Upload metrics_session_002.jsonl with huggingface_hub
Browse files
metrics_session_002.jsonl
CHANGED
|
@@ -3668,3 +3668,6 @@
|
|
| 3668 |
{"timestamp": 1787540796.983581, "elapsed_hours": 3.9369708540042243, "session_id": 2, "event": "train_step", "step": 32470, "epoch": 1, "loss": 2.2043755328655243, "loss_raw": 2.4435410499572754, "perplexity": 9.06458926199682, "muon_lr": 0.009091514289167608, "adamw_lr": 0.00027274542867502816, "tokens_per_sec": 150259.52856695262, "vram_gb": 27.710740089416504, "grad_norm": 0.1435546875, "tokens_processed": 2127953920, "tokens_processed_billions": 2.12795392}
|
| 3669 |
{"timestamp": 1787540801.1206102, "elapsed_hours": 3.9381200287077163, "session_id": 2, "event": "train_step", "step": 32480, "epoch": 1, "loss": 2.1551775419712067, "loss_raw": 2.1655261516571045, "perplexity": 8.629422127587047, "muon_lr": 0.009090593246462126, "adamw_lr": 0.0002727177973938638, "tokens_per_sec": 150261.90896963637, "vram_gb": 27.299458026885986, "grad_norm": 0.1494140625, "tokens_processed": 2128609280, "tokens_processed_billions": 2.12860928}
|
| 3670 |
{"timestamp": 1787540805.093338, "elapsed_hours": 3.9392235640684765, "session_id": 2, "event": "train_step", "step": 32490, "epoch": 1, "loss": 2.0868545782566073, "loss_raw": 1.8690316677093506, "perplexity": 8.059524649226665, "muon_lr": 0.00909013256978762, "adamw_lr": 0.0002727039770936285, "tokens_per_sec": 150266.03108220134, "vram_gb": 27.710740089416504, "grad_norm": 0.154296875, "tokens_processed": 2129264640, "tokens_processed_billions": 2.12926464}
|
|
|
|
|
|
|
|
|
|
|
|
| 3668 |
{"timestamp": 1787540796.983581, "elapsed_hours": 3.9369708540042243, "session_id": 2, "event": "train_step", "step": 32470, "epoch": 1, "loss": 2.2043755328655243, "loss_raw": 2.4435410499572754, "perplexity": 9.06458926199682, "muon_lr": 0.009091514289167608, "adamw_lr": 0.00027274542867502816, "tokens_per_sec": 150259.52856695262, "vram_gb": 27.710740089416504, "grad_norm": 0.1435546875, "tokens_processed": 2127953920, "tokens_processed_billions": 2.12795392}
|
| 3669 |
{"timestamp": 1787540801.1206102, "elapsed_hours": 3.9381200287077163, "session_id": 2, "event": "train_step", "step": 32480, "epoch": 1, "loss": 2.1551775419712067, "loss_raw": 2.1655261516571045, "perplexity": 8.629422127587047, "muon_lr": 0.009090593246462126, "adamw_lr": 0.0002727177973938638, "tokens_per_sec": 150261.90896963637, "vram_gb": 27.299458026885986, "grad_norm": 0.1494140625, "tokens_processed": 2128609280, "tokens_processed_billions": 2.12860928}
|
| 3670 |
{"timestamp": 1787540805.093338, "elapsed_hours": 3.9392235640684765, "session_id": 2, "event": "train_step", "step": 32490, "epoch": 1, "loss": 2.0868545782566073, "loss_raw": 1.8690316677093506, "perplexity": 8.059524649226665, "muon_lr": 0.00909013256978762, "adamw_lr": 0.0002727039770936285, "tokens_per_sec": 150266.03108220134, "vram_gb": 27.710740089416504, "grad_norm": 0.154296875, "tokens_processed": 2129264640, "tokens_processed_billions": 2.12926464}
|
| 3671 |
+
{"timestamp": 1787540809.7900612, "elapsed_hours": 3.940528209540579, "session_id": 2, "event": "train_step", "step": 32500, "epoch": 1, "loss": 1.9808004879951477, "loss_raw": 1.344995379447937, "perplexity": 7.2485430350980335, "muon_lr": 0.00908967178958298, "adamw_lr": 0.0002726901536874893, "tokens_per_sec": 150262.4767691014, "vram_gb": 27.710740089416504, "grad_norm": 0.150390625, "tokens_processed": 2129920000, "tokens_processed_billions": 2.12992}
|
| 3672 |
+
{"timestamp": 1787540809.7933142, "elapsed_hours": 3.9405291131469937, "session_id": 2, "event": "lambda_stats", "step": 32500, "epoch": 1, "lambda_min": 1.109375, "lambda_max": 2.015625, "lambda_mean": 1.1970486111111112, "lambda_std": 0.20254918979414713, "lambda_all": [2.015625, 1.265625, 1.2109375, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.125, 1.140625, 1.109375, 1.140625, 1.109375, 1.125, 1.1328125], "local_layers": 9, "global_layers": 9}
|
| 3673 |
+
{"timestamp": 1787540823.3487804, "elapsed_hours": 3.944294520417849, "session_id": 2, "event": "validation", "step": 32500, "epoch": 1, "val_loss": 2.0221471470594405, "val_perplexity": 7.554528215065266, "is_best": true, "best_loss": 2.0221471470594405, "best_perplexity": 7.554528215065266, "train_loss": 1.344995379447937, "train_loss_avg": 1.9808004879951477}
|