| { | |
| "best_global_step": 4360, | |
| "best_metric": 1.0, | |
| "best_model_checkpoint": "gcbert_ckpt/checkpoint-4360", | |
| "epoch": 4.0, | |
| "eval_steps": 500, | |
| "global_step": 17440, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.022938410368161487, | |
| "grad_norm": 29.002798080444336, | |
| "learning_rate": 2.809633027522936e-06, | |
| "loss": 1.1878, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.045876820736322974, | |
| "grad_norm": 68.14753723144531, | |
| "learning_rate": 5.676605504587156e-06, | |
| "loss": 0.6556, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.06881523110448445, | |
| "grad_norm": 54.466773986816406, | |
| "learning_rate": 8.543577981651376e-06, | |
| "loss": 0.288, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.09175364147264595, | |
| "grad_norm": 0.0029387406539171934, | |
| "learning_rate": 1.1410550458715597e-05, | |
| "loss": 0.0993, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.11469205184080743, | |
| "grad_norm": 0.00012727684224955738, | |
| "learning_rate": 1.4277522935779817e-05, | |
| "loss": 0.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1376304622089689, | |
| "grad_norm": 0.0007764559122733772, | |
| "learning_rate": 1.714449541284404e-05, | |
| "loss": 0.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.1605688725771304, | |
| "grad_norm": 7.87975950515829e-05, | |
| "learning_rate": 2.0011467889908257e-05, | |
| "loss": 0.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.1835072829452919, | |
| "grad_norm": 0.0004955148906446993, | |
| "learning_rate": 2.2878440366972478e-05, | |
| "loss": 0.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.2064456933134534, | |
| "grad_norm": 0.00011743771756300703, | |
| "learning_rate": 2.57454128440367e-05, | |
| "loss": 0.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.22938410368161485, | |
| "grad_norm": 0.0005010354216210544, | |
| "learning_rate": 2.861238532110092e-05, | |
| "loss": 0.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.25232251404977635, | |
| "grad_norm": 0.00021343142725527287, | |
| "learning_rate": 3.147935779816514e-05, | |
| "loss": 0.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.2752609244179378, | |
| "grad_norm": 0.00011905051360372454, | |
| "learning_rate": 3.434633027522936e-05, | |
| "loss": 0.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.29819933478609933, | |
| "grad_norm": 0.00038216536631807685, | |
| "learning_rate": 3.7213302752293576e-05, | |
| "loss": 0.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.3211377451542608, | |
| "grad_norm": 4.422829078976065e-05, | |
| "learning_rate": 4.00802752293578e-05, | |
| "loss": 0.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.3440761555224223, | |
| "grad_norm": 0.0002208059886470437, | |
| "learning_rate": 4.2947247706422025e-05, | |
| "loss": 0.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.3670145658905838, | |
| "grad_norm": 5.1115232054144144e-05, | |
| "learning_rate": 4.581422018348624e-05, | |
| "loss": 0.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.38995297625874525, | |
| "grad_norm": 7.042491051834077e-05, | |
| "learning_rate": 4.868119266055046e-05, | |
| "loss": 0.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.4128913866269068, | |
| "grad_norm": 3.786469824262895e-05, | |
| "learning_rate": 4.982798165137615e-05, | |
| "loss": 0.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.43582979699506824, | |
| "grad_norm": 1.468727168685291e-05, | |
| "learning_rate": 4.950942915392457e-05, | |
| "loss": 0.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.4587682073632297, | |
| "grad_norm": 3.0084795071161352e-05, | |
| "learning_rate": 4.919087665647299e-05, | |
| "loss": 0.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.48170661773139123, | |
| "grad_norm": 2.5331471988465637e-05, | |
| "learning_rate": 4.887232415902141e-05, | |
| "loss": 0.0, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.5046450280995527, | |
| "grad_norm": 2.618041617097333e-05, | |
| "learning_rate": 4.855377166156983e-05, | |
