grad_norm": 0.6574527025222778, "learning_rate": 0.0001, "elbo": 3.58408, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0758855266244427}
{"step": 52455, "loss": 3.5750946044921874, "grad_norm": 0.6756500601768494, "learning_rate": 0.0001, "elbo": 3.57509, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0761787473610136}
{"step": 52460, "loss": 3.5807769775390623, "grad_norm": 0.712331235408783, "learning_rate": 0.0001, "elbo": 3.58078, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.076471968097584}
{"step": 52465, "loss": 3.5725143432617186, "grad_norm": 0.6625824570655823, "learning_rate": 0.0001, "elbo": 3.57251, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0767651888341545}
{"step": 52470, "loss": 3.5679969787597656, "grad_norm": 0.8332529664039612, "learning_rate": 0.0001, "elbo": 3.568, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.077058409570725}
{"step": 52475, "loss": 3.576895904541016, "grad_norm": 0.7354783415794373, "learning_rate": 0.0001, "elbo": 3.5769, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0773516303072954}
{"step": 52480, "loss": 3.5794704437255858, "grad_norm": 0.6657166481018066, "learning_rate": 0.0001, "elbo": 3.57947, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.077644851043866}
{"step": 52485, "loss": 3.589642333984375, "grad_norm": 0.6885325312614441, "learning_rate": 0.0001, "elbo": 3.58964, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0779380717804363}
{"step": 52490, "loss": 3.58050537109375, "grad_norm": 0.6920002698898315, "learning_rate": 0.0001, "elbo": 3.58051, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.078231292517007}
{"step": 52495, "loss": 3.562085723876953, "grad_norm": 0.6862826347351074, "learning_rate": 0.0001, "elbo": 3.56209, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0785245132535772}
{"step": 52500, "loss": 3.580612564086914, "grad_norm": 0.6759748458862305, "learning_rate": 0.0001, "elbo": 3.58061, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0788177339901477}
{"step": 52505, "loss": 3.5794082641601563, "grad_norm": 0.7195490598678589, "learning_rate": 0.0001, "elbo": 3.57941, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.079110954726718}
{"step": 52510, "loss": 3.584528350830078, "grad_norm": 0.7165812849998474, "learning_rate": 0.0001, "elbo": 3.58453, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0794041754632886}
{"step": 52515, "loss": 3.575818634033203, "grad_norm": 0.7145020365715027, "learning_rate": 0.0001, "elbo": 3.57582, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.079697396199859}
{"step": 52520, "loss": 3.581524658203125, "grad_norm": 0.6645650267601013, "learning_rate": 0.0001, "elbo": 3.58152, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.07999061693643}
{"step": 52525, "loss": 3.583463668823242, "grad_norm": 0.7180963754653931, "learning_rate": 0.0001, "elbo": 3.58346, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0802838376730004}
{"step": 52530, "loss": 3.568989944458008, "grad_norm": 0.67719566822052, "learning_rate": 0.0001, "elbo": 3.56899, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.080577058409571}
{"step": 52535, "loss": 3.5933914184570312, "grad_norm": 0.7367391586303711, "learning_rate": 0.0001, "elbo": 3.59339, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0808702791461413}
{"step": 52540, "loss": 3.563389205932617, "grad_norm": 0.7372133135795593, "learning_rate": 0.0001, "elbo": 3.56339, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0811634998827118}
{"step": 52545, "loss": 3.5771152496337892, "grad_norm": 0.6776637434959412, "learning_rate": 0.0001, "elbo": 3.57712, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.081456720619282}
{"step": 52550, "loss": 3.561419677734375, "grad_norm": 0.6596894264221191, "learning_rate": 0.0001, "elbo": 3.56142, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0817499413558527}
{"step": 52555, "loss": 3.576368713378906, "grad_norm": 0.7818202972412109, "learning_rate": 0.0001, "elbo": 3.57637, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.082043162092423}
{"step": 52560, "loss": 3.5876354217529296, "grad_norm": 0.7290863394737244, "learning_rate": 0.0001, "elbo": 3.58764, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0823363828289936}
{"step": 52565, "loss": 3.5828418731689453, "grad_norm": 0.6940244436264038, "learning_rate": 0.0001, "elbo": 3.58284, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.082629603565564}
{"step": 52570, "loss": 3.5783843994140625, "grad_norm": 0.7708474397659302, "learning_rate": 0.0001, "elbo": 3.57838, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0829228243021345}
{"step": 52575, "loss": 3.578116226196289, "grad_norm": 0.9246379137039185, "learning_rate": 0.0001, "elbo": 3.57812, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.083216045038705}
{"step": 52580, "loss": 3.572210693359375, "grad_norm": 0.6590070724487305, "learning_rate": 0.0001, "elbo": 3.57221, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.083509265775276}
{"step": 52585, "loss": 3.570947265625, "grad_norm": 0.6907062530517578, "learning_rate": 0.0001, "elbo": 3.57095, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0838024865118463}
{"step": 52590, "loss": 3.5795433044433596, "grad_norm": 0.8608920574188232, "learning_rate": 0.0001, "elbo": 3.57954, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0840957072484168}
{"step": 52595, "loss": 3.5716506958007814, "grad_norm": 0.918984055519104, "learning_rate": 0.0001, "elbo": 3.57165, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.084388927984987}
{"step": 52600, "loss": 3.5791439056396483, "grad_norm": 0.6726623177528381, "learning_rate": 0.0001, "elbo": 3.57914, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0846821487215577}
{"step": 52605, "loss": 3.5806480407714845, "grad_norm": 0.7349584102630615, "learning_rate": 0.0001, "elbo": 3.58065, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.084975369458128}
{"step": 52610, "loss": 3.5823394775390627, "grad_norm": 0.6567876935005188, "learning_rate": 0.0001, "elbo": 3.58234, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0852685901946986}
{"step": 52615, "loss": 3.5771259307861327, "grad_norm": 0.6897738575935364, "learning_rate": 0.0001, "elbo": 3.57713, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.085561810931269}
{"step": 52620, "loss": 3.5673446655273438, "grad_norm": 0.7075570225715637, "learning_rate": 0.0001, "elbo": 3.56734, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0858550316678395}
{"step": 52625, "loss": 3.5802234649658202, "grad_norm": 0.6644666194915771, "learning_rate": 0.0001, "elbo": 3.58022, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.08614825240441}
{"step": 52630, "loss": 3.5628326416015623, "grad_norm": 0.6595425605773926, "learning_rate": 0.0001, "elbo": 3.56283, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0864414731409804}
{"step": 52635, "loss": 3.5987388610839846, "grad_norm": 0.7047057151794434, "learning_rate": 0.0001, "elbo": 3.59874, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.086734693877551}
{"step": 52640, "loss": 3.5857852935791015, "grad_norm": 0.7714177966117859, "learning_rate": 0.0001, "elbo": 3.58579, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0870279146141213}
{"step": 52645, "loss": 3.5677116394042967, "grad_norm": 0.7351278066635132, "learning_rate": 0.0001, "elbo": 3.56771, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.087321135350692}
{"step": 52650, "loss": 3.578858184814453, "grad_norm": 0.7209094166755676, "learning_rate": 0.0001, "elbo": 3.57886, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0876143560872626}
{"step": 52655, "loss": 3.5784461975097654, "grad_norm": 0.667221188545227, "learning_rate": 0.0001, "elbo": 3.57845, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.087907576823833}
{"step": 52660, "loss": 3.5676448822021483, "grad_norm": 0.7527750134468079, "learning_rate": 0.0001, "elbo": 3.56765, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0882007975604036}
{"step": 52665, "loss": 3.56004638671875, "grad_norm": 0.759269118309021, "learning_rate": 0.0001, "elbo": 3.56005, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.088494018296974}
{"step": 52670, "loss": 3.5645599365234375, "grad_norm": 0.687675416469574, "learning_rate": 0.0001, "elbo": 3.56456, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0887872390335445}
{"step": 52675, "loss": 3.569940185546875, "grad_norm": 0.7038063406944275, "learning_rate": 0.0001, "elbo": 3.56994, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.089080459770115}
{"step": 52680, "loss": 3.5756904602050783, "grad_norm": 0.7490256428718567, "learning_rate": 0.0001, "elbo": 3.57569, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0893736805066854}
{"step": 52685, "loss": 3.5722816467285154, "grad_norm": 0.8224132061004639, "learning_rate": 0.0001, "elbo": 3.57228, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.089666901243256}
{"step": 52690, "loss": 3.5691246032714843, "grad_norm": 0.6941092014312744, "learning_rate": 0.0001, "elbo": 3.56912, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0899601219798263}
{"step": 52695, "loss": 3.5665157318115233, "grad_norm": 0.6709944605827332, "learning_rate": 0.0001, "elbo": 3.56652, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0902533427163967}
{"step": 52700, "loss": 3.5657321929931642, "grad_norm": 0.8199750185012817, "learning_rate": 0.0001, "elbo": 3.56573, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.090546563452967}
{"step": 52705, "loss": 3.580449676513672, "grad_norm": 0.6663733124732971, "learning_rate": 0.0001, "elbo": 3.58045, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.090839784189538}
{"step": 52710, "loss": 3.558425521850586, "grad_norm": 0.6684651970863342, "learning_rate": 0.0001, "elbo": 3.55843, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0911330049261085}
{"step": 52715, "loss": 3.5798465728759767, "grad_norm": 0.6699816584587097, "learning_rate": 0.0001, "elbo": 3.57985, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.091426225662679}
{"step": 52720, "loss": 3.57379035949707, "grad_norm": 0.699176013469696, "learning_rate": 0.0001, "elbo": 3.57379, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0917194463992494}
{"step": 52725, "loss": 3.568797302246094, "grad_norm": 0.7679451107978821, "learning_rate": 0.0001, "elbo": 3.5688, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.09201266713582}
{"step": 52730, "loss": 3.572792816162109, "grad_norm": 0.7088760137557983, "learning_rate": 0.0001, "elbo": 3.57279, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0923058878723904}
{"step": 52735, "loss": 3.5593982696533204, "grad_norm": 0.7122936248779297, "learning_rate": 0.0001, "elbo": 3.5594, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.092599108608961}
{"step": 52740, "loss": 3.582345199584961, "grad_norm": 0.8285267949104309, "learning_rate": 0.0001, "elbo": 3.58235, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0928923293455313}
{"step": 52745, "loss": 3.57950439453125, "grad_norm": 0.7350279688835144, "learning_rate": 0.0001, "elbo": 3.5795, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0931855500821017}
{"step": 52750, "loss": 3.573736572265625, "grad_norm": 0.730125367641449, "learning_rate": 0.0001, "elbo": 3.57374, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.093478770818672}
{"step": 52755, "loss": 3.5860626220703127, "grad_norm": 0.7835559248924255, "learning_rate": 0.0001, "elbo": 3.58606, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0937719915552426}
{"step": 52760, "loss": 3.5784645080566406, "grad_norm": 0.7230752110481262, "learning_rate": 0.0001, "elbo": 3.57846, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.094065212291813}
{"step": 52765, "loss": 3.565590667724609, "grad_norm": 0.693067193031311, "learning_rate": 0.0001, "elbo": 3.56559, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.094358433028384}
{"step": 52770, "loss": 3.5710422515869142, "grad_norm": 0.6817396283149719, "learning_rate": 0.0001, "elbo": 3.57104, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0946516537649544}
{"step": 52775, "loss": 3.578837585449219, "grad_norm": 0.9525138735771179, "learning_rate": 0.0001, "elbo": 3.57884, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.094944874501525}
{"step": 52780, "loss": 3.5679161071777346, "grad_norm": 0.7916432023048401, "learning_rate": 0.0001, "elbo": 3.56792, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0952380952380953}
{"step": 52785, "loss": 3.576333236694336, "grad_norm": 0.7227956056594849, "learning_rate": 0.0001, "elbo": 3.57633, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.095531315974666}
{"step": 52790, "loss": 3.5624649047851564, "grad_norm": 0.695868194103241, "learning_rate": 0.0001, "elbo": 3.56246, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0958245367112363}
{"step": 52795, "loss": 3.5696155548095705, "grad_norm": 0.7560513019561768, "learning_rate": 0.0001, "elbo": 3.56962, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0961177574478067}
{"step": 52800, "loss": 3.5741836547851564, "grad_norm": 0.6722013354301453, "learning_rate": 0.0001, "elbo": 3.57418, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.096410978184377}
{"step": 52805, "loss": 3.575483703613281, "grad_norm": 0.8598158359527588, "learning_rate": 0.0001, "elbo": 3.57548, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0967041989209476}
{"step": 52810, "loss": 3.565607452392578, "grad_norm": 1.3320114612579346, "learning_rate": 0.0001, "elbo": 3.56561, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.096997419657518}
{"step": 52815, "loss": 3.5775245666503905, "grad_norm": 0.6666989922523499, "learning_rate": 0.0001, "elbo": 3.57752, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0972906403940885}
{"step": 52820, "loss": 3.5757450103759765, "grad_norm": 0.6573977470397949, "learning_rate": 0.0001, "elbo": 3.57575, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.097583861130659}
{"step": 52825, "loss": 3.5599666595458985, "grad_norm": 0.6490680575370789, "learning_rate": 0.0001, "elbo": 3.55997, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.09787708186723}
{"step": 52830, "loss": 3.565827560424805, "grad_norm": 0.6421453952789307, "learning_rate": 0.0001, "elbo": 3.56583, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0981703026038003}
{"step": 52835, "loss": 3.5662689208984375, "grad_norm": 0.7720713019371033, "learning_rate": 0.0001, "elbo": 3.56627, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.098463523340371}
{"step": 52840, "loss": 3.5660629272460938, "grad_norm": 0.6999223828315735, "learning_rate": 0.0001, "elbo": 3.56606, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0987567440769412}
{"step": 52845, "loss": 3.568361282348633, "grad_norm": 0.9545673131942749, "learning_rate": 0.0001, "elbo": 3.56836, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0990499648135117}
{"step": 52850, "loss": 3.5677558898925783, "grad_norm": 0.9048020839691162, "learning_rate": 0.0001, "elbo": 3.56776, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.099343185550082}
{"step": 52855, "loss": 3.5656192779541014, "grad_norm": 0.7279791235923767, "learning_rate": 0.0001, "elbo": 3.56562, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0996364062866526}
{"step": 52855, "loss": 3.5656192779541014, "grad_norm": 0.7279791235923767, "learning_rate": 0.0001, "elbo": 3.56562, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.0996364062866526}
{"step": 52860, "loss": 3.589855194091797, "grad_norm": 0.7233893871307373, "learning_rate": 0.0001, "elbo": 3.58986, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.099929627023223}
{"step": 52865, "loss": 3.5616661071777345, "grad_norm": 0.7136251926422119, "learning_rate": 0.0001, "elbo": 3.56167, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1002228477597935}
{"step": 52870, "loss": 3.553993988037109, "grad_norm": 0.7127516865730286, "learning_rate": 0.0001, "elbo": 3.55399, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.100516068496364}
{"step": 52875, "loss": 3.573286437988281, "grad_norm": 0.7241997718811035, "learning_rate": 0.0001, "elbo": 3.57329, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1008092892329344}
{"step": 52880, "loss": 3.576076126098633, "grad_norm": 0.6688114404678345, "learning_rate": 0.0001, "elbo": 3.57608, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.101102509969505}
{"step": 52885, "loss": 3.5763587951660156, "grad_norm": 0.7084934115409851, "learning_rate": 0.0001, "elbo": 3.57636, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1013957307060753}
{"step": 52890, "loss": 3.596417999267578, "grad_norm": 0.6729161143302917, "learning_rate": 0.0001, "elbo": 3.59642, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1016889514426462}
{"step": 52895, "loss": 3.5740127563476562, "grad_norm": 0.6986616849899292, "learning_rate": 0.0001, "elbo": 3.57401, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1019821721792167}
{"step": 52900, "loss": 3.5657249450683595, "grad_norm": 0.6486588716506958, "learning_rate": 0.0001, "elbo": 3.56573, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.102275392915787}
{"step": 52905, "loss": 3.5773162841796875, "grad_norm": 0.6824879050254822, "learning_rate": 0.0001, "elbo": 3.57732, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1025686136523576}
{"step": 52910, "loss": 3.5651885986328127, "grad_norm": 0.6558549404144287, "learning_rate": 0.0001, "elbo": 3.56519, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.102861834388928}
{"step": 52915, "loss": 3.5798568725585938, "grad_norm": 0.6824628114700317, "learning_rate": 0.0001, "elbo": 3.57986, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1031550551254985}
{"step": 52920, "loss": 3.5768203735351562, "grad_norm": 0.7848303318023682, "learning_rate": 0.0001, "elbo": 3.57682, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.103448275862069}
{"step": 52925, "loss": 3.5696647644042967, "grad_norm": 0.6621848344802856, "learning_rate": 0.0001, "elbo": 3.56966, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1037414965986394}
{"step": 52930, "loss": 3.562112045288086, "grad_norm": 0.7972036600112915, "learning_rate": 0.0001, "elbo": 3.56211, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.10403471733521}
{"step": 52935, "loss": 3.565312957763672, "grad_norm": 0.8310911655426025, "learning_rate": 0.0001, "elbo": 3.56531, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1043279380717803}
{"step": 52940, "loss": 3.5798175811767576, "grad_norm": 0.7122983932495117, "learning_rate": 0.0001, "elbo": 3.57982, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1046211588083508}
{"step": 52945, "loss": 3.5748870849609373, "grad_norm": 0.7026382684707642, "learning_rate": 0.0001, "elbo": 3.57489, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.104914379544921}
{"step": 52950, "loss": 3.5810638427734376, "grad_norm": 0.6550308465957642, "learning_rate": 0.0001, "elbo": 3.58106, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.105207600281492}
{"step": 52955, "loss": 3.5603199005126953, "grad_norm": 0.6748479008674622, "learning_rate": 0.0001, "elbo": 3.56032, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1055008210180626}
{"step": 52960, "loss": 3.5804698944091795, "grad_norm": 0.6530507802963257, "learning_rate": 0.0001, "elbo": 3.58047, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.105794041754633}
{"step": 52965, "loss": 3.572055435180664, "grad_norm": 0.7133496999740601, "learning_rate": 0.0001, "elbo": 3.57206, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1060872624912035}
{"step": 52970, "loss": 3.565761947631836, "grad_norm": 0.7073965072631836, "learning_rate": 0.0001, "elbo": 3.56576, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.106380483227774}
{"step": 52975, "loss": 3.5739917755126953, "grad_norm": 0.8864296078681946, "learning_rate": 0.0001, "elbo": 3.57399, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1066737039643444}
{"step": 52980, "loss": 3.556835174560547, "grad_norm": 0.7836724519729614, "learning_rate": 0.0001, "elbo": 3.55683, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.106966924700915}
{"step": 52985, "loss": 3.5887725830078123, "grad_norm": 0.6783468127250671, "learning_rate": 0.0001, "elbo": 3.58877, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1072601454374853}
{"step": 52990, "loss": 3.584996795654297, "grad_norm": 0.7242286801338196, "learning_rate": 0.0001, "elbo": 3.585, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1075533661740558}
{"step": 52995, "loss": 3.5717044830322267, "grad_norm": 0.8549706935882568, "learning_rate": 0.0001, "elbo": 3.5717, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.107846586910626}
{"step": 53000, "loss": 3.570222854614258, "grad_norm": 0.6726264357566833, "learning_rate": 0.0001, "elbo": 3.57022, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1081398076471967}
{"step": 53005, "loss": 3.550815963745117, "grad_norm": 0.7914369702339172, "learning_rate": 0.0001, "elbo": 3.55082, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.108433028383767}
{"step": 53010, "loss": 3.588129425048828, "grad_norm": 0.6895486116409302, "learning_rate": 0.0001, "elbo": 3.58813, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1087262491203376}
{"step": 53010, "loss": 3.588129425048828, "grad_norm": 0.6895486116409302, "learning_rate": 0.0001, "elbo": 3.58813, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1087262491203376}
{"step": 53015, "loss": 3.5777870178222657, "grad_norm": 0.7126567959785461, "learning_rate": 0.0001, "elbo": 3.57779, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1090194698569085}
{"step": 53020, "loss": 3.571277618408203, "grad_norm": 0.7380403876304626, "learning_rate": 0.0001, "elbo": 3.57128, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.109312690593479}
{"step": 53025, "loss": 3.582025909423828, "grad_norm": 0.7629966139793396, "learning_rate": 0.0001, "elbo": 3.58203, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1096059113300494}
{"step": 53030, "loss": 3.5763473510742188, "grad_norm": 0.7372886538505554, "learning_rate": 0.0001, "elbo": 3.57635, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.10989913206662}
{"step": 53035, "loss": 3.5833335876464845, "grad_norm": 0.6814829111099243, "learning_rate": 0.0001, "elbo": 3.58333, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1101923528031903}
{"step": 53040, "loss": 3.562860107421875, "grad_norm": 0.678961455821991, "learning_rate": 0.0001, "elbo": 3.56286, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1104855735397607}
{"step": 53045, "loss": 3.559539031982422, "grad_norm": 0.7358661890029907, "learning_rate": 0.0001, "elbo": 3.55954, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.110778794276331}
{"step": 53050, "loss": 3.5639442443847655, "grad_norm": 0.6732444763183594, "learning_rate": 0.0001, "elbo": 3.56394, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1110720150129016}
{"step": 53055, "loss": 3.5633544921875, "grad_norm": 0.7057403922080994, "learning_rate": 0.0001, "elbo": 3.56335, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.111365235749472}
{"step": 53060, "loss": 3.571905517578125, "grad_norm": 0.6539748907089233, "learning_rate": 0.0001, "elbo": 3.57191, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1116584564860426}
{"step": 53065, "loss": 3.5832714080810546, "grad_norm": 0.6798259615898132, "learning_rate": 0.0001, "elbo": 3.58327, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.111951677222613}
{"step": 53070, "loss": 3.571791076660156, "grad_norm": 0.6824942827224731, "learning_rate": 0.0001, "elbo": 3.57179, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1122448979591835}
{"step": 53075, "loss": 3.57388916015625, "grad_norm": 0.7011913657188416, "learning_rate": 0.0001, "elbo": 3.57389, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1125381186957544}
{"step": 53080, "loss": 3.5700984954833985, "grad_norm": 0.7055895328521729, "learning_rate": 0.0001, "elbo": 3.5701, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.112831339432325}
{"step": 53085, "loss": 3.579388427734375, "grad_norm": 0.9476108551025391, "learning_rate": 0.0001, "elbo": 3.57939, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1131245601688953}
{"step": 53090, "loss": 3.585105133056641, "grad_norm": 0.6974799036979675, "learning_rate": 0.0001, "elbo": 3.58511, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1134177809054657}
{"step": 53095, "loss": 3.575909423828125, "grad_norm": 0.6592090129852295, "learning_rate": 0.0001, "elbo": 3.57591, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.113711001642036}
{"step": 53100, "loss": 3.583722686767578, "grad_norm": 0.7823113203048706, "learning_rate": 0.0001, "elbo": 3.58372, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1140042223786066}
{"step": 53105, "loss": 3.5805007934570314, "grad_norm": 0.7374553680419922, "learning_rate": 0.0001, "elbo": 3.5805, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.114297443115177}
{"step": 53110, "loss": 3.568583679199219, "grad_norm": 0.723786473274231, "learning_rate": 0.0001, "elbo": 3.56858, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1145906638517475}
{"step": 53115, "loss": 3.583229827880859, "grad_norm": 0.6941244006156921, "learning_rate": 0.0001, "elbo": 3.58323, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.114883884588318}
{"step": 53120, "loss": 3.56900634765625, "grad_norm": 0.6943771839141846, "learning_rate": 0.0001, "elbo": 3.56901, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1151771053248885}
{"step": 53125, "loss": 3.5589572906494142, "grad_norm": 0.6871389746665955, "learning_rate": 0.0001, "elbo": 3.55896, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.115470326061459}
{"step": 53130, "loss": 3.561701202392578, "grad_norm": 0.6861976981163025, "learning_rate": 0.0001, "elbo": 3.5617, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1157635467980294}
{"step": 53135, "loss": 3.5783149719238283, "grad_norm": 0.7361125349998474, "learning_rate": 0.0001, "elbo": 3.57831, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1160567675346003}
{"step": 53140, "loss": 3.5739959716796874, "grad_norm": 0.7133641839027405, "learning_rate": 0.0001, "elbo": 3.574, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1163499882711707}
{"step": 53145, "loss": 3.5677726745605467, "grad_norm": 0.7513201236724854, "learning_rate": 0.0001, "elbo": 3.56777, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.116643209007741}
{"step": 53150, "loss": 3.5839458465576173, "grad_norm": 0.6781113147735596, "learning_rate": 0.0001, "elbo": 3.58395, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1169364297443116}
{"step": 53155, "loss": 3.5796173095703123, "grad_norm": 0.8297727704048157, "learning_rate": 0.0001, "elbo": 3.57962, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.117229650480882}
{"step": 53160, "loss": 3.5750053405761717, "grad_norm": 0.6940068602561951, "learning_rate": 0.0001, "elbo": 3.57501, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1175228712174525}
{"step": 53165, "loss": 3.571454620361328, "grad_norm": 0.7906637191772461, "learning_rate": 0.0001, "elbo": 3.57145, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.117816091954023}
{"step": 53170, "loss": 3.572947311401367, "grad_norm": 0.7300173044204712, "learning_rate": 0.0001, "elbo": 3.57295, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1181093126905934}
{"step": 53175, "loss": 3.562908935546875, "grad_norm": 0.804557204246521, "learning_rate": 0.0001, "elbo": 3.56291, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.118402533427164}
{"step": 53180, "loss": 3.560852813720703, "grad_norm": 0.7256456613540649, "learning_rate": 0.0001, "elbo": 3.56085, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1186957541637343}
{"step": 53180, "loss": 3.560852813720703, "grad_norm": 0.7256456613540649, "learning_rate": 0.0001, "elbo": 3.56085, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1186957541637343}
{"step": 53185, "loss": 3.5522632598876953, "grad_norm": 0.7025969624519348, "learning_rate": 0.0001, "elbo": 3.55226, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.118988974900305}
{"step": 53190, "loss": 3.5693737030029298, "grad_norm": 0.9434466361999512, "learning_rate": 0.0001, "elbo": 3.56937, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1192821956368753}
{"step": 53195, "loss": 3.5713951110839846, "grad_norm": 0.7127590179443359, "learning_rate": 0.0001, "elbo": 3.5714, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.119575416373446}
{"step": 53200, "loss": 3.575419616699219, "grad_norm": 0.7202595472335815, "learning_rate": 0.0001, "elbo": 3.57542, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1198686371100166}
{"step": 53205, "loss": 3.559223937988281, "grad_norm": 0.7483828067779541, "learning_rate": 0.0001, "elbo": 3.55922, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.120161857846587}
{"step": 53210, "loss": 3.5595108032226563, "grad_norm": 0.7096110582351685, "learning_rate": 0.0001, "elbo": 3.55951, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1204550785831575}
{"step": 53215, "loss": 3.5904220581054687, "grad_norm": 0.7247487306594849, "learning_rate": 0.0001, "elbo": 3.59042, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.120748299319728}
{"step": 53220, "loss": 3.5649208068847655, "grad_norm": 0.7097029089927673, "learning_rate": 0.0001, "elbo": 3.56492, "mask_frac": 0.4995, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1210415200562984}
{"step": 53225, "loss": 3.552146148681641, "grad_norm": 0.7117350101470947, "learning_rate": 0.0001, "elbo": 3.55215, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.121334740792869}
{"step": 53230, "loss": 3.5796493530273437, "grad_norm": 0.6677435040473938, "learning_rate": 0.0001, "elbo": 3.57965, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1216279615294393}
{"step": 53235, "loss": 3.573506164550781, "grad_norm": 0.8494304418563843, "learning_rate": 0.0001, "elbo": 3.57351, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.12192118226601}
{"step": 53240, "loss": 3.5832893371582033, "grad_norm": 0.6502277851104736, "learning_rate": 0.0001, "elbo": 3.58329, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1222144030025802}
{"step": 53245, "loss": 3.572147750854492, "grad_norm": 0.7067283391952515, "learning_rate": 0.0001, "elbo": 3.57215, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1225076237391507}
{"step": 53250, "loss": 3.557828903198242, "grad_norm": 0.6664783954620361, "learning_rate": 0.0001, "elbo": 3.55783, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.122800844475721}
{"step": 53255, "loss": 3.5673828125, "grad_norm": 0.6978808641433716, "learning_rate": 0.0001, "elbo": 3.56738, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1230940652122916}
{"step": 53260, "loss": 3.5540187835693358, "grad_norm": 0.6582996845245361, "learning_rate": 0.0001, "elbo": 3.55402, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1233872859488625}
{"step": 53265, "loss": 3.5600360870361327, "grad_norm": 0.6798586249351501, "learning_rate": 0.0001, "elbo": 3.56004, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.123680506685433}
{"step": 53270, "loss": 3.5726776123046875, "grad_norm": 0.6705754995346069, "learning_rate": 0.0001, "elbo": 3.57268, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1239737274220034}
{"step": 53275, "loss": 3.5751556396484374, "grad_norm": 0.7170507907867432, "learning_rate": 0.0001, "elbo": 3.57516, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.124266948158574}
{"step": 53280, "loss": 3.5664363861083985, "grad_norm": 0.6937814950942993, "learning_rate": 0.0001, "elbo": 3.56644, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1245601688951443}
{"step": 53285, "loss": 3.554343414306641, "grad_norm": 0.8608508110046387, "learning_rate": 0.0001, "elbo": 3.55434, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1248533896317148}
{"step": 53290, "loss": 3.577503967285156, "grad_norm": 0.6716479659080505, "learning_rate": 0.0001, "elbo": 3.5775, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1251466103682852}
{"step": 53295, "loss": 3.569693756103516, "grad_norm": 0.785637617111206, "learning_rate": 0.0001, "elbo": 3.56969, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1254398311048557}
{"step": 53300, "loss": 3.555582046508789, "grad_norm": 0.7162772417068481, "learning_rate": 0.0001, "elbo": 3.55558, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.125733051841426}
{"step": 53305, "loss": 3.580432891845703, "grad_norm": 0.7472555637359619, "learning_rate": 0.0001, "elbo": 3.58043, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1260262725779966}
{"step": 53310, "loss": 3.5524471282958983, "grad_norm": 0.7067493796348572, "learning_rate": 0.0001, "elbo": 3.55245, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.126319493314567}
{"step": 53315, "loss": 3.5826095581054687, "grad_norm": 0.7032702565193176, "learning_rate": 0.0001, "elbo": 3.58261, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1266127140511375}
{"step": 53320, "loss": 3.56251220703125, "grad_norm": 0.68878173828125, "learning_rate": 0.0001, "elbo": 3.56251, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1269059347877084}
{"step": 53325, "loss": 3.5809097290039062, "grad_norm": 0.7076185345649719, "learning_rate": 0.0001, "elbo": 3.58091, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.127199155524279}
{"step": 53330, "loss": 3.5515823364257812, "grad_norm": 0.7997881174087524, "learning_rate": 0.0001, "elbo": 3.55158, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1274923762608493}
{"step": 53335, "loss": 3.568406677246094, "grad_norm": 0.7407731413841248, "learning_rate": 0.0001, "elbo": 3.56841, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1277855969974198}
{"step": 53340, "loss": 3.563434600830078, "grad_norm": 0.6763028502464294, "learning_rate": 0.0001, "elbo": 3.56343, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.12807881773399}
{"step": 53340, "loss": 3.563434600830078, "grad_norm": 0.6763028502464294, "learning_rate": 0.0001, "elbo": 3.56343, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.12807881773399}
{"step": 53345, "loss": 3.5755683898925783, "grad_norm": 0.6879563331604004, "learning_rate": 0.0001, "elbo": 3.57557, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1283720384705607}
{"step": 53350, "loss": 3.5760570526123048, "grad_norm": 0.8548485040664673, "learning_rate": 0.0001, "elbo": 3.57606, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.128665259207131}
{"step": 53355, "loss": 3.572736358642578, "grad_norm": 0.6470704674720764, "learning_rate": 0.0001, "elbo": 3.57274, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1289584799437016}
{"step": 53360, "loss": 3.5719367980957033, "grad_norm": 0.693481981754303, "learning_rate": 0.0001, "elbo": 3.57194, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.129251700680272}
{"step": 53365, "loss": 3.581035614013672, "grad_norm": 0.6734133958816528, "learning_rate": 0.0001, "elbo": 3.58104, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1295449214168425}
{"step": 53370, "loss": 3.583457183837891, "grad_norm": 0.7279799580574036, "learning_rate": 0.0001, "elbo": 3.58346, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.129838142153413}
{"step": 53375, "loss": 3.5739166259765627, "grad_norm": 0.7262740731239319, "learning_rate": 0.0001, "elbo": 3.57392, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1301313628899834}
{"step": 53380, "loss": 3.575128936767578, "grad_norm": 0.7545194029808044, "learning_rate": 0.0001, "elbo": 3.57513, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.130424583626554}
{"step": 53385, "loss": 3.5604316711425783, "grad_norm": 0.7086681127548218, "learning_rate": 0.0001, "elbo": 3.56043, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1307178043631247}
{"step": 53390, "loss": 3.579396057128906, "grad_norm": 0.6802968978881836, "learning_rate": 0.0001, "elbo": 3.5794, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.131011025099695}
{"step": 53395, "loss": 3.5659812927246093, "grad_norm": 0.6980357766151428, "learning_rate": 0.0001, "elbo": 3.56598, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1313042458362657}
{"step": 53400, "loss": 3.572172927856445, "grad_norm": 0.6942346096038818, "learning_rate": 0.0001, "elbo": 3.57217, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.131597466572836}
{"step": 53405, "loss": 3.5895843505859375, "grad_norm": 0.6769207715988159, "learning_rate": 0.0001, "elbo": 3.58958, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1318906873094066}
{"step": 53410, "loss": 3.575605773925781, "grad_norm": 0.6662019491195679, "learning_rate": 0.0001, "elbo": 3.57561, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.132183908045977}
{"step": 53415, "loss": 3.572156524658203, "grad_norm": 0.6849483847618103, "learning_rate": 0.0001, "elbo": 3.57216, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1324771287825475}
{"step": 53420, "loss": 3.565983200073242, "grad_norm": 1.0977953672409058, "learning_rate": 0.0001, "elbo": 3.56598, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.132770349519118}
{"step": 53425, "loss": 3.567168426513672, "grad_norm": 0.6861910223960876, "learning_rate": 0.0001, "elbo": 3.56717, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1330635702556884}
{"step": 53430, "loss": 3.5526702880859373, "grad_norm": 0.6695674657821655, "learning_rate": 0.0001, "elbo": 3.55267, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.133356790992259}
{"step": 53435, "loss": 3.5620243072509767, "grad_norm": 0.6846078634262085, "learning_rate": 0.0001, "elbo": 3.56202, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1336500117288293}
{"step": 53440, "loss": 3.563017654418945, "grad_norm": 0.7133296728134155, "learning_rate": 0.0001, "elbo": 3.56302, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1339432324654}
{"step": 53445, "loss": 3.5841407775878906, "grad_norm": 1.0991491079330444, "learning_rate": 0.0001, "elbo": 3.58414, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1342364532019706}
{"step": 53450, "loss": 3.577983856201172, "grad_norm": 0.9536111354827881, "learning_rate": 0.0001, "elbo": 3.57798, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.134529673938541}
{"step": 53455, "loss": 3.560464859008789, "grad_norm": 0.671893835067749, "learning_rate": 0.0001, "elbo": 3.56046, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1348228946751115}
{"step": 53460, "loss": 3.573471450805664, "grad_norm": 0.7676236629486084, "learning_rate": 0.0001, "elbo": 3.57347, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.135116115411682}
{"step": 53465, "loss": 3.5762237548828124, "grad_norm": 0.679813027381897, "learning_rate": 0.0001, "elbo": 3.57622, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1354093361482525}
{"step": 53470, "loss": 3.571520233154297, "grad_norm": 0.7233260273933411, "learning_rate": 0.0001, "elbo": 3.57152, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.135702556884823}
{"step": 53475, "loss": 3.565700149536133, "grad_norm": 0.717212438583374, "learning_rate": 0.0001, "elbo": 3.5657, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1359957776213934}
{"step": 53480, "loss": 3.5719085693359376, "grad_norm": 0.7115318179130554, "learning_rate": 0.0001, "elbo": 3.57191, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.136288998357964}
{"step": 53485, "loss": 3.5747085571289063, "grad_norm": 0.7299917340278625, "learning_rate": 0.0001, "elbo": 3.57471, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1365822190945343}
{"step": 53490, "loss": 3.561726379394531, "grad_norm": 0.6800261735916138, "learning_rate": 0.0001, "elbo": 3.56173, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1368754398311047}
{"step": 53495, "loss": 3.570846176147461, "grad_norm": 0.7065168619155884, "learning_rate": 0.0001, "elbo": 3.57085, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.137168660567675}
{"step": 53500, "loss": 3.553310775756836, "grad_norm": 0.9515967965126038, "learning_rate": 0.0001, "elbo": 3.55331, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1374618813042456}
{"step": 53505, "loss": 3.579847717285156, "grad_norm": 0.7001774907112122, "learning_rate": 0.0001, "elbo": 3.57985, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.137755102040816}
{"step": 53510, "loss": 3.563790130615234, "grad_norm": 0.6763089895248413, "learning_rate": 0.0001, "elbo": 3.56379, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.138048322777387}
{"step": 53515, "loss": 3.5775508880615234, "grad_norm": 0.6548404693603516, "learning_rate": 0.0001, "elbo": 3.57755, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1383415435139574}
{"step": 53520, "loss": 3.582920455932617, "grad_norm": 0.7336520552635193, "learning_rate": 0.0001, "elbo": 3.58292, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.138634764250528}
{"step": 53525, "loss": 3.578507995605469, "grad_norm": 0.6975864768028259, "learning_rate": 0.0001, "elbo": 3.57851, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1389279849870984}
{"step": 53530, "loss": 3.5553867340087892, "grad_norm": 0.9185505509376526, "learning_rate": 0.0001, "elbo": 3.55539, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.139221205723669}
{"step": 53535, "loss": 3.5712600708007813, "grad_norm": 0.6691422462463379, "learning_rate": 0.0001, "elbo": 3.57126, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1395144264602393}
{"step": 53540, "loss": 3.583745574951172, "grad_norm": 0.685556948184967, "learning_rate": 0.0001, "elbo": 3.58375, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1398076471968097}
{"step": 53545, "loss": 3.544263458251953, "grad_norm": 0.6959215998649597, "learning_rate": 0.0001, "elbo": 3.54426, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.14010086793338}
{"step": 53550, "loss": 3.581177520751953, "grad_norm": 0.7393833994865417, "learning_rate": 0.0001, "elbo": 3.58118, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1403940886699506}
{"step": 53555, "loss": 3.5775070190429688, "grad_norm": 0.7214245796203613, "learning_rate": 0.0001, "elbo": 3.57751, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.140687309406521}
{"step": 53560, "loss": 3.5720809936523437, "grad_norm": 0.8285368084907532, "learning_rate": 0.0001, "elbo": 3.57208, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1409805301430915}
{"step": 53565, "loss": 3.5663131713867187, "grad_norm": 0.7107077836990356, "learning_rate": 0.0001, "elbo": 3.56631, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1412737508796624}
{"step": 53570, "loss": 3.5674812316894533, "grad_norm": 0.6979948878288269, "learning_rate": 0.0001, "elbo": 3.56748, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.141566971616233}
{"step": 53575, "loss": 3.5881343841552735, "grad_norm": 0.7164989709854126, "learning_rate": 0.0001, "elbo": 3.58813, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1418601923528033}
{"step": 53580, "loss": 3.57838134765625, "grad_norm": 0.7948983311653137, "learning_rate": 0.0001, "elbo": 3.57838, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.142153413089374}
{"step": 53585, "loss": 3.5642791748046876, "grad_norm": 0.7015117406845093, "learning_rate": 0.0001, "elbo": 3.56428, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1424466338259442}
{"step": 53590, "loss": 3.573382568359375, "grad_norm": 0.951022207736969, "learning_rate": 0.0001, "elbo": 3.57338, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1427398545625147}
{"step": 53595, "loss": 3.5862010955810546, "grad_norm": 0.6670653820037842, "learning_rate": 0.0001, "elbo": 3.5862, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.143033075299085}
{"step": 53600, "loss": 3.5542259216308594, "grad_norm": 0.9607940912246704, "learning_rate": 0.0001, "elbo": 3.55423, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1433262960356556}
{"step": 53605, "loss": 3.564224624633789, "grad_norm": 0.6987255811691284, "learning_rate": 0.0001, "elbo": 3.56422, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.143619516772226}
{"step": 53610, "loss": 3.559071731567383, "grad_norm": 0.718651533126831, "learning_rate": 0.0001, "elbo": 3.55907, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1439127375087965}
{"step": 53615, "loss": 3.5733783721923826, "grad_norm": 0.6970219016075134, "learning_rate": 0.0001, "elbo": 3.57338, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.144205958245367}
{"step": 53620, "loss": 3.568513107299805, "grad_norm": 0.7478666305541992, "learning_rate": 0.0001, "elbo": 3.56851, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1444991789819374}
{"step": 53625, "loss": 3.583598327636719, "grad_norm": 0.6778309345245361, "learning_rate": 0.0001, "elbo": 3.5836, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.144792399718508}
{"step": 53630, "loss": 3.571961212158203, "grad_norm": 0.9806245565414429, "learning_rate": 0.0001, "elbo": 3.57196, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.145085620455079}
{"step": 53635, "loss": 3.5777862548828123, "grad_norm": 0.676101803779602, "learning_rate": 0.0001, "elbo": 3.57779, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1453788411916492}
{"step": 53640, "loss": 3.559309387207031, "grad_norm": 0.6968616247177124, "learning_rate": 0.0001, "elbo": 3.55931, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1456720619282197}
{"step": 53645, "loss": 3.5657680511474608, "grad_norm": 0.7251642942428589, "learning_rate": 0.0001, "elbo": 3.56577, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.14596528266479}
{"step": 53650, "loss": 3.5625186920166017, "grad_norm": 0.6530556082725525, "learning_rate": 0.0001, "elbo": 3.56252, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1462585034013606}
{"step": 53655, "loss": 3.5581069946289063, "grad_norm": 0.7323484420776367, "learning_rate": 0.0001, "elbo": 3.55811, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.146551724137931}
{"step": 53660, "loss": 3.579541015625, "grad_norm": 0.7802800536155701, "learning_rate": 0.0001, "elbo": 3.57954, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1468449448745015}
{"step": 53660, "loss": 3.579541015625, "grad_norm": 0.7802800536155701, "learning_rate": 0.0001, "elbo": 3.57954, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1468449448745015}
{"step": 53665, "loss": 3.5708686828613283, "grad_norm": 0.7076047658920288, "learning_rate": 0.0001, "elbo": 3.57087, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.147138165611072}
{"step": 53670, "loss": 3.5805896759033202, "grad_norm": 0.687474250793457, "learning_rate": 0.0001, "elbo": 3.58059, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1474313863476424}
{"step": 53675, "loss": 3.580413055419922, "grad_norm": 0.6742506623268127, "learning_rate": 0.0001, "elbo": 3.58041, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.147724607084213}
{"step": 53680, "loss": 3.568172836303711, "grad_norm": 0.7030426263809204, "learning_rate": 0.0001, "elbo": 3.56817, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1480178278207833}
{"step": 53685, "loss": 3.5747600555419923, "grad_norm": 0.6606660485267639, "learning_rate": 0.0001, "elbo": 3.57476, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1483110485573538}
{"step": 53690, "loss": 3.5721694946289064, "grad_norm": 0.7113873958587646, "learning_rate": 0.0001, "elbo": 3.57217, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1486042692939247}
{"step": 53695, "loss": 3.548063278198242, "grad_norm": 0.7256361842155457, "learning_rate": 0.0001, "elbo": 3.54806, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.148897490030495}
{"step": 53700, "loss": 3.5737579345703123, "grad_norm": 0.8325251340866089, "learning_rate": 0.0001, "elbo": 3.57376, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1491907107670656}
{"step": 53705, "loss": 3.575323486328125, "grad_norm": 0.8228395581245422, "learning_rate": 0.0001, "elbo": 3.57532, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.149483931503636}
{"step": 53710, "loss": 3.560163879394531, "grad_norm": 0.700380802154541, "learning_rate": 0.0001, "elbo": 3.56016, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1497771522402065}
{"step": 53715, "loss": 3.5578163146972654, "grad_norm": 0.7302960157394409, "learning_rate": 0.0001, "elbo": 3.55782, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.150070372976777}
{"step": 53720, "loss": 3.5695003509521483, "grad_norm": 0.6615743041038513, "learning_rate": 0.0001, "elbo": 3.5695, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1503635937133474}
{"step": 53725, "loss": 3.565380096435547, "grad_norm": 0.6899813413619995, "learning_rate": 0.0001, "elbo": 3.56538, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.150656814449918}
{"step": 53730, "loss": 3.573124313354492, "grad_norm": 0.6495481729507446, "learning_rate": 0.0001, "elbo": 3.57312, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1509500351864883}
{"step": 53730, "loss": 3.573124313354492, "grad_norm": 0.6495481729507446, "learning_rate": 0.0001, "elbo": 3.57312, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1509500351864883}
{"step": 53735, "loss": 3.5632606506347657, "grad_norm": 0.7276567220687866, "learning_rate": 0.0001, "elbo": 3.56326, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1512432559230588}
{"step": 53740, "loss": 3.560626220703125, "grad_norm": 0.6927101016044617, "learning_rate": 0.0001, "elbo": 3.56063, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.151536476659629}
{"step": 53745, "loss": 3.59132080078125, "grad_norm": 0.6954818964004517, "learning_rate": 0.0001, "elbo": 3.59132, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1518296973961997}
{"step": 53750, "loss": 3.5648326873779297, "grad_norm": 0.6755163073539734, "learning_rate": 0.0001, "elbo": 3.56483, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.15212291813277}
{"step": 53755, "loss": 3.5549247741699217, "grad_norm": 0.676043689250946, "learning_rate": 0.0001, "elbo": 3.55493, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.152416138869341}
{"step": 53760, "loss": 3.5723503112792967, "grad_norm": 0.6779683232307434, "learning_rate": 0.0001, "elbo": 3.57235, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1527093596059115}
{"step": 53765, "loss": 3.564474105834961, "grad_norm": 0.6622159481048584, "learning_rate": 0.0001, "elbo": 3.56447, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.153002580342482}
{"step": 53770, "loss": 3.568428802490234, "grad_norm": 0.6965364813804626, "learning_rate": 0.0001, "elbo": 3.56843, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1532958010790524}
{"step": 53775, "loss": 3.568826675415039, "grad_norm": 0.7289210557937622, "learning_rate": 0.0001, "elbo": 3.56883, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.153589021815623}
{"step": 53780, "loss": 3.563590240478516, "grad_norm": 0.9006705284118652, "learning_rate": 0.0001, "elbo": 3.56359, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1538822425521933}
{"step": 53785, "loss": 3.574707794189453, "grad_norm": 0.7111136317253113, "learning_rate": 0.0001, "elbo": 3.57471, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1541754632887637}
{"step": 53790, "loss": 3.569289779663086, "grad_norm": 0.9393768310546875, "learning_rate": 0.0001, "elbo": 3.56929, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.154468684025334}
{"step": 53795, "loss": 3.5758220672607424, "grad_norm": 0.7408807873725891, "learning_rate": 0.0001, "elbo": 3.57582, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1547619047619047}
{"step": 53800, "loss": 3.5643074035644533, "grad_norm": 0.7312901616096497, "learning_rate": 0.0001, "elbo": 3.56431, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.155055125498475}
{"step": 53805, "loss": 3.590550994873047, "grad_norm": 0.6571970582008362, "learning_rate": 0.0001, "elbo": 3.59055, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1553483462350456}
{"step": 53810, "loss": 3.5600696563720704, "grad_norm": 1.1608806848526, "learning_rate": 0.0001, "elbo": 3.56007, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1556415669716165}
{"step": 53815, "loss": 3.5686439514160155, "grad_norm": 0.6831344366073608, "learning_rate": 0.0001, "elbo": 3.56864, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.155934787708187}
{"step": 53820, "loss": 3.5784698486328126, "grad_norm": 0.6730788946151733, "learning_rate": 0.0001, "elbo": 3.57847, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1562280084447574}
{"step": 53825, "loss": 3.580037307739258, "grad_norm": 0.717528760433197, "learning_rate": 0.0001, "elbo": 3.58004, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.156521229181328}
{"step": 53830, "loss": 3.5685218811035155, "grad_norm": 0.6916400790214539, "learning_rate": 0.0001, "elbo": 3.56852, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1568144499178983}
{"step": 53835, "loss": 3.5755855560302736, "grad_norm": 0.7772387862205505, "learning_rate": 0.0001, "elbo": 3.57559, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1571076706544687}
{"step": 53840, "loss": 3.570753479003906, "grad_norm": 0.6598213911056519, "learning_rate": 0.0001, "elbo": 3.57075, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.157400891391039}
{"step": 53845, "loss": 3.582530975341797, "grad_norm": 0.7432729005813599, "learning_rate": 0.0001, "elbo": 3.58253, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1576941121276096}
{"step": 53850, "loss": 3.5604209899902344, "grad_norm": 0.7192383408546448, "learning_rate": 0.0001, "elbo": 3.56042, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.15798733286418}
{"step": 53855, "loss": 3.5674938201904296, "grad_norm": 0.8291058540344238, "learning_rate": 0.0001, "elbo": 3.56749, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1582805536007506}
{"step": 53860, "loss": 3.582081604003906, "grad_norm": 0.7145835161209106, "learning_rate": 0.0001, "elbo": 3.58208, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.158573774337321}
{"step": 53865, "loss": 3.571844482421875, "grad_norm": 0.9032264947891235, "learning_rate": 0.0001, "elbo": 3.57184, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1588669950738915}
{"step": 53870, "loss": 3.5743202209472655, "grad_norm": 0.8294546008110046, "learning_rate": 0.0001, "elbo": 3.57432, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.159160215810462}
{"step": 53875, "loss": 3.5808006286621095, "grad_norm": 0.7167227268218994, "learning_rate": 0.0001, "elbo": 3.5808, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.159453436547033}
{"step": 53880, "loss": 3.5671066284179687, "grad_norm": 0.6970821619033813, "learning_rate": 0.0001, "elbo": 3.56711, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1597466572836033}
{"step": 53885, "loss": 3.572804641723633, "grad_norm": 0.749015212059021, "learning_rate": 0.0001, "elbo": 3.5728, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1600398780201737}
{"step": 53890, "loss": 3.5574745178222655, "grad_norm": 1.1292692422866821, "learning_rate": 0.0001, "elbo": 3.55747, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.160333098756744}
{"step": 53895, "loss": 3.595244598388672, "grad_norm": 0.6908118724822998, "learning_rate": 0.0001, "elbo": 3.59524, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1606263194933146}
{"step": 53900, "loss": 3.579656219482422, "grad_norm": 0.8648118376731873, "learning_rate": 0.0001, "elbo": 3.57966, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.160919540229885}
{"step": 53905, "loss": 3.5902332305908202, "grad_norm": 0.7398338913917542, "learning_rate": 0.0001, "elbo": 3.59023, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1612127609664555}
{"step": 53910, "loss": 3.561529541015625, "grad_norm": 0.6781489253044128, "learning_rate": 0.0001, "elbo": 3.56153, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.161505981703026}
{"step": 53910, "loss": 3.561529541015625, "grad_norm": 0.6781489253044128, "learning_rate": 0.0001, "elbo": 3.56153, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.161505981703026}
{"step": 53915, "loss": 3.578749084472656, "grad_norm": 0.6964839100837708, "learning_rate": 0.0001, "elbo": 3.57875, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1617992024395964}
{"step": 53920, "loss": 3.5631103515625, "grad_norm": 0.706264078617096, "learning_rate": 0.0001, "elbo": 3.56311, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.162092423176167}
{"step": 53925, "loss": 3.578595733642578, "grad_norm": 0.6633836030960083, "learning_rate": 0.0001, "elbo": 3.5786, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1623856439127374}
{"step": 53930, "loss": 3.5847450256347657, "grad_norm": 0.8120667934417725, "learning_rate": 0.0001, "elbo": 3.58475, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.162678864649308}
{"step": 53935, "loss": 3.5688312530517576, "grad_norm": 0.6808140277862549, "learning_rate": 0.0001, "elbo": 3.56883, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1629720853858787}
{"step": 53940, "loss": 3.563762664794922, "grad_norm": 0.6577943563461304, "learning_rate": 0.0001, "elbo": 3.56376, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.163265306122449}
{"step": 53945, "loss": 3.5729133605957033, "grad_norm": 0.6674379110336304, "learning_rate": 0.0001, "elbo": 3.57291, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1635585268590196}
{"step": 53950, "loss": 3.564597320556641, "grad_norm": 0.6568810939788818, "learning_rate": 0.0001, "elbo": 3.5646, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.16385174759559}
{"step": 53955, "loss": 3.5682308197021486, "grad_norm": 0.8519838452339172, "learning_rate": 0.0001, "elbo": 3.56823, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1641449683321605}
{"step": 53960, "loss": 3.564714813232422, "grad_norm": 0.7398799061775208, "learning_rate": 0.0001, "elbo": 3.56471, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.164438189068731}
{"step": 53965, "loss": 3.5733985900878906, "grad_norm": 0.8743413686752319, "learning_rate": 0.0001, "elbo": 3.5734, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1647314098053014}
{"step": 53970, "loss": 3.5762649536132813, "grad_norm": 0.7104215025901794, "learning_rate": 0.0001, "elbo": 3.57626, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.165024630541872}
{"step": 53975, "loss": 3.5678695678710937, "grad_norm": 0.6815334558486938, "learning_rate": 0.0001, "elbo": 3.56787, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1653178512784423}
{"step": 53980, "loss": 3.569477844238281, "grad_norm": 1.1331770420074463, "learning_rate": 0.0001, "elbo": 3.56948, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.165611072015013}
{"step": 53985, "loss": 3.5800796508789063, "grad_norm": 0.6501846313476562, "learning_rate": 0.0001, "elbo": 3.58008, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1659042927515832}
{"step": 53990, "loss": 3.5775970458984374, "grad_norm": 0.8888270258903503, "learning_rate": 0.0001, "elbo": 3.5776, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1661975134881537}
{"step": 53995, "loss": 3.589011383056641, "grad_norm": 0.6903966069221497, "learning_rate": 0.0001, "elbo": 3.58901, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.166490734224724}
{"step": 54000, "loss": 3.576161193847656, "grad_norm": 0.672920286655426, "learning_rate": 0.0001, "elbo": 3.57616, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.166783954961295}
{"step": 54005, "loss": 3.561280059814453, "grad_norm": 0.6843581795692444, "learning_rate": 0.0001, "elbo": 3.56128, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1670771756978655}
{"step": 54010, "loss": 3.5630443572998045, "grad_norm": 0.7028799653053284, "learning_rate": 0.0001, "elbo": 3.56304, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.167370396434436}
{"step": 54015, "loss": 3.5825389862060546, "grad_norm": 0.6805089712142944, "learning_rate": 0.0001, "elbo": 3.58254, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1676636171710064}
{"step": 54020, "loss": 3.576598358154297, "grad_norm": 0.6996119618415833, "learning_rate": 0.0001, "elbo": 3.5766, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.167956837907577}
{"step": 54025, "loss": 3.5589488983154296, "grad_norm": 0.6721960306167603, "learning_rate": 0.0001, "elbo": 3.55895, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1682500586441473}
{"step": 54030, "loss": 3.5640625, "grad_norm": 0.6907825469970703, "learning_rate": 0.0001, "elbo": 3.56406, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.168543279380718}
{"step": 54035, "loss": 3.5800449371337892, "grad_norm": 0.692480742931366, "learning_rate": 0.0001, "elbo": 3.58005, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1688365001172882}
{"step": 54040, "loss": 3.5828777313232423, "grad_norm": 0.6613513827323914, "learning_rate": 0.0001, "elbo": 3.58288, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1691297208538587}
{"step": 54045, "loss": 3.5669715881347654, "grad_norm": 0.6932708621025085, "learning_rate": 0.0001, "elbo": 3.56697, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.169422941590429}
{"step": 54050, "loss": 3.5646873474121095, "grad_norm": 0.6733993291854858, "learning_rate": 0.0001, "elbo": 3.56469, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1697161623269996}
{"step": 54055, "loss": 3.590024948120117, "grad_norm": 0.7273387908935547, "learning_rate": 0.0001, "elbo": 3.59002, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.17000938306357}
{"step": 54060, "loss": 3.5761157989501955, "grad_norm": 0.6918448209762573, "learning_rate": 0.0001, "elbo": 3.57612, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.170302603800141}
{"step": 54065, "loss": 3.5584922790527345, "grad_norm": 0.6686532497406006, "learning_rate": 0.0001, "elbo": 3.55849, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1705958245367114}
{"step": 54070, "loss": 3.5637054443359375, "grad_norm": 0.8380449414253235, "learning_rate": 0.0001, "elbo": 3.56371, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.170889045273282}
{"step": 54075, "loss": 3.571233367919922, "grad_norm": 0.7084761261940002, "learning_rate": 0.0001, "elbo": 3.57123, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1711822660098523}
{"step": 54080, "loss": 3.5694320678710936, "grad_norm": 0.7000414729118347, "learning_rate": 0.0001, "elbo": 3.56943, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1714754867464228}
{"step": 54085, "loss": 3.5771194458007813, "grad_norm": 0.7246248126029968, "learning_rate": 0.0001, "elbo": 3.57712, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.171768707482993}
{"step": 54090, "loss": 3.5690208435058595, "grad_norm": 0.7161228060722351, "learning_rate": 0.0001, "elbo": 3.56902, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1720619282195637}
{"step": 54095, "loss": 3.5620681762695314, "grad_norm": 0.6928377151489258, "learning_rate": 0.0001, "elbo": 3.56207, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.172355148956134}
{"step": 54100, "loss": 3.5589763641357424, "grad_norm": 0.6946984529495239, "learning_rate": 0.0001, "elbo": 3.55898, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1726483696927046}
{"step": 54105, "loss": 3.570299530029297, "grad_norm": 0.6919687390327454, "learning_rate": 0.0001, "elbo": 3.5703, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.172941590429275}
{"step": 54110, "loss": 3.5440834045410154, "grad_norm": 0.6971273422241211, "learning_rate": 0.0001, "elbo": 3.54408, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1732348111658455}
{"step": 54115, "loss": 3.5759204864501952, "grad_norm": 0.712883472442627, "learning_rate": 0.0001, "elbo": 3.57592, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.173528031902416}
{"step": 54120, "loss": 3.5533946990966796, "grad_norm": 0.7081864476203918, "learning_rate": 0.0001, "elbo": 3.55339, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1738212526389864}
{"step": 54125, "loss": 3.563810348510742, "grad_norm": 1.296557903289795, "learning_rate": 0.0001, "elbo": 3.56381, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1741144733755573}
{"step": 54130, "loss": 3.5586795806884766, "grad_norm": 0.8846720457077026, "learning_rate": 0.0001, "elbo": 3.55868, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1744076941121278}
{"step": 54135, "loss": 3.5688838958740234, "grad_norm": 0.6990101337432861, "learning_rate": 0.0001, "elbo": 3.56888, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.174700914848698}
{"step": 54140, "loss": 3.5856906890869142, "grad_norm": 0.7334867119789124, "learning_rate": 0.0001, "elbo": 3.58569, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1749941355852687}
{"step": 54145, "loss": 3.579123687744141, "grad_norm": 0.6584200859069824, "learning_rate": 0.0001, "elbo": 3.57912, "mask_frac": 0.5013, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.175287356321839}
{"step": 54150, "loss": 3.5784751892089846, "grad_norm": 0.7121255397796631, "learning_rate": 0.0001, "elbo": 3.57847, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1755805770584096}
{"step": 54155, "loss": 3.5764698028564452, "grad_norm": 0.6710751056671143, "learning_rate": 0.0001, "elbo": 3.57647, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.17587379779498}
{"step": 54160, "loss": 3.5615470886230467, "grad_norm": 0.7366690635681152, "learning_rate": 0.0001, "elbo": 3.56155, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1761670185315505}
{"step": 54165, "loss": 3.5898529052734376, "grad_norm": 0.6751739382743835, "learning_rate": 0.0001, "elbo": 3.58985, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.176460239268121}
{"step": 54170, "loss": 3.564276123046875, "grad_norm": 0.7510302662849426, "learning_rate": 0.0001, "elbo": 3.56428, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1767534600046914}
{"step": 54175, "loss": 3.5695533752441406, "grad_norm": 0.6756675839424133, "learning_rate": 0.0001, "elbo": 3.56955, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.177046680741262}
{"step": 54180, "loss": 3.566884994506836, "grad_norm": 0.7318519949913025, "learning_rate": 0.0001, "elbo": 3.56689, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1773399014778327}
{"step": 54185, "loss": 3.5735267639160155, "grad_norm": 0.695559561252594, "learning_rate": 0.0001, "elbo": 3.57353, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.177633122214403}
{"step": 54190, "loss": 3.56279296875, "grad_norm": 0.7073996663093567, "learning_rate": 0.0001, "elbo": 3.56279, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1779263429509736}
{"step": 54195, "loss": 3.5552753448486327, "grad_norm": 0.9091264605522156, "learning_rate": 0.0001, "elbo": 3.55528, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.178219563687544}
{"step": 54200, "loss": 3.5730140686035154, "grad_norm": 0.7752029895782471, "learning_rate": 0.0001, "elbo": 3.57301, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1785127844241146}
{"step": 54205, "loss": 3.5827178955078125, "grad_norm": 0.6868271827697754, "learning_rate": 0.0001, "elbo": 3.58272, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.178806005160685}
{"step": 54210, "loss": 3.5697982788085936, "grad_norm": 1.0645734071731567, "learning_rate": 0.0001, "elbo": 3.5698, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1790992258972555}
{"step": 54215, "loss": 3.568775177001953, "grad_norm": 0.6640730500221252, "learning_rate": 0.0001, "elbo": 3.56878, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.179392446633826}
{"step": 54220, "loss": 3.5663131713867187, "grad_norm": 0.677299439907074, "learning_rate": 0.0001, "elbo": 3.56631, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1796856673703964}
{"step": 54225, "loss": 3.569928741455078, "grad_norm": 1.394024133682251, "learning_rate": 0.0001, "elbo": 3.56993, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.179978888106967}
{"step": 54230, "loss": 3.5675697326660156, "grad_norm": 0.6885582804679871, "learning_rate": 0.0001, "elbo": 3.56757, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1802721088435373}
{"step": 54235, "loss": 3.5560909271240235, "grad_norm": 0.6451284289360046, "learning_rate": 0.0001, "elbo": 3.55609, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1805653295801077}
{"step": 54240, "loss": 3.5688758850097657, "grad_norm": 0.7163417339324951, "learning_rate": 0.0001, "elbo": 3.56888, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.180858550316678}
{"step": 54245, "loss": 3.570940399169922, "grad_norm": 1.042523980140686, "learning_rate": 0.0001, "elbo": 3.57094, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.181151771053249}
{"step": 54250, "loss": 3.5663238525390626, "grad_norm": 0.8345338106155396, "learning_rate": 0.0001, "elbo": 3.56632, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1814449917898195}
{"step": 54250, "loss": 3.5663238525390626, "grad_norm": 0.8345338106155396, "learning_rate": 0.0001, "elbo": 3.56632, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1814449917898195}
{"step": 54255, "loss": 3.5816665649414063, "grad_norm": 0.7440675497055054, "learning_rate": 0.0001, "elbo": 3.58167, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.18173821252639}
{"step": 54260, "loss": 3.569384765625, "grad_norm": 0.6761777400970459, "learning_rate": 0.0001, "elbo": 3.56938, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1820314332629605}
{"step": 54265, "loss": 3.5584144592285156, "grad_norm": 0.9166335463523865, "learning_rate": 0.0001, "elbo": 3.55841, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.182324653999531}
{"step": 54270, "loss": 3.5680877685546877, "grad_norm": 0.6838987469673157, "learning_rate": 0.0001, "elbo": 3.56809, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1826178747361014}
{"step": 54275, "loss": 3.5852005004882814, "grad_norm": 0.6827452182769775, "learning_rate": 0.0001, "elbo": 3.5852, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.182911095472672}
{"step": 54280, "loss": 3.5679840087890624, "grad_norm": 0.6810460686683655, "learning_rate": 0.0001, "elbo": 3.56798, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1832043162092423}
{"step": 54285, "loss": 3.571543884277344, "grad_norm": 0.6897807121276855, "learning_rate": 0.0001, "elbo": 3.57154, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1834975369458127}
{"step": 54290, "loss": 3.5624046325683594, "grad_norm": 0.7079785466194153, "learning_rate": 0.0001, "elbo": 3.5624, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.183790757682383}
{"step": 54295, "loss": 3.5663101196289064, "grad_norm": 0.8316157460212708, "learning_rate": 0.0001, "elbo": 3.56631, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1840839784189536}
{"step": 54300, "loss": 3.572321319580078, "grad_norm": 0.838802695274353, "learning_rate": 0.0001, "elbo": 3.57232, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.184377199155524}
{"step": 54305, "loss": 3.562377166748047, "grad_norm": 0.7480934858322144, "learning_rate": 0.0001, "elbo": 3.56238, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.184670419892095}
{"step": 54310, "loss": 3.5531925201416015, "grad_norm": 0.7670677900314331, "learning_rate": 0.0001, "elbo": 3.55319, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1849636406286654}
{"step": 54315, "loss": 3.5782989501953124, "grad_norm": 0.6755360960960388, "learning_rate": 0.0001, "elbo": 3.5783, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.185256861365236}
{"step": 54320, "loss": 3.554429626464844, "grad_norm": 0.7107255458831787, "learning_rate": 0.0001, "elbo": 3.55443, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1855500821018063}
{"step": 54325, "loss": 3.5654197692871095, "grad_norm": 0.7077248692512512, "learning_rate": 0.0001, "elbo": 3.56542, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.185843302838377}
{"step": 54330, "loss": 3.5727954864501954, "grad_norm": 0.6959778070449829, "learning_rate": 0.0001, "elbo": 3.5728, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1861365235749473}
{"step": 54335, "loss": 3.5442256927490234, "grad_norm": 1.06756591796875, "learning_rate": 0.0001, "elbo": 3.54423, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1864297443115177}
{"step": 54340, "loss": 3.543351745605469, "grad_norm": 0.6904060244560242, "learning_rate": 0.0001, "elbo": 3.54335, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.186722965048088}
{"step": 54345, "loss": 3.5739467620849608, "grad_norm": 0.9417255520820618, "learning_rate": 0.0001, "elbo": 3.57395, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1870161857846586}
{"step": 54350, "loss": 3.565531921386719, "grad_norm": 0.7231259942054749, "learning_rate": 0.0001, "elbo": 3.56553, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.187309406521229}
{"step": 54355, "loss": 3.571819305419922, "grad_norm": 0.6721532940864563, "learning_rate": 0.0001, "elbo": 3.57182, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1876026272577995}
{"step": 54360, "loss": 3.557560348510742, "grad_norm": 0.766523003578186, "learning_rate": 0.0001, "elbo": 3.55756, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.18789584799437}
{"step": 54365, "loss": 3.559099578857422, "grad_norm": 0.7518160343170166, "learning_rate": 0.0001, "elbo": 3.5591, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1881890687309404}
{"step": 54370, "loss": 3.5624351501464844, "grad_norm": 0.8597721457481384, "learning_rate": 0.0001, "elbo": 3.56244, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1884822894675113}
{"step": 54375, "loss": 3.550886535644531, "grad_norm": 0.703666090965271, "learning_rate": 0.0001, "elbo": 3.55089, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.188775510204082}
{"step": 54380, "loss": 3.572550964355469, "grad_norm": 0.710382342338562, "learning_rate": 0.0001, "elbo": 3.57255, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1890687309406522}
{"step": 54385, "loss": 3.567838668823242, "grad_norm": 0.6592812538146973, "learning_rate": 0.0001, "elbo": 3.56784, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1893619516772227}
{"step": 54390, "loss": 3.5581787109375, "grad_norm": 0.6570707559585571, "learning_rate": 0.0001, "elbo": 3.55818, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.189655172413793}
{"step": 54395, "loss": 3.5740383148193358, "grad_norm": 0.7168452739715576, "learning_rate": 0.0001, "elbo": 3.57404, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1899483931503636}
{"step": 54400, "loss": 3.5633621215820312, "grad_norm": 0.7177601456642151, "learning_rate": 0.0001, "elbo": 3.56336, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.190241613886934}
{"step": 54405, "loss": 3.5587268829345704, "grad_norm": 0.7448674440383911, "learning_rate": 0.0001, "elbo": 3.55873, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1905348346235045}
{"step": 54410, "loss": 3.5596542358398438, "grad_norm": 0.9410935044288635, "learning_rate": 0.0001, "elbo": 3.55965, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.190828055360075}
{"step": 54415, "loss": 3.5637042999267576, "grad_norm": 0.7393444180488586, "learning_rate": 0.0001, "elbo": 3.5637, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1911212760966454}
{"step": 54420, "loss": 3.5730354309082033, "grad_norm": 0.6877230405807495, "learning_rate": 0.0001, "elbo": 3.57304, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.191414496833216}
{"step": 54425, "loss": 3.5641159057617187, "grad_norm": 0.862925112247467, "learning_rate": 0.0001, "elbo": 3.56412, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1917077175697863}
{"step": 54430, "loss": 3.5543144226074217, "grad_norm": 0.6828035712242126, "learning_rate": 0.0001, "elbo": 3.55431, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1920009383063572}
{"step": 54435, "loss": 3.5625679016113283, "grad_norm": 0.70830237865448, "learning_rate": 0.0001, "elbo": 3.56257, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1922941590429277}
{"step": 54440, "loss": 3.55709228515625, "grad_norm": 0.8213289976119995, "learning_rate": 0.0001, "elbo": 3.55709, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.192587379779498}
{"step": 54445, "loss": 3.5647113800048826, "grad_norm": 0.6843976378440857, "learning_rate": 0.0001, "elbo": 3.56471, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1928806005160686}
{"step": 54450, "loss": 3.5675453186035155, "grad_norm": 0.6578646898269653, "learning_rate": 0.0001, "elbo": 3.56754, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.193173821252639}
{"step": 54455, "loss": 3.572398376464844, "grad_norm": 0.651325523853302, "learning_rate": 0.0001, "elbo": 3.5724, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1934670419892095}
{"step": 54460, "loss": 3.575816345214844, "grad_norm": 0.6973528265953064, "learning_rate": 0.0001, "elbo": 3.57582, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.19376026272578}
{"step": 54465, "loss": 3.563663101196289, "grad_norm": 0.6688058376312256, "learning_rate": 0.0001, "elbo": 3.56366, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1940534834623504}
{"step": 54470, "loss": 3.5626598358154298, "grad_norm": 0.746268630027771, "learning_rate": 0.0001, "elbo": 3.56266, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.194346704198921}
{"step": 54475, "loss": 3.5671760559082033, "grad_norm": 0.7323517799377441, "learning_rate": 0.0001, "elbo": 3.56718, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1946399249354913}
{"step": 54480, "loss": 3.5559364318847657, "grad_norm": 0.7152879238128662, "learning_rate": 0.0001, "elbo": 3.55594, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1949331456720618}
{"step": 54485, "loss": 3.559395599365234, "grad_norm": 0.6728652715682983, "learning_rate": 0.0001, "elbo": 3.5594, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1952263664086322}
{"step": 54490, "loss": 3.5738037109375, "grad_norm": 0.6663034558296204, "learning_rate": 0.0001, "elbo": 3.5738, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1955195871452027}
{"step": 54495, "loss": 3.572215270996094, "grad_norm": 0.9027531147003174, "learning_rate": 0.0001, "elbo": 3.57222, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1958128078817736}
{"step": 54500, "loss": 3.55981559753418, "grad_norm": 0.67169588804245, "learning_rate": 0.0001, "elbo": 3.55982, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.196106028618344}
{"step": 54505, "loss": 3.5649242401123047, "grad_norm": 0.6906611919403076, "learning_rate": 0.0001, "elbo": 3.56492, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1963992493549145}
{"step": 54510, "loss": 3.5793331146240233, "grad_norm": 0.7368524670600891, "learning_rate": 0.0001, "elbo": 3.57933, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.196692470091485}
{"step": 54515, "loss": 3.555988311767578, "grad_norm": 0.670109748840332, "learning_rate": 0.0001, "elbo": 3.55599, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1969856908280554}
{"step": 54520, "loss": 3.5741771697998046, "grad_norm": 0.7124074697494507, "learning_rate": 0.0001, "elbo": 3.57418, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.197278911564626}
{"step": 54525, "loss": 3.5671981811523437, "grad_norm": 0.6842990517616272, "learning_rate": 0.0001, "elbo": 3.5672, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1975721323011963}
{"step": 54530, "loss": 3.5662765502929688, "grad_norm": 0.7010749578475952, "learning_rate": 0.0001, "elbo": 3.56628, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1978653530377668}
{"step": 54535, "loss": 3.5626453399658202, "grad_norm": 0.6888487339019775, "learning_rate": 0.0001, "elbo": 3.56265, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.198158573774337}
{"step": 54540, "loss": 3.590106201171875, "grad_norm": 0.6732085347175598, "learning_rate": 0.0001, "elbo": 3.59011, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1984517945109077}
{"step": 54540, "loss": 3.590106201171875, "grad_norm": 0.6732085347175598, "learning_rate": 0.0001, "elbo": 3.59011, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1984517945109077}
{"step": 54545, "loss": 3.5468963623046874, "grad_norm": 0.6501871943473816, "learning_rate": 0.0001, "elbo": 3.5469, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.198745015247478}
{"step": 54550, "loss": 3.5661834716796874, "grad_norm": 0.7071883082389832, "learning_rate": 0.0001, "elbo": 3.56618, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.199038235984049}
{"step": 54550, "loss": 3.5661834716796874, "grad_norm": 0.7071883082389832, "learning_rate": 0.0001, "elbo": 3.56618, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.199038235984049}
{"step": 54555, "loss": 3.5559112548828127, "grad_norm": 0.6515109539031982, "learning_rate": 0.0001, "elbo": 3.55591, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1993314567206195}
{"step": 54560, "loss": 3.5689620971679688, "grad_norm": 0.6860781908035278, "learning_rate": 0.0001, "elbo": 3.56896, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.19962467745719}
{"step": 54565, "loss": 3.5495864868164064, "grad_norm": 0.7220248579978943, "learning_rate": 0.0001, "elbo": 3.54959, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.1999178981937604}
{"step": 54570, "loss": 3.5735153198242187, "grad_norm": 0.7038000226020813, "learning_rate": 0.0001, "elbo": 3.57352, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.200211118930331}
{"step": 54575, "loss": 3.563141632080078, "grad_norm": 0.7019408941268921, "learning_rate": 0.0001, "elbo": 3.56314, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2005043396669013}
{"step": 54580, "loss": 3.564311218261719, "grad_norm": 0.6603039503097534, "learning_rate": 0.0001, "elbo": 3.56431, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2007975604034717}
{"step": 54585, "loss": 3.5464569091796876, "grad_norm": 0.7807794809341431, "learning_rate": 0.0001, "elbo": 3.54646, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.201090781140042}
{"step": 54590, "loss": 3.571742630004883, "grad_norm": 0.6588220000267029, "learning_rate": 0.0001, "elbo": 3.57174, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2013840018766127}
{"step": 54595, "loss": 3.5575481414794923, "grad_norm": 0.6705687642097473, "learning_rate": 0.0001, "elbo": 3.55755, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.201677222613183}
{"step": 54600, "loss": 3.568011474609375, "grad_norm": 0.6818604469299316, "learning_rate": 0.0001, "elbo": 3.56801, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2019704433497536}
{"step": 54605, "loss": 3.5638202667236327, "grad_norm": 0.678922176361084, "learning_rate": 0.0001, "elbo": 3.56382, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.202263664086324}
{"step": 54610, "loss": 3.5754676818847657, "grad_norm": 0.6903141736984253, "learning_rate": 0.0001, "elbo": 3.57547, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2025568848228945}
{"step": 54615, "loss": 3.570180892944336, "grad_norm": 0.6708196401596069, "learning_rate": 0.0001, "elbo": 3.57018, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2028501055594654}
{"step": 54620, "loss": 3.5566139221191406, "grad_norm": 0.6719704866409302, "learning_rate": 0.0001, "elbo": 3.55661, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.203143326296036}
{"step": 54625, "loss": 3.563843536376953, "grad_norm": 0.7568725347518921, "learning_rate": 0.0001, "elbo": 3.56384, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2034365470326063}
{"step": 54630, "loss": 3.5603874206542967, "grad_norm": 0.7695173025131226, "learning_rate": 0.0001, "elbo": 3.56039, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2037297677691767}
{"step": 54635, "loss": 3.584478759765625, "grad_norm": 0.723599374294281, "learning_rate": 0.0001, "elbo": 3.58448, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.204022988505747}
{"step": 54640, "loss": 3.580419921875, "grad_norm": 0.7087476253509521, "learning_rate": 0.0001, "elbo": 3.58042, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2043162092423176}
{"step": 54645, "loss": 3.5843910217285155, "grad_norm": 0.7275738716125488, "learning_rate": 0.0001, "elbo": 3.58439, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.204609429978888}
{"step": 54650, "loss": 3.5668701171875, "grad_norm": 0.6776401996612549, "learning_rate": 0.0001, "elbo": 3.56687, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2049026507154585}
{"step": 54655, "loss": 3.567462158203125, "grad_norm": 0.6618093252182007, "learning_rate": 0.0001, "elbo": 3.56746, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.205195871452029}
{"step": 54660, "loss": 3.5657135009765626, "grad_norm": 0.6830075979232788, "learning_rate": 0.0001, "elbo": 3.56571, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2054890921885995}
{"step": 54665, "loss": 3.575286865234375, "grad_norm": 0.6905840039253235, "learning_rate": 0.0001, "elbo": 3.57529, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.20578231292517}
{"step": 54670, "loss": 3.5641082763671874, "grad_norm": 0.9486885666847229, "learning_rate": 0.0001, "elbo": 3.56411, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2060755336617404}
{"step": 54675, "loss": 3.5754505157470704, "grad_norm": 0.937565267086029, "learning_rate": 0.0001, "elbo": 3.57545, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2063687543983113}
{"step": 54680, "loss": 3.5700870513916017, "grad_norm": 0.755422055721283, "learning_rate": 0.0001, "elbo": 3.57009, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2066619751348817}
{"step": 54685, "loss": 3.5753494262695313, "grad_norm": 0.6924984455108643, "learning_rate": 0.0001, "elbo": 3.57535, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.206955195871452}
{"step": 54690, "loss": 3.57882080078125, "grad_norm": 0.8373575210571289, "learning_rate": 0.0001, "elbo": 3.57882, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2072484166080226}
{"step": 54695, "loss": 3.5729953765869142, "grad_norm": 0.6939065456390381, "learning_rate": 0.0001, "elbo": 3.573, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.207541637344593}
{"step": 54700, "loss": 3.5676666259765626, "grad_norm": 0.665317714214325, "learning_rate": 0.0001, "elbo": 3.56767, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2078348580811635}
{"step": 54705, "loss": 3.572500228881836, "grad_norm": 0.6781452894210815, "learning_rate": 0.0001, "elbo": 3.5725, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.208128078817734}
{"step": 54710, "loss": 3.5498477935791017, "grad_norm": 1.110099196434021, "learning_rate": 0.0001, "elbo": 3.54985, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2084212995543044}
{"step": 54715, "loss": 3.572930908203125, "grad_norm": 1.0288019180297852, "learning_rate": 0.0001, "elbo": 3.57293, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.208714520290875}
{"step": 54720, "loss": 3.5662418365478517, "grad_norm": 0.6948168873786926, "learning_rate": 0.0001, "elbo": 3.56624, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2090077410274453}
{"step": 54725, "loss": 3.5622161865234374, "grad_norm": 0.7531216144561768, "learning_rate": 0.0001, "elbo": 3.56222, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.209300961764016}
{"step": 54730, "loss": 3.5688438415527344, "grad_norm": 0.7882469296455383, "learning_rate": 0.0001, "elbo": 3.56884, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2095941825005863}
{"step": 54735, "loss": 3.557661437988281, "grad_norm": 0.7226279377937317, "learning_rate": 0.0001, "elbo": 3.55766, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2098874032371567}
{"step": 54740, "loss": 3.560072326660156, "grad_norm": 0.6898573040962219, "learning_rate": 0.0001, "elbo": 3.56007, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2101806239737276}
{"step": 54740, "loss": 3.560072326660156, "grad_norm": 0.6898573040962219, "learning_rate": 0.0001, "elbo": 3.56007, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2101806239737276}
{"step": 54745, "loss": 3.556928253173828, "grad_norm": 0.7946425676345825, "learning_rate": 0.0001, "elbo": 3.55693, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.210473844710298}
{"step": 54750, "loss": 3.55478515625, "grad_norm": 0.6404977440834045, "learning_rate": 0.0001, "elbo": 3.55479, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2107670654468685}
{"step": 54750, "loss": 3.55478515625, "grad_norm": 0.6404977440834045, "learning_rate": 0.0001, "elbo": 3.55479, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2107670654468685}
{"step": 54755, "loss": 3.5714031219482423, "grad_norm": 0.7809826731681824, "learning_rate": 0.0001, "elbo": 3.5714, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.211060286183439}
{"step": 54755, "loss": 3.5714031219482423, "grad_norm": 0.7809826731681824, "learning_rate": 0.0001, "elbo": 3.5714, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.211060286183439}
{"step": 54760, "loss": 3.570517730712891, "grad_norm": 0.696419358253479, "learning_rate": 0.0001, "elbo": 3.57052, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2113535069200094}
{"step": 54765, "loss": 3.583953094482422, "grad_norm": 0.6738449931144714, "learning_rate": 0.0001, "elbo": 3.58395, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.21164672765658}
{"step": 54770, "loss": 3.5758460998535155, "grad_norm": 0.69973224401474, "learning_rate": 0.0001, "elbo": 3.57585, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2119399483931503}
{"step": 54770, "loss": 3.5758460998535155, "grad_norm": 0.69973224401474, "learning_rate": 0.0001, "elbo": 3.57585, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2119399483931503}
{"step": 54775, "loss": 3.562185287475586, "grad_norm": 0.6839447021484375, "learning_rate": 0.0001, "elbo": 3.56219, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.212233169129721}
{"step": 54780, "loss": 3.5627941131591796, "grad_norm": 0.6707912087440491, "learning_rate": 0.0001, "elbo": 3.56279, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2125263898662912}
{"step": 54785, "loss": 3.576639175415039, "grad_norm": 0.7919597029685974, "learning_rate": 0.0001, "elbo": 3.57664, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2128196106028617}
{"step": 54790, "loss": 3.565602111816406, "grad_norm": 0.689525842666626, "learning_rate": 0.0001, "elbo": 3.5656, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.213112831339432}
{"step": 54795, "loss": 3.5707122802734377, "grad_norm": 0.6966115832328796, "learning_rate": 0.0001, "elbo": 3.57071, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.213406052076003}
{"step": 54800, "loss": 3.548895263671875, "grad_norm": 0.6925580501556396, "learning_rate": 0.0001, "elbo": 3.5489, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2136992728125735}
{"step": 54805, "loss": 3.585115432739258, "grad_norm": 0.716497004032135, "learning_rate": 0.0001, "elbo": 3.58512, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.213992493549144}
{"step": 54810, "loss": 3.5588584899902345, "grad_norm": 0.674859881401062, "learning_rate": 0.0001, "elbo": 3.55886, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2142857142857144}
{"step": 54815, "loss": 3.560752105712891, "grad_norm": 0.726249635219574, "learning_rate": 0.0001, "elbo": 3.56075, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.214578935022285}
{"step": 54820, "loss": 3.5575183868408202, "grad_norm": 0.6955384016036987, "learning_rate": 0.0001, "elbo": 3.55752, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2148721557588553}
{"step": 54825, "loss": 3.5780826568603517, "grad_norm": 0.6780775189399719, "learning_rate": 0.0001, "elbo": 3.57808, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2151653764954258}
{"step": 54830, "loss": 3.5650146484375, "grad_norm": 0.6765990257263184, "learning_rate": 0.0001, "elbo": 3.56501, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2154585972319962}
{"step": 54835, "loss": 3.5514678955078125, "grad_norm": 0.6879351139068604, "learning_rate": 0.0001, "elbo": 3.55147, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2157518179685667}
{"step": 54840, "loss": 3.5581016540527344, "grad_norm": 0.7216294407844543, "learning_rate": 0.0001, "elbo": 3.5581, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.216045038705137}
{"step": 54845, "loss": 3.5676200866699217, "grad_norm": 0.8369361758232117, "learning_rate": 0.0001, "elbo": 3.56762, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2163382594417076}
{"step": 54850, "loss": 3.5770332336425783, "grad_norm": 0.7201043367385864, "learning_rate": 0.0001, "elbo": 3.57703, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.216631480178278}
{"step": 54855, "loss": 3.5800670623779296, "grad_norm": 0.6918678283691406, "learning_rate": 0.0001, "elbo": 3.58007, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2169247009148485}
{"step": 54860, "loss": 3.5720439910888673, "grad_norm": 0.6983292698860168, "learning_rate": 0.0001, "elbo": 3.57204, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.217217921651419}
{"step": 54865, "loss": 3.569301223754883, "grad_norm": 0.7832351922988892, "learning_rate": 0.0001, "elbo": 3.5693, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.21751114238799}
{"step": 54865, "loss": 3.569301223754883, "grad_norm": 0.7832351922988892, "learning_rate": 0.0001, "elbo": 3.5693, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.21751114238799}
{"step": 54870, "loss": 3.559019088745117, "grad_norm": 0.7298718094825745, "learning_rate": 0.0001, "elbo": 3.55902, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2178043631245603}
{"step": 54875, "loss": 3.5710121154785157, "grad_norm": 0.7230970859527588, "learning_rate": 0.0001, "elbo": 3.57101, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2180975838611308}
{"step": 54880, "loss": 3.5726303100585937, "grad_norm": 0.6614026427268982, "learning_rate": 0.0001, "elbo": 3.57263, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.218390804597701}
{"step": 54885, "loss": 3.54333381652832, "grad_norm": 0.6596877574920654, "learning_rate": 0.0001, "elbo": 3.54333, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2186840253342717}
{"step": 54890, "loss": 3.5718650817871094, "grad_norm": 0.7136844992637634, "learning_rate": 0.0001, "elbo": 3.57187, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.218977246070842}
{"step": 54895, "loss": 3.563639831542969, "grad_norm": 0.7663823366165161, "learning_rate": 0.0001, "elbo": 3.56364, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2192704668074126}
{"step": 54900, "loss": 3.5704139709472655, "grad_norm": 0.692425012588501, "learning_rate": 0.0001, "elbo": 3.57041, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.219563687543983}
{"step": 54905, "loss": 3.5718852996826174, "grad_norm": 0.6887781023979187, "learning_rate": 0.0001, "elbo": 3.57189, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2198569082805535}
{"step": 54910, "loss": 3.566395568847656, "grad_norm": 0.9395390152931213, "learning_rate": 0.0001, "elbo": 3.5664, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.220150129017124}
{"step": 54915, "loss": 3.566448974609375, "grad_norm": 0.7022786736488342, "learning_rate": 0.0001, "elbo": 3.56645, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2204433497536944}
{"step": 54920, "loss": 3.5653606414794923, "grad_norm": 0.7198572158813477, "learning_rate": 0.0001, "elbo": 3.56536, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2207365704902653}
{"step": 54925, "loss": 3.5586078643798826, "grad_norm": 0.6797523498535156, "learning_rate": 0.0001, "elbo": 3.55861, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2210297912268357}
{"step": 54930, "loss": 3.5732986450195314, "grad_norm": 0.6784231066703796, "learning_rate": 0.0001, "elbo": 3.5733, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.221323011963406}
{"step": 54935, "loss": 3.575945281982422, "grad_norm": 0.6525140404701233, "learning_rate": 0.0001, "elbo": 3.57595, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2216162326999767}
{"step": 54935, "loss": 3.575945281982422, "grad_norm": 0.6525140404701233, "learning_rate": 0.0001, "elbo": 3.57595, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2216162326999767}
{"step": 54940, "loss": 3.5720703125, "grad_norm": 0.7200990915298462, "learning_rate": 0.0001, "elbo": 3.57207, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.221909453436547}
{"step": 54945, "loss": 3.56243896484375, "grad_norm": 0.6994655728340149, "learning_rate": 0.0001, "elbo": 3.56244, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2222026741731176}
{"step": 54950, "loss": 3.565174865722656, "grad_norm": 0.683003306388855, "learning_rate": 0.0001, "elbo": 3.56517, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.222495894909688}
{"step": 54955, "loss": 3.5570396423339843, "grad_norm": 0.7633432149887085, "learning_rate": 0.0001, "elbo": 3.55704, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2227891156462585}
{"step": 54955, "loss": 3.5570396423339843, "grad_norm": 0.7633432149887085, "learning_rate": 0.0001, "elbo": 3.55704, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2227891156462585}
{"step": 54960, "loss": 3.5700393676757813, "grad_norm": 0.7551867365837097, "learning_rate": 0.0001, "elbo": 3.57004, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.223082336382829}
{"step": 54965, "loss": 3.5542320251464843, "grad_norm": 0.6910257339477539, "learning_rate": 0.0001, "elbo": 3.55423, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2233755571193994}
{"step": 54970, "loss": 3.559850311279297, "grad_norm": 0.6596450805664062, "learning_rate": 0.0001, "elbo": 3.55985, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.22366877785597}
{"step": 54975, "loss": 3.5618576049804687, "grad_norm": 0.6749019622802734, "learning_rate": 0.0001, "elbo": 3.56186, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2239619985925403}
{"step": 54980, "loss": 3.5645774841308593, "grad_norm": 0.7055002450942993, "learning_rate": 0.0001, "elbo": 3.56458, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2242552193291107}
{"step": 54985, "loss": 3.563425827026367, "grad_norm": 0.6845706105232239, "learning_rate": 0.0001, "elbo": 3.56343, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2245484400656816}
{"step": 54990, "loss": 3.5604042053222655, "grad_norm": 0.6842349767684937, "learning_rate": 0.0001, "elbo": 3.5604, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.224841660802252}
{"step": 54995, "loss": 3.5656410217285157, "grad_norm": 0.7258972525596619, "learning_rate": 0.0001, "elbo": 3.56564, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2251348815388226}
{"step": 55000, "loss": 3.564325714111328, "grad_norm": 0.8820784687995911, "learning_rate": 0.0001, "elbo": 3.56433, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.225428102275393}
{"step": 55005, "loss": 3.574696350097656, "grad_norm": 0.7023350596427917, "learning_rate": 0.0001, "elbo": 3.5747, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2257213230119635}
{"step": 55010, "loss": 3.5582210540771486, "grad_norm": 0.6889705061912537, "learning_rate": 0.0001, "elbo": 3.55822, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.226014543748534}
{"step": 55015, "loss": 3.5471397399902345, "grad_norm": 0.8029380440711975, "learning_rate": 0.0001, "elbo": 3.54714, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2263077644851044}
{"step": 55020, "loss": 3.5681041717529296, "grad_norm": 0.6775016784667969, "learning_rate": 0.0001, "elbo": 3.5681, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.226600985221675}
{"step": 55025, "loss": 3.5645286560058596, "grad_norm": 0.7045935392379761, "learning_rate": 0.0001, "elbo": 3.56453, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2268942059582453}
{"step": 55030, "loss": 3.5550769805908202, "grad_norm": 0.6543006896972656, "learning_rate": 0.0001, "elbo": 3.55508, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2271874266948157}
{"step": 55035, "loss": 3.562689208984375, "grad_norm": 0.6752088665962219, "learning_rate": 0.0001, "elbo": 3.56269, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.227480647431386}
{"step": 55040, "loss": 3.573110580444336, "grad_norm": 0.6765960454940796, "learning_rate": 0.0001, "elbo": 3.57311, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2277738681679566}
{"step": 55045, "loss": 3.562641143798828, "grad_norm": 0.6739447712898254, "learning_rate": 0.0001, "elbo": 3.56264, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2280670889045275}
{"step": 55050, "loss": 3.548687744140625, "grad_norm": 0.6992753148078918, "learning_rate": 0.0001, "elbo": 3.54869, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.228360309641098}
{"step": 55055, "loss": 3.5559066772460937, "grad_norm": 0.6744359731674194, "learning_rate": 0.0001, "elbo": 3.55591, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2286535303776684}
{"step": 55060, "loss": 3.5325607299804687, "grad_norm": 0.7499207854270935, "learning_rate": 0.0001, "elbo": 3.53256, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.228946751114239}
{"step": 55065, "loss": 3.55407829284668, "grad_norm": 0.7828043699264526, "learning_rate": 0.0001, "elbo": 3.55408, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2292399718508094}
{"step": 55070, "loss": 3.5590023040771483, "grad_norm": 0.747745156288147, "learning_rate": 0.0001, "elbo": 3.559, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.22953319258738}
{"step": 55075, "loss": 3.5797752380371093, "grad_norm": 0.6984683871269226, "learning_rate": 0.0001, "elbo": 3.57978, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2298264133239503}
{"step": 55080, "loss": 3.566606903076172, "grad_norm": 0.6868270039558411, "learning_rate": 0.0001, "elbo": 3.56661, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2301196340605207}
{"step": 55085, "loss": 3.566607666015625, "grad_norm": 0.7229235172271729, "learning_rate": 0.0001, "elbo": 3.56661, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.230412854797091}
{"step": 55090, "loss": 3.572348403930664, "grad_norm": 0.6706461906433105, "learning_rate": 0.0001, "elbo": 3.57235, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2307060755336616}
{"step": 55095, "loss": 3.5662574768066406, "grad_norm": 0.6808755993843079, "learning_rate": 0.0001, "elbo": 3.56626, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.230999296270232}
{"step": 55100, "loss": 3.5619720458984374, "grad_norm": 0.6789929866790771, "learning_rate": 0.0001, "elbo": 3.56197, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2312925170068025}
{"step": 55105, "loss": 3.5670143127441407, "grad_norm": 0.7009625434875488, "learning_rate": 0.0001, "elbo": 3.56701, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.231585737743373}
{"step": 55110, "loss": 3.562788391113281, "grad_norm": 0.7364581823348999, "learning_rate": 0.0001, "elbo": 3.56279, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.231878958479944}
{"step": 55115, "loss": 3.5661731719970704, "grad_norm": 0.8301020860671997, "learning_rate": 0.0001, "elbo": 3.56617, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2321721792165143}
{"step": 55120, "loss": 3.5592769622802733, "grad_norm": 0.6920110583305359, "learning_rate": 0.0001, "elbo": 3.55928, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.232465399953085}
{"step": 55125, "loss": 3.5731731414794923, "grad_norm": 0.7112056612968445, "learning_rate": 0.0001, "elbo": 3.57317, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2327586206896552}
{"step": 55130, "loss": 3.5659912109375, "grad_norm": 0.6763913631439209, "learning_rate": 0.0001, "elbo": 3.56599, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2330518414262257}
{"step": 55135, "loss": 3.568832015991211, "grad_norm": 0.7591719031333923, "learning_rate": 0.0001, "elbo": 3.56883, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.233345062162796}
{"step": 55140, "loss": 3.5743480682373048, "grad_norm": 0.6702461838722229, "learning_rate": 0.0001, "elbo": 3.57435, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2336382828993666}
{"step": 55145, "loss": 3.548134613037109, "grad_norm": 0.7142418026924133, "learning_rate": 0.0001, "elbo": 3.54813, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.233931503635937}
{"step": 55150, "loss": 3.560919189453125, "grad_norm": 0.6795613169670105, "learning_rate": 0.0001, "elbo": 3.56092, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2342247243725075}
{"step": 55155, "loss": 3.5745170593261717, "grad_norm": 0.7229123711585999, "learning_rate": 0.0001, "elbo": 3.57452, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.234517945109078}
{"step": 55160, "loss": 3.5558982849121095, "grad_norm": 0.6675318479537964, "learning_rate": 0.0001, "elbo": 3.5559, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2348111658456484}
{"step": 55165, "loss": 3.5715274810791016, "grad_norm": 0.6827724575996399, "learning_rate": 0.0001, "elbo": 3.57153, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2351043865822193}
{"step": 55170, "loss": 3.5485580444335936, "grad_norm": 0.7419335246086121, "learning_rate": 0.0001, "elbo": 3.54856, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.23539760731879}
{"step": 55175, "loss": 3.5593677520751954, "grad_norm": 0.8037093877792358, "learning_rate": 0.0001, "elbo": 3.55937, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2356908280553602}
{"step": 55180, "loss": 3.569805145263672, "grad_norm": 0.7456163167953491, "learning_rate": 0.0001, "elbo": 3.56981, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2359840487919307}
{"step": 55185, "loss": 3.5630867004394533, "grad_norm": 0.6478729844093323, "learning_rate": 0.0001, "elbo": 3.56309, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.236277269528501}
{"step": 55190, "loss": 3.5518836975097656, "grad_norm": 0.6656932234764099, "learning_rate": 0.0001, "elbo": 3.55188, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2365704902650716}
{"step": 55195, "loss": 3.5616355895996095, "grad_norm": 0.6703127026557922, "learning_rate": 0.0001, "elbo": 3.56164, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.236863711001642}
{"step": 55200, "loss": 3.5749160766601564, "grad_norm": 0.7331976890563965, "learning_rate": 0.0001, "elbo": 3.57492, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2371569317382125}
{"step": 55205, "loss": 3.5486595153808596, "grad_norm": 0.7177446484565735, "learning_rate": 0.0001, "elbo": 3.54866, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.237450152474783}
{"step": 55210, "loss": 3.547682189941406, "grad_norm": 0.7499780058860779, "learning_rate": 0.0001, "elbo": 3.54768, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2377433732113534}
{"step": 55215, "loss": 3.582794952392578, "grad_norm": 1.0607415437698364, "learning_rate": 0.0001, "elbo": 3.58279, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.238036593947924}
{"step": 55220, "loss": 3.5657188415527346, "grad_norm": 0.7029023766517639, "learning_rate": 0.0001, "elbo": 3.56572, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2383298146844943}
{"step": 55225, "loss": 3.5766712188720704, "grad_norm": 0.6551176905632019, "learning_rate": 0.0001, "elbo": 3.57667, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2386230354210648}
{"step": 55230, "loss": 3.5747749328613283, "grad_norm": 0.7170352935791016, "learning_rate": 0.0001, "elbo": 3.57477, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2389162561576357}
{"step": 55235, "loss": 3.5650680541992186, "grad_norm": 0.7010229825973511, "learning_rate": 0.0001, "elbo": 3.56507, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.239209476894206}
{"step": 55240, "loss": 3.5717498779296877, "grad_norm": 0.7709454298019409, "learning_rate": 0.0001, "elbo": 3.57175, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2395026976307766}
{"step": 55245, "loss": 3.570317840576172, "grad_norm": 0.7523738145828247, "learning_rate": 0.0001, "elbo": 3.57032, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.239795918367347}
{"step": 55250, "loss": 3.5685806274414062, "grad_norm": 0.6893542408943176, "learning_rate": 0.0001, "elbo": 3.56858, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2400891391039175}
{"step": 55255, "loss": 3.557337188720703, "grad_norm": 0.6785489916801453, "learning_rate": 0.0001, "elbo": 3.55734, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.240382359840488}
{"step": 55260, "loss": 3.5556396484375, "grad_norm": 0.7565011382102966, "learning_rate": 0.0001, "elbo": 3.55564, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2406755805770584}
{"step": 55265, "loss": 3.55894775390625, "grad_norm": 0.6626760363578796, "learning_rate": 0.0001, "elbo": 3.55895, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.240968801313629}
{"step": 55270, "loss": 3.5576725006103516, "grad_norm": 0.7170599102973938, "learning_rate": 0.0001, "elbo": 3.55767, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2412620220501993}
{"step": 55275, "loss": 3.566774368286133, "grad_norm": 0.9912078976631165, "learning_rate": 0.0001, "elbo": 3.56677, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2415552427867698}
{"step": 55280, "loss": 3.587222671508789, "grad_norm": 0.6961654424667358, "learning_rate": 0.0001, "elbo": 3.58722, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.24184846352334}
{"step": 55285, "loss": 3.557794952392578, "grad_norm": 0.6522783637046814, "learning_rate": 0.0001, "elbo": 3.5578, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2421416842599107}
{"step": 55290, "loss": 3.5734375, "grad_norm": 1.1790111064910889, "learning_rate": 0.0001, "elbo": 3.57344, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2424349049964816}
{"step": 55295, "loss": 3.5765060424804687, "grad_norm": 0.7328315377235413, "learning_rate": 0.0001, "elbo": 3.57651, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.242728125733052}
{"step": 55300, "loss": 3.555175018310547, "grad_norm": 0.6813474297523499, "learning_rate": 0.0001, "elbo": 3.55517, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2430213464696225}
{"step": 55305, "loss": 3.5756607055664062, "grad_norm": 0.7439313530921936, "learning_rate": 0.0001, "elbo": 3.57566, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.243314567206193}
{"step": 55310, "loss": 3.5750579833984375, "grad_norm": 0.7029082775115967, "learning_rate": 0.0001, "elbo": 3.57506, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2436077879427634}
{"step": 55315, "loss": 3.5674957275390624, "grad_norm": 1.0700874328613281, "learning_rate": 0.0001, "elbo": 3.5675, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.243901008679334}
{"step": 55320, "loss": 3.555799102783203, "grad_norm": 0.7435696125030518, "learning_rate": 0.0001, "elbo": 3.5558, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2441942294159043}
{"step": 55325, "loss": 3.561296844482422, "grad_norm": 0.766310453414917, "learning_rate": 0.0001, "elbo": 3.5613, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2444874501524748}
{"step": 55330, "loss": 3.568348693847656, "grad_norm": 0.7867034077644348, "learning_rate": 0.0001, "elbo": 3.56835, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.244780670889045}
{"step": 55335, "loss": 3.5532020568847655, "grad_norm": 0.7596644163131714, "learning_rate": 0.0001, "elbo": 3.5532, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2450738916256157}
{"step": 55340, "loss": 3.574937438964844, "grad_norm": 0.6892598867416382, "learning_rate": 0.0001, "elbo": 3.57494, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.245367112362186}
{"step": 55345, "loss": 3.565296936035156, "grad_norm": 0.6666233539581299, "learning_rate": 0.0001, "elbo": 3.5653, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2456603330987566}
{"step": 55350, "loss": 3.547595977783203, "grad_norm": 0.7152532339096069, "learning_rate": 0.0001, "elbo": 3.5476, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.245953553835327}
{"step": 55355, "loss": 3.5611927032470705, "grad_norm": 0.7279210686683655, "learning_rate": 0.0001, "elbo": 3.56119, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.246246774571898}
{"step": 55360, "loss": 3.5729366302490235, "grad_norm": 0.8245474100112915, "learning_rate": 0.0001, "elbo": 3.57294, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2465399953084684}
{"step": 55365, "loss": 3.5694015502929686, "grad_norm": 0.6715060472488403, "learning_rate": 0.0001, "elbo": 3.5694, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.246833216045039}
{"step": 55370, "loss": 3.5690174102783203, "grad_norm": 0.8620275855064392, "learning_rate": 0.0001, "elbo": 3.56902, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2471264367816093}
{"step": 55375, "loss": 3.577708053588867, "grad_norm": 0.6904018521308899, "learning_rate": 0.0001, "elbo": 3.57771, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2474196575181797}
{"step": 55380, "loss": 3.5746490478515627, "grad_norm": 0.7063127160072327, "learning_rate": 0.0001, "elbo": 3.57465, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.24771287825475}
{"step": 55385, "loss": 3.5576366424560546, "grad_norm": 0.6530652642250061, "learning_rate": 0.0001, "elbo": 3.55764, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2480060989913206}
{"step": 55390, "loss": 3.564014434814453, "grad_norm": 0.6578876972198486, "learning_rate": 0.0001, "elbo": 3.56401, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.248299319727891}
{"step": 55395, "loss": 3.570466232299805, "grad_norm": 0.751534640789032, "learning_rate": 0.0001, "elbo": 3.57047, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2485925404644616}
{"step": 55400, "loss": 3.553972625732422, "grad_norm": 0.6794927716255188, "learning_rate": 0.0001, "elbo": 3.55397, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.248885761201032}
{"step": 55405, "loss": 3.5581809997558596, "grad_norm": 0.8387231230735779, "learning_rate": 0.0001, "elbo": 3.55818, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2491789819376025}
{"step": 55410, "loss": 3.574207305908203, "grad_norm": 0.7092329263687134, "learning_rate": 0.0001, "elbo": 3.57421, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.249472202674173}
{"step": 55415, "loss": 3.565663146972656, "grad_norm": 0.6931992173194885, "learning_rate": 0.0001, "elbo": 3.56566, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.249765423410744}
{"step": 55420, "loss": 3.5645816802978514, "grad_norm": 0.7114363312721252, "learning_rate": 0.0001, "elbo": 3.56458, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2500586441473143}
{"step": 55425, "loss": 3.556426239013672, "grad_norm": 0.672355592250824, "learning_rate": 0.0001, "elbo": 3.55643, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2503518648838847}
{"step": 55430, "loss": 3.5643409729003905, "grad_norm": 0.7823783159255981, "learning_rate": 0.0001, "elbo": 3.56434, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.250645085620455}
{"step": 55435, "loss": 3.563372802734375, "grad_norm": 0.6547651886940002, "learning_rate": 0.0001, "elbo": 3.56337, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2509383063570256}
{"step": 55440, "loss": 3.561266326904297, "grad_norm": 0.6970401406288147, "learning_rate": 0.0001, "elbo": 3.56127, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.251231527093596}
{"step": 55445, "loss": 3.556958770751953, "grad_norm": 0.8198404908180237, "learning_rate": 0.0001, "elbo": 3.55696, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2515247478301665}
{"step": 55450, "loss": 3.5663928985595703, "grad_norm": 0.7111942768096924, "learning_rate": 0.0001, "elbo": 3.56639, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.251817968566737}
{"step": 55455, "loss": 3.5528594970703127, "grad_norm": 0.6959227919578552, "learning_rate": 0.0001, "elbo": 3.55286, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2521111893033074}
{"step": 55460, "loss": 3.5570652008056642, "grad_norm": 0.7878590822219849, "learning_rate": 0.0001, "elbo": 3.55707, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.252404410039878}
{"step": 55465, "loss": 3.5603851318359374, "grad_norm": 0.7562596797943115, "learning_rate": 0.0001, "elbo": 3.56039, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2526976307764484}
{"step": 55465, "loss": 3.5603851318359374, "grad_norm": 0.7562596797943115, "learning_rate": 0.0001, "elbo": 3.56039, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2526976307764484}
{"step": 55470, "loss": 3.571807861328125, "grad_norm": 0.8086750507354736, "learning_rate": 0.0001, "elbo": 3.57181, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.252990851513019}
{"step": 55475, "loss": 3.570603942871094, "grad_norm": 0.7153914570808411, "learning_rate": 0.0001, "elbo": 3.5706, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2532840722495893}
{"step": 55480, "loss": 3.564221954345703, "grad_norm": 0.6699938774108887, "learning_rate": 0.0001, "elbo": 3.56422, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.25357729298616}
{"step": 55485, "loss": 3.557996368408203, "grad_norm": 0.7880168557167053, "learning_rate": 0.0001, "elbo": 3.558, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2538705137227306}
{"step": 55490, "loss": 3.548261260986328, "grad_norm": 0.7052163481712341, "learning_rate": 0.0001, "elbo": 3.54826, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.254163734459301}
{"step": 55495, "loss": 3.5478446960449217, "grad_norm": 0.7090892195701599, "learning_rate": 0.0001, "elbo": 3.54784, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2544569551958715}
{"step": 55500, "loss": 3.559463882446289, "grad_norm": 0.8395180702209473, "learning_rate": 0.0001, "elbo": 3.55946, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.254750175932442}
{"step": 55505, "loss": 3.5763168334960938, "grad_norm": 0.6574802398681641, "learning_rate": 0.0001, "elbo": 3.57632, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2550433966690124}
{"step": 55510, "loss": 3.5576324462890625, "grad_norm": 0.6904844045639038, "learning_rate": 0.0001, "elbo": 3.55763, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.255336617405583}
{"step": 55515, "loss": 3.555852508544922, "grad_norm": 0.7568352818489075, "learning_rate": 0.0001, "elbo": 3.55585, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2556298381421533}
{"step": 55520, "loss": 3.580645751953125, "grad_norm": 0.7151001691818237, "learning_rate": 0.0001, "elbo": 3.58065, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.255923058878724}
{"step": 55525, "loss": 3.5658836364746094, "grad_norm": 1.0740894079208374, "learning_rate": 0.0001, "elbo": 3.56588, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2562162796152943}
{"step": 55530, "loss": 3.566048812866211, "grad_norm": 0.9212937951087952, "learning_rate": 0.0001, "elbo": 3.56605, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2565095003518647}
{"step": 55535, "loss": 3.570448303222656, "grad_norm": 1.1755015850067139, "learning_rate": 0.0001, "elbo": 3.57045, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2568027210884356}
{"step": 55535, "loss": 3.570448303222656, "grad_norm": 1.1755015850067139, "learning_rate": 0.0001, "elbo": 3.57045, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2568027210884356}
{"step": 55540, "loss": 3.562744903564453, "grad_norm": 0.7871220707893372, "learning_rate": 0.0001, "elbo": 3.56274, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.257095941825006}
{"step": 55545, "loss": 3.5679096221923827, "grad_norm": 0.8411405086517334, "learning_rate": 0.0001, "elbo": 3.56791, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2573891625615765}
{"step": 55550, "loss": 3.5684555053710936, "grad_norm": 0.7199994325637817, "learning_rate": 0.0001, "elbo": 3.56846, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.257682383298147}
{"step": 55555, "loss": 3.552906799316406, "grad_norm": 0.7715380191802979, "learning_rate": 0.0001, "elbo": 3.55291, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2579756040347174}
{"step": 55560, "loss": 3.5484542846679688, "grad_norm": 0.6851903200149536, "learning_rate": 0.0001, "elbo": 3.54845, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.258268824771288}
{"step": 55565, "loss": 3.5817901611328127, "grad_norm": 1.0801948308944702, "learning_rate": 0.0001, "elbo": 3.58179, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2585620455078583}
{"step": 55570, "loss": 3.569132995605469, "grad_norm": 0.718411386013031, "learning_rate": 0.0001, "elbo": 3.56913, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.258855266244429}
{"step": 55575, "loss": 3.552363967895508, "grad_norm": 0.9358389377593994, "learning_rate": 0.0001, "elbo": 3.55236, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2591484869809992}
{"step": 55580, "loss": 3.5632858276367188, "grad_norm": 0.7800005674362183, "learning_rate": 0.0001, "elbo": 3.56329, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2594417077175697}
{"step": 55585, "loss": 3.574815368652344, "grad_norm": 0.9255848526954651, "learning_rate": 0.0001, "elbo": 3.57482, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.25973492845414}
{"step": 55590, "loss": 3.5649650573730467, "grad_norm": 0.6679272651672363, "learning_rate": 0.0001, "elbo": 3.56496, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2600281491907106}
{"step": 55590, "loss": 3.5649650573730467, "grad_norm": 0.6679272651672363, "learning_rate": 0.0001, "elbo": 3.56496, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2600281491907106}
{"step": 55595, "loss": 3.573809051513672, "grad_norm": 0.757188618183136, "learning_rate": 0.0001, "elbo": 3.57381, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.260321369927281}
{"step": 55600, "loss": 3.5799102783203125, "grad_norm": 1.1691794395446777, "learning_rate": 0.0001, "elbo": 3.57991, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2606145906638515}
{"step": 55605, "loss": 3.5801639556884766, "grad_norm": 0.7672944068908691, "learning_rate": 0.0001, "elbo": 3.58016, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2609078114004224}
{"step": 55610, "loss": 3.563287353515625, "grad_norm": 0.7037083506584167, "learning_rate": 0.0001, "elbo": 3.56329, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.261201032136993}
{"step": 55615, "loss": 3.5487960815429687, "grad_norm": 0.8696531057357788, "learning_rate": 0.0001, "elbo": 3.5488, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2614942528735633}
{"step": 55620, "loss": 3.577414703369141, "grad_norm": 0.7110711336135864, "learning_rate": 0.0001, "elbo": 3.57741, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2617874736101338}
{"step": 55625, "loss": 3.546847915649414, "grad_norm": 0.7744240760803223, "learning_rate": 0.0001, "elbo": 3.54685, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2620806943467042}
{"step": 55630, "loss": 3.569654846191406, "grad_norm": 0.7012459635734558, "learning_rate": 0.0001, "elbo": 3.56965, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2623739150832747}
{"step": 55635, "loss": 3.5574485778808596, "grad_norm": 0.713513970375061, "learning_rate": 0.0001, "elbo": 3.55745, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.262667135819845}
{"step": 55640, "loss": 3.571269989013672, "grad_norm": 0.7157384753227234, "learning_rate": 0.0001, "elbo": 3.57127, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2629603565564156}
{"step": 55645, "loss": 3.5588287353515624, "grad_norm": 0.6736477017402649, "learning_rate": 0.0001, "elbo": 3.55883, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.263253577292986}
{"step": 55650, "loss": 3.548276901245117, "grad_norm": 0.7100265622138977, "learning_rate": 0.0001, "elbo": 3.54828, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2635467980295565}
{"step": 55655, "loss": 3.550081253051758, "grad_norm": 0.7840102910995483, "learning_rate": 0.0001, "elbo": 3.55008, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.263840018766127}
{"step": 55660, "loss": 3.569353485107422, "grad_norm": 0.6925424933433533, "learning_rate": 0.0001, "elbo": 3.56935, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.264133239502698}
{"step": 55665, "loss": 3.552602767944336, "grad_norm": 0.6973095536231995, "learning_rate": 0.0001, "elbo": 3.5526, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2644264602392683}
{"step": 55670, "loss": 3.5710708618164064, "grad_norm": 0.7498317360877991, "learning_rate": 0.0001, "elbo": 3.57107, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2647196809758388}
{"step": 55675, "loss": 3.561991500854492, "grad_norm": 0.9039867520332336, "learning_rate": 0.0001, "elbo": 3.56199, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.265012901712409}
{"step": 55680, "loss": 3.5632034301757813, "grad_norm": 0.7513254880905151, "learning_rate": 0.0001, "elbo": 3.5632, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2653061224489797}
{"step": 55685, "loss": 3.560158538818359, "grad_norm": 0.6869245767593384, "learning_rate": 0.0001, "elbo": 3.56016, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.26559934318555}
{"step": 55690, "loss": 3.5577156066894533, "grad_norm": 0.7310155630111694, "learning_rate": 0.0001, "elbo": 3.55772, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2658925639221206}
{"step": 55695, "loss": 3.584156036376953, "grad_norm": 0.6602386832237244, "learning_rate": 0.0001, "elbo": 3.58416, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.266185784658691}
{"step": 55700, "loss": 3.5660614013671874, "grad_norm": 0.6727311015129089, "learning_rate": 0.0001, "elbo": 3.56606, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2664790053952615}
{"step": 55705, "loss": 3.5630729675292967, "grad_norm": 0.6966738700866699, "learning_rate": 0.0001, "elbo": 3.56307, "mask_frac": 0.5013, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.266772226131832}
{"step": 55710, "loss": 3.5470123291015625, "grad_norm": 0.6805984377861023, "learning_rate": 0.0001, "elbo": 3.54701, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2670654468684024}
{"step": 55715, "loss": 3.564421844482422, "grad_norm": 0.7680298686027527, "learning_rate": 0.0001, "elbo": 3.56442, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.267358667604973}
{"step": 55720, "loss": 3.5534473419189454, "grad_norm": 0.6722098588943481, "learning_rate": 0.0001, "elbo": 3.55345, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2676518883415433}
{"step": 55725, "loss": 3.577809143066406, "grad_norm": 0.7333428263664246, "learning_rate": 0.0001, "elbo": 3.57781, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.267945109078114}
{"step": 55730, "loss": 3.536944580078125, "grad_norm": 0.8096815347671509, "learning_rate": 0.0001, "elbo": 3.53694, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2682383298146847}
{"step": 55735, "loss": 3.5568981170654297, "grad_norm": 0.7279720902442932, "learning_rate": 0.0001, "elbo": 3.5569, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.268531550551255}
{"step": 55740, "loss": 3.567788314819336, "grad_norm": 0.7055878639221191, "learning_rate": 0.0001, "elbo": 3.56779, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2688247712878256}
{"step": 55745, "loss": 3.560862350463867, "grad_norm": 0.7287388443946838, "learning_rate": 0.0001, "elbo": 3.56086, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.269117992024396}
{"step": 55750, "loss": 3.564646911621094, "grad_norm": 0.6894174218177795, "learning_rate": 0.0001, "elbo": 3.56465, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2694112127609665}
{"step": 55755, "loss": 3.5614219665527345, "grad_norm": 0.6835975646972656, "learning_rate": 0.0001, "elbo": 3.56142, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.269704433497537}
{"step": 55760, "loss": 3.5576438903808594, "grad_norm": 0.6861121654510498, "learning_rate": 0.0001, "elbo": 3.55764, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2699976542341074}
{"step": 55765, "loss": 3.561936950683594, "grad_norm": 0.7070083618164062, "learning_rate": 0.0001, "elbo": 3.56194, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.270290874970678}
{"step": 55770, "loss": 3.545709228515625, "grad_norm": 0.6864756345748901, "learning_rate": 0.0001, "elbo": 3.54571, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2705840957072483}
{"step": 55775, "loss": 3.563091278076172, "grad_norm": 0.7740105390548706, "learning_rate": 0.0001, "elbo": 3.56309, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2708773164438187}
{"step": 55780, "loss": 3.555014419555664, "grad_norm": 0.875053882598877, "learning_rate": 0.0001, "elbo": 3.55501, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2711705371803896}
{"step": 55785, "loss": 3.5633827209472657, "grad_norm": 0.7306950092315674, "learning_rate": 0.0001, "elbo": 3.56338, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.27146375791696}
{"step": 55790, "loss": 3.572887420654297, "grad_norm": 0.7001461386680603, "learning_rate": 0.0001, "elbo": 3.57289, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2717569786535305}
{"step": 55795, "loss": 3.563825225830078, "grad_norm": 0.6835307478904724, "learning_rate": 0.0001, "elbo": 3.56383, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.272050199390101}
{"step": 55800, "loss": 3.56849365234375, "grad_norm": 0.7784952521324158, "learning_rate": 0.0001, "elbo": 3.56849, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2723434201266715}
{"step": 55805, "loss": 3.567920684814453, "grad_norm": 0.6840937733650208, "learning_rate": 0.0001, "elbo": 3.56792, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.272636640863242}
{"step": 55810, "loss": 3.566731262207031, "grad_norm": 0.8417302966117859, "learning_rate": 0.0001, "elbo": 3.56673, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2729298615998124}
{"step": 55815, "loss": 3.5531005859375, "grad_norm": 0.6863730549812317, "learning_rate": 0.0001, "elbo": 3.5531, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.273223082336383}
{"step": 55815, "loss": 3.5531005859375, "grad_norm": 0.6863730549812317, "learning_rate": 0.0001, "elbo": 3.5531, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.273223082336383}
{"step": 55820, "loss": 3.5492935180664062, "grad_norm": 0.7783557772636414, "learning_rate": 0.0001, "elbo": 3.54929, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2735163030729533}
{"step": 55825, "loss": 3.560625457763672, "grad_norm": 0.7134937047958374, "learning_rate": 0.0001, "elbo": 3.56063, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2738095238095237}
{"step": 55830, "loss": 3.5556121826171876, "grad_norm": 0.6682528853416443, "learning_rate": 0.0001, "elbo": 3.55561, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.274102744546094}
{"step": 55835, "loss": 3.5846920013427734, "grad_norm": 0.7570863962173462, "learning_rate": 0.0001, "elbo": 3.58469, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2743959652826646}
{"step": 55840, "loss": 3.560158538818359, "grad_norm": 0.750391960144043, "learning_rate": 0.0001, "elbo": 3.56016, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.274689186019235}
{"step": 55845, "loss": 3.5709407806396483, "grad_norm": 0.7294567227363586, "learning_rate": 0.0001, "elbo": 3.57094, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2749824067558055}
{"step": 55850, "loss": 3.57188720703125, "grad_norm": 0.7099334597587585, "learning_rate": 0.0001, "elbo": 3.57189, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2752756274923764}
{"step": 55855, "loss": 3.5531105041503905, "grad_norm": 0.674382209777832, "learning_rate": 0.0001, "elbo": 3.55311, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.275568848228947}
{"step": 55860, "loss": 3.562911605834961, "grad_norm": 0.6950679421424866, "learning_rate": 0.0001, "elbo": 3.56291, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2758620689655173}
{"step": 55865, "loss": 3.556742858886719, "grad_norm": 0.6831448674201965, "learning_rate": 0.0001, "elbo": 3.55674, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.276155289702088}
{"step": 55870, "loss": 3.562460708618164, "grad_norm": 0.726514458656311, "learning_rate": 0.0001, "elbo": 3.56246, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2764485104386583}
{"step": 55870, "loss": 3.562460708618164, "grad_norm": 0.726514458656311, "learning_rate": 0.0001, "elbo": 3.56246, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2764485104386583}
{"step": 55875, "loss": 3.548426055908203, "grad_norm": 0.6921448707580566, "learning_rate": 0.0001, "elbo": 3.54843, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2767417311752287}
{"step": 55880, "loss": 3.5787620544433594, "grad_norm": 0.8564311265945435, "learning_rate": 0.0001, "elbo": 3.57876, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.277034951911799}
{"step": 55885, "loss": 3.55052490234375, "grad_norm": 0.6664489507675171, "learning_rate": 0.0001, "elbo": 3.55052, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2773281726483696}
{"step": 55890, "loss": 3.559398651123047, "grad_norm": 0.7863872051239014, "learning_rate": 0.0001, "elbo": 3.5594, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.27762139338494}
{"step": 55890, "loss": 3.559398651123047, "grad_norm": 0.7863872051239014, "learning_rate": 0.0001, "elbo": 3.5594, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.27762139338494}
{"step": 55895, "loss": 3.569139862060547, "grad_norm": 0.6953293681144714, "learning_rate": 0.0001, "elbo": 3.56914, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2779146141215105}
{"step": 55900, "loss": 3.57440299987793, "grad_norm": 0.847360372543335, "learning_rate": 0.0001, "elbo": 3.5744, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.278207834858081}
{"step": 55905, "loss": 3.559629440307617, "grad_norm": 0.7770601511001587, "learning_rate": 0.0001, "elbo": 3.55963, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.278501055594652}
{"step": 55910, "loss": 3.5655441284179688, "grad_norm": 0.695214033126831, "learning_rate": 0.0001, "elbo": 3.56554, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2787942763312223}
{"step": 55910, "loss": 3.5655441284179688, "grad_norm": 0.695214033126831, "learning_rate": 0.0001, "elbo": 3.56554, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2787942763312223}
{"step": 55915, "loss": 3.5710685729980467, "grad_norm": 0.7053542137145996, "learning_rate": 0.0001, "elbo": 3.57107, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.279087497067793}
{"step": 55915, "loss": 3.5710685729980467, "grad_norm": 0.7053542137145996, "learning_rate": 0.0001, "elbo": 3.57107, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.279087497067793}
{"step": 55920, "loss": 3.5502159118652346, "grad_norm": 1.4833290576934814, "learning_rate": 0.0001, "elbo": 3.55022, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2793807178043632}
{"step": 55925, "loss": 3.5805587768554688, "grad_norm": 0.7072127461433411, "learning_rate": 0.0001, "elbo": 3.58056, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2796739385409337}
{"step": 55930, "loss": 3.5688514709472656, "grad_norm": 0.7541842460632324, "learning_rate": 0.0001, "elbo": 3.56885, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.279967159277504}
{"step": 55935, "loss": 3.5678707122802735, "grad_norm": 0.6968423128128052, "learning_rate": 0.0001, "elbo": 3.56787, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2802603800140746}
{"step": 55940, "loss": 3.551488494873047, "grad_norm": 0.8580779433250427, "learning_rate": 0.0001, "elbo": 3.55149, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.280553600750645}
{"step": 55945, "loss": 3.564105987548828, "grad_norm": 0.6667040586471558, "learning_rate": 0.0001, "elbo": 3.56411, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2808468214872155}
{"step": 55950, "loss": 3.5718547821044924, "grad_norm": 0.6774272322654724, "learning_rate": 0.0001, "elbo": 3.57185, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.281140042223786}
{"step": 55950, "loss": 3.5718547821044924, "grad_norm": 0.6774272322654724, "learning_rate": 0.0001, "elbo": 3.57185, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.281140042223786}
{"step": 55955, "loss": 3.5590847015380858, "grad_norm": 0.6924669742584229, "learning_rate": 0.0001, "elbo": 3.55908, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2814332629603564}
{"step": 55960, "loss": 3.563999557495117, "grad_norm": 0.701897382736206, "learning_rate": 0.0001, "elbo": 3.564, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.281726483696927}
{"step": 55965, "loss": 3.569490432739258, "grad_norm": 1.040439248085022, "learning_rate": 0.0001, "elbo": 3.56949, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2820197044334973}
{"step": 55970, "loss": 3.5581123352050783, "grad_norm": 0.7156822681427002, "learning_rate": 0.0001, "elbo": 3.55811, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.282312925170068}
{"step": 55975, "loss": 3.552751922607422, "grad_norm": 0.7231509685516357, "learning_rate": 0.0001, "elbo": 3.55275, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2826061459066387}
{"step": 55980, "loss": 3.562422180175781, "grad_norm": 0.6843556761741638, "learning_rate": 0.0001, "elbo": 3.56242, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.282899366643209}
{"step": 55985, "loss": 3.5520484924316404, "grad_norm": 0.7011542916297913, "learning_rate": 0.0001, "elbo": 3.55205, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2831925873797796}
{"step": 55990, "loss": 3.547626495361328, "grad_norm": 0.6933794617652893, "learning_rate": 0.0001, "elbo": 3.54763, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.28348580811635}
{"step": 55995, "loss": 3.553878402709961, "grad_norm": 0.8880337476730347, "learning_rate": 0.0001, "elbo": 3.55388, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2837790288529205}
{"step": 56000, "loss": 3.553053283691406, "grad_norm": 0.7279173135757446, "learning_rate": 0.0001, "elbo": 3.55305, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.284072249589491}
{"step": 56005, "loss": 3.5716064453125, "grad_norm": 0.7036846876144409, "learning_rate": 0.0001, "elbo": 3.57161, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2843654703260614}
{"step": 56010, "loss": 3.57824592590332, "grad_norm": 0.6977311968803406, "learning_rate": 0.0001, "elbo": 3.57825, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.284658691062632}
{"step": 56015, "loss": 3.558563232421875, "grad_norm": 0.6917601227760315, "learning_rate": 0.0001, "elbo": 3.55856, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2849519117992023}
{"step": 56020, "loss": 3.562378692626953, "grad_norm": 0.7260198593139648, "learning_rate": 0.0001, "elbo": 3.56238, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2852451325357728}
{"step": 56025, "loss": 3.5641571044921876, "grad_norm": 0.7321420907974243, "learning_rate": 0.0001, "elbo": 3.56416, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2855383532723432}
{"step": 56030, "loss": 3.569416809082031, "grad_norm": 0.7240027189254761, "learning_rate": 0.0001, "elbo": 3.56942, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.285831574008914}
{"step": 56035, "loss": 3.5488296508789063, "grad_norm": 1.1417732238769531, "learning_rate": 0.0001, "elbo": 3.54883, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2861247947454846}
{"step": 56040, "loss": 3.563230514526367, "grad_norm": 0.7176854610443115, "learning_rate": 0.0001, "elbo": 3.56323, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.286418015482055}
{"step": 56040, "loss": 3.563230514526367, "grad_norm": 0.7176854610443115, "learning_rate": 0.0001, "elbo": 3.56323, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.286418015482055}
{"step": 56045, "loss": 3.5791767120361326, "grad_norm": 0.7385318279266357, "learning_rate": 0.0001, "elbo": 3.57918, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2867112362186255}
{"step": 56050, "loss": 3.545366668701172, "grad_norm": 0.7030066847801208, "learning_rate": 0.0001, "elbo": 3.54537, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.287004456955196}
{"step": 56055, "loss": 3.5680694580078125, "grad_norm": 0.6614353060722351, "learning_rate": 0.0001, "elbo": 3.56807, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2872976776917664}
{"step": 56060, "loss": 3.56654052734375, "grad_norm": 0.6945545077323914, "learning_rate": 0.0001, "elbo": 3.56654, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.287590898428337}
{"step": 56065, "loss": 3.566151428222656, "grad_norm": 0.6909814476966858, "learning_rate": 0.0001, "elbo": 3.56615, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2878841191649073}
{"step": 56070, "loss": 3.573909378051758, "grad_norm": 0.6648910641670227, "learning_rate": 0.0001, "elbo": 3.57391, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2881773399014778}
{"step": 56075, "loss": 3.5506248474121094, "grad_norm": 0.6897405982017517, "learning_rate": 0.0001, "elbo": 3.55062, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.288470560638048}
{"step": 56080, "loss": 3.5660400390625, "grad_norm": 0.6737955808639526, "learning_rate": 0.0001, "elbo": 3.56604, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2887637813746187}
{"step": 56085, "loss": 3.5659080505371095, "grad_norm": 0.6520373821258545, "learning_rate": 0.0001, "elbo": 3.56591, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.289057002111189}
{"step": 56090, "loss": 3.5713470458984373, "grad_norm": 0.7552919983863831, "learning_rate": 0.0001, "elbo": 3.57135, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2893502228477596}
{"step": 56095, "loss": 3.572226715087891, "grad_norm": 0.7468201518058777, "learning_rate": 0.0001, "elbo": 3.57223, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2896434435843305}
{"step": 56100, "loss": 3.5633171081542967, "grad_norm": 0.6801267862319946, "learning_rate": 0.0001, "elbo": 3.56332, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.289936664320901}
{"step": 56105, "loss": 3.5814590454101562, "grad_norm": 0.654647707939148, "learning_rate": 0.0001, "elbo": 3.58146, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2902298850574714}
{"step": 56110, "loss": 3.545964813232422, "grad_norm": 0.758091926574707, "learning_rate": 0.0001, "elbo": 3.54597, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.290523105794042}
{"step": 56115, "loss": 3.5759620666503906, "grad_norm": 0.948489785194397, "learning_rate": 0.0001, "elbo": 3.57596, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2908163265306123}
{"step": 56120, "loss": 3.5757213592529298, "grad_norm": 0.8236152529716492, "learning_rate": 0.0001, "elbo": 3.57572, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2911095472671827}
{"step": 56125, "loss": 3.545025634765625, "grad_norm": 0.755967378616333, "learning_rate": 0.0001, "elbo": 3.54503, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.291402768003753}
{"step": 56130, "loss": 3.559468460083008, "grad_norm": 0.7560310959815979, "learning_rate": 0.0001, "elbo": 3.55947, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2916959887403237}
{"step": 56135, "loss": 3.5641075134277345, "grad_norm": 1.1191331148147583, "learning_rate": 0.0001, "elbo": 3.56411, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.291989209476894}
{"step": 56140, "loss": 3.5597206115722657, "grad_norm": 0.742942214012146, "learning_rate": 0.0001, "elbo": 3.55972, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2922824302134646}
{"step": 56145, "loss": 3.577585220336914, "grad_norm": 0.719785213470459, "learning_rate": 0.0001, "elbo": 3.57759, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.292575650950035}
{"step": 56150, "loss": 3.562452697753906, "grad_norm": 0.7369282841682434, "learning_rate": 0.0001, "elbo": 3.56245, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.292868871686606}
{"step": 56155, "loss": 3.57447509765625, "grad_norm": 0.6830887794494629, "learning_rate": 0.0001, "elbo": 3.57448, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2931620924231764}
{"step": 56160, "loss": 3.560759735107422, "grad_norm": 0.6502081751823425, "learning_rate": 0.0001, "elbo": 3.56076, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.293455313159747}
{"step": 56165, "loss": 3.5681549072265626, "grad_norm": 0.7068232297897339, "learning_rate": 0.0001, "elbo": 3.56816, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2937485338963173}
{"step": 56170, "loss": 3.559917449951172, "grad_norm": 0.7154000997543335, "learning_rate": 0.0001, "elbo": 3.55992, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2940417546328877}
{"step": 56175, "loss": 3.566355514526367, "grad_norm": 0.7492866516113281, "learning_rate": 0.0001, "elbo": 3.56636, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.294334975369458}
{"step": 56180, "loss": 3.5469650268554687, "grad_norm": 0.6498070359230042, "learning_rate": 0.0001, "elbo": 3.54696, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2946281961060286}
{"step": 56185, "loss": 3.569191741943359, "grad_norm": 0.7181676626205444, "learning_rate": 0.0001, "elbo": 3.56919, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.294921416842599}
{"step": 56190, "loss": 3.5643989562988283, "grad_norm": 0.7411574721336365, "learning_rate": 0.0001, "elbo": 3.5644, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2952146375791695}
{"step": 56195, "loss": 3.5642906188964845, "grad_norm": 0.7264013886451721, "learning_rate": 0.0001, "elbo": 3.56429, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.29550785831574}
{"step": 56200, "loss": 3.559987258911133, "grad_norm": 0.7334200739860535, "learning_rate": 0.0001, "elbo": 3.55999, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2958010790523105}
{"step": 56205, "loss": 3.557862091064453, "grad_norm": 0.685702919960022, "learning_rate": 0.0001, "elbo": 3.55786, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.296094299788881}
{"step": 56210, "loss": 3.563140106201172, "grad_norm": 0.7177128791809082, "learning_rate": 0.0001, "elbo": 3.56314, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2963875205254514}
{"step": 56215, "loss": 3.5596580505371094, "grad_norm": 0.7732466459274292, "learning_rate": 0.0001, "elbo": 3.55966, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.296680741262022}
{"step": 56220, "loss": 3.558367156982422, "grad_norm": 0.7294368147850037, "learning_rate": 0.0001, "elbo": 3.55837, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2969739619985927}
{"step": 56225, "loss": 3.5626659393310547, "grad_norm": 0.686430811882019, "learning_rate": 0.0001, "elbo": 3.56267, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.297267182735163}
{"step": 56230, "loss": 3.5472206115722655, "grad_norm": 0.6842270493507385, "learning_rate": 0.0001, "elbo": 3.54722, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2975604034717336}
{"step": 56235, "loss": 3.5587554931640626, "grad_norm": 0.8458678722381592, "learning_rate": 0.0001, "elbo": 3.55876, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.297853624208304}
{"step": 56240, "loss": 3.5511924743652346, "grad_norm": 0.7443445324897766, "learning_rate": 0.0001, "elbo": 3.55119, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2981468449448745}
{"step": 56245, "loss": 3.581734466552734, "grad_norm": 0.7830658555030823, "learning_rate": 0.0001, "elbo": 3.58173, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.298440065681445}
{"step": 56245, "loss": 3.581734466552734, "grad_norm": 0.7830658555030823, "learning_rate": 0.0001, "elbo": 3.58173, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.298440065681445}
{"step": 56250, "loss": 3.564097595214844, "grad_norm": 0.7024058103561401, "learning_rate": 0.0001, "elbo": 3.5641, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2987332864180154}
{"step": 56255, "loss": 3.578645706176758, "grad_norm": 0.7529752254486084, "learning_rate": 0.0001, "elbo": 3.57865, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.299026507154586}
{"step": 56260, "loss": 3.5553253173828123, "grad_norm": 0.6885033845901489, "learning_rate": 0.0001, "elbo": 3.55533, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2993197278911564}
{"step": 56260, "loss": 3.5553253173828123, "grad_norm": 0.6885033845901489, "learning_rate": 0.0001, "elbo": 3.55533, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2993197278911564}
{"step": 56265, "loss": 3.5746036529541017, "grad_norm": 0.7445930242538452, "learning_rate": 0.0001, "elbo": 3.5746, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.299612948627727}
{"step": 56270, "loss": 3.5703392028808594, "grad_norm": 0.7044776678085327, "learning_rate": 0.0001, "elbo": 3.57034, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.2999061693642973}
{"step": 56275, "loss": 3.5622772216796874, "grad_norm": 0.6528300046920776, "learning_rate": 0.0001, "elbo": 3.56228, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.300199390100868}
{"step": 56280, "loss": 3.562326431274414, "grad_norm": 0.6777749061584473, "learning_rate": 0.0001, "elbo": 3.56233, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3004926108374386}
{"step": 56285, "loss": 3.555216598510742, "grad_norm": 0.6754454374313354, "learning_rate": 0.0001, "elbo": 3.55522, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.300785831574009}
{"step": 56290, "loss": 3.5676525115966795, "grad_norm": 0.6393699645996094, "learning_rate": 0.0001, "elbo": 3.56765, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3010790523105795}
{"step": 56295, "loss": 3.5600612640380858, "grad_norm": 0.7322795987129211, "learning_rate": 0.0001, "elbo": 3.56006, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.30137227304715}
{"step": 56300, "loss": 3.555403137207031, "grad_norm": 0.695277214050293, "learning_rate": 0.0001, "elbo": 3.5554, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3016654937837204}
{"step": 56305, "loss": 3.549729919433594, "grad_norm": 0.7205774188041687, "learning_rate": 0.0001, "elbo": 3.54973, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.301958714520291}
{"step": 56305, "loss": 3.549729919433594, "grad_norm": 0.7205774188041687, "learning_rate": 0.0001, "elbo": 3.54973, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.301958714520291}
{"step": 56310, "loss": 3.546366882324219, "grad_norm": 0.7631832361221313, "learning_rate": 0.0001, "elbo": 3.54637, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3022519352568613}
{"step": 56315, "loss": 3.5608795166015623, "grad_norm": 0.7396910786628723, "learning_rate": 0.0001, "elbo": 3.56088, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.302545155993432}
{"step": 56320, "loss": 3.5670497894287108, "grad_norm": 0.8825476169586182, "learning_rate": 0.0001, "elbo": 3.56705, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3028383767300022}
{"step": 56325, "loss": 3.5650299072265623, "grad_norm": 0.8403443694114685, "learning_rate": 0.0001, "elbo": 3.56503, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3031315974665727}
{"step": 56325, "loss": 3.5650299072265623, "grad_norm": 0.8403443694114685, "learning_rate": 0.0001, "elbo": 3.56503, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3031315974665727}
{"step": 56330, "loss": 3.581198501586914, "grad_norm": 0.7180913090705872, "learning_rate": 0.0001, "elbo": 3.5812, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.303424818203143}
{"step": 56335, "loss": 3.5689815521240233, "grad_norm": 1.7245335578918457, "learning_rate": 0.0001, "elbo": 3.56898, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3037180389397136}
{"step": 56340, "loss": 3.5764041900634767, "grad_norm": 0.760356605052948, "learning_rate": 0.0001, "elbo": 3.5764, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.304011259676284}
{"step": 56345, "loss": 3.551957702636719, "grad_norm": 0.6889657378196716, "learning_rate": 0.0001, "elbo": 3.55196, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.304304480412855}
{"step": 56350, "loss": 3.5670753479003907, "grad_norm": 0.7226096391677856, "learning_rate": 0.0001, "elbo": 3.56708, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3045977011494254}
{"step": 56355, "loss": 3.5643997192382812, "grad_norm": 0.6730265617370605, "learning_rate": 0.0001, "elbo": 3.5644, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.304890921885996}
{"step": 56360, "loss": 3.567103958129883, "grad_norm": 0.6719805002212524, "learning_rate": 0.0001, "elbo": 3.5671, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3051841426225663}
{"step": 56365, "loss": 3.5703666687011717, "grad_norm": 0.7422103881835938, "learning_rate": 0.0001, "elbo": 3.57037, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3054773633591368}
{"step": 56370, "loss": 3.567555236816406, "grad_norm": 0.8142765760421753, "learning_rate": 0.0001, "elbo": 3.56756, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3057705840957072}
{"step": 56375, "loss": 3.55450325012207, "grad_norm": 0.6747686862945557, "learning_rate": 0.0001, "elbo": 3.5545, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3060638048322777}
{"step": 56380, "loss": 3.57158203125, "grad_norm": 0.6939190030097961, "learning_rate": 0.0001, "elbo": 3.57158, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.306357025568848}
{"step": 56385, "loss": 3.5674057006835938, "grad_norm": 0.6968770623207092, "learning_rate": 0.0001, "elbo": 3.56741, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3066502463054186}
{"step": 56390, "loss": 3.5643661499023436, "grad_norm": 0.819935142993927, "learning_rate": 0.0001, "elbo": 3.56437, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.306943467041989}
{"step": 56395, "loss": 3.5703704833984373, "grad_norm": 0.6832893490791321, "learning_rate": 0.0001, "elbo": 3.57037, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.30723668777856}
{"step": 56400, "loss": 3.553818130493164, "grad_norm": 0.7390276789665222, "learning_rate": 0.0001, "elbo": 3.55382, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3075299085151304}
{"step": 56405, "loss": 3.5602317810058595, "grad_norm": 0.697059690952301, "learning_rate": 0.0001, "elbo": 3.56023, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.307823129251701}
{"step": 56410, "loss": 3.558426284790039, "grad_norm": 0.7536987066268921, "learning_rate": 0.0001, "elbo": 3.55843, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3081163499882713}
{"step": 56415, "loss": 3.5758331298828123, "grad_norm": 0.7668787837028503, "learning_rate": 0.0001, "elbo": 3.57583, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3084095707248418}
{"step": 56420, "loss": 3.583448028564453, "grad_norm": 0.6836137771606445, "learning_rate": 0.0001, "elbo": 3.58345, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.308702791461412}
{"step": 56425, "loss": 3.5687793731689452, "grad_norm": 0.7423394322395325, "learning_rate": 0.0001, "elbo": 3.56878, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3089960121979827}
{"step": 56430, "loss": 3.582530212402344, "grad_norm": 0.8125085830688477, "learning_rate": 0.0001, "elbo": 3.58253, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.309289232934553}
{"step": 56435, "loss": 3.5556968688964843, "grad_norm": 0.7292197942733765, "learning_rate": 0.0001, "elbo": 3.5557, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3095824536711236}
{"step": 56440, "loss": 3.551861572265625, "grad_norm": 0.7355933785438538, "learning_rate": 0.0001, "elbo": 3.55186, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.309875674407694}
{"step": 56445, "loss": 3.583283233642578, "grad_norm": 0.7183300852775574, "learning_rate": 0.0001, "elbo": 3.58328, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3101688951442645}
{"step": 56450, "loss": 3.548810577392578, "grad_norm": 0.7186785340309143, "learning_rate": 0.0001, "elbo": 3.54881, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.310462115880835}
{"step": 56455, "loss": 3.547211456298828, "grad_norm": 0.6902256011962891, "learning_rate": 0.0001, "elbo": 3.54721, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3107553366174054}
{"step": 56460, "loss": 3.560960388183594, "grad_norm": 0.680672287940979, "learning_rate": 0.0001, "elbo": 3.56096, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.311048557353976}
{"step": 56460, "loss": 3.560960388183594, "grad_norm": 0.680672287940979, "learning_rate": 0.0001, "elbo": 3.56096, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.311048557353976}
{"step": 56465, "loss": 3.577275848388672, "grad_norm": 0.6841368079185486, "learning_rate": 0.0001, "elbo": 3.57728, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3113417780905468}
{"step": 56470, "loss": 3.5540672302246095, "grad_norm": 0.7524047493934631, "learning_rate": 0.0001, "elbo": 3.55407, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.311634998827117}
{"step": 56475, "loss": 3.5594966888427733, "grad_norm": 0.7649461627006531, "learning_rate": 0.0001, "elbo": 3.5595, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3119282195636877}
{"step": 56480, "loss": 3.5655479431152344, "grad_norm": 0.6946454644203186, "learning_rate": 0.0001, "elbo": 3.56555, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.312221440300258}
{"step": 56485, "loss": 3.5469505310058596, "grad_norm": 0.6808149814605713, "learning_rate": 0.0001, "elbo": 3.54695, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3125146610368286}
{"step": 56490, "loss": 3.5686195373535154, "grad_norm": 0.6647868156433105, "learning_rate": 0.0001, "elbo": 3.56862, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.312807881773399}
{"step": 56495, "loss": 3.576851654052734, "grad_norm": 0.6796044111251831, "learning_rate": 0.0001, "elbo": 3.57685, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3131011025099695}
{"step": 56500, "loss": 3.5519023895263673, "grad_norm": 0.7470144033432007, "learning_rate": 0.0001, "elbo": 3.5519, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.31339432324654}
{"step": 56505, "loss": 3.552101898193359, "grad_norm": 0.757502555847168, "learning_rate": 0.0001, "elbo": 3.5521, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3136875439831104}
{"step": 56510, "loss": 3.5718063354492187, "grad_norm": 0.7437593340873718, "learning_rate": 0.0001, "elbo": 3.57181, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.313980764719681}
{"step": 56510, "loss": 3.5718063354492187, "grad_norm": 0.7437593340873718, "learning_rate": 0.0001, "elbo": 3.57181, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.313980764719681}
{"step": 56515, "loss": 3.554051971435547, "grad_norm": 0.7011553049087524, "learning_rate": 0.0001, "elbo": 3.55405, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3142739854562513}
{"step": 56520, "loss": 3.5644744873046874, "grad_norm": 0.7648007869720459, "learning_rate": 0.0001, "elbo": 3.56447, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.314567206192822}
{"step": 56525, "loss": 3.5846466064453124, "grad_norm": 0.7346631288528442, "learning_rate": 0.0001, "elbo": 3.58465, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3148604269293926}
{"step": 56530, "loss": 3.5586013793945312, "grad_norm": 0.6692244410514832, "learning_rate": 0.0001, "elbo": 3.5586, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.315153647665963}
{"step": 56530, "loss": 3.5586013793945312, "grad_norm": 0.6692244410514832, "learning_rate": 0.0001, "elbo": 3.5586, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.315153647665963}
{"step": 56535, "loss": 3.5644248962402343, "grad_norm": 0.8948888182640076, "learning_rate": 0.0001, "elbo": 3.56443, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3154468684025336}
{"step": 56540, "loss": 3.544694519042969, "grad_norm": 0.8222179412841797, "learning_rate": 0.0001, "elbo": 3.54469, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.315740089139104}
{"step": 56540, "loss": 3.544694519042969, "grad_norm": 0.8222179412841797, "learning_rate": 0.0001, "elbo": 3.54469, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.315740089139104}
{"step": 56545, "loss": 3.5775413513183594, "grad_norm": 0.7525594234466553, "learning_rate": 0.0001, "elbo": 3.57754, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3160333098756745}
{"step": 56550, "loss": 3.5784473419189453, "grad_norm": 0.8570679426193237, "learning_rate": 0.0001, "elbo": 3.57845, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.316326530612245}
{"step": 56555, "loss": 3.5793933868408203, "grad_norm": 0.719294548034668, "learning_rate": 0.0001, "elbo": 3.57939, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3166197513488154}
{"step": 56555, "loss": 3.5793933868408203, "grad_norm": 0.719294548034668, "learning_rate": 0.0001, "elbo": 3.57939, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3166197513488154}
{"step": 56560, "loss": 3.576465606689453, "grad_norm": 0.6894740462303162, "learning_rate": 0.0001, "elbo": 3.57647, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.316912972085386}
{"step": 56565, "loss": 3.563252639770508, "grad_norm": 0.9281858801841736, "learning_rate": 0.0001, "elbo": 3.56325, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3172061928219563}
{"step": 56565, "loss": 3.563252639770508, "grad_norm": 0.9281858801841736, "learning_rate": 0.0001, "elbo": 3.56325, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3172061928219563}
{"step": 56570, "loss": 3.5521812438964844, "grad_norm": 0.6990526914596558, "learning_rate": 0.0001, "elbo": 3.55218, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3174994135585267}
{"step": 56575, "loss": 3.56865234375, "grad_norm": 0.7289517521858215, "learning_rate": 0.0001, "elbo": 3.56865, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.317792634295097}
{"step": 56580, "loss": 3.5579147338867188, "grad_norm": 0.7433049082756042, "learning_rate": 0.0001, "elbo": 3.55791, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3180858550316676}
{"step": 56585, "loss": 3.5607391357421876, "grad_norm": 0.7257826328277588, "learning_rate": 0.0001, "elbo": 3.56074, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.318379075768238}
{"step": 56590, "loss": 3.5592636108398437, "grad_norm": 0.7177814841270447, "learning_rate": 0.0001, "elbo": 3.55926, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.318672296504809}
{"step": 56595, "loss": 3.539785385131836, "grad_norm": 0.9547663331031799, "learning_rate": 0.0001, "elbo": 3.53979, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3189655172413794}
{"step": 56600, "loss": 3.5757545471191405, "grad_norm": 0.7432172894477844, "learning_rate": 0.0001, "elbo": 3.57575, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.31925873797795}
{"step": 56605, "loss": 3.552644729614258, "grad_norm": 0.687117874622345, "learning_rate": 0.0001, "elbo": 3.55264, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3195519587145204}
{"step": 56610, "loss": 3.578723907470703, "grad_norm": 0.748434841632843, "learning_rate": 0.0001, "elbo": 3.57872, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.319845179451091}
{"step": 56615, "loss": 3.5591331481933595, "grad_norm": 0.7994866371154785, "learning_rate": 0.0001, "elbo": 3.55913, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3201384001876613}
{"step": 56620, "loss": 3.5579803466796873, "grad_norm": 0.7786325812339783, "learning_rate": 0.0001, "elbo": 3.55798, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3204316209242317}
{"step": 56625, "loss": 3.559345245361328, "grad_norm": 0.7883908748626709, "learning_rate": 0.0001, "elbo": 3.55935, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.320724841660802}
{"step": 56630, "loss": 3.550820159912109, "grad_norm": 0.6832528114318848, "learning_rate": 0.0001, "elbo": 3.55082, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3210180623973726}
{"step": 56635, "loss": 3.559048080444336, "grad_norm": 0.6706544756889343, "learning_rate": 0.0001, "elbo": 3.55905, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.321311283133943}
{"step": 56640, "loss": 3.5515819549560548, "grad_norm": 0.7577689290046692, "learning_rate": 0.0001, "elbo": 3.55158, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3216045038705135}
{"step": 56645, "loss": 3.5660140991210936, "grad_norm": 0.7234012484550476, "learning_rate": 0.0001, "elbo": 3.56601, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3218977246070844}
{"step": 56650, "loss": 3.565288543701172, "grad_norm": 0.6810848712921143, "learning_rate": 0.0001, "elbo": 3.56529, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.322190945343655}
{"step": 56655, "loss": 3.5569198608398436, "grad_norm": 0.7455958724021912, "learning_rate": 0.0001, "elbo": 3.55692, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3224841660802253}
{"step": 56660, "loss": 3.5640567779541015, "grad_norm": 0.7192831039428711, "learning_rate": 0.0001, "elbo": 3.56406, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.322777386816796}
{"step": 56665, "loss": 3.5551692962646486, "grad_norm": 0.7174906134605408, "learning_rate": 0.0001, "elbo": 3.55517, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3230706075533663}
{"step": 56670, "loss": 3.562515640258789, "grad_norm": 0.8472068905830383, "learning_rate": 0.0001, "elbo": 3.56252, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3233638282899367}
{"step": 56675, "loss": 3.556143951416016, "grad_norm": 0.6868982315063477, "learning_rate": 0.0001, "elbo": 3.55614, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.323657049026507}
{"step": 56680, "loss": 3.5749790191650392, "grad_norm": 0.6784007549285889, "learning_rate": 0.0001, "elbo": 3.57498, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3239502697630776}
{"step": 56680, "loss": 3.5749790191650392, "grad_norm": 0.6784007549285889, "learning_rate": 0.0001, "elbo": 3.57498, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3239502697630776}
{"step": 56685, "loss": 3.5671951293945314, "grad_norm": 0.6566354036331177, "learning_rate": 0.0001, "elbo": 3.5672, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.324243490499648}
{"step": 56690, "loss": 3.5640274047851563, "grad_norm": 0.7969095706939697, "learning_rate": 0.0001, "elbo": 3.56403, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3245367112362185}
{"step": 56695, "loss": 3.5568748474121095, "grad_norm": 0.6739397644996643, "learning_rate": 0.0001, "elbo": 3.55687, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.324829931972789}
{"step": 56695, "loss": 3.5568748474121095, "grad_norm": 0.6739397644996643, "learning_rate": 0.0001, "elbo": 3.55687, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.324829931972789}
{"step": 56700, "loss": 3.5749446868896486, "grad_norm": 0.7905771136283875, "learning_rate": 0.0001, "elbo": 3.57494, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3251231527093594}
{"step": 56705, "loss": 3.5535472869873046, "grad_norm": 0.6672411561012268, "learning_rate": 0.0001, "elbo": 3.55355, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.32541637344593}
{"step": 56710, "loss": 3.5518264770507812, "grad_norm": 1.1277742385864258, "learning_rate": 0.0001, "elbo": 3.55183, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3257095941825003}
{"step": 56715, "loss": 3.550927734375, "grad_norm": 0.7037115693092346, "learning_rate": 0.0001, "elbo": 3.55093, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3260028149190712}
{"step": 56720, "loss": 3.572714614868164, "grad_norm": 0.7211564183235168, "learning_rate": 0.0001, "elbo": 3.57271, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3262960356556417}
{"step": 56725, "loss": 3.5567893981933594, "grad_norm": 0.776782214641571, "learning_rate": 0.0001, "elbo": 3.55679, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.326589256392212}
{"step": 56730, "loss": 3.5720123291015624, "grad_norm": 0.6713465452194214, "learning_rate": 0.0001, "elbo": 3.57201, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3268824771287826}
{"step": 56735, "loss": 3.5634315490722654, "grad_norm": 0.7648506760597229, "learning_rate": 0.0001, "elbo": 3.56343, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.327175697865353}
{"step": 56740, "loss": 3.5691795349121094, "grad_norm": 0.6834145188331604, "learning_rate": 0.0001, "elbo": 3.56918, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3274689186019235}
{"step": 56745, "loss": 3.545054626464844, "grad_norm": 0.6810574531555176, "learning_rate": 0.0001, "elbo": 3.54505, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.327762139338494}
{"step": 56750, "loss": 3.5648868560791014, "grad_norm": 0.6747118234634399, "learning_rate": 0.0001, "elbo": 3.56489, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3280553600750644}
{"step": 56755, "loss": 3.5496376037597654, "grad_norm": 0.9896610379219055, "learning_rate": 0.0001, "elbo": 3.54964, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.328348580811635}
{"step": 56760, "loss": 3.5840385437011717, "grad_norm": 1.0691770315170288, "learning_rate": 0.0001, "elbo": 3.58404, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3286418015482053}
{"step": 56765, "loss": 3.5608139038085938, "grad_norm": 0.6903071403503418, "learning_rate": 0.0001, "elbo": 3.56081, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3289350222847762}
{"step": 56770, "loss": 3.5515159606933593, "grad_norm": 0.6976721286773682, "learning_rate": 0.0001, "elbo": 3.55152, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3292282430213467}
{"step": 56775, "loss": 3.568574142456055, "grad_norm": 0.7770286202430725, "learning_rate": 0.0001, "elbo": 3.56857, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.329521463757917}
{"step": 56780, "loss": 3.5792686462402346, "grad_norm": 1.076593041419983, "learning_rate": 0.0001, "elbo": 3.57927, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3298146844944876}
{"step": 56785, "loss": 3.562705230712891, "grad_norm": 0.6891635060310364, "learning_rate": 0.0001, "elbo": 3.56271, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.330107905231058}
{"step": 56790, "loss": 3.5622276306152343, "grad_norm": 0.892260730266571, "learning_rate": 0.0001, "elbo": 3.56223, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3304011259676285}
{"step": 56795, "loss": 3.5579132080078124, "grad_norm": 0.9541571140289307, "learning_rate": 0.0001, "elbo": 3.55791, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.330694346704199}
{"step": 56800, "loss": 3.5756568908691406, "grad_norm": 0.6834717988967896, "learning_rate": 0.0001, "elbo": 3.57566, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3309875674407694}
{"step": 56805, "loss": 3.558912658691406, "grad_norm": 1.3899991512298584, "learning_rate": 0.0001, "elbo": 3.55891, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.33128078817734}
{"step": 56810, "loss": 3.5759765625, "grad_norm": 0.6801031231880188, "learning_rate": 0.0001, "elbo": 3.57598, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3315740089139103}
{"step": 56815, "loss": 3.5611148834228517, "grad_norm": 0.6853992342948914, "learning_rate": 0.0001, "elbo": 3.56112, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3318672296504808}
{"step": 56820, "loss": 3.563090515136719, "grad_norm": 1.2644377946853638, "learning_rate": 0.0001, "elbo": 3.56309, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.332160450387051}
{"step": 56825, "loss": 3.567060089111328, "grad_norm": 0.7330054044723511, "learning_rate": 0.0001, "elbo": 3.56706, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3324536711236217}
{"step": 56830, "loss": 3.5507476806640623, "grad_norm": 0.8622913360595703, "learning_rate": 0.0001, "elbo": 3.55075, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.332746891860192}
{"step": 56835, "loss": 3.561223602294922, "grad_norm": 0.9259289503097534, "learning_rate": 0.0001, "elbo": 3.56122, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.333040112596763}
{"step": 56840, "loss": 3.5745571136474608, "grad_norm": 0.6972899436950684, "learning_rate": 0.0001, "elbo": 3.57456, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3333333333333335}
{"step": 56845, "loss": 3.555330276489258, "grad_norm": 1.1601642370224, "learning_rate": 0.0001, "elbo": 3.55533, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.333626554069904}
{"step": 56850, "loss": 3.545454406738281, "grad_norm": 0.8343071341514587, "learning_rate": 0.0001, "elbo": 3.54545, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3339197748064744}
{"step": 56855, "loss": 3.5585235595703124, "grad_norm": 0.799742579460144, "learning_rate": 0.0001, "elbo": 3.55852, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.334212995543045}
{"step": 56860, "loss": 3.569599151611328, "grad_norm": 0.6866136789321899, "learning_rate": 0.0001, "elbo": 3.5696, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3345062162796153}
{"step": 56865, "loss": 3.560749816894531, "grad_norm": 0.8682196140289307, "learning_rate": 0.0001, "elbo": 3.56075, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3347994370161858}
{"step": 56870, "loss": 3.5667980194091795, "grad_norm": 0.6775894165039062, "learning_rate": 0.0001, "elbo": 3.5668, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.335092657752756}
{"step": 56875, "loss": 3.5757484436035156, "grad_norm": 0.7289643883705139, "learning_rate": 0.0001, "elbo": 3.57575, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3353858784893267}
{"step": 56880, "loss": 3.5586406707763674, "grad_norm": 0.7108612060546875, "learning_rate": 0.0001, "elbo": 3.55864, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.335679099225897}
{"step": 56885, "loss": 3.560839080810547, "grad_norm": 0.6783987283706665, "learning_rate": 0.0001, "elbo": 3.56084, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3359723199624676}
{"step": 56890, "loss": 3.545384979248047, "grad_norm": 0.678878903388977, "learning_rate": 0.0001, "elbo": 3.54538, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3362655406990385}
{"step": 56895, "loss": 3.5616153717041015, "grad_norm": 0.6783525347709656, "learning_rate": 0.0001, "elbo": 3.56162, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.336558761435609}
{"step": 56900, "loss": 3.543770599365234, "grad_norm": 0.6957241296768188, "learning_rate": 0.0001, "elbo": 3.54377, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3368519821721794}
{"step": 56905, "loss": 3.5649642944335938, "grad_norm": 0.7400594353675842, "learning_rate": 0.0001, "elbo": 3.56496, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.33714520290875}
{"step": 56910, "loss": 3.5597660064697267, "grad_norm": 0.7017709016799927, "learning_rate": 0.0001, "elbo": 3.55977, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3374384236453203}
{"step": 56915, "loss": 3.553887939453125, "grad_norm": 0.7061389088630676, "learning_rate": 0.0001, "elbo": 3.55389, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3377316443818907}
{"step": 56920, "loss": 3.569300079345703, "grad_norm": 0.667262852191925, "learning_rate": 0.0001, "elbo": 3.5693, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.338024865118461}
{"step": 56925, "loss": 3.5616554260253905, "grad_norm": 0.7277445197105408, "learning_rate": 0.0001, "elbo": 3.56166, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3383180858550316}
{"step": 56930, "loss": 3.5795494079589845, "grad_norm": 0.7124346494674683, "learning_rate": 0.0001, "elbo": 3.57955, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.338611306591602}
{"step": 56935, "loss": 3.5520404815673827, "grad_norm": 0.7045832276344299, "learning_rate": 0.0001, "elbo": 3.55204, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3389045273281726}
{"step": 56940, "loss": 3.5559154510498048, "grad_norm": 0.6696414351463318, "learning_rate": 0.0001, "elbo": 3.55592, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.339197748064743}
{"step": 56945, "loss": 3.5508934020996095, "grad_norm": 1.200421929359436, "learning_rate": 0.0001, "elbo": 3.55089, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3394909688013135}
{"step": 56950, "loss": 3.548567199707031, "grad_norm": 0.9756233096122742, "learning_rate": 0.0001, "elbo": 3.54857, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.339784189537884}
{"step": 56955, "loss": 3.558222198486328, "grad_norm": 0.7048570513725281, "learning_rate": 0.0001, "elbo": 3.55822, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3400774102744544}
{"step": 56960, "loss": 3.571250152587891, "grad_norm": 0.6986525654792786, "learning_rate": 0.0001, "elbo": 3.57125, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3403706310110253}
{"step": 56965, "loss": 3.5698226928710937, "grad_norm": 0.8513568043708801, "learning_rate": 0.0001, "elbo": 3.56982, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3406638517475957}
{"step": 56970, "loss": 3.5746780395507813, "grad_norm": 0.7045423984527588, "learning_rate": 0.0001, "elbo": 3.57468, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.340957072484166}
{"step": 56975, "loss": 3.5603736877441405, "grad_norm": 0.6720499396324158, "learning_rate": 0.0001, "elbo": 3.56037, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3412502932207366}
{"step": 56980, "loss": 3.568543243408203, "grad_norm": 0.6721798777580261, "learning_rate": 0.0001, "elbo": 3.56854, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.341543513957307}
{"step": 56985, "loss": 3.574188995361328, "grad_norm": 0.6972417235374451, "learning_rate": 0.0001, "elbo": 3.57419, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3418367346938775}
{"step": 56990, "loss": 3.5687652587890626, "grad_norm": 0.6882370114326477, "learning_rate": 0.0001, "elbo": 3.56877, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.342129955430448}
{"step": 56995, "loss": 3.5518115997314452, "grad_norm": 0.6777524948120117, "learning_rate": 0.0001, "elbo": 3.55181, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3424231761670185}
{"step": 57000, "loss": 3.5751380920410156, "grad_norm": 0.6666369438171387, "learning_rate": 0.0001, "elbo": 3.57514, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.342716396903589}
{"step": 57005, "loss": 3.559336853027344, "grad_norm": 0.6878275275230408, "learning_rate": 0.0001, "elbo": 3.55934, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3430096176401594}
{"step": 57010, "loss": 3.5508514404296876, "grad_norm": 0.7785510420799255, "learning_rate": 0.0001, "elbo": 3.55085, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.34330283837673}
{"step": 57015, "loss": 3.554860305786133, "grad_norm": 0.7678015232086182, "learning_rate": 0.0001, "elbo": 3.55486, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3435960591133007}
{"step": 57020, "loss": 3.5573829650878905, "grad_norm": 0.8496002554893494, "learning_rate": 0.0001, "elbo": 3.55738, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.343889279849871}
{"step": 57025, "loss": 3.560161590576172, "grad_norm": 0.6876189112663269, "learning_rate": 0.0001, "elbo": 3.56016, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3441825005864416}
{"step": 57025, "loss": 3.560161590576172, "grad_norm": 0.6876189112663269, "learning_rate": 0.0001, "elbo": 3.56016, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3441825005864416}
{"step": 57030, "loss": 3.5550445556640624, "grad_norm": 0.7284418940544128, "learning_rate": 0.0001, "elbo": 3.55504, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.344475721323012}
{"step": 57035, "loss": 3.562786865234375, "grad_norm": 0.671251654624939, "learning_rate": 0.0001, "elbo": 3.56279, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3447689420595825}
{"step": 57040, "loss": 3.53739013671875, "grad_norm": 0.7391193509101868, "learning_rate": 0.0001, "elbo": 3.53739, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.345062162796153}
{"step": 57045, "loss": 3.5487945556640623, "grad_norm": 0.7001640200614929, "learning_rate": 0.0001, "elbo": 3.54879, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3453553835327234}
{"step": 57050, "loss": 3.555744171142578, "grad_norm": 0.8990784883499146, "learning_rate": 0.0001, "elbo": 3.55574, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.345648604269294}
{"step": 57055, "loss": 3.5612937927246096, "grad_norm": 0.7060675024986267, "learning_rate": 0.0001, "elbo": 3.56129, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3459418250058643}
{"step": 57060, "loss": 3.5738685607910154, "grad_norm": 0.7184357047080994, "learning_rate": 0.0001, "elbo": 3.57387, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.346235045742435}
{"step": 57065, "loss": 3.5658123016357424, "grad_norm": 0.7227848768234253, "learning_rate": 0.0001, "elbo": 3.56581, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3465282664790053}
{"step": 57070, "loss": 3.555707550048828, "grad_norm": 0.8934095501899719, "learning_rate": 0.0001, "elbo": 3.55571, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3468214872155757}
{"step": 57075, "loss": 3.5563209533691404, "grad_norm": 0.8187363743782043, "learning_rate": 0.0001, "elbo": 3.55632, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.347114707952146}
{"step": 57080, "loss": 3.5632225036621095, "grad_norm": 0.6586480140686035, "learning_rate": 0.0001, "elbo": 3.56322, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.347407928688717}
{"step": 57085, "loss": 3.560848617553711, "grad_norm": 0.7277722954750061, "learning_rate": 0.0001, "elbo": 3.56085, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3477011494252875}
{"step": 57090, "loss": 3.5847015380859375, "grad_norm": 0.7748512625694275, "learning_rate": 0.0001, "elbo": 3.5847, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.347994370161858}
{"step": 57095, "loss": 3.560158538818359, "grad_norm": 0.7190665602684021, "learning_rate": 0.0001, "elbo": 3.56016, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3482875908984284}
{"step": 57100, "loss": 3.553516387939453, "grad_norm": 0.6997724175453186, "learning_rate": 0.0001, "elbo": 3.55352, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.348580811634999}
{"step": 57105, "loss": 3.5699195861816406, "grad_norm": 0.6756231188774109, "learning_rate": 0.0001, "elbo": 3.56992, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3488740323715693}
{"step": 57110, "loss": 3.554198455810547, "grad_norm": 0.7247840166091919, "learning_rate": 0.0001, "elbo": 3.5542, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.34916725310814}
{"step": 57115, "loss": 3.5579383850097654, "grad_norm": 0.7655575275421143, "learning_rate": 0.0001, "elbo": 3.55794, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3494604738447102}
{"step": 57120, "loss": 3.562881851196289, "grad_norm": 0.6972232460975647, "learning_rate": 0.0001, "elbo": 3.56288, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3497536945812807}
{"step": 57125, "loss": 3.5464714050292967, "grad_norm": 0.6873944997787476, "learning_rate": 0.0001, "elbo": 3.54647, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.350046915317851}
{"step": 57130, "loss": 3.544184112548828, "grad_norm": 0.6779519319534302, "learning_rate": 0.0001, "elbo": 3.54418, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3503401360544216}
{"step": 57135, "loss": 3.5471839904785156, "grad_norm": 0.6852023601531982, "learning_rate": 0.0001, "elbo": 3.54718, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3506333567909925}
{"step": 57140, "loss": 3.542316436767578, "grad_norm": 0.690089762210846, "learning_rate": 0.0001, "elbo": 3.54232, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.350926577527563}
{"step": 57145, "loss": 3.571035385131836, "grad_norm": 0.6872338056564331, "learning_rate": 0.0001, "elbo": 3.57104, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3512197982641334}
{"step": 57150, "loss": 3.5612861633300783, "grad_norm": 0.8876450061798096, "learning_rate": 0.0001, "elbo": 3.56129, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.351513019000704}
{"step": 57155, "loss": 3.570269775390625, "grad_norm": 0.6786047220230103, "learning_rate": 0.0001, "elbo": 3.57027, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3518062397372743}
{"step": 57160, "loss": 3.5385162353515627, "grad_norm": 0.7140464186668396, "learning_rate": 0.0001, "elbo": 3.53852, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3520994604738448}
{"step": 57165, "loss": 3.563579559326172, "grad_norm": 0.7518627047538757, "learning_rate": 0.0001, "elbo": 3.56358, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3523926812104152}
{"step": 57170, "loss": 3.5574390411376955, "grad_norm": 0.6957663297653198, "learning_rate": 0.0001, "elbo": 3.55744, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3526859019469857}
{"step": 57175, "loss": 3.5610008239746094, "grad_norm": 0.6659530997276306, "learning_rate": 0.0001, "elbo": 3.561, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.352979122683556}
{"step": 57180, "loss": 3.5480987548828127, "grad_norm": 0.6551220417022705, "learning_rate": 0.0001, "elbo": 3.5481, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3532723434201266}
{"step": 57185, "loss": 3.5558708190917967, "grad_norm": 0.6628631353378296, "learning_rate": 0.0001, "elbo": 3.55587, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.353565564156697}
{"step": 57190, "loss": 3.5653404235839843, "grad_norm": 0.6757544279098511, "learning_rate": 0.0001, "elbo": 3.56534, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3538587848932675}
{"step": 57195, "loss": 3.55611572265625, "grad_norm": 0.7856950759887695, "learning_rate": 0.0001, "elbo": 3.55612, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.354152005629838}
{"step": 57200, "loss": 3.530912399291992, "grad_norm": 0.7342948317527771, "learning_rate": 0.0001, "elbo": 3.53091, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3544452263664084}
{"step": 57205, "loss": 3.556976318359375, "grad_norm": 0.9996392726898193, "learning_rate": 0.0001, "elbo": 3.55698, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3547384471029793}
{"step": 57210, "loss": 3.555072021484375, "grad_norm": 0.7004933953285217, "learning_rate": 0.0001, "elbo": 3.55507, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3550316678395498}
{"step": 57215, "loss": 3.563153839111328, "grad_norm": 0.6680963039398193, "learning_rate": 0.0001, "elbo": 3.56315, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.35532488857612}
{"step": 57220, "loss": 3.5569629669189453, "grad_norm": 0.7735221982002258, "learning_rate": 0.0001, "elbo": 3.55696, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3556181093126907}
{"step": 57225, "loss": 3.5600330352783205, "grad_norm": 0.7290048003196716, "learning_rate": 0.0001, "elbo": 3.56003, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.355911330049261}
{"step": 57230, "loss": 3.5641075134277345, "grad_norm": 0.7192227840423584, "learning_rate": 0.0001, "elbo": 3.56411, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3562045507858316}
{"step": 57235, "loss": 3.5619880676269533, "grad_norm": 0.7573182582855225, "learning_rate": 0.0001, "elbo": 3.56199, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.356497771522402}
{"step": 57240, "loss": 3.5636886596679687, "grad_norm": 0.6570965051651001, "learning_rate": 0.0001, "elbo": 3.56369, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3567909922589725}
{"step": 57245, "loss": 3.546944427490234, "grad_norm": 0.655109703540802, "learning_rate": 0.0001, "elbo": 3.54694, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.357084212995543}
{"step": 57250, "loss": 3.5381919860839846, "grad_norm": 0.9029128551483154, "learning_rate": 0.0001, "elbo": 3.53819, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3573774337321134}
{"step": 57255, "loss": 3.543743896484375, "grad_norm": 0.8273648023605347, "learning_rate": 0.0001, "elbo": 3.54374, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.357670654468684}
{"step": 57260, "loss": 3.545252227783203, "grad_norm": 0.6788288354873657, "learning_rate": 0.0001, "elbo": 3.54525, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3579638752052547}
{"step": 57265, "loss": 3.567919921875, "grad_norm": 0.6872929334640503, "learning_rate": 0.0001, "elbo": 3.56792, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.358257095941825}
{"step": 57270, "loss": 3.549906921386719, "grad_norm": 0.7024210095405579, "learning_rate": 0.0001, "elbo": 3.54991, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3585503166783957}
{"step": 57275, "loss": 3.572321319580078, "grad_norm": 0.7028018832206726, "learning_rate": 0.0001, "elbo": 3.57232, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.358843537414966}
{"step": 57280, "loss": 3.568651580810547, "grad_norm": 0.716781497001648, "learning_rate": 0.0001, "elbo": 3.56865, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3591367581515366}
{"step": 57285, "loss": 3.5529190063476563, "grad_norm": 0.6693084239959717, "learning_rate": 0.0001, "elbo": 3.55292, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.359429978888107}
{"step": 57290, "loss": 3.5577091217041015, "grad_norm": 0.723803699016571, "learning_rate": 0.0001, "elbo": 3.55771, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3597231996246775}
{"step": 57295, "loss": 3.5483299255371095, "grad_norm": 0.6839653253555298, "learning_rate": 0.0001, "elbo": 3.54833, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.360016420361248}
{"step": 57300, "loss": 3.564625549316406, "grad_norm": 0.8087407946586609, "learning_rate": 0.0001, "elbo": 3.56463, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3603096410978184}
{"step": 57305, "loss": 3.5583332061767576, "grad_norm": 0.6803871393203735, "learning_rate": 0.0001, "elbo": 3.55833, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.360602861834389}
{"step": 57310, "loss": 3.5668258666992188, "grad_norm": 0.7298595905303955, "learning_rate": 0.0001, "elbo": 3.56683, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3608960825709593}
{"step": 57315, "loss": 3.569366455078125, "grad_norm": 0.7336028218269348, "learning_rate": 0.0001, "elbo": 3.56937, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3611893033075297}
{"step": 57315, "loss": 3.569366455078125, "grad_norm": 0.7336028218269348, "learning_rate": 0.0001, "elbo": 3.56937, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3611893033075297}
{"step": 57320, "loss": 3.552125930786133, "grad_norm": 0.7044371962547302, "learning_rate": 0.0001, "elbo": 3.55213, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3614825240441}
{"step": 57325, "loss": 3.5566970825195314, "grad_norm": 0.6970503926277161, "learning_rate": 0.0001, "elbo": 3.5567, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3617757447806706}
{"step": 57330, "loss": 3.566541290283203, "grad_norm": 0.8684537410736084, "learning_rate": 0.0001, "elbo": 3.56654, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3620689655172415}
{"step": 57335, "loss": 3.559009552001953, "grad_norm": 1.0775775909423828, "learning_rate": 0.0001, "elbo": 3.55901, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.362362186253812}
{"step": 57340, "loss": 3.5565200805664063, "grad_norm": 0.691051721572876, "learning_rate": 0.0001, "elbo": 3.55652, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3626554069903825}
{"step": 57345, "loss": 3.5545467376708983, "grad_norm": 0.6818177700042725, "learning_rate": 0.0001, "elbo": 3.55455, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.362948627726953}
{"step": 57350, "loss": 3.537134552001953, "grad_norm": 0.7969956398010254, "learning_rate": 0.0001, "elbo": 3.53713, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3632418484635234}
{"step": 57355, "loss": 3.558802032470703, "grad_norm": 0.651343584060669, "learning_rate": 0.0001, "elbo": 3.5588, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.363535069200094}
{"step": 57360, "loss": 3.573053741455078, "grad_norm": 0.6594053506851196, "learning_rate": 0.0001, "elbo": 3.57305, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3638282899366643}
{"step": 57365, "loss": 3.5488433837890625, "grad_norm": 0.680414617061615, "learning_rate": 0.0001, "elbo": 3.54884, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3641215106732347}
{"step": 57370, "loss": 3.558371734619141, "grad_norm": 0.6649258136749268, "learning_rate": 0.0001, "elbo": 3.55837, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.364414731409805}
{"step": 57375, "loss": 3.5641265869140626, "grad_norm": 0.689739465713501, "learning_rate": 0.0001, "elbo": 3.56413, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3647079521463756}
{"step": 57380, "loss": 3.5726600646972657, "grad_norm": 0.7483814358711243, "learning_rate": 0.0001, "elbo": 3.57266, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.365001172882946}
{"step": 57385, "loss": 3.5700912475585938, "grad_norm": 0.7146350145339966, "learning_rate": 0.0001, "elbo": 3.57009, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.365294393619517}
{"step": 57390, "loss": 3.5687835693359373, "grad_norm": 0.6892557740211487, "learning_rate": 0.0001, "elbo": 3.56878, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3655876143560874}
{"step": 57395, "loss": 3.554468536376953, "grad_norm": 0.8438229560852051, "learning_rate": 0.0001, "elbo": 3.55447, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.365880835092658}
{"step": 57400, "loss": 3.553640365600586, "grad_norm": 0.6712538003921509, "learning_rate": 0.0001, "elbo": 3.55364, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3661740558292284}
{"step": 57405, "loss": 3.5631668090820314, "grad_norm": 0.7389649748802185, "learning_rate": 0.0001, "elbo": 3.56317, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.366467276565799}
{"step": 57410, "loss": 3.5669586181640627, "grad_norm": 0.7786416411399841, "learning_rate": 0.0001, "elbo": 3.56696, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3667604973023693}
{"step": 57415, "loss": 3.5549697875976562, "grad_norm": 0.6489974856376648, "learning_rate": 0.0001, "elbo": 3.55497, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3670537180389397}
{"step": 57420, "loss": 3.5604496002197266, "grad_norm": 0.7202868461608887, "learning_rate": 0.0001, "elbo": 3.56045, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.36734693877551}
{"step": 57425, "loss": 3.5578586578369142, "grad_norm": 0.8079659342765808, "learning_rate": 0.0001, "elbo": 3.55786, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3676401595120806}
{"step": 57430, "loss": 3.5728469848632813, "grad_norm": 0.7997249364852905, "learning_rate": 0.0001, "elbo": 3.57285, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.367933380248651}
{"step": 57435, "loss": 3.573577117919922, "grad_norm": 0.7794562578201294, "learning_rate": 0.0001, "elbo": 3.57358, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3682266009852215}
{"step": 57440, "loss": 3.563650131225586, "grad_norm": 0.7407667636871338, "learning_rate": 0.0001, "elbo": 3.56365, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.368519821721792}
{"step": 57445, "loss": 3.555191421508789, "grad_norm": 0.8882958889007568, "learning_rate": 0.0001, "elbo": 3.55519, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3688130424583624}
{"step": 57450, "loss": 3.570995330810547, "grad_norm": 0.7048227787017822, "learning_rate": 0.0001, "elbo": 3.571, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3691062631949333}
{"step": 57455, "loss": 3.5527599334716795, "grad_norm": 0.6904202103614807, "learning_rate": 0.0001, "elbo": 3.55276, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.369399483931504}
{"step": 57460, "loss": 3.5404312133789064, "grad_norm": 0.8082515597343445, "learning_rate": 0.0001, "elbo": 3.54043, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3696927046680742}
{"step": 57465, "loss": 3.578624725341797, "grad_norm": 0.8837271928787231, "learning_rate": 0.0001, "elbo": 3.57862, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3699859254046447}
{"step": 57470, "loss": 3.5534347534179687, "grad_norm": 0.6674989461898804, "learning_rate": 0.0001, "elbo": 3.55343, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.370279146141215}
{"step": 57475, "loss": 3.5638053894042967, "grad_norm": 0.7478354573249817, "learning_rate": 0.0001, "elbo": 3.56381, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3705723668777856}
{"step": 57480, "loss": 3.5618354797363283, "grad_norm": 1.1356863975524902, "learning_rate": 0.0001, "elbo": 3.56184, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.370865587614356}
{"step": 57485, "loss": 3.5468231201171876, "grad_norm": 0.8934560418128967, "learning_rate": 0.0001, "elbo": 3.54682, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3711588083509265}
{"step": 57490, "loss": 3.557550048828125, "grad_norm": 0.8920041918754578, "learning_rate": 0.0001, "elbo": 3.55755, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.371452029087497}
{"step": 57495, "loss": 3.5604812622070314, "grad_norm": 0.651430070400238, "learning_rate": 0.0001, "elbo": 3.56048, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3717452498240674}
{"step": 57500, "loss": 3.5482513427734377, "grad_norm": 0.7342904806137085, "learning_rate": 0.0001, "elbo": 3.54825, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.372038470560638}
{"step": 57505, "loss": 3.5548572540283203, "grad_norm": 1.1050673723220825, "learning_rate": 0.0001, "elbo": 3.55486, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3723316912972088}
{"step": 57510, "loss": 3.548541259765625, "grad_norm": 0.6541875004768372, "learning_rate": 0.0001, "elbo": 3.54854, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3726249120337792}
{"step": 57515, "loss": 3.5440132141113283, "grad_norm": 0.7546998262405396, "learning_rate": 0.0001, "elbo": 3.54401, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3729181327703497}
{"step": 57520, "loss": 3.5554161071777344, "grad_norm": 0.6765900254249573, "learning_rate": 0.0001, "elbo": 3.55542, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.37321135350692}
{"step": 57525, "loss": 3.567150115966797, "grad_norm": 0.9965264797210693, "learning_rate": 0.0001, "elbo": 3.56715, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3735045742434906}
{"step": 57530, "loss": 3.575644683837891, "grad_norm": 0.7123537063598633, "learning_rate": 0.0001, "elbo": 3.57564, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.373797794980061}
{"step": 57535, "loss": 3.564766693115234, "grad_norm": 0.6788949966430664, "learning_rate": 0.0001, "elbo": 3.56477, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3740910157166315}
{"step": 57540, "loss": 3.5707828521728517, "grad_norm": 0.7249955534934998, "learning_rate": 0.0001, "elbo": 3.57078, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.374384236453202}
{"step": 57545, "loss": 3.5609695434570314, "grad_norm": 0.6963655948638916, "learning_rate": 0.0001, "elbo": 3.56097, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3746774571897724}
{"step": 57550, "loss": 3.574590301513672, "grad_norm": 0.7072802186012268, "learning_rate": 0.0001, "elbo": 3.57459, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.374970677926343}
{"step": 57555, "loss": 3.566209411621094, "grad_norm": 0.7639380097389221, "learning_rate": 0.0001, "elbo": 3.56621, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3752638986629133}
{"step": 57560, "loss": 3.571731185913086, "grad_norm": 0.8250190019607544, "learning_rate": 0.0001, "elbo": 3.57173, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3755571193994838}
{"step": 57565, "loss": 3.5430973052978514, "grad_norm": 0.6588225364685059, "learning_rate": 0.0001, "elbo": 3.5431, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3758503401360542}
{"step": 57570, "loss": 3.562314987182617, "grad_norm": 0.7395082116127014, "learning_rate": 0.0001, "elbo": 3.56232, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3761435608726247}
{"step": 57575, "loss": 3.5472637176513673, "grad_norm": 0.7444648146629333, "learning_rate": 0.0001, "elbo": 3.54726, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3764367816091956}
{"step": 57580, "loss": 3.5635807037353517, "grad_norm": 0.7452011704444885, "learning_rate": 0.0001, "elbo": 3.56358, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.376730002345766}
{"step": 57585, "loss": 3.5758800506591797, "grad_norm": 0.6718188524246216, "learning_rate": 0.0001, "elbo": 3.57588, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3770232230823365}
{"step": 57590, "loss": 3.5522529602050783, "grad_norm": 0.6299058198928833, "learning_rate": 0.0001, "elbo": 3.55225, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.377316443818907}
{"step": 57595, "loss": 3.548584747314453, "grad_norm": 0.6915322542190552, "learning_rate": 0.0001, "elbo": 3.54858, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3776096645554774}
{"step": 57600, "loss": 3.557086944580078, "grad_norm": 0.6876702308654785, "learning_rate": 0.0001, "elbo": 3.55709, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.377902885292048}
{"step": 57605, "loss": 3.5425346374511717, "grad_norm": 0.6754694581031799, "learning_rate": 0.0001, "elbo": 3.54253, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3781961060286183}
{"step": 57610, "loss": 3.5621471405029297, "grad_norm": 0.7024765014648438, "learning_rate": 0.0001, "elbo": 3.56215, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3784893267651888}
{"step": 57615, "loss": 3.568315124511719, "grad_norm": 0.7407827377319336, "learning_rate": 0.0001, "elbo": 3.56832, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.378782547501759}
{"step": 57620, "loss": 3.547105407714844, "grad_norm": 0.7067498564720154, "learning_rate": 0.0001, "elbo": 3.54711, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3790757682383297}
{"step": 57625, "loss": 3.5592109680175783, "grad_norm": 0.9903122186660767, "learning_rate": 0.0001, "elbo": 3.55921, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3793689889749}
{"step": 57630, "loss": 3.5551502227783205, "grad_norm": 0.694512128829956, "learning_rate": 0.0001, "elbo": 3.55515, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.379662209711471}
{"step": 57635, "loss": 3.575764846801758, "grad_norm": 0.8042615652084351, "learning_rate": 0.0001, "elbo": 3.57576, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3799554304480415}
{"step": 57640, "loss": 3.5551437377929687, "grad_norm": 0.8071399927139282, "learning_rate": 0.0001, "elbo": 3.55514, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.380248651184612}
{"step": 57645, "loss": 3.555231475830078, "grad_norm": 0.667144238948822, "learning_rate": 0.0001, "elbo": 3.55523, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3805418719211824}
{"step": 57650, "loss": 3.5632591247558594, "grad_norm": 0.673629879951477, "learning_rate": 0.0001, "elbo": 3.56326, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.380835092657753}
{"step": 57655, "loss": 3.577595520019531, "grad_norm": 0.6934192180633545, "learning_rate": 0.0001, "elbo": 3.5776, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3811283133943233}
{"step": 57660, "loss": 3.5583038330078125, "grad_norm": 0.7489389181137085, "learning_rate": 0.0001, "elbo": 3.5583, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3814215341308937}
{"step": 57660, "loss": 3.5583038330078125, "grad_norm": 0.7489389181137085, "learning_rate": 0.0001, "elbo": 3.5583, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3814215341308937}
{"step": 57665, "loss": 3.5385116577148437, "grad_norm": 0.7144345045089722, "learning_rate": 0.0001, "elbo": 3.53851, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.381714754867464}
{"step": 57670, "loss": 3.5661247253417967, "grad_norm": 0.7216898202896118, "learning_rate": 0.0001, "elbo": 3.56612, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3820079756040347}
{"step": 57675, "loss": 3.559813690185547, "grad_norm": 0.7980236411094666, "learning_rate": 0.0001, "elbo": 3.55981, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.382301196340605}
{"step": 57680, "loss": 3.567584228515625, "grad_norm": 0.6691491603851318, "learning_rate": 0.0001, "elbo": 3.56758, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3825944170771756}
{"step": 57685, "loss": 3.558573532104492, "grad_norm": 0.749927282333374, "learning_rate": 0.0001, "elbo": 3.55857, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.382887637813746}
{"step": 57690, "loss": 3.5628944396972657, "grad_norm": 0.7605962157249451, "learning_rate": 0.0001, "elbo": 3.56289, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3831808585503165}
{"step": 57695, "loss": 3.551683044433594, "grad_norm": 0.7108414769172668, "learning_rate": 0.0001, "elbo": 3.55168, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.383474079286887}
{"step": 57700, "loss": 3.5614036560058593, "grad_norm": 0.7270975708961487, "learning_rate": 0.0001, "elbo": 3.5614, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.383767300023458}
{"step": 57705, "loss": 3.568708038330078, "grad_norm": 0.7108666300773621, "learning_rate": 0.0001, "elbo": 3.56871, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3840605207600283}
{"step": 57710, "loss": 3.5559364318847657, "grad_norm": 0.6876630187034607, "learning_rate": 0.0001, "elbo": 3.55594, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3843537414965987}
{"step": 57715, "loss": 3.5554580688476562, "grad_norm": 0.7124261856079102, "learning_rate": 0.0001, "elbo": 3.55546, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.384646962233169}
{"step": 57720, "loss": 3.5771743774414064, "grad_norm": 0.6698286533355713, "learning_rate": 0.0001, "elbo": 3.57717, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3849401829697396}
{"step": 57725, "loss": 3.5533073425292967, "grad_norm": 0.6428592205047607, "learning_rate": 0.0001, "elbo": 3.55331, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.38523340370631}
{"step": 57725, "loss": 3.5533073425292967, "grad_norm": 0.6428592205047607, "learning_rate": 0.0001, "elbo": 3.55331, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.38523340370631}
{"step": 57730, "loss": 3.5654323577880858, "grad_norm": 0.6761758327484131, "learning_rate": 0.0001, "elbo": 3.56543, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3855266244428805}
{"step": 57735, "loss": 3.58228759765625, "grad_norm": 0.6816237568855286, "learning_rate": 0.0001, "elbo": 3.58229, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.385819845179451}
{"step": 57740, "loss": 3.564034652709961, "grad_norm": 0.7523243427276611, "learning_rate": 0.0001, "elbo": 3.56403, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3861130659160215}
{"step": 57745, "loss": 3.5737014770507813, "grad_norm": 0.6799706220626831, "learning_rate": 0.0001, "elbo": 3.5737, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.386406286652592}
{"step": 57745, "loss": 3.5737014770507813, "grad_norm": 0.6799706220626831, "learning_rate": 0.0001, "elbo": 3.5737, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.386406286652592}
{"step": 57750, "loss": 3.5602874755859375, "grad_norm": 0.6684004664421082, "learning_rate": 0.0001, "elbo": 3.56029, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.386699507389163}
{"step": 57755, "loss": 3.5547657012939453, "grad_norm": 0.6747609972953796, "learning_rate": 0.0001, "elbo": 3.55477, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3869927281257333}
{"step": 57760, "loss": 3.5667537689208983, "grad_norm": 0.8243337869644165, "learning_rate": 0.0001, "elbo": 3.56675, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3872859488623037}
{"step": 57765, "loss": 3.5489547729492186, "grad_norm": 0.7147829532623291, "learning_rate": 0.0001, "elbo": 3.54895, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.387579169598874}
{"step": 57770, "loss": 3.545488739013672, "grad_norm": 0.6873825192451477, "learning_rate": 0.0001, "elbo": 3.54549, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3878723903354446}
{"step": 57775, "loss": 3.559083938598633, "grad_norm": 0.7305142283439636, "learning_rate": 0.0001, "elbo": 3.55908, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.388165611072015}
{"step": 57780, "loss": 3.5505146026611327, "grad_norm": 0.7315555810928345, "learning_rate": 0.0001, "elbo": 3.55051, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3884588318085855}
{"step": 57785, "loss": 3.560272216796875, "grad_norm": 0.6982461810112, "learning_rate": 0.0001, "elbo": 3.56027, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.388752052545156}
{"step": 57790, "loss": 3.563831329345703, "grad_norm": 0.7176558375358582, "learning_rate": 0.0001, "elbo": 3.56383, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3890452732817264}
{"step": 57795, "loss": 3.5347618103027343, "grad_norm": 0.6800187826156616, "learning_rate": 0.0001, "elbo": 3.53476, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.389338494018297}
{"step": 57800, "loss": 3.546368408203125, "grad_norm": 0.9792927503585815, "learning_rate": 0.0001, "elbo": 3.54637, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3896317147548674}
{"step": 57805, "loss": 3.5923595428466797, "grad_norm": 1.0635349750518799, "learning_rate": 0.0001, "elbo": 3.59236, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.389924935491438}
{"step": 57810, "loss": 3.556201171875, "grad_norm": 0.7046578526496887, "learning_rate": 0.0001, "elbo": 3.5562, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3902181562280083}
{"step": 57815, "loss": 3.545307922363281, "grad_norm": 1.0313786268234253, "learning_rate": 0.0001, "elbo": 3.54531, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3905113769645787}
{"step": 57820, "loss": 3.553502655029297, "grad_norm": 0.7412582039833069, "learning_rate": 0.0001, "elbo": 3.5535, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3908045977011496}
{"step": 57825, "loss": 3.5409774780273438, "grad_norm": 0.7116574645042419, "learning_rate": 0.0001, "elbo": 3.54098, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.39109781843772}
{"step": 57830, "loss": 3.552611541748047, "grad_norm": 0.6614967584609985, "learning_rate": 0.0001, "elbo": 3.55261, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3913910391742905}
{"step": 57835, "loss": 3.557481384277344, "grad_norm": 0.6929231882095337, "learning_rate": 0.0001, "elbo": 3.55748, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.391684259910861}
{"step": 57840, "loss": 3.568825531005859, "grad_norm": 0.7141860723495483, "learning_rate": 0.0001, "elbo": 3.56883, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3919774806474314}
{"step": 57845, "loss": 3.5572441101074217, "grad_norm": 0.7165820002555847, "learning_rate": 0.0001, "elbo": 3.55724, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.392270701384002}
{"step": 57850, "loss": 3.554181671142578, "grad_norm": 0.7673075199127197, "learning_rate": 0.0001, "elbo": 3.55418, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3925639221205723}
{"step": 57855, "loss": 3.5578948974609377, "grad_norm": 0.7079187631607056, "learning_rate": 0.0001, "elbo": 3.55789, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.392857142857143}
{"step": 57860, "loss": 3.554804229736328, "grad_norm": 0.6895796656608582, "learning_rate": 0.0001, "elbo": 3.5548, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3931503635937132}
{"step": 57865, "loss": 3.5735591888427733, "grad_norm": 0.7054401636123657, "learning_rate": 0.0001, "elbo": 3.57356, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3934435843302837}
{"step": 57870, "loss": 3.561440277099609, "grad_norm": 0.6807212233543396, "learning_rate": 0.0001, "elbo": 3.56144, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.393736805066854}
{"step": 57875, "loss": 3.553916168212891, "grad_norm": 0.7553732991218567, "learning_rate": 0.0001, "elbo": 3.55392, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.394030025803425}
{"step": 57880, "loss": 3.559986114501953, "grad_norm": 0.8208699226379395, "learning_rate": 0.0001, "elbo": 3.55999, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3943232465399955}
{"step": 57885, "loss": 3.5628387451171877, "grad_norm": 1.0176113843917847, "learning_rate": 0.0001, "elbo": 3.56284, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.394616467276566}
{"step": 57890, "loss": 3.5455745697021483, "grad_norm": 0.697568416595459, "learning_rate": 0.0001, "elbo": 3.54557, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3949096880131364}
{"step": 57895, "loss": 3.5461769104003906, "grad_norm": 0.700722336769104, "learning_rate": 0.0001, "elbo": 3.54618, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.395202908749707}
{"step": 57900, "loss": 3.561786651611328, "grad_norm": 0.7501123547554016, "learning_rate": 0.0001, "elbo": 3.56179, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3954961294862773}
{"step": 57905, "loss": 3.559416961669922, "grad_norm": 0.6728450655937195, "learning_rate": 0.0001, "elbo": 3.55942, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.395789350222848}
{"step": 57910, "loss": 3.5413578033447264, "grad_norm": 0.6735222339630127, "learning_rate": 0.0001, "elbo": 3.54136, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3960825709594182}
{"step": 57915, "loss": 3.5553268432617187, "grad_norm": 0.8461042642593384, "learning_rate": 0.0001, "elbo": 3.55533, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3963757916959887}
{"step": 57920, "loss": 3.548270416259766, "grad_norm": 0.6955400109291077, "learning_rate": 0.0001, "elbo": 3.54827, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.396669012432559}
{"step": 57925, "loss": 3.5464431762695314, "grad_norm": 0.8447873592376709, "learning_rate": 0.0001, "elbo": 3.54644, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3969622331691296}
{"step": 57930, "loss": 3.5462646484375, "grad_norm": 0.7067402601242065, "learning_rate": 0.0001, "elbo": 3.54626, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3972554539057}
{"step": 57935, "loss": 3.5631603240966796, "grad_norm": 1.0511010885238647, "learning_rate": 0.0001, "elbo": 3.56316, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3975486746422705}
{"step": 57940, "loss": 3.5617610931396486, "grad_norm": 0.747902512550354, "learning_rate": 0.0001, "elbo": 3.56176, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.397841895378841}
{"step": 57945, "loss": 3.5357311248779295, "grad_norm": 0.6908131241798401, "learning_rate": 0.0001, "elbo": 3.53573, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.398135116115412}
{"step": 57950, "loss": 3.5552490234375, "grad_norm": 1.3865467309951782, "learning_rate": 0.0001, "elbo": 3.55525, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3984283368519823}
{"step": 57955, "loss": 3.545964813232422, "grad_norm": 0.6468862295150757, "learning_rate": 0.0001, "elbo": 3.54596, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3987215575885528}
{"step": 57960, "loss": 3.5788848876953123, "grad_norm": 0.681578516960144, "learning_rate": 0.0001, "elbo": 3.57888, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.399014778325123}
{"step": 57965, "loss": 3.5642818450927733, "grad_norm": 0.6870055794715881, "learning_rate": 0.0001, "elbo": 3.56428, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3993079990616937}
{"step": 57970, "loss": 3.5659923553466797, "grad_norm": 0.6774486899375916, "learning_rate": 0.0001, "elbo": 3.56599, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.399601219798264}
{"step": 57975, "loss": 3.5499240875244142, "grad_norm": 0.7718704342842102, "learning_rate": 0.0001, "elbo": 3.54992, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.3998944405348346}
{"step": 57980, "loss": 3.5538406372070312, "grad_norm": 0.655829131603241, "learning_rate": 0.0001, "elbo": 3.55384, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.400187661271405}
{"step": 57985, "loss": 3.5532211303710937, "grad_norm": 0.6886131167411804, "learning_rate": 0.0001, "elbo": 3.55322, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4004808820079755}
{"step": 57990, "loss": 3.554848480224609, "grad_norm": 0.6584862470626831, "learning_rate": 0.0001, "elbo": 3.55485, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.400774102744546}
{"step": 57995, "loss": 3.555047607421875, "grad_norm": 0.7427956461906433, "learning_rate": 0.0001, "elbo": 3.55505, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4010673234811164}
{"step": 57995, "loss": 3.555047607421875, "grad_norm": 0.7427956461906433, "learning_rate": 0.0001, "elbo": 3.55505, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4010673234811164}
{"step": 58000, "loss": 3.5431251525878906, "grad_norm": 0.7197657227516174, "learning_rate": 0.0001, "elbo": 3.54313, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4013605442176873}
{"step": 58005, "loss": 3.557847595214844, "grad_norm": 0.7365754842758179, "learning_rate": 0.0001, "elbo": 3.55785, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4016537649542578}
{"step": 58010, "loss": 3.560314178466797, "grad_norm": 0.7977468371391296, "learning_rate": 0.0001, "elbo": 3.56031, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.401946985690828}
{"step": 58015, "loss": 3.5567790985107424, "grad_norm": 0.7094628810882568, "learning_rate": 0.0001, "elbo": 3.55678, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4022402064273987}
{"step": 58020, "loss": 3.551444244384766, "grad_norm": 0.7639268040657043, "learning_rate": 0.0001, "elbo": 3.55144, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.402533427163969}
{"step": 58025, "loss": 3.5536548614501955, "grad_norm": 0.7001312971115112, "learning_rate": 0.0001, "elbo": 3.55365, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4028266479005396}
{"step": 58030, "loss": 3.5574344635009765, "grad_norm": 0.6810432076454163, "learning_rate": 0.0001, "elbo": 3.55743, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.40311986863711}
{"step": 58035, "loss": 3.5518550872802734, "grad_norm": 0.6481006741523743, "learning_rate": 0.0001, "elbo": 3.55186, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4034130893736805}
{"step": 58040, "loss": 3.560810852050781, "grad_norm": 0.6680108904838562, "learning_rate": 0.0001, "elbo": 3.56081, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.403706310110251}
{"step": 58045, "loss": 3.554693603515625, "grad_norm": 0.6928247809410095, "learning_rate": 0.0001, "elbo": 3.55469, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4039995308468214}
{"step": 58050, "loss": 3.5570938110351564, "grad_norm": 0.7409119606018066, "learning_rate": 0.0001, "elbo": 3.55709, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.404292751583392}
{"step": 58050, "loss": 3.5570938110351564, "grad_norm": 0.7409119606018066, "learning_rate": 0.0001, "elbo": 3.55709, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.404292751583392}
{"step": 58055, "loss": 3.5676780700683595, "grad_norm": 0.6877477765083313, "learning_rate": 0.0001, "elbo": 3.56768, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4045859723199623}
{"step": 58060, "loss": 3.5449161529541016, "grad_norm": 0.686501681804657, "learning_rate": 0.0001, "elbo": 3.54492, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4048791930565327}
{"step": 58065, "loss": 3.55938720703125, "grad_norm": 0.7210977673530579, "learning_rate": 0.0001, "elbo": 3.55939, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.405172413793103}
{"step": 58070, "loss": 3.5603050231933593, "grad_norm": 0.7208223938941956, "learning_rate": 0.0001, "elbo": 3.5603, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.405465634529674}
{"step": 58075, "loss": 3.5545280456542967, "grad_norm": 0.6937981247901917, "learning_rate": 0.0001, "elbo": 3.55453, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4057588552662446}
{"step": 58080, "loss": 3.56226806640625, "grad_norm": 0.6600181460380554, "learning_rate": 0.0001, "elbo": 3.56227, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.406052076002815}
{"step": 58085, "loss": 3.5452835083007814, "grad_norm": 0.6966693997383118, "learning_rate": 0.0001, "elbo": 3.54528, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4063452967393855}
{"step": 58090, "loss": 3.5613113403320313, "grad_norm": 1.012834906578064, "learning_rate": 0.0001, "elbo": 3.56131, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.406638517475956}
{"step": 58095, "loss": 3.563446807861328, "grad_norm": 0.9350534081459045, "learning_rate": 0.0001, "elbo": 3.56345, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4069317382125264}
{"step": 58100, "loss": 3.5616722106933594, "grad_norm": 0.6840706467628479, "learning_rate": 0.0001, "elbo": 3.56167, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.407224958949097}
{"step": 58105, "loss": 3.552104949951172, "grad_norm": 0.6847455501556396, "learning_rate": 0.0001, "elbo": 3.5521, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4075181796856673}
{"step": 58110, "loss": 3.551483154296875, "grad_norm": 0.7463201880455017, "learning_rate": 0.0001, "elbo": 3.55148, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4078114004222377}
{"step": 58115, "loss": 3.544658660888672, "grad_norm": 0.7037579417228699, "learning_rate": 0.0001, "elbo": 3.54466, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.408104621158808}
{"step": 58120, "loss": 3.5711044311523437, "grad_norm": 0.6938067078590393, "learning_rate": 0.0001, "elbo": 3.5711, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.408397841895379}
{"step": 58125, "loss": 3.5458065032958985, "grad_norm": 0.6681879162788391, "learning_rate": 0.0001, "elbo": 3.54581, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4086910626319495}
{"step": 58130, "loss": 3.55814208984375, "grad_norm": 0.7368525862693787, "learning_rate": 0.0001, "elbo": 3.55814, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.40898428336852}
{"step": 58130, "loss": 3.55814208984375, "grad_norm": 0.7368525862693787, "learning_rate": 0.0001, "elbo": 3.55814, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.40898428336852}
{"step": 58135, "loss": 3.5599708557128906, "grad_norm": 0.7365512251853943, "learning_rate": 0.0001, "elbo": 3.55997, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4092775041050905}
{"step": 58140, "loss": 3.559794616699219, "grad_norm": 0.686712384223938, "learning_rate": 0.0001, "elbo": 3.55979, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.409570724841661}
{"step": 58145, "loss": 3.5681495666503906, "grad_norm": 0.7809984683990479, "learning_rate": 0.0001, "elbo": 3.56815, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4098639455782314}
{"step": 58150, "loss": 3.562725830078125, "grad_norm": 0.6812350749969482, "learning_rate": 0.0001, "elbo": 3.56273, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.410157166314802}
{"step": 58155, "loss": 3.5549079895019533, "grad_norm": 0.7155652046203613, "learning_rate": 0.0001, "elbo": 3.55491, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4104503870513723}
{"step": 58160, "loss": 3.5578819274902345, "grad_norm": 0.6686462163925171, "learning_rate": 0.0001, "elbo": 3.55788, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4107436077879427}
{"step": 58165, "loss": 3.555278778076172, "grad_norm": 0.6922915577888489, "learning_rate": 0.0001, "elbo": 3.55528, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.411036828524513}
{"step": 58170, "loss": 3.548188018798828, "grad_norm": 0.760645866394043, "learning_rate": 0.0001, "elbo": 3.54819, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4113300492610836}
{"step": 58175, "loss": 3.567708206176758, "grad_norm": 0.7101655602455139, "learning_rate": 0.0001, "elbo": 3.56771, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.411623269997654}
{"step": 58180, "loss": 3.5557811737060545, "grad_norm": 1.168461561203003, "learning_rate": 0.0001, "elbo": 3.55578, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4119164907342245}
{"step": 58185, "loss": 3.561909484863281, "grad_norm": 0.684146523475647, "learning_rate": 0.0001, "elbo": 3.56191, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.412209711470795}
{"step": 58185, "loss": 3.561909484863281, "grad_norm": 0.684146523475647, "learning_rate": 0.0001, "elbo": 3.56191, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.412209711470795}
{"step": 58190, "loss": 3.5510513305664064, "grad_norm": 0.6822232604026794, "learning_rate": 0.0001, "elbo": 3.55105, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.412502932207366}
{"step": 58195, "loss": 3.556182098388672, "grad_norm": 0.9392198920249939, "learning_rate": 0.0001, "elbo": 3.55618, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4127961529439363}
{"step": 58200, "loss": 3.540879821777344, "grad_norm": 0.6923622488975525, "learning_rate": 0.0001, "elbo": 3.54088, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.413089373680507}
{"step": 58205, "loss": 3.5636276245117187, "grad_norm": 0.699459969997406, "learning_rate": 0.0001, "elbo": 3.56363, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4133825944170773}
{"step": 58210, "loss": 3.5768123626708985, "grad_norm": 0.7256134748458862, "learning_rate": 0.0001, "elbo": 3.57681, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4136758151536477}
{"step": 58215, "loss": 3.5653934478759766, "grad_norm": 0.7202855944633484, "learning_rate": 0.0001, "elbo": 3.56539, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.413969035890218}
{"step": 58220, "loss": 3.5619346618652346, "grad_norm": 0.6834633946418762, "learning_rate": 0.0001, "elbo": 3.56193, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4142622566267886}
{"step": 58225, "loss": 3.549925994873047, "grad_norm": 0.809782862663269, "learning_rate": 0.0001, "elbo": 3.54993, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.414555477363359}
{"step": 58230, "loss": 3.566031265258789, "grad_norm": 0.691618800163269, "learning_rate": 0.0001, "elbo": 3.56603, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4148486980999295}
{"step": 58235, "loss": 3.552927017211914, "grad_norm": 0.7153083682060242, "learning_rate": 0.0001, "elbo": 3.55293, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4151419188365}
{"step": 58240, "loss": 3.559517669677734, "grad_norm": 0.7021186351776123, "learning_rate": 0.0001, "elbo": 3.55952, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4154351395730704}
{"step": 58245, "loss": 3.5734100341796875, "grad_norm": 0.7273426651954651, "learning_rate": 0.0001, "elbo": 3.57341, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4157283603096413}
{"step": 58250, "loss": 3.562194061279297, "grad_norm": 0.6883758902549744, "learning_rate": 0.0001, "elbo": 3.56219, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.416021581046212}
{"step": 58255, "loss": 3.5547271728515626, "grad_norm": 0.7143163084983826, "learning_rate": 0.0001, "elbo": 3.55473, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4163148017827822}
{"step": 58260, "loss": 3.558919143676758, "grad_norm": 0.7357825636863708, "learning_rate": 0.0001, "elbo": 3.55892, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4166080225193527}
{"step": 58265, "loss": 3.5594203948974608, "grad_norm": 0.8227221369743347, "learning_rate": 0.0001, "elbo": 3.55942, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.416901243255923}
{"step": 58270, "loss": 3.566476058959961, "grad_norm": 0.684140145778656, "learning_rate": 0.0001, "elbo": 3.56648, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4171944639924936}
{"step": 58275, "loss": 3.54068603515625, "grad_norm": 0.7289742827415466, "learning_rate": 0.0001, "elbo": 3.54069, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.417487684729064}
{"step": 58280, "loss": 3.5591896057128904, "grad_norm": 0.7592390179634094, "learning_rate": 0.0001, "elbo": 3.55919, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4177809054656345}
{"step": 58285, "loss": 3.540605163574219, "grad_norm": 0.7086852192878723, "learning_rate": 0.0001, "elbo": 3.54061, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.418074126202205}
{"step": 58290, "loss": 3.5569393157958986, "grad_norm": 0.7373051047325134, "learning_rate": 0.0001, "elbo": 3.55694, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4183673469387754}
{"step": 58295, "loss": 3.561534881591797, "grad_norm": 0.7142503261566162, "learning_rate": 0.0001, "elbo": 3.56154, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.418660567675346}
{"step": 58300, "loss": 3.5386993408203127, "grad_norm": 0.7271692156791687, "learning_rate": 0.0001, "elbo": 3.5387, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4189537884119163}
{"step": 58305, "loss": 3.5572978973388674, "grad_norm": 0.7586933970451355, "learning_rate": 0.0001, "elbo": 3.5573, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.419247009148487}
{"step": 58310, "loss": 3.5625545501708986, "grad_norm": 0.7850267291069031, "learning_rate": 0.0001, "elbo": 3.56255, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4195402298850572}
{"step": 58310, "loss": 3.5625545501708986, "grad_norm": 0.7850267291069031, "learning_rate": 0.0001, "elbo": 3.56255, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4195402298850572}
{"step": 58315, "loss": 3.5574989318847656, "grad_norm": 0.7869694828987122, "learning_rate": 0.0001, "elbo": 3.5575, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.419833450621628}
{"step": 58320, "loss": 3.5576515197753906, "grad_norm": 0.6975448727607727, "learning_rate": 0.0001, "elbo": 3.55765, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4201266713581986}
{"step": 58325, "loss": 3.549077606201172, "grad_norm": 0.7789697051048279, "learning_rate": 0.0001, "elbo": 3.54908, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.420419892094769}
{"step": 58330, "loss": 3.5398910522460936, "grad_norm": 0.6987757086753845, "learning_rate": 0.0001, "elbo": 3.53989, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4207131128313395}
{"step": 58335, "loss": 3.548492431640625, "grad_norm": 0.690105140209198, "learning_rate": 0.0001, "elbo": 3.54849, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.42100633356791}
{"step": 58340, "loss": 3.5496597290039062, "grad_norm": 0.7251065373420715, "learning_rate": 0.0001, "elbo": 3.54966, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4212995543044804}
{"step": 58345, "loss": 3.5587425231933594, "grad_norm": 0.7002195119857788, "learning_rate": 0.0001, "elbo": 3.55874, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.421592775041051}
{"step": 58350, "loss": 3.5681568145751954, "grad_norm": 0.783071756362915, "learning_rate": 0.0001, "elbo": 3.56816, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4218859957776213}
{"step": 58355, "loss": 3.5630752563476564, "grad_norm": 1.042206883430481, "learning_rate": 0.0001, "elbo": 3.56307, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4221792165141918}
{"step": 58360, "loss": 3.5614112854003905, "grad_norm": 0.6527202725410461, "learning_rate": 0.0001, "elbo": 3.56141, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4224724372507622}
{"step": 58365, "loss": 3.5706718444824217, "grad_norm": 0.7116274237632751, "learning_rate": 0.0001, "elbo": 3.57067, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4227656579873327}
{"step": 58370, "loss": 3.5585563659667967, "grad_norm": 0.7355973124504089, "learning_rate": 0.0001, "elbo": 3.55856, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4230588787239036}
{"step": 58375, "loss": 3.567675018310547, "grad_norm": 0.7789509296417236, "learning_rate": 0.0001, "elbo": 3.56768, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.423352099460474}
{"step": 58380, "loss": 3.5578819274902345, "grad_norm": 0.7125851511955261, "learning_rate": 0.0001, "elbo": 3.55788, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4236453201970445}
{"step": 58385, "loss": 3.559477996826172, "grad_norm": 0.7596051096916199, "learning_rate": 0.0001, "elbo": 3.55948, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.423938540933615}
{"step": 58390, "loss": 3.554804229736328, "grad_norm": 0.7485576272010803, "learning_rate": 0.0001, "elbo": 3.5548, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4242317616701854}
{"step": 58395, "loss": 3.5485191345214844, "grad_norm": 0.679578959941864, "learning_rate": 0.0001, "elbo": 3.54852, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.424524982406756}
{"step": 58400, "loss": 3.550701141357422, "grad_norm": 0.6946837902069092, "learning_rate": 0.0001, "elbo": 3.5507, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4248182031433263}
{"step": 58405, "loss": 3.5607040405273436, "grad_norm": 0.7070502042770386, "learning_rate": 0.0001, "elbo": 3.5607, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4251114238798968}
{"step": 58410, "loss": 3.564009094238281, "grad_norm": 0.7140207886695862, "learning_rate": 0.0001, "elbo": 3.56401, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.425404644616467}
{"step": 58415, "loss": 3.5330856323242186, "grad_norm": 0.6559897661209106, "learning_rate": 0.0001, "elbo": 3.53309, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4256978653530377}
{"step": 58420, "loss": 3.5589553833007814, "grad_norm": 0.7309849858283997, "learning_rate": 0.0001, "elbo": 3.55896, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.425991086089608}
{"step": 58425, "loss": 3.5452178955078124, "grad_norm": 0.6908763647079468, "learning_rate": 0.0001, "elbo": 3.54522, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4262843068261786}
{"step": 58430, "loss": 3.557697296142578, "grad_norm": 0.6967878937721252, "learning_rate": 0.0001, "elbo": 3.5577, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.426577527562749}
{"step": 58435, "loss": 3.555242156982422, "grad_norm": 0.6475088596343994, "learning_rate": 0.0001, "elbo": 3.55524, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4268707482993195}
{"step": 58440, "loss": 3.5622512817382814, "grad_norm": 0.673086404800415, "learning_rate": 0.0001, "elbo": 3.56225, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4271639690358904}
{"step": 58440, "loss": 3.5622512817382814, "grad_norm": 0.673086404800415, "learning_rate": 0.0001, "elbo": 3.56225, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4271639690358904}
{"step": 58445, "loss": 3.565143585205078, "grad_norm": 1.060441255569458, "learning_rate": 0.0001, "elbo": 3.56514, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.427457189772461}
{"step": 58450, "loss": 3.547430419921875, "grad_norm": 0.6765907406806946, "learning_rate": 0.0001, "elbo": 3.54743, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4277504105090313}
{"step": 58455, "loss": 3.5690940856933593, "grad_norm": 0.8130660057067871, "learning_rate": 0.0001, "elbo": 3.56909, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4280436312456017}
{"step": 58460, "loss": 3.5446754455566407, "grad_norm": 0.6933584213256836, "learning_rate": 0.0001, "elbo": 3.54468, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.428336851982172}
{"step": 58465, "loss": 3.5597763061523438, "grad_norm": 0.6802690029144287, "learning_rate": 0.0001, "elbo": 3.55978, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4286300727187426}
{"step": 58470, "loss": 3.560879135131836, "grad_norm": 0.6918153762817383, "learning_rate": 0.0001, "elbo": 3.56088, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.428923293455313}
{"step": 58475, "loss": 3.5602149963378906, "grad_norm": 0.6797125339508057, "learning_rate": 0.0001, "elbo": 3.56021, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4292165141918836}
{"step": 58480, "loss": 3.5531780242919924, "grad_norm": 0.8420781493186951, "learning_rate": 0.0001, "elbo": 3.55318, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.429509734928454}
{"step": 58485, "loss": 3.5507373809814453, "grad_norm": 0.6498050093650818, "learning_rate": 0.0001, "elbo": 3.55074, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4298029556650245}
{"step": 58490, "loss": 3.5538501739501953, "grad_norm": 0.6578094959259033, "learning_rate": 0.0001, "elbo": 3.55385, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4300961764015954}
{"step": 58495, "loss": 3.5602317810058595, "grad_norm": 0.7947867512702942, "learning_rate": 0.0001, "elbo": 3.56023, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.430389397138166}
{"step": 58500, "loss": 3.551660919189453, "grad_norm": 0.7149396538734436, "learning_rate": 0.0001, "elbo": 3.55166, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4306826178747363}
{"step": 58505, "loss": 3.562430191040039, "grad_norm": 0.7229892015457153, "learning_rate": 0.0001, "elbo": 3.56243, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4309758386113067}
{"step": 58510, "loss": 3.563800811767578, "grad_norm": 0.6989124417304993, "learning_rate": 0.0001, "elbo": 3.5638, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.431269059347877}
{"step": 58515, "loss": 3.5542186737060546, "grad_norm": 0.7189054489135742, "learning_rate": 0.0001, "elbo": 3.55422, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4315622800844476}
{"step": 58520, "loss": 3.5538875579833986, "grad_norm": 0.6702147126197815, "learning_rate": 0.0001, "elbo": 3.55389, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.431855500821018}
{"step": 58525, "loss": 3.552224349975586, "grad_norm": 0.6646218299865723, "learning_rate": 0.0001, "elbo": 3.55222, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4321487215575885}
{"step": 58530, "loss": 3.5464641571044924, "grad_norm": 0.7178872227668762, "learning_rate": 0.0001, "elbo": 3.54646, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.432441942294159}
{"step": 58535, "loss": 3.571881866455078, "grad_norm": 0.7137379050254822, "learning_rate": 0.0001, "elbo": 3.57188, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4327351630307295}
{"step": 58540, "loss": 3.5603004455566407, "grad_norm": 0.6630682349205017, "learning_rate": 0.0001, "elbo": 3.5603, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4330283837673}
{"step": 58545, "loss": 3.5503997802734375, "grad_norm": 0.8411538600921631, "learning_rate": 0.0001, "elbo": 3.5504, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4333216045038704}
{"step": 58550, "loss": 3.5626461029052736, "grad_norm": 0.8005650043487549, "learning_rate": 0.0001, "elbo": 3.56265, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.433614825240441}
{"step": 58555, "loss": 3.5426727294921876, "grad_norm": 0.6906574964523315, "learning_rate": 0.0001, "elbo": 3.54267, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4339080459770113}
{"step": 58560, "loss": 3.549250030517578, "grad_norm": 0.9570605158805847, "learning_rate": 0.0001, "elbo": 3.54925, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.434201266713582}
{"step": 58565, "loss": 3.5458335876464844, "grad_norm": 0.9881100654602051, "learning_rate": 0.0001, "elbo": 3.54583, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4344944874501526}
{"step": 58570, "loss": 3.550129699707031, "grad_norm": 0.7541899681091309, "learning_rate": 0.0001, "elbo": 3.55013, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.434787708186723}
{"step": 58575, "loss": 3.5535350799560548, "grad_norm": 0.684656023979187, "learning_rate": 0.0001, "elbo": 3.55353, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4350809289232935}
{"step": 58580, "loss": 3.549803924560547, "grad_norm": 0.6993165612220764, "learning_rate": 0.0001, "elbo": 3.5498, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.435374149659864}
{"step": 58585, "loss": 3.550641632080078, "grad_norm": 0.9103145003318787, "learning_rate": 0.0001, "elbo": 3.55064, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4356673703964344}
{"step": 58590, "loss": 3.5538978576660156, "grad_norm": 0.6822353005409241, "learning_rate": 0.0001, "elbo": 3.5539, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.435960591133005}
{"step": 58595, "loss": 3.549131393432617, "grad_norm": 0.7308610081672668, "learning_rate": 0.0001, "elbo": 3.54913, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4362538118695753}
{"step": 58600, "loss": 3.546225738525391, "grad_norm": 0.68470299243927, "learning_rate": 0.0001, "elbo": 3.54623, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.436547032606146}
{"step": 58600, "loss": 3.546225738525391, "grad_norm": 0.68470299243927, "learning_rate": 0.0001, "elbo": 3.54623, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.436547032606146}
{"step": 58605, "loss": 3.5574623107910157, "grad_norm": 0.893639326095581, "learning_rate": 0.0001, "elbo": 3.55746, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4368402533427163}
{"step": 58610, "loss": 3.5555751800537108, "grad_norm": 0.7407665848731995, "learning_rate": 0.0001, "elbo": 3.55558, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4371334740792867}
{"step": 58615, "loss": 3.5560672760009764, "grad_norm": 0.6924730539321899, "learning_rate": 0.0001, "elbo": 3.55607, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4374266948158576}
{"step": 58620, "loss": 3.55240478515625, "grad_norm": 0.6683436632156372, "learning_rate": 0.0001, "elbo": 3.5524, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.437719915552428}
{"step": 58625, "loss": 3.5517406463623047, "grad_norm": 1.2508121728897095, "learning_rate": 0.0001, "elbo": 3.55174, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4380131362889985}
{"step": 58630, "loss": 3.5541534423828125, "grad_norm": 0.8073762655258179, "learning_rate": 0.0001, "elbo": 3.55415, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.438306357025569}
{"step": 58635, "loss": 3.547382354736328, "grad_norm": 0.7299594879150391, "learning_rate": 0.0001, "elbo": 3.54738, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4385995777621394}
{"step": 58640, "loss": 3.5614269256591795, "grad_norm": 0.7644125819206238, "learning_rate": 0.0001, "elbo": 3.56143, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.43889279849871}
{"step": 58645, "loss": 3.5496147155761717, "grad_norm": 0.7045992016792297, "learning_rate": 0.0001, "elbo": 3.54962, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4391860192352803}
{"step": 58650, "loss": 3.551814651489258, "grad_norm": 0.6830912232398987, "learning_rate": 0.0001, "elbo": 3.55181, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.439479239971851}
{"step": 58655, "loss": 3.5542198181152345, "grad_norm": 0.7148744463920593, "learning_rate": 0.0001, "elbo": 3.55422, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4397724607084212}
{"step": 58660, "loss": 3.5514888763427734, "grad_norm": 0.7036123871803284, "learning_rate": 0.0001, "elbo": 3.55149, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4400656814449917}
{"step": 58665, "loss": 3.549481201171875, "grad_norm": 0.8073199391365051, "learning_rate": 0.0001, "elbo": 3.54948, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.440358902181562}
{"step": 58670, "loss": 3.5529518127441406, "grad_norm": 0.6753301024436951, "learning_rate": 0.0001, "elbo": 3.55295, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4406521229181326}
{"step": 58675, "loss": 3.5628387451171877, "grad_norm": 0.6796602606773376, "learning_rate": 0.0001, "elbo": 3.56284, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.440945343654703}
{"step": 58680, "loss": 3.563763427734375, "grad_norm": 0.7003058791160583, "learning_rate": 0.0001, "elbo": 3.56376, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4412385643912735}
{"step": 58685, "loss": 3.553050231933594, "grad_norm": 0.7100707292556763, "learning_rate": 0.0001, "elbo": 3.55305, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4415317851278444}
{"step": 58690, "loss": 3.5549575805664064, "grad_norm": 0.6992815136909485, "learning_rate": 0.0001, "elbo": 3.55496, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.441825005864415}
{"step": 58695, "loss": 3.5662364959716797, "grad_norm": 0.7441521883010864, "learning_rate": 0.0001, "elbo": 3.56624, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4421182266009853}
{"step": 58700, "loss": 3.5650894165039064, "grad_norm": 0.8084855079650879, "learning_rate": 0.0001, "elbo": 3.56509, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4424114473375558}
{"step": 58705, "loss": 3.537073516845703, "grad_norm": 0.7044907212257385, "learning_rate": 0.0001, "elbo": 3.53707, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4427046680741262}
{"step": 58710, "loss": 3.56225700378418, "grad_norm": 0.6711843609809875, "learning_rate": 0.0001, "elbo": 3.56226, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4429978888106967}
{"step": 58715, "loss": 3.551946258544922, "grad_norm": 0.9003037214279175, "learning_rate": 0.0001, "elbo": 3.55195, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.443291109547267}
{"step": 58720, "loss": 3.5672046661376955, "grad_norm": 0.7053080201148987, "learning_rate": 0.0001, "elbo": 3.5672, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4435843302838376}
{"step": 58725, "loss": 3.5558521270751955, "grad_norm": 0.6943953633308411, "learning_rate": 0.0001, "elbo": 3.55585, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.443877551020408}
{"step": 58730, "loss": 3.555219268798828, "grad_norm": 0.7392224669456482, "learning_rate": 0.0001, "elbo": 3.55522, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4441707717569785}
{"step": 58735, "loss": 3.544941711425781, "grad_norm": 0.7329788208007812, "learning_rate": 0.0001, "elbo": 3.54494, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.444463992493549}
{"step": 58735, "loss": 3.544941711425781, "grad_norm": 0.7329788208007812, "learning_rate": 0.0001, "elbo": 3.54494, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.444463992493549}
{"step": 58740, "loss": 3.5505142211914062, "grad_norm": 0.8512547612190247, "learning_rate": 0.0001, "elbo": 3.55051, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.44475721323012}
{"step": 58745, "loss": 3.5639263153076173, "grad_norm": 0.8237454891204834, "learning_rate": 0.0001, "elbo": 3.56393, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4450504339666903}
{"step": 58750, "loss": 3.5613075256347657, "grad_norm": 0.7002089023590088, "learning_rate": 0.0001, "elbo": 3.56131, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4453436547032608}
{"step": 58755, "loss": 3.553582763671875, "grad_norm": 0.6725953817367554, "learning_rate": 0.0001, "elbo": 3.55358, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.445636875439831}
{"step": 58760, "loss": 3.5543746948242188, "grad_norm": 0.7638955116271973, "learning_rate": 0.0001, "elbo": 3.55437, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4459300961764017}
{"step": 58765, "loss": 3.539617919921875, "grad_norm": 0.6600393056869507, "learning_rate": 0.0001, "elbo": 3.53962, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.446223316912972}
{"step": 58770, "loss": 3.557782745361328, "grad_norm": 0.7005050182342529, "learning_rate": 0.0001, "elbo": 3.55778, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4465165376495426}
{"step": 58775, "loss": 3.5513774871826174, "grad_norm": 0.7148239016532898, "learning_rate": 0.0001, "elbo": 3.55138, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.446809758386113}
{"step": 58780, "loss": 3.535248565673828, "grad_norm": 0.6637187004089355, "learning_rate": 0.0001, "elbo": 3.53525, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4471029791226835}
{"step": 58785, "loss": 3.562632751464844, "grad_norm": 0.6901348233222961, "learning_rate": 0.0001, "elbo": 3.56263, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.447396199859254}
{"step": 58790, "loss": 3.5597305297851562, "grad_norm": 0.7913781404495239, "learning_rate": 0.0001, "elbo": 3.55973, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4476894205958244}
{"step": 58795, "loss": 3.5595657348632814, "grad_norm": 0.6575794219970703, "learning_rate": 0.0001, "elbo": 3.55957, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.447982641332395}
{"step": 58800, "loss": 3.5455135345458983, "grad_norm": 0.7385984063148499, "learning_rate": 0.0001, "elbo": 3.54551, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4482758620689653}
{"step": 58805, "loss": 3.5743812561035155, "grad_norm": 0.6698121428489685, "learning_rate": 0.0001, "elbo": 3.57438, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.448569082805536}
{"step": 58810, "loss": 3.5809120178222655, "grad_norm": 0.6760990619659424, "learning_rate": 0.0001, "elbo": 3.58091, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4488623035421067}
{"step": 58815, "loss": 3.5580322265625, "grad_norm": 0.7110137343406677, "learning_rate": 0.0001, "elbo": 3.55803, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.449155524278677}
{"step": 58820, "loss": 3.549103546142578, "grad_norm": 0.6786755919456482, "learning_rate": 0.0001, "elbo": 3.5491, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4494487450152476}
{"step": 58825, "loss": 3.5444313049316407, "grad_norm": 0.7282513380050659, "learning_rate": 0.0001, "elbo": 3.54443, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.449741965751818}
{"step": 58830, "loss": 3.5491031646728515, "grad_norm": 0.7212134599685669, "learning_rate": 0.0001, "elbo": 3.5491, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4500351864883885}
{"step": 58835, "loss": 3.5565181732177735, "grad_norm": 0.7137324810028076, "learning_rate": 0.0001, "elbo": 3.55652, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.450328407224959}
{"step": 58840, "loss": 3.540782165527344, "grad_norm": 0.6802271008491516, "learning_rate": 0.0001, "elbo": 3.54078, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4506216279615294}
{"step": 58845, "loss": 3.5525562286376955, "grad_norm": 0.6896800398826599, "learning_rate": 0.0001, "elbo": 3.55256, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4509148486981}
{"step": 58850, "loss": 3.5730308532714843, "grad_norm": 0.7048074007034302, "learning_rate": 0.0001, "elbo": 3.57303, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4512080694346703}
{"step": 58855, "loss": 3.551194763183594, "grad_norm": 0.7478089928627014, "learning_rate": 0.0001, "elbo": 3.55119, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4515012901712407}
{"step": 58860, "loss": 3.5630489349365235, "grad_norm": 0.7085444331169128, "learning_rate": 0.0001, "elbo": 3.56305, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4517945109078116}
{"step": 58865, "loss": 3.5689998626708985, "grad_norm": 0.8396775126457214, "learning_rate": 0.0001, "elbo": 3.569, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.452087731644382}
{"step": 58870, "loss": 3.54716796875, "grad_norm": 0.666122317314148, "learning_rate": 0.0001, "elbo": 3.54717, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4523809523809526}
{"step": 58875, "loss": 3.546727752685547, "grad_norm": 0.8695266842842102, "learning_rate": 0.0001, "elbo": 3.54673, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.452674173117523}
{"step": 58880, "loss": 3.5629981994628905, "grad_norm": 0.6984922885894775, "learning_rate": 0.0001, "elbo": 3.563, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4529673938540935}
{"step": 58885, "loss": 3.5632530212402345, "grad_norm": 0.7211897373199463, "learning_rate": 0.0001, "elbo": 3.56325, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.453260614590664}
{"step": 58890, "loss": 3.5609344482421874, "grad_norm": 0.6749489307403564, "learning_rate": 0.0001, "elbo": 3.56093, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4535538353272344}
{"step": 58895, "loss": 3.5583038330078125, "grad_norm": 0.6914863586425781, "learning_rate": 0.0001, "elbo": 3.5583, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.453847056063805}
{"step": 58895, "loss": 3.5583038330078125, "grad_norm": 0.6914863586425781, "learning_rate": 0.0001, "elbo": 3.5583, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.453847056063805}
{"step": 58900, "loss": 3.562969207763672, "grad_norm": 0.7088289856910706, "learning_rate": 0.0001, "elbo": 3.56297, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4541402768003753}
{"step": 58905, "loss": 3.5579269409179686, "grad_norm": 0.6974284648895264, "learning_rate": 0.0001, "elbo": 3.55793, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4544334975369457}
{"step": 58910, "loss": 3.546339416503906, "grad_norm": 0.6864500641822815, "learning_rate": 0.0001, "elbo": 3.54634, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.454726718273516}
{"step": 58915, "loss": 3.54449462890625, "grad_norm": 0.6857253909111023, "learning_rate": 0.0001, "elbo": 3.54449, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4550199390100866}
{"step": 58920, "loss": 3.559235382080078, "grad_norm": 0.7120055556297302, "learning_rate": 0.0001, "elbo": 3.55924, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.455313159746657}
{"step": 58925, "loss": 3.5621334075927735, "grad_norm": 0.7108193039894104, "learning_rate": 0.0001, "elbo": 3.56213, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4556063804832275}
{"step": 58930, "loss": 3.5528800964355467, "grad_norm": 0.6770282983779907, "learning_rate": 0.0001, "elbo": 3.55288, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4558996012197984}
{"step": 58935, "loss": 3.544413375854492, "grad_norm": 0.6739916205406189, "learning_rate": 0.0001, "elbo": 3.54441, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.456192821956369}
{"step": 58940, "loss": 3.5439605712890625, "grad_norm": 0.7014084458351135, "learning_rate": 0.0001, "elbo": 3.54396, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4564860426929394}
{"step": 58945, "loss": 3.565038299560547, "grad_norm": 0.6702946424484253, "learning_rate": 0.0001, "elbo": 3.56504, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.45677926342951}
{"step": 58950, "loss": 3.54124755859375, "grad_norm": 0.7087289690971375, "learning_rate": 0.0001, "elbo": 3.54125, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4570724841660803}
{"step": 58955, "loss": 3.545989227294922, "grad_norm": 0.6650399565696716, "learning_rate": 0.0001, "elbo": 3.54599, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4573657049026507}
{"step": 58960, "loss": 3.557259368896484, "grad_norm": 0.6664719581604004, "learning_rate": 0.0001, "elbo": 3.55726, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.457658925639221}
{"step": 58965, "loss": 3.5618125915527346, "grad_norm": 0.6895319819450378, "learning_rate": 0.0001, "elbo": 3.56181, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4579521463757916}
{"step": 58970, "loss": 3.552342987060547, "grad_norm": 0.7483119368553162, "learning_rate": 0.0001, "elbo": 3.55234, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.458245367112362}
{"step": 58975, "loss": 3.538921356201172, "grad_norm": 0.7399975061416626, "learning_rate": 0.0001, "elbo": 3.53892, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4585385878489325}
{"step": 58980, "loss": 3.532418060302734, "grad_norm": 0.692574679851532, "learning_rate": 0.0001, "elbo": 3.53242, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.458831808585503}
{"step": 58985, "loss": 3.5598777770996093, "grad_norm": 0.6903777122497559, "learning_rate": 0.0001, "elbo": 3.55988, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.459125029322074}
{"step": 58990, "loss": 3.5581920623779295, "grad_norm": 0.6997254490852356, "learning_rate": 0.0001, "elbo": 3.55819, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4594182500586443}
{"step": 58995, "loss": 3.545208740234375, "grad_norm": 0.6935576796531677, "learning_rate": 0.0001, "elbo": 3.54521, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.459711470795215}
{"step": 59000, "loss": 3.554251480102539, "grad_norm": 0.92003333568573, "learning_rate": 0.0001, "elbo": 3.55425, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4600046915317852}
{"step": 59005, "loss": 3.559983825683594, "grad_norm": 0.8434139490127563, "learning_rate": 0.0001, "elbo": 3.55998, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4602979122683557}
{"step": 59010, "loss": 3.5659248352050783, "grad_norm": 0.6903170347213745, "learning_rate": 0.0001, "elbo": 3.56592, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.460591133004926}
{"step": 59015, "loss": 3.5533885955810547, "grad_norm": 0.7302172183990479, "learning_rate": 0.0001, "elbo": 3.55339, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4608843537414966}
{"step": 59020, "loss": 3.547524261474609, "grad_norm": 0.6828420162200928, "learning_rate": 0.0001, "elbo": 3.54752, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.461177574478067}
{"step": 59025, "loss": 3.550737762451172, "grad_norm": 0.743961751461029, "learning_rate": 0.0001, "elbo": 3.55074, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4614707952146375}
{"step": 59030, "loss": 3.557700347900391, "grad_norm": 0.6760351657867432, "learning_rate": 0.0001, "elbo": 3.5577, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.461764015951208}
{"step": 59035, "loss": 3.5579246520996093, "grad_norm": 0.9124823212623596, "learning_rate": 0.0001, "elbo": 3.55792, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4620572366877784}
{"step": 59040, "loss": 3.5493228912353514, "grad_norm": 0.6892824769020081, "learning_rate": 0.0001, "elbo": 3.54932, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.462350457424349}
{"step": 59045, "loss": 3.5662792205810545, "grad_norm": 0.7408720850944519, "learning_rate": 0.0001, "elbo": 3.56628, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4626436781609193}
{"step": 59050, "loss": 3.554594802856445, "grad_norm": 0.778425931930542, "learning_rate": 0.0001, "elbo": 3.55459, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.46293689889749}
{"step": 59055, "loss": 3.5560813903808595, "grad_norm": 0.7330389618873596, "learning_rate": 0.0001, "elbo": 3.55608, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4632301196340607}
{"step": 59060, "loss": 3.551596832275391, "grad_norm": 0.6705703735351562, "learning_rate": 0.0001, "elbo": 3.5516, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.463523340370631}
{"step": 59065, "loss": 3.5522705078125, "grad_norm": 0.9360243678092957, "learning_rate": 0.0001, "elbo": 3.55227, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4638165611072016}
{"step": 59070, "loss": 3.547588348388672, "grad_norm": 0.7298485636711121, "learning_rate": 0.0001, "elbo": 3.54759, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.464109781843772}
{"step": 59075, "loss": 3.5323951721191404, "grad_norm": 0.7465964555740356, "learning_rate": 0.0001, "elbo": 3.5324, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4644030025803425}
{"step": 59080, "loss": 3.5317726135253906, "grad_norm": 0.7379348278045654, "learning_rate": 0.0001, "elbo": 3.53177, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.464696223316913}
{"step": 59085, "loss": 3.5444210052490233, "grad_norm": 0.9357768893241882, "learning_rate": 0.0001, "elbo": 3.54442, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4649894440534834}
{"step": 59090, "loss": 3.5628036499023437, "grad_norm": 0.6827823519706726, "learning_rate": 0.0001, "elbo": 3.5628, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.465282664790054}
{"step": 59095, "loss": 3.5603614807128907, "grad_norm": 0.7184917330741882, "learning_rate": 0.0001, "elbo": 3.56036, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4655758855266243}
{"step": 59100, "loss": 3.5374069213867188, "grad_norm": 0.6976986527442932, "learning_rate": 0.0001, "elbo": 3.53741, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4658691062631948}
{"step": 59105, "loss": 3.557350921630859, "grad_norm": 0.7746171951293945, "learning_rate": 0.0001, "elbo": 3.55735, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4661623269997652}
{"step": 59110, "loss": 3.554451751708984, "grad_norm": 0.6888314485549927, "learning_rate": 0.0001, "elbo": 3.55445, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.466455547736336}
{"step": 59115, "loss": 3.5605560302734376, "grad_norm": 0.7637216448783875, "learning_rate": 0.0001, "elbo": 3.56056, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4667487684729066}
{"step": 59120, "loss": 3.5430721282958983, "grad_norm": 0.7791652679443359, "learning_rate": 0.0001, "elbo": 3.54307, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.467041989209477}
{"step": 59125, "loss": 3.568045806884766, "grad_norm": 0.6643416285514832, "learning_rate": 0.0001, "elbo": 3.56805, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4673352099460475}
{"step": 59130, "loss": 3.5631412506103515, "grad_norm": 0.7093870639801025, "learning_rate": 0.0001, "elbo": 3.56314, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.467628430682618}
{"step": 59135, "loss": 3.5509933471679687, "grad_norm": 0.6803162693977356, "learning_rate": 0.0001, "elbo": 3.55099, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4679216514191884}
{"step": 59140, "loss": 3.5345306396484375, "grad_norm": 0.7357742786407471, "learning_rate": 0.0001, "elbo": 3.53453, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.468214872155759}
{"step": 59145, "loss": 3.561013031005859, "grad_norm": 0.7080689072608948, "learning_rate": 0.0001, "elbo": 3.56101, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4685080928923293}
{"step": 59150, "loss": 3.555702972412109, "grad_norm": 0.706911027431488, "learning_rate": 0.0001, "elbo": 3.5557, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4688013136288998}
{"step": 59155, "loss": 3.559429931640625, "grad_norm": 0.9341155886650085, "learning_rate": 0.0001, "elbo": 3.55943, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.46909453436547}
{"step": 59160, "loss": 3.5442882537841798, "grad_norm": 0.899720311164856, "learning_rate": 0.0001, "elbo": 3.54429, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4693877551020407}
{"step": 59165, "loss": 3.5527130126953126, "grad_norm": 0.6675933599472046, "learning_rate": 0.0001, "elbo": 3.55271, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.469680975838611}
{"step": 59170, "loss": 3.545852279663086, "grad_norm": 0.8455995321273804, "learning_rate": 0.0001, "elbo": 3.54585, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4699741965751816}
{"step": 59175, "loss": 3.549005889892578, "grad_norm": 0.7764289379119873, "learning_rate": 0.0001, "elbo": 3.54901, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4702674173117525}
{"step": 59180, "loss": 3.5631973266601564, "grad_norm": 0.6797381043434143, "learning_rate": 0.0001, "elbo": 3.5632, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.470560638048323}
{"step": 59185, "loss": 3.5618213653564452, "grad_norm": 0.6926940679550171, "learning_rate": 0.0001, "elbo": 3.56182, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4708538587848934}
{"step": 59190, "loss": 3.5565727233886717, "grad_norm": 0.666106641292572, "learning_rate": 0.0001, "elbo": 3.55657, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.471147079521464}
{"step": 59195, "loss": 3.537431335449219, "grad_norm": 0.667654812335968, "learning_rate": 0.0001, "elbo": 3.53743, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4714403002580343}
{"step": 59200, "loss": 3.5504459381103515, "grad_norm": 0.6879162192344666, "learning_rate": 0.0001, "elbo": 3.55045, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4717335209946047}
{"step": 59205, "loss": 3.5557296752929686, "grad_norm": 0.7854077816009521, "learning_rate": 0.0001, "elbo": 3.55573, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.472026741731175}
{"step": 59210, "loss": 3.5475677490234374, "grad_norm": 0.7280727624893188, "learning_rate": 0.0001, "elbo": 3.54757, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4723199624677457}
{"step": 59215, "loss": 3.5486404418945314, "grad_norm": 0.7218875288963318, "learning_rate": 0.0001, "elbo": 3.54864, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.472613183204316}
{"step": 59220, "loss": 3.558765411376953, "grad_norm": 0.6844956278800964, "learning_rate": 0.0001, "elbo": 3.55877, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4729064039408866}
{"step": 59225, "loss": 3.551352691650391, "grad_norm": 0.6877913475036621, "learning_rate": 0.0001, "elbo": 3.55135, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.473199624677457}
{"step": 59230, "loss": 3.5667491912841798, "grad_norm": 0.6631686091423035, "learning_rate": 0.0001, "elbo": 3.56675, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.473492845414028}
{"step": 59235, "loss": 3.5380401611328125, "grad_norm": 0.666758120059967, "learning_rate": 0.0001, "elbo": 3.53804, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4737860661505984}
{"step": 59240, "loss": 3.5636314392089843, "grad_norm": 0.6351545453071594, "learning_rate": 0.0001, "elbo": 3.56363, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.474079286887169}
{"step": 59245, "loss": 3.559168243408203, "grad_norm": 0.9260515570640564, "learning_rate": 0.0001, "elbo": 3.55917, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4743725076237393}
{"step": 59250, "loss": 3.5620628356933595, "grad_norm": 0.6975736021995544, "learning_rate": 0.0001, "elbo": 3.56206, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4746657283603097}
{"step": 59255, "loss": 3.5521121978759767, "grad_norm": 0.6942167282104492, "learning_rate": 0.0001, "elbo": 3.55211, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.47495894909688}
{"step": 59260, "loss": 3.542676544189453, "grad_norm": 0.6875826120376587, "learning_rate": 0.0001, "elbo": 3.54268, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4752521698334506}
{"step": 59265, "loss": 3.5362049102783204, "grad_norm": 0.6718564033508301, "learning_rate": 0.0001, "elbo": 3.5362, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.475545390570021}
{"step": 59270, "loss": 3.5500869750976562, "grad_norm": 0.6899814605712891, "learning_rate": 0.0001, "elbo": 3.55009, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4758386113065916}
{"step": 59275, "loss": 3.563036346435547, "grad_norm": 0.7362548112869263, "learning_rate": 0.0001, "elbo": 3.56304, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.476131832043162}
{"step": 59280, "loss": 3.550750732421875, "grad_norm": 0.69673091173172, "learning_rate": 0.0001, "elbo": 3.55075, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4764250527797325}
{"step": 59285, "loss": 3.53212890625, "grad_norm": 0.6868339776992798, "learning_rate": 0.0001, "elbo": 3.53213, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.476718273516303}
{"step": 59290, "loss": 3.5591197967529298, "grad_norm": 0.8081368207931519, "learning_rate": 0.0001, "elbo": 3.55912, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4770114942528734}
{"step": 59295, "loss": 3.5568077087402346, "grad_norm": 0.7055739164352417, "learning_rate": 0.0001, "elbo": 3.55681, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.477304714989444}
{"step": 59300, "loss": 3.5551101684570314, "grad_norm": 0.745719313621521, "learning_rate": 0.0001, "elbo": 3.55511, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4775979357260147}
{"step": 59305, "loss": 3.5623374938964845, "grad_norm": 0.760765016078949, "learning_rate": 0.0001, "elbo": 3.56234, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.477891156462585}
{"step": 59310, "loss": 3.53856201171875, "grad_norm": 0.6695340275764465, "learning_rate": 0.0001, "elbo": 3.53856, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4781843771991556}
{"step": 59310, "loss": 3.53856201171875, "grad_norm": 0.6695340275764465, "learning_rate": 0.0001, "elbo": 3.53856, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4781843771991556}
{"step": 59315, "loss": 3.553325653076172, "grad_norm": 0.7291684150695801, "learning_rate": 0.0001, "elbo": 3.55333, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.478477597935726}
{"step": 59320, "loss": 3.553117370605469, "grad_norm": 0.8291836380958557, "learning_rate": 0.0001, "elbo": 3.55312, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4787708186722965}
{"step": 59325, "loss": 3.541800308227539, "grad_norm": 0.7165263891220093, "learning_rate": 0.0001, "elbo": 3.5418, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.479064039408867}
{"step": 59330, "loss": 3.5655963897705076, "grad_norm": 0.6883885264396667, "learning_rate": 0.0001, "elbo": 3.5656, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4793572601454374}
{"step": 59335, "loss": 3.562833786010742, "grad_norm": 0.6741454005241394, "learning_rate": 0.0001, "elbo": 3.56283, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.479650480882008}
{"step": 59340, "loss": 3.5422271728515624, "grad_norm": 0.6609501242637634, "learning_rate": 0.0001, "elbo": 3.54223, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4799437016185784}
{"step": 59345, "loss": 3.5520263671875, "grad_norm": 0.6996846199035645, "learning_rate": 0.0001, "elbo": 3.55203, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.480236922355149}
{"step": 59350, "loss": 3.562407302856445, "grad_norm": 0.6637309789657593, "learning_rate": 0.0001, "elbo": 3.56241, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4805301430917193}
{"step": 59355, "loss": 3.544198226928711, "grad_norm": 0.7123394012451172, "learning_rate": 0.0001, "elbo": 3.5442, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.48082336382829}
{"step": 59360, "loss": 3.5386566162109374, "grad_norm": 0.7618007659912109, "learning_rate": 0.0001, "elbo": 3.53866, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4811165845648606}
{"step": 59365, "loss": 3.557582473754883, "grad_norm": 0.7171748876571655, "learning_rate": 0.0001, "elbo": 3.55758, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.481409805301431}
{"step": 59370, "loss": 3.5433490753173826, "grad_norm": 0.6830166578292847, "learning_rate": 0.0001, "elbo": 3.54335, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4817030260380015}
{"step": 59375, "loss": 3.5431358337402346, "grad_norm": 0.6909419298171997, "learning_rate": 0.0001, "elbo": 3.54314, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.481996246774572}
{"step": 59380, "loss": 3.5604434967041017, "grad_norm": 0.6961118578910828, "learning_rate": 0.0001, "elbo": 3.56044, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4822894675111424}
{"step": 59385, "loss": 3.553044891357422, "grad_norm": 0.7117277979850769, "learning_rate": 0.0001, "elbo": 3.55304, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.482582688247713}
{"step": 59385, "loss": 3.553044891357422, "grad_norm": 0.7117277979850769, "learning_rate": 0.0001, "elbo": 3.55304, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.482582688247713}
{"step": 59390, "loss": 3.5518192291259765, "grad_norm": 0.6848514080047607, "learning_rate": 0.0001, "elbo": 3.55182, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4828759089842833}
{"step": 59395, "loss": 3.539866256713867, "grad_norm": 0.6501471996307373, "learning_rate": 0.0001, "elbo": 3.53987, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.483169129720854}
{"step": 59395, "loss": 3.539866256713867, "grad_norm": 0.6501471996307373, "learning_rate": 0.0001, "elbo": 3.53987, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.483169129720854}
{"step": 59400, "loss": 3.5581783294677733, "grad_norm": 0.8997578620910645, "learning_rate": 0.0001, "elbo": 3.55818, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4834623504574243}
{"step": 59405, "loss": 3.560133361816406, "grad_norm": 0.6914204955101013, "learning_rate": 0.0001, "elbo": 3.56013, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4837555711939947}
{"step": 59405, "loss": 3.560133361816406, "grad_norm": 0.6914204955101013, "learning_rate": 0.0001, "elbo": 3.56013, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4837555711939947}
{"step": 59405, "loss": 3.560133361816406, "grad_norm": 0.6914204955101013, "learning_rate": 0.0001, "elbo": 3.56013, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4837555711939947}
{"step": 59410, "loss": 3.5483367919921873, "grad_norm": 0.8951625227928162, "learning_rate": 0.0001, "elbo": 3.54834, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.484048791930565}
{"step": 59415, "loss": 3.5587196350097656, "grad_norm": 0.7771540284156799, "learning_rate": 0.0001, "elbo": 3.55872, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4843420126671356}
{"step": 59420, "loss": 3.540793228149414, "grad_norm": 0.6866902112960815, "learning_rate": 0.0001, "elbo": 3.54079, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.484635233403706}
{"step": 59425, "loss": 3.535710906982422, "grad_norm": 0.6814448237419128, "learning_rate": 0.0001, "elbo": 3.53571, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.484928454140277}
{"step": 59425, "loss": 3.535710906982422, "grad_norm": 0.6814448237419128, "learning_rate": 0.0001, "elbo": 3.53571, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.484928454140277}
{"step": 59430, "loss": 3.535089111328125, "grad_norm": 0.7639457583427429, "learning_rate": 0.0001, "elbo": 3.53509, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4852216748768474}
{"step": 59435, "loss": 3.550212860107422, "grad_norm": 0.7548116445541382, "learning_rate": 0.0001, "elbo": 3.55021, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.485514895613418}
{"step": 59440, "loss": 3.5538658142089843, "grad_norm": 0.8801460862159729, "learning_rate": 0.0001, "elbo": 3.55387, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4858081163499883}
{"step": 59445, "loss": 3.578124237060547, "grad_norm": 0.7070714235305786, "learning_rate": 0.0001, "elbo": 3.57812, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.486101337086559}
{"step": 59450, "loss": 3.537508010864258, "grad_norm": 0.7247936129570007, "learning_rate": 0.0001, "elbo": 3.53751, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4863945578231292}
{"step": 59455, "loss": 3.548056411743164, "grad_norm": 0.6997556090354919, "learning_rate": 0.0001, "elbo": 3.54806, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4866877785596997}
{"step": 59460, "loss": 3.5409011840820312, "grad_norm": 0.965654730796814, "learning_rate": 0.0001, "elbo": 3.5409, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.48698099929627}
{"step": 59465, "loss": 3.5330963134765625, "grad_norm": 0.7630809545516968, "learning_rate": 0.0001, "elbo": 3.5331, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4872742200328406}
{"step": 59470, "loss": 3.568647766113281, "grad_norm": 0.6892653107643127, "learning_rate": 0.0001, "elbo": 3.56865, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.487567440769411}
{"step": 59475, "loss": 3.561750030517578, "grad_norm": 0.6763750910758972, "learning_rate": 0.0001, "elbo": 3.56175, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.487860661505982}
{"step": 59480, "loss": 3.549813079833984, "grad_norm": 0.7626108527183533, "learning_rate": 0.0001, "elbo": 3.54981, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4881538822425524}
{"step": 59485, "loss": 3.5425460815429686, "grad_norm": 0.6553239822387695, "learning_rate": 0.0001, "elbo": 3.54255, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.488447102979123}
{"step": 59490, "loss": 3.530400848388672, "grad_norm": 0.7522243857383728, "learning_rate": 0.0001, "elbo": 3.5304, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4887403237156933}
{"step": 59495, "loss": 3.556072235107422, "grad_norm": 0.6708889603614807, "learning_rate": 0.0001, "elbo": 3.55607, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4890335444522638}
{"step": 59500, "loss": 3.5464237213134764, "grad_norm": 0.8589130640029907, "learning_rate": 0.0001, "elbo": 3.54642, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4893267651888342}
{"step": 59505, "loss": 3.541748046875, "grad_norm": 0.7361703515052795, "learning_rate": 0.0001, "elbo": 3.54175, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4896199859254047}
{"step": 59510, "loss": 3.565143585205078, "grad_norm": 0.7459994554519653, "learning_rate": 0.0001, "elbo": 3.56514, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.489913206661975}
{"step": 59515, "loss": 3.568799591064453, "grad_norm": 0.713320255279541, "learning_rate": 0.0001, "elbo": 3.5688, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4902064273985456}
{"step": 59515, "loss": 3.568799591064453, "grad_norm": 0.713320255279541, "learning_rate": 0.0001, "elbo": 3.5688, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4902064273985456}
{"step": 59520, "loss": 3.565233612060547, "grad_norm": 0.6696553826332092, "learning_rate": 0.0001, "elbo": 3.56523, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.490499648135116}
{"step": 59525, "loss": 3.5450603485107424, "grad_norm": 0.7385558485984802, "learning_rate": 0.0001, "elbo": 3.54506, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4907928688716865}
{"step": 59530, "loss": 3.550578308105469, "grad_norm": 0.9055801033973694, "learning_rate": 0.0001, "elbo": 3.55058, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.491086089608257}
{"step": 59535, "loss": 3.5496482849121094, "grad_norm": 0.7370660901069641, "learning_rate": 0.0001, "elbo": 3.54965, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4913793103448274}
{"step": 59540, "loss": 3.5381710052490236, "grad_norm": 1.333118200302124, "learning_rate": 0.0001, "elbo": 3.53817, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.491672531081398}
{"step": 59545, "loss": 3.5547271728515626, "grad_norm": 0.6970662474632263, "learning_rate": 0.0001, "elbo": 3.55473, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4919657518179688}
{"step": 59550, "loss": 3.551996612548828, "grad_norm": 0.6457952857017517, "learning_rate": 0.0001, "elbo": 3.552, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.492258972554539}
{"step": 59555, "loss": 3.565069580078125, "grad_norm": 0.765369713306427, "learning_rate": 0.0001, "elbo": 3.56507, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4925521932911097}
{"step": 59560, "loss": 3.557900619506836, "grad_norm": 0.7774991393089294, "learning_rate": 0.0001, "elbo": 3.5579, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.49284541402768}
{"step": 59565, "loss": 3.554975891113281, "grad_norm": 0.7319149374961853, "learning_rate": 0.0001, "elbo": 3.55498, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4931386347642506}
{"step": 59570, "loss": 3.5360755920410156, "grad_norm": 0.6804933547973633, "learning_rate": 0.0001, "elbo": 3.53608, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.493431855500821}
{"step": 59575, "loss": 3.5585365295410156, "grad_norm": 0.7579757571220398, "learning_rate": 0.0001, "elbo": 3.55854, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4937250762373915}
{"step": 59580, "loss": 3.5670455932617187, "grad_norm": 0.7169012427330017, "learning_rate": 0.0001, "elbo": 3.56705, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.494018296973962}
{"step": 59585, "loss": 3.5705642700195312, "grad_norm": 0.7139979004859924, "learning_rate": 0.0001, "elbo": 3.57056, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4943115177105324}
{"step": 59590, "loss": 3.5543628692626954, "grad_norm": 0.6978907585144043, "learning_rate": 0.0001, "elbo": 3.55436, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.494604738447103}
{"step": 59595, "loss": 3.5468112945556642, "grad_norm": 0.7128797173500061, "learning_rate": 0.0001, "elbo": 3.54681, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4948979591836733}
{"step": 59600, "loss": 3.5292076110839843, "grad_norm": 0.7128285765647888, "learning_rate": 0.0001, "elbo": 3.52921, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.495191179920244}
{"step": 59605, "loss": 3.5508186340332033, "grad_norm": 0.6752123832702637, "learning_rate": 0.0001, "elbo": 3.55082, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4954844006568146}
{"step": 59610, "loss": 3.5540775299072265, "grad_norm": 0.6790813207626343, "learning_rate": 0.0001, "elbo": 3.55408, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.495777621393385}
{"step": 59615, "loss": 3.534537506103516, "grad_norm": 0.6893134713172913, "learning_rate": 0.0001, "elbo": 3.53454, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4960708421299556}
{"step": 59620, "loss": 3.5555755615234377, "grad_norm": 0.6582687497138977, "learning_rate": 0.0001, "elbo": 3.55558, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.496364062866526}
{"step": 59625, "loss": 3.574884033203125, "grad_norm": 0.6908361315727234, "learning_rate": 0.0001, "elbo": 3.57488, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4966572836030965}
{"step": 59630, "loss": 3.5429008483886717, "grad_norm": 0.6746684312820435, "learning_rate": 0.0001, "elbo": 3.5429, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.496950504339667}
{"step": 59630, "loss": 3.5429008483886717, "grad_norm": 0.6746684312820435, "learning_rate": 0.0001, "elbo": 3.5429, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.496950504339667}
{"step": 59635, "loss": 3.542554473876953, "grad_norm": 0.9537563920021057, "learning_rate": 0.0001, "elbo": 3.54255, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4972437250762374}
{"step": 59640, "loss": 3.547601318359375, "grad_norm": 0.7103888392448425, "learning_rate": 0.0001, "elbo": 3.5476, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.497536945812808}
{"step": 59645, "loss": 3.5644386291503904, "grad_norm": 0.6755993366241455, "learning_rate": 0.0001, "elbo": 3.56444, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4978301665493783}
{"step": 59650, "loss": 3.551583099365234, "grad_norm": 0.7176380753517151, "learning_rate": 0.0001, "elbo": 3.55158, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4981233872859487}
{"step": 59650, "loss": 3.551583099365234, "grad_norm": 0.7176380753517151, "learning_rate": 0.0001, "elbo": 3.55158, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4981233872859487}
{"step": 59655, "loss": 3.559295654296875, "grad_norm": 0.7170291543006897, "learning_rate": 0.0001, "elbo": 3.5593, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.498416608022519}
{"step": 59660, "loss": 3.561480712890625, "grad_norm": 0.799177348613739, "learning_rate": 0.0001, "elbo": 3.56148, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4987098287590896}
{"step": 59665, "loss": 3.565510559082031, "grad_norm": 1.0001897811889648, "learning_rate": 0.0001, "elbo": 3.56551, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.49900304949566}
{"step": 59670, "loss": 3.552103042602539, "grad_norm": 0.6968648433685303, "learning_rate": 0.0001, "elbo": 3.5521, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.499296270232231}
{"step": 59675, "loss": 3.549319839477539, "grad_norm": 0.683026909828186, "learning_rate": 0.0001, "elbo": 3.54932, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.4995894909688015}
{"step": 59680, "loss": 3.553852081298828, "grad_norm": 0.717694878578186, "learning_rate": 0.0001, "elbo": 3.55385, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.499882711705372}
{"step": 59685, "loss": 3.556784820556641, "grad_norm": 0.6559667587280273, "learning_rate": 0.0001, "elbo": 3.55679, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5001759324419424}
{"step": 59690, "loss": 3.5546985626220704, "grad_norm": 0.675006091594696, "learning_rate": 0.0001, "elbo": 3.5547, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.500469153178513}
{"step": 59695, "loss": 3.562081146240234, "grad_norm": 0.7061629295349121, "learning_rate": 0.0001, "elbo": 3.56208, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5007623739150833}
{"step": 59700, "loss": 3.5506961822509764, "grad_norm": 0.6469566226005554, "learning_rate": 0.0001, "elbo": 3.5507, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5010555946516537}
{"step": 59705, "loss": 3.546946716308594, "grad_norm": 1.1518195867538452, "learning_rate": 0.0001, "elbo": 3.54695, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.501348815388224}
{"step": 59710, "loss": 3.55775146484375, "grad_norm": 0.6731184124946594, "learning_rate": 0.0001, "elbo": 3.55775, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5016420361247946}
{"step": 59715, "loss": 3.558258056640625, "grad_norm": 0.7475491166114807, "learning_rate": 0.0001, "elbo": 3.55826, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.501935256861365}
{"step": 59720, "loss": 3.5418327331542967, "grad_norm": 0.7027626633644104, "learning_rate": 0.0001, "elbo": 3.54183, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.502228477597936}
{"step": 59725, "loss": 3.5616439819335937, "grad_norm": 0.7156632542610168, "learning_rate": 0.0001, "elbo": 3.56164, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5025216983345064}
{"step": 59730, "loss": 3.545388031005859, "grad_norm": 0.6790697574615479, "learning_rate": 0.0001, "elbo": 3.54539, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.502814919071077}
{"step": 59735, "loss": 3.5472667694091795, "grad_norm": 0.6613271236419678, "learning_rate": 0.0001, "elbo": 3.54727, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5031081398076473}
{"step": 59740, "loss": 3.564571762084961, "grad_norm": 0.6823980808258057, "learning_rate": 0.0001, "elbo": 3.56457, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.503401360544218}
{"step": 59745, "loss": 3.551594543457031, "grad_norm": 0.7284021377563477, "learning_rate": 0.0001, "elbo": 3.55159, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5036945812807883}
{"step": 59750, "loss": 3.5639389038085936, "grad_norm": 0.7154670357704163, "learning_rate": 0.0001, "elbo": 3.56394, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5039878020173587}
{"step": 59755, "loss": 3.554448699951172, "grad_norm": 0.7741260528564453, "learning_rate": 0.0001, "elbo": 3.55445, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.504281022753929}
{"step": 59755, "loss": 3.554448699951172, "grad_norm": 0.7741260528564453, "learning_rate": 0.0001, "elbo": 3.55445, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.504281022753929}
{"step": 59760, "loss": 3.5662303924560548, "grad_norm": 0.7137305736541748, "learning_rate": 0.0001, "elbo": 3.56623, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5045742434904996}
{"step": 59765, "loss": 3.5415443420410155, "grad_norm": 0.8478940725326538, "learning_rate": 0.0001, "elbo": 3.54154, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.50486746422707}
{"step": 59770, "loss": 3.554209899902344, "grad_norm": 0.6737676858901978, "learning_rate": 0.0001, "elbo": 3.55421, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5051606849636405}
{"step": 59775, "loss": 3.5561180114746094, "grad_norm": 0.7475729584693909, "learning_rate": 0.0001, "elbo": 3.55612, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.505453905700211}
{"step": 59780, "loss": 3.560056686401367, "grad_norm": 0.7178146243095398, "learning_rate": 0.0001, "elbo": 3.56006, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5057471264367814}
{"step": 59785, "loss": 3.559944915771484, "grad_norm": 0.7261857986450195, "learning_rate": 0.0001, "elbo": 3.55994, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.506040347173352}
{"step": 59785, "loss": 3.559944915771484, "grad_norm": 0.7261857986450195, "learning_rate": 0.0001, "elbo": 3.55994, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.506040347173352}
{"step": 59790, "loss": 3.5392967224121095, "grad_norm": 0.7716071605682373, "learning_rate": 0.0001, "elbo": 3.5393, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5063335679099223}
{"step": 59795, "loss": 3.5512866973876953, "grad_norm": 1.0188897848129272, "learning_rate": 0.0001, "elbo": 3.55129, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5066267886464932}
{"step": 59800, "loss": 3.5442291259765626, "grad_norm": 0.69343101978302, "learning_rate": 0.0001, "elbo": 3.54423, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5069200093830637}
{"step": 59805, "loss": 3.554058074951172, "grad_norm": 0.6920691132545471, "learning_rate": 0.0001, "elbo": 3.55406, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.507213230119634}
{"step": 59810, "loss": 3.552505874633789, "grad_norm": 0.683509111404419, "learning_rate": 0.0001, "elbo": 3.55251, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5075064508562046}
{"step": 59815, "loss": 3.540235900878906, "grad_norm": 0.7406607866287231, "learning_rate": 0.0001, "elbo": 3.54024, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.507799671592775}
{"step": 59820, "loss": 3.543585205078125, "grad_norm": 0.724035918712616, "learning_rate": 0.0001, "elbo": 3.54359, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5080928923293455}
{"step": 59825, "loss": 3.543809509277344, "grad_norm": 0.6904953122138977, "learning_rate": 0.0001, "elbo": 3.54381, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.508386113065916}
{"step": 59830, "loss": 3.552539825439453, "grad_norm": 0.7397781014442444, "learning_rate": 0.0001, "elbo": 3.55254, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5086793338024864}
{"step": 59835, "loss": 3.5551036834716796, "grad_norm": 0.6551571488380432, "learning_rate": 0.0001, "elbo": 3.5551, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.508972554539057}
{"step": 59835, "loss": 3.5551036834716796, "grad_norm": 0.6551571488380432, "learning_rate": 0.0001, "elbo": 3.5551, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.508972554539057}
{"step": 59840, "loss": 3.561676025390625, "grad_norm": 0.7591592073440552, "learning_rate": 0.0001, "elbo": 3.56168, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5092657752756273}
{"step": 59840, "loss": 3.561676025390625, "grad_norm": 0.7591592073440552, "learning_rate": 0.0001, "elbo": 3.56168, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5092657752756273}
{"step": 59845, "loss": 3.5569999694824217, "grad_norm": 0.7551183700561523, "learning_rate": 0.0001, "elbo": 3.557, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5095589960121982}
{"step": 59850, "loss": 3.5669189453125, "grad_norm": 0.7294395565986633, "learning_rate": 0.0001, "elbo": 3.56692, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5098522167487687}
{"step": 59855, "loss": 3.552813720703125, "grad_norm": 0.9127936959266663, "learning_rate": 0.0001, "elbo": 3.55281, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.510145437485339}
{"step": 59860, "loss": 3.572032165527344, "grad_norm": 0.6915512084960938, "learning_rate": 0.0001, "elbo": 3.57203, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5104386582219096}
{"step": 59865, "loss": 3.541859817504883, "grad_norm": 0.6942735314369202, "learning_rate": 0.0001, "elbo": 3.54186, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.51073187895848}
{"step": 59870, "loss": 3.5352462768554687, "grad_norm": 0.6859984993934631, "learning_rate": 0.0001, "elbo": 3.53525, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5110250996950505}
{"step": 59875, "loss": 3.5526451110839843, "grad_norm": 0.6959188580513, "learning_rate": 0.0001, "elbo": 3.55265, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.511318320431621}
{"step": 59880, "loss": 3.5753379821777345, "grad_norm": 0.7779905796051025, "learning_rate": 0.0001, "elbo": 3.57534, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5116115411681914}
{"step": 59885, "loss": 3.5442081451416017, "grad_norm": 0.7161455750465393, "learning_rate": 0.0001, "elbo": 3.54421, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.511904761904762}
{"step": 59890, "loss": 3.5552001953125, "grad_norm": 0.7169786095619202, "learning_rate": 0.0001, "elbo": 3.5552, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5121979826413323}
{"step": 59895, "loss": 3.552787017822266, "grad_norm": 0.705773115158081, "learning_rate": 0.0001, "elbo": 3.55279, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5124912033779028}
{"step": 59900, "loss": 3.562535858154297, "grad_norm": 0.876835286617279, "learning_rate": 0.0001, "elbo": 3.56254, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5127844241144732}
{"step": 59905, "loss": 3.5527523040771483, "grad_norm": 1.0641779899597168, "learning_rate": 0.0001, "elbo": 3.55275, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5130776448510437}
{"step": 59910, "loss": 3.5439163208007813, "grad_norm": 0.6969111561775208, "learning_rate": 0.0001, "elbo": 3.54392, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.513370865587614}
{"step": 59915, "loss": 3.5658294677734377, "grad_norm": 0.6468503475189209, "learning_rate": 0.0001, "elbo": 3.56583, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5136640863241846}
{"step": 59920, "loss": 3.5663944244384767, "grad_norm": 0.7329648733139038, "learning_rate": 0.0001, "elbo": 3.56639, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5139573070607555}
{"step": 59925, "loss": 3.5580482482910156, "grad_norm": 0.6929481029510498, "learning_rate": 0.0001, "elbo": 3.55805, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.514250527797326}
{"step": 59930, "loss": 3.555978012084961, "grad_norm": 0.7524442076683044, "learning_rate": 0.0001, "elbo": 3.55598, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5145437485338964}
{"step": 59935, "loss": 3.540028381347656, "grad_norm": 0.673686683177948, "learning_rate": 0.0001, "elbo": 3.54003, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.514836969270467}
{"step": 59940, "loss": 3.5698150634765624, "grad_norm": 0.8642544746398926, "learning_rate": 0.0001, "elbo": 3.56981, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5151301900070373}
{"step": 59945, "loss": 3.535685729980469, "grad_norm": 0.774239718914032, "learning_rate": 0.0001, "elbo": 3.53569, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5154234107436078}
{"step": 59950, "loss": 3.5520782470703125, "grad_norm": 0.7043202519416809, "learning_rate": 0.0001, "elbo": 3.55208, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.515716631480178}
{"step": 59950, "loss": 3.5520782470703125, "grad_norm": 0.7043202519416809, "learning_rate": 0.0001, "elbo": 3.55208, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.515716631480178}
{"step": 59955, "loss": 3.547205352783203, "grad_norm": 0.7074666023254395, "learning_rate": 0.0001, "elbo": 3.54721, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5160098522167487}
{"step": 59960, "loss": 3.5302993774414064, "grad_norm": 0.911186158657074, "learning_rate": 0.0001, "elbo": 3.5303, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.516303072953319}
{"step": 59965, "loss": 3.5580474853515627, "grad_norm": 0.8414456248283386, "learning_rate": 0.0001, "elbo": 3.55805, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.51659629368989}
{"step": 59970, "loss": 3.551827239990234, "grad_norm": 0.6772462725639343, "learning_rate": 0.0001, "elbo": 3.55183, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5168895144264605}
{"step": 59975, "loss": 3.5606475830078126, "grad_norm": 0.7346320152282715, "learning_rate": 0.0001, "elbo": 3.56065, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.517182735163031}
{"step": 59980, "loss": 3.544008255004883, "grad_norm": 0.6885896921157837, "learning_rate": 0.0001, "elbo": 3.54401, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5174759558996014}
{"step": 59985, "loss": 3.540972137451172, "grad_norm": 0.6654924750328064, "learning_rate": 0.0001, "elbo": 3.54097, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.517769176636172}
{"step": 59990, "loss": 3.5367080688476564, "grad_norm": 0.7179813385009766, "learning_rate": 0.0001, "elbo": 3.53671, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5180623973727423}
{"step": 59995, "loss": 3.5480995178222656, "grad_norm": 0.6801615953445435, "learning_rate": 0.0001, "elbo": 3.5481, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5183556181093127}
{"step": 60000, "loss": 3.5685401916503907, "grad_norm": 0.7939176559448242, "learning_rate": 0.0001, "elbo": 3.56854, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.518648838845883}
{"step": 60005, "loss": 3.558563995361328, "grad_norm": 0.7171475887298584, "learning_rate": 0.0001, "elbo": 3.55856, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5189420595824537}
{"step": 60010, "loss": 3.539434051513672, "grad_norm": 0.7168681621551514, "learning_rate": 0.0001, "elbo": 3.53943, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.519235280319024}
{"step": 60015, "loss": 3.5615280151367186, "grad_norm": 0.6771365404129028, "learning_rate": 0.0001, "elbo": 3.56153, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5195285010555946}
{"step": 60020, "loss": 3.554932403564453, "grad_norm": 0.7307010293006897, "learning_rate": 0.0001, "elbo": 3.55493, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.519821721792165}
{"step": 60025, "loss": 3.5552276611328124, "grad_norm": 0.6650322675704956, "learning_rate": 0.0001, "elbo": 3.55523, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5201149425287355}
{"step": 60030, "loss": 3.5426345825195313, "grad_norm": 0.7116929292678833, "learning_rate": 0.0001, "elbo": 3.54263, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.520408163265306}
{"step": 60035, "loss": 3.5365310668945313, "grad_norm": 0.6781905889511108, "learning_rate": 0.0001, "elbo": 3.53653, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5207013840018764}
{"step": 60040, "loss": 3.5510944366455077, "grad_norm": 0.668225884437561, "learning_rate": 0.0001, "elbo": 3.55109, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.520994604738447}
{"step": 60045, "loss": 3.5333892822265627, "grad_norm": 0.8332979679107666, "learning_rate": 0.0001, "elbo": 3.53339, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5212878254750177}
{"step": 60050, "loss": 3.543570709228516, "grad_norm": 0.6813583374023438, "learning_rate": 0.0001, "elbo": 3.54357, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.521581046211588}
{"step": 60055, "loss": 3.546042251586914, "grad_norm": 0.7097212076187134, "learning_rate": 0.0001, "elbo": 3.54604, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5218742669481586}
{"step": 60060, "loss": 3.5476966857910157, "grad_norm": 0.6723546385765076, "learning_rate": 0.0001, "elbo": 3.5477, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.522167487684729}
{"step": 60065, "loss": 3.551465606689453, "grad_norm": 0.7043619155883789, "learning_rate": 0.0001, "elbo": 3.55147, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5224607084212995}
{"step": 60070, "loss": 3.5588050842285157, "grad_norm": 1.3160624504089355, "learning_rate": 0.0001, "elbo": 3.55881, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.52275392915787}
{"step": 60075, "loss": 3.5581409454345705, "grad_norm": 0.6950883865356445, "learning_rate": 0.0001, "elbo": 3.55814, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5230471498944405}
{"step": 60080, "loss": 3.5393470764160155, "grad_norm": 0.6723226308822632, "learning_rate": 0.0001, "elbo": 3.53935, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.523340370631011}
{"step": 60085, "loss": 3.5500579833984376, "grad_norm": 0.6693165302276611, "learning_rate": 0.0001, "elbo": 3.55006, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5236335913675814}
{"step": 60090, "loss": 3.5548377990722657, "grad_norm": 0.6952590942382812, "learning_rate": 0.0001, "elbo": 3.55484, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5239268121041523}
{"step": 60095, "loss": 3.55220947265625, "grad_norm": 0.7205098867416382, "learning_rate": 0.0001, "elbo": 3.55221, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5242200328407227}
{"step": 60100, "loss": 3.5480731964111327, "grad_norm": 0.675145149230957, "learning_rate": 0.0001, "elbo": 3.54807, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.524513253577293}
{"step": 60105, "loss": 3.5574169158935547, "grad_norm": 0.6895129680633545, "learning_rate": 0.0001, "elbo": 3.55742, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5248064743138636}
{"step": 60110, "loss": 3.5428558349609376, "grad_norm": 0.8472908735275269, "learning_rate": 0.0001, "elbo": 3.54286, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.525099695050434}
{"step": 60115, "loss": 3.5597244262695313, "grad_norm": 0.7013183832168579, "learning_rate": 0.0001, "elbo": 3.55972, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5253929157870045}
{"step": 60120, "loss": 3.5547954559326174, "grad_norm": 0.6969496011734009, "learning_rate": 0.0001, "elbo": 3.5548, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.525686136523575}
{"step": 60125, "loss": 3.550730895996094, "grad_norm": 0.7306959629058838, "learning_rate": 0.0001, "elbo": 3.55073, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5259793572601454}
{"step": 60130, "loss": 3.5569095611572266, "grad_norm": 0.6730806231498718, "learning_rate": 0.0001, "elbo": 3.55691, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.526272577996716}
{"step": 60135, "loss": 3.5412940979003906, "grad_norm": 0.6860778331756592, "learning_rate": 0.0001, "elbo": 3.54129, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5265657987332863}
{"step": 60140, "loss": 3.5609310150146483, "grad_norm": 0.6926429867744446, "learning_rate": 0.0001, "elbo": 3.56093, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.526859019469857}
{"step": 60145, "loss": 3.553697967529297, "grad_norm": 0.7110360264778137, "learning_rate": 0.0001, "elbo": 3.5537, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5271522402064273}
{"step": 60150, "loss": 3.5571266174316407, "grad_norm": 0.705615222454071, "learning_rate": 0.0001, "elbo": 3.55713, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5274454609429977}
{"step": 60155, "loss": 3.5561294555664062, "grad_norm": 0.7784249782562256, "learning_rate": 0.0001, "elbo": 3.55613, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.527738681679568}
{"step": 60160, "loss": 3.554047393798828, "grad_norm": 0.6992836594581604, "learning_rate": 0.0001, "elbo": 3.55405, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5280319024161386}
{"step": 60165, "loss": 3.544988250732422, "grad_norm": 0.8720636367797852, "learning_rate": 0.0001, "elbo": 3.54499, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5283251231527095}
{"step": 60170, "loss": 3.5508193969726562, "grad_norm": 0.7309010028839111, "learning_rate": 0.0001, "elbo": 3.55082, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.52861834388928}
{"step": 60175, "loss": 3.538984680175781, "grad_norm": 0.6624676585197449, "learning_rate": 0.0001, "elbo": 3.53898, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5289115646258504}
{"step": 60180, "loss": 3.545737075805664, "grad_norm": 0.6947776079177856, "learning_rate": 0.0001, "elbo": 3.54574, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.529204785362421}
{"step": 60185, "loss": 3.5475120544433594, "grad_norm": 0.691063642501831, "learning_rate": 0.0001, "elbo": 3.54751, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5294980060989913}
{"step": 60190, "loss": 3.544569396972656, "grad_norm": 0.6361842155456543, "learning_rate": 0.0001, "elbo": 3.54457, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.529791226835562}
{"step": 60195, "loss": 3.5681968688964845, "grad_norm": 0.6856551766395569, "learning_rate": 0.0001, "elbo": 3.5682, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5300844475721322}
{"step": 60200, "loss": 3.5684776306152344, "grad_norm": 0.6825821995735168, "learning_rate": 0.0001, "elbo": 3.56848, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5303776683087027}
{"step": 60205, "loss": 3.540740966796875, "grad_norm": 0.7010215520858765, "learning_rate": 0.0001, "elbo": 3.54074, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.530670889045273}
{"step": 60210, "loss": 3.530176544189453, "grad_norm": 0.6626652479171753, "learning_rate": 0.0001, "elbo": 3.53018, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5309641097818436}
{"step": 60215, "loss": 3.553188705444336, "grad_norm": 0.6551195979118347, "learning_rate": 0.0001, "elbo": 3.55319, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5312573305184145}
{"step": 60220, "loss": 3.5549541473388673, "grad_norm": 0.730000376701355, "learning_rate": 0.0001, "elbo": 3.55495, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.531550551254985}
{"step": 60225, "loss": 3.544171905517578, "grad_norm": 0.689717173576355, "learning_rate": 0.0001, "elbo": 3.54417, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5318437719915554}
{"step": 60230, "loss": 3.548539733886719, "grad_norm": 0.7347942590713501, "learning_rate": 0.0001, "elbo": 3.54854, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.532136992728126}
{"step": 60235, "loss": 3.5382129669189455, "grad_norm": 0.6920469999313354, "learning_rate": 0.0001, "elbo": 3.53821, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5324302134646963}
{"step": 60240, "loss": 3.561446762084961, "grad_norm": 0.7846113443374634, "learning_rate": 0.0001, "elbo": 3.56145, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5327234342012668}
{"step": 60245, "loss": 3.542348861694336, "grad_norm": 0.698034942150116, "learning_rate": 0.0001, "elbo": 3.54235, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5330166549378372}
{"step": 60245, "loss": 3.542348861694336, "grad_norm": 0.698034942150116, "learning_rate": 0.0001, "elbo": 3.54235, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5330166549378372}
{"step": 60250, "loss": 3.5596664428710936, "grad_norm": 0.6919831037521362, "learning_rate": 0.0001, "elbo": 3.55967, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5333098756744077}
{"step": 60255, "loss": 3.544412612915039, "grad_norm": 0.729148268699646, "learning_rate": 0.0001, "elbo": 3.54441, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.533603096410978}
{"step": 60260, "loss": 3.5582805633544923, "grad_norm": 0.6965429186820984, "learning_rate": 0.0001, "elbo": 3.55828, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5338963171475486}
{"step": 60265, "loss": 3.5468849182128905, "grad_norm": 0.8564801216125488, "learning_rate": 0.0001, "elbo": 3.54689, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.534189537884119}
{"step": 60270, "loss": 3.542041015625, "grad_norm": 0.7032811641693115, "learning_rate": 0.0001, "elbo": 3.54204, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5344827586206895}
{"step": 60275, "loss": 3.5471260070800783, "grad_norm": 0.7076833844184875, "learning_rate": 0.0001, "elbo": 3.54713, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.53477597935726}
{"step": 60280, "loss": 3.5505115509033205, "grad_norm": 0.736838698387146, "learning_rate": 0.0001, "elbo": 3.55051, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5350692000938304}
{"step": 60285, "loss": 3.520768737792969, "grad_norm": 0.727504551410675, "learning_rate": 0.0001, "elbo": 3.52077, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.535362420830401}
{"step": 60290, "loss": 3.540454864501953, "grad_norm": 0.997589111328125, "learning_rate": 0.0001, "elbo": 3.54046, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5356556415669718}
{"step": 60295, "loss": 3.5291072845458986, "grad_norm": 0.6939070820808411, "learning_rate": 0.0001, "elbo": 3.52911, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.535948862303542}
{"step": 60300, "loss": 3.5468963623046874, "grad_norm": 0.7279933094978333, "learning_rate": 0.0001, "elbo": 3.5469, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5362420830401127}
{"step": 60305, "loss": 3.54661865234375, "grad_norm": 0.745332658290863, "learning_rate": 0.0001, "elbo": 3.54662, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.536535303776683}
{"step": 60310, "loss": 3.556610107421875, "grad_norm": 0.6635200381278992, "learning_rate": 0.0001, "elbo": 3.55661, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5368285245132536}
{"step": 60315, "loss": 3.5368003845214844, "grad_norm": 0.6768028736114502, "learning_rate": 0.0001, "elbo": 3.5368, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.537121745249824}
{"step": 60320, "loss": 3.5535694122314454, "grad_norm": 0.6858395338058472, "learning_rate": 0.0001, "elbo": 3.55357, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5374149659863945}
{"step": 60325, "loss": 3.546819305419922, "grad_norm": 0.7000064253807068, "learning_rate": 0.0001, "elbo": 3.54682, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.537708186722965}
{"step": 60330, "loss": 3.5535552978515623, "grad_norm": 0.6946702003479004, "learning_rate": 0.0001, "elbo": 3.55356, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5380014074595354}
{"step": 60335, "loss": 3.544824981689453, "grad_norm": 0.7013086080551147, "learning_rate": 0.0001, "elbo": 3.54482, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5382946281961063}
{"step": 60340, "loss": 3.5629150390625, "grad_norm": 0.6998552083969116, "learning_rate": 0.0001, "elbo": 3.56292, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5385878489326767}
{"step": 60345, "loss": 3.536856842041016, "grad_norm": 0.7394546270370483, "learning_rate": 0.0001, "elbo": 3.53686, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.538881069669247}
{"step": 60345, "loss": 3.536856842041016, "grad_norm": 0.7394546270370483, "learning_rate": 0.0001, "elbo": 3.53686, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.538881069669247}
{"step": 60350, "loss": 3.555153656005859, "grad_norm": 0.7421427369117737, "learning_rate": 0.0001, "elbo": 3.55515, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5391742904058177}
{"step": 60355, "loss": 3.5415115356445312, "grad_norm": 0.7778930068016052, "learning_rate": 0.0001, "elbo": 3.54151, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.539467511142388}
{"step": 60360, "loss": 3.5360145568847656, "grad_norm": 0.6959293484687805, "learning_rate": 0.0001, "elbo": 3.53601, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5397607318789586}
{"step": 60365, "loss": 3.558472442626953, "grad_norm": 0.7565951943397522, "learning_rate": 0.0001, "elbo": 3.55847, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.540053952615529}
{"step": 60370, "loss": 3.5513965606689455, "grad_norm": 0.6843603849411011, "learning_rate": 0.0001, "elbo": 3.5514, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5403471733520995}
{"step": 60370, "loss": 3.5513965606689455, "grad_norm": 0.6843603849411011, "learning_rate": 0.0001, "elbo": 3.5514, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5403471733520995}
{"step": 60375, "loss": 3.543135070800781, "grad_norm": 0.6839815378189087, "learning_rate": 0.0001, "elbo": 3.54313, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.54064039408867}
{"step": 60375, "loss": 3.543135070800781, "grad_norm": 0.6839815378189087, "learning_rate": 0.0001, "elbo": 3.54313, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.54064039408867}
{"step": 60380, "loss": 3.537643051147461, "grad_norm": 0.6873817443847656, "learning_rate": 0.0001, "elbo": 3.53764, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5409336148252404}
{"step": 60385, "loss": 3.560223388671875, "grad_norm": 0.702225923538208, "learning_rate": 0.0001, "elbo": 3.56022, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.541226835561811}
{"step": 60390, "loss": 3.5457717895507814, "grad_norm": 0.7101702094078064, "learning_rate": 0.0001, "elbo": 3.54577, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5415200562983813}
{"step": 60395, "loss": 3.558446502685547, "grad_norm": 0.7180946469306946, "learning_rate": 0.0001, "elbo": 3.55845, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5418132770349517}
{"step": 60400, "loss": 3.5460479736328123, "grad_norm": 0.7412029504776001, "learning_rate": 0.0001, "elbo": 3.54605, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.542106497771522}
{"step": 60405, "loss": 3.543051910400391, "grad_norm": 0.6952941417694092, "learning_rate": 0.0001, "elbo": 3.54305, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5423997185080927}
{"step": 60410, "loss": 3.5539474487304688, "grad_norm": 0.7294812798500061, "learning_rate": 0.0001, "elbo": 3.55395, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.542692939244663}
{"step": 60415, "loss": 3.5380317687988283, "grad_norm": 0.6698266267776489, "learning_rate": 0.0001, "elbo": 3.53803, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.542986159981234}
{"step": 60420, "loss": 3.561579132080078, "grad_norm": 0.7873949408531189, "learning_rate": 0.0001, "elbo": 3.56158, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5432793807178045}
{"step": 60420, "loss": 3.561579132080078, "grad_norm": 0.7873949408531189, "learning_rate": 0.0001, "elbo": 3.56158, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5432793807178045}
{"step": 60425, "loss": 3.5492897033691406, "grad_norm": 0.7227686643600464, "learning_rate": 0.0001, "elbo": 3.54929, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.543572601454375}
{"step": 60430, "loss": 3.5538143157958983, "grad_norm": 0.7117652297019958, "learning_rate": 0.0001, "elbo": 3.55381, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5438658221909454}
{"step": 60435, "loss": 3.5481285095214843, "grad_norm": 0.7418449521064758, "learning_rate": 0.0001, "elbo": 3.54813, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.544159042927516}
{"step": 60440, "loss": 3.5459800720214845, "grad_norm": 0.7302481532096863, "learning_rate": 0.0001, "elbo": 3.54598, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5444522636640863}
{"step": 60445, "loss": 3.5454856872558596, "grad_norm": 0.7382329702377319, "learning_rate": 0.0001, "elbo": 3.54549, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5447454844006567}
{"step": 60450, "loss": 3.5702590942382812, "grad_norm": 0.6864395141601562, "learning_rate": 0.0001, "elbo": 3.57026, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.545038705137227}
{"step": 60455, "loss": 3.55579833984375, "grad_norm": 0.6408926844596863, "learning_rate": 0.0001, "elbo": 3.5558, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5453319258737976}
{"step": 60460, "loss": 3.538863754272461, "grad_norm": 0.7146194577217102, "learning_rate": 0.0001, "elbo": 3.53886, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5456251466103685}
{"step": 60465, "loss": 3.553000640869141, "grad_norm": 0.7195953726768494, "learning_rate": 0.0001, "elbo": 3.553, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.545918367346939}
{"step": 60470, "loss": 3.5529632568359375, "grad_norm": 0.7103243470191956, "learning_rate": 0.0001, "elbo": 3.55296, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5462115880835094}
{"step": 60475, "loss": 3.5699478149414063, "grad_norm": 0.7417773008346558, "learning_rate": 0.0001, "elbo": 3.56995, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.54650480882008}
{"step": 60480, "loss": 3.552734375, "grad_norm": 0.6833259463310242, "learning_rate": 0.0001, "elbo": 3.55273, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5467980295566504}
{"step": 60485, "loss": 3.545429992675781, "grad_norm": 0.7004729509353638, "learning_rate": 0.0001, "elbo": 3.54543, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.547091250293221}
{"step": 60490, "loss": 3.5535301208496093, "grad_norm": 0.7028326392173767, "learning_rate": 0.0001, "elbo": 3.55353, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5473844710297913}
{"step": 60495, "loss": 3.5504207611083984, "grad_norm": 0.7776774764060974, "learning_rate": 0.0001, "elbo": 3.55042, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5476776917663617}
{"step": 60500, "loss": 3.5417190551757813, "grad_norm": 0.6468698978424072, "learning_rate": 0.0001, "elbo": 3.54172, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.547970912502932}
{"step": 60505, "loss": 3.546892547607422, "grad_norm": 0.8161579370498657, "learning_rate": 0.0001, "elbo": 3.54689, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5482641332395026}
{"step": 60510, "loss": 3.552538299560547, "grad_norm": 0.7025499939918518, "learning_rate": 0.0001, "elbo": 3.55254, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.548557353976073}
{"step": 60515, "loss": 3.544460678100586, "grad_norm": 0.7139423489570618, "learning_rate": 0.0001, "elbo": 3.54446, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5488505747126435}
{"step": 60520, "loss": 3.5548534393310547, "grad_norm": 0.7045080065727234, "learning_rate": 0.0001, "elbo": 3.55485, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.549143795449214}
{"step": 60525, "loss": 3.5546119689941404, "grad_norm": 0.6895504593849182, "learning_rate": 0.0001, "elbo": 3.55461, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5494370161857844}
{"step": 60530, "loss": 3.5438743591308595, "grad_norm": 0.678723931312561, "learning_rate": 0.0001, "elbo": 3.54387, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.549730236922355}
{"step": 60535, "loss": 3.550368881225586, "grad_norm": 0.7157477140426636, "learning_rate": 0.0001, "elbo": 3.55037, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.550023457658926}
{"step": 60540, "loss": 3.551347351074219, "grad_norm": 0.7021814584732056, "learning_rate": 0.0001, "elbo": 3.55135, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5503166783954963}
{"step": 60545, "loss": 3.5465618133544923, "grad_norm": 0.7063227295875549, "learning_rate": 0.0001, "elbo": 3.54656, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5506098991320667}
{"step": 60550, "loss": 3.539753723144531, "grad_norm": 0.744930624961853, "learning_rate": 0.0001, "elbo": 3.53975, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.550903119868637}
{"step": 60555, "loss": 3.548096466064453, "grad_norm": 0.683772623538971, "learning_rate": 0.0001, "elbo": 3.5481, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5511963406052076}
{"step": 60560, "loss": 3.557665252685547, "grad_norm": 0.6690140962600708, "learning_rate": 0.0001, "elbo": 3.55767, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.551489561341778}
{"step": 60565, "loss": 3.5539134979248046, "grad_norm": 1.1491986513137817, "learning_rate": 0.0001, "elbo": 3.55391, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5517827820783485}
{"step": 60570, "loss": 3.545794677734375, "grad_norm": 0.7672260403633118, "learning_rate": 0.0001, "elbo": 3.54579, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.552076002814919}
{"step": 60575, "loss": 3.5567043304443358, "grad_norm": 0.7159115672111511, "learning_rate": 0.0001, "elbo": 3.5567, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5523692235514894}
{"step": 60580, "loss": 3.5458377838134765, "grad_norm": 0.6844088435173035, "learning_rate": 0.0001, "elbo": 3.54584, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5526624442880603}
{"step": 60585, "loss": 3.5562767028808593, "grad_norm": 0.7434127330780029, "learning_rate": 0.0001, "elbo": 3.55628, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.552955665024631}
{"step": 60590, "loss": 3.553362274169922, "grad_norm": 0.6952248811721802, "learning_rate": 0.0001, "elbo": 3.55336, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5532488857612012}
{"step": 60595, "loss": 3.5360744476318358, "grad_norm": 0.6544626951217651, "learning_rate": 0.0001, "elbo": 3.53607, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5535421064977717}
{"step": 60600, "loss": 3.5493545532226562, "grad_norm": 0.7028843760490417, "learning_rate": 0.0001, "elbo": 3.54935, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.553835327234342}
{"step": 60605, "loss": 3.55322265625, "grad_norm": 0.6969972252845764, "learning_rate": 0.0001, "elbo": 3.55322, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5541285479709126}
{"step": 60610, "loss": 3.5531883239746094, "grad_norm": 0.7027406096458435, "learning_rate": 0.0001, "elbo": 3.55319, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.554421768707483}
{"step": 60615, "loss": 3.5562267303466797, "grad_norm": 0.7438938617706299, "learning_rate": 0.0001, "elbo": 3.55623, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5547149894440535}
{"step": 60620, "loss": 3.5585826873779296, "grad_norm": 0.7082472443580627, "learning_rate": 0.0001, "elbo": 3.55858, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.555008210180624}
{"step": 60625, "loss": 3.529772186279297, "grad_norm": 0.9189040660858154, "learning_rate": 0.0001, "elbo": 3.52977, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5553014309171944}
{"step": 60630, "loss": 3.557967758178711, "grad_norm": 0.7460905313491821, "learning_rate": 0.0001, "elbo": 3.55797, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.555594651653765}
{"step": 60635, "loss": 3.5486526489257812, "grad_norm": 0.7878003716468811, "learning_rate": 0.0001, "elbo": 3.54865, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5558878723903353}
{"step": 60640, "loss": 3.534881591796875, "grad_norm": 0.6890853047370911, "learning_rate": 0.0001, "elbo": 3.53488, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.556181093126906}
{"step": 60645, "loss": 3.5546405792236326, "grad_norm": 0.7096038460731506, "learning_rate": 0.0001, "elbo": 3.55464, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5564743138634762}
{"step": 60650, "loss": 3.560108947753906, "grad_norm": 0.7134593725204468, "learning_rate": 0.0001, "elbo": 3.56011, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5567675346000467}
{"step": 60655, "loss": 3.5433860778808595, "grad_norm": 1.007428526878357, "learning_rate": 0.0001, "elbo": 3.54339, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.557060755336617}
{"step": 60660, "loss": 3.536376190185547, "grad_norm": 0.7186315655708313, "learning_rate": 0.0001, "elbo": 3.53638, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.557353976073188}
{"step": 60665, "loss": 3.5388431549072266, "grad_norm": 0.77530437707901, "learning_rate": 0.0001, "elbo": 3.53884, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5576471968097585}
{"step": 60670, "loss": 3.557851028442383, "grad_norm": 0.7048792839050293, "learning_rate": 0.0001, "elbo": 3.55785, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.557940417546329}
{"step": 60675, "loss": 3.551422882080078, "grad_norm": 0.8448194265365601, "learning_rate": 0.0001, "elbo": 3.55142, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5582336382828994}
{"step": 60680, "loss": 3.5427013397216798, "grad_norm": 0.7025855183601379, "learning_rate": 0.0001, "elbo": 3.5427, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.55852685901947}
{"step": 60685, "loss": 3.54365234375, "grad_norm": 0.8872684240341187, "learning_rate": 0.0001, "elbo": 3.54365, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5588200797560403}
{"step": 60690, "loss": 3.5622329711914062, "grad_norm": 0.679585874080658, "learning_rate": 0.0001, "elbo": 3.56223, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5591133004926108}
{"step": 60695, "loss": 3.5557228088378907, "grad_norm": 0.7183945775032043, "learning_rate": 0.0001, "elbo": 3.55572, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.559406521229181}
{"step": 60700, "loss": 3.547138214111328, "grad_norm": 0.8181209564208984, "learning_rate": 0.0001, "elbo": 3.54714, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5596997419657517}
{"step": 60705, "loss": 3.541477584838867, "grad_norm": 0.7656498551368713, "learning_rate": 0.0001, "elbo": 3.54148, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5599929627023226}
{"step": 60710, "loss": 3.5602088928222657, "grad_norm": 0.7728385329246521, "learning_rate": 0.0001, "elbo": 3.56021, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.560286183438893}
{"step": 60715, "loss": 3.5512008666992188, "grad_norm": 0.7672723531723022, "learning_rate": 0.0001, "elbo": 3.5512, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5605794041754635}
{"step": 60720, "loss": 3.5484596252441407, "grad_norm": 0.7261948585510254, "learning_rate": 0.0001, "elbo": 3.54846, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.560872624912034}
{"step": 60725, "loss": 3.559027099609375, "grad_norm": 0.704383909702301, "learning_rate": 0.0001, "elbo": 3.55903, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5611658456486044}
{"step": 60730, "loss": 3.5596603393554687, "grad_norm": 0.6519037485122681, "learning_rate": 0.0001, "elbo": 3.55966, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.561459066385175}
{"step": 60735, "loss": 3.543292236328125, "grad_norm": 0.6620931029319763, "learning_rate": 0.0001, "elbo": 3.54329, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5617522871217453}
{"step": 60740, "loss": 3.5434696197509767, "grad_norm": 0.6861976385116577, "learning_rate": 0.0001, "elbo": 3.54347, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5620455078583158}
{"step": 60745, "loss": 3.5634178161621093, "grad_norm": 0.792665421962738, "learning_rate": 0.0001, "elbo": 3.56342, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.562338728594886}
{"step": 60745, "loss": 3.5634178161621093, "grad_norm": 0.792665421962738, "learning_rate": 0.0001, "elbo": 3.56342, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.562338728594886}
{"step": 60745, "loss": 3.5634178161621093, "grad_norm": 0.792665421962738, "learning_rate": 0.0001, "elbo": 3.56342, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.562338728594886}
{"step": 60750, "loss": 3.545674514770508, "grad_norm": 0.6552209258079529, "learning_rate": 0.0001, "elbo": 3.54567, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5626319493314567}
{"step": 60755, "loss": 3.5387245178222657, "grad_norm": 1.0082993507385254, "learning_rate": 0.0001, "elbo": 3.53872, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.562925170068027}
{"step": 60760, "loss": 3.538047027587891, "grad_norm": 0.7186508774757385, "learning_rate": 0.0001, "elbo": 3.53805, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5632183908045976}
{"step": 60765, "loss": 3.551474380493164, "grad_norm": 0.7468442320823669, "learning_rate": 0.0001, "elbo": 3.55147, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.563511611541168}
{"step": 60770, "loss": 3.534711456298828, "grad_norm": 0.7849432229995728, "learning_rate": 0.0001, "elbo": 3.53471, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5638048322777385}
{"step": 60775, "loss": 3.5556068420410156, "grad_norm": 0.6923912763595581, "learning_rate": 0.0001, "elbo": 3.55561, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.564098053014309}
{"step": 60780, "loss": 3.5519943237304688, "grad_norm": 0.6685783863067627, "learning_rate": 0.0001, "elbo": 3.55199, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5643912737508794}
{"step": 60780, "loss": 3.5519943237304688, "grad_norm": 0.6685783863067627, "learning_rate": 0.0001, "elbo": 3.55199, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5643912737508794}
{"step": 60785, "loss": 3.5376220703125, "grad_norm": 0.7036601305007935, "learning_rate": 0.0001, "elbo": 3.53762, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5646844944874503}
{"step": 60785, "loss": 3.5376220703125, "grad_norm": 0.7036601305007935, "learning_rate": 0.0001, "elbo": 3.53762, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5646844944874503}
{"step": 60790, "loss": 3.548618698120117, "grad_norm": 0.6794653534889221, "learning_rate": 0.0001, "elbo": 3.54862, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5649777152240207}
{"step": 60790, "loss": 3.548618698120117, "grad_norm": 0.6794653534889221, "learning_rate": 0.0001, "elbo": 3.54862, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5649777152240207}
{"step": 60795, "loss": 3.549610137939453, "grad_norm": 0.7347942590713501, "learning_rate": 0.0001, "elbo": 3.54961, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.565270935960591}
{"step": 60800, "loss": 3.5537139892578127, "grad_norm": 0.7713341116905212, "learning_rate": 0.0001, "elbo": 3.55371, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5655641566971616}
{"step": 60805, "loss": 3.5604007720947264, "grad_norm": 0.6733575463294983, "learning_rate": 0.0001, "elbo": 3.5604, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.565857377433732}
{"step": 60810, "loss": 3.5499156951904296, "grad_norm": 0.969165027141571, "learning_rate": 0.0001, "elbo": 3.54992, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5661505981703026}
{"step": 60815, "loss": 3.549558639526367, "grad_norm": 0.8621199131011963, "learning_rate": 0.0001, "elbo": 3.54956, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.566443818906873}
{"step": 60820, "loss": 3.5597789764404295, "grad_norm": 0.8013912439346313, "learning_rate": 0.0001, "elbo": 3.55978, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5667370396434435}
{"step": 60825, "loss": 3.545513916015625, "grad_norm": 0.7212707996368408, "learning_rate": 0.0001, "elbo": 3.54551, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.567030260380014}
{"step": 60830, "loss": 3.541215515136719, "grad_norm": 0.7989212870597839, "learning_rate": 0.0001, "elbo": 3.54122, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.567323481116585}
{"step": 60835, "loss": 3.5426143646240233, "grad_norm": 0.7621257901191711, "learning_rate": 0.0001, "elbo": 3.54261, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5676167018531553}
{"step": 60840, "loss": 3.551908493041992, "grad_norm": 0.697367250919342, "learning_rate": 0.0001, "elbo": 3.55191, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5679099225897257}
{"step": 60845, "loss": 3.5523963928222657, "grad_norm": 0.7387575507164001, "learning_rate": 0.0001, "elbo": 3.5524, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.568203143326296}
{"step": 60850, "loss": 3.5575241088867187, "grad_norm": 0.6546139717102051, "learning_rate": 0.0001, "elbo": 3.55752, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5684963640628666}
{"step": 60855, "loss": 3.554098129272461, "grad_norm": 0.6794045567512512, "learning_rate": 0.0001, "elbo": 3.5541, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.568789584799437}
{"step": 60860, "loss": 3.552852249145508, "grad_norm": 1.7012532949447632, "learning_rate": 0.0001, "elbo": 3.55285, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5690828055360075}
{"step": 60865, "loss": 3.5536239624023436, "grad_norm": 0.7091650366783142, "learning_rate": 0.0001, "elbo": 3.55362, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.569376026272578}
{"step": 60870, "loss": 3.5500537872314455, "grad_norm": 0.7046858668327332, "learning_rate": 0.0001, "elbo": 3.55005, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5696692470091484}
{"step": 60875, "loss": 3.5656993865966795, "grad_norm": 0.7335031628608704, "learning_rate": 0.0001, "elbo": 3.5657, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.569962467745719}
{"step": 60880, "loss": 3.5554168701171873, "grad_norm": 0.7417726516723633, "learning_rate": 0.0001, "elbo": 3.55542, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5702556884822894}
{"step": 60885, "loss": 3.529134750366211, "grad_norm": 0.7398912906646729, "learning_rate": 0.0001, "elbo": 3.52913, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.57054890921886}
{"step": 60890, "loss": 3.5647010803222656, "grad_norm": 0.7407322525978088, "learning_rate": 0.0001, "elbo": 3.5647, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5708421299554303}
{"step": 60895, "loss": 3.5470115661621096, "grad_norm": 0.689990758895874, "learning_rate": 0.0001, "elbo": 3.54701, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5711353506920007}
{"step": 60900, "loss": 3.54071044921875, "grad_norm": 0.674950361251831, "learning_rate": 0.0001, "elbo": 3.54071, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.571428571428571}
{"step": 60905, "loss": 3.5334667205810546, "grad_norm": 0.760790228843689, "learning_rate": 0.0001, "elbo": 3.53347, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.571721792165142}
{"step": 60910, "loss": 3.5361602783203123, "grad_norm": 0.8320514559745789, "learning_rate": 0.0001, "elbo": 3.53616, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5720150129017125}
{"step": 60915, "loss": 3.5615840911865235, "grad_norm": 0.7281244397163391, "learning_rate": 0.0001, "elbo": 3.56158, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.572308233638283}
{"step": 60920, "loss": 3.550209808349609, "grad_norm": 0.7199740409851074, "learning_rate": 0.0001, "elbo": 3.55021, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5726014543748534}
{"step": 60925, "loss": 3.559670257568359, "grad_norm": 0.6562511920928955, "learning_rate": 0.0001, "elbo": 3.55967, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.572894675111424}
{"step": 60930, "loss": 3.5557254791259765, "grad_norm": 0.868263304233551, "learning_rate": 0.0001, "elbo": 3.55573, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5731878958479943}
{"step": 60935, "loss": 3.5373291015625, "grad_norm": 0.7291758060455322, "learning_rate": 0.0001, "elbo": 3.53733, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.573481116584565}
{"step": 60940, "loss": 3.537638854980469, "grad_norm": 0.6827875971794128, "learning_rate": 0.0001, "elbo": 3.53764, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5737743373211353}
{"step": 60945, "loss": 3.541587066650391, "grad_norm": 0.6886453032493591, "learning_rate": 0.0001, "elbo": 3.54159, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5740675580577057}
{"step": 60950, "loss": 3.5555992126464844, "grad_norm": 0.692984402179718, "learning_rate": 0.0001, "elbo": 3.5556, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5743607787942766}
{"step": 60955, "loss": 3.541122817993164, "grad_norm": 0.7011786103248596, "learning_rate": 0.0001, "elbo": 3.54112, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.574653999530847}
{"step": 60960, "loss": 3.5505882263183595, "grad_norm": 0.6979993581771851, "learning_rate": 0.0001, "elbo": 3.55059, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5749472202674175}
{"step": 60965, "loss": 3.5570812225341797, "grad_norm": 0.8797410130500793, "learning_rate": 0.0001, "elbo": 3.55708, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.575240441003988}
{"step": 60970, "loss": 3.5364952087402344, "grad_norm": 0.7041700482368469, "learning_rate": 0.0001, "elbo": 3.5365, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5755336617405584}
{"step": 60975, "loss": 3.533835220336914, "grad_norm": 0.6535214781761169, "learning_rate": 0.0001, "elbo": 3.53384, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.575826882477129}
{"step": 60980, "loss": 3.5511688232421874, "grad_norm": 0.6785179972648621, "learning_rate": 0.0001, "elbo": 3.55117, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5761201032136993}
{"step": 60985, "loss": 3.5512054443359373, "grad_norm": 0.750465452671051, "learning_rate": 0.0001, "elbo": 3.55121, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.57641332395027}
{"step": 60990, "loss": 3.544733428955078, "grad_norm": 0.7076869010925293, "learning_rate": 0.0001, "elbo": 3.54473, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5767065446868402}
{"step": 60995, "loss": 3.5473564147949217, "grad_norm": 0.6897724270820618, "learning_rate": 0.0001, "elbo": 3.54736, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5769997654234107}
{"step": 61000, "loss": 3.5300933837890627, "grad_norm": 0.8193148970603943, "learning_rate": 0.0001, "elbo": 3.53009, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.577292986159981}
{"step": 61005, "loss": 3.543074035644531, "grad_norm": 0.7106198072433472, "learning_rate": 0.0001, "elbo": 3.54307, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5775862068965516}
{"step": 61010, "loss": 3.540386199951172, "grad_norm": 0.7915470004081726, "learning_rate": 0.0001, "elbo": 3.54039, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.577879427633122}
{"step": 61015, "loss": 3.5629981994628905, "grad_norm": 0.6958827376365662, "learning_rate": 0.0001, "elbo": 3.563, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5781726483696925}
{"step": 61020, "loss": 3.5524002075195313, "grad_norm": 0.7516231536865234, "learning_rate": 0.0001, "elbo": 3.5524, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.578465869106263}
{"step": 61025, "loss": 3.5512771606445312, "grad_norm": 0.7370320558547974, "learning_rate": 0.0001, "elbo": 3.55128, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5787590898428334}
{"step": 61030, "loss": 3.533071517944336, "grad_norm": 0.6955842971801758, "learning_rate": 0.0001, "elbo": 3.53307, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5790523105794043}
{"step": 61035, "loss": 3.5525962829589846, "grad_norm": 0.8001784682273865, "learning_rate": 0.0001, "elbo": 3.5526, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5793455313159748}
{"step": 61040, "loss": 3.5616580963134767, "grad_norm": 0.6785815954208374, "learning_rate": 0.0001, "elbo": 3.56166, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5796387520525452}
{"step": 61045, "loss": 3.5594688415527345, "grad_norm": 0.822092592716217, "learning_rate": 0.0001, "elbo": 3.55947, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5799319727891157}
{"step": 61050, "loss": 3.5527416229248048, "grad_norm": 0.7403733134269714, "learning_rate": 0.0001, "elbo": 3.55274, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.580225193525686}
{"step": 61055, "loss": 3.5432106018066407, "grad_norm": 0.8321990966796875, "learning_rate": 0.0001, "elbo": 3.54321, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5805184142622566}
{"step": 61060, "loss": 3.5452075958251954, "grad_norm": 0.7361461520195007, "learning_rate": 0.0001, "elbo": 3.54521, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.580811634998827}
{"step": 61065, "loss": 3.5417549133300783, "grad_norm": 0.7104020118713379, "learning_rate": 0.0001, "elbo": 3.54175, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5811048557353975}
{"step": 61070, "loss": 3.5313507080078126, "grad_norm": 0.698090672492981, "learning_rate": 0.0001, "elbo": 3.53135, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.581398076471968}
{"step": 61075, "loss": 3.538859558105469, "grad_norm": 1.0532106161117554, "learning_rate": 0.0001, "elbo": 3.53886, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.581691297208539}
{"step": 61080, "loss": 3.5482227325439455, "grad_norm": 0.680268406867981, "learning_rate": 0.0001, "elbo": 3.54822, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5819845179451093}
{"step": 61085, "loss": 3.551812744140625, "grad_norm": 0.7359430193901062, "learning_rate": 0.0001, "elbo": 3.55181, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5822777386816798}
{"step": 61085, "loss": 3.551812744140625, "grad_norm": 0.7359430193901062, "learning_rate": 0.0001, "elbo": 3.55181, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5822777386816798}
{"step": 61090, "loss": 3.5475181579589843, "grad_norm": 0.7205405235290527, "learning_rate": 0.0001, "elbo": 3.54752, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.58257095941825}
{"step": 61095, "loss": 3.533612060546875, "grad_norm": 0.6505975723266602, "learning_rate": 0.0001, "elbo": 3.53361, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5828641801548207}
{"step": 61100, "loss": 3.5594371795654296, "grad_norm": 0.6622087359428406, "learning_rate": 0.0001, "elbo": 3.55944, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.583157400891391}
{"step": 61105, "loss": 3.5427410125732424, "grad_norm": 0.7404146790504456, "learning_rate": 0.0001, "elbo": 3.54274, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5834506216279616}
{"step": 61110, "loss": 3.5379707336425783, "grad_norm": 0.6487672328948975, "learning_rate": 0.0001, "elbo": 3.53797, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.583743842364532}
{"step": 61115, "loss": 3.55554084777832, "grad_norm": 0.7299860119819641, "learning_rate": 0.0001, "elbo": 3.55554, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5840370631011025}
{"step": 61120, "loss": 3.5491371154785156, "grad_norm": 0.6924890875816345, "learning_rate": 0.0001, "elbo": 3.54914, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.584330283837673}
{"step": 61125, "loss": 3.5541202545166017, "grad_norm": 0.7422490119934082, "learning_rate": 0.0001, "elbo": 3.55412, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5846235045742434}
{"step": 61130, "loss": 3.5393733978271484, "grad_norm": 0.7152783274650574, "learning_rate": 0.0001, "elbo": 3.53937, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.584916725310814}
{"step": 61135, "loss": 3.549763488769531, "grad_norm": 0.6985276937484741, "learning_rate": 0.0001, "elbo": 3.54976, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5852099460473843}
{"step": 61140, "loss": 3.525870513916016, "grad_norm": 0.708832859992981, "learning_rate": 0.0001, "elbo": 3.52587, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5855031667839548}
{"step": 61145, "loss": 3.538813018798828, "grad_norm": 0.7146233320236206, "learning_rate": 0.0001, "elbo": 3.53881, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.585796387520525}
{"step": 61150, "loss": 3.547269821166992, "grad_norm": 0.6807522177696228, "learning_rate": 0.0001, "elbo": 3.54727, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.586089608257096}
{"step": 61155, "loss": 3.5348072052001953, "grad_norm": 0.6960017681121826, "learning_rate": 0.0001, "elbo": 3.53481, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5863828289936666}
{"step": 61160, "loss": 3.54583854675293, "grad_norm": 0.6722108125686646, "learning_rate": 0.0001, "elbo": 3.54584, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.586676049730237}
{"step": 61165, "loss": 3.54809684753418, "grad_norm": 0.7045308351516724, "learning_rate": 0.0001, "elbo": 3.5481, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5869692704668075}
{"step": 61170, "loss": 3.5321922302246094, "grad_norm": 0.6757416725158691, "learning_rate": 0.0001, "elbo": 3.53219, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.587262491203378}
{"step": 61170, "loss": 3.5321922302246094, "grad_norm": 0.6757416725158691, "learning_rate": 0.0001, "elbo": 3.53219, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.587262491203378}
{"step": 61175, "loss": 3.5423038482666014, "grad_norm": 0.772392213344574, "learning_rate": 0.0001, "elbo": 3.5423, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5875557119399484}
{"step": 61180, "loss": 3.547705078125, "grad_norm": 0.6810935139656067, "learning_rate": 0.0001, "elbo": 3.54771, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.587848932676519}
{"step": 61185, "loss": 3.5315219879150392, "grad_norm": 0.6760885715484619, "learning_rate": 0.0001, "elbo": 3.53152, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5881421534130893}
{"step": 61190, "loss": 3.536028289794922, "grad_norm": 0.7337087392807007, "learning_rate": 0.0001, "elbo": 3.53603, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5884353741496597}
{"step": 61195, "loss": 3.5437763214111326, "grad_norm": 0.6571835279464722, "learning_rate": 0.0001, "elbo": 3.54378, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.58872859488623}
{"step": 61200, "loss": 3.542737579345703, "grad_norm": 0.7090574502944946, "learning_rate": 0.0001, "elbo": 3.54274, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.589021815622801}
{"step": 61205, "loss": 3.549843597412109, "grad_norm": 0.6584180593490601, "learning_rate": 0.0001, "elbo": 3.54984, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5893150363593715}
{"step": 61210, "loss": 3.538850021362305, "grad_norm": 0.6934762597084045, "learning_rate": 0.0001, "elbo": 3.53885, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.589608257095942}
{"step": 61215, "loss": 3.5217945098876955, "grad_norm": 0.7285548448562622, "learning_rate": 0.0001, "elbo": 3.52179, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5899014778325125}
{"step": 61220, "loss": 3.5460670471191404, "grad_norm": 0.6593888401985168, "learning_rate": 0.0001, "elbo": 3.54607, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.590194698569083}
{"step": 61225, "loss": 3.53565673828125, "grad_norm": 0.7278172373771667, "learning_rate": 0.0001, "elbo": 3.53566, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5904879193056534}
{"step": 61230, "loss": 3.5313072204589844, "grad_norm": 0.8060168623924255, "learning_rate": 0.0001, "elbo": 3.53131, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.590781140042224}
{"step": 61235, "loss": 3.550893020629883, "grad_norm": 0.6755350828170776, "learning_rate": 0.0001, "elbo": 3.55089, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5910743607787943}
{"step": 61240, "loss": 3.5533370971679688, "grad_norm": 0.7498945593833923, "learning_rate": 0.0001, "elbo": 3.55334, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5913675815153647}
{"step": 61240, "loss": 3.5533370971679688, "grad_norm": 0.7498945593833923, "learning_rate": 0.0001, "elbo": 3.55334, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5913675815153647}
{"step": 61245, "loss": 3.5575767517089845, "grad_norm": 0.7119483947753906, "learning_rate": 0.0001, "elbo": 3.55758, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.591660802251935}
{"step": 61250, "loss": 3.5467979431152346, "grad_norm": 0.7164870500564575, "learning_rate": 0.0001, "elbo": 3.5468, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5919540229885056}
{"step": 61255, "loss": 3.553419494628906, "grad_norm": 0.7184237837791443, "learning_rate": 0.0001, "elbo": 3.55342, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.592247243725076}
{"step": 61260, "loss": 3.558880615234375, "grad_norm": 0.7294433116912842, "learning_rate": 0.0001, "elbo": 3.55888, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5925404644616465}
{"step": 61260, "loss": 3.558880615234375, "grad_norm": 0.7294433116912842, "learning_rate": 0.0001, "elbo": 3.55888, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5925404644616465}
{"step": 61265, "loss": 3.5499420166015625, "grad_norm": 0.8437855839729309, "learning_rate": 0.0001, "elbo": 3.54994, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.592833685198217}
{"step": 61270, "loss": 3.55460205078125, "grad_norm": 0.7261134386062622, "learning_rate": 0.0001, "elbo": 3.5546, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5931269059347875}
{"step": 61275, "loss": 3.5383560180664064, "grad_norm": 0.690022349357605, "learning_rate": 0.0001, "elbo": 3.53836, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5934201266713584}
{"step": 61280, "loss": 3.552079772949219, "grad_norm": 0.691619336605072, "learning_rate": 0.0001, "elbo": 3.55208, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.593713347407929}
{"step": 61285, "loss": 3.54031982421875, "grad_norm": 0.6980036497116089, "learning_rate": 0.0001, "elbo": 3.54032, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5940065681444993}
{"step": 61290, "loss": 3.541242980957031, "grad_norm": 0.7125937342643738, "learning_rate": 0.0001, "elbo": 3.54124, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5942997888810697}
{"step": 61295, "loss": 3.5344253540039063, "grad_norm": 0.7424493432044983, "learning_rate": 0.0001, "elbo": 3.53443, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.59459300961764}
{"step": 61300, "loss": 3.543781280517578, "grad_norm": 0.9567355513572693, "learning_rate": 0.0001, "elbo": 3.54378, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5948862303542106}
{"step": 61305, "loss": 3.555415725708008, "grad_norm": 0.717352032661438, "learning_rate": 0.0001, "elbo": 3.55542, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.595179451090781}
{"step": 61310, "loss": 3.558094787597656, "grad_norm": 0.6882677674293518, "learning_rate": 0.0001, "elbo": 3.55809, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5954726718273515}
{"step": 61315, "loss": 3.551341247558594, "grad_norm": 0.7149361968040466, "learning_rate": 0.0001, "elbo": 3.55134, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.595765892563922}
{"step": 61320, "loss": 3.5368934631347657, "grad_norm": 0.6937105655670166, "learning_rate": 0.0001, "elbo": 3.53689, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.596059113300493}
{"step": 61325, "loss": 3.5531917572021485, "grad_norm": 0.7795735597610474, "learning_rate": 0.0001, "elbo": 3.55319, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5963523340370633}
{"step": 61330, "loss": 3.5452178955078124, "grad_norm": 0.6786300539970398, "learning_rate": 0.0001, "elbo": 3.54522, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.596645554773634}
{"step": 61335, "loss": 3.5515247344970704, "grad_norm": 0.7417582273483276, "learning_rate": 0.0001, "elbo": 3.55152, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5969387755102042}
{"step": 61340, "loss": 3.5541683197021485, "grad_norm": 0.7111726999282837, "learning_rate": 0.0001, "elbo": 3.55417, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5972319962467747}
{"step": 61345, "loss": 3.5447509765625, "grad_norm": 0.7323240637779236, "learning_rate": 0.0001, "elbo": 3.54475, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.597525216983345}
{"step": 61350, "loss": 3.5508186340332033, "grad_norm": 0.7109841108322144, "learning_rate": 0.0001, "elbo": 3.55082, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5978184377199156}
{"step": 61355, "loss": 3.5247730255126952, "grad_norm": 0.7529364824295044, "learning_rate": 0.0001, "elbo": 3.52477, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.598111658456486}
{"step": 61360, "loss": 3.553790283203125, "grad_norm": 0.7328883409500122, "learning_rate": 0.0001, "elbo": 3.55379, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5984048791930565}
{"step": 61365, "loss": 3.533107376098633, "grad_norm": 0.7719948291778564, "learning_rate": 0.0001, "elbo": 3.53311, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.598698099929627}
{"step": 61370, "loss": 3.551671600341797, "grad_norm": 0.6944632530212402, "learning_rate": 0.0001, "elbo": 3.55167, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5989913206661974}
{"step": 61375, "loss": 3.546796417236328, "grad_norm": 0.8177424669265747, "learning_rate": 0.0001, "elbo": 3.5468, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.599284541402768}
{"step": 61380, "loss": 3.553235626220703, "grad_norm": 0.7175855040550232, "learning_rate": 0.0001, "elbo": 3.55324, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.5995777621393383}
{"step": 61385, "loss": 3.557956314086914, "grad_norm": 0.6975995898246765, "learning_rate": 0.0001, "elbo": 3.55796, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.599870982875909}
{"step": 61390, "loss": 3.557292938232422, "grad_norm": 0.8065992593765259, "learning_rate": 0.0001, "elbo": 3.55729, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6001642036124792}
{"step": 61395, "loss": 3.5368816375732424, "grad_norm": 0.7273654341697693, "learning_rate": 0.0001, "elbo": 3.53688, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6004574243490497}
{"step": 61395, "loss": 3.5368816375732424, "grad_norm": 0.7273654341697693, "learning_rate": 0.0001, "elbo": 3.53688, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6004574243490497}
{"step": 61400, "loss": 3.5400764465332033, "grad_norm": 0.6690455079078674, "learning_rate": 0.0001, "elbo": 3.54008, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6007506450856206}
{"step": 61405, "loss": 3.53314094543457, "grad_norm": 0.7143991589546204, "learning_rate": 0.0001, "elbo": 3.53314, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.601043865822191}
{"step": 61410, "loss": 3.5424789428710937, "grad_norm": 0.7325950860977173, "learning_rate": 0.0001, "elbo": 3.54248, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6013370865587615}
{"step": 61415, "loss": 3.5467681884765625, "grad_norm": 0.6991627812385559, "learning_rate": 0.0001, "elbo": 3.54677, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.601630307295332}
{"step": 61420, "loss": 3.5401626586914063, "grad_norm": 0.7054100036621094, "learning_rate": 0.0001, "elbo": 3.54016, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6019235280319024}
{"step": 61425, "loss": 3.5404090881347656, "grad_norm": 0.688983678817749, "learning_rate": 0.0001, "elbo": 3.54041, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.602216748768473}
{"step": 61430, "loss": 3.535277557373047, "grad_norm": 0.7789533734321594, "learning_rate": 0.0001, "elbo": 3.53528, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6025099695050433}
{"step": 61435, "loss": 3.554888916015625, "grad_norm": 0.6703556180000305, "learning_rate": 0.0001, "elbo": 3.55489, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6028031902416138}
{"step": 61440, "loss": 3.5424915313720704, "grad_norm": 0.7009469270706177, "learning_rate": 0.0001, "elbo": 3.54249, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6030964109781842}
{"step": 61445, "loss": 3.5442520141601563, "grad_norm": 0.7343506217002869, "learning_rate": 0.0001, "elbo": 3.54425, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.603389631714755}
{"step": 61450, "loss": 3.552256774902344, "grad_norm": 0.7084499001502991, "learning_rate": 0.0001, "elbo": 3.55226, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6036828524513256}
{"step": 61455, "loss": 3.535514068603516, "grad_norm": 0.6827598810195923, "learning_rate": 0.0001, "elbo": 3.53551, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.603976073187896}
{"step": 61460, "loss": 3.5486495971679686, "grad_norm": 0.6387110948562622, "learning_rate": 0.0001, "elbo": 3.54865, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6042692939244665}
{"step": 61465, "loss": 3.54705810546875, "grad_norm": 0.7361999154090881, "learning_rate": 0.0001, "elbo": 3.54706, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.604562514661037}
{"step": 61470, "loss": 3.5401309967041015, "grad_norm": 0.7516878247261047, "learning_rate": 0.0001, "elbo": 3.54013, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6048557353976074}
{"step": 61475, "loss": 3.5512428283691406, "grad_norm": 0.7209867238998413, "learning_rate": 0.0001, "elbo": 3.55124, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.605148956134178}
{"step": 61480, "loss": 3.545515441894531, "grad_norm": 0.7280404567718506, "learning_rate": 0.0001, "elbo": 3.54552, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6054421768707483}
{"step": 61485, "loss": 3.5443088531494142, "grad_norm": 0.7282228469848633, "learning_rate": 0.0001, "elbo": 3.54431, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6057353976073188}
{"step": 61490, "loss": 3.5440605163574217, "grad_norm": 0.7724253535270691, "learning_rate": 0.0001, "elbo": 3.54406, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.606028618343889}
{"step": 61495, "loss": 3.524394989013672, "grad_norm": 0.7869176268577576, "learning_rate": 0.0001, "elbo": 3.5244, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6063218390804597}
{"step": 61500, "loss": 3.5340484619140624, "grad_norm": 0.9463815093040466, "learning_rate": 0.0001, "elbo": 3.53405, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.60661505981703}
{"step": 61505, "loss": 3.5336498260498046, "grad_norm": 0.9542508125305176, "learning_rate": 0.0001, "elbo": 3.53365, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6069082805536006}
{"step": 61510, "loss": 3.541271209716797, "grad_norm": 0.7038689851760864, "learning_rate": 0.0001, "elbo": 3.54127, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.607201501290171}
{"step": 61515, "loss": 3.5607894897460937, "grad_norm": 0.7313372492790222, "learning_rate": 0.0001, "elbo": 3.56079, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6074947220267415}
{"step": 61520, "loss": 3.5635936737060545, "grad_norm": 0.868631899356842, "learning_rate": 0.0001, "elbo": 3.56359, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6077879427633124}
{"step": 61525, "loss": 3.5454444885253906, "grad_norm": 0.7198165059089661, "learning_rate": 0.0001, "elbo": 3.54544, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.608081163499883}
{"step": 61530, "loss": 3.542070770263672, "grad_norm": 0.684975802898407, "learning_rate": 0.0001, "elbo": 3.54207, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6083743842364533}
{"step": 61535, "loss": 3.546242523193359, "grad_norm": 0.6867387890815735, "learning_rate": 0.0001, "elbo": 3.54624, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6086676049730237}
{"step": 61540, "loss": 3.5532955169677733, "grad_norm": 0.6607108116149902, "learning_rate": 0.0001, "elbo": 3.5533, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.608960825709594}
{"step": 61545, "loss": 3.526081848144531, "grad_norm": 0.7198011875152588, "learning_rate": 0.0001, "elbo": 3.52608, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6092540464461647}
{"step": 61550, "loss": 3.5340476989746095, "grad_norm": 0.7974942326545715, "learning_rate": 0.0001, "elbo": 3.53405, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.609547267182735}
{"step": 61555, "loss": 3.5546749114990233, "grad_norm": 0.677314817905426, "learning_rate": 0.0001, "elbo": 3.55467, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6098404879193056}
{"step": 61560, "loss": 3.535472106933594, "grad_norm": 0.6763891577720642, "learning_rate": 0.0001, "elbo": 3.53547, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.610133708655876}
{"step": 61565, "loss": 3.5248291015625, "grad_norm": 0.7550790309906006, "learning_rate": 0.0001, "elbo": 3.52483, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6104269293924465}
{"step": 61570, "loss": 3.5467464447021486, "grad_norm": 0.716048538684845, "learning_rate": 0.0001, "elbo": 3.54675, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6107201501290174}
{"step": 61575, "loss": 3.5341781616210937, "grad_norm": 0.7520862221717834, "learning_rate": 0.0001, "elbo": 3.53418, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.611013370865588}
{"step": 61580, "loss": 3.5515785217285156, "grad_norm": 0.8419016003608704, "learning_rate": 0.0001, "elbo": 3.55158, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6113065916021583}
{"step": 61585, "loss": 3.5266746520996093, "grad_norm": 0.6809203028678894, "learning_rate": 0.0001, "elbo": 3.52667, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6115998123387287}
{"step": 61590, "loss": 3.5495872497558594, "grad_norm": 0.8193085789680481, "learning_rate": 0.0001, "elbo": 3.54959, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.611893033075299}
{"step": 61595, "loss": 3.549420928955078, "grad_norm": 0.7185049057006836, "learning_rate": 0.0001, "elbo": 3.54942, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6121862538118696}
{"step": 61595, "loss": 3.549420928955078, "grad_norm": 0.7185049057006836, "learning_rate": 0.0001, "elbo": 3.54942, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6121862538118696}
{"step": 61600, "loss": 3.552347183227539, "grad_norm": 0.7135487794876099, "learning_rate": 0.0001, "elbo": 3.55235, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.61247947454844}
{"step": 61605, "loss": 3.5320411682128907, "grad_norm": 0.6749423146247864, "learning_rate": 0.0001, "elbo": 3.53204, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6127726952850105}
{"step": 61610, "loss": 3.5420433044433595, "grad_norm": 0.7333958148956299, "learning_rate": 0.0001, "elbo": 3.54204, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.613065916021581}
{"step": 61615, "loss": 3.5537704467773437, "grad_norm": 0.8767337203025818, "learning_rate": 0.0001, "elbo": 3.55377, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6133591367581515}
{"step": 61620, "loss": 3.5611244201660157, "grad_norm": 0.7170356512069702, "learning_rate": 0.0001, "elbo": 3.56112, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.613652357494722}
{"step": 61625, "loss": 3.5451728820800783, "grad_norm": 0.7071133255958557, "learning_rate": 0.0001, "elbo": 3.54517, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6139455782312924}
{"step": 61625, "loss": 3.5451728820800783, "grad_norm": 0.7071133255958557, "learning_rate": 0.0001, "elbo": 3.54517, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6139455782312924}
{"step": 61630, "loss": 3.524658203125, "grad_norm": 0.6886772513389587, "learning_rate": 0.0001, "elbo": 3.52466, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.614238798967863}
{"step": 61635, "loss": 3.5446502685546877, "grad_norm": 0.7499132752418518, "learning_rate": 0.0001, "elbo": 3.54465, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6145320197044333}
{"step": 61640, "loss": 3.550462341308594, "grad_norm": 0.9664430022239685, "learning_rate": 0.0001, "elbo": 3.55046, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6148252404410037}
{"step": 61640, "loss": 3.550462341308594, "grad_norm": 0.9664430022239685, "learning_rate": 0.0001, "elbo": 3.55046, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6148252404410037}
{"step": 61645, "loss": 3.5480777740478517, "grad_norm": 0.6715177893638611, "learning_rate": 0.0001, "elbo": 3.54808, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6151184611775746}
{"step": 61650, "loss": 3.5401206970214845, "grad_norm": 0.6934229135513306, "learning_rate": 0.0001, "elbo": 3.54012, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.615411681914145}
{"step": 61655, "loss": 3.546944427490234, "grad_norm": 0.7641434669494629, "learning_rate": 0.0001, "elbo": 3.54694, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6157049026507155}
{"step": 61660, "loss": 3.5335254669189453, "grad_norm": 0.7039433121681213, "learning_rate": 0.0001, "elbo": 3.53353, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.615998123387286}
{"step": 61665, "loss": 3.552346038818359, "grad_norm": 0.7186976075172424, "learning_rate": 0.0001, "elbo": 3.55235, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6162913441238564}
{"step": 61670, "loss": 3.542301559448242, "grad_norm": 0.7018582224845886, "learning_rate": 0.0001, "elbo": 3.5423, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.616584564860427}
{"step": 61675, "loss": 3.550364685058594, "grad_norm": 0.7496719360351562, "learning_rate": 0.0001, "elbo": 3.55036, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6168777855969974}
{"step": 61680, "loss": 3.5571125030517576, "grad_norm": 0.7224340438842773, "learning_rate": 0.0001, "elbo": 3.55711, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.617171006333568}
{"step": 61685, "loss": 3.5480899810791016, "grad_norm": 0.6817736029624939, "learning_rate": 0.0001, "elbo": 3.54809, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6174642270701383}
{"step": 61690, "loss": 3.539989471435547, "grad_norm": 0.8452966213226318, "learning_rate": 0.0001, "elbo": 3.53999, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.617757447806709}
{"step": 61695, "loss": 3.5470199584960938, "grad_norm": 0.7363327145576477, "learning_rate": 0.0001, "elbo": 3.54702, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6180506685432796}
{"step": 61700, "loss": 3.5382049560546873, "grad_norm": 0.6541802287101746, "learning_rate": 0.0001, "elbo": 3.5382, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.61834388927985}
{"step": 61705, "loss": 3.54652099609375, "grad_norm": 0.6760163307189941, "learning_rate": 0.0001, "elbo": 3.54652, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6186371100164205}
{"step": 61710, "loss": 3.5312774658203123, "grad_norm": 0.6679306626319885, "learning_rate": 0.0001, "elbo": 3.53128, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.618930330752991}
{"step": 61715, "loss": 3.543354034423828, "grad_norm": 0.7048177719116211, "learning_rate": 0.0001, "elbo": 3.54335, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6192235514895614}
{"step": 61720, "loss": 3.543814468383789, "grad_norm": 0.7111198306083679, "learning_rate": 0.0001, "elbo": 3.54381, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.619516772226132}
{"step": 61720, "loss": 3.543814468383789, "grad_norm": 0.7111198306083679, "learning_rate": 0.0001, "elbo": 3.54381, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.619516772226132}
{"step": 61725, "loss": 3.5435775756835937, "grad_norm": 0.7210767865180969, "learning_rate": 0.0001, "elbo": 3.54358, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6198099929627023}
{"step": 61730, "loss": 3.5399658203125, "grad_norm": 0.6872220635414124, "learning_rate": 0.0001, "elbo": 3.53997, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.620103213699273}
{"step": 61735, "loss": 3.5352264404296876, "grad_norm": 0.7904726266860962, "learning_rate": 0.0001, "elbo": 3.53523, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6203964344358432}
{"step": 61740, "loss": 3.5487552642822267, "grad_norm": 0.7528403401374817, "learning_rate": 0.0001, "elbo": 3.54876, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6206896551724137}
{"step": 61745, "loss": 3.5566917419433595, "grad_norm": 0.7047684192657471, "learning_rate": 0.0001, "elbo": 3.55669, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.620982875908984}
{"step": 61750, "loss": 3.5403427124023437, "grad_norm": 0.7042796611785889, "learning_rate": 0.0001, "elbo": 3.54034, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6212760966455546}
{"step": 61755, "loss": 3.543578338623047, "grad_norm": 0.6569521427154541, "learning_rate": 0.0001, "elbo": 3.54358, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.621569317382125}
{"step": 61760, "loss": 3.5321720123291014, "grad_norm": 0.8977743983268738, "learning_rate": 0.0001, "elbo": 3.53217, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6218625381186955}
{"step": 61765, "loss": 3.5552719116210936, "grad_norm": 1.0902864933013916, "learning_rate": 0.0001, "elbo": 3.55527, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.622155758855266}
{"step": 61770, "loss": 3.5358596801757813, "grad_norm": 0.7336269617080688, "learning_rate": 0.0001, "elbo": 3.53586, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.622448979591837}
{"step": 61775, "loss": 3.551818084716797, "grad_norm": 0.778049886226654, "learning_rate": 0.0001, "elbo": 3.55182, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6227422003284073}
{"step": 61780, "loss": 3.5451332092285157, "grad_norm": 0.6863589286804199, "learning_rate": 0.0001, "elbo": 3.54513, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.623035421064978}
{"step": 61785, "loss": 3.5423629760742186, "grad_norm": 0.6827208399772644, "learning_rate": 0.0001, "elbo": 3.54236, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6233286418015482}
{"step": 61790, "loss": 3.5491992950439455, "grad_norm": 0.7036159634590149, "learning_rate": 0.0001, "elbo": 3.5492, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6236218625381187}
{"step": 61795, "loss": 3.5367523193359376, "grad_norm": 0.730327844619751, "learning_rate": 0.0001, "elbo": 3.53675, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.623915083274689}
{"step": 61800, "loss": 3.557709503173828, "grad_norm": 0.6905478835105896, "learning_rate": 0.0001, "elbo": 3.55771, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6242083040112596}
{"step": 61805, "loss": 3.5481903076171877, "grad_norm": 0.7653486728668213, "learning_rate": 0.0001, "elbo": 3.54819, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.62450152474783}
{"step": 61810, "loss": 3.5419071197509764, "grad_norm": 0.7959215641021729, "learning_rate": 0.0001, "elbo": 3.54191, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6247947454844005}
{"step": 61815, "loss": 3.5530494689941405, "grad_norm": 0.8261363506317139, "learning_rate": 0.0001, "elbo": 3.55305, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6250879662209714}
{"step": 61820, "loss": 3.5499713897705076, "grad_norm": 0.9723787307739258, "learning_rate": 0.0001, "elbo": 3.54997, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.625381186957542}
{"step": 61825, "loss": 3.5263519287109375, "grad_norm": 0.7294108271598816, "learning_rate": 0.0001, "elbo": 3.52635, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6256744076941123}
{"step": 61830, "loss": 3.55068359375, "grad_norm": 0.6936460733413696, "learning_rate": 0.0001, "elbo": 3.55068, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6259676284306828}
{"step": 61835, "loss": 3.5426048278808593, "grad_norm": 0.7415587306022644, "learning_rate": 0.0001, "elbo": 3.5426, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.626260849167253}
{"step": 61840, "loss": 3.5540283203125, "grad_norm": 0.6866254210472107, "learning_rate": 0.0001, "elbo": 3.55403, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6265540699038237}
{"step": 61845, "loss": 3.545518493652344, "grad_norm": 0.7463994026184082, "learning_rate": 0.0001, "elbo": 3.54552, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.626847290640394}
{"step": 61850, "loss": 3.545967864990234, "grad_norm": 1.0188244581222534, "learning_rate": 0.0001, "elbo": 3.54597, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6271405113769646}
{"step": 61850, "loss": 3.545967864990234, "grad_norm": 1.0188244581222534, "learning_rate": 0.0001, "elbo": 3.54597, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6271405113769646}
{"step": 61855, "loss": 3.5464649200439453, "grad_norm": 0.6756244897842407, "learning_rate": 0.0001, "elbo": 3.54646, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.627433732113535}
{"step": 61860, "loss": 3.5429100036621093, "grad_norm": 0.7180142402648926, "learning_rate": 0.0001, "elbo": 3.54291, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6277269528501055}
{"step": 61865, "loss": 3.5553340911865234, "grad_norm": 0.6824293732643127, "learning_rate": 0.0001, "elbo": 3.55533, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.628020173586676}
{"step": 61865, "loss": 3.5553340911865234, "grad_norm": 0.6824293732643127, "learning_rate": 0.0001, "elbo": 3.55533, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.628020173586676}
{"step": 61870, "loss": 3.5369049072265626, "grad_norm": 0.6397925019264221, "learning_rate": 0.0001, "elbo": 3.5369, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6283133943232464}
{"step": 61875, "loss": 3.5504127502441407, "grad_norm": 0.7609722018241882, "learning_rate": 0.0001, "elbo": 3.55041, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.628606615059817}
{"step": 61880, "loss": 3.553118896484375, "grad_norm": 0.6949121356010437, "learning_rate": 0.0001, "elbo": 3.55312, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6288998357963873}
{"step": 61885, "loss": 3.535115432739258, "grad_norm": 0.6798626184463501, "learning_rate": 0.0001, "elbo": 3.53512, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6291930565329578}
{"step": 61890, "loss": 3.554076385498047, "grad_norm": 0.6744180917739868, "learning_rate": 0.0001, "elbo": 3.55408, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6294862772695287}
{"step": 61895, "loss": 3.549700927734375, "grad_norm": 0.6608065962791443, "learning_rate": 0.0001, "elbo": 3.5497, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.629779498006099}
{"step": 61900, "loss": 3.543338394165039, "grad_norm": 0.6771669387817383, "learning_rate": 0.0001, "elbo": 3.54334, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6300727187426696}
{"step": 61905, "loss": 3.5329742431640625, "grad_norm": 0.7635283470153809, "learning_rate": 0.0001, "elbo": 3.53297, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.63036593947924}
{"step": 61910, "loss": 3.5356449127197265, "grad_norm": 0.7056130170822144, "learning_rate": 0.0001, "elbo": 3.53564, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6306591602158105}
{"step": 61915, "loss": 3.5606369018554687, "grad_norm": 0.7064652442932129, "learning_rate": 0.0001, "elbo": 3.56064, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.630952380952381}
{"step": 61920, "loss": 3.5449790954589844, "grad_norm": 0.6941337585449219, "learning_rate": 0.0001, "elbo": 3.54498, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6312456016889514}
{"step": 61925, "loss": 3.5625675201416014, "grad_norm": 0.6726281046867371, "learning_rate": 0.0001, "elbo": 3.56257, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.631538822425522}
{"step": 61930, "loss": 3.5396751403808593, "grad_norm": 0.705136239528656, "learning_rate": 0.0001, "elbo": 3.53968, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6318320431620923}
{"step": 61930, "loss": 3.5396751403808593, "grad_norm": 0.705136239528656, "learning_rate": 0.0001, "elbo": 3.53968, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6318320431620923}
{"step": 61935, "loss": 3.530369186401367, "grad_norm": 0.7475705146789551, "learning_rate": 0.0001, "elbo": 3.53037, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6321252638986627}
{"step": 61940, "loss": 3.5448333740234377, "grad_norm": 0.6845372915267944, "learning_rate": 0.0001, "elbo": 3.54483, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6324184846352336}
{"step": 61945, "loss": 3.537734603881836, "grad_norm": 0.8178939819335938, "learning_rate": 0.0001, "elbo": 3.53773, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.632711705371804}
{"step": 61950, "loss": 3.5453445434570314, "grad_norm": 0.7136349678039551, "learning_rate": 0.0001, "elbo": 3.54534, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6330049261083746}
{"step": 61955, "loss": 3.5213973999023436, "grad_norm": 0.7196937799453735, "learning_rate": 0.0001, "elbo": 3.5214, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.633298146844945}
{"step": 61960, "loss": 3.529567337036133, "grad_norm": 0.7121350765228271, "learning_rate": 0.0001, "elbo": 3.52957, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6335913675815155}
{"step": 61960, "loss": 3.529567337036133, "grad_norm": 0.7121350765228271, "learning_rate": 0.0001, "elbo": 3.52957, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6335913675815155}
{"step": 61965, "loss": 3.5551216125488283, "grad_norm": 0.7470837831497192, "learning_rate": 0.0001, "elbo": 3.55512, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.633884588318086}
{"step": 61965, "loss": 3.5551216125488283, "grad_norm": 0.7470837831497192, "learning_rate": 0.0001, "elbo": 3.55512, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.633884588318086}
{"step": 61970, "loss": 3.537389373779297, "grad_norm": 0.8071713447570801, "learning_rate": 0.0001, "elbo": 3.53739, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6341778090546564}
{"step": 61975, "loss": 3.5365997314453126, "grad_norm": 1.208485722541809, "learning_rate": 0.0001, "elbo": 3.5366, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.634471029791227}
{"step": 61980, "loss": 3.538685607910156, "grad_norm": 0.7195376753807068, "learning_rate": 0.0001, "elbo": 3.53869, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6347642505277973}
{"step": 61985, "loss": 3.5555488586425783, "grad_norm": 0.8032395243644714, "learning_rate": 0.0001, "elbo": 3.55555, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6350574712643677}
{"step": 61990, "loss": 3.5572818756103515, "grad_norm": 0.869097888469696, "learning_rate": 0.0001, "elbo": 3.55728, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.635350692000938}
{"step": 61995, "loss": 3.5502574920654295, "grad_norm": 0.6838667988777161, "learning_rate": 0.0001, "elbo": 3.55026, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6356439127375086}
{"step": 62000, "loss": 3.5344329833984376, "grad_norm": 0.6735482215881348, "learning_rate": 0.0001, "elbo": 3.53443, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.635937133474079}
{"step": 62005, "loss": 3.5455177307128904, "grad_norm": 0.7148231267929077, "learning_rate": 0.0001, "elbo": 3.54552, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6362303542106496}
{"step": 62005, "loss": 3.5455177307128904, "grad_norm": 0.7148231267929077, "learning_rate": 0.0001, "elbo": 3.54552, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6362303542106496}
{"step": 62010, "loss": 3.53936767578125, "grad_norm": 0.7008134126663208, "learning_rate": 0.0001, "elbo": 3.53937, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.63652357494722}
{"step": 62015, "loss": 3.5398143768310546, "grad_norm": 0.8183714151382446, "learning_rate": 0.0001, "elbo": 3.53981, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.636816795683791}
{"step": 62020, "loss": 3.5429744720458984, "grad_norm": 0.6882288455963135, "learning_rate": 0.0001, "elbo": 3.54297, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6371100164203614}
{"step": 62020, "loss": 3.5429744720458984, "grad_norm": 0.6882288455963135, "learning_rate": 0.0001, "elbo": 3.54297, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6371100164203614}
{"step": 62025, "loss": 3.547176742553711, "grad_norm": 0.6741423606872559, "learning_rate": 0.0001, "elbo": 3.54718, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.637403237156932}
{"step": 62030, "loss": 3.55411376953125, "grad_norm": 0.652627170085907, "learning_rate": 0.0001, "elbo": 3.55411, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6376964578935023}
{"step": 62035, "loss": 3.537201690673828, "grad_norm": 1.0123239755630493, "learning_rate": 0.0001, "elbo": 3.5372, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6379896786300727}
{"step": 62040, "loss": 3.5546470642089845, "grad_norm": 0.8428470492362976, "learning_rate": 0.0001, "elbo": 3.55465, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.638282899366643}
{"step": 62045, "loss": 3.571986770629883, "grad_norm": 0.7012345790863037, "learning_rate": 0.0001, "elbo": 3.57199, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6385761201032136}
{"step": 62050, "loss": 3.5404586791992188, "grad_norm": 0.6976999044418335, "learning_rate": 0.0001, "elbo": 3.54046, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.638869340839784}
{"step": 62055, "loss": 3.5443866729736326, "grad_norm": 0.6907680034637451, "learning_rate": 0.0001, "elbo": 3.54439, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6391625615763545}
{"step": 62060, "loss": 3.549468994140625, "grad_norm": 0.8328830003738403, "learning_rate": 0.0001, "elbo": 3.54947, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6394557823129254}
{"step": 62065, "loss": 3.555297088623047, "grad_norm": 0.6999901533126831, "learning_rate": 0.0001, "elbo": 3.5553, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.639749003049496}
{"step": 62070, "loss": 3.5302196502685548, "grad_norm": 0.685987651348114, "learning_rate": 0.0001, "elbo": 3.53022, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6400422237860663}
{"step": 62075, "loss": 3.5399757385253907, "grad_norm": 0.7140650749206543, "learning_rate": 0.0001, "elbo": 3.53998, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.640335444522637}
{"step": 62080, "loss": 3.556158447265625, "grad_norm": 0.7177784442901611, "learning_rate": 0.0001, "elbo": 3.55616, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6406286652592073}
{"step": 62080, "loss": 3.556158447265625, "grad_norm": 0.7177784442901611, "learning_rate": 0.0001, "elbo": 3.55616, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6406286652592073}
{"step": 62085, "loss": 3.558066558837891, "grad_norm": 0.7704235315322876, "learning_rate": 0.0001, "elbo": 3.55807, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6409218859957777}
{"step": 62085, "loss": 3.558066558837891, "grad_norm": 0.7704235315322876, "learning_rate": 0.0001, "elbo": 3.55807, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6409218859957777}
{"step": 62090, "loss": 3.55135498046875, "grad_norm": 0.7240363955497742, "learning_rate": 0.0001, "elbo": 3.55135, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.641215106732348}
{"step": 62095, "loss": 3.54654541015625, "grad_norm": 0.6881858706474304, "learning_rate": 0.0001, "elbo": 3.54655, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6415083274689186}
{"step": 62100, "loss": 3.5380474090576173, "grad_norm": 0.7049658298492432, "learning_rate": 0.0001, "elbo": 3.53805, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.641801548205489}
{"step": 62105, "loss": 3.542990875244141, "grad_norm": 0.6927238702774048, "learning_rate": 0.0001, "elbo": 3.54299, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6420947689420595}
{"step": 62110, "loss": 3.54439697265625, "grad_norm": 0.6732694506645203, "learning_rate": 0.0001, "elbo": 3.5444, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.64238798967863}
{"step": 62115, "loss": 3.5485908508300783, "grad_norm": 0.7092100381851196, "learning_rate": 0.0001, "elbo": 3.54859, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6426812104152004}
{"step": 62120, "loss": 3.550459289550781, "grad_norm": 0.6548717021942139, "learning_rate": 0.0001, "elbo": 3.55046, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.642974431151771}
{"step": 62125, "loss": 3.5363731384277344, "grad_norm": 0.6953261494636536, "learning_rate": 0.0001, "elbo": 3.53637, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6432676518883413}
{"step": 62130, "loss": 3.5396595001220703, "grad_norm": 0.7456033825874329, "learning_rate": 0.0001, "elbo": 3.53966, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.643560872624912}
{"step": 62135, "loss": 3.528034973144531, "grad_norm": 0.6810417175292969, "learning_rate": 0.0001, "elbo": 3.52804, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6438540933614822}
{"step": 62135, "loss": 3.528034973144531, "grad_norm": 0.6810417175292969, "learning_rate": 0.0001, "elbo": 3.52804, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6438540933614822}
{"step": 62140, "loss": 3.5448684692382812, "grad_norm": 0.7146539688110352, "learning_rate": 0.0001, "elbo": 3.54487, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.644147314098053}
{"step": 62145, "loss": 3.550495910644531, "grad_norm": 0.7153854966163635, "learning_rate": 0.0001, "elbo": 3.5505, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6444405348346236}
{"step": 62150, "loss": 3.5552806854248047, "grad_norm": 0.7512069344520569, "learning_rate": 0.0001, "elbo": 3.55528, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.644733755571194}
{"step": 62155, "loss": 3.5506355285644533, "grad_norm": 0.7458616495132446, "learning_rate": 0.0001, "elbo": 3.55064, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6450269763077645}
{"step": 62160, "loss": 3.538718414306641, "grad_norm": 0.6938252449035645, "learning_rate": 0.0001, "elbo": 3.53872, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.645320197044335}
{"step": 62165, "loss": 3.543282318115234, "grad_norm": 0.7198061347007751, "learning_rate": 0.0001, "elbo": 3.54328, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6456134177809054}
{"step": 62170, "loss": 3.5521976470947267, "grad_norm": 0.836132287979126, "learning_rate": 0.0001, "elbo": 3.5522, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.645906638517476}
{"step": 62175, "loss": 3.5324260711669924, "grad_norm": 0.7145594954490662, "learning_rate": 0.0001, "elbo": 3.53243, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6461998592540463}
{"step": 62180, "loss": 3.5512626647949217, "grad_norm": 0.7799324989318848, "learning_rate": 0.0001, "elbo": 3.55126, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.646493079990617}
{"step": 62185, "loss": 3.5449752807617188, "grad_norm": 0.7915980815887451, "learning_rate": 0.0001, "elbo": 3.54498, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6467863007271877}
{"step": 62190, "loss": 3.5640289306640627, "grad_norm": 0.708416759967804, "learning_rate": 0.0001, "elbo": 3.56403, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.647079521463758}
{"step": 62195, "loss": 3.535753631591797, "grad_norm": 0.7792155146598816, "learning_rate": 0.0001, "elbo": 3.53575, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6473727422003286}
{"step": 62200, "loss": 3.5421661376953124, "grad_norm": 0.8803780674934387, "learning_rate": 0.0001, "elbo": 3.54217, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.647665962936899}
{"step": 62205, "loss": 3.541707992553711, "grad_norm": 0.6857552528381348, "learning_rate": 0.0001, "elbo": 3.54171, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6479591836734695}
{"step": 62210, "loss": 3.528156280517578, "grad_norm": 0.7367101907730103, "learning_rate": 0.0001, "elbo": 3.52816, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.64825240441004}
{"step": 62210, "loss": 3.528156280517578, "grad_norm": 0.7367101907730103, "learning_rate": 0.0001, "elbo": 3.52816, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.64825240441004}
{"step": 62215, "loss": 3.5472232818603517, "grad_norm": 0.7193992137908936, "learning_rate": 0.0001, "elbo": 3.54722, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6485456251466104}
{"step": 62215, "loss": 3.5472232818603517, "grad_norm": 0.7193992137908936, "learning_rate": 0.0001, "elbo": 3.54722, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6485456251466104}
{"step": 62220, "loss": 3.5387317657470705, "grad_norm": 0.6852574348449707, "learning_rate": 0.0001, "elbo": 3.53873, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.648838845883181}
{"step": 62225, "loss": 3.5414527893066405, "grad_norm": 0.686941921710968, "learning_rate": 0.0001, "elbo": 3.54145, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6491320666197513}
{"step": 62230, "loss": 3.5430229187011717, "grad_norm": 0.7040523290634155, "learning_rate": 0.0001, "elbo": 3.54302, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6494252873563218}
{"step": 62235, "loss": 3.5216636657714844, "grad_norm": 0.6819650530815125, "learning_rate": 0.0001, "elbo": 3.52166, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.649718508092892}
{"step": 62240, "loss": 3.5477943420410156, "grad_norm": 0.6930019855499268, "learning_rate": 0.0001, "elbo": 3.54779, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6500117288294627}
{"step": 62245, "loss": 3.546255874633789, "grad_norm": 0.6438373327255249, "learning_rate": 0.0001, "elbo": 3.54626, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.650304949566033}
{"step": 62250, "loss": 3.5310882568359374, "grad_norm": 0.6542989611625671, "learning_rate": 0.0001, "elbo": 3.53109, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6505981703026036}
{"step": 62255, "loss": 3.5504249572753905, "grad_norm": 0.6712542176246643, "learning_rate": 0.0001, "elbo": 3.55043, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.650891391039174}
{"step": 62260, "loss": 3.550833892822266, "grad_norm": 0.638705849647522, "learning_rate": 0.0001, "elbo": 3.55083, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.651184611775745}
{"step": 62265, "loss": 3.528851318359375, "grad_norm": 1.0937507152557373, "learning_rate": 0.0001, "elbo": 3.52885, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6514778325123154}
{"step": 62270, "loss": 3.55062255859375, "grad_norm": 0.7053216695785522, "learning_rate": 0.0001, "elbo": 3.55062, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.651771053248886}
{"step": 62275, "loss": 3.526703643798828, "grad_norm": 0.8713911175727844, "learning_rate": 0.0001, "elbo": 3.5267, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6520642739854563}
{"step": 62280, "loss": 3.542253875732422, "grad_norm": 0.6610298752784729, "learning_rate": 0.0001, "elbo": 3.54225, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6523574947220268}
{"step": 62285, "loss": 3.539484405517578, "grad_norm": 0.6957689523696899, "learning_rate": 0.0001, "elbo": 3.53948, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.652650715458597}
{"step": 62290, "loss": 3.52735595703125, "grad_norm": 0.6927235126495361, "learning_rate": 0.0001, "elbo": 3.52736, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6529439361951677}
{"step": 62295, "loss": 3.551396942138672, "grad_norm": 0.7165480256080627, "learning_rate": 0.0001, "elbo": 3.5514, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.653237156931738}
{"step": 62300, "loss": 3.540850067138672, "grad_norm": 0.7296523451805115, "learning_rate": 0.0001, "elbo": 3.54085, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6535303776683086}
{"step": 62300, "loss": 3.540850067138672, "grad_norm": 0.7296523451805115, "learning_rate": 0.0001, "elbo": 3.54085, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6535303776683086}
{"step": 62305, "loss": 3.5447097778320313, "grad_norm": 0.7651423811912537, "learning_rate": 0.0001, "elbo": 3.54471, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6538235984048795}
{"step": 62310, "loss": 3.530413818359375, "grad_norm": 0.6627573370933533, "learning_rate": 0.0001, "elbo": 3.53041, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.65411681914145}
{"step": 62315, "loss": 3.5322544097900392, "grad_norm": 0.6843674778938293, "learning_rate": 0.0001, "elbo": 3.53225, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6544100398780204}
{"step": 62320, "loss": 3.548358917236328, "grad_norm": 0.8724521398544312, "learning_rate": 0.0001, "elbo": 3.54836, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.654703260614591}
{"step": 62325, "loss": 3.537609100341797, "grad_norm": 0.8778114318847656, "learning_rate": 0.0001, "elbo": 3.53761, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6549964813511613}
{"step": 62330, "loss": 3.551202392578125, "grad_norm": 0.7520251870155334, "learning_rate": 0.0001, "elbo": 3.5512, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6552897020877317}
{"step": 62335, "loss": 3.546503448486328, "grad_norm": 0.7819077968597412, "learning_rate": 0.0001, "elbo": 3.5465, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.655582922824302}
{"step": 62340, "loss": 3.55665283203125, "grad_norm": 0.7935145497322083, "learning_rate": 0.0001, "elbo": 3.55665, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6558761435608726}
{"step": 62345, "loss": 3.539175033569336, "grad_norm": 0.6979513764381409, "learning_rate": 0.0001, "elbo": 3.53917, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.656169364297443}
{"step": 62350, "loss": 3.5493873596191405, "grad_norm": 0.7574179768562317, "learning_rate": 0.0001, "elbo": 3.54939, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6564625850340136}
{"step": 62355, "loss": 3.547491455078125, "grad_norm": 0.7144895195960999, "learning_rate": 0.0001, "elbo": 3.54749, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.656755805770584}
{"step": 62360, "loss": 3.5569339752197267, "grad_norm": 0.6732656359672546, "learning_rate": 0.0001, "elbo": 3.55693, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6570490265071545}
{"step": 62365, "loss": 3.533173751831055, "grad_norm": 0.7242539525032043, "learning_rate": 0.0001, "elbo": 3.53317, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.657342247243725}
{"step": 62370, "loss": 3.5426708221435548, "grad_norm": 0.7238839268684387, "learning_rate": 0.0001, "elbo": 3.54267, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6576354679802954}
{"step": 62375, "loss": 3.5400344848632814, "grad_norm": 0.7106769680976868, "learning_rate": 0.0001, "elbo": 3.54003, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.657928688716866}
{"step": 62380, "loss": 3.529926300048828, "grad_norm": 0.7613103985786438, "learning_rate": 0.0001, "elbo": 3.52993, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6582219094534363}
{"step": 62385, "loss": 3.5548152923583984, "grad_norm": 0.8841870427131653, "learning_rate": 0.0001, "elbo": 3.55482, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.658515130190007}
{"step": 62390, "loss": 3.548771286010742, "grad_norm": 0.8519050478935242, "learning_rate": 0.0001, "elbo": 3.54877, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6588083509265776}
{"step": 62395, "loss": 3.548556900024414, "grad_norm": 0.6867910623550415, "learning_rate": 0.0001, "elbo": 3.54856, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.659101571663148}
{"step": 62400, "loss": 3.539262390136719, "grad_norm": 0.6633703112602234, "learning_rate": 0.0001, "elbo": 3.53926, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6593947923997185}
{"step": 62405, "loss": 3.5298484802246093, "grad_norm": 0.7142220139503479, "learning_rate": 0.0001, "elbo": 3.52985, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.659688013136289}
{"step": 62410, "loss": 3.5452770233154296, "grad_norm": 0.676095724105835, "learning_rate": 0.0001, "elbo": 3.54528, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6599812338728595}
{"step": 62415, "loss": 3.5505802154541017, "grad_norm": 0.7204373478889465, "learning_rate": 0.0001, "elbo": 3.55058, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.66027445460943}
{"step": 62420, "loss": 3.54480094909668, "grad_norm": 0.7044503688812256, "learning_rate": 0.0001, "elbo": 3.5448, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6605676753460004}
{"step": 62425, "loss": 3.54208984375, "grad_norm": 0.7226014733314514, "learning_rate": 0.0001, "elbo": 3.54209, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.660860896082571}
{"step": 62430, "loss": 3.5334442138671873, "grad_norm": 0.7211677432060242, "learning_rate": 0.0001, "elbo": 3.53344, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6611541168191417}
{"step": 62435, "loss": 3.543461227416992, "grad_norm": 0.684969425201416, "learning_rate": 0.0001, "elbo": 3.54346, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.661447337555712}
{"step": 62440, "loss": 3.553801345825195, "grad_norm": 0.9811635613441467, "learning_rate": 0.0001, "elbo": 3.5538, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6617405582922826}
{"step": 62445, "loss": 3.5327835083007812, "grad_norm": 0.6771824955940247, "learning_rate": 0.0001, "elbo": 3.53278, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.662033779028853}
{"step": 62450, "loss": 3.5476917266845702, "grad_norm": 0.7225237488746643, "learning_rate": 0.0001, "elbo": 3.54769, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6623269997654235}
{"step": 62455, "loss": 3.54228515625, "grad_norm": 0.669731616973877, "learning_rate": 0.0001, "elbo": 3.54229, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.662620220501994}
{"step": 62460, "loss": 3.5328121185302734, "grad_norm": 0.6815795302391052, "learning_rate": 0.0001, "elbo": 3.53281, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6629134412385644}
{"step": 62465, "loss": 3.5318931579589843, "grad_norm": 0.6819602847099304, "learning_rate": 0.0001, "elbo": 3.53189, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.663206661975135}
{"step": 62470, "loss": 3.5269886016845704, "grad_norm": 0.8107524514198303, "learning_rate": 0.0001, "elbo": 3.52699, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6634998827117053}
{"step": 62475, "loss": 3.549136734008789, "grad_norm": 0.7083452939987183, "learning_rate": 0.0001, "elbo": 3.54914, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.663793103448276}
{"step": 62480, "loss": 3.533209228515625, "grad_norm": 0.9608038067817688, "learning_rate": 0.0001, "elbo": 3.53321, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6640863241848463}
{"step": 62485, "loss": 3.5393341064453123, "grad_norm": 0.7413594126701355, "learning_rate": 0.0001, "elbo": 3.53933, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6643795449214167}
{"step": 62490, "loss": 3.5465042114257814, "grad_norm": 0.7182730436325073, "learning_rate": 0.0001, "elbo": 3.5465, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.664672765657987}
{"step": 62495, "loss": 3.5499534606933594, "grad_norm": 0.6851015090942383, "learning_rate": 0.0001, "elbo": 3.54995, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6649659863945576}
{"step": 62500, "loss": 3.5491207122802733, "grad_norm": 0.7476056218147278, "learning_rate": 0.0001, "elbo": 3.54912, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.665259207131128}
{"step": 62505, "loss": 3.5307106018066405, "grad_norm": 0.8976298570632935, "learning_rate": 0.0001, "elbo": 3.53071, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6655524278676985}
{"step": 62510, "loss": 3.5385498046875, "grad_norm": 0.6711603999137878, "learning_rate": 0.0001, "elbo": 3.53855, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6658456486042694}
{"step": 62515, "loss": 3.539112854003906, "grad_norm": 0.6347636580467224, "learning_rate": 0.0001, "elbo": 3.53911, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.66613886934084}
{"step": 62520, "loss": 3.543523406982422, "grad_norm": 0.6856376528739929, "learning_rate": 0.0001, "elbo": 3.54352, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6664320900774103}
{"step": 62525, "loss": 3.523664093017578, "grad_norm": 0.8503066897392273, "learning_rate": 0.0001, "elbo": 3.52366, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.666725310813981}
{"step": 62530, "loss": 3.552472686767578, "grad_norm": 0.6651796102523804, "learning_rate": 0.0001, "elbo": 3.55247, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6670185315505512}
{"step": 62535, "loss": 3.5463977813720704, "grad_norm": 0.7726905345916748, "learning_rate": 0.0001, "elbo": 3.5464, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6673117522871217}
{"step": 62540, "loss": 3.5357131958007812, "grad_norm": 0.924530029296875, "learning_rate": 0.0001, "elbo": 3.53571, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.667604973023692}
{"step": 62545, "loss": 3.54981689453125, "grad_norm": 0.7032610774040222, "learning_rate": 0.0001, "elbo": 3.54982, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6678981937602626}
{"step": 62550, "loss": 3.5431449890136717, "grad_norm": 0.7436990141868591, "learning_rate": 0.0001, "elbo": 3.54314, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.668191414496833}
{"step": 62555, "loss": 3.550434875488281, "grad_norm": 0.7056344747543335, "learning_rate": 0.0001, "elbo": 3.55044, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.668484635233404}
{"step": 62560, "loss": 3.541639709472656, "grad_norm": 0.7233642935752869, "learning_rate": 0.0001, "elbo": 3.54164, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6687778559699744}
{"step": 62565, "loss": 3.557303619384766, "grad_norm": 0.9865368604660034, "learning_rate": 0.0001, "elbo": 3.5573, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.669071076706545}
{"step": 62570, "loss": 3.5437198638916017, "grad_norm": 0.7685739398002625, "learning_rate": 0.0001, "elbo": 3.54372, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6693642974431153}
{"step": 62575, "loss": 3.5526973724365236, "grad_norm": 0.6558006405830383, "learning_rate": 0.0001, "elbo": 3.5527, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6696575181796858}
{"step": 62575, "loss": 3.5526973724365236, "grad_norm": 0.6558006405830383, "learning_rate": 0.0001, "elbo": 3.5527, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6696575181796858}
{"step": 62580, "loss": 3.5461448669433593, "grad_norm": 0.7017831206321716, "learning_rate": 0.0001, "elbo": 3.54615, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6699507389162562}
{"step": 62585, "loss": 3.5244537353515626, "grad_norm": 0.6718209981918335, "learning_rate": 0.0001, "elbo": 3.52445, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6702439596528267}
{"step": 62585, "loss": 3.5244537353515626, "grad_norm": 0.6718209981918335, "learning_rate": 0.0001, "elbo": 3.52445, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6702439596528267}
{"step": 62590, "loss": 3.543679046630859, "grad_norm": 0.6647379994392395, "learning_rate": 0.0001, "elbo": 3.54368, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.670537180389397}
{"step": 62595, "loss": 3.5285552978515624, "grad_norm": 0.6897945404052734, "learning_rate": 0.0001, "elbo": 3.52856, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6708304011259676}
{"step": 62600, "loss": 3.555808258056641, "grad_norm": 0.7384170889854431, "learning_rate": 0.0001, "elbo": 3.55581, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.671123621862538}
{"step": 62605, "loss": 3.5377830505371093, "grad_norm": 0.7310494780540466, "learning_rate": 0.0001, "elbo": 3.53778, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6714168425991085}
{"step": 62610, "loss": 3.54931526184082, "grad_norm": 0.7673078179359436, "learning_rate": 0.0001, "elbo": 3.54932, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.671710063335679}
{"step": 62615, "loss": 3.540607452392578, "grad_norm": 0.716329038143158, "learning_rate": 0.0001, "elbo": 3.54061, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6720032840722494}
{"step": 62620, "loss": 3.5199234008789064, "grad_norm": 0.6505017280578613, "learning_rate": 0.0001, "elbo": 3.51992, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.67229650480882}
{"step": 62625, "loss": 3.5295173645019533, "grad_norm": 0.6878196001052856, "learning_rate": 0.0001, "elbo": 3.52952, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6725897255453903}
{"step": 62630, "loss": 3.5276107788085938, "grad_norm": 0.7166459560394287, "learning_rate": 0.0001, "elbo": 3.52761, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.672882946281961}
{"step": 62635, "loss": 3.542222595214844, "grad_norm": 0.7405932545661926, "learning_rate": 0.0001, "elbo": 3.54222, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6731761670185317}
{"step": 62640, "loss": 3.5480300903320314, "grad_norm": 0.6574193239212036, "learning_rate": 0.0001, "elbo": 3.54803, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.673469387755102}
{"step": 62645, "loss": 3.5378372192382814, "grad_norm": 0.7132097482681274, "learning_rate": 0.0001, "elbo": 3.53784, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6737626084916726}
{"step": 62650, "loss": 3.542789077758789, "grad_norm": 0.7041287422180176, "learning_rate": 0.0001, "elbo": 3.54279, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.674055829228243}
{"step": 62655, "loss": 3.532387542724609, "grad_norm": 0.68053138256073, "learning_rate": 0.0001, "elbo": 3.53239, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6743490499648135}
{"step": 62660, "loss": 3.5416393280029297, "grad_norm": 0.6845463514328003, "learning_rate": 0.0001, "elbo": 3.54164, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.674642270701384}
{"step": 62665, "loss": 3.5425086975097657, "grad_norm": 0.6929260492324829, "learning_rate": 0.0001, "elbo": 3.54251, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6749354914379544}
{"step": 62670, "loss": 3.530247116088867, "grad_norm": 0.7231706380844116, "learning_rate": 0.0001, "elbo": 3.53025, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.675228712174525}
{"step": 62675, "loss": 3.525986099243164, "grad_norm": 0.681999683380127, "learning_rate": 0.0001, "elbo": 3.52599, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6755219329110957}
{"step": 62680, "loss": 3.5388256072998048, "grad_norm": 0.7855013608932495, "learning_rate": 0.0001, "elbo": 3.53883, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.675815153647666}
{"step": 62685, "loss": 3.5517047882080077, "grad_norm": 0.741604208946228, "learning_rate": 0.0001, "elbo": 3.5517, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6761083743842367}
{"step": 62690, "loss": 3.5379730224609376, "grad_norm": 0.7051004767417908, "learning_rate": 0.0001, "elbo": 3.53797, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.676401595120807}
{"step": 62695, "loss": 3.5440799713134767, "grad_norm": 0.7175267338752747, "learning_rate": 0.0001, "elbo": 3.54408, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6766948158573776}
{"step": 62700, "loss": 3.5450714111328123, "grad_norm": 0.6904166340827942, "learning_rate": 0.0001, "elbo": 3.54507, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.676988036593948}
{"step": 62705, "loss": 3.5444580078125, "grad_norm": 0.7318413853645325, "learning_rate": 0.0001, "elbo": 3.54446, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6772812573305185}
{"step": 62710, "loss": 3.544580841064453, "grad_norm": 0.8948801159858704, "learning_rate": 0.0001, "elbo": 3.54458, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.677574478067089}
{"step": 62710, "loss": 3.544580841064453, "grad_norm": 0.8948801159858704, "learning_rate": 0.0001, "elbo": 3.54458, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.677574478067089}
{"step": 62715, "loss": 3.53758544921875, "grad_norm": 0.7168539762496948, "learning_rate": 0.0001, "elbo": 3.53759, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6778676988036594}
{"step": 62720, "loss": 3.5498790740966797, "grad_norm": 0.7699257135391235, "learning_rate": 0.0001, "elbo": 3.54988, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.67816091954023}
{"step": 62725, "loss": 3.535875701904297, "grad_norm": 0.7215974926948547, "learning_rate": 0.0001, "elbo": 3.53588, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6784541402768003}
{"step": 62730, "loss": 3.526286315917969, "grad_norm": 0.6834127902984619, "learning_rate": 0.0001, "elbo": 3.52629, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6787473610133707}
{"step": 62735, "loss": 3.540375518798828, "grad_norm": 0.6874310970306396, "learning_rate": 0.0001, "elbo": 3.54038, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.679040581749941}
{"step": 62740, "loss": 3.5428733825683594, "grad_norm": 0.8637378811836243, "learning_rate": 0.0001, "elbo": 3.54287, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6793338024865117}
{"step": 62745, "loss": 3.5378456115722656, "grad_norm": 0.7025232315063477, "learning_rate": 0.0001, "elbo": 3.53785, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.679627023223082}
{"step": 62750, "loss": 3.5458404541015627, "grad_norm": 0.9240885376930237, "learning_rate": 0.0001, "elbo": 3.54584, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6799202439596526}
{"step": 62755, "loss": 3.544111633300781, "grad_norm": 0.6856746673583984, "learning_rate": 0.0001, "elbo": 3.54411, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6802134646962235}
{"step": 62760, "loss": 3.552313232421875, "grad_norm": 0.719031810760498, "learning_rate": 0.0001, "elbo": 3.55231, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.680506685432794}
{"step": 62765, "loss": 3.562489318847656, "grad_norm": 0.6722660660743713, "learning_rate": 0.0001, "elbo": 3.56249, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6807999061693644}
{"step": 62770, "loss": 3.545368957519531, "grad_norm": 0.7011364698410034, "learning_rate": 0.0001, "elbo": 3.54537, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.681093126905935}
{"step": 62770, "loss": 3.545368957519531, "grad_norm": 0.7011364698410034, "learning_rate": 0.0001, "elbo": 3.54537, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.681093126905935}
{"step": 62775, "loss": 3.5423370361328126, "grad_norm": 0.6908385157585144, "learning_rate": 0.0001, "elbo": 3.54234, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6813863476425053}
{"step": 62780, "loss": 3.543285369873047, "grad_norm": 0.8125247359275818, "learning_rate": 0.0001, "elbo": 3.54329, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6816795683790757}
{"step": 62785, "loss": 3.5453304290771483, "grad_norm": 0.7351455688476562, "learning_rate": 0.0001, "elbo": 3.54533, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.681972789115646}
{"step": 62790, "loss": 3.540692138671875, "grad_norm": 0.6951602101325989, "learning_rate": 0.0001, "elbo": 3.54069, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6822660098522166}
{"step": 62795, "loss": 3.538430023193359, "grad_norm": 0.7250333428382874, "learning_rate": 0.0001, "elbo": 3.53843, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.682559230588787}
{"step": 62795, "loss": 3.538430023193359, "grad_norm": 0.7250333428382874, "learning_rate": 0.0001, "elbo": 3.53843, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.682559230588787}
{"step": 62800, "loss": 3.5402153015136717, "grad_norm": 0.6893710494041443, "learning_rate": 0.0001, "elbo": 3.54022, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.682852451325358}
{"step": 62805, "loss": 3.5394363403320312, "grad_norm": 0.6920145750045776, "learning_rate": 0.0001, "elbo": 3.53944, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6831456720619284}
{"step": 62810, "loss": 3.547319793701172, "grad_norm": 0.6837589740753174, "learning_rate": 0.0001, "elbo": 3.54732, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.683438892798499}
{"step": 62815, "loss": 3.534006881713867, "grad_norm": 0.709766685962677, "learning_rate": 0.0001, "elbo": 3.53401, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6837321135350694}
{"step": 62820, "loss": 3.530717468261719, "grad_norm": 0.6860741376876831, "learning_rate": 0.0001, "elbo": 3.53072, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.68402533427164}
{"step": 62825, "loss": 3.546518325805664, "grad_norm": 1.3063405752182007, "learning_rate": 0.0001, "elbo": 3.54652, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6843185550082103}
{"step": 62830, "loss": 3.5317535400390625, "grad_norm": 0.769827127456665, "learning_rate": 0.0001, "elbo": 3.53175, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6846117757447807}
{"step": 62835, "loss": 3.5693374633789063, "grad_norm": 0.6872356534004211, "learning_rate": 0.0001, "elbo": 3.56934, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.684904996481351}
{"step": 62840, "loss": 3.5485328674316405, "grad_norm": 0.7173646092414856, "learning_rate": 0.0001, "elbo": 3.54853, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6851982172179216}
{"step": 62845, "loss": 3.5432430267333985, "grad_norm": 0.7540317177772522, "learning_rate": 0.0001, "elbo": 3.54324, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.685491437954492}
{"step": 62850, "loss": 3.535436248779297, "grad_norm": 0.7081130146980286, "learning_rate": 0.0001, "elbo": 3.53544, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6857846586910625}
{"step": 62855, "loss": 3.550489044189453, "grad_norm": 0.7088128328323364, "learning_rate": 0.0001, "elbo": 3.55049, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.686077879427633}
{"step": 62860, "loss": 3.548212432861328, "grad_norm": 0.7291253805160522, "learning_rate": 0.0001, "elbo": 3.54821, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6863711001642034}
{"step": 62865, "loss": 3.5320762634277343, "grad_norm": 0.7646637558937073, "learning_rate": 0.0001, "elbo": 3.53208, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.686664320900774}
{"step": 62865, "loss": 3.5320762634277343, "grad_norm": 0.7646637558937073, "learning_rate": 0.0001, "elbo": 3.53208, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.686664320900774}
{"step": 62870, "loss": 3.544420623779297, "grad_norm": 0.6979618668556213, "learning_rate": 0.0001, "elbo": 3.54442, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6869575416373443}
{"step": 62875, "loss": 3.5503028869628905, "grad_norm": 0.8529872298240662, "learning_rate": 0.0001, "elbo": 3.5503, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6872507623739152}
{"step": 62880, "loss": 3.5428176879882813, "grad_norm": 0.7119101881980896, "learning_rate": 0.0001, "elbo": 3.54282, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6875439831104857}
{"step": 62885, "loss": 3.5370323181152346, "grad_norm": 0.7278937101364136, "learning_rate": 0.0001, "elbo": 3.53703, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.687837203847056}
{"step": 62890, "loss": 3.554512786865234, "grad_norm": 0.7109202146530151, "learning_rate": 0.0001, "elbo": 3.55451, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6881304245836266}
{"step": 62895, "loss": 3.548898696899414, "grad_norm": 0.8053101897239685, "learning_rate": 0.0001, "elbo": 3.5489, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.688423645320197}
{"step": 62900, "loss": 3.535974884033203, "grad_norm": 0.6836144328117371, "learning_rate": 0.0001, "elbo": 3.53597, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6887168660567675}
{"step": 62905, "loss": 3.5467533111572265, "grad_norm": 0.7333166599273682, "learning_rate": 0.0001, "elbo": 3.54675, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.689010086793338}
{"step": 62910, "loss": 3.5423439025878904, "grad_norm": 0.6882577538490295, "learning_rate": 0.0001, "elbo": 3.54234, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6893033075299084}
{"step": 62915, "loss": 3.546918487548828, "grad_norm": 1.3859477043151855, "learning_rate": 0.0001, "elbo": 3.54692, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.689596528266479}
{"step": 62920, "loss": 3.530598449707031, "grad_norm": 0.7963322401046753, "learning_rate": 0.0001, "elbo": 3.5306, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6898897490030493}
{"step": 62925, "loss": 3.535131072998047, "grad_norm": 0.6937516927719116, "learning_rate": 0.0001, "elbo": 3.53513, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6901829697396202}
{"step": 62930, "loss": 3.535715866088867, "grad_norm": 0.7235760688781738, "learning_rate": 0.0001, "elbo": 3.53572, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6904761904761907}
{"step": 62935, "loss": 3.532323455810547, "grad_norm": 0.6408975124359131, "learning_rate": 0.0001, "elbo": 3.53232, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.690769411212761}
{"step": 62940, "loss": 3.5333187103271486, "grad_norm": 0.6658571362495422, "learning_rate": 0.0001, "elbo": 3.53332, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6910626319493316}
{"step": 62945, "loss": 3.539260482788086, "grad_norm": 0.6992797255516052, "learning_rate": 0.0001, "elbo": 3.53926, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.691355852685902}
{"step": 62950, "loss": 3.539080810546875, "grad_norm": 0.6806314587593079, "learning_rate": 0.0001, "elbo": 3.53908, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6916490734224725}
{"step": 62955, "loss": 3.533827209472656, "grad_norm": 0.8834949135780334, "learning_rate": 0.0001, "elbo": 3.53383, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.691942294159043}
{"step": 62960, "loss": 3.5402320861816405, "grad_norm": 0.9090861082077026, "learning_rate": 0.0001, "elbo": 3.54023, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6922355148956134}
{"step": 62965, "loss": 3.544721984863281, "grad_norm": 0.6904098987579346, "learning_rate": 0.0001, "elbo": 3.54472, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.692528735632184}
{"step": 62970, "loss": 3.5462310791015623, "grad_norm": 0.7293350696563721, "learning_rate": 0.0001, "elbo": 3.54623, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6928219563687543}
{"step": 62975, "loss": 3.5408363342285156, "grad_norm": 0.7203574180603027, "learning_rate": 0.0001, "elbo": 3.54084, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6931151771053248}
{"step": 62980, "loss": 3.5533206939697264, "grad_norm": 0.7193601131439209, "learning_rate": 0.0001, "elbo": 3.55332, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6934083978418952}
{"step": 62985, "loss": 3.555916595458984, "grad_norm": 0.7714620232582092, "learning_rate": 0.0001, "elbo": 3.55592, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6937016185784657}
{"step": 62990, "loss": 3.537511444091797, "grad_norm": 0.6976107954978943, "learning_rate": 0.0001, "elbo": 3.53751, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.693994839315036}
{"step": 62995, "loss": 3.524424743652344, "grad_norm": 0.7006282806396484, "learning_rate": 0.0001, "elbo": 3.52442, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6942880600516066}
{"step": 63000, "loss": 3.5511985778808595, "grad_norm": 0.7222654223442078, "learning_rate": 0.0001, "elbo": 3.5512, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6945812807881775}
{"step": 63005, "loss": 3.5401187896728517, "grad_norm": 0.6978408694267273, "learning_rate": 0.0001, "elbo": 3.54012, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.694874501524748}
{"step": 63010, "loss": 3.5386821746826174, "grad_norm": 0.6767491102218628, "learning_rate": 0.0001, "elbo": 3.53868, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6951677222613184}
{"step": 63015, "loss": 3.5340621948242186, "grad_norm": 0.7630857825279236, "learning_rate": 0.0001, "elbo": 3.53406, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.695460942997889}
{"step": 63020, "loss": 3.5358158111572267, "grad_norm": 0.6692056059837341, "learning_rate": 0.0001, "elbo": 3.53582, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6957541637344593}
{"step": 63025, "loss": 3.5245574951171874, "grad_norm": 0.6741533279418945, "learning_rate": 0.0001, "elbo": 3.52456, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6960473844710298}
{"step": 63030, "loss": 3.552902603149414, "grad_norm": 0.6931955218315125, "learning_rate": 0.0001, "elbo": 3.5529, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6963406052076}
{"step": 63035, "loss": 3.550928497314453, "grad_norm": 0.7743481993675232, "learning_rate": 0.0001, "elbo": 3.55093, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6966338259441707}
{"step": 63040, "loss": 3.545946502685547, "grad_norm": 0.6978697776794434, "learning_rate": 0.0001, "elbo": 3.54595, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.696927046680741}
{"step": 63045, "loss": 3.5265342712402346, "grad_norm": 0.701562225818634, "learning_rate": 0.0001, "elbo": 3.52653, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.697220267417312}
{"step": 63045, "loss": 3.5265342712402346, "grad_norm": 0.701562225818634, "learning_rate": 0.0001, "elbo": 3.52653, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.697220267417312}
{"step": 63050, "loss": 3.5254878997802734, "grad_norm": 0.9725357890129089, "learning_rate": 0.0001, "elbo": 3.52549, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6975134881538825}
{"step": 63055, "loss": 3.536014938354492, "grad_norm": 0.7349487543106079, "learning_rate": 0.0001, "elbo": 3.53601, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.697806708890453}
{"step": 63060, "loss": 3.5452362060546876, "grad_norm": 0.7706467509269714, "learning_rate": 0.0001, "elbo": 3.54524, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6980999296270234}
{"step": 63065, "loss": 3.5341697692871095, "grad_norm": 0.6497294902801514, "learning_rate": 0.0001, "elbo": 3.53417, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.698393150363594}
{"step": 63070, "loss": 3.543927764892578, "grad_norm": 0.744202196598053, "learning_rate": 0.0001, "elbo": 3.54393, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6986863711001643}
{"step": 63075, "loss": 3.5481971740722655, "grad_norm": 0.6772361993789673, "learning_rate": 0.0001, "elbo": 3.5482, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6989795918367347}
{"step": 63080, "loss": 3.5399497985839843, "grad_norm": 0.7670198082923889, "learning_rate": 0.0001, "elbo": 3.53995, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.699272812573305}
{"step": 63085, "loss": 3.5443355560302736, "grad_norm": 0.7021400928497314, "learning_rate": 0.0001, "elbo": 3.54434, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.6995660333098757}
{"step": 63090, "loss": 3.537432861328125, "grad_norm": 0.7944764494895935, "learning_rate": 0.0001, "elbo": 3.53743, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.699859254046446}
{"step": 63095, "loss": 3.549479675292969, "grad_norm": 0.7018036842346191, "learning_rate": 0.0001, "elbo": 3.54948, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7001524747830166}
{"step": 63100, "loss": 3.5432037353515624, "grad_norm": 0.7921850085258484, "learning_rate": 0.0001, "elbo": 3.5432, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.700445695519587}
{"step": 63105, "loss": 3.534406280517578, "grad_norm": 0.7074124217033386, "learning_rate": 0.0001, "elbo": 3.53441, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7007389162561575}
{"step": 63110, "loss": 3.5303478240966797, "grad_norm": 0.6965275406837463, "learning_rate": 0.0001, "elbo": 3.53035, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.701032136992728}
{"step": 63115, "loss": 3.542192077636719, "grad_norm": 0.6630905866622925, "learning_rate": 0.0001, "elbo": 3.54219, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7013253577292984}
{"step": 63120, "loss": 3.5447296142578124, "grad_norm": 0.7800570130348206, "learning_rate": 0.0001, "elbo": 3.54473, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.701618578465869}
{"step": 63125, "loss": 3.532012939453125, "grad_norm": 0.7052522897720337, "learning_rate": 0.0001, "elbo": 3.53201, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7019117992024397}
{"step": 63130, "loss": 3.5515636444091796, "grad_norm": 0.7256901860237122, "learning_rate": 0.0001, "elbo": 3.55156, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.70220501993901}
{"step": 63135, "loss": 3.5423728942871096, "grad_norm": 0.6971667408943176, "learning_rate": 0.0001, "elbo": 3.54237, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7024982406755806}
{"step": 63140, "loss": 3.5601375579833983, "grad_norm": 0.6634718775749207, "learning_rate": 0.0001, "elbo": 3.56014, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.702791461412151}
{"step": 63145, "loss": 3.5344924926757812, "grad_norm": 0.6967130303382874, "learning_rate": 0.0001, "elbo": 3.53449, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7030846821487216}
{"step": 63145, "loss": 3.5344924926757812, "grad_norm": 0.6967130303382874, "learning_rate": 0.0001, "elbo": 3.53449, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7030846821487216}
{"step": 63150, "loss": 3.5473838806152345, "grad_norm": 0.8096219301223755, "learning_rate": 0.0001, "elbo": 3.54738, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.703377902885292}
{"step": 63155, "loss": 3.5503536224365235, "grad_norm": 0.6888339519500732, "learning_rate": 0.0001, "elbo": 3.55035, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7036711236218625}
{"step": 63160, "loss": 3.5438488006591795, "grad_norm": 0.8091912865638733, "learning_rate": 0.0001, "elbo": 3.54385, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.703964344358433}
{"step": 63165, "loss": 3.546255111694336, "grad_norm": 0.7617764472961426, "learning_rate": 0.0001, "elbo": 3.54626, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7042575650950034}
{"step": 63170, "loss": 3.5434619903564455, "grad_norm": 0.7742331624031067, "learning_rate": 0.0001, "elbo": 3.54346, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7045507858315743}
{"step": 63175, "loss": 3.557937240600586, "grad_norm": 0.7069260478019714, "learning_rate": 0.0001, "elbo": 3.55794, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7048440065681447}
{"step": 63180, "loss": 3.5586711883544924, "grad_norm": 0.723548173904419, "learning_rate": 0.0001, "elbo": 3.55867, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.705137227304715}
{"step": 63185, "loss": 3.5401676177978514, "grad_norm": 0.717019259929657, "learning_rate": 0.0001, "elbo": 3.54017, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7054304480412856}
{"step": 63190, "loss": 3.553603744506836, "grad_norm": 0.7253292202949524, "learning_rate": 0.0001, "elbo": 3.5536, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.705723668777856}
{"step": 63195, "loss": 3.547496795654297, "grad_norm": 0.8108681440353394, "learning_rate": 0.0001, "elbo": 3.5475, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7060168895144265}
{"step": 63200, "loss": 3.529928970336914, "grad_norm": 0.7596390843391418, "learning_rate": 0.0001, "elbo": 3.52993, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.706310110250997}
{"step": 63205, "loss": 3.537459945678711, "grad_norm": 0.7864209413528442, "learning_rate": 0.0001, "elbo": 3.53746, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7066033309875674}
{"step": 63210, "loss": 3.522917938232422, "grad_norm": 0.8418396711349487, "learning_rate": 0.0001, "elbo": 3.52292, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.706896551724138}
{"step": 63215, "loss": 3.5195243835449217, "grad_norm": 0.6799673438072205, "learning_rate": 0.0001, "elbo": 3.51952, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7071897724607084}
{"step": 63220, "loss": 3.5467197418212892, "grad_norm": 0.7487685680389404, "learning_rate": 0.0001, "elbo": 3.54672, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.707482993197279}
{"step": 63225, "loss": 3.542303466796875, "grad_norm": 0.743789792060852, "learning_rate": 0.0001, "elbo": 3.5423, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7077762139338493}
{"step": 63230, "loss": 3.5141441345214846, "grad_norm": 0.7064143419265747, "learning_rate": 0.0001, "elbo": 3.51414, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7080694346704197}
{"step": 63235, "loss": 3.5441490173339845, "grad_norm": 0.729095458984375, "learning_rate": 0.0001, "elbo": 3.54415, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.70836265540699}
{"step": 63240, "loss": 3.552857208251953, "grad_norm": 0.6875685453414917, "learning_rate": 0.0001, "elbo": 3.55286, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7086558761435606}
{"step": 63245, "loss": 3.543268585205078, "grad_norm": 0.7034415006637573, "learning_rate": 0.0001, "elbo": 3.54327, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7089490968801315}
{"step": 63250, "loss": 3.5400821685791017, "grad_norm": 1.0198915004730225, "learning_rate": 0.0001, "elbo": 3.54008, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.709242317616702}
{"step": 63255, "loss": 3.529161071777344, "grad_norm": 0.7150043845176697, "learning_rate": 0.0001, "elbo": 3.52916, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7095355383532724}
{"step": 63260, "loss": 3.5295921325683595, "grad_norm": 0.7354272603988647, "learning_rate": 0.0001, "elbo": 3.52959, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.709828759089843}
{"step": 63265, "loss": 3.549774932861328, "grad_norm": 0.6749369502067566, "learning_rate": 0.0001, "elbo": 3.54977, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7101219798264133}
{"step": 63270, "loss": 3.549208068847656, "grad_norm": 0.8006024956703186, "learning_rate": 0.0001, "elbo": 3.54921, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.710415200562984}
{"step": 63275, "loss": 3.526800537109375, "grad_norm": 0.6706823110580444, "learning_rate": 0.0001, "elbo": 3.5268, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7107084212995542}
{"step": 63280, "loss": 3.5444255828857423, "grad_norm": 0.7194271087646484, "learning_rate": 0.0001, "elbo": 3.54443, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7110016420361247}
{"step": 63285, "loss": 3.5431941986083983, "grad_norm": 0.6888388395309448, "learning_rate": 0.0001, "elbo": 3.54319, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.711294862772695}
{"step": 63290, "loss": 3.5520561218261717, "grad_norm": 0.6744880676269531, "learning_rate": 0.0001, "elbo": 3.55206, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7115880835092656}
{"step": 63295, "loss": 3.543762969970703, "grad_norm": 0.7140408158302307, "learning_rate": 0.0001, "elbo": 3.54376, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7118813042458365}
{"step": 63300, "loss": 3.5375537872314453, "grad_norm": 0.7006812691688538, "learning_rate": 0.0001, "elbo": 3.53755, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.712174524982407}
{"step": 63305, "loss": 3.5436019897460938, "grad_norm": 0.7401311993598938, "learning_rate": 0.0001, "elbo": 3.5436, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7124677457189774}
{"step": 63305, "loss": 3.5436019897460938, "grad_norm": 0.7401311993598938, "learning_rate": 0.0001, "elbo": 3.5436, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7124677457189774}
{"step": 63310, "loss": 3.5420387268066404, "grad_norm": 0.7901619672775269, "learning_rate": 0.0001, "elbo": 3.54204, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.712760966455548}
{"step": 63315, "loss": 3.5389957427978516, "grad_norm": 0.7138994336128235, "learning_rate": 0.0001, "elbo": 3.539, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7130541871921183}
{"step": 63320, "loss": 3.5360092163085937, "grad_norm": 0.7561114430427551, "learning_rate": 0.0001, "elbo": 3.53601, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.713347407928689}
{"step": 63325, "loss": 3.559015655517578, "grad_norm": 0.7766606211662292, "learning_rate": 0.0001, "elbo": 3.55902, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7136406286652592}
{"step": 63330, "loss": 3.5508045196533202, "grad_norm": 0.7280809879302979, "learning_rate": 0.0001, "elbo": 3.5508, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7139338494018297}
{"step": 63335, "loss": 3.551380157470703, "grad_norm": 0.8729190826416016, "learning_rate": 0.0001, "elbo": 3.55138, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7142270701384}
{"step": 63340, "loss": 3.537244415283203, "grad_norm": 0.6944563388824463, "learning_rate": 0.0001, "elbo": 3.53724, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7145202908749706}
{"step": 63345, "loss": 3.534906005859375, "grad_norm": 0.7790590524673462, "learning_rate": 0.0001, "elbo": 3.53491, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.714813511611541}
{"step": 63350, "loss": 3.549762725830078, "grad_norm": 0.7668648362159729, "learning_rate": 0.0001, "elbo": 3.54976, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7151067323481115}
{"step": 63355, "loss": 3.5461483001708984, "grad_norm": 0.9798856377601624, "learning_rate": 0.0001, "elbo": 3.54615, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.715399953084682}
{"step": 63360, "loss": 3.5401260375976564, "grad_norm": 0.7087020874023438, "learning_rate": 0.0001, "elbo": 3.54013, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7156931738212524}
{"step": 63365, "loss": 3.552568817138672, "grad_norm": 0.7273761034011841, "learning_rate": 0.0001, "elbo": 3.55257, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.715986394557823}
{"step": 63370, "loss": 3.5385814666748048, "grad_norm": 0.7783149480819702, "learning_rate": 0.0001, "elbo": 3.53858, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7162796152943938}
{"step": 63375, "loss": 3.5283851623535156, "grad_norm": 0.7311549782752991, "learning_rate": 0.0001, "elbo": 3.52838, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.716572836030964}
{"step": 63380, "loss": 3.5334720611572266, "grad_norm": 0.6622556447982788, "learning_rate": 0.0001, "elbo": 3.53347, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7168660567675347}
{"step": 63385, "loss": 3.5348876953125, "grad_norm": 0.691363513469696, "learning_rate": 0.0001, "elbo": 3.53489, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.717159277504105}
{"step": 63390, "loss": 3.546091842651367, "grad_norm": 0.9604025483131409, "learning_rate": 0.0001, "elbo": 3.54609, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7174524982406756}
{"step": 63395, "loss": 3.539427947998047, "grad_norm": 0.6919891238212585, "learning_rate": 0.0001, "elbo": 3.53943, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.717745718977246}
{"step": 63400, "loss": 3.5447113037109377, "grad_norm": 0.6767886877059937, "learning_rate": 0.0001, "elbo": 3.54471, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7180389397138165}
{"step": 63405, "loss": 3.5376941680908205, "grad_norm": 0.7149869203567505, "learning_rate": 0.0001, "elbo": 3.53769, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.718332160450387}
{"step": 63410, "loss": 3.5315505981445314, "grad_norm": 0.6952175498008728, "learning_rate": 0.0001, "elbo": 3.53155, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7186253811869574}
{"step": 63415, "loss": 3.5407379150390623, "grad_norm": 0.6850206851959229, "learning_rate": 0.0001, "elbo": 3.54074, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7189186019235283}
{"step": 63420, "loss": 3.5523033142089844, "grad_norm": 0.7117623090744019, "learning_rate": 0.0001, "elbo": 3.5523, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7192118226600988}
{"step": 63425, "loss": 3.5351146697998046, "grad_norm": 0.6831132173538208, "learning_rate": 0.0001, "elbo": 3.53511, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.719505043396669}
{"step": 63430, "loss": 3.527743911743164, "grad_norm": 0.67576664686203, "learning_rate": 0.0001, "elbo": 3.52774, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7197982641332397}
{"step": 63435, "loss": 3.521913528442383, "grad_norm": 0.7775284647941589, "learning_rate": 0.0001, "elbo": 3.52191, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.72009148486981}
{"step": 63440, "loss": 3.5524166107177733, "grad_norm": 0.7181352376937866, "learning_rate": 0.0001, "elbo": 3.55242, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7203847056063806}
{"step": 63445, "loss": 3.5322006225585936, "grad_norm": 0.684815526008606, "learning_rate": 0.0001, "elbo": 3.5322, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.720677926342951}
{"step": 63450, "loss": 3.5410530090332033, "grad_norm": 0.8478922247886658, "learning_rate": 0.0001, "elbo": 3.54105, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7209711470795215}
{"step": 63455, "loss": 3.5426563262939452, "grad_norm": 0.6937932968139648, "learning_rate": 0.0001, "elbo": 3.54266, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.721264367816092}
{"step": 63460, "loss": 3.5300048828125, "grad_norm": 0.6781395077705383, "learning_rate": 0.0001, "elbo": 3.53, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7215575885526624}
{"step": 63465, "loss": 3.5420791625976564, "grad_norm": 0.7759298086166382, "learning_rate": 0.0001, "elbo": 3.54208, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.721850809289233}
{"step": 63470, "loss": 3.553579330444336, "grad_norm": 0.679162323474884, "learning_rate": 0.0001, "elbo": 3.55358, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7221440300258033}
{"step": 63475, "loss": 3.5532108306884767, "grad_norm": 0.6834045052528381, "learning_rate": 0.0001, "elbo": 3.55321, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7224372507623738}
{"step": 63480, "loss": 3.5200241088867186, "grad_norm": 0.6939464807510376, "learning_rate": 0.0001, "elbo": 3.52002, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.722730471498944}
{"step": 63485, "loss": 3.5451519012451174, "grad_norm": 0.7057629227638245, "learning_rate": 0.0001, "elbo": 3.54515, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7230236922355147}
{"step": 63490, "loss": 3.5436935424804688, "grad_norm": 0.687250554561615, "learning_rate": 0.0001, "elbo": 3.54369, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.723316912972085}
{"step": 63495, "loss": 3.5360336303710938, "grad_norm": 0.6713403463363647, "learning_rate": 0.0001, "elbo": 3.53603, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.723610133708656}
{"step": 63495, "loss": 3.5360336303710938, "grad_norm": 0.6713403463363647, "learning_rate": 0.0001, "elbo": 3.53603, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.723610133708656}
{"step": 63500, "loss": 3.5434165954589845, "grad_norm": 0.6613478660583496, "learning_rate": 0.0001, "elbo": 3.54342, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7239033544452265}
{"step": 63505, "loss": 3.533023452758789, "grad_norm": 0.7284743189811707, "learning_rate": 0.0001, "elbo": 3.53302, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.724196575181797}
{"step": 63510, "loss": 3.5406490325927735, "grad_norm": 0.7429021596908569, "learning_rate": 0.0001, "elbo": 3.54065, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7244897959183674}
{"step": 63515, "loss": 3.5402530670166015, "grad_norm": 0.7263238430023193, "learning_rate": 0.0001, "elbo": 3.54025, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.724783016654938}
{"step": 63520, "loss": 3.552349853515625, "grad_norm": 0.8115300536155701, "learning_rate": 0.0001, "elbo": 3.55235, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7250762373915083}
{"step": 63525, "loss": 3.548097610473633, "grad_norm": 0.6894472241401672, "learning_rate": 0.0001, "elbo": 3.5481, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7253694581280787}
{"step": 63530, "loss": 3.547925567626953, "grad_norm": 0.6891664266586304, "learning_rate": 0.0001, "elbo": 3.54793, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.725662678864649}
{"step": 63535, "loss": 3.539389801025391, "grad_norm": 0.6779371500015259, "learning_rate": 0.0001, "elbo": 3.53939, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7259558996012196}
{"step": 63540, "loss": 3.5495338439941406, "grad_norm": 0.6643427610397339, "learning_rate": 0.0001, "elbo": 3.54953, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7262491203377905}
{"step": 63545, "loss": 3.532939147949219, "grad_norm": 0.6940811276435852, "learning_rate": 0.0001, "elbo": 3.53294, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.726542341074361}
{"step": 63550, "loss": 3.5451278686523438, "grad_norm": 0.7126700282096863, "learning_rate": 0.0001, "elbo": 3.54513, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7268355618109315}
{"step": 63555, "loss": 3.533040237426758, "grad_norm": 0.696252703666687, "learning_rate": 0.0001, "elbo": 3.53304, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.727128782547502}
{"step": 63560, "loss": 3.54290771484375, "grad_norm": 0.7207469344139099, "learning_rate": 0.0001, "elbo": 3.54291, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7274220032840724}
{"step": 63565, "loss": 3.550947570800781, "grad_norm": 0.7437089085578918, "learning_rate": 0.0001, "elbo": 3.55095, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.727715224020643}
{"step": 63570, "loss": 3.5486099243164064, "grad_norm": 0.7136542201042175, "learning_rate": 0.0001, "elbo": 3.54861, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7280084447572133}
{"step": 63575, "loss": 3.5221248626708985, "grad_norm": 0.7381241917610168, "learning_rate": 0.0001, "elbo": 3.52212, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7283016654937837}
{"step": 63580, "loss": 3.541728210449219, "grad_norm": 0.654549777507782, "learning_rate": 0.0001, "elbo": 3.54173, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.728594886230354}
{"step": 63580, "loss": 3.541728210449219, "grad_norm": 0.654549777507782, "learning_rate": 0.0001, "elbo": 3.54173, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.728594886230354}
{"step": 63585, "loss": 3.527301788330078, "grad_norm": 0.7032653093338013, "learning_rate": 0.0001, "elbo": 3.5273, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7288881069669246}
{"step": 63590, "loss": 3.5277076721191407, "grad_norm": 0.7119872570037842, "learning_rate": 0.0001, "elbo": 3.52771, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.729181327703495}
{"step": 63595, "loss": 3.5340782165527345, "grad_norm": 0.7669461965560913, "learning_rate": 0.0001, "elbo": 3.53408, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7294745484400655}
{"step": 63600, "loss": 3.534864807128906, "grad_norm": 0.7670844793319702, "learning_rate": 0.0001, "elbo": 3.53486, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.729767769176636}
{"step": 63605, "loss": 3.5420692443847654, "grad_norm": 0.6718153357505798, "learning_rate": 0.0001, "elbo": 3.54207, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7300609899132064}
{"step": 63610, "loss": 3.529451370239258, "grad_norm": 0.6872665286064148, "learning_rate": 0.0001, "elbo": 3.52945, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.730354210649777}
{"step": 63615, "loss": 3.5444263458251952, "grad_norm": 0.7012892365455627, "learning_rate": 0.0001, "elbo": 3.54443, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.730647431386348}
{"step": 63620, "loss": 3.5411510467529297, "grad_norm": 0.6868194341659546, "learning_rate": 0.0001, "elbo": 3.54115, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7309406521229183}
{"step": 63625, "loss": 3.5267486572265625, "grad_norm": 0.7205625772476196, "learning_rate": 0.0001, "elbo": 3.52675, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7312338728594887}
{"step": 63630, "loss": 3.530013656616211, "grad_norm": 0.6842942237854004, "learning_rate": 0.0001, "elbo": 3.53001, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.731527093596059}
{"step": 63635, "loss": 3.5411964416503907, "grad_norm": 0.6850343346595764, "learning_rate": 0.0001, "elbo": 3.5412, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7318203143326296}
{"step": 63640, "loss": 3.552654266357422, "grad_norm": 0.6999786496162415, "learning_rate": 0.0001, "elbo": 3.55265, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7321135350692}
{"step": 63645, "loss": 3.5465671539306642, "grad_norm": 0.680246114730835, "learning_rate": 0.0001, "elbo": 3.54657, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7324067558057705}
{"step": 63650, "loss": 3.5443321228027345, "grad_norm": 0.68453449010849, "learning_rate": 0.0001, "elbo": 3.54433, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.732699976542341}
{"step": 63655, "loss": 3.5333526611328123, "grad_norm": 0.8633993268013, "learning_rate": 0.0001, "elbo": 3.53335, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7329931972789114}
{"step": 63660, "loss": 3.55780143737793, "grad_norm": 1.364089012145996, "learning_rate": 0.0001, "elbo": 3.5578, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.733286418015482}
{"step": 63665, "loss": 3.5329071044921876, "grad_norm": 0.7164285182952881, "learning_rate": 0.0001, "elbo": 3.53291, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.733579638752053}
{"step": 63670, "loss": 3.5456092834472654, "grad_norm": 0.7074563503265381, "learning_rate": 0.0001, "elbo": 3.54561, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7338728594886232}
{"step": 63675, "loss": 3.5398704528808596, "grad_norm": 0.6884918212890625, "learning_rate": 0.0001, "elbo": 3.53987, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7341660802251937}
{"step": 63680, "loss": 3.5315925598144533, "grad_norm": 0.6781802177429199, "learning_rate": 0.0001, "elbo": 3.53159, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.734459300961764}
{"step": 63685, "loss": 3.537861633300781, "grad_norm": 0.7534199357032776, "learning_rate": 0.0001, "elbo": 3.53786, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7347525216983346}
{"step": 63690, "loss": 3.5457916259765625, "grad_norm": 0.8532191514968872, "learning_rate": 0.0001, "elbo": 3.54579, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.735045742434905}
{"step": 63695, "loss": 3.5506324768066406, "grad_norm": 0.6821960806846619, "learning_rate": 0.0001, "elbo": 3.55063, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7353389631714755}
{"step": 63700, "loss": 3.5501174926757812, "grad_norm": 0.6700047850608826, "learning_rate": 0.0001, "elbo": 3.55012, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.735632183908046}
{"step": 63705, "loss": 3.5428325653076174, "grad_norm": 0.7101459503173828, "learning_rate": 0.0001, "elbo": 3.54283, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7359254046446164}
{"step": 63710, "loss": 3.5453212738037108, "grad_norm": 0.7284455895423889, "learning_rate": 0.0001, "elbo": 3.54532, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.736218625381187}
{"step": 63715, "loss": 3.521167755126953, "grad_norm": 0.9883810877799988, "learning_rate": 0.0001, "elbo": 3.52117, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7365118461177573}
{"step": 63720, "loss": 3.548157501220703, "grad_norm": 0.6829943656921387, "learning_rate": 0.0001, "elbo": 3.54816, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.736805066854328}
{"step": 63725, "loss": 3.537982940673828, "grad_norm": 0.7246162295341492, "learning_rate": 0.0001, "elbo": 3.53798, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7370982875908982}
{"step": 63730, "loss": 3.543499755859375, "grad_norm": 0.7208631038665771, "learning_rate": 0.0001, "elbo": 3.5435, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7373915083274687}
{"step": 63735, "loss": 3.5314407348632812, "grad_norm": 0.6659636497497559, "learning_rate": 0.0001, "elbo": 3.53144, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.737684729064039}
{"step": 63740, "loss": 3.5216079711914063, "grad_norm": 0.6656898856163025, "learning_rate": 0.0001, "elbo": 3.52161, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.73797794980061}
{"step": 63745, "loss": 3.539520263671875, "grad_norm": 0.6996064782142639, "learning_rate": 0.0001, "elbo": 3.53952, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7382711705371805}
{"step": 63750, "loss": 3.551708984375, "grad_norm": 0.6764928102493286, "learning_rate": 0.0001, "elbo": 3.55171, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.738564391273751}
{"step": 63755, "loss": 3.527325439453125, "grad_norm": 0.7680856585502625, "learning_rate": 0.0001, "elbo": 3.52733, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7388576120103214}
{"step": 63760, "loss": 3.5213905334472657, "grad_norm": 0.6924045085906982, "learning_rate": 0.0001, "elbo": 3.52139, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.739150832746892}
{"step": 63765, "loss": 3.5238800048828125, "grad_norm": 0.7014198899269104, "learning_rate": 0.0001, "elbo": 3.52388, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7394440534834623}
{"step": 63770, "loss": 3.5432449340820313, "grad_norm": 1.0483882427215576, "learning_rate": 0.0001, "elbo": 3.54325, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7397372742200328}
{"step": 63770, "loss": 3.5432449340820313, "grad_norm": 1.0483882427215576, "learning_rate": 0.0001, "elbo": 3.54325, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7397372742200328}
{"step": 63775, "loss": 3.530875015258789, "grad_norm": 0.7653962969779968, "learning_rate": 0.0001, "elbo": 3.53088, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7400304949566032}
{"step": 63780, "loss": 3.5358341217041014, "grad_norm": 0.6815351843833923, "learning_rate": 0.0001, "elbo": 3.53583, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7403237156931737}
{"step": 63785, "loss": 3.523748016357422, "grad_norm": 0.6664395928382874, "learning_rate": 0.0001, "elbo": 3.52375, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7406169364297446}
{"step": 63790, "loss": 3.541058349609375, "grad_norm": 0.7061586380004883, "learning_rate": 0.0001, "elbo": 3.54106, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.740910157166315}
{"step": 63795, "loss": 3.510183334350586, "grad_norm": 0.877106249332428, "learning_rate": 0.0001, "elbo": 3.51018, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7412033779028855}
{"step": 63795, "loss": 3.510183334350586, "grad_norm": 0.877106249332428, "learning_rate": 0.0001, "elbo": 3.51018, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7412033779028855}
{"step": 63800, "loss": 3.53533935546875, "grad_norm": 0.7026598453521729, "learning_rate": 0.0001, "elbo": 3.53534, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.741496598639456}
{"step": 63805, "loss": 3.5236892700195312, "grad_norm": 0.6951350569725037, "learning_rate": 0.0001, "elbo": 3.52369, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7417898193760264}
{"step": 63810, "loss": 3.515637969970703, "grad_norm": 0.7074527740478516, "learning_rate": 0.0001, "elbo": 3.51564, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.742083040112597}
{"step": 63810, "loss": 3.515637969970703, "grad_norm": 0.7074527740478516, "learning_rate": 0.0001, "elbo": 3.51564, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.742083040112597}
{"step": 63815, "loss": 3.542802429199219, "grad_norm": 0.7273288369178772, "learning_rate": 0.0001, "elbo": 3.5428, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7423762608491673}
{"step": 63820, "loss": 3.525710678100586, "grad_norm": 0.7678694128990173, "learning_rate": 0.0001, "elbo": 3.52571, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7426694815857378}
{"step": 63825, "loss": 3.551471710205078, "grad_norm": 0.6933841705322266, "learning_rate": 0.0001, "elbo": 3.55147, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.742962702322308}
{"step": 63830, "loss": 3.5569061279296874, "grad_norm": 0.7166821360588074, "learning_rate": 0.0001, "elbo": 3.55691, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7432559230588787}
{"step": 63835, "loss": 3.538285827636719, "grad_norm": 0.7231382131576538, "learning_rate": 0.0001, "elbo": 3.53829, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.743549143795449}
{"step": 63840, "loss": 3.531280517578125, "grad_norm": 0.6889990568161011, "learning_rate": 0.0001, "elbo": 3.53128, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7438423645320196}
{"step": 63845, "loss": 3.5255531311035155, "grad_norm": 0.7061555981636047, "learning_rate": 0.0001, "elbo": 3.52555, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.74413558526859}
{"step": 63850, "loss": 3.526296615600586, "grad_norm": 0.6887428164482117, "learning_rate": 0.0001, "elbo": 3.5263, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7444288060051605}
{"step": 63850, "loss": 3.526296615600586, "grad_norm": 0.6887428164482117, "learning_rate": 0.0001, "elbo": 3.5263, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7444288060051605}
{"step": 63855, "loss": 3.5434303283691406, "grad_norm": 0.7394241690635681, "learning_rate": 0.0001, "elbo": 3.54343, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.744722026741731}
{"step": 63860, "loss": 3.53851432800293, "grad_norm": 0.7763609886169434, "learning_rate": 0.0001, "elbo": 3.53851, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7450152474783014}
{"step": 63865, "loss": 3.5360111236572265, "grad_norm": 0.6979455351829529, "learning_rate": 0.0001, "elbo": 3.53601, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7453084682148723}
{"step": 63870, "loss": 3.5390071868896484, "grad_norm": 0.7789244055747986, "learning_rate": 0.0001, "elbo": 3.53901, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7456016889514427}
{"step": 63875, "loss": 3.5430828094482423, "grad_norm": 0.7331435084342957, "learning_rate": 0.0001, "elbo": 3.54308, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.745894909688013}
{"step": 63880, "loss": 3.5512977600097657, "grad_norm": 1.0550998449325562, "learning_rate": 0.0001, "elbo": 3.5513, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7461881304245837}
{"step": 63885, "loss": 3.5469871520996095, "grad_norm": 0.6801666617393494, "learning_rate": 0.0001, "elbo": 3.54699, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.746481351161154}
{"step": 63890, "loss": 3.530564880371094, "grad_norm": 0.8490800857543945, "learning_rate": 0.0001, "elbo": 3.53056, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7467745718977246}
{"step": 63895, "loss": 3.5482833862304686, "grad_norm": 0.750555694103241, "learning_rate": 0.0001, "elbo": 3.54828, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.747067792634295}
{"step": 63900, "loss": 3.5487045288085937, "grad_norm": 0.830801248550415, "learning_rate": 0.0001, "elbo": 3.5487, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7473610133708655}
{"step": 63905, "loss": 3.526511383056641, "grad_norm": 0.7221892476081848, "learning_rate": 0.0001, "elbo": 3.52651, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.747654234107436}
{"step": 63910, "loss": 3.5367427825927735, "grad_norm": 0.6954988241195679, "learning_rate": 0.0001, "elbo": 3.53674, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.747947454844007}
{"step": 63915, "loss": 3.5455169677734375, "grad_norm": 0.6791878342628479, "learning_rate": 0.0001, "elbo": 3.54552, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7482406755805773}
{"step": 63920, "loss": 3.533679962158203, "grad_norm": 0.6741350293159485, "learning_rate": 0.0001, "elbo": 3.53368, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7485338963171477}
{"step": 63925, "loss": 3.546580505371094, "grad_norm": 0.7277024984359741, "learning_rate": 0.0001, "elbo": 3.54658, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.748827117053718}
{"step": 63930, "loss": 3.5482181549072265, "grad_norm": 0.7935628890991211, "learning_rate": 0.0001, "elbo": 3.54822, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7491203377902886}
{"step": 63935, "loss": 3.544479751586914, "grad_norm": 0.7930071353912354, "learning_rate": 0.0001, "elbo": 3.54448, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.749413558526859}
{"step": 63940, "loss": 3.5430839538574217, "grad_norm": 0.6981143951416016, "learning_rate": 0.0001, "elbo": 3.54308, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7497067792634295}
{"step": 63945, "loss": 3.5399639129638674, "grad_norm": 0.7067269682884216, "learning_rate": 0.0001, "elbo": 3.53996, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.75}
{"step": 63950, "loss": 3.546367645263672, "grad_norm": 0.7311785221099854, "learning_rate": 0.0001, "elbo": 3.54637, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7502932207365705}
{"step": 63955, "loss": 3.5403823852539062, "grad_norm": 0.7771093249320984, "learning_rate": 0.0001, "elbo": 3.54038, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.750586441473141}
{"step": 63960, "loss": 3.5325775146484375, "grad_norm": 0.8270292282104492, "learning_rate": 0.0001, "elbo": 3.53258, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7508796622097114}
{"step": 63965, "loss": 3.5275924682617186, "grad_norm": 0.6432191133499146, "learning_rate": 0.0001, "elbo": 3.52759, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.751172882946282}
{"step": 63965, "loss": 3.5275924682617186, "grad_norm": 0.6432191133499146, "learning_rate": 0.0001, "elbo": 3.52759, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.751172882946282}
{"step": 63970, "loss": 3.541735076904297, "grad_norm": 0.7208658456802368, "learning_rate": 0.0001, "elbo": 3.54174, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7514661036828523}
{"step": 63975, "loss": 3.5466598510742187, "grad_norm": 0.7637589573860168, "learning_rate": 0.0001, "elbo": 3.54666, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7517593244194227}
{"step": 63980, "loss": 3.5562274932861326, "grad_norm": 0.7676288485527039, "learning_rate": 0.0001, "elbo": 3.55623, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.752052545155993}
{"step": 63980, "loss": 3.5562274932861326, "grad_norm": 0.7676288485527039, "learning_rate": 0.0001, "elbo": 3.55623, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.752052545155993}
{"step": 63985, "loss": 3.55084342956543, "grad_norm": 0.6675389409065247, "learning_rate": 0.0001, "elbo": 3.55084, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.752345765892564}
{"step": 63990, "loss": 3.5278526306152345, "grad_norm": 0.7146257758140564, "learning_rate": 0.0001, "elbo": 3.52785, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7526389866291345}
{"step": 63995, "loss": 3.540994644165039, "grad_norm": 0.6779991388320923, "learning_rate": 0.0001, "elbo": 3.54099, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.752932207365705}
{"step": 64000, "loss": 3.5454338073730467, "grad_norm": 0.7022238373756409, "learning_rate": 0.0001, "elbo": 3.54543, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7532254281022754}
{"step": 64005, "loss": 3.5237144470214843, "grad_norm": 0.7006150484085083, "learning_rate": 0.0001, "elbo": 3.52371, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.753518648838846}
{"step": 64010, "loss": 3.5375022888183594, "grad_norm": 0.7286060452461243, "learning_rate": 0.0001, "elbo": 3.5375, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7538118695754163}
{"step": 64015, "loss": 3.5297531127929687, "grad_norm": 0.7116047143936157, "learning_rate": 0.0001, "elbo": 3.52975, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.754105090311987}
{"step": 64020, "loss": 3.5402908325195312, "grad_norm": 0.6443189978599548, "learning_rate": 0.0001, "elbo": 3.54029, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7543983110485573}
{"step": 64025, "loss": 3.534865951538086, "grad_norm": 0.7255871295928955, "learning_rate": 0.0001, "elbo": 3.53487, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7546915317851277}
{"step": 64030, "loss": 3.538951873779297, "grad_norm": 0.6602746248245239, "learning_rate": 0.0001, "elbo": 3.53895, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7549847525216986}
{"step": 64035, "loss": 3.5445823669433594, "grad_norm": 0.6957009434700012, "learning_rate": 0.0001, "elbo": 3.54458, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.755277973258269}
{"step": 64040, "loss": 3.535432052612305, "grad_norm": 0.6880505681037903, "learning_rate": 0.0001, "elbo": 3.53543, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7555711939948395}
{"step": 64045, "loss": 3.5371654510498045, "grad_norm": 0.6438236832618713, "learning_rate": 0.0001, "elbo": 3.53717, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.75586441473141}
{"step": 64050, "loss": 3.5324447631835936, "grad_norm": 0.7628311514854431, "learning_rate": 0.0001, "elbo": 3.53244, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7561576354679804}
{"step": 64055, "loss": 3.5326438903808595, "grad_norm": 0.7247661352157593, "learning_rate": 0.0001, "elbo": 3.53264, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.756450856204551}
{"step": 64060, "loss": 3.523429107666016, "grad_norm": 0.7094405889511108, "learning_rate": 0.0001, "elbo": 3.52343, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7567440769411213}
{"step": 64065, "loss": 3.547850799560547, "grad_norm": 0.7079247832298279, "learning_rate": 0.0001, "elbo": 3.54785, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.757037297677692}
{"step": 64070, "loss": 3.546403503417969, "grad_norm": 0.7518497705459595, "learning_rate": 0.0001, "elbo": 3.5464, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7573305184142622}
{"step": 64075, "loss": 3.5499229431152344, "grad_norm": 0.7152770161628723, "learning_rate": 0.0001, "elbo": 3.54992, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7576237391508327}
{"step": 64080, "loss": 3.557251739501953, "grad_norm": 0.6856565475463867, "learning_rate": 0.0001, "elbo": 3.55725, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.757916959887403}
{"step": 64085, "loss": 3.5258468627929687, "grad_norm": 0.7087967395782471, "learning_rate": 0.0001, "elbo": 3.52585, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7582101806239736}
{"step": 64090, "loss": 3.532246398925781, "grad_norm": 0.8007420897483826, "learning_rate": 0.0001, "elbo": 3.53225, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.758503401360544}
{"step": 64095, "loss": 3.521162414550781, "grad_norm": 0.6991416811943054, "learning_rate": 0.0001, "elbo": 3.52116, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7587966220971145}
{"step": 64100, "loss": 3.5413803100585937, "grad_norm": 0.7039730548858643, "learning_rate": 0.0001, "elbo": 3.54138, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.759089842833685}
{"step": 64105, "loss": 3.538519287109375, "grad_norm": 0.9071669578552246, "learning_rate": 0.0001, "elbo": 3.53852, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7593830635702554}
{"step": 64110, "loss": 3.5320388793945314, "grad_norm": 0.6771467924118042, "learning_rate": 0.0001, "elbo": 3.53204, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7596762843068263}
{"step": 64115, "loss": 3.5494434356689455, "grad_norm": 0.6593222618103027, "learning_rate": 0.0001, "elbo": 3.54944, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7599695050433968}
{"step": 64120, "loss": 3.548380661010742, "grad_norm": 0.7055714130401611, "learning_rate": 0.0001, "elbo": 3.54838, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7602627257799672}
{"step": 64125, "loss": 3.54398193359375, "grad_norm": 0.6826880574226379, "learning_rate": 0.0001, "elbo": 3.54398, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7605559465165377}
{"step": 64130, "loss": 3.5462249755859374, "grad_norm": 0.6805208325386047, "learning_rate": 0.0001, "elbo": 3.54623, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.760849167253108}
{"step": 64135, "loss": 3.536722946166992, "grad_norm": 0.719569742679596, "learning_rate": 0.0001, "elbo": 3.53672, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7611423879896786}
{"step": 64140, "loss": 3.5371353149414064, "grad_norm": 0.6830166578292847, "learning_rate": 0.0001, "elbo": 3.53713, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.761435608726249}
{"step": 64145, "loss": 3.551148223876953, "grad_norm": 0.7687208652496338, "learning_rate": 0.0001, "elbo": 3.55115, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7617288294628195}
{"step": 64150, "loss": 3.5280895233154297, "grad_norm": 0.655910849571228, "learning_rate": 0.0001, "elbo": 3.52809, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.76202205019939}
{"step": 64155, "loss": 3.518119049072266, "grad_norm": 0.6815053224563599, "learning_rate": 0.0001, "elbo": 3.51812, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.762315270935961}
{"step": 64160, "loss": 3.5428131103515623, "grad_norm": 0.7053964138031006, "learning_rate": 0.0001, "elbo": 3.54281, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7626084916725313}
{"step": 64165, "loss": 3.541770172119141, "grad_norm": 0.723495602607727, "learning_rate": 0.0001, "elbo": 3.54177, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7629017124091018}
{"step": 64170, "loss": 3.539716339111328, "grad_norm": 0.7642624974250793, "learning_rate": 0.0001, "elbo": 3.53972, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.763194933145672}
{"step": 64175, "loss": 3.5352401733398438, "grad_norm": 0.7089694738388062, "learning_rate": 0.0001, "elbo": 3.53524, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7634881538822427}
{"step": 64180, "loss": 3.540760040283203, "grad_norm": 0.7924526929855347, "learning_rate": 0.0001, "elbo": 3.54076, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.763781374618813}
{"step": 64185, "loss": 3.5238834381103517, "grad_norm": 0.8000220656394958, "learning_rate": 0.0001, "elbo": 3.52388, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7640745953553836}
{"step": 64190, "loss": 3.538835906982422, "grad_norm": 0.7185026407241821, "learning_rate": 0.0001, "elbo": 3.53884, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.764367816091954}
{"step": 64195, "loss": 3.5171478271484373, "grad_norm": 0.7012100219726562, "learning_rate": 0.0001, "elbo": 3.51715, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7646610368285245}
{"step": 64200, "loss": 3.542724609375, "grad_norm": 0.7240144610404968, "learning_rate": 0.0001, "elbo": 3.54272, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.764954257565095}
{"step": 64205, "loss": 3.5431407928466796, "grad_norm": 0.7100223302841187, "learning_rate": 0.0001, "elbo": 3.54314, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7652474783016654}
{"step": 64210, "loss": 3.5389915466308595, "grad_norm": 0.922039806842804, "learning_rate": 0.0001, "elbo": 3.53899, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.765540699038236}
{"step": 64215, "loss": 3.525181198120117, "grad_norm": 0.6852046251296997, "learning_rate": 0.0001, "elbo": 3.52518, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7658339197748063}
{"step": 64220, "loss": 3.540579986572266, "grad_norm": 0.7222055792808533, "learning_rate": 0.0001, "elbo": 3.54058, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7661271405113768}
{"step": 64220, "loss": 3.540579986572266, "grad_norm": 0.7222055792808533, "learning_rate": 0.0001, "elbo": 3.54058, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7661271405113768}
{"step": 64225, "loss": 3.541914367675781, "grad_norm": 0.7321732640266418, "learning_rate": 0.0001, "elbo": 3.54191, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.766420361247947}
{"step": 64230, "loss": 3.523101043701172, "grad_norm": 0.7017710208892822, "learning_rate": 0.0001, "elbo": 3.5231, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.766713581984518}
{"step": 64230, "loss": 3.523101043701172, "grad_norm": 0.7017710208892822, "learning_rate": 0.0001, "elbo": 3.5231, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.766713581984518}
{"step": 64235, "loss": 3.5289581298828123, "grad_norm": 0.7088244557380676, "learning_rate": 0.0001, "elbo": 3.52896, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7670068027210886}
{"step": 64240, "loss": 3.5458335876464844, "grad_norm": 0.6954361200332642, "learning_rate": 0.0001, "elbo": 3.54583, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.767300023457659}
{"step": 64245, "loss": 3.5607425689697267, "grad_norm": 0.6899042725563049, "learning_rate": 0.0001, "elbo": 3.56074, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7675932441942295}
{"step": 64250, "loss": 3.5401542663574217, "grad_norm": 0.8204365372657776, "learning_rate": 0.0001, "elbo": 3.54015, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7678864649308}
{"step": 64255, "loss": 3.5293140411376953, "grad_norm": 0.7125844359397888, "learning_rate": 0.0001, "elbo": 3.52931, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7681796856673704}
{"step": 64260, "loss": 3.5308326721191405, "grad_norm": 0.7321808934211731, "learning_rate": 0.0001, "elbo": 3.53083, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.768472906403941}
{"step": 64265, "loss": 3.5476444244384764, "grad_norm": 0.8832089304924011, "learning_rate": 0.0001, "elbo": 3.54764, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7687661271405113}
{"step": 64270, "loss": 3.525157165527344, "grad_norm": 0.7034226059913635, "learning_rate": 0.0001, "elbo": 3.52516, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7690593478770817}
{"step": 64275, "loss": 3.550202178955078, "grad_norm": 0.7068387866020203, "learning_rate": 0.0001, "elbo": 3.5502, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.769352568613652}
{"step": 64280, "loss": 3.5502532958984374, "grad_norm": 0.6789007186889648, "learning_rate": 0.0001, "elbo": 3.55025, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.769645789350223}
{"step": 64285, "loss": 3.5286849975585937, "grad_norm": 0.7146271467208862, "learning_rate": 0.0001, "elbo": 3.52869, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7699390100867936}
{"step": 64290, "loss": 3.5418601989746095, "grad_norm": 0.6927796006202698, "learning_rate": 0.0001, "elbo": 3.54186, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.770232230823364}
{"step": 64295, "loss": 3.5514289855957033, "grad_norm": 0.7719367742538452, "learning_rate": 0.0001, "elbo": 3.55143, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7705254515599345}
{"step": 64300, "loss": 3.54710693359375, "grad_norm": 0.6659088730812073, "learning_rate": 0.0001, "elbo": 3.54711, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.770818672296505}
{"step": 64305, "loss": 3.5214111328125, "grad_norm": 0.6853068470954895, "learning_rate": 0.0001, "elbo": 3.52141, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7711118930330754}
{"step": 64310, "loss": 3.520650100708008, "grad_norm": 0.6952170729637146, "learning_rate": 0.0001, "elbo": 3.52065, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.771405113769646}
{"step": 64315, "loss": 3.5408748626708983, "grad_norm": 0.6592414975166321, "learning_rate": 0.0001, "elbo": 3.54087, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7716983345062163}
{"step": 64320, "loss": 3.5403480529785156, "grad_norm": 0.873197078704834, "learning_rate": 0.0001, "elbo": 3.54035, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7719915552427867}
{"step": 64325, "loss": 3.532975769042969, "grad_norm": 0.6971511840820312, "learning_rate": 0.0001, "elbo": 3.53298, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.772284775979357}
{"step": 64330, "loss": 3.539717102050781, "grad_norm": 0.7332791090011597, "learning_rate": 0.0001, "elbo": 3.53972, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7725779967159276}
{"step": 64335, "loss": 3.5343292236328123, "grad_norm": 0.6769413948059082, "learning_rate": 0.0001, "elbo": 3.53433, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.772871217452498}
{"step": 64340, "loss": 3.5373867034912108, "grad_norm": 0.7980308532714844, "learning_rate": 0.0001, "elbo": 3.53739, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7731644381890685}
{"step": 64345, "loss": 3.5250244140625, "grad_norm": 0.7195084691047668, "learning_rate": 0.0001, "elbo": 3.52502, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.773457658925639}
{"step": 64350, "loss": 3.541333770751953, "grad_norm": 0.7568008303642273, "learning_rate": 0.0001, "elbo": 3.54133, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7737508796622095}
{"step": 64355, "loss": 3.545709228515625, "grad_norm": 0.6946949362754822, "learning_rate": 0.0001, "elbo": 3.54571, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7740441003987804}
{"step": 64360, "loss": 3.542021560668945, "grad_norm": 0.7244269251823425, "learning_rate": 0.0001, "elbo": 3.54202, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.774337321135351}
{"step": 64365, "loss": 3.526968765258789, "grad_norm": 0.6594988703727722, "learning_rate": 0.0001, "elbo": 3.52697, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7746305418719213}
{"step": 64370, "loss": 3.5304447174072267, "grad_norm": 0.6850500702857971, "learning_rate": 0.0001, "elbo": 3.53044, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7749237626084917}
{"step": 64375, "loss": 3.5314285278320314, "grad_norm": 0.6882866621017456, "learning_rate": 0.0001, "elbo": 3.53143, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.775216983345062}
{"step": 64380, "loss": 3.5464691162109374, "grad_norm": 0.7105638980865479, "learning_rate": 0.0001, "elbo": 3.54647, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7755102040816326}
{"step": 64385, "loss": 3.5248863220214846, "grad_norm": 0.8848962783813477, "learning_rate": 0.0001, "elbo": 3.52489, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.775803424818203}
{"step": 64390, "loss": 3.5385578155517576, "grad_norm": 0.7098239660263062, "learning_rate": 0.0001, "elbo": 3.53856, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7760966455547735}
{"step": 64395, "loss": 3.5357597351074217, "grad_norm": 0.9521554708480835, "learning_rate": 0.0001, "elbo": 3.53576, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.776389866291344}
{"step": 64400, "loss": 3.549620819091797, "grad_norm": 0.668496310710907, "learning_rate": 0.0001, "elbo": 3.54962, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.776683087027915}
{"step": 64405, "loss": 3.540325164794922, "grad_norm": 0.6732004880905151, "learning_rate": 0.0001, "elbo": 3.54033, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7769763077644853}
{"step": 64410, "loss": 3.5336875915527344, "grad_norm": 0.723577618598938, "learning_rate": 0.0001, "elbo": 3.53369, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.777269528501056}
{"step": 64415, "loss": 3.5464778900146485, "grad_norm": 0.7064642310142517, "learning_rate": 0.0001, "elbo": 3.54648, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7775627492376262}
{"step": 64420, "loss": 3.5495338439941406, "grad_norm": 0.9410941004753113, "learning_rate": 0.0001, "elbo": 3.54953, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7778559699741967}
{"step": 64425, "loss": 3.559640121459961, "grad_norm": 0.6781846284866333, "learning_rate": 0.0001, "elbo": 3.55964, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.778149190710767}
{"step": 64430, "loss": 3.532530975341797, "grad_norm": 0.711081862449646, "learning_rate": 0.0001, "elbo": 3.53253, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7784424114473376}
{"step": 64435, "loss": 3.527130126953125, "grad_norm": 0.6797820925712585, "learning_rate": 0.0001, "elbo": 3.52713, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.778735632183908}
{"step": 64440, "loss": 3.533064270019531, "grad_norm": 0.6888227462768555, "learning_rate": 0.0001, "elbo": 3.53306, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7790288529204785}
{"step": 64445, "loss": 3.550270843505859, "grad_norm": 0.7203930020332336, "learning_rate": 0.0001, "elbo": 3.55027, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.779322073657049}
{"step": 64450, "loss": 3.548427963256836, "grad_norm": 0.7073801755905151, "learning_rate": 0.0001, "elbo": 3.54843, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7796152943936194}
{"step": 64455, "loss": 3.5517784118652345, "grad_norm": 0.7251977920532227, "learning_rate": 0.0001, "elbo": 3.55178, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.77990851513019}
{"step": 64460, "loss": 3.529430389404297, "grad_norm": 0.7016087770462036, "learning_rate": 0.0001, "elbo": 3.52943, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7802017358667603}
{"step": 64465, "loss": 3.547568511962891, "grad_norm": 0.8907968401908875, "learning_rate": 0.0001, "elbo": 3.54757, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.780494956603331}
{"step": 64470, "loss": 3.5382369995117187, "grad_norm": 0.7567027807235718, "learning_rate": 0.0001, "elbo": 3.53824, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7807881773399012}
{"step": 64475, "loss": 3.538899230957031, "grad_norm": 0.7565916180610657, "learning_rate": 0.0001, "elbo": 3.5389, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7810813980764717}
{"step": 64480, "loss": 3.5354293823242187, "grad_norm": 0.7445358633995056, "learning_rate": 0.0001, "elbo": 3.53543, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7813746188130426}
{"step": 64485, "loss": 3.5420581817626955, "grad_norm": 0.6938493847846985, "learning_rate": 0.0001, "elbo": 3.54206, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.781667839549613}
{"step": 64490, "loss": 3.5514312744140626, "grad_norm": 0.7645183205604553, "learning_rate": 0.0001, "elbo": 3.55143, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7819610602861835}
{"step": 64495, "loss": 3.5374069213867188, "grad_norm": 0.6814413070678711, "learning_rate": 0.0001, "elbo": 3.53741, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.782254281022754}
{"step": 64500, "loss": 3.5234474182128905, "grad_norm": 0.7339409589767456, "learning_rate": 0.0001, "elbo": 3.52345, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7825475017593244}
{"step": 64505, "loss": 3.5385536193847655, "grad_norm": 0.7378242611885071, "learning_rate": 0.0001, "elbo": 3.53855, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.782840722495895}
{"step": 64510, "loss": 3.535778045654297, "grad_norm": 0.7447808384895325, "learning_rate": 0.0001, "elbo": 3.53578, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7831339432324653}
{"step": 64515, "loss": 3.5374118804931642, "grad_norm": 0.7303798794746399, "learning_rate": 0.0001, "elbo": 3.53741, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.783427163969036}
{"step": 64520, "loss": 3.5327228546142577, "grad_norm": 0.8836902379989624, "learning_rate": 0.0001, "elbo": 3.53272, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7837203847056062}
{"step": 64525, "loss": 3.5384498596191407, "grad_norm": 0.6879097819328308, "learning_rate": 0.0001, "elbo": 3.53845, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.784013605442177}
{"step": 64530, "loss": 3.5261207580566407, "grad_norm": 0.7513536214828491, "learning_rate": 0.0001, "elbo": 3.52612, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7843068261787476}
{"step": 64535, "loss": 3.537340545654297, "grad_norm": 0.7751734852790833, "learning_rate": 0.0001, "elbo": 3.53734, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.784600046915318}
{"step": 64540, "loss": 3.539772796630859, "grad_norm": 0.6788269877433777, "learning_rate": 0.0001, "elbo": 3.53977, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7848932676518885}
{"step": 64545, "loss": 3.5362316131591798, "grad_norm": 0.7869969606399536, "learning_rate": 0.0001, "elbo": 3.53623, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.785186488388459}
{"step": 64550, "loss": 3.525311279296875, "grad_norm": 0.6798513531684875, "learning_rate": 0.0001, "elbo": 3.52531, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7854797091250294}
{"step": 64555, "loss": 3.539641571044922, "grad_norm": 0.7047956585884094, "learning_rate": 0.0001, "elbo": 3.53964, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7857729298616}
{"step": 64560, "loss": 3.543656921386719, "grad_norm": 0.7652573585510254, "learning_rate": 0.0001, "elbo": 3.54366, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7860661505981703}
{"step": 64560, "loss": 3.543656921386719, "grad_norm": 0.7652573585510254, "learning_rate": 0.0001, "elbo": 3.54366, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7860661505981703}
