Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -3093,3 +3093,15 @@
 {"timestamp": 1787649410.4710007, "elapsed_hours": 6.153767804967033, "session_id": 2, "event": "train_step", "step": 121390, "epoch": 1, "loss": 1.6691827285289764, "loss_raw": 2.1320412158966064, "perplexity": 5.307828087593213, "muon_lr": 0.0018050284597364934, "adamw_lr": 5.41508537920948e-05, "tokens_per_sec": 720771.9910997802, "vram_gb": 2.785593032836914, "grad_norm": 0.08984375, "tokens_processed": 15910830080, "tokens_processed_billions": 15.91083008}
 {"timestamp": 1787649418.0009794, "elapsed_hours": 6.155859465797742, "session_id": 2, "event": "train_step", "step": 121400, "epoch": 1, "loss": 1.6677476871013641, "loss_raw": 1.7530258893966675, "perplexity": 5.300216597105184, "muon_lr": 0.0018041563294557126, "adamw_lr": 5.412468988367137e-05, "tokens_per_sec": 720585.5652693575, "vram_gb": 2.363645076751709, "grad_norm": 0.09375, "tokens_processed": 15912140800, "tokens_processed_billions": 15.9121408}
 {"timestamp": 1787649418.0028183, "elapsed_hours": 6.155859976609548, "session_id": 2, "event": "lambda_stats", "step": 121400, "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": 1787649426.00738, "elapsed_hours": 6.15808346596029, "session_id": 2, "event": "train_step", "step": 121410, "epoch": 1, "loss": 1.6486177933216095, "loss_raw": 1.580083966255188, "perplexity": 5.199787676552838, "muon_lr": 0.0018037204241768867, "adamw_lr": 5.411161272530659e-05, "tokens_per_sec": 720383.732356658, "vram_gb": 2.785593032836914, "grad_norm": 0.09765625, "tokens_processed": 15913451520, "tokens_processed_billions": 15.91345152}
+{"timestamp": 1787649433.468517, "elapsed_hours": 6.160156004097726, "session_id": 2, "event": "train_step", "step": 121420, "epoch": 1, "loss": 1.6707930648326874, "loss_raw": 1.7538492679595947, "perplexity": 5.316382361636977, "muon_lr": 0.0018032846254892004, "adamw_lr": 5.4098538764676e-05, "tokens_per_sec": 720199.8213407228, "vram_gb": 2.785593032836914, "grad_norm": 0.09619140625, "tokens_processed": 15914762240, "tokens_processed_billions": 15.91476224}
+{"timestamp": 1787649440.947087, "elapsed_hours": 6.1622333845827315, "session_id": 2, "event": "train_step", "step": 121430, "epoch": 1, "loss": 1.707235722541809, "loss_raw": 1.7171483039855957, "perplexity": 5.513698998626202, "muon_lr": 0.0018028489334052196, "adamw_lr": 5.408546800215658e-05, "tokens_per_sec": 720015.4530438927, "vram_gb": 2.785593032836914, "grad_norm": 0.09912109375, "tokens_processed": 15916072960, "tokens_processed_billions": 15.91607296}
+{"timestamp": 1787649448.5268137, "elapsed_hours": 6.164338864154286, "session_id": 2, "event": "train_step", "step": 121440, "epoch": 1, "loss": 1.7179648935794831, "loss_raw": 1.9451918601989746, "perplexity": 5.573174911305664, "muon_lr": 0.0018019778690986274, "adamw_lr": 5.4059336072958815e-05, "tokens_per_sec": 719827.9216128689, "vram_gb": 2.363645076751709, "grad_norm": 0.0986328125, "tokens_processed": 15917383680, "tokens_processed_billions": 15.91738368}
+{"timestamp": 1787649456.0180275, "elapsed_hours": 6.166419756942325, "session_id": 2, "event": "train_step", "step": 121450, "epoch": 1, "loss": 1.7050596845149995, "loss_raw": 0.9344683885574341, "perplexity": 5.501714024545696, "muon_lr": 0.0018015424969011363, "adamw_lr": 5.404627490703408e-05, "tokens_per_sec": 719643.4022952226, "vram_gb": 2.785593032836914, "grad_norm": 0.0927734375, "tokens_processed": 15918694400, "tokens_processed_billions": 15.9186944}
+{"timestamp": 1787649463.4685419, "elapsed_hours": 6.168489344384935, "session_id": 2, "event": "train_step", "step": 121460, "epoch": 1, "loss": 1.6509090840816498, "loss_raw": 2.031215190887451, "perplexity": 5.211715561918595, "muon_lr": 0.0018011072313575893, "adamw_lr": 5.403321694072767e-05, "tokens_per_sec": 719460.3282207763, "vram_gb": 2.785593032836914, "grad_norm": 0.08740234375, "tokens_processed": 15920005120, "tokens_processed_billions": 15.92000512}
+{"timestamp": 1787649470.9302878, "elapsed_hours": 6.170562051468425, "session_id": 2, "event": "train_step", "step": 121470, "epoch": 1, "loss": 1.623445256948471, "loss_raw": 1.9617935419082642, "perplexity": 5.070529535132742, "muon_lr": 0.0018006720724805373, "adamw_lr": 5.402016217441611e-05, "tokens_per_sec": 719277.0126589967, "vram_gb": 2.785593032836914, "grad_norm": 0.0966796875, "tokens_processed": 15921315840, "tokens_processed_billions": 15.92131584}
+{"timestamp": 1787649478.462039, "elapsed_hours": 6.172654204501046, "session_id": 2, "event": "train_step", "step": 121480, "epoch": 1, "loss": 1.582441588640213, "loss_raw": 1.9092624187469482, "perplexity": 4.8668240990372995, "muon_lr": 0.001799802074776116, "adamw_lr": 5.399406224328347e-05, "tokens_per_sec": 719091.546090735, "vram_gb": 2.363645076751709, "grad_norm": 0.09033203125, "tokens_processed": 15922626560, "tokens_processed_billions": 15.92262656}
+{"timestamp": 1787649485.9218884, "elapsed_hours": 6.174726384878158, "session_id": 2, "event": "train_step", "step": 121490, "epoch": 1, "loss": 1.5613203573226928, "loss_raw": 2.113929033279419, "perplexity": 4.7651087395724785, "muon_lr": 0.0017993672359738326, "adamw_lr": 5.398101707921497e-05, "tokens_per_sec": 718908.542278036, "vram_gb": 2.785593032836914, "grad_norm": 0.09716796875, "tokens_processed": 15923937280, "tokens_processed_billions": 15.92393728}
+{"timestamp": 1787649493.3618815, "elapsed_hours": 6.176793049706353, "session_id": 2, "event": "train_step", "step": 121500, "epoch": 1, "loss": 1.581110360622406, "loss_raw": 2.0223326683044434, "perplexity": 4.860349556941008, "muon_lr": 0.0017989325038882242, "adamw_lr": 5.396797511664671e-05, "tokens_per_sec": 718726.3045816424, "vram_gb": 2.785593032836914, "grad_norm": 0.09814453125, "tokens_processed": 15925248000, "tokens_processed_billions": 15.925248}
+{"timestamp": 1787649493.3636644, "elapsed_hours": 6.176793544954724, "session_id": 2, "event": "lambda_stats", "step": 121500, "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": 1787649519.1006048, "elapsed_hours": 6.183942695061366, "session_id": 2, "event": "validation", "step": 121500, "epoch": 1, "val_loss": 1.7510874903202056, "val_perplexity": 5.760864154744188, "is_best": false, "best_loss": 1.6233880519866943, "best_perplexity": 5.070239483980759, "train_loss": 2.0223326683044434, "train_loss_avg": 1.581110360622406}