Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -3345,3 +3345,45 @@
 {"timestamp": 1787651214.3883007, "elapsed_hours": 6.6548559438520005, "session_id": 2, "event": "train_step", "step": 123610, "epoch": 1, "loss": 1.682463870048523, "loss_raw": 0.9594517946243286, "perplexity": 5.378792303378095, "muon_lr": 0.0016879344338973606, "adamw_lr": 5.063803301692082e-05, "tokens_per_sec": 678506.8574814305, "vram_gb": 2.785593032836914, "grad_norm": 0.091796875, "tokens_processed": 16201809920, "tokens_processed_billions": 16.20180992}
 {"timestamp": 1787651221.8313766, "elapsed_hours": 6.656923464934031, "session_id": 2, "event": "train_step", "step": 123620, "epoch": 1, "loss": 1.6451995396614074, "loss_raw": 2.551044464111328, "perplexity": 5.182043827061691, "muon_lr": 0.0016875284139620224, "adamw_lr": 5.062585241886067e-05, "tokens_per_sec": 678350.3036357679, "vram_gb": 2.785593032836914, "grad_norm": 0.0927734375, "tokens_processed": 16203120640, "tokens_processed_billions": 16.20312064}
 {"timestamp": 1787651229.29318, "elapsed_hours": 6.658996188110775, "session_id": 2, "event": "train_step", "step": 123630, "epoch": 1, "loss": 1.6716508162021637, "loss_raw": 2.609039545059204, "perplexity": 5.320944452178328, "muon_lr": 0.0016871225039685007, "adamw_lr": 5.061367511905502e-05, "tokens_per_sec": 678193.3154322824, "vram_gb": 2.785593032836914, "grad_norm": 0.09521484375, "tokens_processed": 16204431360, "tokens_processed_billions": 16.20443136}
+{"timestamp": 1787651237.3880413, "elapsed_hours": 6.661244760751725, "session_id": 2, "event": "train_step", "step": 123640, "epoch": 1, "loss": 1.6733055317401886, "loss_raw": 1.3896799087524414, "perplexity": 5.32975639025509, "muon_lr": 0.0016863110138537198, "adamw_lr": 5.058933041561159e-05, "tokens_per_sec": 678018.4663616546, "vram_gb": 2.363645076751709, "grad_norm": 0.09619140625, "tokens_processed": 16205742080, "tokens_processed_billions": 16.20574208}
+{"timestamp": 1787651244.827421, "elapsed_hours": 6.663311255110635, "session_id": 2, "event": "train_step", "step": 123650, "epoch": 1, "loss": 1.7107633030414582, "loss_raw": 1.4763718843460083, "perplexity": 5.533183361819519, "muon_lr": 0.0016859054337558634, "adamw_lr": 5.0577163012675895e-05, "tokens_per_sec": 677862.3213349892, "vram_gb": 2.785593032836914, "grad_norm": 0.1044921875, "tokens_processed": 16207052800, "tokens_processed_billions": 16.2070528}
+{"timestamp": 1787651252.2992694, "elapsed_hours": 6.665386768579483, "session_id": 2, "event": "train_step", "step": 123660, "epoch": 1, "loss": 1.7458736670017243, "loss_raw": 1.38898766040802, "perplexity": 5.730906192505251, "muon_lr": 0.0016854999636466243, "adamw_lr": 5.056499890939872e-05, "tokens_per_sec": 677705.3517329606, "vram_gb": 2.785593032836914, "grad_norm": 0.09130859375, "tokens_processed": 16208363520, "tokens_processed_billions": 16.20836352}
+{"timestamp": 1787651259.7406056, "elapsed_hours": 6.6674538064665265, "session_id": 2, "event": "train_step", "step": 123670, "epoch": 1, "loss": 1.7550165474414825, "loss_raw": 1.7378058433532715, "perplexity": 5.783543444011661, "muon_lr": 0.001685094603537695, "adamw_lr": 5.055283810613084e-05, "tokens_per_sec": 677549.3447818011, "vram_gb": 2.785593032836914, "grad_norm": 0.1015625, "tokens_processed": 16209674240, "tokens_processed_billions": 16.20967424}
+{"timestamp": 1787651267.2848759, "elapsed_hours": 6.6695494369003505, "session_id": 2, "event": "train_step", "step": 123680, "epoch": 1, "loss": 1.7436883556842804, "loss_raw": 1.6111952066421509, "perplexity": 5.718396052597325, "muon_lr": 0.0016842842133675234, "adamw_lr": 5.0528526401025696e-05, "tokens_per_sec": 677390.5192847275, "vram_gb": 2.363645076751709, "grad_norm": 0.09326171875, "tokens_processed": 16210984960, "tokens_processed_billions": 16.21098496}
+{"timestamp": 1787651274.776086, "elapsed_hours": 6.671630328827434, "session_id": 2, "event": "train_step", "step": 123690, "epoch": 1, "loss": 1.7639069664478302, "loss_raw": 2.0341830253601074, "perplexity": 5.835190811758652, "muon_lr": 0.001683879183329651, "adamw_lr": 5.0516375499889526e-05, "tokens_per_sec": 677233.2971829291, "vram_gb": 2.785593032836914, "grad_norm": 0.09716796875, "tokens_processed": 16212295680, "tokens_processed_billions": 16.21229568}
+{"timestamp": 1787651282.2204554, "elapsed_hours": 6.6736982091267905, "session_id": 2, "event": "train_step", "step": 123700, "epoch": 1, "loss": 1.7110215258598327, "loss_raw": 2.0755691528320312, "perplexity": 5.534612340511304, "muon_lr": 0.0016834742633388272, "adamw_lr": 5.050422790016481e-05, "tokens_per_sec": 677077.4980381322, "vram_gb": 2.785593032836914, "grad_norm": 0.09619140625, "tokens_processed": 16213606400, "tokens_processed_billions": 16.2136064}
+{"timestamp": 1787651282.2222538, "elapsed_hours": 6.673698708613713, "session_id": 2, "event": "lambda_stats", "step": 123700, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1892361111111112, "lambda_std": 0.20446532211957086, "lambda_all": [2.015625, 1.25, 1.1875, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.1171875, 1.140625, 1.109375, 1.109375, 1.078125, 1.1015625, 1.125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787651289.6861782, "elapsed_hours": 6.675772021081713, "session_id": 2, "event": "train_step", "step": 123710, "epoch": 1, "loss": 1.737176502943039, "loss_raw": 1.8717423677444458, "perplexity": 5.68127967892404, "muon_lr": 0.0016830694534067305, "adamw_lr": 5.0492083602201907e-05, "tokens_per_sec": 676921.1913191152, "vram_gb": 2.785593032836914, "grad_norm": 0.09228515625, "tokens_processed": 16214917120, "tokens_processed_billions": 16.21491712}
+{"timestamp": 1787651297.2302747, "elapsed_hours": 6.677867603368229, "session_id": 2, "event": "train_step", "step": 123720, "epoch": 1, "loss": 1.732541743516922, "loss_raw": 0.9680756330490112, "perplexity": 5.655009239988228, "muon_lr": 0.0016822601637654076, "adamw_lr": 5.046780491296222e-05, "tokens_per_sec": 676762.7678255569, "vram_gb": 2.363645076751709, "grad_norm": 0.09130859375, "tokens_processed": 16216227840, "tokens_processed_billions": 16.21622784}
+{"timestamp": 1787651304.678868, "elapsed_hours": 6.679936657018132, "session_id": 2, "event": "train_step", "step": 123730, "epoch": 1, "loss": 1.7047544538974762, "loss_raw": 1.5091444253921509, "perplexity": 5.500034989236075, "muon_lr": 0.0016818556840795197, "adamw_lr": 5.0455670522385585e-05, "tokens_per_sec": 676607.1466823304, "vram_gb": 2.785593032836914, "grad_norm": 0.09326171875, "tokens_processed": 16217538560, "tokens_processed_billions": 16.21753856}
+{"timestamp": 1787651312.1468797, "elapsed_hours": 6.682011104755931, "session_id": 2, "event": "train_step", "step": 123740, "epoch": 1, "loss": 1.6498605597019196, "loss_raw": 1.6737768650054932, "perplexity": 5.206253814979503, "muon_lr": 0.0016814513144990328, "adamw_lr": 5.0443539434970974e-05, "tokens_per_sec": 676451.0688002481, "vram_gb": 2.785593032836914, "grad_norm": 0.09033203125, "tokens_processed": 16218849280, "tokens_processed_billions": 16.21884928}
+{"timestamp": 1787651319.602787, "elapsed_hours": 6.684082190129492, "session_id": 2, "event": "train_step", "step": 123750, "epoch": 1, "loss": 1.645585254430771, "loss_raw": 1.258878469467163, "perplexity": 5.184043003432419, "muon_lr": 0.00168104705503561, "adamw_lr": 5.0431411651068293e-05, "tokens_per_sec": 676295.4247055355, "vram_gb": 2.785593032836914, "grad_norm": 0.09375, "tokens_processed": 16220160000, "tokens_processed_billions": 16.22016}
+{"timestamp": 1787651327.2108662, "elapsed_hours": 6.686195545395216, "session_id": 2, "event": "train_step", "step": 123760, "epoch": 1, "loss": 1.632105438709259, "loss_raw": 1.6422315835952759, "perplexity": 5.114631934291789, "muon_lr": 0.0016802388665065804, "adamw_lr": 5.0407165995197406e-05, "tokens_per_sec": 676135.5912102743, "vram_gb": 2.363645076751709, "grad_norm": 0.10205078125, "tokens_processed": 16221470720, "tokens_processed_billions": 16.22147072}
+{"timestamp": 1787651334.6998417, "elapsed_hours": 6.688275816440583, "session_id": 2, "event": "train_step", "step": 123770, "epoch": 1, "loss": 1.6414652848243714, "loss_raw": 1.7676914930343628, "perplexity": 5.162728840820223, "muon_lr": 0.001679834937464281, "adamw_lr": 5.039504812392842e-05, "tokens_per_sec": 675979.2173097653, "vram_gb": 2.785593032836914, "grad_norm": 0.095703125, "tokens_processed": 16222781440, "tokens_processed_billions": 16.22278144}
+{"timestamp": 1787651342.1448514, "elapsed_hours": 6.690343874560462, "session_id": 2, "event": "train_step", "step": 123780, "epoch": 1, "loss": 1.6538648581504822, "loss_raw": 2.0238654613494873, "perplexity": 5.227143004415937, "muon_lr": 0.0016794311185856549, "adamw_lr": 5.0382933557569634e-05, "tokens_per_sec": 675824.1768303813, "vram_gb": 2.785593032836914, "grad_norm": 0.10498046875, "tokens_processed": 16224092160, "tokens_processed_billions": 16.22409216}
+{"timestamp": 1787651349.5981646, "elapsed_hours": 6.692414239446322, "session_id": 2, "event": "train_step", "step": 123790, "epoch": 1, "loss": 1.7036150884628296, "loss_raw": 2.230060338973999, "perplexity": 5.493772008069648, "muon_lr": 0.0016790274098823508, "adamw_lr": 5.037082229647052e-05, "tokens_per_sec": 675669.0033299411, "vram_gb": 2.785593032836914, "grad_norm": 0.09423828125, "tokens_processed": 16225402880, "tokens_processed_billions": 16.22540288}
+{"timestamp": 1787651357.1881151, "elapsed_hours": 6.694522558980518, "session_id": 2, "event": "train_step", "step": 123800, "epoch": 1, "loss": 1.6893562030792237, "loss_raw": 1.3978400230407715, "perplexity": 5.4159927830182655, "muon_lr": 0.0016782203230482663, "adamw_lr": 5.0346609691447985e-05, "tokens_per_sec": 675510.0738888134, "vram_gb": 2.363645076751709, "grad_norm": 0.09130859375, "tokens_processed": 16226713600, "tokens_processed_billions": 16.2267136}
+{"timestamp": 1787651357.1901479, "elapsed_hours": 6.694523123635186, "session_id": 2, "event": "lambda_stats", "step": 123800, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1892361111111112, "lambda_std": 0.20446532211957086, "lambda_all": [2.015625, 1.25, 1.1875, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.1171875, 1.140625, 1.109375, 1.109375, 1.078125, 1.1015625, 1.125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787651364.635045, "elapsed_hours": 6.696591150628196, "session_id": 2, "event": "train_step", "step": 123810, "epoch": 1, "loss": 1.6767964351177216, "loss_raw": 1.5742778778076172, "perplexity": 5.348394567950296, "muon_lr": 0.001677816944940762, "adamw_lr": 5.033450834822286e-05, "tokens_per_sec": 675355.2721763937, "vram_gb": 2.785593032836914, "grad_norm": 0.08984375, "tokens_processed": 16228024320, "tokens_processed_billions": 16.22802432}
+{"timestamp": 1787651372.105679, "elapsed_hours": 6.698666326800982, "session_id": 2, "event": "train_step", "step": 123820, "epoch": 1, "loss": 1.647212792634964, "loss_raw": 0.8563253879547119, "perplexity": 5.192487101154555, "muon_lr": 0.0016774136770551248, "adamw_lr": 5.032241031165374e-05, "tokens_per_sec": 675199.8997278231, "vram_gb": 2.785593032836914, "grad_norm": 0.095703125, "tokens_processed": 16229335040, "tokens_processed_billions": 16.22933504}
+{"timestamp": 1787651379.5685508, "elapsed_hours": 6.700739346808858, "session_id": 2, "event": "train_step", "step": 123830, "epoch": 1, "loss": 1.6655467128753663, "loss_raw": 1.7486664056777954, "perplexity": 5.2885637854559295, "muon_lr": 0.001677010519402987, "adamw_lr": 5.0310315582089606e-05, "tokens_per_sec": 675044.8403861995, "vram_gb": 2.785593032836914, "grad_norm": 0.09375, "tokens_processed": 16230645760, "tokens_processed_billions": 16.23064576}
+{"timestamp": 1787651387.0975473, "elapsed_hours": 6.702830734584067, "session_id": 2, "event": "train_step", "step": 123840, "epoch": 1, "loss": 1.6268955516815184, "loss_raw": 2.3742918968200684, "perplexity": 5.088054572367794, "muon_lr": 0.0016762045348457056, "adamw_lr": 5.028613604537116e-05, "tokens_per_sec": 674888.0195056411, "vram_gb": 2.363645076751709, "grad_norm": 0.099609375, "tokens_processed": 16231956480, "tokens_processed_billions": 16.23195648}
+{"timestamp": 1787651394.5546749, "elapsed_hours": 6.704902158909374, "session_id": 2, "event": "train_step", "step": 123850, "epoch": 1, "loss": 1.6206975162029267, "loss_raw": 0.975600004196167, "perplexity": 5.056616158457495, "muon_lr": 0.0016758017079638054, "adamw_lr": 5.027405123891415e-05, "tokens_per_sec": 674733.3196967279, "vram_gb": 2.785593032836914, "grad_norm": 0.09521484375, "tokens_processed": 16233267200, "tokens_processed_billions": 16.2332672}
+{"timestamp": 1787651402.04561, "elapsed_hours": 6.706982974211375, "session_id": 2, "event": "train_step", "step": 123860, "epoch": 1, "loss": 1.6478363692760467, "loss_raw": 1.1486055850982666, "perplexity": 5.1957260245734975, "muon_lr": 0.0016753989913618882, "adamw_lr": 5.026196974085664e-05, "tokens_per_sec": 674577.75936184, "vram_gb": 2.785593032836914, "grad_norm": 0.09521484375, "tokens_processed": 16234577920, "tokens_processed_billions": 16.23457792}
+{"timestamp": 1787651409.5145218, "elapsed_hours": 6.709057672023773, "session_id": 2, "event": "train_step", "step": 123870, "epoch": 1, "loss": 1.696505800485611, "loss_raw": 1.3900580406188965, "perplexity": 5.454853705410878, "muon_lr": 0.0016749963850515672, "adamw_lr": 5.024989155154701e-05, "tokens_per_sec": 674422.9186998159, "vram_gb": 2.785593032836914, "grad_norm": 0.09423828125, "tokens_processed": 16235888640, "tokens_processed_billions": 16.23588864}
+{"timestamp": 1787651417.1104574, "elapsed_hours": 6.711167654063966, "session_id": 2, "event": "train_step", "step": 123880, "epoch": 1, "loss": 1.7315442419052125, "loss_raw": 1.8205593824386597, "perplexity": 5.649371171615747, "muon_lr": 0.0016741915033521512, "adamw_lr": 5.022574510056453e-05, "tokens_per_sec": 674264.6139013241, "vram_gb": 2.363645076751709, "grad_norm": 0.10107421875, "tokens_processed": 16237199360, "tokens_processed_billions": 16.23719936}
+{"timestamp": 1787651424.5553236, "elapsed_hours": 6.713235672579871, "session_id": 2, "event": "train_step", "step": 123890, "epoch": 1, "loss": 1.7704596757888793, "loss_raw": 1.5194377899169922, "perplexity": 5.873552670888972, "muon_lr": 0.0016737892279862672, "adamw_lr": 5.021367683958801e-05, "tokens_per_sec": 674110.6381628724, "vram_gb": 2.785593032836914, "grad_norm": 0.095703125, "tokens_processed": 16238510080, "tokens_processed_billions": 16.23851008}
+{"timestamp": 1787651432.0080063, "elapsed_hours": 6.7153058620956205, "session_id": 2, "event": "train_step", "step": 123900, "epoch": 1, "loss": 1.7729676687717437, "loss_raw": 1.903686285018921, "perplexity": 5.888301987632358, "muon_lr": 0.0016733870629584004, "adamw_lr": 5.0201611888752004e-05, "tokens_per_sec": 673956.5388528641, "vram_gb": 2.785593032836914, "grad_norm": 0.0966796875, "tokens_processed": 16239820800, "tokens_processed_billions": 16.2398208}
+{"timestamp": 1787651432.0104792, "elapsed_hours": 6.715306550065677, "session_id": 2, "event": "lambda_stats", "step": 123900, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1892361111111112, "lambda_std": 0.20446532211957086, "lambda_all": [2.015625, 1.25, 1.1875, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.1171875, 1.140625, 1.109375, 1.109375, 1.078125, 1.1015625, 1.125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787651439.4805079, "elapsed_hours": 6.717381556961271, "session_id": 2, "event": "train_step", "step": 123910, "epoch": 1, "loss": 1.7717522943019868, "loss_raw": 2.468973159790039, "perplexity": 5.881149842873834, "muon_lr": 0.0016729850082801497, "adamw_lr": 5.018955024840448e-05, "tokens_per_sec": 673801.9807020734, "vram_gb": 2.785593032836914, "grad_norm": 0.09375, "tokens_processed": 16241131520, "tokens_processed_billions": 16.24113152}
+{"timestamp": 1787651447.016981, "elapsed_hours": 6.719475021759669, "session_id": 2, "event": "train_step", "step": 123920, "epoch": 1, "loss": 1.7278267967700958, "loss_raw": 1.2502532005310059, "perplexity": 5.628408931366829, "muon_lr": 0.001672181230018866, "adamw_lr": 5.016543690056597e-05, "tokens_per_sec": 673645.7304571474, "vram_gb": 2.363645076751709, "grad_norm": 0.08740234375, "tokens_processed": 16242442240, "tokens_processed_billions": 16.24244224}
+{"timestamp": 1787651454.4779742, "elapsed_hours": 6.721547519961993, "session_id": 2, "event": "train_step", "step": 123930, "epoch": 1, "loss": 1.707259794473648, "loss_raw": 1.4088019132614136, "perplexity": 5.513831725610168, "muon_lr": 0.0016717795064590136, "adamw_lr": 5.01533851937704e-05, "tokens_per_sec": 673491.6874767465, "vram_gb": 2.785593032836914, "grad_norm": 0.10986328125, "tokens_processed": 16243752960, "tokens_processed_billions": 16.24375296}
+{"timestamp": 1787651461.9414444, "elapsed_hours": 6.723620706068145, "session_id": 2, "event": "train_step", "step": 123940, "epoch": 1, "loss": 1.6620393753051759, "loss_raw": 1.4967973232269287, "perplexity": 5.27004749741658, "muon_lr": 0.001671377893295133, "adamw_lr": 5.0141336798853976e-05, "tokens_per_sec": 673337.669537816, "vram_gb": 2.785593032836914, "grad_norm": 0.0927734375, "tokens_processed": 16245063680, "tokens_processed_billions": 16.24506368}
+{"timestamp": 1787651469.3795407, "elapsed_hours": 6.725686843858825, "session_id": 2, "event": "train_step", "step": 123950, "epoch": 1, "loss": 1.673229306936264, "loss_raw": 1.3806713819503784, "perplexity": 5.329350146102414, "muon_lr": 0.0016709763905388073, "adamw_lr": 5.012929171616421e-05, "tokens_per_sec": 673184.4539836658, "vram_gb": 2.785593032836914, "grad_norm": 0.091796875, "tokens_processed": 16246374400, "tokens_processed_billions": 16.2463744}
+{"timestamp": 1787651476.951757, "elapsed_hours": 6.727790237334039, "session_id": 2, "event": "train_step", "step": 123960, "epoch": 1, "loss": 1.653754504919052, "loss_raw": 1.846056580543518, "perplexity": 5.226566204120726, "muon_lr": 0.0016701737162951292, "adamw_lr": 5.010521148885387e-05, "tokens_per_sec": 673027.5936445728, "vram_gb": 2.363645076751709, "grad_norm": 0.09228515625, "tokens_processed": 16247685120, "tokens_processed_billions": 16.24768512}
+{"timestamp": 1787651484.4079177, "elapsed_hours": 6.729861393173536, "session_id": 2, "event": "train_step", "step": 123970, "epoch": 1, "loss": 1.635907415151596, "loss_raw": 1.355485439300537, "perplexity": 5.134114657375691, "muon_lr": 0.0016697725448309244, "adamw_lr": 5.0093176344927725e-05, "tokens_per_sec": 672874.0672413062, "vram_gb": 2.785593032836914, "grad_norm": 0.09521484375, "tokens_processed": 16248995840, "tokens_processed_billions": 16.24899584}
+{"timestamp": 1787651491.8441336, "elapsed_hours": 6.731927008628845, "session_id": 2, "event": "train_step", "step": 123980, "epoch": 1, "loss": 1.6426461136341095, "loss_raw": 1.6791977882385254, "perplexity": 5.168828740532201, "muon_lr": 0.0016693714838205667, "adamw_lr": 5.0081144514617e-05, "tokens_per_sec": 672721.1898232455, "vram_gb": 2.785593032836914, "grad_norm": 0.09326171875, "tokens_processed": 16250306560, "tokens_processed_billions": 16.25030656}
+{"timestamp": 1787651499.2929132, "elapsed_hours": 6.733996114002334, "session_id": 2, "event": "train_step", "step": 123990, "epoch": 1, "loss": 1.7113628602027893, "loss_raw": 2.01387619972229, "perplexity": 5.536501816231203, "muon_lr": 0.0016689705332756243, "adamw_lr": 5.006911599826873e-05, "tokens_per_sec": 672568.0570933536, "vram_gb": 2.785593032836914, "grad_norm": 0.10205078125, "tokens_processed": 16251617280, "tokens_processed_billions": 16.25161728}
+{"timestamp": 1787651506.8450673, "elapsed_hours": 6.736093934575717, "session_id": 2, "event": "train_step", "step": 124000, "epoch": 1, "loss": 1.6836278176307677, "loss_raw": 1.8295034170150757, "perplexity": 5.385056580612956, "muon_lr": 0.0016681689636282274, "adamw_lr": 5.004506890884682e-05, "tokens_per_sec": 672412.1417115917, "vram_gb": 2.363645076751709, "grad_norm": 0.0888671875, "tokens_processed": 16252928000, "tokens_processed_billions": 16.252928}
+{"timestamp": 1787651506.847008, "elapsed_hours": 6.736094473865297, "session_id": 2, "event": "lambda_stats", "step": 124000, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1892361111111112, "lambda_std": 0.20446532211957086, "lambda_all": [2.015625, 1.25, 1.1875, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.1171875, 1.140625, 1.109375, 1.109375, 1.078125, 1.1015625, 1.125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787651532.5604532, "elapsed_hours": 6.743237097395791, "session_id": 2, "event": "validation", "step": 124000, "epoch": 1, "val_loss": 1.750065842270851, "val_perplexity": 5.754981584587728, "is_best": false, "best_loss": 1.6233880519866943, "best_perplexity": 5.070239483980759, "train_loss": 1.8295034170150757, "train_loss_avg": 1.6836278176307677}