{ "best_metric": null, "best_model_checkpoint": null, "epoch": 4.984251968503937, "eval_steps": 500, "global_step": 1266, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.03937007874015748, "grad_norm": 6.157460689544678, "learning_rate": 3.125e-05, "loss": 1.4374, "step": 10 }, { "epoch": 0.07874015748031496, "grad_norm": 8.225975036621094, "learning_rate": 6.25e-05, "loss": 0.5746, "step": 20 }, { "epoch": 0.11811023622047244, "grad_norm": 2.298436403274536, "learning_rate": 9.375e-05, "loss": 0.3466, "step": 30 }, { "epoch": 0.15748031496062992, "grad_norm": 1.926019549369812, "learning_rate": 0.000125, "loss": 0.3016, "step": 40 }, { "epoch": 0.1968503937007874, "grad_norm": 1.7295136451721191, "learning_rate": 0.00015625, "loss": 0.2241, "step": 50 }, { "epoch": 0.23622047244094488, "grad_norm": 1.6235859394073486, "learning_rate": 0.0001875, "loss": 0.2474, "step": 60 }, { "epoch": 0.2755905511811024, "grad_norm": 2.1525580883026123, "learning_rate": 0.00019998770426725326, "loss": 0.1819, "step": 70 }, { "epoch": 0.31496062992125984, "grad_norm": 1.2614878416061401, "learning_rate": 0.00019991257462799398, "loss": 0.1753, "step": 80 }, { "epoch": 0.3543307086614173, "grad_norm": 1.5198956727981567, "learning_rate": 0.00019976919756880076, "loss": 0.1734, "step": 90 }, { "epoch": 0.3937007874015748, "grad_norm": 1.3017441034317017, "learning_rate": 0.00019955767102643884, "loss": 0.1558, "step": 100 }, { "epoch": 0.4330708661417323, "grad_norm": 1.3634755611419678, "learning_rate": 0.00019927813948863483, "loss": 0.1396, "step": 110 }, { "epoch": 0.47244094488188976, "grad_norm": 1.0837377309799194, "learning_rate": 0.00019893079389538115, "loss": 0.151, "step": 120 }, { "epoch": 0.5118110236220472, "grad_norm": 0.7570620179176331, "learning_rate": 0.0001985158715085105, "loss": 0.1345, "step": 130 }, { "epoch": 0.5511811023622047, "grad_norm": 1.1616774797439575, "learning_rate": 0.00019803365574962904, "loss": 0.1193, "step": 140 }, { "epoch": 0.5905511811023622, "grad_norm": 1.0171620845794678, "learning_rate": 0.00019748447600651922, "loss": 0.1168, "step": 150 }, { "epoch": 0.6299212598425197, "grad_norm": 1.0031143426895142, "learning_rate": 0.00019686870740814438, "loss": 0.1325, "step": 160 }, { "epoch": 0.6692913385826772, "grad_norm": 0.6090546250343323, "learning_rate": 0.00019618677056840896, "loss": 0.1154, "step": 170 }, { "epoch": 0.7086614173228346, "grad_norm": 1.0556284189224243, "learning_rate": 0.00019543913129884918, "loss": 0.1006, "step": 180 }, { "epoch": 0.7480314960629921, "grad_norm": 0.7367269396781921, "learning_rate": 0.00019462630029045035, "loss": 0.1182, "step": 190 }, { "epoch": 0.7874015748031497, "grad_norm": 17.055007934570312, "learning_rate": 0.00019374883276480865, "loss": 0.1464, "step": 200 }, { "epoch": 0.8267716535433071, "grad_norm": 8.562920570373535, "learning_rate": 0.00019280732809487521, "loss": 0.2103, "step": 210 }, { "epoch": 0.8661417322834646, "grad_norm": 1.041045069694519, "learning_rate": 0.00019180242939554137, "loss": 0.1697, "step": 220 }, { "epoch": 0.905511811023622, "grad_norm": 0.8244247436523438, "learning_rate": 0.00019073482308434544, "loss": 0.1478, "step": 230 }, { "epoch": 0.9448818897637795, "grad_norm": 0.9097272753715515, "learning_rate": 0.00018960523841260053, "loss": 0.1712, "step": 240 }, { "epoch": 0.984251968503937, "grad_norm": 1.7566819190979004, "learning_rate": 0.00018841444696726388, "loss": 0.1788, "step": 250 }, { "epoch": 1.0236220472440944, "grad_norm": 0.7728438973426819, "learning_rate": 0.00018716326214388777, "loss": 0.1519, "step": 260 }, { "epoch": 1.0629921259842519, "grad_norm": 1.4151078462600708, "learning_rate": 0.00018585253859101247, "loss": 0.178, "step": 270 }, { "epoch": 1.1023622047244095, "grad_norm": 0.7322055101394653, "learning_rate": 0.0001844831716263801, "loss": 0.1493, "step": 280 }, { "epoch": 1.141732283464567, "grad_norm": 1.238358974456787, "learning_rate": 0.00018305609662536888, "loss": 0.1366, "step": 290 }, { "epoch": 1.1811023622047245, "grad_norm": 1.0000638961791992, "learning_rate": 0.0001815722883820649, "loss": 0.1483, "step": 300 }, { "epoch": 1.220472440944882, "grad_norm": 0.9049971699714661, "learning_rate": 0.00018003276044340814, "loss": 0.1455, "step": 310 }, { "epoch": 1.2598425196850394, "grad_norm": 1.2286065816879272, "learning_rate": 0.00017843856441686763, "loss": 0.1519, "step": 320 }, { "epoch": 1.2992125984251968, "grad_norm": 0.9089654088020325, "learning_rate": 0.00017679078925211851, "loss": 0.1487, "step": 330 }, { "epoch": 1.3385826771653544, "grad_norm": 1.4565386772155762, "learning_rate": 0.0001750905604972116, "loss": 0.1299, "step": 340 }, { "epoch": 1.3779527559055118, "grad_norm": 1.1536667346954346, "learning_rate": 0.00017333903952974374, "loss": 0.1333, "step": 350 }, { "epoch": 1.4173228346456692, "grad_norm": 1.241052508354187, "learning_rate": 0.0001715374227635541, "loss": 0.1422, "step": 360 }, { "epoch": 1.4566929133858268, "grad_norm": 0.700495719909668, "learning_rate": 0.00016968694083148805, "loss": 0.1241, "step": 370 }, { "epoch": 1.4960629921259843, "grad_norm": 0.8857589960098267, "learning_rate": 0.00016778885774478713, "loss": 0.1312, "step": 380 }, { "epoch": 1.5354330708661417, "grad_norm": 0.8474507927894592, "learning_rate": 0.00016584447002967903, "loss": 0.1459, "step": 390 }, { "epoch": 1.574803149606299, "grad_norm": 1.0673331022262573, "learning_rate": 0.00016385510584175785, "loss": 0.1255, "step": 400 }, { "epoch": 1.6141732283464567, "grad_norm": 1.2963683605194092, "learning_rate": 0.0001618221240587588, "loss": 0.1443, "step": 410 }, { "epoch": 1.6535433070866141, "grad_norm": 0.6176678538322449, "learning_rate": 0.0001597469133523479, "loss": 0.1328, "step": 420 }, { "epoch": 1.6929133858267718, "grad_norm": 0.8610302805900574, "learning_rate": 0.00015763089123956004, "loss": 0.1315, "step": 430 }, { "epoch": 1.7322834645669292, "grad_norm": 0.8733415603637695, "learning_rate": 0.0001554755031145336, "loss": 0.1354, "step": 440 }, { "epoch": 1.7716535433070866, "grad_norm": 0.6525992155075073, "learning_rate": 0.00015328222126120315, "loss": 0.1306, "step": 450 }, { "epoch": 1.811023622047244, "grad_norm": 0.6734092831611633, "learning_rate": 0.00015105254384762453, "loss": 0.1272, "step": 460 }, { "epoch": 1.8503937007874016, "grad_norm": 0.9078968167304993, "learning_rate": 0.0001487879939026189, "loss": 0.1084, "step": 470 }, { "epoch": 1.889763779527559, "grad_norm": 0.6580955982208252, "learning_rate": 0.0001464901182754356, "loss": 0.1116, "step": 480 }, { "epoch": 1.9291338582677167, "grad_norm": 0.8749574422836304, "learning_rate": 0.00014416048657914352, "loss": 0.1345, "step": 490 }, { "epoch": 1.968503937007874, "grad_norm": 0.8025640845298767, "learning_rate": 0.00014180069011847347, "loss": 0.096, "step": 500 }, { "epoch": 2.0078740157480315, "grad_norm": 0.5700274705886841, "learning_rate": 0.00013941234080284328, "loss": 0.1201, "step": 510 }, { "epoch": 2.047244094488189, "grad_norm": 1.1415060758590698, "learning_rate": 0.00013699707004530883, "loss": 0.1206, "step": 520 }, { "epoch": 2.0866141732283463, "grad_norm": 0.8418262600898743, "learning_rate": 0.0001345565276481924, "loss": 0.1257, "step": 530 }, { "epoch": 2.1259842519685037, "grad_norm": 0.8720746040344238, "learning_rate": 0.00013209238067614989, "loss": 0.1169, "step": 540 }, { "epoch": 2.1653543307086616, "grad_norm": 1.166407585144043, "learning_rate": 0.0001296063123174467, "loss": 0.1327, "step": 550 }, { "epoch": 2.204724409448819, "grad_norm": 0.8546102046966553, "learning_rate": 0.00012710002073421995, "loss": 0.1287, "step": 560 }, { "epoch": 2.2440944881889764, "grad_norm": 0.6846320033073425, "learning_rate": 0.00012457521790251272, "loss": 0.1143, "step": 570 }, { "epoch": 2.283464566929134, "grad_norm": 1.0204697847366333, "learning_rate": 0.00012203362844287196, "loss": 0.1261, "step": 580 }, { "epoch": 2.322834645669291, "grad_norm": 0.7173842191696167, "learning_rate": 0.00011947698844230972, "loss": 0.1002, "step": 590 }, { "epoch": 2.362204724409449, "grad_norm": 0.7630149722099304, "learning_rate": 0.0001169070442684318, "loss": 0.1127, "step": 600 }, { "epoch": 2.4015748031496065, "grad_norm": 0.819685697555542, "learning_rate": 0.00011432555137654386, "loss": 0.1149, "step": 610 }, { "epoch": 2.440944881889764, "grad_norm": 0.5875028371810913, "learning_rate": 0.00011173427311055041, "loss": 0.1099, "step": 620 }, { "epoch": 2.4803149606299213, "grad_norm": 0.5276161432266235, "learning_rate": 0.00010913497949846483, "loss": 0.0935, "step": 630 }, { "epoch": 2.5196850393700787, "grad_norm": 0.8867918848991394, "learning_rate": 0.00010652944604335413, "loss": 0.1024, "step": 640 }, { "epoch": 2.559055118110236, "grad_norm": 0.7481799125671387, "learning_rate": 0.00010391945251054331, "loss": 0.1001, "step": 650 }, { "epoch": 2.5984251968503935, "grad_norm": 1.034694790840149, "learning_rate": 0.00010130678171190893, "loss": 0.1066, "step": 660 }, { "epoch": 2.637795275590551, "grad_norm": 0.744408369064331, "learning_rate": 9.869321828809109e-05, "loss": 0.0911, "step": 670 }, { "epoch": 2.677165354330709, "grad_norm": 0.6600897908210754, "learning_rate": 9.60805474894567e-05, "loss": 0.0957, "step": 680 }, { "epoch": 2.716535433070866, "grad_norm": 0.8174748420715332, "learning_rate": 9.347055395664589e-05, "loss": 0.1009, "step": 690 }, { "epoch": 2.7559055118110236, "grad_norm": 1.1754448413848877, "learning_rate": 9.08650205015352e-05, "loss": 0.0994, "step": 700 }, { "epoch": 2.795275590551181, "grad_norm": 0.6818444728851318, "learning_rate": 8.826572688944966e-05, "loss": 0.0978, "step": 710 }, { "epoch": 2.8346456692913384, "grad_norm": 0.7322255969047546, "learning_rate": 8.56744486234562e-05, "loss": 0.0917, "step": 720 }, { "epoch": 2.8740157480314963, "grad_norm": 0.4844493567943573, "learning_rate": 8.309295573156825e-05, "loss": 0.1054, "step": 730 }, { "epoch": 2.9133858267716537, "grad_norm": 0.4901849031448364, "learning_rate": 8.052301155769029e-05, "loss": 0.0933, "step": 740 }, { "epoch": 2.952755905511811, "grad_norm": 0.6136536002159119, "learning_rate": 7.796637155712802e-05, "loss": 0.0898, "step": 750 }, { "epoch": 2.9921259842519685, "grad_norm": 0.5571744441986084, "learning_rate": 7.542478209748727e-05, "loss": 0.0932, "step": 760 }, { "epoch": 3.031496062992126, "grad_norm": 0.6652244925498962, "learning_rate": 7.289997926578003e-05, "loss": 0.0894, "step": 770 }, { "epoch": 3.0708661417322833, "grad_norm": 0.4437718391418457, "learning_rate": 7.039368768255335e-05, "loss": 0.0862, "step": 780 }, { "epoch": 3.1102362204724407, "grad_norm": 0.6460803747177124, "learning_rate": 6.790761932385014e-05, "loss": 0.0909, "step": 790 }, { "epoch": 3.1496062992125986, "grad_norm": 0.7623637914657593, "learning_rate": 6.544347235180761e-05, "loss": 0.1056, "step": 800 }, { "epoch": 3.188976377952756, "grad_norm": 0.45258432626724243, "learning_rate": 6.300292995469119e-05, "loss": 0.0828, "step": 810 }, { "epoch": 3.2283464566929134, "grad_norm": 0.8195613026618958, "learning_rate": 6.058765919715674e-05, "loss": 0.1034, "step": 820 }, { "epoch": 3.267716535433071, "grad_norm": 0.5008795857429504, "learning_rate": 5.819930988152658e-05, "loss": 0.0997, "step": 830 }, { "epoch": 3.3070866141732282, "grad_norm": 0.593920886516571, "learning_rate": 5.5839513420856494e-05, "loss": 0.085, "step": 840 }, { "epoch": 3.3464566929133857, "grad_norm": 0.5025109052658081, "learning_rate": 5.3509881724564426e-05, "loss": 0.0909, "step": 850 }, { "epoch": 3.3858267716535435, "grad_norm": 0.4997495412826538, "learning_rate": 5.121200609738113e-05, "loss": 0.0968, "step": 860 }, { "epoch": 3.425196850393701, "grad_norm": 1.036964774131775, "learning_rate": 4.894745615237551e-05, "loss": 0.1025, "step": 870 }, { "epoch": 3.4645669291338583, "grad_norm": 0.4552832841873169, "learning_rate": 4.6717778738796883e-05, "loss": 0.0839, "step": 880 }, { "epoch": 3.5039370078740157, "grad_norm": 0.5048990845680237, "learning_rate": 4.4524496885466436e-05, "loss": 0.0919, "step": 890 }, { "epoch": 3.543307086614173, "grad_norm": 0.4812853932380676, "learning_rate": 4.236910876043999e-05, "loss": 0.0945, "step": 900 }, { "epoch": 3.5826771653543306, "grad_norm": 0.6418349146842957, "learning_rate": 4.02530866476521e-05, "loss": 0.0893, "step": 910 }, { "epoch": 3.622047244094488, "grad_norm": 0.9326638579368591, "learning_rate": 3.817787594124122e-05, "loss": 0.1007, "step": 920 }, { "epoch": 3.661417322834646, "grad_norm": 0.9185812473297119, "learning_rate": 3.6144894158242184e-05, "loss": 0.0912, "step": 930 }, { "epoch": 3.7007874015748032, "grad_norm": 0.6465328335762024, "learning_rate": 3.4155529970320986e-05, "loss": 0.078, "step": 940 }, { "epoch": 3.7401574803149606, "grad_norm": 0.5492585897445679, "learning_rate": 3.22111422552129e-05, "loss": 0.078, "step": 950 }, { "epoch": 3.779527559055118, "grad_norm": 0.638911783695221, "learning_rate": 3.031305916851196e-05, "loss": 0.083, "step": 960 }, { "epoch": 3.8188976377952755, "grad_norm": 0.47551003098487854, "learning_rate": 2.846257723644592e-05, "loss": 0.0847, "step": 970 }, { "epoch": 3.8582677165354333, "grad_norm": 0.8244979381561279, "learning_rate": 2.666096047025628e-05, "loss": 0.0871, "step": 980 }, { "epoch": 3.8976377952755907, "grad_norm": 0.2906242609024048, "learning_rate": 2.4909439502788423e-05, "loss": 0.0757, "step": 990 }, { "epoch": 3.937007874015748, "grad_norm": 0.6135310530662537, "learning_rate": 2.3209210747881515e-05, "loss": 0.0812, "step": 1000 }, { "epoch": 3.9763779527559056, "grad_norm": 0.5390735864639282, "learning_rate": 2.1561435583132395e-05, "loss": 0.0852, "step": 1010 }, { "epoch": 4.015748031496063, "grad_norm": 0.3508215546607971, "learning_rate": 1.996723955659189e-05, "loss": 0.0962, "step": 1020 }, { "epoch": 4.05511811023622, "grad_norm": 0.5338996648788452, "learning_rate": 1.842771161793513e-05, "loss": 0.0863, "step": 1030 }, { "epoch": 4.094488188976378, "grad_norm": 0.45414066314697266, "learning_rate": 1.6943903374631143e-05, "loss": 0.0822, "step": 1040 }, { "epoch": 4.133858267716535, "grad_norm": 0.8013415336608887, "learning_rate": 1.5516828373619918e-05, "loss": 0.0717, "step": 1050 }, { "epoch": 4.173228346456693, "grad_norm": 0.8064348697662354, "learning_rate": 1.4147461408987561e-05, "loss": 0.0945, "step": 1060 }, { "epoch": 4.21259842519685, "grad_norm": 0.4887343645095825, "learning_rate": 1.2836737856112246e-05, "loss": 0.0851, "step": 1070 }, { "epoch": 4.251968503937007, "grad_norm": 0.39578166604042053, "learning_rate": 1.158555303273614e-05, "loss": 0.0715, "step": 1080 }, { "epoch": 4.291338582677166, "grad_norm": 0.5196364521980286, "learning_rate": 1.0394761587399493e-05, "loss": 0.079, "step": 1090 }, { "epoch": 4.330708661417323, "grad_norm": 0.3454650044441223, "learning_rate": 9.265176915654595e-06, "loss": 0.0778, "step": 1100 }, { "epoch": 4.3700787401574805, "grad_norm": 0.44942712783813477, "learning_rate": 8.19757060445865e-06, "loss": 0.0725, "step": 1110 }, { "epoch": 4.409448818897638, "grad_norm": 0.5285565257072449, "learning_rate": 7.192671905124792e-06, "loss": 0.0685, "step": 1120 }, { "epoch": 4.448818897637795, "grad_norm": 0.5806540250778198, "learning_rate": 6.251167235191336e-06, "loss": 0.0759, "step": 1130 }, { "epoch": 4.488188976377953, "grad_norm": 0.592225968837738, "learning_rate": 5.373699709549673e-06, "loss": 0.0806, "step": 1140 }, { "epoch": 4.52755905511811, "grad_norm": 0.7170082330703735, "learning_rate": 4.560868701150844e-06, "loss": 0.0735, "step": 1150 }, { "epoch": 4.566929133858268, "grad_norm": 0.4995642304420471, "learning_rate": 3.8132294315910256e-06, "loss": 0.073, "step": 1160 }, { "epoch": 4.606299212598425, "grad_norm": 0.6979544162750244, "learning_rate": 3.131292591855628e-06, "loss": 0.0685, "step": 1170 }, { "epoch": 4.645669291338582, "grad_norm": 0.8452284932136536, "learning_rate": 2.515523993480806e-06, "loss": 0.0827, "step": 1180 }, { "epoch": 4.68503937007874, "grad_norm": 0.5433554649353027, "learning_rate": 1.966344250370977e-06, "loss": 0.0839, "step": 1190 }, { "epoch": 4.724409448818898, "grad_norm": 0.6554785370826721, "learning_rate": 1.4841284914895116e-06, "loss": 0.0906, "step": 1200 }, { "epoch": 4.7637795275590555, "grad_norm": 0.7909722924232483, "learning_rate": 1.0692061046188696e-06, "loss": 0.0926, "step": 1210 }, { "epoch": 4.803149606299213, "grad_norm": 0.6115571856498718, "learning_rate": 7.2186051136518e-07, "loss": 0.0949, "step": 1220 }, { "epoch": 4.84251968503937, "grad_norm": 0.39121922850608826, "learning_rate": 4.4232897356116795e-07, "loss": 0.0866, "step": 1230 }, { "epoch": 4.881889763779528, "grad_norm": 0.506991446018219, "learning_rate": 2.308024311992618e-07, "loss": 0.0713, "step": 1240 }, { "epoch": 4.921259842519685, "grad_norm": 0.5368971228599548, "learning_rate": 8.742537200603362e-08, "loss": 0.0811, "step": 1250 }, { "epoch": 4.960629921259843, "grad_norm": 0.7474204897880554, "learning_rate": 1.2295732746747668e-08, "loss": 0.0754, "step": 1260 }, { "epoch": 4.984251968503937, "step": 1266, "total_flos": 4.27570483010112e+16, "train_loss": 0.12990388043495527, "train_runtime": 571.8261, "train_samples_per_second": 35.423, "train_steps_per_second": 2.214 } ], "logging_steps": 10, "max_steps": 1266, "num_input_tokens_seen": 0, "num_train_epochs": 5, "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": 4.27570483010112e+16, "train_batch_size": 16, "trial_name": null, "trial_params": null }