{"step": 64565, "loss": 3.525349807739258, "grad_norm": 0.6623102426528931, "learning_rate": 0.0001, "elbo": 3.52535, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7863593713347408}
{"step": 64565, "loss": 3.525349807739258, "grad_norm": 0.6623102426528931, "learning_rate": 0.0001, "elbo": 3.52535, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7863593713347408}
{"step": 64570, "loss": 3.528704071044922, "grad_norm": 0.8101609945297241, "learning_rate": 0.0001, "elbo": 3.5287, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.786652592071311}
{"step": 64575, "loss": 3.5519065856933594, "grad_norm": 0.6850087642669678, "learning_rate": 0.0001, "elbo": 3.55191, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7869458128078817}
{"step": 64580, "loss": 3.5441856384277344, "grad_norm": 0.7744864821434021, "learning_rate": 0.0001, "elbo": 3.54419, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.787239033544452}
{"step": 64585, "loss": 3.5409427642822267, "grad_norm": 0.7584472298622131, "learning_rate": 0.0001, "elbo": 3.54094, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7875322542810226}
{"step": 64590, "loss": 3.5365028381347656, "grad_norm": 0.6620359420776367, "learning_rate": 0.0001, "elbo": 3.5365, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.787825475017593}
{"step": 64595, "loss": 3.555445098876953, "grad_norm": 0.6894088387489319, "learning_rate": 0.0001, "elbo": 3.55545, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7881186957541635}
{"step": 64600, "loss": 3.5438919067382812, "grad_norm": 0.7355021238327026, "learning_rate": 0.0001, "elbo": 3.54389, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7884119164907344}
{"step": 64605, "loss": 3.5359176635742187, "grad_norm": 0.826659083366394, "learning_rate": 0.0001, "elbo": 3.53592, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.788705137227305}
{"step": 64610, "loss": 3.5416698455810547, "grad_norm": 0.7364063262939453, "learning_rate": 0.0001, "elbo": 3.54167, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7889983579638753}
{"step": 64615, "loss": 3.537841033935547, "grad_norm": 0.7324159145355225, "learning_rate": 0.0001, "elbo": 3.53784, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7892915787004458}
{"step": 64620, "loss": 3.5216400146484377, "grad_norm": 0.7065147757530212, "learning_rate": 0.0001, "elbo": 3.52164, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.789584799437016}
{"step": 64625, "loss": 3.5358512878417967, "grad_norm": 0.8093392252922058, "learning_rate": 0.0001, "elbo": 3.53585, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7898780201735867}
{"step": 64630, "loss": 3.536494827270508, "grad_norm": 0.8466895222663879, "learning_rate": 0.0001, "elbo": 3.53649, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.790171240910157}
{"step": 64635, "loss": 3.5456398010253904, "grad_norm": 0.7033378481864929, "learning_rate": 0.0001, "elbo": 3.54564, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7904644616467276}
{"step": 64640, "loss": 3.5548377990722657, "grad_norm": 0.801994264125824, "learning_rate": 0.0001, "elbo": 3.55484, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.790757682383298}
{"step": 64645, "loss": 3.553974151611328, "grad_norm": 0.8400310277938843, "learning_rate": 0.0001, "elbo": 3.55397, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7910509031198685}
{"step": 64650, "loss": 3.5366329193115233, "grad_norm": 0.7514961957931519, "learning_rate": 0.0001, "elbo": 3.53663, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7913441238564394}
{"step": 64655, "loss": 3.525889587402344, "grad_norm": 0.7135504484176636, "learning_rate": 0.0001, "elbo": 3.52589, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.79163734459301}
{"step": 64655, "loss": 3.525889587402344, "grad_norm": 0.7135504484176636, "learning_rate": 0.0001, "elbo": 3.52589, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.79163734459301}
{"step": 64660, "loss": 3.5334503173828127, "grad_norm": 1.5197316408157349, "learning_rate": 0.0001, "elbo": 3.53345, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7919305653295803}
{"step": 64665, "loss": 3.525225067138672, "grad_norm": 0.7173155546188354, "learning_rate": 0.0001, "elbo": 3.52523, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7922237860661507}
{"step": 64670, "loss": 3.5337993621826174, "grad_norm": 0.8113857507705688, "learning_rate": 0.0001, "elbo": 3.5338, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.792517006802721}
{"step": 64675, "loss": 3.551915740966797, "grad_norm": 0.7503695487976074, "learning_rate": 0.0001, "elbo": 3.55192, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7928102275392916}
{"step": 64680, "loss": 3.5501113891601563, "grad_norm": 0.6819034814834595, "learning_rate": 0.0001, "elbo": 3.55011, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.793103448275862}
{"step": 64685, "loss": 3.5431236267089843, "grad_norm": 0.6815372705459595, "learning_rate": 0.0001, "elbo": 3.54312, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7933966690124326}
{"step": 64690, "loss": 3.5410011291503904, "grad_norm": 0.6908155679702759, "learning_rate": 0.0001, "elbo": 3.541, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.793689889749003}
{"step": 64695, "loss": 3.5194515228271483, "grad_norm": 0.7052582502365112, "learning_rate": 0.0001, "elbo": 3.51945, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7939831104855735}
{"step": 64700, "loss": 3.5338550567626954, "grad_norm": 0.7247929573059082, "learning_rate": 0.0001, "elbo": 3.53386, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.794276331222144}
{"step": 64705, "loss": 3.5434173583984374, "grad_norm": 0.7461820244789124, "learning_rate": 0.0001, "elbo": 3.54342, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7945695519587144}
{"step": 64710, "loss": 3.535575103759766, "grad_norm": 0.6859113574028015, "learning_rate": 0.0001, "elbo": 3.53557, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.794862772695285}
{"step": 64715, "loss": 3.5456192016601564, "grad_norm": 0.7475017309188843, "learning_rate": 0.0001, "elbo": 3.54562, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7951559934318553}
{"step": 64720, "loss": 3.5513587951660157, "grad_norm": 0.8174607753753662, "learning_rate": 0.0001, "elbo": 3.55136, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7954492141684257}
{"step": 64725, "loss": 3.546955108642578, "grad_norm": 0.6749640107154846, "learning_rate": 0.0001, "elbo": 3.54696, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7957424349049966}
{"step": 64730, "loss": 3.542990875244141, "grad_norm": 0.7164029479026794, "learning_rate": 0.0001, "elbo": 3.54299, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.796035655641567}
{"step": 64735, "loss": 3.5357147216796876, "grad_norm": 0.6818727850914001, "learning_rate": 0.0001, "elbo": 3.53571, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7963288763781375}
{"step": 64740, "loss": 3.5290863037109377, "grad_norm": 0.7429769039154053, "learning_rate": 0.0001, "elbo": 3.52909, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.796622097114708}
{"step": 64745, "loss": 3.5369625091552734, "grad_norm": 0.692249059677124, "learning_rate": 0.0001, "elbo": 3.53696, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7969153178512784}
{"step": 64750, "loss": 3.535852813720703, "grad_norm": 0.6957517266273499, "learning_rate": 0.0001, "elbo": 3.53585, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.797208538587849}
{"step": 64755, "loss": 3.5326160430908202, "grad_norm": 0.7052175998687744, "learning_rate": 0.0001, "elbo": 3.53262, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7975017593244194}
{"step": 64760, "loss": 3.5183795928955077, "grad_norm": 0.6785051822662354, "learning_rate": 0.0001, "elbo": 3.51838, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.79779498006099}
{"step": 64765, "loss": 3.522999572753906, "grad_norm": 0.6896841526031494, "learning_rate": 0.0001, "elbo": 3.523, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7980882007975603}
{"step": 64770, "loss": 3.5336490631103517, "grad_norm": 0.7782326340675354, "learning_rate": 0.0001, "elbo": 3.53365, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.798381421534131}
{"step": 64775, "loss": 3.5371780395507812, "grad_norm": 0.7186218500137329, "learning_rate": 0.0001, "elbo": 3.53718, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7986746422707016}
{"step": 64780, "loss": 3.551649475097656, "grad_norm": 0.748481810092926, "learning_rate": 0.0001, "elbo": 3.55165, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.798967863007272}
{"step": 64785, "loss": 3.543609619140625, "grad_norm": 1.55363929271698, "learning_rate": 0.0001, "elbo": 3.54361, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7992610837438425}
{"step": 64790, "loss": 3.5366451263427736, "grad_norm": 1.0265812873840332, "learning_rate": 0.0001, "elbo": 3.53665, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.799554304480413}
{"step": 64795, "loss": 3.5376747131347654, "grad_norm": 0.7455295920372009, "learning_rate": 0.0001, "elbo": 3.53767, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.7998475252169834}
{"step": 64800, "loss": 3.5253963470458984, "grad_norm": 0.7083068490028381, "learning_rate": 0.0001, "elbo": 3.5254, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.800140745953554}
{"step": 64805, "loss": 3.5466293334960937, "grad_norm": 0.7029122710227966, "learning_rate": 0.0001, "elbo": 3.54663, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8004339666901243}
{"step": 64810, "loss": 3.529433822631836, "grad_norm": 0.6617216467857361, "learning_rate": 0.0001, "elbo": 3.52943, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.800727187426695}
{"step": 64815, "loss": 3.5478561401367186, "grad_norm": 0.7271144986152649, "learning_rate": 0.0001, "elbo": 3.54786, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8010204081632653}
{"step": 64820, "loss": 3.5195472717285154, "grad_norm": 0.6715351343154907, "learning_rate": 0.0001, "elbo": 3.51955, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8013136288998357}
{"step": 64825, "loss": 3.540287399291992, "grad_norm": 0.7639340758323669, "learning_rate": 0.0001, "elbo": 3.54029, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.801606849636406}
{"step": 64830, "loss": 3.540623092651367, "grad_norm": 0.7270562052726746, "learning_rate": 0.0001, "elbo": 3.54062, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8019000703729766}
{"step": 64835, "loss": 3.5329299926757813, "grad_norm": 0.7180764079093933, "learning_rate": 0.0001, "elbo": 3.53293, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.802193291109547}
{"step": 64840, "loss": 3.529842758178711, "grad_norm": 0.8063609600067139, "learning_rate": 0.0001, "elbo": 3.52984, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8024865118461175}
{"step": 64845, "loss": 3.5209556579589845, "grad_norm": 0.7507511973381042, "learning_rate": 0.0001, "elbo": 3.52096, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.802779732582688}
{"step": 64850, "loss": 3.542373275756836, "grad_norm": 1.0139522552490234, "learning_rate": 0.0001, "elbo": 3.54237, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.803072953319259}
{"step": 64855, "loss": 3.5234676361083985, "grad_norm": 0.6815670728683472, "learning_rate": 0.0001, "elbo": 3.52347, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8033661740558293}
{"step": 64860, "loss": 3.5385673522949217, "grad_norm": 0.7176384925842285, "learning_rate": 0.0001, "elbo": 3.53857, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8036593947924}
{"step": 64865, "loss": 3.5402244567871093, "grad_norm": 0.7346540689468384, "learning_rate": 0.0001, "elbo": 3.54022, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8039526155289702}
{"step": 64870, "loss": 3.539469528198242, "grad_norm": 0.8469979763031006, "learning_rate": 0.0001, "elbo": 3.53947, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8042458362655407}
{"step": 64875, "loss": 3.5379051208496093, "grad_norm": 0.7754248976707458, "learning_rate": 0.0001, "elbo": 3.53791, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.804539057002111}
{"step": 64880, "loss": 3.535976791381836, "grad_norm": 0.7334629893302917, "learning_rate": 0.0001, "elbo": 3.53598, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8048322777386816}
{"step": 64885, "loss": 3.5321388244628906, "grad_norm": 1.1229207515716553, "learning_rate": 0.0001, "elbo": 3.53214, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.805125498475252}
{"step": 64890, "loss": 3.540425109863281, "grad_norm": 0.6559163331985474, "learning_rate": 0.0001, "elbo": 3.54043, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8054187192118225}
{"step": 64895, "loss": 3.5306171417236327, "grad_norm": 1.0558655261993408, "learning_rate": 0.0001, "elbo": 3.53062, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8057119399483934}
{"step": 64900, "loss": 3.546772003173828, "grad_norm": 0.7221511602401733, "learning_rate": 0.0001, "elbo": 3.54677, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.806005160684964}
{"step": 64905, "loss": 3.526575469970703, "grad_norm": 0.7313975095748901, "learning_rate": 0.0001, "elbo": 3.52658, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8062983814215343}
{"step": 64910, "loss": 3.5269596099853517, "grad_norm": 0.7883651256561279, "learning_rate": 0.0001, "elbo": 3.52696, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8065916021581048}
{"step": 64915, "loss": 3.5481468200683595, "grad_norm": 0.7177932262420654, "learning_rate": 0.0001, "elbo": 3.54815, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8068848228946752}
{"step": 64920, "loss": 3.5376201629638673, "grad_norm": 0.6747309565544128, "learning_rate": 0.0001, "elbo": 3.53762, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8071780436312457}
{"step": 64925, "loss": 3.55774040222168, "grad_norm": 0.7669373154640198, "learning_rate": 0.0001, "elbo": 3.55774, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.807471264367816}
{"step": 64930, "loss": 3.5416305541992186, "grad_norm": 0.7832884788513184, "learning_rate": 0.0001, "elbo": 3.54163, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8077644851043866}
{"step": 64935, "loss": 3.53182373046875, "grad_norm": 0.6929073333740234, "learning_rate": 0.0001, "elbo": 3.53182, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.808057705840957}
{"step": 64940, "loss": 3.5444786071777346, "grad_norm": 0.7089134454727173, "learning_rate": 0.0001, "elbo": 3.54448, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8083509265775275}
{"step": 64945, "loss": 3.5471900939941405, "grad_norm": 0.8093032240867615, "learning_rate": 0.0001, "elbo": 3.54719, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.808644147314098}
{"step": 64950, "loss": 3.5389808654785155, "grad_norm": 0.6903636455535889, "learning_rate": 0.0001, "elbo": 3.53898, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8089373680506684}
{"step": 64955, "loss": 3.5367668151855467, "grad_norm": 0.6650686264038086, "learning_rate": 0.0001, "elbo": 3.53677, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.809230588787239}
{"step": 64960, "loss": 3.5262596130371096, "grad_norm": 0.6619276404380798, "learning_rate": 0.0001, "elbo": 3.52626, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8095238095238093}
{"step": 64965, "loss": 3.527873992919922, "grad_norm": 0.8940737843513489, "learning_rate": 0.0001, "elbo": 3.52787, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8098170302603798}
{"step": 64970, "loss": 3.5445526123046873, "grad_norm": 0.6964415311813354, "learning_rate": 0.0001, "elbo": 3.54455, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8101102509969507}
{"step": 64975, "loss": 3.524068832397461, "grad_norm": 0.7659860849380493, "learning_rate": 0.0001, "elbo": 3.52407, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.810403471733521}
{"step": 64980, "loss": 3.5315177917480467, "grad_norm": 0.789584755897522, "learning_rate": 0.0001, "elbo": 3.53152, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8106966924700916}
{"step": 64985, "loss": 3.5340301513671877, "grad_norm": 0.706227719783783, "learning_rate": 0.0001, "elbo": 3.53403, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.810989913206662}
{"step": 64990, "loss": 3.5514339447021483, "grad_norm": 0.7419474720954895, "learning_rate": 0.0001, "elbo": 3.55143, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8112831339432325}
{"step": 64995, "loss": 3.5274730682373048, "grad_norm": 0.7041275501251221, "learning_rate": 0.0001, "elbo": 3.52747, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.811576354679803}
{"step": 65000, "loss": 3.5381256103515626, "grad_norm": 0.7593043446540833, "learning_rate": 0.0001, "elbo": 3.53813, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8118695754163734}
{"step": 65005, "loss": 3.546859359741211, "grad_norm": 0.7142918705940247, "learning_rate": 0.0001, "elbo": 3.54686, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.812162796152944}
{"step": 65005, "loss": 3.546859359741211, "grad_norm": 0.7142918705940247, "learning_rate": 0.0001, "elbo": 3.54686, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.812162796152944}
{"step": 65010, "loss": 3.552932357788086, "grad_norm": 0.7002834677696228, "learning_rate": 0.0001, "elbo": 3.55293, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8124560168895143}
{"step": 65010, "loss": 3.552932357788086, "grad_norm": 0.7002834677696228, "learning_rate": 0.0001, "elbo": 3.55293, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8124560168895143}
{"step": 65015, "loss": 3.5453147888183594, "grad_norm": 1.0233736038208008, "learning_rate": 0.0001, "elbo": 3.54531, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8127492376260848}
{"step": 65020, "loss": 3.5337703704833983, "grad_norm": 1.2593415975570679, "learning_rate": 0.0001, "elbo": 3.53377, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8130424583626557}
{"step": 65025, "loss": 3.5328086853027343, "grad_norm": 0.7407713532447815, "learning_rate": 0.0001, "elbo": 3.53281, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.813335679099226}
{"step": 65030, "loss": 3.54186897277832, "grad_norm": 0.6835176944732666, "learning_rate": 0.0001, "elbo": 3.54187, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8136288998357966}
{"step": 65035, "loss": 3.542044830322266, "grad_norm": 0.7067581415176392, "learning_rate": 0.0001, "elbo": 3.54204, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.813922120572367}
{"step": 65040, "loss": 3.5401889801025392, "grad_norm": 0.6812849640846252, "learning_rate": 0.0001, "elbo": 3.54019, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8142153413089375}
{"step": 65045, "loss": 3.5390556335449217, "grad_norm": 0.709730327129364, "learning_rate": 0.0001, "elbo": 3.53906, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.814508562045508}
{"step": 65050, "loss": 3.544694519042969, "grad_norm": 0.7391626834869385, "learning_rate": 0.0001, "elbo": 3.54469, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8148017827820784}
{"step": 65055, "loss": 3.5403606414794924, "grad_norm": 0.7633344531059265, "learning_rate": 0.0001, "elbo": 3.54036, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.815095003518649}
{"step": 65060, "loss": 3.527976226806641, "grad_norm": 0.7471288442611694, "learning_rate": 0.0001, "elbo": 3.52798, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8153882242552193}
{"step": 65065, "loss": 3.5632381439208984, "grad_norm": 0.6958476305007935, "learning_rate": 0.0001, "elbo": 3.56324, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8156814449917897}
{"step": 65070, "loss": 3.557376480102539, "grad_norm": 0.7125701904296875, "learning_rate": 0.0001, "elbo": 3.55738, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.81597466572836}
{"step": 65075, "loss": 3.5289493560791017, "grad_norm": 0.6835156083106995, "learning_rate": 0.0001, "elbo": 3.52895, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8162678864649306}
{"step": 65080, "loss": 3.5517765045166017, "grad_norm": 0.7081135511398315, "learning_rate": 0.0001, "elbo": 3.55178, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.816561107201501}
{"step": 65085, "loss": 3.5264358520507812, "grad_norm": 0.8098145127296448, "learning_rate": 0.0001, "elbo": 3.52644, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8168543279380716}
{"step": 65090, "loss": 3.5394332885742186, "grad_norm": 0.7251561880111694, "learning_rate": 0.0001, "elbo": 3.53943, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.817147548674642}
{"step": 65095, "loss": 3.5330322265625, "grad_norm": 0.6988842487335205, "learning_rate": 0.0001, "elbo": 3.53303, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.817440769411213}
{"step": 65100, "loss": 3.518768310546875, "grad_norm": 0.6937569379806519, "learning_rate": 0.0001, "elbo": 3.51877, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8177339901477834}
{"step": 65105, "loss": 3.530204010009766, "grad_norm": 0.6427940726280212, "learning_rate": 0.0001, "elbo": 3.5302, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.818027210884354}
{"step": 65110, "loss": 3.5363849639892577, "grad_norm": 0.6480488777160645, "learning_rate": 0.0001, "elbo": 3.53638, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8183204316209243}
{"step": 65115, "loss": 3.5484855651855467, "grad_norm": 1.105719804763794, "learning_rate": 0.0001, "elbo": 3.54849, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8186136523574947}
{"step": 65120, "loss": 3.5527271270751952, "grad_norm": 0.8864952921867371, "learning_rate": 0.0001, "elbo": 3.55273, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.818906873094065}
{"step": 65120, "loss": 3.5527271270751952, "grad_norm": 0.8864952921867371, "learning_rate": 0.0001, "elbo": 3.55273, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.818906873094065}
{"step": 65125, "loss": 3.5472999572753907, "grad_norm": 0.7969110608100891, "learning_rate": 0.0001, "elbo": 3.5473, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8192000938306356}
{"step": 65130, "loss": 3.5477214813232423, "grad_norm": 0.6797446608543396, "learning_rate": 0.0001, "elbo": 3.54772, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.819493314567206}
{"step": 65135, "loss": 3.535556411743164, "grad_norm": 0.7351347208023071, "learning_rate": 0.0001, "elbo": 3.53556, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8197865353037765}
{"step": 65140, "loss": 3.549713897705078, "grad_norm": 0.6778602600097656, "learning_rate": 0.0001, "elbo": 3.54971, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8200797560403474}
{"step": 65145, "loss": 3.5310989379882813, "grad_norm": 0.6831026673316956, "learning_rate": 0.0001, "elbo": 3.5311, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.820372976776918}
{"step": 65150, "loss": 3.5399200439453127, "grad_norm": 0.6756490468978882, "learning_rate": 0.0001, "elbo": 3.53992, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8206661975134883}
{"step": 65155, "loss": 3.5474922180175783, "grad_norm": 0.7178313732147217, "learning_rate": 0.0001, "elbo": 3.54749, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.820959418250059}
{"step": 65160, "loss": 3.543585205078125, "grad_norm": 0.7079309225082397, "learning_rate": 0.0001, "elbo": 3.54359, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8212526389866293}
{"step": 65165, "loss": 3.5155593872070314, "grad_norm": 0.7683488130569458, "learning_rate": 0.0001, "elbo": 3.51556, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8215458597231997}
{"step": 65170, "loss": 3.548486328125, "grad_norm": 0.7605178356170654, "learning_rate": 0.0001, "elbo": 3.54849, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.82183908045977}
{"step": 65175, "loss": 3.5268184661865236, "grad_norm": 0.7024928331375122, "learning_rate": 0.0001, "elbo": 3.52682, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8221323011963406}
{"step": 65180, "loss": 3.5464797973632813, "grad_norm": 0.7337398529052734, "learning_rate": 0.0001, "elbo": 3.54648, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.822425521932911}
{"step": 65185, "loss": 3.535089874267578, "grad_norm": 0.6781879663467407, "learning_rate": 0.0001, "elbo": 3.53509, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8227187426694815}
{"step": 65190, "loss": 3.521897888183594, "grad_norm": 0.707484245300293, "learning_rate": 0.0001, "elbo": 3.5219, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.823011963406052}
{"step": 65195, "loss": 3.541545104980469, "grad_norm": 0.7164794206619263, "learning_rate": 0.0001, "elbo": 3.54155, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8233051841426224}
{"step": 65200, "loss": 3.5211456298828123, "grad_norm": 0.7320765852928162, "learning_rate": 0.0001, "elbo": 3.52115, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.823598404879193}
{"step": 65205, "loss": 3.535419464111328, "grad_norm": 0.6792246699333191, "learning_rate": 0.0001, "elbo": 3.53542, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8238916256157633}
{"step": 65210, "loss": 3.5324123382568358, "grad_norm": 0.8392769694328308, "learning_rate": 0.0001, "elbo": 3.53241, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.824184846352334}
{"step": 65215, "loss": 3.530619812011719, "grad_norm": 1.0372222661972046, "learning_rate": 0.0001, "elbo": 3.53062, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8244780670889043}
{"step": 65220, "loss": 3.5523170471191405, "grad_norm": 0.6831365823745728, "learning_rate": 0.0001, "elbo": 3.55232, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.824771287825475}
{"step": 65225, "loss": 3.5256637573242187, "grad_norm": 0.8329692482948303, "learning_rate": 0.0001, "elbo": 3.52566, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8250645085620456}
{"step": 65230, "loss": 3.531249237060547, "grad_norm": 0.6841952800750732, "learning_rate": 0.0001, "elbo": 3.53125, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.825357729298616}
{"step": 65235, "loss": 3.5440372467041015, "grad_norm": 0.6863594055175781, "learning_rate": 0.0001, "elbo": 3.54404, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8256509500351865}
{"step": 65240, "loss": 3.5459476470947267, "grad_norm": 0.6987881064414978, "learning_rate": 0.0001, "elbo": 3.54595, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.825944170771757}
{"step": 65245, "loss": 3.549418640136719, "grad_norm": 0.6939511299133301, "learning_rate": 0.0001, "elbo": 3.54942, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8262373915083274}
{"step": 65250, "loss": 3.534180450439453, "grad_norm": 0.7963654398918152, "learning_rate": 0.0001, "elbo": 3.53418, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.826530612244898}
{"step": 65255, "loss": 3.542690658569336, "grad_norm": 0.7005694508552551, "learning_rate": 0.0001, "elbo": 3.54269, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8268238329814683}
{"step": 65260, "loss": 3.53406982421875, "grad_norm": 0.7334955334663391, "learning_rate": 0.0001, "elbo": 3.53407, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.827117053718039}
{"step": 65265, "loss": 3.533155822753906, "grad_norm": 0.7532551884651184, "learning_rate": 0.0001, "elbo": 3.53316, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8274102744546097}
{"step": 65270, "loss": 3.5437103271484376, "grad_norm": 0.7389511466026306, "learning_rate": 0.0001, "elbo": 3.54371, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.82770349519118}
{"step": 65275, "loss": 3.525502014160156, "grad_norm": 0.7639499306678772, "learning_rate": 0.0001, "elbo": 3.5255, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8279967159277506}
{"step": 65280, "loss": 3.531873321533203, "grad_norm": 0.7029274702072144, "learning_rate": 0.0001, "elbo": 3.53187, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.828289936664321}
{"step": 65285, "loss": 3.533041000366211, "grad_norm": 0.7137957811355591, "learning_rate": 0.0001, "elbo": 3.53304, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8285831574008915}
{"step": 65290, "loss": 3.546129608154297, "grad_norm": 0.6929609775543213, "learning_rate": 0.0001, "elbo": 3.54613, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.828876378137462}
{"step": 65295, "loss": 3.5569076538085938, "grad_norm": 0.711306631565094, "learning_rate": 0.0001, "elbo": 3.55691, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8291695988740324}
{"step": 65300, "loss": 3.5321735382080077, "grad_norm": 0.6940627098083496, "learning_rate": 0.0001, "elbo": 3.53217, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.829462819610603}
{"step": 65305, "loss": 3.5370079040527345, "grad_norm": 0.6688646078109741, "learning_rate": 0.0001, "elbo": 3.53701, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8297560403471733}
{"step": 65310, "loss": 3.5357315063476564, "grad_norm": 0.7181284427642822, "learning_rate": 0.0001, "elbo": 3.53573, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8300492610837438}
{"step": 65315, "loss": 3.539756011962891, "grad_norm": 0.7250913977622986, "learning_rate": 0.0001, "elbo": 3.53976, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8303424818203142}
{"step": 65320, "loss": 3.5442481994628907, "grad_norm": 0.7210379242897034, "learning_rate": 0.0001, "elbo": 3.54425, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8306357025568847}
{"step": 65325, "loss": 3.521654510498047, "grad_norm": 0.7274755835533142, "learning_rate": 0.0001, "elbo": 3.52165, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.830928923293455}
{"step": 65330, "loss": 3.542671966552734, "grad_norm": 0.7614715099334717, "learning_rate": 0.0001, "elbo": 3.54267, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8312221440300256}
{"step": 65335, "loss": 3.5458568572998046, "grad_norm": 0.6873741745948792, "learning_rate": 0.0001, "elbo": 3.54586, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.831515364766596}
{"step": 65340, "loss": 3.529069519042969, "grad_norm": 0.7521064281463623, "learning_rate": 0.0001, "elbo": 3.52907, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.831808585503167}
{"step": 65345, "loss": 3.5276641845703125, "grad_norm": 0.9631592035293579, "learning_rate": 0.0001, "elbo": 3.52766, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8321018062397374}
{"step": 65350, "loss": 3.530739593505859, "grad_norm": 0.8389905691146851, "learning_rate": 0.0001, "elbo": 3.53074, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.832395026976308}
{"step": 65350, "loss": 3.530739593505859, "grad_norm": 0.8389905691146851, "learning_rate": 0.0001, "elbo": 3.53074, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.832395026976308}
{"step": 65355, "loss": 3.527912902832031, "grad_norm": 0.9778133034706116, "learning_rate": 0.0001, "elbo": 3.52791, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8326882477128783}
{"step": 65360, "loss": 3.525645446777344, "grad_norm": 0.6889169216156006, "learning_rate": 0.0001, "elbo": 3.52565, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8329814684494488}
{"step": 65365, "loss": 3.5362098693847654, "grad_norm": 0.70350182056427, "learning_rate": 0.0001, "elbo": 3.53621, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.833274689186019}
{"step": 65370, "loss": 3.540366363525391, "grad_norm": 0.6793226599693298, "learning_rate": 0.0001, "elbo": 3.54037, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8335679099225897}
{"step": 65375, "loss": 3.538920593261719, "grad_norm": 0.6608847975730896, "learning_rate": 0.0001, "elbo": 3.53892, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.83386113065916}
{"step": 65380, "loss": 3.530439758300781, "grad_norm": 0.8155019283294678, "learning_rate": 0.0001, "elbo": 3.53044, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8341543513957306}
{"step": 65385, "loss": 3.542044830322266, "grad_norm": 0.6917913556098938, "learning_rate": 0.0001, "elbo": 3.54204, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8344475721323015}
{"step": 65390, "loss": 3.5169837951660154, "grad_norm": 0.6813197135925293, "learning_rate": 0.0001, "elbo": 3.51698, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.834740792868872}
{"step": 65395, "loss": 3.54862060546875, "grad_norm": 0.6936659216880798, "learning_rate": 0.0001, "elbo": 3.54862, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8350340136054424}
{"step": 65400, "loss": 3.512311553955078, "grad_norm": 0.7179564237594604, "learning_rate": 0.0001, "elbo": 3.51231, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.835327234342013}
{"step": 65405, "loss": 3.535722351074219, "grad_norm": 0.7684251070022583, "learning_rate": 0.0001, "elbo": 3.53572, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8356204550785833}
{"step": 65410, "loss": 3.5344268798828127, "grad_norm": 0.6928349137306213, "learning_rate": 0.0001, "elbo": 3.53443, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8359136758151537}
{"step": 65415, "loss": 3.5392974853515624, "grad_norm": 0.6882538795471191, "learning_rate": 0.0001, "elbo": 3.5393, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.836206896551724}
{"step": 65420, "loss": 3.546214294433594, "grad_norm": 0.6954171657562256, "learning_rate": 0.0001, "elbo": 3.54621, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8365001172882947}
{"step": 65425, "loss": 3.5283439636230467, "grad_norm": 0.6638173460960388, "learning_rate": 0.0001, "elbo": 3.52834, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.836793338024865}
{"step": 65430, "loss": 3.5249618530273437, "grad_norm": 0.8355957269668579, "learning_rate": 0.0001, "elbo": 3.52496, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8370865587614356}
{"step": 65435, "loss": 3.54019775390625, "grad_norm": 0.7181715965270996, "learning_rate": 0.0001, "elbo": 3.5402, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.837379779498006}
{"step": 65440, "loss": 3.535230255126953, "grad_norm": 0.6714567542076111, "learning_rate": 0.0001, "elbo": 3.53523, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8376730002345765}
{"step": 65445, "loss": 3.5359840393066406, "grad_norm": 0.7038928866386414, "learning_rate": 0.0001, "elbo": 3.53598, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.837966220971147}
{"step": 65450, "loss": 3.5499908447265627, "grad_norm": 0.7439233064651489, "learning_rate": 0.0001, "elbo": 3.54999, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8382594417077174}
{"step": 65455, "loss": 3.5432567596435547, "grad_norm": 0.948475182056427, "learning_rate": 0.0001, "elbo": 3.54326, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.838552662444288}
{"step": 65460, "loss": 3.5493747711181642, "grad_norm": 0.7944881319999695, "learning_rate": 0.0001, "elbo": 3.54937, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8388458831808583}
{"step": 65465, "loss": 3.540329360961914, "grad_norm": 0.7761710286140442, "learning_rate": 0.0001, "elbo": 3.54033, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.839139103917429}
{"step": 65470, "loss": 3.5445411682128904, "grad_norm": 0.7409703731536865, "learning_rate": 0.0001, "elbo": 3.54454, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8394323246539996}
{"step": 65475, "loss": 3.541211700439453, "grad_norm": 0.8748078346252441, "learning_rate": 0.0001, "elbo": 3.54121, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.83972554539057}
{"step": 65480, "loss": 3.539316940307617, "grad_norm": 0.6958019733428955, "learning_rate": 0.0001, "elbo": 3.53932, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8400187661271405}
{"step": 65480, "loss": 3.539316940307617, "grad_norm": 0.6958019733428955, "learning_rate": 0.0001, "elbo": 3.53932, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8400187661271405}
{"step": 65485, "loss": 3.5334331512451174, "grad_norm": 0.7592263221740723, "learning_rate": 0.0001, "elbo": 3.53343, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.840311986863711}
{"step": 65490, "loss": 3.5438827514648437, "grad_norm": 0.7173005938529968, "learning_rate": 0.0001, "elbo": 3.54388, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8406052076002815}
{"step": 65495, "loss": 3.5475162506103515, "grad_norm": 0.8293552994728088, "learning_rate": 0.0001, "elbo": 3.54752, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.840898428336852}
{"step": 65495, "loss": 3.5475162506103515, "grad_norm": 0.8293552994728088, "learning_rate": 0.0001, "elbo": 3.54752, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.840898428336852}
{"step": 65500, "loss": 3.547873306274414, "grad_norm": 0.6599446535110474, "learning_rate": 0.0001, "elbo": 3.54787, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8411916490734224}
{"step": 65505, "loss": 3.529081344604492, "grad_norm": 0.7075316905975342, "learning_rate": 0.0001, "elbo": 3.52908, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.841484869809993}
{"step": 65510, "loss": 3.5304496765136717, "grad_norm": 0.7430455684661865, "learning_rate": 0.0001, "elbo": 3.53045, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8417780905465637}
{"step": 65515, "loss": 3.534316635131836, "grad_norm": 0.7386600971221924, "learning_rate": 0.0001, "elbo": 3.53432, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.842071311283134}
{"step": 65520, "loss": 3.524919891357422, "grad_norm": 0.762353241443634, "learning_rate": 0.0001, "elbo": 3.52492, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8423645320197046}
{"step": 65525, "loss": 3.538258361816406, "grad_norm": 0.7912049889564514, "learning_rate": 0.0001, "elbo": 3.53826, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.842657752756275}
{"step": 65530, "loss": 3.5437042236328127, "grad_norm": 0.8009229302406311, "learning_rate": 0.0001, "elbo": 3.5437, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8429509734928455}
{"step": 65535, "loss": 3.5282924652099608, "grad_norm": 0.6803135871887207, "learning_rate": 0.0001, "elbo": 3.52829, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.843244194229416}
{"step": 65540, "loss": 3.5363040924072267, "grad_norm": 0.7417299151420593, "learning_rate": 0.0001, "elbo": 3.5363, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8435374149659864}
{"step": 65545, "loss": 3.5224044799804686, "grad_norm": 0.7386024594306946, "learning_rate": 0.0001, "elbo": 3.5224, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.843830635702557}
{"step": 65550, "loss": 3.514647674560547, "grad_norm": 0.716157853603363, "learning_rate": 0.0001, "elbo": 3.51465, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8441238564391274}
{"step": 65555, "loss": 3.5209930419921873, "grad_norm": 0.9801916480064392, "learning_rate": 0.0001, "elbo": 3.52099, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.844417077175698}
{"step": 65560, "loss": 3.5240852355957033, "grad_norm": 0.6454463601112366, "learning_rate": 0.0001, "elbo": 3.52409, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8447102979122683}
{"step": 65565, "loss": 3.5480266571044923, "grad_norm": 1.2408537864685059, "learning_rate": 0.0001, "elbo": 3.54803, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8450035186488387}
{"step": 65570, "loss": 3.528891754150391, "grad_norm": 0.7331063747406006, "learning_rate": 0.0001, "elbo": 3.52889, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.845296739385409}
{"step": 65575, "loss": 3.529704284667969, "grad_norm": 0.6950545310974121, "learning_rate": 0.0001, "elbo": 3.5297, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8455899601219796}
{"step": 65580, "loss": 3.5360206604003905, "grad_norm": 0.8710334897041321, "learning_rate": 0.0001, "elbo": 3.53602, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.84588318085855}
{"step": 65585, "loss": 3.531048583984375, "grad_norm": 0.7679834961891174, "learning_rate": 0.0001, "elbo": 3.53105, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8461764015951205}
{"step": 65590, "loss": 3.537931442260742, "grad_norm": 0.6993146538734436, "learning_rate": 0.0001, "elbo": 3.53793, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8464696223316914}
{"step": 65595, "loss": 3.5434329986572264, "grad_norm": 0.773324728012085, "learning_rate": 0.0001, "elbo": 3.54343, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.846762843068262}
{"step": 65600, "loss": 3.5246044158935548, "grad_norm": 0.7521956562995911, "learning_rate": 0.0001, "elbo": 3.5246, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8470560638048323}
{"step": 65605, "loss": 3.531449890136719, "grad_norm": 0.7808548212051392, "learning_rate": 0.0001, "elbo": 3.53145, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.847349284541403}
{"step": 65610, "loss": 3.5310787200927733, "grad_norm": 0.7147254943847656, "learning_rate": 0.0001, "elbo": 3.53108, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8476425052779732}
{"step": 65615, "loss": 3.5280826568603514, "grad_norm": 0.7543236017227173, "learning_rate": 0.0001, "elbo": 3.52808, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8479357260145437}
{"step": 65620, "loss": 3.5338108062744142, "grad_norm": 0.7619392275810242, "learning_rate": 0.0001, "elbo": 3.53381, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.848228946751114}
{"step": 65625, "loss": 3.539488983154297, "grad_norm": 0.6636199951171875, "learning_rate": 0.0001, "elbo": 3.53949, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8485221674876846}
{"step": 65630, "loss": 3.5264297485351563, "grad_norm": 0.8712741136550903, "learning_rate": 0.0001, "elbo": 3.52643, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.848815388224255}
{"step": 65635, "loss": 3.525948333740234, "grad_norm": 0.6832417845726013, "learning_rate": 0.0001, "elbo": 3.52595, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.849108608960826}
{"step": 65640, "loss": 3.5325668334960936, "grad_norm": 0.6924120783805847, "learning_rate": 0.0001, "elbo": 3.53257, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8494018296973964}
{"step": 65645, "loss": 3.530649185180664, "grad_norm": 0.7037412524223328, "learning_rate": 0.0001, "elbo": 3.53065, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.849695050433967}
{"step": 65650, "loss": 3.5671974182128907, "grad_norm": 0.7086496353149414, "learning_rate": 0.0001, "elbo": 3.5672, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8499882711705373}
{"step": 65655, "loss": 3.523295593261719, "grad_norm": 0.7436656355857849, "learning_rate": 0.0001, "elbo": 3.5233, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.850281491907108}
{"step": 65660, "loss": 3.537419891357422, "grad_norm": 0.7205966114997864, "learning_rate": 0.0001, "elbo": 3.53742, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8505747126436782}
{"step": 65665, "loss": 3.535900115966797, "grad_norm": 0.7540314197540283, "learning_rate": 0.0001, "elbo": 3.5359, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8508679333802487}
{"step": 65670, "loss": 3.533613586425781, "grad_norm": 0.7242107391357422, "learning_rate": 0.0001, "elbo": 3.53361, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.851161154116819}
{"step": 65675, "loss": 3.5257450103759767, "grad_norm": 0.6691657304763794, "learning_rate": 0.0001, "elbo": 3.52574, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8514543748533896}
{"step": 65680, "loss": 3.534958267211914, "grad_norm": 1.2700893878936768, "learning_rate": 0.0001, "elbo": 3.53496, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.85174759558996}
{"step": 65685, "loss": 3.5379104614257812, "grad_norm": 0.7944965362548828, "learning_rate": 0.0001, "elbo": 3.53791, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8520408163265305}
{"step": 65690, "loss": 3.534185028076172, "grad_norm": 0.6737174391746521, "learning_rate": 0.0001, "elbo": 3.53419, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.852334037063101}
{"step": 65695, "loss": 3.526060104370117, "grad_norm": 0.6970821022987366, "learning_rate": 0.0001, "elbo": 3.52606, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8526272577996714}
{"step": 65700, "loss": 3.5422882080078124, "grad_norm": 0.7025783658027649, "learning_rate": 0.0001, "elbo": 3.54229, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.852920478536242}
{"step": 65705, "loss": 3.5371856689453125, "grad_norm": 0.8271937966346741, "learning_rate": 0.0001, "elbo": 3.53719, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8532136992728123}
{"step": 65710, "loss": 3.5337547302246093, "grad_norm": 1.0035502910614014, "learning_rate": 0.0001, "elbo": 3.53375, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.853506920009383}
{"step": 65715, "loss": 3.5419681549072264, "grad_norm": 0.6899223923683167, "learning_rate": 0.0001, "elbo": 3.54197, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8538001407459537}
{"step": 65720, "loss": 3.544261169433594, "grad_norm": 0.8152384161949158, "learning_rate": 0.0001, "elbo": 3.54426, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.854093361482524}
{"step": 65725, "loss": 3.541440963745117, "grad_norm": 0.7020266056060791, "learning_rate": 0.0001, "elbo": 3.54144, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8543865822190946}
{"step": 65730, "loss": 3.5276042938232424, "grad_norm": 0.6789050102233887, "learning_rate": 0.0001, "elbo": 3.5276, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.854679802955665}
{"step": 65735, "loss": 3.540226364135742, "grad_norm": 0.6781801581382751, "learning_rate": 0.0001, "elbo": 3.54023, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8549730236922355}
{"step": 65740, "loss": 3.535664749145508, "grad_norm": 0.7202655076980591, "learning_rate": 0.0001, "elbo": 3.53566, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.855266244428806}
{"step": 65745, "loss": 3.5435035705566404, "grad_norm": 0.7690470218658447, "learning_rate": 0.0001, "elbo": 3.5435, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8555594651653764}
{"step": 65750, "loss": 3.5221477508544923, "grad_norm": 0.7652839422225952, "learning_rate": 0.0001, "elbo": 3.52215, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.855852685901947}
{"step": 65750, "loss": 3.5221477508544923, "grad_norm": 0.7652839422225952, "learning_rate": 0.0001, "elbo": 3.52215, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.855852685901947}
{"step": 65755, "loss": 3.5352020263671875, "grad_norm": 0.8121801614761353, "learning_rate": 0.0001, "elbo": 3.5352, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8561459066385178}
{"step": 65760, "loss": 3.5341835021972656, "grad_norm": 0.7882749438285828, "learning_rate": 0.0001, "elbo": 3.53418, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.856439127375088}
{"step": 65765, "loss": 3.5548683166503907, "grad_norm": 0.6935861110687256, "learning_rate": 0.0001, "elbo": 3.55487, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8567323481116587}
{"step": 65770, "loss": 3.5294116973876952, "grad_norm": 0.7235329747200012, "learning_rate": 0.0001, "elbo": 3.52941, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.857025568848229}
{"step": 65775, "loss": 3.539250946044922, "grad_norm": 0.6756983995437622, "learning_rate": 0.0001, "elbo": 3.53925, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8573187895847996}
{"step": 65780, "loss": 3.5344703674316404, "grad_norm": 0.684215784072876, "learning_rate": 0.0001, "elbo": 3.53447, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.85761201032137}
{"step": 65785, "loss": 3.5380035400390626, "grad_norm": 0.7425819635391235, "learning_rate": 0.0001, "elbo": 3.538, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8579052310579405}
{"step": 65790, "loss": 3.544801712036133, "grad_norm": 0.8887625336647034, "learning_rate": 0.0001, "elbo": 3.5448, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.858198451794511}
{"step": 65795, "loss": 3.538095474243164, "grad_norm": 0.6851227879524231, "learning_rate": 0.0001, "elbo": 3.5381, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8584916725310814}
{"step": 65800, "loss": 3.5401351928710936, "grad_norm": 0.6881622076034546, "learning_rate": 0.0001, "elbo": 3.54013, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.858784893267652}
{"step": 65805, "loss": 3.534148406982422, "grad_norm": 1.0488638877868652, "learning_rate": 0.0001, "elbo": 3.53415, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8590781140042223}
{"step": 65810, "loss": 3.5248138427734377, "grad_norm": 0.70234215259552, "learning_rate": 0.0001, "elbo": 3.52481, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8593713347407927}
{"step": 65815, "loss": 3.5192493438720702, "grad_norm": 0.8015759587287903, "learning_rate": 0.0001, "elbo": 3.51925, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.859664555477363}
{"step": 65820, "loss": 3.5429664611816407, "grad_norm": 0.8040105104446411, "learning_rate": 0.0001, "elbo": 3.54297, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8599577762139337}
{"step": 65825, "loss": 3.530437469482422, "grad_norm": 0.7395426034927368, "learning_rate": 0.0001, "elbo": 3.53044, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.860250996950504}
{"step": 65830, "loss": 3.5531517028808595, "grad_norm": 0.7366990447044373, "learning_rate": 0.0001, "elbo": 3.55315, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8605442176870746}
{"step": 65835, "loss": 3.5453857421875, "grad_norm": 0.7017528414726257, "learning_rate": 0.0001, "elbo": 3.54539, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8608374384236455}
{"step": 65840, "loss": 3.5245948791503907, "grad_norm": 0.6972848773002625, "learning_rate": 0.0001, "elbo": 3.52459, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.861130659160216}
{"step": 65845, "loss": 3.5196266174316406, "grad_norm": 0.7379311323165894, "learning_rate": 0.0001, "elbo": 3.51963, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8614238798967864}
{"step": 65850, "loss": 3.5401531219482423, "grad_norm": 0.6894329786300659, "learning_rate": 0.0001, "elbo": 3.54015, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.861717100633357}
{"step": 65855, "loss": 3.540966796875, "grad_norm": 0.795364499092102, "learning_rate": 0.0001, "elbo": 3.54097, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8620103213699273}
{"step": 65860, "loss": 3.5285903930664064, "grad_norm": 0.9071423411369324, "learning_rate": 0.0001, "elbo": 3.52859, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8623035421064977}
{"step": 65865, "loss": 3.5422409057617186, "grad_norm": 0.7111104130744934, "learning_rate": 0.0001, "elbo": 3.54224, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.862596762843068}
{"step": 65870, "loss": 3.5162631988525392, "grad_norm": 0.8720347881317139, "learning_rate": 0.0001, "elbo": 3.51626, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8628899835796386}
{"step": 65875, "loss": 3.5321983337402343, "grad_norm": 0.6977188587188721, "learning_rate": 0.0001, "elbo": 3.5322, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.863183204316209}
{"step": 65880, "loss": 3.53505859375, "grad_norm": 0.7074283957481384, "learning_rate": 0.0001, "elbo": 3.53506, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.86347642505278}
{"step": 65885, "loss": 3.5386756896972655, "grad_norm": 0.7761239409446716, "learning_rate": 0.0001, "elbo": 3.53868, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8637696457893504}
{"step": 65890, "loss": 3.5318103790283204, "grad_norm": 0.6750353574752808, "learning_rate": 0.0001, "elbo": 3.53181, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.864062866525921}
{"step": 65895, "loss": 3.533211898803711, "grad_norm": 0.7078261971473694, "learning_rate": 0.0001, "elbo": 3.53321, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8643560872624914}
{"step": 65900, "loss": 3.537540817260742, "grad_norm": 0.6839458346366882, "learning_rate": 0.0001, "elbo": 3.53754, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.864649307999062}
{"step": 65905, "loss": 3.527251434326172, "grad_norm": 0.7652115821838379, "learning_rate": 0.0001, "elbo": 3.52725, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8649425287356323}
{"step": 65910, "loss": 3.5359565734863283, "grad_norm": 0.713119626045227, "learning_rate": 0.0001, "elbo": 3.53596, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8652357494722027}
{"step": 65915, "loss": 3.5224960327148436, "grad_norm": 0.9127316474914551, "learning_rate": 0.0001, "elbo": 3.5225, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.865528970208773}
{"step": 65920, "loss": 3.5176513671875, "grad_norm": 0.6814295649528503, "learning_rate": 0.0001, "elbo": 3.51765, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8658221909453436}
{"step": 65920, "loss": 3.5176513671875, "grad_norm": 0.6814295649528503, "learning_rate": 0.0001, "elbo": 3.51765, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8658221909453436}
{"step": 65925, "loss": 3.5416519165039064, "grad_norm": 0.7646814584732056, "learning_rate": 0.0001, "elbo": 3.54165, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.866115411681914}
{"step": 65930, "loss": 3.521653747558594, "grad_norm": 0.7816916704177856, "learning_rate": 0.0001, "elbo": 3.52165, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8664086324184845}
{"step": 65935, "loss": 3.525440216064453, "grad_norm": 0.6675489544868469, "learning_rate": 0.0001, "elbo": 3.52544, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.866701853155055}
{"step": 65935, "loss": 3.525440216064453, "grad_norm": 0.6675489544868469, "learning_rate": 0.0001, "elbo": 3.52544, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.866701853155055}
{"step": 65940, "loss": 3.539231872558594, "grad_norm": 0.6811489462852478, "learning_rate": 0.0001, "elbo": 3.53923, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8669950738916254}
{"step": 65945, "loss": 3.552301788330078, "grad_norm": 0.7430552244186401, "learning_rate": 0.0001, "elbo": 3.5523, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.867288294628196}
{"step": 65950, "loss": 3.5250732421875, "grad_norm": 0.6443474292755127, "learning_rate": 0.0001, "elbo": 3.52507, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8675815153647664}
{"step": 65955, "loss": 3.5330574035644533, "grad_norm": 0.6799867153167725, "learning_rate": 0.0001, "elbo": 3.53306, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8678747361013373}
{"step": 65960, "loss": 3.5463653564453126, "grad_norm": 1.0602481365203857, "learning_rate": 0.0001, "elbo": 3.54637, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8681679568379077}
{"step": 65965, "loss": 3.5231842041015624, "grad_norm": 0.7137045860290527, "learning_rate": 0.0001, "elbo": 3.52318, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.868461177574478}
{"step": 65965, "loss": 3.5231842041015624, "grad_norm": 0.7137045860290527, "learning_rate": 0.0001, "elbo": 3.52318, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.868461177574478}
{"step": 65970, "loss": 3.545981216430664, "grad_norm": 0.7284444570541382, "learning_rate": 0.0001, "elbo": 3.54598, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8687543983110486}
{"step": 65975, "loss": 3.5416709899902346, "grad_norm": 0.7187010049819946, "learning_rate": 0.0001, "elbo": 3.54167, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.869047619047619}
{"step": 65980, "loss": 3.527410125732422, "grad_norm": 0.7723988890647888, "learning_rate": 0.0001, "elbo": 3.52741, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8693408397841895}
{"step": 65985, "loss": 3.5280723571777344, "grad_norm": 0.6922793984413147, "learning_rate": 0.0001, "elbo": 3.52807, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.86963406052076}
{"step": 65990, "loss": 3.5345287322998047, "grad_norm": 0.6712702512741089, "learning_rate": 0.0001, "elbo": 3.53453, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8699272812573304}
{"step": 65995, "loss": 3.527852249145508, "grad_norm": 0.689990222454071, "learning_rate": 0.0001, "elbo": 3.52785, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.870220501993901}
{"step": 66000, "loss": 3.5217544555664064, "grad_norm": 0.6513160467147827, "learning_rate": 0.0001, "elbo": 3.52175, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8705137227304713}
{"step": 66005, "loss": 3.5320247650146483, "grad_norm": 0.7566707134246826, "learning_rate": 0.0001, "elbo": 3.53202, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8708069434670422}
{"step": 66010, "loss": 3.514618682861328, "grad_norm": 0.7713261246681213, "learning_rate": 0.0001, "elbo": 3.51462, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8711001642036127}
{"step": 66015, "loss": 3.542075347900391, "grad_norm": 0.7141388058662415, "learning_rate": 0.0001, "elbo": 3.54208, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.871393384940183}
{"step": 66020, "loss": 3.519661712646484, "grad_norm": 0.6654677391052246, "learning_rate": 0.0001, "elbo": 3.51966, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8716866056767536}
{"step": 66025, "loss": 3.5350883483886717, "grad_norm": 0.7998248934745789, "learning_rate": 0.0001, "elbo": 3.53509, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.871979826413324}
{"step": 66030, "loss": 3.5336837768554688, "grad_norm": 0.6842971444129944, "learning_rate": 0.0001, "elbo": 3.53368, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8722730471498945}
{"step": 66035, "loss": 3.5566856384277346, "grad_norm": 0.7215332984924316, "learning_rate": 0.0001, "elbo": 3.55669, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.872566267886465}
{"step": 66040, "loss": 3.531385040283203, "grad_norm": 0.9976816773414612, "learning_rate": 0.0001, "elbo": 3.53139, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8728594886230354}
{"step": 66045, "loss": 3.5351829528808594, "grad_norm": 0.8279722332954407, "learning_rate": 0.0001, "elbo": 3.53518, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.873152709359606}
{"step": 66050, "loss": 3.5227394104003906, "grad_norm": 1.0224264860153198, "learning_rate": 0.0001, "elbo": 3.52274, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8734459300961763}
{"step": 66055, "loss": 3.541673278808594, "grad_norm": 0.7486671805381775, "learning_rate": 0.0001, "elbo": 3.54167, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.873739150832747}
{"step": 66060, "loss": 3.5465904235839845, "grad_norm": 0.7535833716392517, "learning_rate": 0.0001, "elbo": 3.54659, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8740323715693172}
{"step": 66065, "loss": 3.532360076904297, "grad_norm": 0.7928683757781982, "learning_rate": 0.0001, "elbo": 3.53236, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8743255923058877}
{"step": 66070, "loss": 3.5337684631347654, "grad_norm": 0.7756752371788025, "learning_rate": 0.0001, "elbo": 3.53377, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.874618813042458}
{"step": 66075, "loss": 3.5225509643554687, "grad_norm": 0.7410898804664612, "learning_rate": 0.0001, "elbo": 3.52255, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8749120337790286}
{"step": 66080, "loss": 3.543248748779297, "grad_norm": 0.7743664979934692, "learning_rate": 0.0001, "elbo": 3.54325, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8752052545155995}
{"step": 66085, "loss": 3.5234657287597657, "grad_norm": 0.733053982257843, "learning_rate": 0.0001, "elbo": 3.52347, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.87549847525217}
{"step": 66090, "loss": 3.5305686950683595, "grad_norm": 0.671268105506897, "learning_rate": 0.0001, "elbo": 3.53057, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8757916959887404}
{"step": 66090, "loss": 3.5305686950683595, "grad_norm": 0.671268105506897, "learning_rate": 0.0001, "elbo": 3.53057, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8757916959887404}
{"step": 66095, "loss": 3.53001708984375, "grad_norm": 0.7027599811553955, "learning_rate": 0.0001, "elbo": 3.53002, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.876084916725311}
{"step": 66095, "loss": 3.53001708984375, "grad_norm": 0.7027599811553955, "learning_rate": 0.0001, "elbo": 3.53002, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.876084916725311}
{"step": 66100, "loss": 3.5487960815429687, "grad_norm": 0.827621579170227, "learning_rate": 0.0001, "elbo": 3.5488, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8763781374618813}
{"step": 66105, "loss": 3.5308204650878907, "grad_norm": 0.7276654243469238, "learning_rate": 0.0001, "elbo": 3.53082, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8766713581984518}
{"step": 66110, "loss": 3.527790069580078, "grad_norm": 0.7191690802574158, "learning_rate": 0.0001, "elbo": 3.52779, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.876964578935022}
{"step": 66115, "loss": 3.542281723022461, "grad_norm": 0.6937997341156006, "learning_rate": 0.0001, "elbo": 3.54228, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8772577996715927}
{"step": 66120, "loss": 3.5376518249511717, "grad_norm": 0.7007031440734863, "learning_rate": 0.0001, "elbo": 3.53765, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.877551020408163}
{"step": 66125, "loss": 3.525798797607422, "grad_norm": 0.7086317539215088, "learning_rate": 0.0001, "elbo": 3.5258, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.877844241144734}
{"step": 66130, "loss": 3.5360931396484374, "grad_norm": 0.7269660830497742, "learning_rate": 0.0001, "elbo": 3.53609, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8781374618813045}
{"step": 66135, "loss": 3.528372955322266, "grad_norm": 0.7135431170463562, "learning_rate": 0.0001, "elbo": 3.52837, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.878430682617875}
{"step": 66140, "loss": 3.5220985412597656, "grad_norm": 0.7309295535087585, "learning_rate": 0.0001, "elbo": 3.5221, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8787239033544454}
{"step": 66145, "loss": 3.533181381225586, "grad_norm": 0.7740094661712646, "learning_rate": 0.0001, "elbo": 3.53318, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.879017124091016}
{"step": 66150, "loss": 3.524779510498047, "grad_norm": 0.6869810223579407, "learning_rate": 0.0001, "elbo": 3.52478, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8793103448275863}
{"step": 66155, "loss": 3.524062728881836, "grad_norm": 0.6976109743118286, "learning_rate": 0.0001, "elbo": 3.52406, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8796035655641568}
{"step": 66160, "loss": 3.5443141937255858, "grad_norm": 0.7369714379310608, "learning_rate": 0.0001, "elbo": 3.54431, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.879896786300727}
{"step": 66165, "loss": 3.5466964721679686, "grad_norm": 0.6872313618659973, "learning_rate": 0.0001, "elbo": 3.5467, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8801900070372977}
{"step": 66170, "loss": 3.5407447814941406, "grad_norm": 0.884069561958313, "learning_rate": 0.0001, "elbo": 3.54075, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.880483227773868}
{"step": 66175, "loss": 3.5355712890625, "grad_norm": 0.7220389246940613, "learning_rate": 0.0001, "elbo": 3.53557, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8807764485104386}
{"step": 66180, "loss": 3.5344383239746096, "grad_norm": 0.7663004994392395, "learning_rate": 0.0001, "elbo": 3.53444, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.881069669247009}
{"step": 66185, "loss": 3.543000411987305, "grad_norm": 0.8936104774475098, "learning_rate": 0.0001, "elbo": 3.543, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8813628899835795}
{"step": 66190, "loss": 3.540594482421875, "grad_norm": 0.7029989957809448, "learning_rate": 0.0001, "elbo": 3.54059, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.88165611072015}
{"step": 66195, "loss": 3.5262161254882813, "grad_norm": 0.6996025443077087, "learning_rate": 0.0001, "elbo": 3.52622, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8819493314567204}
{"step": 66200, "loss": 3.544842529296875, "grad_norm": 0.7023802399635315, "learning_rate": 0.0001, "elbo": 3.54484, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.882242552193291}
{"step": 66205, "loss": 3.530295562744141, "grad_norm": 0.6651259660720825, "learning_rate": 0.0001, "elbo": 3.5303, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8825357729298617}
{"step": 66210, "loss": 3.5439308166503904, "grad_norm": 0.821500301361084, "learning_rate": 0.0001, "elbo": 3.54393, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.882828993666432}
{"step": 66215, "loss": 3.535846710205078, "grad_norm": 0.9066175222396851, "learning_rate": 0.0001, "elbo": 3.53585, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8831222144030026}
{"step": 66220, "loss": 3.5348506927490235, "grad_norm": 0.7165675163269043, "learning_rate": 0.0001, "elbo": 3.53485, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.883415435139573}
{"step": 66225, "loss": 3.5223922729492188, "grad_norm": 0.7569131851196289, "learning_rate": 0.0001, "elbo": 3.52239, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8837086558761436}
{"step": 66230, "loss": 3.5350715637207033, "grad_norm": 0.926369845867157, "learning_rate": 0.0001, "elbo": 3.53507, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.884001876612714}
{"step": 66235, "loss": 3.5288009643554688, "grad_norm": 0.7370875477790833, "learning_rate": 0.0001, "elbo": 3.5288, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8842950973492845}
{"step": 66240, "loss": 3.548484039306641, "grad_norm": 0.9628193378448486, "learning_rate": 0.0001, "elbo": 3.54848, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.884588318085855}
{"step": 66245, "loss": 3.540716552734375, "grad_norm": 0.7110203504562378, "learning_rate": 0.0001, "elbo": 3.54072, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8848815388224254}
{"step": 66250, "loss": 3.5184417724609376, "grad_norm": 0.6776650547981262, "learning_rate": 0.0001, "elbo": 3.51844, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8851747595589963}
{"step": 66255, "loss": 3.5498016357421873, "grad_norm": 0.6559662818908691, "learning_rate": 0.0001, "elbo": 3.5498, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8854679802955667}
{"step": 66260, "loss": 3.532684326171875, "grad_norm": 0.8742985725402832, "learning_rate": 0.0001, "elbo": 3.53268, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.885761201032137}
{"step": 66265, "loss": 3.536981201171875, "grad_norm": 0.8676958680152893, "learning_rate": 0.0001, "elbo": 3.53698, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8860544217687076}
{"step": 66270, "loss": 3.543171691894531, "grad_norm": 0.710322916507721, "learning_rate": 0.0001, "elbo": 3.54317, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.886347642505278}
{"step": 66275, "loss": 3.5380420684814453, "grad_norm": 0.6772940158843994, "learning_rate": 0.0001, "elbo": 3.53804, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8866408632418485}
{"step": 66280, "loss": 3.5445625305175783, "grad_norm": 0.7045649290084839, "learning_rate": 0.0001, "elbo": 3.54456, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.886934083978419}
{"step": 66285, "loss": 3.5562381744384766, "grad_norm": 0.7033419013023376, "learning_rate": 0.0001, "elbo": 3.55624, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8872273047149895}
{"step": 66290, "loss": 3.5365760803222654, "grad_norm": 0.7486846446990967, "learning_rate": 0.0001, "elbo": 3.53658, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.88752052545156}
{"step": 66295, "loss": 3.540974426269531, "grad_norm": 0.7709804773330688, "learning_rate": 0.0001, "elbo": 3.54097, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8878137461881304}
{"step": 66300, "loss": 3.540949249267578, "grad_norm": 0.675638735294342, "learning_rate": 0.0001, "elbo": 3.54095, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.888106966924701}
{"step": 66305, "loss": 3.536895751953125, "grad_norm": 0.7069257497787476, "learning_rate": 0.0001, "elbo": 3.5369, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8884001876612713}
{"step": 66310, "loss": 3.5225643157958983, "grad_norm": 0.7040923237800598, "learning_rate": 0.0001, "elbo": 3.52256, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8886934083978417}
{"step": 66315, "loss": 3.5230762481689455, "grad_norm": 0.7382227778434753, "learning_rate": 0.0001, "elbo": 3.52308, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.888986629134412}
{"step": 66320, "loss": 3.5316131591796873, "grad_norm": 0.7247005105018616, "learning_rate": 0.0001, "elbo": 3.53161, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8892798498709826}
{"step": 66325, "loss": 3.5317657470703123, "grad_norm": 0.6856099963188171, "learning_rate": 0.0001, "elbo": 3.53177, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8895730706075535}
{"step": 66330, "loss": 3.518865203857422, "grad_norm": 0.8882044553756714, "learning_rate": 0.0001, "elbo": 3.51887, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.889866291344124}
{"step": 66335, "loss": 3.521922302246094, "grad_norm": 0.656951367855072, "learning_rate": 0.0001, "elbo": 3.52192, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8901595120806944}
{"step": 66340, "loss": 3.539112091064453, "grad_norm": 0.6403961181640625, "learning_rate": 0.0001, "elbo": 3.53911, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.890452732817265}
{"step": 66345, "loss": 3.5247520446777343, "grad_norm": 0.776032567024231, "learning_rate": 0.0001, "elbo": 3.52475, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8907459535538353}
{"step": 66350, "loss": 3.538176727294922, "grad_norm": 0.6776276230812073, "learning_rate": 0.0001, "elbo": 3.53818, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.891039174290406}
{"step": 66355, "loss": 3.5435325622558596, "grad_norm": 0.7098913788795471, "learning_rate": 0.0001, "elbo": 3.54353, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8913323950269763}
{"step": 66360, "loss": 3.5400733947753906, "grad_norm": 0.6528884172439575, "learning_rate": 0.0001, "elbo": 3.54007, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8916256157635467}
{"step": 66360, "loss": 3.5400733947753906, "grad_norm": 0.6528884172439575, "learning_rate": 0.0001, "elbo": 3.54007, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8916256157635467}
{"step": 66365, "loss": 3.5368064880371093, "grad_norm": 0.8440459370613098, "learning_rate": 0.0001, "elbo": 3.53681, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.891918836500117}
{"step": 66370, "loss": 3.5400321960449217, "grad_norm": 0.8502891063690186, "learning_rate": 0.0001, "elbo": 3.54003, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8922120572366876}
{"step": 66375, "loss": 3.5295005798339845, "grad_norm": 0.7164618372917175, "learning_rate": 0.0001, "elbo": 3.5295, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8925052779732585}
{"step": 66380, "loss": 3.537572479248047, "grad_norm": 0.7553449869155884, "learning_rate": 0.0001, "elbo": 3.53757, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.892798498709829}
{"step": 66385, "loss": 3.537760543823242, "grad_norm": 1.0114574432373047, "learning_rate": 0.0001, "elbo": 3.53776, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8930917194463994}
{"step": 66390, "loss": 3.5317543029785154, "grad_norm": 0.9572991728782654, "learning_rate": 0.0001, "elbo": 3.53175, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.89338494018297}
{"step": 66390, "loss": 3.5317543029785154, "grad_norm": 0.9572991728782654, "learning_rate": 0.0001, "elbo": 3.53175, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.89338494018297}
{"step": 66395, "loss": 3.5422378540039063, "grad_norm": 0.7846444249153137, "learning_rate": 0.0001, "elbo": 3.54224, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8936781609195403}
{"step": 66400, "loss": 3.531377410888672, "grad_norm": 0.6955051422119141, "learning_rate": 0.0001, "elbo": 3.53138, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.893971381656111}
{"step": 66405, "loss": 3.529732513427734, "grad_norm": 0.7059365510940552, "learning_rate": 0.0001, "elbo": 3.52973, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8942646023926812}
{"step": 66410, "loss": 3.5192291259765627, "grad_norm": 0.710852324962616, "learning_rate": 0.0001, "elbo": 3.51923, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8945578231292517}
{"step": 66415, "loss": 3.5403480529785156, "grad_norm": 0.8007206320762634, "learning_rate": 0.0001, "elbo": 3.54035, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.894851043865822}
{"step": 66420, "loss": 3.522958755493164, "grad_norm": 0.7238007187843323, "learning_rate": 0.0001, "elbo": 3.52296, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8951442646023926}
{"step": 66425, "loss": 3.5251079559326173, "grad_norm": 0.6903444528579712, "learning_rate": 0.0001, "elbo": 3.52511, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.895437485338963}
{"step": 66430, "loss": 3.5243534088134765, "grad_norm": 0.7808787226676941, "learning_rate": 0.0001, "elbo": 3.52435, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8957307060755335}
{"step": 66435, "loss": 3.5309158325195313, "grad_norm": 0.9281153082847595, "learning_rate": 0.0001, "elbo": 3.53092, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.896023926812104}
{"step": 66440, "loss": 3.523912811279297, "grad_norm": 0.7113375663757324, "learning_rate": 0.0001, "elbo": 3.52391, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8963171475486744}
{"step": 66445, "loss": 3.508553314208984, "grad_norm": 0.8801913261413574, "learning_rate": 0.0001, "elbo": 3.50855, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.896610368285245}
{"step": 66450, "loss": 3.556793212890625, "grad_norm": 1.028404951095581, "learning_rate": 0.0001, "elbo": 3.55679, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8969035890218158}
{"step": 66455, "loss": 3.5292678833007813, "grad_norm": 0.7163583636283875, "learning_rate": 0.0001, "elbo": 3.52927, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8971968097583862}
{"step": 66460, "loss": 3.5334625244140625, "grad_norm": 0.6975409984588623, "learning_rate": 0.0001, "elbo": 3.53346, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8974900304949567}
{"step": 66465, "loss": 3.543359375, "grad_norm": 0.7114014029502869, "learning_rate": 0.0001, "elbo": 3.54336, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.897783251231527}
{"step": 66470, "loss": 3.5310997009277343, "grad_norm": 0.6958719491958618, "learning_rate": 0.0001, "elbo": 3.5311, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8980764719680976}
{"step": 66475, "loss": 3.5302886962890625, "grad_norm": 0.6583641171455383, "learning_rate": 0.0001, "elbo": 3.53029, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.898369692704668}
{"step": 66480, "loss": 3.549894332885742, "grad_norm": 0.8104987144470215, "learning_rate": 0.0001, "elbo": 3.54989, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8986629134412385}
{"step": 66485, "loss": 3.525341033935547, "grad_norm": 0.7140715718269348, "learning_rate": 0.0001, "elbo": 3.52534, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.898956134177809}
{"step": 66490, "loss": 3.540660858154297, "grad_norm": 0.753134548664093, "learning_rate": 0.0001, "elbo": 3.54066, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8992493549143794}
{"step": 66495, "loss": 3.5338871002197267, "grad_norm": 0.7587047815322876, "learning_rate": 0.0001, "elbo": 3.53389, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8995425756509503}
{"step": 66500, "loss": 3.5161052703857423, "grad_norm": 0.6802947521209717, "learning_rate": 0.0001, "elbo": 3.51611, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.8998357963875208}
{"step": 66505, "loss": 3.525016021728516, "grad_norm": 0.7471619248390198, "learning_rate": 0.0001, "elbo": 3.52502, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.900129017124091}
{"step": 66510, "loss": 3.526418685913086, "grad_norm": 0.7192603945732117, "learning_rate": 0.0001, "elbo": 3.52642, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9004222378606617}
{"step": 66515, "loss": 3.5272445678710938, "grad_norm": 0.7070393562316895, "learning_rate": 0.0001, "elbo": 3.52724, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.900715458597232}
{"step": 66520, "loss": 3.5274627685546873, "grad_norm": 0.7675148844718933, "learning_rate": 0.0001, "elbo": 3.52746, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9010086793338026}
{"step": 66525, "loss": 3.524795150756836, "grad_norm": 0.6722388863563538, "learning_rate": 0.0001, "elbo": 3.52479, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.901301900070373}
{"step": 66530, "loss": 3.528142547607422, "grad_norm": 0.8953087329864502, "learning_rate": 0.0001, "elbo": 3.52814, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9015951208069435}
{"step": 66535, "loss": 3.5316234588623048, "grad_norm": 0.7109457850456238, "learning_rate": 0.0001, "elbo": 3.53162, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.901888341543514}
{"step": 66540, "loss": 3.5237453460693358, "grad_norm": 0.8473291397094727, "learning_rate": 0.0001, "elbo": 3.52375, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9021815622800844}
{"step": 66545, "loss": 3.528064727783203, "grad_norm": 0.7282488346099854, "learning_rate": 0.0001, "elbo": 3.52806, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.902474783016655}
{"step": 66550, "loss": 3.5442771911621094, "grad_norm": 0.6742932796478271, "learning_rate": 0.0001, "elbo": 3.54428, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9027680037532253}
{"step": 66555, "loss": 3.5394969940185548, "grad_norm": 0.6954500079154968, "learning_rate": 0.0001, "elbo": 3.5395, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9030612244897958}
{"step": 66560, "loss": 3.538343811035156, "grad_norm": 0.820564329624176, "learning_rate": 0.0001, "elbo": 3.53834, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.903354445226366}
{"step": 66565, "loss": 3.5235565185546873, "grad_norm": 0.7011953592300415, "learning_rate": 0.0001, "elbo": 3.52356, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9036476659629367}
{"step": 66570, "loss": 3.523599624633789, "grad_norm": 0.7093856334686279, "learning_rate": 0.0001, "elbo": 3.5236, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.903940886699507}
{"step": 66575, "loss": 3.517882537841797, "grad_norm": 0.7149736285209656, "learning_rate": 0.0001, "elbo": 3.51788, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.904234107436078}
{"step": 66580, "loss": 3.5489593505859376, "grad_norm": 0.6774686574935913, "learning_rate": 0.0001, "elbo": 3.54896, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9045273281726485}
{"step": 66585, "loss": 3.526236724853516, "grad_norm": 0.8335371017456055, "learning_rate": 0.0001, "elbo": 3.52624, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.904820548909219}
{"step": 66590, "loss": 3.5267276763916016, "grad_norm": 0.7381018400192261, "learning_rate": 0.0001, "elbo": 3.52673, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9051137696457894}
{"step": 66595, "loss": 3.5268436431884767, "grad_norm": 0.7499582767486572, "learning_rate": 0.0001, "elbo": 3.52684, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.90540699038236}
{"step": 66600, "loss": 3.537479019165039, "grad_norm": 0.6909469962120056, "learning_rate": 0.0001, "elbo": 3.53748, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9057002111189303}
{"step": 66605, "loss": 3.524347686767578, "grad_norm": 0.9684601426124573, "learning_rate": 0.0001, "elbo": 3.52435, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9059934318555007}
{"step": 66610, "loss": 3.5275779724121095, "grad_norm": 0.8076096773147583, "learning_rate": 0.0001, "elbo": 3.52758, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.906286652592071}
{"step": 66615, "loss": 3.535061264038086, "grad_norm": 0.6996058225631714, "learning_rate": 0.0001, "elbo": 3.53506, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9065798733286416}
{"step": 66620, "loss": 3.525885009765625, "grad_norm": 0.7521900534629822, "learning_rate": 0.0001, "elbo": 3.52589, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9068730940652125}
{"step": 66625, "loss": 3.521797943115234, "grad_norm": 0.882498025894165, "learning_rate": 0.0001, "elbo": 3.5218, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.907166314801783}
{"step": 66630, "loss": 3.534381866455078, "grad_norm": 0.7180343270301819, "learning_rate": 0.0001, "elbo": 3.53438, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9074595355383535}
{"step": 66635, "loss": 3.5373867034912108, "grad_norm": 0.8567283749580383, "learning_rate": 0.0001, "elbo": 3.53739, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.907752756274924}
{"step": 66640, "loss": 3.552830123901367, "grad_norm": 0.7173887491226196, "learning_rate": 0.0001, "elbo": 3.55283, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9080459770114944}
{"step": 66645, "loss": 3.5490341186523438, "grad_norm": 0.7823272347450256, "learning_rate": 0.0001, "elbo": 3.54903, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.908339197748065}
{"step": 66650, "loss": 3.5412223815917967, "grad_norm": 0.781472384929657, "learning_rate": 0.0001, "elbo": 3.54122, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9086324184846353}
{"step": 66655, "loss": 3.542125701904297, "grad_norm": 0.7087079286575317, "learning_rate": 0.0001, "elbo": 3.54213, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9089256392212057}
{"step": 66660, "loss": 3.5403404235839844, "grad_norm": 0.7017649412155151, "learning_rate": 0.0001, "elbo": 3.54034, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.909218859957776}
{"step": 66665, "loss": 3.537683868408203, "grad_norm": 0.7250874638557434, "learning_rate": 0.0001, "elbo": 3.53768, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9095120806943466}
{"step": 66670, "loss": 3.528429412841797, "grad_norm": 0.6708617806434631, "learning_rate": 0.0001, "elbo": 3.52843, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.909805301430917}
{"step": 66675, "loss": 3.5430198669433595, "grad_norm": 0.7469051480293274, "learning_rate": 0.0001, "elbo": 3.54302, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9100985221674875}
{"step": 66680, "loss": 3.531706619262695, "grad_norm": 0.8831713199615479, "learning_rate": 0.0001, "elbo": 3.53171, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.910391742904058}
{"step": 66680, "loss": 3.531706619262695, "grad_norm": 0.8831713199615479, "learning_rate": 0.0001, "elbo": 3.53171, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.910391742904058}
{"step": 66685, "loss": 3.533141326904297, "grad_norm": 0.7022347450256348, "learning_rate": 0.0001, "elbo": 3.53314, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9106849636406285}
{"step": 66690, "loss": 3.5316864013671876, "grad_norm": 0.7075362205505371, "learning_rate": 0.0001, "elbo": 3.53169, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.910978184377199}
{"step": 66695, "loss": 3.5256683349609377, "grad_norm": 0.9158682227134705, "learning_rate": 0.0001, "elbo": 3.52567, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.91127140511377}
{"step": 66700, "loss": 3.5198135375976562, "grad_norm": 0.656760573387146, "learning_rate": 0.0001, "elbo": 3.51981, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9115646258503403}
{"step": 66705, "loss": 3.530841064453125, "grad_norm": 0.7564519643783569, "learning_rate": 0.0001, "elbo": 3.53084, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9118578465869107}
{"step": 66710, "loss": 3.531512451171875, "grad_norm": 0.7102305889129639, "learning_rate": 0.0001, "elbo": 3.53151, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.912151067323481}
{"step": 66715, "loss": 3.5494083404541015, "grad_norm": 1.0872184038162231, "learning_rate": 0.0001, "elbo": 3.54941, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9124442880600516}
{"step": 66720, "loss": 3.533244323730469, "grad_norm": 0.7014857530593872, "learning_rate": 0.0001, "elbo": 3.53324, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.912737508796622}
{"step": 66725, "loss": 3.5361869812011717, "grad_norm": 0.7803862690925598, "learning_rate": 0.0001, "elbo": 3.53619, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9130307295331925}
{"step": 66730, "loss": 3.5376655578613283, "grad_norm": 0.7681678533554077, "learning_rate": 0.0001, "elbo": 3.53767, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.913323950269763}
{"step": 66735, "loss": 3.5335247039794924, "grad_norm": 0.6705787777900696, "learning_rate": 0.0001, "elbo": 3.53352, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9136171710063334}
{"step": 66740, "loss": 3.5336002349853515, "grad_norm": 0.7331548929214478, "learning_rate": 0.0001, "elbo": 3.5336, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.913910391742904}
{"step": 66745, "loss": 3.5272499084472657, "grad_norm": 0.7074489593505859, "learning_rate": 0.0001, "elbo": 3.52725, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.914203612479475}
{"step": 66750, "loss": 3.5305442810058594, "grad_norm": 0.6538788080215454, "learning_rate": 0.0001, "elbo": 3.53054, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9144968332160452}
{"step": 66755, "loss": 3.5342769622802734, "grad_norm": 0.6898793578147888, "learning_rate": 0.0001, "elbo": 3.53428, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9147900539526157}
{"step": 66760, "loss": 3.535512924194336, "grad_norm": 1.3635762929916382, "learning_rate": 0.0001, "elbo": 3.53551, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.915083274689186}
{"step": 66765, "loss": 3.5268505096435545, "grad_norm": 0.7126175761222839, "learning_rate": 0.0001, "elbo": 3.52685, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9153764954257566}
{"step": 66770, "loss": 3.5338680267333986, "grad_norm": 0.7208251953125, "learning_rate": 0.0001, "elbo": 3.53387, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.915669716162327}
{"step": 66775, "loss": 3.5192886352539063, "grad_norm": 0.7180764079093933, "learning_rate": 0.0001, "elbo": 3.51929, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9159629368988975}
{"step": 66780, "loss": 3.527302551269531, "grad_norm": 0.7368962168693542, "learning_rate": 0.0001, "elbo": 3.5273, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.916256157635468}
{"step": 66785, "loss": 3.5478057861328125, "grad_norm": 0.6840592622756958, "learning_rate": 0.0001, "elbo": 3.54781, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9165493783720384}
{"step": 66790, "loss": 3.5425693511962892, "grad_norm": 0.7520357966423035, "learning_rate": 0.0001, "elbo": 3.54257, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.916842599108609}
{"step": 66795, "loss": 3.532168960571289, "grad_norm": 0.8161503076553345, "learning_rate": 0.0001, "elbo": 3.53217, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9171358198451793}
{"step": 66800, "loss": 3.5281093597412108, "grad_norm": 0.7417991161346436, "learning_rate": 0.0001, "elbo": 3.52811, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.91742904058175}
{"step": 66805, "loss": 3.529893493652344, "grad_norm": 0.6835423111915588, "learning_rate": 0.0001, "elbo": 3.52989, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9177222613183202}
{"step": 66810, "loss": 3.5287544250488283, "grad_norm": 0.6934860348701477, "learning_rate": 0.0001, "elbo": 3.52875, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9180154820548907}
{"step": 66815, "loss": 3.515065002441406, "grad_norm": 0.730879545211792, "learning_rate": 0.0001, "elbo": 3.51507, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.918308702791461}
{"step": 66820, "loss": 3.5475692749023438, "grad_norm": 0.71302330493927, "learning_rate": 0.0001, "elbo": 3.54757, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.918601923528032}
{"step": 66825, "loss": 3.540449523925781, "grad_norm": 0.6893100142478943, "learning_rate": 0.0001, "elbo": 3.54045, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9188951442646025}
{"step": 66830, "loss": 3.5298797607421877, "grad_norm": 0.768714189529419, "learning_rate": 0.0001, "elbo": 3.52988, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.919188365001173}
{"step": 66835, "loss": 3.537404251098633, "grad_norm": 0.728012204170227, "learning_rate": 0.0001, "elbo": 3.5374, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9194815857377434}
{"step": 66840, "loss": 3.5368942260742187, "grad_norm": 0.7063949704170227, "learning_rate": 0.0001, "elbo": 3.53689, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.919774806474314}
{"step": 66845, "loss": 3.5185287475585936, "grad_norm": 0.6606659293174744, "learning_rate": 0.0001, "elbo": 3.51853, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9200680272108843}
{"step": 66850, "loss": 3.534321594238281, "grad_norm": 0.665773868560791, "learning_rate": 0.0001, "elbo": 3.53432, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9203612479474548}
{"step": 66855, "loss": 3.5496952056884767, "grad_norm": 0.9972841143608093, "learning_rate": 0.0001, "elbo": 3.5497, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9206544686840252}
{"step": 66860, "loss": 3.536376190185547, "grad_norm": 0.6764012575149536, "learning_rate": 0.0001, "elbo": 3.53638, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9209476894205957}
{"step": 66860, "loss": 3.536376190185547, "grad_norm": 0.6764012575149536, "learning_rate": 0.0001, "elbo": 3.53638, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9209476894205957}
{"step": 66865, "loss": 3.5456260681152343, "grad_norm": 0.6589949131011963, "learning_rate": 0.0001, "elbo": 3.54563, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9212409101571666}
{"step": 66870, "loss": 3.5236042022705076, "grad_norm": 0.7037134170532227, "learning_rate": 0.0001, "elbo": 3.5236, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.921534130893737}
{"step": 66870, "loss": 3.5236042022705076, "grad_norm": 0.7037134170532227, "learning_rate": 0.0001, "elbo": 3.5236, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.921534130893737}
{"step": 66875, "loss": 3.5211166381835937, "grad_norm": 0.6930006146430969, "learning_rate": 0.0001, "elbo": 3.52112, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9218273516303075}
{"step": 66880, "loss": 3.5264625549316406, "grad_norm": 0.7896579504013062, "learning_rate": 0.0001, "elbo": 3.52646, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.922120572366878}
{"step": 66885, "loss": 3.5385726928710937, "grad_norm": 0.7300685048103333, "learning_rate": 0.0001, "elbo": 3.53857, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9224137931034484}
{"step": 66890, "loss": 3.5313838958740233, "grad_norm": 0.6996926069259644, "learning_rate": 0.0001, "elbo": 3.53138, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.922707013840019}
{"step": 66895, "loss": 3.5419540405273438, "grad_norm": 0.7277916073799133, "learning_rate": 0.0001, "elbo": 3.54195, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9230002345765893}
{"step": 66900, "loss": 3.5314441680908204, "grad_norm": 0.7288804650306702, "learning_rate": 0.0001, "elbo": 3.53144, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9232934553131598}
{"step": 66905, "loss": 3.5349651336669923, "grad_norm": 0.7278265953063965, "learning_rate": 0.0001, "elbo": 3.53497, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.92358667604973}
{"step": 66910, "loss": 3.5445877075195313, "grad_norm": 0.6928666830062866, "learning_rate": 0.0001, "elbo": 3.54459, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9238798967863007}
{"step": 66915, "loss": 3.550687789916992, "grad_norm": 0.68609619140625, "learning_rate": 0.0001, "elbo": 3.55069, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.924173117522871}
{"step": 66920, "loss": 3.522868347167969, "grad_norm": 0.7874521613121033, "learning_rate": 0.0001, "elbo": 3.52287, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9244663382594416}
{"step": 66925, "loss": 3.530303192138672, "grad_norm": 0.6997737884521484, "learning_rate": 0.0001, "elbo": 3.5303, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.924759558996012}
{"step": 66930, "loss": 3.5224700927734376, "grad_norm": 0.7310830950737, "learning_rate": 0.0001, "elbo": 3.52247, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9250527797325825}
{"step": 66935, "loss": 3.524597930908203, "grad_norm": 0.7175084352493286, "learning_rate": 0.0001, "elbo": 3.5246, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.925346000469153}
{"step": 66940, "loss": 3.528983306884766, "grad_norm": 0.7276287078857422, "learning_rate": 0.0001, "elbo": 3.52898, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9256392212057234}
{"step": 66945, "loss": 3.5396995544433594, "grad_norm": 0.7541543841362, "learning_rate": 0.0001, "elbo": 3.5397, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9259324419422943}
{"step": 66950, "loss": 3.537000274658203, "grad_norm": 0.7837680578231812, "learning_rate": 0.0001, "elbo": 3.537, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9262256626788647}
{"step": 66955, "loss": 3.5285377502441406, "grad_norm": 0.8429615497589111, "learning_rate": 0.0001, "elbo": 3.52854, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.926518883415435}
{"step": 66960, "loss": 3.5289047241210936, "grad_norm": 0.733250081539154, "learning_rate": 0.0001, "elbo": 3.5289, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9268121041520057}
{"step": 66960, "loss": 3.5289047241210936, "grad_norm": 0.733250081539154, "learning_rate": 0.0001, "elbo": 3.5289, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9268121041520057}
{"step": 66965, "loss": 3.5496841430664063, "grad_norm": 0.7543813586235046, "learning_rate": 0.0001, "elbo": 3.54968, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.927105324888576}
{"step": 66970, "loss": 3.5276504516601563, "grad_norm": 0.7252792716026306, "learning_rate": 0.0001, "elbo": 3.52765, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9273985456251466}
{"step": 66975, "loss": 3.5371475219726562, "grad_norm": 0.689039409160614, "learning_rate": 0.0001, "elbo": 3.53715, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.927691766361717}
{"step": 66980, "loss": 3.514250946044922, "grad_norm": 0.7126486897468567, "learning_rate": 0.0001, "elbo": 3.51425, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9279849870982875}
{"step": 66980, "loss": 3.514250946044922, "grad_norm": 0.7126486897468567, "learning_rate": 0.0001, "elbo": 3.51425, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9279849870982875}
{"step": 66985, "loss": 3.528788757324219, "grad_norm": 0.7085966467857361, "learning_rate": 0.0001, "elbo": 3.52879, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.928278207834858}
{"step": 66990, "loss": 3.5316020965576174, "grad_norm": 0.7202442288398743, "learning_rate": 0.0001, "elbo": 3.5316, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.928571428571429}
{"step": 66995, "loss": 3.537969207763672, "grad_norm": 0.6986455917358398, "learning_rate": 0.0001, "elbo": 3.53797, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9288646493079993}
{"step": 67000, "loss": 3.5160243988037108, "grad_norm": 0.6848418116569519, "learning_rate": 0.0001, "elbo": 3.51602, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9291578700445697}
{"step": 67000, "loss": 3.5160243988037108, "grad_norm": 0.6848418116569519, "learning_rate": 0.0001, "elbo": 3.51602, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9291578700445697}
{"step": 67005, "loss": 3.535146713256836, "grad_norm": 0.6719955801963806, "learning_rate": 0.0001, "elbo": 3.53515, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.92945109078114}
{"step": 67010, "loss": 3.533155059814453, "grad_norm": 0.6767012476921082, "learning_rate": 0.0001, "elbo": 3.53316, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9297443115177106}
{"step": 67015, "loss": 3.521661376953125, "grad_norm": 0.7202010154724121, "learning_rate": 0.0001, "elbo": 3.52166, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.930037532254281}
{"step": 67020, "loss": 3.5213966369628906, "grad_norm": 0.7652978301048279, "learning_rate": 0.0001, "elbo": 3.5214, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9303307529908516}
{"step": 67025, "loss": 3.5215003967285154, "grad_norm": 0.7236722707748413, "learning_rate": 0.0001, "elbo": 3.5215, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.930623973727422}
{"step": 67030, "loss": 3.538638687133789, "grad_norm": 0.6818348169326782, "learning_rate": 0.0001, "elbo": 3.53864, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9309171944639925}
{"step": 67035, "loss": 3.5118507385253905, "grad_norm": 0.7400518655776978, "learning_rate": 0.0001, "elbo": 3.51185, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.931210415200563}
{"step": 67040, "loss": 3.5323326110839846, "grad_norm": 0.7760882377624512, "learning_rate": 0.0001, "elbo": 3.53233, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9315036359371334}
{"step": 67045, "loss": 3.526134490966797, "grad_norm": 0.7940205931663513, "learning_rate": 0.0001, "elbo": 3.52613, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.931796856673704}
{"step": 67050, "loss": 3.5468822479248048, "grad_norm": 0.6722761392593384, "learning_rate": 0.0001, "elbo": 3.54688, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9320900774102743}
{"step": 67055, "loss": 3.5481151580810546, "grad_norm": 1.2989692687988281, "learning_rate": 0.0001, "elbo": 3.54812, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9323832981468447}
{"step": 67060, "loss": 3.5442562103271484, "grad_norm": 0.6935951113700867, "learning_rate": 0.0001, "elbo": 3.54426, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.932676518883415}
{"step": 67065, "loss": 3.5436355590820314, "grad_norm": 0.7212878465652466, "learning_rate": 0.0001, "elbo": 3.54364, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.932969739619986}
{"step": 67070, "loss": 3.525297164916992, "grad_norm": 0.6971125602722168, "learning_rate": 0.0001, "elbo": 3.5253, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9332629603565565}
{"step": 67075, "loss": 3.5180328369140623, "grad_norm": 0.7809571027755737, "learning_rate": 0.0001, "elbo": 3.51803, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.933556181093127}
{"step": 67080, "loss": 3.5173717498779298, "grad_norm": 0.7263521552085876, "learning_rate": 0.0001, "elbo": 3.51737, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9338494018296974}
{"step": 67085, "loss": 3.540789794921875, "grad_norm": 0.6880214810371399, "learning_rate": 0.0001, "elbo": 3.54079, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.934142622566268}
{"step": 67090, "loss": 3.5341590881347655, "grad_norm": 0.6719668507575989, "learning_rate": 0.0001, "elbo": 3.53416, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9344358433028384}
{"step": 67095, "loss": 3.5269264221191405, "grad_norm": 0.7268670201301575, "learning_rate": 0.0001, "elbo": 3.52693, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.934729064039409}
{"step": 67100, "loss": 3.5397377014160156, "grad_norm": 0.7069598436355591, "learning_rate": 0.0001, "elbo": 3.53974, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9350222847759793}
{"step": 67105, "loss": 3.5262100219726564, "grad_norm": 0.6614602208137512, "learning_rate": 0.0001, "elbo": 3.52621, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9353155055125497}
{"step": 67110, "loss": 3.5161853790283204, "grad_norm": 0.6508194208145142, "learning_rate": 0.0001, "elbo": 3.51619, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9356087262491206}
{"step": 67115, "loss": 3.542970657348633, "grad_norm": 0.9169372320175171, "learning_rate": 0.0001, "elbo": 3.54297, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.935901946985691}
{"step": 67115, "loss": 3.542970657348633, "grad_norm": 0.9169372320175171, "learning_rate": 0.0001, "elbo": 3.54297, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.935901946985691}
{"step": 67120, "loss": 3.5222755432128907, "grad_norm": 0.7065184712409973, "learning_rate": 0.0001, "elbo": 3.52228, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9361951677222615}
{"step": 67125, "loss": 3.5375930786132814, "grad_norm": 0.6670771837234497, "learning_rate": 0.0001, "elbo": 3.53759, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.936488388458832}
{"step": 67130, "loss": 3.5256019592285157, "grad_norm": 0.6989396214485168, "learning_rate": 0.0001, "elbo": 3.5256, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9367816091954024}
{"step": 67135, "loss": 3.523981475830078, "grad_norm": 0.6889670491218567, "learning_rate": 0.0001, "elbo": 3.52398, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.937074829931973}
{"step": 67140, "loss": 3.5415931701660157, "grad_norm": 0.6494815349578857, "learning_rate": 0.0001, "elbo": 3.54159, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9373680506685433}
{"step": 67145, "loss": 3.537467193603516, "grad_norm": 0.6504238247871399, "learning_rate": 0.0001, "elbo": 3.53747, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.937661271405114}
{"step": 67150, "loss": 3.548747253417969, "grad_norm": 0.7166029810905457, "learning_rate": 0.0001, "elbo": 3.54875, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9379544921416842}
{"step": 67155, "loss": 3.5434051513671876, "grad_norm": 0.7275719046592712, "learning_rate": 0.0001, "elbo": 3.5434, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9382477128782547}
{"step": 67160, "loss": 3.5220573425292967, "grad_norm": 0.6931394338607788, "learning_rate": 0.0001, "elbo": 3.52206, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.938540933614825}
{"step": 67165, "loss": 3.5506355285644533, "grad_norm": 0.6934859752655029, "learning_rate": 0.0001, "elbo": 3.55064, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9388341543513956}
{"step": 67170, "loss": 3.538924789428711, "grad_norm": 0.7704156637191772, "learning_rate": 0.0001, "elbo": 3.53892, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.939127375087966}
{"step": 67175, "loss": 3.5357017517089844, "grad_norm": 0.688503086566925, "learning_rate": 0.0001, "elbo": 3.5357, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9394205958245365}
{"step": 67180, "loss": 3.5321327209472657, "grad_norm": 0.6838008761405945, "learning_rate": 0.0001, "elbo": 3.53213, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.939713816561107}
{"step": 67185, "loss": 3.51678466796875, "grad_norm": 0.7442471385002136, "learning_rate": 0.0001, "elbo": 3.51678, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9400070372976774}
{"step": 67190, "loss": 3.5312156677246094, "grad_norm": 0.7289016842842102, "learning_rate": 0.0001, "elbo": 3.53122, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9403002580342483}
{"step": 67195, "loss": 3.5266448974609377, "grad_norm": 0.6856465339660645, "learning_rate": 0.0001, "elbo": 3.52664, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.940593478770819}
{"step": 67200, "loss": 3.5176971435546873, "grad_norm": 0.7363020777702332, "learning_rate": 0.0001, "elbo": 3.5177, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9408866995073892}
{"step": 67205, "loss": 3.5267494201660154, "grad_norm": 0.8519018888473511, "learning_rate": 0.0001, "elbo": 3.52675, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9411799202439597}
{"step": 67205, "loss": 3.5267494201660154, "grad_norm": 0.8519018888473511, "learning_rate": 0.0001, "elbo": 3.52675, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9411799202439597}
{"step": 67210, "loss": 3.530431365966797, "grad_norm": 1.0619471073150635, "learning_rate": 0.0001, "elbo": 3.53043, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.94147314098053}
{"step": 67210, "loss": 3.530431365966797, "grad_norm": 1.0619471073150635, "learning_rate": 0.0001, "elbo": 3.53043, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.94147314098053}
{"step": 67215, "loss": 3.5374645233154296, "grad_norm": 0.9081307053565979, "learning_rate": 0.0001, "elbo": 3.53746, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9417663617171006}
{"step": 67220, "loss": 3.546588897705078, "grad_norm": 0.8085101842880249, "learning_rate": 0.0001, "elbo": 3.54659, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.942059582453671}
{"step": 67225, "loss": 3.5365272521972657, "grad_norm": 0.7058022022247314, "learning_rate": 0.0001, "elbo": 3.53653, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9423528031902415}
{"step": 67230, "loss": 3.55090217590332, "grad_norm": 0.6896950602531433, "learning_rate": 0.0001, "elbo": 3.5509, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.942646023926812}
{"step": 67230, "loss": 3.55090217590332, "grad_norm": 0.6896950602531433, "learning_rate": 0.0001, "elbo": 3.5509, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.942646023926812}
{"step": 67235, "loss": 3.521856689453125, "grad_norm": 0.7068308591842651, "learning_rate": 0.0001, "elbo": 3.52186, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.942939244663383}
{"step": 67240, "loss": 3.515873336791992, "grad_norm": 0.7199627161026001, "learning_rate": 0.0001, "elbo": 3.51587, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9432324653999533}
{"step": 67245, "loss": 3.5286022186279298, "grad_norm": 0.6825968623161316, "learning_rate": 0.0001, "elbo": 3.5286, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9435256861365238}
{"step": 67250, "loss": 3.511476516723633, "grad_norm": 0.7232487201690674, "learning_rate": 0.0001, "elbo": 3.51148, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.943818906873094}
{"step": 67255, "loss": 3.5372238159179688, "grad_norm": 0.7722967267036438, "learning_rate": 0.0001, "elbo": 3.53722, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9441121276096647}
{"step": 67260, "loss": 3.526502227783203, "grad_norm": 0.6762264966964722, "learning_rate": 0.0001, "elbo": 3.5265, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.944405348346235}
{"step": 67265, "loss": 3.5229137420654295, "grad_norm": 0.727021336555481, "learning_rate": 0.0001, "elbo": 3.52291, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9446985690828056}
{"step": 67270, "loss": 3.534423828125, "grad_norm": 0.7143157124519348, "learning_rate": 0.0001, "elbo": 3.53442, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.944991789819376}
{"step": 67275, "loss": 3.539933776855469, "grad_norm": 0.7920246124267578, "learning_rate": 0.0001, "elbo": 3.53993, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9452850105559465}
{"step": 67280, "loss": 3.5223323822021486, "grad_norm": 0.7316409945487976, "learning_rate": 0.0001, "elbo": 3.52233, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.945578231292517}
{"step": 67285, "loss": 3.5349693298339844, "grad_norm": 0.8209580183029175, "learning_rate": 0.0001, "elbo": 3.53497, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9458714520290874}
{"step": 67290, "loss": 3.537256622314453, "grad_norm": 0.9108660221099854, "learning_rate": 0.0001, "elbo": 3.53726, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.946164672765658}
{"step": 67295, "loss": 3.522289276123047, "grad_norm": 0.6749763488769531, "learning_rate": 0.0001, "elbo": 3.52229, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9464578935022283}
{"step": 67300, "loss": 3.525461959838867, "grad_norm": 0.7454687356948853, "learning_rate": 0.0001, "elbo": 3.52546, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9467511142387988}
{"step": 67305, "loss": 3.534527587890625, "grad_norm": 0.7027873396873474, "learning_rate": 0.0001, "elbo": 3.53453, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.947044334975369}
{"step": 67310, "loss": 3.53359375, "grad_norm": 0.7084314227104187, "learning_rate": 0.0001, "elbo": 3.53359, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9473375557119397}
{"step": 67315, "loss": 3.535540771484375, "grad_norm": 1.1938310861587524, "learning_rate": 0.0001, "elbo": 3.53554, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9476307764485106}
{"step": 67320, "loss": 3.541094207763672, "grad_norm": 0.7035397887229919, "learning_rate": 0.0001, "elbo": 3.54109, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.947923997185081}
{"step": 67325, "loss": 3.5482078552246095, "grad_norm": 0.711767852306366, "learning_rate": 0.0001, "elbo": 3.54821, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9482172179216515}
{"step": 67330, "loss": 3.525887298583984, "grad_norm": 0.7303645610809326, "learning_rate": 0.0001, "elbo": 3.52589, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.948510438658222}
{"step": 67335, "loss": 3.540648651123047, "grad_norm": 0.7949559092521667, "learning_rate": 0.0001, "elbo": 3.54065, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9488036593947924}
{"step": 67340, "loss": 3.5290607452392577, "grad_norm": 0.7342726588249207, "learning_rate": 0.0001, "elbo": 3.52906, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.949096880131363}
{"step": 67345, "loss": 3.5367828369140626, "grad_norm": 0.6999111175537109, "learning_rate": 0.0001, "elbo": 3.53678, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9493901008679333}
{"step": 67350, "loss": 3.5454872131347654, "grad_norm": 0.7041749358177185, "learning_rate": 0.0001, "elbo": 3.54549, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9496833216045037}
{"step": 67355, "loss": 3.527782440185547, "grad_norm": 0.727705717086792, "learning_rate": 0.0001, "elbo": 3.52778, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.949976542341074}
{"step": 67360, "loss": 3.516173553466797, "grad_norm": 0.7027808427810669, "learning_rate": 0.0001, "elbo": 3.51617, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.950269763077645}
{"step": 67365, "loss": 3.5374263763427733, "grad_norm": 0.7099030017852783, "learning_rate": 0.0001, "elbo": 3.53743, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9505629838142156}
{"step": 67370, "loss": 3.537298583984375, "grad_norm": 0.6875413656234741, "learning_rate": 0.0001, "elbo": 3.5373, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.950856204550786}
{"step": 67375, "loss": 3.529680633544922, "grad_norm": 0.703315794467926, "learning_rate": 0.0001, "elbo": 3.52968, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9511494252873565}
{"step": 67380, "loss": 3.531641387939453, "grad_norm": 0.7155434489250183, "learning_rate": 0.0001, "elbo": 3.53164, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.951442646023927}
{"step": 67385, "loss": 3.5302215576171876, "grad_norm": 0.9232339859008789, "learning_rate": 0.0001, "elbo": 3.53022, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9517358667604974}
{"step": 67390, "loss": 3.528477096557617, "grad_norm": 0.6648262739181519, "learning_rate": 0.0001, "elbo": 3.52848, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.952029087497068}
{"step": 67395, "loss": 3.5226184844970705, "grad_norm": 0.6878048777580261, "learning_rate": 0.0001, "elbo": 3.52262, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9523223082336383}
{"step": 67400, "loss": 3.5408340454101563, "grad_norm": 0.7431942820549011, "learning_rate": 0.0001, "elbo": 3.54083, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9526155289702087}
{"step": 67405, "loss": 3.5475257873535155, "grad_norm": 0.7096605896949768, "learning_rate": 0.0001, "elbo": 3.54753, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.952908749706779}
{"step": 67410, "loss": 3.5176406860351563, "grad_norm": 0.8122241497039795, "learning_rate": 0.0001, "elbo": 3.51764, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9532019704433496}
{"step": 67415, "loss": 3.544957733154297, "grad_norm": 0.747931718826294, "learning_rate": 0.0001, "elbo": 3.54496, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.95349519117992}
{"step": 67420, "loss": 3.5233177185058593, "grad_norm": 0.7021058797836304, "learning_rate": 0.0001, "elbo": 3.52332, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9537884119164906}
{"step": 67425, "loss": 3.528154754638672, "grad_norm": 0.8889223337173462, "learning_rate": 0.0001, "elbo": 3.52815, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.954081632653061}
{"step": 67430, "loss": 3.533739471435547, "grad_norm": 0.6867729425430298, "learning_rate": 0.0001, "elbo": 3.53374, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9543748533896315}
{"step": 67435, "loss": 3.5334510803222656, "grad_norm": 0.6958186030387878, "learning_rate": 0.0001, "elbo": 3.53345, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9546680741262024}
{"step": 67440, "loss": 3.522807312011719, "grad_norm": 0.6931484341621399, "learning_rate": 0.0001, "elbo": 3.52281, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.954961294862773}
{"step": 67445, "loss": 3.5259471893310548, "grad_norm": 0.8698386549949646, "learning_rate": 0.0001, "elbo": 3.52595, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9552545155993433}
{"step": 67450, "loss": 3.527220916748047, "grad_norm": 0.7169552445411682, "learning_rate": 0.0001, "elbo": 3.52722, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9555477363359137}
{"step": 67455, "loss": 3.527922821044922, "grad_norm": 0.7353943586349487, "learning_rate": 0.0001, "elbo": 3.52792, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.955840957072484}
{"step": 67460, "loss": 3.5165481567382812, "grad_norm": 0.6815414428710938, "learning_rate": 0.0001, "elbo": 3.51655, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9561341778090546}
{"step": 67465, "loss": 3.5251060485839845, "grad_norm": 0.7034084796905518, "learning_rate": 0.0001, "elbo": 3.52511, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.956427398545625}
{"step": 67470, "loss": 3.5162734985351562, "grad_norm": 0.854590654373169, "learning_rate": 0.0001, "elbo": 3.51627, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9567206192821955}
{"step": 67475, "loss": 3.520519256591797, "grad_norm": 0.7179608941078186, "learning_rate": 0.0001, "elbo": 3.52052, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.957013840018766}
{"step": 67480, "loss": 3.535328674316406, "grad_norm": 0.7084949016571045, "learning_rate": 0.0001, "elbo": 3.53533, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.957307060755337}
{"step": 67485, "loss": 3.5129207611083983, "grad_norm": 0.7598932981491089, "learning_rate": 0.0001, "elbo": 3.51292, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9576002814919073}
{"step": 67490, "loss": 3.5312263488769533, "grad_norm": 0.7082149386405945, "learning_rate": 0.0001, "elbo": 3.53123, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.957893502228478}
{"step": 67495, "loss": 3.537738037109375, "grad_norm": 0.703334391117096, "learning_rate": 0.0001, "elbo": 3.53774, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9581867229650483}
{"step": 67500, "loss": 3.5177379608154298, "grad_norm": 0.7582656145095825, "learning_rate": 0.0001, "elbo": 3.51774, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9584799437016187}
{"step": 67505, "loss": 3.539310836791992, "grad_norm": 0.678042471408844, "learning_rate": 0.0001, "elbo": 3.53931, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.958773164438189}
{"step": 67510, "loss": 3.5311065673828126, "grad_norm": 0.6746205687522888, "learning_rate": 0.0001, "elbo": 3.53111, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9590663851747596}
{"step": 67515, "loss": 3.539784240722656, "grad_norm": 0.7883338928222656, "learning_rate": 0.0001, "elbo": 3.53978, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.95935960591133}
{"step": 67520, "loss": 3.5513240814208986, "grad_norm": 0.6946893334388733, "learning_rate": 0.0001, "elbo": 3.55132, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9596528266479005}
{"step": 67525, "loss": 3.5203392028808596, "grad_norm": 0.7133524417877197, "learning_rate": 0.0001, "elbo": 3.52034, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.959946047384471}
{"step": 67530, "loss": 3.545146942138672, "grad_norm": 0.7153357267379761, "learning_rate": 0.0001, "elbo": 3.54515, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9602392681210414}
{"step": 67535, "loss": 3.5331825256347655, "grad_norm": 0.7230713963508606, "learning_rate": 0.0001, "elbo": 3.53318, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.960532488857612}
{"step": 67540, "loss": 3.5240127563476564, "grad_norm": 0.6489564180374146, "learning_rate": 0.0001, "elbo": 3.52401, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9608257095941823}
{"step": 67545, "loss": 3.5394737243652346, "grad_norm": 0.7006896138191223, "learning_rate": 0.0001, "elbo": 3.53947, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.961118930330753}
{"step": 67550, "loss": 3.524156188964844, "grad_norm": 0.6978446841239929, "learning_rate": 0.0001, "elbo": 3.52416, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9614121510673233}
{"step": 67555, "loss": 3.5249168395996096, "grad_norm": 0.719049334526062, "learning_rate": 0.0001, "elbo": 3.52492, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9617053718038937}
{"step": 67560, "loss": 3.533417510986328, "grad_norm": 0.6823652982711792, "learning_rate": 0.0001, "elbo": 3.53342, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9619985925404646}
{"step": 67565, "loss": 3.5301513671875, "grad_norm": 0.7230926752090454, "learning_rate": 0.0001, "elbo": 3.53015, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.962291813277035}
{"step": 67570, "loss": 3.5256675720214843, "grad_norm": 0.7088972926139832, "learning_rate": 0.0001, "elbo": 3.52567, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9625850340136055}
{"step": 67575, "loss": 3.5207965850830076, "grad_norm": 0.6960496306419373, "learning_rate": 0.0001, "elbo": 3.5208, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.962878254750176}
{"step": 67580, "loss": 3.5373844146728515, "grad_norm": 0.6852053999900818, "learning_rate": 0.0001, "elbo": 3.53738, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9631714754867464}
{"step": 67585, "loss": 3.5371307373046874, "grad_norm": 0.6736066937446594, "learning_rate": 0.0001, "elbo": 3.53713, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.963464696223317}
{"step": 67590, "loss": 3.5159374237060548, "grad_norm": 0.7268811464309692, "learning_rate": 0.0001, "elbo": 3.51594, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9637579169598873}
{"step": 67595, "loss": 3.5324562072753904, "grad_norm": 0.7463269233703613, "learning_rate": 0.0001, "elbo": 3.53246, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.964051137696458}
{"step": 67600, "loss": 3.5351482391357423, "grad_norm": 0.6532989144325256, "learning_rate": 0.0001, "elbo": 3.53515, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9643443584330282}
{"step": 67600, "loss": 3.5351482391357423, "grad_norm": 0.6532989144325256, "learning_rate": 0.0001, "elbo": 3.53515, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9643443584330282}
{"step": 67605, "loss": 3.523891067504883, "grad_norm": 0.8279296159744263, "learning_rate": 0.0001, "elbo": 3.52389, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.964637579169599}
{"step": 67610, "loss": 3.5329814910888673, "grad_norm": 0.801986038684845, "learning_rate": 0.0001, "elbo": 3.53298, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9649307999061696}
{"step": 67615, "loss": 3.545781707763672, "grad_norm": 0.9542908668518066, "learning_rate": 0.0001, "elbo": 3.54578, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.96522402064274}
{"step": 67620, "loss": 3.530561828613281, "grad_norm": 0.7062807083129883, "learning_rate": 0.0001, "elbo": 3.53056, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9655172413793105}
{"step": 67625, "loss": 3.5313533782958983, "grad_norm": 0.6981046795845032, "learning_rate": 0.0001, "elbo": 3.53135, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.965810462115881}
{"step": 67630, "loss": 3.518670654296875, "grad_norm": 0.7568649053573608, "learning_rate": 0.0001, "elbo": 3.51867, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9661036828524514}
{"step": 67635, "loss": 3.525285339355469, "grad_norm": 0.7619077563285828, "learning_rate": 0.0001, "elbo": 3.52529, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.966396903589022}
{"step": 67640, "loss": 3.537609100341797, "grad_norm": 0.7044460773468018, "learning_rate": 0.0001, "elbo": 3.53761, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9666901243255923}
{"step": 67645, "loss": 3.5316703796386717, "grad_norm": 0.7031469345092773, "learning_rate": 0.0001, "elbo": 3.53167, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9669833450621628}
{"step": 67650, "loss": 3.5360084533691407, "grad_norm": 0.6689953804016113, "learning_rate": 0.0001, "elbo": 3.53601, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9672765657987332}
{"step": 67650, "loss": 3.5360084533691407, "grad_norm": 0.6689953804016113, "learning_rate": 0.0001, "elbo": 3.53601, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9672765657987332}
{"step": 67655, "loss": 3.5385292053222654, "grad_norm": 0.7868643403053284, "learning_rate": 0.0001, "elbo": 3.53853, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9675697865353037}
{"step": 67655, "loss": 3.5385292053222654, "grad_norm": 0.7868643403053284, "learning_rate": 0.0001, "elbo": 3.53853, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9675697865353037}
{"step": 67660, "loss": 3.5441959381103514, "grad_norm": 0.7158452272415161, "learning_rate": 0.0001, "elbo": 3.5442, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.967863007271874}
{"step": 67665, "loss": 3.54107666015625, "grad_norm": 0.7140789031982422, "learning_rate": 0.0001, "elbo": 3.54108, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9681562280084446}
{"step": 67670, "loss": 3.537541961669922, "grad_norm": 0.7389650940895081, "learning_rate": 0.0001, "elbo": 3.53754, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.968449448745015}
{"step": 67675, "loss": 3.5472747802734377, "grad_norm": 0.7609999179840088, "learning_rate": 0.0001, "elbo": 3.54727, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9687426694815855}
{"step": 67680, "loss": 3.5201717376708985, "grad_norm": 0.680425763130188, "learning_rate": 0.0001, "elbo": 3.52017, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9690358902181564}
{"step": 67685, "loss": 3.535025405883789, "grad_norm": 0.698214590549469, "learning_rate": 0.0001, "elbo": 3.53503, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.969329110954727}
{"step": 67690, "loss": 3.534490966796875, "grad_norm": 0.6890137791633606, "learning_rate": 0.0001, "elbo": 3.53449, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9696223316912973}
{"step": 67695, "loss": 3.525403594970703, "grad_norm": 0.6812366247177124, "learning_rate": 0.0001, "elbo": 3.5254, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9699155524278678}
{"step": 67700, "loss": 3.528997039794922, "grad_norm": 0.7005388140678406, "learning_rate": 0.0001, "elbo": 3.529, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.970208773164438}
{"step": 67705, "loss": 3.5238380432128906, "grad_norm": 0.6749547719955444, "learning_rate": 0.0001, "elbo": 3.52384, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9705019939010087}
{"step": 67710, "loss": 3.5101978302001955, "grad_norm": 0.6847259998321533, "learning_rate": 0.0001, "elbo": 3.5102, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.970795214637579}
{"step": 67715, "loss": 3.526658630371094, "grad_norm": 0.7239387631416321, "learning_rate": 0.0001, "elbo": 3.52666, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9710884353741496}
{"step": 67720, "loss": 3.515978240966797, "grad_norm": 0.6834794282913208, "learning_rate": 0.0001, "elbo": 3.51598, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.97138165611072}
{"step": 67725, "loss": 3.5341087341308595, "grad_norm": 0.7282978892326355, "learning_rate": 0.0001, "elbo": 3.53411, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9716748768472905}
{"step": 67730, "loss": 3.541240692138672, "grad_norm": 0.7072240114212036, "learning_rate": 0.0001, "elbo": 3.54124, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9719680975838614}
{"step": 67735, "loss": 3.512493133544922, "grad_norm": 0.9647208452224731, "learning_rate": 0.0001, "elbo": 3.51249, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.972261318320432}
{"step": 67740, "loss": 3.5252960205078123, "grad_norm": 0.6874728202819824, "learning_rate": 0.0001, "elbo": 3.5253, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9725545390570023}
{"step": 67745, "loss": 3.5540714263916016, "grad_norm": 0.715774655342102, "learning_rate": 0.0001, "elbo": 3.55407, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9728477597935727}
{"step": 67750, "loss": 3.518366241455078, "grad_norm": 0.7653635144233704, "learning_rate": 0.0001, "elbo": 3.51837, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.973140980530143}
{"step": 67755, "loss": 3.5246456146240233, "grad_norm": 0.7885869145393372, "learning_rate": 0.0001, "elbo": 3.52465, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9734342012667137}
{"step": 67760, "loss": 3.529995346069336, "grad_norm": 0.8268393874168396, "learning_rate": 0.0001, "elbo": 3.53, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.973727422003284}
{"step": 67765, "loss": 3.5350715637207033, "grad_norm": 1.0238795280456543, "learning_rate": 0.0001, "elbo": 3.53507, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9740206427398546}
{"step": 67770, "loss": 3.5376922607421877, "grad_norm": 0.6681895852088928, "learning_rate": 0.0001, "elbo": 3.53769, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.974313863476425}
{"step": 67775, "loss": 3.5341156005859373, "grad_norm": 0.6839442849159241, "learning_rate": 0.0001, "elbo": 3.53412, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9746070842129955}
{"step": 67780, "loss": 3.522734832763672, "grad_norm": 0.7116854786872864, "learning_rate": 0.0001, "elbo": 3.52273, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.974900304949566}
{"step": 67785, "loss": 3.5418769836425783, "grad_norm": 0.7241150140762329, "learning_rate": 0.0001, "elbo": 3.54188, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9751935256861364}
{"step": 67790, "loss": 3.525725555419922, "grad_norm": 0.6652106046676636, "learning_rate": 0.0001, "elbo": 3.52573, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.975486746422707}
{"step": 67795, "loss": 3.515766143798828, "grad_norm": 0.6835373640060425, "learning_rate": 0.0001, "elbo": 3.51577, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9757799671592773}
{"step": 67800, "loss": 3.545763397216797, "grad_norm": 0.6736641526222229, "learning_rate": 0.0001, "elbo": 3.54576, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9760731878958477}
{"step": 67805, "loss": 3.5135269165039062, "grad_norm": 0.7178554534912109, "learning_rate": 0.0001, "elbo": 3.51353, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9763664086324186}
{"step": 67810, "loss": 3.513046646118164, "grad_norm": 0.709065854549408, "learning_rate": 0.0001, "elbo": 3.51305, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.976659629368989}
{"step": 67815, "loss": 3.5350826263427733, "grad_norm": 0.7040804624557495, "learning_rate": 0.0001, "elbo": 3.53508, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9769528501055595}
{"step": 67820, "loss": 3.522322082519531, "grad_norm": 0.901703953742981, "learning_rate": 0.0001, "elbo": 3.52232, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.97724607084213}
{"step": 67825, "loss": 3.5362174987792967, "grad_norm": 0.7500001788139343, "learning_rate": 0.0001, "elbo": 3.53622, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9775392915787005}
{"step": 67830, "loss": 3.525987243652344, "grad_norm": 0.7869118452072144, "learning_rate": 0.0001, "elbo": 3.52599, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.977832512315271}
{"step": 67835, "loss": 3.5254730224609374, "grad_norm": 0.7440451979637146, "learning_rate": 0.0001, "elbo": 3.52547, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9781257330518414}
{"step": 67840, "loss": 3.542400360107422, "grad_norm": 0.7058584690093994, "learning_rate": 0.0001, "elbo": 3.5424, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.978418953788412}
{"step": 67845, "loss": 3.5398345947265626, "grad_norm": 0.7963032126426697, "learning_rate": 0.0001, "elbo": 3.53983, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9787121745249823}
{"step": 67850, "loss": 3.5384117126464845, "grad_norm": 0.7322286367416382, "learning_rate": 0.0001, "elbo": 3.53841, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.979005395261553}
{"step": 67855, "loss": 3.5194446563720705, "grad_norm": 0.6936789155006409, "learning_rate": 0.0001, "elbo": 3.51944, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9792986159981236}
{"step": 67860, "loss": 3.517011260986328, "grad_norm": 0.6854458451271057, "learning_rate": 0.0001, "elbo": 3.51701, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.979591836734694}
{"step": 67865, "loss": 3.526868438720703, "grad_norm": 0.8093498349189758, "learning_rate": 0.0001, "elbo": 3.52687, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9798850574712645}
{"step": 67870, "loss": 3.530457305908203, "grad_norm": 0.7241671085357666, "learning_rate": 0.0001, "elbo": 3.53046, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.980178278207835}
{"step": 67875, "loss": 3.5453411102294923, "grad_norm": 0.7126458287239075, "learning_rate": 0.0001, "elbo": 3.54534, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9804714989444054}
{"step": 67875, "loss": 3.5453411102294923, "grad_norm": 0.7126458287239075, "learning_rate": 0.0001, "elbo": 3.54534, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9804714989444054}
{"step": 67880, "loss": 3.544420623779297, "grad_norm": 0.6844565272331238, "learning_rate": 0.0001, "elbo": 3.54442, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.980764719680976}
{"step": 67885, "loss": 3.525441360473633, "grad_norm": 0.6860659718513489, "learning_rate": 0.0001, "elbo": 3.52544, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9810579404175463}
{"step": 67890, "loss": 3.521923828125, "grad_norm": 0.6665843725204468, "learning_rate": 0.0001, "elbo": 3.52192, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.981351161154117}
{"step": 67895, "loss": 3.5317516326904297, "grad_norm": 0.7299568057060242, "learning_rate": 0.0001, "elbo": 3.53175, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9816443818906873}
{"step": 67900, "loss": 3.534449005126953, "grad_norm": 0.7078137993812561, "learning_rate": 0.0001, "elbo": 3.53445, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9819376026272577}
{"step": 67905, "loss": 3.534198760986328, "grad_norm": 0.6811066269874573, "learning_rate": 0.0001, "elbo": 3.5342, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.982230823363828}
{"step": 67910, "loss": 3.5189666748046875, "grad_norm": 0.8082450032234192, "learning_rate": 0.0001, "elbo": 3.51897, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9825240441003986}
{"step": 67915, "loss": 3.546207046508789, "grad_norm": 0.7167889475822449, "learning_rate": 0.0001, "elbo": 3.54621, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.982817264836969}
{"step": 67920, "loss": 3.5205307006835938, "grad_norm": 0.6997002363204956, "learning_rate": 0.0001, "elbo": 3.52053, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9831104855735395}
{"step": 67925, "loss": 3.5296043395996093, "grad_norm": 0.7247229814529419, "learning_rate": 0.0001, "elbo": 3.5296, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.98340370631011}
{"step": 67930, "loss": 3.5481971740722655, "grad_norm": 0.7909563183784485, "learning_rate": 0.0001, "elbo": 3.5482, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.983696927046681}
{"step": 67935, "loss": 3.513311004638672, "grad_norm": 0.7101646661758423, "learning_rate": 0.0001, "elbo": 3.51331, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9839901477832513}
{"step": 67940, "loss": 3.5191551208496095, "grad_norm": 0.7397934794425964, "learning_rate": 0.0001, "elbo": 3.51916, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.984283368519822}
{"step": 67945, "loss": 3.524456024169922, "grad_norm": 0.6938247680664062, "learning_rate": 0.0001, "elbo": 3.52446, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9845765892563922}
{"step": 67950, "loss": 3.526972198486328, "grad_norm": 0.6969705820083618, "learning_rate": 0.0001, "elbo": 3.52697, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9848698099929627}
{"step": 67955, "loss": 3.534986114501953, "grad_norm": 0.7455706596374512, "learning_rate": 0.0001, "elbo": 3.53499, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.985163030729533}
{"step": 67960, "loss": 3.5261913299560548, "grad_norm": 0.6933209896087646, "learning_rate": 0.0001, "elbo": 3.52619, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9854562514661036}
{"step": 67965, "loss": 3.526328277587891, "grad_norm": 0.7931668758392334, "learning_rate": 0.0001, "elbo": 3.52633, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.985749472202674}
{"step": 67970, "loss": 3.5036689758300783, "grad_norm": 0.7672345638275146, "learning_rate": 0.0001, "elbo": 3.50367, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9860426929392445}
{"step": 67975, "loss": 3.542829132080078, "grad_norm": 0.7096822261810303, "learning_rate": 0.0001, "elbo": 3.54283, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9863359136758154}
{"step": 67975, "loss": 3.542829132080078, "grad_norm": 0.7096822261810303, "learning_rate": 0.0001, "elbo": 3.54283, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9863359136758154}
{"step": 67980, "loss": 3.5197402954101564, "grad_norm": 0.6971853375434875, "learning_rate": 0.0001, "elbo": 3.51974, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.986629134412386}
{"step": 67985, "loss": 3.529405975341797, "grad_norm": 0.7990260124206543, "learning_rate": 0.0001, "elbo": 3.52941, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9869223551489563}
{"step": 67990, "loss": 3.5420181274414064, "grad_norm": 0.8985153436660767, "learning_rate": 0.0001, "elbo": 3.54202, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9872155758855268}
{"step": 67995, "loss": 3.5297012329101562, "grad_norm": 0.6837660074234009, "learning_rate": 0.0001, "elbo": 3.5297, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9875087966220972}
{"step": 68000, "loss": 3.51407470703125, "grad_norm": 0.7332605719566345, "learning_rate": 0.0001, "elbo": 3.51407, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9878020173586677}
{"step": 68005, "loss": 3.533289337158203, "grad_norm": 0.7413703203201294, "learning_rate": 0.0001, "elbo": 3.53329, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.988095238095238}
{"step": 68010, "loss": 3.5196273803710936, "grad_norm": 0.7833863496780396, "learning_rate": 0.0001, "elbo": 3.51963, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9883884588318086}
{"step": 68015, "loss": 3.521105194091797, "grad_norm": 0.707497775554657, "learning_rate": 0.0001, "elbo": 3.52111, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.988681679568379}
{"step": 68020, "loss": 3.536765289306641, "grad_norm": 0.7090973854064941, "learning_rate": 0.0001, "elbo": 3.53677, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9889749003049495}
{"step": 68025, "loss": 3.5290016174316405, "grad_norm": 1.4345208406448364, "learning_rate": 0.0001, "elbo": 3.529, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.98926812104152}
{"step": 68030, "loss": 3.516138458251953, "grad_norm": 0.7003225088119507, "learning_rate": 0.0001, "elbo": 3.51614, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9895613417780904}
{"step": 68035, "loss": 3.5317756652832033, "grad_norm": 0.7041432857513428, "learning_rate": 0.0001, "elbo": 3.53178, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.989854562514661}
{"step": 68040, "loss": 3.5345306396484375, "grad_norm": 0.7168282270431519, "learning_rate": 0.0001, "elbo": 3.53453, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9901477832512313}
{"step": 68045, "loss": 3.5310325622558594, "grad_norm": 1.0528141260147095, "learning_rate": 0.0001, "elbo": 3.53103, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9904410039878018}
{"step": 68050, "loss": 3.5094879150390623, "grad_norm": 0.720457136631012, "learning_rate": 0.0001, "elbo": 3.50949, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9907342247243727}
{"step": 68055, "loss": 3.524100494384766, "grad_norm": 0.7326381802558899, "learning_rate": 0.0001, "elbo": 3.5241, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.991027445460943}
{"step": 68060, "loss": 3.5261497497558594, "grad_norm": 0.750900149345398, "learning_rate": 0.0001, "elbo": 3.52615, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9913206661975136}
{"step": 68065, "loss": 3.54434814453125, "grad_norm": 0.687347412109375, "learning_rate": 0.0001, "elbo": 3.54435, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.991613886934084}
{"step": 68070, "loss": 3.525194549560547, "grad_norm": 0.8165856599807739, "learning_rate": 0.0001, "elbo": 3.52519, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9919071076706545}
{"step": 68075, "loss": 3.5266139984130858, "grad_norm": 0.7101457715034485, "learning_rate": 0.0001, "elbo": 3.52661, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.992200328407225}
{"step": 68075, "loss": 3.5266139984130858, "grad_norm": 0.7101457715034485, "learning_rate": 0.0001, "elbo": 3.52661, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.992200328407225}
{"step": 68080, "loss": 3.5385822296142577, "grad_norm": 0.8257655501365662, "learning_rate": 0.0001, "elbo": 3.53858, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9924935491437954}
{"step": 68085, "loss": 3.5336231231689452, "grad_norm": 0.7329773902893066, "learning_rate": 0.0001, "elbo": 3.53362, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.992786769880366}
{"step": 68090, "loss": 3.521486282348633, "grad_norm": 0.7077417969703674, "learning_rate": 0.0001, "elbo": 3.52149, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9930799906169363}
{"step": 68095, "loss": 3.530302047729492, "grad_norm": 0.6684582233428955, "learning_rate": 0.0001, "elbo": 3.5303, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9933732113535068}
{"step": 68100, "loss": 3.5257545471191407, "grad_norm": 0.7049340605735779, "learning_rate": 0.0001, "elbo": 3.52575, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9936664320900777}
{"step": 68105, "loss": 3.53330078125, "grad_norm": 0.728725790977478, "learning_rate": 0.0001, "elbo": 3.5333, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.993959652826648}
{"step": 68110, "loss": 3.527815246582031, "grad_norm": 0.7003486752510071, "learning_rate": 0.0001, "elbo": 3.52782, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9942528735632186}
{"step": 68115, "loss": 3.5262424468994142, "grad_norm": 0.6856144666671753, "learning_rate": 0.0001, "elbo": 3.52624, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.994546094299789}
{"step": 68120, "loss": 3.5346755981445312, "grad_norm": 0.697912335395813, "learning_rate": 0.0001, "elbo": 3.53468, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9948393150363595}
{"step": 68125, "loss": 3.523382568359375, "grad_norm": 0.7911473512649536, "learning_rate": 0.0001, "elbo": 3.52338, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.99513253577293}
{"step": 68130, "loss": 3.541751480102539, "grad_norm": 0.8113813996315002, "learning_rate": 0.0001, "elbo": 3.54175, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9954257565095004}
{"step": 68135, "loss": 3.532009506225586, "grad_norm": 0.6598060131072998, "learning_rate": 0.0001, "elbo": 3.53201, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.995718977246071}
{"step": 68140, "loss": 3.5235664367675783, "grad_norm": 0.7439688444137573, "learning_rate": 0.0001, "elbo": 3.52357, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9960121979826413}
{"step": 68145, "loss": 3.5329532623291016, "grad_norm": 0.7063445448875427, "learning_rate": 0.0001, "elbo": 3.53295, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9963054187192117}
{"step": 68150, "loss": 3.5258026123046875, "grad_norm": 0.724232017993927, "learning_rate": 0.0001, "elbo": 3.5258, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.996598639455782}
{"step": 68155, "loss": 3.5154644012451173, "grad_norm": 0.6802322864532471, "learning_rate": 0.0001, "elbo": 3.51546, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9968918601923527}
{"step": 68160, "loss": 3.529798126220703, "grad_norm": 0.7491930723190308, "learning_rate": 0.0001, "elbo": 3.5298, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.997185080928923}
{"step": 68165, "loss": 3.5212963104248045, "grad_norm": 0.6973106265068054, "learning_rate": 0.0001, "elbo": 3.5213, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9974783016654936}
{"step": 68170, "loss": 3.536082458496094, "grad_norm": 0.6616246700286865, "learning_rate": 0.0001, "elbo": 3.53608, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.997771522402064}
{"step": 68175, "loss": 3.522479248046875, "grad_norm": 0.734013020992279, "learning_rate": 0.0001, "elbo": 3.52248, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.998064743138635}
{"step": 68180, "loss": 3.5161285400390625, "grad_norm": 0.7480552196502686, "learning_rate": 0.0001, "elbo": 3.51613, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9983579638752054}
{"step": 68185, "loss": 3.521514129638672, "grad_norm": 0.7354402542114258, "learning_rate": 0.0001, "elbo": 3.52151, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.998651184611776}
{"step": 68190, "loss": 3.53427734375, "grad_norm": 0.7406812310218811, "learning_rate": 0.0001, "elbo": 3.53428, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9989444053483463}
{"step": 68195, "loss": 3.5358325958251955, "grad_norm": 0.7064200639724731, "learning_rate": 0.0001, "elbo": 3.53583, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9992376260849167}
{"step": 68200, "loss": 3.5144378662109377, "grad_norm": 0.727534294128418, "learning_rate": 0.0001, "elbo": 3.51444, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.999530846821487}
{"step": 68205, "loss": 3.521251678466797, "grad_norm": 0.8218148350715637, "learning_rate": 0.0001, "elbo": 3.52125, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 3.9998240675580576}
{"step": 68210, "loss": 3.5425514221191405, "grad_norm": 0.7413842082023621, "learning_rate": 0.0001, "elbo": 3.54255, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0001172882946285}
{"step": 68215, "loss": 3.5230781555175783, "grad_norm": 0.6717231273651123, "learning_rate": 0.0001, "elbo": 3.52308, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.000410509031199}
{"step": 68215, "loss": 3.5230781555175783, "grad_norm": 0.6717231273651123, "learning_rate": 0.0001, "elbo": 3.52308, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.000410509031199}
{"step": 68220, "loss": 3.547381591796875, "grad_norm": 0.6789609789848328, "learning_rate": 0.0001, "elbo": 3.54738, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0007037297677694}
{"step": 68225, "loss": 3.538724899291992, "grad_norm": 0.8263278603553772, "learning_rate": 0.0001, "elbo": 3.53872, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.00099695050434}
{"step": 68230, "loss": 3.548558807373047, "grad_norm": 0.7034105658531189, "learning_rate": 0.0001, "elbo": 3.54856, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.00129017124091}
{"step": 68235, "loss": 3.528398895263672, "grad_norm": 0.7004755139350891, "learning_rate": 0.0001, "elbo": 3.5284, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.001583391977481}
{"step": 68240, "loss": 3.537561798095703, "grad_norm": 0.8051543235778809, "learning_rate": 0.0001, "elbo": 3.53756, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.001876612714051}
{"step": 68245, "loss": 3.5090919494628907, "grad_norm": 0.6778067350387573, "learning_rate": 0.0001, "elbo": 3.50909, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.002169833450622}
{"step": 68250, "loss": 3.5301223754882813, "grad_norm": 1.1432456970214844, "learning_rate": 0.0001, "elbo": 3.53012, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.002463054187192}
{"step": 68255, "loss": 3.5338897705078125, "grad_norm": 0.7287644743919373, "learning_rate": 0.0001, "elbo": 3.53389, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.002756274923763}
{"step": 68260, "loss": 3.5229423522949217, "grad_norm": 0.8087698817253113, "learning_rate": 0.0001, "elbo": 3.52294, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.003049495660333}
{"step": 68265, "loss": 3.5323501586914063, "grad_norm": 0.7186852097511292, "learning_rate": 0.0001, "elbo": 3.53235, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0033427163969035}
{"step": 68270, "loss": 3.5232337951660155, "grad_norm": 0.6998754739761353, "learning_rate": 0.0001, "elbo": 3.52323, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.003635937133474}
{"step": 68275, "loss": 3.5326465606689452, "grad_norm": 0.7654930949211121, "learning_rate": 0.0001, "elbo": 3.53265, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.003929157870044}
{"step": 68280, "loss": 3.529897689819336, "grad_norm": 0.719240128993988, "learning_rate": 0.0001, "elbo": 3.5299, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.004222378606615}
{"step": 68285, "loss": 3.5119361877441406, "grad_norm": 0.6910674571990967, "learning_rate": 0.0001, "elbo": 3.51194, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.004515599343185}
{"step": 68290, "loss": 3.519532012939453, "grad_norm": 0.8724251985549927, "learning_rate": 0.0001, "elbo": 3.51953, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.004808820079756}
{"step": 68295, "loss": 3.5237991333007814, "grad_norm": 0.7116338014602661, "learning_rate": 0.0001, "elbo": 3.5238, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.005102040816326}
{"step": 68300, "loss": 3.5228321075439455, "grad_norm": 0.6796684861183167, "learning_rate": 0.0001, "elbo": 3.52283, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.005395261552897}
{"step": 68300, "loss": 3.5228321075439455, "grad_norm": 0.6796684861183167, "learning_rate": 0.0001, "elbo": 3.52283, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.005395261552897}
{"step": 68305, "loss": 3.5232147216796874, "grad_norm": 0.6884989142417908, "learning_rate": 0.0001, "elbo": 3.52321, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.005688482289467}
{"step": 68310, "loss": 3.5175552368164062, "grad_norm": 0.6760279536247253, "learning_rate": 0.0001, "elbo": 3.51756, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.005981703026038}
{"step": 68310, "loss": 3.5175552368164062, "grad_norm": 0.6760279536247253, "learning_rate": 0.0001, "elbo": 3.51756, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.005981703026038}
{"step": 68315, "loss": 3.5080802917480467, "grad_norm": 0.6929786801338196, "learning_rate": 0.0001, "elbo": 3.50808, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.006274923762608}
{"step": 68320, "loss": 3.538069152832031, "grad_norm": 0.7361977100372314, "learning_rate": 0.0001, "elbo": 3.53807, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.006568144499179}
{"step": 68325, "loss": 3.5275550842285157, "grad_norm": 0.7418628931045532, "learning_rate": 0.0001, "elbo": 3.52756, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.00686136523575}
{"step": 68330, "loss": 3.5387241363525392, "grad_norm": 0.689639687538147, "learning_rate": 0.0001, "elbo": 3.53872, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.00715458597232}
{"step": 68335, "loss": 3.528535842895508, "grad_norm": 0.7697540521621704, "learning_rate": 0.0001, "elbo": 3.52854, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.007447806708891}
{"step": 68340, "loss": 3.532143783569336, "grad_norm": 0.7953962683677673, "learning_rate": 0.0001, "elbo": 3.53214, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.007741027445461}
{"step": 68345, "loss": 3.5077205657958985, "grad_norm": 0.7644192576408386, "learning_rate": 0.0001, "elbo": 3.50772, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.008034248182032}
{"step": 68350, "loss": 3.528656768798828, "grad_norm": 0.7653250694274902, "learning_rate": 0.0001, "elbo": 3.52866, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.008327468918602}
{"step": 68355, "loss": 3.5249813079833983, "grad_norm": 0.8104366660118103, "learning_rate": 0.0001, "elbo": 3.52498, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.008620689655173}
{"step": 68360, "loss": 3.526144790649414, "grad_norm": 0.7557517290115356, "learning_rate": 0.0001, "elbo": 3.52614, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.008913910391743}
{"step": 68365, "loss": 3.5166046142578127, "grad_norm": 1.0568501949310303, "learning_rate": 0.0001, "elbo": 3.5166, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0092071311283135}
{"step": 68370, "loss": 3.515901565551758, "grad_norm": 0.676112174987793, "learning_rate": 0.0001, "elbo": 3.5159, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.009500351864884}
{"step": 68375, "loss": 3.5512069702148437, "grad_norm": 0.7038561105728149, "learning_rate": 0.0001, "elbo": 3.55121, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.009793572601454}
{"step": 68380, "loss": 3.5257301330566406, "grad_norm": 0.6829789876937866, "learning_rate": 0.0001, "elbo": 3.52573, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.010086793338025}
{"step": 68385, "loss": 3.539065933227539, "grad_norm": 0.6677377820014954, "learning_rate": 0.0001, "elbo": 3.53907, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.010380014074595}
{"step": 68390, "loss": 3.5188625335693358, "grad_norm": 0.7802451848983765, "learning_rate": 0.0001, "elbo": 3.51886, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.010673234811166}
{"step": 68395, "loss": 3.534131622314453, "grad_norm": 0.716411828994751, "learning_rate": 0.0001, "elbo": 3.53413, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.010966455547736}
{"step": 68400, "loss": 3.5190757751464843, "grad_norm": 0.7509880065917969, "learning_rate": 0.0001, "elbo": 3.51908, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.011259676284307}
{"step": 68405, "loss": 3.5321388244628906, "grad_norm": 0.7285151481628418, "learning_rate": 0.0001, "elbo": 3.53214, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.011552897020877}
{"step": 68410, "loss": 3.526742935180664, "grad_norm": 0.7297635078430176, "learning_rate": 0.0001, "elbo": 3.52674, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.011846117757448}
{"step": 68415, "loss": 3.517279052734375, "grad_norm": 0.6727946400642395, "learning_rate": 0.0001, "elbo": 3.51728, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.012139338494018}
{"step": 68420, "loss": 3.522814178466797, "grad_norm": 0.6826040744781494, "learning_rate": 0.0001, "elbo": 3.52281, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0124325592305885}
{"step": 68425, "loss": 3.531068801879883, "grad_norm": 0.6997442245483398, "learning_rate": 0.0001, "elbo": 3.53107, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.012725779967159}
{"step": 68430, "loss": 3.5158824920654297, "grad_norm": 0.7519645094871521, "learning_rate": 0.0001, "elbo": 3.51588, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.013019000703729}
{"step": 68435, "loss": 3.537815475463867, "grad_norm": 0.7080072164535522, "learning_rate": 0.0001, "elbo": 3.53782, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0133122214403}
{"step": 68440, "loss": 3.541775131225586, "grad_norm": 1.0479201078414917, "learning_rate": 0.0001, "elbo": 3.54178, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.01360544217687}
{"step": 68445, "loss": 3.5369972229003905, "grad_norm": 0.6374230980873108, "learning_rate": 0.0001, "elbo": 3.537, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.013898662913442}
{"step": 68450, "loss": 3.541820526123047, "grad_norm": 0.734954833984375, "learning_rate": 0.0001, "elbo": 3.54182, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.014191883650012}
{"step": 68455, "loss": 3.529816436767578, "grad_norm": 0.7265633940696716, "learning_rate": 0.0001, "elbo": 3.52982, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.014485104386583}
{"step": 68460, "loss": 3.5276824951171877, "grad_norm": 0.8132797479629517, "learning_rate": 0.0001, "elbo": 3.52768, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.014778325123153}
{"step": 68465, "loss": 3.5375431060791014, "grad_norm": 0.7847738862037659, "learning_rate": 0.0001, "elbo": 3.53754, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0150715458597235}
{"step": 68470, "loss": 3.5242591857910157, "grad_norm": 0.6728383302688599, "learning_rate": 0.0001, "elbo": 3.52426, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.015364766596294}
{"step": 68475, "loss": 3.5297523498535157, "grad_norm": 0.7063596248626709, "learning_rate": 0.0001, "elbo": 3.52975, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.015657987332864}
{"step": 68480, "loss": 3.528644561767578, "grad_norm": 0.6423032283782959, "learning_rate": 0.0001, "elbo": 3.52864, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.015951208069435}
{"step": 68485, "loss": 3.5100772857666014, "grad_norm": 0.7990788817405701, "learning_rate": 0.0001, "elbo": 3.51008, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.016244428806005}
{"step": 68490, "loss": 3.528957748413086, "grad_norm": 0.6665604114532471, "learning_rate": 0.0001, "elbo": 3.52896, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.016537649542576}
{"step": 68495, "loss": 3.5147354125976564, "grad_norm": 0.8319060206413269, "learning_rate": 0.0001, "elbo": 3.51474, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.016830870279146}
{"step": 68500, "loss": 3.5395786285400392, "grad_norm": 0.6928401589393616, "learning_rate": 0.0001, "elbo": 3.53958, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.017124091015717}
{"step": 68505, "loss": 3.507548141479492, "grad_norm": 0.7343305349349976, "learning_rate": 0.0001, "elbo": 3.50755, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.017417311752287}
{"step": 68510, "loss": 3.515147399902344, "grad_norm": 0.7587316632270813, "learning_rate": 0.0001, "elbo": 3.51515, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.017710532488858}
{"step": 68515, "loss": 3.5366374969482424, "grad_norm": 0.7045930624008179, "learning_rate": 0.0001, "elbo": 3.53664, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.018003753225428}
{"step": 68520, "loss": 3.5283180236816407, "grad_norm": 0.711245059967041, "learning_rate": 0.0001, "elbo": 3.52832, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0182969739619985}
{"step": 68520, "loss": 3.5283180236816407, "grad_norm": 0.711245059967041, "learning_rate": 0.0001, "elbo": 3.52832, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0182969739619985}
{"step": 68525, "loss": 3.5222900390625, "grad_norm": 0.7383648753166199, "learning_rate": 0.0001, "elbo": 3.52229, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.018590194698569}
{"step": 68530, "loss": 3.5253150939941404, "grad_norm": 0.6870134472846985, "learning_rate": 0.0001, "elbo": 3.52532, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.018883415435139}
{"step": 68535, "loss": 3.5264751434326174, "grad_norm": 0.7624296545982361, "learning_rate": 0.0001, "elbo": 3.52648, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.01917663617171}
{"step": 68540, "loss": 3.5334381103515624, "grad_norm": 0.9320566654205322, "learning_rate": 0.0001, "elbo": 3.53344, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.01946985690828}
{"step": 68545, "loss": 3.522683334350586, "grad_norm": 0.7591361999511719, "learning_rate": 0.0001, "elbo": 3.52268, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.019763077644851}
{"step": 68550, "loss": 3.527867889404297, "grad_norm": 0.8977959156036377, "learning_rate": 0.0001, "elbo": 3.52787, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.020056298381421}
{"step": 68555, "loss": 3.5270759582519533, "grad_norm": 0.749184250831604, "learning_rate": 0.0001, "elbo": 3.52708, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.020349519117992}
{"step": 68560, "loss": 3.5255203247070312, "grad_norm": 0.8090632557868958, "learning_rate": 0.0001, "elbo": 3.52552, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.020642739854562}
{"step": 68565, "loss": 3.520925521850586, "grad_norm": 0.6923753023147583, "learning_rate": 0.0001, "elbo": 3.52093, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.020935960591133}
{"step": 68570, "loss": 3.527460479736328, "grad_norm": 0.702865719795227, "learning_rate": 0.0001, "elbo": 3.52746, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.021229181327704}
{"step": 68575, "loss": 3.527945327758789, "grad_norm": 0.691925048828125, "learning_rate": 0.0001, "elbo": 3.52795, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.021522402064274}
{"step": 68580, "loss": 3.522216796875, "grad_norm": 0.725287914276123, "learning_rate": 0.0001, "elbo": 3.52222, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.021815622800845}
{"step": 68585, "loss": 3.5216957092285157, "grad_norm": 0.7774960398674011, "learning_rate": 0.0001, "elbo": 3.5217, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.022108843537415}
{"step": 68590, "loss": 3.5247859954833984, "grad_norm": 0.6552923321723938, "learning_rate": 0.0001, "elbo": 3.52479, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.022402064273986}
{"step": 68595, "loss": 3.5140617370605467, "grad_norm": 0.6799347400665283, "learning_rate": 0.0001, "elbo": 3.51406, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.022695285010556}
{"step": 68600, "loss": 3.5190948486328124, "grad_norm": 0.7179925441741943, "learning_rate": 0.0001, "elbo": 3.51909, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.022988505747127}
{"step": 68605, "loss": 3.5303855895996095, "grad_norm": 0.7067606449127197, "learning_rate": 0.0001, "elbo": 3.53039, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.023281726483697}
{"step": 68610, "loss": 3.534075164794922, "grad_norm": 0.7430659532546997, "learning_rate": 0.0001, "elbo": 3.53408, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0235749472202675}
{"step": 68615, "loss": 3.531676483154297, "grad_norm": 0.6942981481552124, "learning_rate": 0.0001, "elbo": 3.53168, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.023868167956838}
{"step": 68620, "loss": 3.544819641113281, "grad_norm": 0.6543576717376709, "learning_rate": 0.0001, "elbo": 3.54482, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0241613886934084}
{"step": 68625, "loss": 3.5216995239257813, "grad_norm": 0.7717915773391724, "learning_rate": 0.0001, "elbo": 3.5217, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.024454609429979}
{"step": 68630, "loss": 3.533772277832031, "grad_norm": 0.6977272629737854, "learning_rate": 0.0001, "elbo": 3.53377, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.024747830166549}
{"step": 68635, "loss": 3.518054962158203, "grad_norm": 0.706566333770752, "learning_rate": 0.0001, "elbo": 3.51805, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.02504105090312}
{"step": 68640, "loss": 3.523876953125, "grad_norm": 0.6758961081504822, "learning_rate": 0.0001, "elbo": 3.52388, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.02533427163969}
{"step": 68645, "loss": 3.5141220092773438, "grad_norm": 0.7249482870101929, "learning_rate": 0.0001, "elbo": 3.51412, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.025627492376261}
{"step": 68650, "loss": 3.517311859130859, "grad_norm": 0.71006178855896, "learning_rate": 0.0001, "elbo": 3.51731, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.025920713112831}
{"step": 68655, "loss": 3.5072376251220705, "grad_norm": 0.7007507681846619, "learning_rate": 0.0001, "elbo": 3.50724, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.026213933849402}
{"step": 68660, "loss": 3.525243377685547, "grad_norm": 0.7252350449562073, "learning_rate": 0.0001, "elbo": 3.52524, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.026507154585972}
{"step": 68665, "loss": 3.535568618774414, "grad_norm": 0.8020566701889038, "learning_rate": 0.0001, "elbo": 3.53557, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0268003753225425}
{"step": 68670, "loss": 3.5274116516113283, "grad_norm": 0.7108193635940552, "learning_rate": 0.0001, "elbo": 3.52741, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.027093596059113}
{"step": 68675, "loss": 3.5334381103515624, "grad_norm": 0.7395958304405212, "learning_rate": 0.0001, "elbo": 3.53344, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.027386816795683}
{"step": 68680, "loss": 3.527177429199219, "grad_norm": 0.7213907837867737, "learning_rate": 0.0001, "elbo": 3.52718, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.027680037532254}
{"step": 68685, "loss": 3.5301395416259767, "grad_norm": 1.029176115989685, "learning_rate": 0.0001, "elbo": 3.53014, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.027973258268824}
{"step": 68685, "loss": 3.5301395416259767, "grad_norm": 1.029176115989685, "learning_rate": 0.0001, "elbo": 3.53014, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.027973258268824}
{"step": 68690, "loss": 3.519337844848633, "grad_norm": 0.711153507232666, "learning_rate": 0.0001, "elbo": 3.51934, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.028266479005396}
{"step": 68695, "loss": 3.543126678466797, "grad_norm": 0.7902964353561401, "learning_rate": 0.0001, "elbo": 3.54313, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.028559699741966}
{"step": 68700, "loss": 3.5286468505859374, "grad_norm": 0.7162643671035767, "learning_rate": 0.0001, "elbo": 3.52865, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.028852920478537}
{"step": 68705, "loss": 3.5196517944335937, "grad_norm": 0.6873751282691956, "learning_rate": 0.0001, "elbo": 3.51965, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.029146141215107}
{"step": 68710, "loss": 3.5364730834960936, "grad_norm": 0.6864062547683716, "learning_rate": 0.0001, "elbo": 3.53647, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0294393619516775}
{"step": 68715, "loss": 3.5261783599853516, "grad_norm": 0.7276838421821594, "learning_rate": 0.0001, "elbo": 3.52618, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.029732582688248}
{"step": 68720, "loss": 3.519511032104492, "grad_norm": 0.7463076710700989, "learning_rate": 0.0001, "elbo": 3.51951, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.030025803424818}
{"step": 68725, "loss": 3.533769226074219, "grad_norm": 0.7010526657104492, "learning_rate": 0.0001, "elbo": 3.53377, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.030319024161389}
{"step": 68730, "loss": 3.5141937255859377, "grad_norm": 0.6851775646209717, "learning_rate": 0.0001, "elbo": 3.51419, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.030612244897959}
{"step": 68735, "loss": 3.525251007080078, "grad_norm": 0.70219886302948, "learning_rate": 0.0001, "elbo": 3.52525, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.03090546563453}
{"step": 68740, "loss": 3.5191635131835937, "grad_norm": 0.7309705018997192, "learning_rate": 0.0001, "elbo": 3.51916, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0311986863711}
{"step": 68745, "loss": 3.529275894165039, "grad_norm": 1.1614974737167358, "learning_rate": 0.0001, "elbo": 3.52928, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.031491907107671}
{"step": 68750, "loss": 3.525775909423828, "grad_norm": 0.6712201237678528, "learning_rate": 0.0001, "elbo": 3.52578, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.031785127844241}
{"step": 68755, "loss": 3.515950012207031, "grad_norm": 0.690194845199585, "learning_rate": 0.0001, "elbo": 3.51595, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.032078348580812}
{"step": 68760, "loss": 3.5385555267333983, "grad_norm": 0.6671754121780396, "learning_rate": 0.0001, "elbo": 3.53856, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.032371569317382}
{"step": 68765, "loss": 3.5379981994628906, "grad_norm": 0.6658838987350464, "learning_rate": 0.0001, "elbo": 3.538, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0326647900539525}
{"step": 68770, "loss": 3.524517059326172, "grad_norm": 0.7911379933357239, "learning_rate": 0.0001, "elbo": 3.52452, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.032958010790523}
{"step": 68775, "loss": 3.5321922302246094, "grad_norm": 0.7891284227371216, "learning_rate": 0.0001, "elbo": 3.53219, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.033251231527093}
{"step": 68780, "loss": 3.523699188232422, "grad_norm": 0.717610239982605, "learning_rate": 0.0001, "elbo": 3.5237, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.033544452263664}
{"step": 68785, "loss": 3.5436744689941406, "grad_norm": 0.9250388741493225, "learning_rate": 0.0001, "elbo": 3.54367, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.033837673000234}
{"step": 68790, "loss": 3.5321891784667967, "grad_norm": 0.6893518567085266, "learning_rate": 0.0001, "elbo": 3.53219, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.034130893736805}
{"step": 68795, "loss": 3.5237903594970703, "grad_norm": 0.7064315676689148, "learning_rate": 0.0001, "elbo": 3.52379, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.034424114473375}
{"step": 68800, "loss": 3.506071853637695, "grad_norm": 0.7668181657791138, "learning_rate": 0.0001, "elbo": 3.50607, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.034717335209946}
{"step": 68805, "loss": 3.5422706604003906, "grad_norm": 0.6878803968429565, "learning_rate": 0.0001, "elbo": 3.54227, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.035010555946516}
{"step": 68805, "loss": 3.5422706604003906, "grad_norm": 0.6878803968429565, "learning_rate": 0.0001, "elbo": 3.54227, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.035010555946516}
{"step": 68810, "loss": 3.5260929107666015, "grad_norm": 0.7947851419448853, "learning_rate": 0.0001, "elbo": 3.52609, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.035303776683087}
{"step": 68815, "loss": 3.536576843261719, "grad_norm": 0.7016162276268005, "learning_rate": 0.0001, "elbo": 3.53658, "mask_frac": 0.5013, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.035596997419658}
{"step": 68820, "loss": 3.5304527282714844, "grad_norm": 0.6950825452804565, "learning_rate": 0.0001, "elbo": 3.53045, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.035890218156228}
{"step": 68825, "loss": 3.518707275390625, "grad_norm": 0.7489219903945923, "learning_rate": 0.0001, "elbo": 3.51871, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.036183438892799}
{"step": 68830, "loss": 3.531555938720703, "grad_norm": 0.7026576995849609, "learning_rate": 0.0001, "elbo": 3.53156, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.036476659629369}
{"step": 68835, "loss": 3.5297592163085936, "grad_norm": 0.6433469653129578, "learning_rate": 0.0001, "elbo": 3.52976, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.03676988036594}
{"step": 68840, "loss": 3.5283626556396483, "grad_norm": 0.6756136417388916, "learning_rate": 0.0001, "elbo": 3.52836, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.03706310110251}
{"step": 68845, "loss": 3.520800018310547, "grad_norm": 0.963748037815094, "learning_rate": 0.0001, "elbo": 3.5208, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.037356321839081}
{"step": 68850, "loss": 3.521006393432617, "grad_norm": 0.7798053622245789, "learning_rate": 0.0001, "elbo": 3.52101, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.037649542575651}
{"step": 68850, "loss": 3.521006393432617, "grad_norm": 0.7798053622245789, "learning_rate": 0.0001, "elbo": 3.52101, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.037649542575651}
{"step": 68855, "loss": 3.5309158325195313, "grad_norm": 0.70836341381073, "learning_rate": 0.0001, "elbo": 3.53092, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.037942763312222}
{"step": 68860, "loss": 3.520982360839844, "grad_norm": 0.7318494915962219, "learning_rate": 0.0001, "elbo": 3.52098, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.038235984048792}
{"step": 68865, "loss": 3.530043029785156, "grad_norm": 1.1951638460159302, "learning_rate": 0.0001, "elbo": 3.53004, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0385292047853625}
{"step": 68865, "loss": 3.530043029785156, "grad_norm": 1.1951638460159302, "learning_rate": 0.0001, "elbo": 3.53004, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0385292047853625}
{"step": 68870, "loss": 3.5186763763427735, "grad_norm": 0.6856955289840698, "learning_rate": 0.0001, "elbo": 3.51868, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.038822425521933}
{"step": 68875, "loss": 3.5392669677734374, "grad_norm": 0.9171363711357117, "learning_rate": 0.0001, "elbo": 3.53927, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.039115646258503}
{"step": 68880, "loss": 3.5292995452880858, "grad_norm": 0.7116744518280029, "learning_rate": 0.0001, "elbo": 3.5293, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.039408866995074}
{"step": 68880, "loss": 3.5292995452880858, "grad_norm": 0.7116744518280029, "learning_rate": 0.0001, "elbo": 3.5293, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.039408866995074}
{"step": 68885, "loss": 3.529039764404297, "grad_norm": 0.8077967166900635, "learning_rate": 0.0001, "elbo": 3.52904, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.039702087731644}
{"step": 68890, "loss": 3.5208114624023437, "grad_norm": 0.6872479915618896, "learning_rate": 0.0001, "elbo": 3.52081, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.039995308468215}
{"step": 68895, "loss": 3.521367645263672, "grad_norm": 0.6770041584968567, "learning_rate": 0.0001, "elbo": 3.52137, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.040288529204785}
{"step": 68900, "loss": 3.5178409576416017, "grad_norm": 0.6931946277618408, "learning_rate": 0.0001, "elbo": 3.51784, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.040581749941356}
{"step": 68905, "loss": 3.5398269653320313, "grad_norm": 0.7113351225852966, "learning_rate": 0.0001, "elbo": 3.53983, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.040874970677926}
{"step": 68910, "loss": 3.5126182556152346, "grad_norm": 0.6958041787147522, "learning_rate": 0.0001, "elbo": 3.51262, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.041168191414497}
{"step": 68915, "loss": 3.5194679260253907, "grad_norm": 0.7953738570213318, "learning_rate": 0.0001, "elbo": 3.51947, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.041461412151067}
{"step": 68920, "loss": 3.5203285217285156, "grad_norm": 0.7260415554046631, "learning_rate": 0.0001, "elbo": 3.52033, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0417546328876375}
{"step": 68925, "loss": 3.5316017150878904, "grad_norm": 0.7385101914405823, "learning_rate": 0.0001, "elbo": 3.5316, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.042047853624208}
{"step": 68925, "loss": 3.5316017150878904, "grad_norm": 0.7385101914405823, "learning_rate": 0.0001, "elbo": 3.5316, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.042047853624208}
{"step": 68930, "loss": 3.5098495483398438, "grad_norm": 0.7114570736885071, "learning_rate": 0.0001, "elbo": 3.50985, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.042341074360778}
{"step": 68935, "loss": 3.5197525024414062, "grad_norm": 0.683113157749176, "learning_rate": 0.0001, "elbo": 3.51975, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.042634295097349}
{"step": 68940, "loss": 3.5324050903320314, "grad_norm": 0.7273460030555725, "learning_rate": 0.0001, "elbo": 3.5324, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.04292751583392}
{"step": 68945, "loss": 3.5317916870117188, "grad_norm": 0.6898680329322815, "learning_rate": 0.0001, "elbo": 3.53179, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.043220736570491}
{"step": 68950, "loss": 3.5243499755859373, "grad_norm": 0.7048788070678711, "learning_rate": 0.0001, "elbo": 3.52435, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.043513957307061}
{"step": 68955, "loss": 3.5268539428710937, "grad_norm": 0.7238062620162964, "learning_rate": 0.0001, "elbo": 3.52685, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0438071780436315}
{"step": 68960, "loss": 3.548326873779297, "grad_norm": 1.312605857849121, "learning_rate": 0.0001, "elbo": 3.54833, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.044100398780202}
{"step": 68965, "loss": 3.5445648193359376, "grad_norm": 0.7918976545333862, "learning_rate": 0.0001, "elbo": 3.54456, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0443936195167725}
{"step": 68970, "loss": 3.526662063598633, "grad_norm": 0.7376769781112671, "learning_rate": 0.0001, "elbo": 3.52666, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.044686840253343}
{"step": 68975, "loss": 3.529680633544922, "grad_norm": 0.8307173252105713, "learning_rate": 0.0001, "elbo": 3.52968, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.044980060989913}
{"step": 68980, "loss": 3.5386726379394533, "grad_norm": 0.7632460594177246, "learning_rate": 0.0001, "elbo": 3.53867, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.045273281726484}
{"step": 68985, "loss": 3.528255081176758, "grad_norm": 0.7013185024261475, "learning_rate": 0.0001, "elbo": 3.52826, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.045566502463054}
{"step": 68990, "loss": 3.522755432128906, "grad_norm": 0.7122322916984558, "learning_rate": 0.0001, "elbo": 3.52276, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.045859723199625}
{"step": 68995, "loss": 3.5224403381347655, "grad_norm": 0.7104150056838989, "learning_rate": 0.0001, "elbo": 3.52244, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.046152943936195}
{"step": 69000, "loss": 3.5316558837890626, "grad_norm": 0.7228320837020874, "learning_rate": 0.0001, "elbo": 3.53166, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.046446164672766}
{"step": 69005, "loss": 3.5229232788085936, "grad_norm": 0.6969403028488159, "learning_rate": 0.0001, "elbo": 3.52292, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.046739385409336}
{"step": 69010, "loss": 3.5075164794921876, "grad_norm": 0.686741828918457, "learning_rate": 0.0001, "elbo": 3.50752, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0470326061459065}
{"step": 69010, "loss": 3.5075164794921876, "grad_norm": 0.686741828918457, "learning_rate": 0.0001, "elbo": 3.50752, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0470326061459065}
{"step": 69015, "loss": 3.5253448486328125, "grad_norm": 0.8401845097541809, "learning_rate": 0.0001, "elbo": 3.52534, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.047325826882477}
{"step": 69020, "loss": 3.5314987182617186, "grad_norm": 0.7757130861282349, "learning_rate": 0.0001, "elbo": 3.5315, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0476190476190474}
{"step": 69025, "loss": 3.5102783203125, "grad_norm": 0.7581372857093811, "learning_rate": 0.0001, "elbo": 3.51028, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.047912268355618}
{"step": 69030, "loss": 3.512246322631836, "grad_norm": 0.9019796252250671, "learning_rate": 0.0001, "elbo": 3.51225, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.048205489092188}
{"step": 69035, "loss": 3.530598449707031, "grad_norm": 0.7487102150917053, "learning_rate": 0.0001, "elbo": 3.5306, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.048498709828759}
{"step": 69040, "loss": 3.5236488342285157, "grad_norm": 0.6936251521110535, "learning_rate": 0.0001, "elbo": 3.52365, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.048791930565329}
{"step": 69045, "loss": 3.5175018310546875, "grad_norm": 0.6863886713981628, "learning_rate": 0.0001, "elbo": 3.5175, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0490851513019}
{"step": 69050, "loss": 3.528331756591797, "grad_norm": 0.6658847332000732, "learning_rate": 0.0001, "elbo": 3.52833, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.04937837203847}
{"step": 69055, "loss": 3.533072662353516, "grad_norm": 0.7809204459190369, "learning_rate": 0.0001, "elbo": 3.53307, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.049671592775041}
{"step": 69060, "loss": 3.523682403564453, "grad_norm": 0.7390019297599792, "learning_rate": 0.0001, "elbo": 3.52368, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.049964813511612}
{"step": 69065, "loss": 3.524080276489258, "grad_norm": 0.7256963849067688, "learning_rate": 0.0001, "elbo": 3.52408, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.050258034248182}
{"step": 69070, "loss": 3.5319080352783203, "grad_norm": 0.709965705871582, "learning_rate": 0.0001, "elbo": 3.53191, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.050551254984753}
{"step": 69075, "loss": 3.51947021484375, "grad_norm": 0.8126018047332764, "learning_rate": 0.0001, "elbo": 3.51947, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.050844475721323}
{"step": 69080, "loss": 3.515655517578125, "grad_norm": 0.7963288426399231, "learning_rate": 0.0001, "elbo": 3.51566, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.051137696457894}
{"step": 69080, "loss": 3.515655517578125, "grad_norm": 0.7963288426399231, "learning_rate": 0.0001, "elbo": 3.51566, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.051137696457894}
{"step": 69085, "loss": 3.5164676666259767, "grad_norm": 0.6803886890411377, "learning_rate": 0.0001, "elbo": 3.51647, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.051430917194464}
{"step": 69085, "loss": 3.5164676666259767, "grad_norm": 0.6803886890411377, "learning_rate": 0.0001, "elbo": 3.51647, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.051430917194464}
{"step": 69090, "loss": 3.533547210693359, "grad_norm": 0.686433732509613, "learning_rate": 0.0001, "elbo": 3.53355, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.051724137931035}
{"step": 69095, "loss": 3.5252616882324217, "grad_norm": 0.8099698424339294, "learning_rate": 0.0001, "elbo": 3.52526, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.052017358667605}
{"step": 69100, "loss": 3.5417076110839845, "grad_norm": 0.6754162907600403, "learning_rate": 0.0001, "elbo": 3.54171, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.052310579404176}
{"step": 69105, "loss": 3.507893371582031, "grad_norm": 0.6699059009552002, "learning_rate": 0.0001, "elbo": 3.50789, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.052603800140746}
{"step": 69110, "loss": 3.537479782104492, "grad_norm": 1.0308008193969727, "learning_rate": 0.0001, "elbo": 3.53748, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0528970208773165}
{"step": 69115, "loss": 3.524504852294922, "grad_norm": 0.7023804783821106, "learning_rate": 0.0001, "elbo": 3.5245, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.053190241613887}
{"step": 69120, "loss": 3.529927062988281, "grad_norm": 0.6770064830780029, "learning_rate": 0.0001, "elbo": 3.52993, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.053483462350457}
{"step": 69125, "loss": 3.524980926513672, "grad_norm": 0.6875690221786499, "learning_rate": 0.0001, "elbo": 3.52498, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.053776683087028}
{"step": 69130, "loss": 3.517778015136719, "grad_norm": 0.6754056811332703, "learning_rate": 0.0001, "elbo": 3.51778, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.054069903823598}
{"step": 69135, "loss": 3.540113067626953, "grad_norm": 0.7234363555908203, "learning_rate": 0.0001, "elbo": 3.54011, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.054363124560169}
{"step": 69140, "loss": 3.5560157775878904, "grad_norm": 0.7060703039169312, "learning_rate": 0.0001, "elbo": 3.55602, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.054656345296739}
{"step": 69145, "loss": 3.5098506927490236, "grad_norm": 0.8998488783836365, "learning_rate": 0.0001, "elbo": 3.50985, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.05494956603331}
{"step": 69150, "loss": 3.529701995849609, "grad_norm": 0.7976580262184143, "learning_rate": 0.0001, "elbo": 3.5297, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.05524278676988}
{"step": 69155, "loss": 3.5129104614257813, "grad_norm": 0.6879797577857971, "learning_rate": 0.0001, "elbo": 3.51291, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.055536007506451}
{"step": 69160, "loss": 3.52118034362793, "grad_norm": 0.7205038666725159, "learning_rate": 0.0001, "elbo": 3.52118, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.055829228243021}
{"step": 69160, "loss": 3.52118034362793, "grad_norm": 0.7205038666725159, "learning_rate": 0.0001, "elbo": 3.52118, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.055829228243021}
{"step": 69165, "loss": 3.496438980102539, "grad_norm": 0.7030870318412781, "learning_rate": 0.0001, "elbo": 3.49644, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0561224489795915}
{"step": 69170, "loss": 3.549348831176758, "grad_norm": 0.8456965684890747, "learning_rate": 0.0001, "elbo": 3.54935, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.056415669716162}
{"step": 69175, "loss": 3.51007194519043, "grad_norm": 0.6760765314102173, "learning_rate": 0.0001, "elbo": 3.51007, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.056708890452732}
{"step": 69180, "loss": 3.5293563842773437, "grad_norm": 0.7705157399177551, "learning_rate": 0.0001, "elbo": 3.52936, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.057002111189303}
{"step": 69185, "loss": 3.53001708984375, "grad_norm": 0.793933629989624, "learning_rate": 0.0001, "elbo": 3.53002, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.057295331925874}
{"step": 69190, "loss": 3.512572479248047, "grad_norm": 0.6910988688468933, "learning_rate": 0.0001, "elbo": 3.51257, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.057588552662445}
{"step": 69195, "loss": 3.527876281738281, "grad_norm": 0.9113794565200806, "learning_rate": 0.0001, "elbo": 3.52788, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.057881773399015}
{"step": 69200, "loss": 3.527788543701172, "grad_norm": 0.9821180701255798, "learning_rate": 0.0001, "elbo": 3.52779, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.058174994135586}
{"step": 69205, "loss": 3.5242523193359374, "grad_norm": 0.6841451525688171, "learning_rate": 0.0001, "elbo": 3.52425, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.058468214872156}
{"step": 69210, "loss": 3.519919586181641, "grad_norm": 0.6962699890136719, "learning_rate": 0.0001, "elbo": 3.51992, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0587614356087265}
{"step": 69215, "loss": 3.518141174316406, "grad_norm": 0.8217174410820007, "learning_rate": 0.0001, "elbo": 3.51814, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.059054656345297}
{"step": 69220, "loss": 3.522151565551758, "grad_norm": 0.7055100798606873, "learning_rate": 0.0001, "elbo": 3.52215, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.059347877081867}
{"step": 69225, "loss": 3.5195613861083985, "grad_norm": 0.7070708870887756, "learning_rate": 0.0001, "elbo": 3.51956, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.059641097818438}
{"step": 69230, "loss": 3.5227802276611326, "grad_norm": 0.6968700885772705, "learning_rate": 0.0001, "elbo": 3.52278, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.059934318555008}
{"step": 69235, "loss": 3.519181823730469, "grad_norm": 0.759323000907898, "learning_rate": 0.0001, "elbo": 3.51918, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.060227539291579}
{"step": 69240, "loss": 3.53486328125, "grad_norm": 0.7238364219665527, "learning_rate": 0.0001, "elbo": 3.53486, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.060520760028149}
{"step": 69245, "loss": 3.5306472778320312, "grad_norm": 0.701674222946167, "learning_rate": 0.0001, "elbo": 3.53065, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.06081398076472}
{"step": 69250, "loss": 3.5195392608642577, "grad_norm": 0.690642774105072, "learning_rate": 0.0001, "elbo": 3.51954, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.06110720150129}
{"step": 69255, "loss": 3.5377124786376952, "grad_norm": 0.6902784109115601, "learning_rate": 0.0001, "elbo": 3.53771, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.061400422237861}
{"step": 69260, "loss": 3.5274276733398438, "grad_norm": 0.6743520498275757, "learning_rate": 0.0001, "elbo": 3.52743, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.061693642974431}
{"step": 69265, "loss": 3.515196990966797, "grad_norm": 0.6681549549102783, "learning_rate": 0.0001, "elbo": 3.5152, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0619868637110015}
{"step": 69270, "loss": 3.511937713623047, "grad_norm": 0.6719506978988647, "learning_rate": 0.0001, "elbo": 3.51194, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.062280084447572}
{"step": 69275, "loss": 3.508169937133789, "grad_norm": 0.9795756936073303, "learning_rate": 0.0001, "elbo": 3.50817, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.062573305184142}
{"step": 69280, "loss": 3.5225906372070312, "grad_norm": 0.7097118496894836, "learning_rate": 0.0001, "elbo": 3.52259, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.062866525920713}
{"step": 69285, "loss": 3.52572021484375, "grad_norm": 0.7039212584495544, "learning_rate": 0.0001, "elbo": 3.52572, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.063159746657283}
{"step": 69290, "loss": 3.5417236328125, "grad_norm": 0.6906551718711853, "learning_rate": 0.0001, "elbo": 3.54172, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.063452967393854}
{"step": 69295, "loss": 3.5257835388183594, "grad_norm": 0.7722495794296265, "learning_rate": 0.0001, "elbo": 3.52578, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.063746188130424}
{"step": 69300, "loss": 3.5333595275878906, "grad_norm": 0.861769437789917, "learning_rate": 0.0001, "elbo": 3.53336, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.064039408866995}
{"step": 69305, "loss": 3.50976448059082, "grad_norm": 0.6720251441001892, "learning_rate": 0.0001, "elbo": 3.50976, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.064332629603566}
{"step": 69310, "loss": 3.53282470703125, "grad_norm": 0.6843902468681335, "learning_rate": 0.0001, "elbo": 3.53282, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0646258503401365}
{"step": 69315, "loss": 3.5313743591308593, "grad_norm": 0.7257785201072693, "learning_rate": 0.0001, "elbo": 3.53137, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.064919071076707}
{"step": 69320, "loss": 3.532628631591797, "grad_norm": 0.7479230165481567, "learning_rate": 0.0001, "elbo": 3.53263, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.065212291813277}
{"step": 69320, "loss": 3.532628631591797, "grad_norm": 0.7479230165481567, "learning_rate": 0.0001, "elbo": 3.53263, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.065212291813277}
{"step": 69325, "loss": 3.524200439453125, "grad_norm": 0.6893579363822937, "learning_rate": 0.0001, "elbo": 3.5242, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.065505512549848}
{"step": 69330, "loss": 3.5061744689941405, "grad_norm": 0.7792049050331116, "learning_rate": 0.0001, "elbo": 3.50617, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.065798733286418}
{"step": 69335, "loss": 3.5417686462402345, "grad_norm": 0.7023629546165466, "learning_rate": 0.0001, "elbo": 3.54177, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.066091954022989}
{"step": 69340, "loss": 3.5354049682617186, "grad_norm": 0.7083736658096313, "learning_rate": 0.0001, "elbo": 3.53541, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.066385174759559}
{"step": 69345, "loss": 3.539631652832031, "grad_norm": 0.7542581558227539, "learning_rate": 0.0001, "elbo": 3.53963, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.06667839549613}
{"step": 69350, "loss": 3.5180027008056642, "grad_norm": 0.7335514426231384, "learning_rate": 0.0001, "elbo": 3.518, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0669716162327}
{"step": 69350, "loss": 3.5180027008056642, "grad_norm": 0.7335514426231384, "learning_rate": 0.0001, "elbo": 3.518, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0669716162327}
{"step": 69355, "loss": 3.5354263305664064, "grad_norm": 0.8030176758766174, "learning_rate": 0.0001, "elbo": 3.53543, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0672648369692705}
{"step": 69360, "loss": 3.5211090087890624, "grad_norm": 0.7616880536079407, "learning_rate": 0.0001, "elbo": 3.52111, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.067558057705841}
{"step": 69365, "loss": 3.5076026916503906, "grad_norm": 0.7218407988548279, "learning_rate": 0.0001, "elbo": 3.5076, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0678512784424115}
{"step": 69370, "loss": 3.523178482055664, "grad_norm": 0.829444944858551, "learning_rate": 0.0001, "elbo": 3.52318, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.068144499178982}
{"step": 69375, "loss": 3.5301666259765625, "grad_norm": 0.6755050420761108, "learning_rate": 0.0001, "elbo": 3.53017, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.068437719915552}
{"step": 69375, "loss": 3.5301666259765625, "grad_norm": 0.6755050420761108, "learning_rate": 0.0001, "elbo": 3.53017, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.068437719915552}
{"step": 69380, "loss": 3.5299148559570312, "grad_norm": 0.7679403424263, "learning_rate": 0.0001, "elbo": 3.52991, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.068730940652123}
{"step": 69385, "loss": 3.52032585144043, "grad_norm": 0.8614417314529419, "learning_rate": 0.0001, "elbo": 3.52033, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.069024161388693}
{"step": 69390, "loss": 3.5335582733154296, "grad_norm": 0.7532561421394348, "learning_rate": 0.0001, "elbo": 3.53356, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.069317382125264}
{"step": 69395, "loss": 3.5223506927490233, "grad_norm": 0.7031834721565247, "learning_rate": 0.0001, "elbo": 3.52235, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.069610602861834}
{"step": 69400, "loss": 3.528711700439453, "grad_norm": 0.6865445375442505, "learning_rate": 0.0001, "elbo": 3.52871, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.069903823598405}
{"step": 69405, "loss": 3.5145023345947264, "grad_norm": 0.6993146538734436, "learning_rate": 0.0001, "elbo": 3.5145, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.070197044334975}
{"step": 69410, "loss": 3.518083953857422, "grad_norm": 0.7134683728218079, "learning_rate": 0.0001, "elbo": 3.51808, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0704902650715455}
{"step": 69410, "loss": 3.518083953857422, "grad_norm": 0.7134683728218079, "learning_rate": 0.0001, "elbo": 3.51808, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0704902650715455}
{"step": 69415, "loss": 3.525679016113281, "grad_norm": 0.7501569986343384, "learning_rate": 0.0001, "elbo": 3.52568, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.070783485808116}
{"step": 69420, "loss": 3.525669479370117, "grad_norm": 0.6854051351547241, "learning_rate": 0.0001, "elbo": 3.52567, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0710767065446865}
{"step": 69425, "loss": 3.5222713470458986, "grad_norm": 0.6761801242828369, "learning_rate": 0.0001, "elbo": 3.52227, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.071369927281257}
{"step": 69430, "loss": 3.5355186462402344, "grad_norm": 0.6582493782043457, "learning_rate": 0.0001, "elbo": 3.53552, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.071663148017828}
{"step": 69435, "loss": 3.5076751708984375, "grad_norm": 0.6667070388793945, "learning_rate": 0.0001, "elbo": 3.50768, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.071956368754399}
{"step": 69440, "loss": 3.5156394958496096, "grad_norm": 0.7175023555755615, "learning_rate": 0.0001, "elbo": 3.51564, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.072249589490969}
{"step": 69440, "loss": 3.5156394958496096, "grad_norm": 0.7175023555755615, "learning_rate": 0.0001, "elbo": 3.51564, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.072249589490969}
{"step": 69445, "loss": 3.5213958740234377, "grad_norm": 0.7514098286628723, "learning_rate": 0.0001, "elbo": 3.5214, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.07254281022754}
{"step": 69450, "loss": 3.521468734741211, "grad_norm": 0.869396984577179, "learning_rate": 0.0001, "elbo": 3.52147, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.07283603096411}
{"step": 69455, "loss": 3.5225284576416014, "grad_norm": 0.6813545823097229, "learning_rate": 0.0001, "elbo": 3.52253, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0731292517006805}
{"step": 69460, "loss": 3.528788757324219, "grad_norm": 0.773506224155426, "learning_rate": 0.0001, "elbo": 3.52879, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.073422472437251}
{"step": 69465, "loss": 3.5177734375, "grad_norm": 0.7477135062217712, "learning_rate": 0.0001, "elbo": 3.51777, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.073715693173821}
{"step": 69470, "loss": 3.5231094360351562, "grad_norm": 0.7026020884513855, "learning_rate": 0.0001, "elbo": 3.52311, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.074008913910392}
{"step": 69475, "loss": 3.5435455322265623, "grad_norm": 0.7858796715736389, "learning_rate": 0.0001, "elbo": 3.54355, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.074302134646962}
{"step": 69480, "loss": 3.512703704833984, "grad_norm": 0.7206375002861023, "learning_rate": 0.0001, "elbo": 3.5127, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.074595355383533}
{"step": 69485, "loss": 3.530755615234375, "grad_norm": 0.7424150109291077, "learning_rate": 0.0001, "elbo": 3.53076, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.074888576120103}
{"step": 69490, "loss": 3.521570587158203, "grad_norm": 0.6615983843803406, "learning_rate": 0.0001, "elbo": 3.52157, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.075181796856674}
{"step": 69495, "loss": 3.5237350463867188, "grad_norm": 0.7277663946151733, "learning_rate": 0.0001, "elbo": 3.52374, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.075475017593244}
{"step": 69500, "loss": 3.5047378540039062, "grad_norm": 0.7481765151023865, "learning_rate": 0.0001, "elbo": 3.50474, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.075768238329815}
{"step": 69500, "loss": 3.5047378540039062, "grad_norm": 0.7481765151023865, "learning_rate": 0.0001, "elbo": 3.50474, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.075768238329815}
{"step": 69505, "loss": 3.5207042694091797, "grad_norm": 0.7188695073127747, "learning_rate": 0.0001, "elbo": 3.5207, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.076061459066385}
{"step": 69510, "loss": 3.51898193359375, "grad_norm": 0.7187264561653137, "learning_rate": 0.0001, "elbo": 3.51898, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0763546798029555}
{"step": 69515, "loss": 3.526112747192383, "grad_norm": 0.6566768884658813, "learning_rate": 0.0001, "elbo": 3.52611, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.076647900539526}
{"step": 69520, "loss": 3.52563591003418, "grad_norm": 0.7464175224304199, "learning_rate": 0.0001, "elbo": 3.52564, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.076941121276096}
{"step": 69525, "loss": 3.522611618041992, "grad_norm": 0.7106600999832153, "learning_rate": 0.0001, "elbo": 3.52261, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.077234342012667}
{"step": 69530, "loss": 3.5234493255615233, "grad_norm": 0.7313686013221741, "learning_rate": 0.0001, "elbo": 3.52345, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.077527562749237}
{"step": 69535, "loss": 3.5149497985839844, "grad_norm": 0.7278717160224915, "learning_rate": 0.0001, "elbo": 3.51495, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.077820783485808}
{"step": 69540, "loss": 3.5301944732666017, "grad_norm": 0.8335527181625366, "learning_rate": 0.0001, "elbo": 3.53019, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.078114004222378}
{"step": 69545, "loss": 3.5226531982421876, "grad_norm": 0.7503741979598999, "learning_rate": 0.0001, "elbo": 3.52265, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.078407224958949}
{"step": 69550, "loss": 3.525556945800781, "grad_norm": 0.7402135133743286, "learning_rate": 0.0001, "elbo": 3.52556, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.078700445695519}
{"step": 69555, "loss": 3.5211578369140626, "grad_norm": 0.6924052834510803, "learning_rate": 0.0001, "elbo": 3.52116, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0789936664320905}
{"step": 69560, "loss": 3.5229888916015626, "grad_norm": 0.7008086442947388, "learning_rate": 0.0001, "elbo": 3.52299, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.079286887168661}
{"step": 69565, "loss": 3.523079681396484, "grad_norm": 0.7074921131134033, "learning_rate": 0.0001, "elbo": 3.52308, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.079580107905231}
{"step": 69565, "loss": 3.523079681396484, "grad_norm": 0.7074921131134033, "learning_rate": 0.0001, "elbo": 3.52308, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.079580107905231}
{"step": 69570, "loss": 3.5154449462890627, "grad_norm": 0.7253401875495911, "learning_rate": 0.0001, "elbo": 3.51545, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.079873328641802}
{"step": 69570, "loss": 3.5154449462890627, "grad_norm": 0.7253401875495911, "learning_rate": 0.0001, "elbo": 3.51545, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.079873328641802}
{"step": 69575, "loss": 3.5310546875, "grad_norm": 0.7518014311790466, "learning_rate": 0.0001, "elbo": 3.53106, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.080166549378372}
{"step": 69580, "loss": 3.5240074157714845, "grad_norm": 0.7964984774589539, "learning_rate": 0.0001, "elbo": 3.52401, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.080459770114943}
{"step": 69585, "loss": 3.541478729248047, "grad_norm": 0.7070304751396179, "learning_rate": 0.0001, "elbo": 3.54148, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.080752990851513}
{"step": 69590, "loss": 3.52996826171875, "grad_norm": 0.6993497014045715, "learning_rate": 0.0001, "elbo": 3.52997, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.081046211588084}
{"step": 69595, "loss": 3.5193885803222655, "grad_norm": 0.6976282596588135, "learning_rate": 0.0001, "elbo": 3.51939, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.081339432324654}
{"step": 69600, "loss": 3.5224884033203123, "grad_norm": 0.7127286791801453, "learning_rate": 0.0001, "elbo": 3.52249, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.081632653061225}
{"step": 69605, "loss": 3.5342498779296876, "grad_norm": 0.693732500076294, "learning_rate": 0.0001, "elbo": 3.53425, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.081925873797795}
{"step": 69610, "loss": 3.535778045654297, "grad_norm": 1.0012673139572144, "learning_rate": 0.0001, "elbo": 3.53578, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0822190945343655}
{"step": 69615, "loss": 3.508391189575195, "grad_norm": 0.9016138911247253, "learning_rate": 0.0001, "elbo": 3.50839, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.082512315270936}
{"step": 69620, "loss": 3.511321258544922, "grad_norm": 0.8243393898010254, "learning_rate": 0.0001, "elbo": 3.51132, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.082805536007506}
{"step": 69625, "loss": 3.5319427490234374, "grad_norm": 0.7013752460479736, "learning_rate": 0.0001, "elbo": 3.53194, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.083098756744077}
{"step": 69630, "loss": 3.5320911407470703, "grad_norm": 0.7176796197891235, "learning_rate": 0.0001, "elbo": 3.53209, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.083391977480647}
{"step": 69635, "loss": 3.517383575439453, "grad_norm": 0.9245555996894836, "learning_rate": 0.0001, "elbo": 3.51738, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.083685198217218}
{"step": 69640, "loss": 3.5228897094726563, "grad_norm": 0.8026831150054932, "learning_rate": 0.0001, "elbo": 3.52289, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.083978418953788}
{"step": 69645, "loss": 3.5286277770996093, "grad_norm": 0.753970742225647, "learning_rate": 0.0001, "elbo": 3.52863, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.084271639690359}
{"step": 69650, "loss": 3.5309852600097655, "grad_norm": 0.8878111243247986, "learning_rate": 0.0001, "elbo": 3.53099, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.084564860426929}
{"step": 69655, "loss": 3.523751068115234, "grad_norm": 0.8282293677330017, "learning_rate": 0.0001, "elbo": 3.52375, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0848580811635}
{"step": 69660, "loss": 3.5447803497314454, "grad_norm": 0.9275667667388916, "learning_rate": 0.0001, "elbo": 3.54478, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.08515130190007}
{"step": 69665, "loss": 3.5311000823974608, "grad_norm": 0.6979474425315857, "learning_rate": 0.0001, "elbo": 3.5311, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0854445226366405}
{"step": 69670, "loss": 3.5247146606445314, "grad_norm": 0.6806753277778625, "learning_rate": 0.0001, "elbo": 3.52471, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.085737743373211}
{"step": 69675, "loss": 3.5386947631835937, "grad_norm": 0.6915302276611328, "learning_rate": 0.0001, "elbo": 3.53869, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.086030964109781}
{"step": 69680, "loss": 3.5306011199951173, "grad_norm": 0.6656501889228821, "learning_rate": 0.0001, "elbo": 3.5306, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.086324184846353}
{"step": 69685, "loss": 3.5448238372802736, "grad_norm": 0.7071230411529541, "learning_rate": 0.0001, "elbo": 3.54482, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.086617405582923}
{"step": 69690, "loss": 3.5407806396484376, "grad_norm": 0.7439368963241577, "learning_rate": 0.0001, "elbo": 3.54078, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.086910626319494}
{"step": 69695, "loss": 3.5265331268310547, "grad_norm": 0.7496601939201355, "learning_rate": 0.0001, "elbo": 3.52653, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.087203847056064}
{"step": 69700, "loss": 3.5267990112304686, "grad_norm": 0.7072383165359497, "learning_rate": 0.0001, "elbo": 3.5268, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0874970677926346}
{"step": 69705, "loss": 3.501036834716797, "grad_norm": 0.6771000623703003, "learning_rate": 0.0001, "elbo": 3.50104, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.087790288529205}
{"step": 69710, "loss": 3.5086875915527345, "grad_norm": 0.9271690249443054, "learning_rate": 0.0001, "elbo": 3.50869, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0880835092657755}
{"step": 69715, "loss": 3.507513427734375, "grad_norm": 0.7554251551628113, "learning_rate": 0.0001, "elbo": 3.50751, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.088376730002346}
{"step": 69720, "loss": 3.538867950439453, "grad_norm": 0.7191652059555054, "learning_rate": 0.0001, "elbo": 3.53887, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.088669950738916}
{"step": 69725, "loss": 3.524736022949219, "grad_norm": 0.6967228055000305, "learning_rate": 0.0001, "elbo": 3.52474, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.088963171475487}
{"step": 69730, "loss": 3.5216583251953124, "grad_norm": 0.7060099840164185, "learning_rate": 0.0001, "elbo": 3.52166, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.089256392212057}
{"step": 69735, "loss": 3.523362731933594, "grad_norm": 0.6955483555793762, "learning_rate": 0.0001, "elbo": 3.52336, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.089549612948628}
{"step": 69740, "loss": 3.530363845825195, "grad_norm": 0.7202267050743103, "learning_rate": 0.0001, "elbo": 3.53036, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.089842833685198}
{"step": 69745, "loss": 3.528006744384766, "grad_norm": 0.7056312561035156, "learning_rate": 0.0001, "elbo": 3.52801, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.090136054421769}
{"step": 69750, "loss": 3.524039077758789, "grad_norm": 0.816145658493042, "learning_rate": 0.0001, "elbo": 3.52404, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.090429275158339}
{"step": 69750, "loss": 3.524039077758789, "grad_norm": 0.816145658493042, "learning_rate": 0.0001, "elbo": 3.52404, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.090429275158339}
{"step": 69755, "loss": 3.5228992462158204, "grad_norm": 0.713267982006073, "learning_rate": 0.0001, "elbo": 3.5229, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0907224958949095}
{"step": 69760, "loss": 3.526558685302734, "grad_norm": 0.7192965149879456, "learning_rate": 0.0001, "elbo": 3.52656, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.09101571663148}
{"step": 69765, "loss": 3.5210494995117188, "grad_norm": 0.6943581104278564, "learning_rate": 0.0001, "elbo": 3.52105, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0913089373680505}
{"step": 69770, "loss": 3.5365943908691406, "grad_norm": 0.722545862197876, "learning_rate": 0.0001, "elbo": 3.53659, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.091602158104621}
{"step": 69775, "loss": 3.5270614624023438, "grad_norm": 0.6923571228981018, "learning_rate": 0.0001, "elbo": 3.52706, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.091895378841191}
{"step": 69780, "loss": 3.520840072631836, "grad_norm": 0.7253362536430359, "learning_rate": 0.0001, "elbo": 3.52084, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.092188599577762}
{"step": 69785, "loss": 3.5104297637939452, "grad_norm": 0.7812734246253967, "learning_rate": 0.0001, "elbo": 3.51043, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.092481820314332}
{"step": 69790, "loss": 3.5327194213867186, "grad_norm": 0.7192696928977966, "learning_rate": 0.0001, "elbo": 3.53272, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.092775041050903}
{"step": 69795, "loss": 3.53021240234375, "grad_norm": 0.7280863523483276, "learning_rate": 0.0001, "elbo": 3.53021, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.093068261787473}
{"step": 69800, "loss": 3.530648040771484, "grad_norm": 0.7986682653427124, "learning_rate": 0.0001, "elbo": 3.53065, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0933614825240445}
{"step": 69805, "loss": 3.522188186645508, "grad_norm": 0.7409440875053406, "learning_rate": 0.0001, "elbo": 3.52219, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.093654703260615}
{"step": 69810, "loss": 3.5241722106933593, "grad_norm": 0.7042338848114014, "learning_rate": 0.0001, "elbo": 3.52417, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.093947923997185}
{"step": 69815, "loss": 3.533189392089844, "grad_norm": 0.7190090417861938, "learning_rate": 0.0001, "elbo": 3.53319, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.094241144733756}
{"step": 69815, "loss": 3.533189392089844, "grad_norm": 0.7190090417861938, "learning_rate": 0.0001, "elbo": 3.53319, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.094241144733756}
{"step": 69820, "loss": 3.5108707427978514, "grad_norm": 0.7329421043395996, "learning_rate": 0.0001, "elbo": 3.51087, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.094534365470326}
{"step": 69825, "loss": 3.527510070800781, "grad_norm": 0.7151918411254883, "learning_rate": 0.0001, "elbo": 3.52751, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.094827586206897}
{"step": 69830, "loss": 3.5180953979492187, "grad_norm": 0.7080411314964294, "learning_rate": 0.0001, "elbo": 3.5181, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.095120806943467}
{"step": 69835, "loss": 3.538042449951172, "grad_norm": 0.696759045124054, "learning_rate": 0.0001, "elbo": 3.53804, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.095414027680038}
{"step": 69840, "loss": 3.493895721435547, "grad_norm": 0.8454843759536743, "learning_rate": 0.0001, "elbo": 3.4939, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.095707248416608}
{"step": 69845, "loss": 3.516231155395508, "grad_norm": 0.674404501914978, "learning_rate": 0.0001, "elbo": 3.51623, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.096000469153179}
{"step": 69850, "loss": 3.5413719177246095, "grad_norm": 0.7154898047447205, "learning_rate": 0.0001, "elbo": 3.54137, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.096293689889749}
{"step": 69855, "loss": 3.535805511474609, "grad_norm": 0.6938677430152893, "learning_rate": 0.0001, "elbo": 3.53581, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0965869106263195}
{"step": 69860, "loss": 3.503531646728516, "grad_norm": 0.7269615530967712, "learning_rate": 0.0001, "elbo": 3.50353, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.09688013136289}
{"step": 69865, "loss": 3.51524658203125, "grad_norm": 0.7101343870162964, "learning_rate": 0.0001, "elbo": 3.51525, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.09717335209946}
{"step": 69870, "loss": 3.5164608001708983, "grad_norm": 0.792041003704071, "learning_rate": 0.0001, "elbo": 3.51646, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.097466572836031}
{"step": 69870, "loss": 3.5164608001708983, "grad_norm": 0.792041003704071, "learning_rate": 0.0001, "elbo": 3.51646, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.097466572836031}
{"step": 69875, "loss": 3.523626708984375, "grad_norm": 0.7309128046035767, "learning_rate": 0.0001, "elbo": 3.52363, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.097759793572601}
{"step": 69880, "loss": 3.540034866333008, "grad_norm": 0.6604055762290955, "learning_rate": 0.0001, "elbo": 3.54003, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.098053014309172}
{"step": 69885, "loss": 3.528104782104492, "grad_norm": 0.688629150390625, "learning_rate": 0.0001, "elbo": 3.5281, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.098346235045742}
{"step": 69890, "loss": 3.5324573516845703, "grad_norm": 0.7389536499977112, "learning_rate": 0.0001, "elbo": 3.53246, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.098639455782313}
{"step": 69895, "loss": 3.517137145996094, "grad_norm": 0.7164667844772339, "learning_rate": 0.0001, "elbo": 3.51714, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.098932676518883}
{"step": 69900, "loss": 3.526646041870117, "grad_norm": 0.6754703521728516, "learning_rate": 0.0001, "elbo": 3.52665, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.099225897255454}
{"step": 69905, "loss": 3.5205440521240234, "grad_norm": 0.7175151109695435, "learning_rate": 0.0001, "elbo": 3.52054, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.099519117992024}
{"step": 69905, "loss": 3.5205440521240234, "grad_norm": 0.7175151109695435, "learning_rate": 0.0001, "elbo": 3.52054, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.099519117992024}
{"step": 69910, "loss": 3.5105941772460936, "grad_norm": 0.7546835541725159, "learning_rate": 0.0001, "elbo": 3.51059, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.0998123387285945}
{"step": 69915, "loss": 3.5262542724609376, "grad_norm": 0.7789263725280762, "learning_rate": 0.0001, "elbo": 3.52625, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.100105559465165}
{"step": 69920, "loss": 3.5329822540283202, "grad_norm": 0.6736083030700684, "learning_rate": 0.0001, "elbo": 3.53298, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.100398780201736}
{"step": 69925, "loss": 3.509585952758789, "grad_norm": 0.7032128572463989, "learning_rate": 0.0001, "elbo": 3.50959, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.100692000938307}
{"step": 69925, "loss": 3.509585952758789, "grad_norm": 0.7032128572463989, "learning_rate": 0.0001, "elbo": 3.50959, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.100692000938307}
{"step": 69930, "loss": 3.5376632690429686, "grad_norm": 0.9794004559516907, "learning_rate": 0.0001, "elbo": 3.53766, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.100985221674877}
{"step": 69935, "loss": 3.5065532684326173, "grad_norm": 0.7735645771026611, "learning_rate": 0.0001, "elbo": 3.50655, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.101278442411448}
{"step": 69940, "loss": 3.5224365234375, "grad_norm": 0.7266467213630676, "learning_rate": 0.0001, "elbo": 3.52244, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.101571663148018}
{"step": 69945, "loss": 3.5415508270263674, "grad_norm": 0.900154709815979, "learning_rate": 0.0001, "elbo": 3.54155, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.101864883884589}
{"step": 69950, "loss": 3.533507537841797, "grad_norm": 0.7476345896720886, "learning_rate": 0.0001, "elbo": 3.53351, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.102158104621159}
{"step": 69955, "loss": 3.516704559326172, "grad_norm": 0.7211520075798035, "learning_rate": 0.0001, "elbo": 3.5167, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1024513253577295}
{"step": 69960, "loss": 3.5202327728271485, "grad_norm": 0.7424417734146118, "learning_rate": 0.0001, "elbo": 3.52023, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1027445460943}
{"step": 69965, "loss": 3.5422904968261717, "grad_norm": 0.7912165522575378, "learning_rate": 0.0001, "elbo": 3.54229, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.10303776683087}
{"step": 69970, "loss": 3.5451465606689454, "grad_norm": 0.7299239039421082, "learning_rate": 0.0001, "elbo": 3.54515, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.103330987567441}
{"step": 69975, "loss": 3.5078826904296876, "grad_norm": 0.7699220180511475, "learning_rate": 0.0001, "elbo": 3.50788, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.103624208304011}
{"step": 69980, "loss": 3.510224533081055, "grad_norm": 0.6777770519256592, "learning_rate": 0.0001, "elbo": 3.51022, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.103917429040582}
{"step": 69985, "loss": 3.5258773803710937, "grad_norm": 0.6732104420661926, "learning_rate": 0.0001, "elbo": 3.52588, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.104210649777152}
{"step": 69990, "loss": 3.5204139709472657, "grad_norm": 0.7053664922714233, "learning_rate": 0.0001, "elbo": 3.52041, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.104503870513723}
{"step": 69995, "loss": 3.5214702606201174, "grad_norm": 0.7387738823890686, "learning_rate": 0.0001, "elbo": 3.52147, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.104797091250293}
{"step": 70000, "loss": 3.5189865112304686, "grad_norm": 0.7808171510696411, "learning_rate": 0.0001, "elbo": 3.51899, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.105090311986864}
{"step": 70005, "loss": 3.5201263427734375, "grad_norm": 0.7463241815567017, "learning_rate": 0.0001, "elbo": 3.52013, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.105383532723434}
{"step": 70010, "loss": 3.524971771240234, "grad_norm": 0.9291953444480896, "learning_rate": 0.0001, "elbo": 3.52497, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1056767534600045}
{"step": 70010, "loss": 3.524971771240234, "grad_norm": 0.9291953444480896, "learning_rate": 0.0001, "elbo": 3.52497, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1056767534600045}
{"step": 70015, "loss": 3.4989768981933596, "grad_norm": 0.711214542388916, "learning_rate": 0.0001, "elbo": 3.49898, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.105969974196575}
{"step": 70020, "loss": 3.508757400512695, "grad_norm": 0.6858602166175842, "learning_rate": 0.0001, "elbo": 3.50876, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.106263194933145}
{"step": 70025, "loss": 3.5406681060791017, "grad_norm": 0.7122630476951599, "learning_rate": 0.0001, "elbo": 3.54067, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.106556415669716}
{"step": 70030, "loss": 3.525290298461914, "grad_norm": 0.7183310985565186, "learning_rate": 0.0001, "elbo": 3.52529, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.106849636406286}
{"step": 70035, "loss": 3.5165641784667967, "grad_norm": 0.7333645224571228, "learning_rate": 0.0001, "elbo": 3.51656, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.107142857142857}
{"step": 70040, "loss": 3.53565673828125, "grad_norm": 0.930014431476593, "learning_rate": 0.0001, "elbo": 3.53566, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.107436077879427}
{"step": 70045, "loss": 3.5271240234375, "grad_norm": 1.0887949466705322, "learning_rate": 0.0001, "elbo": 3.52712, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.107729298615999}
{"step": 70050, "loss": 3.517079162597656, "grad_norm": 0.7430993318557739, "learning_rate": 0.0001, "elbo": 3.51708, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.108022519352569}
{"step": 70055, "loss": 3.5154354095458986, "grad_norm": 0.8666597008705139, "learning_rate": 0.0001, "elbo": 3.51544, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1083157400891395}
{"step": 70060, "loss": 3.520001220703125, "grad_norm": 0.683289647102356, "learning_rate": 0.0001, "elbo": 3.52, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.10860896082571}
{"step": 70065, "loss": 3.5230026245117188, "grad_norm": 0.7000361680984497, "learning_rate": 0.0001, "elbo": 3.523, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.10890218156228}
{"step": 70070, "loss": 3.5137710571289062, "grad_norm": 0.6950773596763611, "learning_rate": 0.0001, "elbo": 3.51377, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.109195402298851}
{"step": 70075, "loss": 3.5245494842529297, "grad_norm": 0.7534593939781189, "learning_rate": 0.0001, "elbo": 3.52455, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.109488623035421}
{"step": 70080, "loss": 3.518904113769531, "grad_norm": 0.7143594026565552, "learning_rate": 0.0001, "elbo": 3.5189, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.109781843771992}
{"step": 70085, "loss": 3.5190189361572264, "grad_norm": 0.6780999898910522, "learning_rate": 0.0001, "elbo": 3.51902, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.110075064508562}
{"step": 70090, "loss": 3.5280521392822264, "grad_norm": 0.7211626768112183, "learning_rate": 0.0001, "elbo": 3.52805, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.110368285245133}
{"step": 70095, "loss": 3.5218505859375, "grad_norm": 0.6838076710700989, "learning_rate": 0.0001, "elbo": 3.52185, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.110661505981703}
{"step": 70100, "loss": 3.510961151123047, "grad_norm": 0.7039856314659119, "learning_rate": 0.0001, "elbo": 3.51096, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1109547267182736}
{"step": 70105, "loss": 3.527882385253906, "grad_norm": 0.7114004492759705, "learning_rate": 0.0001, "elbo": 3.52788, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.111247947454844}
{"step": 70110, "loss": 3.5137908935546873, "grad_norm": 0.704963207244873, "learning_rate": 0.0001, "elbo": 3.51379, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1115411681914145}
{"step": 70115, "loss": 3.512940216064453, "grad_norm": 0.6753562688827515, "learning_rate": 0.0001, "elbo": 3.51294, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.111834388927985}
{"step": 70120, "loss": 3.5338699340820314, "grad_norm": 0.7833207249641418, "learning_rate": 0.0001, "elbo": 3.53387, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.112127609664555}
{"step": 70125, "loss": 3.5180923461914064, "grad_norm": 0.8065345287322998, "learning_rate": 0.0001, "elbo": 3.51809, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.112420830401126}
{"step": 70130, "loss": 3.5145286560058593, "grad_norm": 0.6950792074203491, "learning_rate": 0.0001, "elbo": 3.51453, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.112714051137696}
{"step": 70135, "loss": 3.536016082763672, "grad_norm": 0.7680288553237915, "learning_rate": 0.0001, "elbo": 3.53602, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.113007271874267}
{"step": 70140, "loss": 3.5313671112060545, "grad_norm": 0.7040247321128845, "learning_rate": 0.0001, "elbo": 3.53137, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.113300492610837}
{"step": 70145, "loss": 3.5192100524902346, "grad_norm": 0.7457690238952637, "learning_rate": 0.0001, "elbo": 3.51921, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.113593713347408}
{"step": 70150, "loss": 3.5012271881103514, "grad_norm": 0.7489700317382812, "learning_rate": 0.0001, "elbo": 3.50123, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.113886934083978}
{"step": 70155, "loss": 3.5163074493408204, "grad_norm": 1.3448982238769531, "learning_rate": 0.0001, "elbo": 3.51631, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1141801548205486}
{"step": 70160, "loss": 3.5210525512695314, "grad_norm": 0.6977010369300842, "learning_rate": 0.0001, "elbo": 3.52105, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.114473375557119}
{"step": 70160, "loss": 3.5210525512695314, "grad_norm": 0.6977010369300842, "learning_rate": 0.0001, "elbo": 3.52105, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.114473375557119}
{"step": 70165, "loss": 3.5158226013183596, "grad_norm": 0.6794028878211975, "learning_rate": 0.0001, "elbo": 3.51582, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1147665962936895}
{"step": 70170, "loss": 3.513874053955078, "grad_norm": 1.0688462257385254, "learning_rate": 0.0001, "elbo": 3.51387, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.115059817030261}
{"step": 70175, "loss": 3.5243209838867187, "grad_norm": 0.6673528552055359, "learning_rate": 0.0001, "elbo": 3.52432, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.115353037766831}
{"step": 70180, "loss": 3.523653030395508, "grad_norm": 0.6761338710784912, "learning_rate": 0.0001, "elbo": 3.52365, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.115646258503402}
{"step": 70185, "loss": 3.5279396057128904, "grad_norm": 0.7515548467636108, "learning_rate": 0.0001, "elbo": 3.52794, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.115939479239972}
{"step": 70190, "loss": 3.5225894927978514, "grad_norm": 0.7614634037017822, "learning_rate": 0.0001, "elbo": 3.52259, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.116232699976543}
{"step": 70195, "loss": 3.524253082275391, "grad_norm": 0.702566921710968, "learning_rate": 0.0001, "elbo": 3.52425, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.116525920713113}
{"step": 70200, "loss": 3.516238784790039, "grad_norm": 0.7435930967330933, "learning_rate": 0.0001, "elbo": 3.51624, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1168191414496835}
{"step": 70205, "loss": 3.519241714477539, "grad_norm": 0.726240336894989, "learning_rate": 0.0001, "elbo": 3.51924, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.117112362186254}
{"step": 70210, "loss": 3.5097381591796877, "grad_norm": 0.7030265927314758, "learning_rate": 0.0001, "elbo": 3.50974, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.117405582922824}
{"step": 70215, "loss": 3.5373268127441406, "grad_norm": 0.7144138216972351, "learning_rate": 0.0001, "elbo": 3.53733, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.117698803659395}
{"step": 70220, "loss": 3.541962432861328, "grad_norm": 0.8556627035140991, "learning_rate": 0.0001, "elbo": 3.54196, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.117992024395965}
{"step": 70225, "loss": 3.520923614501953, "grad_norm": 0.72356116771698, "learning_rate": 0.0001, "elbo": 3.52092, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.118285245132536}
{"step": 70230, "loss": 3.5386016845703123, "grad_norm": 0.7151498794555664, "learning_rate": 0.0001, "elbo": 3.5386, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.118578465869106}
{"step": 70235, "loss": 3.522787094116211, "grad_norm": 0.7271479368209839, "learning_rate": 0.0001, "elbo": 3.52279, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.118871686605677}
{"step": 70240, "loss": 3.5191463470458983, "grad_norm": 0.7594993710517883, "learning_rate": 0.0001, "elbo": 3.51915, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.119164907342247}
{"step": 70245, "loss": 3.493989944458008, "grad_norm": 0.7294753789901733, "learning_rate": 0.0001, "elbo": 3.49399, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.119458128078818}
{"step": 70250, "loss": 3.524156951904297, "grad_norm": 0.6783348917961121, "learning_rate": 0.0001, "elbo": 3.52416, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.119751348815388}
{"step": 70255, "loss": 3.52874755859375, "grad_norm": 0.740507960319519, "learning_rate": 0.0001, "elbo": 3.52875, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1200445695519585}
{"step": 70260, "loss": 3.523398590087891, "grad_norm": 0.6778705716133118, "learning_rate": 0.0001, "elbo": 3.5234, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.120337790288529}
{"step": 70265, "loss": 3.5244251251220704, "grad_norm": 0.7257683277130127, "learning_rate": 0.0001, "elbo": 3.52443, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.120631011025099}
{"step": 70270, "loss": 3.52938232421875, "grad_norm": 0.8008663058280945, "learning_rate": 0.0001, "elbo": 3.52938, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.12092423176167}
{"step": 70275, "loss": 3.514692687988281, "grad_norm": 0.7293993234634399, "learning_rate": 0.0001, "elbo": 3.51469, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.12121745249824}
{"step": 70275, "loss": 3.514692687988281, "grad_norm": 0.7293993234634399, "learning_rate": 0.0001, "elbo": 3.51469, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.12121745249824}
{"step": 70280, "loss": 3.523375701904297, "grad_norm": 0.7033304572105408, "learning_rate": 0.0001, "elbo": 3.52338, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.121510673234811}
{"step": 70285, "loss": 3.513602066040039, "grad_norm": 0.6606383919715881, "learning_rate": 0.0001, "elbo": 3.5136, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.121803893971381}
{"step": 70290, "loss": 3.5338905334472654, "grad_norm": 0.7373203635215759, "learning_rate": 0.0001, "elbo": 3.53389, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.122097114707952}
{"step": 70295, "loss": 3.5318527221679688, "grad_norm": 0.7064128518104553, "learning_rate": 0.0001, "elbo": 3.53185, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.122390335444523}
{"step": 70300, "loss": 3.5199378967285155, "grad_norm": 0.6841039061546326, "learning_rate": 0.0001, "elbo": 3.51994, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1226835561810935}
{"step": 70305, "loss": 3.532335662841797, "grad_norm": 0.7229723930358887, "learning_rate": 0.0001, "elbo": 3.53234, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.122976776917664}
{"step": 70310, "loss": 3.518462371826172, "grad_norm": 0.6581581830978394, "learning_rate": 0.0001, "elbo": 3.51846, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.123269997654234}
{"step": 70315, "loss": 3.514351654052734, "grad_norm": 0.7371236681938171, "learning_rate": 0.0001, "elbo": 3.51435, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.123563218390805}
{"step": 70320, "loss": 3.5171279907226562, "grad_norm": 0.7081913352012634, "learning_rate": 0.0001, "elbo": 3.51713, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.123856439127375}
{"step": 70325, "loss": 3.523085021972656, "grad_norm": 0.7039929032325745, "learning_rate": 0.0001, "elbo": 3.52309, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.124149659863946}
{"step": 70330, "loss": 3.52906494140625, "grad_norm": 0.672927737236023, "learning_rate": 0.0001, "elbo": 3.52907, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.124442880600516}
{"step": 70335, "loss": 3.5081573486328126, "grad_norm": 0.6991022229194641, "learning_rate": 0.0001, "elbo": 3.50816, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.124736101337087}
{"step": 70340, "loss": 3.5175197601318358, "grad_norm": 0.6843193769454956, "learning_rate": 0.0001, "elbo": 3.51752, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.125029322073657}
{"step": 70345, "loss": 3.5379600524902344, "grad_norm": 0.6984013319015503, "learning_rate": 0.0001, "elbo": 3.53796, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.125322542810228}
{"step": 70350, "loss": 3.5452911376953127, "grad_norm": 0.7207199335098267, "learning_rate": 0.0001, "elbo": 3.54529, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.125615763546798}
{"step": 70355, "loss": 3.5397777557373047, "grad_norm": 0.6899224519729614, "learning_rate": 0.0001, "elbo": 3.53978, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1259089842833685}
{"step": 70360, "loss": 3.5108669281005858, "grad_norm": 0.6991674900054932, "learning_rate": 0.0001, "elbo": 3.51087, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.126202205019939}
{"step": 70365, "loss": 3.536412811279297, "grad_norm": 0.6955630779266357, "learning_rate": 0.0001, "elbo": 3.53641, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.126495425756509}
{"step": 70370, "loss": 3.51961669921875, "grad_norm": 0.950329601764679, "learning_rate": 0.0001, "elbo": 3.51962, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.12678864649308}
{"step": 70375, "loss": 3.5324974060058594, "grad_norm": 0.7231833934783936, "learning_rate": 0.0001, "elbo": 3.5325, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.12708186722965}
{"step": 70380, "loss": 3.517561340332031, "grad_norm": 0.7082018256187439, "learning_rate": 0.0001, "elbo": 3.51756, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.127375087966221}
{"step": 70385, "loss": 3.5282608032226563, "grad_norm": 0.7177559733390808, "learning_rate": 0.0001, "elbo": 3.52826, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.127668308702791}
{"step": 70385, "loss": 3.5282608032226563, "grad_norm": 0.7177559733390808, "learning_rate": 0.0001, "elbo": 3.52826, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.127668308702791}
{"step": 70390, "loss": 3.5297584533691406, "grad_norm": 0.6762408018112183, "learning_rate": 0.0001, "elbo": 3.52976, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.127961529439362}
{"step": 70395, "loss": 3.5175567626953126, "grad_norm": 0.7300204038619995, "learning_rate": 0.0001, "elbo": 3.51756, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.128254750175932}
{"step": 70400, "loss": 3.5259674072265623, "grad_norm": 0.9004344940185547, "learning_rate": 0.0001, "elbo": 3.52597, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.128547970912503}
{"step": 70405, "loss": 3.5329856872558594, "grad_norm": 0.7028091549873352, "learning_rate": 0.0001, "elbo": 3.53299, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.128841191649073}
{"step": 70410, "loss": 3.529534912109375, "grad_norm": 0.6948854327201843, "learning_rate": 0.0001, "elbo": 3.52953, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1291344123856435}
{"step": 70415, "loss": 3.5306957244873045, "grad_norm": 0.8089805841445923, "learning_rate": 0.0001, "elbo": 3.5307, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.129427633122214}
{"step": 70420, "loss": 3.5311607360839843, "grad_norm": 0.6817411780357361, "learning_rate": 0.0001, "elbo": 3.53116, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.129720853858785}
{"step": 70425, "loss": 3.506867218017578, "grad_norm": 0.763945460319519, "learning_rate": 0.0001, "elbo": 3.50687, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.130014074595356}
{"step": 70430, "loss": 3.5264919281005858, "grad_norm": 1.1206281185150146, "learning_rate": 0.0001, "elbo": 3.52649, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.130307295331926}
{"step": 70435, "loss": 3.5447242736816404, "grad_norm": 0.7351824641227722, "learning_rate": 0.0001, "elbo": 3.54472, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.130600516068497}
{"step": 70440, "loss": 3.521007537841797, "grad_norm": 0.6786352396011353, "learning_rate": 0.0001, "elbo": 3.52101, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.130893736805067}
{"step": 70445, "loss": 3.518060302734375, "grad_norm": 0.6928900480270386, "learning_rate": 0.0001, "elbo": 3.51806, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.131186957541638}
{"step": 70450, "loss": 3.5317882537841796, "grad_norm": 1.1185106039047241, "learning_rate": 0.0001, "elbo": 3.53179, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.131480178278208}
{"step": 70455, "loss": 3.517567443847656, "grad_norm": 0.788202702999115, "learning_rate": 0.0001, "elbo": 3.51757, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1317733990147785}
{"step": 70460, "loss": 3.5243492126464844, "grad_norm": 0.7405319809913635, "learning_rate": 0.0001, "elbo": 3.52435, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.132066619751349}
{"step": 70465, "loss": 3.525206756591797, "grad_norm": 0.7000473141670227, "learning_rate": 0.0001, "elbo": 3.52521, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.132359840487919}
{"step": 70470, "loss": 3.5215080261230467, "grad_norm": 0.765515923500061, "learning_rate": 0.0001, "elbo": 3.52151, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.13265306122449}
{"step": 70475, "loss": 3.5213241577148438, "grad_norm": 0.7192223072052002, "learning_rate": 0.0001, "elbo": 3.52132, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.13294628196106}
{"step": 70480, "loss": 3.524492645263672, "grad_norm": 0.6890037059783936, "learning_rate": 0.0001, "elbo": 3.52449, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.133239502697631}
{"step": 70485, "loss": 3.518191909790039, "grad_norm": 0.7114001512527466, "learning_rate": 0.0001, "elbo": 3.51819, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.133532723434201}
{"step": 70490, "loss": 3.523351287841797, "grad_norm": 0.7429178953170776, "learning_rate": 0.0001, "elbo": 3.52335, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.133825944170772}
{"step": 70495, "loss": 3.5213241577148438, "grad_norm": 0.7099688053131104, "learning_rate": 0.0001, "elbo": 3.52132, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.134119164907342}
{"step": 70500, "loss": 3.5276966094970703, "grad_norm": 0.7605887055397034, "learning_rate": 0.0001, "elbo": 3.5277, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.134412385643913}
{"step": 70505, "loss": 3.5068702697753906, "grad_norm": 0.7405754923820496, "learning_rate": 0.0001, "elbo": 3.50687, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.134705606380483}
{"step": 70510, "loss": 3.5369941711425783, "grad_norm": 0.7879779934883118, "learning_rate": 0.0001, "elbo": 3.53699, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1349988271170535}
{"step": 70515, "loss": 3.5079067230224608, "grad_norm": 0.7725157737731934, "learning_rate": 0.0001, "elbo": 3.50791, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.135292047853624}
{"step": 70520, "loss": 3.497611236572266, "grad_norm": 0.81959068775177, "learning_rate": 0.0001, "elbo": 3.49761, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.135585268590194}
{"step": 70525, "loss": 3.534649658203125, "grad_norm": 0.820609450340271, "learning_rate": 0.0001, "elbo": 3.53465, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.135878489326765}
{"step": 70530, "loss": 3.5280380249023438, "grad_norm": 0.7626843452453613, "learning_rate": 0.0001, "elbo": 3.52804, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.136171710063335}
{"step": 70535, "loss": 3.520232391357422, "grad_norm": 0.757651686668396, "learning_rate": 0.0001, "elbo": 3.52023, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.136464930799907}
{"step": 70540, "loss": 3.5179637908935546, "grad_norm": 0.7044602632522583, "learning_rate": 0.0001, "elbo": 3.51796, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.136758151536477}
{"step": 70545, "loss": 3.5170143127441404, "grad_norm": 0.7171953916549683, "learning_rate": 0.0001, "elbo": 3.51701, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1370513722730475}
{"step": 70550, "loss": 3.5065944671630858, "grad_norm": 0.7659813165664673, "learning_rate": 0.0001, "elbo": 3.50659, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.137344593009618}
{"step": 70555, "loss": 3.519975280761719, "grad_norm": 0.7843422293663025, "learning_rate": 0.0001, "elbo": 3.51998, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.137637813746188}
{"step": 70560, "loss": 3.5323867797851562, "grad_norm": 0.7340086102485657, "learning_rate": 0.0001, "elbo": 3.53239, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.137931034482759}
{"step": 70565, "loss": 3.5180084228515627, "grad_norm": 0.7249903082847595, "learning_rate": 0.0001, "elbo": 3.51801, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.138224255219329}
{"step": 70570, "loss": 3.5376243591308594, "grad_norm": 0.6985774040222168, "learning_rate": 0.0001, "elbo": 3.53762, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1385174759559}
{"step": 70575, "loss": 3.53092041015625, "grad_norm": 1.2092382907867432, "learning_rate": 0.0001, "elbo": 3.53092, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.13881069669247}
{"step": 70580, "loss": 3.519367980957031, "grad_norm": 0.6978518962860107, "learning_rate": 0.0001, "elbo": 3.51937, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.139103917429041}
{"step": 70585, "loss": 3.5204978942871095, "grad_norm": 0.7028524279594421, "learning_rate": 0.0001, "elbo": 3.5205, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.139397138165611}
{"step": 70590, "loss": 3.522594451904297, "grad_norm": 0.7014788389205933, "learning_rate": 0.0001, "elbo": 3.52259, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.139690358902182}
{"step": 70595, "loss": 3.5121665954589845, "grad_norm": 0.6816940307617188, "learning_rate": 0.0001, "elbo": 3.51217, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.139983579638752}
{"step": 70600, "loss": 3.523773193359375, "grad_norm": 0.7571083307266235, "learning_rate": 0.0001, "elbo": 3.52377, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1402768003753225}
