{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.6455026455026456, "global_step": 25000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.0000000000000003e-07, "loss": 0.0016, "step": 10 }, { "epoch": 0.0, "learning_rate": 8.000000000000001e-07, "loss": 0.0016, "step": 20 }, { "epoch": 0.0, "learning_rate": 1.2000000000000002e-06, "loss": 0.0014, "step": 30 }, { "epoch": 0.0, "learning_rate": 1.6000000000000001e-06, "loss": 0.0013, "step": 40 }, { "epoch": 0.01, "learning_rate": 2.0000000000000003e-06, "loss": 0.0013, "step": 50 }, { "epoch": 0.01, "learning_rate": 2.4000000000000003e-06, "loss": 0.0011, "step": 60 }, { "epoch": 0.01, "learning_rate": 2.8000000000000003e-06, "loss": 0.0012, "step": 70 }, { "epoch": 0.01, "learning_rate": 3.2000000000000003e-06, "loss": 0.0012, "step": 80 }, { "epoch": 0.01, "learning_rate": 3.6000000000000003e-06, "loss": 0.0013, "step": 90 }, { "epoch": 0.01, "learning_rate": 4.000000000000001e-06, "loss": 0.0012, "step": 100 }, { "epoch": 0.01, "learning_rate": 4.4e-06, "loss": 0.001, "step": 110 }, { "epoch": 0.01, "learning_rate": 4.800000000000001e-06, "loss": 0.0011, "step": 120 }, { "epoch": 0.01, "learning_rate": 5.2e-06, "loss": 0.0011, "step": 130 }, { "epoch": 0.01, "learning_rate": 5.600000000000001e-06, "loss": 0.0012, "step": 140 }, { "epoch": 0.02, "learning_rate": 6e-06, "loss": 0.001, "step": 150 }, { "epoch": 0.02, "learning_rate": 6.4000000000000006e-06, "loss": 0.0013, "step": 160 }, { "epoch": 0.02, "learning_rate": 6.800000000000001e-06, "loss": 0.0013, "step": 170 }, { "epoch": 0.02, "learning_rate": 7.2000000000000005e-06, "loss": 0.001, "step": 180 }, { "epoch": 0.02, "learning_rate": 7.600000000000001e-06, "loss": 0.0011, "step": 190 }, { "epoch": 0.02, "learning_rate": 8.000000000000001e-06, "loss": 0.0011, "step": 200 }, { "epoch": 0.02, "learning_rate": 8.400000000000001e-06, "loss": 0.0009, "step": 210 }, { "epoch": 0.02, "learning_rate": 8.8e-06, "loss": 0.0014, "step": 220 }, { "epoch": 0.02, "learning_rate": 9.200000000000002e-06, "loss": 0.0011, "step": 230 }, { "epoch": 0.03, "learning_rate": 9.600000000000001e-06, "loss": 0.0012, "step": 240 }, { "epoch": 0.03, "learning_rate": 1e-05, "loss": 0.0014, "step": 250 }, { "epoch": 0.03, "learning_rate": 1.04e-05, "loss": 0.0014, "step": 260 }, { "epoch": 0.03, "learning_rate": 1.0800000000000002e-05, "loss": 0.0012, "step": 270 }, { "epoch": 0.03, "learning_rate": 1.1200000000000001e-05, "loss": 0.0011, "step": 280 }, { "epoch": 0.03, "learning_rate": 1.16e-05, "loss": 0.0032, "step": 290 }, { "epoch": 0.03, "learning_rate": 1.2e-05, "loss": 0.0018, "step": 300 }, { "epoch": 0.03, "learning_rate": 1.2400000000000002e-05, "loss": 0.0015, "step": 310 }, { "epoch": 0.03, "learning_rate": 1.2800000000000001e-05, "loss": 0.0012, "step": 320 }, { "epoch": 0.03, "learning_rate": 1.3200000000000002e-05, "loss": 0.0013, "step": 330 }, { "epoch": 0.04, "learning_rate": 1.3600000000000002e-05, "loss": 0.0012, "step": 340 }, { "epoch": 0.04, "learning_rate": 1.4e-05, "loss": 0.001, "step": 350 }, { "epoch": 0.04, "learning_rate": 1.4400000000000001e-05, "loss": 0.0012, "step": 360 }, { "epoch": 0.04, "learning_rate": 1.48e-05, "loss": 0.0012, "step": 370 }, { "epoch": 0.04, "learning_rate": 1.5200000000000002e-05, "loss": 0.0011, "step": 380 }, { "epoch": 0.04, "learning_rate": 1.5600000000000003e-05, "loss": 0.0013, "step": 390 }, { "epoch": 0.04, "learning_rate": 1.6000000000000003e-05, "loss": 0.0013, "step": 400 }, { "epoch": 0.04, "learning_rate": 1.64e-05, "loss": 0.0014, "step": 410 }, { "epoch": 0.04, "learning_rate": 1.6800000000000002e-05, "loss": 0.0012, "step": 420 }, { "epoch": 0.05, "learning_rate": 1.72e-05, "loss": 0.0013, "step": 430 }, { "epoch": 0.05, "learning_rate": 1.76e-05, "loss": 0.001, "step": 440 }, { "epoch": 0.05, "learning_rate": 1.8e-05, "loss": 0.0015, "step": 450 }, { "epoch": 0.05, "learning_rate": 1.8400000000000003e-05, "loss": 0.0016, "step": 460 }, { "epoch": 0.05, "learning_rate": 1.88e-05, "loss": 0.0015, "step": 470 }, { "epoch": 0.05, "learning_rate": 1.9200000000000003e-05, "loss": 0.0013, "step": 480 }, { "epoch": 0.05, "learning_rate": 1.9600000000000002e-05, "loss": 0.0011, "step": 490 }, { "epoch": 0.05, "learning_rate": 2e-05, "loss": 0.0014, "step": 500 }, { "epoch": 0.05, "learning_rate": 1.999281867145422e-05, "loss": 0.0013, "step": 510 }, { "epoch": 0.06, "learning_rate": 1.998563734290844e-05, "loss": 0.0012, "step": 520 }, { "epoch": 0.06, "learning_rate": 1.9978456014362658e-05, "loss": 0.0013, "step": 530 }, { "epoch": 0.06, "learning_rate": 1.9971274685816877e-05, "loss": 0.0014, "step": 540 }, { "epoch": 0.06, "learning_rate": 1.9964093357271096e-05, "loss": 0.0014, "step": 550 }, { "epoch": 0.06, "learning_rate": 1.9956912028725315e-05, "loss": 0.0016, "step": 560 }, { "epoch": 0.06, "learning_rate": 1.9949730700179533e-05, "loss": 0.0012, "step": 570 }, { "epoch": 0.06, "learning_rate": 1.9942549371633756e-05, "loss": 0.0013, "step": 580 }, { "epoch": 0.06, "learning_rate": 1.993536804308797e-05, "loss": 0.0023, "step": 590 }, { "epoch": 0.06, "learning_rate": 1.9928186714542193e-05, "loss": 0.0019, "step": 600 }, { "epoch": 0.06, "learning_rate": 1.992100538599641e-05, "loss": 0.0049, "step": 610 }, { "epoch": 0.07, "learning_rate": 1.991382405745063e-05, "loss": 0.0044, "step": 620 }, { "epoch": 0.07, "learning_rate": 1.9906642728904846e-05, "loss": 0.0016, "step": 630 }, { "epoch": 0.07, "learning_rate": 1.989946140035907e-05, "loss": 0.0014, "step": 640 }, { "epoch": 0.07, "learning_rate": 1.9892280071813287e-05, "loss": 0.0018, "step": 650 }, { "epoch": 0.07, "learning_rate": 1.9885098743267506e-05, "loss": 0.0018, "step": 660 }, { "epoch": 0.07, "learning_rate": 1.9877917414721725e-05, "loss": 0.0013, "step": 670 }, { "epoch": 0.07, "learning_rate": 1.9870736086175944e-05, "loss": 0.0012, "step": 680 }, { "epoch": 0.07, "learning_rate": 1.9863554757630163e-05, "loss": 0.0021, "step": 690 }, { "epoch": 0.07, "learning_rate": 1.985637342908438e-05, "loss": 0.0015, "step": 700 }, { "epoch": 0.08, "learning_rate": 1.98491921005386e-05, "loss": 0.0018, "step": 710 }, { "epoch": 0.08, "learning_rate": 1.9842010771992823e-05, "loss": 0.002, "step": 720 }, { "epoch": 0.08, "learning_rate": 1.9834829443447038e-05, "loss": 0.0016, "step": 730 }, { "epoch": 0.08, "learning_rate": 1.982764811490126e-05, "loss": 0.0051, "step": 740 }, { "epoch": 0.08, "learning_rate": 1.9820466786355476e-05, "loss": 0.0018, "step": 750 }, { "epoch": 0.08, "learning_rate": 1.9813285457809698e-05, "loss": 0.0016, "step": 760 }, { "epoch": 0.08, "learning_rate": 1.9806104129263913e-05, "loss": 0.0012, "step": 770 }, { "epoch": 0.08, "learning_rate": 1.9798922800718136e-05, "loss": 0.0011, "step": 780 }, { "epoch": 0.08, "learning_rate": 1.9791741472172355e-05, "loss": 0.0012, "step": 790 }, { "epoch": 0.08, "learning_rate": 1.9784560143626573e-05, "loss": 0.0009, "step": 800 }, { "epoch": 0.09, "learning_rate": 1.9777378815080792e-05, "loss": 0.0012, "step": 810 }, { "epoch": 0.09, "learning_rate": 1.977019748653501e-05, "loss": 0.0013, "step": 820 }, { "epoch": 0.09, "learning_rate": 1.976301615798923e-05, "loss": 0.0012, "step": 830 }, { "epoch": 0.09, "learning_rate": 1.975583482944345e-05, "loss": 0.0012, "step": 840 }, { "epoch": 0.09, "learning_rate": 1.9748653500897668e-05, "loss": 0.0011, "step": 850 }, { "epoch": 0.09, "learning_rate": 1.9741472172351886e-05, "loss": 0.0015, "step": 860 }, { "epoch": 0.09, "learning_rate": 1.9734290843806105e-05, "loss": 0.0013, "step": 870 }, { "epoch": 0.09, "learning_rate": 1.9727109515260324e-05, "loss": 0.0012, "step": 880 }, { "epoch": 0.09, "learning_rate": 1.9719928186714543e-05, "loss": 0.0013, "step": 890 }, { "epoch": 0.1, "learning_rate": 1.971274685816876e-05, "loss": 0.0013, "step": 900 }, { "epoch": 0.1, "learning_rate": 1.970556552962298e-05, "loss": 0.0012, "step": 910 }, { "epoch": 0.1, "learning_rate": 1.96983842010772e-05, "loss": 0.0013, "step": 920 }, { "epoch": 0.1, "learning_rate": 1.969120287253142e-05, "loss": 0.0013, "step": 930 }, { "epoch": 0.1, "learning_rate": 1.9684021543985637e-05, "loss": 0.0012, "step": 940 }, { "epoch": 0.1, "learning_rate": 1.967684021543986e-05, "loss": 0.0012, "step": 950 }, { "epoch": 0.1, "learning_rate": 1.9669658886894075e-05, "loss": 0.0011, "step": 960 }, { "epoch": 0.1, "learning_rate": 1.9662477558348297e-05, "loss": 0.0012, "step": 970 }, { "epoch": 0.1, "learning_rate": 1.9655296229802516e-05, "loss": 0.0013, "step": 980 }, { "epoch": 0.1, "learning_rate": 1.9648114901256735e-05, "loss": 0.0013, "step": 990 }, { "epoch": 0.11, "learning_rate": 1.9640933572710953e-05, "loss": 0.0013, "step": 1000 }, { "epoch": 0.11, "learning_rate": 1.9633752244165172e-05, "loss": 0.0012, "step": 1010 }, { "epoch": 0.11, "learning_rate": 1.962657091561939e-05, "loss": 0.0013, "step": 1020 }, { "epoch": 0.11, "learning_rate": 1.961938958707361e-05, "loss": 0.0012, "step": 1030 }, { "epoch": 0.11, "learning_rate": 1.961220825852783e-05, "loss": 0.0012, "step": 1040 }, { "epoch": 0.11, "learning_rate": 1.9605026929982048e-05, "loss": 0.0014, "step": 1050 }, { "epoch": 0.11, "learning_rate": 1.9597845601436266e-05, "loss": 0.0012, "step": 1060 }, { "epoch": 0.11, "learning_rate": 1.959066427289049e-05, "loss": 0.0013, "step": 1070 }, { "epoch": 0.11, "learning_rate": 1.9583482944344704e-05, "loss": 0.0012, "step": 1080 }, { "epoch": 0.12, "learning_rate": 1.9576301615798926e-05, "loss": 0.0016, "step": 1090 }, { "epoch": 0.12, "learning_rate": 1.9569120287253142e-05, "loss": 0.0012, "step": 1100 }, { "epoch": 0.12, "learning_rate": 1.9561938958707364e-05, "loss": 0.0013, "step": 1110 }, { "epoch": 0.12, "learning_rate": 1.955475763016158e-05, "loss": 0.0011, "step": 1120 }, { "epoch": 0.12, "learning_rate": 1.95475763016158e-05, "loss": 0.0013, "step": 1130 }, { "epoch": 0.12, "learning_rate": 1.954039497307002e-05, "loss": 0.0009, "step": 1140 }, { "epoch": 0.12, "learning_rate": 1.953321364452424e-05, "loss": 0.0013, "step": 1150 }, { "epoch": 0.12, "learning_rate": 1.9526032315978458e-05, "loss": 0.0017, "step": 1160 }, { "epoch": 0.12, "learning_rate": 1.9518850987432677e-05, "loss": 0.0013, "step": 1170 }, { "epoch": 0.12, "learning_rate": 1.9511669658886896e-05, "loss": 0.0012, "step": 1180 }, { "epoch": 0.13, "learning_rate": 1.9504488330341115e-05, "loss": 0.0011, "step": 1190 }, { "epoch": 0.13, "learning_rate": 1.9497307001795333e-05, "loss": 0.0014, "step": 1200 }, { "epoch": 0.13, "learning_rate": 1.9490125673249552e-05, "loss": 0.0012, "step": 1210 }, { "epoch": 0.13, "learning_rate": 1.948294434470377e-05, "loss": 0.0014, "step": 1220 }, { "epoch": 0.13, "learning_rate": 1.947576301615799e-05, "loss": 0.0013, "step": 1230 }, { "epoch": 0.13, "learning_rate": 1.946858168761221e-05, "loss": 0.0013, "step": 1240 }, { "epoch": 0.13, "learning_rate": 1.9461400359066428e-05, "loss": 0.0012, "step": 1250 }, { "epoch": 0.13, "learning_rate": 1.9454219030520646e-05, "loss": 0.0014, "step": 1260 }, { "epoch": 0.13, "learning_rate": 1.9447037701974865e-05, "loss": 0.0012, "step": 1270 }, { "epoch": 0.14, "learning_rate": 1.9439856373429087e-05, "loss": 0.0012, "step": 1280 }, { "epoch": 0.14, "learning_rate": 1.9432675044883306e-05, "loss": 0.0014, "step": 1290 }, { "epoch": 0.14, "learning_rate": 1.9425493716337525e-05, "loss": 0.0012, "step": 1300 }, { "epoch": 0.14, "learning_rate": 1.9418312387791744e-05, "loss": 0.001, "step": 1310 }, { "epoch": 0.14, "learning_rate": 1.9411131059245963e-05, "loss": 0.0011, "step": 1320 }, { "epoch": 0.14, "learning_rate": 1.940394973070018e-05, "loss": 0.0011, "step": 1330 }, { "epoch": 0.14, "learning_rate": 1.93967684021544e-05, "loss": 0.0012, "step": 1340 }, { "epoch": 0.14, "learning_rate": 1.938958707360862e-05, "loss": 0.0011, "step": 1350 }, { "epoch": 0.14, "learning_rate": 1.9382405745062838e-05, "loss": 0.0015, "step": 1360 }, { "epoch": 0.14, "learning_rate": 1.9375224416517057e-05, "loss": 0.0012, "step": 1370 }, { "epoch": 0.15, "learning_rate": 1.9368043087971276e-05, "loss": 0.0012, "step": 1380 }, { "epoch": 0.15, "learning_rate": 1.9360861759425495e-05, "loss": 0.0012, "step": 1390 }, { "epoch": 0.15, "learning_rate": 1.9353680430879713e-05, "loss": 0.0012, "step": 1400 }, { "epoch": 0.15, "learning_rate": 1.9346499102333932e-05, "loss": 0.0011, "step": 1410 }, { "epoch": 0.15, "learning_rate": 1.9339317773788154e-05, "loss": 0.0013, "step": 1420 }, { "epoch": 0.15, "learning_rate": 1.933213644524237e-05, "loss": 0.0013, "step": 1430 }, { "epoch": 0.15, "learning_rate": 1.9324955116696592e-05, "loss": 0.0014, "step": 1440 }, { "epoch": 0.15, "learning_rate": 1.9317773788150808e-05, "loss": 0.0013, "step": 1450 }, { "epoch": 0.15, "learning_rate": 1.931059245960503e-05, "loss": 0.0012, "step": 1460 }, { "epoch": 0.16, "learning_rate": 1.9303411131059245e-05, "loss": 0.0014, "step": 1470 }, { "epoch": 0.16, "learning_rate": 1.9296229802513467e-05, "loss": 0.001, "step": 1480 }, { "epoch": 0.16, "learning_rate": 1.9289048473967686e-05, "loss": 0.0012, "step": 1490 }, { "epoch": 0.16, "learning_rate": 1.9281867145421905e-05, "loss": 0.0013, "step": 1500 }, { "epoch": 0.16, "learning_rate": 1.9274685816876124e-05, "loss": 0.0012, "step": 1510 }, { "epoch": 0.16, "learning_rate": 1.9267504488330343e-05, "loss": 0.0014, "step": 1520 }, { "epoch": 0.16, "learning_rate": 1.926032315978456e-05, "loss": 0.0011, "step": 1530 }, { "epoch": 0.16, "learning_rate": 1.925314183123878e-05, "loss": 0.0011, "step": 1540 }, { "epoch": 0.16, "learning_rate": 1.9245960502693e-05, "loss": 0.0009, "step": 1550 }, { "epoch": 0.17, "learning_rate": 1.9238779174147218e-05, "loss": 0.0013, "step": 1560 }, { "epoch": 0.17, "learning_rate": 1.9231597845601437e-05, "loss": 0.0013, "step": 1570 }, { "epoch": 0.17, "learning_rate": 1.9224416517055656e-05, "loss": 0.0013, "step": 1580 }, { "epoch": 0.17, "learning_rate": 1.9217235188509875e-05, "loss": 0.001, "step": 1590 }, { "epoch": 0.17, "learning_rate": 1.9210053859964093e-05, "loss": 0.001, "step": 1600 }, { "epoch": 0.17, "learning_rate": 1.9202872531418312e-05, "loss": 0.0012, "step": 1610 }, { "epoch": 0.17, "learning_rate": 1.9195691202872534e-05, "loss": 0.0013, "step": 1620 }, { "epoch": 0.17, "learning_rate": 1.9188509874326753e-05, "loss": 0.0013, "step": 1630 }, { "epoch": 0.17, "learning_rate": 1.9181328545780972e-05, "loss": 0.0012, "step": 1640 }, { "epoch": 0.17, "learning_rate": 1.917414721723519e-05, "loss": 0.0013, "step": 1650 }, { "epoch": 0.18, "learning_rate": 1.916696588868941e-05, "loss": 0.0012, "step": 1660 }, { "epoch": 0.18, "learning_rate": 1.915978456014363e-05, "loss": 0.0014, "step": 1670 }, { "epoch": 0.18, "learning_rate": 1.9152603231597847e-05, "loss": 0.0014, "step": 1680 }, { "epoch": 0.18, "learning_rate": 1.9145421903052066e-05, "loss": 0.0012, "step": 1690 }, { "epoch": 0.18, "learning_rate": 1.9138240574506285e-05, "loss": 0.0012, "step": 1700 }, { "epoch": 0.18, "learning_rate": 1.9131059245960504e-05, "loss": 0.0012, "step": 1710 }, { "epoch": 0.18, "learning_rate": 1.9123877917414723e-05, "loss": 0.0011, "step": 1720 }, { "epoch": 0.18, "learning_rate": 1.911669658886894e-05, "loss": 0.0011, "step": 1730 }, { "epoch": 0.18, "learning_rate": 1.910951526032316e-05, "loss": 0.0013, "step": 1740 }, { "epoch": 0.19, "learning_rate": 1.910233393177738e-05, "loss": 0.0013, "step": 1750 }, { "epoch": 0.19, "learning_rate": 1.9095152603231598e-05, "loss": 0.0015, "step": 1760 }, { "epoch": 0.19, "learning_rate": 1.908797127468582e-05, "loss": 0.0013, "step": 1770 }, { "epoch": 0.19, "learning_rate": 1.9080789946140036e-05, "loss": 0.0013, "step": 1780 }, { "epoch": 0.19, "learning_rate": 1.9073608617594258e-05, "loss": 0.001, "step": 1790 }, { "epoch": 0.19, "learning_rate": 1.9066427289048473e-05, "loss": 0.0011, "step": 1800 }, { "epoch": 0.19, "learning_rate": 1.9059245960502696e-05, "loss": 0.0012, "step": 1810 }, { "epoch": 0.19, "learning_rate": 1.9052064631956915e-05, "loss": 0.0011, "step": 1820 }, { "epoch": 0.19, "learning_rate": 1.9044883303411133e-05, "loss": 0.0013, "step": 1830 }, { "epoch": 0.19, "learning_rate": 1.9037701974865352e-05, "loss": 