Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -619,3 +619,11 @@
 {"timestamp": 1787529002.6720486, "elapsed_hours": 0.6607732059558232, "session_id": 2, "event": "train_step", "step": 5500, "epoch": 1, "loss": 2.7483660340309144, "loss_raw": 2.843714714050293, "perplexity": 15.617093226353395, "muon_lr": 0.009987621899887528, "adamw_lr": 0.00029962865699662575, "tokens_per_sec": 152245.44949650436, "vram_gb": 27.710740089416504, "grad_norm": 0.2294921875, "tokens_processed": 360448000, "tokens_processed_billions": 0.360448}
 {"timestamp": 1787529002.6819, "elapsed_hours": 0.6607759425375197, "session_id": 2, "event": "lambda_stats", "step": 5500, "epoch": 1, "lambda_min": 1.140625, "lambda_max": 1.6953125, "lambda_mean": 1.2287326388888888, "lambda_std": 0.13214203612304679, "lambda_all": [1.6953125, 1.390625, 1.3046875, 1.296875, 1.1875, 1.1796875, 1.140625, 1.25, 1.1875, 1.140625, 1.15625, 1.140625, 1.140625, 1.1875, 1.140625, 1.21875, 1.140625, 1.21875], "local_layers": 9, "global_layers": 9}
 {"timestamp": 1787529016.0739782, "elapsed_hours": 0.6644959642489752, "session_id": 2, "event": "validation", "step": 5500, "epoch": 1, "val_loss": 2.756121916770935, "val_perplexity": 15.738688499965537, "is_best": true, "best_loss": 2.756121916770935, "best_perplexity": 15.738688499965537, "train_loss": 2.843714714050293, "train_loss_avg": 2.7483660340309144}
+{"timestamp": 1787529020.5284762, "elapsed_hours": 0.6657333246866862, "session_id": 2, "event": "train_step", "step": 5510, "epoch": 1, "loss": 2.670591638088226, "loss_raw": 2.819758653640747, "perplexity": 14.448514956320475, "muon_lr": 0.009987565194622346, "adamw_lr": 0.0002996269558386703, "tokens_per_sec": 151380.52293193366, "vram_gb": 27.710740089416504, "grad_norm": 0.2294921875, "tokens_processed": 361103360, "tokens_processed_billions": 0.36110336}
+{"timestamp": 1787529024.595297, "elapsed_hours": 0.6668629972139994, "session_id": 2, "event": "train_step", "step": 5520, "epoch": 1, "loss": 2.493804783821106, "loss_raw": 2.4754271507263184, "perplexity": 12.107254081324331, "muon_lr": 0.009987451395864315, "adamw_lr": 0.0002996235418759294, "tokens_per_sec": 151397.140544717, "vram_gb": 27.299458026885986, "grad_norm": 0.4140625, "tokens_processed": 361758720, "tokens_processed_billions": 0.36175872}
+{"timestamp": 1787529028.5336971, "elapsed_hours": 0.6679569970899158, "session_id": 2, "event": "train_step", "step": 5530, "epoch": 1, "loss": 2.5574221658706664, "loss_raw": 1.4669238328933716, "perplexity": 12.902513868000511, "muon_lr": 0.00998739430237475, "adamw_lr": 0.00029962182907124245, "tokens_per_sec": 151421.83580108502, "vram_gb": 27.710740089416504, "grad_norm": 0.232421875, "tokens_processed": 362414080, "tokens_processed_billions": 0.36241408}
+{"timestamp": 1787529032.4473848, "elapsed_hours": 0.6690441326300303, "session_id": 2, "event": "train_step", "step": 5540, "epoch": 1, "loss": 2.584288227558136, "loss_raw": 3.177670478820801, "perplexity": 13.253852005232167, "muon_lr": 0.009987337079480345, "adamw_lr": 0.00029962011238441027, "tokens_per_sec": 151448.01116632708, "vram_gb": 27.710740089416504, "grad_norm": 0.244140625, "tokens_processed": 363069440, "tokens_processed_billions": 0.36306944}
+{"timestamp": 1787529036.4004455, "elapsed_hours": 0.6701422050264146, "session_id": 2, "event": "train_step", "step": 5550, "epoch": 1, "loss": 2.4391718912124634, "loss_raw": 2.729775905609131, "perplexity": 11.463543749750382, "muon_lr": 0.00998727972718275, "adamw_lr": 0.00029961839181548243, "tokens_per_sec": 151471.61117378683, "vram_gb": 27.710740089416504, "grad_norm": 0.2109375, "tokens_processed": 363724800, "tokens_processed_billions": 0.3637248}
+{"timestamp": 1787529040.500608, "elapsed_hours": 0.6712811391221153, "session_id": 2, "event": "train_step", "step": 5560, "epoch": 1, "loss": 2.5526931262016297, "loss_raw": 2.9633867740631104, "perplexity": 12.841641415654374, "muon_lr": 0.00998716463438461, "adamw_lr": 0.0002996149390315382, "tokens_per_sec": 151485.87175594547, "vram_gb": 27.299458026885986, "grad_norm": 0.232421875, "tokens_processed": 364380160, "tokens_processed_billions": 0.36438016}
+{"timestamp": 1787529044.4326587, "elapsed_hours": 0.6723733752965927, "session_id": 2, "event": "train_step", "step": 5570, "epoch": 1, "loss": 2.57450975894928, "loss_raw": 2.8271102905273438, "perplexity": 13.124881225477077, "muon_lr": 0.009987106893887383, "adamw_lr": 0.00029961320681662145, "tokens_per_sec": 151510.6516098949, "vram_gb": 27.710740089416504, "grad_norm": 0.1953125, "tokens_processed": 365035520, "tokens_processed_billions": 0.36503552}
+{"timestamp": 1787529048.3997927, "elapsed_hours": 0.6734753569629457, "session_id": 2, "event": "train_step", "step": 5580, "epoch": 1, "loss": 2.631426742076874, "loss_raw": 1.450372338294983, "perplexity": 13.893578320806187, "muon_lr": 0.009987049023993604, "adamw_lr": 0.00029961147071980805, "tokens_per_sec": 151533.15642863564, "vram_gb": 27.710740089416504, "grad_norm": 0.2412109375, "tokens_processed": 365690880, "tokens_processed_billions": 0.36569088}