| { | |
| "best_metric": 0.3763463497161865, | |
| "best_model_checkpoint": "PhoBert-SingleLabel-KMeans/checkpoint-2862", | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 2862, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02620545073375262, | |
| "grad_norm": 3.5584876537323, | |
| "learning_rate": 4.355400696864112e-06, | |
| "loss": 1.7744, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.05241090146750524, | |
| "grad_norm": 3.515974283218384, | |
| "learning_rate": 8.710801393728225e-06, | |
| "loss": 1.6878, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.07861635220125786, | |
| "grad_norm": 3.664238691329956, | |
| "learning_rate": 1.3066202090592336e-05, | |
| "loss": 1.5996, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.10482180293501048, | |
| "grad_norm": 4.6186723709106445, | |
| "learning_rate": 1.742160278745645e-05, | |
| "loss": 1.4775, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1310272536687631, | |
| "grad_norm": 5.333361625671387, | |
| "learning_rate": 2.1777003484320557e-05, | |
| "loss": 1.3713, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.15723270440251572, | |
| "grad_norm": 10.412248611450195, | |
| "learning_rate": 2.6132404181184672e-05, | |
| "loss": 1.0538, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.18343815513626835, | |
| "grad_norm": 3.342909812927246, | |
| "learning_rate": 3.048780487804878e-05, | |
| "loss": 0.9225, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.20964360587002095, | |
| "grad_norm": 5.6379547119140625, | |
| "learning_rate": 3.48432055749129e-05, | |
| "loss": 0.8722, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.2358490566037736, | |
| "grad_norm": 14.341384887695312, | |
| "learning_rate": 3.9198606271777003e-05, | |
| "loss": 0.8813, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.2620545073375262, | |
| "grad_norm": 8.06952953338623, | |
| "learning_rate": 4.3554006968641115e-05, | |
| "loss": 0.8815, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.2882599580712788, | |
| "grad_norm": 16.64231300354004, | |
| "learning_rate": 4.7909407665505226e-05, | |
| "loss": 0.6877, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.31446540880503143, | |
| "grad_norm": 10.036324501037598, | |
| "learning_rate": 4.974757281553398e-05, | |
| "loss": 0.612, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.34067085953878407, | |
| "grad_norm": 28.696367263793945, | |
| "learning_rate": 4.9262135922330097e-05, | |
| "loss": 0.5523, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.3668763102725367, | |
| "grad_norm": 2.96999192237854, | |
| "learning_rate": 4.877669902912622e-05, | |
| "loss": 0.5393, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.39308176100628933, | |
| "grad_norm": 0.7660526037216187, | |
| "learning_rate": 4.829126213592233e-05, | |
| "loss": 0.6644, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.4192872117400419, | |
| "grad_norm": 23.782712936401367, | |
| "learning_rate": 4.780582524271845e-05, | |
| "loss": 0.5915, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.44549266247379454, | |
| "grad_norm": 14.538418769836426, | |
| "learning_rate": 4.732038834951457e-05, | |
| "loss": 0.4766, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.4716981132075472, | |
| "grad_norm": 1.3607491254806519, | |
| "learning_rate": 4.683495145631068e-05, | |
| "loss": 0.592, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.4979035639412998, | |
| "grad_norm": 2.8048417568206787, | |
| "learning_rate": 4.63495145631068e-05, | |
| "loss": 0.6274, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.5241090146750524, | |
| "grad_norm": 294.476806640625, | |
| "learning_rate": 4.586407766990292e-05, | |
| "loss": 0.4636, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.550314465408805, | |
| "grad_norm": 3.9831271171569824, | |
| "learning_rate": 4.5378640776699033e-05, | |
| "loss": 0.6935, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.5765199161425576, | |
| "grad_norm": 8.449265480041504, | |
| "learning_rate": 4.489320388349515e-05, | |
| "loss": 0.5617, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.6027253668763103, | |
| "grad_norm": 0.5198196172714233, | |
| "learning_rate": 4.440776699029127e-05, | |
| "loss": 0.6695, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.6289308176100629, | |
| "grad_norm": 0.511917769908905, | |
| "learning_rate": 4.3922330097087384e-05, | |
| "loss": 0.6177, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.6551362683438156, | |
