{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.9995119967466448, "global_step": 6146, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.999967339434433e-05, "loss": 1.8952, "step": 10 }, { "epoch": 0.01, "learning_rate": 4.999869358591101e-05, "loss": 1.7279, "step": 20 }, { "epoch": 0.01, "learning_rate": 4.999706060030093e-05, "loss": 1.6717, "step": 30 }, { "epoch": 0.01, "learning_rate": 4.9994774480181473e-05, "loss": 1.64, "step": 40 }, { "epoch": 0.02, "learning_rate": 4.999183528528541e-05, "loss": 1.5903, "step": 50 }, { "epoch": 0.02, "learning_rate": 4.998824309240937e-05, "loss": 1.5936, "step": 60 }, { "epoch": 0.02, "learning_rate": 4.998399799541179e-05, "loss": 1.5633, "step": 70 }, { "epoch": 0.03, "learning_rate": 4.9979100105210475e-05, "loss": 1.5244, "step": 80 }, { "epoch": 0.03, "learning_rate": 4.997354954977972e-05, "loss": 1.5517, "step": 90 }, { "epoch": 0.03, "learning_rate": 4.996734647414696e-05, "loss": 1.5226, "step": 100 }, { "epoch": 0.04, "learning_rate": 4.996049104038895e-05, "loss": 1.5107, "step": 110 }, { "epoch": 0.04, "learning_rate": 4.9952983427627566e-05, "loss": 1.5111, "step": 120 }, { "epoch": 0.04, "learning_rate": 4.9944823832025115e-05, "loss": 1.5107, "step": 130 }, { "epoch": 0.05, "learning_rate": 4.99360124667792e-05, "loss": 1.5035, "step": 140 }, { "epoch": 0.05, "learning_rate": 4.9926549562117164e-05, "loss": 1.4973, "step": 150 }, { "epoch": 0.05, "learning_rate": 4.991643536529006e-05, "loss": 1.4968, "step": 160 }, { "epoch": 0.06, "learning_rate": 4.99056701405662e-05, "loss": 1.479, "step": 170 }, { "epoch": 0.06, "learning_rate": 4.989425416922423e-05, "loss": 1.4879, "step": 180 }, { "epoch": 0.06, "learning_rate": 4.988218774954583e-05, "loss": 1.4943, "step": 190 }, { "epoch": 0.07, "learning_rate": 4.986947119680788e-05, "loss": 1.4952, "step": 200 }, { "epoch": 0.07, "learning_rate": 4.985610484327421e-05, "loss": 1.4588, "step": 210 }, { "epoch": 0.07, "learning_rate": 4.9842089038186954e-05, "loss": 1.4869, "step": 220 }, { "epoch": 0.07, "learning_rate": 4.9827424147757405e-05, "loss": 1.4762, "step": 230 }, { "epoch": 0.08, "learning_rate": 4.981211055515648e-05, "loss": 1.4909, "step": 240 }, { "epoch": 0.08, "learning_rate": 4.9796148660504624e-05, "loss": 1.4659, "step": 250 }, { "epoch": 0.08, "learning_rate": 4.9779538880861455e-05, "loss": 1.4683, "step": 260 }, { "epoch": 0.09, "learning_rate": 4.976228165021482e-05, "loss": 1.4581, "step": 270 }, { "epoch": 0.09, "learning_rate": 4.9744377419469426e-05, "loss": 1.4481, "step": 280 }, { "epoch": 0.09, "learning_rate": 4.972582665643514e-05, "loss": 1.4571, "step": 290 }, { "epoch": 0.1, "learning_rate": 4.970662984581468e-05, "loss": 1.4683, "step": 300 }, { "epoch": 0.1, "learning_rate": 4.9686787489191e-05, "loss": 1.4528, "step": 310 }, { "epoch": 0.1, "learning_rate": 4.966630010501417e-05, "loss": 1.4548, "step": 320 }, { "epoch": 0.11, "learning_rate": 4.964516822858784e-05, "loss": 1.4487, "step": 330 }, { "epoch": 0.11, "learning_rate": 4.962339241205523e-05, "loss": 1.4468, "step": 340 }, { "epoch": 0.11, "learning_rate": 4.960097322438474e-05, "loss": 1.468, "step": 350 }, { "epoch": 0.12, "learning_rate": 4.957791125135503e-05, "loss": 1.4601, "step": 360 }, { "epoch": 0.12, "learning_rate": 4.955420709553979e-05, "loss": 1.4604, "step": 370 }, { "epoch": 0.12, "learning_rate": 4.9529861376291895e-05, "loss": 1.4528, "step": 380 }, { "epoch": 0.13, "learning_rate": 4.9504874729727344e-05, "loss": 1.4478, "step": 390 }, { "epoch": 0.13, "learning_rate": 4.947924780870853e-05, "loss": 1.4697, "step": 400 }, { "epoch": 0.13, "learning_rate": 4.945298128282724e-05, "loss": 1.45, "step": 410 }, { "epoch": 0.14, "learning_rate": 4.942607583838715e-05, "loss": 1.4368, "step": 420 }, { "epoch": 0.14, "learning_rate": 4.939853217838588e-05, "loss": 1.4279, "step": 430 }, { "epoch": 0.14, "learning_rate": 4.937035102249665e-05, "loss": 1.4384, "step": 440 }, { "epoch": 0.15, "learning_rate": 4.934153310704945e-05, "loss": 1.4543, "step": 450 }, { "epoch": 0.15, "learning_rate": 4.931207918501181e-05, "loss": 1.4581, "step": 460 }, { "epoch": 0.15, "learning_rate": 4.9281990025969135e-05, "loss": 1.4503, "step": 470 }, { "epoch": 0.16, "learning_rate": 4.925126641610458e-05, "loss": 1.4333, "step": 480 }, { "epoch": 0.16, "learning_rate": 4.921990915817852e-05, "loss": 1.4506, "step": 490 }, { "epoch": 0.16, "learning_rate": 4.91879190715076e-05, "loss": 1.4368, "step": 500 }, { "epoch": 0.17, "learning_rate": 4.915529699194326e-05, "loss": 1.4428, "step": 510 }, { "epoch": 0.17, "learning_rate": 4.9122043771849957e-05, "loss": 1.4521, "step": 520 }, { "epoch": 0.17, "learning_rate": 4.9088160280082875e-05, "loss": 1.4296, "step": 530 }, { "epoch": 0.18, "learning_rate": 4.905364740196521e-05, "loss": 1.4456, "step": 540 }, { "epoch": 0.18, "learning_rate": 4.901850603926507e-05, "loss": 1.425, "step": 550 }, { "epoch": 0.18, "learning_rate": 4.898273711017187e-05, "loss": 1.4604, "step": 560 }, { "epoch": 0.19, "learning_rate": 4.894634154927238e-05, "loss": 1.4328, "step": 570 }, { "epoch": 0.19, "learning_rate": 4.890932030752627e-05, "loss": 1.4232, "step": 580 }, { "epoch": 0.19, "learning_rate": 4.887167435224131e-05, "loss": 1.4344, "step": 590 }, { "epoch": 0.2, "learning_rate": 4.8833404667048045e-05, "loss": 1.4459, "step": 600 }, { "epoch": 0.2, "learning_rate": 4.8794512251874125e-05, "loss": 1.4308, "step": 610 }, { "epoch": 0.2, "learning_rate": 4.875499812291817e-05, "loss": 1.442, "step": 620 }, { "epoch": 0.2, "learning_rate": 4.871486331262322e-05, "loss": 1.4297, "step": 630 }, { "epoch": 0.21, "learning_rate": 4.867410886964976e-05, "loss": 1.442, "step": 640 }, { "epoch": 0.21, "learning_rate": 4.863273585884831e-05, "loss": 1.436, "step": 650 }, { "epoch": 0.21, "learning_rate": 4.8590745361231626e-05, "loss": 1.4153, "step": 660 }, { "epoch": 0.22, "learning_rate": 4.8548138473946424e-05, "loss": 1.4344, "step": 670 }, { "epoch": 0.22, "learning_rate": 4.850491631024473e-05, "loss": 1.4287, "step": 680 }, { "epoch": 0.22, "learning_rate": 4.846107999945479e-05, "loss": 1.4492, "step": 690 }, { "epoch": 0.23, "learning_rate": 4.841663068695157e-05, "loss": 1.4636, "step": 700 }, { "epoch": 0.23, "learning_rate": 4.837156953412682e-05, "loss": 1.4465, "step": 710 }, { "epoch": 0.23, "learning_rate": 4.8325897718358725e-05, "loss": 1.4127, "step": 720 }, { "epoch": 0.24, "learning_rate": 4.8279616432981154e-05, "loss": 1.4162, "step": 730 }, { "epoch": 0.24, "learning_rate": 4.823272688725248e-05, "loss": 1.4313, "step": 740 }, { "epoch": 0.24, "learning_rate": 4.818523030632396e-05, "loss": 1.4191, "step": 750 }, { "epoch": 0.25, "learning_rate": 4.8137127931207745e-05, "loss": 1.4212, "step": 760 }, { "epoch": 0.25, "learning_rate": 4.808842101874447e-05, "loss": 1.4179, "step": 770 }, { "epoch": 0.25, "learning_rate": 4.803911084157037e-05, "loss": 1.4353, "step": 780 }, { "epoch": 0.26, "learning_rate": 4.798919868808406e-05, "loss": 1.4294, "step": 790 }, { "epoch": 0.26, "learning_rate": 4.793868586241289e-05, "loss": 1.437, "step": 800 }, { "epoch": 0.26, "learning_rate": 4.7887573684378815e-05, "loss": 1.4265, "step": 810 }, { "epoch": 0.27, "learning_rate": 4.783586348946393e-05, "loss": 1.4121, "step": 820 }, { "epoch": 0.27, "learning_rate": 4.778355662877564e-05, "loss": 1.414, "step": 830 }, { "epoch": 0.27, "learning_rate": 4.773065446901123e-05, "loss": 1.4205, "step": 840 }, { "epoch": 0.28, "learning_rate": 4.76771583924223e-05, "loss": 1.4289, "step": 850 }, { "epoch": 0.28, "learning_rate": 4.7623069796778506e-05, "loss": 1.4088, "step": 860 }, { "epoch": 0.28, "learning_rate": 4.756839009533118e-05, "loss": 1.4206, "step": 870 }, { "epoch": 0.29, "learning_rate": 4.7513120716776294e-05, "loss": 1.4196, "step": 880 }, { "epoch": 0.29, "learning_rate": 4.745726310521717e-05, "loss": 1.4409, "step": 890 }, { "epoch": 0.29, "learning_rate": 4.740081872012676e-05, "loss": 1.4221, "step": 900 }, { "epoch": 0.3, "learning_rate": 4.734378903630949e-05, "loss": 1.4097, "step": 910 }, { "epoch": 0.3, "learning_rate": 4.728617554386276e-05, "loss": 1.4256, "step": 920 }, { "epoch": 0.3, "learning_rate": 4.722797974813795e-05, "loss": 1.4139, "step": 930 }, { "epoch": 0.31, "learning_rate": 4.716920316970115e-05, "loss": 1.4258, "step": 940 }, { "epoch": 0.31, "learning_rate": 4.710984734429339e-05, "loss": 1.4362, "step": 950 }, { "epoch": 0.31, "learning_rate": 4.7049913822790526e-05, "loss": 1.4329, "step": 960 }, { "epoch": 0.32, "learning_rate": 4.6989404171162745e-05, "loss": 1.4058, "step": 970 }, { "epoch": 0.32, "learning_rate": 4.692831997043359e-05, "loss": 1.4412, "step": 980 }, { "epoch": 0.32, "learning_rate": 4.6866662816638694e-05, "loss": 1.4148, "step": 990 }, { "epoch": 0.33, "learning_rate": 4.680443432078407e-05, "loss": 1.4253, "step": 1000 }, { "epoch": 0.33, "learning_rate": 4.6741636108804016e-05, "loss": 1.425, "step": 1010 }, { "epoch": 0.33, "learning_rate": 4.6678269821518626e-05, "loss": 1.4159, "step": 1020 }, { "epoch": 0.34, "learning_rate": 4.661433711459092e-05, "loss": 1.3995, "step": 1030 }, { "epoch": 0.34, "learning_rate": 4.654983965848361e-05, "loss": 1.4065, "step": 1040 }, { "epoch": 0.34, "learning_rate": 4.648477913841538e-05, "loss": 1.4023, "step": 1050 }, { "epoch": 0.34, "learning_rate": 4.641915725431696e-05, "loss": 1.39, "step": 1060 }, { "epoch": 0.35, "learning_rate": 4.6352975720786617e-05, "loss": 1.4028, "step": 1070 }, { "epoch": 0.35, "learning_rate": 4.6286236267045415e-05, "loss": 1.419, "step": 1080 }, { "epoch": 0.35, "learning_rate": 4.621894063689198e-05, "loss": 1.4118, "step": 1090 }, { "epoch": 0.36, "learning_rate": 4.6151090588657e-05, "loss": 1.4, "step": 1100 }, { "epoch": 0.36, "learning_rate": 4.608268789515723e-05, "loss": 1.4156, "step": 1110 }, { "epoch": 0.36, "learning_rate": 4.60137343436492e-05, "loss": 1.4186, "step": 1120 }, { "epoch": 0.37, "learning_rate": 4.5944231735782486e-05, "loss": 1.4164, "step": 1130 }, { "epoch": 0.37, "learning_rate": 4.587418188755268e-05, "loss": 1.4155, "step": 1140 }, { "epoch": 0.37, "learning_rate": 4.580358662925392e-05, "loss": 1.4117, "step": 1150 }, { "epoch": 0.38, "learning_rate": 4.5732447805431044e-05, "loss": 1.399, "step": 1160 }, { "epoch": 0.38, "learning_rate": 4.566076727483144e-05, "loss": 1.4182, "step": 1170 }, { "epoch": 0.38, "learning_rate": 4.558854691035644e-05, "loss": 1.4179, "step": 1180 }, { "epoch": 0.39, "learning_rate": 4.55157885990124e-05, "loss": 1.4144, "step": 1190 }, { "epoch": 0.39, "learning_rate": 4.544249424186139e-05, "loss": 1.4162, "step": 1200 }, { "epoch": 0.39, "learning_rate": 4.5368665753971554e-05, "loss": 1.4031, "step": 1210 }, { "epoch": 0.4, "learning_rate": 4.5294305064367014e-05, "loss": 1.4086, "step": 1220 }, { "epoch": 0.4, "learning_rate": 4.521941411597752e-05, "loss": 1.4069, "step": 1230 }, { "epoch": 0.4, "learning_rate": 4.5143994865587655e-05, "loss": 1.4146, "step": 1240 }, { "epoch": 0.41, "learning_rate": 4.506804928378571e-05, "loss": 1.4365, "step": 1250 }, { "epoch": 0.41, "learning_rate": 4.499157935491222e-05, "loss": 1.4197, "step": 1260 }, { "epoch": 0.41, "learning_rate": 4.491458707700807e-05, "loss": 1.3969, "step": 1270 }, { "epoch": 0.42, "learning_rate": 4.483707446176235e-05, "loss": 1.4089, "step": 1280 }, { "epoch": 0.42, "learning_rate": 4.475904353445973e-05, "loss": 1.4079, "step": 1290 }, { "epoch": 0.42, "learning_rate": 4.46804963339276e-05, "loss": 1.414, "step": 1300 }, { "epoch": 0.43, "learning_rate": 4.460143491248273e-05, "loss": 1.3883, "step": 1310 }, { "epoch": 0.43, "learning_rate": 4.452186133587773e-05, "loss": 1.407, "step": 1320 }, { "epoch": 0.43, "learning_rate": 4.4441777683247e-05, "loss": 1.4026, "step": 1330 }, { "epoch": 0.44, "learning_rate": 4.436118604705246e-05, "loss": 1.396, "step": 1340 }, { "epoch": 0.44, "learning_rate": 4.428008853302884e-05, "loss": 1.4037, "step": 1350 }, { "epoch": 0.44, "learning_rate": 4.419848726012868e-05, "loss": 1.3998, "step": 1360 }, { "epoch": 0.45, "learning_rate": 4.411638436046696e-05, "loss": 1.4212, "step": 1370 }, { "epoch": 0.45, "learning_rate": 4.403378197926539e-05, "loss": 1.3948, "step": 1380 }, { "epoch": 0.45, "learning_rate": 4.395068227479636e-05, "loss": 1.3882, "step": 1390 }, { "epoch": 0.46, "learning_rate": 4.3867087418326543e-05, "loss": 1.4053, "step": 1400 }, { "epoch": 0.46, "learning_rate": 4.378299959406018e-05, "loss": 1.414, "step": 1410 }, { "epoch": 0.46, "learning_rate": 4.369842099908199e-05, "loss": 1.4386, "step": 1420 }, { "epoch": 0.47, "learning_rate": 4.361335384329975e-05, "loss": 1.4114, "step": 1430 }, { "epoch": 0.47, "learning_rate": 4.352780034938663e-05, "loss": 1.4185, "step": 1440 }, { "epoch": 0.47, "learning_rate": 4.3441762752722995e-05, "loss": 1.4209, "step": 1450 }, { "epoch": 0.47, "learning_rate": 4.335524330133811e-05, "loss": 1.4071, "step": 1460 }, { "epoch": 0.48, "learning_rate": 4.326824425585135e-05, "loss": 1.3998, "step": 1470 }, { "epoch": 0.48, "learning_rate": 4.3180767889413144e-05, "loss": 1.3928, "step": 1480 }, { "epoch": 0.48, "learning_rate": 4.309281648764556e-05, "loss": 1.406, "step": 1490 }, { "epoch": 0.49, "learning_rate": 4.300439234858262e-05, "loss": 1.3937, "step": 1500 }, { "epoch": 0.49, "learning_rate": 4.291549778261025e-05, "loss": 1.3943, "step": 1510 }, { "epoch": 0.49, "learning_rate": 4.2826135112405875e-05, "loss": 1.393, "step": 1520 }, { "epoch": 0.5, "learning_rate": 4.273630667287779e-05, "loss": 1.3956, "step": 1530 }, { "epoch": 0.5, "learning_rate": 4.264601481110409e-05, "loss": 1.3972, "step": 1540 }, { "epoch": 0.5, "learning_rate": 4.2555261886271405e-05, "loss": 1.4232, "step": 1550 }, { "epoch": 0.51, "learning_rate": 4.2464050269613205e-05, "loss": 1.4231, "step": 1560 }, { "epoch": 0.51, "learning_rate": 4.237238234434789e-05, "loss": 1.3968, "step": 1570 }, { "epoch": 0.51, "learning_rate": 4.2280260505616477e-05, "loss": 1.3858, "step": 1580 }, { "epoch": 0.52, "learning_rate": 4.218768716042005e-05, "loss": 1.3965, "step": 1590 }, { "epoch": 0.52, "learning_rate": 4.209466472755687e-05, "loss": 1.4159, "step": 1600 }, { "epoch": 0.52, "learning_rate": 4.200119563755915e-05, "loss": 1.4228, "step": 1610 }, { "epoch": 0.53, "learning_rate": 4.190728233262955e-05, "loss": 1.391, "step": 1620 }, { "epoch": 0.53, "learning_rate": 4.18129272665774e-05, "loss": 1.401, "step": 1630 }, { "epoch": 0.53, "learning_rate": 4.171813290475455e-05, "loss": 1.4074, "step": 1640 }, { "epoch": 0.54, "learning_rate": 4.162290172399099e-05, "loss": 1.4002, "step": 1650 }, { "epoch": 0.54, "learning_rate": 4.152723621253008e-05, "loss": 1.3952, "step": 1660 }, { "epoch": 0.54, "learning_rate": 4.1431138869963614e-05, "loss": 1.3901, "step": 1670 }, { "epoch": 0.55, "learning_rate": 4.133461220716642e-05, "loss": 1.3889, "step": 1680 }, { "epoch": 0.55, "learning_rate": 4.1237658746230814e-05, "loss": 1.4069, "step": 1690 }, { "epoch": 0.55, "learning_rate": 4.11402810204007e-05, "loss": 1.3851, "step": 1700 }, { "epoch": 0.56, "learning_rate": 4.1042481574005355e-05, "loss": 1.4234, "step": 1710 }, { "epoch": 0.56, "learning_rate": 4.094426296239296e-05, "loss": 1.4101, "step": 1720 }, { "epoch": 0.56, "learning_rate": 4.0845627751863843e-05, "loss": 1.4122, "step": 1730 }, { "epoch": 0.57, "learning_rate": 4.0746578519603404e-05, "loss": 1.407, "step": 1740 }, { "epoch": 0.57, "learning_rate": 4.064711785361482e-05, "loss": 1.3968, "step": 1750 }, { "epoch": 0.57, "learning_rate": 4.0547248352651354e-05, "loss": 1.3994, "step": 1760 }, { "epoch": 0.58, "learning_rate": 4.044697262614852e-05, "loss": 1.3806, "step": 1770 }, { "epoch": 0.58, "learning_rate": 4.0346293294155865e-05, "loss": 1.4131, "step": 1780 }, { "epoch": 0.58, "learning_rate": 4.0245212987268533e-05, "loss": 1.3817, "step": 1790 }, { "epoch": 0.59, "learning_rate": 4.0143734346558515e-05, "loss": 1.3835, "step": 1800 }, { "epoch": 0.59, "learning_rate": 4.004186002350565e-05, "loss": 1.399, "step": 1810 }, { "epoch": 0.59, "learning_rate": 3.993959267992835e-05, "loss": 1.4, "step": 1820 }, { "epoch": 0.6, "learning_rate": 3.983693498791403e-05, "loss": 1.3731, "step": 1830 }, { "epoch": 0.6, "learning_rate": 3.9733889629749324e-05, "loss": 1.4064, "step": 1840 }, { "epoch": 0.6, "learning_rate": 3.963045929784996e-05, "loss": 1.4022, "step": 1850 }, { "epoch": 0.61, "learning_rate": 3.952664669469046e-05, "loss": 1.4068, "step": 1860 }, { "epoch": 0.61, "learning_rate": 3.942245453273348e-05, "loss": 1.4103, "step": 1870 }, { "epoch": 0.61, "learning_rate": 3.931788553435898e-05, "loss": 1.4096, "step": 1880 }, { "epoch": 0.61, "learning_rate": 3.9212942431793056e-05, "loss": 1.3996, "step": 1890 }, { "epoch": 0.62, "learning_rate": 3.9107627967036576e-05, "loss": 1.4124, "step": 1900 }, { "epoch": 0.62, "learning_rate": 3.900194489179352e-05, "loss": 1.4065, "step": 1910 }, { "epoch": 0.62, "learning_rate": 3.88958959673991e-05, "loss": 1.3858, "step": 1920 }, { "epoch": 0.63, "learning_rate": 3.8789483964747595e-05, "loss": 1.3778, "step": 1930 }, { "epoch": 0.63, "learning_rate": 3.868271166421996e-05, "loss": 1.4049, "step": 1940 }, { "epoch": 0.63, "learning_rate": 3.857558185561117e-05, "loss": 1.3862, "step": 1950 }, { "epoch": 0.64, "learning_rate": 3.846809733805732e-05, "loss": 1.371, "step": 1960 }, { "epoch": 0.64, "learning_rate": 3.8360260919962535e-05, "loss": 1.3884, "step": 1970 }, { "epoch": 0.64, "learning_rate": 3.8252075418925525e-05, "loss": 1.3748, "step": 1980 }, { "epoch": 0.65, "learning_rate": 3.814354366166603e-05, "loss": 1.3801, "step": 1990 }, { "epoch": 0.65, "learning_rate": 3.8034668483950895e-05, "loss": 1.3609, "step": 2000 }, { "epoch": 0.65, "learning_rate": 3.7925452730520025e-05, "loss": 1.3885, "step": 2010 }, { "epoch": 0.66, "learning_rate": 3.7815899255012044e-05, "loss": 1.3876, "step": 2020 }, { "epoch": 0.66, "learning_rate": 3.770601091988972e-05, "loss": 1.3982, "step": 2030 }, { "epoch": 0.66, "learning_rate": 3.7595790596365204e-05, "loss": 1.3781, "step": 2040 }, { "epoch": 0.67, "learning_rate": 3.748524116432498e-05, "loss": 1.3844, "step": 2050 }, { "epoch": 0.67, "learning_rate": 3.73743655122546e-05, "loss": 1.3946, "step": 2060 }, { "epoch": 0.67, "learning_rate": 3.726316653716331e-05, "loss": 1.3988, "step": 2070 }, { "epoch": 0.68, "learning_rate": 3.715164714450821e-05, "loss": 1.4002, "step": 2080 }, { "epoch": 0.68, "learning_rate": 3.703981024811846e-05, "loss": 1.4044, "step": 2090 }, { "epoch": 0.68, "learning_rate": 3.692765877011909e-05, "loss": 1.3789, "step": 2100 }, { "epoch": 0.69, "learning_rate": 3.681519564085466e-05, "loss": 1.4105, "step": 2110 }, { "epoch": 0.69, "learning_rate": 3.67024237988127e-05, "loss": 1.4161, "step": 2120 }, { "epoch": 0.69, "learning_rate": 3.658934619054692e-05, "loss": 1.3894, "step": 2130 }, { "epoch": 0.7, "learning_rate": 3.647596577060022e-05, "loss": 1.3934, "step": 2140 }, { "epoch": 0.7, "learning_rate": 3.636228550142753e-05, "loss": 1.3988, "step": 2150 }, { "epoch": 0.7, "learning_rate": 3.6248308353318345e-05, "loss": 1.3945, "step": 2160 }, { "epoch": 0.71, "learning_rate": 3.613403730431917e-05, "loss": 1.3892, "step": 2170 }, { "epoch": 0.71, "learning_rate": 3.6019475340155666e-05, "loss": 1.3831, "step": 2180 }, { "epoch": 0.71, "learning_rate": 3.5904625454154675e-05, "loss": 1.4009, "step": 2190 }, { "epoch": 0.72, "learning_rate": 3.578949064716598e-05, "loss": 1.3689, "step": 2200 }, { "epoch": 0.72, "learning_rate": 3.567407392748391e-05, "loss": 1.3776, "step": 2210 }, { "epoch": 0.72, "learning_rate": 3.555837831076873e-05, "loss": 1.3939, "step": 2220 }, { "epoch": 0.73, "learning_rate": 3.5442406819967874e-05, "loss": 1.3888, "step": 2230 }, { "epoch": 0.73, "learning_rate": 3.532616248523692e-05, "loss": 1.3723, "step": 2240 }, { "epoch": 0.73, "learning_rate": 3.520964834386044e-05, "loss": 1.3849, "step": 2250 }, { "epoch": 0.74, "learning_rate": 3.509286744017264e-05, "loss": 1.3768, "step": 2260 }, { "epoch": 0.74, "learning_rate": 3.497582282547781e-05, "loss": 1.3777, "step": 2270 }, { "epoch": 0.74, "learning_rate": 3.4858517557970586e-05, "loss": 1.3897, "step": 2280 }, { "epoch": 0.75, "learning_rate": 3.474095470265609e-05, "loss": 1.3823, "step": 2290 }, { "epoch": 0.75, "learning_rate": 3.462313733126979e-05, "loss": 1.4039, "step": 2300 }, { "epoch": 0.75, "learning_rate": 3.450506852219728e-05, "loss": 1.4062, "step": 2310 }, { "epoch": 0.75, "learning_rate": 3.438675136039381e-05, "loss": 1.3909, "step": 2320 }, { "epoch": 0.76, "learning_rate": 3.426818893730373e-05, "loss": 1.4088, "step": 2330 }, { "epoch": 0.76, "learning_rate": 3.4149384350779666e-05, "loss": 1.4013, "step": 2340 }, { "epoch": 0.76, "learning_rate": 3.403034070500161e-05, "loss": 1.377, "step": 2350 }, { "epoch": 0.77, "learning_rate": 3.391106111039581e-05, "loss": 1.3647, "step": 2360 }, { "epoch": 0.77, "learning_rate": 3.379154868355347e-05, "loss": 1.4084, "step": 2370 }, { "epoch": 0.77, "learning_rate": 3.367180654714936e-05, "loss": 1.3842, "step": 2380 }, { "epoch": 0.78, "learning_rate": 3.355183782986018e-05, "loss": 1.3993, "step": 2390 }, { "epoch": 0.78, "learning_rate": 3.343164566628288e-05, "loss": 1.3933, "step": 2400 }, { "epoch": 0.78, "learning_rate": 3.3311233196852684e-05, "loss": 1.3612, "step": 2410 }, { "epoch": 0.79, "learning_rate": 3.319060356776108e-05, "loss": 1.3781, "step": 2420 }, { "epoch": 0.79, "learning_rate": 3.306975993087357e-05, "loss": 1.3966, "step": 2430 }, { "epoch": 0.79, "learning_rate": 3.29487054436474e-05, "loss": 1.3985, "step": 2440 }, { "epoch": 0.8, "learning_rate": 3.2827443269048985e-05, "loss": 1.3798, "step": 2450 }, { "epoch": 0.8, "learning_rate": 3.270597657547127e-05, "loss": 1.3878, "step": 2460 }, { "epoch": 0.8, "learning_rate": 3.2584308536651e-05, "loss": 1.3912, "step": 2470 }, { "epoch": 0.81, "learning_rate": 3.246244233158573e-05, "loss": 1.4125, "step": 2480 }, { "epoch": 0.81, "learning_rate": 3.2340381144450804e-05, "loss": 1.3724, "step": 2490 }, { "epoch": 0.81, "learning_rate": 3.2218128164516166e-05, "loss": 1.3851, "step": 2500 }, { "epoch": 0.82, "learning_rate": 3.209568658606297e-05, "loss": 1.3954, "step": 2510 }, { "epoch": 0.82, "learning_rate": 3.197305960830018e-05, "loss": 1.4135, "step": 2520 }, { "epoch": 0.82, "learning_rate": 3.185025043528096e-05, "loss": 1.3955, "step": 2530 }, { "epoch": 0.83, "learning_rate": 3.1727262275818945e-05, "loss": 1.3892, "step": 2540 }, { "epoch": 0.83, "learning_rate": 3.160409834340442e-05, "loss": 1.3654, "step": 2550 }, { "epoch": 0.83, "learning_rate": 3.148076185612032e-05, "loss": 1.3759, "step": 2560 }, { "epoch": 0.84, "learning_rate": 3.1357256036558205e-05, "loss": 1.3986, "step": 2570 }, { "epoch": 0.84, "learning_rate": 3.1233584111734e-05, "loss": 1.3876, "step": 2580 }, { "epoch": 0.84, "learning_rate": 3.1109749313003704e-05, "loss": 1.3983, "step": 2590 }, { "epoch": 0.85, "learning_rate": 3.0985754875979e-05, "loss": 1.3682, "step": 2600 }, { "epoch": 0.85, "learning_rate": 3.0861604040442605e-05, "loss": 1.4011, "step": 2610 }, { "epoch": 0.85, "learning_rate": 3.073730005026374e-05, "loss": 1.3981, "step": 2620 }, { "epoch": 0.86, "learning_rate": 3.0612846153313294e-05, "loss": 1.3812, "step": 2630 }, { "epoch": 0.86, "learning_rate": 3.0488245601379e-05, "loss": 1.3782, "step": 2640 }, { "epoch": 0.86, "learning_rate": 3.0363501650080468e-05, "loss": 1.3981, "step": 2650 }, { "epoch": 0.87, "learning_rate": 3.0238617558784076e-05, "loss": 1.3931, "step": 2660 }, { "epoch": 0.87, "learning_rate": 3.0113596590517877e-05, "loss": 1.3929, "step": 2670 }, { "epoch": 0.87, "learning_rate": 2.99884420118863e-05, "loss": 1.3919, "step": 2680 }, { "epoch": 0.88, "learning_rate": 2.9863157092984788e-05, "loss": 1.3934, "step": 2690 }, { "epoch": 0.88, "learning_rate": 2.9737745107314407e-05, "loss": 1.3747, "step": 2700 }, { "epoch": 0.88, "learning_rate": 2.9612209331696237e-05, "loss": 1.375, "step": 2710 }, { "epoch": 0.88, "learning_rate": 2.9486553046185862e-05, "loss": 1.3875, "step": 2720 }, { "epoch": 0.89, "learning_rate": 2.9360779533987515e-05, "loss": 1.3689, "step": 2730 }, { "epoch": 0.89, "learning_rate": 2.923489208136846e-05, "loss": 1.3643, "step": 2740 }, { "epoch": 0.89, "learning_rate": 2.9108893977573016e-05, "loss": 1.3559, "step": 2750 }, { "epoch": 0.9, "learning_rate": 2.898278851473663e-05, "loss": 1.3875, "step": 2760 }, { "epoch": 0.9, "learning_rate": 2.8856578987799913e-05, "loss": 1.3711, "step": 2770 }, { "epoch": 0.9, "learning_rate": 2.873026869442248e-05, "loss": 1.3982, "step": 2780 }, { "epoch": 0.91, "learning_rate": 2.8603860934896813e-05, "loss": 1.3914, "step": 2790 }, { "epoch": 0.91, "learning_rate": 2.8477359012062065e-05, "loss": 1.3714, "step": 2800 }, { "epoch": 0.91, "learning_rate": 2.835076623121771e-05, "loss": 1.3948, "step": 2810 }, { "epoch": 0.92, "learning_rate": 2.82240859000372e-05, "loss": 1.3789, "step": 2820 }, { "epoch": 0.92, "learning_rate": 2.8097321328481535e-05, "loss": 1.3839, "step": 2830 }, { "epoch": 0.92, "learning_rate": 2.797047582871281e-05, "loss": 1.3897, "step": 2840 }, { "epoch": 0.93, "learning_rate": 2.784355271500763e-05, "loss": 1.3806, "step": 2850 }, { "epoch": 0.93, "learning_rate": 2.7716555303670533e-05, "loss": 1.3792, "step": 2860 }, { "epoch": 0.93, "learning_rate": 2.7589486912947353e-05, "loss": 1.3847, "step": 2870 }, { "epoch": 0.94, "learning_rate": 2.746235086293848e-05, "loss": 1.3965, "step": 2880 }, { "epoch": 0.94, "learning_rate": 2.7335150475512162e-05, "loss": 1.3608, "step": 2890 }, { "epoch": 0.94, "learning_rate": 2.7207889074217678e-05, "loss": 1.3808, "step": 2900 }, { "epoch": 0.95, "learning_rate": 2.708056998419849e-05, "loss": 1.3923, "step": 2910 }, { "epoch": 0.95, "learning_rate": 2.6953196532105397e-05, "loss": 1.3836, "step": 2920 }, { "epoch": 0.95, "learning_rate": 2.682577204600958e-05, "loss": 1.3984, "step": 2930 }, { "epoch": 0.96, "learning_rate": 2.669829985531566e-05, "loss": 1.3784, "step": 2940 }, { "epoch": 0.96, "learning_rate": 2.6570783290674728e-05, "loss": 1.3891, "step": 2950 }, { "epoch": 0.96, "learning_rate": 2.644322568389727e-05, "loss": 1.3882, "step": 2960 }, { "epoch": 0.97, "learning_rate": 2.631563036786615e-05, "loss": 1.3884, "step": 2970 }, { "epoch": 0.97, "learning_rate": 2.6188000676449508e-05, "loss": 1.3885, "step": 2980 }, { "epoch": 0.97, "learning_rate": 2.6060339944413688e-05, "loss": 1.3753, "step": 2990 }, { "epoch": 0.98, "learning_rate": 2.5932651507336038e-05, "loss": 1.3881, "step": 3000 }, { "epoch": 0.98, "learning_rate": 2.580493870151782e-05, "loss": 1.3944, "step": 3010 }, { "epoch": 0.98, "learning_rate": 2.567720486389702e-05, "loss": 1.3844, "step": 3020 }, { "epoch": 0.99, "learning_rate": 2.5549453331961128e-05, "loss": 1.3867, "step": 3030 }, { "epoch": 0.99, "learning_rate": 2.5421687443659976e-05, "loss": 1.4, "step": 3040 }, { "epoch": 0.99, "learning_rate": 2.52939105373185e-05, "loss": 1.3971, "step": 3050 }, { "epoch": 1.0, "learning_rate": 2.516612595154953e-05, "loss": 1.385, "step": 3060 }, { "epoch": 1.0, "learning_rate": 2.503833702516653e-05, "loss": 1.3773, "step": 3070 }, { "epoch": 1.0, "learning_rate": 2.4910547097096386e-05, "loss": 1.3871, "step": 3080 }, { "epoch": 1.01, "learning_rate": 2.478275950629217e-05, "loss": 1.3734, "step": 3090 }, { "epoch": 1.01, "learning_rate": 2.4654977591645863e-05, "loss": 1.3553, "step": 3100 }, { "epoch": 1.01, "learning_rate": 2.452720469190115e-05, "loss": 1.3624, "step": 3110 }, { "epoch": 1.02, "learning_rate": 2.4399444145566155e-05, "loss": 1.3871, "step": 3120 }, { "epoch": 1.02, "learning_rate": 2.427169929082626e-05, "loss": 1.3805, "step": 3130 }, { "epoch": 1.02, "learning_rate": 2.4143973465456813e-05, "loss": 1.3822, "step": 3140 }, { "epoch": 1.02, "learning_rate": 2.4016270006735966e-05, "loss": 1.382, "step": 3150 }, { "epoch": 1.03, "learning_rate": 2.3888592251357483e-05, "loss": 1.3752, "step": 3160 }, { "epoch": 1.03, "learning_rate": 2.3760943535343517e-05, "loss": 1.3826, "step": 3170 }, { "epoch": 1.03, "learning_rate": 2.363332719395747e-05, "loss": 1.3658, "step": 3180 }, { "epoch": 1.04, "learning_rate": 2.350574656161686e-05, "loss": 1.3807, "step": 3190 }, { "epoch": 1.04, "learning_rate": 2.3378204971806172e-05, "loss": 1.3903, "step": 3200 }, { "epoch": 1.04, "learning_rate": 2.325070575698977e-05, "loss": 1.3759, "step": 3210 }, { "epoch": 1.05, "learning_rate": 2.312325224852481e-05, "loss": 1.3725, "step": 3220 }, { "epoch": 1.05, "learning_rate": 2.2995847776574253e-05, "loss": 1.3888, "step": 3230 }, { "epoch": 1.05, "learning_rate": 2.286849567001978e-05, "loss": 1.3832, "step": 3240 }, { "epoch": 1.06, "learning_rate": 2.2741199256374832e-05, "loss": 1.3807, "step": 3250 }, { "epoch": 1.06, "learning_rate": 2.261396186169773e-05, "loss": 1.3774, "step": 3260 }, { "epoch": 1.06, "learning_rate": 2.2486786810504676e-05, "loss": 1.4039, "step": 3270 }, { "epoch": 1.07, "learning_rate": 2.2359677425682936e-05, "loss": 1.3742, "step": 3280 }, { "epoch": 1.07, "learning_rate": 2.223263702840406e-05, "loss": 1.375, "step": 3290 }, { "epoch": 1.07, "learning_rate": 2.2105668938037005e-05, "loss": 1.3824, "step": 3300 }, { "epoch": 1.08, "learning_rate": 2.1978776472061485e-05, "loss": 1.3707, "step": 3310 }, { "epoch": 1.08, "learning_rate": 2.1851962945981262e-05, "loss": 1.3684, "step": 3320 }, { "epoch": 1.08, "learning_rate": 2.1725231673237536e-05, "loss": 1.3748, "step": 3330 }, { "epoch": 1.09, "learning_rate": 2.159858596512233e-05, "loss": 1.3866, "step": 3340 }, { "epoch": 1.09, "learning_rate": 2.1472029130692006e-05, "loss": 1.3767, "step": 3350 }, { "epoch": 1.09, "learning_rate": 2.1345564476680804e-05, "loss": 1.3521, "step": 3360 }, { "epoch": 1.1, "learning_rate": 2.1219195307414418e-05, "loss": 1.3645, "step": 3370 }, { "epoch": 1.1, "learning_rate": 2.1092924924723675e-05, "loss": 1.3621, "step": 3380 }, { "epoch": 1.1, "learning_rate": 2.0966756627858277e-05, "loss": 1.3843, "step": 3390 }, { "epoch": 1.11, "learning_rate": 2.0840693713400557e-05, "loss": 1.3738, "step": 3400 }, { "epoch": 1.11, "learning_rate": 2.071473947517939e-05, "loss": 1.3834, "step": 3410 }, { "epoch": 1.11, "learning_rate": 2.0588897204184092e-05, "loss": 1.3695, "step": 3420 }, { "epoch": 1.12, "learning_rate": 2.046317018847847e-05, "loss": 1.3589, "step": 3430 }, { "epoch": 1.12, "learning_rate": 2.0337561713114866e-05, "loss": 1.3772, "step": 3440 }, { "epoch": 1.12, "learning_rate": 2.021207506004836e-05, "loss": 1.3755, "step": 3450 }, { "epoch": 1.13, "learning_rate": 2.0086713508050996e-05, "loss": 1.3571, "step": 3460 }, { "epoch": 1.13, "learning_rate": 1.9961480332626136e-05, "loss": 1.377, "step": 3470 }, { "epoch": 1.13, "learning_rate": 1.9836378805922835e-05, "loss": 1.3748, "step": 3480 }, { "epoch": 1.14, "learning_rate": 1.9711412196650396e-05, "loss": 1.3708, "step": 3490 }, { "epoch": 1.14, "learning_rate": 1.9586583769992926e-05, "loss": 1.3691, "step": 3500 }, { "epoch": 1.14, "learning_rate": 1.946189678752404e-05, "loss": 1.3837, "step": 3510 }, { "epoch": 1.15, "learning_rate": 1.9337354507121607e-05, "loss": 1.3702, "step": 3520 }, { "epoch": 1.15, "learning_rate": 1.9212960182882712e-05, "loss": 1.3873, "step": 3530 }, { "epoch": 1.15, "learning_rate": 1.908871706503853e-05, "loss": 1.3766, "step": 3540 }, { "epoch": 1.15, "learning_rate": 1.896462839986944e-05, "loss": 1.3657, "step": 3550 }, { "epoch": 1.16, "learning_rate": 1.8840697429620253e-05, "loss": 1.3658, "step": 3560 }, { "epoch": 1.16, "learning_rate": 1.8716927392415427e-05, "loss": 1.3808, "step": 3570 }, { "epoch": 1.16, "learning_rate": 1.859332152217448e-05, "loss": 1.3772, "step": 3580 }, { "epoch": 1.17, "learning_rate": 1.846988304852754e-05, "loss": 1.3532, "step": 3590 }, { "epoch": 1.17, "learning_rate": 1.834661519673089e-05, "loss": 1.362, "step": 3600 }, { "epoch": 1.17, "learning_rate": 1.8223521187582714e-05, "loss": 1.3757, "step": 3610 }, { "epoch": 1.18, "learning_rate": 1.8100604237338993e-05, "loss": 1.3616, "step": 3620 }, { "epoch": 1.18, "learning_rate": 1.7977867557629425e-05, "loss": 1.3524, "step": 3630 }, { "epoch": 1.18, "learning_rate": 1.78553143553735e-05, "loss": 1.3708, "step": 3640 }, { "epoch": 1.19, "learning_rate": 1.7732947832696724e-05, "loss": 1.3939, "step": 3650 }, { "epoch": 1.19, "learning_rate": 1.7610771186846996e-05, "loss": 1.3592, "step": 3660 }, { "epoch": 1.19, "learning_rate": 1.7488787610110983e-05, "loss": 1.379, "step": 3670 }, { "epoch": 1.2, "learning_rate": 1.7367000289730762e-05, "loss": 1.3834, "step": 3680 }, { "epoch": 1.2, "learning_rate": 1.7245412407820552e-05, "loss": 1.3925, "step": 3690 }, { "epoch": 1.2, "learning_rate": 1.7124027141283545e-05, "loss": 1.3929, "step": 3700 }, { "epoch": 1.21, "learning_rate": 1.7002847661728906e-05, "loss": 1.3819, "step": 3710 }, { "epoch": 1.21, "learning_rate": 1.6881877135388895e-05, "loss": 1.3693, "step": 3720 }, { "epoch": 1.21, "learning_rate": 1.676111872303617e-05, "loss": 1.3912, "step": 3730 }, { "epoch": 1.22, "learning_rate": 1.6640575579901164e-05, "loss": 1.3653, "step": 3740 }, { "epoch": 1.22, "learning_rate": 1.6520250855589652e-05, "loss": 1.381, "step": 3750 }, { "epoch": 1.22, "learning_rate": 1.6400147694000488e-05, "loss": 1.3702, "step": 3760 }, { "epoch": 1.23, "learning_rate": 1.6280269233243413e-05, "loss": 1.3719, "step": 3770 }, { "epoch": 1.23, "learning_rate": 1.6160618605557076e-05, "loss": 1.3634, "step": 3780 }, { "epoch": 1.23, "learning_rate": 1.6041198937227234e-05, "loss": 1.3681, "step": 3790 }, { "epoch": 1.24, "learning_rate": 1.5922013348505e-05, "loss": 1.3696, "step": 3800 }, { "epoch": 1.24, "learning_rate": 1.5803064953525368e-05, "loss": 1.3756, "step": 3810 }, { "epoch": 1.24, "learning_rate": 1.5684356860225812e-05, "loss": 1.3825, "step": 3820 }, { "epoch": 1.25, "learning_rate": 1.5565892170265116e-05, "loss": 1.3762, "step": 3830 }, { "epoch": 1.25, "learning_rate": 1.544767397894229e-05, "loss": 1.3652, "step": 3840 }, { "epoch": 1.25, "learning_rate": 1.5329705375115727e-05, "loss": 1.373, "step": 3850 }, { "epoch": 1.26, "learning_rate": 1.5211989441122488e-05, "loss": 1.3666, "step": 3860 }, { "epoch": 1.26, "learning_rate": 1.509452925269775e-05, "loss": 1.3688, "step": 3870 }, { "epoch": 1.26, "learning_rate": 1.4977327878894465e-05, "loss": 1.359, "step": 3880 }, { "epoch": 1.27, "learning_rate": 1.4860388382003157e-05, "loss": 1.3677, "step": 3890 }, { "epoch": 1.27, "learning_rate": 1.4743713817471916e-05, "loss": 1.3872, "step": 3900 }, { "epoch": 1.27, "learning_rate": 1.4627307233826542e-05, "loss": 1.3626, "step": 3910 }, { "epoch": 1.28, "learning_rate": 1.4511171672590923e-05, "loss": 1.3524, "step": 3920 }, { "epoch": 1.28, "learning_rate": 1.4395310168207552e-05, "loss": 1.3627, "step": 3930 }, { "epoch": 1.28, "learning_rate": 1.4279725747958244e-05, "loss": 1.376, "step": 3940 }, { "epoch": 1.29, "learning_rate": 1.4164421431885017e-05, "loss": 1.3671, "step": 3950 }, { "epoch": 1.29, "learning_rate": 1.4049400232711216e-05, "loss": 1.3871, "step": 3960 }, { "epoch": 1.29, "learning_rate": 1.3934665155762782e-05, "loss": 1.3874, "step": 3970 }, { "epoch": 1.29, "learning_rate": 1.38202191988897e-05, "loss": 1.3879, "step": 3980 }, { "epoch": 1.3, "learning_rate": 1.3706065352387726e-05, "loss": 1.3702, "step": 3990 }, { "epoch": 1.3, "learning_rate": 1.3592206598920219e-05, "loss": 1.3674, "step": 4000 }, { "epoch": 1.3, "learning_rate": 1.347864591344018e-05, "loss": 1.3566, "step": 4010 }, { "epoch": 1.31, "learning_rate": 1.33653862631126e-05, "loss": 1.369, "step": 4020 }, { "epoch": 1.31, "learning_rate": 1.3252430607236869e-05, "loss": 1.3647, "step": 4030 }, { "epoch": 1.31, "learning_rate": 1.3139781897169461e-05, "loss": 1.3856, "step": 4040 }, { "epoch": 1.32, "learning_rate": 1.3027443076246832e-05, "loss": 1.3812, "step": 4050 }, { "epoch": 1.32, "learning_rate": 1.291541707970855e-05, "loss": 1.3782, "step": 4060 }, { "epoch": 1.32, "learning_rate": 1.2803706834620524e-05, "loss": 1.3616, "step": 4070 }, { "epoch": 1.33, "learning_rate": 1.2692315259798566e-05, "loss": 1.3692, "step": 4080 }, { "epoch": 1.33, "learning_rate": 1.2581245265732173e-05, "loss": 1.3801, "step": 4090 }, { "epoch": 1.33, "learning_rate": 1.2470499754508392e-05, "loss": 1.3896, "step": 4100 }, { "epoch": 1.34, "learning_rate": 1.2360081619736035e-05, "loss": 1.3845, "step": 4110 }, { "epoch": 1.34, "learning_rate": 1.2249993746470096e-05, "loss": 1.3785, "step": 4120 }, { "epoch": 1.34, "learning_rate": 1.2140239011136348e-05, "loss": 1.365, "step": 4130 }, { "epoch": 1.35, "learning_rate": 1.2030820281456151e-05, "loss": 1.3647, "step": 4140 }, { "epoch": 1.35, "learning_rate": 1.1921740416371599e-05, "loss": 1.3774, "step": 4150 }, { "epoch": 1.35, "learning_rate": 1.1813002265970764e-05, "loss": 1.3574, "step": 4160 }, { "epoch": 1.36, "learning_rate": 1.170460867141322e-05, "loss": 1.3415, "step": 4170 }, { "epoch": 1.36, "learning_rate": 1.1596562464855858e-05, "loss": 1.3702, "step": 4180 }, { "epoch": 1.36, "learning_rate": 1.1488866469378862e-05, "loss": 1.3748, "step": 4190 }, { "epoch": 1.37, "learning_rate": 1.1381523498911905e-05, "loss": 1.3777, "step": 4200 }, { "epoch": 1.37, "learning_rate": 1.1274536358160711e-05, "loss": 1.3694, "step": 4210 }, { "epoch": 1.37, "learning_rate": 1.1167907842533682e-05, "loss": 1.3656, "step": 4220 }, { "epoch": 1.38, "learning_rate": 1.106164073806892e-05, "loss": 1.3951, "step": 4230 }, { "epoch": 1.38, "learning_rate": 1.095573782136143e-05, "loss": 1.3613, "step": 4240 }, { "epoch": 1.38, "learning_rate": 1.085020185949051e-05, "loss": 1.3749, "step": 4250 }, { "epoch": 1.39, "learning_rate": 1.074503560994754e-05, "loss": 1.359, "step": 4260 }, { "epoch": 1.39, "learning_rate": 1.064024182056387e-05, "loss": 1.3684, "step": 4270 }, { "epoch": 1.39, "learning_rate": 1.0535823229439033e-05, "loss": 1.3549, "step": 4280 }, { "epoch": 1.4, "learning_rate": 1.0431782564869228e-05, "loss": 1.3783, "step": 4290 }, { "epoch": 1.4, "learning_rate": 1.032812254527602e-05, "loss": 1.3749, "step": 4300 }, { "epoch": 1.4, "learning_rate": 1.0224845879135286e-05, "loss": 1.367, "step": 4310 }, { "epoch": 1.41, "learning_rate": 1.0121955264906496e-05, "loss": 1.3607, "step": 4320 }, { "epoch": 1.41, "learning_rate": 1.0019453390962183e-05, "loss": 1.3881, "step": 4330 }, { "epoch": 1.41, "learning_rate": 9.917342935517665e-06, "loss": 1.3595, "step": 4340 }, { "epoch": 1.42, "learning_rate": 9.815626566561134e-06, "loss": 1.3645, "step": 4350 }, { "epoch": 1.42, "learning_rate": 9.714306941783907e-06, "loss": 1.3812, "step": 4360 }, { "epoch": 1.42, "learning_rate": 9.613386708510966e-06, "loss": 1.3675, "step": 4370 }, { "epoch": 1.42, "learning_rate": 9.512868503631836e-06, "loss": 1.3587, "step": 4380 }, { "epoch": 1.43, "learning_rate": 9.412754953531663e-06, "loss": 1.3704, "step": 4390 }, { "epoch": 1.43, "learning_rate": 9.313048674022565e-06, "loss": 1.3753, "step": 4400 }, { "epoch": 1.43, "learning_rate": 9.213752270275339e-06, "loss": 1.3599, "step": 4410 }, { "epoch": 1.44, "learning_rate": 9.114868336751334e-06, "loss": 1.3751, "step": 4420 }, { "epoch": 1.44, "learning_rate": 9.016399457134723e-06, "loss": 1.3736, "step": 4430 }, { "epoch": 1.44, "learning_rate": 8.918348204264926e-06, "loss": 1.3734, "step": 4440 }, { "epoch": 1.45, "learning_rate": 8.82071714006945e-06, "loss": 1.3707, "step": 4450 }, { "epoch": 1.45, "learning_rate": 8.723508815496929e-06, "loss": 1.358, "step": 4460 }, { "epoch": 1.45, "learning_rate": 8.626725770450425e-06, "loss": 1.3532, "step": 4470 }, { "epoch": 1.46, "learning_rate": 8.530370533721144e-06, "loss": 1.3782, "step": 4480 }, { "epoch": 1.46, "learning_rate": 8.434445622922307e-06, "loss": 1.3956, "step": 4490 }, { "epoch": 1.46, "learning_rate": 8.338953544423379e-06, "loss": 1.3779, "step": 4500 }, { "epoch": 1.47, "learning_rate": 8.243896793284597e-06, "loss": 1.3669, "step": 4510 }, { "epoch": 1.47, "learning_rate": 8.149277853191758e-06, "loss": 1.3599, "step": 4520 }, { "epoch": 1.47, "learning_rate": 8.055099196391344e-06, "loss": 1.3644, "step": 4530 }, { "epoch": 1.48, "learning_rate": 7.96136328362592e-06, "loss": 1.3836, "step": 4540 }, { "epoch": 1.48, "learning_rate": 7.868072564069806e-06, "loss": 1.3765, "step": 4550 }, { "epoch": 1.48, "learning_rate": 7.775229475265142e-06, "loss": 1.3769, "step": 4560 }, { "epoch": 1.49, "learning_rate": 7.68283644305817e-06, "loss": 1.3446, "step": 4570 }, { "epoch": 1.49, "learning_rate": 7.590895881535823e-06, "loss": 1.3628, "step": 4580 }, { "epoch": 1.49, "learning_rate": 7.499410192962697e-06, "loss": 1.3612, "step": 4590 }, { "epoch": 1.5, "learning_rate": 7.408381767718264e-06, "loss": 1.3687, "step": 4600 }, { "epoch": 1.5, "learning_rate": 7.31781298423439e-06, "loss": 1.3706, "step": 4610 }, { "epoch": 1.5, "learning_rate": 7.227706208933238e-06, "loss": 1.3763, "step": 4620 }, { "epoch": 1.51, "learning_rate": 7.138063796165406e-06, "loss": 1.3758, "step": 4630 }, { "epoch": 1.51, "learning_rate": 7.048888088148403e-06, "loss": 1.3736, "step": 4640 }, { "epoch": 1.51, "learning_rate": 6.960181414905479e-06, "loss": 1.3707, "step": 4650 }, { "epoch": 1.52, "learning_rate": 6.871946094204737e-06, "loss": 1.3686, "step": 4660 }, { "epoch": 1.52, "learning_rate": 6.784184431498547e-06, "loss": 1.3549, "step": 4670 }, { "epoch": 1.52, "learning_rate": 6.696898719863343e-06, "loss": 1.3768, "step": 4680 }, { "epoch": 1.53, "learning_rate": 6.610091239939703e-06, "loss": 1.3805, "step": 4690 }, { "epoch": 1.53, "learning_rate": 6.523764259872719e-06, "loss": 1.3811, "step": 4700 }, { "epoch": 1.53, "learning_rate": 6.4379200352528035e-06, "loss": 1.3782, "step": 4710 }, { "epoch": 1.54, "learning_rate": 6.352560809056685e-06, "loss": 1.3944, "step": 4720 }, { "epoch": 1.54, "learning_rate": 6.267688811588848e-06, "loss": 1.381, "step": 4730 }, { "epoch": 1.54, "learning_rate": 6.183306260423258e-06, "loss": 1.3751, "step": 4740 }, { "epoch": 1.55, "learning_rate": 6.09941536034537e-06, "loss": 1.3785, "step": 4750 }, { "epoch": 1.55, "learning_rate": 6.0160183032945876e-06, "loss": 1.3573, "step": 4760 }, { "epoch": 1.55, "learning_rate": 5.933117268306957e-06, "loss": 1.3768, "step": 4770 }, { "epoch": 1.56, "learning_rate": 5.850714421458217e-06, "loss": 1.373, "step": 4780 }, { "epoch": 1.56, "learning_rate": 5.76881191580724e-06, "loss": 1.3418, "step": 4790 }, { "epoch": 1.56, "learning_rate": 5.687411891339756e-06, "loss": 1.3751, "step": 4800 }, { "epoch": 1.56, "learning_rate": 5.60651647491243e-06, "loss": 1.3603, "step": 4810 }, { "epoch": 1.57, "learning_rate": 5.526127780197285e-06, "loss": 1.3825, "step": 4820 }, { "epoch": 1.57, "learning_rate": 5.4462479076265425e-06, "loss": 1.3876, "step": 4830 }, { "epoch": 1.57, "learning_rate": 5.366878944337636e-06, "loss": 1.364, "step": 4840 }, { "epoch": 1.58, "learning_rate": 5.288022964118747e-06, "loss": 1.3762, "step": 4850 }, { "epoch": 1.58, "learning_rate": 5.209682027354609e-06, "loss": 1.3586, "step": 4860 }, { "epoch": 1.58, "learning_rate": 5.131858180972671e-06, "loss": 1.3729, "step": 4870 }, { "epoch": 1.59, "learning_rate": 5.054553458389588e-06, "loss": 1.3661, "step": 4880 }, { "epoch": 1.59, "learning_rate": 4.977769879458139e-06, "loss": 1.3684, "step": 4890 }, { "epoch": 1.59, "learning_rate": 4.901509450414418e-06, "loss": 1.3713, "step": 4900 }, { "epoch": 1.6, "learning_rate": 4.825774163825408e-06, "loss": 1.3628, "step": 4910 }, { "epoch": 1.6, "learning_rate": 4.750565998536949e-06, "loss": 1.3797, "step": 4920 }, { "epoch": 1.6, "learning_rate": 4.675886919622022e-06, "loss": 1.3815, "step": 4930 }, { "epoch": 1.61, "learning_rate": 4.601738878329371e-06, "loss": 1.3734, "step": 4940 }, { "epoch": 1.61, "learning_rate": 4.528123812032578e-06, "loss": 1.391, "step": 4950 }, { "epoch": 1.61, "learning_rate": 4.45504364417941e-06, "loss": 1.3436, "step": 4960 }, { "epoch": 1.62, "learning_rate": 4.382500284241542e-06, "loss": 1.3582, "step": 4970 }, { "epoch": 1.62, "learning_rate": 4.3104956276647125e-06, "loss": 1.353, "step": 4980 }, { "epoch": 1.62, "learning_rate": 4.239031555819173e-06, "loss": 1.3724, "step": 4990 }, { "epoch": 1.63, "learning_rate": 4.168109935950518e-06, "loss": 1.3635, "step": 5000 }, { "epoch": 1.63, "learning_rate": 4.097732621130926e-06, "loss": 1.3946, "step": 5010 }, { "epoch": 1.63, "learning_rate": 4.027901450210716e-06, "loss": 1.3586, "step": 5020 }, { "epoch": 1.64, "learning_rate": 3.958618247770318e-06, "loss": 1.3666, "step": 5030 }, { "epoch": 1.64, "learning_rate": 3.889884824072601e-06, "loss": 1.3442, "step": 5040 }, { "epoch": 1.64, "learning_rate": 3.821702975015545e-06, "loss": 1.3656, "step": 5050 }, { "epoch": 1.65, "learning_rate": 3.7540744820853595e-06, "loss": 1.3828, "step": 5060 }, { "epoch": 1.65, "learning_rate": 3.6870011123099097e-06, "loss": 1.3589, "step": 5070 }, { "epoch": 1.65, "learning_rate": 3.6204846182125386e-06, "loss": 1.3532, "step": 5080 }, { "epoch": 1.66, "learning_rate": 3.554526737766306e-06, "loss": 1.3886, "step": 5090 }, { "epoch": 1.66, "learning_rate": 3.4891291943485587e-06, "loss": 1.3721, "step": 5100 }, { "epoch": 1.66, "learning_rate": 3.424293696695896e-06, "loss": 1.3809, "step": 5110 }, { "epoch": 1.67, "learning_rate": 3.3600219388595256e-06, "loss": 1.3511, "step": 5120 }, { "epoch": 1.67, "learning_rate": 3.2963156001610374e-06, "loss": 1.3584, "step": 5130 }, { "epoch": 1.67, "learning_rate": 3.2331763451484615e-06, "loss": 1.3515, "step": 5140 }, { "epoch": 1.68, "learning_rate": 3.170605823552808e-06, "loss": 1.3756, "step": 5150 }, { "epoch": 1.68, "learning_rate": 3.1086056702450015e-06, "loss": 1.3506, "step": 5160 }, { "epoch": 1.68, "learning_rate": 3.0471775051930857e-06, "loss": 1.3462, "step": 5170 }, { "epoch": 1.69, "learning_rate": 2.986322933419941e-06, "loss": 1.3958, "step": 5180 }, { "epoch": 1.69, "learning_rate": 2.926043544961374e-06, "loss": 1.3659, "step": 5190 }, { "epoch": 1.69, "learning_rate": 2.8663409148245056e-06, "loss": 1.3688, "step": 5200 }, { "epoch": 1.69, "learning_rate": 2.807216602946672e-06, "loss": 1.3554, "step": 5210 }, { "epoch": 1.7, "learning_rate": 2.748672154154641e-06, "loss": 1.3817, "step": 5220 }, { "epoch": 1.7, "learning_rate": 2.6907090981242696e-06, "loss": 1.3588, "step": 5230 }, { "epoch": 1.7, "learning_rate": 2.6333289493405023e-06, "loss": 1.3604, "step": 5240 }, { "epoch": 1.71, "learning_rate": 2.5765332070578296e-06, "loss": 1.3603, "step": 5250 }, { "epoch": 1.71, "learning_rate": 2.520323355261109e-06, "loss": 1.357, "step": 5260 }, { "epoch": 1.71, "learning_rate": 2.4647008626267703e-06, "loss": 1.3592, "step": 5270 }, { "epoch": 1.72, "learning_rate": 2.4096671824844763e-06, "loss": 1.3918, "step": 5280 }, { "epoch": 1.72, "learning_rate": 2.355223752779123e-06, "loss": 1.3583, "step": 5290 }, { "epoch": 1.72, "learning_rate": 2.301371996033269e-06, "loss": 1.3667, "step": 5300 }, { "epoch": 1.73, "learning_rate": 2.2481133193099855e-06, "loss": 1.3801, "step": 5310 }, { "epoch": 1.73, "learning_rate": 2.1954491141760654e-06, "loss": 1.3825, "step": 5320 }, { "epoch": 1.73, "learning_rate": 2.1433807566656973e-06, "loss": 1.348, "step": 5330 }, { "epoch": 1.74, "learning_rate": 2.0919096072444855e-06, "loss": 1.3647, "step": 5340 }, { "epoch": 1.74, "learning_rate": 2.041037010773905e-06, "loss": 1.3553, "step": 5350 }, { "epoch": 1.74, "learning_rate": 1.9907642964761745e-06, "loss": 1.3665, "step": 5360 }, { "epoch": 1.75, "learning_rate": 1.941092777899525e-06, "loss": 1.3685, "step": 5370 }, { "epoch": 1.75, "learning_rate": 1.892023752883862e-06, "loss": 1.3898, "step": 5380 }, { "epoch": 1.75, "learning_rate": 1.843558503526871e-06, "loss": 1.361, "step": 5390 }, { "epoch": 1.76, "learning_rate": 1.7956982961505275e-06, "loss": 1.3769, "step": 5400 }, { "epoch": 1.76, "learning_rate": 1.7484443812679718e-06, "loss": 1.3609, "step": 5410 }, { "epoch": 1.76, "learning_rate": 1.7017979935508715e-06, "loss": 1.3611, "step": 5420 }, { "epoch": 1.77, "learning_rate": 1.6557603517971615e-06, "loss": 1.3874, "step": 5430 }, { "epoch": 1.77, "learning_rate": 1.6103326588991663e-06, "loss": 1.3839, "step": 5440 }, { "epoch": 1.77, "learning_rate": 1.5655161018121978e-06, "loss": 1.3764, "step": 5450 }, { "epoch": 1.78, "learning_rate": 1.5213118515235492e-06, "loss": 1.3663, "step": 5460 }, { "epoch": 1.78, "learning_rate": 1.4777210630218642e-06, "loss": 1.3684, "step": 5470 }, { "epoch": 1.78, "learning_rate": 1.4347448752669802e-06, "loss": 1.3692, "step": 5480 }, { "epoch": 1.79, "learning_rate": 1.3923844111601914e-06, "loss": 1.3716, "step": 5490 }, { "epoch": 1.79, "learning_rate": 1.350640777514864e-06, "loss": 1.3837, "step": 5500 }, { "epoch": 1.79, "learning_rate": 1.3095150650275362e-06, "loss": 1.3721, "step": 5510 }, { "epoch": 1.8, "learning_rate": 1.2690083482494396e-06, "loss": 1.3542, "step": 5520 }, { "epoch": 1.8, "learning_rate": 1.2291216855584014e-06, "loss": 1.366, "step": 5530 }, { "epoch": 1.8, "learning_rate": 1.189856119131183e-06, "loss": 1.3597, "step": 5540 }, { "epoch": 1.81, "learning_rate": 1.1512126749162717e-06, "loss": 1.3784, "step": 5550 }, { "epoch": 1.81, "learning_rate": 1.1131923626070683e-06, "loss": 1.3439, "step": 5560 }, { "epoch": 1.81, "learning_rate": 1.075796175615487e-06, "loss": 1.3833, "step": 5570 }, { "epoch": 1.82, "learning_rate": 1.0390250910460252e-06, "loss": 1.3677, "step": 5580 }, { "epoch": 1.82, "learning_rate": 1.0028800696702228e-06, "loss": 1.3428, "step": 5590 }, { "epoch": 1.82, "learning_rate": 9.673620559015411e-07, "loss": 1.3763, "step": 5600 }, { "epoch": 1.83, "learning_rate": 9.324719777707241e-07, "loss": 1.3841, "step": 5610 }, { "epoch": 1.83, "learning_rate": 8.98210746901515e-07, "loss": 1.3711, "step": 5620 }, { "epoch": 1.83, "learning_rate": 8.645792584868556e-07, "loss": 1.3583, "step": 5630 }, { "epoch": 1.83, "learning_rate": 8.315783912654945e-07, "loss": 1.359, "step": 5640 }, { "epoch": 1.84, "learning_rate": 7.992090074990161e-07, "loss": 1.3609, "step": 5650 }, { "epoch": 1.84, "learning_rate": 7.674719529493279e-07, "loss": 1.3607, "step": 5660 }, { "epoch": 1.84, "learning_rate": 7.363680568565568e-07, "loss": 1.3597, "step": 5670 }, { "epoch": 1.85, "learning_rate": 7.058981319173653e-07, "loss": 1.3716, "step": 5680 }, { "epoch": 1.85, "learning_rate": 6.760629742637387e-07, "loss": 1.3618, "step": 5690 }, { "epoch": 1.85, "learning_rate": 6.468633634421822e-07, "loss": 1.371, "step": 5700 }, { "epoch": 1.86, "learning_rate": 6.183000623933288e-07, "loss": 1.3795, "step": 5710 }, { "epoch": 1.86, "learning_rate": 5.903738174320384e-07, "loss": 1.3657, "step": 5720 }, { "epoch": 1.86, "learning_rate": 5.630853582278773e-07, "loss": 1.3638, "step": 5730 }, { "epoch": 1.87, "learning_rate": 5.364353977860503e-07, "loss": 1.3451, "step": 5740 }, { "epoch": 1.87, "learning_rate": 5.104246324287742e-07, "loss": 1.3623, "step": 5750 }, { "epoch": 1.87, "learning_rate": 4.850537417771112e-07, "loss": 1.3863, "step": 5760 }, { "epoch": 1.88, "learning_rate": 4.6032338873316095e-07, "loss": 1.3741, "step": 5770 }, { "epoch": 1.88, "learning_rate": 4.36234219462775e-07, "loss": 1.3653, "step": 5780 }, { "epoch": 1.88, "learning_rate": 4.1278686337867247e-07, "loss": 1.3724, "step": 5790 }, { "epoch": 1.89, "learning_rate": 3.89981933123984e-07, "loss": 1.3689, "step": 5800 }, { "epoch": 1.89, "learning_rate": 3.6782002455623685e-07, "loss": 1.3527, "step": 5810 }, { "epoch": 1.89, "learning_rate": 3.463017167318172e-07, "loss": 1.3828, "step": 5820 }, { "epoch": 1.9, "learning_rate": 3.25427571890799e-07, "loss": 1.3761, "step": 5830 }, { "epoch": 1.9, "learning_rate": 3.0519813544228335e-07, "loss": 1.3655, "step": 5840 }, { "epoch": 1.9, "learning_rate": 2.8561393595014085e-07, "loss": 1.3767, "step": 5850 }, { "epoch": 1.91, "learning_rate": 2.666754851191999e-07, "loss": 1.3507, "step": 5860 }, { "epoch": 1.91, "learning_rate": 2.48383277781869e-07, "loss": 1.3729, "step": 5870 }, { "epoch": 1.91, "learning_rate": 2.3073779188521605e-07, "loss": 1.3858, "step": 5880 }, { "epoch": 1.92, "learning_rate": 2.1373948847848436e-07, "loss": 1.3318, "step": 5890 }, { "epoch": 1.92, "learning_rate": 1.973888117010353e-07, "loss": 1.3734, "step": 5900 }, { "epoch": 1.92, "learning_rate": 1.8168618877074662e-07, "loss": 1.3747, "step": 5910 }, { "epoch": 1.93, "learning_rate": 1.666320299728602e-07, "loss": 1.3582, "step": 5920 }, { "epoch": 1.93, "learning_rate": 1.5222672864924347e-07, "loss": 1.3617, "step": 5930 }, { "epoch": 1.93, "learning_rate": 1.3847066118812812e-07, "loss": 1.3681, "step": 5940 }, { "epoch": 1.94, "learning_rate": 1.2536418701427078e-07, "loss": 1.3815, "step": 5950 }, { "epoch": 1.94, "learning_rate": 1.1290764857956048e-07, "loss": 1.3744, "step": 5960 }, { "epoch": 1.94, "learning_rate": 1.0110137135406484e-07, "loss": 1.3752, "step": 5970 }, { "epoch": 1.95, "learning_rate": 8.994566381753666e-08, "loss": 1.361, "step": 5980 }, { "epoch": 1.95, "learning_rate": 7.94408174513539e-08, "loss": 1.3812, "step": 5990 }, { "epoch": 1.95, "learning_rate": 6.958710673089508e-08, "loss": 1.3791, "step": 6000 }, { "epoch": 1.96, "learning_rate": 6.038478911837008e-08, "loss": 1.3813, "step": 6010 }, { "epoch": 1.96, "learning_rate": 5.1834105056094985e-08, "loss": 1.3731, "step": 6020 }, { "epoch": 1.96, "learning_rate": 4.3935277960216456e-08, "loss": 1.3714, "step": 6030 }, { "epoch": 1.97, "learning_rate": 3.668851421486097e-08, "loss": 1.379, "step": 6040 }, { "epoch": 1.97, "learning_rate": 3.0094003166752924e-08, "loss": 1.3665, "step": 6050 }, { "epoch": 1.97, "learning_rate": 2.4151917120254773e-08, "loss": 1.3722, "step": 6060 }, { "epoch": 1.97, "learning_rate": 1.8862411332884467e-08, "loss": 1.3783, "step": 6070 }, { "epoch": 1.98, "learning_rate": 1.422562401123817e-08, "loss": 1.3749, "step": 6080 }, { "epoch": 1.98, "learning_rate": 1.0241676307398695e-08, "loss": 1.3732, "step": 6090 }, { "epoch": 1.98, "learning_rate": 6.910672315749156e-09, "loss": 1.3683, "step": 6100 }, { "epoch": 1.99, "learning_rate": 4.232699070269574e-09, "loss": 1.3613, "step": 6110 }, { "epoch": 1.99, "learning_rate": 2.207826542258151e-09, "loss": 1.3622, "step": 6120 }, { "epoch": 1.99, "learning_rate": 8.361076385021705e-10, "loss": 1.3642, "step": 6130 }, { "epoch": 2.0, "learning_rate": 1.1757819989022255e-10, "loss": 1.3579, "step": 6140 }, { "epoch": 2.0, "step": 6146, "total_flos": 1.0786406469474451e+19, "train_loss": 1.394172432598756, "train_runtime": 79574.7503, "train_samples_per_second": 7.416, "train_steps_per_second": 0.077 } ], "max_steps": 6146, "num_train_epochs": 2, "total_flos": 1.0786406469474451e+19, "trial_name": null, "trial_params": null }