Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -9885,3 +9885,11 @@
 {"timestamp": 1787564928.6986785, "elapsed_hours": 10.640225047535367, "session_id": 2, "event": "train_step", "step": 87500, "epoch": 1, "loss": 1.7823043644428254, "loss_raw": 2.4452171325683594, "perplexity": 5.94353672502542, "muon_lr": 0.004412844775729655, "adamw_lr": 0.00013238534327188962, "tokens_per_sec": 149748.3708163774, "vram_gb": 27.710740089416504, "grad_norm": 0.1337890625, "tokens_processed": 5734400000, "tokens_processed_billions": 5.7344}
 {"timestamp": 1787564928.7022176, "elapsed_hours": 10.640226030813324, "session_id": 2, "event": "lambda_stats", "step": 87500, "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": 1787564942.2452385, "elapsed_hours": 10.643987981279691, "session_id": 2, "event": "validation", "step": 87500, "epoch": 1, "val_loss": 1.8211108577251434, "val_perplexity": 6.178718315983297, "is_best": true, "best_loss": 1.8211108577251434, "best_perplexity": 6.178718315983297, "train_loss": 2.4452171325683594, "train_loss_avg": 1.7823043644428254}
+{"timestamp": 1787564947.2459807, "elapsed_hours": 10.645377076003287, "session_id": 2, "event": "train_step", "step": 87510, "epoch": 1, "loss": 1.7918007838726044, "loss_raw": 1.0616458654403687, "perplexity": 6.0002478929880665, "muon_lr": 0.004412103256192387, "adamw_lr": 0.0001323630976857716, "tokens_per_sec": 149692.9819336723, "vram_gb": 27.710740089416504, "grad_norm": 0.1376953125, "tokens_processed": 5735055360, "tokens_processed_billions": 5.73505536}
+{"timestamp": 1787564951.3841386, "elapsed_hours": 10.64652656422721, "session_id": 2, "event": "train_step", "step": 87520, "epoch": 1, "loss": 1.8201546859741211, "loss_raw": 2.4518649578094482, "perplexity": 6.172813223663163, "muon_lr": 0.004410620311255614, "adamw_lr": 0.0001323186093376684, "tokens_per_sec": 149693.91861931636, "vram_gb": 27.299458026885986, "grad_norm": 0.146484375, "tokens_processed": 5735710720, "tokens_processed_billions": 5.73571072}
+{"timestamp": 1787564955.3702815, "elapsed_hours": 10.647633826269043, "session_id": 2, "event": "train_step", "step": 87530, "epoch": 1, "loss": 1.845431901216507, "loss_raw": 1.4933489561080933, "perplexity": 6.3308334947023335, "muon_lr": 0.004409878885898872, "adamw_lr": 0.00013229636657696614, "tokens_per_sec": 149695.45010167273, "vram_gb": 27.710740089416504, "grad_norm": 0.1416015625, "tokens_processed": 5736366080, "tokens_processed_billions": 5.73636608}
+{"timestamp": 1787564959.5162401, "elapsed_hours": 10.648785481386714, "session_id": 2, "event": "train_step", "step": 87540, "epoch": 1, "loss": 1.7871901381015778, "loss_raw": 2.237830877304077, "perplexity": 5.972646554308757, "muon_lr": 0.004409137491978405, "adamw_lr": 0.00013227412475935212, "tokens_per_sec": 149696.35565268586, "vram_gb": 27.710740089416504, "grad_norm": 0.1376953125, "tokens_processed": 5737021440, "tokens_processed_billions": 5.73702144}
+{"timestamp": 1787564963.5002542, "elapsed_hours": 10.649892152084245, "session_id": 2, "event": "train_step", "step": 87550, "epoch": 1, "loss": 1.8070262336730958, "loss_raw": 0.8731624484062195, "perplexity": 6.092303380869894, "muon_lr": 0.004408396129515589, "adamw_lr": 0.00013225188388546764, "tokens_per_sec": 149697.89478181672, "vram_gb": 27.710740089416504, "grad_norm": 0.138671875, "tokens_processed": 5737676800, "tokens_processed_billions": 5.7376768}
+{"timestamp": 1787564967.7574763, "elapsed_hours": 10.651074713601007, "session_id": 2, "event": "train_step", "step": 87560, "epoch": 1, "loss": 1.768518133163452, "loss_raw": 1.2078866958618164, "perplexity": 5.862159981290797, "muon_lr": 0.004406913499048426, "adamw_lr": 0.00013220740497145276, "tokens_per_sec": 149698.3656336748, "vram_gb": 27.299458026885986, "grad_norm": 0.1318359375, "tokens_processed": 5738332160, "tokens_processed_billions": 5.73833216}
+{"timestamp": 1787564971.855847, "elapsed_hours": 10.652213149865469, "session_id": 2, "event": "train_step", "step": 87570, "epoch": 1, "loss": 1.7325332129001618, "loss_raw": 1.7205485105514526, "perplexity": 5.654960999477388, "muon_lr": 0.004406172231086835, "adamw_lr": 0.00013218516693260505, "tokens_per_sec": 149699.4570343949, "vram_gb": 27.710740089416504, "grad_norm": 0.138671875, "tokens_processed": 5738987520, "tokens_processed_billions": 5.73898752}
+{"timestamp": 1787564975.8563972, "elapsed_hours": 10.653324413961835, "session_id": 2, "event": "train_step", "step": 87580, "epoch": 1, "loss": 1.7301462030410766, "loss_raw": 2.5616397857666016, "perplexity": 5.641478649471749, "muon_lr": 0.0044054309946684085, "adamw_lr": 0.00013216292984005222, "tokens_per_sec": 149700.93055693613, "vram_gb": 27.710740089416504, "grad_norm": 0.134765625, "tokens_processed": 5739642880, "tokens_processed_billions": 5.73964288}