Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -450,3 +450,23 @@
 {"timestamp": 1787528348.2902775, "elapsed_hours": 0.4790004918310377, "session_id": 2, "event": "lambda_stats", "step": 4000, "epoch": 1, "lambda_min": 1.140625, "lambda_max": 1.625, "lambda_mean": 1.2443576388888888, "lambda_std": 0.11598940662977776, "lambda_all": [1.625, 1.390625, 1.3359375, 1.328125, 1.203125, 1.21875, 1.140625, 1.2578125, 1.203125, 1.171875, 1.171875, 1.140625, 1.140625, 1.203125, 1.1484375, 1.2421875, 1.203125, 1.2734375], "local_layers": 9, "global_layers": 9}
 {"timestamp": 1787528361.6996825, "elapsed_hours": 0.482725326816241, "session_id": 2, "event": "validation", "step": 4000, "epoch": 1, "val_loss": 3.0368866670131682, "val_perplexity": 20.840259462282727, "is_best": true, "best_loss": 3.0368866670131682, "best_perplexity": 20.840259462282727, "train_loss": 2.572169780731201, "train_loss_avg": 3.1422323060035704}
 {"timestamp": 1787528374.4575255, "elapsed_hours": 0.4862691720989015, "session_id": 2, "event": "checkpoint", "step": 4000, "epoch": 0, "checkpoint_loss": 2.572169780731201, "checkpoint_name": "codva2_s002_step0004000.pt", "best_loss": 3.0368866670131682, "val_loss": null, "is_best": false}
+{"timestamp": 1787528374.9975245, "elapsed_hours": 0.4864191714260313, "session_id": 2, "event": "checkpoint_save", "step": 4000, "epoch": 1, "checkpoint_type": "scheduled", "time_since_last_ckpt": 6.67572021484375e-06, "current_loss": 2.572169780731201, "best_loss": 3.0368866670131682, "vram_gb": 27.299458026885986}
+{"timestamp": 1787528378.99228, "elapsed_hours": 0.4875288258658515, "session_id": 2, "event": "train_step", "step": 4010, "epoch": 1, "loss": 3.2105546307563784, "loss_raw": 2.250105857849121, "perplexity": 24.79283327984023, "muon_lr": 0.009995916919987476, "adamw_lr": 0.0002998775075996242, "tokens_per_sec": 150699.33339216956, "vram_gb": 27.710740089416504, "grad_norm": 0.248046875, "tokens_processed": 262799360, "tokens_processed_billions": 0.26279936}
+{"timestamp": 1787528382.983231, "elapsed_hours": 0.4886374231841829, "session_id": 2, "event": "train_step", "step": 4020, "epoch": 1, "loss": 3.056169054508209, "loss_raw": 2.6334786415100098, "perplexity": 21.24600875654894, "muon_lr": 0.0099958843093731, "adamw_lr": 0.00029987652928119296, "tokens_per_sec": 150730.18450608876, "vram_gb": 27.710740089416504, "grad_norm": 0.244140625, "tokens_processed": 263454720, "tokens_processed_billions": 0.26345472}
+{"timestamp": 1787528386.9910784, "elapsed_hours": 0.4897507142358356, "session_id": 2, "event": "train_step", "step": 4030, "epoch": 1, "loss": 3.0227677273750304, "loss_raw": 3.996128559112549, "perplexity": 20.548084550468097, "muon_lr": 0.009995851569109059, "adamw_lr": 0.0002998755470732717, "tokens_per_sec": 150759.44016460725, "vram_gb": 27.710740089416504, "grad_norm": 0.208984375, "tokens_processed": 264110080, "tokens_processed_billions": 0.26411008}
+{"timestamp": 1787528391.0833273, "elapsed_hours": 0.4908874500460095, "session_id": 2, "event": "train_step", "step": 4040, "epoch": 1, "loss": 3.004849519729614, "loss_raw": 3.0409250259399414, "perplexity": 20.183178697271245, "muon_lr": 0.009995785699635742, "adamw_lr": 0.00029987357098907225, "tokens_per_sec": 150781.31044886977, "vram_gb": 27.299458026885986, "grad_norm": 0.298828125, "tokens_processed": 264765440, "tokens_processed_billions": 0.26476544}
