Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -2993,3 +2993,16 @@
 {"timestamp": 1787648690.6277814, "elapsed_hours": 5.953811355166965, "session_id": 2, "event": "train_step", "step": 120500, "epoch": 1, "loss": 1.6548359680175782, "loss_raw": 1.216089129447937, "perplexity": 5.232221600102322, "muon_lr": 0.0018540970546030687, "adamw_lr": 5.5622911638092046e-05, "tokens_per_sec": 739605.5815642223, "vram_gb": 2.785593032836914, "grad_norm": 0.10009765625, "tokens_processed": 15794176000, "tokens_processed_billions": 15.794176}
 {"timestamp": 1787648690.6297991, "elapsed_hours": 5.953811915781763, "session_id": 2, "event": "lambda_stats", "step": 120500, "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": 1787648716.356595, "elapsed_hours": 5.960958247913255, "session_id": 2, "event": "validation", "step": 120500, "epoch": 1, "val_loss": 1.7515667796134948, "val_perplexity": 5.7636259370471254, "is_best": false, "best_loss": 1.6233880519866943, "best_perplexity": 5.070239483980759, "train_loss": 1.216089129447937, "train_loss_avg": 1.6548359680175782}
+{"timestamp": 1787648724.4684877, "elapsed_hours": 5.963211551374859, "session_id": 2, "event": "train_step", "step": 120510, "epoch": 1, "loss": 1.668072419166565, "loss_raw": 2.206441879272461, "perplexity": 5.301938026873368, "muon_lr": 0.0018536491862280577, "adamw_lr": 5.560947558684172e-05, "tokens_per_sec": 738496.6767615768, "vram_gb": 2.785593032836914, "grad_norm": 0.091796875, "tokens_processed": 15795486720, "tokens_processed_billions": 15.79548672}
+{"timestamp": 1787648731.9968574, "elapsed_hours": 5.965302765170733, "session_id": 2, "event": "train_step", "step": 120520, "epoch": 1, "loss": 1.667767573595047, "loss_raw": 0.9142382144927979, "perplexity": 5.300322000877112, "muon_lr": 0.0018527537649449125, "adamw_lr": 5.558261294834736e-05, "tokens_per_sec": 738298.090692642, "vram_gb": 2.363645076751709, "grad_norm": 0.09228515625, "tokens_processed": 15796797440, "tokens_processed_billions": 15.79679744}
+{"timestamp": 1787648739.466436, "elapsed_hours": 5.967377648154894, "session_id": 2, "event": "train_step", "step": 120530, "epoch": 1, "loss": 1.653720281124115, "loss_raw": 1.7640538215637207, "perplexity": 5.226387334251551, "muon_lr": 0.0018523062120626016, "adamw_lr": 5.556918636187803e-05, "tokens_per_sec": 738101.6790008968, "vram_gb": 2.785593032836914, "grad_norm": 0.09375, "tokens_processed": 15798108160, "tokens_processed_billions": 15.79810816}
+{"timestamp": 1787648746.9102967, "elapsed_hours": 5.96944538725747, "session_id": 2, "event": "train_step", "step": 120540, "epoch": 1, "loss": 1.64486971616745, "loss_raw": 1.458123803138733, "perplexity": 5.180334949090351, "muon_lr": 0.0018518587643703388, "adamw_lr": 5.555576293111015e-05, "tokens_per_sec": 737906.280608494, "vram_gb": 2.785593032836914, "grad_norm": 0.09765625, "tokens_processed": 15799418880, "tokens_processed_billions": 15.79941888}
+{"timestamp": 1787648754.3703208, "elapsed_hours": 5.971517616113027, "session_id": 2, "event": "train_step", "step": 120550, "epoch": 1, "loss": 1.6660165357589722, "loss_raw": 1.8722820281982422, "perplexity": 5.291049057516875, "muon_lr": 0.0018514114218810308, "adamw_lr": 5.554234265643092e-05, "tokens_per_sec": 737710.4663967848, "vram_gb": 2.785593032836914, "grad_norm": 0.09619140625, "tokens_processed": 15800729600, "tokens_processed_billions": 15.8007296}
+{"timestamp": 1787648761.9445412, "elapsed_hours": 5.973621566229396, "session_id": 2, "event": "train_step", "step": 120560, "epoch": 1, "loss": 1.6785102105140686, "loss_raw": 1.8284169435501099, "perplexity": 5.357568373646743, "muon_lr": 0.0018505170525628702, "adamw_lr": 5.55155115768861e-05, "tokens_per_sec": 737510.8556821002, "vram_gb": 2.363645076751709, "grad_norm": 0.0947265625, "tokens_processed": 15802040320, "tokens_processed_billions": 15.80204032}
+{"timestamp": 1787648769.3942509, "elapsed_hours": 5.975690930022134, "session_id": 2, "event": "train_step", "step": 120570, "epoch": 1, "loss": 1.6930589640140534, "loss_raw": 1.2607165575027466, "perplexity": 5.436084083204235, "muon_lr": 0.0018500700257598108, "adamw_lr": 5.5502100772794316e-05, "tokens_per_sec": 737315.6729064154, "vram_gb": 2.785593032836914, "grad_norm": 0.0927734375, "tokens_processed": 15803351040, "tokens_processed_billions": 15.80335104}
+{"timestamp": 1787648776.879098, "elapsed_hours": 5.977770054207908, "session_id": 2, "event": "train_step", "step": 120580, "epoch": 1, "loss": 1.721484537124634, "loss_raw": 1.877375602722168, "perplexity": 5.592825060883978, "muon_lr": 0.0018496231042112858, "adamw_lr": 5.548869312633856e-05, "tokens_per_sec": 737119.4155983964, "vram_gb": 2.785593032836914, "grad_norm": 0.0947265625, "tokens_processed": 15804661760, "tokens_processed_billions": 15.80466176}
+{"timestamp": 1787648784.3730266, "elapsed_hours": 5.979851701127158, "session_id": 2, "event": "train_step", "step": 120590, "epoch": 1, "loss": 1.708468816280365, "loss_raw": 1.2126914262771606, "perplexity": 5.520502099906464, "muon_lr": 0.0018491762879301852, "adamw_lr": 5.547528863790555e-05, "tokens_per_sec": 736922.9870270881, "vram_gb": 2.785593032836914, "grad_norm": 0.095703125, "tokens_processed": 15805972480, "tokens_processed_billions": 15.80597248}
+{"timestamp": 1787648791.9285293, "elapsed_hours": 5.981950451864137, "session_id": 2, "event": "train_step", "step": 120600, "epoch": 1, "loss": 1.6746001851558685, "loss_raw": 1.8049649000167847, "perplexity": 5.336661046174237, "muon_lr": 0.001848282971221792, "adamw_lr": 5.544848913665375e-05, "tokens_per_sec": 736724.5713537184, "vram_gb": 2.363645076751709, "grad_norm": 0.09716796875, "tokens_processed": 15807283200, "tokens_processed_billions": 15.8072832}
+{"timestamp": 1787648791.9301646, "elapsed_hours": 5.981950906183984, "session_id": 2, "event": "lambda_stats", "step": 120600, "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": 1787648799.3946898, "elapsed_hours": 5.984024385346307, "session_id": 2, "event": "train_step", "step": 120610, "epoch": 1, "loss": 1.6311170077323913, "loss_raw": 1.0416420698165894, "perplexity": 5.109578971316616, "muon_lr": 0.001847836470820261, "adamw_lr": 5.543509412460782e-05, "tokens_per_sec": 736529.3690508152, "vram_gb": 2.785593032836914, "grad_norm": 0.08837890625, "tokens_processed": 15808593920, "tokens_processed_billions": 15.80859392}
+{"timestamp": 1787648806.8435323, "elapsed_hours": 5.986093508270052, "session_id": 2, "event": "train_step", "step": 120620, "epoch": 1, "loss": 1.6675921821594237, "loss_raw": 1.3062455654144287, "perplexity": 5.299392451312012, "muon_lr": 0.0018473900757376746, "adamw_lr": 5.542170227213023e-05, "tokens_per_sec": 736334.894726407, "vram_gb": 2.785593032836914, "grad_norm": 0.1025390625, "tokens_processed": 15809904640, "tokens_processed_billions": 15.80990464}