Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -3049,3 +3049,47 @@
 {"timestamp": 1787649092.109277, "elapsed_hours": 6.065333992905087, "session_id": 2, "event": "train_step", "step": 121000, "epoch": 1, "loss": 1.6964381551742553, "loss_raw": 0.9465924501419067, "perplexity": 5.454484722613695, "muon_lr": 0.0018260872184617792, "adamw_lr": 5.478261655385336e-05, "tokens_per_sec": 728969.5147896284, "vram_gb": 2.363645076751709, "grad_norm": 0.09326171875, "tokens_processed": 15859712000, "tokens_processed_billions": 15.859712}
 {"timestamp": 1787649092.111032, "elapsed_hours": 6.065334480404854, "session_id": 2, "event": "lambda_stats", "step": 121000, "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": 1787649117.8083577, "elapsed_hours": 6.072472626566887, "session_id": 2, "event": "validation", "step": 121000, "epoch": 1, "val_loss": 1.7513439029455184, "val_perplexity": 5.762341502443383, "is_best": false, "best_loss": 1.6233880519866943, "best_perplexity": 5.070239483980759, "train_loss": 0.9465924501419067, "train_loss_avg": 1.6964381551742553}
+{"timestamp": 1787649126.0142276, "elapsed_hours": 6.074752034743627, "session_id": 2, "event": "train_step", "step": 121010, "epoch": 1, "loss": 1.6820567727088929, "loss_raw": 1.3263062238693237, "perplexity": 5.376603056989391, "muon_lr": 0.0018256459997171885, "adamw_lr": 5.476937999151565e-05, "tokens_per_sec": 727895.4125964717, "vram_gb": 2.785593032836914, "grad_norm": 0.09326171875, "tokens_processed": 15861022720, "tokens_processed_billions": 15.86102272}
+{"timestamp": 1787649133.465741, "elapsed_hours": 6.076821899546517, "session_id": 2, "event": "train_step", "step": 121020, "epoch": 1, "loss": 1.649389500617981, "loss_raw": 1.5780837535858154, "perplexity": 5.203801939361147, "muon_lr": 0.001825204886931458, "adamw_lr": 5.475614660794374e-05, "tokens_per_sec": 727706.7134362991, "vram_gb": 2.785593032836914, "grad_norm": 0.09716796875, "tokens_processed": 15862333440, "tokens_processed_billions": 15.86233344}
+{"timestamp": 1787649140.917513, "elapsed_hours": 6.078891836272346, "session_id": 2, "event": "train_step", "step": 121030, "epoch": 1, "loss": 1.6302888488769531, "loss_raw": 1.6835377216339111, "perplexity": 5.1053491799552955, "muon_lr": 0.00182476388011731, "adamw_lr": 5.47429164035193e-05, "tokens_per_sec": 727518.1342723307, "vram_gb": 2.785593032836914, "grad_norm": 0.09228515625, "tokens_processed": 15863644160, "tokens_processed_billions": 15.86364416}
+{"timestamp": 1787649148.4664881, "elapsed_hours": 6.0809887737698025, "session_id": 2, "event": "train_step", "step": 121040, "epoch": 1, "loss": 1.6828792321681976, "loss_raw": 3.0909950733184814, "perplexity": 5.381026914004691, "muon_lr": 0.0018238821844546271, "adamw_lr": 5.47164655336388e-05, "tokens_per_sec": 727326.4463969792, "vram_gb": 2.363645076751709, "grad_norm": 0.095703125, "tokens_processed": 15864954880, "tokens_processed_billions": 15.86495488}
+{"timestamp": 1787649155.932524, "elapsed_hours": 6.083062672548824, "session_id": 2, "event": "train_step", "step": 121050, "epoch": 1, "loss": 1.6769173169136047, "loss_raw": 1.8238195180892944, "perplexity": 5.3490411305688, "muon_lr": 0.0018234414956315181, "adamw_lr": 5.4703244868945536e-05, "tokens_per_sec": 727137.6497326811, "vram_gb": 2.785593032836914, "grad_norm": 0.09326171875, "tokens_processed": 15866265600, "tokens_processed_billions": 15.8662656}
+{"timestamp": 1787649163.3757641, "elapsed_hours": 6.085130239393976, "session_id": 2, "event": "train_step", "step": 121060, "epoch": 1, "loss": 1.6509172737598419, "loss_raw": 2.4075632095336914, "perplexity": 5.211758244366653, "muon_lr": 0.0018230009128308421, "adamw_lr": 5.469002738492525e-05, "tokens_per_sec": 726949.743668123, "vram_gb": 2.785593032836914, "grad_norm": 0.08984375, "tokens_processed": 15867576320, "tokens_processed_billions": 15.86757632}
+{"timestamp": 1787649170.881306, "elapsed_hours": 6.087215111984147, "session_id": 2, "event": "train_step", "step": 121070, "epoch": 1, "loss": 1.6724808990955353, "loss_raw": 0.9201948642730713, "perplexity": 5.325363110817417, "muon_lr": 0.0018225604360653056, "adamw_lr": 5.467681308195916e-05, "tokens_per_sec": 726759.8922495174, "vram_gb": 2.785593032836914, "grad_norm": 0.09375, "tokens_processed": 15868887040, "tokens_processed_billions": 15.86888704}
+{"timestamp": 1787649178.4422204, "elapsed_hours": 6.089315366016494, "session_id": 2, "event": "train_step", "step": 121080, "epoch": 1, "loss": 1.6912324559688567, "loss_raw": 1.8237301111221313, "perplexity": 5.426164094119698, "muon_lr": 0.0018216798006904546, "adamw_lr": 5.4650394020713626e-05, "tokens_per_sec": 726568.326859953, "vram_gb": 2.363645076751709, "grad_norm": 0.09619140625, "tokens_processed": 15870197760, "tokens_processed_billions": 15.87019776}
+{"timestamp": 1787649185.9526656, "elapsed_hours": 6.091401600837708, "session_id": 2, "event": "train_step", "step": 121090, "epoch": 1, "loss": 1.6193292331695557, "loss_raw": 2.2663486003875732, "perplexity": 5.049702007697918, "muon_lr": 0.0018212396421065347, "adamw_lr": 5.463718926319603e-05, "tokens_per_sec": 726378.5765434483, "vram_gb": 2.785593032836914, "grad_norm": 0.095703125, "tokens_processed": 15871508480, "tokens_processed_billions": 15.87150848}
+{"timestamp": 1787649193.3993618, "elapsed_hours": 6.09347012758255, "session_id": 2, "event": "train_step", "step": 121100, "epoch": 1, "loss": 1.6266150403022765, "loss_raw": 1.3713990449905396, "perplexity": 5.086627515324269, "muon_lr": 0.0018207995896085457, "adamw_lr": 5.462398768825637e-05, "tokens_per_sec": 726191.072411059, "vram_gb": 2.785593032836914, "grad_norm": 0.09716796875, "tokens_processed": 15872819200, "tokens_processed_billions": 15.8728192}
+{"timestamp": 1787649193.4011004, "elapsed_hours": 6.093470610578855, "session_id": 2, "event": "lambda_stats", "step": 121100, "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": 1787649200.887705, "elapsed_hours": 6.095550222926669, "session_id": 2, "event": "train_step", "step": 121110, "epoch": 1, "loss": 1.659514387845993, "loss_raw": 1.3945186138153076, "perplexity": 5.2567574792019105, "muon_lr": 0.0018203596432091757, "adamw_lr": 5.4610789296275264e-05, "tokens_per_sec": 726002.3127825377, "vram_gb": 2.785593032836914, "grad_norm": 0.1015625, "tokens_processed": 15874129920, "tokens_processed_billions": 15.87412992}
+{"timestamp": 1787649208.479494, "elapsed_hours": 6.097659053140217, "session_id": 2, "event": "train_step", "step": 121120, "epoch": 1, "loss": 1.6461013925075532, "loss_raw": 1.2515709400177002, "perplexity": 5.186719376047658, "muon_lr": 0.0018194800687570374, "adamw_lr": 5.458440206271112e-05, "tokens_per_sec": 725810.247745402, "vram_gb": 2.363645076751709, "grad_norm": 0.0908203125, "tokens_processed": 15875440640, "tokens_processed_billions": 15.87544064}
+{"timestamp": 1787649215.9872885, "elapsed_hours": 6.099744551711613, "session_id": 2, "event": "train_step", "step": 121130, "epoch": 1, "loss": 1.641336749792099, "loss_raw": 1.4574122428894043, "perplexity": 5.162065291947606, "muon_lr": 0.0018190404407296338, "adamw_lr": 5.4571213221889014e-05, "tokens_per_sec": 725621.1064397, "vram_gb": 2.785593032836914, "grad_norm": 0.09716796875, "tokens_processed": 15876751360, "tokens_processed_billions": 15.87675136}
+{"timestamp": 1787649223.4409113, "elapsed_hours": 6.101815002361933, "session_id": 2, "event": "train_step", "step": 121140, "epoch": 1, "loss": 1.6719102263450623, "loss_raw": 1.776532530784607, "perplexity": 5.322324938187317, "muon_lr": 0.0018186009188515804, "adamw_lr": 5.455802756554741e-05, "tokens_per_sec": 725433.8873633505, "vram_gb": 2.785593032836914, "grad_norm": 0.09326171875, "tokens_processed": 15878062080, "tokens_processed_billions": 15.87806208}
+{"timestamp": 1787649230.9046881, "elapsed_hours": 6.103888273702728, "session_id": 2, "event": "train_step", "step": 121150, "epoch": 1, "loss": 1.7275656986236572, "loss_raw": 1.388953685760498, "perplexity": 5.626939556061383, "muon_lr": 0.0018181615031355497, "adamw_lr": 5.454484509406649e-05, "tokens_per_sec": 725246.4582560943, "vram_gb": 2.785593032836914, "grad_norm": 0.095703125, "tokens_processed": 15879372800, "tokens_processed_billions": 15.8793728}
+{"timestamp": 1787649238.4339046, "elapsed_hours": 6.105979722804493, "session_id": 2, "event": "train_step", "step": 121160, "epoch": 1, "loss": 1.7420697021484375, "loss_raw": 2.1879026889801025, "perplexity": 5.709147437778289, "muon_lr": 0.0018172829902402409, "adamw_lr": 5.451848970720722e-05, "tokens_per_sec": 725056.9907764381, "vram_gb": 2.363645076751709, "grad_norm": 0.1015625, "tokens_processed": 15880683520, "tokens_processed_billions": 15.88068352}
+{"timestamp": 1787649245.8763962, "elapsed_hours": 6.108047081563208, "session_id": 2, "event": "train_step", "step": 121170, "epoch": 1, "loss": 1.7155135011672973, "loss_raw": 0.9461504817008972, "perplexity": 5.559529604450404, "muon_lr": 0.001816843893086298, "adamw_lr": 5.4505316792588934e-05, "tokens_per_sec": 724870.5255067036, "vram_gb": 2.785593032836914, "grad_norm": 0.09033203125, "tokens_processed": 15881994240, "tokens_processed_billions": 15.88199424}
+{"timestamp": 1787649253.3193626, "elapsed_hours": 6.110114572180642, "session_id": 2, "event": "train_step", "step": 121180, "epoch": 1, "loss": 1.7329978108406068, "loss_raw": 1.793026328086853, "perplexity": 5.65758889312076, "muon_lr": 0.001816404902145045, "adamw_lr": 5.4492147064351344e-05, "tokens_per_sec": 724684.1707188034, "vram_gb": 2.785593032836914, "grad_norm": 0.09521484375, "tokens_processed": 15883304960, "tokens_processed_billions": 15.88330496}
+{"timestamp": 1787649260.7687287, "elapsed_hours": 6.112183842791452, "session_id": 2, "event": "train_step", "step": 121190, "epoch": 1, "loss": 1.7608433294296264, "loss_raw": 1.5603681802749634, "perplexity": 5.817341261412196, "muon_lr": 0.0018159660174291445, "adamw_lr": 5.447898052287432e-05, "tokens_per_sec": 724497.7384130378, "vram_gb": 2.785593032836914, "grad_norm": 0.1005859375, "tokens_processed": 15884615680, "tokens_processed_billions": 15.88461568}
