Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -5131,3 +5131,11 @@
 {"timestamp": 1787546536.6046553, "elapsed_hours": 5.53131004108323, "session_id": 2, "event": "train_step", "step": 45420, "epoch": 1, "loss": 1.9713290274143218, "loss_raw": 1.7151119709014893, "perplexity": 7.1802128496179565, "muon_lr": 0.008220569205758872, "adamw_lr": 0.00024661707617276615, "tokens_per_sec": 149569.0487369863, "vram_gb": 27.710740089416504, "grad_norm": 0.1455078125, "tokens_processed": 2976645120, "tokens_processed_billions": 2.97664512}
 {"timestamp": 1787546540.6154113, "elapsed_hours": 5.532424139910274, "session_id": 2, "event": "train_step", "step": 45430, "epoch": 1, "loss": 1.9762127327919006, "loss_raw": 2.218130350112915, "perplexity": 7.215364659399137, "muon_lr": 0.008219960464977562, "adamw_lr": 0.0002465988139493268, "tokens_per_sec": 149571.83508718145, "vram_gb": 27.710740089416504, "grad_norm": 0.15234375, "tokens_processed": 2977300480, "tokens_processed_billions": 2.97730048}
 {"timestamp": 1787546544.7454917, "elapsed_hours": 5.533571384482913, "session_id": 2, "event": "train_step", "step": 45440, "epoch": 1, "loss": 2.0018708968162535, "loss_raw": 2.122863292694092, "perplexity": 7.40289320031961, "muon_lr": 0.008218742748122631, "adamw_lr": 0.0002465622824436789, "tokens_per_sec": 149573.72510410906, "vram_gb": 27.299458026885986, "grad_norm": 0.1552734375, "tokens_processed": 2977955840, "tokens_processed_billions": 2.97795584}
+{"timestamp": 1787546549.5067966, "elapsed_hours": 5.534893969231182, "session_id": 2, "event": "train_step", "step": 45450, "epoch": 1, "loss": 1.954073907136917, "loss_raw": 0.8947054743766785, "perplexity": 7.057380206951061, "muon_lr": 0.008218133772084128, "adamw_lr": 0.0002465440131625238, "tokens_per_sec": 149570.87244405702, "vram_gb": 27.710740089416504, "grad_norm": 0.1337890625, "tokens_processed": 2978611200, "tokens_processed_billions": 2.9786112}
+{"timestamp": 1787546553.4851182, "elapsed_hours": 5.535999058683713, "session_id": 2, "event": "train_step", "step": 45460, "epoch": 1, "loss": 1.9027068150043487, "loss_raw": 1.6929481029510498, "perplexity": 6.704016437052046, "muon_lr": 0.00821752471766168, "adamw_lr": 0.00024652574152985037, "tokens_per_sec": 149573.90116816867, "vram_gb": 27.710740089416504, "grad_norm": 0.1396484375, "tokens_processed": 2979266560, "tokens_processed_billions": 2.97926656}
+{"timestamp": 1787546557.4427345, "elapsed_hours": 5.537098396486706, "session_id": 2, "event": "train_step", "step": 45470, "epoch": 1, "loss": 1.896302229166031, "loss_raw": 1.2890639305114746, "perplexity": 6.661217190341565, "muon_lr": 0.008216915584872853, "adamw_lr": 0.0002465074675461855, "tokens_per_sec": 149577.083359007, "vram_gb": 27.710740089416504, "grad_norm": 0.14453125, "tokens_processed": 2979921920, "tokens_processed_billions": 2.97992192}
+{"timestamp": 1787546561.5660968, "elapsed_hours": 5.5382437750366, "session_id": 2, "event": "train_step", "step": 45480, "epoch": 1, "loss": 1.8549529659748076, "loss_raw": 1.8632861375808716, "perplexity": 6.391397630394103, "muon_lr": 0.00821569708426632, "adamw_lr": 0.0002464709125279895, "tokens_per_sec": 149579.0201809728, "vram_gb": 27.299458026885986, "grad_norm": 0.173828125, "tokens_processed": 2980577280, "tokens_processed_billions": 2.98057728}
+{"timestamp": 1787546565.5548193, "elapsed_hours": 5.539351753327582, "session_id": 2, "event": "train_step", "step": 45490, "epoch": 1, "loss": 1.775053061246872, "loss_raw": 1.3651982545852661, "perplexity": 5.9005942208992606, "muon_lr": 0.008215087716483752, "adamw_lr": 0.00024645263149451253, "tokens_per_sec": 149581.9677637195, "vram_gb": 27.710740089416504, "grad_norm": 0.17578125, "tokens_processed": 2981232640, "tokens_processed_billions": 2.98123264}
+{"timestamp": 1787546569.5658875, "elapsed_hours": 5.540465938912497, "session_id": 2, "event": "train_step", "step": 45500, "epoch": 1, "loss": 1.911975017786026, "loss_raw": 2.7600529193878174, "perplexity": 6.7664394485649, "muon_lr": 0.008214478270405081, "adamw_lr": 0.00024643434811215235, "tokens_per_sec": 149584.74540529266, "vram_gb": 27.710740089416504, "grad_norm": 0.1552734375, "tokens_processed": 2981888000, "tokens_processed_billions": 2.981888}
+{"timestamp": 1787546569.5711746, "elapsed_hours": 5.540467407968309, "session_id": 2, "event": "lambda_stats", "step": 45500, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1896701388888888, "lambda_std": 0.2043699421791281, "lambda_all": [2.015625, 1.2421875, 1.203125, 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": 1787546583.022972, "elapsed_hours": 5.544204018447134, "session_id": 2, "event": "validation", "step": 45500, "epoch": 1, "val_loss": 1.9575361496210097, "val_perplexity": 7.081856916230893, "is_best": false, "best_loss": 1.9518783861398696, "best_perplexity": 7.041902577470059, "train_loss": 2.7600529193878174, "train_loss_avg": 1.911975017786026}