{ "best_metric": 0.9591836734693877, "best_model_checkpoint": "./model_temp/checkpoint-101000", "epoch": 48.69916027747353, "eval_steps": 500, "global_step": 134000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.08305647840531562, "grad_norm": 7.692755222320557, "learning_rate": 5.000000000000001e-07, "loss": 0.5721, "step": 50 }, { "epoch": 0.16611295681063123, "grad_norm": 0.5308930277824402, "learning_rate": 1.0000000000000002e-06, "loss": 0.119, "step": 100 }, { "epoch": 0.24916943521594684, "grad_norm": 0.05565977841615677, "learning_rate": 1.5e-06, "loss": 0.0306, "step": 150 }, { "epoch": 0.33222591362126247, "grad_norm": 0.0325307697057724, "learning_rate": 2.0000000000000003e-06, "loss": 0.0272, "step": 200 }, { "epoch": 0.4152823920265781, "grad_norm": 0.057338133454322815, "learning_rate": 2.5e-06, "loss": 0.0274, "step": 250 }, { "epoch": 0.4983388704318937, "grad_norm": 0.021708570420742035, "learning_rate": 3e-06, "loss": 0.0272, "step": 300 }, { "epoch": 0.5813953488372093, "grad_norm": 0.029700901359319687, "learning_rate": 3.5e-06, "loss": 0.0275, "step": 350 }, { "epoch": 0.6644518272425249, "grad_norm": 0.025190476328134537, "learning_rate": 4.000000000000001e-06, "loss": 0.0263, "step": 400 }, { "epoch": 0.7475083056478405, "grad_norm": 0.03320654481649399, "learning_rate": 4.5e-06, "loss": 0.0274, "step": 450 }, { "epoch": 0.8305647840531561, "grad_norm": 0.03724764287471771, "learning_rate": 5e-06, "loss": 0.0262, "step": 500 }, { "epoch": 0.9136212624584718, "grad_norm": 0.04773814603686333, "learning_rate": 5.500000000000001e-06, "loss": 0.0265, "step": 550 }, { "epoch": 0.9966777408637874, "grad_norm": 0.05851120129227638, "learning_rate": 6e-06, "loss": 0.0264, "step": 600 }, { "epoch": 1.079734219269103, "grad_norm": 0.044609878212213516, "learning_rate": 6.5000000000000004e-06, "loss": 0.0266, "step": 650 }, { "epoch": 1.1627906976744187, "grad_norm": 0.048703815788030624, "learning_rate": 7e-06, "loss": 0.0261, "step": 700 }, { "epoch": 1.2458471760797343, "grad_norm": 0.032963819801807404, "learning_rate": 7.500000000000001e-06, "loss": 0.0254, "step": 750 }, { "epoch": 1.3289036544850499, "grad_norm": 0.061456166207790375, "learning_rate": 8.000000000000001e-06, "loss": 0.0261, "step": 800 }, { "epoch": 1.4119601328903655, "grad_norm": 0.04576973244547844, "learning_rate": 8.5e-06, "loss": 0.0244, "step": 850 }, { "epoch": 1.495016611295681, "grad_norm": 0.04015668109059334, "learning_rate": 9e-06, "loss": 0.0252, "step": 900 }, { "epoch": 1.5780730897009967, "grad_norm": 0.055900804698467255, "learning_rate": 9.5e-06, "loss": 0.025, "step": 950 }, { "epoch": 1.6611295681063123, "grad_norm": 0.03412498161196709, "learning_rate": 1e-05, "loss": 0.0252, "step": 1000 }, { "epoch": 1.744186046511628, "grad_norm": 0.047917820513248444, "learning_rate": 9.991554054054055e-06, "loss": 0.0249, "step": 1050 }, { "epoch": 1.8272425249169435, "grad_norm": 0.0429849848151207, "learning_rate": 9.983108108108109e-06, "loss": 0.0247, "step": 1100 }, { "epoch": 1.910299003322259, "grad_norm": 0.043424028903245926, "learning_rate": 9.974662162162163e-06, "loss": 0.0249, "step": 1150 }, { "epoch": 1.9933554817275747, "grad_norm": 0.04757978767156601, "learning_rate": 9.966216216216217e-06, "loss": 0.0251, "step": 1200 }, { "epoch": 2.0764119601328903, "grad_norm": 0.042466793209314346, "learning_rate": 9.957770270270271e-06, "loss": 0.0247, "step": 1250 }, { "epoch": 2.159468438538206, "grad_norm": 0.04843801632523537, "learning_rate": 9.949324324324325e-06, "loss": 0.0245, "step": 1300 }, { "epoch": 2.2425249169435215, "grad_norm": 0.05055597424507141, "learning_rate": 9.940878378378379e-06, "loss": 0.0242, "step": 1350 }, { "epoch": 2.3255813953488373, "grad_norm": 0.03343519568443298, "learning_rate": 9.932432432432433e-06, "loss": 0.0245, "step": 1400 }, { "epoch": 2.4086378737541527, "grad_norm": 0.044828835874795914, "learning_rate": 9.923986486486487e-06, "loss": 0.0243, "step": 1450 }, { "epoch": 2.4916943521594686, "grad_norm": 0.03191625326871872, "learning_rate": 9.915540540540541e-06, "loss": 0.0244, "step": 1500 }, { "epoch": 2.574750830564784, "grad_norm": 0.06024698168039322, "learning_rate": 9.907094594594595e-06, "loss": 0.0238, "step": 1550 }, { "epoch": 2.6578073089700998, "grad_norm": 0.04001982510089874, "learning_rate": 9.89864864864865e-06, "loss": 0.0244, "step": 1600 }, { "epoch": 2.740863787375415, "grad_norm": 0.1221691444516182, "learning_rate": 9.890202702702703e-06, "loss": 0.0239, "step": 1650 }, { "epoch": 2.823920265780731, "grad_norm": 0.04491797834634781, "learning_rate": 9.881756756756757e-06, "loss": 0.0235, "step": 1700 }, { "epoch": 2.9069767441860463, "grad_norm": 0.08072737604379654, "learning_rate": 9.873310810810811e-06, "loss": 0.0235, "step": 1750 }, { "epoch": 2.990033222591362, "grad_norm": 0.12076549977064133, "learning_rate": 9.864864864864865e-06, "loss": 0.0231, "step": 1800 }, { "epoch": 3.0730897009966776, "grad_norm": 0.047267742455005646, "learning_rate": 9.85641891891892e-06, "loss": 0.0237, "step": 1850 }, { "epoch": 3.1561461794019934, "grad_norm": 0.04982184246182442, "learning_rate": 9.847972972972973e-06, "loss": 0.0236, "step": 1900 }, { "epoch": 3.2392026578073088, "grad_norm": 0.025792259722948074, "learning_rate": 9.839527027027027e-06, "loss": 0.0236, "step": 1950 }, { "epoch": 3.3222591362126246, "grad_norm": 0.08177588880062103, "learning_rate": 9.831081081081081e-06, "loss": 0.0226, "step": 2000 }, { "epoch": 3.4053156146179404, "grad_norm": 0.04007701203227043, "learning_rate": 9.822635135135135e-06, "loss": 0.0238, "step": 2050 }, { "epoch": 3.488372093023256, "grad_norm": 0.060702063143253326, "learning_rate": 9.814189189189191e-06, "loss": 0.0238, "step": 2100 }, { "epoch": 3.571428571428571, "grad_norm": 0.05167702957987785, "learning_rate": 9.805743243243244e-06, "loss": 0.0232, "step": 2150 }, { "epoch": 3.654485049833887, "grad_norm": 0.060260944068431854, "learning_rate": 9.797297297297298e-06, "loss": 0.0234, "step": 2200 }, { "epoch": 3.737541528239203, "grad_norm": 0.04363139718770981, "learning_rate": 9.788851351351352e-06, "loss": 0.0236, "step": 2250 }, { "epoch": 3.820598006644518, "grad_norm": 0.07498718798160553, "learning_rate": 9.780405405405407e-06, "loss": 0.0232, "step": 2300 }, { "epoch": 3.903654485049834, "grad_norm": 0.031206348910927773, "learning_rate": 9.77195945945946e-06, "loss": 0.0233, "step": 2350 }, { "epoch": 3.9867109634551494, "grad_norm": 0.09429947286844254, "learning_rate": 9.763513513513514e-06, "loss": 0.0236, "step": 2400 }, { "epoch": 4.069767441860465, "grad_norm": 0.04899977147579193, "learning_rate": 9.75506756756757e-06, "loss": 0.0229, "step": 2450 }, { "epoch": 4.152823920265781, "grad_norm": 0.0407995842397213, "learning_rate": 9.746621621621622e-06, "loss": 0.0226, "step": 2500 }, { "epoch": 4.235880398671096, "grad_norm": 0.08796069771051407, "learning_rate": 9.738175675675676e-06, "loss": 0.0234, "step": 2550 }, { "epoch": 4.318936877076412, "grad_norm": 0.05589350312948227, "learning_rate": 9.729729729729732e-06, "loss": 0.0228, "step": 2600 }, { "epoch": 4.401993355481728, "grad_norm": 0.05050768330693245, "learning_rate": 9.721283783783786e-06, "loss": 0.0229, "step": 2650 }, { "epoch": 4.485049833887043, "grad_norm": 0.056809961795806885, "learning_rate": 9.712837837837838e-06, "loss": 0.0231, "step": 2700 }, { "epoch": 4.568106312292358, "grad_norm": 0.04864858463406563, "learning_rate": 9.704391891891892e-06, "loss": 0.0227, "step": 2750 }, { "epoch": 4.651162790697675, "grad_norm": 0.04045893996953964, "learning_rate": 9.695945945945948e-06, "loss": 0.0232, "step": 2800 }, { "epoch": 4.73421926910299, "grad_norm": 0.057447399944067, "learning_rate": 9.6875e-06, "loss": 0.023, "step": 2850 }, { "epoch": 4.8172757475083055, "grad_norm": 0.046988341957330704, "learning_rate": 9.679054054054054e-06, "loss": 0.0229, "step": 2900 }, { "epoch": 4.900332225913621, "grad_norm": 0.049161553382873535, "learning_rate": 9.67060810810811e-06, "loss": 0.0235, "step": 2950 }, { "epoch": 4.983388704318937, "grad_norm": 0.06303807348012924, "learning_rate": 9.662162162162164e-06, "loss": 0.0231, "step": 3000 }, { "epoch": 5.0664451827242525, "grad_norm": 0.062485501170158386, "learning_rate": 9.653716216216216e-06, "loss": 0.022, "step": 3050 }, { "epoch": 5.149501661129568, "grad_norm": 0.06548245251178741, "learning_rate": 9.64527027027027e-06, "loss": 0.0228, "step": 3100 }, { "epoch": 5.232558139534884, "grad_norm": 0.05610573664307594, "learning_rate": 9.636824324324326e-06, "loss": 0.023, "step": 3150 }, { "epoch": 5.3156146179401995, "grad_norm": 0.06676553189754486, "learning_rate": 9.62837837837838e-06, "loss": 0.0222, "step": 3200 }, { "epoch": 5.398671096345515, "grad_norm": 0.042792513966560364, "learning_rate": 9.619932432432432e-06, "loss": 0.0223, "step": 3250 }, { "epoch": 5.48172757475083, "grad_norm": 0.10044345259666443, "learning_rate": 9.611486486486488e-06, "loss": 0.0221, "step": 3300 }, { "epoch": 5.5647840531561465, "grad_norm": 0.08114602416753769, "learning_rate": 9.603040540540542e-06, "loss": 0.022, "step": 3350 }, { "epoch": 5.647840531561462, "grad_norm": 0.08317956328392029, "learning_rate": 9.594594594594594e-06, "loss": 0.022, "step": 3400 }, { "epoch": 5.730897009966777, "grad_norm": 0.11240950226783752, "learning_rate": 9.58614864864865e-06, "loss": 0.0225, "step": 3450 }, { "epoch": 5.813953488372093, "grad_norm": 0.10367760807275772, "learning_rate": 9.577702702702704e-06, "loss": 0.0218, "step": 3500 }, { "epoch": 5.897009966777409, "grad_norm": 0.10737499594688416, "learning_rate": 9.569256756756758e-06, "loss": 0.0216, "step": 3550 }, { "epoch": 5.980066445182724, "grad_norm": 0.14091305434703827, "learning_rate": 9.56081081081081e-06, "loss": 0.0206, "step": 3600 }, { "epoch": 6.06312292358804, "grad_norm": 0.1119067594408989, "learning_rate": 9.552364864864866e-06, "loss": 0.0195, "step": 3650 }, { "epoch": 6.146179401993355, "grad_norm": 0.1258949488401413, "learning_rate": 9.54391891891892e-06, "loss": 0.0188, "step": 3700 }, { "epoch": 6.229235880398671, "grad_norm": 0.12416691333055496, "learning_rate": 9.535472972972973e-06, "loss": 0.0176, "step": 3750 }, { "epoch": 6.312292358803987, "grad_norm": 0.1567380279302597, "learning_rate": 9.527027027027028e-06, "loss": 0.0162, "step": 3800 }, { "epoch": 6.395348837209302, "grad_norm": 0.1346510350704193, "learning_rate": 9.518581081081082e-06, "loss": 0.0154, "step": 3850 }, { "epoch": 6.4784053156146175, "grad_norm": 0.13163822889328003, "learning_rate": 9.510135135135136e-06, "loss": 0.0151, "step": 3900 }, { "epoch": 6.561461794019934, "grad_norm": 0.141840398311615, "learning_rate": 9.50168918918919e-06, "loss": 0.0148, "step": 3950 }, { "epoch": 6.644518272425249, "grad_norm": 0.12375494092702866, "learning_rate": 9.493243243243244e-06, "loss": 0.0143, "step": 4000 }, { "epoch": 6.7275747508305646, "grad_norm": 0.1005747988820076, "learning_rate": 9.484797297297298e-06, "loss": 0.014, "step": 4050 }, { "epoch": 6.810631229235881, "grad_norm": 0.10704457014799118, "learning_rate": 9.476351351351352e-06, "loss": 0.0137, "step": 4100 }, { "epoch": 6.893687707641196, "grad_norm": 0.10846409201622009, "learning_rate": 9.467905405405407e-06, "loss": 0.0141, "step": 4150 }, { "epoch": 6.976744186046512, "grad_norm": 0.15824003517627716, "learning_rate": 9.45945945945946e-06, "loss": 0.0135, "step": 4200 }, { "epoch": 7.059800664451827, "grad_norm": 0.12364927679300308, "learning_rate": 9.451013513513515e-06, "loss": 0.0126, "step": 4250 }, { "epoch": 7.142857142857143, "grad_norm": 0.10351550579071045, "learning_rate": 9.442567567567569e-06, "loss": 0.0128, "step": 4300 }, { "epoch": 7.225913621262459, "grad_norm": 0.1284281462430954, "learning_rate": 9.434121621621623e-06, "loss": 0.0123, "step": 4350 }, { "epoch": 7.308970099667774, "grad_norm": 0.11880920082330704, "learning_rate": 9.425675675675677e-06, "loss": 0.012, "step": 4400 }, { "epoch": 7.392026578073089, "grad_norm": 0.11804422736167908, "learning_rate": 9.41722972972973e-06, "loss": 0.0118, "step": 4450 }, { "epoch": 7.475083056478406, "grad_norm": 0.13938409090042114, "learning_rate": 9.408783783783785e-06, "loss": 0.0124, "step": 4500 }, { "epoch": 7.558139534883721, "grad_norm": 0.11719731986522675, "learning_rate": 9.400337837837839e-06, "loss": 0.0117, "step": 4550 }, { "epoch": 7.641196013289036, "grad_norm": 0.11749731004238129, "learning_rate": 9.391891891891893e-06, "loss": 0.0122, "step": 4600 }, { "epoch": 7.724252491694352, "grad_norm": 0.08036090433597565, "learning_rate": 9.383445945945947e-06, "loss": 0.0115, "step": 4650 }, { "epoch": 7.807308970099668, "grad_norm": 0.12178517878055573, "learning_rate": 9.375000000000001e-06, "loss": 0.0118, "step": 4700 }, { "epoch": 7.8903654485049834, "grad_norm": 0.1416023224592209, "learning_rate": 9.366554054054055e-06, "loss": 0.0118, "step": 4750 }, { "epoch": 7.973421926910299, "grad_norm": 0.14340586960315704, "learning_rate": 9.358108108108109e-06, "loss": 0.0115, "step": 4800 }, { "epoch": 8.056478405315614, "grad_norm": 0.1212538629770279, "learning_rate": 9.349662162162163e-06, "loss": 0.0109, "step": 4850 }, { "epoch": 8.13953488372093, "grad_norm": 0.08581560105085373, "learning_rate": 9.341216216216217e-06, "loss": 0.01, "step": 4900 }, { "epoch": 8.222591362126245, "grad_norm": 0.09330979734659195, "learning_rate": 9.332770270270271e-06, "loss": 0.0107, "step": 4950 }, { "epoch": 8.305647840531561, "grad_norm": 0.15124256908893585, "learning_rate": 9.324324324324325e-06, "loss": 0.0108, "step": 5000 }, { "epoch": 8.388704318936878, "grad_norm": 0.12136714160442352, "learning_rate": 9.315878378378379e-06, "loss": 0.0108, "step": 5050 }, { "epoch": 8.471760797342192, "grad_norm": 0.09980884194374084, "learning_rate": 9.307432432432433e-06, "loss": 0.0101, "step": 5100 }, { "epoch": 8.554817275747508, "grad_norm": 0.11205988377332687, "learning_rate": 9.298986486486487e-06, "loss": 0.0107, "step": 5150 }, { "epoch": 8.637873754152825, "grad_norm": 0.1103920042514801, "learning_rate": 9.290540540540541e-06, "loss": 0.0107, "step": 5200 }, { "epoch": 8.720930232558139, "grad_norm": 0.10186067223548889, "learning_rate": 9.282094594594595e-06, "loss": 0.0108, "step": 5250 }, { "epoch": 8.803986710963455, "grad_norm": 0.1278827041387558, "learning_rate": 9.27364864864865e-06, "loss": 0.0108, "step": 5300 }, { "epoch": 8.887043189368772, "grad_norm": 0.08402027189731598, "learning_rate": 9.265202702702703e-06, "loss": 0.0105, "step": 5350 }, { "epoch": 8.970099667774086, "grad_norm": 0.14129339158535004, "learning_rate": 9.256756756756757e-06, "loss": 0.0104, "step": 5400 }, { "epoch": 9.053156146179402, "grad_norm": 0.1324179470539093, "learning_rate": 9.248310810810811e-06, "loss": 0.01, "step": 5450 }, { "epoch": 9.136212624584717, "grad_norm": 0.09467537701129913, "learning_rate": 9.239864864864865e-06, "loss": 0.0095, "step": 5500 }, { "epoch": 9.219269102990033, "grad_norm": 0.10648459941148758, "learning_rate": 9.23141891891892e-06, "loss": 0.0098, "step": 5550 }, { "epoch": 9.30232558139535, "grad_norm": 0.11457883566617966, "learning_rate": 9.222972972972973e-06, "loss": 0.0095, "step": 5600 }, { "epoch": 9.385382059800664, "grad_norm": 0.1026664674282074, "learning_rate": 9.214527027027027e-06, "loss": 0.0092, "step": 5650 }, { "epoch": 9.46843853820598, "grad_norm": 0.10952220112085342, "learning_rate": 9.206081081081082e-06, "loss": 0.0098, "step": 5700 }, { "epoch": 9.551495016611296, "grad_norm": 0.12173211574554443, "learning_rate": 9.197635135135136e-06, "loss": 0.0099, "step": 5750 }, { "epoch": 9.634551495016611, "grad_norm": 0.14872153103351593, "learning_rate": 9.189189189189191e-06, "loss": 0.0098, "step": 5800 }, { "epoch": 9.717607973421927, "grad_norm": 0.11462219059467316, "learning_rate": 9.180743243243244e-06, "loss": 0.0098, "step": 5850 }, { "epoch": 9.800664451827242, "grad_norm": 0.0995425134897232, "learning_rate": 9.172297297297298e-06, "loss": 0.0096, "step": 5900 }, { "epoch": 9.883720930232558, "grad_norm": 0.14007925987243652, "learning_rate": 9.163851351351352e-06, "loss": 0.0095, "step": 5950 }, { "epoch": 9.966777408637874, "grad_norm": 0.1499241441488266, "learning_rate": 9.155405405405406e-06, "loss": 0.0097, "step": 6000 }, { "epoch": 10.049833887043189, "grad_norm": 0.10661840438842773, "learning_rate": 9.14695945945946e-06, "loss": 0.0092, "step": 6050 }, { "epoch": 10.132890365448505, "grad_norm": 0.1181366890668869, "learning_rate": 9.138513513513514e-06, "loss": 0.0086, "step": 6100 }, { "epoch": 10.215946843853821, "grad_norm": 0.14740252494812012, "learning_rate": 9.13006756756757e-06, "loss": 0.0089, "step": 6150 }, { "epoch": 10.299003322259136, "grad_norm": 0.19841988384723663, "learning_rate": 9.121621621621622e-06, "loss": 0.0086, "step": 6200 }, { "epoch": 10.382059800664452, "grad_norm": 0.08790756016969681, "learning_rate": 9.113175675675676e-06, "loss": 0.009, "step": 6250 }, { "epoch": 10.465116279069768, "grad_norm": 0.103313148021698, "learning_rate": 9.104729729729732e-06, "loss": 0.0088, "step": 6300 }, { "epoch": 10.548172757475083, "grad_norm": 0.11116333305835724, "learning_rate": 9.096283783783784e-06, "loss": 0.0092, "step": 6350 }, { "epoch": 10.631229235880399, "grad_norm": 0.10039646178483963, "learning_rate": 9.087837837837838e-06, "loss": 0.0091, "step": 6400 }, { "epoch": 10.714285714285714, "grad_norm": 0.09815255552530289, "learning_rate": 9.079391891891892e-06, "loss": 0.0091, "step": 6450 }, { "epoch": 10.79734219269103, "grad_norm": 0.10742838680744171, "learning_rate": 9.070945945945948e-06, "loss": 0.0093, "step": 6500 }, { "epoch": 10.880398671096346, "grad_norm": 0.12387531250715256, "learning_rate": 9.0625e-06, "loss": 0.009, "step": 6550 }, { "epoch": 10.96345514950166, "grad_norm": 0.10993634909391403, "learning_rate": 9.054054054054054e-06, "loss": 0.0089, "step": 6600 }, { "epoch": 11.046511627906977, "grad_norm": 0.13742654025554657, "learning_rate": 9.04560810810811e-06, "loss": 0.0086, "step": 6650 }, { "epoch": 11.129568106312293, "grad_norm": 0.09748993813991547, "learning_rate": 9.037162162162164e-06, "loss": 0.008, "step": 6700 }, { "epoch": 11.212624584717608, "grad_norm": 0.1164650246500969, "learning_rate": 9.028716216216216e-06, "loss": 0.0083, "step": 6750 }, { "epoch": 11.295681063122924, "grad_norm": 0.10681592673063278, "learning_rate": 9.02027027027027e-06, "loss": 0.0085, "step": 6800 }, { "epoch": 11.378737541528238, "grad_norm": 0.09350414574146271, "learning_rate": 9.011824324324326e-06, "loss": 0.0083, "step": 6850 }, { "epoch": 11.461794019933555, "grad_norm": 0.13445159792900085, "learning_rate": 9.003378378378378e-06, "loss": 0.0088, "step": 6900 }, { "epoch": 11.544850498338871, "grad_norm": 0.11919626593589783, "learning_rate": 8.994932432432432e-06, "loss": 0.0081, "step": 6950 }, { "epoch": 11.627906976744185, "grad_norm": 0.1174791157245636, "learning_rate": 8.986486486486488e-06, "loss": 0.0085, "step": 7000 }, { "epoch": 11.710963455149502, "grad_norm": 0.11531531810760498, "learning_rate": 8.978040540540542e-06, "loss": 0.008, "step": 7050 }, { "epoch": 11.794019933554818, "grad_norm": 0.14403431117534637, "learning_rate": 8.969594594594594e-06, "loss": 0.0081, "step": 7100 }, { "epoch": 11.877076411960132, "grad_norm": 0.1350933164358139, "learning_rate": 8.96114864864865e-06, "loss": 0.0084, "step": 7150 }, { "epoch": 11.960132890365449, "grad_norm": 0.12807510793209076, "learning_rate": 8.952702702702704e-06, "loss": 0.0081, "step": 7200 }, { "epoch": 12.043189368770765, "grad_norm": 0.12483798712491989, "learning_rate": 8.944256756756756e-06, "loss": 0.0076, "step": 7250 }, { "epoch": 12.12624584717608, "grad_norm": 0.09776129573583603, "learning_rate": 8.93581081081081e-06, "loss": 0.0076, "step": 7300 }, { "epoch": 12.209302325581396, "grad_norm": 0.1335456818342209, "learning_rate": 8.927364864864866e-06, "loss": 0.0075, "step": 7350 }, { "epoch": 12.29235880398671, "grad_norm": 0.11351911723613739, "learning_rate": 8.91891891891892e-06, "loss": 0.0077, "step": 7400 }, { "epoch": 12.375415282392026, "grad_norm": 0.10579736530780792, "learning_rate": 8.910472972972973e-06, "loss": 0.0078, "step": 7450 }, { "epoch": 12.458471760797343, "grad_norm": 0.10290402173995972, "learning_rate": 8.902027027027028e-06, "loss": 0.0079, "step": 7500 }, { "epoch": 12.541528239202657, "grad_norm": 0.08668718487024307, "learning_rate": 8.893581081081082e-06, "loss": 0.0076, "step": 7550 }, { "epoch": 12.624584717607974, "grad_norm": 0.11149685084819794, "learning_rate": 8.885135135135136e-06, "loss": 0.0077, "step": 7600 }, { "epoch": 12.70764119601329, "grad_norm": 0.1043517217040062, "learning_rate": 8.87668918918919e-06, "loss": 0.0077, "step": 7650 }, { "epoch": 12.790697674418604, "grad_norm": 0.12131992727518082, "learning_rate": 8.868243243243244e-06, "loss": 0.0077, "step": 7700 }, { "epoch": 12.87375415282392, "grad_norm": 0.12205377221107483, "learning_rate": 8.859797297297299e-06, "loss": 0.0077, "step": 7750 }, { "epoch": 12.956810631229235, "grad_norm": 0.14293748140335083, "learning_rate": 8.851351351351351e-06, "loss": 0.0075, "step": 7800 }, { "epoch": 13.039867109634551, "grad_norm": 0.10758055746555328, "learning_rate": 8.842905405405407e-06, "loss": 0.0076, "step": 7850 }, { "epoch": 13.122923588039868, "grad_norm": 0.13439467549324036, "learning_rate": 8.83445945945946e-06, "loss": 0.0073, "step": 7900 }, { "epoch": 13.205980066445182, "grad_norm": 0.11394330114126205, "learning_rate": 8.826013513513515e-06, "loss": 0.0073, "step": 7950 }, { "epoch": 13.289036544850498, "grad_norm": 0.087078757584095, "learning_rate": 8.817567567567569e-06, "loss": 0.0069, "step": 8000 }, { "epoch": 13.372093023255815, "grad_norm": 0.12730620801448822, "learning_rate": 8.809121621621623e-06, "loss": 0.007, "step": 8050 }, { "epoch": 13.455149501661129, "grad_norm": 0.12950794398784637, "learning_rate": 8.800675675675677e-06, "loss": 0.0071, "step": 8100 }, { "epoch": 13.538205980066445, "grad_norm": 0.14103956520557404, "learning_rate": 8.79222972972973e-06, "loss": 0.0071, "step": 8150 }, { "epoch": 13.621262458471762, "grad_norm": 0.13010290265083313, "learning_rate": 8.783783783783785e-06, "loss": 0.0068, "step": 8200 }, { "epoch": 13.704318936877076, "grad_norm": 0.11448788642883301, "learning_rate": 8.775337837837839e-06, "loss": 0.0068, "step": 8250 }, { "epoch": 13.787375415282392, "grad_norm": 0.10770177096128464, "learning_rate": 8.766891891891893e-06, "loss": 0.0069, "step": 8300 }, { "epoch": 13.870431893687707, "grad_norm": 0.1083039715886116, "learning_rate": 8.758445945945947e-06, "loss": 0.007, "step": 8350 }, { "epoch": 13.953488372093023, "grad_norm": 0.09664791077375412, "learning_rate": 8.750000000000001e-06, "loss": 0.0074, "step": 8400 }, { "epoch": 14.03654485049834, "grad_norm": 0.21872323751449585, "learning_rate": 8.741554054054055e-06, "loss": 0.007, "step": 8450 }, { "epoch": 14.119601328903654, "grad_norm": 0.12858127057552338, "learning_rate": 8.733108108108109e-06, "loss": 0.0063, "step": 8500 }, { "epoch": 14.20265780730897, "grad_norm": 0.1704428791999817, "learning_rate": 8.724662162162163e-06, "loss": 0.0065, "step": 8550 }, { "epoch": 14.285714285714286, "grad_norm": 0.13034966588020325, "learning_rate": 8.716216216216217e-06, "loss": 0.0063, "step": 8600 }, { "epoch": 14.368770764119601, "grad_norm": 0.09468378126621246, "learning_rate": 8.707770270270271e-06, "loss": 0.0067, "step": 8650 }, { "epoch": 14.451827242524917, "grad_norm": 0.18818555772304535, "learning_rate": 8.699324324324325e-06, "loss": 0.0063, "step": 8700 }, { "epoch": 14.534883720930232, "grad_norm": 0.10597342997789383, "learning_rate": 8.690878378378379e-06, "loss": 0.0069, "step": 8750 }, { "epoch": 14.617940199335548, "grad_norm": 0.15204182267189026, "learning_rate": 8.682432432432433e-06, "loss": 0.0067, "step": 8800 }, { "epoch": 14.700996677740864, "grad_norm": 0.11198179423809052, "learning_rate": 8.673986486486487e-06, "loss": 0.0067, "step": 8850 }, { "epoch": 14.784053156146179, "grad_norm": 0.09852922707796097, "learning_rate": 8.665540540540541e-06, "loss": 0.0068, "step": 8900 }, { "epoch": 14.867109634551495, "grad_norm": 0.18433059751987457, "learning_rate": 8.657094594594595e-06, "loss": 0.0065, "step": 8950 }, { "epoch": 14.950166112956811, "grad_norm": 0.09405363351106644, "learning_rate": 8.64864864864865e-06, "loss": 0.0065, "step": 9000 }, { "epoch": 15.033222591362126, "grad_norm": 0.16450953483581543, "learning_rate": 8.640202702702703e-06, "loss": 0.0061, "step": 9050 }, { "epoch": 15.116279069767442, "grad_norm": 0.1192869246006012, "learning_rate": 8.631756756756757e-06, "loss": 0.0059, "step": 9100 }, { "epoch": 15.199335548172758, "grad_norm": 0.1115177720785141, "learning_rate": 8.623310810810811e-06, "loss": 0.006, "step": 9150 }, { "epoch": 15.282392026578073, "grad_norm": 0.12159425020217896, "learning_rate": 8.614864864864865e-06, "loss": 0.0061, "step": 9200 }, { "epoch": 15.365448504983389, "grad_norm": 0.13033558428287506, "learning_rate": 8.60641891891892e-06, "loss": 0.0059, "step": 9250 }, { "epoch": 15.448504983388704, "grad_norm": 0.12953154742717743, "learning_rate": 8.597972972972974e-06, "loss": 0.0059, "step": 9300 }, { "epoch": 15.53156146179402, "grad_norm": 0.15541379153728485, "learning_rate": 8.589527027027028e-06, "loss": 0.006, "step": 9350 }, { "epoch": 15.614617940199336, "grad_norm": 0.11866448819637299, "learning_rate": 8.581081081081082e-06, "loss": 0.0059, "step": 9400 }, { "epoch": 15.69767441860465, "grad_norm": 0.16709576547145844, "learning_rate": 8.572635135135136e-06, "loss": 0.006, "step": 9450 }, { "epoch": 15.780730897009967, "grad_norm": 0.11866389960050583, "learning_rate": 8.56418918918919e-06, "loss": 0.0062, "step": 9500 }, { "epoch": 15.863787375415283, "grad_norm": 0.1888553947210312, "learning_rate": 8.555743243243244e-06, "loss": 0.0062, "step": 9550 }, { "epoch": 15.946843853820598, "grad_norm": 0.14204780757427216, "learning_rate": 8.547297297297298e-06, "loss": 0.0064, "step": 9600 }, { "epoch": 16.029900332225914, "grad_norm": 0.1339997798204422, "learning_rate": 8.538851351351352e-06, "loss": 0.0059, "step": 9650 }, { "epoch": 16.11295681063123, "grad_norm": 0.1062059998512268, "learning_rate": 8.530405405405406e-06, "loss": 0.0055, "step": 9700 }, { "epoch": 16.196013289036546, "grad_norm": 0.1264619082212448, "learning_rate": 8.52195945945946e-06, "loss": 0.0056, "step": 9750 }, { "epoch": 16.27906976744186, "grad_norm": 0.09743476659059525, "learning_rate": 8.513513513513514e-06, "loss": 0.0055, "step": 9800 }, { "epoch": 16.362126245847175, "grad_norm": 0.16417229175567627, "learning_rate": 8.505067567567568e-06, "loss": 0.0053, "step": 9850 }, { "epoch": 16.44518272425249, "grad_norm": 0.12796853482723236, "learning_rate": 8.496621621621622e-06, "loss": 0.0054, "step": 9900 }, { "epoch": 16.528239202657808, "grad_norm": 0.1504344791173935, "learning_rate": 8.488175675675676e-06, "loss": 0.0056, "step": 9950 }, { "epoch": 16.611295681063122, "grad_norm": 0.1360059231519699, "learning_rate": 8.479729729729732e-06, "loss": 0.0056, "step": 10000 }, { "epoch": 16.694352159468437, "grad_norm": 0.12313506752252579, "learning_rate": 8.471283783783784e-06, "loss": 0.0056, "step": 10050 }, { "epoch": 16.777408637873755, "grad_norm": 0.16136495769023895, "learning_rate": 8.462837837837838e-06, "loss": 0.0056, "step": 10100 }, { "epoch": 16.86046511627907, "grad_norm": 0.15183241665363312, "learning_rate": 8.454391891891892e-06, "loss": 0.0057, "step": 10150 }, { "epoch": 16.943521594684384, "grad_norm": 0.11542920768260956, "learning_rate": 8.445945945945948e-06, "loss": 0.0057, "step": 10200 }, { "epoch": 17.026578073089702, "grad_norm": 0.14199940860271454, "learning_rate": 8.4375e-06, "loss": 0.0055, "step": 10250 }, { "epoch": 17.109634551495017, "grad_norm": 0.19965875148773193, "learning_rate": 8.429054054054054e-06, "loss": 0.005, "step": 10300 }, { "epoch": 17.19269102990033, "grad_norm": 0.14374370872974396, "learning_rate": 8.42060810810811e-06, "loss": 0.0052, "step": 10350 }, { "epoch": 17.27574750830565, "grad_norm": 0.10513945668935776, "learning_rate": 8.412162162162162e-06, "loss": 0.0052, "step": 10400 }, { "epoch": 17.358803986710964, "grad_norm": 0.14402873814105988, "learning_rate": 8.403716216216216e-06, "loss": 0.0052, "step": 10450 }, { "epoch": 17.441860465116278, "grad_norm": 0.10594220459461212, "learning_rate": 8.39527027027027e-06, "loss": 0.0051, "step": 10500 }, { "epoch": 17.524916943521596, "grad_norm": 0.13597488403320312, "learning_rate": 8.386824324324326e-06, "loss": 0.0051, "step": 10550 }, { "epoch": 17.60797342192691, "grad_norm": 0.17150713503360748, "learning_rate": 8.378378378378378e-06, "loss": 0.0053, "step": 10600 }, { "epoch": 17.691029900332225, "grad_norm": 0.12565460801124573, "learning_rate": 8.369932432432432e-06, "loss": 0.0051, "step": 10650 }, { "epoch": 17.774086378737543, "grad_norm": 0.10341569781303406, "learning_rate": 8.361486486486488e-06, "loss": 0.0053, "step": 10700 }, { "epoch": 17.857142857142858, "grad_norm": 0.10507813096046448, "learning_rate": 8.35304054054054e-06, "loss": 0.0052, "step": 10750 }, { "epoch": 17.940199335548172, "grad_norm": 0.15510645508766174, "learning_rate": 8.344594594594594e-06, "loss": 0.0054, "step": 10800 }, { "epoch": 18.023255813953487, "grad_norm": 0.11788327246904373, "learning_rate": 8.33614864864865e-06, "loss": 0.0048, "step": 10850 }, { "epoch": 18.106312292358805, "grad_norm": 0.14898274838924408, "learning_rate": 8.327702702702704e-06, "loss": 0.0049, "step": 10900 }, { "epoch": 18.18936877076412, "grad_norm": 0.1194600835442543, "learning_rate": 8.319256756756757e-06, "loss": 0.0046, "step": 10950 }, { "epoch": 18.272425249169434, "grad_norm": 0.1447572112083435, "learning_rate": 8.31081081081081e-06, "loss": 0.0047, "step": 11000 }, { "epoch": 18.35548172757475, "grad_norm": 0.2115248292684555, "learning_rate": 8.302364864864866e-06, "loss": 0.0045, "step": 11050 }, { "epoch": 18.438538205980066, "grad_norm": 0.1299593299627304, "learning_rate": 8.29391891891892e-06, "loss": 0.0048, "step": 11100 }, { "epoch": 18.52159468438538, "grad_norm": 0.1801750361919403, "learning_rate": 8.285472972972973e-06, "loss": 0.005, "step": 11150 }, { "epoch": 18.6046511627907, "grad_norm": 0.14780189096927643, "learning_rate": 8.277027027027028e-06, "loss": 0.0051, "step": 11200 }, { "epoch": 18.687707641196013, "grad_norm": 0.11595986038446426, "learning_rate": 8.268581081081082e-06, "loss": 0.0048, "step": 11250 }, { "epoch": 18.770764119601328, "grad_norm": 0.10885494947433472, "learning_rate": 8.260135135135135e-06, "loss": 0.005, "step": 11300 }, { "epoch": 18.853820598006646, "grad_norm": 0.15778137743473053, "learning_rate": 8.25168918918919e-06, "loss": 0.005, "step": 11350 }, { "epoch": 18.93687707641196, "grad_norm": 0.1174725666642189, "learning_rate": 8.243243243243245e-06, "loss": 0.0048, "step": 11400 }, { "epoch": 19.019933554817275, "grad_norm": 0.13217081129550934, "learning_rate": 8.234797297297299e-06, "loss": 0.0045, "step": 11450 }, { "epoch": 19.102990033222593, "grad_norm": 0.13525256514549255, "learning_rate": 8.226351351351351e-06, "loss": 0.0045, "step": 11500 }, { "epoch": 19.186046511627907, "grad_norm": 0.10065272450447083, "learning_rate": 8.217905405405407e-06, "loss": 0.0044, "step": 11550 }, { "epoch": 19.269102990033222, "grad_norm": 0.20510923862457275, "learning_rate": 8.20945945945946e-06, "loss": 0.0043, "step": 11600 }, { "epoch": 19.35215946843854, "grad_norm": 0.14453084766864777, "learning_rate": 8.201013513513513e-06, "loss": 0.0044, "step": 11650 }, { "epoch": 19.435215946843854, "grad_norm": 0.18659614026546478, "learning_rate": 8.192567567567569e-06, "loss": 0.0044, "step": 11700 }, { "epoch": 19.51827242524917, "grad_norm": 0.07844480127096176, "learning_rate": 8.184121621621623e-06, "loss": 0.0044, "step": 11750 }, { "epoch": 19.601328903654483, "grad_norm": 0.09800627827644348, "learning_rate": 8.175675675675677e-06, "loss": 0.0046, "step": 11800 }, { "epoch": 19.6843853820598, "grad_norm": 0.13740216195583344, "learning_rate": 8.16722972972973e-06, "loss": 0.0044, "step": 11850 }, { "epoch": 19.767441860465116, "grad_norm": 0.16714897751808167, "learning_rate": 8.158783783783785e-06, "loss": 0.0042, "step": 11900 }, { "epoch": 19.85049833887043, "grad_norm": 0.2006695568561554, "learning_rate": 8.150337837837839e-06, "loss": 0.0045, "step": 11950 }, { "epoch": 19.93355481727575, "grad_norm": 0.1745181381702423, "learning_rate": 8.141891891891893e-06, "loss": 0.0044, "step": 12000 }, { "epoch": 20.016611295681063, "grad_norm": 0.12740877270698547, "learning_rate": 8.133445945945947e-06, "loss": 0.0044, "step": 12050 }, { "epoch": 20.099667774086377, "grad_norm": 0.14775437116622925, "learning_rate": 8.125000000000001e-06, "loss": 0.0039, "step": 12100 }, { "epoch": 20.182724252491695, "grad_norm": 0.14270846545696259, "learning_rate": 8.116554054054055e-06, "loss": 0.0041, "step": 12150 }, { "epoch": 20.26578073089701, "grad_norm": 0.112132228910923, "learning_rate": 8.108108108108109e-06, "loss": 0.0042, "step": 12200 }, { "epoch": 20.348837209302324, "grad_norm": 0.18052621185779572, "learning_rate": 8.099662162162163e-06, "loss": 0.0039, "step": 12250 }, { "epoch": 20.431893687707642, "grad_norm": 0.16352351009845734, "learning_rate": 8.091216216216217e-06, "loss": 0.0038, "step": 12300 }, { "epoch": 20.514950166112957, "grad_norm": 0.13323716819286346, "learning_rate": 8.082770270270271e-06, "loss": 0.0042, "step": 12350 }, { "epoch": 20.59800664451827, "grad_norm": 0.16940850019454956, "learning_rate": 8.074324324324325e-06, "loss": 0.0039, "step": 12400 }, { "epoch": 20.68106312292359, "grad_norm": 0.19970548152923584, "learning_rate": 8.06587837837838e-06, "loss": 0.0042, "step": 12450 }, { "epoch": 20.764119601328904, "grad_norm": 0.1452200710773468, "learning_rate": 8.057432432432433e-06, "loss": 0.004, "step": 12500 }, { "epoch": 20.84717607973422, "grad_norm": 0.20347419381141663, "learning_rate": 8.048986486486487e-06, "loss": 0.004, "step": 12550 }, { "epoch": 20.930232558139537, "grad_norm": 0.19756294786930084, "learning_rate": 8.040540540540541e-06, "loss": 0.0043, "step": 12600 }, { "epoch": 21.01328903654485, "grad_norm": 0.14150208234786987, "learning_rate": 8.032094594594595e-06, "loss": 0.0041, "step": 12650 }, { "epoch": 21.096345514950166, "grad_norm": 0.13472872972488403, "learning_rate": 8.02364864864865e-06, "loss": 0.0034, "step": 12700 }, { "epoch": 21.17940199335548, "grad_norm": 0.19563962519168854, "learning_rate": 8.015202702702703e-06, "loss": 0.0038, "step": 12750 }, { "epoch": 21.262458471760798, "grad_norm": 0.17354747653007507, "learning_rate": 8.006756756756757e-06, "loss": 0.0038, "step": 12800 }, { "epoch": 21.345514950166113, "grad_norm": 0.13766075670719147, "learning_rate": 7.998310810810811e-06, "loss": 0.0038, "step": 12850 }, { "epoch": 21.428571428571427, "grad_norm": 0.16376228630542755, "learning_rate": 7.989864864864865e-06, "loss": 0.0037, "step": 12900 }, { "epoch": 21.511627906976745, "grad_norm": 0.07772815227508545, "learning_rate": 7.98141891891892e-06, "loss": 0.0038, "step": 12950 }, { "epoch": 21.59468438538206, "grad_norm": 0.15981240570545197, "learning_rate": 7.972972972972974e-06, "loss": 0.0036, "step": 13000 }, { "epoch": 21.677740863787374, "grad_norm": 0.15515024960041046, "learning_rate": 7.964527027027028e-06, "loss": 0.004, "step": 13050 }, { "epoch": 21.760797342192692, "grad_norm": 0.14390234649181366, "learning_rate": 7.956081081081082e-06, "loss": 0.0037, "step": 13100 }, { "epoch": 21.843853820598007, "grad_norm": 0.16870105266571045, "learning_rate": 7.947635135135136e-06, "loss": 0.0039, "step": 13150 }, { "epoch": 21.92691029900332, "grad_norm": 0.13909319043159485, "learning_rate": 7.93918918918919e-06, "loss": 0.0035, "step": 13200 }, { "epoch": 22.00996677740864, "grad_norm": 0.0985972061753273, "learning_rate": 7.930743243243244e-06, "loss": 0.0038, "step": 13250 }, { "epoch": 22.093023255813954, "grad_norm": 0.1900232881307602, "learning_rate": 7.922297297297298e-06, "loss": 0.0035, "step": 13300 }, { "epoch": 22.176079734219268, "grad_norm": 0.12009528279304504, "learning_rate": 7.913851351351352e-06, "loss": 0.0035, "step": 13350 }, { "epoch": 22.259136212624586, "grad_norm": 0.2472468614578247, "learning_rate": 7.905405405405406e-06, "loss": 0.0035, "step": 13400 }, { "epoch": 22.3421926910299, "grad_norm": 0.206543430685997, "learning_rate": 7.89695945945946e-06, "loss": 0.0032, "step": 13450 }, { "epoch": 22.425249169435215, "grad_norm": 0.11098843067884445, "learning_rate": 7.888513513513514e-06, "loss": 0.0035, "step": 13500 }, { "epoch": 22.508305647840533, "grad_norm": 0.11390508711338043, "learning_rate": 7.880067567567568e-06, "loss": 0.0034, "step": 13550 }, { "epoch": 22.591362126245848, "grad_norm": 0.15061669051647186, "learning_rate": 7.871621621621622e-06, "loss": 0.0035, "step": 13600 }, { "epoch": 22.674418604651162, "grad_norm": 0.11270301789045334, "learning_rate": 7.863175675675676e-06, "loss": 0.0035, "step": 13650 }, { "epoch": 22.757475083056477, "grad_norm": 0.24368135631084442, "learning_rate": 7.854729729729732e-06, "loss": 0.0033, "step": 13700 }, { "epoch": 22.840531561461795, "grad_norm": 0.1437971293926239, "learning_rate": 7.846283783783784e-06, "loss": 0.0035, "step": 13750 }, { "epoch": 22.92358803986711, "grad_norm": 0.11598620563745499, "learning_rate": 7.837837837837838e-06, "loss": 0.0037, "step": 13800 }, { "epoch": 23.006644518272424, "grad_norm": 0.14135406911373138, "learning_rate": 7.829391891891892e-06, "loss": 0.0037, "step": 13850 }, { "epoch": 23.089700996677742, "grad_norm": 0.21696355938911438, "learning_rate": 7.820945945945946e-06, "loss": 0.0031, "step": 13900 }, { "epoch": 23.172757475083056, "grad_norm": 0.17200914025306702, "learning_rate": 7.8125e-06, "loss": 0.0034, "step": 13950 }, { "epoch": 23.25581395348837, "grad_norm": 0.19111374020576477, "learning_rate": 7.804054054054054e-06, "loss": 0.0035, "step": 14000 }, { "epoch": 23.33887043189369, "grad_norm": 0.17747582495212555, "learning_rate": 7.79560810810811e-06, "loss": 0.0034, "step": 14050 }, { "epoch": 23.421926910299003, "grad_norm": 0.129624605178833, "learning_rate": 7.787162162162162e-06, "loss": 0.0033, "step": 14100 }, { "epoch": 23.504983388704318, "grad_norm": 0.18598869442939758, "learning_rate": 7.778716216216216e-06, "loss": 0.0033, "step": 14150 }, { "epoch": 23.588039867109636, "grad_norm": 0.11749470978975296, "learning_rate": 7.77027027027027e-06, "loss": 0.0034, "step": 14200 }, { "epoch": 23.67109634551495, "grad_norm": 0.11158709973096848, "learning_rate": 7.761824324324324e-06, "loss": 0.0033, "step": 14250 }, { "epoch": 23.754152823920265, "grad_norm": 0.11481600254774094, "learning_rate": 7.753378378378378e-06, "loss": 0.0031, "step": 14300 }, { "epoch": 23.837209302325583, "grad_norm": 0.16647398471832275, "learning_rate": 7.744932432432432e-06, "loss": 0.0033, "step": 14350 }, { "epoch": 23.920265780730897, "grad_norm": 0.14807546138763428, "learning_rate": 7.736486486486488e-06, "loss": 0.0031, "step": 14400 }, { "epoch": 24.003322259136212, "grad_norm": 0.1168646588921547, "learning_rate": 7.72804054054054e-06, "loss": 0.0032, "step": 14450 }, { "epoch": 24.08637873754153, "grad_norm": 0.1923869401216507, "learning_rate": 7.719594594594595e-06, "loss": 0.0032, "step": 14500 }, { "epoch": 24.169435215946844, "grad_norm": 0.088302381336689, "learning_rate": 7.71114864864865e-06, "loss": 0.003, "step": 14550 }, { "epoch": 24.25249169435216, "grad_norm": 0.1589769572019577, "learning_rate": 7.702702702702704e-06, "loss": 0.0029, "step": 14600 }, { "epoch": 24.335548172757473, "grad_norm": 0.12336576730012894, "learning_rate": 7.694256756756757e-06, "loss": 0.0029, "step": 14650 }, { "epoch": 24.41860465116279, "grad_norm": 0.18861009180545807, "learning_rate": 7.68581081081081e-06, "loss": 0.0031, "step": 14700 }, { "epoch": 24.501661129568106, "grad_norm": 0.15962085127830505, "learning_rate": 7.677364864864866e-06, "loss": 0.0029, "step": 14750 }, { "epoch": 24.58471760797342, "grad_norm": 0.12614157795906067, "learning_rate": 7.668918918918919e-06, "loss": 0.0032, "step": 14800 }, { "epoch": 24.66777408637874, "grad_norm": 0.14915503561496735, "learning_rate": 7.660472972972973e-06, "loss": 0.0031, "step": 14850 }, { "epoch": 24.750830564784053, "grad_norm": 0.1845034956932068, "learning_rate": 7.652027027027028e-06, "loss": 0.0031, "step": 14900 }, { "epoch": 24.833887043189367, "grad_norm": 0.2132125347852707, "learning_rate": 7.643581081081082e-06, "loss": 0.003, "step": 14950 }, { "epoch": 24.916943521594686, "grad_norm": 0.1357172429561615, "learning_rate": 7.635135135135135e-06, "loss": 0.0032, "step": 15000 }, { "epoch": 25.0, "grad_norm": 0.1253499686717987, "learning_rate": 7.6266891891891906e-06, "loss": 0.0032, "step": 15050 }, { "epoch": 25.083056478405314, "grad_norm": 0.1296907216310501, "learning_rate": 7.618243243243244e-06, "loss": 0.0028, "step": 15100 }, { "epoch": 25.166112956810633, "grad_norm": 0.10292603075504303, "learning_rate": 7.609797297297298e-06, "loss": 0.0028, "step": 15150 }, { "epoch": 25.249169435215947, "grad_norm": 0.17277835309505463, "learning_rate": 7.601351351351352e-06, "loss": 0.003, "step": 15200 }, { "epoch": 25.33222591362126, "grad_norm": 0.13872121274471283, "learning_rate": 7.592905405405407e-06, "loss": 0.0026, "step": 15250 }, { "epoch": 25.41528239202658, "grad_norm": 0.16662010550498962, "learning_rate": 7.58445945945946e-06, "loss": 0.003, "step": 15300 }, { "epoch": 25.498338870431894, "grad_norm": 0.17448820173740387, "learning_rate": 7.576013513513514e-06, "loss": 0.0028, "step": 15350 }, { "epoch": 25.58139534883721, "grad_norm": 0.09598555415868759, "learning_rate": 7.567567567567569e-06, "loss": 0.003, "step": 15400 }, { "epoch": 25.664451827242527, "grad_norm": 0.14937862753868103, "learning_rate": 7.559121621621622e-06, "loss": 0.0028, "step": 15450 }, { "epoch": 25.74750830564784, "grad_norm": 0.15394462645053864, "learning_rate": 7.550675675675676e-06, "loss": 0.0029, "step": 15500 }, { "epoch": 25.830564784053156, "grad_norm": 0.14284346997737885, "learning_rate": 7.542229729729731e-06, "loss": 0.0027, "step": 15550 }, { "epoch": 25.91362126245847, "grad_norm": 0.15943670272827148, "learning_rate": 7.533783783783785e-06, "loss": 0.0028, "step": 15600 }, { "epoch": 25.996677740863788, "grad_norm": 0.11656994372606277, "learning_rate": 7.525337837837838e-06, "loss": 0.0027, "step": 15650 }, { "epoch": 26.079734219269103, "grad_norm": 0.13571897149085999, "learning_rate": 7.516891891891892e-06, "loss": 0.0028, "step": 15700 }, { "epoch": 26.162790697674417, "grad_norm": 0.14880968630313873, "learning_rate": 7.508445945945947e-06, "loss": 0.0025, "step": 15750 }, { "epoch": 26.245847176079735, "grad_norm": 0.14208616316318512, "learning_rate": 7.500000000000001e-06, "loss": 0.0026, "step": 15800 }, { "epoch": 26.32890365448505, "grad_norm": 0.11586987972259521, "learning_rate": 7.491554054054054e-06, "loss": 0.0025, "step": 15850 }, { "epoch": 26.411960132890364, "grad_norm": 0.14217934012413025, "learning_rate": 7.483108108108109e-06, "loss": 0.0027, "step": 15900 }, { "epoch": 26.495016611295682, "grad_norm": 0.17761094868183136, "learning_rate": 7.474662162162163e-06, "loss": 0.0025, "step": 15950 }, { "epoch": 26.578073089700997, "grad_norm": 0.10167348384857178, "learning_rate": 7.466216216216216e-06, "loss": 0.0024, "step": 16000 }, { "epoch": 26.66112956810631, "grad_norm": 0.14244844019412994, "learning_rate": 7.45777027027027e-06, "loss": 0.0028, "step": 16050 }, { "epoch": 26.74418604651163, "grad_norm": 0.1568807065486908, "learning_rate": 7.449324324324325e-06, "loss": 0.0026, "step": 16100 }, { "epoch": 26.827242524916944, "grad_norm": 0.13260778784751892, "learning_rate": 7.440878378378379e-06, "loss": 0.0026, "step": 16150 }, { "epoch": 26.910299003322258, "grad_norm": 0.1154264584183693, "learning_rate": 7.4324324324324324e-06, "loss": 0.0027, "step": 16200 }, { "epoch": 26.993355481727576, "grad_norm": 0.12289530783891678, "learning_rate": 7.423986486486487e-06, "loss": 0.0029, "step": 16250 }, { "epoch": 27.07641196013289, "grad_norm": 0.12922519445419312, "learning_rate": 7.415540540540541e-06, "loss": 0.0023, "step": 16300 }, { "epoch": 27.159468438538205, "grad_norm": 0.13929371535778046, "learning_rate": 7.4070945945945945e-06, "loss": 0.0025, "step": 16350 }, { "epoch": 27.242524916943523, "grad_norm": 0.1668965071439743, "learning_rate": 7.398648648648649e-06, "loss": 0.0026, "step": 16400 }, { "epoch": 27.325581395348838, "grad_norm": 0.1570366621017456, "learning_rate": 7.3902027027027035e-06, "loss": 0.0025, "step": 16450 }, { "epoch": 27.408637873754152, "grad_norm": 0.1284647434949875, "learning_rate": 7.3817567567567575e-06, "loss": 0.0024, "step": 16500 }, { "epoch": 27.491694352159467, "grad_norm": 0.15990597009658813, "learning_rate": 7.373310810810811e-06, "loss": 0.0024, "step": 16550 }, { "epoch": 27.574750830564785, "grad_norm": 0.12897512316703796, "learning_rate": 7.3648648648648655e-06, "loss": 0.0028, "step": 16600 }, { "epoch": 27.6578073089701, "grad_norm": 0.13328291475772858, "learning_rate": 7.35641891891892e-06, "loss": 0.0025, "step": 16650 }, { "epoch": 27.740863787375414, "grad_norm": 0.07590755075216293, "learning_rate": 7.347972972972973e-06, "loss": 0.0025, "step": 16700 }, { "epoch": 27.823920265780732, "grad_norm": 0.14616797864437103, "learning_rate": 7.339527027027028e-06, "loss": 0.0025, "step": 16750 }, { "epoch": 27.906976744186046, "grad_norm": 0.10863177478313446, "learning_rate": 7.331081081081082e-06, "loss": 0.0025, "step": 16800 }, { "epoch": 27.99003322259136, "grad_norm": 0.11199267953634262, "learning_rate": 7.322635135135136e-06, "loss": 0.0027, "step": 16850 }, { "epoch": 28.07308970099668, "grad_norm": 0.13767418265342712, "learning_rate": 7.314189189189191e-06, "loss": 0.0022, "step": 16900 }, { "epoch": 28.156146179401993, "grad_norm": 0.13357830047607422, "learning_rate": 7.305743243243244e-06, "loss": 0.0023, "step": 16950 }, { "epoch": 28.239202657807308, "grad_norm": 0.22289451956748962, "learning_rate": 7.297297297297298e-06, "loss": 0.0021, "step": 17000 }, { "epoch": 28.322259136212626, "grad_norm": 0.12740780413150787, "learning_rate": 7.288851351351352e-06, "loss": 0.0023, "step": 17050 }, { "epoch": 28.40531561461794, "grad_norm": 0.16222257912158966, "learning_rate": 7.280405405405407e-06, "loss": 0.0024, "step": 17100 }, { "epoch": 28.488372093023255, "grad_norm": 0.11356110870838165, "learning_rate": 7.27195945945946e-06, "loss": 0.0023, "step": 17150 }, { "epoch": 28.571428571428573, "grad_norm": 0.11776330322027206, "learning_rate": 7.263513513513514e-06, "loss": 0.0024, "step": 17200 }, { "epoch": 28.654485049833887, "grad_norm": 0.16843853890895844, "learning_rate": 7.255067567567569e-06, "loss": 0.0023, "step": 17250 }, { "epoch": 28.737541528239202, "grad_norm": 0.1754775196313858, "learning_rate": 7.246621621621622e-06, "loss": 0.0023, "step": 17300 }, { "epoch": 28.82059800664452, "grad_norm": 0.1168438196182251, "learning_rate": 7.238175675675676e-06, "loss": 0.0023, "step": 17350 }, { "epoch": 28.903654485049834, "grad_norm": 0.07629954069852829, "learning_rate": 7.229729729729731e-06, "loss": 0.0023, "step": 17400 }, { "epoch": 28.98671096345515, "grad_norm": 0.16367056965827942, "learning_rate": 7.221283783783785e-06, "loss": 0.0025, "step": 17450 }, { "epoch": 29.069767441860463, "grad_norm": 0.15453363955020905, "learning_rate": 7.212837837837838e-06, "loss": 0.0022, "step": 17500 }, { "epoch": 29.15282392026578, "grad_norm": 0.1271519809961319, "learning_rate": 7.204391891891892e-06, "loss": 0.0022, "step": 17550 }, { "epoch": 29.235880398671096, "grad_norm": 0.07987678050994873, "learning_rate": 7.195945945945947e-06, "loss": 0.0018, "step": 17600 }, { "epoch": 29.31893687707641, "grad_norm": 0.1695636510848999, "learning_rate": 7.1875e-06, "loss": 0.0022, "step": 17650 }, { "epoch": 29.40199335548173, "grad_norm": 0.1446850448846817, "learning_rate": 7.179054054054054e-06, "loss": 0.0023, "step": 17700 }, { "epoch": 29.485049833887043, "grad_norm": 0.134325310587883, "learning_rate": 7.170608108108109e-06, "loss": 0.0022, "step": 17750 }, { "epoch": 29.568106312292358, "grad_norm": 0.18675336241722107, "learning_rate": 7.162162162162163e-06, "loss": 0.0023, "step": 17800 }, { "epoch": 29.651162790697676, "grad_norm": 0.22028237581253052, "learning_rate": 7.153716216216216e-06, "loss": 0.0021, "step": 17850 }, { "epoch": 29.73421926910299, "grad_norm": 0.1827327311038971, "learning_rate": 7.14527027027027e-06, "loss": 0.0024, "step": 17900 }, { "epoch": 29.817275747508305, "grad_norm": 0.18215274810791016, "learning_rate": 7.136824324324325e-06, "loss": 0.0024, "step": 17950 }, { "epoch": 29.900332225913623, "grad_norm": 0.20356135070323944, "learning_rate": 7.128378378378379e-06, "loss": 0.0022, "step": 18000 }, { "epoch": 29.983388704318937, "grad_norm": 0.14500537514686584, "learning_rate": 7.1199324324324325e-06, "loss": 0.0021, "step": 18050 }, { "epoch": 30.06644518272425, "grad_norm": 0.1145324632525444, "learning_rate": 7.111486486486487e-06, "loss": 0.002, "step": 18100 }, { "epoch": 30.14950166112957, "grad_norm": 0.14700041711330414, "learning_rate": 7.103040540540541e-06, "loss": 0.0021, "step": 18150 }, { "epoch": 30.232558139534884, "grad_norm": 0.20703761279582977, "learning_rate": 7.0945945945945946e-06, "loss": 0.002, "step": 18200 }, { "epoch": 30.3156146179402, "grad_norm": 0.13587066531181335, "learning_rate": 7.0861486486486494e-06, "loss": 0.002, "step": 18250 }, { "epoch": 30.398671096345517, "grad_norm": 0.18110080063343048, "learning_rate": 7.0777027027027035e-06, "loss": 0.0019, "step": 18300 }, { "epoch": 30.48172757475083, "grad_norm": 0.16752487421035767, "learning_rate": 7.0692567567567575e-06, "loss": 0.0022, "step": 18350 }, { "epoch": 30.564784053156146, "grad_norm": 0.13479451835155487, "learning_rate": 7.060810810810811e-06, "loss": 0.0021, "step": 18400 }, { "epoch": 30.64784053156146, "grad_norm": 0.1294214129447937, "learning_rate": 7.0523648648648656e-06, "loss": 0.002, "step": 18450 }, { "epoch": 30.730897009966778, "grad_norm": 0.15154747664928436, "learning_rate": 7.04391891891892e-06, "loss": 0.0022, "step": 18500 }, { "epoch": 30.813953488372093, "grad_norm": 0.18404319882392883, "learning_rate": 7.035472972972973e-06, "loss": 0.0021, "step": 18550 }, { "epoch": 30.897009966777407, "grad_norm": 0.13255277276039124, "learning_rate": 7.027027027027028e-06, "loss": 0.0023, "step": 18600 }, { "epoch": 30.980066445182725, "grad_norm": 0.18859443068504333, "learning_rate": 7.018581081081082e-06, "loss": 0.0021, "step": 18650 }, { "epoch": 31.06312292358804, "grad_norm": 0.19044509530067444, "learning_rate": 7.010135135135136e-06, "loss": 0.0021, "step": 18700 }, { "epoch": 31.146179401993354, "grad_norm": 0.15139712393283844, "learning_rate": 7.001689189189191e-06, "loss": 0.002, "step": 18750 }, { "epoch": 31.229235880398672, "grad_norm": 0.06357304006814957, "learning_rate": 6.993243243243244e-06, "loss": 0.002, "step": 18800 }, { "epoch": 31.312292358803987, "grad_norm": 0.10291548818349838, "learning_rate": 6.984797297297298e-06, "loss": 0.0019, "step": 18850 }, { "epoch": 31.3953488372093, "grad_norm": 0.19652868807315826, "learning_rate": 6.976351351351352e-06, "loss": 0.0019, "step": 18900 }, { "epoch": 31.47840531561462, "grad_norm": 0.13265159726142883, "learning_rate": 6.967905405405406e-06, "loss": 0.0019, "step": 18950 }, { "epoch": 31.561461794019934, "grad_norm": 0.14125719666481018, "learning_rate": 6.95945945945946e-06, "loss": 0.0018, "step": 19000 }, { "epoch": 31.64451827242525, "grad_norm": 0.1850222498178482, "learning_rate": 6.951013513513514e-06, "loss": 0.002, "step": 19050 }, { "epoch": 31.727574750830566, "grad_norm": 0.23264606297016144, "learning_rate": 6.942567567567569e-06, "loss": 0.002, "step": 19100 }, { "epoch": 31.81063122923588, "grad_norm": 0.13076111674308777, "learning_rate": 6.934121621621622e-06, "loss": 0.0021, "step": 19150 }, { "epoch": 31.893687707641195, "grad_norm": 0.10948067903518677, "learning_rate": 6.925675675675676e-06, "loss": 0.002, "step": 19200 }, { "epoch": 31.97674418604651, "grad_norm": 0.21915070712566376, "learning_rate": 6.917229729729731e-06, "loss": 0.0019, "step": 19250 }, { "epoch": 32.05980066445183, "grad_norm": 0.135627880692482, "learning_rate": 6.908783783783785e-06, "loss": 0.002, "step": 19300 }, { "epoch": 32.142857142857146, "grad_norm": 0.14064209163188934, "learning_rate": 6.900337837837838e-06, "loss": 0.0019, "step": 19350 }, { "epoch": 32.22591362126246, "grad_norm": 0.12070033699274063, "learning_rate": 6.891891891891892e-06, "loss": 0.0017, "step": 19400 }, { "epoch": 32.308970099667775, "grad_norm": 0.08246263116598129, "learning_rate": 6.883445945945947e-06, "loss": 0.0017, "step": 19450 }, { "epoch": 32.39202657807309, "grad_norm": 0.2247394323348999, "learning_rate": 6.875e-06, "loss": 0.0019, "step": 19500 }, { "epoch": 32.475083056478404, "grad_norm": 0.1050836518406868, "learning_rate": 6.866554054054054e-06, "loss": 0.0019, "step": 19550 }, { "epoch": 32.55813953488372, "grad_norm": 0.17155566811561584, "learning_rate": 6.858108108108109e-06, "loss": 0.0018, "step": 19600 }, { "epoch": 32.64119601328904, "grad_norm": 0.16185705363750458, "learning_rate": 6.849662162162163e-06, "loss": 0.0018, "step": 19650 }, { "epoch": 32.72425249169435, "grad_norm": 0.14572490751743317, "learning_rate": 6.841216216216216e-06, "loss": 0.0019, "step": 19700 }, { "epoch": 32.80730897009967, "grad_norm": 0.2703460454940796, "learning_rate": 6.83277027027027e-06, "loss": 0.002, "step": 19750 }, { "epoch": 32.89036544850498, "grad_norm": 0.11189229041337967, "learning_rate": 6.824324324324325e-06, "loss": 0.0017, "step": 19800 }, { "epoch": 32.9734219269103, "grad_norm": 0.09773913025856018, "learning_rate": 6.8158783783783785e-06, "loss": 0.002, "step": 19850 }, { "epoch": 33.056478405315616, "grad_norm": 0.13506102561950684, "learning_rate": 6.8074324324324325e-06, "loss": 0.0019, "step": 19900 }, { "epoch": 33.13953488372093, "grad_norm": 0.14404888451099396, "learning_rate": 6.798986486486487e-06, "loss": 0.0016, "step": 19950 }, { "epoch": 33.222591362126245, "grad_norm": 0.12977443635463715, "learning_rate": 6.790540540540541e-06, "loss": 0.0018, "step": 20000 }, { "epoch": 33.30564784053156, "grad_norm": 0.0847565308213234, "learning_rate": 6.782094594594595e-06, "loss": 0.0018, "step": 20050 }, { "epoch": 33.388704318936874, "grad_norm": 0.12460016459226608, "learning_rate": 6.7736486486486495e-06, "loss": 0.0017, "step": 20100 }, { "epoch": 33.47176079734219, "grad_norm": 0.10600145906209946, "learning_rate": 6.7652027027027035e-06, "loss": 0.0018, "step": 20150 }, { "epoch": 33.55481727574751, "grad_norm": 0.19188791513442993, "learning_rate": 6.7567567567567575e-06, "loss": 0.0018, "step": 20200 }, { "epoch": 33.63787375415282, "grad_norm": 0.1580614298582077, "learning_rate": 6.748310810810811e-06, "loss": 0.0018, "step": 20250 }, { "epoch": 33.72093023255814, "grad_norm": 0.201605886220932, "learning_rate": 6.739864864864866e-06, "loss": 0.0018, "step": 20300 }, { "epoch": 33.80398671096346, "grad_norm": 0.15210430324077606, "learning_rate": 6.73141891891892e-06, "loss": 0.0017, "step": 20350 }, { "epoch": 33.88704318936877, "grad_norm": 0.11763100326061249, "learning_rate": 6.722972972972973e-06, "loss": 0.0019, "step": 20400 }, { "epoch": 33.970099667774086, "grad_norm": 0.1792268007993698, "learning_rate": 6.714527027027028e-06, "loss": 0.0017, "step": 20450 }, { "epoch": 34.053156146179404, "grad_norm": 0.11926724761724472, "learning_rate": 6.706081081081082e-06, "loss": 0.0017, "step": 20500 }, { "epoch": 34.136212624584715, "grad_norm": 0.10554293543100357, "learning_rate": 6.697635135135136e-06, "loss": 0.0016, "step": 20550 }, { "epoch": 34.21926910299003, "grad_norm": 0.13360972702503204, "learning_rate": 6.689189189189191e-06, "loss": 0.0017, "step": 20600 }, { "epoch": 34.30232558139535, "grad_norm": 0.1783204972743988, "learning_rate": 6.680743243243244e-06, "loss": 0.0016, "step": 20650 }, { "epoch": 34.38538205980066, "grad_norm": 0.09073592722415924, "learning_rate": 6.672297297297298e-06, "loss": 0.0019, "step": 20700 }, { "epoch": 34.46843853820598, "grad_norm": 0.25335633754730225, "learning_rate": 6.663851351351351e-06, "loss": 0.0017, "step": 20750 }, { "epoch": 34.5514950166113, "grad_norm": 0.1067514419555664, "learning_rate": 6.655405405405406e-06, "loss": 0.0017, "step": 20800 }, { "epoch": 34.63455149501661, "grad_norm": 0.15787842869758606, "learning_rate": 6.64695945945946e-06, "loss": 0.0017, "step": 20850 }, { "epoch": 34.71760797342193, "grad_norm": 0.1340426504611969, "learning_rate": 6.638513513513514e-06, "loss": 0.0018, "step": 20900 }, { "epoch": 34.800664451827245, "grad_norm": 0.2136928290128708, "learning_rate": 6.630067567567569e-06, "loss": 0.0016, "step": 20950 }, { "epoch": 34.883720930232556, "grad_norm": 0.12659946084022522, "learning_rate": 6.621621621621622e-06, "loss": 0.0016, "step": 21000 }, { "epoch": 34.966777408637874, "grad_norm": 0.13203801214694977, "learning_rate": 6.613175675675676e-06, "loss": 0.0016, "step": 21050 }, { "epoch": 35.04983388704319, "grad_norm": 0.13505145907402039, "learning_rate": 6.604729729729731e-06, "loss": 0.0015, "step": 21100 }, { "epoch": 35.1328903654485, "grad_norm": 0.12108131498098373, "learning_rate": 6.596283783783784e-06, "loss": 0.0016, "step": 21150 }, { "epoch": 35.21594684385382, "grad_norm": 0.06628163158893585, "learning_rate": 6.587837837837838e-06, "loss": 0.0014, "step": 21200 }, { "epoch": 35.29900332225914, "grad_norm": 0.0889243558049202, "learning_rate": 6.579391891891892e-06, "loss": 0.0017, "step": 21250 }, { "epoch": 35.38205980066445, "grad_norm": 0.11748132109642029, "learning_rate": 6.570945945945947e-06, "loss": 0.0014, "step": 21300 }, { "epoch": 35.46511627906977, "grad_norm": 0.16354264318943024, "learning_rate": 6.5625e-06, "loss": 0.0016, "step": 21350 }, { "epoch": 35.548172757475086, "grad_norm": 0.05732898786664009, "learning_rate": 6.554054054054054e-06, "loss": 0.0015, "step": 21400 }, { "epoch": 35.6312292358804, "grad_norm": 0.2699935734272003, "learning_rate": 6.545608108108109e-06, "loss": 0.0017, "step": 21450 }, { "epoch": 35.714285714285715, "grad_norm": 0.18786229193210602, "learning_rate": 6.537162162162163e-06, "loss": 0.0016, "step": 21500 }, { "epoch": 35.79734219269103, "grad_norm": 0.087208591401577, "learning_rate": 6.528716216216216e-06, "loss": 0.0016, "step": 21550 }, { "epoch": 35.880398671096344, "grad_norm": 0.0993415042757988, "learning_rate": 6.5202702702702704e-06, "loss": 0.0017, "step": 21600 }, { "epoch": 35.96345514950166, "grad_norm": 0.2238815277814865, "learning_rate": 6.511824324324325e-06, "loss": 0.0016, "step": 21650 }, { "epoch": 36.04651162790697, "grad_norm": 0.13334788382053375, "learning_rate": 6.5033783783783785e-06, "loss": 0.0015, "step": 21700 }, { "epoch": 36.12956810631229, "grad_norm": 0.18202228844165802, "learning_rate": 6.4949324324324325e-06, "loss": 0.0014, "step": 21750 }, { "epoch": 36.21262458471761, "grad_norm": 0.06453296542167664, "learning_rate": 6.486486486486487e-06, "loss": 0.0014, "step": 21800 }, { "epoch": 36.29568106312292, "grad_norm": 0.12124403566122055, "learning_rate": 6.4780405405405414e-06, "loss": 0.0013, "step": 21850 }, { "epoch": 36.37873754152824, "grad_norm": 0.24454446136951447, "learning_rate": 6.469594594594595e-06, "loss": 0.0014, "step": 21900 }, { "epoch": 36.461794019933556, "grad_norm": 0.13677524030208588, "learning_rate": 6.4611486486486495e-06, "loss": 0.0015, "step": 21950 }, { "epoch": 36.54485049833887, "grad_norm": 0.25657689571380615, "learning_rate": 6.4527027027027035e-06, "loss": 0.0015, "step": 22000 }, { "epoch": 36.627906976744185, "grad_norm": 0.08691773563623428, "learning_rate": 6.444256756756757e-06, "loss": 0.0014, "step": 22050 }, { "epoch": 36.7109634551495, "grad_norm": 0.11459925025701523, "learning_rate": 6.435810810810811e-06, "loss": 0.0017, "step": 22100 }, { "epoch": 36.794019933554814, "grad_norm": 0.15285205841064453, "learning_rate": 6.427364864864866e-06, "loss": 0.0016, "step": 22150 }, { "epoch": 36.87707641196013, "grad_norm": 0.06666266173124313, "learning_rate": 6.41891891891892e-06, "loss": 0.0014, "step": 22200 }, { "epoch": 36.96013289036545, "grad_norm": 0.15892595052719116, "learning_rate": 6.410472972972973e-06, "loss": 0.0015, "step": 22250 }, { "epoch": 37.04318936877076, "grad_norm": 0.07195648550987244, "learning_rate": 6.402027027027028e-06, "loss": 0.0016, "step": 22300 }, { "epoch": 37.12624584717608, "grad_norm": 0.13736845552921295, "learning_rate": 6.393581081081082e-06, "loss": 0.0015, "step": 22350 }, { "epoch": 37.2093023255814, "grad_norm": 0.23577900230884552, "learning_rate": 6.385135135135136e-06, "loss": 0.0014, "step": 22400 }, { "epoch": 37.29235880398671, "grad_norm": 0.14850834012031555, "learning_rate": 6.37668918918919e-06, "loss": 0.0013, "step": 22450 }, { "epoch": 37.37541528239203, "grad_norm": 0.10560604929924011, "learning_rate": 6.368243243243244e-06, "loss": 0.0015, "step": 22500 }, { "epoch": 37.458471760797345, "grad_norm": 0.14007191359996796, "learning_rate": 6.359797297297298e-06, "loss": 0.0014, "step": 22550 }, { "epoch": 37.541528239202655, "grad_norm": 0.19559958577156067, "learning_rate": 6.351351351351351e-06, "loss": 0.0014, "step": 22600 }, { "epoch": 37.62458471760797, "grad_norm": 0.10152676701545715, "learning_rate": 6.342905405405406e-06, "loss": 0.0014, "step": 22650 }, { "epoch": 37.70764119601329, "grad_norm": 0.1924515813589096, "learning_rate": 6.33445945945946e-06, "loss": 0.0014, "step": 22700 }, { "epoch": 37.7906976744186, "grad_norm": 0.12958408892154694, "learning_rate": 6.326013513513514e-06, "loss": 0.0016, "step": 22750 }, { "epoch": 37.87375415282392, "grad_norm": 0.12392258644104004, "learning_rate": 6.317567567567569e-06, "loss": 0.0015, "step": 22800 }, { "epoch": 37.95681063122924, "grad_norm": 0.17323195934295654, "learning_rate": 6.309121621621622e-06, "loss": 0.0015, "step": 22850 }, { "epoch": 38.03986710963455, "grad_norm": 0.16793905198574066, "learning_rate": 6.300675675675676e-06, "loss": 0.0014, "step": 22900 }, { "epoch": 38.12292358803987, "grad_norm": 0.06940795481204987, "learning_rate": 6.292229729729731e-06, "loss": 0.0014, "step": 22950 }, { "epoch": 38.205980066445186, "grad_norm": 0.19214914739131927, "learning_rate": 6.283783783783784e-06, "loss": 0.0013, "step": 23000 }, { "epoch": 38.2890365448505, "grad_norm": 0.08065056800842285, "learning_rate": 6.275337837837838e-06, "loss": 0.0014, "step": 23050 }, { "epoch": 38.372093023255815, "grad_norm": 0.14864496886730194, "learning_rate": 6.266891891891892e-06, "loss": 0.0014, "step": 23100 }, { "epoch": 38.45514950166113, "grad_norm": 0.08823973685503006, "learning_rate": 6.258445945945947e-06, "loss": 0.0012, "step": 23150 }, { "epoch": 38.538205980066444, "grad_norm": 0.10207533836364746, "learning_rate": 6.25e-06, "loss": 0.0013, "step": 23200 }, { "epoch": 38.62126245847176, "grad_norm": 0.19492359459400177, "learning_rate": 6.241554054054054e-06, "loss": 0.0014, "step": 23250 }, { "epoch": 38.70431893687708, "grad_norm": 0.18092162907123566, "learning_rate": 6.233108108108109e-06, "loss": 0.0014, "step": 23300 }, { "epoch": 38.78737541528239, "grad_norm": 0.10780604183673859, "learning_rate": 6.224662162162162e-06, "loss": 0.0014, "step": 23350 }, { "epoch": 38.87043189368771, "grad_norm": 0.07147455215454102, "learning_rate": 6.2162162162162164e-06, "loss": 0.0012, "step": 23400 }, { "epoch": 38.95348837209303, "grad_norm": 0.14428885281085968, "learning_rate": 6.2077702702702705e-06, "loss": 0.0012, "step": 23450 }, { "epoch": 39.03654485049834, "grad_norm": 0.19951941072940826, "learning_rate": 6.199324324324325e-06, "loss": 0.0013, "step": 23500 }, { "epoch": 15.032658393207054, "grad_norm": 0.296187162399292, "learning_rate": 5.000000000000001e-07, "loss": 0.0393, "step": 23550 }, { "epoch": 15.065316786414108, "grad_norm": 0.06856995820999146, "learning_rate": 1.0000000000000002e-06, "loss": 0.029, "step": 23600 }, { "epoch": 15.097975179621162, "grad_norm": 0.07123560458421707, "learning_rate": 1.5e-06, "loss": 0.0282, "step": 23650 }, { "epoch": 15.130633572828216, "grad_norm": 0.08933639526367188, "learning_rate": 2.0000000000000003e-06, "loss": 0.0269, "step": 23700 }, { "epoch": 15.16329196603527, "grad_norm": 0.09562961757183075, "learning_rate": 2.5e-06, "loss": 0.0266, "step": 23750 }, { "epoch": 15.195950359242325, "grad_norm": 0.12030792236328125, "learning_rate": 3e-06, "loss": 0.025, "step": 23800 }, { "epoch": 15.228608752449379, "grad_norm": 0.10658561438322067, "learning_rate": 3.5e-06, "loss": 0.0234, "step": 23850 }, { "epoch": 15.261267145656433, "grad_norm": 0.1295558512210846, "learning_rate": 4.000000000000001e-06, "loss": 0.0229, "step": 23900 }, { "epoch": 15.293925538863489, "grad_norm": 0.13431811332702637, "learning_rate": 4.5e-06, "loss": 0.023, "step": 23950 }, { "epoch": 15.326583932070543, "grad_norm": 0.12976951897144318, "learning_rate": 5e-06, "loss": 0.0217, "step": 24000 }, { "epoch": 15.359242325277597, "grad_norm": 0.12544389069080353, "learning_rate": 5.500000000000001e-06, "loss": 0.022, "step": 24050 }, { "epoch": 15.391900718484651, "grad_norm": 0.11983750015497208, "learning_rate": 6e-06, "loss": 0.0204, "step": 24100 }, { "epoch": 15.424559111691705, "grad_norm": 0.10992011427879333, "learning_rate": 6.5000000000000004e-06, "loss": 0.0207, "step": 24150 }, { "epoch": 15.45721750489876, "grad_norm": 0.11979202926158905, "learning_rate": 7e-06, "loss": 0.02, "step": 24200 }, { "epoch": 15.489875898105813, "grad_norm": 0.11368417739868164, "learning_rate": 7.500000000000001e-06, "loss": 0.0199, "step": 24250 }, { "epoch": 15.522534291312867, "grad_norm": 0.11034137010574341, "learning_rate": 8.000000000000001e-06, "loss": 0.0197, "step": 24300 }, { "epoch": 15.555192684519922, "grad_norm": 0.12382794916629791, "learning_rate": 8.5e-06, "loss": 0.0192, "step": 24350 }, { "epoch": 15.587851077726976, "grad_norm": 0.12503217160701752, "learning_rate": 9e-06, "loss": 0.0187, "step": 24400 }, { "epoch": 15.62050947093403, "grad_norm": 0.13080088794231415, "learning_rate": 9.5e-06, "loss": 0.0185, "step": 24450 }, { "epoch": 15.653167864141084, "grad_norm": 0.09857973456382751, "learning_rate": 1e-05, "loss": 0.0184, "step": 24500 }, { "epoch": 15.685826257348138, "grad_norm": 0.12122488766908646, "learning_rate": 9.996712689020382e-06, "loss": 0.0182, "step": 24550 }, { "epoch": 15.718484650555192, "grad_norm": 0.1259155422449112, "learning_rate": 9.993425378040764e-06, "loss": 0.0178, "step": 24600 }, { "epoch": 15.751143043762246, "grad_norm": 0.12195907533168793, "learning_rate": 9.990138067061145e-06, "loss": 0.0179, "step": 24650 }, { "epoch": 15.783801436969302, "grad_norm": 0.1145223081111908, "learning_rate": 9.986850756081527e-06, "loss": 0.0173, "step": 24700 }, { "epoch": 15.816459830176356, "grad_norm": 0.09642328321933746, "learning_rate": 9.983563445101906e-06, "loss": 0.0168, "step": 24750 }, { "epoch": 15.84911822338341, "grad_norm": 0.10971999168395996, "learning_rate": 9.980276134122288e-06, "loss": 0.0174, "step": 24800 }, { "epoch": 15.881776616590464, "grad_norm": 0.10064239054918289, "learning_rate": 9.976988823142671e-06, "loss": 0.0167, "step": 24850 }, { "epoch": 15.914435009797518, "grad_norm": 0.0871860682964325, "learning_rate": 9.97370151216305e-06, "loss": 0.0166, "step": 24900 }, { "epoch": 15.947093403004573, "grad_norm": 0.10426429659128189, "learning_rate": 9.970414201183432e-06, "loss": 0.0167, "step": 24950 }, { "epoch": 15.979751796211627, "grad_norm": 0.10435881465673447, "learning_rate": 9.967126890203814e-06, "loss": 0.0161, "step": 25000 }, { "epoch": 16.01241018941868, "grad_norm": 0.09956376999616623, "learning_rate": 9.963839579224195e-06, "loss": 0.0163, "step": 25050 }, { "epoch": 16.045068582625735, "grad_norm": 0.09757275134325027, "learning_rate": 9.960552268244577e-06, "loss": 0.0167, "step": 25100 }, { "epoch": 16.07772697583279, "grad_norm": 0.09212671220302582, "learning_rate": 9.957264957264958e-06, "loss": 0.0156, "step": 25150 }, { "epoch": 16.110385369039843, "grad_norm": 0.11294306069612503, "learning_rate": 9.95397764628534e-06, "loss": 0.0158, "step": 25200 }, { "epoch": 16.143043762246897, "grad_norm": 0.1380755454301834, "learning_rate": 9.950690335305721e-06, "loss": 0.0159, "step": 25250 }, { "epoch": 16.17570215545395, "grad_norm": 0.09846784174442291, "learning_rate": 9.947403024326102e-06, "loss": 0.0156, "step": 25300 }, { "epoch": 16.208360548661005, "grad_norm": 0.10531885176897049, "learning_rate": 9.944115713346484e-06, "loss": 0.0151, "step": 25350 }, { "epoch": 16.24101894186806, "grad_norm": 0.10934318602085114, "learning_rate": 9.940828402366864e-06, "loss": 0.0153, "step": 25400 }, { "epoch": 16.273677335075114, "grad_norm": 0.08792321383953094, "learning_rate": 9.937541091387247e-06, "loss": 0.0155, "step": 25450 }, { "epoch": 16.306335728282168, "grad_norm": 0.09419489651918411, "learning_rate": 9.934253780407628e-06, "loss": 0.0151, "step": 25500 }, { "epoch": 16.338994121489222, "grad_norm": 0.09185058623552322, "learning_rate": 9.930966469428008e-06, "loss": 0.0152, "step": 25550 }, { "epoch": 16.371652514696276, "grad_norm": 0.0971304401755333, "learning_rate": 9.92767915844839e-06, "loss": 0.0152, "step": 25600 }, { "epoch": 16.40431090790333, "grad_norm": 0.1052471175789833, "learning_rate": 9.924391847468771e-06, "loss": 0.0149, "step": 25650 }, { "epoch": 16.436969301110384, "grad_norm": 0.0822051614522934, "learning_rate": 9.921104536489152e-06, "loss": 0.0151, "step": 25700 }, { "epoch": 16.46962769431744, "grad_norm": 0.08607298135757446, "learning_rate": 9.917817225509534e-06, "loss": 0.015, "step": 25750 }, { "epoch": 16.502286087524492, "grad_norm": 0.11428522318601608, "learning_rate": 9.914529914529915e-06, "loss": 0.0149, "step": 25800 }, { "epoch": 16.534944480731546, "grad_norm": 0.10517720878124237, "learning_rate": 9.911242603550297e-06, "loss": 0.0145, "step": 25850 }, { "epoch": 16.5676028739386, "grad_norm": 0.08069934695959091, "learning_rate": 9.907955292570678e-06, "loss": 0.0147, "step": 25900 }, { "epoch": 16.600261267145655, "grad_norm": 0.070205457508564, "learning_rate": 9.90466798159106e-06, "loss": 0.0148, "step": 25950 }, { "epoch": 16.632919660352712, "grad_norm": 0.10561920702457428, "learning_rate": 9.90138067061144e-06, "loss": 0.0151, "step": 26000 }, { "epoch": 16.665578053559766, "grad_norm": 0.09528687596321106, "learning_rate": 9.898093359631823e-06, "loss": 0.0148, "step": 26050 }, { "epoch": 16.69823644676682, "grad_norm": 0.08869145065546036, "learning_rate": 9.894806048652204e-06, "loss": 0.0141, "step": 26100 }, { "epoch": 16.730894839973875, "grad_norm": 0.10699061304330826, "learning_rate": 9.891518737672584e-06, "loss": 0.0144, "step": 26150 }, { "epoch": 16.76355323318093, "grad_norm": 0.0824965164065361, "learning_rate": 9.888231426692965e-06, "loss": 0.0141, "step": 26200 }, { "epoch": 16.796211626387983, "grad_norm": 0.09220811724662781, "learning_rate": 9.884944115713347e-06, "loss": 0.0145, "step": 26250 }, { "epoch": 16.828870019595037, "grad_norm": 0.08117767423391342, "learning_rate": 9.88165680473373e-06, "loss": 0.0145, "step": 26300 }, { "epoch": 16.86152841280209, "grad_norm": 0.09427885711193085, "learning_rate": 9.87836949375411e-06, "loss": 0.014, "step": 26350 }, { "epoch": 16.894186806009145, "grad_norm": 0.07976298779249191, "learning_rate": 9.875082182774491e-06, "loss": 0.0139, "step": 26400 }, { "epoch": 16.9268451992162, "grad_norm": 0.09682408720254898, "learning_rate": 9.871794871794872e-06, "loss": 0.014, "step": 26450 }, { "epoch": 16.959503592423253, "grad_norm": 0.0674891471862793, "learning_rate": 9.868507560815254e-06, "loss": 0.0138, "step": 26500 }, { "epoch": 16.992161985630307, "grad_norm": 0.07873521000146866, "learning_rate": 9.865220249835635e-06, "loss": 0.0138, "step": 26550 }, { "epoch": 17.02482037883736, "grad_norm": 0.0872296616435051, "learning_rate": 9.861932938856017e-06, "loss": 0.0137, "step": 26600 }, { "epoch": 17.057478772044416, "grad_norm": 0.080799400806427, "learning_rate": 9.858645627876398e-06, "loss": 0.014, "step": 26650 }, { "epoch": 17.09013716525147, "grad_norm": 0.10432834923267365, "learning_rate": 9.85535831689678e-06, "loss": 0.0139, "step": 26700 }, { "epoch": 17.122795558458524, "grad_norm": 0.10217973589897156, "learning_rate": 9.852071005917161e-06, "loss": 0.0137, "step": 26750 }, { "epoch": 17.155453951665578, "grad_norm": 0.08877214789390564, "learning_rate": 9.848783694937541e-06, "loss": 0.0137, "step": 26800 }, { "epoch": 17.188112344872632, "grad_norm": 0.10301809012889862, "learning_rate": 9.845496383957922e-06, "loss": 0.0135, "step": 26850 }, { "epoch": 17.220770738079686, "grad_norm": 0.1152462363243103, "learning_rate": 9.842209072978306e-06, "loss": 0.0135, "step": 26900 }, { "epoch": 17.25342913128674, "grad_norm": 0.08223509788513184, "learning_rate": 9.838921761998685e-06, "loss": 0.0137, "step": 26950 }, { "epoch": 17.286087524493794, "grad_norm": 0.09059303998947144, "learning_rate": 9.835634451019067e-06, "loss": 0.0138, "step": 27000 }, { "epoch": 17.31874591770085, "grad_norm": 0.07942213863134384, "learning_rate": 9.832347140039448e-06, "loss": 0.014, "step": 27050 }, { "epoch": 17.351404310907903, "grad_norm": 0.0942160040140152, "learning_rate": 9.82905982905983e-06, "loss": 0.0133, "step": 27100 }, { "epoch": 17.384062704114957, "grad_norm": 0.09322342276573181, "learning_rate": 9.825772518080211e-06, "loss": 0.014, "step": 27150 }, { "epoch": 17.41672109732201, "grad_norm": 0.0794295072555542, "learning_rate": 9.822485207100593e-06, "loss": 0.0133, "step": 27200 }, { "epoch": 17.449379490529065, "grad_norm": 0.09018723666667938, "learning_rate": 9.819197896120974e-06, "loss": 0.0138, "step": 27250 }, { "epoch": 17.48203788373612, "grad_norm": 0.07898092269897461, "learning_rate": 9.815910585141355e-06, "loss": 0.0138, "step": 27300 }, { "epoch": 17.514696276943173, "grad_norm": 0.08042161166667938, "learning_rate": 9.812623274161737e-06, "loss": 0.0132, "step": 27350 }, { "epoch": 17.547354670150227, "grad_norm": 0.08876854926347733, "learning_rate": 9.809335963182118e-06, "loss": 0.0133, "step": 27400 }, { "epoch": 17.58001306335728, "grad_norm": 0.11614891141653061, "learning_rate": 9.806048652202498e-06, "loss": 0.0135, "step": 27450 }, { "epoch": 17.612671456564335, "grad_norm": 0.11040101200342178, "learning_rate": 9.802761341222881e-06, "loss": 0.0134, "step": 27500 }, { "epoch": 17.645329849771393, "grad_norm": 0.07913624495267868, "learning_rate": 9.799474030243263e-06, "loss": 0.0132, "step": 27550 }, { "epoch": 17.677988242978447, "grad_norm": 0.0810326486825943, "learning_rate": 9.796186719263643e-06, "loss": 0.0133, "step": 27600 }, { "epoch": 17.7106466361855, "grad_norm": 0.08804348856210709, "learning_rate": 9.792899408284024e-06, "loss": 0.0128, "step": 27650 }, { "epoch": 17.743305029392555, "grad_norm": 0.10724814981222153, "learning_rate": 9.789612097304405e-06, "loss": 0.0131, "step": 27700 }, { "epoch": 17.77596342259961, "grad_norm": 0.09030251204967499, "learning_rate": 9.786324786324787e-06, "loss": 0.0137, "step": 27750 }, { "epoch": 17.808621815806664, "grad_norm": 0.08435764908790588, "learning_rate": 9.783037475345168e-06, "loss": 0.0134, "step": 27800 }, { "epoch": 17.841280209013718, "grad_norm": 0.06447231024503708, "learning_rate": 9.77975016436555e-06, "loss": 0.0132, "step": 27850 }, { "epoch": 17.873938602220772, "grad_norm": 0.07145614176988602, "learning_rate": 9.776462853385931e-06, "loss": 0.0128, "step": 27900 }, { "epoch": 17.906596995427826, "grad_norm": 0.09933588653802872, "learning_rate": 9.773175542406313e-06, "loss": 0.0128, "step": 27950 }, { "epoch": 17.93925538863488, "grad_norm": 0.09780805557966232, "learning_rate": 9.769888231426694e-06, "loss": 0.0133, "step": 28000 }, { "epoch": 17.971913781841934, "grad_norm": 0.0783376470208168, "learning_rate": 9.766600920447074e-06, "loss": 0.0126, "step": 28050 }, { "epoch": 18.004572175048988, "grad_norm": 0.08968358486890793, "learning_rate": 9.763313609467457e-06, "loss": 0.0127, "step": 28100 }, { "epoch": 18.037230568256042, "grad_norm": 0.09786919504404068, "learning_rate": 9.760026298487839e-06, "loss": 0.0129, "step": 28150 }, { "epoch": 18.069888961463096, "grad_norm": 0.07008064538240433, "learning_rate": 9.756738987508218e-06, "loss": 0.0128, "step": 28200 }, { "epoch": 18.10254735467015, "grad_norm": 0.09454207122325897, "learning_rate": 9.7534516765286e-06, "loss": 0.0129, "step": 28250 }, { "epoch": 18.135205747877205, "grad_norm": 0.07008275389671326, "learning_rate": 9.750164365548981e-06, "loss": 0.0126, "step": 28300 }, { "epoch": 18.16786414108426, "grad_norm": 0.09741360694169998, "learning_rate": 9.746877054569363e-06, "loss": 0.0127, "step": 28350 }, { "epoch": 18.200522534291313, "grad_norm": 0.092861607670784, "learning_rate": 9.743589743589744e-06, "loss": 0.0125, "step": 28400 }, { "epoch": 18.233180927498367, "grad_norm": 0.07552482187747955, "learning_rate": 9.740302432610126e-06, "loss": 0.0128, "step": 28450 }, { "epoch": 18.26583932070542, "grad_norm": 0.08930996060371399, "learning_rate": 9.737015121630507e-06, "loss": 0.0128, "step": 28500 }, { "epoch": 18.298497713912475, "grad_norm": 0.08118516951799393, "learning_rate": 9.733727810650888e-06, "loss": 0.0131, "step": 28550 }, { "epoch": 18.33115610711953, "grad_norm": 0.09239883720874786, "learning_rate": 9.73044049967127e-06, "loss": 0.013, "step": 28600 }, { "epoch": 18.363814500326583, "grad_norm": 0.0745251476764679, "learning_rate": 9.727153188691651e-06, "loss": 0.013, "step": 28650 }, { "epoch": 18.396472893533637, "grad_norm": 0.08314284682273865, "learning_rate": 9.723865877712033e-06, "loss": 0.0123, "step": 28700 }, { "epoch": 18.42913128674069, "grad_norm": 0.07523535937070847, "learning_rate": 9.720578566732414e-06, "loss": 0.0127, "step": 28750 }, { "epoch": 18.461789679947746, "grad_norm": 0.09558777511119843, "learning_rate": 9.717291255752796e-06, "loss": 0.0129, "step": 28800 }, { "epoch": 18.4944480731548, "grad_norm": 0.08536320924758911, "learning_rate": 9.714003944773175e-06, "loss": 0.0127, "step": 28850 }, { "epoch": 18.527106466361854, "grad_norm": 0.07737783342599869, "learning_rate": 9.710716633793557e-06, "loss": 0.0122, "step": 28900 }, { "epoch": 18.559764859568908, "grad_norm": 0.08645167946815491, "learning_rate": 9.70742932281394e-06, "loss": 0.012, "step": 28950 }, { "epoch": 18.592423252775962, "grad_norm": 0.06443276256322861, "learning_rate": 9.70414201183432e-06, "loss": 0.0128, "step": 29000 }, { "epoch": 18.625081645983016, "grad_norm": 0.08362197875976562, "learning_rate": 9.700854700854701e-06, "loss": 0.0128, "step": 29050 }, { "epoch": 18.65774003919007, "grad_norm": 0.08388455957174301, "learning_rate": 9.697567389875083e-06, "loss": 0.0129, "step": 29100 }, { "epoch": 18.690398432397124, "grad_norm": 0.08566679060459137, "learning_rate": 9.694280078895464e-06, "loss": 0.0126, "step": 29150 }, { "epoch": 18.723056825604182, "grad_norm": 0.08044921606779099, "learning_rate": 9.690992767915846e-06, "loss": 0.0128, "step": 29200 }, { "epoch": 18.755715218811236, "grad_norm": 0.07965279370546341, "learning_rate": 9.687705456936227e-06, "loss": 0.0125, "step": 29250 }, { "epoch": 18.78837361201829, "grad_norm": 0.077300526201725, "learning_rate": 9.684418145956609e-06, "loss": 0.0123, "step": 29300 }, { "epoch": 18.821032005225344, "grad_norm": 0.08230002969503403, "learning_rate": 9.68113083497699e-06, "loss": 0.0124, "step": 29350 }, { "epoch": 18.8536903984324, "grad_norm": 0.08845552057027817, "learning_rate": 9.677843523997371e-06, "loss": 0.0126, "step": 29400 }, { "epoch": 18.886348791639453, "grad_norm": 0.09909436851739883, "learning_rate": 9.674556213017751e-06, "loss": 0.0125, "step": 29450 }, { "epoch": 18.919007184846507, "grad_norm": 0.06949647516012192, "learning_rate": 9.671268902038133e-06, "loss": 0.012, "step": 29500 }, { "epoch": 18.95166557805356, "grad_norm": 0.07505377382040024, "learning_rate": 9.667981591058516e-06, "loss": 0.0124, "step": 29550 }, { "epoch": 18.984323971260615, "grad_norm": 0.10071639716625214, "learning_rate": 9.664694280078897e-06, "loss": 0.0123, "step": 29600 }, { "epoch": 19.01698236446767, "grad_norm": 0.06843467056751251, "learning_rate": 9.661406969099277e-06, "loss": 0.0121, "step": 29650 }, { "epoch": 19.049640757674723, "grad_norm": 0.08361203968524933, "learning_rate": 9.658119658119659e-06, "loss": 0.0123, "step": 29700 }, { "epoch": 19.082299150881777, "grad_norm": 0.0840635672211647, "learning_rate": 9.65483234714004e-06, "loss": 0.0123, "step": 29750 }, { "epoch": 19.11495754408883, "grad_norm": 0.07549753040075302, "learning_rate": 9.651545036160421e-06, "loss": 0.0121, "step": 29800 }, { "epoch": 19.147615937295885, "grad_norm": 0.08807700872421265, "learning_rate": 9.648257725180803e-06, "loss": 0.0122, "step": 29850 }, { "epoch": 19.18027433050294, "grad_norm": 0.07554081082344055, "learning_rate": 9.644970414201184e-06, "loss": 0.0124, "step": 29900 }, { "epoch": 19.212932723709994, "grad_norm": 0.07595884799957275, "learning_rate": 9.641683103221566e-06, "loss": 0.0119, "step": 29950 }, { "epoch": 19.245591116917048, "grad_norm": 0.09061748534440994, "learning_rate": 9.638395792241947e-06, "loss": 0.0123, "step": 30000 }, { "epoch": 19.278249510124102, "grad_norm": 0.079330675303936, "learning_rate": 9.635108481262329e-06, "loss": 0.0118, "step": 30050 }, { "epoch": 19.310907903331156, "grad_norm": 0.08413425087928772, "learning_rate": 9.631821170282708e-06, "loss": 0.0123, "step": 30100 }, { "epoch": 19.34356629653821, "grad_norm": 0.08711785823106766, "learning_rate": 9.628533859303092e-06, "loss": 0.0121, "step": 30150 }, { "epoch": 19.376224689745264, "grad_norm": 0.06705302745103836, "learning_rate": 9.625246548323473e-06, "loss": 0.0118, "step": 30200 }, { "epoch": 19.40888308295232, "grad_norm": 0.08774611353874207, "learning_rate": 9.621959237343853e-06, "loss": 0.0121, "step": 30250 }, { "epoch": 19.441541476159372, "grad_norm": 0.07749926298856735, "learning_rate": 9.618671926364234e-06, "loss": 0.0123, "step": 30300 }, { "epoch": 19.474199869366426, "grad_norm": 0.07043664157390594, "learning_rate": 9.615384615384616e-06, "loss": 0.0117, "step": 30350 }, { "epoch": 19.50685826257348, "grad_norm": 0.09341815859079361, "learning_rate": 9.612097304404997e-06, "loss": 0.0121, "step": 30400 }, { "epoch": 19.539516655780535, "grad_norm": 0.06324969977140427, "learning_rate": 9.608809993425379e-06, "loss": 0.0121, "step": 30450 }, { "epoch": 19.57217504898759, "grad_norm": 0.0659535601735115, "learning_rate": 9.60552268244576e-06, "loss": 0.0124, "step": 30500 }, { "epoch": 19.604833442194643, "grad_norm": 0.08671879023313522, "learning_rate": 9.602235371466142e-06, "loss": 0.0124, "step": 30550 }, { "epoch": 19.637491835401697, "grad_norm": 0.08552669733762741, "learning_rate": 9.598948060486523e-06, "loss": 0.0122, "step": 30600 }, { "epoch": 19.67015022860875, "grad_norm": 0.07820162922143936, "learning_rate": 9.595660749506904e-06, "loss": 0.0118, "step": 30650 }, { "epoch": 19.702808621815805, "grad_norm": 0.07291736453771591, "learning_rate": 9.592373438527286e-06, "loss": 0.0122, "step": 30700 }, { "epoch": 19.73546701502286, "grad_norm": 0.07649485766887665, "learning_rate": 9.589086127547667e-06, "loss": 0.0121, "step": 30750 }, { "epoch": 19.768125408229913, "grad_norm": 0.09021729230880737, "learning_rate": 9.585798816568049e-06, "loss": 0.0121, "step": 30800 }, { "epoch": 19.80078380143697, "grad_norm": 0.10497128963470459, "learning_rate": 9.58251150558843e-06, "loss": 0.0121, "step": 30850 }, { "epoch": 19.833442194644025, "grad_norm": 0.07241571694612503, "learning_rate": 9.57922419460881e-06, "loss": 0.0121, "step": 30900 }, { "epoch": 19.86610058785108, "grad_norm": 0.09371298551559448, "learning_rate": 9.575936883629191e-06, "loss": 0.012, "step": 30950 }, { "epoch": 19.898758981058133, "grad_norm": 0.06992656737565994, "learning_rate": 9.572649572649575e-06, "loss": 0.012, "step": 31000 }, { "epoch": 19.931417374265187, "grad_norm": 0.0780627653002739, "learning_rate": 9.569362261669954e-06, "loss": 0.0121, "step": 31050 }, { "epoch": 19.96407576747224, "grad_norm": 0.06465399265289307, "learning_rate": 9.566074950690336e-06, "loss": 0.0116, "step": 31100 }, { "epoch": 19.996734160679296, "grad_norm": 0.0792844220995903, "learning_rate": 9.562787639710717e-06, "loss": 0.0116, "step": 31150 }, { "epoch": 20.02939255388635, "grad_norm": 0.07747771590948105, "learning_rate": 9.559500328731099e-06, "loss": 0.0116, "step": 31200 }, { "epoch": 20.062050947093404, "grad_norm": 0.07056711614131927, "learning_rate": 9.55621301775148e-06, "loss": 0.0117, "step": 31250 }, { "epoch": 20.094709340300458, "grad_norm": 0.07007728517055511, "learning_rate": 9.552925706771862e-06, "loss": 0.0114, "step": 31300 }, { "epoch": 20.127367733507512, "grad_norm": 0.06829990446567535, "learning_rate": 9.549638395792243e-06, "loss": 0.0123, "step": 31350 }, { "epoch": 20.160026126714566, "grad_norm": 0.07261107116937637, "learning_rate": 9.546351084812625e-06, "loss": 0.0113, "step": 31400 }, { "epoch": 20.19268451992162, "grad_norm": 0.0791664719581604, "learning_rate": 9.543063773833006e-06, "loss": 0.0116, "step": 31450 }, { "epoch": 20.225342913128674, "grad_norm": 0.08574667572975159, "learning_rate": 9.539776462853386e-06, "loss": 0.0119, "step": 31500 }, { "epoch": 20.25800130633573, "grad_norm": 0.0764293447136879, "learning_rate": 9.536489151873767e-06, "loss": 0.0117, "step": 31550 }, { "epoch": 20.290659699542783, "grad_norm": 0.0843268409371376, "learning_rate": 9.53320184089415e-06, "loss": 0.0116, "step": 31600 }, { "epoch": 20.323318092749837, "grad_norm": 0.08600505441427231, "learning_rate": 9.52991452991453e-06, "loss": 0.0118, "step": 31650 }, { "epoch": 20.35597648595689, "grad_norm": 0.08768628537654877, "learning_rate": 9.526627218934912e-06, "loss": 0.0117, "step": 31700 }, { "epoch": 20.388634879163945, "grad_norm": 0.08994566649198532, "learning_rate": 9.523339907955293e-06, "loss": 0.0121, "step": 31750 }, { "epoch": 20.421293272371, "grad_norm": 0.07525193691253662, "learning_rate": 9.520052596975674e-06, "loss": 0.0121, "step": 31800 }, { "epoch": 20.453951665578053, "grad_norm": 0.09931448847055435, "learning_rate": 9.516765285996056e-06, "loss": 0.0112, "step": 31850 }, { "epoch": 20.486610058785107, "grad_norm": 0.06344626843929291, "learning_rate": 9.513477975016437e-06, "loss": 0.0119, "step": 31900 }, { "epoch": 20.51926845199216, "grad_norm": 0.0673469752073288, "learning_rate": 9.510190664036819e-06, "loss": 0.0114, "step": 31950 }, { "epoch": 20.551926845199215, "grad_norm": 0.08714130520820618, "learning_rate": 9.5069033530572e-06, "loss": 0.0116, "step": 32000 }, { "epoch": 20.58458523840627, "grad_norm": 0.06989229470491409, "learning_rate": 9.503616042077582e-06, "loss": 0.0123, "step": 32050 }, { "epoch": 20.617243631613324, "grad_norm": 0.0799252986907959, "learning_rate": 9.500328731097963e-06, "loss": 0.0115, "step": 32100 }, { "epoch": 20.649902024820378, "grad_norm": 0.07591370493173599, "learning_rate": 9.497041420118343e-06, "loss": 0.012, "step": 32150 }, { "epoch": 20.682560418027432, "grad_norm": 0.08256584405899048, "learning_rate": 9.493754109138726e-06, "loss": 0.0115, "step": 32200 }, { "epoch": 20.715218811234486, "grad_norm": 0.07658610492944717, "learning_rate": 9.490466798159108e-06, "loss": 0.0117, "step": 32250 }, { "epoch": 20.74787720444154, "grad_norm": 0.08041895925998688, "learning_rate": 9.487179487179487e-06, "loss": 0.0114, "step": 32300 }, { "epoch": 20.780535597648594, "grad_norm": 0.08026766777038574, "learning_rate": 9.483892176199869e-06, "loss": 0.0115, "step": 32350 }, { "epoch": 20.81319399085565, "grad_norm": 0.07113494724035263, "learning_rate": 9.48060486522025e-06, "loss": 0.012, "step": 32400 }, { "epoch": 20.845852384062702, "grad_norm": 0.09265585243701935, "learning_rate": 9.477317554240632e-06, "loss": 0.0115, "step": 32450 }, { "epoch": 20.87851077726976, "grad_norm": 0.0856519415974617, "learning_rate": 9.474030243261013e-06, "loss": 0.0116, "step": 32500 }, { "epoch": 20.911169170476814, "grad_norm": 0.08142653852701187, "learning_rate": 9.470742932281395e-06, "loss": 0.0116, "step": 32550 }, { "epoch": 20.943827563683868, "grad_norm": 0.06179051101207733, "learning_rate": 9.467455621301776e-06, "loss": 0.0121, "step": 32600 }, { "epoch": 20.976485956890922, "grad_norm": 0.07312364876270294, "learning_rate": 9.464168310322158e-06, "loss": 0.0113, "step": 32650 }, { "epoch": 21.009144350097976, "grad_norm": 0.06998687982559204, "learning_rate": 9.460880999342539e-06, "loss": 0.0112, "step": 32700 }, { "epoch": 21.04180274330503, "grad_norm": 0.07416413724422455, "learning_rate": 9.457593688362919e-06, "loss": 0.0117, "step": 32750 }, { "epoch": 21.074461136512085, "grad_norm": 0.07157248258590698, "learning_rate": 9.4543063773833e-06, "loss": 0.0111, "step": 32800 }, { "epoch": 21.10711952971914, "grad_norm": 0.07252523303031921, "learning_rate": 9.451019066403683e-06, "loss": 0.0111, "step": 32850 }, { "epoch": 21.139777922926193, "grad_norm": 0.08431415259838104, "learning_rate": 9.447731755424065e-06, "loss": 0.0115, "step": 32900 }, { "epoch": 21.172436316133247, "grad_norm": 0.06362010538578033, "learning_rate": 9.444444444444445e-06, "loss": 0.0114, "step": 32950 }, { "epoch": 21.2050947093403, "grad_norm": 0.07112806290388107, "learning_rate": 9.441157133464826e-06, "loss": 0.0116, "step": 33000 }, { "epoch": 21.237753102547355, "grad_norm": 0.08300979435443878, "learning_rate": 9.43786982248521e-06, "loss": 0.0109, "step": 33050 }, { "epoch": 21.27041149575441, "grad_norm": 0.07988585531711578, "learning_rate": 9.434582511505589e-06, "loss": 0.0112, "step": 33100 }, { "epoch": 21.303069888961463, "grad_norm": 0.09664539992809296, "learning_rate": 9.43129520052597e-06, "loss": 0.0109, "step": 33150 }, { "epoch": 21.335728282168517, "grad_norm": 0.08490543812513351, "learning_rate": 9.428007889546352e-06, "loss": 0.0118, "step": 33200 }, { "epoch": 21.36838667537557, "grad_norm": 0.06111636757850647, "learning_rate": 9.424720578566733e-06, "loss": 0.011, "step": 33250 }, { "epoch": 21.401045068582626, "grad_norm": 0.07294650375843048, "learning_rate": 9.421433267587115e-06, "loss": 0.0114, "step": 33300 }, { "epoch": 21.43370346178968, "grad_norm": 0.07111258059740067, "learning_rate": 9.418145956607496e-06, "loss": 0.0112, "step": 33350 }, { "epoch": 21.466361854996734, "grad_norm": 0.09822231531143188, "learning_rate": 9.414858645627876e-06, "loss": 0.0118, "step": 33400 }, { "epoch": 21.499020248203788, "grad_norm": 0.06523829698562622, "learning_rate": 9.411571334648259e-06, "loss": 0.0113, "step": 33450 }, { "epoch": 21.531678641410842, "grad_norm": 0.06331858038902283, "learning_rate": 9.40828402366864e-06, "loss": 0.0114, "step": 33500 }, { "epoch": 21.564337034617896, "grad_norm": 0.07485601305961609, "learning_rate": 9.40499671268902e-06, "loss": 0.0118, "step": 33550 }, { "epoch": 21.59699542782495, "grad_norm": 0.07524490356445312, "learning_rate": 9.401709401709402e-06, "loss": 0.0111, "step": 33600 }, { "epoch": 21.629653821032004, "grad_norm": 0.07027073204517365, "learning_rate": 9.398422090729783e-06, "loss": 0.0114, "step": 33650 }, { "epoch": 21.66231221423906, "grad_norm": 0.06455384194850922, "learning_rate": 9.395134779750165e-06, "loss": 0.0114, "step": 33700 }, { "epoch": 21.694970607446113, "grad_norm": 0.07016217708587646, "learning_rate": 9.391847468770546e-06, "loss": 0.0113, "step": 33750 }, { "epoch": 21.727629000653167, "grad_norm": 0.08227501064538956, "learning_rate": 9.388560157790928e-06, "loss": 0.0116, "step": 33800 }, { "epoch": 21.76028739386022, "grad_norm": 0.06575901806354523, "learning_rate": 9.385272846811309e-06, "loss": 0.0108, "step": 33850 }, { "epoch": 21.792945787067275, "grad_norm": 0.07008541375398636, "learning_rate": 9.38198553583169e-06, "loss": 0.0115, "step": 33900 }, { "epoch": 21.82560418027433, "grad_norm": 0.07634240388870239, "learning_rate": 9.378698224852072e-06, "loss": 0.0113, "step": 33950 }, { "epoch": 21.858262573481383, "grad_norm": 0.06252789497375488, "learning_rate": 9.375410913872453e-06, "loss": 0.0116, "step": 34000 }, { "epoch": 21.890920966688437, "grad_norm": 0.07221711426973343, "learning_rate": 9.372123602892835e-06, "loss": 0.0114, "step": 34050 }, { "epoch": 21.92357935989549, "grad_norm": 0.07890020310878754, "learning_rate": 9.368836291913216e-06, "loss": 0.0114, "step": 34100 }, { "epoch": 21.95623775310255, "grad_norm": 0.0640934631228447, "learning_rate": 9.365548980933598e-06, "loss": 0.011, "step": 34150 }, { "epoch": 21.988896146309603, "grad_norm": 0.06607390195131302, "learning_rate": 9.362261669953978e-06, "loss": 0.0108, "step": 34200 }, { "epoch": 22.021554539516657, "grad_norm": 0.07007242739200592, "learning_rate": 9.358974358974359e-06, "loss": 0.0111, "step": 34250 }, { "epoch": 22.05421293272371, "grad_norm": 0.07252377271652222, "learning_rate": 9.355687047994742e-06, "loss": 0.0108, "step": 34300 }, { "epoch": 22.086871325930765, "grad_norm": 0.07591360062360764, "learning_rate": 9.352399737015122e-06, "loss": 0.0108, "step": 34350 }, { "epoch": 22.11952971913782, "grad_norm": 0.06834853440523148, "learning_rate": 9.349112426035503e-06, "loss": 0.0112, "step": 34400 }, { "epoch": 22.152188112344874, "grad_norm": 0.06629569083452225, "learning_rate": 9.345825115055885e-06, "loss": 0.0111, "step": 34450 }, { "epoch": 22.184846505551928, "grad_norm": 0.09552879631519318, "learning_rate": 9.342537804076266e-06, "loss": 0.0106, "step": 34500 }, { "epoch": 22.217504898758982, "grad_norm": 0.0836816355586052, "learning_rate": 9.339250493096648e-06, "loss": 0.0108, "step": 34550 }, { "epoch": 22.250163291966036, "grad_norm": 0.08128397911787033, "learning_rate": 9.33596318211703e-06, "loss": 0.0114, "step": 34600 }, { "epoch": 22.28282168517309, "grad_norm": 0.09292303025722504, "learning_rate": 9.33267587113741e-06, "loss": 0.0112, "step": 34650 }, { "epoch": 22.315480078380144, "grad_norm": 0.06572126597166061, "learning_rate": 9.329388560157792e-06, "loss": 0.0114, "step": 34700 }, { "epoch": 22.3481384715872, "grad_norm": 0.06700612604618073, "learning_rate": 9.326101249178174e-06, "loss": 0.011, "step": 34750 }, { "epoch": 22.380796864794252, "grad_norm": 0.06546659022569656, "learning_rate": 9.322813938198553e-06, "loss": 0.0112, "step": 34800 }, { "epoch": 22.413455258001306, "grad_norm": 0.07440585643053055, "learning_rate": 9.319526627218935e-06, "loss": 0.011, "step": 34850 }, { "epoch": 22.44611365120836, "grad_norm": 0.0840035080909729, "learning_rate": 9.316239316239318e-06, "loss": 0.0114, "step": 34900 }, { "epoch": 22.478772044415415, "grad_norm": 0.06784361600875854, "learning_rate": 9.312952005259698e-06, "loss": 0.0108, "step": 34950 }, { "epoch": 22.51143043762247, "grad_norm": 0.05288035422563553, "learning_rate": 9.309664694280079e-06, "loss": 0.0109, "step": 35000 }, { "epoch": 22.544088830829523, "grad_norm": 0.08191203325986862, "learning_rate": 9.30637738330046e-06, "loss": 0.0109, "step": 35050 }, { "epoch": 22.576747224036577, "grad_norm": 0.06293505430221558, "learning_rate": 9.303090072320842e-06, "loss": 0.0109, "step": 35100 }, { "epoch": 22.60940561724363, "grad_norm": 0.058216482400894165, "learning_rate": 9.299802761341223e-06, "loss": 0.0109, "step": 35150 }, { "epoch": 22.642064010450685, "grad_norm": 0.07931092381477356, "learning_rate": 9.296515450361605e-06, "loss": 0.0109, "step": 35200 }, { "epoch": 22.67472240365774, "grad_norm": 0.09247830510139465, "learning_rate": 9.293228139381986e-06, "loss": 0.0112, "step": 35250 }, { "epoch": 22.707380796864793, "grad_norm": 0.05982724577188492, "learning_rate": 9.289940828402368e-06, "loss": 0.0114, "step": 35300 }, { "epoch": 22.740039190071847, "grad_norm": 0.06627553701400757, "learning_rate": 9.28665351742275e-06, "loss": 0.0109, "step": 35350 }, { "epoch": 22.7726975832789, "grad_norm": 0.07697000354528427, "learning_rate": 9.28336620644313e-06, "loss": 0.0111, "step": 35400 }, { "epoch": 22.805355976485956, "grad_norm": 0.06429773569107056, "learning_rate": 9.28007889546351e-06, "loss": 0.0111, "step": 35450 }, { "epoch": 22.83801436969301, "grad_norm": 0.07256831228733063, "learning_rate": 9.276791584483894e-06, "loss": 0.0106, "step": 35500 }, { "epoch": 22.870672762900064, "grad_norm": 0.06597444415092468, "learning_rate": 9.273504273504275e-06, "loss": 0.0114, "step": 35550 }, { "epoch": 22.903331156107118, "grad_norm": 0.06517421454191208, "learning_rate": 9.270216962524655e-06, "loss": 0.011, "step": 35600 }, { "epoch": 22.935989549314172, "grad_norm": 0.07113608717918396, "learning_rate": 9.266929651545036e-06, "loss": 0.0111, "step": 35650 }, { "epoch": 22.96864794252123, "grad_norm": 0.07608538866043091, "learning_rate": 9.263642340565418e-06, "loss": 0.0107, "step": 35700 }, { "epoch": 23.001306335728284, "grad_norm": 0.09357374906539917, "learning_rate": 9.2603550295858e-06, "loss": 0.0108, "step": 35750 }, { "epoch": 23.033964728935338, "grad_norm": 0.07360487431287766, "learning_rate": 9.25706771860618e-06, "loss": 0.0107, "step": 35800 }, { "epoch": 23.066623122142392, "grad_norm": 0.0792042464017868, "learning_rate": 9.253780407626562e-06, "loss": 0.011, "step": 35850 }, { "epoch": 23.099281515349446, "grad_norm": 0.06482048332691193, "learning_rate": 9.250493096646944e-06, "loss": 0.0112, "step": 35900 }, { "epoch": 23.1319399085565, "grad_norm": 0.06572939455509186, "learning_rate": 9.247205785667325e-06, "loss": 0.0107, "step": 35950 }, { "epoch": 23.164598301763554, "grad_norm": 0.06663377583026886, "learning_rate": 9.243918474687706e-06, "loss": 0.0107, "step": 36000 }, { "epoch": 23.19725669497061, "grad_norm": 0.06091673672199249, "learning_rate": 9.240631163708086e-06, "loss": 0.0108, "step": 36050 }, { "epoch": 23.229915088177663, "grad_norm": 0.08492511510848999, "learning_rate": 9.23734385272847e-06, "loss": 0.0112, "step": 36100 }, { "epoch": 23.262573481384717, "grad_norm": 0.08291380107402802, "learning_rate": 9.234056541748851e-06, "loss": 0.0108, "step": 36150 }, { "epoch": 23.29523187459177, "grad_norm": 0.0624297559261322, "learning_rate": 9.230769230769232e-06, "loss": 0.0109, "step": 36200 }, { "epoch": 23.327890267798825, "grad_norm": 0.0811142697930336, "learning_rate": 9.227481919789612e-06, "loss": 0.0108, "step": 36250 }, { "epoch": 23.36054866100588, "grad_norm": 0.0685185045003891, "learning_rate": 9.224194608809994e-06, "loss": 0.0103, "step": 36300 }, { "epoch": 23.393207054212933, "grad_norm": 0.07936639338731766, "learning_rate": 9.220907297830377e-06, "loss": 0.011, "step": 36350 }, { "epoch": 23.425865447419987, "grad_norm": 0.07991329580545425, "learning_rate": 9.217619986850756e-06, "loss": 0.0109, "step": 36400 }, { "epoch": 23.45852384062704, "grad_norm": 0.06367255747318268, "learning_rate": 9.214332675871138e-06, "loss": 0.0103, "step": 36450 }, { "epoch": 23.491182233834095, "grad_norm": 0.0692523941397667, "learning_rate": 9.21104536489152e-06, "loss": 0.011, "step": 36500 }, { "epoch": 23.52384062704115, "grad_norm": 0.07745783030986786, "learning_rate": 9.2077580539119e-06, "loss": 0.0107, "step": 36550 }, { "epoch": 23.556499020248204, "grad_norm": 0.07395331561565399, "learning_rate": 9.204470742932282e-06, "loss": 0.0103, "step": 36600 }, { "epoch": 23.589157413455258, "grad_norm": 0.07659647613763809, "learning_rate": 9.201183431952664e-06, "loss": 0.0107, "step": 36650 }, { "epoch": 23.621815806662312, "grad_norm": 0.08041811734437943, "learning_rate": 9.197896120973045e-06, "loss": 0.0108, "step": 36700 }, { "epoch": 23.654474199869366, "grad_norm": 0.0716087818145752, "learning_rate": 9.194608809993427e-06, "loss": 0.011, "step": 36750 }, { "epoch": 23.68713259307642, "grad_norm": 0.08389966189861298, "learning_rate": 9.191321499013808e-06, "loss": 0.0108, "step": 36800 }, { "epoch": 23.719790986283474, "grad_norm": 0.06117480620741844, "learning_rate": 9.188034188034188e-06, "loss": 0.0108, "step": 36850 }, { "epoch": 23.75244937949053, "grad_norm": 0.07161203771829605, "learning_rate": 9.18474687705457e-06, "loss": 0.0107, "step": 36900 }, { "epoch": 23.785107772697582, "grad_norm": 0.06900031864643097, "learning_rate": 9.181459566074952e-06, "loss": 0.0107, "step": 36950 }, { "epoch": 23.817766165904636, "grad_norm": 0.0728040263056755, "learning_rate": 9.178172255095332e-06, "loss": 0.0108, "step": 37000 }, { "epoch": 23.85042455911169, "grad_norm": 0.06457971781492233, "learning_rate": 9.174884944115714e-06, "loss": 0.0104, "step": 37050 }, { "epoch": 23.883082952318745, "grad_norm": 0.06282978504896164, "learning_rate": 9.171597633136095e-06, "loss": 0.0109, "step": 37100 }, { "epoch": 23.9157413455258, "grad_norm": 0.08687738329172134, "learning_rate": 9.168310322156477e-06, "loss": 0.0107, "step": 37150 }, { "epoch": 23.948399738732853, "grad_norm": 0.06982048600912094, "learning_rate": 9.165023011176858e-06, "loss": 0.0106, "step": 37200 }, { "epoch": 23.981058131939907, "grad_norm": 0.08296804875135422, "learning_rate": 9.16173570019724e-06, "loss": 0.0104, "step": 37250 }, { "epoch": 24.01371652514696, "grad_norm": 0.06745100021362305, "learning_rate": 9.158448389217621e-06, "loss": 0.0104, "step": 37300 }, { "epoch": 24.04637491835402, "grad_norm": 0.05570142716169357, "learning_rate": 9.155161078238002e-06, "loss": 0.0109, "step": 37350 }, { "epoch": 24.079033311561073, "grad_norm": 0.05718238651752472, "learning_rate": 9.151873767258384e-06, "loss": 0.0104, "step": 37400 }, { "epoch": 24.111691704768127, "grad_norm": 0.0746530145406723, "learning_rate": 9.148586456278765e-06, "loss": 0.0106, "step": 37450 }, { "epoch": 24.14435009797518, "grad_norm": 0.07946937531232834, "learning_rate": 9.145299145299145e-06, "loss": 0.0111, "step": 37500 }, { "epoch": 24.177008491182235, "grad_norm": 0.06598696112632751, "learning_rate": 9.142011834319528e-06, "loss": 0.0107, "step": 37550 }, { "epoch": 24.20966688438929, "grad_norm": 0.0551881343126297, "learning_rate": 9.13872452333991e-06, "loss": 0.0108, "step": 37600 }, { "epoch": 24.242325277596343, "grad_norm": 0.06613214313983917, "learning_rate": 9.13543721236029e-06, "loss": 0.0105, "step": 37650 }, { "epoch": 24.274983670803397, "grad_norm": 0.055688247084617615, "learning_rate": 9.132149901380671e-06, "loss": 0.0103, "step": 37700 }, { "epoch": 24.30764206401045, "grad_norm": 0.08066407591104507, "learning_rate": 9.128862590401052e-06, "loss": 0.0102, "step": 37750 }, { "epoch": 24.340300457217506, "grad_norm": 0.052430395036935806, "learning_rate": 9.125575279421434e-06, "loss": 0.0104, "step": 37800 }, { "epoch": 24.37295885042456, "grad_norm": 0.058990783989429474, "learning_rate": 9.122287968441815e-06, "loss": 0.011, "step": 37850 }, { "epoch": 24.405617243631614, "grad_norm": 0.08820757269859314, "learning_rate": 9.119000657462197e-06, "loss": 0.0097, "step": 37900 }, { "epoch": 24.438275636838668, "grad_norm": 0.07525777071714401, "learning_rate": 9.115713346482578e-06, "loss": 0.0104, "step": 37950 }, { "epoch": 24.470934030045722, "grad_norm": 0.0593407116830349, "learning_rate": 9.11242603550296e-06, "loss": 0.0102, "step": 38000 }, { "epoch": 24.503592423252776, "grad_norm": 0.08133421093225479, "learning_rate": 9.109138724523341e-06, "loss": 0.0109, "step": 38050 }, { "epoch": 24.53625081645983, "grad_norm": 0.06059282645583153, "learning_rate": 9.10585141354372e-06, "loss": 0.0104, "step": 38100 }, { "epoch": 24.568909209666884, "grad_norm": 0.06982138752937317, "learning_rate": 9.102564102564104e-06, "loss": 0.0106, "step": 38150 }, { "epoch": 24.60156760287394, "grad_norm": 0.061989493668079376, "learning_rate": 9.099276791584485e-06, "loss": 0.0108, "step": 38200 }, { "epoch": 24.634225996080993, "grad_norm": 0.07107903063297272, "learning_rate": 9.095989480604865e-06, "loss": 0.0103, "step": 38250 }, { "epoch": 24.666884389288047, "grad_norm": 0.092849500477314, "learning_rate": 9.092702169625247e-06, "loss": 0.0102, "step": 38300 }, { "epoch": 24.6995427824951, "grad_norm": 0.07604721933603287, "learning_rate": 9.089414858645628e-06, "loss": 0.0105, "step": 38350 }, { "epoch": 24.732201175702155, "grad_norm": 0.07060103863477707, "learning_rate": 9.086127547666011e-06, "loss": 0.01, "step": 38400 }, { "epoch": 24.76485956890921, "grad_norm": 0.062143828719854355, "learning_rate": 9.082840236686391e-06, "loss": 0.0106, "step": 38450 }, { "epoch": 24.797517962116263, "grad_norm": 0.06545909494161606, "learning_rate": 9.079552925706772e-06, "loss": 0.0105, "step": 38500 }, { "epoch": 24.830176355323317, "grad_norm": 0.07944540679454803, "learning_rate": 9.076265614727154e-06, "loss": 0.0103, "step": 38550 }, { "epoch": 24.86283474853037, "grad_norm": 0.06381931900978088, "learning_rate": 9.072978303747535e-06, "loss": 0.01, "step": 38600 }, { "epoch": 24.895493141737425, "grad_norm": 0.07727362215518951, "learning_rate": 9.069690992767917e-06, "loss": 0.0106, "step": 38650 }, { "epoch": 24.92815153494448, "grad_norm": 0.07862541079521179, "learning_rate": 9.066403681788298e-06, "loss": 0.0104, "step": 38700 }, { "epoch": 24.960809928151534, "grad_norm": 0.0978250578045845, "learning_rate": 9.06311637080868e-06, "loss": 0.0107, "step": 38750 }, { "epoch": 24.993468321358588, "grad_norm": 0.08668365329504013, "learning_rate": 9.059829059829061e-06, "loss": 0.0101, "step": 38800 }, { "epoch": 25.026126714565642, "grad_norm": 0.06107616424560547, "learning_rate": 9.056541748849443e-06, "loss": 0.0105, "step": 38850 }, { "epoch": 25.058785107772696, "grad_norm": 0.1027405709028244, "learning_rate": 9.053254437869822e-06, "loss": 0.0105, "step": 38900 }, { "epoch": 25.09144350097975, "grad_norm": 0.06902670115232468, "learning_rate": 9.049967126890204e-06, "loss": 0.0104, "step": 38950 }, { "epoch": 25.124101894186808, "grad_norm": 0.08010412007570267, "learning_rate": 9.046679815910587e-06, "loss": 0.0102, "step": 39000 }, { "epoch": 25.156760287393862, "grad_norm": 0.06517750769853592, "learning_rate": 9.043392504930967e-06, "loss": 0.0101, "step": 39050 }, { "epoch": 25.189418680600916, "grad_norm": 0.057309724390506744, "learning_rate": 9.040105193951348e-06, "loss": 0.0102, "step": 39100 }, { "epoch": 25.22207707380797, "grad_norm": 0.06081010773777962, "learning_rate": 9.03681788297173e-06, "loss": 0.0104, "step": 39150 }, { "epoch": 25.254735467015024, "grad_norm": 0.06606275588274002, "learning_rate": 9.033530571992111e-06, "loss": 0.0104, "step": 39200 }, { "epoch": 25.28739386022208, "grad_norm": 0.06801149994134903, "learning_rate": 9.030243261012493e-06, "loss": 0.0104, "step": 39250 }, { "epoch": 25.320052253429132, "grad_norm": 0.08606895059347153, "learning_rate": 9.026955950032874e-06, "loss": 0.0105, "step": 39300 }, { "epoch": 25.352710646636186, "grad_norm": 0.06489425897598267, "learning_rate": 9.023668639053255e-06, "loss": 0.0104, "step": 39350 }, { "epoch": 25.38536903984324, "grad_norm": 0.0781024694442749, "learning_rate": 9.020381328073637e-06, "loss": 0.01, "step": 39400 }, { "epoch": 25.418027433050295, "grad_norm": 0.07859282940626144, "learning_rate": 9.017094017094018e-06, "loss": 0.0101, "step": 39450 }, { "epoch": 25.45068582625735, "grad_norm": 0.07220655679702759, "learning_rate": 9.0138067061144e-06, "loss": 0.0102, "step": 39500 }, { "epoch": 25.483344219464403, "grad_norm": 0.08303755521774292, "learning_rate": 9.01051939513478e-06, "loss": 0.0099, "step": 39550 }, { "epoch": 25.516002612671457, "grad_norm": 0.1362331360578537, "learning_rate": 9.007232084155163e-06, "loss": 0.0101, "step": 39600 }, { "epoch": 25.54866100587851, "grad_norm": 0.06204262748360634, "learning_rate": 9.003944773175544e-06, "loss": 0.0099, "step": 39650 }, { "epoch": 25.581319399085565, "grad_norm": 0.05903197452425957, "learning_rate": 9.000657462195924e-06, "loss": 0.0101, "step": 39700 }, { "epoch": 25.61397779229262, "grad_norm": 0.085459403693676, "learning_rate": 8.997370151216305e-06, "loss": 0.01, "step": 39750 }, { "epoch": 25.646636185499673, "grad_norm": 0.06661771982908249, "learning_rate": 8.994082840236687e-06, "loss": 0.0104, "step": 39800 }, { "epoch": 25.679294578706727, "grad_norm": 0.09459152817726135, "learning_rate": 8.990795529257068e-06, "loss": 0.0106, "step": 39850 }, { "epoch": 25.71195297191378, "grad_norm": 0.05796539783477783, "learning_rate": 8.98750821827745e-06, "loss": 0.0101, "step": 39900 }, { "epoch": 25.744611365120836, "grad_norm": 0.05809909850358963, "learning_rate": 8.984220907297831e-06, "loss": 0.0103, "step": 39950 }, { "epoch": 25.77726975832789, "grad_norm": 0.07010135799646378, "learning_rate": 8.980933596318213e-06, "loss": 0.0105, "step": 40000 }, { "epoch": 25.809928151534944, "grad_norm": 0.07605331391096115, "learning_rate": 8.977646285338594e-06, "loss": 0.0101, "step": 40050 }, { "epoch": 25.842586544741998, "grad_norm": 0.06494385004043579, "learning_rate": 8.974358974358976e-06, "loss": 0.0098, "step": 40100 }, { "epoch": 25.875244937949052, "grad_norm": 0.07195784151554108, "learning_rate": 8.971071663379355e-06, "loss": 0.0107, "step": 40150 }, { "epoch": 25.907903331156106, "grad_norm": 0.07367152720689774, "learning_rate": 8.967784352399738e-06, "loss": 0.0102, "step": 40200 }, { "epoch": 25.94056172436316, "grad_norm": 0.060549940913915634, "learning_rate": 8.96449704142012e-06, "loss": 0.0101, "step": 40250 }, { "epoch": 25.973220117570214, "grad_norm": 0.074057936668396, "learning_rate": 8.9612097304405e-06, "loss": 0.0102, "step": 40300 }, { "epoch": 26.00587851077727, "grad_norm": 0.0657525509595871, "learning_rate": 8.957922419460881e-06, "loss": 0.01, "step": 40350 }, { "epoch": 26.038536903984323, "grad_norm": 0.05703706294298172, "learning_rate": 8.954635108481263e-06, "loss": 0.0099, "step": 40400 }, { "epoch": 26.071195297191377, "grad_norm": 0.05926249548792839, "learning_rate": 8.951347797501644e-06, "loss": 0.0103, "step": 40450 }, { "epoch": 26.10385369039843, "grad_norm": 0.06457608938217163, "learning_rate": 8.948060486522026e-06, "loss": 0.0102, "step": 40500 }, { "epoch": 26.136512083605485, "grad_norm": 0.09410393983125687, "learning_rate": 8.944773175542407e-06, "loss": 0.01, "step": 40550 }, { "epoch": 26.16917047681254, "grad_norm": 0.06839783489704132, "learning_rate": 8.941485864562788e-06, "loss": 0.0099, "step": 40600 }, { "epoch": 26.201828870019597, "grad_norm": 0.06976661086082458, "learning_rate": 8.93819855358317e-06, "loss": 0.0095, "step": 40650 }, { "epoch": 26.23448726322665, "grad_norm": 0.07970596104860306, "learning_rate": 8.934911242603551e-06, "loss": 0.0098, "step": 40700 }, { "epoch": 26.267145656433705, "grad_norm": 0.07029850780963898, "learning_rate": 8.931623931623933e-06, "loss": 0.0095, "step": 40750 }, { "epoch": 26.29980404964076, "grad_norm": 0.05546072497963905, "learning_rate": 8.928336620644314e-06, "loss": 0.0099, "step": 40800 }, { "epoch": 26.332462442847813, "grad_norm": 0.07101114839315414, "learning_rate": 8.925049309664696e-06, "loss": 0.01, "step": 40850 }, { "epoch": 26.365120836054867, "grad_norm": 0.0728398784995079, "learning_rate": 8.921761998685077e-06, "loss": 0.0104, "step": 40900 }, { "epoch": 26.39777922926192, "grad_norm": 0.06518243253231049, "learning_rate": 8.918474687705457e-06, "loss": 0.0102, "step": 40950 }, { "epoch": 26.430437622468975, "grad_norm": 0.07374660670757294, "learning_rate": 8.915187376725838e-06, "loss": 0.01, "step": 41000 }, { "epoch": 26.46309601567603, "grad_norm": 0.06757120788097382, "learning_rate": 8.911900065746222e-06, "loss": 0.01, "step": 41050 }, { "epoch": 26.495754408883084, "grad_norm": 0.07450166344642639, "learning_rate": 8.908612754766601e-06, "loss": 0.0099, "step": 41100 }, { "epoch": 26.528412802090138, "grad_norm": 0.06532027572393417, "learning_rate": 8.905325443786983e-06, "loss": 0.0097, "step": 41150 }, { "epoch": 26.561071195297192, "grad_norm": 0.05976063758134842, "learning_rate": 8.902038132807364e-06, "loss": 0.0101, "step": 41200 }, { "epoch": 26.593729588504246, "grad_norm": 0.07502580434083939, "learning_rate": 8.898750821827746e-06, "loss": 0.0101, "step": 41250 }, { "epoch": 26.6263879817113, "grad_norm": 0.06817267835140228, "learning_rate": 8.895463510848127e-06, "loss": 0.0103, "step": 41300 }, { "epoch": 26.659046374918354, "grad_norm": 0.060340505093336105, "learning_rate": 8.892176199868509e-06, "loss": 0.0102, "step": 41350 }, { "epoch": 26.69170476812541, "grad_norm": 0.08452873677015305, "learning_rate": 8.888888888888888e-06, "loss": 0.0101, "step": 41400 }, { "epoch": 26.724363161332462, "grad_norm": 0.06396488845348358, "learning_rate": 8.885601577909271e-06, "loss": 0.0104, "step": 41450 }, { "epoch": 26.757021554539516, "grad_norm": 0.0851375088095665, "learning_rate": 8.882314266929653e-06, "loss": 0.0102, "step": 41500 }, { "epoch": 26.78967994774657, "grad_norm": 0.07479476183652878, "learning_rate": 8.879026955950033e-06, "loss": 0.0101, "step": 41550 }, { "epoch": 26.822338340953625, "grad_norm": 0.061094850301742554, "learning_rate": 8.875739644970414e-06, "loss": 0.01, "step": 41600 }, { "epoch": 26.85499673416068, "grad_norm": 0.0774194747209549, "learning_rate": 8.872452333990797e-06, "loss": 0.0097, "step": 41650 }, { "epoch": 26.887655127367733, "grad_norm": 0.0612763874232769, "learning_rate": 8.869165023011179e-06, "loss": 0.0102, "step": 41700 }, { "epoch": 26.920313520574787, "grad_norm": 0.06663808971643448, "learning_rate": 8.865877712031558e-06, "loss": 0.0102, "step": 41750 }, { "epoch": 26.95297191378184, "grad_norm": 0.06964550912380219, "learning_rate": 8.86259040105194e-06, "loss": 0.0098, "step": 41800 }, { "epoch": 26.985630306988895, "grad_norm": 0.05729374289512634, "learning_rate": 8.859303090072321e-06, "loss": 0.0101, "step": 41850 }, { "epoch": 27.01828870019595, "grad_norm": 0.05539995804429054, "learning_rate": 8.856015779092703e-06, "loss": 0.0098, "step": 41900 }, { "epoch": 27.050947093403003, "grad_norm": 0.06388706713914871, "learning_rate": 8.852728468113084e-06, "loss": 0.0097, "step": 41950 }, { "epoch": 27.083605486610058, "grad_norm": 0.06247853860259056, "learning_rate": 8.849441157133466e-06, "loss": 0.0096, "step": 42000 }, { "epoch": 27.11626387981711, "grad_norm": 0.10026439279317856, "learning_rate": 8.846153846153847e-06, "loss": 0.0095, "step": 42050 }, { "epoch": 27.148922273024166, "grad_norm": 0.06085231155157089, "learning_rate": 8.842866535174229e-06, "loss": 0.01, "step": 42100 }, { "epoch": 27.18158066623122, "grad_norm": 0.06942855566740036, "learning_rate": 8.83957922419461e-06, "loss": 0.0098, "step": 42150 }, { "epoch": 27.214239059438274, "grad_norm": 0.07817970961332321, "learning_rate": 8.83629191321499e-06, "loss": 0.0101, "step": 42200 }, { "epoch": 27.24689745264533, "grad_norm": 0.062483951449394226, "learning_rate": 8.833004602235371e-06, "loss": 0.0098, "step": 42250 }, { "epoch": 27.279555845852386, "grad_norm": 0.05964931845664978, "learning_rate": 8.829717291255754e-06, "loss": 0.01, "step": 42300 }, { "epoch": 27.31221423905944, "grad_norm": 0.0791887491941452, "learning_rate": 8.826429980276134e-06, "loss": 0.0099, "step": 42350 }, { "epoch": 27.344872632266494, "grad_norm": 0.09129365533590317, "learning_rate": 8.823142669296516e-06, "loss": 0.0097, "step": 42400 }, { "epoch": 27.377531025473548, "grad_norm": 0.07373974472284317, "learning_rate": 8.819855358316897e-06, "loss": 0.0098, "step": 42450 }, { "epoch": 27.410189418680602, "grad_norm": 0.07452785223722458, "learning_rate": 8.816568047337279e-06, "loss": 0.0099, "step": 42500 }, { "epoch": 27.442847811887656, "grad_norm": 0.06308138370513916, "learning_rate": 8.81328073635766e-06, "loss": 0.0101, "step": 42550 }, { "epoch": 27.47550620509471, "grad_norm": 0.08655289560556412, "learning_rate": 8.809993425378042e-06, "loss": 0.0097, "step": 42600 }, { "epoch": 27.508164598301764, "grad_norm": 0.08159057050943375, "learning_rate": 8.806706114398423e-06, "loss": 0.01, "step": 42650 }, { "epoch": 27.54082299150882, "grad_norm": 0.06101159006357193, "learning_rate": 8.803418803418804e-06, "loss": 0.01, "step": 42700 }, { "epoch": 27.573481384715873, "grad_norm": 0.07626356184482574, "learning_rate": 8.800131492439186e-06, "loss": 0.0097, "step": 42750 }, { "epoch": 27.606139777922927, "grad_norm": 0.07007754594087601, "learning_rate": 8.796844181459567e-06, "loss": 0.01, "step": 42800 }, { "epoch": 27.63879817112998, "grad_norm": 0.06938617676496506, "learning_rate": 8.793556870479947e-06, "loss": 0.0097, "step": 42850 }, { "epoch": 27.671456564337035, "grad_norm": 0.0703434869647026, "learning_rate": 8.79026955950033e-06, "loss": 0.0101, "step": 42900 }, { "epoch": 27.70411495754409, "grad_norm": 0.07527665048837662, "learning_rate": 8.786982248520712e-06, "loss": 0.0096, "step": 42950 }, { "epoch": 27.736773350751143, "grad_norm": 0.06230452284216881, "learning_rate": 8.783694937541091e-06, "loss": 0.0101, "step": 43000 }, { "epoch": 27.769431743958197, "grad_norm": 0.06309953331947327, "learning_rate": 8.780407626561473e-06, "loss": 0.01, "step": 43050 }, { "epoch": 27.80209013716525, "grad_norm": 0.08436129242181778, "learning_rate": 8.777120315581854e-06, "loss": 0.0096, "step": 43100 }, { "epoch": 27.834748530372305, "grad_norm": 0.07899975031614304, "learning_rate": 8.773833004602236e-06, "loss": 0.0097, "step": 43150 }, { "epoch": 27.86740692357936, "grad_norm": 0.07137970626354218, "learning_rate": 8.770545693622617e-06, "loss": 0.0096, "step": 43200 }, { "epoch": 27.900065316786414, "grad_norm": 0.07216981053352356, "learning_rate": 8.767258382642999e-06, "loss": 0.0095, "step": 43250 }, { "epoch": 27.932723709993468, "grad_norm": 0.09067603945732117, "learning_rate": 8.76397107166338e-06, "loss": 0.0093, "step": 43300 }, { "epoch": 27.965382103200522, "grad_norm": 0.06958699971437454, "learning_rate": 8.760683760683762e-06, "loss": 0.0096, "step": 43350 }, { "epoch": 27.998040496407576, "grad_norm": 0.060856375843286514, "learning_rate": 8.757396449704143e-06, "loss": 0.0097, "step": 43400 }, { "epoch": 28.03069888961463, "grad_norm": 0.0756077691912651, "learning_rate": 8.754109138724523e-06, "loss": 0.0092, "step": 43450 }, { "epoch": 28.063357282821684, "grad_norm": 0.05710800737142563, "learning_rate": 8.750821827744906e-06, "loss": 0.0092, "step": 43500 }, { "epoch": 28.09601567602874, "grad_norm": 0.08190213888883591, "learning_rate": 8.747534516765287e-06, "loss": 0.0094, "step": 43550 }, { "epoch": 28.128674069235792, "grad_norm": 0.07220468670129776, "learning_rate": 8.744247205785667e-06, "loss": 0.0095, "step": 43600 }, { "epoch": 28.161332462442846, "grad_norm": 0.08410811424255371, "learning_rate": 8.740959894806049e-06, "loss": 0.0096, "step": 43650 }, { "epoch": 28.1939908556499, "grad_norm": 0.057440534234046936, "learning_rate": 8.73767258382643e-06, "loss": 0.0092, "step": 43700 }, { "epoch": 28.226649248856955, "grad_norm": 0.06163407862186432, "learning_rate": 8.734385272846812e-06, "loss": 0.0099, "step": 43750 }, { "epoch": 28.25930764206401, "grad_norm": 0.08130808174610138, "learning_rate": 8.731097961867193e-06, "loss": 0.0094, "step": 43800 }, { "epoch": 28.291966035271066, "grad_norm": 0.08615657687187195, "learning_rate": 8.727810650887574e-06, "loss": 0.0099, "step": 43850 }, { "epoch": 28.32462442847812, "grad_norm": 0.07963917404413223, "learning_rate": 8.724523339907956e-06, "loss": 0.0097, "step": 43900 }, { "epoch": 28.357282821685175, "grad_norm": 0.06758115440607071, "learning_rate": 8.721236028928337e-06, "loss": 0.01, "step": 43950 }, { "epoch": 28.38994121489223, "grad_norm": 0.07085925340652466, "learning_rate": 8.717948717948719e-06, "loss": 0.0092, "step": 44000 }, { "epoch": 28.422599608099283, "grad_norm": 0.0618162676692009, "learning_rate": 8.7146614069691e-06, "loss": 0.0097, "step": 44050 }, { "epoch": 28.455258001306337, "grad_norm": 0.06311200559139252, "learning_rate": 8.711374095989482e-06, "loss": 0.0091, "step": 44100 }, { "epoch": 28.48791639451339, "grad_norm": 0.0744459480047226, "learning_rate": 8.708086785009863e-06, "loss": 0.0098, "step": 44150 }, { "epoch": 28.520574787720445, "grad_norm": 0.07541000843048096, "learning_rate": 8.704799474030245e-06, "loss": 0.0097, "step": 44200 }, { "epoch": 28.5532331809275, "grad_norm": 0.08303575962781906, "learning_rate": 8.701512163050624e-06, "loss": 0.0094, "step": 44250 }, { "epoch": 28.585891574134553, "grad_norm": 0.06145607680082321, "learning_rate": 8.698224852071006e-06, "loss": 0.0095, "step": 44300 }, { "epoch": 28.618549967341607, "grad_norm": 0.054414454847574234, "learning_rate": 8.694937541091389e-06, "loss": 0.0102, "step": 44350 }, { "epoch": 28.65120836054866, "grad_norm": 0.07142601162195206, "learning_rate": 8.691650230111769e-06, "loss": 0.0096, "step": 44400 }, { "epoch": 28.683866753755716, "grad_norm": 0.07634475827217102, "learning_rate": 8.68836291913215e-06, "loss": 0.0098, "step": 44450 }, { "epoch": 28.71652514696277, "grad_norm": 0.06411303579807281, "learning_rate": 8.685075608152532e-06, "loss": 0.0099, "step": 44500 }, { "epoch": 29.032658393207054, "grad_norm": 0.0766330137848854, "learning_rate": 8.681788297172913e-06, "loss": 0.0093, "step": 44550 }, { "epoch": 29.065316786414108, "grad_norm": 0.07081354409456253, "learning_rate": 8.678500986193295e-06, "loss": 0.0099, "step": 44600 }, { "epoch": 29.097975179621162, "grad_norm": 0.05489321053028107, "learning_rate": 8.675213675213676e-06, "loss": 0.0096, "step": 44650 }, { "epoch": 29.130633572828216, "grad_norm": 0.07282430678606033, "learning_rate": 8.671926364234058e-06, "loss": 0.0092, "step": 44700 }, { "epoch": 29.16329196603527, "grad_norm": 0.08215247094631195, "learning_rate": 8.668639053254439e-06, "loss": 0.0094, "step": 44750 }, { "epoch": 29.195950359242325, "grad_norm": 0.06071177497506142, "learning_rate": 8.66535174227482e-06, "loss": 0.0093, "step": 44800 }, { "epoch": 29.22860875244938, "grad_norm": 0.08762010186910629, "learning_rate": 8.6620644312952e-06, "loss": 0.0095, "step": 44850 }, { "epoch": 29.261267145656433, "grad_norm": 0.06701937317848206, "learning_rate": 8.658777120315582e-06, "loss": 0.0093, "step": 44900 }, { "epoch": 29.293925538863487, "grad_norm": 0.08069388568401337, "learning_rate": 8.655489809335965e-06, "loss": 0.0093, "step": 44950 }, { "epoch": 29.32658393207054, "grad_norm": 0.07930802553892136, "learning_rate": 8.652202498356346e-06, "loss": 0.0099, "step": 45000 }, { "epoch": 29.032658393207054, "grad_norm": 0.07764122635126114, "learning_rate": 8.648915187376726e-06, "loss": 0.0095, "step": 45050 }, { "epoch": 29.065316786414108, "grad_norm": 0.07332498580217361, "learning_rate": 8.645627876397107e-06, "loss": 0.0099, "step": 45100 }, { "epoch": 29.097975179621162, "grad_norm": 0.06437701731920242, "learning_rate": 8.642340565417489e-06, "loss": 0.0095, "step": 45150 }, { "epoch": 29.130633572828216, "grad_norm": 0.07691024988889694, "learning_rate": 8.63905325443787e-06, "loss": 0.0093, "step": 45200 }, { "epoch": 29.16329196603527, "grad_norm": 0.10297652333974838, "learning_rate": 8.635765943458252e-06, "loss": 0.0094, "step": 45250 }, { "epoch": 29.195950359242325, "grad_norm": 0.07019223272800446, "learning_rate": 8.632478632478633e-06, "loss": 0.0092, "step": 45300 }, { "epoch": 29.22860875244938, "grad_norm": 0.06949339807033539, "learning_rate": 8.629191321499015e-06, "loss": 0.0095, "step": 45350 }, { "epoch": 29.261267145656433, "grad_norm": 0.07147526741027832, "learning_rate": 8.625904010519396e-06, "loss": 0.0092, "step": 45400 }, { "epoch": 29.293925538863487, "grad_norm": 0.061854057013988495, "learning_rate": 8.622616699539778e-06, "loss": 0.009, "step": 45450 }, { "epoch": 29.32658393207054, "grad_norm": 0.07520570605993271, "learning_rate": 8.619329388560157e-06, "loss": 0.0098, "step": 45500 }, { "epoch": 29.359242325277595, "grad_norm": 0.07743249088525772, "learning_rate": 8.61604207758054e-06, "loss": 0.0096, "step": 45550 }, { "epoch": 29.39190071848465, "grad_norm": 0.08658551424741745, "learning_rate": 8.612754766600922e-06, "loss": 0.0095, "step": 45600 }, { "epoch": 29.424559111691703, "grad_norm": 0.07984712719917297, "learning_rate": 8.609467455621302e-06, "loss": 0.0097, "step": 45650 }, { "epoch": 29.457217504898757, "grad_norm": 0.06923504918813705, "learning_rate": 8.606180144641683e-06, "loss": 0.0093, "step": 45700 }, { "epoch": 29.48987589810581, "grad_norm": 0.09051218628883362, "learning_rate": 8.602892833662065e-06, "loss": 0.0096, "step": 45750 }, { "epoch": 29.522534291312866, "grad_norm": 0.07055158913135529, "learning_rate": 8.599605522682446e-06, "loss": 0.0095, "step": 45800 }, { "epoch": 29.555192684519923, "grad_norm": 0.06876330822706223, "learning_rate": 8.596318211702828e-06, "loss": 0.0091, "step": 45850 }, { "epoch": 29.587851077726977, "grad_norm": 0.07638917118310928, "learning_rate": 8.593030900723209e-06, "loss": 0.0096, "step": 45900 }, { "epoch": 29.62050947093403, "grad_norm": 0.08202219754457474, "learning_rate": 8.58974358974359e-06, "loss": 0.0094, "step": 45950 }, { "epoch": 29.653167864141086, "grad_norm": 0.06962180137634277, "learning_rate": 8.586456278763972e-06, "loss": 0.0093, "step": 46000 }, { "epoch": 29.68582625734814, "grad_norm": 0.061191633343696594, "learning_rate": 8.583168967784353e-06, "loss": 0.0098, "step": 46050 }, { "epoch": 29.718484650555194, "grad_norm": 0.0624864287674427, "learning_rate": 8.579881656804735e-06, "loss": 0.0096, "step": 46100 }, { "epoch": 29.751143043762248, "grad_norm": 0.08512195944786072, "learning_rate": 8.576594345825116e-06, "loss": 0.0096, "step": 46150 }, { "epoch": 29.783801436969302, "grad_norm": 0.09398037940263748, "learning_rate": 8.573307034845498e-06, "loss": 0.0095, "step": 46200 }, { "epoch": 29.816459830176356, "grad_norm": 0.07401812076568604, "learning_rate": 8.57001972386588e-06, "loss": 0.0092, "step": 46250 }, { "epoch": 29.84911822338341, "grad_norm": 0.06938875466585159, "learning_rate": 8.566732412886259e-06, "loss": 0.0094, "step": 46300 }, { "epoch": 29.881776616590464, "grad_norm": 0.07812067866325378, "learning_rate": 8.56344510190664e-06, "loss": 0.0091, "step": 46350 }, { "epoch": 29.91443500979752, "grad_norm": 0.06064947694540024, "learning_rate": 8.560157790927024e-06, "loss": 0.0091, "step": 46400 }, { "epoch": 29.947093403004573, "grad_norm": 0.08336592465639114, "learning_rate": 8.556870479947403e-06, "loss": 0.0091, "step": 46450 }, { "epoch": 29.979751796211627, "grad_norm": 0.04857954755425453, "learning_rate": 8.553583168967785e-06, "loss": 0.0092, "step": 46500 }, { "epoch": 30.01241018941868, "grad_norm": 0.0645669624209404, "learning_rate": 8.550295857988166e-06, "loss": 0.0093, "step": 46550 }, { "epoch": 30.045068582625735, "grad_norm": 0.08509334921836853, "learning_rate": 8.547008547008548e-06, "loss": 0.0092, "step": 46600 }, { "epoch": 30.07772697583279, "grad_norm": 0.06356937438249588, "learning_rate": 8.543721236028929e-06, "loss": 0.0089, "step": 46650 }, { "epoch": 30.110385369039843, "grad_norm": 0.07010962069034576, "learning_rate": 8.54043392504931e-06, "loss": 0.0092, "step": 46700 }, { "epoch": 30.143043762246897, "grad_norm": 0.14091841876506805, "learning_rate": 8.537146614069692e-06, "loss": 0.0093, "step": 46750 }, { "epoch": 30.17570215545395, "grad_norm": 0.07473936676979065, "learning_rate": 8.533859303090073e-06, "loss": 0.0089, "step": 46800 }, { "epoch": 30.208360548661005, "grad_norm": 0.08768052607774734, "learning_rate": 8.530571992110455e-06, "loss": 0.009, "step": 46850 }, { "epoch": 30.24101894186806, "grad_norm": 0.08124729990959167, "learning_rate": 8.527284681130835e-06, "loss": 0.0089, "step": 46900 }, { "epoch": 30.273677335075114, "grad_norm": 0.07920121401548386, "learning_rate": 8.523997370151216e-06, "loss": 0.0088, "step": 46950 }, { "epoch": 30.306335728282168, "grad_norm": 0.06729952991008759, "learning_rate": 8.5207100591716e-06, "loss": 0.0095, "step": 47000 }, { "epoch": 30.338994121489222, "grad_norm": 0.10546492040157318, "learning_rate": 8.517422748191979e-06, "loss": 0.0091, "step": 47050 }, { "epoch": 30.371652514696276, "grad_norm": 0.1032395213842392, "learning_rate": 8.51413543721236e-06, "loss": 0.0089, "step": 47100 }, { "epoch": 30.40431090790333, "grad_norm": 0.0650549978017807, "learning_rate": 8.510848126232742e-06, "loss": 0.0091, "step": 47150 }, { "epoch": 30.436969301110384, "grad_norm": 0.05851263180375099, "learning_rate": 8.507560815253123e-06, "loss": 0.0094, "step": 47200 }, { "epoch": 30.46962769431744, "grad_norm": 0.07420110702514648, "learning_rate": 8.504273504273505e-06, "loss": 0.0091, "step": 47250 }, { "epoch": 30.502286087524492, "grad_norm": 0.07141097635030746, "learning_rate": 8.500986193293886e-06, "loss": 0.0091, "step": 47300 }, { "epoch": 30.534944480731546, "grad_norm": 0.059118326753377914, "learning_rate": 8.497698882314268e-06, "loss": 0.0092, "step": 47350 }, { "epoch": 30.5676028739386, "grad_norm": 0.06489855796098709, "learning_rate": 8.49441157133465e-06, "loss": 0.0092, "step": 47400 }, { "epoch": 30.600261267145655, "grad_norm": 0.07056184858083725, "learning_rate": 8.49112426035503e-06, "loss": 0.0088, "step": 47450 }, { "epoch": 30.632919660352712, "grad_norm": 0.08145038038492203, "learning_rate": 8.487836949375412e-06, "loss": 0.009, "step": 47500 }, { "epoch": 30.665578053559766, "grad_norm": 0.07086652517318726, "learning_rate": 8.484549638395792e-06, "loss": 0.0088, "step": 47550 }, { "epoch": 30.69823644676682, "grad_norm": 0.05630020424723625, "learning_rate": 8.481262327416175e-06, "loss": 0.0086, "step": 47600 }, { "epoch": 30.730894839973875, "grad_norm": 0.05787571519613266, "learning_rate": 8.477975016436557e-06, "loss": 0.0091, "step": 47650 }, { "epoch": 30.76355323318093, "grad_norm": 0.06763002276420593, "learning_rate": 8.474687705456936e-06, "loss": 0.0093, "step": 47700 }, { "epoch": 30.796211626387983, "grad_norm": 0.05904902517795563, "learning_rate": 8.471400394477318e-06, "loss": 0.0093, "step": 47750 }, { "epoch": 30.828870019595037, "grad_norm": 0.05922796577215195, "learning_rate": 8.4681130834977e-06, "loss": 0.0088, "step": 47800 }, { "epoch": 30.86152841280209, "grad_norm": 0.0699804499745369, "learning_rate": 8.46482577251808e-06, "loss": 0.0087, "step": 47850 }, { "epoch": 30.894186806009145, "grad_norm": 0.05228590965270996, "learning_rate": 8.461538461538462e-06, "loss": 0.0088, "step": 47900 }, { "epoch": 30.9268451992162, "grad_norm": 0.07833770662546158, "learning_rate": 8.458251150558844e-06, "loss": 0.0089, "step": 47950 }, { "epoch": 30.959503592423253, "grad_norm": 0.06715569645166397, "learning_rate": 8.454963839579225e-06, "loss": 0.009, "step": 48000 }, { "epoch": 30.992161985630307, "grad_norm": 0.08389867097139359, "learning_rate": 8.451676528599606e-06, "loss": 0.009, "step": 48050 }, { "epoch": 31.02482037883736, "grad_norm": 0.06357283145189285, "learning_rate": 8.448389217619988e-06, "loss": 0.009, "step": 48100 }, { "epoch": 31.057478772044416, "grad_norm": 0.07123304158449173, "learning_rate": 8.445101906640368e-06, "loss": 0.0084, "step": 48150 }, { "epoch": 31.09013716525147, "grad_norm": 0.058402303606271744, "learning_rate": 8.44181459566075e-06, "loss": 0.0087, "step": 48200 }, { "epoch": 31.122795558458524, "grad_norm": 0.09926163405179977, "learning_rate": 8.438527284681132e-06, "loss": 0.0088, "step": 48250 }, { "epoch": 31.155453951665578, "grad_norm": 0.06271667778491974, "learning_rate": 8.435239973701514e-06, "loss": 0.0089, "step": 48300 }, { "epoch": 31.188112344872632, "grad_norm": 0.06636612862348557, "learning_rate": 8.431952662721893e-06, "loss": 0.0088, "step": 48350 }, { "epoch": 31.220770738079686, "grad_norm": 0.056099530309438705, "learning_rate": 8.428665351742275e-06, "loss": 0.0086, "step": 48400 }, { "epoch": 31.25342913128674, "grad_norm": 0.08813437819480896, "learning_rate": 8.425378040762658e-06, "loss": 0.0089, "step": 48450 }, { "epoch": 31.286087524493794, "grad_norm": 0.053726114332675934, "learning_rate": 8.422090729783038e-06, "loss": 0.0088, "step": 48500 }, { "epoch": 31.31874591770085, "grad_norm": 0.06936299055814743, "learning_rate": 8.41880341880342e-06, "loss": 0.0083, "step": 48550 }, { "epoch": 31.351404310907903, "grad_norm": 0.09912844747304916, "learning_rate": 8.4155161078238e-06, "loss": 0.0084, "step": 48600 }, { "epoch": 31.384062704114957, "grad_norm": 0.11457721143960953, "learning_rate": 8.412228796844182e-06, "loss": 0.009, "step": 48650 }, { "epoch": 31.41672109732201, "grad_norm": 0.06133641302585602, "learning_rate": 8.408941485864564e-06, "loss": 0.0091, "step": 48700 }, { "epoch": 31.449379490529065, "grad_norm": 0.069750115275383, "learning_rate": 8.405654174884945e-06, "loss": 0.0089, "step": 48750 }, { "epoch": 31.48203788373612, "grad_norm": 0.06444136053323746, "learning_rate": 8.402366863905327e-06, "loss": 0.0085, "step": 48800 }, { "epoch": 31.514696276943173, "grad_norm": 0.0707576796412468, "learning_rate": 8.399079552925708e-06, "loss": 0.0086, "step": 48850 }, { "epoch": 31.547354670150227, "grad_norm": 0.09656653553247452, "learning_rate": 8.39579224194609e-06, "loss": 0.0085, "step": 48900 }, { "epoch": 31.58001306335728, "grad_norm": 0.06256120651960373, "learning_rate": 8.39250493096647e-06, "loss": 0.009, "step": 48950 }, { "epoch": 31.612671456564335, "grad_norm": 0.06921864300966263, "learning_rate": 8.38921761998685e-06, "loss": 0.0089, "step": 49000 }, { "epoch": 31.645329849771393, "grad_norm": 0.05597410723567009, "learning_rate": 8.385930309007234e-06, "loss": 0.0085, "step": 49050 }, { "epoch": 31.677988242978447, "grad_norm": 0.10604964196681976, "learning_rate": 8.382642998027614e-06, "loss": 0.0085, "step": 49100 }, { "epoch": 31.7106466361855, "grad_norm": 0.06222546473145485, "learning_rate": 8.379355687047995e-06, "loss": 0.0084, "step": 49150 }, { "epoch": 31.743305029392555, "grad_norm": 0.08166509121656418, "learning_rate": 8.376068376068377e-06, "loss": 0.009, "step": 49200 }, { "epoch": 31.77596342259961, "grad_norm": 0.07103191316127777, "learning_rate": 8.372781065088758e-06, "loss": 0.0087, "step": 49250 }, { "epoch": 31.808621815806664, "grad_norm": 0.08163702487945557, "learning_rate": 8.36949375410914e-06, "loss": 0.0088, "step": 49300 }, { "epoch": 31.841280209013718, "grad_norm": 0.07810921967029572, "learning_rate": 8.366206443129521e-06, "loss": 0.009, "step": 49350 }, { "epoch": 31.873938602220772, "grad_norm": 0.0793294832110405, "learning_rate": 8.362919132149902e-06, "loss": 0.0091, "step": 49400 }, { "epoch": 31.906596995427826, "grad_norm": 0.09774003177881241, "learning_rate": 8.359631821170284e-06, "loss": 0.0086, "step": 49450 }, { "epoch": 31.93925538863488, "grad_norm": 0.06942662596702576, "learning_rate": 8.356344510190665e-06, "loss": 0.0087, "step": 49500 }, { "epoch": 31.971913781841934, "grad_norm": 0.07087003439664841, "learning_rate": 8.353057199211047e-06, "loss": 0.0086, "step": 49550 }, { "epoch": 32.004572175048985, "grad_norm": 0.06099782511591911, "learning_rate": 8.349769888231426e-06, "loss": 0.009, "step": 49600 }, { "epoch": 32.03723056825604, "grad_norm": 0.07634332776069641, "learning_rate": 8.34648257725181e-06, "loss": 0.0081, "step": 49650 }, { "epoch": 32.06988896146309, "grad_norm": 0.06689299643039703, "learning_rate": 8.343195266272191e-06, "loss": 0.0085, "step": 49700 }, { "epoch": 32.10254735467015, "grad_norm": 0.06661924719810486, "learning_rate": 8.33990795529257e-06, "loss": 0.0082, "step": 49750 }, { "epoch": 32.1352057478772, "grad_norm": 0.09380127489566803, "learning_rate": 8.336620644312952e-06, "loss": 0.0085, "step": 49800 }, { "epoch": 32.16786414108426, "grad_norm": 0.10260223597288132, "learning_rate": 8.333333333333334e-06, "loss": 0.0081, "step": 49850 }, { "epoch": 32.200522534291316, "grad_norm": 0.06356624513864517, "learning_rate": 8.330046022353715e-06, "loss": 0.0087, "step": 49900 }, { "epoch": 32.23318092749837, "grad_norm": 0.07155543565750122, "learning_rate": 8.326758711374097e-06, "loss": 0.0088, "step": 49950 }, { "epoch": 32.265839320705425, "grad_norm": 0.10149654746055603, "learning_rate": 8.323471400394478e-06, "loss": 0.0088, "step": 50000 }, { "epoch": 32.298497713912475, "grad_norm": 0.06213049963116646, "learning_rate": 8.32018408941486e-06, "loss": 0.0082, "step": 50050 }, { "epoch": 32.33115610711953, "grad_norm": 0.07706457376480103, "learning_rate": 8.316896778435241e-06, "loss": 0.0085, "step": 50100 }, { "epoch": 32.36381450032658, "grad_norm": 0.08293071389198303, "learning_rate": 8.313609467455622e-06, "loss": 0.0086, "step": 50150 }, { "epoch": 32.39647289353364, "grad_norm": 0.0799180418252945, "learning_rate": 8.310322156476002e-06, "loss": 0.0085, "step": 50200 }, { "epoch": 32.42913128674069, "grad_norm": 0.08061999827623367, "learning_rate": 8.307034845496384e-06, "loss": 0.009, "step": 50250 }, { "epoch": 32.46178967994775, "grad_norm": 0.059159934520721436, "learning_rate": 8.303747534516767e-06, "loss": 0.0083, "step": 50300 }, { "epoch": 32.4944480731548, "grad_norm": 0.09864724427461624, "learning_rate": 8.300460223537147e-06, "loss": 0.0084, "step": 50350 }, { "epoch": 32.52710646636186, "grad_norm": 0.06515152007341385, "learning_rate": 8.297172912557528e-06, "loss": 0.0085, "step": 50400 }, { "epoch": 32.55976485956891, "grad_norm": 0.05703325942158699, "learning_rate": 8.29388560157791e-06, "loss": 0.0082, "step": 50450 }, { "epoch": 32.592423252775966, "grad_norm": 0.06730402261018753, "learning_rate": 8.290598290598293e-06, "loss": 0.0087, "step": 50500 }, { "epoch": 32.625081645983016, "grad_norm": 0.08516230434179306, "learning_rate": 8.287310979618672e-06, "loss": 0.0085, "step": 50550 }, { "epoch": 32.657740039190074, "grad_norm": 0.07388225942850113, "learning_rate": 8.284023668639054e-06, "loss": 0.0088, "step": 50600 }, { "epoch": 32.690398432397124, "grad_norm": 0.0825827568769455, "learning_rate": 8.280736357659435e-06, "loss": 0.0086, "step": 50650 }, { "epoch": 32.72305682560418, "grad_norm": 0.08514679968357086, "learning_rate": 8.277449046679817e-06, "loss": 0.0086, "step": 50700 }, { "epoch": 32.75571521881123, "grad_norm": 0.05570615082979202, "learning_rate": 8.274161735700198e-06, "loss": 0.0087, "step": 50750 }, { "epoch": 32.78837361201829, "grad_norm": 0.09362334758043289, "learning_rate": 8.27087442472058e-06, "loss": 0.0081, "step": 50800 }, { "epoch": 32.82103200522534, "grad_norm": 0.07868677377700806, "learning_rate": 8.26758711374096e-06, "loss": 0.0084, "step": 50850 }, { "epoch": 32.8536903984324, "grad_norm": 0.06164439022541046, "learning_rate": 8.264299802761343e-06, "loss": 0.0085, "step": 50900 }, { "epoch": 32.88634879163945, "grad_norm": 0.08538579940795898, "learning_rate": 8.261012491781724e-06, "loss": 0.0086, "step": 50950 }, { "epoch": 32.91900718484651, "grad_norm": 0.060735203325748444, "learning_rate": 8.257725180802104e-06, "loss": 0.0086, "step": 51000 }, { "epoch": 32.95166557805356, "grad_norm": 0.06922443956136703, "learning_rate": 8.254437869822485e-06, "loss": 0.0091, "step": 51050 }, { "epoch": 32.984323971260615, "grad_norm": 0.0882391557097435, "learning_rate": 8.251150558842867e-06, "loss": 0.008, "step": 51100 }, { "epoch": 33.016982364467665, "grad_norm": 0.07642501592636108, "learning_rate": 8.247863247863248e-06, "loss": 0.0084, "step": 51150 }, { "epoch": 33.04964075767472, "grad_norm": 0.08096608519554138, "learning_rate": 8.24457593688363e-06, "loss": 0.008, "step": 51200 }, { "epoch": 33.082299150881774, "grad_norm": 0.05849987268447876, "learning_rate": 8.241288625904011e-06, "loss": 0.0082, "step": 51250 }, { "epoch": 33.11495754408883, "grad_norm": 0.13448837399482727, "learning_rate": 8.238001314924393e-06, "loss": 0.0083, "step": 51300 }, { "epoch": 33.14761593729588, "grad_norm": 0.06831568479537964, "learning_rate": 8.234714003944774e-06, "loss": 0.0084, "step": 51350 }, { "epoch": 33.18027433050294, "grad_norm": 0.06570922583341599, "learning_rate": 8.231426692965155e-06, "loss": 0.0086, "step": 51400 }, { "epoch": 33.21293272370999, "grad_norm": 0.062486033886671066, "learning_rate": 8.228139381985535e-06, "loss": 0.0083, "step": 51450 }, { "epoch": 33.24559111691705, "grad_norm": 0.07701075077056885, "learning_rate": 8.224852071005918e-06, "loss": 0.0083, "step": 51500 }, { "epoch": 33.278249510124105, "grad_norm": 0.06391100585460663, "learning_rate": 8.2215647600263e-06, "loss": 0.0082, "step": 51550 }, { "epoch": 33.310907903331156, "grad_norm": 0.06537971645593643, "learning_rate": 8.218277449046681e-06, "loss": 0.0082, "step": 51600 }, { "epoch": 33.343566296538214, "grad_norm": 0.08056467771530151, "learning_rate": 8.214990138067061e-06, "loss": 0.0081, "step": 51650 }, { "epoch": 33.376224689745264, "grad_norm": 0.08163385093212128, "learning_rate": 8.211702827087442e-06, "loss": 0.0081, "step": 51700 }, { "epoch": 33.40888308295232, "grad_norm": 0.09512770175933838, "learning_rate": 8.208415516107826e-06, "loss": 0.0079, "step": 51750 }, { "epoch": 33.44154147615937, "grad_norm": 0.060141269117593765, "learning_rate": 8.205128205128205e-06, "loss": 0.0084, "step": 51800 }, { "epoch": 33.47419986936643, "grad_norm": 0.04651719331741333, "learning_rate": 8.201840894148587e-06, "loss": 0.0083, "step": 51850 }, { "epoch": 33.50685826257348, "grad_norm": 0.0626654326915741, "learning_rate": 8.198553583168968e-06, "loss": 0.0084, "step": 51900 }, { "epoch": 33.53951665578054, "grad_norm": 0.060986001044511795, "learning_rate": 8.19526627218935e-06, "loss": 0.0087, "step": 51950 }, { "epoch": 33.57217504898759, "grad_norm": 0.08697523176670074, "learning_rate": 8.191978961209731e-06, "loss": 0.0086, "step": 52000 }, { "epoch": 33.604833442194646, "grad_norm": 0.06833694875240326, "learning_rate": 8.188691650230113e-06, "loss": 0.0082, "step": 52050 }, { "epoch": 33.6374918354017, "grad_norm": 0.06158861890435219, "learning_rate": 8.185404339250494e-06, "loss": 0.0083, "step": 52100 }, { "epoch": 33.670150228608755, "grad_norm": 0.08568635582923889, "learning_rate": 8.182117028270876e-06, "loss": 0.0084, "step": 52150 }, { "epoch": 33.702808621815805, "grad_norm": 0.0662611722946167, "learning_rate": 8.178829717291257e-06, "loss": 0.0083, "step": 52200 }, { "epoch": 33.73546701502286, "grad_norm": 0.06524196267127991, "learning_rate": 8.175542406311637e-06, "loss": 0.0084, "step": 52250 }, { "epoch": 33.76812540822991, "grad_norm": 0.07562558352947235, "learning_rate": 8.172255095332018e-06, "loss": 0.0086, "step": 52300 }, { "epoch": 33.80078380143697, "grad_norm": 0.0772087424993515, "learning_rate": 8.168967784352401e-06, "loss": 0.0081, "step": 52350 }, { "epoch": 33.83344219464402, "grad_norm": 0.08937671035528183, "learning_rate": 8.165680473372781e-06, "loss": 0.0083, "step": 52400 }, { "epoch": 33.86610058785108, "grad_norm": 0.07191990315914154, "learning_rate": 8.162393162393163e-06, "loss": 0.0085, "step": 52450 }, { "epoch": 33.89875898105813, "grad_norm": 0.04881991446018219, "learning_rate": 8.159105851413544e-06, "loss": 0.0081, "step": 52500 }, { "epoch": 33.93141737426519, "grad_norm": 0.07393967360258102, "learning_rate": 8.155818540433925e-06, "loss": 0.0084, "step": 52550 }, { "epoch": 33.96407576747224, "grad_norm": 0.08467967063188553, "learning_rate": 8.152531229454307e-06, "loss": 0.0081, "step": 52600 }, { "epoch": 33.996734160679296, "grad_norm": 0.058565691113471985, "learning_rate": 8.149243918474688e-06, "loss": 0.0083, "step": 52650 }, { "epoch": 34.029392553886346, "grad_norm": 0.07146356254816055, "learning_rate": 8.14595660749507e-06, "loss": 0.0079, "step": 52700 }, { "epoch": 34.062050947093404, "grad_norm": 0.06826076656579971, "learning_rate": 8.142669296515451e-06, "loss": 0.008, "step": 52750 }, { "epoch": 34.094709340300454, "grad_norm": 0.08253225684165955, "learning_rate": 8.139381985535833e-06, "loss": 0.0082, "step": 52800 }, { "epoch": 34.12736773350751, "grad_norm": 0.07931675016880035, "learning_rate": 8.136094674556214e-06, "loss": 0.0081, "step": 52850 }, { "epoch": 34.16002612671456, "grad_norm": 0.05465434119105339, "learning_rate": 8.132807363576594e-06, "loss": 0.0079, "step": 52900 }, { "epoch": 34.19268451992162, "grad_norm": 0.06749475002288818, "learning_rate": 8.129520052596977e-06, "loss": 0.0079, "step": 52950 }, { "epoch": 34.22534291312867, "grad_norm": 0.07008467614650726, "learning_rate": 8.126232741617359e-06, "loss": 0.0082, "step": 53000 }, { "epoch": 34.25800130633573, "grad_norm": 0.07345925271511078, "learning_rate": 8.122945430637738e-06, "loss": 0.0084, "step": 53050 }, { "epoch": 34.29065969954278, "grad_norm": 0.0824049711227417, "learning_rate": 8.11965811965812e-06, "loss": 0.0081, "step": 53100 }, { "epoch": 34.32331809274984, "grad_norm": 0.07596274465322495, "learning_rate": 8.116370808678501e-06, "loss": 0.0083, "step": 53150 }, { "epoch": 34.355976485956894, "grad_norm": 0.05257643014192581, "learning_rate": 8.113083497698883e-06, "loss": 0.0081, "step": 53200 }, { "epoch": 34.388634879163945, "grad_norm": 0.07520149648189545, "learning_rate": 8.109796186719264e-06, "loss": 0.0078, "step": 53250 }, { "epoch": 34.421293272371, "grad_norm": 0.08370786160230637, "learning_rate": 8.106508875739646e-06, "loss": 0.0078, "step": 53300 }, { "epoch": 34.45395166557805, "grad_norm": 0.08980908989906311, "learning_rate": 8.103221564760027e-06, "loss": 0.0079, "step": 53350 }, { "epoch": 34.48661005878511, "grad_norm": 0.04652412608265877, "learning_rate": 8.099934253780409e-06, "loss": 0.0085, "step": 53400 }, { "epoch": 34.51926845199216, "grad_norm": 0.06980093568563461, "learning_rate": 8.09664694280079e-06, "loss": 0.0082, "step": 53450 }, { "epoch": 34.55192684519922, "grad_norm": 0.056630637496709824, "learning_rate": 8.09335963182117e-06, "loss": 0.0079, "step": 53500 }, { "epoch": 34.58458523840627, "grad_norm": 0.059379786252975464, "learning_rate": 8.090072320841553e-06, "loss": 0.0081, "step": 53550 }, { "epoch": 34.61724363161333, "grad_norm": 0.07385461032390594, "learning_rate": 8.086785009861934e-06, "loss": 0.0078, "step": 53600 }, { "epoch": 34.64990202482038, "grad_norm": 0.07951278239488602, "learning_rate": 8.083497698882314e-06, "loss": 0.0082, "step": 53650 }, { "epoch": 34.682560418027435, "grad_norm": 0.08047601580619812, "learning_rate": 8.080210387902696e-06, "loss": 0.0077, "step": 53700 }, { "epoch": 34.715218811234486, "grad_norm": 0.08462663739919662, "learning_rate": 8.076923076923077e-06, "loss": 0.0083, "step": 53750 }, { "epoch": 34.747877204441544, "grad_norm": 0.06819440424442291, "learning_rate": 8.07363576594346e-06, "loss": 0.008, "step": 53800 }, { "epoch": 34.780535597648594, "grad_norm": 0.07236823439598083, "learning_rate": 8.07034845496384e-06, "loss": 0.0083, "step": 53850 }, { "epoch": 34.81319399085565, "grad_norm": 0.07484553754329681, "learning_rate": 8.067061143984221e-06, "loss": 0.0082, "step": 53900 }, { "epoch": 34.8458523840627, "grad_norm": 0.07134470343589783, "learning_rate": 8.063773833004603e-06, "loss": 0.0082, "step": 53950 }, { "epoch": 34.87851077726976, "grad_norm": 0.06985644996166229, "learning_rate": 8.060486522024984e-06, "loss": 0.0086, "step": 54000 }, { "epoch": 34.91116917047681, "grad_norm": 0.06322511285543442, "learning_rate": 8.057199211045366e-06, "loss": 0.0078, "step": 54050 }, { "epoch": 34.94382756368387, "grad_norm": 0.10806406289339066, "learning_rate": 8.053911900065747e-06, "loss": 0.008, "step": 54100 }, { "epoch": 34.97648595689092, "grad_norm": 0.05752309784293175, "learning_rate": 8.050624589086129e-06, "loss": 0.0082, "step": 54150 }, { "epoch": 35.009144350097976, "grad_norm": 0.07018327713012695, "learning_rate": 8.04733727810651e-06, "loss": 0.0085, "step": 54200 }, { "epoch": 35.04180274330503, "grad_norm": 0.0636325404047966, "learning_rate": 8.044049967126892e-06, "loss": 0.0081, "step": 54250 }, { "epoch": 35.074461136512085, "grad_norm": 0.09761761873960495, "learning_rate": 8.040762656147271e-06, "loss": 0.0079, "step": 54300 }, { "epoch": 35.107119529719135, "grad_norm": 0.06210368871688843, "learning_rate": 8.037475345167653e-06, "loss": 0.008, "step": 54350 }, { "epoch": 35.13977792292619, "grad_norm": 0.057588595896959305, "learning_rate": 8.034188034188036e-06, "loss": 0.0081, "step": 54400 }, { "epoch": 35.17243631613324, "grad_norm": 0.09066003561019897, "learning_rate": 8.030900723208416e-06, "loss": 0.0081, "step": 54450 }, { "epoch": 35.2050947093403, "grad_norm": 0.07244653254747391, "learning_rate": 8.027613412228797e-06, "loss": 0.0079, "step": 54500 }, { "epoch": 36.03331112591606, "grad_norm": 0.06708467751741409, "learning_rate": 7.984574111334675e-06, "loss": 0.0076, "step": 54550 }, { "epoch": 36.066622251832115, "grad_norm": 0.07713846862316132, "learning_rate": 7.981220657276996e-06, "loss": 0.0079, "step": 54600 }, { "epoch": 36.099933377748165, "grad_norm": 0.09249703586101532, "learning_rate": 7.977867203219316e-06, "loss": 0.0079, "step": 54650 }, { "epoch": 36.13324450366422, "grad_norm": 0.07342194020748138, "learning_rate": 7.974513749161638e-06, "loss": 0.0081, "step": 54700 }, { "epoch": 36.16655562958028, "grad_norm": 0.060539375990629196, "learning_rate": 7.971160295103958e-06, "loss": 0.0076, "step": 54750 }, { "epoch": 36.19986675549634, "grad_norm": 0.07849667221307755, "learning_rate": 7.967806841046278e-06, "loss": 0.0077, "step": 54800 }, { "epoch": 36.233177881412395, "grad_norm": 0.07112744450569153, "learning_rate": 7.964453386988599e-06, "loss": 0.0081, "step": 54850 }, { "epoch": 36.266489007328445, "grad_norm": 0.09162628650665283, "learning_rate": 7.961099932930919e-06, "loss": 0.0078, "step": 54900 }, { "epoch": 36.2998001332445, "grad_norm": 0.177199587225914, "learning_rate": 7.95774647887324e-06, "loss": 0.0079, "step": 54950 }, { "epoch": 36.33311125916056, "grad_norm": 0.06649234890937805, "learning_rate": 7.95439302481556e-06, "loss": 0.0079, "step": 55000 }, { "epoch": 20.018254837531945, "grad_norm": 0.06797217577695847, "learning_rate": 8.88054232319531e-06, "loss": 0.0094, "step": 55050 }, { "epoch": 20.036509675063893, "grad_norm": 0.06051107123494148, "learning_rate": 8.878710150238184e-06, "loss": 0.0092, "step": 55100 }, { "epoch": 20.05476451259584, "grad_norm": 0.08235818892717361, "learning_rate": 8.876877977281056e-06, "loss": 0.0094, "step": 55150 }, { "epoch": 20.073019350127783, "grad_norm": 0.0761820450425148, "learning_rate": 8.875045804323929e-06, "loss": 0.0093, "step": 55200 }, { "epoch": 20.091274187659728, "grad_norm": 0.05336988344788551, "learning_rate": 8.873213631366801e-06, "loss": 0.0093, "step": 55250 }, { "epoch": 20.109529025191677, "grad_norm": 0.0651998296380043, "learning_rate": 8.871381458409675e-06, "loss": 0.0092, "step": 55300 }, { "epoch": 20.12778386272362, "grad_norm": 0.05962739884853363, "learning_rate": 8.869549285452547e-06, "loss": 0.009, "step": 55350 }, { "epoch": 20.146038700255566, "grad_norm": 0.05445632338523865, "learning_rate": 8.86771711249542e-06, "loss": 0.0092, "step": 55400 }, { "epoch": 20.164293537787515, "grad_norm": 0.06587453186511993, "learning_rate": 8.865884939538293e-06, "loss": 0.0091, "step": 55450 }, { "epoch": 20.18254837531946, "grad_norm": 0.0614401176571846, "learning_rate": 8.864052766581166e-06, "loss": 0.0094, "step": 55500 }, { "epoch": 20.200803212851405, "grad_norm": 0.06000984460115433, "learning_rate": 8.86222059362404e-06, "loss": 0.0091, "step": 55550 }, { "epoch": 20.219058050383353, "grad_norm": 0.05823017284274101, "learning_rate": 8.860388420666912e-06, "loss": 0.0092, "step": 55600 }, { "epoch": 20.237312887915298, "grad_norm": 0.05388309061527252, "learning_rate": 8.858556247709784e-06, "loss": 0.0095, "step": 55650 }, { "epoch": 20.255567725447243, "grad_norm": 0.06236082687973976, "learning_rate": 8.856724074752657e-06, "loss": 0.0091, "step": 55700 }, { "epoch": 20.27382256297919, "grad_norm": 0.06299027055501938, "learning_rate": 8.85489190179553e-06, "loss": 0.0091, "step": 55750 }, { "epoch": 20.292077400511136, "grad_norm": 0.06052004173398018, "learning_rate": 8.853059728838403e-06, "loss": 0.0091, "step": 55800 }, { "epoch": 20.31033223804308, "grad_norm": 0.08799801021814346, "learning_rate": 8.851227555881275e-06, "loss": 0.0086, "step": 55850 }, { "epoch": 20.328587075575026, "grad_norm": 0.06422752887010574, "learning_rate": 8.84939538292415e-06, "loss": 0.0095, "step": 55900 }, { "epoch": 20.346841913106974, "grad_norm": 0.08138249814510345, "learning_rate": 8.847563209967022e-06, "loss": 0.009, "step": 55950 }, { "epoch": 20.36509675063892, "grad_norm": 0.07572277635335922, "learning_rate": 8.845731037009896e-06, "loss": 0.0091, "step": 56000 }, { "epoch": 20.383351588170864, "grad_norm": 0.06524857133626938, "learning_rate": 8.843898864052766e-06, "loss": 0.0092, "step": 56050 }, { "epoch": 20.401606425702813, "grad_norm": 0.06799105554819107, "learning_rate": 8.84206669109564e-06, "loss": 0.0089, "step": 56100 }, { "epoch": 20.419861263234758, "grad_norm": 0.06698976457118988, "learning_rate": 8.840234518138512e-06, "loss": 0.009, "step": 56150 }, { "epoch": 20.438116100766702, "grad_norm": 0.06604661047458649, "learning_rate": 8.838402345181386e-06, "loss": 0.0094, "step": 56200 }, { "epoch": 20.456370938298647, "grad_norm": 0.04851552098989487, "learning_rate": 8.836570172224259e-06, "loss": 0.0088, "step": 56250 }, { "epoch": 20.474625775830596, "grad_norm": 0.07488046586513519, "learning_rate": 8.834737999267131e-06, "loss": 0.0093, "step": 56300 }, { "epoch": 20.49288061336254, "grad_norm": 0.06332817673683167, "learning_rate": 8.832905826310005e-06, "loss": 0.0093, "step": 56350 }, { "epoch": 20.511135450894486, "grad_norm": 0.04852518066763878, "learning_rate": 8.831073653352877e-06, "loss": 0.0094, "step": 56400 }, { "epoch": 20.529390288426434, "grad_norm": 0.06774722039699554, "learning_rate": 8.82924148039575e-06, "loss": 0.0093, "step": 56450 }, { "epoch": 20.54764512595838, "grad_norm": 0.059105101972818375, "learning_rate": 8.827409307438622e-06, "loss": 0.0091, "step": 56500 }, { "epoch": 20.565899963490324, "grad_norm": 0.052642662078142166, "learning_rate": 8.825577134481496e-06, "loss": 0.0086, "step": 56550 }, { "epoch": 20.584154801022272, "grad_norm": 0.07305848598480225, "learning_rate": 8.823744961524368e-06, "loss": 0.0091, "step": 56600 }, { "epoch": 20.602409638554217, "grad_norm": 0.05794287100434303, "learning_rate": 8.821912788567242e-06, "loss": 0.0093, "step": 56650 }, { "epoch": 20.620664476086162, "grad_norm": 0.05902906507253647, "learning_rate": 8.820080615610115e-06, "loss": 0.009, "step": 56700 }, { "epoch": 20.63891931361811, "grad_norm": 0.04885618016123772, "learning_rate": 8.818248442652987e-06, "loss": 0.0095, "step": 56750 }, { "epoch": 20.657174151150056, "grad_norm": 0.0719715803861618, "learning_rate": 8.81641626969586e-06, "loss": 0.0092, "step": 56800 }, { "epoch": 20.675428988682, "grad_norm": 0.0657823383808136, "learning_rate": 8.814584096738733e-06, "loss": 0.0092, "step": 56850 }, { "epoch": 20.693683826213945, "grad_norm": 0.13935545086860657, "learning_rate": 8.812751923781605e-06, "loss": 0.0092, "step": 56900 }, { "epoch": 20.711938663745894, "grad_norm": 0.0742974728345871, "learning_rate": 8.810919750824478e-06, "loss": 0.0092, "step": 56950 }, { "epoch": 20.73019350127784, "grad_norm": 0.09424673765897751, "learning_rate": 8.809087577867352e-06, "loss": 0.0088, "step": 57000 }, { "epoch": 20.748448338809784, "grad_norm": 0.10960248857736588, "learning_rate": 8.807255404910224e-06, "loss": 0.0092, "step": 57050 }, { "epoch": 20.766703176341732, "grad_norm": 0.05283306911587715, "learning_rate": 8.805423231953098e-06, "loss": 0.009, "step": 57100 }, { "epoch": 20.784958013873677, "grad_norm": 0.060610584914684296, "learning_rate": 8.803591058995969e-06, "loss": 0.0092, "step": 57150 }, { "epoch": 20.803212851405622, "grad_norm": 0.05835588276386261, "learning_rate": 8.801758886038843e-06, "loss": 0.0091, "step": 57200 }, { "epoch": 20.821467688937567, "grad_norm": 0.06471339613199234, "learning_rate": 8.799926713081715e-06, "loss": 0.0093, "step": 57250 }, { "epoch": 20.839722526469515, "grad_norm": 0.05594169721007347, "learning_rate": 8.798094540124589e-06, "loss": 0.0092, "step": 57300 }, { "epoch": 20.85797736400146, "grad_norm": 0.05852265655994415, "learning_rate": 8.796262367167461e-06, "loss": 0.0092, "step": 57350 }, { "epoch": 20.876232201533405, "grad_norm": 0.07459704577922821, "learning_rate": 8.794430194210334e-06, "loss": 0.0094, "step": 57400 }, { "epoch": 20.894487039065353, "grad_norm": 0.051926735788583755, "learning_rate": 8.792598021253208e-06, "loss": 0.0093, "step": 57450 }, { "epoch": 20.9127418765973, "grad_norm": 0.07277660816907883, "learning_rate": 8.79076584829608e-06, "loss": 0.0093, "step": 57500 }, { "epoch": 20.930996714129243, "grad_norm": 0.059177327901124954, "learning_rate": 8.788933675338954e-06, "loss": 0.009, "step": 57550 }, { "epoch": 20.94925155166119, "grad_norm": 0.06189264729619026, "learning_rate": 8.787101502381825e-06, "loss": 0.0095, "step": 57600 }, { "epoch": 20.967506389193137, "grad_norm": 0.061540741473436356, "learning_rate": 8.785269329424699e-06, "loss": 0.0091, "step": 57650 }, { "epoch": 20.98576122672508, "grad_norm": 0.06274130940437317, "learning_rate": 8.78343715646757e-06, "loss": 0.0092, "step": 57700 }, { "epoch": 21.004016064257026, "grad_norm": 0.07412123680114746, "learning_rate": 8.781604983510445e-06, "loss": 0.009, "step": 57750 }, { "epoch": 21.022270901788975, "grad_norm": 0.06972913444042206, "learning_rate": 8.779772810553317e-06, "loss": 0.0091, "step": 57800 }, { "epoch": 21.04052573932092, "grad_norm": 0.05870004743337631, "learning_rate": 8.77794063759619e-06, "loss": 0.0092, "step": 57850 }, { "epoch": 21.058780576852865, "grad_norm": 0.060268133878707886, "learning_rate": 8.776108464639063e-06, "loss": 0.0087, "step": 57900 }, { "epoch": 21.077035414384813, "grad_norm": 0.08082205802202225, "learning_rate": 8.774276291681936e-06, "loss": 0.0089, "step": 57950 }, { "epoch": 21.095290251916758, "grad_norm": 0.06273343414068222, "learning_rate": 8.772444118724808e-06, "loss": 0.0088, "step": 58000 }, { "epoch": 21.113545089448703, "grad_norm": 0.06352389603853226, "learning_rate": 8.77061194576768e-06, "loss": 0.0092, "step": 58050 }, { "epoch": 21.13179992698065, "grad_norm": 0.09286611527204514, "learning_rate": 8.768779772810554e-06, "loss": 0.0089, "step": 58100 }, { "epoch": 21.150054764512596, "grad_norm": 0.04642708972096443, "learning_rate": 8.766947599853427e-06, "loss": 0.0086, "step": 58150 }, { "epoch": 21.16830960204454, "grad_norm": 0.06841983646154404, "learning_rate": 8.7651154268963e-06, "loss": 0.0092, "step": 58200 }, { "epoch": 21.186564439576486, "grad_norm": 0.13237148523330688, "learning_rate": 8.763283253939173e-06, "loss": 0.0091, "step": 58250 }, { "epoch": 21.204819277108435, "grad_norm": 0.0589657723903656, "learning_rate": 8.761451080982045e-06, "loss": 0.009, "step": 58300 }, { "epoch": 21.22307411464038, "grad_norm": 0.0630548968911171, "learning_rate": 8.759618908024918e-06, "loss": 0.0086, "step": 58350 }, { "epoch": 21.241328952172324, "grad_norm": 0.05287938192486763, "learning_rate": 8.757786735067792e-06, "loss": 0.0089, "step": 58400 }, { "epoch": 21.259583789704273, "grad_norm": 0.09002646803855896, "learning_rate": 8.755954562110664e-06, "loss": 0.0088, "step": 58450 }, { "epoch": 21.277838627236218, "grad_norm": 0.09573077410459518, "learning_rate": 8.754122389153536e-06, "loss": 0.009, "step": 58500 }, { "epoch": 21.296093464768163, "grad_norm": 0.09754585474729538, "learning_rate": 8.75229021619641e-06, "loss": 0.0089, "step": 58550 }, { "epoch": 21.31434830230011, "grad_norm": 0.07476762682199478, "learning_rate": 8.750458043239282e-06, "loss": 0.0088, "step": 58600 }, { "epoch": 21.332603139832056, "grad_norm": 0.07495803385972977, "learning_rate": 8.748625870282156e-06, "loss": 0.0095, "step": 58650 }, { "epoch": 21.350857977364, "grad_norm": 0.057302817702293396, "learning_rate": 8.746793697325027e-06, "loss": 0.0088, "step": 58700 }, { "epoch": 21.369112814895946, "grad_norm": 0.06529004126787186, "learning_rate": 8.744961524367901e-06, "loss": 0.0089, "step": 58750 }, { "epoch": 21.387367652427894, "grad_norm": 0.07001931965351105, "learning_rate": 8.743129351410773e-06, "loss": 0.009, "step": 58800 }, { "epoch": 21.40562248995984, "grad_norm": 0.056651655584573746, "learning_rate": 8.741297178453647e-06, "loss": 0.0089, "step": 58850 }, { "epoch": 21.423877327491784, "grad_norm": 0.08324690163135529, "learning_rate": 8.73946500549652e-06, "loss": 0.0084, "step": 58900 }, { "epoch": 21.442132165023732, "grad_norm": 0.062204282730817795, "learning_rate": 8.737632832539392e-06, "loss": 0.0091, "step": 58950 }, { "epoch": 21.460387002555677, "grad_norm": 0.058759935200214386, "learning_rate": 8.735800659582266e-06, "loss": 0.0087, "step": 59000 }, { "epoch": 21.478641840087622, "grad_norm": 0.09147629886865616, "learning_rate": 8.733968486625138e-06, "loss": 0.0089, "step": 59050 }, { "epoch": 21.49689667761957, "grad_norm": 0.050570517778396606, "learning_rate": 8.732136313668012e-06, "loss": 0.0089, "step": 59100 }, { "epoch": 21.515151515151516, "grad_norm": 0.11068356782197952, "learning_rate": 8.730304140710883e-06, "loss": 0.0089, "step": 59150 }, { "epoch": 21.53340635268346, "grad_norm": 0.05365932732820511, "learning_rate": 8.728471967753757e-06, "loss": 0.0087, "step": 59200 }, { "epoch": 21.551661190215405, "grad_norm": 0.07228037714958191, "learning_rate": 8.726639794796629e-06, "loss": 0.0088, "step": 59250 }, { "epoch": 21.569916027747354, "grad_norm": 0.05023273453116417, "learning_rate": 8.724807621839503e-06, "loss": 0.0087, "step": 59300 }, { "epoch": 21.5881708652793, "grad_norm": 0.05931396409869194, "learning_rate": 8.722975448882375e-06, "loss": 0.0092, "step": 59350 }, { "epoch": 21.606425702811244, "grad_norm": 0.08150698989629745, "learning_rate": 8.721143275925248e-06, "loss": 0.009, "step": 59400 }, { "epoch": 21.624680540343192, "grad_norm": 0.05365688353776932, "learning_rate": 8.719311102968122e-06, "loss": 0.0089, "step": 59450 }, { "epoch": 21.642935377875137, "grad_norm": 0.056187257170677185, "learning_rate": 8.717478930010994e-06, "loss": 0.0091, "step": 59500 }, { "epoch": 21.661190215407082, "grad_norm": 0.058979302644729614, "learning_rate": 8.715646757053866e-06, "loss": 0.009, "step": 59550 }, { "epoch": 21.67944505293903, "grad_norm": 0.057258591055870056, "learning_rate": 8.713814584096739e-06, "loss": 0.0091, "step": 59600 }, { "epoch": 21.697699890470975, "grad_norm": 0.049408867955207825, "learning_rate": 8.711982411139613e-06, "loss": 0.0085, "step": 59650 }, { "epoch": 21.71595472800292, "grad_norm": 0.06210870295763016, "learning_rate": 8.710150238182485e-06, "loss": 0.0086, "step": 59700 }, { "epoch": 21.734209565534865, "grad_norm": 0.06718998402357101, "learning_rate": 8.708318065225359e-06, "loss": 0.009, "step": 59750 }, { "epoch": 21.752464403066814, "grad_norm": 0.07591931521892548, "learning_rate": 8.706485892268231e-06, "loss": 0.0092, "step": 59800 }, { "epoch": 21.77071924059876, "grad_norm": 0.07207608222961426, "learning_rate": 8.704653719311104e-06, "loss": 0.0087, "step": 59850 }, { "epoch": 21.788974078130703, "grad_norm": 0.07853410392999649, "learning_rate": 8.702821546353976e-06, "loss": 0.0087, "step": 59900 }, { "epoch": 21.80722891566265, "grad_norm": 0.08041689544916153, "learning_rate": 8.70098937339685e-06, "loss": 0.009, "step": 59950 }, { "epoch": 21.825483753194597, "grad_norm": 0.08932217210531235, "learning_rate": 8.699157200439722e-06, "loss": 0.0087, "step": 60000 }, { "epoch": 21.84373859072654, "grad_norm": 0.07520779222249985, "learning_rate": 8.697325027482594e-06, "loss": 0.0085, "step": 60050 }, { "epoch": 21.86199342825849, "grad_norm": 0.05378332734107971, "learning_rate": 8.695492854525468e-06, "loss": 0.0089, "step": 60100 }, { "epoch": 21.880248265790435, "grad_norm": 0.06476875394582748, "learning_rate": 8.69366068156834e-06, "loss": 0.0084, "step": 60150 }, { "epoch": 21.89850310332238, "grad_norm": 0.0820067971944809, "learning_rate": 8.691828508611213e-06, "loss": 0.009, "step": 60200 }, { "epoch": 21.916757940854325, "grad_norm": 0.05638987198472023, "learning_rate": 8.689996335654085e-06, "loss": 0.0089, "step": 60250 }, { "epoch": 21.935012778386273, "grad_norm": 0.07554665952920914, "learning_rate": 8.68816416269696e-06, "loss": 0.0087, "step": 60300 }, { "epoch": 21.953267615918218, "grad_norm": 0.0646883025765419, "learning_rate": 8.686331989739832e-06, "loss": 0.0089, "step": 60350 }, { "epoch": 21.971522453450163, "grad_norm": 0.055584002286195755, "learning_rate": 8.684499816782706e-06, "loss": 0.0089, "step": 60400 }, { "epoch": 21.98977729098211, "grad_norm": 0.04451603814959526, "learning_rate": 8.682667643825578e-06, "loss": 0.0086, "step": 60450 }, { "epoch": 22.008032128514056, "grad_norm": 0.0625370591878891, "learning_rate": 8.68083547086845e-06, "loss": 0.0088, "step": 60500 }, { "epoch": 22.026286966046, "grad_norm": 0.06393207609653473, "learning_rate": 8.679003297911324e-06, "loss": 0.009, "step": 60550 }, { "epoch": 22.04454180357795, "grad_norm": 0.06288832426071167, "learning_rate": 8.677171124954197e-06, "loss": 0.0086, "step": 60600 }, { "epoch": 22.062796641109895, "grad_norm": 0.08870762586593628, "learning_rate": 8.675338951997069e-06, "loss": 0.0086, "step": 60650 }, { "epoch": 22.08105147864184, "grad_norm": 0.06099590286612511, "learning_rate": 8.673506779039941e-06, "loss": 0.0083, "step": 60700 }, { "epoch": 22.099306316173784, "grad_norm": 0.06576605886220932, "learning_rate": 8.671674606082815e-06, "loss": 0.0085, "step": 60750 }, { "epoch": 22.117561153705733, "grad_norm": 0.06348154693841934, "learning_rate": 8.669842433125687e-06, "loss": 0.0086, "step": 60800 }, { "epoch": 22.135815991237678, "grad_norm": 0.07181636989116669, "learning_rate": 8.668010260168561e-06, "loss": 0.0086, "step": 60850 }, { "epoch": 22.154070828769623, "grad_norm": 0.06625356525182724, "learning_rate": 8.666178087211434e-06, "loss": 0.0089, "step": 60900 }, { "epoch": 22.17232566630157, "grad_norm": 0.0668405219912529, "learning_rate": 8.664345914254306e-06, "loss": 0.009, "step": 60950 }, { "epoch": 22.190580503833516, "grad_norm": 0.06813893467187881, "learning_rate": 8.66251374129718e-06, "loss": 0.0085, "step": 61000 }, { "epoch": 22.20883534136546, "grad_norm": 0.06065446883440018, "learning_rate": 8.660681568340052e-06, "loss": 0.009, "step": 61050 }, { "epoch": 22.22709017889741, "grad_norm": 0.06655614823102951, "learning_rate": 8.658849395382925e-06, "loss": 0.0083, "step": 61100 }, { "epoch": 22.245345016429354, "grad_norm": 0.05601873621344566, "learning_rate": 8.657017222425797e-06, "loss": 0.009, "step": 61150 }, { "epoch": 22.2635998539613, "grad_norm": 0.05184505879878998, "learning_rate": 8.655185049468671e-06, "loss": 0.0082, "step": 61200 }, { "epoch": 22.281854691493244, "grad_norm": 0.07137031853199005, "learning_rate": 8.653352876511543e-06, "loss": 0.0084, "step": 61250 }, { "epoch": 22.300109529025193, "grad_norm": 0.1407080590724945, "learning_rate": 8.651520703554416e-06, "loss": 0.0085, "step": 61300 }, { "epoch": 22.318364366557137, "grad_norm": 0.08005314320325851, "learning_rate": 8.64968853059729e-06, "loss": 0.0088, "step": 61350 }, { "epoch": 22.336619204089082, "grad_norm": 0.06484245508909225, "learning_rate": 8.647856357640162e-06, "loss": 0.0085, "step": 61400 }, { "epoch": 22.35487404162103, "grad_norm": 0.05882987752556801, "learning_rate": 8.646024184683036e-06, "loss": 0.0089, "step": 61450 }, { "epoch": 22.373128879152976, "grad_norm": 0.07112240791320801, "learning_rate": 8.644192011725908e-06, "loss": 0.0089, "step": 61500 }, { "epoch": 22.39138371668492, "grad_norm": 0.05574699118733406, "learning_rate": 8.64235983876878e-06, "loss": 0.0084, "step": 61550 }, { "epoch": 22.40963855421687, "grad_norm": 0.08208150416612625, "learning_rate": 8.640527665811653e-06, "loss": 0.0087, "step": 61600 }, { "epoch": 22.427893391748814, "grad_norm": 0.07382607460021973, "learning_rate": 8.638695492854527e-06, "loss": 0.0087, "step": 61650 }, { "epoch": 22.44614822928076, "grad_norm": 0.09085138887166977, "learning_rate": 8.636863319897399e-06, "loss": 0.0084, "step": 61700 }, { "epoch": 22.464403066812704, "grad_norm": 0.06320594251155853, "learning_rate": 8.635031146940271e-06, "loss": 0.0088, "step": 61750 }, { "epoch": 22.482657904344652, "grad_norm": 0.0756024494767189, "learning_rate": 8.633198973983145e-06, "loss": 0.0088, "step": 61800 }, { "epoch": 22.500912741876597, "grad_norm": 0.057025641202926636, "learning_rate": 8.631366801026018e-06, "loss": 0.0087, "step": 61850 }, { "epoch": 22.519167579408542, "grad_norm": 0.06510153412818909, "learning_rate": 8.62953462806889e-06, "loss": 0.0089, "step": 61900 }, { "epoch": 22.53742241694049, "grad_norm": 0.05341318994760513, "learning_rate": 8.627702455111762e-06, "loss": 0.0082, "step": 61950 }, { "epoch": 22.555677254472435, "grad_norm": 0.0649246796965599, "learning_rate": 8.625870282154636e-06, "loss": 0.0088, "step": 62000 }, { "epoch": 22.57393209200438, "grad_norm": 0.07860676199197769, "learning_rate": 8.624038109197509e-06, "loss": 0.0084, "step": 62050 }, { "epoch": 22.59218692953633, "grad_norm": 0.06481479853391647, "learning_rate": 8.622205936240383e-06, "loss": 0.0083, "step": 62100 }, { "epoch": 22.610441767068274, "grad_norm": 0.06072891876101494, "learning_rate": 8.620373763283255e-06, "loss": 0.0086, "step": 62150 }, { "epoch": 22.62869660460022, "grad_norm": 0.09615935385227203, "learning_rate": 8.618541590326127e-06, "loss": 0.0087, "step": 62200 }, { "epoch": 22.646951442132163, "grad_norm": 0.0503789559006691, "learning_rate": 8.616709417369e-06, "loss": 0.0084, "step": 62250 }, { "epoch": 22.665206279664112, "grad_norm": 0.07485286891460419, "learning_rate": 8.614877244411874e-06, "loss": 0.0083, "step": 62300 }, { "epoch": 22.683461117196057, "grad_norm": 0.06894497573375702, "learning_rate": 8.613045071454746e-06, "loss": 0.0082, "step": 62350 }, { "epoch": 22.701715954728, "grad_norm": 0.07024747878313065, "learning_rate": 8.611212898497618e-06, "loss": 0.0087, "step": 62400 }, { "epoch": 22.71997079225995, "grad_norm": 0.08370257169008255, "learning_rate": 8.609380725540492e-06, "loss": 0.0088, "step": 62450 }, { "epoch": 22.738225629791895, "grad_norm": 0.10005588829517365, "learning_rate": 8.607548552583364e-06, "loss": 0.0089, "step": 62500 }, { "epoch": 22.75648046732384, "grad_norm": 0.11333871632814407, "learning_rate": 8.605716379626238e-06, "loss": 0.009, "step": 62550 }, { "epoch": 22.77473530485579, "grad_norm": 0.0648239403963089, "learning_rate": 8.603884206669109e-06, "loss": 0.0085, "step": 62600 }, { "epoch": 22.792990142387733, "grad_norm": 0.0639931783080101, "learning_rate": 8.602052033711983e-06, "loss": 0.0087, "step": 62650 }, { "epoch": 22.811244979919678, "grad_norm": 0.0740325078368187, "learning_rate": 8.600219860754855e-06, "loss": 0.0085, "step": 62700 }, { "epoch": 22.829499817451623, "grad_norm": 0.05029847472906113, "learning_rate": 8.59838768779773e-06, "loss": 0.0087, "step": 62750 }, { "epoch": 22.84775465498357, "grad_norm": 0.07053235918283463, "learning_rate": 8.596555514840602e-06, "loss": 0.0088, "step": 62800 }, { "epoch": 22.866009492515516, "grad_norm": 0.0815574899315834, "learning_rate": 8.594723341883474e-06, "loss": 0.0086, "step": 62850 }, { "epoch": 22.88426433004746, "grad_norm": 0.05690452829003334, "learning_rate": 8.592891168926348e-06, "loss": 0.0089, "step": 62900 }, { "epoch": 22.90251916757941, "grad_norm": 0.07596134394407272, "learning_rate": 8.59105899596922e-06, "loss": 0.0086, "step": 62950 }, { "epoch": 22.920774005111355, "grad_norm": 0.04925605282187462, "learning_rate": 8.589226823012094e-06, "loss": 0.0087, "step": 63000 }, { "epoch": 22.9390288426433, "grad_norm": 0.08094919472932816, "learning_rate": 8.587394650054965e-06, "loss": 0.0085, "step": 63050 }, { "epoch": 22.957283680175248, "grad_norm": 0.08631262183189392, "learning_rate": 8.585562477097839e-06, "loss": 0.0089, "step": 63100 }, { "epoch": 22.975538517707193, "grad_norm": 0.07808315753936768, "learning_rate": 8.583730304140711e-06, "loss": 0.0087, "step": 63150 }, { "epoch": 22.993793355239138, "grad_norm": 0.05618629232048988, "learning_rate": 8.581898131183585e-06, "loss": 0.0089, "step": 63200 }, { "epoch": 23.012048192771083, "grad_norm": 0.07608875632286072, "learning_rate": 8.580065958226457e-06, "loss": 0.0081, "step": 63250 }, { "epoch": 23.03030303030303, "grad_norm": 0.08526889979839325, "learning_rate": 8.57823378526933e-06, "loss": 0.0085, "step": 63300 }, { "epoch": 23.048557867834976, "grad_norm": 0.06414763629436493, "learning_rate": 8.576401612312204e-06, "loss": 0.0084, "step": 63350 }, { "epoch": 23.06681270536692, "grad_norm": 0.06458277255296707, "learning_rate": 8.574569439355076e-06, "loss": 0.0087, "step": 63400 }, { "epoch": 23.08506754289887, "grad_norm": 0.15176260471343994, "learning_rate": 8.572737266397948e-06, "loss": 0.0084, "step": 63450 }, { "epoch": 23.103322380430814, "grad_norm": 0.05909590795636177, "learning_rate": 8.57090509344082e-06, "loss": 0.0082, "step": 63500 }, { "epoch": 23.12157721796276, "grad_norm": 0.07062385976314545, "learning_rate": 8.569072920483695e-06, "loss": 0.0084, "step": 63550 }, { "epoch": 23.139832055494708, "grad_norm": 0.05863451212644577, "learning_rate": 8.567240747526567e-06, "loss": 0.0086, "step": 63600 }, { "epoch": 23.158086893026653, "grad_norm": 0.059923697263002396, "learning_rate": 8.565408574569441e-06, "loss": 0.0086, "step": 63650 }, { "epoch": 23.176341730558597, "grad_norm": 0.06300569325685501, "learning_rate": 8.563576401612313e-06, "loss": 0.0083, "step": 63700 }, { "epoch": 23.194596568090542, "grad_norm": 0.06146559491753578, "learning_rate": 8.561744228655186e-06, "loss": 0.0081, "step": 63750 }, { "epoch": 23.21285140562249, "grad_norm": 0.058338746428489685, "learning_rate": 8.559912055698058e-06, "loss": 0.0087, "step": 63800 }, { "epoch": 23.231106243154436, "grad_norm": 0.06301118433475494, "learning_rate": 8.558079882740932e-06, "loss": 0.0087, "step": 63850 }, { "epoch": 23.24936108068638, "grad_norm": 0.05700257793068886, "learning_rate": 8.556247709783804e-06, "loss": 0.0083, "step": 63900 }, { "epoch": 23.26761591821833, "grad_norm": 0.05440147593617439, "learning_rate": 8.554415536826676e-06, "loss": 0.0081, "step": 63950 }, { "epoch": 23.285870755750274, "grad_norm": 0.05918889865279198, "learning_rate": 8.55258336386955e-06, "loss": 0.0083, "step": 64000 }, { "epoch": 23.30412559328222, "grad_norm": 0.06032286956906319, "learning_rate": 8.550751190912423e-06, "loss": 0.0084, "step": 64050 }, { "epoch": 23.322380430814167, "grad_norm": 0.052777327597141266, "learning_rate": 8.548919017955297e-06, "loss": 0.0086, "step": 64100 }, { "epoch": 23.340635268346112, "grad_norm": 0.062323469668626785, "learning_rate": 8.547086844998167e-06, "loss": 0.008, "step": 64150 }, { "epoch": 23.358890105878057, "grad_norm": 0.06405339390039444, "learning_rate": 8.545254672041041e-06, "loss": 0.0086, "step": 64200 }, { "epoch": 23.377144943410002, "grad_norm": 0.0659857988357544, "learning_rate": 8.543422499083914e-06, "loss": 0.0082, "step": 64250 }, { "epoch": 23.39539978094195, "grad_norm": 0.08874934911727905, "learning_rate": 8.541590326126788e-06, "loss": 0.0082, "step": 64300 }, { "epoch": 23.413654618473895, "grad_norm": 0.05576704069972038, "learning_rate": 8.53975815316966e-06, "loss": 0.0085, "step": 64350 }, { "epoch": 23.43190945600584, "grad_norm": 0.07811632007360458, "learning_rate": 8.537925980212532e-06, "loss": 0.0088, "step": 64400 }, { "epoch": 23.45016429353779, "grad_norm": 0.05863044038414955, "learning_rate": 8.536093807255406e-06, "loss": 0.0083, "step": 64450 }, { "epoch": 23.468419131069734, "grad_norm": 0.056595828384160995, "learning_rate": 8.534261634298279e-06, "loss": 0.0082, "step": 64500 }, { "epoch": 23.48667396860168, "grad_norm": 0.07028651237487793, "learning_rate": 8.532429461341153e-06, "loss": 0.0081, "step": 64550 }, { "epoch": 23.504928806133627, "grad_norm": 0.08953356742858887, "learning_rate": 8.530597288384023e-06, "loss": 0.0087, "step": 64600 }, { "epoch": 23.523183643665572, "grad_norm": 0.07956796884536743, "learning_rate": 8.528765115426897e-06, "loss": 0.0085, "step": 64650 }, { "epoch": 23.541438481197517, "grad_norm": 0.07422539591789246, "learning_rate": 8.52693294246977e-06, "loss": 0.0086, "step": 64700 }, { "epoch": 23.55969331872946, "grad_norm": 0.060287415981292725, "learning_rate": 8.525100769512643e-06, "loss": 0.0084, "step": 64750 }, { "epoch": 23.57794815626141, "grad_norm": 0.07086237519979477, "learning_rate": 8.523268596555516e-06, "loss": 0.008, "step": 64800 }, { "epoch": 23.596202993793355, "grad_norm": 0.08128505945205688, "learning_rate": 8.521436423598388e-06, "loss": 0.0087, "step": 64850 }, { "epoch": 23.6144578313253, "grad_norm": 0.06881418824195862, "learning_rate": 8.519604250641262e-06, "loss": 0.0083, "step": 64900 }, { "epoch": 23.63271266885725, "grad_norm": 0.06325821578502655, "learning_rate": 8.517772077684134e-06, "loss": 0.0083, "step": 64950 }, { "epoch": 23.650967506389193, "grad_norm": 0.07646933943033218, "learning_rate": 8.515939904727007e-06, "loss": 0.0085, "step": 65000 }, { "epoch": 23.669222343921138, "grad_norm": 0.0678785890340805, "learning_rate": 8.514107731769879e-06, "loss": 0.0085, "step": 65050 }, { "epoch": 23.687477181453087, "grad_norm": 0.08116672933101654, "learning_rate": 8.512275558812753e-06, "loss": 0.0083, "step": 65100 }, { "epoch": 23.70573201898503, "grad_norm": 0.05456022545695305, "learning_rate": 8.510443385855625e-06, "loss": 0.0081, "step": 65150 }, { "epoch": 23.723986856516976, "grad_norm": 0.06983570009469986, "learning_rate": 8.5086112128985e-06, "loss": 0.0086, "step": 65200 }, { "epoch": 23.74224169404892, "grad_norm": 0.06777353584766388, "learning_rate": 8.506779039941372e-06, "loss": 0.0085, "step": 65250 }, { "epoch": 23.76049653158087, "grad_norm": 0.05610644444823265, "learning_rate": 8.504946866984244e-06, "loss": 0.0085, "step": 65300 }, { "epoch": 23.778751369112815, "grad_norm": 0.05237841606140137, "learning_rate": 8.503114694027116e-06, "loss": 0.0085, "step": 65350 }, { "epoch": 23.79700620664476, "grad_norm": 0.07181138545274734, "learning_rate": 8.50128252106999e-06, "loss": 0.0085, "step": 65400 }, { "epoch": 23.815261044176708, "grad_norm": 0.08305675536394119, "learning_rate": 8.499450348112862e-06, "loss": 0.0084, "step": 65450 }, { "epoch": 23.833515881708653, "grad_norm": 0.06492849439382553, "learning_rate": 8.497618175155735e-06, "loss": 0.0081, "step": 65500 }, { "epoch": 23.851770719240598, "grad_norm": 0.07611887156963348, "learning_rate": 8.495786002198609e-06, "loss": 0.0078, "step": 65550 }, { "epoch": 23.870025556772546, "grad_norm": 0.05179259926080704, "learning_rate": 8.493953829241481e-06, "loss": 0.0083, "step": 65600 }, { "epoch": 23.88828039430449, "grad_norm": 0.06830105930566788, "learning_rate": 8.492121656284355e-06, "loss": 0.0084, "step": 65650 }, { "epoch": 23.906535231836436, "grad_norm": 0.04747121408581734, "learning_rate": 8.490289483327226e-06, "loss": 0.0081, "step": 65700 }, { "epoch": 23.92479006936838, "grad_norm": 0.05353684350848198, "learning_rate": 8.4884573103701e-06, "loss": 0.008, "step": 65750 }, { "epoch": 23.94304490690033, "grad_norm": 0.06180991232395172, "learning_rate": 8.486625137412972e-06, "loss": 0.0087, "step": 65800 }, { "epoch": 23.961299744432274, "grad_norm": 0.057827144861221313, "learning_rate": 8.484792964455846e-06, "loss": 0.0088, "step": 65850 }, { "epoch": 23.97955458196422, "grad_norm": 0.06048797070980072, "learning_rate": 8.482960791498718e-06, "loss": 0.0087, "step": 65900 }, { "epoch": 23.997809419496168, "grad_norm": 0.062790647149086, "learning_rate": 8.48112861854159e-06, "loss": 0.0088, "step": 65950 }, { "epoch": 24.016064257028113, "grad_norm": 0.06070517748594284, "learning_rate": 8.479296445584465e-06, "loss": 0.0082, "step": 66000 }, { "epoch": 24.034319094560058, "grad_norm": 0.046685487031936646, "learning_rate": 8.477464272627337e-06, "loss": 0.0082, "step": 66050 }, { "epoch": 24.052573932092006, "grad_norm": 0.06198562681674957, "learning_rate": 8.47563209967021e-06, "loss": 0.0083, "step": 66100 }, { "epoch": 24.07082876962395, "grad_norm": 0.16641733050346375, "learning_rate": 8.473799926713081e-06, "loss": 0.0081, "step": 66150 }, { "epoch": 24.089083607155896, "grad_norm": 0.08327002078294754, "learning_rate": 8.471967753755955e-06, "loss": 0.0085, "step": 66200 }, { "epoch": 24.10733844468784, "grad_norm": 0.07346319407224655, "learning_rate": 8.470135580798828e-06, "loss": 0.0081, "step": 66250 }, { "epoch": 24.12559328221979, "grad_norm": 0.04904026538133621, "learning_rate": 8.468303407841702e-06, "loss": 0.0082, "step": 66300 }, { "epoch": 24.143848119751734, "grad_norm": 0.045337796211242676, "learning_rate": 8.466471234884574e-06, "loss": 0.0083, "step": 66350 }, { "epoch": 24.16210295728368, "grad_norm": 0.07809699326753616, "learning_rate": 8.464639061927446e-06, "loss": 0.0079, "step": 66400 }, { "epoch": 24.180357794815627, "grad_norm": 0.05970840901136398, "learning_rate": 8.46280688897032e-06, "loss": 0.0082, "step": 66450 }, { "epoch": 24.198612632347572, "grad_norm": 0.054344937205314636, "learning_rate": 8.460974716013193e-06, "loss": 0.0085, "step": 66500 }, { "epoch": 24.216867469879517, "grad_norm": 0.05814411863684654, "learning_rate": 8.459142543056065e-06, "loss": 0.0083, "step": 66550 }, { "epoch": 24.235122307411466, "grad_norm": 0.056382980197668076, "learning_rate": 8.457310370098937e-06, "loss": 0.0081, "step": 66600 }, { "epoch": 24.25337714494341, "grad_norm": 0.06226412206888199, "learning_rate": 8.455478197141811e-06, "loss": 0.0081, "step": 66650 }, { "epoch": 24.271631982475355, "grad_norm": 0.0644901692867279, "learning_rate": 8.453646024184684e-06, "loss": 0.0083, "step": 66700 }, { "epoch": 24.2898868200073, "grad_norm": 0.060136761516332626, "learning_rate": 8.451813851227556e-06, "loss": 0.0083, "step": 66750 }, { "epoch": 24.30814165753925, "grad_norm": 0.06622643768787384, "learning_rate": 8.44998167827043e-06, "loss": 0.0081, "step": 66800 }, { "epoch": 24.326396495071194, "grad_norm": 0.07891649752855301, "learning_rate": 8.448149505313302e-06, "loss": 0.0082, "step": 66850 }, { "epoch": 24.34465133260314, "grad_norm": 0.06354216486215591, "learning_rate": 8.446317332356175e-06, "loss": 0.0082, "step": 66900 }, { "epoch": 24.362906170135087, "grad_norm": 0.05686216801404953, "learning_rate": 8.444485159399048e-06, "loss": 0.0082, "step": 66950 }, { "epoch": 24.381161007667032, "grad_norm": 0.058685146272182465, "learning_rate": 8.44265298644192e-06, "loss": 0.008, "step": 67000 }, { "epoch": 24.399415845198977, "grad_norm": 0.09632483869791031, "learning_rate": 8.440820813484793e-06, "loss": 0.0082, "step": 67050 }, { "epoch": 24.417670682730925, "grad_norm": 0.05267462879419327, "learning_rate": 8.438988640527667e-06, "loss": 0.0082, "step": 67100 }, { "epoch": 24.43592552026287, "grad_norm": 0.06621171534061432, "learning_rate": 8.43715646757054e-06, "loss": 0.0083, "step": 67150 }, { "epoch": 24.454180357794815, "grad_norm": 0.07872860133647919, "learning_rate": 8.435324294613412e-06, "loss": 0.0081, "step": 67200 }, { "epoch": 24.47243519532676, "grad_norm": 0.05196872353553772, "learning_rate": 8.433492121656284e-06, "loss": 0.0084, "step": 67250 }, { "epoch": 24.49069003285871, "grad_norm": 0.0724998190999031, "learning_rate": 8.431659948699158e-06, "loss": 0.0081, "step": 67300 }, { "epoch": 24.508944870390653, "grad_norm": 0.07765519618988037, "learning_rate": 8.42982777574203e-06, "loss": 0.0085, "step": 67350 }, { "epoch": 24.5271997079226, "grad_norm": 0.06245283782482147, "learning_rate": 8.427995602784903e-06, "loss": 0.008, "step": 67400 }, { "epoch": 24.545454545454547, "grad_norm": 0.05998021364212036, "learning_rate": 8.426163429827777e-06, "loss": 0.0085, "step": 67450 }, { "epoch": 24.56370938298649, "grad_norm": 0.06283596158027649, "learning_rate": 8.424331256870649e-06, "loss": 0.0082, "step": 67500 }, { "epoch": 24.581964220518437, "grad_norm": 0.07417424768209457, "learning_rate": 8.422499083913523e-06, "loss": 0.0078, "step": 67550 }, { "epoch": 24.600219058050385, "grad_norm": 0.06474895775318146, "learning_rate": 8.420666910956395e-06, "loss": 0.0086, "step": 67600 }, { "epoch": 24.61847389558233, "grad_norm": 0.07617438584566116, "learning_rate": 8.418834737999268e-06, "loss": 0.0079, "step": 67650 }, { "epoch": 24.636728733114275, "grad_norm": 0.09005995094776154, "learning_rate": 8.41700256504214e-06, "loss": 0.0083, "step": 67700 }, { "epoch": 24.65498357064622, "grad_norm": 0.06088002398610115, "learning_rate": 8.415170392085014e-06, "loss": 0.0083, "step": 67750 }, { "epoch": 24.673238408178168, "grad_norm": 0.06126929074525833, "learning_rate": 8.413338219127886e-06, "loss": 0.0086, "step": 67800 }, { "epoch": 24.691493245710113, "grad_norm": 0.07581337541341782, "learning_rate": 8.411506046170758e-06, "loss": 0.0081, "step": 67850 }, { "epoch": 24.709748083242058, "grad_norm": 0.062375716865062714, "learning_rate": 8.409673873213632e-06, "loss": 0.0082, "step": 67900 }, { "epoch": 24.728002920774006, "grad_norm": 0.07420395314693451, "learning_rate": 8.407841700256505e-06, "loss": 0.0086, "step": 67950 }, { "epoch": 24.74625775830595, "grad_norm": 0.05925365164875984, "learning_rate": 8.406009527299379e-06, "loss": 0.0079, "step": 68000 }, { "epoch": 24.764512595837896, "grad_norm": 0.06616424769163132, "learning_rate": 8.404177354342251e-06, "loss": 0.0084, "step": 68050 }, { "epoch": 24.782767433369845, "grad_norm": 0.058155301958322525, "learning_rate": 8.402345181385123e-06, "loss": 0.0082, "step": 68100 }, { "epoch": 24.80102227090179, "grad_norm": 0.06643568724393845, "learning_rate": 8.400513008427996e-06, "loss": 0.008, "step": 68150 }, { "epoch": 24.819277108433734, "grad_norm": 0.07352788001298904, "learning_rate": 8.39868083547087e-06, "loss": 0.0082, "step": 68200 }, { "epoch": 24.83753194596568, "grad_norm": 0.07120870798826218, "learning_rate": 8.396848662513742e-06, "loss": 0.0082, "step": 68250 }, { "epoch": 24.855786783497628, "grad_norm": 0.05157686024904251, "learning_rate": 8.395016489556614e-06, "loss": 0.0081, "step": 68300 }, { "epoch": 24.874041621029573, "grad_norm": 0.053273286670446396, "learning_rate": 8.393184316599488e-06, "loss": 0.0084, "step": 68350 }, { "epoch": 24.892296458561518, "grad_norm": 0.054891299456357956, "learning_rate": 8.39135214364236e-06, "loss": 0.0084, "step": 68400 }, { "epoch": 24.910551296093466, "grad_norm": 0.06483190506696701, "learning_rate": 8.389519970685235e-06, "loss": 0.0086, "step": 68450 }, { "epoch": 24.92880613362541, "grad_norm": 0.06376541405916214, "learning_rate": 8.387687797728105e-06, "loss": 0.008, "step": 68500 }, { "epoch": 24.947060971157356, "grad_norm": 0.06917565315961838, "learning_rate": 8.385855624770979e-06, "loss": 0.0079, "step": 68550 }, { "epoch": 24.965315808689304, "grad_norm": 0.045247890055179596, "learning_rate": 8.384023451813851e-06, "loss": 0.0083, "step": 68600 }, { "epoch": 24.98357064622125, "grad_norm": 0.05585048720240593, "learning_rate": 8.382191278856725e-06, "loss": 0.0083, "step": 68650 }, { "epoch": 25.001825483753194, "grad_norm": 0.0665009468793869, "learning_rate": 8.380359105899598e-06, "loss": 0.0085, "step": 68700 }, { "epoch": 25.02008032128514, "grad_norm": 0.055849842727184296, "learning_rate": 8.37852693294247e-06, "loss": 0.0079, "step": 68750 }, { "epoch": 25.038335158817087, "grad_norm": 0.07151801884174347, "learning_rate": 8.376694759985344e-06, "loss": 0.0079, "step": 68800 }, { "epoch": 25.056589996349032, "grad_norm": 0.08591381460428238, "learning_rate": 8.374862587028216e-06, "loss": 0.0082, "step": 68850 }, { "epoch": 25.074844833880977, "grad_norm": 0.06130188703536987, "learning_rate": 8.373030414071089e-06, "loss": 0.008, "step": 68900 }, { "epoch": 25.093099671412926, "grad_norm": 0.1278325319290161, "learning_rate": 8.371198241113961e-06, "loss": 0.0076, "step": 68950 }, { "epoch": 25.11135450894487, "grad_norm": 0.06064237654209137, "learning_rate": 8.369366068156835e-06, "loss": 0.0078, "step": 69000 }, { "epoch": 25.129609346476816, "grad_norm": 0.07630188763141632, "learning_rate": 8.367533895199707e-06, "loss": 0.0081, "step": 69050 }, { "epoch": 25.147864184008764, "grad_norm": 0.08553296327590942, "learning_rate": 8.365701722242581e-06, "loss": 0.0077, "step": 69100 }, { "epoch": 25.16611902154071, "grad_norm": 0.06878108531236649, "learning_rate": 8.363869549285452e-06, "loss": 0.0076, "step": 69150 }, { "epoch": 25.184373859072654, "grad_norm": 0.08523499965667725, "learning_rate": 8.362037376328326e-06, "loss": 0.0083, "step": 69200 }, { "epoch": 25.2026286966046, "grad_norm": 0.0612613670527935, "learning_rate": 8.360205203371198e-06, "loss": 0.008, "step": 69250 }, { "epoch": 25.220883534136547, "grad_norm": 0.05561048164963722, "learning_rate": 8.358373030414072e-06, "loss": 0.0079, "step": 69300 }, { "epoch": 25.239138371668492, "grad_norm": 0.06551104784011841, "learning_rate": 8.356540857456944e-06, "loss": 0.0082, "step": 69350 }, { "epoch": 25.257393209200437, "grad_norm": 0.07042861729860306, "learning_rate": 8.354708684499817e-06, "loss": 0.0083, "step": 69400 }, { "epoch": 25.275648046732385, "grad_norm": 0.0665983110666275, "learning_rate": 8.35287651154269e-06, "loss": 0.0083, "step": 69450 }, { "epoch": 25.29390288426433, "grad_norm": 0.06401227414608002, "learning_rate": 8.351044338585563e-06, "loss": 0.008, "step": 69500 }, { "epoch": 25.312157721796275, "grad_norm": 0.055021982640028, "learning_rate": 8.349212165628437e-06, "loss": 0.0079, "step": 69550 }, { "epoch": 25.330412559328224, "grad_norm": 0.06465829908847809, "learning_rate": 8.347379992671308e-06, "loss": 0.0083, "step": 69600 }, { "epoch": 25.34866739686017, "grad_norm": 0.05454101413488388, "learning_rate": 8.345547819714182e-06, "loss": 0.0077, "step": 69650 }, { "epoch": 25.366922234392113, "grad_norm": 0.061470936983823776, "learning_rate": 8.343715646757054e-06, "loss": 0.0079, "step": 69700 }, { "epoch": 25.38517707192406, "grad_norm": 0.07804445177316666, "learning_rate": 8.341883473799928e-06, "loss": 0.0079, "step": 69750 }, { "epoch": 25.403431909456007, "grad_norm": 0.06513529270887375, "learning_rate": 8.3400513008428e-06, "loss": 0.0081, "step": 69800 }, { "epoch": 25.42168674698795, "grad_norm": 0.06795467436313629, "learning_rate": 8.338219127885673e-06, "loss": 0.0081, "step": 69850 }, { "epoch": 25.439941584519897, "grad_norm": 0.06848642975091934, "learning_rate": 8.336386954928547e-06, "loss": 0.008, "step": 69900 }, { "epoch": 25.458196422051845, "grad_norm": 0.0702468678355217, "learning_rate": 8.334554781971419e-06, "loss": 0.0078, "step": 69950 }, { "epoch": 25.47645125958379, "grad_norm": 0.07004186511039734, "learning_rate": 8.332722609014293e-06, "loss": 0.0082, "step": 70000 }, { "epoch": 25.494706097115735, "grad_norm": 0.05611666291952133, "learning_rate": 8.330890436057163e-06, "loss": 0.0078, "step": 70050 }, { "epoch": 25.512960934647683, "grad_norm": 0.06685035675764084, "learning_rate": 8.329058263100037e-06, "loss": 0.0078, "step": 70100 }, { "epoch": 25.53121577217963, "grad_norm": 0.06934567540884018, "learning_rate": 8.32722609014291e-06, "loss": 0.0081, "step": 70150 }, { "epoch": 25.549470609711573, "grad_norm": 0.128873810172081, "learning_rate": 8.325393917185784e-06, "loss": 0.0081, "step": 70200 }, { "epoch": 25.567725447243518, "grad_norm": 0.05134338513016701, "learning_rate": 8.323561744228656e-06, "loss": 0.008, "step": 70250 }, { "epoch": 25.585980284775466, "grad_norm": 0.06192426383495331, "learning_rate": 8.321729571271528e-06, "loss": 0.0082, "step": 70300 }, { "epoch": 25.60423512230741, "grad_norm": 0.05665076896548271, "learning_rate": 8.319897398314402e-06, "loss": 0.0082, "step": 70350 }, { "epoch": 25.622489959839356, "grad_norm": 0.0713886246085167, "learning_rate": 8.318065225357275e-06, "loss": 0.0082, "step": 70400 }, { "epoch": 25.640744797371305, "grad_norm": 0.0546143501996994, "learning_rate": 8.316233052400147e-06, "loss": 0.008, "step": 70450 }, { "epoch": 25.65899963490325, "grad_norm": 0.09072823822498322, "learning_rate": 8.31440087944302e-06, "loss": 0.008, "step": 70500 }, { "epoch": 25.677254472435195, "grad_norm": 0.05573221668601036, "learning_rate": 8.312568706485893e-06, "loss": 0.0084, "step": 70550 }, { "epoch": 25.695509309967143, "grad_norm": 0.07449289411306381, "learning_rate": 8.310736533528766e-06, "loss": 0.0081, "step": 70600 }, { "epoch": 25.713764147499088, "grad_norm": 0.057186707854270935, "learning_rate": 8.30890436057164e-06, "loss": 0.0079, "step": 70650 }, { "epoch": 25.732018985031033, "grad_norm": 0.05291341617703438, "learning_rate": 8.307072187614512e-06, "loss": 0.0079, "step": 70700 }, { "epoch": 25.750273822562978, "grad_norm": 0.05352197587490082, "learning_rate": 8.305240014657384e-06, "loss": 0.0081, "step": 70750 }, { "epoch": 25.768528660094926, "grad_norm": 0.07027698308229446, "learning_rate": 8.303407841700256e-06, "loss": 0.0079, "step": 70800 }, { "epoch": 25.78678349762687, "grad_norm": 0.06517873704433441, "learning_rate": 8.30157566874313e-06, "loss": 0.0078, "step": 70850 }, { "epoch": 25.805038335158816, "grad_norm": 0.06365886330604553, "learning_rate": 8.299743495786003e-06, "loss": 0.0079, "step": 70900 }, { "epoch": 25.823293172690764, "grad_norm": 0.06816490739583969, "learning_rate": 8.297911322828875e-06, "loss": 0.0079, "step": 70950 }, { "epoch": 25.84154801022271, "grad_norm": 0.08994780480861664, "learning_rate": 8.296079149871749e-06, "loss": 0.0082, "step": 71000 }, { "epoch": 25.859802847754654, "grad_norm": 0.06243437901139259, "learning_rate": 8.294246976914621e-06, "loss": 0.0083, "step": 71050 }, { "epoch": 25.878057685286603, "grad_norm": 0.05098967254161835, "learning_rate": 8.292414803957495e-06, "loss": 0.0082, "step": 71100 }, { "epoch": 25.896312522818548, "grad_norm": 0.06109187379479408, "learning_rate": 8.290582631000366e-06, "loss": 0.008, "step": 71150 }, { "epoch": 25.914567360350492, "grad_norm": 0.0692925900220871, "learning_rate": 8.28875045804324e-06, "loss": 0.0079, "step": 71200 }, { "epoch": 25.932822197882437, "grad_norm": 0.07059754431247711, "learning_rate": 8.286918285086112e-06, "loss": 0.0078, "step": 71250 }, { "epoch": 25.951077035414386, "grad_norm": 0.06998153775930405, "learning_rate": 8.285086112128986e-06, "loss": 0.0081, "step": 71300 }, { "epoch": 25.96933187294633, "grad_norm": 0.06631340086460114, "learning_rate": 8.283253939171859e-06, "loss": 0.008, "step": 71350 }, { "epoch": 25.987586710478276, "grad_norm": 0.056150730699300766, "learning_rate": 8.281421766214731e-06, "loss": 0.0082, "step": 71400 }, { "epoch": 26.005841548010224, "grad_norm": 0.050910934805870056, "learning_rate": 8.279589593257605e-06, "loss": 0.008, "step": 71450 }, { "epoch": 26.02409638554217, "grad_norm": 0.05750512704253197, "learning_rate": 8.277757420300477e-06, "loss": 0.0082, "step": 71500 }, { "epoch": 26.042351223074114, "grad_norm": 0.05809541419148445, "learning_rate": 8.27592524734335e-06, "loss": 0.008, "step": 71550 }, { "epoch": 26.060606060606062, "grad_norm": 0.05806922912597656, "learning_rate": 8.274093074386222e-06, "loss": 0.0076, "step": 71600 }, { "epoch": 26.078860898138007, "grad_norm": 0.06896824389696121, "learning_rate": 8.272260901429096e-06, "loss": 0.0077, "step": 71650 }, { "epoch": 26.097115735669952, "grad_norm": 0.05349760130047798, "learning_rate": 8.270428728471968e-06, "loss": 0.0077, "step": 71700 }, { "epoch": 26.115370573201897, "grad_norm": 0.10557685047388077, "learning_rate": 8.268596555514842e-06, "loss": 0.0077, "step": 71750 }, { "epoch": 26.133625410733845, "grad_norm": 0.04997897148132324, "learning_rate": 8.266764382557714e-06, "loss": 0.008, "step": 71800 }, { "epoch": 26.15188024826579, "grad_norm": 0.09700646251440048, "learning_rate": 8.264932209600587e-06, "loss": 0.0077, "step": 71850 }, { "epoch": 26.170135085797735, "grad_norm": 0.08959037065505981, "learning_rate": 8.26310003664346e-06, "loss": 0.0075, "step": 71900 }, { "epoch": 26.188389923329684, "grad_norm": 0.05606921389698982, "learning_rate": 8.261267863686333e-06, "loss": 0.008, "step": 71950 }, { "epoch": 26.20664476086163, "grad_norm": 0.10053978115320206, "learning_rate": 8.259435690729205e-06, "loss": 0.0077, "step": 72000 }, { "epoch": 26.224899598393574, "grad_norm": 0.05436232313513756, "learning_rate": 8.257603517772078e-06, "loss": 0.0079, "step": 72050 }, { "epoch": 26.243154435925522, "grad_norm": 0.06855732202529907, "learning_rate": 8.255771344814952e-06, "loss": 0.008, "step": 72100 }, { "epoch": 26.261409273457467, "grad_norm": 0.061214715242385864, "learning_rate": 8.253939171857824e-06, "loss": 0.0078, "step": 72150 }, { "epoch": 26.279664110989412, "grad_norm": 0.07048065215349197, "learning_rate": 8.252106998900698e-06, "loss": 0.0074, "step": 72200 }, { "epoch": 26.297918948521357, "grad_norm": 0.06310497969388962, "learning_rate": 8.25027482594357e-06, "loss": 0.0078, "step": 72250 }, { "epoch": 26.316173786053305, "grad_norm": 0.09356817603111267, "learning_rate": 8.248442652986443e-06, "loss": 0.0078, "step": 72300 }, { "epoch": 26.33442862358525, "grad_norm": 0.0704420953989029, "learning_rate": 8.246610480029315e-06, "loss": 0.0079, "step": 72350 }, { "epoch": 26.352683461117195, "grad_norm": 0.06630934774875641, "learning_rate": 8.244778307072189e-06, "loss": 0.0078, "step": 72400 }, { "epoch": 26.370938298649143, "grad_norm": 0.08426592499017715, "learning_rate": 8.242946134115061e-06, "loss": 0.0077, "step": 72450 }, { "epoch": 26.38919313618109, "grad_norm": 0.05927763879299164, "learning_rate": 8.241113961157933e-06, "loss": 0.0079, "step": 72500 }, { "epoch": 26.407447973713033, "grad_norm": 0.10463499277830124, "learning_rate": 8.239281788200807e-06, "loss": 0.0079, "step": 72550 }, { "epoch": 26.42570281124498, "grad_norm": 0.08069847524166107, "learning_rate": 8.23744961524368e-06, "loss": 0.0077, "step": 72600 }, { "epoch": 26.443957648776927, "grad_norm": 0.06566139310598373, "learning_rate": 8.235617442286552e-06, "loss": 0.0076, "step": 72650 }, { "epoch": 26.46221248630887, "grad_norm": 0.057224418967962265, "learning_rate": 8.233785269329424e-06, "loss": 0.0076, "step": 72700 }, { "epoch": 26.480467323840816, "grad_norm": 0.09072554856538773, "learning_rate": 8.231953096372298e-06, "loss": 0.0078, "step": 72750 }, { "epoch": 26.498722161372765, "grad_norm": 0.056058816611766815, "learning_rate": 8.23012092341517e-06, "loss": 0.0078, "step": 72800 }, { "epoch": 26.51697699890471, "grad_norm": 0.23570749163627625, "learning_rate": 8.228288750458045e-06, "loss": 0.008, "step": 72850 }, { "epoch": 26.535231836436655, "grad_norm": 0.056156229227781296, "learning_rate": 8.226456577500917e-06, "loss": 0.0079, "step": 72900 }, { "epoch": 26.553486673968603, "grad_norm": 0.1209225207567215, "learning_rate": 8.22462440454379e-06, "loss": 0.0082, "step": 72950 }, { "epoch": 26.571741511500548, "grad_norm": 0.07513442635536194, "learning_rate": 8.222792231586663e-06, "loss": 0.0075, "step": 73000 }, { "epoch": 26.589996349032493, "grad_norm": 0.06366512924432755, "learning_rate": 8.220960058629536e-06, "loss": 0.0083, "step": 73050 }, { "epoch": 26.608251186564438, "grad_norm": 0.08948852121829987, "learning_rate": 8.219127885672408e-06, "loss": 0.0078, "step": 73100 }, { "epoch": 26.626506024096386, "grad_norm": 0.05680718272924423, "learning_rate": 8.21729571271528e-06, "loss": 0.0075, "step": 73150 }, { "epoch": 26.64476086162833, "grad_norm": 0.05975888669490814, "learning_rate": 8.215463539758154e-06, "loss": 0.0077, "step": 73200 }, { "epoch": 26.663015699160276, "grad_norm": 0.06697451323270798, "learning_rate": 8.213631366801026e-06, "loss": 0.008, "step": 73250 }, { "epoch": 26.681270536692224, "grad_norm": 0.08459766954183578, "learning_rate": 8.211799193843899e-06, "loss": 0.0077, "step": 73300 }, { "epoch": 26.69952537422417, "grad_norm": 0.06454908102750778, "learning_rate": 8.209967020886773e-06, "loss": 0.0082, "step": 73350 }, { "epoch": 26.717780211756114, "grad_norm": 0.061517130583524704, "learning_rate": 8.208134847929645e-06, "loss": 0.0076, "step": 73400 }, { "epoch": 26.736035049288063, "grad_norm": 0.0776604562997818, "learning_rate": 8.206302674972519e-06, "loss": 0.0077, "step": 73450 }, { "epoch": 26.754289886820008, "grad_norm": 0.10406628251075745, "learning_rate": 8.204470502015391e-06, "loss": 0.0074, "step": 73500 }, { "epoch": 26.772544724351953, "grad_norm": 0.08300726860761642, "learning_rate": 8.202638329058264e-06, "loss": 0.0076, "step": 73550 }, { "epoch": 26.7907995618839, "grad_norm": 0.0867338627576828, "learning_rate": 8.200806156101136e-06, "loss": 0.0082, "step": 73600 }, { "epoch": 26.809054399415846, "grad_norm": 0.12747009098529816, "learning_rate": 8.19897398314401e-06, "loss": 0.0077, "step": 73650 }, { "epoch": 26.82730923694779, "grad_norm": 0.07793665677309036, "learning_rate": 8.197141810186882e-06, "loss": 0.0074, "step": 73700 }, { "epoch": 26.845564074479736, "grad_norm": 0.07117403298616409, "learning_rate": 8.195309637229755e-06, "loss": 0.008, "step": 73750 }, { "epoch": 26.863818912011684, "grad_norm": 0.05632568895816803, "learning_rate": 8.193477464272629e-06, "loss": 0.0075, "step": 73800 }, { "epoch": 26.88207374954363, "grad_norm": 0.09333931654691696, "learning_rate": 8.1916452913155e-06, "loss": 0.0078, "step": 73850 }, { "epoch": 26.900328587075574, "grad_norm": 0.06050976738333702, "learning_rate": 8.189813118358375e-06, "loss": 0.008, "step": 73900 }, { "epoch": 26.918583424607522, "grad_norm": 0.06008275970816612, "learning_rate": 8.187980945401245e-06, "loss": 0.0082, "step": 73950 }, { "epoch": 26.936838262139467, "grad_norm": 0.05773244798183441, "learning_rate": 8.18614877244412e-06, "loss": 0.0083, "step": 74000 }, { "epoch": 26.955093099671412, "grad_norm": 0.05469867214560509, "learning_rate": 8.184316599486992e-06, "loss": 0.0076, "step": 74050 }, { "epoch": 26.973347937203357, "grad_norm": 0.06356282532215118, "learning_rate": 8.182484426529866e-06, "loss": 0.0076, "step": 74100 }, { "epoch": 26.991602774735306, "grad_norm": 0.07441577315330505, "learning_rate": 8.180652253572738e-06, "loss": 0.0082, "step": 74150 }, { "epoch": 27.00985761226725, "grad_norm": 0.06272673606872559, "learning_rate": 8.17882008061561e-06, "loss": 0.0081, "step": 74200 }, { "epoch": 27.028112449799195, "grad_norm": 0.05756441503763199, "learning_rate": 8.176987907658483e-06, "loss": 0.0079, "step": 74250 }, { "epoch": 27.046367287331144, "grad_norm": 0.09137877821922302, "learning_rate": 8.175155734701357e-06, "loss": 0.0077, "step": 74300 }, { "epoch": 27.06462212486309, "grad_norm": 0.07936594635248184, "learning_rate": 8.173323561744229e-06, "loss": 0.0075, "step": 74350 }, { "epoch": 27.082876962395034, "grad_norm": 0.07034654915332794, "learning_rate": 8.171491388787101e-06, "loss": 0.0082, "step": 74400 }, { "epoch": 27.101131799926982, "grad_norm": 0.07509841024875641, "learning_rate": 8.169659215829975e-06, "loss": 0.0074, "step": 74450 }, { "epoch": 27.119386637458927, "grad_norm": 0.05667085573077202, "learning_rate": 8.167827042872848e-06, "loss": 0.0078, "step": 74500 }, { "epoch": 27.137641474990872, "grad_norm": 0.06890314817428589, "learning_rate": 8.165994869915722e-06, "loss": 0.0079, "step": 74550 }, { "epoch": 27.15589631252282, "grad_norm": 0.06592974811792374, "learning_rate": 8.164162696958594e-06, "loss": 0.0081, "step": 74600 }, { "epoch": 27.174151150054765, "grad_norm": 0.05735612288117409, "learning_rate": 8.162330524001466e-06, "loss": 0.0077, "step": 74650 }, { "epoch": 27.19240598758671, "grad_norm": 0.07535567134618759, "learning_rate": 8.160498351044338e-06, "loss": 0.0075, "step": 74700 }, { "epoch": 27.210660825118655, "grad_norm": 0.0629306361079216, "learning_rate": 8.158666178087212e-06, "loss": 0.0078, "step": 74750 }, { "epoch": 27.228915662650603, "grad_norm": 0.08967940509319305, "learning_rate": 8.156834005130085e-06, "loss": 0.0076, "step": 74800 }, { "epoch": 27.24717050018255, "grad_norm": 0.08157164603471756, "learning_rate": 8.155001832172957e-06, "loss": 0.0077, "step": 74850 }, { "epoch": 27.265425337714493, "grad_norm": 0.09233804047107697, "learning_rate": 8.153169659215831e-06, "loss": 0.0075, "step": 74900 }, { "epoch": 27.28368017524644, "grad_norm": 0.05763858184218407, "learning_rate": 8.151337486258703e-06, "loss": 0.0075, "step": 74950 }, { "epoch": 27.301935012778387, "grad_norm": 0.09383340179920197, "learning_rate": 8.149505313301577e-06, "loss": 0.0075, "step": 75000 }, { "epoch": 27.32018985031033, "grad_norm": 0.06458237022161484, "learning_rate": 8.147673140344448e-06, "loss": 0.0074, "step": 75050 }, { "epoch": 27.338444687842276, "grad_norm": 0.05898362770676613, "learning_rate": 8.145840967387322e-06, "loss": 0.0082, "step": 75100 }, { "epoch": 27.356699525374225, "grad_norm": 0.06436064839363098, "learning_rate": 8.144008794430194e-06, "loss": 0.0078, "step": 75150 }, { "epoch": 27.37495436290617, "grad_norm": 0.07442861795425415, "learning_rate": 8.142176621473068e-06, "loss": 0.0078, "step": 75200 }, { "epoch": 27.393209200438115, "grad_norm": 0.0688323974609375, "learning_rate": 8.14034444851594e-06, "loss": 0.0076, "step": 75250 }, { "epoch": 27.411464037970063, "grad_norm": 0.08945580571889877, "learning_rate": 8.138512275558813e-06, "loss": 0.0078, "step": 75300 }, { "epoch": 27.429718875502008, "grad_norm": 0.05201173573732376, "learning_rate": 8.136680102601687e-06, "loss": 0.0077, "step": 75350 }, { "epoch": 27.447973713033953, "grad_norm": 0.06862752139568329, "learning_rate": 8.13484792964456e-06, "loss": 0.0076, "step": 75400 }, { "epoch": 27.4662285505659, "grad_norm": 0.08229533582925797, "learning_rate": 8.133015756687433e-06, "loss": 0.0074, "step": 75450 }, { "epoch": 27.484483388097846, "grad_norm": 0.06273151934146881, "learning_rate": 8.131183583730304e-06, "loss": 0.0076, "step": 75500 }, { "epoch": 27.50273822562979, "grad_norm": 0.06511589139699936, "learning_rate": 8.129351410773178e-06, "loss": 0.0076, "step": 75550 }, { "epoch": 27.52099306316174, "grad_norm": 0.06538913398981094, "learning_rate": 8.12751923781605e-06, "loss": 0.0073, "step": 75600 }, { "epoch": 27.539247900693685, "grad_norm": 0.051664456725120544, "learning_rate": 8.125687064858924e-06, "loss": 0.0077, "step": 75650 }, { "epoch": 27.55750273822563, "grad_norm": 0.06950704008340836, "learning_rate": 8.123854891901796e-06, "loss": 0.0075, "step": 75700 }, { "epoch": 27.575757575757574, "grad_norm": 0.06633534282445908, "learning_rate": 8.122022718944669e-06, "loss": 0.0074, "step": 75750 }, { "epoch": 27.594012413289523, "grad_norm": 0.06463894993066788, "learning_rate": 8.120190545987543e-06, "loss": 0.0076, "step": 75800 }, { "epoch": 27.612267250821468, "grad_norm": 0.0565531961619854, "learning_rate": 8.118358373030415e-06, "loss": 0.0076, "step": 75850 }, { "epoch": 27.630522088353413, "grad_norm": 0.05665811896324158, "learning_rate": 8.116526200073287e-06, "loss": 0.0071, "step": 75900 }, { "epoch": 27.64877692588536, "grad_norm": 0.055220719426870346, "learning_rate": 8.11469402711616e-06, "loss": 0.0078, "step": 75950 }, { "epoch": 27.667031763417306, "grad_norm": 0.0556454174220562, "learning_rate": 8.112861854159034e-06, "loss": 0.0074, "step": 76000 }, { "epoch": 27.68528660094925, "grad_norm": 0.0677880197763443, "learning_rate": 8.111029681201906e-06, "loss": 0.0076, "step": 76050 }, { "epoch": 27.703541438481196, "grad_norm": 0.07601721584796906, "learning_rate": 8.10919750824478e-06, "loss": 0.0077, "step": 76100 }, { "epoch": 27.721796276013144, "grad_norm": 0.07077684253454208, "learning_rate": 8.107365335287652e-06, "loss": 0.0074, "step": 76150 }, { "epoch": 27.74005111354509, "grad_norm": 0.06813836097717285, "learning_rate": 8.105533162330524e-06, "loss": 0.0074, "step": 76200 }, { "epoch": 27.758305951077034, "grad_norm": 0.06732522696256638, "learning_rate": 8.103700989373397e-06, "loss": 0.0079, "step": 76250 }, { "epoch": 27.776560788608982, "grad_norm": 0.07012120634317398, "learning_rate": 8.10186881641627e-06, "loss": 0.0077, "step": 76300 }, { "epoch": 27.794815626140927, "grad_norm": 0.09761738032102585, "learning_rate": 8.100036643459143e-06, "loss": 0.0073, "step": 76350 }, { "epoch": 27.813070463672872, "grad_norm": 0.07090756297111511, "learning_rate": 8.098204470502015e-06, "loss": 0.0079, "step": 76400 }, { "epoch": 27.83132530120482, "grad_norm": 0.0863017737865448, "learning_rate": 8.09637229754489e-06, "loss": 0.0074, "step": 76450 }, { "epoch": 27.849580138736766, "grad_norm": 0.05838792398571968, "learning_rate": 8.094540124587762e-06, "loss": 0.0081, "step": 76500 }, { "epoch": 27.86783497626871, "grad_norm": 0.09894339740276337, "learning_rate": 8.092707951630636e-06, "loss": 0.0081, "step": 76550 }, { "epoch": 27.88608981380066, "grad_norm": 0.05869816988706589, "learning_rate": 8.090875778673506e-06, "loss": 0.0075, "step": 76600 }, { "epoch": 27.904344651332604, "grad_norm": 0.1206921711564064, "learning_rate": 8.08904360571638e-06, "loss": 0.0077, "step": 76650 }, { "epoch": 27.92259948886455, "grad_norm": 0.06278300285339355, "learning_rate": 8.087211432759253e-06, "loss": 0.0078, "step": 76700 }, { "epoch": 27.940854326396494, "grad_norm": 0.07514593005180359, "learning_rate": 8.085379259802127e-06, "loss": 0.007, "step": 76750 }, { "epoch": 27.959109163928442, "grad_norm": 0.05882091820240021, "learning_rate": 8.083547086844999e-06, "loss": 0.0079, "step": 76800 }, { "epoch": 27.977364001460387, "grad_norm": 0.06605298072099686, "learning_rate": 8.081714913887871e-06, "loss": 0.0072, "step": 76850 }, { "epoch": 27.995618838992332, "grad_norm": 0.05800521373748779, "learning_rate": 8.079882740930745e-06, "loss": 0.0074, "step": 76900 }, { "epoch": 28.01387367652428, "grad_norm": 0.051720041781663895, "learning_rate": 8.078050567973618e-06, "loss": 0.0073, "step": 76950 }, { "epoch": 28.032128514056225, "grad_norm": 0.06925223767757416, "learning_rate": 8.076218395016492e-06, "loss": 0.0076, "step": 77000 }, { "epoch": 28.05038335158817, "grad_norm": 0.05824137106537819, "learning_rate": 8.074386222059362e-06, "loss": 0.0077, "step": 77050 }, { "epoch": 28.068638189120115, "grad_norm": 0.07645922899246216, "learning_rate": 8.072554049102236e-06, "loss": 0.0075, "step": 77100 }, { "epoch": 28.086893026652064, "grad_norm": 0.06242457032203674, "learning_rate": 8.070721876145108e-06, "loss": 0.0074, "step": 77150 }, { "epoch": 28.10514786418401, "grad_norm": 0.09106332808732986, "learning_rate": 8.068889703187982e-06, "loss": 0.0076, "step": 77200 }, { "epoch": 28.123402701715953, "grad_norm": 0.073928602039814, "learning_rate": 8.067057530230855e-06, "loss": 0.0075, "step": 77250 }, { "epoch": 28.141657539247902, "grad_norm": 0.07682711631059647, "learning_rate": 8.065225357273727e-06, "loss": 0.0074, "step": 77300 }, { "epoch": 28.159912376779847, "grad_norm": 0.05180603265762329, "learning_rate": 8.063393184316601e-06, "loss": 0.0073, "step": 77350 }, { "epoch": 28.17816721431179, "grad_norm": 0.06351172178983688, "learning_rate": 8.061561011359473e-06, "loss": 0.0072, "step": 77400 }, { "epoch": 28.19642205184374, "grad_norm": 0.06141902878880501, "learning_rate": 8.059728838402346e-06, "loss": 0.0074, "step": 77450 }, { "epoch": 28.214676889375685, "grad_norm": 0.07650449126958847, "learning_rate": 8.057896665445218e-06, "loss": 0.0078, "step": 77500 }, { "epoch": 28.23293172690763, "grad_norm": 0.08064411580562592, "learning_rate": 8.056064492488092e-06, "loss": 0.0075, "step": 77550 }, { "epoch": 28.251186564439575, "grad_norm": 0.053219303488731384, "learning_rate": 8.054232319530964e-06, "loss": 0.0074, "step": 77600 }, { "epoch": 28.269441401971523, "grad_norm": 0.08709681779146194, "learning_rate": 8.052400146573838e-06, "loss": 0.0074, "step": 77650 }, { "epoch": 28.287696239503468, "grad_norm": 0.0644121915102005, "learning_rate": 8.05056797361671e-06, "loss": 0.0071, "step": 77700 }, { "epoch": 28.305951077035413, "grad_norm": 0.07835350930690765, "learning_rate": 8.048735800659583e-06, "loss": 0.0074, "step": 77750 }, { "epoch": 28.32420591456736, "grad_norm": 0.06490179896354675, "learning_rate": 8.046903627702455e-06, "loss": 0.0079, "step": 77800 }, { "epoch": 28.342460752099306, "grad_norm": 0.06255075335502625, "learning_rate": 8.045071454745329e-06, "loss": 0.0072, "step": 77850 }, { "epoch": 28.36071558963125, "grad_norm": 0.05514924228191376, "learning_rate": 8.043239281788201e-06, "loss": 0.0076, "step": 77900 }, { "epoch": 28.3789704271632, "grad_norm": 0.0911376103758812, "learning_rate": 8.041407108831074e-06, "loss": 0.0072, "step": 77950 }, { "epoch": 28.397225264695145, "grad_norm": 0.05838305130600929, "learning_rate": 8.039574935873948e-06, "loss": 0.0076, "step": 78000 }, { "epoch": 28.41548010222709, "grad_norm": 0.07033880054950714, "learning_rate": 8.03774276291682e-06, "loss": 0.0076, "step": 78050 }, { "epoch": 28.433734939759034, "grad_norm": 0.08116988837718964, "learning_rate": 8.035910589959692e-06, "loss": 0.0075, "step": 78100 }, { "epoch": 28.451989777290983, "grad_norm": 0.05829693749547005, "learning_rate": 8.034078417002565e-06, "loss": 0.0078, "step": 78150 }, { "epoch": 28.470244614822928, "grad_norm": 0.07058205455541611, "learning_rate": 8.032246244045439e-06, "loss": 0.0074, "step": 78200 }, { "epoch": 28.488499452354873, "grad_norm": 0.05450369045138359, "learning_rate": 8.030414071088311e-06, "loss": 0.0075, "step": 78250 }, { "epoch": 28.50675428988682, "grad_norm": 0.06194787099957466, "learning_rate": 8.028581898131185e-06, "loss": 0.0076, "step": 78300 }, { "epoch": 28.525009127418766, "grad_norm": 0.06342658400535583, "learning_rate": 8.026749725174057e-06, "loss": 0.0072, "step": 78350 }, { "epoch": 28.54326396495071, "grad_norm": 0.07188816368579865, "learning_rate": 8.02491755221693e-06, "loss": 0.0078, "step": 78400 }, { "epoch": 28.56151880248266, "grad_norm": 0.07719308137893677, "learning_rate": 8.023085379259804e-06, "loss": 0.0075, "step": 78450 }, { "epoch": 28.579773640014604, "grad_norm": 0.061608050018548965, "learning_rate": 8.021253206302676e-06, "loss": 0.0072, "step": 78500 }, { "epoch": 28.59802847754655, "grad_norm": 0.08127917349338531, "learning_rate": 8.019421033345548e-06, "loss": 0.0074, "step": 78550 }, { "epoch": 28.616283315078494, "grad_norm": 0.09360869973897934, "learning_rate": 8.01758886038842e-06, "loss": 0.0075, "step": 78600 }, { "epoch": 28.634538152610443, "grad_norm": 0.08539122343063354, "learning_rate": 8.015756687431294e-06, "loss": 0.0077, "step": 78650 }, { "epoch": 28.652792990142387, "grad_norm": 0.07136587053537369, "learning_rate": 8.013924514474167e-06, "loss": 0.0073, "step": 78700 }, { "epoch": 28.671047827674332, "grad_norm": 0.06686566770076752, "learning_rate": 8.01209234151704e-06, "loss": 0.0079, "step": 78750 }, { "epoch": 28.68930266520628, "grad_norm": 0.06917273998260498, "learning_rate": 8.010260168559913e-06, "loss": 0.0076, "step": 78800 }, { "epoch": 28.707557502738226, "grad_norm": 0.056814976036548615, "learning_rate": 8.008427995602785e-06, "loss": 0.0077, "step": 78850 }, { "epoch": 28.72581234027017, "grad_norm": 0.06712199002504349, "learning_rate": 8.00659582264566e-06, "loss": 0.0075, "step": 78900 }, { "epoch": 28.74406717780212, "grad_norm": 0.08391757309436798, "learning_rate": 8.004763649688532e-06, "loss": 0.0073, "step": 78950 }, { "epoch": 28.762322015334064, "grad_norm": 0.06136947497725487, "learning_rate": 8.002931476731404e-06, "loss": 0.0078, "step": 79000 }, { "epoch": 28.78057685286601, "grad_norm": 0.0736754760146141, "learning_rate": 8.001099303774276e-06, "loss": 0.0073, "step": 79050 }, { "epoch": 28.798831690397954, "grad_norm": 0.062265489250421524, "learning_rate": 7.99926713081715e-06, "loss": 0.0072, "step": 79100 }, { "epoch": 28.817086527929902, "grad_norm": 0.07650555670261383, "learning_rate": 7.997434957860023e-06, "loss": 0.0071, "step": 79150 }, { "epoch": 28.835341365461847, "grad_norm": 0.07041070610284805, "learning_rate": 7.995602784902895e-06, "loss": 0.0074, "step": 79200 }, { "epoch": 28.853596202993792, "grad_norm": 0.07315370440483093, "learning_rate": 7.993770611945769e-06, "loss": 0.0075, "step": 79250 }, { "epoch": 28.87185104052574, "grad_norm": 0.0719771757721901, "learning_rate": 7.991938438988641e-06, "loss": 0.0072, "step": 79300 }, { "epoch": 28.890105878057685, "grad_norm": 0.08390133827924728, "learning_rate": 7.990106266031513e-06, "loss": 0.0073, "step": 79350 }, { "epoch": 28.90836071558963, "grad_norm": 0.06938619166612625, "learning_rate": 7.988274093074387e-06, "loss": 0.0077, "step": 79400 }, { "epoch": 28.92661555312158, "grad_norm": 0.08508136868476868, "learning_rate": 7.98644192011726e-06, "loss": 0.0076, "step": 79450 }, { "epoch": 28.944870390653524, "grad_norm": 0.0766613781452179, "learning_rate": 7.984609747160132e-06, "loss": 0.0073, "step": 79500 }, { "epoch": 29.018254837531945, "grad_norm": 0.06182140111923218, "learning_rate": 7.982777574203006e-06, "loss": 0.0074, "step": 79550 }, { "epoch": 29.036509675063893, "grad_norm": 0.06543870270252228, "learning_rate": 7.980945401245878e-06, "loss": 0.0072, "step": 79600 }, { "epoch": 29.05476451259584, "grad_norm": 0.08262419700622559, "learning_rate": 7.97911322828875e-06, "loss": 0.0074, "step": 79650 }, { "epoch": 29.073019350127783, "grad_norm": 0.07292219996452332, "learning_rate": 7.977281055331623e-06, "loss": 0.0072, "step": 79700 }, { "epoch": 29.091274187659728, "grad_norm": 0.06897533684968948, "learning_rate": 7.975448882374497e-06, "loss": 0.0068, "step": 79750 }, { "epoch": 29.109529025191677, "grad_norm": 0.09884458780288696, "learning_rate": 7.97361670941737e-06, "loss": 0.0072, "step": 79800 }, { "epoch": 29.12778386272362, "grad_norm": 0.058845311403274536, "learning_rate": 7.971784536460242e-06, "loss": 0.0073, "step": 79850 }, { "epoch": 29.146038700255566, "grad_norm": 0.06181783229112625, "learning_rate": 7.969952363503116e-06, "loss": 0.0067, "step": 79900 }, { "epoch": 29.164293537787515, "grad_norm": 0.06043371185660362, "learning_rate": 7.968120190545988e-06, "loss": 0.0074, "step": 79950 }, { "epoch": 29.18254837531946, "grad_norm": 0.08618185669183731, "learning_rate": 7.966288017588862e-06, "loss": 0.0072, "step": 80000 }, { "epoch": 29.200803212851405, "grad_norm": 0.06627660244703293, "learning_rate": 7.964455844631734e-06, "loss": 0.0071, "step": 80050 }, { "epoch": 29.219058050383353, "grad_norm": 0.08391069620847702, "learning_rate": 7.962623671674606e-06, "loss": 0.007, "step": 80100 }, { "epoch": 29.237312887915298, "grad_norm": 0.06872262060642242, "learning_rate": 7.960791498717479e-06, "loss": 0.0076, "step": 80150 }, { "epoch": 29.255567725447243, "grad_norm": 0.059014998376369476, "learning_rate": 7.958959325760353e-06, "loss": 0.0073, "step": 80200 }, { "epoch": 29.27382256297919, "grad_norm": 0.06593556702136993, "learning_rate": 7.957127152803225e-06, "loss": 0.0074, "step": 80250 }, { "epoch": 29.292077400511136, "grad_norm": 0.2186076045036316, "learning_rate": 7.955294979846097e-06, "loss": 0.007, "step": 80300 }, { "epoch": 29.31033223804308, "grad_norm": 0.07353591173887253, "learning_rate": 7.953462806888971e-06, "loss": 0.0074, "step": 80350 }, { "epoch": 29.328587075575026, "grad_norm": 0.06211809813976288, "learning_rate": 7.951630633931844e-06, "loss": 0.0071, "step": 80400 }, { "epoch": 29.346841913106974, "grad_norm": 0.06414536386728287, "learning_rate": 7.949798460974718e-06, "loss": 0.0072, "step": 80450 }, { "epoch": 29.36509675063892, "grad_norm": 0.05896814167499542, "learning_rate": 7.947966288017588e-06, "loss": 0.0073, "step": 80500 }, { "epoch": 29.383351588170864, "grad_norm": 0.0623180978000164, "learning_rate": 7.946134115060462e-06, "loss": 0.0072, "step": 80550 }, { "epoch": 29.401606425702813, "grad_norm": 0.06735369563102722, "learning_rate": 7.944301942103335e-06, "loss": 0.0073, "step": 80600 }, { "epoch": 29.419861263234758, "grad_norm": 0.06385549157857895, "learning_rate": 7.942469769146209e-06, "loss": 0.0073, "step": 80650 }, { "epoch": 29.438116100766702, "grad_norm": 0.06826087832450867, "learning_rate": 7.940637596189081e-06, "loss": 0.0073, "step": 80700 }, { "epoch": 29.456370938298647, "grad_norm": 0.0658302903175354, "learning_rate": 7.938805423231953e-06, "loss": 0.0075, "step": 80750 }, { "epoch": 29.474625775830596, "grad_norm": 0.07737997174263, "learning_rate": 7.936973250274827e-06, "loss": 0.0076, "step": 80800 }, { "epoch": 29.49288061336254, "grad_norm": 0.0661461353302002, "learning_rate": 7.9351410773177e-06, "loss": 0.0073, "step": 80850 }, { "epoch": 29.511135450894486, "grad_norm": 0.06623401492834091, "learning_rate": 7.933308904360573e-06, "loss": 0.0074, "step": 80900 }, { "epoch": 29.529390288426434, "grad_norm": 0.1918768435716629, "learning_rate": 7.931476731403444e-06, "loss": 0.0075, "step": 80950 }, { "epoch": 29.54764512595838, "grad_norm": 0.06365705281496048, "learning_rate": 7.929644558446318e-06, "loss": 0.0074, "step": 81000 }, { "epoch": 29.565899963490324, "grad_norm": 0.06181911379098892, "learning_rate": 7.92781238548919e-06, "loss": 0.0077, "step": 81050 }, { "epoch": 29.584154801022272, "grad_norm": 0.08525305986404419, "learning_rate": 7.925980212532064e-06, "loss": 0.0071, "step": 81100 }, { "epoch": 29.602409638554217, "grad_norm": 0.10943957418203354, "learning_rate": 7.924148039574937e-06, "loss": 0.0071, "step": 81150 }, { "epoch": 29.620664476086162, "grad_norm": 0.07574176043272018, "learning_rate": 7.922315866617809e-06, "loss": 0.0072, "step": 81200 }, { "epoch": 29.63891931361811, "grad_norm": 0.05273205786943436, "learning_rate": 7.920483693660681e-06, "loss": 0.0076, "step": 81250 }, { "epoch": 29.657174151150056, "grad_norm": 0.07380454987287521, "learning_rate": 7.918651520703555e-06, "loss": 0.0075, "step": 81300 }, { "epoch": 29.675428988682, "grad_norm": 0.051318127661943436, "learning_rate": 7.916819347746428e-06, "loss": 0.0075, "step": 81350 }, { "epoch": 29.693683826213945, "grad_norm": 0.0764152929186821, "learning_rate": 7.9149871747893e-06, "loss": 0.0075, "step": 81400 }, { "epoch": 29.711938663745894, "grad_norm": 0.05802054703235626, "learning_rate": 7.913155001832174e-06, "loss": 0.007, "step": 81450 }, { "epoch": 29.73019350127784, "grad_norm": 0.07457083463668823, "learning_rate": 7.911322828875046e-06, "loss": 0.0071, "step": 81500 }, { "epoch": 29.748448338809784, "grad_norm": 0.06422644853591919, "learning_rate": 7.90949065591792e-06, "loss": 0.0073, "step": 81550 }, { "epoch": 29.766703176341732, "grad_norm": 0.05865848809480667, "learning_rate": 7.90765848296079e-06, "loss": 0.0074, "step": 81600 }, { "epoch": 29.784958013873677, "grad_norm": 0.08266253769397736, "learning_rate": 7.905826310003665e-06, "loss": 0.0075, "step": 81650 }, { "epoch": 29.803212851405622, "grad_norm": 0.06528962403535843, "learning_rate": 7.903994137046537e-06, "loss": 0.0071, "step": 81700 }, { "epoch": 29.821467688937567, "grad_norm": 0.06314411759376526, "learning_rate": 7.902161964089411e-06, "loss": 0.0074, "step": 81750 }, { "epoch": 29.839722526469515, "grad_norm": 0.0889691486954689, "learning_rate": 7.900329791132283e-06, "loss": 0.0073, "step": 81800 }, { "epoch": 29.85797736400146, "grad_norm": 0.10467179864645004, "learning_rate": 7.898497618175156e-06, "loss": 0.0073, "step": 81850 }, { "epoch": 29.876232201533405, "grad_norm": 0.06854028254747391, "learning_rate": 7.89666544521803e-06, "loss": 0.0075, "step": 81900 }, { "epoch": 29.894487039065353, "grad_norm": 0.11528389155864716, "learning_rate": 7.894833272260902e-06, "loss": 0.007, "step": 81950 }, { "epoch": 29.9127418765973, "grad_norm": 0.08205393701791763, "learning_rate": 7.893001099303776e-06, "loss": 0.0076, "step": 82000 }, { "epoch": 29.930996714129243, "grad_norm": 0.0832001119852066, "learning_rate": 7.891168926346647e-06, "loss": 0.0075, "step": 82050 }, { "epoch": 29.94925155166119, "grad_norm": 0.07379259914159775, "learning_rate": 7.88933675338952e-06, "loss": 0.0073, "step": 82100 }, { "epoch": 29.967506389193137, "grad_norm": 0.07361481338739395, "learning_rate": 7.887504580432393e-06, "loss": 0.0073, "step": 82150 }, { "epoch": 29.98576122672508, "grad_norm": 0.0669662281870842, "learning_rate": 7.885672407475267e-06, "loss": 0.0074, "step": 82200 }, { "epoch": 30.004016064257026, "grad_norm": 0.06385339051485062, "learning_rate": 7.88384023451814e-06, "loss": 0.0076, "step": 82250 }, { "epoch": 30.022270901788975, "grad_norm": 0.16685138642787933, "learning_rate": 7.882008061561012e-06, "loss": 0.0071, "step": 82300 }, { "epoch": 30.04052573932092, "grad_norm": 0.061589423567056656, "learning_rate": 7.880175888603886e-06, "loss": 0.0068, "step": 82350 }, { "epoch": 30.058780576852865, "grad_norm": 0.09114423394203186, "learning_rate": 7.878343715646758e-06, "loss": 0.0071, "step": 82400 }, { "epoch": 30.077035414384813, "grad_norm": 0.08508852124214172, "learning_rate": 7.876511542689632e-06, "loss": 0.0076, "step": 82450 }, { "epoch": 30.095290251916758, "grad_norm": 0.08360867947340012, "learning_rate": 7.874679369732502e-06, "loss": 0.007, "step": 82500 }, { "epoch": 30.113545089448703, "grad_norm": 0.07298173755407333, "learning_rate": 7.872847196775376e-06, "loss": 0.0069, "step": 82550 }, { "epoch": 30.13179992698065, "grad_norm": 0.07659643143415451, "learning_rate": 7.871015023818249e-06, "loss": 0.0067, "step": 82600 }, { "epoch": 30.150054764512596, "grad_norm": 0.07533630728721619, "learning_rate": 7.869182850861123e-06, "loss": 0.0074, "step": 82650 }, { "epoch": 30.16830960204454, "grad_norm": 0.06985796242952347, "learning_rate": 7.867350677903995e-06, "loss": 0.0073, "step": 82700 }, { "epoch": 30.186564439576486, "grad_norm": 0.07252094149589539, "learning_rate": 7.865518504946867e-06, "loss": 0.0074, "step": 82750 }, { "epoch": 30.204819277108435, "grad_norm": 0.0684509426355362, "learning_rate": 7.863686331989741e-06, "loss": 0.0068, "step": 82800 }, { "epoch": 30.22307411464038, "grad_norm": 0.08677452057600021, "learning_rate": 7.861854159032614e-06, "loss": 0.0069, "step": 82850 }, { "epoch": 30.241328952172324, "grad_norm": 0.07157012820243835, "learning_rate": 7.860021986075486e-06, "loss": 0.0069, "step": 82900 }, { "epoch": 30.259583789704273, "grad_norm": 0.07455437630414963, "learning_rate": 7.858189813118358e-06, "loss": 0.0069, "step": 82950 }, { "epoch": 30.277838627236218, "grad_norm": 0.08211082220077515, "learning_rate": 7.856357640161232e-06, "loss": 0.0071, "step": 83000 }, { "epoch": 30.296093464768163, "grad_norm": 0.07487872242927551, "learning_rate": 7.854525467204105e-06, "loss": 0.0071, "step": 83050 }, { "epoch": 30.31434830230011, "grad_norm": 0.07527283579111099, "learning_rate": 7.852693294246979e-06, "loss": 0.0071, "step": 83100 }, { "epoch": 30.332603139832056, "grad_norm": 0.0807192251086235, "learning_rate": 7.85086112128985e-06, "loss": 0.0071, "step": 83150 }, { "epoch": 30.350857977364, "grad_norm": 0.07028758525848389, "learning_rate": 7.849028948332723e-06, "loss": 0.0075, "step": 83200 }, { "epoch": 30.369112814895946, "grad_norm": 0.07940153777599335, "learning_rate": 7.847196775375595e-06, "loss": 0.007, "step": 83250 }, { "epoch": 30.387367652427894, "grad_norm": 0.08903021365404129, "learning_rate": 7.84536460241847e-06, "loss": 0.0071, "step": 83300 }, { "epoch": 30.40562248995984, "grad_norm": 0.0699775442481041, "learning_rate": 7.843532429461342e-06, "loss": 0.0068, "step": 83350 }, { "epoch": 30.423877327491784, "grad_norm": 0.17301498353481293, "learning_rate": 7.841700256504214e-06, "loss": 0.007, "step": 83400 }, { "epoch": 30.442132165023732, "grad_norm": 0.07502716034650803, "learning_rate": 7.839868083547088e-06, "loss": 0.0073, "step": 83450 }, { "epoch": 30.460387002555677, "grad_norm": 0.09460007399320602, "learning_rate": 7.83803591058996e-06, "loss": 0.0069, "step": 83500 }, { "epoch": 30.478641840087622, "grad_norm": 0.07436145842075348, "learning_rate": 7.836203737632834e-06, "loss": 0.0072, "step": 83550 }, { "epoch": 30.49689667761957, "grad_norm": 0.0753907635807991, "learning_rate": 7.834371564675705e-06, "loss": 0.007, "step": 83600 }, { "epoch": 30.515151515151516, "grad_norm": 0.07714949548244476, "learning_rate": 7.832539391718579e-06, "loss": 0.0071, "step": 83650 }, { "epoch": 30.53340635268346, "grad_norm": 0.0818701833486557, "learning_rate": 7.830707218761451e-06, "loss": 0.007, "step": 83700 }, { "epoch": 30.551661190215405, "grad_norm": 0.06792210787534714, "learning_rate": 7.828875045804325e-06, "loss": 0.0071, "step": 83750 }, { "epoch": 30.569916027747354, "grad_norm": 0.0595746673643589, "learning_rate": 7.827042872847198e-06, "loss": 0.0072, "step": 83800 }, { "epoch": 30.5881708652793, "grad_norm": 0.08002226054668427, "learning_rate": 7.82521069989007e-06, "loss": 0.007, "step": 83850 }, { "epoch": 30.606425702811244, "grad_norm": 0.060037169605493546, "learning_rate": 7.823378526932944e-06, "loss": 0.0072, "step": 83900 }, { "epoch": 30.624680540343192, "grad_norm": 0.06481093913316727, "learning_rate": 7.821546353975816e-06, "loss": 0.0074, "step": 83950 }, { "epoch": 30.642935377875137, "grad_norm": 0.058170221745967865, "learning_rate": 7.819714181018688e-06, "loss": 0.0073, "step": 84000 }, { "epoch": 30.661190215407082, "grad_norm": 0.06569618731737137, "learning_rate": 7.81788200806156e-06, "loss": 0.0072, "step": 84050 }, { "epoch": 30.67944505293903, "grad_norm": 0.09736990928649902, "learning_rate": 7.816049835104435e-06, "loss": 0.0068, "step": 84100 }, { "epoch": 30.697699890470975, "grad_norm": 0.10132616758346558, "learning_rate": 7.814217662147307e-06, "loss": 0.0072, "step": 84150 }, { "epoch": 30.71595472800292, "grad_norm": 0.0747184008359909, "learning_rate": 7.812385489190181e-06, "loss": 0.0073, "step": 84200 }, { "epoch": 30.734209565534865, "grad_norm": 0.07132124900817871, "learning_rate": 7.810553316233053e-06, "loss": 0.0071, "step": 84250 }, { "epoch": 30.752464403066814, "grad_norm": 0.06775057315826416, "learning_rate": 7.808721143275926e-06, "loss": 0.0071, "step": 84300 }, { "epoch": 30.77071924059876, "grad_norm": 0.08591961860656738, "learning_rate": 7.8068889703188e-06, "loss": 0.0072, "step": 84350 }, { "epoch": 30.788974078130703, "grad_norm": 0.07291033864021301, "learning_rate": 7.805056797361672e-06, "loss": 0.0073, "step": 84400 }, { "epoch": 30.80722891566265, "grad_norm": 0.11974475532770157, "learning_rate": 7.803224624404544e-06, "loss": 0.007, "step": 84450 }, { "epoch": 30.825483753194597, "grad_norm": 0.07179722934961319, "learning_rate": 7.801392451447417e-06, "loss": 0.0071, "step": 84500 }, { "epoch": 30.84373859072654, "grad_norm": 0.06474166363477707, "learning_rate": 7.79956027849029e-06, "loss": 0.0071, "step": 84550 }, { "epoch": 30.86199342825849, "grad_norm": 0.05868005380034447, "learning_rate": 7.797728105533163e-06, "loss": 0.0067, "step": 84600 }, { "epoch": 30.880248265790435, "grad_norm": 0.08048903942108154, "learning_rate": 7.795895932576035e-06, "loss": 0.0069, "step": 84650 }, { "epoch": 30.89850310332238, "grad_norm": 0.06953826546669006, "learning_rate": 7.79406375961891e-06, "loss": 0.0072, "step": 84700 }, { "epoch": 30.916757940854325, "grad_norm": 0.0686163604259491, "learning_rate": 7.792231586661781e-06, "loss": 0.0066, "step": 84750 }, { "epoch": 30.935012778386273, "grad_norm": 0.06322144716978073, "learning_rate": 7.790399413704654e-06, "loss": 0.0072, "step": 84800 }, { "epoch": 30.953267615918218, "grad_norm": 0.05418592691421509, "learning_rate": 7.788567240747528e-06, "loss": 0.0074, "step": 84850 }, { "epoch": 30.971522453450163, "grad_norm": 0.06316572427749634, "learning_rate": 7.7867350677904e-06, "loss": 0.0068, "step": 84900 }, { "epoch": 30.98977729098211, "grad_norm": 0.06772512197494507, "learning_rate": 7.784902894833272e-06, "loss": 0.0068, "step": 84950 }, { "epoch": 31.008032128514056, "grad_norm": 0.06311652809381485, "learning_rate": 7.783070721876146e-06, "loss": 0.0074, "step": 85000 }, { "epoch": 31.026286966046, "grad_norm": 0.09480763971805573, "learning_rate": 7.781238548919019e-06, "loss": 0.0067, "step": 85050 }, { "epoch": 31.04454180357795, "grad_norm": 0.11540386080741882, "learning_rate": 7.779406375961891e-06, "loss": 0.0071, "step": 85100 }, { "epoch": 31.062796641109895, "grad_norm": 0.07928896695375443, "learning_rate": 7.777574203004763e-06, "loss": 0.0069, "step": 85150 }, { "epoch": 31.08105147864184, "grad_norm": 0.07635536044836044, "learning_rate": 7.775742030047637e-06, "loss": 0.0072, "step": 85200 }, { "epoch": 31.099306316173784, "grad_norm": 0.0827091783285141, "learning_rate": 7.77390985709051e-06, "loss": 0.0071, "step": 85250 }, { "epoch": 31.117561153705733, "grad_norm": 0.0769251137971878, "learning_rate": 7.772077684133384e-06, "loss": 0.0067, "step": 85300 }, { "epoch": 31.135815991237678, "grad_norm": 0.1027689129114151, "learning_rate": 7.770245511176256e-06, "loss": 0.0068, "step": 85350 }, { "epoch": 31.154070828769623, "grad_norm": 0.08222832530736923, "learning_rate": 7.768413338219128e-06, "loss": 0.0066, "step": 85400 }, { "epoch": 31.17232566630157, "grad_norm": 0.07502683997154236, "learning_rate": 7.766581165262002e-06, "loss": 0.0073, "step": 85450 }, { "epoch": 31.190580503833516, "grad_norm": 0.05718762055039406, "learning_rate": 7.764748992304874e-06, "loss": 0.0068, "step": 85500 }, { "epoch": 31.20883534136546, "grad_norm": 0.06570791453123093, "learning_rate": 7.762916819347747e-06, "loss": 0.0069, "step": 85550 }, { "epoch": 31.22709017889741, "grad_norm": 0.12292777746915817, "learning_rate": 7.761084646390619e-06, "loss": 0.0067, "step": 85600 }, { "epoch": 31.245345016429354, "grad_norm": 0.08384477347135544, "learning_rate": 7.759252473433493e-06, "loss": 0.0067, "step": 85650 }, { "epoch": 31.2635998539613, "grad_norm": 0.0846899226307869, "learning_rate": 7.757420300476365e-06, "loss": 0.0073, "step": 85700 }, { "epoch": 31.281854691493244, "grad_norm": 0.0630795881152153, "learning_rate": 7.755588127519238e-06, "loss": 0.0068, "step": 85750 }, { "epoch": 31.300109529025193, "grad_norm": 0.06208505854010582, "learning_rate": 7.753755954562112e-06, "loss": 0.0071, "step": 85800 }, { "epoch": 31.318364366557137, "grad_norm": 0.06848796457052231, "learning_rate": 7.751923781604984e-06, "loss": 0.0071, "step": 85850 }, { "epoch": 31.336619204089082, "grad_norm": 0.08801737427711487, "learning_rate": 7.750091608647858e-06, "loss": 0.0069, "step": 85900 }, { "epoch": 31.35487404162103, "grad_norm": 0.08753516525030136, "learning_rate": 7.74825943569073e-06, "loss": 0.0067, "step": 85950 }, { "epoch": 31.373128879152976, "grad_norm": 0.0878089964389801, "learning_rate": 7.746427262733603e-06, "loss": 0.0069, "step": 86000 }, { "epoch": 31.39138371668492, "grad_norm": 0.06123606488108635, "learning_rate": 7.744595089776475e-06, "loss": 0.0068, "step": 86050 }, { "epoch": 31.40963855421687, "grad_norm": 0.08994601666927338, "learning_rate": 7.742762916819349e-06, "loss": 0.0067, "step": 86100 }, { "epoch": 31.427893391748814, "grad_norm": 0.09004167467355728, "learning_rate": 7.740930743862221e-06, "loss": 0.0066, "step": 86150 }, { "epoch": 31.44614822928076, "grad_norm": 0.09477866441011429, "learning_rate": 7.739098570905094e-06, "loss": 0.0071, "step": 86200 }, { "epoch": 31.464403066812704, "grad_norm": 0.09493061900138855, "learning_rate": 7.737266397947968e-06, "loss": 0.007, "step": 86250 }, { "epoch": 31.482657904344652, "grad_norm": 0.05244487151503563, "learning_rate": 7.73543422499084e-06, "loss": 0.0069, "step": 86300 }, { "epoch": 31.500912741876597, "grad_norm": 0.06595064699649811, "learning_rate": 7.733602052033712e-06, "loss": 0.0072, "step": 86350 }, { "epoch": 31.519167579408542, "grad_norm": 0.0859295055270195, "learning_rate": 7.731769879076584e-06, "loss": 0.007, "step": 86400 }, { "epoch": 31.53742241694049, "grad_norm": 0.05661967396736145, "learning_rate": 7.729937706119458e-06, "loss": 0.0068, "step": 86450 }, { "epoch": 31.555677254472435, "grad_norm": 0.07679593563079834, "learning_rate": 7.72810553316233e-06, "loss": 0.0071, "step": 86500 }, { "epoch": 31.57393209200438, "grad_norm": 0.07280939072370529, "learning_rate": 7.726273360205205e-06, "loss": 0.0071, "step": 86550 }, { "epoch": 31.59218692953633, "grad_norm": 0.08821633458137512, "learning_rate": 7.724441187248077e-06, "loss": 0.0067, "step": 86600 }, { "epoch": 31.610441767068274, "grad_norm": 0.07913122326135635, "learning_rate": 7.72260901429095e-06, "loss": 0.0066, "step": 86650 }, { "epoch": 31.62869660460022, "grad_norm": 0.07386074215173721, "learning_rate": 7.720776841333822e-06, "loss": 0.0072, "step": 86700 }, { "epoch": 31.646951442132163, "grad_norm": 0.0783054456114769, "learning_rate": 7.718944668376696e-06, "loss": 0.0067, "step": 86750 }, { "epoch": 31.665206279664112, "grad_norm": 0.07334897667169571, "learning_rate": 7.717112495419568e-06, "loss": 0.007, "step": 86800 }, { "epoch": 31.683461117196057, "grad_norm": 0.08848831802606583, "learning_rate": 7.71528032246244e-06, "loss": 0.007, "step": 86850 }, { "epoch": 31.701715954728, "grad_norm": 0.08435018360614777, "learning_rate": 7.713448149505314e-06, "loss": 0.0068, "step": 86900 }, { "epoch": 31.71997079225995, "grad_norm": 0.11135094612836838, "learning_rate": 7.711615976548187e-06, "loss": 0.0071, "step": 86950 }, { "epoch": 31.738225629791895, "grad_norm": 0.06772398948669434, "learning_rate": 7.70978380359106e-06, "loss": 0.0069, "step": 87000 }, { "epoch": 31.75648046732384, "grad_norm": 0.06990749388933182, "learning_rate": 7.707951630633931e-06, "loss": 0.007, "step": 87050 }, { "epoch": 31.77473530485579, "grad_norm": 0.07143830507993698, "learning_rate": 7.706119457676805e-06, "loss": 0.0072, "step": 87100 }, { "epoch": 31.792990142387733, "grad_norm": 0.0700405016541481, "learning_rate": 7.704287284719677e-06, "loss": 0.0068, "step": 87150 }, { "epoch": 31.811244979919678, "grad_norm": 0.10221503674983978, "learning_rate": 7.702455111762551e-06, "loss": 0.007, "step": 87200 }, { "epoch": 31.829499817451623, "grad_norm": 0.07455871254205704, "learning_rate": 7.700622938805424e-06, "loss": 0.0069, "step": 87250 }, { "epoch": 31.84775465498357, "grad_norm": 0.09439749270677567, "learning_rate": 7.698790765848296e-06, "loss": 0.0067, "step": 87300 }, { "epoch": 31.866009492515516, "grad_norm": 0.05693293362855911, "learning_rate": 7.69695859289117e-06, "loss": 0.0072, "step": 87350 }, { "epoch": 31.88426433004746, "grad_norm": 0.09308143705129623, "learning_rate": 7.695126419934042e-06, "loss": 0.0069, "step": 87400 }, { "epoch": 31.90251916757941, "grad_norm": 0.07922778278589249, "learning_rate": 7.693294246976916e-06, "loss": 0.0072, "step": 87450 }, { "epoch": 31.920774005111355, "grad_norm": 0.06627195328474045, "learning_rate": 7.691462074019787e-06, "loss": 0.0067, "step": 87500 }, { "epoch": 31.9390288426433, "grad_norm": 0.1097969114780426, "learning_rate": 7.689629901062661e-06, "loss": 0.0068, "step": 87550 }, { "epoch": 31.957283680175248, "grad_norm": 0.07041894644498825, "learning_rate": 7.687797728105533e-06, "loss": 0.0073, "step": 87600 }, { "epoch": 31.975538517707193, "grad_norm": 0.07094933092594147, "learning_rate": 7.685965555148407e-06, "loss": 0.007, "step": 87650 }, { "epoch": 31.993793355239138, "grad_norm": 0.08098384737968445, "learning_rate": 7.68413338219128e-06, "loss": 0.007, "step": 87700 }, { "epoch": 32.01204819277108, "grad_norm": 0.0577063225209713, "learning_rate": 7.682301209234152e-06, "loss": 0.0066, "step": 87750 }, { "epoch": 32.03030303030303, "grad_norm": 0.054525069892406464, "learning_rate": 7.680469036277026e-06, "loss": 0.0066, "step": 87800 }, { "epoch": 32.04855786783498, "grad_norm": 0.0784829631447792, "learning_rate": 7.678636863319898e-06, "loss": 0.0065, "step": 87850 }, { "epoch": 32.06681270536692, "grad_norm": 0.061458442360162735, "learning_rate": 7.676804690362772e-06, "loss": 0.007, "step": 87900 }, { "epoch": 32.08506754289887, "grad_norm": 0.1012117937207222, "learning_rate": 7.674972517405643e-06, "loss": 0.0066, "step": 87950 }, { "epoch": 32.10332238043081, "grad_norm": 0.07040351629257202, "learning_rate": 7.673140344448517e-06, "loss": 0.0068, "step": 88000 }, { "epoch": 32.12157721796276, "grad_norm": 0.0767582356929779, "learning_rate": 7.671308171491389e-06, "loss": 0.0067, "step": 88050 }, { "epoch": 32.13983205549471, "grad_norm": 0.07797151058912277, "learning_rate": 7.669475998534263e-06, "loss": 0.0067, "step": 88100 }, { "epoch": 32.15808689302665, "grad_norm": 0.0579819530248642, "learning_rate": 7.667643825577135e-06, "loss": 0.0065, "step": 88150 }, { "epoch": 32.1763417305586, "grad_norm": 0.08333556354045868, "learning_rate": 7.665811652620008e-06, "loss": 0.0071, "step": 88200 }, { "epoch": 32.194596568090546, "grad_norm": 0.07710177451372147, "learning_rate": 7.663979479662882e-06, "loss": 0.0067, "step": 88250 }, { "epoch": 32.21285140562249, "grad_norm": 0.08976130187511444, "learning_rate": 7.662147306705754e-06, "loss": 0.0069, "step": 88300 }, { "epoch": 32.231106243154436, "grad_norm": 0.057984720915555954, "learning_rate": 7.660315133748626e-06, "loss": 0.0068, "step": 88350 }, { "epoch": 32.249361080686384, "grad_norm": 0.055399633944034576, "learning_rate": 7.658482960791499e-06, "loss": 0.007, "step": 88400 }, { "epoch": 32.267615918218326, "grad_norm": 0.0700848400592804, "learning_rate": 7.656650787834373e-06, "loss": 0.0065, "step": 88450 }, { "epoch": 32.285870755750274, "grad_norm": 0.06455592811107635, "learning_rate": 7.654818614877245e-06, "loss": 0.0068, "step": 88500 }, { "epoch": 32.30412559328222, "grad_norm": 0.11038955301046371, "learning_rate": 7.652986441920119e-06, "loss": 0.0065, "step": 88550 }, { "epoch": 32.322380430814164, "grad_norm": 0.06685930490493774, "learning_rate": 7.65115426896299e-06, "loss": 0.0069, "step": 88600 }, { "epoch": 32.34063526834611, "grad_norm": 0.08404071629047394, "learning_rate": 7.649322096005863e-06, "loss": 0.0069, "step": 88650 }, { "epoch": 32.35889010587806, "grad_norm": 0.06650456041097641, "learning_rate": 7.647489923048736e-06, "loss": 0.0069, "step": 88700 }, { "epoch": 32.37714494341, "grad_norm": 0.06137634441256523, "learning_rate": 7.64565775009161e-06, "loss": 0.0068, "step": 88750 }, { "epoch": 32.39539978094195, "grad_norm": 0.07725632935762405, "learning_rate": 7.643825577134482e-06, "loss": 0.0069, "step": 88800 }, { "epoch": 32.4136546184739, "grad_norm": 0.07307960838079453, "learning_rate": 7.641993404177354e-06, "loss": 0.0065, "step": 88850 }, { "epoch": 32.43190945600584, "grad_norm": 0.06974013149738312, "learning_rate": 7.640161231220228e-06, "loss": 0.0069, "step": 88900 }, { "epoch": 32.45016429353779, "grad_norm": 0.07056721299886703, "learning_rate": 7.6383290582631e-06, "loss": 0.0068, "step": 88950 }, { "epoch": 32.46841913106973, "grad_norm": 0.0880282074213028, "learning_rate": 7.636496885305975e-06, "loss": 0.007, "step": 89000 }, { "epoch": 32.48667396860168, "grad_norm": 0.0785638615489006, "learning_rate": 7.634664712348845e-06, "loss": 0.0066, "step": 89050 }, { "epoch": 32.50492880613363, "grad_norm": 0.1087317168712616, "learning_rate": 7.63283253939172e-06, "loss": 0.0069, "step": 89100 }, { "epoch": 32.52318364366557, "grad_norm": 0.0878078043460846, "learning_rate": 7.631000366434592e-06, "loss": 0.0066, "step": 89150 }, { "epoch": 32.54143848119752, "grad_norm": 0.08884698152542114, "learning_rate": 7.6291681934774656e-06, "loss": 0.0063, "step": 89200 }, { "epoch": 32.559693318729465, "grad_norm": 0.06762262433767319, "learning_rate": 7.627336020520337e-06, "loss": 0.0065, "step": 89250 }, { "epoch": 32.57794815626141, "grad_norm": 0.07849127799272537, "learning_rate": 7.62550384756321e-06, "loss": 0.0069, "step": 89300 }, { "epoch": 32.596202993793355, "grad_norm": 0.09534306824207306, "learning_rate": 7.623671674606083e-06, "loss": 0.007, "step": 89350 }, { "epoch": 32.6144578313253, "grad_norm": 0.08199641853570938, "learning_rate": 7.6218395016489565e-06, "loss": 0.0069, "step": 89400 }, { "epoch": 32.632712668857245, "grad_norm": 0.06695178151130676, "learning_rate": 7.62000732869183e-06, "loss": 0.0069, "step": 89450 }, { "epoch": 32.65096750638919, "grad_norm": 0.08887915313243866, "learning_rate": 7.618175155734702e-06, "loss": 0.0071, "step": 89500 }, { "epoch": 32.66922234392114, "grad_norm": 0.06490937620401382, "learning_rate": 7.616342982777575e-06, "loss": 0.0068, "step": 89550 }, { "epoch": 32.68747718145308, "grad_norm": 0.06714525073766708, "learning_rate": 7.614510809820448e-06, "loss": 0.0069, "step": 89600 }, { "epoch": 32.70573201898503, "grad_norm": 0.06591560691595078, "learning_rate": 7.612678636863321e-06, "loss": 0.007, "step": 89650 }, { "epoch": 32.72398685651698, "grad_norm": 0.05745172128081322, "learning_rate": 7.610846463906193e-06, "loss": 0.0068, "step": 89700 }, { "epoch": 32.74224169404892, "grad_norm": 0.09448807686567307, "learning_rate": 7.609014290949066e-06, "loss": 0.0071, "step": 89750 }, { "epoch": 32.76049653158087, "grad_norm": 0.06272632628679276, "learning_rate": 7.607182117991939e-06, "loss": 0.0069, "step": 89800 }, { "epoch": 32.77875136911282, "grad_norm": 0.06873849779367447, "learning_rate": 7.605349945034812e-06, "loss": 0.0068, "step": 89850 }, { "epoch": 32.79700620664476, "grad_norm": 0.05152475833892822, "learning_rate": 7.603517772077685e-06, "loss": 0.0069, "step": 89900 }, { "epoch": 32.81526104417671, "grad_norm": 0.0759289488196373, "learning_rate": 7.601685599120558e-06, "loss": 0.0068, "step": 89950 }, { "epoch": 32.83351588170865, "grad_norm": 0.06016925349831581, "learning_rate": 7.599853426163431e-06, "loss": 0.007, "step": 90000 }, { "epoch": 32.8517707192406, "grad_norm": 0.06483444571495056, "learning_rate": 7.598021253206303e-06, "loss": 0.0067, "step": 90050 }, { "epoch": 32.870025556772546, "grad_norm": 0.0779428482055664, "learning_rate": 7.596189080249176e-06, "loss": 0.0069, "step": 90100 }, { "epoch": 32.88828039430449, "grad_norm": 0.06382060796022415, "learning_rate": 7.594356907292049e-06, "loss": 0.0066, "step": 90150 }, { "epoch": 32.906535231836436, "grad_norm": 0.08806990832090378, "learning_rate": 7.592524734334922e-06, "loss": 0.0067, "step": 90200 }, { "epoch": 32.924790069368385, "grad_norm": 0.06399805098772049, "learning_rate": 7.590692561377795e-06, "loss": 0.0069, "step": 90250 }, { "epoch": 32.943044906900326, "grad_norm": 0.11611298471689224, "learning_rate": 7.588860388420668e-06, "loss": 0.0065, "step": 90300 }, { "epoch": 32.961299744432274, "grad_norm": 0.08025538921356201, "learning_rate": 7.5870282154635395e-06, "loss": 0.0067, "step": 90350 }, { "epoch": 32.97955458196422, "grad_norm": 0.07093523442745209, "learning_rate": 7.585196042506413e-06, "loss": 0.0065, "step": 90400 }, { "epoch": 32.997809419496164, "grad_norm": 0.07254663854837418, "learning_rate": 7.583363869549286e-06, "loss": 0.0067, "step": 90450 }, { "epoch": 33.01606425702811, "grad_norm": 0.07248546928167343, "learning_rate": 7.581531696592159e-06, "loss": 0.0063, "step": 90500 }, { "epoch": 33.03431909456006, "grad_norm": 0.09890737384557724, "learning_rate": 7.579699523635031e-06, "loss": 0.0067, "step": 90550 }, { "epoch": 33.052573932092, "grad_norm": 0.0862441211938858, "learning_rate": 7.5778673506779044e-06, "loss": 0.0064, "step": 90600 }, { "epoch": 33.07082876962395, "grad_norm": 0.07000481337308884, "learning_rate": 7.576035177720778e-06, "loss": 0.0063, "step": 90650 }, { "epoch": 33.0890836071559, "grad_norm": 0.09363103657960892, "learning_rate": 7.574203004763651e-06, "loss": 0.0066, "step": 90700 }, { "epoch": 33.10733844468784, "grad_norm": 0.0768694132566452, "learning_rate": 7.572370831806524e-06, "loss": 0.0069, "step": 90750 }, { "epoch": 33.12559328221979, "grad_norm": 0.06516299396753311, "learning_rate": 7.570538658849395e-06, "loss": 0.0066, "step": 90800 }, { "epoch": 33.14384811975174, "grad_norm": 0.05921657383441925, "learning_rate": 7.5687064858922685e-06, "loss": 0.0063, "step": 90850 }, { "epoch": 33.16210295728368, "grad_norm": 0.07661553472280502, "learning_rate": 7.566874312935142e-06, "loss": 0.0066, "step": 90900 }, { "epoch": 33.18035779481563, "grad_norm": 0.07710602134466171, "learning_rate": 7.565042139978015e-06, "loss": 0.0067, "step": 90950 }, { "epoch": 33.19861263234757, "grad_norm": 0.10175678879022598, "learning_rate": 7.563209967020887e-06, "loss": 0.0064, "step": 91000 }, { "epoch": 33.01825483753195, "grad_norm": 0.08640611916780472, "learning_rate": 7.56137779406376e-06, "loss": 0.0065, "step": 91050 }, { "epoch": 33.03650967506389, "grad_norm": 0.07877669483423233, "learning_rate": 7.559545621106633e-06, "loss": 0.0062, "step": 91100 }, { "epoch": 33.05476451259584, "grad_norm": 0.06036520376801491, "learning_rate": 7.5577134481495066e-06, "loss": 0.006, "step": 91150 }, { "epoch": 33.07301935012779, "grad_norm": 0.07644917070865631, "learning_rate": 7.555881275192378e-06, "loss": 0.0062, "step": 91200 }, { "epoch": 33.09127418765973, "grad_norm": 0.08299405872821808, "learning_rate": 7.554049102235251e-06, "loss": 0.0062, "step": 91250 }, { "epoch": 33.10952902519168, "grad_norm": 0.053021352738142014, "learning_rate": 7.552216929278124e-06, "loss": 0.0063, "step": 91300 }, { "epoch": 33.127783862723625, "grad_norm": 0.07107523083686829, "learning_rate": 7.5503847563209975e-06, "loss": 0.0062, "step": 91350 }, { "epoch": 33.146038700255566, "grad_norm": 0.11849170178174973, "learning_rate": 7.548552583363871e-06, "loss": 0.0062, "step": 91400 }, { "epoch": 33.164293537787515, "grad_norm": 0.07564707845449448, "learning_rate": 7.546720410406743e-06, "loss": 0.0063, "step": 91450 }, { "epoch": 33.182548375319456, "grad_norm": 0.07467735558748245, "learning_rate": 7.544888237449616e-06, "loss": 0.0065, "step": 91500 }, { "epoch": 33.200803212851405, "grad_norm": 0.08215086907148361, "learning_rate": 7.543056064492489e-06, "loss": 0.0065, "step": 91550 }, { "epoch": 33.21905805038335, "grad_norm": 0.10069339722394943, "learning_rate": 7.541223891535362e-06, "loss": 0.0066, "step": 91600 }, { "epoch": 33.237312887915294, "grad_norm": 0.07324201613664627, "learning_rate": 7.539391718578234e-06, "loss": 0.0068, "step": 91650 }, { "epoch": 33.25556772544724, "grad_norm": 0.07787135988473892, "learning_rate": 7.537559545621107e-06, "loss": 0.0065, "step": 91700 }, { "epoch": 33.27382256297919, "grad_norm": 0.0801265761256218, "learning_rate": 7.53572737266398e-06, "loss": 0.0068, "step": 91750 }, { "epoch": 33.29207740051113, "grad_norm": 0.0759592056274414, "learning_rate": 7.533895199706853e-06, "loss": 0.0067, "step": 91800 }, { "epoch": 33.31033223804308, "grad_norm": 0.07302507758140564, "learning_rate": 7.532063026749726e-06, "loss": 0.0067, "step": 91850 }, { "epoch": 33.32858707557503, "grad_norm": 0.06013672426342964, "learning_rate": 7.530230853792599e-06, "loss": 0.0065, "step": 91900 }, { "epoch": 33.34684191310697, "grad_norm": 0.06420207023620605, "learning_rate": 7.528398680835471e-06, "loss": 0.0068, "step": 91950 }, { "epoch": 33.36509675063892, "grad_norm": 0.11937499791383743, "learning_rate": 7.526566507878344e-06, "loss": 0.007, "step": 92000 }, { "epoch": 33.38335158817087, "grad_norm": 0.07446599751710892, "learning_rate": 7.524734334921217e-06, "loss": 0.0067, "step": 92050 }, { "epoch": 33.40160642570281, "grad_norm": 0.06346719712018967, "learning_rate": 7.52290216196409e-06, "loss": 0.0069, "step": 92100 }, { "epoch": 33.41986126323476, "grad_norm": 0.06799725443124771, "learning_rate": 7.521069989006963e-06, "loss": 0.0067, "step": 92150 }, { "epoch": 33.438116100766706, "grad_norm": 0.07583191990852356, "learning_rate": 7.519237816049836e-06, "loss": 0.007, "step": 92200 }, { "epoch": 33.45637093829865, "grad_norm": 0.07636619359254837, "learning_rate": 7.517405643092709e-06, "loss": 0.0064, "step": 92250 }, { "epoch": 33.474625775830596, "grad_norm": 0.08070795983076096, "learning_rate": 7.5155734701355805e-06, "loss": 0.0063, "step": 92300 }, { "epoch": 33.492880613362544, "grad_norm": 0.06665036827325821, "learning_rate": 7.513741297178454e-06, "loss": 0.0069, "step": 92350 }, { "epoch": 33.511135450894486, "grad_norm": 0.0569748729467392, "learning_rate": 7.511909124221327e-06, "loss": 0.0064, "step": 92400 }, { "epoch": 33.529390288426434, "grad_norm": 0.06580673903226852, "learning_rate": 7.5100769512642e-06, "loss": 0.0066, "step": 92450 }, { "epoch": 33.54764512595838, "grad_norm": 0.06324420869350433, "learning_rate": 7.508244778307073e-06, "loss": 0.0064, "step": 92500 }, { "epoch": 33.565899963490324, "grad_norm": 0.07890833914279938, "learning_rate": 7.5064126053499454e-06, "loss": 0.0065, "step": 92550 }, { "epoch": 33.58415480102227, "grad_norm": 0.0710713267326355, "learning_rate": 7.504580432392819e-06, "loss": 0.0065, "step": 92600 }, { "epoch": 33.602409638554214, "grad_norm": 0.08788053691387177, "learning_rate": 7.502748259435692e-06, "loss": 0.0064, "step": 92650 }, { "epoch": 33.62066447608616, "grad_norm": 0.09910070896148682, "learning_rate": 7.500916086478565e-06, "loss": 0.0068, "step": 92700 }, { "epoch": 33.63891931361811, "grad_norm": 0.09052541851997375, "learning_rate": 7.499083913521436e-06, "loss": 0.0067, "step": 92750 }, { "epoch": 33.65717415115005, "grad_norm": 0.07365219295024872, "learning_rate": 7.4972517405643095e-06, "loss": 0.0067, "step": 92800 }, { "epoch": 33.675428988682, "grad_norm": 0.08162645250558853, "learning_rate": 7.495419567607183e-06, "loss": 0.0065, "step": 92850 }, { "epoch": 33.69368382621395, "grad_norm": 0.07561594247817993, "learning_rate": 7.493587394650056e-06, "loss": 0.0066, "step": 92900 }, { "epoch": 33.71193866374589, "grad_norm": 0.10299866646528244, "learning_rate": 7.491755221692928e-06, "loss": 0.0068, "step": 92950 }, { "epoch": 33.73019350127784, "grad_norm": 0.11611182987689972, "learning_rate": 7.489923048735801e-06, "loss": 0.0068, "step": 93000 }, { "epoch": 33.74844833880979, "grad_norm": 0.05956384167075157, "learning_rate": 7.488090875778674e-06, "loss": 0.0068, "step": 93050 }, { "epoch": 33.76670317634173, "grad_norm": 0.08203823119401932, "learning_rate": 7.4862587028215475e-06, "loss": 0.0063, "step": 93100 }, { "epoch": 33.78495801387368, "grad_norm": 0.10844368487596512, "learning_rate": 7.484426529864421e-06, "loss": 0.0065, "step": 93150 }, { "epoch": 33.803212851405625, "grad_norm": 0.07232064008712769, "learning_rate": 7.482594356907292e-06, "loss": 0.0064, "step": 93200 }, { "epoch": 33.82146768893757, "grad_norm": 0.08342086523771286, "learning_rate": 7.480762183950165e-06, "loss": 0.0066, "step": 93250 }, { "epoch": 33.839722526469515, "grad_norm": 0.08903005719184875, "learning_rate": 7.4789300109930385e-06, "loss": 0.0069, "step": 93300 }, { "epoch": 33.857977364001464, "grad_norm": 0.0657573789358139, "learning_rate": 7.477097838035912e-06, "loss": 0.0066, "step": 93350 }, { "epoch": 33.876232201533405, "grad_norm": 0.08723960071802139, "learning_rate": 7.475265665078784e-06, "loss": 0.0066, "step": 93400 }, { "epoch": 33.89448703906535, "grad_norm": 0.08263439685106277, "learning_rate": 7.473433492121657e-06, "loss": 0.0065, "step": 93450 }, { "epoch": 33.912741876597295, "grad_norm": 0.0738154947757721, "learning_rate": 7.47160131916453e-06, "loss": 0.0065, "step": 93500 }, { "epoch": 33.93099671412924, "grad_norm": 0.09733784198760986, "learning_rate": 7.4697691462074025e-06, "loss": 0.0064, "step": 93550 }, { "epoch": 33.94925155166119, "grad_norm": 0.05863521993160248, "learning_rate": 7.467936973250275e-06, "loss": 0.0064, "step": 93600 }, { "epoch": 33.96750638919313, "grad_norm": 0.08290941268205643, "learning_rate": 7.466104800293148e-06, "loss": 0.0068, "step": 93650 }, { "epoch": 33.98576122672508, "grad_norm": 0.07428315281867981, "learning_rate": 7.464272627336021e-06, "loss": 0.0069, "step": 93700 }, { "epoch": 34.00401606425703, "grad_norm": 0.06447955965995789, "learning_rate": 7.462440454378894e-06, "loss": 0.0064, "step": 93750 }, { "epoch": 34.02227090178897, "grad_norm": 0.0819171667098999, "learning_rate": 7.460608281421767e-06, "loss": 0.0061, "step": 93800 }, { "epoch": 34.04052573932092, "grad_norm": 0.0662824809551239, "learning_rate": 7.458776108464639e-06, "loss": 0.0065, "step": 93850 }, { "epoch": 34.05878057685287, "grad_norm": 0.09369952976703644, "learning_rate": 7.456943935507512e-06, "loss": 0.0063, "step": 93900 }, { "epoch": 34.07703541438481, "grad_norm": 0.08420363068580627, "learning_rate": 7.455111762550385e-06, "loss": 0.0065, "step": 93950 }, { "epoch": 34.09529025191676, "grad_norm": 0.0757431611418724, "learning_rate": 7.453279589593258e-06, "loss": 0.0061, "step": 94000 }, { "epoch": 34.11354508944871, "grad_norm": 0.07587788999080658, "learning_rate": 7.451447416636131e-06, "loss": 0.0064, "step": 94050 }, { "epoch": 34.13179992698065, "grad_norm": 0.08815893530845642, "learning_rate": 7.449615243679004e-06, "loss": 0.0062, "step": 94100 }, { "epoch": 34.150054764512596, "grad_norm": 0.09660379588603973, "learning_rate": 7.447783070721877e-06, "loss": 0.0062, "step": 94150 }, { "epoch": 34.168309602044545, "grad_norm": 0.08910347521305084, "learning_rate": 7.44595089776475e-06, "loss": 0.0062, "step": 94200 }, { "epoch": 34.186564439576486, "grad_norm": 0.07158197462558746, "learning_rate": 7.444118724807623e-06, "loss": 0.0062, "step": 94250 }, { "epoch": 34.204819277108435, "grad_norm": 0.07484482228755951, "learning_rate": 7.442286551850495e-06, "loss": 0.0066, "step": 94300 }, { "epoch": 34.22307411464038, "grad_norm": 0.08587847650051117, "learning_rate": 7.440454378893368e-06, "loss": 0.0067, "step": 94350 }, { "epoch": 34.241328952172324, "grad_norm": 0.08244587481021881, "learning_rate": 7.438622205936241e-06, "loss": 0.0067, "step": 94400 }, { "epoch": 34.25958378970427, "grad_norm": 0.0655352771282196, "learning_rate": 7.436790032979114e-06, "loss": 0.0065, "step": 94450 }, { "epoch": 34.277838627236214, "grad_norm": 0.07686389982700348, "learning_rate": 7.4349578600219864e-06, "loss": 0.0063, "step": 94500 }, { "epoch": 34.29609346476816, "grad_norm": 0.10769946128129959, "learning_rate": 7.4331256870648596e-06, "loss": 0.0065, "step": 94550 }, { "epoch": 34.31434830230011, "grad_norm": 0.0671880766749382, "learning_rate": 7.431293514107733e-06, "loss": 0.0062, "step": 94600 }, { "epoch": 34.33260313983205, "grad_norm": 0.07192417979240417, "learning_rate": 7.429461341150606e-06, "loss": 0.0067, "step": 94650 }, { "epoch": 34.350857977364, "grad_norm": 0.10501667857170105, "learning_rate": 7.427629168193477e-06, "loss": 0.0064, "step": 94700 }, { "epoch": 34.36911281489595, "grad_norm": 0.1024191826581955, "learning_rate": 7.4257969952363505e-06, "loss": 0.0063, "step": 94750 }, { "epoch": 34.38736765242789, "grad_norm": 0.06977224349975586, "learning_rate": 7.423964822279224e-06, "loss": 0.0059, "step": 94800 }, { "epoch": 34.40562248995984, "grad_norm": 0.09432656317949295, "learning_rate": 7.422132649322097e-06, "loss": 0.0064, "step": 94850 }, { "epoch": 34.42387732749179, "grad_norm": 0.07734890282154083, "learning_rate": 7.42030047636497e-06, "loss": 0.0066, "step": 94900 }, { "epoch": 34.44213216502373, "grad_norm": 0.08781347423791885, "learning_rate": 7.418468303407842e-06, "loss": 0.0063, "step": 94950 }, { "epoch": 34.46038700255568, "grad_norm": 0.07764413207769394, "learning_rate": 7.416636130450715e-06, "loss": 0.0062, "step": 95000 }, { "epoch": 34.478641840087626, "grad_norm": 0.0753496065735817, "learning_rate": 7.4148039574935885e-06, "loss": 0.0065, "step": 95050 }, { "epoch": 34.49689667761957, "grad_norm": 0.10010667145252228, "learning_rate": 7.412971784536462e-06, "loss": 0.0067, "step": 95100 }, { "epoch": 34.515151515151516, "grad_norm": 0.09689407795667648, "learning_rate": 7.411139611579333e-06, "loss": 0.0064, "step": 95150 }, { "epoch": 34.533406352683464, "grad_norm": 0.08131708949804306, "learning_rate": 7.409307438622206e-06, "loss": 0.0063, "step": 95200 }, { "epoch": 34.551661190215405, "grad_norm": 0.10950660705566406, "learning_rate": 7.4074752656650794e-06, "loss": 0.0067, "step": 95250 }, { "epoch": 34.569916027747354, "grad_norm": 0.07988095283508301, "learning_rate": 7.405643092707953e-06, "loss": 0.0068, "step": 95300 }, { "epoch": 34.5881708652793, "grad_norm": 0.0752549096941948, "learning_rate": 7.403810919750825e-06, "loss": 0.0061, "step": 95350 }, { "epoch": 34.606425702811244, "grad_norm": 0.06627549231052399, "learning_rate": 7.401978746793698e-06, "loss": 0.0063, "step": 95400 }, { "epoch": 34.62468054034319, "grad_norm": 0.05696071311831474, "learning_rate": 7.40014657383657e-06, "loss": 0.0067, "step": 95450 }, { "epoch": 34.64293537787513, "grad_norm": 0.08637910336256027, "learning_rate": 7.3983144008794435e-06, "loss": 0.0063, "step": 95500 }, { "epoch": 34.66119021540708, "grad_norm": 0.06776952743530273, "learning_rate": 7.396482227922317e-06, "loss": 0.0063, "step": 95550 }, { "epoch": 34.67944505293903, "grad_norm": 0.08339675515890121, "learning_rate": 7.394650054965189e-06, "loss": 0.0066, "step": 95600 }, { "epoch": 34.69769989047097, "grad_norm": 0.08814459294080734, "learning_rate": 7.392817882008062e-06, "loss": 0.0064, "step": 95650 }, { "epoch": 34.71595472800292, "grad_norm": 0.07799631357192993, "learning_rate": 7.390985709050935e-06, "loss": 0.0065, "step": 95700 }, { "epoch": 34.73420956553487, "grad_norm": 0.08527465909719467, "learning_rate": 7.389153536093808e-06, "loss": 0.0066, "step": 95750 }, { "epoch": 34.75246440306681, "grad_norm": 0.0828857421875, "learning_rate": 7.38732136313668e-06, "loss": 0.0062, "step": 95800 }, { "epoch": 34.77071924059876, "grad_norm": 0.0705450177192688, "learning_rate": 7.385489190179553e-06, "loss": 0.0063, "step": 95850 }, { "epoch": 34.78897407813071, "grad_norm": 0.0812784880399704, "learning_rate": 7.383657017222426e-06, "loss": 0.0065, "step": 95900 }, { "epoch": 34.80722891566265, "grad_norm": 0.09739243984222412, "learning_rate": 7.381824844265299e-06, "loss": 0.0064, "step": 95950 }, { "epoch": 34.8254837531946, "grad_norm": 0.0860084667801857, "learning_rate": 7.379992671308172e-06, "loss": 0.0065, "step": 96000 }, { "epoch": 34.843738590726545, "grad_norm": 0.0991797149181366, "learning_rate": 7.378160498351045e-06, "loss": 0.0063, "step": 96050 }, { "epoch": 34.86199342825849, "grad_norm": 0.07097810506820679, "learning_rate": 7.376328325393918e-06, "loss": 0.0064, "step": 96100 }, { "epoch": 34.880248265790435, "grad_norm": 0.0673302710056305, "learning_rate": 7.374496152436791e-06, "loss": 0.0063, "step": 96150 }, { "epoch": 34.89850310332238, "grad_norm": 0.09015652537345886, "learning_rate": 7.372663979479664e-06, "loss": 0.0064, "step": 96200 }, { "epoch": 34.916757940854325, "grad_norm": 0.0800556093454361, "learning_rate": 7.370831806522536e-06, "loss": 0.0066, "step": 96250 }, { "epoch": 34.93501277838627, "grad_norm": 0.08675851672887802, "learning_rate": 7.368999633565409e-06, "loss": 0.0062, "step": 96300 }, { "epoch": 34.95326761591822, "grad_norm": 0.0763222873210907, "learning_rate": 7.367167460608282e-06, "loss": 0.0063, "step": 96350 }, { "epoch": 34.97152245345016, "grad_norm": 0.06653405725955963, "learning_rate": 7.365335287651155e-06, "loss": 0.0064, "step": 96400 }, { "epoch": 34.98977729098211, "grad_norm": 0.06374112516641617, "learning_rate": 7.363503114694027e-06, "loss": 0.0065, "step": 96450 }, { "epoch": 35.00803212851405, "grad_norm": 0.06980693340301514, "learning_rate": 7.3616709417369006e-06, "loss": 0.0058, "step": 96500 }, { "epoch": 35.026286966046, "grad_norm": 0.09170663356781006, "learning_rate": 7.359838768779774e-06, "loss": 0.0063, "step": 96550 }, { "epoch": 35.04454180357795, "grad_norm": 0.08213771134614944, "learning_rate": 7.358006595822647e-06, "loss": 0.0063, "step": 96600 }, { "epoch": 35.06279664110989, "grad_norm": 0.1013873815536499, "learning_rate": 7.35617442286552e-06, "loss": 0.0063, "step": 96650 }, { "epoch": 35.08105147864184, "grad_norm": 0.10413961857557297, "learning_rate": 7.3543422499083915e-06, "loss": 0.0064, "step": 96700 }, { "epoch": 35.09930631617379, "grad_norm": 0.05678106099367142, "learning_rate": 7.352510076951265e-06, "loss": 0.0061, "step": 96750 }, { "epoch": 35.11756115370573, "grad_norm": 0.09337053447961807, "learning_rate": 7.350677903994138e-06, "loss": 0.0063, "step": 96800 }, { "epoch": 35.13581599123768, "grad_norm": 0.09263963252305984, "learning_rate": 7.348845731037011e-06, "loss": 0.0061, "step": 96850 }, { "epoch": 35.154070828769626, "grad_norm": 0.06468808650970459, "learning_rate": 7.347013558079883e-06, "loss": 0.0063, "step": 96900 }, { "epoch": 35.17232566630157, "grad_norm": 0.09230858832597733, "learning_rate": 7.345181385122756e-06, "loss": 0.006, "step": 96950 }, { "epoch": 35.190580503833516, "grad_norm": 0.08486556261777878, "learning_rate": 7.3433492121656295e-06, "loss": 0.0063, "step": 97000 }, { "epoch": 35.208835341365464, "grad_norm": 0.0800599753856659, "learning_rate": 7.341517039208502e-06, "loss": 0.0059, "step": 97050 }, { "epoch": 35.227090178897406, "grad_norm": 0.08433115482330322, "learning_rate": 7.339684866251374e-06, "loss": 0.006, "step": 97100 }, { "epoch": 35.245345016429354, "grad_norm": 0.07007022202014923, "learning_rate": 7.337852693294247e-06, "loss": 0.0064, "step": 97150 }, { "epoch": 35.2635998539613, "grad_norm": 0.05833178758621216, "learning_rate": 7.3360205203371204e-06, "loss": 0.0062, "step": 97200 }, { "epoch": 35.281854691493244, "grad_norm": 0.06445662677288055, "learning_rate": 7.334188347379994e-06, "loss": 0.0061, "step": 97250 }, { "epoch": 35.30010952902519, "grad_norm": 0.14155003428459167, "learning_rate": 7.332356174422867e-06, "loss": 0.0066, "step": 97300 }, { "epoch": 35.31836436655714, "grad_norm": 0.07279349118471146, "learning_rate": 7.330524001465739e-06, "loss": 0.0062, "step": 97350 }, { "epoch": 35.33661920408908, "grad_norm": 0.07542844116687775, "learning_rate": 7.328691828508611e-06, "loss": 0.0062, "step": 97400 }, { "epoch": 35.35487404162103, "grad_norm": 0.1268591731786728, "learning_rate": 7.3268596555514845e-06, "loss": 0.0063, "step": 97450 }, { "epoch": 35.37312887915297, "grad_norm": 0.08298996835947037, "learning_rate": 7.325027482594358e-06, "loss": 0.0061, "step": 97500 }, { "epoch": 35.39138371668492, "grad_norm": 0.08568397164344788, "learning_rate": 7.32319530963723e-06, "loss": 0.0064, "step": 97550 }, { "epoch": 35.40963855421687, "grad_norm": 0.07693557441234589, "learning_rate": 7.321363136680103e-06, "loss": 0.006, "step": 97600 }, { "epoch": 35.42789339174881, "grad_norm": 0.06171392276883125, "learning_rate": 7.319530963722976e-06, "loss": 0.0059, "step": 97650 }, { "epoch": 35.44614822928076, "grad_norm": 0.10071246325969696, "learning_rate": 7.317698790765849e-06, "loss": 0.0064, "step": 97700 }, { "epoch": 35.46440306681271, "grad_norm": 0.07874522358179092, "learning_rate": 7.315866617808721e-06, "loss": 0.0064, "step": 97750 }, { "epoch": 35.48265790434465, "grad_norm": 0.06351500749588013, "learning_rate": 7.314034444851594e-06, "loss": 0.0064, "step": 97800 }, { "epoch": 35.5009127418766, "grad_norm": 0.06227762624621391, "learning_rate": 7.312202271894467e-06, "loss": 0.0062, "step": 97850 }, { "epoch": 35.519167579408546, "grad_norm": 0.08105426281690598, "learning_rate": 7.31037009893734e-06, "loss": 0.0061, "step": 97900 }, { "epoch": 35.53742241694049, "grad_norm": 0.07497590035200119, "learning_rate": 7.3085379259802134e-06, "loss": 0.0059, "step": 97950 }, { "epoch": 35.555677254472435, "grad_norm": 0.08010555803775787, "learning_rate": 7.306705753023086e-06, "loss": 0.0063, "step": 98000 }, { "epoch": 35.573932092004384, "grad_norm": 0.06692805141210556, "learning_rate": 7.304873580065959e-06, "loss": 0.0065, "step": 98050 }, { "epoch": 35.592186929536325, "grad_norm": 0.06786669045686722, "learning_rate": 7.303041407108832e-06, "loss": 0.0061, "step": 98100 }, { "epoch": 35.610441767068274, "grad_norm": 0.08806033432483673, "learning_rate": 7.301209234151705e-06, "loss": 0.0061, "step": 98150 }, { "epoch": 35.62869660460022, "grad_norm": 0.06242065131664276, "learning_rate": 7.299377061194577e-06, "loss": 0.0065, "step": 98200 }, { "epoch": 35.64695144213216, "grad_norm": 0.08658912777900696, "learning_rate": 7.29754488823745e-06, "loss": 0.0062, "step": 98250 }, { "epoch": 35.66520627966411, "grad_norm": 0.06975515931844711, "learning_rate": 7.295712715280323e-06, "loss": 0.0061, "step": 98300 }, { "epoch": 35.68346111719606, "grad_norm": 0.10880664736032486, "learning_rate": 7.293880542323196e-06, "loss": 0.0064, "step": 98350 }, { "epoch": 35.701715954728, "grad_norm": 0.0828244537115097, "learning_rate": 7.292048369366069e-06, "loss": 0.0061, "step": 98400 }, { "epoch": 35.71997079225995, "grad_norm": 0.08215038478374481, "learning_rate": 7.2902161964089416e-06, "loss": 0.0064, "step": 98450 }, { "epoch": 35.73822562979189, "grad_norm": 0.07113134860992432, "learning_rate": 7.288384023451815e-06, "loss": 0.0063, "step": 98500 }, { "epoch": 35.75648046732384, "grad_norm": 0.07428980618715286, "learning_rate": 7.286551850494688e-06, "loss": 0.0064, "step": 98550 }, { "epoch": 35.77473530485579, "grad_norm": 0.05782924219965935, "learning_rate": 7.284719677537561e-06, "loss": 0.0063, "step": 98600 }, { "epoch": 35.79299014238773, "grad_norm": 0.09295880049467087, "learning_rate": 7.2828875045804325e-06, "loss": 0.0061, "step": 98650 }, { "epoch": 35.81124497991968, "grad_norm": 0.0801381841301918, "learning_rate": 7.281055331623306e-06, "loss": 0.006, "step": 98700 }, { "epoch": 35.82949981745163, "grad_norm": 0.06812888383865356, "learning_rate": 7.279223158666179e-06, "loss": 0.0059, "step": 98750 }, { "epoch": 35.84775465498357, "grad_norm": 0.06402938812971115, "learning_rate": 7.277390985709052e-06, "loss": 0.0062, "step": 98800 }, { "epoch": 35.866009492515516, "grad_norm": 0.09592340886592865, "learning_rate": 7.275558812751924e-06, "loss": 0.0063, "step": 98850 }, { "epoch": 35.884264330047465, "grad_norm": 0.10497952997684479, "learning_rate": 7.273726639794797e-06, "loss": 0.0063, "step": 98900 }, { "epoch": 35.902519167579406, "grad_norm": 0.07185162603855133, "learning_rate": 7.27189446683767e-06, "loss": 0.0063, "step": 98950 }, { "epoch": 35.920774005111355, "grad_norm": 0.0840897187590599, "learning_rate": 7.270062293880543e-06, "loss": 0.006, "step": 99000 }, { "epoch": 35.9390288426433, "grad_norm": 0.13000895082950592, "learning_rate": 7.268230120923416e-06, "loss": 0.0064, "step": 99050 }, { "epoch": 35.957283680175244, "grad_norm": 0.07287903130054474, "learning_rate": 7.266397947966288e-06, "loss": 0.0062, "step": 99100 }, { "epoch": 35.97553851770719, "grad_norm": 0.08225014805793762, "learning_rate": 7.264565775009161e-06, "loss": 0.0064, "step": 99150 }, { "epoch": 35.99379335523914, "grad_norm": 0.07806675881147385, "learning_rate": 7.2627336020520346e-06, "loss": 0.0063, "step": 99200 }, { "epoch": 36.01204819277108, "grad_norm": 0.08090275526046753, "learning_rate": 7.260901429094908e-06, "loss": 0.0059, "step": 99250 }, { "epoch": 36.03030303030303, "grad_norm": 0.0847168043255806, "learning_rate": 7.259069256137779e-06, "loss": 0.0061, "step": 99300 }, { "epoch": 36.04855786783498, "grad_norm": 0.08830001950263977, "learning_rate": 7.257237083180652e-06, "loss": 0.0058, "step": 99350 }, { "epoch": 36.06681270536692, "grad_norm": 0.06367868185043335, "learning_rate": 7.2554049102235255e-06, "loss": 0.0062, "step": 99400 }, { "epoch": 36.08506754289887, "grad_norm": 0.08511219918727875, "learning_rate": 7.253572737266399e-06, "loss": 0.006, "step": 99450 }, { "epoch": 36.10332238043081, "grad_norm": 0.06778503209352493, "learning_rate": 7.251740564309271e-06, "loss": 0.0062, "step": 99500 }, { "epoch": 36.12157721796276, "grad_norm": 0.12021316587924957, "learning_rate": 7.249908391352144e-06, "loss": 0.0058, "step": 99550 }, { "epoch": 36.13983205549471, "grad_norm": 0.07956379652023315, "learning_rate": 7.248076218395017e-06, "loss": 0.006, "step": 99600 }, { "epoch": 36.15808689302665, "grad_norm": 0.0703323483467102, "learning_rate": 7.24624404543789e-06, "loss": 0.0061, "step": 99650 }, { "epoch": 36.1763417305586, "grad_norm": 0.06540928035974503, "learning_rate": 7.2444118724807635e-06, "loss": 0.0057, "step": 99700 }, { "epoch": 36.194596568090546, "grad_norm": 0.09054423868656158, "learning_rate": 7.242579699523635e-06, "loss": 0.0058, "step": 99750 }, { "epoch": 36.21285140562249, "grad_norm": 0.11251653730869293, "learning_rate": 7.240747526566508e-06, "loss": 0.0059, "step": 99800 }, { "epoch": 36.231106243154436, "grad_norm": 0.07803277671337128, "learning_rate": 7.238915353609381e-06, "loss": 0.0061, "step": 99850 }, { "epoch": 36.249361080686384, "grad_norm": 0.060590025037527084, "learning_rate": 7.2370831806522544e-06, "loss": 0.006, "step": 99900 }, { "epoch": 36.267615918218326, "grad_norm": 0.10745778679847717, "learning_rate": 7.235251007695127e-06, "loss": 0.0063, "step": 99950 }, { "epoch": 36.285870755750274, "grad_norm": 0.08331143110990524, "learning_rate": 7.233418834738e-06, "loss": 0.0059, "step": 100000 }, { "epoch": 36.30412559328222, "grad_norm": 0.07380392402410507, "learning_rate": 7.231586661780873e-06, "loss": 0.0059, "step": 100050 }, { "epoch": 36.322380430814164, "grad_norm": 0.13845571875572205, "learning_rate": 7.229754488823746e-06, "loss": 0.0062, "step": 100100 }, { "epoch": 36.34063526834611, "grad_norm": 0.073224738240242, "learning_rate": 7.227922315866618e-06, "loss": 0.0062, "step": 100150 }, { "epoch": 36.35889010587806, "grad_norm": 0.10338149964809418, "learning_rate": 7.226090142909491e-06, "loss": 0.0062, "step": 100200 }, { "epoch": 36.37714494341, "grad_norm": 0.05641770362854004, "learning_rate": 7.224257969952364e-06, "loss": 0.0059, "step": 100250 }, { "epoch": 36.39539978094195, "grad_norm": 0.07054747641086578, "learning_rate": 7.222425796995237e-06, "loss": 0.0061, "step": 100300 }, { "epoch": 36.4136546184739, "grad_norm": 0.08552581816911697, "learning_rate": 7.22059362403811e-06, "loss": 0.0059, "step": 100350 }, { "epoch": 36.43190945600584, "grad_norm": 0.08262912929058075, "learning_rate": 7.2187614510809825e-06, "loss": 0.0059, "step": 100400 }, { "epoch": 36.45016429353779, "grad_norm": 0.08468727022409439, "learning_rate": 7.216929278123856e-06, "loss": 0.0058, "step": 100450 }, { "epoch": 36.46841913106973, "grad_norm": 0.09433956444263458, "learning_rate": 7.215097105166729e-06, "loss": 0.0059, "step": 100500 }, { "epoch": 36.48667396860168, "grad_norm": 0.08190920203924179, "learning_rate": 7.213264932209601e-06, "loss": 0.0062, "step": 100550 }, { "epoch": 36.50492880613363, "grad_norm": 0.0770258978009224, "learning_rate": 7.2114327592524735e-06, "loss": 0.0062, "step": 100600 }, { "epoch": 36.52318364366557, "grad_norm": 0.09810082614421844, "learning_rate": 7.209600586295347e-06, "loss": 0.0061, "step": 100650 }, { "epoch": 36.54143848119752, "grad_norm": 0.06831865757703781, "learning_rate": 7.20776841333822e-06, "loss": 0.006, "step": 100700 }, { "epoch": 36.559693318729465, "grad_norm": 0.06771672517061234, "learning_rate": 7.205936240381093e-06, "loss": 0.006, "step": 100750 }, { "epoch": 36.57794815626141, "grad_norm": 0.10460344702005386, "learning_rate": 7.204104067423966e-06, "loss": 0.0062, "step": 100800 }, { "epoch": 36.596202993793355, "grad_norm": 0.06301269680261612, "learning_rate": 7.202271894466838e-06, "loss": 0.0061, "step": 100850 }, { "epoch": 36.6144578313253, "grad_norm": 0.07760588079690933, "learning_rate": 7.200439721509711e-06, "loss": 0.006, "step": 100900 }, { "epoch": 36.632712668857245, "grad_norm": 0.07670927792787552, "learning_rate": 7.198607548552584e-06, "loss": 0.006, "step": 100950 }, { "epoch": 36.65096750638919, "grad_norm": 0.0999051108956337, "learning_rate": 7.196775375595457e-06, "loss": 0.0059, "step": 101000 }, { "epoch": 36.66922234392114, "grad_norm": 0.084406778216362, "learning_rate": 7.194943202638329e-06, "loss": 0.0058, "step": 101050 }, { "epoch": 36.68747718145308, "grad_norm": 0.093657486140728, "learning_rate": 7.193111029681202e-06, "loss": 0.0059, "step": 101100 }, { "epoch": 36.70573201898503, "grad_norm": 0.06331384927034378, "learning_rate": 7.1912788567240756e-06, "loss": 0.0058, "step": 101150 }, { "epoch": 36.72398685651698, "grad_norm": 0.09734392911195755, "learning_rate": 7.189446683766949e-06, "loss": 0.0059, "step": 101200 }, { "epoch": 36.74224169404892, "grad_norm": 0.06250515580177307, "learning_rate": 7.18761451080982e-06, "loss": 0.0061, "step": 101250 }, { "epoch": 36.76049653158087, "grad_norm": 0.07793083041906357, "learning_rate": 7.185782337852693e-06, "loss": 0.0061, "step": 101300 }, { "epoch": 36.77875136911282, "grad_norm": 0.08349274098873138, "learning_rate": 7.1839501648955665e-06, "loss": 0.0062, "step": 101350 }, { "epoch": 36.79700620664476, "grad_norm": 0.07904528826475143, "learning_rate": 7.18211799193844e-06, "loss": 0.0063, "step": 101400 }, { "epoch": 36.81526104417671, "grad_norm": 0.09504704177379608, "learning_rate": 7.180285818981313e-06, "loss": 0.0059, "step": 101450 }, { "epoch": 36.83351588170865, "grad_norm": 0.080413818359375, "learning_rate": 7.178453646024185e-06, "loss": 0.0064, "step": 101500 }, { "epoch": 36.8517707192406, "grad_norm": 0.07363329082727432, "learning_rate": 7.176621473067058e-06, "loss": 0.0061, "step": 101550 }, { "epoch": 36.870025556772546, "grad_norm": 0.08098061382770538, "learning_rate": 7.174789300109931e-06, "loss": 0.0062, "step": 101600 }, { "epoch": 36.88828039430449, "grad_norm": 0.06893512606620789, "learning_rate": 7.1729571271528045e-06, "loss": 0.0062, "step": 101650 }, { "epoch": 36.906535231836436, "grad_norm": 0.08279791474342346, "learning_rate": 7.171124954195676e-06, "loss": 0.0061, "step": 101700 }, { "epoch": 36.924790069368385, "grad_norm": 0.11987020075321198, "learning_rate": 7.169292781238549e-06, "loss": 0.0062, "step": 101750 }, { "epoch": 36.943044906900326, "grad_norm": 0.12745238840579987, "learning_rate": 7.167460608281422e-06, "loss": 0.0061, "step": 101800 }, { "epoch": 36.961299744432274, "grad_norm": 0.0693376213312149, "learning_rate": 7.1656284353242954e-06, "loss": 0.0062, "step": 101850 }, { "epoch": 36.97955458196422, "grad_norm": 0.09628033638000488, "learning_rate": 7.163796262367168e-06, "loss": 0.0064, "step": 101900 }, { "epoch": 36.997809419496164, "grad_norm": 0.0808599665760994, "learning_rate": 7.161964089410041e-06, "loss": 0.0063, "step": 101950 }, { "epoch": 37.01606425702811, "grad_norm": 0.10416191816329956, "learning_rate": 7.160131916452914e-06, "loss": 0.0061, "step": 102000 }, { "epoch": 37.03431909456006, "grad_norm": 0.054223742336034775, "learning_rate": 7.158299743495787e-06, "loss": 0.0057, "step": 102050 }, { "epoch": 37.052573932092, "grad_norm": 0.0729273110628128, "learning_rate": 7.15646757053866e-06, "loss": 0.0058, "step": 102100 }, { "epoch": 37.07082876962395, "grad_norm": 0.09902309626340866, "learning_rate": 7.154635397581532e-06, "loss": 0.0057, "step": 102150 }, { "epoch": 37.0890836071559, "grad_norm": 0.07513562589883804, "learning_rate": 7.152803224624405e-06, "loss": 0.0058, "step": 102200 }, { "epoch": 37.10733844468784, "grad_norm": 0.08582492172718048, "learning_rate": 7.150971051667278e-06, "loss": 0.0058, "step": 102250 }, { "epoch": 37.12559328221979, "grad_norm": 0.0826055184006691, "learning_rate": 7.149138878710151e-06, "loss": 0.0057, "step": 102300 }, { "epoch": 37.14384811975174, "grad_norm": 0.09341729432344437, "learning_rate": 7.1473067057530235e-06, "loss": 0.0061, "step": 102350 }, { "epoch": 37.16210295728368, "grad_norm": 0.09186813235282898, "learning_rate": 7.145474532795897e-06, "loss": 0.0057, "step": 102400 }, { "epoch": 37.18035779481563, "grad_norm": 0.09176471829414368, "learning_rate": 7.143642359838769e-06, "loss": 0.006, "step": 102450 }, { "epoch": 37.19861263234757, "grad_norm": 0.1099560484290123, "learning_rate": 7.141810186881642e-06, "loss": 0.0057, "step": 102500 }, { "epoch": 37.21686746987952, "grad_norm": 0.10375861078500748, "learning_rate": 7.1399780139245144e-06, "loss": 0.0059, "step": 102550 }, { "epoch": 37.235122307411466, "grad_norm": 0.07808709144592285, "learning_rate": 7.138145840967388e-06, "loss": 0.0058, "step": 102600 }, { "epoch": 37.25337714494341, "grad_norm": 0.06815973669290543, "learning_rate": 7.136313668010261e-06, "loss": 0.006, "step": 102650 }, { "epoch": 37.271631982475355, "grad_norm": 0.08833569288253784, "learning_rate": 7.134481495053134e-06, "loss": 0.0062, "step": 102700 }, { "epoch": 37.289886820007304, "grad_norm": 0.07679002732038498, "learning_rate": 7.132649322096007e-06, "loss": 0.0058, "step": 102750 }, { "epoch": 37.308141657539245, "grad_norm": 0.08442166447639465, "learning_rate": 7.1308171491388785e-06, "loss": 0.0061, "step": 102800 }, { "epoch": 37.326396495071194, "grad_norm": 0.06310170888900757, "learning_rate": 7.128984976181752e-06, "loss": 0.0059, "step": 102850 }, { "epoch": 37.34465133260314, "grad_norm": 0.08076153695583344, "learning_rate": 7.127152803224625e-06, "loss": 0.0057, "step": 102900 }, { "epoch": 37.36290617013508, "grad_norm": 0.08187290281057358, "learning_rate": 7.125320630267498e-06, "loss": 0.0062, "step": 102950 }, { "epoch": 37.38116100766703, "grad_norm": 0.0739421471953392, "learning_rate": 7.12348845731037e-06, "loss": 0.0058, "step": 103000 }, { "epoch": 37.39941584519898, "grad_norm": 0.06982476264238358, "learning_rate": 7.121656284353243e-06, "loss": 0.0057, "step": 103050 }, { "epoch": 37.41767068273092, "grad_norm": 0.0734749510884285, "learning_rate": 7.1198241113961166e-06, "loss": 0.0058, "step": 103100 }, { "epoch": 37.43592552026287, "grad_norm": 0.14415009319782257, "learning_rate": 7.11799193843899e-06, "loss": 0.0058, "step": 103150 }, { "epoch": 37.45418035779482, "grad_norm": 0.07229702174663544, "learning_rate": 7.116159765481863e-06, "loss": 0.006, "step": 103200 }, { "epoch": 37.47243519532676, "grad_norm": 0.07358714938163757, "learning_rate": 7.114327592524734e-06, "loss": 0.0057, "step": 103250 }, { "epoch": 37.49069003285871, "grad_norm": 0.11026856303215027, "learning_rate": 7.1124954195676075e-06, "loss": 0.0061, "step": 103300 }, { "epoch": 37.50894487039066, "grad_norm": 0.11759543418884277, "learning_rate": 7.110663246610481e-06, "loss": 0.0061, "step": 103350 }, { "epoch": 37.5271997079226, "grad_norm": 0.10979145020246506, "learning_rate": 7.108831073653354e-06, "loss": 0.0058, "step": 103400 }, { "epoch": 37.54545454545455, "grad_norm": 0.07655245810747147, "learning_rate": 7.106998900696226e-06, "loss": 0.0059, "step": 103450 }, { "epoch": 37.56370938298649, "grad_norm": 0.08144721388816833, "learning_rate": 7.105166727739099e-06, "loss": 0.0058, "step": 103500 }, { "epoch": 37.58196422051844, "grad_norm": 0.11915040761232376, "learning_rate": 7.103334554781972e-06, "loss": 0.0056, "step": 103550 }, { "epoch": 37.600219058050385, "grad_norm": 0.06583862006664276, "learning_rate": 7.1015023818248455e-06, "loss": 0.0059, "step": 103600 }, { "epoch": 37.618473895582326, "grad_norm": 0.10096468776464462, "learning_rate": 7.099670208867717e-06, "loss": 0.0061, "step": 103650 }, { "epoch": 37.636728733114275, "grad_norm": 0.13261179625988007, "learning_rate": 7.09783803591059e-06, "loss": 0.0057, "step": 103700 }, { "epoch": 37.65498357064622, "grad_norm": 0.09574093669652939, "learning_rate": 7.096005862953463e-06, "loss": 0.0061, "step": 103750 }, { "epoch": 37.673238408178165, "grad_norm": 0.08889744430780411, "learning_rate": 7.094173689996336e-06, "loss": 0.006, "step": 103800 }, { "epoch": 37.69149324571011, "grad_norm": 0.07086731493473053, "learning_rate": 7.0923415170392096e-06, "loss": 0.0058, "step": 103850 }, { "epoch": 37.70974808324206, "grad_norm": 0.09308399260044098, "learning_rate": 7.090509344082082e-06, "loss": 0.0058, "step": 103900 }, { "epoch": 37.728002920774, "grad_norm": 0.1022915318608284, "learning_rate": 7.088677171124955e-06, "loss": 0.0057, "step": 103950 }, { "epoch": 37.74625775830595, "grad_norm": 0.09854364395141602, "learning_rate": 7.086844998167828e-06, "loss": 0.006, "step": 104000 }, { "epoch": 37.7645125958379, "grad_norm": 0.0795750841498375, "learning_rate": 7.0850128252107005e-06, "loss": 0.0057, "step": 104050 }, { "epoch": 37.78276743336984, "grad_norm": 0.07381213456392288, "learning_rate": 7.083180652253573e-06, "loss": 0.0058, "step": 104100 }, { "epoch": 37.80102227090179, "grad_norm": 0.09518029540777206, "learning_rate": 7.081348479296446e-06, "loss": 0.006, "step": 104150 }, { "epoch": 37.81927710843374, "grad_norm": 0.06364358216524124, "learning_rate": 7.079516306339319e-06, "loss": 0.006, "step": 104200 }, { "epoch": 37.83753194596568, "grad_norm": 0.0636284202337265, "learning_rate": 7.077684133382192e-06, "loss": 0.0058, "step": 104250 }, { "epoch": 37.85578678349763, "grad_norm": 0.07945403456687927, "learning_rate": 7.0758519604250645e-06, "loss": 0.0059, "step": 104300 }, { "epoch": 37.874041621029576, "grad_norm": 0.07059254497289658, "learning_rate": 7.074019787467938e-06, "loss": 0.0057, "step": 104350 }, { "epoch": 37.89229645856152, "grad_norm": 0.10135543346405029, "learning_rate": 7.07218761451081e-06, "loss": 0.0058, "step": 104400 }, { "epoch": 37.910551296093466, "grad_norm": 0.078563392162323, "learning_rate": 7.070355441553683e-06, "loss": 0.006, "step": 104450 }, { "epoch": 37.92880613362541, "grad_norm": 0.0885709822177887, "learning_rate": 7.068523268596556e-06, "loss": 0.0058, "step": 104500 }, { "epoch": 37.947060971157356, "grad_norm": 0.2267555594444275, "learning_rate": 7.066691095639429e-06, "loss": 0.0058, "step": 104550 }, { "epoch": 37.965315808689304, "grad_norm": 0.0938604325056076, "learning_rate": 7.064858922682302e-06, "loss": 0.0061, "step": 104600 }, { "epoch": 37.983570646221246, "grad_norm": 0.10156997293233871, "learning_rate": 7.063026749725175e-06, "loss": 0.006, "step": 104650 }, { "epoch": 38.001825483753194, "grad_norm": 0.07141386717557907, "learning_rate": 7.061194576768048e-06, "loss": 0.006, "step": 104700 }, { "epoch": 38.02008032128514, "grad_norm": 0.07558689266443253, "learning_rate": 7.0593624038109195e-06, "loss": 0.0052, "step": 104750 }, { "epoch": 38.038335158817084, "grad_norm": 0.10832683742046356, "learning_rate": 7.057530230853793e-06, "loss": 0.0055, "step": 104800 }, { "epoch": 38.05658999634903, "grad_norm": 0.10129398852586746, "learning_rate": 7.055698057896666e-06, "loss": 0.0055, "step": 104850 }, { "epoch": 38.07484483388098, "grad_norm": 0.09860972315073013, "learning_rate": 7.053865884939539e-06, "loss": 0.0057, "step": 104900 }, { "epoch": 38.09309967141292, "grad_norm": 0.07783135026693344, "learning_rate": 7.052033711982412e-06, "loss": 0.0056, "step": 104950 }, { "epoch": 38.11135450894487, "grad_norm": 0.08689619600772858, "learning_rate": 7.050201539025284e-06, "loss": 0.0059, "step": 105000 }, { "epoch": 38.12960934647682, "grad_norm": 0.07707160711288452, "learning_rate": 7.0483693660681575e-06, "loss": 0.0057, "step": 105050 }, { "epoch": 38.14786418400876, "grad_norm": 0.09588942676782608, "learning_rate": 7.046537193111031e-06, "loss": 0.0058, "step": 105100 }, { "epoch": 38.16611902154071, "grad_norm": 0.0914759412407875, "learning_rate": 7.044705020153904e-06, "loss": 0.0063, "step": 105150 }, { "epoch": 38.18437385907266, "grad_norm": 0.05774582177400589, "learning_rate": 7.042872847196775e-06, "loss": 0.0059, "step": 105200 }, { "epoch": 38.2026286966046, "grad_norm": 0.13030464947223663, "learning_rate": 7.0410406742396485e-06, "loss": 0.0059, "step": 105250 }, { "epoch": 38.22088353413655, "grad_norm": 0.0963662788271904, "learning_rate": 7.039208501282522e-06, "loss": 0.0057, "step": 105300 }, { "epoch": 38.239138371668496, "grad_norm": 0.11424513161182404, "learning_rate": 7.037376328325395e-06, "loss": 0.0059, "step": 105350 }, { "epoch": 38.25739320920044, "grad_norm": 0.08271259814500809, "learning_rate": 7.035544155368267e-06, "loss": 0.0053, "step": 105400 }, { "epoch": 38.275648046732385, "grad_norm": 0.05827345699071884, "learning_rate": 7.03371198241114e-06, "loss": 0.0056, "step": 105450 }, { "epoch": 38.29390288426433, "grad_norm": 0.07844190299510956, "learning_rate": 7.031879809454013e-06, "loss": 0.0058, "step": 105500 }, { "epoch": 38.312157721796275, "grad_norm": 0.07474935054779053, "learning_rate": 7.0300476364968865e-06, "loss": 0.0053, "step": 105550 }, { "epoch": 38.330412559328224, "grad_norm": 0.07028050720691681, "learning_rate": 7.02821546353976e-06, "loss": 0.0057, "step": 105600 }, { "epoch": 38.348667396860165, "grad_norm": 0.0745229497551918, "learning_rate": 7.026383290582631e-06, "loss": 0.0056, "step": 105650 }, { "epoch": 38.36692223439211, "grad_norm": 0.05398118868470192, "learning_rate": 7.024551117625504e-06, "loss": 0.0056, "step": 105700 }, { "epoch": 38.38517707192406, "grad_norm": 0.09702519327402115, "learning_rate": 7.022718944668377e-06, "loss": 0.0054, "step": 105750 }, { "epoch": 38.403431909456, "grad_norm": 0.08073694258928299, "learning_rate": 7.0208867717112506e-06, "loss": 0.0056, "step": 105800 }, { "epoch": 38.42168674698795, "grad_norm": 0.09093575924634933, "learning_rate": 7.019054598754123e-06, "loss": 0.0054, "step": 105850 }, { "epoch": 38.4399415845199, "grad_norm": 0.07572072744369507, "learning_rate": 7.017222425796996e-06, "loss": 0.0059, "step": 105900 }, { "epoch": 38.45819642205184, "grad_norm": 0.06884942203760147, "learning_rate": 7.015390252839869e-06, "loss": 0.0059, "step": 105950 }, { "epoch": 38.47645125958379, "grad_norm": 0.10885024070739746, "learning_rate": 7.0135580798827415e-06, "loss": 0.0055, "step": 106000 }, { "epoch": 38.49470609711574, "grad_norm": 0.0830949917435646, "learning_rate": 7.011725906925614e-06, "loss": 0.0055, "step": 106050 }, { "epoch": 38.51296093464768, "grad_norm": 0.09013762325048447, "learning_rate": 7.009893733968487e-06, "loss": 0.0056, "step": 106100 }, { "epoch": 38.53121577217963, "grad_norm": 0.09971994161605835, "learning_rate": 7.00806156101136e-06, "loss": 0.0057, "step": 106150 }, { "epoch": 38.54947060971158, "grad_norm": 0.09773129224777222, "learning_rate": 7.006229388054233e-06, "loss": 0.0058, "step": 106200 }, { "epoch": 38.56772544724352, "grad_norm": 0.07375184446573257, "learning_rate": 7.004397215097106e-06, "loss": 0.0056, "step": 106250 }, { "epoch": 38.58598028477547, "grad_norm": 0.07753311097621918, "learning_rate": 7.002565042139978e-06, "loss": 0.0059, "step": 106300 }, { "epoch": 38.604235122307415, "grad_norm": 0.10288874059915543, "learning_rate": 7.000732869182851e-06, "loss": 0.0055, "step": 106350 }, { "epoch": 38.622489959839356, "grad_norm": 0.08462907373905182, "learning_rate": 6.998900696225724e-06, "loss": 0.0057, "step": 106400 }, { "epoch": 38.640744797371305, "grad_norm": 0.08914110064506531, "learning_rate": 6.997068523268597e-06, "loss": 0.0056, "step": 106450 }, { "epoch": 38.658999634903246, "grad_norm": 0.08008231967687607, "learning_rate": 6.99523635031147e-06, "loss": 0.0061, "step": 106500 }, { "epoch": 38.677254472435195, "grad_norm": 0.09229998290538788, "learning_rate": 6.993404177354343e-06, "loss": 0.0056, "step": 106550 }, { "epoch": 38.69550930996714, "grad_norm": 0.09555326402187347, "learning_rate": 6.991572004397216e-06, "loss": 0.0056, "step": 106600 }, { "epoch": 38.713764147499084, "grad_norm": 0.08962292224168777, "learning_rate": 6.989739831440089e-06, "loss": 0.0061, "step": 106650 }, { "epoch": 38.73201898503103, "grad_norm": 0.10741361230611801, "learning_rate": 6.9879076584829605e-06, "loss": 0.0056, "step": 106700 }, { "epoch": 38.75027382256298, "grad_norm": 0.07065148651599884, "learning_rate": 6.986075485525834e-06, "loss": 0.0057, "step": 106750 }, { "epoch": 38.76852866009492, "grad_norm": 0.07458830624818802, "learning_rate": 6.984243312568707e-06, "loss": 0.0058, "step": 106800 }, { "epoch": 38.78678349762687, "grad_norm": 0.10816897451877594, "learning_rate": 6.98241113961158e-06, "loss": 0.0056, "step": 106850 }, { "epoch": 38.80503833515882, "grad_norm": 0.08799444884061813, "learning_rate": 6.980578966654453e-06, "loss": 0.0056, "step": 106900 }, { "epoch": 38.82329317269076, "grad_norm": 0.0818234235048294, "learning_rate": 6.978746793697325e-06, "loss": 0.0059, "step": 106950 }, { "epoch": 38.84154801022271, "grad_norm": 0.06351131200790405, "learning_rate": 6.9769146207401985e-06, "loss": 0.0059, "step": 107000 }, { "epoch": 38.85980284775466, "grad_norm": 0.0928465723991394, "learning_rate": 6.975082447783072e-06, "loss": 0.0057, "step": 107050 }, { "epoch": 38.8780576852866, "grad_norm": 0.09231691062450409, "learning_rate": 6.973250274825945e-06, "loss": 0.006, "step": 107100 }, { "epoch": 38.89631252281855, "grad_norm": 0.09538191556930542, "learning_rate": 6.971418101868816e-06, "loss": 0.0061, "step": 107150 }, { "epoch": 38.914567360350496, "grad_norm": 0.0805901437997818, "learning_rate": 6.9695859289116894e-06, "loss": 0.0056, "step": 107200 }, { "epoch": 38.93282219788244, "grad_norm": 0.09985378384590149, "learning_rate": 6.967753755954563e-06, "loss": 0.0058, "step": 107250 }, { "epoch": 38.951077035414386, "grad_norm": 0.06330256164073944, "learning_rate": 6.965921582997436e-06, "loss": 0.0064, "step": 107300 }, { "epoch": 38.969331872946334, "grad_norm": 0.06845296174287796, "learning_rate": 6.964089410040309e-06, "loss": 0.0057, "step": 107350 }, { "epoch": 38.987586710478276, "grad_norm": 0.09338348358869553, "learning_rate": 6.962257237083181e-06, "loss": 0.0059, "step": 107400 }, { "epoch": 39.005841548010224, "grad_norm": 0.11043407768011093, "learning_rate": 6.960425064126054e-06, "loss": 0.0055, "step": 107450 }, { "epoch": 39.024096385542165, "grad_norm": 0.10748768597841263, "learning_rate": 6.9585928911689275e-06, "loss": 0.0056, "step": 107500 }, { "epoch": 39.042351223074114, "grad_norm": 0.10336379706859589, "learning_rate": 6.9567607182118e-06, "loss": 0.0055, "step": 107550 }, { "epoch": 39.06060606060606, "grad_norm": 0.08351026475429535, "learning_rate": 6.954928545254672e-06, "loss": 0.0056, "step": 107600 }, { "epoch": 39.078860898138004, "grad_norm": 0.10501637309789658, "learning_rate": 6.953096372297545e-06, "loss": 0.0055, "step": 107650 }, { "epoch": 39.09711573566995, "grad_norm": 0.11452414095401764, "learning_rate": 6.951264199340418e-06, "loss": 0.0052, "step": 107700 }, { "epoch": 39.1153705732019, "grad_norm": 0.08445675671100616, "learning_rate": 6.9494320263832915e-06, "loss": 0.0057, "step": 107750 }, { "epoch": 39.13362541073384, "grad_norm": 0.0833323746919632, "learning_rate": 6.947599853426164e-06, "loss": 0.0054, "step": 107800 }, { "epoch": 39.15188024826579, "grad_norm": 0.07796288281679153, "learning_rate": 6.945767680469037e-06, "loss": 0.0051, "step": 107850 }, { "epoch": 39.17013508579774, "grad_norm": 0.09570194780826569, "learning_rate": 6.943935507511909e-06, "loss": 0.0057, "step": 107900 }, { "epoch": 39.18838992332968, "grad_norm": 0.08860123157501221, "learning_rate": 6.9421033345547825e-06, "loss": 0.0052, "step": 107950 }, { "epoch": 39.20664476086163, "grad_norm": 0.09336325526237488, "learning_rate": 6.940271161597656e-06, "loss": 0.0059, "step": 108000 }, { "epoch": 39.22489959839358, "grad_norm": 0.07894650101661682, "learning_rate": 6.938438988640528e-06, "loss": 0.0055, "step": 108050 }, { "epoch": 39.24315443592552, "grad_norm": 0.07193988561630249, "learning_rate": 6.936606815683401e-06, "loss": 0.0058, "step": 108100 }, { "epoch": 39.26140927345747, "grad_norm": 0.09570521116256714, "learning_rate": 6.934774642726274e-06, "loss": 0.0052, "step": 108150 }, { "epoch": 39.279664110989415, "grad_norm": 0.08886415511369705, "learning_rate": 6.932942469769147e-06, "loss": 0.0055, "step": 108200 }, { "epoch": 39.29791894852136, "grad_norm": 0.0973355621099472, "learning_rate": 6.931110296812019e-06, "loss": 0.0055, "step": 108250 }, { "epoch": 39.316173786053305, "grad_norm": 0.0875801146030426, "learning_rate": 6.929278123854892e-06, "loss": 0.0058, "step": 108300 }, { "epoch": 39.334428623585254, "grad_norm": 0.10668575018644333, "learning_rate": 6.927445950897765e-06, "loss": 0.0056, "step": 108350 }, { "epoch": 39.352683461117195, "grad_norm": 0.07983995974063873, "learning_rate": 6.925613777940638e-06, "loss": 0.0055, "step": 108400 }, { "epoch": 39.37093829864914, "grad_norm": 0.07771609723567963, "learning_rate": 6.9237816049835106e-06, "loss": 0.0055, "step": 108450 }, { "epoch": 39.389193136181085, "grad_norm": 0.05969242751598358, "learning_rate": 6.921949432026384e-06, "loss": 0.0052, "step": 108500 }, { "epoch": 39.40744797371303, "grad_norm": 0.10155881196260452, "learning_rate": 6.920117259069257e-06, "loss": 0.0053, "step": 108550 }, { "epoch": 39.42570281124498, "grad_norm": 0.09336396306753159, "learning_rate": 6.91828508611213e-06, "loss": 0.0055, "step": 108600 }, { "epoch": 39.44395764877692, "grad_norm": 0.07840098440647125, "learning_rate": 6.916452913155003e-06, "loss": 0.006, "step": 108650 }, { "epoch": 39.46221248630887, "grad_norm": 0.06942208111286163, "learning_rate": 6.914620740197875e-06, "loss": 0.0054, "step": 108700 }, { "epoch": 39.48046732384082, "grad_norm": 0.09940773993730545, "learning_rate": 6.912788567240748e-06, "loss": 0.0056, "step": 108750 }, { "epoch": 39.49872216137276, "grad_norm": 0.07615727186203003, "learning_rate": 6.910956394283621e-06, "loss": 0.0054, "step": 108800 }, { "epoch": 39.51697699890471, "grad_norm": 0.07418935745954514, "learning_rate": 6.909124221326494e-06, "loss": 0.0057, "step": 108850 }, { "epoch": 39.53523183643666, "grad_norm": 0.08329157531261444, "learning_rate": 6.907292048369366e-06, "loss": 0.0056, "step": 108900 }, { "epoch": 39.5534866739686, "grad_norm": 0.10732761770486832, "learning_rate": 6.9054598754122395e-06, "loss": 0.0052, "step": 108950 }, { "epoch": 39.57174151150055, "grad_norm": 0.12449579685926437, "learning_rate": 6.903627702455113e-06, "loss": 0.0055, "step": 109000 }, { "epoch": 39.589996349032496, "grad_norm": 0.0727342814207077, "learning_rate": 6.901795529497986e-06, "loss": 0.0056, "step": 109050 }, { "epoch": 39.60825118656444, "grad_norm": 0.09954434633255005, "learning_rate": 6.899963356540857e-06, "loss": 0.0053, "step": 109100 }, { "epoch": 39.626506024096386, "grad_norm": 0.09712584316730499, "learning_rate": 6.8981311835837304e-06, "loss": 0.0058, "step": 109150 }, { "epoch": 39.644760861628335, "grad_norm": 0.10896305739879608, "learning_rate": 6.896299010626604e-06, "loss": 0.0055, "step": 109200 }, { "epoch": 39.663015699160276, "grad_norm": 0.08060245960950851, "learning_rate": 6.894466837669477e-06, "loss": 0.0057, "step": 109250 }, { "epoch": 39.681270536692224, "grad_norm": 0.13593117892742157, "learning_rate": 6.89263466471235e-06, "loss": 0.0056, "step": 109300 }, { "epoch": 39.699525374224166, "grad_norm": 0.08266246318817139, "learning_rate": 6.890802491755222e-06, "loss": 0.0052, "step": 109350 }, { "epoch": 39.717780211756114, "grad_norm": 0.07467716932296753, "learning_rate": 6.888970318798095e-06, "loss": 0.0057, "step": 109400 }, { "epoch": 39.73603504928806, "grad_norm": 0.10146960616111755, "learning_rate": 6.8871381458409685e-06, "loss": 0.0055, "step": 109450 }, { "epoch": 39.754289886820004, "grad_norm": 0.07878026366233826, "learning_rate": 6.885305972883841e-06, "loss": 0.0054, "step": 109500 }, { "epoch": 39.77254472435195, "grad_norm": 0.07913444936275482, "learning_rate": 6.883473799926713e-06, "loss": 0.0054, "step": 109550 }, { "epoch": 39.7907995618839, "grad_norm": 0.07976087927818298, "learning_rate": 6.881641626969586e-06, "loss": 0.0058, "step": 109600 }, { "epoch": 39.80905439941584, "grad_norm": 0.08506577461957932, "learning_rate": 6.879809454012459e-06, "loss": 0.0056, "step": 109650 }, { "epoch": 39.82730923694779, "grad_norm": 0.09638950228691101, "learning_rate": 6.8779772810553325e-06, "loss": 0.0061, "step": 109700 }, { "epoch": 39.84556407447974, "grad_norm": 0.10849089175462723, "learning_rate": 6.876145108098206e-06, "loss": 0.0057, "step": 109750 }, { "epoch": 39.86381891201168, "grad_norm": 0.06655406206846237, "learning_rate": 6.874312935141077e-06, "loss": 0.0058, "step": 109800 }, { "epoch": 39.88207374954363, "grad_norm": 0.10428298264741898, "learning_rate": 6.87248076218395e-06, "loss": 0.0057, "step": 109850 }, { "epoch": 39.90032858707558, "grad_norm": 0.08007663488388062, "learning_rate": 6.8706485892268234e-06, "loss": 0.0058, "step": 109900 }, { "epoch": 39.91858342460752, "grad_norm": 0.084381103515625, "learning_rate": 6.868816416269697e-06, "loss": 0.0056, "step": 109950 }, { "epoch": 39.93683826213947, "grad_norm": 0.07506196200847626, "learning_rate": 6.866984243312569e-06, "loss": 0.0056, "step": 110000 }, { "epoch": 39.955093099671416, "grad_norm": 0.09177088737487793, "learning_rate": 6.865152070355442e-06, "loss": 0.0056, "step": 110050 }, { "epoch": 39.97334793720336, "grad_norm": 0.11367994546890259, "learning_rate": 6.863319897398315e-06, "loss": 0.0057, "step": 110100 }, { "epoch": 39.991602774735306, "grad_norm": 0.08052871376276016, "learning_rate": 6.861487724441188e-06, "loss": 0.0054, "step": 110150 }, { "epoch": 40.009857612267254, "grad_norm": 0.09983289241790771, "learning_rate": 6.85965555148406e-06, "loss": 0.0055, "step": 110200 }, { "epoch": 40.028112449799195, "grad_norm": 0.1016269102692604, "learning_rate": 6.857823378526933e-06, "loss": 0.0054, "step": 110250 }, { "epoch": 40.046367287331144, "grad_norm": 0.09276389330625534, "learning_rate": 6.855991205569806e-06, "loss": 0.0053, "step": 110300 }, { "epoch": 40.06462212486309, "grad_norm": 0.1079629510641098, "learning_rate": 6.854159032612679e-06, "loss": 0.0054, "step": 110350 }, { "epoch": 40.082876962395034, "grad_norm": 0.0910111665725708, "learning_rate": 6.852326859655552e-06, "loss": 0.0054, "step": 110400 }, { "epoch": 40.10113179992698, "grad_norm": 0.07992445677518845, "learning_rate": 6.850494686698425e-06, "loss": 0.0054, "step": 110450 }, { "epoch": 40.11938663745892, "grad_norm": 0.11348450928926468, "learning_rate": 6.848662513741298e-06, "loss": 0.0057, "step": 110500 }, { "epoch": 40.13764147499087, "grad_norm": 0.07800240069627762, "learning_rate": 6.846830340784171e-06, "loss": 0.005, "step": 110550 }, { "epoch": 40.15589631252282, "grad_norm": 0.06344039738178253, "learning_rate": 6.844998167827044e-06, "loss": 0.0055, "step": 110600 }, { "epoch": 40.17415115005476, "grad_norm": 0.12245462089776993, "learning_rate": 6.843165994869916e-06, "loss": 0.0053, "step": 110650 }, { "epoch": 40.19240598758671, "grad_norm": 0.09026476740837097, "learning_rate": 6.841333821912789e-06, "loss": 0.0057, "step": 110700 }, { "epoch": 40.21066082511866, "grad_norm": 0.08638624846935272, "learning_rate": 6.839501648955662e-06, "loss": 0.0055, "step": 110750 }, { "epoch": 40.2289156626506, "grad_norm": 0.09534577280282974, "learning_rate": 6.837669475998535e-06, "loss": 0.0053, "step": 110800 }, { "epoch": 40.24717050018255, "grad_norm": 0.1316833198070526, "learning_rate": 6.835837303041407e-06, "loss": 0.0053, "step": 110850 }, { "epoch": 40.2654253377145, "grad_norm": 0.08303643763065338, "learning_rate": 6.8340051300842805e-06, "loss": 0.0053, "step": 110900 }, { "epoch": 40.28368017524644, "grad_norm": 0.09234058856964111, "learning_rate": 6.832172957127154e-06, "loss": 0.0055, "step": 110950 }, { "epoch": 40.30193501277839, "grad_norm": 0.07637283951044083, "learning_rate": 6.830340784170027e-06, "loss": 0.0057, "step": 111000 }, { "epoch": 40.320189850310335, "grad_norm": 0.09585154056549072, "learning_rate": 6.828508611212899e-06, "loss": 0.0057, "step": 111050 }, { "epoch": 40.338444687842276, "grad_norm": 0.08937186747789383, "learning_rate": 6.826676438255771e-06, "loss": 0.0054, "step": 111100 }, { "epoch": 40.356699525374225, "grad_norm": 0.09191879630088806, "learning_rate": 6.8248442652986446e-06, "loss": 0.0052, "step": 111150 }, { "epoch": 40.37495436290617, "grad_norm": 0.0663374811410904, "learning_rate": 6.823012092341518e-06, "loss": 0.0052, "step": 111200 }, { "epoch": 40.393209200438115, "grad_norm": 0.08493369072675705, "learning_rate": 6.821179919384391e-06, "loss": 0.0051, "step": 111250 }, { "epoch": 40.41146403797006, "grad_norm": 0.12028602510690689, "learning_rate": 6.819347746427263e-06, "loss": 0.0056, "step": 111300 }, { "epoch": 40.429718875502004, "grad_norm": 0.08799166232347488, "learning_rate": 6.817515573470136e-06, "loss": 0.0052, "step": 111350 }, { "epoch": 40.44797371303395, "grad_norm": 0.08820776641368866, "learning_rate": 6.815683400513009e-06, "loss": 0.0051, "step": 111400 }, { "epoch": 40.4662285505659, "grad_norm": 0.10800744593143463, "learning_rate": 6.813851227555882e-06, "loss": 0.0054, "step": 111450 }, { "epoch": 40.48448338809784, "grad_norm": 0.07887940108776093, "learning_rate": 6.812019054598755e-06, "loss": 0.0053, "step": 111500 }, { "epoch": 40.50273822562979, "grad_norm": 0.07590191066265106, "learning_rate": 6.810186881641627e-06, "loss": 0.0054, "step": 111550 }, { "epoch": 40.52099306316174, "grad_norm": 0.0931200459599495, "learning_rate": 6.8083547086845e-06, "loss": 0.005, "step": 111600 }, { "epoch": 40.53924790069368, "grad_norm": 0.08735565841197968, "learning_rate": 6.8065225357273735e-06, "loss": 0.0054, "step": 111650 }, { "epoch": 40.55750273822563, "grad_norm": 0.07800202071666718, "learning_rate": 6.804690362770247e-06, "loss": 0.0054, "step": 111700 }, { "epoch": 40.57575757575758, "grad_norm": 0.10140656679868698, "learning_rate": 6.802858189813118e-06, "loss": 0.0051, "step": 111750 }, { "epoch": 40.59401241328952, "grad_norm": 0.1425938457250595, "learning_rate": 6.801026016855991e-06, "loss": 0.0054, "step": 111800 }, { "epoch": 40.61226725082147, "grad_norm": 0.10176616162061691, "learning_rate": 6.7991938438988644e-06, "loss": 0.0053, "step": 111850 }, { "epoch": 40.630522088353416, "grad_norm": 0.10816904902458191, "learning_rate": 6.797361670941738e-06, "loss": 0.0054, "step": 111900 }, { "epoch": 40.64877692588536, "grad_norm": 0.1073964387178421, "learning_rate": 6.79552949798461e-06, "loss": 0.0053, "step": 111950 }, { "epoch": 40.667031763417306, "grad_norm": 0.09485726803541183, "learning_rate": 6.793697325027483e-06, "loss": 0.0056, "step": 112000 }, { "epoch": 40.685286600949254, "grad_norm": 0.08419659733772278, "learning_rate": 6.791865152070356e-06, "loss": 0.0052, "step": 112050 }, { "epoch": 40.703541438481196, "grad_norm": 0.11649110168218613, "learning_rate": 6.790032979113229e-06, "loss": 0.0053, "step": 112100 }, { "epoch": 40.721796276013144, "grad_norm": 0.09259484708309174, "learning_rate": 6.7882008061561025e-06, "loss": 0.0056, "step": 112150 }, { "epoch": 40.74005111354509, "grad_norm": 0.08090998977422714, "learning_rate": 6.786368633198974e-06, "loss": 0.0054, "step": 112200 }, { "epoch": 40.758305951077034, "grad_norm": 0.09317684173583984, "learning_rate": 6.784536460241847e-06, "loss": 0.0054, "step": 112250 }, { "epoch": 40.77656078860898, "grad_norm": 0.10803469270467758, "learning_rate": 6.78270428728472e-06, "loss": 0.0053, "step": 112300 }, { "epoch": 40.79481562614093, "grad_norm": 0.07808904349803925, "learning_rate": 6.780872114327593e-06, "loss": 0.0053, "step": 112350 }, { "epoch": 40.81307046367287, "grad_norm": 0.09340313822031021, "learning_rate": 6.779039941370466e-06, "loss": 0.0055, "step": 112400 }, { "epoch": 40.83132530120482, "grad_norm": 0.08585655689239502, "learning_rate": 6.777207768413339e-06, "loss": 0.0054, "step": 112450 }, { "epoch": 40.84958013873676, "grad_norm": 0.08001890033483505, "learning_rate": 6.775375595456212e-06, "loss": 0.0054, "step": 112500 }, { "epoch": 40.86783497626871, "grad_norm": 0.13029362261295319, "learning_rate": 6.773543422499085e-06, "loss": 0.0054, "step": 112550 }, { "epoch": 40.88608981380066, "grad_norm": 0.0936351791024208, "learning_rate": 6.771711249541957e-06, "loss": 0.0057, "step": 112600 }, { "epoch": 40.9043446513326, "grad_norm": 0.061671484261751175, "learning_rate": 6.76987907658483e-06, "loss": 0.0056, "step": 112650 }, { "epoch": 40.92259948886455, "grad_norm": 0.09203708171844482, "learning_rate": 6.768046903627703e-06, "loss": 0.0054, "step": 112700 }, { "epoch": 40.9408543263965, "grad_norm": 0.08330999314785004, "learning_rate": 6.766214730670576e-06, "loss": 0.0053, "step": 112750 }, { "epoch": 40.95910916392844, "grad_norm": 0.12374526262283325, "learning_rate": 6.764382557713449e-06, "loss": 0.0055, "step": 112800 }, { "epoch": 40.97736400146039, "grad_norm": 0.16492806375026703, "learning_rate": 6.7625503847563215e-06, "loss": 0.0056, "step": 112850 }, { "epoch": 40.995618838992335, "grad_norm": 0.09608200192451477, "learning_rate": 6.760718211799195e-06, "loss": 0.0054, "step": 112900 }, { "epoch": 41.01387367652428, "grad_norm": 0.08785369992256165, "learning_rate": 6.758886038842068e-06, "loss": 0.0053, "step": 112950 }, { "epoch": 41.032128514056225, "grad_norm": 0.10524538159370422, "learning_rate": 6.75705386588494e-06, "loss": 0.0051, "step": 113000 }, { "epoch": 41.050383351588174, "grad_norm": 0.11223378777503967, "learning_rate": 6.755221692927812e-06, "loss": 0.0053, "step": 113050 }, { "epoch": 41.068638189120115, "grad_norm": 0.09873943775892258, "learning_rate": 6.7533895199706856e-06, "loss": 0.0049, "step": 113100 }, { "epoch": 41.086893026652064, "grad_norm": 0.09510210901498795, "learning_rate": 6.751557347013559e-06, "loss": 0.005, "step": 113150 }, { "epoch": 41.10514786418401, "grad_norm": 0.10416844487190247, "learning_rate": 6.749725174056432e-06, "loss": 0.0054, "step": 113200 }, { "epoch": 41.12340270171595, "grad_norm": 0.08831951767206192, "learning_rate": 6.747893001099304e-06, "loss": 0.0052, "step": 113250 }, { "epoch": 41.1416575392479, "grad_norm": 0.12266934663057327, "learning_rate": 6.7460608281421765e-06, "loss": 0.0052, "step": 113300 }, { "epoch": 41.15991237677984, "grad_norm": 0.06362644582986832, "learning_rate": 6.74422865518505e-06, "loss": 0.0052, "step": 113350 }, { "epoch": 41.17816721431179, "grad_norm": 0.12486229836940765, "learning_rate": 6.742396482227923e-06, "loss": 0.005, "step": 113400 }, { "epoch": 41.19642205184374, "grad_norm": 0.10706956684589386, "learning_rate": 6.740564309270796e-06, "loss": 0.0052, "step": 113450 }, { "epoch": 41.21467688937568, "grad_norm": 0.10666733980178833, "learning_rate": 6.738732136313668e-06, "loss": 0.0051, "step": 113500 }, { "epoch": 41.23293172690763, "grad_norm": 0.07920535653829575, "learning_rate": 6.736899963356541e-06, "loss": 0.0053, "step": 113550 }, { "epoch": 41.25118656443958, "grad_norm": 0.08798008412122726, "learning_rate": 6.7350677903994145e-06, "loss": 0.0053, "step": 113600 }, { "epoch": 41.26944140197152, "grad_norm": 0.0852372869849205, "learning_rate": 6.733235617442288e-06, "loss": 0.0051, "step": 113650 }, { "epoch": 41.28769623950347, "grad_norm": 0.0948399230837822, "learning_rate": 6.731403444485159e-06, "loss": 0.005, "step": 113700 }, { "epoch": 41.30595107703542, "grad_norm": 0.06313244998455048, "learning_rate": 6.729571271528032e-06, "loss": 0.005, "step": 113750 }, { "epoch": 41.32420591456736, "grad_norm": 0.09295839816331863, "learning_rate": 6.7277390985709054e-06, "loss": 0.0052, "step": 113800 }, { "epoch": 41.342460752099306, "grad_norm": 0.10179838538169861, "learning_rate": 6.725906925613779e-06, "loss": 0.0053, "step": 113850 }, { "epoch": 41.360715589631255, "grad_norm": 0.10780146718025208, "learning_rate": 6.724074752656652e-06, "loss": 0.0055, "step": 113900 }, { "epoch": 41.378970427163196, "grad_norm": 0.09591986984014511, "learning_rate": 6.722242579699524e-06, "loss": 0.0052, "step": 113950 }, { "epoch": 41.397225264695145, "grad_norm": 0.10673645883798599, "learning_rate": 6.720410406742397e-06, "loss": 0.0052, "step": 114000 }, { "epoch": 41.41548010222709, "grad_norm": 0.06676339358091354, "learning_rate": 6.71857823378527e-06, "loss": 0.0054, "step": 114050 }, { "epoch": 41.433734939759034, "grad_norm": 0.11513575166463852, "learning_rate": 6.7167460608281435e-06, "loss": 0.0051, "step": 114100 }, { "epoch": 41.45198977729098, "grad_norm": 0.10171222686767578, "learning_rate": 6.714913887871015e-06, "loss": 0.0051, "step": 114150 }, { "epoch": 41.47024461482293, "grad_norm": 0.09004172682762146, "learning_rate": 6.713081714913888e-06, "loss": 0.0053, "step": 114200 }, { "epoch": 41.48849945235487, "grad_norm": 0.11554663628339767, "learning_rate": 6.711249541956761e-06, "loss": 0.0051, "step": 114250 }, { "epoch": 41.50675428988682, "grad_norm": 0.10601811856031418, "learning_rate": 6.709417368999634e-06, "loss": 0.0053, "step": 114300 }, { "epoch": 41.52500912741876, "grad_norm": 0.07002589106559753, "learning_rate": 6.707585196042507e-06, "loss": 0.0054, "step": 114350 }, { "epoch": 41.54326396495071, "grad_norm": 0.0755009725689888, "learning_rate": 6.70575302308538e-06, "loss": 0.0053, "step": 114400 }, { "epoch": 41.56151880248266, "grad_norm": 0.1059579998254776, "learning_rate": 6.703920850128253e-06, "loss": 0.0053, "step": 114450 }, { "epoch": 41.5797736400146, "grad_norm": 0.09456472098827362, "learning_rate": 6.702088677171126e-06, "loss": 0.0048, "step": 114500 }, { "epoch": 41.59802847754655, "grad_norm": 0.11582711338996887, "learning_rate": 6.700256504213999e-06, "loss": 0.0053, "step": 114550 }, { "epoch": 41.6162833150785, "grad_norm": 0.07862074673175812, "learning_rate": 6.698424331256871e-06, "loss": 0.0053, "step": 114600 }, { "epoch": 41.63453815261044, "grad_norm": 0.08456604182720184, "learning_rate": 6.696592158299744e-06, "loss": 0.0052, "step": 114650 }, { "epoch": 41.65279299014239, "grad_norm": 0.07232081890106201, "learning_rate": 6.694759985342617e-06, "loss": 0.0053, "step": 114700 }, { "epoch": 41.671047827674336, "grad_norm": 0.08778981864452362, "learning_rate": 6.69292781238549e-06, "loss": 0.0052, "step": 114750 }, { "epoch": 41.68930266520628, "grad_norm": 0.11182025820016861, "learning_rate": 6.6910956394283625e-06, "loss": 0.005, "step": 114800 }, { "epoch": 41.707557502738226, "grad_norm": 0.11483198404312134, "learning_rate": 6.689263466471236e-06, "loss": 0.0053, "step": 114850 }, { "epoch": 41.725812340270174, "grad_norm": 0.09406556189060211, "learning_rate": 6.687431293514108e-06, "loss": 0.0054, "step": 114900 }, { "epoch": 41.744067177802116, "grad_norm": 0.09629793465137482, "learning_rate": 6.685599120556981e-06, "loss": 0.0052, "step": 114950 }, { "epoch": 41.762322015334064, "grad_norm": 0.12486808747053146, "learning_rate": 6.683766947599853e-06, "loss": 0.0052, "step": 115000 }, { "epoch": 41.78057685286601, "grad_norm": 0.0869254618883133, "learning_rate": 6.6819347746427266e-06, "loss": 0.0056, "step": 115050 }, { "epoch": 41.798831690397954, "grad_norm": 0.06979994475841522, "learning_rate": 6.6801026016856e-06, "loss": 0.0053, "step": 115100 }, { "epoch": 41.8170865279299, "grad_norm": 0.09904607385396957, "learning_rate": 6.678270428728473e-06, "loss": 0.0055, "step": 115150 }, { "epoch": 41.83534136546185, "grad_norm": 0.10609513521194458, "learning_rate": 6.676438255771346e-06, "loss": 0.0054, "step": 115200 }, { "epoch": 41.85359620299379, "grad_norm": 0.0911189541220665, "learning_rate": 6.6746060828142175e-06, "loss": 0.0055, "step": 115250 }, { "epoch": 41.87185104052574, "grad_norm": 0.0792144313454628, "learning_rate": 6.672773909857091e-06, "loss": 0.0055, "step": 115300 }, { "epoch": 41.89010587805768, "grad_norm": 0.1093355193734169, "learning_rate": 6.670941736899964e-06, "loss": 0.0051, "step": 115350 }, { "epoch": 41.90836071558963, "grad_norm": 0.06736687570810318, "learning_rate": 6.669109563942837e-06, "loss": 0.0053, "step": 115400 }, { "epoch": 41.92661555312158, "grad_norm": 0.09530867636203766, "learning_rate": 6.667277390985709e-06, "loss": 0.005, "step": 115450 }, { "epoch": 41.94487039065352, "grad_norm": 0.09418410062789917, "learning_rate": 6.665445218028582e-06, "loss": 0.0053, "step": 115500 }, { "epoch": 41.96312522818547, "grad_norm": 0.08674073219299316, "learning_rate": 6.6636130450714555e-06, "loss": 0.0049, "step": 115550 }, { "epoch": 41.98138006571742, "grad_norm": 0.05825132504105568, "learning_rate": 6.661780872114329e-06, "loss": 0.0051, "step": 115600 }, { "epoch": 41.99963490324936, "grad_norm": 0.08353707194328308, "learning_rate": 6.6599486991572e-06, "loss": 0.0055, "step": 115650 }, { "epoch": 42.01788974078131, "grad_norm": 0.09194663912057877, "learning_rate": 6.658116526200073e-06, "loss": 0.0049, "step": 115700 }, { "epoch": 42.036144578313255, "grad_norm": 0.09772255271673203, "learning_rate": 6.656284353242946e-06, "loss": 0.0053, "step": 115750 }, { "epoch": 42.0543994158452, "grad_norm": 0.07787705212831497, "learning_rate": 6.6544521802858196e-06, "loss": 0.0048, "step": 115800 }, { "epoch": 42.072654253377145, "grad_norm": 0.1105140969157219, "learning_rate": 6.652620007328693e-06, "loss": 0.0048, "step": 115850 }, { "epoch": 42.09090909090909, "grad_norm": 0.08834534138441086, "learning_rate": 6.650787834371565e-06, "loss": 0.005, "step": 115900 }, { "epoch": 42.109163928441035, "grad_norm": 0.11384642869234085, "learning_rate": 6.648955661414438e-06, "loss": 0.005, "step": 115950 }, { "epoch": 42.12741876597298, "grad_norm": 0.073062963783741, "learning_rate": 6.647123488457311e-06, "loss": 0.0048, "step": 116000 }, { "epoch": 42.14567360350493, "grad_norm": 0.08729217201471329, "learning_rate": 6.6452913155001845e-06, "loss": 0.0048, "step": 116050 }, { "epoch": 42.16392844103687, "grad_norm": 0.08440303057432175, "learning_rate": 6.643459142543056e-06, "loss": 0.0049, "step": 116100 }, { "epoch": 42.18218327856882, "grad_norm": 0.08654425293207169, "learning_rate": 6.641626969585929e-06, "loss": 0.005, "step": 116150 }, { "epoch": 42.20043811610077, "grad_norm": 0.0675879493355751, "learning_rate": 6.639794796628802e-06, "loss": 0.005, "step": 116200 }, { "epoch": 42.21869295363271, "grad_norm": 0.08536078035831451, "learning_rate": 6.637962623671675e-06, "loss": 0.005, "step": 116250 }, { "epoch": 42.23694779116466, "grad_norm": 0.13917163014411926, "learning_rate": 6.6361304507145485e-06, "loss": 0.0052, "step": 116300 }, { "epoch": 42.2552026286966, "grad_norm": 0.13478752970695496, "learning_rate": 6.634298277757421e-06, "loss": 0.0051, "step": 116350 }, { "epoch": 42.27345746622855, "grad_norm": 0.07896340638399124, "learning_rate": 6.632466104800294e-06, "loss": 0.0051, "step": 116400 }, { "epoch": 42.2917123037605, "grad_norm": 0.10201392322778702, "learning_rate": 6.630633931843167e-06, "loss": 0.0051, "step": 116450 }, { "epoch": 42.30996714129244, "grad_norm": 0.12439602613449097, "learning_rate": 6.6288017588860394e-06, "loss": 0.0051, "step": 116500 }, { "epoch": 42.32822197882439, "grad_norm": 0.0901709496974945, "learning_rate": 6.626969585928912e-06, "loss": 0.0053, "step": 116550 }, { "epoch": 42.346476816356336, "grad_norm": 0.09743001312017441, "learning_rate": 6.625137412971785e-06, "loss": 0.005, "step": 116600 }, { "epoch": 42.36473165388828, "grad_norm": 0.07297075539827347, "learning_rate": 6.623305240014658e-06, "loss": 0.0052, "step": 116650 }, { "epoch": 42.382986491420226, "grad_norm": 0.1150427907705307, "learning_rate": 6.621473067057531e-06, "loss": 0.0051, "step": 116700 }, { "epoch": 42.401241328952175, "grad_norm": 0.09779272973537445, "learning_rate": 6.6196408941004035e-06, "loss": 0.0053, "step": 116750 }, { "epoch": 42.419496166484116, "grad_norm": 0.11598919332027435, "learning_rate": 6.617808721143276e-06, "loss": 0.0049, "step": 116800 }, { "epoch": 42.437751004016064, "grad_norm": 0.1324910819530487, "learning_rate": 6.615976548186149e-06, "loss": 0.005, "step": 116850 }, { "epoch": 42.45600584154801, "grad_norm": 0.07964885234832764, "learning_rate": 6.614144375229022e-06, "loss": 0.0049, "step": 116900 }, { "epoch": 42.474260679079954, "grad_norm": 0.12287833541631699, "learning_rate": 6.612312202271895e-06, "loss": 0.005, "step": 116950 }, { "epoch": 42.4925155166119, "grad_norm": 0.08588016033172607, "learning_rate": 6.6104800293147675e-06, "loss": 0.0049, "step": 117000 }, { "epoch": 42.51077035414385, "grad_norm": 0.098282590508461, "learning_rate": 6.608647856357641e-06, "loss": 0.0054, "step": 117050 }, { "epoch": 42.52902519167579, "grad_norm": 0.08994124084711075, "learning_rate": 6.606815683400514e-06, "loss": 0.005, "step": 117100 }, { "epoch": 42.54728002920774, "grad_norm": 0.0671117901802063, "learning_rate": 6.604983510443387e-06, "loss": 0.0053, "step": 117150 }, { "epoch": 42.56553486673969, "grad_norm": 0.09800095111131668, "learning_rate": 6.6031513374862585e-06, "loss": 0.0051, "step": 117200 }, { "epoch": 42.58378970427163, "grad_norm": 0.05866605415940285, "learning_rate": 6.601319164529132e-06, "loss": 0.005, "step": 117250 }, { "epoch": 42.60204454180358, "grad_norm": 0.08472153544425964, "learning_rate": 6.599486991572005e-06, "loss": 0.0051, "step": 117300 }, { "epoch": 42.62029937933552, "grad_norm": 0.08061538636684418, "learning_rate": 6.597654818614878e-06, "loss": 0.0049, "step": 117350 }, { "epoch": 42.63855421686747, "grad_norm": 0.06918776780366898, "learning_rate": 6.59582264565775e-06, "loss": 0.0053, "step": 117400 }, { "epoch": 42.65680905439942, "grad_norm": 0.08320360630750656, "learning_rate": 6.593990472700623e-06, "loss": 0.0051, "step": 117450 }, { "epoch": 42.67506389193136, "grad_norm": 0.1047106385231018, "learning_rate": 6.5921582997434965e-06, "loss": 0.0051, "step": 117500 }, { "epoch": 42.69331872946331, "grad_norm": 0.11748748272657394, "learning_rate": 6.59032612678637e-06, "loss": 0.0049, "step": 117550 }, { "epoch": 42.711573566995256, "grad_norm": 0.08307626843452454, "learning_rate": 6.588493953829243e-06, "loss": 0.0048, "step": 117600 }, { "epoch": 42.7298284045272, "grad_norm": 0.09040186554193497, "learning_rate": 6.586661780872114e-06, "loss": 0.005, "step": 117650 }, { "epoch": 42.748083242059145, "grad_norm": 0.08989427238702774, "learning_rate": 6.584829607914987e-06, "loss": 0.0055, "step": 117700 }, { "epoch": 42.766338079591094, "grad_norm": 0.10779829323291779, "learning_rate": 6.5829974349578606e-06, "loss": 0.0051, "step": 117750 }, { "epoch": 42.784592917123035, "grad_norm": 0.08712311834096909, "learning_rate": 6.581165262000734e-06, "loss": 0.0057, "step": 117800 }, { "epoch": 42.802847754654984, "grad_norm": 0.0907873660326004, "learning_rate": 6.579333089043606e-06, "loss": 0.0051, "step": 117850 }, { "epoch": 42.82110259218693, "grad_norm": 0.07150555402040482, "learning_rate": 6.577500916086479e-06, "loss": 0.005, "step": 117900 }, { "epoch": 42.83935742971887, "grad_norm": 0.09530286490917206, "learning_rate": 6.575668743129352e-06, "loss": 0.0051, "step": 117950 }, { "epoch": 42.85761226725082, "grad_norm": 0.09744761884212494, "learning_rate": 6.5738365701722255e-06, "loss": 0.0053, "step": 118000 }, { "epoch": 42.87586710478277, "grad_norm": 0.10499196499586105, "learning_rate": 6.572004397215099e-06, "loss": 0.0052, "step": 118050 }, { "epoch": 42.89412194231471, "grad_norm": 0.0927620530128479, "learning_rate": 6.57017222425797e-06, "loss": 0.005, "step": 118100 }, { "epoch": 42.91237677984666, "grad_norm": 0.11338216811418533, "learning_rate": 6.568340051300843e-06, "loss": 0.0051, "step": 118150 }, { "epoch": 42.93063161737861, "grad_norm": 0.10481079667806625, "learning_rate": 6.566507878343716e-06, "loss": 0.0047, "step": 118200 }, { "epoch": 42.94888645491055, "grad_norm": 0.11322116106748581, "learning_rate": 6.5646757053865895e-06, "loss": 0.0053, "step": 118250 }, { "epoch": 42.9671412924425, "grad_norm": 0.09435020387172699, "learning_rate": 6.562843532429462e-06, "loss": 0.0053, "step": 118300 }, { "epoch": 42.98539612997444, "grad_norm": 0.1249832957983017, "learning_rate": 6.561011359472335e-06, "loss": 0.0051, "step": 118350 }, { "epoch": 43.00365096750639, "grad_norm": 0.0877332091331482, "learning_rate": 6.559179186515207e-06, "loss": 0.0051, "step": 118400 }, { "epoch": 43.02190580503834, "grad_norm": 0.07476967573165894, "learning_rate": 6.5573470135580804e-06, "loss": 0.0047, "step": 118450 }, { "epoch": 43.04016064257028, "grad_norm": 0.08570320904254913, "learning_rate": 6.555514840600953e-06, "loss": 0.005, "step": 118500 }, { "epoch": 43.05841548010223, "grad_norm": 0.09007178246974945, "learning_rate": 6.553682667643826e-06, "loss": 0.0045, "step": 118550 }, { "epoch": 43.076670317634175, "grad_norm": 0.09660544246435165, "learning_rate": 6.551850494686699e-06, "loss": 0.0054, "step": 118600 }, { "epoch": 43.094925155166116, "grad_norm": 0.11541305482387543, "learning_rate": 6.550018321729572e-06, "loss": 0.0048, "step": 118650 }, { "epoch": 43.113179992698065, "grad_norm": 0.10734861344099045, "learning_rate": 6.548186148772445e-06, "loss": 0.0051, "step": 118700 }, { "epoch": 43.13143483023001, "grad_norm": 0.09853185713291168, "learning_rate": 6.546353975815317e-06, "loss": 0.0049, "step": 118750 }, { "epoch": 43.149689667761955, "grad_norm": 0.08131727576255798, "learning_rate": 6.54452180285819e-06, "loss": 0.0046, "step": 118800 }, { "epoch": 43.1679445052939, "grad_norm": 0.09476389735937119, "learning_rate": 6.542689629901063e-06, "loss": 0.0047, "step": 118850 }, { "epoch": 43.18619934282585, "grad_norm": 0.09595343470573425, "learning_rate": 6.540857456943936e-06, "loss": 0.0048, "step": 118900 }, { "epoch": 43.20445418035779, "grad_norm": 0.08065979927778244, "learning_rate": 6.5390252839868085e-06, "loss": 0.0049, "step": 118950 }, { "epoch": 43.22270901788974, "grad_norm": 0.12298467755317688, "learning_rate": 6.537193111029682e-06, "loss": 0.0049, "step": 119000 }, { "epoch": 43.24096385542169, "grad_norm": 0.10628730803728104, "learning_rate": 6.535360938072555e-06, "loss": 0.0048, "step": 119050 }, { "epoch": 43.25921869295363, "grad_norm": 0.07544521242380142, "learning_rate": 6.533528765115428e-06, "loss": 0.0048, "step": 119100 }, { "epoch": 43.27747353048558, "grad_norm": 0.1205558255314827, "learning_rate": 6.5316965921582994e-06, "loss": 0.0049, "step": 119150 }, { "epoch": 43.29572836801753, "grad_norm": 0.07054748386144638, "learning_rate": 6.529864419201173e-06, "loss": 0.0047, "step": 119200 }, { "epoch": 43.31398320554947, "grad_norm": 0.0925961583852768, "learning_rate": 6.528032246244046e-06, "loss": 0.0045, "step": 119250 }, { "epoch": 43.33223804308142, "grad_norm": 0.1105981096625328, "learning_rate": 6.526200073286919e-06, "loss": 0.0049, "step": 119300 }, { "epoch": 43.35049288061336, "grad_norm": 0.08765171468257904, "learning_rate": 6.524367900329792e-06, "loss": 0.0047, "step": 119350 }, { "epoch": 43.36874771814531, "grad_norm": 0.07528301328420639, "learning_rate": 6.522535727372664e-06, "loss": 0.0049, "step": 119400 }, { "epoch": 43.387002555677256, "grad_norm": 0.082547627389431, "learning_rate": 6.5207035544155375e-06, "loss": 0.0054, "step": 119450 }, { "epoch": 43.4052573932092, "grad_norm": 0.09582000970840454, "learning_rate": 6.518871381458411e-06, "loss": 0.0052, "step": 119500 }, { "epoch": 43.423512230741146, "grad_norm": 0.10432495921850204, "learning_rate": 6.517039208501284e-06, "loss": 0.0049, "step": 119550 }, { "epoch": 43.441767068273094, "grad_norm": 0.10597340017557144, "learning_rate": 6.515207035544155e-06, "loss": 0.0051, "step": 119600 }, { "epoch": 43.460021905805036, "grad_norm": 0.08753450959920883, "learning_rate": 6.513374862587028e-06, "loss": 0.0048, "step": 119650 }, { "epoch": 43.478276743336984, "grad_norm": 0.10873019695281982, "learning_rate": 6.5115426896299016e-06, "loss": 0.0048, "step": 119700 }, { "epoch": 43.49653158086893, "grad_norm": 0.08238986879587173, "learning_rate": 6.509710516672775e-06, "loss": 0.0048, "step": 119750 }, { "epoch": 43.514786418400874, "grad_norm": 0.10150475054979324, "learning_rate": 6.507878343715647e-06, "loss": 0.0046, "step": 119800 }, { "epoch": 43.53304125593282, "grad_norm": 0.07756687700748444, "learning_rate": 6.50604617075852e-06, "loss": 0.0049, "step": 119850 }, { "epoch": 43.55129609346477, "grad_norm": 0.08646021783351898, "learning_rate": 6.504213997801393e-06, "loss": 0.0048, "step": 119900 }, { "epoch": 43.56955093099671, "grad_norm": 0.056060872972011566, "learning_rate": 6.5023818248442665e-06, "loss": 0.0048, "step": 119950 }, { "epoch": 43.58780576852866, "grad_norm": 0.07329832017421722, "learning_rate": 6.500549651887139e-06, "loss": 0.0052, "step": 120000 }, { "epoch": 43.60606060606061, "grad_norm": 0.08801274746656418, "learning_rate": 6.498717478930011e-06, "loss": 0.0051, "step": 120050 }, { "epoch": 43.62431544359255, "grad_norm": 0.11767558008432388, "learning_rate": 6.496885305972884e-06, "loss": 0.0048, "step": 120100 }, { "epoch": 43.6425702811245, "grad_norm": 0.09081073850393295, "learning_rate": 6.495053133015757e-06, "loss": 0.005, "step": 120150 }, { "epoch": 43.66082511865645, "grad_norm": 0.05393541604280472, "learning_rate": 6.4932209600586305e-06, "loss": 0.0047, "step": 120200 }, { "epoch": 43.67907995618839, "grad_norm": 0.15293632447719574, "learning_rate": 6.491388787101503e-06, "loss": 0.0049, "step": 120250 }, { "epoch": 43.69733479372034, "grad_norm": 0.07996521145105362, "learning_rate": 6.489556614144376e-06, "loss": 0.0049, "step": 120300 }, { "epoch": 43.71558963125228, "grad_norm": 0.08945640921592712, "learning_rate": 6.487724441187248e-06, "loss": 0.005, "step": 120350 }, { "epoch": 43.73384446878423, "grad_norm": 0.07870243489742279, "learning_rate": 6.485892268230121e-06, "loss": 0.005, "step": 120400 }, { "epoch": 43.752099306316175, "grad_norm": 0.1164698526263237, "learning_rate": 6.4840600952729946e-06, "loss": 0.0049, "step": 120450 }, { "epoch": 43.77035414384812, "grad_norm": 0.09787051379680634, "learning_rate": 6.482227922315867e-06, "loss": 0.0053, "step": 120500 }, { "epoch": 43.788608981380065, "grad_norm": 0.10332025587558746, "learning_rate": 6.48039574935874e-06, "loss": 0.0052, "step": 120550 }, { "epoch": 43.806863818912014, "grad_norm": 0.08075930178165436, "learning_rate": 6.478563576401613e-06, "loss": 0.0052, "step": 120600 }, { "epoch": 43.825118656443955, "grad_norm": 0.06612218916416168, "learning_rate": 6.476731403444486e-06, "loss": 0.0052, "step": 120650 }, { "epoch": 43.8433734939759, "grad_norm": 0.09319305419921875, "learning_rate": 6.474899230487358e-06, "loss": 0.005, "step": 120700 }, { "epoch": 43.86162833150785, "grad_norm": 0.1021774634718895, "learning_rate": 6.473067057530231e-06, "loss": 0.0048, "step": 120750 }, { "epoch": 43.87988316903979, "grad_norm": 0.0935245156288147, "learning_rate": 6.471234884573104e-06, "loss": 0.0051, "step": 120800 }, { "epoch": 43.89813800657174, "grad_norm": 0.09391063451766968, "learning_rate": 6.469402711615977e-06, "loss": 0.005, "step": 120850 }, { "epoch": 43.91639284410369, "grad_norm": 0.1297447681427002, "learning_rate": 6.4675705386588495e-06, "loss": 0.0051, "step": 120900 }, { "epoch": 43.93464768163563, "grad_norm": 0.0745924711227417, "learning_rate": 6.465738365701723e-06, "loss": 0.0049, "step": 120950 }, { "epoch": 43.95290251916758, "grad_norm": 0.07661570608615875, "learning_rate": 6.463906192744596e-06, "loss": 0.0051, "step": 121000 }, { "epoch": 43.97115735669953, "grad_norm": 0.08804076910018921, "learning_rate": 6.462074019787469e-06, "loss": 0.0048, "step": 121050 }, { "epoch": 43.98941219423147, "grad_norm": 0.10511090606451035, "learning_rate": 6.460241846830342e-06, "loss": 0.0049, "step": 121100 }, { "epoch": 44.00766703176342, "grad_norm": 0.05866978317499161, "learning_rate": 6.458409673873214e-06, "loss": 0.0047, "step": 121150 }, { "epoch": 44.02592186929537, "grad_norm": 0.09909752756357193, "learning_rate": 6.456577500916087e-06, "loss": 0.0046, "step": 121200 }, { "epoch": 44.04417670682731, "grad_norm": 0.0851065143942833, "learning_rate": 6.45474532795896e-06, "loss": 0.0045, "step": 121250 }, { "epoch": 44.06243154435926, "grad_norm": 0.07917455583810806, "learning_rate": 6.452913155001833e-06, "loss": 0.0046, "step": 121300 }, { "epoch": 44.0806863818912, "grad_norm": 0.08574696630239487, "learning_rate": 6.451080982044705e-06, "loss": 0.0047, "step": 121350 }, { "epoch": 44.098941219423146, "grad_norm": 0.11335451900959015, "learning_rate": 6.4492488090875785e-06, "loss": 0.0046, "step": 121400 }, { "epoch": 44.117196056955095, "grad_norm": 0.09198125451803207, "learning_rate": 6.447416636130452e-06, "loss": 0.0047, "step": 121450 }, { "epoch": 44.135450894487036, "grad_norm": 0.08669835329055786, "learning_rate": 6.445584463173325e-06, "loss": 0.0051, "step": 121500 }, { "epoch": 44.153705732018985, "grad_norm": 0.06756642460823059, "learning_rate": 6.443752290216196e-06, "loss": 0.0046, "step": 121550 }, { "epoch": 44.17196056955093, "grad_norm": 0.0963951051235199, "learning_rate": 6.441920117259069e-06, "loss": 0.0046, "step": 121600 }, { "epoch": 44.190215407082874, "grad_norm": 0.09031671285629272, "learning_rate": 6.4400879443019425e-06, "loss": 0.005, "step": 121650 }, { "epoch": 44.20847024461482, "grad_norm": 0.08938317745923996, "learning_rate": 6.438255771344816e-06, "loss": 0.0046, "step": 121700 }, { "epoch": 44.22672508214677, "grad_norm": 0.07930494844913483, "learning_rate": 6.436423598387689e-06, "loss": 0.0046, "step": 121750 }, { "epoch": 44.24497991967871, "grad_norm": 0.11311503499746323, "learning_rate": 6.434591425430561e-06, "loss": 0.0049, "step": 121800 }, { "epoch": 44.26323475721066, "grad_norm": 0.0936070904135704, "learning_rate": 6.432759252473434e-06, "loss": 0.0046, "step": 121850 }, { "epoch": 44.28148959474261, "grad_norm": 0.08987881243228912, "learning_rate": 6.430927079516307e-06, "loss": 0.0048, "step": 121900 }, { "epoch": 44.29974443227455, "grad_norm": 0.10625980794429779, "learning_rate": 6.42909490655918e-06, "loss": 0.0047, "step": 121950 }, { "epoch": 44.3179992698065, "grad_norm": 0.08153087645769119, "learning_rate": 6.427262733602052e-06, "loss": 0.0045, "step": 122000 }, { "epoch": 44.33625410733845, "grad_norm": 0.08635585755109787, "learning_rate": 6.425430560644925e-06, "loss": 0.0048, "step": 122050 }, { "epoch": 44.35450894487039, "grad_norm": 0.10117514431476593, "learning_rate": 6.423598387687798e-06, "loss": 0.0047, "step": 122100 }, { "epoch": 44.37276378240234, "grad_norm": 0.11264174431562424, "learning_rate": 6.4217662147306715e-06, "loss": 0.0045, "step": 122150 }, { "epoch": 44.391018619934286, "grad_norm": 0.08618224412202835, "learning_rate": 6.419934041773544e-06, "loss": 0.0045, "step": 122200 }, { "epoch": 44.40927345746623, "grad_norm": 0.09823068231344223, "learning_rate": 6.418101868816416e-06, "loss": 0.0046, "step": 122250 }, { "epoch": 44.427528294998176, "grad_norm": 0.07930401712656021, "learning_rate": 6.416269695859289e-06, "loss": 0.0048, "step": 122300 }, { "epoch": 44.44578313253012, "grad_norm": 0.10463091731071472, "learning_rate": 6.414437522902162e-06, "loss": 0.0047, "step": 122350 }, { "epoch": 44.464037970062066, "grad_norm": 0.08113061636686325, "learning_rate": 6.4126053499450356e-06, "loss": 0.0048, "step": 122400 }, { "epoch": 44.482292807594014, "grad_norm": 0.11078689992427826, "learning_rate": 6.410773176987908e-06, "loss": 0.0047, "step": 122450 }, { "epoch": 44.500547645125955, "grad_norm": 0.08755557239055634, "learning_rate": 6.408941004030781e-06, "loss": 0.0048, "step": 122500 }, { "epoch": 44.518802482657904, "grad_norm": 0.094663105905056, "learning_rate": 6.407108831073654e-06, "loss": 0.0048, "step": 122550 }, { "epoch": 44.53705732018985, "grad_norm": 0.08095148950815201, "learning_rate": 6.405276658116527e-06, "loss": 0.0046, "step": 122600 }, { "epoch": 44.555312157721794, "grad_norm": 0.08783454447984695, "learning_rate": 6.403444485159399e-06, "loss": 0.0049, "step": 122650 }, { "epoch": 44.57356699525374, "grad_norm": 0.3185591995716095, "learning_rate": 6.401612312202272e-06, "loss": 0.0045, "step": 122700 }, { "epoch": 44.59182183278569, "grad_norm": 0.07423868030309677, "learning_rate": 6.399780139245145e-06, "loss": 0.0048, "step": 122750 }, { "epoch": 44.61007667031763, "grad_norm": 0.06885489821434021, "learning_rate": 6.397947966288018e-06, "loss": 0.0046, "step": 122800 }, { "epoch": 44.62833150784958, "grad_norm": 0.11765241622924805, "learning_rate": 6.396115793330891e-06, "loss": 0.005, "step": 122850 }, { "epoch": 44.64658634538153, "grad_norm": 0.07743636518716812, "learning_rate": 6.394283620373764e-06, "loss": 0.0049, "step": 122900 }, { "epoch": 44.66484118291347, "grad_norm": 0.10201650857925415, "learning_rate": 6.392451447416637e-06, "loss": 0.0049, "step": 122950 }, { "epoch": 44.68309602044542, "grad_norm": 0.0937248170375824, "learning_rate": 6.39061927445951e-06, "loss": 0.0048, "step": 123000 }, { "epoch": 44.70135085797737, "grad_norm": 0.12085042893886566, "learning_rate": 6.388787101502383e-06, "loss": 0.0046, "step": 123050 }, { "epoch": 44.71960569550931, "grad_norm": 0.09201015532016754, "learning_rate": 6.3869549285452546e-06, "loss": 0.0046, "step": 123100 }, { "epoch": 44.73786053304126, "grad_norm": 0.10921952873468399, "learning_rate": 6.385122755588128e-06, "loss": 0.0048, "step": 123150 }, { "epoch": 44.756115370573205, "grad_norm": 0.0725519061088562, "learning_rate": 6.383290582631001e-06, "loss": 0.0049, "step": 123200 }, { "epoch": 44.77437020810515, "grad_norm": 0.09321488440036774, "learning_rate": 6.381458409673874e-06, "loss": 0.0048, "step": 123250 }, { "epoch": 44.792625045637095, "grad_norm": 0.0863138809800148, "learning_rate": 6.379626236716746e-06, "loss": 0.0047, "step": 123300 }, { "epoch": 44.81087988316904, "grad_norm": 0.1432102620601654, "learning_rate": 6.3777940637596195e-06, "loss": 0.0045, "step": 123350 }, { "epoch": 44.829134720700985, "grad_norm": 0.08834115415811539, "learning_rate": 6.375961890802493e-06, "loss": 0.005, "step": 123400 }, { "epoch": 44.84738955823293, "grad_norm": 0.12708266079425812, "learning_rate": 6.374129717845366e-06, "loss": 0.0049, "step": 123450 }, { "epoch": 44.865644395764875, "grad_norm": 0.13521406054496765, "learning_rate": 6.372297544888238e-06, "loss": 0.0048, "step": 123500 }, { "epoch": 44.88389923329682, "grad_norm": 0.0976947769522667, "learning_rate": 6.37046537193111e-06, "loss": 0.0049, "step": 123550 }, { "epoch": 44.90215407082877, "grad_norm": 0.07056766748428345, "learning_rate": 6.3686331989739835e-06, "loss": 0.005, "step": 123600 }, { "epoch": 44.92040890836071, "grad_norm": 0.09044899791479111, "learning_rate": 6.366801026016857e-06, "loss": 0.0047, "step": 123650 }, { "epoch": 44.93866374589266, "grad_norm": 0.07794420421123505, "learning_rate": 6.36496885305973e-06, "loss": 0.0049, "step": 123700 }, { "epoch": 44.95691858342461, "grad_norm": 0.0828799307346344, "learning_rate": 6.363136680102602e-06, "loss": 0.0045, "step": 123750 }, { "epoch": 44.97517342095655, "grad_norm": 0.0873388722538948, "learning_rate": 6.361304507145475e-06, "loss": 0.0048, "step": 123800 }, { "epoch": 44.9934282584885, "grad_norm": 0.08319572359323502, "learning_rate": 6.359472334188348e-06, "loss": 0.0051, "step": 123850 }, { "epoch": 45.01168309602045, "grad_norm": 0.06485471874475479, "learning_rate": 6.357640161231221e-06, "loss": 0.0046, "step": 123900 }, { "epoch": 45.02993793355239, "grad_norm": 0.1460689753293991, "learning_rate": 6.355807988274093e-06, "loss": 0.0047, "step": 123950 }, { "epoch": 45.04819277108434, "grad_norm": 0.1042620837688446, "learning_rate": 6.353975815316966e-06, "loss": 0.0045, "step": 124000 }, { "epoch": 45.066447608616286, "grad_norm": 0.08859159797430038, "learning_rate": 6.352143642359839e-06, "loss": 0.0044, "step": 124050 }, { "epoch": 45.08470244614823, "grad_norm": 0.09520501643419266, "learning_rate": 6.3503114694027125e-06, "loss": 0.0046, "step": 124100 }, { "epoch": 45.102957283680176, "grad_norm": 0.1257169246673584, "learning_rate": 6.348479296445586e-06, "loss": 0.0043, "step": 124150 }, { "epoch": 45.121212121212125, "grad_norm": 0.07608997076749802, "learning_rate": 6.346647123488457e-06, "loss": 0.0044, "step": 124200 }, { "epoch": 45.139466958744066, "grad_norm": 0.06328009814023972, "learning_rate": 6.34481495053133e-06, "loss": 0.0043, "step": 124250 }, { "epoch": 45.157721796276014, "grad_norm": 0.09853637218475342, "learning_rate": 6.342982777574203e-06, "loss": 0.0049, "step": 124300 }, { "epoch": 45.175976633807956, "grad_norm": 0.08201096206903458, "learning_rate": 6.3411506046170765e-06, "loss": 0.0048, "step": 124350 }, { "epoch": 45.194231471339904, "grad_norm": 0.0877637192606926, "learning_rate": 6.339318431659949e-06, "loss": 0.0047, "step": 124400 }, { "epoch": 45.21248630887185, "grad_norm": 0.09361255168914795, "learning_rate": 6.337486258702822e-06, "loss": 0.0045, "step": 124450 }, { "epoch": 45.230741146403794, "grad_norm": 0.11702991276979446, "learning_rate": 6.335654085745695e-06, "loss": 0.0043, "step": 124500 }, { "epoch": 45.24899598393574, "grad_norm": 0.09021913260221481, "learning_rate": 6.333821912788568e-06, "loss": 0.0048, "step": 124550 }, { "epoch": 45.26725082146769, "grad_norm": 0.07573777437210083, "learning_rate": 6.33198973983144e-06, "loss": 0.0044, "step": 124600 }, { "epoch": 45.28550565899963, "grad_norm": 0.09693477302789688, "learning_rate": 6.330157566874313e-06, "loss": 0.0047, "step": 124650 }, { "epoch": 45.30376049653158, "grad_norm": 0.07110164314508438, "learning_rate": 6.328325393917186e-06, "loss": 0.0048, "step": 124700 }, { "epoch": 45.32201533406353, "grad_norm": 0.0971999317407608, "learning_rate": 6.326493220960059e-06, "loss": 0.0043, "step": 124750 }, { "epoch": 45.34027017159547, "grad_norm": 0.09390491247177124, "learning_rate": 6.324661048002932e-06, "loss": 0.0044, "step": 124800 }, { "epoch": 45.35852500912742, "grad_norm": 0.08418618142604828, "learning_rate": 6.322828875045805e-06, "loss": 0.0045, "step": 124850 }, { "epoch": 45.37677984665937, "grad_norm": 0.056122470647096634, "learning_rate": 6.320996702088678e-06, "loss": 0.0046, "step": 124900 }, { "epoch": 45.39503468419131, "grad_norm": 0.08667424321174622, "learning_rate": 6.319164529131551e-06, "loss": 0.0048, "step": 124950 }, { "epoch": 45.41328952172326, "grad_norm": 0.11606257408857346, "learning_rate": 6.317332356174424e-06, "loss": 0.0045, "step": 125000 }, { "epoch": 45.431544359255206, "grad_norm": 0.09905866533517838, "learning_rate": 6.3155001832172956e-06, "loss": 0.0048, "step": 125050 }, { "epoch": 45.44979919678715, "grad_norm": 0.06187543645501137, "learning_rate": 6.313668010260169e-06, "loss": 0.0044, "step": 125100 }, { "epoch": 45.468054034319096, "grad_norm": 0.11465045064687729, "learning_rate": 6.311835837303042e-06, "loss": 0.0046, "step": 125150 }, { "epoch": 45.486308871851044, "grad_norm": 0.08560871332883835, "learning_rate": 6.310003664345915e-06, "loss": 0.0046, "step": 125200 }, { "epoch": 45.504563709382985, "grad_norm": 0.08498185873031616, "learning_rate": 6.308171491388788e-06, "loss": 0.0047, "step": 125250 }, { "epoch": 45.522818546914934, "grad_norm": 0.1327318549156189, "learning_rate": 6.3063393184316605e-06, "loss": 0.0043, "step": 125300 }, { "epoch": 45.541073384446875, "grad_norm": 0.08194301277399063, "learning_rate": 6.304507145474534e-06, "loss": 0.0049, "step": 125350 }, { "epoch": 45.559328221978824, "grad_norm": 0.09320477396249771, "learning_rate": 6.302674972517406e-06, "loss": 0.0047, "step": 125400 }, { "epoch": 45.57758305951077, "grad_norm": 0.10656895488500595, "learning_rate": 6.300842799560279e-06, "loss": 0.0046, "step": 125450 }, { "epoch": 45.59583789704271, "grad_norm": 0.06291146576404572, "learning_rate": 6.299010626603151e-06, "loss": 0.0045, "step": 125500 }, { "epoch": 45.61409273457466, "grad_norm": 0.10364147275686264, "learning_rate": 6.2971784536460245e-06, "loss": 0.0046, "step": 125550 }, { "epoch": 45.63234757210661, "grad_norm": 0.07790575921535492, "learning_rate": 6.295346280688898e-06, "loss": 0.0047, "step": 125600 }, { "epoch": 45.65060240963855, "grad_norm": 0.10303451865911484, "learning_rate": 6.293514107731771e-06, "loss": 0.0048, "step": 125650 }, { "epoch": 45.6688572471705, "grad_norm": 0.09793249517679214, "learning_rate": 6.291681934774643e-06, "loss": 0.0047, "step": 125700 }, { "epoch": 45.68711208470245, "grad_norm": 0.13290570676326752, "learning_rate": 6.2898497618175154e-06, "loss": 0.0047, "step": 125750 }, { "epoch": 45.70536692223439, "grad_norm": 0.07874773442745209, "learning_rate": 6.288017588860389e-06, "loss": 0.0046, "step": 125800 }, { "epoch": 45.72362175976634, "grad_norm": 0.11661013960838318, "learning_rate": 6.286185415903262e-06, "loss": 0.005, "step": 125850 }, { "epoch": 45.74187659729829, "grad_norm": 0.07244199514389038, "learning_rate": 6.284353242946135e-06, "loss": 0.0046, "step": 125900 }, { "epoch": 45.76013143483023, "grad_norm": 0.1314825862646103, "learning_rate": 6.282521069989007e-06, "loss": 0.0049, "step": 125950 }, { "epoch": 45.77838627236218, "grad_norm": 0.0916513055562973, "learning_rate": 6.28068889703188e-06, "loss": 0.0043, "step": 126000 }, { "epoch": 45.796641109894125, "grad_norm": 0.09225677698850632, "learning_rate": 6.2788567240747535e-06, "loss": 0.0047, "step": 126050 }, { "epoch": 45.814895947426066, "grad_norm": 0.07904773950576782, "learning_rate": 6.277024551117627e-06, "loss": 0.0044, "step": 126100 }, { "epoch": 45.833150784958015, "grad_norm": 0.08428318798542023, "learning_rate": 6.275192378160498e-06, "loss": 0.0049, "step": 126150 }, { "epoch": 45.85140562248996, "grad_norm": 0.10649510473012924, "learning_rate": 6.273360205203371e-06, "loss": 0.0046, "step": 126200 }, { "epoch": 45.869660460021905, "grad_norm": 0.12103387713432312, "learning_rate": 6.271528032246244e-06, "loss": 0.0047, "step": 126250 }, { "epoch": 45.88791529755385, "grad_norm": 0.09783146530389786, "learning_rate": 6.2696958592891175e-06, "loss": 0.0048, "step": 126300 }, { "epoch": 45.906170135085794, "grad_norm": 0.1110522523522377, "learning_rate": 6.26786368633199e-06, "loss": 0.0046, "step": 126350 }, { "epoch": 45.92442497261774, "grad_norm": 0.12016516178846359, "learning_rate": 6.266031513374863e-06, "loss": 0.0047, "step": 126400 }, { "epoch": 45.94267981014969, "grad_norm": 0.09590303897857666, "learning_rate": 6.264199340417736e-06, "loss": 0.0045, "step": 126450 }, { "epoch": 45.96093464768163, "grad_norm": 0.10076398402452469, "learning_rate": 6.262367167460609e-06, "loss": 0.0049, "step": 126500 }, { "epoch": 45.97918948521358, "grad_norm": 0.09686345607042313, "learning_rate": 6.2605349945034824e-06, "loss": 0.0046, "step": 126550 }, { "epoch": 45.99744432274553, "grad_norm": 0.12283005565404892, "learning_rate": 6.258702821546354e-06, "loss": 0.0046, "step": 126600 }, { "epoch": 46.01569916027747, "grad_norm": 0.12778128683567047, "learning_rate": 6.256870648589227e-06, "loss": 0.0044, "step": 126650 }, { "epoch": 46.03395399780942, "grad_norm": 0.06719617545604706, "learning_rate": 6.2550384756321e-06, "loss": 0.0045, "step": 126700 }, { "epoch": 46.05220883534137, "grad_norm": 0.09697455167770386, "learning_rate": 6.253206302674973e-06, "loss": 0.0044, "step": 126750 }, { "epoch": 46.07046367287331, "grad_norm": 0.10436306148767471, "learning_rate": 6.251374129717846e-06, "loss": 0.0045, "step": 126800 }, { "epoch": 46.08871851040526, "grad_norm": 0.06990363448858261, "learning_rate": 6.249541956760719e-06, "loss": 0.0046, "step": 126850 }, { "epoch": 46.106973347937206, "grad_norm": 0.15162351727485657, "learning_rate": 6.247709783803592e-06, "loss": 0.0044, "step": 126900 }, { "epoch": 46.12522818546915, "grad_norm": 0.08683234453201294, "learning_rate": 6.245877610846465e-06, "loss": 0.0044, "step": 126950 }, { "epoch": 46.143483023001096, "grad_norm": 0.11583215743303299, "learning_rate": 6.244045437889337e-06, "loss": 0.0045, "step": 127000 }, { "epoch": 46.161737860533044, "grad_norm": 0.11069400608539581, "learning_rate": 6.24221326493221e-06, "loss": 0.0044, "step": 127050 }, { "epoch": 46.179992698064986, "grad_norm": 0.09422094374895096, "learning_rate": 6.240381091975083e-06, "loss": 0.0042, "step": 127100 }, { "epoch": 46.198247535596934, "grad_norm": 0.08521673828363419, "learning_rate": 6.238548919017956e-06, "loss": 0.0044, "step": 127150 }, { "epoch": 46.21650237312888, "grad_norm": 0.10951060056686401, "learning_rate": 6.236716746060829e-06, "loss": 0.0044, "step": 127200 }, { "epoch": 46.234757210660824, "grad_norm": 0.09834132343530655, "learning_rate": 6.2348845731037015e-06, "loss": 0.0045, "step": 127250 }, { "epoch": 46.25301204819277, "grad_norm": 0.0985632911324501, "learning_rate": 6.233052400146575e-06, "loss": 0.0047, "step": 127300 }, { "epoch": 46.271266885724714, "grad_norm": 0.08229799568653107, "learning_rate": 6.231220227189447e-06, "loss": 0.0043, "step": 127350 }, { "epoch": 46.28952172325666, "grad_norm": 0.08428385108709335, "learning_rate": 6.22938805423232e-06, "loss": 0.0043, "step": 127400 }, { "epoch": 46.30777656078861, "grad_norm": 0.08850458264350891, "learning_rate": 6.227555881275192e-06, "loss": 0.0044, "step": 127450 }, { "epoch": 46.32603139832055, "grad_norm": 0.07611207664012909, "learning_rate": 6.2257237083180655e-06, "loss": 0.0043, "step": 127500 }, { "epoch": 46.3442862358525, "grad_norm": 0.11391718685626984, "learning_rate": 6.223891535360939e-06, "loss": 0.0045, "step": 127550 }, { "epoch": 46.36254107338445, "grad_norm": 0.1047099307179451, "learning_rate": 6.222059362403812e-06, "loss": 0.0046, "step": 127600 }, { "epoch": 46.38079591091639, "grad_norm": 0.09547355026006699, "learning_rate": 6.220227189446685e-06, "loss": 0.0045, "step": 127650 }, { "epoch": 46.39905074844834, "grad_norm": 0.07315968722105026, "learning_rate": 6.218395016489556e-06, "loss": 0.0042, "step": 127700 }, { "epoch": 46.41730558598029, "grad_norm": 0.17775288224220276, "learning_rate": 6.2165628435324296e-06, "loss": 0.0046, "step": 127750 }, { "epoch": 46.43556042351223, "grad_norm": 0.07889262586832047, "learning_rate": 6.214730670575303e-06, "loss": 0.0044, "step": 127800 }, { "epoch": 46.45381526104418, "grad_norm": 0.13150930404663086, "learning_rate": 6.212898497618176e-06, "loss": 0.0043, "step": 127850 }, { "epoch": 46.472070098576125, "grad_norm": 0.1039467453956604, "learning_rate": 6.211066324661048e-06, "loss": 0.0042, "step": 127900 }, { "epoch": 46.49032493610807, "grad_norm": 0.09393088519573212, "learning_rate": 6.209234151703921e-06, "loss": 0.0045, "step": 127950 }, { "epoch": 46.508579773640015, "grad_norm": 0.09139914810657501, "learning_rate": 6.2074019787467945e-06, "loss": 0.0045, "step": 128000 }, { "epoch": 46.526834611171964, "grad_norm": 0.09246904402971268, "learning_rate": 6.205569805789668e-06, "loss": 0.0044, "step": 128050 }, { "epoch": 46.545089448703905, "grad_norm": 0.0982731506228447, "learning_rate": 6.203737632832539e-06, "loss": 0.0044, "step": 128100 }, { "epoch": 46.56334428623585, "grad_norm": 0.10178857296705246, "learning_rate": 6.201905459875412e-06, "loss": 0.0045, "step": 128150 }, { "epoch": 46.5815991237678, "grad_norm": 0.07372350245714188, "learning_rate": 6.200073286918285e-06, "loss": 0.0046, "step": 128200 }, { "epoch": 46.59985396129974, "grad_norm": 0.08808635175228119, "learning_rate": 6.1982411139611585e-06, "loss": 0.0048, "step": 128250 }, { "epoch": 46.61810879883169, "grad_norm": 0.08808941394090652, "learning_rate": 6.196408941004032e-06, "loss": 0.0046, "step": 128300 }, { "epoch": 46.63636363636363, "grad_norm": 0.11742013692855835, "learning_rate": 6.194576768046904e-06, "loss": 0.0042, "step": 128350 }, { "epoch": 46.65461847389558, "grad_norm": 0.09015604853630066, "learning_rate": 6.192744595089777e-06, "loss": 0.0044, "step": 128400 }, { "epoch": 46.67287331142753, "grad_norm": 0.12545983493328094, "learning_rate": 6.19091242213265e-06, "loss": 0.0046, "step": 128450 }, { "epoch": 46.69112814895947, "grad_norm": 0.13544654846191406, "learning_rate": 6.1890802491755234e-06, "loss": 0.0045, "step": 128500 }, { "epoch": 46.70938298649142, "grad_norm": 0.11101964116096497, "learning_rate": 6.187248076218395e-06, "loss": 0.0042, "step": 128550 }, { "epoch": 46.72763782402337, "grad_norm": 0.11416248232126236, "learning_rate": 6.185415903261268e-06, "loss": 0.0043, "step": 128600 }, { "epoch": 46.74589266155531, "grad_norm": 0.09690819680690765, "learning_rate": 6.183583730304141e-06, "loss": 0.0046, "step": 128650 }, { "epoch": 46.76414749908726, "grad_norm": 0.10517092794179916, "learning_rate": 6.181751557347014e-06, "loss": 0.0048, "step": 128700 }, { "epoch": 46.78240233661921, "grad_norm": 0.09808546304702759, "learning_rate": 6.179919384389887e-06, "loss": 0.0042, "step": 128750 }, { "epoch": 46.80065717415115, "grad_norm": 0.09419416636228561, "learning_rate": 6.17808721143276e-06, "loss": 0.0045, "step": 128800 }, { "epoch": 46.818912011683096, "grad_norm": 0.07550405710935593, "learning_rate": 6.176255038475633e-06, "loss": 0.0044, "step": 128850 }, { "epoch": 46.837166849215045, "grad_norm": 0.10497239977121353, "learning_rate": 6.174422865518506e-06, "loss": 0.0042, "step": 128900 }, { "epoch": 46.855421686746986, "grad_norm": 0.12031522393226624, "learning_rate": 6.172590692561378e-06, "loss": 0.0046, "step": 128950 }, { "epoch": 46.873676524278935, "grad_norm": 0.08948862552642822, "learning_rate": 6.170758519604251e-06, "loss": 0.0046, "step": 129000 }, { "epoch": 46.89193136181088, "grad_norm": 0.10355645418167114, "learning_rate": 6.168926346647124e-06, "loss": 0.0044, "step": 129050 }, { "epoch": 46.910186199342824, "grad_norm": 0.1214522123336792, "learning_rate": 6.167094173689997e-06, "loss": 0.0047, "step": 129100 }, { "epoch": 46.92844103687477, "grad_norm": 0.10329955071210861, "learning_rate": 6.16526200073287e-06, "loss": 0.0044, "step": 129150 }, { "epoch": 46.946695874406714, "grad_norm": 0.09071590006351471, "learning_rate": 6.1634298277757424e-06, "loss": 0.0046, "step": 129200 }, { "epoch": 46.96495071193866, "grad_norm": 0.1131206676363945, "learning_rate": 6.161597654818615e-06, "loss": 0.0047, "step": 129250 }, { "epoch": 46.98320554947061, "grad_norm": 0.1070469543337822, "learning_rate": 6.159765481861488e-06, "loss": 0.0047, "step": 129300 }, { "epoch": 47.00146038700255, "grad_norm": 0.09909582883119583, "learning_rate": 6.157933308904361e-06, "loss": 0.0048, "step": 129350 }, { "epoch": 47.0197152245345, "grad_norm": 0.08149278163909912, "learning_rate": 6.156101135947234e-06, "loss": 0.0043, "step": 129400 }, { "epoch": 47.03797006206645, "grad_norm": 0.11414974927902222, "learning_rate": 6.1542689629901065e-06, "loss": 0.0045, "step": 129450 }, { "epoch": 47.05622489959839, "grad_norm": 0.1014004722237587, "learning_rate": 6.15243679003298e-06, "loss": 0.0043, "step": 129500 }, { "epoch": 47.07447973713034, "grad_norm": 0.07217259705066681, "learning_rate": 6.150604617075853e-06, "loss": 0.0044, "step": 129550 }, { "epoch": 47.09273457466229, "grad_norm": 0.1096716970205307, "learning_rate": 6.148772444118726e-06, "loss": 0.0046, "step": 129600 }, { "epoch": 47.11098941219423, "grad_norm": 0.11197841912508011, "learning_rate": 6.146940271161597e-06, "loss": 0.0043, "step": 129650 }, { "epoch": 47.12924424972618, "grad_norm": 0.08715730905532837, "learning_rate": 6.1451080982044706e-06, "loss": 0.0044, "step": 129700 }, { "epoch": 47.147499087258126, "grad_norm": 0.07300163060426712, "learning_rate": 6.143275925247344e-06, "loss": 0.0041, "step": 129750 }, { "epoch": 47.16575392479007, "grad_norm": 0.11157883703708649, "learning_rate": 6.141443752290217e-06, "loss": 0.0045, "step": 129800 }, { "epoch": 47.184008762322016, "grad_norm": 0.08178367465734482, "learning_rate": 6.139611579333089e-06, "loss": 0.0041, "step": 129850 }, { "epoch": 47.202263599853964, "grad_norm": 0.10576021671295166, "learning_rate": 6.137779406375962e-06, "loss": 0.0042, "step": 129900 }, { "epoch": 47.220518437385905, "grad_norm": 0.06411070376634598, "learning_rate": 6.1359472334188355e-06, "loss": 0.004, "step": 129950 }, { "epoch": 47.238773274917854, "grad_norm": 0.08004731684923172, "learning_rate": 6.134115060461709e-06, "loss": 0.0041, "step": 130000 }, { "epoch": 47.2570281124498, "grad_norm": 0.12736456096172333, "learning_rate": 6.132282887504582e-06, "loss": 0.0042, "step": 130050 }, { "epoch": 47.275282949981744, "grad_norm": 0.13504065573215485, "learning_rate": 6.130450714547453e-06, "loss": 0.0044, "step": 130100 }, { "epoch": 47.29353778751369, "grad_norm": 0.12139490991830826, "learning_rate": 6.128618541590326e-06, "loss": 0.0041, "step": 130150 }, { "epoch": 47.31179262504564, "grad_norm": 0.12800097465515137, "learning_rate": 6.1267863686331995e-06, "loss": 0.0043, "step": 130200 }, { "epoch": 47.33004746257758, "grad_norm": 0.09835221618413925, "learning_rate": 6.124954195676073e-06, "loss": 0.0043, "step": 130250 }, { "epoch": 47.34830230010953, "grad_norm": 0.071064792573452, "learning_rate": 6.123122022718945e-06, "loss": 0.0042, "step": 130300 }, { "epoch": 47.36655713764147, "grad_norm": 0.0949312373995781, "learning_rate": 6.121289849761818e-06, "loss": 0.0042, "step": 130350 }, { "epoch": 47.38481197517342, "grad_norm": 0.1112135723233223, "learning_rate": 6.119457676804691e-06, "loss": 0.0043, "step": 130400 }, { "epoch": 47.40306681270537, "grad_norm": 0.08167316764593124, "learning_rate": 6.117625503847564e-06, "loss": 0.0042, "step": 130450 }, { "epoch": 47.42132165023731, "grad_norm": 0.06050172448158264, "learning_rate": 6.115793330890436e-06, "loss": 0.0044, "step": 130500 }, { "epoch": 47.43957648776926, "grad_norm": 0.09045983850955963, "learning_rate": 6.113961157933309e-06, "loss": 0.0046, "step": 130550 }, { "epoch": 47.45783132530121, "grad_norm": 0.10721985995769501, "learning_rate": 6.112128984976182e-06, "loss": 0.0046, "step": 130600 }, { "epoch": 47.47608616283315, "grad_norm": 0.07332080602645874, "learning_rate": 6.110296812019055e-06, "loss": 0.0044, "step": 130650 }, { "epoch": 47.4943410003651, "grad_norm": 0.09104850888252258, "learning_rate": 6.1084646390619285e-06, "loss": 0.0045, "step": 130700 }, { "epoch": 47.512595837897045, "grad_norm": 0.11164005100727081, "learning_rate": 6.106632466104801e-06, "loss": 0.0044, "step": 130750 }, { "epoch": 47.53085067542899, "grad_norm": 0.09923738986253738, "learning_rate": 6.104800293147674e-06, "loss": 0.0045, "step": 130800 }, { "epoch": 47.549105512960935, "grad_norm": 0.10474247485399246, "learning_rate": 6.102968120190546e-06, "loss": 0.0042, "step": 130850 }, { "epoch": 47.56736035049288, "grad_norm": 0.12509803473949432, "learning_rate": 6.101135947233419e-06, "loss": 0.0043, "step": 130900 }, { "epoch": 47.585615188024825, "grad_norm": 0.18994233012199402, "learning_rate": 6.099303774276292e-06, "loss": 0.0042, "step": 130950 }, { "epoch": 47.60387002555677, "grad_norm": 0.1005072370171547, "learning_rate": 6.097471601319165e-06, "loss": 0.0044, "step": 131000 }, { "epoch": 47.62212486308872, "grad_norm": 0.07798033952713013, "learning_rate": 6.095639428362038e-06, "loss": 0.0042, "step": 131050 }, { "epoch": 47.64037970062066, "grad_norm": 0.09219565987586975, "learning_rate": 6.093807255404911e-06, "loss": 0.0045, "step": 131100 }, { "epoch": 47.65863453815261, "grad_norm": 0.1207447350025177, "learning_rate": 6.0919750824477834e-06, "loss": 0.0042, "step": 131150 }, { "epoch": 47.67688937568455, "grad_norm": 0.11846964061260223, "learning_rate": 6.090142909490656e-06, "loss": 0.0044, "step": 131200 }, { "epoch": 47.6951442132165, "grad_norm": 0.09729281812906265, "learning_rate": 6.088310736533529e-06, "loss": 0.0042, "step": 131250 }, { "epoch": 47.71339905074845, "grad_norm": 0.10580186545848846, "learning_rate": 6.086478563576402e-06, "loss": 0.0046, "step": 131300 }, { "epoch": 47.73165388828039, "grad_norm": 0.1222720742225647, "learning_rate": 6.084646390619275e-06, "loss": 0.0044, "step": 131350 }, { "epoch": 47.74990872581234, "grad_norm": 0.11015098541975021, "learning_rate": 6.0828142176621475e-06, "loss": 0.0042, "step": 131400 }, { "epoch": 47.76816356334429, "grad_norm": 0.09386037290096283, "learning_rate": 6.080982044705021e-06, "loss": 0.0046, "step": 131450 }, { "epoch": 47.78641840087623, "grad_norm": 0.08783093839883804, "learning_rate": 6.079149871747894e-06, "loss": 0.0044, "step": 131500 }, { "epoch": 47.80467323840818, "grad_norm": 0.11424906551837921, "learning_rate": 6.077317698790767e-06, "loss": 0.0045, "step": 131550 }, { "epoch": 47.822928075940126, "grad_norm": 0.07912424951791763, "learning_rate": 6.075485525833638e-06, "loss": 0.0041, "step": 131600 }, { "epoch": 47.84118291347207, "grad_norm": 0.0752551257610321, "learning_rate": 6.0736533528765116e-06, "loss": 0.0043, "step": 131650 }, { "epoch": 47.859437751004016, "grad_norm": 0.11501950770616531, "learning_rate": 6.071821179919385e-06, "loss": 0.0045, "step": 131700 }, { "epoch": 47.877692588535965, "grad_norm": 0.10269715636968613, "learning_rate": 6.069989006962258e-06, "loss": 0.0043, "step": 131750 }, { "epoch": 47.895947426067906, "grad_norm": 0.1120321974158287, "learning_rate": 6.068156834005131e-06, "loss": 0.0043, "step": 131800 }, { "epoch": 47.914202263599854, "grad_norm": 0.1135217621922493, "learning_rate": 6.066324661048003e-06, "loss": 0.0044, "step": 131850 }, { "epoch": 47.9324571011318, "grad_norm": 0.12219502031803131, "learning_rate": 6.0644924880908765e-06, "loss": 0.0043, "step": 131900 }, { "epoch": 47.950711938663744, "grad_norm": 0.11285781860351562, "learning_rate": 6.06266031513375e-06, "loss": 0.0041, "step": 131950 }, { "epoch": 47.96896677619569, "grad_norm": 0.10152813792228699, "learning_rate": 6.060828142176623e-06, "loss": 0.0044, "step": 132000 }, { "epoch": 47.98722161372764, "grad_norm": 0.19300232827663422, "learning_rate": 6.058995969219494e-06, "loss": 0.0041, "step": 132050 }, { "epoch": 48.00547645125958, "grad_norm": 0.10919605940580368, "learning_rate": 6.057163796262367e-06, "loss": 0.0043, "step": 132100 }, { "epoch": 48.02373128879153, "grad_norm": 0.10010013729333878, "learning_rate": 6.0553316233052405e-06, "loss": 0.0043, "step": 132150 }, { "epoch": 48.04198612632347, "grad_norm": 0.07365663349628448, "learning_rate": 6.053499450348114e-06, "loss": 0.0041, "step": 132200 }, { "epoch": 48.06024096385542, "grad_norm": 0.10735583305358887, "learning_rate": 6.051667277390986e-06, "loss": 0.0045, "step": 132250 }, { "epoch": 48.07849580138737, "grad_norm": 0.07005506008863449, "learning_rate": 6.049835104433859e-06, "loss": 0.0042, "step": 132300 }, { "epoch": 48.09675063891931, "grad_norm": 0.07254713773727417, "learning_rate": 6.048002931476732e-06, "loss": 0.004, "step": 132350 }, { "epoch": 48.11500547645126, "grad_norm": 0.11322945356369019, "learning_rate": 6.046170758519605e-06, "loss": 0.0044, "step": 132400 }, { "epoch": 48.13326031398321, "grad_norm": 0.12890049815177917, "learning_rate": 6.044338585562478e-06, "loss": 0.0041, "step": 132450 }, { "epoch": 48.15151515151515, "grad_norm": 0.10722081363201141, "learning_rate": 6.04250641260535e-06, "loss": 0.0043, "step": 132500 }, { "epoch": 48.1697699890471, "grad_norm": 0.08698159456253052, "learning_rate": 6.040674239648223e-06, "loss": 0.004, "step": 132550 }, { "epoch": 48.188024826579046, "grad_norm": 0.10645138472318649, "learning_rate": 6.038842066691096e-06, "loss": 0.0039, "step": 132600 }, { "epoch": 48.20627966411099, "grad_norm": 0.11375320702791214, "learning_rate": 6.0370098937339695e-06, "loss": 0.0041, "step": 132650 }, { "epoch": 48.224534501642935, "grad_norm": 0.12982425093650818, "learning_rate": 6.035177720776842e-06, "loss": 0.0042, "step": 132700 }, { "epoch": 48.242789339174884, "grad_norm": 0.11918260157108307, "learning_rate": 6.033345547819714e-06, "loss": 0.004, "step": 132750 }, { "epoch": 48.261044176706825, "grad_norm": 0.07436525076627731, "learning_rate": 6.031513374862587e-06, "loss": 0.0041, "step": 132800 }, { "epoch": 48.279299014238774, "grad_norm": 0.11533653736114502, "learning_rate": 6.02968120190546e-06, "loss": 0.0043, "step": 132850 }, { "epoch": 48.29755385177072, "grad_norm": 0.10609728842973709, "learning_rate": 6.027849028948333e-06, "loss": 0.0038, "step": 132900 }, { "epoch": 48.31580868930266, "grad_norm": 0.09817364811897278, "learning_rate": 6.026016855991206e-06, "loss": 0.0041, "step": 132950 }, { "epoch": 48.33406352683461, "grad_norm": 0.06099587678909302, "learning_rate": 6.024184683034079e-06, "loss": 0.0042, "step": 133000 }, { "epoch": 48.35231836436656, "grad_norm": 0.10092119127511978, "learning_rate": 6.022352510076952e-06, "loss": 0.0042, "step": 133050 }, { "epoch": 48.3705732018985, "grad_norm": 0.10637925565242767, "learning_rate": 6.020520337119825e-06, "loss": 0.0041, "step": 133100 }, { "epoch": 48.38882803943045, "grad_norm": 0.10733850300312042, "learning_rate": 6.018688164162697e-06, "loss": 0.0041, "step": 133150 }, { "epoch": 48.40708287696239, "grad_norm": 0.08428292721509933, "learning_rate": 6.01685599120557e-06, "loss": 0.0042, "step": 133200 }, { "epoch": 48.42533771449434, "grad_norm": 0.09039627760648727, "learning_rate": 6.015023818248443e-06, "loss": 0.0045, "step": 133250 }, { "epoch": 48.44359255202629, "grad_norm": 0.09594766795635223, "learning_rate": 6.013191645291316e-06, "loss": 0.0042, "step": 133300 }, { "epoch": 48.46184738955823, "grad_norm": 0.10535372793674469, "learning_rate": 6.0113594723341885e-06, "loss": 0.0041, "step": 133350 }, { "epoch": 48.48010222709018, "grad_norm": 0.07930446416139603, "learning_rate": 6.009527299377062e-06, "loss": 0.0044, "step": 133400 }, { "epoch": 48.49835706462213, "grad_norm": 0.12783053517341614, "learning_rate": 6.007695126419935e-06, "loss": 0.0045, "step": 133450 }, { "epoch": 48.51661190215407, "grad_norm": 0.112743079662323, "learning_rate": 6.005862953462808e-06, "loss": 0.0043, "step": 133500 }, { "epoch": 48.53486673968602, "grad_norm": 0.1292959749698639, "learning_rate": 6.004030780505681e-06, "loss": 0.004, "step": 133550 }, { "epoch": 48.553121577217965, "grad_norm": 0.08584164083003998, "learning_rate": 6.0021986075485525e-06, "loss": 0.0041, "step": 133600 }, { "epoch": 48.571376414749906, "grad_norm": 0.09789010137319565, "learning_rate": 6.000366434591426e-06, "loss": 0.0039, "step": 133650 }, { "epoch": 48.589631252281855, "grad_norm": 0.10573101788759232, "learning_rate": 5.998534261634299e-06, "loss": 0.004, "step": 133700 }, { "epoch": 48.6078860898138, "grad_norm": 0.08038768917322159, "learning_rate": 5.996702088677172e-06, "loss": 0.0037, "step": 133750 }, { "epoch": 48.626140927345745, "grad_norm": 0.10818087309598923, "learning_rate": 5.994869915720044e-06, "loss": 0.0042, "step": 133800 }, { "epoch": 48.64439576487769, "grad_norm": 0.1978798359632492, "learning_rate": 5.9930377427629174e-06, "loss": 0.0041, "step": 133850 }, { "epoch": 48.66265060240964, "grad_norm": 0.10325589030981064, "learning_rate": 5.991205569805791e-06, "loss": 0.0039, "step": 133900 }, { "epoch": 48.68090543994158, "grad_norm": 0.0939413532614708, "learning_rate": 5.989373396848664e-06, "loss": 0.0039, "step": 133950 }, { "epoch": 48.69916027747353, "grad_norm": 0.13589763641357422, "learning_rate": 5.987541223891535e-06, "loss": 0.0043, "step": 134000 } ], "logging_steps": 50, "max_steps": 273900, "num_input_tokens_seen": 0, "num_train_epochs": 100, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": true, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 16, "trial_name": null, "trial_params": null }