0.0011, "step": 1840 }, { "epoch": 0.2, "learning_rate": 1.903052064631957e-05, "loss": 0.0013, "step": 1850 }, { "epoch": 0.2, "learning_rate": 1.902333931777379e-05, "loss": 0.0012, "step": 1860 }, { "epoch": 0.2, "learning_rate": 1.901615798922801e-05, "loss": 0.0012, "step": 1870 }, { "epoch": 0.2, "learning_rate": 1.9008976660682228e-05, "loss": 0.0012, "step": 1880 }, { "epoch": 0.2, "learning_rate": 1.9001795332136446e-05, "loss": 0.0012, "step": 1890 }, { "epoch": 0.2, "learning_rate": 1.8994614003590665e-05, "loss": 0.0013, "step": 1900 }, { "epoch": 0.2, "learning_rate": 1.8987432675044884e-05, "loss": 0.0012, "step": 1910 }, { "epoch": 0.2, "learning_rate": 1.8980251346499103e-05, "loss": 0.0013, "step": 1920 }, { "epoch": 0.2, "learning_rate": 1.8973070017953325e-05, "loss": 0.0012, "step": 1930 }, { "epoch": 0.21, "learning_rate": 1.896588868940754e-05, "loss": 0.0015, "step": 1940 }, { "epoch": 0.21, "learning_rate": 1.8958707360861763e-05, "loss": 0.0013, "step": 1950 }, { "epoch": 0.21, "learning_rate": 1.895152603231598e-05, "loss": 0.0014, "step": 1960 }, { "epoch": 0.21, "learning_rate": 1.89443447037702e-05, "loss": 0.0012, "step": 1970 }, { "epoch": 0.21, "learning_rate": 1.893716337522442e-05, "loss": 0.0014, "step": 1980 }, { "epoch": 0.21, "learning_rate": 1.8929982046678638e-05, "loss": 0.0013, "step": 1990 }, { "epoch": 0.21, "learning_rate": 1.8922800718132857e-05, "loss": 0.0013, "step": 2000 }, { "epoch": 0.21, "learning_rate": 1.8915619389587076e-05, "loss": 0.0013, "step": 2010 }, { "epoch": 0.21, "learning_rate": 1.8908438061041295e-05, "loss": 0.0012, "step": 2020 }, { "epoch": 0.21, "learning_rate": 1.8901256732495513e-05, "loss": 0.0012, "step": 2030 }, { "epoch": 0.22, "learning_rate": 1.8894075403949732e-05, "loss": 0.0015, "step": 2040 }, { "epoch": 0.22, "learning_rate": 1.888689407540395e-05, "loss": 0.0012, "step": 2050 }, { "epoch": 0.22, "learning_rate": 1.887971274685817e-05, "loss": 0.0011, "step": 2060 }, { "epoch": 0.22, "learning_rate": 1.887253141831239e-05, "loss": 0.0015, "step": 2070 }, { "epoch": 0.22, "learning_rate": 1.8865350089766608e-05, "loss": 0.0011, "step": 2080 }, { "epoch": 0.22, "learning_rate": 1.8858168761220826e-05, "loss": 0.0013, "step": 2090 }, { "epoch": 0.22, "learning_rate": 1.885098743267505e-05, "loss": 0.0013, "step": 2100 }, { "epoch": 0.22, "learning_rate": 1.8843806104129264e-05, "loss": 0.0015, "step": 2110 }, { "epoch": 0.22, "learning_rate": 1.8836624775583486e-05, "loss": 0.001, "step": 2120 }, { "epoch": 0.23, "learning_rate": 1.88294434470377e-05, "loss": 0.0012, "step": 2130 }, { "epoch": 0.23, "learning_rate": 1.8822262118491924e-05, "loss": 0.0012, "step": 2140 }, { "epoch": 0.23, "learning_rate": 1.881508078994614e-05, "loss": 0.0012, "step": 2150 }, { "epoch": 0.23, "learning_rate": 1.880789946140036e-05, "loss": 0.0012, "step": 2160 }, { "epoch": 0.23, "learning_rate": 1.880071813285458e-05, "loss": 0.0012, "step": 2170 }, { "epoch": 0.23, "learning_rate": 1.87935368043088e-05, "loss": 0.001, "step": 2180 }, { "epoch": 0.23, "learning_rate": 1.8786355475763018e-05, "loss": 0.0014, "step": 2190 }, { "epoch": 0.23, "learning_rate": 1.8779174147217237e-05, "loss": 0.0013, "step": 2200 }, { "epoch": 0.23, "learning_rate": 1.8771992818671456e-05, "loss": 0.0012, "step": 2210 }, { "epoch": 0.23, "learning_rate": 1.8764811490125675e-05, "loss": 0.0013, "step": 2220 }, { "epoch": 0.24, "learning_rate": 1.8757630161579893e-05, "loss": 0.0012, "step": 2230 }, { "epoch": 0.24, "learning_rate": 1.8750448833034112e-05, "loss": 0.0014, "step": 2240 }, { "epoch": 0.24, "learning_rate": 1.874326750448833e-05, "loss": 0.0012, "step": 2250 }, { "epoch": 0.24, "learning_rate": 1.8736086175942553e-05, "loss": 0.0012, "step": 2260 }, { "epoch": 0.24, "learning_rate": 1.872890484739677e-05, "loss": 0.0015, "step": 2270 }, { "epoch": 0.24, "learning_rate": 1.872172351885099e-05, "loss": 0.0014, "step": 2280 }, { "epoch": 0.24, "learning_rate": 1.8714542190305206e-05, "loss": 0.0012, "step": 2290 }, { "epoch": 0.24, "learning_rate": 1.870736086175943e-05, "loss": 0.0011, "step": 2300 }, { "epoch": 0.24, "learning_rate": 1.8700179533213647e-05, "loss": 0.0011, "step": 2310 }, { "epoch": 0.25, "learning_rate": 1.8692998204667866e-05, "loss": 0.001, "step": 2320 }, { "epoch": 0.25, "learning_rate": 1.8685816876122085e-05, "loss": 0.0012, "step": 2330 }, { "epoch": 0.25, "learning_rate": 1.8678635547576304e-05, "loss": 0.0015, "step": 2340 }, { "epoch": 0.25, "learning_rate": 1.8671454219030523e-05, "loss": 0.0015, "step": 2350 }, { "epoch": 0.25, "learning_rate": 1.866427289048474e-05, "loss": 0.0015, "step": 2360 }, { "epoch": 0.25, "learning_rate": 1.865709156193896e-05, "loss": 0.0013, "step": 2370 }, { "epoch": 0.25, "learning_rate": 1.864991023339318e-05, "loss": 0.0011, "step": 2380 }, { "epoch": 0.25, "learning_rate": 1.8642728904847398e-05, "loss": 0.0013, "step": 2390 }, { "epoch": 0.25, "learning_rate": 1.8635547576301617e-05, "loss": 0.0011, "step": 2400 }, { "epoch": 0.26, "learning_rate": 1.8628366247755836e-05, "loss": 0.0012, "step": 2410 }, { "epoch": 0.26, "learning_rate": 1.8621184919210055e-05, "loss": 0.0009, "step": 2420 }, { "epoch": 0.26, "learning_rate": 1.8614003590664273e-05, "loss": 0.0013, "step": 2430 }, { "epoch": 0.26, "learning_rate": 1.8606822262118492e-05, "loss": 0.0013, "step": 2440 }, { "epoch": 0.26, "learning_rate": 1.8599640933572714e-05, "loss": 0.0013, "step": 2450 }, { "epoch": 0.26, "learning_rate": 1.859245960502693e-05, "loss": 0.0013, "step": 2460 }, { "epoch": 0.26, "learning_rate": 1.8585278276481152e-05, "loss": 0.0015, "step": 2470 }, { "epoch": 0.26, "learning_rate": 1.8578096947935368e-05, "loss": 0.0013, "step": 2480 }, { "epoch": 0.26, "learning_rate": 1.857091561938959e-05, "loss": 0.0009, "step": 2490 }, { "epoch": 0.26, "learning_rate": 1.8563734290843805e-05, "loss": 0.0014, "step": 2500 }, { "epoch": 0.27, "learning_rate": 1.8556552962298027e-05, "loss": 0.0013, "step": 2510 }, { "epoch": 0.27, "learning_rate": 1.8549371633752246e-05, "loss": 0.0012, "step": 2520 }, { "epoch": 0.27, "learning_rate": 1.8542190305206465e-05, "loss": 0.0012, "step": 2530 }, { "epoch": 0.27, "learning_rate": 1.8535008976660684e-05, "loss": 0.0013, "step": 2540 }, { "epoch": 0.27, "learning_rate": 1.8527827648114903e-05, "loss": 0.0011, "step": 2550 }, { "epoch": 0.27, "learning_rate": 1.852064631956912e-05, "loss": 0.0011, "step": 2560 }, { "epoch": 0.27, "learning_rate": 1.851346499102334e-05, "loss": 0.0013, "step": 2570 }, { "epoch": 0.27, "learning_rate": 1.850628366247756e-05, "loss": 0.0014, "step": 2580 }, { "epoch": 0.27, "learning_rate": 1.849910233393178e-05, "loss": 0.0012, "step": 2590 }, { "epoch": 0.28, "learning_rate": 1.8491921005385997e-05, "loss": 0.0013, "step": 2600 }, { "epoch": 0.28, "learning_rate": 1.848473967684022e-05, "loss": 0.0012, "step": 2610 }, { "epoch": 0.28, "learning_rate": 1.8477558348294435e-05, "loss": 0.0012, "step": 2620 }, { "epoch": 0.28, "learning_rate": 1.8470377019748657e-05, "loss": 0.001, "step": 2630 }, { "epoch": 0.28, "learning_rate": 1.8463195691202872e-05, "loss": 0.0012, "step": 2640 }, { "epoch": 0.28, "learning_rate": 1.8456014362657094e-05, "loss": 0.0014, "step": 2650 }, { "epoch": 0.28, "learning_rate": 1.8448833034111313e-05, "loss": 0.0013, "step": 2660 }, { "epoch": 0.28, "learning_rate": 1.8441651705565532e-05, "loss": 0.0015, "step": 2670 }, { "epoch": 0.28, "learning_rate": 1.843447037701975e-05, "loss": 0.0011, "step": 2680 }, { "epoch": 0.28, "learning_rate": 1.842728904847397e-05, "loss": 0.0011, "step": 2690 }, { "epoch": 0.29, "learning_rate": 1.842010771992819e-05, "loss": 0.0012, "step": 2700 }, { "epoch": 0.29, "learning_rate": 1.8412926391382407e-05, "loss": 0.0013, "step": 2710 }, { "epoch": 0.29, "learning_rate": 1.8405745062836626e-05, "loss": 0.0013, "step": 2720 }, { "epoch": 0.29, "learning_rate": 1.8398563734290845e-05, "loss": 0.0012, "step": 2730 }, { "epoch": 0.29, "learning_rate": 1.8391382405745064e-05, "loss": 0.0012, "step": 2740 }, { "epoch": 