| "loss": 0.0, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.5275834384677142, | |
| "grad_norm": 2.4747334464336745e-05, | |
| "learning_rate": 4.823521916411825e-05, | |
| "loss": 0.0, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.5505218488358756, | |
| "grad_norm": 1.4968503819545731e-05, | |
| "learning_rate": 4.791666666666667e-05, | |
| "loss": 0.0, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.5734602592040372, | |
| "grad_norm": 0.00038964845589362085, | |
| "learning_rate": 4.759811416921509e-05, | |
| "loss": 0.0, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.5963986695721987, | |
| "grad_norm": 2.8935370210092515e-05, | |
| "learning_rate": 4.7279561671763515e-05, | |
| "loss": 0.0, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.6193370799403601, | |
| "grad_norm": 8.388165042561013e-06, | |
| "learning_rate": 4.6961009174311924e-05, | |
| "loss": 0.0, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.6422754903085216, | |
| "grad_norm": 1.768140464264434e-05, | |
| "learning_rate": 4.664245667686035e-05, | |
| "loss": 0.0, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.6652139006766831, | |
| "grad_norm": 1.36974013003055e-05, | |
| "learning_rate": 4.6323904179408764e-05, | |
| "loss": 0.0, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.6881523110448446, | |
| "grad_norm": 9.414131454832386e-06, | |
| "learning_rate": 4.600535168195719e-05, | |
| "loss": 0.0, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.7110907214130061, | |
| "grad_norm": 4.622045162250288e-05, | |
| "learning_rate": 4.568679918450561e-05, | |
| "loss": 0.0, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.7340291317811676, | |
| "grad_norm": 2.692650923563633e-05, | |
| "learning_rate": 4.536824668705403e-05, | |
| "loss": 0.0, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.756967542149329, | |
| "grad_norm": 1.0936350008705631e-05, | |
| "learning_rate": 4.504969418960245e-05, | |
| "loss": 0.0, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.7799059525174905, | |
| "grad_norm": 5.966558546788292e-06, | |
| "learning_rate": 4.473114169215087e-05, | |
| "loss": 0.0, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.802844362885652, | |
| "grad_norm": 9.507278264209162e-06, | |
| "learning_rate": 4.441258919469929e-05, | |
| "loss": 0.0, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.8257827732538136, | |
| "grad_norm": 1.0006630873249378e-05, | |
| "learning_rate": 4.409403669724771e-05, | |
| "loss": 0.0, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.848721183621975, | |
| "grad_norm": 5.595240509137511e-06, | |
| "learning_rate": 4.377548419979613e-05, | |
| "loss": 0.0, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.8716595939901365, | |
| "grad_norm": 8.78509945323458e-06, | |
| "learning_rate": 4.3456931702344547e-05, | |
| "loss": 0.0, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.894598004358298, | |
| "grad_norm": 1.4489376553683542e-05, | |
| "learning_rate": 4.313837920489297e-05, | |
| "loss": 0.0, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.9175364147264594, | |
| "grad_norm": 0.0006358507671393454, | |
| "learning_rate": 4.281982670744139e-05, | |
| "loss": 0.0, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.940474825094621, | |
| "grad_norm": 2.2141068257042207e-05, | |
| "learning_rate": 4.250127420998981e-05, | |
| "loss": 0.0, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.9634132354627825, | |
| "grad_norm": 2.3798056645318866e-06, | |
| "learning_rate": 4.2182721712538226e-05, | |
| "loss": 0.0, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.9863516458309439, | |
| "grad_norm": 2.4955927074188367e-05, | |
| "learning_rate": 4.186416921508665e-05, | |
| "loss": 0.0, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 1.0, | |
| "eval_f1": 1.0, | |
| "eval_loss": 3.4182854463660206e-11, | |
| "eval_precision": 1.0, | |
| "eval_recall": 1.0, | |
| "eval_runtime": 342.7946, | |
| "eval_samples_per_second": 50.867, | |
| "eval_steps_per_second": 3.18, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 1.0091753641472645, | |
