Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -3727,3 +3727,20 @@
 {"timestamp": 1787541026.833551, "elapsed_hours": 4.000818067656623, "session_id": 2, "event": "train_step", "step": 33000, "epoch": 1, "loss": 2.0204446506500244, "loss_raw": 2.1008999347686768, "perplexity": 7.54167760007118, "muon_lr": 0.009060434502366831, "adamw_lr": 0.0002718130350710049, "tokens_per_sec": 150273.21141929785, "vram_gb": 27.299458026885986, "grad_norm": 0.1591796875, "tokens_processed": 2162688000, "tokens_processed_billions": 2.162688}
 {"timestamp": 1787541026.8368824, "elapsed_hours": 4.000818993316757, "session_id": 2, "event": "lambda_stats", "step": 33000, "epoch": 1, "lambda_min": 1.109375, "lambda_max": 2.015625, "lambda_mean": 1.1966145833333333, "lambda_std": 0.20269468977080954, "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, 1.125], "local_layers": 9, "global_layers": 9}
 {"timestamp": 1787541040.3042943, "elapsed_hours": 4.0045599412255815, "session_id": 2, "event": "validation", "step": 33000, "epoch": 1, "val_loss": 2.0201363903284073, "val_perplexity": 7.5393531583936255, "is_best": true, "best_loss": 2.0201363903284073, "best_perplexity": 7.5393531583936255, "train_loss": 2.1008999347686768, "train_loss_avg": 2.0204446506500244}
+{"timestamp": 1787541045.161962, "elapsed_hours": 4.005909292962816, "session_id": 2, "event": "train_step", "step": 33010, "epoch": 1, "loss": 1.991683634519577, "loss_raw": 2.130929946899414, "perplexity": 7.327860821700397, "muon_lr": 0.009059967124002224, "adamw_lr": 0.00027179901372006667, "tokens_per_sec": 150127.55658694686, "vram_gb": 27.710740089416504, "grad_norm": 0.1689453125, "tokens_processed": 2163343360, "tokens_processed_billions": 2.16334336}
+{"timestamp": 1787541049.1546345, "elapsed_hours": 4.0070183686415355, "session_id": 2, "event": "train_step", "step": 33020, "epoch": 1, "loss": 2.0430376756191255, "loss_raw": 2.4528377056121826, "perplexity": 7.714006293339019, "muon_lr": 0.009059499642977378, "adamw_lr": 0.00027178498928932126, "tokens_per_sec": 150131.437774283, "vram_gb": 27.710740089416504, "grad_norm": 0.1474609375, "tokens_processed": 2163998720, "tokens_processed_billions": 2.16399872}
+{"timestamp": 1787541053.131435, "elapsed_hours": 4.008123035497135, "session_id": 2, "event": "train_step", "step": 33030, "epoch": 1, "loss": 2.0784998869895936, "loss_raw": 2.502650260925293, "perplexity": 7.992470308219242, "muon_lr": 0.00905903205930577, "adamw_lr": 0.0002717709617791731, "tokens_per_sec": 150135.48249999533, "vram_gb": 27.710740089416504, "grad_norm": 0.23828125, "tokens_processed": 2164654080, "tokens_processed_billions": 2.16465408}
+{"timestamp": 1787541057.259198, "elapsed_hours": 4.009269636339612, "session_id": 2, "event": "train_step", "step": 33040, "epoch": 1, "loss": 2.0290070748329163, "loss_raw": 2.417379379272461, "perplexity": 7.606529892933377, "muon_lr": 0.00905809658407622, "adamw_lr": 0.00027174289752228653, "tokens_per_sec": 150137.9524578204, "vram_gb": 27.299458026885986, "grad_norm": 0.1484375, "tokens_processed": 2165309440, "tokens_processed_billions": 2.16530944}
+{"timestamp": 1787541061.2480514, "elapsed_hours": 4.010377651055654, "session_id": 2, "event": "train_step", "step": 33050, "epoch": 1, "loss": 1.9411774289608001, "loss_raw": 2.080911159515381, "perplexity": 6.966949231048584, "muon_lr": 0.009057628692545252, "adamw_lr": 0.00027172886077635747, "tokens_per_sec": 150141.8682226518, "vram_gb": 27.710740089416504, "grad_norm": 0.1572265625, "tokens_processed": 2165964800, "tokens_processed_billions": 2.1659648}
+{"timestamp": 1787541065.2833862, "elapsed_hours": 4.0114985775285295, "session_id": 2, "event": "train_step", "step": 33060, "epoch": 1, "loss": 1.9831329798698425, "loss_raw": 1.579886794090271, "perplexity": 7.265469936086025, "muon_lr": 0.009057160698421479, "adamw_lr": 0.0002717148209526443, "tokens_per_sec": 150145.29722891716, "vram_gb": 27.710740089416504, "grad_norm": 0.55078125, "tokens_processed": 2166620160, "tokens_processed_billions": 2.16662016}
+{"timestamp": 1787541069.2507873, "elapsed_hours": 4.012600633435779, "session_id": 2, "event": "train_step", "step": 33070, "epoch": 1, "loss": 1.9817612171173096, "loss_raw": 1.1632922887802124, "perplexity": 7.255510267760859, "muon_lr": 0.009056692601718395, "adamw_lr": 0.0002717007780515518, "tokens_per_sec": 150149.43216667735, "vram_gb": 27.710740089416504, "grad_norm": 0.1669921875, "tokens_processed": 2167275520, "tokens_processed_billions": 2.16727552}
+{"timestamp": 1787541073.417261, "elapsed_hours": 4.013757987221082, "session_id": 2, "event": "train_step", "step": 33080, "epoch": 1, "loss": 1.99486811876297, "loss_raw": 2.492245674133301, "perplexity": 7.351233474194958, "muon_lr": 0.009055756100628296, "adamw_lr": 0.00027167268301884877, "tokens_per_sec": 150151.49399321797, "vram_gb": 27.299458026885986, "grad_norm": 0.1630859375, "tokens_processed": 2167930880, "tokens_processed_billions": 2.16793088}
+{"timestamp": 1787541077.3883877, "elapsed_hours": 4.014861077864965, "session_id": 2, "event": "train_step", "step": 33090, "epoch": 1, "loss": 2.0985755848884584, "loss_raw": 1.8769466876983643, "perplexity": 8.154546177218254, "muon_lr": 0.009055287696268285, "adamw_lr": 0.0002716586308880485, "tokens_per_sec": 150155.5856699989, "vram_gb": 27.710740089416504, "grad_norm": 0.158203125, "tokens_processed": 2168586240, "tokens_processed_billions": 2.16858624}
+{"timestamp": 1787541081.3570485, "elapsed_hours": 4.015963483783934, "session_id": 2, "event": "train_step", "step": 33100, "epoch": 1, "loss": 2.0239615297317504, "loss_raw": 1.2673070430755615, "perplexity": 7.568247462374825, "muon_lr": 0.00905481918938298, "adamw_lr": 0.0002716445756814893, "tokens_per_sec": 150159.69933420772, "vram_gb": 27.710740089416504, "grad_norm": 0.138671875, "tokens_processed": 2169241600, "tokens_processed_billions": 2.1692416}
+{"timestamp": 1787541081.360464, "elapsed_hours": 4.015964432822334, "session_id": 2, "event": "lambda_stats", "step": 33100, "epoch": 1, "lambda_min": 1.109375, "lambda_max": 2.015625, "lambda_mean": 1.1966145833333333, "lambda_std": 0.20269468977080954, "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, 1.125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787541085.3558805, "elapsed_hours": 4.017074270248413, "session_id": 2, "event": "train_step", "step": 33110, "epoch": 1, "loss": 1.9955676317214965, "loss_raw": 1.2228535413742065, "perplexity": 7.356377556237578, "muon_lr": 0.009054350579985885, "adamw_lr": 0.0002716305173995765, "tokens_per_sec": 150163.49952286796, "vram_gb": 27.710740089416504, "grad_norm": 0.126953125, "tokens_processed": 2169896960, "tokens_processed_billions": 2.16989696}
+{"timestamp": 1787541089.5822794, "elapsed_hours": 4.0182482700877715, "session_id": 2, "event": "train_step", "step": 33120, "epoch": 1, "loss": 1.8786216914653777, "loss_raw": 1.1925150156021118, "perplexity": 6.544478332608284, "muon_lr": 0.009053413053710397, "adamw_lr": 0.0002716023916113118, "tokens_per_sec": 150164.93143735826, "vram_gb": 27.299458026885986, "grad_norm": 0.13671875, "tokens_processed": 2170552320, "tokens_processed_billions": 2.17055232}
+{"timestamp": 1787541093.5946462, "elapsed_hours": 4.019362816545699, "session_id": 2, "event": "train_step", "step": 33130, "epoch": 1, "loss": 1.889720183610916, "loss_raw": 2.1252193450927734, "perplexity": 6.617516732315619, "muon_lr": 0.009052944136859035, "adamw_lr": 0.000271588324105771, "tokens_per_sec": 150168.5873652073, "vram_gb": 27.710740089416504, "grad_norm": 0.1455078125, "tokens_processed": 2171207680, "tokens_processed_billions": 2.17120768}
+{"timestamp": 1787541097.586105, "elapsed_hours": 4.020471554928355, "session_id": 2, "event": "train_step", "step": 33140, "epoch": 1, "loss": 1.9376701939105987, "loss_raw": 2.3959147930145264, "perplexity": 6.942557301630516, "muon_lr": 0.009052475117549957, "adamw_lr": 0.0002715742535264987, "tokens_per_sec": 150172.45644972337, "vram_gb": 27.710740089416504, "grad_norm": 0.1650390625, "tokens_processed": 2171863040, "tokens_processed_billions": 2.17186304}
+{"timestamp": 1787541101.6418967, "elapsed_hours": 4.021598163776927, "session_id": 2, "event": "train_step", "step": 33150, "epoch": 1, "loss": 2.090186823606491, "loss_raw": 2.2243900299072266, "perplexity": 8.086425758417157, "muon_lr": 0.009052005995796693, "adamw_lr": 0.0002715601798739007, "tokens_per_sec": 150175.656407658, "vram_gb": 27.710740089416504, "grad_norm": 0.173828125, "tokens_processed": 2172518400, "tokens_processed_billions": 2.1725184}
+{"timestamp": 1787541105.7880845, "elapsed_hours": 4.022749882472886, "session_id": 2, "event": "train_step", "step": 33160, "epoch": 1, "loss": 2.042172040939331, "loss_raw": 0.9378709197044373, "perplexity": 7.707331671280303, "muon_lr": 0.009051067445011707, "adamw_lr": 0.00027153202335035117, "tokens_per_sec": 150177.91597800463, "vram_gb": 27.299458026885986, "grad_norm": 0.138671875, "tokens_processed": 2173173760, "tokens_processed_billions": 2.17317376}