{"step": 70605, "loss": 3.5365966796875, "grad_norm": 0.7112175822257996, "learning_rate": 0.0001, "elbo": 3.5366, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.140570021111893}
{"step": 70610, "loss": 3.507684326171875, "grad_norm": 0.7082422971725464, "learning_rate": 0.0001, "elbo": 3.50768, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.140863241848463}
{"step": 70615, "loss": 3.5359153747558594, "grad_norm": 0.7383677363395691, "learning_rate": 0.0001, "elbo": 3.53592, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.141156462585034}
{"step": 70620, "loss": 3.512896728515625, "grad_norm": 0.697433590888977, "learning_rate": 0.0001, "elbo": 3.5129, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.141449683321604}
{"step": 70625, "loss": 3.509812927246094, "grad_norm": 0.7191446423530579, "learning_rate": 0.0001, "elbo": 3.50981, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.141742904058175}
{"step": 70630, "loss": 3.5110710144042967, "grad_norm": 0.8320949673652649, "learning_rate": 0.0001, "elbo": 3.51107, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.142036124794745}
{"step": 70635, "loss": 3.50987548828125, "grad_norm": 0.7175428867340088, "learning_rate": 0.0001, "elbo": 3.50988, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.142329345531316}
{"step": 70640, "loss": 3.525679016113281, "grad_norm": 0.6895644664764404, "learning_rate": 0.0001, "elbo": 3.52568, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.142622566267886}
{"step": 70645, "loss": 3.5168201446533205, "grad_norm": 0.7459155321121216, "learning_rate": 0.0001, "elbo": 3.51682, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.142915787004457}
{"step": 70650, "loss": 3.501321792602539, "grad_norm": 0.7045091390609741, "learning_rate": 0.0001, "elbo": 3.50132, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.143209007741027}
{"step": 70655, "loss": 3.5182113647460938, "grad_norm": 0.844856858253479, "learning_rate": 0.0001, "elbo": 3.51821, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1435022284775975}
{"step": 70660, "loss": 3.5132156372070313, "grad_norm": 0.7878125309944153, "learning_rate": 0.0001, "elbo": 3.51322, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.143795449214169}
{"step": 70665, "loss": 3.5303268432617188, "grad_norm": 0.7004331350326538, "learning_rate": 0.0001, "elbo": 3.53033, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.144088669950739}
{"step": 70670, "loss": 3.534105682373047, "grad_norm": 0.9852310419082642, "learning_rate": 0.0001, "elbo": 3.53411, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.14438189068731}
{"step": 70675, "loss": 3.5177066802978514, "grad_norm": 0.6783068776130676, "learning_rate": 0.0001, "elbo": 3.51771, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.14467511142388}
{"step": 70680, "loss": 3.5231918334960937, "grad_norm": 0.735512912273407, "learning_rate": 0.0001, "elbo": 3.52319, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.144968332160451}
{"step": 70685, "loss": 3.518108367919922, "grad_norm": 0.703755795955658, "learning_rate": 0.0001, "elbo": 3.51811, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.145261552897021}
{"step": 70690, "loss": 3.5238372802734377, "grad_norm": 0.7013466358184814, "learning_rate": 0.0001, "elbo": 3.52384, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.145554773633592}
{"step": 70695, "loss": 3.515778350830078, "grad_norm": 0.7326196432113647, "learning_rate": 0.0001, "elbo": 3.51578, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.145847994370162}
{"step": 70700, "loss": 3.5169818878173826, "grad_norm": 0.7151956558227539, "learning_rate": 0.0001, "elbo": 3.51698, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1461412151067325}
{"step": 70705, "loss": 3.5228435516357424, "grad_norm": 0.6891708374023438, "learning_rate": 0.0001, "elbo": 3.52284, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.146434435843303}
{"step": 70710, "loss": 3.5279640197753905, "grad_norm": 0.7143595218658447, "learning_rate": 0.0001, "elbo": 3.52796, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.146727656579873}
{"step": 70715, "loss": 3.51829833984375, "grad_norm": 0.6794449090957642, "learning_rate": 0.0001, "elbo": 3.5183, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.147020877316444}
{"step": 70720, "loss": 3.5097930908203123, "grad_norm": 0.6941893696784973, "learning_rate": 0.0001, "elbo": 3.50979, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.147314098053014}
{"step": 70725, "loss": 3.515493392944336, "grad_norm": 0.681702733039856, "learning_rate": 0.0001, "elbo": 3.51549, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.147607318789585}
{"step": 70730, "loss": 3.5400081634521485, "grad_norm": 0.8970866799354553, "learning_rate": 0.0001, "elbo": 3.54001, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.147900539526155}
{"step": 70735, "loss": 3.515869140625, "grad_norm": 0.6918036937713623, "learning_rate": 0.0001, "elbo": 3.51587, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.148193760262726}
{"step": 70740, "loss": 3.5245502471923826, "grad_norm": 0.6705933809280396, "learning_rate": 0.0001, "elbo": 3.52455, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.148486980999296}
{"step": 70745, "loss": 3.5089691162109373, "grad_norm": 0.6831783056259155, "learning_rate": 0.0001, "elbo": 3.50897, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.148780201735867}
{"step": 70750, "loss": 3.5231143951416017, "grad_norm": 0.7739636301994324, "learning_rate": 0.0001, "elbo": 3.52311, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.149073422472437}
{"step": 70755, "loss": 3.5176712036132813, "grad_norm": 0.7081432342529297, "learning_rate": 0.0001, "elbo": 3.51767, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1493666432090075}
{"step": 70760, "loss": 3.516985321044922, "grad_norm": 0.7445741891860962, "learning_rate": 0.0001, "elbo": 3.51699, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.149659863945578}
{"step": 70765, "loss": 3.5260066986083984, "grad_norm": 0.7830975651741028, "learning_rate": 0.0001, "elbo": 3.52601, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.149953084682148}
{"step": 70770, "loss": 3.520407867431641, "grad_norm": 1.020302653312683, "learning_rate": 0.0001, "elbo": 3.52041, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.150246305418719}
{"step": 70775, "loss": 3.525318908691406, "grad_norm": 0.732803463935852, "learning_rate": 0.0001, "elbo": 3.52532, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.150539526155289}
{"step": 70780, "loss": 3.521017074584961, "grad_norm": 0.7046334743499756, "learning_rate": 0.0001, "elbo": 3.52102, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.15083274689186}
{"step": 70785, "loss": 3.522910308837891, "grad_norm": 0.7132290601730347, "learning_rate": 0.0001, "elbo": 3.52291, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.151125967628431}
{"step": 70790, "loss": 3.536894607543945, "grad_norm": 0.6967388987541199, "learning_rate": 0.0001, "elbo": 3.53689, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.151419188365002}
{"step": 70795, "loss": 3.5259300231933595, "grad_norm": 0.7218701839447021, "learning_rate": 0.0001, "elbo": 3.52593, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.151712409101572}
{"step": 70800, "loss": 3.525818634033203, "grad_norm": 0.7222533226013184, "learning_rate": 0.0001, "elbo": 3.52582, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1520056298381425}
{"step": 70805, "loss": 3.510193634033203, "grad_norm": 0.7237854599952698, "learning_rate": 0.0001, "elbo": 3.51019, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.152298850574713}
{"step": 70810, "loss": 3.518896484375, "grad_norm": 0.7283839583396912, "learning_rate": 0.0001, "elbo": 3.5189, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.152592071311283}
{"step": 70815, "loss": 3.50869140625, "grad_norm": 0.6827893853187561, "learning_rate": 0.0001, "elbo": 3.50869, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.152885292047854}
{"step": 70820, "loss": 3.523184967041016, "grad_norm": 0.6652540564537048, "learning_rate": 0.0001, "elbo": 3.52318, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.153178512784424}
{"step": 70825, "loss": 3.50640869140625, "grad_norm": 0.9198005795478821, "learning_rate": 0.0001, "elbo": 3.50641, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.153471733520995}
{"step": 70830, "loss": 3.5143177032470705, "grad_norm": 0.7562311291694641, "learning_rate": 0.0001, "elbo": 3.51432, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.153764954257565}
{"step": 70835, "loss": 3.5186111450195314, "grad_norm": 0.657437264919281, "learning_rate": 0.0001, "elbo": 3.51861, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.154058174994136}
{"step": 70840, "loss": 3.508727264404297, "grad_norm": 0.6898141503334045, "learning_rate": 0.0001, "elbo": 3.50873, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.154351395730706}
{"step": 70845, "loss": 3.52122802734375, "grad_norm": 0.7418447136878967, "learning_rate": 0.0001, "elbo": 3.52123, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.154644616467277}
{"step": 70850, "loss": 3.5220710754394533, "grad_norm": 0.6856141090393066, "learning_rate": 0.0001, "elbo": 3.52207, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.154937837203847}
{"step": 70855, "loss": 3.5169677734375, "grad_norm": 0.7600729465484619, "learning_rate": 0.0001, "elbo": 3.51697, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1552310579404175}
{"step": 70860, "loss": 3.5287681579589845, "grad_norm": 0.7242444753646851, "learning_rate": 0.0001, "elbo": 3.52877, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.155524278676988}
{"step": 70865, "loss": 3.5221817016601564, "grad_norm": 0.730505645275116, "learning_rate": 0.0001, "elbo": 3.52218, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.155817499413558}
{"step": 70870, "loss": 3.524494934082031, "grad_norm": 0.8063749670982361, "learning_rate": 0.0001, "elbo": 3.5245, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.156110720150129}
{"step": 70875, "loss": 3.5262622833251953, "grad_norm": 0.7432971596717834, "learning_rate": 0.0001, "elbo": 3.52626, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.156403940886699}
{"step": 70880, "loss": 3.5261962890625, "grad_norm": 0.7056147456169128, "learning_rate": 0.0001, "elbo": 3.5262, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.15669716162327}
{"step": 70885, "loss": 3.537514877319336, "grad_norm": 0.7366565465927124, "learning_rate": 0.0001, "elbo": 3.53751, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.15699038235984}
{"step": 70890, "loss": 3.5068038940429687, "grad_norm": 0.6563681364059448, "learning_rate": 0.0001, "elbo": 3.5068, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.157283603096411}
{"step": 70895, "loss": 3.5270294189453124, "grad_norm": 0.7252171039581299, "learning_rate": 0.0001, "elbo": 3.52703, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.157576823832981}
{"step": 70895, "loss": 3.5270294189453124, "grad_norm": 0.7252171039581299, "learning_rate": 0.0001, "elbo": 3.52703, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.157576823832981}
{"step": 70900, "loss": 3.517905426025391, "grad_norm": 0.7326792478561401, "learning_rate": 0.0001, "elbo": 3.51791, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.157870044569552}
{"step": 70905, "loss": 3.540605926513672, "grad_norm": 0.6818289160728455, "learning_rate": 0.0001, "elbo": 3.54061, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.158163265306122}
{"step": 70910, "loss": 3.50305061340332, "grad_norm": 0.7244137525558472, "learning_rate": 0.0001, "elbo": 3.50305, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.158456486042693}
{"step": 70915, "loss": 3.512469482421875, "grad_norm": 0.7012174725532532, "learning_rate": 0.0001, "elbo": 3.51247, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.158749706779264}
{"step": 70920, "loss": 3.5142311096191405, "grad_norm": 0.7215733528137207, "learning_rate": 0.0001, "elbo": 3.51423, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.159042927515834}
{"step": 70925, "loss": 3.522432327270508, "grad_norm": 0.7224892377853394, "learning_rate": 0.0001, "elbo": 3.52243, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.159336148252405}
{"step": 70930, "loss": 3.5242919921875, "grad_norm": 0.6959614753723145, "learning_rate": 0.0001, "elbo": 3.52429, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.159629368988975}
{"step": 70935, "loss": 3.5143272399902346, "grad_norm": 0.7387284636497498, "learning_rate": 0.0001, "elbo": 3.51433, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.159922589725546}
{"step": 70940, "loss": 3.536682891845703, "grad_norm": 0.6982139945030212, "learning_rate": 0.0001, "elbo": 3.53668, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.160215810462116}
{"step": 70945, "loss": 3.5170867919921873, "grad_norm": 0.7275204658508301, "learning_rate": 0.0001, "elbo": 3.51709, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1605090311986865}
{"step": 70950, "loss": 3.502711868286133, "grad_norm": 0.8475897312164307, "learning_rate": 0.0001, "elbo": 3.50271, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.160802251935257}
{"step": 70955, "loss": 3.5063007354736326, "grad_norm": 0.6653411388397217, "learning_rate": 0.0001, "elbo": 3.5063, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1610954726718274}
{"step": 70960, "loss": 3.513849639892578, "grad_norm": 0.7104340195655823, "learning_rate": 0.0001, "elbo": 3.51385, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.161388693408398}
{"step": 70965, "loss": 3.518901824951172, "grad_norm": 0.6797156929969788, "learning_rate": 0.0001, "elbo": 3.5189, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.161681914144968}
{"step": 70970, "loss": 3.5066871643066406, "grad_norm": 0.6980127692222595, "learning_rate": 0.0001, "elbo": 3.50669, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.161975134881539}
{"step": 70975, "loss": 3.5399547576904298, "grad_norm": 0.7681192755699158, "learning_rate": 0.0001, "elbo": 3.53995, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.162268355618109}
{"step": 70980, "loss": 3.5261646270751954, "grad_norm": 1.0885319709777832, "learning_rate": 0.0001, "elbo": 3.52616, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.16256157635468}
{"step": 70985, "loss": 3.5099021911621096, "grad_norm": 0.708660364151001, "learning_rate": 0.0001, "elbo": 3.5099, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.16285479709125}
{"step": 70990, "loss": 3.508738708496094, "grad_norm": 0.6902353167533875, "learning_rate": 0.0001, "elbo": 3.50874, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.163148017827821}
{"step": 70995, "loss": 3.526109313964844, "grad_norm": 0.6986948847770691, "learning_rate": 0.0001, "elbo": 3.52611, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.163441238564391}
{"step": 71000, "loss": 3.5268325805664062, "grad_norm": 0.6857922077178955, "learning_rate": 0.0001, "elbo": 3.52683, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1637344593009615}
{"step": 71005, "loss": 3.5178802490234373, "grad_norm": 0.7299038767814636, "learning_rate": 0.0001, "elbo": 3.51788, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.164027680037532}
{"step": 71010, "loss": 3.5143199920654298, "grad_norm": 0.8276795744895935, "learning_rate": 0.0001, "elbo": 3.51432, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.164320900774102}
{"step": 71015, "loss": 3.5318092346191405, "grad_norm": 0.721724271774292, "learning_rate": 0.0001, "elbo": 3.53181, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.164614121510673}
{"step": 71020, "loss": 3.523886871337891, "grad_norm": 0.9673044681549072, "learning_rate": 0.0001, "elbo": 3.52389, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.164907342247243}
{"step": 71025, "loss": 3.508331298828125, "grad_norm": 0.7639600038528442, "learning_rate": 0.0001, "elbo": 3.50833, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.165200562983814}
{"step": 71030, "loss": 3.530462646484375, "grad_norm": 0.6800568103790283, "learning_rate": 0.0001, "elbo": 3.53046, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.165493783720384}
{"step": 71035, "loss": 3.5316322326660154, "grad_norm": 0.7222330570220947, "learning_rate": 0.0001, "elbo": 3.53163, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.165787004456956}
{"step": 71040, "loss": 3.5372188568115233, "grad_norm": 0.7210832834243774, "learning_rate": 0.0001, "elbo": 3.53722, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.166080225193526}
{"step": 71045, "loss": 3.527213287353516, "grad_norm": 0.7089282274246216, "learning_rate": 0.0001, "elbo": 3.52721, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1663734459300965}
{"step": 71050, "loss": 3.521527099609375, "grad_norm": 0.7442150115966797, "learning_rate": 0.0001, "elbo": 3.52153, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.166666666666667}
{"step": 71055, "loss": 3.5089744567871093, "grad_norm": 0.7150724530220032, "learning_rate": 0.0001, "elbo": 3.50897, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.166959887403237}
{"step": 71060, "loss": 3.5162891387939452, "grad_norm": 0.7284064888954163, "learning_rate": 0.0001, "elbo": 3.51629, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.167253108139808}
{"step": 71065, "loss": 3.5059799194335937, "grad_norm": 0.669411838054657, "learning_rate": 0.0001, "elbo": 3.50598, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.167546328876378}
{"step": 71065, "loss": 3.5059799194335937, "grad_norm": 0.669411838054657, "learning_rate": 0.0001, "elbo": 3.50598, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.167546328876378}
{"step": 71070, "loss": 3.5254196166992187, "grad_norm": 0.7192993760108948, "learning_rate": 0.0001, "elbo": 3.52542, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.167839549612949}
{"step": 71075, "loss": 3.5149166107177736, "grad_norm": 0.7529439926147461, "learning_rate": 0.0001, "elbo": 3.51492, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.168132770349519}
{"step": 71080, "loss": 3.532196807861328, "grad_norm": 0.7143658399581909, "learning_rate": 0.0001, "elbo": 3.5322, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.16842599108609}
{"step": 71085, "loss": 3.5235149383544924, "grad_norm": 0.8074488043785095, "learning_rate": 0.0001, "elbo": 3.52352, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.16871921182266}
{"step": 71090, "loss": 3.501270294189453, "grad_norm": 0.6794261932373047, "learning_rate": 0.0001, "elbo": 3.50127, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.169012432559231}
{"step": 71095, "loss": 3.508327865600586, "grad_norm": 0.7704169154167175, "learning_rate": 0.0001, "elbo": 3.50833, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.169305653295801}
{"step": 71095, "loss": 3.508327865600586, "grad_norm": 0.7704169154167175, "learning_rate": 0.0001, "elbo": 3.50833, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.169305653295801}
{"step": 71100, "loss": 3.5240768432617187, "grad_norm": 0.7537615299224854, "learning_rate": 0.0001, "elbo": 3.52408, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1695988740323715}
{"step": 71105, "loss": 3.531169128417969, "grad_norm": 0.7121542096138, "learning_rate": 0.0001, "elbo": 3.53117, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.169892094768942}
{"step": 71110, "loss": 3.5160758972167967, "grad_norm": 0.6789908409118652, "learning_rate": 0.0001, "elbo": 3.51608, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.170185315505512}
{"step": 71115, "loss": 3.5343719482421876, "grad_norm": 0.71053147315979, "learning_rate": 0.0001, "elbo": 3.53437, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.170478536242083}
{"step": 71120, "loss": 3.529422378540039, "grad_norm": 0.7131709456443787, "learning_rate": 0.0001, "elbo": 3.52942, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.170771756978653}
{"step": 71125, "loss": 3.511123275756836, "grad_norm": 0.7927117943763733, "learning_rate": 0.0001, "elbo": 3.51112, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.171064977715224}
{"step": 71130, "loss": 3.507489013671875, "grad_norm": 0.8040004372596741, "learning_rate": 0.0001, "elbo": 3.50749, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.171358198451794}
{"step": 71135, "loss": 3.5134429931640625, "grad_norm": 0.6813932657241821, "learning_rate": 0.0001, "elbo": 3.51344, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.171651419188365}
{"step": 71140, "loss": 3.522624969482422, "grad_norm": 0.7907101511955261, "learning_rate": 0.0001, "elbo": 3.52262, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.171944639924935}
{"step": 71145, "loss": 3.5245620727539064, "grad_norm": 0.8352736830711365, "learning_rate": 0.0001, "elbo": 3.52456, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.172237860661506}
{"step": 71150, "loss": 3.522394561767578, "grad_norm": 0.698925793170929, "learning_rate": 0.0001, "elbo": 3.52239, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.172531081398076}
{"step": 71155, "loss": 3.5260467529296875, "grad_norm": 0.7841092348098755, "learning_rate": 0.0001, "elbo": 3.52605, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.172824302134647}
{"step": 71160, "loss": 3.5377708435058595, "grad_norm": 0.7079237699508667, "learning_rate": 0.0001, "elbo": 3.53777, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.173117522871218}
{"step": 71165, "loss": 3.5271862030029295, "grad_norm": 0.7932537198066711, "learning_rate": 0.0001, "elbo": 3.52719, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.173410743607788}
{"step": 71170, "loss": 3.522105407714844, "grad_norm": 0.6756956577301025, "learning_rate": 0.0001, "elbo": 3.52211, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.173703964344359}
{"step": 71175, "loss": 3.5348178863525392, "grad_norm": 0.6954969167709351, "learning_rate": 0.0001, "elbo": 3.53482, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.173997185080929}
{"step": 71180, "loss": 3.510347366333008, "grad_norm": 0.6926673650741577, "learning_rate": 0.0001, "elbo": 3.51035, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1742904058175}
{"step": 71185, "loss": 3.505453109741211, "grad_norm": 0.6706109046936035, "learning_rate": 0.0001, "elbo": 3.50545, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.17458362655407}
{"step": 71190, "loss": 3.518939971923828, "grad_norm": 0.9736464619636536, "learning_rate": 0.0001, "elbo": 3.51894, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.174876847290641}
{"step": 71195, "loss": 3.5223724365234377, "grad_norm": 0.8042181134223938, "learning_rate": 0.0001, "elbo": 3.52237, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.175170068027211}
{"step": 71200, "loss": 3.524400329589844, "grad_norm": 0.7943275570869446, "learning_rate": 0.0001, "elbo": 3.5244, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1754632887637815}
{"step": 71205, "loss": 3.529102325439453, "grad_norm": 0.668955385684967, "learning_rate": 0.0001, "elbo": 3.5291, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.175756509500352}
{"step": 71210, "loss": 3.531665802001953, "grad_norm": 0.7145248055458069, "learning_rate": 0.0001, "elbo": 3.53167, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.176049730236922}
{"step": 71215, "loss": 3.517104721069336, "grad_norm": 0.6828036308288574, "learning_rate": 0.0001, "elbo": 3.5171, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.176342950973493}
{"step": 71220, "loss": 3.5222434997558594, "grad_norm": 0.6914744973182678, "learning_rate": 0.0001, "elbo": 3.52224, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.176636171710063}
{"step": 71225, "loss": 3.499957275390625, "grad_norm": 0.7090744972229004, "learning_rate": 0.0001, "elbo": 3.49996, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.176929392446634}
{"step": 71230, "loss": 3.5067459106445313, "grad_norm": 0.7146416902542114, "learning_rate": 0.0001, "elbo": 3.50675, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.177222613183204}
{"step": 71235, "loss": 3.4954864501953127, "grad_norm": 0.7321534156799316, "learning_rate": 0.0001, "elbo": 3.49549, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.177515833919775}
{"step": 71240, "loss": 3.5155635833740235, "grad_norm": 0.8743309378623962, "learning_rate": 0.0001, "elbo": 3.51556, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.177809054656345}
{"step": 71245, "loss": 3.5118595123291017, "grad_norm": 0.7097800970077515, "learning_rate": 0.0001, "elbo": 3.51186, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.178102275392916}
{"step": 71250, "loss": 3.5157325744628904, "grad_norm": 0.7202214598655701, "learning_rate": 0.0001, "elbo": 3.51573, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.178395496129486}
{"step": 71255, "loss": 3.540363311767578, "grad_norm": 0.7345111966133118, "learning_rate": 0.0001, "elbo": 3.54036, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1786887168660565}
{"step": 71260, "loss": 3.531100845336914, "grad_norm": 0.7302848100662231, "learning_rate": 0.0001, "elbo": 3.5311, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.178981937602627}
{"step": 71265, "loss": 3.5357173919677733, "grad_norm": 0.7329025268554688, "learning_rate": 0.0001, "elbo": 3.53572, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.179275158339197}
{"step": 71270, "loss": 3.5284027099609374, "grad_norm": 0.7113733291625977, "learning_rate": 0.0001, "elbo": 3.5284, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.179568379075768}
{"step": 71275, "loss": 3.527640151977539, "grad_norm": 0.7070951461791992, "learning_rate": 0.0001, "elbo": 3.52764, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.179861599812339}
{"step": 71280, "loss": 3.524429702758789, "grad_norm": 0.693382978439331, "learning_rate": 0.0001, "elbo": 3.52443, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.18015482054891}
{"step": 71285, "loss": 3.524891662597656, "grad_norm": 0.6884968280792236, "learning_rate": 0.0001, "elbo": 3.52489, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.18044804128548}
{"step": 71290, "loss": 3.5222442626953123, "grad_norm": 0.7835355997085571, "learning_rate": 0.0001, "elbo": 3.52224, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1807412620220505}
{"step": 71295, "loss": 3.505872344970703, "grad_norm": 0.7998694777488708, "learning_rate": 0.0001, "elbo": 3.50587, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.181034482758621}
{"step": 71300, "loss": 3.518646240234375, "grad_norm": 0.847219705581665, "learning_rate": 0.0001, "elbo": 3.51865, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1813277034951915}
{"step": 71305, "loss": 3.515065002441406, "grad_norm": 0.7297831177711487, "learning_rate": 0.0001, "elbo": 3.51507, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.181620924231762}
{"step": 71310, "loss": 3.5292335510253907, "grad_norm": 0.735226571559906, "learning_rate": 0.0001, "elbo": 3.52923, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.181914144968332}
{"step": 71310, "loss": 3.5292335510253907, "grad_norm": 0.735226571559906, "learning_rate": 0.0001, "elbo": 3.52923, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.181914144968332}
{"step": 71315, "loss": 3.5174652099609376, "grad_norm": 0.7193965315818787, "learning_rate": 0.0001, "elbo": 3.51747, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.182207365704903}
{"step": 71320, "loss": 3.5285987854003906, "grad_norm": 0.7240450978279114, "learning_rate": 0.0001, "elbo": 3.5286, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.182500586441473}
{"step": 71325, "loss": 3.510428619384766, "grad_norm": 0.7680525779724121, "learning_rate": 0.0001, "elbo": 3.51043, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.182793807178044}
{"step": 71330, "loss": 3.5262519836425783, "grad_norm": 0.69712895154953, "learning_rate": 0.0001, "elbo": 3.52625, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.183087027914614}
{"step": 71335, "loss": 3.526579666137695, "grad_norm": 0.742196261882782, "learning_rate": 0.0001, "elbo": 3.52658, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.183380248651185}
{"step": 71340, "loss": 3.5111434936523436, "grad_norm": 0.682210385799408, "learning_rate": 0.0001, "elbo": 3.51114, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.183673469387755}
{"step": 71345, "loss": 3.527051544189453, "grad_norm": 0.6913385987281799, "learning_rate": 0.0001, "elbo": 3.52705, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1839666901243255}
{"step": 71350, "loss": 3.5173377990722656, "grad_norm": 0.7103206515312195, "learning_rate": 0.0001, "elbo": 3.51734, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.184259910860896}
{"step": 71355, "loss": 3.536054992675781, "grad_norm": 0.7133890986442566, "learning_rate": 0.0001, "elbo": 3.53605, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1845531315974664}
{"step": 71360, "loss": 3.517342376708984, "grad_norm": 0.6733543872833252, "learning_rate": 0.0001, "elbo": 3.51734, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.184846352334037}
{"step": 71365, "loss": 3.510968780517578, "grad_norm": 0.7516573667526245, "learning_rate": 0.0001, "elbo": 3.51097, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.185139573070607}
{"step": 71370, "loss": 3.517700958251953, "grad_norm": 0.7112732529640198, "learning_rate": 0.0001, "elbo": 3.5177, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.185432793807178}
{"step": 71375, "loss": 3.5314399719238283, "grad_norm": 0.9618305563926697, "learning_rate": 0.0001, "elbo": 3.53144, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.185726014543748}
{"step": 71380, "loss": 3.5285629272460937, "grad_norm": 0.7728192806243896, "learning_rate": 0.0001, "elbo": 3.52856, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.186019235280319}
{"step": 71385, "loss": 3.543528747558594, "grad_norm": 0.7750314474105835, "learning_rate": 0.0001, "elbo": 3.54353, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.186312456016889}
{"step": 71390, "loss": 3.5328155517578126, "grad_norm": 0.7578602433204651, "learning_rate": 0.0001, "elbo": 3.53282, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.18660567675346}
{"step": 71395, "loss": 3.5348587036132812, "grad_norm": 0.7285816669464111, "learning_rate": 0.0001, "elbo": 3.53486, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.18689889749003}
{"step": 71395, "loss": 3.5348587036132812, "grad_norm": 0.7285816669464111, "learning_rate": 0.0001, "elbo": 3.53486, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.18689889749003}
{"step": 71400, "loss": 3.5184783935546875, "grad_norm": 0.764735758304596, "learning_rate": 0.0001, "elbo": 3.51848, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.187192118226601}
{"step": 71405, "loss": 3.5025398254394533, "grad_norm": 0.7260761260986328, "learning_rate": 0.0001, "elbo": 3.50254, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.187485338963172}
{"step": 71410, "loss": 3.5188255310058594, "grad_norm": 0.7249879240989685, "learning_rate": 0.0001, "elbo": 3.51883, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.187778559699742}
{"step": 71415, "loss": 3.515245056152344, "grad_norm": 0.6792135834693909, "learning_rate": 0.0001, "elbo": 3.51525, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.188071780436313}
{"step": 71420, "loss": 3.5160655975341797, "grad_norm": 0.7349459528923035, "learning_rate": 0.0001, "elbo": 3.51607, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.188365001172883}
{"step": 71420, "loss": 3.5160655975341797, "grad_norm": 0.7349459528923035, "learning_rate": 0.0001, "elbo": 3.51607, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.188365001172883}
{"step": 71425, "loss": 3.5253501892089845, "grad_norm": 0.7320236563682556, "learning_rate": 0.0001, "elbo": 3.52535, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.188658221909454}
{"step": 71430, "loss": 3.5233253479003905, "grad_norm": 0.7003273367881775, "learning_rate": 0.0001, "elbo": 3.52333, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.188951442646024}
{"step": 71435, "loss": 3.5228855133056642, "grad_norm": 0.7479419112205505, "learning_rate": 0.0001, "elbo": 3.52289, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.189244663382595}
{"step": 71440, "loss": 3.5252899169921874, "grad_norm": 0.8152149319648743, "learning_rate": 0.0001, "elbo": 3.52529, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.189537884119165}
{"step": 71445, "loss": 3.5348365783691404, "grad_norm": 0.804011881351471, "learning_rate": 0.0001, "elbo": 3.53484, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1898311048557355}
{"step": 71450, "loss": 3.5096817016601562, "grad_norm": 0.7590265870094299, "learning_rate": 0.0001, "elbo": 3.50968, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.190124325592306}
{"step": 71455, "loss": 3.5229888916015626, "grad_norm": 0.7130147814750671, "learning_rate": 0.0001, "elbo": 3.52299, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.190417546328876}
{"step": 71460, "loss": 3.5258224487304686, "grad_norm": 0.7172946333885193, "learning_rate": 0.0001, "elbo": 3.52582, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.190710767065447}
{"step": 71465, "loss": 3.523375701904297, "grad_norm": 0.7116552591323853, "learning_rate": 0.0001, "elbo": 3.52338, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.191003987802017}
{"step": 71470, "loss": 3.525775146484375, "grad_norm": 0.8043828010559082, "learning_rate": 0.0001, "elbo": 3.52577, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.191297208538588}
{"step": 71475, "loss": 3.522426223754883, "grad_norm": 0.856492817401886, "learning_rate": 0.0001, "elbo": 3.52243, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.191590429275158}
{"step": 71480, "loss": 3.5179283142089846, "grad_norm": 0.7509045004844666, "learning_rate": 0.0001, "elbo": 3.51793, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.191883650011729}
{"step": 71485, "loss": 3.511824035644531, "grad_norm": 0.6958862543106079, "learning_rate": 0.0001, "elbo": 3.51182, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.192176870748299}
{"step": 71490, "loss": 3.5083984375, "grad_norm": 0.7228474617004395, "learning_rate": 0.0001, "elbo": 3.5084, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.19247009148487}
{"step": 71495, "loss": 3.5314605712890623, "grad_norm": 0.6984007358551025, "learning_rate": 0.0001, "elbo": 3.53146, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.19276331222144}
{"step": 71500, "loss": 3.5034202575683593, "grad_norm": 0.7165602445602417, "learning_rate": 0.0001, "elbo": 3.50342, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1930565329580105}
{"step": 71505, "loss": 3.5358978271484376, "grad_norm": 0.7149394750595093, "learning_rate": 0.0001, "elbo": 3.5359, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.193349753694581}
{"step": 71510, "loss": 3.5327838897705077, "grad_norm": 0.7137343287467957, "learning_rate": 0.0001, "elbo": 3.53278, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.193642974431151}
{"step": 71515, "loss": 3.5091407775878904, "grad_norm": 0.6984958052635193, "learning_rate": 0.0001, "elbo": 3.50914, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.193936195167722}
{"step": 71520, "loss": 3.5155014038085937, "grad_norm": 1.0311251878738403, "learning_rate": 0.0001, "elbo": 3.5155, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.194229415904292}
{"step": 71525, "loss": 3.504755401611328, "grad_norm": 0.8387483358383179, "learning_rate": 0.0001, "elbo": 3.50476, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.194522636640864}
{"step": 71530, "loss": 3.526233673095703, "grad_norm": 0.7369638085365295, "learning_rate": 0.0001, "elbo": 3.52623, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.194815857377434}
{"step": 71535, "loss": 3.522411346435547, "grad_norm": 0.6841012835502625, "learning_rate": 0.0001, "elbo": 3.52241, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.195109078114005}
{"step": 71540, "loss": 3.516565704345703, "grad_norm": 0.719559907913208, "learning_rate": 0.0001, "elbo": 3.51657, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.195402298850575}
{"step": 71545, "loss": 3.5050464630126954, "grad_norm": 0.7151613831520081, "learning_rate": 0.0001, "elbo": 3.50505, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1956955195871455}
{"step": 71550, "loss": 3.527173614501953, "grad_norm": 0.6632137298583984, "learning_rate": 0.0001, "elbo": 3.52717, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.195988740323716}
{"step": 71555, "loss": 3.5116867065429687, "grad_norm": 0.7218337059020996, "learning_rate": 0.0001, "elbo": 3.51169, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.196281961060286}
{"step": 71560, "loss": 3.5254058837890625, "grad_norm": 0.7153039574623108, "learning_rate": 0.0001, "elbo": 3.52541, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.196575181796857}
{"step": 71565, "loss": 3.518677520751953, "grad_norm": 0.699328601360321, "learning_rate": 0.0001, "elbo": 3.51868, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.196868402533427}
{"step": 71570, "loss": 3.5268013000488283, "grad_norm": 0.7491840124130249, "learning_rate": 0.0001, "elbo": 3.5268, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.197161623269998}
{"step": 71575, "loss": 3.5259056091308594, "grad_norm": 0.7731537222862244, "learning_rate": 0.0001, "elbo": 3.52591, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.197454844006568}
{"step": 71580, "loss": 3.512313461303711, "grad_norm": 0.8525902628898621, "learning_rate": 0.0001, "elbo": 3.51231, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.197748064743139}
{"step": 71585, "loss": 3.5240104675292967, "grad_norm": 0.6992959380149841, "learning_rate": 0.0001, "elbo": 3.52401, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.198041285479709}
{"step": 71590, "loss": 3.527204132080078, "grad_norm": 0.6828991770744324, "learning_rate": 0.0001, "elbo": 3.5272, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.19833450621628}
{"step": 71595, "loss": 3.51748046875, "grad_norm": 0.7395270466804504, "learning_rate": 0.0001, "elbo": 3.51748, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.19862772695285}
{"step": 71600, "loss": 3.5381744384765623, "grad_norm": 0.7380528450012207, "learning_rate": 0.0001, "elbo": 3.53817, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.1989209476894205}
{"step": 71605, "loss": 3.522621917724609, "grad_norm": 0.7623915076255798, "learning_rate": 0.0001, "elbo": 3.52262, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.199214168425991}
{"step": 71610, "loss": 3.5228950500488283, "grad_norm": 0.6912259459495544, "learning_rate": 0.0001, "elbo": 3.52289, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.199507389162561}
{"step": 71615, "loss": 3.5192291259765627, "grad_norm": 0.7003616690635681, "learning_rate": 0.0001, "elbo": 3.51923, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.199800609899132}
{"step": 71620, "loss": 3.52098388671875, "grad_norm": 1.0702803134918213, "learning_rate": 0.0001, "elbo": 3.52098, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.200093830635702}
{"step": 71625, "loss": 3.5300621032714843, "grad_norm": 0.6843385100364685, "learning_rate": 0.0001, "elbo": 3.53006, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.200387051372273}
{"step": 71630, "loss": 3.515848922729492, "grad_norm": 0.7263797521591187, "learning_rate": 0.0001, "elbo": 3.51585, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.200680272108843}
{"step": 71635, "loss": 3.525709533691406, "grad_norm": 0.7032963633537292, "learning_rate": 0.0001, "elbo": 3.52571, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.200973492845414}
{"step": 71640, "loss": 3.5196937561035155, "grad_norm": 0.7143682837486267, "learning_rate": 0.0001, "elbo": 3.51969, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.201266713581984}
{"step": 71645, "loss": 3.5301727294921874, "grad_norm": 0.7838099002838135, "learning_rate": 0.0001, "elbo": 3.53017, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.201559934318555}
{"step": 71650, "loss": 3.5294204711914063, "grad_norm": 0.8000434041023254, "learning_rate": 0.0001, "elbo": 3.52942, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.201853155055126}
{"step": 71655, "loss": 3.5218414306640624, "grad_norm": 0.6996827721595764, "learning_rate": 0.0001, "elbo": 3.52184, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.202146375791696}
{"step": 71660, "loss": 3.5194633483886717, "grad_norm": 0.6738932728767395, "learning_rate": 0.0001, "elbo": 3.51946, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.202439596528267}
{"step": 71665, "loss": 3.521067428588867, "grad_norm": 0.7371233701705933, "learning_rate": 0.0001, "elbo": 3.52107, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.202732817264837}
{"step": 71670, "loss": 3.521318817138672, "grad_norm": 0.7553502321243286, "learning_rate": 0.0001, "elbo": 3.52132, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.203026038001408}
{"step": 71675, "loss": 3.527456283569336, "grad_norm": 0.6709184050559998, "learning_rate": 0.0001, "elbo": 3.52746, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.203319258737978}
{"step": 71680, "loss": 3.532392120361328, "grad_norm": 0.7901330590248108, "learning_rate": 0.0001, "elbo": 3.53239, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.203612479474549}
{"step": 71685, "loss": 3.521253204345703, "grad_norm": 0.8085594177246094, "learning_rate": 0.0001, "elbo": 3.52125, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.203905700211119}
{"step": 71690, "loss": 3.5125221252441405, "grad_norm": 1.1339913606643677, "learning_rate": 0.0001, "elbo": 3.51252, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2041989209476895}
{"step": 71695, "loss": 3.5155990600585936, "grad_norm": 0.7000226974487305, "learning_rate": 0.0001, "elbo": 3.5156, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.20449214168426}
{"step": 71700, "loss": 3.528268051147461, "grad_norm": 1.116557240486145, "learning_rate": 0.0001, "elbo": 3.52827, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2047853624208305}
{"step": 71705, "loss": 3.5206443786621096, "grad_norm": 0.7196235060691833, "learning_rate": 0.0001, "elbo": 3.52064, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.205078583157401}
{"step": 71710, "loss": 3.524923324584961, "grad_norm": 0.7125318646430969, "learning_rate": 0.0001, "elbo": 3.52492, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.205371803893971}
{"step": 71715, "loss": 3.519147491455078, "grad_norm": 0.7575127482414246, "learning_rate": 0.0001, "elbo": 3.51915, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.205665024630542}
{"step": 71720, "loss": 3.5232601165771484, "grad_norm": 0.8299963474273682, "learning_rate": 0.0001, "elbo": 3.52326, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.205958245367112}
{"step": 71725, "loss": 3.512332534790039, "grad_norm": 0.6729838252067566, "learning_rate": 0.0001, "elbo": 3.51233, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.206251466103683}
{"step": 71730, "loss": 3.4989177703857424, "grad_norm": 1.063877820968628, "learning_rate": 0.0001, "elbo": 3.49892, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.206544686840253}
{"step": 71735, "loss": 3.496063995361328, "grad_norm": 0.6815043687820435, "learning_rate": 0.0001, "elbo": 3.49606, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.206837907576824}
{"step": 71740, "loss": 3.5273780822753906, "grad_norm": 0.7060582637786865, "learning_rate": 0.0001, "elbo": 3.52738, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.207131128313394}
{"step": 71745, "loss": 3.514243316650391, "grad_norm": 0.6657801270484924, "learning_rate": 0.0001, "elbo": 3.51424, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2074243490499645}
{"step": 71750, "loss": 3.521376037597656, "grad_norm": 0.6875557899475098, "learning_rate": 0.0001, "elbo": 3.52138, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.207717569786535}
{"step": 71750, "loss": 3.521376037597656, "grad_norm": 0.6875557899475098, "learning_rate": 0.0001, "elbo": 3.52138, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.207717569786535}
{"step": 71755, "loss": 3.51190299987793, "grad_norm": 0.7094330191612244, "learning_rate": 0.0001, "elbo": 3.5119, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2080107905231054}
{"step": 71760, "loss": 3.531916046142578, "grad_norm": 0.6713265776634216, "learning_rate": 0.0001, "elbo": 3.53192, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.208304011259676}
{"step": 71765, "loss": 3.5022598266601563, "grad_norm": 0.6895536184310913, "learning_rate": 0.0001, "elbo": 3.50226, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.208597231996246}
{"step": 71770, "loss": 3.516804504394531, "grad_norm": 0.6942233443260193, "learning_rate": 0.0001, "elbo": 3.5168, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.208890452732817}
{"step": 71775, "loss": 3.529110336303711, "grad_norm": 0.6744110584259033, "learning_rate": 0.0001, "elbo": 3.52911, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.209183673469388}
{"step": 71780, "loss": 3.513304901123047, "grad_norm": 0.6942861676216125, "learning_rate": 0.0001, "elbo": 3.51331, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.209476894205959}
{"step": 71785, "loss": 3.5060672760009766, "grad_norm": 0.7428838014602661, "learning_rate": 0.0001, "elbo": 3.50607, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.209770114942529}
{"step": 71790, "loss": 3.5240554809570312, "grad_norm": 0.8115437626838684, "learning_rate": 0.0001, "elbo": 3.52406, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2100633356790995}
{"step": 71795, "loss": 3.5282798767089845, "grad_norm": 0.7107411026954651, "learning_rate": 0.0001, "elbo": 3.52828, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.21035655641567}
{"step": 71800, "loss": 3.525945281982422, "grad_norm": 0.7350165843963623, "learning_rate": 0.0001, "elbo": 3.52595, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.21064977715224}
{"step": 71805, "loss": 3.516786575317383, "grad_norm": 0.667091965675354, "learning_rate": 0.0001, "elbo": 3.51679, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.210942997888811}
{"step": 71810, "loss": 3.5172584533691404, "grad_norm": 0.6946382522583008, "learning_rate": 0.0001, "elbo": 3.51726, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.211236218625381}
{"step": 71815, "loss": 3.509756088256836, "grad_norm": 0.7861582636833191, "learning_rate": 0.0001, "elbo": 3.50976, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.211529439361952}
{"step": 71820, "loss": 3.5317291259765624, "grad_norm": 0.6906731128692627, "learning_rate": 0.0001, "elbo": 3.53173, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.211822660098522}
{"step": 71825, "loss": 3.521455764770508, "grad_norm": 0.6966649889945984, "learning_rate": 0.0001, "elbo": 3.52146, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.212115880835093}
{"step": 71830, "loss": 3.5247291564941405, "grad_norm": 0.7225567102432251, "learning_rate": 0.0001, "elbo": 3.52473, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.212409101571663}
{"step": 71835, "loss": 3.5165390014648437, "grad_norm": 0.7509150505065918, "learning_rate": 0.0001, "elbo": 3.51654, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.212702322308234}
{"step": 71840, "loss": 3.540785217285156, "grad_norm": 0.7241514921188354, "learning_rate": 0.0001, "elbo": 3.54079, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.212995543044804}
{"step": 71845, "loss": 3.522932434082031, "grad_norm": 0.7180904746055603, "learning_rate": 0.0001, "elbo": 3.52293, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2132887637813745}
{"step": 71850, "loss": 3.5175247192382812, "grad_norm": 0.7090245485305786, "learning_rate": 0.0001, "elbo": 3.51753, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.213581984517945}
{"step": 71855, "loss": 3.5263851165771483, "grad_norm": 0.7066617608070374, "learning_rate": 0.0001, "elbo": 3.52639, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.213875205254515}
{"step": 71860, "loss": 3.527385711669922, "grad_norm": 0.6883047223091125, "learning_rate": 0.0001, "elbo": 3.52739, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.214168425991086}
{"step": 71865, "loss": 3.539293670654297, "grad_norm": 0.7574986815452576, "learning_rate": 0.0001, "elbo": 3.53929, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.214461646727656}
{"step": 71870, "loss": 3.5116298675537108, "grad_norm": 0.6986973881721497, "learning_rate": 0.0001, "elbo": 3.51163, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.214754867464227}
{"step": 71875, "loss": 3.51324348449707, "grad_norm": 0.7523255348205566, "learning_rate": 0.0001, "elbo": 3.51324, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.215048088200797}
{"step": 71880, "loss": 3.52026481628418, "grad_norm": 0.6994036436080933, "learning_rate": 0.0001, "elbo": 3.52026, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.215341308937368}
{"step": 71885, "loss": 3.5136028289794923, "grad_norm": 0.7175858020782471, "learning_rate": 0.0001, "elbo": 3.5136, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.215634529673938}
{"step": 71885, "loss": 3.5136028289794923, "grad_norm": 0.7175858020782471, "learning_rate": 0.0001, "elbo": 3.5136, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.215634529673938}
{"step": 71890, "loss": 3.531964111328125, "grad_norm": 0.7706011533737183, "learning_rate": 0.0001, "elbo": 3.53196, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2159277504105095}
{"step": 71895, "loss": 3.531378173828125, "grad_norm": 1.1245777606964111, "learning_rate": 0.0001, "elbo": 3.53138, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.21622097114708}
{"step": 71900, "loss": 3.522394561767578, "grad_norm": 1.0414319038391113, "learning_rate": 0.0001, "elbo": 3.52239, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.21651419188365}
{"step": 71905, "loss": 3.5140289306640624, "grad_norm": 0.8097901344299316, "learning_rate": 0.0001, "elbo": 3.51403, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.216807412620221}
{"step": 71910, "loss": 3.520397186279297, "grad_norm": 0.6826269030570984, "learning_rate": 0.0001, "elbo": 3.5204, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.217100633356791}
{"step": 71915, "loss": 3.509926986694336, "grad_norm": 0.6984622478485107, "learning_rate": 0.0001, "elbo": 3.50993, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.217393854093362}
{"step": 71920, "loss": 3.5334190368652343, "grad_norm": 0.7374111413955688, "learning_rate": 0.0001, "elbo": 3.53342, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.217687074829932}
{"step": 71925, "loss": 3.522998809814453, "grad_norm": 0.6949003338813782, "learning_rate": 0.0001, "elbo": 3.523, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.217980295566503}
{"step": 71930, "loss": 3.532756805419922, "grad_norm": 0.7444165349006653, "learning_rate": 0.0001, "elbo": 3.53276, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.218273516303073}
{"step": 71935, "loss": 3.5238441467285155, "grad_norm": 0.7719361186027527, "learning_rate": 0.0001, "elbo": 3.52384, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.218566737039644}
{"step": 71940, "loss": 3.517089080810547, "grad_norm": 0.7572569847106934, "learning_rate": 0.0001, "elbo": 3.51709, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.218859957776214}
{"step": 71945, "loss": 3.5100452423095705, "grad_norm": 0.6844813227653503, "learning_rate": 0.0001, "elbo": 3.51005, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2191531785127845}
{"step": 71950, "loss": 3.518514633178711, "grad_norm": 0.7354766726493835, "learning_rate": 0.0001, "elbo": 3.51851, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.219446399249355}
{"step": 71955, "loss": 3.51522216796875, "grad_norm": 0.7200459241867065, "learning_rate": 0.0001, "elbo": 3.51522, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.219739619985925}
{"step": 71960, "loss": 3.512739562988281, "grad_norm": 0.7601869106292725, "learning_rate": 0.0001, "elbo": 3.51274, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.220032840722496}
{"step": 71965, "loss": 3.5090801239013674, "grad_norm": 0.7220714092254639, "learning_rate": 0.0001, "elbo": 3.50908, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.220326061459066}
{"step": 71970, "loss": 3.5314544677734374, "grad_norm": 0.8124855756759644, "learning_rate": 0.0001, "elbo": 3.53145, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.220619282195637}
{"step": 71975, "loss": 3.5158611297607423, "grad_norm": 0.69765704870224, "learning_rate": 0.0001, "elbo": 3.51586, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.220912502932207}
{"step": 71980, "loss": 3.5182525634765627, "grad_norm": 0.70247882604599, "learning_rate": 0.0001, "elbo": 3.51825, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.221205723668778}
{"step": 71985, "loss": 3.511466217041016, "grad_norm": 0.6759345531463623, "learning_rate": 0.0001, "elbo": 3.51147, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.221498944405348}
{"step": 71990, "loss": 3.521720123291016, "grad_norm": 0.6918809413909912, "learning_rate": 0.0001, "elbo": 3.52172, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.221792165141919}
{"step": 71995, "loss": 3.516729736328125, "grad_norm": 0.7122191786766052, "learning_rate": 0.0001, "elbo": 3.51673, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.222085385878489}
{"step": 72000, "loss": 3.5064010620117188, "grad_norm": 0.7096893787384033, "learning_rate": 0.0001, "elbo": 3.5064, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2223786066150595}
{"step": 72005, "loss": 3.525513458251953, "grad_norm": 0.7381235361099243, "learning_rate": 0.0001, "elbo": 3.52551, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.22267182735163}
{"step": 72010, "loss": 3.522747039794922, "grad_norm": 0.6993657350540161, "learning_rate": 0.0001, "elbo": 3.52275, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2229650480882}
{"step": 72015, "loss": 3.527077865600586, "grad_norm": 0.7643724679946899, "learning_rate": 0.0001, "elbo": 3.52708, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.223258268824772}
{"step": 72020, "loss": 3.530797576904297, "grad_norm": 0.8264267444610596, "learning_rate": 0.0001, "elbo": 3.5308, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.223551489561342}
{"step": 72025, "loss": 3.5231254577636717, "grad_norm": 0.7370678782463074, "learning_rate": 0.0001, "elbo": 3.52313, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.223844710297913}
{"step": 72030, "loss": 3.499961090087891, "grad_norm": 0.9283866882324219, "learning_rate": 0.0001, "elbo": 3.49996, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.224137931034483}
{"step": 72035, "loss": 3.523631286621094, "grad_norm": 0.69200199842453, "learning_rate": 0.0001, "elbo": 3.52363, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2244311517710535}
{"step": 72040, "loss": 3.5333385467529297, "grad_norm": 0.7450782656669617, "learning_rate": 0.0001, "elbo": 3.53334, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.224724372507624}
{"step": 72045, "loss": 3.5185699462890625, "grad_norm": 0.6783275604248047, "learning_rate": 0.0001, "elbo": 3.51857, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2250175932441945}
{"step": 72050, "loss": 3.516118621826172, "grad_norm": 0.6942176818847656, "learning_rate": 0.0001, "elbo": 3.51612, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.225310813980765}
{"step": 72055, "loss": 3.514569091796875, "grad_norm": 0.709770143032074, "learning_rate": 0.0001, "elbo": 3.51457, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.225604034717335}
{"step": 72055, "loss": 3.514569091796875, "grad_norm": 0.709770143032074, "learning_rate": 0.0001, "elbo": 3.51457, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.225604034717335}
{"step": 72060, "loss": 3.513562774658203, "grad_norm": 0.7078791856765747, "learning_rate": 0.0001, "elbo": 3.51356, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.225897255453906}
{"step": 72060, "loss": 3.513562774658203, "grad_norm": 0.7078791856765747, "learning_rate": 0.0001, "elbo": 3.51356, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.225897255453906}
{"step": 72065, "loss": 3.503874588012695, "grad_norm": 0.6915645003318787, "learning_rate": 0.0001, "elbo": 3.50387, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.226190476190476}
{"step": 72070, "loss": 3.515546417236328, "grad_norm": 0.8512511253356934, "learning_rate": 0.0001, "elbo": 3.51555, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.226483696927047}
{"step": 72075, "loss": 3.518601989746094, "grad_norm": 0.7203465700149536, "learning_rate": 0.0001, "elbo": 3.5186, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.226776917663617}
{"step": 72080, "loss": 3.531376266479492, "grad_norm": 0.6879317760467529, "learning_rate": 0.0001, "elbo": 3.53138, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.227070138400188}
{"step": 72085, "loss": 3.5065956115722656, "grad_norm": 0.9380319714546204, "learning_rate": 0.0001, "elbo": 3.5066, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.227363359136758}
{"step": 72090, "loss": 3.5219356536865236, "grad_norm": 0.6816263794898987, "learning_rate": 0.0001, "elbo": 3.52194, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2276565798733285}
{"step": 72095, "loss": 3.538580322265625, "grad_norm": 0.7101632952690125, "learning_rate": 0.0001, "elbo": 3.53858, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.227949800609899}
{"step": 72100, "loss": 3.524250030517578, "grad_norm": 0.7661917209625244, "learning_rate": 0.0001, "elbo": 3.52425, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2282430213464695}
{"step": 72105, "loss": 3.5240882873535155, "grad_norm": 0.7417920231819153, "learning_rate": 0.0001, "elbo": 3.52409, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.22853624208304}
{"step": 72110, "loss": 3.513367462158203, "grad_norm": 0.8302273154258728, "learning_rate": 0.0001, "elbo": 3.51337, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.22882946281961}
{"step": 72115, "loss": 3.528363037109375, "grad_norm": 0.7607215642929077, "learning_rate": 0.0001, "elbo": 3.52836, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.229122683556181}
{"step": 72120, "loss": 3.5267906188964844, "grad_norm": 0.6913912892341614, "learning_rate": 0.0001, "elbo": 3.52679, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.229415904292751}
{"step": 72125, "loss": 3.5277629852294923, "grad_norm": 0.7461686134338379, "learning_rate": 0.0001, "elbo": 3.52776, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.229709125029322}
{"step": 72125, "loss": 3.5277629852294923, "grad_norm": 0.7461686134338379, "learning_rate": 0.0001, "elbo": 3.52776, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.229709125029322}
{"step": 72130, "loss": 3.5088211059570313, "grad_norm": 0.6825863718986511, "learning_rate": 0.0001, "elbo": 3.50882, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.230002345765892}
{"step": 72135, "loss": 3.5221134185791017, "grad_norm": 0.7719951868057251, "learning_rate": 0.0001, "elbo": 3.52211, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.230295566502463}
{"step": 72140, "loss": 3.5255523681640626, "grad_norm": 0.6606012582778931, "learning_rate": 0.0001, "elbo": 3.52555, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.230588787239034}
{"step": 72145, "loss": 3.5251846313476562, "grad_norm": 0.6776821613311768, "learning_rate": 0.0001, "elbo": 3.52518, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.230882007975604}
{"step": 72150, "loss": 3.516851806640625, "grad_norm": 0.6794067621231079, "learning_rate": 0.0001, "elbo": 3.51685, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.231175228712175}
{"step": 72155, "loss": 3.5235061645507812, "grad_norm": 0.7883570194244385, "learning_rate": 0.0001, "elbo": 3.52351, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.231468449448745}
{"step": 72160, "loss": 3.5182838439941406, "grad_norm": 0.7861312627792358, "learning_rate": 0.0001, "elbo": 3.51828, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.231761670185316}
{"step": 72165, "loss": 3.514177703857422, "grad_norm": 0.8109487295150757, "learning_rate": 0.0001, "elbo": 3.51418, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.232054890921886}
{"step": 72170, "loss": 3.5189334869384767, "grad_norm": 0.758201003074646, "learning_rate": 0.0001, "elbo": 3.51893, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.232348111658457}
{"step": 72175, "loss": 3.526736831665039, "grad_norm": 0.7300199270248413, "learning_rate": 0.0001, "elbo": 3.52674, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.232641332395027}
{"step": 72180, "loss": 3.5054904937744142, "grad_norm": 0.699749231338501, "learning_rate": 0.0001, "elbo": 3.50549, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.232934553131598}
{"step": 72185, "loss": 3.5223773956298827, "grad_norm": 0.7246810793876648, "learning_rate": 0.0001, "elbo": 3.52238, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.233227773868168}
{"step": 72190, "loss": 3.517182159423828, "grad_norm": 0.83482426404953, "learning_rate": 0.0001, "elbo": 3.51718, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2335209946047385}
{"step": 72195, "loss": 3.518755340576172, "grad_norm": 0.7365930080413818, "learning_rate": 0.0001, "elbo": 3.51876, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.233814215341309}
{"step": 72200, "loss": 3.516398620605469, "grad_norm": 0.6627480983734131, "learning_rate": 0.0001, "elbo": 3.5164, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.234107436077879}
{"step": 72205, "loss": 3.512256622314453, "grad_norm": 0.6897915601730347, "learning_rate": 0.0001, "elbo": 3.51226, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.23440065681445}
{"step": 72210, "loss": 3.5186180114746093, "grad_norm": 0.7417623996734619, "learning_rate": 0.0001, "elbo": 3.51862, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.23469387755102}
{"step": 72215, "loss": 3.4987430572509766, "grad_norm": 0.7791767716407776, "learning_rate": 0.0001, "elbo": 3.49874, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.234987098287591}
{"step": 72220, "loss": 3.5106300354003905, "grad_norm": 0.6887990832328796, "learning_rate": 0.0001, "elbo": 3.51063, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.235280319024161}
{"step": 72225, "loss": 3.5168220520019533, "grad_norm": 0.7794340252876282, "learning_rate": 0.0001, "elbo": 3.51682, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.235573539760732}
{"step": 72230, "loss": 3.5131462097167967, "grad_norm": 0.7268936038017273, "learning_rate": 0.0001, "elbo": 3.51315, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.235866760497302}
{"step": 72235, "loss": 3.5003330230712892, "grad_norm": 0.6908372044563293, "learning_rate": 0.0001, "elbo": 3.50033, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.236159981233873}
{"step": 72240, "loss": 3.5131397247314453, "grad_norm": 0.77850341796875, "learning_rate": 0.0001, "elbo": 3.51314, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.236453201970443}
{"step": 72245, "loss": 3.5238578796386717, "grad_norm": 0.6997823715209961, "learning_rate": 0.0001, "elbo": 3.52386, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2367464227070135}
{"step": 72250, "loss": 3.5096767425537108, "grad_norm": 0.8606195449829102, "learning_rate": 0.0001, "elbo": 3.50968, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.237039643443584}
{"step": 72255, "loss": 3.532134246826172, "grad_norm": 0.7320334315299988, "learning_rate": 0.0001, "elbo": 3.53213, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.237332864180154}
{"step": 72255, "loss": 3.532134246826172, "grad_norm": 0.7320334315299988, "learning_rate": 0.0001, "elbo": 3.53213, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.237332864180154}
{"step": 72260, "loss": 3.509102630615234, "grad_norm": 0.8336273431777954, "learning_rate": 0.0001, "elbo": 3.5091, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.237626084916725}
{"step": 72265, "loss": 3.5185386657714846, "grad_norm": 0.7005213499069214, "learning_rate": 0.0001, "elbo": 3.51854, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.237919305653296}
{"step": 72270, "loss": 3.5389652252197266, "grad_norm": 0.7550356388092041, "learning_rate": 0.0001, "elbo": 3.53897, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.238212526389867}
{"step": 72275, "loss": 3.5106922149658204, "grad_norm": 0.6962000131607056, "learning_rate": 0.0001, "elbo": 3.51069, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.238505747126437}
{"step": 72275, "loss": 3.5106922149658204, "grad_norm": 0.6962000131607056, "learning_rate": 0.0001, "elbo": 3.51069, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.238505747126437}
{"step": 72280, "loss": 3.497824859619141, "grad_norm": 0.7351276278495789, "learning_rate": 0.0001, "elbo": 3.49782, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.238798967863008}
{"step": 72285, "loss": 3.5265125274658202, "grad_norm": 0.7724061608314514, "learning_rate": 0.0001, "elbo": 3.52651, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.239092188599578}
{"step": 72290, "loss": 3.5227035522460937, "grad_norm": 0.6913667917251587, "learning_rate": 0.0001, "elbo": 3.5227, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2393854093361485}
{"step": 72295, "loss": 3.5322429656982424, "grad_norm": 0.7082213759422302, "learning_rate": 0.0001, "elbo": 3.53224, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.239678630072719}
{"step": 72300, "loss": 3.519870376586914, "grad_norm": 0.6841785311698914, "learning_rate": 0.0001, "elbo": 3.51987, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.239971850809289}
{"step": 72305, "loss": 3.517772674560547, "grad_norm": 0.7229426503181458, "learning_rate": 0.0001, "elbo": 3.51777, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.24026507154586}
{"step": 72310, "loss": 3.5080810546875, "grad_norm": 0.7307026386260986, "learning_rate": 0.0001, "elbo": 3.50808, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.24055829228243}
{"step": 72315, "loss": 3.5268394470214846, "grad_norm": 0.9817514419555664, "learning_rate": 0.0001, "elbo": 3.52684, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.240851513019001}
{"step": 72320, "loss": 3.5234642028808594, "grad_norm": 0.7927499413490295, "learning_rate": 0.0001, "elbo": 3.52346, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.241144733755571}
{"step": 72325, "loss": 3.5202159881591797, "grad_norm": 0.6999422311782837, "learning_rate": 0.0001, "elbo": 3.52022, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.241437954492142}
{"step": 72330, "loss": 3.520314025878906, "grad_norm": 0.7118374705314636, "learning_rate": 0.0001, "elbo": 3.52031, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.241731175228712}
{"step": 72335, "loss": 3.5187347412109373, "grad_norm": 0.742156445980072, "learning_rate": 0.0001, "elbo": 3.51873, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.242024395965283}
{"step": 72340, "loss": 3.5241043090820314, "grad_norm": 0.7219850420951843, "learning_rate": 0.0001, "elbo": 3.5241, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.242317616701853}
{"step": 72345, "loss": 3.510057067871094, "grad_norm": 0.6773050427436829, "learning_rate": 0.0001, "elbo": 3.51006, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2426108374384235}
{"step": 72350, "loss": 3.533026123046875, "grad_norm": 1.0611780881881714, "learning_rate": 0.0001, "elbo": 3.53303, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.242904058174994}
{"step": 72355, "loss": 3.4948558807373047, "grad_norm": 0.8877174854278564, "learning_rate": 0.0001, "elbo": 3.49486, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.243197278911564}
{"step": 72360, "loss": 3.5027084350585938, "grad_norm": 0.7371232509613037, "learning_rate": 0.0001, "elbo": 3.50271, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.243490499648135}
{"step": 72365, "loss": 3.522653579711914, "grad_norm": 0.7297257781028748, "learning_rate": 0.0001, "elbo": 3.52265, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.243783720384705}
{"step": 72370, "loss": 3.503215789794922, "grad_norm": 0.832772970199585, "learning_rate": 0.0001, "elbo": 3.50322, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.244076941121276}
{"step": 72375, "loss": 3.5180328369140623, "grad_norm": 0.7468961477279663, "learning_rate": 0.0001, "elbo": 3.51803, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.244370161857846}
{"step": 72380, "loss": 3.5099536895751955, "grad_norm": 0.7426177263259888, "learning_rate": 0.0001, "elbo": 3.50995, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.244663382594417}
{"step": 72385, "loss": 3.5193557739257812, "grad_norm": 0.7144277691841125, "learning_rate": 0.0001, "elbo": 3.51936, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.244956603330987}
{"step": 72390, "loss": 3.5269813537597656, "grad_norm": 0.7289484739303589, "learning_rate": 0.0001, "elbo": 3.52698, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2452498240675585}
{"step": 72395, "loss": 3.538946533203125, "grad_norm": 1.1052372455596924, "learning_rate": 0.0001, "elbo": 3.53895, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.245543044804129}
{"step": 72400, "loss": 3.518010711669922, "grad_norm": 0.6809982657432556, "learning_rate": 0.0001, "elbo": 3.51801, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.245836265540699}
{"step": 72405, "loss": 3.527667999267578, "grad_norm": 0.6942684650421143, "learning_rate": 0.0001, "elbo": 3.52767, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.24612948627727}
{"step": 72410, "loss": 3.516063690185547, "grad_norm": 0.9731834530830383, "learning_rate": 0.0001, "elbo": 3.51606, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.24642270701384}
{"step": 72415, "loss": 3.5131717681884767, "grad_norm": 0.6960932612419128, "learning_rate": 0.0001, "elbo": 3.51317, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.246715927750411}
{"step": 72420, "loss": 3.5136009216308595, "grad_norm": 0.7177357077598572, "learning_rate": 0.0001, "elbo": 3.5136, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.247009148486981}
{"step": 72425, "loss": 3.511612319946289, "grad_norm": 0.7038917541503906, "learning_rate": 0.0001, "elbo": 3.51161, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.247302369223552}
{"step": 72430, "loss": 3.5114261627197267, "grad_norm": 0.6767722964286804, "learning_rate": 0.0001, "elbo": 3.51143, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.247595589960122}
{"step": 72435, "loss": 3.520466995239258, "grad_norm": 0.7051925659179688, "learning_rate": 0.0001, "elbo": 3.52047, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2478888106966926}
{"step": 72440, "loss": 3.509970855712891, "grad_norm": 0.7253870368003845, "learning_rate": 0.0001, "elbo": 3.50997, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.248182031433263}
{"step": 72445, "loss": 3.516616439819336, "grad_norm": 0.7184137105941772, "learning_rate": 0.0001, "elbo": 3.51662, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2484752521698335}
{"step": 72450, "loss": 3.5194522857666017, "grad_norm": 0.8408230543136597, "learning_rate": 0.0001, "elbo": 3.51945, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.248768472906404}
{"step": 72450, "loss": 3.5194522857666017, "grad_norm": 0.8408230543136597, "learning_rate": 0.0001, "elbo": 3.51945, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.248768472906404}
{"step": 72455, "loss": 3.5252429962158205, "grad_norm": 0.6892476677894592, "learning_rate": 0.0001, "elbo": 3.52524, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.249061693642974}
{"step": 72460, "loss": 3.5060958862304688, "grad_norm": 0.8123564720153809, "learning_rate": 0.0001, "elbo": 3.5061, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.249354914379545}
{"step": 72465, "loss": 3.5252410888671877, "grad_norm": 0.7261009216308594, "learning_rate": 0.0001, "elbo": 3.52524, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.249648135116115}
{"step": 72470, "loss": 3.536199951171875, "grad_norm": 1.0920801162719727, "learning_rate": 0.0001, "elbo": 3.5362, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.249941355852686}
{"step": 72475, "loss": 3.5083168029785154, "grad_norm": 0.749557614326477, "learning_rate": 0.0001, "elbo": 3.50832, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.250234576589256}
{"step": 72480, "loss": 3.503923034667969, "grad_norm": 0.7539618015289307, "learning_rate": 0.0001, "elbo": 3.50392, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.250527797325827}
{"step": 72485, "loss": 3.5172893524169924, "grad_norm": 0.7425560355186462, "learning_rate": 0.0001, "elbo": 3.51729, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.250821018062397}
{"step": 72490, "loss": 3.5237274169921875, "grad_norm": 0.6570227742195129, "learning_rate": 0.0001, "elbo": 3.52373, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2511142387989675}
{"step": 72495, "loss": 3.5035018920898438, "grad_norm": 0.6771951913833618, "learning_rate": 0.0001, "elbo": 3.5035, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.251407459535538}
{"step": 72500, "loss": 3.5298858642578126, "grad_norm": 1.020544409751892, "learning_rate": 0.0001, "elbo": 3.52989, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2517006802721085}
{"step": 72505, "loss": 3.5176773071289062, "grad_norm": 0.7050957083702087, "learning_rate": 0.0001, "elbo": 3.51768, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.25199390100868}
{"step": 72510, "loss": 3.5241928100585938, "grad_norm": 0.6826708912849426, "learning_rate": 0.0001, "elbo": 3.52419, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.252287121745249}
{"step": 72515, "loss": 3.5219802856445312, "grad_norm": 0.690957248210907, "learning_rate": 0.0001, "elbo": 3.52198, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.252580342481821}
{"step": 72520, "loss": 3.519039535522461, "grad_norm": 0.6857047080993652, "learning_rate": 0.0001, "elbo": 3.51904, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.252873563218391}
{"step": 72525, "loss": 3.5281982421875, "grad_norm": 0.7475564479827881, "learning_rate": 0.0001, "elbo": 3.5282, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.253166783954962}
{"step": 72530, "loss": 3.525908660888672, "grad_norm": 0.6870490312576294, "learning_rate": 0.0001, "elbo": 3.52591, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.253460004691532}
{"step": 72535, "loss": 3.527783203125, "grad_norm": 0.7777984142303467, "learning_rate": 0.0001, "elbo": 3.52778, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2537532254281025}
{"step": 72540, "loss": 3.519636535644531, "grad_norm": 0.7445845007896423, "learning_rate": 0.0001, "elbo": 3.51964, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.254046446164673}
{"step": 72545, "loss": 3.531572723388672, "grad_norm": 0.677039623260498, "learning_rate": 0.0001, "elbo": 3.53157, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.254339666901243}
{"step": 72550, "loss": 3.5055557250976563, "grad_norm": 0.7070922255516052, "learning_rate": 0.0001, "elbo": 3.50556, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.254632887637814}
{"step": 72555, "loss": 3.5336593627929687, "grad_norm": 0.6690186262130737, "learning_rate": 0.0001, "elbo": 3.53366, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.254926108374384}
{"step": 72560, "loss": 3.5078857421875, "grad_norm": 0.7133090496063232, "learning_rate": 0.0001, "elbo": 3.50789, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.255219329110955}
{"step": 72565, "loss": 3.518576431274414, "grad_norm": 0.6815499663352966, "learning_rate": 0.0001, "elbo": 3.51858, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.255512549847525}
{"step": 72570, "loss": 3.4977127075195313, "grad_norm": 0.7240555882453918, "learning_rate": 0.0001, "elbo": 3.49771, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.255805770584096}
{"step": 72575, "loss": 3.535849761962891, "grad_norm": 0.700194776058197, "learning_rate": 0.0001, "elbo": 3.53585, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.256098991320666}
{"step": 72580, "loss": 3.531093978881836, "grad_norm": 0.6994568705558777, "learning_rate": 0.0001, "elbo": 3.53109, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.256392212057237}
{"step": 72585, "loss": 3.52905158996582, "grad_norm": 0.6923392415046692, "learning_rate": 0.0001, "elbo": 3.52905, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.256685432793807}
{"step": 72590, "loss": 3.5301170349121094, "grad_norm": 0.7141810059547424, "learning_rate": 0.0001, "elbo": 3.53012, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2569786535303775}
{"step": 72595, "loss": 3.5098342895507812, "grad_norm": 0.7474148869514465, "learning_rate": 0.0001, "elbo": 3.50983, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.257271874266948}
{"step": 72600, "loss": 3.5284095764160157, "grad_norm": 0.7987759709358215, "learning_rate": 0.0001, "elbo": 3.52841, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.257565095003518}
{"step": 72605, "loss": 3.5141063690185548, "grad_norm": 0.7282266020774841, "learning_rate": 0.0001, "elbo": 3.51411, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.257858315740089}
{"step": 72610, "loss": 3.513968658447266, "grad_norm": 0.683153510093689, "learning_rate": 0.0001, "elbo": 3.51397, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.258151536476659}
{"step": 72615, "loss": 3.5368118286132812, "grad_norm": 0.795654296875, "learning_rate": 0.0001, "elbo": 3.53681, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.25844475721323}
{"step": 72620, "loss": 3.528407669067383, "grad_norm": 0.967511773109436, "learning_rate": 0.0001, "elbo": 3.52841, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2587379779498}
{"step": 72625, "loss": 3.5191978454589843, "grad_norm": 0.6929070949554443, "learning_rate": 0.0001, "elbo": 3.5192, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.259031198686371}
{"step": 72630, "loss": 3.5139801025390627, "grad_norm": 0.7161511182785034, "learning_rate": 0.0001, "elbo": 3.51398, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.259324419422942}
{"step": 72635, "loss": 3.508046340942383, "grad_norm": 0.7312045693397522, "learning_rate": 0.0001, "elbo": 3.50805, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.259617640159512}
{"step": 72640, "loss": 3.5066902160644533, "grad_norm": 0.6749855875968933, "learning_rate": 0.0001, "elbo": 3.50669, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.259910860896083}
{"step": 72645, "loss": 3.505845642089844, "grad_norm": 0.7528389096260071, "learning_rate": 0.0001, "elbo": 3.50585, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.260204081632653}
{"step": 72650, "loss": 3.5115196228027346, "grad_norm": 0.729243814945221, "learning_rate": 0.0001, "elbo": 3.51152, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.260497302369224}
{"step": 72655, "loss": 3.5344600677490234, "grad_norm": 0.7774548530578613, "learning_rate": 0.0001, "elbo": 3.53446, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.260790523105794}
{"step": 72660, "loss": 3.5124221801757813, "grad_norm": 0.9249125123023987, "learning_rate": 0.0001, "elbo": 3.51242, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.261083743842365}
{"step": 72665, "loss": 3.515715789794922, "grad_norm": 0.7342432737350464, "learning_rate": 0.0001, "elbo": 3.51572, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.261376964578935}
{"step": 72670, "loss": 3.5076068878173827, "grad_norm": 0.6756508946418762, "learning_rate": 0.0001, "elbo": 3.50761, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.261670185315506}
{"step": 72675, "loss": 3.5191383361816406, "grad_norm": 0.7572547793388367, "learning_rate": 0.0001, "elbo": 3.51914, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.261963406052076}
{"step": 72680, "loss": 3.5302749633789063, "grad_norm": 0.6864510178565979, "learning_rate": 0.0001, "elbo": 3.53027, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.262256626788647}
{"step": 72685, "loss": 3.5272216796875, "grad_norm": 0.7427898049354553, "learning_rate": 0.0001, "elbo": 3.52722, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.262549847525217}
{"step": 72690, "loss": 3.508802032470703, "grad_norm": 0.7024586200714111, "learning_rate": 0.0001, "elbo": 3.5088, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2628430682617875}
{"step": 72695, "loss": 3.516519546508789, "grad_norm": 0.7046793103218079, "learning_rate": 0.0001, "elbo": 3.51652, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.263136288998358}
{"step": 72700, "loss": 3.5226669311523438, "grad_norm": 0.8255181908607483, "learning_rate": 0.0001, "elbo": 3.52267, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.263429509734928}
{"step": 72705, "loss": 3.5092071533203124, "grad_norm": 0.7151389718055725, "learning_rate": 0.0001, "elbo": 3.50921, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.263722730471499}
{"step": 72710, "loss": 3.5428531646728514, "grad_norm": 0.8136974573135376, "learning_rate": 0.0001, "elbo": 3.54285, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.264015951208069}
{"step": 72715, "loss": 3.5228981018066405, "grad_norm": 0.7216929793357849, "learning_rate": 0.0001, "elbo": 3.5229, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.26430917194464}
{"step": 72720, "loss": 3.5403160095214843, "grad_norm": 0.7409573197364807, "learning_rate": 0.0001, "elbo": 3.54032, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.26460239268121}
{"step": 72725, "loss": 3.5233200073242186, "grad_norm": 0.8522782325744629, "learning_rate": 0.0001, "elbo": 3.52332, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.264895613417781}
{"step": 72730, "loss": 3.522055816650391, "grad_norm": 0.7192168235778809, "learning_rate": 0.0001, "elbo": 3.52206, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.265188834154351}
{"step": 72735, "loss": 3.533477020263672, "grad_norm": 0.8494004011154175, "learning_rate": 0.0001, "elbo": 3.53348, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.265482054890922}
{"step": 72740, "loss": 3.5279388427734375, "grad_norm": 0.7139859795570374, "learning_rate": 0.0001, "elbo": 3.52794, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.265775275627492}
{"step": 72745, "loss": 3.5168907165527346, "grad_norm": 0.6990256905555725, "learning_rate": 0.0001, "elbo": 3.51689, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2660684963640625}
{"step": 72750, "loss": 3.5168182373046877, "grad_norm": 0.7167084217071533, "learning_rate": 0.0001, "elbo": 3.51682, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.266361717100633}
{"step": 72755, "loss": 3.5182876586914062, "grad_norm": 0.7280315160751343, "learning_rate": 0.0001, "elbo": 3.51829, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.266654937837204}
{"step": 72760, "loss": 3.510081481933594, "grad_norm": 0.7525776624679565, "learning_rate": 0.0001, "elbo": 3.51008, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.266948158573775}
{"step": 72765, "loss": 3.517180633544922, "grad_norm": 1.0031787157058716, "learning_rate": 0.0001, "elbo": 3.51718, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.267241379310345}
{"step": 72770, "loss": 3.530366897583008, "grad_norm": 0.7041363716125488, "learning_rate": 0.0001, "elbo": 3.53037, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.267534600046916}
{"step": 72775, "loss": 3.524214172363281, "grad_norm": 0.7205834984779358, "learning_rate": 0.0001, "elbo": 3.52421, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.267827820783486}
{"step": 72780, "loss": 3.510894012451172, "grad_norm": 0.8656210899353027, "learning_rate": 0.0001, "elbo": 3.51089, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.268121041520057}
{"step": 72780, "loss": 3.510894012451172, "grad_norm": 0.8656210899353027, "learning_rate": 0.0001, "elbo": 3.51089, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.268121041520057}
{"step": 72785, "loss": 3.505840301513672, "grad_norm": 1.2679845094680786, "learning_rate": 0.0001, "elbo": 3.50584, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.268414262256627}
{"step": 72790, "loss": 3.5302574157714846, "grad_norm": 0.6881465315818787, "learning_rate": 0.0001, "elbo": 3.53026, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2687074829931975}
{"step": 72795, "loss": 3.5032779693603517, "grad_norm": 0.7284318208694458, "learning_rate": 0.0001, "elbo": 3.50328, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.269000703729768}
{"step": 72800, "loss": 3.520904541015625, "grad_norm": 0.7193819880485535, "learning_rate": 0.0001, "elbo": 3.5209, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.269293924466338}
{"step": 72805, "loss": 3.5244346618652345, "grad_norm": 0.6989263296127319, "learning_rate": 0.0001, "elbo": 3.52443, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.269587145202909}
{"step": 72810, "loss": 3.532997894287109, "grad_norm": 1.8880970478057861, "learning_rate": 0.0001, "elbo": 3.533, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.269880365939479}
{"step": 72810, "loss": 3.532997894287109, "grad_norm": 1.8880970478057861, "learning_rate": 0.0001, "elbo": 3.533, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.269880365939479}
{"step": 72815, "loss": 3.5294551849365234, "grad_norm": 0.9026040434837341, "learning_rate": 0.0001, "elbo": 3.52946, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.27017358667605}
{"step": 72820, "loss": 3.5271347045898436, "grad_norm": 0.861577570438385, "learning_rate": 0.0001, "elbo": 3.52713, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.27046680741262}
{"step": 72825, "loss": 3.5205894470214845, "grad_norm": 0.7151982188224792, "learning_rate": 0.0001, "elbo": 3.52059, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.270760028149191}
{"step": 72830, "loss": 3.5167377471923826, "grad_norm": 0.6817612648010254, "learning_rate": 0.0001, "elbo": 3.51674, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.271053248885761}
{"step": 72835, "loss": 3.507923889160156, "grad_norm": 0.7675847411155701, "learning_rate": 0.0001, "elbo": 3.50792, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2713464696223316}
{"step": 72840, "loss": 3.5200096130371095, "grad_norm": 0.7174720764160156, "learning_rate": 0.0001, "elbo": 3.52001, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.271639690358902}
{"step": 72845, "loss": 3.498107147216797, "grad_norm": 0.7298344373703003, "learning_rate": 0.0001, "elbo": 3.49811, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2719329110954725}
{"step": 72850, "loss": 3.5148643493652343, "grad_norm": 0.673744797706604, "learning_rate": 0.0001, "elbo": 3.51486, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.272226131832043}
{"step": 72855, "loss": 3.524934768676758, "grad_norm": 0.6863477826118469, "learning_rate": 0.0001, "elbo": 3.52493, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.272519352568613}
{"step": 72860, "loss": 3.5014316558837892, "grad_norm": 0.7070218324661255, "learning_rate": 0.0001, "elbo": 3.50143, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.272812573305184}
{"step": 72865, "loss": 3.5150318145751953, "grad_norm": 0.7067010402679443, "learning_rate": 0.0001, "elbo": 3.51503, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.273105794041754}
{"step": 72870, "loss": 3.5191307067871094, "grad_norm": 0.7422610521316528, "learning_rate": 0.0001, "elbo": 3.51913, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.273399014778325}
{"step": 72875, "loss": 3.5168670654296874, "grad_norm": 0.7266939282417297, "learning_rate": 0.0001, "elbo": 3.51687, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.273692235514895}
{"step": 72880, "loss": 3.5207427978515624, "grad_norm": 0.7204188108444214, "learning_rate": 0.0001, "elbo": 3.52074, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2739854562514665}
{"step": 72885, "loss": 3.5268150329589845, "grad_norm": 0.7326986789703369, "learning_rate": 0.0001, "elbo": 3.52682, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.274278676988037}
{"step": 72890, "loss": 3.5258235931396484, "grad_norm": 0.7004417181015015, "learning_rate": 0.0001, "elbo": 3.52582, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.274571897724607}
{"step": 72895, "loss": 3.5047203063964845, "grad_norm": 0.7447075247764587, "learning_rate": 0.0001, "elbo": 3.50472, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.274865118461178}
{"step": 72900, "loss": 3.523426055908203, "grad_norm": 0.9143052101135254, "learning_rate": 0.0001, "elbo": 3.52343, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.275158339197748}
{"step": 72905, "loss": 3.5070213317871093, "grad_norm": 0.8244414329528809, "learning_rate": 0.0001, "elbo": 3.50702, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.275451559934319}
{"step": 72910, "loss": 3.5091957092285155, "grad_norm": 0.8227055072784424, "learning_rate": 0.0001, "elbo": 3.5092, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.275744780670889}
{"step": 72915, "loss": 3.518502044677734, "grad_norm": 0.7805238962173462, "learning_rate": 0.0001, "elbo": 3.5185, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.27603800140746}
{"step": 72920, "loss": 3.516478729248047, "grad_norm": 0.7025121450424194, "learning_rate": 0.0001, "elbo": 3.51648, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.27633122214403}
{"step": 72925, "loss": 3.5235023498535156, "grad_norm": 0.7057742476463318, "learning_rate": 0.0001, "elbo": 3.5235, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.276624442880601}
{"step": 72930, "loss": 3.510865020751953, "grad_norm": 0.7164615392684937, "learning_rate": 0.0001, "elbo": 3.51087, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.276917663617171}
{"step": 72935, "loss": 3.535547637939453, "grad_norm": 1.0006004571914673, "learning_rate": 0.0001, "elbo": 3.53555, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2772108843537415}
{"step": 72940, "loss": 3.524617385864258, "grad_norm": 0.8318625688552856, "learning_rate": 0.0001, "elbo": 3.52462, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.277504105090312}
{"step": 72945, "loss": 3.517958068847656, "grad_norm": 0.8754159808158875, "learning_rate": 0.0001, "elbo": 3.51796, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.277797325826882}
{"step": 72950, "loss": 3.5090190887451174, "grad_norm": 0.7872068285942078, "learning_rate": 0.0001, "elbo": 3.50902, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.278090546563453}
{"step": 72955, "loss": 3.500675582885742, "grad_norm": 0.7311024069786072, "learning_rate": 0.0001, "elbo": 3.50068, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.278383767300023}
{"step": 72960, "loss": 3.5327373504638673, "grad_norm": 0.6658381223678589, "learning_rate": 0.0001, "elbo": 3.53274, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.278676988036594}
{"step": 72960, "loss": 3.5327373504638673, "grad_norm": 0.6658381223678589, "learning_rate": 0.0001, "elbo": 3.53274, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.278676988036594}
{"step": 72965, "loss": 3.5067264556884767, "grad_norm": 0.712515115737915, "learning_rate": 0.0001, "elbo": 3.50673, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.278970208773164}
{"step": 72970, "loss": 3.513245391845703, "grad_norm": 0.6925068497657776, "learning_rate": 0.0001, "elbo": 3.51325, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.279263429509735}
{"step": 72975, "loss": 3.513629150390625, "grad_norm": 0.6751078963279724, "learning_rate": 0.0001, "elbo": 3.51363, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.279556650246305}
{"step": 72980, "loss": 3.5024272918701174, "grad_norm": 0.7037095427513123, "learning_rate": 0.0001, "elbo": 3.50243, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.279849870982876}
{"step": 72980, "loss": 3.5024272918701174, "grad_norm": 0.7037095427513123, "learning_rate": 0.0001, "elbo": 3.50243, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.279849870982876}
{"step": 72985, "loss": 3.5254837036132813, "grad_norm": 0.7228723764419556, "learning_rate": 0.0001, "elbo": 3.52548, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.280143091719446}
{"step": 72990, "loss": 3.5196456909179688, "grad_norm": 0.6658527255058289, "learning_rate": 0.0001, "elbo": 3.51965, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2804363124560165}
{"step": 72995, "loss": 3.5087989807128905, "grad_norm": 0.653659462928772, "learning_rate": 0.0001, "elbo": 3.5088, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.280729533192587}
{"step": 73000, "loss": 3.5157737731933594, "grad_norm": 0.6830219030380249, "learning_rate": 0.0001, "elbo": 3.51577, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.281022753929157}
{"step": 73005, "loss": 3.5147377014160157, "grad_norm": 0.6867336630821228, "learning_rate": 0.0001, "elbo": 3.51474, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.281315974665729}
{"step": 73010, "loss": 3.496856689453125, "grad_norm": 0.7098380923271179, "learning_rate": 0.0001, "elbo": 3.49686, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.281609195402299}
{"step": 73010, "loss": 3.496856689453125, "grad_norm": 0.7098380923271179, "learning_rate": 0.0001, "elbo": 3.49686, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.281609195402299}
{"step": 73015, "loss": 3.5292953491210937, "grad_norm": 0.6714481711387634, "learning_rate": 0.0001, "elbo": 3.5293, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.28190241613887}
{"step": 73020, "loss": 3.518581008911133, "grad_norm": 0.6873718500137329, "learning_rate": 0.0001, "elbo": 3.51858, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.28219563687544}
{"step": 73025, "loss": 3.5089340209960938, "grad_norm": 0.9168137907981873, "learning_rate": 0.0001, "elbo": 3.50893, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.282488857612011}
{"step": 73030, "loss": 3.5197265625, "grad_norm": 0.7258184552192688, "learning_rate": 0.0001, "elbo": 3.51973, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.282782078348581}
{"step": 73035, "loss": 3.5166011810302735, "grad_norm": 0.6476573944091797, "learning_rate": 0.0001, "elbo": 3.5166, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2830752990851515}
{"step": 73040, "loss": 3.4960403442382812, "grad_norm": 0.6805199384689331, "learning_rate": 0.0001, "elbo": 3.49604, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.283368519821722}
{"step": 73045, "loss": 3.5157936096191404, "grad_norm": 0.6713719964027405, "learning_rate": 0.0001, "elbo": 3.51579, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.283661740558292}
{"step": 73050, "loss": 3.5157745361328123, "grad_norm": 0.7182546854019165, "learning_rate": 0.0001, "elbo": 3.51577, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.283954961294863}
{"step": 73055, "loss": 3.524729919433594, "grad_norm": 0.6954464912414551, "learning_rate": 0.0001, "elbo": 3.52473, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.284248182031433}
{"step": 73060, "loss": 3.5273773193359377, "grad_norm": 0.7387630939483643, "learning_rate": 0.0001, "elbo": 3.52738, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.284541402768004}
{"step": 73065, "loss": 3.5201560974121096, "grad_norm": 1.0109202861785889, "learning_rate": 0.0001, "elbo": 3.52016, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.284834623504574}
{"step": 73070, "loss": 3.523065185546875, "grad_norm": 0.7439244389533997, "learning_rate": 0.0001, "elbo": 3.52307, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.285127844241145}
{"step": 73070, "loss": 3.523065185546875, "grad_norm": 0.7439244389533997, "learning_rate": 0.0001, "elbo": 3.52307, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.285127844241145}
{"step": 73075, "loss": 3.515947723388672, "grad_norm": 0.7439035773277283, "learning_rate": 0.0001, "elbo": 3.51595, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.285421064977715}
{"step": 73080, "loss": 3.5355567932128906, "grad_norm": 0.7762873768806458, "learning_rate": 0.0001, "elbo": 3.53556, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.285714285714286}
{"step": 73085, "loss": 3.506414794921875, "grad_norm": 0.7420251965522766, "learning_rate": 0.0001, "elbo": 3.50641, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.286007506450856}
{"step": 73090, "loss": 3.51199951171875, "grad_norm": 0.8854435086250305, "learning_rate": 0.0001, "elbo": 3.512, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2863007271874265}
{"step": 73095, "loss": 3.5203697204589846, "grad_norm": 0.8088045716285706, "learning_rate": 0.0001, "elbo": 3.52037, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.286593947923997}
{"step": 73100, "loss": 3.514128875732422, "grad_norm": 0.7130889892578125, "learning_rate": 0.0001, "elbo": 3.51413, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.286887168660567}
{"step": 73105, "loss": 3.51712646484375, "grad_norm": 0.8106832504272461, "learning_rate": 0.0001, "elbo": 3.51713, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.287180389397138}
{"step": 73110, "loss": 3.5179908752441404, "grad_norm": 0.6873992085456848, "learning_rate": 0.0001, "elbo": 3.51799, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.287473610133708}
{"step": 73115, "loss": 3.5137069702148436, "grad_norm": 0.6927464008331299, "learning_rate": 0.0001, "elbo": 3.51371, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.287766830870279}
{"step": 73120, "loss": 3.5279502868652344, "grad_norm": 0.7350447177886963, "learning_rate": 0.0001, "elbo": 3.52795, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.288060051606849}
{"step": 73125, "loss": 3.512866973876953, "grad_norm": 0.7374989986419678, "learning_rate": 0.0001, "elbo": 3.51287, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.28835327234342}
{"step": 73130, "loss": 3.522142791748047, "grad_norm": 0.6857773065567017, "learning_rate": 0.0001, "elbo": 3.52214, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.288646493079991}
{"step": 73135, "loss": 3.5223434448242186, "grad_norm": 0.7274559736251831, "learning_rate": 0.0001, "elbo": 3.52234, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2889397138165615}
{"step": 73140, "loss": 3.526318359375, "grad_norm": 0.7147220373153687, "learning_rate": 0.0001, "elbo": 3.52632, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.289232934553132}
{"step": 73145, "loss": 3.5011547088623045, "grad_norm": 0.7532903552055359, "learning_rate": 0.0001, "elbo": 3.50115, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.289526155289702}
{"step": 73150, "loss": 3.507640075683594, "grad_norm": 0.6776424050331116, "learning_rate": 0.0001, "elbo": 3.50764, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.289819376026273}
{"step": 73155, "loss": 3.503279113769531, "grad_norm": 0.6973073482513428, "learning_rate": 0.0001, "elbo": 3.50328, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.290112596762843}
{"step": 73160, "loss": 3.5133987426757813, "grad_norm": 0.6835080981254578, "learning_rate": 0.0001, "elbo": 3.5134, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.290405817499414}
{"step": 73160, "loss": 3.5133987426757813, "grad_norm": 0.6835080981254578, "learning_rate": 0.0001, "elbo": 3.5134, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.290405817499414}
{"step": 73165, "loss": 3.5116222381591795, "grad_norm": 0.8584544658660889, "learning_rate": 0.0001, "elbo": 3.51162, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.290699038235984}
{"step": 73170, "loss": 3.5290374755859375, "grad_norm": 0.6801099181175232, "learning_rate": 0.0001, "elbo": 3.52904, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.290992258972555}
{"step": 73175, "loss": 3.4905242919921875, "grad_norm": 0.7977195382118225, "learning_rate": 0.0001, "elbo": 3.49052, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.291285479709125}
{"step": 73180, "loss": 3.5268043518066405, "grad_norm": 0.834290087223053, "learning_rate": 0.0001, "elbo": 3.5268, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.291578700445696}
{"step": 73185, "loss": 3.4942901611328123, "grad_norm": 0.6740142703056335, "learning_rate": 0.0001, "elbo": 3.49429, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.291871921182266}
{"step": 73190, "loss": 3.5233436584472657, "grad_norm": 0.7544289827346802, "learning_rate": 0.0001, "elbo": 3.52334, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2921651419188365}
{"step": 73195, "loss": 3.524403381347656, "grad_norm": 0.7523499727249146, "learning_rate": 0.0001, "elbo": 3.5244, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.292458362655407}
{"step": 73200, "loss": 3.510173797607422, "grad_norm": 0.7512105703353882, "learning_rate": 0.0001, "elbo": 3.51017, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.292751583391977}
{"step": 73205, "loss": 3.5189983367919924, "grad_norm": 0.7464486956596375, "learning_rate": 0.0001, "elbo": 3.519, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.293044804128548}
{"step": 73210, "loss": 3.5123409271240233, "grad_norm": 0.774648129940033, "learning_rate": 0.0001, "elbo": 3.51234, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.293338024865118}
{"step": 73215, "loss": 3.5272117614746095, "grad_norm": 0.828252375125885, "learning_rate": 0.0001, "elbo": 3.52721, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.293631245601689}
{"step": 73220, "loss": 3.506320571899414, "grad_norm": 0.7032307386398315, "learning_rate": 0.0001, "elbo": 3.50632, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.293924466338259}
{"step": 73225, "loss": 3.506531524658203, "grad_norm": 0.9219020009040833, "learning_rate": 0.0001, "elbo": 3.50653, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.29421768707483}
{"step": 73230, "loss": 3.5210105895996096, "grad_norm": 0.705162763595581, "learning_rate": 0.0001, "elbo": 3.52101, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2945109078114}
{"step": 73235, "loss": 3.5145572662353515, "grad_norm": 0.707615852355957, "learning_rate": 0.0001, "elbo": 3.51456, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.294804128547971}
{"step": 73240, "loss": 3.5117820739746093, "grad_norm": 0.6919819116592407, "learning_rate": 0.0001, "elbo": 3.51178, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.295097349284541}
{"step": 73245, "loss": 3.52282829284668, "grad_norm": 0.8452864289283752, "learning_rate": 0.0001, "elbo": 3.52283, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.295390570021112}
{"step": 73250, "loss": 3.5190963745117188, "grad_norm": 0.7319586277008057, "learning_rate": 0.0001, "elbo": 3.5191, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.295683790757682}
{"step": 73255, "loss": 3.511457061767578, "grad_norm": 0.7669845223426819, "learning_rate": 0.0001, "elbo": 3.51146, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.295977011494253}
{"step": 73260, "loss": 3.5195972442626955, "grad_norm": 0.6834975481033325, "learning_rate": 0.0001, "elbo": 3.5196, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.296270232230824}
{"step": 73265, "loss": 3.5204620361328125, "grad_norm": 0.7494569420814514, "learning_rate": 0.0001, "elbo": 3.52046, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.296563452967394}
{"step": 73270, "loss": 3.501464080810547, "grad_norm": 0.8690378069877625, "learning_rate": 0.0001, "elbo": 3.50146, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.296856673703965}
{"step": 73275, "loss": 3.494135284423828, "grad_norm": 0.8013065457344055, "learning_rate": 0.0001, "elbo": 3.49414, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.297149894440535}
{"step": 73280, "loss": 3.518100357055664, "grad_norm": 0.743767261505127, "learning_rate": 0.0001, "elbo": 3.5181, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.2974431151771055}
{"step": 73285, "loss": 3.512835693359375, "grad_norm": 0.7269346714019775, "learning_rate": 0.0001, "elbo": 3.51284, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.297736335913676}
{"step": 73290, "loss": 3.5250221252441407, "grad_norm": 0.9104739427566528, "learning_rate": 0.0001, "elbo": 3.52502, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.298029556650246}
{"step": 73295, "loss": 3.5343696594238283, "grad_norm": 0.7318243384361267, "learning_rate": 0.0001, "elbo": 3.53437, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.298322777386817}
{"step": 73295, "loss": 3.5343696594238283, "grad_norm": 0.7318243384361267, "learning_rate": 0.0001, "elbo": 3.53437, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.298322777386817}
{"step": 73300, "loss": 3.518238067626953, "grad_norm": 0.682939350605011, "learning_rate": 0.0001, "elbo": 3.51824, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.298615998123387}
{"step": 73305, "loss": 3.533642578125, "grad_norm": 0.6820069551467896, "learning_rate": 0.0001, "elbo": 3.53364, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.298909218859958}
{"step": 73310, "loss": 3.5017189025878905, "grad_norm": 0.6756234765052795, "learning_rate": 0.0001, "elbo": 3.50172, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.299202439596528}
{"step": 73310, "loss": 3.5017189025878905, "grad_norm": 0.6756234765052795, "learning_rate": 0.0001, "elbo": 3.50172, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.299202439596528}
{"step": 73315, "loss": 3.513782501220703, "grad_norm": 0.714418351650238, "learning_rate": 0.0001, "elbo": 3.51378, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.299495660333099}
{"step": 73320, "loss": 3.509769821166992, "grad_norm": 0.6874726414680481, "learning_rate": 0.0001, "elbo": 3.50977, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.299788881069669}
{"step": 73325, "loss": 3.527567672729492, "grad_norm": 0.7138535976409912, "learning_rate": 0.0001, "elbo": 3.52757, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.30008210180624}
{"step": 73330, "loss": 3.520447540283203, "grad_norm": 0.7208107709884644, "learning_rate": 0.0001, "elbo": 3.52045, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.30037532254281}
{"step": 73335, "loss": 3.5069374084472655, "grad_norm": 0.6850228905677795, "learning_rate": 0.0001, "elbo": 3.50694, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3006685432793805}
{"step": 73340, "loss": 3.505879211425781, "grad_norm": 0.6946961879730225, "learning_rate": 0.0001, "elbo": 3.50588, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.300961764015951}
{"step": 73345, "loss": 3.515597915649414, "grad_norm": 0.7247371673583984, "learning_rate": 0.0001, "elbo": 3.5156, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.301254984752521}
{"step": 73350, "loss": 3.5066902160644533, "grad_norm": 0.716053307056427, "learning_rate": 0.0001, "elbo": 3.50669, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.301548205489092}
{"step": 73355, "loss": 3.5202716827392577, "grad_norm": 0.6910130381584167, "learning_rate": 0.0001, "elbo": 3.52027, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.301841426225662}
{"step": 73360, "loss": 3.5127418518066404, "grad_norm": 0.6892812848091125, "learning_rate": 0.0001, "elbo": 3.51274, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.302134646962233}
{"step": 73365, "loss": 3.5194633483886717, "grad_norm": 1.0625667572021484, "learning_rate": 0.0001, "elbo": 3.51946, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.302427867698803}
{"step": 73370, "loss": 3.532733154296875, "grad_norm": 0.745166540145874, "learning_rate": 0.0001, "elbo": 3.53273, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.302721088435375}
{"step": 73375, "loss": 3.5035972595214844, "grad_norm": 0.7868924140930176, "learning_rate": 0.0001, "elbo": 3.5036, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.303014309171945}
{"step": 73380, "loss": 3.538195037841797, "grad_norm": 0.7075830101966858, "learning_rate": 0.0001, "elbo": 3.5382, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3033075299085155}
{"step": 73385, "loss": 3.5114234924316405, "grad_norm": 0.7454249262809753, "learning_rate": 0.0001, "elbo": 3.51142, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.303600750645086}
{"step": 73390, "loss": 3.519720458984375, "grad_norm": 0.8535715937614441, "learning_rate": 0.0001, "elbo": 3.51972, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.303893971381656}
{"step": 73395, "loss": 3.513134765625, "grad_norm": 0.8877889513969421, "learning_rate": 0.0001, "elbo": 3.51313, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.304187192118227}
{"step": 73400, "loss": 3.5018333435058593, "grad_norm": 0.7091515064239502, "learning_rate": 0.0001, "elbo": 3.50183, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.304480412854797}
{"step": 73405, "loss": 3.5211658477783203, "grad_norm": 0.690548300743103, "learning_rate": 0.0001, "elbo": 3.52117, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.304773633591368}
{"step": 73410, "loss": 3.519414520263672, "grad_norm": 0.8772597312927246, "learning_rate": 0.0001, "elbo": 3.51941, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.305066854327938}
{"step": 73415, "loss": 3.5328643798828123, "grad_norm": 0.7612760663032532, "learning_rate": 0.0001, "elbo": 3.53286, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.305360075064509}
{"step": 73420, "loss": 3.5063861846923827, "grad_norm": 0.7120363116264343, "learning_rate": 0.0001, "elbo": 3.50639, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.305653295801079}
{"step": 73425, "loss": 3.526792526245117, "grad_norm": 0.7537283897399902, "learning_rate": 0.0001, "elbo": 3.52679, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.30594651653765}
{"step": 73430, "loss": 3.518381118774414, "grad_norm": 0.8379678726196289, "learning_rate": 0.0001, "elbo": 3.51838, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.30623973727422}
{"step": 73435, "loss": 3.5243717193603517, "grad_norm": 0.7400804758071899, "learning_rate": 0.0001, "elbo": 3.52437, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3065329580107905}
{"step": 73440, "loss": 3.5180099487304686, "grad_norm": 0.958086371421814, "learning_rate": 0.0001, "elbo": 3.51801, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.306826178747361}
{"step": 73445, "loss": 3.521726608276367, "grad_norm": 0.7965397834777832, "learning_rate": 0.0001, "elbo": 3.52173, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.307119399483931}
{"step": 73450, "loss": 3.5179420471191407, "grad_norm": 0.6966766715049744, "learning_rate": 0.0001, "elbo": 3.51794, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.307412620220502}
{"step": 73455, "loss": 3.5119178771972654, "grad_norm": 0.7740170359611511, "learning_rate": 0.0001, "elbo": 3.51192, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.307705840957072}
{"step": 73460, "loss": 3.5223762512207033, "grad_norm": 0.7640484571456909, "learning_rate": 0.0001, "elbo": 3.52238, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.307999061693643}
{"step": 73465, "loss": 3.5113353729248047, "grad_norm": 0.7006640434265137, "learning_rate": 0.0001, "elbo": 3.51134, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.308292282430213}
{"step": 73470, "loss": 3.5055118560791017, "grad_norm": 0.7270671129226685, "learning_rate": 0.0001, "elbo": 3.50551, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.308585503166784}
{"step": 73475, "loss": 3.5155494689941404, "grad_norm": 1.1187471151351929, "learning_rate": 0.0001, "elbo": 3.51555, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.308878723903354}
{"step": 73480, "loss": 3.511038589477539, "grad_norm": 1.0670756101608276, "learning_rate": 0.0001, "elbo": 3.51104, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.309171944639925}
{"step": 73485, "loss": 3.536904144287109, "grad_norm": 0.7133246660232544, "learning_rate": 0.0001, "elbo": 3.5369, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.309465165376495}
{"step": 73490, "loss": 3.5203094482421875, "grad_norm": 0.7242630124092102, "learning_rate": 0.0001, "elbo": 3.52031, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3097583861130655}
{"step": 73495, "loss": 3.5112468719482424, "grad_norm": 0.7509570717811584, "learning_rate": 0.0001, "elbo": 3.51125, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.310051606849637}
{"step": 73500, "loss": 3.5280399322509766, "grad_norm": 0.6962301731109619, "learning_rate": 0.0001, "elbo": 3.52804, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.310344827586207}
{"step": 73505, "loss": 3.5171875, "grad_norm": 0.8325303196907043, "learning_rate": 0.0001, "elbo": 3.51719, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.310638048322778}
{"step": 73510, "loss": 3.5222686767578124, "grad_norm": 0.7201026678085327, "learning_rate": 0.0001, "elbo": 3.52227, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.310931269059348}
{"step": 73515, "loss": 3.5039634704589844, "grad_norm": 0.708005428314209, "learning_rate": 0.0001, "elbo": 3.50396, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.311224489795919}
{"step": 73520, "loss": 3.507598114013672, "grad_norm": 0.6738067865371704, "learning_rate": 0.0001, "elbo": 3.5076, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.311517710532489}
{"step": 73525, "loss": 3.511335754394531, "grad_norm": 0.716602087020874, "learning_rate": 0.0001, "elbo": 3.51134, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.31181093126906}
{"step": 73530, "loss": 3.533267593383789, "grad_norm": 0.7135238647460938, "learning_rate": 0.0001, "elbo": 3.53327, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.31210415200563}
{"step": 73535, "loss": 3.5069122314453125, "grad_norm": 0.6989660859107971, "learning_rate": 0.0001, "elbo": 3.50691, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3123973727422005}
{"step": 73540, "loss": 3.5330215454101563, "grad_norm": 0.7809861302375793, "learning_rate": 0.0001, "elbo": 3.53302, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.312690593478771}
{"step": 73545, "loss": 3.5239768981933595, "grad_norm": 0.7186340689659119, "learning_rate": 0.0001, "elbo": 3.52398, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.312983814215341}
{"step": 73550, "loss": 3.5264286041259765, "grad_norm": 0.6873016357421875, "learning_rate": 0.0001, "elbo": 3.52643, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.313277034951912}
{"step": 73555, "loss": 3.5208126068115235, "grad_norm": 0.7004764676094055, "learning_rate": 0.0001, "elbo": 3.52081, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.313570255688482}
{"step": 73560, "loss": 3.519105911254883, "grad_norm": 0.7411516904830933, "learning_rate": 0.0001, "elbo": 3.51911, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.313863476425053}
{"step": 73565, "loss": 3.519257736206055, "grad_norm": 0.6862341165542603, "learning_rate": 0.0001, "elbo": 3.51926, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.314156697161623}
{"step": 73570, "loss": 3.5098731994628904, "grad_norm": 0.7076048851013184, "learning_rate": 0.0001, "elbo": 3.50987, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.314449917898194}
{"step": 73575, "loss": 3.5206722259521483, "grad_norm": 0.7323577404022217, "learning_rate": 0.0001, "elbo": 3.52067, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.314743138634764}
{"step": 73580, "loss": 3.538745880126953, "grad_norm": 0.675940752029419, "learning_rate": 0.0001, "elbo": 3.53875, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.315036359371335}
{"step": 73585, "loss": 3.5320323944091796, "grad_norm": 0.6898025870323181, "learning_rate": 0.0001, "elbo": 3.53203, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.315329580107905}
{"step": 73590, "loss": 3.5175376892089845, "grad_norm": 0.6868788599967957, "learning_rate": 0.0001, "elbo": 3.51754, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3156228008444755}
{"step": 73595, "loss": 3.510980987548828, "grad_norm": 0.7387495040893555, "learning_rate": 0.0001, "elbo": 3.51098, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.315916021581046}
{"step": 73600, "loss": 3.5246963500976562, "grad_norm": 0.7264190316200256, "learning_rate": 0.0001, "elbo": 3.5247, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.316209242317616}
{"step": 73605, "loss": 3.5136734008789063, "grad_norm": 0.681171715259552, "learning_rate": 0.0001, "elbo": 3.51367, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.316502463054187}
{"step": 73610, "loss": 3.5223365783691407, "grad_norm": 0.7230070233345032, "learning_rate": 0.0001, "elbo": 3.52234, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.316795683790757}
{"step": 73615, "loss": 3.4995712280273437, "grad_norm": 0.766193151473999, "learning_rate": 0.0001, "elbo": 3.49957, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.317088904527328}
{"step": 73620, "loss": 3.5267669677734377, "grad_norm": 0.7124881148338318, "learning_rate": 0.0001, "elbo": 3.52677, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.317382125263899}
{"step": 73625, "loss": 3.509682464599609, "grad_norm": 0.7710884809494019, "learning_rate": 0.0001, "elbo": 3.50968, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3176753460004695}
{"step": 73630, "loss": 3.517431640625, "grad_norm": 0.7080263495445251, "learning_rate": 0.0001, "elbo": 3.51743, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.31796856673704}
{"step": 73635, "loss": 3.5185821533203123, "grad_norm": 0.7509694695472717, "learning_rate": 0.0001, "elbo": 3.51858, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3182617874736104}
{"step": 73640, "loss": 3.525909423828125, "grad_norm": 1.0006781816482544, "learning_rate": 0.0001, "elbo": 3.52591, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.318555008210181}
{"step": 73645, "loss": 3.5237968444824217, "grad_norm": 0.7566245198249817, "learning_rate": 0.0001, "elbo": 3.5238, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.318848228946751}
{"step": 73650, "loss": 3.513600158691406, "grad_norm": 1.0615134239196777, "learning_rate": 0.0001, "elbo": 3.5136, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.319141449683322}
{"step": 73655, "loss": 3.512165069580078, "grad_norm": 0.6879251003265381, "learning_rate": 0.0001, "elbo": 3.51216, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.319434670419892}
{"step": 73660, "loss": 3.5120086669921875, "grad_norm": 0.7216015458106995, "learning_rate": 0.0001, "elbo": 3.51201, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.319727891156463}
{"step": 73665, "loss": 3.5074447631835937, "grad_norm": 0.6964594125747681, "learning_rate": 0.0001, "elbo": 3.50744, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.320021111893033}
{"step": 73670, "loss": 3.5180828094482424, "grad_norm": 0.7423706650733948, "learning_rate": 0.0001, "elbo": 3.51808, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.320314332629604}
{"step": 73675, "loss": 3.509965515136719, "grad_norm": 0.695342481136322, "learning_rate": 0.0001, "elbo": 3.50997, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.320607553366174}
{"step": 73680, "loss": 3.5185821533203123, "grad_norm": 0.6809669733047485, "learning_rate": 0.0001, "elbo": 3.51858, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3209007741027445}
{"step": 73685, "loss": 3.510015106201172, "grad_norm": 0.7467294931411743, "learning_rate": 0.0001, "elbo": 3.51002, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.321193994839315}
{"step": 73690, "loss": 3.5277095794677735, "grad_norm": 0.8208884596824646, "learning_rate": 0.0001, "elbo": 3.52771, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.321487215575885}
{"step": 73695, "loss": 3.517439270019531, "grad_norm": 0.7124963998794556, "learning_rate": 0.0001, "elbo": 3.51744, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.321780436312456}
{"step": 73700, "loss": 3.4992912292480467, "grad_norm": 0.6783416867256165, "learning_rate": 0.0001, "elbo": 3.49929, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.322073657049026}
{"step": 73705, "loss": 3.5293312072753906, "grad_norm": 1.098924160003662, "learning_rate": 0.0001, "elbo": 3.52933, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.322366877785597}
{"step": 73710, "loss": 3.525833511352539, "grad_norm": 0.6778470873832703, "learning_rate": 0.0001, "elbo": 3.52583, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.322660098522167}
{"step": 73715, "loss": 3.540876007080078, "grad_norm": 0.7763432860374451, "learning_rate": 0.0001, "elbo": 3.54088, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.322953319258738}
{"step": 73720, "loss": 3.514551544189453, "grad_norm": 0.6926620602607727, "learning_rate": 0.0001, "elbo": 3.51455, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.323246539995308}
{"step": 73725, "loss": 3.517400360107422, "grad_norm": 0.877631664276123, "learning_rate": 0.0001, "elbo": 3.5174, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.323539760731879}
{"step": 73730, "loss": 3.5149234771728515, "grad_norm": 0.8951324820518494, "learning_rate": 0.0001, "elbo": 3.51492, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.323832981468449}
{"step": 73735, "loss": 3.524283599853516, "grad_norm": 0.7383452653884888, "learning_rate": 0.0001, "elbo": 3.52428, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3241262022050195}
{"step": 73740, "loss": 3.529235076904297, "grad_norm": 0.6828734874725342, "learning_rate": 0.0001, "elbo": 3.52924, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.32441942294159}
{"step": 73745, "loss": 3.512920379638672, "grad_norm": 0.7703872323036194, "learning_rate": 0.0001, "elbo": 3.51292, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.324712643678161}
{"step": 73750, "loss": 3.519413375854492, "grad_norm": 0.7035225629806519, "learning_rate": 0.0001, "elbo": 3.51941, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.325005864414732}
{"step": 73755, "loss": 3.52357177734375, "grad_norm": 0.6935181617736816, "learning_rate": 0.0001, "elbo": 3.52357, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.325299085151302}
{"step": 73760, "loss": 3.501195526123047, "grad_norm": 0.7225706577301025, "learning_rate": 0.0001, "elbo": 3.5012, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.325592305887873}
{"step": 73765, "loss": 3.5178451538085938, "grad_norm": 0.732409656047821, "learning_rate": 0.0001, "elbo": 3.51785, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.325885526624443}
{"step": 73770, "loss": 3.528849792480469, "grad_norm": 0.7039245367050171, "learning_rate": 0.0001, "elbo": 3.52885, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.326178747361014}
{"step": 73775, "loss": 3.496469497680664, "grad_norm": 0.7045339941978455, "learning_rate": 0.0001, "elbo": 3.49647, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.326471968097584}
{"step": 73780, "loss": 3.5237266540527346, "grad_norm": 0.7356504201889038, "learning_rate": 0.0001, "elbo": 3.52373, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3267651888341545}
{"step": 73785, "loss": 3.5097110748291014, "grad_norm": 0.7361985445022583, "learning_rate": 0.0001, "elbo": 3.50971, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.327058409570725}
{"step": 73790, "loss": 3.5219051361083986, "grad_norm": 0.7187505960464478, "learning_rate": 0.0001, "elbo": 3.5219, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.327351630307295}
{"step": 73795, "loss": 3.5213985443115234, "grad_norm": 0.6877291202545166, "learning_rate": 0.0001, "elbo": 3.5214, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.327644851043866}
{"step": 73800, "loss": 3.5099315643310547, "grad_norm": 0.6821111440658569, "learning_rate": 0.0001, "elbo": 3.50993, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.327938071780436}
{"step": 73805, "loss": 3.515760803222656, "grad_norm": 0.7212586402893066, "learning_rate": 0.0001, "elbo": 3.51576, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.328231292517007}
{"step": 73810, "loss": 3.527478790283203, "grad_norm": 0.7169346213340759, "learning_rate": 0.0001, "elbo": 3.52748, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.328524513253577}
{"step": 73815, "loss": 3.5117626190185547, "grad_norm": 0.898028552532196, "learning_rate": 0.0001, "elbo": 3.51176, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.328817733990148}
{"step": 73820, "loss": 3.5188838958740236, "grad_norm": 0.7197499871253967, "learning_rate": 0.0001, "elbo": 3.51888, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.329110954726718}
{"step": 73825, "loss": 3.5147064208984373, "grad_norm": 0.6942550539970398, "learning_rate": 0.0001, "elbo": 3.51471, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.329404175463289}
{"step": 73830, "loss": 3.522246551513672, "grad_norm": 0.8307819366455078, "learning_rate": 0.0001, "elbo": 3.52225, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.329697396199859}
{"step": 73835, "loss": 3.5022811889648438, "grad_norm": 0.7624865174293518, "learning_rate": 0.0001, "elbo": 3.50228, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3299906169364295}
{"step": 73835, "loss": 3.5022811889648438, "grad_norm": 0.7624865174293518, "learning_rate": 0.0001, "elbo": 3.50228, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3299906169364295}
{"step": 73840, "loss": 3.5258247375488283, "grad_norm": 1.0433340072631836, "learning_rate": 0.0001, "elbo": 3.52582, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.330283837673}
{"step": 73845, "loss": 3.512450408935547, "grad_norm": 0.7734172940254211, "learning_rate": 0.0001, "elbo": 3.51245, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.33057705840957}
{"step": 73850, "loss": 3.5219585418701174, "grad_norm": 0.7336252331733704, "learning_rate": 0.0001, "elbo": 3.52196, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.330870279146141}
{"step": 73855, "loss": 3.5257858276367187, "grad_norm": 0.6911879181861877, "learning_rate": 0.0001, "elbo": 3.52579, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.331163499882711}
{"step": 73860, "loss": 3.521292877197266, "grad_norm": 0.7274250984191895, "learning_rate": 0.0001, "elbo": 3.52129, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.331456720619283}
{"step": 73865, "loss": 3.518976593017578, "grad_norm": 0.6871824264526367, "learning_rate": 0.0001, "elbo": 3.51898, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.331749941355852}
{"step": 73870, "loss": 3.5040870666503907, "grad_norm": 0.8704083561897278, "learning_rate": 0.0001, "elbo": 3.50409, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.332043162092424}
{"step": 73875, "loss": 3.528656768798828, "grad_norm": 0.6803309321403503, "learning_rate": 0.0001, "elbo": 3.52866, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.332336382828994}
{"step": 73880, "loss": 3.5144290924072266, "grad_norm": 0.7422829866409302, "learning_rate": 0.0001, "elbo": 3.51443, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3326296035655645}
{"step": 73885, "loss": 3.5115711212158205, "grad_norm": 0.675101637840271, "learning_rate": 0.0001, "elbo": 3.51157, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.332922824302135}
{"step": 73890, "loss": 3.5078163146972656, "grad_norm": 0.8669534921646118, "learning_rate": 0.0001, "elbo": 3.50782, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.333216045038705}
{"step": 73895, "loss": 3.514810562133789, "grad_norm": 0.7007149457931519, "learning_rate": 0.0001, "elbo": 3.51481, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.333509265775276}
{"step": 73900, "loss": 3.527099609375, "grad_norm": 0.7145814299583435, "learning_rate": 0.0001, "elbo": 3.5271, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.333802486511846}
{"step": 73905, "loss": 3.496781921386719, "grad_norm": 0.7103971838951111, "learning_rate": 0.0001, "elbo": 3.49678, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.334095707248417}
{"step": 73910, "loss": 3.5108436584472655, "grad_norm": 0.7418954968452454, "learning_rate": 0.0001, "elbo": 3.51084, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.334388927984987}
{"step": 73915, "loss": 3.5226707458496094, "grad_norm": 0.7180626392364502, "learning_rate": 0.0001, "elbo": 3.52267, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.334682148721558}
{"step": 73920, "loss": 3.5319801330566407, "grad_norm": 0.7566553950309753, "learning_rate": 0.0001, "elbo": 3.53198, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.334975369458128}
{"step": 73925, "loss": 3.517591094970703, "grad_norm": 0.6731072664260864, "learning_rate": 0.0001, "elbo": 3.51759, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.335268590194699}
{"step": 73930, "loss": 3.512103271484375, "grad_norm": 0.6704660058021545, "learning_rate": 0.0001, "elbo": 3.5121, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.335561810931269}
{"step": 73935, "loss": 3.509991455078125, "grad_norm": 0.6952834129333496, "learning_rate": 0.0001, "elbo": 3.50999, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3358550316678395}
{"step": 73940, "loss": 3.509877014160156, "grad_norm": 0.6985257267951965, "learning_rate": 0.0001, "elbo": 3.50988, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.33614825240441}
{"step": 73945, "loss": 3.5139228820800783, "grad_norm": 0.7063623070716858, "learning_rate": 0.0001, "elbo": 3.51392, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.33644147314098}
{"step": 73950, "loss": 3.5151927947998045, "grad_norm": 0.7078595161437988, "learning_rate": 0.0001, "elbo": 3.51519, "mask_frac": 0.5013, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.336734693877551}
{"step": 73955, "loss": 3.520699310302734, "grad_norm": 0.692660391330719, "learning_rate": 0.0001, "elbo": 3.5207, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.337027914614121}
{"step": 73960, "loss": 3.5339935302734373, "grad_norm": 0.8105550408363342, "learning_rate": 0.0001, "elbo": 3.53399, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.337321135350692}
{"step": 73965, "loss": 3.50756950378418, "grad_norm": 0.6696321368217468, "learning_rate": 0.0001, "elbo": 3.50757, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.337614356087262}
{"step": 73970, "loss": 3.5005722045898438, "grad_norm": 0.8001870512962341, "learning_rate": 0.0001, "elbo": 3.50057, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.337907576823833}
{"step": 73975, "loss": 3.517705535888672, "grad_norm": 0.7803751826286316, "learning_rate": 0.0001, "elbo": 3.51771, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.338200797560403}
{"step": 73980, "loss": 3.521464157104492, "grad_norm": 0.7183602452278137, "learning_rate": 0.0001, "elbo": 3.52146, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.338494018296974}
{"step": 73985, "loss": 3.52221565246582, "grad_norm": 0.7127934098243713, "learning_rate": 0.0001, "elbo": 3.52222, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.338787239033545}
{"step": 73990, "loss": 3.5228908538818358, "grad_norm": 0.713833212852478, "learning_rate": 0.0001, "elbo": 3.52289, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3390804597701145}
{"step": 73995, "loss": 3.5217689514160155, "grad_norm": 0.8920149207115173, "learning_rate": 0.0001, "elbo": 3.52177, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.339373680506686}
{"step": 74000, "loss": 3.519337844848633, "grad_norm": 0.7173039317131042, "learning_rate": 0.0001, "elbo": 3.51934, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.339666901243256}
{"step": 74005, "loss": 3.517609786987305, "grad_norm": 0.7091137170791626, "learning_rate": 0.0001, "elbo": 3.51761, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.339960121979827}
{"step": 74010, "loss": 3.5086318969726564, "grad_norm": 0.6787360310554504, "learning_rate": 0.0001, "elbo": 3.50863, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.340253342716397}
{"step": 74015, "loss": 3.5182449340820314, "grad_norm": 1.2542742490768433, "learning_rate": 0.0001, "elbo": 3.51824, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.340546563452968}
{"step": 74020, "loss": 3.5333282470703127, "grad_norm": 0.6785882711410522, "learning_rate": 0.0001, "elbo": 3.53333, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.340839784189538}
{"step": 74025, "loss": 3.5157665252685546, "grad_norm": 0.8296142220497131, "learning_rate": 0.0001, "elbo": 3.51577, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3411330049261085}
{"step": 74030, "loss": 3.522001266479492, "grad_norm": 0.7173623442649841, "learning_rate": 0.0001, "elbo": 3.522, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.341426225662679}
{"step": 74035, "loss": 3.5260257720947266, "grad_norm": 0.6633121967315674, "learning_rate": 0.0001, "elbo": 3.52603, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3417194463992494}
{"step": 74040, "loss": 3.521596145629883, "grad_norm": 0.6745756268501282, "learning_rate": 0.0001, "elbo": 3.5216, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.34201266713582}
{"step": 74045, "loss": 3.501703643798828, "grad_norm": 0.7026430368423462, "learning_rate": 0.0001, "elbo": 3.5017, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.34230588787239}
{"step": 74050, "loss": 3.522161102294922, "grad_norm": 0.7243507504463196, "learning_rate": 0.0001, "elbo": 3.52216, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.342599108608961}
{"step": 74055, "loss": 3.5073089599609375, "grad_norm": 0.702130913734436, "learning_rate": 0.0001, "elbo": 3.50731, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.342892329345531}
{"step": 74060, "loss": 3.500285339355469, "grad_norm": 1.0025842189788818, "learning_rate": 0.0001, "elbo": 3.50029, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.343185550082102}
{"step": 74065, "loss": 3.5279777526855467, "grad_norm": 0.6856529116630554, "learning_rate": 0.0001, "elbo": 3.52798, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.343478770818672}
{"step": 74070, "loss": 3.519354248046875, "grad_norm": 0.6824727058410645, "learning_rate": 0.0001, "elbo": 3.51935, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.343771991555243}
{"step": 74075, "loss": 3.5059585571289062, "grad_norm": 0.7139407396316528, "learning_rate": 0.0001, "elbo": 3.50596, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.344065212291813}
{"step": 74080, "loss": 3.5069808959960938, "grad_norm": 0.6957073211669922, "learning_rate": 0.0001, "elbo": 3.50698, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3443584330283835}
{"step": 74085, "loss": 3.5191253662109374, "grad_norm": 0.8436151146888733, "learning_rate": 0.0001, "elbo": 3.51913, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.344651653764954}
{"step": 74090, "loss": 3.5364761352539062, "grad_norm": 0.7753088474273682, "learning_rate": 0.0001, "elbo": 3.53648, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3449448745015244}
{"step": 74095, "loss": 3.511100006103516, "grad_norm": 0.7030577063560486, "learning_rate": 0.0001, "elbo": 3.5111, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.345238095238095}
{"step": 74100, "loss": 3.519174575805664, "grad_norm": 0.7119950652122498, "learning_rate": 0.0001, "elbo": 3.51917, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.345531315974665}
{"step": 74105, "loss": 3.504359817504883, "grad_norm": 0.7285299301147461, "learning_rate": 0.0001, "elbo": 3.50436, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.345824536711236}
{"step": 74110, "loss": 3.5321884155273438, "grad_norm": 0.7636001706123352, "learning_rate": 0.0001, "elbo": 3.53219, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.346117757447807}
{"step": 74115, "loss": 3.5077075958251953, "grad_norm": 0.7181294560432434, "learning_rate": 0.0001, "elbo": 3.50771, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.346410978184378}
{"step": 74120, "loss": 3.5112350463867186, "grad_norm": 0.7224797010421753, "learning_rate": 0.0001, "elbo": 3.51124, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.346704198920948}
{"step": 74125, "loss": 3.5275665283203126, "grad_norm": 0.6875473856925964, "learning_rate": 0.0001, "elbo": 3.52757, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3469974196575185}
{"step": 74130, "loss": 3.505736541748047, "grad_norm": 0.712234377861023, "learning_rate": 0.0001, "elbo": 3.50574, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.347290640394089}
{"step": 74135, "loss": 3.50439453125, "grad_norm": 0.6903103590011597, "learning_rate": 0.0001, "elbo": 3.50439, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.347583861130659}
{"step": 74140, "loss": 3.5173976898193358, "grad_norm": 0.6883595585823059, "learning_rate": 0.0001, "elbo": 3.5174, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.34787708186723}
{"step": 74145, "loss": 3.5081382751464845, "grad_norm": 0.7914496064186096, "learning_rate": 0.0001, "elbo": 3.50814, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3481703026038}
{"step": 74150, "loss": 3.4933040618896483, "grad_norm": 0.874704122543335, "learning_rate": 0.0001, "elbo": 3.4933, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.348463523340371}
{"step": 74155, "loss": 3.5138313293457033, "grad_norm": 0.7872418165206909, "learning_rate": 0.0001, "elbo": 3.51383, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.348756744076941}
{"step": 74160, "loss": 3.5263999938964843, "grad_norm": 0.8283246159553528, "learning_rate": 0.0001, "elbo": 3.5264, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.349049964813512}
{"step": 74165, "loss": 3.503742218017578, "grad_norm": 0.7452386021614075, "learning_rate": 0.0001, "elbo": 3.50374, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.349343185550082}
{"step": 74170, "loss": 3.518752670288086, "grad_norm": 0.7177326083183289, "learning_rate": 0.0001, "elbo": 3.51875, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.349636406286653}
{"step": 74170, "loss": 3.518752670288086, "grad_norm": 0.7177326083183289, "learning_rate": 0.0001, "elbo": 3.51875, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.349636406286653}
{"step": 74175, "loss": 3.5184089660644533, "grad_norm": 0.7496650218963623, "learning_rate": 0.0001, "elbo": 3.51841, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.349929627023223}
{"step": 74180, "loss": 3.5079181671142576, "grad_norm": 0.7151774764060974, "learning_rate": 0.0001, "elbo": 3.50792, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3502228477597935}
{"step": 74185, "loss": 3.496441650390625, "grad_norm": 0.7144696712493896, "learning_rate": 0.0001, "elbo": 3.49644, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.350516068496364}
{"step": 74190, "loss": 3.5103721618652344, "grad_norm": 0.7213782072067261, "learning_rate": 0.0001, "elbo": 3.51037, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.350809289232934}
{"step": 74195, "loss": 3.4992649078369142, "grad_norm": 0.7874227166175842, "learning_rate": 0.0001, "elbo": 3.49926, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.351102509969505}
{"step": 74200, "loss": 3.5187046051025392, "grad_norm": 0.6928166151046753, "learning_rate": 0.0001, "elbo": 3.5187, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.351395730706075}
{"step": 74205, "loss": 3.521279525756836, "grad_norm": 0.7456478476524353, "learning_rate": 0.0001, "elbo": 3.52128, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.351688951442646}
{"step": 74210, "loss": 3.5096588134765625, "grad_norm": 0.7436798810958862, "learning_rate": 0.0001, "elbo": 3.50966, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.351982172179216}
{"step": 74215, "loss": 3.521657943725586, "grad_norm": 0.7362247705459595, "learning_rate": 0.0001, "elbo": 3.52166, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.352275392915787}
{"step": 74220, "loss": 3.518876647949219, "grad_norm": 0.685420036315918, "learning_rate": 0.0001, "elbo": 3.51888, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.352568613652357}
{"step": 74225, "loss": 3.5100017547607423, "grad_norm": 0.7168280482292175, "learning_rate": 0.0001, "elbo": 3.51, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.352861834388928}
{"step": 74230, "loss": 3.513530731201172, "grad_norm": 0.7843111157417297, "learning_rate": 0.0001, "elbo": 3.51353, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.353155055125498}
{"step": 74235, "loss": 3.506877136230469, "grad_norm": 0.7052873373031616, "learning_rate": 0.0001, "elbo": 3.50688, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.353448275862069}
{"step": 74240, "loss": 3.5066009521484376, "grad_norm": 0.7826579809188843, "learning_rate": 0.0001, "elbo": 3.5066, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.35374149659864}
{"step": 74245, "loss": 3.511290740966797, "grad_norm": 0.772649884223938, "learning_rate": 0.0001, "elbo": 3.51129, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.35403471733521}
{"step": 74250, "loss": 3.5238338470458985, "grad_norm": 0.7006256580352783, "learning_rate": 0.0001, "elbo": 3.52383, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.354327938071781}
{"step": 74255, "loss": 3.525246429443359, "grad_norm": 0.7023031115531921, "learning_rate": 0.0001, "elbo": 3.52525, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.354621158808351}
{"step": 74260, "loss": 3.518895721435547, "grad_norm": 0.7229139804840088, "learning_rate": 0.0001, "elbo": 3.5189, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.354914379544922}
{"step": 74260, "loss": 3.518895721435547, "grad_norm": 0.7229139804840088, "learning_rate": 0.0001, "elbo": 3.5189, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.354914379544922}
{"step": 74265, "loss": 3.5151351928710937, "grad_norm": 0.6994923949241638, "learning_rate": 0.0001, "elbo": 3.51513, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.355207600281492}
{"step": 74270, "loss": 3.5231651306152343, "grad_norm": 0.7365462779998779, "learning_rate": 0.0001, "elbo": 3.52316, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.355500821018063}
{"step": 74275, "loss": 3.522654724121094, "grad_norm": 0.7275243401527405, "learning_rate": 0.0001, "elbo": 3.52265, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.355794041754633}
{"step": 74280, "loss": 3.5175197601318358, "grad_norm": 0.7698432803153992, "learning_rate": 0.0001, "elbo": 3.51752, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3560872624912035}
{"step": 74285, "loss": 3.5175926208496096, "grad_norm": 0.7340586185455322, "learning_rate": 0.0001, "elbo": 3.51759, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.356380483227774}
{"step": 74290, "loss": 3.5122093200683593, "grad_norm": 0.7039063572883606, "learning_rate": 0.0001, "elbo": 3.51221, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.356673703964344}
{"step": 74295, "loss": 3.5261383056640625, "grad_norm": 0.7099990248680115, "learning_rate": 0.0001, "elbo": 3.52614, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.356966924700915}
{"step": 74300, "loss": 3.519398498535156, "grad_norm": 0.7139229774475098, "learning_rate": 0.0001, "elbo": 3.5194, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.357260145437485}
{"step": 74305, "loss": 3.5255313873291017, "grad_norm": 0.7049698233604431, "learning_rate": 0.0001, "elbo": 3.52553, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.357553366174056}
{"step": 74310, "loss": 3.508129119873047, "grad_norm": 0.6778505444526672, "learning_rate": 0.0001, "elbo": 3.50813, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.357846586910626}
{"step": 74315, "loss": 3.5108047485351563, "grad_norm": 0.7385414242744446, "learning_rate": 0.0001, "elbo": 3.5108, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.358139807647197}
{"step": 74320, "loss": 3.4999141693115234, "grad_norm": 0.6826586127281189, "learning_rate": 0.0001, "elbo": 3.49991, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.358433028383767}
{"step": 74325, "loss": 3.5216522216796875, "grad_norm": 0.7732667922973633, "learning_rate": 0.0001, "elbo": 3.52165, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.358726249120338}
{"step": 74330, "loss": 3.518983840942383, "grad_norm": 0.6636443734169006, "learning_rate": 0.0001, "elbo": 3.51898, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.359019469856908}
{"step": 74335, "loss": 3.500414276123047, "grad_norm": 0.7488798499107361, "learning_rate": 0.0001, "elbo": 3.50041, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3593126905934785}
{"step": 74340, "loss": 3.510939025878906, "grad_norm": 0.7295804619789124, "learning_rate": 0.0001, "elbo": 3.51094, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.359605911330049}
{"step": 74345, "loss": 3.503228759765625, "grad_norm": 0.7333144545555115, "learning_rate": 0.0001, "elbo": 3.50323, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.359899132066619}
{"step": 74350, "loss": 3.5177555084228516, "grad_norm": 0.7000775337219238, "learning_rate": 0.0001, "elbo": 3.51776, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.36019235280319}
{"step": 74355, "loss": 3.5218185424804687, "grad_norm": 0.7153839468955994, "learning_rate": 0.0001, "elbo": 3.52182, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.36048557353976}
{"step": 74360, "loss": 3.5144287109375, "grad_norm": 0.7554564476013184, "learning_rate": 0.0001, "elbo": 3.51443, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.360778794276332}
{"step": 74365, "loss": 3.514950180053711, "grad_norm": 0.8780285716056824, "learning_rate": 0.0001, "elbo": 3.51495, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.361072015012902}
{"step": 74365, "loss": 3.514950180053711, "grad_norm": 0.8780285716056824, "learning_rate": 0.0001, "elbo": 3.51495, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.361072015012902}
{"step": 74370, "loss": 3.5080184936523438, "grad_norm": 0.8296698331832886, "learning_rate": 0.0001, "elbo": 3.50802, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3613652357494725}
{"step": 74375, "loss": 3.5138111114501953, "grad_norm": 0.8866240978240967, "learning_rate": 0.0001, "elbo": 3.51381, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.361658456486043}
{"step": 74380, "loss": 3.5090904235839844, "grad_norm": 0.7003546357154846, "learning_rate": 0.0001, "elbo": 3.50909, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3619516772226135}
{"step": 74385, "loss": 3.522126388549805, "grad_norm": 0.7049694657325745, "learning_rate": 0.0001, "elbo": 3.52213, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.362244897959184}
{"step": 74390, "loss": 3.505146789550781, "grad_norm": 0.6922836303710938, "learning_rate": 0.0001, "elbo": 3.50515, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.362538118695754}
{"step": 74395, "loss": 3.5161033630371095, "grad_norm": 0.727287769317627, "learning_rate": 0.0001, "elbo": 3.5161, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.362831339432325}
{"step": 74400, "loss": 3.5113300323486327, "grad_norm": 0.6945982575416565, "learning_rate": 0.0001, "elbo": 3.51133, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.363124560168895}
{"step": 74405, "loss": 3.5225486755371094, "grad_norm": 0.6643656492233276, "learning_rate": 0.0001, "elbo": 3.52255, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.363417780905466}
{"step": 74410, "loss": 3.520630645751953, "grad_norm": 0.7098678946495056, "learning_rate": 0.0001, "elbo": 3.52063, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.363711001642036}
{"step": 74415, "loss": 3.512654113769531, "grad_norm": 1.0045771598815918, "learning_rate": 0.0001, "elbo": 3.51265, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.364004222378607}
{"step": 74420, "loss": 3.517325592041016, "grad_norm": 0.8152011632919312, "learning_rate": 0.0001, "elbo": 3.51733, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.364297443115177}
{"step": 74425, "loss": 3.5128814697265627, "grad_norm": 0.7142595648765564, "learning_rate": 0.0001, "elbo": 3.51288, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3645906638517475}
{"step": 74430, "loss": 3.5115306854248045, "grad_norm": 0.7140349745750427, "learning_rate": 0.0001, "elbo": 3.51153, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.364883884588318}
{"step": 74435, "loss": 3.504139709472656, "grad_norm": 0.7749380469322205, "learning_rate": 0.0001, "elbo": 3.50414, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3651771053248885}
{"step": 74440, "loss": 3.512828063964844, "grad_norm": 0.8214526176452637, "learning_rate": 0.0001, "elbo": 3.51283, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.365470326061459}
{"step": 74445, "loss": 3.521392059326172, "grad_norm": 0.74159836769104, "learning_rate": 0.0001, "elbo": 3.52139, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.365763546798029}
{"step": 74450, "loss": 3.5206974029541014, "grad_norm": 0.7598081231117249, "learning_rate": 0.0001, "elbo": 3.5207, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3660567675346}
{"step": 74455, "loss": 3.522045135498047, "grad_norm": 0.7496464252471924, "learning_rate": 0.0001, "elbo": 3.52205, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.36634998827117}
{"step": 74460, "loss": 3.498320770263672, "grad_norm": 0.6686844825744629, "learning_rate": 0.0001, "elbo": 3.49832, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.366643209007741}
{"step": 74465, "loss": 3.514556884765625, "grad_norm": 0.8786264657974243, "learning_rate": 0.0001, "elbo": 3.51456, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.366936429744311}
{"step": 74470, "loss": 3.489635467529297, "grad_norm": 0.762844979763031, "learning_rate": 0.0001, "elbo": 3.48964, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.367229650480882}
{"step": 74475, "loss": 3.5138072967529297, "grad_norm": 0.7094547748565674, "learning_rate": 0.0001, "elbo": 3.51381, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.367522871217452}
{"step": 74480, "loss": 3.5170852661132814, "grad_norm": 0.7123326659202576, "learning_rate": 0.0001, "elbo": 3.51709, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3678160919540225}
{"step": 74480, "loss": 3.5170852661132814, "grad_norm": 0.7123326659202576, "learning_rate": 0.0001, "elbo": 3.51709, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3678160919540225}
{"step": 74485, "loss": 3.5054359436035156, "grad_norm": 0.8942315578460693, "learning_rate": 0.0001, "elbo": 3.50544, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.368109312690594}
{"step": 74490, "loss": 3.5036651611328127, "grad_norm": 0.8617202639579773, "learning_rate": 0.0001, "elbo": 3.50366, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.368402533427164}
{"step": 74490, "loss": 3.5036651611328127, "grad_norm": 0.8617202639579773, "learning_rate": 0.0001, "elbo": 3.50366, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.368402533427164}
{"step": 74495, "loss": 3.5135986328125, "grad_norm": 0.6744463443756104, "learning_rate": 0.0001, "elbo": 3.5136, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.368695754163735}
{"step": 74500, "loss": 3.5183258056640625, "grad_norm": 0.6921822428703308, "learning_rate": 0.0001, "elbo": 3.51833, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.368988974900305}
{"step": 74505, "loss": 3.5121337890625, "grad_norm": 0.6866913437843323, "learning_rate": 0.0001, "elbo": 3.51213, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.369282195636876}
{"step": 74510, "loss": 3.5136795043945312, "grad_norm": 0.7647704482078552, "learning_rate": 0.0001, "elbo": 3.51368, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.369575416373446}
{"step": 74515, "loss": 3.5161602020263674, "grad_norm": 0.7096232175827026, "learning_rate": 0.0001, "elbo": 3.51616, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.369868637110017}
{"step": 74520, "loss": 3.5048484802246094, "grad_norm": 0.738980770111084, "learning_rate": 0.0001, "elbo": 3.50485, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.370161857846587}
{"step": 74525, "loss": 3.5256481170654297, "grad_norm": 0.7140520215034485, "learning_rate": 0.0001, "elbo": 3.52565, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3704550785831575}
{"step": 74530, "loss": 3.5289688110351562, "grad_norm": 0.7245184183120728, "learning_rate": 0.0001, "elbo": 3.52897, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.370748299319728}
{"step": 74535, "loss": 3.519910430908203, "grad_norm": 0.7273339629173279, "learning_rate": 0.0001, "elbo": 3.51991, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.371041520056298}
{"step": 74540, "loss": 3.5171695709228517, "grad_norm": 0.7887666821479797, "learning_rate": 0.0001, "elbo": 3.51717, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.371334740792869}
{"step": 74545, "loss": 3.5057914733886717, "grad_norm": 0.7493743896484375, "learning_rate": 0.0001, "elbo": 3.50579, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.371627961529439}
{"step": 74550, "loss": 3.515140914916992, "grad_norm": 0.9297628402709961, "learning_rate": 0.0001, "elbo": 3.51514, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.37192118226601}
{"step": 74555, "loss": 3.5098594665527343, "grad_norm": 0.7298381328582764, "learning_rate": 0.0001, "elbo": 3.50986, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.37221440300258}
{"step": 74560, "loss": 3.506435012817383, "grad_norm": 0.7689158916473389, "learning_rate": 0.0001, "elbo": 3.50644, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.372507623739151}
{"step": 74565, "loss": 3.51905517578125, "grad_norm": 1.2872709035873413, "learning_rate": 0.0001, "elbo": 3.51906, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.372800844475721}
{"step": 74570, "loss": 3.522594451904297, "grad_norm": 0.6903151273727417, "learning_rate": 0.0001, "elbo": 3.52259, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.373094065212292}
{"step": 74575, "loss": 3.5060165405273436, "grad_norm": 0.7325198650360107, "learning_rate": 0.0001, "elbo": 3.50602, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.373387285948862}
{"step": 74580, "loss": 3.529595947265625, "grad_norm": 0.6761026978492737, "learning_rate": 0.0001, "elbo": 3.5296, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3736805066854325}
{"step": 74585, "loss": 3.511300277709961, "grad_norm": 0.7042760252952576, "learning_rate": 0.0001, "elbo": 3.5113, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.373973727422003}
{"step": 74590, "loss": 3.5163516998291016, "grad_norm": 0.7717597484588623, "learning_rate": 0.0001, "elbo": 3.51635, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.374266948158573}
{"step": 74595, "loss": 3.507102203369141, "grad_norm": 0.7638332843780518, "learning_rate": 0.0001, "elbo": 3.5071, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.374560168895144}
{"step": 74600, "loss": 3.5364288330078124, "grad_norm": 0.7754348516464233, "learning_rate": 0.0001, "elbo": 3.53643, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.374853389631715}
{"step": 74605, "loss": 3.4941612243652345, "grad_norm": 0.6847243905067444, "learning_rate": 0.0001, "elbo": 3.49416, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.375146610368285}
{"step": 74610, "loss": 3.511846160888672, "grad_norm": 0.7150566577911377, "learning_rate": 0.0001, "elbo": 3.51185, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.375439831104856}
{"step": 74615, "loss": 3.517837905883789, "grad_norm": 0.8553587198257446, "learning_rate": 0.0001, "elbo": 3.51784, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.375733051841427}
{"step": 74615, "loss": 3.517837905883789, "grad_norm": 0.8553587198257446, "learning_rate": 0.0001, "elbo": 3.51784, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.375733051841427}
{"step": 74620, "loss": 3.524730682373047, "grad_norm": 1.2122223377227783, "learning_rate": 0.0001, "elbo": 3.52473, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.376026272577997}
{"step": 74625, "loss": 3.5255836486816405, "grad_norm": 0.7002572417259216, "learning_rate": 0.0001, "elbo": 3.52558, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3763194933145675}
{"step": 74630, "loss": 3.516193389892578, "grad_norm": 0.7060086131095886, "learning_rate": 0.0001, "elbo": 3.51619, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.376612714051138}
{"step": 74635, "loss": 3.525225067138672, "grad_norm": 0.7272665500640869, "learning_rate": 0.0001, "elbo": 3.52523, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.376905934787708}
{"step": 74640, "loss": 3.525065231323242, "grad_norm": 0.7261706590652466, "learning_rate": 0.0001, "elbo": 3.52507, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.377199155524279}
{"step": 74645, "loss": 3.5281646728515623, "grad_norm": 0.7236299514770508, "learning_rate": 0.0001, "elbo": 3.52816, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.377492376260849}
{"step": 74650, "loss": 3.537789535522461, "grad_norm": 0.8408378958702087, "learning_rate": 0.0001, "elbo": 3.53779, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.37778559699742}
{"step": 74655, "loss": 3.5268890380859377, "grad_norm": 0.7231740355491638, "learning_rate": 0.0001, "elbo": 3.52689, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.37807881773399}
{"step": 74660, "loss": 3.519824981689453, "grad_norm": 0.7629920840263367, "learning_rate": 0.0001, "elbo": 3.51983, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.378372038470561}
{"step": 74665, "loss": 3.5128662109375, "grad_norm": 0.6895455718040466, "learning_rate": 0.0001, "elbo": 3.51287, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.378665259207131}
{"step": 74670, "loss": 3.5089138031005858, "grad_norm": 0.7215750217437744, "learning_rate": 0.0001, "elbo": 3.50891, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.378958479943702}
{"step": 74675, "loss": 3.5092960357666017, "grad_norm": 0.6931756138801575, "learning_rate": 0.0001, "elbo": 3.5093, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.379251700680272}
{"step": 74680, "loss": 3.513050079345703, "grad_norm": 0.7039379477500916, "learning_rate": 0.0001, "elbo": 3.51305, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3795449214168425}
{"step": 74685, "loss": 3.51669921875, "grad_norm": 0.7293657064437866, "learning_rate": 0.0001, "elbo": 3.5167, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.379838142153413}
{"step": 74690, "loss": 3.508961486816406, "grad_norm": 0.7611442804336548, "learning_rate": 0.0001, "elbo": 3.50896, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.380131362889983}
{"step": 74695, "loss": 3.538965606689453, "grad_norm": 0.7379270195960999, "learning_rate": 0.0001, "elbo": 3.53897, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.380424583626554}
{"step": 74700, "loss": 3.506114196777344, "grad_norm": 0.7818901538848877, "learning_rate": 0.0001, "elbo": 3.50611, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.380717804363124}
{"step": 74705, "loss": 3.5294868469238283, "grad_norm": 0.7146034836769104, "learning_rate": 0.0001, "elbo": 3.52949, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.381011025099695}
{"step": 74710, "loss": 3.5196884155273436, "grad_norm": 0.7012056112289429, "learning_rate": 0.0001, "elbo": 3.51969, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.381304245836265}
{"step": 74715, "loss": 3.4976600646972655, "grad_norm": 0.7079399824142456, "learning_rate": 0.0001, "elbo": 3.49766, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.381597466572836}
{"step": 74720, "loss": 3.519415283203125, "grad_norm": 0.6950187683105469, "learning_rate": 0.0001, "elbo": 3.51942, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.381890687309406}
{"step": 74725, "loss": 3.4991077423095702, "grad_norm": 1.1170698404312134, "learning_rate": 0.0001, "elbo": 3.49911, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3821839080459775}
{"step": 74730, "loss": 3.5276870727539062, "grad_norm": 0.7051054835319519, "learning_rate": 0.0001, "elbo": 3.52769, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.382477128782548}
{"step": 74735, "loss": 3.5115676879882813, "grad_norm": 0.7420344352722168, "learning_rate": 0.0001, "elbo": 3.51157, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.382770349519118}
{"step": 74740, "loss": 3.5226982116699217, "grad_norm": 0.7760487198829651, "learning_rate": 0.0001, "elbo": 3.5227, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.383063570255689}
{"step": 74745, "loss": 3.514053726196289, "grad_norm": 0.7890843152999878, "learning_rate": 0.0001, "elbo": 3.51405, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.383356790992259}
{"step": 74750, "loss": 3.504896545410156, "grad_norm": 0.721508800983429, "learning_rate": 0.0001, "elbo": 3.5049, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.38365001172883}
{"step": 74755, "loss": 3.521681213378906, "grad_norm": 0.7001185417175293, "learning_rate": 0.0001, "elbo": 3.52168, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3839432324654}
{"step": 74760, "loss": 3.492368698120117, "grad_norm": 0.8032774329185486, "learning_rate": 0.0001, "elbo": 3.49237, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.384236453201971}
{"step": 74765, "loss": 3.500990295410156, "grad_norm": 0.7019054293632507, "learning_rate": 0.0001, "elbo": 3.50099, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.384529673938541}
{"step": 74770, "loss": 3.5049819946289062, "grad_norm": 0.7828423976898193, "learning_rate": 0.0001, "elbo": 3.50498, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3848228946751115}
{"step": 74775, "loss": 3.5217670440673827, "grad_norm": 0.6938155293464661, "learning_rate": 0.0001, "elbo": 3.52177, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.385116115411682}
{"step": 74780, "loss": 3.505071258544922, "grad_norm": 0.7110802531242371, "learning_rate": 0.0001, "elbo": 3.50507, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3854093361482525}
{"step": 74785, "loss": 3.512290191650391, "grad_norm": 0.747810423374176, "learning_rate": 0.0001, "elbo": 3.51229, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.385702556884823}
{"step": 74790, "loss": 3.5086273193359374, "grad_norm": 0.6626715660095215, "learning_rate": 0.0001, "elbo": 3.50863, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.385995777621393}
{"step": 74795, "loss": 3.502019500732422, "grad_norm": 0.6867491602897644, "learning_rate": 0.0001, "elbo": 3.50202, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.386288998357964}
{"step": 74800, "loss": 3.5207187652587892, "grad_norm": 1.275887370109558, "learning_rate": 0.0001, "elbo": 3.52072, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.386582219094534}
{"step": 74800, "loss": 3.5207187652587892, "grad_norm": 1.275887370109558, "learning_rate": 0.0001, "elbo": 3.52072, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.386582219094534}
{"step": 74805, "loss": 3.5035667419433594, "grad_norm": 0.692326009273529, "learning_rate": 0.0001, "elbo": 3.50357, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.386875439831105}
{"step": 74810, "loss": 3.515423583984375, "grad_norm": 0.7377825975418091, "learning_rate": 0.0001, "elbo": 3.51542, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.387168660567675}
{"step": 74810, "loss": 3.515423583984375, "grad_norm": 0.7377825975418091, "learning_rate": 0.0001, "elbo": 3.51542, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.387168660567675}
{"step": 74815, "loss": 3.5044883728027343, "grad_norm": 0.7461667060852051, "learning_rate": 0.0001, "elbo": 3.50449, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.387461881304246}
{"step": 74820, "loss": 3.519203948974609, "grad_norm": 0.7136273384094238, "learning_rate": 0.0001, "elbo": 3.5192, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.387755102040816}
{"step": 74825, "loss": 3.523876190185547, "grad_norm": 0.7407708764076233, "learning_rate": 0.0001, "elbo": 3.52388, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3880483227773865}
{"step": 74830, "loss": 3.5141677856445312, "grad_norm": 0.7010111808776855, "learning_rate": 0.0001, "elbo": 3.51417, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.388341543513957}
{"step": 74835, "loss": 3.512279510498047, "grad_norm": 0.7474290132522583, "learning_rate": 0.0001, "elbo": 3.51228, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3886347642505275}
{"step": 74840, "loss": 3.522635269165039, "grad_norm": 0.7102977633476257, "learning_rate": 0.0001, "elbo": 3.52264, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.388927984987098}
{"step": 74845, "loss": 3.4936431884765624, "grad_norm": 0.6953792572021484, "learning_rate": 0.0001, "elbo": 3.49364, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.389221205723668}
{"step": 74850, "loss": 3.505994415283203, "grad_norm": 0.7521603107452393, "learning_rate": 0.0001, "elbo": 3.50599, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.38951442646024}
{"step": 74855, "loss": 3.500026321411133, "grad_norm": 0.7875263094902039, "learning_rate": 0.0001, "elbo": 3.50003, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.38980764719681}
{"step": 74860, "loss": 3.5062946319580077, "grad_norm": 0.8054408431053162, "learning_rate": 0.0001, "elbo": 3.50629, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.390100867933381}
{"step": 74865, "loss": 3.511214828491211, "grad_norm": 0.7405241131782532, "learning_rate": 0.0001, "elbo": 3.51121, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.390394088669951}
{"step": 74870, "loss": 3.5028030395507814, "grad_norm": 0.6987341642379761, "learning_rate": 0.0001, "elbo": 3.5028, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3906873094065215}
{"step": 74875, "loss": 3.504743194580078, "grad_norm": 0.8383501768112183, "learning_rate": 0.0001, "elbo": 3.50474, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.390980530143092}
{"step": 74875, "loss": 3.504743194580078, "grad_norm": 0.8383501768112183, "learning_rate": 0.0001, "elbo": 3.50474, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.390980530143092}
{"step": 74880, "loss": 3.5226425170898437, "grad_norm": 0.6801250576972961, "learning_rate": 0.0001, "elbo": 3.52264, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.391273750879662}
{"step": 74880, "loss": 3.5226425170898437, "grad_norm": 0.6801250576972961, "learning_rate": 0.0001, "elbo": 3.52264, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.391273750879662}
{"step": 74885, "loss": 3.519378662109375, "grad_norm": 0.6920798420906067, "learning_rate": 0.0001, "elbo": 3.51938, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.391566971616233}
{"step": 74890, "loss": 3.5081432342529295, "grad_norm": 0.6944644451141357, "learning_rate": 0.0001, "elbo": 3.50814, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.391860192352803}
{"step": 74895, "loss": 3.515158843994141, "grad_norm": 0.7026676535606384, "learning_rate": 0.0001, "elbo": 3.51516, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.392153413089374}
{"step": 74900, "loss": 3.5206935882568358, "grad_norm": 0.7226628065109253, "learning_rate": 0.0001, "elbo": 3.52069, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.392446633825944}
{"step": 74905, "loss": 3.5220191955566404, "grad_norm": 0.7390846610069275, "learning_rate": 0.0001, "elbo": 3.52202, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.392739854562515}
{"step": 74910, "loss": 3.4979400634765625, "grad_norm": 0.7121742963790894, "learning_rate": 0.0001, "elbo": 3.49794, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.393033075299085}
{"step": 74915, "loss": 3.524479293823242, "grad_norm": 0.8566808104515076, "learning_rate": 0.0001, "elbo": 3.52448, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.393326296035656}
{"step": 74920, "loss": 3.5055938720703126, "grad_norm": 0.9918278455734253, "learning_rate": 0.0001, "elbo": 3.50559, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.393619516772226}
{"step": 74925, "loss": 3.5125137329101563, "grad_norm": 0.8077285289764404, "learning_rate": 0.0001, "elbo": 3.51251, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3939127375087965}
{"step": 74930, "loss": 3.518558120727539, "grad_norm": 0.7220702767372131, "learning_rate": 0.0001, "elbo": 3.51856, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.394205958245367}
{"step": 74935, "loss": 3.5112186431884767, "grad_norm": 0.7010600566864014, "learning_rate": 0.0001, "elbo": 3.51122, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.394499178981937}
{"step": 74940, "loss": 3.515745162963867, "grad_norm": 0.7118887901306152, "learning_rate": 0.0001, "elbo": 3.51575, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.394792399718508}
{"step": 74945, "loss": 3.5092803955078127, "grad_norm": 0.7005414366722107, "learning_rate": 0.0001, "elbo": 3.50928, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.395085620455078}
{"step": 74950, "loss": 3.5138641357421876, "grad_norm": 0.6783912181854248, "learning_rate": 0.0001, "elbo": 3.51386, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.395378841191649}
{"step": 74955, "loss": 3.5116539001464844, "grad_norm": 0.7888386249542236, "learning_rate": 0.0001, "elbo": 3.51165, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.395672061928219}
{"step": 74960, "loss": 3.5123191833496095, "grad_norm": 0.7040369510650635, "learning_rate": 0.0001, "elbo": 3.51232, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.39596528266479}
{"step": 74965, "loss": 3.5262550354003905, "grad_norm": 0.6892680525779724, "learning_rate": 0.0001, "elbo": 3.52625, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.39625850340136}
{"step": 74970, "loss": 3.506592559814453, "grad_norm": 0.8025016188621521, "learning_rate": 0.0001, "elbo": 3.50659, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.396551724137931}
{"step": 74975, "loss": 3.5082916259765624, "grad_norm": 0.7432042360305786, "learning_rate": 0.0001, "elbo": 3.50829, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.396844944874502}
{"step": 74980, "loss": 3.50194091796875, "grad_norm": 0.9663299918174744, "learning_rate": 0.0001, "elbo": 3.50194, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.397138165611072}
{"step": 74985, "loss": 3.517689514160156, "grad_norm": 0.7153448462486267, "learning_rate": 0.0001, "elbo": 3.51769, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.397431386347643}
{"step": 74990, "loss": 3.5195858001708986, "grad_norm": 0.7249438166618347, "learning_rate": 0.0001, "elbo": 3.51959, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.397724607084213}
{"step": 74995, "loss": 3.5179428100585937, "grad_norm": 0.7623172998428345, "learning_rate": 0.0001, "elbo": 3.51794, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.398017827820784}
{"step": 75000, "loss": 3.5146514892578127, "grad_norm": 0.7300318479537964, "learning_rate": 0.0001, "elbo": 3.51465, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.398311048557354}
{"step": 75005, "loss": 3.5174617767333984, "grad_norm": 0.6920112371444702, "learning_rate": 0.0001, "elbo": 3.51746, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.398604269293925}
{"step": 75010, "loss": 3.502874755859375, "grad_norm": 0.8535717129707336, "learning_rate": 0.0001, "elbo": 3.50287, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.398897490030495}
{"step": 75015, "loss": 3.5082122802734377, "grad_norm": 0.7080475687980652, "learning_rate": 0.0001, "elbo": 3.50821, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.399190710767066}
{"step": 75015, "loss": 3.5082122802734377, "grad_norm": 0.7080475687980652, "learning_rate": 0.0001, "elbo": 3.50821, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.399190710767066}
{"step": 75020, "loss": 3.505280685424805, "grad_norm": 0.688117265701294, "learning_rate": 0.0001, "elbo": 3.50528, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.399483931503636}
{"step": 75025, "loss": 3.5230148315429686, "grad_norm": 0.6992023587226868, "learning_rate": 0.0001, "elbo": 3.52301, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.3997771522402065}
{"step": 75030, "loss": 3.519261932373047, "grad_norm": 0.7576079964637756, "learning_rate": 0.0001, "elbo": 3.51926, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.400070372976777}
{"step": 75035, "loss": 3.502846527099609, "grad_norm": 1.1941057443618774, "learning_rate": 0.0001, "elbo": 3.50285, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.400363593713347}
{"step": 75040, "loss": 3.5169368743896485, "grad_norm": 0.723465085029602, "learning_rate": 0.0001, "elbo": 3.51694, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.400656814449918}
{"step": 75045, "loss": 3.5102840423583985, "grad_norm": 0.7157350182533264, "learning_rate": 0.0001, "elbo": 3.51028, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.400950035186488}
{"step": 75050, "loss": 3.5197879791259767, "grad_norm": 0.7266099452972412, "learning_rate": 0.0001, "elbo": 3.51979, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.401243255923059}
{"step": 75055, "loss": 3.514939880371094, "grad_norm": 0.7185099124908447, "learning_rate": 0.0001, "elbo": 3.51494, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.401536476659629}
{"step": 75060, "loss": 3.519637680053711, "grad_norm": 0.8399905562400818, "learning_rate": 0.0001, "elbo": 3.51964, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4018296973962}
{"step": 75065, "loss": 3.52392578125, "grad_norm": 0.7058844566345215, "learning_rate": 0.0001, "elbo": 3.52393, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.40212291813277}
{"step": 75070, "loss": 3.5192001342773436, "grad_norm": 0.7608922123908997, "learning_rate": 0.0001, "elbo": 3.5192, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.402416138869341}
{"step": 75075, "loss": 3.5061649322509765, "grad_norm": 0.7711437344551086, "learning_rate": 0.0001, "elbo": 3.50616, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.402709359605911}
{"step": 75075, "loss": 3.5061649322509765, "grad_norm": 0.7711437344551086, "learning_rate": 0.0001, "elbo": 3.50616, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.402709359605911}
{"step": 75080, "loss": 3.516001892089844, "grad_norm": 0.6866549849510193, "learning_rate": 0.0001, "elbo": 3.516, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4030025803424815}
{"step": 75085, "loss": 3.513775634765625, "grad_norm": 0.6891373991966248, "learning_rate": 0.0001, "elbo": 3.51378, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.403295801079052}
{"step": 75090, "loss": 3.5039268493652345, "grad_norm": 0.6455075740814209, "learning_rate": 0.0001, "elbo": 3.50393, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.403589021815622}
{"step": 75095, "loss": 3.504896545410156, "grad_norm": 0.7197374701499939, "learning_rate": 0.0001, "elbo": 3.5049, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.403882242552193}
{"step": 75100, "loss": 3.5063941955566404, "grad_norm": 0.7649376392364502, "learning_rate": 0.0001, "elbo": 3.50639, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.404175463288764}
{"step": 75105, "loss": 3.511256790161133, "grad_norm": 0.7062184810638428, "learning_rate": 0.0001, "elbo": 3.51126, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.404468684025335}
{"step": 75110, "loss": 3.5092567443847655, "grad_norm": 0.6872088313102722, "learning_rate": 0.0001, "elbo": 3.50926, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.404761904761905}
{"step": 75115, "loss": 3.513457107543945, "grad_norm": 0.8739298582077026, "learning_rate": 0.0001, "elbo": 3.51346, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4050551254984756}
{"step": 75120, "loss": 3.5187530517578125, "grad_norm": 0.7164340615272522, "learning_rate": 0.0001, "elbo": 3.51875, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.405348346235046}
{"step": 75125, "loss": 3.528150939941406, "grad_norm": 1.1697266101837158, "learning_rate": 0.0001, "elbo": 3.52815, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4056415669716165}
{"step": 75130, "loss": 3.504621124267578, "grad_norm": 0.8653194904327393, "learning_rate": 0.0001, "elbo": 3.50462, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.405934787708187}
{"step": 75135, "loss": 3.5007816314697267, "grad_norm": 0.8760267496109009, "learning_rate": 0.0001, "elbo": 3.50078, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.406228008444757}
{"step": 75135, "loss": 3.5007816314697267, "grad_norm": 0.8760267496109009, "learning_rate": 0.0001, "elbo": 3.50078, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.406228008444757}
{"step": 75140, "loss": 3.5374423980712892, "grad_norm": 0.9957741498947144, "learning_rate": 0.0001, "elbo": 3.53744, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.406521229181328}
{"step": 75145, "loss": 3.527478790283203, "grad_norm": 0.7516846060752869, "learning_rate": 0.0001, "elbo": 3.52748, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.406814449917898}
{"step": 75150, "loss": 3.5297847747802735, "grad_norm": 0.7364745736122131, "learning_rate": 0.0001, "elbo": 3.52978, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.407107670654469}
{"step": 75155, "loss": 3.524353790283203, "grad_norm": 0.6921266913414001, "learning_rate": 0.0001, "elbo": 3.52435, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.407400891391039}
{"step": 75160, "loss": 3.5320083618164064, "grad_norm": 0.6574822068214417, "learning_rate": 0.0001, "elbo": 3.53201, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.40769411212761}
{"step": 75160, "loss": 3.5320083618164064, "grad_norm": 0.6574822068214417, "learning_rate": 0.0001, "elbo": 3.53201, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.40769411212761}
{"step": 75165, "loss": 3.5140335083007814, "grad_norm": 0.6883637309074402, "learning_rate": 0.0001, "elbo": 3.51403, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.40798733286418}
{"step": 75170, "loss": 3.521971893310547, "grad_norm": 0.7311243414878845, "learning_rate": 0.0001, "elbo": 3.52197, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4082805536007506}
{"step": 75175, "loss": 3.523516845703125, "grad_norm": 0.7250657677650452, "learning_rate": 0.0001, "elbo": 3.52352, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.408573774337321}
{"step": 75180, "loss": 3.498174285888672, "grad_norm": 0.7777181267738342, "learning_rate": 0.0001, "elbo": 3.49817, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4088669950738915}
{"step": 75185, "loss": 3.5200122833251952, "grad_norm": 0.7285817861557007, "learning_rate": 0.0001, "elbo": 3.52001, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.409160215810462}
{"step": 75190, "loss": 3.5303794860839846, "grad_norm": 0.7301869988441467, "learning_rate": 0.0001, "elbo": 3.53038, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.409453436547032}
{"step": 75195, "loss": 3.4989837646484374, "grad_norm": 0.7024852633476257, "learning_rate": 0.0001, "elbo": 3.49898, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.409746657283603}
{"step": 75200, "loss": 3.526464080810547, "grad_norm": 0.7409976720809937, "learning_rate": 0.0001, "elbo": 3.52646, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.410039878020173}
{"step": 75205, "loss": 3.522705078125, "grad_norm": 0.6753254532814026, "learning_rate": 0.0001, "elbo": 3.52271, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.410333098756744}
{"step": 75210, "loss": 3.524231719970703, "grad_norm": 0.6666776537895203, "learning_rate": 0.0001, "elbo": 3.52423, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.410626319493314}
{"step": 75215, "loss": 3.528584289550781, "grad_norm": 0.7817361354827881, "learning_rate": 0.0001, "elbo": 3.52858, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4109195402298855}
{"step": 75215, "loss": 3.528584289550781, "grad_norm": 0.7817361354827881, "learning_rate": 0.0001, "elbo": 3.52858, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4109195402298855}
{"step": 75220, "loss": 3.5143558502197267, "grad_norm": 0.7310211658477783, "learning_rate": 0.0001, "elbo": 3.51436, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.411212760966455}
{"step": 75220, "loss": 3.5143558502197267, "grad_norm": 0.7310211658477783, "learning_rate": 0.0001, "elbo": 3.51436, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.411212760966455}
{"step": 75225, "loss": 3.4995216369628905, "grad_norm": 0.7768194675445557, "learning_rate": 0.0001, "elbo": 3.49952, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.411505981703026}
{"step": 75230, "loss": 3.5184215545654296, "grad_norm": 0.7656598687171936, "learning_rate": 0.0001, "elbo": 3.51842, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.411799202439597}
{"step": 75235, "loss": 3.5168956756591796, "grad_norm": 0.7262354493141174, "learning_rate": 0.0001, "elbo": 3.5169, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.412092423176167}
{"step": 75240, "loss": 3.513593292236328, "grad_norm": 0.6762012243270874, "learning_rate": 0.0001, "elbo": 3.51359, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.412385643912738}
{"step": 75245, "loss": 3.5031005859375, "grad_norm": 0.6597881317138672, "learning_rate": 0.0001, "elbo": 3.5031, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.412678864649308}
{"step": 75245, "loss": 3.5031005859375, "grad_norm": 0.6597881317138672, "learning_rate": 0.0001, "elbo": 3.5031, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.412678864649308}
{"step": 75250, "loss": 3.501418685913086, "grad_norm": 0.6864160895347595, "learning_rate": 0.0001, "elbo": 3.50142, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.412972085385879}
{"step": 75255, "loss": 3.5111457824707033, "grad_norm": 0.7120133638381958, "learning_rate": 0.0001, "elbo": 3.51115, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.413265306122449}
{"step": 75260, "loss": 3.512043762207031, "grad_norm": 0.8517955541610718, "learning_rate": 0.0001, "elbo": 3.51204, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.41355852685902}
{"step": 75265, "loss": 3.5186477661132813, "grad_norm": 0.7989848256111145, "learning_rate": 0.0001, "elbo": 3.51865, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.41385174759559}
{"step": 75270, "loss": 3.5249114990234376, "grad_norm": 0.744392454624176, "learning_rate": 0.0001, "elbo": 3.52491, "mask_frac": 0.5013, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4141449683321605}
{"step": 75275, "loss": 3.515633392333984, "grad_norm": 0.7127866148948669, "learning_rate": 0.0001, "elbo": 3.51563, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.414438189068731}
{"step": 75280, "loss": 3.5049415588378907, "grad_norm": 0.6976674795150757, "learning_rate": 0.0001, "elbo": 3.50494, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.414731409805301}
{"step": 75285, "loss": 3.507654571533203, "grad_norm": 0.6960811614990234, "learning_rate": 0.0001, "elbo": 3.50765, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.415024630541872}
{"step": 75290, "loss": 3.509661102294922, "grad_norm": 0.7220016121864319, "learning_rate": 0.0001, "elbo": 3.50966, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.415317851278442}
{"step": 75295, "loss": 3.5070899963378905, "grad_norm": 0.7127872705459595, "learning_rate": 0.0001, "elbo": 3.50709, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.415611072015013}
{"step": 75300, "loss": 3.5015697479248047, "grad_norm": 0.8084959983825684, "learning_rate": 0.0001, "elbo": 3.50157, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.415904292751583}
{"step": 75305, "loss": 3.5108722686767577, "grad_norm": 0.8426725268363953, "learning_rate": 0.0001, "elbo": 3.51087, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.416197513488154}
{"step": 75310, "loss": 3.5050785064697267, "grad_norm": 0.9881159067153931, "learning_rate": 0.0001, "elbo": 3.50508, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.416490734224724}
{"step": 75315, "loss": 3.523741912841797, "grad_norm": 0.6809250712394714, "learning_rate": 0.0001, "elbo": 3.52374, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.416783954961295}
{"step": 75320, "loss": 3.522050476074219, "grad_norm": 0.8239424228668213, "learning_rate": 0.0001, "elbo": 3.52205, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.417077175697865}
{"step": 75325, "loss": 3.5199813842773438, "grad_norm": 0.7092623710632324, "learning_rate": 0.0001, "elbo": 3.51998, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4173703964344355}
{"step": 75330, "loss": 3.513232421875, "grad_norm": 0.7393897175788879, "learning_rate": 0.0001, "elbo": 3.51323, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.417663617171006}
{"step": 75335, "loss": 3.5222068786621095, "grad_norm": 0.7496473789215088, "learning_rate": 0.0001, "elbo": 3.52221, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.417956837907576}
{"step": 75340, "loss": 3.5071029663085938, "grad_norm": 0.7581872940063477, "learning_rate": 0.0001, "elbo": 3.5071, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.418250058644148}
{"step": 75345, "loss": 3.5216670989990235, "grad_norm": 0.742725670337677, "learning_rate": 0.0001, "elbo": 3.52167, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.418543279380717}
{"step": 75350, "loss": 3.5258445739746094, "grad_norm": 0.7032456398010254, "learning_rate": 0.0001, "elbo": 3.52584, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.418836500117289}
{"step": 75355, "loss": 3.526382827758789, "grad_norm": 0.7503858208656311, "learning_rate": 0.0001, "elbo": 3.52638, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.419129720853859}
{"step": 75360, "loss": 3.508126068115234, "grad_norm": 0.6954192519187927, "learning_rate": 0.0001, "elbo": 3.50813, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.41942294159043}
{"step": 75365, "loss": 3.508836364746094, "grad_norm": 0.704491376876831, "learning_rate": 0.0001, "elbo": 3.50884, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.419716162327}
{"step": 75370, "loss": 3.513340377807617, "grad_norm": 0.9412086009979248, "learning_rate": 0.0001, "elbo": 3.51334, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4200093830635705}
{"step": 75375, "loss": 3.523295593261719, "grad_norm": 0.6976296901702881, "learning_rate": 0.0001, "elbo": 3.5233, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.420302603800141}
{"step": 75380, "loss": 3.521715545654297, "grad_norm": 1.2412641048431396, "learning_rate": 0.0001, "elbo": 3.52172, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.420595824536711}
{"step": 75385, "loss": 3.524341583251953, "grad_norm": 0.7924606800079346, "learning_rate": 0.0001, "elbo": 3.52434, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.420889045273282}
{"step": 75390, "loss": 3.507726287841797, "grad_norm": 0.6878613233566284, "learning_rate": 0.0001, "elbo": 3.50773, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.421182266009852}
{"step": 75395, "loss": 3.5122039794921873, "grad_norm": 0.7553679943084717, "learning_rate": 0.0001, "elbo": 3.5122, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.421475486746423}
{"step": 75400, "loss": 3.5047550201416016, "grad_norm": 0.7374364733695984, "learning_rate": 0.0001, "elbo": 3.50476, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.421768707482993}
{"step": 75405, "loss": 3.5119308471679687, "grad_norm": 0.7214546203613281, "learning_rate": 0.0001, "elbo": 3.51193, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.422061928219564}
{"step": 75410, "loss": 3.510985565185547, "grad_norm": 0.7154744863510132, "learning_rate": 0.0001, "elbo": 3.51099, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.422355148956134}
{"step": 75415, "loss": 3.5107460021972656, "grad_norm": 0.736640989780426, "learning_rate": 0.0001, "elbo": 3.51075, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.422648369692705}
{"step": 75420, "loss": 3.5188438415527346, "grad_norm": 0.856988251209259, "learning_rate": 0.0001, "elbo": 3.51884, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.422941590429275}
{"step": 75425, "loss": 3.499583053588867, "grad_norm": 0.701488196849823, "learning_rate": 0.0001, "elbo": 3.49958, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4232348111658455}
{"step": 75430, "loss": 3.52091064453125, "grad_norm": 0.7362241148948669, "learning_rate": 0.0001, "elbo": 3.52091, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.423528031902416}
{"step": 75435, "loss": 3.513132095336914, "grad_norm": 0.8471929430961609, "learning_rate": 0.0001, "elbo": 3.51313, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.423821252638986}
{"step": 75440, "loss": 3.5303413391113283, "grad_norm": 0.6833757162094116, "learning_rate": 0.0001, "elbo": 3.53034, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.424114473375557}
{"step": 75445, "loss": 3.500057601928711, "grad_norm": 0.6821146607398987, "learning_rate": 0.0001, "elbo": 3.50006, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.424407694112127}
{"step": 75450, "loss": 3.517487716674805, "grad_norm": 0.7128725647926331, "learning_rate": 0.0001, "elbo": 3.51749, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.424700914848698}
{"step": 75455, "loss": 3.5259986877441407, "grad_norm": 0.6964712738990784, "learning_rate": 0.0001, "elbo": 3.526, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.424994135585268}
{"step": 75460, "loss": 3.5124843597412108, "grad_norm": 0.6952791810035706, "learning_rate": 0.0001, "elbo": 3.51248, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.425287356321839}
{"step": 75465, "loss": 3.5047462463378904, "grad_norm": 0.6864350438117981, "learning_rate": 0.0001, "elbo": 3.50475, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.42558057705841}
{"step": 75470, "loss": 3.504633331298828, "grad_norm": 0.7646743059158325, "learning_rate": 0.0001, "elbo": 3.50463, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4258737977949805}
{"step": 75475, "loss": 3.5040340423583984, "grad_norm": 0.6905701160430908, "learning_rate": 0.0001, "elbo": 3.50403, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.426167018531551}
{"step": 75480, "loss": 3.5089210510253905, "grad_norm": 0.706866443157196, "learning_rate": 0.0001, "elbo": 3.50892, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.426460239268121}
{"step": 75485, "loss": 3.481836700439453, "grad_norm": 0.6633985638618469, "learning_rate": 0.0001, "elbo": 3.48184, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.426753460004692}
{"step": 75490, "loss": 3.509088134765625, "grad_norm": 0.7586512565612793, "learning_rate": 0.0001, "elbo": 3.50909, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.427046680741262}
{"step": 75495, "loss": 3.515782928466797, "grad_norm": 0.7554756999015808, "learning_rate": 0.0001, "elbo": 3.51578, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.427339901477833}
{"step": 75500, "loss": 3.5254981994628904, "grad_norm": 0.7221004366874695, "learning_rate": 0.0001, "elbo": 3.5255, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.427633122214403}
{"step": 75505, "loss": 3.5059322357177733, "grad_norm": 0.7293547987937927, "learning_rate": 0.0001, "elbo": 3.50593, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.427926342950974}
{"step": 75510, "loss": 3.510957717895508, "grad_norm": 1.127020239830017, "learning_rate": 0.0001, "elbo": 3.51096, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.428219563687544}
{"step": 75515, "loss": 3.510295867919922, "grad_norm": 0.7833192348480225, "learning_rate": 0.0001, "elbo": 3.5103, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.428512784424115}
{"step": 75520, "loss": 3.5125286102294924, "grad_norm": 0.7769187092781067, "learning_rate": 0.0001, "elbo": 3.51253, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.428806005160685}
{"step": 75525, "loss": 3.516836166381836, "grad_norm": 0.7081180214881897, "learning_rate": 0.0001, "elbo": 3.51684, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4290992258972555}
{"step": 75530, "loss": 3.5039741516113283, "grad_norm": 0.8552872538566589, "learning_rate": 0.0001, "elbo": 3.50397, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.429392446633826}
{"step": 75535, "loss": 3.500977325439453, "grad_norm": 0.7052826881408691, "learning_rate": 0.0001, "elbo": 3.50098, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.429685667370396}
{"step": 75535, "loss": 3.500977325439453, "grad_norm": 0.7052826881408691, "learning_rate": 0.0001, "elbo": 3.50098, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.429685667370396}
{"step": 75540, "loss": 3.512586975097656, "grad_norm": 0.9507489800453186, "learning_rate": 0.0001, "elbo": 3.51259, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.429978888106967}
{"step": 75540, "loss": 3.512586975097656, "grad_norm": 0.9507489800453186, "learning_rate": 0.0001, "elbo": 3.51259, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.429978888106967}
{"step": 75545, "loss": 3.502946472167969, "grad_norm": 0.832089364528656, "learning_rate": 0.0001, "elbo": 3.50295, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.430272108843537}
{"step": 75550, "loss": 3.5060985565185545, "grad_norm": 0.7686390280723572, "learning_rate": 0.0001, "elbo": 3.5061, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.430565329580108}
{"step": 75550, "loss": 3.5060985565185545, "grad_norm": 0.7686390280723572, "learning_rate": 0.0001, "elbo": 3.5061, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.430565329580108}
{"step": 75555, "loss": 3.502930450439453, "grad_norm": 0.7238073945045471, "learning_rate": 0.0001, "elbo": 3.50293, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.430858550316678}
{"step": 75560, "loss": 3.5125728607177735, "grad_norm": 0.7231084108352661, "learning_rate": 0.0001, "elbo": 3.51257, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.431151771053249}
{"step": 75565, "loss": 3.5129371643066407, "grad_norm": 0.6642207503318787, "learning_rate": 0.0001, "elbo": 3.51294, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.431444991789819}
{"step": 75570, "loss": 3.5162559509277345, "grad_norm": 0.713132917881012, "learning_rate": 0.0001, "elbo": 3.51626, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4317382125263896}
{"step": 75575, "loss": 3.5141021728515627, "grad_norm": 0.7417935729026794, "learning_rate": 0.0001, "elbo": 3.5141, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.43203143326296}
{"step": 75580, "loss": 3.518393707275391, "grad_norm": 0.6802161931991577, "learning_rate": 0.0001, "elbo": 3.51839, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4323246539995305}
{"step": 75585, "loss": 3.515699005126953, "grad_norm": 0.8695662021636963, "learning_rate": 0.0001, "elbo": 3.5157, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.432617874736101}
{"step": 75590, "loss": 3.525765228271484, "grad_norm": 0.7618371248245239, "learning_rate": 0.0001, "elbo": 3.52577, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.432911095472672}
{"step": 75595, "loss": 3.505019760131836, "grad_norm": 0.7408501505851746, "learning_rate": 0.0001, "elbo": 3.50502, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.433204316209243}
{"step": 75600, "loss": 3.517013931274414, "grad_norm": 0.7442191243171692, "learning_rate": 0.0001, "elbo": 3.51701, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.433497536945813}
{"step": 75605, "loss": 3.5101593017578123, "grad_norm": 0.7164586782455444, "learning_rate": 0.0001, "elbo": 3.51016, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.433790757682384}
{"step": 75610, "loss": 3.517751693725586, "grad_norm": 0.7459684014320374, "learning_rate": 0.0001, "elbo": 3.51775, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.434083978418954}
{"step": 75615, "loss": 3.5080360412597655, "grad_norm": 0.7632352709770203, "learning_rate": 0.0001, "elbo": 3.50804, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4343771991555245}
{"step": 75620, "loss": 3.535297393798828, "grad_norm": 0.7042441368103027, "learning_rate": 0.0001, "elbo": 3.5353, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.434670419892095}
{"step": 75625, "loss": 3.5150306701660154, "grad_norm": 0.6853291988372803, "learning_rate": 0.0001, "elbo": 3.51503, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.434963640628665}
{"step": 75630, "loss": 3.49609375, "grad_norm": 0.7622460126876831, "learning_rate": 0.0001, "elbo": 3.49609, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.435256861365236}
{"step": 75635, "loss": 3.519389343261719, "grad_norm": 0.749898374080658, "learning_rate": 0.0001, "elbo": 3.51939, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.435550082101806}
{"step": 75640, "loss": 3.5078704833984373, "grad_norm": 0.7808369398117065, "learning_rate": 0.0001, "elbo": 3.50787, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.435843302838377}
{"step": 75645, "loss": 3.5189437866210938, "grad_norm": 0.6610159873962402, "learning_rate": 0.0001, "elbo": 3.51894, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.436136523574947}
{"step": 75650, "loss": 3.5167011260986327, "grad_norm": 0.6587793827056885, "learning_rate": 0.0001, "elbo": 3.5167, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.436429744311518}
{"step": 75655, "loss": 3.4894203186035155, "grad_norm": 1.0869569778442383, "learning_rate": 0.0001, "elbo": 3.48942, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.436722965048088}
{"step": 75660, "loss": 3.5144554138183595, "grad_norm": 0.7045968174934387, "learning_rate": 0.0001, "elbo": 3.51446, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.437016185784659}
{"step": 75665, "loss": 3.4974395751953127, "grad_norm": 0.6717132925987244, "learning_rate": 0.0001, "elbo": 3.49744, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.437309406521229}
{"step": 75670, "loss": 3.531951141357422, "grad_norm": 0.6691161394119263, "learning_rate": 0.0001, "elbo": 3.53195, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4376026272577995}
{"step": 75670, "loss": 3.531951141357422, "grad_norm": 0.6691161394119263, "learning_rate": 0.0001, "elbo": 3.53195, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4376026272577995}
{"step": 75675, "loss": 3.511001205444336, "grad_norm": 0.6774275302886963, "learning_rate": 0.0001, "elbo": 3.511, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.43789584799437}
{"step": 75680, "loss": 3.5147003173828124, "grad_norm": 1.285687804222107, "learning_rate": 0.0001, "elbo": 3.5147, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.43818906873094}
{"step": 75685, "loss": 3.50753173828125, "grad_norm": 0.6969834566116333, "learning_rate": 0.0001, "elbo": 3.50753, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.438482289467511}
{"step": 75690, "loss": 3.5132591247558596, "grad_norm": 0.7263457775115967, "learning_rate": 0.0001, "elbo": 3.51326, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.438775510204081}
{"step": 75695, "loss": 3.506749725341797, "grad_norm": 0.7031629681587219, "learning_rate": 0.0001, "elbo": 3.50675, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.439068730940652}
{"step": 75700, "loss": 3.5004539489746094, "grad_norm": 0.6916260719299316, "learning_rate": 0.0001, "elbo": 3.50045, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.439361951677222}
{"step": 75705, "loss": 3.5083595275878907, "grad_norm": 0.685657262802124, "learning_rate": 0.0001, "elbo": 3.50836, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.439655172413793}
{"step": 75710, "loss": 3.5099422454833986, "grad_norm": 0.7167017459869385, "learning_rate": 0.0001, "elbo": 3.50994, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.439948393150363}
{"step": 75715, "loss": 3.5100303649902345, "grad_norm": 0.7816689014434814, "learning_rate": 0.0001, "elbo": 3.51003, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4402416138869345}
{"step": 75720, "loss": 3.5071083068847657, "grad_norm": 0.7551908493041992, "learning_rate": 0.0001, "elbo": 3.50711, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.440534834623505}
{"step": 75725, "loss": 3.5080223083496094, "grad_norm": 0.7158167958259583, "learning_rate": 0.0001, "elbo": 3.50802, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.440828055360075}
{"step": 75730, "loss": 3.516743469238281, "grad_norm": 0.7140142321586609, "learning_rate": 0.0001, "elbo": 3.51674, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.441121276096646}
{"step": 75735, "loss": 3.5237667083740236, "grad_norm": 0.7325767874717712, "learning_rate": 0.0001, "elbo": 3.52377, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.441414496833216}
{"step": 75740, "loss": 3.515947723388672, "grad_norm": 0.7179368734359741, "learning_rate": 0.0001, "elbo": 3.51595, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.441707717569787}
{"step": 75745, "loss": 3.522571563720703, "grad_norm": 0.7502896785736084, "learning_rate": 0.0001, "elbo": 3.52257, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.442000938306357}
{"step": 75750, "loss": 3.519140625, "grad_norm": 0.6756474375724792, "learning_rate": 0.0001, "elbo": 3.51914, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.442294159042928}
{"step": 75755, "loss": 3.500487518310547, "grad_norm": 0.9186156988143921, "learning_rate": 0.0001, "elbo": 3.50049, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.442587379779498}
{"step": 75760, "loss": 3.5206424713134767, "grad_norm": 0.8497604131698608, "learning_rate": 0.0001, "elbo": 3.52064, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.442880600516069}
{"step": 75765, "loss": 3.5058826446533202, "grad_norm": 0.788048505783081, "learning_rate": 0.0001, "elbo": 3.50588, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.443173821252639}
{"step": 75770, "loss": 3.523162078857422, "grad_norm": 0.7647744417190552, "learning_rate": 0.0001, "elbo": 3.52316, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4434670419892095}
{"step": 75775, "loss": 3.5091041564941405, "grad_norm": 0.6880605816841125, "learning_rate": 0.0001, "elbo": 3.5091, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.44376026272578}
{"step": 75780, "loss": 3.512384033203125, "grad_norm": 0.7293076515197754, "learning_rate": 0.0001, "elbo": 3.51238, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.44405348346235}
{"step": 75785, "loss": 3.508250427246094, "grad_norm": 0.7232928276062012, "learning_rate": 0.0001, "elbo": 3.50825, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.444346704198921}
{"step": 75790, "loss": 3.5105430603027346, "grad_norm": 0.7082463502883911, "learning_rate": 0.0001, "elbo": 3.51054, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.444639924935491}
{"step": 75795, "loss": 3.5236747741699217, "grad_norm": 0.6890546083450317, "learning_rate": 0.0001, "elbo": 3.52367, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.444933145672062}
{"step": 75800, "loss": 3.517286682128906, "grad_norm": 0.7425275444984436, "learning_rate": 0.0001, "elbo": 3.51729, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.445226366408632}
{"step": 75805, "loss": 3.5148746490478517, "grad_norm": 0.7534109950065613, "learning_rate": 0.0001, "elbo": 3.51487, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.445519587145203}
{"step": 75810, "loss": 3.5060672760009766, "grad_norm": 0.7088820338249207, "learning_rate": 0.0001, "elbo": 3.50607, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.445812807881773}
{"step": 75815, "loss": 3.5207904815673827, "grad_norm": 0.6696543097496033, "learning_rate": 0.0001, "elbo": 3.52079, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.446106028618344}
{"step": 75820, "loss": 3.5123046875, "grad_norm": 0.6848786473274231, "learning_rate": 0.0001, "elbo": 3.5123, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.446399249354914}
{"step": 75820, "loss": 3.5123046875, "grad_norm": 0.6848786473274231, "learning_rate": 0.0001, "elbo": 3.5123, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.446399249354914}
{"step": 75825, "loss": 3.527614974975586, "grad_norm": 0.7172051072120667, "learning_rate": 0.0001, "elbo": 3.52761, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4466924700914845}
{"step": 75830, "loss": 3.513585662841797, "grad_norm": 0.7608337998390198, "learning_rate": 0.0001, "elbo": 3.51359, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.446985690828055}
{"step": 75835, "loss": 3.494272232055664, "grad_norm": 0.726716160774231, "learning_rate": 0.0001, "elbo": 3.49427, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.447278911564625}
{"step": 75840, "loss": 3.524626922607422, "grad_norm": 0.7048805356025696, "learning_rate": 0.0001, "elbo": 3.52463, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.447572132301197}
{"step": 75845, "loss": 3.51558837890625, "grad_norm": 0.6908621191978455, "learning_rate": 0.0001, "elbo": 3.51559, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.447865353037767}
{"step": 75850, "loss": 3.511469268798828, "grad_norm": 0.6890934705734253, "learning_rate": 0.0001, "elbo": 3.51147, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.448158573774338}
{"step": 75855, "loss": 3.5127708435058596, "grad_norm": 0.7232103943824768, "learning_rate": 0.0001, "elbo": 3.51277, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.448451794510908}
{"step": 75860, "loss": 3.5087863922119142, "grad_norm": 1.1295703649520874, "learning_rate": 0.0001, "elbo": 3.50879, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.448745015247479}
{"step": 75865, "loss": 3.5191696166992186, "grad_norm": 0.6884049773216248, "learning_rate": 0.0001, "elbo": 3.51917, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.449038235984049}
{"step": 75870, "loss": 3.520154571533203, "grad_norm": 0.6948040723800659, "learning_rate": 0.0001, "elbo": 3.52015, "mask_frac": 0.5013, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4493314567206195}
{"step": 75875, "loss": 3.510883331298828, "grad_norm": 0.693165123462677, "learning_rate": 0.0001, "elbo": 3.51088, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.44962467745719}
{"step": 75880, "loss": 3.494764709472656, "grad_norm": 0.7003865838050842, "learning_rate": 0.0001, "elbo": 3.49476, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.44991789819376}
{"step": 75885, "loss": 3.5266864776611326, "grad_norm": 0.7371214628219604, "learning_rate": 0.0001, "elbo": 3.52669, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.450211118930331}
{"step": 75890, "loss": 3.513057327270508, "grad_norm": 0.7248039245605469, "learning_rate": 0.0001, "elbo": 3.51306, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.450504339666901}
{"step": 75895, "loss": 3.519369125366211, "grad_norm": 0.763611376285553, "learning_rate": 0.0001, "elbo": 3.51937, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.450797560403472}
{"step": 75900, "loss": 3.5076236724853516, "grad_norm": 0.6951736807823181, "learning_rate": 0.0001, "elbo": 3.50762, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.451090781140042}
{"step": 75905, "loss": 3.5181678771972655, "grad_norm": 0.6671916842460632, "learning_rate": 0.0001, "elbo": 3.51817, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.451384001876613}
{"step": 75910, "loss": 3.522195053100586, "grad_norm": 0.9279901385307312, "learning_rate": 0.0001, "elbo": 3.5222, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.451677222613183}
{"step": 75915, "loss": 3.4942085266113283, "grad_norm": 0.7025877237319946, "learning_rate": 0.0001, "elbo": 3.49421, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.451970443349754}
{"step": 75920, "loss": 3.5273696899414064, "grad_norm": 0.7232971787452698, "learning_rate": 0.0001, "elbo": 3.52737, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.452263664086324}
{"step": 75925, "loss": 3.50811767578125, "grad_norm": 0.6872389316558838, "learning_rate": 0.0001, "elbo": 3.50812, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4525568848228945}
{"step": 75925, "loss": 3.50811767578125, "grad_norm": 0.6872389316558838, "learning_rate": 0.0001, "elbo": 3.50812, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4525568848228945}
{"step": 75930, "loss": 3.4995964050292967, "grad_norm": 0.85318922996521, "learning_rate": 0.0001, "elbo": 3.4996, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.452850105559465}
{"step": 75930, "loss": 3.4995964050292967, "grad_norm": 0.85318922996521, "learning_rate": 0.0001, "elbo": 3.4996, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.452850105559465}
{"step": 75935, "loss": 3.501636505126953, "grad_norm": 0.6891077756881714, "learning_rate": 0.0001, "elbo": 3.50164, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.453143326296035}
{"step": 75940, "loss": 3.517594909667969, "grad_norm": 0.6843128800392151, "learning_rate": 0.0001, "elbo": 3.51759, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.453436547032606}
{"step": 75945, "loss": 3.532075881958008, "grad_norm": 0.7400769591331482, "learning_rate": 0.0001, "elbo": 3.53208, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.453729767769176}
{"step": 75950, "loss": 3.5218170166015623, "grad_norm": 0.7194498181343079, "learning_rate": 0.0001, "elbo": 3.52182, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.454022988505747}
{"step": 75955, "loss": 3.5153266906738283, "grad_norm": 0.7152003645896912, "learning_rate": 0.0001, "elbo": 3.51533, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.454316209242318}
{"step": 75960, "loss": 3.5013702392578123, "grad_norm": 0.697912335395813, "learning_rate": 0.0001, "elbo": 3.50137, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.454609429978888}
{"step": 75965, "loss": 3.5246604919433593, "grad_norm": 0.7297993302345276, "learning_rate": 0.0001, "elbo": 3.52466, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.454902650715459}
{"step": 75970, "loss": 3.530763626098633, "grad_norm": 0.7138503789901733, "learning_rate": 0.0001, "elbo": 3.53076, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.455195871452029}
{"step": 75975, "loss": 3.5055683135986326, "grad_norm": 0.8335002660751343, "learning_rate": 0.0001, "elbo": 3.50557, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4554890921886}
{"step": 75980, "loss": 3.49395751953125, "grad_norm": 0.7211291790008545, "learning_rate": 0.0001, "elbo": 3.49396, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.45578231292517}
{"step": 75985, "loss": 3.5242000579833985, "grad_norm": 0.7488017082214355, "learning_rate": 0.0001, "elbo": 3.5242, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.456075533661741}
{"step": 75990, "loss": 3.515589141845703, "grad_norm": 0.7315305471420288, "learning_rate": 0.0001, "elbo": 3.51559, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.456368754398311}
{"step": 75995, "loss": 3.510811614990234, "grad_norm": 0.7181836366653442, "learning_rate": 0.0001, "elbo": 3.51081, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.456661975134882}
{"step": 76000, "loss": 3.5020767211914063, "grad_norm": 0.6669309735298157, "learning_rate": 0.0001, "elbo": 3.50208, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.456955195871452}
{"step": 76005, "loss": 3.508466339111328, "grad_norm": 0.7091178894042969, "learning_rate": 0.0001, "elbo": 3.50847, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.457248416608023}
{"step": 76010, "loss": 3.510905456542969, "grad_norm": 0.7163596153259277, "learning_rate": 0.0001, "elbo": 3.51091, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.457541637344593}
{"step": 76015, "loss": 3.5072509765625, "grad_norm": 0.7275465726852417, "learning_rate": 0.0001, "elbo": 3.50725, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4578348580811635}
{"step": 76020, "loss": 3.5040870666503907, "grad_norm": 0.736291229724884, "learning_rate": 0.0001, "elbo": 3.50409, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.458128078817734}
{"step": 76025, "loss": 3.531803512573242, "grad_norm": 0.7013710737228394, "learning_rate": 0.0001, "elbo": 3.5318, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.458421299554304}
{"step": 76030, "loss": 3.503975677490234, "grad_norm": 0.7230982780456543, "learning_rate": 0.0001, "elbo": 3.50398, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.458714520290875}
{"step": 76035, "loss": 3.5158660888671873, "grad_norm": 0.6960762143135071, "learning_rate": 0.0001, "elbo": 3.51587, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.459007741027445}
{"step": 76040, "loss": 3.512449264526367, "grad_norm": 0.7538782358169556, "learning_rate": 0.0001, "elbo": 3.51245, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.459300961764016}
{"step": 76045, "loss": 3.500417709350586, "grad_norm": 0.767326295375824, "learning_rate": 0.0001, "elbo": 3.50042, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.459594182500586}
{"step": 76050, "loss": 3.5076034545898436, "grad_norm": 0.7051838040351868, "learning_rate": 0.0001, "elbo": 3.5076, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.459887403237157}
{"step": 76055, "loss": 3.502239990234375, "grad_norm": 0.7086490392684937, "learning_rate": 0.0001, "elbo": 3.50224, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.460180623973727}
{"step": 76060, "loss": 3.5000518798828124, "grad_norm": 0.8394383788108826, "learning_rate": 0.0001, "elbo": 3.50005, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.460473844710298}
{"step": 76060, "loss": 3.5000518798828124, "grad_norm": 0.8394383788108826, "learning_rate": 0.0001, "elbo": 3.50005, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.460473844710298}
{"step": 76065, "loss": 3.502025604248047, "grad_norm": 0.7145818471908569, "learning_rate": 0.0001, "elbo": 3.50203, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.460767065446868}
{"step": 76065, "loss": 3.502025604248047, "grad_norm": 0.7145818471908569, "learning_rate": 0.0001, "elbo": 3.50203, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.460767065446868}
{"step": 76070, "loss": 3.4994438171386717, "grad_norm": 0.7151293754577637, "learning_rate": 0.0001, "elbo": 3.49944, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4610602861834385}
{"step": 76075, "loss": 3.516133499145508, "grad_norm": 0.6913013458251953, "learning_rate": 0.0001, "elbo": 3.51613, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.461353506920009}
{"step": 76080, "loss": 3.5265243530273436, "grad_norm": 0.7285663485527039, "learning_rate": 0.0001, "elbo": 3.52652, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.46164672765658}
{"step": 76085, "loss": 3.515995407104492, "grad_norm": 0.7172811031341553, "learning_rate": 0.0001, "elbo": 3.516, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.461939948393151}
{"step": 76090, "loss": 3.5047298431396485, "grad_norm": 0.7269892692565918, "learning_rate": 0.0001, "elbo": 3.50473, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.462233169129721}
{"step": 76095, "loss": 3.51367073059082, "grad_norm": 0.7478865385055542, "learning_rate": 0.0001, "elbo": 3.51367, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.462526389866292}
{"step": 76100, "loss": 3.4991531372070312, "grad_norm": 0.7309165596961975, "learning_rate": 0.0001, "elbo": 3.49915, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.462819610602862}
{"step": 76105, "loss": 3.5247871398925783, "grad_norm": 0.8189268708229065, "learning_rate": 0.0001, "elbo": 3.52479, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.463112831339433}
{"step": 76110, "loss": 3.5179672241210938, "grad_norm": 0.7295137047767639, "learning_rate": 0.0001, "elbo": 3.51797, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.463406052076003}
{"step": 76115, "loss": 3.5235740661621096, "grad_norm": 0.8268997073173523, "learning_rate": 0.0001, "elbo": 3.52357, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4636992728125735}
{"step": 76115, "loss": 3.5235740661621096, "grad_norm": 0.8268997073173523, "learning_rate": 0.0001, "elbo": 3.52357, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4636992728125735}
{"step": 76120, "loss": 3.512806701660156, "grad_norm": 0.7458462715148926, "learning_rate": 0.0001, "elbo": 3.51281, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.463992493549144}
{"step": 76125, "loss": 3.5007709503173827, "grad_norm": 0.6888450980186462, "learning_rate": 0.0001, "elbo": 3.50077, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.464285714285714}
{"step": 76130, "loss": 3.512917709350586, "grad_norm": 0.7477227449417114, "learning_rate": 0.0001, "elbo": 3.51292, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.464578935022285}
{"step": 76130, "loss": 3.512917709350586, "grad_norm": 0.7477227449417114, "learning_rate": 0.0001, "elbo": 3.51292, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.464578935022285}
{"step": 76135, "loss": 3.519109344482422, "grad_norm": 0.8315013647079468, "learning_rate": 0.0001, "elbo": 3.51911, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.464872155758855}
{"step": 76135, "loss": 3.519109344482422, "grad_norm": 0.8315013647079468, "learning_rate": 0.0001, "elbo": 3.51911, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.464872155758855}
{"step": 76140, "loss": 3.4999099731445313, "grad_norm": 0.7856372594833374, "learning_rate": 0.0001, "elbo": 3.49991, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.465165376495426}
{"step": 76145, "loss": 3.510573959350586, "grad_norm": 0.7433690428733826, "learning_rate": 0.0001, "elbo": 3.51057, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.465458597231996}
{"step": 76150, "loss": 3.5297340393066405, "grad_norm": 0.7528077363967896, "learning_rate": 0.0001, "elbo": 3.52973, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.465751817968567}
{"step": 76155, "loss": 3.499333953857422, "grad_norm": 0.6934328079223633, "learning_rate": 0.0001, "elbo": 3.49933, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.466045038705137}
{"step": 76160, "loss": 3.5080245971679687, "grad_norm": 0.8673765063285828, "learning_rate": 0.0001, "elbo": 3.50802, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.466338259441708}
{"step": 76165, "loss": 3.524713134765625, "grad_norm": 0.7183874845504761, "learning_rate": 0.0001, "elbo": 3.52471, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.466631480178278}
{"step": 76170, "loss": 3.5121631622314453, "grad_norm": 0.7429375052452087, "learning_rate": 0.0001, "elbo": 3.51216, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4669247009148485}
{"step": 76175, "loss": 3.5351524353027344, "grad_norm": 0.7687734365463257, "learning_rate": 0.0001, "elbo": 3.53515, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.467217921651419}
{"step": 76180, "loss": 3.517210006713867, "grad_norm": 0.7129563689231873, "learning_rate": 0.0001, "elbo": 3.51721, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.467511142387989}
{"step": 76185, "loss": 3.5299732208251955, "grad_norm": 0.7512298226356506, "learning_rate": 0.0001, "elbo": 3.52997, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.46780436312456}
{"step": 76190, "loss": 3.507146453857422, "grad_norm": 0.7989796996116638, "learning_rate": 0.0001, "elbo": 3.50715, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.46809758386113}
{"step": 76195, "loss": 3.4996646881103515, "grad_norm": 0.7878916263580322, "learning_rate": 0.0001, "elbo": 3.49966, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.468390804597701}
{"step": 76200, "loss": 3.513456344604492, "grad_norm": 0.768458902835846, "learning_rate": 0.0001, "elbo": 3.51346, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.468684025334271}
{"step": 76205, "loss": 3.501414489746094, "grad_norm": 0.6899943947792053, "learning_rate": 0.0001, "elbo": 3.50141, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.468977246070843}
{"step": 76210, "loss": 3.5141475677490233, "grad_norm": 0.731561005115509, "learning_rate": 0.0001, "elbo": 3.51415, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.469270466807413}
{"step": 76215, "loss": 3.506961441040039, "grad_norm": 0.7062264084815979, "learning_rate": 0.0001, "elbo": 3.50696, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4695636875439835}
{"step": 76220, "loss": 3.5080780029296874, "grad_norm": 0.6832058429718018, "learning_rate": 0.0001, "elbo": 3.50808, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.469856908280554}
{"step": 76225, "loss": 3.5143154144287108, "grad_norm": 0.7175360918045044, "learning_rate": 0.0001, "elbo": 3.51432, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.470150129017124}
{"step": 76230, "loss": 3.519178009033203, "grad_norm": 0.7814382910728455, "learning_rate": 0.0001, "elbo": 3.51918, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.470443349753695}
{"step": 76235, "loss": 3.5182174682617187, "grad_norm": 0.6809108257293701, "learning_rate": 0.0001, "elbo": 3.51822, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.470736570490265}
{"step": 76240, "loss": 3.502564239501953, "grad_norm": 0.6790540814399719, "learning_rate": 0.0001, "elbo": 3.50256, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.471029791226836}
{"step": 76245, "loss": 3.504121017456055, "grad_norm": 0.7013258337974548, "learning_rate": 0.0001, "elbo": 3.50412, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.471323011963406}
{"step": 76250, "loss": 3.5056976318359374, "grad_norm": 0.6815553903579712, "learning_rate": 0.0001, "elbo": 3.5057, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.471616232699977}
{"step": 76255, "loss": 3.4815532684326174, "grad_norm": 0.706822395324707, "learning_rate": 0.0001, "elbo": 3.48155, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.471909453436547}
{"step": 76260, "loss": 3.507619857788086, "grad_norm": 0.753381073474884, "learning_rate": 0.0001, "elbo": 3.50762, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.472202674173118}
{"step": 76265, "loss": 3.505484771728516, "grad_norm": 0.6994053721427917, "learning_rate": 0.0001, "elbo": 3.50548, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.472495894909688}
{"step": 76270, "loss": 3.503984069824219, "grad_norm": 0.7461299896240234, "learning_rate": 0.0001, "elbo": 3.50398, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4727891156462585}
{"step": 76275, "loss": 3.4946922302246093, "grad_norm": 0.861283540725708, "learning_rate": 0.0001, "elbo": 3.49469, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.473082336382829}
{"step": 76280, "loss": 3.5175743103027344, "grad_norm": 0.8157468438148499, "learning_rate": 0.0001, "elbo": 3.51757, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.473375557119399}
{"step": 76285, "loss": 3.514209747314453, "grad_norm": 0.7188019752502441, "learning_rate": 0.0001, "elbo": 3.51421, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.47366877785597}
{"step": 76290, "loss": 3.5175193786621093, "grad_norm": 0.7370134592056274, "learning_rate": 0.0001, "elbo": 3.51752, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.47396199859254}
{"step": 76295, "loss": 3.525978851318359, "grad_norm": 0.9300975799560547, "learning_rate": 0.0001, "elbo": 3.52598, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.474255219329111}
{"step": 76300, "loss": 3.5019821166992187, "grad_norm": 0.7539979219436646, "learning_rate": 0.0001, "elbo": 3.50198, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.474548440065681}
{"step": 76305, "loss": 3.516061782836914, "grad_norm": 0.6879957914352417, "learning_rate": 0.0001, "elbo": 3.51606, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.474841660802252}
{"step": 76310, "loss": 3.5094932556152343, "grad_norm": 0.6839496493339539, "learning_rate": 0.0001, "elbo": 3.50949, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.475134881538822}
{"step": 76315, "loss": 3.5039363861083985, "grad_norm": 0.7356796860694885, "learning_rate": 0.0001, "elbo": 3.50394, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.475428102275393}
{"step": 76320, "loss": 3.5174945831298827, "grad_norm": 0.7336319088935852, "learning_rate": 0.0001, "elbo": 3.51749, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.475721323011963}
{"step": 76325, "loss": 3.516770935058594, "grad_norm": 0.8058266639709473, "learning_rate": 0.0001, "elbo": 3.51677, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4760145437485335}
{"step": 76330, "loss": 3.5311737060546875, "grad_norm": 0.705876886844635, "learning_rate": 0.0001, "elbo": 3.53117, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.476307764485105}
{"step": 76335, "loss": 3.5018936157226563, "grad_norm": 0.677796483039856, "learning_rate": 0.0001, "elbo": 3.50189, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.476600985221675}
{"step": 76340, "loss": 3.512799072265625, "grad_norm": 0.6951708197593689, "learning_rate": 0.0001, "elbo": 3.5128, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.476894205958246}
{"step": 76345, "loss": 3.4965087890625, "grad_norm": 0.821354329586029, "learning_rate": 0.0001, "elbo": 3.49651, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.477187426694816}
{"step": 76350, "loss": 3.495956802368164, "grad_norm": 1.2699717283248901, "learning_rate": 0.0001, "elbo": 3.49596, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.477480647431387}
{"step": 76355, "loss": 3.522265625, "grad_norm": 0.6951881647109985, "learning_rate": 0.0001, "elbo": 3.52227, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.477773868167957}
{"step": 76360, "loss": 3.5039321899414064, "grad_norm": 0.7776698470115662, "learning_rate": 0.0001, "elbo": 3.50393, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4780670889045275}
{"step": 76365, "loss": 3.5268882751464843, "grad_norm": 0.7111011147499084, "learning_rate": 0.0001, "elbo": 3.52689, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.478360309641098}
{"step": 76365, "loss": 3.5268882751464843, "grad_norm": 0.7111011147499084, "learning_rate": 0.0001, "elbo": 3.52689, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.478360309641098}
{"step": 76370, "loss": 3.5070205688476563, "grad_norm": 0.8027349710464478, "learning_rate": 0.0001, "elbo": 3.50702, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4786535303776684}
{"step": 76370, "loss": 3.5070205688476563, "grad_norm": 0.8027349710464478, "learning_rate": 0.0001, "elbo": 3.50702, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4786535303776684}
{"step": 76375, "loss": 3.501298522949219, "grad_norm": 0.7192277312278748, "learning_rate": 0.0001, "elbo": 3.5013, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.478946751114239}
{"step": 76380, "loss": 3.4975841522216795, "grad_norm": 0.743562638759613, "learning_rate": 0.0001, "elbo": 3.49758, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.479239971850809}
{"step": 76385, "loss": 3.505606460571289, "grad_norm": 0.8498740792274475, "learning_rate": 0.0001, "elbo": 3.50561, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.47953319258738}
{"step": 76390, "loss": 3.522413635253906, "grad_norm": 1.0956943035125732, "learning_rate": 0.0001, "elbo": 3.52241, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.47982641332395}
{"step": 76395, "loss": 3.5123783111572267, "grad_norm": 0.7350074648857117, "learning_rate": 0.0001, "elbo": 3.51238, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.480119634060521}
{"step": 76400, "loss": 3.5011112213134767, "grad_norm": 0.7236309051513672, "learning_rate": 0.0001, "elbo": 3.50111, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.480412854797091}
{"step": 76405, "loss": 3.5078880310058596, "grad_norm": 0.6929017901420593, "learning_rate": 0.0001, "elbo": 3.50789, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.480706075533662}
{"step": 76410, "loss": 3.5026523590087892, "grad_norm": 0.7521820664405823, "learning_rate": 0.0001, "elbo": 3.50265, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.480999296270232}
{"step": 76415, "loss": 3.5306968688964844, "grad_norm": 0.7143911719322205, "learning_rate": 0.0001, "elbo": 3.5307, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4812925170068025}
{"step": 76420, "loss": 3.5088264465332033, "grad_norm": 0.7752344608306885, "learning_rate": 0.0001, "elbo": 3.50883, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.481585737743373}
{"step": 76425, "loss": 3.521473693847656, "grad_norm": 0.7945564985275269, "learning_rate": 0.0001, "elbo": 3.52147, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.481878958479943}
{"step": 76430, "loss": 3.517792892456055, "grad_norm": 1.0062024593353271, "learning_rate": 0.0001, "elbo": 3.51779, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.482172179216514}
{"step": 76435, "loss": 3.504137420654297, "grad_norm": 0.7623945474624634, "learning_rate": 0.0001, "elbo": 3.50414, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.482465399953084}
{"step": 76440, "loss": 3.4973819732666014, "grad_norm": 0.7428390979766846, "learning_rate": 0.0001, "elbo": 3.49738, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.482758620689655}
{"step": 76445, "loss": 3.501280975341797, "grad_norm": 0.739592432975769, "learning_rate": 0.0001, "elbo": 3.50128, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.483051841426225}
{"step": 76450, "loss": 3.506483459472656, "grad_norm": 0.7076727151870728, "learning_rate": 0.0001, "elbo": 3.50648, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.483345062162796}
{"step": 76455, "loss": 3.5058006286621093, "grad_norm": 0.7403756380081177, "learning_rate": 0.0001, "elbo": 3.5058, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.483638282899367}
{"step": 76460, "loss": 3.498813247680664, "grad_norm": 0.7412127256393433, "learning_rate": 0.0001, "elbo": 3.49881, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4839315036359375}
{"step": 76465, "loss": 3.517334747314453, "grad_norm": 0.6834481358528137, "learning_rate": 0.0001, "elbo": 3.51733, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.484224724372508}
{"step": 76470, "loss": 3.5027801513671877, "grad_norm": 0.7238460779190063, "learning_rate": 0.0001, "elbo": 3.50278, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.484517945109078}
{"step": 76475, "loss": 3.5196632385253905, "grad_norm": 0.7480072975158691, "learning_rate": 0.0001, "elbo": 3.51966, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.484811165845649}
{"step": 76480, "loss": 3.513580322265625, "grad_norm": 0.770666778087616, "learning_rate": 0.0001, "elbo": 3.51358, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.485104386582219}
{"step": 76485, "loss": 3.5180381774902343, "grad_norm": 0.7019604444503784, "learning_rate": 0.0001, "elbo": 3.51804, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.48539760731879}
{"step": 76490, "loss": 3.52115478515625, "grad_norm": 0.6975023150444031, "learning_rate": 0.0001, "elbo": 3.52115, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.48569082805536}
{"step": 76495, "loss": 3.519769287109375, "grad_norm": 0.7469876408576965, "learning_rate": 0.0001, "elbo": 3.51977, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.485984048791931}
{"step": 76500, "loss": 3.5137649536132813, "grad_norm": 0.8025696277618408, "learning_rate": 0.0001, "elbo": 3.51376, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.486277269528501}
{"step": 76505, "loss": 3.5289794921875, "grad_norm": 0.6836710572242737, "learning_rate": 0.0001, "elbo": 3.52898, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.486570490265072}
{"step": 76510, "loss": 3.4937671661376952, "grad_norm": 0.7232559323310852, "learning_rate": 0.0001, "elbo": 3.49377, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.486863711001642}
{"step": 76515, "loss": 3.5209259033203124, "grad_norm": 0.7409156560897827, "learning_rate": 0.0001, "elbo": 3.52093, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4871569317382125}
{"step": 76520, "loss": 3.503678894042969, "grad_norm": 0.9556098580360413, "learning_rate": 0.0001, "elbo": 3.50368, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.487450152474783}
{"step": 76525, "loss": 3.530661773681641, "grad_norm": 0.6652798652648926, "learning_rate": 0.0001, "elbo": 3.53066, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.487743373211353}
{"step": 76530, "loss": 3.4941665649414064, "grad_norm": 0.8379178047180176, "learning_rate": 0.0001, "elbo": 3.49417, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.488036593947924}
{"step": 76535, "loss": 3.513477325439453, "grad_norm": 1.1930434703826904, "learning_rate": 0.0001, "elbo": 3.51348, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.488329814684494}
{"step": 76540, "loss": 3.4986915588378906, "grad_norm": 0.7305260896682739, "learning_rate": 0.0001, "elbo": 3.49869, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.488623035421065}
{"step": 76545, "loss": 3.5075950622558594, "grad_norm": 0.8328405618667603, "learning_rate": 0.0001, "elbo": 3.50759, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.488916256157635}
{"step": 76550, "loss": 3.5028099060058593, "grad_norm": 0.6901484131813049, "learning_rate": 0.0001, "elbo": 3.50281, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.489209476894206}
{"step": 76555, "loss": 3.503063201904297, "grad_norm": 0.716083824634552, "learning_rate": 0.0001, "elbo": 3.50306, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.489502697630776}
{"step": 76560, "loss": 3.5097686767578127, "grad_norm": 0.969302773475647, "learning_rate": 0.0001, "elbo": 3.50977, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.489795918367347}
{"step": 76565, "loss": 3.5140312194824217, "grad_norm": 0.7432448863983154, "learning_rate": 0.0001, "elbo": 3.51403, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.490089139103917}
{"step": 76570, "loss": 3.5158157348632812, "grad_norm": 0.7154826521873474, "learning_rate": 0.0001, "elbo": 3.51582, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.490382359840488}
{"step": 76575, "loss": 3.501629638671875, "grad_norm": 0.6814610958099365, "learning_rate": 0.0001, "elbo": 3.50163, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.490675580577058}
{"step": 76580, "loss": 3.4890228271484376, "grad_norm": 0.728435754776001, "learning_rate": 0.0001, "elbo": 3.48902, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.490968801313629}
{"step": 76585, "loss": 3.512158203125, "grad_norm": 0.7047481536865234, "learning_rate": 0.0001, "elbo": 3.51216, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4912620220502}
{"step": 76590, "loss": 3.5208820343017577, "grad_norm": 0.7083985209465027, "learning_rate": 0.0001, "elbo": 3.52088, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.49155524278677}
{"step": 76595, "loss": 3.523479461669922, "grad_norm": 0.9177606105804443, "learning_rate": 0.0001, "elbo": 3.52348, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.491848463523341}
{"step": 76600, "loss": 3.5183856964111326, "grad_norm": 0.7079989910125732, "learning_rate": 0.0001, "elbo": 3.51839, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.492141684259911}
{"step": 76605, "loss": 3.505901336669922, "grad_norm": 0.7072926163673401, "learning_rate": 0.0001, "elbo": 3.5059, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.492434904996482}
{"step": 76610, "loss": 3.510102462768555, "grad_norm": 0.9598580002784729, "learning_rate": 0.0001, "elbo": 3.5101, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.492728125733052}
{"step": 76615, "loss": 3.5233665466308595, "grad_norm": 0.7726212739944458, "learning_rate": 0.0001, "elbo": 3.52337, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4930213464696225}
{"step": 76620, "loss": 3.5109931945800783, "grad_norm": 0.7538571953773499, "learning_rate": 0.0001, "elbo": 3.51099, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.493314567206193}
{"step": 76625, "loss": 3.5378189086914062, "grad_norm": 1.015123963356018, "learning_rate": 0.0001, "elbo": 3.53782, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.493607787942763}
{"step": 76630, "loss": 3.5172828674316405, "grad_norm": 0.8507131934165955, "learning_rate": 0.0001, "elbo": 3.51728, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.493901008679334}
{"step": 76635, "loss": 3.516147994995117, "grad_norm": 0.7719905972480774, "learning_rate": 0.0001, "elbo": 3.51615, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.494194229415904}
{"step": 76635, "loss": 3.516147994995117, "grad_norm": 0.7719905972480774, "learning_rate": 0.0001, "elbo": 3.51615, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.494194229415904}
{"step": 76640, "loss": 3.5158916473388673, "grad_norm": 1.3056542873382568, "learning_rate": 0.0001, "elbo": 3.51589, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.494487450152475}
{"step": 76645, "loss": 3.4995826721191405, "grad_norm": 0.7282090187072754, "learning_rate": 0.0001, "elbo": 3.49958, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.494780670889045}
{"step": 76650, "loss": 3.5125598907470703, "grad_norm": 0.780730128288269, "learning_rate": 0.0001, "elbo": 3.51256, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.495073891625616}
{"step": 76655, "loss": 3.524669647216797, "grad_norm": 0.6619304418563843, "learning_rate": 0.0001, "elbo": 3.52467, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.495367112362186}
{"step": 76660, "loss": 3.514490509033203, "grad_norm": 0.6985988616943359, "learning_rate": 0.0001, "elbo": 3.51449, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.495660333098757}
{"step": 76665, "loss": 3.520357131958008, "grad_norm": 0.7006269693374634, "learning_rate": 0.0001, "elbo": 3.52036, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.495953553835327}
{"step": 76670, "loss": 3.500368118286133, "grad_norm": 0.698922336101532, "learning_rate": 0.0001, "elbo": 3.50037, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4962467745718975}
{"step": 76675, "loss": 3.5023704528808595, "grad_norm": 0.6798631548881531, "learning_rate": 0.0001, "elbo": 3.50237, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.496539995308468}