| "grad_norm": 2.645613676577341e-06, | |
| "learning_rate": 4.1545616717635066e-05, | |
| "loss": 0.0, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.032113774515426, | |
| "grad_norm": 1.391179011989152e-05, | |
| "learning_rate": 4.122706422018349e-05, | |
| "loss": 0.0, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.0550521848835877, | |
| "grad_norm": 6.799902621423826e-06, | |
| "learning_rate": 4.0908511722731906e-05, | |
| "loss": 0.0, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.077990595251749, | |
| "grad_norm": 5.317841896612663e-06, | |
| "learning_rate": 4.058995922528033e-05, | |
| "loss": 0.0, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.1009290056199106, | |
| "grad_norm": 6.453754849644611e-06, | |
| "learning_rate": 4.0271406727828745e-05, | |
| "loss": 0.0, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.123867415988072, | |
| "grad_norm": 3.003899337272742e-06, | |
| "learning_rate": 3.995285423037717e-05, | |
| "loss": 0.0, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.1468058263562335, | |
| "grad_norm": 5.224228516453877e-06, | |
| "learning_rate": 3.9634301732925585e-05, | |
| "loss": 0.0, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.169744236724395, | |
| "grad_norm": 2.482443960616365e-06, | |
| "learning_rate": 3.931574923547401e-05, | |
| "loss": 0.0, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.1926826470925564, | |
| "grad_norm": 1.988922804230242e-06, | |
| "learning_rate": 3.899719673802243e-05, | |
| "loss": 0.0, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.215621057460718, | |
| "grad_norm": 2.143744268323644e-06, | |
| "learning_rate": 3.867864424057085e-05, | |
| "loss": 0.0, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 1.2385594678288794, | |
| "grad_norm": 5.1592205636552535e-06, | |
| "learning_rate": 3.836009174311927e-05, | |
| "loss": 0.0, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.261497878197041, | |
| "grad_norm": 3.7546260500676e-06, | |
| "learning_rate": 3.804153924566769e-05, | |
| "loss": 0.0, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.2844362885652023, | |
| "grad_norm": 2.9266414003359387e-06, | |
| "learning_rate": 3.772298674821611e-05, | |
| "loss": 0.0, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.3073746989333639, | |
| "grad_norm": 9.482760106038768e-06, | |
| "learning_rate": 3.740443425076453e-05, | |
| "loss": 0.0, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.3303131093015255, | |
| "grad_norm": 5.139393579156604e-06, | |
| "learning_rate": 3.7085881753312944e-05, | |
| "loss": 0.0, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.3532515196696868, | |
| "grad_norm": 1.1174703104188666e-05, | |
| "learning_rate": 3.676732925586137e-05, | |
| "loss": 0.0, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.3761899300378484, | |
| "grad_norm": 1.985909420909593e-06, | |
| "learning_rate": 3.6448776758409784e-05, | |
| "loss": 0.0, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.3991283404060098, | |
| "grad_norm": 9.205145761370659e-05, | |
| "learning_rate": 3.613022426095821e-05, | |
| "loss": 0.0, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 1.4220667507741713, | |
| "grad_norm": 4.935003289574524e-06, | |
| "learning_rate": 3.5811671763506624e-05, | |
| "loss": 0.0, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 1.445005161142333, | |
| "grad_norm": 1.7747262290868093e-06, | |
| "learning_rate": 3.549630479102956e-05, | |
| "loss": 0.002, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.4679435715104943, | |
| "grad_norm": 2.3939119273563847e-06, | |
| "learning_rate": 3.517775229357798e-05, | |
| "loss": 0.0, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 1.4908819818786558, | |
| "grad_norm": 1.2233704183017835e-06, | |
| "learning_rate": 3.48591997961264e-05, | |
| "loss": 0.0, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.5138203922468172, | |
| "grad_norm": 1.0319820830773097e-05, | |
| "learning_rate": 3.454064729867482e-05, | |
| "loss": 0.0, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 1.5367588026149788, | |
| "grad_norm": 2.717318920986145e-06, | |
| "learning_rate": 3.422209480122324e-05, | |
| "loss": 0.0, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 1.5596972129831403, | |
| "grad_norm": 1.162458488579432e-06, | |
| "learning_rate": 3.390354230377166e-05, | |
| "loss": 0.0, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 1.5826356233513017, | |
| "grad_norm": 1.7770501017366769e-06, | |
| "learning_rate": 3.3584989806320086e-05, | |
| "loss": 0.0, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 1.6055740337194633, | |
| "grad_norm": 1.4451183005803614e-06, | |
| "learning_rate": 3.32664373088685e-05, | |
| "loss": 0.0, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.6285124440876246, | |
| "grad_norm": 3.033783741557272e-06, | |
| "learning_rate": 3.2947884811416926e-05, | |
| "loss": 0.0, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 1.6514508544557862, | |
| "grad_norm": 1.1340368928358657e-06, | |
| "learning_rate": 3.262933231396534e-05, | |
| "loss": 0.0, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 1.6743892648239478, | |
| "grad_norm": 1.2689773711827002e-06, | |
| "learning_rate": 3.2310779816513766e-05, | |
| "loss": 0.0, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 1.6973276751921091, | |
| "grad_norm": 2.7127709927299293e-06, | |
| "learning_rate": 3.199222731906218e-05, | |
| "loss": 0.0, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 1.7202660855602707, | |
| "grad_norm": 1.0890358908000053e-06, | |
| "learning_rate": 3.1673674821610605e-05, | |
| "loss": 0.0, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.743204495928432, | |
| "grad_norm": 1.0123217180080246e-06, | |
| "learning_rate": 3.135512232415902e-05, | |
| "loss": 0.0, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 1.7661429062965937, | |
| "grad_norm": 1.0123065976586076e-06, | |
| "learning_rate": 3.1036569826707445e-05, | |
| "loss": 0.0, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 1.7890813166647552, | |
| "grad_norm": 1.1159958148709848e-06, | |
| "learning_rate": 3.071801732925586e-05, | |
| "loss": 0.0, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 1.8120197270329166, | |
| "grad_norm": 7.618372137585538e-07, | |
| "learning_rate": 3.039946483180428e-05, | |
| "loss": 0.0, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 1.8349581374010782, | |
| "grad_norm": 1.1276981695118593e-06, | |
| "learning_rate": 3.00809123343527e-05, | |
| "loss": 0.0, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.8578965477692395, | |
| "grad_norm": 1.2691349411397823e-06, | |
| "learning_rate": 2.976235983690112e-05, | |
| "loss": 0.0, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 1.880834958137401, | |
| "grad_norm": 1.1746386689992505e-06, | |
| "learning_rate": 2.944380733944954e-05, | |
| "loss": 0.0, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 1.9037733685055627, | |
| "grad_norm": 2.057612618955318e-06, | |
| "learning_rate": 2.912525484199796e-05, | |
| "loss": 0.0, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 1.926711778873724, | |
| "grad_norm": 8.718508865968033e-07, | |
| "learning_rate": 2.880670234454638e-05, | |
| "loss": 0.0, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 1.9496501892418854, | |
| "grad_norm": 1.6281912849080982e-06, | |
| "learning_rate": 2.8488149847094804e-05, | |
| "loss": 0.0, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.972588599610047, | |
| "grad_norm": 6.946272037566814e-07, | |
| "learning_rate": 2.8169597349643224e-05, | |
| "loss": 0.0, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 1.9955270099782085, | |
| "grad_norm": 7.4337913247291e-07, | |
| "learning_rate": 2.7851044852191644e-05, | |
| "loss": 0.0, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 1.0, | |
| "eval_f1": 1.0, | |
| "eval_loss": 0.0, | |
| "eval_precision": 1.0, | |
| "eval_recall": 1.0, | |
| "eval_runtime": 343.3577, | |
| "eval_samples_per_second": 50.784, | |
| "eval_steps_per_second": 3.175, | |
| "step": 8720 | |
