{ "best_metric": 0.3200523853302002, "best_model_checkpoint": "/leonardo_work/IscrC_AGENT/PROFES2025/results/balanced_per_dataset/oss_large/checkpoint-9756", "epoch": 5.0, "eval_steps": 500, "global_step": 24390, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.9995899958999593e-05, "loss": 0.6952, "step": 10 }, { "epoch": 0.0, "learning_rate": 1.999179991799918e-05, "loss": 0.6623, "step": 20 }, { "epoch": 0.01, "learning_rate": 1.9987699876998773e-05, "loss": 0.6305, "step": 30 }, { "epoch": 0.01, "learning_rate": 1.998359983599836e-05, "loss": 0.652, "step": 40 }, { "epoch": 0.01, "learning_rate": 1.9979499794997953e-05, "loss": 0.6033, "step": 50 }, { "epoch": 0.01, "learning_rate": 1.997539975399754e-05, "loss": 0.5846, "step": 60 }, { "epoch": 0.01, "learning_rate": 1.9971299712997133e-05, "loss": 0.6314, "step": 70 }, { "epoch": 0.02, "learning_rate": 1.996719967199672e-05, "loss": 0.6384, "step": 80 }, { "epoch": 0.02, "learning_rate": 1.9963099630996312e-05, "loss": 0.5717, "step": 90 }, { "epoch": 0.02, "learning_rate": 1.99589995899959e-05, "loss": 0.5783, "step": 100 }, { "epoch": 0.02, "learning_rate": 1.9954899548995492e-05, "loss": 0.6091, "step": 110 }, { "epoch": 0.02, "learning_rate": 1.9950799507995084e-05, "loss": 0.602, "step": 120 }, { "epoch": 0.03, "learning_rate": 1.9946699466994672e-05, "loss": 0.6114, "step": 130 }, { "epoch": 0.03, "learning_rate": 1.9942599425994264e-05, "loss": 0.5192, "step": 140 }, { "epoch": 0.03, "learning_rate": 1.9938499384993852e-05, "loss": 0.5235, "step": 150 }, { "epoch": 0.03, "learning_rate": 1.9934399343993443e-05, "loss": 0.5575, "step": 160 }, { "epoch": 0.03, "learning_rate": 1.993029930299303e-05, "loss": 0.5503, "step": 170 }, { "epoch": 0.04, "learning_rate": 1.9926199261992623e-05, "loss": 0.5836, "step": 180 }, { "epoch": 0.04, "learning_rate": 1.992209922099221e-05, "loss": 0.5566, "step": 190 }, { "epoch": 0.04, "learning_rate": 1.99179991799918e-05, "loss": 0.5367, "step": 200 }, { "epoch": 0.04, "learning_rate": 1.991389913899139e-05, "loss": 0.574, "step": 210 }, { "epoch": 0.05, "learning_rate": 1.990979909799098e-05, "loss": 0.6571, "step": 220 }, { "epoch": 0.05, "learning_rate": 1.990569905699057e-05, "loss": 0.5375, "step": 230 }, { "epoch": 0.05, "learning_rate": 1.990159901599016e-05, "loss": 0.5235, "step": 240 }, { "epoch": 0.05, "learning_rate": 1.989749897498975e-05, "loss": 0.5618, "step": 250 }, { "epoch": 0.05, "learning_rate": 1.9893398933989342e-05, "loss": 0.55, "step": 260 }, { "epoch": 0.06, "learning_rate": 1.988929889298893e-05, "loss": 0.5692, "step": 270 }, { "epoch": 0.06, "learning_rate": 1.9885198851988522e-05, "loss": 0.5843, "step": 280 }, { "epoch": 0.06, "learning_rate": 1.988109881098811e-05, "loss": 0.537, "step": 290 }, { "epoch": 0.06, "learning_rate": 1.9876998769987702e-05, "loss": 0.4374, "step": 300 }, { "epoch": 0.06, "learning_rate": 1.987289872898729e-05, "loss": 0.5931, "step": 310 }, { "epoch": 0.07, "learning_rate": 1.9868798687986882e-05, "loss": 0.5185, "step": 320 }, { "epoch": 0.07, "learning_rate": 1.986469864698647e-05, "loss": 0.5998, "step": 330 }, { "epoch": 0.07, "learning_rate": 1.986059860598606e-05, "loss": 0.4671, "step": 340 }, { "epoch": 0.07, "learning_rate": 1.985649856498565e-05, "loss": 0.5078, "step": 350 }, { "epoch": 0.07, "learning_rate": 1.985239852398524e-05, "loss": 0.4383, "step": 360 }, { "epoch": 0.08, "learning_rate": 1.9848298482984833e-05, "loss": 0.4909, "step": 370 }, { "epoch": 0.08, "learning_rate": 1.984419844198442e-05, "loss": 0.6681, "step": 380 }, { "epoch": 0.08, "learning_rate": 1.9840098400984013e-05, "loss": 0.5045, "step": 390 }, { "epoch": 0.08, "learning_rate": 1.98359983599836e-05, "loss": 0.511, "step": 400 }, { "epoch": 0.08, "learning_rate": 1.9831898318983192e-05, "loss": 0.4996, "step": 410 }, { "epoch": 0.09, "learning_rate": 1.982779827798278e-05, "loss": 0.4917, "step": 420 }, { "epoch": 0.09, "learning_rate": 1.9823698236982372e-05, "loss": 0.4922, "step": 430 }, { "epoch": 0.09, "learning_rate": 1.981959819598196e-05, "loss": 0.5111, "step": 440 }, { "epoch": 0.09, "learning_rate": 1.9815498154981552e-05, "loss": 0.4353, "step": 450 }, { "epoch": 0.09, "learning_rate": 1.981139811398114e-05, "loss": 0.4498, "step": 460 }, { "epoch": 0.1, "learning_rate": 1.9807298072980732e-05, "loss": 0.5585, "step": 470 }, { "epoch": 0.1, "learning_rate": 1.9803198031980323e-05, "loss": 0.4362, "step": 480 }, { "epoch": 0.1, "learning_rate": 1.979909799097991e-05, "loss": 0.5014, "step": 490 }, { "epoch": 0.1, "learning_rate": 1.9794997949979503e-05, "loss": 0.495, "step": 500 }, { "epoch": 0.1, "learning_rate": 1.979089790897909e-05, "loss": 0.4548, "step": 510 }, { "epoch": 0.11, "learning_rate": 1.9786797867978683e-05, "loss": 0.5371, "step": 520 }, { "epoch": 0.11, "learning_rate": 1.978269782697827e-05, "loss": 0.474, "step": 530 }, { "epoch": 0.11, "learning_rate": 1.9778597785977863e-05, "loss": 0.4915, "step": 540 }, { "epoch": 0.11, "learning_rate": 1.977449774497745e-05, "loss": 0.4911, "step": 550 }, { "epoch": 0.11, "learning_rate": 1.9770397703977043e-05, "loss": 0.4057, "step": 560 }, { "epoch": 0.12, "learning_rate": 1.976629766297663e-05, "loss": 0.4819, "step": 570 }, { "epoch": 0.12, "learning_rate": 1.9762197621976222e-05, "loss": 0.4343, "step": 580 }, { "epoch": 0.12, "learning_rate": 1.9758097580975814e-05, "loss": 0.4961, "step": 590 }, { "epoch": 0.12, "learning_rate": 1.97539975399754e-05, "loss": 0.4207, "step": 600 }, { "epoch": 0.13, "learning_rate": 1.974989749897499e-05, "loss": 0.4589, "step": 610 }, { "epoch": 0.13, "learning_rate": 1.9745797457974582e-05, "loss": 0.551, "step": 620 }, { "epoch": 0.13, "learning_rate": 1.974169741697417e-05, "loss": 0.4839, "step": 630 }, { "epoch": 0.13, "learning_rate": 1.9737597375973762e-05, "loss": 0.4895, "step": 640 }, { "epoch": 0.13, "learning_rate": 1.973349733497335e-05, "loss": 0.4869, "step": 650 }, { "epoch": 0.14, "learning_rate": 1.972939729397294e-05, "loss": 0.3975, "step": 660 }, { "epoch": 0.14, "learning_rate": 1.972529725297253e-05, "loss": 0.4899, "step": 670 }, { "epoch": 0.14, "learning_rate": 1.972119721197212e-05, "loss": 0.5121, "step": 680 }, { "epoch": 0.14, "learning_rate": 1.971709717097171e-05, "loss": 0.4149, "step": 690 }, { "epoch": 0.14, "learning_rate": 1.97129971299713e-05, "loss": 0.4827, "step": 700 }, { "epoch": 0.15, "learning_rate": 1.970889708897089e-05, "loss": 0.5512, "step": 710 }, { "epoch": 0.15, "learning_rate": 1.970479704797048e-05, "loss": 0.3603, "step": 720 }, { "epoch": 0.15, "learning_rate": 1.9700697006970073e-05, "loss": 0.4038, "step": 730 }, { "epoch": 0.15, "learning_rate": 1.969659696596966e-05, "loss": 0.3994, "step": 740 }, { "epoch": 0.15, "learning_rate": 1.9692496924969252e-05, "loss": 0.3928, "step": 750 }, { "epoch": 0.16, "learning_rate": 1.968839688396884e-05, "loss": 0.5249, "step": 760 }, { "epoch": 0.16, "learning_rate": 1.9684296842968432e-05, "loss": 0.4877, "step": 770 }, { "epoch": 0.16, "learning_rate": 1.968019680196802e-05, "loss": 0.4392, "step": 780 }, { "epoch": 0.16, "learning_rate": 1.9676096760967612e-05, "loss": 0.3942, "step": 790 }, { "epoch": 0.16, "learning_rate": 1.96719967199672e-05, "loss": 0.452, "step": 800 }, { "epoch": 0.17, "learning_rate": 1.9667896678966792e-05, "loss": 0.4028, "step": 810 }, { "epoch": 0.17, "learning_rate": 1.966379663796638e-05, "loss": 0.4262, "step": 820 }, { "epoch": 0.17, "learning_rate": 1.965969659696597e-05, "loss": 0.5593, "step": 830 }, { "epoch": 0.17, "learning_rate": 1.9655596555965563e-05, "loss": 0.4746, "step": 840 }, { "epoch": 0.17, "learning_rate": 1.965149651496515e-05, "loss": 0.3833, "step": 850 }, { "epoch": 0.18, "learning_rate": 1.9647396473964743e-05, "loss": 0.4843, "step": 860 }, { "epoch": 0.18, "learning_rate": 1.964329643296433e-05, "loss": 0.4538, "step": 870 }, { "epoch": 0.18, "learning_rate": 1.9639196391963923e-05, "loss": 0.4718, "step": 880 }, { "epoch": 0.18, "learning_rate": 1.963509635096351e-05, "loss": 0.479, "step": 890 }, { "epoch": 0.18, "learning_rate": 1.9630996309963103e-05, "loss": 0.3487, "step": 900 }, { "epoch": 0.19, "learning_rate": 1.962689626896269e-05, "loss": 0.3925, "step": 910 }, { "epoch": 0.19, "learning_rate": 1.9622796227962282e-05, "loss": 0.4928, "step": 920 }, { "epoch": 0.19, "learning_rate": 1.961869618696187e-05, "loss": 0.4217, "step": 930 }, { "epoch": 0.19, "learning_rate": 1.9614596145961462e-05, "loss": 0.3855, "step": 940 }, { "epoch": 0.19, "learning_rate": 1.9610496104961054e-05, "loss": 0.3871, "step": 950 }, { "epoch": 0.2, "learning_rate": 1.9606396063960642e-05, "loss": 0.5699, "step": 960 }, { "epoch": 0.2, "learning_rate": 1.9602296022960233e-05, "loss": 0.4557, "step": 970 }, { "epoch": 0.2, "learning_rate": 1.9598195981959822e-05, "loss": 0.4698, "step": 980 }, { "epoch": 0.2, "learning_rate": 1.9594095940959413e-05, "loss": 0.5331, "step": 990 }, { "epoch": 0.21, "learning_rate": 1.9589995899959e-05, "loss": 0.5127, "step": 1000 }, { "epoch": 0.21, "learning_rate": 1.958589585895859e-05, "loss": 0.419, "step": 1010 }, { "epoch": 0.21, "learning_rate": 1.958179581795818e-05, "loss": 0.4543, "step": 1020 }, { "epoch": 0.21, "learning_rate": 1.957769577695777e-05, "loss": 0.3658, "step": 1030 }, { "epoch": 0.21, "learning_rate": 1.957359573595736e-05, "loss": 0.4056, "step": 1040 }, { "epoch": 0.22, "learning_rate": 1.956949569495695e-05, "loss": 0.4788, "step": 1050 }, { "epoch": 0.22, "learning_rate": 1.956539565395654e-05, "loss": 0.4935, "step": 1060 }, { "epoch": 0.22, "learning_rate": 1.956129561295613e-05, "loss": 0.5355, "step": 1070 }, { "epoch": 0.22, "learning_rate": 1.955719557195572e-05, "loss": 0.3749, "step": 1080 }, { "epoch": 0.22, "learning_rate": 1.955309553095531e-05, "loss": 0.415, "step": 1090 }, { "epoch": 0.23, "learning_rate": 1.95489954899549e-05, "loss": 0.4198, "step": 1100 }, { "epoch": 0.23, "learning_rate": 1.9544895448954492e-05, "loss": 0.5377, "step": 1110 }, { "epoch": 0.23, "learning_rate": 1.954079540795408e-05, "loss": 0.5349, "step": 1120 }, { "epoch": 0.23, "learning_rate": 1.9536695366953672e-05, "loss": 0.4057, "step": 1130 }, { "epoch": 0.23, "learning_rate": 1.953259532595326e-05, "loss": 0.4291, "step": 1140 }, { "epoch": 0.24, "learning_rate": 1.952849528495285e-05, "loss": 0.4422, "step": 1150 }, { "epoch": 0.24, "learning_rate": 1.952439524395244e-05, "loss": 0.4176, "step": 1160 }, { "epoch": 0.24, "learning_rate": 1.952029520295203e-05, "loss": 0.3885, "step": 1170 }, { "epoch": 0.24, "learning_rate": 1.951619516195162e-05, "loss": 0.4426, "step": 1180 }, { "epoch": 0.24, "learning_rate": 1.951209512095121e-05, "loss": 0.3789, "step": 1190 }, { "epoch": 0.25, "learning_rate": 1.95079950799508e-05, "loss": 0.4486, "step": 1200 }, { "epoch": 0.25, "learning_rate": 1.950389503895039e-05, "loss": 0.388, "step": 1210 }, { "epoch": 0.25, "learning_rate": 1.9499794997949983e-05, "loss": 0.4251, "step": 1220 }, { "epoch": 0.25, "learning_rate": 1.949569495694957e-05, "loss": 0.384, "step": 1230 }, { "epoch": 0.25, "learning_rate": 1.9491594915949162e-05, "loss": 0.5137, "step": 1240 }, { "epoch": 0.26, "learning_rate": 1.948749487494875e-05, "loss": 0.3817, "step": 1250 }, { "epoch": 0.26, "learning_rate": 1.9483394833948342e-05, "loss": 0.4796, "step": 1260 }, { "epoch": 0.26, "learning_rate": 1.947929479294793e-05, "loss": 0.3736, "step": 1270 }, { "epoch": 0.26, "learning_rate": 1.9475194751947522e-05, "loss": 0.4271, "step": 1280 }, { "epoch": 0.26, "learning_rate": 1.947109471094711e-05, "loss": 0.4886, "step": 1290 }, { "epoch": 0.27, "learning_rate": 1.9466994669946702e-05, "loss": 0.4524, "step": 1300 }, { "epoch": 0.27, "learning_rate": 1.946289462894629e-05, "loss": 0.3401, "step": 1310 }, { "epoch": 0.27, "learning_rate": 1.945879458794588e-05, "loss": 0.4076, "step": 1320 }, { "epoch": 0.27, "learning_rate": 1.9454694546945473e-05, "loss": 0.3766, "step": 1330 }, { "epoch": 0.27, "learning_rate": 1.945059450594506e-05, "loss": 0.4107, "step": 1340 }, { "epoch": 0.28, "learning_rate": 1.9446494464944653e-05, "loss": 0.3432, "step": 1350 }, { "epoch": 0.28, "learning_rate": 1.944239442394424e-05, "loss": 0.413, "step": 1360 }, { "epoch": 0.28, "learning_rate": 1.9438294382943833e-05, "loss": 0.4329, "step": 1370 }, { "epoch": 0.28, "learning_rate": 1.943419434194342e-05, "loss": 0.375, "step": 1380 }, { "epoch": 0.28, "learning_rate": 1.9430094300943013e-05, "loss": 0.371, "step": 1390 }, { "epoch": 0.29, "learning_rate": 1.94259942599426e-05, "loss": 0.4312, "step": 1400 }, { "epoch": 0.29, "learning_rate": 1.942189421894219e-05, "loss": 0.3903, "step": 1410 }, { "epoch": 0.29, "learning_rate": 1.941779417794178e-05, "loss": 0.3033, "step": 1420 }, { "epoch": 0.29, "learning_rate": 1.941369413694137e-05, "loss": 0.3614, "step": 1430 }, { "epoch": 0.3, "learning_rate": 1.940959409594096e-05, "loss": 0.4235, "step": 1440 }, { "epoch": 0.3, "learning_rate": 1.940549405494055e-05, "loss": 0.3408, "step": 1450 }, { "epoch": 0.3, "learning_rate": 1.940139401394014e-05, "loss": 0.498, "step": 1460 }, { "epoch": 0.3, "learning_rate": 1.9397293972939732e-05, "loss": 0.3689, "step": 1470 }, { "epoch": 0.3, "learning_rate": 1.939319393193932e-05, "loss": 0.4651, "step": 1480 }, { "epoch": 0.31, "learning_rate": 1.938909389093891e-05, "loss": 0.3589, "step": 1490 }, { "epoch": 0.31, "learning_rate": 1.93849938499385e-05, "loss": 0.4562, "step": 1500 }, { "epoch": 0.31, "learning_rate": 1.938089380893809e-05, "loss": 0.4628, "step": 1510 }, { "epoch": 0.31, "learning_rate": 1.937679376793768e-05, "loss": 0.3542, "step": 1520 }, { "epoch": 0.31, "learning_rate": 1.937269372693727e-05, "loss": 0.3694, "step": 1530 }, { "epoch": 0.32, "learning_rate": 1.936859368593686e-05, "loss": 0.3261, "step": 1540 }, { "epoch": 0.32, "learning_rate": 1.936449364493645e-05, "loss": 0.5337, "step": 1550 }, { "epoch": 0.32, "learning_rate": 1.936039360393604e-05, "loss": 0.4264, "step": 1560 }, { "epoch": 0.32, "learning_rate": 1.935629356293563e-05, "loss": 0.4346, "step": 1570 }, { "epoch": 0.32, "learning_rate": 1.9352193521935222e-05, "loss": 0.3785, "step": 1580 }, { "epoch": 0.33, "learning_rate": 1.934809348093481e-05, "loss": 0.3452, "step": 1590 }, { "epoch": 0.33, "learning_rate": 1.9343993439934402e-05, "loss": 0.431, "step": 1600 }, { "epoch": 0.33, "learning_rate": 1.933989339893399e-05, "loss": 0.4159, "step": 1610 }, { "epoch": 0.33, "learning_rate": 1.9335793357933582e-05, "loss": 0.3585, "step": 1620 }, { "epoch": 0.33, "learning_rate": 1.933169331693317e-05, "loss": 0.4739, "step": 1630 }, { "epoch": 0.34, "learning_rate": 1.932759327593276e-05, "loss": 0.3671, "step": 1640 }, { "epoch": 0.34, "learning_rate": 1.932349323493235e-05, "loss": 0.3701, "step": 1650 }, { "epoch": 0.34, "learning_rate": 1.931939319393194e-05, "loss": 0.4008, "step": 1660 }, { "epoch": 0.34, "learning_rate": 1.931529315293153e-05, "loss": 0.31, "step": 1670 }, { "epoch": 0.34, "learning_rate": 1.931119311193112e-05, "loss": 0.3813, "step": 1680 }, { "epoch": 0.35, "learning_rate": 1.9307093070930713e-05, "loss": 0.4187, "step": 1690 }, { "epoch": 0.35, "learning_rate": 1.93029930299303e-05, "loss": 0.4278, "step": 1700 }, { "epoch": 0.35, "learning_rate": 1.9298892988929893e-05, "loss": 0.3975, "step": 1710 }, { "epoch": 0.35, "learning_rate": 1.929479294792948e-05, "loss": 0.3793, "step": 1720 }, { "epoch": 0.35, "learning_rate": 1.9290692906929072e-05, "loss": 0.3289, "step": 1730 }, { "epoch": 0.36, "learning_rate": 1.928659286592866e-05, "loss": 0.4047, "step": 1740 }, { "epoch": 0.36, "learning_rate": 1.9282492824928252e-05, "loss": 0.4442, "step": 1750 }, { "epoch": 0.36, "learning_rate": 1.927839278392784e-05, "loss": 0.3982, "step": 1760 }, { "epoch": 0.36, "learning_rate": 1.9274292742927432e-05, "loss": 0.3918, "step": 1770 }, { "epoch": 0.36, "learning_rate": 1.927019270192702e-05, "loss": 0.3104, "step": 1780 }, { "epoch": 0.37, "learning_rate": 1.9266092660926612e-05, "loss": 0.2937, "step": 1790 }, { "epoch": 0.37, "learning_rate": 1.92619926199262e-05, "loss": 0.5006, "step": 1800 }, { "epoch": 0.37, "learning_rate": 1.9257892578925788e-05, "loss": 0.35, "step": 1810 }, { "epoch": 0.37, "learning_rate": 1.925379253792538e-05, "loss": 0.3959, "step": 1820 }, { "epoch": 0.38, "learning_rate": 1.924969249692497e-05, "loss": 0.3601, "step": 1830 }, { "epoch": 0.38, "learning_rate": 1.924559245592456e-05, "loss": 0.4325, "step": 1840 }, { "epoch": 0.38, "learning_rate": 1.924149241492415e-05, "loss": 0.4425, "step": 1850 }, { "epoch": 0.38, "learning_rate": 1.923739237392374e-05, "loss": 0.3957, "step": 1860 }, { "epoch": 0.38, "learning_rate": 1.923329233292333e-05, "loss": 0.3811, "step": 1870 }, { "epoch": 0.39, "learning_rate": 1.922919229192292e-05, "loss": 0.454, "step": 1880 }, { "epoch": 0.39, "learning_rate": 1.922509225092251e-05, "loss": 0.4165, "step": 1890 }, { "epoch": 0.39, "learning_rate": 1.92209922099221e-05, "loss": 0.4604, "step": 1900 }, { "epoch": 0.39, "learning_rate": 1.921689216892169e-05, "loss": 0.4977, "step": 1910 }, { "epoch": 0.39, "learning_rate": 1.921279212792128e-05, "loss": 0.3596, "step": 1920 }, { "epoch": 0.4, "learning_rate": 1.920869208692087e-05, "loss": 0.4288, "step": 1930 }, { "epoch": 0.4, "learning_rate": 1.9204592045920462e-05, "loss": 0.406, "step": 1940 }, { "epoch": 0.4, "learning_rate": 1.920049200492005e-05, "loss": 0.3541, "step": 1950 }, { "epoch": 0.4, "learning_rate": 1.9196391963919642e-05, "loss": 0.3518, "step": 1960 }, { "epoch": 0.4, "learning_rate": 1.919229192291923e-05, "loss": 0.3384, "step": 1970 }, { "epoch": 0.41, "learning_rate": 1.918819188191882e-05, "loss": 0.3992, "step": 1980 }, { "epoch": 0.41, "learning_rate": 1.918409184091841e-05, "loss": 0.3795, "step": 1990 }, { "epoch": 0.41, "learning_rate": 1.9179991799918e-05, "loss": 0.3343, "step": 2000 }, { "epoch": 0.41, "learning_rate": 1.917589175891759e-05, "loss": 0.3388, "step": 2010 }, { "epoch": 0.41, "learning_rate": 1.917179171791718e-05, "loss": 0.4125, "step": 2020 }, { "epoch": 0.42, "learning_rate": 1.916769167691677e-05, "loss": 0.3592, "step": 2030 }, { "epoch": 0.42, "learning_rate": 1.916359163591636e-05, "loss": 0.4324, "step": 2040 }, { "epoch": 0.42, "learning_rate": 1.9159491594915953e-05, "loss": 0.3951, "step": 2050 }, { "epoch": 0.42, "learning_rate": 1.915539155391554e-05, "loss": 0.373, "step": 2060 }, { "epoch": 0.42, "learning_rate": 1.9151291512915132e-05, "loss": 0.4003, "step": 2070 }, { "epoch": 0.43, "learning_rate": 1.914719147191472e-05, "loss": 0.2606, "step": 2080 }, { "epoch": 0.43, "learning_rate": 1.9143091430914312e-05, "loss": 0.4446, "step": 2090 }, { "epoch": 0.43, "learning_rate": 1.91389913899139e-05, "loss": 0.3145, "step": 2100 }, { "epoch": 0.43, "learning_rate": 1.9134891348913492e-05, "loss": 0.3647, "step": 2110 }, { "epoch": 0.43, "learning_rate": 1.913079130791308e-05, "loss": 0.2904, "step": 2120 }, { "epoch": 0.44, "learning_rate": 1.9126691266912672e-05, "loss": 0.328, "step": 2130 }, { "epoch": 0.44, "learning_rate": 1.912259122591226e-05, "loss": 0.4403, "step": 2140 }, { "epoch": 0.44, "learning_rate": 1.911849118491185e-05, "loss": 0.3826, "step": 2150 }, { "epoch": 0.44, "learning_rate": 1.9114391143911443e-05, "loss": 0.4112, "step": 2160 }, { "epoch": 0.44, "learning_rate": 1.911029110291103e-05, "loss": 0.4231, "step": 2170 }, { "epoch": 0.45, "learning_rate": 1.9106191061910623e-05, "loss": 0.4089, "step": 2180 }, { "epoch": 0.45, "learning_rate": 1.910209102091021e-05, "loss": 0.3894, "step": 2190 }, { "epoch": 0.45, "learning_rate": 1.90979909799098e-05, "loss": 0.3441, "step": 2200 }, { "epoch": 0.45, "learning_rate": 1.909389093890939e-05, "loss": 0.3391, "step": 2210 }, { "epoch": 0.46, "learning_rate": 1.908979089790898e-05, "loss": 0.355, "step": 2220 }, { "epoch": 0.46, "learning_rate": 1.908569085690857e-05, "loss": 0.488, "step": 2230 }, { "epoch": 0.46, "learning_rate": 1.908159081590816e-05, "loss": 0.363, "step": 2240 }, { "epoch": 0.46, "learning_rate": 1.907749077490775e-05, "loss": 0.4267, "step": 2250 }, { "epoch": 0.46, "learning_rate": 1.907339073390734e-05, "loss": 0.2989, "step": 2260 }, { "epoch": 0.47, "learning_rate": 1.906929069290693e-05, "loss": 0.329, "step": 2270 }, { "epoch": 0.47, "learning_rate": 1.906519065190652e-05, "loss": 0.4069, "step": 2280 }, { "epoch": 0.47, "learning_rate": 1.906109061090611e-05, "loss": 0.4457, "step": 2290 }, { "epoch": 0.47, "learning_rate": 1.90569905699057e-05, "loss": 0.3793, "step": 2300 }, { "epoch": 0.47, "learning_rate": 1.905289052890529e-05, "loss": 0.4791, "step": 2310 }, { "epoch": 0.48, "learning_rate": 1.904879048790488e-05, "loss": 0.4547, "step": 2320 }, { "epoch": 0.48, "learning_rate": 1.904469044690447e-05, "loss": 0.3992, "step": 2330 }, { "epoch": 0.48, "learning_rate": 1.904059040590406e-05, "loss": 0.3682, "step": 2340 }, { "epoch": 0.48, "learning_rate": 1.903649036490365e-05, "loss": 0.3343, "step": 2350 }, { "epoch": 0.48, "learning_rate": 1.903239032390324e-05, "loss": 0.4526, "step": 2360 }, { "epoch": 0.49, "learning_rate": 1.902829028290283e-05, "loss": 0.3218, "step": 2370 }, { "epoch": 0.49, "learning_rate": 1.902419024190242e-05, "loss": 0.388, "step": 2380 }, { "epoch": 0.49, "learning_rate": 1.902009020090201e-05, "loss": 0.3778, "step": 2390 }, { "epoch": 0.49, "learning_rate": 1.90159901599016e-05, "loss": 0.4118, "step": 2400 }, { "epoch": 0.49, "learning_rate": 1.9011890118901192e-05, "loss": 0.3844, "step": 2410 }, { "epoch": 0.5, "learning_rate": 1.900779007790078e-05, "loss": 0.39, "step": 2420 }, { "epoch": 0.5, "learning_rate": 1.9003690036900372e-05, "loss": 0.3641, "step": 2430 }, { "epoch": 0.5, "learning_rate": 1.899958999589996e-05, "loss": 0.4327, "step": 2440 }, { "epoch": 0.5, "learning_rate": 1.8995489954899552e-05, "loss": 0.3913, "step": 2450 }, { "epoch": 0.5, "learning_rate": 1.899138991389914e-05, "loss": 0.4313, "step": 2460 }, { "epoch": 0.51, "learning_rate": 1.898728987289873e-05, "loss": 0.4, "step": 2470 }, { "epoch": 0.51, "learning_rate": 1.898318983189832e-05, "loss": 0.3025, "step": 2480 }, { "epoch": 0.51, "learning_rate": 1.897908979089791e-05, "loss": 0.3735, "step": 2490 }, { "epoch": 0.51, "learning_rate": 1.89749897498975e-05, "loss": 0.2513, "step": 2500 }, { "epoch": 0.51, "learning_rate": 1.897088970889709e-05, "loss": 0.3939, "step": 2510 }, { "epoch": 0.52, "learning_rate": 1.8966789667896683e-05, "loss": 0.3392, "step": 2520 }, { "epoch": 0.52, "learning_rate": 1.896268962689627e-05, "loss": 0.4682, "step": 2530 }, { "epoch": 0.52, "learning_rate": 1.8958589585895863e-05, "loss": 0.3597, "step": 2540 }, { "epoch": 0.52, "learning_rate": 1.895448954489545e-05, "loss": 0.3051, "step": 2550 }, { "epoch": 0.52, "learning_rate": 1.8950389503895042e-05, "loss": 0.3468, "step": 2560 }, { "epoch": 0.53, "learning_rate": 1.894628946289463e-05, "loss": 0.3561, "step": 2570 }, { "epoch": 0.53, "learning_rate": 1.8942189421894222e-05, "loss": 0.2583, "step": 2580 }, { "epoch": 0.53, "learning_rate": 1.893808938089381e-05, "loss": 0.4674, "step": 2590 }, { "epoch": 0.53, "learning_rate": 1.89339893398934e-05, "loss": 0.3904, "step": 2600 }, { "epoch": 0.54, "learning_rate": 1.892988929889299e-05, "loss": 0.3552, "step": 2610 }, { "epoch": 0.54, "learning_rate": 1.892578925789258e-05, "loss": 0.3013, "step": 2620 }, { "epoch": 0.54, "learning_rate": 1.892168921689217e-05, "loss": 0.2935, "step": 2630 }, { "epoch": 0.54, "learning_rate": 1.8917589175891758e-05, "loss": 0.3028, "step": 2640 }, { "epoch": 0.54, "learning_rate": 1.891348913489135e-05, "loss": 0.41, "step": 2650 }, { "epoch": 0.55, "learning_rate": 1.890938909389094e-05, "loss": 0.3567, "step": 2660 }, { "epoch": 0.55, "learning_rate": 1.890528905289053e-05, "loss": 0.3715, "step": 2670 }, { "epoch": 0.55, "learning_rate": 1.890118901189012e-05, "loss": 0.3209, "step": 2680 }, { "epoch": 0.55, "learning_rate": 1.889708897088971e-05, "loss": 0.4302, "step": 2690 }, { "epoch": 0.55, "learning_rate": 1.88929889298893e-05, "loss": 0.3665, "step": 2700 }, { "epoch": 0.56, "learning_rate": 1.888888888888889e-05, "loss": 0.3547, "step": 2710 }, { "epoch": 0.56, "learning_rate": 1.888478884788848e-05, "loss": 0.3081, "step": 2720 }, { "epoch": 0.56, "learning_rate": 1.888068880688807e-05, "loss": 0.3733, "step": 2730 }, { "epoch": 0.56, "learning_rate": 1.887658876588766e-05, "loss": 0.4772, "step": 2740 }, { "epoch": 0.56, "learning_rate": 1.887248872488725e-05, "loss": 0.3243, "step": 2750 }, { "epoch": 0.57, "learning_rate": 1.886838868388684e-05, "loss": 0.3465, "step": 2760 }, { "epoch": 0.57, "learning_rate": 1.8864288642886432e-05, "loss": 0.3157, "step": 2770 }, { "epoch": 0.57, "learning_rate": 1.886018860188602e-05, "loss": 0.4832, "step": 2780 }, { "epoch": 0.57, "learning_rate": 1.885608856088561e-05, "loss": 0.3207, "step": 2790 }, { "epoch": 0.57, "learning_rate": 1.88519885198852e-05, "loss": 0.403, "step": 2800 }, { "epoch": 0.58, "learning_rate": 1.884788847888479e-05, "loss": 0.3779, "step": 2810 }, { "epoch": 0.58, "learning_rate": 1.884378843788438e-05, "loss": 0.3478, "step": 2820 }, { "epoch": 0.58, "learning_rate": 1.883968839688397e-05, "loss": 0.3809, "step": 2830 }, { "epoch": 0.58, "learning_rate": 1.883558835588356e-05, "loss": 0.3046, "step": 2840 }, { "epoch": 0.58, "learning_rate": 1.883148831488315e-05, "loss": 0.4257, "step": 2850 }, { "epoch": 0.59, "learning_rate": 1.882738827388274e-05, "loss": 0.3584, "step": 2860 }, { "epoch": 0.59, "learning_rate": 1.882328823288233e-05, "loss": 0.3601, "step": 2870 }, { "epoch": 0.59, "learning_rate": 1.8819188191881922e-05, "loss": 0.2878, "step": 2880 }, { "epoch": 0.59, "learning_rate": 1.881508815088151e-05, "loss": 0.2665, "step": 2890 }, { "epoch": 0.59, "learning_rate": 1.8810988109881102e-05, "loss": 0.4221, "step": 2900 }, { "epoch": 0.6, "learning_rate": 1.880688806888069e-05, "loss": 0.2811, "step": 2910 }, { "epoch": 0.6, "learning_rate": 1.8802788027880282e-05, "loss": 0.4015, "step": 2920 }, { "epoch": 0.6, "learning_rate": 1.879868798687987e-05, "loss": 0.394, "step": 2930 }, { "epoch": 0.6, "learning_rate": 1.8794587945879462e-05, "loss": 0.442, "step": 2940 }, { "epoch": 0.6, "learning_rate": 1.879048790487905e-05, "loss": 0.4203, "step": 2950 }, { "epoch": 0.61, "learning_rate": 1.878638786387864e-05, "loss": 0.332, "step": 2960 }, { "epoch": 0.61, "learning_rate": 1.878228782287823e-05, "loss": 0.3904, "step": 2970 }, { "epoch": 0.61, "learning_rate": 1.877818778187782e-05, "loss": 0.3535, "step": 2980 }, { "epoch": 0.61, "learning_rate": 1.8774087740877413e-05, "loss": 0.3322, "step": 2990 }, { "epoch": 0.62, "learning_rate": 1.8769987699876998e-05, "loss": 0.3849, "step": 3000 }, { "epoch": 0.62, "learning_rate": 1.876588765887659e-05, "loss": 0.3877, "step": 3010 }, { "epoch": 0.62, "learning_rate": 1.8761787617876178e-05, "loss": 0.3631, "step": 3020 }, { "epoch": 0.62, "learning_rate": 1.875768757687577e-05, "loss": 0.3865, "step": 3030 }, { "epoch": 0.62, "learning_rate": 1.875358753587536e-05, "loss": 0.2927, "step": 3040 }, { "epoch": 0.63, "learning_rate": 1.874948749487495e-05, "loss": 0.3235, "step": 3050 }, { "epoch": 0.63, "learning_rate": 1.874538745387454e-05, "loss": 0.2994, "step": 3060 }, { "epoch": 0.63, "learning_rate": 1.874128741287413e-05, "loss": 0.3737, "step": 3070 }, { "epoch": 0.63, "learning_rate": 1.873718737187372e-05, "loss": 0.2974, "step": 3080 }, { "epoch": 0.63, "learning_rate": 1.873308733087331e-05, "loss": 0.3977, "step": 3090 }, { "epoch": 0.64, "learning_rate": 1.87289872898729e-05, "loss": 0.3058, "step": 3100 }, { "epoch": 0.64, "learning_rate": 1.872488724887249e-05, "loss": 0.3365, "step": 3110 }, { "epoch": 0.64, "learning_rate": 1.872078720787208e-05, "loss": 0.4264, "step": 3120 }, { "epoch": 0.64, "learning_rate": 1.8716687166871668e-05, "loss": 0.3617, "step": 3130 }, { "epoch": 0.64, "learning_rate": 1.871258712587126e-05, "loss": 0.2847, "step": 3140 }, { "epoch": 0.65, "learning_rate": 1.870848708487085e-05, "loss": 0.3311, "step": 3150 }, { "epoch": 0.65, "learning_rate": 1.870438704387044e-05, "loss": 0.4281, "step": 3160 }, { "epoch": 0.65, "learning_rate": 1.870028700287003e-05, "loss": 0.3363, "step": 3170 }, { "epoch": 0.65, "learning_rate": 1.869618696186962e-05, "loss": 0.389, "step": 3180 }, { "epoch": 0.65, "learning_rate": 1.869208692086921e-05, "loss": 0.2844, "step": 3190 }, { "epoch": 0.66, "learning_rate": 1.86879868798688e-05, "loss": 0.3841, "step": 3200 }, { "epoch": 0.66, "learning_rate": 1.868388683886839e-05, "loss": 0.4762, "step": 3210 }, { "epoch": 0.66, "learning_rate": 1.867978679786798e-05, "loss": 0.2833, "step": 3220 }, { "epoch": 0.66, "learning_rate": 1.867568675686757e-05, "loss": 0.3782, "step": 3230 }, { "epoch": 0.66, "learning_rate": 1.867158671586716e-05, "loss": 0.3348, "step": 3240 }, { "epoch": 0.67, "learning_rate": 1.866748667486675e-05, "loss": 0.4847, "step": 3250 }, { "epoch": 0.67, "learning_rate": 1.8663386633866342e-05, "loss": 0.2847, "step": 3260 }, { "epoch": 0.67, "learning_rate": 1.865928659286593e-05, "loss": 0.3441, "step": 3270 }, { "epoch": 0.67, "learning_rate": 1.8655186551865522e-05, "loss": 0.4207, "step": 3280 }, { "epoch": 0.67, "learning_rate": 1.865108651086511e-05, "loss": 0.3452, "step": 3290 }, { "epoch": 0.68, "learning_rate": 1.86469864698647e-05, "loss": 0.291, "step": 3300 }, { "epoch": 0.68, "learning_rate": 1.864288642886429e-05, "loss": 0.459, "step": 3310 }, { "epoch": 0.68, "learning_rate": 1.863878638786388e-05, "loss": 0.3293, "step": 3320 }, { "epoch": 0.68, "learning_rate": 1.863468634686347e-05, "loss": 0.3378, "step": 3330 }, { "epoch": 0.68, "learning_rate": 1.863058630586306e-05, "loss": 0.315, "step": 3340 }, { "epoch": 0.69, "learning_rate": 1.862648626486265e-05, "loss": 0.2893, "step": 3350 }, { "epoch": 0.69, "learning_rate": 1.862238622386224e-05, "loss": 0.3943, "step": 3360 }, { "epoch": 0.69, "learning_rate": 1.8618286182861832e-05, "loss": 0.352, "step": 3370 }, { "epoch": 0.69, "learning_rate": 1.861418614186142e-05, "loss": 0.3584, "step": 3380 }, { "epoch": 0.69, "learning_rate": 1.8610086100861012e-05, "loss": 0.3142, "step": 3390 }, { "epoch": 0.7, "learning_rate": 1.86059860598606e-05, "loss": 0.3015, "step": 3400 }, { "epoch": 0.7, "learning_rate": 1.860188601886019e-05, "loss": 0.288, "step": 3410 }, { "epoch": 0.7, "learning_rate": 1.859778597785978e-05, "loss": 0.3055, "step": 3420 }, { "epoch": 0.7, "learning_rate": 1.859368593685937e-05, "loss": 0.292, "step": 3430 }, { "epoch": 0.71, "learning_rate": 1.858958589585896e-05, "loss": 0.3304, "step": 3440 }, { "epoch": 0.71, "learning_rate": 1.8585485854858548e-05, "loss": 0.2711, "step": 3450 }, { "epoch": 0.71, "learning_rate": 1.858138581385814e-05, "loss": 0.3784, "step": 3460 }, { "epoch": 0.71, "learning_rate": 1.8577285772857728e-05, "loss": 0.3167, "step": 3470 }, { "epoch": 0.71, "learning_rate": 1.857318573185732e-05, "loss": 0.2965, "step": 3480 }, { "epoch": 0.72, "learning_rate": 1.8569085690856908e-05, "loss": 0.3876, "step": 3490 }, { "epoch": 0.72, "learning_rate": 1.85649856498565e-05, "loss": 0.3658, "step": 3500 }, { "epoch": 0.72, "learning_rate": 1.856088560885609e-05, "loss": 0.293, "step": 3510 }, { "epoch": 0.72, "learning_rate": 1.855678556785568e-05, "loss": 0.3525, "step": 3520 }, { "epoch": 0.72, "learning_rate": 1.855268552685527e-05, "loss": 0.2794, "step": 3530 }, { "epoch": 0.73, "learning_rate": 1.854858548585486e-05, "loss": 0.3475, "step": 3540 }, { "epoch": 0.73, "learning_rate": 1.854448544485445e-05, "loss": 0.344, "step": 3550 }, { "epoch": 0.73, "learning_rate": 1.854038540385404e-05, "loss": 0.3576, "step": 3560 }, { "epoch": 0.73, "learning_rate": 1.853628536285363e-05, "loss": 0.3306, "step": 3570 }, { "epoch": 0.73, "learning_rate": 1.853218532185322e-05, "loss": 0.3678, "step": 3580 }, { "epoch": 0.74, "learning_rate": 1.852808528085281e-05, "loss": 0.3443, "step": 3590 }, { "epoch": 0.74, "learning_rate": 1.85239852398524e-05, "loss": 0.3886, "step": 3600 }, { "epoch": 0.74, "learning_rate": 1.851988519885199e-05, "loss": 0.4374, "step": 3610 }, { "epoch": 0.74, "learning_rate": 1.851578515785158e-05, "loss": 0.254, "step": 3620 }, { "epoch": 0.74, "learning_rate": 1.851168511685117e-05, "loss": 0.2735, "step": 3630 }, { "epoch": 0.75, "learning_rate": 1.850758507585076e-05, "loss": 0.3738, "step": 3640 }, { "epoch": 0.75, "learning_rate": 1.850348503485035e-05, "loss": 0.3978, "step": 3650 }, { "epoch": 0.75, "learning_rate": 1.849938499384994e-05, "loss": 0.2642, "step": 3660 }, { "epoch": 0.75, "learning_rate": 1.849528495284953e-05, "loss": 0.3331, "step": 3670 }, { "epoch": 0.75, "learning_rate": 1.849118491184912e-05, "loss": 0.3334, "step": 3680 }, { "epoch": 0.76, "learning_rate": 1.848708487084871e-05, "loss": 0.3869, "step": 3690 }, { "epoch": 0.76, "learning_rate": 1.84829848298483e-05, "loss": 0.3432, "step": 3700 }, { "epoch": 0.76, "learning_rate": 1.847888478884789e-05, "loss": 0.3247, "step": 3710 }, { "epoch": 0.76, "learning_rate": 1.847478474784748e-05, "loss": 0.414, "step": 3720 }, { "epoch": 0.76, "learning_rate": 1.8470684706847072e-05, "loss": 0.3464, "step": 3730 }, { "epoch": 0.77, "learning_rate": 1.846658466584666e-05, "loss": 0.2769, "step": 3740 }, { "epoch": 0.77, "learning_rate": 1.8462484624846252e-05, "loss": 0.3754, "step": 3750 }, { "epoch": 0.77, "learning_rate": 1.845838458384584e-05, "loss": 0.4192, "step": 3760 }, { "epoch": 0.77, "learning_rate": 1.8454284542845432e-05, "loss": 0.408, "step": 3770 }, { "epoch": 0.77, "learning_rate": 1.845018450184502e-05, "loss": 0.3512, "step": 3780 }, { "epoch": 0.78, "learning_rate": 1.844608446084461e-05, "loss": 0.3656, "step": 3790 }, { "epoch": 0.78, "learning_rate": 1.84419844198442e-05, "loss": 0.3509, "step": 3800 }, { "epoch": 0.78, "learning_rate": 1.8437884378843788e-05, "loss": 0.3464, "step": 3810 }, { "epoch": 0.78, "learning_rate": 1.843378433784338e-05, "loss": 0.3245, "step": 3820 }, { "epoch": 0.79, "learning_rate": 1.8429684296842968e-05, "loss": 0.3572, "step": 3830 }, { "epoch": 0.79, "learning_rate": 1.842558425584256e-05, "loss": 0.3937, "step": 3840 }, { "epoch": 0.79, "learning_rate": 1.8421484214842148e-05, "loss": 0.3498, "step": 3850 }, { "epoch": 0.79, "learning_rate": 1.841738417384174e-05, "loss": 0.351, "step": 3860 }, { "epoch": 0.79, "learning_rate": 1.841328413284133e-05, "loss": 0.3201, "step": 3870 }, { "epoch": 0.8, "learning_rate": 1.840918409184092e-05, "loss": 0.4047, "step": 3880 }, { "epoch": 0.8, "learning_rate": 1.840508405084051e-05, "loss": 0.3068, "step": 3890 }, { "epoch": 0.8, "learning_rate": 1.84009840098401e-05, "loss": 0.3902, "step": 3900 }, { "epoch": 0.8, "learning_rate": 1.839688396883969e-05, "loss": 0.2978, "step": 3910 }, { "epoch": 0.8, "learning_rate": 1.839278392783928e-05, "loss": 0.4323, "step": 3920 }, { "epoch": 0.81, "learning_rate": 1.838868388683887e-05, "loss": 0.2951, "step": 3930 }, { "epoch": 0.81, "learning_rate": 1.8384583845838458e-05, "loss": 0.2599, "step": 3940 }, { "epoch": 0.81, "learning_rate": 1.838048380483805e-05, "loss": 0.3699, "step": 3950 }, { "epoch": 0.81, "learning_rate": 1.8376383763837638e-05, "loss": 0.2749, "step": 3960 }, { "epoch": 0.81, "learning_rate": 1.837228372283723e-05, "loss": 0.3461, "step": 3970 }, { "epoch": 0.82, "learning_rate": 1.836818368183682e-05, "loss": 0.3475, "step": 3980 }, { "epoch": 0.82, "learning_rate": 1.836408364083641e-05, "loss": 0.3171, "step": 3990 }, { "epoch": 0.82, "learning_rate": 1.8359983599836e-05, "loss": 0.324, "step": 4000 }, { "epoch": 0.82, "learning_rate": 1.835588355883559e-05, "loss": 0.3223, "step": 4010 }, { "epoch": 0.82, "learning_rate": 1.835178351783518e-05, "loss": 0.3894, "step": 4020 }, { "epoch": 0.83, "learning_rate": 1.834768347683477e-05, "loss": 0.3299, "step": 4030 }, { "epoch": 0.83, "learning_rate": 1.834358343583436e-05, "loss": 0.3812, "step": 4040 }, { "epoch": 0.83, "learning_rate": 1.833948339483395e-05, "loss": 0.2824, "step": 4050 }, { "epoch": 0.83, "learning_rate": 1.833538335383354e-05, "loss": 0.326, "step": 4060 }, { "epoch": 0.83, "learning_rate": 1.833128331283313e-05, "loss": 0.3923, "step": 4070 }, { "epoch": 0.84, "learning_rate": 1.832718327183272e-05, "loss": 0.2765, "step": 4080 }, { "epoch": 0.84, "learning_rate": 1.8323083230832312e-05, "loss": 0.3132, "step": 4090 }, { "epoch": 0.84, "learning_rate": 1.83189831898319e-05, "loss": 0.3454, "step": 4100 }, { "epoch": 0.84, "learning_rate": 1.831488314883149e-05, "loss": 0.348, "step": 4110 }, { "epoch": 0.84, "learning_rate": 1.831078310783108e-05, "loss": 0.3597, "step": 4120 }, { "epoch": 0.85, "learning_rate": 1.830668306683067e-05, "loss": 0.3299, "step": 4130 }, { "epoch": 0.85, "learning_rate": 1.830258302583026e-05, "loss": 0.3702, "step": 4140 }, { "epoch": 0.85, "learning_rate": 1.829848298482985e-05, "loss": 0.3299, "step": 4150 }, { "epoch": 0.85, "learning_rate": 1.829438294382944e-05, "loss": 0.3386, "step": 4160 }, { "epoch": 0.85, "learning_rate": 1.829028290282903e-05, "loss": 0.346, "step": 4170 }, { "epoch": 0.86, "learning_rate": 1.828618286182862e-05, "loss": 0.3859, "step": 4180 }, { "epoch": 0.86, "learning_rate": 1.828208282082821e-05, "loss": 0.2902, "step": 4190 }, { "epoch": 0.86, "learning_rate": 1.82779827798278e-05, "loss": 0.2623, "step": 4200 }, { "epoch": 0.86, "learning_rate": 1.8273882738827387e-05, "loss": 0.2224, "step": 4210 }, { "epoch": 0.87, "learning_rate": 1.826978269782698e-05, "loss": 0.3405, "step": 4220 }, { "epoch": 0.87, "learning_rate": 1.826568265682657e-05, "loss": 0.4493, "step": 4230 }, { "epoch": 0.87, "learning_rate": 1.826158261582616e-05, "loss": 0.5084, "step": 4240 }, { "epoch": 0.87, "learning_rate": 1.825748257482575e-05, "loss": 0.3786, "step": 4250 }, { "epoch": 0.87, "learning_rate": 1.825338253382534e-05, "loss": 0.3467, "step": 4260 }, { "epoch": 0.88, "learning_rate": 1.824928249282493e-05, "loss": 0.2461, "step": 4270 }, { "epoch": 0.88, "learning_rate": 1.8245182451824518e-05, "loss": 0.3894, "step": 4280 }, { "epoch": 0.88, "learning_rate": 1.824108241082411e-05, "loss": 0.3834, "step": 4290 }, { "epoch": 0.88, "learning_rate": 1.8236982369823698e-05, "loss": 0.2901, "step": 4300 }, { "epoch": 0.88, "learning_rate": 1.823288232882329e-05, "loss": 0.3063, "step": 4310 }, { "epoch": 0.89, "learning_rate": 1.8228782287822878e-05, "loss": 0.3667, "step": 4320 }, { "epoch": 0.89, "learning_rate": 1.822468224682247e-05, "loss": 0.3286, "step": 4330 }, { "epoch": 0.89, "learning_rate": 1.822058220582206e-05, "loss": 0.4017, "step": 4340 }, { "epoch": 0.89, "learning_rate": 1.821648216482165e-05, "loss": 0.4211, "step": 4350 }, { "epoch": 0.89, "learning_rate": 1.821238212382124e-05, "loss": 0.3602, "step": 4360 }, { "epoch": 0.9, "learning_rate": 1.820828208282083e-05, "loss": 0.2824, "step": 4370 }, { "epoch": 0.9, "learning_rate": 1.820418204182042e-05, "loss": 0.349, "step": 4380 }, { "epoch": 0.9, "learning_rate": 1.820008200082001e-05, "loss": 0.395, "step": 4390 }, { "epoch": 0.9, "learning_rate": 1.81959819598196e-05, "loss": 0.4579, "step": 4400 }, { "epoch": 0.9, "learning_rate": 1.819188191881919e-05, "loss": 0.3512, "step": 4410 }, { "epoch": 0.91, "learning_rate": 1.818778187781878e-05, "loss": 0.3371, "step": 4420 }, { "epoch": 0.91, "learning_rate": 1.818368183681837e-05, "loss": 0.4146, "step": 4430 }, { "epoch": 0.91, "learning_rate": 1.817958179581796e-05, "loss": 0.4092, "step": 4440 }, { "epoch": 0.91, "learning_rate": 1.817548175481755e-05, "loss": 0.4054, "step": 4450 }, { "epoch": 0.91, "learning_rate": 1.817138171381714e-05, "loss": 0.3423, "step": 4460 }, { "epoch": 0.92, "learning_rate": 1.816728167281673e-05, "loss": 0.3465, "step": 4470 }, { "epoch": 0.92, "learning_rate": 1.816318163181632e-05, "loss": 0.359, "step": 4480 }, { "epoch": 0.92, "learning_rate": 1.815908159081591e-05, "loss": 0.2925, "step": 4490 }, { "epoch": 0.92, "learning_rate": 1.81549815498155e-05, "loss": 0.3959, "step": 4500 }, { "epoch": 0.92, "learning_rate": 1.815088150881509e-05, "loss": 0.3031, "step": 4510 }, { "epoch": 0.93, "learning_rate": 1.814678146781468e-05, "loss": 0.4207, "step": 4520 }, { "epoch": 0.93, "learning_rate": 1.814268142681427e-05, "loss": 0.2117, "step": 4530 }, { "epoch": 0.93, "learning_rate": 1.813858138581386e-05, "loss": 0.3808, "step": 4540 }, { "epoch": 0.93, "learning_rate": 1.813448134481345e-05, "loss": 0.3372, "step": 4550 }, { "epoch": 0.93, "learning_rate": 1.8130381303813042e-05, "loss": 0.3473, "step": 4560 }, { "epoch": 0.94, "learning_rate": 1.812628126281263e-05, "loss": 0.3071, "step": 4570 }, { "epoch": 0.94, "learning_rate": 1.8122181221812222e-05, "loss": 0.2296, "step": 4580 }, { "epoch": 0.94, "learning_rate": 1.811808118081181e-05, "loss": 0.371, "step": 4590 }, { "epoch": 0.94, "learning_rate": 1.8113981139811398e-05, "loss": 0.2518, "step": 4600 }, { "epoch": 0.95, "learning_rate": 1.810988109881099e-05, "loss": 0.3031, "step": 4610 }, { "epoch": 0.95, "learning_rate": 1.8105781057810578e-05, "loss": 0.2891, "step": 4620 }, { "epoch": 0.95, "learning_rate": 1.810168101681017e-05, "loss": 0.3685, "step": 4630 }, { "epoch": 0.95, "learning_rate": 1.8097580975809758e-05, "loss": 0.3854, "step": 4640 }, { "epoch": 0.95, "learning_rate": 1.809348093480935e-05, "loss": 0.3065, "step": 4650 }, { "epoch": 0.96, "learning_rate": 1.8089380893808938e-05, "loss": 0.338, "step": 4660 }, { "epoch": 0.96, "learning_rate": 1.808528085280853e-05, "loss": 0.3672, "step": 4670 }, { "epoch": 0.96, "learning_rate": 1.8081180811808117e-05, "loss": 0.2906, "step": 4680 }, { "epoch": 0.96, "learning_rate": 1.807708077080771e-05, "loss": 0.388, "step": 4690 }, { "epoch": 0.96, "learning_rate": 1.80729807298073e-05, "loss": 0.4207, "step": 4700 }, { "epoch": 0.97, "learning_rate": 1.806888068880689e-05, "loss": 0.252, "step": 4710 }, { "epoch": 0.97, "learning_rate": 1.806478064780648e-05, "loss": 0.3254, "step": 4720 }, { "epoch": 0.97, "learning_rate": 1.806068060680607e-05, "loss": 0.2907, "step": 4730 }, { "epoch": 0.97, "learning_rate": 1.805658056580566e-05, "loss": 0.4041, "step": 4740 }, { "epoch": 0.97, "learning_rate": 1.805248052480525e-05, "loss": 0.3741, "step": 4750 }, { "epoch": 0.98, "learning_rate": 1.804838048380484e-05, "loss": 0.3189, "step": 4760 }, { "epoch": 0.98, "learning_rate": 1.8044280442804428e-05, "loss": 0.411, "step": 4770 }, { "epoch": 0.98, "learning_rate": 1.804018040180402e-05, "loss": 0.3392, "step": 4780 }, { "epoch": 0.98, "learning_rate": 1.8036080360803608e-05, "loss": 0.321, "step": 4790 }, { "epoch": 0.98, "learning_rate": 1.80319803198032e-05, "loss": 0.2386, "step": 4800 }, { "epoch": 0.99, "learning_rate": 1.802788027880279e-05, "loss": 0.3475, "step": 4810 }, { "epoch": 0.99, "learning_rate": 1.802378023780238e-05, "loss": 0.3495, "step": 4820 }, { "epoch": 0.99, "learning_rate": 1.801968019680197e-05, "loss": 0.3028, "step": 4830 }, { "epoch": 0.99, "learning_rate": 1.801558015580156e-05, "loss": 0.3384, "step": 4840 }, { "epoch": 0.99, "learning_rate": 1.801148011480115e-05, "loss": 0.2545, "step": 4850 }, { "epoch": 1.0, "learning_rate": 1.800738007380074e-05, "loss": 0.2674, "step": 4860 }, { "epoch": 1.0, "learning_rate": 1.800328003280033e-05, "loss": 0.3288, "step": 4870 }, { "epoch": 1.0, "eval_FN": 498, "eval_FP": 2849, "eval_TN": 12910, "eval_TP": 4408, "eval_accuracy": 0.8380353254294701, "eval_f1": 0.7248211789854476, "eval_loss": 0.37436580657958984, "eval_precision": 0.6074135317624363, "eval_recall": 0.8984916428862617, "eval_runtime": 141.9732, "eval_samples_per_second": 145.556, "eval_steps_per_second": 9.1, "step": 4878 }, { "epoch": 1.0, "learning_rate": 1.799917999179992e-05, "loss": 0.3284, "step": 4880 }, { "epoch": 1.0, "learning_rate": 1.799507995079951e-05, "loss": 0.3418, "step": 4890 }, { "epoch": 1.0, "learning_rate": 1.79909799097991e-05, "loss": 0.2969, "step": 4900 }, { "epoch": 1.01, "learning_rate": 1.798687986879869e-05, "loss": 0.3028, "step": 4910 }, { "epoch": 1.01, "learning_rate": 1.7982779827798282e-05, "loss": 0.3554, "step": 4920 }, { "epoch": 1.01, "learning_rate": 1.797867978679787e-05, "loss": 0.2822, "step": 4930 }, { "epoch": 1.01, "learning_rate": 1.797457974579746e-05, "loss": 0.246, "step": 4940 }, { "epoch": 1.01, "learning_rate": 1.797047970479705e-05, "loss": 0.2855, "step": 4950 }, { "epoch": 1.02, "learning_rate": 1.796637966379664e-05, "loss": 0.2581, "step": 4960 }, { "epoch": 1.02, "learning_rate": 1.796227962279623e-05, "loss": 0.3964, "step": 4970 }, { "epoch": 1.02, "learning_rate": 1.795817958179582e-05, "loss": 0.3397, "step": 4980 }, { "epoch": 1.02, "learning_rate": 1.795407954079541e-05, "loss": 0.3216, "step": 4990 }, { "epoch": 1.03, "learning_rate": 1.7949979499794998e-05, "loss": 0.2889, "step": 5000 }, { "epoch": 1.03, "learning_rate": 1.794587945879459e-05, "loss": 0.3164, "step": 5010 }, { "epoch": 1.03, "learning_rate": 1.7941779417794177e-05, "loss": 0.1978, "step": 5020 }, { "epoch": 1.03, "learning_rate": 1.793767937679377e-05, "loss": 0.3155, "step": 5030 }, { "epoch": 1.03, "learning_rate": 1.7933579335793357e-05, "loss": 0.3994, "step": 5040 }, { "epoch": 1.04, "learning_rate": 1.792947929479295e-05, "loss": 0.3924, "step": 5050 }, { "epoch": 1.04, "learning_rate": 1.792537925379254e-05, "loss": 0.2885, "step": 5060 }, { "epoch": 1.04, "learning_rate": 1.792127921279213e-05, "loss": 0.2513, "step": 5070 }, { "epoch": 1.04, "learning_rate": 1.791717917179172e-05, "loss": 0.3564, "step": 5080 }, { "epoch": 1.04, "learning_rate": 1.7913079130791308e-05, "loss": 0.3365, "step": 5090 }, { "epoch": 1.05, "learning_rate": 1.79089790897909e-05, "loss": 0.3096, "step": 5100 }, { "epoch": 1.05, "learning_rate": 1.7904879048790488e-05, "loss": 0.2817, "step": 5110 }, { "epoch": 1.05, "learning_rate": 1.790077900779008e-05, "loss": 0.3356, "step": 5120 }, { "epoch": 1.05, "learning_rate": 1.7896678966789668e-05, "loss": 0.2387, "step": 5130 }, { "epoch": 1.05, "learning_rate": 1.789257892578926e-05, "loss": 0.3089, "step": 5140 }, { "epoch": 1.06, "learning_rate": 1.7888478884788848e-05, "loss": 0.3273, "step": 5150 }, { "epoch": 1.06, "learning_rate": 1.788437884378844e-05, "loss": 0.3296, "step": 5160 }, { "epoch": 1.06, "learning_rate": 1.788027880278803e-05, "loss": 0.2439, "step": 5170 }, { "epoch": 1.06, "learning_rate": 1.787617876178762e-05, "loss": 0.2815, "step": 5180 }, { "epoch": 1.06, "learning_rate": 1.787207872078721e-05, "loss": 0.3219, "step": 5190 }, { "epoch": 1.07, "learning_rate": 1.78679786797868e-05, "loss": 0.4333, "step": 5200 }, { "epoch": 1.07, "learning_rate": 1.786387863878639e-05, "loss": 0.421, "step": 5210 }, { "epoch": 1.07, "learning_rate": 1.785977859778598e-05, "loss": 0.2905, "step": 5220 }, { "epoch": 1.07, "learning_rate": 1.785567855678557e-05, "loss": 0.28, "step": 5230 }, { "epoch": 1.07, "learning_rate": 1.785157851578516e-05, "loss": 0.2568, "step": 5240 }, { "epoch": 1.08, "learning_rate": 1.784747847478475e-05, "loss": 0.2842, "step": 5250 }, { "epoch": 1.08, "learning_rate": 1.7843378433784338e-05, "loss": 0.3305, "step": 5260 }, { "epoch": 1.08, "learning_rate": 1.783927839278393e-05, "loss": 0.3087, "step": 5270 }, { "epoch": 1.08, "learning_rate": 1.783517835178352e-05, "loss": 0.2193, "step": 5280 }, { "epoch": 1.08, "learning_rate": 1.783107831078311e-05, "loss": 0.2543, "step": 5290 }, { "epoch": 1.09, "learning_rate": 1.78269782697827e-05, "loss": 0.398, "step": 5300 }, { "epoch": 1.09, "learning_rate": 1.782287822878229e-05, "loss": 0.2998, "step": 5310 }, { "epoch": 1.09, "learning_rate": 1.781877818778188e-05, "loss": 0.3083, "step": 5320 }, { "epoch": 1.09, "learning_rate": 1.781467814678147e-05, "loss": 0.2771, "step": 5330 }, { "epoch": 1.09, "learning_rate": 1.781057810578106e-05, "loss": 0.3999, "step": 5340 }, { "epoch": 1.1, "learning_rate": 1.780647806478065e-05, "loss": 0.3322, "step": 5350 }, { "epoch": 1.1, "learning_rate": 1.780237802378024e-05, "loss": 0.3916, "step": 5360 }, { "epoch": 1.1, "learning_rate": 1.779827798277983e-05, "loss": 0.2909, "step": 5370 }, { "epoch": 1.1, "learning_rate": 1.779417794177942e-05, "loss": 0.3624, "step": 5380 }, { "epoch": 1.1, "learning_rate": 1.7790077900779012e-05, "loss": 0.3306, "step": 5390 }, { "epoch": 1.11, "learning_rate": 1.7785977859778597e-05, "loss": 0.3452, "step": 5400 }, { "epoch": 1.11, "learning_rate": 1.778187781877819e-05, "loss": 0.3368, "step": 5410 }, { "epoch": 1.11, "learning_rate": 1.7777777777777777e-05, "loss": 0.3049, "step": 5420 }, { "epoch": 1.11, "learning_rate": 1.7773677736777368e-05, "loss": 0.348, "step": 5430 }, { "epoch": 1.12, "learning_rate": 1.776957769577696e-05, "loss": 0.3074, "step": 5440 }, { "epoch": 1.12, "learning_rate": 1.7765477654776548e-05, "loss": 0.2911, "step": 5450 }, { "epoch": 1.12, "learning_rate": 1.776137761377614e-05, "loss": 0.3609, "step": 5460 }, { "epoch": 1.12, "learning_rate": 1.7757277572775728e-05, "loss": 0.3289, "step": 5470 }, { "epoch": 1.12, "learning_rate": 1.775317753177532e-05, "loss": 0.228, "step": 5480 }, { "epoch": 1.13, "learning_rate": 1.7749077490774908e-05, "loss": 0.2611, "step": 5490 }, { "epoch": 1.13, "learning_rate": 1.77449774497745e-05, "loss": 0.3434, "step": 5500 }, { "epoch": 1.13, "learning_rate": 1.7740877408774087e-05, "loss": 0.2415, "step": 5510 }, { "epoch": 1.13, "learning_rate": 1.773677736777368e-05, "loss": 0.2905, "step": 5520 }, { "epoch": 1.13, "learning_rate": 1.7732677326773267e-05, "loss": 0.3105, "step": 5530 }, { "epoch": 1.14, "learning_rate": 1.772857728577286e-05, "loss": 0.3383, "step": 5540 }, { "epoch": 1.14, "learning_rate": 1.772447724477245e-05, "loss": 0.3079, "step": 5550 }, { "epoch": 1.14, "learning_rate": 1.772037720377204e-05, "loss": 0.3205, "step": 5560 }, { "epoch": 1.14, "learning_rate": 1.771627716277163e-05, "loss": 0.2755, "step": 5570 }, { "epoch": 1.14, "learning_rate": 1.771217712177122e-05, "loss": 0.2114, "step": 5580 }, { "epoch": 1.15, "learning_rate": 1.770807708077081e-05, "loss": 0.3217, "step": 5590 }, { "epoch": 1.15, "learning_rate": 1.7703977039770398e-05, "loss": 0.3502, "step": 5600 }, { "epoch": 1.15, "learning_rate": 1.769987699876999e-05, "loss": 0.3439, "step": 5610 }, { "epoch": 1.15, "learning_rate": 1.7695776957769578e-05, "loss": 0.3429, "step": 5620 }, { "epoch": 1.15, "learning_rate": 1.769167691676917e-05, "loss": 0.3215, "step": 5630 }, { "epoch": 1.16, "learning_rate": 1.7687576875768758e-05, "loss": 0.2368, "step": 5640 }, { "epoch": 1.16, "learning_rate": 1.768347683476835e-05, "loss": 0.3334, "step": 5650 }, { "epoch": 1.16, "learning_rate": 1.767937679376794e-05, "loss": 0.3228, "step": 5660 }, { "epoch": 1.16, "learning_rate": 1.767527675276753e-05, "loss": 0.1951, "step": 5670 }, { "epoch": 1.16, "learning_rate": 1.767117671176712e-05, "loss": 0.351, "step": 5680 }, { "epoch": 1.17, "learning_rate": 1.766707667076671e-05, "loss": 0.2962, "step": 5690 }, { "epoch": 1.17, "learning_rate": 1.76629766297663e-05, "loss": 0.2563, "step": 5700 }, { "epoch": 1.17, "learning_rate": 1.765887658876589e-05, "loss": 0.3504, "step": 5710 }, { "epoch": 1.17, "learning_rate": 1.765477654776548e-05, "loss": 0.367, "step": 5720 }, { "epoch": 1.17, "learning_rate": 1.765067650676507e-05, "loss": 0.3128, "step": 5730 }, { "epoch": 1.18, "learning_rate": 1.764657646576466e-05, "loss": 0.2638, "step": 5740 }, { "epoch": 1.18, "learning_rate": 1.7642476424764248e-05, "loss": 0.2669, "step": 5750 }, { "epoch": 1.18, "learning_rate": 1.763837638376384e-05, "loss": 0.3203, "step": 5760 }, { "epoch": 1.18, "learning_rate": 1.763427634276343e-05, "loss": 0.2889, "step": 5770 }, { "epoch": 1.18, "learning_rate": 1.763017630176302e-05, "loss": 0.3688, "step": 5780 }, { "epoch": 1.19, "learning_rate": 1.762607626076261e-05, "loss": 0.2677, "step": 5790 }, { "epoch": 1.19, "learning_rate": 1.76219762197622e-05, "loss": 0.2674, "step": 5800 }, { "epoch": 1.19, "learning_rate": 1.7617876178761788e-05, "loss": 0.2329, "step": 5810 }, { "epoch": 1.19, "learning_rate": 1.761377613776138e-05, "loss": 0.229, "step": 5820 }, { "epoch": 1.2, "learning_rate": 1.7609676096760967e-05, "loss": 0.3313, "step": 5830 }, { "epoch": 1.2, "learning_rate": 1.760557605576056e-05, "loss": 0.3511, "step": 5840 }, { "epoch": 1.2, "learning_rate": 1.7601476014760147e-05, "loss": 0.2949, "step": 5850 }, { "epoch": 1.2, "learning_rate": 1.759737597375974e-05, "loss": 0.2977, "step": 5860 }, { "epoch": 1.2, "learning_rate": 1.7593275932759327e-05, "loss": 0.3473, "step": 5870 }, { "epoch": 1.21, "learning_rate": 1.758917589175892e-05, "loss": 0.3431, "step": 5880 }, { "epoch": 1.21, "learning_rate": 1.7585075850758507e-05, "loss": 0.2931, "step": 5890 }, { "epoch": 1.21, "learning_rate": 1.75809758097581e-05, "loss": 0.3298, "step": 5900 }, { "epoch": 1.21, "learning_rate": 1.757687576875769e-05, "loss": 0.2823, "step": 5910 }, { "epoch": 1.21, "learning_rate": 1.7572775727757278e-05, "loss": 0.3349, "step": 5920 }, { "epoch": 1.22, "learning_rate": 1.756867568675687e-05, "loss": 0.4149, "step": 5930 }, { "epoch": 1.22, "learning_rate": 1.7564575645756458e-05, "loss": 0.3266, "step": 5940 }, { "epoch": 1.22, "learning_rate": 1.756047560475605e-05, "loss": 0.3168, "step": 5950 }, { "epoch": 1.22, "learning_rate": 1.7556375563755638e-05, "loss": 0.36, "step": 5960 }, { "epoch": 1.22, "learning_rate": 1.755227552275523e-05, "loss": 0.2926, "step": 5970 }, { "epoch": 1.23, "learning_rate": 1.7548175481754818e-05, "loss": 0.335, "step": 5980 }, { "epoch": 1.23, "learning_rate": 1.754407544075441e-05, "loss": 0.2463, "step": 5990 }, { "epoch": 1.23, "learning_rate": 1.7539975399753997e-05, "loss": 0.3547, "step": 6000 }, { "epoch": 1.23, "learning_rate": 1.753587535875359e-05, "loss": 0.356, "step": 6010 }, { "epoch": 1.23, "learning_rate": 1.753177531775318e-05, "loss": 0.3162, "step": 6020 }, { "epoch": 1.24, "learning_rate": 1.752767527675277e-05, "loss": 0.2353, "step": 6030 }, { "epoch": 1.24, "learning_rate": 1.752357523575236e-05, "loss": 0.2587, "step": 6040 }, { "epoch": 1.24, "learning_rate": 1.751947519475195e-05, "loss": 0.3122, "step": 6050 }, { "epoch": 1.24, "learning_rate": 1.751537515375154e-05, "loss": 0.3817, "step": 6060 }, { "epoch": 1.24, "learning_rate": 1.751127511275113e-05, "loss": 0.2055, "step": 6070 }, { "epoch": 1.25, "learning_rate": 1.750717507175072e-05, "loss": 0.3934, "step": 6080 }, { "epoch": 1.25, "learning_rate": 1.7503075030750308e-05, "loss": 0.3387, "step": 6090 }, { "epoch": 1.25, "learning_rate": 1.74989749897499e-05, "loss": 0.2728, "step": 6100 }, { "epoch": 1.25, "learning_rate": 1.7494874948749488e-05, "loss": 0.321, "step": 6110 }, { "epoch": 1.25, "learning_rate": 1.749077490774908e-05, "loss": 0.4601, "step": 6120 }, { "epoch": 1.26, "learning_rate": 1.748667486674867e-05, "loss": 0.1881, "step": 6130 }, { "epoch": 1.26, "learning_rate": 1.748257482574826e-05, "loss": 0.2268, "step": 6140 }, { "epoch": 1.26, "learning_rate": 1.747847478474785e-05, "loss": 0.2673, "step": 6150 }, { "epoch": 1.26, "learning_rate": 1.747437474374744e-05, "loss": 0.3285, "step": 6160 }, { "epoch": 1.26, "learning_rate": 1.747027470274703e-05, "loss": 0.3812, "step": 6170 }, { "epoch": 1.27, "learning_rate": 1.746617466174662e-05, "loss": 0.3462, "step": 6180 }, { "epoch": 1.27, "learning_rate": 1.746207462074621e-05, "loss": 0.3705, "step": 6190 }, { "epoch": 1.27, "learning_rate": 1.74579745797458e-05, "loss": 0.3146, "step": 6200 }, { "epoch": 1.27, "learning_rate": 1.7453874538745387e-05, "loss": 0.3776, "step": 6210 }, { "epoch": 1.28, "learning_rate": 1.744977449774498e-05, "loss": 0.3036, "step": 6220 }, { "epoch": 1.28, "learning_rate": 1.7445674456744567e-05, "loss": 0.3429, "step": 6230 }, { "epoch": 1.28, "learning_rate": 1.744157441574416e-05, "loss": 0.2378, "step": 6240 }, { "epoch": 1.28, "learning_rate": 1.7437474374743747e-05, "loss": 0.3468, "step": 6250 }, { "epoch": 1.28, "learning_rate": 1.7433374333743338e-05, "loss": 0.3352, "step": 6260 }, { "epoch": 1.29, "learning_rate": 1.742927429274293e-05, "loss": 0.2819, "step": 6270 }, { "epoch": 1.29, "learning_rate": 1.7425174251742518e-05, "loss": 0.2886, "step": 6280 }, { "epoch": 1.29, "learning_rate": 1.742107421074211e-05, "loss": 0.3138, "step": 6290 }, { "epoch": 1.29, "learning_rate": 1.7416974169741698e-05, "loss": 0.2873, "step": 6300 }, { "epoch": 1.29, "learning_rate": 1.741287412874129e-05, "loss": 0.206, "step": 6310 }, { "epoch": 1.3, "learning_rate": 1.7408774087740877e-05, "loss": 0.2187, "step": 6320 }, { "epoch": 1.3, "learning_rate": 1.740467404674047e-05, "loss": 0.356, "step": 6330 }, { "epoch": 1.3, "learning_rate": 1.7400574005740057e-05, "loss": 0.1848, "step": 6340 }, { "epoch": 1.3, "learning_rate": 1.739647396473965e-05, "loss": 0.3037, "step": 6350 }, { "epoch": 1.3, "learning_rate": 1.7392373923739237e-05, "loss": 0.3712, "step": 6360 }, { "epoch": 1.31, "learning_rate": 1.738827388273883e-05, "loss": 0.2197, "step": 6370 }, { "epoch": 1.31, "learning_rate": 1.738417384173842e-05, "loss": 0.249, "step": 6380 }, { "epoch": 1.31, "learning_rate": 1.738007380073801e-05, "loss": 0.2463, "step": 6390 }, { "epoch": 1.31, "learning_rate": 1.73759737597376e-05, "loss": 0.3051, "step": 6400 }, { "epoch": 1.31, "learning_rate": 1.7371873718737188e-05, "loss": 0.2001, "step": 6410 }, { "epoch": 1.32, "learning_rate": 1.736777367773678e-05, "loss": 0.2584, "step": 6420 }, { "epoch": 1.32, "learning_rate": 1.7363673636736368e-05, "loss": 0.2244, "step": 6430 }, { "epoch": 1.32, "learning_rate": 1.735957359573596e-05, "loss": 0.3401, "step": 6440 }, { "epoch": 1.32, "learning_rate": 1.7355473554735548e-05, "loss": 0.4253, "step": 6450 }, { "epoch": 1.32, "learning_rate": 1.735137351373514e-05, "loss": 0.2315, "step": 6460 }, { "epoch": 1.33, "learning_rate": 1.7347273472734728e-05, "loss": 0.2716, "step": 6470 }, { "epoch": 1.33, "learning_rate": 1.734317343173432e-05, "loss": 0.2736, "step": 6480 }, { "epoch": 1.33, "learning_rate": 1.733907339073391e-05, "loss": 0.333, "step": 6490 }, { "epoch": 1.33, "learning_rate": 1.73349733497335e-05, "loss": 0.269, "step": 6500 }, { "epoch": 1.33, "learning_rate": 1.733087330873309e-05, "loss": 0.1981, "step": 6510 }, { "epoch": 1.34, "learning_rate": 1.732677326773268e-05, "loss": 0.2551, "step": 6520 }, { "epoch": 1.34, "learning_rate": 1.732267322673227e-05, "loss": 0.3273, "step": 6530 }, { "epoch": 1.34, "learning_rate": 1.731857318573186e-05, "loss": 0.221, "step": 6540 }, { "epoch": 1.34, "learning_rate": 1.731447314473145e-05, "loss": 0.3006, "step": 6550 }, { "epoch": 1.34, "learning_rate": 1.731037310373104e-05, "loss": 0.248, "step": 6560 }, { "epoch": 1.35, "learning_rate": 1.730627306273063e-05, "loss": 0.2826, "step": 6570 }, { "epoch": 1.35, "learning_rate": 1.7302173021730218e-05, "loss": 0.288, "step": 6580 }, { "epoch": 1.35, "learning_rate": 1.729807298072981e-05, "loss": 0.3256, "step": 6590 }, { "epoch": 1.35, "learning_rate": 1.7293972939729398e-05, "loss": 0.2441, "step": 6600 }, { "epoch": 1.36, "learning_rate": 1.7289872898728986e-05, "loss": 0.2653, "step": 6610 }, { "epoch": 1.36, "learning_rate": 1.7285772857728578e-05, "loss": 0.2495, "step": 6620 }, { "epoch": 1.36, "learning_rate": 1.728167281672817e-05, "loss": 0.3632, "step": 6630 }, { "epoch": 1.36, "learning_rate": 1.7277572775727758e-05, "loss": 0.3414, "step": 6640 }, { "epoch": 1.36, "learning_rate": 1.727347273472735e-05, "loss": 0.2849, "step": 6650 }, { "epoch": 1.37, "learning_rate": 1.7269372693726937e-05, "loss": 0.2769, "step": 6660 }, { "epoch": 1.37, "learning_rate": 1.726527265272653e-05, "loss": 0.32, "step": 6670 }, { "epoch": 1.37, "learning_rate": 1.7261172611726117e-05, "loss": 0.3103, "step": 6680 }, { "epoch": 1.37, "learning_rate": 1.725707257072571e-05, "loss": 0.3826, "step": 6690 }, { "epoch": 1.37, "learning_rate": 1.7252972529725297e-05, "loss": 0.3074, "step": 6700 }, { "epoch": 1.38, "learning_rate": 1.724887248872489e-05, "loss": 0.2312, "step": 6710 }, { "epoch": 1.38, "learning_rate": 1.7244772447724477e-05, "loss": 0.2156, "step": 6720 }, { "epoch": 1.38, "learning_rate": 1.724067240672407e-05, "loss": 0.2845, "step": 6730 }, { "epoch": 1.38, "learning_rate": 1.723657236572366e-05, "loss": 0.3659, "step": 6740 }, { "epoch": 1.38, "learning_rate": 1.7232472324723248e-05, "loss": 0.3317, "step": 6750 }, { "epoch": 1.39, "learning_rate": 1.722837228372284e-05, "loss": 0.2844, "step": 6760 }, { "epoch": 1.39, "learning_rate": 1.7224272242722428e-05, "loss": 0.2582, "step": 6770 }, { "epoch": 1.39, "learning_rate": 1.722017220172202e-05, "loss": 0.2412, "step": 6780 }, { "epoch": 1.39, "learning_rate": 1.7216072160721608e-05, "loss": 0.3141, "step": 6790 }, { "epoch": 1.39, "learning_rate": 1.72119721197212e-05, "loss": 0.2593, "step": 6800 }, { "epoch": 1.4, "learning_rate": 1.7207872078720788e-05, "loss": 0.3165, "step": 6810 }, { "epoch": 1.4, "learning_rate": 1.720377203772038e-05, "loss": 0.2699, "step": 6820 }, { "epoch": 1.4, "learning_rate": 1.7199671996719967e-05, "loss": 0.2408, "step": 6830 }, { "epoch": 1.4, "learning_rate": 1.719557195571956e-05, "loss": 0.2683, "step": 6840 }, { "epoch": 1.4, "learning_rate": 1.719147191471915e-05, "loss": 0.3512, "step": 6850 }, { "epoch": 1.41, "learning_rate": 1.718737187371874e-05, "loss": 0.3394, "step": 6860 }, { "epoch": 1.41, "learning_rate": 1.718327183271833e-05, "loss": 0.2291, "step": 6870 }, { "epoch": 1.41, "learning_rate": 1.717917179171792e-05, "loss": 0.259, "step": 6880 }, { "epoch": 1.41, "learning_rate": 1.717507175071751e-05, "loss": 0.3811, "step": 6890 }, { "epoch": 1.41, "learning_rate": 1.7170971709717098e-05, "loss": 0.3299, "step": 6900 }, { "epoch": 1.42, "learning_rate": 1.716687166871669e-05, "loss": 0.1881, "step": 6910 }, { "epoch": 1.42, "learning_rate": 1.7162771627716278e-05, "loss": 0.2819, "step": 6920 }, { "epoch": 1.42, "learning_rate": 1.715867158671587e-05, "loss": 0.4115, "step": 6930 }, { "epoch": 1.42, "learning_rate": 1.7154571545715458e-05, "loss": 0.2734, "step": 6940 }, { "epoch": 1.42, "learning_rate": 1.715047150471505e-05, "loss": 0.3075, "step": 6950 }, { "epoch": 1.43, "learning_rate": 1.714637146371464e-05, "loss": 0.3945, "step": 6960 }, { "epoch": 1.43, "learning_rate": 1.714227142271423e-05, "loss": 0.2939, "step": 6970 }, { "epoch": 1.43, "learning_rate": 1.713817138171382e-05, "loss": 0.2139, "step": 6980 }, { "epoch": 1.43, "learning_rate": 1.713407134071341e-05, "loss": 0.3366, "step": 6990 }, { "epoch": 1.44, "learning_rate": 1.7129971299712997e-05, "loss": 0.2558, "step": 7000 }, { "epoch": 1.44, "learning_rate": 1.712587125871259e-05, "loss": 0.3908, "step": 7010 }, { "epoch": 1.44, "learning_rate": 1.7121771217712177e-05, "loss": 0.2021, "step": 7020 }, { "epoch": 1.44, "learning_rate": 1.711767117671177e-05, "loss": 0.2952, "step": 7030 }, { "epoch": 1.44, "learning_rate": 1.7113571135711357e-05, "loss": 0.2837, "step": 7040 }, { "epoch": 1.45, "learning_rate": 1.710947109471095e-05, "loss": 0.1982, "step": 7050 }, { "epoch": 1.45, "learning_rate": 1.7105371053710537e-05, "loss": 0.1707, "step": 7060 }, { "epoch": 1.45, "learning_rate": 1.7101271012710128e-05, "loss": 0.3063, "step": 7070 }, { "epoch": 1.45, "learning_rate": 1.7097170971709716e-05, "loss": 0.3494, "step": 7080 }, { "epoch": 1.45, "learning_rate": 1.7093070930709308e-05, "loss": 0.3126, "step": 7090 }, { "epoch": 1.46, "learning_rate": 1.70889708897089e-05, "loss": 0.2733, "step": 7100 }, { "epoch": 1.46, "learning_rate": 1.7084870848708488e-05, "loss": 0.2817, "step": 7110 }, { "epoch": 1.46, "learning_rate": 1.708077080770808e-05, "loss": 0.2862, "step": 7120 }, { "epoch": 1.46, "learning_rate": 1.7076670766707668e-05, "loss": 0.3164, "step": 7130 }, { "epoch": 1.46, "learning_rate": 1.707257072570726e-05, "loss": 0.2959, "step": 7140 }, { "epoch": 1.47, "learning_rate": 1.7068470684706847e-05, "loss": 0.3699, "step": 7150 }, { "epoch": 1.47, "learning_rate": 1.706437064370644e-05, "loss": 0.3196, "step": 7160 }, { "epoch": 1.47, "learning_rate": 1.7060270602706027e-05, "loss": 0.234, "step": 7170 }, { "epoch": 1.47, "learning_rate": 1.705617056170562e-05, "loss": 0.3088, "step": 7180 }, { "epoch": 1.47, "learning_rate": 1.7052070520705207e-05, "loss": 0.2093, "step": 7190 }, { "epoch": 1.48, "learning_rate": 1.70479704797048e-05, "loss": 0.2776, "step": 7200 }, { "epoch": 1.48, "learning_rate": 1.704387043870439e-05, "loss": 0.3036, "step": 7210 }, { "epoch": 1.48, "learning_rate": 1.703977039770398e-05, "loss": 0.477, "step": 7220 }, { "epoch": 1.48, "learning_rate": 1.703567035670357e-05, "loss": 0.2399, "step": 7230 }, { "epoch": 1.48, "learning_rate": 1.7031570315703158e-05, "loss": 0.3689, "step": 7240 }, { "epoch": 1.49, "learning_rate": 1.702747027470275e-05, "loss": 0.3276, "step": 7250 }, { "epoch": 1.49, "learning_rate": 1.7023370233702338e-05, "loss": 0.3262, "step": 7260 }, { "epoch": 1.49, "learning_rate": 1.701927019270193e-05, "loss": 0.2819, "step": 7270 }, { "epoch": 1.49, "learning_rate": 1.7015170151701518e-05, "loss": 0.2502, "step": 7280 }, { "epoch": 1.49, "learning_rate": 1.701107011070111e-05, "loss": 0.3086, "step": 7290 }, { "epoch": 1.5, "learning_rate": 1.7006970069700698e-05, "loss": 0.3462, "step": 7300 }, { "epoch": 1.5, "learning_rate": 1.700287002870029e-05, "loss": 0.2893, "step": 7310 }, { "epoch": 1.5, "learning_rate": 1.699876998769988e-05, "loss": 0.256, "step": 7320 }, { "epoch": 1.5, "learning_rate": 1.699466994669947e-05, "loss": 0.3429, "step": 7330 }, { "epoch": 1.5, "learning_rate": 1.699056990569906e-05, "loss": 0.3109, "step": 7340 }, { "epoch": 1.51, "learning_rate": 1.698646986469865e-05, "loss": 0.2516, "step": 7350 }, { "epoch": 1.51, "learning_rate": 1.698236982369824e-05, "loss": 0.306, "step": 7360 }, { "epoch": 1.51, "learning_rate": 1.697826978269783e-05, "loss": 0.2639, "step": 7370 }, { "epoch": 1.51, "learning_rate": 1.697416974169742e-05, "loss": 0.195, "step": 7380 }, { "epoch": 1.51, "learning_rate": 1.697006970069701e-05, "loss": 0.3782, "step": 7390 }, { "epoch": 1.52, "learning_rate": 1.6965969659696597e-05, "loss": 0.2585, "step": 7400 }, { "epoch": 1.52, "learning_rate": 1.6961869618696188e-05, "loss": 0.3153, "step": 7410 }, { "epoch": 1.52, "learning_rate": 1.6957769577695776e-05, "loss": 0.3057, "step": 7420 }, { "epoch": 1.52, "learning_rate": 1.6953669536695368e-05, "loss": 0.4063, "step": 7430 }, { "epoch": 1.53, "learning_rate": 1.6949569495694956e-05, "loss": 0.2961, "step": 7440 }, { "epoch": 1.53, "learning_rate": 1.6945469454694548e-05, "loss": 0.2161, "step": 7450 }, { "epoch": 1.53, "learning_rate": 1.6941369413694136e-05, "loss": 0.3482, "step": 7460 }, { "epoch": 1.53, "learning_rate": 1.6937269372693727e-05, "loss": 0.283, "step": 7470 }, { "epoch": 1.53, "learning_rate": 1.693316933169332e-05, "loss": 0.2814, "step": 7480 }, { "epoch": 1.54, "learning_rate": 1.6929069290692907e-05, "loss": 0.226, "step": 7490 }, { "epoch": 1.54, "learning_rate": 1.69249692496925e-05, "loss": 0.2942, "step": 7500 }, { "epoch": 1.54, "learning_rate": 1.6920869208692087e-05, "loss": 0.2179, "step": 7510 }, { "epoch": 1.54, "learning_rate": 1.691676916769168e-05, "loss": 0.1906, "step": 7520 }, { "epoch": 1.54, "learning_rate": 1.6912669126691267e-05, "loss": 0.3422, "step": 7530 }, { "epoch": 1.55, "learning_rate": 1.690856908569086e-05, "loss": 0.3053, "step": 7540 }, { "epoch": 1.55, "learning_rate": 1.6904469044690447e-05, "loss": 0.3584, "step": 7550 }, { "epoch": 1.55, "learning_rate": 1.6900369003690038e-05, "loss": 0.2177, "step": 7560 }, { "epoch": 1.55, "learning_rate": 1.6896268962689626e-05, "loss": 0.3147, "step": 7570 }, { "epoch": 1.55, "learning_rate": 1.6892168921689218e-05, "loss": 0.2615, "step": 7580 }, { "epoch": 1.56, "learning_rate": 1.688806888068881e-05, "loss": 0.3669, "step": 7590 }, { "epoch": 1.56, "learning_rate": 1.6883968839688398e-05, "loss": 0.3666, "step": 7600 }, { "epoch": 1.56, "learning_rate": 1.687986879868799e-05, "loss": 0.2498, "step": 7610 }, { "epoch": 1.56, "learning_rate": 1.6875768757687578e-05, "loss": 0.2949, "step": 7620 }, { "epoch": 1.56, "learning_rate": 1.687166871668717e-05, "loss": 0.2601, "step": 7630 }, { "epoch": 1.57, "learning_rate": 1.6867568675686757e-05, "loss": 0.3201, "step": 7640 }, { "epoch": 1.57, "learning_rate": 1.686346863468635e-05, "loss": 0.3365, "step": 7650 }, { "epoch": 1.57, "learning_rate": 1.6859368593685937e-05, "loss": 0.4379, "step": 7660 }, { "epoch": 1.57, "learning_rate": 1.685526855268553e-05, "loss": 0.2451, "step": 7670 }, { "epoch": 1.57, "learning_rate": 1.6851168511685117e-05, "loss": 0.3495, "step": 7680 }, { "epoch": 1.58, "learning_rate": 1.684706847068471e-05, "loss": 0.2656, "step": 7690 }, { "epoch": 1.58, "learning_rate": 1.68429684296843e-05, "loss": 0.231, "step": 7700 }, { "epoch": 1.58, "learning_rate": 1.683886838868389e-05, "loss": 0.2906, "step": 7710 }, { "epoch": 1.58, "learning_rate": 1.683476834768348e-05, "loss": 0.2177, "step": 7720 }, { "epoch": 1.58, "learning_rate": 1.6830668306683068e-05, "loss": 0.3514, "step": 7730 }, { "epoch": 1.59, "learning_rate": 1.682656826568266e-05, "loss": 0.2403, "step": 7740 }, { "epoch": 1.59, "learning_rate": 1.6822468224682248e-05, "loss": 0.3533, "step": 7750 }, { "epoch": 1.59, "learning_rate": 1.681836818368184e-05, "loss": 0.3011, "step": 7760 }, { "epoch": 1.59, "learning_rate": 1.6814268142681428e-05, "loss": 0.2738, "step": 7770 }, { "epoch": 1.59, "learning_rate": 1.681016810168102e-05, "loss": 0.2004, "step": 7780 }, { "epoch": 1.6, "learning_rate": 1.6806068060680608e-05, "loss": 0.3944, "step": 7790 }, { "epoch": 1.6, "learning_rate": 1.6801968019680196e-05, "loss": 0.2799, "step": 7800 }, { "epoch": 1.6, "learning_rate": 1.6797867978679787e-05, "loss": 0.3361, "step": 7810 }, { "epoch": 1.6, "learning_rate": 1.6793767937679376e-05, "loss": 0.2155, "step": 7820 }, { "epoch": 1.61, "learning_rate": 1.6789667896678967e-05, "loss": 0.2744, "step": 7830 }, { "epoch": 1.61, "learning_rate": 1.678556785567856e-05, "loss": 0.2172, "step": 7840 }, { "epoch": 1.61, "learning_rate": 1.6781467814678147e-05, "loss": 0.2873, "step": 7850 }, { "epoch": 1.61, "learning_rate": 1.677736777367774e-05, "loss": 0.2722, "step": 7860 }, { "epoch": 1.61, "learning_rate": 1.6773267732677327e-05, "loss": 0.348, "step": 7870 }, { "epoch": 1.62, "learning_rate": 1.676916769167692e-05, "loss": 0.22, "step": 7880 }, { "epoch": 1.62, "learning_rate": 1.6765067650676507e-05, "loss": 0.3468, "step": 7890 }, { "epoch": 1.62, "learning_rate": 1.6760967609676098e-05, "loss": 0.3295, "step": 7900 }, { "epoch": 1.62, "learning_rate": 1.6756867568675686e-05, "loss": 0.2088, "step": 7910 }, { "epoch": 1.62, "learning_rate": 1.6752767527675278e-05, "loss": 0.2528, "step": 7920 }, { "epoch": 1.63, "learning_rate": 1.6748667486674866e-05, "loss": 0.3237, "step": 7930 }, { "epoch": 1.63, "learning_rate": 1.6744567445674458e-05, "loss": 0.2157, "step": 7940 }, { "epoch": 1.63, "learning_rate": 1.674046740467405e-05, "loss": 0.3716, "step": 7950 }, { "epoch": 1.63, "learning_rate": 1.6736367363673638e-05, "loss": 0.2515, "step": 7960 }, { "epoch": 1.63, "learning_rate": 1.673226732267323e-05, "loss": 0.2072, "step": 7970 }, { "epoch": 1.64, "learning_rate": 1.6728167281672817e-05, "loss": 0.2695, "step": 7980 }, { "epoch": 1.64, "learning_rate": 1.672406724067241e-05, "loss": 0.3962, "step": 7990 }, { "epoch": 1.64, "learning_rate": 1.6719967199671997e-05, "loss": 0.2756, "step": 8000 }, { "epoch": 1.64, "learning_rate": 1.671586715867159e-05, "loss": 0.2858, "step": 8010 }, { "epoch": 1.64, "learning_rate": 1.6711767117671177e-05, "loss": 0.299, "step": 8020 }, { "epoch": 1.65, "learning_rate": 1.670766707667077e-05, "loss": 0.2674, "step": 8030 }, { "epoch": 1.65, "learning_rate": 1.6703567035670357e-05, "loss": 0.2132, "step": 8040 }, { "epoch": 1.65, "learning_rate": 1.6699466994669948e-05, "loss": 0.2141, "step": 8050 }, { "epoch": 1.65, "learning_rate": 1.669536695366954e-05, "loss": 0.21, "step": 8060 }, { "epoch": 1.65, "learning_rate": 1.6691266912669128e-05, "loss": 0.4325, "step": 8070 }, { "epoch": 1.66, "learning_rate": 1.668716687166872e-05, "loss": 0.3819, "step": 8080 }, { "epoch": 1.66, "learning_rate": 1.6683066830668308e-05, "loss": 0.2897, "step": 8090 }, { "epoch": 1.66, "learning_rate": 1.66789667896679e-05, "loss": 0.2217, "step": 8100 }, { "epoch": 1.66, "learning_rate": 1.6674866748667488e-05, "loss": 0.2511, "step": 8110 }, { "epoch": 1.66, "learning_rate": 1.667076670766708e-05, "loss": 0.1945, "step": 8120 }, { "epoch": 1.67, "learning_rate": 1.6666666666666667e-05, "loss": 0.2625, "step": 8130 }, { "epoch": 1.67, "learning_rate": 1.666256662566626e-05, "loss": 0.4433, "step": 8140 }, { "epoch": 1.67, "learning_rate": 1.6658466584665847e-05, "loss": 0.2573, "step": 8150 }, { "epoch": 1.67, "learning_rate": 1.665436654366544e-05, "loss": 0.3167, "step": 8160 }, { "epoch": 1.67, "learning_rate": 1.665026650266503e-05, "loss": 0.2928, "step": 8170 }, { "epoch": 1.68, "learning_rate": 1.664616646166462e-05, "loss": 0.2686, "step": 8180 }, { "epoch": 1.68, "learning_rate": 1.664206642066421e-05, "loss": 0.279, "step": 8190 }, { "epoch": 1.68, "learning_rate": 1.66379663796638e-05, "loss": 0.2573, "step": 8200 }, { "epoch": 1.68, "learning_rate": 1.6633866338663387e-05, "loss": 0.2654, "step": 8210 }, { "epoch": 1.69, "learning_rate": 1.6629766297662978e-05, "loss": 0.218, "step": 8220 }, { "epoch": 1.69, "learning_rate": 1.6625666256662566e-05, "loss": 0.2693, "step": 8230 }, { "epoch": 1.69, "learning_rate": 1.6621566215662158e-05, "loss": 0.2958, "step": 8240 }, { "epoch": 1.69, "learning_rate": 1.6617466174661746e-05, "loss": 0.361, "step": 8250 }, { "epoch": 1.69, "learning_rate": 1.6613366133661338e-05, "loss": 0.3522, "step": 8260 }, { "epoch": 1.7, "learning_rate": 1.6609266092660926e-05, "loss": 0.4026, "step": 8270 }, { "epoch": 1.7, "learning_rate": 1.6605166051660518e-05, "loss": 0.3462, "step": 8280 }, { "epoch": 1.7, "learning_rate": 1.6601066010660106e-05, "loss": 0.2767, "step": 8290 }, { "epoch": 1.7, "learning_rate": 1.6596965969659697e-05, "loss": 0.2609, "step": 8300 }, { "epoch": 1.7, "learning_rate": 1.659286592865929e-05, "loss": 0.3064, "step": 8310 }, { "epoch": 1.71, "learning_rate": 1.6588765887658877e-05, "loss": 0.3346, "step": 8320 }, { "epoch": 1.71, "learning_rate": 1.658466584665847e-05, "loss": 0.3, "step": 8330 }, { "epoch": 1.71, "learning_rate": 1.6580565805658057e-05, "loss": 0.2839, "step": 8340 }, { "epoch": 1.71, "learning_rate": 1.657646576465765e-05, "loss": 0.2761, "step": 8350 }, { "epoch": 1.71, "learning_rate": 1.6572365723657237e-05, "loss": 0.2375, "step": 8360 }, { "epoch": 1.72, "learning_rate": 1.656826568265683e-05, "loss": 0.2904, "step": 8370 }, { "epoch": 1.72, "learning_rate": 1.6564165641656417e-05, "loss": 0.2985, "step": 8380 }, { "epoch": 1.72, "learning_rate": 1.6560065600656008e-05, "loss": 0.376, "step": 8390 }, { "epoch": 1.72, "learning_rate": 1.6555965559655596e-05, "loss": 0.2942, "step": 8400 }, { "epoch": 1.72, "learning_rate": 1.6551865518655188e-05, "loss": 0.2319, "step": 8410 }, { "epoch": 1.73, "learning_rate": 1.654776547765478e-05, "loss": 0.2768, "step": 8420 }, { "epoch": 1.73, "learning_rate": 1.6543665436654368e-05, "loss": 0.2296, "step": 8430 }, { "epoch": 1.73, "learning_rate": 1.653956539565396e-05, "loss": 0.3506, "step": 8440 }, { "epoch": 1.73, "learning_rate": 1.6535465354653548e-05, "loss": 0.3109, "step": 8450 }, { "epoch": 1.73, "learning_rate": 1.653136531365314e-05, "loss": 0.372, "step": 8460 }, { "epoch": 1.74, "learning_rate": 1.6527265272652727e-05, "loss": 0.2823, "step": 8470 }, { "epoch": 1.74, "learning_rate": 1.652316523165232e-05, "loss": 0.2477, "step": 8480 }, { "epoch": 1.74, "learning_rate": 1.6519065190651907e-05, "loss": 0.2999, "step": 8490 }, { "epoch": 1.74, "learning_rate": 1.65149651496515e-05, "loss": 0.2838, "step": 8500 }, { "epoch": 1.74, "learning_rate": 1.6510865108651087e-05, "loss": 0.3108, "step": 8510 }, { "epoch": 1.75, "learning_rate": 1.650676506765068e-05, "loss": 0.3199, "step": 8520 }, { "epoch": 1.75, "learning_rate": 1.650266502665027e-05, "loss": 0.2724, "step": 8530 }, { "epoch": 1.75, "learning_rate": 1.649856498564986e-05, "loss": 0.2525, "step": 8540 }, { "epoch": 1.75, "learning_rate": 1.649446494464945e-05, "loss": 0.3621, "step": 8550 }, { "epoch": 1.75, "learning_rate": 1.6490364903649038e-05, "loss": 0.3264, "step": 8560 }, { "epoch": 1.76, "learning_rate": 1.648626486264863e-05, "loss": 0.3217, "step": 8570 }, { "epoch": 1.76, "learning_rate": 1.6482164821648218e-05, "loss": 0.2111, "step": 8580 }, { "epoch": 1.76, "learning_rate": 1.647806478064781e-05, "loss": 0.3078, "step": 8590 }, { "epoch": 1.76, "learning_rate": 1.6473964739647398e-05, "loss": 0.2764, "step": 8600 }, { "epoch": 1.77, "learning_rate": 1.6469864698646986e-05, "loss": 0.3139, "step": 8610 }, { "epoch": 1.77, "learning_rate": 1.6465764657646577e-05, "loss": 0.3134, "step": 8620 }, { "epoch": 1.77, "learning_rate": 1.6461664616646166e-05, "loss": 0.2694, "step": 8630 }, { "epoch": 1.77, "learning_rate": 1.6457564575645757e-05, "loss": 0.228, "step": 8640 }, { "epoch": 1.77, "learning_rate": 1.6453464534645345e-05, "loss": 0.2447, "step": 8650 }, { "epoch": 1.78, "learning_rate": 1.6449364493644937e-05, "loss": 0.2809, "step": 8660 }, { "epoch": 1.78, "learning_rate": 1.644526445264453e-05, "loss": 0.2893, "step": 8670 }, { "epoch": 1.78, "learning_rate": 1.6441164411644117e-05, "loss": 0.2355, "step": 8680 }, { "epoch": 1.78, "learning_rate": 1.643706437064371e-05, "loss": 0.3254, "step": 8690 }, { "epoch": 1.78, "learning_rate": 1.6432964329643297e-05, "loss": 0.1866, "step": 8700 }, { "epoch": 1.79, "learning_rate": 1.6428864288642888e-05, "loss": 0.4173, "step": 8710 }, { "epoch": 1.79, "learning_rate": 1.6424764247642476e-05, "loss": 0.1702, "step": 8720 }, { "epoch": 1.79, "learning_rate": 1.6420664206642068e-05, "loss": 0.2863, "step": 8730 }, { "epoch": 1.79, "learning_rate": 1.6416564165641656e-05, "loss": 0.2541, "step": 8740 }, { "epoch": 1.79, "learning_rate": 1.6412464124641248e-05, "loss": 0.3285, "step": 8750 }, { "epoch": 1.8, "learning_rate": 1.6408364083640836e-05, "loss": 0.2366, "step": 8760 }, { "epoch": 1.8, "learning_rate": 1.6404264042640428e-05, "loss": 0.2923, "step": 8770 }, { "epoch": 1.8, "learning_rate": 1.640016400164002e-05, "loss": 0.3123, "step": 8780 }, { "epoch": 1.8, "learning_rate": 1.6396063960639607e-05, "loss": 0.361, "step": 8790 }, { "epoch": 1.8, "learning_rate": 1.63919639196392e-05, "loss": 0.271, "step": 8800 }, { "epoch": 1.81, "learning_rate": 1.6387863878638787e-05, "loss": 0.3007, "step": 8810 }, { "epoch": 1.81, "learning_rate": 1.638376383763838e-05, "loss": 0.3458, "step": 8820 }, { "epoch": 1.81, "learning_rate": 1.6379663796637967e-05, "loss": 0.2152, "step": 8830 }, { "epoch": 1.81, "learning_rate": 1.637556375563756e-05, "loss": 0.2732, "step": 8840 }, { "epoch": 1.81, "learning_rate": 1.6371463714637147e-05, "loss": 0.3559, "step": 8850 }, { "epoch": 1.82, "learning_rate": 1.636736367363674e-05, "loss": 0.3072, "step": 8860 }, { "epoch": 1.82, "learning_rate": 1.6363263632636327e-05, "loss": 0.3059, "step": 8870 }, { "epoch": 1.82, "learning_rate": 1.6359163591635918e-05, "loss": 0.234, "step": 8880 }, { "epoch": 1.82, "learning_rate": 1.635506355063551e-05, "loss": 0.3019, "step": 8890 }, { "epoch": 1.82, "learning_rate": 1.6350963509635098e-05, "loss": 0.2738, "step": 8900 }, { "epoch": 1.83, "learning_rate": 1.634686346863469e-05, "loss": 0.2718, "step": 8910 }, { "epoch": 1.83, "learning_rate": 1.6342763427634278e-05, "loss": 0.1871, "step": 8920 }, { "epoch": 1.83, "learning_rate": 1.633866338663387e-05, "loss": 0.2605, "step": 8930 }, { "epoch": 1.83, "learning_rate": 1.6334563345633458e-05, "loss": 0.2876, "step": 8940 }, { "epoch": 1.83, "learning_rate": 1.633046330463305e-05, "loss": 0.2929, "step": 8950 }, { "epoch": 1.84, "learning_rate": 1.6326363263632637e-05, "loss": 0.2511, "step": 8960 }, { "epoch": 1.84, "learning_rate": 1.632226322263223e-05, "loss": 0.2828, "step": 8970 }, { "epoch": 1.84, "learning_rate": 1.6318163181631817e-05, "loss": 0.2246, "step": 8980 }, { "epoch": 1.84, "learning_rate": 1.631406314063141e-05, "loss": 0.3066, "step": 8990 }, { "epoch": 1.85, "learning_rate": 1.6309963099630997e-05, "loss": 0.2639, "step": 9000 }, { "epoch": 1.85, "learning_rate": 1.6305863058630585e-05, "loss": 0.2602, "step": 9010 }, { "epoch": 1.85, "learning_rate": 1.6301763017630177e-05, "loss": 0.2986, "step": 9020 }, { "epoch": 1.85, "learning_rate": 1.629766297662977e-05, "loss": 0.2612, "step": 9030 }, { "epoch": 1.85, "learning_rate": 1.6293562935629357e-05, "loss": 0.313, "step": 9040 }, { "epoch": 1.86, "learning_rate": 1.6289462894628948e-05, "loss": 0.2938, "step": 9050 }, { "epoch": 1.86, "learning_rate": 1.6285362853628536e-05, "loss": 0.2961, "step": 9060 }, { "epoch": 1.86, "learning_rate": 1.6281262812628128e-05, "loss": 0.3112, "step": 9070 }, { "epoch": 1.86, "learning_rate": 1.6277162771627716e-05, "loss": 0.252, "step": 9080 }, { "epoch": 1.86, "learning_rate": 1.6273062730627308e-05, "loss": 0.267, "step": 9090 }, { "epoch": 1.87, "learning_rate": 1.6268962689626896e-05, "loss": 0.2124, "step": 9100 }, { "epoch": 1.87, "learning_rate": 1.6264862648626488e-05, "loss": 0.2857, "step": 9110 }, { "epoch": 1.87, "learning_rate": 1.6260762607626076e-05, "loss": 0.282, "step": 9120 }, { "epoch": 1.87, "learning_rate": 1.6256662566625667e-05, "loss": 0.2042, "step": 9130 }, { "epoch": 1.87, "learning_rate": 1.625256252562526e-05, "loss": 0.241, "step": 9140 }, { "epoch": 1.88, "learning_rate": 1.6248462484624847e-05, "loss": 0.213, "step": 9150 }, { "epoch": 1.88, "learning_rate": 1.624436244362444e-05, "loss": 0.2417, "step": 9160 }, { "epoch": 1.88, "learning_rate": 1.6240262402624027e-05, "loss": 0.2919, "step": 9170 }, { "epoch": 1.88, "learning_rate": 1.623616236162362e-05, "loss": 0.3472, "step": 9180 }, { "epoch": 1.88, "learning_rate": 1.6232062320623207e-05, "loss": 0.2728, "step": 9190 }, { "epoch": 1.89, "learning_rate": 1.6227962279622798e-05, "loss": 0.2013, "step": 9200 }, { "epoch": 1.89, "learning_rate": 1.6223862238622386e-05, "loss": 0.2083, "step": 9210 }, { "epoch": 1.89, "learning_rate": 1.6219762197621978e-05, "loss": 0.1891, "step": 9220 }, { "epoch": 1.89, "learning_rate": 1.6215662156621566e-05, "loss": 0.3281, "step": 9230 }, { "epoch": 1.89, "learning_rate": 1.6211562115621158e-05, "loss": 0.2679, "step": 9240 }, { "epoch": 1.9, "learning_rate": 1.620746207462075e-05, "loss": 0.2518, "step": 9250 }, { "epoch": 1.9, "learning_rate": 1.6203362033620338e-05, "loss": 0.2237, "step": 9260 }, { "epoch": 1.9, "learning_rate": 1.619926199261993e-05, "loss": 0.3227, "step": 9270 }, { "epoch": 1.9, "learning_rate": 1.6195161951619517e-05, "loss": 0.3425, "step": 9280 }, { "epoch": 1.9, "learning_rate": 1.619106191061911e-05, "loss": 0.3436, "step": 9290 }, { "epoch": 1.91, "learning_rate": 1.6186961869618697e-05, "loss": 0.3434, "step": 9300 }, { "epoch": 1.91, "learning_rate": 1.618286182861829e-05, "loss": 0.359, "step": 9310 }, { "epoch": 1.91, "learning_rate": 1.6178761787617877e-05, "loss": 0.2839, "step": 9320 }, { "epoch": 1.91, "learning_rate": 1.617466174661747e-05, "loss": 0.3169, "step": 9330 }, { "epoch": 1.91, "learning_rate": 1.6170561705617057e-05, "loss": 0.3363, "step": 9340 }, { "epoch": 1.92, "learning_rate": 1.616646166461665e-05, "loss": 0.265, "step": 9350 }, { "epoch": 1.92, "learning_rate": 1.616236162361624e-05, "loss": 0.2993, "step": 9360 }, { "epoch": 1.92, "learning_rate": 1.6158261582615828e-05, "loss": 0.2394, "step": 9370 }, { "epoch": 1.92, "learning_rate": 1.615416154161542e-05, "loss": 0.3384, "step": 9380 }, { "epoch": 1.92, "learning_rate": 1.6150061500615008e-05, "loss": 0.2233, "step": 9390 }, { "epoch": 1.93, "learning_rate": 1.6145961459614596e-05, "loss": 0.2887, "step": 9400 }, { "epoch": 1.93, "learning_rate": 1.6141861418614188e-05, "loss": 0.2957, "step": 9410 }, { "epoch": 1.93, "learning_rate": 1.6137761377613776e-05, "loss": 0.3324, "step": 9420 }, { "epoch": 1.93, "learning_rate": 1.6133661336613368e-05, "loss": 0.3194, "step": 9430 }, { "epoch": 1.94, "learning_rate": 1.6129561295612956e-05, "loss": 0.2916, "step": 9440 }, { "epoch": 1.94, "learning_rate": 1.6125461254612547e-05, "loss": 0.2697, "step": 9450 }, { "epoch": 1.94, "learning_rate": 1.6121361213612136e-05, "loss": 0.2893, "step": 9460 }, { "epoch": 1.94, "learning_rate": 1.6117261172611727e-05, "loss": 0.3424, "step": 9470 }, { "epoch": 1.94, "learning_rate": 1.6113161131611315e-05, "loss": 0.2918, "step": 9480 }, { "epoch": 1.95, "learning_rate": 1.6109061090610907e-05, "loss": 0.2721, "step": 9490 }, { "epoch": 1.95, "learning_rate": 1.6104961049610495e-05, "loss": 0.2785, "step": 9500 }, { "epoch": 1.95, "learning_rate": 1.6100861008610087e-05, "loss": 0.2027, "step": 9510 }, { "epoch": 1.95, "learning_rate": 1.609676096760968e-05, "loss": 0.2636, "step": 9520 }, { "epoch": 1.95, "learning_rate": 1.6092660926609267e-05, "loss": 0.3672, "step": 9530 }, { "epoch": 1.96, "learning_rate": 1.6088560885608858e-05, "loss": 0.1914, "step": 9540 }, { "epoch": 1.96, "learning_rate": 1.6084460844608446e-05, "loss": 0.1814, "step": 9550 }, { "epoch": 1.96, "learning_rate": 1.6080360803608038e-05, "loss": 0.3411, "step": 9560 }, { "epoch": 1.96, "learning_rate": 1.6076260762607626e-05, "loss": 0.198, "step": 9570 }, { "epoch": 1.96, "learning_rate": 1.6072160721607218e-05, "loss": 0.2893, "step": 9580 }, { "epoch": 1.97, "learning_rate": 1.6068060680606806e-05, "loss": 0.3293, "step": 9590 }, { "epoch": 1.97, "learning_rate": 1.6063960639606398e-05, "loss": 0.208, "step": 9600 }, { "epoch": 1.97, "learning_rate": 1.6059860598605986e-05, "loss": 0.2478, "step": 9610 }, { "epoch": 1.97, "learning_rate": 1.6055760557605577e-05, "loss": 0.2557, "step": 9620 }, { "epoch": 1.97, "learning_rate": 1.605166051660517e-05, "loss": 0.2971, "step": 9630 }, { "epoch": 1.98, "learning_rate": 1.6047560475604757e-05, "loss": 0.2347, "step": 9640 }, { "epoch": 1.98, "learning_rate": 1.604346043460435e-05, "loss": 0.3245, "step": 9650 }, { "epoch": 1.98, "learning_rate": 1.6039360393603937e-05, "loss": 0.2784, "step": 9660 }, { "epoch": 1.98, "learning_rate": 1.603526035260353e-05, "loss": 0.3224, "step": 9670 }, { "epoch": 1.98, "learning_rate": 1.6031160311603117e-05, "loss": 0.2963, "step": 9680 }, { "epoch": 1.99, "learning_rate": 1.602706027060271e-05, "loss": 0.2522, "step": 9690 }, { "epoch": 1.99, "learning_rate": 1.6022960229602297e-05, "loss": 0.2716, "step": 9700 }, { "epoch": 1.99, "learning_rate": 1.6018860188601888e-05, "loss": 0.3401, "step": 9710 }, { "epoch": 1.99, "learning_rate": 1.6014760147601476e-05, "loss": 0.2346, "step": 9720 }, { "epoch": 1.99, "learning_rate": 1.6010660106601068e-05, "loss": 0.3074, "step": 9730 }, { "epoch": 2.0, "learning_rate": 1.600656006560066e-05, "loss": 0.309, "step": 9740 }, { "epoch": 2.0, "learning_rate": 1.6002460024600248e-05, "loss": 0.2596, "step": 9750 }, { "epoch": 2.0, "eval_FN": 547, "eval_FP": 2203, "eval_TN": 13556, "eval_TP": 4359, "eval_accuracy": 0.8669247519961287, "eval_f1": 0.7602023020579002, "eval_loss": 0.3200523853302002, "eval_precision": 0.664279183175861, "eval_recall": 0.8885038728088055, "eval_runtime": 141.8219, "eval_samples_per_second": 145.711, "eval_steps_per_second": 9.11, "step": 9756 }, { "epoch": 2.0, "learning_rate": 1.599835998359984e-05, "loss": 0.2661, "step": 9760 }, { "epoch": 2.0, "learning_rate": 1.5994259942599428e-05, "loss": 0.1798, "step": 9770 }, { "epoch": 2.0, "learning_rate": 1.599015990159902e-05, "loss": 0.1933, "step": 9780 }, { "epoch": 2.01, "learning_rate": 1.5986059860598607e-05, "loss": 0.2763, "step": 9790 }, { "epoch": 2.01, "learning_rate": 1.5981959819598196e-05, "loss": 0.2088, "step": 9800 }, { "epoch": 2.01, "learning_rate": 1.5977859778597787e-05, "loss": 0.2027, "step": 9810 }, { "epoch": 2.01, "learning_rate": 1.5973759737597375e-05, "loss": 0.1984, "step": 9820 }, { "epoch": 2.02, "learning_rate": 1.5969659696596967e-05, "loss": 0.2224, "step": 9830 }, { "epoch": 2.02, "learning_rate": 1.5965559655596555e-05, "loss": 0.28, "step": 9840 }, { "epoch": 2.02, "learning_rate": 1.5961459614596147e-05, "loss": 0.3387, "step": 9850 }, { "epoch": 2.02, "learning_rate": 1.5957359573595735e-05, "loss": 0.1445, "step": 9860 }, { "epoch": 2.02, "learning_rate": 1.5953259532595326e-05, "loss": 0.2059, "step": 9870 }, { "epoch": 2.03, "learning_rate": 1.5949159491594918e-05, "loss": 0.2499, "step": 9880 }, { "epoch": 2.03, "learning_rate": 1.5945059450594506e-05, "loss": 0.2106, "step": 9890 }, { "epoch": 2.03, "learning_rate": 1.5940959409594098e-05, "loss": 0.1347, "step": 9900 }, { "epoch": 2.03, "learning_rate": 1.5936859368593686e-05, "loss": 0.1636, "step": 9910 }, { "epoch": 2.03, "learning_rate": 1.5932759327593278e-05, "loss": 0.1683, "step": 9920 }, { "epoch": 2.04, "learning_rate": 1.5928659286592866e-05, "loss": 0.3341, "step": 9930 }, { "epoch": 2.04, "learning_rate": 1.5924559245592457e-05, "loss": 0.3377, "step": 9940 }, { "epoch": 2.04, "learning_rate": 1.5920459204592046e-05, "loss": 0.2643, "step": 9950 }, { "epoch": 2.04, "learning_rate": 1.5916359163591637e-05, "loss": 0.2341, "step": 9960 }, { "epoch": 2.04, "learning_rate": 1.5912259122591225e-05, "loss": 0.2009, "step": 9970 }, { "epoch": 2.05, "learning_rate": 1.5908159081590817e-05, "loss": 0.2459, "step": 9980 }, { "epoch": 2.05, "learning_rate": 1.590405904059041e-05, "loss": 0.2195, "step": 9990 }, { "epoch": 2.05, "learning_rate": 1.5899958999589997e-05, "loss": 0.2228, "step": 10000 }, { "epoch": 2.05, "learning_rate": 1.589585895858959e-05, "loss": 0.268, "step": 10010 }, { "epoch": 2.05, "learning_rate": 1.5891758917589177e-05, "loss": 0.2342, "step": 10020 }, { "epoch": 2.06, "learning_rate": 1.5887658876588768e-05, "loss": 0.1817, "step": 10030 }, { "epoch": 2.06, "learning_rate": 1.5883558835588356e-05, "loss": 0.239, "step": 10040 }, { "epoch": 2.06, "learning_rate": 1.5879458794587948e-05, "loss": 0.1946, "step": 10050 }, { "epoch": 2.06, "learning_rate": 1.5875358753587536e-05, "loss": 0.2734, "step": 10060 }, { "epoch": 2.06, "learning_rate": 1.5871258712587128e-05, "loss": 0.3131, "step": 10070 }, { "epoch": 2.07, "learning_rate": 1.5867158671586716e-05, "loss": 0.2231, "step": 10080 }, { "epoch": 2.07, "learning_rate": 1.5863058630586308e-05, "loss": 0.1896, "step": 10090 }, { "epoch": 2.07, "learning_rate": 1.58589585895859e-05, "loss": 0.2269, "step": 10100 }, { "epoch": 2.07, "learning_rate": 1.5854858548585487e-05, "loss": 0.2926, "step": 10110 }, { "epoch": 2.07, "learning_rate": 1.585075850758508e-05, "loss": 0.3849, "step": 10120 }, { "epoch": 2.08, "learning_rate": 1.5846658466584667e-05, "loss": 0.249, "step": 10130 }, { "epoch": 2.08, "learning_rate": 1.584255842558426e-05, "loss": 0.201, "step": 10140 }, { "epoch": 2.08, "learning_rate": 1.5838458384583847e-05, "loss": 0.2842, "step": 10150 }, { "epoch": 2.08, "learning_rate": 1.583435834358344e-05, "loss": 0.1842, "step": 10160 }, { "epoch": 2.08, "learning_rate": 1.5830258302583027e-05, "loss": 0.3367, "step": 10170 }, { "epoch": 2.09, "learning_rate": 1.582615826158262e-05, "loss": 0.1745, "step": 10180 }, { "epoch": 2.09, "learning_rate": 1.5822058220582207e-05, "loss": 0.2534, "step": 10190 }, { "epoch": 2.09, "learning_rate": 1.5817958179581795e-05, "loss": 0.2448, "step": 10200 }, { "epoch": 2.09, "learning_rate": 1.5813858138581386e-05, "loss": 0.2567, "step": 10210 }, { "epoch": 2.1, "learning_rate": 1.5809758097580975e-05, "loss": 0.2998, "step": 10220 }, { "epoch": 2.1, "learning_rate": 1.5805658056580566e-05, "loss": 0.2234, "step": 10230 }, { "epoch": 2.1, "learning_rate": 1.5801558015580158e-05, "loss": 0.1947, "step": 10240 }, { "epoch": 2.1, "learning_rate": 1.5797457974579746e-05, "loss": 0.1565, "step": 10250 }, { "epoch": 2.1, "learning_rate": 1.5793357933579338e-05, "loss": 0.255, "step": 10260 }, { "epoch": 2.11, "learning_rate": 1.5789257892578926e-05, "loss": 0.1667, "step": 10270 }, { "epoch": 2.11, "learning_rate": 1.5785157851578517e-05, "loss": 0.3159, "step": 10280 }, { "epoch": 2.11, "learning_rate": 1.5781057810578106e-05, "loss": 0.241, "step": 10290 }, { "epoch": 2.11, "learning_rate": 1.5776957769577697e-05, "loss": 0.1454, "step": 10300 }, { "epoch": 2.11, "learning_rate": 1.5772857728577285e-05, "loss": 0.1936, "step": 10310 }, { "epoch": 2.12, "learning_rate": 1.5768757687576877e-05, "loss": 0.2079, "step": 10320 }, { "epoch": 2.12, "learning_rate": 1.5764657646576465e-05, "loss": 0.2289, "step": 10330 }, { "epoch": 2.12, "learning_rate": 1.5760557605576057e-05, "loss": 0.1089, "step": 10340 }, { "epoch": 2.12, "learning_rate": 1.5756457564575648e-05, "loss": 0.2705, "step": 10350 }, { "epoch": 2.12, "learning_rate": 1.5752357523575237e-05, "loss": 0.1477, "step": 10360 }, { "epoch": 2.13, "learning_rate": 1.5748257482574828e-05, "loss": 0.3715, "step": 10370 }, { "epoch": 2.13, "learning_rate": 1.5744157441574416e-05, "loss": 0.2487, "step": 10380 }, { "epoch": 2.13, "learning_rate": 1.5740057400574008e-05, "loss": 0.1802, "step": 10390 }, { "epoch": 2.13, "learning_rate": 1.5735957359573596e-05, "loss": 0.1736, "step": 10400 }, { "epoch": 2.13, "learning_rate": 1.5731857318573188e-05, "loss": 0.2157, "step": 10410 }, { "epoch": 2.14, "learning_rate": 1.5727757277572776e-05, "loss": 0.3245, "step": 10420 }, { "epoch": 2.14, "learning_rate": 1.5723657236572367e-05, "loss": 0.2563, "step": 10430 }, { "epoch": 2.14, "learning_rate": 1.5719557195571956e-05, "loss": 0.1857, "step": 10440 }, { "epoch": 2.14, "learning_rate": 1.5715457154571547e-05, "loss": 0.2505, "step": 10450 }, { "epoch": 2.14, "learning_rate": 1.571135711357114e-05, "loss": 0.2125, "step": 10460 }, { "epoch": 2.15, "learning_rate": 1.5707257072570727e-05, "loss": 0.2368, "step": 10470 }, { "epoch": 2.15, "learning_rate": 1.570315703157032e-05, "loss": 0.2723, "step": 10480 }, { "epoch": 2.15, "learning_rate": 1.5699056990569907e-05, "loss": 0.1742, "step": 10490 }, { "epoch": 2.15, "learning_rate": 1.56949569495695e-05, "loss": 0.2199, "step": 10500 }, { "epoch": 2.15, "learning_rate": 1.5690856908569087e-05, "loss": 0.3337, "step": 10510 }, { "epoch": 2.16, "learning_rate": 1.5686756867568678e-05, "loss": 0.2653, "step": 10520 }, { "epoch": 2.16, "learning_rate": 1.5682656826568266e-05, "loss": 0.298, "step": 10530 }, { "epoch": 2.16, "learning_rate": 1.5678556785567858e-05, "loss": 0.1943, "step": 10540 }, { "epoch": 2.16, "learning_rate": 1.5674456744567446e-05, "loss": 0.2348, "step": 10550 }, { "epoch": 2.16, "learning_rate": 1.5670356703567038e-05, "loss": 0.1987, "step": 10560 }, { "epoch": 2.17, "learning_rate": 1.566625666256663e-05, "loss": 0.2876, "step": 10570 }, { "epoch": 2.17, "learning_rate": 1.5662156621566218e-05, "loss": 0.2224, "step": 10580 }, { "epoch": 2.17, "learning_rate": 1.565805658056581e-05, "loss": 0.2753, "step": 10590 }, { "epoch": 2.17, "learning_rate": 1.5653956539565397e-05, "loss": 0.2978, "step": 10600 }, { "epoch": 2.18, "learning_rate": 1.5649856498564986e-05, "loss": 0.2399, "step": 10610 }, { "epoch": 2.18, "learning_rate": 1.5645756457564577e-05, "loss": 0.324, "step": 10620 }, { "epoch": 2.18, "learning_rate": 1.5641656416564165e-05, "loss": 0.2841, "step": 10630 }, { "epoch": 2.18, "learning_rate": 1.5637556375563757e-05, "loss": 0.2137, "step": 10640 }, { "epoch": 2.18, "learning_rate": 1.5633456334563345e-05, "loss": 0.3054, "step": 10650 }, { "epoch": 2.19, "learning_rate": 1.5629356293562937e-05, "loss": 0.1887, "step": 10660 }, { "epoch": 2.19, "learning_rate": 1.5625256252562525e-05, "loss": 0.1515, "step": 10670 }, { "epoch": 2.19, "learning_rate": 1.5621156211562117e-05, "loss": 0.2668, "step": 10680 }, { "epoch": 2.19, "learning_rate": 1.5617056170561705e-05, "loss": 0.2393, "step": 10690 }, { "epoch": 2.19, "learning_rate": 1.5612956129561296e-05, "loss": 0.2224, "step": 10700 }, { "epoch": 2.2, "learning_rate": 1.5608856088560888e-05, "loss": 0.3261, "step": 10710 }, { "epoch": 2.2, "learning_rate": 1.5604756047560476e-05, "loss": 0.1507, "step": 10720 }, { "epoch": 2.2, "learning_rate": 1.5600656006560068e-05, "loss": 0.2152, "step": 10730 }, { "epoch": 2.2, "learning_rate": 1.5596555965559656e-05, "loss": 0.1515, "step": 10740 }, { "epoch": 2.2, "learning_rate": 1.5592455924559248e-05, "loss": 0.2073, "step": 10750 }, { "epoch": 2.21, "learning_rate": 1.5588355883558836e-05, "loss": 0.1672, "step": 10760 }, { "epoch": 2.21, "learning_rate": 1.5584255842558427e-05, "loss": 0.2314, "step": 10770 }, { "epoch": 2.21, "learning_rate": 1.5580155801558016e-05, "loss": 0.301, "step": 10780 }, { "epoch": 2.21, "learning_rate": 1.5576055760557607e-05, "loss": 0.2549, "step": 10790 }, { "epoch": 2.21, "learning_rate": 1.5571955719557195e-05, "loss": 0.2023, "step": 10800 }, { "epoch": 2.22, "learning_rate": 1.5567855678556787e-05, "loss": 0.201, "step": 10810 }, { "epoch": 2.22, "learning_rate": 1.556375563755638e-05, "loss": 0.1818, "step": 10820 }, { "epoch": 2.22, "learning_rate": 1.5559655596555967e-05, "loss": 0.3767, "step": 10830 }, { "epoch": 2.22, "learning_rate": 1.555555555555556e-05, "loss": 0.2321, "step": 10840 }, { "epoch": 2.22, "learning_rate": 1.5551455514555147e-05, "loss": 0.1917, "step": 10850 }, { "epoch": 2.23, "learning_rate": 1.5547355473554738e-05, "loss": 0.1714, "step": 10860 }, { "epoch": 2.23, "learning_rate": 1.5543255432554326e-05, "loss": 0.2768, "step": 10870 }, { "epoch": 2.23, "learning_rate": 1.5539155391553918e-05, "loss": 0.2687, "step": 10880 }, { "epoch": 2.23, "learning_rate": 1.5535055350553506e-05, "loss": 0.2686, "step": 10890 }, { "epoch": 2.23, "learning_rate": 1.5530955309553098e-05, "loss": 0.201, "step": 10900 }, { "epoch": 2.24, "learning_rate": 1.5526855268552686e-05, "loss": 0.1773, "step": 10910 }, { "epoch": 2.24, "learning_rate": 1.5522755227552278e-05, "loss": 0.3226, "step": 10920 }, { "epoch": 2.24, "learning_rate": 1.551865518655187e-05, "loss": 0.1681, "step": 10930 }, { "epoch": 2.24, "learning_rate": 1.5514555145551457e-05, "loss": 0.1741, "step": 10940 }, { "epoch": 2.24, "learning_rate": 1.551045510455105e-05, "loss": 0.2206, "step": 10950 }, { "epoch": 2.25, "learning_rate": 1.5506355063550637e-05, "loss": 0.3214, "step": 10960 }, { "epoch": 2.25, "learning_rate": 1.550225502255023e-05, "loss": 0.2533, "step": 10970 }, { "epoch": 2.25, "learning_rate": 1.5498154981549817e-05, "loss": 0.3544, "step": 10980 }, { "epoch": 2.25, "learning_rate": 1.549405494054941e-05, "loss": 0.222, "step": 10990 }, { "epoch": 2.26, "learning_rate": 1.5489954899548997e-05, "loss": 0.2215, "step": 11000 }, { "epoch": 2.26, "learning_rate": 1.5485854858548585e-05, "loss": 0.2671, "step": 11010 }, { "epoch": 2.26, "learning_rate": 1.5481754817548176e-05, "loss": 0.3292, "step": 11020 }, { "epoch": 2.26, "learning_rate": 1.5477654776547765e-05, "loss": 0.3364, "step": 11030 }, { "epoch": 2.26, "learning_rate": 1.5473554735547356e-05, "loss": 0.3196, "step": 11040 }, { "epoch": 2.27, "learning_rate": 1.5469454694546944e-05, "loss": 0.1962, "step": 11050 }, { "epoch": 2.27, "learning_rate": 1.5465354653546536e-05, "loss": 0.1304, "step": 11060 }, { "epoch": 2.27, "learning_rate": 1.5461254612546128e-05, "loss": 0.3595, "step": 11070 }, { "epoch": 2.27, "learning_rate": 1.5457154571545716e-05, "loss": 0.2036, "step": 11080 }, { "epoch": 2.27, "learning_rate": 1.5453054530545307e-05, "loss": 0.1992, "step": 11090 }, { "epoch": 2.28, "learning_rate": 1.5448954489544896e-05, "loss": 0.2918, "step": 11100 }, { "epoch": 2.28, "learning_rate": 1.5444854448544487e-05, "loss": 0.3134, "step": 11110 }, { "epoch": 2.28, "learning_rate": 1.5440754407544075e-05, "loss": 0.1989, "step": 11120 }, { "epoch": 2.28, "learning_rate": 1.5436654366543667e-05, "loss": 0.1745, "step": 11130 }, { "epoch": 2.28, "learning_rate": 1.5432554325543255e-05, "loss": 0.2083, "step": 11140 }, { "epoch": 2.29, "learning_rate": 1.5428454284542847e-05, "loss": 0.2667, "step": 11150 }, { "epoch": 2.29, "learning_rate": 1.5424354243542435e-05, "loss": 0.1723, "step": 11160 }, { "epoch": 2.29, "learning_rate": 1.5420254202542027e-05, "loss": 0.265, "step": 11170 }, { "epoch": 2.29, "learning_rate": 1.5416154161541618e-05, "loss": 0.1964, "step": 11180 }, { "epoch": 2.29, "learning_rate": 1.5412054120541206e-05, "loss": 0.2361, "step": 11190 }, { "epoch": 2.3, "learning_rate": 1.5407954079540798e-05, "loss": 0.2264, "step": 11200 }, { "epoch": 2.3, "learning_rate": 1.5403854038540386e-05, "loss": 0.2653, "step": 11210 }, { "epoch": 2.3, "learning_rate": 1.5399753997539978e-05, "loss": 0.2895, "step": 11220 }, { "epoch": 2.3, "learning_rate": 1.5395653956539566e-05, "loss": 0.3363, "step": 11230 }, { "epoch": 2.3, "learning_rate": 1.5391553915539158e-05, "loss": 0.2911, "step": 11240 }, { "epoch": 2.31, "learning_rate": 1.5387453874538746e-05, "loss": 0.2376, "step": 11250 }, { "epoch": 2.31, "learning_rate": 1.5383353833538337e-05, "loss": 0.2621, "step": 11260 }, { "epoch": 2.31, "learning_rate": 1.5379253792537926e-05, "loss": 0.245, "step": 11270 }, { "epoch": 2.31, "learning_rate": 1.5375153751537517e-05, "loss": 0.2225, "step": 11280 }, { "epoch": 2.31, "learning_rate": 1.537105371053711e-05, "loss": 0.2378, "step": 11290 }, { "epoch": 2.32, "learning_rate": 1.5366953669536697e-05, "loss": 0.2617, "step": 11300 }, { "epoch": 2.32, "learning_rate": 1.536285362853629e-05, "loss": 0.2535, "step": 11310 }, { "epoch": 2.32, "learning_rate": 1.5358753587535877e-05, "loss": 0.2708, "step": 11320 }, { "epoch": 2.32, "learning_rate": 1.535465354653547e-05, "loss": 0.2043, "step": 11330 }, { "epoch": 2.32, "learning_rate": 1.5350553505535057e-05, "loss": 0.1841, "step": 11340 }, { "epoch": 2.33, "learning_rate": 1.5346453464534648e-05, "loss": 0.1768, "step": 11350 }, { "epoch": 2.33, "learning_rate": 1.5342353423534236e-05, "loss": 0.3239, "step": 11360 }, { "epoch": 2.33, "learning_rate": 1.5338253382533828e-05, "loss": 0.2818, "step": 11370 }, { "epoch": 2.33, "learning_rate": 1.5334153341533416e-05, "loss": 0.2335, "step": 11380 }, { "epoch": 2.33, "learning_rate": 1.5330053300533008e-05, "loss": 0.3696, "step": 11390 }, { "epoch": 2.34, "learning_rate": 1.5325953259532596e-05, "loss": 0.2409, "step": 11400 }, { "epoch": 2.34, "learning_rate": 1.5321853218532184e-05, "loss": 0.2433, "step": 11410 }, { "epoch": 2.34, "learning_rate": 1.5317753177531776e-05, "loss": 0.2597, "step": 11420 }, { "epoch": 2.34, "learning_rate": 1.5313653136531367e-05, "loss": 0.3951, "step": 11430 }, { "epoch": 2.35, "learning_rate": 1.5309553095530956e-05, "loss": 0.1427, "step": 11440 }, { "epoch": 2.35, "learning_rate": 1.5305453054530547e-05, "loss": 0.2319, "step": 11450 }, { "epoch": 2.35, "learning_rate": 1.5301353013530135e-05, "loss": 0.4135, "step": 11460 }, { "epoch": 2.35, "learning_rate": 1.5297252972529727e-05, "loss": 0.183, "step": 11470 }, { "epoch": 2.35, "learning_rate": 1.5293152931529315e-05, "loss": 0.1778, "step": 11480 }, { "epoch": 2.36, "learning_rate": 1.5289052890528907e-05, "loss": 0.2594, "step": 11490 }, { "epoch": 2.36, "learning_rate": 1.5284952849528495e-05, "loss": 0.2846, "step": 11500 }, { "epoch": 2.36, "learning_rate": 1.5280852808528087e-05, "loss": 0.1374, "step": 11510 }, { "epoch": 2.36, "learning_rate": 1.5276752767527675e-05, "loss": 0.2367, "step": 11520 }, { "epoch": 2.36, "learning_rate": 1.5272652726527266e-05, "loss": 0.3575, "step": 11530 }, { "epoch": 2.37, "learning_rate": 1.5268552685526858e-05, "loss": 0.1898, "step": 11540 }, { "epoch": 2.37, "learning_rate": 1.5264452644526446e-05, "loss": 0.1588, "step": 11550 }, { "epoch": 2.37, "learning_rate": 1.5260352603526038e-05, "loss": 0.2387, "step": 11560 }, { "epoch": 2.37, "learning_rate": 1.5256252562525626e-05, "loss": 0.2197, "step": 11570 }, { "epoch": 2.37, "learning_rate": 1.5252152521525216e-05, "loss": 0.2214, "step": 11580 }, { "epoch": 2.38, "learning_rate": 1.5248052480524806e-05, "loss": 0.2563, "step": 11590 }, { "epoch": 2.38, "learning_rate": 1.5243952439524397e-05, "loss": 0.2317, "step": 11600 }, { "epoch": 2.38, "learning_rate": 1.5239852398523987e-05, "loss": 0.1931, "step": 11610 }, { "epoch": 2.38, "learning_rate": 1.5235752357523577e-05, "loss": 0.1926, "step": 11620 }, { "epoch": 2.38, "learning_rate": 1.5231652316523167e-05, "loss": 0.1417, "step": 11630 }, { "epoch": 2.39, "learning_rate": 1.5227552275522757e-05, "loss": 0.201, "step": 11640 }, { "epoch": 2.39, "learning_rate": 1.5223452234522347e-05, "loss": 0.2805, "step": 11650 }, { "epoch": 2.39, "learning_rate": 1.5219352193521937e-05, "loss": 0.2224, "step": 11660 }, { "epoch": 2.39, "learning_rate": 1.5215252152521527e-05, "loss": 0.2466, "step": 11670 }, { "epoch": 2.39, "learning_rate": 1.5211152111521116e-05, "loss": 0.2268, "step": 11680 }, { "epoch": 2.4, "learning_rate": 1.5207052070520706e-05, "loss": 0.269, "step": 11690 }, { "epoch": 2.4, "learning_rate": 1.5202952029520296e-05, "loss": 0.191, "step": 11700 }, { "epoch": 2.4, "learning_rate": 1.5198851988519888e-05, "loss": 0.1349, "step": 11710 }, { "epoch": 2.4, "learning_rate": 1.5194751947519478e-05, "loss": 0.2739, "step": 11720 }, { "epoch": 2.4, "learning_rate": 1.5190651906519068e-05, "loss": 0.2273, "step": 11730 }, { "epoch": 2.41, "learning_rate": 1.5186551865518658e-05, "loss": 0.3194, "step": 11740 }, { "epoch": 2.41, "learning_rate": 1.5182451824518247e-05, "loss": 0.1801, "step": 11750 }, { "epoch": 2.41, "learning_rate": 1.5178351783517837e-05, "loss": 0.2901, "step": 11760 }, { "epoch": 2.41, "learning_rate": 1.5174251742517427e-05, "loss": 0.2195, "step": 11770 }, { "epoch": 2.41, "learning_rate": 1.5170151701517017e-05, "loss": 0.2136, "step": 11780 }, { "epoch": 2.42, "learning_rate": 1.5166051660516607e-05, "loss": 0.2798, "step": 11790 }, { "epoch": 2.42, "learning_rate": 1.5161951619516195e-05, "loss": 0.1719, "step": 11800 }, { "epoch": 2.42, "learning_rate": 1.5157851578515785e-05, "loss": 0.2679, "step": 11810 }, { "epoch": 2.42, "learning_rate": 1.5153751537515375e-05, "loss": 0.2328, "step": 11820 }, { "epoch": 2.43, "learning_rate": 1.5149651496514965e-05, "loss": 0.3476, "step": 11830 }, { "epoch": 2.43, "learning_rate": 1.5145551455514555e-05, "loss": 0.2368, "step": 11840 }, { "epoch": 2.43, "learning_rate": 1.5141451414514146e-05, "loss": 0.2692, "step": 11850 }, { "epoch": 2.43, "learning_rate": 1.5137351373513736e-05, "loss": 0.241, "step": 11860 }, { "epoch": 2.43, "learning_rate": 1.5133251332513326e-05, "loss": 0.2552, "step": 11870 }, { "epoch": 2.44, "learning_rate": 1.5129151291512916e-05, "loss": 0.1713, "step": 11880 }, { "epoch": 2.44, "learning_rate": 1.5125051250512506e-05, "loss": 0.1215, "step": 11890 }, { "epoch": 2.44, "learning_rate": 1.5120951209512096e-05, "loss": 0.2664, "step": 11900 }, { "epoch": 2.44, "learning_rate": 1.5116851168511686e-05, "loss": 0.2556, "step": 11910 }, { "epoch": 2.44, "learning_rate": 1.5112751127511276e-05, "loss": 0.2222, "step": 11920 }, { "epoch": 2.45, "learning_rate": 1.5108651086510866e-05, "loss": 0.188, "step": 11930 }, { "epoch": 2.45, "learning_rate": 1.5104551045510455e-05, "loss": 0.2723, "step": 11940 }, { "epoch": 2.45, "learning_rate": 1.5100451004510045e-05, "loss": 0.2944, "step": 11950 }, { "epoch": 2.45, "learning_rate": 1.5096350963509637e-05, "loss": 0.2212, "step": 11960 }, { "epoch": 2.45, "learning_rate": 1.5092250922509227e-05, "loss": 0.2671, "step": 11970 }, { "epoch": 2.46, "learning_rate": 1.5088150881508817e-05, "loss": 0.2521, "step": 11980 }, { "epoch": 2.46, "learning_rate": 1.5084050840508407e-05, "loss": 0.2679, "step": 11990 }, { "epoch": 2.46, "learning_rate": 1.5079950799507997e-05, "loss": 0.1764, "step": 12000 }, { "epoch": 2.46, "learning_rate": 1.5075850758507586e-05, "loss": 0.2392, "step": 12010 }, { "epoch": 2.46, "learning_rate": 1.5071750717507176e-05, "loss": 0.2, "step": 12020 }, { "epoch": 2.47, "learning_rate": 1.5067650676506766e-05, "loss": 0.1725, "step": 12030 }, { "epoch": 2.47, "learning_rate": 1.5063550635506356e-05, "loss": 0.2718, "step": 12040 }, { "epoch": 2.47, "learning_rate": 1.5059450594505946e-05, "loss": 0.3752, "step": 12050 }, { "epoch": 2.47, "learning_rate": 1.5055350553505536e-05, "loss": 0.2506, "step": 12060 }, { "epoch": 2.47, "learning_rate": 1.5051250512505128e-05, "loss": 0.1697, "step": 12070 }, { "epoch": 2.48, "learning_rate": 1.5047150471504717e-05, "loss": 0.2188, "step": 12080 }, { "epoch": 2.48, "learning_rate": 1.5043050430504307e-05, "loss": 0.2313, "step": 12090 }, { "epoch": 2.48, "learning_rate": 1.5038950389503897e-05, "loss": 0.2553, "step": 12100 }, { "epoch": 2.48, "learning_rate": 1.5034850348503487e-05, "loss": 0.2869, "step": 12110 }, { "epoch": 2.48, "learning_rate": 1.5030750307503077e-05, "loss": 0.2451, "step": 12120 }, { "epoch": 2.49, "learning_rate": 1.5026650266502667e-05, "loss": 0.123, "step": 12130 }, { "epoch": 2.49, "learning_rate": 1.5022550225502257e-05, "loss": 0.387, "step": 12140 }, { "epoch": 2.49, "learning_rate": 1.5018450184501847e-05, "loss": 0.2443, "step": 12150 }, { "epoch": 2.49, "learning_rate": 1.5014350143501437e-05, "loss": 0.2713, "step": 12160 }, { "epoch": 2.49, "learning_rate": 1.5010250102501026e-05, "loss": 0.2945, "step": 12170 }, { "epoch": 2.5, "learning_rate": 1.5006150061500618e-05, "loss": 0.306, "step": 12180 }, { "epoch": 2.5, "learning_rate": 1.5002050020500208e-05, "loss": 0.1265, "step": 12190 }, { "epoch": 2.5, "learning_rate": 1.4997949979499794e-05, "loss": 0.3091, "step": 12200 }, { "epoch": 2.5, "learning_rate": 1.4993849938499384e-05, "loss": 0.2731, "step": 12210 }, { "epoch": 2.51, "learning_rate": 1.4989749897498976e-05, "loss": 0.2758, "step": 12220 }, { "epoch": 2.51, "learning_rate": 1.4985649856498566e-05, "loss": 0.1954, "step": 12230 }, { "epoch": 2.51, "learning_rate": 1.4981549815498156e-05, "loss": 0.339, "step": 12240 }, { "epoch": 2.51, "learning_rate": 1.4977449774497746e-05, "loss": 0.288, "step": 12250 }, { "epoch": 2.51, "learning_rate": 1.4973349733497336e-05, "loss": 0.2654, "step": 12260 }, { "epoch": 2.52, "learning_rate": 1.4969249692496925e-05, "loss": 0.2902, "step": 12270 }, { "epoch": 2.52, "learning_rate": 1.4965149651496515e-05, "loss": 0.2766, "step": 12280 }, { "epoch": 2.52, "learning_rate": 1.4961049610496105e-05, "loss": 0.2892, "step": 12290 }, { "epoch": 2.52, "learning_rate": 1.4956949569495695e-05, "loss": 0.2177, "step": 12300 }, { "epoch": 2.52, "learning_rate": 1.4952849528495285e-05, "loss": 0.2084, "step": 12310 }, { "epoch": 2.53, "learning_rate": 1.4948749487494875e-05, "loss": 0.2797, "step": 12320 }, { "epoch": 2.53, "learning_rate": 1.4944649446494467e-05, "loss": 0.2399, "step": 12330 }, { "epoch": 2.53, "learning_rate": 1.4940549405494056e-05, "loss": 0.299, "step": 12340 }, { "epoch": 2.53, "learning_rate": 1.4936449364493646e-05, "loss": 0.2487, "step": 12350 }, { "epoch": 2.53, "learning_rate": 1.4932349323493236e-05, "loss": 0.3337, "step": 12360 }, { "epoch": 2.54, "learning_rate": 1.4928249282492826e-05, "loss": 0.2206, "step": 12370 }, { "epoch": 2.54, "learning_rate": 1.4924149241492416e-05, "loss": 0.2423, "step": 12380 }, { "epoch": 2.54, "learning_rate": 1.4920049200492006e-05, "loss": 0.3143, "step": 12390 }, { "epoch": 2.54, "learning_rate": 1.4915949159491596e-05, "loss": 0.2628, "step": 12400 }, { "epoch": 2.54, "learning_rate": 1.4911849118491186e-05, "loss": 0.1958, "step": 12410 }, { "epoch": 2.55, "learning_rate": 1.4907749077490776e-05, "loss": 0.2812, "step": 12420 }, { "epoch": 2.55, "learning_rate": 1.4903649036490365e-05, "loss": 0.2763, "step": 12430 }, { "epoch": 2.55, "learning_rate": 1.4899548995489957e-05, "loss": 0.1635, "step": 12440 }, { "epoch": 2.55, "learning_rate": 1.4895448954489547e-05, "loss": 0.2962, "step": 12450 }, { "epoch": 2.55, "learning_rate": 1.4891348913489137e-05, "loss": 0.2658, "step": 12460 }, { "epoch": 2.56, "learning_rate": 1.4887248872488727e-05, "loss": 0.1758, "step": 12470 }, { "epoch": 2.56, "learning_rate": 1.4883148831488317e-05, "loss": 0.2505, "step": 12480 }, { "epoch": 2.56, "learning_rate": 1.4879048790487907e-05, "loss": 0.2773, "step": 12490 }, { "epoch": 2.56, "learning_rate": 1.4874948749487496e-05, "loss": 0.2714, "step": 12500 }, { "epoch": 2.56, "learning_rate": 1.4870848708487086e-05, "loss": 0.2357, "step": 12510 }, { "epoch": 2.57, "learning_rate": 1.4866748667486676e-05, "loss": 0.2487, "step": 12520 }, { "epoch": 2.57, "learning_rate": 1.4862648626486266e-05, "loss": 0.3095, "step": 12530 }, { "epoch": 2.57, "learning_rate": 1.4858548585485856e-05, "loss": 0.2534, "step": 12540 }, { "epoch": 2.57, "learning_rate": 1.4854448544485448e-05, "loss": 0.1805, "step": 12550 }, { "epoch": 2.57, "learning_rate": 1.4850348503485038e-05, "loss": 0.245, "step": 12560 }, { "epoch": 2.58, "learning_rate": 1.4846248462484627e-05, "loss": 0.2859, "step": 12570 }, { "epoch": 2.58, "learning_rate": 1.4842148421484217e-05, "loss": 0.1856, "step": 12580 }, { "epoch": 2.58, "learning_rate": 1.4838048380483807e-05, "loss": 0.2082, "step": 12590 }, { "epoch": 2.58, "learning_rate": 1.4833948339483395e-05, "loss": 0.2198, "step": 12600 }, { "epoch": 2.59, "learning_rate": 1.4829848298482985e-05, "loss": 0.1851, "step": 12610 }, { "epoch": 2.59, "learning_rate": 1.4825748257482575e-05, "loss": 0.1555, "step": 12620 }, { "epoch": 2.59, "learning_rate": 1.4821648216482165e-05, "loss": 0.214, "step": 12630 }, { "epoch": 2.59, "learning_rate": 1.4817548175481755e-05, "loss": 0.2376, "step": 12640 }, { "epoch": 2.59, "learning_rate": 1.4813448134481345e-05, "loss": 0.2399, "step": 12650 }, { "epoch": 2.6, "learning_rate": 1.4809348093480935e-05, "loss": 0.2386, "step": 12660 }, { "epoch": 2.6, "learning_rate": 1.4805248052480525e-05, "loss": 0.1609, "step": 12670 }, { "epoch": 2.6, "learning_rate": 1.4801148011480115e-05, "loss": 0.2197, "step": 12680 }, { "epoch": 2.6, "learning_rate": 1.4797047970479706e-05, "loss": 0.3129, "step": 12690 }, { "epoch": 2.6, "learning_rate": 1.4792947929479296e-05, "loss": 0.4621, "step": 12700 }, { "epoch": 2.61, "learning_rate": 1.4788847888478886e-05, "loss": 0.2398, "step": 12710 }, { "epoch": 2.61, "learning_rate": 1.4784747847478476e-05, "loss": 0.2021, "step": 12720 }, { "epoch": 2.61, "learning_rate": 1.4780647806478066e-05, "loss": 0.2416, "step": 12730 }, { "epoch": 2.61, "learning_rate": 1.4776547765477656e-05, "loss": 0.2349, "step": 12740 }, { "epoch": 2.61, "learning_rate": 1.4772447724477246e-05, "loss": 0.2135, "step": 12750 }, { "epoch": 2.62, "learning_rate": 1.4768347683476835e-05, "loss": 0.2879, "step": 12760 }, { "epoch": 2.62, "learning_rate": 1.4764247642476425e-05, "loss": 0.2871, "step": 12770 }, { "epoch": 2.62, "learning_rate": 1.4760147601476015e-05, "loss": 0.3173, "step": 12780 }, { "epoch": 2.62, "learning_rate": 1.4756047560475605e-05, "loss": 0.3347, "step": 12790 }, { "epoch": 2.62, "learning_rate": 1.4751947519475197e-05, "loss": 0.2251, "step": 12800 }, { "epoch": 2.63, "learning_rate": 1.4747847478474787e-05, "loss": 0.2286, "step": 12810 }, { "epoch": 2.63, "learning_rate": 1.4743747437474377e-05, "loss": 0.241, "step": 12820 }, { "epoch": 2.63, "learning_rate": 1.4739647396473966e-05, "loss": 0.2357, "step": 12830 }, { "epoch": 2.63, "learning_rate": 1.4735547355473556e-05, "loss": 0.3214, "step": 12840 }, { "epoch": 2.63, "learning_rate": 1.4731447314473146e-05, "loss": 0.3069, "step": 12850 }, { "epoch": 2.64, "learning_rate": 1.4727347273472736e-05, "loss": 0.2408, "step": 12860 }, { "epoch": 2.64, "learning_rate": 1.4723247232472326e-05, "loss": 0.2392, "step": 12870 }, { "epoch": 2.64, "learning_rate": 1.4719147191471916e-05, "loss": 0.2179, "step": 12880 }, { "epoch": 2.64, "learning_rate": 1.4715047150471506e-05, "loss": 0.2294, "step": 12890 }, { "epoch": 2.64, "learning_rate": 1.4710947109471096e-05, "loss": 0.2818, "step": 12900 }, { "epoch": 2.65, "learning_rate": 1.4706847068470687e-05, "loss": 0.2433, "step": 12910 }, { "epoch": 2.65, "learning_rate": 1.4702747027470277e-05, "loss": 0.2592, "step": 12920 }, { "epoch": 2.65, "learning_rate": 1.4698646986469867e-05, "loss": 0.3014, "step": 12930 }, { "epoch": 2.65, "learning_rate": 1.4694546945469457e-05, "loss": 0.2467, "step": 12940 }, { "epoch": 2.65, "learning_rate": 1.4690446904469047e-05, "loss": 0.2936, "step": 12950 }, { "epoch": 2.66, "learning_rate": 1.4686346863468637e-05, "loss": 0.2236, "step": 12960 }, { "epoch": 2.66, "learning_rate": 1.4682246822468227e-05, "loss": 0.244, "step": 12970 }, { "epoch": 2.66, "learning_rate": 1.4678146781467817e-05, "loss": 0.2917, "step": 12980 }, { "epoch": 2.66, "learning_rate": 1.4674046740467407e-05, "loss": 0.283, "step": 12990 }, { "epoch": 2.67, "learning_rate": 1.4669946699466995e-05, "loss": 0.2247, "step": 13000 }, { "epoch": 2.67, "learning_rate": 1.4665846658466585e-05, "loss": 0.3347, "step": 13010 }, { "epoch": 2.67, "learning_rate": 1.4661746617466175e-05, "loss": 0.2258, "step": 13020 }, { "epoch": 2.67, "learning_rate": 1.4657646576465764e-05, "loss": 0.1915, "step": 13030 }, { "epoch": 2.67, "learning_rate": 1.4653546535465354e-05, "loss": 0.2468, "step": 13040 }, { "epoch": 2.68, "learning_rate": 1.4649446494464946e-05, "loss": 0.204, "step": 13050 }, { "epoch": 2.68, "learning_rate": 1.4645346453464536e-05, "loss": 0.2696, "step": 13060 }, { "epoch": 2.68, "learning_rate": 1.4641246412464126e-05, "loss": 0.313, "step": 13070 }, { "epoch": 2.68, "learning_rate": 1.4637146371463716e-05, "loss": 0.1766, "step": 13080 }, { "epoch": 2.68, "learning_rate": 1.4633046330463305e-05, "loss": 0.3411, "step": 13090 }, { "epoch": 2.69, "learning_rate": 1.4628946289462895e-05, "loss": 0.1764, "step": 13100 }, { "epoch": 2.69, "learning_rate": 1.4624846248462485e-05, "loss": 0.2924, "step": 13110 }, { "epoch": 2.69, "learning_rate": 1.4620746207462075e-05, "loss": 0.2323, "step": 13120 }, { "epoch": 2.69, "learning_rate": 1.4616646166461665e-05, "loss": 0.2937, "step": 13130 }, { "epoch": 2.69, "learning_rate": 1.4612546125461255e-05, "loss": 0.2406, "step": 13140 }, { "epoch": 2.7, "learning_rate": 1.4608446084460845e-05, "loss": 0.1941, "step": 13150 }, { "epoch": 2.7, "learning_rate": 1.4604346043460436e-05, "loss": 0.2796, "step": 13160 }, { "epoch": 2.7, "learning_rate": 1.4600246002460026e-05, "loss": 0.179, "step": 13170 }, { "epoch": 2.7, "learning_rate": 1.4596145961459616e-05, "loss": 0.2052, "step": 13180 }, { "epoch": 2.7, "learning_rate": 1.4592045920459206e-05, "loss": 0.2663, "step": 13190 }, { "epoch": 2.71, "learning_rate": 1.4587945879458796e-05, "loss": 0.1485, "step": 13200 }, { "epoch": 2.71, "learning_rate": 1.4583845838458386e-05, "loss": 0.2972, "step": 13210 }, { "epoch": 2.71, "learning_rate": 1.4579745797457976e-05, "loss": 0.2272, "step": 13220 }, { "epoch": 2.71, "learning_rate": 1.4575645756457566e-05, "loss": 0.1889, "step": 13230 }, { "epoch": 2.71, "learning_rate": 1.4571545715457156e-05, "loss": 0.2116, "step": 13240 }, { "epoch": 2.72, "learning_rate": 1.4567445674456746e-05, "loss": 0.1798, "step": 13250 }, { "epoch": 2.72, "learning_rate": 1.4563345633456335e-05, "loss": 0.2597, "step": 13260 }, { "epoch": 2.72, "learning_rate": 1.4559245592455927e-05, "loss": 0.207, "step": 13270 }, { "epoch": 2.72, "learning_rate": 1.4555145551455517e-05, "loss": 0.229, "step": 13280 }, { "epoch": 2.72, "learning_rate": 1.4551045510455107e-05, "loss": 0.2578, "step": 13290 }, { "epoch": 2.73, "learning_rate": 1.4546945469454697e-05, "loss": 0.2915, "step": 13300 }, { "epoch": 2.73, "learning_rate": 1.4542845428454287e-05, "loss": 0.2398, "step": 13310 }, { "epoch": 2.73, "learning_rate": 1.4538745387453876e-05, "loss": 0.3455, "step": 13320 }, { "epoch": 2.73, "learning_rate": 1.4534645346453466e-05, "loss": 0.2235, "step": 13330 }, { "epoch": 2.73, "learning_rate": 1.4530545305453056e-05, "loss": 0.3005, "step": 13340 }, { "epoch": 2.74, "learning_rate": 1.4526445264452646e-05, "loss": 0.286, "step": 13350 }, { "epoch": 2.74, "learning_rate": 1.4522345223452236e-05, "loss": 0.2008, "step": 13360 }, { "epoch": 2.74, "learning_rate": 1.4518245182451826e-05, "loss": 0.2752, "step": 13370 }, { "epoch": 2.74, "learning_rate": 1.4514145141451418e-05, "loss": 0.2251, "step": 13380 }, { "epoch": 2.74, "learning_rate": 1.4510045100451007e-05, "loss": 0.2353, "step": 13390 }, { "epoch": 2.75, "learning_rate": 1.4505945059450594e-05, "loss": 0.2107, "step": 13400 }, { "epoch": 2.75, "learning_rate": 1.4501845018450184e-05, "loss": 0.2722, "step": 13410 }, { "epoch": 2.75, "learning_rate": 1.4497744977449775e-05, "loss": 0.2484, "step": 13420 }, { "epoch": 2.75, "learning_rate": 1.4493644936449365e-05, "loss": 0.3519, "step": 13430 }, { "epoch": 2.76, "learning_rate": 1.4489544895448955e-05, "loss": 0.3241, "step": 13440 }, { "epoch": 2.76, "learning_rate": 1.4485444854448545e-05, "loss": 0.2198, "step": 13450 }, { "epoch": 2.76, "learning_rate": 1.4481344813448135e-05, "loss": 0.1991, "step": 13460 }, { "epoch": 2.76, "learning_rate": 1.4477244772447725e-05, "loss": 0.2887, "step": 13470 }, { "epoch": 2.76, "learning_rate": 1.4473144731447315e-05, "loss": 0.1369, "step": 13480 }, { "epoch": 2.77, "learning_rate": 1.4469044690446905e-05, "loss": 0.2444, "step": 13490 }, { "epoch": 2.77, "learning_rate": 1.4464944649446495e-05, "loss": 0.2519, "step": 13500 }, { "epoch": 2.77, "learning_rate": 1.4460844608446085e-05, "loss": 0.2206, "step": 13510 }, { "epoch": 2.77, "learning_rate": 1.4456744567445674e-05, "loss": 0.2699, "step": 13520 }, { "epoch": 2.77, "learning_rate": 1.4452644526445266e-05, "loss": 0.1902, "step": 13530 }, { "epoch": 2.78, "learning_rate": 1.4448544485444856e-05, "loss": 0.3103, "step": 13540 }, { "epoch": 2.78, "learning_rate": 1.4444444444444446e-05, "loss": 0.1573, "step": 13550 }, { "epoch": 2.78, "learning_rate": 1.4440344403444036e-05, "loss": 0.2185, "step": 13560 }, { "epoch": 2.78, "learning_rate": 1.4436244362443626e-05, "loss": 0.199, "step": 13570 }, { "epoch": 2.78, "learning_rate": 1.4432144321443216e-05, "loss": 0.2354, "step": 13580 }, { "epoch": 2.79, "learning_rate": 1.4428044280442805e-05, "loss": 0.2421, "step": 13590 }, { "epoch": 2.79, "learning_rate": 1.4423944239442395e-05, "loss": 0.2359, "step": 13600 }, { "epoch": 2.79, "learning_rate": 1.4419844198441985e-05, "loss": 0.238, "step": 13610 }, { "epoch": 2.79, "learning_rate": 1.4415744157441575e-05, "loss": 0.2709, "step": 13620 }, { "epoch": 2.79, "learning_rate": 1.4411644116441165e-05, "loss": 0.2248, "step": 13630 }, { "epoch": 2.8, "learning_rate": 1.4407544075440757e-05, "loss": 0.1662, "step": 13640 }, { "epoch": 2.8, "learning_rate": 1.4403444034440346e-05, "loss": 0.3331, "step": 13650 }, { "epoch": 2.8, "learning_rate": 1.4399343993439936e-05, "loss": 0.3492, "step": 13660 }, { "epoch": 2.8, "learning_rate": 1.4395243952439526e-05, "loss": 0.2345, "step": 13670 }, { "epoch": 2.8, "learning_rate": 1.4391143911439116e-05, "loss": 0.2134, "step": 13680 }, { "epoch": 2.81, "learning_rate": 1.4387043870438706e-05, "loss": 0.2394, "step": 13690 }, { "epoch": 2.81, "learning_rate": 1.4382943829438296e-05, "loss": 0.2333, "step": 13700 }, { "epoch": 2.81, "learning_rate": 1.4378843788437886e-05, "loss": 0.3961, "step": 13710 }, { "epoch": 2.81, "learning_rate": 1.4374743747437476e-05, "loss": 0.2372, "step": 13720 }, { "epoch": 2.81, "learning_rate": 1.4370643706437066e-05, "loss": 0.1901, "step": 13730 }, { "epoch": 2.82, "learning_rate": 1.4366543665436656e-05, "loss": 0.2647, "step": 13740 }, { "epoch": 2.82, "learning_rate": 1.4362443624436247e-05, "loss": 0.2979, "step": 13750 }, { "epoch": 2.82, "learning_rate": 1.4358343583435837e-05, "loss": 0.3136, "step": 13760 }, { "epoch": 2.82, "learning_rate": 1.4354243542435427e-05, "loss": 0.2567, "step": 13770 }, { "epoch": 2.82, "learning_rate": 1.4350143501435017e-05, "loss": 0.2645, "step": 13780 }, { "epoch": 2.83, "learning_rate": 1.4346043460434607e-05, "loss": 0.1999, "step": 13790 }, { "epoch": 2.83, "learning_rate": 1.4341943419434195e-05, "loss": 0.2296, "step": 13800 }, { "epoch": 2.83, "learning_rate": 1.4337843378433785e-05, "loss": 0.2616, "step": 13810 }, { "epoch": 2.83, "learning_rate": 1.4333743337433375e-05, "loss": 0.2499, "step": 13820 }, { "epoch": 2.84, "learning_rate": 1.4329643296432965e-05, "loss": 0.2761, "step": 13830 }, { "epoch": 2.84, "learning_rate": 1.4325543255432555e-05, "loss": 0.2712, "step": 13840 }, { "epoch": 2.84, "learning_rate": 1.4321443214432144e-05, "loss": 0.2663, "step": 13850 }, { "epoch": 2.84, "learning_rate": 1.4317343173431734e-05, "loss": 0.3846, "step": 13860 }, { "epoch": 2.84, "learning_rate": 1.4313243132431324e-05, "loss": 0.239, "step": 13870 }, { "epoch": 2.85, "learning_rate": 1.4309143091430914e-05, "loss": 0.3131, "step": 13880 }, { "epoch": 2.85, "learning_rate": 1.4305043050430506e-05, "loss": 0.1867, "step": 13890 }, { "epoch": 2.85, "learning_rate": 1.4300943009430096e-05, "loss": 0.1866, "step": 13900 }, { "epoch": 2.85, "learning_rate": 1.4296842968429685e-05, "loss": 0.2681, "step": 13910 }, { "epoch": 2.85, "learning_rate": 1.4292742927429275e-05, "loss": 0.2433, "step": 13920 }, { "epoch": 2.86, "learning_rate": 1.4288642886428865e-05, "loss": 0.3126, "step": 13930 }, { "epoch": 2.86, "learning_rate": 1.4284542845428455e-05, "loss": 0.3139, "step": 13940 }, { "epoch": 2.86, "learning_rate": 1.4280442804428045e-05, "loss": 0.2418, "step": 13950 }, { "epoch": 2.86, "learning_rate": 1.4276342763427635e-05, "loss": 0.4344, "step": 13960 }, { "epoch": 2.86, "learning_rate": 1.4272242722427225e-05, "loss": 0.2607, "step": 13970 }, { "epoch": 2.87, "learning_rate": 1.4268142681426815e-05, "loss": 0.1852, "step": 13980 }, { "epoch": 2.87, "learning_rate": 1.4264042640426405e-05, "loss": 0.2672, "step": 13990 }, { "epoch": 2.87, "learning_rate": 1.4259942599425996e-05, "loss": 0.1766, "step": 14000 }, { "epoch": 2.87, "learning_rate": 1.4255842558425586e-05, "loss": 0.2636, "step": 14010 }, { "epoch": 2.87, "learning_rate": 1.4251742517425176e-05, "loss": 0.363, "step": 14020 }, { "epoch": 2.88, "learning_rate": 1.4247642476424766e-05, "loss": 0.2693, "step": 14030 }, { "epoch": 2.88, "learning_rate": 1.4243542435424356e-05, "loss": 0.2885, "step": 14040 }, { "epoch": 2.88, "learning_rate": 1.4239442394423946e-05, "loss": 0.2903, "step": 14050 }, { "epoch": 2.88, "learning_rate": 1.4235342353423536e-05, "loss": 0.2083, "step": 14060 }, { "epoch": 2.88, "learning_rate": 1.4231242312423126e-05, "loss": 0.2296, "step": 14070 }, { "epoch": 2.89, "learning_rate": 1.4227142271422715e-05, "loss": 0.3051, "step": 14080 }, { "epoch": 2.89, "learning_rate": 1.4223042230422305e-05, "loss": 0.2493, "step": 14090 }, { "epoch": 2.89, "learning_rate": 1.4218942189421895e-05, "loss": 0.2061, "step": 14100 }, { "epoch": 2.89, "learning_rate": 1.4214842148421487e-05, "loss": 0.2323, "step": 14110 }, { "epoch": 2.89, "learning_rate": 1.4210742107421077e-05, "loss": 0.3514, "step": 14120 }, { "epoch": 2.9, "learning_rate": 1.4206642066420667e-05, "loss": 0.2198, "step": 14130 }, { "epoch": 2.9, "learning_rate": 1.4202542025420257e-05, "loss": 0.2339, "step": 14140 }, { "epoch": 2.9, "learning_rate": 1.4198441984419846e-05, "loss": 0.2709, "step": 14150 }, { "epoch": 2.9, "learning_rate": 1.4194341943419436e-05, "loss": 0.2239, "step": 14160 }, { "epoch": 2.9, "learning_rate": 1.4190241902419026e-05, "loss": 0.2722, "step": 14170 }, { "epoch": 2.91, "learning_rate": 1.4186141861418616e-05, "loss": 0.3064, "step": 14180 }, { "epoch": 2.91, "learning_rate": 1.4182041820418206e-05, "loss": 0.2481, "step": 14190 }, { "epoch": 2.91, "learning_rate": 1.4177941779417794e-05, "loss": 0.2025, "step": 14200 }, { "epoch": 2.91, "learning_rate": 1.4173841738417384e-05, "loss": 0.3067, "step": 14210 }, { "epoch": 2.92, "learning_rate": 1.4169741697416974e-05, "loss": 0.1655, "step": 14220 }, { "epoch": 2.92, "learning_rate": 1.4165641656416564e-05, "loss": 0.2721, "step": 14230 }, { "epoch": 2.92, "learning_rate": 1.4161541615416154e-05, "loss": 0.2612, "step": 14240 }, { "epoch": 2.92, "learning_rate": 1.4157441574415744e-05, "loss": 0.218, "step": 14250 }, { "epoch": 2.92, "learning_rate": 1.4153341533415335e-05, "loss": 0.2344, "step": 14260 }, { "epoch": 2.93, "learning_rate": 1.4149241492414925e-05, "loss": 0.2268, "step": 14270 }, { "epoch": 2.93, "learning_rate": 1.4145141451414515e-05, "loss": 0.2741, "step": 14280 }, { "epoch": 2.93, "learning_rate": 1.4141041410414105e-05, "loss": 0.1863, "step": 14290 }, { "epoch": 2.93, "learning_rate": 1.4136941369413695e-05, "loss": 0.2043, "step": 14300 }, { "epoch": 2.93, "learning_rate": 1.4132841328413285e-05, "loss": 0.2441, "step": 14310 }, { "epoch": 2.94, "learning_rate": 1.4128741287412875e-05, "loss": 0.3462, "step": 14320 }, { "epoch": 2.94, "learning_rate": 1.4124641246412465e-05, "loss": 0.2571, "step": 14330 }, { "epoch": 2.94, "learning_rate": 1.4120541205412054e-05, "loss": 0.218, "step": 14340 }, { "epoch": 2.94, "learning_rate": 1.4116441164411644e-05, "loss": 0.224, "step": 14350 }, { "epoch": 2.94, "learning_rate": 1.4112341123411234e-05, "loss": 0.239, "step": 14360 }, { "epoch": 2.95, "learning_rate": 1.4108241082410826e-05, "loss": 0.2527, "step": 14370 }, { "epoch": 2.95, "learning_rate": 1.4104141041410416e-05, "loss": 0.203, "step": 14380 }, { "epoch": 2.95, "learning_rate": 1.4100041000410006e-05, "loss": 0.3635, "step": 14390 }, { "epoch": 2.95, "learning_rate": 1.4095940959409596e-05, "loss": 0.3163, "step": 14400 }, { "epoch": 2.95, "learning_rate": 1.4091840918409185e-05, "loss": 0.1913, "step": 14410 }, { "epoch": 2.96, "learning_rate": 1.4087740877408775e-05, "loss": 0.1123, "step": 14420 }, { "epoch": 2.96, "learning_rate": 1.4083640836408365e-05, "loss": 0.1971, "step": 14430 }, { "epoch": 2.96, "learning_rate": 1.4079540795407955e-05, "loss": 0.2543, "step": 14440 }, { "epoch": 2.96, "learning_rate": 1.4075440754407545e-05, "loss": 0.1916, "step": 14450 }, { "epoch": 2.96, "learning_rate": 1.4071340713407135e-05, "loss": 0.2376, "step": 14460 }, { "epoch": 2.97, "learning_rate": 1.4067240672406725e-05, "loss": 0.2129, "step": 14470 }, { "epoch": 2.97, "learning_rate": 1.4063140631406316e-05, "loss": 0.2098, "step": 14480 }, { "epoch": 2.97, "learning_rate": 1.4059040590405906e-05, "loss": 0.1912, "step": 14490 }, { "epoch": 2.97, "learning_rate": 1.4054940549405496e-05, "loss": 0.2144, "step": 14500 }, { "epoch": 2.97, "learning_rate": 1.4050840508405086e-05, "loss": 0.2544, "step": 14510 }, { "epoch": 2.98, "learning_rate": 1.4046740467404676e-05, "loss": 0.2113, "step": 14520 }, { "epoch": 2.98, "learning_rate": 1.4042640426404266e-05, "loss": 0.2256, "step": 14530 }, { "epoch": 2.98, "learning_rate": 1.4038540385403856e-05, "loss": 0.2777, "step": 14540 }, { "epoch": 2.98, "learning_rate": 1.4034440344403446e-05, "loss": 0.3704, "step": 14550 }, { "epoch": 2.98, "learning_rate": 1.4030340303403036e-05, "loss": 0.1825, "step": 14560 }, { "epoch": 2.99, "learning_rate": 1.4026240262402625e-05, "loss": 0.3311, "step": 14570 }, { "epoch": 2.99, "learning_rate": 1.4022140221402215e-05, "loss": 0.2192, "step": 14580 }, { "epoch": 2.99, "learning_rate": 1.4018040180401807e-05, "loss": 0.2621, "step": 14590 }, { "epoch": 2.99, "learning_rate": 1.4013940139401393e-05, "loss": 0.229, "step": 14600 }, { "epoch": 3.0, "learning_rate": 1.4009840098400983e-05, "loss": 0.244, "step": 14610 }, { "epoch": 3.0, "learning_rate": 1.4005740057400575e-05, "loss": 0.2566, "step": 14620 }, { "epoch": 3.0, "learning_rate": 1.4001640016400165e-05, "loss": 0.2285, "step": 14630 }, { "epoch": 3.0, "eval_FN": 374, "eval_FP": 2527, "eval_TN": 13232, "eval_TP": 4532, "eval_accuracy": 0.8596177111057344, "eval_f1": 0.7575428332636859, "eval_loss": 0.3749154508113861, "eval_precision": 0.6420172829012608, "eval_recall": 0.9237668161434978, "eval_runtime": 141.9185, "eval_samples_per_second": 145.612, "eval_steps_per_second": 9.104, "step": 14634 }, { "epoch": 3.0, "learning_rate": 1.3997539975399755e-05, "loss": 0.2153, "step": 14640 }, { "epoch": 3.0, "learning_rate": 1.3993439934399345e-05, "loss": 0.1955, "step": 14650 }, { "epoch": 3.01, "learning_rate": 1.3989339893398935e-05, "loss": 0.1659, "step": 14660 }, { "epoch": 3.01, "learning_rate": 1.3985239852398524e-05, "loss": 0.1856, "step": 14670 }, { "epoch": 3.01, "learning_rate": 1.3981139811398114e-05, "loss": 0.0993, "step": 14680 }, { "epoch": 3.01, "learning_rate": 1.3977039770397704e-05, "loss": 0.2422, "step": 14690 }, { "epoch": 3.01, "learning_rate": 1.3972939729397294e-05, "loss": 0.1988, "step": 14700 }, { "epoch": 3.02, "learning_rate": 1.3968839688396884e-05, "loss": 0.2326, "step": 14710 }, { "epoch": 3.02, "learning_rate": 1.3964739647396474e-05, "loss": 0.1282, "step": 14720 }, { "epoch": 3.02, "learning_rate": 1.3960639606396066e-05, "loss": 0.1991, "step": 14730 }, { "epoch": 3.02, "learning_rate": 1.3956539565395655e-05, "loss": 0.1342, "step": 14740 }, { "epoch": 3.02, "learning_rate": 1.3952439524395245e-05, "loss": 0.1364, "step": 14750 }, { "epoch": 3.03, "learning_rate": 1.3948339483394835e-05, "loss": 0.1597, "step": 14760 }, { "epoch": 3.03, "learning_rate": 1.3944239442394425e-05, "loss": 0.1931, "step": 14770 }, { "epoch": 3.03, "learning_rate": 1.3940139401394015e-05, "loss": 0.2508, "step": 14780 }, { "epoch": 3.03, "learning_rate": 1.3936039360393605e-05, "loss": 0.1138, "step": 14790 }, { "epoch": 3.03, "learning_rate": 1.3931939319393195e-05, "loss": 0.1133, "step": 14800 }, { "epoch": 3.04, "learning_rate": 1.3927839278392785e-05, "loss": 0.1661, "step": 14810 }, { "epoch": 3.04, "learning_rate": 1.3923739237392375e-05, "loss": 0.3196, "step": 14820 }, { "epoch": 3.04, "learning_rate": 1.3919639196391964e-05, "loss": 0.2274, "step": 14830 }, { "epoch": 3.04, "learning_rate": 1.3915539155391556e-05, "loss": 0.2236, "step": 14840 }, { "epoch": 3.04, "learning_rate": 1.3911439114391146e-05, "loss": 0.238, "step": 14850 }, { "epoch": 3.05, "learning_rate": 1.3907339073390736e-05, "loss": 0.1846, "step": 14860 }, { "epoch": 3.05, "learning_rate": 1.3903239032390326e-05, "loss": 0.2052, "step": 14870 }, { "epoch": 3.05, "learning_rate": 1.3899138991389916e-05, "loss": 0.2224, "step": 14880 }, { "epoch": 3.05, "learning_rate": 1.3895038950389506e-05, "loss": 0.1808, "step": 14890 }, { "epoch": 3.05, "learning_rate": 1.3890938909389095e-05, "loss": 0.1866, "step": 14900 }, { "epoch": 3.06, "learning_rate": 1.3886838868388685e-05, "loss": 0.078, "step": 14910 }, { "epoch": 3.06, "learning_rate": 1.3882738827388275e-05, "loss": 0.162, "step": 14920 }, { "epoch": 3.06, "learning_rate": 1.3878638786387865e-05, "loss": 0.1429, "step": 14930 }, { "epoch": 3.06, "learning_rate": 1.3874538745387455e-05, "loss": 0.184, "step": 14940 }, { "epoch": 3.06, "learning_rate": 1.3870438704387047e-05, "loss": 0.2498, "step": 14950 }, { "epoch": 3.07, "learning_rate": 1.3866338663386637e-05, "loss": 0.2195, "step": 14960 }, { "epoch": 3.07, "learning_rate": 1.3862238622386226e-05, "loss": 0.2308, "step": 14970 }, { "epoch": 3.07, "learning_rate": 1.3858138581385816e-05, "loss": 0.2247, "step": 14980 }, { "epoch": 3.07, "learning_rate": 1.3854038540385406e-05, "loss": 0.2171, "step": 14990 }, { "epoch": 3.08, "learning_rate": 1.3849938499384994e-05, "loss": 0.2481, "step": 15000 }, { "epoch": 3.08, "learning_rate": 1.3845838458384584e-05, "loss": 0.1489, "step": 15010 }, { "epoch": 3.08, "learning_rate": 1.3841738417384174e-05, "loss": 0.2111, "step": 15020 }, { "epoch": 3.08, "learning_rate": 1.3837638376383764e-05, "loss": 0.1919, "step": 15030 }, { "epoch": 3.08, "learning_rate": 1.3833538335383354e-05, "loss": 0.2022, "step": 15040 }, { "epoch": 3.09, "learning_rate": 1.3829438294382944e-05, "loss": 0.2465, "step": 15050 }, { "epoch": 3.09, "learning_rate": 1.3825338253382534e-05, "loss": 0.2294, "step": 15060 }, { "epoch": 3.09, "learning_rate": 1.3821238212382124e-05, "loss": 0.2202, "step": 15070 }, { "epoch": 3.09, "learning_rate": 1.3817138171381714e-05, "loss": 0.2652, "step": 15080 }, { "epoch": 3.09, "learning_rate": 1.3813038130381305e-05, "loss": 0.2449, "step": 15090 }, { "epoch": 3.1, "learning_rate": 1.3808938089380895e-05, "loss": 0.0803, "step": 15100 }, { "epoch": 3.1, "learning_rate": 1.3804838048380485e-05, "loss": 0.2488, "step": 15110 }, { "epoch": 3.1, "learning_rate": 1.3800738007380075e-05, "loss": 0.1559, "step": 15120 }, { "epoch": 3.1, "learning_rate": 1.3796637966379665e-05, "loss": 0.2432, "step": 15130 }, { "epoch": 3.1, "learning_rate": 1.3792537925379255e-05, "loss": 0.1592, "step": 15140 }, { "epoch": 3.11, "learning_rate": 1.3788437884378845e-05, "loss": 0.1474, "step": 15150 }, { "epoch": 3.11, "learning_rate": 1.3784337843378434e-05, "loss": 0.2052, "step": 15160 }, { "epoch": 3.11, "learning_rate": 1.3780237802378024e-05, "loss": 0.2261, "step": 15170 }, { "epoch": 3.11, "learning_rate": 1.3776137761377614e-05, "loss": 0.253, "step": 15180 }, { "epoch": 3.11, "learning_rate": 1.3772037720377204e-05, "loss": 0.2773, "step": 15190 }, { "epoch": 3.12, "learning_rate": 1.3767937679376796e-05, "loss": 0.1592, "step": 15200 }, { "epoch": 3.12, "learning_rate": 1.3763837638376386e-05, "loss": 0.2043, "step": 15210 }, { "epoch": 3.12, "learning_rate": 1.3759737597375976e-05, "loss": 0.1608, "step": 15220 }, { "epoch": 3.12, "learning_rate": 1.3755637556375565e-05, "loss": 0.3201, "step": 15230 }, { "epoch": 3.12, "learning_rate": 1.3751537515375155e-05, "loss": 0.2018, "step": 15240 }, { "epoch": 3.13, "learning_rate": 1.3747437474374745e-05, "loss": 0.2028, "step": 15250 }, { "epoch": 3.13, "learning_rate": 1.3743337433374335e-05, "loss": 0.1647, "step": 15260 }, { "epoch": 3.13, "learning_rate": 1.3739237392373925e-05, "loss": 0.2474, "step": 15270 }, { "epoch": 3.13, "learning_rate": 1.3735137351373515e-05, "loss": 0.169, "step": 15280 }, { "epoch": 3.13, "learning_rate": 1.3731037310373105e-05, "loss": 0.1682, "step": 15290 }, { "epoch": 3.14, "learning_rate": 1.3726937269372695e-05, "loss": 0.1709, "step": 15300 }, { "epoch": 3.14, "learning_rate": 1.3722837228372286e-05, "loss": 0.349, "step": 15310 }, { "epoch": 3.14, "learning_rate": 1.3718737187371876e-05, "loss": 0.2747, "step": 15320 }, { "epoch": 3.14, "learning_rate": 1.3714637146371466e-05, "loss": 0.2056, "step": 15330 }, { "epoch": 3.14, "learning_rate": 1.3710537105371056e-05, "loss": 0.1954, "step": 15340 }, { "epoch": 3.15, "learning_rate": 1.3706437064370646e-05, "loss": 0.2222, "step": 15350 }, { "epoch": 3.15, "learning_rate": 1.3702337023370236e-05, "loss": 0.2288, "step": 15360 }, { "epoch": 3.15, "learning_rate": 1.3698236982369826e-05, "loss": 0.1882, "step": 15370 }, { "epoch": 3.15, "learning_rate": 1.3694136941369416e-05, "loss": 0.165, "step": 15380 }, { "epoch": 3.15, "learning_rate": 1.3690036900369005e-05, "loss": 0.206, "step": 15390 }, { "epoch": 3.16, "learning_rate": 1.3685936859368594e-05, "loss": 0.2199, "step": 15400 }, { "epoch": 3.16, "learning_rate": 1.3681836818368184e-05, "loss": 0.1266, "step": 15410 }, { "epoch": 3.16, "learning_rate": 1.3677736777367773e-05, "loss": 0.219, "step": 15420 }, { "epoch": 3.16, "learning_rate": 1.3673636736367363e-05, "loss": 0.2371, "step": 15430 }, { "epoch": 3.17, "learning_rate": 1.3669536695366953e-05, "loss": 0.28, "step": 15440 }, { "epoch": 3.17, "learning_rate": 1.3665436654366543e-05, "loss": 0.1519, "step": 15450 }, { "epoch": 3.17, "learning_rate": 1.3661336613366135e-05, "loss": 0.2207, "step": 15460 }, { "epoch": 3.17, "learning_rate": 1.3657236572365725e-05, "loss": 0.2068, "step": 15470 }, { "epoch": 3.17, "learning_rate": 1.3653136531365315e-05, "loss": 0.1454, "step": 15480 }, { "epoch": 3.18, "learning_rate": 1.3649036490364904e-05, "loss": 0.2221, "step": 15490 }, { "epoch": 3.18, "learning_rate": 1.3644936449364494e-05, "loss": 0.1538, "step": 15500 }, { "epoch": 3.18, "learning_rate": 1.3640836408364084e-05, "loss": 0.1629, "step": 15510 }, { "epoch": 3.18, "learning_rate": 1.3636736367363674e-05, "loss": 0.1653, "step": 15520 }, { "epoch": 3.18, "learning_rate": 1.3632636326363264e-05, "loss": 0.1481, "step": 15530 }, { "epoch": 3.19, "learning_rate": 1.3628536285362854e-05, "loss": 0.178, "step": 15540 }, { "epoch": 3.19, "learning_rate": 1.3624436244362444e-05, "loss": 0.2081, "step": 15550 }, { "epoch": 3.19, "learning_rate": 1.3620336203362034e-05, "loss": 0.1836, "step": 15560 }, { "epoch": 3.19, "learning_rate": 1.3616236162361625e-05, "loss": 0.2609, "step": 15570 }, { "epoch": 3.19, "learning_rate": 1.3612136121361215e-05, "loss": 0.1638, "step": 15580 }, { "epoch": 3.2, "learning_rate": 1.3608036080360805e-05, "loss": 0.0903, "step": 15590 }, { "epoch": 3.2, "learning_rate": 1.3603936039360395e-05, "loss": 0.2893, "step": 15600 }, { "epoch": 3.2, "learning_rate": 1.3599835998359985e-05, "loss": 0.3104, "step": 15610 }, { "epoch": 3.2, "learning_rate": 1.3595735957359575e-05, "loss": 0.2895, "step": 15620 }, { "epoch": 3.2, "learning_rate": 1.3591635916359165e-05, "loss": 0.205, "step": 15630 }, { "epoch": 3.21, "learning_rate": 1.3587535875358755e-05, "loss": 0.2055, "step": 15640 }, { "epoch": 3.21, "learning_rate": 1.3583435834358344e-05, "loss": 0.2227, "step": 15650 }, { "epoch": 3.21, "learning_rate": 1.3579335793357934e-05, "loss": 0.1584, "step": 15660 }, { "epoch": 3.21, "learning_rate": 1.3575235752357524e-05, "loss": 0.2667, "step": 15670 }, { "epoch": 3.21, "learning_rate": 1.3571135711357116e-05, "loss": 0.1052, "step": 15680 }, { "epoch": 3.22, "learning_rate": 1.3567035670356706e-05, "loss": 0.2447, "step": 15690 }, { "epoch": 3.22, "learning_rate": 1.3562935629356296e-05, "loss": 0.1515, "step": 15700 }, { "epoch": 3.22, "learning_rate": 1.3558835588355886e-05, "loss": 0.3312, "step": 15710 }, { "epoch": 3.22, "learning_rate": 1.3554735547355475e-05, "loss": 0.1673, "step": 15720 }, { "epoch": 3.22, "learning_rate": 1.3550635506355065e-05, "loss": 0.1298, "step": 15730 }, { "epoch": 3.23, "learning_rate": 1.3546535465354655e-05, "loss": 0.1485, "step": 15740 }, { "epoch": 3.23, "learning_rate": 1.3542435424354245e-05, "loss": 0.1872, "step": 15750 }, { "epoch": 3.23, "learning_rate": 1.3538335383353835e-05, "loss": 0.3186, "step": 15760 }, { "epoch": 3.23, "learning_rate": 1.3534235342353425e-05, "loss": 0.1909, "step": 15770 }, { "epoch": 3.23, "learning_rate": 1.3530135301353015e-05, "loss": 0.2007, "step": 15780 }, { "epoch": 3.24, "learning_rate": 1.3526035260352606e-05, "loss": 0.1824, "step": 15790 }, { "epoch": 3.24, "learning_rate": 1.3521935219352193e-05, "loss": 0.1678, "step": 15800 }, { "epoch": 3.24, "learning_rate": 1.3517835178351783e-05, "loss": 0.2561, "step": 15810 }, { "epoch": 3.24, "learning_rate": 1.3513735137351374e-05, "loss": 0.2103, "step": 15820 }, { "epoch": 3.25, "learning_rate": 1.3509635096350964e-05, "loss": 0.0986, "step": 15830 }, { "epoch": 3.25, "learning_rate": 1.3505535055350554e-05, "loss": 0.1285, "step": 15840 }, { "epoch": 3.25, "learning_rate": 1.3501435014350144e-05, "loss": 0.2606, "step": 15850 }, { "epoch": 3.25, "learning_rate": 1.3497334973349734e-05, "loss": 0.3352, "step": 15860 }, { "epoch": 3.25, "learning_rate": 1.3493234932349324e-05, "loss": 0.2573, "step": 15870 }, { "epoch": 3.26, "learning_rate": 1.3489134891348914e-05, "loss": 0.1945, "step": 15880 }, { "epoch": 3.26, "learning_rate": 1.3485034850348504e-05, "loss": 0.1507, "step": 15890 }, { "epoch": 3.26, "learning_rate": 1.3480934809348094e-05, "loss": 0.1861, "step": 15900 }, { "epoch": 3.26, "learning_rate": 1.3476834768347684e-05, "loss": 0.3014, "step": 15910 }, { "epoch": 3.26, "learning_rate": 1.3472734727347273e-05, "loss": 0.214, "step": 15920 }, { "epoch": 3.27, "learning_rate": 1.3468634686346865e-05, "loss": 0.1452, "step": 15930 }, { "epoch": 3.27, "learning_rate": 1.3464534645346455e-05, "loss": 0.2187, "step": 15940 }, { "epoch": 3.27, "learning_rate": 1.3460434604346045e-05, "loss": 0.2853, "step": 15950 }, { "epoch": 3.27, "learning_rate": 1.3456334563345635e-05, "loss": 0.1913, "step": 15960 }, { "epoch": 3.27, "learning_rate": 1.3452234522345225e-05, "loss": 0.2199, "step": 15970 }, { "epoch": 3.28, "learning_rate": 1.3448134481344814e-05, "loss": 0.1893, "step": 15980 }, { "epoch": 3.28, "learning_rate": 1.3444034440344404e-05, "loss": 0.1489, "step": 15990 }, { "epoch": 3.28, "learning_rate": 1.3439934399343994e-05, "loss": 0.1788, "step": 16000 }, { "epoch": 3.28, "learning_rate": 1.3435834358343584e-05, "loss": 0.2196, "step": 16010 }, { "epoch": 3.28, "learning_rate": 1.3431734317343174e-05, "loss": 0.1734, "step": 16020 }, { "epoch": 3.29, "learning_rate": 1.3427634276342764e-05, "loss": 0.213, "step": 16030 }, { "epoch": 3.29, "learning_rate": 1.3423534235342356e-05, "loss": 0.201, "step": 16040 }, { "epoch": 3.29, "learning_rate": 1.3419434194341945e-05, "loss": 0.2691, "step": 16050 }, { "epoch": 3.29, "learning_rate": 1.3415334153341535e-05, "loss": 0.202, "step": 16060 }, { "epoch": 3.29, "learning_rate": 1.3411234112341125e-05, "loss": 0.2162, "step": 16070 }, { "epoch": 3.3, "learning_rate": 1.3407134071340715e-05, "loss": 0.2455, "step": 16080 }, { "epoch": 3.3, "learning_rate": 1.3403034030340305e-05, "loss": 0.3127, "step": 16090 }, { "epoch": 3.3, "learning_rate": 1.3398933989339895e-05, "loss": 0.1871, "step": 16100 }, { "epoch": 3.3, "learning_rate": 1.3394833948339485e-05, "loss": 0.1408, "step": 16110 }, { "epoch": 3.3, "learning_rate": 1.3390733907339075e-05, "loss": 0.1671, "step": 16120 }, { "epoch": 3.31, "learning_rate": 1.3386633866338665e-05, "loss": 0.1589, "step": 16130 }, { "epoch": 3.31, "learning_rate": 1.3382533825338255e-05, "loss": 0.1987, "step": 16140 }, { "epoch": 3.31, "learning_rate": 1.3378433784337846e-05, "loss": 0.3567, "step": 16150 }, { "epoch": 3.31, "learning_rate": 1.3374333743337436e-05, "loss": 0.3121, "step": 16160 }, { "epoch": 3.31, "learning_rate": 1.3370233702337026e-05, "loss": 0.1233, "step": 16170 }, { "epoch": 3.32, "learning_rate": 1.3366133661336616e-05, "loss": 0.1716, "step": 16180 }, { "epoch": 3.32, "learning_rate": 1.3362033620336206e-05, "loss": 0.1619, "step": 16190 }, { "epoch": 3.32, "learning_rate": 1.3357933579335794e-05, "loss": 0.2205, "step": 16200 }, { "epoch": 3.32, "learning_rate": 1.3353833538335384e-05, "loss": 0.131, "step": 16210 }, { "epoch": 3.33, "learning_rate": 1.3349733497334974e-05, "loss": 0.2465, "step": 16220 }, { "epoch": 3.33, "learning_rate": 1.3345633456334564e-05, "loss": 0.2513, "step": 16230 }, { "epoch": 3.33, "learning_rate": 1.3341533415334154e-05, "loss": 0.1678, "step": 16240 }, { "epoch": 3.33, "learning_rate": 1.3337433374333743e-05, "loss": 0.2419, "step": 16250 }, { "epoch": 3.33, "learning_rate": 1.3333333333333333e-05, "loss": 0.2665, "step": 16260 }, { "epoch": 3.34, "learning_rate": 1.3329233292332923e-05, "loss": 0.1344, "step": 16270 }, { "epoch": 3.34, "learning_rate": 1.3325133251332513e-05, "loss": 0.2393, "step": 16280 }, { "epoch": 3.34, "learning_rate": 1.3321033210332105e-05, "loss": 0.1295, "step": 16290 }, { "epoch": 3.34, "learning_rate": 1.3316933169331695e-05, "loss": 0.1852, "step": 16300 }, { "epoch": 3.34, "learning_rate": 1.3312833128331284e-05, "loss": 0.2283, "step": 16310 }, { "epoch": 3.35, "learning_rate": 1.3308733087330874e-05, "loss": 0.2871, "step": 16320 }, { "epoch": 3.35, "learning_rate": 1.3304633046330464e-05, "loss": 0.2195, "step": 16330 }, { "epoch": 3.35, "learning_rate": 1.3300533005330054e-05, "loss": 0.1678, "step": 16340 }, { "epoch": 3.35, "learning_rate": 1.3296432964329644e-05, "loss": 0.1797, "step": 16350 }, { "epoch": 3.35, "learning_rate": 1.3292332923329234e-05, "loss": 0.2435, "step": 16360 }, { "epoch": 3.36, "learning_rate": 1.3288232882328824e-05, "loss": 0.2179, "step": 16370 }, { "epoch": 3.36, "learning_rate": 1.3284132841328414e-05, "loss": 0.2317, "step": 16380 }, { "epoch": 3.36, "learning_rate": 1.3280032800328004e-05, "loss": 0.1974, "step": 16390 }, { "epoch": 3.36, "learning_rate": 1.3275932759327595e-05, "loss": 0.1561, "step": 16400 }, { "epoch": 3.36, "learning_rate": 1.3271832718327185e-05, "loss": 0.2463, "step": 16410 }, { "epoch": 3.37, "learning_rate": 1.3267732677326775e-05, "loss": 0.2064, "step": 16420 }, { "epoch": 3.37, "learning_rate": 1.3263632636326365e-05, "loss": 0.2635, "step": 16430 }, { "epoch": 3.37, "learning_rate": 1.3259532595325955e-05, "loss": 0.1574, "step": 16440 }, { "epoch": 3.37, "learning_rate": 1.3255432554325545e-05, "loss": 0.2732, "step": 16450 }, { "epoch": 3.37, "learning_rate": 1.3251332513325135e-05, "loss": 0.1334, "step": 16460 }, { "epoch": 3.38, "learning_rate": 1.3247232472324725e-05, "loss": 0.2526, "step": 16470 }, { "epoch": 3.38, "learning_rate": 1.3243132431324314e-05, "loss": 0.2176, "step": 16480 }, { "epoch": 3.38, "learning_rate": 1.3239032390323904e-05, "loss": 0.1758, "step": 16490 }, { "epoch": 3.38, "learning_rate": 1.3234932349323494e-05, "loss": 0.238, "step": 16500 }, { "epoch": 3.38, "learning_rate": 1.3230832308323086e-05, "loss": 0.227, "step": 16510 }, { "epoch": 3.39, "learning_rate": 1.3226732267322676e-05, "loss": 0.225, "step": 16520 }, { "epoch": 3.39, "learning_rate": 1.3222632226322266e-05, "loss": 0.1466, "step": 16530 }, { "epoch": 3.39, "learning_rate": 1.3218532185321855e-05, "loss": 0.2727, "step": 16540 }, { "epoch": 3.39, "learning_rate": 1.3214432144321445e-05, "loss": 0.2362, "step": 16550 }, { "epoch": 3.39, "learning_rate": 1.3210332103321035e-05, "loss": 0.2248, "step": 16560 }, { "epoch": 3.4, "learning_rate": 1.3206232062320625e-05, "loss": 0.1908, "step": 16570 }, { "epoch": 3.4, "learning_rate": 1.3202132021320215e-05, "loss": 0.2507, "step": 16580 }, { "epoch": 3.4, "learning_rate": 1.3198031980319805e-05, "loss": 0.2225, "step": 16590 }, { "epoch": 3.4, "learning_rate": 1.3193931939319393e-05, "loss": 0.1629, "step": 16600 }, { "epoch": 3.41, "learning_rate": 1.3189831898318983e-05, "loss": 0.1655, "step": 16610 }, { "epoch": 3.41, "learning_rate": 1.3185731857318573e-05, "loss": 0.3558, "step": 16620 }, { "epoch": 3.41, "learning_rate": 1.3181631816318163e-05, "loss": 0.2549, "step": 16630 }, { "epoch": 3.41, "learning_rate": 1.3177531775317753e-05, "loss": 0.1512, "step": 16640 }, { "epoch": 3.41, "learning_rate": 1.3173431734317343e-05, "loss": 0.1421, "step": 16650 }, { "epoch": 3.42, "learning_rate": 1.3169331693316934e-05, "loss": 0.3825, "step": 16660 }, { "epoch": 3.42, "learning_rate": 1.3165231652316524e-05, "loss": 0.2246, "step": 16670 }, { "epoch": 3.42, "learning_rate": 1.3161131611316114e-05, "loss": 0.1937, "step": 16680 }, { "epoch": 3.42, "learning_rate": 1.3157031570315704e-05, "loss": 0.2306, "step": 16690 }, { "epoch": 3.42, "learning_rate": 1.3152931529315294e-05, "loss": 0.2048, "step": 16700 }, { "epoch": 3.43, "learning_rate": 1.3148831488314884e-05, "loss": 0.1839, "step": 16710 }, { "epoch": 3.43, "learning_rate": 1.3144731447314474e-05, "loss": 0.2613, "step": 16720 }, { "epoch": 3.43, "learning_rate": 1.3140631406314064e-05, "loss": 0.1434, "step": 16730 }, { "epoch": 3.43, "learning_rate": 1.3136531365313653e-05, "loss": 0.1817, "step": 16740 }, { "epoch": 3.43, "learning_rate": 1.3132431324313243e-05, "loss": 0.2976, "step": 16750 }, { "epoch": 3.44, "learning_rate": 1.3128331283312833e-05, "loss": 0.1898, "step": 16760 }, { "epoch": 3.44, "learning_rate": 1.3124231242312425e-05, "loss": 0.3718, "step": 16770 }, { "epoch": 3.44, "learning_rate": 1.3120131201312015e-05, "loss": 0.2394, "step": 16780 }, { "epoch": 3.44, "learning_rate": 1.3116031160311605e-05, "loss": 0.1393, "step": 16790 }, { "epoch": 3.44, "learning_rate": 1.3111931119311195e-05, "loss": 0.2073, "step": 16800 }, { "epoch": 3.45, "learning_rate": 1.3107831078310784e-05, "loss": 0.206, "step": 16810 }, { "epoch": 3.45, "learning_rate": 1.3103731037310374e-05, "loss": 0.1594, "step": 16820 }, { "epoch": 3.45, "learning_rate": 1.3099630996309964e-05, "loss": 0.2032, "step": 16830 }, { "epoch": 3.45, "learning_rate": 1.3095530955309554e-05, "loss": 0.2068, "step": 16840 }, { "epoch": 3.45, "learning_rate": 1.3091430914309144e-05, "loss": 0.1079, "step": 16850 }, { "epoch": 3.46, "learning_rate": 1.3087330873308734e-05, "loss": 0.2738, "step": 16860 }, { "epoch": 3.46, "learning_rate": 1.3083230832308324e-05, "loss": 0.2796, "step": 16870 }, { "epoch": 3.46, "learning_rate": 1.3079130791307915e-05, "loss": 0.1562, "step": 16880 }, { "epoch": 3.46, "learning_rate": 1.3075030750307505e-05, "loss": 0.2479, "step": 16890 }, { "epoch": 3.46, "learning_rate": 1.3070930709307095e-05, "loss": 0.1376, "step": 16900 }, { "epoch": 3.47, "learning_rate": 1.3066830668306685e-05, "loss": 0.1724, "step": 16910 }, { "epoch": 3.47, "learning_rate": 1.3062730627306275e-05, "loss": 0.1277, "step": 16920 }, { "epoch": 3.47, "learning_rate": 1.3058630586305865e-05, "loss": 0.2764, "step": 16930 }, { "epoch": 3.47, "learning_rate": 1.3054530545305455e-05, "loss": 0.1736, "step": 16940 }, { "epoch": 3.47, "learning_rate": 1.3050430504305045e-05, "loss": 0.231, "step": 16950 }, { "epoch": 3.48, "learning_rate": 1.3046330463304635e-05, "loss": 0.3482, "step": 16960 }, { "epoch": 3.48, "learning_rate": 1.3042230422304224e-05, "loss": 0.2101, "step": 16970 }, { "epoch": 3.48, "learning_rate": 1.3038130381303814e-05, "loss": 0.1538, "step": 16980 }, { "epoch": 3.48, "learning_rate": 1.3034030340303406e-05, "loss": 0.3042, "step": 16990 }, { "epoch": 3.49, "learning_rate": 1.3029930299302992e-05, "loss": 0.1579, "step": 17000 }, { "epoch": 3.49, "learning_rate": 1.3025830258302582e-05, "loss": 0.1359, "step": 17010 }, { "epoch": 3.49, "learning_rate": 1.3021730217302174e-05, "loss": 0.2506, "step": 17020 }, { "epoch": 3.49, "learning_rate": 1.3017630176301764e-05, "loss": 0.1924, "step": 17030 }, { "epoch": 3.49, "learning_rate": 1.3013530135301354e-05, "loss": 0.2246, "step": 17040 }, { "epoch": 3.5, "learning_rate": 1.3009430094300944e-05, "loss": 0.1519, "step": 17050 }, { "epoch": 3.5, "learning_rate": 1.3005330053300534e-05, "loss": 0.1584, "step": 17060 }, { "epoch": 3.5, "learning_rate": 1.3001230012300123e-05, "loss": 0.2155, "step": 17070 }, { "epoch": 3.5, "learning_rate": 1.2997129971299713e-05, "loss": 0.2321, "step": 17080 }, { "epoch": 3.5, "learning_rate": 1.2993029930299303e-05, "loss": 0.2769, "step": 17090 }, { "epoch": 3.51, "learning_rate": 1.2988929889298893e-05, "loss": 0.2199, "step": 17100 }, { "epoch": 3.51, "learning_rate": 1.2984829848298483e-05, "loss": 0.3218, "step": 17110 }, { "epoch": 3.51, "learning_rate": 1.2980729807298073e-05, "loss": 0.2748, "step": 17120 }, { "epoch": 3.51, "learning_rate": 1.2976629766297664e-05, "loss": 0.2647, "step": 17130 }, { "epoch": 3.51, "learning_rate": 1.2972529725297254e-05, "loss": 0.2272, "step": 17140 }, { "epoch": 3.52, "learning_rate": 1.2968429684296844e-05, "loss": 0.1944, "step": 17150 }, { "epoch": 3.52, "learning_rate": 1.2964329643296434e-05, "loss": 0.1748, "step": 17160 }, { "epoch": 3.52, "learning_rate": 1.2960229602296024e-05, "loss": 0.1613, "step": 17170 }, { "epoch": 3.52, "learning_rate": 1.2956129561295614e-05, "loss": 0.1372, "step": 17180 }, { "epoch": 3.52, "learning_rate": 1.2952029520295204e-05, "loss": 0.2104, "step": 17190 }, { "epoch": 3.53, "learning_rate": 1.2947929479294794e-05, "loss": 0.3052, "step": 17200 }, { "epoch": 3.53, "learning_rate": 1.2943829438294384e-05, "loss": 0.303, "step": 17210 }, { "epoch": 3.53, "learning_rate": 1.2939729397293974e-05, "loss": 0.2498, "step": 17220 }, { "epoch": 3.53, "learning_rate": 1.2935629356293563e-05, "loss": 0.2008, "step": 17230 }, { "epoch": 3.53, "learning_rate": 1.2931529315293155e-05, "loss": 0.2021, "step": 17240 }, { "epoch": 3.54, "learning_rate": 1.2927429274292745e-05, "loss": 0.2257, "step": 17250 }, { "epoch": 3.54, "learning_rate": 1.2923329233292335e-05, "loss": 0.1523, "step": 17260 }, { "epoch": 3.54, "learning_rate": 1.2919229192291925e-05, "loss": 0.1701, "step": 17270 }, { "epoch": 3.54, "learning_rate": 1.2915129151291515e-05, "loss": 0.2965, "step": 17280 }, { "epoch": 3.54, "learning_rate": 1.2911029110291105e-05, "loss": 0.212, "step": 17290 }, { "epoch": 3.55, "learning_rate": 1.2906929069290694e-05, "loss": 0.1471, "step": 17300 }, { "epoch": 3.55, "learning_rate": 1.2902829028290284e-05, "loss": 0.1974, "step": 17310 }, { "epoch": 3.55, "learning_rate": 1.2898728987289874e-05, "loss": 0.2132, "step": 17320 }, { "epoch": 3.55, "learning_rate": 1.2894628946289464e-05, "loss": 0.184, "step": 17330 }, { "epoch": 3.55, "learning_rate": 1.2890528905289054e-05, "loss": 0.1307, "step": 17340 }, { "epoch": 3.56, "learning_rate": 1.2886428864288646e-05, "loss": 0.1978, "step": 17350 }, { "epoch": 3.56, "learning_rate": 1.2882328823288236e-05, "loss": 0.2283, "step": 17360 }, { "epoch": 3.56, "learning_rate": 1.2878228782287825e-05, "loss": 0.2062, "step": 17370 }, { "epoch": 3.56, "learning_rate": 1.2874128741287415e-05, "loss": 0.3099, "step": 17380 }, { "epoch": 3.56, "learning_rate": 1.2870028700287005e-05, "loss": 0.2107, "step": 17390 }, { "epoch": 3.57, "learning_rate": 1.2865928659286593e-05, "loss": 0.1999, "step": 17400 }, { "epoch": 3.57, "learning_rate": 1.2861828618286183e-05, "loss": 0.1624, "step": 17410 }, { "epoch": 3.57, "learning_rate": 1.2857728577285773e-05, "loss": 0.2423, "step": 17420 }, { "epoch": 3.57, "learning_rate": 1.2853628536285363e-05, "loss": 0.2142, "step": 17430 }, { "epoch": 3.58, "learning_rate": 1.2849528495284953e-05, "loss": 0.1842, "step": 17440 }, { "epoch": 3.58, "learning_rate": 1.2845428454284543e-05, "loss": 0.3072, "step": 17450 }, { "epoch": 3.58, "learning_rate": 1.2841328413284133e-05, "loss": 0.2234, "step": 17460 }, { "epoch": 3.58, "learning_rate": 1.2837228372283723e-05, "loss": 0.2337, "step": 17470 }, { "epoch": 3.58, "learning_rate": 1.2833128331283313e-05, "loss": 0.1502, "step": 17480 }, { "epoch": 3.59, "learning_rate": 1.2829028290282902e-05, "loss": 0.3544, "step": 17490 }, { "epoch": 3.59, "learning_rate": 1.2824928249282494e-05, "loss": 0.1137, "step": 17500 }, { "epoch": 3.59, "learning_rate": 1.2820828208282084e-05, "loss": 0.2452, "step": 17510 }, { "epoch": 3.59, "learning_rate": 1.2816728167281674e-05, "loss": 0.1383, "step": 17520 }, { "epoch": 3.59, "learning_rate": 1.2812628126281264e-05, "loss": 0.2029, "step": 17530 }, { "epoch": 3.6, "learning_rate": 1.2808528085280854e-05, "loss": 0.1714, "step": 17540 }, { "epoch": 3.6, "learning_rate": 1.2804428044280444e-05, "loss": 0.2308, "step": 17550 }, { "epoch": 3.6, "learning_rate": 1.2800328003280033e-05, "loss": 0.2267, "step": 17560 }, { "epoch": 3.6, "learning_rate": 1.2796227962279623e-05, "loss": 0.1443, "step": 17570 }, { "epoch": 3.6, "learning_rate": 1.2792127921279213e-05, "loss": 0.2416, "step": 17580 }, { "epoch": 3.61, "learning_rate": 1.2788027880278803e-05, "loss": 0.1805, "step": 17590 }, { "epoch": 3.61, "learning_rate": 1.2783927839278393e-05, "loss": 0.1884, "step": 17600 }, { "epoch": 3.61, "learning_rate": 1.2779827798277985e-05, "loss": 0.2292, "step": 17610 }, { "epoch": 3.61, "learning_rate": 1.2775727757277575e-05, "loss": 0.1655, "step": 17620 }, { "epoch": 3.61, "learning_rate": 1.2771627716277164e-05, "loss": 0.211, "step": 17630 }, { "epoch": 3.62, "learning_rate": 1.2767527675276754e-05, "loss": 0.2479, "step": 17640 }, { "epoch": 3.62, "learning_rate": 1.2763427634276344e-05, "loss": 0.2709, "step": 17650 }, { "epoch": 3.62, "learning_rate": 1.2759327593275934e-05, "loss": 0.1912, "step": 17660 }, { "epoch": 3.62, "learning_rate": 1.2755227552275524e-05, "loss": 0.189, "step": 17670 }, { "epoch": 3.62, "learning_rate": 1.2751127511275114e-05, "loss": 0.2363, "step": 17680 }, { "epoch": 3.63, "learning_rate": 1.2747027470274704e-05, "loss": 0.1995, "step": 17690 }, { "epoch": 3.63, "learning_rate": 1.2742927429274294e-05, "loss": 0.1682, "step": 17700 }, { "epoch": 3.63, "learning_rate": 1.2738827388273884e-05, "loss": 0.1792, "step": 17710 }, { "epoch": 3.63, "learning_rate": 1.2734727347273475e-05, "loss": 0.2512, "step": 17720 }, { "epoch": 3.63, "learning_rate": 1.2730627306273065e-05, "loss": 0.245, "step": 17730 }, { "epoch": 3.64, "learning_rate": 1.2726527265272655e-05, "loss": 0.1909, "step": 17740 }, { "epoch": 3.64, "learning_rate": 1.2722427224272245e-05, "loss": 0.183, "step": 17750 }, { "epoch": 3.64, "learning_rate": 1.2718327183271835e-05, "loss": 0.2979, "step": 17760 }, { "epoch": 3.64, "learning_rate": 1.2714227142271425e-05, "loss": 0.2248, "step": 17770 }, { "epoch": 3.64, "learning_rate": 1.2710127101271015e-05, "loss": 0.1703, "step": 17780 }, { "epoch": 3.65, "learning_rate": 1.2706027060270604e-05, "loss": 0.1378, "step": 17790 }, { "epoch": 3.65, "learning_rate": 1.2701927019270193e-05, "loss": 0.1453, "step": 17800 }, { "epoch": 3.65, "learning_rate": 1.2697826978269783e-05, "loss": 0.2136, "step": 17810 }, { "epoch": 3.65, "learning_rate": 1.2693726937269372e-05, "loss": 0.1901, "step": 17820 }, { "epoch": 3.66, "learning_rate": 1.2689626896268962e-05, "loss": 0.1447, "step": 17830 }, { "epoch": 3.66, "learning_rate": 1.2685526855268552e-05, "loss": 0.3334, "step": 17840 }, { "epoch": 3.66, "learning_rate": 1.2681426814268142e-05, "loss": 0.22, "step": 17850 }, { "epoch": 3.66, "learning_rate": 1.2677326773267734e-05, "loss": 0.2339, "step": 17860 }, { "epoch": 3.66, "learning_rate": 1.2673226732267324e-05, "loss": 0.2729, "step": 17870 }, { "epoch": 3.67, "learning_rate": 1.2669126691266914e-05, "loss": 0.2618, "step": 17880 }, { "epoch": 3.67, "learning_rate": 1.2665026650266503e-05, "loss": 0.2575, "step": 17890 }, { "epoch": 3.67, "learning_rate": 1.2660926609266093e-05, "loss": 0.2168, "step": 17900 }, { "epoch": 3.67, "learning_rate": 1.2656826568265683e-05, "loss": 0.2632, "step": 17910 }, { "epoch": 3.67, "learning_rate": 1.2652726527265273e-05, "loss": 0.2265, "step": 17920 }, { "epoch": 3.68, "learning_rate": 1.2648626486264863e-05, "loss": 0.2419, "step": 17930 }, { "epoch": 3.68, "learning_rate": 1.2644526445264453e-05, "loss": 0.1901, "step": 17940 }, { "epoch": 3.68, "learning_rate": 1.2640426404264043e-05, "loss": 0.2182, "step": 17950 }, { "epoch": 3.68, "learning_rate": 1.2636326363263633e-05, "loss": 0.1844, "step": 17960 }, { "epoch": 3.68, "learning_rate": 1.2632226322263224e-05, "loss": 0.1911, "step": 17970 }, { "epoch": 3.69, "learning_rate": 1.2628126281262814e-05, "loss": 0.2367, "step": 17980 }, { "epoch": 3.69, "learning_rate": 1.2624026240262404e-05, "loss": 0.1727, "step": 17990 }, { "epoch": 3.69, "learning_rate": 1.2619926199261994e-05, "loss": 0.2902, "step": 18000 }, { "epoch": 3.69, "learning_rate": 1.2615826158261584e-05, "loss": 0.1271, "step": 18010 }, { "epoch": 3.69, "learning_rate": 1.2611726117261174e-05, "loss": 0.2098, "step": 18020 }, { "epoch": 3.7, "learning_rate": 1.2607626076260764e-05, "loss": 0.1108, "step": 18030 }, { "epoch": 3.7, "learning_rate": 1.2603526035260354e-05, "loss": 0.1568, "step": 18040 }, { "epoch": 3.7, "learning_rate": 1.2599425994259943e-05, "loss": 0.2111, "step": 18050 }, { "epoch": 3.7, "learning_rate": 1.2595325953259533e-05, "loss": 0.1514, "step": 18060 }, { "epoch": 3.7, "learning_rate": 1.2591225912259123e-05, "loss": 0.153, "step": 18070 }, { "epoch": 3.71, "learning_rate": 1.2587125871258715e-05, "loss": 0.1709, "step": 18080 }, { "epoch": 3.71, "learning_rate": 1.2583025830258305e-05, "loss": 0.1666, "step": 18090 }, { "epoch": 3.71, "learning_rate": 1.2578925789257895e-05, "loss": 0.1817, "step": 18100 }, { "epoch": 3.71, "learning_rate": 1.2574825748257485e-05, "loss": 0.2181, "step": 18110 }, { "epoch": 3.71, "learning_rate": 1.2570725707257074e-05, "loss": 0.3226, "step": 18120 }, { "epoch": 3.72, "learning_rate": 1.2566625666256664e-05, "loss": 0.3006, "step": 18130 }, { "epoch": 3.72, "learning_rate": 1.2562525625256254e-05, "loss": 0.2288, "step": 18140 }, { "epoch": 3.72, "learning_rate": 1.2558425584255844e-05, "loss": 0.1923, "step": 18150 }, { "epoch": 3.72, "learning_rate": 1.2554325543255434e-05, "loss": 0.2906, "step": 18160 }, { "epoch": 3.72, "learning_rate": 1.2550225502255024e-05, "loss": 0.2959, "step": 18170 }, { "epoch": 3.73, "learning_rate": 1.2546125461254614e-05, "loss": 0.1248, "step": 18180 }, { "epoch": 3.73, "learning_rate": 1.2542025420254205e-05, "loss": 0.2097, "step": 18190 }, { "epoch": 3.73, "learning_rate": 1.2537925379253792e-05, "loss": 0.1881, "step": 18200 }, { "epoch": 3.73, "learning_rate": 1.2533825338253382e-05, "loss": 0.2489, "step": 18210 }, { "epoch": 3.74, "learning_rate": 1.2529725297252973e-05, "loss": 0.1771, "step": 18220 }, { "epoch": 3.74, "learning_rate": 1.2525625256252563e-05, "loss": 0.1738, "step": 18230 }, { "epoch": 3.74, "learning_rate": 1.2521525215252153e-05, "loss": 0.2423, "step": 18240 }, { "epoch": 3.74, "learning_rate": 1.2517425174251743e-05, "loss": 0.1495, "step": 18250 }, { "epoch": 3.74, "learning_rate": 1.2513325133251333e-05, "loss": 0.149, "step": 18260 }, { "epoch": 3.75, "learning_rate": 1.2509225092250923e-05, "loss": 0.2459, "step": 18270 }, { "epoch": 3.75, "learning_rate": 1.2505125051250513e-05, "loss": 0.1603, "step": 18280 }, { "epoch": 3.75, "learning_rate": 1.2501025010250103e-05, "loss": 0.2612, "step": 18290 }, { "epoch": 3.75, "learning_rate": 1.2496924969249693e-05, "loss": 0.1681, "step": 18300 }, { "epoch": 3.75, "learning_rate": 1.2492824928249282e-05, "loss": 0.2686, "step": 18310 }, { "epoch": 3.76, "learning_rate": 1.2488724887248872e-05, "loss": 0.0866, "step": 18320 }, { "epoch": 3.76, "learning_rate": 1.2484624846248464e-05, "loss": 0.2563, "step": 18330 }, { "epoch": 3.76, "learning_rate": 1.2480524805248054e-05, "loss": 0.1519, "step": 18340 }, { "epoch": 3.76, "learning_rate": 1.2476424764247644e-05, "loss": 0.178, "step": 18350 }, { "epoch": 3.76, "learning_rate": 1.2472324723247234e-05, "loss": 0.1608, "step": 18360 }, { "epoch": 3.77, "learning_rate": 1.2468224682246824e-05, "loss": 0.2954, "step": 18370 }, { "epoch": 3.77, "learning_rate": 1.2464124641246413e-05, "loss": 0.1859, "step": 18380 }, { "epoch": 3.77, "learning_rate": 1.2460024600246003e-05, "loss": 0.2852, "step": 18390 }, { "epoch": 3.77, "learning_rate": 1.2455924559245593e-05, "loss": 0.2755, "step": 18400 }, { "epoch": 3.77, "learning_rate": 1.2451824518245183e-05, "loss": 0.2755, "step": 18410 }, { "epoch": 3.78, "learning_rate": 1.2447724477244773e-05, "loss": 0.1686, "step": 18420 }, { "epoch": 3.78, "learning_rate": 1.2443624436244363e-05, "loss": 0.2709, "step": 18430 }, { "epoch": 3.78, "learning_rate": 1.2439524395243955e-05, "loss": 0.167, "step": 18440 }, { "epoch": 3.78, "learning_rate": 1.2435424354243544e-05, "loss": 0.1123, "step": 18450 }, { "epoch": 3.78, "learning_rate": 1.2431324313243134e-05, "loss": 0.3171, "step": 18460 }, { "epoch": 3.79, "learning_rate": 1.2427224272242724e-05, "loss": 0.2523, "step": 18470 }, { "epoch": 3.79, "learning_rate": 1.2423124231242314e-05, "loss": 0.2179, "step": 18480 }, { "epoch": 3.79, "learning_rate": 1.2419024190241904e-05, "loss": 0.1872, "step": 18490 }, { "epoch": 3.79, "learning_rate": 1.2414924149241494e-05, "loss": 0.2506, "step": 18500 }, { "epoch": 3.79, "learning_rate": 1.2410824108241084e-05, "loss": 0.2424, "step": 18510 }, { "epoch": 3.8, "learning_rate": 1.2406724067240674e-05, "loss": 0.1871, "step": 18520 }, { "epoch": 3.8, "learning_rate": 1.2402624026240264e-05, "loss": 0.1887, "step": 18530 }, { "epoch": 3.8, "learning_rate": 1.2398523985239854e-05, "loss": 0.2808, "step": 18540 }, { "epoch": 3.8, "learning_rate": 1.2394423944239445e-05, "loss": 0.1826, "step": 18550 }, { "epoch": 3.8, "learning_rate": 1.2390323903239035e-05, "loss": 0.1531, "step": 18560 }, { "epoch": 3.81, "learning_rate": 1.2386223862238625e-05, "loss": 0.2008, "step": 18570 }, { "epoch": 3.81, "learning_rate": 1.2382123821238215e-05, "loss": 0.2098, "step": 18580 }, { "epoch": 3.81, "learning_rate": 1.2378023780237805e-05, "loss": 0.2397, "step": 18590 }, { "epoch": 3.81, "learning_rate": 1.2373923739237393e-05, "loss": 0.2315, "step": 18600 }, { "epoch": 3.82, "learning_rate": 1.2369823698236983e-05, "loss": 0.2327, "step": 18610 }, { "epoch": 3.82, "learning_rate": 1.2365723657236573e-05, "loss": 0.1574, "step": 18620 }, { "epoch": 3.82, "learning_rate": 1.2361623616236163e-05, "loss": 0.1699, "step": 18630 }, { "epoch": 3.82, "learning_rate": 1.2357523575235752e-05, "loss": 0.2242, "step": 18640 }, { "epoch": 3.82, "learning_rate": 1.2353423534235342e-05, "loss": 0.1989, "step": 18650 }, { "epoch": 3.83, "learning_rate": 1.2349323493234932e-05, "loss": 0.1988, "step": 18660 }, { "epoch": 3.83, "learning_rate": 1.2345223452234522e-05, "loss": 0.2733, "step": 18670 }, { "epoch": 3.83, "learning_rate": 1.2341123411234112e-05, "loss": 0.1097, "step": 18680 }, { "epoch": 3.83, "learning_rate": 1.2337023370233702e-05, "loss": 0.277, "step": 18690 }, { "epoch": 3.83, "learning_rate": 1.2332923329233294e-05, "loss": 0.204, "step": 18700 }, { "epoch": 3.84, "learning_rate": 1.2328823288232883e-05, "loss": 0.2171, "step": 18710 }, { "epoch": 3.84, "learning_rate": 1.2324723247232473e-05, "loss": 0.2856, "step": 18720 }, { "epoch": 3.84, "learning_rate": 1.2320623206232063e-05, "loss": 0.3374, "step": 18730 }, { "epoch": 3.84, "learning_rate": 1.2316523165231653e-05, "loss": 0.2076, "step": 18740 }, { "epoch": 3.84, "learning_rate": 1.2312423124231243e-05, "loss": 0.2281, "step": 18750 }, { "epoch": 3.85, "learning_rate": 1.2308323083230833e-05, "loss": 0.1502, "step": 18760 }, { "epoch": 3.85, "learning_rate": 1.2304223042230423e-05, "loss": 0.1901, "step": 18770 }, { "epoch": 3.85, "learning_rate": 1.2300123001230013e-05, "loss": 0.1913, "step": 18780 }, { "epoch": 3.85, "learning_rate": 1.2296022960229603e-05, "loss": 0.2013, "step": 18790 }, { "epoch": 3.85, "learning_rate": 1.2291922919229193e-05, "loss": 0.2289, "step": 18800 }, { "epoch": 3.86, "learning_rate": 1.2287822878228784e-05, "loss": 0.164, "step": 18810 }, { "epoch": 3.86, "learning_rate": 1.2283722837228374e-05, "loss": 0.1676, "step": 18820 }, { "epoch": 3.86, "learning_rate": 1.2279622796227964e-05, "loss": 0.3149, "step": 18830 }, { "epoch": 3.86, "learning_rate": 1.2275522755227554e-05, "loss": 0.2224, "step": 18840 }, { "epoch": 3.86, "learning_rate": 1.2271422714227144e-05, "loss": 0.1316, "step": 18850 }, { "epoch": 3.87, "learning_rate": 1.2267322673226734e-05, "loss": 0.2389, "step": 18860 }, { "epoch": 3.87, "learning_rate": 1.2263222632226323e-05, "loss": 0.2827, "step": 18870 }, { "epoch": 3.87, "learning_rate": 1.2259122591225913e-05, "loss": 0.1975, "step": 18880 }, { "epoch": 3.87, "learning_rate": 1.2255022550225503e-05, "loss": 0.2095, "step": 18890 }, { "epoch": 3.87, "learning_rate": 1.2250922509225093e-05, "loss": 0.2329, "step": 18900 }, { "epoch": 3.88, "learning_rate": 1.2246822468224683e-05, "loss": 0.1609, "step": 18910 }, { "epoch": 3.88, "learning_rate": 1.2242722427224275e-05, "loss": 0.119, "step": 18920 }, { "epoch": 3.88, "learning_rate": 1.2238622386223865e-05, "loss": 0.2598, "step": 18930 }, { "epoch": 3.88, "learning_rate": 1.2234522345223454e-05, "loss": 0.2501, "step": 18940 }, { "epoch": 3.88, "learning_rate": 1.2230422304223044e-05, "loss": 0.1686, "step": 18950 }, { "epoch": 3.89, "learning_rate": 1.2226322263222634e-05, "loss": 0.1529, "step": 18960 }, { "epoch": 3.89, "learning_rate": 1.2222222222222224e-05, "loss": 0.1965, "step": 18970 }, { "epoch": 3.89, "learning_rate": 1.2218122181221814e-05, "loss": 0.1138, "step": 18980 }, { "epoch": 3.89, "learning_rate": 1.2214022140221404e-05, "loss": 0.2964, "step": 18990 }, { "epoch": 3.9, "learning_rate": 1.2209922099220992e-05, "loss": 0.2028, "step": 19000 }, { "epoch": 3.9, "learning_rate": 1.2205822058220582e-05, "loss": 0.2218, "step": 19010 }, { "epoch": 3.9, "learning_rate": 1.2201722017220172e-05, "loss": 0.2105, "step": 19020 }, { "epoch": 3.9, "learning_rate": 1.2197621976219762e-05, "loss": 0.1752, "step": 19030 }, { "epoch": 3.9, "learning_rate": 1.2193521935219352e-05, "loss": 0.195, "step": 19040 }, { "epoch": 3.91, "learning_rate": 1.2189421894218942e-05, "loss": 0.2423, "step": 19050 }, { "epoch": 3.91, "learning_rate": 1.2185321853218533e-05, "loss": 0.1085, "step": 19060 }, { "epoch": 3.91, "learning_rate": 1.2181221812218123e-05, "loss": 0.1895, "step": 19070 }, { "epoch": 3.91, "learning_rate": 1.2177121771217713e-05, "loss": 0.1776, "step": 19080 }, { "epoch": 3.91, "learning_rate": 1.2173021730217303e-05, "loss": 0.2983, "step": 19090 }, { "epoch": 3.92, "learning_rate": 1.2168921689216893e-05, "loss": 0.1767, "step": 19100 }, { "epoch": 3.92, "learning_rate": 1.2164821648216483e-05, "loss": 0.1512, "step": 19110 }, { "epoch": 3.92, "learning_rate": 1.2160721607216073e-05, "loss": 0.1276, "step": 19120 }, { "epoch": 3.92, "learning_rate": 1.2156621566215663e-05, "loss": 0.2002, "step": 19130 }, { "epoch": 3.92, "learning_rate": 1.2152521525215252e-05, "loss": 0.3322, "step": 19140 }, { "epoch": 3.93, "learning_rate": 1.2148421484214842e-05, "loss": 0.3129, "step": 19150 }, { "epoch": 3.93, "learning_rate": 1.2144321443214432e-05, "loss": 0.2359, "step": 19160 }, { "epoch": 3.93, "learning_rate": 1.2140221402214024e-05, "loss": 0.2598, "step": 19170 }, { "epoch": 3.93, "learning_rate": 1.2136121361213614e-05, "loss": 0.1981, "step": 19180 }, { "epoch": 3.93, "learning_rate": 1.2132021320213204e-05, "loss": 0.2695, "step": 19190 }, { "epoch": 3.94, "learning_rate": 1.2127921279212793e-05, "loss": 0.2796, "step": 19200 }, { "epoch": 3.94, "learning_rate": 1.2123821238212383e-05, "loss": 0.1695, "step": 19210 }, { "epoch": 3.94, "learning_rate": 1.2119721197211973e-05, "loss": 0.1914, "step": 19220 }, { "epoch": 3.94, "learning_rate": 1.2115621156211563e-05, "loss": 0.2023, "step": 19230 }, { "epoch": 3.94, "learning_rate": 1.2111521115211153e-05, "loss": 0.2512, "step": 19240 }, { "epoch": 3.95, "learning_rate": 1.2107421074210743e-05, "loss": 0.2124, "step": 19250 }, { "epoch": 3.95, "learning_rate": 1.2103321033210333e-05, "loss": 0.114, "step": 19260 }, { "epoch": 3.95, "learning_rate": 1.2099220992209923e-05, "loss": 0.196, "step": 19270 }, { "epoch": 3.95, "learning_rate": 1.2095120951209514e-05, "loss": 0.2562, "step": 19280 }, { "epoch": 3.95, "learning_rate": 1.2091020910209104e-05, "loss": 0.1943, "step": 19290 }, { "epoch": 3.96, "learning_rate": 1.2086920869208694e-05, "loss": 0.2126, "step": 19300 }, { "epoch": 3.96, "learning_rate": 1.2082820828208284e-05, "loss": 0.1365, "step": 19310 }, { "epoch": 3.96, "learning_rate": 1.2078720787207874e-05, "loss": 0.2421, "step": 19320 }, { "epoch": 3.96, "learning_rate": 1.2074620746207464e-05, "loss": 0.2463, "step": 19330 }, { "epoch": 3.96, "learning_rate": 1.2070520705207054e-05, "loss": 0.1509, "step": 19340 }, { "epoch": 3.97, "learning_rate": 1.2066420664206644e-05, "loss": 0.1498, "step": 19350 }, { "epoch": 3.97, "learning_rate": 1.2062320623206234e-05, "loss": 0.2418, "step": 19360 }, { "epoch": 3.97, "learning_rate": 1.2058220582205823e-05, "loss": 0.2824, "step": 19370 }, { "epoch": 3.97, "learning_rate": 1.2054120541205413e-05, "loss": 0.2832, "step": 19380 }, { "epoch": 3.97, "learning_rate": 1.2050020500205005e-05, "loss": 0.2365, "step": 19390 }, { "epoch": 3.98, "learning_rate": 1.2045920459204591e-05, "loss": 0.3041, "step": 19400 }, { "epoch": 3.98, "learning_rate": 1.2041820418204181e-05, "loss": 0.1685, "step": 19410 }, { "epoch": 3.98, "learning_rate": 1.2037720377203773e-05, "loss": 0.162, "step": 19420 }, { "epoch": 3.98, "learning_rate": 1.2033620336203363e-05, "loss": 0.1477, "step": 19430 }, { "epoch": 3.99, "learning_rate": 1.2029520295202953e-05, "loss": 0.2249, "step": 19440 }, { "epoch": 3.99, "learning_rate": 1.2025420254202543e-05, "loss": 0.193, "step": 19450 }, { "epoch": 3.99, "learning_rate": 1.2021320213202133e-05, "loss": 0.1905, "step": 19460 }, { "epoch": 3.99, "learning_rate": 1.2017220172201722e-05, "loss": 0.1851, "step": 19470 }, { "epoch": 3.99, "learning_rate": 1.2013120131201312e-05, "loss": 0.2525, "step": 19480 }, { "epoch": 4.0, "learning_rate": 1.2009020090200902e-05, "loss": 0.3105, "step": 19490 }, { "epoch": 4.0, "learning_rate": 1.2004920049200492e-05, "loss": 0.1298, "step": 19500 }, { "epoch": 4.0, "learning_rate": 1.2000820008200082e-05, "loss": 0.2857, "step": 19510 }, { "epoch": 4.0, "eval_FN": 479, "eval_FP": 2088, "eval_TN": 13671, "eval_TP": 4427, "eval_accuracy": 0.8757803048632954, "eval_f1": 0.7752385955695648, "eval_loss": 0.33523041009902954, "eval_precision": 0.6795088257866462, "eval_recall": 0.9023644516918059, "eval_runtime": 141.852, "eval_samples_per_second": 145.68, "eval_steps_per_second": 9.108, "step": 19512 }, { "epoch": 4.0, "learning_rate": 1.1996719967199672e-05, "loss": 0.1371, "step": 19520 }, { "epoch": 4.0, "learning_rate": 1.1992619926199263e-05, "loss": 0.1274, "step": 19530 }, { "epoch": 4.01, "learning_rate": 1.1988519885198853e-05, "loss": 0.1769, "step": 19540 }, { "epoch": 4.01, "learning_rate": 1.1984419844198443e-05, "loss": 0.0334, "step": 19550 }, { "epoch": 4.01, "learning_rate": 1.1980319803198033e-05, "loss": 0.0994, "step": 19560 }, { "epoch": 4.01, "learning_rate": 1.1976219762197623e-05, "loss": 0.2865, "step": 19570 }, { "epoch": 4.01, "learning_rate": 1.1972119721197213e-05, "loss": 0.1241, "step": 19580 }, { "epoch": 4.02, "learning_rate": 1.1968019680196803e-05, "loss": 0.2888, "step": 19590 }, { "epoch": 4.02, "learning_rate": 1.1963919639196393e-05, "loss": 0.1588, "step": 19600 }, { "epoch": 4.02, "learning_rate": 1.1959819598195983e-05, "loss": 0.09, "step": 19610 }, { "epoch": 4.02, "learning_rate": 1.1955719557195573e-05, "loss": 0.1183, "step": 19620 }, { "epoch": 4.02, "learning_rate": 1.1951619516195162e-05, "loss": 0.0962, "step": 19630 }, { "epoch": 4.03, "learning_rate": 1.1947519475194754e-05, "loss": 0.1242, "step": 19640 }, { "epoch": 4.03, "learning_rate": 1.1943419434194344e-05, "loss": 0.2024, "step": 19650 }, { "epoch": 4.03, "learning_rate": 1.1939319393193934e-05, "loss": 0.1098, "step": 19660 }, { "epoch": 4.03, "learning_rate": 1.1935219352193524e-05, "loss": 0.4222, "step": 19670 }, { "epoch": 4.03, "learning_rate": 1.1931119311193114e-05, "loss": 0.204, "step": 19680 }, { "epoch": 4.04, "learning_rate": 1.1927019270192704e-05, "loss": 0.1585, "step": 19690 }, { "epoch": 4.04, "learning_rate": 1.1922919229192293e-05, "loss": 0.1498, "step": 19700 }, { "epoch": 4.04, "learning_rate": 1.1918819188191883e-05, "loss": 0.1759, "step": 19710 }, { "epoch": 4.04, "learning_rate": 1.1914719147191473e-05, "loss": 0.1839, "step": 19720 }, { "epoch": 4.04, "learning_rate": 1.1910619106191063e-05, "loss": 0.1044, "step": 19730 }, { "epoch": 4.05, "learning_rate": 1.1906519065190653e-05, "loss": 0.0821, "step": 19740 }, { "epoch": 4.05, "learning_rate": 1.1902419024190245e-05, "loss": 0.1106, "step": 19750 }, { "epoch": 4.05, "learning_rate": 1.1898318983189834e-05, "loss": 0.1164, "step": 19760 }, { "epoch": 4.05, "learning_rate": 1.1894218942189424e-05, "loss": 0.2521, "step": 19770 }, { "epoch": 4.05, "learning_rate": 1.1890118901189014e-05, "loss": 0.1195, "step": 19780 }, { "epoch": 4.06, "learning_rate": 1.1886018860188604e-05, "loss": 0.2051, "step": 19790 }, { "epoch": 4.06, "learning_rate": 1.1881918819188192e-05, "loss": 0.2517, "step": 19800 }, { "epoch": 4.06, "learning_rate": 1.1877818778187782e-05, "loss": 0.2673, "step": 19810 }, { "epoch": 4.06, "learning_rate": 1.1873718737187372e-05, "loss": 0.2674, "step": 19820 }, { "epoch": 4.07, "learning_rate": 1.1869618696186962e-05, "loss": 0.1966, "step": 19830 }, { "epoch": 4.07, "learning_rate": 1.1865518655186552e-05, "loss": 0.1046, "step": 19840 }, { "epoch": 4.07, "learning_rate": 1.1861418614186142e-05, "loss": 0.2629, "step": 19850 }, { "epoch": 4.07, "learning_rate": 1.1857318573185732e-05, "loss": 0.1897, "step": 19860 }, { "epoch": 4.07, "learning_rate": 1.1853218532185322e-05, "loss": 0.1477, "step": 19870 }, { "epoch": 4.08, "learning_rate": 1.1849118491184912e-05, "loss": 0.1682, "step": 19880 }, { "epoch": 4.08, "learning_rate": 1.1845018450184501e-05, "loss": 0.1287, "step": 19890 }, { "epoch": 4.08, "learning_rate": 1.1840918409184093e-05, "loss": 0.1875, "step": 19900 }, { "epoch": 4.08, "learning_rate": 1.1836818368183683e-05, "loss": 0.2704, "step": 19910 }, { "epoch": 4.08, "learning_rate": 1.1832718327183273e-05, "loss": 0.1491, "step": 19920 }, { "epoch": 4.09, "learning_rate": 1.1828618286182863e-05, "loss": 0.2184, "step": 19930 }, { "epoch": 4.09, "learning_rate": 1.1824518245182453e-05, "loss": 0.0643, "step": 19940 }, { "epoch": 4.09, "learning_rate": 1.1820418204182043e-05, "loss": 0.2177, "step": 19950 }, { "epoch": 4.09, "learning_rate": 1.1816318163181632e-05, "loss": 0.1171, "step": 19960 }, { "epoch": 4.09, "learning_rate": 1.1812218122181222e-05, "loss": 0.1821, "step": 19970 }, { "epoch": 4.1, "learning_rate": 1.1808118081180812e-05, "loss": 0.2271, "step": 19980 }, { "epoch": 4.1, "learning_rate": 1.1804018040180402e-05, "loss": 0.2652, "step": 19990 }, { "epoch": 4.1, "learning_rate": 1.1799917999179992e-05, "loss": 0.2873, "step": 20000 }, { "epoch": 4.1, "learning_rate": 1.1795817958179584e-05, "loss": 0.2018, "step": 20010 }, { "epoch": 4.1, "learning_rate": 1.1791717917179174e-05, "loss": 0.206, "step": 20020 }, { "epoch": 4.11, "learning_rate": 1.1787617876178763e-05, "loss": 0.1798, "step": 20030 }, { "epoch": 4.11, "learning_rate": 1.1783517835178353e-05, "loss": 0.1882, "step": 20040 }, { "epoch": 4.11, "learning_rate": 1.1779417794177943e-05, "loss": 0.1463, "step": 20050 }, { "epoch": 4.11, "learning_rate": 1.1775317753177533e-05, "loss": 0.1273, "step": 20060 }, { "epoch": 4.11, "learning_rate": 1.1771217712177123e-05, "loss": 0.1597, "step": 20070 }, { "epoch": 4.12, "learning_rate": 1.1767117671176713e-05, "loss": 0.1404, "step": 20080 }, { "epoch": 4.12, "learning_rate": 1.1763017630176303e-05, "loss": 0.1409, "step": 20090 }, { "epoch": 4.12, "learning_rate": 1.1758917589175893e-05, "loss": 0.1477, "step": 20100 }, { "epoch": 4.12, "learning_rate": 1.1754817548175483e-05, "loss": 0.1958, "step": 20110 }, { "epoch": 4.12, "learning_rate": 1.1750717507175074e-05, "loss": 0.126, "step": 20120 }, { "epoch": 4.13, "learning_rate": 1.1746617466174664e-05, "loss": 0.2874, "step": 20130 }, { "epoch": 4.13, "learning_rate": 1.1742517425174254e-05, "loss": 0.2115, "step": 20140 }, { "epoch": 4.13, "learning_rate": 1.1738417384173844e-05, "loss": 0.0538, "step": 20150 }, { "epoch": 4.13, "learning_rate": 1.1734317343173434e-05, "loss": 0.1833, "step": 20160 }, { "epoch": 4.13, "learning_rate": 1.1730217302173024e-05, "loss": 0.1104, "step": 20170 }, { "epoch": 4.14, "learning_rate": 1.1726117261172614e-05, "loss": 0.2213, "step": 20180 }, { "epoch": 4.14, "learning_rate": 1.1722017220172203e-05, "loss": 0.1455, "step": 20190 }, { "epoch": 4.14, "learning_rate": 1.1717917179171792e-05, "loss": 0.0932, "step": 20200 }, { "epoch": 4.14, "learning_rate": 1.1713817138171382e-05, "loss": 0.2128, "step": 20210 }, { "epoch": 4.15, "learning_rate": 1.1709717097170971e-05, "loss": 0.1086, "step": 20220 }, { "epoch": 4.15, "learning_rate": 1.1705617056170561e-05, "loss": 0.1657, "step": 20230 }, { "epoch": 4.15, "learning_rate": 1.1701517015170151e-05, "loss": 0.2714, "step": 20240 }, { "epoch": 4.15, "learning_rate": 1.1697416974169741e-05, "loss": 0.2281, "step": 20250 }, { "epoch": 4.15, "learning_rate": 1.1693316933169333e-05, "loss": 0.1765, "step": 20260 }, { "epoch": 4.16, "learning_rate": 1.1689216892168923e-05, "loss": 0.1773, "step": 20270 }, { "epoch": 4.16, "learning_rate": 1.1685116851168513e-05, "loss": 0.1574, "step": 20280 }, { "epoch": 4.16, "learning_rate": 1.1681016810168102e-05, "loss": 0.1631, "step": 20290 }, { "epoch": 4.16, "learning_rate": 1.1676916769167692e-05, "loss": 0.1943, "step": 20300 }, { "epoch": 4.16, "learning_rate": 1.1672816728167282e-05, "loss": 0.2347, "step": 20310 }, { "epoch": 4.17, "learning_rate": 1.1668716687166872e-05, "loss": 0.107, "step": 20320 }, { "epoch": 4.17, "learning_rate": 1.1664616646166462e-05, "loss": 0.2684, "step": 20330 }, { "epoch": 4.17, "learning_rate": 1.1660516605166052e-05, "loss": 0.1375, "step": 20340 }, { "epoch": 4.17, "learning_rate": 1.1656416564165642e-05, "loss": 0.0787, "step": 20350 }, { "epoch": 4.17, "learning_rate": 1.1652316523165232e-05, "loss": 0.2697, "step": 20360 }, { "epoch": 4.18, "learning_rate": 1.1648216482164823e-05, "loss": 0.2137, "step": 20370 }, { "epoch": 4.18, "learning_rate": 1.1644116441164413e-05, "loss": 0.1388, "step": 20380 }, { "epoch": 4.18, "learning_rate": 1.1640016400164003e-05, "loss": 0.1568, "step": 20390 }, { "epoch": 4.18, "learning_rate": 1.1635916359163593e-05, "loss": 0.1304, "step": 20400 }, { "epoch": 4.18, "learning_rate": 1.1631816318163183e-05, "loss": 0.1661, "step": 20410 }, { "epoch": 4.19, "learning_rate": 1.1627716277162773e-05, "loss": 0.1252, "step": 20420 }, { "epoch": 4.19, "learning_rate": 1.1623616236162363e-05, "loss": 0.1957, "step": 20430 }, { "epoch": 4.19, "learning_rate": 1.1619516195161953e-05, "loss": 0.2279, "step": 20440 }, { "epoch": 4.19, "learning_rate": 1.1615416154161542e-05, "loss": 0.2439, "step": 20450 }, { "epoch": 4.19, "learning_rate": 1.1611316113161132e-05, "loss": 0.2369, "step": 20460 }, { "epoch": 4.2, "learning_rate": 1.1607216072160722e-05, "loss": 0.2531, "step": 20470 }, { "epoch": 4.2, "learning_rate": 1.1603116031160314e-05, "loss": 0.1928, "step": 20480 }, { "epoch": 4.2, "learning_rate": 1.1599015990159904e-05, "loss": 0.1831, "step": 20490 }, { "epoch": 4.2, "learning_rate": 1.1594915949159494e-05, "loss": 0.1544, "step": 20500 }, { "epoch": 4.2, "learning_rate": 1.1590815908159084e-05, "loss": 0.163, "step": 20510 }, { "epoch": 4.21, "learning_rate": 1.1586715867158673e-05, "loss": 0.1087, "step": 20520 }, { "epoch": 4.21, "learning_rate": 1.1582615826158263e-05, "loss": 0.1531, "step": 20530 }, { "epoch": 4.21, "learning_rate": 1.1578515785157853e-05, "loss": 0.1892, "step": 20540 }, { "epoch": 4.21, "learning_rate": 1.1574415744157443e-05, "loss": 0.175, "step": 20550 }, { "epoch": 4.21, "learning_rate": 1.1570315703157033e-05, "loss": 0.3216, "step": 20560 }, { "epoch": 4.22, "learning_rate": 1.1566215662156623e-05, "loss": 0.1772, "step": 20570 }, { "epoch": 4.22, "learning_rate": 1.1562115621156213e-05, "loss": 0.2532, "step": 20580 }, { "epoch": 4.22, "learning_rate": 1.1558015580155804e-05, "loss": 0.2548, "step": 20590 }, { "epoch": 4.22, "learning_rate": 1.1553915539155391e-05, "loss": 0.1801, "step": 20600 }, { "epoch": 4.23, "learning_rate": 1.154981549815498e-05, "loss": 0.1845, "step": 20610 }, { "epoch": 4.23, "learning_rate": 1.154571545715457e-05, "loss": 0.1446, "step": 20620 }, { "epoch": 4.23, "learning_rate": 1.1541615416154162e-05, "loss": 0.1278, "step": 20630 }, { "epoch": 4.23, "learning_rate": 1.1537515375153752e-05, "loss": 0.295, "step": 20640 }, { "epoch": 4.23, "learning_rate": 1.1533415334153342e-05, "loss": 0.123, "step": 20650 }, { "epoch": 4.24, "learning_rate": 1.1529315293152932e-05, "loss": 0.1446, "step": 20660 }, { "epoch": 4.24, "learning_rate": 1.1525215252152522e-05, "loss": 0.219, "step": 20670 }, { "epoch": 4.24, "learning_rate": 1.1521115211152112e-05, "loss": 0.2148, "step": 20680 }, { "epoch": 4.24, "learning_rate": 1.1517015170151702e-05, "loss": 0.1635, "step": 20690 }, { "epoch": 4.24, "learning_rate": 1.1512915129151292e-05, "loss": 0.1803, "step": 20700 }, { "epoch": 4.25, "learning_rate": 1.1508815088150881e-05, "loss": 0.1803, "step": 20710 }, { "epoch": 4.25, "learning_rate": 1.1504715047150471e-05, "loss": 0.2562, "step": 20720 }, { "epoch": 4.25, "learning_rate": 1.1500615006150061e-05, "loss": 0.1013, "step": 20730 }, { "epoch": 4.25, "learning_rate": 1.1496514965149653e-05, "loss": 0.157, "step": 20740 }, { "epoch": 4.25, "learning_rate": 1.1492414924149243e-05, "loss": 0.2014, "step": 20750 }, { "epoch": 4.26, "learning_rate": 1.1488314883148833e-05, "loss": 0.1692, "step": 20760 }, { "epoch": 4.26, "learning_rate": 1.1484214842148423e-05, "loss": 0.1588, "step": 20770 }, { "epoch": 4.26, "learning_rate": 1.1480114801148012e-05, "loss": 0.1527, "step": 20780 }, { "epoch": 4.26, "learning_rate": 1.1476014760147602e-05, "loss": 0.1155, "step": 20790 }, { "epoch": 4.26, "learning_rate": 1.1471914719147192e-05, "loss": 0.1893, "step": 20800 }, { "epoch": 4.27, "learning_rate": 1.1467814678146782e-05, "loss": 0.1157, "step": 20810 }, { "epoch": 4.27, "learning_rate": 1.1463714637146372e-05, "loss": 0.182, "step": 20820 }, { "epoch": 4.27, "learning_rate": 1.1459614596145962e-05, "loss": 0.249, "step": 20830 }, { "epoch": 4.27, "learning_rate": 1.1455514555145552e-05, "loss": 0.2082, "step": 20840 }, { "epoch": 4.27, "learning_rate": 1.1451414514145143e-05, "loss": 0.1168, "step": 20850 }, { "epoch": 4.28, "learning_rate": 1.1447314473144733e-05, "loss": 0.1597, "step": 20860 }, { "epoch": 4.28, "learning_rate": 1.1443214432144323e-05, "loss": 0.2155, "step": 20870 }, { "epoch": 4.28, "learning_rate": 1.1439114391143913e-05, "loss": 0.1206, "step": 20880 }, { "epoch": 4.28, "learning_rate": 1.1435014350143503e-05, "loss": 0.1426, "step": 20890 }, { "epoch": 4.28, "learning_rate": 1.1430914309143093e-05, "loss": 0.1129, "step": 20900 }, { "epoch": 4.29, "learning_rate": 1.1426814268142683e-05, "loss": 0.1279, "step": 20910 }, { "epoch": 4.29, "learning_rate": 1.1422714227142273e-05, "loss": 0.2392, "step": 20920 }, { "epoch": 4.29, "learning_rate": 1.1418614186141863e-05, "loss": 0.1364, "step": 20930 }, { "epoch": 4.29, "learning_rate": 1.1414514145141452e-05, "loss": 0.1675, "step": 20940 }, { "epoch": 4.29, "learning_rate": 1.1410414104141042e-05, "loss": 0.1486, "step": 20950 }, { "epoch": 4.3, "learning_rate": 1.1406314063140634e-05, "loss": 0.1526, "step": 20960 }, { "epoch": 4.3, "learning_rate": 1.1402214022140224e-05, "loss": 0.1091, "step": 20970 }, { "epoch": 4.3, "learning_rate": 1.1398113981139814e-05, "loss": 0.2329, "step": 20980 }, { "epoch": 4.3, "learning_rate": 1.1394013940139404e-05, "loss": 0.2758, "step": 20990 }, { "epoch": 4.31, "learning_rate": 1.1389913899138992e-05, "loss": 0.1558, "step": 21000 }, { "epoch": 4.31, "learning_rate": 1.1385813858138582e-05, "loss": 0.32, "step": 21010 }, { "epoch": 4.31, "learning_rate": 1.1381713817138172e-05, "loss": 0.1263, "step": 21020 }, { "epoch": 4.31, "learning_rate": 1.1377613776137762e-05, "loss": 0.0691, "step": 21030 }, { "epoch": 4.31, "learning_rate": 1.1373513735137351e-05, "loss": 0.1975, "step": 21040 }, { "epoch": 4.32, "learning_rate": 1.1369413694136941e-05, "loss": 0.1527, "step": 21050 }, { "epoch": 4.32, "learning_rate": 1.1365313653136531e-05, "loss": 0.1668, "step": 21060 }, { "epoch": 4.32, "learning_rate": 1.1361213612136121e-05, "loss": 0.1663, "step": 21070 }, { "epoch": 4.32, "learning_rate": 1.1357113571135711e-05, "loss": 0.2357, "step": 21080 }, { "epoch": 4.32, "learning_rate": 1.1353013530135301e-05, "loss": 0.1797, "step": 21090 }, { "epoch": 4.33, "learning_rate": 1.1348913489134893e-05, "loss": 0.254, "step": 21100 }, { "epoch": 4.33, "learning_rate": 1.1344813448134482e-05, "loss": 0.1241, "step": 21110 }, { "epoch": 4.33, "learning_rate": 1.1340713407134072e-05, "loss": 0.1098, "step": 21120 }, { "epoch": 4.33, "learning_rate": 1.1336613366133662e-05, "loss": 0.1302, "step": 21130 }, { "epoch": 4.33, "learning_rate": 1.1332513325133252e-05, "loss": 0.1138, "step": 21140 }, { "epoch": 4.34, "learning_rate": 1.1328413284132842e-05, "loss": 0.2215, "step": 21150 }, { "epoch": 4.34, "learning_rate": 1.1324313243132432e-05, "loss": 0.2137, "step": 21160 }, { "epoch": 4.34, "learning_rate": 1.1320213202132022e-05, "loss": 0.1513, "step": 21170 }, { "epoch": 4.34, "learning_rate": 1.1316113161131612e-05, "loss": 0.1936, "step": 21180 }, { "epoch": 4.34, "learning_rate": 1.1312013120131202e-05, "loss": 0.0953, "step": 21190 }, { "epoch": 4.35, "learning_rate": 1.1307913079130792e-05, "loss": 0.1605, "step": 21200 }, { "epoch": 4.35, "learning_rate": 1.1303813038130383e-05, "loss": 0.1563, "step": 21210 }, { "epoch": 4.35, "learning_rate": 1.1299712997129973e-05, "loss": 0.2455, "step": 21220 }, { "epoch": 4.35, "learning_rate": 1.1295612956129563e-05, "loss": 0.1304, "step": 21230 }, { "epoch": 4.35, "learning_rate": 1.1291512915129153e-05, "loss": 0.2813, "step": 21240 }, { "epoch": 4.36, "learning_rate": 1.1287412874128743e-05, "loss": 0.2066, "step": 21250 }, { "epoch": 4.36, "learning_rate": 1.1283312833128333e-05, "loss": 0.2644, "step": 21260 }, { "epoch": 4.36, "learning_rate": 1.1279212792127922e-05, "loss": 0.1616, "step": 21270 }, { "epoch": 4.36, "learning_rate": 1.1275112751127512e-05, "loss": 0.1561, "step": 21280 }, { "epoch": 4.36, "learning_rate": 1.1271012710127102e-05, "loss": 0.135, "step": 21290 }, { "epoch": 4.37, "learning_rate": 1.1266912669126692e-05, "loss": 0.2639, "step": 21300 }, { "epoch": 4.37, "learning_rate": 1.1262812628126282e-05, "loss": 0.1693, "step": 21310 }, { "epoch": 4.37, "learning_rate": 1.1258712587125874e-05, "loss": 0.17, "step": 21320 }, { "epoch": 4.37, "learning_rate": 1.1254612546125464e-05, "loss": 0.2043, "step": 21330 }, { "epoch": 4.37, "learning_rate": 1.1250512505125053e-05, "loss": 0.3191, "step": 21340 }, { "epoch": 4.38, "learning_rate": 1.1246412464124643e-05, "loss": 0.1409, "step": 21350 }, { "epoch": 4.38, "learning_rate": 1.1242312423124233e-05, "loss": 0.2001, "step": 21360 }, { "epoch": 4.38, "learning_rate": 1.1238212382123823e-05, "loss": 0.2163, "step": 21370 }, { "epoch": 4.38, "learning_rate": 1.1234112341123413e-05, "loss": 0.1531, "step": 21380 }, { "epoch": 4.38, "learning_rate": 1.1230012300123003e-05, "loss": 0.1186, "step": 21390 }, { "epoch": 4.39, "learning_rate": 1.1225912259122591e-05, "loss": 0.0913, "step": 21400 }, { "epoch": 4.39, "learning_rate": 1.1221812218122181e-05, "loss": 0.1604, "step": 21410 }, { "epoch": 4.39, "learning_rate": 1.1217712177121771e-05, "loss": 0.1634, "step": 21420 }, { "epoch": 4.39, "learning_rate": 1.121361213612136e-05, "loss": 0.251, "step": 21430 }, { "epoch": 4.4, "learning_rate": 1.120951209512095e-05, "loss": 0.0838, "step": 21440 }, { "epoch": 4.4, "learning_rate": 1.120541205412054e-05, "loss": 0.2192, "step": 21450 }, { "epoch": 4.4, "learning_rate": 1.1201312013120132e-05, "loss": 0.2353, "step": 21460 }, { "epoch": 4.4, "learning_rate": 1.1197211972119722e-05, "loss": 0.076, "step": 21470 }, { "epoch": 4.4, "learning_rate": 1.1193111931119312e-05, "loss": 0.1837, "step": 21480 }, { "epoch": 4.41, "learning_rate": 1.1189011890118902e-05, "loss": 0.2637, "step": 21490 }, { "epoch": 4.41, "learning_rate": 1.1184911849118492e-05, "loss": 0.1846, "step": 21500 }, { "epoch": 4.41, "learning_rate": 1.1180811808118082e-05, "loss": 0.1999, "step": 21510 }, { "epoch": 4.41, "learning_rate": 1.1176711767117672e-05, "loss": 0.2924, "step": 21520 }, { "epoch": 4.41, "learning_rate": 1.1172611726117261e-05, "loss": 0.1574, "step": 21530 }, { "epoch": 4.42, "learning_rate": 1.1168511685116851e-05, "loss": 0.159, "step": 21540 }, { "epoch": 4.42, "learning_rate": 1.1164411644116441e-05, "loss": 0.2479, "step": 21550 }, { "epoch": 4.42, "learning_rate": 1.1160311603116031e-05, "loss": 0.1457, "step": 21560 }, { "epoch": 4.42, "learning_rate": 1.1156211562115623e-05, "loss": 0.2515, "step": 21570 }, { "epoch": 4.42, "learning_rate": 1.1152111521115213e-05, "loss": 0.1655, "step": 21580 }, { "epoch": 4.43, "learning_rate": 1.1148011480114803e-05, "loss": 0.1512, "step": 21590 }, { "epoch": 4.43, "learning_rate": 1.1143911439114392e-05, "loss": 0.1566, "step": 21600 }, { "epoch": 4.43, "learning_rate": 1.1139811398113982e-05, "loss": 0.1907, "step": 21610 }, { "epoch": 4.43, "learning_rate": 1.1135711357113572e-05, "loss": 0.1449, "step": 21620 }, { "epoch": 4.43, "learning_rate": 1.1131611316113162e-05, "loss": 0.1442, "step": 21630 }, { "epoch": 4.44, "learning_rate": 1.1127511275112752e-05, "loss": 0.1186, "step": 21640 }, { "epoch": 4.44, "learning_rate": 1.1123411234112342e-05, "loss": 0.1921, "step": 21650 }, { "epoch": 4.44, "learning_rate": 1.1119311193111932e-05, "loss": 0.3467, "step": 21660 }, { "epoch": 4.44, "learning_rate": 1.1115211152111522e-05, "loss": 0.134, "step": 21670 }, { "epoch": 4.44, "learning_rate": 1.1111111111111113e-05, "loss": 0.1572, "step": 21680 }, { "epoch": 4.45, "learning_rate": 1.1107011070110703e-05, "loss": 0.239, "step": 21690 }, { "epoch": 4.45, "learning_rate": 1.1102911029110293e-05, "loss": 0.1847, "step": 21700 }, { "epoch": 4.45, "learning_rate": 1.1098810988109883e-05, "loss": 0.1682, "step": 21710 }, { "epoch": 4.45, "learning_rate": 1.1094710947109473e-05, "loss": 0.2746, "step": 21720 }, { "epoch": 4.45, "learning_rate": 1.1090610906109063e-05, "loss": 0.1148, "step": 21730 }, { "epoch": 4.46, "learning_rate": 1.1086510865108653e-05, "loss": 0.193, "step": 21740 }, { "epoch": 4.46, "learning_rate": 1.1082410824108243e-05, "loss": 0.1792, "step": 21750 }, { "epoch": 4.46, "learning_rate": 1.1078310783107833e-05, "loss": 0.1901, "step": 21760 }, { "epoch": 4.46, "learning_rate": 1.1074210742107422e-05, "loss": 0.1714, "step": 21770 }, { "epoch": 4.46, "learning_rate": 1.1070110701107012e-05, "loss": 0.2769, "step": 21780 }, { "epoch": 4.47, "learning_rate": 1.1066010660106604e-05, "loss": 0.1975, "step": 21790 }, { "epoch": 4.47, "learning_rate": 1.106191061910619e-05, "loss": 0.2048, "step": 21800 }, { "epoch": 4.47, "learning_rate": 1.105781057810578e-05, "loss": 0.2011, "step": 21810 }, { "epoch": 4.47, "learning_rate": 1.105371053710537e-05, "loss": 0.094, "step": 21820 }, { "epoch": 4.48, "learning_rate": 1.1049610496104962e-05, "loss": 0.1856, "step": 21830 }, { "epoch": 4.48, "learning_rate": 1.1045510455104552e-05, "loss": 0.2625, "step": 21840 }, { "epoch": 4.48, "learning_rate": 1.1041410414104142e-05, "loss": 0.1089, "step": 21850 }, { "epoch": 4.48, "learning_rate": 1.1037310373103731e-05, "loss": 0.283, "step": 21860 }, { "epoch": 4.48, "learning_rate": 1.1033210332103321e-05, "loss": 0.2326, "step": 21870 }, { "epoch": 4.49, "learning_rate": 1.1029110291102911e-05, "loss": 0.0884, "step": 21880 }, { "epoch": 4.49, "learning_rate": 1.1025010250102501e-05, "loss": 0.105, "step": 21890 }, { "epoch": 4.49, "learning_rate": 1.1020910209102091e-05, "loss": 0.2508, "step": 21900 }, { "epoch": 4.49, "learning_rate": 1.1016810168101681e-05, "loss": 0.1309, "step": 21910 }, { "epoch": 4.49, "learning_rate": 1.1012710127101271e-05, "loss": 0.1383, "step": 21920 }, { "epoch": 4.5, "learning_rate": 1.100861008610086e-05, "loss": 0.3003, "step": 21930 }, { "epoch": 4.5, "learning_rate": 1.1004510045100452e-05, "loss": 0.2951, "step": 21940 }, { "epoch": 4.5, "learning_rate": 1.1000410004100042e-05, "loss": 0.2211, "step": 21950 }, { "epoch": 4.5, "learning_rate": 1.0996309963099632e-05, "loss": 0.2572, "step": 21960 }, { "epoch": 4.5, "learning_rate": 1.0992209922099222e-05, "loss": 0.2779, "step": 21970 }, { "epoch": 4.51, "learning_rate": 1.0988109881098812e-05, "loss": 0.1427, "step": 21980 }, { "epoch": 4.51, "learning_rate": 1.0984009840098402e-05, "loss": 0.1766, "step": 21990 }, { "epoch": 4.51, "learning_rate": 1.0979909799097992e-05, "loss": 0.124, "step": 22000 }, { "epoch": 4.51, "learning_rate": 1.0975809758097582e-05, "loss": 0.1176, "step": 22010 }, { "epoch": 4.51, "learning_rate": 1.0971709717097172e-05, "loss": 0.1362, "step": 22020 }, { "epoch": 4.52, "learning_rate": 1.0967609676096761e-05, "loss": 0.142, "step": 22030 }, { "epoch": 4.52, "learning_rate": 1.0963509635096351e-05, "loss": 0.2215, "step": 22040 }, { "epoch": 4.52, "learning_rate": 1.0959409594095943e-05, "loss": 0.124, "step": 22050 }, { "epoch": 4.52, "learning_rate": 1.0955309553095533e-05, "loss": 0.15, "step": 22060 }, { "epoch": 4.52, "learning_rate": 1.0951209512095123e-05, "loss": 0.1934, "step": 22070 }, { "epoch": 4.53, "learning_rate": 1.0947109471094713e-05, "loss": 0.2463, "step": 22080 }, { "epoch": 4.53, "learning_rate": 1.0943009430094303e-05, "loss": 0.1635, "step": 22090 }, { "epoch": 4.53, "learning_rate": 1.0938909389093892e-05, "loss": 0.1376, "step": 22100 }, { "epoch": 4.53, "learning_rate": 1.0934809348093482e-05, "loss": 0.213, "step": 22110 }, { "epoch": 4.53, "learning_rate": 1.0930709307093072e-05, "loss": 0.1986, "step": 22120 }, { "epoch": 4.54, "learning_rate": 1.0926609266092662e-05, "loss": 0.1614, "step": 22130 }, { "epoch": 4.54, "learning_rate": 1.0922509225092252e-05, "loss": 0.2121, "step": 22140 }, { "epoch": 4.54, "learning_rate": 1.0918409184091842e-05, "loss": 0.139, "step": 22150 }, { "epoch": 4.54, "learning_rate": 1.0914309143091433e-05, "loss": 0.1523, "step": 22160 }, { "epoch": 4.54, "learning_rate": 1.0910209102091023e-05, "loss": 0.1966, "step": 22170 }, { "epoch": 4.55, "learning_rate": 1.0906109061090613e-05, "loss": 0.2792, "step": 22180 }, { "epoch": 4.55, "learning_rate": 1.0902009020090203e-05, "loss": 0.2165, "step": 22190 }, { "epoch": 4.55, "learning_rate": 1.0897908979089791e-05, "loss": 0.2712, "step": 22200 }, { "epoch": 4.55, "learning_rate": 1.0893808938089381e-05, "loss": 0.1507, "step": 22210 }, { "epoch": 4.56, "learning_rate": 1.0889708897088971e-05, "loss": 0.225, "step": 22220 }, { "epoch": 4.56, "learning_rate": 1.0885608856088561e-05, "loss": 0.1037, "step": 22230 }, { "epoch": 4.56, "learning_rate": 1.0881508815088151e-05, "loss": 0.1147, "step": 22240 }, { "epoch": 4.56, "learning_rate": 1.0877408774087741e-05, "loss": 0.2359, "step": 22250 }, { "epoch": 4.56, "learning_rate": 1.087330873308733e-05, "loss": 0.1954, "step": 22260 }, { "epoch": 4.57, "learning_rate": 1.086920869208692e-05, "loss": 0.1682, "step": 22270 }, { "epoch": 4.57, "learning_rate": 1.086510865108651e-05, "loss": 0.1446, "step": 22280 }, { "epoch": 4.57, "learning_rate": 1.08610086100861e-05, "loss": 0.1518, "step": 22290 }, { "epoch": 4.57, "learning_rate": 1.0856908569085692e-05, "loss": 0.1739, "step": 22300 }, { "epoch": 4.57, "learning_rate": 1.0852808528085282e-05, "loss": 0.275, "step": 22310 }, { "epoch": 4.58, "learning_rate": 1.0848708487084872e-05, "loss": 0.2375, "step": 22320 }, { "epoch": 4.58, "learning_rate": 1.0844608446084462e-05, "loss": 0.3204, "step": 22330 }, { "epoch": 4.58, "learning_rate": 1.0840508405084052e-05, "loss": 0.1623, "step": 22340 }, { "epoch": 4.58, "learning_rate": 1.0836408364083642e-05, "loss": 0.1361, "step": 22350 }, { "epoch": 4.58, "learning_rate": 1.0832308323083231e-05, "loss": 0.1498, "step": 22360 }, { "epoch": 4.59, "learning_rate": 1.0828208282082821e-05, "loss": 0.1682, "step": 22370 }, { "epoch": 4.59, "learning_rate": 1.0824108241082411e-05, "loss": 0.1819, "step": 22380 }, { "epoch": 4.59, "learning_rate": 1.0820008200082001e-05, "loss": 0.147, "step": 22390 }, { "epoch": 4.59, "learning_rate": 1.0815908159081591e-05, "loss": 0.1801, "step": 22400 }, { "epoch": 4.59, "learning_rate": 1.0811808118081183e-05, "loss": 0.1707, "step": 22410 }, { "epoch": 4.6, "learning_rate": 1.0807708077080772e-05, "loss": 0.143, "step": 22420 }, { "epoch": 4.6, "learning_rate": 1.0803608036080362e-05, "loss": 0.2171, "step": 22430 }, { "epoch": 4.6, "learning_rate": 1.0799507995079952e-05, "loss": 0.0855, "step": 22440 }, { "epoch": 4.6, "learning_rate": 1.0795407954079542e-05, "loss": 0.1405, "step": 22450 }, { "epoch": 4.6, "learning_rate": 1.0791307913079132e-05, "loss": 0.1278, "step": 22460 }, { "epoch": 4.61, "learning_rate": 1.0787207872078722e-05, "loss": 0.206, "step": 22470 }, { "epoch": 4.61, "learning_rate": 1.0783107831078312e-05, "loss": 0.1346, "step": 22480 }, { "epoch": 4.61, "learning_rate": 1.0779007790077902e-05, "loss": 0.2939, "step": 22490 }, { "epoch": 4.61, "learning_rate": 1.0774907749077492e-05, "loss": 0.2001, "step": 22500 }, { "epoch": 4.61, "learning_rate": 1.0770807708077082e-05, "loss": 0.1247, "step": 22510 }, { "epoch": 4.62, "learning_rate": 1.0766707667076673e-05, "loss": 0.2372, "step": 22520 }, { "epoch": 4.62, "learning_rate": 1.0762607626076263e-05, "loss": 0.0601, "step": 22530 }, { "epoch": 4.62, "learning_rate": 1.0758507585075853e-05, "loss": 0.2674, "step": 22540 }, { "epoch": 4.62, "learning_rate": 1.0754407544075443e-05, "loss": 0.1485, "step": 22550 }, { "epoch": 4.62, "learning_rate": 1.0750307503075033e-05, "loss": 0.1703, "step": 22560 }, { "epoch": 4.63, "learning_rate": 1.0746207462074623e-05, "loss": 0.2582, "step": 22570 }, { "epoch": 4.63, "learning_rate": 1.0742107421074213e-05, "loss": 0.1221, "step": 22580 }, { "epoch": 4.63, "learning_rate": 1.0738007380073802e-05, "loss": 0.1734, "step": 22590 }, { "epoch": 4.63, "learning_rate": 1.073390733907339e-05, "loss": 0.0955, "step": 22600 }, { "epoch": 4.64, "learning_rate": 1.072980729807298e-05, "loss": 0.1016, "step": 22610 }, { "epoch": 4.64, "learning_rate": 1.072570725707257e-05, "loss": 0.2169, "step": 22620 }, { "epoch": 4.64, "learning_rate": 1.072160721607216e-05, "loss": 0.1688, "step": 22630 }, { "epoch": 4.64, "learning_rate": 1.071750717507175e-05, "loss": 0.2092, "step": 22640 }, { "epoch": 4.64, "learning_rate": 1.071340713407134e-05, "loss": 0.3326, "step": 22650 }, { "epoch": 4.65, "learning_rate": 1.070930709307093e-05, "loss": 0.3489, "step": 22660 }, { "epoch": 4.65, "learning_rate": 1.0705207052070522e-05, "loss": 0.1948, "step": 22670 }, { "epoch": 4.65, "learning_rate": 1.0701107011070112e-05, "loss": 0.2059, "step": 22680 }, { "epoch": 4.65, "learning_rate": 1.0697006970069701e-05, "loss": 0.2321, "step": 22690 }, { "epoch": 4.65, "learning_rate": 1.0692906929069291e-05, "loss": 0.0994, "step": 22700 }, { "epoch": 4.66, "learning_rate": 1.0688806888068881e-05, "loss": 0.1249, "step": 22710 }, { "epoch": 4.66, "learning_rate": 1.0684706847068471e-05, "loss": 0.1166, "step": 22720 }, { "epoch": 4.66, "learning_rate": 1.0680606806068061e-05, "loss": 0.1115, "step": 22730 }, { "epoch": 4.66, "learning_rate": 1.0676506765067651e-05, "loss": 0.1561, "step": 22740 }, { "epoch": 4.66, "learning_rate": 1.067240672406724e-05, "loss": 0.2413, "step": 22750 }, { "epoch": 4.67, "learning_rate": 1.066830668306683e-05, "loss": 0.1138, "step": 22760 }, { "epoch": 4.67, "learning_rate": 1.0664206642066422e-05, "loss": 0.1381, "step": 22770 }, { "epoch": 4.67, "learning_rate": 1.0660106601066012e-05, "loss": 0.2833, "step": 22780 }, { "epoch": 4.67, "learning_rate": 1.0656006560065602e-05, "loss": 0.1874, "step": 22790 }, { "epoch": 4.67, "learning_rate": 1.0651906519065192e-05, "loss": 0.2023, "step": 22800 }, { "epoch": 4.68, "learning_rate": 1.0647806478064782e-05, "loss": 0.2501, "step": 22810 }, { "epoch": 4.68, "learning_rate": 1.0643706437064372e-05, "loss": 0.2089, "step": 22820 }, { "epoch": 4.68, "learning_rate": 1.0639606396063962e-05, "loss": 0.1344, "step": 22830 }, { "epoch": 4.68, "learning_rate": 1.0635506355063552e-05, "loss": 0.1082, "step": 22840 }, { "epoch": 4.68, "learning_rate": 1.0631406314063141e-05, "loss": 0.2448, "step": 22850 }, { "epoch": 4.69, "learning_rate": 1.0627306273062731e-05, "loss": 0.2076, "step": 22860 }, { "epoch": 4.69, "learning_rate": 1.0623206232062321e-05, "loss": 0.1369, "step": 22870 }, { "epoch": 4.69, "learning_rate": 1.0619106191061913e-05, "loss": 0.1574, "step": 22880 }, { "epoch": 4.69, "learning_rate": 1.0615006150061503e-05, "loss": 0.1418, "step": 22890 }, { "epoch": 4.69, "learning_rate": 1.0610906109061093e-05, "loss": 0.1596, "step": 22900 }, { "epoch": 4.7, "learning_rate": 1.0606806068060683e-05, "loss": 0.2486, "step": 22910 }, { "epoch": 4.7, "learning_rate": 1.0602706027060272e-05, "loss": 0.3464, "step": 22920 }, { "epoch": 4.7, "learning_rate": 1.0598605986059862e-05, "loss": 0.0833, "step": 22930 }, { "epoch": 4.7, "learning_rate": 1.0594505945059452e-05, "loss": 0.1993, "step": 22940 }, { "epoch": 4.7, "learning_rate": 1.0590405904059042e-05, "loss": 0.1033, "step": 22950 }, { "epoch": 4.71, "learning_rate": 1.0586305863058632e-05, "loss": 0.2492, "step": 22960 }, { "epoch": 4.71, "learning_rate": 1.0582205822058222e-05, "loss": 0.1268, "step": 22970 }, { "epoch": 4.71, "learning_rate": 1.0578105781057812e-05, "loss": 0.2161, "step": 22980 }, { "epoch": 4.71, "learning_rate": 1.0574005740057403e-05, "loss": 0.28, "step": 22990 }, { "epoch": 4.72, "learning_rate": 1.056990569905699e-05, "loss": 0.1492, "step": 23000 }, { "epoch": 4.72, "learning_rate": 1.056580565805658e-05, "loss": 0.1865, "step": 23010 }, { "epoch": 4.72, "learning_rate": 1.056170561705617e-05, "loss": 0.2466, "step": 23020 }, { "epoch": 4.72, "learning_rate": 1.0557605576055761e-05, "loss": 0.1418, "step": 23030 }, { "epoch": 4.72, "learning_rate": 1.0553505535055351e-05, "loss": 0.1508, "step": 23040 }, { "epoch": 4.73, "learning_rate": 1.0549405494054941e-05, "loss": 0.1471, "step": 23050 }, { "epoch": 4.73, "learning_rate": 1.0545305453054531e-05, "loss": 0.1934, "step": 23060 }, { "epoch": 4.73, "learning_rate": 1.0541205412054121e-05, "loss": 0.1782, "step": 23070 }, { "epoch": 4.73, "learning_rate": 1.053710537105371e-05, "loss": 0.1469, "step": 23080 }, { "epoch": 4.73, "learning_rate": 1.05330053300533e-05, "loss": 0.1109, "step": 23090 }, { "epoch": 4.74, "learning_rate": 1.052890528905289e-05, "loss": 0.2122, "step": 23100 }, { "epoch": 4.74, "learning_rate": 1.052480524805248e-05, "loss": 0.3454, "step": 23110 }, { "epoch": 4.74, "learning_rate": 1.052070520705207e-05, "loss": 0.2334, "step": 23120 }, { "epoch": 4.74, "learning_rate": 1.051660516605166e-05, "loss": 0.1402, "step": 23130 }, { "epoch": 4.74, "learning_rate": 1.0512505125051252e-05, "loss": 0.2203, "step": 23140 }, { "epoch": 4.75, "learning_rate": 1.0508405084050842e-05, "loss": 0.1038, "step": 23150 }, { "epoch": 4.75, "learning_rate": 1.0504305043050432e-05, "loss": 0.1063, "step": 23160 }, { "epoch": 4.75, "learning_rate": 1.0500205002050022e-05, "loss": 0.151, "step": 23170 }, { "epoch": 4.75, "learning_rate": 1.0496104961049611e-05, "loss": 0.1652, "step": 23180 }, { "epoch": 4.75, "learning_rate": 1.0492004920049201e-05, "loss": 0.1474, "step": 23190 }, { "epoch": 4.76, "learning_rate": 1.0487904879048791e-05, "loss": 0.1111, "step": 23200 }, { "epoch": 4.76, "learning_rate": 1.0483804838048381e-05, "loss": 0.3027, "step": 23210 }, { "epoch": 4.76, "learning_rate": 1.0479704797047971e-05, "loss": 0.1459, "step": 23220 }, { "epoch": 4.76, "learning_rate": 1.0475604756047561e-05, "loss": 0.1253, "step": 23230 }, { "epoch": 4.76, "learning_rate": 1.047150471504715e-05, "loss": 0.2071, "step": 23240 }, { "epoch": 4.77, "learning_rate": 1.0467404674046742e-05, "loss": 0.2566, "step": 23250 }, { "epoch": 4.77, "learning_rate": 1.0463304633046332e-05, "loss": 0.2492, "step": 23260 }, { "epoch": 4.77, "learning_rate": 1.0459204592045922e-05, "loss": 0.1948, "step": 23270 }, { "epoch": 4.77, "learning_rate": 1.0455104551045512e-05, "loss": 0.1673, "step": 23280 }, { "epoch": 4.77, "learning_rate": 1.0451004510045102e-05, "loss": 0.177, "step": 23290 }, { "epoch": 4.78, "learning_rate": 1.0446904469044692e-05, "loss": 0.0936, "step": 23300 }, { "epoch": 4.78, "learning_rate": 1.0442804428044282e-05, "loss": 0.2738, "step": 23310 }, { "epoch": 4.78, "learning_rate": 1.0438704387043872e-05, "loss": 0.1877, "step": 23320 }, { "epoch": 4.78, "learning_rate": 1.0434604346043462e-05, "loss": 0.16, "step": 23330 }, { "epoch": 4.78, "learning_rate": 1.0430504305043051e-05, "loss": 0.1339, "step": 23340 }, { "epoch": 4.79, "learning_rate": 1.0426404264042641e-05, "loss": 0.1234, "step": 23350 }, { "epoch": 4.79, "learning_rate": 1.0422304223042233e-05, "loss": 0.11, "step": 23360 }, { "epoch": 4.79, "learning_rate": 1.0418204182041823e-05, "loss": 0.1729, "step": 23370 }, { "epoch": 4.79, "learning_rate": 1.0414104141041413e-05, "loss": 0.1736, "step": 23380 }, { "epoch": 4.79, "learning_rate": 1.0410004100041003e-05, "loss": 0.1925, "step": 23390 }, { "epoch": 4.8, "learning_rate": 1.0405904059040591e-05, "loss": 0.1865, "step": 23400 }, { "epoch": 4.8, "learning_rate": 1.040180401804018e-05, "loss": 0.2069, "step": 23410 }, { "epoch": 4.8, "learning_rate": 1.039770397703977e-05, "loss": 0.2378, "step": 23420 }, { "epoch": 4.8, "learning_rate": 1.039360393603936e-05, "loss": 0.2041, "step": 23430 }, { "epoch": 4.81, "learning_rate": 1.038950389503895e-05, "loss": 0.1705, "step": 23440 }, { "epoch": 4.81, "learning_rate": 1.038540385403854e-05, "loss": 0.1155, "step": 23450 }, { "epoch": 4.81, "learning_rate": 1.038130381303813e-05, "loss": 0.1849, "step": 23460 }, { "epoch": 4.81, "learning_rate": 1.037720377203772e-05, "loss": 0.1528, "step": 23470 }, { "epoch": 4.81, "learning_rate": 1.037310373103731e-05, "loss": 0.2489, "step": 23480 }, { "epoch": 4.82, "learning_rate": 1.03690036900369e-05, "loss": 0.2154, "step": 23490 }, { "epoch": 4.82, "learning_rate": 1.0364903649036492e-05, "loss": 0.1732, "step": 23500 }, { "epoch": 4.82, "learning_rate": 1.0360803608036081e-05, "loss": 0.1801, "step": 23510 }, { "epoch": 4.82, "learning_rate": 1.0356703567035671e-05, "loss": 0.2023, "step": 23520 }, { "epoch": 4.82, "learning_rate": 1.0352603526035261e-05, "loss": 0.0984, "step": 23530 }, { "epoch": 4.83, "learning_rate": 1.0348503485034851e-05, "loss": 0.2116, "step": 23540 }, { "epoch": 4.83, "learning_rate": 1.0344403444034441e-05, "loss": 0.19, "step": 23550 }, { "epoch": 4.83, "learning_rate": 1.0340303403034031e-05, "loss": 0.2636, "step": 23560 }, { "epoch": 4.83, "learning_rate": 1.033620336203362e-05, "loss": 0.2117, "step": 23570 }, { "epoch": 4.83, "learning_rate": 1.033210332103321e-05, "loss": 0.1379, "step": 23580 }, { "epoch": 4.84, "learning_rate": 1.03280032800328e-05, "loss": 0.2062, "step": 23590 }, { "epoch": 4.84, "learning_rate": 1.032390323903239e-05, "loss": 0.0982, "step": 23600 }, { "epoch": 4.84, "learning_rate": 1.0319803198031982e-05, "loss": 0.2289, "step": 23610 }, { "epoch": 4.84, "learning_rate": 1.0315703157031572e-05, "loss": 0.1757, "step": 23620 }, { "epoch": 4.84, "learning_rate": 1.0311603116031162e-05, "loss": 0.1376, "step": 23630 }, { "epoch": 4.85, "learning_rate": 1.0307503075030752e-05, "loss": 0.1098, "step": 23640 }, { "epoch": 4.85, "learning_rate": 1.0303403034030342e-05, "loss": 0.1735, "step": 23650 }, { "epoch": 4.85, "learning_rate": 1.0299302993029932e-05, "loss": 0.2323, "step": 23660 }, { "epoch": 4.85, "learning_rate": 1.0295202952029521e-05, "loss": 0.1703, "step": 23670 }, { "epoch": 4.85, "learning_rate": 1.0291102911029111e-05, "loss": 0.2324, "step": 23680 }, { "epoch": 4.86, "learning_rate": 1.0287002870028701e-05, "loss": 0.2524, "step": 23690 }, { "epoch": 4.86, "learning_rate": 1.0282902829028291e-05, "loss": 0.1131, "step": 23700 }, { "epoch": 4.86, "learning_rate": 1.0278802788027881e-05, "loss": 0.1252, "step": 23710 }, { "epoch": 4.86, "learning_rate": 1.0274702747027473e-05, "loss": 0.2131, "step": 23720 }, { "epoch": 4.86, "learning_rate": 1.0270602706027063e-05, "loss": 0.1862, "step": 23730 }, { "epoch": 4.87, "learning_rate": 1.0266502665026652e-05, "loss": 0.1816, "step": 23740 }, { "epoch": 4.87, "learning_rate": 1.0262402624026242e-05, "loss": 0.1667, "step": 23750 }, { "epoch": 4.87, "learning_rate": 1.0258302583025832e-05, "loss": 0.1468, "step": 23760 }, { "epoch": 4.87, "learning_rate": 1.0254202542025422e-05, "loss": 0.2233, "step": 23770 }, { "epoch": 4.87, "learning_rate": 1.0250102501025012e-05, "loss": 0.0958, "step": 23780 }, { "epoch": 4.88, "learning_rate": 1.0246002460024602e-05, "loss": 0.1937, "step": 23790 }, { "epoch": 4.88, "learning_rate": 1.024190241902419e-05, "loss": 0.1207, "step": 23800 }, { "epoch": 4.88, "learning_rate": 1.023780237802378e-05, "loss": 0.1364, "step": 23810 }, { "epoch": 4.88, "learning_rate": 1.023370233702337e-05, "loss": 0.1973, "step": 23820 }, { "epoch": 4.89, "learning_rate": 1.022960229602296e-05, "loss": 0.1368, "step": 23830 }, { "epoch": 4.89, "learning_rate": 1.022550225502255e-05, "loss": 0.0771, "step": 23840 }, { "epoch": 4.89, "learning_rate": 1.022140221402214e-05, "loss": 0.2154, "step": 23850 }, { "epoch": 4.89, "learning_rate": 1.021730217302173e-05, "loss": 0.1601, "step": 23860 }, { "epoch": 4.89, "learning_rate": 1.0213202132021321e-05, "loss": 0.1212, "step": 23870 }, { "epoch": 4.9, "learning_rate": 1.0209102091020911e-05, "loss": 0.2988, "step": 23880 }, { "epoch": 4.9, "learning_rate": 1.0205002050020501e-05, "loss": 0.1363, "step": 23890 }, { "epoch": 4.9, "learning_rate": 1.020090200902009e-05, "loss": 0.0973, "step": 23900 }, { "epoch": 4.9, "learning_rate": 1.019680196801968e-05, "loss": 0.079, "step": 23910 }, { "epoch": 4.9, "learning_rate": 1.019270192701927e-05, "loss": 0.1559, "step": 23920 }, { "epoch": 4.91, "learning_rate": 1.018860188601886e-05, "loss": 0.1584, "step": 23930 }, { "epoch": 4.91, "learning_rate": 1.018450184501845e-05, "loss": 0.1654, "step": 23940 }, { "epoch": 4.91, "learning_rate": 1.018040180401804e-05, "loss": 0.1395, "step": 23950 }, { "epoch": 4.91, "learning_rate": 1.017630176301763e-05, "loss": 0.0628, "step": 23960 }, { "epoch": 4.91, "learning_rate": 1.017220172201722e-05, "loss": 0.1793, "step": 23970 }, { "epoch": 4.92, "learning_rate": 1.0168101681016812e-05, "loss": 0.2599, "step": 23980 }, { "epoch": 4.92, "learning_rate": 1.0164001640016402e-05, "loss": 0.3726, "step": 23990 }, { "epoch": 4.92, "learning_rate": 1.0159901599015991e-05, "loss": 0.2904, "step": 24000 }, { "epoch": 4.92, "learning_rate": 1.0155801558015581e-05, "loss": 0.1318, "step": 24010 }, { "epoch": 4.92, "learning_rate": 1.0151701517015171e-05, "loss": 0.2192, "step": 24020 }, { "epoch": 4.93, "learning_rate": 1.0147601476014761e-05, "loss": 0.2298, "step": 24030 }, { "epoch": 4.93, "learning_rate": 1.0143501435014351e-05, "loss": 0.1492, "step": 24040 }, { "epoch": 4.93, "learning_rate": 1.0139401394013941e-05, "loss": 0.1728, "step": 24050 }, { "epoch": 4.93, "learning_rate": 1.013530135301353e-05, "loss": 0.2202, "step": 24060 }, { "epoch": 4.93, "learning_rate": 1.013120131201312e-05, "loss": 0.2823, "step": 24070 }, { "epoch": 4.94, "learning_rate": 1.012710127101271e-05, "loss": 0.1171, "step": 24080 }, { "epoch": 4.94, "learning_rate": 1.0123001230012302e-05, "loss": 0.1885, "step": 24090 }, { "epoch": 4.94, "learning_rate": 1.0118901189011892e-05, "loss": 0.1091, "step": 24100 }, { "epoch": 4.94, "learning_rate": 1.0114801148011482e-05, "loss": 0.147, "step": 24110 }, { "epoch": 4.94, "learning_rate": 1.0110701107011072e-05, "loss": 0.0979, "step": 24120 }, { "epoch": 4.95, "learning_rate": 1.0106601066010662e-05, "loss": 0.2093, "step": 24130 }, { "epoch": 4.95, "learning_rate": 1.0102501025010252e-05, "loss": 0.1505, "step": 24140 }, { "epoch": 4.95, "learning_rate": 1.0098400984009842e-05, "loss": 0.1363, "step": 24150 }, { "epoch": 4.95, "learning_rate": 1.0094300943009431e-05, "loss": 0.1472, "step": 24160 }, { "epoch": 4.95, "learning_rate": 1.0090200902009021e-05, "loss": 0.1134, "step": 24170 }, { "epoch": 4.96, "learning_rate": 1.0086100861008611e-05, "loss": 0.1978, "step": 24180 }, { "epoch": 4.96, "learning_rate": 1.0082000820008201e-05, "loss": 0.2345, "step": 24190 }, { "epoch": 4.96, "learning_rate": 1.007790077900779e-05, "loss": 0.1892, "step": 24200 }, { "epoch": 4.96, "learning_rate": 1.007380073800738e-05, "loss": 0.194, "step": 24210 }, { "epoch": 4.97, "learning_rate": 1.006970069700697e-05, "loss": 0.2456, "step": 24220 }, { "epoch": 4.97, "learning_rate": 1.006560065600656e-05, "loss": 0.256, "step": 24230 }, { "epoch": 4.97, "learning_rate": 1.006150061500615e-05, "loss": 0.129, "step": 24240 }, { "epoch": 4.97, "learning_rate": 1.005740057400574e-05, "loss": 0.1975, "step": 24250 }, { "epoch": 4.97, "learning_rate": 1.005330053300533e-05, "loss": 0.0472, "step": 24260 }, { "epoch": 4.98, "learning_rate": 1.004920049200492e-05, "loss": 0.1439, "step": 24270 }, { "epoch": 4.98, "learning_rate": 1.004510045100451e-05, "loss": 0.2375, "step": 24280 }, { "epoch": 4.98, "learning_rate": 1.00410004100041e-05, "loss": 0.1684, "step": 24290 }, { "epoch": 4.98, "learning_rate": 1.003690036900369e-05, "loss": 0.2107, "step": 24300 }, { "epoch": 4.98, "learning_rate": 1.003280032800328e-05, "loss": 0.1036, "step": 24310 }, { "epoch": 4.99, "learning_rate": 1.002870028700287e-05, "loss": 0.0528, "step": 24320 }, { "epoch": 4.99, "learning_rate": 1.002460024600246e-05, "loss": 0.2652, "step": 24330 }, { "epoch": 4.99, "learning_rate": 1.0020500205002051e-05, "loss": 0.2093, "step": 24340 }, { "epoch": 4.99, "learning_rate": 1.0016400164001641e-05, "loss": 0.1825, "step": 24350 }, { "epoch": 4.99, "learning_rate": 1.0012300123001231e-05, "loss": 0.2022, "step": 24360 }, { "epoch": 5.0, "learning_rate": 1.0008200082000821e-05, "loss": 0.2515, "step": 24370 }, { "epoch": 5.0, "learning_rate": 1.0004100041000411e-05, "loss": 0.1545, "step": 24380 }, { "epoch": 5.0, "learning_rate": 1e-05, "loss": 0.1364, "step": 24390 }, { "epoch": 5.0, "eval_FN": 544, "eval_FP": 1877, "eval_TN": 13882, "eval_TP": 4362, "eval_accuracy": 0.8828453907573192, "eval_f1": 0.7827725437415882, "eval_loss": 0.3944579064846039, "eval_precision": 0.699150504888604, "eval_recall": 0.8891153689359967, "eval_runtime": 141.8353, "eval_samples_per_second": 145.697, "eval_steps_per_second": 9.109, "step": 24390 } ], "logging_steps": 10, "max_steps": 48780, "num_train_epochs": 10, "save_steps": 500, "total_flos": 1.026738271331328e+17, "trial_name": null, "trial_params": null }