Upload sft_metrics.jsonl with huggingface_hub

sft_metrics.jsonl

@@ -193,3 +193,20 @@
 {"timestamp": 1787699035.1204128, "elapsed_hours": 0.422458801401986, "event": "train_step", "step": 9820, "epoch": 1, "loss": 1.4632143592834472, "perplexity": 4.319822697997016, "muon_lr": 0.0011333816300503835, "adamw_lr": 2.833454075125959e-05, "tokens_per_sec": 865409.2224103145, "vram_gb": 2.785654067993164, "tokens_processed": 1287127040}
 {"timestamp": 1787699042.558752, "elapsed_hours": 0.4245250067445967, "event": "train_step", "step": 9830, "epoch": 1, "loss": 1.4091460883617402, "perplexity": 4.092459312970722, "muon_lr": 0.0011322457292209684, "adamw_lr": 2.8306143230524212e-05, "tokens_per_sec": 861979.5219407637, "vram_gb": 2.785654067993164, "tokens_processed": 1288437760}
 {"timestamp": 1787699050.074713, "elapsed_hours": 0.4266127736700906, "event": "train_step", "step": 9840, "epoch": 1, "loss": 1.370722531080246, "perplexity": 3.9381951361342655, "muon_lr": 0.0011299737750736659, "adamw_lr": 2.8249344376841646e-05, "tokens_per_sec": 858539.4086041743, "vram_gb": 2.363706111907959, "tokens_processed": 1289748480}
+{"timestamp": 1787699058.1711488, "elapsed_hours": 0.42886178347799514, "event": "train_step", "step": 9850, "epoch": 1, "loss": 1.378437248468399, "perplexity": 3.9686946950981827, "muon_lr": 0.0011288377253796521, "adamw_lr": 2.8220943134491306e-05, "tokens_per_sec": 854804.9435874373, "vram_gb": 2.785654067993164, "tokens_processed": 1291059200}
+{"timestamp": 1787699065.599445, "elapsed_hours": 0.43092519919077554, "event": "train_step", "step": 9860, "epoch": 1, "loss": 1.4037262141704558, "perplexity": 4.070338697986059, "muon_lr": 0.0011277016296881088, "adamw_lr": 2.8192540742202717e-05, "tokens_per_sec": 851484.9701715624, "vram_gb": 2.785654067993164, "tokens_processed": 1292369920}
+{"timestamp": 1787699073.0629811, "elapsed_hours": 0.43299840364191267, "event": "train_step", "step": 9870, "epoch": 1, "loss": 1.4012152206897737, "perplexity": 4.060130925240064, "muon_lr": 0.0011265654898111614, "adamw_lr": 2.8164137245279035e-05, "tokens_per_sec": 848177.7271822846, "vram_gb": 2.785654067993164, "tokens_processed": 1293680640}
+{"timestamp": 1787699080.586811, "elapsed_hours": 0.4350883563359578, "event": "train_step", "step": 9880, "epoch": 1, "loss": 1.4178740799427032, "perplexity": 4.128334595239071, "muon_lr": 0.0011242930847499095, "adamw_lr": 2.810732711874774e-05, "tokens_per_sec": 844869.4263163911, "vram_gb": 2.363706111907959, "tokens_processed": 1294991360}
+{"timestamp": 1787699088.018597, "elapsed_hours": 0.4371527413527171, "event": "train_step", "step": 9890, "epoch": 1, "loss": 1.4377399718761443, "perplexity": 4.211167696713391, "muon_lr": 0.0011231568231901978, "adamw_lr": 2.8078920579754948e-05, "tokens_per_sec": 841643.8204527886, "vram_gb": 2.785654067993164, "tokens_processed": 1296302080}
+{"timestamp": 1787699095.4502351, "elapsed_hours": 0.4392170853084988, "event": "train_step", "step": 9900, "epoch": 1, "loss": 1.4387905395030975, "perplexity": 4.215594137898015, "muon_lr": 0.001122020524694263, "adamw_lr": 2.805051311735657e-05, "tokens_per_sec": 838449.198131598, "vram_gb": 2.785654067993164, "tokens_processed": 1297612800}
+{"timestamp": 1787699102.8823388, "elapsed_hours": 0.4412815585401323, "event": "train_step", "step": 9910, "epoch": 1, "loss": 1.4488852536678314, "perplexity": 4.258364871715076, "muon_lr": 0.0011208841910745541, "adamw_lr": 2.8022104776863855e-05, "tokens_per_sec": 835284.8795742714, "vram_gb": 2.785654067993164, "tokens_processed": 1298923520}