| }, | |
| { | |
| "epoch": 2.018350728294529, | |
| "grad_norm": 2.445955033181235e-06, | |
| "learning_rate": 2.7532492354740064e-05, | |
| "loss": 0.0, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 2.0412891386626906, | |
| "grad_norm": 8.972128284767678e-07, | |
| "learning_rate": 2.7213939857288484e-05, | |
| "loss": 0.0, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 2.064227549030852, | |
| "grad_norm": 5.759555961049045e-07, | |
| "learning_rate": 2.6895387359836904e-05, | |
| "loss": 0.0, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.0871659593990137, | |
| "grad_norm": 7.570725415462221e-07, | |
| "learning_rate": 2.6576834862385324e-05, | |
| "loss": 0.0, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 2.1101043697671753, | |
| "grad_norm": 6.959637630643556e-07, | |
| "learning_rate": 2.6258282364933743e-05, | |
| "loss": 0.0, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 2.1330427801353364, | |
| "grad_norm": 6.756280868103204e-07, | |
| "learning_rate": 2.593972986748216e-05, | |
| "loss": 0.0, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 2.155981190503498, | |
| "grad_norm": 7.722316581748601e-07, | |
| "learning_rate": 2.562117737003058e-05, | |
| "loss": 0.0, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 2.1789196008716596, | |
| "grad_norm": 5.774248279521998e-07, | |
| "learning_rate": 2.5302624872579e-05, | |
| "loss": 0.0, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 2.201858011239821, | |
| "grad_norm": 5.61651461339352e-07, | |
| "learning_rate": 2.498407237512742e-05, | |
| "loss": 0.0, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 2.2247964216079827, | |
| "grad_norm": 3.2662783269188367e-06, | |
| "learning_rate": 2.466551987767584e-05, | |
| "loss": 0.0, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 2.247734831976144, | |
| "grad_norm": 5.740246820096218e-07, | |
| "learning_rate": 2.4346967380224263e-05, | |
| "loss": 0.0, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 2.2706732423443055, | |
| "grad_norm": 8.198475143217365e-07, | |
| "learning_rate": 2.4028414882772683e-05, | |
| "loss": 0.0, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 2.293611652712467, | |
| "grad_norm": 5.943483074588585e-07, | |
| "learning_rate": 2.3709862385321103e-05, | |
| "loss": 0.0, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.3165500630806286, | |
| "grad_norm": 4.994344067199563e-07, | |
| "learning_rate": 2.3391309887869522e-05, | |
| "loss": 0.0, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 2.33948847344879, | |
| "grad_norm": 5.787817372038262e-07, | |
| "learning_rate": 2.3072757390417942e-05, | |
| "loss": 0.0, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 2.3624268838169513, | |
| "grad_norm": 6.035662067915837e-07, | |
| "learning_rate": 2.2754204892966362e-05, | |
| "loss": 0.0, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 2.385365294185113, | |
| "grad_norm": 6.968316483835224e-07, | |
| "learning_rate": 2.2435652395514782e-05, | |
| "loss": 0.0, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 2.4083037045532745, | |
| "grad_norm": 5.578097557190631e-07, | |
| "learning_rate": 2.2117099898063202e-05, | |
| "loss": 0.0, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.431242114921436, | |
| "grad_norm": 5.246223508947878e-07, | |
| "learning_rate": 2.1798547400611622e-05, | |
| "loss": 0.0, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 2.4541805252895976, | |
| "grad_norm": 5.867889285582351e-07, | |
| "learning_rate": 2.1479994903160042e-05, | |
| "loss": 0.0, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 2.4771189356577588, | |
| "grad_norm": 4.947526122123236e-07, | |
| "learning_rate": 2.116144240570846e-05, | |
| "loss": 0.0, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 2.5000573460259203, | |
| "grad_norm": 7.670053605579596e-07, | |
| "learning_rate": 2.084288990825688e-05, | |
| "loss": 0.0, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 2.522995756394082, | |
| "grad_norm": 6.710503157592029e-07, | |
| "learning_rate": 2.05243374108053e-05, | |
| "loss": 0.0, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.5459341667622435, | |
| "grad_norm": 6.09717119459674e-07, | |
| "learning_rate": 2.0205784913353725e-05, | |
| "loss": 0.0, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 2.5688725771304046, | |
| "grad_norm": 4.0628717101753864e-07, | |
| "learning_rate": 1.988723241590214e-05, | |
| "loss": 0.0, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 2.591810987498566, | |
| "grad_norm": 5.194257823859516e-07, | |
| "learning_rate": 1.956867991845056e-05, | |
| "loss": 0.0, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 2.6147493978667278, | |
| "grad_norm": 5.950200261395366e-07, | |
| "learning_rate": 1.925012742099898e-05, | |
| "loss": 0.0, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 2.6376878082348894, | |
| "grad_norm": 5.413147619037773e-07, | |
| "learning_rate": 1.89315749235474e-05, | |
| "loss": 0.0, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.660626218603051, | |
| "grad_norm": 4.3072932953691634e-07, | |
| "learning_rate": 1.861302242609582e-05, | |
| "loss": 0.0, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 2.6835646289712125, | |
| "grad_norm": 5.671867597811797e-07, | |
| "learning_rate": 1.829446992864424e-05, | |
| "loss": 0.0, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 2.7065030393393736, | |
| "grad_norm": 4.88942816900817e-07, | |
| "learning_rate": 1.7979102956167177e-05, | |
| "loss": 0.0041, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 2.729441449707535, | |
| "grad_norm": 5.162837055650016e-07, | |
| "learning_rate": 1.7660550458715597e-05, | |
| "loss": 0.0, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 2.752379860075697, | |
| "grad_norm": 5.953765480626316e-07, | |
| "learning_rate": 1.7341997961264017e-05, | |
| "loss": 0.0, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.7753182704438584, | |
| "grad_norm": 2.1004482277930947e-06, | |
| "learning_rate": 1.7023445463812437e-05, | |
| "loss": 0.0, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 2.7982566808120195, | |
| "grad_norm": 4.830233706343279e-07, | |
| "learning_rate": 1.6704892966360856e-05, | |
| "loss": 0.0, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 2.821195091180181, | |
| "grad_norm": 3.652270947895886e-07, | |
| "learning_rate": 1.638634046890928e-05, | |
| "loss": 0.0, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 2.8441335015483427, | |
| "grad_norm": 1.8413612679069047e-06, | |
| "learning_rate": 1.6067787971457696e-05, | |
| "loss": 0.0, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 2.8670719119165042, | |
| "grad_norm": 4.999230327484838e-07, | |
| "learning_rate": 1.5749235474006116e-05, | |
| "loss": 0.0, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 2.890010322284666, | |
| "grad_norm": 5.655538188875653e-07, | |
| "learning_rate": 1.5430682976554536e-05, | |
| "loss": 0.0, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 2.9129487326528274, | |
| "grad_norm": 1.9848837382596685e-06, | |
| "learning_rate": 1.5112130479102956e-05, | |
| "loss": 0.0, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 2.9358871430209885, | |
| "grad_norm": 6.09657490713289e-07, | |
| "learning_rate": 1.4793577981651377e-05, | |
| "loss": 0.0, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 2.95882555338915, | |
| "grad_norm": 5.037222763348836e-07, | |
| "learning_rate": 1.4475025484199797e-05, | |
| "loss": 0.0, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 2.9817639637573117, | |
| "grad_norm": 4.1535062678121903e-07, | |
| "learning_rate": 1.4156472986748217e-05, | |
| "loss": 0.0, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 1.0, | |
| "eval_f1": 1.0, | |
| "eval_loss": 0.0, | |
| "eval_precision": 1.0, | |
| "eval_recall": 1.0, | |
| "eval_runtime": 344.2572, | |
| "eval_samples_per_second": 50.651, | |
| "eval_steps_per_second": 3.166, | |
| "step": 13080 | |
| }, | |
| { | |
| "epoch": 3.004587682073632, | |
| "grad_norm": 6.05821981025656e-07, | |
| "learning_rate": 1.3837920489296639e-05, | |
