Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -2767,3 +2767,31 @@
 {"timestamp": 1787647077.2064545, "elapsed_hours": 5.505638764368163, "session_id": 2, "event": "train_step", "step": 118500, "epoch": 1, "loss": 1.690584535598755, "loss_raw": 2.2069568634033203, "perplexity": 5.4226495105796895, "muon_lr": 0.0019693287382833474, "adamw_lr": 5.907986214850041e-05, "tokens_per_sec": 786772.9274725106, "vram_gb": 2.785593032836914, "grad_norm": 0.09716796875, "tokens_processed": 15532032000, "tokens_processed_billions": 15.532032}
 {"timestamp": 1787647077.2083695, "elapsed_hours": 5.505639296438959, "session_id": 2, "event": "lambda_stats", "step": 118500, "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": 1787647102.8670166, "elapsed_hours": 5.5127666983339525, "session_id": 2, "event": "validation", "step": 118500, "epoch": 1, "val_loss": 1.752578262090683, "val_perplexity": 5.769458693055474, "is_best": false, "best_loss": 1.6233880519866943, "best_perplexity": 5.070239483980759, "train_loss": 2.2069568634033203, "train_loss_avg": 1.690584535598755}
+{"timestamp": 1787647111.418405, "elapsed_hours": 5.5151420840289855, "session_id": 2, "event": "train_step", "step": 118510, "epoch": 1, "loss": 1.7259255611896516, "loss_raw": 1.8545289039611816, "perplexity": 5.617718166097204, "muon_lr": 0.001968854995132071, "adamw_lr": 5.906564985396212e-05, "tokens_per_sec": 785478.088878096, "vram_gb": 2.785593032836914, "grad_norm": 0.095703125, "tokens_processed": 15533342720, "tokens_processed_billions": 15.53334272}
+{"timestamp": 1787647118.957925, "elapsed_hours": 5.517236395080884, "session_id": 2, "event": "train_step", "step": 118520, "epoch": 1, "loss": 1.7030861032009126, "loss_raw": 1.7974631786346436, "perplexity": 5.490866652158025, "muon_lr": 0.001967907814330429, "adamw_lr": 5.903723442991286e-05, "tokens_per_sec": 785244.9893464762, "vram_gb": 2.363645076751709, "grad_norm": 0.09765625, "tokens_processed": 15534653440, "tokens_processed_billions": 15.53465344}
+{"timestamp": 1787647126.3979983, "elapsed_hours": 5.519303082227707, "session_id": 2, "event": "train_step", "step": 118530, "epoch": 1, "loss": 1.6526921844482423, "loss_raw": 1.4971561431884766, "perplexity": 5.221016863960752, "muon_lr": 0.0019674343767073785, "adamw_lr": 5.902303130122135e-05, "tokens_per_sec": 785016.014774244, "vram_gb": 2.785593032836914, "grad_norm": 0.09130859375, "tokens_processed": 15535964160, "tokens_processed_billions": 15.53596416}
+{"timestamp": 1787647133.825919, "elapsed_hours": 5.521366393433677, "session_id": 2, "event": "train_step", "step": 118540, "epoch": 1, "loss": 1.6616701412200927, "loss_raw": 1.2892708778381348, "perplexity": 5.268101975449142, "muon_lr": 0.0019669610409543802, "adamw_lr": 5.900883122863139e-05, "tokens_per_sec": 784787.6935998144, "vram_gb": 2.785593032836914, "grad_norm": 0.09619140625, "tokens_processed": 15537274880, "tokens_processed_billions": 15.53727488}
+{"timestamp": 1787647141.3089051, "elapsed_hours": 5.523445000780954, "session_id": 2, "event": "train_step", "step": 118550, "epoch": 1, "loss": 1.6117999291419982, "loss_raw": 2.0462729930877686, "perplexity": 5.01182404233495, "muon_lr": 0.0019664878070850818, "adamw_lr": 5.899463421255244e-05, "tokens_per_sec": 784557.363856497, "vram_gb": 2.785593032836914, "grad_norm": 0.08935546875, "tokens_processed": 15538585600, "tokens_processed_billions": 15.5385856}
+{"timestamp": 1787647148.8342257, "elapsed_hours": 5.52553536746237, "session_id": 2, "event": "train_step", "step": 118560, "epoch": 1, "loss": 1.5309086847305298, "loss_raw": 1.1727712154388428, "perplexity": 4.622375196271366, "muon_lr": 0.0019655416450521748, "adamw_lr": 5.896624935156523e-05, "tokens_per_sec": 784325.528542575, "vram_gb": 2.363645076751709, "grad_norm": 0.09033203125, "tokens_processed": 15539896320, "tokens_processed_billions": 15.53989632}
+{"timestamp": 1787647156.2731922, "elapsed_hours": 5.527601747115453, "session_id": 2, "event": "train_step", "step": 118570, "epoch": 1, "loss": 1.5510423064231873, "loss_raw": 2.1252944469451904, "perplexity": 4.716383538391042, "muon_lr": 0.001965068716915851, "adamw_lr": 5.895206150747551e-05, "tokens_per_sec": 784097.2897335375, "vram_gb": 2.785593032836914, "grad_norm": 0.095703125, "tokens_processed": 15541207040, "tokens_processed_billions": 15.54120704}
+{"timestamp": 1787647163.7364357, "elapsed_hours": 5.529674870371818, "session_id": 2, "event": "train_step", "step": 118580, "epoch": 1, "loss": 1.5840167033672332, "loss_raw": 1.2960747480392456, "perplexity": 4.874495945782761, "muon_lr": 0.0019645958907177977, "adamw_lr": 5.8937876721533915e-05, "tokens_per_sec": 783868.2593466999, "vram_gb": 2.785593032836914, "grad_norm": 0.0947265625, "tokens_processed": 15542517760, "tokens_processed_billions": 15.54251776}
+{"timestamp": 1787647171.1823604, "elapsed_hours": 5.531743182672395, "session_id": 2, "event": "train_step", "step": 118590, "epoch": 1, "loss": 1.5483188354969024, "loss_raw": 1.5989702939987183, "perplexity": 4.703556080480117, "muon_lr": 0.0019641231664716497, "adamw_lr": 5.892369499414948e-05, "tokens_per_sec": 783640.087528051, "vram_gb": 2.785593032836914, "grad_norm": 0.09033203125, "tokens_processed": 15543828480, "tokens_processed_billions": 15.54382848}
+{"timestamp": 1787647178.7464159, "elapsed_hours": 5.533844309184286, "session_id": 2, "event": "train_step", "step": 118600, "epoch": 1, "loss": 1.577284187078476, "loss_raw": 1.5448293685913086, "perplexity": 4.841788547517404, "muon_lr": 0.001963178023889599, "adamw_lr": 5.889534071668796e-05, "tokens_per_sec": 783407.4195817569, "vram_gb": 2.363645076751709, "grad_norm": 0.09521484375, "tokens_processed": 15545139200, "tokens_processed_billions": 15.5451392}
+{"timestamp": 1787647178.748051, "elapsed_hours": 5.53384476337168, "session_id": 2, "event": "lambda_stats", "step": 118600, "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": 1787647186.1804478, "elapsed_hours": 5.535909318129222, "session_id": 2, "event": "train_step", "step": 118610, "epoch": 1, "loss": 1.6323824667930602, "loss_raw": 1.262042760848999, "perplexity": 5.116049027254105, "muon_lr": 0.001962705605580951, "adamw_lr": 5.888116816742853e-05, "tokens_per_sec": 783180.06353375, "vram_gb": 2.785593032836914, "grad_norm": 0.1005859375, "tokens_processed": 15546449920, "tokens_processed_billions": 15.54644992}
+{"timestamp": 1787647193.6261015, "elapsed_hours": 5.537977555327945, "session_id": 2, "event": "train_step", "step": 118620, "epoch": 1, "loss": 1.6456483912467956, "loss_raw": 2.0536365509033203, "perplexity": 5.184370317734473, "muon_lr": 0.0019622332892787215, "adamw_lr": 5.8866998678361635e-05, "tokens_per_sec": 782952.4168508386, "vram_gb": 2.785593032836914, "grad_norm": 0.0927734375, "tokens_processed": 15547760640, "tokens_processed_billions": 15.54776064}
+{"timestamp": 1787647201.072112, "elapsed_hours": 5.540045891536606, "session_id": 2, "event": "train_step", "step": 118630, "epoch": 1, "loss": 1.660806964635849, "loss_raw": 1.385420799255371, "perplexity": 5.2635566351783885, "muon_lr": 0.001961761074996527, "adamw_lr": 5.88528322498958e-05, "tokens_per_sec": 782724.9270999193, "vram_gb": 2.785593032836914, "grad_norm": 0.09423828125, "tokens_processed": 15549071360, "tokens_processed_billions": 15.54907136}
+{"timestamp": 1787647208.5946388, "elapsed_hours": 5.542135482231776, "session_id": 2, "event": "train_step", "step": 118640, "epoch": 1, "loss": 1.6841987621784211, "loss_raw": 1.6872035264968872, "perplexity": 5.388132027177642, "muon_lr": 0.0019608169525467192, "adamw_lr": 5.882450857640158e-05, "tokens_per_sec": 782494.5930652436, "vram_gb": 2.363645076751709, "grad_norm": 0.09423828125, "tokens_processed": 15550382080, "tokens_processed_billions": 15.55038208}
+{"timestamp": 1787647216.0355146, "elapsed_hours": 5.544202392167516, "session_id": 2, "event": "train_step", "step": 118650, "epoch": 1, "loss": 1.7022926342487334, "loss_raw": 1.7159126996994019, "perplexity": 5.486511547997445, "muon_lr": 0.0019603450444063315, "adamw_lr": 5.8810351332189945e-05, "tokens_per_sec": 782267.6504037271, "vram_gb": 2.785593032836914, "grad_norm": 0.09765625, "tokens_processed": 15551692800, "tokens_processed_billions": 15.5516928}
+{"timestamp": 1787647223.4859257, "elapsed_hours": 5.546271950801214, "session_id": 2, "event": "train_step", "step": 118660, "epoch": 1, "loss": 1.6635880362987518, "loss_raw": 1.5317682027816772, "perplexity": 5.278215337385257, "muon_lr": 0.0019598732383404338, "adamw_lr": 5.879619715021301e-05, "tokens_per_sec": 782040.5000824507, "vram_gb": 2.785593032836914, "grad_norm": 0.09423828125, "tokens_processed": 15553003520, "tokens_processed_billions": 15.55300352}
+{"timestamp": 1787647230.925877, "elapsed_hours": 5.548338603973389, "session_id": 2, "event": "train_step", "step": 118670, "epoch": 1, "loss": 1.6553427469730377, "loss_raw": 2.211235284805298, "perplexity": 5.234873851895488, "muon_lr": 0.00195940153436263, "adamw_lr": 5.878204603087889e-05, "tokens_per_sec": 781813.9342639936, "vram_gb": 2.785593032836914, "grad_norm": 0.0927734375, "tokens_processed": 15554314240, "tokens_processed_billions": 15.55431424}
+{"timestamp": 1787647238.4608788, "elapsed_hours": 5.550431660016378, "session_id": 2, "event": "train_step", "step": 118680, "epoch": 1, "loss": 1.6919376063346863, "loss_raw": 1.8386008739471436, "perplexity": 5.429991705077748, "muon_lr": 0.0019584584327257165, "adamw_lr": 5.875375298177148e-05, "tokens_per_sec": 781583.7976743535, "vram_gb": 2.363645076751709, "grad_norm": 0.09912109375, "tokens_processed": 15555624960, "tokens_processed_billions": 15.55562496}
+{"timestamp": 1787647245.8979166, "elapsed_hours": 5.552497503889931, "session_id": 2, "event": "train_step", "step": 118690, "epoch": 1, "loss": 1.7692861700057982, "loss_raw": 1.850093960762024, "perplexity": 5.86666406556208, "muon_lr": 0.0019579870350938037, "adamw_lr": 5.8739611052814096e-05, "tokens_per_sec": 781357.6859376238, "vram_gb": 2.785593032836914, "grad_norm": 0.1005859375, "tokens_processed": 15556935680, "tokens_processed_billions": 15.55693568}
+{"timestamp": 1787647253.339736, "elapsed_hours": 5.554564676086108, "session_id": 2, "event": "train_step", "step": 118700, "epoch": 1, "loss": 1.7214273929595947, "loss_raw": 0.9690784811973572, "perplexity": 5.5925054726970505, "muon_lr": 0.0019575157396043794, "adamw_lr": 5.872547218813137e-05, "tokens_per_sec": 781131.55385166, "vram_gb": 2.785593032836914, "grad_norm": 0.0927734375, "tokens_processed": 15558246400, "tokens_processed_billions": 15.5582464}
+{"timestamp": 1787647253.341769, "elapsed_hours": 5.554565240740776, "session_id": 2, "event": "lambda_stats", "step": 118700, "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": 1787647260.8118958, "elapsed_hours": 5.556640275849237, "session_id": 2, "event": "train_step", "step": 118710, "epoch": 1, "loss": 1.7846691417694092, "loss_raw": 1.7522056102752686, "perplexity": 5.957608497659266, "muon_lr": 0.001957044546271033, "adamw_lr": 5.871133638813099e-05, "tokens_per_sec": 780904.4009212843, "vram_gb": 2.785593032836914, "grad_norm": 0.09033203125, "tokens_processed": 15559557120, "tokens_processed_billions": 15.55955712}
+{"timestamp": 1787647268.3495193, "elapsed_hours": 5.558734060194757, "session_id": 2, "event": "train_step", "step": 118720, "epoch": 1, "loss": 1.770236840248108, "loss_raw": 2.0755538940429688, "perplexity": 5.8722439804200945, "muon_lr": 0.001956102466126929, "adamw_lr": 5.868307398380786e-05, "tokens_per_sec": 780674.8541548905, "vram_gb": 2.363645076751709, "grad_norm": 0.09228515625, "tokens_processed": 15560867840, "tokens_processed_billions": 15.56086784}
+{"timestamp": 1787647275.79826, "elapsed_hours": 5.56080315483941, "session_id": 2, "event": "train_step", "step": 118730, "epoch": 1, "loss": 1.720800609588623, "loss_raw": 1.4966754913330078, "perplexity": 5.589001281563782, "muon_lr": 0.0019556315793433383, "adamw_lr": 5.866894738030014e-05, "tokens_per_sec": 780448.9637706801, "vram_gb": 2.785593032836914, "grad_norm": 0.09326171875, "tokens_processed": 15562178560, "tokens_processed_billions": 15.56217856}
+{"timestamp": 1787647283.232239, "elapsed_hours": 5.562868148949411, "session_id": 2, "event": "train_step", "step": 118740, "epoch": 1, "loss": 1.6533761262893676, "loss_raw": 2.178661823272705, "perplexity": 5.224588957260025, "muon_lr": 0.001955160794770161, "adamw_lr": 5.865482384310483e-05, "tokens_per_sec": 780223.8157724342, "vram_gb": 2.785593032836914, "grad_norm": 0.0986328125, "tokens_processed": 15563489280, "tokens_processed_billions": 15.56348928}
+{"timestamp": 1787647290.6740403, "elapsed_hours": 5.564935316112306, "session_id": 2, "event": "train_step", "step": 118750, "epoch": 1, "loss": 1.684439709186554, "loss_raw": 1.6103296279907227, "perplexity": 5.389430437886821, "muon_lr": 0.001954690112420973, "adamw_lr": 5.864070337262918e-05, "tokens_per_sec": 779998.5293223877, "vram_gb": 2.785593032836914, "grad_norm": 0.099609375, "tokens_processed": 15564800000, "tokens_processed_billions": 15.5648}
+{"timestamp": 1787647298.2112072, "elapsed_hours": 5.567028973433707, "session_id": 2, "event": "train_step", "step": 118760, "epoch": 1, "loss": 1.5994374668598175, "loss_raw": 0.53460294008255, "perplexity": 4.950246963042608, "muon_lr": 0.001953749054448859, "adamw_lr": 5.861247163346576e-05, "tokens_per_sec": 779769.6858408054, "vram_gb": 2.363645076751709, "grad_norm": 0.087890625, "tokens_processed": 15566110720, "tokens_processed_billions": 15.56611072}