{"step": 76680, "loss": 3.488341522216797, "grad_norm": 0.7058756351470947, "learning_rate": 0.0001, "elbo": 3.48834, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.496833216045038}
{"step": 76685, "loss": 3.492275619506836, "grad_norm": 0.7060806155204773, "learning_rate": 0.0001, "elbo": 3.49228, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.497126436781609}
{"step": 76690, "loss": 3.50952033996582, "grad_norm": 0.7728673815727234, "learning_rate": 0.0001, "elbo": 3.50952, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.497419657518179}
{"step": 76695, "loss": 3.5116783142089845, "grad_norm": 0.7353109121322632, "learning_rate": 0.0001, "elbo": 3.51168, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.497712878254751}
{"step": 76700, "loss": 3.4993080139160155, "grad_norm": 0.7336726188659668, "learning_rate": 0.0001, "elbo": 3.49931, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.49800609899132}
{"step": 76705, "loss": 3.523512268066406, "grad_norm": 0.8775678873062134, "learning_rate": 0.0001, "elbo": 3.52351, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4982993197278915}
{"step": 76710, "loss": 3.5096168518066406, "grad_norm": 0.7392969131469727, "learning_rate": 0.0001, "elbo": 3.50962, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.498592540464462}
{"step": 76715, "loss": 3.520557403564453, "grad_norm": 0.7369555830955505, "learning_rate": 0.0001, "elbo": 3.52056, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.4988857612010325}
{"step": 76720, "loss": 3.500357437133789, "grad_norm": 0.7261866331100464, "learning_rate": 0.0001, "elbo": 3.50036, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.499178981937603}
{"step": 76725, "loss": 3.5239028930664062, "grad_norm": 0.6817086935043335, "learning_rate": 0.0001, "elbo": 3.5239, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.499472202674173}
{"step": 76730, "loss": 3.4985816955566404, "grad_norm": 0.6702409386634827, "learning_rate": 0.0001, "elbo": 3.49858, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.499765423410744}
{"step": 76735, "loss": 3.5126373291015627, "grad_norm": 0.7160630226135254, "learning_rate": 0.0001, "elbo": 3.51264, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.500058644147314}
{"step": 76740, "loss": 3.5044918060302734, "grad_norm": 0.875606894493103, "learning_rate": 0.0001, "elbo": 3.50449, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.500351864883885}
{"step": 76745, "loss": 3.514044189453125, "grad_norm": 0.7985852360725403, "learning_rate": 0.0001, "elbo": 3.51404, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.500645085620455}
{"step": 76750, "loss": 3.5095588684082033, "grad_norm": 0.7293421626091003, "learning_rate": 0.0001, "elbo": 3.50956, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.500938306357026}
{"step": 76750, "loss": 3.5095588684082033, "grad_norm": 0.7293421626091003, "learning_rate": 0.0001, "elbo": 3.50956, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.500938306357026}
{"step": 76755, "loss": 3.514649200439453, "grad_norm": 0.7279789447784424, "learning_rate": 0.0001, "elbo": 3.51465, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.501231527093596}
{"step": 76760, "loss": 3.5180923461914064, "grad_norm": 0.6933088898658752, "learning_rate": 0.0001, "elbo": 3.51809, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5015247478301665}
{"step": 76765, "loss": 3.500062942504883, "grad_norm": 1.0776054859161377, "learning_rate": 0.0001, "elbo": 3.50006, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.501817968566737}
{"step": 76770, "loss": 3.5005111694335938, "grad_norm": 0.6979413628578186, "learning_rate": 0.0001, "elbo": 3.50051, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5021111893033074}
{"step": 76775, "loss": 3.504291534423828, "grad_norm": 0.7048174738883972, "learning_rate": 0.0001, "elbo": 3.50429, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.502404410039878}
{"step": 76780, "loss": 3.511347198486328, "grad_norm": 0.6862559914588928, "learning_rate": 0.0001, "elbo": 3.51135, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.502697630776448}
{"step": 76785, "loss": 3.511327362060547, "grad_norm": 0.7660694718360901, "learning_rate": 0.0001, "elbo": 3.51133, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.502990851513019}
{"step": 76790, "loss": 3.5140716552734377, "grad_norm": 0.8180075287818909, "learning_rate": 0.0001, "elbo": 3.51407, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.503284072249589}
{"step": 76795, "loss": 3.5089820861816405, "grad_norm": 0.707169234752655, "learning_rate": 0.0001, "elbo": 3.50898, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.50357729298616}
{"step": 76800, "loss": 3.5101478576660154, "grad_norm": 0.7481217384338379, "learning_rate": 0.0001, "elbo": 3.51015, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.50387051372273}
{"step": 76805, "loss": 3.517145538330078, "grad_norm": 0.7236731648445129, "learning_rate": 0.0001, "elbo": 3.51715, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.504163734459301}
{"step": 76810, "loss": 3.5053741455078127, "grad_norm": 0.8267025947570801, "learning_rate": 0.0001, "elbo": 3.50537, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.504456955195871}
{"step": 76815, "loss": 3.5132484436035156, "grad_norm": 1.0325932502746582, "learning_rate": 0.0001, "elbo": 3.51325, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5047501759324415}
{"step": 76820, "loss": 3.5204189300537108, "grad_norm": 0.7959835529327393, "learning_rate": 0.0001, "elbo": 3.52042, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.505043396669013}
{"step": 76825, "loss": 3.4969276428222655, "grad_norm": 0.8731761574745178, "learning_rate": 0.0001, "elbo": 3.49693, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5053366174055824}
{"step": 76830, "loss": 3.507244873046875, "grad_norm": 0.6935184597969055, "learning_rate": 0.0001, "elbo": 3.50724, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.505629838142154}
{"step": 76835, "loss": 3.506454086303711, "grad_norm": 0.6809908747673035, "learning_rate": 0.0001, "elbo": 3.50645, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.505923058878724}
{"step": 76840, "loss": 3.5233837127685548, "grad_norm": 0.6935476660728455, "learning_rate": 0.0001, "elbo": 3.52338, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.506216279615295}
{"step": 76845, "loss": 3.5324638366699217, "grad_norm": 0.7172329425811768, "learning_rate": 0.0001, "elbo": 3.53246, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.506509500351865}
{"step": 76850, "loss": 3.5163242340087892, "grad_norm": 0.6904316544532776, "learning_rate": 0.0001, "elbo": 3.51632, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.506802721088436}
{"step": 76855, "loss": 3.5171890258789062, "grad_norm": 0.7462272047996521, "learning_rate": 0.0001, "elbo": 3.51719, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.507095941825006}
{"step": 76860, "loss": 3.512464141845703, "grad_norm": 0.6741729974746704, "learning_rate": 0.0001, "elbo": 3.51246, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5073891625615765}
{"step": 76865, "loss": 3.51612548828125, "grad_norm": 0.6988689303398132, "learning_rate": 0.0001, "elbo": 3.51613, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.507682383298147}
{"step": 76870, "loss": 3.5019241333007813, "grad_norm": 0.6884700059890747, "learning_rate": 0.0001, "elbo": 3.50192, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.507975604034717}
{"step": 76875, "loss": 3.5153228759765627, "grad_norm": 0.7041682600975037, "learning_rate": 0.0001, "elbo": 3.51532, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.508268824771288}
{"step": 76880, "loss": 3.524100494384766, "grad_norm": 0.6998571753501892, "learning_rate": 0.0001, "elbo": 3.5241, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.508562045507858}
{"step": 76885, "loss": 3.499021530151367, "grad_norm": 0.8118394017219543, "learning_rate": 0.0001, "elbo": 3.49902, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.508855266244429}
{"step": 76890, "loss": 3.5068668365478515, "grad_norm": 0.7172042727470398, "learning_rate": 0.0001, "elbo": 3.50687, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.509148486980999}
{"step": 76895, "loss": 3.499104690551758, "grad_norm": 0.7046072483062744, "learning_rate": 0.0001, "elbo": 3.4991, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.50944170771757}
{"step": 76900, "loss": 3.508809280395508, "grad_norm": 0.6962236762046814, "learning_rate": 0.0001, "elbo": 3.50881, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.50973492845414}
{"step": 76900, "loss": 3.508809280395508, "grad_norm": 0.6962236762046814, "learning_rate": 0.0001, "elbo": 3.50881, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.50973492845414}
{"step": 76905, "loss": 3.528014373779297, "grad_norm": 0.6982805728912354, "learning_rate": 0.0001, "elbo": 3.52801, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.510028149190711}
{"step": 76910, "loss": 3.5141185760498046, "grad_norm": 0.7389975190162659, "learning_rate": 0.0001, "elbo": 3.51412, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.510321369927281}
{"step": 76915, "loss": 3.509529876708984, "grad_norm": 1.142928123474121, "learning_rate": 0.0001, "elbo": 3.50953, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5106145906638515}
{"step": 76920, "loss": 3.5063095092773438, "grad_norm": 0.8812909126281738, "learning_rate": 0.0001, "elbo": 3.50631, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.510907811400422}
{"step": 76925, "loss": 3.5211395263671874, "grad_norm": 0.688099205493927, "learning_rate": 0.0001, "elbo": 3.52114, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.511201032136992}
{"step": 76930, "loss": 3.516265869140625, "grad_norm": 0.7114261388778687, "learning_rate": 0.0001, "elbo": 3.51627, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.511494252873563}
{"step": 76935, "loss": 3.509600067138672, "grad_norm": 0.6782441139221191, "learning_rate": 0.0001, "elbo": 3.5096, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.511787473610133}
{"step": 76940, "loss": 3.49971923828125, "grad_norm": 0.6903418898582458, "learning_rate": 0.0001, "elbo": 3.49972, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.512080694346704}
{"step": 76945, "loss": 3.4986545562744142, "grad_norm": 0.9062210321426392, "learning_rate": 0.0001, "elbo": 3.49865, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.512373915083275}
{"step": 76950, "loss": 3.509374237060547, "grad_norm": 0.7088158130645752, "learning_rate": 0.0001, "elbo": 3.50937, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.512667135819845}
{"step": 76955, "loss": 3.5204116821289064, "grad_norm": 0.7225848436355591, "learning_rate": 0.0001, "elbo": 3.52041, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.512960356556416}
{"step": 76960, "loss": 3.515353775024414, "grad_norm": 0.7337468266487122, "learning_rate": 0.0001, "elbo": 3.51535, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5132535772929865}
{"step": 76965, "loss": 3.5194698333740235, "grad_norm": 0.7284289002418518, "learning_rate": 0.0001, "elbo": 3.51947, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.513546798029557}
{"step": 76970, "loss": 3.5242076873779298, "grad_norm": 0.6984471678733826, "learning_rate": 0.0001, "elbo": 3.52421, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.513840018766127}
{"step": 76975, "loss": 3.5116146087646483, "grad_norm": 1.2257716655731201, "learning_rate": 0.0001, "elbo": 3.51161, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.514133239502698}
{"step": 76975, "loss": 3.5116146087646483, "grad_norm": 1.2257716655731201, "learning_rate": 0.0001, "elbo": 3.51161, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.514133239502698}
{"step": 76980, "loss": 3.5316261291503905, "grad_norm": 0.6946609020233154, "learning_rate": 0.0001, "elbo": 3.53163, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.514426460239268}
{"step": 76985, "loss": 3.504781723022461, "grad_norm": 0.9320554733276367, "learning_rate": 0.0001, "elbo": 3.50478, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.514719680975839}
{"step": 76990, "loss": 3.507465362548828, "grad_norm": 0.6645298004150391, "learning_rate": 0.0001, "elbo": 3.50747, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.515012901712409}
{"step": 76995, "loss": 3.5167694091796875, "grad_norm": 0.7609760165214539, "learning_rate": 0.0001, "elbo": 3.51677, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.51530612244898}
{"step": 77000, "loss": 3.500775146484375, "grad_norm": 0.7241359353065491, "learning_rate": 0.0001, "elbo": 3.50078, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.51559934318555}
{"step": 77005, "loss": 3.513505554199219, "grad_norm": 0.7305540442466736, "learning_rate": 0.0001, "elbo": 3.51351, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.515892563922121}
{"step": 77010, "loss": 3.516715240478516, "grad_norm": 0.9419698715209961, "learning_rate": 0.0001, "elbo": 3.51672, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.516185784658691}
{"step": 77015, "loss": 3.526234436035156, "grad_norm": 0.8532364368438721, "learning_rate": 0.0001, "elbo": 3.52623, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5164790053952615}
{"step": 77015, "loss": 3.526234436035156, "grad_norm": 0.8532364368438721, "learning_rate": 0.0001, "elbo": 3.52623, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5164790053952615}
{"step": 77020, "loss": 3.527251052856445, "grad_norm": 0.7238015532493591, "learning_rate": 0.0001, "elbo": 3.52725, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.516772226131832}
{"step": 77025, "loss": 3.510034942626953, "grad_norm": 0.7691675424575806, "learning_rate": 0.0001, "elbo": 3.51004, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.517065446868402}
{"step": 77030, "loss": 3.509487533569336, "grad_norm": 0.7139210104942322, "learning_rate": 0.0001, "elbo": 3.50949, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.517358667604973}
{"step": 77035, "loss": 3.4968875885009765, "grad_norm": 0.7157754898071289, "learning_rate": 0.0001, "elbo": 3.49689, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.517651888341543}
{"step": 77040, "loss": 3.507562255859375, "grad_norm": 0.6903956532478333, "learning_rate": 0.0001, "elbo": 3.50756, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.517945109078114}
{"step": 77045, "loss": 3.5326217651367187, "grad_norm": 0.745195209980011, "learning_rate": 0.0001, "elbo": 3.53262, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.518238329814684}
{"step": 77050, "loss": 3.499785232543945, "grad_norm": 0.7518689632415771, "learning_rate": 0.0001, "elbo": 3.49979, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.518531550551255}
{"step": 77055, "loss": 3.5056602478027346, "grad_norm": 0.674569845199585, "learning_rate": 0.0001, "elbo": 3.50566, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.518824771287825}
{"step": 77060, "loss": 3.5070846557617186, "grad_norm": 0.7093705534934998, "learning_rate": 0.0001, "elbo": 3.50708, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5191179920243965}
{"step": 77060, "loss": 3.5070846557617186, "grad_norm": 0.7093705534934998, "learning_rate": 0.0001, "elbo": 3.50708, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5191179920243965}
{"step": 77065, "loss": 3.527372360229492, "grad_norm": 0.7050554752349854, "learning_rate": 0.0001, "elbo": 3.52737, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.519411212760966}
{"step": 77070, "loss": 3.49316520690918, "grad_norm": 0.7205063700675964, "learning_rate": 0.0001, "elbo": 3.49317, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.519704433497537}
{"step": 77070, "loss": 3.49316520690918, "grad_norm": 0.7205063700675964, "learning_rate": 0.0001, "elbo": 3.49317, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.519704433497537}
{"step": 77075, "loss": 3.513936233520508, "grad_norm": 0.7494041919708252, "learning_rate": 0.0001, "elbo": 3.51394, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.519997654234108}
{"step": 77080, "loss": 3.513819122314453, "grad_norm": 0.7002464532852173, "learning_rate": 0.0001, "elbo": 3.51382, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.520290874970678}
{"step": 77080, "loss": 3.513819122314453, "grad_norm": 0.7002464532852173, "learning_rate": 0.0001, "elbo": 3.51382, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.520290874970678}
{"step": 77085, "loss": 3.511973571777344, "grad_norm": 0.7262354493141174, "learning_rate": 0.0001, "elbo": 3.51197, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.520584095707249}
{"step": 77090, "loss": 3.5158924102783202, "grad_norm": 0.7512372732162476, "learning_rate": 0.0001, "elbo": 3.51589, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.520877316443819}
{"step": 77095, "loss": 3.5088218688964843, "grad_norm": 0.740583062171936, "learning_rate": 0.0001, "elbo": 3.50882, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.52117053718039}
{"step": 77100, "loss": 3.504264068603516, "grad_norm": 0.7122786641120911, "learning_rate": 0.0001, "elbo": 3.50426, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.52146375791696}
{"step": 77105, "loss": 3.5122188568115233, "grad_norm": 0.7494135499000549, "learning_rate": 0.0001, "elbo": 3.51222, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5217569786535305}
{"step": 77110, "loss": 3.512067413330078, "grad_norm": 0.6888070702552795, "learning_rate": 0.0001, "elbo": 3.51207, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.522050199390101}
{"step": 77115, "loss": 3.5096710205078123, "grad_norm": 0.693449854850769, "learning_rate": 0.0001, "elbo": 3.50967, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5223434201266715}
{"step": 77120, "loss": 3.486350250244141, "grad_norm": 0.6955979466438293, "learning_rate": 0.0001, "elbo": 3.48635, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.522636640863242}
{"step": 77125, "loss": 3.5184562683105467, "grad_norm": 0.7162903547286987, "learning_rate": 0.0001, "elbo": 3.51846, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.522929861599812}
{"step": 77130, "loss": 3.5215774536132813, "grad_norm": 0.7121332287788391, "learning_rate": 0.0001, "elbo": 3.52158, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.523223082336383}
{"step": 77135, "loss": 3.507914733886719, "grad_norm": 0.7405914664268494, "learning_rate": 0.0001, "elbo": 3.50791, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.523516303072953}
{"step": 77140, "loss": 3.4974021911621094, "grad_norm": 0.7237595319747925, "learning_rate": 0.0001, "elbo": 3.4974, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.523809523809524}
{"step": 77145, "loss": 3.5049678802490236, "grad_norm": 0.7325975298881531, "learning_rate": 0.0001, "elbo": 3.50497, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.524102744546094}
{"step": 77150, "loss": 3.507456970214844, "grad_norm": 0.7163906097412109, "learning_rate": 0.0001, "elbo": 3.50746, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.524395965282665}
{"step": 77155, "loss": 3.495412063598633, "grad_norm": 0.7094535827636719, "learning_rate": 0.0001, "elbo": 3.49541, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.524689186019235}
{"step": 77160, "loss": 3.5149662017822267, "grad_norm": 0.7436463832855225, "learning_rate": 0.0001, "elbo": 3.51497, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5249824067558055}
{"step": 77165, "loss": 3.518489456176758, "grad_norm": 0.7965438961982727, "learning_rate": 0.0001, "elbo": 3.51849, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.525275627492376}
{"step": 77170, "loss": 3.510389709472656, "grad_norm": 0.7338392734527588, "learning_rate": 0.0001, "elbo": 3.51039, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5255688482289465}
{"step": 77175, "loss": 3.5112510681152345, "grad_norm": 0.6836997866630554, "learning_rate": 0.0001, "elbo": 3.51125, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.525862068965517}
{"step": 77180, "loss": 3.528264617919922, "grad_norm": 0.6920868754386902, "learning_rate": 0.0001, "elbo": 3.52826, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.526155289702087}
{"step": 77180, "loss": 3.528264617919922, "grad_norm": 0.6920868754386902, "learning_rate": 0.0001, "elbo": 3.52826, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.526155289702087}
{"step": 77185, "loss": 3.501421356201172, "grad_norm": 0.72662353515625, "learning_rate": 0.0001, "elbo": 3.50142, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.526448510438659}
{"step": 77190, "loss": 3.5097023010253907, "grad_norm": 0.7757880091667175, "learning_rate": 0.0001, "elbo": 3.5097, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.526741731175228}
{"step": 77195, "loss": 3.4977676391601564, "grad_norm": 0.7233269214630127, "learning_rate": 0.0001, "elbo": 3.49777, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5270349519118}
{"step": 77200, "loss": 3.5036964416503906, "grad_norm": 0.8590745329856873, "learning_rate": 0.0001, "elbo": 3.5037, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.52732817264837}
{"step": 77205, "loss": 3.4931259155273438, "grad_norm": 0.7092183828353882, "learning_rate": 0.0001, "elbo": 3.49313, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5276213933849405}
{"step": 77210, "loss": 3.4935012817382813, "grad_norm": 1.0178731679916382, "learning_rate": 0.0001, "elbo": 3.4935, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.527914614121511}
{"step": 77210, "loss": 3.4935012817382813, "grad_norm": 1.0178731679916382, "learning_rate": 0.0001, "elbo": 3.4935, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.527914614121511}
{"step": 77215, "loss": 3.508536529541016, "grad_norm": 0.6549215316772461, "learning_rate": 0.0001, "elbo": 3.50854, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.528207834858081}
{"step": 77220, "loss": 3.512430191040039, "grad_norm": 0.7276098132133484, "learning_rate": 0.0001, "elbo": 3.51243, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.528501055594652}
{"step": 77225, "loss": 3.5062240600585937, "grad_norm": 0.6776543855667114, "learning_rate": 0.0001, "elbo": 3.50622, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.528794276331222}
{"step": 77230, "loss": 3.523906707763672, "grad_norm": 0.6898388266563416, "learning_rate": 0.0001, "elbo": 3.52391, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.529087497067793}
{"step": 77235, "loss": 3.4992095947265627, "grad_norm": 0.7062645554542542, "learning_rate": 0.0001, "elbo": 3.49921, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.529380717804363}
{"step": 77240, "loss": 3.5051666259765626, "grad_norm": 0.6854614019393921, "learning_rate": 0.0001, "elbo": 3.50517, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.529673938540934}
{"step": 77245, "loss": 3.52078857421875, "grad_norm": 1.4653329849243164, "learning_rate": 0.0001, "elbo": 3.52079, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.529967159277504}
{"step": 77250, "loss": 3.5066879272460936, "grad_norm": 0.6916836500167847, "learning_rate": 0.0001, "elbo": 3.50669, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.530260380014075}
{"step": 77255, "loss": 3.5134456634521483, "grad_norm": 0.7172386050224304, "learning_rate": 0.0001, "elbo": 3.51345, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.530553600750645}
{"step": 77260, "loss": 3.509724426269531, "grad_norm": 0.7224082350730896, "learning_rate": 0.0001, "elbo": 3.50972, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5308468214872155}
{"step": 77265, "loss": 3.5083194732666017, "grad_norm": 0.7560349702835083, "learning_rate": 0.0001, "elbo": 3.50832, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.531140042223786}
{"step": 77270, "loss": 3.498228073120117, "grad_norm": 0.6725802421569824, "learning_rate": 0.0001, "elbo": 3.49823, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.531433262960356}
{"step": 77275, "loss": 3.5083465576171875, "grad_norm": 0.7024297714233398, "learning_rate": 0.0001, "elbo": 3.50835, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.531726483696927}
{"step": 77280, "loss": 3.5044528961181642, "grad_norm": 0.6871629953384399, "learning_rate": 0.0001, "elbo": 3.50445, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.532019704433497}
{"step": 77285, "loss": 3.507823944091797, "grad_norm": 0.8998180031776428, "learning_rate": 0.0001, "elbo": 3.50782, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.532312925170068}
{"step": 77290, "loss": 3.5141334533691406, "grad_norm": 0.6828287243843079, "learning_rate": 0.0001, "elbo": 3.51413, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.532606145906638}
{"step": 77295, "loss": 3.518943023681641, "grad_norm": 0.8337611556053162, "learning_rate": 0.0001, "elbo": 3.51894, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.532899366643209}
{"step": 77300, "loss": 3.506983184814453, "grad_norm": 0.7402724623680115, "learning_rate": 0.0001, "elbo": 3.50698, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.533192587379779}
{"step": 77305, "loss": 3.5067489624023436, "grad_norm": 0.6827004551887512, "learning_rate": 0.0001, "elbo": 3.50675, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.53348580811635}
{"step": 77310, "loss": 3.5070068359375, "grad_norm": 0.6808591485023499, "learning_rate": 0.0001, "elbo": 3.50701, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.533779028852921}
{"step": 77315, "loss": 3.5093963623046873, "grad_norm": 0.6866750717163086, "learning_rate": 0.0001, "elbo": 3.5094, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5340722495894905}
{"step": 77320, "loss": 3.513522720336914, "grad_norm": 0.6819763779640198, "learning_rate": 0.0001, "elbo": 3.51352, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.534365470326062}
{"step": 77325, "loss": 3.505519485473633, "grad_norm": 0.8186436891555786, "learning_rate": 0.0001, "elbo": 3.50552, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.534658691062632}
{"step": 77330, "loss": 3.5065025329589843, "grad_norm": 1.1999924182891846, "learning_rate": 0.0001, "elbo": 3.5065, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.534951911799203}
{"step": 77335, "loss": 3.511858367919922, "grad_norm": 0.7495024800300598, "learning_rate": 0.0001, "elbo": 3.51186, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.535245132535773}
{"step": 77340, "loss": 3.5083648681640627, "grad_norm": 0.8147763609886169, "learning_rate": 0.0001, "elbo": 3.50837, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.535538353272344}
{"step": 77345, "loss": 3.4967437744140626, "grad_norm": 0.677914023399353, "learning_rate": 0.0001, "elbo": 3.49674, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.535831574008914}
{"step": 77350, "loss": 3.5229549407958984, "grad_norm": 0.6830534338951111, "learning_rate": 0.0001, "elbo": 3.52295, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.536124794745485}
{"step": 77355, "loss": 3.5015792846679688, "grad_norm": 0.670303463935852, "learning_rate": 0.0001, "elbo": 3.50158, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.536418015482055}
{"step": 77360, "loss": 3.5068824768066404, "grad_norm": 0.7529987692832947, "learning_rate": 0.0001, "elbo": 3.50688, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5367112362186255}
{"step": 77365, "loss": 3.49537353515625, "grad_norm": 0.8736445307731628, "learning_rate": 0.0001, "elbo": 3.49537, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.537004456955196}
{"step": 77370, "loss": 3.5023094177246095, "grad_norm": 0.7768110632896423, "learning_rate": 0.0001, "elbo": 3.50231, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.537297677691766}
{"step": 77375, "loss": 3.524741363525391, "grad_norm": 0.7503593564033508, "learning_rate": 0.0001, "elbo": 3.52474, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.537590898428337}
{"step": 77380, "loss": 3.4863685607910155, "grad_norm": 0.7381894588470459, "learning_rate": 0.0001, "elbo": 3.48637, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.537884119164907}
{"step": 77385, "loss": 3.517177963256836, "grad_norm": 0.7191596031188965, "learning_rate": 0.0001, "elbo": 3.51718, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.538177339901478}
{"step": 77390, "loss": 3.522313690185547, "grad_norm": 0.7102220058441162, "learning_rate": 0.0001, "elbo": 3.52231, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.538470560638048}
{"step": 77395, "loss": 3.539864349365234, "grad_norm": 0.7374573945999146, "learning_rate": 0.0001, "elbo": 3.53986, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.538763781374619}
{"step": 77400, "loss": 3.515130615234375, "grad_norm": 0.6847487092018127, "learning_rate": 0.0001, "elbo": 3.51513, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.539057002111189}
{"step": 77405, "loss": 3.508827972412109, "grad_norm": 0.736167848110199, "learning_rate": 0.0001, "elbo": 3.50883, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.53935022284776}
{"step": 77410, "loss": 3.51490478515625, "grad_norm": 0.6539896130561829, "learning_rate": 0.0001, "elbo": 3.5149, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.53964344358433}
{"step": 77415, "loss": 3.5216659545898437, "grad_norm": 0.7171700596809387, "learning_rate": 0.0001, "elbo": 3.52167, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5399366643209005}
{"step": 77420, "loss": 3.506616973876953, "grad_norm": 0.6845569014549255, "learning_rate": 0.0001, "elbo": 3.50662, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.540229885057471}
{"step": 77425, "loss": 3.508441162109375, "grad_norm": 0.7575859427452087, "learning_rate": 0.0001, "elbo": 3.50844, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.540523105794041}
{"step": 77430, "loss": 3.5039878845214845, "grad_norm": 0.7844319939613342, "learning_rate": 0.0001, "elbo": 3.50399, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.540816326530612}
{"step": 77435, "loss": 3.516027069091797, "grad_norm": 0.7191071510314941, "learning_rate": 0.0001, "elbo": 3.51603, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.541109547267183}
{"step": 77440, "loss": 3.509539794921875, "grad_norm": 0.7433908581733704, "learning_rate": 0.0001, "elbo": 3.50954, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.541402768003753}
{"step": 77445, "loss": 3.5158927917480467, "grad_norm": 0.7096583247184753, "learning_rate": 0.0001, "elbo": 3.51589, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.541695988740324}
{"step": 77450, "loss": 3.4996307373046873, "grad_norm": 0.708698570728302, "learning_rate": 0.0001, "elbo": 3.49963, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5419892094768946}
{"step": 77455, "loss": 3.5002708435058594, "grad_norm": 0.7547594904899597, "learning_rate": 0.0001, "elbo": 3.50027, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.542282430213465}
{"step": 77460, "loss": 3.519205093383789, "grad_norm": 0.6694480180740356, "learning_rate": 0.0001, "elbo": 3.51921, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5425756509500355}
{"step": 77465, "loss": 3.521265411376953, "grad_norm": 0.7568486928939819, "learning_rate": 0.0001, "elbo": 3.52127, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.542868871686606}
{"step": 77470, "loss": 3.5178596496582033, "grad_norm": 0.7640539407730103, "learning_rate": 0.0001, "elbo": 3.51786, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.543162092423176}
{"step": 77475, "loss": 3.509130859375, "grad_norm": 0.747134804725647, "learning_rate": 0.0001, "elbo": 3.50913, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.543455313159747}
{"step": 77480, "loss": 3.509527587890625, "grad_norm": 0.6957683563232422, "learning_rate": 0.0001, "elbo": 3.50953, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.543748533896317}
{"step": 77485, "loss": 3.5171913146972655, "grad_norm": 0.6908571124076843, "learning_rate": 0.0001, "elbo": 3.51719, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.544041754632888}
{"step": 77490, "loss": 3.5045093536376952, "grad_norm": 0.7592782974243164, "learning_rate": 0.0001, "elbo": 3.50451, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.544334975369458}
{"step": 77495, "loss": 3.512466049194336, "grad_norm": 0.7572650909423828, "learning_rate": 0.0001, "elbo": 3.51247, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.544628196106029}
{"step": 77500, "loss": 3.504058074951172, "grad_norm": 0.6884270310401917, "learning_rate": 0.0001, "elbo": 3.50406, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.544921416842599}
{"step": 77505, "loss": 3.515100860595703, "grad_norm": 0.7674673199653625, "learning_rate": 0.0001, "elbo": 3.5151, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5452146375791695}
{"step": 77510, "loss": 3.5046623229980467, "grad_norm": 0.8193740844726562, "learning_rate": 0.0001, "elbo": 3.50466, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.54550785831574}
{"step": 77515, "loss": 3.50609130859375, "grad_norm": 0.8143638968467712, "learning_rate": 0.0001, "elbo": 3.50609, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5458010790523105}
{"step": 77520, "loss": 3.514506530761719, "grad_norm": 0.6930839419364929, "learning_rate": 0.0001, "elbo": 3.51451, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.546094299788881}
{"step": 77525, "loss": 3.5183765411376955, "grad_norm": 0.7226105332374573, "learning_rate": 0.0001, "elbo": 3.51838, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.546387520525451}
{"step": 77530, "loss": 3.5045516967773436, "grad_norm": 0.6608906388282776, "learning_rate": 0.0001, "elbo": 3.50455, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.546680741262022}
{"step": 77535, "loss": 3.5116245269775392, "grad_norm": 0.7456580400466919, "learning_rate": 0.0001, "elbo": 3.51162, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.546973961998592}
{"step": 77540, "loss": 3.5006740570068358, "grad_norm": 0.7240123748779297, "learning_rate": 0.0001, "elbo": 3.50067, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.547267182735163}
{"step": 77545, "loss": 3.505656433105469, "grad_norm": 0.9648629426956177, "learning_rate": 0.0001, "elbo": 3.50566, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.547560403471733}
{"step": 77550, "loss": 3.514909362792969, "grad_norm": 0.6934379935264587, "learning_rate": 0.0001, "elbo": 3.51491, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.547853624208304}
{"step": 77555, "loss": 3.4994598388671876, "grad_norm": 0.6952312588691711, "learning_rate": 0.0001, "elbo": 3.49946, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.548146844944874}
{"step": 77560, "loss": 3.5151657104492187, "grad_norm": 0.7659438848495483, "learning_rate": 0.0001, "elbo": 3.51517, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.548440065681445}
{"step": 77565, "loss": 3.506163787841797, "grad_norm": 0.724563479423523, "learning_rate": 0.0001, "elbo": 3.50616, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.548733286418015}
{"step": 77570, "loss": 3.5133060455322265, "grad_norm": 0.8092935085296631, "learning_rate": 0.0001, "elbo": 3.51331, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.549026507154586}
{"step": 77575, "loss": 3.4994178771972657, "grad_norm": 0.7263480424880981, "learning_rate": 0.0001, "elbo": 3.49942, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.549319727891157}
{"step": 77580, "loss": 3.5161956787109374, "grad_norm": 0.8108857274055481, "learning_rate": 0.0001, "elbo": 3.5162, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.549612948627727}
{"step": 77585, "loss": 3.4949623107910157, "grad_norm": 0.6877633333206177, "learning_rate": 0.0001, "elbo": 3.49496, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.549906169364298}
{"step": 77590, "loss": 3.504975128173828, "grad_norm": 0.6850259304046631, "learning_rate": 0.0001, "elbo": 3.50498, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.550199390100868}
{"step": 77595, "loss": 3.5109161376953124, "grad_norm": 0.7009154558181763, "learning_rate": 0.0001, "elbo": 3.51092, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.550492610837439}
{"step": 77600, "loss": 3.4934715270996093, "grad_norm": 0.7673010230064392, "learning_rate": 0.0001, "elbo": 3.49347, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.550785831574009}
{"step": 77605, "loss": 3.5092491149902343, "grad_norm": 0.7190008759498596, "learning_rate": 0.0001, "elbo": 3.50925, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5510790523105795}
{"step": 77610, "loss": 3.5034698486328124, "grad_norm": 0.7273066639900208, "learning_rate": 0.0001, "elbo": 3.50347, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.55137227304715}
{"step": 77615, "loss": 3.4957984924316405, "grad_norm": 0.7273382544517517, "learning_rate": 0.0001, "elbo": 3.4958, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.55166549378372}
{"step": 77620, "loss": 3.5218395233154296, "grad_norm": 0.8498069643974304, "learning_rate": 0.0001, "elbo": 3.52184, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.551958714520291}
{"step": 77625, "loss": 3.517289733886719, "grad_norm": 0.7037745714187622, "learning_rate": 0.0001, "elbo": 3.51729, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.552251935256861}
{"step": 77630, "loss": 3.4829010009765624, "grad_norm": 0.6931753158569336, "learning_rate": 0.0001, "elbo": 3.4829, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.552545155993432}
{"step": 77635, "loss": 3.5053367614746094, "grad_norm": 0.8059091567993164, "learning_rate": 0.0001, "elbo": 3.50534, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.552838376730002}
{"step": 77640, "loss": 3.507302474975586, "grad_norm": 0.7122102975845337, "learning_rate": 0.0001, "elbo": 3.5073, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.553131597466573}
{"step": 77645, "loss": 3.500627136230469, "grad_norm": 0.7871349453926086, "learning_rate": 0.0001, "elbo": 3.50063, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.553424818203143}
{"step": 77650, "loss": 3.4984466552734377, "grad_norm": 0.7425616383552551, "learning_rate": 0.0001, "elbo": 3.49845, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.553718038939714}
{"step": 77655, "loss": 3.5130615234375, "grad_norm": 0.7983860373497009, "learning_rate": 0.0001, "elbo": 3.51306, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.554011259676284}
{"step": 77660, "loss": 3.508707809448242, "grad_norm": 0.8009473085403442, "learning_rate": 0.0001, "elbo": 3.50871, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5543044804128545}
{"step": 77665, "loss": 3.499655532836914, "grad_norm": 0.7569555044174194, "learning_rate": 0.0001, "elbo": 3.49966, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.554597701149425}
{"step": 77670, "loss": 3.523126220703125, "grad_norm": 0.7304511666297913, "learning_rate": 0.0001, "elbo": 3.52313, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.554890921885995}
{"step": 77675, "loss": 3.5083152770996096, "grad_norm": 0.7734395265579224, "learning_rate": 0.0001, "elbo": 3.50832, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.555184142622566}
{"step": 77680, "loss": 3.504987335205078, "grad_norm": 0.6980082392692566, "learning_rate": 0.0001, "elbo": 3.50499, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.555477363359136}
{"step": 77685, "loss": 3.523609161376953, "grad_norm": 0.8056023120880127, "learning_rate": 0.0001, "elbo": 3.52361, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.555770584095708}
{"step": 77690, "loss": 3.514311981201172, "grad_norm": 0.7228282690048218, "learning_rate": 0.0001, "elbo": 3.51431, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.556063804832277}
{"step": 77695, "loss": 3.515730285644531, "grad_norm": 0.6731023788452148, "learning_rate": 0.0001, "elbo": 3.51573, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.556357025568849}
{"step": 77700, "loss": 3.489977264404297, "grad_norm": 0.9730126857757568, "learning_rate": 0.0001, "elbo": 3.48998, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.556650246305419}
{"step": 77705, "loss": 3.5001922607421876, "grad_norm": 0.7105048298835754, "learning_rate": 0.0001, "elbo": 3.50019, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5569434670419895}
{"step": 77710, "loss": 3.501173400878906, "grad_norm": 0.9177659749984741, "learning_rate": 0.0001, "elbo": 3.50117, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.55723668777856}
{"step": 77715, "loss": 3.511344146728516, "grad_norm": 0.7367253303527832, "learning_rate": 0.0001, "elbo": 3.51134, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.55752990851513}
{"step": 77720, "loss": 3.5002471923828127, "grad_norm": 0.9450350403785706, "learning_rate": 0.0001, "elbo": 3.50025, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.557823129251701}
{"step": 77725, "loss": 3.508440399169922, "grad_norm": 0.7309806942939758, "learning_rate": 0.0001, "elbo": 3.50844, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.558116349988271}
{"step": 77730, "loss": 3.5057609558105467, "grad_norm": 0.706674337387085, "learning_rate": 0.0001, "elbo": 3.50576, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.558409570724842}
{"step": 77735, "loss": 3.516218566894531, "grad_norm": 0.6922524571418762, "learning_rate": 0.0001, "elbo": 3.51622, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.558702791461412}
{"step": 77740, "loss": 3.5040794372558595, "grad_norm": 0.7106207013130188, "learning_rate": 0.0001, "elbo": 3.50408, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.558996012197983}
{"step": 77745, "loss": 3.5025489807128904, "grad_norm": 0.7118871808052063, "learning_rate": 0.0001, "elbo": 3.50255, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.559289232934553}
{"step": 77750, "loss": 3.505287933349609, "grad_norm": 0.7099802494049072, "learning_rate": 0.0001, "elbo": 3.50529, "mask_frac": 0.5014, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.559582453671124}
{"step": 77755, "loss": 3.4959266662597654, "grad_norm": 0.7401968240737915, "learning_rate": 0.0001, "elbo": 3.49593, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.559875674407694}
{"step": 77760, "loss": 3.5070281982421876, "grad_norm": 0.6835998296737671, "learning_rate": 0.0001, "elbo": 3.50703, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5601688951442645}
{"step": 77765, "loss": 3.509499359130859, "grad_norm": 0.7180825471878052, "learning_rate": 0.0001, "elbo": 3.5095, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.560462115880835}
{"step": 77770, "loss": 3.5097007751464844, "grad_norm": 0.8390259742736816, "learning_rate": 0.0001, "elbo": 3.5097, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.560755336617405}
{"step": 77775, "loss": 3.5034751892089844, "grad_norm": 0.6752333045005798, "learning_rate": 0.0001, "elbo": 3.50348, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.561048557353976}
{"step": 77780, "loss": 3.517655944824219, "grad_norm": 0.7365804314613342, "learning_rate": 0.0001, "elbo": 3.51766, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.561341778090546}
{"step": 77785, "loss": 3.492523193359375, "grad_norm": 0.765626072883606, "learning_rate": 0.0001, "elbo": 3.49252, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.561634998827117}
{"step": 77790, "loss": 3.523493194580078, "grad_norm": 0.7476115226745605, "learning_rate": 0.0001, "elbo": 3.52349, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.561928219563687}
{"step": 77795, "loss": 3.497843933105469, "grad_norm": 0.7904585599899292, "learning_rate": 0.0001, "elbo": 3.49784, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.562221440300258}
{"step": 77800, "loss": 3.51036376953125, "grad_norm": 0.8149873614311218, "learning_rate": 0.0001, "elbo": 3.51036, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.562514661036829}
{"step": 77805, "loss": 3.4948577880859375, "grad_norm": 0.7815074324607849, "learning_rate": 0.0001, "elbo": 3.49486, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.562807881773399}
{"step": 77810, "loss": 3.5129135131835936, "grad_norm": 0.7167510390281677, "learning_rate": 0.0001, "elbo": 3.51291, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.56310110250997}
{"step": 77815, "loss": 3.496571731567383, "grad_norm": 0.7112523317337036, "learning_rate": 0.0001, "elbo": 3.49657, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.56339432324654}
{"step": 77820, "loss": 3.505048370361328, "grad_norm": 0.6744894981384277, "learning_rate": 0.0001, "elbo": 3.50505, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.563687543983111}
{"step": 77825, "loss": 3.5069820404052736, "grad_norm": 0.9654517769813538, "learning_rate": 0.0001, "elbo": 3.50698, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.563980764719681}
{"step": 77830, "loss": 3.52221565246582, "grad_norm": 0.7213881015777588, "learning_rate": 0.0001, "elbo": 3.52222, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.564273985456252}
{"step": 77835, "loss": 3.5014625549316407, "grad_norm": 0.7647746801376343, "learning_rate": 0.0001, "elbo": 3.50146, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.564567206192822}
{"step": 77840, "loss": 3.525651550292969, "grad_norm": 0.7315642833709717, "learning_rate": 0.0001, "elbo": 3.52565, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.564860426929393}
{"step": 77845, "loss": 3.505316162109375, "grad_norm": 0.7874823212623596, "learning_rate": 0.0001, "elbo": 3.50532, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.565153647665963}
{"step": 77850, "loss": 3.5064552307128904, "grad_norm": 0.7213563323020935, "learning_rate": 0.0001, "elbo": 3.50646, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5654468684025336}
{"step": 77855, "loss": 3.498231887817383, "grad_norm": 0.9444727301597595, "learning_rate": 0.0001, "elbo": 3.49823, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.565740089139104}
{"step": 77855, "loss": 3.498231887817383, "grad_norm": 0.9444727301597595, "learning_rate": 0.0001, "elbo": 3.49823, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.565740089139104}
{"step": 77860, "loss": 3.505858612060547, "grad_norm": 0.7085891962051392, "learning_rate": 0.0001, "elbo": 3.50586, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5660333098756745}
{"step": 77865, "loss": 3.509975814819336, "grad_norm": 0.7478018999099731, "learning_rate": 0.0001, "elbo": 3.50998, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.566326530612245}
{"step": 77870, "loss": 3.4974475860595704, "grad_norm": 0.9870670437812805, "learning_rate": 0.0001, "elbo": 3.49745, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.566619751348815}
{"step": 77875, "loss": 3.496957778930664, "grad_norm": 0.7212457060813904, "learning_rate": 0.0001, "elbo": 3.49696, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.566912972085386}
{"step": 77880, "loss": 3.511326217651367, "grad_norm": 0.8522908091545105, "learning_rate": 0.0001, "elbo": 3.51133, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.567206192821956}
{"step": 77885, "loss": 3.5039627075195314, "grad_norm": 0.7348929047584534, "learning_rate": 0.0001, "elbo": 3.50396, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.567499413558527}
{"step": 77890, "loss": 3.5174400329589846, "grad_norm": 0.8176198601722717, "learning_rate": 0.0001, "elbo": 3.51744, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.567792634295097}
{"step": 77895, "loss": 3.4994407653808595, "grad_norm": 0.7158116698265076, "learning_rate": 0.0001, "elbo": 3.49944, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.568085855031668}
{"step": 77900, "loss": 3.506111907958984, "grad_norm": 0.7432318925857544, "learning_rate": 0.0001, "elbo": 3.50611, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.568379075768238}
{"step": 77905, "loss": 3.5102386474609375, "grad_norm": 0.7463908791542053, "learning_rate": 0.0001, "elbo": 3.51024, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5686722965048085}
{"step": 77910, "loss": 3.494153594970703, "grad_norm": 0.7131738662719727, "learning_rate": 0.0001, "elbo": 3.49415, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.568965517241379}
{"step": 77915, "loss": 3.5153656005859375, "grad_norm": 0.7124939560890198, "learning_rate": 0.0001, "elbo": 3.51537, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5692587379779495}
{"step": 77920, "loss": 3.5100772857666014, "grad_norm": 0.707889199256897, "learning_rate": 0.0001, "elbo": 3.51008, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.56955195871452}
{"step": 77925, "loss": 3.517115020751953, "grad_norm": 0.7425910830497742, "learning_rate": 0.0001, "elbo": 3.51712, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.569845179451091}
{"step": 77930, "loss": 3.5186626434326174, "grad_norm": 0.8731242418289185, "learning_rate": 0.0001, "elbo": 3.51866, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.570138400187661}
{"step": 77935, "loss": 3.5163978576660155, "grad_norm": 0.757344663143158, "learning_rate": 0.0001, "elbo": 3.5164, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.570431620924232}
{"step": 77940, "loss": 3.4985469818115233, "grad_norm": 0.7023836970329285, "learning_rate": 0.0001, "elbo": 3.49855, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.570724841660803}
{"step": 77945, "loss": 3.5076370239257812, "grad_norm": 0.723504900932312, "learning_rate": 0.0001, "elbo": 3.50764, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.571018062397373}
{"step": 77950, "loss": 3.5129024505615236, "grad_norm": 0.705592155456543, "learning_rate": 0.0001, "elbo": 3.5129, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5713112831339435}
{"step": 77955, "loss": 3.5142112731933595, "grad_norm": 0.7049347162246704, "learning_rate": 0.0001, "elbo": 3.51421, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.571604503870514}
{"step": 77960, "loss": 3.5204429626464844, "grad_norm": 0.6710987091064453, "learning_rate": 0.0001, "elbo": 3.52044, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.571897724607084}
{"step": 77965, "loss": 3.494080352783203, "grad_norm": 0.7000253200531006, "learning_rate": 0.0001, "elbo": 3.49408, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.572190945343655}
{"step": 77970, "loss": 3.515589141845703, "grad_norm": 0.7439082264900208, "learning_rate": 0.0001, "elbo": 3.51559, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.572484166080225}
{"step": 77975, "loss": 3.483974838256836, "grad_norm": 0.7053062915802002, "learning_rate": 0.0001, "elbo": 3.48397, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.572777386816796}
{"step": 77980, "loss": 3.5103355407714845, "grad_norm": 0.7108309864997864, "learning_rate": 0.0001, "elbo": 3.51034, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.573070607553366}
{"step": 77985, "loss": 3.498937225341797, "grad_norm": 0.7457601428031921, "learning_rate": 0.0001, "elbo": 3.49894, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.573363828289937}
{"step": 77985, "loss": 3.498937225341797, "grad_norm": 0.7457601428031921, "learning_rate": 0.0001, "elbo": 3.49894, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.573363828289937}
{"step": 77990, "loss": 3.504603958129883, "grad_norm": 0.7054901719093323, "learning_rate": 0.0001, "elbo": 3.5046, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.573657049026507}
{"step": 77995, "loss": 3.496154022216797, "grad_norm": 0.7031883001327515, "learning_rate": 0.0001, "elbo": 3.49615, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.573950269763078}
{"step": 78000, "loss": 3.4917221069335938, "grad_norm": 0.7096061110496521, "learning_rate": 0.0001, "elbo": 3.49172, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.574243490499648}
{"step": 78005, "loss": 3.5095008850097655, "grad_norm": 0.699950098991394, "learning_rate": 0.0001, "elbo": 3.5095, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5745367112362185}
{"step": 78010, "loss": 3.496221160888672, "grad_norm": 0.7651364803314209, "learning_rate": 0.0001, "elbo": 3.49622, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.574829931972789}
{"step": 78015, "loss": 3.5048507690429687, "grad_norm": 0.740051805973053, "learning_rate": 0.0001, "elbo": 3.50485, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.575123152709359}
{"step": 78020, "loss": 3.5218238830566406, "grad_norm": 0.7353801131248474, "learning_rate": 0.0001, "elbo": 3.52182, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.57541637344593}
{"step": 78025, "loss": 3.523997497558594, "grad_norm": 0.741409957408905, "learning_rate": 0.0001, "elbo": 3.524, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5757095941825}
{"step": 78030, "loss": 3.515717315673828, "grad_norm": 0.7208964824676514, "learning_rate": 0.0001, "elbo": 3.51572, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.576002814919071}
{"step": 78035, "loss": 3.5003639221191407, "grad_norm": 0.6525745391845703, "learning_rate": 0.0001, "elbo": 3.50036, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.576296035655641}
{"step": 78040, "loss": 3.490126037597656, "grad_norm": 0.7123515009880066, "learning_rate": 0.0001, "elbo": 3.49013, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.576589256392212}
{"step": 78045, "loss": 3.5045814514160156, "grad_norm": 0.8545267581939697, "learning_rate": 0.0001, "elbo": 3.50458, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.576882477128782}
{"step": 78050, "loss": 3.5080402374267576, "grad_norm": 0.7416085600852966, "learning_rate": 0.0001, "elbo": 3.50804, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5771756978653535}
{"step": 78055, "loss": 3.5070541381835936, "grad_norm": 0.6987077593803406, "learning_rate": 0.0001, "elbo": 3.50705, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.577468918601923}
{"step": 78060, "loss": 3.5062713623046875, "grad_norm": 0.6821638345718384, "learning_rate": 0.0001, "elbo": 3.50627, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.577762139338494}
{"step": 78065, "loss": 3.518511962890625, "grad_norm": 0.7343224883079529, "learning_rate": 0.0001, "elbo": 3.51851, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.578055360075065}
{"step": 78070, "loss": 3.52198486328125, "grad_norm": 0.7596026659011841, "learning_rate": 0.0001, "elbo": 3.52198, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.578348580811635}
{"step": 78075, "loss": 3.4997394561767576, "grad_norm": 0.7400758266448975, "learning_rate": 0.0001, "elbo": 3.49974, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.578641801548206}
{"step": 78075, "loss": 3.4997394561767576, "grad_norm": 0.7400758266448975, "learning_rate": 0.0001, "elbo": 3.49974, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.578641801548206}
{"step": 78080, "loss": 3.505361557006836, "grad_norm": 0.6957271695137024, "learning_rate": 0.0001, "elbo": 3.50536, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.578935022284776}
{"step": 78085, "loss": 3.4942062377929686, "grad_norm": 0.659949004650116, "learning_rate": 0.0001, "elbo": 3.49421, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.579228243021347}
{"step": 78090, "loss": 3.4988937377929688, "grad_norm": 0.6912256479263306, "learning_rate": 0.0001, "elbo": 3.49889, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.579521463757917}
{"step": 78095, "loss": 3.5175891876220704, "grad_norm": 0.7239737510681152, "learning_rate": 0.0001, "elbo": 3.51759, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.579814684494488}
{"step": 78100, "loss": 3.4852066040039062, "grad_norm": 0.6677687764167786, "learning_rate": 0.0001, "elbo": 3.48521, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.580107905231058}
{"step": 78100, "loss": 3.4852066040039062, "grad_norm": 0.6677687764167786, "learning_rate": 0.0001, "elbo": 3.48521, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.580107905231058}
{"step": 78105, "loss": 3.4957206726074217, "grad_norm": 0.8303158283233643, "learning_rate": 0.0001, "elbo": 3.49572, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5804011259676285}
{"step": 78110, "loss": 3.5052120208740236, "grad_norm": 0.8267475962638855, "learning_rate": 0.0001, "elbo": 3.50521, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.580694346704199}
{"step": 78115, "loss": 3.5183971405029295, "grad_norm": 0.7072977423667908, "learning_rate": 0.0001, "elbo": 3.5184, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.580987567440769}
{"step": 78120, "loss": 3.513144302368164, "grad_norm": 0.6842820644378662, "learning_rate": 0.0001, "elbo": 3.51314, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.58128078817734}
{"step": 78125, "loss": 3.513884735107422, "grad_norm": 0.6854609847068787, "learning_rate": 0.0001, "elbo": 3.51388, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.58157400891391}
{"step": 78130, "loss": 3.4957481384277345, "grad_norm": 0.9163661003112793, "learning_rate": 0.0001, "elbo": 3.49575, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.581867229650481}
{"step": 78135, "loss": 3.5108406066894533, "grad_norm": 0.7242386937141418, "learning_rate": 0.0001, "elbo": 3.51084, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.582160450387051}
{"step": 78140, "loss": 3.523863983154297, "grad_norm": 0.7341777682304382, "learning_rate": 0.0001, "elbo": 3.52386, "mask_frac": 0.5014, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.582453671123622}
{"step": 78145, "loss": 3.5210102081298826, "grad_norm": 0.7093915939331055, "learning_rate": 0.0001, "elbo": 3.52101, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.582746891860192}
{"step": 78145, "loss": 3.5210102081298826, "grad_norm": 0.7093915939331055, "learning_rate": 0.0001, "elbo": 3.52101, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.582746891860192}
{"step": 78150, "loss": 3.5101207733154296, "grad_norm": 0.7398319840431213, "learning_rate": 0.0001, "elbo": 3.51012, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.583040112596763}
{"step": 78155, "loss": 3.5094886779785157, "grad_norm": 0.7071359753608704, "learning_rate": 0.0001, "elbo": 3.50949, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.583333333333333}
{"step": 78155, "loss": 3.5094886779785157, "grad_norm": 0.7071359753608704, "learning_rate": 0.0001, "elbo": 3.50949, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.583333333333333}
{"step": 78160, "loss": 3.5123111724853517, "grad_norm": 0.7375873327255249, "learning_rate": 0.0001, "elbo": 3.51231, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5836265540699035}
{"step": 78165, "loss": 3.491472625732422, "grad_norm": 0.7164886593818665, "learning_rate": 0.0001, "elbo": 3.49147, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.583919774806474}
{"step": 78170, "loss": 3.5147144317626955, "grad_norm": 0.7607476115226746, "learning_rate": 0.0001, "elbo": 3.51471, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.584212995543044}
{"step": 78175, "loss": 3.5048938751220704, "grad_norm": 0.6939202547073364, "learning_rate": 0.0001, "elbo": 3.50489, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.584506216279616}
{"step": 78180, "loss": 3.50567626953125, "grad_norm": 0.8082186579704285, "learning_rate": 0.0001, "elbo": 3.50568, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.584799437016185}
{"step": 78185, "loss": 3.5064533233642576, "grad_norm": 0.729148268699646, "learning_rate": 0.0001, "elbo": 3.50645, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.585092657752757}
{"step": 78190, "loss": 3.505609130859375, "grad_norm": 0.8437155485153198, "learning_rate": 0.0001, "elbo": 3.50561, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.585385878489327}
{"step": 78195, "loss": 3.5104736328125, "grad_norm": 0.7271426916122437, "learning_rate": 0.0001, "elbo": 3.51047, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.585679099225898}
{"step": 78200, "loss": 3.5088844299316406, "grad_norm": 0.7318366169929504, "learning_rate": 0.0001, "elbo": 3.50888, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.585972319962468}
{"step": 78205, "loss": 3.5201969146728516, "grad_norm": 0.8089037537574768, "learning_rate": 0.0001, "elbo": 3.5202, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5862655406990385}
{"step": 78210, "loss": 3.5077003479003905, "grad_norm": 0.7989599704742432, "learning_rate": 0.0001, "elbo": 3.5077, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.586558761435609}
{"step": 78215, "loss": 3.4971542358398438, "grad_norm": 0.7237948775291443, "learning_rate": 0.0001, "elbo": 3.49715, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.586851982172179}
{"step": 78220, "loss": 3.5073486328125, "grad_norm": 0.7055532336235046, "learning_rate": 0.0001, "elbo": 3.50735, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.58714520290875}
{"step": 78225, "loss": 3.522736358642578, "grad_norm": 0.7324481010437012, "learning_rate": 0.0001, "elbo": 3.52274, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.58743842364532}
{"step": 78230, "loss": 3.516290283203125, "grad_norm": 0.6705663800239563, "learning_rate": 0.0001, "elbo": 3.51629, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.587731644381891}
{"step": 78235, "loss": 3.5184974670410156, "grad_norm": 0.6892001032829285, "learning_rate": 0.0001, "elbo": 3.5185, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.588024865118461}
{"step": 78235, "loss": 3.5184974670410156, "grad_norm": 0.6892001032829285, "learning_rate": 0.0001, "elbo": 3.5185, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.588024865118461}
{"step": 78240, "loss": 3.5038028717041017, "grad_norm": 0.7077376842498779, "learning_rate": 0.0001, "elbo": 3.5038, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.588318085855032}
{"step": 78245, "loss": 3.515401840209961, "grad_norm": 0.7097645401954651, "learning_rate": 0.0001, "elbo": 3.5154, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.588611306591602}
{"step": 78245, "loss": 3.515401840209961, "grad_norm": 0.7097645401954651, "learning_rate": 0.0001, "elbo": 3.5154, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.588611306591602}
{"step": 78250, "loss": 3.518354797363281, "grad_norm": 0.7671791315078735, "learning_rate": 0.0001, "elbo": 3.51835, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5889045273281726}
{"step": 78255, "loss": 3.4927494049072267, "grad_norm": 0.7424026131629944, "learning_rate": 0.0001, "elbo": 3.49275, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.589197748064743}
{"step": 78260, "loss": 3.5022956848144533, "grad_norm": 0.7326239943504333, "learning_rate": 0.0001, "elbo": 3.5023, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5894909688013135}
{"step": 78265, "loss": 3.4970603942871095, "grad_norm": 0.7151759266853333, "learning_rate": 0.0001, "elbo": 3.49706, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.589784189537884}
{"step": 78270, "loss": 3.509496307373047, "grad_norm": 0.7109814882278442, "learning_rate": 0.0001, "elbo": 3.5095, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.590077410274454}
{"step": 78275, "loss": 3.5178768157958986, "grad_norm": 0.8229937553405762, "learning_rate": 0.0001, "elbo": 3.51788, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.590370631011025}
{"step": 78280, "loss": 3.521615982055664, "grad_norm": 0.7514160871505737, "learning_rate": 0.0001, "elbo": 3.52162, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.590663851747595}
{"step": 78285, "loss": 3.509851837158203, "grad_norm": 0.690117597579956, "learning_rate": 0.0001, "elbo": 3.50985, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.590957072484166}
{"step": 78290, "loss": 3.5116065979003905, "grad_norm": 0.7136713266372681, "learning_rate": 0.0001, "elbo": 3.51161, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.591250293220736}
{"step": 78295, "loss": 3.4980392456054688, "grad_norm": 0.6764603853225708, "learning_rate": 0.0001, "elbo": 3.49804, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.591543513957307}
{"step": 78300, "loss": 3.508761978149414, "grad_norm": 0.7256003022193909, "learning_rate": 0.0001, "elbo": 3.50876, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.591836734693878}
{"step": 78305, "loss": 3.493155670166016, "grad_norm": 0.6941810846328735, "learning_rate": 0.0001, "elbo": 3.49316, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5921299554304476}
{"step": 78305, "loss": 3.493155670166016, "grad_norm": 0.6941810846328735, "learning_rate": 0.0001, "elbo": 3.49316, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5921299554304476}
{"step": 78310, "loss": 3.506306457519531, "grad_norm": 0.7072019577026367, "learning_rate": 0.0001, "elbo": 3.50631, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.592423176167019}
{"step": 78315, "loss": 3.4857711791992188, "grad_norm": 0.7292153835296631, "learning_rate": 0.0001, "elbo": 3.48577, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.592716396903589}
{"step": 78320, "loss": 3.5073123931884767, "grad_norm": 0.6910949349403381, "learning_rate": 0.0001, "elbo": 3.50731, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.59300961764016}
{"step": 78325, "loss": 3.5069061279296876, "grad_norm": 0.812594473361969, "learning_rate": 0.0001, "elbo": 3.50691, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.59330283837673}
{"step": 78330, "loss": 3.5023284912109376, "grad_norm": 0.6784692406654358, "learning_rate": 0.0001, "elbo": 3.50233, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.593596059113301}
{"step": 78335, "loss": 3.507175827026367, "grad_norm": 0.6888951659202576, "learning_rate": 0.0001, "elbo": 3.50718, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.593889279849871}
{"step": 78340, "loss": 3.4980289459228517, "grad_norm": 0.8154110312461853, "learning_rate": 0.0001, "elbo": 3.49803, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.594182500586442}
{"step": 78345, "loss": 3.5062862396240235, "grad_norm": 0.7026134729385376, "learning_rate": 0.0001, "elbo": 3.50629, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.594475721323012}
{"step": 78350, "loss": 3.515241241455078, "grad_norm": 0.7028296589851379, "learning_rate": 0.0001, "elbo": 3.51524, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5947689420595825}
{"step": 78350, "loss": 3.515241241455078, "grad_norm": 0.7028296589851379, "learning_rate": 0.0001, "elbo": 3.51524, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5947689420595825}
{"step": 78355, "loss": 3.4970882415771483, "grad_norm": 0.6639123558998108, "learning_rate": 0.0001, "elbo": 3.49709, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.595062162796153}
{"step": 78360, "loss": 3.5163902282714843, "grad_norm": 0.8454596400260925, "learning_rate": 0.0001, "elbo": 3.51639, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.595355383532723}
{"step": 78360, "loss": 3.5163902282714843, "grad_norm": 0.8454596400260925, "learning_rate": 0.0001, "elbo": 3.51639, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.595355383532723}
{"step": 78365, "loss": 3.497175598144531, "grad_norm": 0.7678095698356628, "learning_rate": 0.0001, "elbo": 3.49718, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.595648604269294}
{"step": 78365, "loss": 3.497175598144531, "grad_norm": 0.7678095698356628, "learning_rate": 0.0001, "elbo": 3.49718, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.595648604269294}
{"step": 78370, "loss": 3.5174701690673826, "grad_norm": 0.7206065058708191, "learning_rate": 0.0001, "elbo": 3.51747, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.595941825005864}
{"step": 78375, "loss": 3.5158313751220702, "grad_norm": 0.9406298398971558, "learning_rate": 0.0001, "elbo": 3.51583, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.596235045742435}
{"step": 78380, "loss": 3.5014183044433596, "grad_norm": 0.7249432802200317, "learning_rate": 0.0001, "elbo": 3.50142, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.596528266479005}
{"step": 78385, "loss": 3.498031997680664, "grad_norm": 0.8052276372909546, "learning_rate": 0.0001, "elbo": 3.49803, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.596821487215576}
{"step": 78390, "loss": 3.5028820037841797, "grad_norm": 0.77901291847229, "learning_rate": 0.0001, "elbo": 3.50288, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.597114707952146}
{"step": 78390, "loss": 3.5028820037841797, "grad_norm": 0.77901291847229, "learning_rate": 0.0001, "elbo": 3.50288, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.597114707952146}
{"step": 78395, "loss": 3.503472900390625, "grad_norm": 0.7192253470420837, "learning_rate": 0.0001, "elbo": 3.50347, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.597407928688717}
{"step": 78400, "loss": 3.5160797119140623, "grad_norm": 0.7259536385536194, "learning_rate": 0.0001, "elbo": 3.51608, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.597701149425287}
{"step": 78405, "loss": 3.4993927001953127, "grad_norm": 0.7034116983413696, "learning_rate": 0.0001, "elbo": 3.49939, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.5979943701618575}
{"step": 78410, "loss": 3.5106266021728514, "grad_norm": 0.6829613447189331, "learning_rate": 0.0001, "elbo": 3.51063, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.598287590898428}
{"step": 78415, "loss": 3.507867431640625, "grad_norm": 0.6895706653594971, "learning_rate": 0.0001, "elbo": 3.50787, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.598580811634999}
{"step": 78420, "loss": 3.495494842529297, "grad_norm": 0.6911014318466187, "learning_rate": 0.0001, "elbo": 3.4955, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.598874032371569}
{"step": 78425, "loss": 3.504866027832031, "grad_norm": 0.8561073541641235, "learning_rate": 0.0001, "elbo": 3.50487, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.59916725310814}
{"step": 78430, "loss": 3.5032073974609377, "grad_norm": 0.712289571762085, "learning_rate": 0.0001, "elbo": 3.50321, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.599460473844711}
{"step": 78435, "loss": 3.5124526977539063, "grad_norm": 0.8592230677604675, "learning_rate": 0.0001, "elbo": 3.51245, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.599753694581281}
{"step": 78435, "loss": 3.5124526977539063, "grad_norm": 0.8592230677604675, "learning_rate": 0.0001, "elbo": 3.51245, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.599753694581281}
{"step": 78440, "loss": 3.495183563232422, "grad_norm": 0.6551735401153564, "learning_rate": 0.0001, "elbo": 3.49518, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.600046915317852}
{"step": 78445, "loss": 3.4955066680908202, "grad_norm": 0.6851696372032166, "learning_rate": 0.0001, "elbo": 3.49551, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.600340136054422}
{"step": 78445, "loss": 3.4955066680908202, "grad_norm": 0.6851696372032166, "learning_rate": 0.0001, "elbo": 3.49551, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.600340136054422}
{"step": 78450, "loss": 3.513141632080078, "grad_norm": 0.7029662132263184, "learning_rate": 0.0001, "elbo": 3.51314, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6006333567909925}
{"step": 78455, "loss": 3.5089576721191404, "grad_norm": 0.7044890522956848, "learning_rate": 0.0001, "elbo": 3.50896, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.600926577527563}
{"step": 78460, "loss": 3.5056686401367188, "grad_norm": 0.7490048408508301, "learning_rate": 0.0001, "elbo": 3.50567, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.601219798264133}
{"step": 78465, "loss": 3.506175231933594, "grad_norm": 0.7761040925979614, "learning_rate": 0.0001, "elbo": 3.50618, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.601513019000704}
{"step": 78470, "loss": 3.4932735443115233, "grad_norm": 0.7481188774108887, "learning_rate": 0.0001, "elbo": 3.49327, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.601806239737274}
{"step": 78475, "loss": 3.505870819091797, "grad_norm": 0.7026783227920532, "learning_rate": 0.0001, "elbo": 3.50587, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.602099460473845}
{"step": 78480, "loss": 3.512739562988281, "grad_norm": 0.6853311657905579, "learning_rate": 0.0001, "elbo": 3.51274, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.602392681210415}
{"step": 78485, "loss": 3.5128543853759764, "grad_norm": 1.0835545063018799, "learning_rate": 0.0001, "elbo": 3.51285, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.602685901946986}
{"step": 78490, "loss": 3.521428680419922, "grad_norm": 0.8139243721961975, "learning_rate": 0.0001, "elbo": 3.52143, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.602979122683556}
{"step": 78495, "loss": 3.5024627685546874, "grad_norm": 0.7335967421531677, "learning_rate": 0.0001, "elbo": 3.50246, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.603272343420127}
{"step": 78500, "loss": 3.517417144775391, "grad_norm": 0.7043386101722717, "learning_rate": 0.0001, "elbo": 3.51742, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.603565564156697}
{"step": 78505, "loss": 3.50087890625, "grad_norm": 0.7077152132987976, "learning_rate": 0.0001, "elbo": 3.50088, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6038587848932675}
{"step": 78505, "loss": 3.50087890625, "grad_norm": 0.7077152132987976, "learning_rate": 0.0001, "elbo": 3.50088, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6038587848932675}
{"step": 78510, "loss": 3.5261795043945314, "grad_norm": 0.7026678919792175, "learning_rate": 0.0001, "elbo": 3.52618, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.604152005629838}
{"step": 78515, "loss": 3.5175891876220704, "grad_norm": 0.7766039371490479, "learning_rate": 0.0001, "elbo": 3.51759, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.604445226366408}
{"step": 78520, "loss": 3.5028038024902344, "grad_norm": 0.7003365159034729, "learning_rate": 0.0001, "elbo": 3.5028, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.604738447102979}
{"step": 78525, "loss": 3.500437545776367, "grad_norm": 1.3014562129974365, "learning_rate": 0.0001, "elbo": 3.50044, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.605031667839549}
{"step": 78530, "loss": 3.5097206115722654, "grad_norm": 0.7791943550109863, "learning_rate": 0.0001, "elbo": 3.50972, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.60532488857612}
{"step": 78535, "loss": 3.504874420166016, "grad_norm": 0.8052260279655457, "learning_rate": 0.0001, "elbo": 3.50487, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.60561810931269}
{"step": 78535, "loss": 3.504874420166016, "grad_norm": 0.8052260279655457, "learning_rate": 0.0001, "elbo": 3.50487, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.60561810931269}
{"step": 78540, "loss": 3.528470993041992, "grad_norm": 0.7408875823020935, "learning_rate": 0.0001, "elbo": 3.52847, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.605911330049262}
{"step": 78545, "loss": 3.4926994323730467, "grad_norm": 0.7248877882957458, "learning_rate": 0.0001, "elbo": 3.4927, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.606204550785831}
{"step": 78550, "loss": 3.492906951904297, "grad_norm": 0.7263860106468201, "learning_rate": 0.0001, "elbo": 3.49291, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6064977715224025}
{"step": 78555, "loss": 3.510251998901367, "grad_norm": 0.7382439374923706, "learning_rate": 0.0001, "elbo": 3.51025, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.606790992258973}
{"step": 78560, "loss": 3.5157760620117187, "grad_norm": 0.6984705924987793, "learning_rate": 0.0001, "elbo": 3.51578, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.607084212995543}
{"step": 78565, "loss": 3.4902969360351563, "grad_norm": 0.9396276473999023, "learning_rate": 0.0001, "elbo": 3.4903, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.607377433732114}
{"step": 78570, "loss": 3.493827056884766, "grad_norm": 0.6980092525482178, "learning_rate": 0.0001, "elbo": 3.49383, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.607670654468684}
{"step": 78575, "loss": 3.5088150024414064, "grad_norm": 0.6797126531600952, "learning_rate": 0.0001, "elbo": 3.50881, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.607963875205255}
{"step": 78580, "loss": 3.5126537322998046, "grad_norm": 0.9567604064941406, "learning_rate": 0.0001, "elbo": 3.51265, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.608257095941825}
{"step": 78585, "loss": 3.51563720703125, "grad_norm": 0.714477002620697, "learning_rate": 0.0001, "elbo": 3.51564, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.608550316678396}
{"step": 78590, "loss": 3.5038490295410156, "grad_norm": 0.6965215802192688, "learning_rate": 0.0001, "elbo": 3.50385, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.608843537414966}
{"step": 78595, "loss": 3.494508743286133, "grad_norm": 0.6847130656242371, "learning_rate": 0.0001, "elbo": 3.49451, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.609136758151537}
{"step": 78600, "loss": 3.5174701690673826, "grad_norm": 0.7476619482040405, "learning_rate": 0.0001, "elbo": 3.51747, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.609429978888107}
{"step": 78605, "loss": 3.5129638671875, "grad_norm": 0.7628684043884277, "learning_rate": 0.0001, "elbo": 3.51296, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6097231996246775}
{"step": 78610, "loss": 3.5215595245361326, "grad_norm": 0.7051560878753662, "learning_rate": 0.0001, "elbo": 3.52156, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.610016420361248}
{"step": 78615, "loss": 3.4877655029296877, "grad_norm": 0.7929007411003113, "learning_rate": 0.0001, "elbo": 3.48777, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.610309641097818}
{"step": 78615, "loss": 3.4877655029296877, "grad_norm": 0.7929007411003113, "learning_rate": 0.0001, "elbo": 3.48777, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.610309641097818}
{"step": 78620, "loss": 3.5052413940429688, "grad_norm": 0.6876562833786011, "learning_rate": 0.0001, "elbo": 3.50524, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.610602861834389}
{"step": 78625, "loss": 3.5120330810546876, "grad_norm": 0.7356464862823486, "learning_rate": 0.0001, "elbo": 3.51203, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.610896082570959}
{"step": 78630, "loss": 3.5021514892578125, "grad_norm": 0.7346659302711487, "learning_rate": 0.0001, "elbo": 3.50215, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.61118930330753}
{"step": 78630, "loss": 3.5021514892578125, "grad_norm": 0.7346659302711487, "learning_rate": 0.0001, "elbo": 3.50215, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.61118930330753}
{"step": 78635, "loss": 3.498189163208008, "grad_norm": 0.6942813396453857, "learning_rate": 0.0001, "elbo": 3.49819, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6114825240441}
{"step": 78640, "loss": 3.51649169921875, "grad_norm": 0.7174779176712036, "learning_rate": 0.0001, "elbo": 3.51649, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.611775744780671}
{"step": 78645, "loss": 3.5158668518066407, "grad_norm": 0.69269859790802, "learning_rate": 0.0001, "elbo": 3.51587, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.612068965517241}
{"step": 78650, "loss": 3.4980445861816407, "grad_norm": 0.7836251854896545, "learning_rate": 0.0001, "elbo": 3.49804, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.612362186253812}
{"step": 78650, "loss": 3.4980445861816407, "grad_norm": 0.7836251854896545, "learning_rate": 0.0001, "elbo": 3.49804, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.612362186253812}
{"step": 78655, "loss": 3.5191581726074217, "grad_norm": 0.7192851305007935, "learning_rate": 0.0001, "elbo": 3.51916, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.612655406990382}
{"step": 78660, "loss": 3.5090667724609377, "grad_norm": 0.6969399452209473, "learning_rate": 0.0001, "elbo": 3.50907, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6129486277269525}
{"step": 78665, "loss": 3.5254226684570313, "grad_norm": 0.7778639197349548, "learning_rate": 0.0001, "elbo": 3.52542, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.613241848463524}
{"step": 78670, "loss": 3.514963150024414, "grad_norm": 0.7410294413566589, "learning_rate": 0.0001, "elbo": 3.51496, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.613535069200093}
{"step": 78675, "loss": 3.504171371459961, "grad_norm": 0.6884911060333252, "learning_rate": 0.0001, "elbo": 3.50417, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.613828289936665}
{"step": 78675, "loss": 3.504171371459961, "grad_norm": 0.6884911060333252, "learning_rate": 0.0001, "elbo": 3.50417, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.613828289936665}
{"step": 78680, "loss": 3.4971282958984373, "grad_norm": 0.7255966067314148, "learning_rate": 0.0001, "elbo": 3.49713, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.614121510673235}
{"step": 78680, "loss": 3.4971282958984373, "grad_norm": 0.7255966067314148, "learning_rate": 0.0001, "elbo": 3.49713, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.614121510673235}
{"step": 78685, "loss": 3.487766647338867, "grad_norm": 0.668989896774292, "learning_rate": 0.0001, "elbo": 3.48777, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.614414731409806}
{"step": 78690, "loss": 3.517384719848633, "grad_norm": 0.7364254593849182, "learning_rate": 0.0001, "elbo": 3.51738, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.614707952146376}
{"step": 78695, "loss": 3.4993656158447264, "grad_norm": 0.6976008415222168, "learning_rate": 0.0001, "elbo": 3.49937, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6150011728829465}
{"step": 78700, "loss": 3.5012351989746096, "grad_norm": 0.8044875264167786, "learning_rate": 0.0001, "elbo": 3.50124, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.615294393619517}
{"step": 78705, "loss": 3.492198181152344, "grad_norm": 0.825020432472229, "learning_rate": 0.0001, "elbo": 3.4922, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.615587614356087}
{"step": 78710, "loss": 3.4958614349365233, "grad_norm": 0.7005448341369629, "learning_rate": 0.0001, "elbo": 3.49586, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.615880835092658}
{"step": 78715, "loss": 3.50877685546875, "grad_norm": 0.9639052152633667, "learning_rate": 0.0001, "elbo": 3.50878, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.616174055829228}
{"step": 78720, "loss": 3.493113708496094, "grad_norm": 0.7694165110588074, "learning_rate": 0.0001, "elbo": 3.49311, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.616467276565799}
{"step": 78725, "loss": 3.5010284423828124, "grad_norm": 0.7699421048164368, "learning_rate": 0.0001, "elbo": 3.50103, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.616760497302369}
{"step": 78730, "loss": 3.529210662841797, "grad_norm": 0.7705625891685486, "learning_rate": 0.0001, "elbo": 3.52921, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.61705371803894}
{"step": 78735, "loss": 3.4916534423828125, "grad_norm": 0.7376546263694763, "learning_rate": 0.0001, "elbo": 3.49165, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.61734693877551}
{"step": 78740, "loss": 3.5044090270996096, "grad_norm": 0.7376705408096313, "learning_rate": 0.0001, "elbo": 3.50441, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.617640159512081}
{"step": 78745, "loss": 3.499886322021484, "grad_norm": 0.6843603253364563, "learning_rate": 0.0001, "elbo": 3.49989, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.617933380248651}
{"step": 78750, "loss": 3.4973381042480467, "grad_norm": 0.77020663022995, "learning_rate": 0.0001, "elbo": 3.49734, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6182266009852215}
{"step": 78755, "loss": 3.5121929168701174, "grad_norm": 0.7699633836746216, "learning_rate": 0.0001, "elbo": 3.51219, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.618519821721792}
{"step": 78760, "loss": 3.510659027099609, "grad_norm": 0.7113292217254639, "learning_rate": 0.0001, "elbo": 3.51066, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.618813042458362}
{"step": 78765, "loss": 3.50609130859375, "grad_norm": 0.8020523190498352, "learning_rate": 0.0001, "elbo": 3.50609, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.619106263194933}
{"step": 78770, "loss": 3.5124168395996094, "grad_norm": 0.7211715579032898, "learning_rate": 0.0001, "elbo": 3.51242, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.619399483931503}
{"step": 78775, "loss": 3.513846206665039, "grad_norm": 0.7398109436035156, "learning_rate": 0.0001, "elbo": 3.51385, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.619692704668074}
{"step": 78780, "loss": 3.501287841796875, "grad_norm": 0.7239030003547668, "learning_rate": 0.0001, "elbo": 3.50129, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.619985925404644}
{"step": 78785, "loss": 3.5022727966308596, "grad_norm": 0.7897602915763855, "learning_rate": 0.0001, "elbo": 3.50227, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.620279146141215}
{"step": 78790, "loss": 3.5203834533691407, "grad_norm": 0.7872306704521179, "learning_rate": 0.0001, "elbo": 3.52038, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.620572366877786}
{"step": 78795, "loss": 3.499140167236328, "grad_norm": 1.0316572189331055, "learning_rate": 0.0001, "elbo": 3.49914, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.620865587614356}
{"step": 78800, "loss": 3.495330047607422, "grad_norm": 0.7386228442192078, "learning_rate": 0.0001, "elbo": 3.49533, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.621158808350927}
{"step": 78805, "loss": 3.5039344787597657, "grad_norm": 0.6669430732727051, "learning_rate": 0.0001, "elbo": 3.50393, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.621452029087497}
{"step": 78810, "loss": 3.539705276489258, "grad_norm": 0.7635819315910339, "learning_rate": 0.0001, "elbo": 3.53971, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.621745249824068}
{"step": 78815, "loss": 3.520556640625, "grad_norm": 0.707279622554779, "learning_rate": 0.0001, "elbo": 3.52056, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.622038470560638}
{"step": 78820, "loss": 3.5059345245361326, "grad_norm": 0.7018076777458191, "learning_rate": 0.0001, "elbo": 3.50593, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.622331691297209}
{"step": 78825, "loss": 3.4954071044921875, "grad_norm": 0.7795493006706238, "learning_rate": 0.0001, "elbo": 3.49541, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.622624912033779}
{"step": 78830, "loss": 3.502029037475586, "grad_norm": 0.7671645879745483, "learning_rate": 0.0001, "elbo": 3.50203, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.62291813277035}
{"step": 78835, "loss": 3.5146804809570313, "grad_norm": 0.8413058519363403, "learning_rate": 0.0001, "elbo": 3.51468, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.62321135350692}
{"step": 78840, "loss": 3.5112594604492187, "grad_norm": 0.7319909334182739, "learning_rate": 0.0001, "elbo": 3.51126, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.623504574243491}
{"step": 78845, "loss": 3.5051788330078124, "grad_norm": 0.6819359064102173, "learning_rate": 0.0001, "elbo": 3.50518, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.623797794980061}
{"step": 78850, "loss": 3.5159591674804687, "grad_norm": 0.6572902202606201, "learning_rate": 0.0001, "elbo": 3.51596, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6240910157166315}
{"step": 78855, "loss": 3.5213134765625, "grad_norm": 0.7145289778709412, "learning_rate": 0.0001, "elbo": 3.52131, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.624384236453202}
{"step": 78860, "loss": 3.490270233154297, "grad_norm": 0.9775885939598083, "learning_rate": 0.0001, "elbo": 3.49027, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.624677457189772}
{"step": 78865, "loss": 3.5173011779785157, "grad_norm": 0.7331218719482422, "learning_rate": 0.0001, "elbo": 3.5173, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.624970677926343}
{"step": 78870, "loss": 3.514853286743164, "grad_norm": 0.7018293142318726, "learning_rate": 0.0001, "elbo": 3.51485, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.625263898662913}
{"step": 78875, "loss": 3.49437255859375, "grad_norm": 0.7755162119865417, "learning_rate": 0.0001, "elbo": 3.49437, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.625557119399484}
{"step": 78880, "loss": 3.499001312255859, "grad_norm": 0.7005775570869446, "learning_rate": 0.0001, "elbo": 3.499, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.625850340136054}
{"step": 78885, "loss": 3.490011215209961, "grad_norm": 0.7355653047561646, "learning_rate": 0.0001, "elbo": 3.49001, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.626143560872625}
{"step": 78890, "loss": 3.492005157470703, "grad_norm": 0.8156576156616211, "learning_rate": 0.0001, "elbo": 3.492, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.626436781609195}
{"step": 78895, "loss": 3.5201061248779295, "grad_norm": 0.7386396527290344, "learning_rate": 0.0001, "elbo": 3.52011, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.626730002345766}
{"step": 78900, "loss": 3.4907054901123047, "grad_norm": 0.7073176503181458, "learning_rate": 0.0001, "elbo": 3.49071, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.627023223082336}
{"step": 78905, "loss": 3.5014503479003904, "grad_norm": 0.7227940559387207, "learning_rate": 0.0001, "elbo": 3.50145, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6273164438189065}
{"step": 78910, "loss": 3.5031471252441406, "grad_norm": 0.7561850547790527, "learning_rate": 0.0001, "elbo": 3.50315, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.627609664555477}
{"step": 78915, "loss": 3.5032066345214843, "grad_norm": 0.7077221274375916, "learning_rate": 0.0001, "elbo": 3.50321, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.627902885292048}
{"step": 78920, "loss": 3.488910675048828, "grad_norm": 0.7315926551818848, "learning_rate": 0.0001, "elbo": 3.48891, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.628196106028618}
{"step": 78925, "loss": 3.5072349548339843, "grad_norm": 0.707384467124939, "learning_rate": 0.0001, "elbo": 3.50724, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.628489326765189}
{"step": 78930, "loss": 3.491191101074219, "grad_norm": 1.0723973512649536, "learning_rate": 0.0001, "elbo": 3.49119, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.62878254750176}
{"step": 78935, "loss": 3.5029132843017576, "grad_norm": 0.7193158864974976, "learning_rate": 0.0001, "elbo": 3.50291, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.62907576823833}
{"step": 78940, "loss": 3.488993835449219, "grad_norm": 0.6893064975738525, "learning_rate": 0.0001, "elbo": 3.48899, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.629368988974901}
{"step": 78945, "loss": 3.4947540283203127, "grad_norm": 0.728434681892395, "learning_rate": 0.0001, "elbo": 3.49475, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.629662209711471}
{"step": 78950, "loss": 3.5149856567382813, "grad_norm": 0.8689287304878235, "learning_rate": 0.0001, "elbo": 3.51499, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6299554304480415}
{"step": 78955, "loss": 3.50909423828125, "grad_norm": 0.7066247463226318, "learning_rate": 0.0001, "elbo": 3.50909, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.630248651184612}
{"step": 78960, "loss": 3.500929260253906, "grad_norm": 0.9186940789222717, "learning_rate": 0.0001, "elbo": 3.50093, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.630541871921182}
{"step": 78965, "loss": 3.5052581787109376, "grad_norm": 0.7383268475532532, "learning_rate": 0.0001, "elbo": 3.50526, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.630835092657753}
{"step": 78970, "loss": 3.508820343017578, "grad_norm": 0.7173483967781067, "learning_rate": 0.0001, "elbo": 3.50882, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.631128313394323}
{"step": 78975, "loss": 3.5081348419189453, "grad_norm": 0.6616376042366028, "learning_rate": 0.0001, "elbo": 3.50813, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.631421534130894}
{"step": 78980, "loss": 3.5178123474121095, "grad_norm": 0.6630650162696838, "learning_rate": 0.0001, "elbo": 3.51781, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.631714754867464}
{"step": 78985, "loss": 3.514236068725586, "grad_norm": 0.7134419083595276, "learning_rate": 0.0001, "elbo": 3.51424, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.632007975604035}
{"step": 78990, "loss": 3.520418167114258, "grad_norm": 0.7609713077545166, "learning_rate": 0.0001, "elbo": 3.52042, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.632301196340605}
{"step": 78995, "loss": 3.5025733947753905, "grad_norm": 0.6852350831031799, "learning_rate": 0.0001, "elbo": 3.50257, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.632594417077176}
{"step": 79000, "loss": 3.5098724365234375, "grad_norm": 0.6773492693901062, "learning_rate": 0.0001, "elbo": 3.50987, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.632887637813746}
{"step": 79000, "loss": 3.5098724365234375, "grad_norm": 0.6773492693901062, "learning_rate": 0.0001, "elbo": 3.50987, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.632887637813746}
{"step": 79005, "loss": 3.495214080810547, "grad_norm": 0.8753371238708496, "learning_rate": 0.0001, "elbo": 3.49521, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6331808585503165}
{"step": 79010, "loss": 3.504355621337891, "grad_norm": 0.7097605466842651, "learning_rate": 0.0001, "elbo": 3.50436, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.633474079286887}
{"step": 79015, "loss": 3.501300048828125, "grad_norm": 0.691631019115448, "learning_rate": 0.0001, "elbo": 3.5013, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.633767300023457}
{"step": 79020, "loss": 3.4982032775878906, "grad_norm": 0.721452534198761, "learning_rate": 0.0001, "elbo": 3.4982, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.634060520760028}
{"step": 79025, "loss": 3.494645690917969, "grad_norm": 0.7781996726989746, "learning_rate": 0.0001, "elbo": 3.49465, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.634353741496598}
{"step": 79030, "loss": 3.491616058349609, "grad_norm": 0.7101780772209167, "learning_rate": 0.0001, "elbo": 3.49162, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.634646962233169}
{"step": 79035, "loss": 3.4950546264648437, "grad_norm": 0.7270394563674927, "learning_rate": 0.0001, "elbo": 3.49505, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.634940182969739}
{"step": 79035, "loss": 3.4950546264648437, "grad_norm": 0.7270394563674927, "learning_rate": 0.0001, "elbo": 3.49505, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.634940182969739}
{"step": 79040, "loss": 3.4958282470703126, "grad_norm": 0.9042748808860779, "learning_rate": 0.0001, "elbo": 3.49583, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6352334037063105}
{"step": 79045, "loss": 3.50654296875, "grad_norm": 1.049821138381958, "learning_rate": 0.0001, "elbo": 3.50654, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.63552662444288}
{"step": 79045, "loss": 3.50654296875, "grad_norm": 1.049821138381958, "learning_rate": 0.0001, "elbo": 3.50654, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.63552662444288}
{"step": 79050, "loss": 3.4897388458251952, "grad_norm": 0.7147135138511658, "learning_rate": 0.0001, "elbo": 3.48974, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6358198451794514}
{"step": 79055, "loss": 3.495308685302734, "grad_norm": 0.6788126230239868, "learning_rate": 0.0001, "elbo": 3.49531, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.636113065916022}
{"step": 79060, "loss": 3.486866760253906, "grad_norm": 0.7718093991279602, "learning_rate": 0.0001, "elbo": 3.48687, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.636406286652592}
{"step": 79065, "loss": 3.5029838562011717, "grad_norm": 0.713362991809845, "learning_rate": 0.0001, "elbo": 3.50298, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.636699507389163}
{"step": 79070, "loss": 3.509295654296875, "grad_norm": 0.7129486203193665, "learning_rate": 0.0001, "elbo": 3.5093, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.636992728125733}
{"step": 79075, "loss": 3.505512237548828, "grad_norm": 0.6936065554618835, "learning_rate": 0.0001, "elbo": 3.50551, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.637285948862304}
{"step": 79080, "loss": 3.5053043365478516, "grad_norm": 0.7850320339202881, "learning_rate": 0.0001, "elbo": 3.5053, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.637579169598874}
{"step": 79085, "loss": 3.505626678466797, "grad_norm": 0.7379435300827026, "learning_rate": 0.0001, "elbo": 3.50563, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.637872390335445}
{"step": 79090, "loss": 3.5135353088378904, "grad_norm": 0.6676219701766968, "learning_rate": 0.0001, "elbo": 3.51354, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.638165611072015}
{"step": 79095, "loss": 3.497400665283203, "grad_norm": 0.7322250604629517, "learning_rate": 0.0001, "elbo": 3.4974, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6384588318085855}
{"step": 79100, "loss": 3.5100311279296874, "grad_norm": 0.8063048124313354, "learning_rate": 0.0001, "elbo": 3.51003, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.638752052545156}
{"step": 79105, "loss": 3.5034332275390625, "grad_norm": 0.6904877424240112, "learning_rate": 0.0001, "elbo": 3.50343, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6390452732817264}
{"step": 79110, "loss": 3.5082939147949217, "grad_norm": 0.7161382436752319, "learning_rate": 0.0001, "elbo": 3.50829, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.639338494018297}
{"step": 79115, "loss": 3.4908897399902346, "grad_norm": 0.7085037231445312, "learning_rate": 0.0001, "elbo": 3.49089, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.639631714754867}
{"step": 79120, "loss": 3.507883071899414, "grad_norm": 0.7381271719932556, "learning_rate": 0.0001, "elbo": 3.50788, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.639924935491438}
{"step": 79120, "loss": 3.507883071899414, "grad_norm": 0.7381271719932556, "learning_rate": 0.0001, "elbo": 3.50788, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.639924935491438}
{"step": 79125, "loss": 3.485626220703125, "grad_norm": 0.7559977173805237, "learning_rate": 0.0001, "elbo": 3.48563, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.640218156228008}
{"step": 79125, "loss": 3.485626220703125, "grad_norm": 0.7559977173805237, "learning_rate": 0.0001, "elbo": 3.48563, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.640218156228008}
{"step": 79130, "loss": 3.4853050231933596, "grad_norm": 0.6993411779403687, "learning_rate": 0.0001, "elbo": 3.48531, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.640511376964579}
{"step": 79135, "loss": 3.509740447998047, "grad_norm": 0.9159318208694458, "learning_rate": 0.0001, "elbo": 3.50974, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.640804597701149}
{"step": 79140, "loss": 3.506635284423828, "grad_norm": 0.7589596509933472, "learning_rate": 0.0001, "elbo": 3.50664, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.64109781843772}
{"step": 79145, "loss": 3.509968566894531, "grad_norm": 0.6971330046653748, "learning_rate": 0.0001, "elbo": 3.50997, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.64139103917429}
{"step": 79145, "loss": 3.509968566894531, "grad_norm": 0.6971330046653748, "learning_rate": 0.0001, "elbo": 3.50997, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.64139103917429}
{"step": 79150, "loss": 3.511820602416992, "grad_norm": 0.699823796749115, "learning_rate": 0.0001, "elbo": 3.51182, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6416842599108605}
{"step": 79155, "loss": 3.5000236511230467, "grad_norm": 0.6817774772644043, "learning_rate": 0.0001, "elbo": 3.50002, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.641977480647432}
{"step": 79160, "loss": 3.50308837890625, "grad_norm": 0.7443119883537292, "learning_rate": 0.0001, "elbo": 3.50309, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.642270701384001}
{"step": 79165, "loss": 3.5078685760498045, "grad_norm": 0.823053777217865, "learning_rate": 0.0001, "elbo": 3.50787, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.642563922120573}
{"step": 79170, "loss": 3.5009082794189452, "grad_norm": 0.6573986411094666, "learning_rate": 0.0001, "elbo": 3.50091, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.642857142857143}
{"step": 79175, "loss": 3.487387847900391, "grad_norm": 0.7003309726715088, "learning_rate": 0.0001, "elbo": 3.48739, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.643150363593714}
{"step": 79180, "loss": 3.4959903717041017, "grad_norm": 0.7173780798912048, "learning_rate": 0.0001, "elbo": 3.49599, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.643443584330284}
{"step": 79185, "loss": 3.508350372314453, "grad_norm": 0.8195241093635559, "learning_rate": 0.0001, "elbo": 3.50835, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.643736805066855}
{"step": 79190, "loss": 3.532598876953125, "grad_norm": 0.7551085352897644, "learning_rate": 0.0001, "elbo": 3.5326, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.644030025803425}
{"step": 79195, "loss": 3.4884620666503907, "grad_norm": 0.707988440990448, "learning_rate": 0.0001, "elbo": 3.48846, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6443232465399955}
{"step": 79200, "loss": 3.4907905578613283, "grad_norm": 0.7077558040618896, "learning_rate": 0.0001, "elbo": 3.49079, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.644616467276566}
{"step": 79205, "loss": 3.4971084594726562, "grad_norm": 0.7407774329185486, "learning_rate": 0.0001, "elbo": 3.49711, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.644909688013136}
{"step": 79210, "loss": 3.5144004821777344, "grad_norm": 0.7169172763824463, "learning_rate": 0.0001, "elbo": 3.5144, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.645202908749707}
{"step": 79215, "loss": 3.503759765625, "grad_norm": 0.7711322903633118, "learning_rate": 0.0001, "elbo": 3.50376, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.645496129486277}
{"step": 79215, "loss": 3.503759765625, "grad_norm": 0.7711322903633118, "learning_rate": 0.0001, "elbo": 3.50376, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.645496129486277}
{"step": 79220, "loss": 3.52703857421875, "grad_norm": 0.7433931827545166, "learning_rate": 0.0001, "elbo": 3.52704, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.645789350222848}
{"step": 79225, "loss": 3.497431182861328, "grad_norm": 0.7641577124595642, "learning_rate": 0.0001, "elbo": 3.49743, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.646082570959418}
{"step": 79230, "loss": 3.4977989196777344, "grad_norm": 0.7692484855651855, "learning_rate": 0.0001, "elbo": 3.4978, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.646375791695989}
{"step": 79235, "loss": 3.511709976196289, "grad_norm": 0.7349342703819275, "learning_rate": 0.0001, "elbo": 3.51171, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.646669012432559}
{"step": 79240, "loss": 3.511399841308594, "grad_norm": 0.6535371541976929, "learning_rate": 0.0001, "elbo": 3.5114, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.64696223316913}
{"step": 79245, "loss": 3.4985641479492187, "grad_norm": 0.6829402446746826, "learning_rate": 0.0001, "elbo": 3.49856, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6472554539057}
{"step": 79250, "loss": 3.520100402832031, "grad_norm": 0.7528032064437866, "learning_rate": 0.0001, "elbo": 3.5201, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6475486746422705}
{"step": 79255, "loss": 3.5119003295898437, "grad_norm": 0.8306946158409119, "learning_rate": 0.0001, "elbo": 3.5119, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.647841895378841}
{"step": 79260, "loss": 3.50616340637207, "grad_norm": 0.9209882020950317, "learning_rate": 0.0001, "elbo": 3.50616, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.648135116115411}
{"step": 79265, "loss": 3.5154544830322267, "grad_norm": 1.0346630811691284, "learning_rate": 0.0001, "elbo": 3.51545, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.648428336851982}
{"step": 79270, "loss": 3.486602020263672, "grad_norm": 0.7957315444946289, "learning_rate": 0.0001, "elbo": 3.4866, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.648721557588552}
{"step": 79275, "loss": 3.5104816436767576, "grad_norm": 1.246868371963501, "learning_rate": 0.0001, "elbo": 3.51048, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.649014778325123}
{"step": 79280, "loss": 3.509160614013672, "grad_norm": 0.67353755235672, "learning_rate": 0.0001, "elbo": 3.50916, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.649307999061694}
{"step": 79285, "loss": 3.5029067993164062, "grad_norm": 0.7197202444076538, "learning_rate": 0.0001, "elbo": 3.50291, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.649601219798264}
{"step": 79290, "loss": 3.5128494262695313, "grad_norm": 0.728082537651062, "learning_rate": 0.0001, "elbo": 3.51285, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.649894440534835}
{"step": 79295, "loss": 3.4988998413085937, "grad_norm": 0.7248988151550293, "learning_rate": 0.0001, "elbo": 3.4989, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6501876612714055}
{"step": 79300, "loss": 3.50814208984375, "grad_norm": 0.7928777933120728, "learning_rate": 0.0001, "elbo": 3.50814, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.650480882007976}
{"step": 79305, "loss": 3.5067596435546875, "grad_norm": 0.8154139518737793, "learning_rate": 0.0001, "elbo": 3.50676, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.650774102744546}
{"step": 79310, "loss": 3.5153602600097655, "grad_norm": 0.6854592561721802, "learning_rate": 0.0001, "elbo": 3.51536, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.651067323481117}
{"step": 79315, "loss": 3.501727294921875, "grad_norm": 0.7419270873069763, "learning_rate": 0.0001, "elbo": 3.50173, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.651360544217687}
{"step": 79320, "loss": 3.5033489227294923, "grad_norm": 0.7191694974899292, "learning_rate": 0.0001, "elbo": 3.50335, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.651653764954258}
{"step": 79325, "loss": 3.4992294311523438, "grad_norm": 0.6662322282791138, "learning_rate": 0.0001, "elbo": 3.49923, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.651946985690828}
{"step": 79330, "loss": 3.524112319946289, "grad_norm": 0.6720314025878906, "learning_rate": 0.0001, "elbo": 3.52411, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.652240206427399}
{"step": 79335, "loss": 3.5118038177490236, "grad_norm": 0.7149754762649536, "learning_rate": 0.0001, "elbo": 3.5118, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.652533427163969}
{"step": 79340, "loss": 3.4834217071533202, "grad_norm": 0.7989128232002258, "learning_rate": 0.0001, "elbo": 3.48342, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.65282664790054}
{"step": 79345, "loss": 3.5022598266601563, "grad_norm": 0.685156524181366, "learning_rate": 0.0001, "elbo": 3.50226, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.65311986863711}
{"step": 79350, "loss": 3.497911071777344, "grad_norm": 0.770607054233551, "learning_rate": 0.0001, "elbo": 3.49791, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6534130893736805}
{"step": 79355, "loss": 3.512198638916016, "grad_norm": 0.8992806673049927, "learning_rate": 0.0001, "elbo": 3.5122, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.653706310110251}
{"step": 79360, "loss": 3.5076507568359374, "grad_norm": 0.7194623947143555, "learning_rate": 0.0001, "elbo": 3.50765, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.653999530846821}
{"step": 79365, "loss": 3.502829742431641, "grad_norm": 0.7330939173698425, "learning_rate": 0.0001, "elbo": 3.50283, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.654292751583392}
{"step": 79370, "loss": 3.506999969482422, "grad_norm": 0.7022042274475098, "learning_rate": 0.0001, "elbo": 3.507, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.654585972319962}
{"step": 79375, "loss": 3.4889862060546877, "grad_norm": 0.6606658697128296, "learning_rate": 0.0001, "elbo": 3.48899, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.654879193056533}
{"step": 79380, "loss": 3.504467010498047, "grad_norm": 0.7501862049102783, "learning_rate": 0.0001, "elbo": 3.50447, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.655172413793103}
{"step": 79385, "loss": 3.5080497741699217, "grad_norm": 0.6998899579048157, "learning_rate": 0.0001, "elbo": 3.50805, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.655465634529674}
{"step": 79390, "loss": 3.4969482421875, "grad_norm": 0.6991854906082153, "learning_rate": 0.0001, "elbo": 3.49695, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.655758855266244}
{"step": 79395, "loss": 3.510388946533203, "grad_norm": 0.7025598883628845, "learning_rate": 0.0001, "elbo": 3.51039, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.656052076002815}
{"step": 79400, "loss": 3.500537109375, "grad_norm": 0.7532203793525696, "learning_rate": 0.0001, "elbo": 3.50054, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.656345296739385}
{"step": 79405, "loss": 3.5045391082763673, "grad_norm": 0.8175888657569885, "learning_rate": 0.0001, "elbo": 3.50454, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.656638517475956}
{"step": 79410, "loss": 3.5066207885742187, "grad_norm": 0.7499168515205383, "learning_rate": 0.0001, "elbo": 3.50662, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.656931738212526}
{"step": 79415, "loss": 3.491100311279297, "grad_norm": 0.7995452284812927, "learning_rate": 0.0001, "elbo": 3.4911, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.657224958949097}
{"step": 79415, "loss": 3.491100311279297, "grad_norm": 0.7995452284812927, "learning_rate": 0.0001, "elbo": 3.4911, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.657224958949097}
{"step": 79420, "loss": 3.4880615234375, "grad_norm": 0.879960834980011, "learning_rate": 0.0001, "elbo": 3.48806, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.657518179685668}
{"step": 79420, "loss": 3.4880615234375, "grad_norm": 0.879960834980011, "learning_rate": 0.0001, "elbo": 3.48806, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.657518179685668}
{"step": 79425, "loss": 3.503343200683594, "grad_norm": 0.7133286595344543, "learning_rate": 0.0001, "elbo": 3.50334, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.657811400422238}
{"step": 79430, "loss": 3.5019203186035157, "grad_norm": 0.7665083408355713, "learning_rate": 0.0001, "elbo": 3.50192, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.658104621158809}
{"step": 79435, "loss": 3.5001995086669924, "grad_norm": 0.7233391404151917, "learning_rate": 0.0001, "elbo": 3.5002, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.658397841895379}
{"step": 79440, "loss": 3.501470184326172, "grad_norm": 0.7774868011474609, "learning_rate": 0.0001, "elbo": 3.50147, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6586910626319495}
{"step": 79445, "loss": 3.5168224334716798, "grad_norm": 0.7150290608406067, "learning_rate": 0.0001, "elbo": 3.51682, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.65898428336852}
{"step": 79450, "loss": 3.5152359008789062, "grad_norm": 1.0456392765045166, "learning_rate": 0.0001, "elbo": 3.51524, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6592775041050905}
{"step": 79455, "loss": 3.514379119873047, "grad_norm": 0.6706328988075256, "learning_rate": 0.0001, "elbo": 3.51438, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.659570724841661}
{"step": 79460, "loss": 3.5063404083251952, "grad_norm": 0.6823697090148926, "learning_rate": 0.0001, "elbo": 3.50634, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.659863945578231}
{"step": 79465, "loss": 3.5047653198242186, "grad_norm": 0.7633388042449951, "learning_rate": 0.0001, "elbo": 3.50477, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.660157166314802}
{"step": 79470, "loss": 3.5127220153808594, "grad_norm": 0.7803422808647156, "learning_rate": 0.0001, "elbo": 3.51272, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.660450387051372}
{"step": 79475, "loss": 3.5136791229248048, "grad_norm": 0.740179717540741, "learning_rate": 0.0001, "elbo": 3.51368, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.660743607787943}
{"step": 79480, "loss": 3.5103214263916014, "grad_norm": 0.7937623858451843, "learning_rate": 0.0001, "elbo": 3.51032, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.661036828524513}
{"step": 79485, "loss": 3.5137660980224608, "grad_norm": 0.7304553389549255, "learning_rate": 0.0001, "elbo": 3.51377, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.661330049261084}
{"step": 79490, "loss": 3.5123641967773436, "grad_norm": 0.7020325064659119, "learning_rate": 0.0001, "elbo": 3.51236, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.661623269997654}
{"step": 79495, "loss": 3.5140819549560547, "grad_norm": 0.787810206413269, "learning_rate": 0.0001, "elbo": 3.51408, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6619164907342245}
{"step": 79500, "loss": 3.502699279785156, "grad_norm": 1.087835431098938, "learning_rate": 0.0001, "elbo": 3.5027, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.662209711470795}
{"step": 79500, "loss": 3.502699279785156, "grad_norm": 1.087835431098938, "learning_rate": 0.0001, "elbo": 3.5027, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.662209711470795}
{"step": 79505, "loss": 3.5047473907470703, "grad_norm": 0.6919652819633484, "learning_rate": 0.0001, "elbo": 3.50475, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6625029322073654}
{"step": 79510, "loss": 3.499679183959961, "grad_norm": 0.7297512888908386, "learning_rate": 0.0001, "elbo": 3.49968, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.662796152943936}
{"step": 79515, "loss": 3.5028976440429687, "grad_norm": 0.7217453718185425, "learning_rate": 0.0001, "elbo": 3.5029, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.663089373680506}
{"step": 79520, "loss": 3.487186050415039, "grad_norm": 0.7139999866485596, "learning_rate": 0.0001, "elbo": 3.48719, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.663382594417077}
{"step": 79520, "loss": 3.487186050415039, "grad_norm": 0.7139999866485596, "learning_rate": 0.0001, "elbo": 3.48719, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.663382594417077}
{"step": 79525, "loss": 3.5004920959472656, "grad_norm": 0.736564040184021, "learning_rate": 0.0001, "elbo": 3.50049, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.663675815153647}
{"step": 79530, "loss": 3.4925521850585937, "grad_norm": 0.6801234483718872, "learning_rate": 0.0001, "elbo": 3.49255, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.663969035890219}
{"step": 79535, "loss": 3.516923522949219, "grad_norm": 0.7404575347900391, "learning_rate": 0.0001, "elbo": 3.51692, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.664262256626788}
{"step": 79540, "loss": 3.5024440765380858, "grad_norm": 0.6725673675537109, "learning_rate": 0.0001, "elbo": 3.50244, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6645554773633595}
{"step": 79545, "loss": 3.5128692626953124, "grad_norm": 0.7143264412879944, "learning_rate": 0.0001, "elbo": 3.51287, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.66484869809993}
{"step": 79550, "loss": 3.5201370239257814, "grad_norm": 0.7110668420791626, "learning_rate": 0.0001, "elbo": 3.52014, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6651419188365}
{"step": 79555, "loss": 3.5038101196289064, "grad_norm": 0.7410750389099121, "learning_rate": 0.0001, "elbo": 3.50381, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.665435139573071}
{"step": 79560, "loss": 3.5102874755859377, "grad_norm": 0.8102977871894836, "learning_rate": 0.0001, "elbo": 3.51029, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.665728360309641}
{"step": 79565, "loss": 3.513372802734375, "grad_norm": 1.322070837020874, "learning_rate": 0.0001, "elbo": 3.51337, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.666021581046212}
{"step": 79570, "loss": 3.507090759277344, "grad_norm": 0.7202051877975464, "learning_rate": 0.0001, "elbo": 3.50709, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.666314801782782}
{"step": 79575, "loss": 3.5125873565673826, "grad_norm": 0.8939930200576782, "learning_rate": 0.0001, "elbo": 3.51259, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.666608022519353}
{"step": 79580, "loss": 3.5023422241210938, "grad_norm": 0.8827503323554993, "learning_rate": 0.0001, "elbo": 3.50234, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.666901243255923}
{"step": 79585, "loss": 3.5035354614257814, "grad_norm": 0.6862427592277527, "learning_rate": 0.0001, "elbo": 3.50354, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.667194463992494}
{"step": 79590, "loss": 3.4970497131347655, "grad_norm": 0.7311634421348572, "learning_rate": 0.0001, "elbo": 3.49705, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.667487684729064}
{"step": 79595, "loss": 3.503701019287109, "grad_norm": 0.6834868788719177, "learning_rate": 0.0001, "elbo": 3.5037, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6677809054656345}
{"step": 79600, "loss": 3.5074172973632813, "grad_norm": 0.7426291704177856, "learning_rate": 0.0001, "elbo": 3.50742, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.668074126202205}
{"step": 79605, "loss": 3.5220081329345705, "grad_norm": 0.7508535385131836, "learning_rate": 0.0001, "elbo": 3.52201, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.668367346938775}
{"step": 79610, "loss": 3.500989532470703, "grad_norm": 0.7119829058647156, "learning_rate": 0.0001, "elbo": 3.50099, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.668660567675346}
{"step": 79615, "loss": 3.5160472869873045, "grad_norm": 0.7390881776809692, "learning_rate": 0.0001, "elbo": 3.51605, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.668953788411916}
{"step": 79620, "loss": 3.502655792236328, "grad_norm": 0.7630338072776794, "learning_rate": 0.0001, "elbo": 3.50266, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.669247009148487}
{"step": 79620, "loss": 3.502655792236328, "grad_norm": 0.7630338072776794, "learning_rate": 0.0001, "elbo": 3.50266, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.669247009148487}
{"step": 79625, "loss": 3.5043716430664062, "grad_norm": 0.7999889254570007, "learning_rate": 0.0001, "elbo": 3.50437, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.669540229885057}
{"step": 79630, "loss": 3.5073646545410155, "grad_norm": 0.7252858877182007, "learning_rate": 0.0001, "elbo": 3.50736, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.669833450621628}
{"step": 79635, "loss": 3.5233760833740235, "grad_norm": 0.7331845164299011, "learning_rate": 0.0001, "elbo": 3.52338, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.670126671358198}
{"step": 79640, "loss": 3.500724029541016, "grad_norm": 0.6799783706665039, "learning_rate": 0.0001, "elbo": 3.50072, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.670419892094769}
{"step": 79645, "loss": 3.5087379455566405, "grad_norm": 0.705295741558075, "learning_rate": 0.0001, "elbo": 3.50874, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.670713112831339}
{"step": 79650, "loss": 3.5245849609375, "grad_norm": 0.7229405641555786, "learning_rate": 0.0001, "elbo": 3.52458, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6710063335679095}
{"step": 79655, "loss": 3.515714645385742, "grad_norm": 0.7999027371406555, "learning_rate": 0.0001, "elbo": 3.51571, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.671299554304481}
{"step": 79660, "loss": 3.511518096923828, "grad_norm": 0.7155731916427612, "learning_rate": 0.0001, "elbo": 3.51152, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.67159277504105}
{"step": 79665, "loss": 3.4996124267578126, "grad_norm": 0.6980746388435364, "learning_rate": 0.0001, "elbo": 3.49961, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.671885995777622}
{"step": 79670, "loss": 3.4958282470703126, "grad_norm": 0.7255568504333496, "learning_rate": 0.0001, "elbo": 3.49583, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.672179216514192}
{"step": 79675, "loss": 3.4959880828857424, "grad_norm": 0.6892760396003723, "learning_rate": 0.0001, "elbo": 3.49599, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.672472437250763}
{"step": 79680, "loss": 3.4847183227539062, "grad_norm": 0.9641043543815613, "learning_rate": 0.0001, "elbo": 3.48472, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.672765657987333}
{"step": 79685, "loss": 3.5270877838134767, "grad_norm": 0.7638351321220398, "learning_rate": 0.0001, "elbo": 3.52709, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.673058878723904}
{"step": 79685, "loss": 3.5270877838134767, "grad_norm": 0.7638351321220398, "learning_rate": 0.0001, "elbo": 3.52709, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.673058878723904}
{"step": 79690, "loss": 3.5076877593994142, "grad_norm": 0.7161949872970581, "learning_rate": 0.0001, "elbo": 3.50769, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.673352099460474}
{"step": 79695, "loss": 3.512731170654297, "grad_norm": 0.7299127578735352, "learning_rate": 0.0001, "elbo": 3.51273, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6736453201970445}
{"step": 79700, "loss": 3.496209716796875, "grad_norm": 0.7298763394355774, "learning_rate": 0.0001, "elbo": 3.49621, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.673938540933615}
{"step": 79705, "loss": 3.4946441650390625, "grad_norm": 0.7191400527954102, "learning_rate": 0.0001, "elbo": 3.49464, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.674231761670185}
{"step": 79710, "loss": 3.5037437438964845, "grad_norm": 0.7257404923439026, "learning_rate": 0.0001, "elbo": 3.50374, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.674524982406756}
{"step": 79715, "loss": 3.496107482910156, "grad_norm": 0.7580703496932983, "learning_rate": 0.0001, "elbo": 3.49611, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.674818203143326}
{"step": 79720, "loss": 3.5129714965820313, "grad_norm": 0.7453655004501343, "learning_rate": 0.0001, "elbo": 3.51297, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.675111423879897}
{"step": 79720, "loss": 3.5129714965820313, "grad_norm": 0.7453655004501343, "learning_rate": 0.0001, "elbo": 3.51297, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.675111423879897}
{"step": 79725, "loss": 3.4983062744140625, "grad_norm": 0.8239636421203613, "learning_rate": 0.0001, "elbo": 3.49831, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.675404644616467}
{"step": 79730, "loss": 3.4920669555664063, "grad_norm": 0.8345986008644104, "learning_rate": 0.0001, "elbo": 3.49207, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.675697865353038}
{"step": 79735, "loss": 3.4892658233642577, "grad_norm": 0.7097815275192261, "learning_rate": 0.0001, "elbo": 3.48927, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.675991086089608}
{"step": 79740, "loss": 3.5059440612792967, "grad_norm": 0.705639660358429, "learning_rate": 0.0001, "elbo": 3.50594, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.676284306826179}
{"step": 79745, "loss": 3.499591827392578, "grad_norm": 0.7050177454948425, "learning_rate": 0.0001, "elbo": 3.49959, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.676577527562749}
{"step": 79750, "loss": 3.501005935668945, "grad_norm": 0.733878493309021, "learning_rate": 0.0001, "elbo": 3.50101, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6768707482993195}
{"step": 79755, "loss": 3.498307800292969, "grad_norm": 0.7374208569526672, "learning_rate": 0.0001, "elbo": 3.49831, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.67716396903589}
{"step": 79760, "loss": 3.509287643432617, "grad_norm": 0.7109943628311157, "learning_rate": 0.0001, "elbo": 3.50929, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.67745718977246}
{"step": 79765, "loss": 3.496169662475586, "grad_norm": 0.7705706357955933, "learning_rate": 0.0001, "elbo": 3.49617, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.677750410509031}
{"step": 79770, "loss": 3.4955398559570314, "grad_norm": 0.7322376370429993, "learning_rate": 0.0001, "elbo": 3.49554, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.678043631245602}
{"step": 79775, "loss": 3.4926162719726563, "grad_norm": 0.7614668607711792, "learning_rate": 0.0001, "elbo": 3.49262, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.678336851982172}
{"step": 79780, "loss": 3.497456741333008, "grad_norm": 0.7097830772399902, "learning_rate": 0.0001, "elbo": 3.49746, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.678630072718743}
{"step": 79785, "loss": 3.513530731201172, "grad_norm": 0.6978088617324829, "learning_rate": 0.0001, "elbo": 3.51353, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6789232934553135}
{"step": 79790, "loss": 3.499521255493164, "grad_norm": 0.688836395740509, "learning_rate": 0.0001, "elbo": 3.49952, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.679216514191884}
{"step": 79795, "loss": 3.4894752502441406, "grad_norm": 0.6832610964775085, "learning_rate": 0.0001, "elbo": 3.48948, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6795097349284545}
{"step": 79800, "loss": 3.5041950225830076, "grad_norm": 0.7013833522796631, "learning_rate": 0.0001, "elbo": 3.5042, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.679802955665025}
{"step": 79805, "loss": 3.51005859375, "grad_norm": 0.7134585380554199, "learning_rate": 0.0001, "elbo": 3.51006, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.680096176401595}
{"step": 79810, "loss": 3.5078765869140627, "grad_norm": 0.7255485653877258, "learning_rate": 0.0001, "elbo": 3.50788, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.680389397138166}
{"step": 79810, "loss": 3.5078765869140627, "grad_norm": 0.7255485653877258, "learning_rate": 0.0001, "elbo": 3.50788, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.680389397138166}
{"step": 79815, "loss": 3.5126422882080077, "grad_norm": 0.7858722805976868, "learning_rate": 0.0001, "elbo": 3.51264, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.680682617874736}
{"step": 79820, "loss": 3.517591094970703, "grad_norm": 0.7704407572746277, "learning_rate": 0.0001, "elbo": 3.51759, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.680975838611307}
{"step": 79825, "loss": 3.5065258026123045, "grad_norm": 1.3504632711410522, "learning_rate": 0.0001, "elbo": 3.50653, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.681269059347877}
{"step": 79830, "loss": 3.4876827239990233, "grad_norm": 0.6750524044036865, "learning_rate": 0.0001, "elbo": 3.48768, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.681562280084448}
{"step": 79835, "loss": 3.512023162841797, "grad_norm": 0.6990869045257568, "learning_rate": 0.0001, "elbo": 3.51202, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.681855500821018}
{"step": 79840, "loss": 3.504651641845703, "grad_norm": 0.9288235306739807, "learning_rate": 0.0001, "elbo": 3.50465, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6821487215575885}
{"step": 79840, "loss": 3.504651641845703, "grad_norm": 0.9288235306739807, "learning_rate": 0.0001, "elbo": 3.50465, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6821487215575885}
{"step": 79845, "loss": 3.510029602050781, "grad_norm": 0.7353043556213379, "learning_rate": 0.0001, "elbo": 3.51003, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.682441942294159}
{"step": 79850, "loss": 3.5088523864746093, "grad_norm": 0.7501158714294434, "learning_rate": 0.0001, "elbo": 3.50885, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6827351630307295}
{"step": 79855, "loss": 3.4903907775878906, "grad_norm": 0.7420675754547119, "learning_rate": 0.0001, "elbo": 3.49039, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6830283837673}
{"step": 79860, "loss": 3.5074478149414063, "grad_norm": 0.8335906863212585, "learning_rate": 0.0001, "elbo": 3.50745, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.68332160450387}
{"step": 79865, "loss": 3.508538818359375, "grad_norm": 0.7245909571647644, "learning_rate": 0.0001, "elbo": 3.50854, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.683614825240441}
{"step": 79870, "loss": 3.500157928466797, "grad_norm": 0.702254056930542, "learning_rate": 0.0001, "elbo": 3.50016, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.683908045977011}
{"step": 79875, "loss": 3.5048614501953126, "grad_norm": 1.3683916330337524, "learning_rate": 0.0001, "elbo": 3.50486, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.684201266713582}
{"step": 79880, "loss": 3.4927906036376952, "grad_norm": 0.7039327621459961, "learning_rate": 0.0001, "elbo": 3.49279, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.684494487450152}
{"step": 79885, "loss": 3.511022186279297, "grad_norm": 0.8393599987030029, "learning_rate": 0.0001, "elbo": 3.51102, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.684787708186723}
{"step": 79890, "loss": 3.4983360290527346, "grad_norm": 0.6673380732536316, "learning_rate": 0.0001, "elbo": 3.49834, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.685080928923293}
{"step": 79895, "loss": 3.5125732421875, "grad_norm": 0.6977286338806152, "learning_rate": 0.0001, "elbo": 3.51257, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.685374149659864}
{"step": 79900, "loss": 3.497728729248047, "grad_norm": 0.6986075043678284, "learning_rate": 0.0001, "elbo": 3.49773, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.685667370396434}
{"step": 79905, "loss": 3.5068714141845705, "grad_norm": 0.6902090311050415, "learning_rate": 0.0001, "elbo": 3.50687, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.685960591133005}
{"step": 79910, "loss": 3.499523162841797, "grad_norm": 1.1277178525924683, "learning_rate": 0.0001, "elbo": 3.49952, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.686253811869576}
{"step": 79915, "loss": 3.499867248535156, "grad_norm": 0.7953989505767822, "learning_rate": 0.0001, "elbo": 3.49987, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.686547032606146}
{"step": 79920, "loss": 3.4952579498291017, "grad_norm": 0.8476555943489075, "learning_rate": 0.0001, "elbo": 3.49526, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.686840253342717}
{"step": 79925, "loss": 3.5099586486816405, "grad_norm": 0.7235207557678223, "learning_rate": 0.0001, "elbo": 3.50996, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.687133474079287}
{"step": 79930, "loss": 3.5094337463378906, "grad_norm": 0.7222785353660583, "learning_rate": 0.0001, "elbo": 3.50943, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.687426694815858}
{"step": 79935, "loss": 3.4901214599609376, "grad_norm": 0.7402734160423279, "learning_rate": 0.0001, "elbo": 3.49012, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.687719915552428}
{"step": 79940, "loss": 3.507050323486328, "grad_norm": 0.6803302764892578, "learning_rate": 0.0001, "elbo": 3.50705, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6880131362889985}
{"step": 79945, "loss": 3.4915000915527346, "grad_norm": 0.7229049205780029, "learning_rate": 0.0001, "elbo": 3.4915, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.688306357025569}
{"step": 79950, "loss": 3.5149192810058594, "grad_norm": 0.6922242045402527, "learning_rate": 0.0001, "elbo": 3.51492, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.688599577762139}
{"step": 79955, "loss": 3.4877758026123047, "grad_norm": 0.8753876090049744, "learning_rate": 0.0001, "elbo": 3.48778, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.68889279849871}
{"step": 79960, "loss": 3.4943111419677733, "grad_norm": 0.6686311364173889, "learning_rate": 0.0001, "elbo": 3.49431, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.68918601923528}
{"step": 79965, "loss": 3.492111587524414, "grad_norm": 0.8780876398086548, "learning_rate": 0.0001, "elbo": 3.49211, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.689479239971851}
{"step": 79970, "loss": 3.5022918701171877, "grad_norm": 0.7276010513305664, "learning_rate": 0.0001, "elbo": 3.50229, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.689772460708421}
{"step": 79975, "loss": 3.5086814880371096, "grad_norm": 0.6954821348190308, "learning_rate": 0.0001, "elbo": 3.50868, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.690065681444992}
{"step": 79980, "loss": 3.5101524353027345, "grad_norm": 0.7754164338111877, "learning_rate": 0.0001, "elbo": 3.51015, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.690358902181562}
{"step": 79985, "loss": 3.5064788818359376, "grad_norm": 0.6842482089996338, "learning_rate": 0.0001, "elbo": 3.50648, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.690652122918133}
{"step": 79990, "loss": 3.49981689453125, "grad_norm": 0.7140610218048096, "learning_rate": 0.0001, "elbo": 3.49982, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.690945343654703}
{"step": 79995, "loss": 3.4943817138671873, "grad_norm": 0.7763784527778625, "learning_rate": 0.0001, "elbo": 3.49438, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6912385643912735}
{"step": 80000, "loss": 3.510184478759766, "grad_norm": 0.7222234010696411, "learning_rate": 0.0001, "elbo": 3.51018, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.691531785127844}
{"step": 80005, "loss": 3.50465087890625, "grad_norm": 0.6968851089477539, "learning_rate": 0.0001, "elbo": 3.50465, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.691825005864414}
{"step": 80010, "loss": 3.5066131591796874, "grad_norm": 0.7286607623100281, "learning_rate": 0.0001, "elbo": 3.50661, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.692118226600985}
{"step": 80015, "loss": 3.517116165161133, "grad_norm": 0.7231301069259644, "learning_rate": 0.0001, "elbo": 3.51712, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.692411447337555}
{"step": 80020, "loss": 3.504697799682617, "grad_norm": 0.719089686870575, "learning_rate": 0.0001, "elbo": 3.5047, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.692704668074127}
{"step": 80025, "loss": 3.5191646575927735, "grad_norm": 0.7377385497093201, "learning_rate": 0.0001, "elbo": 3.51916, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.692997888810696}
{"step": 80030, "loss": 3.5051242828369142, "grad_norm": 0.6800757050514221, "learning_rate": 0.0001, "elbo": 3.50512, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.693291109547268}
{"step": 80035, "loss": 3.512312316894531, "grad_norm": 0.7189757227897644, "learning_rate": 0.0001, "elbo": 3.51231, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.693584330283838}
{"step": 80040, "loss": 3.4899566650390623, "grad_norm": 0.7590543031692505, "learning_rate": 0.0001, "elbo": 3.48996, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6938775510204085}
{"step": 80045, "loss": 3.5174144744873046, "grad_norm": 0.7044867277145386, "learning_rate": 0.0001, "elbo": 3.51741, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.694170771756979}
{"step": 80050, "loss": 3.5093368530273437, "grad_norm": 0.791554868221283, "learning_rate": 0.0001, "elbo": 3.50934, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.694463992493549}
{"step": 80055, "loss": 3.5013267517089846, "grad_norm": 0.7320224642753601, "learning_rate": 0.0001, "elbo": 3.50133, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.69475721323012}
{"step": 80060, "loss": 3.5037349700927733, "grad_norm": 0.7267554998397827, "learning_rate": 0.0001, "elbo": 3.50374, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.69505043396669}
{"step": 80065, "loss": 3.500362014770508, "grad_norm": 0.6916806101799011, "learning_rate": 0.0001, "elbo": 3.50036, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.695343654703261}
{"step": 80070, "loss": 3.506163787841797, "grad_norm": 1.3945655822753906, "learning_rate": 0.0001, "elbo": 3.50616, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.695636875439831}
{"step": 80070, "loss": 3.506163787841797, "grad_norm": 1.3945655822753906, "learning_rate": 0.0001, "elbo": 3.50616, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.695636875439831}
{"step": 80075, "loss": 3.5162681579589843, "grad_norm": 0.7253872156143188, "learning_rate": 0.0001, "elbo": 3.51627, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.695930096176402}
{"step": 80080, "loss": 3.5036102294921876, "grad_norm": 0.731141209602356, "learning_rate": 0.0001, "elbo": 3.50361, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.696223316912972}
{"step": 80085, "loss": 3.508000946044922, "grad_norm": 0.7070441246032715, "learning_rate": 0.0001, "elbo": 3.508, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.696516537649543}
{"step": 80090, "loss": 3.514297103881836, "grad_norm": 0.8430715799331665, "learning_rate": 0.0001, "elbo": 3.5143, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.696809758386113}
{"step": 80095, "loss": 3.515245819091797, "grad_norm": 0.7754707932472229, "learning_rate": 0.0001, "elbo": 3.51525, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.6971029791226835}
{"step": 80100, "loss": 3.500115966796875, "grad_norm": 0.6942248940467834, "learning_rate": 0.0001, "elbo": 3.50012, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.697396199859254}
{"step": 80105, "loss": 3.5005420684814452, "grad_norm": 0.7240422964096069, "learning_rate": 0.0001, "elbo": 3.50054, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.697689420595824}
{"step": 80110, "loss": 3.4993247985839844, "grad_norm": 0.819251298904419, "learning_rate": 0.0001, "elbo": 3.49933, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.697982641332395}
{"step": 80115, "loss": 3.5002552032470704, "grad_norm": 0.7426996827125549, "learning_rate": 0.0001, "elbo": 3.50026, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.698275862068965}
{"step": 80120, "loss": 3.485617828369141, "grad_norm": 0.7071532011032104, "learning_rate": 0.0001, "elbo": 3.48562, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.698569082805536}
{"step": 80125, "loss": 3.4846145629882814, "grad_norm": 0.7109741568565369, "learning_rate": 0.0001, "elbo": 3.48461, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.698862303542106}
{"step": 80130, "loss": 3.510376739501953, "grad_norm": 0.7044121623039246, "learning_rate": 0.0001, "elbo": 3.51038, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.699155524278677}
{"step": 80135, "loss": 3.501190185546875, "grad_norm": 0.7092306017875671, "learning_rate": 0.0001, "elbo": 3.50119, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.699448745015247}
{"step": 80140, "loss": 3.5110740661621094, "grad_norm": 0.7233164310455322, "learning_rate": 0.0001, "elbo": 3.51107, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.699741965751818}
{"step": 80145, "loss": 3.524166488647461, "grad_norm": 0.7133433818817139, "learning_rate": 0.0001, "elbo": 3.52417, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.700035186488389}
{"step": 80150, "loss": 3.5049453735351563, "grad_norm": 0.9040846824645996, "learning_rate": 0.0001, "elbo": 3.50495, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7003284072249585}
{"step": 80155, "loss": 3.4938915252685545, "grad_norm": 0.7098549604415894, "learning_rate": 0.0001, "elbo": 3.49389, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.70062162796153}
{"step": 80160, "loss": 3.4949180603027346, "grad_norm": 0.6783439517021179, "learning_rate": 0.0001, "elbo": 3.49492, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7009148486981}
{"step": 80165, "loss": 3.5159751892089846, "grad_norm": 0.663626492023468, "learning_rate": 0.0001, "elbo": 3.51598, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.701208069434671}
{"step": 80170, "loss": 3.5000972747802734, "grad_norm": 1.2174839973449707, "learning_rate": 0.0001, "elbo": 3.5001, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.701501290171241}
{"step": 80175, "loss": 3.4945545196533203, "grad_norm": 0.7471510767936707, "learning_rate": 0.0001, "elbo": 3.49455, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.701794510907812}
{"step": 80180, "loss": 3.5215736389160157, "grad_norm": 0.6941272020339966, "learning_rate": 0.0001, "elbo": 3.52157, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.702087731644382}
{"step": 80185, "loss": 3.487811279296875, "grad_norm": 0.7281994819641113, "learning_rate": 0.0001, "elbo": 3.48781, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7023809523809526}
{"step": 80185, "loss": 3.487811279296875, "grad_norm": 0.7281994819641113, "learning_rate": 0.0001, "elbo": 3.48781, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7023809523809526}
{"step": 80190, "loss": 3.5126811981201174, "grad_norm": 0.7211306691169739, "learning_rate": 0.0001, "elbo": 3.51268, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.702674173117523}
{"step": 80195, "loss": 3.5058563232421873, "grad_norm": 0.7212650179862976, "learning_rate": 0.0001, "elbo": 3.50586, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7029673938540935}
{"step": 80200, "loss": 3.5007354736328127, "grad_norm": 0.8570291996002197, "learning_rate": 0.0001, "elbo": 3.50074, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.703260614590664}
{"step": 80205, "loss": 3.504983901977539, "grad_norm": 0.7893831133842468, "learning_rate": 0.0001, "elbo": 3.50498, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.703553835327234}
{"step": 80210, "loss": 3.507773590087891, "grad_norm": 0.7184688448905945, "learning_rate": 0.0001, "elbo": 3.50777, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.703847056063805}
{"step": 80210, "loss": 3.507773590087891, "grad_norm": 0.7184688448905945, "learning_rate": 0.0001, "elbo": 3.50777, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.703847056063805}
{"step": 80215, "loss": 3.5018470764160154, "grad_norm": 0.7450094819068909, "learning_rate": 0.0001, "elbo": 3.50185, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.704140276800375}
{"step": 80220, "loss": 3.499603271484375, "grad_norm": 0.7776232957839966, "learning_rate": 0.0001, "elbo": 3.4996, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.704433497536946}
{"step": 80225, "loss": 3.4975791931152345, "grad_norm": 0.7424328327178955, "learning_rate": 0.0001, "elbo": 3.49758, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.704726718273516}
{"step": 80230, "loss": 3.4912040710449217, "grad_norm": 0.7485687732696533, "learning_rate": 0.0001, "elbo": 3.4912, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.705019939010087}
{"step": 80235, "loss": 3.483531188964844, "grad_norm": 0.7251995205879211, "learning_rate": 0.0001, "elbo": 3.48353, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.705313159746657}
{"step": 80240, "loss": 3.504290008544922, "grad_norm": 0.7304826378822327, "learning_rate": 0.0001, "elbo": 3.50429, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7056063804832275}
{"step": 80245, "loss": 3.5098358154296876, "grad_norm": 0.7098493576049805, "learning_rate": 0.0001, "elbo": 3.50984, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.705899601219798}
{"step": 80250, "loss": 3.4994876861572264, "grad_norm": 0.6675599217414856, "learning_rate": 0.0001, "elbo": 3.49949, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7061928219563685}
{"step": 80255, "loss": 3.502223587036133, "grad_norm": 0.7463135719299316, "learning_rate": 0.0001, "elbo": 3.50222, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.706486042692939}
{"step": 80260, "loss": 3.5073501586914064, "grad_norm": 0.7031821608543396, "learning_rate": 0.0001, "elbo": 3.50735, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.706779263429509}
{"step": 80265, "loss": 3.492388916015625, "grad_norm": 0.684395968914032, "learning_rate": 0.0001, "elbo": 3.49239, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.70707248416608}
{"step": 80270, "loss": 3.504261016845703, "grad_norm": 0.7174364328384399, "learning_rate": 0.0001, "elbo": 3.50426, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.707365704902651}
{"step": 80275, "loss": 3.5194412231445313, "grad_norm": 0.7575901746749878, "learning_rate": 0.0001, "elbo": 3.51944, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.707658925639221}
{"step": 80280, "loss": 3.5113761901855467, "grad_norm": 0.6910285949707031, "learning_rate": 0.0001, "elbo": 3.51138, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.707952146375792}
{"step": 80285, "loss": 3.502989959716797, "grad_norm": 0.7310961484909058, "learning_rate": 0.0001, "elbo": 3.50299, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7082453671123625}
{"step": 80290, "loss": 3.490825653076172, "grad_norm": 0.7400485873222351, "learning_rate": 0.0001, "elbo": 3.49083, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.708538587848933}
{"step": 80290, "loss": 3.490825653076172, "grad_norm": 0.7400485873222351, "learning_rate": 0.0001, "elbo": 3.49083, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.708538587848933}
{"step": 80295, "loss": 3.5049190521240234, "grad_norm": 0.6981561183929443, "learning_rate": 0.0001, "elbo": 3.50492, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.708831808585503}
{"step": 80300, "loss": 3.5081787109375, "grad_norm": 0.9367825388908386, "learning_rate": 0.0001, "elbo": 3.50818, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.709125029322074}
{"step": 80305, "loss": 3.501827621459961, "grad_norm": 0.7245745062828064, "learning_rate": 0.0001, "elbo": 3.50183, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.709418250058644}
{"step": 80310, "loss": 3.5111602783203124, "grad_norm": 0.7390846610069275, "learning_rate": 0.0001, "elbo": 3.51116, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.709711470795215}
{"step": 80315, "loss": 3.4989578247070314, "grad_norm": 0.7670460939407349, "learning_rate": 0.0001, "elbo": 3.49896, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.710004691531785}
{"step": 80320, "loss": 3.5081512451171877, "grad_norm": 0.7232096195220947, "learning_rate": 0.0001, "elbo": 3.50815, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.710297912268356}
{"step": 80325, "loss": 3.5174251556396485, "grad_norm": 0.6588019132614136, "learning_rate": 0.0001, "elbo": 3.51743, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.710591133004926}
{"step": 80330, "loss": 3.4962112426757814, "grad_norm": 0.8039198517799377, "learning_rate": 0.0001, "elbo": 3.49621, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.710884353741497}
{"step": 80335, "loss": 3.5058006286621093, "grad_norm": 0.7573667764663696, "learning_rate": 0.0001, "elbo": 3.5058, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.711177574478067}
{"step": 80340, "loss": 3.511915588378906, "grad_norm": 0.7491524815559387, "learning_rate": 0.0001, "elbo": 3.51192, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7114707952146375}
{"step": 80345, "loss": 3.5039676666259765, "grad_norm": 0.7483307123184204, "learning_rate": 0.0001, "elbo": 3.50397, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.711764015951208}
{"step": 80350, "loss": 3.506722640991211, "grad_norm": 0.813116729259491, "learning_rate": 0.0001, "elbo": 3.50672, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.712057236687778}
{"step": 80355, "loss": 3.4992576599121095, "grad_norm": 1.054788589477539, "learning_rate": 0.0001, "elbo": 3.49926, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.712350457424349}
{"step": 80360, "loss": 3.4956680297851563, "grad_norm": 0.7523441910743713, "learning_rate": 0.0001, "elbo": 3.49567, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.712643678160919}
{"step": 80365, "loss": 3.497209167480469, "grad_norm": 0.7950422763824463, "learning_rate": 0.0001, "elbo": 3.49721, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.71293689889749}
{"step": 80370, "loss": 3.506513977050781, "grad_norm": 0.9847031235694885, "learning_rate": 0.0001, "elbo": 3.50651, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.71323011963406}
{"step": 80375, "loss": 3.4944538116455077, "grad_norm": 0.7394375801086426, "learning_rate": 0.0001, "elbo": 3.49445, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.713523340370631}
{"step": 80380, "loss": 3.504705810546875, "grad_norm": 0.7438253164291382, "learning_rate": 0.0001, "elbo": 3.50471, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.713816561107201}
{"step": 80385, "loss": 3.5059173583984373, "grad_norm": 0.8512063026428223, "learning_rate": 0.0001, "elbo": 3.50592, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.714109781843772}
{"step": 80390, "loss": 3.5005645751953125, "grad_norm": 0.7097195386886597, "learning_rate": 0.0001, "elbo": 3.50056, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.714403002580342}
{"step": 80395, "loss": 3.5034709930419923, "grad_norm": 0.6958618760108948, "learning_rate": 0.0001, "elbo": 3.50347, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.714696223316913}
{"step": 80400, "loss": 3.502759552001953, "grad_norm": 0.6870436668395996, "learning_rate": 0.0001, "elbo": 3.50276, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.714989444053483}
{"step": 80405, "loss": 3.5098915100097656, "grad_norm": 0.6979489326477051, "learning_rate": 0.0001, "elbo": 3.50989, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.715282664790054}
{"step": 80410, "loss": 3.4996002197265623, "grad_norm": 0.7075219750404358, "learning_rate": 0.0001, "elbo": 3.4996, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.715575885526625}
{"step": 80415, "loss": 3.4910377502441405, "grad_norm": 0.7002007961273193, "learning_rate": 0.0001, "elbo": 3.49104, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.715869106263195}
{"step": 80420, "loss": 3.517633819580078, "grad_norm": 0.7239531874656677, "learning_rate": 0.0001, "elbo": 3.51763, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.716162326999766}
{"step": 80425, "loss": 3.520979309082031, "grad_norm": 0.6980880498886108, "learning_rate": 0.0001, "elbo": 3.52098, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.716455547736336}
{"step": 80430, "loss": 3.4959091186523437, "grad_norm": 0.7143178582191467, "learning_rate": 0.0001, "elbo": 3.49591, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.716748768472907}
{"step": 80435, "loss": 3.5056663513183595, "grad_norm": 0.7114428281784058, "learning_rate": 0.0001, "elbo": 3.50567, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.717041989209477}
{"step": 80440, "loss": 3.497732925415039, "grad_norm": 0.7580488920211792, "learning_rate": 0.0001, "elbo": 3.49773, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7173352099460475}
{"step": 80445, "loss": 3.5043666839599608, "grad_norm": 0.7771273255348206, "learning_rate": 0.0001, "elbo": 3.50437, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.717628430682618}
{"step": 80450, "loss": 3.510247802734375, "grad_norm": 0.7551497220993042, "learning_rate": 0.0001, "elbo": 3.51025, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.717921651419188}
{"step": 80455, "loss": 3.5068782806396483, "grad_norm": 0.7243039011955261, "learning_rate": 0.0001, "elbo": 3.50688, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.718214872155759}
{"step": 80460, "loss": 3.495554733276367, "grad_norm": 0.7405259013175964, "learning_rate": 0.0001, "elbo": 3.49555, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.718508092892329}
{"step": 80465, "loss": 3.5007068634033205, "grad_norm": 0.7393847703933716, "learning_rate": 0.0001, "elbo": 3.50071, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7188013136289}
{"step": 80470, "loss": 3.4978694915771484, "grad_norm": 0.7114002108573914, "learning_rate": 0.0001, "elbo": 3.49787, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.71909453436547}
{"step": 80475, "loss": 3.5134132385253904, "grad_norm": 0.8518075942993164, "learning_rate": 0.0001, "elbo": 3.51341, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.719387755102041}
{"step": 80480, "loss": 3.5111045837402344, "grad_norm": 0.6964306831359863, "learning_rate": 0.0001, "elbo": 3.5111, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.719680975838611}
{"step": 80485, "loss": 3.512931060791016, "grad_norm": 0.6839780807495117, "learning_rate": 0.0001, "elbo": 3.51293, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.719974196575182}
{"step": 80490, "loss": 3.498611068725586, "grad_norm": 0.7018902897834778, "learning_rate": 0.0001, "elbo": 3.49861, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.720267417311752}
{"step": 80495, "loss": 3.5082672119140623, "grad_norm": 0.8286707997322083, "learning_rate": 0.0001, "elbo": 3.50827, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7205606380483225}
{"step": 80500, "loss": 3.504222106933594, "grad_norm": 0.7252658605575562, "learning_rate": 0.0001, "elbo": 3.50422, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.720853858784893}
{"step": 80505, "loss": 3.513713073730469, "grad_norm": 0.7219420075416565, "learning_rate": 0.0001, "elbo": 3.51371, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.721147079521463}
{"step": 80510, "loss": 3.5060302734375, "grad_norm": 0.6823296546936035, "learning_rate": 0.0001, "elbo": 3.50603, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.721440300258035}
{"step": 80515, "loss": 3.5081329345703125, "grad_norm": 0.6984823346138, "learning_rate": 0.0001, "elbo": 3.50813, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.721733520994604}
{"step": 80520, "loss": 3.508375549316406, "grad_norm": 0.7754910588264465, "learning_rate": 0.0001, "elbo": 3.50838, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.722026741731176}
{"step": 80525, "loss": 3.502716064453125, "grad_norm": 0.7095422148704529, "learning_rate": 0.0001, "elbo": 3.50272, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.722319962467746}
{"step": 80530, "loss": 3.5059463500976564, "grad_norm": 0.7066531777381897, "learning_rate": 0.0001, "elbo": 3.50595, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.722613183204317}
{"step": 80535, "loss": 3.508340835571289, "grad_norm": 0.710012674331665, "learning_rate": 0.0001, "elbo": 3.50834, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.722906403940887}
{"step": 80540, "loss": 3.4968162536621095, "grad_norm": 0.8566920161247253, "learning_rate": 0.0001, "elbo": 3.49682, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7231996246774575}
{"step": 80545, "loss": 3.5175121307373045, "grad_norm": 0.6883630752563477, "learning_rate": 0.0001, "elbo": 3.51751, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.723492845414028}
{"step": 80550, "loss": 3.4865623474121095, "grad_norm": 0.7035428285598755, "learning_rate": 0.0001, "elbo": 3.48656, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.723786066150598}
{"step": 80555, "loss": 3.487234878540039, "grad_norm": 0.6732141375541687, "learning_rate": 0.0001, "elbo": 3.48723, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.724079286887169}
{"step": 80555, "loss": 3.487234878540039, "grad_norm": 0.6732141375541687, "learning_rate": 0.0001, "elbo": 3.48723, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.724079286887169}
{"step": 80560, "loss": 3.4851638793945314, "grad_norm": 0.797953724861145, "learning_rate": 0.0001, "elbo": 3.48516, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.724372507623739}
{"step": 80565, "loss": 3.5081607818603517, "grad_norm": 0.7669845223426819, "learning_rate": 0.0001, "elbo": 3.50816, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.72466572836031}
{"step": 80570, "loss": 3.509848785400391, "grad_norm": 0.6835321187973022, "learning_rate": 0.0001, "elbo": 3.50985, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.72495894909688}
{"step": 80575, "loss": 3.5011550903320314, "grad_norm": 0.7014092206954956, "learning_rate": 0.0001, "elbo": 3.50116, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.725252169833451}
{"step": 80580, "loss": 3.4971893310546873, "grad_norm": 0.8185959458351135, "learning_rate": 0.0001, "elbo": 3.49719, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.725545390570021}
{"step": 80585, "loss": 3.510036087036133, "grad_norm": 0.7666522264480591, "learning_rate": 0.0001, "elbo": 3.51004, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7258386113065916}
{"step": 80590, "loss": 3.497933197021484, "grad_norm": 0.7138675451278687, "learning_rate": 0.0001, "elbo": 3.49793, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.726131832043162}
{"step": 80595, "loss": 3.499043273925781, "grad_norm": 0.7986481189727783, "learning_rate": 0.0001, "elbo": 3.49904, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7264250527797325}
{"step": 80600, "loss": 3.4977725982666015, "grad_norm": 0.717441201210022, "learning_rate": 0.0001, "elbo": 3.49777, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.726718273516303}
{"step": 80605, "loss": 3.491656494140625, "grad_norm": 0.7535456418991089, "learning_rate": 0.0001, "elbo": 3.49166, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.727011494252873}
{"step": 80610, "loss": 3.5144561767578124, "grad_norm": 0.7085912227630615, "learning_rate": 0.0001, "elbo": 3.51446, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.727304714989444}
{"step": 80615, "loss": 3.5001415252685546, "grad_norm": 0.8064004182815552, "learning_rate": 0.0001, "elbo": 3.50014, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.727597935726014}
{"step": 80620, "loss": 3.483590316772461, "grad_norm": 0.713679313659668, "learning_rate": 0.0001, "elbo": 3.48359, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.727891156462585}
{"step": 80625, "loss": 3.49853630065918, "grad_norm": 0.7907750010490417, "learning_rate": 0.0001, "elbo": 3.49854, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.728184377199155}
{"step": 80630, "loss": 3.4925857543945313, "grad_norm": 0.8902507424354553, "learning_rate": 0.0001, "elbo": 3.49259, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.728477597935726}
{"step": 80635, "loss": 3.4960037231445313, "grad_norm": 0.7021297216415405, "learning_rate": 0.0001, "elbo": 3.496, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.728770818672297}
{"step": 80640, "loss": 3.50035400390625, "grad_norm": 0.7209758162498474, "learning_rate": 0.0001, "elbo": 3.50035, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7290640394088665}
{"step": 80645, "loss": 3.5008316040039062, "grad_norm": 0.7918068170547485, "learning_rate": 0.0001, "elbo": 3.50083, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.729357260145438}
{"step": 80650, "loss": 3.5053539276123047, "grad_norm": 0.7014638185501099, "learning_rate": 0.0001, "elbo": 3.50535, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.729650480882008}
{"step": 80655, "loss": 3.5103309631347654, "grad_norm": 0.7220627069473267, "learning_rate": 0.0001, "elbo": 3.51033, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.729943701618579}
{"step": 80660, "loss": 3.503144073486328, "grad_norm": 0.740580141544342, "learning_rate": 0.0001, "elbo": 3.50314, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.730236922355149}
{"step": 80665, "loss": 3.5188350677490234, "grad_norm": 0.9071096777915955, "learning_rate": 0.0001, "elbo": 3.51884, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.73053014309172}
{"step": 80670, "loss": 3.497059631347656, "grad_norm": 0.7894760966300964, "learning_rate": 0.0001, "elbo": 3.49706, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.73082336382829}
{"step": 80675, "loss": 3.5131294250488283, "grad_norm": 0.6582077145576477, "learning_rate": 0.0001, "elbo": 3.51313, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.731116584564861}
{"step": 80680, "loss": 3.510456848144531, "grad_norm": 0.8245974779129028, "learning_rate": 0.0001, "elbo": 3.51046, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.731409805301431}
{"step": 80685, "loss": 3.5057464599609376, "grad_norm": 0.7296723127365112, "learning_rate": 0.0001, "elbo": 3.50575, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7317030260380015}
{"step": 80690, "loss": 3.4961814880371094, "grad_norm": 0.7541786432266235, "learning_rate": 0.0001, "elbo": 3.49618, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.731996246774572}
{"step": 80695, "loss": 3.4901378631591795, "grad_norm": 0.7138178944587708, "learning_rate": 0.0001, "elbo": 3.49014, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.732289467511142}
{"step": 80700, "loss": 3.5151771545410155, "grad_norm": 0.7599385380744934, "learning_rate": 0.0001, "elbo": 3.51518, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.732582688247713}
{"step": 80705, "loss": 3.5197681427001952, "grad_norm": 0.7053600549697876, "learning_rate": 0.0001, "elbo": 3.51977, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.732875908984283}
{"step": 80710, "loss": 3.490127944946289, "grad_norm": 0.7273715734481812, "learning_rate": 0.0001, "elbo": 3.49013, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.733169129720854}
{"step": 80715, "loss": 3.5041412353515624, "grad_norm": 0.7199766039848328, "learning_rate": 0.0001, "elbo": 3.50414, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.733462350457424}
{"step": 80720, "loss": 3.509014129638672, "grad_norm": 1.109266757965088, "learning_rate": 0.0001, "elbo": 3.50901, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.733755571193995}
{"step": 80725, "loss": 3.5031944274902345, "grad_norm": 0.7105690836906433, "learning_rate": 0.0001, "elbo": 3.50319, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.734048791930565}
{"step": 80730, "loss": 3.494002914428711, "grad_norm": 0.736901044845581, "learning_rate": 0.0001, "elbo": 3.494, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.734342012667136}
{"step": 80735, "loss": 3.493798828125, "grad_norm": 0.7108481526374817, "learning_rate": 0.0001, "elbo": 3.4938, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.734635233403706}
{"step": 80740, "loss": 3.5131221771240235, "grad_norm": 0.7068557143211365, "learning_rate": 0.0001, "elbo": 3.51312, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7349284541402765}
{"step": 80745, "loss": 3.493510437011719, "grad_norm": 0.7726331949234009, "learning_rate": 0.0001, "elbo": 3.49351, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.735221674876847}
{"step": 80750, "loss": 3.502532958984375, "grad_norm": 0.7041304111480713, "learning_rate": 0.0001, "elbo": 3.50253, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.735514895613417}
{"step": 80755, "loss": 3.502140426635742, "grad_norm": 0.6985434889793396, "learning_rate": 0.0001, "elbo": 3.50214, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.735808116349988}
{"step": 80760, "loss": 3.515670394897461, "grad_norm": 0.7628113031387329, "learning_rate": 0.0001, "elbo": 3.51567, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.736101337086559}
{"step": 80765, "loss": 3.490566635131836, "grad_norm": 0.7250422239303589, "learning_rate": 0.0001, "elbo": 3.49057, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.736394557823129}
{"step": 80770, "loss": 3.506245422363281, "grad_norm": 0.7166914939880371, "learning_rate": 0.0001, "elbo": 3.50625, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7366877785597}
{"step": 80775, "loss": 3.500701904296875, "grad_norm": 0.7302894592285156, "learning_rate": 0.0001, "elbo": 3.5007, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.736980999296271}
{"step": 80780, "loss": 3.497336196899414, "grad_norm": 0.6858701109886169, "learning_rate": 0.0001, "elbo": 3.49734, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.737274220032841}
{"step": 80785, "loss": 3.5040077209472655, "grad_norm": 0.8908755779266357, "learning_rate": 0.0001, "elbo": 3.50401, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7375674407694115}
{"step": 80790, "loss": 3.5123321533203127, "grad_norm": 0.685218334197998, "learning_rate": 0.0001, "elbo": 3.51233, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.737860661505982}
{"step": 80795, "loss": 3.5234840393066404, "grad_norm": 0.7394178509712219, "learning_rate": 0.0001, "elbo": 3.52348, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.738153882242552}
{"step": 80800, "loss": 3.5182228088378906, "grad_norm": 0.8113955855369568, "learning_rate": 0.0001, "elbo": 3.51822, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.738447102979123}
{"step": 80805, "loss": 3.51695556640625, "grad_norm": 0.8080675601959229, "learning_rate": 0.0001, "elbo": 3.51696, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.738740323715693}
{"step": 80810, "loss": 3.5058692932128905, "grad_norm": 0.7116182446479797, "learning_rate": 0.0001, "elbo": 3.50587, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.739033544452264}
{"step": 80815, "loss": 3.496891403198242, "grad_norm": 0.7434635758399963, "learning_rate": 0.0001, "elbo": 3.49689, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.739326765188834}
{"step": 80820, "loss": 3.4970272064208983, "grad_norm": 0.8492260575294495, "learning_rate": 0.0001, "elbo": 3.49703, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.739619985925405}
{"step": 80825, "loss": 3.504743957519531, "grad_norm": 0.7256568074226379, "learning_rate": 0.0001, "elbo": 3.50474, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.739913206661975}
{"step": 80830, "loss": 3.5225852966308593, "grad_norm": 0.7252415418624878, "learning_rate": 0.0001, "elbo": 3.52259, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.740206427398546}
{"step": 80835, "loss": 3.50633544921875, "grad_norm": 0.7628656029701233, "learning_rate": 0.0001, "elbo": 3.50634, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.740499648135116}
{"step": 80840, "loss": 3.501628875732422, "grad_norm": 0.6921344995498657, "learning_rate": 0.0001, "elbo": 3.50163, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7407928688716865}
{"step": 80845, "loss": 3.4978370666503906, "grad_norm": 0.7004403471946716, "learning_rate": 0.0001, "elbo": 3.49784, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.741086089608257}
{"step": 80850, "loss": 3.4946224212646486, "grad_norm": 0.7184292078018188, "learning_rate": 0.0001, "elbo": 3.49462, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.741379310344827}
{"step": 80855, "loss": 3.4971992492675783, "grad_norm": 0.7204610705375671, "learning_rate": 0.0001, "elbo": 3.4972, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.741672531081398}
{"step": 80860, "loss": 3.499704360961914, "grad_norm": 0.7518836855888367, "learning_rate": 0.0001, "elbo": 3.4997, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.741965751817968}
{"step": 80865, "loss": 3.4971477508544924, "grad_norm": 0.6814929842948914, "learning_rate": 0.0001, "elbo": 3.49715, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.742258972554539}
{"step": 80870, "loss": 3.500994110107422, "grad_norm": 0.7471417188644409, "learning_rate": 0.0001, "elbo": 3.50099, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.742552193291109}
{"step": 80875, "loss": 3.506906509399414, "grad_norm": 0.9906296730041504, "learning_rate": 0.0001, "elbo": 3.50691, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.74284541402768}
{"step": 80875, "loss": 3.506906509399414, "grad_norm": 0.9906296730041504, "learning_rate": 0.0001, "elbo": 3.50691, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.74284541402768}
{"step": 80880, "loss": 3.51124267578125, "grad_norm": 0.7079421877861023, "learning_rate": 0.0001, "elbo": 3.51124, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.74313863476425}
{"step": 80885, "loss": 3.5034759521484373, "grad_norm": 0.6897385120391846, "learning_rate": 0.0001, "elbo": 3.50348, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7434318555008215}
{"step": 80890, "loss": 3.5028831481933596, "grad_norm": 0.7242690920829773, "learning_rate": 0.0001, "elbo": 3.50288, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.743725076237391}
{"step": 80895, "loss": 3.497426986694336, "grad_norm": 0.7230762243270874, "learning_rate": 0.0001, "elbo": 3.49743, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.744018296973962}
{"step": 80900, "loss": 3.4874008178710936, "grad_norm": 0.7737448215484619, "learning_rate": 0.0001, "elbo": 3.4874, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.744311517710533}
{"step": 80905, "loss": 3.505484771728516, "grad_norm": 0.7859628796577454, "learning_rate": 0.0001, "elbo": 3.50548, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.744604738447103}
{"step": 80910, "loss": 3.4913352966308593, "grad_norm": 0.9438437819480896, "learning_rate": 0.0001, "elbo": 3.49133, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.744897959183674}
{"step": 80915, "loss": 3.5170059204101562, "grad_norm": 0.681056797504425, "learning_rate": 0.0001, "elbo": 3.51701, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.745191179920244}
{"step": 80920, "loss": 3.5043590545654295, "grad_norm": 0.7064907550811768, "learning_rate": 0.0001, "elbo": 3.50436, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.745484400656815}
{"step": 80925, "loss": 3.509259796142578, "grad_norm": 0.7054713368415833, "learning_rate": 0.0001, "elbo": 3.50926, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.745777621393385}
{"step": 80930, "loss": 3.503919219970703, "grad_norm": 0.721832275390625, "learning_rate": 0.0001, "elbo": 3.50392, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.746070842129956}
{"step": 80935, "loss": 3.496603012084961, "grad_norm": 0.7957392930984497, "learning_rate": 0.0001, "elbo": 3.4966, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.746364062866526}
{"step": 80940, "loss": 3.4966121673583985, "grad_norm": 0.7122108936309814, "learning_rate": 0.0001, "elbo": 3.49661, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7466572836030965}
{"step": 80945, "loss": 3.513296127319336, "grad_norm": 0.7048272490501404, "learning_rate": 0.0001, "elbo": 3.5133, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.746950504339667}
{"step": 80945, "loss": 3.513296127319336, "grad_norm": 0.7048272490501404, "learning_rate": 0.0001, "elbo": 3.5133, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.746950504339667}
{"step": 80950, "loss": 3.5004364013671876, "grad_norm": 0.7433361411094666, "learning_rate": 0.0001, "elbo": 3.50044, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.747243725076237}
{"step": 80955, "loss": 3.506661224365234, "grad_norm": 0.6854705810546875, "learning_rate": 0.0001, "elbo": 3.50666, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.747536945812808}
{"step": 80960, "loss": 3.505604553222656, "grad_norm": 0.753872275352478, "learning_rate": 0.0001, "elbo": 3.5056, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.747830166549378}
{"step": 80965, "loss": 3.5066513061523437, "grad_norm": 0.9029955267906189, "learning_rate": 0.0001, "elbo": 3.50665, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.748123387285949}
{"step": 80970, "loss": 3.48663215637207, "grad_norm": 0.7462477087974548, "learning_rate": 0.0001, "elbo": 3.48663, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.748416608022519}
{"step": 80975, "loss": 3.5123111724853517, "grad_norm": 0.7869090437889099, "learning_rate": 0.0001, "elbo": 3.51231, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.74870982875909}
{"step": 80980, "loss": 3.49384765625, "grad_norm": 0.879457950592041, "learning_rate": 0.0001, "elbo": 3.49385, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.74900304949566}
{"step": 80985, "loss": 3.5053829193115233, "grad_norm": 0.745499849319458, "learning_rate": 0.0001, "elbo": 3.50538, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7492962702322306}
{"step": 80990, "loss": 3.510713577270508, "grad_norm": 0.6862531304359436, "learning_rate": 0.0001, "elbo": 3.51071, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.749589490968801}
{"step": 80995, "loss": 3.4877178192138674, "grad_norm": 0.7166483998298645, "learning_rate": 0.0001, "elbo": 3.48772, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7498827117053715}
{"step": 81000, "loss": 3.5140403747558593, "grad_norm": 0.7169978618621826, "learning_rate": 0.0001, "elbo": 3.51404, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.750175932441942}
{"step": 81005, "loss": 3.492778015136719, "grad_norm": 0.6918312311172485, "learning_rate": 0.0001, "elbo": 3.49278, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.750469153178512}
{"step": 81010, "loss": 3.497333526611328, "grad_norm": 0.7096925377845764, "learning_rate": 0.0001, "elbo": 3.49733, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.750762373915084}
{"step": 81015, "loss": 3.5110668182373046, "grad_norm": 0.6804244518280029, "learning_rate": 0.0001, "elbo": 3.51107, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.751055594651653}
{"step": 81020, "loss": 3.492418670654297, "grad_norm": 0.6556565165519714, "learning_rate": 0.0001, "elbo": 3.49242, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.751348815388225}
{"step": 81025, "loss": 3.500133514404297, "grad_norm": 1.3438676595687866, "learning_rate": 0.0001, "elbo": 3.50013, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.751642036124795}
{"step": 81030, "loss": 3.5043006896972657, "grad_norm": 0.7025637030601501, "learning_rate": 0.0001, "elbo": 3.5043, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7519352568613655}
{"step": 81035, "loss": 3.5136077880859373, "grad_norm": 0.7614501714706421, "learning_rate": 0.0001, "elbo": 3.51361, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.752228477597936}
{"step": 81040, "loss": 3.497264862060547, "grad_norm": 0.6858429312705994, "learning_rate": 0.0001, "elbo": 3.49726, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.752521698334506}
{"step": 81045, "loss": 3.5175216674804686, "grad_norm": 0.7083976864814758, "learning_rate": 0.0001, "elbo": 3.51752, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.752814919071077}
{"step": 81050, "loss": 3.4855396270751955, "grad_norm": 0.6913496255874634, "learning_rate": 0.0001, "elbo": 3.48554, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.753108139807647}
{"step": 81055, "loss": 3.5117603302001954, "grad_norm": 0.7357509136199951, "learning_rate": 0.0001, "elbo": 3.51176, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.753401360544218}
{"step": 81060, "loss": 3.4931709289550783, "grad_norm": 0.7327359914779663, "learning_rate": 0.0001, "elbo": 3.49317, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.753694581280788}
{"step": 81065, "loss": 3.508892822265625, "grad_norm": 0.7244605422019958, "learning_rate": 0.0001, "elbo": 3.50889, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.753987802017359}
{"step": 81070, "loss": 3.497550201416016, "grad_norm": 0.7260718941688538, "learning_rate": 0.0001, "elbo": 3.49755, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.754281022753929}
{"step": 81075, "loss": 3.509865570068359, "grad_norm": 0.6839861273765564, "learning_rate": 0.0001, "elbo": 3.50987, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7545742434905}
{"step": 81080, "loss": 3.485575866699219, "grad_norm": 0.8772090673446655, "learning_rate": 0.0001, "elbo": 3.48558, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.75486746422707}
{"step": 81085, "loss": 3.483389663696289, "grad_norm": 0.7461671829223633, "learning_rate": 0.0001, "elbo": 3.48339, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7551606849636405}
{"step": 81090, "loss": 3.4950523376464844, "grad_norm": 0.6828694939613342, "learning_rate": 0.0001, "elbo": 3.49505, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.755453905700211}
{"step": 81095, "loss": 3.4887344360351564, "grad_norm": 0.7842863202095032, "learning_rate": 0.0001, "elbo": 3.48873, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.755747126436781}
{"step": 81100, "loss": 3.482678985595703, "grad_norm": 0.7780275940895081, "learning_rate": 0.0001, "elbo": 3.48268, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.756040347173352}
{"step": 81105, "loss": 3.5016082763671874, "grad_norm": 0.7706718444824219, "learning_rate": 0.0001, "elbo": 3.50161, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.756333567909922}
{"step": 81110, "loss": 3.5046485900878905, "grad_norm": 0.7613305449485779, "learning_rate": 0.0001, "elbo": 3.50465, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.756626788646493}
{"step": 81115, "loss": 3.4876567840576174, "grad_norm": 0.7154537439346313, "learning_rate": 0.0001, "elbo": 3.48766, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.756920009383063}
{"step": 81120, "loss": 3.4850730895996094, "grad_norm": 0.702334463596344, "learning_rate": 0.0001, "elbo": 3.48507, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.757213230119634}
{"step": 81125, "loss": 3.5071571350097654, "grad_norm": 0.7146297693252563, "learning_rate": 0.0001, "elbo": 3.50716, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.757506450856205}
{"step": 81130, "loss": 3.509404754638672, "grad_norm": 0.7404878735542297, "learning_rate": 0.0001, "elbo": 3.50941, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.757799671592775}
{"step": 81135, "loss": 3.5081031799316404, "grad_norm": 0.810817539691925, "learning_rate": 0.0001, "elbo": 3.5081, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.758092892329346}
{"step": 81140, "loss": 3.500945281982422, "grad_norm": 0.7275287508964539, "learning_rate": 0.0001, "elbo": 3.50095, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7583861130659155}
{"step": 81145, "loss": 3.504119873046875, "grad_norm": 0.766973614692688, "learning_rate": 0.0001, "elbo": 3.50412, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.758679333802487}
{"step": 81150, "loss": 3.4733219146728516, "grad_norm": 0.7373847365379333, "learning_rate": 0.0001, "elbo": 3.47332, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.758972554539057}
{"step": 81155, "loss": 3.4901031494140624, "grad_norm": 0.7375724911689758, "learning_rate": 0.0001, "elbo": 3.4901, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.759265775275628}
{"step": 81160, "loss": 3.5237075805664064, "grad_norm": 0.7014908790588379, "learning_rate": 0.0001, "elbo": 3.52371, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.759558996012198}
{"step": 81165, "loss": 3.5087001800537108, "grad_norm": 0.6885985136032104, "learning_rate": 0.0001, "elbo": 3.5087, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.759852216748769}
{"step": 81170, "loss": 3.5048377990722654, "grad_norm": 0.73432457447052, "learning_rate": 0.0001, "elbo": 3.50484, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.760145437485339}
{"step": 81175, "loss": 3.5058830261230467, "grad_norm": 0.8334564566612244, "learning_rate": 0.0001, "elbo": 3.50588, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.76043865822191}
{"step": 81180, "loss": 3.5103328704833983, "grad_norm": 1.1642324924468994, "learning_rate": 0.0001, "elbo": 3.51033, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.76073187895848}
{"step": 81185, "loss": 3.5123870849609373, "grad_norm": 0.7151902914047241, "learning_rate": 0.0001, "elbo": 3.51239, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7610250996950505}
{"step": 81190, "loss": 3.5126007080078123, "grad_norm": 0.7107980251312256, "learning_rate": 0.0001, "elbo": 3.5126, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.761318320431621}
{"step": 81195, "loss": 3.499658966064453, "grad_norm": 0.7485274076461792, "learning_rate": 0.0001, "elbo": 3.49966, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.761611541168191}
{"step": 81200, "loss": 3.503862762451172, "grad_norm": 0.7513425946235657, "learning_rate": 0.0001, "elbo": 3.50386, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.761904761904762}
{"step": 81205, "loss": 3.491870880126953, "grad_norm": 0.7342187762260437, "learning_rate": 0.0001, "elbo": 3.49187, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.762197982641332}
{"step": 81210, "loss": 3.5023128509521486, "grad_norm": 0.8998834490776062, "learning_rate": 0.0001, "elbo": 3.50231, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.762491203377903}
{"step": 81215, "loss": 3.5066490173339844, "grad_norm": 0.7190325856208801, "learning_rate": 0.0001, "elbo": 3.50665, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.762784424114473}
{"step": 81220, "loss": 3.5012786865234373, "grad_norm": 0.8047042489051819, "learning_rate": 0.0001, "elbo": 3.50128, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.763077644851044}
{"step": 81225, "loss": 3.508788299560547, "grad_norm": 0.7002905607223511, "learning_rate": 0.0001, "elbo": 3.50879, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.763370865587614}
{"step": 81230, "loss": 3.498566436767578, "grad_norm": 0.7783265709877014, "learning_rate": 0.0001, "elbo": 3.49857, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.763664086324185}
{"step": 81235, "loss": 3.5026275634765627, "grad_norm": 0.6922489404678345, "learning_rate": 0.0001, "elbo": 3.50263, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.763957307060755}
{"step": 81240, "loss": 3.516621398925781, "grad_norm": 0.7735004425048828, "learning_rate": 0.0001, "elbo": 3.51662, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7642505277973255}
{"step": 81245, "loss": 3.5106075286865233, "grad_norm": 0.7369652986526489, "learning_rate": 0.0001, "elbo": 3.51061, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.764543748533896}
{"step": 81250, "loss": 3.4866336822509765, "grad_norm": 0.7752947211265564, "learning_rate": 0.0001, "elbo": 3.48663, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.764836969270467}
{"step": 81255, "loss": 3.5082374572753907, "grad_norm": 0.7351750731468201, "learning_rate": 0.0001, "elbo": 3.50824, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.765130190007037}
{"step": 81260, "loss": 3.5075393676757813, "grad_norm": 0.7953506708145142, "learning_rate": 0.0001, "elbo": 3.50754, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.765423410743608}
{"step": 81265, "loss": 3.4936614990234376, "grad_norm": 0.7182575464248657, "learning_rate": 0.0001, "elbo": 3.49366, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.765716631480179}
{"step": 81270, "loss": 3.5235084533691405, "grad_norm": 0.9633017778396606, "learning_rate": 0.0001, "elbo": 3.52351, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.766009852216749}
{"step": 81275, "loss": 3.5102649688720704, "grad_norm": 0.7076072692871094, "learning_rate": 0.0001, "elbo": 3.51026, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.76630307295332}
{"step": 81280, "loss": 3.5068565368652345, "grad_norm": 0.7452673316001892, "learning_rate": 0.0001, "elbo": 3.50686, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.76659629368989}
{"step": 81285, "loss": 3.500706100463867, "grad_norm": 0.7297503352165222, "learning_rate": 0.0001, "elbo": 3.50071, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7668895144264605}
{"step": 81290, "loss": 3.5100555419921875, "grad_norm": 0.9290285706520081, "learning_rate": 0.0001, "elbo": 3.51006, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.767182735163031}
{"step": 81295, "loss": 3.5051429748535154, "grad_norm": 1.0400407314300537, "learning_rate": 0.0001, "elbo": 3.50514, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.767475955899601}
{"step": 81300, "loss": 3.497272491455078, "grad_norm": 0.9774971604347229, "learning_rate": 0.0001, "elbo": 3.49727, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.767769176636172}
{"step": 81305, "loss": 3.48632926940918, "grad_norm": 0.7172747850418091, "learning_rate": 0.0001, "elbo": 3.48633, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.768062397372742}
{"step": 81310, "loss": 3.502020263671875, "grad_norm": 0.7486938238143921, "learning_rate": 0.0001, "elbo": 3.50202, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.768355618109313}
{"step": 81315, "loss": 3.5042869567871096, "grad_norm": 0.7072286009788513, "learning_rate": 0.0001, "elbo": 3.50429, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.768648838845883}
{"step": 81320, "loss": 3.488883209228516, "grad_norm": 0.7452565431594849, "learning_rate": 0.0001, "elbo": 3.48888, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.768942059582454}
{"step": 81325, "loss": 3.4826629638671873, "grad_norm": 0.6609402298927307, "learning_rate": 0.0001, "elbo": 3.48266, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.769235280319024}
{"step": 81330, "loss": 3.4939319610595705, "grad_norm": 0.7927611470222473, "learning_rate": 0.0001, "elbo": 3.49393, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.769528501055595}
{"step": 81335, "loss": 3.4948650360107423, "grad_norm": 0.7226362228393555, "learning_rate": 0.0001, "elbo": 3.49486, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.769821721792165}
{"step": 81340, "loss": 3.506191635131836, "grad_norm": 0.7594317197799683, "learning_rate": 0.0001, "elbo": 3.50619, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7701149425287355}
{"step": 81345, "loss": 3.496697998046875, "grad_norm": 0.7336443066596985, "learning_rate": 0.0001, "elbo": 3.4967, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.770408163265306}
{"step": 81350, "loss": 3.5011886596679687, "grad_norm": 0.7419514656066895, "learning_rate": 0.0001, "elbo": 3.50119, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.770701384001876}
{"step": 81355, "loss": 3.499958801269531, "grad_norm": 0.8463663458824158, "learning_rate": 0.0001, "elbo": 3.49996, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.770994604738447}
{"step": 81360, "loss": 3.500636672973633, "grad_norm": 0.7123203277587891, "learning_rate": 0.0001, "elbo": 3.50064, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.771287825475017}
{"step": 81365, "loss": 3.4902252197265624, "grad_norm": 0.935506284236908, "learning_rate": 0.0001, "elbo": 3.49023, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.771581046211588}
{"step": 81370, "loss": 3.495531463623047, "grad_norm": 0.7063588500022888, "learning_rate": 0.0001, "elbo": 3.49553, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.771874266948158}
{"step": 81375, "loss": 3.5180679321289063, "grad_norm": 0.6880717277526855, "learning_rate": 0.0001, "elbo": 3.51807, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7721674876847295}
{"step": 81380, "loss": 3.508832550048828, "grad_norm": 0.8815809488296509, "learning_rate": 0.0001, "elbo": 3.50883, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.772460708421299}
{"step": 81385, "loss": 3.4928962707519533, "grad_norm": 0.8418274521827698, "learning_rate": 0.0001, "elbo": 3.4929, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7727539291578704}
{"step": 81390, "loss": 3.5109222412109373, "grad_norm": 0.8307536840438843, "learning_rate": 0.0001, "elbo": 3.51092, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.773047149894441}
{"step": 81395, "loss": 3.501880645751953, "grad_norm": 0.7452172636985779, "learning_rate": 0.0001, "elbo": 3.50188, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.773340370631011}
{"step": 81400, "loss": 3.520447540283203, "grad_norm": 0.7202000617980957, "learning_rate": 0.0001, "elbo": 3.52045, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.773633591367582}
{"step": 81405, "loss": 3.510955047607422, "grad_norm": 0.8002910614013672, "learning_rate": 0.0001, "elbo": 3.51095, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.773926812104152}
{"step": 81410, "loss": 3.4980117797851564, "grad_norm": 0.7316591739654541, "learning_rate": 0.0001, "elbo": 3.49801, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.774220032840723}
{"step": 81415, "loss": 3.5083454132080076, "grad_norm": 0.7618487477302551, "learning_rate": 0.0001, "elbo": 3.50835, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.774513253577293}
{"step": 81420, "loss": 3.5091148376464845, "grad_norm": 0.7439682483673096, "learning_rate": 0.0001, "elbo": 3.50912, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.774806474313864}
{"step": 81425, "loss": 3.490831756591797, "grad_norm": 0.7531256079673767, "learning_rate": 0.0001, "elbo": 3.49083, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.775099695050434}
{"step": 81430, "loss": 3.51550178527832, "grad_norm": 0.706790566444397, "learning_rate": 0.0001, "elbo": 3.5155, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7753929157870045}
{"step": 81435, "loss": 3.513280487060547, "grad_norm": 0.6797500848770142, "learning_rate": 0.0001, "elbo": 3.51328, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.775686136523575}
{"step": 81440, "loss": 3.5017425537109377, "grad_norm": 0.7154797315597534, "learning_rate": 0.0001, "elbo": 3.50174, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.775979357260145}
{"step": 81445, "loss": 3.4860794067382814, "grad_norm": 0.6999051570892334, "learning_rate": 0.0001, "elbo": 3.48608, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.776272577996716}
{"step": 81450, "loss": 3.4899307250976563, "grad_norm": 0.7503952980041504, "learning_rate": 0.0001, "elbo": 3.48993, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.776565798733286}
{"step": 81455, "loss": 3.5038909912109375, "grad_norm": 0.6981238126754761, "learning_rate": 0.0001, "elbo": 3.50389, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.776859019469857}
{"step": 81460, "loss": 3.4886871337890626, "grad_norm": 0.7500169277191162, "learning_rate": 0.0001, "elbo": 3.48869, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.777152240206427}
{"step": 81465, "loss": 3.4997932434082033, "grad_norm": 0.7272275686264038, "learning_rate": 0.0001, "elbo": 3.49979, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.777445460942998}
{"step": 81470, "loss": 3.4964439392089846, "grad_norm": 0.7668419480323792, "learning_rate": 0.0001, "elbo": 3.49644, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.777738681679568}
{"step": 81475, "loss": 3.5136852264404297, "grad_norm": 0.6938400864601135, "learning_rate": 0.0001, "elbo": 3.51369, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.778031902416139}
{"step": 81480, "loss": 3.48260498046875, "grad_norm": 0.7958836555480957, "learning_rate": 0.0001, "elbo": 3.4826, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.778325123152709}
{"step": 81485, "loss": 3.4767913818359375, "grad_norm": 1.183810830116272, "learning_rate": 0.0001, "elbo": 3.47679, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7786183438892795}
{"step": 81490, "loss": 3.5008644104003905, "grad_norm": 0.6933057904243469, "learning_rate": 0.0001, "elbo": 3.50086, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.77891156462585}
{"step": 81495, "loss": 3.5001087188720703, "grad_norm": 0.7337667346000671, "learning_rate": 0.0001, "elbo": 3.50011, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.77920478536242}
{"step": 81500, "loss": 3.491749572753906, "grad_norm": 0.6989210247993469, "learning_rate": 0.0001, "elbo": 3.49175, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.779498006098992}
{"step": 81505, "loss": 3.4935958862304686, "grad_norm": 0.8267550468444824, "learning_rate": 0.0001, "elbo": 3.4936, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.779791226835561}
{"step": 81505, "loss": 3.4935958862304686, "grad_norm": 0.8267550468444824, "learning_rate": 0.0001, "elbo": 3.4936, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.779791226835561}
{"step": 81510, "loss": 3.4878379821777346, "grad_norm": 0.7009814977645874, "learning_rate": 0.0001, "elbo": 3.48784, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.780084447572133}
{"step": 81515, "loss": 3.5112834930419923, "grad_norm": 0.7002226114273071, "learning_rate": 0.0001, "elbo": 3.51128, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.780377668308703}
{"step": 81520, "loss": 3.4871070861816404, "grad_norm": 0.6912429332733154, "learning_rate": 0.0001, "elbo": 3.48711, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.780670889045274}
{"step": 81525, "loss": 3.4950210571289064, "grad_norm": 0.7394366264343262, "learning_rate": 0.0001, "elbo": 3.49502, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.780964109781844}
{"step": 81530, "loss": 3.485179901123047, "grad_norm": 0.7253645062446594, "learning_rate": 0.0001, "elbo": 3.48518, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7812573305184145}
{"step": 81535, "loss": 3.496590423583984, "grad_norm": 0.8186101317405701, "learning_rate": 0.0001, "elbo": 3.49659, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.781550551254985}
{"step": 81540, "loss": 3.515095520019531, "grad_norm": 0.663086473941803, "learning_rate": 0.0001, "elbo": 3.5151, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.781843771991555}
{"step": 81545, "loss": 3.506417465209961, "grad_norm": 0.7212277054786682, "learning_rate": 0.0001, "elbo": 3.50642, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.782136992728126}
{"step": 81550, "loss": 3.5050533294677733, "grad_norm": 0.7012271881103516, "learning_rate": 0.0001, "elbo": 3.50505, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.782430213464696}
{"step": 81555, "loss": 3.4864707946777345, "grad_norm": 0.7357451915740967, "learning_rate": 0.0001, "elbo": 3.48647, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.782723434201267}
{"step": 81560, "loss": 3.4912994384765623, "grad_norm": 0.8433835506439209, "learning_rate": 0.0001, "elbo": 3.4913, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.783016654937837}
{"step": 81565, "loss": 3.4850589752197267, "grad_norm": 0.6999570727348328, "learning_rate": 0.0001, "elbo": 3.48506, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.783309875674408}
{"step": 81570, "loss": 3.5097335815429687, "grad_norm": 0.7262588739395142, "learning_rate": 0.0001, "elbo": 3.50973, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.783603096410978}
{"step": 81570, "loss": 3.5097335815429687, "grad_norm": 0.7262588739395142, "learning_rate": 0.0001, "elbo": 3.50973, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.783603096410978}
{"step": 81575, "loss": 3.4906906127929687, "grad_norm": 0.6993164420127869, "learning_rate": 0.0001, "elbo": 3.49069, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.783896317147549}
{"step": 81580, "loss": 3.4895774841308596, "grad_norm": 0.8368316888809204, "learning_rate": 0.0001, "elbo": 3.48958, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.784189537884119}
{"step": 81585, "loss": 3.5023483276367187, "grad_norm": 0.7577714920043945, "learning_rate": 0.0001, "elbo": 3.50235, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7844827586206895}
{"step": 81590, "loss": 3.497425079345703, "grad_norm": 0.747840166091919, "learning_rate": 0.0001, "elbo": 3.49742, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.78477597935726}
{"step": 81595, "loss": 3.510408401489258, "grad_norm": 0.7853243350982666, "learning_rate": 0.0001, "elbo": 3.51041, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.78506920009383}
{"step": 81595, "loss": 3.510408401489258, "grad_norm": 0.7853243350982666, "learning_rate": 0.0001, "elbo": 3.51041, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.78506920009383}
{"step": 81600, "loss": 3.492967224121094, "grad_norm": 0.8637517094612122, "learning_rate": 0.0001, "elbo": 3.49297, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.785362420830401}
{"step": 81605, "loss": 3.5067825317382812, "grad_norm": 0.6935271620750427, "learning_rate": 0.0001, "elbo": 3.50678, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.785655641566971}
{"step": 81610, "loss": 3.507202911376953, "grad_norm": 0.7463288903236389, "learning_rate": 0.0001, "elbo": 3.5072, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.785948862303542}
{"step": 81615, "loss": 3.505199432373047, "grad_norm": 0.6849341988563538, "learning_rate": 0.0001, "elbo": 3.5052, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.786242083040112}
{"step": 81620, "loss": 3.5068153381347655, "grad_norm": 0.6944092512130737, "learning_rate": 0.0001, "elbo": 3.50682, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.786535303776683}
{"step": 81625, "loss": 3.5117931365966797, "grad_norm": 0.6784520149230957, "learning_rate": 0.0001, "elbo": 3.51179, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.786828524513254}
{"step": 81630, "loss": 3.487129974365234, "grad_norm": 0.8399888277053833, "learning_rate": 0.0001, "elbo": 3.48713, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.787121745249824}
{"step": 81635, "loss": 3.4962310791015625, "grad_norm": 0.8349555134773254, "learning_rate": 0.0001, "elbo": 3.49623, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.787414965986395}
{"step": 81640, "loss": 3.5057090759277343, "grad_norm": 0.7044963836669922, "learning_rate": 0.0001, "elbo": 3.50571, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.787708186722965}
{"step": 81645, "loss": 3.510325241088867, "grad_norm": 0.9657048583030701, "learning_rate": 0.0001, "elbo": 3.51033, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.788001407459536}
{"step": 81650, "loss": 3.5123138427734375, "grad_norm": 0.7138317823410034, "learning_rate": 0.0001, "elbo": 3.51231, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.788294628196106}
{"step": 81655, "loss": 3.5089256286621096, "grad_norm": 0.7272332310676575, "learning_rate": 0.0001, "elbo": 3.50893, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.788587848932677}
{"step": 81660, "loss": 3.4824195861816407, "grad_norm": 0.728778064250946, "learning_rate": 0.0001, "elbo": 3.48242, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.788881069669247}
{"step": 81665, "loss": 3.5009326934814453, "grad_norm": 0.7014809250831604, "learning_rate": 0.0001, "elbo": 3.50093, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.789174290405818}
{"step": 81670, "loss": 3.491961669921875, "grad_norm": 0.6962727904319763, "learning_rate": 0.0001, "elbo": 3.49196, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.789467511142388}
{"step": 81675, "loss": 3.4866706848144533, "grad_norm": 0.7462141513824463, "learning_rate": 0.0001, "elbo": 3.48667, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.789760731878959}
{"step": 81680, "loss": 3.5035175323486327, "grad_norm": 0.7117552161216736, "learning_rate": 0.0001, "elbo": 3.50352, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.790053952615529}
{"step": 81685, "loss": 3.485718536376953, "grad_norm": 0.7013458609580994, "learning_rate": 0.0001, "elbo": 3.48572, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7903471733520995}
{"step": 81690, "loss": 3.506427001953125, "grad_norm": 0.8272136449813843, "learning_rate": 0.0001, "elbo": 3.50643, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.79064039408867}
{"step": 81695, "loss": 3.495350646972656, "grad_norm": 0.7866753339767456, "learning_rate": 0.0001, "elbo": 3.49535, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.79093361482524}
{"step": 81700, "loss": 3.501129150390625, "grad_norm": 0.7689346075057983, "learning_rate": 0.0001, "elbo": 3.50113, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.791226835561811}
{"step": 81705, "loss": 3.487163543701172, "grad_norm": 0.7195982933044434, "learning_rate": 0.0001, "elbo": 3.48716, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.791520056298381}
{"step": 81710, "loss": 3.4969223022460936, "grad_norm": 0.7117783427238464, "learning_rate": 0.0001, "elbo": 3.49692, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.791813277034952}
{"step": 81715, "loss": 3.4927047729492187, "grad_norm": 0.8192117214202881, "learning_rate": 0.0001, "elbo": 3.4927, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.792106497771522}
{"step": 81720, "loss": 3.5062911987304686, "grad_norm": 0.7701025605201721, "learning_rate": 0.0001, "elbo": 3.50629, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.792399718508093}
{"step": 81725, "loss": 3.4941864013671875, "grad_norm": 0.7114535570144653, "learning_rate": 0.0001, "elbo": 3.49419, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.792692939244663}
{"step": 81730, "loss": 3.510194778442383, "grad_norm": 0.6787610650062561, "learning_rate": 0.0001, "elbo": 3.51019, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.792986159981234}
{"step": 81735, "loss": 3.4921363830566405, "grad_norm": 0.7115873694419861, "learning_rate": 0.0001, "elbo": 3.49214, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.793279380717804}
{"step": 81740, "loss": 3.4986595153808593, "grad_norm": 0.7456432580947876, "learning_rate": 0.0001, "elbo": 3.49866, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7935726014543745}
{"step": 81745, "loss": 3.5158340454101564, "grad_norm": 0.7503935098648071, "learning_rate": 0.0001, "elbo": 3.51583, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.793865822190945}
{"step": 81750, "loss": 3.505975341796875, "grad_norm": 0.7094406485557556, "learning_rate": 0.0001, "elbo": 3.50598, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.794159042927516}
{"step": 81755, "loss": 3.509225845336914, "grad_norm": 0.6918758153915405, "learning_rate": 0.0001, "elbo": 3.50923, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.794452263664086}
{"step": 81760, "loss": 3.510741424560547, "grad_norm": 0.7054944634437561, "learning_rate": 0.0001, "elbo": 3.51074, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.794745484400657}
{"step": 81765, "loss": 3.4907283782958984, "grad_norm": 1.1118394136428833, "learning_rate": 0.0001, "elbo": 3.49073, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.795038705137228}
{"step": 81770, "loss": 3.5020572662353517, "grad_norm": 0.8387624025344849, "learning_rate": 0.0001, "elbo": 3.50206, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.795331925873798}
{"step": 81775, "loss": 3.494287872314453, "grad_norm": 0.6793244481086731, "learning_rate": 0.0001, "elbo": 3.49429, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7956251466103685}
{"step": 81780, "loss": 3.5032867431640624, "grad_norm": 0.6984506845474243, "learning_rate": 0.0001, "elbo": 3.50329, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.795918367346939}
{"step": 81785, "loss": 3.4923690795898437, "grad_norm": 0.7119935154914856, "learning_rate": 0.0001, "elbo": 3.49237, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7962115880835094}
{"step": 81790, "loss": 3.5100631713867188, "grad_norm": 0.684182345867157, "learning_rate": 0.0001, "elbo": 3.51006, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.79650480882008}
{"step": 81795, "loss": 3.5050838470458983, "grad_norm": 0.7317017912864685, "learning_rate": 0.0001, "elbo": 3.50508, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.79679802955665}
{"step": 81800, "loss": 3.4949745178222655, "grad_norm": 0.751891553401947, "learning_rate": 0.0001, "elbo": 3.49497, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.797091250293221}
{"step": 81805, "loss": 3.503062438964844, "grad_norm": 0.7831185460090637, "learning_rate": 0.0001, "elbo": 3.50306, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.797384471029791}
{"step": 81810, "loss": 3.4940086364746095, "grad_norm": 0.7062956094741821, "learning_rate": 0.0001, "elbo": 3.49401, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.797677691766362}
{"step": 81815, "loss": 3.4840473175048827, "grad_norm": 0.7166984677314758, "learning_rate": 0.0001, "elbo": 3.48405, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.797970912502932}
{"step": 81820, "loss": 3.500416564941406, "grad_norm": 0.7364304065704346, "learning_rate": 0.0001, "elbo": 3.50042, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.798264133239503}
{"step": 81825, "loss": 3.5035785675048827, "grad_norm": 0.7054039239883423, "learning_rate": 0.0001, "elbo": 3.50358, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.798557353976073}
{"step": 81830, "loss": 3.490594482421875, "grad_norm": 0.6996270418167114, "learning_rate": 0.0001, "elbo": 3.49059, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.7988505747126435}
{"step": 81835, "loss": 3.5065383911132812, "grad_norm": 0.7024675607681274, "learning_rate": 0.0001, "elbo": 3.50654, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.799143795449214}
{"step": 81840, "loss": 3.507227325439453, "grad_norm": 0.7645071148872375, "learning_rate": 0.0001, "elbo": 3.50723, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.799437016185784}
{"step": 81845, "loss": 3.4957046508789062, "grad_norm": 0.7001795768737793, "learning_rate": 0.0001, "elbo": 3.4957, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.799730236922355}
{"step": 81850, "loss": 3.5053791046142577, "grad_norm": 0.7790802717208862, "learning_rate": 0.0001, "elbo": 3.50538, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.800023457658925}
{"step": 81855, "loss": 3.499408721923828, "grad_norm": 0.9566268920898438, "learning_rate": 0.0001, "elbo": 3.49941, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.800316678395496}
{"step": 81860, "loss": 3.505072021484375, "grad_norm": 1.1180754899978638, "learning_rate": 0.0001, "elbo": 3.50507, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.800609899132066}
{"step": 81865, "loss": 3.5081050872802733, "grad_norm": 0.7418840527534485, "learning_rate": 0.0001, "elbo": 3.50811, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.800903119868638}
{"step": 81870, "loss": 3.5058925628662108, "grad_norm": 0.721964955329895, "learning_rate": 0.0001, "elbo": 3.50589, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.801196340605207}
{"step": 81875, "loss": 3.501788330078125, "grad_norm": 1.1753795146942139, "learning_rate": 0.0001, "elbo": 3.50179, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8014895613417785}
{"step": 81880, "loss": 3.503199005126953, "grad_norm": 0.7152441740036011, "learning_rate": 0.0001, "elbo": 3.5032, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.801782782078349}
{"step": 81885, "loss": 3.5100471496582033, "grad_norm": 0.6747017502784729, "learning_rate": 0.0001, "elbo": 3.51005, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.802076002814919}
{"step": 81890, "loss": 3.493244934082031, "grad_norm": 0.6719353199005127, "learning_rate": 0.0001, "elbo": 3.49325, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.80236922355149}
{"step": 81895, "loss": 3.505146026611328, "grad_norm": 0.7055104374885559, "learning_rate": 0.0001, "elbo": 3.50515, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.80266244428806}
{"step": 81900, "loss": 3.5117897033691405, "grad_norm": 0.8570101857185364, "learning_rate": 0.0001, "elbo": 3.51179, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.802955665024631}
{"step": 81905, "loss": 3.5042762756347656, "grad_norm": 0.9215444922447205, "learning_rate": 0.0001, "elbo": 3.50428, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.803248885761201}
{"step": 81910, "loss": 3.506713104248047, "grad_norm": 0.7499775886535645, "learning_rate": 0.0001, "elbo": 3.50671, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.803542106497772}
{"step": 81915, "loss": 3.4908302307128904, "grad_norm": 0.7076546549797058, "learning_rate": 0.0001, "elbo": 3.49083, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.803835327234342}
{"step": 81920, "loss": 3.5011928558349608, "grad_norm": 0.7098448872566223, "learning_rate": 0.0001, "elbo": 3.50119, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.804128547970913}
{"step": 81920, "loss": 3.5011928558349608, "grad_norm": 0.7098448872566223, "learning_rate": 0.0001, "elbo": 3.50119, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.804128547970913}
{"step": 81925, "loss": 3.4944999694824217, "grad_norm": 0.797143816947937, "learning_rate": 0.0001, "elbo": 3.4945, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.804421768707483}
{"step": 81930, "loss": 3.504381561279297, "grad_norm": 0.7192833423614502, "learning_rate": 0.0001, "elbo": 3.50438, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8047149894440535}
{"step": 81935, "loss": 3.510986328125, "grad_norm": 0.6932404637336731, "learning_rate": 0.0001, "elbo": 3.51099, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.805008210180624}
{"step": 81940, "loss": 3.4990379333496096, "grad_norm": 0.9499828815460205, "learning_rate": 0.0001, "elbo": 3.49904, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.805301430917194}
{"step": 81945, "loss": 3.505207061767578, "grad_norm": 0.6855639815330505, "learning_rate": 0.0001, "elbo": 3.50521, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.805594651653765}
{"step": 81950, "loss": 3.5152923583984377, "grad_norm": 0.7685412168502808, "learning_rate": 0.0001, "elbo": 3.51529, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.805887872390335}
{"step": 81955, "loss": 3.496686553955078, "grad_norm": 0.7014902830123901, "learning_rate": 0.0001, "elbo": 3.49669, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.806181093126906}
{"step": 81960, "loss": 3.5112682342529298, "grad_norm": 0.7096416354179382, "learning_rate": 0.0001, "elbo": 3.51127, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.806474313863476}
{"step": 81965, "loss": 3.498548889160156, "grad_norm": 0.8088763356208801, "learning_rate": 0.0001, "elbo": 3.49855, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.806767534600047}
{"step": 81970, "loss": 3.500230407714844, "grad_norm": 0.8538104891777039, "learning_rate": 0.0001, "elbo": 3.50023, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.807060755336617}
{"step": 81975, "loss": 3.4856746673583983, "grad_norm": 0.7140405178070068, "learning_rate": 0.0001, "elbo": 3.48567, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.807353976073188}
{"step": 81980, "loss": 3.4920700073242186, "grad_norm": 0.7169234156608582, "learning_rate": 0.0001, "elbo": 3.49207, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.807647196809758}
{"step": 81985, "loss": 3.497508239746094, "grad_norm": 0.7116979360580444, "learning_rate": 0.0001, "elbo": 3.49751, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8079404175463285}
{"step": 81985, "loss": 3.497508239746094, "grad_norm": 0.7116979360580444, "learning_rate": 0.0001, "elbo": 3.49751, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8079404175463285}
{"step": 81990, "loss": 3.5127494812011717, "grad_norm": 0.7044779658317566, "learning_rate": 0.0001, "elbo": 3.51275, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8082336382829}
{"step": 81995, "loss": 3.493439483642578, "grad_norm": 0.7340468764305115, "learning_rate": 0.0001, "elbo": 3.49344, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.808526859019469}
{"step": 82000, "loss": 3.4909671783447265, "grad_norm": 0.8444455862045288, "learning_rate": 0.0001, "elbo": 3.49097, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.808820079756041}
{"step": 82005, "loss": 3.5066177368164064, "grad_norm": 0.8937965035438538, "learning_rate": 0.0001, "elbo": 3.50662, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.809113300492611}
{"step": 82010, "loss": 3.5026275634765627, "grad_norm": 0.6998552680015564, "learning_rate": 0.0001, "elbo": 3.50263, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.809406521229182}
{"step": 82015, "loss": 3.4972789764404295, "grad_norm": 0.7890214323997498, "learning_rate": 0.0001, "elbo": 3.49728, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.809699741965752}
{"step": 82020, "loss": 3.5006195068359376, "grad_norm": 0.749086856842041, "learning_rate": 0.0001, "elbo": 3.50062, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.809992962702323}
{"step": 82025, "loss": 3.5117095947265624, "grad_norm": 0.744408369064331, "learning_rate": 0.0001, "elbo": 3.51171, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.810286183438893}
{"step": 82030, "loss": 3.5090293884277344, "grad_norm": 0.9413191080093384, "learning_rate": 0.0001, "elbo": 3.50903, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8105794041754635}
{"step": 82035, "loss": 3.5092891693115233, "grad_norm": 0.9323511719703674, "learning_rate": 0.0001, "elbo": 3.50929, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.810872624912034}
{"step": 82040, "loss": 3.4986995697021483, "grad_norm": 0.7431425452232361, "learning_rate": 0.0001, "elbo": 3.4987, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.811165845648604}
{"step": 82045, "loss": 3.4950695037841797, "grad_norm": 0.7882381081581116, "learning_rate": 0.0001, "elbo": 3.49507, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.811459066385175}
{"step": 82050, "loss": 3.4926265716552733, "grad_norm": 0.6878766417503357, "learning_rate": 0.0001, "elbo": 3.49263, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.811752287121745}
{"step": 82055, "loss": 3.5017799377441405, "grad_norm": 0.7270488739013672, "learning_rate": 0.0001, "elbo": 3.50178, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.812045507858316}
{"step": 82060, "loss": 3.4911472320556642, "grad_norm": 0.7186300158500671, "learning_rate": 0.0001, "elbo": 3.49115, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.812338728594886}
{"step": 82065, "loss": 3.492181396484375, "grad_norm": 0.7755951285362244, "learning_rate": 0.0001, "elbo": 3.49218, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.812631949331457}
{"step": 82070, "loss": 3.4969520568847656, "grad_norm": 0.7221736311912537, "learning_rate": 0.0001, "elbo": 3.49695, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.812925170068027}
{"step": 82075, "loss": 3.505451202392578, "grad_norm": 0.7311655282974243, "learning_rate": 0.0001, "elbo": 3.50545, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.813218390804598}
{"step": 82075, "loss": 3.505451202392578, "grad_norm": 0.7311655282974243, "learning_rate": 0.0001, "elbo": 3.50545, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.813218390804598}
{"step": 82080, "loss": 3.5008193969726564, "grad_norm": 0.6726439595222473, "learning_rate": 0.0001, "elbo": 3.50082, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.813511611541168}
{"step": 82085, "loss": 3.5150001525878904, "grad_norm": 0.8959616422653198, "learning_rate": 0.0001, "elbo": 3.515, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8138048322777385}
{"step": 82090, "loss": 3.495452117919922, "grad_norm": 0.7104964256286621, "learning_rate": 0.0001, "elbo": 3.49545, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.814098053014309}
{"step": 82095, "loss": 3.5097991943359377, "grad_norm": 0.703653872013092, "learning_rate": 0.0001, "elbo": 3.5098, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.814391273750879}
{"step": 82100, "loss": 3.506644439697266, "grad_norm": 0.6917216777801514, "learning_rate": 0.0001, "elbo": 3.50664, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.81468449448745}
{"step": 82105, "loss": 3.505628204345703, "grad_norm": 0.6800988912582397, "learning_rate": 0.0001, "elbo": 3.50563, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.81497771522402}
{"step": 82110, "loss": 3.5012741088867188, "grad_norm": 0.7584213614463806, "learning_rate": 0.0001, "elbo": 3.50127, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.815270935960591}
{"step": 82115, "loss": 3.4926448822021485, "grad_norm": 0.7179413437843323, "learning_rate": 0.0001, "elbo": 3.49264, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.815564156697162}
{"step": 82120, "loss": 3.497012710571289, "grad_norm": 0.8758588433265686, "learning_rate": 0.0001, "elbo": 3.49701, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.815857377433732}
{"step": 82125, "loss": 3.502853012084961, "grad_norm": 0.742293655872345, "learning_rate": 0.0001, "elbo": 3.50285, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.816150598170303}
{"step": 82130, "loss": 3.49434814453125, "grad_norm": 0.7591082453727722, "learning_rate": 0.0001, "elbo": 3.49435, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8164438189068735}
{"step": 82135, "loss": 3.4989990234375, "grad_norm": 0.7280119061470032, "learning_rate": 0.0001, "elbo": 3.499, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.816737039643444}
{"step": 82140, "loss": 3.485888671875, "grad_norm": 0.7328113913536072, "learning_rate": 0.0001, "elbo": 3.48589, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.817030260380014}
{"step": 82145, "loss": 3.5082202911376954, "grad_norm": 0.6897926926612854, "learning_rate": 0.0001, "elbo": 3.50822, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.817323481116585}
{"step": 82150, "loss": 3.4929359436035154, "grad_norm": 0.7143139839172363, "learning_rate": 0.0001, "elbo": 3.49294, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.817616701853155}
{"step": 82155, "loss": 3.499217987060547, "grad_norm": 0.8246227502822876, "learning_rate": 0.0001, "elbo": 3.49922, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.817909922589726}
{"step": 82160, "loss": 3.4965187072753907, "grad_norm": 0.703891396522522, "learning_rate": 0.0001, "elbo": 3.49652, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.818203143326296}
{"step": 82165, "loss": 3.523558807373047, "grad_norm": 0.7976599931716919, "learning_rate": 0.0001, "elbo": 3.52356, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.818496364062867}
{"step": 82170, "loss": 3.5026344299316405, "grad_norm": 1.1468919515609741, "learning_rate": 0.0001, "elbo": 3.50263, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.818789584799437}
{"step": 82175, "loss": 3.502062225341797, "grad_norm": 0.7376304864883423, "learning_rate": 0.0001, "elbo": 3.50206, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8190828055360075}
{"step": 82180, "loss": 3.48311767578125, "grad_norm": 0.7498365640640259, "learning_rate": 0.0001, "elbo": 3.48312, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.819376026272578}
{"step": 82185, "loss": 3.4901519775390626, "grad_norm": 0.9033701419830322, "learning_rate": 0.0001, "elbo": 3.49015, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8196692470091484}
{"step": 82190, "loss": 3.498651885986328, "grad_norm": 0.6601925492286682, "learning_rate": 0.0001, "elbo": 3.49865, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.819962467745719}
{"step": 82195, "loss": 3.491994857788086, "grad_norm": 0.6886245012283325, "learning_rate": 0.0001, "elbo": 3.49199, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.820255688482289}
{"step": 82200, "loss": 3.5011497497558595, "grad_norm": 0.7329045534133911, "learning_rate": 0.0001, "elbo": 3.50115, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.82054890921886}
{"step": 82205, "loss": 3.4898414611816406, "grad_norm": 0.7185266017913818, "learning_rate": 0.0001, "elbo": 3.48984, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.82084212995543}
{"step": 82210, "loss": 3.4962837219238283, "grad_norm": 0.7450628876686096, "learning_rate": 0.0001, "elbo": 3.49628, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.821135350692001}
{"step": 82215, "loss": 3.504715347290039, "grad_norm": 0.7183606028556824, "learning_rate": 0.0001, "elbo": 3.50472, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.821428571428571}
{"step": 82220, "loss": 3.4964836120605467, "grad_norm": 0.7243531942367554, "learning_rate": 0.0001, "elbo": 3.49648, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.821721792165142}
{"step": 82225, "loss": 3.494969940185547, "grad_norm": 0.7187226414680481, "learning_rate": 0.0001, "elbo": 3.49497, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.822015012901712}
{"step": 82230, "loss": 3.5169639587402344, "grad_norm": 0.7147025465965271, "learning_rate": 0.0001, "elbo": 3.51696, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8223082336382825}
{"step": 82235, "loss": 3.496063995361328, "grad_norm": 0.705420970916748, "learning_rate": 0.0001, "elbo": 3.49606, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.822601454374853}
{"step": 82240, "loss": 3.503311538696289, "grad_norm": 0.7149346470832825, "learning_rate": 0.0001, "elbo": 3.50331, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.822894675111424}
{"step": 82245, "loss": 3.505867767333984, "grad_norm": 0.7856586575508118, "learning_rate": 0.0001, "elbo": 3.50587, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.823187895847994}
{"step": 82250, "loss": 3.493047332763672, "grad_norm": 1.985969066619873, "learning_rate": 0.0001, "elbo": 3.49305, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.823481116584565}
{"step": 82255, "loss": 3.505290985107422, "grad_norm": 0.7246591448783875, "learning_rate": 0.0001, "elbo": 3.50529, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.823774337321136}
{"step": 82260, "loss": 3.49112548828125, "grad_norm": 1.089227557182312, "learning_rate": 0.0001, "elbo": 3.49113, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.824067558057706}
{"step": 82265, "loss": 3.510208511352539, "grad_norm": 0.7503218054771423, "learning_rate": 0.0001, "elbo": 3.51021, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.824360778794277}
{"step": 82270, "loss": 3.486024856567383, "grad_norm": 0.8493953943252563, "learning_rate": 0.0001, "elbo": 3.48602, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.824653999530847}
{"step": 82275, "loss": 3.498738098144531, "grad_norm": 0.7141093015670776, "learning_rate": 0.0001, "elbo": 3.49874, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8249472202674175}
{"step": 82280, "loss": 3.489643096923828, "grad_norm": 0.8160353302955627, "learning_rate": 0.0001, "elbo": 3.48964, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.825240441003988}
{"step": 82285, "loss": 3.5026344299316405, "grad_norm": 0.7136300802230835, "learning_rate": 0.0001, "elbo": 3.50263, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.825533661740558}
{"step": 82290, "loss": 3.5093833923339846, "grad_norm": 0.7486375570297241, "learning_rate": 0.0001, "elbo": 3.50938, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.825826882477129}
{"step": 82295, "loss": 3.499228286743164, "grad_norm": 0.7350409626960754, "learning_rate": 0.0001, "elbo": 3.49923, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.826120103213699}
{"step": 82300, "loss": 3.5079402923583984, "grad_norm": 0.6999494433403015, "learning_rate": 0.0001, "elbo": 3.50794, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.82641332395027}
{"step": 82305, "loss": 3.490406799316406, "grad_norm": 0.7811657190322876, "learning_rate": 0.0001, "elbo": 3.49041, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.82670654468684}
{"step": 82310, "loss": 3.518767547607422, "grad_norm": 0.7852181196212769, "learning_rate": 0.0001, "elbo": 3.51877, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.826999765423411}
{"step": 82315, "loss": 3.4925071716308596, "grad_norm": 0.7302037477493286, "learning_rate": 0.0001, "elbo": 3.49251, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.827292986159981}
{"step": 82320, "loss": 3.4901981353759766, "grad_norm": 0.7236536741256714, "learning_rate": 0.0001, "elbo": 3.4902, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.827586206896552}
{"step": 82325, "loss": 3.506845474243164, "grad_norm": 0.6978249549865723, "learning_rate": 0.0001, "elbo": 3.50685, "mask_frac": 0.5013, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.827879427633122}
{"step": 82325, "loss": 3.506845474243164, "grad_norm": 0.6978249549865723, "learning_rate": 0.0001, "elbo": 3.50685, "mask_frac": 0.5013, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.827879427633122}
{"step": 82330, "loss": 3.4964523315429688, "grad_norm": 0.699604868888855, "learning_rate": 0.0001, "elbo": 3.49645, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8281726483696925}
{"step": 82335, "loss": 3.4946914672851563, "grad_norm": 0.7623808979988098, "learning_rate": 0.0001, "elbo": 3.49469, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.828465869106263}
{"step": 82340, "loss": 3.4865901947021483, "grad_norm": 0.8464837074279785, "learning_rate": 0.0001, "elbo": 3.48659, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.828759089842833}
{"step": 82345, "loss": 3.499995803833008, "grad_norm": 0.7003794312477112, "learning_rate": 0.0001, "elbo": 3.5, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.829052310579404}
{"step": 82350, "loss": 3.4990631103515626, "grad_norm": 0.7117127180099487, "learning_rate": 0.0001, "elbo": 3.49906, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.829345531315974}
{"step": 82355, "loss": 3.4855934143066407, "grad_norm": 0.7797174453735352, "learning_rate": 0.0001, "elbo": 3.48559, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.829638752052545}
{"step": 82360, "loss": 3.4872802734375, "grad_norm": 0.6829498410224915, "learning_rate": 0.0001, "elbo": 3.48728, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.829931972789115}
{"step": 82365, "loss": 3.5022502899169923, "grad_norm": 0.7951180934906006, "learning_rate": 0.0001, "elbo": 3.50225, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.830225193525687}
{"step": 82370, "loss": 3.5030776977539064, "grad_norm": 0.6909217238426208, "learning_rate": 0.0001, "elbo": 3.50308, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.830518414262256}
{"step": 82375, "loss": 3.510967254638672, "grad_norm": 0.6955861449241638, "learning_rate": 0.0001, "elbo": 3.51097, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8308116349988275}
{"step": 82380, "loss": 3.4927032470703123, "grad_norm": 0.68367999792099, "learning_rate": 0.0001, "elbo": 3.4927, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.831104855735398}
{"step": 82385, "loss": 3.49793701171875, "grad_norm": 0.7020344138145447, "learning_rate": 0.0001, "elbo": 3.49794, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.831398076471968}
{"step": 82390, "loss": 3.493103790283203, "grad_norm": 0.7192131876945496, "learning_rate": 0.0001, "elbo": 3.4931, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.831691297208539}
{"step": 82395, "loss": 3.499261474609375, "grad_norm": 1.0555670261383057, "learning_rate": 0.0001, "elbo": 3.49926, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.831984517945109}
{"step": 82400, "loss": 3.4957626342773436, "grad_norm": 0.7029805183410645, "learning_rate": 0.0001, "elbo": 3.49576, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.83227773868168}
{"step": 82405, "loss": 3.5067779541015627, "grad_norm": 0.7030124068260193, "learning_rate": 0.0001, "elbo": 3.50678, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.83257095941825}
{"step": 82410, "loss": 3.5070098876953124, "grad_norm": 0.7672543525695801, "learning_rate": 0.0001, "elbo": 3.50701, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.832864180154821}
{"step": 82415, "loss": 3.4941726684570313, "grad_norm": 0.6749554872512817, "learning_rate": 0.0001, "elbo": 3.49417, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.833157400891391}
{"step": 82420, "loss": 3.4965003967285155, "grad_norm": 0.7307361960411072, "learning_rate": 0.0001, "elbo": 3.4965, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.833450621627962}
{"step": 82425, "loss": 3.512640380859375, "grad_norm": 0.7345834374427795, "learning_rate": 0.0001, "elbo": 3.51264, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.833743842364532}
{"step": 82430, "loss": 3.4950355529785155, "grad_norm": 0.784797191619873, "learning_rate": 0.0001, "elbo": 3.49504, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8340370631011025}
{"step": 82435, "loss": 3.517719268798828, "grad_norm": 0.7118232250213623, "learning_rate": 0.0001, "elbo": 3.51772, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.834330283837673}
{"step": 82440, "loss": 3.4980819702148436, "grad_norm": 0.7417336702346802, "learning_rate": 0.0001, "elbo": 3.49808, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.834623504574243}
{"step": 82445, "loss": 3.5048851013183593, "grad_norm": 0.7637699246406555, "learning_rate": 0.0001, "elbo": 3.50488, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.834916725310814}
{"step": 82450, "loss": 3.496776580810547, "grad_norm": 0.6917741298675537, "learning_rate": 0.0001, "elbo": 3.49678, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.835209946047384}
{"step": 82455, "loss": 3.4966503143310548, "grad_norm": 0.7453714609146118, "learning_rate": 0.0001, "elbo": 3.49665, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.835503166783955}
{"step": 82460, "loss": 3.499623489379883, "grad_norm": 0.7270147204399109, "learning_rate": 0.0001, "elbo": 3.49962, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.835796387520525}
{"step": 82465, "loss": 3.52210693359375, "grad_norm": 0.7876772880554199, "learning_rate": 0.0001, "elbo": 3.52211, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.836089608257096}
{"step": 82470, "loss": 3.509683609008789, "grad_norm": 0.698666512966156, "learning_rate": 0.0001, "elbo": 3.50968, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.836382828993666}
{"step": 82475, "loss": 3.4832443237304687, "grad_norm": 0.7033211588859558, "learning_rate": 0.0001, "elbo": 3.48324, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.836676049730237}
{"step": 82480, "loss": 3.5228092193603517, "grad_norm": 0.7519789338111877, "learning_rate": 0.0001, "elbo": 3.52281, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.836969270466807}
{"step": 82485, "loss": 3.501877212524414, "grad_norm": 0.688926100730896, "learning_rate": 0.0001, "elbo": 3.50188, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8372624912033775}
{"step": 82490, "loss": 3.5024757385253906, "grad_norm": 0.7409124374389648, "learning_rate": 0.0001, "elbo": 3.50248, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.837555711939949}
{"step": 82495, "loss": 3.505876159667969, "grad_norm": 0.7424275279045105, "learning_rate": 0.0001, "elbo": 3.50588, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.837848932676518}
{"step": 82500, "loss": 3.5092796325683593, "grad_norm": 0.7425430417060852, "learning_rate": 0.0001, "elbo": 3.50928, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.83814215341309}
{"step": 82505, "loss": 3.4962799072265627, "grad_norm": 1.0315333604812622, "learning_rate": 0.0001, "elbo": 3.49628, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.83843537414966}
{"step": 82510, "loss": 3.5101444244384767, "grad_norm": 0.7088991403579712, "learning_rate": 0.0001, "elbo": 3.51014, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.838728594886231}
{"step": 82515, "loss": 3.5147674560546873, "grad_norm": 0.7240776419639587, "learning_rate": 0.0001, "elbo": 3.51477, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.839021815622801}
{"step": 82520, "loss": 3.507341003417969, "grad_norm": 0.6922324299812317, "learning_rate": 0.0001, "elbo": 3.50734, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8393150363593715}
{"step": 82525, "loss": 3.492737579345703, "grad_norm": 0.7236474752426147, "learning_rate": 0.0001, "elbo": 3.49274, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.839608257095942}
{"step": 82530, "loss": 3.4924110412597655, "grad_norm": 0.6854555010795593, "learning_rate": 0.0001, "elbo": 3.49241, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8399014778325125}
{"step": 82535, "loss": 3.5131462097167967, "grad_norm": 0.7625141143798828, "learning_rate": 0.0001, "elbo": 3.51315, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.840194698569083}
{"step": 82540, "loss": 3.494853973388672, "grad_norm": 0.7461697459220886, "learning_rate": 0.0001, "elbo": 3.49485, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.840487919305653}
{"step": 82540, "loss": 3.494853973388672, "grad_norm": 0.7461697459220886, "learning_rate": 0.0001, "elbo": 3.49485, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.840487919305653}
{"step": 82545, "loss": 3.505479431152344, "grad_norm": 0.7315459847450256, "learning_rate": 0.0001, "elbo": 3.50548, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.840781140042224}
{"step": 82550, "loss": 3.489268493652344, "grad_norm": 0.6871811151504517, "learning_rate": 0.0001, "elbo": 3.48927, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.841074360778794}
{"step": 82555, "loss": 3.49805908203125, "grad_norm": 0.7116191983222961, "learning_rate": 0.0001, "elbo": 3.49806, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.841367581515365}
{"step": 82560, "loss": 3.4999603271484374, "grad_norm": 0.8332063555717468, "learning_rate": 0.0001, "elbo": 3.49996, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.841660802251935}
{"step": 82565, "loss": 3.50751953125, "grad_norm": 0.9157353639602661, "learning_rate": 0.0001, "elbo": 3.50752, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.841954022988506}
{"step": 82570, "loss": 3.494702911376953, "grad_norm": 0.8448835015296936, "learning_rate": 0.0001, "elbo": 3.4947, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.842247243725076}
{"step": 82575, "loss": 3.4987964630126953, "grad_norm": 0.7019237279891968, "learning_rate": 0.0001, "elbo": 3.4988, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8425404644616465}
{"step": 82580, "loss": 3.5046424865722656, "grad_norm": 0.7516725659370422, "learning_rate": 0.0001, "elbo": 3.50464, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.842833685198217}
{"step": 82580, "loss": 3.5046424865722656, "grad_norm": 0.7516725659370422, "learning_rate": 0.0001, "elbo": 3.50464, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.842833685198217}
{"step": 82585, "loss": 3.5304054260253905, "grad_norm": 0.7082789540290833, "learning_rate": 0.0001, "elbo": 3.53041, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8431269059347875}
{"step": 82585, "loss": 3.5304054260253905, "grad_norm": 0.7082789540290833, "learning_rate": 0.0001, "elbo": 3.53041, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8431269059347875}
{"step": 82590, "loss": 3.502826690673828, "grad_norm": 0.7741211652755737, "learning_rate": 0.0001, "elbo": 3.50283, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.843420126671358}
{"step": 82595, "loss": 3.499932861328125, "grad_norm": 0.7406070232391357, "learning_rate": 0.0001, "elbo": 3.49993, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.843713347407928}
{"step": 82600, "loss": 3.4955562591552733, "grad_norm": 0.6787058115005493, "learning_rate": 0.0001, "elbo": 3.49556, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.844006568144499}
{"step": 82605, "loss": 3.496022415161133, "grad_norm": 0.7534496784210205, "learning_rate": 0.0001, "elbo": 3.49602, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.84429978888107}
{"step": 82610, "loss": 3.4976600646972655, "grad_norm": 0.6914529204368591, "learning_rate": 0.0001, "elbo": 3.49766, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.84459300961764}
{"step": 82615, "loss": 3.4905601501464845, "grad_norm": 0.7036914825439453, "learning_rate": 0.0001, "elbo": 3.49056, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.844886230354211}
{"step": 82620, "loss": 3.5040462493896483, "grad_norm": 0.7395647764205933, "learning_rate": 0.0001, "elbo": 3.50405, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8451794510907815}
{"step": 82625, "loss": 3.510661315917969, "grad_norm": 0.7360348701477051, "learning_rate": 0.0001, "elbo": 3.51066, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.845472671827352}
{"step": 82630, "loss": 3.508373260498047, "grad_norm": 0.7708554267883301, "learning_rate": 0.0001, "elbo": 3.50837, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.845765892563922}
{"step": 82635, "loss": 3.5066226959228515, "grad_norm": 0.8625624179840088, "learning_rate": 0.0001, "elbo": 3.50662, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.846059113300493}
{"step": 82640, "loss": 3.5056396484375, "grad_norm": 0.7363947629928589, "learning_rate": 0.0001, "elbo": 3.50564, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.846352334037063}
{"step": 82645, "loss": 3.4860843658447265, "grad_norm": 0.7602166533470154, "learning_rate": 0.0001, "elbo": 3.48608, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.846645554773634}
{"step": 82650, "loss": 3.4894729614257813, "grad_norm": 0.7023420929908752, "learning_rate": 0.0001, "elbo": 3.48947, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.846938775510204}
{"step": 82650, "loss": 3.4894729614257813, "grad_norm": 0.7023420929908752, "learning_rate": 0.0001, "elbo": 3.48947, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.846938775510204}
{"step": 82655, "loss": 3.4970401763916015, "grad_norm": 0.7973001003265381, "learning_rate": 0.0001, "elbo": 3.49704, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.847231996246775}
{"step": 82660, "loss": 3.5021575927734374, "grad_norm": 0.7197923064231873, "learning_rate": 0.0001, "elbo": 3.50216, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.847525216983345}
{"step": 82665, "loss": 3.511333465576172, "grad_norm": 0.699234664440155, "learning_rate": 0.0001, "elbo": 3.51133, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.847818437719916}
{"step": 82670, "loss": 3.49200439453125, "grad_norm": 0.7801106572151184, "learning_rate": 0.0001, "elbo": 3.492, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.848111658456486}
{"step": 82675, "loss": 3.508026123046875, "grad_norm": 0.7089040279388428, "learning_rate": 0.0001, "elbo": 3.50803, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8484048791930565}
{"step": 82680, "loss": 3.498524856567383, "grad_norm": 0.8065527081489563, "learning_rate": 0.0001, "elbo": 3.49852, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.848698099929627}
{"step": 82685, "loss": 3.506696319580078, "grad_norm": 0.7996636033058167, "learning_rate": 0.0001, "elbo": 3.5067, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.848991320666197}
{"step": 82690, "loss": 3.509413146972656, "grad_norm": 0.8388538956642151, "learning_rate": 0.0001, "elbo": 3.50941, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.849284541402768}
{"step": 82695, "loss": 3.5004600524902343, "grad_norm": 0.7297490835189819, "learning_rate": 0.0001, "elbo": 3.50046, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.849577762139338}
{"step": 82700, "loss": 3.495418930053711, "grad_norm": 0.6748296618461609, "learning_rate": 0.0001, "elbo": 3.49542, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.849870982875909}
{"step": 82705, "loss": 3.4915084838867188, "grad_norm": 0.934857964515686, "learning_rate": 0.0001, "elbo": 3.49151, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.850164203612479}