0.29, "learning_rate": 1.8384201077199283e-05, "loss": 0.001, "step": 2750 }, { "epoch": 0.29, "learning_rate": 1.83770197486535e-05, "loss": 0.0015, "step": 2760 }, { "epoch": 0.29, "learning_rate": 1.836983842010772e-05, "loss": 0.0013, "step": 2770 }, { "epoch": 0.29, "learning_rate": 1.836265709156194e-05, "loss": 0.0012, "step": 2780 }, { "epoch": 0.3, "learning_rate": 1.8355475763016158e-05, "loss": 0.0014, "step": 2790 }, { "epoch": 0.3, "learning_rate": 1.834829443447038e-05, "loss": 0.0013, "step": 2800 }, { "epoch": 0.3, "learning_rate": 1.8341113105924596e-05, "loss": 0.0012, "step": 2810 }, { "epoch": 0.3, "learning_rate": 1.8333931777378818e-05, "loss": 0.0012, "step": 2820 }, { "epoch": 0.3, "learning_rate": 1.8326750448833033e-05, "loss": 0.0011, "step": 2830 }, { "epoch": 0.3, "learning_rate": 1.8319569120287256e-05, "loss": 0.0014, "step": 2840 }, { "epoch": 0.3, "learning_rate": 1.831238779174147e-05, "loss": 0.0013, "step": 2850 }, { "epoch": 0.3, "learning_rate": 1.8305206463195693e-05, "loss": 0.0015, "step": 2860 }, { "epoch": 0.3, "learning_rate": 1.8298025134649912e-05, "loss": 0.0013, "step": 2870 }, { "epoch": 0.3, "learning_rate": 1.829084380610413e-05, "loss": 0.0011, "step": 2880 }, { "epoch": 0.31, "learning_rate": 1.828366247755835e-05, "loss": 0.0012, "step": 2890 }, { "epoch": 0.31, "learning_rate": 1.827648114901257e-05, "loss": 0.0013, "step": 2900 }, { "epoch": 0.31, "learning_rate": 1.8269299820466787e-05, "loss": 0.0013, "step": 2910 }, { "epoch": 0.31, "learning_rate": 1.8262118491921006e-05, "loss": 0.0016, "step": 2920 }, { "epoch": 0.31, "learning_rate": 1.8254937163375225e-05, "loss": 0.001, "step": 2930 }, { "epoch": 0.31, "learning_rate": 1.8247755834829447e-05, "loss": 0.0014, "step": 2940 }, { "epoch": 0.31, "learning_rate": 1.8240574506283663e-05, "loss": 0.0011, "step": 2950 }, { "epoch": 0.31, "learning_rate": 1.8233393177737885e-05, "loss": 0.0011, "step": 2960 }, { "epoch": 0.31, "learning_rate": 1.82262118491921e-05, "loss": 0.0013, "step": 2970 }, { "epoch": 0.32, "learning_rate": 1.8219030520646323e-05, "loss": 0.0012, "step": 2980 }, { "epoch": 0.32, "learning_rate": 1.8211849192100538e-05, "loss": 0.0013, "step": 2990 }, { "epoch": 0.32, "learning_rate": 1.820466786355476e-05, "loss": 0.0015, "step": 3000 }, { "epoch": 0.32, "learning_rate": 1.819748653500898e-05, "loss": 0.0011, "step": 3010 }, { "epoch": 0.32, "learning_rate": 1.8190305206463198e-05, "loss": 0.0012, "step": 3020 }, { "epoch": 0.32, "learning_rate": 1.8183123877917417e-05, "loss": 0.0013, "step": 3030 }, { "epoch": 0.32, "learning_rate": 1.8175942549371636e-05, "loss": 0.0013, "step": 3040 }, { "epoch": 0.32, "learning_rate": 1.8168761220825855e-05, "loss": 0.0013, "step": 3050 }, { "epoch": 0.32, "learning_rate": 1.8161579892280073e-05, "loss": 0.0012, "step": 3060 }, { "epoch": 0.32, "learning_rate": 1.8154398563734292e-05, "loss": 0.0014, "step": 3070 }, { "epoch": 0.33, "learning_rate": 1.814721723518851e-05, "loss": 0.001, "step": 3080 }, { "epoch": 0.33, "learning_rate": 1.814003590664273e-05, "loss": 0.0013, "step": 3090 }, { "epoch": 0.33, "learning_rate": 1.813285457809695e-05, "loss": 0.0014, "step": 3100 }, { "epoch": 0.33, "learning_rate": 1.8125673249551168e-05, "loss": 0.001, "step": 3110 }, { "epoch": 0.33, "learning_rate": 1.8118491921005386e-05, "loss": 0.0012, "step": 3120 }, { "epoch": 0.33, "learning_rate": 1.8111310592459605e-05, "loss": 0.0012, "step": 3130 }, { "epoch": 0.33, "learning_rate": 1.8104129263913824e-05, "loss": 0.0014, "step": 3140 }, { "epoch": 0.33, "learning_rate": 1.8096947935368046e-05, "loss": 0.0014, "step": 3150 }, { "epoch": 0.33, "learning_rate": 1.808976660682226e-05, "loss": 0.0013, "step": 3160 }, { "epoch": 0.34, "learning_rate": 1.8082585278276484e-05, "loss": 0.0011, "step": 3170 }, { "epoch": 0.34, "learning_rate": 1.80754039497307e-05, "loss": 0.0012, "step": 3180 }, { "epoch": 0.34, "learning_rate": 1.806822262118492e-05, "loss": 0.0012, "step": 3190 }, { "epoch": 0.34, "learning_rate": 1.806104129263914e-05, "loss": 0.0016, "step": 3200 }, { "epoch": 0.34, "learning_rate": 1.805385996409336e-05, "loss": 0.0012, "step": 3210 }, { "epoch": 0.34, "learning_rate": 1.8046678635547578e-05, "loss": 0.0011, "step": 3220 }, { "epoch": 0.34, "learning_rate": 1.8039497307001797e-05, "loss": 0.0013, "step": 3230 }, { "epoch": 0.34, "learning_rate": 1.8032315978456016e-05, "loss": 0.0013, "step": 3240 }, { "epoch": 0.34, "learning_rate": 1.8025134649910235e-05, "loss": 0.0011, "step": 3250 }, { "epoch": 0.34, "learning_rate": 1.8017953321364453e-05, "loss": 0.0013, "step": 3260 }, { "epoch": 0.35, "learning_rate": 1.8010771992818672e-05, "loss": 0.0012, "step": 3270 }, { "epoch": 0.35, "learning_rate": 1.800359066427289e-05, "loss": 0.0012, "step": 3280 }, { "epoch": 0.35, "learning_rate": 1.7996409335727113e-05, "loss": 0.001, "step": 3290 }, { "epoch": 0.35, "learning_rate": 1.798922800718133e-05, "loss": 0.0011, "step": 3300 }, { "epoch": 0.35, "learning_rate": 1.798204667863555e-05, "loss": 0.0013, "step": 3310 }, { "epoch": 0.35, "learning_rate": 1.7974865350089766e-05, "loss": 0.0016, "step": 3320 }, { "epoch": 0.35, "learning_rate": 1.796768402154399e-05, "loss": 0.0013, "step": 3330 }, { "epoch": 0.35, "learning_rate": 1.7960502692998204e-05, "loss": 0.0012, "step": 3340 }, { "epoch": 0.35, "learning_rate": 1.7953321364452426e-05, "loss": 0.0011, "step": 3350 }, { "epoch": 0.36, "learning_rate": 1.7946140035906645e-05, "loss": 0.001, "step": 3360 }, { "epoch": 0.36, "learning_rate": 1.7938958707360864e-05, "loss": 0.0014, "step": 3370 }, { "epoch": 0.36, "learning_rate": 1.7931777378815083e-05, "loss": 0.0013, "step": 3380 }, { "epoch": 0.36, "learning_rate": 1.79245960502693e-05, "loss": 0.0014, "step": 3390 }, { "epoch": 0.36, "learning_rate": 1.791741472172352e-05, "loss": 0.0011, "step": 3400 }, { "epoch": 0.36, "learning_rate": 1.791023339317774e-05, "loss": 0.0011, "step": 3410 }, { "epoch": 0.36, "learning_rate": 1.7903052064631958e-05, "loss": 0.0013, "step": 3420 }, { "epoch": 0.36, "learning_rate": 1.7895870736086177e-05, "loss": 0.0011, "step": 3430 }, { "epoch": 0.36, "learning_rate": 1.7888689407540396e-05, "loss": 0.0014, "step": 3440 }, { "epoch": 0.37, "learning_rate": 1.7881508078994615e-05, "loss": 0.0016, "step": 3450 }, { "epoch": 0.37, "learning_rate": 1.7874326750448833e-05, "loss": 0.0014, "step": 3460 }, { "epoch": 0.37, "learning_rate": 1.7867145421903052e-05, "loss": 0.0014, "step": 3470 }, { "epoch": 0.37, "learning_rate": 1.785996409335727e-05, "loss": 0.0012, "step": 3480 }, { "epoch": 0.37, "learning_rate": 1.785278276481149e-05, "loss": 0.0013, "step": 3490 }, { "epoch": 0.37, "learning_rate": 1.7845601436265712e-05, "loss": 0.0012, "step": 3500 }, { "epoch": 0.37, "learning_rate": 1.783842010771993e-05, "loss": 0.0011, "step": 3510 }, { "epoch": 0.37, "learning_rate": 1.783123877917415e-05, "loss": 0.0012, "step": 3520 }, { "epoch": 0.37, "learning_rate": 1.782405745062837e-05, "loss": 0.0012, "step": 3530 }, { "epoch": 0.37, "learning_rate": 1.7816876122082587e-05, "loss": 0.0012, "step": 3540 }, { "epoch": 0.38, "learning_rate": 1.7809694793536806e-05, "loss": 0.0012, "step": 3550 }, { "epoch": 0.38, "learning_rate": 1.7802513464991025e-05, "loss": 0.0015, "step": 3560 }, { "epoch": 0.38, "learning_rate": 1.7795332136445244e-05, "loss": 0.0011, "step": 3570 }, { "epoch": 0.38, "learning_rate": 1.7788150807899463e-05, "loss": 0.0013, "step": 3580 }, { "epoch": 0.38, "learning_rate": 1.778096947935368e-05, "loss": 0.0012, "step": 3590 }, { "epoch": 0.38, "learning_rate": 1.77737881508079e-05, "loss": 0.0013, "step": 3600 }, { "epoch": 0.38, "learning_rate": 1.776660682226212e-05, "loss": 0.0012, "step": 3610 }, { "epoch": 0.38, "learning_rate": 1.7759425493716338e-05, "loss": 0.0013, "step": 3620 }, { "epoch": 0.38, "learning_rate": 1.7752244165170557e-05, "loss": 0.001, "step": 3630 }, { "epoch": 0.39, "learning_rate": 1.774506283662478e-05, "loss": 0.0012, "step": 3640 }, { "epoch": 0.39, "learning_rate": 1.7737881508078995e-05, "loss": 0.0012, "step": 