Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -4285,3 +4285,9 @@
 {"timestamp": 1787543204.3011646, "elapsed_hours": 4.605670182704926, "session_id": 2, "event": "train_step", "step": 37940, "epoch": 1, "loss": 1.961496546268463, "loss_raw": 1.3462512493133545, "perplexity": 7.109959490558461, "muon_lr": 0.008753095501965461, "adamw_lr": 0.0002625928650589638, "tokens_per_sec": 150063.95586875602, "vram_gb": 27.710740089416504, "grad_norm": 0.302734375, "tokens_processed": 2486435840, "tokens_processed_billions": 2.48643584}
 {"timestamp": 1787543208.346204, "elapsed_hours": 4.606793804764748, "session_id": 2, "event": "train_step", "step": 37950, "epoch": 1, "loss": 1.912974396944046, "loss_raw": 1.2650855779647827, "perplexity": 6.773205067269861, "muon_lr": 0.008752567457026806, "adamw_lr": 0.0002625770237108041, "tokens_per_sec": 150066.87257173422, "vram_gb": 27.710740089416504, "grad_norm": 0.1474609375, "tokens_processed": 2487091200, "tokens_processed_billions": 2.4870912}
 {"timestamp": 1787543212.4829512, "elapsed_hours": 4.607942901055018, "session_id": 2, "event": "train_step", "step": 37960, "epoch": 1, "loss": 1.942787810564041, "loss_raw": 3.588768482208252, "perplexity": 6.978177716567638, "muon_lr": 0.008751511085777843, "adamw_lr": 0.0002625453325733352, "tokens_per_sec": 150068.95759737212, "vram_gb": 27.299458026885986, "grad_norm": 0.162109375, "tokens_processed": 2487746560, "tokens_processed_billions": 2.48774656}
+{"timestamp": 1787543217.0962403, "elapsed_hours": 4.609224370453092, "session_id": 2, "event": "train_step", "step": 37970, "epoch": 1, "loss": 1.936397718191147, "loss_raw": 1.2699451446533203, "perplexity": 6.933728684325137, "muon_lr": 0.008750982759497993, "adamw_lr": 0.00026252948278493975, "tokens_per_sec": 150066.7298805391, "vram_gb": 27.710740089416504, "grad_norm": 0.14453125, "tokens_processed": 2488401920, "tokens_processed_billions": 2.48840192}
+{"timestamp": 1787543221.1289098, "elapsed_hours": 4.610344556238916, "session_id": 2, "event": "train_step", "step": 37980, "epoch": 1, "loss": 1.9427249240875244, "loss_raw": 0.6626055836677551, "perplexity": 6.977738897356578, "muon_lr": 0.008750454339468215, "adamw_lr": 0.0002625136301840464, "tokens_per_sec": 150069.75521888616, "vram_gb": 27.710740089416504, "grad_norm": 0.15234375, "tokens_processed": 2489057280, "tokens_processed_billions": 2.48905728}
+{"timestamp": 1787543225.116452, "elapsed_hours": 4.611452206969261, "session_id": 2, "event": "train_step", "step": 37990, "epoch": 1, "loss": 2.109716844558716, "loss_raw": 2.227025270462036, "perplexity": 8.245906080854326, "muon_lr": 0.008749925825703745, "adamw_lr": 0.0002624977747711123, "tokens_per_sec": 150073.18821390433, "vram_gb": 27.710740089416504, "grad_norm": 0.1533203125, "tokens_processed": 2489712640, "tokens_processed_billions": 2.48971264}
+{"timestamp": 1787543229.2395773, "elapsed_hours": 4.612597519424226, "session_id": 2, "event": "train_step", "step": 38000, "epoch": 1, "loss": 2.1310717165470123, "loss_raw": 1.6460421085357666, "perplexity": 8.42388999772823, "muon_lr": 0.008748868517031695, "adamw_lr": 0.0002624660555109508, "tokens_per_sec": 150075.39279513888, "vram_gb": 27.299458026885986, "grad_norm": 0.14453125, "tokens_processed": 2490368000, "tokens_processed_billions": 2.490368}
+{"timestamp": 1787543229.2461166, "elapsed_hours": 4.612599336041344, "session_id": 2, "event": "lambda_stats", "step": 38000, "epoch": 1, "lambda_min": 1.109375, "lambda_max": 2.015625, "lambda_mean": 1.1957465277777777, "lambda_std": 0.2024305741673805, "lambda_all": [2.015625, 1.25, 1.2109375, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.125, 1.140625, 1.109375, 1.140625, 1.109375, 1.125, 1.125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787543242.8706596, "elapsed_hours": 4.616383931570583, "session_id": 2, "event": "validation", "step": 38000, "epoch": 1, "val_loss": 1.9904787677526474, "val_perplexity": 7.319037042530245, "is_best": true, "best_loss": 1.9904787677526474, "best_perplexity": 7.319037042530245, "train_loss": 1.6460421085357666, "train_loss_avg": 2.1310717165470123}