+{"timestamp": 1787699110.4056056, "elapsed_hours": 0.44337135480509865, "event": "train_step", "step": 9920, "epoch": 1, "loss": 1.4784575021266937, "perplexity": 4.386174794929506, "muon_lr": 0.0011186114257138906, "adamw_lr": 2.7965285642847267e-05, "tokens_per_sec": 832102.0903302351, "vram_gb": 2.363706111907959, "tokens_processed": 1300234240}
+{"timestamp": 1787699117.8304036, "elapsed_hours": 0.4454337987634871, "event": "train_step", "step": 9930, "epoch": 1, "loss": 1.5080088818073272, "perplexity": 4.5177265055527736, "muon_lr": 0.0011174749975981038, "adamw_lr": 2.7936874939952593e-05, "tokens_per_sec": 829001.8513034013, "vram_gb": 2.785654067993164, "tokens_processed": 1301544960}
+{"timestamp": 1787699125.2623854, "elapsed_hours": 0.44749823815292783, "event": "train_step", "step": 9940, "epoch": 1, "loss": 1.500021730661392, "perplexity": 4.4817864614639, "muon_lr": 0.001116338541608873, "adamw_lr": 2.7908463540221823e-05, "tokens_per_sec": 825926.9892187574, "vram_gb": 2.785654067993164, "tokens_processed": 1302855680}
+{"timestamp": 1787699132.6953602, "elapsed_hours": 0.4495629533794191, "event": "train_step", "step": 9950, "epoch": 1, "loss": 1.5152058231830596, "perplexity": 4.550357599553642, "muon_lr": 0.0011152020595588986, "adamw_lr": 2.788005148897247e-05, "tokens_per_sec": 822880.4768431089, "vram_gb": 2.785654067993164, "tokens_processed": 1304166400}
+{"timestamp": 1787699140.2049813, "elapsed_hours": 0.451648959186342, "event": "train_step", "step": 9960, "epoch": 1, "loss": 1.486575697660446, "perplexity": 4.421927547119361, "muon_lr": 0.001112929024527729, "adamw_lr": 2.7823225613193225e-05, "tokens_per_sec": 819822.9123207147, "vram_gb": 2.363706111907959, "tokens_processed": 1305477120}
+{"timestamp": 1787699147.6289103, "elapsed_hours": 0.4537111616796917, "event": "train_step", "step": 9970, "epoch": 1, "loss": 1.4864662992954254, "perplexity": 4.421443821935321, "muon_lr": 0.0011117924751721314, "adamw_lr": 2.7794811879303283e-05, "tokens_per_sec": 816837.8278943086, "vram_gb": 2.785654067993164, "tokens_processed": 1306787840}
+{"timestamp": 1787699155.0551727, "elapsed_hours": 0.4557740124066671, "event": "train_step", "step": 9980, "epoch": 1, "loss": 1.4906876873970032, "perplexity": 4.440147903094124, "muon_lr": 0.0011106559070069807, "adamw_lr": 2.7766397675174517e-05, "tokens_per_sec": 813879.1452927937, "vram_gb": 2.785654067993164, "tokens_processed": 1308098560}
+{"timestamp": 1787699162.4824002, "elapsed_hours": 0.45783713108963436, "event": "train_step", "step": 9990, "epoch": 1, "loss": 1.4981549382209778, "perplexity": 4.473427700841122, "muon_lr": 0.0011095193218451562, "adamw_lr": 2.7737983046128904e-05, "tokens_per_sec": 810947.166061217, "vram_gb": 2.785654067993164, "tokens_processed": 1309409280}
+{"timestamp": 1787699170.0018644, "elapsed_hours": 0.4599258712265227, "event": "train_step", "step": 10000, "epoch": 1, "loss": 1.454041965007782, "perplexity": 4.280380746106711, "muon_lr": 0.0011072461077831388, "adamw_lr": 2.7681152694578473e-05, "tokens_per_sec": 807996.0931751023, "vram_gb": 2.363706111907959, "tokens_processed": 1310720000}
+{"timestamp": 1787699182.825166, "elapsed_hours": 0.46348789950211844, "event": "validation", "step": 10000, "val_loss": 1.4727324509620667, "val_perplexity": 4.361135463928254, "best_loss": 1.4726950550079345}