3650 }, { "epoch": 0.39, "learning_rate": 1.7730700179533217e-05, "loss": 0.0014, "step": 3660 }, { "epoch": 0.39, "learning_rate": 1.7723518850987432e-05, "loss": 0.0012, "step": 3670 }, { "epoch": 0.39, "learning_rate": 1.7716337522441654e-05, "loss": 0.0011, "step": 3680 }, { "epoch": 0.39, "learning_rate": 1.770915619389587e-05, "loss": 0.0013, "step": 3690 }, { "epoch": 0.39, "learning_rate": 1.7701974865350092e-05, "loss": 0.0011, "step": 3700 }, { "epoch": 0.39, "learning_rate": 1.769479353680431e-05, "loss": 0.0012, "step": 3710 }, { "epoch": 0.39, "learning_rate": 1.768761220825853e-05, "loss": 0.0011, "step": 3720 }, { "epoch": 0.39, "learning_rate": 1.768043087971275e-05, "loss": 0.0014, "step": 3730 }, { "epoch": 0.4, "learning_rate": 1.7673249551166967e-05, "loss": 0.0011, "step": 3740 }, { "epoch": 0.4, "learning_rate": 1.7666068222621186e-05, "loss": 0.0013, "step": 3750 }, { "epoch": 0.4, "learning_rate": 1.7658886894075405e-05, "loss": 0.0013, "step": 3760 }, { "epoch": 0.4, "learning_rate": 1.7651705565529624e-05, "loss": 0.0008, "step": 3770 }, { "epoch": 0.4, "learning_rate": 1.7644524236983843e-05, "loss": 0.0012, "step": 3780 }, { "epoch": 0.4, "learning_rate": 1.763734290843806e-05, "loss": 0.0014, "step": 3790 }, { "epoch": 0.4, "learning_rate": 1.763016157989228e-05, "loss": 0.0013, "step": 3800 }, { "epoch": 0.4, "learning_rate": 1.76229802513465e-05, "loss": 0.0011, "step": 3810 }, { "epoch": 0.4, "learning_rate": 1.7615798922800718e-05, "loss": 0.0013, "step": 3820 }, { "epoch": 0.41, "learning_rate": 1.7608617594254937e-05, "loss": 0.0013, "step": 3830 }, { "epoch": 0.41, "learning_rate": 1.760143626570916e-05, "loss": 0.0014, "step": 3840 }, { "epoch": 0.41, "learning_rate": 1.7594254937163378e-05, "loss": 0.0012, "step": 3850 }, { "epoch": 0.41, "learning_rate": 1.7587073608617597e-05, "loss": 0.0012, "step": 3860 }, { "epoch": 0.41, "learning_rate": 1.7579892280071816e-05, "loss": 0.0015, "step": 3870 }, { "epoch": 0.41, "learning_rate": 1.7572710951526034e-05, "loss": 0.001, "step": 3880 }, { "epoch": 0.41, "learning_rate": 1.7565529622980253e-05, "loss": 0.0013, "step": 3890 }, { "epoch": 0.41, "learning_rate": 1.7558348294434472e-05, "loss": 0.0013, "step": 3900 }, { "epoch": 0.41, "learning_rate": 1.755116696588869e-05, "loss": 0.0013, "step": 3910 }, { "epoch": 0.41, "learning_rate": 1.754398563734291e-05, "loss": 0.0013, "step": 3920 }, { "epoch": 0.42, "learning_rate": 1.753680430879713e-05, "loss": 0.0012, "step": 3930 }, { "epoch": 0.42, "learning_rate": 1.7529622980251347e-05, "loss": 0.0011, "step": 3940 }, { "epoch": 0.42, "learning_rate": 1.7522441651705566e-05, "loss": 0.0014, "step": 3950 }, { "epoch": 0.42, "learning_rate": 1.7515260323159785e-05, "loss": 0.0013, "step": 3960 }, { "epoch": 0.42, "learning_rate": 1.7508078994614004e-05, "loss": 0.0011, "step": 3970 }, { "epoch": 0.42, "learning_rate": 1.7500897666068223e-05, "loss": 0.0013, "step": 3980 }, { "epoch": 0.42, "learning_rate": 1.7493716337522445e-05, "loss": 0.0015, "step": 3990 }, { "epoch": 0.42, "learning_rate": 1.748653500897666e-05, "loss": 0.0009, "step": 4000 }, { "epoch": 0.42, "learning_rate": 1.7479353680430883e-05, "loss": 0.0012, "step": 4010 }, { "epoch": 0.43, "learning_rate": 1.7472172351885098e-05, "loss": 0.0011, "step": 4020 }, { "epoch": 0.43, "learning_rate": 1.746499102333932e-05, "loss": 0.0013, "step": 4030 }, { "epoch": 0.43, "learning_rate": 1.7457809694793536e-05, "loss": 0.0011, "step": 4040 }, { "epoch": 0.43, "learning_rate": 1.7450628366247758e-05, "loss": 0.0011, "step": 4050 }, { "epoch": 0.43, "learning_rate": 1.7443447037701977e-05, "loss": 0.0013, "step": 4060 }, { "epoch": 0.43, "learning_rate": 1.7436265709156196e-05, "loss": 0.0012, "step": 4070 }, { "epoch": 0.43, "learning_rate": 1.7429084380610415e-05, "loss": 0.0012, "step": 4080 }, { "epoch": 0.43, "learning_rate": 1.7421903052064633e-05, "loss": 0.0015, "step": 4090 }, { "epoch": 0.43, "learning_rate": 1.7414721723518852e-05, "loss": 0.0012, "step": 4100 }, { "epoch": 0.43, "learning_rate": 1.740754039497307e-05, "loss": 0.0012, "step": 4110 }, { "epoch": 0.44, "learning_rate": 1.740035906642729e-05, "loss": 0.0014, "step": 4120 }, { "epoch": 0.44, "learning_rate": 1.739317773788151e-05, "loss": 0.0014, "step": 4130 }, { "epoch": 0.44, "learning_rate": 1.7385996409335728e-05, "loss": 0.0014, "step": 4140 }, { "epoch": 0.44, "learning_rate": 1.737881508078995e-05, "loss": 0.0013, "step": 4150 }, { "epoch": 0.44, "learning_rate": 1.7371633752244165e-05, "loss": 0.0012, "step": 4160 }, { "epoch": 0.44, "learning_rate": 1.7364452423698387e-05, "loss": 0.0011, "step": 4170 }, { "epoch": 0.44, "learning_rate": 1.7357271095152603e-05, "loss": 0.0011, "step": 4180 }, { "epoch": 0.44, "learning_rate": 1.7350089766606825e-05, "loss": 0.0012, "step": 4190 }, { "epoch": 0.44, "learning_rate": 1.7342908438061044e-05, "loss": 0.0012, "step": 4200 }, { "epoch": 0.45, "learning_rate": 1.7335727109515263e-05, "loss": 0.0012, "step": 4210 }, { "epoch": 0.45, "learning_rate": 1.732854578096948e-05, "loss": 0.0014, "step": 4220 }, { "epoch": 0.45, "learning_rate": 1.73213644524237e-05, "loss": 0.0014, "step": 4230 }, { "epoch": 0.45, "learning_rate": 1.731418312387792e-05, "loss": 0.0013, "step": 4240 }, { "epoch": 0.45, "learning_rate": 1.7307001795332138e-05, "loss": 0.001, "step": 4250 }, { "epoch": 0.45, "learning_rate": 1.7299820466786357e-05, "loss": 0.0012, "step": 4260 }, { "epoch": 0.45, "learning_rate": 1.7292639138240576e-05, "loss": 0.0013, "step": 4270 }, { "epoch": 0.45, "learning_rate": 1.7285457809694795e-05, "loss": 0.0011, "step": 4280 }, { "epoch": 0.45, "learning_rate": 1.7278276481149013e-05, "loss": 0.0013, "step": 4290 }, { "epoch": 0.46, "learning_rate": 1.7271095152603232e-05, "loss": 0.0012, "step": 4300 }, { "epoch": 0.46, "learning_rate": 1.726391382405745e-05, "loss": 0.001, "step": 4310 }, { "epoch": 0.46, "learning_rate": 1.7256732495511673e-05, "loss": 0.0014, "step": 4320 }, { "epoch": 0.46, "learning_rate": 1.724955116696589e-05, "loss": 0.0012, "step": 4330 }, { "epoch": 0.46, "learning_rate": 1.724236983842011e-05, "loss": 0.0013, "step": 4340 }, { "epoch": 0.46, "learning_rate": 1.7235188509874326e-05, "loss": 0.0012, "step": 4350 }, { "epoch": 0.46, "learning_rate": 1.722800718132855e-05, "loss": 0.0012, "step": 4360 }, { "epoch": 0.46, "learning_rate": 1.7220825852782764e-05, "loss": 0.0013, "step": 4370 }, { "epoch": 0.46, "learning_rate": 1.7213644524236986e-05, "loss": 0.0013, "step": 4380 }, { "epoch": 0.46, "learning_rate": 1.7206463195691205e-05, "loss": 0.0012, "step": 4390 }, { "epoch": 0.47, "learning_rate": 1.7199281867145424e-05, "loss": 0.0015, "step": 4400 }, { "epoch": 0.47, "learning_rate": 1.7192100538599643e-05, "loss": 0.0015, "step": 4410 }, { "epoch": 0.47, "learning_rate": 1.718491921005386e-05, "loss": 0.0012, "step": 4420 }, { "epoch": 0.47, "learning_rate": 1.717773788150808e-05, "loss": 0.0013, "step": 4430 }, { "epoch": 0.47, "learning_rate": 1.71705565529623e-05, "loss": 0.0012, "step": 4440 }, { "epoch": 0.47, "learning_rate": 1.7163375224416518e-05, "loss": 0.0013, "step": 4450 }, { "epoch": 0.47, "learning_rate": 1.7156193895870737e-05, "loss": 0.0012, "step": 4460 }, { "epoch": 0.47, "learning_rate": 1.7149012567324956e-05, "loss": 0.0011, "step": 4470 }, { "epoch": 0.47, "learning_rate": 1.7141831238779178e-05, "loss": 0.0014, "step": 4480 }, { "epoch": 0.48, "learning_rate": 1.7134649910233393e-05, "loss": 0.0013, "step": 4490 }, { "epoch": 0.48, "learning_rate": 1.7127468581687616e-05, "loss": 0.001, "step": 4500 }, { "epoch": 0.48, "learning_rate": 1.712028725314183e-05, "loss": 0.0014, "step": 4510 }, { "epoch": 0.48, "learning_rate": 1.7113105924596053e-05, "loss": 0.001, "step": 4520 }, { "epoch": 0.48, "learning_rate": 1.7105924596050272e-05, "loss": 0.0012, "step": 4530 }, { "epoch": 0.48, "learning_rate": 1.709874326750449e-05, "loss": 0.0015, "step": 4540 }, { "epoch": 0.48, "learning_rate": 1.709156193895871e-05, "loss": 0.001, "step": 4550 }, { "epoch": 0.48, "learning_rate": 