{"step": 82710, "loss": 3.5057395935058593, "grad_norm": 0.7751373648643494, "learning_rate": 0.0001, "elbo": 3.50574, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.85045742434905}
{"step": 82715, "loss": 3.510234832763672, "grad_norm": 0.7799235582351685, "learning_rate": 0.0001, "elbo": 3.51024, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.85075064508562}
{"step": 82720, "loss": 3.4977161407470705, "grad_norm": 0.8645039796829224, "learning_rate": 0.0001, "elbo": 3.49772, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.851043865822191}
{"step": 82725, "loss": 3.5149566650390627, "grad_norm": 0.7627499103546143, "learning_rate": 0.0001, "elbo": 3.51496, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.851337086558761}
{"step": 82730, "loss": 3.4867568969726563, "grad_norm": 0.7156972289085388, "learning_rate": 0.0001, "elbo": 3.48676, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.851630307295332}
{"step": 82735, "loss": 3.5082160949707033, "grad_norm": 0.7245692610740662, "learning_rate": 0.0001, "elbo": 3.50822, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.851923528031902}
{"step": 82740, "loss": 3.5023101806640624, "grad_norm": 0.779106855392456, "learning_rate": 0.0001, "elbo": 3.50231, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.852216748768473}
{"step": 82745, "loss": 3.5052032470703125, "grad_norm": 0.7006189823150635, "learning_rate": 0.0001, "elbo": 3.5052, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.852509969505044}
{"step": 82750, "loss": 3.506952667236328, "grad_norm": 0.7251173257827759, "learning_rate": 0.0001, "elbo": 3.50695, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.852803190241614}
{"step": 82755, "loss": 3.502777099609375, "grad_norm": 0.6979060173034668, "learning_rate": 0.0001, "elbo": 3.50278, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.853096410978185}
{"step": 82760, "loss": 3.505501556396484, "grad_norm": 0.7172515988349915, "learning_rate": 0.0001, "elbo": 3.5055, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.853389631714755}
{"step": 82765, "loss": 3.5001384735107424, "grad_norm": 0.7618721723556519, "learning_rate": 0.0001, "elbo": 3.50014, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.853682852451326}
{"step": 82770, "loss": 3.5049156188964843, "grad_norm": 0.7244382500648499, "learning_rate": 0.0001, "elbo": 3.50492, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.853976073187896}
{"step": 82775, "loss": 3.501361083984375, "grad_norm": 0.7152321934700012, "learning_rate": 0.0001, "elbo": 3.50136, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8542692939244665}
{"step": 82780, "loss": 3.4986923217773436, "grad_norm": 0.6977238655090332, "learning_rate": 0.0001, "elbo": 3.49869, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.854562514661037}
{"step": 82785, "loss": 3.5066349029541017, "grad_norm": 0.6721159815788269, "learning_rate": 0.0001, "elbo": 3.50664, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.854855735397607}
{"step": 82790, "loss": 3.5013011932373046, "grad_norm": 0.7908779382705688, "learning_rate": 0.0001, "elbo": 3.5013, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.855148956134178}
{"step": 82795, "loss": 3.5087989807128905, "grad_norm": 0.8280273675918579, "learning_rate": 0.0001, "elbo": 3.5088, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.855442176870748}
{"step": 82800, "loss": 3.5041656494140625, "grad_norm": 0.8156489729881287, "learning_rate": 0.0001, "elbo": 3.50417, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.855735397607319}
{"step": 82805, "loss": 3.5104690551757813, "grad_norm": 0.8613218069076538, "learning_rate": 0.0001, "elbo": 3.51047, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.856028618343889}
{"step": 82810, "loss": 3.488280487060547, "grad_norm": 0.7244704365730286, "learning_rate": 0.0001, "elbo": 3.48828, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.85632183908046}
{"step": 82815, "loss": 3.493693542480469, "grad_norm": 0.788361668586731, "learning_rate": 0.0001, "elbo": 3.49369, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.85661505981703}
{"step": 82815, "loss": 3.493693542480469, "grad_norm": 0.788361668586731, "learning_rate": 0.0001, "elbo": 3.49369, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.85661505981703}
{"step": 82820, "loss": 3.4870853424072266, "grad_norm": 0.7874772548675537, "learning_rate": 0.0001, "elbo": 3.48709, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.856908280553601}
{"step": 82825, "loss": 3.494465637207031, "grad_norm": 0.7036300301551819, "learning_rate": 0.0001, "elbo": 3.49447, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.857201501290171}
{"step": 82830, "loss": 3.521525573730469, "grad_norm": 0.7341167330741882, "learning_rate": 0.0001, "elbo": 3.52153, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8574947220267415}
{"step": 82835, "loss": 3.5040637969970705, "grad_norm": 0.6951919198036194, "learning_rate": 0.0001, "elbo": 3.50406, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.857787942763312}
{"step": 82840, "loss": 3.500836944580078, "grad_norm": 0.6927070617675781, "learning_rate": 0.0001, "elbo": 3.50084, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.858081163499882}
{"step": 82845, "loss": 3.4874198913574217, "grad_norm": 0.6731276512145996, "learning_rate": 0.0001, "elbo": 3.48742, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.858374384236453}
{"step": 82850, "loss": 3.5056419372558594, "grad_norm": 0.7095922231674194, "learning_rate": 0.0001, "elbo": 3.50564, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.858667604973023}
{"step": 82855, "loss": 3.4815528869628904, "grad_norm": 0.7238131165504456, "learning_rate": 0.0001, "elbo": 3.48155, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.858960825709595}
{"step": 82860, "loss": 3.5006065368652344, "grad_norm": 0.692209780216217, "learning_rate": 0.0001, "elbo": 3.50061, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.859254046446164}
{"step": 82865, "loss": 3.490642547607422, "grad_norm": 0.804098904132843, "learning_rate": 0.0001, "elbo": 3.49064, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8595472671827356}
{"step": 82870, "loss": 3.4994598388671876, "grad_norm": 0.7076186537742615, "learning_rate": 0.0001, "elbo": 3.49946, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.859840487919306}
{"step": 82875, "loss": 3.5014923095703123, "grad_norm": 0.6947638392448425, "learning_rate": 0.0001, "elbo": 3.50149, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8601337086558765}
{"step": 82880, "loss": 3.5011886596679687, "grad_norm": 0.8349363207817078, "learning_rate": 0.0001, "elbo": 3.50119, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.860426929392447}
{"step": 82885, "loss": 3.486185073852539, "grad_norm": 0.6565987467765808, "learning_rate": 0.0001, "elbo": 3.48619, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.860720150129017}
{"step": 82890, "loss": 3.5041908264160155, "grad_norm": 0.7263829112052917, "learning_rate": 0.0001, "elbo": 3.50419, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.861013370865588}
{"step": 82895, "loss": 3.517649841308594, "grad_norm": 0.7601986527442932, "learning_rate": 0.0001, "elbo": 3.51765, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.861306591602158}
{"step": 82900, "loss": 3.511820983886719, "grad_norm": 1.1306405067443848, "learning_rate": 0.0001, "elbo": 3.51182, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.861599812338729}
{"step": 82905, "loss": 3.4768512725830076, "grad_norm": 0.7246087789535522, "learning_rate": 0.0001, "elbo": 3.47685, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.861893033075299}
{"step": 82910, "loss": 3.500379943847656, "grad_norm": 0.7293224334716797, "learning_rate": 0.0001, "elbo": 3.50038, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.86218625381187}
{"step": 82915, "loss": 3.504412078857422, "grad_norm": 0.79685378074646, "learning_rate": 0.0001, "elbo": 3.50441, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.86247947454844}
{"step": 82920, "loss": 3.508322906494141, "grad_norm": 0.7458924651145935, "learning_rate": 0.0001, "elbo": 3.50832, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8627726952850105}
{"step": 82925, "loss": 3.4993629455566406, "grad_norm": 0.7304468154907227, "learning_rate": 0.0001, "elbo": 3.49936, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.863065916021581}
{"step": 82930, "loss": 3.500012969970703, "grad_norm": 0.7221946120262146, "learning_rate": 0.0001, "elbo": 3.50001, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8633591367581515}
{"step": 82935, "loss": 3.4890846252441405, "grad_norm": 0.8212392926216125, "learning_rate": 0.0001, "elbo": 3.48908, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.863652357494722}
{"step": 82940, "loss": 3.511247253417969, "grad_norm": 0.7246456146240234, "learning_rate": 0.0001, "elbo": 3.51125, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.863945578231292}
{"step": 82945, "loss": 3.4992019653320314, "grad_norm": 0.7000014781951904, "learning_rate": 0.0001, "elbo": 3.4992, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.864238798967863}
{"step": 82950, "loss": 3.50805778503418, "grad_norm": 0.7153626680374146, "learning_rate": 0.0001, "elbo": 3.50806, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.864532019704433}
{"step": 82955, "loss": 3.5001625061035155, "grad_norm": 0.7169026136398315, "learning_rate": 0.0001, "elbo": 3.50016, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.864825240441004}
{"step": 82955, "loss": 3.5001625061035155, "grad_norm": 0.7169026136398315, "learning_rate": 0.0001, "elbo": 3.50016, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.864825240441004}
{"step": 82960, "loss": 3.5092399597167967, "grad_norm": 0.7550463676452637, "learning_rate": 0.0001, "elbo": 3.50924, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.865118461177574}
{"step": 82965, "loss": 3.4974475860595704, "grad_norm": 0.7251470685005188, "learning_rate": 0.0001, "elbo": 3.49745, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.865411681914145}
{"step": 82970, "loss": 3.520094299316406, "grad_norm": 0.7613388895988464, "learning_rate": 0.0001, "elbo": 3.52009, "mask_frac": 0.5016, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.865704902650715}
{"step": 82975, "loss": 3.493429946899414, "grad_norm": 1.0375711917877197, "learning_rate": 0.0001, "elbo": 3.49343, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8659981233872855}
{"step": 82980, "loss": 3.4963058471679687, "grad_norm": 0.7427486777305603, "learning_rate": 0.0001, "elbo": 3.49631, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.866291344123857}
{"step": 82985, "loss": 3.513691711425781, "grad_norm": 0.6933839321136475, "learning_rate": 0.0001, "elbo": 3.51369, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8665845648604265}
{"step": 82990, "loss": 3.4978561401367188, "grad_norm": 0.7205021977424622, "learning_rate": 0.0001, "elbo": 3.49786, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.866877785596998}
{"step": 82995, "loss": 3.505066680908203, "grad_norm": 0.8077682256698608, "learning_rate": 0.0001, "elbo": 3.50507, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.867171006333568}
{"step": 83000, "loss": 3.487244415283203, "grad_norm": 0.7420603036880493, "learning_rate": 0.0001, "elbo": 3.48724, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.867464227070139}
{"step": 83005, "loss": 3.4918357849121096, "grad_norm": 0.7038300037384033, "learning_rate": 0.0001, "elbo": 3.49184, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.867757447806709}
{"step": 83010, "loss": 3.4933441162109373, "grad_norm": 0.7346378564834595, "learning_rate": 0.0001, "elbo": 3.49334, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.86805066854328}
{"step": 83015, "loss": 3.4990440368652345, "grad_norm": 0.7146027088165283, "learning_rate": 0.0001, "elbo": 3.49904, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.86834388927985}
{"step": 83020, "loss": 3.500008392333984, "grad_norm": 0.7280206084251404, "learning_rate": 0.0001, "elbo": 3.50001, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8686371100164205}
{"step": 83025, "loss": 3.4948081970214844, "grad_norm": 0.839990496635437, "learning_rate": 0.0001, "elbo": 3.49481, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.868930330752991}
{"step": 83030, "loss": 3.498172378540039, "grad_norm": 0.6803855299949646, "learning_rate": 0.0001, "elbo": 3.49817, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.869223551489561}
{"step": 83035, "loss": 3.473455810546875, "grad_norm": 0.8085395693778992, "learning_rate": 0.0001, "elbo": 3.47346, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.869516772226132}
{"step": 83040, "loss": 3.4752471923828123, "grad_norm": 0.8315317034721375, "learning_rate": 0.0001, "elbo": 3.47525, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.869809992962702}
{"step": 83045, "loss": 3.4814098358154295, "grad_norm": 0.7085815072059631, "learning_rate": 0.0001, "elbo": 3.48141, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.870103213699273}
{"step": 83050, "loss": 3.485676956176758, "grad_norm": 0.7864637970924377, "learning_rate": 0.0001, "elbo": 3.48568, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.870396434435843}
{"step": 83055, "loss": 3.5136558532714846, "grad_norm": 0.774797260761261, "learning_rate": 0.0001, "elbo": 3.51366, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.870689655172414}
{"step": 83060, "loss": 3.496254730224609, "grad_norm": 0.8562466502189636, "learning_rate": 0.0001, "elbo": 3.49625, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.870982875908984}
{"step": 83065, "loss": 3.492730712890625, "grad_norm": 0.7356386184692383, "learning_rate": 0.0001, "elbo": 3.49273, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.871276096645555}
{"step": 83070, "loss": 3.5037384033203125, "grad_norm": 0.8809407949447632, "learning_rate": 0.0001, "elbo": 3.50374, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.871569317382125}
{"step": 83075, "loss": 3.4985301971435545, "grad_norm": 0.7271133065223694, "learning_rate": 0.0001, "elbo": 3.49853, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8718625381186955}
{"step": 83080, "loss": 3.487443542480469, "grad_norm": 0.7292047142982483, "learning_rate": 0.0001, "elbo": 3.48744, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.872155758855266}
{"step": 83085, "loss": 3.493682861328125, "grad_norm": 0.7151183485984802, "learning_rate": 0.0001, "elbo": 3.49368, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.872448979591836}
{"step": 83090, "loss": 3.500248336791992, "grad_norm": 1.3007742166519165, "learning_rate": 0.0001, "elbo": 3.50025, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.872742200328407}
{"step": 83095, "loss": 3.519336700439453, "grad_norm": 0.791511595249176, "learning_rate": 0.0001, "elbo": 3.51934, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.873035421064977}
{"step": 83100, "loss": 3.5100387573242187, "grad_norm": 0.7525977492332458, "learning_rate": 0.0001, "elbo": 3.51004, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.873328641801548}
{"step": 83105, "loss": 3.518324279785156, "grad_norm": 0.6906404495239258, "learning_rate": 0.0001, "elbo": 3.51832, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.873621862538119}
{"step": 83110, "loss": 3.504069519042969, "grad_norm": 0.6689178347587585, "learning_rate": 0.0001, "elbo": 3.50407, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.873915083274689}
{"step": 83115, "loss": 3.4955673217773438, "grad_norm": 0.717320442199707, "learning_rate": 0.0001, "elbo": 3.49557, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.87420830401126}
{"step": 83120, "loss": 3.4998950958251953, "grad_norm": 0.7127060890197754, "learning_rate": 0.0001, "elbo": 3.4999, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8745015247478305}
{"step": 83125, "loss": 3.492654037475586, "grad_norm": 0.6966835260391235, "learning_rate": 0.0001, "elbo": 3.49265, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.874794745484401}
{"step": 83130, "loss": 3.5020862579345704, "grad_norm": 0.7869977951049805, "learning_rate": 0.0001, "elbo": 3.50209, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.875087966220971}
{"step": 83135, "loss": 3.5124065399169924, "grad_norm": 0.7111396193504333, "learning_rate": 0.0001, "elbo": 3.51241, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.875381186957542}
{"step": 83140, "loss": 3.5014415740966798, "grad_norm": 0.6881728172302246, "learning_rate": 0.0001, "elbo": 3.50144, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.875674407694112}
{"step": 83145, "loss": 3.4821510314941406, "grad_norm": 0.6980489492416382, "learning_rate": 0.0001, "elbo": 3.48215, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.875967628430683}
{"step": 83150, "loss": 3.5099769592285157, "grad_norm": 0.8214017748832703, "learning_rate": 0.0001, "elbo": 3.50998, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.876260849167253}
{"step": 83155, "loss": 3.477112579345703, "grad_norm": 0.7563839554786682, "learning_rate": 0.0001, "elbo": 3.47711, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.876554069903824}
{"step": 83160, "loss": 3.4876712799072265, "grad_norm": 0.7582468390464783, "learning_rate": 0.0001, "elbo": 3.48767, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.876847290640394}
{"step": 83165, "loss": 3.5075363159179687, "grad_norm": 0.9102630019187927, "learning_rate": 0.0001, "elbo": 3.50754, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.877140511376965}
{"step": 83165, "loss": 3.5075363159179687, "grad_norm": 0.9102630019187927, "learning_rate": 0.0001, "elbo": 3.50754, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.877140511376965}
{"step": 83170, "loss": 3.509547805786133, "grad_norm": 0.8568761348724365, "learning_rate": 0.0001, "elbo": 3.50955, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.877433732113535}
{"step": 83175, "loss": 3.462047576904297, "grad_norm": 0.7249038219451904, "learning_rate": 0.0001, "elbo": 3.46205, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8777269528501055}
{"step": 83180, "loss": 3.499233627319336, "grad_norm": 0.7804973721504211, "learning_rate": 0.0001, "elbo": 3.49923, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.878020173586676}
{"step": 83185, "loss": 3.5024505615234376, "grad_norm": 0.682937741279602, "learning_rate": 0.0001, "elbo": 3.50245, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.878313394323246}
{"step": 83190, "loss": 3.499163818359375, "grad_norm": 0.7134289741516113, "learning_rate": 0.0001, "elbo": 3.49916, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.878606615059817}
{"step": 83195, "loss": 3.5076995849609376, "grad_norm": 0.7116966247558594, "learning_rate": 0.0001, "elbo": 3.5077, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.878899835796387}
{"step": 83200, "loss": 3.4848606109619142, "grad_norm": 0.7435152530670166, "learning_rate": 0.0001, "elbo": 3.48486, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.879193056532958}
{"step": 83205, "loss": 3.5003623962402344, "grad_norm": 0.7212510108947754, "learning_rate": 0.0001, "elbo": 3.50036, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.879486277269528}
{"step": 83210, "loss": 3.4796215057373048, "grad_norm": 0.8436720967292786, "learning_rate": 0.0001, "elbo": 3.47962, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.879779498006099}
{"step": 83215, "loss": 3.5001480102539064, "grad_norm": 0.6793888211250305, "learning_rate": 0.0001, "elbo": 3.50015, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.880072718742669}
{"step": 83220, "loss": 3.499968719482422, "grad_norm": 0.709691047668457, "learning_rate": 0.0001, "elbo": 3.49997, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8803659394792405}
{"step": 83225, "loss": 3.496550750732422, "grad_norm": 0.7733244299888611, "learning_rate": 0.0001, "elbo": 3.49655, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.88065916021581}
{"step": 83230, "loss": 3.4979225158691407, "grad_norm": 0.7424618005752563, "learning_rate": 0.0001, "elbo": 3.49792, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.880952380952381}
{"step": 83235, "loss": 3.4892711639404297, "grad_norm": 0.7544719576835632, "learning_rate": 0.0001, "elbo": 3.48927, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.881245601688952}
{"step": 83240, "loss": 3.504510498046875, "grad_norm": 0.7534650564193726, "learning_rate": 0.0001, "elbo": 3.50451, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.881538822425522}
{"step": 83245, "loss": 3.482727813720703, "grad_norm": 0.6688796281814575, "learning_rate": 0.0001, "elbo": 3.48273, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.881832043162093}
{"step": 83250, "loss": 3.4849502563476564, "grad_norm": 0.7566953301429749, "learning_rate": 0.0001, "elbo": 3.48495, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.882125263898663}
{"step": 83255, "loss": 3.485587310791016, "grad_norm": 0.7834171652793884, "learning_rate": 0.0001, "elbo": 3.48559, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.882418484635234}
{"step": 83260, "loss": 3.5036575317382814, "grad_norm": 0.7329478859901428, "learning_rate": 0.0001, "elbo": 3.50366, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.882711705371804}
{"step": 83265, "loss": 3.4842460632324217, "grad_norm": 0.712327778339386, "learning_rate": 0.0001, "elbo": 3.48425, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8830049261083746}
{"step": 83270, "loss": 3.505626678466797, "grad_norm": 0.7093786001205444, "learning_rate": 0.0001, "elbo": 3.50563, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.883298146844945}
{"step": 83275, "loss": 3.498801422119141, "grad_norm": 0.6974519491195679, "learning_rate": 0.0001, "elbo": 3.4988, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8835913675815155}
{"step": 83280, "loss": 3.4925521850585937, "grad_norm": 0.6932833194732666, "learning_rate": 0.0001, "elbo": 3.49255, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.883884588318086}
{"step": 83285, "loss": 3.489126968383789, "grad_norm": 0.7230855822563171, "learning_rate": 0.0001, "elbo": 3.48913, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.884177809054656}
{"step": 83290, "loss": 3.5076763153076174, "grad_norm": 0.7033738493919373, "learning_rate": 0.0001, "elbo": 3.50768, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.884471029791227}
{"step": 83295, "loss": 3.476409912109375, "grad_norm": 0.7762954831123352, "learning_rate": 0.0001, "elbo": 3.47641, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.884764250527797}
{"step": 83300, "loss": 3.4981548309326174, "grad_norm": 0.724222719669342, "learning_rate": 0.0001, "elbo": 3.49815, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.885057471264368}
{"step": 83305, "loss": 3.510614013671875, "grad_norm": 0.7068251967430115, "learning_rate": 0.0001, "elbo": 3.51061, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.885350692000938}
{"step": 83310, "loss": 3.4968605041503906, "grad_norm": 0.6782235503196716, "learning_rate": 0.0001, "elbo": 3.49686, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.885643912737509}
{"step": 83315, "loss": 3.5059852600097656, "grad_norm": 0.6965068578720093, "learning_rate": 0.0001, "elbo": 3.50598, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.885937133474079}
{"step": 83320, "loss": 3.472481918334961, "grad_norm": 0.7071253061294556, "learning_rate": 0.0001, "elbo": 3.47248, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8862303542106496}
{"step": 83325, "loss": 3.4844741821289062, "grad_norm": 0.7221196889877319, "learning_rate": 0.0001, "elbo": 3.48447, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.88652357494722}
{"step": 83330, "loss": 3.5067214965820312, "grad_norm": 0.7149008512496948, "learning_rate": 0.0001, "elbo": 3.50672, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8868167956837905}
{"step": 83335, "loss": 3.4758750915527346, "grad_norm": 0.6974737048149109, "learning_rate": 0.0001, "elbo": 3.47588, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.887110016420361}
{"step": 83340, "loss": 3.5079750061035155, "grad_norm": 0.6910603642463684, "learning_rate": 0.0001, "elbo": 3.50797, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.887403237156931}
{"step": 83345, "loss": 3.4883964538574217, "grad_norm": 0.6981201767921448, "learning_rate": 0.0001, "elbo": 3.4884, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.887696457893503}
{"step": 83350, "loss": 3.4930248260498047, "grad_norm": 0.6951878666877747, "learning_rate": 0.0001, "elbo": 3.49302, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.887989678630072}
{"step": 83355, "loss": 3.4837390899658205, "grad_norm": 0.6596813797950745, "learning_rate": 0.0001, "elbo": 3.48374, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.888282899366644}
{"step": 83360, "loss": 3.4937294006347654, "grad_norm": 0.6859157681465149, "learning_rate": 0.0001, "elbo": 3.49373, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.888576120103214}
{"step": 83365, "loss": 3.5041824340820313, "grad_norm": 0.7164590954780579, "learning_rate": 0.0001, "elbo": 3.50418, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8888693408397845}
{"step": 83370, "loss": 3.486491394042969, "grad_norm": 0.6728254556655884, "learning_rate": 0.0001, "elbo": 3.48649, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.889162561576355}
{"step": 83375, "loss": 3.497707748413086, "grad_norm": 0.7595943212509155, "learning_rate": 0.0001, "elbo": 3.49771, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.889455782312925}
{"step": 83380, "loss": 3.4939762115478517, "grad_norm": 0.7038776874542236, "learning_rate": 0.0001, "elbo": 3.49398, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.889749003049496}
{"step": 83385, "loss": 3.4935646057128906, "grad_norm": 0.6638256311416626, "learning_rate": 0.0001, "elbo": 3.49356, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.890042223786066}
{"step": 83390, "loss": 3.4957447052001953, "grad_norm": 0.6994006633758545, "learning_rate": 0.0001, "elbo": 3.49574, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.890335444522637}
{"step": 83395, "loss": 3.508045196533203, "grad_norm": 0.7388743162155151, "learning_rate": 0.0001, "elbo": 3.50804, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.890628665259207}
{"step": 83395, "loss": 3.508045196533203, "grad_norm": 0.7388743162155151, "learning_rate": 0.0001, "elbo": 3.50804, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.890628665259207}
{"step": 83400, "loss": 3.471612548828125, "grad_norm": 0.6920526623725891, "learning_rate": 0.0001, "elbo": 3.47161, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.890921885995778}
{"step": 83405, "loss": 3.506687545776367, "grad_norm": 0.782055139541626, "learning_rate": 0.0001, "elbo": 3.50669, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.891215106732348}
{"step": 83410, "loss": 3.515023040771484, "grad_norm": 0.7299430966377258, "learning_rate": 0.0001, "elbo": 3.51502, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.891508327468919}
{"step": 83410, "loss": 3.515023040771484, "grad_norm": 0.7299430966377258, "learning_rate": 0.0001, "elbo": 3.51502, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.891508327468919}
{"step": 83415, "loss": 3.501251220703125, "grad_norm": 0.8293590545654297, "learning_rate": 0.0001, "elbo": 3.50125, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.891801548205489}
{"step": 83420, "loss": 3.4692726135253906, "grad_norm": 0.7426204681396484, "learning_rate": 0.0001, "elbo": 3.46927, "mask_frac": 0.4971, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8920947689420595}
{"step": 83425, "loss": 3.4867897033691406, "grad_norm": 0.7449508905410767, "learning_rate": 0.0001, "elbo": 3.48679, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.89238798967863}
{"step": 83425, "loss": 3.4867897033691406, "grad_norm": 0.7449508905410767, "learning_rate": 0.0001, "elbo": 3.48679, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.89238798967863}
{"step": 83430, "loss": 3.5278366088867186, "grad_norm": 0.7249566912651062, "learning_rate": 0.0001, "elbo": 3.52784, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8926812104152}
{"step": 83435, "loss": 3.5080570220947265, "grad_norm": 0.8704858422279358, "learning_rate": 0.0001, "elbo": 3.50806, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.892974431151771}
{"step": 83440, "loss": 3.497848129272461, "grad_norm": 0.7376478314399719, "learning_rate": 0.0001, "elbo": 3.49785, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.893267651888341}
{"step": 83445, "loss": 3.5001705169677733, "grad_norm": 0.7113161683082581, "learning_rate": 0.0001, "elbo": 3.50017, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.893560872624912}
{"step": 83450, "loss": 3.492644500732422, "grad_norm": 0.7505766749382019, "learning_rate": 0.0001, "elbo": 3.49264, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.893854093361482}
{"step": 83455, "loss": 3.484447479248047, "grad_norm": 0.8976273536682129, "learning_rate": 0.0001, "elbo": 3.48445, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.894147314098053}
{"step": 83460, "loss": 3.4903987884521483, "grad_norm": 0.7027177810668945, "learning_rate": 0.0001, "elbo": 3.4904, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.894440534834623}
{"step": 83465, "loss": 3.499620819091797, "grad_norm": 0.7289538383483887, "learning_rate": 0.0001, "elbo": 3.49962, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.894733755571194}
{"step": 83470, "loss": 3.4857254028320312, "grad_norm": 0.7177911400794983, "learning_rate": 0.0001, "elbo": 3.48573, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.895026976307765}
{"step": 83475, "loss": 3.487323760986328, "grad_norm": 0.7301998138427734, "learning_rate": 0.0001, "elbo": 3.48732, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8953201970443345}
{"step": 83480, "loss": 3.493150329589844, "grad_norm": 0.8268824815750122, "learning_rate": 0.0001, "elbo": 3.49315, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.895613417780906}
{"step": 83485, "loss": 3.502548599243164, "grad_norm": 0.7452496290206909, "learning_rate": 0.0001, "elbo": 3.50255, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.895906638517476}
{"step": 83490, "loss": 3.5000648498535156, "grad_norm": 0.7000595927238464, "learning_rate": 0.0001, "elbo": 3.50007, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.896199859254047}
{"step": 83495, "loss": 3.4968048095703126, "grad_norm": 0.7913991212844849, "learning_rate": 0.0001, "elbo": 3.4968, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.896493079990617}
{"step": 83500, "loss": 3.501905822753906, "grad_norm": 0.7310631275177002, "learning_rate": 0.0001, "elbo": 3.50191, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.896786300727188}
{"step": 83505, "loss": 3.4973991394042967, "grad_norm": 0.7942534685134888, "learning_rate": 0.0001, "elbo": 3.4974, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.897079521463758}
{"step": 83510, "loss": 3.498265838623047, "grad_norm": 0.6907618641853333, "learning_rate": 0.0001, "elbo": 3.49827, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.897372742200329}
{"step": 83515, "loss": 3.495853042602539, "grad_norm": 0.7959171533584595, "learning_rate": 0.0001, "elbo": 3.49585, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.897665962936899}
{"step": 83520, "loss": 3.497491455078125, "grad_norm": 0.7228512763977051, "learning_rate": 0.0001, "elbo": 3.49749, "mask_frac": 0.5014, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.8979591836734695}
{"step": 83525, "loss": 3.5011276245117187, "grad_norm": 0.841781735420227, "learning_rate": 0.0001, "elbo": 3.50113, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.89825240441004}
{"step": 83530, "loss": 3.516423797607422, "grad_norm": 0.7288888692855835, "learning_rate": 0.0001, "elbo": 3.51642, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.89854562514661}
{"step": 83535, "loss": 3.500827407836914, "grad_norm": 0.7836936712265015, "learning_rate": 0.0001, "elbo": 3.50083, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.898838845883181}
{"step": 83540, "loss": 3.4933395385742188, "grad_norm": 0.6896732449531555, "learning_rate": 0.0001, "elbo": 3.49334, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.899132066619751}
{"step": 83545, "loss": 3.5005874633789062, "grad_norm": 0.8682079315185547, "learning_rate": 0.0001, "elbo": 3.50059, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.899425287356322}
{"step": 83550, "loss": 3.4916713714599608, "grad_norm": 0.8327159285545349, "learning_rate": 0.0001, "elbo": 3.49167, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.899718508092892}
{"step": 83555, "loss": 3.4908233642578126, "grad_norm": 0.7638708353042603, "learning_rate": 0.0001, "elbo": 3.49082, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.900011728829463}
{"step": 83560, "loss": 3.504075622558594, "grad_norm": 0.8847616314888, "learning_rate": 0.0001, "elbo": 3.50408, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.900304949566033}
{"step": 83565, "loss": 3.4954448699951173, "grad_norm": 0.763765275478363, "learning_rate": 0.0001, "elbo": 3.49544, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.900598170302604}
{"step": 83570, "loss": 3.5057018280029295, "grad_norm": 0.7475773692131042, "learning_rate": 0.0001, "elbo": 3.5057, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.900891391039174}
{"step": 83570, "loss": 3.5057018280029295, "grad_norm": 0.7475773692131042, "learning_rate": 0.0001, "elbo": 3.5057, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.900891391039174}
{"step": 83575, "loss": 3.489049530029297, "grad_norm": 0.7153265476226807, "learning_rate": 0.0001, "elbo": 3.48905, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9011846117757445}
{"step": 83580, "loss": 3.4957504272460938, "grad_norm": 0.7062188982963562, "learning_rate": 0.0001, "elbo": 3.49575, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.901477832512315}
{"step": 83585, "loss": 3.5127590179443358, "grad_norm": 0.7646172642707825, "learning_rate": 0.0001, "elbo": 3.51276, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.901771053248885}
{"step": 83590, "loss": 3.4929244995117186, "grad_norm": 0.7293347716331482, "learning_rate": 0.0001, "elbo": 3.49292, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.902064273985456}
{"step": 83590, "loss": 3.4929244995117186, "grad_norm": 0.7293347716331482, "learning_rate": 0.0001, "elbo": 3.49292, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.902064273985456}
{"step": 83595, "loss": 3.4864456176757814, "grad_norm": 0.6935465931892395, "learning_rate": 0.0001, "elbo": 3.48645, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.902357494722027}
{"step": 83600, "loss": 3.5114364624023438, "grad_norm": 0.7334343791007996, "learning_rate": 0.0001, "elbo": 3.51144, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.902650715458597}
{"step": 83605, "loss": 3.4979022979736327, "grad_norm": 0.8232148885726929, "learning_rate": 0.0001, "elbo": 3.4979, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.902943936195168}
{"step": 83610, "loss": 3.498303985595703, "grad_norm": 0.6961170434951782, "learning_rate": 0.0001, "elbo": 3.4983, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.903237156931739}
{"step": 83615, "loss": 3.511296844482422, "grad_norm": 0.6638327240943909, "learning_rate": 0.0001, "elbo": 3.5113, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.903530377668309}
{"step": 83620, "loss": 3.4833446502685548, "grad_norm": 1.1093182563781738, "learning_rate": 0.0001, "elbo": 3.48334, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9038235984048795}
{"step": 83620, "loss": 3.4833446502685548, "grad_norm": 1.1093182563781738, "learning_rate": 0.0001, "elbo": 3.48334, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9038235984048795}
{"step": 83625, "loss": 3.5030845642089843, "grad_norm": 0.6953575015068054, "learning_rate": 0.0001, "elbo": 3.50308, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.90411681914145}
{"step": 83630, "loss": 3.500415802001953, "grad_norm": 0.7178224921226501, "learning_rate": 0.0001, "elbo": 3.50042, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.90441003987802}
{"step": 83635, "loss": 3.4865367889404295, "grad_norm": 0.7274589538574219, "learning_rate": 0.0001, "elbo": 3.48654, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.904703260614591}
{"step": 83640, "loss": 3.487117385864258, "grad_norm": 0.7438669800758362, "learning_rate": 0.0001, "elbo": 3.48712, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.904996481351161}
{"step": 83645, "loss": 3.497303009033203, "grad_norm": 0.8163936734199524, "learning_rate": 0.0001, "elbo": 3.4973, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.905289702087732}
{"step": 83650, "loss": 3.499214935302734, "grad_norm": 0.792824923992157, "learning_rate": 0.0001, "elbo": 3.49921, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.905582922824302}
{"step": 83655, "loss": 3.5007530212402345, "grad_norm": 0.7069565653800964, "learning_rate": 0.0001, "elbo": 3.50075, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.905876143560873}
{"step": 83660, "loss": 3.497661590576172, "grad_norm": 0.7416119575500488, "learning_rate": 0.0001, "elbo": 3.49766, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.906169364297443}
{"step": 83665, "loss": 3.504804229736328, "grad_norm": 0.7099340558052063, "learning_rate": 0.0001, "elbo": 3.5048, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.906462585034014}
{"step": 83670, "loss": 3.5156017303466798, "grad_norm": 0.6891763210296631, "learning_rate": 0.0001, "elbo": 3.5156, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.906755805770584}
{"step": 83670, "loss": 3.5156017303466798, "grad_norm": 0.6891763210296631, "learning_rate": 0.0001, "elbo": 3.5156, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.906755805770584}
{"step": 83675, "loss": 3.4901451110839843, "grad_norm": 0.7535320520401001, "learning_rate": 0.0001, "elbo": 3.49015, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9070490265071545}
{"step": 83680, "loss": 3.4918399810791017, "grad_norm": 0.686909019947052, "learning_rate": 0.0001, "elbo": 3.49184, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.907342247243725}
{"step": 83680, "loss": 3.4918399810791017, "grad_norm": 0.686909019947052, "learning_rate": 0.0001, "elbo": 3.49184, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.907342247243725}
{"step": 83685, "loss": 3.4878494262695314, "grad_norm": 0.7436433434486389, "learning_rate": 0.0001, "elbo": 3.48785, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.907635467980295}
{"step": 83690, "loss": 3.493427276611328, "grad_norm": 0.7163469791412354, "learning_rate": 0.0001, "elbo": 3.49343, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.907928688716866}
{"step": 83695, "loss": 3.4958641052246096, "grad_norm": 0.7528257966041565, "learning_rate": 0.0001, "elbo": 3.49586, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.908221909453436}
{"step": 83700, "loss": 3.497180938720703, "grad_norm": 0.7707349061965942, "learning_rate": 0.0001, "elbo": 3.49718, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.908515130190007}
{"step": 83705, "loss": 3.5036991119384764, "grad_norm": 0.7424963712692261, "learning_rate": 0.0001, "elbo": 3.5037, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.908808350926577}
{"step": 83710, "loss": 3.506255340576172, "grad_norm": 0.7139885425567627, "learning_rate": 0.0001, "elbo": 3.50626, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.909101571663148}
{"step": 83715, "loss": 3.5034881591796876, "grad_norm": 0.7198249697685242, "learning_rate": 0.0001, "elbo": 3.50349, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.909394792399718}
{"step": 83720, "loss": 3.494484710693359, "grad_norm": 0.7425593137741089, "learning_rate": 0.0001, "elbo": 3.49448, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.909688013136289}
{"step": 83725, "loss": 3.492043304443359, "grad_norm": 0.8383042812347412, "learning_rate": 0.0001, "elbo": 3.49204, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.909981233872859}
{"step": 83730, "loss": 3.501612091064453, "grad_norm": 0.6979873180389404, "learning_rate": 0.0001, "elbo": 3.50161, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.91027445460943}
{"step": 83735, "loss": 3.4808998107910156, "grad_norm": 0.7854178547859192, "learning_rate": 0.0001, "elbo": 3.4809, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.910567675346001}
{"step": 83740, "loss": 3.4826988220214843, "grad_norm": 0.8574392795562744, "learning_rate": 0.0001, "elbo": 3.4827, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.910860896082571}
{"step": 83745, "loss": 3.4953292846679687, "grad_norm": 0.7535037994384766, "learning_rate": 0.0001, "elbo": 3.49533, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.911154116819142}
{"step": 83750, "loss": 3.5115478515625, "grad_norm": 0.7183007597923279, "learning_rate": 0.0001, "elbo": 3.51155, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.911447337555712}
{"step": 83755, "loss": 3.4967720031738283, "grad_norm": 0.6900169253349304, "learning_rate": 0.0001, "elbo": 3.49677, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.911740558292283}
{"step": 83760, "loss": 3.5125755310058593, "grad_norm": 0.7065039277076721, "learning_rate": 0.0001, "elbo": 3.51258, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.912033779028853}
{"step": 83765, "loss": 3.5010498046875, "grad_norm": 0.7812932729721069, "learning_rate": 0.0001, "elbo": 3.50105, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9123269997654235}
{"step": 83770, "loss": 3.4841163635253904, "grad_norm": 0.7891620397567749, "learning_rate": 0.0001, "elbo": 3.48412, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.912620220501994}
{"step": 83775, "loss": 3.500015640258789, "grad_norm": 0.6953384280204773, "learning_rate": 0.0001, "elbo": 3.50002, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.912913441238564}
{"step": 83780, "loss": 3.5082672119140623, "grad_norm": 0.7122406959533691, "learning_rate": 0.0001, "elbo": 3.50827, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.913206661975135}
{"step": 83785, "loss": 3.4863822937011717, "grad_norm": 0.7794806957244873, "learning_rate": 0.0001, "elbo": 3.48638, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.913499882711705}
{"step": 83790, "loss": 3.4979698181152346, "grad_norm": 0.7928766012191772, "learning_rate": 0.0001, "elbo": 3.49797, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.913793103448276}
{"step": 83795, "loss": 3.4974990844726563, "grad_norm": 0.7513214349746704, "learning_rate": 0.0001, "elbo": 3.4975, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.914086324184846}
{"step": 83800, "loss": 3.500993347167969, "grad_norm": 0.7240204811096191, "learning_rate": 0.0001, "elbo": 3.50099, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.914379544921417}
{"step": 83805, "loss": 3.492808532714844, "grad_norm": 0.6997705698013306, "learning_rate": 0.0001, "elbo": 3.49281, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.914672765657987}
{"step": 83810, "loss": 3.5034637451171875, "grad_norm": 1.0613662004470825, "learning_rate": 0.0001, "elbo": 3.50346, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.914965986394558}
{"step": 83815, "loss": 3.5040077209472655, "grad_norm": 0.7981284260749817, "learning_rate": 0.0001, "elbo": 3.50401, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.915259207131128}
{"step": 83820, "loss": 3.497515869140625, "grad_norm": 0.7784470915794373, "learning_rate": 0.0001, "elbo": 3.49752, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9155524278676985}
{"step": 83825, "loss": 3.4866458892822267, "grad_norm": 0.7273731231689453, "learning_rate": 0.0001, "elbo": 3.48665, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.915845648604269}
{"step": 83830, "loss": 3.5189529418945313, "grad_norm": 0.7482563853263855, "learning_rate": 0.0001, "elbo": 3.51895, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.916138869340839}
{"step": 83835, "loss": 3.4953384399414062, "grad_norm": 0.7057592272758484, "learning_rate": 0.0001, "elbo": 3.49534, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.91643209007741}
{"step": 83840, "loss": 3.4907630920410155, "grad_norm": 0.7468502521514893, "learning_rate": 0.0001, "elbo": 3.49076, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.91672531081398}
{"step": 83845, "loss": 3.4988975524902344, "grad_norm": 0.7354614734649658, "learning_rate": 0.0001, "elbo": 3.4989, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.917018531550552}
{"step": 83850, "loss": 3.498706817626953, "grad_norm": 0.7373555302619934, "learning_rate": 0.0001, "elbo": 3.49871, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.917311752287121}
{"step": 83855, "loss": 3.5097259521484374, "grad_norm": 0.7166746854782104, "learning_rate": 0.0001, "elbo": 3.50973, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.917604973023693}
{"step": 83860, "loss": 3.4931991577148436, "grad_norm": 0.6920308470726013, "learning_rate": 0.0001, "elbo": 3.4932, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.917898193760263}
{"step": 83865, "loss": 3.48270263671875, "grad_norm": 0.7047123908996582, "learning_rate": 0.0001, "elbo": 3.4827, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9181914144968335}
{"step": 83870, "loss": 3.4968032836914062, "grad_norm": 0.7316688299179077, "learning_rate": 0.0001, "elbo": 3.4968, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.918484635233404}
{"step": 83875, "loss": 3.490580749511719, "grad_norm": 0.7602513432502747, "learning_rate": 0.0001, "elbo": 3.49058, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.918777855969974}
{"step": 83880, "loss": 3.487540435791016, "grad_norm": 0.7087960243225098, "learning_rate": 0.0001, "elbo": 3.48754, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.919071076706545}
{"step": 83885, "loss": 3.4927047729492187, "grad_norm": 0.7015985250473022, "learning_rate": 0.0001, "elbo": 3.4927, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.919364297443115}
{"step": 83890, "loss": 3.491834259033203, "grad_norm": 0.7337155342102051, "learning_rate": 0.0001, "elbo": 3.49183, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.919657518179686}
{"step": 83895, "loss": 3.4977882385253904, "grad_norm": 0.7034449577331543, "learning_rate": 0.0001, "elbo": 3.49779, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.919950738916256}
{"step": 83900, "loss": 3.489837646484375, "grad_norm": 0.7164922952651978, "learning_rate": 0.0001, "elbo": 3.48984, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.920243959652827}
{"step": 83905, "loss": 3.497454833984375, "grad_norm": 0.7382147908210754, "learning_rate": 0.0001, "elbo": 3.49746, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.920537180389397}
{"step": 83910, "loss": 3.4882179260253907, "grad_norm": 0.6894165277481079, "learning_rate": 0.0001, "elbo": 3.48822, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.920830401125968}
{"step": 83915, "loss": 3.484907531738281, "grad_norm": 0.7712368965148926, "learning_rate": 0.0001, "elbo": 3.48491, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.921123621862538}
{"step": 83920, "loss": 3.497695541381836, "grad_norm": 0.8066070079803467, "learning_rate": 0.0001, "elbo": 3.4977, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9214168425991085}
{"step": 83925, "loss": 3.488026428222656, "grad_norm": 0.6993116736412048, "learning_rate": 0.0001, "elbo": 3.48803, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.921710063335679}
{"step": 83930, "loss": 3.4825321197509767, "grad_norm": 0.7184838652610779, "learning_rate": 0.0001, "elbo": 3.48253, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.922003284072249}
{"step": 83935, "loss": 3.500971221923828, "grad_norm": 0.7809992432594299, "learning_rate": 0.0001, "elbo": 3.50097, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.92229650480882}
{"step": 83940, "loss": 3.5019649505615233, "grad_norm": 0.7324598431587219, "learning_rate": 0.0001, "elbo": 3.50196, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.92258972554539}
{"step": 83945, "loss": 3.4991588592529297, "grad_norm": 0.7909055352210999, "learning_rate": 0.0001, "elbo": 3.49916, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.922882946281961}
{"step": 83950, "loss": 3.508345031738281, "grad_norm": 0.7286643981933594, "learning_rate": 0.0001, "elbo": 3.50834, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.923176167018531}
{"step": 83955, "loss": 3.4934928894042967, "grad_norm": 0.8073352575302124, "learning_rate": 0.0001, "elbo": 3.49349, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.923469387755102}
{"step": 83960, "loss": 3.512704849243164, "grad_norm": 0.7014712691307068, "learning_rate": 0.0001, "elbo": 3.5127, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.923762608491673}
{"step": 83965, "loss": 3.4854202270507812, "grad_norm": 0.7203994989395142, "learning_rate": 0.0001, "elbo": 3.48542, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.924055829228243}
{"step": 83970, "loss": 3.4969459533691407, "grad_norm": 0.7283181548118591, "learning_rate": 0.0001, "elbo": 3.49695, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.924349049964814}
{"step": 83975, "loss": 3.499530029296875, "grad_norm": 0.7696388959884644, "learning_rate": 0.0001, "elbo": 3.49953, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.924642270701384}
{"step": 83980, "loss": 3.506898880004883, "grad_norm": 0.6729936599731445, "learning_rate": 0.0001, "elbo": 3.5069, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.924935491437955}
{"step": 83985, "loss": 3.490509033203125, "grad_norm": 0.7691330313682556, "learning_rate": 0.0001, "elbo": 3.49051, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.925228712174525}
{"step": 83990, "loss": 3.495008087158203, "grad_norm": 0.7162608504295349, "learning_rate": 0.0001, "elbo": 3.49501, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.925521932911096}
{"step": 83995, "loss": 3.483935546875, "grad_norm": 0.6894747614860535, "learning_rate": 0.0001, "elbo": 3.48394, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.925815153647666}
{"step": 84000, "loss": 3.500143051147461, "grad_norm": 0.7377825379371643, "learning_rate": 0.0001, "elbo": 3.50014, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.926108374384237}
{"step": 84005, "loss": 3.4924575805664064, "grad_norm": 0.7222529053688049, "learning_rate": 0.0001, "elbo": 3.49246, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.926401595120807}
{"step": 84010, "loss": 3.4887786865234376, "grad_norm": 0.6929872035980225, "learning_rate": 0.0001, "elbo": 3.48878, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.926694815857378}
{"step": 84015, "loss": 3.4869758605957033, "grad_norm": 0.8053116202354431, "learning_rate": 0.0001, "elbo": 3.48698, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.926988036593948}
{"step": 84020, "loss": 3.4889129638671874, "grad_norm": 0.6956626176834106, "learning_rate": 0.0001, "elbo": 3.48891, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9272812573305185}
{"step": 84025, "loss": 3.491933822631836, "grad_norm": 0.7646806836128235, "learning_rate": 0.0001, "elbo": 3.49193, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.927574478067089}
{"step": 84030, "loss": 3.505419921875, "grad_norm": 0.6946018934249878, "learning_rate": 0.0001, "elbo": 3.50542, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.927867698803659}
{"step": 84035, "loss": 3.503845977783203, "grad_norm": 0.706814706325531, "learning_rate": 0.0001, "elbo": 3.50385, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.92816091954023}
{"step": 84035, "loss": 3.503845977783203, "grad_norm": 0.706814706325531, "learning_rate": 0.0001, "elbo": 3.50385, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.92816091954023}
{"step": 84040, "loss": 3.508905792236328, "grad_norm": 0.7175288200378418, "learning_rate": 0.0001, "elbo": 3.50891, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9284541402768}
{"step": 84045, "loss": 3.5027797698974608, "grad_norm": 0.7285048961639404, "learning_rate": 0.0001, "elbo": 3.50278, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.928747361013371}
{"step": 84050, "loss": 3.492330551147461, "grad_norm": 0.7448208332061768, "learning_rate": 0.0001, "elbo": 3.49233, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.929040581749941}
{"step": 84055, "loss": 3.4995040893554688, "grad_norm": 0.7628218531608582, "learning_rate": 0.0001, "elbo": 3.4995, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.929333802486512}
{"step": 84060, "loss": 3.513418197631836, "grad_norm": 1.163370132446289, "learning_rate": 0.0001, "elbo": 3.51342, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.929627023223082}
{"step": 84065, "loss": 3.492961120605469, "grad_norm": 0.6938697695732117, "learning_rate": 0.0001, "elbo": 3.49296, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.929920243959653}
{"step": 84070, "loss": 3.5109973907470704, "grad_norm": 0.7906933426856995, "learning_rate": 0.0001, "elbo": 3.511, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.930213464696223}
{"step": 84075, "loss": 3.506866455078125, "grad_norm": 0.6460565328598022, "learning_rate": 0.0001, "elbo": 3.50687, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9305066854327935}
{"step": 84080, "loss": 3.4878631591796876, "grad_norm": 0.9401043057441711, "learning_rate": 0.0001, "elbo": 3.48786, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.930799906169364}
{"step": 84085, "loss": 3.494994354248047, "grad_norm": 0.6905501484870911, "learning_rate": 0.0001, "elbo": 3.49499, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.931093126905935}
{"step": 84090, "loss": 3.487052154541016, "grad_norm": 0.942162036895752, "learning_rate": 0.0001, "elbo": 3.48705, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.931386347642505}
{"step": 84095, "loss": 3.4848587036132814, "grad_norm": 0.6723483204841614, "learning_rate": 0.0001, "elbo": 3.48486, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.931679568379076}
{"step": 84100, "loss": 3.492344284057617, "grad_norm": 0.7030876278877258, "learning_rate": 0.0001, "elbo": 3.49234, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.931972789115647}
{"step": 84105, "loss": 3.4956295013427736, "grad_norm": 0.739811897277832, "learning_rate": 0.0001, "elbo": 3.49563, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.932266009852217}
{"step": 84110, "loss": 3.4942550659179688, "grad_norm": 0.8673384785652161, "learning_rate": 0.0001, "elbo": 3.49426, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9325592305887875}
{"step": 84115, "loss": 3.50174560546875, "grad_norm": 0.8105519413948059, "learning_rate": 0.0001, "elbo": 3.50175, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.932852451325358}
{"step": 84120, "loss": 3.5106231689453127, "grad_norm": 0.8609432578086853, "learning_rate": 0.0001, "elbo": 3.51062, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.933145672061928}
{"step": 84125, "loss": 3.5000320434570313, "grad_norm": 0.7652453184127808, "learning_rate": 0.0001, "elbo": 3.50003, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.933438892798499}
{"step": 84125, "loss": 3.5000320434570313, "grad_norm": 0.7652453184127808, "learning_rate": 0.0001, "elbo": 3.50003, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.933438892798499}
{"step": 84130, "loss": 3.5033279418945313, "grad_norm": 0.6923068165779114, "learning_rate": 0.0001, "elbo": 3.50333, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.933732113535069}
{"step": 84135, "loss": 3.492204284667969, "grad_norm": 0.6776753664016724, "learning_rate": 0.0001, "elbo": 3.4922, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.93402533427164}
{"step": 84140, "loss": 3.489678955078125, "grad_norm": 0.7222438454627991, "learning_rate": 0.0001, "elbo": 3.48968, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.93431855500821}
{"step": 84145, "loss": 3.5088127136230467, "grad_norm": 0.7389116883277893, "learning_rate": 0.0001, "elbo": 3.50881, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.934611775744781}
{"step": 84150, "loss": 3.484740447998047, "grad_norm": 0.6952423453330994, "learning_rate": 0.0001, "elbo": 3.48474, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.934904996481351}
{"step": 84155, "loss": 3.4753486633300783, "grad_norm": 0.6952046751976013, "learning_rate": 0.0001, "elbo": 3.47535, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.935198217217922}
{"step": 84160, "loss": 3.494904708862305, "grad_norm": 0.7458503842353821, "learning_rate": 0.0001, "elbo": 3.4949, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.935491437954492}
{"step": 84165, "loss": 3.4860130310058595, "grad_norm": 0.7070327401161194, "learning_rate": 0.0001, "elbo": 3.48601, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9357846586910625}
{"step": 84170, "loss": 3.4891563415527345, "grad_norm": 0.68682461977005, "learning_rate": 0.0001, "elbo": 3.48916, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.936077879427633}
{"step": 84175, "loss": 3.5041149139404295, "grad_norm": 0.7781051993370056, "learning_rate": 0.0001, "elbo": 3.50411, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.936371100164203}
{"step": 84180, "loss": 3.487099456787109, "grad_norm": 0.6995980739593506, "learning_rate": 0.0001, "elbo": 3.4871, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.936664320900774}
{"step": 84185, "loss": 3.497239685058594, "grad_norm": 0.7458749413490295, "learning_rate": 0.0001, "elbo": 3.49724, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.936957541637344}
{"step": 84190, "loss": 3.5006332397460938, "grad_norm": 0.7354344129562378, "learning_rate": 0.0001, "elbo": 3.50063, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.937250762373915}
{"step": 84195, "loss": 3.502781295776367, "grad_norm": 0.9252400398254395, "learning_rate": 0.0001, "elbo": 3.50278, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.937543983110485}
{"step": 84200, "loss": 3.496855926513672, "grad_norm": 0.6776512265205383, "learning_rate": 0.0001, "elbo": 3.49686, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.937837203847056}
{"step": 84205, "loss": 3.488560104370117, "grad_norm": 0.6886759996414185, "learning_rate": 0.0001, "elbo": 3.48856, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.938130424583626}
{"step": 84210, "loss": 3.4902008056640623, "grad_norm": 0.7435890436172485, "learning_rate": 0.0001, "elbo": 3.4902, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9384236453201975}
{"step": 84215, "loss": 3.4983306884765626, "grad_norm": 0.7689142227172852, "learning_rate": 0.0001, "elbo": 3.49833, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.938716866056767}
{"step": 84220, "loss": 3.5023670196533203, "grad_norm": 0.7357152700424194, "learning_rate": 0.0001, "elbo": 3.50237, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.939010086793338}
{"step": 84225, "loss": 3.5021736145019533, "grad_norm": 0.7806232571601868, "learning_rate": 0.0001, "elbo": 3.50217, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.939303307529909}
{"step": 84230, "loss": 3.5001392364501953, "grad_norm": 0.9404172301292419, "learning_rate": 0.0001, "elbo": 3.50014, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.939596528266479}
{"step": 84235, "loss": 3.481394577026367, "grad_norm": 0.7027052640914917, "learning_rate": 0.0001, "elbo": 3.48139, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.93988974900305}
{"step": 84240, "loss": 3.4950218200683594, "grad_norm": 0.6737762093544006, "learning_rate": 0.0001, "elbo": 3.49502, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.94018296973962}
{"step": 84245, "loss": 3.4964523315429688, "grad_norm": 0.8003794550895691, "learning_rate": 0.0001, "elbo": 3.49645, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.940476190476191}
{"step": 84250, "loss": 3.493097686767578, "grad_norm": 0.7694846391677856, "learning_rate": 0.0001, "elbo": 3.4931, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.940769411212761}
{"step": 84255, "loss": 3.4799468994140623, "grad_norm": 0.7766886353492737, "learning_rate": 0.0001, "elbo": 3.47995, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.941062631949332}
{"step": 84260, "loss": 3.488303375244141, "grad_norm": 0.7399080395698547, "learning_rate": 0.0001, "elbo": 3.4883, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.941355852685902}
{"step": 84265, "loss": 3.5002635955810546, "grad_norm": 0.7116584181785583, "learning_rate": 0.0001, "elbo": 3.50026, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9416490734224725}
{"step": 84270, "loss": 3.4987789154052735, "grad_norm": 0.7359263896942139, "learning_rate": 0.0001, "elbo": 3.49878, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.941942294159043}
{"step": 84275, "loss": 3.4996383666992186, "grad_norm": 0.7415289282798767, "learning_rate": 0.0001, "elbo": 3.49964, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.942235514895613}
{"step": 84280, "loss": 3.4967140197753905, "grad_norm": 0.757480800151825, "learning_rate": 0.0001, "elbo": 3.49671, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.942528735632184}
{"step": 84285, "loss": 3.486777496337891, "grad_norm": 0.6886850595474243, "learning_rate": 0.0001, "elbo": 3.48678, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.942821956368754}
{"step": 84290, "loss": 3.499216079711914, "grad_norm": 0.7868561148643494, "learning_rate": 0.0001, "elbo": 3.49922, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.943115177105325}
{"step": 84295, "loss": 3.489885711669922, "grad_norm": 0.7612530589103699, "learning_rate": 0.0001, "elbo": 3.48989, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.943408397841895}
{"step": 84300, "loss": 3.486883544921875, "grad_norm": 0.7157261967658997, "learning_rate": 0.0001, "elbo": 3.48688, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.943701618578466}
{"step": 84305, "loss": 3.5051284790039063, "grad_norm": 0.9106029868125916, "learning_rate": 0.0001, "elbo": 3.50513, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.943994839315036}
{"step": 84310, "loss": 3.5005119323730467, "grad_norm": 0.7651035189628601, "learning_rate": 0.0001, "elbo": 3.50051, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.944288060051607}
{"step": 84315, "loss": 3.504121017456055, "grad_norm": 0.6900913715362549, "learning_rate": 0.0001, "elbo": 3.50412, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.944581280788177}
{"step": 84320, "loss": 3.4880199432373047, "grad_norm": 0.7272610068321228, "learning_rate": 0.0001, "elbo": 3.48802, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9448745015247475}
{"step": 84325, "loss": 3.503406524658203, "grad_norm": 1.136365294456482, "learning_rate": 0.0001, "elbo": 3.50341, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.945167722261318}
{"step": 84330, "loss": 3.5005565643310548, "grad_norm": 0.7430132627487183, "learning_rate": 0.0001, "elbo": 3.50056, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.945460942997888}
{"step": 84335, "loss": 3.5064556121826174, "grad_norm": 0.7172634601593018, "learning_rate": 0.0001, "elbo": 3.50646, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.94575416373446}
{"step": 84340, "loss": 3.4881965637207033, "grad_norm": 0.6917935013771057, "learning_rate": 0.0001, "elbo": 3.4882, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.946047384471029}
{"step": 84345, "loss": 3.5007770538330076, "grad_norm": 1.1784179210662842, "learning_rate": 0.0001, "elbo": 3.50078, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.946340605207601}
{"step": 84350, "loss": 3.488362121582031, "grad_norm": 0.6839824318885803, "learning_rate": 0.0001, "elbo": 3.48836, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.946633825944171}
{"step": 84355, "loss": 3.488523483276367, "grad_norm": 0.7868561148643494, "learning_rate": 0.0001, "elbo": 3.48852, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.946927046680742}
{"step": 84360, "loss": 3.4834991455078126, "grad_norm": 0.7271528244018555, "learning_rate": 0.0001, "elbo": 3.4835, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.947220267417312}
{"step": 84365, "loss": 3.48917236328125, "grad_norm": 0.682527482509613, "learning_rate": 0.0001, "elbo": 3.48917, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9475134881538825}
{"step": 84370, "loss": 3.5048389434814453, "grad_norm": 0.9132340550422668, "learning_rate": 0.0001, "elbo": 3.50484, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.947806708890453}
{"step": 84375, "loss": 3.4865371704101564, "grad_norm": 1.2583773136138916, "learning_rate": 0.0001, "elbo": 3.48654, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.948099929627023}
{"step": 84380, "loss": 3.5099231719970705, "grad_norm": 0.716225266456604, "learning_rate": 0.0001, "elbo": 3.50992, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.948393150363594}
{"step": 84385, "loss": 3.4866958618164063, "grad_norm": 0.7581651210784912, "learning_rate": 0.0001, "elbo": 3.4867, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.948686371100164}
{"step": 84390, "loss": 3.5055633544921876, "grad_norm": 0.8934904932975769, "learning_rate": 0.0001, "elbo": 3.50556, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.948979591836735}
{"step": 84395, "loss": 3.5052513122558593, "grad_norm": 0.7038634419441223, "learning_rate": 0.0001, "elbo": 3.50525, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.949272812573305}
{"step": 84400, "loss": 3.4983028411865233, "grad_norm": 0.7172678709030151, "learning_rate": 0.0001, "elbo": 3.4983, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.949566033309876}
{"step": 84405, "loss": 3.5088958740234375, "grad_norm": 0.7176845073699951, "learning_rate": 0.0001, "elbo": 3.5089, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.949859254046446}
{"step": 84410, "loss": 3.5081634521484375, "grad_norm": 0.7173415422439575, "learning_rate": 0.0001, "elbo": 3.50816, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.950152474783017}
{"step": 84415, "loss": 3.483257293701172, "grad_norm": 0.7080876231193542, "learning_rate": 0.0001, "elbo": 3.48326, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.950445695519587}
{"step": 84420, "loss": 3.4934120178222656, "grad_norm": 0.790016770362854, "learning_rate": 0.0001, "elbo": 3.49341, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9507389162561575}
{"step": 84425, "loss": 3.497597503662109, "grad_norm": 0.8020818829536438, "learning_rate": 0.0001, "elbo": 3.4976, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.951032136992728}
{"step": 84430, "loss": 3.486377716064453, "grad_norm": 0.7507262229919434, "learning_rate": 0.0001, "elbo": 3.48638, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.951325357729298}
{"step": 84435, "loss": 3.4899307250976563, "grad_norm": 0.7361263632774353, "learning_rate": 0.0001, "elbo": 3.48993, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.951618578465869}
{"step": 84440, "loss": 3.482147216796875, "grad_norm": 0.9080649614334106, "learning_rate": 0.0001, "elbo": 3.48215, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.951911799202439}
{"step": 84445, "loss": 3.502522277832031, "grad_norm": 0.7067587375640869, "learning_rate": 0.0001, "elbo": 3.50252, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.95220501993901}
{"step": 84450, "loss": 3.4987060546875, "grad_norm": 0.7844610214233398, "learning_rate": 0.0001, "elbo": 3.49871, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.95249824067558}
{"step": 84455, "loss": 3.5063987731933595, "grad_norm": 0.7004805207252502, "learning_rate": 0.0001, "elbo": 3.5064, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.952791461412151}
{"step": 84460, "loss": 3.503917694091797, "grad_norm": 0.68178790807724, "learning_rate": 0.0001, "elbo": 3.50392, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.953084682148722}
{"step": 84465, "loss": 3.493256378173828, "grad_norm": 0.7257590293884277, "learning_rate": 0.0001, "elbo": 3.49326, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.953377902885292}
{"step": 84470, "loss": 3.4936496734619142, "grad_norm": 0.7033894658088684, "learning_rate": 0.0001, "elbo": 3.49365, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.953671123621863}
{"step": 84475, "loss": 3.4879425048828123, "grad_norm": 0.7756401896476746, "learning_rate": 0.0001, "elbo": 3.48794, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.953964344358433}
{"step": 84480, "loss": 3.4950798034667967, "grad_norm": 0.8424174189567566, "learning_rate": 0.0001, "elbo": 3.49508, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.954257565095004}
{"step": 84485, "loss": 3.491883087158203, "grad_norm": 0.683523952960968, "learning_rate": 0.0001, "elbo": 3.49188, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.954550785831574}
{"step": 84490, "loss": 3.4954471588134766, "grad_norm": 0.7266746759414673, "learning_rate": 0.0001, "elbo": 3.49545, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.954844006568145}
{"step": 84495, "loss": 3.4833343505859373, "grad_norm": 0.7650356888771057, "learning_rate": 0.0001, "elbo": 3.48333, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.955137227304715}
{"step": 84500, "loss": 3.4797508239746096, "grad_norm": 0.9166083931922913, "learning_rate": 0.0001, "elbo": 3.47975, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.955430448041286}
{"step": 84505, "loss": 3.4970783233642577, "grad_norm": 0.7057799100875854, "learning_rate": 0.0001, "elbo": 3.49708, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.955723668777856}
{"step": 84510, "loss": 3.5009685516357423, "grad_norm": 0.721642255783081, "learning_rate": 0.0001, "elbo": 3.50097, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9560168895144265}
{"step": 84515, "loss": 3.500733184814453, "grad_norm": 0.7244537472724915, "learning_rate": 0.0001, "elbo": 3.50073, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.956310110250997}
{"step": 84520, "loss": 3.4816864013671873, "grad_norm": 0.6945092678070068, "learning_rate": 0.0001, "elbo": 3.48169, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9566033309875674}
{"step": 84525, "loss": 3.4845252990722657, "grad_norm": 0.757924497127533, "learning_rate": 0.0001, "elbo": 3.48453, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.956896551724138}
{"step": 84525, "loss": 3.4845252990722657, "grad_norm": 0.757924497127533, "learning_rate": 0.0001, "elbo": 3.48453, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.956896551724138}
{"step": 84530, "loss": 3.5105636596679686, "grad_norm": 0.7764188647270203, "learning_rate": 0.0001, "elbo": 3.51056, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.957189772460708}
{"step": 84535, "loss": 3.4865966796875, "grad_norm": 0.720194935798645, "learning_rate": 0.0001, "elbo": 3.4866, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.957482993197279}
{"step": 84540, "loss": 3.5017940521240236, "grad_norm": 1.1988614797592163, "learning_rate": 0.0001, "elbo": 3.50179, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.957776213933849}
{"step": 84545, "loss": 3.5087684631347655, "grad_norm": 0.7401886582374573, "learning_rate": 0.0001, "elbo": 3.50877, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.95806943467042}
{"step": 84550, "loss": 3.5129356384277344, "grad_norm": 0.9758875966072083, "learning_rate": 0.0001, "elbo": 3.51294, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.95836265540699}
{"step": 84555, "loss": 3.513916015625, "grad_norm": 0.7663328647613525, "learning_rate": 0.0001, "elbo": 3.51392, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.958655876143561}
{"step": 84560, "loss": 3.485875701904297, "grad_norm": 0.7534436583518982, "learning_rate": 0.0001, "elbo": 3.48588, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.958949096880131}
{"step": 84565, "loss": 3.4965278625488283, "grad_norm": 0.6888957023620605, "learning_rate": 0.0001, "elbo": 3.49653, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9592423176167015}
{"step": 84565, "loss": 3.4965278625488283, "grad_norm": 0.6888957023620605, "learning_rate": 0.0001, "elbo": 3.49653, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9592423176167015}
{"step": 84570, "loss": 3.4908546447753905, "grad_norm": 1.0607208013534546, "learning_rate": 0.0001, "elbo": 3.49085, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.959535538353272}
{"step": 84575, "loss": 3.493621826171875, "grad_norm": 0.7067367434501648, "learning_rate": 0.0001, "elbo": 3.49362, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.959828759089843}
{"step": 84580, "loss": 3.488346481323242, "grad_norm": 0.7133687138557434, "learning_rate": 0.0001, "elbo": 3.48835, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.960121979826413}
{"step": 84585, "loss": 3.5031482696533205, "grad_norm": 0.730217695236206, "learning_rate": 0.0001, "elbo": 3.50315, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.960415200562984}
{"step": 84585, "loss": 3.5031482696533205, "grad_norm": 0.730217695236206, "learning_rate": 0.0001, "elbo": 3.50315, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.960415200562984}
{"step": 84590, "loss": 3.477425384521484, "grad_norm": 0.7610498666763306, "learning_rate": 0.0001, "elbo": 3.47743, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.960708421299554}
{"step": 84595, "loss": 3.483265686035156, "grad_norm": 0.7052412033081055, "learning_rate": 0.0001, "elbo": 3.48327, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.961001642036125}
{"step": 84600, "loss": 3.512412261962891, "grad_norm": 0.8321250677108765, "learning_rate": 0.0001, "elbo": 3.51241, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.961294862772696}
{"step": 84605, "loss": 3.492864227294922, "grad_norm": 0.7471271753311157, "learning_rate": 0.0001, "elbo": 3.49286, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.961588083509266}
{"step": 84610, "loss": 3.4862117767333984, "grad_norm": 0.7323802709579468, "learning_rate": 0.0001, "elbo": 3.48621, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9618813042458365}
{"step": 84615, "loss": 3.4786003112792967, "grad_norm": 0.7145363688468933, "learning_rate": 0.0001, "elbo": 3.4786, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.962174524982407}
{"step": 84620, "loss": 3.4974365234375, "grad_norm": 0.7037729620933533, "learning_rate": 0.0001, "elbo": 3.49744, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.962467745718977}
{"step": 84625, "loss": 3.4990673065185547, "grad_norm": 0.6946083307266235, "learning_rate": 0.0001, "elbo": 3.49907, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.962760966455548}
{"step": 84630, "loss": 3.501603698730469, "grad_norm": 0.7535758018493652, "learning_rate": 0.0001, "elbo": 3.5016, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.963054187192118}
{"step": 84635, "loss": 3.483829879760742, "grad_norm": 0.9276089072227478, "learning_rate": 0.0001, "elbo": 3.48383, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.963347407928689}
{"step": 84640, "loss": 3.508774185180664, "grad_norm": 0.7169049382209778, "learning_rate": 0.0001, "elbo": 3.50877, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.963640628665259}
{"step": 84645, "loss": 3.487558364868164, "grad_norm": 0.7067000865936279, "learning_rate": 0.0001, "elbo": 3.48756, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.96393384940183}
{"step": 84650, "loss": 3.4913379669189455, "grad_norm": 0.7261976599693298, "learning_rate": 0.0001, "elbo": 3.49134, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9642270701384}
{"step": 84655, "loss": 3.49508056640625, "grad_norm": 0.6803321242332458, "learning_rate": 0.0001, "elbo": 3.49508, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.964520290874971}
{"step": 84660, "loss": 3.498277282714844, "grad_norm": 0.6863381862640381, "learning_rate": 0.0001, "elbo": 3.49828, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.964813511611541}
{"step": 84665, "loss": 3.491493988037109, "grad_norm": 0.7053043246269226, "learning_rate": 0.0001, "elbo": 3.49149, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9651067323481115}
{"step": 84670, "loss": 3.4995540618896483, "grad_norm": 0.85332852602005, "learning_rate": 0.0001, "elbo": 3.49955, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.965399953084682}
{"step": 84675, "loss": 3.5060935974121095, "grad_norm": 0.6876665353775024, "learning_rate": 0.0001, "elbo": 3.50609, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.965693173821252}
{"step": 84680, "loss": 3.4902366638183593, "grad_norm": 1.1529394388198853, "learning_rate": 0.0001, "elbo": 3.49024, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.965986394557823}
{"step": 84685, "loss": 3.492145538330078, "grad_norm": 0.7050274610519409, "learning_rate": 0.0001, "elbo": 3.49215, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.966279615294393}
{"step": 84690, "loss": 3.4998027801513674, "grad_norm": 0.7194541096687317, "learning_rate": 0.0001, "elbo": 3.4998, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.966572836030964}
{"step": 84695, "loss": 3.497359848022461, "grad_norm": 0.7341470718383789, "learning_rate": 0.0001, "elbo": 3.49736, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.966866056767534}
{"step": 84700, "loss": 3.49249382019043, "grad_norm": 0.7537081241607666, "learning_rate": 0.0001, "elbo": 3.49249, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.967159277504106}
{"step": 84705, "loss": 3.480263519287109, "grad_norm": 0.8039843440055847, "learning_rate": 0.0001, "elbo": 3.48026, "mask_frac": 0.4995, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.967452498240675}
{"step": 84710, "loss": 3.500320816040039, "grad_norm": 0.8037208914756775, "learning_rate": 0.0001, "elbo": 3.50032, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9677457189772465}
{"step": 84715, "loss": 3.4983055114746096, "grad_norm": 0.9441390037536621, "learning_rate": 0.0001, "elbo": 3.49831, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.968038939713817}
{"step": 84720, "loss": 3.5039512634277346, "grad_norm": 0.7079611420631409, "learning_rate": 0.0001, "elbo": 3.50395, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.968332160450387}
{"step": 84725, "loss": 3.497791290283203, "grad_norm": 0.7675697207450867, "learning_rate": 0.0001, "elbo": 3.49779, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.968625381186958}
{"step": 84730, "loss": 3.4957038879394533, "grad_norm": 0.6937035918235779, "learning_rate": 0.0001, "elbo": 3.4957, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.968918601923528}
{"step": 84735, "loss": 3.4793350219726564, "grad_norm": 0.693690836429596, "learning_rate": 0.0001, "elbo": 3.47933, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.969211822660099}
{"step": 84740, "loss": 3.50899658203125, "grad_norm": 0.7673118710517883, "learning_rate": 0.0001, "elbo": 3.509, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.969505043396669}
{"step": 84745, "loss": 3.488439178466797, "grad_norm": 0.7244746088981628, "learning_rate": 0.0001, "elbo": 3.48844, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.96979826413324}
{"step": 84750, "loss": 3.5108909606933594, "grad_norm": 0.7119032144546509, "learning_rate": 0.0001, "elbo": 3.51089, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.97009148486981}
{"step": 84755, "loss": 3.489320755004883, "grad_norm": 0.7915988564491272, "learning_rate": 0.0001, "elbo": 3.48932, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.970384705606381}
{"step": 84760, "loss": 3.495848846435547, "grad_norm": 0.7747343182563782, "learning_rate": 0.0001, "elbo": 3.49585, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.970677926342951}
{"step": 84765, "loss": 3.4946990966796876, "grad_norm": 0.7239279747009277, "learning_rate": 0.0001, "elbo": 3.4947, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9709711470795215}
{"step": 84770, "loss": 3.498444366455078, "grad_norm": 0.8771815896034241, "learning_rate": 0.0001, "elbo": 3.49844, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.971264367816092}
{"step": 84775, "loss": 3.496937561035156, "grad_norm": 0.7665401101112366, "learning_rate": 0.0001, "elbo": 3.49694, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.971557588552662}
{"step": 84780, "loss": 3.483441925048828, "grad_norm": 0.7471559047698975, "learning_rate": 0.0001, "elbo": 3.48344, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.971850809289233}
{"step": 84785, "loss": 3.4925785064697266, "grad_norm": 0.7969912886619568, "learning_rate": 0.0001, "elbo": 3.49258, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.972144030025803}
{"step": 84790, "loss": 3.4865169525146484, "grad_norm": 0.8023171424865723, "learning_rate": 0.0001, "elbo": 3.48652, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.972437250762374}
{"step": 84795, "loss": 3.4962226867675783, "grad_norm": 0.7354856729507446, "learning_rate": 0.0001, "elbo": 3.49622, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.972730471498944}
{"step": 84800, "loss": 3.501384735107422, "grad_norm": 0.7617905735969543, "learning_rate": 0.0001, "elbo": 3.50138, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.973023692235515}
{"step": 84805, "loss": 3.5061981201171877, "grad_norm": 0.7360045313835144, "learning_rate": 0.0001, "elbo": 3.5062, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.973316912972085}
{"step": 84810, "loss": 3.479471969604492, "grad_norm": 0.7352116703987122, "learning_rate": 0.0001, "elbo": 3.47947, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.973610133708656}
{"step": 84815, "loss": 3.4972274780273436, "grad_norm": 0.7650007605552673, "learning_rate": 0.0001, "elbo": 3.49723, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.973903354445226}
{"step": 84820, "loss": 3.4916969299316407, "grad_norm": 0.7159690856933594, "learning_rate": 0.0001, "elbo": 3.4917, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9741965751817965}
{"step": 84825, "loss": 3.489554214477539, "grad_norm": 0.6768202185630798, "learning_rate": 0.0001, "elbo": 3.48955, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.974489795918368}
{"step": 84830, "loss": 3.5049110412597657, "grad_norm": 0.731970489025116, "learning_rate": 0.0001, "elbo": 3.50491, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.974783016654937}
{"step": 84835, "loss": 3.496625518798828, "grad_norm": 0.7337032556533813, "learning_rate": 0.0001, "elbo": 3.49663, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.975076237391509}
{"step": 84840, "loss": 3.4883541107177733, "grad_norm": 0.7173529863357544, "learning_rate": 0.0001, "elbo": 3.48835, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.975369458128079}
{"step": 84845, "loss": 3.5050022125244142, "grad_norm": 0.7851679921150208, "learning_rate": 0.0001, "elbo": 3.505, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.97566267886465}
{"step": 84850, "loss": 3.513357162475586, "grad_norm": 0.7457929253578186, "learning_rate": 0.0001, "elbo": 3.51336, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.97595589960122}
{"step": 84855, "loss": 3.4797210693359375, "grad_norm": 0.7042253017425537, "learning_rate": 0.0001, "elbo": 3.47972, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9762491203377905}
{"step": 84860, "loss": 3.4879440307617187, "grad_norm": 0.769505500793457, "learning_rate": 0.0001, "elbo": 3.48794, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.976542341074361}
{"step": 84865, "loss": 3.49429931640625, "grad_norm": 0.7110253572463989, "learning_rate": 0.0001, "elbo": 3.4943, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9768355618109315}
{"step": 84870, "loss": 3.4946704864501954, "grad_norm": 0.7155652642250061, "learning_rate": 0.0001, "elbo": 3.49467, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.977128782547502}
{"step": 84875, "loss": 3.496392822265625, "grad_norm": 0.9536972045898438, "learning_rate": 0.0001, "elbo": 3.49639, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.977422003284072}
{"step": 84880, "loss": 3.5073760986328124, "grad_norm": 0.6726523041725159, "learning_rate": 0.0001, "elbo": 3.50738, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.977715224020643}
{"step": 84885, "loss": 3.487848663330078, "grad_norm": 1.203295111656189, "learning_rate": 0.0001, "elbo": 3.48785, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.978008444757213}
{"step": 84890, "loss": 3.468303680419922, "grad_norm": 0.6978746056556702, "learning_rate": 0.0001, "elbo": 3.4683, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.978301665493784}
{"step": 84895, "loss": 3.4804412841796877, "grad_norm": 0.7518463730812073, "learning_rate": 0.0001, "elbo": 3.48044, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.978594886230354}
{"step": 84900, "loss": 3.5132518768310548, "grad_norm": 0.6897491812705994, "learning_rate": 0.0001, "elbo": 3.51325, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.978888106966925}
{"step": 84905, "loss": 3.4919281005859375, "grad_norm": 0.7051055431365967, "learning_rate": 0.0001, "elbo": 3.49193, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.979181327703495}
{"step": 84910, "loss": 3.4993865966796873, "grad_norm": 0.7385523319244385, "learning_rate": 0.0001, "elbo": 3.49939, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9794745484400655}
{"step": 84915, "loss": 3.499066162109375, "grad_norm": 0.7689725756645203, "learning_rate": 0.0001, "elbo": 3.49907, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.979767769176636}
{"step": 84920, "loss": 3.497029113769531, "grad_norm": 0.7378342747688293, "learning_rate": 0.0001, "elbo": 3.49703, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9800609899132064}
{"step": 84925, "loss": 3.4901771545410156, "grad_norm": 0.75833660364151, "learning_rate": 0.0001, "elbo": 3.49018, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.980354210649777}
{"step": 84930, "loss": 3.4847785949707033, "grad_norm": 0.709050178527832, "learning_rate": 0.0001, "elbo": 3.48478, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.980647431386347}
{"step": 84935, "loss": 3.483509063720703, "grad_norm": 0.7177845239639282, "learning_rate": 0.0001, "elbo": 3.48351, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.980940652122918}
{"step": 84940, "loss": 3.51025390625, "grad_norm": 0.7011963129043579, "learning_rate": 0.0001, "elbo": 3.51025, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.981233872859488}
{"step": 84945, "loss": 3.5006298065185546, "grad_norm": 0.7532287240028381, "learning_rate": 0.0001, "elbo": 3.50063, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.981527093596059}
{"step": 84950, "loss": 3.480389404296875, "grad_norm": 0.759091317653656, "learning_rate": 0.0001, "elbo": 3.48039, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.98182031433263}
{"step": 84955, "loss": 3.50552978515625, "grad_norm": 1.337170958518982, "learning_rate": 0.0001, "elbo": 3.50553, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9821135350692}
{"step": 84960, "loss": 3.491172027587891, "grad_norm": 0.782489001750946, "learning_rate": 0.0001, "elbo": 3.49117, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.982406755805771}
{"step": 84965, "loss": 3.492174530029297, "grad_norm": 0.7373577952384949, "learning_rate": 0.0001, "elbo": 3.49217, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.982699976542341}
{"step": 84970, "loss": 3.489118194580078, "grad_norm": 0.7027736902236938, "learning_rate": 0.0001, "elbo": 3.48912, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.982993197278912}
{"step": 84975, "loss": 3.50682373046875, "grad_norm": 0.8256726861000061, "learning_rate": 0.0001, "elbo": 3.50682, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.983286418015482}
{"step": 84980, "loss": 3.5006046295166016, "grad_norm": 0.7868703007698059, "learning_rate": 0.0001, "elbo": 3.5006, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.983579638752053}
{"step": 84985, "loss": 3.495588684082031, "grad_norm": 0.8152238130569458, "learning_rate": 0.0001, "elbo": 3.49559, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.983872859488623}
{"step": 84985, "loss": 3.495588684082031, "grad_norm": 0.8152238130569458, "learning_rate": 0.0001, "elbo": 3.49559, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.983872859488623}
{"step": 84990, "loss": 3.491838836669922, "grad_norm": 0.7236382365226746, "learning_rate": 0.0001, "elbo": 3.49184, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.984166080225194}
{"step": 84995, "loss": 3.4850341796875, "grad_norm": 0.7158610820770264, "learning_rate": 0.0001, "elbo": 3.48503, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.984459300961764}
{"step": 85000, "loss": 3.4855815887451174, "grad_norm": 0.7596676349639893, "learning_rate": 0.0001, "elbo": 3.48558, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.984752521698335}
{"step": 85005, "loss": 3.492877960205078, "grad_norm": 0.8909169435501099, "learning_rate": 0.0001, "elbo": 3.49288, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.985045742434905}
{"step": 85010, "loss": 3.4791725158691404, "grad_norm": 0.6710993647575378, "learning_rate": 0.0001, "elbo": 3.47917, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9853389631714755}
{"step": 85015, "loss": 3.5083953857421877, "grad_norm": 0.88726407289505, "learning_rate": 0.0001, "elbo": 3.5084, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.985632183908046}
{"step": 85020, "loss": 3.5162525177001953, "grad_norm": 0.7146100401878357, "learning_rate": 0.0001, "elbo": 3.51625, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.985925404644616}
{"step": 85025, "loss": 3.495679473876953, "grad_norm": 0.694575846195221, "learning_rate": 0.0001, "elbo": 3.49568, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.986218625381187}
{"step": 85030, "loss": 3.492628479003906, "grad_norm": 0.9719349145889282, "learning_rate": 0.0001, "elbo": 3.49263, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.986511846117757}
{"step": 85035, "loss": 3.5131610870361327, "grad_norm": 0.731257975101471, "learning_rate": 0.0001, "elbo": 3.51316, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.986805066854328}
{"step": 85040, "loss": 3.487101364135742, "grad_norm": 0.6748554110527039, "learning_rate": 0.0001, "elbo": 3.4871, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.987098287590898}
{"step": 85045, "loss": 3.488628387451172, "grad_norm": 0.697977602481842, "learning_rate": 0.0001, "elbo": 3.48863, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.987391508327469}
{"step": 85050, "loss": 3.4751205444335938, "grad_norm": 0.7053354382514954, "learning_rate": 0.0001, "elbo": 3.47512, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.987684729064039}
{"step": 85055, "loss": 3.4908172607421877, "grad_norm": 0.7024611830711365, "learning_rate": 0.0001, "elbo": 3.49082, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.98797794980061}
{"step": 85060, "loss": 3.4908424377441407, "grad_norm": 0.6903920769691467, "learning_rate": 0.0001, "elbo": 3.49084, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.98827117053718}
{"step": 85065, "loss": 3.500995635986328, "grad_norm": 0.7074596285820007, "learning_rate": 0.0001, "elbo": 3.501, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9885643912737505}
{"step": 85070, "loss": 3.4910465240478517, "grad_norm": 0.7226463556289673, "learning_rate": 0.0001, "elbo": 3.49105, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.988857612010321}
{"step": 85075, "loss": 3.4932235717773437, "grad_norm": 0.6755537390708923, "learning_rate": 0.0001, "elbo": 3.49322, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.989150832746892}
{"step": 85080, "loss": 3.4818511962890626, "grad_norm": 0.7289196252822876, "learning_rate": 0.0001, "elbo": 3.48185, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.989444053483462}
{"step": 85085, "loss": 3.481291961669922, "grad_norm": 0.7192835211753845, "learning_rate": 0.0001, "elbo": 3.48129, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.989737274220033}
{"step": 85090, "loss": 3.4907501220703123, "grad_norm": 0.7655505537986755, "learning_rate": 0.0001, "elbo": 3.49075, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.990030494956604}
{"step": 85095, "loss": 3.482369232177734, "grad_norm": 0.8441669940948486, "learning_rate": 0.0001, "elbo": 3.48237, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.990323715693174}
{"step": 85100, "loss": 3.488576889038086, "grad_norm": 0.6798157691955566, "learning_rate": 0.0001, "elbo": 3.48858, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.990616936429745}
{"step": 85105, "loss": 3.4903621673583984, "grad_norm": 0.689929187297821, "learning_rate": 0.0001, "elbo": 3.49036, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.990910157166315}
{"step": 85110, "loss": 3.5109489440917967, "grad_norm": 0.721196174621582, "learning_rate": 0.0001, "elbo": 3.51095, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9912033779028855}
{"step": 85110, "loss": 3.5109489440917967, "grad_norm": 0.721196174621582, "learning_rate": 0.0001, "elbo": 3.51095, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9912033779028855}
{"step": 85115, "loss": 3.5071380615234373, "grad_norm": 0.758506715297699, "learning_rate": 0.0001, "elbo": 3.50714, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.991496598639456}
{"step": 85120, "loss": 3.495467758178711, "grad_norm": 0.9290997982025146, "learning_rate": 0.0001, "elbo": 3.49547, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.991789819376026}
{"step": 85125, "loss": 3.511119079589844, "grad_norm": 0.7212998867034912, "learning_rate": 0.0001, "elbo": 3.51112, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.992083040112597}
{"step": 85130, "loss": 3.5002540588378905, "grad_norm": 0.752830445766449, "learning_rate": 0.0001, "elbo": 3.50025, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.992376260849167}
{"step": 85135, "loss": 3.495697784423828, "grad_norm": 0.805228590965271, "learning_rate": 0.0001, "elbo": 3.4957, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.992669481585738}
{"step": 85140, "loss": 3.4988166809082033, "grad_norm": 0.7603645324707031, "learning_rate": 0.0001, "elbo": 3.49882, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.992962702322308}
{"step": 85145, "loss": 3.4962142944335937, "grad_norm": 0.9481931924819946, "learning_rate": 0.0001, "elbo": 3.49621, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.993255923058879}
{"step": 85150, "loss": 3.490937042236328, "grad_norm": 1.4393482208251953, "learning_rate": 0.0001, "elbo": 3.49094, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.993549143795449}
{"step": 85155, "loss": 3.500782012939453, "grad_norm": 0.7404546737670898, "learning_rate": 0.0001, "elbo": 3.50078, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.99384236453202}
{"step": 85160, "loss": 3.486851119995117, "grad_norm": 0.8778678774833679, "learning_rate": 0.0001, "elbo": 3.48685, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.99413558526859}
{"step": 85165, "loss": 3.500397491455078, "grad_norm": 0.7424072027206421, "learning_rate": 0.0001, "elbo": 3.5004, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9944288060051605}
{"step": 85170, "loss": 3.488330078125, "grad_norm": 0.8548754453659058, "learning_rate": 0.0001, "elbo": 3.48833, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.994722026741731}
{"step": 85175, "loss": 3.4806587219238283, "grad_norm": 0.7380369305610657, "learning_rate": 0.0001, "elbo": 3.48066, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.995015247478301}
{"step": 85180, "loss": 3.4903907775878906, "grad_norm": 0.7026453018188477, "learning_rate": 0.0001, "elbo": 3.49039, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.995308468214872}
{"step": 85185, "loss": 3.4870071411132812, "grad_norm": 0.715212881565094, "learning_rate": 0.0001, "elbo": 3.48701, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.995601688951442}
{"step": 85190, "loss": 3.4826385498046877, "grad_norm": 0.7172300815582275, "learning_rate": 0.0001, "elbo": 3.48264, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.995894909688013}
{"step": 85195, "loss": 3.506632614135742, "grad_norm": 0.7455288767814636, "learning_rate": 0.0001, "elbo": 3.50663, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.996188130424583}
{"step": 85200, "loss": 3.4999134063720705, "grad_norm": 0.8393425941467285, "learning_rate": 0.0001, "elbo": 3.49991, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9964813511611545}
{"step": 85205, "loss": 3.486764144897461, "grad_norm": 0.7411839365959167, "learning_rate": 0.0001, "elbo": 3.48676, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.996774571897724}
{"step": 85210, "loss": 3.496540069580078, "grad_norm": 0.7131595611572266, "learning_rate": 0.0001, "elbo": 3.49654, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9970677926342955}
{"step": 85215, "loss": 3.507435607910156, "grad_norm": 0.7482709884643555, "learning_rate": 0.0001, "elbo": 3.50744, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.997361013370866}
{"step": 85220, "loss": 3.5081100463867188, "grad_norm": 0.7026323676109314, "learning_rate": 0.0001, "elbo": 3.50811, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.997654234107436}
{"step": 85225, "loss": 3.4958877563476562, "grad_norm": 0.7577832937240601, "learning_rate": 0.0001, "elbo": 3.49589, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.997947454844007}
{"step": 85230, "loss": 3.5088497161865235, "grad_norm": 0.6717084646224976, "learning_rate": 0.0001, "elbo": 3.50885, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.998240675580577}
{"step": 85235, "loss": 3.4897727966308594, "grad_norm": 0.6968570351600647, "learning_rate": 0.0001, "elbo": 3.48977, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.998533896317148}
{"step": 85240, "loss": 3.501057434082031, "grad_norm": 0.7238262295722961, "learning_rate": 0.0001, "elbo": 3.50106, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.998827117053718}
{"step": 85245, "loss": 3.4964889526367187, "grad_norm": 0.7058759331703186, "learning_rate": 0.0001, "elbo": 3.49649, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.999120337790289}
{"step": 85250, "loss": 3.5030853271484377, "grad_norm": 0.7366542816162109, "learning_rate": 0.0001, "elbo": 3.50308, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.999413558526859}
{"step": 85255, "loss": 3.506400299072266, "grad_norm": 0.7099444270133972, "learning_rate": 0.0001, "elbo": 3.5064, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 4.9997067792634295}
{"step": 85260, "loss": 3.4986778259277345, "grad_norm": 0.7123327255249023, "learning_rate": 0.0001, "elbo": 3.49868, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0}
{"step": 85265, "loss": 3.496165466308594, "grad_norm": 0.7440895438194275, "learning_rate": 0.0001, "elbo": 3.49617, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0002932207365705}
{"step": 85270, "loss": 3.4993061065673827, "grad_norm": 0.7706648111343384, "learning_rate": 0.0001, "elbo": 3.49931, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.000586441473141}
{"step": 85275, "loss": 3.487689208984375, "grad_norm": 0.8039621114730835, "learning_rate": 0.0001, "elbo": 3.48769, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.000879662209711}
{"step": 85280, "loss": 3.5072582244873045, "grad_norm": 0.8599281907081604, "learning_rate": 0.0001, "elbo": 3.50726, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.001172882946282}
{"step": 85285, "loss": 3.487715148925781, "grad_norm": 0.6830798387527466, "learning_rate": 0.0001, "elbo": 3.48772, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.001466103682852}
{"step": 85290, "loss": 3.482117462158203, "grad_norm": 0.7029753923416138, "learning_rate": 0.0001, "elbo": 3.48212, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.001759324419423}
{"step": 85295, "loss": 3.497159957885742, "grad_norm": 0.7636390328407288, "learning_rate": 0.0001, "elbo": 3.49716, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.002052545155993}
{"step": 85300, "loss": 3.4923728942871093, "grad_norm": 0.8082646727561951, "learning_rate": 0.0001, "elbo": 3.49237, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.002345765892564}
{"step": 85305, "loss": 3.470441436767578, "grad_norm": 0.7356592416763306, "learning_rate": 0.0001, "elbo": 3.47044, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.002638986629134}
{"step": 85310, "loss": 3.484730529785156, "grad_norm": 0.6988801956176758, "learning_rate": 0.0001, "elbo": 3.48473, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0029322073657045}
{"step": 85315, "loss": 3.4945083618164063, "grad_norm": 0.9513856172561646, "learning_rate": 0.0001, "elbo": 3.49451, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.003225428102275}
{"step": 85320, "loss": 3.502180480957031, "grad_norm": 0.737683892250061, "learning_rate": 0.0001, "elbo": 3.50218, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0035186488388455}
{"step": 85325, "loss": 3.5035247802734375, "grad_norm": 0.7186552882194519, "learning_rate": 0.0001, "elbo": 3.50352, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.003811869575417}
{"step": 85330, "loss": 3.4994480133056642, "grad_norm": 0.7317719459533691, "learning_rate": 0.0001, "elbo": 3.49945, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.004105090311987}
{"step": 85335, "loss": 3.480143737792969, "grad_norm": 0.7282878756523132, "learning_rate": 0.0001, "elbo": 3.48014, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.004398311048558}
{"step": 85340, "loss": 3.4984199523925783, "grad_norm": 0.7090898156166077, "learning_rate": 0.0001, "elbo": 3.49842, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.004691531785128}
{"step": 85345, "loss": 3.4802261352539063, "grad_norm": 0.7064197659492493, "learning_rate": 0.0001, "elbo": 3.48023, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.004984752521699}
{"step": 85350, "loss": 3.501072311401367, "grad_norm": 0.6844185590744019, "learning_rate": 0.0001, "elbo": 3.50107, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.005277973258269}
{"step": 85355, "loss": 3.5018390655517577, "grad_norm": 0.7703579068183899, "learning_rate": 0.0001, "elbo": 3.50184, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0055711939948395}
{"step": 85360, "loss": 3.495774841308594, "grad_norm": 0.7198833227157593, "learning_rate": 0.0001, "elbo": 3.49577, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.00586441473141}
{"step": 85365, "loss": 3.498762512207031, "grad_norm": 0.7170880436897278, "learning_rate": 0.0001, "elbo": 3.49876, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.00615763546798}
{"step": 85370, "loss": 3.4852508544921874, "grad_norm": 0.677372932434082, "learning_rate": 0.0001, "elbo": 3.48525, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.006450856204551}
{"step": 85375, "loss": 3.5099746704101564, "grad_norm": 0.7462885975837708, "learning_rate": 0.0001, "elbo": 3.50997, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.006744076941121}
{"step": 85380, "loss": 3.477519226074219, "grad_norm": 0.7528747320175171, "learning_rate": 0.0001, "elbo": 3.47752, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.007037297677692}
{"step": 85385, "loss": 3.5141273498535157, "grad_norm": 0.7312489151954651, "learning_rate": 0.0001, "elbo": 3.51413, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.007330518414262}
{"step": 85390, "loss": 3.500722885131836, "grad_norm": 0.6828374862670898, "learning_rate": 0.0001, "elbo": 3.50072, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.007623739150833}
{"step": 85395, "loss": 3.490458679199219, "grad_norm": 0.7038590908050537, "learning_rate": 0.0001, "elbo": 3.49046, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.007916959887403}
{"step": 85400, "loss": 3.4891197204589846, "grad_norm": 0.7677688002586365, "learning_rate": 0.0001, "elbo": 3.48912, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.008210180623974}
{"step": 85405, "loss": 3.4971435546875, "grad_norm": 0.8217777013778687, "learning_rate": 0.0001, "elbo": 3.49714, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.008503401360544}
{"step": 85410, "loss": 3.493550109863281, "grad_norm": 0.7367779612541199, "learning_rate": 0.0001, "elbo": 3.49355, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0087966220971145}
{"step": 85415, "loss": 3.488336944580078, "grad_norm": 0.7191951274871826, "learning_rate": 0.0001, "elbo": 3.48834, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.009089842833685}
{"step": 85420, "loss": 3.4960609436035157, "grad_norm": 0.8543798327445984, "learning_rate": 0.0001, "elbo": 3.49606, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.009383063570255}
{"step": 85425, "loss": 3.483713150024414, "grad_norm": 0.7586880922317505, "learning_rate": 0.0001, "elbo": 3.48371, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.009676284306826}
{"step": 85430, "loss": 3.4889244079589843, "grad_norm": 0.7020854949951172, "learning_rate": 0.0001, "elbo": 3.48892, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.009969505043396}
{"step": 85435, "loss": 3.4805038452148436, "grad_norm": 0.8672019839286804, "learning_rate": 0.0001, "elbo": 3.4805, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.010262725779967}
{"step": 85440, "loss": 3.4904273986816405, "grad_norm": 0.7135500311851501, "learning_rate": 0.0001, "elbo": 3.49043, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.010555946516537}
{"step": 85445, "loss": 3.4883163452148436, "grad_norm": 0.7606443762779236, "learning_rate": 0.0001, "elbo": 3.48832, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.010849167253109}
{"step": 85450, "loss": 3.4982273101806642, "grad_norm": 0.7719488143920898, "learning_rate": 0.0001, "elbo": 3.49823, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.011142387989679}
{"step": 85450, "loss": 3.4982273101806642, "grad_norm": 0.7719488143920898, "learning_rate": 0.0001, "elbo": 3.49823, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.011142387989679}
{"step": 85455, "loss": 3.496839904785156, "grad_norm": 0.7068092823028564, "learning_rate": 0.0001, "elbo": 3.49684, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0114356087262495}
{"step": 85460, "loss": 3.50035400390625, "grad_norm": 0.7041181325912476, "learning_rate": 0.0001, "elbo": 3.50035, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.01172882946282}
{"step": 85465, "loss": 3.480803298950195, "grad_norm": 0.7092248797416687, "learning_rate": 0.0001, "elbo": 3.4808, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.01202205019939}
{"step": 85470, "loss": 3.4952537536621096, "grad_norm": 0.7156093716621399, "learning_rate": 0.0001, "elbo": 3.49525, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.012315270935961}
{"step": 85475, "loss": 3.490522003173828, "grad_norm": 0.7108166217803955, "learning_rate": 0.0001, "elbo": 3.49052, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.012608491672531}
{"step": 85480, "loss": 3.5017715454101563, "grad_norm": 0.688278317451477, "learning_rate": 0.0001, "elbo": 3.50177, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.012901712409102}
{"step": 85485, "loss": 3.489679718017578, "grad_norm": 0.7239162921905518, "learning_rate": 0.0001, "elbo": 3.48968, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.013194933145672}
{"step": 85490, "loss": 3.494795227050781, "grad_norm": 0.7110889554023743, "learning_rate": 0.0001, "elbo": 3.4948, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.013488153882243}
{"step": 85490, "loss": 3.494795227050781, "grad_norm": 0.7110889554023743, "learning_rate": 0.0001, "elbo": 3.4948, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.013488153882243}
{"step": 85495, "loss": 3.5022514343261717, "grad_norm": 0.8092741370201111, "learning_rate": 0.0001, "elbo": 3.50225, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.013781374618813}
{"step": 85500, "loss": 3.4997913360595705, "grad_norm": 0.7088068723678589, "learning_rate": 0.0001, "elbo": 3.49979, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.014074595355384}
{"step": 85505, "loss": 3.4768199920654297, "grad_norm": 0.7325724959373474, "learning_rate": 0.0001, "elbo": 3.47682, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.014367816091954}
{"step": 85510, "loss": 3.509463882446289, "grad_norm": 0.7199116945266724, "learning_rate": 0.0001, "elbo": 3.50946, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0146610368285245}
{"step": 85515, "loss": 3.4969547271728514, "grad_norm": 0.795682966709137, "learning_rate": 0.0001, "elbo": 3.49695, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.014954257565095}
{"step": 85520, "loss": 3.48968505859375, "grad_norm": 0.6896303296089172, "learning_rate": 0.0001, "elbo": 3.48969, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.015247478301665}
{"step": 85525, "loss": 3.4832832336425783, "grad_norm": 0.7389583587646484, "learning_rate": 0.0001, "elbo": 3.48328, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.015540699038236}
{"step": 85525, "loss": 3.4832832336425783, "grad_norm": 0.7389583587646484, "learning_rate": 0.0001, "elbo": 3.48328, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.015540699038236}
{"step": 85530, "loss": 3.484976959228516, "grad_norm": 0.8191300630569458, "learning_rate": 0.0001, "elbo": 3.48498, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.015833919774806}
{"step": 85535, "loss": 3.4967918395996094, "grad_norm": 0.7631197571754456, "learning_rate": 0.0001, "elbo": 3.49679, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.016127140511377}
{"step": 85540, "loss": 3.503042221069336, "grad_norm": 0.7776023149490356, "learning_rate": 0.0001, "elbo": 3.50304, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.016420361247947}
{"step": 85545, "loss": 3.487294006347656, "grad_norm": 0.7255433797836304, "learning_rate": 0.0001, "elbo": 3.48729, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.016713581984518}
{"step": 85550, "loss": 3.491918182373047, "grad_norm": 0.7820001244544983, "learning_rate": 0.0001, "elbo": 3.49192, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.017006802721088}
{"step": 85555, "loss": 3.5027587890625, "grad_norm": 0.7485724091529846, "learning_rate": 0.0001, "elbo": 3.50276, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.017300023457659}
{"step": 85560, "loss": 3.5042045593261717, "grad_norm": 0.7657606601715088, "learning_rate": 0.0001, "elbo": 3.5042, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.017593244194229}
{"step": 85565, "loss": 3.4916408538818358, "grad_norm": 0.8457076549530029, "learning_rate": 0.0001, "elbo": 3.49164, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0178864649307995}
{"step": 85570, "loss": 3.5058868408203123, "grad_norm": 0.7405111789703369, "learning_rate": 0.0001, "elbo": 3.50589, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.018179685667371}
{"step": 85575, "loss": 3.4913631439208985, "grad_norm": 0.7141016125679016, "learning_rate": 0.0001, "elbo": 3.49136, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.018472906403941}
{"step": 85580, "loss": 3.511210250854492, "grad_norm": 0.72526615858078, "learning_rate": 0.0001, "elbo": 3.51121, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.018766127140512}
{"step": 85585, "loss": 3.511005401611328, "grad_norm": 0.7348822951316833, "learning_rate": 0.0001, "elbo": 3.51101, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.019059347877082}
{"step": 85590, "loss": 3.5081813812255858, "grad_norm": 0.85504549741745, "learning_rate": 0.0001, "elbo": 3.50818, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.019352568613653}
{"step": 85595, "loss": 3.4995471954345705, "grad_norm": 0.727153480052948, "learning_rate": 0.0001, "elbo": 3.49955, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.019645789350223}
{"step": 85600, "loss": 3.4945022583007814, "grad_norm": 0.7767379283905029, "learning_rate": 0.0001, "elbo": 3.4945, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0199390100867936}
{"step": 85600, "loss": 3.4945022583007814, "grad_norm": 0.7767379283905029, "learning_rate": 0.0001, "elbo": 3.4945, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0199390100867936}
{"step": 85605, "loss": 3.4862991333007813, "grad_norm": 0.9373137950897217, "learning_rate": 0.0001, "elbo": 3.4863, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.020232230823364}
{"step": 85610, "loss": 3.4867000579833984, "grad_norm": 0.7691923975944519, "learning_rate": 0.0001, "elbo": 3.4867, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0205254515599345}
{"step": 85615, "loss": 3.5045402526855467, "grad_norm": 0.7141373157501221, "learning_rate": 0.0001, "elbo": 3.50454, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.020818672296505}
{"step": 85620, "loss": 3.499297332763672, "grad_norm": 0.7314848899841309, "learning_rate": 0.0001, "elbo": 3.4993, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.021111893033075}
{"step": 85625, "loss": 3.4622894287109376, "grad_norm": 0.7362356185913086, "learning_rate": 0.0001, "elbo": 3.46229, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.021405113769646}
{"step": 85630, "loss": 3.4824012756347655, "grad_norm": 0.9115835428237915, "learning_rate": 0.0001, "elbo": 3.4824, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.021698334506216}
{"step": 85635, "loss": 3.4958026885986326, "grad_norm": 0.7642645239830017, "learning_rate": 0.0001, "elbo": 3.4958, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.021991555242787}
{"step": 85640, "loss": 3.4935455322265625, "grad_norm": 0.696487545967102, "learning_rate": 0.0001, "elbo": 3.49355, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.022284775979357}
{"step": 85645, "loss": 3.490998077392578, "grad_norm": 0.834984540939331, "learning_rate": 0.0001, "elbo": 3.491, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.022577996715928}
{"step": 85650, "loss": 3.4944076538085938, "grad_norm": 0.7103579044342041, "learning_rate": 0.0001, "elbo": 3.49441, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.022871217452498}
{"step": 85655, "loss": 3.4890869140625, "grad_norm": 0.6887298822402954, "learning_rate": 0.0001, "elbo": 3.48909, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0231644381890685}
{"step": 85660, "loss": 3.486520004272461, "grad_norm": 0.6825159192085266, "learning_rate": 0.0001, "elbo": 3.48652, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.023457658925639}
{"step": 85665, "loss": 3.495611572265625, "grad_norm": 0.7129165530204773, "learning_rate": 0.0001, "elbo": 3.49561, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0237508796622095}
{"step": 85670, "loss": 3.4959774017333984, "grad_norm": 0.8521513938903809, "learning_rate": 0.0001, "elbo": 3.49598, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.02404410039878}
{"step": 85675, "loss": 3.484754943847656, "grad_norm": 0.7730120420455933, "learning_rate": 0.0001, "elbo": 3.48476, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.02433732113535}
{"step": 85680, "loss": 3.507642364501953, "grad_norm": 0.6994727849960327, "learning_rate": 0.0001, "elbo": 3.50764, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.024630541871921}
{"step": 85685, "loss": 3.487040328979492, "grad_norm": 0.724234402179718, "learning_rate": 0.0001, "elbo": 3.48704, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.024923762608491}
{"step": 85690, "loss": 3.513315963745117, "grad_norm": 0.7638520002365112, "learning_rate": 0.0001, "elbo": 3.51332, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.025216983345062}
{"step": 85695, "loss": 3.486127471923828, "grad_norm": 0.71805340051651, "learning_rate": 0.0001, "elbo": 3.48613, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.025510204081633}
{"step": 85700, "loss": 3.4940654754638674, "grad_norm": 0.7387213110923767, "learning_rate": 0.0001, "elbo": 3.49407, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0258034248182035}
{"step": 85705, "loss": 3.4895095825195312, "grad_norm": 0.7230295538902283, "learning_rate": 0.0001, "elbo": 3.48951, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.026096645554774}
{"step": 85710, "loss": 3.498120880126953, "grad_norm": 0.7291050553321838, "learning_rate": 0.0001, "elbo": 3.49812, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.026389866291344}
{"step": 85715, "loss": 3.4930572509765625, "grad_norm": 0.690266489982605, "learning_rate": 0.0001, "elbo": 3.49306, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.026683087027915}
{"step": 85720, "loss": 3.487993621826172, "grad_norm": 0.6845406293869019, "learning_rate": 0.0001, "elbo": 3.48799, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.026976307764485}
{"step": 85725, "loss": 3.50240478515625, "grad_norm": 0.7079451084136963, "learning_rate": 0.0001, "elbo": 3.5024, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.027269528501056}
{"step": 85730, "loss": 3.498235321044922, "grad_norm": 0.7990830540657043, "learning_rate": 0.0001, "elbo": 3.49824, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.027562749237626}
{"step": 85735, "loss": 3.496303939819336, "grad_norm": 0.7711077332496643, "learning_rate": 0.0001, "elbo": 3.4963, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.027855969974197}
{"step": 85740, "loss": 3.4922256469726562, "grad_norm": 0.7077798247337341, "learning_rate": 0.0001, "elbo": 3.49223, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.028149190710767}
{"step": 85745, "loss": 3.5045612335205076, "grad_norm": 0.7130663394927979, "learning_rate": 0.0001, "elbo": 3.50456, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.028442411447338}
{"step": 85750, "loss": 3.4846099853515624, "grad_norm": 0.9970715045928955, "learning_rate": 0.0001, "elbo": 3.48461, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.028735632183908}
{"step": 85755, "loss": 3.50006103515625, "grad_norm": 0.7246869802474976, "learning_rate": 0.0001, "elbo": 3.50006, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0290288529204785}
{"step": 85760, "loss": 3.501139831542969, "grad_norm": 0.7493387460708618, "learning_rate": 0.0001, "elbo": 3.50114, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.029322073657049}
{"step": 85765, "loss": 3.5077354431152346, "grad_norm": 0.8143362998962402, "learning_rate": 0.0001, "elbo": 3.50774, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.029615294393619}
{"step": 85770, "loss": 3.500849151611328, "grad_norm": 0.7502116560935974, "learning_rate": 0.0001, "elbo": 3.50085, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.02990851513019}
{"step": 85775, "loss": 3.5146553039550783, "grad_norm": 0.7129412293434143, "learning_rate": 0.0001, "elbo": 3.51466, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.03020173586676}
{"step": 85780, "loss": 3.513380432128906, "grad_norm": 1.3212045431137085, "learning_rate": 0.0001, "elbo": 3.51338, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.030494956603331}
{"step": 85785, "loss": 3.4872802734375, "grad_norm": 0.7291779518127441, "learning_rate": 0.0001, "elbo": 3.48728, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.030788177339901}
{"step": 85785, "loss": 3.4872802734375, "grad_norm": 0.7291779518127441, "learning_rate": 0.0001, "elbo": 3.48728, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.030788177339901}
{"step": 85790, "loss": 3.4967792510986326, "grad_norm": 0.7078720331192017, "learning_rate": 0.0001, "elbo": 3.49678, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.031081398076472}
{"step": 85795, "loss": 3.4802520751953123, "grad_norm": 0.752583920955658, "learning_rate": 0.0001, "elbo": 3.48025, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.031374618813042}
{"step": 85800, "loss": 3.4867538452148437, "grad_norm": 0.7145740985870361, "learning_rate": 0.0001, "elbo": 3.48675, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.031667839549613}
{"step": 85805, "loss": 3.511457061767578, "grad_norm": 0.7188911437988281, "learning_rate": 0.0001, "elbo": 3.51146, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.031961060286183}
{"step": 85810, "loss": 3.5075363159179687, "grad_norm": 0.7469533681869507, "learning_rate": 0.0001, "elbo": 3.50754, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0322542810227535}
{"step": 85815, "loss": 3.486470413208008, "grad_norm": 1.2317472696304321, "learning_rate": 0.0001, "elbo": 3.48647, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.032547501759325}
{"step": 85820, "loss": 3.4876663208007814, "grad_norm": 0.7078695893287659, "learning_rate": 0.0001, "elbo": 3.48767, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.032840722495895}
{"step": 85825, "loss": 3.491547775268555, "grad_norm": 0.6851500272750854, "learning_rate": 0.0001, "elbo": 3.49155, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.033133943232466}
{"step": 85830, "loss": 3.4881343841552734, "grad_norm": 0.8212084174156189, "learning_rate": 0.0001, "elbo": 3.48813, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.033427163969036}
{"step": 85835, "loss": 3.4815650939941407, "grad_norm": 0.6926077008247375, "learning_rate": 0.0001, "elbo": 3.48156, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.033720384705607}
{"step": 85840, "loss": 3.4848251342773438, "grad_norm": 0.7738738059997559, "learning_rate": 0.0001, "elbo": 3.48482, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.034013605442177}
{"step": 85845, "loss": 3.4881412506103517, "grad_norm": 0.7323895692825317, "learning_rate": 0.0001, "elbo": 3.48814, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.034306826178748}
{"step": 85850, "loss": 3.4985366821289063, "grad_norm": 0.7163553833961487, "learning_rate": 0.0001, "elbo": 3.49854, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.034600046915318}
{"step": 85855, "loss": 3.4852561950683594, "grad_norm": 0.7373071312904358, "learning_rate": 0.0001, "elbo": 3.48526, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0348932676518885}
{"step": 85860, "loss": 3.490946960449219, "grad_norm": 0.7427639961242676, "learning_rate": 0.0001, "elbo": 3.49095, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.035186488388459}
{"step": 85865, "loss": 3.4982608795166015, "grad_norm": 0.7161475419998169, "learning_rate": 0.0001, "elbo": 3.49826, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.035479709125029}
{"step": 85870, "loss": 3.470641326904297, "grad_norm": 1.2262521982192993, "learning_rate": 0.0001, "elbo": 3.47064, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0357729298616}
{"step": 85875, "loss": 3.4863327026367186, "grad_norm": 0.776430606842041, "learning_rate": 0.0001, "elbo": 3.48633, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.03606615059817}
{"step": 85880, "loss": 3.4939285278320313, "grad_norm": 0.7165768146514893, "learning_rate": 0.0001, "elbo": 3.49393, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.036359371334741}
{"step": 85885, "loss": 3.484958267211914, "grad_norm": 0.8368454575538635, "learning_rate": 0.0001, "elbo": 3.48496, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.036652592071311}
{"step": 85890, "loss": 3.5172348022460938, "grad_norm": 0.6851004958152771, "learning_rate": 0.0001, "elbo": 3.51723, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.036945812807882}
{"step": 85895, "loss": 3.5069698333740233, "grad_norm": 1.124287724494934, "learning_rate": 0.0001, "elbo": 3.50697, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.037239033544452}
{"step": 85900, "loss": 3.4868186950683593, "grad_norm": 0.8257710933685303, "learning_rate": 0.0001, "elbo": 3.48682, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.037532254281023}
{"step": 85905, "loss": 3.475524139404297, "grad_norm": 0.731660008430481, "learning_rate": 0.0001, "elbo": 3.47552, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.037825475017593}
{"step": 85910, "loss": 3.5120620727539062, "grad_norm": 0.797292947769165, "learning_rate": 0.0001, "elbo": 3.51206, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0381186957541635}
{"step": 85915, "loss": 3.4895843505859374, "grad_norm": 0.7540197968482971, "learning_rate": 0.0001, "elbo": 3.48958, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.038411916490734}
{"step": 85920, "loss": 3.4919807434082033, "grad_norm": 0.7796333432197571, "learning_rate": 0.0001, "elbo": 3.49198, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.038705137227304}
{"step": 85925, "loss": 3.484752655029297, "grad_norm": 0.7169269323348999, "learning_rate": 0.0001, "elbo": 3.48475, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.038998357963875}
{"step": 85930, "loss": 3.5030712127685546, "grad_norm": 0.8821631669998169, "learning_rate": 0.0001, "elbo": 3.50307, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.039291578700445}
{"step": 85935, "loss": 3.504840850830078, "grad_norm": 0.7667545676231384, "learning_rate": 0.0001, "elbo": 3.50484, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.039584799437016}
{"step": 85940, "loss": 3.4835693359375, "grad_norm": 0.7319152355194092, "learning_rate": 0.0001, "elbo": 3.48357, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.039878020173587}
{"step": 85945, "loss": 3.4800750732421877, "grad_norm": 0.696704626083374, "learning_rate": 0.0001, "elbo": 3.48008, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.040171240910158}
{"step": 85950, "loss": 3.479070281982422, "grad_norm": 0.7567092776298523, "learning_rate": 0.0001, "elbo": 3.47907, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.040464461646728}
{"step": 85955, "loss": 3.4801902770996094, "grad_norm": 0.7863401770591736, "learning_rate": 0.0001, "elbo": 3.48019, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0407576823832985}
{"step": 85960, "loss": 3.4775062561035157, "grad_norm": 0.850484311580658, "learning_rate": 0.0001, "elbo": 3.47751, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.041050903119869}
{"step": 85965, "loss": 3.489349365234375, "grad_norm": 0.7343650460243225, "learning_rate": 0.0001, "elbo": 3.48935, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.041344123856439}
{"step": 85970, "loss": 3.4845436096191404, "grad_norm": 0.7278706431388855, "learning_rate": 0.0001, "elbo": 3.48454, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.04163734459301}
{"step": 85975, "loss": 3.485173797607422, "grad_norm": 0.7169832587242126, "learning_rate": 0.0001, "elbo": 3.48517, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.04193056532958}
{"step": 85980, "loss": 3.4751914978027343, "grad_norm": 0.6933351159095764, "learning_rate": 0.0001, "elbo": 3.47519, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.042223786066151}
{"step": 85985, "loss": 3.5088546752929686, "grad_norm": 0.7103496193885803, "learning_rate": 0.0001, "elbo": 3.50885, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.042517006802721}
{"step": 85990, "loss": 3.487994384765625, "grad_norm": 0.7401042580604553, "learning_rate": 0.0001, "elbo": 3.48799, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.042810227539292}
{"step": 85995, "loss": 3.4930137634277343, "grad_norm": 0.8906069993972778, "learning_rate": 0.0001, "elbo": 3.49301, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.043103448275862}
{"step": 86000, "loss": 3.4880931854248045, "grad_norm": 0.7482863664627075, "learning_rate": 0.0001, "elbo": 3.48809, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0433966690124326}
{"step": 86005, "loss": 3.498424530029297, "grad_norm": 0.7109706997871399, "learning_rate": 0.0001, "elbo": 3.49842, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.043689889749003}
{"step": 86010, "loss": 3.5107803344726562, "grad_norm": 0.8245043754577637, "learning_rate": 0.0001, "elbo": 3.51078, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0439831104855735}
{"step": 86015, "loss": 3.4910118103027346, "grad_norm": 0.7120184302330017, "learning_rate": 0.0001, "elbo": 3.49101, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.044276331222144}
{"step": 86020, "loss": 3.505410003662109, "grad_norm": 0.727390706539154, "learning_rate": 0.0001, "elbo": 3.50541, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.044569551958714}
{"step": 86025, "loss": 3.494140625, "grad_norm": 0.7389621138572693, "learning_rate": 0.0001, "elbo": 3.49414, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.044862772695285}
{"step": 86030, "loss": 3.498352813720703, "grad_norm": 0.7157852649688721, "learning_rate": 0.0001, "elbo": 3.49835, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.045155993431855}
{"step": 86035, "loss": 3.510114288330078, "grad_norm": 0.7297243475914001, "learning_rate": 0.0001, "elbo": 3.51011, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.045449214168426}
{"step": 86040, "loss": 3.4880435943603514, "grad_norm": 0.7002964615821838, "learning_rate": 0.0001, "elbo": 3.48804, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.045742434904996}
{"step": 86045, "loss": 3.473577117919922, "grad_norm": 0.8264890909194946, "learning_rate": 0.0001, "elbo": 3.47358, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.046035655641567}
{"step": 86050, "loss": 3.500418853759766, "grad_norm": 0.6934701204299927, "learning_rate": 0.0001, "elbo": 3.50042, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.046328876378137}
{"step": 86055, "loss": 3.5079833984375, "grad_norm": 0.7194511294364929, "learning_rate": 0.0001, "elbo": 3.50798, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0466220971147076}
{"step": 86060, "loss": 3.4969844818115234, "grad_norm": 0.7465511560440063, "learning_rate": 0.0001, "elbo": 3.49698, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.046915317851278}
{"step": 86065, "loss": 3.5185287475585936, "grad_norm": 0.7123536467552185, "learning_rate": 0.0001, "elbo": 3.51853, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.047208538587849}
{"step": 86070, "loss": 3.5024654388427736, "grad_norm": 0.7229093313217163, "learning_rate": 0.0001, "elbo": 3.50247, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.04750175932442}
{"step": 86075, "loss": 3.4850234985351562, "grad_norm": 0.7809585332870483, "learning_rate": 0.0001, "elbo": 3.48502, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.04779498006099}
{"step": 86080, "loss": 3.4855072021484377, "grad_norm": 0.6592080593109131, "learning_rate": 0.0001, "elbo": 3.48551, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.048088200797561}
{"step": 86085, "loss": 3.496363067626953, "grad_norm": 0.7141802310943604, "learning_rate": 0.0001, "elbo": 3.49636, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.048381421534131}
{"step": 86090, "loss": 3.498874282836914, "grad_norm": 0.7005739212036133, "learning_rate": 0.0001, "elbo": 3.49887, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.048674642270702}
{"step": 86095, "loss": 3.490504837036133, "grad_norm": 0.8501392602920532, "learning_rate": 0.0001, "elbo": 3.4905, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.048967863007272}
{"step": 86100, "loss": 3.4827953338623048, "grad_norm": 0.731737494468689, "learning_rate": 0.0001, "elbo": 3.4828, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0492610837438425}
{"step": 86105, "loss": 3.4827125549316404, "grad_norm": 0.697520911693573, "learning_rate": 0.0001, "elbo": 3.48271, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.049554304480413}
{"step": 86110, "loss": 3.486937713623047, "grad_norm": 0.7613348960876465, "learning_rate": 0.0001, "elbo": 3.48694, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.049847525216983}
{"step": 86115, "loss": 3.498317337036133, "grad_norm": 0.7456720471382141, "learning_rate": 0.0001, "elbo": 3.49832, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.050140745953554}
{"step": 86120, "loss": 3.4996807098388674, "grad_norm": 0.7417953014373779, "learning_rate": 0.0001, "elbo": 3.49968, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.050433966690124}
{"step": 86125, "loss": 3.5025875091552736, "grad_norm": 0.8060399293899536, "learning_rate": 0.0001, "elbo": 3.50259, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.050727187426695}
{"step": 86130, "loss": 3.49923095703125, "grad_norm": 0.7193518877029419, "learning_rate": 0.0001, "elbo": 3.49923, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.051020408163265}
{"step": 86135, "loss": 3.4779434204101562, "grad_norm": 0.7220958471298218, "learning_rate": 0.0001, "elbo": 3.47794, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.051313628899836}
{"step": 86140, "loss": 3.4852256774902344, "grad_norm": 0.7113251686096191, "learning_rate": 0.0001, "elbo": 3.48523, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.051606849636406}
{"step": 86145, "loss": 3.500702667236328, "grad_norm": 0.6849411725997925, "learning_rate": 0.0001, "elbo": 3.5007, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.051900070372977}
{"step": 86150, "loss": 3.4966033935546874, "grad_norm": 0.7379515171051025, "learning_rate": 0.0001, "elbo": 3.4966, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.052193291109547}
{"step": 86155, "loss": 3.499083709716797, "grad_norm": 0.7471200227737427, "learning_rate": 0.0001, "elbo": 3.49908, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0524865118461175}
{"step": 86160, "loss": 3.490948486328125, "grad_norm": 0.7723972797393799, "learning_rate": 0.0001, "elbo": 3.49095, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.052779732582688}
{"step": 86165, "loss": 3.512212371826172, "grad_norm": 0.6953061819076538, "learning_rate": 0.0001, "elbo": 3.51221, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.053072953319258}
{"step": 86170, "loss": 3.478049468994141, "grad_norm": 0.728092610836029, "learning_rate": 0.0001, "elbo": 3.47805, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.053366174055829}
{"step": 86175, "loss": 3.5060958862304688, "grad_norm": 0.7161690592765808, "learning_rate": 0.0001, "elbo": 3.5061, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.053659394792399}
{"step": 86180, "loss": 3.5046722412109377, "grad_norm": 0.7310041785240173, "learning_rate": 0.0001, "elbo": 3.50467, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.05395261552897}
{"step": 86185, "loss": 3.4991500854492186, "grad_norm": 0.7283422350883484, "learning_rate": 0.0001, "elbo": 3.49915, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.054245836265541}
{"step": 86190, "loss": 3.4930130004882813, "grad_norm": 0.7957198619842529, "learning_rate": 0.0001, "elbo": 3.49301, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.054539057002112}
{"step": 86195, "loss": 3.4943199157714844, "grad_norm": 0.7117906212806702, "learning_rate": 0.0001, "elbo": 3.49432, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.054832277738682}
{"step": 86200, "loss": 3.489793395996094, "grad_norm": 0.75150066614151, "learning_rate": 0.0001, "elbo": 3.48979, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0551254984752525}
{"step": 86205, "loss": 3.4986557006835937, "grad_norm": 0.7650513052940369, "learning_rate": 0.0001, "elbo": 3.49866, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.055418719211823}
{"step": 86210, "loss": 3.4873199462890625, "grad_norm": 0.6848899722099304, "learning_rate": 0.0001, "elbo": 3.48732, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.055711939948393}
{"step": 86215, "loss": 3.477265548706055, "grad_norm": 0.7862704396247864, "learning_rate": 0.0001, "elbo": 3.47727, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.056005160684964}
{"step": 86220, "loss": 3.483272933959961, "grad_norm": 0.745416522026062, "learning_rate": 0.0001, "elbo": 3.48327, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.056298381421534}
{"step": 86225, "loss": 3.494007110595703, "grad_norm": 0.755672037601471, "learning_rate": 0.0001, "elbo": 3.49401, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.056591602158105}
{"step": 86230, "loss": 3.480891799926758, "grad_norm": 0.7431982159614563, "learning_rate": 0.0001, "elbo": 3.48089, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.056884822894675}
{"step": 86235, "loss": 3.493367385864258, "grad_norm": 0.7215965390205383, "learning_rate": 0.0001, "elbo": 3.49337, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.057178043631246}
{"step": 86240, "loss": 3.492139434814453, "grad_norm": 0.7079049944877625, "learning_rate": 0.0001, "elbo": 3.49214, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.057471264367816}
{"step": 86245, "loss": 3.4930320739746095, "grad_norm": 0.7908869981765747, "learning_rate": 0.0001, "elbo": 3.49303, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.057764485104387}
{"step": 86250, "loss": 3.4942562103271486, "grad_norm": 0.7615487575531006, "learning_rate": 0.0001, "elbo": 3.49426, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.058057705840957}
{"step": 86255, "loss": 3.5027511596679686, "grad_norm": 0.7558637261390686, "learning_rate": 0.0001, "elbo": 3.50275, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0583509265775275}
{"step": 86260, "loss": 3.4814952850341796, "grad_norm": 0.7136462926864624, "learning_rate": 0.0001, "elbo": 3.4815, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.058644147314098}
{"step": 86265, "loss": 3.5132633209228517, "grad_norm": 0.7236130237579346, "learning_rate": 0.0001, "elbo": 3.51326, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.058937368050668}
{"step": 86270, "loss": 3.4848281860351564, "grad_norm": 0.6903407573699951, "learning_rate": 0.0001, "elbo": 3.48483, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.059230588787239}
{"step": 86275, "loss": 3.5009876251220704, "grad_norm": 0.7087825536727905, "learning_rate": 0.0001, "elbo": 3.50099, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.059523809523809}
{"step": 86280, "loss": 3.4925437927246095, "grad_norm": 0.7142060995101929, "learning_rate": 0.0001, "elbo": 3.49254, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.05981703026038}
{"step": 86285, "loss": 3.4811481475830077, "grad_norm": 0.7112402319908142, "learning_rate": 0.0001, "elbo": 3.48115, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.06011025099695}
{"step": 86290, "loss": 3.4872886657714846, "grad_norm": 0.7542006969451904, "learning_rate": 0.0001, "elbo": 3.48729, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.060403471733521}
{"step": 86295, "loss": 3.4734893798828126, "grad_norm": 0.7245470881462097, "learning_rate": 0.0001, "elbo": 3.47349, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.060696692470091}
{"step": 86300, "loss": 3.477303314208984, "grad_norm": 0.7244335412979126, "learning_rate": 0.0001, "elbo": 3.4773, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.060989913206662}
{"step": 86305, "loss": 3.4923004150390624, "grad_norm": 0.7342992424964905, "learning_rate": 0.0001, "elbo": 3.4923, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.061283133943232}
{"step": 86310, "loss": 3.483171844482422, "grad_norm": 0.6880614161491394, "learning_rate": 0.0001, "elbo": 3.48317, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.061576354679803}
{"step": 86315, "loss": 3.4751739501953125, "grad_norm": 0.773354709148407, "learning_rate": 0.0001, "elbo": 3.47517, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.061869575416374}
{"step": 86320, "loss": 3.497529220581055, "grad_norm": 0.7965545654296875, "learning_rate": 0.0001, "elbo": 3.49753, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.062162796152944}
{"step": 86320, "loss": 3.497529220581055, "grad_norm": 0.7965545654296875, "learning_rate": 0.0001, "elbo": 3.49753, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.062162796152944}
{"step": 86325, "loss": 3.5052444458007814, "grad_norm": 0.7348421812057495, "learning_rate": 0.0001, "elbo": 3.50524, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.062456016889515}
{"step": 86330, "loss": 3.4974319458007814, "grad_norm": 0.6862909197807312, "learning_rate": 0.0001, "elbo": 3.49743, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.062749237626085}
{"step": 86335, "loss": 3.4802833557128907, "grad_norm": 0.7154505252838135, "learning_rate": 0.0001, "elbo": 3.48028, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.063042458362656}
{"step": 86340, "loss": 3.5031753540039063, "grad_norm": 0.7710074782371521, "learning_rate": 0.0001, "elbo": 3.50318, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.063335679099226}
{"step": 86345, "loss": 3.498282241821289, "grad_norm": 0.7405030727386475, "learning_rate": 0.0001, "elbo": 3.49828, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.063628899835797}
{"step": 86350, "loss": 3.489645004272461, "grad_norm": 0.7702907919883728, "learning_rate": 0.0001, "elbo": 3.48965, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.063922120572367}
{"step": 86350, "loss": 3.489645004272461, "grad_norm": 0.7702907919883728, "learning_rate": 0.0001, "elbo": 3.48965, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.063922120572367}
{"step": 86355, "loss": 3.487047576904297, "grad_norm": 0.7334175705909729, "learning_rate": 0.0001, "elbo": 3.48705, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0642153413089375}
{"step": 86355, "loss": 3.487047576904297, "grad_norm": 0.7334175705909729, "learning_rate": 0.0001, "elbo": 3.48705, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0642153413089375}
{"step": 86360, "loss": 3.491045379638672, "grad_norm": 0.7231292128562927, "learning_rate": 0.0001, "elbo": 3.49105, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.064508562045508}
{"step": 86365, "loss": 3.4859539031982423, "grad_norm": 0.6768295764923096, "learning_rate": 0.0001, "elbo": 3.48595, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.064801782782078}
{"step": 86370, "loss": 3.485390472412109, "grad_norm": 0.677182674407959, "learning_rate": 0.0001, "elbo": 3.48539, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.065095003518649}
{"step": 86375, "loss": 3.483995819091797, "grad_norm": 0.7319377660751343, "learning_rate": 0.0001, "elbo": 3.484, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.065388224255219}
{"step": 86380, "loss": 3.4802410125732424, "grad_norm": 0.7135044932365417, "learning_rate": 0.0001, "elbo": 3.48024, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.06568144499179}
{"step": 86385, "loss": 3.477197265625, "grad_norm": 0.7404744625091553, "learning_rate": 0.0001, "elbo": 3.4772, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.06597466572836}
{"step": 86390, "loss": 3.497653579711914, "grad_norm": 0.6796216368675232, "learning_rate": 0.0001, "elbo": 3.49765, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.066267886464931}
{"step": 86395, "loss": 3.505567169189453, "grad_norm": 0.8068524599075317, "learning_rate": 0.0001, "elbo": 3.50557, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.066561107201501}
{"step": 86400, "loss": 3.497140884399414, "grad_norm": 0.703333854675293, "learning_rate": 0.0001, "elbo": 3.49714, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.066854327938072}
{"step": 86405, "loss": 3.5040496826171874, "grad_norm": 0.6792892217636108, "learning_rate": 0.0001, "elbo": 3.50405, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.067147548674642}
{"step": 86410, "loss": 3.487439727783203, "grad_norm": 0.8404490351676941, "learning_rate": 0.0001, "elbo": 3.48744, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0674407694112125}
{"step": 86415, "loss": 3.483658218383789, "grad_norm": 0.7824705839157104, "learning_rate": 0.0001, "elbo": 3.48366, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.067733990147783}
{"step": 86420, "loss": 3.478342819213867, "grad_norm": 0.7263408899307251, "learning_rate": 0.0001, "elbo": 3.47834, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.068027210884353}
{"step": 86425, "loss": 3.477187728881836, "grad_norm": 0.800427258014679, "learning_rate": 0.0001, "elbo": 3.47719, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.068320431620924}
{"step": 86430, "loss": 3.5013092041015623, "grad_norm": 0.7213951945304871, "learning_rate": 0.0001, "elbo": 3.50131, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.068613652357495}
{"step": 86430, "loss": 3.5013092041015623, "grad_norm": 0.7213951945304871, "learning_rate": 0.0001, "elbo": 3.50131, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.068613652357495}
{"step": 86435, "loss": 3.4933700561523438, "grad_norm": 0.7045235633850098, "learning_rate": 0.0001, "elbo": 3.49337, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.068906873094066}
{"step": 86440, "loss": 3.511859130859375, "grad_norm": 0.7041993141174316, "learning_rate": 0.0001, "elbo": 3.51186, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.069200093830636}
{"step": 86445, "loss": 3.4935981750488283, "grad_norm": 0.7582682371139526, "learning_rate": 0.0001, "elbo": 3.4936, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0694933145672065}
{"step": 86450, "loss": 3.511796569824219, "grad_norm": 0.7029348015785217, "learning_rate": 0.0001, "elbo": 3.5118, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.069786535303777}
{"step": 86455, "loss": 3.5018470764160154, "grad_norm": 0.677709698677063, "learning_rate": 0.0001, "elbo": 3.50185, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.070079756040347}
{"step": 86460, "loss": 3.486713409423828, "grad_norm": 0.7054473161697388, "learning_rate": 0.0001, "elbo": 3.48671, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.070372976776918}
{"step": 86465, "loss": 3.480355072021484, "grad_norm": 0.6999825239181519, "learning_rate": 0.0001, "elbo": 3.48036, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.070666197513488}
{"step": 86470, "loss": 3.498113250732422, "grad_norm": 0.8073397278785706, "learning_rate": 0.0001, "elbo": 3.49811, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.070959418250059}
{"step": 86475, "loss": 3.5038246154785155, "grad_norm": 0.7396459579467773, "learning_rate": 0.0001, "elbo": 3.50382, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.071252638986629}
{"step": 86480, "loss": 3.4906387329101562, "grad_norm": 0.7978968620300293, "learning_rate": 0.0001, "elbo": 3.49064, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0715458597232}
{"step": 86485, "loss": 3.4853111267089845, "grad_norm": 0.6718019247055054, "learning_rate": 0.0001, "elbo": 3.48531, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.07183908045977}
{"step": 86490, "loss": 3.4978504180908203, "grad_norm": 0.8262595534324646, "learning_rate": 0.0001, "elbo": 3.49785, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.072132301196341}
{"step": 86495, "loss": 3.4815101623535156, "grad_norm": 0.7243471741676331, "learning_rate": 0.0001, "elbo": 3.48151, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.072425521932911}
{"step": 86500, "loss": 3.482522964477539, "grad_norm": 0.7097600698471069, "learning_rate": 0.0001, "elbo": 3.48252, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0727187426694815}
{"step": 86505, "loss": 3.4842498779296873, "grad_norm": 0.667453408241272, "learning_rate": 0.0001, "elbo": 3.48425, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.073011963406052}
{"step": 86510, "loss": 3.476556396484375, "grad_norm": 0.7702915072441101, "learning_rate": 0.0001, "elbo": 3.47656, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.073305184142622}
{"step": 86515, "loss": 3.492496871948242, "grad_norm": 0.7146273255348206, "learning_rate": 0.0001, "elbo": 3.4925, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.073598404879193}
{"step": 86520, "loss": 3.5056663513183595, "grad_norm": 0.7521244883537292, "learning_rate": 0.0001, "elbo": 3.50567, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.073891625615763}
{"step": 86525, "loss": 3.4913261413574217, "grad_norm": 0.7854811549186707, "learning_rate": 0.0001, "elbo": 3.49133, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.074184846352334}
{"step": 86530, "loss": 3.4836990356445314, "grad_norm": 0.8395960330963135, "learning_rate": 0.0001, "elbo": 3.4837, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.074478067088904}
{"step": 86535, "loss": 3.4888141632080076, "grad_norm": 0.7304263710975647, "learning_rate": 0.0001, "elbo": 3.48881, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.074771287825475}
{"step": 86540, "loss": 3.503635787963867, "grad_norm": 0.7523913383483887, "learning_rate": 0.0001, "elbo": 3.50364, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.075064508562045}
{"step": 86545, "loss": 3.488298034667969, "grad_norm": 0.8314472436904907, "learning_rate": 0.0001, "elbo": 3.4883, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.075357729298616}
{"step": 86550, "loss": 3.493317413330078, "grad_norm": 0.7023553848266602, "learning_rate": 0.0001, "elbo": 3.49332, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.075650950035186}
{"step": 86555, "loss": 3.5178367614746096, "grad_norm": 0.744490385055542, "learning_rate": 0.0001, "elbo": 3.51784, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.075944170771757}
{"step": 86560, "loss": 3.501496124267578, "grad_norm": 0.7072452902793884, "learning_rate": 0.0001, "elbo": 3.5015, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.076237391508328}
{"step": 86565, "loss": 3.485340118408203, "grad_norm": 0.7287592887878418, "learning_rate": 0.0001, "elbo": 3.48534, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.076530612244898}
{"step": 86570, "loss": 3.488961029052734, "grad_norm": 0.7374773621559143, "learning_rate": 0.0001, "elbo": 3.48896, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.076823832981469}
{"step": 86575, "loss": 3.49635009765625, "grad_norm": 0.7077294588088989, "learning_rate": 0.0001, "elbo": 3.49635, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.077117053718039}
{"step": 86580, "loss": 3.494210052490234, "grad_norm": 0.7321678400039673, "learning_rate": 0.0001, "elbo": 3.49421, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.07741027445461}
{"step": 86585, "loss": 3.5017425537109377, "grad_norm": 0.7894037961959839, "learning_rate": 0.0001, "elbo": 3.50174, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.07770349519118}
{"step": 86590, "loss": 3.5091747283935546, "grad_norm": 0.7260892391204834, "learning_rate": 0.0001, "elbo": 3.50917, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.077996715927751}
{"step": 86595, "loss": 3.496209716796875, "grad_norm": 0.745501697063446, "learning_rate": 0.0001, "elbo": 3.49621, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.078289936664321}
{"step": 86600, "loss": 3.4766616821289062, "grad_norm": 0.6892572045326233, "learning_rate": 0.0001, "elbo": 3.47666, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0785831574008915}
{"step": 86605, "loss": 3.486395263671875, "grad_norm": 0.6972854137420654, "learning_rate": 0.0001, "elbo": 3.4864, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.078876378137462}
{"step": 86610, "loss": 3.5015533447265623, "grad_norm": 0.7388630509376526, "learning_rate": 0.0001, "elbo": 3.50155, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.079169598874032}
{"step": 86615, "loss": 3.4880630493164064, "grad_norm": 0.7064297199249268, "learning_rate": 0.0001, "elbo": 3.48806, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.079462819610603}
{"step": 86620, "loss": 3.487357330322266, "grad_norm": 0.6808844208717346, "learning_rate": 0.0001, "elbo": 3.48736, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.079756040347173}
{"step": 86625, "loss": 3.5081050872802733, "grad_norm": 0.7293896675109863, "learning_rate": 0.0001, "elbo": 3.50811, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.080049261083744}
{"step": 86630, "loss": 3.4951595306396483, "grad_norm": 0.7109367251396179, "learning_rate": 0.0001, "elbo": 3.49516, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.080342481820314}
{"step": 86635, "loss": 3.485565185546875, "grad_norm": 0.9989640712738037, "learning_rate": 0.0001, "elbo": 3.48557, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.080635702556885}
{"step": 86640, "loss": 3.4959373474121094, "grad_norm": 0.7939566373825073, "learning_rate": 0.0001, "elbo": 3.49594, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.080928923293455}
{"step": 86645, "loss": 3.49820556640625, "grad_norm": 0.7269980311393738, "learning_rate": 0.0001, "elbo": 3.49821, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.081222144030026}
{"step": 86650, "loss": 3.505023193359375, "grad_norm": 0.7253599762916565, "learning_rate": 0.0001, "elbo": 3.50502, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.081515364766596}
{"step": 86655, "loss": 3.4908870697021483, "grad_norm": 0.7170113921165466, "learning_rate": 0.0001, "elbo": 3.49089, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0818085855031665}
{"step": 86660, "loss": 3.490156555175781, "grad_norm": 0.8223321437835693, "learning_rate": 0.0001, "elbo": 3.49016, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.082101806239737}
{"step": 86665, "loss": 3.4871559143066406, "grad_norm": 0.7275827527046204, "learning_rate": 0.0001, "elbo": 3.48716, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.082395026976307}
{"step": 86670, "loss": 3.4881614685058593, "grad_norm": 0.7317677140235901, "learning_rate": 0.0001, "elbo": 3.48816, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.082688247712878}
{"step": 86675, "loss": 3.4820518493652344, "grad_norm": 0.6982287168502808, "learning_rate": 0.0001, "elbo": 3.48205, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.082981468449448}
{"step": 86680, "loss": 3.5037151336669923, "grad_norm": 0.8110769391059875, "learning_rate": 0.0001, "elbo": 3.50371, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.08327468918602}
{"step": 86685, "loss": 3.504747772216797, "grad_norm": 0.7384265661239624, "learning_rate": 0.0001, "elbo": 3.50475, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.08356790992259}
{"step": 86690, "loss": 3.504346466064453, "grad_norm": 0.7257641553878784, "learning_rate": 0.0001, "elbo": 3.50435, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.083861130659161}
{"step": 86695, "loss": 3.490240478515625, "grad_norm": 0.8630391955375671, "learning_rate": 0.0001, "elbo": 3.49024, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.084154351395731}
{"step": 86700, "loss": 3.4902854919433595, "grad_norm": 0.7191653251647949, "learning_rate": 0.0001, "elbo": 3.49029, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0844475721323015}
{"step": 86705, "loss": 3.4991851806640626, "grad_norm": 0.7214303016662598, "learning_rate": 0.0001, "elbo": 3.49919, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.084740792868872}
{"step": 86710, "loss": 3.498548126220703, "grad_norm": 0.7689287662506104, "learning_rate": 0.0001, "elbo": 3.49855, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.085034013605442}
{"step": 86715, "loss": 3.4799789428710937, "grad_norm": 0.7731877565383911, "learning_rate": 0.0001, "elbo": 3.47998, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.085327234342013}
{"step": 86720, "loss": 3.495954895019531, "grad_norm": 0.7082096338272095, "learning_rate": 0.0001, "elbo": 3.49595, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.085620455078583}
{"step": 86725, "loss": 3.493206024169922, "grad_norm": 0.8072870969772339, "learning_rate": 0.0001, "elbo": 3.49321, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.085913675815154}
{"step": 86730, "loss": 3.4902099609375, "grad_norm": 0.7076336145401001, "learning_rate": 0.0001, "elbo": 3.49021, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.086206896551724}
{"step": 86735, "loss": 3.4975559234619142, "grad_norm": 0.7132862210273743, "learning_rate": 0.0001, "elbo": 3.49756, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.086500117288295}
{"step": 86740, "loss": 3.4872589111328125, "grad_norm": 0.7139790058135986, "learning_rate": 0.0001, "elbo": 3.48726, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.086793338024865}
{"step": 86745, "loss": 3.4850868225097655, "grad_norm": 0.7158461213111877, "learning_rate": 0.0001, "elbo": 3.48509, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.087086558761436}
{"step": 86750, "loss": 3.4834720611572267, "grad_norm": 0.7154690623283386, "learning_rate": 0.0001, "elbo": 3.48347, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.087379779498006}
{"step": 86755, "loss": 3.4909294128417967, "grad_norm": 0.7304204702377319, "learning_rate": 0.0001, "elbo": 3.49093, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0876730002345765}
{"step": 86760, "loss": 3.482746887207031, "grad_norm": 0.6953125596046448, "learning_rate": 0.0001, "elbo": 3.48275, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.087966220971147}
{"step": 86765, "loss": 3.4890625, "grad_norm": 1.4226255416870117, "learning_rate": 0.0001, "elbo": 3.48906, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.088259441707717}
{"step": 86770, "loss": 3.486103820800781, "grad_norm": 0.9345836043357849, "learning_rate": 0.0001, "elbo": 3.4861, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.088552662444288}
{"step": 86775, "loss": 3.48673095703125, "grad_norm": 0.77520751953125, "learning_rate": 0.0001, "elbo": 3.48673, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.088845883180858}
{"step": 86780, "loss": 3.4917972564697264, "grad_norm": 0.7907167673110962, "learning_rate": 0.0001, "elbo": 3.4918, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.089139103917429}
{"step": 86785, "loss": 3.4815216064453125, "grad_norm": 0.7031881809234619, "learning_rate": 0.0001, "elbo": 3.48152, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.089432324653999}
{"step": 86790, "loss": 3.497770309448242, "grad_norm": 0.7634459733963013, "learning_rate": 0.0001, "elbo": 3.49777, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.08972554539057}
{"step": 86795, "loss": 3.4874767303466796, "grad_norm": 0.7402170896530151, "learning_rate": 0.0001, "elbo": 3.48748, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.09001876612714}
{"step": 86795, "loss": 3.4874767303466796, "grad_norm": 0.7402170896530151, "learning_rate": 0.0001, "elbo": 3.48748, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.09001876612714}
{"step": 86800, "loss": 3.4892669677734376, "grad_norm": 0.7034540772438049, "learning_rate": 0.0001, "elbo": 3.48927, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.090311986863711}
{"step": 86805, "loss": 3.480490493774414, "grad_norm": 0.7300239205360413, "learning_rate": 0.0001, "elbo": 3.48049, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.090605207600282}
{"step": 86810, "loss": 3.485340881347656, "grad_norm": 0.7356367707252502, "learning_rate": 0.0001, "elbo": 3.48534, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.090898428336852}
{"step": 86815, "loss": 3.496419143676758, "grad_norm": 0.6763433218002319, "learning_rate": 0.0001, "elbo": 3.49642, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.091191649073423}
{"step": 86820, "loss": 3.494293212890625, "grad_norm": 0.8362470269203186, "learning_rate": 0.0001, "elbo": 3.49429, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.091484869809993}
{"step": 86825, "loss": 3.4948558807373047, "grad_norm": 0.6983798742294312, "learning_rate": 0.0001, "elbo": 3.49486, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.091778090546564}
{"step": 86830, "loss": 3.501447296142578, "grad_norm": 0.7486053109169006, "learning_rate": 0.0001, "elbo": 3.50145, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.092071311283134}
{"step": 86835, "loss": 3.500616455078125, "grad_norm": 0.7490627765655518, "learning_rate": 0.0001, "elbo": 3.50062, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.092364532019705}
{"step": 86840, "loss": 3.4927108764648436, "grad_norm": 0.7068608403205872, "learning_rate": 0.0001, "elbo": 3.49271, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.092657752756275}
{"step": 86845, "loss": 3.4876174926757812, "grad_norm": 0.6985561847686768, "learning_rate": 0.0001, "elbo": 3.48762, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0929509734928455}
{"step": 86850, "loss": 3.4927139282226562, "grad_norm": 1.022764801979065, "learning_rate": 0.0001, "elbo": 3.49271, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.093244194229416}
{"step": 86855, "loss": 3.481732177734375, "grad_norm": 0.7776220440864563, "learning_rate": 0.0001, "elbo": 3.48173, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.093537414965986}
{"step": 86860, "loss": 3.5108692169189455, "grad_norm": 0.7285887598991394, "learning_rate": 0.0001, "elbo": 3.51087, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.093830635702557}
{"step": 86865, "loss": 3.4951637268066404, "grad_norm": 0.7278507351875305, "learning_rate": 0.0001, "elbo": 3.49516, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.094123856439127}
{"step": 86870, "loss": 3.4713790893554686, "grad_norm": 0.7693235874176025, "learning_rate": 0.0001, "elbo": 3.47138, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.094417077175698}
{"step": 86875, "loss": 3.502996826171875, "grad_norm": 0.7473641633987427, "learning_rate": 0.0001, "elbo": 3.503, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.094710297912268}
{"step": 86880, "loss": 3.4851898193359374, "grad_norm": 0.7380861043930054, "learning_rate": 0.0001, "elbo": 3.48519, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.095003518648839}
{"step": 86885, "loss": 3.4883663177490236, "grad_norm": 1.037563681602478, "learning_rate": 0.0001, "elbo": 3.48837, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.095296739385409}
{"step": 86890, "loss": 3.4956943511962892, "grad_norm": 1.5486518144607544, "learning_rate": 0.0001, "elbo": 3.49569, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.09558996012198}
{"step": 86895, "loss": 3.493601989746094, "grad_norm": 0.691408097743988, "learning_rate": 0.0001, "elbo": 3.4936, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.09588318085855}
{"step": 86900, "loss": 3.496054840087891, "grad_norm": 0.7013993263244629, "learning_rate": 0.0001, "elbo": 3.49605, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0961764015951205}
{"step": 86905, "loss": 3.503630828857422, "grad_norm": 0.7034774422645569, "learning_rate": 0.0001, "elbo": 3.50363, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.096469622331691}
{"step": 86910, "loss": 3.499528503417969, "grad_norm": 0.735780656337738, "learning_rate": 0.0001, "elbo": 3.49953, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.096762843068261}
{"step": 86915, "loss": 3.491896057128906, "grad_norm": 0.6951791644096375, "learning_rate": 0.0001, "elbo": 3.4919, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.097056063804832}
{"step": 86920, "loss": 3.486994171142578, "grad_norm": 0.7030433416366577, "learning_rate": 0.0001, "elbo": 3.48699, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.097349284541402}
{"step": 86925, "loss": 3.497278594970703, "grad_norm": 0.6909311413764954, "learning_rate": 0.0001, "elbo": 3.49728, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.097642505277974}
{"step": 86930, "loss": 3.485364532470703, "grad_norm": 0.7483514547348022, "learning_rate": 0.0001, "elbo": 3.48536, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.097935726014544}
{"step": 86935, "loss": 3.482974624633789, "grad_norm": 0.764713704586029, "learning_rate": 0.0001, "elbo": 3.48297, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.098228946751115}
{"step": 86940, "loss": 3.494329071044922, "grad_norm": 0.7090026140213013, "learning_rate": 0.0001, "elbo": 3.49433, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.098522167487685}
{"step": 86945, "loss": 3.5098167419433595, "grad_norm": 0.7251377105712891, "learning_rate": 0.0001, "elbo": 3.50982, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.0988153882242555}
{"step": 86950, "loss": 3.475385284423828, "grad_norm": 0.7449083924293518, "learning_rate": 0.0001, "elbo": 3.47539, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.099108608960826}
{"step": 86955, "loss": 3.4844322204589844, "grad_norm": 0.8166438341140747, "learning_rate": 0.0001, "elbo": 3.48443, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.099401829697396}
{"step": 86960, "loss": 3.485485076904297, "grad_norm": 0.9039192199707031, "learning_rate": 0.0001, "elbo": 3.48549, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.099695050433967}
{"step": 86965, "loss": 3.4973480224609377, "grad_norm": 0.7244311571121216, "learning_rate": 0.0001, "elbo": 3.49735, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.099988271170537}
{"step": 86970, "loss": 3.514719009399414, "grad_norm": 0.7319359183311462, "learning_rate": 0.0001, "elbo": 3.51472, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.100281491907108}
{"step": 86975, "loss": 3.4962490081787108, "grad_norm": 0.6815712451934814, "learning_rate": 0.0001, "elbo": 3.49625, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.100574712643678}
{"step": 86980, "loss": 3.48673095703125, "grad_norm": 0.7001824975013733, "learning_rate": 0.0001, "elbo": 3.48673, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.100867933380249}
{"step": 86985, "loss": 3.5094009399414063, "grad_norm": 0.6891467571258545, "learning_rate": 0.0001, "elbo": 3.5094, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.101161154116819}
{"step": 86990, "loss": 3.494581604003906, "grad_norm": 0.7384427785873413, "learning_rate": 0.0001, "elbo": 3.49458, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.10145437485339}
{"step": 86995, "loss": 3.4775016784667967, "grad_norm": 0.7550068497657776, "learning_rate": 0.0001, "elbo": 3.4775, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.10174759558996}
{"step": 87000, "loss": 3.4939414978027346, "grad_norm": 0.7088379263877869, "learning_rate": 0.0001, "elbo": 3.49394, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1020408163265305}
{"step": 87005, "loss": 3.4938926696777344, "grad_norm": 0.7311163544654846, "learning_rate": 0.0001, "elbo": 3.49389, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.102334037063101}
{"step": 87010, "loss": 3.493958282470703, "grad_norm": 0.7895059585571289, "learning_rate": 0.0001, "elbo": 3.49396, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.102627257799671}
{"step": 87015, "loss": 3.4929275512695312, "grad_norm": 0.8833079934120178, "learning_rate": 0.0001, "elbo": 3.49293, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.102920478536242}
{"step": 87020, "loss": 3.4856014251708984, "grad_norm": 0.7028881907463074, "learning_rate": 0.0001, "elbo": 3.4856, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.103213699272812}
{"step": 87025, "loss": 3.4904457092285157, "grad_norm": 0.733610212802887, "learning_rate": 0.0001, "elbo": 3.49045, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.103506920009383}
{"step": 87030, "loss": 3.47203369140625, "grad_norm": 0.9400174021720886, "learning_rate": 0.0001, "elbo": 3.47203, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.103800140745953}
{"step": 87035, "loss": 3.492176055908203, "grad_norm": 0.7354387640953064, "learning_rate": 0.0001, "elbo": 3.49218, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.104093361482524}
{"step": 87040, "loss": 3.4946434020996096, "grad_norm": 0.7035471796989441, "learning_rate": 0.0001, "elbo": 3.49464, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.104386582219094}
{"step": 87045, "loss": 3.492784881591797, "grad_norm": 0.7197986245155334, "learning_rate": 0.0001, "elbo": 3.49279, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.104679802955665}
{"step": 87050, "loss": 3.4786453247070312, "grad_norm": 0.9000102877616882, "learning_rate": 0.0001, "elbo": 3.47865, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.104973023692236}
{"step": 87055, "loss": 3.493745803833008, "grad_norm": 0.7315071821212769, "learning_rate": 0.0001, "elbo": 3.49375, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.105266244428806}
{"step": 87055, "loss": 3.493745803833008, "grad_norm": 0.7315071821212769, "learning_rate": 0.0001, "elbo": 3.49375, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.105266244428806}
{"step": 87060, "loss": 3.4903587341308593, "grad_norm": 0.8199558258056641, "learning_rate": 0.0001, "elbo": 3.49036, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.105559465165377}
{"step": 87065, "loss": 3.4832199096679686, "grad_norm": 0.7447496056556702, "learning_rate": 0.0001, "elbo": 3.48322, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.105852685901947}
{"step": 87070, "loss": 3.477779769897461, "grad_norm": 0.6897432804107666, "learning_rate": 0.0001, "elbo": 3.47778, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.106145906638518}
{"step": 87075, "loss": 3.4707508087158203, "grad_norm": 0.6860402226448059, "learning_rate": 0.0001, "elbo": 3.47075, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.106439127375088}
{"step": 87080, "loss": 3.4968338012695312, "grad_norm": 0.7602118253707886, "learning_rate": 0.0001, "elbo": 3.49683, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.106732348111659}
{"step": 87085, "loss": 3.4755149841308595, "grad_norm": 0.703068733215332, "learning_rate": 0.0001, "elbo": 3.47551, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.107025568848229}
{"step": 87090, "loss": 3.495919036865234, "grad_norm": 0.7934556603431702, "learning_rate": 0.0001, "elbo": 3.49592, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1073187895848}
{"step": 87095, "loss": 3.505189895629883, "grad_norm": 0.7791651487350464, "learning_rate": 0.0001, "elbo": 3.50519, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.10761201032137}
{"step": 87100, "loss": 3.5009220123291014, "grad_norm": 0.7472738027572632, "learning_rate": 0.0001, "elbo": 3.50092, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1079052310579405}
{"step": 87105, "loss": 3.499306488037109, "grad_norm": 0.7624058127403259, "learning_rate": 0.0001, "elbo": 3.49931, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.108198451794511}
{"step": 87110, "loss": 3.501848602294922, "grad_norm": 0.6767935752868652, "learning_rate": 0.0001, "elbo": 3.50185, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.108491672531081}
{"step": 87115, "loss": 3.4956050872802735, "grad_norm": 0.7208005785942078, "learning_rate": 0.0001, "elbo": 3.49561, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.108784893267652}
{"step": 87120, "loss": 3.490924835205078, "grad_norm": 0.7652122974395752, "learning_rate": 0.0001, "elbo": 3.49092, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.109078114004222}
{"step": 87125, "loss": 3.4939414978027346, "grad_norm": 0.6898097395896912, "learning_rate": 0.0001, "elbo": 3.49394, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.109371334740793}
{"step": 87130, "loss": 3.487485885620117, "grad_norm": 0.7384491562843323, "learning_rate": 0.0001, "elbo": 3.48749, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.109664555477363}
{"step": 87135, "loss": 3.4897090911865236, "grad_norm": 0.733733057975769, "learning_rate": 0.0001, "elbo": 3.48971, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.109957776213934}
{"step": 87140, "loss": 3.489274597167969, "grad_norm": 0.6950270533561707, "learning_rate": 0.0001, "elbo": 3.48927, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.110250996950504}
{"step": 87145, "loss": 3.505078887939453, "grad_norm": 0.7689886093139648, "learning_rate": 0.0001, "elbo": 3.50508, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.110544217687075}
{"step": 87150, "loss": 3.499768829345703, "grad_norm": 0.725059986114502, "learning_rate": 0.0001, "elbo": 3.49977, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.110837438423645}
{"step": 87155, "loss": 3.479308319091797, "grad_norm": 0.8737389445304871, "learning_rate": 0.0001, "elbo": 3.47931, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1111306591602155}
{"step": 87160, "loss": 3.488941192626953, "grad_norm": 0.7691349983215332, "learning_rate": 0.0001, "elbo": 3.48894, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.111423879896786}
{"step": 87165, "loss": 3.4794456481933596, "grad_norm": 0.6873666644096375, "learning_rate": 0.0001, "elbo": 3.47945, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.111717100633356}
{"step": 87170, "loss": 3.5056106567382814, "grad_norm": 0.7361232042312622, "learning_rate": 0.0001, "elbo": 3.50561, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.112010321369928}
{"step": 87175, "loss": 3.5019580841064455, "grad_norm": 0.7063719630241394, "learning_rate": 0.0001, "elbo": 3.50196, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.112303542106498}
{"step": 87180, "loss": 3.485005187988281, "grad_norm": 0.7105388045310974, "learning_rate": 0.0001, "elbo": 3.48501, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.112596762843069}
{"step": 87185, "loss": 3.492231750488281, "grad_norm": 0.7476130723953247, "learning_rate": 0.0001, "elbo": 3.49223, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.112889983579639}
{"step": 87190, "loss": 3.491812896728516, "grad_norm": 0.6987453103065491, "learning_rate": 0.0001, "elbo": 3.49181, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1131832043162095}
{"step": 87190, "loss": 3.491812896728516, "grad_norm": 0.6987453103065491, "learning_rate": 0.0001, "elbo": 3.49181, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1131832043162095}
{"step": 87195, "loss": 3.4929489135742187, "grad_norm": 0.6884957551956177, "learning_rate": 0.0001, "elbo": 3.49295, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.11347642505278}
{"step": 87200, "loss": 3.483582305908203, "grad_norm": 0.7268067002296448, "learning_rate": 0.0001, "elbo": 3.48358, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1137696457893504}
{"step": 87205, "loss": 3.498149871826172, "grad_norm": 0.793344259262085, "learning_rate": 0.0001, "elbo": 3.49815, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.114062866525921}
{"step": 87210, "loss": 3.4863971710205077, "grad_norm": 0.7205674648284912, "learning_rate": 0.0001, "elbo": 3.4864, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.114356087262491}
{"step": 87215, "loss": 3.493073272705078, "grad_norm": 0.7298335433006287, "learning_rate": 0.0001, "elbo": 3.49307, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.114649307999062}
{"step": 87220, "loss": 3.4837051391601563, "grad_norm": 0.7881079316139221, "learning_rate": 0.0001, "elbo": 3.48371, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.114942528735632}
{"step": 87225, "loss": 3.50301513671875, "grad_norm": 0.7666898965835571, "learning_rate": 0.0001, "elbo": 3.50301, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.115235749472203}
{"step": 87230, "loss": 3.4815807342529297, "grad_norm": 0.8994235992431641, "learning_rate": 0.0001, "elbo": 3.48158, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.115528970208773}
{"step": 87235, "loss": 3.511585998535156, "grad_norm": 0.8244777321815491, "learning_rate": 0.0001, "elbo": 3.51159, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.115822190945344}
{"step": 87240, "loss": 3.4850765228271485, "grad_norm": 0.6808497309684753, "learning_rate": 0.0001, "elbo": 3.48508, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.116115411681914}
{"step": 87245, "loss": 3.493377685546875, "grad_norm": 0.7053165435791016, "learning_rate": 0.0001, "elbo": 3.49338, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1164086324184845}
{"step": 87250, "loss": 3.4753883361816404, "grad_norm": 0.6985166072845459, "learning_rate": 0.0001, "elbo": 3.47539, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.116701853155055}
{"step": 87255, "loss": 3.486772918701172, "grad_norm": 0.7550391554832458, "learning_rate": 0.0001, "elbo": 3.48677, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1169950738916254}
{"step": 87260, "loss": 3.505488967895508, "grad_norm": 0.7068614363670349, "learning_rate": 0.0001, "elbo": 3.50549, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.117288294628196}
{"step": 87265, "loss": 3.5010753631591798, "grad_norm": 0.6874985694885254, "learning_rate": 0.0001, "elbo": 3.50108, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.117581515364766}
{"step": 87270, "loss": 3.4851364135742187, "grad_norm": 0.7419730424880981, "learning_rate": 0.0001, "elbo": 3.48514, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.117874736101337}
{"step": 87275, "loss": 3.4989654541015627, "grad_norm": 0.731177806854248, "learning_rate": 0.0001, "elbo": 3.49897, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.118167956837907}
{"step": 87280, "loss": 3.490906524658203, "grad_norm": 0.8463844060897827, "learning_rate": 0.0001, "elbo": 3.49091, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.118461177574478}
{"step": 87285, "loss": 3.5039440155029298, "grad_norm": 0.7157147526741028, "learning_rate": 0.0001, "elbo": 3.50394, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.118754398311048}
{"step": 87285, "loss": 3.5039440155029298, "grad_norm": 0.7157147526741028, "learning_rate": 0.0001, "elbo": 3.50394, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.118754398311048}
{"step": 87290, "loss": 3.485692596435547, "grad_norm": 0.7203572392463684, "learning_rate": 0.0001, "elbo": 3.48569, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.119047619047619}
{"step": 87290, "loss": 3.485692596435547, "grad_norm": 0.7203572392463684, "learning_rate": 0.0001, "elbo": 3.48569, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.119047619047619}
{"step": 87295, "loss": 3.49297981262207, "grad_norm": 0.7876144051551819, "learning_rate": 0.0001, "elbo": 3.49298, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.11934083978419}
{"step": 87300, "loss": 3.4952312469482423, "grad_norm": 0.7295973300933838, "learning_rate": 0.0001, "elbo": 3.49523, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.11963406052076}
{"step": 87300, "loss": 3.4952312469482423, "grad_norm": 0.7295973300933838, "learning_rate": 0.0001, "elbo": 3.49523, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.11963406052076}
{"step": 87305, "loss": 3.494567108154297, "grad_norm": 0.8849341869354248, "learning_rate": 0.0001, "elbo": 3.49457, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.119927281257331}
{"step": 87310, "loss": 3.492536163330078, "grad_norm": 0.8636398911476135, "learning_rate": 0.0001, "elbo": 3.49254, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.120220501993901}
{"step": 87315, "loss": 3.4914173126220702, "grad_norm": 0.72340989112854, "learning_rate": 0.0001, "elbo": 3.49142, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.120513722730472}
{"step": 87320, "loss": 3.489675521850586, "grad_norm": 1.2842193841934204, "learning_rate": 0.0001, "elbo": 3.48968, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.120806943467042}
{"step": 87325, "loss": 3.47451171875, "grad_norm": 0.709736168384552, "learning_rate": 0.0001, "elbo": 3.47451, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.121100164203613}
{"step": 87330, "loss": 3.4816864013671873, "grad_norm": 0.7715714573860168, "learning_rate": 0.0001, "elbo": 3.48169, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.121393384940183}
{"step": 87335, "loss": 3.486199951171875, "grad_norm": 0.7207276821136475, "learning_rate": 0.0001, "elbo": 3.4862, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.121686605676754}
{"step": 87340, "loss": 3.486209487915039, "grad_norm": 0.8956354260444641, "learning_rate": 0.0001, "elbo": 3.48621, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.121979826413324}
{"step": 87345, "loss": 3.4837631225585937, "grad_norm": 0.739240825176239, "learning_rate": 0.0001, "elbo": 3.48376, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1222730471498945}
{"step": 87350, "loss": 3.4931404113769533, "grad_norm": 0.7021640539169312, "learning_rate": 0.0001, "elbo": 3.49314, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.122566267886465}
{"step": 87355, "loss": 3.478974151611328, "grad_norm": 0.7923546433448792, "learning_rate": 0.0001, "elbo": 3.47897, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.122859488623035}
{"step": 87360, "loss": 3.489710235595703, "grad_norm": 0.7504963874816895, "learning_rate": 0.0001, "elbo": 3.48971, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.123152709359606}
{"step": 87365, "loss": 3.477922058105469, "grad_norm": 0.7212617993354797, "learning_rate": 0.0001, "elbo": 3.47792, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.123445930096176}
{"step": 87370, "loss": 3.4920112609863283, "grad_norm": 0.7406001687049866, "learning_rate": 0.0001, "elbo": 3.49201, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.123739150832747}
{"step": 87375, "loss": 3.5049468994140627, "grad_norm": 0.7271462678909302, "learning_rate": 0.0001, "elbo": 3.50495, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.124032371569317}
{"step": 87380, "loss": 3.5124229431152343, "grad_norm": 0.7674070596694946, "learning_rate": 0.0001, "elbo": 3.51242, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.124325592305888}
{"step": 87385, "loss": 3.4733543395996094, "grad_norm": 0.7244125008583069, "learning_rate": 0.0001, "elbo": 3.47335, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.124618813042458}
{"step": 87390, "loss": 3.474303436279297, "grad_norm": 0.9627926349639893, "learning_rate": 0.0001, "elbo": 3.4743, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.124912033779029}
{"step": 87395, "loss": 3.4742691040039064, "grad_norm": 0.7066783308982849, "learning_rate": 0.0001, "elbo": 3.47427, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.125205254515599}
{"step": 87400, "loss": 3.5033153533935546, "grad_norm": 0.813732385635376, "learning_rate": 0.0001, "elbo": 3.50332, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1254984752521695}
{"step": 87405, "loss": 3.503701019287109, "grad_norm": 0.8050731420516968, "learning_rate": 0.0001, "elbo": 3.5037, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.12579169598874}
{"step": 87410, "loss": 3.49450798034668, "grad_norm": 0.7182983756065369, "learning_rate": 0.0001, "elbo": 3.49451, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.12608491672531}
{"step": 87415, "loss": 3.4882881164550783, "grad_norm": 0.6859229207038879, "learning_rate": 0.0001, "elbo": 3.48829, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.126378137461881}
{"step": 87420, "loss": 3.4735691070556642, "grad_norm": 0.7129722237586975, "learning_rate": 0.0001, "elbo": 3.47357, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.126671358198452}
{"step": 87425, "loss": 3.4817153930664064, "grad_norm": 0.702642560005188, "learning_rate": 0.0001, "elbo": 3.48172, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.126964578935023}
{"step": 87430, "loss": 3.478866958618164, "grad_norm": 0.7008329629898071, "learning_rate": 0.0001, "elbo": 3.47887, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.127257799671593}
{"step": 87430, "loss": 3.478866958618164, "grad_norm": 0.7008329629898071, "learning_rate": 0.0001, "elbo": 3.47887, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.127257799671593}
{"step": 87435, "loss": 3.4955875396728517, "grad_norm": 0.6793635487556458, "learning_rate": 0.0001, "elbo": 3.49559, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.127551020408164}
{"step": 87440, "loss": 3.4820091247558596, "grad_norm": 0.7069128751754761, "learning_rate": 0.0001, "elbo": 3.48201, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.127844241144734}
{"step": 87445, "loss": 3.491514205932617, "grad_norm": 0.7323958873748779, "learning_rate": 0.0001, "elbo": 3.49151, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1281374618813045}
{"step": 87450, "loss": 3.483935546875, "grad_norm": 0.7063316106796265, "learning_rate": 0.0001, "elbo": 3.48394, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.128430682617875}
{"step": 87455, "loss": 3.4944747924804687, "grad_norm": 1.2102471590042114, "learning_rate": 0.0001, "elbo": 3.49447, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.128723903354445}
{"step": 87460, "loss": 3.4885223388671873, "grad_norm": 0.7582646012306213, "learning_rate": 0.0001, "elbo": 3.48852, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.129017124091016}
{"step": 87465, "loss": 3.4986946105957033, "grad_norm": 0.732912540435791, "learning_rate": 0.0001, "elbo": 3.49869, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.129310344827586}
{"step": 87470, "loss": 3.5039955139160157, "grad_norm": 0.7241246104240417, "learning_rate": 0.0001, "elbo": 3.504, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.129603565564157}
{"step": 87475, "loss": 3.4946884155273437, "grad_norm": 0.6905436515808105, "learning_rate": 0.0001, "elbo": 3.49469, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.129896786300727}
{"step": 87480, "loss": 3.4979198455810545, "grad_norm": 0.9044784903526306, "learning_rate": 0.0001, "elbo": 3.49792, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.130190007037298}
{"step": 87480, "loss": 3.4979198455810545, "grad_norm": 0.9044784903526306, "learning_rate": 0.0001, "elbo": 3.49792, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.130190007037298}
{"step": 87485, "loss": 3.497498321533203, "grad_norm": 0.7555994987487793, "learning_rate": 0.0001, "elbo": 3.4975, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.130483227773868}
{"step": 87490, "loss": 3.490626907348633, "grad_norm": 0.7718856334686279, "learning_rate": 0.0001, "elbo": 3.49063, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.130776448510439}
{"step": 87495, "loss": 3.4904243469238283, "grad_norm": 0.75355064868927, "learning_rate": 0.0001, "elbo": 3.49042, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.131069669247009}
{"step": 87500, "loss": 3.504591369628906, "grad_norm": 0.7168633341789246, "learning_rate": 0.0001, "elbo": 3.50459, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1313628899835795}
{"step": 87505, "loss": 3.494310760498047, "grad_norm": 0.7060973644256592, "learning_rate": 0.0001, "elbo": 3.49431, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.13165611072015}
{"step": 87510, "loss": 3.478474426269531, "grad_norm": 0.7247517704963684, "learning_rate": 0.0001, "elbo": 3.47847, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.13194933145672}
{"step": 87515, "loss": 3.491057586669922, "grad_norm": 0.7275717258453369, "learning_rate": 0.0001, "elbo": 3.49106, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.132242552193291}
{"step": 87520, "loss": 3.4828392028808595, "grad_norm": 0.7357333898544312, "learning_rate": 0.0001, "elbo": 3.48284, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.132535772929861}
{"step": 87525, "loss": 3.484322357177734, "grad_norm": 0.7439438700675964, "learning_rate": 0.0001, "elbo": 3.48432, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.132828993666432}
{"step": 87530, "loss": 3.5003101348876955, "grad_norm": 0.8053299188613892, "learning_rate": 0.0001, "elbo": 3.50031, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.133122214403002}
{"step": 87535, "loss": 3.495756912231445, "grad_norm": 0.7347097992897034, "learning_rate": 0.0001, "elbo": 3.49576, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.133415435139573}
{"step": 87540, "loss": 3.4873611450195314, "grad_norm": 0.7123584747314453, "learning_rate": 0.0001, "elbo": 3.48736, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.133708655876143}
{"step": 87545, "loss": 3.4888641357421877, "grad_norm": 0.8893691301345825, "learning_rate": 0.0001, "elbo": 3.48886, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1340018766127145}
{"step": 87550, "loss": 3.4759746551513673, "grad_norm": 0.7828158736228943, "learning_rate": 0.0001, "elbo": 3.47597, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.134295097349285}
{"step": 87555, "loss": 3.471248245239258, "grad_norm": 0.753452479839325, "learning_rate": 0.0001, "elbo": 3.47125, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.134588318085855}
{"step": 87560, "loss": 3.5071342468261717, "grad_norm": 0.7433784604072571, "learning_rate": 0.0001, "elbo": 3.50713, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.134881538822426}
{"step": 87565, "loss": 3.500208282470703, "grad_norm": 0.7007013559341431, "learning_rate": 0.0001, "elbo": 3.50021, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.135174759558996}
{"step": 87570, "loss": 3.489360809326172, "grad_norm": 0.7041525840759277, "learning_rate": 0.0001, "elbo": 3.48936, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.135467980295567}
{"step": 87575, "loss": 3.4753421783447265, "grad_norm": 0.7991258502006531, "learning_rate": 0.0001, "elbo": 3.47534, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.135761201032137}
{"step": 87580, "loss": 3.478700637817383, "grad_norm": 0.6918730139732361, "learning_rate": 0.0001, "elbo": 3.4787, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.136054421768708}
{"step": 87585, "loss": 3.490019607543945, "grad_norm": 0.7100299596786499, "learning_rate": 0.0001, "elbo": 3.49002, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.136347642505278}
{"step": 87590, "loss": 3.5017677307128907, "grad_norm": 0.7483601570129395, "learning_rate": 0.0001, "elbo": 3.50177, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1366408632418485}
{"step": 87595, "loss": 3.486585998535156, "grad_norm": 0.7137129306793213, "learning_rate": 0.0001, "elbo": 3.48659, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.136934083978419}
{"step": 87600, "loss": 3.499163055419922, "grad_norm": 0.683422327041626, "learning_rate": 0.0001, "elbo": 3.49916, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1372273047149895}
{"step": 87605, "loss": 3.4904964447021483, "grad_norm": 0.7492619156837463, "learning_rate": 0.0001, "elbo": 3.4905, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.13752052545156}
{"step": 87610, "loss": 3.4878807067871094, "grad_norm": 0.8349434733390808, "learning_rate": 0.0001, "elbo": 3.48788, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.13781374618813}
{"step": 87615, "loss": 3.4787933349609377, "grad_norm": 0.7328993678092957, "learning_rate": 0.0001, "elbo": 3.47879, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.138106966924701}
{"step": 87615, "loss": 3.4787933349609377, "grad_norm": 0.7328993678092957, "learning_rate": 0.0001, "elbo": 3.47879, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.138106966924701}
{"step": 87620, "loss": 3.490251159667969, "grad_norm": 0.7448720335960388, "learning_rate": 0.0001, "elbo": 3.49025, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.138400187661271}
{"step": 87625, "loss": 3.4946178436279296, "grad_norm": 0.7014850378036499, "learning_rate": 0.0001, "elbo": 3.49462, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.138693408397842}
{"step": 87625, "loss": 3.4946178436279296, "grad_norm": 0.7014850378036499, "learning_rate": 0.0001, "elbo": 3.49462, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.138693408397842}
{"step": 87630, "loss": 3.498050308227539, "grad_norm": 0.744259774684906, "learning_rate": 0.0001, "elbo": 3.49805, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.138986629134412}
{"step": 87635, "loss": 3.4967472076416017, "grad_norm": 0.7483105659484863, "learning_rate": 0.0001, "elbo": 3.49675, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.139279849870983}
{"step": 87635, "loss": 3.4967472076416017, "grad_norm": 0.7483105659484863, "learning_rate": 0.0001, "elbo": 3.49675, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.139279849870983}
{"step": 87640, "loss": 3.4875648498535154, "grad_norm": 0.7303162813186646, "learning_rate": 0.0001, "elbo": 3.48756, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.139573070607553}
{"step": 87645, "loss": 3.4857719421386717, "grad_norm": 0.7924672961235046, "learning_rate": 0.0001, "elbo": 3.48577, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1398662913441235}
{"step": 87650, "loss": 3.4941951751708986, "grad_norm": 0.7174649834632874, "learning_rate": 0.0001, "elbo": 3.4942, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.140159512080694}
{"step": 87655, "loss": 3.501361083984375, "grad_norm": 0.793372392654419, "learning_rate": 0.0001, "elbo": 3.50136, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1404527328172644}
{"step": 87660, "loss": 3.4853858947753906, "grad_norm": 0.7084168791770935, "learning_rate": 0.0001, "elbo": 3.48539, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.140745953553835}
{"step": 87665, "loss": 3.4934417724609377, "grad_norm": 0.7376020550727844, "learning_rate": 0.0001, "elbo": 3.49344, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.141039174290406}
{"step": 87670, "loss": 3.489214324951172, "grad_norm": 0.7462089657783508, "learning_rate": 0.0001, "elbo": 3.48921, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.141332395026977}
{"step": 87670, "loss": 3.489214324951172, "grad_norm": 0.7462089657783508, "learning_rate": 0.0001, "elbo": 3.48921, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.141332395026977}
{"step": 87675, "loss": 3.5090614318847657, "grad_norm": 0.8287941813468933, "learning_rate": 0.0001, "elbo": 3.50906, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.141625615763547}
{"step": 87680, "loss": 3.4870872497558594, "grad_norm": 0.8210650086402893, "learning_rate": 0.0001, "elbo": 3.48709, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.141918836500118}
{"step": 87680, "loss": 3.4870872497558594, "grad_norm": 0.8210650086402893, "learning_rate": 0.0001, "elbo": 3.48709, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.141918836500118}
{"step": 87685, "loss": 3.4915847778320312, "grad_norm": 0.6978834271430969, "learning_rate": 0.0001, "elbo": 3.49158, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.142212057236688}
{"step": 87690, "loss": 3.486285400390625, "grad_norm": 0.7455378174781799, "learning_rate": 0.0001, "elbo": 3.48629, "mask_frac": 0.5013, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1425052779732585}
{"step": 87695, "loss": 3.4829002380371095, "grad_norm": 0.7183319330215454, "learning_rate": 0.0001, "elbo": 3.4829, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.142798498709829}
{"step": 87700, "loss": 3.475980758666992, "grad_norm": 0.7868285179138184, "learning_rate": 0.0001, "elbo": 3.47598, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.143091719446399}
{"step": 87705, "loss": 3.487824249267578, "grad_norm": 0.7058730125427246, "learning_rate": 0.0001, "elbo": 3.48782, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.14338494018297}
{"step": 87710, "loss": 3.4954200744628907, "grad_norm": 0.7781455516815186, "learning_rate": 0.0001, "elbo": 3.49542, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.14367816091954}
{"step": 87715, "loss": 3.4912765502929686, "grad_norm": 0.7525991201400757, "learning_rate": 0.0001, "elbo": 3.49128, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.143971381656111}
{"step": 87720, "loss": 3.4986454010009767, "grad_norm": 0.71949702501297, "learning_rate": 0.0001, "elbo": 3.49865, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.144264602392681}
{"step": 87725, "loss": 3.49996337890625, "grad_norm": 0.702470064163208, "learning_rate": 0.0001, "elbo": 3.49996, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.144557823129252}
{"step": 87730, "loss": 3.4792545318603514, "grad_norm": 0.7083473801612854, "learning_rate": 0.0001, "elbo": 3.47925, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.144851043865822}
{"step": 87730, "loss": 3.4792545318603514, "grad_norm": 0.7083473801612854, "learning_rate": 0.0001, "elbo": 3.47925, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.144851043865822}
{"step": 87735, "loss": 3.4973224639892577, "grad_norm": 0.6924882531166077, "learning_rate": 0.0001, "elbo": 3.49732, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.145144264602393}
{"step": 87740, "loss": 3.4824161529541016, "grad_norm": 0.6847884058952332, "learning_rate": 0.0001, "elbo": 3.48242, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.145437485338963}
{"step": 87745, "loss": 3.4921051025390626, "grad_norm": 0.7428405284881592, "learning_rate": 0.0001, "elbo": 3.49211, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1457307060755335}
{"step": 87750, "loss": 3.483834075927734, "grad_norm": 0.7259519696235657, "learning_rate": 0.0001, "elbo": 3.48383, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.146023926812104}
{"step": 87755, "loss": 3.49359245300293, "grad_norm": 0.7140014171600342, "learning_rate": 0.0001, "elbo": 3.49359, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.146317147548674}
{"step": 87760, "loss": 3.480924606323242, "grad_norm": 0.7885544300079346, "learning_rate": 0.0001, "elbo": 3.48092, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.146610368285245}
{"step": 87765, "loss": 3.491310882568359, "grad_norm": 0.7201652526855469, "learning_rate": 0.0001, "elbo": 3.49131, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.146903589021815}
{"step": 87770, "loss": 3.475342559814453, "grad_norm": 0.783609926700592, "learning_rate": 0.0001, "elbo": 3.47534, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.147196809758386}
{"step": 87770, "loss": 3.475342559814453, "grad_norm": 0.783609926700592, "learning_rate": 0.0001, "elbo": 3.47534, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.147196809758386}
{"step": 87775, "loss": 3.4791507720947266, "grad_norm": 0.7741116881370544, "learning_rate": 0.0001, "elbo": 3.47915, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.147490030494956}
{"step": 87775, "loss": 3.4791507720947266, "grad_norm": 0.7741116881370544, "learning_rate": 0.0001, "elbo": 3.47915, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.147490030494956}
{"step": 87780, "loss": 3.5072772979736326, "grad_norm": 0.967717707157135, "learning_rate": 0.0001, "elbo": 3.50728, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.147783251231527}
{"step": 87780, "loss": 3.5072772979736326, "grad_norm": 0.967717707157135, "learning_rate": 0.0001, "elbo": 3.50728, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.147783251231527}
{"step": 87785, "loss": 3.491228485107422, "grad_norm": 0.7307957410812378, "learning_rate": 0.0001, "elbo": 3.49123, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.148076471968098}
{"step": 87790, "loss": 3.4847572326660154, "grad_norm": 0.8061869740486145, "learning_rate": 0.0001, "elbo": 3.48476, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1483696927046685}
{"step": 87795, "loss": 3.4852088928222655, "grad_norm": 0.7911774516105652, "learning_rate": 0.0001, "elbo": 3.48521, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.148662913441239}
{"step": 87800, "loss": 3.480641174316406, "grad_norm": 0.7912414073944092, "learning_rate": 0.0001, "elbo": 3.48064, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.148956134177809}
{"step": 87805, "loss": 3.4895973205566406, "grad_norm": 0.7131090760231018, "learning_rate": 0.0001, "elbo": 3.4896, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.14924935491438}
{"step": 87805, "loss": 3.4895973205566406, "grad_norm": 0.7131090760231018, "learning_rate": 0.0001, "elbo": 3.4896, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.14924935491438}
{"step": 87810, "loss": 3.481375503540039, "grad_norm": 0.7457010746002197, "learning_rate": 0.0001, "elbo": 3.48138, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.14954257565095}
{"step": 87815, "loss": 3.492264175415039, "grad_norm": 0.7580292820930481, "learning_rate": 0.0001, "elbo": 3.49226, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.149835796387521}
{"step": 87820, "loss": 3.4985443115234376, "grad_norm": 0.7197738289833069, "learning_rate": 0.0001, "elbo": 3.49854, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.150129017124091}
{"step": 87825, "loss": 3.492666244506836, "grad_norm": 0.7290742993354797, "learning_rate": 0.0001, "elbo": 3.49267, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.150422237860662}
{"step": 87825, "loss": 3.492666244506836, "grad_norm": 0.7290742993354797, "learning_rate": 0.0001, "elbo": 3.49267, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.150422237860662}
{"step": 87830, "loss": 3.4971473693847654, "grad_norm": 0.7002468705177307, "learning_rate": 0.0001, "elbo": 3.49715, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.150715458597232}
{"step": 87835, "loss": 3.492723846435547, "grad_norm": 1.6422069072723389, "learning_rate": 0.0001, "elbo": 3.49272, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.151008679333803}
{"step": 87840, "loss": 3.477625274658203, "grad_norm": 0.7583425641059875, "learning_rate": 0.0001, "elbo": 3.47763, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.151301900070373}
{"step": 87845, "loss": 3.4895530700683595, "grad_norm": 0.7477911114692688, "learning_rate": 0.0001, "elbo": 3.48955, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1515951208069435}
{"step": 87850, "loss": 3.479584503173828, "grad_norm": 0.7268543243408203, "learning_rate": 0.0001, "elbo": 3.47958, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.151888341543514}
{"step": 87855, "loss": 3.4946685791015626, "grad_norm": 0.7803806662559509, "learning_rate": 0.0001, "elbo": 3.49467, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.152181562280084}
{"step": 87855, "loss": 3.4946685791015626, "grad_norm": 0.7803806662559509, "learning_rate": 0.0001, "elbo": 3.49467, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.152181562280084}
{"step": 87860, "loss": 3.5013484954833984, "grad_norm": 0.7304154634475708, "learning_rate": 0.0001, "elbo": 3.50135, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.152474783016655}
{"step": 87865, "loss": 3.4941715240478515, "grad_norm": 0.738945722579956, "learning_rate": 0.0001, "elbo": 3.49417, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.152768003753225}
{"step": 87870, "loss": 3.4820266723632813, "grad_norm": 0.7820231318473816, "learning_rate": 0.0001, "elbo": 3.48203, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.153061224489796}
{"step": 87870, "loss": 3.4820266723632813, "grad_norm": 0.7820231318473816, "learning_rate": 0.0001, "elbo": 3.48203, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.153061224489796}
{"step": 87875, "loss": 3.4785633087158203, "grad_norm": 0.6880634427070618, "learning_rate": 0.0001, "elbo": 3.47856, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.153354445226366}
{"step": 87880, "loss": 3.4844627380371094, "grad_norm": 0.7653266191482544, "learning_rate": 0.0001, "elbo": 3.48446, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.153647665962937}
{"step": 87885, "loss": 3.476824188232422, "grad_norm": 0.7040687799453735, "learning_rate": 0.0001, "elbo": 3.47682, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.153940886699507}
{"step": 87890, "loss": 3.4752853393554686, "grad_norm": 0.7177233099937439, "learning_rate": 0.0001, "elbo": 3.47529, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.154234107436078}
{"step": 87895, "loss": 3.4883399963378907, "grad_norm": 0.7030388712882996, "learning_rate": 0.0001, "elbo": 3.48834, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.154527328172648}
{"step": 87900, "loss": 3.5173629760742187, "grad_norm": 0.7482062578201294, "learning_rate": 0.0001, "elbo": 3.51736, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1548205489092185}
{"step": 87900, "loss": 3.5173629760742187, "grad_norm": 0.7482062578201294, "learning_rate": 0.0001, "elbo": 3.51736, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1548205489092185}
{"step": 87905, "loss": 3.4949607849121094, "grad_norm": 1.1814130544662476, "learning_rate": 0.0001, "elbo": 3.49496, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.155113769645789}
{"step": 87910, "loss": 3.4917701721191405, "grad_norm": 0.732904314994812, "learning_rate": 0.0001, "elbo": 3.49177, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.15540699038236}
{"step": 87915, "loss": 3.5011455535888674, "grad_norm": 0.6956230998039246, "learning_rate": 0.0001, "elbo": 3.50115, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.155700211118931}
{"step": 87920, "loss": 3.492584228515625, "grad_norm": 1.1614246368408203, "learning_rate": 0.0001, "elbo": 3.49258, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.155993431855501}
{"step": 87925, "loss": 3.4919204711914062, "grad_norm": 0.7838830351829529, "learning_rate": 0.0001, "elbo": 3.49192, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.156286652592072}
{"step": 87930, "loss": 3.5104907989501952, "grad_norm": 0.701204776763916, "learning_rate": 0.0001, "elbo": 3.51049, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.156579873328642}
{"step": 87935, "loss": 3.5012752532958986, "grad_norm": 0.7650004029273987, "learning_rate": 0.0001, "elbo": 3.50128, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1568730940652125}
{"step": 87935, "loss": 3.5012752532958986, "grad_norm": 0.7650004029273987, "learning_rate": 0.0001, "elbo": 3.50128, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1568730940652125}
{"step": 87940, "loss": 3.492152786254883, "grad_norm": 0.7238743305206299, "learning_rate": 0.0001, "elbo": 3.49215, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.157166314801783}
{"step": 87945, "loss": 3.4864391326904296, "grad_norm": 0.7684135437011719, "learning_rate": 0.0001, "elbo": 3.48644, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1574595355383535}
{"step": 87950, "loss": 3.465628814697266, "grad_norm": 1.0540484189987183, "learning_rate": 0.0001, "elbo": 3.46563, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.157752756274924}
{"step": 87955, "loss": 3.5010971069335937, "grad_norm": 0.7796427607536316, "learning_rate": 0.0001, "elbo": 3.5011, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.158045977011494}
{"step": 87960, "loss": 3.470249557495117, "grad_norm": 0.767920970916748, "learning_rate": 0.0001, "elbo": 3.47025, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.158339197748065}
{"step": 87965, "loss": 3.4902313232421873, "grad_norm": 0.7123130559921265, "learning_rate": 0.0001, "elbo": 3.49023, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.158632418484635}
{"step": 87970, "loss": 3.487742614746094, "grad_norm": 0.6906401515007019, "learning_rate": 0.0001, "elbo": 3.48774, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.158925639221206}
{"step": 87975, "loss": 3.480281448364258, "grad_norm": 0.7287954092025757, "learning_rate": 0.0001, "elbo": 3.48028, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.159218859957776}
{"step": 87980, "loss": 3.4928855895996094, "grad_norm": 0.9016024470329285, "learning_rate": 0.0001, "elbo": 3.49289, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.159512080694347}
{"step": 87985, "loss": 3.497822570800781, "grad_norm": 0.7431044578552246, "learning_rate": 0.0001, "elbo": 3.49782, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.159805301430917}
{"step": 87990, "loss": 3.521485137939453, "grad_norm": 0.6998478770256042, "learning_rate": 0.0001, "elbo": 3.52149, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1600985221674875}
{"step": 87990, "loss": 3.521485137939453, "grad_norm": 0.6998478770256042, "learning_rate": 0.0001, "elbo": 3.52149, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1600985221674875}
{"step": 87995, "loss": 3.477656936645508, "grad_norm": 0.7608762979507446, "learning_rate": 0.0001, "elbo": 3.47766, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.160391742904058}
{"step": 88000, "loss": 3.499982452392578, "grad_norm": 0.804324746131897, "learning_rate": 0.0001, "elbo": 3.49998, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1606849636406285}
{"step": 88005, "loss": 3.4896053314208983, "grad_norm": 1.3465065956115723, "learning_rate": 0.0001, "elbo": 3.48961, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.160978184377199}
{"step": 88010, "loss": 3.512837219238281, "grad_norm": 0.7373413443565369, "learning_rate": 0.0001, "elbo": 3.51284, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.161271405113769}
{"step": 88015, "loss": 3.4912830352783204, "grad_norm": 0.8184685707092285, "learning_rate": 0.0001, "elbo": 3.49128, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.16156462585034}
{"step": 88020, "loss": 3.493977355957031, "grad_norm": 0.7240496277809143, "learning_rate": 0.0001, "elbo": 3.49398, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.16185784658691}
{"step": 88025, "loss": 3.4891292572021486, "grad_norm": 0.7830174565315247, "learning_rate": 0.0001, "elbo": 3.48913, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.162151067323481}
{"step": 88030, "loss": 3.495670700073242, "grad_norm": 0.7335917353630066, "learning_rate": 0.0001, "elbo": 3.49567, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.162444288060051}
{"step": 88035, "loss": 3.495291900634766, "grad_norm": 0.7378957867622375, "learning_rate": 0.0001, "elbo": 3.49529, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1627375087966225}
{"step": 88040, "loss": 3.4831024169921876, "grad_norm": 0.6691046357154846, "learning_rate": 0.0001, "elbo": 3.4831, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.163030729533193}
{"step": 88045, "loss": 3.499169921875, "grad_norm": 0.9234976768493652, "learning_rate": 0.0001, "elbo": 3.49917, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.163323950269763}
{"step": 88050, "loss": 3.475419616699219, "grad_norm": 0.7248950004577637, "learning_rate": 0.0001, "elbo": 3.47542, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.163617171006334}
{"step": 88055, "loss": 3.4941429138183593, "grad_norm": 0.6984317302703857, "learning_rate": 0.0001, "elbo": 3.49414, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.163910391742904}
{"step": 88060, "loss": 3.4908935546875, "grad_norm": 0.8883043527603149, "learning_rate": 0.0001, "elbo": 3.49089, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.164203612479475}
{"step": 88065, "loss": 3.487641143798828, "grad_norm": 0.7264083623886108, "learning_rate": 0.0001, "elbo": 3.48764, "mask_frac": 0.5014, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.164496833216045}
{"step": 88070, "loss": 3.496102142333984, "grad_norm": 0.8208597898483276, "learning_rate": 0.0001, "elbo": 3.4961, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.164790053952616}
{"step": 88075, "loss": 3.481365203857422, "grad_norm": 0.7697203159332275, "learning_rate": 0.0001, "elbo": 3.48136, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.165083274689186}
{"step": 88080, "loss": 3.4919403076171873, "grad_norm": 0.712603747844696, "learning_rate": 0.0001, "elbo": 3.49194, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.165376495425757}
{"step": 88085, "loss": 3.475737762451172, "grad_norm": 0.9215509295463562, "learning_rate": 0.0001, "elbo": 3.47574, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.165669716162327}
{"step": 88090, "loss": 3.480467987060547, "grad_norm": 0.7797982096672058, "learning_rate": 0.0001, "elbo": 3.48047, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1659629368988975}
{"step": 88095, "loss": 3.4883922576904296, "grad_norm": 0.7860404849052429, "learning_rate": 0.0001, "elbo": 3.48839, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.166256157635468}
{"step": 88100, "loss": 3.481029510498047, "grad_norm": 0.7202802300453186, "learning_rate": 0.0001, "elbo": 3.48103, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.166549378372038}
{"step": 88105, "loss": 3.4920345306396485, "grad_norm": 1.067226529121399, "learning_rate": 0.0001, "elbo": 3.49203, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.166842599108609}
{"step": 88110, "loss": 3.505406951904297, "grad_norm": 0.7194721698760986, "learning_rate": 0.0001, "elbo": 3.50541, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.167135819845179}
{"step": 88115, "loss": 3.507184600830078, "grad_norm": 0.7163767218589783, "learning_rate": 0.0001, "elbo": 3.50718, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.16742904058175}
{"step": 88120, "loss": 3.4868160247802735, "grad_norm": 0.7467969059944153, "learning_rate": 0.0001, "elbo": 3.48682, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.16772226131832}
{"step": 88125, "loss": 3.481214141845703, "grad_norm": 0.7342926263809204, "learning_rate": 0.0001, "elbo": 3.48121, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.168015482054891}
{"step": 88130, "loss": 3.50419921875, "grad_norm": 0.7514660954475403, "learning_rate": 0.0001, "elbo": 3.5042, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.168308702791461}
{"step": 88135, "loss": 3.4828720092773438, "grad_norm": 1.0242067575454712, "learning_rate": 0.0001, "elbo": 3.48287, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.168601923528032}
{"step": 88140, "loss": 3.484782409667969, "grad_norm": 0.7452001571655273, "learning_rate": 0.0001, "elbo": 3.48478, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.168895144264602}
{"step": 88145, "loss": 3.480426025390625, "grad_norm": 0.6815628409385681, "learning_rate": 0.0001, "elbo": 3.48043, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1691883650011725}
{"step": 88150, "loss": 3.4919448852539063, "grad_norm": 0.8388279676437378, "learning_rate": 0.0001, "elbo": 3.49194, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.169481585737743}
{"step": 88155, "loss": 3.4935150146484375, "grad_norm": 1.1700657606124878, "learning_rate": 0.0001, "elbo": 3.49352, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.169774806474313}
{"step": 88160, "loss": 3.504597473144531, "grad_norm": 0.6949927806854248, "learning_rate": 0.0001, "elbo": 3.5046, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.170068027210885}
{"step": 88165, "loss": 3.481523132324219, "grad_norm": 0.9587723612785339, "learning_rate": 0.0001, "elbo": 3.48152, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.170361247947455}
{"step": 88170, "loss": 3.493471145629883, "grad_norm": 0.729966402053833, "learning_rate": 0.0001, "elbo": 3.49347, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.170654468684026}
{"step": 88175, "loss": 3.5224334716796877, "grad_norm": 0.746892511844635, "learning_rate": 0.0001, "elbo": 3.52243, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.170947689420596}
{"step": 88180, "loss": 3.4897815704345705, "grad_norm": 0.712192952632904, "learning_rate": 0.0001, "elbo": 3.48978, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.171240910157167}
{"step": 88180, "loss": 3.4897815704345705, "grad_norm": 0.712192952632904, "learning_rate": 0.0001, "elbo": 3.48978, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.171240910157167}
{"step": 88185, "loss": 3.4977981567382814, "grad_norm": 0.7279927730560303, "learning_rate": 0.0001, "elbo": 3.4978, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.171534130893737}
{"step": 88190, "loss": 3.494916534423828, "grad_norm": 0.7138153314590454, "learning_rate": 0.0001, "elbo": 3.49492, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1718273516303075}
{"step": 88195, "loss": 3.4954490661621094, "grad_norm": 0.7392135262489319, "learning_rate": 0.0001, "elbo": 3.49545, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.172120572366878}
{"step": 88200, "loss": 3.489482879638672, "grad_norm": 0.7193393707275391, "learning_rate": 0.0001, "elbo": 3.48948, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.172413793103448}
{"step": 88205, "loss": 3.4766754150390624, "grad_norm": 0.6933830380439758, "learning_rate": 0.0001, "elbo": 3.47668, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.172707013840019}
{"step": 88210, "loss": 3.4724029541015624, "grad_norm": 0.718874990940094, "learning_rate": 0.0001, "elbo": 3.4724, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.173000234576589}
{"step": 88215, "loss": 3.4851428985595705, "grad_norm": 0.621567964553833, "learning_rate": 0.0001, "elbo": 3.48514, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.17329345531316}
{"step": 88220, "loss": 3.4941940307617188, "grad_norm": 0.7330536842346191, "learning_rate": 0.0001, "elbo": 3.49419, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.17358667604973}
{"step": 88225, "loss": 3.4892837524414064, "grad_norm": 0.7060071229934692, "learning_rate": 0.0001, "elbo": 3.48928, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.173879896786301}
{"step": 88230, "loss": 3.499552917480469, "grad_norm": 0.6939659714698792, "learning_rate": 0.0001, "elbo": 3.49955, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.174173117522871}
{"step": 88235, "loss": 3.4876121520996093, "grad_norm": 0.6808873414993286, "learning_rate": 0.0001, "elbo": 3.48761, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.174466338259442}
{"step": 88240, "loss": 3.483592987060547, "grad_norm": 1.130651831626892, "learning_rate": 0.0001, "elbo": 3.48359, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.174759558996012}
{"step": 88245, "loss": 3.4918357849121096, "grad_norm": 0.7926188707351685, "learning_rate": 0.0001, "elbo": 3.49184, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1750527797325825}
{"step": 88250, "loss": 3.4945613861083986, "grad_norm": 0.7479471564292908, "learning_rate": 0.0001, "elbo": 3.49456, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.175346000469153}
{"step": 88255, "loss": 3.467362976074219, "grad_norm": 0.6857585310935974, "learning_rate": 0.0001, "elbo": 3.46736, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.175639221205723}
{"step": 88260, "loss": 3.4886341094970703, "grad_norm": 0.691590428352356, "learning_rate": 0.0001, "elbo": 3.48863, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.175932441942294}
{"step": 88265, "loss": 3.484635925292969, "grad_norm": 0.9889287948608398, "learning_rate": 0.0001, "elbo": 3.48464, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.176225662678864}
{"step": 88270, "loss": 3.4884963989257813, "grad_norm": 0.7075562477111816, "learning_rate": 0.0001, "elbo": 3.4885, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.176518883415435}
{"step": 88275, "loss": 3.486603546142578, "grad_norm": 0.7635281085968018, "learning_rate": 0.0001, "elbo": 3.4866, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.176812104152005}
{"step": 88280, "loss": 3.4971435546875, "grad_norm": 0.8071328401565552, "learning_rate": 0.0001, "elbo": 3.49714, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.177105324888576}
{"step": 88285, "loss": 3.486085891723633, "grad_norm": 0.7510049939155579, "learning_rate": 0.0001, "elbo": 3.48609, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.177398545625147}
{"step": 88290, "loss": 3.4909431457519533, "grad_norm": 0.6741528511047363, "learning_rate": 0.0001, "elbo": 3.49094, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1776917663617175}
{"step": 88295, "loss": 3.499555206298828, "grad_norm": 0.7351251244544983, "learning_rate": 0.0001, "elbo": 3.49955, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.177984987098288}
{"step": 88300, "loss": 3.4923995971679687, "grad_norm": 0.6964091062545776, "learning_rate": 0.0001, "elbo": 3.4924, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.178278207834858}
{"step": 88305, "loss": 3.495409393310547, "grad_norm": 0.8062800765037537, "learning_rate": 0.0001, "elbo": 3.49541, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.178571428571429}
{"step": 88310, "loss": 3.4803672790527345, "grad_norm": 0.7442542314529419, "learning_rate": 0.0001, "elbo": 3.48037, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.178864649307999}
{"step": 88315, "loss": 3.49434814453125, "grad_norm": 0.784665048122406, "learning_rate": 0.0001, "elbo": 3.49435, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.17915787004457}
{"step": 88320, "loss": 3.487591552734375, "grad_norm": 0.68076491355896, "learning_rate": 0.0001, "elbo": 3.48759, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.17945109078114}
{"step": 88325, "loss": 3.4884315490722657, "grad_norm": 0.7132613658905029, "learning_rate": 0.0001, "elbo": 3.48843, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.179744311517711}
{"step": 88330, "loss": 3.48782958984375, "grad_norm": 0.7652518153190613, "learning_rate": 0.0001, "elbo": 3.48783, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.180037532254281}
{"step": 88335, "loss": 3.503146743774414, "grad_norm": 0.7909894585609436, "learning_rate": 0.0001, "elbo": 3.50315, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1803307529908516}
{"step": 88340, "loss": 3.4833351135253907, "grad_norm": 0.7079434394836426, "learning_rate": 0.0001, "elbo": 3.48333, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.180623973727422}
{"step": 88345, "loss": 3.4946517944335938, "grad_norm": 0.7415933012962341, "learning_rate": 0.0001, "elbo": 3.49465, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1809171944639925}
{"step": 88350, "loss": 3.4876682281494142, "grad_norm": 0.7961277365684509, "learning_rate": 0.0001, "elbo": 3.48767, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.181210415200563}
{"step": 88355, "loss": 3.4764610290527345, "grad_norm": 0.7198570966720581, "learning_rate": 0.0001, "elbo": 3.47646, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.181503635937133}
{"step": 88360, "loss": 3.486400604248047, "grad_norm": 0.8963508605957031, "learning_rate": 0.0001, "elbo": 3.4864, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.181796856673704}
{"step": 88365, "loss": 3.4883838653564454, "grad_norm": 0.743114709854126, "learning_rate": 0.0001, "elbo": 3.48838, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.182090077410274}
{"step": 88370, "loss": 3.4799148559570314, "grad_norm": 0.7439061403274536, "learning_rate": 0.0001, "elbo": 3.47992, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.182383298146845}
{"step": 88375, "loss": 3.4943389892578125, "grad_norm": 0.6933882236480713, "learning_rate": 0.0001, "elbo": 3.49434, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.182676518883415}
{"step": 88380, "loss": 3.4861083984375, "grad_norm": 0.7603376507759094, "learning_rate": 0.0001, "elbo": 3.48611, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.182969739619986}
{"step": 88385, "loss": 3.48427734375, "grad_norm": 0.8003337979316711, "learning_rate": 0.0001, "elbo": 3.48428, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.183262960356556}
{"step": 88385, "loss": 3.48427734375, "grad_norm": 0.8003337979316711, "learning_rate": 0.0001, "elbo": 3.48428, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.183262960356556}
{"step": 88390, "loss": 3.495869445800781, "grad_norm": 0.7820383906364441, "learning_rate": 0.0001, "elbo": 3.49587, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1835561810931265}
{"step": 88395, "loss": 3.4910350799560548, "grad_norm": 0.8397331833839417, "learning_rate": 0.0001, "elbo": 3.49104, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.183849401829697}
{"step": 88400, "loss": 3.48966064453125, "grad_norm": 0.7756268382072449, "learning_rate": 0.0001, "elbo": 3.48966, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.184142622566268}
{"step": 88405, "loss": 3.4971961975097656, "grad_norm": 0.8334983587265015, "learning_rate": 0.0001, "elbo": 3.4972, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.184435843302839}
{"step": 88410, "loss": 3.4869125366210936, "grad_norm": 0.7278161644935608, "learning_rate": 0.0001, "elbo": 3.48691, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.184729064039409}
{"step": 88415, "loss": 3.4940853118896484, "grad_norm": 0.722067654132843, "learning_rate": 0.0001, "elbo": 3.49409, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.18502228477598}
{"step": 88420, "loss": 3.4883853912353517, "grad_norm": 0.7072005271911621, "learning_rate": 0.0001, "elbo": 3.48839, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.18531550551255}
{"step": 88425, "loss": 3.4826629638671873, "grad_norm": 0.7218815684318542, "learning_rate": 0.0001, "elbo": 3.48266, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.185608726249121}
{"step": 88430, "loss": 3.485896682739258, "grad_norm": 0.7013407349586487, "learning_rate": 0.0001, "elbo": 3.4859, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.185901946985691}
{"step": 88435, "loss": 3.4902580261230467, "grad_norm": 0.7201808094978333, "learning_rate": 0.0001, "elbo": 3.49026, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1861951677222615}
{"step": 88440, "loss": 3.472267913818359, "grad_norm": 0.8683898448944092, "learning_rate": 0.0001, "elbo": 3.47227, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.186488388458832}
{"step": 88445, "loss": 3.4918338775634767, "grad_norm": 0.7442785501480103, "learning_rate": 0.0001, "elbo": 3.49183, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.186781609195402}
{"step": 88450, "loss": 3.504052734375, "grad_norm": 0.9160317778587341, "learning_rate": 0.0001, "elbo": 3.50405, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.187074829931973}
{"step": 88455, "loss": 3.4868701934814452, "grad_norm": 0.7130618095397949, "learning_rate": 0.0001, "elbo": 3.48687, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.187368050668543}
{"step": 88460, "loss": 3.4653907775878907, "grad_norm": 0.7592759132385254, "learning_rate": 0.0001, "elbo": 3.46539, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.187661271405114}
{"step": 88465, "loss": 3.479120635986328, "grad_norm": 0.7635151147842407, "learning_rate": 0.0001, "elbo": 3.47912, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.187954492141684}
{"step": 88470, "loss": 3.505327606201172, "grad_norm": 0.7088236212730408, "learning_rate": 0.0001, "elbo": 3.50533, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.188247712878255}
{"step": 88475, "loss": 3.5022369384765626, "grad_norm": 0.7360203862190247, "learning_rate": 0.0001, "elbo": 3.50224, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.188540933614825}
{"step": 88480, "loss": 3.495261383056641, "grad_norm": 0.7510446310043335, "learning_rate": 0.0001, "elbo": 3.49526, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.188834154351396}
{"step": 88485, "loss": 3.4834144592285154, "grad_norm": 0.6876512169837952, "learning_rate": 0.0001, "elbo": 3.48341, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.189127375087966}
{"step": 88490, "loss": 3.484722900390625, "grad_norm": 0.7150209546089172, "learning_rate": 0.0001, "elbo": 3.48472, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1894205958245365}
{"step": 88495, "loss": 3.4790687561035156, "grad_norm": 0.7635597586631775, "learning_rate": 0.0001, "elbo": 3.47907, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.189713816561107}
{"step": 88500, "loss": 3.494915008544922, "grad_norm": 0.7985550165176392, "learning_rate": 0.0001, "elbo": 3.49492, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.190007037297677}
{"step": 88500, "loss": 3.494915008544922, "grad_norm": 0.7985550165176392, "learning_rate": 0.0001, "elbo": 3.49492, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.190007037297677}
{"step": 88505, "loss": 3.4887344360351564, "grad_norm": 0.8342147469520569, "learning_rate": 0.0001, "elbo": 3.48873, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.190300258034248}
{"step": 88510, "loss": 3.473252868652344, "grad_norm": 0.7473628520965576, "learning_rate": 0.0001, "elbo": 3.47325, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.190593478770818}
{"step": 88515, "loss": 3.4669815063476563, "grad_norm": 0.747738778591156, "learning_rate": 0.0001, "elbo": 3.46698, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.190886699507389}
{"step": 88520, "loss": 3.4832229614257812, "grad_norm": 0.7609614729881287, "learning_rate": 0.0001, "elbo": 3.48322, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.191179920243959}
{"step": 88525, "loss": 3.4882743835449217, "grad_norm": 1.0783214569091797, "learning_rate": 0.0001, "elbo": 3.48827, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.191473140980531}
{"step": 88530, "loss": 3.5074920654296875, "grad_norm": 0.7074663043022156, "learning_rate": 0.0001, "elbo": 3.50749, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.191766361717101}
{"step": 88535, "loss": 3.487685775756836, "grad_norm": 0.7032899856567383, "learning_rate": 0.0001, "elbo": 3.48769, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1920595824536715}
{"step": 88540, "loss": 3.489190673828125, "grad_norm": 0.7358472347259521, "learning_rate": 0.0001, "elbo": 3.48919, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.192352803190242}
{"step": 88545, "loss": 3.490716552734375, "grad_norm": 0.6955780386924744, "learning_rate": 0.0001, "elbo": 3.49072, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.192646023926812}
{"step": 88550, "loss": 3.488438034057617, "grad_norm": 0.7021082639694214, "learning_rate": 0.0001, "elbo": 3.48844, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.192939244663383}
{"step": 88555, "loss": 3.4904853820800783, "grad_norm": 0.721947968006134, "learning_rate": 0.0001, "elbo": 3.49049, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.193232465399953}
{"step": 88560, "loss": 3.474135971069336, "grad_norm": 0.7352709174156189, "learning_rate": 0.0001, "elbo": 3.47414, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.193525686136524}
{"step": 88565, "loss": 3.5033164978027345, "grad_norm": 1.1139006614685059, "learning_rate": 0.0001, "elbo": 3.50332, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.193818906873094}
{"step": 88570, "loss": 3.4737174987792967, "grad_norm": 0.8472191095352173, "learning_rate": 0.0001, "elbo": 3.47372, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.194112127609665}
{"step": 88575, "loss": 3.487523651123047, "grad_norm": 0.7568939924240112, "learning_rate": 0.0001, "elbo": 3.48752, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.194405348346235}
{"step": 88580, "loss": 3.478061294555664, "grad_norm": 0.7079404592514038, "learning_rate": 0.0001, "elbo": 3.47806, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.194698569082806}
{"step": 88585, "loss": 3.4866451263427733, "grad_norm": 0.9886055588722229, "learning_rate": 0.0001, "elbo": 3.48665, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.194991789819376}
{"step": 88590, "loss": 3.4875442504882814, "grad_norm": 0.7550828456878662, "learning_rate": 0.0001, "elbo": 3.48754, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1952850105559465}
{"step": 88595, "loss": 3.480865478515625, "grad_norm": 0.6951066255569458, "learning_rate": 0.0001, "elbo": 3.48087, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.195578231292517}
{"step": 88600, "loss": 3.4869338989257814, "grad_norm": 0.8783013224601746, "learning_rate": 0.0001, "elbo": 3.48693, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.195871452029087}
{"step": 88605, "loss": 3.4928215026855467, "grad_norm": 0.8324143290519714, "learning_rate": 0.0001, "elbo": 3.49282, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.196164672765658}
{"step": 88610, "loss": 3.4933181762695313, "grad_norm": 0.7092589735984802, "learning_rate": 0.0001, "elbo": 3.49332, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.196457893502228}
{"step": 88615, "loss": 3.4982860565185545, "grad_norm": 0.7227151989936829, "learning_rate": 0.0001, "elbo": 3.49829, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.196751114238799}
{"step": 88620, "loss": 3.5013484954833984, "grad_norm": 0.6882478594779968, "learning_rate": 0.0001, "elbo": 3.50135, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.197044334975369}
{"step": 88625, "loss": 3.4921661376953126, "grad_norm": 0.7433182001113892, "learning_rate": 0.0001, "elbo": 3.49217, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.19733755571194}
{"step": 88630, "loss": 3.4718505859375, "grad_norm": 0.7323467135429382, "learning_rate": 0.0001, "elbo": 3.47185, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.19763077644851}
{"step": 88635, "loss": 3.4892738342285154, "grad_norm": 0.81647789478302, "learning_rate": 0.0001, "elbo": 3.48927, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.197923997185081}
{"step": 88640, "loss": 3.492672348022461, "grad_norm": 0.7376205325126648, "learning_rate": 0.0001, "elbo": 3.49267, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.198217217921651}
{"step": 88645, "loss": 3.4969921112060547, "grad_norm": 0.8730328679084778, "learning_rate": 0.0001, "elbo": 3.49699, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.1985104386582215}
{"step": 88650, "loss": 3.4931495666503904, "grad_norm": 0.7400062084197998, "learning_rate": 0.0001, "elbo": 3.49315, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.198803659394793}
{"step": 88655, "loss": 3.490239715576172, "grad_norm": 0.6965053081512451, "learning_rate": 0.0001, "elbo": 3.49024, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.199096880131363}
{"step": 88660, "loss": 3.4870956420898436, "grad_norm": 0.7445741295814514, "learning_rate": 0.0001, "elbo": 3.4871, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.199390100867934}
{"step": 88665, "loss": 3.481088638305664, "grad_norm": 0.6967968344688416, "learning_rate": 0.0001, "elbo": 3.48109, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.199683321604504}
{"step": 88670, "loss": 3.4815383911132813, "grad_norm": 0.7588348388671875, "learning_rate": 0.0001, "elbo": 3.48154, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.199976542341075}
{"step": 88675, "loss": 3.4932456970214845, "grad_norm": 0.7018738389015198, "learning_rate": 0.0001, "elbo": 3.49325, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.200269763077645}
{"step": 88680, "loss": 3.508693313598633, "grad_norm": 0.7968063950538635, "learning_rate": 0.0001, "elbo": 3.50869, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.200562983814216}
{"step": 88685, "loss": 3.5044898986816406, "grad_norm": 0.838539183139801, "learning_rate": 0.0001, "elbo": 3.50449, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.200856204550786}
{"step": 88690, "loss": 3.4834514617919923, "grad_norm": 0.7156838774681091, "learning_rate": 0.0001, "elbo": 3.48345, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2011494252873565}
{"step": 88695, "loss": 3.499104309082031, "grad_norm": 0.714008092880249, "learning_rate": 0.0001, "elbo": 3.4991, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.201442646023927}
{"step": 88700, "loss": 3.496430206298828, "grad_norm": 0.7552239298820496, "learning_rate": 0.0001, "elbo": 3.49643, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.201735866760497}
{"step": 88705, "loss": 3.483246612548828, "grad_norm": 0.6910672783851624, "learning_rate": 0.0001, "elbo": 3.48325, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.202029087497068}
{"step": 88710, "loss": 3.4863548278808594, "grad_norm": 0.737295925617218, "learning_rate": 0.0001, "elbo": 3.48635, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.202322308233638}
{"step": 88715, "loss": 3.4918121337890624, "grad_norm": 0.7170796394348145, "learning_rate": 0.0001, "elbo": 3.49181, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.202615528970209}
{"step": 88720, "loss": 3.4932743072509767, "grad_norm": 0.7214198112487793, "learning_rate": 0.0001, "elbo": 3.49327, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.202908749706779}
{"step": 88725, "loss": 3.4809593200683593, "grad_norm": 0.7374491095542908, "learning_rate": 0.0001, "elbo": 3.48096, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.20320197044335}
{"step": 88730, "loss": 3.4916053771972657, "grad_norm": 0.729350209236145, "learning_rate": 0.0001, "elbo": 3.49161, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.20349519117992}
{"step": 88735, "loss": 3.4942283630371094, "grad_norm": 0.7548763751983643, "learning_rate": 0.0001, "elbo": 3.49423, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2037884119164906}
{"step": 88740, "loss": 3.493144989013672, "grad_norm": 0.7183869481086731, "learning_rate": 0.0001, "elbo": 3.49315, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.204081632653061}
{"step": 88745, "loss": 3.4830650329589843, "grad_norm": 0.7184027433395386, "learning_rate": 0.0001, "elbo": 3.48306, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2043748533896315}
{"step": 88750, "loss": 3.4952640533447266, "grad_norm": 0.9055362939834595, "learning_rate": 0.0001, "elbo": 3.49526, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.204668074126202}
{"step": 88755, "loss": 3.4912017822265624, "grad_norm": 0.7173839807510376, "learning_rate": 0.0001, "elbo": 3.4912, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.204961294862772}
{"step": 88755, "loss": 3.4912017822265624, "grad_norm": 0.7173839807510376, "learning_rate": 0.0001, "elbo": 3.4912, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.204961294862772}
{"step": 88760, "loss": 3.4841598510742187, "grad_norm": 0.8375054597854614, "learning_rate": 0.0001, "elbo": 3.48416, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.205254515599343}
{"step": 88765, "loss": 3.4889785766601564, "grad_norm": 0.7177475094795227, "learning_rate": 0.0001, "elbo": 3.48898, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.205547736335913}
{"step": 88770, "loss": 3.4916866302490233, "grad_norm": 0.7182839512825012, "learning_rate": 0.0001, "elbo": 3.49169, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.205840957072484}
{"step": 88775, "loss": 3.486952209472656, "grad_norm": 0.770807147026062, "learning_rate": 0.0001, "elbo": 3.48695, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.206134177809055}
{"step": 88780, "loss": 3.482483673095703, "grad_norm": 0.7436372637748718, "learning_rate": 0.0001, "elbo": 3.48248, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2064273985456255}
{"step": 88785, "loss": 3.49279899597168, "grad_norm": 0.7092076539993286, "learning_rate": 0.0001, "elbo": 3.4928, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.206720619282196}
{"step": 88790, "loss": 3.4959503173828126, "grad_norm": 0.7452056407928467, "learning_rate": 0.0001, "elbo": 3.49595, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.207013840018766}
{"step": 88795, "loss": 3.4902370452880858, "grad_norm": 0.68855220079422, "learning_rate": 0.0001, "elbo": 3.49024, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.207307060755337}
{"step": 88800, "loss": 3.4940895080566405, "grad_norm": 0.7162001729011536, "learning_rate": 0.0001, "elbo": 3.49409, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.207600281491907}
{"step": 88805, "loss": 3.474091720581055, "grad_norm": 0.6927406191825867, "learning_rate": 0.0001, "elbo": 3.47409, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.207893502228478}
{"step": 88810, "loss": 3.493198013305664, "grad_norm": 0.7089134454727173, "learning_rate": 0.0001, "elbo": 3.4932, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.208186722965048}
{"step": 88815, "loss": 3.4895709991455077, "grad_norm": 0.7163823246955872, "learning_rate": 0.0001, "elbo": 3.48957, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.208479943701619}
{"step": 88820, "loss": 3.477947998046875, "grad_norm": 0.7023658156394958, "learning_rate": 0.0001, "elbo": 3.47795, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.208773164438189}
{"step": 88825, "loss": 3.495518112182617, "grad_norm": 0.8317610621452332, "learning_rate": 0.0001, "elbo": 3.49552, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.20906638517476}
{"step": 88830, "loss": 3.5001182556152344, "grad_norm": 0.7421348094940186, "learning_rate": 0.0001, "elbo": 3.50012, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.20935960591133}
{"step": 88835, "loss": 3.4691646575927733, "grad_norm": 0.7847150564193726, "learning_rate": 0.0001, "elbo": 3.46916, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2096528266479005}
{"step": 88840, "loss": 3.481642150878906, "grad_norm": 0.7161737084388733, "learning_rate": 0.0001, "elbo": 3.48164, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.209946047384471}
{"step": 88845, "loss": 3.4842735290527345, "grad_norm": 0.730981171131134, "learning_rate": 0.0001, "elbo": 3.48427, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.210239268121041}
{"step": 88850, "loss": 3.469279479980469, "grad_norm": 0.7410921454429626, "learning_rate": 0.0001, "elbo": 3.46928, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.210532488857612}
{"step": 88855, "loss": 3.485854721069336, "grad_norm": 0.7450056076049805, "learning_rate": 0.0001, "elbo": 3.48585, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.210825709594182}
{"step": 88860, "loss": 3.4786582946777345, "grad_norm": 0.734826922416687, "learning_rate": 0.0001, "elbo": 3.47866, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.211118930330753}
{"step": 88865, "loss": 3.498447036743164, "grad_norm": 0.6824775338172913, "learning_rate": 0.0001, "elbo": 3.49845, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.211412151067323}
{"step": 88870, "loss": 3.4765693664550783, "grad_norm": 0.8316807746887207, "learning_rate": 0.0001, "elbo": 3.47657, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.211705371803894}
{"step": 88875, "loss": 3.482901382446289, "grad_norm": 0.7583036422729492, "learning_rate": 0.0001, "elbo": 3.4829, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.211998592540464}
{"step": 88880, "loss": 3.4898422241210936, "grad_norm": 0.7420275211334229, "learning_rate": 0.0001, "elbo": 3.48984, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.212291813277035}
{"step": 88885, "loss": 3.489434814453125, "grad_norm": 0.7209271788597107, "learning_rate": 0.0001, "elbo": 3.48944, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.212585034013605}
{"step": 88890, "loss": 3.479500579833984, "grad_norm": 0.708794116973877, "learning_rate": 0.0001, "elbo": 3.4795, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2128782547501755}
{"step": 88895, "loss": 3.489480972290039, "grad_norm": 0.7191370725631714, "learning_rate": 0.0001, "elbo": 3.48948, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.213171475486746}
{"step": 88900, "loss": 3.475613021850586, "grad_norm": 0.6871649026870728, "learning_rate": 0.0001, "elbo": 3.47561, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.213464696223317}
{"step": 88905, "loss": 3.501982498168945, "grad_norm": 0.7021307349205017, "learning_rate": 0.0001, "elbo": 3.50198, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.213757916959888}
{"step": 88910, "loss": 3.4759029388427733, "grad_norm": 0.7299792170524597, "learning_rate": 0.0001, "elbo": 3.4759, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.214051137696458}
{"step": 88910, "loss": 3.4759029388427733, "grad_norm": 0.7299792170524597, "learning_rate": 0.0001, "elbo": 3.4759, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.214051137696458}
{"step": 88915, "loss": 3.485442352294922, "grad_norm": 0.7666834592819214, "learning_rate": 0.0001, "elbo": 3.48544, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.214344358433029}
{"step": 88915, "loss": 3.485442352294922, "grad_norm": 0.7666834592819214, "learning_rate": 0.0001, "elbo": 3.48544, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.214344358433029}
{"step": 88920, "loss": 3.5020111083984373, "grad_norm": 0.7513196468353271, "learning_rate": 0.0001, "elbo": 3.50201, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.214637579169599}
{"step": 88925, "loss": 3.5010684967041015, "grad_norm": 0.7093088626861572, "learning_rate": 0.0001, "elbo": 3.50107, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.21493079990617}
{"step": 88930, "loss": 3.4981735229492186, "grad_norm": 0.7782589793205261, "learning_rate": 0.0001, "elbo": 3.49817, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.21522402064274}
{"step": 88935, "loss": 3.488469696044922, "grad_norm": 0.7341352105140686, "learning_rate": 0.0001, "elbo": 3.48847, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2155172413793105}
{"step": 88940, "loss": 3.4808372497558593, "grad_norm": 0.7216683030128479, "learning_rate": 0.0001, "elbo": 3.48084, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.215810462115881}
{"step": 88945, "loss": 3.487895965576172, "grad_norm": 0.8281674981117249, "learning_rate": 0.0001, "elbo": 3.4879, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.216103682852451}
{"step": 88950, "loss": 3.477099609375, "grad_norm": 0.7998359799385071, "learning_rate": 0.0001, "elbo": 3.4771, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.216396903589022}
{"step": 88955, "loss": 3.4984054565429688, "grad_norm": 0.7252482771873474, "learning_rate": 0.0001, "elbo": 3.49841, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.216690124325592}
{"step": 88960, "loss": 3.495953369140625, "grad_norm": 0.6890013813972473, "learning_rate": 0.0001, "elbo": 3.49595, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.216983345062163}
{"step": 88965, "loss": 3.4659564971923826, "grad_norm": 0.7049955725669861, "learning_rate": 0.0001, "elbo": 3.46596, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.217276565798733}
{"step": 88970, "loss": 3.482130432128906, "grad_norm": 0.7288843393325806, "learning_rate": 0.0001, "elbo": 3.48213, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.217569786535304}
{"step": 88975, "loss": 3.483254623413086, "grad_norm": 0.7537149786949158, "learning_rate": 0.0001, "elbo": 3.48325, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.217863007271874}
{"step": 88980, "loss": 3.4856014251708984, "grad_norm": 0.6946264505386353, "learning_rate": 0.0001, "elbo": 3.4856, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.218156228008445}
{"step": 88985, "loss": 3.489038848876953, "grad_norm": 0.8582442998886108, "learning_rate": 0.0001, "elbo": 3.48904, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.218449448745015}
{"step": 88990, "loss": 3.471875, "grad_norm": 0.7484270334243774, "learning_rate": 0.0001, "elbo": 3.47187, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2187426694815855}
{"step": 88995, "loss": 3.4953598022460937, "grad_norm": 0.7108559012413025, "learning_rate": 0.0001, "elbo": 3.49536, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.219035890218156}
{"step": 89000, "loss": 3.4999465942382812, "grad_norm": 0.7072818279266357, "learning_rate": 0.0001, "elbo": 3.49995, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.219329110954726}
{"step": 89005, "loss": 3.481739044189453, "grad_norm": 0.8581820726394653, "learning_rate": 0.0001, "elbo": 3.48174, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.219622331691297}
{"step": 89010, "loss": 3.472591018676758, "grad_norm": 0.8606189489364624, "learning_rate": 0.0001, "elbo": 3.47259, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.219915552427867}
{"step": 89015, "loss": 3.485992431640625, "grad_norm": 0.7400376200675964, "learning_rate": 0.0001, "elbo": 3.48599, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.220208773164438}
{"step": 89020, "loss": 3.4986122131347654, "grad_norm": 0.8061652183532715, "learning_rate": 0.0001, "elbo": 3.49861, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.220501993901009}
{"step": 89025, "loss": 3.4906074523925783, "grad_norm": 0.8348166942596436, "learning_rate": 0.0001, "elbo": 3.49061, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.22079521463758}
{"step": 89030, "loss": 3.470720672607422, "grad_norm": 0.731716513633728, "learning_rate": 0.0001, "elbo": 3.47072, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.22108843537415}
{"step": 89035, "loss": 3.505759048461914, "grad_norm": 0.8243607878684998, "learning_rate": 0.0001, "elbo": 3.50576, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2213816561107205}
{"step": 89040, "loss": 3.5005111694335938, "grad_norm": 0.7747385501861572, "learning_rate": 0.0001, "elbo": 3.50051, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.221674876847291}
{"step": 89045, "loss": 3.4971046447753906, "grad_norm": 0.7182047963142395, "learning_rate": 0.0001, "elbo": 3.4971, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.221968097583861}
{"step": 89050, "loss": 3.482351303100586, "grad_norm": 0.8548445701599121, "learning_rate": 0.0001, "elbo": 3.48235, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.222261318320432}
{"step": 89055, "loss": 3.484262466430664, "grad_norm": 0.7402816414833069, "learning_rate": 0.0001, "elbo": 3.48426, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.222554539057002}
{"step": 89060, "loss": 3.4887363433837892, "grad_norm": 0.7360379695892334, "learning_rate": 0.0001, "elbo": 3.48874, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.222847759793573}
{"step": 89065, "loss": 3.4940494537353515, "grad_norm": 0.7546318173408508, "learning_rate": 0.0001, "elbo": 3.49405, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.223140980530143}
{"step": 89070, "loss": 3.484081268310547, "grad_norm": 0.7054810523986816, "learning_rate": 0.0001, "elbo": 3.48408, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.223434201266714}
{"step": 89075, "loss": 3.4832229614257812, "grad_norm": 0.7650458216667175, "learning_rate": 0.0001, "elbo": 3.48322, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.223727422003284}
{"step": 89080, "loss": 3.469365692138672, "grad_norm": 0.7104539275169373, "learning_rate": 0.0001, "elbo": 3.46937, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.224020642739855}
{"step": 89085, "loss": 3.481354522705078, "grad_norm": 0.695959210395813, "learning_rate": 0.0001, "elbo": 3.48135, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.224313863476425}
{"step": 89090, "loss": 3.4910327911376955, "grad_norm": 0.7398886680603027, "learning_rate": 0.0001, "elbo": 3.49103, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2246070842129955}
{"step": 89095, "loss": 3.4917293548583985, "grad_norm": 0.8888364434242249, "learning_rate": 0.0001, "elbo": 3.49173, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.224900304949566}
{"step": 89100, "loss": 3.4776901245117187, "grad_norm": 0.7257879376411438, "learning_rate": 0.0001, "elbo": 3.47769, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.225193525686136}
{"step": 89105, "loss": 3.478632354736328, "grad_norm": 0.809048056602478, "learning_rate": 0.0001, "elbo": 3.47863, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.225486746422707}
{"step": 89110, "loss": 3.484271240234375, "grad_norm": 0.7790243625640869, "learning_rate": 0.0001, "elbo": 3.48427, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.225779967159277}
{"step": 89115, "loss": 3.485612487792969, "grad_norm": 0.7061135172843933, "learning_rate": 0.0001, "elbo": 3.48561, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.226073187895848}
{"step": 89120, "loss": 3.495068359375, "grad_norm": 0.7463957667350769, "learning_rate": 0.0001, "elbo": 3.49507, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.226366408632418}
{"step": 89125, "loss": 3.492755889892578, "grad_norm": 0.7783929705619812, "learning_rate": 0.0001, "elbo": 3.49276, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.226659629368989}
{"step": 89130, "loss": 3.4863433837890625, "grad_norm": 0.8400404453277588, "learning_rate": 0.0001, "elbo": 3.48634, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.226952850105559}
{"step": 89135, "loss": 3.4887351989746094, "grad_norm": 0.7790741324424744, "learning_rate": 0.0001, "elbo": 3.48874, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2272460708421296}
{"step": 89140, "loss": 3.484030532836914, "grad_norm": 0.7349392175674438, "learning_rate": 0.0001, "elbo": 3.48403, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.227539291578701}
{"step": 89145, "loss": 3.5002037048339845, "grad_norm": 0.7234438061714172, "learning_rate": 0.0001, "elbo": 3.5002, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.227832512315271}
{"step": 89150, "loss": 3.4748668670654297, "grad_norm": 0.7157989144325256, "learning_rate": 0.0001, "elbo": 3.47487, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.228125733051842}
{"step": 89150, "loss": 3.4748668670654297, "grad_norm": 0.7157989144325256, "learning_rate": 0.0001, "elbo": 3.47487, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.228125733051842}
{"step": 89155, "loss": 3.4888710021972655, "grad_norm": 0.7821595072746277, "learning_rate": 0.0001, "elbo": 3.48887, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.228418953788412}
{"step": 89160, "loss": 3.477674865722656, "grad_norm": 0.8495057821273804, "learning_rate": 0.0001, "elbo": 3.47767, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.228712174524983}
{"step": 89165, "loss": 3.4830204010009767, "grad_norm": 0.8113957047462463, "learning_rate": 0.0001, "elbo": 3.48302, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.229005395261553}
{"step": 89170, "loss": 3.4851806640625, "grad_norm": 0.7493606805801392, "learning_rate": 0.0001, "elbo": 3.48518, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.229298615998124}
{"step": 89175, "loss": 3.475585174560547, "grad_norm": 1.0409741401672363, "learning_rate": 0.0001, "elbo": 3.47559, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.229591836734694}
{"step": 89180, "loss": 3.479256439208984, "grad_norm": 0.6946520209312439, "learning_rate": 0.0001, "elbo": 3.47926, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2298850574712645}
{"step": 89185, "loss": 3.483358383178711, "grad_norm": 0.7154380083084106, "learning_rate": 0.0001, "elbo": 3.48336, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.230178278207835}
{"step": 89190, "loss": 3.50965576171875, "grad_norm": 0.698441743850708, "learning_rate": 0.0001, "elbo": 3.50966, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.230471498944405}
{"step": 89195, "loss": 3.4851593017578124, "grad_norm": 0.68674236536026, "learning_rate": 0.0001, "elbo": 3.48516, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.230764719680976}
{"step": 89195, "loss": 3.4851593017578124, "grad_norm": 0.68674236536026, "learning_rate": 0.0001, "elbo": 3.48516, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.230764719680976}
{"step": 89200, "loss": 3.4823482513427733, "grad_norm": 0.69399493932724, "learning_rate": 0.0001, "elbo": 3.48235, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.231057940417546}
{"step": 89205, "loss": 3.477734375, "grad_norm": 0.6821098327636719, "learning_rate": 0.0001, "elbo": 3.47773, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.231351161154117}
{"step": 89210, "loss": 3.463720703125, "grad_norm": 0.7555060386657715, "learning_rate": 0.0001, "elbo": 3.46372, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.231644381890687}
{"step": 89215, "loss": 3.4886714935302736, "grad_norm": 0.7161851525306702, "learning_rate": 0.0001, "elbo": 3.48867, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.231937602627258}
{"step": 89220, "loss": 3.4770839691162108, "grad_norm": 0.7000055909156799, "learning_rate": 0.0001, "elbo": 3.47708, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.232230823363828}
{"step": 89225, "loss": 3.479894256591797, "grad_norm": 0.717190146446228, "learning_rate": 0.0001, "elbo": 3.47989, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.232524044100399}
{"step": 89230, "loss": 3.4947410583496095, "grad_norm": 0.730656087398529, "learning_rate": 0.0001, "elbo": 3.49474, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.232817264836969}
{"step": 89235, "loss": 3.4805313110351563, "grad_norm": 0.7352975606918335, "learning_rate": 0.0001, "elbo": 3.48053, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2331104855735395}
{"step": 89240, "loss": 3.4860267639160156, "grad_norm": 0.6943511366844177, "learning_rate": 0.0001, "elbo": 3.48603, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.23340370631011}
{"step": 89245, "loss": 3.484422302246094, "grad_norm": 0.7308266758918762, "learning_rate": 0.0001, "elbo": 3.48442, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.23369692704668}
{"step": 89250, "loss": 3.498871994018555, "grad_norm": 0.7023288011550903, "learning_rate": 0.0001, "elbo": 3.49887, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.233990147783251}
{"step": 89255, "loss": 3.4835578918457033, "grad_norm": 0.8098283410072327, "learning_rate": 0.0001, "elbo": 3.48356, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.234283368519821}
{"step": 89260, "loss": 3.4768226623535154, "grad_norm": 0.7505219578742981, "learning_rate": 0.0001, "elbo": 3.47682, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.234576589256392}
{"step": 89265, "loss": 3.4987377166748046, "grad_norm": 0.8059128522872925, "learning_rate": 0.0001, "elbo": 3.49874, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.234869809992963}
{"step": 89270, "loss": 3.471352767944336, "grad_norm": 0.6748939156532288, "learning_rate": 0.0001, "elbo": 3.47135, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.235163030729534}
{"step": 89275, "loss": 3.468058395385742, "grad_norm": 0.7353588342666626, "learning_rate": 0.0001, "elbo": 3.46806, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.235456251466104}
{"step": 89280, "loss": 3.4832908630371096, "grad_norm": 0.6782299876213074, "learning_rate": 0.0001, "elbo": 3.48329, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2357494722026745}
{"step": 89285, "loss": 3.4768775939941405, "grad_norm": 0.7083786725997925, "learning_rate": 0.0001, "elbo": 3.47688, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.236042692939245}
{"step": 89290, "loss": 3.4944992065429688, "grad_norm": 0.6950035691261292, "learning_rate": 0.0001, "elbo": 3.4945, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.236335913675815}
{"step": 89295, "loss": 3.4802982330322267, "grad_norm": 0.8342201113700867, "learning_rate": 0.0001, "elbo": 3.4803, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.236629134412386}
{"step": 89300, "loss": 3.4829769134521484, "grad_norm": 0.7041761875152588, "learning_rate": 0.0001, "elbo": 3.48298, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.236922355148956}
{"step": 89305, "loss": 3.4661529541015623, "grad_norm": 0.7685415744781494, "learning_rate": 0.0001, "elbo": 3.46615, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.237215575885527}
{"step": 89310, "loss": 3.4884193420410154, "grad_norm": 0.7875835299491882, "learning_rate": 0.0001, "elbo": 3.48842, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.237508796622097}
{"step": 89315, "loss": 3.4696460723876954, "grad_norm": 0.8381107449531555, "learning_rate": 0.0001, "elbo": 3.46965, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.237802017358668}
{"step": 89320, "loss": 3.4906219482421874, "grad_norm": 0.6964763402938843, "learning_rate": 0.0001, "elbo": 3.49062, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.238095238095238}
{"step": 89325, "loss": 3.475197601318359, "grad_norm": 0.778620183467865, "learning_rate": 0.0001, "elbo": 3.4752, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.238388458831809}
{"step": 89330, "loss": 3.4864044189453125, "grad_norm": 0.7218490839004517, "learning_rate": 0.0001, "elbo": 3.4864, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.238681679568379}
{"step": 89335, "loss": 3.4940261840820312, "grad_norm": 0.6985821723937988, "learning_rate": 0.0001, "elbo": 3.49403, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2389749003049495}
{"step": 89340, "loss": 3.485609436035156, "grad_norm": 0.7071012258529663, "learning_rate": 0.0001, "elbo": 3.48561, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.23926812104152}
{"step": 89345, "loss": 3.47803955078125, "grad_norm": 0.707990825176239, "learning_rate": 0.0001, "elbo": 3.47804, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.23956134177809}
{"step": 89350, "loss": 3.4900989532470703, "grad_norm": 0.7780344486236572, "learning_rate": 0.0001, "elbo": 3.4901, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.239854562514661}
{"step": 89355, "loss": 3.491150665283203, "grad_norm": 0.8499073386192322, "learning_rate": 0.0001, "elbo": 3.49115, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.240147783251231}
{"step": 89360, "loss": 3.4804714202880858, "grad_norm": 0.7729161381721497, "learning_rate": 0.0001, "elbo": 3.48047, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.240441003987802}
{"step": 89365, "loss": 3.503934860229492, "grad_norm": 0.7341286540031433, "learning_rate": 0.0001, "elbo": 3.50393, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.240734224724372}
{"step": 89365, "loss": 3.503934860229492, "grad_norm": 0.7341286540031433, "learning_rate": 0.0001, "elbo": 3.50393, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.240734224724372}
{"step": 89370, "loss": 3.4888076782226562, "grad_norm": 0.7787527441978455, "learning_rate": 0.0001, "elbo": 3.48881, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.241027445460943}
{"step": 89375, "loss": 3.492066192626953, "grad_norm": 0.7724950909614563, "learning_rate": 0.0001, "elbo": 3.49207, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.241320666197513}
{"step": 89380, "loss": 3.4984493255615234, "grad_norm": 0.7072004675865173, "learning_rate": 0.0001, "elbo": 3.49845, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.241613886934084}
{"step": 89385, "loss": 3.4985107421875, "grad_norm": 0.8413071632385254, "learning_rate": 0.0001, "elbo": 3.49851, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.241907107670654}
{"step": 89390, "loss": 3.4601478576660156, "grad_norm": 0.688258171081543, "learning_rate": 0.0001, "elbo": 3.46015, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.242200328407225}
{"step": 89395, "loss": 3.4956249237060546, "grad_norm": 0.7610358595848083, "learning_rate": 0.0001, "elbo": 3.49562, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.242493549143796}
{"step": 89400, "loss": 3.474359130859375, "grad_norm": 0.7270467877388, "learning_rate": 0.0001, "elbo": 3.47436, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.242786769880366}
{"step": 89400, "loss": 3.474359130859375, "grad_norm": 0.7270467877388, "learning_rate": 0.0001, "elbo": 3.47436, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.242786769880366}
{"step": 89405, "loss": 3.4611854553222656, "grad_norm": 0.7469112277030945, "learning_rate": 0.0001, "elbo": 3.46119, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.243079990616937}
{"step": 89410, "loss": 3.4811668395996094, "grad_norm": 1.1171259880065918, "learning_rate": 0.0001, "elbo": 3.48117, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.243373211353507}
{"step": 89415, "loss": 3.4877395629882812, "grad_norm": 0.6971688270568848, "learning_rate": 0.0001, "elbo": 3.48774, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.243666432090078}
{"step": 89420, "loss": 3.49017333984375, "grad_norm": 0.7928292155265808, "learning_rate": 0.0001, "elbo": 3.49017, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.243959652826648}
{"step": 89425, "loss": 3.4776195526123046, "grad_norm": 0.7514212727546692, "learning_rate": 0.0001, "elbo": 3.47762, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.244252873563219}
{"step": 89430, "loss": 3.500266265869141, "grad_norm": 0.7406506538391113, "learning_rate": 0.0001, "elbo": 3.50027, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.244546094299789}
{"step": 89435, "loss": 3.4911754608154295, "grad_norm": 0.7879250645637512, "learning_rate": 0.0001, "elbo": 3.49118, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2448393150363595}
{"step": 89440, "loss": 3.482069396972656, "grad_norm": 0.7170851826667786, "learning_rate": 0.0001, "elbo": 3.48207, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.24513253577293}
{"step": 89445, "loss": 3.499014663696289, "grad_norm": 0.8105412721633911, "learning_rate": 0.0001, "elbo": 3.49901, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2454257565095}
{"step": 89450, "loss": 3.498615264892578, "grad_norm": 0.8899857997894287, "learning_rate": 0.0001, "elbo": 3.49862, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.245718977246071}
{"step": 89455, "loss": 3.498628616333008, "grad_norm": 0.9675855040550232, "learning_rate": 0.0001, "elbo": 3.49863, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.246012197982641}
{"step": 89460, "loss": 3.4852226257324217, "grad_norm": 0.7356807589530945, "learning_rate": 0.0001, "elbo": 3.48522, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.246305418719212}
{"step": 89465, "loss": 3.483685302734375, "grad_norm": 0.7287020087242126, "learning_rate": 0.0001, "elbo": 3.48369, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.246598639455782}
{"step": 89470, "loss": 3.483699417114258, "grad_norm": 0.7185847759246826, "learning_rate": 0.0001, "elbo": 3.4837, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.246891860192353}
{"step": 89475, "loss": 3.5015087127685547, "grad_norm": 0.7843150496482849, "learning_rate": 0.0001, "elbo": 3.50151, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.247185080928923}
{"step": 89480, "loss": 3.4918563842773436, "grad_norm": 1.112884283065796, "learning_rate": 0.0001, "elbo": 3.49186, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.247478301665494}
{"step": 89485, "loss": 3.4786972045898437, "grad_norm": 0.709840714931488, "learning_rate": 0.0001, "elbo": 3.4787, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.247771522402064}
{"step": 89490, "loss": 3.488969421386719, "grad_norm": 0.7511613368988037, "learning_rate": 0.0001, "elbo": 3.48897, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2480647431386345}
{"step": 89495, "loss": 3.4861732482910157, "grad_norm": 0.7764087915420532, "learning_rate": 0.0001, "elbo": 3.48617, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.248357963875205}
{"step": 89500, "loss": 3.491857147216797, "grad_norm": 0.7188020944595337, "learning_rate": 0.0001, "elbo": 3.49186, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.248651184611775}
{"step": 89505, "loss": 3.494899368286133, "grad_norm": 0.6875815987586975, "learning_rate": 0.0001, "elbo": 3.4949, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.248944405348346}
{"step": 89510, "loss": 3.4955909729003904, "grad_norm": 0.7241668701171875, "learning_rate": 0.0001, "elbo": 3.49559, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.249237626084916}
{"step": 89515, "loss": 3.501629638671875, "grad_norm": 0.8161542415618896, "learning_rate": 0.0001, "elbo": 3.50163, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.249530846821488}
{"step": 89520, "loss": 3.4846466064453123, "grad_norm": 0.7315508127212524, "learning_rate": 0.0001, "elbo": 3.48465, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.249824067558058}
{"step": 89525, "loss": 3.472796630859375, "grad_norm": 0.7503687143325806, "learning_rate": 0.0001, "elbo": 3.4728, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2501172882946285}
{"step": 89530, "loss": 3.473341369628906, "grad_norm": 0.745326817035675, "learning_rate": 0.0001, "elbo": 3.47334, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.250410509031199}
{"step": 89535, "loss": 3.4916732788085936, "grad_norm": 0.7574915289878845, "learning_rate": 0.0001, "elbo": 3.49167, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2507037297677694}
{"step": 89540, "loss": 3.4854671478271486, "grad_norm": 0.7438552379608154, "learning_rate": 0.0001, "elbo": 3.48547, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.25099695050434}
{"step": 89545, "loss": 3.4929428100585938, "grad_norm": 0.7094183564186096, "learning_rate": 0.0001, "elbo": 3.49294, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.25129017124091}
{"step": 89550, "loss": 3.48133544921875, "grad_norm": 0.70762038230896, "learning_rate": 0.0001, "elbo": 3.48134, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.251583391977481}
{"step": 89555, "loss": 3.4971878051757814, "grad_norm": 0.7040427327156067, "learning_rate": 0.0001, "elbo": 3.49719, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.251876612714051}
{"step": 89560, "loss": 3.481882858276367, "grad_norm": 0.7489978671073914, "learning_rate": 0.0001, "elbo": 3.48188, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.252169833450622}
{"step": 89565, "loss": 3.484170913696289, "grad_norm": 0.7288111448287964, "learning_rate": 0.0001, "elbo": 3.48417, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.252463054187192}
{"step": 89570, "loss": 3.488893508911133, "grad_norm": 0.7123358845710754, "learning_rate": 0.0001, "elbo": 3.48889, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.252756274923763}
{"step": 89575, "loss": 3.489940643310547, "grad_norm": 0.7183284759521484, "learning_rate": 0.0001, "elbo": 3.48994, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.253049495660333}
{"step": 89580, "loss": 3.4943683624267576, "grad_norm": 0.7021887302398682, "learning_rate": 0.0001, "elbo": 3.49437, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2533427163969035}
{"step": 89585, "loss": 3.487689971923828, "grad_norm": 0.7061130404472351, "learning_rate": 0.0001, "elbo": 3.48769, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.253635937133474}
{"step": 89590, "loss": 3.495347595214844, "grad_norm": 0.7360421419143677, "learning_rate": 0.0001, "elbo": 3.49535, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.253929157870044}
{"step": 89595, "loss": 3.490355682373047, "grad_norm": 0.718112051486969, "learning_rate": 0.0001, "elbo": 3.49036, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.254222378606615}
{"step": 89600, "loss": 3.492203140258789, "grad_norm": 0.6771931052207947, "learning_rate": 0.0001, "elbo": 3.4922, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.254515599343185}
{"step": 89605, "loss": 3.4716812133789063, "grad_norm": 0.7353752255439758, "learning_rate": 0.0001, "elbo": 3.47168, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.254808820079756}
{"step": 89610, "loss": 3.4883804321289062, "grad_norm": 0.762610137462616, "learning_rate": 0.0001, "elbo": 3.48838, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.255102040816326}
{"step": 89615, "loss": 3.493022155761719, "grad_norm": 0.71682208776474, "learning_rate": 0.0001, "elbo": 3.49302, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.255395261552897}
{"step": 89620, "loss": 3.5041275024414062, "grad_norm": 0.7346985340118408, "learning_rate": 0.0001, "elbo": 3.50413, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.255688482289467}
{"step": 89625, "loss": 3.485281753540039, "grad_norm": 0.7017743587493896, "learning_rate": 0.0001, "elbo": 3.48528, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.255981703026038}
{"step": 89630, "loss": 3.480260467529297, "grad_norm": 0.7086749076843262, "learning_rate": 0.0001, "elbo": 3.48026, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.256274923762608}
{"step": 89635, "loss": 3.484429931640625, "grad_norm": 0.8223670125007629, "learning_rate": 0.0001, "elbo": 3.48443, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2565681444991785}
{"step": 89640, "loss": 3.460369873046875, "grad_norm": 0.8151262998580933, "learning_rate": 0.0001, "elbo": 3.46037, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.25686136523575}
{"step": 89645, "loss": 3.4921829223632814, "grad_norm": 0.763994038105011, "learning_rate": 0.0001, "elbo": 3.49218, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.25715458597232}
{"step": 89650, "loss": 3.4840057373046873, "grad_norm": 0.7754490971565247, "learning_rate": 0.0001, "elbo": 3.48401, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.257447806708891}
{"step": 89655, "loss": 3.4819915771484373, "grad_norm": 0.8989548683166504, "learning_rate": 0.0001, "elbo": 3.48199, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.257741027445461}
{"step": 89660, "loss": 3.494416046142578, "grad_norm": 0.7550678849220276, "learning_rate": 0.0001, "elbo": 3.49442, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.258034248182032}
{"step": 89665, "loss": 3.4794612884521485, "grad_norm": 0.7104001045227051, "learning_rate": 0.0001, "elbo": 3.47946, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.258327468918602}
{"step": 89670, "loss": 3.4993270874023437, "grad_norm": 0.7515554428100586, "learning_rate": 0.0001, "elbo": 3.49933, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.258620689655173}
{"step": 89675, "loss": 3.4987167358398437, "grad_norm": 0.7817099094390869, "learning_rate": 0.0001, "elbo": 3.49872, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.258913910391743}
{"step": 89680, "loss": 3.482594299316406, "grad_norm": 0.9101185202598572, "learning_rate": 0.0001, "elbo": 3.48259, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2592071311283135}
{"step": 89685, "loss": 3.510842132568359, "grad_norm": 0.740881621837616, "learning_rate": 0.0001, "elbo": 3.51084, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.259500351864884}
{"step": 89685, "loss": 3.510842132568359, "grad_norm": 0.740881621837616, "learning_rate": 0.0001, "elbo": 3.51084, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.259500351864884}
{"step": 89690, "loss": 3.4762645721435548, "grad_norm": 0.736642599105835, "learning_rate": 0.0001, "elbo": 3.47626, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.259793572601454}
{"step": 89695, "loss": 3.4742710113525392, "grad_norm": 0.7481658458709717, "learning_rate": 0.0001, "elbo": 3.47427, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.260086793338025}
{"step": 89700, "loss": 3.4928321838378906, "grad_norm": 0.7279084324836731, "learning_rate": 0.0001, "elbo": 3.49283, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.260380014074595}
{"step": 89705, "loss": 3.4914997100830076, "grad_norm": 0.7314580082893372, "learning_rate": 0.0001, "elbo": 3.4915, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.260673234811166}
{"step": 89705, "loss": 3.4914997100830076, "grad_norm": 0.7314580082893372, "learning_rate": 0.0001, "elbo": 3.4915, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.260673234811166}
{"step": 89710, "loss": 3.4900863647460936, "grad_norm": 0.790614128112793, "learning_rate": 0.0001, "elbo": 3.49009, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.260966455547736}
{"step": 89715, "loss": 3.4984367370605467, "grad_norm": 0.833634614944458, "learning_rate": 0.0001, "elbo": 3.49844, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.261259676284307}
{"step": 89720, "loss": 3.4973804473876955, "grad_norm": 0.8250428438186646, "learning_rate": 0.0001, "elbo": 3.49738, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.261552897020877}
{"step": 89725, "loss": 3.4915809631347656, "grad_norm": 0.9219537973403931, "learning_rate": 0.0001, "elbo": 3.49158, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.261846117757448}
{"step": 89730, "loss": 3.4852035522460936, "grad_norm": 0.7442205548286438, "learning_rate": 0.0001, "elbo": 3.4852, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.262139338494018}
{"step": 89730, "loss": 3.4852035522460936, "grad_norm": 0.7442205548286438, "learning_rate": 0.0001, "elbo": 3.4852, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.262139338494018}
{"step": 89735, "loss": 3.5015853881835937, "grad_norm": 0.6972111463546753, "learning_rate": 0.0001, "elbo": 3.50159, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2624325592305885}
{"step": 89740, "loss": 3.4832080841064452, "grad_norm": 0.7122790813446045, "learning_rate": 0.0001, "elbo": 3.48321, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.262725779967159}
{"step": 89745, "loss": 3.47601203918457, "grad_norm": 0.7316957116127014, "learning_rate": 0.0001, "elbo": 3.47601, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.263019000703729}
{"step": 89750, "loss": 3.491254425048828, "grad_norm": 0.7048377394676208, "learning_rate": 0.0001, "elbo": 3.49125, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2633122214403}
{"step": 89755, "loss": 3.4966506958007812, "grad_norm": 0.745135486125946, "learning_rate": 0.0001, "elbo": 3.49665, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.263605442176871}
{"step": 89760, "loss": 3.481719970703125, "grad_norm": 0.6965519189834595, "learning_rate": 0.0001, "elbo": 3.48172, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.263898662913441}
{"step": 89765, "loss": 3.4817955017089846, "grad_norm": 0.8943572640419006, "learning_rate": 0.0001, "elbo": 3.4818, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.264191883650012}
{"step": 89770, "loss": 3.48394775390625, "grad_norm": 0.769096314907074, "learning_rate": 0.0001, "elbo": 3.48395, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.264485104386583}
{"step": 89775, "loss": 3.478350067138672, "grad_norm": 0.7003143429756165, "learning_rate": 0.0001, "elbo": 3.47835, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.264778325123153}
{"step": 89780, "loss": 3.492567443847656, "grad_norm": 0.7814227342605591, "learning_rate": 0.0001, "elbo": 3.49257, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2650715458597235}
{"step": 89785, "loss": 3.4809860229492187, "grad_norm": 0.7274975776672363, "learning_rate": 0.0001, "elbo": 3.48099, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.265364766596294}
{"step": 89790, "loss": 3.489497756958008, "grad_norm": 0.7117355465888977, "learning_rate": 0.0001, "elbo": 3.4895, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.265657987332864}
{"step": 89795, "loss": 3.475847625732422, "grad_norm": 0.7890966534614563, "learning_rate": 0.0001, "elbo": 3.47585, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.265951208069435}
{"step": 89795, "loss": 3.475847625732422, "grad_norm": 0.7890966534614563, "learning_rate": 0.0001, "elbo": 3.47585, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.265951208069435}
{"step": 89800, "loss": 3.482268524169922, "grad_norm": 0.7459439635276794, "learning_rate": 0.0001, "elbo": 3.48227, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.266244428806005}
{"step": 89805, "loss": 3.4841041564941406, "grad_norm": 0.7028299570083618, "learning_rate": 0.0001, "elbo": 3.4841, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.266537649542576}
{"step": 89810, "loss": 3.493780517578125, "grad_norm": 0.7239704728126526, "learning_rate": 0.0001, "elbo": 3.49378, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.266830870279146}
{"step": 89810, "loss": 3.493780517578125, "grad_norm": 0.7239704728126526, "learning_rate": 0.0001, "elbo": 3.49378, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.266830870279146}
{"step": 89815, "loss": 3.487596130371094, "grad_norm": 0.7548820972442627, "learning_rate": 0.0001, "elbo": 3.4876, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.267124091015717}
{"step": 89815, "loss": 3.487596130371094, "grad_norm": 0.7548820972442627, "learning_rate": 0.0001, "elbo": 3.4876, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.267124091015717}
{"step": 89820, "loss": 3.483490753173828, "grad_norm": 0.7232978940010071, "learning_rate": 0.0001, "elbo": 3.48349, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.267417311752287}
{"step": 89825, "loss": 3.4870460510253904, "grad_norm": 0.8632264733314514, "learning_rate": 0.0001, "elbo": 3.48705, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.267710532488858}
{"step": 89830, "loss": 3.482829284667969, "grad_norm": 0.7094674110412598, "learning_rate": 0.0001, "elbo": 3.48283, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.268003753225428}
{"step": 89835, "loss": 3.4488494873046873, "grad_norm": 0.9825972318649292, "learning_rate": 0.0001, "elbo": 3.44885, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2682969739619985}
{"step": 89840, "loss": 3.4766250610351563, "grad_norm": 0.956436038017273, "learning_rate": 0.0001, "elbo": 3.47663, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.268590194698569}
{"step": 89845, "loss": 3.4846595764160155, "grad_norm": 0.7293418049812317, "learning_rate": 0.0001, "elbo": 3.48466, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.268883415435139}
{"step": 89850, "loss": 3.4877079010009764, "grad_norm": 0.6731114983558655, "learning_rate": 0.0001, "elbo": 3.48771, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.26917663617171}
{"step": 89855, "loss": 3.4730621337890626, "grad_norm": 0.6998454928398132, "learning_rate": 0.0001, "elbo": 3.47306, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.26946985690828}
{"step": 89860, "loss": 3.490587615966797, "grad_norm": 0.6990789771080017, "learning_rate": 0.0001, "elbo": 3.49059, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.269763077644851}
{"step": 89865, "loss": 3.4944538116455077, "grad_norm": 0.7433065176010132, "learning_rate": 0.0001, "elbo": 3.49445, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.270056298381421}
{"step": 89870, "loss": 3.483010101318359, "grad_norm": 0.7187254428863525, "learning_rate": 0.0001, "elbo": 3.48301, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.270349519117992}
{"step": 89870, "loss": 3.483010101318359, "grad_norm": 0.7187254428863525, "learning_rate": 0.0001, "elbo": 3.48301, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.270349519117992}
{"step": 89875, "loss": 3.492686462402344, "grad_norm": 0.7313640713691711, "learning_rate": 0.0001, "elbo": 3.49269, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.270642739854562}
{"step": 89880, "loss": 3.4812896728515623, "grad_norm": 0.6776775121688843, "learning_rate": 0.0001, "elbo": 3.48129, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2709359605911335}
{"step": 89885, "loss": 3.4796287536621096, "grad_norm": 0.9234492778778076, "learning_rate": 0.0001, "elbo": 3.47963, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.271229181327704}
{"step": 89885, "loss": 3.4796287536621096, "grad_norm": 0.9234492778778076, "learning_rate": 0.0001, "elbo": 3.47963, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.271229181327704}
{"step": 89890, "loss": 3.481916046142578, "grad_norm": 0.7380836009979248, "learning_rate": 0.0001, "elbo": 3.48192, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.271522402064274}
{"step": 89895, "loss": 3.4936256408691406, "grad_norm": 0.7455675601959229, "learning_rate": 0.0001, "elbo": 3.49363, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.271815622800845}
{"step": 89895, "loss": 3.4936256408691406, "grad_norm": 0.7455675601959229, "learning_rate": 0.0001, "elbo": 3.49363, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.271815622800845}
{"step": 89900, "loss": 3.492099380493164, "grad_norm": 0.9512071013450623, "learning_rate": 0.0001, "elbo": 3.4921, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.272108843537415}
{"step": 89905, "loss": 3.485517120361328, "grad_norm": 0.6958044767379761, "learning_rate": 0.0001, "elbo": 3.48552, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.272402064273986}
{"step": 89910, "loss": 3.485521697998047, "grad_norm": 0.7724314332008362, "learning_rate": 0.0001, "elbo": 3.48552, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.272695285010556}
{"step": 89915, "loss": 3.483800506591797, "grad_norm": 0.7916786670684814, "learning_rate": 0.0001, "elbo": 3.4838, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.272988505747127}
{"step": 89920, "loss": 3.4906631469726563, "grad_norm": 0.8110035061836243, "learning_rate": 0.0001, "elbo": 3.49066, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.273281726483697}
{"step": 89925, "loss": 3.481561279296875, "grad_norm": 0.7598167657852173, "learning_rate": 0.0001, "elbo": 3.48156, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2735749472202675}
{"step": 89930, "loss": 3.4917953491210936, "grad_norm": 0.7147415280342102, "learning_rate": 0.0001, "elbo": 3.4918, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.273868167956838}
{"step": 89935, "loss": 3.4921459197998046, "grad_norm": 0.7369426488876343, "learning_rate": 0.0001, "elbo": 3.49215, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2741613886934084}
{"step": 89940, "loss": 3.4924488067626953, "grad_norm": 0.7493991255760193, "learning_rate": 0.0001, "elbo": 3.49245, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.274454609429979}
{"step": 89945, "loss": 3.481196975708008, "grad_norm": 0.7153573036193848, "learning_rate": 0.0001, "elbo": 3.4812, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.274747830166549}
{"step": 89950, "loss": 3.487043762207031, "grad_norm": 0.7192988991737366, "learning_rate": 0.0001, "elbo": 3.48704, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.27504105090312}
{"step": 89955, "loss": 3.495378112792969, "grad_norm": 0.6961960196495056, "learning_rate": 0.0001, "elbo": 3.49538, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.27533427163969}
{"step": 89960, "loss": 3.491278076171875, "grad_norm": 0.7976527214050293, "learning_rate": 0.0001, "elbo": 3.49128, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.275627492376261}
{"step": 89965, "loss": 3.498047637939453, "grad_norm": 0.6822313666343689, "learning_rate": 0.0001, "elbo": 3.49805, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.275920713112831}
{"step": 89970, "loss": 3.496576690673828, "grad_norm": 0.7250905632972717, "learning_rate": 0.0001, "elbo": 3.49658, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.276213933849402}
{"step": 89975, "loss": 3.4787761688232424, "grad_norm": 0.9119343161582947, "learning_rate": 0.0001, "elbo": 3.47878, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.276507154585972}
{"step": 89980, "loss": 3.473688507080078, "grad_norm": 0.8086632490158081, "learning_rate": 0.0001, "elbo": 3.47369, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2768003753225425}
{"step": 89985, "loss": 3.4919620513916017, "grad_norm": 0.7030501961708069, "learning_rate": 0.0001, "elbo": 3.49196, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.277093596059113}
{"step": 89990, "loss": 3.490005874633789, "grad_norm": 0.9367402791976929, "learning_rate": 0.0001, "elbo": 3.49001, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.277386816795683}
{"step": 89995, "loss": 3.4970783233642577, "grad_norm": 0.726567804813385, "learning_rate": 0.0001, "elbo": 3.49708, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.277680037532254}
{"step": 90000, "loss": 3.4708683013916017, "grad_norm": 0.7582768797874451, "learning_rate": 0.0001, "elbo": 3.47087, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.277973258268824}
{"step": 90005, "loss": 3.490031433105469, "grad_norm": 0.7305746078491211, "learning_rate": 9.99999605215876e-05, "elbo": 3.49003, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.278266479005396}
{"step": 90005, "loss": 3.490031433105469, "grad_norm": 0.7305746078491211, "learning_rate": 9.99999605215876e-05, "elbo": 3.49003, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.278266479005396}
{"step": 90005, "loss": 3.490031433105469, "grad_norm": 0.7305746078491211, "learning_rate": 9.99999605215876e-05, "elbo": 3.49003, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.278266479005396}
{"step": 90010, "loss": 3.4896484375, "grad_norm": 0.7192012667655945, "learning_rate": 9.999980014064402e-05, "elbo": 3.48965, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.278559699741966}
{"step": 90015, "loss": 3.4859169006347654, "grad_norm": 0.7276408076286316, "learning_rate": 9.999951639016395e-05, "elbo": 3.48592, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.278852920478537}
{"step": 90020, "loss": 3.4721233367919924, "grad_norm": 0.678482711315155, "learning_rate": 9.999910927084749e-05, "elbo": 3.47212, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.279146141215107}
{"step": 90025, "loss": 3.490702819824219, "grad_norm": 0.7483022809028625, "learning_rate": 9.999857878369916e-05, "elbo": 3.4907, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2794393619516775}
{"step": 90030, "loss": 3.483135223388672, "grad_norm": 0.7227422595024109, "learning_rate": 9.999792493002788e-05, "elbo": 3.48314, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.279732582688248}
{"step": 90035, "loss": 3.4805526733398438, "grad_norm": 0.7812101244926453, "learning_rate": 9.999714771144701e-05, "elbo": 3.48055, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.280025803424818}
{"step": 90040, "loss": 3.4854583740234375, "grad_norm": 0.7054675221443176, "learning_rate": 9.99962471298742e-05, "elbo": 3.48546, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.280319024161389}
{"step": 90045, "loss": 3.4760810852050783, "grad_norm": 0.7177801728248596, "learning_rate": 9.99952231875316e-05, "elbo": 3.47608, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.280612244897959}
{"step": 90050, "loss": 3.498097610473633, "grad_norm": 0.7083173990249634, "learning_rate": 9.999407588694567e-05, "elbo": 3.4981, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.28090546563453}
{"step": 90055, "loss": 3.484662628173828, "grad_norm": 0.7213065028190613, "learning_rate": 9.999280523094724e-05, "elbo": 3.48466, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2811986863711}
{"step": 90060, "loss": 3.492448425292969, "grad_norm": 0.7212387323379517, "learning_rate": 9.999141122267154e-05, "elbo": 3.49245, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.281491907107671}
{"step": 90065, "loss": 3.499654006958008, "grad_norm": 0.7175865769386292, "learning_rate": 9.998989386555814e-05, "elbo": 3.49965, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.281785127844241}
{"step": 90070, "loss": 3.499369812011719, "grad_norm": 0.745571494102478, "learning_rate": 9.998825316335099e-05, "elbo": 3.49937, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.282078348580812}
{"step": 90075, "loss": 3.4727046966552733, "grad_norm": 0.7811420559883118, "learning_rate": 9.998648912009835e-05, "elbo": 3.4727, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.282371569317382}
{"step": 90080, "loss": 3.490917205810547, "grad_norm": 0.8903334140777588, "learning_rate": 9.99846017401528e-05, "elbo": 3.49092, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2826647900539525}
{"step": 90085, "loss": 3.495360565185547, "grad_norm": 0.8551634550094604, "learning_rate": 9.998259102817129e-05, "elbo": 3.49536, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.282958010790523}
{"step": 90090, "loss": 3.491851806640625, "grad_norm": 0.7229580283164978, "learning_rate": 9.998045698911505e-05, "elbo": 3.49185, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.283251231527093}
{"step": 90095, "loss": 3.489430236816406, "grad_norm": 0.7409956455230713, "learning_rate": 9.997819962824957e-05, "elbo": 3.48943, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.283544452263664}
{"step": 90100, "loss": 3.461384582519531, "grad_norm": 0.7563133835792542, "learning_rate": 9.997581895114472e-05, "elbo": 3.46138, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.283837673000234}
{"step": 90105, "loss": 3.4993011474609377, "grad_norm": 1.256303071975708, "learning_rate": 9.997331496367455e-05, "elbo": 3.4993, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.284130893736805}
{"step": 90110, "loss": 3.480792236328125, "grad_norm": 0.7175356149673462, "learning_rate": 9.99706876720174e-05, "elbo": 3.48079, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.284424114473375}
{"step": 90115, "loss": 3.4716781616210937, "grad_norm": 0.7122799158096313, "learning_rate": 9.996793708265586e-05, "elbo": 3.47168, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.284717335209946}
{"step": 90120, "loss": 3.468622589111328, "grad_norm": 0.7431893944740295, "learning_rate": 9.996506320237675e-05, "elbo": 3.46862, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.285010555946516}
{"step": 90125, "loss": 3.488331985473633, "grad_norm": 0.7359940409660339, "learning_rate": 9.996206603827105e-05, "elbo": 3.48833, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.285303776683087}
{"step": 90130, "loss": 3.4895614624023437, "grad_norm": 0.7460118532180786, "learning_rate": 9.9958945597734e-05, "elbo": 3.48956, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.285596997419658}
{"step": 90130, "loss": 3.4895614624023437, "grad_norm": 0.7460118532180786, "learning_rate": 9.9958945597734e-05, "elbo": 3.48956, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.285596997419658}
{"step": 90135, "loss": 3.482025146484375, "grad_norm": 0.9129818081855774, "learning_rate": 9.995570188846495e-05, "elbo": 3.48203, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.285890218156228}
{"step": 90140, "loss": 3.482318115234375, "grad_norm": 0.7015721797943115, "learning_rate": 9.995233491846744e-05, "elbo": 3.48232, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.286183438892799}
{"step": 90140, "loss": 3.482318115234375, "grad_norm": 0.7015721797943115, "learning_rate": 9.995233491846744e-05, "elbo": 3.48232, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.286183438892799}
{"step": 90145, "loss": 3.500438690185547, "grad_norm": 0.6803033947944641, "learning_rate": 9.994884469604912e-05, "elbo": 3.50044, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.286476659629369}
{"step": 90150, "loss": 3.4814476013183593, "grad_norm": 0.9061625003814697, "learning_rate": 9.99452312298218e-05, "elbo": 3.48145, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.28676988036594}
{"step": 90155, "loss": 3.490281677246094, "grad_norm": 0.7780604362487793, "learning_rate": 9.994149452870133e-05, "elbo": 3.49028, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.28706310110251}
{"step": 90160, "loss": 3.4840206146240233, "grad_norm": 0.7173513770103455, "learning_rate": 9.993763460190763e-05, "elbo": 3.48402, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.287356321839081}
{"step": 90165, "loss": 3.5046916961669923, "grad_norm": 0.736527144908905, "learning_rate": 9.993365145896473e-05, "elbo": 3.50469, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.287649542575651}
{"step": 90170, "loss": 3.493288040161133, "grad_norm": 0.7053917646408081, "learning_rate": 9.99295451097006e-05, "elbo": 3.49329, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.287942763312222}
{"step": 90175, "loss": 3.4769313812255858, "grad_norm": 1.1662696599960327, "learning_rate": 9.992531556424726e-05, "elbo": 3.47693, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.288235984048792}
{"step": 90180, "loss": 3.483320617675781, "grad_norm": 0.6934157013893127, "learning_rate": 9.992096283304069e-05, "elbo": 3.48332, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2885292047853625}
{"step": 90185, "loss": 3.4953479766845703, "grad_norm": 0.7667604088783264, "learning_rate": 9.991648692682083e-05, "elbo": 3.49535, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.288822425521933}
{"step": 90190, "loss": 3.4927162170410155, "grad_norm": 0.7707346081733704, "learning_rate": 9.991188785663153e-05, "elbo": 3.49272, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.289115646258503}
{"step": 90195, "loss": 3.476532745361328, "grad_norm": 0.7611656785011292, "learning_rate": 9.990716563382055e-05, "elbo": 3.47653, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.289408866995074}
{"step": 90200, "loss": 3.4942703247070312, "grad_norm": 0.8022636771202087, "learning_rate": 9.990232027003947e-05, "elbo": 3.49427, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.289702087731644}
{"step": 90200, "loss": 3.4942703247070312, "grad_norm": 0.8022636771202087, "learning_rate": 9.990232027003947e-05, "elbo": 3.49427, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.289702087731644}
{"step": 90205, "loss": 3.475302886962891, "grad_norm": 0.7433667778968811, "learning_rate": 9.989735177724378e-05, "elbo": 3.4753, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.289995308468215}
{"step": 90205, "loss": 3.475302886962891, "grad_norm": 0.7433667778968811, "learning_rate": 9.989735177724378e-05, "elbo": 3.4753, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.289995308468215}
{"step": 90210, "loss": 3.472156524658203, "grad_norm": 1.211226224899292, "learning_rate": 9.989226016769273e-05, "elbo": 3.47216, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.290288529204785}
{"step": 90215, "loss": 3.485613250732422, "grad_norm": 0.7867431044578552, "learning_rate": 9.988704545394936e-05, "elbo": 3.48561, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.290581749941356}
{"step": 90220, "loss": 3.479473876953125, "grad_norm": 0.8081035017967224, "learning_rate": 9.988170764888045e-05, "elbo": 3.47947, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.290874970677926}
{"step": 90225, "loss": 3.479325866699219, "grad_norm": 0.7031770944595337, "learning_rate": 9.987624676565652e-05, "elbo": 3.47933, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.291168191414497}
{"step": 90230, "loss": 3.4883941650390624, "grad_norm": 0.7680780291557312, "learning_rate": 9.987066281775174e-05, "elbo": 3.48839, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.291461412151067}
{"step": 90235, "loss": 3.4965530395507813, "grad_norm": 0.7584003806114197, "learning_rate": 9.986495581894395e-05, "elbo": 3.49655, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2917546328876375}
{"step": 90240, "loss": 3.482160949707031, "grad_norm": 1.038033366203308, "learning_rate": 9.985912578331463e-05, "elbo": 3.48216, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.292047853624208}
{"step": 90245, "loss": 3.4895057678222656, "grad_norm": 0.749352753162384, "learning_rate": 9.985317272524876e-05, "elbo": 3.48951, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.292341074360778}
{"step": 90250, "loss": 3.484170913696289, "grad_norm": 0.7174981832504272, "learning_rate": 9.984709665943494e-05, "elbo": 3.48417, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.292634295097349}
{"step": 90255, "loss": 3.5004035949707033, "grad_norm": 0.7817439436912537, "learning_rate": 9.98408976008653e-05, "elbo": 3.5004, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.29292751583392}
{"step": 90260, "loss": 3.494402313232422, "grad_norm": 1.1019257307052612, "learning_rate": 9.983457556483536e-05, "elbo": 3.4944, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.293220736570491}
{"step": 90265, "loss": 3.497655487060547, "grad_norm": 0.7928273677825928, "learning_rate": 9.982813056694412e-05, "elbo": 3.49766, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.293513957307061}
{"step": 90270, "loss": 3.4994575500488283, "grad_norm": 0.7479313015937805, "learning_rate": 9.982156262309395e-05, "elbo": 3.49946, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2938071780436315}
{"step": 90275, "loss": 3.494689178466797, "grad_norm": 0.7305800318717957, "learning_rate": 9.981487174949065e-05, "elbo": 3.49469, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.294100398780202}
{"step": 90280, "loss": 3.4913871765136717, "grad_norm": 0.7420605421066284, "learning_rate": 9.980805796264325e-05, "elbo": 3.49139, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2943936195167725}
{"step": 90285, "loss": 3.4823982238769533, "grad_norm": 0.7305528521537781, "learning_rate": 9.980112127936409e-05, "elbo": 3.4824, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.294686840253343}
{"step": 90290, "loss": 3.485626983642578, "grad_norm": 0.7140975594520569, "learning_rate": 9.979406171676877e-05, "elbo": 3.48563, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.294980060989913}
{"step": 90295, "loss": 3.4737472534179688, "grad_norm": 0.7794461846351624, "learning_rate": 9.978687929227606e-05, "elbo": 3.47375, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.295273281726484}
{"step": 90295, "loss": 3.4737472534179688, "grad_norm": 0.7794461846351624, "learning_rate": 9.978687929227606e-05, "elbo": 3.47375, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.295273281726484}
{"step": 90300, "loss": 3.481768035888672, "grad_norm": 0.6964794993400574, "learning_rate": 9.977957402360783e-05, "elbo": 3.48177, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.295566502463054}
{"step": 90305, "loss": 3.4840713500976563, "grad_norm": 0.7538737058639526, "learning_rate": 9.977214592878916e-05, "elbo": 3.48407, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.295859723199625}
{"step": 90310, "loss": 3.4756282806396483, "grad_norm": 0.7162377834320068, "learning_rate": 9.976459502614811e-05, "elbo": 3.47563, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.296152943936195}
{"step": 90315, "loss": 3.4992401123046877, "grad_norm": 0.8201991319656372, "learning_rate": 9.975692133431579e-05, "elbo": 3.49924, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.296446164672766}
{"step": 90320, "loss": 3.4891002655029295, "grad_norm": 0.6875159740447998, "learning_rate": 9.974912487222626e-05, "elbo": 3.4891, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.296739385409336}
{"step": 90325, "loss": 3.511974334716797, "grad_norm": 0.714244544506073, "learning_rate": 9.974120565911652e-05, "elbo": 3.51197, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2970326061459065}
{"step": 90330, "loss": 3.4708793640136717, "grad_norm": 0.7257257103919983, "learning_rate": 9.973316371452646e-05, "elbo": 3.47088, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.297325826882477}
{"step": 90335, "loss": 3.483567810058594, "grad_norm": 0.7177184820175171, "learning_rate": 9.972499905829875e-05, "elbo": 3.48357, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2976190476190474}
{"step": 90340, "loss": 3.484735870361328, "grad_norm": 0.7213311791419983, "learning_rate": 9.971671171057886e-05, "elbo": 3.48474, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.297912268355618}
{"step": 90345, "loss": 3.489639663696289, "grad_norm": 0.7114747762680054, "learning_rate": 9.970830169181505e-05, "elbo": 3.48964, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.298205489092188}
{"step": 90350, "loss": 3.4945030212402344, "grad_norm": 0.7260919213294983, "learning_rate": 9.969976902275815e-05, "elbo": 3.4945, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.298498709828759}
{"step": 90355, "loss": 3.4746063232421873, "grad_norm": 0.7139782309532166, "learning_rate": 9.969111372446171e-05, "elbo": 3.47461, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.298791930565329}
{"step": 90360, "loss": 3.454998779296875, "grad_norm": 0.7111664414405823, "learning_rate": 9.968233581828179e-05, "elbo": 3.455, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2990851513019}
{"step": 90365, "loss": 3.478647232055664, "grad_norm": 0.7191482186317444, "learning_rate": 9.967343532587702e-05, "elbo": 3.47865, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.29937837203847}
{"step": 90370, "loss": 3.482136535644531, "grad_norm": 0.6784351468086243, "learning_rate": 9.966441226920848e-05, "elbo": 3.48214, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.2996715927750415}
{"step": 90375, "loss": 3.4913494110107424, "grad_norm": 0.8071979284286499, "learning_rate": 9.965526667053963e-05, "elbo": 3.49135, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.299964813511611}
{"step": 90380, "loss": 3.4812164306640625, "grad_norm": 0.6976894736289978, "learning_rate": 9.964599855243639e-05, "elbo": 3.48122, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.300258034248182}
{"step": 90385, "loss": 3.483031463623047, "grad_norm": 0.7566171884536743, "learning_rate": 9.963660793776688e-05, "elbo": 3.48303, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.300551254984753}
{"step": 90390, "loss": 3.488486480712891, "grad_norm": 0.7111132740974426, "learning_rate": 9.96270948497015e-05, "elbo": 3.48849, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.300844475721323}
{"step": 90395, "loss": 3.485212707519531, "grad_norm": 0.757370114326477, "learning_rate": 9.961745931171287e-05, "elbo": 3.48521, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.301137696457894}
{"step": 90400, "loss": 3.491407012939453, "grad_norm": 0.7233338356018066, "learning_rate": 9.960770134757572e-05, "elbo": 3.49141, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.301430917194464}
{"step": 90405, "loss": 3.491364669799805, "grad_norm": 0.7438026666641235, "learning_rate": 9.959782098136683e-05, "elbo": 3.49136, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.301724137931035}
{"step": 90410, "loss": 3.477002716064453, "grad_norm": 0.8116443157196045, "learning_rate": 9.958781823746505e-05, "elbo": 3.477, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.302017358667605}
{"step": 90410, "loss": 3.477002716064453, "grad_norm": 0.8116443157196045, "learning_rate": 9.958781823746505e-05, "elbo": 3.477, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.302017358667605}
{"step": 90415, "loss": 3.4897254943847655, "grad_norm": 0.7050266265869141, "learning_rate": 9.957769314055117e-05, "elbo": 3.48973, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.302310579404176}
{"step": 90420, "loss": 3.483795928955078, "grad_norm": 0.8339845538139343, "learning_rate": 9.95674457156078e-05, "elbo": 3.4838, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.302603800140746}
{"step": 90425, "loss": 3.4942073822021484, "grad_norm": 0.7230216860771179, "learning_rate": 9.955707598791952e-05, "elbo": 3.49421, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3028970208773165}
{"step": 90430, "loss": 3.5009002685546875, "grad_norm": 0.7644352912902832, "learning_rate": 9.954658398307255e-05, "elbo": 3.5009, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.303190241613887}
{"step": 90435, "loss": 3.484571838378906, "grad_norm": 0.7765552401542664, "learning_rate": 9.953596972695487e-05, "elbo": 3.48457, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.303483462350457}
{"step": 90440, "loss": 3.495220947265625, "grad_norm": 0.6857165098190308, "learning_rate": 9.952523324575613e-05, "elbo": 3.49522, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.303776683087028}
{"step": 90445, "loss": 3.4683876037597656, "grad_norm": 0.7030212879180908, "learning_rate": 9.95143745659675e-05, "elbo": 3.46839, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.304069903823598}
{"step": 90445, "loss": 3.4683876037597656, "grad_norm": 0.7030212879180908, "learning_rate": 9.95143745659675e-05, "elbo": 3.46839, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.304069903823598}
{"step": 90450, "loss": 3.4938087463378906, "grad_norm": 0.7321531772613525, "learning_rate": 9.950339371438172e-05, "elbo": 3.49381, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.304363124560169}
{"step": 90450, "loss": 3.4938087463378906, "grad_norm": 0.7321531772613525, "learning_rate": 9.950339371438172e-05, "elbo": 3.49381, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.304363124560169}
{"step": 90455, "loss": 3.4883819580078126, "grad_norm": 0.7092421650886536, "learning_rate": 9.949229071809293e-05, "elbo": 3.48838, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.304656345296739}
{"step": 90460, "loss": 3.4781288146972655, "grad_norm": 0.7999354600906372, "learning_rate": 9.948106560449668e-05, "elbo": 3.47813, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.30494956603331}
{"step": 90465, "loss": 3.4898948669433594, "grad_norm": 0.6936834454536438, "learning_rate": 9.946971840128981e-05, "elbo": 3.48989, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.30524278676988}
{"step": 90470, "loss": 3.495256805419922, "grad_norm": 0.7308788299560547, "learning_rate": 9.945824913647044e-05, "elbo": 3.49526, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.305536007506451}
{"step": 90475, "loss": 3.4837913513183594, "grad_norm": 0.7504755854606628, "learning_rate": 9.944665783833782e-05, "elbo": 3.48379, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.305829228243021}
{"step": 90480, "loss": 3.492047882080078, "grad_norm": 0.7155960202217102, "learning_rate": 9.943494453549231e-05, "elbo": 3.49205, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3061224489795915}
{"step": 90485, "loss": 3.4797515869140625, "grad_norm": 0.763738214969635, "learning_rate": 9.942310925683538e-05, "elbo": 3.47975, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.306415669716162}
{"step": 90490, "loss": 3.4840972900390623, "grad_norm": 0.7567737698554993, "learning_rate": 9.941115203156933e-05, "elbo": 3.4841, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.306708890452732}
{"step": 90495, "loss": 3.4841773986816404, "grad_norm": 0.906380832195282, "learning_rate": 9.939907288919747e-05, "elbo": 3.48418, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.307002111189304}
{"step": 90500, "loss": 3.492817687988281, "grad_norm": 0.6736531257629395, "learning_rate": 9.93868718595239e-05, "elbo": 3.49282, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.307295331925873}
{"step": 90505, "loss": 3.4851036071777344, "grad_norm": 0.7065138816833496, "learning_rate": 9.937454897265337e-05, "elbo": 3.4851, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.307588552662445}
{"step": 90510, "loss": 3.482236862182617, "grad_norm": 0.7929270267486572, "learning_rate": 9.936210425899146e-05, "elbo": 3.48224, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.307881773399015}
{"step": 90515, "loss": 3.480356216430664, "grad_norm": 0.7134840488433838, "learning_rate": 9.934953774924424e-05, "elbo": 3.48036, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.308174994135586}
{"step": 90520, "loss": 3.4827423095703125, "grad_norm": 0.7709200978279114, "learning_rate": 9.93368494744183e-05, "elbo": 3.48274, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.308468214872156}
{"step": 90525, "loss": 3.4815441131591798, "grad_norm": 0.7460358142852783, "learning_rate": 9.932403946582072e-05, "elbo": 3.48154, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3087614356087265}
{"step": 90530, "loss": 3.5052650451660154, "grad_norm": 0.7437427639961243, "learning_rate": 9.931110775505892e-05, "elbo": 3.50527, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.309054656345297}
{"step": 90535, "loss": 3.474018096923828, "grad_norm": 0.6820735335350037, "learning_rate": 9.92980543740406e-05, "elbo": 3.47402, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.309347877081867}
{"step": 90540, "loss": 3.4913604736328123, "grad_norm": 0.735523521900177, "learning_rate": 9.928487935497369e-05, "elbo": 3.49136, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.309641097818438}
{"step": 90540, "loss": 3.4913604736328123, "grad_norm": 0.735523521900177, "learning_rate": 9.928487935497369e-05, "elbo": 3.49136, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.309641097818438}
{"step": 90545, "loss": 3.490273284912109, "grad_norm": 0.736879825592041, "learning_rate": 9.927158273036625e-05, "elbo": 3.49027, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.309934318555008}
{"step": 90550, "loss": 3.473551940917969, "grad_norm": 0.7971987128257751, "learning_rate": 9.925816453302634e-05, "elbo": 3.47355, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.310227539291579}
{"step": 90555, "loss": 3.480068588256836, "grad_norm": 1.0509663820266724, "learning_rate": 9.924462479606207e-05, "elbo": 3.48007, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.310520760028149}
{"step": 90560, "loss": 3.4907814025878907, "grad_norm": 0.709572434425354, "learning_rate": 9.923096355288137e-05, "elbo": 3.49078, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.31081398076472}
{"step": 90565, "loss": 3.479828643798828, "grad_norm": 0.8938488364219666, "learning_rate": 9.921718083719203e-05, "elbo": 3.47983, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.31110720150129}
{"step": 90570, "loss": 3.503337097167969, "grad_norm": 0.699074923992157, "learning_rate": 9.92032766830015e-05, "elbo": 3.50334, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.311400422237861}
{"step": 90575, "loss": 3.4819854736328124, "grad_norm": 0.7404154539108276, "learning_rate": 9.918925112461688e-05, "elbo": 3.48199, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.311693642974431}
{"step": 90580, "loss": 3.485973358154297, "grad_norm": 0.7230230569839478, "learning_rate": 9.917510419664491e-05, "elbo": 3.48597, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3119868637110015}
{"step": 90585, "loss": 3.4970108032226563, "grad_norm": 0.7035537958145142, "learning_rate": 9.916083593399166e-05, "elbo": 3.49701, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.312280084447572}
{"step": 90590, "loss": 3.4939559936523437, "grad_norm": 0.7690997123718262, "learning_rate": 9.914644637186269e-05, "elbo": 3.49396, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.312573305184142}
{"step": 90595, "loss": 3.493870162963867, "grad_norm": 0.7256326079368591, "learning_rate": 9.91319355457628e-05, "elbo": 3.49387, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.312866525920713}
{"step": 90600, "loss": 3.4992237091064453, "grad_norm": 0.8909767270088196, "learning_rate": 9.911730349149602e-05, "elbo": 3.49922, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.313159746657283}
{"step": 90605, "loss": 3.4829811096191405, "grad_norm": 0.8025895953178406, "learning_rate": 9.910255024516546e-05, "elbo": 3.48298, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.313452967393854}
{"step": 90610, "loss": 3.4800472259521484, "grad_norm": 0.7187608480453491, "learning_rate": 9.908767584317333e-05, "elbo": 3.48005, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.313746188130424}
{"step": 90615, "loss": 3.480311965942383, "grad_norm": 0.7024217247962952, "learning_rate": 9.907268032222071e-05, "elbo": 3.48031, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.314039408866995}
{"step": 90620, "loss": 3.4820480346679688, "grad_norm": 0.9511444568634033, "learning_rate": 9.905756371930757e-05, "elbo": 3.48205, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.314332629603566}
{"step": 90625, "loss": 3.464604949951172, "grad_norm": 0.8203772306442261, "learning_rate": 9.904232607173262e-05, "elbo": 3.46461, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3146258503401365}
{"step": 90625, "loss": 3.464604949951172, "grad_norm": 0.8203772306442261, "learning_rate": 9.904232607173262e-05, "elbo": 3.46461, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3146258503401365}
{"step": 90630, "loss": 3.486367034912109, "grad_norm": 0.8396435379981995, "learning_rate": 9.902696741709324e-05, "elbo": 3.48637, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.314919071076707}
{"step": 90635, "loss": 3.482940673828125, "grad_norm": 1.0315055847167969, "learning_rate": 9.90114877932854e-05, "elbo": 3.48294, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.315212291813277}
{"step": 90640, "loss": 3.493183898925781, "grad_norm": 0.6925889253616333, "learning_rate": 9.899588723850349e-05, "elbo": 3.49318, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.315505512549848}
{"step": 90645, "loss": 3.5032943725585937, "grad_norm": 0.7561399936676025, "learning_rate": 9.898016579124037e-05, "elbo": 3.50329, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.315798733286418}
{"step": 90650, "loss": 3.4866043090820313, "grad_norm": 0.7440904974937439, "learning_rate": 9.896432349028715e-05, "elbo": 3.4866, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.316091954022989}
{"step": 90655, "loss": 3.480931854248047, "grad_norm": 0.7142893075942993, "learning_rate": 9.89483603747331e-05, "elbo": 3.48093, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.316385174759559}
{"step": 90660, "loss": 3.489835739135742, "grad_norm": 0.74321049451828, "learning_rate": 9.893227648396564e-05, "elbo": 3.48984, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.31667839549613}
{"step": 90665, "loss": 3.48536376953125, "grad_norm": 0.7663556933403015, "learning_rate": 9.891607185767018e-05, "elbo": 3.48536, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3169716162327}
{"step": 90670, "loss": 3.4730186462402344, "grad_norm": 0.800398051738739, "learning_rate": 9.889974653583e-05, "elbo": 3.47302, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3172648369692705}
{"step": 90670, "loss": 3.4730186462402344, "grad_norm": 0.800398051738739, "learning_rate": 9.889974653583e-05, "elbo": 3.47302, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3172648369692705}
{"step": 90675, "loss": 3.486299514770508, "grad_norm": 0.9382002353668213, "learning_rate": 9.888330055872623e-05, "elbo": 3.4863, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.317558057705841}
{"step": 90680, "loss": 3.497076416015625, "grad_norm": 0.7301650047302246, "learning_rate": 9.886673396693768e-05, "elbo": 3.49708, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3178512784424115}
{"step": 90685, "loss": 3.4961647033691405, "grad_norm": 0.8756155967712402, "learning_rate": 9.885004680134076e-05, "elbo": 3.49616, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.318144499178982}
{"step": 90690, "loss": 3.4872108459472657, "grad_norm": 0.7387024760246277, "learning_rate": 9.88332391031094e-05, "elbo": 3.48721, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.318437719915552}
{"step": 90695, "loss": 3.48736572265625, "grad_norm": 0.705194890499115, "learning_rate": 9.881631091371491e-05, "elbo": 3.48737, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.318730940652123}
{"step": 90700, "loss": 3.484369659423828, "grad_norm": 0.7646549940109253, "learning_rate": 9.879926227492595e-05, "elbo": 3.48437, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.319024161388693}
{"step": 90705, "loss": 3.474481201171875, "grad_norm": 0.715785562992096, "learning_rate": 9.87820932288083e-05, "elbo": 3.47448, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.319317382125264}
{"step": 90710, "loss": 3.4735107421875, "grad_norm": 0.7340697646141052, "learning_rate": 9.876480381772491e-05, "elbo": 3.47351, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.319610602861834}
{"step": 90715, "loss": 3.4811237335205076, "grad_norm": 0.696869432926178, "learning_rate": 9.874739408433565e-05, "elbo": 3.48112, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.319903823598405}
{"step": 90720, "loss": 3.487378692626953, "grad_norm": 0.7141573429107666, "learning_rate": 9.872986407159732e-05, "elbo": 3.48738, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.320197044334975}
{"step": 90725, "loss": 3.4840747833251955, "grad_norm": 0.7070984244346619, "learning_rate": 9.87122138227635e-05, "elbo": 3.48407, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3204902650715455}
{"step": 90730, "loss": 3.4945598602294923, "grad_norm": 0.7092905640602112, "learning_rate": 9.869444338138441e-05, "elbo": 3.49456, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.320783485808116}
{"step": 90735, "loss": 3.482460784912109, "grad_norm": 0.7178430557250977, "learning_rate": 9.867655279130683e-05, "elbo": 3.48246, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3210767065446865}
{"step": 90740, "loss": 3.490987014770508, "grad_norm": 0.6967760324478149, "learning_rate": 9.865854209667404e-05, "elbo": 3.49099, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.321369927281257}
{"step": 90745, "loss": 3.510990524291992, "grad_norm": 0.73260098695755, "learning_rate": 9.864041134192563e-05, "elbo": 3.51099, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.321663148017828}
{"step": 90750, "loss": 3.493277740478516, "grad_norm": 0.7058006525039673, "learning_rate": 9.862216057179745e-05, "elbo": 3.49328, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.321956368754399}
{"step": 90755, "loss": 3.4698543548583984, "grad_norm": 0.7180277705192566, "learning_rate": 9.860378983132143e-05, "elbo": 3.46985, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.322249589490969}
{"step": 90760, "loss": 3.483999252319336, "grad_norm": 0.7165917754173279, "learning_rate": 9.858529916582557e-05, "elbo": 3.484, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.32254281022754}
{"step": 90765, "loss": 3.483119583129883, "grad_norm": 0.7403217554092407, "learning_rate": 9.856668862093372e-05, "elbo": 3.48312, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.32283603096411}
{"step": 90770, "loss": 3.4899654388427734, "grad_norm": 0.7295211553573608, "learning_rate": 9.85479582425656e-05, "elbo": 3.48997, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3231292517006805}
{"step": 90775, "loss": 3.4861988067626952, "grad_norm": 0.9831644892692566, "learning_rate": 9.85291080769365e-05, "elbo": 3.4862, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.323422472437251}
{"step": 90780, "loss": 3.4811023712158202, "grad_norm": 0.6873175501823425, "learning_rate": 9.851013817055737e-05, "elbo": 3.4811, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.323715693173821}
{"step": 90785, "loss": 3.4977691650390623, "grad_norm": 0.9229249954223633, "learning_rate": 9.849104857023455e-05, "elbo": 3.49777, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.324008913910392}
{"step": 90790, "loss": 3.4976497650146485, "grad_norm": 0.8252624273300171, "learning_rate": 9.847183932306973e-05, "elbo": 3.49765, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.324302134646962}
{"step": 90795, "loss": 3.483031463623047, "grad_norm": 0.772487998008728, "learning_rate": 9.845251047645983e-05, "elbo": 3.48303, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.324595355383533}
{"step": 90800, "loss": 3.4804187774658204, "grad_norm": 0.9172987937927246, "learning_rate": 9.843306207809687e-05, "elbo": 3.48042, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.324888576120103}
{"step": 90805, "loss": 3.5043624877929687, "grad_norm": 0.7473406791687012, "learning_rate": 9.841349417596779e-05, "elbo": 3.50436, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.325181796856674}
{"step": 90810, "loss": 3.4958587646484376, "grad_norm": 0.8170978426933289, "learning_rate": 9.83938068183545e-05, "elbo": 3.49586, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.325475017593244}
{"step": 90815, "loss": 3.4953075408935548, "grad_norm": 0.7759097814559937, "learning_rate": 9.837400005383354e-05, "elbo": 3.49531, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.325768238329815}
{"step": 90820, "loss": 3.497724151611328, "grad_norm": 0.7539257407188416, "learning_rate": 9.835407393127621e-05, "elbo": 3.49772, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.326061459066385}
{"step": 90825, "loss": 3.4896739959716796, "grad_norm": 0.7868717908859253, "learning_rate": 9.833402849984815e-05, "elbo": 3.48967, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3263546798029555}
{"step": 90830, "loss": 3.4890869140625, "grad_norm": 0.7975499033927917, "learning_rate": 9.831386380900953e-05, "elbo": 3.48909, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.326647900539526}
{"step": 90835, "loss": 3.506764602661133, "grad_norm": 0.8980153799057007, "learning_rate": 9.829357990851468e-05, "elbo": 3.50676, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.326941121276096}
{"step": 90840, "loss": 3.476618194580078, "grad_norm": 0.7079169750213623, "learning_rate": 9.827317684841216e-05, "elbo": 3.47662, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.327234342012667}
{"step": 90845, "loss": 3.4994434356689452, "grad_norm": 1.1211062669754028, "learning_rate": 9.825265467904445e-05, "elbo": 3.49944, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.327527562749237}
{"step": 90850, "loss": 3.493433380126953, "grad_norm": 0.7336397767066956, "learning_rate": 9.823201345104798e-05, "elbo": 3.49343, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.327820783485808}
{"step": 90855, "loss": 3.480514144897461, "grad_norm": 0.7316429615020752, "learning_rate": 9.82112532153529e-05, "elbo": 3.48051, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.328114004222378}
{"step": 90860, "loss": 3.492401885986328, "grad_norm": 0.7297926545143127, "learning_rate": 9.819037402318308e-05, "elbo": 3.4924, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.328407224958949}
{"step": 90865, "loss": 3.4859386444091798, "grad_norm": 0.7863432765007019, "learning_rate": 9.816937592605579e-05, "elbo": 3.48594, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.328700445695519}
{"step": 90870, "loss": 3.489315414428711, "grad_norm": 0.8234909772872925, "learning_rate": 9.814825897578179e-05, "elbo": 3.48932, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3289936664320905}
{"step": 90875, "loss": 3.497724914550781, "grad_norm": 0.7315918803215027, "learning_rate": 9.812702322446505e-05, "elbo": 3.49773, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.329286887168661}
{"step": 90880, "loss": 3.498705291748047, "grad_norm": 0.6732640862464905, "learning_rate": 9.810566872450269e-05, "elbo": 3.49871, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.329580107905231}
{"step": 90885, "loss": 3.506867218017578, "grad_norm": 0.7039766907691956, "learning_rate": 9.808419552858477e-05, "elbo": 3.50687, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.329873328641802}
{"step": 90890, "loss": 3.473709487915039, "grad_norm": 0.7773534059524536, "learning_rate": 9.806260368969429e-05, "elbo": 3.47371, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.330166549378372}
{"step": 90895, "loss": 3.4895153045654297, "grad_norm": 0.7301090359687805, "learning_rate": 9.804089326110697e-05, "elbo": 3.48952, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.330459770114943}
{"step": 90900, "loss": 3.4848609924316407, "grad_norm": 0.734778106212616, "learning_rate": 9.801906429639114e-05, "elbo": 3.48486, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.330752990851513}
{"step": 90905, "loss": 3.487223815917969, "grad_norm": 0.7334897518157959, "learning_rate": 9.79971168494076e-05, "elbo": 3.48722, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.331046211588084}
{"step": 90910, "loss": 3.4881618499755858, "grad_norm": 0.7401747703552246, "learning_rate": 9.797505097430947e-05, "elbo": 3.48816, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.331339432324654}
{"step": 90915, "loss": 3.454907989501953, "grad_norm": 0.7137543559074402, "learning_rate": 9.795286672554213e-05, "elbo": 3.45491, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.331632653061225}
{"step": 90920, "loss": 3.4678371429443358, "grad_norm": 0.7666890025138855, "learning_rate": 9.793056415784298e-05, "elbo": 3.46784, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.331925873797795}
{"step": 90925, "loss": 3.4874862670898437, "grad_norm": 0.7150291800498962, "learning_rate": 9.790814332624143e-05, "elbo": 3.48749, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3322190945343655}
{"step": 90930, "loss": 3.4696067810058593, "grad_norm": 0.7362528443336487, "learning_rate": 9.788560428605863e-05, "elbo": 3.46961, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.332512315270936}
{"step": 90930, "loss": 3.4696067810058593, "grad_norm": 0.7362528443336487, "learning_rate": 9.788560428605863e-05, "elbo": 3.46961, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.332512315270936}
{"step": 90935, "loss": 3.467999267578125, "grad_norm": 1.0358225107192993, "learning_rate": 9.786294709290741e-05, "elbo": 3.468, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.332805536007506}
{"step": 90940, "loss": 3.488368606567383, "grad_norm": 0.8272331953048706, "learning_rate": 9.784017180269216e-05, "elbo": 3.48837, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.333098756744077}
{"step": 90945, "loss": 3.4810501098632813, "grad_norm": 0.7078899145126343, "learning_rate": 9.781727847160865e-05, "elbo": 3.48105, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.333391977480647}
{"step": 90950, "loss": 3.469207000732422, "grad_norm": 0.7317364811897278, "learning_rate": 9.779426715614387e-05, "elbo": 3.46921, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.333685198217218}
{"step": 90955, "loss": 3.487578582763672, "grad_norm": 0.7433924078941345, "learning_rate": 9.777113791307598e-05, "elbo": 3.48758, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.333978418953788}
{"step": 90960, "loss": 3.4679500579833986, "grad_norm": 0.7273712754249573, "learning_rate": 9.774789079947406e-05, "elbo": 3.46795, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.334271639690359}
{"step": 90960, "loss": 3.4679500579833986, "grad_norm": 0.7273712754249573, "learning_rate": 9.774789079947406e-05, "elbo": 3.46795, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.334271639690359}
{"step": 90965, "loss": 3.4958900451660155, "grad_norm": 0.7296302318572998, "learning_rate": 9.772452587269808e-05, "elbo": 3.49589, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.334564860426929}
{"step": 90970, "loss": 3.4954627990722655, "grad_norm": 0.7021084427833557, "learning_rate": 9.770104319039866e-05, "elbo": 3.49546, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3348580811635}
{"step": 90975, "loss": 3.4805625915527343, "grad_norm": 0.699303388595581, "learning_rate": 9.767744281051701e-05, "elbo": 3.48056, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.33515130190007}
{"step": 90980, "loss": 3.478704071044922, "grad_norm": 0.7266038656234741, "learning_rate": 9.765372479128467e-05, "elbo": 3.4787, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3354445226366405}
{"step": 90985, "loss": 3.4810256958007812, "grad_norm": 0.73191237449646, "learning_rate": 9.762988919122355e-05, "elbo": 3.48103, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.335737743373211}
{"step": 90990, "loss": 3.479705047607422, "grad_norm": 0.7259107828140259, "learning_rate": 9.760593606914559e-05, "elbo": 3.47971, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.336030964109781}
{"step": 90995, "loss": 3.480524444580078, "grad_norm": 0.7631645798683167, "learning_rate": 9.758186548415273e-05, "elbo": 3.48052, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.336324184846353}
{"step": 91000, "loss": 3.479758453369141, "grad_norm": 0.768589973449707, "learning_rate": 9.755767749563678e-05, "elbo": 3.47976, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.336617405582923}
{"step": 91005, "loss": 3.480046844482422, "grad_norm": 0.7585333585739136, "learning_rate": 9.753337216327917e-05, "elbo": 3.48005, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.336910626319494}
{"step": 91010, "loss": 3.4814186096191406, "grad_norm": 0.7206116318702698, "learning_rate": 9.75089495470509e-05, "elbo": 3.48142, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.337203847056064}
{"step": 91010, "loss": 3.4814186096191406, "grad_norm": 0.7206116318702698, "learning_rate": 9.75089495470509e-05, "elbo": 3.48142, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.337203847056064}
{"step": 91015, "loss": 3.4773521423339844, "grad_norm": 0.7804847359657288, "learning_rate": 9.748440970721236e-05, "elbo": 3.47735, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3374970677926346}
{"step": 91020, "loss": 3.475360870361328, "grad_norm": 0.6917481422424316, "learning_rate": 9.745975270431313e-05, "elbo": 3.47536, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.337790288529205}
{"step": 91025, "loss": 3.4744190216064452, "grad_norm": 0.7505252957344055, "learning_rate": 9.743497859919196e-05, "elbo": 3.47442, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3380835092657755}
{"step": 91030, "loss": 3.5005931854248047, "grad_norm": 0.7581508755683899, "learning_rate": 9.741008745297646e-05, "elbo": 3.50059, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.338376730002346}
{"step": 91035, "loss": 3.4846134185791016, "grad_norm": 0.7000285983085632, "learning_rate": 9.738507932708307e-05, "elbo": 3.48461, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.338669950738916}
{"step": 91040, "loss": 3.485240173339844, "grad_norm": 1.239105463027954, "learning_rate": 9.735995428321685e-05, "elbo": 3.48524, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.338963171475487}
{"step": 91045, "loss": 3.4975589752197265, "grad_norm": 0.7565001845359802, "learning_rate": 9.733471238337136e-05, "elbo": 3.49756, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.339256392212057}
{"step": 91050, "loss": 3.4898468017578126, "grad_norm": 1.025175929069519, "learning_rate": 9.730935368982846e-05, "elbo": 3.48985, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.339549612948628}
{"step": 91055, "loss": 3.4820884704589843, "grad_norm": 0.7132789492607117, "learning_rate": 9.728387826515819e-05, "elbo": 3.48209, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.339842833685198}
{"step": 91060, "loss": 3.4847900390625, "grad_norm": 0.7133187651634216, "learning_rate": 9.725828617221868e-05, "elbo": 3.48479, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.340136054421769}
{"step": 91065, "loss": 3.502592086791992, "grad_norm": 0.930851399898529, "learning_rate": 9.723257747415584e-05, "elbo": 3.50259, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.340429275158339}
{"step": 91070, "loss": 3.4822837829589846, "grad_norm": 0.7067127227783203, "learning_rate": 9.720675223440334e-05, "elbo": 3.48228, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3407224958949095}
{"step": 91075, "loss": 3.48524169921875, "grad_norm": 0.7261618375778198, "learning_rate": 9.718081051668239e-05, "elbo": 3.48524, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.34101571663148}
{"step": 91080, "loss": 3.4776817321777345, "grad_norm": 0.7147430181503296, "learning_rate": 9.715475238500159e-05, "elbo": 3.47768, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3413089373680505}
{"step": 91080, "loss": 3.4776817321777345, "grad_norm": 0.7147430181503296, "learning_rate": 9.715475238500159e-05, "elbo": 3.47768, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3413089373680505}
{"step": 91085, "loss": 3.4759757995605467, "grad_norm": 0.7408387660980225, "learning_rate": 9.71285779036568e-05, "elbo": 3.47598, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.341602158104621}
{"step": 91090, "loss": 3.494637298583984, "grad_norm": 1.0292131900787354, "learning_rate": 9.710228713723092e-05, "elbo": 3.49464, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.341895378841191}
{"step": 91095, "loss": 3.4891517639160154, "grad_norm": 0.7089919447898865, "learning_rate": 9.707588015059386e-05, "elbo": 3.48915, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.342188599577762}
{"step": 91100, "loss": 3.460490417480469, "grad_norm": 0.7345603704452515, "learning_rate": 9.70493570089022e-05, "elbo": 3.46049, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.342481820314332}
{"step": 91105, "loss": 3.4892288208007813, "grad_norm": 0.6864414811134338, "learning_rate": 9.702271777759916e-05, "elbo": 3.48923, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.342775041050903}
{"step": 91110, "loss": 3.494709014892578, "grad_norm": 0.7036576867103577, "learning_rate": 9.699596252241438e-05, "elbo": 3.49471, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.343068261787474}
{"step": 91115, "loss": 3.4748661041259767, "grad_norm": 0.7477577924728394, "learning_rate": 9.696909130936382e-05, "elbo": 3.47487, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.343361482524044}
{"step": 91120, "loss": 3.470112991333008, "grad_norm": 0.7009323239326477, "learning_rate": 9.69421042047495e-05, "elbo": 3.47011, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.343654703260615}
{"step": 91125, "loss": 3.496600341796875, "grad_norm": 0.7235203981399536, "learning_rate": 9.691500127515945e-05, "elbo": 3.4966, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.343947923997185}
{"step": 91130, "loss": 3.4791210174560545, "grad_norm": 0.7407472133636475, "learning_rate": 9.688778258746743e-05, "elbo": 3.47912, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.344241144733756}
{"step": 91135, "loss": 3.487377166748047, "grad_norm": 0.692244827747345, "learning_rate": 9.686044820883285e-05, "elbo": 3.48738, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.344534365470326}
{"step": 91140, "loss": 3.4825191497802734, "grad_norm": 0.7135477662086487, "learning_rate": 9.683299820670056e-05, "elbo": 3.48252, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.344827586206897}
{"step": 91145, "loss": 3.474198913574219, "grad_norm": 0.8314880132675171, "learning_rate": 9.680543264880076e-05, "elbo": 3.4742, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.345120806943467}
{"step": 91150, "loss": 3.4861297607421875, "grad_norm": 0.7281827330589294, "learning_rate": 9.677775160314865e-05, "elbo": 3.48613, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.345414027680038}
{"step": 91155, "loss": 3.4818077087402344, "grad_norm": 0.7283372282981873, "learning_rate": 9.674995513804452e-05, "elbo": 3.48181, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.345707248416608}
{"step": 91160, "loss": 3.4769142150878904, "grad_norm": 0.8430704474449158, "learning_rate": 9.672204332207336e-05, "elbo": 3.47691, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.346000469153179}
{"step": 91165, "loss": 3.488329315185547, "grad_norm": 0.7428876161575317, "learning_rate": 9.669401622410482e-05, "elbo": 3.48833, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.346293689889749}
{"step": 91170, "loss": 3.467803955078125, "grad_norm": 0.7349466681480408, "learning_rate": 9.666587391329294e-05, "elbo": 3.4678, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3465869106263195}
{"step": 91175, "loss": 3.501372528076172, "grad_norm": 0.8515468835830688, "learning_rate": 9.663761645907609e-05, "elbo": 3.50137, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.34688013136289}
{"step": 91180, "loss": 3.48741455078125, "grad_norm": 0.6891637444496155, "learning_rate": 9.660924393117675e-05, "elbo": 3.48741, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.34717335209946}
{"step": 91185, "loss": 3.4831844329833985, "grad_norm": 0.8683391213417053, "learning_rate": 9.65807563996013e-05, "elbo": 3.48318, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.347466572836031}
{"step": 91190, "loss": 3.466826629638672, "grad_norm": 0.8201948404312134, "learning_rate": 9.655215393463989e-05, "elbo": 3.46683, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.347759793572601}
{"step": 91195, "loss": 3.4984355926513673, "grad_norm": 0.7216342687606812, "learning_rate": 9.652343660686626e-05, "elbo": 3.49844, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.348053014309172}
{"step": 91200, "loss": 3.492975616455078, "grad_norm": 0.7491535544395447, "learning_rate": 9.649460448713755e-05, "elbo": 3.49298, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.348346235045742}
{"step": 91205, "loss": 3.4756900787353517, "grad_norm": 0.7689887285232544, "learning_rate": 9.646565764659417e-05, "elbo": 3.47569, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.348639455782313}
{"step": 91210, "loss": 3.48468132019043, "grad_norm": 0.7191406488418579, "learning_rate": 9.643659615665957e-05, "elbo": 3.48468, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.348932676518883}
{"step": 91215, "loss": 3.4956428527832033, "grad_norm": 0.8545845150947571, "learning_rate": 9.640742008904005e-05, "elbo": 3.49564, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.349225897255454}
{"step": 91220, "loss": 3.476349639892578, "grad_norm": 0.6888216733932495, "learning_rate": 9.637812951572471e-05, "elbo": 3.47635, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.349519117992024}
{"step": 91225, "loss": 3.4634067535400392, "grad_norm": 0.8015666007995605, "learning_rate": 9.63487245089851e-05, "elbo": 3.46341, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3498123387285945}
{"step": 91230, "loss": 3.4873279571533202, "grad_norm": 0.7345951795578003, "learning_rate": 9.631920514137516e-05, "elbo": 3.48733, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.350105559465165}
{"step": 91235, "loss": 3.4961605072021484, "grad_norm": 0.7108643651008606, "learning_rate": 9.628957148573098e-05, "elbo": 3.49616, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.350398780201736}
{"step": 91240, "loss": 3.495245361328125, "grad_norm": 0.6974796652793884, "learning_rate": 9.625982361517068e-05, "elbo": 3.49525, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.350692000938307}
{"step": 91245, "loss": 3.4860153198242188, "grad_norm": 0.859143853187561, "learning_rate": 9.622996160309414e-05, "elbo": 3.48602, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.350985221674877}
{"step": 91250, "loss": 3.50390739440918, "grad_norm": 0.6929617524147034, "learning_rate": 9.619998552318295e-05, "elbo": 3.50391, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.351278442411448}
{"step": 91255, "loss": 3.4718021392822265, "grad_norm": 0.6934723854064941, "learning_rate": 9.616989544940009e-05, "elbo": 3.4718, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.351571663148018}
{"step": 91260, "loss": 3.489209747314453, "grad_norm": 0.7425893545150757, "learning_rate": 9.61396914559898e-05, "elbo": 3.48921, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.351864883884589}
{"step": 91265, "loss": 3.4953487396240233, "grad_norm": 0.7115665078163147, "learning_rate": 9.610937361747748e-05, "elbo": 3.49535, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.352158104621159}
{"step": 91270, "loss": 3.4748191833496094, "grad_norm": 0.704120397567749, "learning_rate": 9.607894200866934e-05, "elbo": 3.47482, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3524513253577295}
{"step": 91275, "loss": 3.4933212280273436, "grad_norm": 0.7844572067260742, "learning_rate": 9.604839670465236e-05, "elbo": 3.49332, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3527445460943}
{"step": 91275, "loss": 3.4933212280273436, "grad_norm": 0.7844572067260742, "learning_rate": 9.604839670465236e-05, "elbo": 3.49332, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3527445460943}
{"step": 91280, "loss": 3.498444366455078, "grad_norm": 0.8477641940116882, "learning_rate": 9.601773778079406e-05, "elbo": 3.49844, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.35303776683087}
{"step": 91285, "loss": 3.478659820556641, "grad_norm": 0.767516016960144, "learning_rate": 9.598696531274227e-05, "elbo": 3.47866, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.353330987567441}
{"step": 91290, "loss": 3.5042884826660154, "grad_norm": 0.8136755228042603, "learning_rate": 9.595607937642502e-05, "elbo": 3.50429, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.353624208304011}
{"step": 91295, "loss": 3.4716728210449217, "grad_norm": 0.7336395978927612, "learning_rate": 9.592508004805023e-05, "elbo": 3.47167, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.353917429040582}
{"step": 91300, "loss": 3.493581771850586, "grad_norm": 1.1801066398620605, "learning_rate": 9.589396740410573e-05, "elbo": 3.49358, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.354210649777152}
{"step": 91305, "loss": 3.4961544036865235, "grad_norm": 0.7468885779380798, "learning_rate": 9.586274152135884e-05, "elbo": 3.49615, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.354503870513723}
{"step": 91310, "loss": 3.457602691650391, "grad_norm": 0.7210195660591125, "learning_rate": 9.583140247685631e-05, "elbo": 3.4576, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.354797091250293}
{"step": 91315, "loss": 3.515188980102539, "grad_norm": 0.7370889186859131, "learning_rate": 9.579995034792414e-05, "elbo": 3.51519, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.355090311986864}
{"step": 91320, "loss": 3.4883861541748047, "grad_norm": 0.7654078602790833, "learning_rate": 9.576838521216734e-05, "elbo": 3.48839, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.355383532723434}
{"step": 91325, "loss": 3.4698486328125, "grad_norm": 0.7310976386070251, "learning_rate": 9.573670714746972e-05, "elbo": 3.46985, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3556767534600045}
{"step": 91330, "loss": 3.5122886657714845, "grad_norm": 0.9420188069343567, "learning_rate": 9.570491623199376e-05, "elbo": 3.51229, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.355969974196575}
{"step": 91335, "loss": 3.4728443145751955, "grad_norm": 0.7233130931854248, "learning_rate": 9.567301254418038e-05, "elbo": 3.47284, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.356263194933145}
{"step": 91340, "loss": 3.4906742095947267, "grad_norm": 0.7144619822502136, "learning_rate": 9.564099616274879e-05, "elbo": 3.49067, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.356556415669716}
{"step": 91345, "loss": 3.4682769775390625, "grad_norm": 0.7588281631469727, "learning_rate": 9.56088671666962e-05, "elbo": 3.46828, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.356849636406286}
{"step": 91350, "loss": 3.485552978515625, "grad_norm": 0.711220383644104, "learning_rate": 9.557662563529771e-05, "elbo": 3.48555, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.357142857142857}
{"step": 91355, "loss": 3.4968814849853516, "grad_norm": 0.7660422325134277, "learning_rate": 9.554427164810611e-05, "elbo": 3.49688, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.357436077879427}
{"step": 91360, "loss": 3.46771240234375, "grad_norm": 0.7443971633911133, "learning_rate": 9.551180528495165e-05, "elbo": 3.46771, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.357729298615999}
{"step": 91365, "loss": 3.483205795288086, "grad_norm": 0.9716806411743164, "learning_rate": 9.547922662594183e-05, "elbo": 3.48321, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.358022519352569}
{"step": 91370, "loss": 3.4788833618164063, "grad_norm": 0.7232659459114075, "learning_rate": 9.544653575146127e-05, "elbo": 3.47888, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3583157400891395}
{"step": 91375, "loss": 3.4731231689453126, "grad_norm": 0.7873700857162476, "learning_rate": 9.541373274217145e-05, "elbo": 3.47312, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.35860896082571}
{"step": 91380, "loss": 3.476479721069336, "grad_norm": 0.6867458820343018, "learning_rate": 9.538081767901054e-05, "elbo": 3.47648, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.35890218156228}
{"step": 91385, "loss": 3.48704833984375, "grad_norm": 0.7753639221191406, "learning_rate": 9.534779064319318e-05, "elbo": 3.48705, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.359195402298851}
{"step": 91390, "loss": 3.475260925292969, "grad_norm": 0.8077570796012878, "learning_rate": 9.53146517162103e-05, "elbo": 3.47526, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.359488623035421}
{"step": 91395, "loss": 3.477106475830078, "grad_norm": 0.7314435839653015, "learning_rate": 9.528140097982889e-05, "elbo": 3.47711, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.359781843771992}
{"step": 91400, "loss": 3.4940582275390626, "grad_norm": 0.701813817024231, "learning_rate": 9.524803851609188e-05, "elbo": 3.49406, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.360075064508562}
{"step": 91405, "loss": 3.4854923248291017, "grad_norm": 0.7334837913513184, "learning_rate": 9.52145644073178e-05, "elbo": 3.48549, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.360368285245133}
{"step": 91410, "loss": 3.487010192871094, "grad_norm": 0.7505595684051514, "learning_rate": 9.518097873610068e-05, "elbo": 3.48701, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.360661505981703}
{"step": 91415, "loss": 3.463887023925781, "grad_norm": 0.7586278319358826, "learning_rate": 9.514728158530983e-05, "elbo": 3.46389, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3609547267182736}
{"step": 91420, "loss": 3.455879974365234, "grad_norm": 0.801490068435669, "learning_rate": 9.511347303808965e-05, "elbo": 3.45588, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.361247947454844}
{"step": 91425, "loss": 3.482697296142578, "grad_norm": 0.7347511053085327, "learning_rate": 9.507955317785934e-05, "elbo": 3.4827, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3615411681914145}
{"step": 91430, "loss": 3.4769287109375, "grad_norm": 0.9655991792678833, "learning_rate": 9.504552208831279e-05, "elbo": 3.47693, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.361834388927985}
{"step": 91435, "loss": 3.504039764404297, "grad_norm": 0.705098032951355, "learning_rate": 9.501137985341832e-05, "elbo": 3.50404, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.362127609664555}
{"step": 91440, "loss": 3.4788318634033204, "grad_norm": 0.7301804423332214, "learning_rate": 9.497712655741851e-05, "elbo": 3.47883, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.362420830401126}
{"step": 91445, "loss": 3.4859603881835937, "grad_norm": 0.7456446290016174, "learning_rate": 9.494276228482998e-05, "elbo": 3.48596, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.362714051137696}
{"step": 91445, "loss": 3.4859603881835937, "grad_norm": 0.7456446290016174, "learning_rate": 9.494276228482998e-05, "elbo": 3.48596, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.362714051137696}
{"step": 91450, "loss": 3.474386978149414, "grad_norm": 0.7834206223487854, "learning_rate": 9.490828712044312e-05, "elbo": 3.47439, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.363007271874267}
{"step": 91450, "loss": 3.474386978149414, "grad_norm": 0.7834206223487854, "learning_rate": 9.490828712044312e-05, "elbo": 3.47439, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.363007271874267}
{"step": 91455, "loss": 3.4788223266601563, "grad_norm": 0.7131314873695374, "learning_rate": 9.487370114932202e-05, "elbo": 3.47882, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.363300492610837}
{"step": 91460, "loss": 3.4930931091308595, "grad_norm": 0.7122820019721985, "learning_rate": 9.483900445680407e-05, "elbo": 3.49309, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.363593713347408}
{"step": 91460, "loss": 3.4930931091308595, "grad_norm": 0.7122820019721985, "learning_rate": 9.483900445680407e-05, "elbo": 3.49309, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.363593713347408}
{"step": 91465, "loss": 3.4759593963623048, "grad_norm": 0.7149441838264465, "learning_rate": 9.480419712849995e-05, "elbo": 3.47596, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.363886934083978}
{"step": 91470, "loss": 3.470353698730469, "grad_norm": 0.7734033465385437, "learning_rate": 9.476927925029326e-05, "elbo": 3.47035, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3641801548205486}
{"step": 91475, "loss": 3.475608062744141, "grad_norm": 0.9522197246551514, "learning_rate": 9.473425090834041e-05, "elbo": 3.47561, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.364473375557119}
{"step": 91480, "loss": 3.4884502410888674, "grad_norm": 0.8710215091705322, "learning_rate": 9.469911218907035e-05, "elbo": 3.48845, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3647665962936895}
{"step": 91485, "loss": 3.486186218261719, "grad_norm": 0.7273626327514648, "learning_rate": 9.466386317918436e-05, "elbo": 3.48619, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.365059817030261}
{"step": 91490, "loss": 3.481883239746094, "grad_norm": 0.7225004434585571, "learning_rate": 9.462850396565588e-05, "elbo": 3.48188, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.365353037766831}
{"step": 91490, "loss": 3.481883239746094, "grad_norm": 0.7225004434585571, "learning_rate": 9.462850396565588e-05, "elbo": 3.48188, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.365353037766831}
{"step": 91495, "loss": 3.474753570556641, "grad_norm": 0.7737979292869568, "learning_rate": 9.459303463573026e-05, "elbo": 3.47475, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.365646258503402}
{"step": 91500, "loss": 3.4737869262695313, "grad_norm": 0.7677534222602844, "learning_rate": 9.455745527692454e-05, "elbo": 3.47379, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.365939479239972}
{"step": 91505, "loss": 3.4806060791015625, "grad_norm": 0.7348681092262268, "learning_rate": 9.452176597702725e-05, "elbo": 3.48061, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.366232699976543}
{"step": 91510, "loss": 3.465160369873047, "grad_norm": 0.73101806640625, "learning_rate": 9.448596682409819e-05, "elbo": 3.46516, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.366525920713113}
{"step": 91515, "loss": 3.47808837890625, "grad_norm": 0.7052220106124878, "learning_rate": 9.445005790646819e-05, "elbo": 3.47809, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3668191414496835}
{"step": 91515, "loss": 3.47808837890625, "grad_norm": 0.7052220106124878, "learning_rate": 9.445005790646819e-05, "elbo": 3.47809, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3668191414496835}
{"step": 91520, "loss": 3.453718566894531, "grad_norm": 0.7686837315559387, "learning_rate": 9.441403931273897e-05, "elbo": 3.45372, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.367112362186254}
{"step": 91525, "loss": 3.490395736694336, "grad_norm": 0.7142157554626465, "learning_rate": 9.437791113178282e-05, "elbo": 3.4904, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.367405582922824}
{"step": 91530, "loss": 3.4692691802978515, "grad_norm": 0.9417300224304199, "learning_rate": 9.434167345274244e-05, "elbo": 3.46927, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.367698803659395}
{"step": 91535, "loss": 3.474170684814453, "grad_norm": 0.7408395409584045, "learning_rate": 9.430532636503068e-05, "elbo": 3.47417, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.367992024395965}
{"step": 91540, "loss": 3.4903274536132813, "grad_norm": 0.7265886664390564, "learning_rate": 9.426886995833038e-05, "elbo": 3.49033, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.368285245132536}
{"step": 91545, "loss": 3.486754608154297, "grad_norm": 0.731991171836853, "learning_rate": 9.423230432259409e-05, "elbo": 3.48675, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.368578465869106}
{"step": 91550, "loss": 3.468577194213867, "grad_norm": 0.7107913494110107, "learning_rate": 9.419562954804388e-05, "elbo": 3.46858, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.368871686605677}
{"step": 91555, "loss": 3.4794822692871095, "grad_norm": 0.761822521686554, "learning_rate": 9.415884572517113e-05, "elbo": 3.47948, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.369164907342247}
{"step": 91560, "loss": 3.476262664794922, "grad_norm": 0.7185661792755127, "learning_rate": 9.412195294473625e-05, "elbo": 3.47626, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.369458128078818}
{"step": 91565, "loss": 3.476160430908203, "grad_norm": 0.7441838979721069, "learning_rate": 9.408495129776851e-05, "elbo": 3.47616, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.369751348815388}
{"step": 91565, "loss": 3.476160430908203, "grad_norm": 0.7441838979721069, "learning_rate": 9.408495129776851e-05, "elbo": 3.47616, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.369751348815388}
{"step": 91570, "loss": 3.485808181762695, "grad_norm": 0.7595805525779724, "learning_rate": 9.404784087556582e-05, "elbo": 3.48581, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3700445695519585}
{"step": 91575, "loss": 3.467087173461914, "grad_norm": 0.6696009039878845, "learning_rate": 9.401062176969442e-05, "elbo": 3.46709, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.370337790288529}
{"step": 91580, "loss": 3.483345794677734, "grad_norm": 0.6827584505081177, "learning_rate": 9.397329407198875e-05, "elbo": 3.48335, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.370631011025099}
{"step": 91585, "loss": 3.4666927337646483, "grad_norm": 0.7377746105194092, "learning_rate": 9.393585787455124e-05, "elbo": 3.46669, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.37092423176167}
{"step": 91590, "loss": 3.4802082061767576, "grad_norm": 0.8012658357620239, "learning_rate": 9.389831326975196e-05, "elbo": 3.48021, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.37121745249824}
{"step": 91595, "loss": 3.4622241973876955, "grad_norm": 0.9458647966384888, "learning_rate": 9.386066035022848e-05, "elbo": 3.46222, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.371510673234811}
{"step": 91600, "loss": 3.466817092895508, "grad_norm": 0.6955516934394836, "learning_rate": 9.382289920888565e-05, "elbo": 3.46682, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.371803893971381}
{"step": 91605, "loss": 3.4873001098632814, "grad_norm": 0.6895300149917603, "learning_rate": 9.378502993889533e-05, "elbo": 3.4873, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.372097114707952}
{"step": 91610, "loss": 3.4758232116699217, "grad_norm": 0.7823479771614075, "learning_rate": 9.374705263369617e-05, "elbo": 3.47582, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.372390335444523}
{"step": 91615, "loss": 3.4663219451904297, "grad_norm": 0.7788943648338318, "learning_rate": 9.370896738699339e-05, "elbo": 3.46632, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3726835561810935}
{"step": 91620, "loss": 3.4748214721679687, "grad_norm": 1.154159665107727, "learning_rate": 9.367077429275859e-05, "elbo": 3.47482, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.372976776917664}
{"step": 91625, "loss": 3.487784576416016, "grad_norm": 0.71209716796875, "learning_rate": 9.363247344522938e-05, "elbo": 3.48778, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.373269997654234}
{"step": 91630, "loss": 3.4799831390380858, "grad_norm": 0.7735114693641663, "learning_rate": 9.359406493890934e-05, "elbo": 3.47998, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.373563218390805}
{"step": 91635, "loss": 3.496446228027344, "grad_norm": 0.6910625100135803, "learning_rate": 9.355554886856762e-05, "elbo": 3.49645, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.373856439127375}
{"step": 91640, "loss": 3.4769744873046875, "grad_norm": 0.7067468166351318, "learning_rate": 9.351692532923879e-05, "elbo": 3.47697, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.374149659863946}
{"step": 91645, "loss": 3.4698081970214845, "grad_norm": 0.7473634481430054, "learning_rate": 9.347819441622261e-05, "elbo": 3.46981, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.374442880600516}
{"step": 91650, "loss": 3.4742172241210936, "grad_norm": 0.6850465536117554, "learning_rate": 9.343935622508373e-05, "elbo": 3.47422, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.374736101337087}
{"step": 91650, "loss": 3.4742172241210936, "grad_norm": 0.6850465536117554, "learning_rate": 9.343935622508373e-05, "elbo": 3.47422, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.374736101337087}
{"step": 91655, "loss": 3.4875556945800783, "grad_norm": 0.6997440457344055, "learning_rate": 9.340041085165156e-05, "elbo": 3.48756, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.375029322073657}
{"step": 91660, "loss": 3.4843204498291014, "grad_norm": 0.7725067734718323, "learning_rate": 9.336135839201992e-05, "elbo": 3.48432, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.375322542810228}
{"step": 91660, "loss": 3.4843204498291014, "grad_norm": 0.7725067734718323, "learning_rate": 9.336135839201992e-05, "elbo": 3.48432, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.375322542810228}
{"step": 91665, "loss": 3.471160888671875, "grad_norm": 0.709932267665863, "learning_rate": 9.332219894254686e-05, "elbo": 3.47116, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.375615763546798}
{"step": 91670, "loss": 3.478092575073242, "grad_norm": 0.6846686005592346, "learning_rate": 9.328293259985446e-05, "elbo": 3.47809, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3759089842833685}
{"step": 91675, "loss": 3.4954158782958986, "grad_norm": 0.6808072924613953, "learning_rate": 9.324355946082848e-05, "elbo": 3.49542, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.376202205019939}
{"step": 91680, "loss": 3.4765830993652345, "grad_norm": 0.7317870855331421, "learning_rate": 9.320407962261826e-05, "elbo": 3.47658, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.376495425756509}
{"step": 91685, "loss": 3.478314971923828, "grad_norm": 0.7765255570411682, "learning_rate": 9.316449318263635e-05, "elbo": 3.47832, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.37678864649308}
{"step": 91690, "loss": 3.4703689575195313, "grad_norm": 0.8075999021530151, "learning_rate": 9.312480023855836e-05, "elbo": 3.47037, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.37708186722965}
{"step": 91695, "loss": 3.468048095703125, "grad_norm": 0.7661924362182617, "learning_rate": 9.308500088832272e-05, "elbo": 3.46805, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.377375087966221}
{"step": 91700, "loss": 3.476732635498047, "grad_norm": 0.7053108215332031, "learning_rate": 9.304509523013031e-05, "elbo": 3.47673, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.377668308702791}
{"step": 91700, "loss": 3.476732635498047, "grad_norm": 0.7053108215332031, "learning_rate": 9.304509523013031e-05, "elbo": 3.47673, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.377668308702791}
{"step": 91705, "loss": 3.4795757293701173, "grad_norm": 0.8123566508293152, "learning_rate": 9.300508336244444e-05, "elbo": 3.47958, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.377961529439362}
{"step": 91710, "loss": 3.474509429931641, "grad_norm": 0.8905639052391052, "learning_rate": 9.296496538399036e-05, "elbo": 3.47451, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.378254750175932}
{"step": 91715, "loss": 3.48707275390625, "grad_norm": 0.6687185764312744, "learning_rate": 9.292474139375522e-05, "elbo": 3.48707, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.378547970912503}
{"step": 91720, "loss": 3.4792633056640625, "grad_norm": 0.7133371829986572, "learning_rate": 9.288441149098773e-05, "elbo": 3.47926, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.378841191649073}
{"step": 91725, "loss": 3.47423095703125, "grad_norm": 0.7321133017539978, "learning_rate": 9.284397577519788e-05, "elbo": 3.47423, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.379134412385644}
{"step": 91730, "loss": 3.477747344970703, "grad_norm": 0.7084339261054993, "learning_rate": 9.280343434615679e-05, "elbo": 3.47775, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.379427633122214}
{"step": 91735, "loss": 3.4858699798583985, "grad_norm": 0.7760383486747742, "learning_rate": 9.276278730389642e-05, "elbo": 3.48587, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.379720853858785}
{"step": 91740, "loss": 3.4778774261474608, "grad_norm": 1.2322545051574707, "learning_rate": 9.272203474870932e-05, "elbo": 3.47788, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.380014074595356}
{"step": 91745, "loss": 3.465715789794922, "grad_norm": 0.7215152382850647, "learning_rate": 9.268117678114834e-05, "elbo": 3.46572, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.380307295331926}
{"step": 91750, "loss": 3.4678443908691405, "grad_norm": 0.7830747365951538, "learning_rate": 9.264021350202646e-05, "elbo": 3.46784, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.380600516068497}
{"step": 91755, "loss": 3.4726234436035157, "grad_norm": 0.7335706949234009, "learning_rate": 9.259914501241652e-05, "elbo": 3.47262, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.380893736805067}
{"step": 91760, "loss": 3.4766315460205077, "grad_norm": 0.7452252507209778, "learning_rate": 9.25579714136509e-05, "elbo": 3.47663, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.381186957541638}
{"step": 91765, "loss": 3.4796737670898437, "grad_norm": 0.7290142774581909, "learning_rate": 9.251669280732137e-05, "elbo": 3.47967, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.381480178278208}
{"step": 91765, "loss": 3.4796737670898437, "grad_norm": 0.7290142774581909, "learning_rate": 9.251669280732137e-05, "elbo": 3.47967, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.381480178278208}
{"step": 91770, "loss": 3.4738719940185545, "grad_norm": 0.7916088104248047, "learning_rate": 9.247530929527881e-05, "elbo": 3.47387, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3817733990147785}
{"step": 91775, "loss": 3.4882747650146486, "grad_norm": 0.7323437333106995, "learning_rate": 9.243382097963291e-05, "elbo": 3.48827, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.382066619751349}
{"step": 91780, "loss": 3.4835174560546873, "grad_norm": 0.7093664407730103, "learning_rate": 9.239222796275195e-05, "elbo": 3.48352, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.382359840487919}
{"step": 91785, "loss": 3.4763198852539063, "grad_norm": 0.7750756740570068, "learning_rate": 9.23505303472626e-05, "elbo": 3.47632, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.38265306122449}
{"step": 91790, "loss": 3.460318374633789, "grad_norm": 0.7673206329345703, "learning_rate": 9.230872823604956e-05, "elbo": 3.46032, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.38294628196106}
{"step": 91795, "loss": 3.4805419921875, "grad_norm": 0.7077853083610535, "learning_rate": 9.226682173225537e-05, "elbo": 3.48054, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.383239502697631}
{"step": 91800, "loss": 3.4817184448242187, "grad_norm": 0.7174766063690186, "learning_rate": 9.22248109392802e-05, "elbo": 3.48172, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.383532723434201}
{"step": 91805, "loss": 3.4712669372558596, "grad_norm": 0.7310473918914795, "learning_rate": 9.218269596078146e-05, "elbo": 3.47127, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.383825944170772}
{"step": 91810, "loss": 3.4833724975585936, "grad_norm": 0.7084946632385254, "learning_rate": 9.214047690067369e-05, "elbo": 3.48337, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.384119164907342}
{"step": 91815, "loss": 3.4734832763671877, "grad_norm": 0.6780222058296204, "learning_rate": 9.209815386312824e-05, "elbo": 3.47348, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.384412385643913}
{"step": 91820, "loss": 3.464725112915039, "grad_norm": 0.702329158782959, "learning_rate": 9.205572695257298e-05, "elbo": 3.46473, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.384705606380483}
{"step": 91825, "loss": 3.482335662841797, "grad_norm": 0.7350054383277893, "learning_rate": 9.201319627369211e-05, "elbo": 3.48234, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3849988271170535}
{"step": 91830, "loss": 3.478718948364258, "grad_norm": 0.7620722651481628, "learning_rate": 9.197056193142583e-05, "elbo": 3.47872, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.385292047853624}
{"step": 91835, "loss": 3.4631507873535154, "grad_norm": 0.7982946038246155, "learning_rate": 9.192782403097018e-05, "elbo": 3.46315, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.385585268590194}
{"step": 91840, "loss": 3.4731903076171875, "grad_norm": 0.8287769556045532, "learning_rate": 9.188498267777664e-05, "elbo": 3.47319, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.385878489326765}
{"step": 91845, "loss": 3.4911033630371096, "grad_norm": 0.7505976557731628, "learning_rate": 9.1842037977552e-05, "elbo": 3.4911, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.386171710063335}
{"step": 91850, "loss": 3.4790714263916014, "grad_norm": 0.8597451448440552, "learning_rate": 9.179899003625804e-05, "elbo": 3.47907, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.386464930799907}
{"step": 91850, "loss": 3.4790714263916014, "grad_norm": 0.8597451448440552, "learning_rate": 9.179899003625804e-05, "elbo": 3.47907, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.386464930799907}
{"step": 91855, "loss": 3.4789703369140623, "grad_norm": 0.7692286968231201, "learning_rate": 9.175583896011131e-05, "elbo": 3.47897, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.386758151536476}
{"step": 91860, "loss": 3.4727756500244142, "grad_norm": 0.723192572593689, "learning_rate": 9.171258485558274e-05, "elbo": 3.47278, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3870513722730475}
{"step": 91865, "loss": 3.4951450347900392, "grad_norm": 0.7553744912147522, "learning_rate": 9.166922782939758e-05, "elbo": 3.49515, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.387344593009618}
{"step": 91870, "loss": 3.475551986694336, "grad_norm": 0.7323023080825806, "learning_rate": 9.162576798853497e-05, "elbo": 3.47555, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.387637813746188}
{"step": 91875, "loss": 3.485586166381836, "grad_norm": 0.7848578691482544, "learning_rate": 9.158220544022773e-05, "elbo": 3.48559, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.387931034482759}
{"step": 91880, "loss": 3.4934558868408203, "grad_norm": 0.695451557636261, "learning_rate": 9.153854029196213e-05, "elbo": 3.49346, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.388224255219329}
{"step": 91885, "loss": 3.4716644287109375, "grad_norm": 0.9862110018730164, "learning_rate": 9.149477265147762e-05, "elbo": 3.47166, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3885174759559}
{"step": 91890, "loss": 3.4868614196777346, "grad_norm": 0.7787746787071228, "learning_rate": 9.145090262676642e-05, "elbo": 3.48686, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.38881069669247}
{"step": 91895, "loss": 3.471627044677734, "grad_norm": 0.7392387986183167, "learning_rate": 9.140693032607353e-05, "elbo": 3.47163, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.389103917429041}
{"step": 91900, "loss": 3.481671142578125, "grad_norm": 0.7342073917388916, "learning_rate": 9.136285585789619e-05, "elbo": 3.48167, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.389397138165611}
{"step": 91905, "loss": 3.4665733337402345, "grad_norm": 0.7563720941543579, "learning_rate": 9.131867933098378e-05, "elbo": 3.46657, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.389690358902182}
{"step": 91910, "loss": 3.46591796875, "grad_norm": 0.7316106557846069, "learning_rate": 9.12744008543375e-05, "elbo": 3.46592, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.389983579638752}
{"step": 91915, "loss": 3.4756095886230467, "grad_norm": 0.7962214946746826, "learning_rate": 9.123002053721005e-05, "elbo": 3.47561, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3902768003753225}
{"step": 91915, "loss": 3.4756095886230467, "grad_norm": 0.7962214946746826, "learning_rate": 9.123002053721005e-05, "elbo": 3.47561, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3902768003753225}
{"step": 91920, "loss": 3.491232681274414, "grad_norm": 0.707310676574707, "learning_rate": 9.118553848910552e-05, "elbo": 3.49123, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.390570021111893}
{"step": 91925, "loss": 3.488688278198242, "grad_norm": 0.7206076979637146, "learning_rate": 9.114095481977888e-05, "elbo": 3.48869, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.390863241848463}
{"step": 91930, "loss": 3.4860565185546877, "grad_norm": 0.796760618686676, "learning_rate": 9.109626963923592e-05, "elbo": 3.48606, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.391156462585034}
{"step": 91935, "loss": 3.4736560821533202, "grad_norm": 0.6985984444618225, "learning_rate": 9.10514830577329e-05, "elbo": 3.47366, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.391449683321604}
{"step": 91935, "loss": 3.4736560821533202, "grad_norm": 0.6985984444618225, "learning_rate": 9.10514830577329e-05, "elbo": 3.47366, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.391449683321604}
{"step": 91940, "loss": 3.4840248107910154, "grad_norm": 0.7571746110916138, "learning_rate": 9.100659518577624e-05, "elbo": 3.48402, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.391742904058175}
{"step": 91945, "loss": 3.4726924896240234, "grad_norm": 0.7556883692741394, "learning_rate": 9.096160613412228e-05, "elbo": 3.47269, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.392036124794745}
{"step": 91950, "loss": 3.4786666870117187, "grad_norm": 0.7867440581321716, "learning_rate": 9.091651601377709e-05, "elbo": 3.47867, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.392329345531316}
{"step": 91955, "loss": 3.479913330078125, "grad_norm": 1.091247797012329, "learning_rate": 9.0871324935996e-05, "elbo": 3.47991, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.392622566267886}
{"step": 91960, "loss": 3.497878646850586, "grad_norm": 0.8571501970291138, "learning_rate": 9.082603301228354e-05, "elbo": 3.49788, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.392915787004457}
{"step": 91960, "loss": 3.497878646850586, "grad_norm": 0.8571501970291138, "learning_rate": 9.082603301228354e-05, "elbo": 3.49788, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.392915787004457}
{"step": 91965, "loss": 3.4807060241699217, "grad_norm": 0.6986537575721741, "learning_rate": 9.078064035439301e-05, "elbo": 3.48071, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.393209007741027}
{"step": 91970, "loss": 3.4737613677978514, "grad_norm": 0.7137166857719421, "learning_rate": 9.073514707432631e-05, "elbo": 3.47376, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3935022284775975}
{"step": 91975, "loss": 3.483782196044922, "grad_norm": 0.725016176700592, "learning_rate": 9.068955328433355e-05, "elbo": 3.48378, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.393795449214169}
{"step": 91980, "loss": 3.4794384002685548, "grad_norm": 0.7740428447723389, "learning_rate": 9.06438590969129e-05, "elbo": 3.47944, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.394088669950739}
{"step": 91985, "loss": 3.460298538208008, "grad_norm": 0.7359550595283508, "learning_rate": 9.059806462481023e-05, "elbo": 3.4603, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.39438189068731}
{"step": 91990, "loss": 3.479186248779297, "grad_norm": 0.7354037761688232, "learning_rate": 9.055216998101879e-05, "elbo": 3.47919, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.39467511142388}
{"step": 91995, "loss": 3.497100067138672, "grad_norm": 0.7522748112678528, "learning_rate": 9.050617527877911e-05, "elbo": 3.4971, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.394968332160451}
{"step": 92000, "loss": 3.4821983337402345, "grad_norm": 0.7642499804496765, "learning_rate": 9.046008063157855e-05, "elbo": 3.4822, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.395261552897021}
{"step": 92005, "loss": 3.490416717529297, "grad_norm": 0.7191197276115417, "learning_rate": 9.041388615315102e-05, "elbo": 3.49042, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.395554773633592}
{"step": 92010, "loss": 3.4677570343017576, "grad_norm": 0.6841768026351929, "learning_rate": 9.036759195747686e-05, "elbo": 3.46776, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.395847994370162}
{"step": 92010, "loss": 3.4677570343017576, "grad_norm": 0.6841768026351929, "learning_rate": 9.036759195747686e-05, "elbo": 3.46776, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.395847994370162}
{"step": 92015, "loss": 3.4796215057373048, "grad_norm": 0.7594135403633118, "learning_rate": 9.032119815878236e-05, "elbo": 3.47962, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3961412151067325}
{"step": 92020, "loss": 3.4558990478515623, "grad_norm": 0.7210913300514221, "learning_rate": 9.027470487153962e-05, "elbo": 3.4559, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.396434435843303}
{"step": 92025, "loss": 3.477404022216797, "grad_norm": 0.6901746392250061, "learning_rate": 9.022811221046618e-05, "elbo": 3.4774, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.396727656579873}
{"step": 92030, "loss": 3.4719474792480467, "grad_norm": 0.7935242652893066, "learning_rate": 9.018142029052483e-05, "elbo": 3.47195, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.397020877316444}
{"step": 92035, "loss": 3.4561248779296876, "grad_norm": 0.7124564051628113, "learning_rate": 9.013462922692324e-05, "elbo": 3.45612, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.397314098053014}
{"step": 92040, "loss": 3.484017181396484, "grad_norm": 0.7484568953514099, "learning_rate": 9.008773913511369e-05, "elbo": 3.48402, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.397607318789585}
{"step": 92045, "loss": 3.475908660888672, "grad_norm": 0.6902848482131958, "learning_rate": 9.004075013079283e-05, "elbo": 3.47591, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.397900539526155}
{"step": 92050, "loss": 3.4872718811035157, "grad_norm": 0.7001262307167053, "learning_rate": 8.999366232990136e-05, "elbo": 3.48727, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.398193760262726}
{"step": 92050, "loss": 3.4872718811035157, "grad_norm": 0.7001262307167053, "learning_rate": 8.999366232990136e-05, "elbo": 3.48727, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.398193760262726}
{"step": 92055, "loss": 3.4656726837158205, "grad_norm": 0.7447909116744995, "learning_rate": 8.994647584862374e-05, "elbo": 3.46567, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.398486980999296}
{"step": 92060, "loss": 3.4604518890380858, "grad_norm": 0.7079881429672241, "learning_rate": 8.989919080338792e-05, "elbo": 3.46045, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.398780201735867}
{"step": 92065, "loss": 3.471033477783203, "grad_norm": 0.6758766174316406, "learning_rate": 8.985180731086505e-05, "elbo": 3.47103, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.399073422472437}
{"step": 92070, "loss": 3.4695106506347657, "grad_norm": 0.7188454270362854, "learning_rate": 8.98043254879692e-05, "elbo": 3.46951, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.3993666432090075}
{"step": 92075, "loss": 3.4586017608642576, "grad_norm": 0.6836944222450256, "learning_rate": 8.975674545185703e-05, "elbo": 3.4586, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.399659863945578}
{"step": 92080, "loss": 3.463145446777344, "grad_norm": 0.8984236717224121, "learning_rate": 8.970906731992759e-05, "elbo": 3.46315, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.399953084682148}
{"step": 92085, "loss": 3.4718204498291017, "grad_norm": 0.7393488883972168, "learning_rate": 8.966129120982188e-05, "elbo": 3.47182, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.400246305418719}
{"step": 92090, "loss": 3.4644866943359376, "grad_norm": 0.9818729758262634, "learning_rate": 8.961341723942271e-05, "elbo": 3.46449, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.400539526155289}
{"step": 92095, "loss": 3.485322952270508, "grad_norm": 0.9726305603981018, "learning_rate": 8.956544552685437e-05, "elbo": 3.48532, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.40083274689186}
{"step": 92100, "loss": 3.485639953613281, "grad_norm": 0.79315185546875, "learning_rate": 8.951737619048228e-05, "elbo": 3.48564, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.401125967628431}
{"step": 92105, "loss": 3.473986053466797, "grad_norm": 0.774063229560852, "learning_rate": 8.946920934891274e-05, "elbo": 3.47399, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.401419188365002}
{"step": 92110, "loss": 3.4649444580078126, "grad_norm": 0.7855973243713379, "learning_rate": 8.942094512099264e-05, "elbo": 3.46494, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.401712409101572}
{"step": 92115, "loss": 3.4891990661621093, "grad_norm": 0.7317979335784912, "learning_rate": 8.937258362580918e-05, "elbo": 3.4892, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4020056298381425}
{"step": 92120, "loss": 3.487313079833984, "grad_norm": 0.762158215045929, "learning_rate": 8.932412498268955e-05, "elbo": 3.48731, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.402298850574713}
{"step": 92125, "loss": 3.4747264862060545, "grad_norm": 0.6948477029800415, "learning_rate": 8.92755693112006e-05, "elbo": 3.47473, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.402592071311283}
{"step": 92130, "loss": 3.4747684478759764, "grad_norm": 0.7059973478317261, "learning_rate": 8.922691673114866e-05, "elbo": 3.47477, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.402885292047854}
{"step": 92135, "loss": 3.4871673583984375, "grad_norm": 0.7074762582778931, "learning_rate": 8.917816736257911e-05, "elbo": 3.48717, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.403178512784424}
{"step": 92140, "loss": 3.4756492614746093, "grad_norm": 0.7047852873802185, "learning_rate": 8.912932132577619e-05, "elbo": 3.47565, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.403471733520995}
{"step": 92145, "loss": 3.469074249267578, "grad_norm": 0.7269728779792786, "learning_rate": 8.908037874126263e-05, "elbo": 3.46907, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.403764954257565}
{"step": 92150, "loss": 3.447684478759766, "grad_norm": 0.794177770614624, "learning_rate": 8.903133972979938e-05, "elbo": 3.44768, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.404058174994136}
{"step": 92155, "loss": 3.4835914611816405, "grad_norm": 0.7530357837677002, "learning_rate": 8.898220441238534e-05, "elbo": 3.48359, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.404351395730706}
{"step": 92160, "loss": 3.4701091766357424, "grad_norm": 0.8215131163597107, "learning_rate": 8.893297291025703e-05, "elbo": 3.47011, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.404644616467277}
{"step": 92165, "loss": 3.4809619903564455, "grad_norm": 0.9159908890724182, "learning_rate": 8.888364534488827e-05, "elbo": 3.48096, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.404937837203847}
{"step": 92170, "loss": 3.4790107727050783, "grad_norm": 0.7655046582221985, "learning_rate": 8.883422183798992e-05, "elbo": 3.47901, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4052310579404175}
{"step": 92175, "loss": 3.482461929321289, "grad_norm": 0.723446249961853, "learning_rate": 8.878470251150959e-05, "elbo": 3.48246, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.405524278676988}
{"step": 92180, "loss": 3.4890342712402345, "grad_norm": 0.9294942021369934, "learning_rate": 8.873508748763129e-05, "elbo": 3.48903, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.405817499413558}
{"step": 92185, "loss": 3.47530517578125, "grad_norm": 0.6895052194595337, "learning_rate": 8.868537688877516e-05, "elbo": 3.47531, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.406110720150129}
{"step": 92190, "loss": 3.4626163482666015, "grad_norm": 0.7478560209274292, "learning_rate": 8.863557083759714e-05, "elbo": 3.46262, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.406403940886699}
{"step": 92195, "loss": 3.479685974121094, "grad_norm": 0.7136788964271545, "learning_rate": 8.858566945698874e-05, "elbo": 3.47969, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.40669716162327}
{"step": 92200, "loss": 3.4734691619873046, "grad_norm": 0.7979550361633301, "learning_rate": 8.853567287007665e-05, "elbo": 3.47347, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.40699038235984}
{"step": 92205, "loss": 3.466553497314453, "grad_norm": 0.6764766573905945, "learning_rate": 8.848558120022246e-05, "elbo": 3.46655, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.407283603096411}
{"step": 92210, "loss": 3.4756935119628904, "grad_norm": 0.753538191318512, "learning_rate": 8.843539457102238e-05, "elbo": 3.47569, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.407576823832981}
{"step": 92215, "loss": 3.479247283935547, "grad_norm": 0.8637574315071106, "learning_rate": 8.838511310630697e-05, "elbo": 3.47925, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.407870044569552}
{"step": 92220, "loss": 3.4773788452148438, "grad_norm": 0.7075116038322449, "learning_rate": 8.833473693014071e-05, "elbo": 3.47738, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.408163265306122}
{"step": 92225, "loss": 3.465494155883789, "grad_norm": 0.7793738842010498, "learning_rate": 8.828426616682184e-05, "elbo": 3.46549, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.408456486042693}
{"step": 92230, "loss": 3.4719478607177736, "grad_norm": 0.7102559208869934, "learning_rate": 8.823370094088191e-05, "elbo": 3.47195, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.408749706779264}
{"step": 92235, "loss": 3.4689208984375, "grad_norm": 0.7316542863845825, "learning_rate": 8.81830413770856e-05, "elbo": 3.46892, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.409042927515834}
{"step": 92240, "loss": 3.4742523193359376, "grad_norm": 0.6698794960975647, "learning_rate": 8.813228760043037e-05, "elbo": 3.47425, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.409336148252405}
{"step": 92245, "loss": 3.490740203857422, "grad_norm": 0.6856736540794373, "learning_rate": 8.808143973614611e-05, "elbo": 3.49074, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.409629368988975}
{"step": 92250, "loss": 3.4909767150878905, "grad_norm": 0.6920049786567688, "learning_rate": 8.803049790969486e-05, "elbo": 3.49098, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.409922589725546}
{"step": 92255, "loss": 3.464218521118164, "grad_norm": 0.9258260726928711, "learning_rate": 8.797946224677052e-05, "elbo": 3.46422, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.410215810462116}
{"step": 92260, "loss": 3.486614227294922, "grad_norm": 0.7728217244148254, "learning_rate": 8.792833287329851e-05, "elbo": 3.48661, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4105090311986865}
{"step": 92265, "loss": 3.467332458496094, "grad_norm": 0.7230608463287354, "learning_rate": 8.787710991543548e-05, "elbo": 3.46733, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.410802251935257}
{"step": 92270, "loss": 3.483226013183594, "grad_norm": 0.8098630905151367, "learning_rate": 8.782579349956897e-05, "elbo": 3.48323, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4110954726718274}
{"step": 92275, "loss": 3.4889068603515625, "grad_norm": 0.7728985548019409, "learning_rate": 8.777438375231717e-05, "elbo": 3.48891, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.411388693408398}
{"step": 92280, "loss": 3.4798519134521486, "grad_norm": 0.8567337393760681, "learning_rate": 8.77228808005285e-05, "elbo": 3.47985, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.411681914144968}
{"step": 92285, "loss": 3.478594207763672, "grad_norm": 0.7226264476776123, "learning_rate": 8.767128477128137e-05, "elbo": 3.47859, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.411975134881539}
{"step": 92290, "loss": 3.4857547760009764, "grad_norm": 0.9767508506774902, "learning_rate": 8.761959579188386e-05, "elbo": 3.48575, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.412268355618109}
{"step": 92295, "loss": 3.4687564849853514, "grad_norm": 0.7317407131195068, "learning_rate": 8.75678139898734e-05, "elbo": 3.46876, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.41256157635468}
{"step": 92300, "loss": 3.476622772216797, "grad_norm": 0.7053695917129517, "learning_rate": 8.75159394930164e-05, "elbo": 3.47662, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.41285479709125}
{"step": 92305, "loss": 3.4730449676513673, "grad_norm": 0.7098397612571716, "learning_rate": 8.746397242930808e-05, "elbo": 3.47305, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.413148017827821}
{"step": 92310, "loss": 3.4724571228027346, "grad_norm": 0.7459536194801331, "learning_rate": 8.741191292697195e-05, "elbo": 3.47246, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.413441238564391}
{"step": 92315, "loss": 3.472342681884766, "grad_norm": 0.7362739443778992, "learning_rate": 8.73597611144597e-05, "elbo": 3.47234, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4137344593009615}
{"step": 92320, "loss": 3.482249450683594, "grad_norm": 0.72696453332901, "learning_rate": 8.73075171204507e-05, "elbo": 3.48225, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.414027680037532}
{"step": 92325, "loss": 3.468913269042969, "grad_norm": 0.8752859830856323, "learning_rate": 8.725518107385187e-05, "elbo": 3.46891, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.414320900774102}
{"step": 92330, "loss": 3.453438568115234, "grad_norm": 0.7850625514984131, "learning_rate": 8.720275310379713e-05, "elbo": 3.45344, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.414614121510673}
{"step": 92335, "loss": 3.4738849639892577, "grad_norm": 0.779895007610321, "learning_rate": 8.715023333964736e-05, "elbo": 3.47388, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.414907342247243}
{"step": 92340, "loss": 3.466979217529297, "grad_norm": 0.7044205069541931, "learning_rate": 8.70976219109898e-05, "elbo": 3.46698, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.415200562983814}
{"step": 92345, "loss": 3.477974700927734, "grad_norm": 0.7168036699295044, "learning_rate": 8.704491894763794e-05, "elbo": 3.47797, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.415493783720384}
{"step": 92350, "loss": 3.4630287170410154, "grad_norm": 0.8300110697746277, "learning_rate": 8.699212457963111e-05, "elbo": 3.46303, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.415787004456956}
{"step": 92355, "loss": 3.465074157714844, "grad_norm": 0.8034656047821045, "learning_rate": 8.693923893723416e-05, "elbo": 3.46507, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.416080225193526}
{"step": 92360, "loss": 3.463134002685547, "grad_norm": 0.724416196346283, "learning_rate": 8.688626215093714e-05, "elbo": 3.46313, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4163734459300965}
{"step": 92365, "loss": 3.4704925537109377, "grad_norm": 0.7451924681663513, "learning_rate": 8.683319435145503e-05, "elbo": 3.47049, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.416666666666667}
{"step": 92370, "loss": 3.478585052490234, "grad_norm": 0.8587011098861694, "learning_rate": 8.678003566972735e-05, "elbo": 3.47858, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.416959887403237}
{"step": 92375, "loss": 3.483966064453125, "grad_norm": 0.690593957901001, "learning_rate": 8.672678623691783e-05, "elbo": 3.48397, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.417253108139808}
{"step": 92380, "loss": 3.475624847412109, "grad_norm": 0.6943471431732178, "learning_rate": 8.667344618441418e-05, "elbo": 3.47562, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.417546328876378}
{"step": 92385, "loss": 3.4794643402099608, "grad_norm": 1.097249984741211, "learning_rate": 8.662001564382767e-05, "elbo": 3.47946, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.417839549612949}
{"step": 92390, "loss": 3.4805999755859376, "grad_norm": 0.7586583495140076, "learning_rate": 8.656649474699283e-05, "elbo": 3.4806, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.418132770349519}
{"step": 92395, "loss": 3.4753677368164064, "grad_norm": 0.6970484852790833, "learning_rate": 8.651288362596718e-05, "elbo": 3.47537, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.41842599108609}
{"step": 92400, "loss": 3.4745792388916015, "grad_norm": 0.7247471213340759, "learning_rate": 8.645918241303084e-05, "elbo": 3.47458, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.41871921182266}
{"step": 92405, "loss": 3.477417755126953, "grad_norm": 0.6771168112754822, "learning_rate": 8.640539124068617e-05, "elbo": 3.47742, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.419012432559231}
{"step": 92410, "loss": 3.472138595581055, "grad_norm": 0.7852901220321655, "learning_rate": 8.635151024165759e-05, "elbo": 3.47214, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.419305653295801}
{"step": 92415, "loss": 3.4774208068847656, "grad_norm": 0.7388373613357544, "learning_rate": 8.629753954889107e-05, "elbo": 3.47742, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4195988740323715}
{"step": 92420, "loss": 3.477935028076172, "grad_norm": 0.9963307976722717, "learning_rate": 8.624347929555394e-05, "elbo": 3.47793, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.419892094768942}
{"step": 92425, "loss": 3.47020263671875, "grad_norm": 0.754767656326294, "learning_rate": 8.618932961503452e-05, "elbo": 3.4702, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.420185315505512}
{"step": 92430, "loss": 3.488642120361328, "grad_norm": 0.793615996837616, "learning_rate": 8.613509064094175e-05, "elbo": 3.48864, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.420478536242083}
{"step": 92435, "loss": 3.475225067138672, "grad_norm": 0.868584930896759, "learning_rate": 8.60807625071049e-05, "elbo": 3.47522, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.420771756978653}
{"step": 92440, "loss": 3.4667869567871095, "grad_norm": 0.7574566602706909, "learning_rate": 8.602634534757326e-05, "elbo": 3.46679, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.421064977715224}
{"step": 92445, "loss": 3.4607025146484376, "grad_norm": 0.8923636078834534, "learning_rate": 8.597183929661573e-05, "elbo": 3.4607, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.421358198451794}
{"step": 92450, "loss": 3.466180419921875, "grad_norm": 0.691555917263031, "learning_rate": 8.591724448872061e-05, "elbo": 3.46618, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.421651419188365}
{"step": 92455, "loss": 3.4688148498535156, "grad_norm": 0.7768640518188477, "learning_rate": 8.586256105859512e-05, "elbo": 3.46882, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.421944639924935}
{"step": 92460, "loss": 3.4823066711425783, "grad_norm": 0.7366205453872681, "learning_rate": 8.580778914116522e-05, "elbo": 3.48231, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.422237860661506}
{"step": 92460, "loss": 3.4823066711425783, "grad_norm": 0.7366205453872681, "learning_rate": 8.580778914116522e-05, "elbo": 3.48231, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.422237860661506}
{"step": 92465, "loss": 3.472063446044922, "grad_norm": 0.6938414573669434, "learning_rate": 8.575292887157516e-05, "elbo": 3.47206, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.422531081398077}
{"step": 92470, "loss": 3.4607791900634766, "grad_norm": 0.7337709665298462, "learning_rate": 8.569798038518719e-05, "elbo": 3.46078, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4228243021346465}
{"step": 92475, "loss": 3.46234016418457, "grad_norm": 0.7392928600311279, "learning_rate": 8.564294381758128e-05, "elbo": 3.46234, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.423117522871218}
{"step": 92480, "loss": 3.4740070343017577, "grad_norm": 0.7538308501243591, "learning_rate": 8.558781930455464e-05, "elbo": 3.47401, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.423410743607788}
{"step": 92485, "loss": 3.4684967041015624, "grad_norm": 0.6878002882003784, "learning_rate": 8.553260698212155e-05, "elbo": 3.4685, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.423703964344359}
{"step": 92490, "loss": 3.471837615966797, "grad_norm": 0.7731730937957764, "learning_rate": 8.547730698651293e-05, "elbo": 3.47184, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.423997185080929}
{"step": 92495, "loss": 3.467800521850586, "grad_norm": 0.7437525391578674, "learning_rate": 8.542191945417601e-05, "elbo": 3.4678, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4242904058175}
{"step": 92500, "loss": 3.4615509033203127, "grad_norm": 0.7332227230072021, "learning_rate": 8.536644452177404e-05, "elbo": 3.46155, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.42458362655407}
{"step": 92500, "loss": 3.4615509033203127, "grad_norm": 0.7332227230072021, "learning_rate": 8.536644452177404e-05, "elbo": 3.46155, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.42458362655407}
{"step": 92505, "loss": 3.4719024658203126, "grad_norm": 0.7155212759971619, "learning_rate": 8.531088232618588e-05, "elbo": 3.4719, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.424876847290641}
{"step": 92505, "loss": 3.4719024658203126, "grad_norm": 0.7155212759971619, "learning_rate": 8.531088232618588e-05, "elbo": 3.4719, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.424876847290641}
{"step": 92510, "loss": 3.45460205078125, "grad_norm": 0.717377245426178, "learning_rate": 8.525523300450571e-05, "elbo": 3.4546, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.425170068027211}
{"step": 92510, "loss": 3.45460205078125, "grad_norm": 0.717377245426178, "learning_rate": 8.525523300450571e-05, "elbo": 3.4546, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.425170068027211}
{"step": 92515, "loss": 3.467847442626953, "grad_norm": 0.7121168971061707, "learning_rate": 8.519949669404274e-05, "elbo": 3.46785, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4254632887637815}
{"step": 92520, "loss": 3.4665077209472654, "grad_norm": 0.826006293296814, "learning_rate": 8.514367353232076e-05, "elbo": 3.46651, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.425756509500352}
{"step": 92525, "loss": 3.4733131408691404, "grad_norm": 0.9767909646034241, "learning_rate": 8.508776365707787e-05, "elbo": 3.47331, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.426049730236922}
{"step": 92530, "loss": 3.4739654541015623, "grad_norm": 0.8002771735191345, "learning_rate": 8.50317672062661e-05, "elbo": 3.47397, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.426342950973493}
{"step": 92535, "loss": 3.466297912597656, "grad_norm": 0.741603672504425, "learning_rate": 8.497568431805119e-05, "elbo": 3.4663, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.426636171710063}
{"step": 92540, "loss": 3.4676929473876954, "grad_norm": 0.7123607397079468, "learning_rate": 8.491951513081203e-05, "elbo": 3.46769, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.426929392446634}
{"step": 92545, "loss": 3.4668914794921877, "grad_norm": 0.7116767764091492, "learning_rate": 8.486325978314053e-05, "elbo": 3.46689, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.427222613183204}
{"step": 92550, "loss": 3.4701217651367187, "grad_norm": 0.6959356665611267, "learning_rate": 8.480691841384118e-05, "elbo": 3.47012, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.427515833919775}
{"step": 92555, "loss": 3.4794971466064455, "grad_norm": 0.6963419318199158, "learning_rate": 8.475049116193071e-05, "elbo": 3.4795, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.427809054656345}
{"step": 92560, "loss": 3.492584991455078, "grad_norm": 0.768295407295227, "learning_rate": 8.469397816663773e-05, "elbo": 3.49259, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.428102275392916}
{"step": 92565, "loss": 3.478559875488281, "grad_norm": 1.109305500984192, "learning_rate": 8.463737956740245e-05, "elbo": 3.47856, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.428395496129486}
{"step": 92570, "loss": 3.4848636627197265, "grad_norm": 0.7792874574661255, "learning_rate": 8.458069550387629e-05, "elbo": 3.48486, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4286887168660565}
{"step": 92575, "loss": 3.4654304504394533, "grad_norm": 0.7146317362785339, "learning_rate": 8.452392611592153e-05, "elbo": 3.46543, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.428981937602627}
{"step": 92580, "loss": 3.4634326934814452, "grad_norm": 0.8387191891670227, "learning_rate": 8.446707154361102e-05, "elbo": 3.46343, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.429275158339197}
{"step": 92585, "loss": 3.4671279907226564, "grad_norm": 0.7554637789726257, "learning_rate": 8.441013192722773e-05, "elbo": 3.46713, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.429568379075768}
{"step": 92590, "loss": 3.471660614013672, "grad_norm": 0.7252991199493408, "learning_rate": 8.435310740726452e-05, "elbo": 3.47166, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.429861599812339}
{"step": 92595, "loss": 3.456790542602539, "grad_norm": 0.7259654998779297, "learning_rate": 8.429599812442372e-05, "elbo": 3.45679, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.43015482054891}
{"step": 92600, "loss": 3.4602996826171877, "grad_norm": 0.7074742317199707, "learning_rate": 8.423880421961683e-05, "elbo": 3.4603, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.43044804128548}
{"step": 92605, "loss": 3.475664520263672, "grad_norm": 0.7389761209487915, "learning_rate": 8.41815258339641e-05, "elbo": 3.47566, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4307412620220505}
{"step": 92605, "loss": 3.475664520263672, "grad_norm": 0.7389761209487915, "learning_rate": 8.41815258339641e-05, "elbo": 3.47566, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4307412620220505}
{"step": 92610, "loss": 3.4807693481445314, "grad_norm": 0.6999717950820923, "learning_rate": 8.412416310879426e-05, "elbo": 3.48077, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.431034482758621}
{"step": 92610, "loss": 3.4807693481445314, "grad_norm": 0.6999717950820923, "learning_rate": 8.412416310879426e-05, "elbo": 3.48077, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.431034482758621}
{"step": 92615, "loss": 3.4607105255126953, "grad_norm": 0.7874998450279236, "learning_rate": 8.406671618564413e-05, "elbo": 3.46071, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4313277034951915}
{"step": 92620, "loss": 3.471820068359375, "grad_norm": 0.7107972502708435, "learning_rate": 8.400918520625829e-05, "elbo": 3.47182, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.431620924231762}
{"step": 92625, "loss": 3.4675315856933593, "grad_norm": 0.782541036605835, "learning_rate": 8.39515703125887e-05, "elbo": 3.46753, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.431914144968332}
{"step": 92630, "loss": 3.4892623901367186, "grad_norm": 0.716630756855011, "learning_rate": 8.389387164679439e-05, "elbo": 3.48926, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.432207365704903}
{"step": 92635, "loss": 3.4609115600585936, "grad_norm": 0.770076334476471, "learning_rate": 8.383608935124109e-05, "elbo": 3.46091, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.432500586441473}
{"step": 92640, "loss": 3.4831314086914062, "grad_norm": 0.8007282018661499, "learning_rate": 8.377822356850084e-05, "elbo": 3.48313, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.432793807178044}
{"step": 92645, "loss": 3.472666931152344, "grad_norm": 0.749302089214325, "learning_rate": 8.372027444135176e-05, "elbo": 3.47267, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.433087027914614}
{"step": 92650, "loss": 3.4659503936767577, "grad_norm": 0.7178457975387573, "learning_rate": 8.366224211277751e-05, "elbo": 3.46595, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.433380248651185}
{"step": 92655, "loss": 3.4616119384765627, "grad_norm": 0.7271606922149658, "learning_rate": 8.360412672596712e-05, "elbo": 3.46161, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.433673469387755}
{"step": 92660, "loss": 3.47570686340332, "grad_norm": 0.7021403312683105, "learning_rate": 8.354592842431452e-05, "elbo": 3.47571, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4339666901243255}
{"step": 92665, "loss": 3.4595245361328124, "grad_norm": 0.6934264898300171, "learning_rate": 8.348764735141823e-05, "elbo": 3.45952, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.434259910860896}
{"step": 92670, "loss": 3.479437255859375, "grad_norm": 3.5944764614105225, "learning_rate": 8.342928365108102e-05, "elbo": 3.47944, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4345531315974664}
{"step": 92675, "loss": 3.45875244140625, "grad_norm": 0.7758259773254395, "learning_rate": 8.33708374673095e-05, "elbo": 3.45875, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.434846352334037}
{"step": 92680, "loss": 3.469755172729492, "grad_norm": 0.7114593982696533, "learning_rate": 8.331230894431381e-05, "elbo": 3.46976, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.435139573070607}
{"step": 92685, "loss": 3.445342254638672, "grad_norm": 0.800359845161438, "learning_rate": 8.32536982265073e-05, "elbo": 3.44534, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.435432793807178}
{"step": 92690, "loss": 3.4801517486572267, "grad_norm": 0.7289351224899292, "learning_rate": 8.319500545850608e-05, "elbo": 3.48015, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.435726014543748}
{"step": 92695, "loss": 3.4799579620361327, "grad_norm": 0.7672290205955505, "learning_rate": 8.313623078512869e-05, "elbo": 3.47996, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.436019235280319}
{"step": 92700, "loss": 3.4981998443603515, "grad_norm": 0.9022172689437866, "learning_rate": 8.30773743513958e-05, "elbo": 3.4982, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.436312456016889}
{"step": 92705, "loss": 3.468502426147461, "grad_norm": 0.7240528464317322, "learning_rate": 8.301843630252985e-05, "elbo": 3.4685, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.43660567675346}
{"step": 92710, "loss": 3.467173385620117, "grad_norm": 0.7524003982543945, "learning_rate": 8.295941678395457e-05, "elbo": 3.46717, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.43689889749003}
{"step": 92715, "loss": 3.4626739501953123, "grad_norm": 0.8731392025947571, "learning_rate": 8.29003159412948e-05, "elbo": 3.46267, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.437192118226601}
{"step": 92720, "loss": 3.482338714599609, "grad_norm": 0.8776924014091492, "learning_rate": 8.284113392037593e-05, "elbo": 3.48234, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.437485338963172}
{"step": 92725, "loss": 3.4643810272216795, "grad_norm": 1.05470871925354, "learning_rate": 8.278187086722377e-05, "elbo": 3.46438, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.437778559699742}
{"step": 92725, "loss": 3.4643810272216795, "grad_norm": 1.05470871925354, "learning_rate": 8.278187086722377e-05, "elbo": 3.46438, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.437778559699742}
{"step": 92730, "loss": 3.485045623779297, "grad_norm": 0.6982539296150208, "learning_rate": 8.272252692806402e-05, "elbo": 3.48505, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.438071780436313}
{"step": 92730, "loss": 3.485045623779297, "grad_norm": 0.6982539296150208, "learning_rate": 8.272252692806402e-05, "elbo": 3.48505, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.438071780436313}
{"step": 92735, "loss": 3.476871871948242, "grad_norm": 0.7501205205917358, "learning_rate": 8.26631022493219e-05, "elbo": 3.47687, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.438365001172883}
{"step": 92740, "loss": 3.4612998962402344, "grad_norm": 0.7090959548950195, "learning_rate": 8.260359697762193e-05, "elbo": 3.4613, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.438658221909454}
{"step": 92740, "loss": 3.4612998962402344, "grad_norm": 0.7090959548950195, "learning_rate": 8.260359697762193e-05, "elbo": 3.4613, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.438658221909454}
{"step": 92745, "loss": 3.483446502685547, "grad_norm": 0.8155034780502319, "learning_rate": 8.254401125978743e-05, "elbo": 3.48345, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.438951442646024}
{"step": 92750, "loss": 3.462248611450195, "grad_norm": 0.6817589998245239, "learning_rate": 8.248434524284028e-05, "elbo": 3.46225, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.439244663382595}
{"step": 92755, "loss": 3.469477081298828, "grad_norm": 0.6900347471237183, "learning_rate": 8.242459907400039e-05, "elbo": 3.46948, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.439537884119165}
{"step": 92760, "loss": 3.4700218200683595, "grad_norm": 0.7062869071960449, "learning_rate": 8.236477290068554e-05, "elbo": 3.47002, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4398311048557355}
{"step": 92765, "loss": 3.482253646850586, "grad_norm": 0.7295018434524536, "learning_rate": 8.230486687051082e-05, "elbo": 3.48225, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.440124325592306}
{"step": 92770, "loss": 3.477037048339844, "grad_norm": 0.7159522771835327, "learning_rate": 8.224488113128845e-05, "elbo": 3.47704, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.440417546328876}
{"step": 92775, "loss": 3.4770816802978515, "grad_norm": 0.7589525580406189, "learning_rate": 8.218481583102726e-05, "elbo": 3.47708, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.440710767065447}
{"step": 92780, "loss": 3.4582305908203126, "grad_norm": 0.7331764698028564, "learning_rate": 8.212467111793239e-05, "elbo": 3.45823, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.441003987802017}
{"step": 92785, "loss": 3.4620628356933594, "grad_norm": 0.6914806365966797, "learning_rate": 8.206444714040495e-05, "elbo": 3.46206, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.441297208538588}
{"step": 92790, "loss": 3.449013519287109, "grad_norm": 0.7010610699653625, "learning_rate": 8.200414404704164e-05, "elbo": 3.44901, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.441590429275158}
{"step": 92795, "loss": 3.448286437988281, "grad_norm": 0.7558433413505554, "learning_rate": 8.194376198663434e-05, "elbo": 3.44829, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.441883650011729}
{"step": 92800, "loss": 3.467235565185547, "grad_norm": 0.7151975035667419, "learning_rate": 8.188330110816976e-05, "elbo": 3.46724, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.442176870748299}
{"step": 92805, "loss": 3.455544662475586, "grad_norm": 0.7245344519615173, "learning_rate": 8.18227615608291e-05, "elbo": 3.45554, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.44247009148487}
{"step": 92810, "loss": 3.4868621826171875, "grad_norm": 0.7358859181404114, "learning_rate": 8.176214349398774e-05, "elbo": 3.48686, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.44276331222144}
{"step": 92810, "loss": 3.4868621826171875, "grad_norm": 0.7358859181404114, "learning_rate": 8.176214349398774e-05, "elbo": 3.48686, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.44276331222144}
{"step": 92815, "loss": 3.4755401611328125, "grad_norm": 0.6956892609596252, "learning_rate": 8.170144705721465e-05, "elbo": 3.47554, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4430565329580105}
{"step": 92820, "loss": 3.4678165435791017, "grad_norm": 0.8239229917526245, "learning_rate": 8.164067240027233e-05, "elbo": 3.46782, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.443349753694581}
{"step": 92825, "loss": 3.449365234375, "grad_norm": 0.7519155740737915, "learning_rate": 8.157981967311614e-05, "elbo": 3.44937, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.443642974431151}
{"step": 92830, "loss": 3.4450424194335936, "grad_norm": 0.7057719230651855, "learning_rate": 8.151888902589419e-05, "elbo": 3.44504, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.443936195167722}
{"step": 92835, "loss": 3.4699165344238283, "grad_norm": 0.804535984992981, "learning_rate": 8.145788060894674e-05, "elbo": 3.46992, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.444229415904292}
{"step": 92840, "loss": 3.44824333190918, "grad_norm": 0.855837881565094, "learning_rate": 8.139679457280605e-05, "elbo": 3.44824, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.444522636640864}
{"step": 92845, "loss": 3.466871643066406, "grad_norm": 0.7662307620048523, "learning_rate": 8.13356310681958e-05, "elbo": 3.46687, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.444815857377434}
{"step": 92850, "loss": 3.4563247680664064, "grad_norm": 0.7198243737220764, "learning_rate": 8.127439024603088e-05, "elbo": 3.45632, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.445109078114005}
{"step": 92855, "loss": 3.4727462768554687, "grad_norm": 0.7399437427520752, "learning_rate": 8.12130722574169e-05, "elbo": 3.47275, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.445402298850575}
{"step": 92860, "loss": 3.4714393615722656, "grad_norm": 0.7807868123054504, "learning_rate": 8.115167725364995e-05, "elbo": 3.47144, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4456955195871455}
{"step": 92865, "loss": 3.454523468017578, "grad_norm": 0.7330150604248047, "learning_rate": 8.109020538621606e-05, "elbo": 3.45452, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.445988740323716}
{"step": 92865, "loss": 3.454523468017578, "grad_norm": 0.7330150604248047, "learning_rate": 8.109020538621606e-05, "elbo": 3.45452, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.445988740323716}
{"step": 92870, "loss": 3.4718841552734374, "grad_norm": 0.7278215885162354, "learning_rate": 8.102865680679099e-05, "elbo": 3.47188, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.446281961060286}
{"step": 92875, "loss": 3.460631561279297, "grad_norm": 0.7497953772544861, "learning_rate": 8.096703166723968e-05, "elbo": 3.46063, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.446575181796857}
{"step": 92880, "loss": 3.459290313720703, "grad_norm": 0.7450479865074158, "learning_rate": 8.090533011961609e-05, "elbo": 3.45929, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.446868402533427}
{"step": 92885, "loss": 3.4729072570800783, "grad_norm": 0.7063602209091187, "learning_rate": 8.084355231616265e-05, "elbo": 3.47291, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.447161623269998}
{"step": 92890, "loss": 3.470958709716797, "grad_norm": 0.7096590399742126, "learning_rate": 8.078169840930993e-05, "elbo": 3.47096, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.447454844006568}
{"step": 92895, "loss": 3.4695934295654296, "grad_norm": 0.7391175627708435, "learning_rate": 8.071976855167629e-05, "elbo": 3.46959, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.447748064743139}
{"step": 92900, "loss": 3.4589366912841797, "grad_norm": 0.7812890410423279, "learning_rate": 8.065776289606752e-05, "elbo": 3.45894, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.448041285479709}
{"step": 92905, "loss": 3.450872802734375, "grad_norm": 1.1108362674713135, "learning_rate": 8.05956815954764e-05, "elbo": 3.45087, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.44833450621628}
{"step": 92910, "loss": 3.4640121459960938, "grad_norm": 0.7871367931365967, "learning_rate": 8.053352480308239e-05, "elbo": 3.46401, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.44862772695285}
{"step": 92915, "loss": 3.486613464355469, "grad_norm": 0.7495467066764832, "learning_rate": 8.047129267225115e-05, "elbo": 3.48661, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4489209476894205}
{"step": 92920, "loss": 3.4685199737548826, "grad_norm": 0.7480962872505188, "learning_rate": 8.040898535653432e-05, "elbo": 3.46852, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.449214168425991}
{"step": 92925, "loss": 3.450487518310547, "grad_norm": 0.7336888313293457, "learning_rate": 8.034660300966898e-05, "elbo": 3.45049, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.449507389162561}
{"step": 92930, "loss": 3.4482452392578127, "grad_norm": 0.700261116027832, "learning_rate": 8.028414578557738e-05, "elbo": 3.44825, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.449800609899132}
{"step": 92935, "loss": 3.465596008300781, "grad_norm": 0.7260714173316956, "learning_rate": 8.022161383836652e-05, "elbo": 3.4656, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.450093830635702}
{"step": 92940, "loss": 3.4668514251708986, "grad_norm": 0.7720879912376404, "learning_rate": 8.015900732232775e-05, "elbo": 3.46685, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.450387051372273}
{"step": 92945, "loss": 3.4693092346191405, "grad_norm": 0.7308007478713989, "learning_rate": 8.009632639193643e-05, "elbo": 3.46931, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.450680272108843}
{"step": 92950, "loss": 3.468259811401367, "grad_norm": 0.6953151226043701, "learning_rate": 8.003357120185154e-05, "elbo": 3.46826, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.450973492845414}
{"step": 92955, "loss": 3.4790802001953125, "grad_norm": 0.7666018605232239, "learning_rate": 7.997074190691522e-05, "elbo": 3.47908, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.451266713581984}
{"step": 92960, "loss": 3.4596229553222657, "grad_norm": 0.6836732029914856, "learning_rate": 7.990783866215259e-05, "elbo": 3.45962, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.451559934318555}
{"step": 92965, "loss": 3.470289611816406, "grad_norm": 0.777256429195404, "learning_rate": 7.98448616227711e-05, "elbo": 3.47029, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.451853155055126}
{"step": 92970, "loss": 3.4674877166748046, "grad_norm": 0.7519968748092651, "learning_rate": 7.978181094416033e-05, "elbo": 3.46749, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.452146375791696}
{"step": 92975, "loss": 3.484545135498047, "grad_norm": 0.7602331042289734, "learning_rate": 7.971868678189161e-05, "elbo": 3.48455, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.452439596528267}
{"step": 92980, "loss": 3.4678768157958983, "grad_norm": 0.7511187195777893, "learning_rate": 7.965548929171746e-05, "elbo": 3.46788, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.452732817264837}
{"step": 92985, "loss": 3.465314483642578, "grad_norm": 0.7221029996871948, "learning_rate": 7.959221862957148e-05, "elbo": 3.46531, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.453026038001408}
{"step": 92990, "loss": 3.4661186218261717, "grad_norm": 0.7072005271911621, "learning_rate": 7.95288749515677e-05, "elbo": 3.46612, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.453319258737978}
{"step": 92995, "loss": 3.4531349182128905, "grad_norm": 0.8836337327957153, "learning_rate": 7.946545841400035e-05, "elbo": 3.45313, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.453612479474549}
{"step": 93000, "loss": 3.447030258178711, "grad_norm": 0.7517287135124207, "learning_rate": 7.940196917334343e-05, "elbo": 3.44703, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.453905700211119}
{"step": 93005, "loss": 3.4630828857421876, "grad_norm": 0.7531781196594238, "learning_rate": 7.933840738625036e-05, "elbo": 3.46308, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4541989209476895}
{"step": 93010, "loss": 3.4716552734375, "grad_norm": 0.8663654923439026, "learning_rate": 7.927477320955351e-05, "elbo": 3.47166, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.45449214168426}
{"step": 93015, "loss": 3.465955352783203, "grad_norm": 0.7395907044410706, "learning_rate": 7.921106680026387e-05, "elbo": 3.46596, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4547853624208305}
{"step": 93020, "loss": 3.470967102050781, "grad_norm": 0.7753058075904846, "learning_rate": 7.91472883155707e-05, "elbo": 3.47097, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.455078583157401}
{"step": 93025, "loss": 3.4536094665527344, "grad_norm": 0.7954479455947876, "learning_rate": 7.908343791284104e-05, "elbo": 3.45361, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.455371803893971}
{"step": 93030, "loss": 3.461710739135742, "grad_norm": 0.7672436237335205, "learning_rate": 7.901951574961943e-05, "elbo": 3.46171, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.455665024630542}
{"step": 93035, "loss": 3.4732589721679688, "grad_norm": 0.7067698836326599, "learning_rate": 7.895552198362748e-05, "elbo": 3.47326, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.455958245367112}
{"step": 93040, "loss": 3.4660003662109373, "grad_norm": 0.7326518297195435, "learning_rate": 7.88914567727634e-05, "elbo": 3.466, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.456251466103683}
{"step": 93045, "loss": 3.4679412841796875, "grad_norm": 0.7162637114524841, "learning_rate": 7.882732027510174e-05, "elbo": 3.46794, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.456544686840253}
{"step": 93050, "loss": 3.453574371337891, "grad_norm": 0.7978881597518921, "learning_rate": 7.876311264889296e-05, "elbo": 3.45357, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.456837907576824}
{"step": 93055, "loss": 3.4608062744140624, "grad_norm": 0.7622168064117432, "learning_rate": 7.869883405256295e-05, "elbo": 3.46081, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.457131128313394}
{"step": 93060, "loss": 3.4787059783935548, "grad_norm": 0.786848783493042, "learning_rate": 7.863448464471281e-05, "elbo": 3.47871, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4574243490499645}
{"step": 93065, "loss": 3.4721389770507813, "grad_norm": 0.7429934144020081, "learning_rate": 7.857006458411826e-05, "elbo": 3.47214, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.457717569786535}
{"step": 93070, "loss": 3.480356979370117, "grad_norm": 1.0447499752044678, "learning_rate": 7.850557402972942e-05, "elbo": 3.48036, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4580107905231054}
{"step": 93075, "loss": 3.459213638305664, "grad_norm": 0.7347282767295837, "learning_rate": 7.844101314067032e-05, "elbo": 3.45921, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.458304011259676}
{"step": 93075, "loss": 3.459213638305664, "grad_norm": 0.7347282767295837, "learning_rate": 7.844101314067032e-05, "elbo": 3.45921, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.458304011259676}
{"step": 93080, "loss": 3.467560577392578, "grad_norm": 0.8785656094551086, "learning_rate": 7.837638207623854e-05, "elbo": 3.46756, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.458597231996247}
{"step": 93085, "loss": 3.4600570678710936, "grad_norm": 0.7243924140930176, "learning_rate": 7.831168099590478e-05, "elbo": 3.46006, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.458890452732817}
{"step": 93090, "loss": 3.470387649536133, "grad_norm": 0.7020840048789978, "learning_rate": 7.824691005931255e-05, "elbo": 3.47039, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.459183673469388}
{"step": 93095, "loss": 3.4669586181640626, "grad_norm": 0.7217199802398682, "learning_rate": 7.81820694262777e-05, "elbo": 3.46696, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.459476894205959}
{"step": 93095, "loss": 3.4669586181640626, "grad_norm": 0.7217199802398682, "learning_rate": 7.81820694262777e-05, "elbo": 3.46696, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.459476894205959}
{"step": 93100, "loss": 3.4584808349609375, "grad_norm": 0.9272100329399109, "learning_rate": 7.811715925678804e-05, "elbo": 3.45848, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.459770114942529}
{"step": 93105, "loss": 3.4639183044433595, "grad_norm": 0.7420610189437866, "learning_rate": 7.805217971100295e-05, "elbo": 3.46392, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4600633356790995}
{"step": 93110, "loss": 3.458574676513672, "grad_norm": 0.8593958616256714, "learning_rate": 7.798713094925301e-05, "elbo": 3.45857, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.46035655641567}
{"step": 93115, "loss": 3.4609771728515626, "grad_norm": 0.7082753777503967, "learning_rate": 7.792201313203957e-05, "elbo": 3.46098, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.46064977715224}
{"step": 93120, "loss": 3.4535064697265625, "grad_norm": 0.7193999886512756, "learning_rate": 7.785682642003437e-05, "elbo": 3.45351, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.460942997888811}
{"step": 93125, "loss": 3.4692432403564455, "grad_norm": 0.7480447888374329, "learning_rate": 7.779157097407915e-05, "elbo": 3.46924, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.461236218625381}
{"step": 93130, "loss": 3.461439514160156, "grad_norm": 0.7272252440452576, "learning_rate": 7.772624695518522e-05, "elbo": 3.46144, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.461529439361952}
{"step": 93135, "loss": 3.4646652221679686, "grad_norm": 0.694148600101471, "learning_rate": 7.766085452453312e-05, "elbo": 3.46467, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.461822660098522}
{"step": 93140, "loss": 3.457117462158203, "grad_norm": 0.7611686587333679, "learning_rate": 7.759539384347216e-05, "elbo": 3.45712, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.462115880835093}
{"step": 93145, "loss": 3.469010925292969, "grad_norm": 0.7590054273605347, "learning_rate": 7.752986507352008e-05, "elbo": 3.46901, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.462409101571663}
{"step": 93150, "loss": 3.481338882446289, "grad_norm": 0.7904376983642578, "learning_rate": 7.746426837636259e-05, "elbo": 3.48134, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.462702322308234}
{"step": 93155, "loss": 3.463998794555664, "grad_norm": 0.6917633414268494, "learning_rate": 7.739860391385303e-05, "elbo": 3.464, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.462995543044804}
{"step": 93160, "loss": 3.471760559082031, "grad_norm": 0.7319877743721008, "learning_rate": 7.733287184801191e-05, "elbo": 3.47176, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4632887637813745}
{"step": 93165, "loss": 3.4554519653320312, "grad_norm": 0.6748159527778625, "learning_rate": 7.726707234102659e-05, "elbo": 3.45545, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.463581984517945}
{"step": 93170, "loss": 3.4626182556152343, "grad_norm": 0.7230044603347778, "learning_rate": 7.720120555525081e-05, "elbo": 3.46262, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.463875205254515}
{"step": 93175, "loss": 3.454778289794922, "grad_norm": 0.7823047637939453, "learning_rate": 7.713527165320432e-05, "elbo": 3.45478, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.464168425991086}
{"step": 93180, "loss": 3.4838516235351564, "grad_norm": 0.7536402940750122, "learning_rate": 7.706927079757247e-05, "elbo": 3.48385, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.464461646727656}
{"step": 93185, "loss": 3.4676647186279297, "grad_norm": 0.7038094401359558, "learning_rate": 7.70032031512058e-05, "elbo": 3.46766, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.464754867464227}
{"step": 93190, "loss": 3.4479820251464846, "grad_norm": 0.9577478170394897, "learning_rate": 7.693706887711965e-05, "elbo": 3.44798, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.465048088200797}
{"step": 93195, "loss": 3.465153121948242, "grad_norm": 0.6844855546951294, "learning_rate": 7.687086813849378e-05, "elbo": 3.46515, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.465341308937368}
{"step": 93200, "loss": 3.4660247802734374, "grad_norm": 0.7146247029304504, "learning_rate": 7.680460109867195e-05, "elbo": 3.46602, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.465634529673938}
{"step": 93205, "loss": 3.4556705474853517, "grad_norm": 0.7693608403205872, "learning_rate": 7.673826792116145e-05, "elbo": 3.45567, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4659277504105095}
{"step": 93210, "loss": 3.461182403564453, "grad_norm": 0.7114871740341187, "learning_rate": 7.667186876963286e-05, "elbo": 3.46118, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.466220971147079}
{"step": 93215, "loss": 3.470905303955078, "grad_norm": 0.7227243781089783, "learning_rate": 7.660540380791942e-05, "elbo": 3.47091, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.46651419188365}
{"step": 93220, "loss": 3.4750701904296877, "grad_norm": 0.7562541365623474, "learning_rate": 7.653887320001688e-05, "elbo": 3.47507, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.466807412620221}
{"step": 93225, "loss": 3.4622482299804687, "grad_norm": 0.7500104308128357, "learning_rate": 7.647227711008287e-05, "elbo": 3.46225, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.467100633356791}
{"step": 93230, "loss": 3.4765090942382812, "grad_norm": 0.7234870195388794, "learning_rate": 7.640561570243661e-05, "elbo": 3.47651, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.467393854093362}
{"step": 93235, "loss": 3.461986541748047, "grad_norm": 0.6822853088378906, "learning_rate": 7.63388891415585e-05, "elbo": 3.46199, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.467687074829932}
{"step": 93240, "loss": 3.4522140502929686, "grad_norm": 0.8777136206626892, "learning_rate": 7.627209759208971e-05, "elbo": 3.45221, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.467980295566503}
{"step": 93245, "loss": 3.4722610473632813, "grad_norm": 0.8441575765609741, "learning_rate": 7.620524121883174e-05, "elbo": 3.47226, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.468273516303073}
{"step": 93250, "loss": 3.4629112243652345, "grad_norm": 0.6901426911354065, "learning_rate": 7.613832018674606e-05, "elbo": 3.46291, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.468566737039644}
{"step": 93255, "loss": 3.468459701538086, "grad_norm": 0.7387069463729858, "learning_rate": 7.607133466095365e-05, "elbo": 3.46846, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.468859957776214}
{"step": 93260, "loss": 3.4711612701416015, "grad_norm": 0.7189489006996155, "learning_rate": 7.600428480673463e-05, "elbo": 3.47116, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4691531785127845}
{"step": 93265, "loss": 3.4754920959472657, "grad_norm": 0.7178307175636292, "learning_rate": 7.593717078952788e-05, "elbo": 3.47549, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.469446399249355}
{"step": 93270, "loss": 3.4706047058105467, "grad_norm": 0.7360876798629761, "learning_rate": 7.586999277493053e-05, "elbo": 3.4706, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.469739619985925}
{"step": 93275, "loss": 3.484613037109375, "grad_norm": 0.719468891620636, "learning_rate": 7.580275092869766e-05, "elbo": 3.48461, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.470032840722496}
{"step": 93280, "loss": 3.4636455535888673, "grad_norm": 0.7610522508621216, "learning_rate": 7.573544541674187e-05, "elbo": 3.46365, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.470326061459066}
{"step": 93285, "loss": 3.4771827697753905, "grad_norm": 0.8448511362075806, "learning_rate": 7.566807640513278e-05, "elbo": 3.47718, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.470619282195637}
{"step": 93290, "loss": 3.4798210144042967, "grad_norm": 0.7648704051971436, "learning_rate": 7.560064406009677e-05, "elbo": 3.47982, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.470912502932207}
{"step": 93295, "loss": 3.4475399017333985, "grad_norm": 0.8001687526702881, "learning_rate": 7.553314854801641e-05, "elbo": 3.44754, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.471205723668778}
{"step": 93300, "loss": 3.4613372802734377, "grad_norm": 0.7067041397094727, "learning_rate": 7.546559003543019e-05, "elbo": 3.46134, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.471498944405348}
{"step": 93305, "loss": 3.4654937744140626, "grad_norm": 0.8657945990562439, "learning_rate": 7.539796868903199e-05, "elbo": 3.46549, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.471792165141919}
{"step": 93310, "loss": 3.4572769165039063, "grad_norm": 0.731242835521698, "learning_rate": 7.533028467567083e-05, "elbo": 3.45728, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.472085385878489}
{"step": 93315, "loss": 3.451830291748047, "grad_norm": 0.6950441598892212, "learning_rate": 7.526253816235023e-05, "elbo": 3.45183, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4723786066150595}
{"step": 93320, "loss": 3.4480804443359374, "grad_norm": 0.7055776119232178, "learning_rate": 7.519472931622797e-05, "elbo": 3.44808, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.47267182735163}
{"step": 93325, "loss": 3.467633056640625, "grad_norm": 0.7010436654090881, "learning_rate": 7.512685830461568e-05, "elbo": 3.46763, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4729650480882}
{"step": 93330, "loss": 3.4603042602539062, "grad_norm": 0.7714749574661255, "learning_rate": 7.50589252949783e-05, "elbo": 3.4603, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.473258268824772}
{"step": 93335, "loss": 3.454949951171875, "grad_norm": 0.7216790914535522, "learning_rate": 7.499093045493379e-05, "elbo": 3.45495, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.473551489561342}
{"step": 93340, "loss": 3.458528518676758, "grad_norm": 0.7422671914100647, "learning_rate": 7.492287395225265e-05, "elbo": 3.45853, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.473844710297913}
{"step": 93345, "loss": 3.460305023193359, "grad_norm": 0.7128708958625793, "learning_rate": 7.485475595485756e-05, "elbo": 3.46031, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.474137931034483}
{"step": 93350, "loss": 3.456407165527344, "grad_norm": 0.8494316935539246, "learning_rate": 7.478657663082287e-05, "elbo": 3.45641, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4744311517710535}
{"step": 93355, "loss": 3.4558563232421875, "grad_norm": 0.6844720840454102, "learning_rate": 7.471833614837431e-05, "elbo": 3.45586, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.474724372507624}
{"step": 93360, "loss": 3.452967071533203, "grad_norm": 0.7287129163742065, "learning_rate": 7.465003467588847e-05, "elbo": 3.45297, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4750175932441945}
{"step": 93365, "loss": 3.4617874145507814, "grad_norm": 0.7731466293334961, "learning_rate": 7.458167238189248e-05, "elbo": 3.46179, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.475310813980765}
{"step": 93370, "loss": 3.4570266723632814, "grad_norm": 0.7110713124275208, "learning_rate": 7.451324943506347e-05, "elbo": 3.45703, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.475604034717335}
{"step": 93375, "loss": 3.4802734375, "grad_norm": 0.7973568439483643, "learning_rate": 7.444476600422828e-05, "elbo": 3.48027, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.475897255453906}
{"step": 93375, "loss": 3.4802734375, "grad_norm": 0.7973568439483643, "learning_rate": 7.444476600422828e-05, "elbo": 3.48027, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.475897255453906}
{"step": 93380, "loss": 3.4476394653320312, "grad_norm": 0.8125832676887512, "learning_rate": 7.437622225836294e-05, "elbo": 3.44764, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.476190476190476}
{"step": 93385, "loss": 3.441086196899414, "grad_norm": 0.7621387839317322, "learning_rate": 7.430761836659235e-05, "elbo": 3.44109, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.476483696927047}
{"step": 93390, "loss": 3.449565124511719, "grad_norm": 0.8932515978813171, "learning_rate": 7.423895449818979e-05, "elbo": 3.44956, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.476776917663617}
{"step": 93395, "loss": 3.4652694702148437, "grad_norm": 0.7104986906051636, "learning_rate": 7.417023082257652e-05, "elbo": 3.46527, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.477070138400188}
{"step": 93400, "loss": 3.452471160888672, "grad_norm": 0.7893259525299072, "learning_rate": 7.41014475093214e-05, "elbo": 3.45247, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.477363359136758}
{"step": 93405, "loss": 3.456686019897461, "grad_norm": 0.7814929485321045, "learning_rate": 7.403260472814039e-05, "elbo": 3.45669, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4776565798733285}
{"step": 93410, "loss": 3.458236312866211, "grad_norm": 0.8056389093399048, "learning_rate": 7.396370264889624e-05, "elbo": 3.45824, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.477949800609899}
{"step": 93415, "loss": 3.461600112915039, "grad_norm": 0.7329570055007935, "learning_rate": 7.389474144159796e-05, "elbo": 3.4616, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4782430213464695}
{"step": 93420, "loss": 3.463622283935547, "grad_norm": 0.7837713956832886, "learning_rate": 7.382572127640048e-05, "elbo": 3.46362, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.47853624208304}
{"step": 93425, "loss": 3.456981658935547, "grad_norm": 0.7165889739990234, "learning_rate": 7.375664232360421e-05, "elbo": 3.45698, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.47882946281961}
{"step": 93430, "loss": 3.472148895263672, "grad_norm": 0.7137104272842407, "learning_rate": 7.368750475365455e-05, "elbo": 3.47215, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.479122683556181}
{"step": 93430, "loss": 3.472148895263672, "grad_norm": 0.7137104272842407, "learning_rate": 7.368750475365455e-05, "elbo": 3.47215, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.479122683556181}
{"step": 93435, "loss": 3.4566810607910154, "grad_norm": 0.702700138092041, "learning_rate": 7.361830873714165e-05, "elbo": 3.45668, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.479415904292751}
{"step": 93440, "loss": 3.449247360229492, "grad_norm": 0.7578864693641663, "learning_rate": 7.354905444479975e-05, "elbo": 3.44925, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.479709125029322}
{"step": 93440, "loss": 3.449247360229492, "grad_norm": 0.7578864693641663, "learning_rate": 7.354905444479975e-05, "elbo": 3.44925, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.479709125029322}
{"step": 93445, "loss": 3.451116180419922, "grad_norm": 0.71245938539505, "learning_rate": 7.347974204750696e-05, "elbo": 3.45112, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.480002345765892}
{"step": 93450, "loss": 3.4611183166503907, "grad_norm": 0.7283444404602051, "learning_rate": 7.34103717162847e-05, "elbo": 3.46112, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.480295566502463}
{"step": 93455, "loss": 3.4582229614257813, "grad_norm": 0.6972630620002747, "learning_rate": 7.33409436222974e-05, "elbo": 3.45822, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.480588787239034}
{"step": 93460, "loss": 3.459309387207031, "grad_norm": 0.6807122826576233, "learning_rate": 7.327145793685197e-05, "elbo": 3.45931, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.480882007975604}
{"step": 93465, "loss": 3.454120635986328, "grad_norm": 0.6938725113868713, "learning_rate": 7.320191483139742e-05, "elbo": 3.45412, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.481175228712175}
{"step": 93470, "loss": 3.4742542266845704, "grad_norm": 0.6902425289154053, "learning_rate": 7.313231447752447e-05, "elbo": 3.47425, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.481468449448745}
{"step": 93475, "loss": 3.450387191772461, "grad_norm": 0.6756194829940796, "learning_rate": 7.306265704696504e-05, "elbo": 3.45039, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.481761670185316}
{"step": 93480, "loss": 3.4418617248535157, "grad_norm": 0.7241092920303345, "learning_rate": 7.299294271159196e-05, "elbo": 3.44186, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.482054890921886}
{"step": 93485, "loss": 3.4589427947998046, "grad_norm": 0.740845263004303, "learning_rate": 7.29231716434184e-05, "elbo": 3.45894, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.482348111658457}
{"step": 93490, "loss": 3.460162353515625, "grad_norm": 0.762941300868988, "learning_rate": 7.285334401459752e-05, "elbo": 3.46016, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.482641332395027}
{"step": 93495, "loss": 3.4652576446533203, "grad_norm": 0.7882974743843079, "learning_rate": 7.278345999742208e-05, "elbo": 3.46526, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.482934553131598}
{"step": 93500, "loss": 3.4669750213623045, "grad_norm": 0.9339042901992798, "learning_rate": 7.271351976432393e-05, "elbo": 3.46698, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.483227773868168}
{"step": 93505, "loss": 3.448019790649414, "grad_norm": 0.7298359870910645, "learning_rate": 7.264352348787364e-05, "elbo": 3.44802, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4835209946047385}
{"step": 93510, "loss": 3.4324501037597654, "grad_norm": 0.8462153077125549, "learning_rate": 7.257347134078007e-05, "elbo": 3.43245, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.483814215341309}
{"step": 93515, "loss": 3.4503746032714844, "grad_norm": 0.7244088649749756, "learning_rate": 7.250336349588994e-05, "elbo": 3.45037, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.484107436077879}
{"step": 93515, "loss": 3.4503746032714844, "grad_norm": 0.7244088649749756, "learning_rate": 7.250336349588994e-05, "elbo": 3.45037, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.484107436077879}
{"step": 93520, "loss": 3.437066650390625, "grad_norm": 0.8822525143623352, "learning_rate": 7.243320012618737e-05, "elbo": 3.43707, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.48440065681445}
{"step": 93525, "loss": 3.460182952880859, "grad_norm": 0.7308211922645569, "learning_rate": 7.236298140479351e-05, "elbo": 3.46018, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.48469387755102}
{"step": 93530, "loss": 3.4587158203125, "grad_norm": 0.6833402514457703, "learning_rate": 7.229270750496606e-05, "elbo": 3.45872, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.484987098287591}
{"step": 93535, "loss": 3.461943817138672, "grad_norm": 0.6951966881752014, "learning_rate": 7.222237860009892e-05, "elbo": 3.46194, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.485280319024161}
{"step": 93540, "loss": 3.466382598876953, "grad_norm": 0.7274866104125977, "learning_rate": 7.215199486372162e-05, "elbo": 3.46638, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.485573539760732}
{"step": 93545, "loss": 3.455069351196289, "grad_norm": 0.7886032462120056, "learning_rate": 7.208155646949908e-05, "elbo": 3.45507, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.485866760497302}
{"step": 93550, "loss": 3.473271942138672, "grad_norm": 0.7321065664291382, "learning_rate": 7.2011063591231e-05, "elbo": 3.47327, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.486159981233873}
{"step": 93555, "loss": 3.4523059844970705, "grad_norm": 0.7356703281402588, "learning_rate": 7.194051640285157e-05, "elbo": 3.45231, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.486453201970443}
{"step": 93560, "loss": 3.460736083984375, "grad_norm": 0.8128666877746582, "learning_rate": 7.186991507842894e-05, "elbo": 3.46074, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4867464227070135}
{"step": 93565, "loss": 3.446916961669922, "grad_norm": 0.755981981754303, "learning_rate": 7.179925979216491e-05, "elbo": 3.44692, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.487039643443584}
{"step": 93570, "loss": 3.4652198791503905, "grad_norm": 0.7921065092086792, "learning_rate": 7.172855071839432e-05, "elbo": 3.46522, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.487332864180154}
{"step": 93575, "loss": 3.463825225830078, "grad_norm": 0.9539082646369934, "learning_rate": 7.16577880315848e-05, "elbo": 3.46383, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.487626084916725}
{"step": 93580, "loss": 3.448020172119141, "grad_norm": 0.7174724340438843, "learning_rate": 7.158697190633627e-05, "elbo": 3.44802, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.487919305653296}
{"step": 93585, "loss": 3.4504066467285157, "grad_norm": 0.7016782164573669, "learning_rate": 7.151610251738045e-05, "elbo": 3.45041, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.488212526389867}
{"step": 93590, "loss": 3.456755828857422, "grad_norm": 0.7358895540237427, "learning_rate": 7.144518003958051e-05, "elbo": 3.45676, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.488505747126437}
{"step": 93595, "loss": 3.4473358154296876, "grad_norm": 0.7466626763343811, "learning_rate": 7.137420464793063e-05, "elbo": 3.44734, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.488798967863008}
{"step": 93595, "loss": 3.4473358154296876, "grad_norm": 0.7466626763343811, "learning_rate": 7.137420464793063e-05, "elbo": 3.44734, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.488798967863008}
{"step": 93600, "loss": 3.4600433349609374, "grad_norm": 0.7465209364891052, "learning_rate": 7.130317651755554e-05, "elbo": 3.46004, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.489092188599578}
{"step": 93605, "loss": 3.4428176879882812, "grad_norm": 0.7378808856010437, "learning_rate": 7.123209582371006e-05, "elbo": 3.44282, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4893854093361485}
{"step": 93610, "loss": 3.46434326171875, "grad_norm": 0.7215643525123596, "learning_rate": 7.116096274177878e-05, "elbo": 3.46434, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.489678630072719}
{"step": 93615, "loss": 3.4844440460205077, "grad_norm": 1.1559772491455078, "learning_rate": 7.108977744727547e-05, "elbo": 3.48444, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.489971850809289}
{"step": 93620, "loss": 3.4517669677734375, "grad_norm": 0.7253875136375427, "learning_rate": 7.101854011584277e-05, "elbo": 3.45177, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.49026507154586}
{"step": 93625, "loss": 3.4700382232666014, "grad_norm": 0.7971612215042114, "learning_rate": 7.094725092325176e-05, "elbo": 3.47004, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.49055829228243}
{"step": 93630, "loss": 3.456945037841797, "grad_norm": 0.7537063956260681, "learning_rate": 7.087591004540137e-05, "elbo": 3.45695, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.490851513019001}
{"step": 93635, "loss": 3.4490562438964845, "grad_norm": 0.6936315894126892, "learning_rate": 7.080451765831817e-05, "elbo": 3.44906, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.491144733755571}
{"step": 93640, "loss": 3.4604568481445312, "grad_norm": 0.7462077140808105, "learning_rate": 7.073307393815576e-05, "elbo": 3.46046, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.491437954492142}
{"step": 93640, "loss": 3.4604568481445312, "grad_norm": 0.7462077140808105, "learning_rate": 7.073307393815576e-05, "elbo": 3.46046, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.491437954492142}
{"step": 93645, "loss": 3.4643993377685547, "grad_norm": 0.7397443056106567, "learning_rate": 7.066157906119441e-05, "elbo": 3.4644, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.491731175228712}
{"step": 93650, "loss": 3.451914978027344, "grad_norm": 0.7259619235992432, "learning_rate": 7.059003320384065e-05, "elbo": 3.45191, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.492024395965283}
{"step": 93655, "loss": 3.453874206542969, "grad_norm": 0.6769172549247742, "learning_rate": 7.051843654262676e-05, "elbo": 3.45387, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.492317616701853}
{"step": 93655, "loss": 3.453874206542969, "grad_norm": 0.6769172549247742, "learning_rate": 7.051843654262676e-05, "elbo": 3.45387, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.492317616701853}
{"step": 93660, "loss": 3.4822532653808596, "grad_norm": 0.7598552107810974, "learning_rate": 7.044678925421036e-05, "elbo": 3.48225, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4926108374384235}
{"step": 93665, "loss": 3.451039505004883, "grad_norm": 0.7488466501235962, "learning_rate": 7.037509151537403e-05, "elbo": 3.45104, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.492904058174994}
{"step": 93670, "loss": 3.451054000854492, "grad_norm": 0.6808662414550781, "learning_rate": 7.030334350302483e-05, "elbo": 3.45105, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.493197278911564}
{"step": 93675, "loss": 3.4519081115722656, "grad_norm": 0.7175337672233582, "learning_rate": 7.023154539419384e-05, "elbo": 3.45191, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.493490499648135}
{"step": 93680, "loss": 3.450743865966797, "grad_norm": 1.0024651288986206, "learning_rate": 7.015969736603573e-05, "elbo": 3.45074, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.493783720384705}
{"step": 93685, "loss": 3.4630401611328123, "grad_norm": 0.7453929781913757, "learning_rate": 7.008779959582837e-05, "elbo": 3.46304, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.494076941121276}
{"step": 93685, "loss": 3.4630401611328123, "grad_norm": 0.7453929781913757, "learning_rate": 7.008779959582837e-05, "elbo": 3.46304, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.494076941121276}
{"step": 93690, "loss": 3.467326354980469, "grad_norm": 0.7473202347755432, "learning_rate": 7.00158522609724e-05, "elbo": 3.46733, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.494370161857846}
{"step": 93695, "loss": 3.462401580810547, "grad_norm": 0.8235034942626953, "learning_rate": 6.994385553899069e-05, "elbo": 3.4624, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.494663382594417}
{"step": 93700, "loss": 3.4647674560546875, "grad_norm": 0.7269942760467529, "learning_rate": 6.9871809607528e-05, "elbo": 3.46477, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.494956603330987}
{"step": 93705, "loss": 3.4663414001464843, "grad_norm": 0.7161886692047119, "learning_rate": 6.97997146443505e-05, "elbo": 3.46634, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4952498240675585}
{"step": 93710, "loss": 3.4603614807128906, "grad_norm": 1.0243207216262817, "learning_rate": 6.972757082734534e-05, "elbo": 3.46036, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.495543044804129}
{"step": 93715, "loss": 3.4431831359863283, "grad_norm": 0.7276797294616699, "learning_rate": 6.965537833452024e-05, "elbo": 3.44318, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.495836265540699}
{"step": 93720, "loss": 3.4570220947265624, "grad_norm": 0.7461420893669128, "learning_rate": 6.958313734400295e-05, "elbo": 3.45702, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.49612948627727}
{"step": 93725, "loss": 3.460759735107422, "grad_norm": 0.755139172077179, "learning_rate": 6.9510848034041e-05, "elbo": 3.46076, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.49642270701384}
{"step": 93730, "loss": 3.466038131713867, "grad_norm": 0.7521592974662781, "learning_rate": 6.943851058300103e-05, "elbo": 3.46604, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.496715927750411}
{"step": 93735, "loss": 3.4608474731445313, "grad_norm": 1.1025002002716064, "learning_rate": 6.936612516936852e-05, "elbo": 3.46085, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.497009148486981}
{"step": 93740, "loss": 3.474468994140625, "grad_norm": 0.6822832226753235, "learning_rate": 6.929369197174728e-05, "elbo": 3.47447, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.497302369223552}
{"step": 93745, "loss": 3.4557125091552736, "grad_norm": 0.722571849822998, "learning_rate": 6.922121116885904e-05, "elbo": 3.45571, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.497595589960122}
{"step": 93750, "loss": 3.4673362731933595, "grad_norm": 0.7201598286628723, "learning_rate": 6.914868293954297e-05, "elbo": 3.46734, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4978888106966926}
{"step": 93755, "loss": 3.4601730346679687, "grad_norm": 0.7259930372238159, "learning_rate": 6.907610746275523e-05, "elbo": 3.46017, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.498182031433263}
{"step": 93760, "loss": 3.4550689697265624, "grad_norm": 1.1596362590789795, "learning_rate": 6.900348491756867e-05, "elbo": 3.45507, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.4984752521698335}
{"step": 93765, "loss": 3.4471240997314454, "grad_norm": 0.670831561088562, "learning_rate": 6.893081548317211e-05, "elbo": 3.44712, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.498768472906404}
{"step": 93770, "loss": 3.4625389099121096, "grad_norm": 0.7102130651473999, "learning_rate": 6.885809933887023e-05, "elbo": 3.46254, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.499061693642974}
{"step": 93775, "loss": 3.457254409790039, "grad_norm": 0.719734787940979, "learning_rate": 6.878533666408286e-05, "elbo": 3.45725, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.499354914379545}
{"step": 93780, "loss": 3.4484458923339845, "grad_norm": 0.7419063448905945, "learning_rate": 6.871252763834465e-05, "elbo": 3.44845, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.499648135116115}
{"step": 93785, "loss": 3.448019790649414, "grad_norm": 0.6900601983070374, "learning_rate": 6.863967244130467e-05, "elbo": 3.44802, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.499941355852686}
{"step": 93790, "loss": 3.459319305419922, "grad_norm": 0.683536171913147, "learning_rate": 6.856677125272585e-05, "elbo": 3.45932, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.500234576589256}
{"step": 93795, "loss": 3.452584457397461, "grad_norm": 0.6863804459571838, "learning_rate": 6.849382425248464e-05, "elbo": 3.45258, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.500527797325827}
{"step": 93800, "loss": 3.437116241455078, "grad_norm": 0.7206366062164307, "learning_rate": 6.842083162057047e-05, "elbo": 3.43712, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.500821018062397}
{"step": 93805, "loss": 3.4464080810546873, "grad_norm": 0.8281248807907104, "learning_rate": 6.834779353708548e-05, "elbo": 3.44641, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5011142387989675}
{"step": 93810, "loss": 3.448011779785156, "grad_norm": 0.7155276536941528, "learning_rate": 6.82747101822438e-05, "elbo": 3.44801, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.501407459535538}
{"step": 93815, "loss": 3.453124237060547, "grad_norm": 0.7756128907203674, "learning_rate": 6.820158173637142e-05, "elbo": 3.45312, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5017006802721085}
{"step": 93820, "loss": 3.469490814208984, "grad_norm": 0.8233893513679504, "learning_rate": 6.812840837990542e-05, "elbo": 3.46949, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.50199390100868}
{"step": 93825, "loss": 3.4635143280029297, "grad_norm": 0.7477840185165405, "learning_rate": 6.805519029339387e-05, "elbo": 3.46351, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.502287121745249}
{"step": 93830, "loss": 3.458607864379883, "grad_norm": 0.7245284914970398, "learning_rate": 6.798192765749508e-05, "elbo": 3.45861, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.502580342481821}
{"step": 93835, "loss": 3.4589996337890625, "grad_norm": 0.7067325711250305, "learning_rate": 6.790862065297732e-05, "elbo": 3.459, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.502873563218391}
{"step": 93840, "loss": 3.4629585266113283, "grad_norm": 0.734388530254364, "learning_rate": 6.783526946071835e-05, "elbo": 3.46296, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.503166783954962}
{"step": 93845, "loss": 3.4557373046875, "grad_norm": 0.7141395807266235, "learning_rate": 6.776187426170493e-05, "elbo": 3.45574, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.503460004691532}
{"step": 93850, "loss": 3.443924331665039, "grad_norm": 0.6973578333854675, "learning_rate": 6.768843523703244e-05, "elbo": 3.44392, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5037532254281025}
{"step": 93855, "loss": 3.448673629760742, "grad_norm": 0.6797424554824829, "learning_rate": 6.761495256790435e-05, "elbo": 3.44867, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.504046446164673}
{"step": 93860, "loss": 3.4748035430908204, "grad_norm": 0.9151451587677002, "learning_rate": 6.754142643563185e-05, "elbo": 3.4748, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.504339666901243}
{"step": 93865, "loss": 3.4578319549560548, "grad_norm": 0.728659451007843, "learning_rate": 6.746785702163336e-05, "elbo": 3.45783, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.504632887637814}
{"step": 93870, "loss": 3.449795150756836, "grad_norm": 0.7944301962852478, "learning_rate": 6.739424450743409e-05, "elbo": 3.4498, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.504926108374384}
{"step": 93875, "loss": 3.4568336486816404, "grad_norm": 0.7304652333259583, "learning_rate": 6.73205890746656e-05, "elbo": 3.45683, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.505219329110955}
{"step": 93880, "loss": 3.4616168975830077, "grad_norm": 1.0822117328643799, "learning_rate": 6.724689090506539e-05, "elbo": 3.46162, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.505512549847525}
{"step": 93885, "loss": 3.4528213500976563, "grad_norm": 0.729826807975769, "learning_rate": 6.71731501804763e-05, "elbo": 3.45282, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.505805770584096}
{"step": 93890, "loss": 3.462061309814453, "grad_norm": 0.7434620261192322, "learning_rate": 6.70993670828463e-05, "elbo": 3.46206, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.506098991320666}
{"step": 93895, "loss": 3.453870391845703, "grad_norm": 0.7258502840995789, "learning_rate": 6.702554179422783e-05, "elbo": 3.45387, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.506392212057237}
{"step": 93900, "loss": 3.456378173828125, "grad_norm": 0.8195850849151611, "learning_rate": 6.69516744967774e-05, "elbo": 3.45638, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.506685432793807}
{"step": 93905, "loss": 3.438477325439453, "grad_norm": 0.8871377110481262, "learning_rate": 6.68777653727553e-05, "elbo": 3.43848, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5069786535303775}
{"step": 93910, "loss": 3.446124267578125, "grad_norm": 0.7133239507675171, "learning_rate": 6.680381460452488e-05, "elbo": 3.44612, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.507271874266948}
{"step": 93915, "loss": 3.455440902709961, "grad_norm": 0.7006372809410095, "learning_rate": 6.672982237455237e-05, "elbo": 3.45544, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.507565095003518}
{"step": 93920, "loss": 3.444562530517578, "grad_norm": 1.0171221494674683, "learning_rate": 6.66557888654062e-05, "elbo": 3.44456, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.507858315740089}
{"step": 93925, "loss": 3.442279052734375, "grad_norm": 1.0243350267410278, "learning_rate": 6.658171425975672e-05, "elbo": 3.44228, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.508151536476659}
{"step": 93930, "loss": 3.461595916748047, "grad_norm": 0.7471895813941956, "learning_rate": 6.650759874037562e-05, "elbo": 3.4616, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.50844475721323}
{"step": 93935, "loss": 3.456865692138672, "grad_norm": 0.7442511320114136, "learning_rate": 6.643344249013562e-05, "elbo": 3.45687, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5087379779498}
{"step": 93940, "loss": 3.458060455322266, "grad_norm": 0.6806083917617798, "learning_rate": 6.635924569200989e-05, "elbo": 3.45806, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.509031198686371}
{"step": 93945, "loss": 3.4451980590820312, "grad_norm": 0.693347156047821, "learning_rate": 6.628500852907161e-05, "elbo": 3.4452, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.509324419422942}
{"step": 93950, "loss": 3.463347625732422, "grad_norm": 1.0739253759384155, "learning_rate": 6.621073118449364e-05, "elbo": 3.46335, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.509617640159512}
{"step": 93955, "loss": 3.4401756286621095, "grad_norm": 0.9634232521057129, "learning_rate": 6.613641384154794e-05, "elbo": 3.44018, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.509910860896083}
{"step": 93960, "loss": 3.4348621368408203, "grad_norm": 0.7605531215667725, "learning_rate": 6.606205668360516e-05, "elbo": 3.43486, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.510204081632653}
{"step": 93965, "loss": 3.4506561279296877, "grad_norm": 1.922269344329834, "learning_rate": 6.598765989413419e-05, "elbo": 3.45066, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.510497302369224}
{"step": 93970, "loss": 3.454302215576172, "grad_norm": 1.3439247608184814, "learning_rate": 6.591322365670171e-05, "elbo": 3.4543, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.510790523105794}
{"step": 93975, "loss": 3.447449493408203, "grad_norm": 0.7664238810539246, "learning_rate": 6.583874815497175e-05, "elbo": 3.44745, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.511083743842365}
{"step": 93980, "loss": 3.4464378356933594, "grad_norm": 0.6981356143951416, "learning_rate": 6.57642335727052e-05, "elbo": 3.44644, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.511376964578935}
{"step": 93985, "loss": 3.456193542480469, "grad_norm": 0.7518587708473206, "learning_rate": 6.568968009375937e-05, "elbo": 3.45619, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.511670185315506}
{"step": 93990, "loss": 3.444963836669922, "grad_norm": 0.7024879455566406, "learning_rate": 6.56150879020876e-05, "elbo": 3.44496, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.511963406052076}
{"step": 93990, "loss": 3.444963836669922, "grad_norm": 0.7024879455566406, "learning_rate": 6.56150879020876e-05, "elbo": 3.44496, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.511963406052076}
{"step": 93995, "loss": 3.4386520385742188, "grad_norm": 0.7153291702270508, "learning_rate": 6.554045718173867e-05, "elbo": 3.43865, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.512256626788647}
{"step": 94000, "loss": 3.4509307861328127, "grad_norm": 0.7312988042831421, "learning_rate": 6.546578811685648e-05, "elbo": 3.45093, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.512549847525217}
{"step": 94005, "loss": 3.4493026733398438, "grad_norm": 0.6655982136726379, "learning_rate": 6.539108089167953e-05, "elbo": 3.4493, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5128430682617875}
{"step": 94010, "loss": 3.4588287353515623, "grad_norm": 0.6855341196060181, "learning_rate": 6.531633569054044e-05, "elbo": 3.45883, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.513136288998358}
{"step": 94015, "loss": 3.455770492553711, "grad_norm": 0.6950114965438843, "learning_rate": 6.52415526978656e-05, "elbo": 3.45577, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.513429509734928}
{"step": 94020, "loss": 3.4424224853515626, "grad_norm": 0.750451385974884, "learning_rate": 6.516673209817458e-05, "elbo": 3.44242, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.513722730471499}
{"step": 94025, "loss": 3.451655960083008, "grad_norm": 0.7027801275253296, "learning_rate": 6.50918740760798e-05, "elbo": 3.45166, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.514015951208069}
{"step": 94030, "loss": 3.4620941162109373, "grad_norm": 0.8617647290229797, "learning_rate": 6.501697881628597e-05, "elbo": 3.46209, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.51430917194464}
{"step": 94035, "loss": 3.458203887939453, "grad_norm": 0.6946031451225281, "learning_rate": 6.494204650358973e-05, "elbo": 3.4582, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.51460239268121}
{"step": 94040, "loss": 3.4575172424316407, "grad_norm": 0.7317050695419312, "learning_rate": 6.486707732287906e-05, "elbo": 3.45752, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.514895613417781}
{"step": 94045, "loss": 3.443299102783203, "grad_norm": 0.680270254611969, "learning_rate": 6.4792071459133e-05, "elbo": 3.4433, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.515188834154351}
{"step": 94050, "loss": 3.4544052124023437, "grad_norm": 1.0976297855377197, "learning_rate": 6.471702909742104e-05, "elbo": 3.45441, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.515482054890922}
{"step": 94055, "loss": 3.4525726318359373, "grad_norm": 0.6884170770645142, "learning_rate": 6.464195042290277e-05, "elbo": 3.45257, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.515775275627492}
{"step": 94060, "loss": 3.449066925048828, "grad_norm": 0.7364899516105652, "learning_rate": 6.456683562082734e-05, "elbo": 3.44907, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5160684963640625}
{"step": 94065, "loss": 3.4380783081054687, "grad_norm": 0.6858014464378357, "learning_rate": 6.449168487653305e-05, "elbo": 3.43808, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.516361717100633}
{"step": 94070, "loss": 3.443416213989258, "grad_norm": 0.697626531124115, "learning_rate": 6.441649837544694e-05, "elbo": 3.44342, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.516654937837204}
{"step": 94075, "loss": 3.441112518310547, "grad_norm": 0.6857654452323914, "learning_rate": 6.434127630308416e-05, "elbo": 3.44111, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.516948158573774}
{"step": 94080, "loss": 3.4491806030273438, "grad_norm": 0.7243415713310242, "learning_rate": 6.426601884504773e-05, "elbo": 3.44918, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.517241379310345}
{"step": 94085, "loss": 3.468183898925781, "grad_norm": 0.7134314775466919, "learning_rate": 6.419072618702794e-05, "elbo": 3.46818, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.517534600046916}
{"step": 94090, "loss": 3.4416393280029296, "grad_norm": 0.740439772605896, "learning_rate": 6.411539851480195e-05, "elbo": 3.44164, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.517827820783486}
{"step": 94095, "loss": 3.4596466064453124, "grad_norm": 0.7016701102256775, "learning_rate": 6.404003601423329e-05, "elbo": 3.45965, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.518121041520057}
{"step": 94100, "loss": 3.4563743591308596, "grad_norm": 0.7066850662231445, "learning_rate": 6.396463887127144e-05, "elbo": 3.45637, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.518414262256627}
{"step": 94105, "loss": 3.4540611267089845, "grad_norm": 0.7128087878227234, "learning_rate": 6.388920727195138e-05, "elbo": 3.45406, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5187074829931975}
{"step": 94110, "loss": 3.456195068359375, "grad_norm": 0.7282987236976624, "learning_rate": 6.381374140239303e-05, "elbo": 3.45619, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.519000703729768}
{"step": 94115, "loss": 3.445844268798828, "grad_norm": 0.7129228711128235, "learning_rate": 6.373824144880098e-05, "elbo": 3.44584, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.519293924466338}
{"step": 94120, "loss": 3.4374343872070314, "grad_norm": 0.9191185235977173, "learning_rate": 6.366270759746382e-05, "elbo": 3.43743, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.519587145202909}
{"step": 94125, "loss": 3.4543975830078124, "grad_norm": 0.6664459109306335, "learning_rate": 6.358714003475384e-05, "elbo": 3.4544, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.519880365939479}
{"step": 94130, "loss": 3.4594375610351564, "grad_norm": 1.024687647819519, "learning_rate": 6.351153894712647e-05, "elbo": 3.45944, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.52017358667605}
{"step": 94135, "loss": 3.449174499511719, "grad_norm": 0.6959266066551208, "learning_rate": 6.34359045211199e-05, "elbo": 3.44917, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.52046680741262}
{"step": 94140, "loss": 3.4670642852783202, "grad_norm": 0.7004629373550415, "learning_rate": 6.336023694335456e-05, "elbo": 3.46706, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.520760028149191}
{"step": 94145, "loss": 3.457141876220703, "grad_norm": 0.6960279941558838, "learning_rate": 6.328453640053263e-05, "elbo": 3.45714, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.521053248885761}
{"step": 94150, "loss": 3.446053314208984, "grad_norm": 0.7080476880073547, "learning_rate": 6.320880307943774e-05, "elbo": 3.44605, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5213464696223316}
{"step": 94155, "loss": 3.4422340393066406, "grad_norm": 0.7498438954353333, "learning_rate": 6.313303716693428e-05, "elbo": 3.44223, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.521639690358902}
{"step": 94160, "loss": 3.447400665283203, "grad_norm": 0.6936725378036499, "learning_rate": 6.305723884996715e-05, "elbo": 3.4474, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5219329110954725}
{"step": 94165, "loss": 3.4350044250488283, "grad_norm": 0.689621090888977, "learning_rate": 6.298140831556112e-05, "elbo": 3.435, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.522226131832043}
{"step": 94170, "loss": 3.4552406311035155, "grad_norm": 0.7572875618934631, "learning_rate": 6.290554575082052e-05, "elbo": 3.45524, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.522519352568613}
{"step": 94175, "loss": 3.45980224609375, "grad_norm": 0.7908656597137451, "learning_rate": 6.282965134292868e-05, "elbo": 3.4598, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.522812573305184}
{"step": 94180, "loss": 3.454188919067383, "grad_norm": 0.8019132018089294, "learning_rate": 6.275372527914752e-05, "elbo": 3.45419, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.523105794041754}
{"step": 94185, "loss": 3.4512779235839846, "grad_norm": 0.847089946269989, "learning_rate": 6.267776774681703e-05, "elbo": 3.45128, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.523399014778326}
{"step": 94185, "loss": 3.4512779235839846, "grad_norm": 0.847089946269989, "learning_rate": 6.267776774681703e-05, "elbo": 3.45128, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.523399014778326}
{"step": 94190, "loss": 3.4475528717041017, "grad_norm": 0.7846642732620239, "learning_rate": 6.260177893335489e-05, "elbo": 3.44755, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.523692235514895}
{"step": 94195, "loss": 3.441400909423828, "grad_norm": 0.7660394310951233, "learning_rate": 6.252575902625595e-05, "elbo": 3.4414, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5239854562514665}
{"step": 94200, "loss": 3.4474327087402346, "grad_norm": 0.8264415264129639, "learning_rate": 6.244970821309174e-05, "elbo": 3.44743, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.524278676988036}
{"step": 94205, "loss": 3.4558692932128907, "grad_norm": 0.6671199798583984, "learning_rate": 6.237362668151012e-05, "elbo": 3.45587, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.524571897724607}
{"step": 94210, "loss": 3.4456390380859374, "grad_norm": 0.7385469079017639, "learning_rate": 6.229751461923468e-05, "elbo": 3.44564, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.524865118461178}
{"step": 94215, "loss": 3.452797698974609, "grad_norm": 0.7226120233535767, "learning_rate": 6.222137221406439e-05, "elbo": 3.4528, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.525158339197748}
{"step": 94220, "loss": 3.455790710449219, "grad_norm": 0.6864637136459351, "learning_rate": 6.214519965387303e-05, "elbo": 3.45579, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.525451559934319}
{"step": 94225, "loss": 3.4762039184570312, "grad_norm": 0.7207406163215637, "learning_rate": 6.206899712660886e-05, "elbo": 3.4762, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.525744780670889}
{"step": 94230, "loss": 3.437998580932617, "grad_norm": 0.6852063536643982, "learning_rate": 6.199276482029401e-05, "elbo": 3.438, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.52603800140746}
{"step": 94235, "loss": 3.449310302734375, "grad_norm": 0.697519063949585, "learning_rate": 6.191650292302414e-05, "elbo": 3.44931, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.52633122214403}
{"step": 94240, "loss": 3.4471702575683594, "grad_norm": 0.7974977493286133, "learning_rate": 6.18402116229679e-05, "elbo": 3.44717, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.526624442880601}
{"step": 94240, "loss": 3.4471702575683594, "grad_norm": 0.7974977493286133, "learning_rate": 6.18402116229679e-05, "elbo": 3.44717, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.526624442880601}
{"step": 94245, "loss": 3.4605804443359376, "grad_norm": 0.7928847670555115, "learning_rate": 6.176389110836647e-05, "elbo": 3.46058, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.526917663617171}
{"step": 94250, "loss": 3.4575469970703123, "grad_norm": 0.6800076365470886, "learning_rate": 6.168754156753316e-05, "elbo": 3.45755, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5272108843537415}
{"step": 94255, "loss": 3.451985168457031, "grad_norm": 0.9107043147087097, "learning_rate": 6.161116318885283e-05, "elbo": 3.45199, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.527504105090312}
{"step": 94260, "loss": 3.445115661621094, "grad_norm": 0.7211392521858215, "learning_rate": 6.153475616078158e-05, "elbo": 3.44512, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.527797325826882}
{"step": 94265, "loss": 3.452532196044922, "grad_norm": 0.6887140870094299, "learning_rate": 6.145832067184614e-05, "elbo": 3.45253, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.528090546563453}
{"step": 94270, "loss": 3.465113067626953, "grad_norm": 0.6986390948295593, "learning_rate": 6.138185691064348e-05, "elbo": 3.46511, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.528383767300023}
{"step": 94275, "loss": 3.465203857421875, "grad_norm": 0.7045629620552063, "learning_rate": 6.130536506584032e-05, "elbo": 3.4652, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.528676988036594}
{"step": 94280, "loss": 3.453777313232422, "grad_norm": 0.6986510753631592, "learning_rate": 6.12288453261727e-05, "elbo": 3.45378, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.528970208773164}
{"step": 94285, "loss": 3.4449462890625, "grad_norm": 0.7018997669219971, "learning_rate": 6.115229788044548e-05, "elbo": 3.44495, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.529263429509735}
{"step": 94290, "loss": 3.4307342529296876, "grad_norm": 0.7042824625968933, "learning_rate": 6.107572291753183e-05, "elbo": 3.43073, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.529556650246305}
{"step": 94295, "loss": 3.4467800140380858, "grad_norm": 0.7123608589172363, "learning_rate": 6.0999120626372894e-05, "elbo": 3.44678, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.529849870982876}
{"step": 94300, "loss": 3.4538600921630858, "grad_norm": 0.6690999865531921, "learning_rate": 6.092249119597721e-05, "elbo": 3.45386, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.530143091719446}
{"step": 94305, "loss": 3.4424728393554687, "grad_norm": 0.6834113597869873, "learning_rate": 6.0845834815420274e-05, "elbo": 3.44247, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5304363124560165}
{"step": 94310, "loss": 3.447233963012695, "grad_norm": 0.7105745673179626, "learning_rate": 6.076915167384407e-05, "elbo": 3.44723, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.530729533192588}
{"step": 94315, "loss": 3.4523738861083983, "grad_norm": 0.7470470666885376, "learning_rate": 6.0692441960456657e-05, "elbo": 3.45237, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.531022753929157}
{"step": 94320, "loss": 3.4474342346191404, "grad_norm": 0.6876434087753296, "learning_rate": 6.0615705864531594e-05, "elbo": 3.44743, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.531315974665729}
{"step": 94325, "loss": 3.45221061706543, "grad_norm": 0.6981390118598938, "learning_rate": 6.05389435754076e-05, "elbo": 3.45221, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.531609195402299}
{"step": 94330, "loss": 3.454698181152344, "grad_norm": 0.6956357359886169, "learning_rate": 6.046215528248798e-05, "elbo": 3.4547, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.53190241613887}
{"step": 94335, "loss": 3.4544746398925783, "grad_norm": 0.6905387043952942, "learning_rate": 6.0385341175240205e-05, "elbo": 3.45447, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.53219563687544}
{"step": 94340, "loss": 3.4420536041259764, "grad_norm": 0.7339234948158264, "learning_rate": 6.0308501443195464e-05, "elbo": 3.44205, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.532488857612011}
{"step": 94345, "loss": 3.4346435546875, "grad_norm": 0.8205516934394836, "learning_rate": 6.0231636275948135e-05, "elbo": 3.43464, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.532782078348581}
{"step": 94350, "loss": 3.467459487915039, "grad_norm": 0.7242521047592163, "learning_rate": 6.0154745863155405e-05, "elbo": 3.46746, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5330752990851515}
{"step": 94355, "loss": 3.432766342163086, "grad_norm": 0.7407072186470032, "learning_rate": 6.00778303945367e-05, "elbo": 3.43277, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.533368519821722}
{"step": 94360, "loss": 3.4385139465332033, "grad_norm": 0.6813199520111084, "learning_rate": 6.000089005987331e-05, "elbo": 3.43851, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.533661740558292}
{"step": 94365, "loss": 3.4548973083496093, "grad_norm": 0.8171948194503784, "learning_rate": 5.9923925049007855e-05, "elbo": 3.4549, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.533954961294863}
{"step": 94370, "loss": 3.4334877014160154, "grad_norm": 0.7138549089431763, "learning_rate": 5.984693555184386e-05, "elbo": 3.43349, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.534248182031433}
{"step": 94375, "loss": 3.450611877441406, "grad_norm": 0.7384381890296936, "learning_rate": 5.9769921758345246e-05, "elbo": 3.45061, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.534541402768004}
{"step": 94380, "loss": 3.451861572265625, "grad_norm": 0.8396598100662231, "learning_rate": 5.9692883858535884e-05, "elbo": 3.45186, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.534834623504574}
{"step": 94385, "loss": 3.44128303527832, "grad_norm": 0.7191562652587891, "learning_rate": 5.9615822042499146e-05, "elbo": 3.44128, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.535127844241145}
{"step": 94390, "loss": 3.4420516967773436, "grad_norm": 0.7274619936943054, "learning_rate": 5.953873650037739e-05, "elbo": 3.44205, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.535421064977715}
{"step": 94395, "loss": 3.4512229919433595, "grad_norm": 0.7697978615760803, "learning_rate": 5.946162742237154e-05, "elbo": 3.45122, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.535714285714286}
{"step": 94400, "loss": 3.4568382263183595, "grad_norm": 0.7085822224617004, "learning_rate": 5.938449499874059e-05, "elbo": 3.45684, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.536007506450856}
{"step": 94405, "loss": 3.438965606689453, "grad_norm": 0.6831241250038147, "learning_rate": 5.930733941980111e-05, "elbo": 3.43897, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5363007271874265}
{"step": 94410, "loss": 3.44649658203125, "grad_norm": 0.6866365075111389, "learning_rate": 5.923016087592681e-05, "elbo": 3.4465, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.536593947923997}
{"step": 94415, "loss": 3.445069122314453, "grad_norm": 1.2272601127624512, "learning_rate": 5.9152959557548117e-05, "elbo": 3.44507, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.536887168660567}
{"step": 94420, "loss": 3.4468528747558596, "grad_norm": 0.707972526550293, "learning_rate": 5.9075735655151575e-05, "elbo": 3.44685, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.537180389397138}
{"step": 94425, "loss": 3.444618988037109, "grad_norm": 0.7318782806396484, "learning_rate": 5.89984893592795e-05, "elbo": 3.44462, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.537473610133708}
{"step": 94430, "loss": 3.4377361297607423, "grad_norm": 0.7394681572914124, "learning_rate": 5.892122086052944e-05, "elbo": 3.43774, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.537766830870279}
{"step": 94435, "loss": 3.4368247985839844, "grad_norm": 0.6962073445320129, "learning_rate": 5.8843930349553736e-05, "elbo": 3.43682, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.53806005160685}
{"step": 94440, "loss": 3.444535827636719, "grad_norm": 0.7709401249885559, "learning_rate": 5.876661801705905e-05, "elbo": 3.44454, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.53835327234342}
{"step": 94445, "loss": 3.430459213256836, "grad_norm": 0.7008699178695679, "learning_rate": 5.868928405380585e-05, "elbo": 3.43046, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.538646493079991}
{"step": 94450, "loss": 3.435797119140625, "grad_norm": 0.7221989631652832, "learning_rate": 5.861192865060805e-05, "elbo": 3.4358, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5389397138165615}
{"step": 94455, "loss": 3.4629592895507812, "grad_norm": 0.7247304320335388, "learning_rate": 5.853455199833238e-05, "elbo": 3.46296, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.539232934553132}
{"step": 94460, "loss": 3.446379852294922, "grad_norm": 0.7582495808601379, "learning_rate": 5.845715428789805e-05, "elbo": 3.44638, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.539526155289702}
{"step": 94465, "loss": 3.4326438903808594, "grad_norm": 0.7259649634361267, "learning_rate": 5.837973571027621e-05, "elbo": 3.43264, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.539819376026273}
{"step": 94470, "loss": 3.451895904541016, "grad_norm": 0.6812506318092346, "learning_rate": 5.830229645648952e-05, "elbo": 3.4519, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.540112596762843}
{"step": 94475, "loss": 3.447452163696289, "grad_norm": 0.6936132311820984, "learning_rate": 5.822483671761163e-05, "elbo": 3.44745, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.540405817499414}
{"step": 94480, "loss": 3.447104644775391, "grad_norm": 0.7967360019683838, "learning_rate": 5.814735668476674e-05, "elbo": 3.4471, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.540699038235984}
{"step": 94485, "loss": 3.4365554809570313, "grad_norm": 0.7045105695724487, "learning_rate": 5.806985654912914e-05, "elbo": 3.43656, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.540992258972555}
{"step": 94490, "loss": 3.457223129272461, "grad_norm": 0.7261384129524231, "learning_rate": 5.79923365019227e-05, "elbo": 3.45722, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.541285479709125}
{"step": 94495, "loss": 3.4365089416503904, "grad_norm": 0.8081353306770325, "learning_rate": 5.791479673442044e-05, "elbo": 3.43651, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.541578700445696}
{"step": 94500, "loss": 3.4456859588623048, "grad_norm": 0.7405845522880554, "learning_rate": 5.783723743794402e-05, "elbo": 3.44569, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.541871921182266}
{"step": 94505, "loss": 3.4489097595214844, "grad_norm": 0.8371793627738953, "learning_rate": 5.7759658803863304e-05, "elbo": 3.44891, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5421651419188365}
{"step": 94510, "loss": 3.4280967712402344, "grad_norm": 0.7124797105789185, "learning_rate": 5.7682061023595845e-05, "elbo": 3.4281, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.542458362655407}
{"step": 94515, "loss": 3.4517711639404296, "grad_norm": 0.7561008930206299, "learning_rate": 5.7604444288606474e-05, "elbo": 3.45177, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.542751583391977}
{"step": 94520, "loss": 3.4423038482666017, "grad_norm": 0.6795347929000854, "learning_rate": 5.752680879040676e-05, "elbo": 3.4423, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.543044804128548}
{"step": 94525, "loss": 3.429853057861328, "grad_norm": 0.6897479891777039, "learning_rate": 5.744915472055456e-05, "elbo": 3.42985, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.543338024865118}
{"step": 94530, "loss": 3.4425209045410154, "grad_norm": 0.7003504633903503, "learning_rate": 5.73714822706536e-05, "elbo": 3.44252, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.543631245601689}
{"step": 94535, "loss": 3.4152984619140625, "grad_norm": 0.6981723308563232, "learning_rate": 5.72937916323529e-05, "elbo": 3.4153, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.543924466338259}
{"step": 94540, "loss": 3.4557018280029297, "grad_norm": 0.688094437122345, "learning_rate": 5.721608299734641e-05, "elbo": 3.4557, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.54421768707483}
{"step": 94545, "loss": 3.4392158508300783, "grad_norm": 0.7088078260421753, "learning_rate": 5.713835655737244e-05, "elbo": 3.43922, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5445109078114}
{"step": 94550, "loss": 3.4470882415771484, "grad_norm": 0.7112588882446289, "learning_rate": 5.706061250421328e-05, "elbo": 3.44709, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.544804128547971}
{"step": 94555, "loss": 3.455036926269531, "grad_norm": 0.6753712892532349, "learning_rate": 5.6982851029694636e-05, "elbo": 3.45504, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.545097349284541}
{"step": 94560, "loss": 3.463213348388672, "grad_norm": 0.7000607252120972, "learning_rate": 5.690507232568524e-05, "elbo": 3.46321, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.545390570021112}
{"step": 94565, "loss": 3.441438674926758, "grad_norm": 0.7613134384155273, "learning_rate": 5.682727658409628e-05, "elbo": 3.44144, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.545683790757682}
{"step": 94570, "loss": 3.4336921691894533, "grad_norm": 0.6719661951065063, "learning_rate": 5.6749463996881026e-05, "elbo": 3.43369, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.545977011494253}
{"step": 94575, "loss": 3.432120513916016, "grad_norm": 0.7335240840911865, "learning_rate": 5.6671634756034295e-05, "elbo": 3.43212, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.546270232230824}
{"step": 94580, "loss": 3.4345062255859373, "grad_norm": 0.6914181113243103, "learning_rate": 5.659378905359203e-05, "elbo": 3.43451, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.546563452967394}
{"step": 94580, "loss": 3.4345062255859373, "grad_norm": 0.6914181113243103, "learning_rate": 5.659378905359203e-05, "elbo": 3.43451, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.546563452967394}
{"step": 94585, "loss": 3.4437286376953127, "grad_norm": 0.6954068541526794, "learning_rate": 5.651592708163074e-05, "elbo": 3.44373, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.546856673703965}
{"step": 94590, "loss": 3.437281036376953, "grad_norm": 0.7419368028640747, "learning_rate": 5.6438049032267095e-05, "elbo": 3.43728, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.547149894440535}
{"step": 94595, "loss": 3.4396999359130858, "grad_norm": 0.6887128353118896, "learning_rate": 5.6360155097657464e-05, "elbo": 3.4397, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5474431151771055}
{"step": 94600, "loss": 3.4482677459716795, "grad_norm": 0.6693927049636841, "learning_rate": 5.628224546999735e-05, "elbo": 3.44827, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.547736335913676}
{"step": 94605, "loss": 3.45718994140625, "grad_norm": 0.8293523192405701, "learning_rate": 5.620432034152106e-05, "elbo": 3.45719, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.548029556650246}
{"step": 94610, "loss": 3.4410541534423826, "grad_norm": 0.7055026292800903, "learning_rate": 5.612637990450107e-05, "elbo": 3.44105, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.548322777386817}
{"step": 94615, "loss": 3.4459739685058595, "grad_norm": 0.6679953336715698, "learning_rate": 5.6048424351247686e-05, "elbo": 3.44597, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.548615998123387}
{"step": 94620, "loss": 3.4409194946289063, "grad_norm": 0.7044491767883301, "learning_rate": 5.597045387410845e-05, "elbo": 3.44092, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.548909218859958}
{"step": 94625, "loss": 3.441693115234375, "grad_norm": 0.6985624432563782, "learning_rate": 5.58924686654678e-05, "elbo": 3.44169, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.549202439596528}
{"step": 94630, "loss": 3.4575450897216795, "grad_norm": 0.7178107500076294, "learning_rate": 5.5814468917746475e-05, "elbo": 3.45754, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.549495660333099}
{"step": 94635, "loss": 3.4373607635498047, "grad_norm": 0.785825788974762, "learning_rate": 5.573645482340107e-05, "elbo": 3.43736, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.549788881069669}
{"step": 94640, "loss": 3.4384521484375, "grad_norm": 0.7123294472694397, "learning_rate": 5.565842657492366e-05, "elbo": 3.43845, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.55008210180624}
{"step": 94645, "loss": 3.4483596801757814, "grad_norm": 0.7354428172111511, "learning_rate": 5.5580384364841157e-05, "elbo": 3.44836, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.55037532254281}
{"step": 94650, "loss": 3.437510681152344, "grad_norm": 0.7220710515975952, "learning_rate": 5.5502328385714954e-05, "elbo": 3.43751, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5506685432793805}
{"step": 94655, "loss": 3.440836715698242, "grad_norm": 0.7306620478630066, "learning_rate": 5.542425883014043e-05, "elbo": 3.44084, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.550961764015951}
{"step": 94660, "loss": 3.4410350799560545, "grad_norm": 0.7454886436462402, "learning_rate": 5.534617589074647e-05, "elbo": 3.44103, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.551254984752521}
{"step": 94665, "loss": 3.429456329345703, "grad_norm": 0.7663138508796692, "learning_rate": 5.526807976019492e-05, "elbo": 3.42946, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.551548205489092}
{"step": 94670, "loss": 3.450054931640625, "grad_norm": 0.7070273756980896, "learning_rate": 5.5189970631180265e-05, "elbo": 3.45005, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.551841426225662}
{"step": 94675, "loss": 3.4335395812988283, "grad_norm": 0.7726504802703857, "learning_rate": 5.5111848696429e-05, "elbo": 3.43354, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.552134646962233}
{"step": 94680, "loss": 3.4337432861328123, "grad_norm": 0.7070298790931702, "learning_rate": 5.503371414869922e-05, "elbo": 3.43374, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.552427867698803}
{"step": 94685, "loss": 3.4537033081054687, "grad_norm": 0.705927848815918, "learning_rate": 5.4955567180780164e-05, "elbo": 3.4537, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.552721088435375}
{"step": 94690, "loss": 3.445968246459961, "grad_norm": 0.7917954325675964, "learning_rate": 5.4877407985491714e-05, "elbo": 3.44597, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.553014309171944}
{"step": 94695, "loss": 3.4359642028808595, "grad_norm": 0.7756563425064087, "learning_rate": 5.4799236755683914e-05, "elbo": 3.43596, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5533075299085155}
{"step": 94700, "loss": 3.428038787841797, "grad_norm": 0.9845840334892273, "learning_rate": 5.472105368423648e-05, "elbo": 3.42804, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.553600750645086}
{"step": 94705, "loss": 3.4346778869628904, "grad_norm": 0.7249080538749695, "learning_rate": 5.46428589640584e-05, "elbo": 3.43468, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.553893971381656}
{"step": 94710, "loss": 3.4464431762695313, "grad_norm": 0.7135946154594421, "learning_rate": 5.4564652788087353e-05, "elbo": 3.44644, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.554187192118227}
{"step": 94715, "loss": 3.444355773925781, "grad_norm": 0.755653977394104, "learning_rate": 5.448643534928931e-05, "elbo": 3.44436, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.554480412854797}
{"step": 94720, "loss": 3.4378997802734377, "grad_norm": 0.7277869582176208, "learning_rate": 5.440820684065804e-05, "elbo": 3.4379, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.554773633591368}
{"step": 94725, "loss": 3.4582252502441406, "grad_norm": 0.7813137173652649, "learning_rate": 5.432996745521458e-05, "elbo": 3.45823, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.555066854327938}
{"step": 94730, "loss": 3.4376754760742188, "grad_norm": 0.7193117737770081, "learning_rate": 5.425171738600686e-05, "elbo": 3.43768, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.555360075064509}
{"step": 94735, "loss": 3.444680404663086, "grad_norm": 0.7162424921989441, "learning_rate": 5.417345682610914e-05, "elbo": 3.44468, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.555653295801079}
{"step": 94740, "loss": 3.4476093292236327, "grad_norm": 0.6958941221237183, "learning_rate": 5.4095185968621575e-05, "elbo": 3.44761, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.55594651653765}
{"step": 94745, "loss": 3.4305320739746095, "grad_norm": 0.7720393538475037, "learning_rate": 5.4016905006669715e-05, "elbo": 3.43053, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.55623973727422}
{"step": 94750, "loss": 3.4386199951171874, "grad_norm": 0.7238444089889526, "learning_rate": 5.3938614133404075e-05, "elbo": 3.43862, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5565329580107905}
{"step": 94755, "loss": 3.437910461425781, "grad_norm": 1.0776509046554565, "learning_rate": 5.386031354199956e-05, "elbo": 3.43791, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.556826178747361}
{"step": 94760, "loss": 3.4404891967773437, "grad_norm": 0.6946107745170593, "learning_rate": 5.378200342565516e-05, "elbo": 3.44049, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.557119399483931}
{"step": 94765, "loss": 3.432081604003906, "grad_norm": 0.8180059194564819, "learning_rate": 5.370368397759324e-05, "elbo": 3.43208, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.557412620220502}
{"step": 94765, "loss": 3.432081604003906, "grad_norm": 0.8180059194564819, "learning_rate": 5.370368397759324e-05, "elbo": 3.43208, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.557412620220502}
{"step": 94770, "loss": 3.4356948852539064, "grad_norm": 0.7125567197799683, "learning_rate": 5.362535539105928e-05, "elbo": 3.43569, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.557705840957072}
{"step": 94775, "loss": 3.4334529876708983, "grad_norm": 0.6987847089767456, "learning_rate": 5.354701785932129e-05, "elbo": 3.43345, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.557999061693643}
{"step": 94780, "loss": 3.4427871704101562, "grad_norm": 0.7979008555412292, "learning_rate": 5.346867157566931e-05, "elbo": 3.44279, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.558292282430213}
{"step": 94785, "loss": 3.4406700134277344, "grad_norm": 0.7409911155700684, "learning_rate": 5.3390316733415044e-05, "elbo": 3.44067, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.558585503166784}
{"step": 94790, "loss": 3.4308685302734374, "grad_norm": 0.729160726070404, "learning_rate": 5.331195352589126e-05, "elbo": 3.43087, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.558878723903354}
{"step": 94795, "loss": 3.436969757080078, "grad_norm": 0.8228051662445068, "learning_rate": 5.3233582146451375e-05, "elbo": 3.43697, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.559171944639925}
{"step": 94800, "loss": 3.432756042480469, "grad_norm": 0.6989761590957642, "learning_rate": 5.315520278846898e-05, "elbo": 3.43276, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.559465165376495}
{"step": 94800, "loss": 3.432756042480469, "grad_norm": 0.6989761590957642, "learning_rate": 5.315520278846898e-05, "elbo": 3.43276, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.559465165376495}
{"step": 94805, "loss": 3.437007522583008, "grad_norm": 0.7007325887680054, "learning_rate": 5.307681564533736e-05, "elbo": 3.43701, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5597583861130655}
{"step": 94810, "loss": 3.434598922729492, "grad_norm": 0.7001990675926208, "learning_rate": 5.299842091046899e-05, "elbo": 3.4346, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.560051606849637}
{"step": 94810, "loss": 3.434598922729492, "grad_norm": 0.7001990675926208, "learning_rate": 5.299842091046899e-05, "elbo": 3.4346, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.560051606849637}
{"step": 94815, "loss": 3.4406715393066407, "grad_norm": 0.7032233476638794, "learning_rate": 5.29200187772951e-05, "elbo": 3.44067, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.560344827586206}
{"step": 94820, "loss": 3.4317367553710936, "grad_norm": 0.9419336318969727, "learning_rate": 5.284160943926515e-05, "elbo": 3.43174, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.560638048322778}
{"step": 94825, "loss": 3.4570308685302735, "grad_norm": 0.7057064771652222, "learning_rate": 5.276319308984636e-05, "elbo": 3.45703, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.560931269059348}
{"step": 94830, "loss": 3.4309593200683595, "grad_norm": 0.8991979956626892, "learning_rate": 5.2684769922523325e-05, "elbo": 3.43096, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.561224489795919}
{"step": 94835, "loss": 3.438581848144531, "grad_norm": 0.7063981294631958, "learning_rate": 5.260634013079737e-05, "elbo": 3.43858, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.561517710532489}
{"step": 94840, "loss": 3.435597229003906, "grad_norm": 0.6834640502929688, "learning_rate": 5.252790390818625e-05, "elbo": 3.4356, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.56181093126906}
{"step": 94845, "loss": 3.4332542419433594, "grad_norm": 0.7469058036804199, "learning_rate": 5.2449461448223517e-05, "elbo": 3.43325, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.56210415200563}
{"step": 94850, "loss": 3.4426956176757812, "grad_norm": 0.8305386304855347, "learning_rate": 5.2371012944458165e-05, "elbo": 3.4427, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5623973727422005}
{"step": 94855, "loss": 3.448954772949219, "grad_norm": 0.7358543276786804, "learning_rate": 5.229255859045406e-05, "elbo": 3.44896, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.562690593478771}
{"step": 94860, "loss": 3.437227249145508, "grad_norm": 1.194671869277954, "learning_rate": 5.2214098579789526e-05, "elbo": 3.43723, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.562983814215341}
{"step": 94865, "loss": 3.4380043029785154, "grad_norm": 0.7244107723236084, "learning_rate": 5.213563310605686e-05, "elbo": 3.438, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.563277034951912}
{"step": 94870, "loss": 3.439360809326172, "grad_norm": 0.7281297445297241, "learning_rate": 5.2057162362861775e-05, "elbo": 3.43936, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.563570255688482}
{"step": 94875, "loss": 3.4208534240722654, "grad_norm": 0.6888887286186218, "learning_rate": 5.1978686543823076e-05, "elbo": 3.42085, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.563863476425053}
{"step": 94880, "loss": 3.441961669921875, "grad_norm": 0.698057234287262, "learning_rate": 5.190020584257201e-05, "elbo": 3.44196, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.564156697161623}
{"step": 94885, "loss": 3.4325904846191406, "grad_norm": 0.6967681050300598, "learning_rate": 5.1821720452751945e-05, "elbo": 3.43259, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.564449917898194}
{"step": 94890, "loss": 3.4397598266601563, "grad_norm": 0.7410631775856018, "learning_rate": 5.174323056801773e-05, "elbo": 3.43976, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.564743138634764}
{"step": 94895, "loss": 3.446573257446289, "grad_norm": 0.9477196335792542, "learning_rate": 5.166473638203539e-05, "elbo": 3.44657, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.565036359371335}
{"step": 94900, "loss": 3.4407684326171877, "grad_norm": 0.9619206190109253, "learning_rate": 5.158623808848152e-05, "elbo": 3.44077, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.565329580107905}
{"step": 94905, "loss": 3.4188282012939455, "grad_norm": 0.7252053618431091, "learning_rate": 5.150773588104284e-05, "elbo": 3.41883, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5656228008444755}
{"step": 94910, "loss": 3.4345897674560546, "grad_norm": 0.7124371528625488, "learning_rate": 5.142922995341576e-05, "elbo": 3.43459, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.565916021581046}
{"step": 94915, "loss": 3.435115432739258, "grad_norm": 0.7049762606620789, "learning_rate": 5.1350720499305835e-05, "elbo": 3.43512, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.566209242317616}
{"step": 94920, "loss": 3.441943359375, "grad_norm": 0.6951492428779602, "learning_rate": 5.127220771242737e-05, "elbo": 3.44194, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.566502463054187}
{"step": 94925, "loss": 3.4154678344726563, "grad_norm": 0.8795234560966492, "learning_rate": 5.119369178650282e-05, "elbo": 3.41547, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.566795683790758}
{"step": 94930, "loss": 3.441507339477539, "grad_norm": 1.125319480895996, "learning_rate": 5.111517291526245e-05, "elbo": 3.44151, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.567088904527328}
{"step": 94935, "loss": 3.437742233276367, "grad_norm": 0.8229036927223206, "learning_rate": 5.1036651292443774e-05, "elbo": 3.43774, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.567382125263899}
{"step": 94940, "loss": 3.440137481689453, "grad_norm": 0.7076430320739746, "learning_rate": 5.0958127111791086e-05, "elbo": 3.44014, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5676753460004695}
{"step": 94945, "loss": 3.4397354125976562, "grad_norm": 0.7639005184173584, "learning_rate": 5.087960056705499e-05, "elbo": 3.43974, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.56796856673704}
{"step": 94950, "loss": 3.4487037658691406, "grad_norm": 0.7297846674919128, "learning_rate": 5.080107185199193e-05, "elbo": 3.4487, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5682617874736104}
{"step": 94955, "loss": 3.4272796630859377, "grad_norm": 0.6983293890953064, "learning_rate": 5.0722541160363726e-05, "elbo": 3.42728, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.568555008210181}
{"step": 94960, "loss": 3.431986999511719, "grad_norm": 0.6750927567481995, "learning_rate": 5.0644008685937005e-05, "elbo": 3.43199, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.568848228946751}
{"step": 94965, "loss": 3.4320873260498046, "grad_norm": 0.7276594638824463, "learning_rate": 5.05654746224829e-05, "elbo": 3.43209, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.569141449683322}
{"step": 94970, "loss": 3.4397747039794924, "grad_norm": 0.7444413304328918, "learning_rate": 5.048693916377635e-05, "elbo": 3.43977, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.569434670419892}
{"step": 94975, "loss": 3.4314224243164064, "grad_norm": 0.7039170265197754, "learning_rate": 5.0408402503595845e-05, "elbo": 3.43142, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.569727891156463}
{"step": 94980, "loss": 3.4502643585205077, "grad_norm": 0.9945472478866577, "learning_rate": 5.032986483572273e-05, "elbo": 3.45026, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.570021111893033}
{"step": 94985, "loss": 3.4317401885986327, "grad_norm": 0.6558318138122559, "learning_rate": 5.025132635394095e-05, "elbo": 3.43174, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.570314332629604}
{"step": 94985, "loss": 3.4317401885986327, "grad_norm": 0.6558318138122559, "learning_rate": 5.025132635394095e-05, "elbo": 3.43174, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.570314332629604}
{"step": 94990, "loss": 3.4294506072998048, "grad_norm": 0.7299344539642334, "learning_rate": 5.017278725203636e-05, "elbo": 3.42945, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.570607553366174}
{"step": 94995, "loss": 3.4164634704589845, "grad_norm": 0.7120507955551147, "learning_rate": 5.0094247723796403e-05, "elbo": 3.41646, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5709007741027445}
{"step": 95000, "loss": 3.437468338012695, "grad_norm": 0.7323763966560364, "learning_rate": 5.0015707963009575e-05, "elbo": 3.43747, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.571193994839315}
{"step": 95005, "loss": 3.4382118225097655, "grad_norm": 1.204250693321228, "learning_rate": 4.9937168163464897e-05, "elbo": 3.43821, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.571487215575885}
{"step": 95010, "loss": 3.431039047241211, "grad_norm": 0.7761061787605286, "learning_rate": 4.985862851895152e-05, "elbo": 3.43104, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.571780436312456}
{"step": 95015, "loss": 3.4272483825683593, "grad_norm": 0.7144140601158142, "learning_rate": 4.978008922325824e-05, "elbo": 3.42725, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.572073657049026}
{"step": 95020, "loss": 3.444178009033203, "grad_norm": 1.080369234085083, "learning_rate": 4.9701550470172914e-05, "elbo": 3.44418, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.572366877785597}
{"step": 95025, "loss": 3.449047088623047, "grad_norm": 0.7762836813926697, "learning_rate": 4.962301245348216e-05, "elbo": 3.44905, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.572660098522167}
{"step": 95030, "loss": 3.438503646850586, "grad_norm": 0.9110128283500671, "learning_rate": 4.954447536697068e-05, "elbo": 3.4385, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.572953319258738}
{"step": 95035, "loss": 3.4360145568847655, "grad_norm": 0.6739652752876282, "learning_rate": 4.9465939404420966e-05, "elbo": 3.43601, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.573246539995308}
{"step": 95040, "loss": 3.4231216430664064, "grad_norm": 0.6931828260421753, "learning_rate": 4.9387404759612674e-05, "elbo": 3.42312, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.573539760731879}
{"step": 95045, "loss": 3.4508148193359376, "grad_norm": 0.877955973148346, "learning_rate": 4.930887162632224e-05, "elbo": 3.45081, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.573832981468449}
{"step": 95050, "loss": 3.430786895751953, "grad_norm": 0.7002754807472229, "learning_rate": 4.923034019832237e-05, "elbo": 3.43079, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.57412620220502}
{"step": 95055, "loss": 3.4317096710205077, "grad_norm": 0.6908903121948242, "learning_rate": 4.9151810669381556e-05, "elbo": 3.43171, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.57441942294159}
{"step": 95060, "loss": 3.4384132385253907, "grad_norm": 0.7433410286903381, "learning_rate": 4.907328323326359e-05, "elbo": 3.43841, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.574712643678161}
{"step": 95065, "loss": 3.43740234375, "grad_norm": 0.7060986161231995, "learning_rate": 4.899475808372714e-05, "elbo": 3.4374, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.575005864414732}
{"step": 95070, "loss": 3.428705596923828, "grad_norm": 0.678432047367096, "learning_rate": 4.891623541452516e-05, "elbo": 3.42871, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.575299085151302}
{"step": 95075, "loss": 3.445281982421875, "grad_norm": 0.7065569758415222, "learning_rate": 4.883771541940459e-05, "elbo": 3.44528, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.575592305887873}
{"step": 95080, "loss": 3.429595184326172, "grad_norm": 0.7497661709785461, "learning_rate": 4.8759198292105655e-05, "elbo": 3.4296, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.575885526624443}
{"step": 95085, "loss": 3.422953796386719, "grad_norm": 0.6748467087745667, "learning_rate": 4.868068422636162e-05, "elbo": 3.42295, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.576178747361014}
{"step": 95090, "loss": 3.4466621398925783, "grad_norm": 0.7608890533447266, "learning_rate": 4.8602173415898076e-05, "elbo": 3.44666, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.576471968097584}
{"step": 95095, "loss": 3.4561824798583984, "grad_norm": 0.7805129289627075, "learning_rate": 4.852366605443271e-05, "elbo": 3.45618, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5767651888341545}
{"step": 95100, "loss": 3.4426403045654297, "grad_norm": 0.7617461681365967, "learning_rate": 4.8445162335674574e-05, "elbo": 3.44264, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.577058409570725}
{"step": 95105, "loss": 3.4312694549560545, "grad_norm": 0.7439035773277283, "learning_rate": 4.836666245332382e-05, "elbo": 3.43127, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.577351630307295}
{"step": 95105, "loss": 3.4312694549560545, "grad_norm": 0.7439035773277283, "learning_rate": 4.836666245332382e-05, "elbo": 3.43127, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.577351630307295}
{"step": 95110, "loss": 3.4317970275878906, "grad_norm": 0.7050695419311523, "learning_rate": 4.828816660107109e-05, "elbo": 3.4318, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.577644851043866}
{"step": 95115, "loss": 3.442376708984375, "grad_norm": 0.7927328944206238, "learning_rate": 4.8209674972597114e-05, "elbo": 3.44238, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.577938071780436}
{"step": 95115, "loss": 3.442376708984375, "grad_norm": 0.7927328944206238, "learning_rate": 4.8209674972597114e-05, "elbo": 3.44238, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.577938071780436}
{"step": 95120, "loss": 3.430794525146484, "grad_norm": 0.700537919998169, "learning_rate": 4.813118776157216e-05, "elbo": 3.43079, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.578231292517007}
{"step": 95125, "loss": 3.4425567626953124, "grad_norm": 0.7402504682540894, "learning_rate": 4.8052705161655644e-05, "elbo": 3.44256, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.578524513253577}
{"step": 95130, "loss": 3.432078552246094, "grad_norm": 0.7310939431190491, "learning_rate": 4.797422736649555e-05, "elbo": 3.43208, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.578817733990148}
{"step": 95135, "loss": 3.4380435943603516, "grad_norm": 0.9101495742797852, "learning_rate": 4.7895754569728066e-05, "elbo": 3.43804, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.579110954726718}
{"step": 95140, "loss": 3.439340591430664, "grad_norm": 0.7355921268463135, "learning_rate": 4.7817286964976995e-05, "elbo": 3.43934, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.579404175463289}
{"step": 95140, "loss": 3.439340591430664, "grad_norm": 0.7355921268463135, "learning_rate": 4.7817286964976995e-05, "elbo": 3.43934, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.579404175463289}
{"step": 95145, "loss": 3.428416442871094, "grad_norm": 0.8362100720405579, "learning_rate": 4.7738824745853374e-05, "elbo": 3.42842, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.579697396199859}
{"step": 95150, "loss": 3.4355850219726562, "grad_norm": 1.3042705059051514, "learning_rate": 4.7660368105954905e-05, "elbo": 3.43558, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5799906169364295}
{"step": 95155, "loss": 3.4201454162597655, "grad_norm": 0.7127723693847656, "learning_rate": 4.758191723886557e-05, "elbo": 3.42015, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.580283837673}
{"step": 95160, "loss": 3.420306396484375, "grad_norm": 0.6894789338111877, "learning_rate": 4.750347233815506e-05, "elbo": 3.42031, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.58057705840957}
{"step": 95160, "loss": 3.420306396484375, "grad_norm": 0.6894789338111877, "learning_rate": 4.750347233815506e-05, "elbo": 3.42031, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.58057705840957}
{"step": 95165, "loss": 3.422171783447266, "grad_norm": 0.6910560727119446, "learning_rate": 4.742503359737841e-05, "elbo": 3.42217, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.580870279146141}
{"step": 95170, "loss": 3.430035400390625, "grad_norm": 0.7278743982315063, "learning_rate": 4.734660121007534e-05, "elbo": 3.43004, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.581163499882711}
{"step": 95175, "loss": 3.4255035400390623, "grad_norm": 0.7238031625747681, "learning_rate": 4.7268175369770066e-05, "elbo": 3.4255, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.581456720619283}
{"step": 95180, "loss": 3.4238258361816407, "grad_norm": 1.0464900732040405, "learning_rate": 4.718975626997047e-05, "elbo": 3.42383, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.581749941355852}
{"step": 95185, "loss": 3.431963348388672, "grad_norm": 0.7029110789299011, "learning_rate": 4.711134410416794e-05, "elbo": 3.43196, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.582043162092424}
{"step": 95190, "loss": 3.445273590087891, "grad_norm": 1.1563975811004639, "learning_rate": 4.7032939065836674e-05, "elbo": 3.44527, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.582336382828994}
{"step": 95195, "loss": 3.441040802001953, "grad_norm": 0.6911770701408386, "learning_rate": 4.6954541348433316e-05, "elbo": 3.44104, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5826296035655645}
{"step": 95200, "loss": 3.43238525390625, "grad_norm": 0.6709239482879639, "learning_rate": 4.687615114539644e-05, "elbo": 3.43239, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.582922824302135}
{"step": 95205, "loss": 3.423748016357422, "grad_norm": 0.7013427019119263, "learning_rate": 4.679776865014608e-05, "elbo": 3.42375, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.583216045038705}
{"step": 95210, "loss": 3.4292930603027343, "grad_norm": 0.7428232431411743, "learning_rate": 4.6719394056083246e-05, "elbo": 3.42929, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.583509265775276}
{"step": 95215, "loss": 3.4202590942382813, "grad_norm": 0.7116729617118835, "learning_rate": 4.664102755658948e-05, "elbo": 3.42026, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.583802486511846}
{"step": 95215, "loss": 3.4202590942382813, "grad_norm": 0.7116729617118835, "learning_rate": 4.664102755658948e-05, "elbo": 3.42026, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.583802486511846}
{"step": 95220, "loss": 3.443426513671875, "grad_norm": 0.6954582333564758, "learning_rate": 4.65626693450263e-05, "elbo": 3.44343, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.584095707248417}
{"step": 95225, "loss": 3.429180908203125, "grad_norm": 0.787631630897522, "learning_rate": 4.6484319614734815e-05, "elbo": 3.42918, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.584388927984987}
{"step": 95230, "loss": 3.4279808044433593, "grad_norm": 0.7068207859992981, "learning_rate": 4.6405978559035195e-05, "elbo": 3.42798, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.584682148721558}
{"step": 95235, "loss": 3.4402793884277343, "grad_norm": 0.7742127776145935, "learning_rate": 4.632764637122622e-05, "elbo": 3.44028, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.584975369458128}
{"step": 95240, "loss": 3.432719421386719, "grad_norm": 0.665091335773468, "learning_rate": 4.6249323244584756e-05, "elbo": 3.43272, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.585268590194699}
{"step": 95245, "loss": 3.431505584716797, "grad_norm": 0.6761242151260376, "learning_rate": 4.617100937236535e-05, "elbo": 3.43151, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.585561810931269}
{"step": 95250, "loss": 3.422578811645508, "grad_norm": 0.707688570022583, "learning_rate": 4.6092704947799684e-05, "elbo": 3.42258, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5858550316678395}
{"step": 95255, "loss": 3.450389099121094, "grad_norm": 0.7609039545059204, "learning_rate": 4.601441016409616e-05, "elbo": 3.45039, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.58614825240441}
{"step": 95260, "loss": 3.4210662841796875, "grad_norm": 0.6725730299949646, "learning_rate": 4.593612521443933e-05, "elbo": 3.42107, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.58644147314098}
{"step": 95265, "loss": 3.4261810302734377, "grad_norm": 0.9200180172920227, "learning_rate": 4.585785029198959e-05, "elbo": 3.42618, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.586734693877551}
{"step": 95270, "loss": 3.446482849121094, "grad_norm": 0.7174587845802307, "learning_rate": 4.577958558988247e-05, "elbo": 3.44648, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.587027914614121}
{"step": 95275, "loss": 3.4361534118652344, "grad_norm": 0.6904364228248596, "learning_rate": 4.570133130122839e-05, "elbo": 3.43615, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.587321135350692}
{"step": 95280, "loss": 3.407904052734375, "grad_norm": 0.7698663473129272, "learning_rate": 4.5623087619111985e-05, "elbo": 3.4079, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.587614356087262}
{"step": 95285, "loss": 3.435022735595703, "grad_norm": 0.6835573315620422, "learning_rate": 4.55448547365918e-05, "elbo": 3.43502, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.587907576823833}
{"step": 95290, "loss": 3.424869155883789, "grad_norm": 0.7164966464042664, "learning_rate": 4.546663284669967e-05, "elbo": 3.42487, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.588200797560403}
{"step": 95295, "loss": 3.434275817871094, "grad_norm": 0.6954115629196167, "learning_rate": 4.538842214244035e-05, "elbo": 3.43428, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.588494018296974}
{"step": 95300, "loss": 3.4539260864257812, "grad_norm": 0.9162835478782654, "learning_rate": 4.5310222816790956e-05, "elbo": 3.45393, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.588787239033545}
{"step": 95305, "loss": 3.43221435546875, "grad_norm": 0.6965005397796631, "learning_rate": 4.5232035062700576e-05, "elbo": 3.43221, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5890804597701145}
{"step": 95310, "loss": 3.4319671630859374, "grad_norm": 0.7223967909812927, "learning_rate": 4.5153859073089696e-05, "elbo": 3.43197, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.589373680506686}
{"step": 95315, "loss": 3.425852966308594, "grad_norm": 0.6951015591621399, "learning_rate": 4.507569504084983e-05, "elbo": 3.42585, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.589666901243256}
{"step": 95320, "loss": 3.425648498535156, "grad_norm": 0.937436580657959, "learning_rate": 4.499754315884293e-05, "elbo": 3.42565, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.589960121979827}
{"step": 95325, "loss": 3.3915435791015627, "grad_norm": 0.7165056467056274, "learning_rate": 4.491940361990101e-05, "elbo": 3.39154, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.590253342716397}
{"step": 95330, "loss": 3.4307022094726562, "grad_norm": 0.7049582600593567, "learning_rate": 4.484127661682561e-05, "elbo": 3.4307, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.590546563452968}
{"step": 95335, "loss": 3.4292469024658203, "grad_norm": 0.6974320411682129, "learning_rate": 4.476316234238734e-05, "elbo": 3.42925, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.590839784189538}
{"step": 95340, "loss": 3.425327682495117, "grad_norm": 0.6945776343345642, "learning_rate": 4.468506098932541e-05, "elbo": 3.42533, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5911330049261085}
{"step": 95345, "loss": 3.426230621337891, "grad_norm": 0.6806023120880127, "learning_rate": 4.460697275034717e-05, "elbo": 3.42623, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.591426225662679}
{"step": 95350, "loss": 3.4342624664306642, "grad_norm": 0.7254838943481445, "learning_rate": 4.452889781812753e-05, "elbo": 3.43426, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5917194463992494}
{"step": 95355, "loss": 3.4285160064697267, "grad_norm": 0.6902813911437988, "learning_rate": 4.445083638530869e-05, "elbo": 3.42852, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.59201266713582}
{"step": 95360, "loss": 3.416035461425781, "grad_norm": 0.6895595192909241, "learning_rate": 4.437278864449941e-05, "elbo": 3.41604, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.59230588787239}
{"step": 95365, "loss": 3.4295722961425783, "grad_norm": 0.702404797077179, "learning_rate": 4.4294754788274796e-05, "elbo": 3.42957, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.592599108608961}
{"step": 95370, "loss": 3.431573486328125, "grad_norm": 0.7075960040092468, "learning_rate": 4.421673500917558e-05, "elbo": 3.43157, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.592892329345531}
{"step": 95375, "loss": 3.4229469299316406, "grad_norm": 0.7501131892204285, "learning_rate": 4.4138729499707844e-05, "elbo": 3.42295, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.593185550082102}
{"step": 95380, "loss": 3.4400840759277345, "grad_norm": 0.710494339466095, "learning_rate": 4.40607384523424e-05, "elbo": 3.44008, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.593478770818672}
{"step": 95385, "loss": 3.4191207885742188, "grad_norm": 0.6642864346504211, "learning_rate": 4.398276205951443e-05, "elbo": 3.41912, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.593771991555243}
{"step": 95390, "loss": 3.4358188629150392, "grad_norm": 0.7348679304122925, "learning_rate": 4.390480051362292e-05, "elbo": 3.43582, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.594065212291813}
{"step": 95395, "loss": 3.4228416442871095, "grad_norm": 0.8885932564735413, "learning_rate": 4.382685400703024e-05, "elbo": 3.42284, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5943584330283835}
{"step": 95400, "loss": 3.4499233245849608, "grad_norm": 0.6784219741821289, "learning_rate": 4.374892273206164e-05, "elbo": 3.44992, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.594651653764954}
{"step": 95405, "loss": 3.4329429626464845, "grad_norm": 1.4651730060577393, "learning_rate": 4.36710068810048e-05, "elbo": 3.43294, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5949448745015244}
{"step": 95410, "loss": 3.4106788635253906, "grad_norm": 0.7133870720863342, "learning_rate": 4.359310664610933e-05, "elbo": 3.41068, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.595238095238095}
{"step": 95415, "loss": 3.443510818481445, "grad_norm": 0.6982282996177673, "learning_rate": 4.3515222219586325e-05, "elbo": 3.44351, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.595531315974665}
{"step": 95420, "loss": 3.4317977905273436, "grad_norm": 0.6961074471473694, "learning_rate": 4.343735379360785e-05, "elbo": 3.4318, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.595824536711236}
{"step": 95425, "loss": 3.425978088378906, "grad_norm": 0.6858260035514832, "learning_rate": 4.3359501560306535e-05, "elbo": 3.42598, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.596117757447807}
{"step": 95430, "loss": 3.419574737548828, "grad_norm": 0.7469817399978638, "learning_rate": 4.3281665711774986e-05, "elbo": 3.41957, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.596410978184377}
{"step": 95435, "loss": 3.4270675659179686, "grad_norm": 0.718562662601471, "learning_rate": 4.320384644006546e-05, "elbo": 3.42707, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.596704198920948}
{"step": 95440, "loss": 3.4517181396484373, "grad_norm": 0.6842883229255676, "learning_rate": 4.312604393718925e-05, "elbo": 3.45172, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5969974196575185}
{"step": 95445, "loss": 3.4366901397705076, "grad_norm": 0.7439223527908325, "learning_rate": 4.304825839511632e-05, "elbo": 3.43669, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.597290640394089}
{"step": 95450, "loss": 3.424958419799805, "grad_norm": 1.1417007446289062, "learning_rate": 4.2970490005774716e-05, "elbo": 3.42496, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.597583861130659}
{"step": 95455, "loss": 3.4241592407226564, "grad_norm": 0.7166954874992371, "learning_rate": 4.289273896105027e-05, "elbo": 3.42416, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.59787708186723}
{"step": 95460, "loss": 3.4199493408203123, "grad_norm": 0.7078045010566711, "learning_rate": 4.28150054527859e-05, "elbo": 3.41995, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.5981703026038}
{"step": 95465, "loss": 3.4153278350830076, "grad_norm": 0.7024790048599243, "learning_rate": 4.273728967278137e-05, "elbo": 3.41533, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.598463523340371}
{"step": 95470, "loss": 3.43363037109375, "grad_norm": 0.6525975465774536, "learning_rate": 4.2659591812792585e-05, "elbo": 3.43363, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.598756744076941}
{"step": 95475, "loss": 3.417340850830078, "grad_norm": 0.7914624810218811, "learning_rate": 4.258191206453131e-05, "elbo": 3.41734, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.599049964813512}
{"step": 95480, "loss": 3.402386474609375, "grad_norm": 0.704647421836853, "learning_rate": 4.250425061966461e-05, "elbo": 3.40239, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.599343185550082}
{"step": 95485, "loss": 3.437225341796875, "grad_norm": 0.6974827647209167, "learning_rate": 4.242660766981438e-05, "elbo": 3.43723, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.599636406286653}
{"step": 95490, "loss": 3.419169616699219, "grad_norm": 0.7364533543586731, "learning_rate": 4.234898340655687e-05, "elbo": 3.41917, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.599929627023223}
{"step": 95495, "loss": 3.4167716979980467, "grad_norm": 0.6982248425483704, "learning_rate": 4.2271378021422246e-05, "elbo": 3.41677, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6002228477597935}
{"step": 95500, "loss": 3.439240264892578, "grad_norm": 0.6991705894470215, "learning_rate": 4.219379170589407e-05, "elbo": 3.43924, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.600516068496364}
{"step": 95505, "loss": 3.4249366760253905, "grad_norm": 0.6977992057800293, "learning_rate": 4.211622465140887e-05, "elbo": 3.42494, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.600809289232934}
{"step": 95510, "loss": 3.4284294128417967, "grad_norm": 0.786027729511261, "learning_rate": 4.203867704935566e-05, "elbo": 3.42843, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.601102509969505}
{"step": 95515, "loss": 3.439159393310547, "grad_norm": 0.737409234046936, "learning_rate": 4.1961149091075376e-05, "elbo": 3.43916, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.601395730706075}
{"step": 95520, "loss": 3.4260009765625, "grad_norm": 0.7313119173049927, "learning_rate": 4.188364096786063e-05, "elbo": 3.426, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.601688951442646}
{"step": 95525, "loss": 3.4232757568359373, "grad_norm": 0.8246030807495117, "learning_rate": 4.1806152870954935e-05, "elbo": 3.42328, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.601982172179216}
{"step": 95530, "loss": 3.4360015869140623, "grad_norm": 1.1883561611175537, "learning_rate": 4.172868499155254e-05, "elbo": 3.436, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.602275392915787}
{"step": 95535, "loss": 3.4166282653808593, "grad_norm": 0.7315513491630554, "learning_rate": 4.165123752079768e-05, "elbo": 3.41663, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.602568613652357}
{"step": 95540, "loss": 3.443050003051758, "grad_norm": 0.732970654964447, "learning_rate": 4.157381064978432e-05, "elbo": 3.44305, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.602861834388928}
{"step": 95545, "loss": 3.410797882080078, "grad_norm": 0.6814823150634766, "learning_rate": 4.1496404569555554e-05, "elbo": 3.4108, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.603155055125498}
{"step": 95550, "loss": 3.420561599731445, "grad_norm": 0.7445895075798035, "learning_rate": 4.14190194711032e-05, "elbo": 3.42056, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.603448275862069}
{"step": 95555, "loss": 3.410540771484375, "grad_norm": 0.7143089175224304, "learning_rate": 4.134165554536728e-05, "elbo": 3.41054, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.603741496598639}
{"step": 95560, "loss": 3.425215148925781, "grad_norm": 0.68724524974823, "learning_rate": 4.126431298323561e-05, "elbo": 3.42522, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.60403471733521}
{"step": 95565, "loss": 3.4318065643310547, "grad_norm": 0.7199627161026001, "learning_rate": 4.118699197554327e-05, "elbo": 3.43181, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.604327938071781}
{"step": 95570, "loss": 3.4275192260742187, "grad_norm": 0.7514951229095459, "learning_rate": 4.110969271307215e-05, "elbo": 3.42752, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.604621158808351}
{"step": 95575, "loss": 3.4167747497558594, "grad_norm": 0.7106547355651855, "learning_rate": 4.103241538655049e-05, "elbo": 3.41677, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.604914379544922}
{"step": 95580, "loss": 3.4195011138916014, "grad_norm": 0.6767146587371826, "learning_rate": 4.095516018665244e-05, "elbo": 3.4195, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.605207600281492}
{"step": 95585, "loss": 3.432404708862305, "grad_norm": 0.7064186334609985, "learning_rate": 4.087792730399749e-05, "elbo": 3.4324, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.605500821018063}
{"step": 95590, "loss": 3.433083343505859, "grad_norm": 0.6842581033706665, "learning_rate": 4.080071692915014e-05, "elbo": 3.43308, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.605794041754633}
{"step": 95595, "loss": 3.4434410095214845, "grad_norm": 0.7277331352233887, "learning_rate": 4.072352925261927e-05, "elbo": 3.44344, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6060872624912035}
{"step": 95600, "loss": 3.4404502868652345, "grad_norm": 0.7249635457992554, "learning_rate": 4.0646364464857854e-05, "elbo": 3.44045, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.606380483227774}
{"step": 95605, "loss": 3.4281707763671876, "grad_norm": 1.0989811420440674, "learning_rate": 4.056922275626227e-05, "elbo": 3.42817, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.606673703964344}
{"step": 95610, "loss": 3.418560028076172, "grad_norm": 0.6949677467346191, "learning_rate": 4.049210431717208e-05, "elbo": 3.41856, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.606966924700915}
{"step": 95615, "loss": 3.4421642303466795, "grad_norm": 0.7067316174507141, "learning_rate": 4.04150093378693e-05, "elbo": 3.44216, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.607260145437485}
{"step": 95620, "loss": 3.4216739654541017, "grad_norm": 0.9067445993423462, "learning_rate": 4.0337938008578204e-05, "elbo": 3.42167, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.607553366174056}
{"step": 95625, "loss": 3.412842559814453, "grad_norm": 0.7855067253112793, "learning_rate": 4.026089051946457e-05, "elbo": 3.41284, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.607846586910626}
{"step": 95630, "loss": 3.4373374938964845, "grad_norm": 0.941713273525238, "learning_rate": 4.0183867060635446e-05, "elbo": 3.43734, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.608139807647197}
{"step": 95635, "loss": 3.4054332733154298, "grad_norm": 0.7208383083343506, "learning_rate": 4.010686782213855e-05, "elbo": 3.40543, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.608433028383767}
{"step": 95640, "loss": 3.4165283203125, "grad_norm": 0.6960484981536865, "learning_rate": 4.002989299396187e-05, "elbo": 3.41653, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.608726249120338}
{"step": 95645, "loss": 3.4350276947021485, "grad_norm": 0.7183679938316345, "learning_rate": 3.9952942766033114e-05, "elbo": 3.43503, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.609019469856908}
{"step": 95650, "loss": 3.429637908935547, "grad_norm": 0.6784607172012329, "learning_rate": 3.987601732821935e-05, "elbo": 3.42964, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6093126905934785}
{"step": 95655, "loss": 3.4348262786865233, "grad_norm": 0.7106271386146545, "learning_rate": 3.9799116870326415e-05, "elbo": 3.43483, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.609605911330049}
{"step": 95660, "loss": 3.4081497192382812, "grad_norm": 0.713602602481842, "learning_rate": 3.9722241582098574e-05, "elbo": 3.40815, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.609899132066619}
{"step": 95665, "loss": 3.403001403808594, "grad_norm": 0.6665305495262146, "learning_rate": 3.9645391653217945e-05, "elbo": 3.403, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.610192352803191}
{"step": 95670, "loss": 3.4132125854492186, "grad_norm": 0.7036917805671692, "learning_rate": 3.9568567273304094e-05, "elbo": 3.41321, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.61048557353976}
{"step": 95675, "loss": 3.418699264526367, "grad_norm": 0.9352855086326599, "learning_rate": 3.949176863191353e-05, "elbo": 3.4187, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.610778794276332}
{"step": 95680, "loss": 3.433758544921875, "grad_norm": 0.6831185221672058, "learning_rate": 3.941499591853928e-05, "elbo": 3.43376, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.611072015012902}
{"step": 95685, "loss": 3.43594970703125, "grad_norm": 0.7229366898536682, "learning_rate": 3.933824932261037e-05, "elbo": 3.43595, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6113652357494725}
{"step": 95690, "loss": 3.422520065307617, "grad_norm": 0.7239450216293335, "learning_rate": 3.926152903349143e-05, "elbo": 3.42252, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.611658456486043}
{"step": 95695, "loss": 3.4107086181640627, "grad_norm": 0.6993048787117004, "learning_rate": 3.918483524048207e-05, "elbo": 3.41071, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6119516772226135}
{"step": 95700, "loss": 3.438798522949219, "grad_norm": 0.6747066974639893, "learning_rate": 3.91081681328167e-05, "elbo": 3.4388, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.612244897959184}
{"step": 95705, "loss": 3.4243759155273437, "grad_norm": 0.7093331217765808, "learning_rate": 3.90315278996637e-05, "elbo": 3.42438, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.612538118695754}
{"step": 95710, "loss": 3.419092559814453, "grad_norm": 0.7243168950080872, "learning_rate": 3.8954914730125304e-05, "elbo": 3.41909, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.612831339432325}
{"step": 95715, "loss": 3.4415748596191404, "grad_norm": 0.6929869651794434, "learning_rate": 3.887832881323681e-05, "elbo": 3.44157, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.613124560168895}
{"step": 95720, "loss": 3.4204673767089844, "grad_norm": 0.8233240246772766, "learning_rate": 3.880177033796643e-05, "elbo": 3.42047, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.613417780905466}
{"step": 95725, "loss": 3.4123077392578125, "grad_norm": 0.713988721370697, "learning_rate": 3.872523949321454e-05, "elbo": 3.41231, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.613711001642036}
{"step": 95730, "loss": 3.4213516235351564, "grad_norm": 0.7632936835289001, "learning_rate": 3.864873646781341e-05, "elbo": 3.42135, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.614004222378607}
{"step": 95735, "loss": 3.445229339599609, "grad_norm": 0.6824045777320862, "learning_rate": 3.857226145052665e-05, "elbo": 3.44523, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.614297443115177}
{"step": 95740, "loss": 3.4185111999511717, "grad_norm": 0.7255529761314392, "learning_rate": 3.849581463004877e-05, "elbo": 3.41851, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6145906638517475}
{"step": 95745, "loss": 3.4215156555175783, "grad_norm": 0.6649523973464966, "learning_rate": 3.841939619500468e-05, "elbo": 3.42152, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.614883884588318}
{"step": 95750, "loss": 3.4406497955322264, "grad_norm": 0.6958101391792297, "learning_rate": 3.83430063339493e-05, "elbo": 3.44065, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6151771053248885}
{"step": 95755, "loss": 3.431432342529297, "grad_norm": 0.6826584339141846, "learning_rate": 3.8266645235367e-05, "elbo": 3.43143, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.615470326061459}
{"step": 95760, "loss": 3.4237743377685548, "grad_norm": 0.7110628485679626, "learning_rate": 3.819031308767119e-05, "elbo": 3.42377, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.615763546798029}
{"step": 95765, "loss": 3.4351707458496095, "grad_norm": 0.7278398871421814, "learning_rate": 3.811401007920388e-05, "elbo": 3.43517, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6160567675346}
{"step": 95770, "loss": 3.424853515625, "grad_norm": 0.7105251550674438, "learning_rate": 3.803773639823515e-05, "elbo": 3.42485, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.61634998827117}
{"step": 95775, "loss": 3.430910110473633, "grad_norm": 0.7240325808525085, "learning_rate": 3.796149223296272e-05, "elbo": 3.43091, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.616643209007741}
{"step": 95780, "loss": 3.4239738464355467, "grad_norm": 0.7687380313873291, "learning_rate": 3.7885277771511495e-05, "elbo": 3.42397, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.616936429744311}
{"step": 95785, "loss": 3.419333648681641, "grad_norm": 0.8297430872917175, "learning_rate": 3.780909320193308e-05, "elbo": 3.41933, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.617229650480882}
{"step": 95790, "loss": 3.401885223388672, "grad_norm": 0.6642237901687622, "learning_rate": 3.7732938712205336e-05, "elbo": 3.40189, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.617522871217453}
{"step": 95795, "loss": 3.4236305236816404, "grad_norm": 0.8189055919647217, "learning_rate": 3.7656814490231865e-05, "elbo": 3.42363, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6178160919540225}
{"step": 95800, "loss": 3.4365882873535156, "grad_norm": 0.7017124891281128, "learning_rate": 3.758072072384167e-05, "elbo": 3.43659, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.618109312690594}
{"step": 95805, "loss": 3.408245849609375, "grad_norm": 0.701330840587616, "learning_rate": 3.7504657600788485e-05, "elbo": 3.40825, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.618402533427164}
{"step": 95810, "loss": 3.4255470275878905, "grad_norm": 0.7059260010719299, "learning_rate": 3.742862530875058e-05, "elbo": 3.42555, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.618695754163735}
{"step": 95815, "loss": 3.41192741394043, "grad_norm": 0.7309697866439819, "learning_rate": 3.735262403533002e-05, "elbo": 3.41193, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.618988974900305}
{"step": 95820, "loss": 3.399666595458984, "grad_norm": 0.914158284664154, "learning_rate": 3.7276653968052434e-05, "elbo": 3.39967, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.619282195636876}
{"step": 95825, "loss": 3.4422542572021486, "grad_norm": 1.1975423097610474, "learning_rate": 3.720071529436637e-05, "elbo": 3.44225, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.619575416373446}
{"step": 95830, "loss": 3.4245330810546877, "grad_norm": 0.6927534937858582, "learning_rate": 3.712480820164299e-05, "elbo": 3.42453, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.619868637110017}
{"step": 95835, "loss": 3.429401397705078, "grad_norm": 0.6826397776603699, "learning_rate": 3.704893287717548e-05, "elbo": 3.4294, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.620161857846587}
{"step": 95840, "loss": 3.4253990173339846, "grad_norm": 0.7011718153953552, "learning_rate": 3.697308950817868e-05, "elbo": 3.4254, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6204550785831575}
{"step": 95845, "loss": 3.417094421386719, "grad_norm": 0.7569675445556641, "learning_rate": 3.689727828178854e-05, "elbo": 3.41709, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.620748299319728}
{"step": 95850, "loss": 3.4327934265136717, "grad_norm": 0.6833641529083252, "learning_rate": 3.6821499385061744e-05, "elbo": 3.43279, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.621041520056298}
{"step": 95855, "loss": 3.4271263122558593, "grad_norm": 0.7063555121421814, "learning_rate": 3.674575300497517e-05, "elbo": 3.42713, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.621334740792869}
{"step": 95860, "loss": 3.4121089935302735, "grad_norm": 0.7079258561134338, "learning_rate": 3.6670039328425505e-05, "elbo": 3.41211, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.621627961529439}
{"step": 95865, "loss": 3.4084625244140625, "grad_norm": 0.81888347864151, "learning_rate": 3.659435854222869e-05, "elbo": 3.40846, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.62192118226601}
{"step": 95870, "loss": 3.4152957916259767, "grad_norm": 0.7425854802131653, "learning_rate": 3.651871083311957e-05, "elbo": 3.4153, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.62221440300258}
{"step": 95875, "loss": 3.407077407836914, "grad_norm": 0.7120069265365601, "learning_rate": 3.644309638775132e-05, "elbo": 3.40708, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.622507623739151}
{"step": 95880, "loss": 3.4315212249755858, "grad_norm": 0.6897122263908386, "learning_rate": 3.636751539269511e-05, "elbo": 3.43152, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.622800844475721}
{"step": 95885, "loss": 3.4134571075439455, "grad_norm": 0.7740967273712158, "learning_rate": 3.629196803443946e-05, "elbo": 3.41346, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.623094065212292}
{"step": 95890, "loss": 3.4319889068603517, "grad_norm": 0.6977787613868713, "learning_rate": 3.621645449939005e-05, "elbo": 3.43199, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.623387285948862}
{"step": 95895, "loss": 3.4311187744140623, "grad_norm": 0.7456921339035034, "learning_rate": 3.614097497386894e-05, "elbo": 3.43112, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6236805066854325}
{"step": 95900, "loss": 3.424778366088867, "grad_norm": 0.6956301927566528, "learning_rate": 3.6065529644114436e-05, "elbo": 3.42478, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.623973727422003}
{"step": 95905, "loss": 3.411589813232422, "grad_norm": 0.7254340052604675, "learning_rate": 3.599011869628033e-05, "elbo": 3.41159, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.624266948158573}
{"step": 95910, "loss": 3.430519866943359, "grad_norm": 0.7131888270378113, "learning_rate": 3.5914742316435665e-05, "elbo": 3.43052, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.624560168895144}
{"step": 95915, "loss": 3.4132938385009766, "grad_norm": 0.7273009419441223, "learning_rate": 3.583940069056415e-05, "elbo": 3.41329, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.624853389631715}
{"step": 95920, "loss": 3.4140907287597657, "grad_norm": 1.1724534034729004, "learning_rate": 3.576409400456379e-05, "elbo": 3.41409, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.625146610368285}
{"step": 95925, "loss": 3.427132415771484, "grad_norm": 0.687127411365509, "learning_rate": 3.5688822444246294e-05, "elbo": 3.42713, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.625439831104856}
{"step": 95930, "loss": 3.414066696166992, "grad_norm": 0.745016872882843, "learning_rate": 3.561358619533681e-05, "elbo": 3.41407, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.625733051841427}
{"step": 95935, "loss": 3.426085662841797, "grad_norm": 0.7403693795204163, "learning_rate": 3.553838544347326e-05, "elbo": 3.42609, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.626026272577997}
{"step": 95940, "loss": 3.4305763244628906, "grad_norm": 0.6671233177185059, "learning_rate": 3.5463220374206055e-05, "elbo": 3.43058, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6263194933145675}
{"step": 95945, "loss": 3.4191463470458983, "grad_norm": 0.6891377568244934, "learning_rate": 3.5388091172997504e-05, "elbo": 3.41915, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.626612714051138}
{"step": 95950, "loss": 3.413995361328125, "grad_norm": 0.747009813785553, "learning_rate": 3.531299802522148e-05, "elbo": 3.414, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.626905934787708}
{"step": 95955, "loss": 3.4196189880371093, "grad_norm": 1.0348314046859741, "learning_rate": 3.523794111616281e-05, "elbo": 3.41962, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.627199155524279}
{"step": 95960, "loss": 3.420340347290039, "grad_norm": 0.7106852531433105, "learning_rate": 3.5162920631017004e-05, "elbo": 3.42034, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.627492376260849}
{"step": 95965, "loss": 3.4241783142089846, "grad_norm": 0.7185989618301392, "learning_rate": 3.508793675488961e-05, "elbo": 3.42418, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.62778559699742}
{"step": 95970, "loss": 3.4169548034667967, "grad_norm": 0.6834771037101746, "learning_rate": 3.501298967279594e-05, "elbo": 3.41695, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.62807881773399}
{"step": 95975, "loss": 3.4140396118164062, "grad_norm": 0.6759915351867676, "learning_rate": 3.49380795696604e-05, "elbo": 3.41404, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.628372038470561}
{"step": 95980, "loss": 3.4202720642089846, "grad_norm": 0.6956703662872314, "learning_rate": 3.486320663031629e-05, "elbo": 3.42027, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.628665259207131}
{"step": 95985, "loss": 3.413155746459961, "grad_norm": 0.7442492842674255, "learning_rate": 3.478837103950509e-05, "elbo": 3.41316, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.628958479943702}
{"step": 95990, "loss": 3.4289485931396486, "grad_norm": 0.7843080759048462, "learning_rate": 3.471357298187624e-05, "elbo": 3.42895, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.629251700680272}
{"step": 95995, "loss": 3.429259490966797, "grad_norm": 0.7018866539001465, "learning_rate": 3.4638812641986454e-05, "elbo": 3.42926, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6295449214168425}
{"step": 96000, "loss": 3.427113342285156, "grad_norm": 0.6877143979072571, "learning_rate": 3.456409020429947e-05, "elbo": 3.42711, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.629838142153413}
{"step": 96005, "loss": 3.4282745361328124, "grad_norm": 0.7228349447250366, "learning_rate": 3.4489405853185465e-05, "elbo": 3.42827, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.630131362889983}
{"step": 96005, "loss": 3.4282745361328124, "grad_norm": 0.7228349447250366, "learning_rate": 3.4489405853185465e-05, "elbo": 3.42827, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.630131362889983}
{"step": 96010, "loss": 3.406328582763672, "grad_norm": 0.6798956394195557, "learning_rate": 3.441475977292066e-05, "elbo": 3.40633, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.630424583626554}
{"step": 96015, "loss": 3.4157184600830077, "grad_norm": 0.7844358086585999, "learning_rate": 3.4340152147686824e-05, "elbo": 3.41572, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.630717804363124}
{"step": 96020, "loss": 3.4053131103515626, "grad_norm": 0.6768108010292053, "learning_rate": 3.426558316157087e-05, "elbo": 3.40531, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.631011025099695}
{"step": 96025, "loss": 3.405816650390625, "grad_norm": 0.706759512424469, "learning_rate": 3.4191052998564344e-05, "elbo": 3.40582, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.631304245836265}
{"step": 96030, "loss": 3.413542938232422, "grad_norm": 0.6823853850364685, "learning_rate": 3.411656184256304e-05, "elbo": 3.41354, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.631597466572836}
{"step": 96035, "loss": 3.4162109375, "grad_norm": 0.6909129619598389, "learning_rate": 3.4042109877366446e-05, "elbo": 3.41621, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.631890687309406}
{"step": 96040, "loss": 3.4219715118408205, "grad_norm": 0.7158769369125366, "learning_rate": 3.3967697286677415e-05, "elbo": 3.42197, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6321839080459775}
{"step": 96045, "loss": 3.4135814666748048, "grad_norm": 0.8108280897140503, "learning_rate": 3.38933242541016e-05, "elbo": 3.41358, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.632477128782547}
{"step": 96050, "loss": 3.4263931274414063, "grad_norm": 0.6930571794509888, "learning_rate": 3.3818990963147065e-05, "elbo": 3.42639, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.632770349519118}
{"step": 96055, "loss": 3.4053276062011717, "grad_norm": 0.755524754524231, "learning_rate": 3.374469759722383e-05, "elbo": 3.40533, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.633063570255689}
{"step": 96060, "loss": 3.4139934539794923, "grad_norm": 0.7334926724433899, "learning_rate": 3.367044433964338e-05, "elbo": 3.41399, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.633356790992259}
{"step": 96065, "loss": 3.415772247314453, "grad_norm": 0.6982980370521545, "learning_rate": 3.359623137361825e-05, "elbo": 3.41577, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.63365001172883}
{"step": 96070, "loss": 3.4222469329833984, "grad_norm": 0.7407723665237427, "learning_rate": 3.352205888226156e-05, "elbo": 3.42225, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6339432324654}
{"step": 96075, "loss": 3.4175186157226562, "grad_norm": 0.6916913390159607, "learning_rate": 3.344792704858654e-05, "elbo": 3.41752, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.634236453201971}
{"step": 96080, "loss": 3.4136074066162108, "grad_norm": 0.7518190145492554, "learning_rate": 3.337383605550615e-05, "elbo": 3.41361, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.634529673938541}
{"step": 96085, "loss": 3.427103042602539, "grad_norm": 0.6946937441825867, "learning_rate": 3.3299786085832516e-05, "elbo": 3.4271, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6348228946751115}
{"step": 96090, "loss": 3.433135223388672, "grad_norm": 0.7262274026870728, "learning_rate": 3.3225777322276616e-05, "elbo": 3.43314, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.635116115411682}
{"step": 96095, "loss": 3.4128955841064452, "grad_norm": 1.5162322521209717, "learning_rate": 3.315180994744769e-05, "elbo": 3.4129, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6354093361482525}
{"step": 96100, "loss": 3.4228595733642577, "grad_norm": 0.7752269506454468, "learning_rate": 3.3077884143852886e-05, "elbo": 3.42286, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.635702556884823}
{"step": 96105, "loss": 3.4211654663085938, "grad_norm": 0.6829132437705994, "learning_rate": 3.300400009389678e-05, "elbo": 3.42117, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.635995777621393}
{"step": 96110, "loss": 3.4160667419433595, "grad_norm": 0.7741963863372803, "learning_rate": 3.2930157979880925e-05, "elbo": 3.41607, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.636288998357964}
{"step": 96115, "loss": 3.4203582763671876, "grad_norm": 0.6958585977554321, "learning_rate": 3.285635798400338e-05, "elbo": 3.42036, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.636582219094534}
{"step": 96120, "loss": 3.4227996826171876, "grad_norm": 0.6761490702629089, "learning_rate": 3.278260028835832e-05, "elbo": 3.4228, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.636875439831105}
{"step": 96125, "loss": 3.407106399536133, "grad_norm": 0.6943035125732422, "learning_rate": 3.2708885074935514e-05, "elbo": 3.40711, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.637168660567675}
{"step": 96130, "loss": 3.43138427734375, "grad_norm": 1.0815024375915527, "learning_rate": 3.263521252561993e-05, "elbo": 3.43138, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.637461881304246}
{"step": 96135, "loss": 3.4028892517089844, "grad_norm": 0.7032763957977295, "learning_rate": 3.2561582822191274e-05, "elbo": 3.40289, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.637755102040816}
{"step": 96140, "loss": 3.4269134521484377, "grad_norm": 0.7135340571403503, "learning_rate": 3.2487996146323465e-05, "elbo": 3.42691, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6380483227773865}
{"step": 96145, "loss": 3.4293441772460938, "grad_norm": 0.7491982579231262, "learning_rate": 3.241445267958438e-05, "elbo": 3.42934, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.638341543513957}
{"step": 96150, "loss": 3.403158950805664, "grad_norm": 0.6735658049583435, "learning_rate": 3.234095260343515e-05, "elbo": 3.40316, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6386347642505275}
{"step": 96155, "loss": 3.4224853515625, "grad_norm": 0.7076454162597656, "learning_rate": 3.226749609922997e-05, "elbo": 3.42249, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.638927984987098}
{"step": 96160, "loss": 3.4249492645263673, "grad_norm": 0.7404710650444031, "learning_rate": 3.21940833482154e-05, "elbo": 3.42495, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.639221205723668}
{"step": 96165, "loss": 3.4165390014648436, "grad_norm": 0.8081525564193726, "learning_rate": 3.2120714531530146e-05, "elbo": 3.41654, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.63951442646024}
{"step": 96170, "loss": 3.4275741577148438, "grad_norm": 0.6690447926521301, "learning_rate": 3.204738983020445e-05, "elbo": 3.42757, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.639807647196809}
{"step": 96175, "loss": 3.430075836181641, "grad_norm": 0.7637477517127991, "learning_rate": 3.197410942515975e-05, "elbo": 3.43008, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.640100867933381}
{"step": 96180, "loss": 3.4185882568359376, "grad_norm": 0.7091873288154602, "learning_rate": 3.190087349720813e-05, "elbo": 3.41859, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.640394088669951}
{"step": 96185, "loss": 3.4095687866210938, "grad_norm": 0.760156512260437, "learning_rate": 3.1827682227051983e-05, "elbo": 3.40957, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6406873094065215}
{"step": 96190, "loss": 3.4372806549072266, "grad_norm": 0.6985432505607605, "learning_rate": 3.175453579528347e-05, "elbo": 3.43728, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.640980530143092}
{"step": 96195, "loss": 3.4213577270507813, "grad_norm": 0.7357938289642334, "learning_rate": 3.1681434382384165e-05, "elbo": 3.42136, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.641273750879662}
{"step": 96200, "loss": 3.4162574768066407, "grad_norm": 0.7018392086029053, "learning_rate": 3.160837816872452e-05, "elbo": 3.41626, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.641566971616233}
{"step": 96200, "loss": 3.4162574768066407, "grad_norm": 0.7018392086029053, "learning_rate": 3.160837816872452e-05, "elbo": 3.41626, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.641566971616233}
{"step": 96205, "loss": 3.4202022552490234, "grad_norm": 0.6866207718849182, "learning_rate": 3.153536733456349e-05, "elbo": 3.4202, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.641860192352803}
{"step": 96210, "loss": 3.42657585144043, "grad_norm": 1.0340995788574219, "learning_rate": 3.146240206004805e-05, "elbo": 3.42658, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.642153413089374}
{"step": 96215, "loss": 3.4340328216552733, "grad_norm": 0.7599058151245117, "learning_rate": 3.138948252521275e-05, "elbo": 3.43403, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.642446633825944}
{"step": 96220, "loss": 3.431257629394531, "grad_norm": 0.7365310192108154, "learning_rate": 3.1316608909979306e-05, "elbo": 3.43126, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.642739854562515}
{"step": 96225, "loss": 3.4186866760253904, "grad_norm": 0.8064673542976379, "learning_rate": 3.124378139415614e-05, "elbo": 3.41869, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.643033075299085}
{"step": 96230, "loss": 3.408052444458008, "grad_norm": 0.7436060905456543, "learning_rate": 3.1171000157437835e-05, "elbo": 3.40805, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.643326296035656}
{"step": 96235, "loss": 3.3998504638671876, "grad_norm": 0.6901317834854126, "learning_rate": 3.1098265379404954e-05, "elbo": 3.39985, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.643619516772226}
{"step": 96240, "loss": 3.4008750915527344, "grad_norm": 0.706277072429657, "learning_rate": 3.102557723952324e-05, "elbo": 3.40087, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6439127375087965}
{"step": 96245, "loss": 3.4185409545898438, "grad_norm": 0.7123451828956604, "learning_rate": 3.0952935917143533e-05, "elbo": 3.41854, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.644205958245367}
{"step": 96250, "loss": 3.4144882202148437, "grad_norm": 0.7259713411331177, "learning_rate": 3.088034159150102e-05, "elbo": 3.41449, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.644499178981937}
{"step": 96255, "loss": 3.392154312133789, "grad_norm": 0.7490668892860413, "learning_rate": 3.0807794441715e-05, "elbo": 3.39215, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.644792399718508}
{"step": 96260, "loss": 3.4049087524414063, "grad_norm": 0.731709361076355, "learning_rate": 3.073529464678835e-05, "elbo": 3.40491, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.645085620455078}
{"step": 96265, "loss": 3.431304931640625, "grad_norm": 0.7395435571670532, "learning_rate": 3.066284238560713e-05, "elbo": 3.43131, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.645378841191649}
{"step": 96270, "loss": 3.434009552001953, "grad_norm": 0.6950244307518005, "learning_rate": 3.059043783694005e-05, "elbo": 3.43401, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.645672061928219}
{"step": 96275, "loss": 3.4201404571533205, "grad_norm": 0.7114172577857971, "learning_rate": 3.051808117943817e-05, "elbo": 3.42014, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.64596528266479}
{"step": 96280, "loss": 3.427045440673828, "grad_norm": 0.7397038340568542, "learning_rate": 3.0445772591634336e-05, "elbo": 3.42705, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.646258503401361}
{"step": 96285, "loss": 3.421692657470703, "grad_norm": 0.7745075225830078, "learning_rate": 3.0373512251942814e-05, "elbo": 3.42169, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.646551724137931}
{"step": 96290, "loss": 3.432590866088867, "grad_norm": 0.7083678841590881, "learning_rate": 3.030130033865879e-05, "elbo": 3.43259, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.646844944874502}
{"step": 96290, "loss": 3.432590866088867, "grad_norm": 0.7083678841590881, "learning_rate": 3.030130033865879e-05, "elbo": 3.43259, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.646844944874502}
{"step": 96295, "loss": 3.4122528076171874, "grad_norm": 1.0844612121582031, "learning_rate": 3.0229137029957993e-05, "elbo": 3.41225, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.647138165611072}
{"step": 96300, "loss": 3.396739959716797, "grad_norm": 0.6671849489212036, "learning_rate": 3.0157022503896203e-05, "elbo": 3.39674, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.647431386347643}
{"step": 96305, "loss": 3.4235885620117186, "grad_norm": 0.6898276805877686, "learning_rate": 3.008495693840887e-05, "elbo": 3.42359, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.647724607084213}
{"step": 96305, "loss": 3.4235885620117186, "grad_norm": 0.6898276805877686, "learning_rate": 3.008495693840887e-05, "elbo": 3.42359, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.647724607084213}
{"step": 96310, "loss": 3.4352191925048827, "grad_norm": 0.7014896273612976, "learning_rate": 3.001294051131058e-05, "elbo": 3.43522, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.648017827820784}
{"step": 96315, "loss": 3.4202484130859374, "grad_norm": 0.6577672958374023, "learning_rate": 2.9940973400294743e-05, "elbo": 3.42025, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.648311048557354}
{"step": 96320, "loss": 3.4256282806396485, "grad_norm": 0.728503406047821, "learning_rate": 2.9869055782933e-05, "elbo": 3.42563, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.648604269293925}
{"step": 96320, "loss": 3.4256282806396485, "grad_norm": 0.728503406047821, "learning_rate": 2.9869055782933e-05, "elbo": 3.42563, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.648604269293925}
{"step": 96325, "loss": 3.3967491149902345, "grad_norm": 0.7524531483650208, "learning_rate": 2.979718783667499e-05, "elbo": 3.39675, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.648897490030495}
{"step": 96330, "loss": 3.4298141479492186, "grad_norm": 0.6803666353225708, "learning_rate": 2.9725369738847653e-05, "elbo": 3.42981, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.649190710767066}
{"step": 96335, "loss": 3.4065704345703125, "grad_norm": 0.6968683004379272, "learning_rate": 2.965360166665508e-05, "elbo": 3.40657, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.649483931503636}
{"step": 96340, "loss": 3.4106113433837892, "grad_norm": 0.8907023072242737, "learning_rate": 2.958188379717779e-05, "elbo": 3.41061, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6497771522402065}
{"step": 96345, "loss": 3.4063156127929686, "grad_norm": 0.8625056743621826, "learning_rate": 2.951021630737255e-05, "elbo": 3.40632, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.650070372976777}
{"step": 96350, "loss": 3.420513153076172, "grad_norm": 0.6671551465988159, "learning_rate": 2.9438599374071725e-05, "elbo": 3.42051, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.650363593713347}
{"step": 96355, "loss": 3.40762939453125, "grad_norm": 0.6976247429847717, "learning_rate": 2.9367033173983006e-05, "elbo": 3.40763, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.650656814449918}
{"step": 96360, "loss": 3.4192214965820313, "grad_norm": 0.7275846600532532, "learning_rate": 2.9295517883688855e-05, "elbo": 3.41922, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.650950035186488}
{"step": 96365, "loss": 3.420730209350586, "grad_norm": 0.6668893694877625, "learning_rate": 2.9224053679646167e-05, "elbo": 3.42073, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.651243255923059}
{"step": 96370, "loss": 3.415117645263672, "grad_norm": 0.6946319937705994, "learning_rate": 2.9152640738185733e-05, "elbo": 3.41512, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.651536476659629}
{"step": 96375, "loss": 3.414545440673828, "grad_norm": 0.7992522120475769, "learning_rate": 2.908127923551194e-05, "elbo": 3.41455, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6518296973962}
{"step": 96380, "loss": 3.413197326660156, "grad_norm": 0.7103984355926514, "learning_rate": 2.9009969347702116e-05, "elbo": 3.4132, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.65212291813277}
{"step": 96385, "loss": 3.409871292114258, "grad_norm": 0.6965145468711853, "learning_rate": 2.8938711250706395e-05, "elbo": 3.40987, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.652416138869341}
{"step": 96390, "loss": 3.4132614135742188, "grad_norm": 0.7107099294662476, "learning_rate": 2.886750512034703e-05, "elbo": 3.41326, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.652709359605911}
{"step": 96390, "loss": 3.4132614135742188, "grad_norm": 0.7107099294662476, "learning_rate": 2.886750512034703e-05, "elbo": 3.41326, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.652709359605911}
{"step": 96395, "loss": 3.420305633544922, "grad_norm": 0.7100651264190674, "learning_rate": 2.8796351132318046e-05, "elbo": 3.42031, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6530025803424815}
{"step": 96400, "loss": 3.3919139862060548, "grad_norm": 0.682243824005127, "learning_rate": 2.872524946218484e-05, "elbo": 3.39191, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.653295801079052}
{"step": 96405, "loss": 3.4116226196289063, "grad_norm": 0.7400405406951904, "learning_rate": 2.8654200285383748e-05, "elbo": 3.41162, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.653589021815623}
{"step": 96405, "loss": 3.4116226196289063, "grad_norm": 0.7400405406951904, "learning_rate": 2.8654200285383748e-05, "elbo": 3.41162, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.653589021815623}
{"step": 96410, "loss": 3.4162479400634767, "grad_norm": 0.6807109117507935, "learning_rate": 2.8583203777221524e-05, "elbo": 3.41625, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.653882242552193}
{"step": 96415, "loss": 3.4271823883056642, "grad_norm": 0.7086926698684692, "learning_rate": 2.8512260112874996e-05, "elbo": 3.42718, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.654175463288764}
{"step": 96420, "loss": 3.3936405181884766, "grad_norm": 0.735729992389679, "learning_rate": 2.844136946739059e-05, "elbo": 3.39364, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.654468684025335}
{"step": 96425, "loss": 3.425550079345703, "grad_norm": 0.782109797000885, "learning_rate": 2.837053201568396e-05, "elbo": 3.42555, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.654761904761905}
{"step": 96430, "loss": 3.408990478515625, "grad_norm": 0.6860557198524475, "learning_rate": 2.8299747932539468e-05, "elbo": 3.40899, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6550551254984756}
{"step": 96435, "loss": 3.405847930908203, "grad_norm": 0.6763004064559937, "learning_rate": 2.822901739260978e-05, "elbo": 3.40585, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.655348346235046}
{"step": 96440, "loss": 3.41925048828125, "grad_norm": 0.6816961169242859, "learning_rate": 2.8158340570415477e-05, "elbo": 3.41925, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6556415669716165}
{"step": 96445, "loss": 3.411263275146484, "grad_norm": 0.7206222414970398, "learning_rate": 2.8087717640344618e-05, "elbo": 3.41126, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.655934787708187}
{"step": 96445, "loss": 3.411263275146484, "grad_norm": 0.7206222414970398, "learning_rate": 2.8087717640344618e-05, "elbo": 3.41126, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.655934787708187}
{"step": 96450, "loss": 3.402609634399414, "grad_norm": 0.7008801698684692, "learning_rate": 2.8017148776652246e-05, "elbo": 3.40261, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.656228008444757}
{"step": 96455, "loss": 3.4067703247070313, "grad_norm": 0.6990771889686584, "learning_rate": 2.7946634153460016e-05, "elbo": 3.40677, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.656521229181328}
{"step": 96460, "loss": 3.419635772705078, "grad_norm": 1.3955069780349731, "learning_rate": 2.7876173944755733e-05, "elbo": 3.41964, "mask_frac": 0.4998, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.656814449917898}
{"step": 96465, "loss": 3.4223236083984374, "grad_norm": 0.7088161706924438, "learning_rate": 2.7805768324393015e-05, "elbo": 3.42232, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.657107670654469}
{"step": 96470, "loss": 3.410099411010742, "grad_norm": 0.6785717010498047, "learning_rate": 2.7735417466090652e-05, "elbo": 3.4101, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.657400891391039}
{"step": 96475, "loss": 3.4211448669433593, "grad_norm": 0.6871699690818787, "learning_rate": 2.766512154343246e-05, "elbo": 3.42115, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.65769411212761}
{"step": 96480, "loss": 3.4077728271484373, "grad_norm": 1.0165574550628662, "learning_rate": 2.759488072986661e-05, "elbo": 3.40777, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.65798733286418}
{"step": 96485, "loss": 3.40941162109375, "grad_norm": 0.7963323593139648, "learning_rate": 2.7524695198705334e-05, "elbo": 3.40941, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6582805536007506}
{"step": 96490, "loss": 3.4192859649658205, "grad_norm": 0.6979706883430481, "learning_rate": 2.7454565123124432e-05, "elbo": 3.41929, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.658573774337321}
{"step": 96495, "loss": 3.4015804290771485, "grad_norm": 0.6954628229141235, "learning_rate": 2.7384490676162933e-05, "elbo": 3.40158, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6588669950738915}
{"step": 96500, "loss": 3.424285125732422, "grad_norm": 0.69565349817276, "learning_rate": 2.731447203072255e-05, "elbo": 3.42429, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.659160215810462}
{"step": 96505, "loss": 3.4055549621582033, "grad_norm": 1.09292471408844, "learning_rate": 2.7244509359567327e-05, "elbo": 3.40555, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.659453436547032}
{"step": 96510, "loss": 3.4032737731933596, "grad_norm": 0.7569338083267212, "learning_rate": 2.7174602835323182e-05, "elbo": 3.40327, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.659746657283603}
{"step": 96515, "loss": 3.404288482666016, "grad_norm": 0.6827086210250854, "learning_rate": 2.710475263047756e-05, "elbo": 3.40429, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.660039878020173}
{"step": 96520, "loss": 3.4246849060058593, "grad_norm": 0.8570061326026917, "learning_rate": 2.7034958917378867e-05, "elbo": 3.42469, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.660333098756744}
{"step": 96525, "loss": 3.409025955200195, "grad_norm": 0.683488130569458, "learning_rate": 2.6965221868236155e-05, "elbo": 3.40903, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.660626319493314}
{"step": 96530, "loss": 3.4176753997802733, "grad_norm": 0.6732129454612732, "learning_rate": 2.6895541655118652e-05, "elbo": 3.41768, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6609195402298855}
{"step": 96535, "loss": 3.408303070068359, "grad_norm": 0.7080658078193665, "learning_rate": 2.6825918449955383e-05, "elbo": 3.4083, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.661212760966455}
{"step": 96540, "loss": 3.401844024658203, "grad_norm": 0.6540197134017944, "learning_rate": 2.6756352424534674e-05, "elbo": 3.40184, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.661505981703026}
{"step": 96545, "loss": 3.411790466308594, "grad_norm": 0.777138352394104, "learning_rate": 2.6686843750503776e-05, "elbo": 3.41179, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.661799202439597}
{"step": 96550, "loss": 3.408632278442383, "grad_norm": 0.7128644585609436, "learning_rate": 2.6617392599368406e-05, "elbo": 3.40863, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.662092423176167}
{"step": 96555, "loss": 3.3958457946777343, "grad_norm": 0.6877671480178833, "learning_rate": 2.6547999142492447e-05, "elbo": 3.39585, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.662385643912738}
{"step": 96555, "loss": 3.3958457946777343, "grad_norm": 0.6877671480178833, "learning_rate": 2.6547999142492447e-05, "elbo": 3.39585, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.662385643912738}
{"step": 96560, "loss": 3.4098182678222657, "grad_norm": 0.7133236527442932, "learning_rate": 2.6478663551097267e-05, "elbo": 3.40982, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.662678864649308}
{"step": 96560, "loss": 3.4098182678222657, "grad_norm": 0.7133236527442932, "learning_rate": 2.6478663551097267e-05, "elbo": 3.40982, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.662678864649308}
{"step": 96565, "loss": 3.402507781982422, "grad_norm": 0.8033431768417358, "learning_rate": 2.6409385996261603e-05, "elbo": 3.40251, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.662972085385879}
{"step": 96570, "loss": 3.4092140197753906, "grad_norm": 0.770582914352417, "learning_rate": 2.6340166648920905e-05, "elbo": 3.40921, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.663265306122449}
{"step": 96570, "loss": 3.4092140197753906, "grad_norm": 0.770582914352417, "learning_rate": 2.6340166648920905e-05, "elbo": 3.40921, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.663265306122449}
{"step": 96575, "loss": 3.3993621826171876, "grad_norm": 0.6936294436454773, "learning_rate": 2.627100567986709e-05, "elbo": 3.39936, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.66355852685902}
{"step": 96580, "loss": 3.4138851165771484, "grad_norm": 0.8126202821731567, "learning_rate": 2.620190325974788e-05, "elbo": 3.41389, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.66385174759559}
{"step": 96585, "loss": 3.3970703125, "grad_norm": 1.1057277917861938, "learning_rate": 2.6132859559066702e-05, "elbo": 3.39707, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6641449683321605}
{"step": 96590, "loss": 3.3978450775146483, "grad_norm": 0.6890211701393127, "learning_rate": 2.6063874748182e-05, "elbo": 3.39784, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.664438189068731}
{"step": 96595, "loss": 3.3998947143554688, "grad_norm": 0.7064433097839355, "learning_rate": 2.5994948997306935e-05, "elbo": 3.39989, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.664731409805301}
{"step": 96600, "loss": 3.4064910888671873, "grad_norm": 0.7086924910545349, "learning_rate": 2.5926082476508928e-05, "elbo": 3.40649, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.665024630541872}
{"step": 96605, "loss": 3.424530029296875, "grad_norm": 0.6866959929466248, "learning_rate": 2.5857275355709317e-05, "elbo": 3.42453, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.665317851278442}
{"step": 96610, "loss": 3.4155250549316407, "grad_norm": 0.866165041923523, "learning_rate": 2.5788527804682804e-05, "elbo": 3.41553, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.665611072015013}
{"step": 96615, "loss": 3.409021759033203, "grad_norm": 1.4402801990509033, "learning_rate": 2.5719839993057142e-05, "elbo": 3.40902, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.665904292751583}
{"step": 96615, "loss": 3.409021759033203, "grad_norm": 1.4402801990509033, "learning_rate": 2.5719839993057142e-05, "elbo": 3.40902, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.665904292751583}
{"step": 96620, "loss": 3.403949737548828, "grad_norm": 0.6688353419303894, "learning_rate": 2.565121209031266e-05, "elbo": 3.40395, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.666197513488154}
{"step": 96625, "loss": 3.4028244018554688, "grad_norm": 1.4325450658798218, "learning_rate": 2.558264426578192e-05, "elbo": 3.40282, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.666490734224724}
{"step": 96630, "loss": 3.4132980346679687, "grad_norm": 0.7646943926811218, "learning_rate": 2.55141366886492e-05, "elbo": 3.4133, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.666783954961295}
{"step": 96635, "loss": 3.3909584045410157, "grad_norm": 0.6932616829872131, "learning_rate": 2.5445689527950133e-05, "elbo": 3.39096, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.667077175697865}
{"step": 96640, "loss": 3.414462661743164, "grad_norm": 0.6814020872116089, "learning_rate": 2.5377302952571257e-05, "elbo": 3.41446, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6673703964344355}
{"step": 96645, "loss": 3.4018180847167967, "grad_norm": 0.7103056311607361, "learning_rate": 2.5308977131249722e-05, "elbo": 3.40182, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.667663617171006}
{"step": 96650, "loss": 3.4062423706054688, "grad_norm": 0.7045295238494873, "learning_rate": 2.5240712232572618e-05, "elbo": 3.40624, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.667956837907576}
{"step": 96655, "loss": 3.4110061645507814, "grad_norm": 1.1516205072402954, "learning_rate": 2.5172508424976838e-05, "elbo": 3.41101, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.668250058644148}
{"step": 96660, "loss": 3.3974334716796877, "grad_norm": 0.6846015453338623, "learning_rate": 2.510436587674848e-05, "elbo": 3.39743, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.668543279380717}
{"step": 96665, "loss": 3.427418518066406, "grad_norm": 0.688734233379364, "learning_rate": 2.503628475602256e-05, "elbo": 3.42742, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.668836500117289}
{"step": 96670, "loss": 3.411387252807617, "grad_norm": 0.7235771417617798, "learning_rate": 2.4968265230782394e-05, "elbo": 3.41139, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.669129720853859}
{"step": 96675, "loss": 3.3962535858154297, "grad_norm": 0.6887260675430298, "learning_rate": 2.4900307468859457e-05, "elbo": 3.39625, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.66942294159043}
{"step": 96680, "loss": 3.419263458251953, "grad_norm": 1.199672818183899, "learning_rate": 2.4832411637932757e-05, "elbo": 3.41926, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.669716162327}
{"step": 96685, "loss": 3.406249237060547, "grad_norm": 0.7068833112716675, "learning_rate": 2.47645779055285e-05, "elbo": 3.40625, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6700093830635705}
{"step": 96690, "loss": 3.3906509399414064, "grad_norm": 0.6700121760368347, "learning_rate": 2.4696806439019664e-05, "elbo": 3.39065, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.670302603800141}
{"step": 96695, "loss": 3.412143325805664, "grad_norm": 0.6931427121162415, "learning_rate": 2.4629097405625647e-05, "elbo": 3.41214, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.670595824536711}
{"step": 96700, "loss": 3.3923221588134767, "grad_norm": 0.7018359899520874, "learning_rate": 2.456145097241173e-05, "elbo": 3.39232, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.670889045273282}
{"step": 96705, "loss": 3.4177734375, "grad_norm": 0.7190477252006531, "learning_rate": 2.449386730628877e-05, "elbo": 3.41777, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.671182266009852}
{"step": 96710, "loss": 3.4079689025878905, "grad_norm": 0.8188689947128296, "learning_rate": 2.442634657401272e-05, "elbo": 3.40797, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.671475486746423}
{"step": 96715, "loss": 3.4024986267089843, "grad_norm": 0.7115371823310852, "learning_rate": 2.435888894218432e-05, "elbo": 3.4025, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.671768707482993}
{"step": 96720, "loss": 3.4039119720458983, "grad_norm": 0.6995206475257874, "learning_rate": 2.4291494577248545e-05, "elbo": 3.40391, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.672061928219564}
{"step": 96725, "loss": 3.409765625, "grad_norm": 0.6562817692756653, "learning_rate": 2.422416364549429e-05, "elbo": 3.40977, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.672355148956134}
{"step": 96730, "loss": 3.3858741760253905, "grad_norm": 0.7067269086837769, "learning_rate": 2.415689631305393e-05, "elbo": 3.38587, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.672648369692705}
{"step": 96735, "loss": 3.4112522125244142, "grad_norm": 0.6753022074699402, "learning_rate": 2.408969274590296e-05, "elbo": 3.41125, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.672941590429275}
{"step": 96740, "loss": 3.4100357055664063, "grad_norm": 0.7983964085578918, "learning_rate": 2.4022553109859435e-05, "elbo": 3.41004, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6732348111658455}
{"step": 96745, "loss": 3.4120101928710938, "grad_norm": 0.7715064883232117, "learning_rate": 2.395547757058379e-05, "elbo": 3.41201, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.673528031902416}
{"step": 96750, "loss": 3.4133914947509765, "grad_norm": 0.6894474029541016, "learning_rate": 2.388846629357821e-05, "elbo": 3.41339, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.673821252638986}
{"step": 96755, "loss": 3.4173023223876955, "grad_norm": 0.7034969925880432, "learning_rate": 2.382151944418642e-05, "elbo": 3.4173, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.674114473375557}
{"step": 96760, "loss": 3.4111892700195314, "grad_norm": 0.6952256560325623, "learning_rate": 2.3754637187593053e-05, "elbo": 3.41119, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.674407694112127}
{"step": 96765, "loss": 3.393844223022461, "grad_norm": 0.823711097240448, "learning_rate": 2.368781968882343e-05, "elbo": 3.39384, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.674700914848698}
{"step": 96770, "loss": 3.4040348052978517, "grad_norm": 0.671708345413208, "learning_rate": 2.3621067112743134e-05, "elbo": 3.40403, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.674994135585268}
{"step": 96775, "loss": 3.402252960205078, "grad_norm": 0.6754640936851501, "learning_rate": 2.3554379624057482e-05, "elbo": 3.40225, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.675287356321839}
{"step": 96780, "loss": 3.406757354736328, "grad_norm": 0.6954602599143982, "learning_rate": 2.3487757387311227e-05, "elbo": 3.40676, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.67558057705841}
{"step": 96785, "loss": 3.407440948486328, "grad_norm": 0.7618499398231506, "learning_rate": 2.3421200566888095e-05, "elbo": 3.40744, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.67587379779498}
{"step": 96790, "loss": 3.4132476806640626, "grad_norm": 0.7546557188034058, "learning_rate": 2.335470932701046e-05, "elbo": 3.41325, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.676167018531551}
{"step": 96795, "loss": 3.416918182373047, "grad_norm": 0.7711251378059387, "learning_rate": 2.3288283831738834e-05, "elbo": 3.41692, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.676460239268121}
{"step": 96800, "loss": 3.4042396545410156, "grad_norm": 0.7322844862937927, "learning_rate": 2.322192424497152e-05, "elbo": 3.40424, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.676753460004692}
{"step": 96805, "loss": 3.4164337158203124, "grad_norm": 0.735867977142334, "learning_rate": 2.3155630730444183e-05, "elbo": 3.41643, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.677046680741262}
{"step": 96810, "loss": 3.4053993225097656, "grad_norm": 0.780759334564209, "learning_rate": 2.3089403451729536e-05, "elbo": 3.4054, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.677339901477833}
{"step": 96815, "loss": 3.4110977172851564, "grad_norm": 0.7076753377914429, "learning_rate": 2.3023242572236724e-05, "elbo": 3.4111, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.677633122214403}
{"step": 96820, "loss": 3.4149078369140624, "grad_norm": 0.7333910465240479, "learning_rate": 2.2957148255211215e-05, "elbo": 3.41491, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.677926342950974}
{"step": 96825, "loss": 3.4241912841796873, "grad_norm": 0.6916677355766296, "learning_rate": 2.2891120663734106e-05, "elbo": 3.42419, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.678219563687544}
{"step": 96830, "loss": 3.3860004425048826, "grad_norm": 0.7342161536216736, "learning_rate": 2.2825159960721988e-05, "elbo": 3.386, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.678512784424115}
{"step": 96835, "loss": 3.4192951202392576, "grad_norm": 0.6925182938575745, "learning_rate": 2.275926630892626e-05, "elbo": 3.4193, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.678806005160685}
{"step": 96840, "loss": 3.4055435180664064, "grad_norm": 0.7138652205467224, "learning_rate": 2.2693439870933007e-05, "elbo": 3.40554, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6790992258972555}
{"step": 96845, "loss": 3.40325927734375, "grad_norm": 0.8301591277122498, "learning_rate": 2.262768080916241e-05, "elbo": 3.40326, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.679392446633826}
{"step": 96850, "loss": 3.4203643798828125, "grad_norm": 0.6827607750892639, "learning_rate": 2.2561989285868417e-05, "elbo": 3.42036, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.679685667370396}
{"step": 96855, "loss": 3.4262939453125, "grad_norm": 0.7394952774047852, "learning_rate": 2.2496365463138308e-05, "elbo": 3.42629, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.679978888106967}
{"step": 96860, "loss": 3.4077224731445312, "grad_norm": 0.7813538908958435, "learning_rate": 2.2430809502892382e-05, "elbo": 3.40772, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.680272108843537}
{"step": 96865, "loss": 3.398027801513672, "grad_norm": 0.7057156562805176, "learning_rate": 2.2365321566883436e-05, "elbo": 3.39803, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.680565329580108}
{"step": 96870, "loss": 3.414356231689453, "grad_norm": 1.1085129976272583, "learning_rate": 2.2299901816696438e-05, "elbo": 3.41436, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.680858550316678}
{"step": 96875, "loss": 3.403017044067383, "grad_norm": 0.7134872078895569, "learning_rate": 2.2234550413748106e-05, "elbo": 3.40302, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.681151771053249}
{"step": 96880, "loss": 3.404291534423828, "grad_norm": 0.7258177399635315, "learning_rate": 2.2169267519286563e-05, "elbo": 3.40429, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.681444991789819}
{"step": 96885, "loss": 3.4110977172851564, "grad_norm": 0.6746819615364075, "learning_rate": 2.2104053294390846e-05, "elbo": 3.4111, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6817382125263896}
{"step": 96890, "loss": 3.4041954040527345, "grad_norm": 0.6640599370002747, "learning_rate": 2.2038907899970568e-05, "elbo": 3.4042, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.68203143326296}
{"step": 96895, "loss": 3.41326789855957, "grad_norm": 0.772817075252533, "learning_rate": 2.1973831496765502e-05, "elbo": 3.41327, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6823246539995305}
{"step": 96900, "loss": 3.4035614013671873, "grad_norm": 0.713890790939331, "learning_rate": 2.1908824245345257e-05, "elbo": 3.40356, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.682617874736101}
{"step": 96905, "loss": 3.4035430908203126, "grad_norm": 0.6842502355575562, "learning_rate": 2.1843886306108686e-05, "elbo": 3.40354, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.682911095472672}
{"step": 96910, "loss": 3.4198726654052733, "grad_norm": 0.6817853450775146, "learning_rate": 2.1779017839283754e-05, "elbo": 3.41987, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.683204316209242}
{"step": 96915, "loss": 3.408661651611328, "grad_norm": 0.9222955703735352, "learning_rate": 2.171421900492692e-05, "elbo": 3.40866, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.683497536945813}
{"step": 96920, "loss": 3.4129409790039062, "grad_norm": 0.725203275680542, "learning_rate": 2.1649489962922924e-05, "elbo": 3.41294, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.683790757682384}
{"step": 96925, "loss": 3.410420227050781, "grad_norm": 0.8103958368301392, "learning_rate": 2.158483087298417e-05, "elbo": 3.41042, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.684083978418954}
{"step": 96930, "loss": 3.407079315185547, "grad_norm": 0.6958894729614258, "learning_rate": 2.1520241894650585e-05, "elbo": 3.40708, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6843771991555245}
{"step": 96935, "loss": 3.412245178222656, "grad_norm": 0.6813204288482666, "learning_rate": 2.145572318728903e-05, "elbo": 3.41225, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.684670419892095}
{"step": 96940, "loss": 3.4000885009765627, "grad_norm": 0.7767226696014404, "learning_rate": 2.139127491009301e-05, "elbo": 3.40009, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.684963640628665}
{"step": 96945, "loss": 3.4114513397216797, "grad_norm": 0.7673693895339966, "learning_rate": 2.1326897222082226e-05, "elbo": 3.41145, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.685256861365236}
{"step": 96950, "loss": 3.4030189514160156, "grad_norm": 0.7153067588806152, "learning_rate": 2.126259028210225e-05, "elbo": 3.40302, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.685550082101806}
{"step": 96955, "loss": 3.3961101531982423, "grad_norm": 0.8899148106575012, "learning_rate": 2.1198354248824054e-05, "elbo": 3.39611, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.685843302838377}
{"step": 96960, "loss": 3.4175853729248047, "grad_norm": 0.6960605978965759, "learning_rate": 2.1134189280743665e-05, "elbo": 3.41759, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.686136523574947}
{"step": 96965, "loss": 3.417516326904297, "grad_norm": 0.690691351890564, "learning_rate": 2.107009553618174e-05, "elbo": 3.41752, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.686429744311518}
{"step": 96970, "loss": 3.4090003967285156, "grad_norm": 0.6849097609519958, "learning_rate": 2.1006073173283257e-05, "elbo": 3.409, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.686722965048088}
{"step": 96975, "loss": 3.407021331787109, "grad_norm": 0.6693049073219299, "learning_rate": 2.0942122350017023e-05, "elbo": 3.40702, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.687016185784659}
{"step": 96980, "loss": 3.4014442443847654, "grad_norm": 0.688686192035675, "learning_rate": 2.0878243224175327e-05, "elbo": 3.40144, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.687309406521229}
{"step": 96985, "loss": 3.408283996582031, "grad_norm": 0.6886667609214783, "learning_rate": 2.081443595337355e-05, "elbo": 3.40828, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6876026272577995}
{"step": 96990, "loss": 3.404405212402344, "grad_norm": 0.8526698350906372, "learning_rate": 2.0750700695049847e-05, "elbo": 3.40441, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.68789584799437}
{"step": 96995, "loss": 3.413372802734375, "grad_norm": 0.8954156041145325, "learning_rate": 2.0687037606464553e-05, "elbo": 3.41337, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.68818906873094}
{"step": 97000, "loss": 3.402635192871094, "grad_norm": 0.8332164287567139, "learning_rate": 2.062344684470007e-05, "elbo": 3.40264, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.688482289467511}
{"step": 97005, "loss": 3.391118621826172, "grad_norm": 0.6912042498588562, "learning_rate": 2.0559928566660237e-05, "elbo": 3.39112, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.688775510204081}
{"step": 97010, "loss": 3.398394012451172, "grad_norm": 0.6803104281425476, "learning_rate": 2.0496482929070155e-05, "elbo": 3.39839, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.689068730940652}
{"step": 97015, "loss": 3.4192291259765626, "grad_norm": 0.6597431898117065, "learning_rate": 2.043311008847555e-05, "elbo": 3.41923, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.689361951677222}
{"step": 97020, "loss": 3.425176239013672, "grad_norm": 0.6831132173538208, "learning_rate": 2.036981020124266e-05, "elbo": 3.42518, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.689655172413794}
{"step": 97025, "loss": 3.400548553466797, "grad_norm": 0.8412230014801025, "learning_rate": 2.0306583423557653e-05, "elbo": 3.40055, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.689948393150363}
{"step": 97030, "loss": 3.4143470764160155, "grad_norm": 0.7455009818077087, "learning_rate": 2.024342991142631e-05, "elbo": 3.41435, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6902416138869345}
{"step": 97035, "loss": 3.3958366394042967, "grad_norm": 1.0038925409317017, "learning_rate": 2.018034982067363e-05, "elbo": 3.39584, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.690534834623505}
{"step": 97040, "loss": 3.4010414123535155, "grad_norm": 1.2235666513442993, "learning_rate": 2.01173433069435e-05, "elbo": 3.40104, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.690828055360075}
{"step": 97045, "loss": 3.4018836975097657, "grad_norm": 0.7116731405258179, "learning_rate": 2.0054410525698215e-05, "elbo": 3.40188, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.691121276096646}
{"step": 97050, "loss": 3.402840805053711, "grad_norm": 0.7062166929244995, "learning_rate": 1.9991551632218154e-05, "elbo": 3.40284, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.691414496833216}
{"step": 97055, "loss": 3.4115100860595704, "grad_norm": 0.6697235107421875, "learning_rate": 1.9928766781601365e-05, "elbo": 3.41151, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.691707717569787}
{"step": 97060, "loss": 3.3792552947998047, "grad_norm": 0.7053769826889038, "learning_rate": 1.9866056128763278e-05, "elbo": 3.37926, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.692000938306357}
{"step": 97065, "loss": 3.4154388427734377, "grad_norm": 0.6948304176330566, "learning_rate": 1.980341982843616e-05, "elbo": 3.41544, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.692294159042928}
{"step": 97070, "loss": 3.398701477050781, "grad_norm": 1.1289150714874268, "learning_rate": 1.9740858035168857e-05, "elbo": 3.3987, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.692587379779498}
{"step": 97075, "loss": 3.4069587707519533, "grad_norm": 1.1054627895355225, "learning_rate": 1.967837090332637e-05, "elbo": 3.40696, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.692880600516069}
{"step": 97080, "loss": 3.4020599365234374, "grad_norm": 0.7043803930282593, "learning_rate": 1.961595858708953e-05, "elbo": 3.40206, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.693173821252639}
{"step": 97080, "loss": 3.4020599365234374, "grad_norm": 0.7043803930282593, "learning_rate": 1.961595858708953e-05, "elbo": 3.40206, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.693173821252639}
{"step": 97085, "loss": 3.3987884521484375, "grad_norm": 0.6863434910774231, "learning_rate": 1.9553621240454452e-05, "elbo": 3.39879, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6934670419892095}
{"step": 97090, "loss": 3.3992069244384764, "grad_norm": 0.7487267851829529, "learning_rate": 1.949135901723239e-05, "elbo": 3.39921, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.69376026272578}
{"step": 97095, "loss": 3.401780700683594, "grad_norm": 0.7945627570152283, "learning_rate": 1.9429172071049168e-05, "elbo": 3.40178, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.69405348346235}
{"step": 97100, "loss": 3.4019065856933595, "grad_norm": 0.6749513149261475, "learning_rate": 1.9367060555344934e-05, "elbo": 3.40191, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.694346704198921}
{"step": 97105, "loss": 3.3937004089355467, "grad_norm": 0.6715604066848755, "learning_rate": 1.9305024623373618e-05, "elbo": 3.3937, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.694639924935491}
{"step": 97110, "loss": 3.407051849365234, "grad_norm": 0.7266059517860413, "learning_rate": 1.924306442820276e-05, "elbo": 3.40705, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.694933145672062}
{"step": 97115, "loss": 3.39559326171875, "grad_norm": 0.6791653633117676, "learning_rate": 1.918118012271297e-05, "elbo": 3.39559, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.695226366408632}
{"step": 97120, "loss": 3.400054168701172, "grad_norm": 0.7236032485961914, "learning_rate": 1.9119371859597623e-05, "elbo": 3.40005, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.695519587145203}
{"step": 97125, "loss": 3.411093902587891, "grad_norm": 0.6854997873306274, "learning_rate": 1.9057639791362437e-05, "elbo": 3.41109, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.695812807881773}
{"step": 97130, "loss": 3.3965213775634764, "grad_norm": 0.7571251392364502, "learning_rate": 1.8995984070325197e-05, "elbo": 3.39652, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.696106028618344}
{"step": 97135, "loss": 3.410321044921875, "grad_norm": 0.748839795589447, "learning_rate": 1.8934404848615246e-05, "elbo": 3.41032, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.696399249354914}
{"step": 97135, "loss": 3.410321044921875, "grad_norm": 0.748839795589447, "learning_rate": 1.8934404848615246e-05, "elbo": 3.41032, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.696399249354914}
{"step": 97140, "loss": 3.4107925415039064, "grad_norm": 0.7177708745002747, "learning_rate": 1.8872902278173188e-05, "elbo": 3.41079, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6966924700914845}
{"step": 97145, "loss": 3.397473907470703, "grad_norm": 0.7379639148712158, "learning_rate": 1.8811476510750488e-05, "elbo": 3.39747, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.696985690828056}
{"step": 97145, "loss": 3.397473907470703, "grad_norm": 0.7379639148712158, "learning_rate": 1.8811476510750488e-05, "elbo": 3.39747, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.696985690828056}
{"step": 97150, "loss": 3.4072898864746093, "grad_norm": 0.7127470374107361, "learning_rate": 1.8750127697909154e-05, "elbo": 3.40729, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.697278911564625}
{"step": 97155, "loss": 3.3906150817871095, "grad_norm": 0.8467569947242737, "learning_rate": 1.868885599102127e-05, "elbo": 3.39062, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.697572132301197}
{"step": 97155, "loss": 3.3906150817871095, "grad_norm": 0.8467569947242737, "learning_rate": 1.868885599102127e-05, "elbo": 3.39062, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.697572132301197}
{"step": 97160, "loss": 3.413905715942383, "grad_norm": 0.7066988348960876, "learning_rate": 1.862766154126868e-05, "elbo": 3.41391, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.697865353037767}
{"step": 97165, "loss": 3.4114654541015623, "grad_norm": 0.7416334748268127, "learning_rate": 1.8566544499642587e-05, "elbo": 3.41147, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.698158573774338}
{"step": 97170, "loss": 3.423004913330078, "grad_norm": 0.7280300259590149, "learning_rate": 1.8505505016943252e-05, "elbo": 3.42301, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.698451794510908}
{"step": 97175, "loss": 3.409021759033203, "grad_norm": 0.6876212954521179, "learning_rate": 1.844454324377951e-05, "elbo": 3.40902, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.698745015247479}
{"step": 97180, "loss": 3.39864501953125, "grad_norm": 0.7210736870765686, "learning_rate": 1.838365933056848e-05, "elbo": 3.39865, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.699038235984049}
{"step": 97185, "loss": 3.4126056671142577, "grad_norm": 0.6977251768112183, "learning_rate": 1.832285342753515e-05, "elbo": 3.41261, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.6993314567206195}
{"step": 97190, "loss": 3.399962615966797, "grad_norm": 0.6752517223358154, "learning_rate": 1.8262125684712084e-05, "elbo": 3.39996, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.69962467745719}
{"step": 97195, "loss": 3.4157684326171873, "grad_norm": 0.6656085848808289, "learning_rate": 1.8201476251938887e-05, "elbo": 3.41577, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.69991789819376}
{"step": 97200, "loss": 3.4113700866699217, "grad_norm": 0.8663877844810486, "learning_rate": 1.814090527886206e-05, "elbo": 3.41137, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.700211118930331}
{"step": 97205, "loss": 3.3914409637451173, "grad_norm": 0.6908027529716492, "learning_rate": 1.8080412914934436e-05, "elbo": 3.39144, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.700504339666901}
{"step": 97210, "loss": 3.407299041748047, "grad_norm": 0.7297345995903015, "learning_rate": 1.8019999309414916e-05, "elbo": 3.4073, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.700797560403472}
{"step": 97215, "loss": 3.40433349609375, "grad_norm": 0.6966882944107056, "learning_rate": 1.7959664611368044e-05, "elbo": 3.40433, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.701090781140042}
{"step": 97220, "loss": 3.406428909301758, "grad_norm": 0.7281152606010437, "learning_rate": 1.7899408969663723e-05, "elbo": 3.40643, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.701384001876613}
{"step": 97225, "loss": 3.39892692565918, "grad_norm": 0.7076948881149292, "learning_rate": 1.7839232532976747e-05, "elbo": 3.39893, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.701677222613183}
{"step": 97230, "loss": 3.388259506225586, "grad_norm": 0.7552923560142517, "learning_rate": 1.7779135449786482e-05, "elbo": 3.38826, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.701970443349754}
{"step": 97235, "loss": 3.420933151245117, "grad_norm": 0.7035332322120667, "learning_rate": 1.77191178683765e-05, "elbo": 3.42093, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.702263664086324}
{"step": 97235, "loss": 3.420933151245117, "grad_norm": 0.7035332322120667, "learning_rate": 1.77191178683765e-05, "elbo": 3.42093, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.702263664086324}
{"step": 97240, "loss": 3.3891014099121093, "grad_norm": 0.680842399597168, "learning_rate": 1.7659179936834246e-05, "elbo": 3.3891, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7025568848228945}
{"step": 97245, "loss": 3.418073272705078, "grad_norm": 0.6903101801872253, "learning_rate": 1.7599321803050596e-05, "elbo": 3.41807, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.702850105559465}
{"step": 97250, "loss": 3.4173229217529295, "grad_norm": 0.660317599773407, "learning_rate": 1.753954361471954e-05, "elbo": 3.41732, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.703143326296035}
{"step": 97250, "loss": 3.4173229217529295, "grad_norm": 0.660317599773407, "learning_rate": 1.753954361471954e-05, "elbo": 3.41732, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.703143326296035}
{"step": 97255, "loss": 3.390188217163086, "grad_norm": 0.7234156727790833, "learning_rate": 1.7479845519337795e-05, "elbo": 3.39019, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.703436547032606}
{"step": 97260, "loss": 3.4019641876220703, "grad_norm": 0.6795283555984497, "learning_rate": 1.7420227664204514e-05, "elbo": 3.40196, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.703729767769176}
{"step": 97265, "loss": 3.3906517028808594, "grad_norm": 0.6823729276657104, "learning_rate": 1.7360690196420815e-05, "elbo": 3.39065, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.704022988505747}
{"step": 97270, "loss": 3.3902122497558596, "grad_norm": 0.6821334958076477, "learning_rate": 1.7301233262889475e-05, "elbo": 3.39021, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.704316209242318}
{"step": 97275, "loss": 3.392675018310547, "grad_norm": 0.7044277787208557, "learning_rate": 1.7241857010314555e-05, "elbo": 3.39267, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.704609429978888}
{"step": 97275, "loss": 3.392675018310547, "grad_norm": 0.7044277787208557, "learning_rate": 1.7241857010314555e-05, "elbo": 3.39267, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.704609429978888}
{"step": 97280, "loss": 3.405498504638672, "grad_norm": 0.6971794366836548, "learning_rate": 1.7182561585201108e-05, "elbo": 3.4055, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.704902650715459}
{"step": 97285, "loss": 3.4084197998046877, "grad_norm": 0.7639333605766296, "learning_rate": 1.712334713385463e-05, "elbo": 3.40842, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.705195871452029}
{"step": 97290, "loss": 3.3990570068359376, "grad_norm": 0.6621721386909485, "learning_rate": 1.7064213802380942e-05, "elbo": 3.39906, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7054890921886}
{"step": 97295, "loss": 3.386919403076172, "grad_norm": 0.6883593201637268, "learning_rate": 1.700516173668565e-05, "elbo": 3.38692, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.70578231292517}
{"step": 97300, "loss": 3.3899456024169923, "grad_norm": 0.671068549156189, "learning_rate": 1.6946191082473846e-05, "elbo": 3.38995, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.706075533661741}
{"step": 97305, "loss": 3.4047920227050783, "grad_norm": 0.6766407489776611, "learning_rate": 1.6887301985249754e-05, "elbo": 3.40479, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.706368754398311}
{"step": 97310, "loss": 3.4038124084472656, "grad_norm": 0.6759745478630066, "learning_rate": 1.682849459031639e-05, "elbo": 3.40381, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.706661975134882}
{"step": 97315, "loss": 3.3937843322753904, "grad_norm": 0.7147045135498047, "learning_rate": 1.6769769042775142e-05, "elbo": 3.39378, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.706955195871452}
{"step": 97320, "loss": 3.4064361572265627, "grad_norm": 0.722418487071991, "learning_rate": 1.6711125487525464e-05, "elbo": 3.40644, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.707248416608023}
{"step": 97325, "loss": 3.386414337158203, "grad_norm": 1.132187008857727, "learning_rate": 1.6652564069264475e-05, "elbo": 3.38641, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.707541637344593}
{"step": 97330, "loss": 3.4139999389648437, "grad_norm": 1.0001349449157715, "learning_rate": 1.6594084932486697e-05, "elbo": 3.414, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7078348580811635}
{"step": 97335, "loss": 3.4179744720458984, "grad_norm": 0.6815939545631409, "learning_rate": 1.653568822148356e-05, "elbo": 3.41797, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.708128078817734}
{"step": 97340, "loss": 3.3993507385253907, "grad_norm": 0.6997297406196594, "learning_rate": 1.647737408034315e-05, "elbo": 3.39935, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.708421299554304}
{"step": 97345, "loss": 3.410186004638672, "grad_norm": 0.7482622861862183, "learning_rate": 1.6419142652949793e-05, "elbo": 3.41019, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.708714520290875}
{"step": 97350, "loss": 3.4092739105224608, "grad_norm": 0.6926854252815247, "learning_rate": 1.6360994082983775e-05, "elbo": 3.40927, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.709007741027445}
{"step": 97355, "loss": 3.390130615234375, "grad_norm": 0.6541157364845276, "learning_rate": 1.6302928513920913e-05, "elbo": 3.39013, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.709300961764016}
{"step": 97360, "loss": 3.386919403076172, "grad_norm": 0.6678148508071899, "learning_rate": 1.6244946089032216e-05, "elbo": 3.38692, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.709594182500586}
{"step": 97365, "loss": 3.388660430908203, "grad_norm": 0.8549166917800903, "learning_rate": 1.618704695138353e-05, "elbo": 3.38866, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.709887403237157}
{"step": 97370, "loss": 3.4, "grad_norm": 0.6841784715652466, "learning_rate": 1.612923124383528e-05, "elbo": 3.4, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.710180623973727}
{"step": 97375, "loss": 3.398203659057617, "grad_norm": 0.721186101436615, "learning_rate": 1.607149910904191e-05, "elbo": 3.3982, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.710473844710298}
{"step": 97380, "loss": 3.4151756286621096, "grad_norm": 0.7603034973144531, "learning_rate": 1.601385068945177e-05, "elbo": 3.41518, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.710767065446868}
{"step": 97385, "loss": 3.4127479553222657, "grad_norm": 0.7010371685028076, "learning_rate": 1.5956286127306592e-05, "elbo": 3.41275, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7110602861834385}
{"step": 97390, "loss": 3.394967269897461, "grad_norm": 0.6854403018951416, "learning_rate": 1.589880556464121e-05, "elbo": 3.39497, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.711353506920009}
{"step": 97395, "loss": 3.3884819030761717, "grad_norm": 0.663730263710022, "learning_rate": 1.5841409143283197e-05, "elbo": 3.38848, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.71164672765658}
{"step": 97400, "loss": 3.4083206176757814, "grad_norm": 0.7019749879837036, "learning_rate": 1.5784097004852505e-05, "elbo": 3.40832, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.71193994839315}
{"step": 97405, "loss": 3.3972667694091796, "grad_norm": 0.7243580222129822, "learning_rate": 1.5726869290761158e-05, "elbo": 3.39727, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.712233169129721}
{"step": 97410, "loss": 3.388536071777344, "grad_norm": 0.6924495697021484, "learning_rate": 1.5669726142212855e-05, "elbo": 3.38854, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.712526389866292}
{"step": 97415, "loss": 3.4072120666503904, "grad_norm": 0.706766664981842, "learning_rate": 1.5612667700202615e-05, "elbo": 3.40721, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.712819610602862}
{"step": 97420, "loss": 3.401668930053711, "grad_norm": 0.7144964933395386, "learning_rate": 1.555569410551647e-05, "elbo": 3.40167, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.713112831339433}
{"step": 97425, "loss": 3.410507583618164, "grad_norm": 0.6764493584632874, "learning_rate": 1.5498805498731144e-05, "elbo": 3.41051, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.713406052076003}
{"step": 97430, "loss": 3.4080322265625, "grad_norm": 0.6957054734230042, "learning_rate": 1.5442002020213558e-05, "elbo": 3.40803, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7136992728125735}
{"step": 97435, "loss": 3.3992523193359374, "grad_norm": 0.731844961643219, "learning_rate": 1.5385283810120688e-05, "elbo": 3.39925, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.713992493549144}
{"step": 97440, "loss": 3.3910945892333983, "grad_norm": 0.7146914601325989, "learning_rate": 1.5328651008399057e-05, "elbo": 3.39109, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.714285714285714}
{"step": 97445, "loss": 3.406560516357422, "grad_norm": 0.6900807619094849, "learning_rate": 1.5272103754784517e-05, "elbo": 3.40656, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.714578935022285}
{"step": 97450, "loss": 3.3888450622558595, "grad_norm": 0.719938337802887, "learning_rate": 1.5215642188801732e-05, "elbo": 3.38885, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.714872155758855}
{"step": 97455, "loss": 3.372599792480469, "grad_norm": 0.6640546917915344, "learning_rate": 1.5159266449764049e-05, "elbo": 3.3726, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.715165376495426}
{"step": 97460, "loss": 3.4063549041748047, "grad_norm": 0.6667646169662476, "learning_rate": 1.510297667677299e-05, "elbo": 3.40635, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.715458597231996}
{"step": 97465, "loss": 3.3938121795654297, "grad_norm": 0.763177752494812, "learning_rate": 1.5046773008717969e-05, "elbo": 3.39381, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.715751817968567}
{"step": 97465, "loss": 3.3938121795654297, "grad_norm": 0.763177752494812, "learning_rate": 1.5046773008717969e-05, "elbo": 3.39381, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.715751817968567}
{"step": 97470, "loss": 3.3961212158203127, "grad_norm": 0.9034613370895386, "learning_rate": 1.4990655584275931e-05, "elbo": 3.39612, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.716045038705137}
{"step": 97475, "loss": 3.393818664550781, "grad_norm": 0.689376175403595, "learning_rate": 1.4934624541911085e-05, "elbo": 3.39382, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.716338259441708}
{"step": 97475, "loss": 3.393818664550781, "grad_norm": 0.689376175403595, "learning_rate": 1.4934624541911085e-05, "elbo": 3.39382, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.716338259441708}
{"step": 97480, "loss": 3.411208724975586, "grad_norm": 0.7008758783340454, "learning_rate": 1.4878680019874431e-05, "elbo": 3.41121, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.716631480178278}
{"step": 97485, "loss": 3.396615219116211, "grad_norm": 0.6816025972366333, "learning_rate": 1.482282215620352e-05, "elbo": 3.39662, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7169247009148485}
{"step": 97490, "loss": 3.3902610778808593, "grad_norm": 0.6897208094596863, "learning_rate": 1.4767051088722067e-05, "elbo": 3.39026, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.717217921651419}
{"step": 97495, "loss": 3.4160087585449217, "grad_norm": 0.6702716946601868, "learning_rate": 1.4711366955039663e-05, "elbo": 3.41601, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.717511142387989}
{"step": 97500, "loss": 3.388393020629883, "grad_norm": 0.7060099840164185, "learning_rate": 1.4655769892551357e-05, "elbo": 3.38839, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.71780436312456}
{"step": 97505, "loss": 3.3965282440185547, "grad_norm": 0.7167207598686218, "learning_rate": 1.4600260038437375e-05, "elbo": 3.39653, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.71809758386113}
{"step": 97510, "loss": 3.3976318359375, "grad_norm": 0.7046646475791931, "learning_rate": 1.4544837529662748e-05, "elbo": 3.39763, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.718390804597701}
{"step": 97515, "loss": 3.4040782928466795, "grad_norm": 0.6955628395080566, "learning_rate": 1.4489502502977037e-05, "elbo": 3.40408, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.718684025334271}
{"step": 97520, "loss": 3.399224853515625, "grad_norm": 0.6795174479484558, "learning_rate": 1.4434255094913906e-05, "elbo": 3.39923, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.718977246070843}
{"step": 97525, "loss": 3.381357192993164, "grad_norm": 0.9472881555557251, "learning_rate": 1.4379095441790846e-05, "elbo": 3.38136, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.719270466807412}
{"step": 97530, "loss": 3.3873939514160156, "grad_norm": 0.7738549113273621, "learning_rate": 1.4324023679708792e-05, "elbo": 3.38739, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7195636875439835}
{"step": 97535, "loss": 3.393513488769531, "grad_norm": 0.7249885201454163, "learning_rate": 1.4269039944551898e-05, "elbo": 3.39351, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.719856908280554}
{"step": 97540, "loss": 3.3936920166015625, "grad_norm": 0.6945469975471497, "learning_rate": 1.4214144371987004e-05, "elbo": 3.39369, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.720150129017124}
{"step": 97545, "loss": 3.389543914794922, "grad_norm": 0.6667041778564453, "learning_rate": 1.4159337097463515e-05, "elbo": 3.38954, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.720443349753695}
{"step": 97550, "loss": 3.4034774780273436, "grad_norm": 0.7257225513458252, "learning_rate": 1.4104618256212926e-05, "elbo": 3.40348, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.720736570490265}
{"step": 97555, "loss": 3.4002670288085937, "grad_norm": 1.1154588460922241, "learning_rate": 1.404998798324853e-05, "elbo": 3.40027, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.721029791226836}
{"step": 97555, "loss": 3.4002670288085937, "grad_norm": 1.1154588460922241, "learning_rate": 1.404998798324853e-05, "elbo": 3.40027, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.721029791226836}
{"step": 97560, "loss": 3.4012981414794923, "grad_norm": 0.650517463684082, "learning_rate": 1.3995446413365093e-05, "elbo": 3.4013, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.721323011963406}
{"step": 97565, "loss": 3.4031944274902344, "grad_norm": 0.7004033923149109, "learning_rate": 1.394099368113853e-05, "elbo": 3.40319, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.721616232699977}
{"step": 97570, "loss": 3.3887893676757814, "grad_norm": 0.6861176490783691, "learning_rate": 1.3886629920925548e-05, "elbo": 3.38879, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.721909453436547}
{"step": 97575, "loss": 3.3899490356445314, "grad_norm": 0.7092711925506592, "learning_rate": 1.3832355266863305e-05, "elbo": 3.38995, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.722202674173118}
{"step": 97580, "loss": 3.395532989501953, "grad_norm": 0.7325199246406555, "learning_rate": 1.377816985286911e-05, "elbo": 3.39553, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.722495894909688}
{"step": 97585, "loss": 3.4101325988769533, "grad_norm": 0.659078061580658, "learning_rate": 1.3724073812640109e-05, "elbo": 3.41013, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7227891156462585}
{"step": 97590, "loss": 3.3975318908691405, "grad_norm": 0.6564510464668274, "learning_rate": 1.3670067279652899e-05, "elbo": 3.39753, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.723082336382829}
{"step": 97595, "loss": 3.3998722076416015, "grad_norm": 0.7680980563163757, "learning_rate": 1.3616150387163217e-05, "elbo": 3.39987, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.723375557119399}
{"step": 97600, "loss": 3.394285202026367, "grad_norm": 1.0234497785568237, "learning_rate": 1.3562323268205629e-05, "elbo": 3.39429, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.72366877785597}
{"step": 97605, "loss": 3.390406036376953, "grad_norm": 0.8004351854324341, "learning_rate": 1.350858605559323e-05, "elbo": 3.39041, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.72396199859254}
{"step": 97610, "loss": 3.414141082763672, "grad_norm": 0.6855174899101257, "learning_rate": 1.345493888191724e-05, "elbo": 3.41414, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.724255219329111}
{"step": 97615, "loss": 3.3859848022460937, "grad_norm": 0.8126305937767029, "learning_rate": 1.3401381879546715e-05, "elbo": 3.38598, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.724548440065681}
{"step": 97620, "loss": 3.395849609375, "grad_norm": 0.7494570016860962, "learning_rate": 1.3347915180628228e-05, "elbo": 3.39585, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.724841660802252}
{"step": 97625, "loss": 3.3946815490722657, "grad_norm": 0.7390186190605164, "learning_rate": 1.3294538917085586e-05, "elbo": 3.39468, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.725134881538822}
{"step": 97630, "loss": 3.400464630126953, "grad_norm": 0.7071839570999146, "learning_rate": 1.3241253220619355e-05, "elbo": 3.40046, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.725428102275393}
{"step": 97635, "loss": 3.3989490509033202, "grad_norm": 0.7001235485076904, "learning_rate": 1.3188058222706734e-05, "elbo": 3.39895, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.725721323011964}
{"step": 97640, "loss": 3.390673828125, "grad_norm": 0.6809281706809998, "learning_rate": 1.3134954054601083e-05, "elbo": 3.39067, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7260145437485335}
{"step": 97645, "loss": 3.3859573364257813, "grad_norm": 0.6897633671760559, "learning_rate": 1.3081940847331659e-05, "elbo": 3.38596, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.726307764485105}
{"step": 97650, "loss": 3.4051929473876954, "grad_norm": 0.6797217130661011, "learning_rate": 1.3029018731703263e-05, "elbo": 3.40519, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.726600985221674}
{"step": 97655, "loss": 3.4144561767578123, "grad_norm": 0.675597608089447, "learning_rate": 1.2976187838295983e-05, "elbo": 3.41446, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.726894205958246}
{"step": 97660, "loss": 3.4011024475097655, "grad_norm": 0.7387408018112183, "learning_rate": 1.2923448297464785e-05, "elbo": 3.4011, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.727187426694816}
{"step": 97665, "loss": 3.406755065917969, "grad_norm": 0.7300507426261902, "learning_rate": 1.2870800239339236e-05, "elbo": 3.40675, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.727480647431387}
{"step": 97670, "loss": 3.4100547790527345, "grad_norm": 0.985194206237793, "learning_rate": 1.2818243793823188e-05, "elbo": 3.41005, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.727773868167957}
{"step": 97675, "loss": 3.3956039428710936, "grad_norm": 0.6997852325439453, "learning_rate": 1.2765779090594454e-05, "elbo": 3.3956, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7280670889045275}
{"step": 97680, "loss": 3.39718017578125, "grad_norm": 0.6525474786758423, "learning_rate": 1.271340625910447e-05, "elbo": 3.39718, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.728360309641098}
{"step": 97685, "loss": 3.398092269897461, "grad_norm": 1.0424323081970215, "learning_rate": 1.2661125428577997e-05, "elbo": 3.39809, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7286535303776684}
{"step": 97690, "loss": 3.38487548828125, "grad_norm": 1.0793206691741943, "learning_rate": 1.260893672801276e-05, "elbo": 3.38488, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.728946751114239}
{"step": 97695, "loss": 3.4015125274658202, "grad_norm": 0.7233640551567078, "learning_rate": 1.2556840286179233e-05, "elbo": 3.40151, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.729239971850809}
{"step": 97700, "loss": 3.3955638885498045, "grad_norm": 0.6776862740516663, "learning_rate": 1.2504836231620176e-05, "elbo": 3.39556, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.72953319258738}
{"step": 97705, "loss": 3.401255416870117, "grad_norm": 0.7045212388038635, "learning_rate": 1.2452924692650442e-05, "elbo": 3.40126, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.72982641332395}
{"step": 97710, "loss": 3.3946197509765623, "grad_norm": 0.6723179221153259, "learning_rate": 1.2401105797356566e-05, "elbo": 3.39462, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.730119634060521}
{"step": 97715, "loss": 3.3823184967041016, "grad_norm": 0.6837292909622192, "learning_rate": 1.2349379673596568e-05, "elbo": 3.38232, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.730412854797091}
{"step": 97720, "loss": 3.4048686981201173, "grad_norm": 0.7135854363441467, "learning_rate": 1.2297746448999453e-05, "elbo": 3.40487, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.730706075533662}
{"step": 97725, "loss": 3.400156021118164, "grad_norm": 0.7178691625595093, "learning_rate": 1.2246206250965125e-05, "elbo": 3.40016, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.730999296270232}
{"step": 97730, "loss": 3.3898239135742188, "grad_norm": 0.6684398651123047, "learning_rate": 1.2194759206663881e-05, "elbo": 3.38982, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7312925170068025}
{"step": 97735, "loss": 3.385370635986328, "grad_norm": 0.649948239326477, "learning_rate": 1.2143405443036182e-05, "elbo": 3.38537, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.731585737743373}
{"step": 97740, "loss": 3.3967880249023437, "grad_norm": 0.7114386558532715, "learning_rate": 1.2092145086792327e-05, "elbo": 3.39679, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.731878958479943}
{"step": 97745, "loss": 3.3844058990478514, "grad_norm": 0.6765696406364441, "learning_rate": 1.2040978264412178e-05, "elbo": 3.38441, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.732172179216514}
{"step": 97750, "loss": 3.3809059143066404, "grad_norm": 0.7119889259338379, "learning_rate": 1.198990510214476e-05, "elbo": 3.38091, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.732465399953084}
{"step": 97755, "loss": 3.3974983215332033, "grad_norm": 0.6828325390815735, "learning_rate": 1.1938925726008038e-05, "elbo": 3.3975, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.732758620689655}
{"step": 97760, "loss": 3.412394714355469, "grad_norm": 0.6933469772338867, "learning_rate": 1.188804026178853e-05, "elbo": 3.41239, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.733051841426226}
{"step": 97765, "loss": 3.4000385284423826, "grad_norm": 0.8578618168830872, "learning_rate": 1.1837248835041092e-05, "elbo": 3.40004, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.733345062162796}
{"step": 97770, "loss": 3.3955238342285154, "grad_norm": 0.725939154624939, "learning_rate": 1.1786551571088506e-05, "elbo": 3.39552, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.733638282899367}
{"step": 97775, "loss": 3.401858901977539, "grad_norm": 0.6965915560722351, "learning_rate": 1.1735948595021235e-05, "elbo": 3.40186, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7339315036359375}
{"step": 97780, "loss": 3.3859935760498048, "grad_norm": 0.8186975717544556, "learning_rate": 1.168544003169707e-05, "elbo": 3.38599, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.734224724372508}
{"step": 97785, "loss": 3.410242462158203, "grad_norm": 0.71891188621521, "learning_rate": 1.16350260057409e-05, "elbo": 3.41024, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.734517945109078}
{"step": 97790, "loss": 3.4175891876220703, "grad_norm": 0.6839141249656677, "learning_rate": 1.1584706641544319e-05, "elbo": 3.41759, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.734811165845649}
{"step": 97795, "loss": 3.38560791015625, "grad_norm": 0.7438153028488159, "learning_rate": 1.1534482063265345e-05, "elbo": 3.38561, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.735104386582219}
{"step": 97800, "loss": 3.397079849243164, "grad_norm": 0.8345142006874084, "learning_rate": 1.1484352394828118e-05, "elbo": 3.39708, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.73539760731879}
{"step": 97805, "loss": 3.40460319519043, "grad_norm": 0.6952674984931946, "learning_rate": 1.1434317759922664e-05, "elbo": 3.4046, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.73569082805536}
{"step": 97810, "loss": 3.404135894775391, "grad_norm": 0.7124033570289612, "learning_rate": 1.1384378282004404e-05, "elbo": 3.40414, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.735984048791931}
{"step": 97815, "loss": 3.3971370697021483, "grad_norm": 0.7567009925842285, "learning_rate": 1.1334534084294086e-05, "elbo": 3.39714, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.736277269528501}
{"step": 97820, "loss": 3.3900405883789064, "grad_norm": 0.7784056663513184, "learning_rate": 1.1284785289777272e-05, "elbo": 3.39004, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.736570490265072}
{"step": 97825, "loss": 3.377248001098633, "grad_norm": 1.296251654624939, "learning_rate": 1.1235132021204226e-05, "elbo": 3.37725, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.736863711001642}
{"step": 97830, "loss": 3.384815979003906, "grad_norm": 0.6777108311653137, "learning_rate": 1.1185574401089383e-05, "elbo": 3.38482, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7371569317382125}
{"step": 97835, "loss": 3.3914955139160154, "grad_norm": 0.6696159839630127, "learning_rate": 1.1136112551711291e-05, "elbo": 3.3915, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.737450152474783}
{"step": 97840, "loss": 3.4080718994140624, "grad_norm": 0.6859172582626343, "learning_rate": 1.1086746595112136e-05, "elbo": 3.40807, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.737743373211353}
{"step": 97845, "loss": 3.3760868072509767, "grad_norm": 1.0235284566879272, "learning_rate": 1.10374766530975e-05, "elbo": 3.37609, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.738036593947924}
{"step": 97850, "loss": 3.392816162109375, "grad_norm": 0.8176283240318298, "learning_rate": 1.0988302847236066e-05, "elbo": 3.39282, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.738329814684494}
{"step": 97855, "loss": 3.3959068298339843, "grad_norm": 0.6760163903236389, "learning_rate": 1.0939225298859323e-05, "elbo": 3.39591, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.738623035421065}
{"step": 97860, "loss": 3.399591827392578, "grad_norm": 0.760288417339325, "learning_rate": 1.0890244129061245e-05, "elbo": 3.39959, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.738916256157635}
{"step": 97865, "loss": 3.3986961364746096, "grad_norm": 0.6948357820510864, "learning_rate": 1.0841359458697986e-05, "elbo": 3.3987, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.739209476894206}
{"step": 97870, "loss": 3.383380126953125, "grad_norm": 0.6896686553955078, "learning_rate": 1.0792571408387608e-05, "elbo": 3.38338, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.739502697630776}
{"step": 97875, "loss": 3.404673767089844, "grad_norm": 0.6921356916427612, "learning_rate": 1.07438800985098e-05, "elbo": 3.40467, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.739795918367347}
{"step": 97875, "loss": 3.404673767089844, "grad_norm": 0.6921356916427612, "learning_rate": 1.07438800985098e-05, "elbo": 3.40467, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.739795918367347}
{"step": 97880, "loss": 3.390245056152344, "grad_norm": 0.7333121299743652, "learning_rate": 1.069528564920551e-05, "elbo": 3.39025, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.740089139103917}
{"step": 97885, "loss": 3.3957469940185545, "grad_norm": 0.7155827879905701, "learning_rate": 1.0646788180376716e-05, "elbo": 3.39575, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.740382359840488}
{"step": 97890, "loss": 3.391124725341797, "grad_norm": 0.6903631091117859, "learning_rate": 1.0598387811686084e-05, "elbo": 3.39112, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.740675580577058}
{"step": 97895, "loss": 3.399370574951172, "grad_norm": 0.6640936732292175, "learning_rate": 1.0550084662556753e-05, "elbo": 3.39937, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.740968801313629}
{"step": 97900, "loss": 3.4011486053466795, "grad_norm": 0.6744409203529358, "learning_rate": 1.0501878852171893e-05, "elbo": 3.40115, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7412620220502}
{"step": 97905, "loss": 3.384412384033203, "grad_norm": 0.8759084939956665, "learning_rate": 1.0453770499474586e-05, "elbo": 3.38441, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.74155524278677}
{"step": 97910, "loss": 3.3937705993652343, "grad_norm": 0.7473598122596741, "learning_rate": 1.0405759723167385e-05, "elbo": 3.39377, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.741848463523341}
{"step": 97915, "loss": 3.3903732299804688, "grad_norm": 0.6517718434333801, "learning_rate": 1.0357846641712144e-05, "elbo": 3.39037, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.742141684259911}
{"step": 97920, "loss": 3.398618698120117, "grad_norm": 1.0082775354385376, "learning_rate": 1.0310031373329576e-05, "elbo": 3.39862, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.742434904996482}
{"step": 97925, "loss": 3.3887355804443358, "grad_norm": 0.721798837184906, "learning_rate": 1.026231403599915e-05, "elbo": 3.38874, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.742728125733052}
{"step": 97930, "loss": 3.4005058288574217, "grad_norm": 0.6939556002616882, "learning_rate": 1.0214694747458631e-05, "elbo": 3.40051, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7430213464696225}
{"step": 97935, "loss": 3.3907264709472655, "grad_norm": 0.6930142045021057, "learning_rate": 1.016717362520388e-05, "elbo": 3.39073, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.743314567206193}
{"step": 97940, "loss": 3.3996917724609377, "grad_norm": 0.6813298463821411, "learning_rate": 1.0119750786488536e-05, "elbo": 3.39969, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.743607787942763}
{"step": 97945, "loss": 3.402443695068359, "grad_norm": 0.6929855942726135, "learning_rate": 1.0072426348323754e-05, "elbo": 3.40244, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.743901008679334}
{"step": 97950, "loss": 3.4174789428710937, "grad_norm": 0.7105804085731506, "learning_rate": 1.0025200427477876e-05, "elbo": 3.41748, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.744194229415904}
{"step": 97955, "loss": 3.381575012207031, "grad_norm": 0.7259080410003662, "learning_rate": 9.978073140476169e-06, "elbo": 3.38157, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.744487450152475}
{"step": 97960, "loss": 3.3929237365722655, "grad_norm": 0.6870777010917664, "learning_rate": 9.931044603600514e-06, "elbo": 3.39292, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.744780670889045}
{"step": 97965, "loss": 3.3853492736816406, "grad_norm": 0.6763260960578918, "learning_rate": 9.884114932889171e-06, "elbo": 3.38535, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.745073891625616}
{"step": 97970, "loss": 3.383222198486328, "grad_norm": 0.6933138370513916, "learning_rate": 9.83728424413643e-06, "elbo": 3.38322, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.745367112362186}
{"step": 97975, "loss": 3.3957454681396486, "grad_norm": 0.7018860578536987, "learning_rate": 9.79055265289236e-06, "elbo": 3.39575, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.745660333098757}
{"step": 97980, "loss": 3.3932212829589843, "grad_norm": 0.6566734313964844, "learning_rate": 9.743920274462498e-06, "elbo": 3.39322, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.745953553835327}
{"step": 97985, "loss": 3.3990169525146485, "grad_norm": 0.6562850475311279, "learning_rate": 9.69738722390765e-06, "elbo": 3.39902, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7462467745718975}
{"step": 97985, "loss": 3.3990169525146485, "grad_norm": 0.6562850475311279, "learning_rate": 9.69738722390765e-06, "elbo": 3.39902, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7462467745718975}
{"step": 97990, "loss": 3.396410369873047, "grad_norm": 0.690162181854248, "learning_rate": 9.65095361604344e-06, "elbo": 3.39641, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.746539995308468}
{"step": 97995, "loss": 3.3903324127197267, "grad_norm": 0.7159439921379089, "learning_rate": 9.60461956544021e-06, "elbo": 3.39033, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.746833216045038}
{"step": 98000, "loss": 3.3914955139160154, "grad_norm": 0.7501589059829712, "learning_rate": 9.558385186422603e-06, "elbo": 3.3915, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.747126436781609}
{"step": 98005, "loss": 3.4111167907714846, "grad_norm": 0.6711862683296204, "learning_rate": 9.512250593069393e-06, "elbo": 3.41112, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.747419657518179}
{"step": 98010, "loss": 3.399507522583008, "grad_norm": 0.7907613515853882, "learning_rate": 9.466215899213065e-06, "elbo": 3.39951, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.747712878254751}
{"step": 98015, "loss": 3.3982192993164064, "grad_norm": 0.6892922520637512, "learning_rate": 9.420281218439648e-06, "elbo": 3.39822, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.74800609899132}
{"step": 98020, "loss": 3.399929428100586, "grad_norm": 0.6608400940895081, "learning_rate": 9.374446664088427e-06, "elbo": 3.39993, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7482993197278915}
{"step": 98025, "loss": 3.390774154663086, "grad_norm": 0.9619752168655396, "learning_rate": 9.32871234925159e-06, "elbo": 3.39077, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.748592540464462}
{"step": 98030, "loss": 3.3854942321777344, "grad_norm": 0.6568264365196228, "learning_rate": 9.283078386774025e-06, "elbo": 3.38549, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7488857612010325}
{"step": 98035, "loss": 3.3855545043945314, "grad_norm": 0.7883116006851196, "learning_rate": 9.237544889252969e-06, "elbo": 3.38555, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.749178981937603}
{"step": 98040, "loss": 3.3994827270507812, "grad_norm": 0.8165348172187805, "learning_rate": 9.192111969037841e-06, "elbo": 3.39948, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.749472202674173}
{"step": 98045, "loss": 3.4038986206054687, "grad_norm": 0.683269202709198, "learning_rate": 9.146779738229838e-06, "elbo": 3.4039, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.749765423410744}
{"step": 98050, "loss": 3.3890914916992188, "grad_norm": 0.6872516870498657, "learning_rate": 9.101548308681734e-06, "elbo": 3.38909, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.750058644147314}
{"step": 98055, "loss": 3.3866287231445313, "grad_norm": 0.7096714973449707, "learning_rate": 9.056417791997568e-06, "elbo": 3.38663, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.750351864883885}
{"step": 98060, "loss": 3.39853515625, "grad_norm": 0.7305824160575867, "learning_rate": 9.011388299532441e-06, "elbo": 3.39853, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.750645085620455}
{"step": 98065, "loss": 3.3946517944335937, "grad_norm": 0.6950092911720276, "learning_rate": 8.966459942392109e-06, "elbo": 3.39465, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.750938306357026}
{"step": 98065, "loss": 3.3946517944335937, "grad_norm": 0.6950092911720276, "learning_rate": 8.966459942392109e-06, "elbo": 3.39465, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.750938306357026}
{"step": 98070, "loss": 3.395848846435547, "grad_norm": 0.8033450841903687, "learning_rate": 8.921632831432852e-06, "elbo": 3.39585, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.751231527093596}
{"step": 98075, "loss": 3.393717575073242, "grad_norm": 0.862835705280304, "learning_rate": 8.876907077261093e-06, "elbo": 3.39372, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7515247478301665}
{"step": 98080, "loss": 3.3949798583984374, "grad_norm": 1.1251115798950195, "learning_rate": 8.832282790233221e-06, "elbo": 3.39498, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.751817968566737}
{"step": 98080, "loss": 3.3949798583984374, "grad_norm": 1.1251115798950195, "learning_rate": 8.832282790233221e-06, "elbo": 3.39498, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.751817968566737}
{"step": 98085, "loss": 3.3841316223144533, "grad_norm": 0.7212384343147278, "learning_rate": 8.78776008045517e-06, "elbo": 3.38413, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7521111893033074}
{"step": 98090, "loss": 3.3892436981201173, "grad_norm": 0.7711348533630371, "learning_rate": 8.743339057782351e-06, "elbo": 3.38924, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.752404410039878}
{"step": 98095, "loss": 3.4031280517578124, "grad_norm": 0.6734252572059631, "learning_rate": 8.699019831819205e-06, "elbo": 3.40313, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.752697630776448}
{"step": 98100, "loss": 3.385858154296875, "grad_norm": 0.6511624455451965, "learning_rate": 8.654802511919013e-06, "elbo": 3.38586, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.752990851513019}
{"step": 98105, "loss": 3.401976776123047, "grad_norm": 0.7828192114830017, "learning_rate": 8.610687207183605e-06, "elbo": 3.40198, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.753284072249589}
{"step": 98110, "loss": 3.4030433654785157, "grad_norm": 0.8587237596511841, "learning_rate": 8.566674026463135e-06, "elbo": 3.40304, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.75357729298616}
{"step": 98115, "loss": 3.3926471710205077, "grad_norm": 0.684048056602478, "learning_rate": 8.52276307835574e-06, "elbo": 3.39265, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.75387051372273}
{"step": 98120, "loss": 3.4007530212402344, "grad_norm": 0.7452256679534912, "learning_rate": 8.478954471207317e-06, "elbo": 3.40075, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.754163734459301}
{"step": 98120, "loss": 3.4007530212402344, "grad_norm": 0.7452256679534912, "learning_rate": 8.478954471207317e-06, "elbo": 3.40075, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.754163734459301}
{"step": 98125, "loss": 3.3935585021972656, "grad_norm": 0.6973762512207031, "learning_rate": 8.435248313111243e-06, "elbo": 3.39356, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.754456955195871}
{"step": 98130, "loss": 3.3844284057617187, "grad_norm": 0.670293927192688, "learning_rate": 8.391644711908136e-06, "elbo": 3.38443, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7547501759324415}
{"step": 98135, "loss": 3.4028347015380858, "grad_norm": 1.1247484683990479, "learning_rate": 8.348143775185536e-06, "elbo": 3.40283, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.755043396669013}
{"step": 98140, "loss": 3.3918472290039063, "grad_norm": 0.9866786003112793, "learning_rate": 8.304745610277687e-06, "elbo": 3.39185, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7553366174055824}
{"step": 98145, "loss": 3.4021461486816404, "grad_norm": 0.667934238910675, "learning_rate": 8.261450324265224e-06, "elbo": 3.40215, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.755629838142154}
{"step": 98150, "loss": 3.3982192993164064, "grad_norm": 0.8451879620552063, "learning_rate": 8.218258023975006e-06, "elbo": 3.39822, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.755923058878724}
{"step": 98155, "loss": 3.3972644805908203, "grad_norm": 0.7047753930091858, "learning_rate": 8.17516881597969e-06, "elbo": 3.39726, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.756216279615295}
{"step": 98160, "loss": 3.390229034423828, "grad_norm": 0.706443727016449, "learning_rate": 8.13218280659765e-06, "elbo": 3.39023, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.756509500351865}
{"step": 98165, "loss": 3.3833274841308594, "grad_norm": 0.6898561716079712, "learning_rate": 8.08930010189256e-06, "elbo": 3.38333, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.756802721088436}
{"step": 98170, "loss": 3.3885753631591795, "grad_norm": 0.7031892538070679, "learning_rate": 8.046520807673274e-06, "elbo": 3.38858, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.757095941825006}
{"step": 98175, "loss": 3.4037315368652346, "grad_norm": 0.6811073422431946, "learning_rate": 8.003845029493407e-06, "elbo": 3.40373, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7573891625615765}
{"step": 98180, "loss": 3.3841171264648438, "grad_norm": 0.701683521270752, "learning_rate": 7.961272872651221e-06, "elbo": 3.38412, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.757682383298147}
{"step": 98185, "loss": 3.375834655761719, "grad_norm": 0.7133597135543823, "learning_rate": 7.91880444218927e-06, "elbo": 3.37583, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.757975604034717}
{"step": 98190, "loss": 3.3889083862304688, "grad_norm": 0.6510565280914307, "learning_rate": 7.876439842894195e-06, "elbo": 3.38891, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.758268824771288}
{"step": 98195, "loss": 3.3954853057861327, "grad_norm": 0.9222056269645691, "learning_rate": 7.834179179296419e-06, "elbo": 3.39549, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.758562045507858}
{"step": 98200, "loss": 3.4034461975097656, "grad_norm": 0.6931716203689575, "learning_rate": 7.792022555669947e-06, "elbo": 3.40345, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.758855266244429}
{"step": 98205, "loss": 3.402337646484375, "grad_norm": 0.6748948693275452, "learning_rate": 7.749970076032049e-06, "elbo": 3.40234, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.759148486980999}
{"step": 98210, "loss": 3.373291778564453, "grad_norm": 0.7542356848716736, "learning_rate": 7.708021844143032e-06, "elbo": 3.37329, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.75944170771757}
{"step": 98215, "loss": 3.3768653869628906, "grad_norm": 0.6769194006919861, "learning_rate": 7.666177963505988e-06, "elbo": 3.37687, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.75973492845414}
{"step": 98220, "loss": 3.3869552612304688, "grad_norm": 0.6460399031639099, "learning_rate": 7.6244385373665465e-06, "elbo": 3.38696, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.760028149190711}
{"step": 98225, "loss": 3.401808166503906, "grad_norm": 0.6846469640731812, "learning_rate": 7.582803668712579e-06, "elbo": 3.40181, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.760321369927281}
{"step": 98230, "loss": 3.403070831298828, "grad_norm": 0.686981201171875, "learning_rate": 7.541273460273996e-06, "elbo": 3.40307, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7606145906638515}
{"step": 98235, "loss": 3.391396713256836, "grad_norm": 0.770503044128418, "learning_rate": 7.499848014522443e-06, "elbo": 3.3914, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.760907811400422}
{"step": 98240, "loss": 3.3706993103027343, "grad_norm": 0.6519693732261658, "learning_rate": 7.458527433671114e-06, "elbo": 3.3707, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.761201032136992}
{"step": 98245, "loss": 3.3780067443847654, "grad_norm": 0.6636600494384766, "learning_rate": 7.4173118196744e-06, "elbo": 3.37801, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.761494252873563}
{"step": 98250, "loss": 3.3722774505615236, "grad_norm": 0.6491532325744629, "learning_rate": 7.376201274227762e-06, "elbo": 3.37228, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.761787473610133}
{"step": 98255, "loss": 3.3918167114257813, "grad_norm": 0.6897863149642944, "learning_rate": 7.335195898767366e-06, "elbo": 3.39182, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.762080694346704}
{"step": 98260, "loss": 3.3704986572265625, "grad_norm": 1.1872220039367676, "learning_rate": 7.294295794469924e-06, "elbo": 3.3705, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.762373915083275}
{"step": 98265, "loss": 3.396638870239258, "grad_norm": 0.6796107292175293, "learning_rate": 7.253501062252338e-06, "elbo": 3.39664, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.762667135819845}
{"step": 98270, "loss": 3.3905868530273438, "grad_norm": 0.904106616973877, "learning_rate": 7.212811802771591e-06, "elbo": 3.39059, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.762960356556416}
{"step": 98275, "loss": 3.384339141845703, "grad_norm": 0.8135324120521545, "learning_rate": 7.172228116424373e-06, "elbo": 3.38434, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7632535772929865}
{"step": 98280, "loss": 3.3859817504882814, "grad_norm": 0.7730229496955872, "learning_rate": 7.1317501033468905e-06, "elbo": 3.38598, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.763546798029557}
{"step": 98285, "loss": 3.3945083618164062, "grad_norm": 0.6891756653785706, "learning_rate": 7.091377863414611e-06, "elbo": 3.39451, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.763840018766127}
{"step": 98290, "loss": 3.387156677246094, "grad_norm": 0.6761906147003174, "learning_rate": 7.051111496242047e-06, "elbo": 3.38716, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.764133239502698}
{"step": 98295, "loss": 3.399341583251953, "grad_norm": 0.6864597201347351, "learning_rate": 7.010951101182439e-06, "elbo": 3.39934, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.764426460239268}
{"step": 98300, "loss": 3.3959430694580077, "grad_norm": 0.7378166913986206, "learning_rate": 6.970896777327573e-06, "elbo": 3.39594, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.764719680975839}
{"step": 98305, "loss": 3.3965282440185547, "grad_norm": 0.6922692656517029, "learning_rate": 6.930948623507505e-06, "elbo": 3.39653, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.765012901712409}
{"step": 98310, "loss": 3.406999206542969, "grad_norm": 0.7549840211868286, "learning_rate": 6.891106738290348e-06, "elbo": 3.407, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.76530612244898}
{"step": 98315, "loss": 3.386034393310547, "grad_norm": 0.6911600828170776, "learning_rate": 6.851371219981989e-06, "elbo": 3.38603, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.76559934318555}
{"step": 98315, "loss": 3.386034393310547, "grad_norm": 0.6911600828170776, "learning_rate": 6.851371219981989e-06, "elbo": 3.38603, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.76559934318555}
{"step": 98320, "loss": 3.395248794555664, "grad_norm": 0.6732039451599121, "learning_rate": 6.811742166625862e-06, "elbo": 3.39525, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.765892563922121}
{"step": 98325, "loss": 3.392116165161133, "grad_norm": 0.6672446727752686, "learning_rate": 6.772219676002716e-06, "elbo": 3.39212, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.766185784658691}
{"step": 98330, "loss": 3.3924301147460936, "grad_norm": 0.7675579190254211, "learning_rate": 6.732803845630392e-06, "elbo": 3.39243, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7664790053952615}
{"step": 98335, "loss": 3.4000709533691404, "grad_norm": 0.6754626035690308, "learning_rate": 6.693494772763486e-06, "elbo": 3.40007, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.766772226131832}
{"step": 98340, "loss": 3.389342498779297, "grad_norm": 0.7179508209228516, "learning_rate": 6.654292554393276e-06, "elbo": 3.38934, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.767065446868402}
{"step": 98345, "loss": 3.379717254638672, "grad_norm": 0.8159048557281494, "learning_rate": 6.615197287247299e-06, "elbo": 3.37972, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.767358667604973}
{"step": 98350, "loss": 3.389414596557617, "grad_norm": 0.6926963329315186, "learning_rate": 6.576209067789285e-06, "elbo": 3.38941, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.767651888341543}
{"step": 98355, "loss": 3.403123474121094, "grad_norm": 0.6830133199691772, "learning_rate": 6.5373279922187445e-06, "elbo": 3.40312, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.767945109078114}
{"step": 98360, "loss": 3.4029891967773436, "grad_norm": 0.6703078746795654, "learning_rate": 6.498554156470899e-06, "elbo": 3.40299, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.768238329814684}
{"step": 98365, "loss": 3.375160980224609, "grad_norm": 0.684491753578186, "learning_rate": 6.459887656216318e-06, "elbo": 3.37516, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.768531550551255}
{"step": 98370, "loss": 3.3946868896484377, "grad_norm": 0.6774111986160278, "learning_rate": 6.421328586860747e-06, "elbo": 3.39469, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.768824771287825}
{"step": 98375, "loss": 3.4003501892089845, "grad_norm": 0.6572479009628296, "learning_rate": 6.382877043544855e-06, "elbo": 3.40035, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7691179920243965}
{"step": 98380, "loss": 3.397625732421875, "grad_norm": 0.695275604724884, "learning_rate": 6.344533121144014e-06, "elbo": 3.39763, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.769411212760966}
{"step": 98385, "loss": 3.3846359252929688, "grad_norm": 0.6680839657783508, "learning_rate": 6.30629691426804e-06, "elbo": 3.38464, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.769704433497537}
{"step": 98390, "loss": 3.4055549621582033, "grad_norm": 0.6757307648658752, "learning_rate": 6.268168517260964e-06, "elbo": 3.40556, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.769997654234108}
{"step": 98395, "loss": 3.407613754272461, "grad_norm": 0.6547414064407349, "learning_rate": 6.23014802420081e-06, "elbo": 3.40761, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.770290874970678}
{"step": 98400, "loss": 3.3882484436035156, "grad_norm": 0.7119588255882263, "learning_rate": 6.192235528899382e-06, "elbo": 3.38825, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.770584095707249}
{"step": 98405, "loss": 3.413613128662109, "grad_norm": 0.6697394847869873, "learning_rate": 6.154431124901982e-06, "elbo": 3.41361, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.770877316443819}
{"step": 98410, "loss": 3.3826011657714843, "grad_norm": 0.6714845895767212, "learning_rate": 6.116734905487226e-06, "elbo": 3.3826, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.77117053718039}
{"step": 98415, "loss": 3.390943145751953, "grad_norm": 0.698344349861145, "learning_rate": 6.079146963666776e-06, "elbo": 3.39094, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.77146375791696}
{"step": 98420, "loss": 3.4051982879638674, "grad_norm": 0.7614485621452332, "learning_rate": 6.0416673921851576e-06, "elbo": 3.4052, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7717569786535305}
{"step": 98425, "loss": 3.384407806396484, "grad_norm": 0.766962468624115, "learning_rate": 6.004296283519478e-06, "elbo": 3.38441, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.772050199390101}
{"step": 98430, "loss": 3.3866920471191406, "grad_norm": 0.6639828085899353, "learning_rate": 5.96703372987924e-06, "elbo": 3.38669, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7723434201266715}
{"step": 98435, "loss": 3.382080078125, "grad_norm": 0.682094931602478, "learning_rate": 5.9298798232060695e-06, "elbo": 3.38208, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.772636640863242}
{"step": 98440, "loss": 3.393436813354492, "grad_norm": 0.8702563047409058, "learning_rate": 5.892834655173579e-06, "elbo": 3.39344, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.772929861599812}
{"step": 98445, "loss": 3.3981300354003907, "grad_norm": 0.7965014576911926, "learning_rate": 5.855898317186992e-06, "elbo": 3.39813, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.773223082336383}
{"step": 98450, "loss": 3.3798442840576173, "grad_norm": 0.7305021286010742, "learning_rate": 5.819070900383078e-06, "elbo": 3.37984, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.773516303072953}
{"step": 98455, "loss": 3.400617980957031, "grad_norm": 0.6587472558021545, "learning_rate": 5.782352495629822e-06, "elbo": 3.40062, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.773809523809524}
{"step": 98460, "loss": 3.3929447174072265, "grad_norm": 0.6674961447715759, "learning_rate": 5.745743193526237e-06, "elbo": 3.39294, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.774102744546094}
{"step": 98465, "loss": 3.3847389221191406, "grad_norm": 0.7002393007278442, "learning_rate": 5.7092430844021275e-06, "elbo": 3.38474, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.774395965282665}
{"step": 98470, "loss": 3.3809654235839846, "grad_norm": 0.6794306635856628, "learning_rate": 5.672852258317901e-06, "elbo": 3.38097, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.774689186019235}
{"step": 98475, "loss": 3.4081100463867187, "grad_norm": 0.680270254611969, "learning_rate": 5.636570805064301e-06, "elbo": 3.40811, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7749824067558055}
{"step": 98480, "loss": 3.3933067321777344, "grad_norm": 0.6481443643569946, "learning_rate": 5.600398814162194e-06, "elbo": 3.39331, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.775275627492376}
{"step": 98485, "loss": 3.4030376434326173, "grad_norm": 0.7274940609931946, "learning_rate": 5.564336374862372e-06, "elbo": 3.40304, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7755688482289465}
{"step": 98490, "loss": 3.3935874938964843, "grad_norm": 0.6478208303451538, "learning_rate": 5.528383576145335e-06, "elbo": 3.39359, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.775862068965517}
{"step": 98495, "loss": 3.4168380737304687, "grad_norm": 0.6743327975273132, "learning_rate": 5.492540506721033e-06, "elbo": 3.41684, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.776155289702087}
{"step": 98500, "loss": 3.3924041748046876, "grad_norm": 0.6715584993362427, "learning_rate": 5.456807255028673e-06, "elbo": 3.3924, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.776448510438659}
{"step": 98505, "loss": 3.3891239166259766, "grad_norm": 0.6784249544143677, "learning_rate": 5.421183909236494e-06, "elbo": 3.38912, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.776741731175228}
{"step": 98510, "loss": 3.384675979614258, "grad_norm": 0.680300235748291, "learning_rate": 5.385670557241579e-06, "elbo": 3.38468, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7770349519118}
{"step": 98515, "loss": 3.388003921508789, "grad_norm": 0.6678885221481323, "learning_rate": 5.350267286669586e-06, "elbo": 3.388, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.77732817264837}
{"step": 98520, "loss": 3.3859832763671873, "grad_norm": 0.6659762263298035, "learning_rate": 5.31497418487456e-06, "elbo": 3.38598, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7776213933849405}
{"step": 98525, "loss": 3.4063438415527343, "grad_norm": 0.7401593327522278, "learning_rate": 5.279791338938716e-06, "elbo": 3.40634, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.777914614121511}
{"step": 98530, "loss": 3.4036659240722655, "grad_norm": 0.6964154243469238, "learning_rate": 5.2447188356722636e-06, "elbo": 3.40367, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.778207834858081}
{"step": 98535, "loss": 3.3806182861328127, "grad_norm": 0.7044309973716736, "learning_rate": 5.209756761613071e-06, "elbo": 3.38062, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.778501055594652}
{"step": 98540, "loss": 3.4008739471435545, "grad_norm": 0.8242059946060181, "learning_rate": 5.174905203026608e-06, "elbo": 3.40087, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.778794276331222}
{"step": 98545, "loss": 3.4049022674560545, "grad_norm": 0.6687778830528259, "learning_rate": 5.140164245905632e-06, "elbo": 3.4049, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.779087497067793}
{"step": 98550, "loss": 3.3880950927734377, "grad_norm": 0.7031693458557129, "learning_rate": 5.10553397596999e-06, "elbo": 3.38809, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.779380717804363}
{"step": 98555, "loss": 3.3964385986328125, "grad_norm": 0.6829410195350647, "learning_rate": 5.071014478666425e-06, "elbo": 3.39644, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.779673938540934}
{"step": 98560, "loss": 3.390086364746094, "grad_norm": 0.6422129273414612, "learning_rate": 5.03660583916839e-06, "elbo": 3.39009, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.779967159277504}
{"step": 98560, "loss": 3.390086364746094, "grad_norm": 0.6422129273414612, "learning_rate": 5.03660583916839e-06, "elbo": 3.39009, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.779967159277504}
{"step": 98565, "loss": 3.4122154235839846, "grad_norm": 0.7050926089286804, "learning_rate": 5.002308142375761e-06, "elbo": 3.41222, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.780260380014075}
{"step": 98565, "loss": 3.4122154235839846, "grad_norm": 0.7050926089286804, "learning_rate": 5.002308142375761e-06, "elbo": 3.41222, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.780260380014075}
{"step": 98570, "loss": 3.3939361572265625, "grad_norm": 1.1082521677017212, "learning_rate": 4.968121472914705e-06, "elbo": 3.39394, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.780553600750645}
{"step": 98570, "loss": 3.3939361572265625, "grad_norm": 1.1082521677017212, "learning_rate": 4.968121472914705e-06, "elbo": 3.39394, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.780553600750645}
{"step": 98575, "loss": 3.3850166320800783, "grad_norm": 0.6599543690681458, "learning_rate": 4.9340459151374196e-06, "elbo": 3.38502, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7808468214872155}
{"step": 98580, "loss": 3.399680328369141, "grad_norm": 0.6904357671737671, "learning_rate": 4.900081553121971e-06, "elbo": 3.39968, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.781140042223786}
{"step": 98585, "loss": 3.4017921447753907, "grad_norm": 0.6555585861206055, "learning_rate": 4.86622847067204e-06, "elbo": 3.40179, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.781433262960356}
{"step": 98585, "loss": 3.4017921447753907, "grad_norm": 0.6555585861206055, "learning_rate": 4.86622847067204e-06, "elbo": 3.40179, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.781433262960356}
{"step": 98590, "loss": 3.404765319824219, "grad_norm": 1.487061619758606, "learning_rate": 4.832486751316745e-06, "elbo": 3.40477, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.781726483696927}
{"step": 98590, "loss": 3.404765319824219, "grad_norm": 1.487061619758606, "learning_rate": 4.832486751316745e-06, "elbo": 3.40477, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.781726483696927}
{"step": 98595, "loss": 3.396337890625, "grad_norm": 0.6941655278205872, "learning_rate": 4.798856478310409e-06, "elbo": 3.39634, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.782019704433497}
{"step": 98600, "loss": 3.3907623291015625, "grad_norm": 0.7589660882949829, "learning_rate": 4.765337734632408e-06, "elbo": 3.39076, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.782312925170068}
{"step": 98605, "loss": 3.401446533203125, "grad_norm": 0.6925349235534668, "learning_rate": 4.731930602986906e-06, "elbo": 3.40145, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.782606145906638}
{"step": 98610, "loss": 3.3700271606445313, "grad_norm": 0.7185423374176025, "learning_rate": 4.698635165802673e-06, "elbo": 3.37003, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.782899366643209}
{"step": 98615, "loss": 3.4015270233154298, "grad_norm": 0.7400160431861877, "learning_rate": 4.665451505232882e-06, "elbo": 3.40153, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.783192587379779}
{"step": 98620, "loss": 3.3860137939453123, "grad_norm": 0.8229462504386902, "learning_rate": 4.632379703154949e-06, "elbo": 3.38601, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.78348580811635}
{"step": 98625, "loss": 3.4039405822753905, "grad_norm": 0.667003333568573, "learning_rate": 4.599419841170216e-06, "elbo": 3.40394, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.783779028852921}
{"step": 98630, "loss": 3.38829345703125, "grad_norm": 0.6775118708610535, "learning_rate": 4.566572000603892e-06, "elbo": 3.38829, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7840722495894905}
{"step": 98635, "loss": 3.3968429565429688, "grad_norm": 0.6463621258735657, "learning_rate": 4.533836262504759e-06, "elbo": 3.39684, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.784365470326062}
{"step": 98640, "loss": 3.394557571411133, "grad_norm": 0.7666015028953552, "learning_rate": 4.501212707644986e-06, "elbo": 3.39456, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.784658691062632}
{"step": 98640, "loss": 3.394557571411133, "grad_norm": 0.7666015028953552, "learning_rate": 4.501212707644986e-06, "elbo": 3.39456, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.784658691062632}
{"step": 98645, "loss": 3.383313751220703, "grad_norm": 0.7821991443634033, "learning_rate": 4.468701416519955e-06, "elbo": 3.38331, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.784951911799203}
{"step": 98650, "loss": 3.3776336669921876, "grad_norm": 0.6432985663414001, "learning_rate": 4.436302469348041e-06, "elbo": 3.37763, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.785245132535773}
{"step": 98655, "loss": 3.3920623779296877, "grad_norm": 0.6579523086547852, "learning_rate": 4.404015946070439e-06, "elbo": 3.39206, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.785538353272344}
{"step": 98660, "loss": 3.4071701049804686, "grad_norm": 0.6836730241775513, "learning_rate": 4.371841926350922e-06, "elbo": 3.40717, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.785831574008914}
{"step": 98665, "loss": 3.3904697418212892, "grad_norm": 0.6554431915283203, "learning_rate": 4.3397804895756955e-06, "elbo": 3.39047, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.786124794745485}
{"step": 98670, "loss": 3.386930465698242, "grad_norm": 0.6711239814758301, "learning_rate": 4.30783171485315e-06, "elbo": 3.38693, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.786418015482055}
{"step": 98675, "loss": 3.3951980590820314, "grad_norm": 0.7843886613845825, "learning_rate": 4.275995681013745e-06, "elbo": 3.3952, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7867112362186255}
{"step": 98680, "loss": 3.381312942504883, "grad_norm": 1.1139129400253296, "learning_rate": 4.244272466609689e-06, "elbo": 3.38131, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.787004456955196}
{"step": 98685, "loss": 3.4051490783691407, "grad_norm": 0.6654572486877441, "learning_rate": 4.212662149914886e-06, "elbo": 3.40515, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.787297677691766}
{"step": 98690, "loss": 3.386544036865234, "grad_norm": 0.6734930276870728, "learning_rate": 4.1811648089246365e-06, "elbo": 3.38654, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.787590898428337}
{"step": 98695, "loss": 3.3737247467041014, "grad_norm": 0.6720353364944458, "learning_rate": 4.149780521355523e-06, "elbo": 3.37372, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.787884119164907}
{"step": 98700, "loss": 3.38671760559082, "grad_norm": 0.8301843404769897, "learning_rate": 4.118509364645118e-06, "elbo": 3.38672, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.788177339901478}
{"step": 98705, "loss": 3.3680530548095704, "grad_norm": 0.6815997958183289, "learning_rate": 4.087351415951918e-06, "elbo": 3.36805, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.788470560638048}
{"step": 98710, "loss": 3.3899612426757812, "grad_norm": 0.6693112850189209, "learning_rate": 4.056306752155059e-06, "elbo": 3.38996, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.788763781374619}
{"step": 98715, "loss": 3.37313232421875, "grad_norm": 0.7467309236526489, "learning_rate": 4.025375449854163e-06, "elbo": 3.37313, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.789057002111189}
{"step": 98720, "loss": 3.383234405517578, "grad_norm": 0.6760557889938354, "learning_rate": 3.994557585369135e-06, "elbo": 3.38323, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.78935022284776}
{"step": 98725, "loss": 3.3894866943359374, "grad_norm": 0.7024657130241394, "learning_rate": 3.963853234740006e-06, "elbo": 3.38949, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.78964344358433}
{"step": 98730, "loss": 3.4137168884277345, "grad_norm": 0.8974186778068542, "learning_rate": 3.933262473726701e-06, "elbo": 3.41372, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7899366643209005}
{"step": 98730, "loss": 3.4137168884277345, "grad_norm": 0.8974186778068542, "learning_rate": 3.933262473726701e-06, "elbo": 3.41372, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7899366643209005}
{"step": 98735, "loss": 3.3750373840332033, "grad_norm": 0.6819778680801392, "learning_rate": 3.902785377808882e-06, "elbo": 3.37504, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.790229885057471}
{"step": 98740, "loss": 3.3861839294433596, "grad_norm": 0.6628040671348572, "learning_rate": 3.872422022185746e-06, "elbo": 3.38618, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.790523105794041}
{"step": 98745, "loss": 3.385567474365234, "grad_norm": 0.6396012902259827, "learning_rate": 3.842172481775874e-06, "elbo": 3.38557, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.790816326530612}
{"step": 98750, "loss": 3.3970890045166016, "grad_norm": 1.276186466217041, "learning_rate": 3.812036831216986e-06, "elbo": 3.39709, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.791109547267183}
{"step": 98750, "loss": 3.3970890045166016, "grad_norm": 1.276186466217041, "learning_rate": 3.812036831216986e-06, "elbo": 3.39709, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.791109547267183}
{"step": 98755, "loss": 3.384885787963867, "grad_norm": 0.7544485926628113, "learning_rate": 3.782015144865808e-06, "elbo": 3.38489, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.791402768003753}
{"step": 98755, "loss": 3.384885787963867, "grad_norm": 0.7544485926628113, "learning_rate": 3.782015144865808e-06, "elbo": 3.38489, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.791402768003753}
{"step": 98760, "loss": 3.3892955780029297, "grad_norm": 0.6763165593147278, "learning_rate": 3.7521074967978577e-06, "elbo": 3.3893, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.791695988740324}
{"step": 98765, "loss": 3.4027988433837892, "grad_norm": 0.6626390218734741, "learning_rate": 3.7223139608073e-06, "elbo": 3.4028, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7919892094768946}
{"step": 98770, "loss": 3.3829425811767577, "grad_norm": 1.3595651388168335, "learning_rate": 3.6926346104067143e-06, "elbo": 3.38294, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.792282430213465}
{"step": 98775, "loss": 3.371878814697266, "grad_norm": 0.6881340742111206, "learning_rate": 3.6630695188269505e-06, "elbo": 3.37188, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7925756509500355}
{"step": 98780, "loss": 3.391509246826172, "grad_norm": 0.8149372935295105, "learning_rate": 3.6336187590169214e-06, "elbo": 3.39151, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.792868871686606}
{"step": 98785, "loss": 3.381700897216797, "grad_norm": 0.7091418504714966, "learning_rate": 3.604282403643472e-06, "elbo": 3.3817, "mask_frac": 0.5011, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.793162092423176}
{"step": 98790, "loss": 3.4034568786621096, "grad_norm": 0.7545256018638611, "learning_rate": 3.575060525091117e-06, "elbo": 3.40346, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.793455313159747}
{"step": 98795, "loss": 3.396055221557617, "grad_norm": 0.7031484246253967, "learning_rate": 3.545953195461954e-06, "elbo": 3.39605, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.793748533896317}
{"step": 98800, "loss": 3.3910343170166017, "grad_norm": 0.6481612920761108, "learning_rate": 3.5169604865754212e-06, "elbo": 3.39103, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.794041754632888}
{"step": 98805, "loss": 3.399739074707031, "grad_norm": 0.7354663014411926, "learning_rate": 3.488082469968146e-06, "elbo": 3.39974, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.794334975369458}
{"step": 98810, "loss": 3.383067321777344, "grad_norm": 0.7086325287818909, "learning_rate": 3.459319216893758e-06, "elbo": 3.38307, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.794628196106029}
{"step": 98815, "loss": 3.383479690551758, "grad_norm": 0.6891538500785828, "learning_rate": 3.4306707983227325e-06, "elbo": 3.38348, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.794921416842599}
{"step": 98820, "loss": 3.3897624969482423, "grad_norm": 0.7473463416099548, "learning_rate": 3.4021372849421972e-06, "elbo": 3.38976, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7952146375791695}
{"step": 98825, "loss": 3.4083507537841795, "grad_norm": 0.6544449925422668, "learning_rate": 3.373718747155752e-06, "elbo": 3.40835, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.79550785831574}
{"step": 98830, "loss": 3.3983413696289064, "grad_norm": 0.686135470867157, "learning_rate": 3.3454152550833063e-06, "elbo": 3.39834, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.7958010790523105}
{"step": 98835, "loss": 3.4012161254882813, "grad_norm": 0.6824153661727905, "learning_rate": 3.3172268785609307e-06, "elbo": 3.40122, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.796094299788881}
{"step": 98840, "loss": 3.399699401855469, "grad_norm": 0.8003321886062622, "learning_rate": 3.2891536871406393e-06, "elbo": 3.3997, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.796387520525451}
{"step": 98845, "loss": 3.3930740356445312, "grad_norm": 0.6592718958854675, "learning_rate": 3.2611957500902347e-06, "elbo": 3.39307, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.796680741262022}
{"step": 98850, "loss": 3.393335723876953, "grad_norm": 0.6538188457489014, "learning_rate": 3.233353136393141e-06, "elbo": 3.39334, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.796973961998592}
{"step": 98855, "loss": 3.3907676696777345, "grad_norm": 0.6663693189620972, "learning_rate": 3.205625914748256e-06, "elbo": 3.39077, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.797267182735163}
{"step": 98860, "loss": 3.3916629791259765, "grad_norm": 0.7282485961914062, "learning_rate": 3.1780141535697363e-06, "elbo": 3.39166, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.797560403471733}
{"step": 98865, "loss": 3.392874526977539, "grad_norm": 0.6756235361099243, "learning_rate": 3.150517920986851e-06, "elbo": 3.39287, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.797853624208304}
{"step": 98870, "loss": 3.3890869140625, "grad_norm": 0.6872718930244446, "learning_rate": 3.1231372848438243e-06, "elbo": 3.38909, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.798146844944874}
{"step": 98875, "loss": 3.387964630126953, "grad_norm": 0.7382343411445618, "learning_rate": 3.095872312699666e-06, "elbo": 3.38796, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.798440065681445}
{"step": 98880, "loss": 3.3727020263671874, "grad_norm": 0.6730659604072571, "learning_rate": 3.0687230718279615e-06, "elbo": 3.3727, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.798733286418015}
{"step": 98885, "loss": 3.3940948486328124, "grad_norm": 0.6800633072853088, "learning_rate": 3.0416896292167875e-06, "elbo": 3.3941, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.799026507154586}
{"step": 98890, "loss": 3.3932266235351562, "grad_norm": 0.6858242750167847, "learning_rate": 3.0147720515684653e-06, "elbo": 3.39323, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.799319727891157}
{"step": 98895, "loss": 3.3750457763671875, "grad_norm": 0.6718384027481079, "learning_rate": 2.9879704052994394e-06, "elbo": 3.37505, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.799612948627727}
{"step": 98900, "loss": 3.3780006408691405, "grad_norm": 0.687758207321167, "learning_rate": 2.9612847565401005e-06, "elbo": 3.378, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.799906169364298}
{"step": 98905, "loss": 3.385654830932617, "grad_norm": 0.8737455010414124, "learning_rate": 2.9347151711346555e-06, "elbo": 3.38565, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.800199390100868}
{"step": 98910, "loss": 3.375144195556641, "grad_norm": 0.6900250315666199, "learning_rate": 2.908261714640903e-06, "elbo": 3.37514, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.800492610837439}
{"step": 98915, "loss": 3.397530364990234, "grad_norm": 0.693878173828125, "learning_rate": 2.88192445233012e-06, "elbo": 3.39753, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.800785831574009}
{"step": 98920, "loss": 3.385603332519531, "grad_norm": 0.6700573563575745, "learning_rate": 2.8557034491868696e-06, "elbo": 3.3856, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8010790523105795}
{"step": 98925, "loss": 3.404344177246094, "grad_norm": 0.6859787702560425, "learning_rate": 2.8295987699088923e-06, "elbo": 3.40434, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.80137227304715}
{"step": 98930, "loss": 3.3835769653320313, "grad_norm": 0.7004175782203674, "learning_rate": 2.803610478906876e-06, "elbo": 3.38358, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.80166549378372}
{"step": 98935, "loss": 3.400704193115234, "grad_norm": 0.7133764624595642, "learning_rate": 2.7777386403043505e-06, "elbo": 3.4007, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.801958714520291}
{"step": 98940, "loss": 3.399724578857422, "grad_norm": 0.6891422867774963, "learning_rate": 2.7519833179374952e-06, "elbo": 3.39972, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.802251935256861}
{"step": 98945, "loss": 3.4008056640625, "grad_norm": 0.6690624356269836, "learning_rate": 2.7263445753550276e-06, "elbo": 3.40081, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.802545155993432}
{"step": 98950, "loss": 3.3930595397949217, "grad_norm": 0.7635258436203003, "learning_rate": 2.700822475817982e-06, "elbo": 3.39306, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.802838376730002}
{"step": 98955, "loss": 3.3919918060302736, "grad_norm": 0.6582872867584229, "learning_rate": 2.675417082299603e-06, "elbo": 3.39199, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.803131597466573}
{"step": 98960, "loss": 3.392706298828125, "grad_norm": 0.7390221953392029, "learning_rate": 2.6501284574851737e-06, "elbo": 3.39271, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.803424818203143}
{"step": 98965, "loss": 3.3946983337402346, "grad_norm": 1.0013371706008911, "learning_rate": 2.6249566637718714e-06, "elbo": 3.3947, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.803718038939714}
{"step": 98970, "loss": 3.3855628967285156, "grad_norm": 0.6688107252120972, "learning_rate": 2.599901763268575e-06, "elbo": 3.38556, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.804011259676284}
{"step": 98975, "loss": 3.387090301513672, "grad_norm": 0.6570725440979004, "learning_rate": 2.5749638177957835e-06, "elbo": 3.38709, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8043044804128545}
{"step": 98980, "loss": 3.3879684448242187, "grad_norm": 0.6544409990310669, "learning_rate": 2.5501428888853816e-06, "elbo": 3.38797, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.804597701149425}
{"step": 98985, "loss": 3.379730224609375, "grad_norm": 0.6753640174865723, "learning_rate": 2.5254390377805583e-06, "elbo": 3.37973, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.804890921885995}
{"step": 98990, "loss": 3.389771270751953, "grad_norm": 0.6960991621017456, "learning_rate": 2.5008523254355878e-06, "elbo": 3.38977, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.805184142622566}
{"step": 98995, "loss": 3.395107650756836, "grad_norm": 0.6867910623550415, "learning_rate": 2.476382812515765e-06, "elbo": 3.39511, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.805477363359136}
{"step": 99000, "loss": 3.4060733795166014, "grad_norm": 0.6986822485923767, "learning_rate": 2.452030559397167e-06, "elbo": 3.40607, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.805770584095708}
{"step": 99005, "loss": 3.3969512939453126, "grad_norm": 0.670098066329956, "learning_rate": 2.4277956261665626e-06, "elbo": 3.39695, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.806063804832277}
{"step": 99010, "loss": 3.3956642150878906, "grad_norm": 0.6752325892448425, "learning_rate": 2.4036780726212237e-06, "elbo": 3.39566, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.806357025568849}
{"step": 99015, "loss": 3.3732154846191404, "grad_norm": 0.8026899099349976, "learning_rate": 2.3796779582688446e-06, "elbo": 3.37322, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.806650246305419}
{"step": 99020, "loss": 3.37896842956543, "grad_norm": 0.8624197244644165, "learning_rate": 2.3557953423273048e-06, "elbo": 3.37897, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8069434670419895}
{"step": 99025, "loss": 3.3875728607177735, "grad_norm": 0.6840929388999939, "learning_rate": 2.3320302837245843e-06, "elbo": 3.38757, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.80723668777856}
{"step": 99030, "loss": 3.373191070556641, "grad_norm": 0.6670916080474854, "learning_rate": 2.3083828410986e-06, "elbo": 3.37319, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.80752990851513}
{"step": 99030, "loss": 3.373191070556641, "grad_norm": 0.6670916080474854, "learning_rate": 2.3083828410986e-06, "elbo": 3.37319, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.80752990851513}
{"step": 99035, "loss": 3.3949005126953127, "grad_norm": 0.6971323490142822, "learning_rate": 2.2848530727970774e-06, "elbo": 3.3949, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.807823129251701}
{"step": 99035, "loss": 3.3949005126953127, "grad_norm": 0.6971323490142822, "learning_rate": 2.2848530727970774e-06, "elbo": 3.3949, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.807823129251701}
{"step": 99040, "loss": 3.3944351196289064, "grad_norm": 0.7374401092529297, "learning_rate": 2.2614410368773765e-06, "elbo": 3.39444, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.808116349988271}
{"step": 99045, "loss": 3.396076202392578, "grad_norm": 0.74111008644104, "learning_rate": 2.238146791106366e-06, "elbo": 3.39608, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.808409570724842}
{"step": 99050, "loss": 3.3880130767822267, "grad_norm": 0.6485305428504944, "learning_rate": 2.214970392960275e-06, "elbo": 3.38801, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.808702791461412}
{"step": 99055, "loss": 3.3860614776611326, "grad_norm": 0.6666707992553711, "learning_rate": 2.191911899624588e-06, "elbo": 3.38606, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.808996012197983}
{"step": 99060, "loss": 3.385116195678711, "grad_norm": 0.7900135517120361, "learning_rate": 2.1689713679938105e-06, "elbo": 3.38512, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.809289232934553}
{"step": 99065, "loss": 3.384374237060547, "grad_norm": 0.8089678883552551, "learning_rate": 2.1461488546714427e-06, "elbo": 3.38437, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.809582453671124}
{"step": 99070, "loss": 3.4030723571777344, "grad_norm": 0.6569461822509766, "learning_rate": 2.1234444159697665e-06, "elbo": 3.40307, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.809875674407694}
{"step": 99075, "loss": 3.370695877075195, "grad_norm": 0.701750636100769, "learning_rate": 2.100858107909731e-06, "elbo": 3.3707, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8101688951442645}
{"step": 99080, "loss": 3.381775665283203, "grad_norm": 0.7336834073066711, "learning_rate": 2.0783899862207967e-06, "elbo": 3.38178, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.810462115880835}
{"step": 99085, "loss": 3.392293167114258, "grad_norm": 0.7227231860160828, "learning_rate": 2.056040106340823e-06, "elbo": 3.39229, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.810755336617405}
{"step": 99090, "loss": 3.390796661376953, "grad_norm": 0.6881696581840515, "learning_rate": 2.03380852341592e-06, "elbo": 3.3908, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.811048557353976}
{"step": 99095, "loss": 3.381146240234375, "grad_norm": 0.6502912044525146, "learning_rate": 2.011695292300314e-06, "elbo": 3.38115, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.811341778090546}
{"step": 99100, "loss": 3.394744873046875, "grad_norm": 0.6943719983100891, "learning_rate": 1.9897004675561825e-06, "elbo": 3.39474, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.811634998827117}
{"step": 99105, "loss": 3.404763031005859, "grad_norm": 0.7254716157913208, "learning_rate": 1.967824103453597e-06, "elbo": 3.40476, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.811928219563687}
{"step": 99110, "loss": 3.3830772399902345, "grad_norm": 0.6872707009315491, "learning_rate": 1.9460662539702968e-06, "elbo": 3.38308, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.812221440300258}
{"step": 99115, "loss": 3.385087585449219, "grad_norm": 0.651383101940155, "learning_rate": 1.9244269727916096e-06, "elbo": 3.38509, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.812514661036829}
{"step": 99120, "loss": 3.399827575683594, "grad_norm": 0.816069483757019, "learning_rate": 1.9029063133103153e-06, "elbo": 3.39983, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.812807881773399}
{"step": 99125, "loss": 3.3915454864501955, "grad_norm": 0.6961846351623535, "learning_rate": 1.8815043286265045e-06, "elbo": 3.39155, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.81310110250997}
{"step": 99130, "loss": 3.379679489135742, "grad_norm": 0.7370973825454712, "learning_rate": 1.8602210715474466e-06, "elbo": 3.37968, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.81339432324654}
{"step": 99135, "loss": 3.38846435546875, "grad_norm": 0.6461139917373657, "learning_rate": 1.8390565945874571e-06, "elbo": 3.38846, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.813687543983111}
{"step": 99140, "loss": 3.3706008911132814, "grad_norm": 0.6557728052139282, "learning_rate": 1.81801094996778e-06, "elbo": 3.3706, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.813980764719681}
{"step": 99145, "loss": 3.3730159759521485, "grad_norm": 0.6703552007675171, "learning_rate": 1.7970841896164658e-06, "elbo": 3.37302, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.814273985456252}
{"step": 99150, "loss": 3.3936607360839846, "grad_norm": 0.7063208222389221, "learning_rate": 1.7762763651681892e-06, "elbo": 3.39366, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.814567206192822}
{"step": 99155, "loss": 3.3851356506347656, "grad_norm": 0.8317947387695312, "learning_rate": 1.7555875279642087e-06, "elbo": 3.38514, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.814860426929393}
{"step": 99160, "loss": 3.384988784790039, "grad_norm": 0.6772720813751221, "learning_rate": 1.735017729052163e-06, "elbo": 3.38499, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.815153647665963}
{"step": 99165, "loss": 3.3959579467773438, "grad_norm": 0.6972953081130981, "learning_rate": 1.7145670191859974e-06, "elbo": 3.39596, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8154468684025336}
{"step": 99170, "loss": 3.376395416259766, "grad_norm": 0.673527717590332, "learning_rate": 1.6942354488257928e-06, "elbo": 3.3764, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.815740089139104}
{"step": 99175, "loss": 3.398673248291016, "grad_norm": 0.762429416179657, "learning_rate": 1.6740230681376867e-06, "elbo": 3.39867, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8160333098756745}
{"step": 99180, "loss": 3.390096664428711, "grad_norm": 0.7083562612533569, "learning_rate": 1.6539299269937192e-06, "elbo": 3.3901, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.816326530612245}
{"step": 99185, "loss": 3.3896141052246094, "grad_norm": 0.6740171313285828, "learning_rate": 1.6339560749717154e-06, "elbo": 3.38961, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.816619751348815}
{"step": 99190, "loss": 3.3878662109375, "grad_norm": 0.6864825487136841, "learning_rate": 1.6141015613551636e-06, "elbo": 3.38787, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.816912972085386}
{"step": 99195, "loss": 3.3635963439941405, "grad_norm": 0.7395142316818237, "learning_rate": 1.59436643513311e-06, "elbo": 3.3636, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.817206192821956}
{"step": 99200, "loss": 3.406157684326172, "grad_norm": 0.6844503283500671, "learning_rate": 1.5747507450000198e-06, "elbo": 3.40616, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.817499413558527}
{"step": 99205, "loss": 3.3872390747070313, "grad_norm": 0.6795655488967896, "learning_rate": 1.555254539355655e-06, "elbo": 3.38724, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.817792634295097}
{"step": 99210, "loss": 3.391602325439453, "grad_norm": 0.7097321152687073, "learning_rate": 1.5358778663049579e-06, "elbo": 3.3916, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.818085855031668}
{"step": 99215, "loss": 3.38134765625, "grad_norm": 0.6769225001335144, "learning_rate": 1.5166207736579562e-06, "elbo": 3.38135, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.818379075768238}
{"step": 99220, "loss": 3.408246612548828, "grad_norm": 0.6742774248123169, "learning_rate": 1.4974833089296037e-06, "elbo": 3.40825, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8186722965048085}
{"step": 99225, "loss": 3.395886993408203, "grad_norm": 0.6760151982307434, "learning_rate": 1.4784655193396946e-06, "elbo": 3.39589, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.818965517241379}
{"step": 99230, "loss": 3.374134063720703, "grad_norm": 0.6838579177856445, "learning_rate": 1.4595674518127268e-06, "elbo": 3.37413, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8192587379779495}
{"step": 99235, "loss": 3.380995178222656, "grad_norm": 0.9503191113471985, "learning_rate": 1.4407891529778172e-06, "elbo": 3.381, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.81955195871452}
{"step": 99240, "loss": 3.373809051513672, "grad_norm": 0.6557364463806152, "learning_rate": 1.4221306691685366e-06, "elbo": 3.37381, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.819845179451091}
{"step": 99245, "loss": 3.373719024658203, "grad_norm": 0.6726007461547852, "learning_rate": 1.4035920464228526e-06, "elbo": 3.37372, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.820138400187661}
{"step": 99250, "loss": 3.399336242675781, "grad_norm": 0.6721754670143127, "learning_rate": 1.385173330482964e-06, "elbo": 3.39934, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.820431620924232}
{"step": 99255, "loss": 3.379867935180664, "grad_norm": 0.6768538355827332, "learning_rate": 1.3668745667952287e-06, "elbo": 3.37987, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.820724841660803}
{"step": 99260, "loss": 3.3675601959228514, "grad_norm": 0.675871729850769, "learning_rate": 1.3486958005100247e-06, "elbo": 3.36756, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.821018062397373}
{"step": 99265, "loss": 3.3992881774902344, "grad_norm": 0.7340127825737, "learning_rate": 1.3306370764816389e-06, "elbo": 3.39929, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8213112831339435}
{"step": 99270, "loss": 3.3955116271972656, "grad_norm": 0.6616964340209961, "learning_rate": 1.3126984392682006e-06, "elbo": 3.39551, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.821604503870514}
{"step": 99275, "loss": 3.3894256591796874, "grad_norm": 0.8426302671432495, "learning_rate": 1.2948799331314932e-06, "elbo": 3.38943, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.821897724607084}
{"step": 99280, "loss": 3.406500244140625, "grad_norm": 0.6694154739379883, "learning_rate": 1.2771816020369253e-06, "elbo": 3.4065, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.822190945343655}
{"step": 99285, "loss": 3.3868885040283203, "grad_norm": 0.7093721032142639, "learning_rate": 1.2596034896533549e-06, "elbo": 3.38689, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.822484166080225}
{"step": 99290, "loss": 3.384485626220703, "grad_norm": 0.661196768283844, "learning_rate": 1.2421456393530372e-06, "elbo": 3.38449, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.822777386816796}
{"step": 99295, "loss": 3.3983131408691407, "grad_norm": 0.6903674006462097, "learning_rate": 1.224808094211477e-06, "elbo": 3.39831, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.823070607553366}
{"step": 99300, "loss": 3.4003326416015627, "grad_norm": 0.7334812879562378, "learning_rate": 1.207590897007349e-06, "elbo": 3.40033, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.823363828289937}
{"step": 99305, "loss": 3.3844272613525392, "grad_norm": 0.6713266372680664, "learning_rate": 1.190494090222366e-06, "elbo": 3.38443, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.823657049026507}
{"step": 99310, "loss": 3.4012592315673826, "grad_norm": 0.6690467596054077, "learning_rate": 1.173517716041217e-06, "elbo": 3.40126, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.823950269763078}
{"step": 99315, "loss": 3.3947769165039063, "grad_norm": 0.6623072028160095, "learning_rate": 1.1566618163513953e-06, "elbo": 3.39478, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.824243490499648}
{"step": 99320, "loss": 3.3758270263671877, "grad_norm": 0.6835634112358093, "learning_rate": 1.1399264327431713e-06, "elbo": 3.37583, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8245367112362185}
{"step": 99325, "loss": 3.3806472778320313, "grad_norm": 0.7108340263366699, "learning_rate": 1.1233116065094362e-06, "elbo": 3.38065, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.824829931972789}
{"step": 99330, "loss": 3.3922164916992186, "grad_norm": 0.6707472205162048, "learning_rate": 1.106817378645636e-06, "elbo": 3.39222, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.825123152709359}
{"step": 99335, "loss": 3.4001399993896486, "grad_norm": 0.7107091546058655, "learning_rate": 1.0904437898496101e-06, "elbo": 3.40014, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.82541637344593}
{"step": 99340, "loss": 3.383405303955078, "grad_norm": 0.7558078765869141, "learning_rate": 1.0741908805215861e-06, "elbo": 3.38341, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8257095941825}
{"step": 99345, "loss": 3.392381286621094, "grad_norm": 0.6673247218132019, "learning_rate": 1.058058690763991e-06, "elbo": 3.39238, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.826002814919071}
{"step": 99350, "loss": 3.386671447753906, "grad_norm": 0.6514536738395691, "learning_rate": 1.042047260381407e-06, "elbo": 3.38667, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.826296035655641}
{"step": 99355, "loss": 3.384840393066406, "grad_norm": 0.7310569286346436, "learning_rate": 1.0261566288804315e-06, "elbo": 3.38484, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.826589256392212}
{"step": 99360, "loss": 3.400217056274414, "grad_norm": 0.8551552891731262, "learning_rate": 1.0103868354696343e-06, "elbo": 3.40022, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.826882477128782}
{"step": 99365, "loss": 3.3957489013671873, "grad_norm": 0.6689968109130859, "learning_rate": 9.947379190594075e-07, "elbo": 3.39575, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8271756978653535}
{"step": 99370, "loss": 3.3761455535888674, "grad_norm": 0.6822552680969238, "learning_rate": 9.792099182618919e-07, "elbo": 3.37615, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.827468918601923}
{"step": 99375, "loss": 3.410289764404297, "grad_norm": 0.6737776398658752, "learning_rate": 9.638028713908897e-07, "elbo": 3.41029, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.827762139338494}
{"step": 99380, "loss": 3.3967617034912108, "grad_norm": 0.7329878807067871, "learning_rate": 9.485168164617642e-07, "elbo": 3.39676, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.828055360075065}
{"step": 99385, "loss": 3.388623809814453, "grad_norm": 0.664669394493103, "learning_rate": 9.333517911913281e-07, "elbo": 3.38862, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.828348580811635}
{"step": 99390, "loss": 3.388001251220703, "grad_norm": 0.6767815947532654, "learning_rate": 9.183078329977723e-07, "elbo": 3.388, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.828641801548206}
{"step": 99395, "loss": 3.387903594970703, "grad_norm": 0.669529378414154, "learning_rate": 9.03384979000571e-07, "elbo": 3.3879, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.828935022284776}
{"step": 99400, "loss": 3.3763858795166017, "grad_norm": 0.7878337502479553, "learning_rate": 8.885832660203875e-07, "elbo": 3.37639, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.829228243021347}
{"step": 99405, "loss": 3.3901004791259766, "grad_norm": 0.6824594736099243, "learning_rate": 8.739027305789683e-07, "elbo": 3.3901, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.829521463757917}
{"step": 99405, "loss": 3.3901004791259766, "grad_norm": 0.6824594736099243, "learning_rate": 8.739027305789683e-07, "elbo": 3.3901, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.829521463757917}
{"step": 99410, "loss": 3.382241439819336, "grad_norm": 0.7140140533447266, "learning_rate": 8.593434088990882e-07, "elbo": 3.38224, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.829814684494488}
{"step": 99410, "loss": 3.382241439819336, "grad_norm": 0.7140140533447266, "learning_rate": 8.593434088990882e-07, "elbo": 3.38224, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.829814684494488}
{"step": 99415, "loss": 3.401720428466797, "grad_norm": 1.0700068473815918, "learning_rate": 8.449053369044058e-07, "elbo": 3.40172, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.830107905231058}
{"step": 99420, "loss": 3.3902664184570312, "grad_norm": 0.8566161394119263, "learning_rate": 8.305885502194577e-07, "elbo": 3.39027, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8304011259676285}
{"step": 99425, "loss": 3.3903854370117186, "grad_norm": 0.6594113707542419, "learning_rate": 8.163930841694589e-07, "elbo": 3.39039, "mask_frac": 0.5012, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.830694346704199}
{"step": 99430, "loss": 3.3870384216308596, "grad_norm": 0.7376753091812134, "learning_rate": 8.02318973780336e-07, "elbo": 3.38704, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.830987567440769}
{"step": 99430, "loss": 3.3870384216308596, "grad_norm": 0.7376753091812134, "learning_rate": 8.02318973780336e-07, "elbo": 3.38704, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.830987567440769}
{"step": 99435, "loss": 3.3895217895507814, "grad_norm": 0.7021214365959167, "learning_rate": 7.883662537785441e-07, "elbo": 3.38952, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.83128078817734}
{"step": 99440, "loss": 3.398085021972656, "grad_norm": 0.6858036518096924, "learning_rate": 7.745349585910389e-07, "elbo": 3.39809, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.83157400891391}
{"step": 99445, "loss": 3.4026126861572266, "grad_norm": 0.6837894916534424, "learning_rate": 7.608251223451602e-07, "elbo": 3.40261, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.831867229650481}
{"step": 99450, "loss": 3.409621810913086, "grad_norm": 1.0188544988632202, "learning_rate": 7.472367788685763e-07, "elbo": 3.40962, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.832160450387051}
{"step": 99455, "loss": 3.3912994384765627, "grad_norm": 0.6577423810958862, "learning_rate": 7.33769961689168e-07, "elbo": 3.3913, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.832453671123622}
{"step": 99460, "loss": 3.372803497314453, "grad_norm": 0.697732150554657, "learning_rate": 7.204247040349665e-07, "elbo": 3.3728, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.832746891860192}
{"step": 99465, "loss": 3.3902801513671874, "grad_norm": 0.6672391891479492, "learning_rate": 7.072010388340655e-07, "elbo": 3.39028, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.833040112596763}
{"step": 99470, "loss": 3.3932403564453124, "grad_norm": 0.7289730906486511, "learning_rate": 6.940989987145485e-07, "elbo": 3.39324, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.833333333333333}
{"step": 99475, "loss": 3.375339889526367, "grad_norm": 0.7255252003669739, "learning_rate": 6.811186160044003e-07, "elbo": 3.37534, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8336265540699035}
{"step": 99480, "loss": 3.4015762329101564, "grad_norm": 0.6894487142562866, "learning_rate": 6.682599227314234e-07, "elbo": 3.40158, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.833919774806474}
{"step": 99485, "loss": 3.363768768310547, "grad_norm": 0.7546783089637756, "learning_rate": 6.555229506231608e-07, "elbo": 3.36377, "mask_frac": 0.4997, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.834212995543044}
{"step": 99490, "loss": 3.393105316162109, "grad_norm": 0.7883924245834351, "learning_rate": 6.429077311068344e-07, "elbo": 3.39311, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.834506216279616}
{"step": 99495, "loss": 3.379067230224609, "grad_norm": 0.6859717965126038, "learning_rate": 6.304142953092284e-07, "elbo": 3.37907, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.834799437016185}
{"step": 99500, "loss": 3.3694934844970703, "grad_norm": 0.7516495585441589, "learning_rate": 6.18042674056668e-07, "elbo": 3.36949, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.835092657752757}
{"step": 99505, "loss": 3.388996887207031, "grad_norm": 0.7007535099983215, "learning_rate": 6.057928978748905e-07, "elbo": 3.389, "mask_frac": 0.4999, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.835385878489327}
{"step": 99510, "loss": 3.385455322265625, "grad_norm": 0.6751935482025146, "learning_rate": 5.936649969890073e-07, "elbo": 3.38546, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.835679099225898}
{"step": 99515, "loss": 3.404938507080078, "grad_norm": 0.66028892993927, "learning_rate": 5.816590013234035e-07, "elbo": 3.40494, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.835972319962468}
{"step": 99520, "loss": 3.3880355834960936, "grad_norm": 0.6860532760620117, "learning_rate": 5.697749405016828e-07, "elbo": 3.38804, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8362655406990385}
{"step": 99525, "loss": 3.3902854919433594, "grad_norm": 0.6904234886169434, "learning_rate": 5.580128438465837e-07, "elbo": 3.39029, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.836558761435609}
{"step": 99530, "loss": 3.3804244995117188, "grad_norm": 0.6947592496871948, "learning_rate": 5.463727403799079e-07, "elbo": 3.38042, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.836851982172179}
{"step": 99535, "loss": 3.379655456542969, "grad_norm": 0.7089954614639282, "learning_rate": 5.348546588224534e-07, "elbo": 3.37966, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.83714520290875}
{"step": 99540, "loss": 3.401600646972656, "grad_norm": 0.6674172878265381, "learning_rate": 5.234586275939424e-07, "elbo": 3.4016, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.83743842364532}
{"step": 99545, "loss": 3.3874286651611327, "grad_norm": 0.6571483016014099, "learning_rate": 5.121846748129544e-07, "elbo": 3.38743, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.837731644381891}
{"step": 99550, "loss": 3.3707462310791017, "grad_norm": 0.6846421360969543, "learning_rate": 5.010328282968435e-07, "elbo": 3.37075, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.838024865118461}
{"step": 99555, "loss": 3.3842327117919924, "grad_norm": 0.6672211289405823, "learning_rate": 4.900031155616769e-07, "elbo": 3.38423, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.838318085855032}
{"step": 99560, "loss": 3.4006404876708984, "grad_norm": 0.9121753573417664, "learning_rate": 4.790955638221795e-07, "elbo": 3.40064, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.838611306591602}
{"step": 99565, "loss": 3.383042907714844, "grad_norm": 0.7502734065055847, "learning_rate": 4.6831019999165617e-07, "elbo": 3.38304, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8389045273281726}
{"step": 99570, "loss": 3.382829284667969, "grad_norm": 0.6868380904197693, "learning_rate": 4.576470506819086e-07, "elbo": 3.38283, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.839197748064743}
{"step": 99575, "loss": 3.4004547119140627, "grad_norm": 0.6848403215408325, "learning_rate": 4.4710614220320746e-07, "elbo": 3.40045, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8394909688013135}
{"step": 99580, "loss": 3.3824657440185546, "grad_norm": 0.6944786906242371, "learning_rate": 4.366875005641924e-07, "elbo": 3.38247, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.839784189537884}
{"step": 99585, "loss": 3.3836082458496093, "grad_norm": 0.8111269474029541, "learning_rate": 4.2639115147182217e-07, "elbo": 3.38361, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.840077410274454}
{"step": 99590, "loss": 3.394729995727539, "grad_norm": 0.6606844663619995, "learning_rate": 4.162171203313192e-07, "elbo": 3.39473, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.840370631011025}
{"step": 99595, "loss": 3.392711639404297, "grad_norm": 0.9145470857620239, "learning_rate": 4.0616543224609726e-07, "elbo": 3.39271, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.840663851747595}
{"step": 99600, "loss": 3.3958465576171877, "grad_norm": 0.7121832966804504, "learning_rate": 3.9623611201769493e-07, "elbo": 3.39585, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.840957072484166}
{"step": 99605, "loss": 3.3774997711181642, "grad_norm": 0.7024653553962708, "learning_rate": 3.864291841457146e-07, "elbo": 3.3775, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.841250293220736}
{"step": 99610, "loss": 3.3960784912109374, "grad_norm": 0.6668214797973633, "learning_rate": 3.76744672827789e-07, "elbo": 3.39608, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.841543513957307}
{"step": 99615, "loss": 3.3877288818359377, "grad_norm": 0.6812378764152527, "learning_rate": 3.67182601959476e-07, "elbo": 3.38773, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.841836734693878}
{"step": 99620, "loss": 3.3833992004394533, "grad_norm": 0.6667124629020691, "learning_rate": 3.5774299513424147e-07, "elbo": 3.3834, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8421299554304476}
{"step": 99625, "loss": 3.3709205627441405, "grad_norm": 0.6601340174674988, "learning_rate": 3.484258756433767e-07, "elbo": 3.37092, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.842423176167019}
{"step": 99630, "loss": 3.3877403259277346, "grad_norm": 0.6370020508766174, "learning_rate": 3.3923126647594785e-07, "elbo": 3.38774, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.842716396903589}
{"step": 99635, "loss": 3.3859893798828127, "grad_norm": 0.8296728134155273, "learning_rate": 3.301591903187351e-07, "elbo": 3.38599, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.84300961764016}
{"step": 99640, "loss": 3.388663482666016, "grad_norm": 0.6712502241134644, "learning_rate": 3.212096695561884e-07, "elbo": 3.38866, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.84330283837673}
{"step": 99645, "loss": 3.3753292083740236, "grad_norm": 0.6812955737113953, "learning_rate": 3.123827262703549e-07, "elbo": 3.37533, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.843596059113301}
{"step": 99650, "loss": 3.3868698120117187, "grad_norm": 0.6628897786140442, "learning_rate": 3.0367838224084623e-07, "elbo": 3.38687, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.843889279849871}
{"step": 99655, "loss": 3.3728790283203125, "grad_norm": 0.6646103858947754, "learning_rate": 2.9509665894476566e-07, "elbo": 3.37288, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.844182500586442}
{"step": 99660, "loss": 3.385675048828125, "grad_norm": 0.6729172468185425, "learning_rate": 2.866375775566643e-07, "elbo": 3.38568, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.844475721323012}
{"step": 99665, "loss": 3.3821247100830076, "grad_norm": 0.6725370287895203, "learning_rate": 2.783011589484741e-07, "elbo": 3.38212, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8447689420595825}
{"step": 99670, "loss": 3.3846366882324217, "grad_norm": 0.679567277431488, "learning_rate": 2.700874236894857e-07, "elbo": 3.38464, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.845062162796153}
{"step": 99675, "loss": 3.3856197357177735, "grad_norm": 0.6300585865974426, "learning_rate": 2.61996392046282e-07, "elbo": 3.38562, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.845355383532723}
{"step": 99680, "loss": 3.38060302734375, "grad_norm": 0.7014582753181458, "learning_rate": 2.540280839826659e-07, "elbo": 3.3806, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.845648604269294}
{"step": 99685, "loss": 3.3854259490966796, "grad_norm": 0.6648176908493042, "learning_rate": 2.4618251915964896e-07, "elbo": 3.38543, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.845941825005864}
{"step": 99690, "loss": 3.390268325805664, "grad_norm": 0.7635915279388428, "learning_rate": 2.3845971693539083e-07, "elbo": 3.39027, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.846235045742435}
{"step": 99695, "loss": 3.3828853607177733, "grad_norm": 0.687385618686676, "learning_rate": 2.3085969636513216e-07, "elbo": 3.38289, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.846528266479005}
{"step": 99700, "loss": 3.3820465087890623, "grad_norm": 0.7318059206008911, "learning_rate": 2.2338247620116714e-07, "elbo": 3.38205, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.846821487215576}
{"step": 99705, "loss": 3.398192596435547, "grad_norm": 0.7119767069816589, "learning_rate": 2.160280748927934e-07, "elbo": 3.39819, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.847114707952146}
{"step": 99710, "loss": 3.3982139587402345, "grad_norm": 0.6867397427558899, "learning_rate": 2.087965105862677e-07, "elbo": 3.39821, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.847407928688717}
{"step": 99715, "loss": 3.390034484863281, "grad_norm": 0.7010703086853027, "learning_rate": 2.0168780112475026e-07, "elbo": 3.39003, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.847701149425287}
{"step": 99720, "loss": 3.3847190856933596, "grad_norm": 0.674843430519104, "learning_rate": 1.9470196404828278e-07, "elbo": 3.38472, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8479943701618575}
{"step": 99725, "loss": 3.3840431213378905, "grad_norm": 0.7429717779159546, "learning_rate": 1.878390165937216e-07, "elbo": 3.38404, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.848287590898428}
{"step": 99730, "loss": 3.3920547485351564, "grad_norm": 0.7458031177520752, "learning_rate": 1.8109897569470457e-07, "elbo": 3.39205, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.848580811634999}
{"step": 99735, "loss": 3.368019866943359, "grad_norm": 0.6744178533554077, "learning_rate": 1.7448185798161764e-07, "elbo": 3.36802, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.848874032371569}
{"step": 99740, "loss": 3.391764831542969, "grad_norm": 0.76072758436203, "learning_rate": 1.6798767978153386e-07, "elbo": 3.39176, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.84916725310814}
{"step": 99745, "loss": 3.389344024658203, "grad_norm": 1.393350601196289, "learning_rate": 1.6161645711819662e-07, "elbo": 3.38934, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.849460473844711}
{"step": 99750, "loss": 3.3695663452148437, "grad_norm": 0.6378800868988037, "learning_rate": 1.553682057119643e-07, "elbo": 3.36957, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.849753694581281}
{"step": 99755, "loss": 3.37685546875, "grad_norm": 0.7285426259040833, "learning_rate": 1.4924294097977686e-07, "elbo": 3.37686, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.850046915317852}
{"step": 99760, "loss": 3.3899467468261717, "grad_norm": 0.7070634365081787, "learning_rate": 1.4324067803511142e-07, "elbo": 3.38995, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.850340136054422}
{"step": 99765, "loss": 3.3789337158203123, "grad_norm": 0.6772640943527222, "learning_rate": 1.3736143168796012e-07, "elbo": 3.37893, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8506333567909925}
{"step": 99770, "loss": 3.3887630462646485, "grad_norm": 0.8287351131439209, "learning_rate": 1.3160521644478008e-07, "elbo": 3.38876, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.850926577527563}
{"step": 99775, "loss": 3.384889602661133, "grad_norm": 0.6595879793167114, "learning_rate": 1.2597204650845463e-07, "elbo": 3.38489, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.851219798264133}
{"step": 99780, "loss": 3.3700904846191406, "grad_norm": 0.6457129716873169, "learning_rate": 1.2046193577827657e-07, "elbo": 3.37009, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.851513019000704}
{"step": 99785, "loss": 3.4048954010009767, "grad_norm": 0.6668005585670471, "learning_rate": 1.1507489784989278e-07, "elbo": 3.4049, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.851806239737274}
{"step": 99790, "loss": 3.380257415771484, "grad_norm": 0.6844305396080017, "learning_rate": 1.0981094601528186e-07, "elbo": 3.38026, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.852099460473845}
{"step": 99795, "loss": 3.3912353515625, "grad_norm": 0.6627162098884583, "learning_rate": 1.0467009326272648e-07, "elbo": 3.39124, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.852392681210415}
{"step": 99800, "loss": 3.393901062011719, "grad_norm": 0.6659650802612305, "learning_rate": 9.965235227676895e-08, "elbo": 3.3939, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.852685901946986}
{"step": 99805, "loss": 3.387664031982422, "grad_norm": 0.6933935880661011, "learning_rate": 9.475773543818344e-08, "elbo": 3.38766, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.852979122683556}
{"step": 99810, "loss": 3.383478546142578, "grad_norm": 0.8422908186912537, "learning_rate": 8.998625482394829e-08, "elbo": 3.38348, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.853272343420127}
{"step": 99815, "loss": 3.384260559082031, "grad_norm": 0.6878654956817627, "learning_rate": 8.53379222072237e-08, "elbo": 3.38426, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.853565564156697}
{"step": 99820, "loss": 3.388111877441406, "grad_norm": 0.6811684966087341, "learning_rate": 8.08127490573074e-08, "elbo": 3.38811, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8538587848932675}
{"step": 99825, "loss": 3.3875377655029295, "grad_norm": 0.7073820233345032, "learning_rate": 7.641074653961244e-08, "elbo": 3.38754, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.854152005629838}
{"step": 99830, "loss": 3.3873886108398437, "grad_norm": 0.6848454475402832, "learning_rate": 7.213192551564496e-08, "elbo": 3.38739, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.854445226366408}
{"step": 99835, "loss": 3.3936866760253905, "grad_norm": 0.6753833293914795, "learning_rate": 6.797629654296534e-08, "elbo": 3.39369, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.854738447102979}
{"step": 99840, "loss": 3.405959701538086, "grad_norm": 0.8155598640441895, "learning_rate": 6.394386987517708e-08, "elbo": 3.40596, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.855031667839549}
{"step": 99845, "loss": 3.375440979003906, "grad_norm": 0.66469407081604, "learning_rate": 6.003465546189358e-08, "elbo": 3.37544, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.85532488857612}
{"step": 99850, "loss": 3.3861244201660154, "grad_norm": 1.105505108833313, "learning_rate": 5.624866294871578e-08, "elbo": 3.38612, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.85561810931269}
{"step": 99855, "loss": 3.3925785064697265, "grad_norm": 0.6605166792869568, "learning_rate": 5.258590167719901e-08, "elbo": 3.39258, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.855911330049262}
{"step": 99860, "loss": 3.3857288360595703, "grad_norm": 0.6588773727416992, "learning_rate": 4.90463806848418e-08, "elbo": 3.38573, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.856204550785831}
{"step": 99860, "loss": 3.3857288360595703, "grad_norm": 0.6588773727416992, "learning_rate": 4.90463806848418e-08, "elbo": 3.38573, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.856204550785831}
{"step": 99865, "loss": 3.3891006469726563, "grad_norm": 0.6836986541748047, "learning_rate": 4.563010870506368e-08, "elbo": 3.3891, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8564977715224025}
{"step": 99870, "loss": 3.386182403564453, "grad_norm": 0.6516706347465515, "learning_rate": 4.233709416717746e-08, "elbo": 3.38618, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.856790992258973}
{"step": 99875, "loss": 3.386862945556641, "grad_norm": 0.6759347319602966, "learning_rate": 3.916734519636145e-08, "elbo": 3.38686, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.857084212995543}
{"step": 99880, "loss": 3.396199035644531, "grad_norm": 0.6875867247581482, "learning_rate": 3.612086961366501e-08, "elbo": 3.3962, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.857377433732114}
{"step": 99885, "loss": 3.390374755859375, "grad_norm": 0.979158878326416, "learning_rate": 3.31976749359586e-08, "elbo": 3.39037, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.857670654468684}
{"step": 99890, "loss": 3.3991870880126953, "grad_norm": 0.7779688835144043, "learning_rate": 3.0397768375939325e-08, "elbo": 3.39919, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.857963875205255}
{"step": 99895, "loss": 3.4064449310302733, "grad_norm": 0.8334016799926758, "learning_rate": 2.772115684209209e-08, "elbo": 3.40644, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.858257095941825}
{"step": 99900, "loss": 3.393933868408203, "grad_norm": 0.6887803077697754, "learning_rate": 2.5167846938689567e-08, "elbo": 3.39393, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.858550316678396}
{"step": 99905, "loss": 3.3970176696777346, "grad_norm": 0.6616532802581787, "learning_rate": 2.273784496577558e-08, "elbo": 3.39702, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.858843537414966}
{"step": 99910, "loss": 3.3756553649902346, "grad_norm": 0.6662558913230896, "learning_rate": 2.0431156919137328e-08, "elbo": 3.37566, "mask_frac": 0.5004, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.859136758151537}
{"step": 99915, "loss": 3.388253021240234, "grad_norm": 0.6972927451133728, "learning_rate": 1.8247788490299844e-08, "elbo": 3.38825, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.859429978888107}
{"step": 99920, "loss": 3.372264862060547, "grad_norm": 0.7969528436660767, "learning_rate": 1.6187745066498228e-08, "elbo": 3.37227, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8597231996246775}
{"step": 99925, "loss": 3.3903079986572267, "grad_norm": 0.6446943879127502, "learning_rate": 1.425103173069986e-08, "elbo": 3.39031, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.860016420361248}
{"step": 99930, "loss": 3.3881717681884767, "grad_norm": 0.7236632704734802, "learning_rate": 1.2437653261537785e-08, "elbo": 3.38817, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.860309641097818}
{"step": 99935, "loss": 3.383544921875, "grad_norm": 0.6622477173805237, "learning_rate": 1.074761413334957e-08, "elbo": 3.38354, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.860602861834389}
{"step": 99940, "loss": 3.3976203918457033, "grad_norm": 0.6858606338500977, "learning_rate": 9.180918516138448e-09, "elbo": 3.39762, "mask_frac": 0.5005, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.860896082570959}
{"step": 99945, "loss": 3.3892585754394533, "grad_norm": 0.7657207250595093, "learning_rate": 7.737570275573313e-09, "elbo": 3.38926, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.86118930330753}
{"step": 99950, "loss": 3.3817070007324217, "grad_norm": 0.8129560351371765, "learning_rate": 6.417572972966524e-09, "elbo": 3.38171, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8614825240441}
{"step": 99955, "loss": 3.3874637603759767, "grad_norm": 0.6751328706741333, "learning_rate": 5.220929865284996e-09, "elbo": 3.38746, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.861775744780671}
{"step": 99960, "loss": 3.3844768524169924, "grad_norm": 0.6705267429351807, "learning_rate": 4.147643905122456e-09, "elbo": 3.38448, "mask_frac": 0.501, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.862068965517241}
{"step": 99965, "loss": 3.378167724609375, "grad_norm": 0.7284144759178162, "learning_rate": 3.1977177407105372e-09, "elbo": 3.37817, "mask_frac": 0.5009, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.862362186253812}
{"step": 99970, "loss": 3.3814132690429686, "grad_norm": 0.6924079060554504, "learning_rate": 2.371153715891028e-09, "elbo": 3.38141, "mask_frac": 0.5, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.862655406990382}
{"step": 99975, "loss": 3.4011157989501952, "grad_norm": 0.7199167609214783, "learning_rate": 1.6679538701325215e-09, "elbo": 3.40112, "mask_frac": 0.5007, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.8629486277269525}
{"step": 99980, "loss": 3.369149017333984, "grad_norm": 0.6546038389205933, "learning_rate": 1.0881199385137652e-09, "elbo": 3.36915, "mask_frac": 0.5001, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.863241848463524}
{"step": 99985, "loss": 3.404438018798828, "grad_norm": 0.6430757641792297, "learning_rate": 6.316533517125578e-10, "elbo": 3.40444, "mask_frac": 0.5008, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.863535069200093}
{"step": 99990, "loss": 3.3729515075683594, "grad_norm": 0.7987287044525146, "learning_rate": 2.985552360168509e-10, "elbo": 3.37295, "mask_frac": 0.5002, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.863828289936665}
{"step": 99995, "loss": 3.38281135559082, "grad_norm": 0.7172755002975464, "learning_rate": 8.882641330809626e-11, "elbo": 3.38281, "mask_frac": 0.5003, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.864121510673235}
{"step": 100000, "loss": 3.3929725646972657, "grad_norm": 0.7117201089859009, "learning_rate": 2.4674010778991117e-12, "elbo": 3.39297, "mask_frac": 0.5006, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.864414731409806}
{"step": 100000, "train_runtime": 19172.0797, "train_samples_per_second": 2670.55, "train_steps_per_second": 5.216, "total_flos": 2.67246555365376e+19, "train_loss": 3.7316875759124755, "mem_alloc_gib": 49.7, "mem_res_gib": 50.0, "epoch": 5.864414731409806}