+{"timestamp": 1787528395.0574434, "elapsed_hours": 0.4919913711150487, "session_id": 2, "event": "train_step", "step": 4050, "epoch": 1, "loss": 2.9344901633262634, "loss_raw": 2.719149589538574, "perplexity": 18.811909688384787, "muon_lr": 0.009995752570428368, "adamw_lr": 0.000299872577112851, "tokens_per_sec": 150813.21650829434, "vram_gb": 27.710740089416504, "grad_norm": 0.2392578125, "tokens_processed": 265420800, "tokens_processed_billions": 0.2654208}
+{"timestamp": 1787528399.0224822, "elapsed_hours": 0.4930927708413866, "session_id": 2, "event": "train_step", "step": 4060, "epoch": 1, "loss": 2.895320792198181, "loss_raw": 3.5833942890167236, "perplexity": 18.089303418000583, "muon_lr": 0.009995719311575127, "adamw_lr": 0.0002998715793472537, "tokens_per_sec": 150845.7505082489, "vram_gb": 27.710740089416504, "grad_norm": 0.259765625, "tokens_processed": 266076160, "tokens_processed_billions": 0.26607616}
+{"timestamp": 1787528402.9870927, "elapsed_hours": 0.4941940514246623, "session_id": 2, "event": "train_step", "step": 4070, "epoch": 1, "loss": 3.0154302453994752, "loss_raw": 2.978327751159668, "perplexity": 20.3978651405201, "muon_lr": 0.009995685923076973, "adamw_lr": 0.00029987057769230907, "tokens_per_sec": 150878.1684023249, "vram_gb": 27.710740089416504, "grad_norm": 0.341796875, "tokens_processed": 266731520, "tokens_processed_billions": 0.26673152}
+{"timestamp": 1787528407.1128125, "elapsed_hours": 0.4953400848309199, "session_id": 2, "event": "train_step", "step": 4080, "epoch": 1, "loss": 2.9700451159477232, "loss_raw": 2.8032288551330566, "perplexity": 19.492799012294338, "muon_lr": 0.009995618757149782, "adamw_lr": 0.0002998685627144934, "tokens_per_sec": 150896.72050302447, "vram_gb": 27.299458026885986, "grad_norm": 0.259765625, "tokens_processed": 267386880, "tokens_processed_billions": 0.26738688}
+{"timestamp": 1787528411.0779982, "elapsed_hours": 0.496441525220871, "session_id": 2, "event": "train_step", "step": 4090, "epoch": 1, "loss": 3.0235792326927187, "loss_raw": 3.4909000396728516, "perplexity": 20.56476619805651, "muon_lr": 0.009995584979722683, "adamw_lr": 0.00029986754939168047, "tokens_per_sec": 150928.82982161554, "vram_gb": 27.710740089416504, "grad_norm": 0.265625, "tokens_processed": 268042240, "tokens_processed_billions": 0.26804224}
+{"timestamp": 1787528415.058794, "elapsed_hours": 0.4975473018487295, "session_id": 2, "event": "train_step", "step": 4100, "epoch": 1, "loss": 3.0358504319190978, "loss_raw": 2.95054030418396, "perplexity": 20.818675239154516, "muon_lr": 0.009995551072654547, "adamw_lr": 0.0002998665321796364, "tokens_per_sec": 150959.47294200488, "vram_gb": 27.710740089416504, "grad_norm": 0.2490234375, "tokens_processed": 268697600, "tokens_processed_billions": 0.2686976}
+{"timestamp": 1787528415.062435, "elapsed_hours": 0.49754831340577865, "session_id": 2, "event": "lambda_stats", "step": 4100, "epoch": 1, "lambda_min": 1.140625, "lambda_max": 1.640625, "lambda_mean": 1.2417534722222223, "lambda_std": 0.11960674762104911, "lambda_all": [1.640625, 1.390625, 1.328125, 1.328125, 1.203125, 1.21875, 1.140625, 1.2578125, 1.203125, 1.1640625, 1.171875, 1.140625, 1.140625, 1.1953125, 1.140625, 1.234375, 1.1875, 1.265625], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787528419.025699, "elapsed_hours": 0.49864921993679473, "session_id": 2, "event": "train_step", "step": 4110, "epoch": 1, "loss": 3.038788936138153, "loss_raw": 3.064258575439453, "perplexity": 20.8799409749066, "muon_lr": 0.009995517035946352, "adamw_lr": 0.0002998655110783905, "tokens_per_sec": 150991.15060134657, "vram_gb": 27.710740089416504, "grad_norm": 0.27734375, "tokens_processed": 269352960, "tokens_processed_billions": 0.26935296}
+{"timestamp": 1787528423.141044, "elapsed_hours": 0.4997923713922501, "session_id": 2, "event": "train_step", "step": 4120, "epoch": 1, "loss": 3.0423877573013307, "loss_raw": 1.7149800062179565, "perplexity": 20.955219524143477, "muon_lr": 0.00999544857361371, "adamw_lr": 0.00029986345720841125, "tokens_per_sec": 151010.15450450592, "vram_gb": 27.299458026885986, "grad_norm": 0.25, "tokens_processed": 270008320, "tokens_processed_billions": 0.27000832}
+{"timestamp": 1787528427.1074133, "elapsed_hours": 0.5008941405349308, "session_id": 2, "event": "train_step", "step": 4130, "epoch": 1, "loss": 3.035190427303314, "loss_raw": 3.3369314670562744, "perplexity": 20.804939350775662, "muon_lr": 0.009995414147991238, "adamw_lr": 0.0002998624244397371, "tokens_per_sec": 151041.62839839954, "vram_gb": 27.710740089416504, "grad_norm": 0.2451171875, "tokens_processed": 270663680, "tokens_processed_billions": 0.27066368}
+{"timestamp": 1787528431.0564086, "elapsed_hours": 0.5019910835557514, "session_id": 2, "event": "train_step", "step": 4140, "epoch": 1, "loss": 2.905520853996277, "loss_raw": 2.2270307540893555, "perplexity": 18.27475965536741, "muon_lr": 0.009995379592732657, "adamw_lr": 0.0002998613877819796, "tokens_per_sec": 151074.42850035, "vram_gb": 27.710740089416504, "grad_norm": 0.255859375, "tokens_processed": 271319040, "tokens_processed_billions": 0.27131904}
+{"timestamp": 1787528435.009925, "elapsed_hours": 0.5030892825788922, "session_id": 2, "event": "train_step", "step": 4150, "epoch": 1, "loss": 3.0193099188804626, "loss_raw": 3.2770252227783203, "perplexity": 20.477155908663615, "muon_lr": 0.009995344907838957, "adamw_lr": 0.00029986034723516867, "tokens_per_sec": 151106.69181844202, "vram_gb": 27.710740089416504, "grad_norm": 0.24609375, "tokens_processed": 271974400, "tokens_processed_billions": 0.2719744}
+{"timestamp": 1787528439.1294062, "elapsed_hours": 0.504233583079444, "session_id": 2, "event": "train_step", "step": 4160, "epoch": 1, "loss": 2.9922337245941164, "loss_raw": 3.2873690128326416, "perplexity": 19.930151276636526, "muon_lr": 0.009995275149150221, "adamw_lr": 0.0002998582544745066, "tokens_per_sec": 151124.91968945725, "vram_gb": 27.299458026885986, "grad_norm": 0.26953125, "tokens_processed": 272629760, "tokens_processed_billions": 0.27262976}
+{"timestamp": 1787528443.1199484, "elapsed_hours": 0.5053420670164956, "session_id": 2, "event": "train_step", "step": 4170, "epoch": 1, "loss": 2.8553618240356444, "loss_raw": 1.2272441387176514, "perplexity": 17.380724834082372, "muon_lr": 0.009995240075357195, "adamw_lr": 0.00029985720226071577, "tokens_per_sec": 151153.84775325947, "vram_gb": 27.710740089416504, "grad_norm": 0.265625, "tokens_processed": 273285120, "tokens_processed_billions": 0.27328512}
+{"timestamp": 1787528447.0973125, "elapsed_hours": 0.5064468903011746, "session_id": 2, "event": "train_step", "step": 4180, "epoch": 1, "loss": 2.888508038520813, "loss_raw": 1.8815553188323975, "perplexity": 17.966484292881297, "muon_lr": 0.009995204871933075, "adamw_lr": 0.0002998561461579922, "tokens_per_sec": 151183.7385355013, "vram_gb": 27.710740089416504, "grad_norm": 0.263671875, "tokens_processed": 273940480, "tokens_processed_billions": 0.27394048}