| "grad_norm": 0.4070751368999481, | |
| "learning_rate": 4.34368932038835e-05, | |
| "loss": 0.4971, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.6813417190775681, | |
| "grad_norm": 24.54846954345703, | |
| "learning_rate": 4.295145631067961e-05, | |
| "loss": 0.3295, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.7075471698113207, | |
| "grad_norm": 26.0633487701416, | |
| "learning_rate": 4.246601941747573e-05, | |
| "loss": 0.4744, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.7337526205450734, | |
| "grad_norm": 4.230802059173584, | |
| "learning_rate": 4.198058252427184e-05, | |
| "loss": 0.4744, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.759958071278826, | |
| "grad_norm": 3.7451541423797607, | |
| "learning_rate": 4.1495145631067964e-05, | |
| "loss": 0.5867, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.7861635220125787, | |
| "grad_norm": 8.754020690917969, | |
| "learning_rate": 4.100970873786408e-05, | |
| "loss": 0.4409, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.8123689727463312, | |
| "grad_norm": 0.14476385712623596, | |
| "learning_rate": 4.052427184466019e-05, | |
| "loss": 0.5824, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.8385744234800838, | |
| "grad_norm": 0.38359129428863525, | |
| "learning_rate": 4.0038834951456314e-05, | |
| "loss": 0.4913, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.8647798742138365, | |
| "grad_norm": 34.098060607910156, | |
| "learning_rate": 3.955339805825243e-05, | |
| "loss": 0.3129, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.8909853249475891, | |
| "grad_norm": 3.0660719871520996, | |
| "learning_rate": 3.906796116504854e-05, | |
| "loss": 0.6103, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.9171907756813418, | |
| "grad_norm": 7.5438385009765625, | |
| "learning_rate": 3.858252427184466e-05, | |
| "loss": 0.495, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.9433962264150944, | |
| "grad_norm": 67.45906829833984, | |
| "learning_rate": 3.809708737864078e-05, | |
| "loss": 0.5724, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.9696016771488469, | |
| "grad_norm": 3.6292409896850586, | |
| "learning_rate": 3.7611650485436894e-05, | |
| "loss": 0.3536, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.9958071278825996, | |
| "grad_norm": 29.442237854003906, | |
| "learning_rate": 3.712621359223301e-05, | |
| "loss": 0.4138, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.9000717188620607, | |
| "eval_f1_macro": 0.8889047304060399, | |
| "eval_f1_micro": 0.9000717188620607, | |
| "eval_f1_weighted": 0.9004594891408181, | |
| "eval_loss": 0.446470707654953, | |
| "eval_precision_macro": 0.8923438114196293, | |
| "eval_precision_micro": 0.9000717188620607, | |
| "eval_precision_weighted": 0.9041587013504236, | |
| "eval_recall_macro": 0.8879970575079902, | |
| "eval_recall_micro": 0.9000717188620607, | |
| "eval_recall_weighted": 0.9000717188620607, | |
| "eval_runtime": 755.376, | |
| "eval_samples_per_second": 5.538, | |
| "eval_steps_per_second": 0.347, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 1.0220125786163523, | |
| "grad_norm": 15.653645515441895, | |
| "learning_rate": 3.664077669902913e-05, | |
| "loss": 0.3952, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.0482180293501049, | |
| "grad_norm": 0.11768877506256104, | |
| "learning_rate": 3.6155339805825244e-05, | |
| "loss": 0.3761, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.0744234800838575, | |
| "grad_norm": 2.6586456298828125, | |
| "learning_rate": 3.566990291262136e-05, | |
| "loss": 0.3241, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.10062893081761, | |
| "grad_norm": 4.2325119972229, | |
| "learning_rate": 3.518446601941748e-05, | |
| "loss": 0.5266, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.1268343815513626, | |
| "grad_norm": 0.2591557502746582, | |
| "learning_rate": 3.4699029126213595e-05, | |
| "loss": 0.3688, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.1530398322851152, | |
| "grad_norm": 7.8002777099609375, | |
| "learning_rate": 3.421359223300971e-05, | |
| "loss": 0.3996, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.179245283018868, | |
| "grad_norm": 7.753486633300781, | |
| "learning_rate": 3.372815533980583e-05, | |
| "loss": 0.5435, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.2054507337526206, | |
| "grad_norm": 4.600539684295654, | |
| "learning_rate": 3.3242718446601945e-05, | |
| "loss": 0.4125, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.2316561844863732, | |
| "grad_norm": 0.507254421710968, | |
| "learning_rate": 3.275728155339806e-05, | |
| "loss": 0.3485, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.2578616352201257, | |
| "grad_norm": 2.2480947971343994, | |
| "learning_rate": 3.227184466019418e-05, | |
| "loss": 0.3006, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.2840670859538785, | |
| "grad_norm": 0.17361511290073395, | |
| "learning_rate": 3.1786407766990296e-05, | |
| "loss": 0.372, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.310272536687631, | |
| "grad_norm": 2.2198197841644287, | |
| "learning_rate": 3.130097087378641e-05, | |
| "loss": 0.4539, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.3364779874213837, | |
| "grad_norm": 2.0315468311309814, | |
| "learning_rate": 3.0815533980582525e-05, | |
| "loss": 0.482, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.3626834381551363, | |
| "grad_norm": 8.348484992980957, | |
| "learning_rate": 3.0330097087378646e-05, | |
| "loss": 0.3263, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.3888888888888888, | |
| "grad_norm": 7.209967136383057, | |
| "learning_rate": 2.9844660194174757e-05, | |
| "loss": 0.332, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.4150943396226414, | |
| "grad_norm": 56.43852615356445, | |
| "learning_rate": 2.9359223300970872e-05, | |
| "loss": 0.5554, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.441299790356394, | |
| "grad_norm": 1.4395267963409424, | |
| "learning_rate": 2.887378640776699e-05, | |
| "loss": 0.4749, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.4675052410901468, | |
| "grad_norm": 70.63890838623047, | |
| "learning_rate": 2.8388349514563105e-05, | |
| "loss": 0.4356, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.4937106918238994, | |
| "grad_norm": 52.49846649169922, | |
| "learning_rate": 2.7902912621359222e-05, | |
| "loss": 0.3524, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.519916142557652, | |
| "grad_norm": 39.48180389404297, | |
| "learning_rate": 2.741747572815534e-05, | |
| "loss": 0.4155, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.5461215932914047, | |
| "grad_norm": 4.6654510498046875, | |
| "learning_rate": 2.6932038834951455e-05, | |
| "loss": 0.3802, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.5723270440251573, | |
| "grad_norm": 2.445298433303833, | |
| "learning_rate": 2.6446601941747573e-05, | |
| "loss": 0.2013, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.59853249475891, | |
| "grad_norm": 0.5410274267196655, | |
| "learning_rate": 2.596116504854369e-05, | |
| "loss": 0.3016, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.6247379454926625, | |
| "grad_norm": 6.175892353057861, | |
| "learning_rate": 2.5475728155339806e-05, | |
| "loss": 0.4059, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.650943396226415, | |
| "grad_norm": 19.350805282592773, | |
| "learning_rate": 2.4990291262135923e-05, | |
| "loss": 0.2649, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.6771488469601676, | |
| "grad_norm": 0.5931205749511719, | |
| "learning_rate": 2.450485436893204e-05, | |
| "loss": 0.2931, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.7033542976939202, | |
| "grad_norm": 20.36564826965332, | |
| "learning_rate": 2.4019417475728156e-05, | |
| "loss": 0.2659, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.7295597484276728, | |
| "grad_norm": 0.1273786723613739, | |
| "learning_rate": 2.3533980582524274e-05, | |
| "loss": 0.2192, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.7557651991614256, | |
| "grad_norm": 39.38661193847656, | |
| "learning_rate": 2.304854368932039e-05, | |
| "loss": 0.4232, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.7819706498951782, | |
| "grad_norm": 35.051902770996094, | |
| "learning_rate": 2.2563106796116507e-05, | |
| "loss": 0.4047, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.808176100628931, | |
| "grad_norm": 10.8402738571167, | |
| "learning_rate": 2.2077669902912624e-05, | |
| "loss": 0.257, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.8343815513626835, | |
| "grad_norm": 18.704975128173828, | |
| "learning_rate": 2.159223300970874e-05, | |
| "loss": 0.3526, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.8605870020964361, | |
| "grad_norm": 0.09916484355926514, | |
| "learning_rate": 2.1106796116504857e-05, | |
| "loss": 0.3903, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.8867924528301887, | |
| "grad_norm": 4.514852046966553, | |
| "learning_rate": 2.062135922330097e-05, | |
| "loss": 0.3345, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.9129979035639413, | |
| "grad_norm": 31.562023162841797, | |
| "learning_rate": 2.0135922330097086e-05, | |
| "loss": 0.5511, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.9392033542976939, | |
| "grad_norm": 13.065159797668457, | |
| "learning_rate": 1.9650485436893204e-05, | |
| "loss": 0.2387, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.9654088050314464, | |
| "grad_norm": 41.689964294433594, | |
| "learning_rate": 1.9165048543689322e-05, | |
| "loss": 0.3024, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.991614255765199, | |
| "grad_norm": 10.776809692382812, | |
| "learning_rate": 1.8679611650485437e-05, | |
| "loss": 0.3133, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9022232847238824, | |
| "eval_f1_macro": 0.8791017264419937, | |
| "eval_f1_micro": 0.9022232847238824, | |
| "eval_f1_weighted": 0.9030036030460146, | |
| "eval_loss": 0.4108516573905945, | |
| "eval_precision_macro": 0.8552051719676482, | |
| "eval_precision_micro": 0.9022232847238824, | |
| "eval_precision_weighted": 0.906337005319111, | |
| "eval_recall_macro": 0.9078028911862802, | |
| "eval_recall_micro": 0.9022232847238824, | |
| "eval_recall_weighted": 0.9022232847238824, | |
| "eval_runtime": 882.9584, | |
| "eval_samples_per_second": 4.737, | |
| "eval_steps_per_second": 0.297, | |
| "step": 1908 | |
| }, | |
| { | |
| "epoch": 2.0178197064989516, | |
| "grad_norm": 7.140912055969238, | |
| "learning_rate": 1.8194174757281555e-05, | |
| "loss": 0.2435, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 2.0440251572327046, | |
| "grad_norm": 13.220586776733398, | |
| "learning_rate": 1.770873786407767e-05, | |
| "loss": 0.2558, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.070230607966457, | |
| "grad_norm": 4.269901275634766, | |
| "learning_rate": 1.7223300970873787e-05, | |
| "loss": 0.2658, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 2.0964360587002098, | |
| "grad_norm": 0.11073017865419388, | |
| "learning_rate": 1.6737864077669905e-05, | |
| "loss": 0.1196, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.1226415094339623, | |
| "grad_norm": 0.16363948583602905, | |
| "learning_rate": 1.625242718446602e-05, | |
| "loss": 0.3133, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 2.148846960167715, | |
| "grad_norm": 0.40826770663261414, | |
| "learning_rate": 1.5766990291262138e-05, | |
| "loss": 0.2075, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.1750524109014675, | |
| "grad_norm": 0.336016982793808, | |
| "learning_rate": 1.5281553398058256e-05, | |
| "loss": 0.2866, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 2.20125786163522, | |
| "grad_norm": 0.0927320122718811, | |
| "learning_rate": 1.479611650485437e-05, | |
| "loss": 0.2213, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.2274633123689727, | |
| "grad_norm": 0.058554600924253464, | |
| "learning_rate": 1.4310679611650485e-05, | |
| "loss": 0.1864, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 2.2536687631027252, | |
| "grad_norm": 0.06633560359477997, | |
| "learning_rate": 1.3825242718446601e-05, | |
| "loss": 0.2221, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.279874213836478, | |
| "grad_norm": 24.724422454833984, | |
| "learning_rate": 1.3339805825242719e-05, | |
| "loss": 0.2335, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 2.3060796645702304, | |
| "grad_norm": 0.0917593464255333, | |
| "learning_rate": 1.2854368932038835e-05, | |
| "loss": 0.2385, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.3322851153039834, | |
| "grad_norm": 0.06267809122800827, | |
| "learning_rate": 1.2368932038834952e-05, | |
| "loss": 0.2185, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 2.358490566037736, | |
| "grad_norm": 0.05579007416963577, | |
| "learning_rate": 1.1883495145631068e-05, | |
| "loss": 0.2324, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 2.3846960167714886, | |
| "grad_norm": 3.492902994155884, | |
| "learning_rate": 1.1398058252427186e-05, | |
| "loss": 0.4154, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 2.410901467505241, | |
| "grad_norm": 0.16507238149642944, | |
| "learning_rate": 1.0912621359223302e-05, | |
| "loss": 0.2711, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.4371069182389937, | |
| "grad_norm": 2.3962414264678955, | |
| "learning_rate": 1.0427184466019418e-05, | |
| "loss": 0.2866, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 2.4633123689727463, | |
| "grad_norm": 2.307986259460449, | |
| "learning_rate": 9.941747572815535e-06, | |
| "loss": 0.4232, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.489517819706499, | |
| "grad_norm": 8.218997955322266, | |
| "learning_rate": 9.45631067961165e-06, | |
| "loss": 0.2528, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 2.5157232704402515, | |
| "grad_norm": 0.10614609718322754, | |
| "learning_rate": 8.970873786407767e-06, | |
| "loss": 0.0532, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.541928721174004, | |
| "grad_norm": 6.59151554107666, | |
| "learning_rate": 8.485436893203883e-06, | |
| "loss": 0.2535, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 2.568134171907757, | |
| "grad_norm": 0.46849825978279114, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.2503, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.5943396226415096, | |
| "grad_norm": 2.4090096950531006, | |
| "learning_rate": 7.514563106796118e-06, | |
| "loss": 0.345, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 2.620545073375262, | |
| "grad_norm": 19.20216941833496, | |
| "learning_rate": 7.029126213592234e-06, | |
| "loss": 0.1694, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.646750524109015, | |
| "grad_norm": 10.444535255432129, | |
| "learning_rate": 6.543689320388349e-06, | |
| "loss": 0.2558, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 2.6729559748427674, | |
| "grad_norm": 0.15079370141029358, | |
| "learning_rate": 6.058252427184466e-06, | |
| "loss": 0.2905, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.69916142557652, | |
| "grad_norm": 0.0647541806101799, | |
| "learning_rate": 5.572815533980583e-06, | |
| "loss": 0.1197, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 2.7253668763102725, | |
| "grad_norm": 50.22942352294922, | |
| "learning_rate": 5.087378640776699e-06, | |
| "loss": 0.159, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.751572327044025, | |
| "grad_norm": 0.06316754221916199, | |
| "learning_rate": 4.601941747572816e-06, | |
| "loss": 0.2032, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 2.7777777777777777, | |
| "grad_norm": 0.06050650402903557, | |
| "learning_rate": 4.116504854368932e-06, | |
| "loss": 0.2446, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.8039832285115303, | |
| "grad_norm": 7.044517993927002, | |
| "learning_rate": 3.6310679611650486e-06, | |
| "loss": 0.2692, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 2.830188679245283, | |
| "grad_norm": 0.36540356278419495, | |
| "learning_rate": 3.1456310679611653e-06, | |
| "loss": 0.3174, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.8563941299790354, | |
| "grad_norm": 1.1024131774902344, | |
| "learning_rate": 2.660194174757282e-06, | |
| "loss": 0.2608, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 2.882599580712788, | |
| "grad_norm": 4.021923065185547, | |
| "learning_rate": 2.1747572815533982e-06, | |
| "loss": 0.0811, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.908805031446541, | |
| "grad_norm": 0.10723225027322769, | |
| "learning_rate": 1.6893203883495145e-06, | |
| "loss": 0.097, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 2.9350104821802936, | |
| "grad_norm": 0.07229924201965332, | |
| "learning_rate": 1.2038834951456312e-06, | |
| "loss": 0.3037, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.961215932914046, | |
| "grad_norm": 38.69488525390625, | |
| "learning_rate": 7.184466019417476e-07, | |
| "loss": 0.2884, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 2.9874213836477987, | |
| "grad_norm": 1.4321260452270508, | |
| "learning_rate": 2.330097087378641e-07, | |
| "loss": 0.3024, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.9194358116184557, | |
| "eval_f1_macro": 0.8947470035372601, | |
| "eval_f1_micro": 0.9194358116184557, | |
| "eval_f1_weighted": 0.9196235681987903, | |
| "eval_loss": 0.3763463497161865, | |
| "eval_precision_macro": 0.8811990938584341, | |
| "eval_precision_micro": 0.9194358116184557, | |
| "eval_precision_weighted": 0.9204284600348198, | |
| "eval_recall_macro": 0.9094848524838759, | |
| "eval_recall_micro": 0.9194358116184557, | |
| "eval_recall_weighted": 0.9194358116184557, | |
| "eval_runtime": 833.3591, | |
| "eval_samples_per_second": 5.019, | |
| "eval_steps_per_second": 0.314, | |
| "step": 2862 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 2862, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.01 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1505115067765248.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |