| { | |
| "best_metric": 0.021347366273403168, | |
| "best_model_checkpoint": "autotrain-78oj2-jlrbv/checkpoint-3639", | |
| "epoch": 2.99938182567484, | |
| "eval_steps": 500, | |
| "global_step": 3639, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0206058108386565, | |
| "grad_norm": 5.651961326599121, | |
| "learning_rate": 3.021978021978022e-06, | |
| "loss": 0.9893, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.041211621677313, | |
| "grad_norm": 3.856781005859375, | |
| "learning_rate": 6.456043956043957e-06, | |
| "loss": 0.8203, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0618174325159695, | |
| "grad_norm": 3.5703439712524414, | |
| "learning_rate": 9.89010989010989e-06, | |
| "loss": 0.608, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.082423243354626, | |
| "grad_norm": 6.108644485473633, | |
| "learning_rate": 1.3324175824175824e-05, | |
| "loss": 0.6063, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.10302905419328251, | |
| "grad_norm": 3.141345739364624, | |
| "learning_rate": 1.6758241758241757e-05, | |
| "loss": 0.5388, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.123634865031939, | |
| "grad_norm": 4.041093349456787, | |
| "learning_rate": 2.0192307692307694e-05, | |
| "loss": 0.4164, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.1442406758705955, | |
| "grad_norm": 5.099579334259033, | |
| "learning_rate": 2.348901098901099e-05, | |
| "loss": 0.3599, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.164846486709252, | |
| "grad_norm": 2.3562018871307373, | |
| "learning_rate": 2.6785714285714288e-05, | |
| "loss": 0.3026, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.1854522975479085, | |
| "grad_norm": 1.3748841285705566, | |
| "learning_rate": 3.021978021978022e-05, | |
| "loss": 0.1953, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.20605810838656502, | |
| "grad_norm": 5.422688961029053, | |
| "learning_rate": 3.365384615384616e-05, | |
| "loss": 0.2107, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.2266639192252215, | |
| "grad_norm": 3.1931090354919434, | |
| "learning_rate": 3.708791208791209e-05, | |
| "loss": 0.1905, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.247269730063878, | |
| "grad_norm": 7.671177387237549, | |
| "learning_rate": 4.052197802197803e-05, | |
| "loss": 0.1454, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.2678755409025345, | |
| "grad_norm": 5.701260089874268, | |
| "learning_rate": 4.3956043956043955e-05, | |
| "loss": 0.1193, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.288481351741191, | |
| "grad_norm": 4.128372669219971, | |
| "learning_rate": 4.739010989010989e-05, | |
| "loss": 0.1242, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.30908716257984753, | |
| "grad_norm": 2.761212110519409, | |
| "learning_rate": 4.990839694656489e-05, | |
| "loss": 0.137, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.329692973418504, | |
| "grad_norm": 5.561426162719727, | |
| "learning_rate": 4.952671755725191e-05, | |
| "loss": 0.1326, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.35029878425716054, | |
| "grad_norm": 0.6887926459312439, | |
| "learning_rate": 4.9145038167938934e-05, | |
| "loss": 0.1149, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.370904595095817, | |
| "grad_norm": 3.1560258865356445, | |
| "learning_rate": 4.8763358778625956e-05, | |
| "loss": 0.1091, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.3915104059344735, | |
| "grad_norm": 3.2499308586120605, | |
| "learning_rate": 4.838167938931298e-05, | |
| "loss": 0.1192, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.41211621677313004, | |
| "grad_norm": 1.6396406888961792, | |
| "learning_rate": 4.8e-05, | |
| "loss": 0.098, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.4327220276117865, | |
| "grad_norm": 1.4734318256378174, | |
| "learning_rate": 4.7618320610687024e-05, | |
| "loss": 0.1064, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.453327838450443, | |
| "grad_norm": 2.2481157779693604, | |
| "learning_rate": 4.723664122137405e-05, | |
| "loss": 0.1099, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.47393364928909953, | |
| "grad_norm": 2.839649200439453, | |
| "learning_rate": 4.6854961832061075e-05, | |
| "loss": 0.0999, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.494539460127756, | |
| "grad_norm": 0.30553099513053894, | |
| "learning_rate": 4.64732824427481e-05, | |
| "loss": 0.0992, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.5151452709664125, | |
| "grad_norm": 0.589301347732544, | |
| "learning_rate": 4.609160305343512e-05, | |
| "loss": 0.0735, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.535751081805069, | |
| "grad_norm": 2.1147384643554688, | |
| "learning_rate": 4.570992366412214e-05, | |
| "loss": 0.0901, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.5563568926437256, | |
| "grad_norm": 0.805992841720581, | |
| "learning_rate": 4.5328244274809165e-05, | |
| "loss": 0.0747, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.576962703482382, | |
| "grad_norm": 1.168734073638916, | |
| "learning_rate": 4.494656488549619e-05, | |
| "loss": 0.067, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.5975685143210385, | |
| "grad_norm": 5.0603790283203125, | |
| "learning_rate": 4.45648854961832e-05, | |
| "loss": 0.0677, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.6181743251596951, | |
| "grad_norm": 5.00893497467041, | |
| "learning_rate": 4.4183206106870225e-05, | |
| "loss": 0.0538, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.6387801359983515, | |
| "grad_norm": 0.6863914132118225, | |
| "learning_rate": 4.3801526717557255e-05, | |
| "loss": 0.0596, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.659385946837008, | |
| "grad_norm": 3.027167797088623, | |
| "learning_rate": 4.341984732824428e-05, | |
| "loss": 0.0476, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.6799917576756646, | |
| "grad_norm": 3.794942617416382, | |
| "learning_rate": 4.30381679389313e-05, | |
| "loss": 0.0922, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.7005975685143211, | |
| "grad_norm": 1.6739851236343384, | |
| "learning_rate": 4.265648854961832e-05, | |
| "loss": 0.0601, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.7212033793529775, | |
| "grad_norm": 4.5667195320129395, | |
| "learning_rate": 4.2274809160305344e-05, | |
| "loss": 0.0606, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.741809190191634, | |
| "grad_norm": 1.4475433826446533, | |
| "learning_rate": 4.189312977099237e-05, | |
| "loss": 0.0488, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.7624150010302906, | |
| "grad_norm": 3.672123908996582, | |
| "learning_rate": 4.151145038167939e-05, | |
| "loss": 0.066, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.783020811868947, | |
| "grad_norm": 0.7629732489585876, | |
| "learning_rate": 4.112977099236641e-05, | |
| "loss": 0.0533, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.8036266227076035, | |
| "grad_norm": 3.216966390609741, | |
| "learning_rate": 4.0748091603053434e-05, | |
| "loss": 0.0524, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.8242324335462601, | |
| "grad_norm": 1.6424092054367065, | |
| "learning_rate": 4.036641221374046e-05, | |
| "loss": 0.0449, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.8448382443849165, | |
| "grad_norm": 3.3341548442840576, | |
| "learning_rate": 3.9984732824427486e-05, | |
| "loss": 0.0595, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.865444055223573, | |
| "grad_norm": 0.1322302669286728, | |
| "learning_rate": 3.960305343511451e-05, | |
| "loss": 0.0717, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.8860498660622296, | |
| "grad_norm": 5.5867414474487305, | |
| "learning_rate": 3.922137404580153e-05, | |
| "loss": 0.0457, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.906655676900886, | |
| "grad_norm": 1.4986252784729004, | |
| "learning_rate": 3.883969465648855e-05, | |
| "loss": 0.0395, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.9272614877395425, | |
| "grad_norm": 0.7124060392379761, | |
| "learning_rate": 3.8458015267175575e-05, | |
| "loss": 0.0463, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.9478672985781991, | |
| "grad_norm": 1.1682511568069458, | |
| "learning_rate": 3.80763358778626e-05, | |
| "loss": 0.0396, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.9684731094168556, | |
| "grad_norm": 4.690053939819336, | |
| "learning_rate": 3.770992366412214e-05, | |
| "loss": 0.0428, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.989078920255512, | |
| "grad_norm": 1.3264271020889282, | |
| "learning_rate": 3.732824427480916e-05, | |
| "loss": 0.06, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.9997939418916134, | |
| "eval_accuracy": 0.9912411767736617, | |
| "eval_f1_macro": 0.9877568131380815, | |
| "eval_f1_micro": 0.9912411767736617, | |
| "eval_f1_weighted": 0.9911910099696459, | |
| "eval_loss": 0.03924937546253204, | |
| "eval_precision_macro": 0.9939377252247231, | |
| "eval_precision_micro": 0.9912411767736617, | |
| "eval_precision_weighted": 0.9912837913018231, | |
| "eval_recall_macro": 0.9818041226442024, | |
| "eval_recall_micro": 0.9912411767736617, | |
| "eval_recall_weighted": 0.9912411767736617, | |
| "eval_runtime": 18.3314, | |
| "eval_samples_per_second": 1058.782, | |
| "eval_steps_per_second": 33.112, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 1.0096847310941686, | |
| "grad_norm": 0.46682170033454895, | |
| "learning_rate": 3.694656488549619e-05, | |
| "loss": 0.0372, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.030290541932825, | |
| "grad_norm": 3.3791985511779785, | |
| "learning_rate": 3.656488549618321e-05, | |
| "loss": 0.0439, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.0508963527714816, | |
| "grad_norm": 3.6968090534210205, | |
| "learning_rate": 3.6183206106870234e-05, | |
| "loss": 0.0382, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.071502163610138, | |
| "grad_norm": 2.349797487258911, | |
| "learning_rate": 3.580152671755726e-05, | |
| "loss": 0.0327, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.0921079744487945, | |
| "grad_norm": 1.1067183017730713, | |
| "learning_rate": 3.54351145038168e-05, | |
| "loss": 0.0315, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.1127137852874511, | |
| "grad_norm": 0.032840073108673096, | |
| "learning_rate": 3.505343511450382e-05, | |
| "loss": 0.0316, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.1333195961261076, | |
| "grad_norm": 0.028824035078287125, | |
| "learning_rate": 3.467175572519084e-05, | |
| "loss": 0.0213, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.153925406964764, | |
| "grad_norm": 0.5763944983482361, | |
| "learning_rate": 3.4290076335877864e-05, | |
| "loss": 0.026, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.1745312178034206, | |
| "grad_norm": 0.08884817361831665, | |
| "learning_rate": 3.3908396946564886e-05, | |
| "loss": 0.025, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.195137028642077, | |
| "grad_norm": 0.41536378860473633, | |
| "learning_rate": 3.352671755725191e-05, | |
| "loss": 0.0346, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.2157428394807335, | |
| "grad_norm": 0.3112834393978119, | |
| "learning_rate": 3.314503816793894e-05, | |
| "loss": 0.0264, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.2363486503193901, | |
| "grad_norm": 0.48356932401657104, | |
| "learning_rate": 3.276335877862596e-05, | |
| "loss": 0.0188, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.2569544611580465, | |
| "grad_norm": 3.8283050060272217, | |
| "learning_rate": 3.238167938931298e-05, | |
| "loss": 0.038, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.2775602719967032, | |
| "grad_norm": 0.06744606792926788, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 0.0299, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.2981660828353596, | |
| "grad_norm": 5.736328125, | |
| "learning_rate": 3.161832061068703e-05, | |
| "loss": 0.0243, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.318771893674016, | |
| "grad_norm": 0.07336083799600601, | |
| "learning_rate": 3.123664122137404e-05, | |
| "loss": 0.0268, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.3393777045126725, | |
| "grad_norm": 0.10824728012084961, | |
| "learning_rate": 3.0854961832061066e-05, | |
| "loss": 0.0277, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.359983515351329, | |
| "grad_norm": 2.3970558643341064, | |
| "learning_rate": 3.047328244274809e-05, | |
| "loss": 0.0247, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.3805893261899855, | |
| "grad_norm": 0.14330969750881195, | |
| "learning_rate": 3.0091603053435114e-05, | |
| "loss": 0.0292, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.4011951370286422, | |
| "grad_norm": 0.5449164509773254, | |
| "learning_rate": 2.970992366412214e-05, | |
| "loss": 0.026, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.4218009478672986, | |
| "grad_norm": 0.13645227253437042, | |
| "learning_rate": 2.9328244274809162e-05, | |
| "loss": 0.0337, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.442406758705955, | |
| "grad_norm": 0.2142195850610733, | |
| "learning_rate": 2.8946564885496185e-05, | |
| "loss": 0.0148, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.4630125695446115, | |
| "grad_norm": 0.053218819200992584, | |
| "learning_rate": 2.8564885496183207e-05, | |
| "loss": 0.0168, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.483618380383268, | |
| "grad_norm": 6.44566011428833, | |
| "learning_rate": 2.818320610687023e-05, | |
| "loss": 0.0177, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.5042241912219247, | |
| "grad_norm": 0.035187333822250366, | |
| "learning_rate": 2.7801526717557252e-05, | |
| "loss": 0.0277, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.5248300020605812, | |
| "grad_norm": 0.042209599167108536, | |
| "learning_rate": 2.7419847328244274e-05, | |
| "loss": 0.0349, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.5454358128992376, | |
| "grad_norm": 4.258108139038086, | |
| "learning_rate": 2.7038167938931297e-05, | |
| "loss": 0.0232, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.566041623737894, | |
| "grad_norm": 0.2930113673210144, | |
| "learning_rate": 2.665648854961832e-05, | |
| "loss": 0.0225, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.5866474345765504, | |
| "grad_norm": 4.373762607574463, | |
| "learning_rate": 2.627480916030535e-05, | |
| "loss": 0.0216, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.607253245415207, | |
| "grad_norm": 1.3070122003555298, | |
| "learning_rate": 2.589312977099237e-05, | |
| "loss": 0.0118, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.6278590562538637, | |
| "grad_norm": 1.7738819122314453, | |
| "learning_rate": 2.5511450381679393e-05, | |
| "loss": 0.024, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.6484648670925202, | |
| "grad_norm": 0.027883194386959076, | |
| "learning_rate": 2.5129770992366412e-05, | |
| "loss": 0.0186, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.6690706779311766, | |
| "grad_norm": 0.030870651826262474, | |
| "learning_rate": 2.4748091603053435e-05, | |
| "loss": 0.0174, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.689676488769833, | |
| "grad_norm": 0.11152543127536774, | |
| "learning_rate": 2.4366412213740457e-05, | |
| "loss": 0.0225, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.7102822996084894, | |
| "grad_norm": 0.09863735735416412, | |
| "learning_rate": 2.3984732824427483e-05, | |
| "loss": 0.0154, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.730888110447146, | |
| "grad_norm": 0.03743008151650429, | |
| "learning_rate": 2.3603053435114505e-05, | |
| "loss": 0.0145, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.7514939212858027, | |
| "grad_norm": 0.037079595029354095, | |
| "learning_rate": 2.3221374045801528e-05, | |
| "loss": 0.0228, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.7720997321244591, | |
| "grad_norm": 0.02248048223555088, | |
| "learning_rate": 2.283969465648855e-05, | |
| "loss": 0.0302, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.7927055429631156, | |
| "grad_norm": 0.5228779911994934, | |
| "learning_rate": 2.2458015267175573e-05, | |
| "loss": 0.0304, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.813311353801772, | |
| "grad_norm": 0.07086365669965744, | |
| "learning_rate": 2.20763358778626e-05, | |
| "loss": 0.0224, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.8339171646404286, | |
| "grad_norm": 0.02493666112422943, | |
| "learning_rate": 2.169465648854962e-05, | |
| "loss": 0.0135, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.854522975479085, | |
| "grad_norm": 1.703371286392212, | |
| "learning_rate": 2.1312977099236643e-05, | |
| "loss": 0.0213, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.8751287863177417, | |
| "grad_norm": 2.37048602104187, | |
| "learning_rate": 2.0931297709923666e-05, | |
| "loss": 0.0245, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.8957345971563981, | |
| "grad_norm": 1.9404609203338623, | |
| "learning_rate": 2.0549618320610688e-05, | |
| "loss": 0.0249, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.9163404079950546, | |
| "grad_norm": 2.3614635467529297, | |
| "learning_rate": 2.016793893129771e-05, | |
| "loss": 0.0155, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.936946218833711, | |
| "grad_norm": 0.010869844816625118, | |
| "learning_rate": 1.9786259541984733e-05, | |
| "loss": 0.011, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.9575520296723676, | |
| "grad_norm": 3.8267769813537598, | |
| "learning_rate": 1.9404580152671755e-05, | |
| "loss": 0.0194, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.978157840511024, | |
| "grad_norm": 1.0191699266433716, | |
| "learning_rate": 1.902290076335878e-05, | |
| "loss": 0.024, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.9987636513496807, | |
| "grad_norm": 2.944854497909546, | |
| "learning_rate": 1.8641221374045803e-05, | |
| "loss": 0.0157, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.9995878837832268, | |
| "eval_accuracy": 0.9954144984285641, | |
| "eval_f1_macro": 0.9934889743246283, | |
| "eval_f1_micro": 0.9954144984285641, | |
| "eval_f1_weighted": 0.9954058869538126, | |
| "eval_loss": 0.02238026075065136, | |
| "eval_precision_macro": 0.9952598218686433, | |
| "eval_precision_micro": 0.9954144984285641, | |
| "eval_precision_weighted": 0.9954125216770092, | |
| "eval_recall_macro": 0.9917415518445051, | |
| "eval_recall_micro": 0.9954144984285641, | |
| "eval_recall_weighted": 0.9954144984285641, | |
| "eval_runtime": 18.0784, | |
| "eval_samples_per_second": 1073.599, | |
| "eval_steps_per_second": 33.576, | |
| "step": 2426 | |
| }, | |
| { | |
| "epoch": 2.019369462188337, | |
| "grad_norm": 0.0203185323625803, | |
| "learning_rate": 1.8259541984732826e-05, | |
| "loss": 0.0163, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.0399752730269936, | |
| "grad_norm": 0.050282154232263565, | |
| "learning_rate": 1.787786259541985e-05, | |
| "loss": 0.023, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 2.06058108386565, | |
| "grad_norm": 1.1421475410461426, | |
| "learning_rate": 1.749618320610687e-05, | |
| "loss": 0.0076, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.0811868947043064, | |
| "grad_norm": 1.4307163953781128, | |
| "learning_rate": 1.712977099236641e-05, | |
| "loss": 0.0237, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 2.1017927055429633, | |
| "grad_norm": 1.7668225765228271, | |
| "learning_rate": 1.6748091603053437e-05, | |
| "loss": 0.0097, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.1223985163816197, | |
| "grad_norm": 0.43096867203712463, | |
| "learning_rate": 1.636641221374046e-05, | |
| "loss": 0.0164, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 2.143004327220276, | |
| "grad_norm": 0.009999148547649384, | |
| "learning_rate": 1.598473282442748e-05, | |
| "loss": 0.005, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.1636101380589325, | |
| "grad_norm": 0.015757331624627113, | |
| "learning_rate": 1.5603053435114504e-05, | |
| "loss": 0.0053, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 2.184215948897589, | |
| "grad_norm": 0.014865142293274403, | |
| "learning_rate": 1.5221374045801528e-05, | |
| "loss": 0.0085, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.2048217597362454, | |
| "grad_norm": 0.25190040469169617, | |
| "learning_rate": 1.483969465648855e-05, | |
| "loss": 0.0108, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 2.2254275705749023, | |
| "grad_norm": 0.048391640186309814, | |
| "learning_rate": 1.4458015267175573e-05, | |
| "loss": 0.0015, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.2460333814135587, | |
| "grad_norm": 0.03182575851678848, | |
| "learning_rate": 1.4076335877862595e-05, | |
| "loss": 0.006, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 2.266639192252215, | |
| "grad_norm": 0.03702850267291069, | |
| "learning_rate": 1.3694656488549621e-05, | |
| "loss": 0.0042, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.2872450030908715, | |
| "grad_norm": 0.010030820034444332, | |
| "learning_rate": 1.3312977099236642e-05, | |
| "loss": 0.0091, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 2.307850813929528, | |
| "grad_norm": 0.42647820711135864, | |
| "learning_rate": 1.2931297709923664e-05, | |
| "loss": 0.019, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.328456624768185, | |
| "grad_norm": 0.012364137917757034, | |
| "learning_rate": 1.2549618320610687e-05, | |
| "loss": 0.0052, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 2.3490624356068412, | |
| "grad_norm": 0.8119286298751831, | |
| "learning_rate": 1.216793893129771e-05, | |
| "loss": 0.0058, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.3696682464454977, | |
| "grad_norm": 0.03282434120774269, | |
| "learning_rate": 1.1786259541984733e-05, | |
| "loss": 0.0114, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 2.390274057284154, | |
| "grad_norm": 0.03597622364759445, | |
| "learning_rate": 1.1404580152671757e-05, | |
| "loss": 0.0122, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.4108798681228105, | |
| "grad_norm": 0.17078496515750885, | |
| "learning_rate": 1.1022900763358778e-05, | |
| "loss": 0.0111, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 2.431485678961467, | |
| "grad_norm": 4.740245342254639, | |
| "learning_rate": 1.0641221374045802e-05, | |
| "loss": 0.0082, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 2.452091489800124, | |
| "grad_norm": 0.05649268627166748, | |
| "learning_rate": 1.0259541984732825e-05, | |
| "loss": 0.0118, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 2.4726973006387802, | |
| "grad_norm": 0.04967363923788071, | |
| "learning_rate": 9.877862595419849e-06, | |
| "loss": 0.0183, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.4933031114774367, | |
| "grad_norm": 0.007585471961647272, | |
| "learning_rate": 9.496183206106871e-06, | |
| "loss": 0.0044, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 2.513908922316093, | |
| "grad_norm": 0.020672131329774857, | |
| "learning_rate": 9.114503816793894e-06, | |
| "loss": 0.0043, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 2.5345147331547495, | |
| "grad_norm": 0.005783187225461006, | |
| "learning_rate": 8.732824427480916e-06, | |
| "loss": 0.0049, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 2.5551205439934064, | |
| "grad_norm": 0.0074052768759429455, | |
| "learning_rate": 8.351145038167938e-06, | |
| "loss": 0.0016, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 2.575726354832063, | |
| "grad_norm": 0.03194005414843559, | |
| "learning_rate": 7.969465648854962e-06, | |
| "loss": 0.0039, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 2.5963321656707192, | |
| "grad_norm": 0.0073048705235123634, | |
| "learning_rate": 7.587786259541985e-06, | |
| "loss": 0.0042, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 2.6169379765093757, | |
| "grad_norm": 3.5076446533203125, | |
| "learning_rate": 7.206106870229008e-06, | |
| "loss": 0.0052, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 2.637543787348032, | |
| "grad_norm": 0.004831337369978428, | |
| "learning_rate": 6.8244274809160305e-06, | |
| "loss": 0.0122, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.6581495981866885, | |
| "grad_norm": 0.06588663160800934, | |
| "learning_rate": 6.442748091603054e-06, | |
| "loss": 0.0126, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 2.678755409025345, | |
| "grad_norm": 0.11881962418556213, | |
| "learning_rate": 6.061068702290076e-06, | |
| "loss": 0.0109, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 2.699361219864002, | |
| "grad_norm": 0.03703455999493599, | |
| "learning_rate": 5.6793893129770995e-06, | |
| "loss": 0.0063, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 2.719967030702658, | |
| "grad_norm": 0.015223699621856213, | |
| "learning_rate": 5.297709923664122e-06, | |
| "loss": 0.0059, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.7405728415413146, | |
| "grad_norm": 2.5166127681732178, | |
| "learning_rate": 4.916030534351145e-06, | |
| "loss": 0.0084, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 2.761178652379971, | |
| "grad_norm": 0.007914497517049313, | |
| "learning_rate": 4.5343511450381684e-06, | |
| "loss": 0.0038, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 2.7817844632186275, | |
| "grad_norm": 0.015695059671998024, | |
| "learning_rate": 4.152671755725191e-06, | |
| "loss": 0.012, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 2.8023902740572844, | |
| "grad_norm": 0.012181021273136139, | |
| "learning_rate": 3.770992366412214e-06, | |
| "loss": 0.0205, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.822996084895941, | |
| "grad_norm": 2.1182310581207275, | |
| "learning_rate": 3.389312977099237e-06, | |
| "loss": 0.0109, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 2.843601895734597, | |
| "grad_norm": 0.04122795909643173, | |
| "learning_rate": 3.0076335877862594e-06, | |
| "loss": 0.0112, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 2.8642077065732536, | |
| "grad_norm": 0.8893581032752991, | |
| "learning_rate": 2.6259541984732826e-06, | |
| "loss": 0.0062, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 2.88481351741191, | |
| "grad_norm": 11.825778007507324, | |
| "learning_rate": 2.2442748091603055e-06, | |
| "loss": 0.0104, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.9054193282505665, | |
| "grad_norm": 0.019555876031517982, | |
| "learning_rate": 1.8625954198473283e-06, | |
| "loss": 0.0095, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 2.926025139089223, | |
| "grad_norm": 0.054326750338077545, | |
| "learning_rate": 1.4809160305343512e-06, | |
| "loss": 0.0117, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 2.9466309499278798, | |
| "grad_norm": 0.01002267561852932, | |
| "learning_rate": 1.099236641221374e-06, | |
| "loss": 0.011, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 2.967236760766536, | |
| "grad_norm": 0.13304032385349274, | |
| "learning_rate": 7.175572519083969e-07, | |
| "loss": 0.0057, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.9878425716051926, | |
| "grad_norm": 0.027920976281166077, | |
| "learning_rate": 3.3587786259541984e-07, | |
| "loss": 0.0033, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 2.99938182567484, | |
| "eval_accuracy": 0.9955175434076975, | |
| "eval_f1_macro": 0.9936121048770471, | |
| "eval_f1_micro": 0.9955175434076975, | |
| "eval_f1_weighted": 0.9955108152465474, | |
| "eval_loss": 0.021347366273403168, | |
| "eval_precision_macro": 0.9946094746601329, | |
| "eval_precision_micro": 0.9955175434076975, | |
| "eval_precision_weighted": 0.9955142956474508, | |
| "eval_recall_macro": 0.9926308276207328, | |
| "eval_recall_micro": 0.9955175434076975, | |
| "eval_recall_weighted": 0.9955175434076975, | |
| "eval_runtime": 18.1429, | |
| "eval_samples_per_second": 1069.787, | |
| "eval_steps_per_second": 33.457, | |
| "step": 3639 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 3639, | |
| "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": 1 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.5317936953288704e+16, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |