Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -549,3 +549,17 @@
 {"timestamp": 1787528737.1936774, "elapsed_hours": 0.5870292139715618, "session_id": 2, "event": "train_step", "step": 4870, "epoch": 1, "loss": 2.839409728050232, "loss_raw": 2.9670908451080322, "perplexity": 17.105665562009754, "muon_lr": 0.009991692547662107, "adamw_lr": 0.0002997507764298631, "tokens_per_sec": 151831.66428765864, "vram_gb": 27.710740089416504, "grad_norm": 0.2109375, "tokens_processed": 319160320, "tokens_processed_billions": 0.31916032}
 {"timestamp": 1787528741.2884698, "elapsed_hours": 0.5881666562292311, "session_id": 2, "event": "train_step", "step": 4880, "epoch": 1, "loss": 2.829335081577301, "loss_raw": 2.06535005569458, "perplexity": 16.934197220561533, "muon_lr": 0.009991599464068936, "adamw_lr": 0.00029974798392206806, "tokens_per_sec": 151847.6314814435, "vram_gb": 27.299458026885986, "grad_norm": 0.2490234375, "tokens_processed": 319815680, "tokens_processed_billions": 0.31981568}
 {"timestamp": 1787528745.2312198, "elapsed_hours": 0.5892618645562065, "session_id": 2, "event": "train_step", "step": 4890, "epoch": 1, "loss": 2.860183184146881, "loss_raw": 3.443005323410034, "perplexity": 17.464725904486265, "muon_lr": 0.009991552727982522, "adamw_lr": 0.00029974658183947557, "tokens_per_sec": 151874.49196729137, "vram_gb": 27.710740089416504, "grad_norm": 0.25, "tokens_processed": 320471040, "tokens_processed_billions": 0.32047104}
+{"timestamp": 1787528749.5827973, "elapsed_hours": 0.5904706360896429, "session_id": 2, "event": "train_step", "step": 4900, "epoch": 1, "loss": 2.819428737163544, "loss_raw": 2.598177433013916, "perplexity": 16.76726941780593, "muon_lr": 0.00999150586237135, "adamw_lr": 0.0002997451758711404, "tokens_per_sec": 151871.86427148804, "vram_gb": 27.710740089416504, "grad_norm": 0.22265625, "tokens_processed": 321126400, "tokens_processed_billions": 0.3211264}
+{"timestamp": 1787528749.5886672, "elapsed_hours": 0.5904722668064966, "session_id": 2, "event": "lambda_stats", "step": 4900, "epoch": 1, "lambda_min": 1.140625, "lambda_max": 1.6484375, "lambda_mean": 1.2317708333333333, "lambda_std": 0.12371847416895523, "lambda_all": [1.6484375, 1.390625, 1.3203125, 1.3125, 1.1875, 1.1875, 1.140625, 1.25, 1.1875, 1.140625, 1.1640625, 1.140625, 1.140625, 1.1953125, 1.140625, 1.234375, 1.140625, 1.25], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787528753.5221615, "elapsed_hours": 0.5915649038553238, "session_id": 2, "event": "train_step", "step": 4910, "epoch": 1, "loss": 2.806440668106079, "loss_raw": 2.370434522628784, "perplexity": 16.55090309662286, "muon_lr": 0.009991458867236767, "adamw_lr": 0.000299743766017103, "tokens_per_sec": 151898.82098997216, "vram_gb": 27.710740089416504, "grad_norm": 0.2373046875, "tokens_processed": 321781760, "tokens_processed_billions": 0.32178176}
+{"timestamp": 1787528757.604219, "elapsed_hours": 0.5926988089084625, "session_id": 2, "event": "train_step", "step": 4920, "epoch": 1, "loss": 2.810022807121277, "loss_raw": 1.7119619846343994, "perplexity": 16.610297047525897, "muon_lr": 0.009991364488402814, "adamw_lr": 0.00029974093465208435, "tokens_per_sec": 151915.4476136982, "vram_gb": 27.299458026885986, "grad_norm": 0.2421875, "tokens_processed": 322437120, "tokens_processed_billions": 0.32243712}
+{"timestamp": 1787528761.5492237, "elapsed_hours": 0.5937946433491177, "session_id": 2, "event": "train_step", "step": 4930, "epoch": 1, "loss": 2.7508825063705444, "loss_raw": 2.3355484008789062, "perplexity": 15.65644269964802, "muon_lr": 0.00999131710470616, "adamw_lr": 0.00029973951314118475, "tokens_per_sec": 151941.812547247, "vram_gb": 27.710740089416504, "grad_norm": 0.224609375, "tokens_processed": 323092480, "tokens_processed_billions": 0.32309248}
+{"timestamp": 1787528765.520572, "elapsed_hours": 0.5948977957831488, "session_id": 2, "event": "train_step", "step": 4940, "epoch": 1, "loss": 2.7133845472335816, "loss_raw": 3.0022284984588623, "perplexity": 15.08022898631741, "muon_lr": 0.00999126959149154, "adamw_lr": 0.00029973808774474617, "tokens_per_sec": 151966.20209051252, "vram_gb": 27.710740089416504, "grad_norm": 0.2265625, "tokens_processed": 323747840, "tokens_processed_billions": 0.32374784}
+{"timestamp": 1787528769.476604, "elapsed_hours": 0.5959966934389538, "session_id": 2, "event": "train_step", "step": 4950, "epoch": 1, "loss": 2.7345314598083497, "loss_raw": 2.79874849319458, "perplexity": 15.402525042972231, "muon_lr": 0.009991221948760332, "adamw_lr": 0.0002997366584628099, "tokens_per_sec": 151991.57942937358, "vram_gb": 27.710740089416504, "grad_norm": 0.2041015625, "tokens_processed": 324403200, "tokens_processed_billions": 0.3244032}
+{"timestamp": 1787528773.569561, "elapsed_hours": 0.5971336260769102, "session_id": 2, "event": "train_step", "step": 4960, "epoch": 1, "loss": 2.802748682498932, "loss_raw": 2.3828744888305664, "perplexity": 16.48991006258849, "muon_lr": 0.009991126274753629, "adamw_lr": 0.0002997337882426088, "tokens_per_sec": 152007.127192082, "vram_gb": 27.299458026885986, "grad_norm": 0.2158203125, "tokens_processed": 325058560, "tokens_processed_billions": 0.32505856}
+{"timestamp": 1787528777.5193684, "elapsed_hours": 0.5982307946681976, "session_id": 2, "event": "train_step", "step": 4970, "epoch": 1, "loss": 2.7843938207626344, "loss_raw": 3.3632779121398926, "perplexity": 16.19000085925336, "muon_lr": 0.009991078243480899, "adamw_lr": 0.0002997323473044269, "tokens_per_sec": 152032.7855402536, "vram_gb": 27.710740089416504, "grad_norm": 0.232421875, "tokens_processed": 325713920, "tokens_processed_billions": 0.32571392}
+{"timestamp": 1787528781.4612262, "elapsed_hours": 0.599325755238533, "session_id": 2, "event": "train_step", "step": 4980, "epoch": 1, "loss": 2.7926758670806886, "loss_raw": 2.0801892280578613, "perplexity": 16.32464398694651, "muon_lr": 0.00999103008269709, "adamw_lr": 0.0002997309024809127, "tokens_per_sec": 152058.91133031272, "vram_gb": 27.710740089416504, "grad_norm": 0.2197265625, "tokens_processed": 326369280, "tokens_processed_billions": 0.32636928}
+{"timestamp": 1787528785.4329662, "elapsed_hours": 0.6004290163516999, "session_id": 2, "event": "train_step", "step": 4990, "epoch": 1, "loss": 2.8285541224479673, "loss_raw": 2.3297739028930664, "perplexity": 16.920977467359567, "muon_lr": 0.009990981792403597, "adamw_lr": 0.0002997294537721079, "tokens_per_sec": 152082.8240255611, "vram_gb": 27.710740089416504, "grad_norm": 0.26171875, "tokens_processed": 327024640, "tokens_processed_billions": 0.32702464}
+{"timestamp": 1787528789.5252805, "elapsed_hours": 0.6015657703081767, "session_id": 2, "event": "train_step", "step": 5000, "epoch": 1, "loss": 2.727938725948334, "loss_raw": 2.1073598861694336, "perplexity": 15.301314288960437, "muon_lr": 0.009990884823293127, "adamw_lr": 0.00029972654469879377, "tokens_per_sec": 152098.13425009273, "vram_gb": 27.299458026885986, "grad_norm": 0.228515625, "tokens_processed": 327680000, "tokens_processed_billions": 0.32768}
+{"timestamp": 1787528789.5309296, "elapsed_hours": 0.601567339764701, "session_id": 2, "event": "lambda_stats", "step": 5000, "epoch": 1, "lambda_min": 1.140625, "lambda_max": 1.6484375, "lambda_mean": 1.2300347222222223, "lambda_std": 0.12358745677232225, "lambda_all": [1.6484375, 1.390625, 1.3125, 1.3125, 1.1875, 1.1875, 1.140625, 1.25, 1.1875, 1.140625, 1.15625, 1.140625, 1.140625, 1.1953125, 1.140625, 1.234375, 1.140625, 1.234375], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787528802.9331946, "elapsed_hours": 0.6052901911735534, "session_id": 2, "event": "validation", "step": 5000, "epoch": 1, "val_loss": 2.8289759135246277, "val_perplexity": 16.928116090061515, "is_best": true, "best_loss": 2.8289759135246277, "best_perplexity": 16.928116090061515, "train_loss": 2.1073598861694336, "train_loss_avg": 2.727938725948334}