| "loss": 0.0, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 3.0275260924417937, | |
| "grad_norm": 6.071956022424274e-07, | |
| "learning_rate": 1.3519367991845055e-05, | |
| "loss": 0.0, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 3.0504645028099553, | |
| "grad_norm": 1.934891088239965e-06, | |
| "learning_rate": 1.3200815494393475e-05, | |
| "loss": 0.0, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 3.073402913178117, | |
| "grad_norm": 2.5065089630516013e-06, | |
| "learning_rate": 1.2882262996941897e-05, | |
| "loss": 0.0, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 3.0963413235462784, | |
| "grad_norm": 5.728885753342183e-06, | |
| "learning_rate": 1.2563710499490317e-05, | |
| "loss": 0.0, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 3.1192797339144396, | |
| "grad_norm": 4.320262178225676e-06, | |
| "learning_rate": 1.2245158002038737e-05, | |
| "loss": 0.0, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 3.142218144282601, | |
| "grad_norm": 4.542172007404588e-07, | |
| "learning_rate": 1.1926605504587156e-05, | |
| "loss": 0.0, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 3.1651565546507627, | |
| "grad_norm": 3.681015243728325e-07, | |
| "learning_rate": 1.1608053007135576e-05, | |
| "loss": 0.0, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 3.1880949650189243, | |
| "grad_norm": 4.951008918396838e-07, | |
| "learning_rate": 1.1289500509683996e-05, | |
| "loss": 0.0, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 3.2110333753870854, | |
| "grad_norm": 4.484388966830011e-07, | |
| "learning_rate": 1.0970948012232416e-05, | |
| "loss": 0.0, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 3.233971785755247, | |
| "grad_norm": 5.063902790425345e-07, | |
| "learning_rate": 1.0652395514780836e-05, | |
| "loss": 0.0, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 3.2569101961234086, | |
| "grad_norm": 4.447717003586149e-07, | |
| "learning_rate": 1.0333843017329256e-05, | |
| "loss": 0.0, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 3.27984860649157, | |
| "grad_norm": 7.843581215638551e-07, | |
| "learning_rate": 1.0015290519877676e-05, | |
| "loss": 0.0, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 3.3027870168597318, | |
| "grad_norm": 4.5617230171046685e-07, | |
| "learning_rate": 9.696738022426097e-06, | |
| "loss": 0.0, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 3.3257254272278933, | |
| "grad_norm": 3.457494130998384e-07, | |
| "learning_rate": 9.378185524974515e-06, | |
| "loss": 0.0, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 3.3486638375960545, | |
| "grad_norm": 4.5762243416902493e-07, | |
| "learning_rate": 9.059633027522935e-06, | |
| "loss": 0.0, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 3.371602247964216, | |
| "grad_norm": 5.967251013316854e-07, | |
| "learning_rate": 8.741080530071357e-06, | |
| "loss": 0.0, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 3.3945406583323776, | |
| "grad_norm": 3.912183217380516e-07, | |
| "learning_rate": 8.422528032619777e-06, | |
| "loss": 0.0, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 3.417479068700539, | |
| "grad_norm": 2.0574229893099982e-06, | |
| "learning_rate": 8.103975535168195e-06, | |
| "loss": 0.0, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 3.4404174790687003, | |
| "grad_norm": 5.268063887342578e-06, | |
| "learning_rate": 7.785423037716617e-06, | |
| "loss": 0.0, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 3.463355889436862, | |
| "grad_norm": 4.4349508243612945e-07, | |
| "learning_rate": 7.4668705402650365e-06, | |
| "loss": 0.0, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 3.4862942998050235, | |
| "grad_norm": 3.2844191082403995e-06, | |
| "learning_rate": 7.148318042813456e-06, | |
| "loss": 0.0, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 3.509232710173185, | |
| "grad_norm": 6.628841902056593e-07, | |
| "learning_rate": 6.8297655453618754e-06, | |
| "loss": 0.0, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 3.5321711205413466, | |
| "grad_norm": 4.5607134779857006e-07, | |
| "learning_rate": 6.511213047910296e-06, | |
| "loss": 0.0, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 3.555109530909508, | |
| "grad_norm": 5.205767479310452e-07, | |
| "learning_rate": 6.192660550458716e-06, | |
| "loss": 0.0, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 3.5780479412776693, | |
| "grad_norm": 5.164214940123202e-07, | |
| "learning_rate": 5.874108053007136e-06, | |
| "loss": 0.0, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 3.600986351645831, | |
| "grad_norm": 4.506584048158402e-07, | |
| "learning_rate": 5.558741080530071e-06, | |
| "loss": 0.0076, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 3.6239247620139925, | |
| "grad_norm": 6.092589615036559e-07, | |
| "learning_rate": 5.240188583078492e-06, | |
| "loss": 0.0, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 3.646863172382154, | |
| "grad_norm": 4.135415110795293e-06, | |
| "learning_rate": 4.921636085626911e-06, | |
| "loss": 0.0, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 3.669801582750315, | |
| "grad_norm": 6.201661335580866e-07, | |
| "learning_rate": 4.603083588175332e-06, | |
| "loss": 0.0, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 3.692739993118477, | |
| "grad_norm": 6.843212077001226e-07, | |
| "learning_rate": 4.284531090723752e-06, | |
| "loss": 0.0, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 3.7156784034866384, | |
| "grad_norm": 6.781843921999098e-07, | |
| "learning_rate": 3.9659785932721716e-06, | |
| "loss": 0.0, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 3.7386168138548, | |
| "grad_norm": 5.762826731370296e-07, | |
| "learning_rate": 3.6474260958205915e-06, | |
| "loss": 0.0, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 3.7615552242229615, | |
| "grad_norm": 5.510862592927879e-07, | |
| "learning_rate": 3.3288735983690118e-06, | |
| "loss": 0.0, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 3.784493634591123, | |
| "grad_norm": 6.336876481327636e-07, | |
| "learning_rate": 3.0103211009174312e-06, | |
| "loss": 0.0, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 3.8074320449592842, | |
| "grad_norm": 4.1163076502925833e-07, | |
| "learning_rate": 2.691768603465851e-06, | |
| "loss": 0.0, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 3.830370455327446, | |
| "grad_norm": 5.610125981547753e-07, | |
| "learning_rate": 2.3732161060142714e-06, | |
| "loss": 0.0, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 3.8533088656956074, | |
| "grad_norm": 5.504385853782878e-07, | |
| "learning_rate": 2.0546636085626913e-06, | |
| "loss": 0.0, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 3.8762472760637685, | |
| "grad_norm": 5.19926686592953e-07, | |
| "learning_rate": 1.7361111111111112e-06, | |
| "loss": 0.0, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 3.89918568643193, | |
| "grad_norm": 6.166213779579266e-07, | |
| "learning_rate": 1.417558613659531e-06, | |
| "loss": 0.0, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 3.9221240968000917, | |
| "grad_norm": 5.947557042418339e-07, | |
| "learning_rate": 1.099006116207951e-06, | |
| "loss": 0.0, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 3.9450625071682532, | |
| "grad_norm": 5.760847443525563e-07, | |
| "learning_rate": 7.804536187563711e-07, | |
| "loss": 0.0, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 3.968000917536415, | |
| "grad_norm": 1.4817189821769716e-06, | |
| "learning_rate": 4.6190112130479106e-07, | |
| "loss": 0.0, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 3.9909393279045764, | |
| "grad_norm": 4.664090056394343e-07, | |
| "learning_rate": 1.4334862385321102e-07, | |
| "loss": 0.0, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 1.0, | |
| "eval_f1": 1.0, | |
| "eval_loss": 0.0, | |
| "eval_precision": 1.0, | |
| "eval_recall": 1.0, | |
| "eval_runtime": 344.4309, | |
| "eval_samples_per_second": 50.626, | |
| "eval_steps_per_second": 3.165, | |
| "step": 17440 | |
| } | |
| ], | |
| "logging_steps": 100, | |
| "max_steps": 17440, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |