Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -2711,3 +2711,47 @@
 {"timestamp": 1787646693.7065382, "elapsed_hours": 5.399111010167334, "session_id": 2, "event": "validation", "step": 118000, "epoch": 1, "val_loss": 1.7528041487932204, "val_perplexity": 5.770762084258903, "is_best": false, "best_loss": 1.6233880519866943, "best_perplexity": 5.070239483980759, "train_loss": 1.3806840181350708, "train_loss_avg": 1.8447966408729553}
 {"timestamp": 1787646701.5956833, "elapsed_hours": 5.401302439173063, "session_id": 2, "event": "checkpoint", "step": 118000, "epoch": 0, "checkpoint_loss": 1.3806840181350708, "checkpoint_name": "codva2_s002_step0118000.pt", "best_loss": 1.6233880519866943, "val_loss": null, "is_best": false}
 {"timestamp": 1787646702.2701476, "elapsed_hours": 5.401489790214432, "session_id": 2, "event": "checkpoint_save", "step": 118000, "epoch": 1, "checkpoint_type": "scheduled", "time_since_last_ckpt": 1.430511474609375e-06, "current_loss": 1.3806840181350708, "best_loss": 1.6233880519866943, "vram_gb": 2.363645076751709}
+{"timestamp": 1787646710.4402015, "elapsed_hours": 5.403759249713686, "session_id": 2, "event": "train_step", "step": 118010, "epoch": 1, "loss": 1.8797098970413209, "loss_raw": 2.717712879180908, "perplexity": 6.551603946784832, "muon_lr": 0.001998419085051154, "adamw_lr": 5.995257255153461e-05, "tokens_per_sec": 798351.7800069232, "vram_gb": 2.785593032836914, "grad_norm": 0.10205078125, "tokens_processed": 15467806720, "tokens_processed_billions": 15.46780672}
+{"timestamp": 1787646717.8820302, "elapsed_hours": 5.405826424294048, "session_id": 2, "event": "train_step", "step": 118020, "epoch": 1, "loss": 1.9277190828323365, "loss_raw": 2.0415594577789307, "perplexity": 6.873813747401193, "muon_lr": 0.0019979391562704872, "adamw_lr": 5.993817468811461e-05, "tokens_per_sec": 798112.8748954197, "vram_gb": 2.785593032836914, "grad_norm": 0.11279296875, "tokens_processed": 15469117440, "tokens_processed_billions": 15.46911744}
+{"timestamp": 1787646725.3232088, "elapsed_hours": 5.407893418404791, "session_id": 2, "event": "train_step", "step": 118030, "epoch": 1, "loss": 1.875941548347473, "loss_raw": 1.4001704454421997, "perplexity": 6.526961678100801, "muon_lr": 0.0019974593284801936, "adamw_lr": 5.99237798544058e-05, "tokens_per_sec": 797874.1809353364, "vram_gb": 2.785593032836914, "grad_norm": 0.09912109375, "tokens_processed": 15470428160, "tokens_processed_billions": 15.47042816}
+{"timestamp": 1787646732.8555748, "elapsed_hours": 5.40998574230406, "session_id": 2, "event": "train_step", "step": 118040, "epoch": 1, "loss": 1.8173159289360046, "loss_raw": 1.2574414014816284, "perplexity": 6.15531495509691, "muon_lr": 0.001996499975926068, "adamw_lr": 5.989499927778203e-05, "tokens_per_sec": 797631.916806572, "vram_gb": 2.363645076751709, "grad_norm": 0.08984375, "tokens_processed": 15471738880, "tokens_processed_billions": 15.47173888}
+{"timestamp": 1787646740.299318, "elapsed_hours": 5.412053448756536, "session_id": 2, "event": "train_step", "step": 118050, "epoch": 1, "loss": 1.8158587980270386, "loss_raw": 1.9070290327072144, "perplexity": 6.146352386824844, "muon_lr": 0.001996020451189901, "adamw_lr": 5.988061353569702e-05, "tokens_per_sec": 797393.4883053007, "vram_gb": 2.785593032836914, "grad_norm": 0.09912109375, "tokens_processed": 15473049600, "tokens_processed_billions": 15.4730496}
+{"timestamp": 1787646747.7384999, "elapsed_hours": 5.414119888080491, "session_id": 2, "event": "train_step", "step": 118060, "epoch": 1, "loss": 1.7514699339866637, "loss_raw": 2.165580987930298, "perplexity": 5.7630677821083, "muon_lr": 0.0019955410274994385, "adamw_lr": 5.986623082498314e-05, "tokens_per_sec": 797155.4258620132, "vram_gb": 2.785593032836914, "grad_norm": 0.08984375, "tokens_processed": 15474360320, "tokens_processed_billions": 15.47436032}
+{"timestamp": 1787646755.1904154, "elapsed_hours": 5.416189864675204, "session_id": 2, "event": "train_step", "step": 118070, "epoch": 1, "loss": 1.683034553527832, "loss_raw": 1.7547392845153809, "perplexity": 5.381862767332461, "muon_lr": 0.001995061704868504, "adamw_lr": 5.985185114605511e-05, "tokens_per_sec": 796917.0264680116, "vram_gb": 2.785593032836914, "grad_norm": 0.09375, "tokens_processed": 15475671040, "tokens_processed_billions": 15.47567104}
+{"timestamp": 1787646762.7249694, "elapsed_hours": 5.41828279627694, "session_id": 2, "event": "train_step", "step": 118080, "epoch": 1, "loss": 1.6888782382011414, "loss_raw": 1.2794960737228394, "perplexity": 5.413404747232409, "muon_lr": 0.001994103362840507, "adamw_lr": 5.98231008852152e-05, "tokens_per_sec": 796675.4165728283, "vram_gb": 2.363645076751709, "grad_norm": 0.0927734375, "tokens_processed": 15476981760, "tokens_processed_billions": 15.47698176}
+{"timestamp": 1787646770.1682692, "elapsed_hours": 5.420350379546483, "session_id": 2, "event": "train_step", "step": 118090, "epoch": 1, "loss": 1.6822868824005126, "loss_raw": 0.8609408140182495, "perplexity": 5.377840407818546, "muon_lr": 0.00199362434347108, "adamw_lr": 5.980873030413239e-05, "tokens_per_sec": 796437.738256475, "vram_gb": 2.785593032836914, "grad_norm": 0.09814453125, "tokens_processed": 15478292480, "tokens_processed_billions": 15.47829248}
+{"timestamp": 1787646777.6204453, "elapsed_hours": 5.422420428461499, "session_id": 2, "event": "train_step", "step": 118100, "epoch": 1, "loss": 1.6555425477027894, "loss_raw": 2.012507915496826, "perplexity": 5.235919888007164, "muon_lr": 0.001993145425216455, "adamw_lr": 5.979436275649363e-05, "tokens_per_sec": 796199.8770118363, "vram_gb": 2.785593032836914, "grad_norm": 0.0908203125, "tokens_processed": 15479603200, "tokens_processed_billions": 15.4796032}
+{"timestamp": 1787646777.622212, "elapsed_hours": 5.422420919272635, "session_id": 2, "event": "lambda_stats", "step": 118100, "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": 1787646785.064974, "elapsed_hours": 5.424488353199429, "session_id": 2, "event": "train_step", "step": 118110, "epoch": 1, "loss": 1.6805913078784942, "loss_raw": 0.8909364342689514, "perplexity": 5.368729604845043, "muon_lr": 0.0019926666080904414, "adamw_lr": 5.9779998242713234e-05, "tokens_per_sec": 795962.5093793416, "vram_gb": 2.785593032836914, "grad_norm": 0.1015625, "tokens_processed": 15480913920, "tokens_processed_billions": 15.48091392}
+{"timestamp": 1787646792.6178136, "elapsed_hours": 5.426586364242766, "session_id": 2, "event": "train_step", "step": 118120, "epoch": 1, "loss": 1.7277838695049286, "loss_raw": 2.1772687435150146, "perplexity": 5.628167324349965, "muon_lr": 0.0019917092772794753, "adamw_lr": 5.9751278318384253e-05, "tokens_per_sec": 795720.8930854026, "vram_gb": 2.363645076751709, "grad_norm": 0.09912109375, "tokens_processed": 15482224640, "tokens_processed_billions": 15.48222464}
+{"timestamp": 1787646800.0624573, "elapsed_hours": 5.42865432076984, "session_id": 2, "event": "train_step", "step": 118130, "epoch": 1, "loss": 1.6999236404895783, "loss_raw": 1.4277989864349365, "perplexity": 5.473529419742564, "muon_lr": 0.00199123076362213, "adamw_lr": 5.973692290866389e-05, "tokens_per_sec": 795483.8909703102, "vram_gb": 2.785593032836914, "grad_norm": 0.0947265625, "tokens_processed": 15483535360, "tokens_processed_billions": 15.48353536}
+{"timestamp": 1787646807.5068877, "elapsed_hours": 5.430722218089634, "session_id": 2, "event": "train_step", "step": 118140, "epoch": 1, "loss": 1.758933435678482, "loss_raw": 2.275407552719116, "perplexity": 5.806241360878171, "muon_lr": 0.001990752351148611, "adamw_lr": 5.972257053445833e-05, "tokens_per_sec": 795247.0754936833, "vram_gb": 2.785593032836914, "grad_norm": 0.099609375, "tokens_processed": 15484846080, "tokens_processed_billions": 15.48484608}
+{"timestamp": 1787646814.950626, "elapsed_hours": 5.432789923085107, "session_id": 2, "event": "train_step", "step": 118150, "epoch": 1, "loss": 1.7349411809444428, "loss_raw": 0.837281346321106, "perplexity": 5.668594372631688, "muon_lr": 0.001990274039872714, "adamw_lr": 5.970822119618141e-05, "tokens_per_sec": 795010.4689437101, "vram_gb": 2.785593032836914, "grad_norm": 0.09375, "tokens_processed": 15486156800, "tokens_processed_billions": 15.4861568}
+{"timestamp": 1787646822.4702322, "elapsed_hours": 5.434878702627288, "session_id": 2, "event": "train_step", "step": 118160, "epoch": 1, "loss": 1.7282178688049317, "loss_raw": 2.0651464462280273, "perplexity": 5.630610475153554, "muon_lr": 0.001989317720968953, "adamw_lr": 5.967953162906859e-05, "tokens_per_sec": 794770.9471889285, "vram_gb": 2.363645076751709, "grad_norm": 0.095703125, "tokens_processed": 15487467520, "tokens_processed_billions": 15.48746752}
+{"timestamp": 1787646829.9205704, "elapsed_hours": 5.436948240995407, "session_id": 2, "event": "train_step", "step": 118170, "epoch": 1, "loss": 1.6878511369228364, "loss_raw": 2.188648223876953, "perplexity": 5.407847486719779, "muon_lr": 0.001988839713368667, "adamw_lr": 5.9665191401060005e-05, "tokens_per_sec": 794534.436785893, "vram_gb": 2.785593032836914, "grad_norm": 0.09228515625, "tokens_processed": 15488778240, "tokens_processed_billions": 15.48877824}
+{"timestamp": 1787646837.3683105, "elapsed_hours": 5.439017057683733, "session_id": 2, "event": "train_step", "step": 118180, "epoch": 1, "loss": 1.7270019042491913, "loss_raw": 1.053288459777832, "perplexity": 5.6237680133272026, "muon_lr": 0.0019883618070211592, "adamw_lr": 5.9650854210634776e-05, "tokens_per_sec": 794298.2120379214, "vram_gb": 2.785593032836914, "grad_norm": 0.0986328125, "tokens_processed": 15490088960, "tokens_processed_billions": 15.49008896}
+{"timestamp": 1787646844.8020382, "elapsed_hours": 5.4410819820563, "session_id": 2, "event": "train_step", "step": 118190, "epoch": 1, "loss": 1.7173670661449432, "loss_raw": 2.215590476989746, "perplexity": 5.569844110167547, "muon_lr": 0.0019878840019402102, "adamw_lr": 5.96365200582063e-05, "tokens_per_sec": 794062.7386257047, "vram_gb": 2.785593032836914, "grad_norm": 0.099609375, "tokens_processed": 15491399680, "tokens_processed_billions": 15.49139968}
+{"timestamp": 1787646852.3374903, "elapsed_hours": 5.443175163269043, "session_id": 2, "event": "train_step", "step": 118200, "epoch": 1, "loss": 1.7034452378749847, "loss_raw": 2.402045488357544, "perplexity": 5.492838966905629, "muon_lr": 0.001986928695633098, "adamw_lr": 5.9607860868992926e-05, "tokens_per_sec": 793823.304374584, "vram_gb": 2.363645076751709, "grad_norm": 0.09130859375, "tokens_processed": 15492710400, "tokens_processed_billions": 15.4927104}
+{"timestamp": 1787646852.3391972, "elapsed_hours": 5.443175637390879, "session_id": 2, "event": "lambda_stats", "step": 118200, "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": 1787646859.7992644, "elapsed_hours": 5.445247878233592, "session_id": 2, "event": "train_step", "step": 118210, "epoch": 1, "loss": 1.684767209291458, "loss_raw": 1.1505630016326904, "perplexity": 5.3911957659773915, "muon_lr": 0.0019864511944344835, "adamw_lr": 5.95935358330345e-05, "tokens_per_sec": 793587.0547119592, "vram_gb": 2.785593032836914, "grad_norm": 0.09716796875, "tokens_processed": 15494021120, "tokens_processed_billions": 15.49402112}
+{"timestamp": 1787646867.2609937, "elapsed_hours": 5.447320580879847, "session_id": 2, "event": "train_step", "step": 118220, "epoch": 1, "loss": 1.7134649705886842, "loss_raw": 2.5796117782592773, "perplexity": 5.5481523953120755, "muon_lr": 0.0019859737945575264, "adamw_lr": 5.957921383672578e-05, "tokens_per_sec": 793350.9826257471, "vram_gb": 2.785593032836914, "grad_norm": 0.0947265625, "tokens_processed": 15495331840, "tokens_processed_billions": 15.49533184}
+{"timestamp": 1787646874.7050524, "elapsed_hours": 5.449388374884923, "session_id": 2, "event": "train_step", "step": 118230, "epoch": 1, "loss": 1.6697146463394166, "loss_raw": 1.5077131986618042, "perplexity": 5.310652166910216, "muon_lr": 0.0019854964960159915, "adamw_lr": 5.956489488047974e-05, "tokens_per_sec": 793115.8120929778, "vram_gb": 2.785593032836914, "grad_norm": 0.0908203125, "tokens_processed": 15496642560, "tokens_processed_billions": 15.49664256}
+{"timestamp": 1787646882.243579, "elapsed_hours": 5.4514824100335435, "session_id": 2, "event": "train_step", "step": 118240, "epoch": 1, "loss": 1.6240328931808472, "loss_raw": 1.4794037342071533, "perplexity": 5.073510037644786, "muon_lr": 0.0019845422029942427, "adamw_lr": 5.9536266089827276e-05, "tokens_per_sec": 792876.9821923842, "vram_gb": 2.363645076751709, "grad_norm": 0.09130859375, "tokens_processed": 15497953280, "tokens_processed_billions": 15.49795328}
+{"timestamp": 1787646889.6913598, "elapsed_hours": 5.453551238046752, "session_id": 2, "event": "train_step", "step": 118250, "epoch": 1, "loss": 1.6288779366016388, "loss_raw": 2.252775192260742, "perplexity": 5.098151059280033, "muon_lr": 0.0019840652085415487, "adamw_lr": 5.952195625624645e-05, "tokens_per_sec": 792642.0206044207, "vram_gb": 2.785593032836914, "grad_norm": 0.09228515625, "tokens_processed": 15499264000, "tokens_processed_billions": 15.499264}
+{"timestamp": 1787646897.1264381, "elapsed_hours": 5.455616537597444, "session_id": 2, "event": "train_step", "step": 118260, "epoch": 1, "loss": 1.669984265565872, "loss_raw": 1.9706416130065918, "perplexity": 5.3120842138844555, "muon_lr": 0.001983588315479316, "adamw_lr": 5.950764946437948e-05, "tokens_per_sec": 792407.7513555534, "vram_gb": 2.785593032836914, "grad_norm": 0.10400390625, "tokens_processed": 15500574720, "tokens_processed_billions": 15.50057472}
+{"timestamp": 1787646904.5788822, "elapsed_hours": 5.457686660952038, "session_id": 2, "event": "train_step", "step": 118270, "epoch": 1, "loss": 1.6933850276470184, "loss_raw": 1.9422390460968018, "perplexity": 5.437856881536333, "muon_lr": 0.0019831115238212963, "adamw_lr": 5.949334571463887e-05, "tokens_per_sec": 792172.9565485886, "vram_gb": 2.785593032836914, "grad_norm": 0.10791015625, "tokens_processed": 15501885440, "tokens_processed_billions": 15.50188544}
+{"timestamp": 1787646912.1182857, "elapsed_hours": 5.459780939684974, "session_id": 2, "event": "train_step", "step": 118280, "epoch": 1, "loss": 1.719272574186325, "loss_raw": 1.346179485321045, "perplexity": 5.580467611277651, "muon_lr": 0.001982158244772894, "adamw_lr": 5.9464747343186825e-05, "tokens_per_sec": 791934.8207989355, "vram_gb": 2.363645076751709, "grad_norm": 0.09130859375, "tokens_processed": 15503196160, "tokens_processed_billions": 15.50319616}
+{"timestamp": 1787646919.5612395, "elapsed_hours": 5.461848426858584, "session_id": 2, "event": "train_step", "step": 118290, "epoch": 1, "loss": 1.7430812323093414, "loss_raw": 1.2222946882247925, "perplexity": 5.714925334370325, "muon_lr": 0.0019816817574100005, "adamw_lr": 5.9450452722300005e-05, "tokens_per_sec": 791700.772211593, "vram_gb": 2.785593032836914, "grad_norm": 0.10107421875, "tokens_processed": 15504506880, "tokens_processed_billions": 15.50450688}
+{"timestamp": 1787646927.013505, "elapsed_hours": 5.463918500675096, "session_id": 2, "event": "train_step", "step": 118300, "epoch": 1, "loss": 1.7358668696880342, "loss_raw": 0.906538724899292, "perplexity": 5.673844156092268, "muon_lr": 0.001981205371506302, "adamw_lr": 5.943616114518905e-05, "tokens_per_sec": 791466.5217797197, "vram_gb": 2.785593032836914, "grad_norm": 0.09130859375, "tokens_processed": 15505817600, "tokens_processed_billions": 15.5058176}
+{"timestamp": 1787646927.0153625, "elapsed_hours": 5.463919016652637, "session_id": 2, "event": "lambda_stats", "step": 118300, "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": 1787646934.463345, "elapsed_hours": 5.465987900627984, "session_id": 2, "event": "train_step", "step": 118310, "epoch": 1, "loss": 1.6843412280082704, "loss_raw": 2.4036879539489746, "perplexity": 5.388899706560973, "muon_lr": 0.001980729087075534, "adamw_lr": 5.9421872612266e-05, "tokens_per_sec": 791232.5487298202, "vram_gb": 2.785593032836914, "grad_norm": 0.091796875, "tokens_processed": 15507128320, "tokens_processed_billions": 15.50712832}
+{"timestamp": 1787646941.994707, "elapsed_hours": 5.468079945643743, "session_id": 2, "event": "train_step", "step": 118320, "epoch": 1, "loss": 1.6418554878234863, "loss_raw": 2.141075849533081, "perplexity": 5.16474374618302, "muon_lr": 0.0019797768226877312, "adamw_lr": 5.9393304680631926e-05, "tokens_per_sec": 790995.464081631, "vram_gb": 2.363645076751709, "grad_norm": 0.091796875, "tokens_processed": 15508439040, "tokens_processed_billions": 15.50843904}
+{"timestamp": 1787646949.4384108, "elapsed_hours": 5.4701476411687, "session_id": 2, "event": "train_step", "step": 118330, "epoch": 1, "loss": 1.6424736046791077, "loss_raw": 1.7883265018463135, "perplexity": 5.167937148193631, "muon_lr": 0.001979300842758155, "adamw_lr": 5.937902528274465e-05, "tokens_per_sec": 790762.0957962788, "vram_gb": 2.785593032836914, "grad_norm": 0.0966796875, "tokens_processed": 15509749760, "tokens_processed_billions": 15.50974976}
+{"timestamp": 1787646956.8848073, "elapsed_hours": 5.472216084599495, "session_id": 2, "event": "train_step", "step": 118340, "epoch": 1, "loss": 1.669414918422699, "loss_raw": 2.0120527744293213, "perplexity": 5.309060654722047, "muon_lr": 0.001978824964356434, "adamw_lr": 5.9364748930693e-05, "tokens_per_sec": 790528.7953018363, "vram_gb": 2.785593032836914, "grad_norm": 0.10205078125, "tokens_processed": 15511060480, "tokens_processed_billions": 15.51106048}
+{"timestamp": 1787646964.3418837, "elapsed_hours": 5.474287494752142, "session_id": 2, "event": "train_step", "step": 118350, "epoch": 1, "loss": 1.7383844292163848, "loss_raw": 1.9283462762832642, "perplexity": 5.688146392321272, "muon_lr": 0.0019783491874962875, "adamw_lr": 5.9350475624888615e-05, "tokens_per_sec": 790295.2416931388, "vram_gb": 2.785593032836914, "grad_norm": 0.09619140625, "tokens_processed": 15512371200, "tokens_processed_billions": 15.5123712}
+{"timestamp": 1787646971.874639, "elapsed_hours": 5.4763799267345, "session_id": 2, "event": "train_step", "step": 118360, "epoch": 1, "loss": 1.7180918788909911, "loss_raw": 1.3616783618927002, "perplexity": 5.57388266759424, "muon_lr": 0.0019773979384556046, "adamw_lr": 5.932193815366814e-05, "tokens_per_sec": 790058.8189534713, "vram_gb": 2.363645076751709, "grad_norm": 0.08740234375, "tokens_processed": 15513681920, "tokens_processed_billions": 15.51368192}
+{"timestamp": 1787646979.3223014, "elapsed_hours": 5.478448721898927, "session_id": 2, "event": "train_step", "step": 118370, "epoch": 1, "loss": 1.6996885454654693, "loss_raw": 1.6373406648635864, "perplexity": 5.472242771459899, "muon_lr": 0.001976922466302497, "adamw_lr": 5.93076739890749e-05, "tokens_per_sec": 789826.0042537047, "vram_gb": 2.785593032836914, "grad_norm": 0.08984375, "tokens_processed": 15514992640, "tokens_processed_billions": 15.51499264}
+{"timestamp": 1787646986.7728508, "elapsed_hours": 5.48051831887828, "session_id": 2, "event": "train_step", "step": 118380, "epoch": 1, "loss": 1.6958967447280884, "loss_raw": 1.78841233253479, "perplexity": 5.451532406885818, "muon_lr": 0.0019764470957458306, "adamw_lr": 5.929341287237491e-05, "tokens_per_sec": 789593.2445418956, "vram_gb": 2.785593032836914, "grad_norm": 0.091796875, "tokens_processed": 15516303360, "tokens_processed_billions": 15.51630336}
+{"timestamp": 1787646994.2143664, "elapsed_hours": 5.482585406634542, "session_id": 2, "event": "train_step", "step": 118390, "epoch": 1, "loss": 1.662173924446106, "loss_raw": 1.7267922163009644, "perplexity": 5.270756625485235, "muon_lr": 0.0019759718267993104, "adamw_lr": 5.92791548039793e-05, "tokens_per_sec": 789361.0284988653, "vram_gb": 2.785593032836914, "grad_norm": 0.09765625, "tokens_processed": 15517614080, "tokens_processed_billions": 15.51761408}
+{"timestamp": 1787647001.7527418, "elapsed_hours": 5.484679399795002, "session_id": 2, "event": "train_step", "step": 118400, "epoch": 1, "loss": 1.6215137767791747, "loss_raw": 1.747532844543457, "perplexity": 5.06074535989977, "muon_lr": 0.0019750215937915363, "adamw_lr": 5.925064781374609e-05, "tokens_per_sec": 789125.0961496081, "vram_gb": 2.363645076751709, "grad_norm": 0.0966796875, "tokens_processed": 15518924800, "tokens_processed_billions": 15.5189248}
+{"timestamp": 1787647001.7544572, "elapsed_hours": 5.484679876301024, "session_id": 2, "event": "lambda_stats", "step": 118400, "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}