Upload sft_metrics.jsonl with huggingface_hub

sft_metrics.jsonl

@@ -348,3 +348,14 @@
 {"timestamp": 1787700234.6585705, "elapsed_hours": 0.7556638451417287, "event": "train_step", "step": 11280, "epoch": 1, "loss": 1.4568647718429566, "perplexity": 4.292480503738766, "muon_lr": 0.0009264746265326828, "adamw_lr": 2.316186566331707e-05, "tokens_per_sec": 550271.1877844172, "vram_gb": 2.363706111907959, "tokens_processed": 1478492160}
 {"timestamp": 1787700242.0943127, "elapsed_hours": 0.7577293290032281, "event": "train_step", "step": 11290, "epoch": 1, "loss": 1.4326110148429871, "perplexity": 4.189624093871692, "muon_lr": 0.0009253594357812361, "adamw_lr": 2.3133985894530904e-05, "tokens_per_sec": 549239.0386710799, "vram_gb": 2.785654067993164, "tokens_processed": 1479802880}
 {"timestamp": 1787700249.5313933, "elapsed_hours": 0.7597951847314834, "event": "train_step", "step": 11300, "epoch": 1, "loss": 1.463402227163315, "perplexity": 4.320634330166113, "muon_lr": 0.0009242445235897127, "adamw_lr": 2.3106113089742817e-05, "tokens_per_sec": 548212.2646218977, "vram_gb": 2.785654067993164, "tokens_processed": 1481113600}
+{"timestamp": 1787700257.9054568, "elapsed_hours": 0.7621213136116664, "event": "train_step", "step": 11310, "epoch": 1, "loss": 1.4670104277133942, "perplexity": 4.336252204600238, "muon_lr": 0.0009231298917364507, "adamw_lr": 2.3078247293411268e-05, "tokens_per_sec": 547001.9607613873, "vram_gb": 2.785654067993164, "tokens_processed": 1482424320}
+{"timestamp": 1787700265.4308887, "elapsed_hours": 0.7642117112212711, "event": "train_step", "step": 11320, "epoch": 1, "loss": 1.4780803406238556, "perplexity": 4.384520810581358, "muon_lr": 0.0009209014761558205, "adamw_lr": 2.3022536903895513e-05, "tokens_per_sec": 545969.5365662657, "vram_gb": 2.363706111907959, "tokens_processed": 1483735040}
+{"timestamp": 1787700272.8528636, "elapsed_hours": 0.7662733709812164, "event": "train_step", "step": 11330, "epoch": 1, "loss": 1.4867072486877442, "perplexity": 4.4225092944947155, "muon_lr": 0.0009197876959828801, "adamw_lr": 2.2994692399572e-05, "tokens_per_sec": 544963.5192240513, "vram_gb": 2.785654067993164, "tokens_processed": 1485045760}
+{"timestamp": 1787700280.2772567, "elapsed_hours": 0.7683357023530536, "event": "train_step", "step": 11340, "epoch": 1, "loss": 1.4659690022468568, "perplexity": 4.331738671786982, "muon_lr": 0.0009186742032570509, "adamw_lr": 2.2966855081426276e-05, "tokens_per_sec": 543962.4752811191, "vram_gb": 2.785654067993164, "tokens_processed": 1486356480}
+{"timestamp": 1787700287.711959, "elapsed_hours": 0.7704008974631628, "event": "train_step", "step": 11350, "epoch": 1, "loss": 1.4565072345733643, "perplexity": 4.290946056307099, "muon_lr": 0.0009175609997544059, "adamw_lr": 2.293902499386015e-05, "tokens_per_sec": 542964.82476155, "vram_gb": 2.785654067993164, "tokens_processed": 1487667200}
+{"timestamp": 1787700295.220643, "elapsed_hours": 0.7724866430626975, "event": "train_step", "step": 11360, "epoch": 1, "loss": 1.4760524177551269, "perplexity": 4.375638350077628, "muon_lr": 0.0009153354675206524, "adamw_lr": 2.288338668801631e-05, "tokens_per_sec": 541957.960357814, "vram_gb": 2.363706111907959, "tokens_processed": 1488977920}
+{"timestamp": 1787700302.6516562, "elapsed_hours": 0.7745508133040534, "event": "train_step", "step": 11370, "epoch": 1, "loss": 1.4697375428676605, "perplexity": 4.348093803062448, "muon_lr": 0.000914223142339373, "adamw_lr": 2.2855578558484325e-05, "tokens_per_sec": 540971.852330008, "vram_gb": 2.785654067993164, "tokens_processed": 1490288640}
+{"timestamp": 1787700310.081428, "elapsed_hours": 0.7766146388318803, "event": "train_step", "step": 11380, "epoch": 1, "loss": 1.4227594578266143, "perplexity": 4.148552415402376, "muon_lr": 0.0009131111134809291, "adamw_lr": 2.282777783702323e-05, "tokens_per_sec": 539991.2874316641, "vram_gb": 2.785654067993164, "tokens_processed": 1491599360}
+{"timestamp": 1787700317.5041513, "elapsed_hours": 0.7786765064795812, "event": "train_step", "step": 11390, "epoch": 1, "loss": 1.4455369102954865, "perplexity": 4.244130248422296, "muon_lr": 0.00091199938271906, "adamw_lr": 2.27999845679765e-05, "tokens_per_sec": 539017.362345298, "vram_gb": 2.785654067993164, "tokens_processed": 1492910080}
+{"timestamp": 1787700325.0195868, "elapsed_hours": 0.780764127307468, "event": "train_step", "step": 11400, "epoch": 1, "loss": 1.4539507269859313, "perplexity": 4.279990230449877, "muon_lr": 0.000909776822577618, "adamw_lr": 2.2744420564440455e-05, "tokens_per_sec": 538030.6660541185, "vram_gb": 2.363706111907959, "tokens_processed": 1494220800}
+{"timestamp": 1787700337.8438919, "elapsed_hours": 0.7843264345328013, "event": "validation", "step": 11400, "val_loss": 1.4708755040168762, "val_perplexity": 4.353044581245841, "best_loss": 1.4708755040168762}