+{"timestamp": 1787649268.3168602, "elapsed_hours": 6.114280543790923, "session_id": 2, "event": "train_step", "step": 121200, "epoch": 1, "loss": 1.6836962699890137, "loss_raw": 1.5941717624664307, "perplexity": 5.3854252130519225, "muon_lr": 0.0018150885667240174, "adamw_lr": 5.445265700172051e-05, "tokens_per_sec": 724308.1510945621, "vram_gb": 2.363645076751709, "grad_norm": 0.0947265625, "tokens_processed": 15885926400, "tokens_processed_billions": 15.8859264}
+{"timestamp": 1787649268.3185492, "elapsed_hours": 6.11428101307816, "session_id": 2, "event": "lambda_stats", "step": 121200, "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": 1787649275.7756782, "elapsed_hours": 6.116352437668376, "session_id": 2, "event": "train_step", "step": 121210, "epoch": 1, "loss": 1.6637189626693725, "loss_raw": 1.852385401725769, "perplexity": 5.2789064402035395, "muon_lr": 0.0018146500007600928, "adamw_lr": 5.443950002280278e-05, "tokens_per_sec": 724121.666331097, "vram_gb": 2.785593032836914, "grad_norm": 0.0927734375, "tokens_processed": 15887237120, "tokens_processed_billions": 15.88723712}
+{"timestamp": 1787649283.2418737, "elapsed_hours": 6.118426380885972, "session_id": 2, "event": "train_step", "step": 121220, "epoch": 1, "loss": 1.6451105654239655, "loss_raw": 1.8192899227142334, "perplexity": 5.181582779174784, "muon_lr": 0.0018142115410721272, "adamw_lr": 5.442634623216381e-05, "tokens_per_sec": 723935.0406949851, "vram_gb": 2.785593032836914, "grad_norm": 0.083984375, "tokens_processed": 15888547840, "tokens_processed_billions": 15.88854784}
+{"timestamp": 1787649290.6802251, "elapsed_hours": 6.1204925896724065, "session_id": 2, "event": "train_step", "step": 121230, "epoch": 1, "loss": 1.6399261915683747, "loss_raw": 1.4216293096542358, "perplexity": 5.154789031299744, "muon_lr": 0.001813773187672764, "adamw_lr": 5.4413195630182906e-05, "tokens_per_sec": 723749.4759426486, "vram_gb": 2.785593032836914, "grad_norm": 0.09912109375, "tokens_processed": 15889858560, "tokens_processed_billions": 15.88985856}
+{"timestamp": 1787649298.262085, "elapsed_hours": 6.122598661780358, "session_id": 2, "event": "train_step", "step": 121240, "epoch": 1, "loss": 1.6017179763317109, "loss_raw": 1.579278588294983, "perplexity": 4.96154893035359, "muon_lr": 0.0018128967997904006, "adamw_lr": 5.4386903993712004e-05, "tokens_per_sec": 723559.2977592013, "vram_gb": 2.363645076751709, "grad_norm": 0.091796875, "tokens_processed": 15891169280, "tokens_processed_billions": 15.89116928}
+{"timestamp": 1787649305.6937263, "elapsed_hours": 6.124663006530867, "session_id": 2, "event": "train_step", "step": 121250, "epoch": 1, "loss": 1.6046716845035554, "loss_raw": 1.5389907360076904, "perplexity": 4.9762255625480165, "muon_lr": 0.001812458765332674, "adamw_lr": 5.437376295998022e-05, "tokens_per_sec": 723374.2045246856, "vram_gb": 2.785593032836914, "grad_norm": 0.0947265625, "tokens_processed": 15892480000, "tokens_processed_billions": 15.89248}
+{"timestamp": 1787649313.150757, "elapsed_hours": 6.126734404100312, "session_id": 2, "event": "train_step", "step": 121260, "epoch": 1, "loss": 1.6388399767875672, "loss_raw": 1.2135918140411377, "perplexity": 5.149192863132517, "muon_lr": 0.0018120208372140961, "adamw_lr": 5.4360625116422874e-05, "tokens_per_sec": 723188.4002799817, "vram_gb": 2.785593032836914, "grad_norm": 0.09765625, "tokens_processed": 15893790720, "tokens_processed_billions": 15.89379072}
+{"timestamp": 1787649320.612862, "elapsed_hours": 6.128807211054696, "session_id": 2, "event": "train_step", "step": 121270, "epoch": 1, "loss": 1.6854880273342132, "loss_raw": 1.5611549615859985, "perplexity": 5.395083238069224, "muon_lr": 0.001811583015447294, "adamw_lr": 5.4347490463418814e-05, "tokens_per_sec": 723002.5540617161, "vram_gb": 2.785593032836914, "grad_norm": 0.09375, "tokens_processed": 15895101440, "tokens_processed_billions": 15.89510144}
+{"timestamp": 1787649328.181024, "elapsed_hours": 6.130909478267034, "session_id": 2, "event": "train_step", "step": 121280, "epoch": 1, "loss": 1.6793428480625152, "loss_raw": 1.6762138605117798, "perplexity": 5.36203114392023, "muon_lr": 0.0018107076910195152, "adamw_lr": 5.4321230730585446e-05, "tokens_per_sec": 722813.347199522, "vram_gb": 2.363645076751709, "grad_norm": 0.09716796875, "tokens_processed": 15896412160, "tokens_processed_billions": 15.89641216}
+{"timestamp": 1787649335.6203814, "elapsed_hours": 6.13297596640057, "session_id": 2, "event": "train_step", "step": 121290, "epoch": 1, "loss": 1.687800874710083, "loss_raw": 1.756060004234314, "perplexity": 5.407575683169646, "muon_lr": 0.0018102701883837813, "adamw_lr": 5.4308105651513426e-05, "tokens_per_sec": 722628.5041371874, "vram_gb": 2.785593032836914, "grad_norm": 0.09033203125, "tokens_processed": 15897722880, "tokens_processed_billions": 15.89772288}
+{"timestamp": 1787649343.1164618, "elapsed_hours": 6.135058210889499, "session_id": 2, "event": "train_step", "step": 121300, "epoch": 1, "loss": 1.7265835523605346, "loss_raw": 1.8358752727508545, "perplexity": 5.621415791420005, "muon_lr": 0.0018098327921503082, "adamw_lr": 5.429498376450924e-05, "tokens_per_sec": 722441.923728947, "vram_gb": 2.785593032836914, "grad_norm": 0.1005859375, "tokens_processed": 15899033600, "tokens_processed_billions": 15.8990336}
+{"timestamp": 1787649343.1183362, "elapsed_hours": 6.135058731569184, "session_id": 2, "event": "lambda_stats", "step": 121300, "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": 1787649350.5701685, "elapsed_hours": 6.137128684984313, "session_id": 2, "event": "train_step", "step": 121310, "epoch": 1, "loss": 1.697241291999817, "loss_raw": 0.9174508452415466, "perplexity": 5.45886717977969, "muon_lr": 0.0018093955023317084, "adamw_lr": 5.428186506995125e-05, "tokens_per_sec": 722256.8596496056, "vram_gb": 2.785593032836914, "grad_norm": 0.09765625, "tokens_processed": 15900344320, "tokens_processed_billions": 15.90034432}
+{"timestamp": 1787649358.1343656, "elapsed_hours": 6.139229850769043, "session_id": 2, "event": "train_step", "step": 121320, "epoch": 1, "loss": 1.6803965485095977, "loss_raw": 2.3750932216644287, "perplexity": 5.367684096270027, "muon_lr": 0.0018085212419895664, "adamw_lr": 5.4255637259686984e-05, "tokens_per_sec": 722068.2973408632, "vram_gb": 2.363645076751709, "grad_norm": 0.091796875, "tokens_processed": 15901655040, "tokens_processed_billions": 15.90165504}
+{"timestamp": 1787649365.579668, "elapsed_hours": 6.141297990414832, "session_id": 2, "event": "train_step", "step": 121330, "epoch": 1, "loss": 1.7330684173107147, "loss_raw": 1.5392851829528809, "perplexity": 5.657988369604471, "muon_lr": 0.0018080842714912355, "adamw_lr": 5.424252814473706e-05, "tokens_per_sec": 721883.7614810233, "vram_gb": 2.785593032836914, "grad_norm": 0.1044921875, "tokens_processed": 15902965760, "tokens_processed_billions": 15.90296576}
+{"timestamp": 1787649373.0530026, "elapsed_hours": 6.143373916678958, "session_id": 2, "event": "train_step", "step": 121340, "epoch": 1, "loss": 1.784141321182251, "loss_raw": 1.0751359462738037, "perplexity": 5.954464778976679, "muon_lr": 0.0018076474074582016, "adamw_lr": 5.422942222374604e-05, "tokens_per_sec": 721698.4342027868, "vram_gb": 2.785593032836914, "grad_norm": 0.10302734375, "tokens_processed": 15904276480, "tokens_processed_billions": 15.90427648}
+{"timestamp": 1787649380.5115256, "elapsed_hours": 6.145445728566911, "session_id": 2, "event": "train_step", "step": 121350, "epoch": 1, "loss": 1.7483252954483033, "loss_raw": 1.094406008720398, "perplexity": 5.744973481989073, "muon_lr": 0.0018072106499030624, "adamw_lr": 5.4216319497091866e-05, "tokens_per_sec": 721513.7126849322, "vram_gb": 2.785593032836914, "grad_norm": 0.0947265625, "tokens_processed": 15905587200, "tokens_processed_billions": 15.9055872}
+{"timestamp": 1787649388.0502257, "elapsed_hours": 6.147539811929067, "session_id": 2, "event": "train_step", "step": 121360, "epoch": 1, "loss": 1.7792345201969146, "loss_raw": 1.592943787574768, "perplexity": 5.92531897015165, "muon_lr": 0.0018063374542768426, "adamw_lr": 5.4190123628305274e-05, "tokens_per_sec": 721326.4953823445, "vram_gb": 2.363645076751709, "grad_norm": 0.1015625, "tokens_processed": 15906897920, "tokens_processed_billions": 15.90689792}
+{"timestamp": 1787649395.5092993, "elapsed_hours": 6.149611776868502, "session_id": 2, "event": "train_step", "step": 121370, "epoch": 1, "loss": 1.7846658957004546, "loss_raw": 1.9834753274917603, "perplexity": 5.957589158882667, "muon_lr": 0.0018059010162309423, "adamw_lr": 5.417703048692826e-05, "tokens_per_sec": 721142.0119846746, "vram_gb": 2.785593032836914, "grad_norm": 0.08984375, "tokens_processed": 15908208640, "tokens_processed_billions": 15.90820864}
+{"timestamp": 1787649402.9983983, "elapsed_hours": 6.151692082219654, "session_id": 2, "event": "train_step", "step": 121380, "epoch": 1, "loss": 1.7068820440769195, "loss_raw": 1.9971802234649658, "perplexity": 5.511749266837901, "muon_lr": 0.0018054646847132988, "adamw_lr": 5.416394054139896e-05, "tokens_per_sec": 720956.6694281338, "vram_gb": 2.785593032836914, "grad_norm": 0.09521484375, "tokens_processed": 15909519360, "tokens_processed_billions": 15.90951936}
+{"timestamp": 1787649410.4710007, "elapsed_hours": 6.153767804967033, "session_id": 2, "event": "train_step", "step": 121390, "epoch": 1, "loss": 1.6691827285289764, "loss_raw": 2.1320412158966064, "perplexity": 5.307828087593213, "muon_lr": 0.0018050284597364934, "adamw_lr": 5.41508537920948e-05, "tokens_per_sec": 720771.9910997802, "vram_gb": 2.785593032836914, "grad_norm": 0.08984375, "tokens_processed": 15910830080, "tokens_processed_billions": 15.91083008}
+{"timestamp": 1787649418.0009794, "elapsed_hours": 6.155859465797742, "session_id": 2, "event": "train_step", "step": 121400, "epoch": 1, "loss": 1.6677476871013641, "loss_raw": 1.7530258893966675, "perplexity": 5.300216597105184, "muon_lr": 0.0018041563294557126, "adamw_lr": 5.412468988367137e-05, "tokens_per_sec": 720585.5652693575, "vram_gb": 2.363645076751709, "grad_norm": 0.09375, "tokens_processed": 15912140800, "tokens_processed_billions": 15.9121408}
+{"timestamp": 1787649418.0028183, "elapsed_hours": 6.155859976609548, "session_id": 2, "event": "lambda_stats", "step": 121400, "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}