1.708438061041293e-05, "loss": 0.0014, "step": 4560 }, { "epoch": 0.48, "learning_rate": 1.7077199281867147e-05, "loss": 0.0013, "step": 4570 }, { "epoch": 0.48, "learning_rate": 1.7070017953321366e-05, "loss": 0.0013, "step": 4580 }, { "epoch": 0.49, "learning_rate": 1.7062836624775585e-05, "loss": 0.0011, "step": 4590 }, { "epoch": 0.49, "learning_rate": 1.7055655296229804e-05, "loss": 0.0013, "step": 4600 }, { "epoch": 0.49, "learning_rate": 1.7048473967684023e-05, "loss": 0.0011, "step": 4610 }, { "epoch": 0.49, "learning_rate": 1.704129263913824e-05, "loss": 0.0011, "step": 4620 }, { "epoch": 0.49, "learning_rate": 1.703411131059246e-05, "loss": 0.0013, "step": 4630 }, { "epoch": 0.49, "learning_rate": 1.702692998204668e-05, "loss": 0.0012, "step": 4640 }, { "epoch": 0.49, "learning_rate": 1.7019748653500898e-05, "loss": 0.0013, "step": 4650 }, { "epoch": 0.49, "learning_rate": 1.7012567324955117e-05, "loss": 0.0012, "step": 4660 }, { "epoch": 0.49, "learning_rate": 1.700538599640934e-05, "loss": 0.0011, "step": 4670 }, { "epoch": 0.5, "learning_rate": 1.6998204667863555e-05, "loss": 0.0012, "step": 4680 }, { "epoch": 0.5, "learning_rate": 1.6991023339317777e-05, "loss": 0.0016, "step": 4690 }, { "epoch": 0.5, "learning_rate": 1.6983842010771992e-05, "loss": 0.0013, "step": 4700 }, { "epoch": 0.5, "learning_rate": 1.6976660682226214e-05, "loss": 0.0013, "step": 4710 }, { "epoch": 0.5, "learning_rate": 1.696947935368043e-05, "loss": 0.0012, "step": 4720 }, { "epoch": 0.5, "learning_rate": 1.6962298025134652e-05, "loss": 0.001, "step": 4730 }, { "epoch": 0.5, "learning_rate": 1.695511669658887e-05, "loss": 0.0013, "step": 4740 }, { "epoch": 0.5, "learning_rate": 1.694793536804309e-05, "loss": 0.0012, "step": 4750 }, { "epoch": 0.5, "learning_rate": 1.694075403949731e-05, "loss": 0.0015, "step": 4760 }, { "epoch": 0.5, "learning_rate": 1.6933572710951527e-05, "loss": 0.0015, "step": 4770 }, { "epoch": 0.51, "learning_rate": 1.6926391382405746e-05, "loss": 0.001, "step": 4780 }, { "epoch": 0.51, "learning_rate": 1.6919210053859965e-05, "loss": 0.0014, "step": 4790 }, { "epoch": 0.51, "learning_rate": 1.6912028725314184e-05, "loss": 0.0011, "step": 4800 }, { "epoch": 0.51, "learning_rate": 1.6904847396768406e-05, "loss": 0.0009, "step": 4810 }, { "epoch": 0.51, "learning_rate": 1.689766606822262e-05, "loss": 0.0013, "step": 4820 }, { "epoch": 0.51, "learning_rate": 1.6890484739676844e-05, "loss": 0.0012, "step": 4830 }, { "epoch": 0.51, "learning_rate": 1.688330341113106e-05, "loss": 0.001, "step": 4840 }, { "epoch": 0.51, "learning_rate": 1.687612208258528e-05, "loss": 0.0014, "step": 4850 }, { "epoch": 0.51, "learning_rate": 1.6868940754039497e-05, "loss": 0.0013, "step": 4860 }, { "epoch": 0.52, "learning_rate": 1.686175942549372e-05, "loss": 0.0013, "step": 4870 }, { "epoch": 0.52, "learning_rate": 1.6854578096947938e-05, "loss": 0.0016, "step": 4880 }, { "epoch": 0.52, "learning_rate": 1.6847396768402157e-05, "loss": 0.0012, "step": 4890 }, { "epoch": 0.52, "learning_rate": 1.6840215439856376e-05, "loss": 0.0013, "step": 4900 }, { "epoch": 0.52, "learning_rate": 1.6833034111310594e-05, "loss": 0.0014, "step": 4910 }, { "epoch": 0.52, "learning_rate": 1.6825852782764813e-05, "loss": 0.0014, "step": 4920 }, { "epoch": 0.52, "learning_rate": 1.6818671454219032e-05, "loss": 0.0013, "step": 4930 }, { "epoch": 0.52, "learning_rate": 1.681149012567325e-05, "loss": 0.0013, "step": 4940 }, { "epoch": 0.52, "learning_rate": 1.680430879712747e-05, "loss": 0.0009, "step": 4950 }, { "epoch": 0.52, "learning_rate": 1.679712746858169e-05, "loss": 0.0012, "step": 4960 }, { "epoch": 0.53, "learning_rate": 1.6789946140035907e-05, "loss": 0.0014, "step": 4970 }, { "epoch": 0.53, "learning_rate": 1.6782764811490126e-05, "loss": 0.0013, "step": 4980 }, { "epoch": 0.53, "learning_rate": 1.6775583482944345e-05, "loss": 0.0013, "step": 4990 }, { "epoch": 0.53, "learning_rate": 1.6768402154398564e-05, "loss": 0.0012, "step": 5000 }, { "epoch": 0.53, "learning_rate": 1.6761220825852783e-05, "loss": 0.001, "step": 5010 }, { "epoch": 0.53, "learning_rate": 1.6754039497307005e-05, "loss": 0.0013, "step": 5020 }, { "epoch": 0.53, "learning_rate": 1.674685816876122e-05, "loss": 0.0014, "step": 5030 }, { "epoch": 0.53, "learning_rate": 1.6739676840215443e-05, "loss": 0.001, "step": 5040 }, { "epoch": 0.53, "learning_rate": 1.6732495511669658e-05, "loss": 0.0011, "step": 5050 }, { "epoch": 0.54, "learning_rate": 1.672531418312388e-05, "loss": 0.0012, "step": 5060 }, { "epoch": 0.54, "learning_rate": 1.6718132854578096e-05, "loss": 0.0012, "step": 5070 }, { "epoch": 0.54, "learning_rate": 1.6710951526032318e-05, "loss": 0.0012, "step": 5080 }, { "epoch": 0.54, "learning_rate": 1.6703770197486537e-05, "loss": 0.0011, "step": 5090 }, { "epoch": 0.54, "learning_rate": 1.6696588868940756e-05, "loss": 0.0012, "step": 5100 }, { "epoch": 0.54, "learning_rate": 1.6689407540394974e-05, "loss": 0.0012, "step": 5110 }, { "epoch": 0.54, "learning_rate": 1.6682226211849193e-05, "loss": 0.0013, "step": 5120 }, { "epoch": 0.54, "learning_rate": 1.6675044883303412e-05, "loss": 0.0014, "step": 5130 }, { "epoch": 0.54, "learning_rate": 1.666786355475763e-05, "loss": 0.0014, "step": 5140 }, { "epoch": 0.54, "learning_rate": 1.666068222621185e-05, "loss": 0.0012, "step": 5150 }, { "epoch": 0.55, "learning_rate": 1.6653500897666072e-05, "loss": 0.0011, "step": 5160 }, { "epoch": 0.55, "learning_rate": 1.6646319569120287e-05, "loss": 0.0012, "step": 5170 }, { "epoch": 0.55, "learning_rate": 1.663913824057451e-05, "loss": 0.0012, "step": 5180 }, { "epoch": 0.55, "learning_rate": 1.6631956912028725e-05, "loss": 0.0011, "step": 5190 }, { "epoch": 0.55, "learning_rate": 1.6624775583482947e-05, "loss": 0.001, "step": 5200 }, { "epoch": 0.55, "learning_rate": 1.6617594254937163e-05, "loss": 0.0011, "step": 5210 }, { "epoch": 0.55, "learning_rate": 1.6610412926391385e-05, "loss": 0.0011, "step": 5220 }, { "epoch": 0.55, "learning_rate": 1.6603231597845604e-05, "loss": 0.0014, "step": 5230 }, { "epoch": 0.55, "learning_rate": 1.6596050269299823e-05, "loss": 0.0014, "step": 5240 }, { "epoch": 0.56, "learning_rate": 1.658886894075404e-05, "loss": 0.0012, "step": 5250 }, { "epoch": 0.56, "learning_rate": 1.658168761220826e-05, "loss": 0.0013, "step": 5260 }, { "epoch": 0.56, "learning_rate": 1.657450628366248e-05, "loss": 0.0012, "step": 5270 }, { "epoch": 0.56, "learning_rate": 1.6567324955116698e-05, "loss": 0.001, "step": 5280 }, { "epoch": 0.56, "learning_rate": 1.6560143626570917e-05, "loss": 0.0013, "step": 5290 }, { "epoch": 0.56, "learning_rate": 1.6552962298025136e-05, "loss": 0.0012, "step": 5300 }, { "epoch": 0.56, "learning_rate": 1.6545780969479355e-05, "loss": 0.0012, "step": 5310 }, { "epoch": 0.56, "learning_rate": 1.6538599640933573e-05, "loss": 0.0013, "step": 5320 }, { "epoch": 0.56, "learning_rate": 1.6531418312387792e-05, "loss": 0.0011, "step": 5330 }, { "epoch": 0.57, "learning_rate": 1.652423698384201e-05, "loss": 0.0013, "step": 5340 }, { "epoch": 0.57, "learning_rate": 1.651705565529623e-05, "loss": 0.0011, "step": 5350 }, { "epoch": 0.57, "learning_rate": 1.650987432675045e-05, "loss": 0.0012, "step": 5360 }, { "epoch": 0.57, "learning_rate": 1.650269299820467e-05, "loss": 0.0015, "step": 5370 }, { "epoch": 0.57, "learning_rate": 1.6495511669658886e-05, "loss": 0.0014, "step": 5380 }, { "epoch": 0.57, "learning_rate": 1.648833034111311e-05, "loss": 0.0012, "step": 5390 }, { "epoch": 0.57, "learning_rate": 1.6481149012567324e-05, "loss": 0.0011, "step": 5400 }, { "epoch": 0.57, "learning_rate": 1.6473967684021546e-05, "loss": 0.0015, "step": 5410 }, { "epoch": 0.57, "learning_rate": 1.6466786355475765e-05, "loss": 0.0011, "step": 5420 }, { "epoch": 0.57, "learning_rate": 1.6459605026929984e-05, "loss": 0.0012, "step": 5430 }, { "epoch": 0.58, "learning_rate": 1.6452423698384203e-05, "loss": 0.0011, "step": 5440 }, { "epoch": 0.58, "learning_rate": 1.644524236983842e-05, "loss": 0.0011, "step": 5450 }, { "epoch": 0.58, "learning_rate": 1.643806104129264e-05, "loss": 0.0011, "step": 5460 }, { "epoch": 0.58, "learning_rate": 1.643087971274686e-05, "loss": 0.0014, "step": 5470 }, { "epoch": 0.58, "learning_rate": 1.6423698384201078e-05, "loss": 0.0013, "step": 5480 }, { "epoch": 0.58, "learning_rate": 1.6416517055655297e-05, "loss": 0.0011, "step": 5490 }, { "epoch": 0.58, "learning_rate": 1.6409335727109516e-05, "loss": 0.0012, "step": 5500 }, { "epoch": 0.58, "learning_rate": 1.6402154398563738e-05, "loss": 0.0015, "step": 5510 }, { "epoch": 0.58, "learning_rate": 1.6394973070017953e-05, "loss": 0.0016, "step": 5520 }, { "epoch": 0.59, "learning_rate": 1.6387791741472176e-05, "loss": 0.0014, "step": 5530 }, { "epoch": 0.59, "learning_rate": 1.638061041292639e-05, "loss": 0.0017, "step": 5540 }, { "epoch": 0.59, "learning_rate": 1.6373429084380613e-05, "loss": 0.0747, "step": 5550 }, { "epoch": 0.59, "learning_rate": 1.636624775583483e-05, "loss": 0.0379, "step": 5560 }, { "epoch": 0.59, "learning_rate": 1.635906642728905e-05, "loss": 0.0016, "step": 5570 }, { "epoch": 0.59, "learning_rate": 1.635188509874327e-05, "loss": 0.0012, "step": 5580 }, { "epoch": 0.59, "learning_rate": 1.634470377019749e-05, "loss": 0.0014, "step": 5590 }, { "epoch": 0.59, "learning_rate": 1.6337522441651707e-05, "loss": 0.0013, "step": 5600 }, { "epoch": 0.59, "learning_rate": 1.6330341113105926e-05, "loss": 0.0013, "step": 5610 }, { "epoch": 0.59, "learning_rate": 1.6323159784560145e-05, "loss": 0.0013, "step": 5620 }, { "epoch": 0.6, "learning_rate": 1.6315978456014364e-05, "loss": 0.0012, "step": 5630 }, { "epoch": 0.6, "learning_rate": 1.6308797127468583e-05, "loss": 0.0014, "step": 5640 }, { "epoch": 0.6, "learning_rate": 1.63016157989228e-05, "loss": 0.0011, "step": 5650 }, { "epoch": 0.6, "learning_rate": 1.629443447037702e-05, "loss": 0.0013, "step": 5660 }, { "epoch": 0.6, "learning_rate": 1.628725314183124e-05, "loss": 0.0012, "step": 5670 }, { "epoch": 0.6, "learning_rate": 1.6280071813285458e-05, "loss": 0.0013, "step": 5680 }, { "epoch": 0.6, "learning_rate": 1.6272890484739677e-05, "loss": 0.0013, "step": 5690 }, { "epoch": 0.6, "learning_rate": 1.6265709156193896e-05, "loss": 0.0015, "step": 5700 }, { "epoch": 0.6, "learning_rate": 1.6258527827648115e-05, "loss": 0.0011, "step": 5710 }, { "epoch": 0.61, "learning_rate": 1.6251346499102337e-05, "loss": 0.0012, "step": 5720 }, { "epoch": 0.61, "learning_rate": 1.6244165170556556e-05, "loss": 0.0014, "step": 5730 }, { "epoch": 0.61, "learning_rate": 1.6236983842010774e-05, "loss": 0.0013, "step": 5740 }, { "epoch": 0.61, "learning_rate": 1.6229802513464993e-05, "loss": 0.0011, "step": 5750 }, { "epoch": 0.61, "learning_rate": 1.6222621184919212e-05, "loss": 0.0013, "step": 5760 }, { "epoch": 0.61, "learning_rate": 1.621543985637343e-05, "loss": 0.001, "step": 5770 }, { "epoch": 0.61, "learning_rate": 1.620825852782765e-05, "loss": 0.0011, "step": 5780 }, { "epoch": 0.61, "learning_rate": 1.620107719928187e-05, "loss": 0.0014, "step": 5790 }, { "epoch": 0.61, "learning_rate": 1.6193895870736087e-05, "loss": 0.0011, "step": 5800 }, { "epoch": 0.61, "learning_rate": 1.6186714542190306e-05, "loss": 0.0012, "step": 5810 }, { "epoch": 0.62, "learning_rate": 1.6179533213644525e-05, "loss": 0.0015, "step": 5820 }, { "epoch": 0.62, "learning_rate": 1.6172351885098744e-05, "loss": 0.001, "step": 5830 }, { "epoch": 0.62, "learning_rate": 1.6165170556552963e-05, "loss": 0.0013, "step": 5840 }, { "epoch": 0.62, "learning_rate": 1.615798922800718e-05, "loss": 0.0014, "step": 5850 }, { "epoch": 0.62, "learning_rate": 1.6150807899461404e-05, "loss": 0.001, "step": 5860 }, { "epoch": 0.62, "learning_rate": 1.614362657091562e-05, "loss": 0.0011, "step": 5870 }, { "epoch": 0.62, "learning_rate": 1.613644524236984e-05, "loss": 0.0013, "step": 5880 }, { "epoch": 0.62, "learning_rate": 1.6129263913824057e-05, "loss": 0.0012, "step": 5890 }, { "epoch": 0.62, "learning_rate": 1.612208258527828e-05, "loss": 0.0011, "step": 5900 }, { "epoch": 0.63, "learning_rate": 1.6114901256732495e-05, "loss": 0.0011, "step": 5910 }, { "epoch": 0.63, "learning_rate": 1.6107719928186717e-05, "loss": 0.0011, "step": 5920 }, { "epoch": 0.63, "learning_rate": 1.6100538599640936e-05, "loss": 0.0011, "step": 5930 }, { "epoch": 0.63, "learning_rate": 1.6093357271095154e-05, "loss": 0.0013, "step": 5940 }, { "epoch": 0.63, "learning_rate": 1.6086175942549373e-05, "loss": 0.0013, "step": 5950 }, { "epoch": 0.63, "learning_rate": 1.6078994614003592e-05, "loss": 0.0012, "step": 5960 }, { "epoch": 0.63, "learning_rate": 1.607181328545781e-05, "loss": 0.0012, "step": 5970 }, { "epoch": 0.63, "learning_rate": 1.606463195691203e-05, "loss": 0.0014, "step": 5980 }, { "epoch": 0.63, "learning_rate": 1.605745062836625e-05, "loss": 0.0012, "step": 5990 }, { "epoch": 0.63, "learning_rate": 1.6050269299820467e-05, "loss": 0.0012, "step": 6000 }, { "epoch": 0.64, "learning_rate": 1.6043087971274686e-05, "loss": 0.0013, "step": 6010 }, { "epoch": 0.64, "learning_rate": 1.6035906642728905e-05, "loss": 0.0013, "step": 6020 }, { "epoch": 0.64, "learning_rate": 1.6028725314183124e-05, "loss": 0.0012, "step": 6030 }, { "epoch": 0.64, "learning_rate": 1.6021543985637346e-05, "loss": 0.001, "step": 6040 }, { "epoch": 0.64, "learning_rate": 1.601436265709156e-05, "loss": 0.0013, "step": 6050 }, { "epoch": 0.64, "learning_rate": 1.6007181328545784e-05, "loss": 0.0013, "step": 6060 }, { "epoch": 0.64, "learning_rate": 1.6000000000000003e-05, "loss": 0.0014, "step": 6070 }, { "epoch": 0.64, "learning_rate": 1.599281867145422e-05, "loss": 0.0012, "step": 6080 }, { "epoch": 0.64, "learning_rate": 1.598563734290844e-05, "loss": 0.0013, "step": 6090 }, { "epoch": 0.65, "learning_rate": 1.597845601436266e-05, "loss": 0.0011, "step": 6100 }, { "epoch": 0.65, "learning_rate": 1.5971274685816878e-05, "loss": 0.0011, "step": 6110 }, { "epoch": 0.65, "learning_rate": 1.5964093357271097e-05, "loss": 0.0012, "step": 6120 }, { "epoch": 0.65, "learning_rate": 1.5956912028725316e-05, "loss": 0.0013, "step": 6130 }, { "epoch": 0.65, "learning_rate": 1.5949730700179534e-05, "loss": 0.0012, "step": 6140 }, { "epoch": 0.65, "learning_rate": 1.5942549371633753e-05, "loss": 0.0011, "step": 6150 }, { "epoch": 0.65, "learning_rate": 1.5935368043087972e-05, "loss": 0.0014, "step": 6160 }, { "epoch": 0.65, "learning_rate": 1.592818671454219e-05, "loss": 0.0014, "step": 6170 }, { "epoch": 0.65, "learning_rate": 1.592100538599641e-05, "loss": 0.0014, "step": 6180 }, { "epoch": 0.66, "learning_rate": 1.591382405745063e-05, "loss": 0.0012, "step": 6190 }, { "epoch": 0.66, "learning_rate": 1.5906642728904847e-05, "loss": 0.0014, "step": 6200 }, { "epoch": 0.66, "learning_rate": 1.589946140035907e-05, "loss": 0.0014, "step": 6210 }, { "epoch": 0.66, "learning_rate": 1.5892280071813285e-05, "loss": 0.0012, "step": 6220 }, { "epoch": 0.66, "learning_rate": 1.5885098743267507e-05, "loss": 0.0015, "step": 6230 }, { "epoch": 0.66, "learning_rate": 1.5877917414721723e-05, "loss": 0.0011, "step": 6240 }, { "epoch": 0.66, "learning_rate": 1.5870736086175945e-05, "loss": 0.0011, "step": 6250 }, { "epoch": 0.66, "learning_rate": 1.586355475763016e-05, "loss": 0.0013, "step": 6260 }, { "epoch": 0.66, "learning_rate": 1.5856373429084383e-05, "loss": 0.0012, "step": 6270 }, { "epoch": 0.66, "learning_rate": 1.58491921005386e-05, "loss": 0.0012, "step": 6280 }, { "epoch": 0.67, "learning_rate": 1.584201077199282e-05, "loss": 0.0014, "step": 6290 }, { "epoch": 0.67, "learning_rate": 1.583482944344704e-05, "loss": 0.0013, "step": 6300 }, { "epoch": 0.67, "learning_rate": 1.5827648114901258e-05, "loss": 0.0013, "step": 6310 }, { "epoch": 0.67, "learning_rate": 1.5820466786355477e-05, "loss": 0.0013, "step": 6320 }, { "epoch": 0.67, "learning_rate": 1.5813285457809696e-05, "loss": 0.0012, "step": 6330 }, { "epoch": 0.67, "learning_rate": 1.5806104129263915e-05, "loss": 0.001, "step": 6340 }, { "epoch": 0.67, "learning_rate": 1.5798922800718133e-05, "loss": 0.0014, "step": 6350 }, { "epoch": 0.67, "learning_rate": 1.5791741472172352e-05, "loss": 0.0012, "step": 6360 }, { "epoch": 0.67, "learning_rate": 1.5784560143626574e-05, "loss": 0.0013, "step": 6370 }, { "epoch": 0.68, "learning_rate": 1.577737881508079e-05, "loss": 0.0012, "step": 6380 }, { "epoch": 0.68, "learning_rate": 1.5770197486535012e-05, "loss": 0.0015, "step": 6390 }, { "epoch": 0.68, "learning_rate": 1.5763016157989227e-05, "loss": 0.0013, "step": 6400 }, { "epoch": 0.68, "learning_rate": 1.575583482944345e-05, "loss": 0.0013, "step": 6410 }, { "epoch": 0.68, "learning_rate": 1.574865350089767e-05, "loss": 0.0013, "step": 6420 }, { "epoch": 0.68, "learning_rate": 1.5741472172351887e-05, "loss": 0.0013, "step": 6430 }, { "epoch": 0.68, "learning_rate": 1.5734290843806106e-05, "loss": 0.0011, "step": 6440 }, { "epoch": 0.68, "learning_rate": 1.5727109515260325e-05, "loss": 0.0013, "step": 6450 }, { "epoch": 0.68, "learning_rate": 1.5719928186714544e-05, "loss": 0.0011, "step": 6460 }, { "epoch": 0.68, "learning_rate": 1.5712746858168763e-05, "loss": 0.0012, "step": 6470 }, { "epoch": 0.69, "learning_rate": 1.570556552962298e-05, "loss": 0.0012, "step": 6480 }, { "epoch": 0.69, "learning_rate": 1.56983842010772e-05, "loss": 0.0012, "step": 6490 }, { "epoch": 0.69, "learning_rate": 1.569120287253142e-05, "loss": 0.0014, "step": 6500 }, { "epoch": 0.69, "learning_rate": 1.5684021543985638e-05, "loss": 0.0013, "step": 6510 }, { "epoch": 0.69, "learning_rate": 1.5676840215439857e-05, "loss": 0.0012, "step": 6520 }, { "epoch": 0.69, "learning_rate": 1.5669658886894076e-05, "loss": 0.0011, "step": 6530 }, { "epoch": 0.69, "learning_rate": 1.5662477558348295e-05, "loss": 0.0012, "step": 6540 }, { "epoch": 0.69, "learning_rate": 1.5655296229802513e-05, "loss": 0.0012, "step": 6550 }, { "epoch": 0.69, "learning_rate": 1.5648114901256736e-05, "loss": 0.0012, "step": 6560 }, { "epoch": 0.7, "learning_rate": 1.564093357271095e-05, "loss": 0.001, "step": 6570 }, { "epoch": 0.7, "learning_rate": 1.5633752244165173e-05, "loss": 0.0012, "step": 6580 }, { "epoch": 0.7, "learning_rate": 1.562657091561939e-05, "loss": 0.0009, "step": 6590 }, { "epoch": 0.7, "learning_rate": 1.561938958707361e-05, "loss": 0.0013, "step": 6600 }, { "epoch": 0.7, "learning_rate": 1.5612208258527826e-05, "loss": 0.0011, "step": 6610 }, { "epoch": 0.7, "learning_rate": 1.560502692998205e-05, "loss": 0.0012, "step": 6620 }, { "epoch": 0.7, "learning_rate": 1.5597845601436267e-05, "loss": 0.0012, "step": 6630 }, { "epoch": 0.7, "learning_rate": 1.5590664272890486e-05, "loss": 0.0014, "step": 6640 }, { "epoch": 0.7, "learning_rate": 1.5583482944344705e-05, "loss": 0.0013, "step": 6650 }, { "epoch": 0.7, "learning_rate": 1.5576301615798924e-05, "loss": 0.0012, "step": 6660 }, { "epoch": 0.71, "learning_rate": 1.5569120287253143e-05, "loss": 0.0011, "step": 6670 }, { "epoch": 0.71, "learning_rate": 1.556193895870736e-05, "loss": 0.0014, "step": 6680 }, { "epoch": 0.71, "learning_rate": 1.555475763016158e-05, "loss": 0.0011, "step": 6690 }, { "epoch": 0.71, "learning_rate": 1.5547576301615803e-05, "loss": 0.0013, "step": 6700 }, { "epoch": 0.71, "learning_rate": 1.5540394973070018e-05, "loss": 0.0011, "step": 6710 }, { "epoch": 0.71, "learning_rate": 1.553321364452424e-05, "loss": 0.0013, "step": 6720 }, { "epoch": 0.71, "learning_rate": 1.5526032315978456e-05, "loss": 0.0013, "step": 6730 }, { "epoch": 0.71, "learning_rate": 1.5518850987432678e-05, "loss": 0.0012, "step": 6740 }, { "epoch": 0.71, "learning_rate": 1.5511669658886893e-05, "loss": 0.0013, "step": 6750 }, { "epoch": 0.72, "learning_rate": 1.5504488330341116e-05, "loss": 0.0012, "step": 6760 }, { "epoch": 0.72, "learning_rate": 1.5497307001795334e-05, "loss": 0.0011, "step": 6770 }, { "epoch": 0.72, "learning_rate": 1.5490125673249553e-05, "loss": 0.0013, "step": 6780 }, { "epoch": 0.72, "learning_rate": 1.5482944344703772e-05, "loss": 0.0013, "step": 6790 }, { "epoch": 0.72, "learning_rate": 1.547576301615799e-05, "loss": 0.0013, "step": 6800 }, { "epoch": 0.72, "learning_rate": 1.546858168761221e-05, "loss": 0.0013, "step": 6810 }, { "epoch": 0.72, "learning_rate": 1.546140035906643e-05, "loss": 0.0012, "step": 6820 }, { "epoch": 0.72, "learning_rate": 1.5454219030520647e-05, "loss": 0.0012, "step": 6830 }, { "epoch": 0.72, "learning_rate": 1.5447037701974866e-05, "loss": 0.0013, "step": 6840 }, { "epoch": 0.72, "learning_rate": 1.5439856373429085e-05, "loss": 0.0013, "step": 6850 }, { "epoch": 0.73, "learning_rate": 1.5432675044883304e-05, "loss": 0.001, "step": 6860 }, { "epoch": 0.73, "learning_rate": 1.5425493716337523e-05, "loss": 0.0011, "step": 6870 }, { "epoch": 0.73, "learning_rate": 1.541831238779174e-05, "loss": 0.0013, "step": 6880 }, { "epoch": 0.73, "learning_rate": 1.5411131059245964e-05, "loss": 0.0013, "step": 6890 }, { "epoch": 0.73, "learning_rate": 1.540394973070018e-05, "loss": 0.0011, "step": 6900 }, { "epoch": 0.73, "learning_rate": 1.53967684021544e-05, "loss": 0.0013, "step": 6910 }, { "epoch": 0.73, "learning_rate": 1.5389587073608617e-05, "loss": 0.0013, "step": 6920 }, { "epoch": 0.73, "learning_rate": 1.538240574506284e-05, "loss": 0.0013, "step": 6930 }, { "epoch": 0.73, "learning_rate": 1.5375224416517055e-05, "loss": 0.0012, "step": 6940 }, { "epoch": 0.74, "learning_rate": 1.5368043087971277e-05, "loss": 0.0012, "step": 6950 }, { "epoch": 0.74, "learning_rate": 1.5360861759425496e-05, "loss": 0.0012, "step": 6960 }, { "epoch": 0.74, "learning_rate": 1.5353680430879714e-05, "loss": 0.0014, "step": 6970 }, { "epoch": 0.74, "learning_rate": 1.5346499102333933e-05, "loss": 0.0012, "step": 6980 }, { "epoch": 0.74, "learning_rate": 1.5339317773788152e-05, "loss": 0.0013, "step": 6990 }, { "epoch": 0.74, "learning_rate": 1.533213644524237e-05, "loss": 0.0012, "step": 7000 }, { "epoch": 0.74, "learning_rate": 1.532495511669659e-05, "loss": 0.0016, "step": 7010 }, { "epoch": 0.74, "learning_rate": 1.531777378815081e-05, "loss": 0.0013, "step": 7020 }, { "epoch": 0.74, "learning_rate": 1.531059245960503e-05, "loss": 0.0014, "step": 7030 }, { "epoch": 0.74, "learning_rate": 1.5303411131059246e-05, "loss": 0.0011, "step": 7040 }, { "epoch": 0.75, "learning_rate": 1.529622980251347e-05, "loss": 0.0013, "step": 7050 }, { "epoch": 0.75, "learning_rate": 1.5289048473967684e-05, "loss": 0.0013, "step": 7060 }, { "epoch": 0.75, "learning_rate": 1.5281867145421906e-05, "loss": 0.0012, "step": 7070 }, { "epoch": 0.75, "learning_rate": 1.527468581687612e-05, "loss": 0.0011, "step": 7080 }, { "epoch": 0.75, "learning_rate": 1.5267504488330344e-05, "loss": 0.0013, "step": 7090 }, { "epoch": 0.75, "learning_rate": 1.5260323159784563e-05, "loss": 0.0013, "step": 7100 }, { "epoch": 0.75, "learning_rate": 1.525314183123878e-05, "loss": 0.0011, "step": 7110 }, { "epoch": 0.75, "learning_rate": 1.5245960502693e-05, "loss": 0.0012, "step": 7120 }, { "epoch": 0.75, "learning_rate": 1.5238779174147217e-05, "loss": 0.0013, "step": 7130 }, { "epoch": 0.76, "learning_rate": 1.5231597845601438e-05, "loss": 0.0011, "step": 7140 }, { "epoch": 0.76, "learning_rate": 1.5224416517055655e-05, "loss": 0.0011, "step": 7150 }, { "epoch": 0.76, "learning_rate": 1.5217235188509876e-05, "loss": 0.0011, "step": 7160 }, { "epoch": 0.76, "learning_rate": 1.5210053859964096e-05, "loss": 0.0013, "step": 7170 }, { "epoch": 0.76, "learning_rate": 1.5202872531418313e-05, "loss": 0.0013, "step": 7180 }, { "epoch": 0.76, "learning_rate": 1.5195691202872534e-05, "loss": 0.0011, "step": 7190 }, { "epoch": 0.76, "learning_rate": 1.5188509874326751e-05, "loss": 0.0014, "step": 7200 }, { "epoch": 0.76, "learning_rate": 1.5181328545780971e-05, "loss": 0.001, "step": 7210 }, { "epoch": 0.76, "learning_rate": 1.5174147217235189e-05, "loss": 0.0013, "step": 7220 }, { "epoch": 0.77, "learning_rate": 1.516696588868941e-05, "loss": 0.0013, "step": 7230 }, { "epoch": 0.77, "learning_rate": 1.515978456014363e-05, "loss": 0.0012, "step": 7240 }, { "epoch": 0.77, "learning_rate": 1.5152603231597847e-05, "loss": 0.0014, "step": 7250 }, { "epoch": 0.77, "learning_rate": 1.5145421903052067e-05, "loss": 0.0012, "step": 7260 }, { "epoch": 0.77, "learning_rate": 1.5138240574506284e-05, "loss": 0.001, "step": 7270 }, { "epoch": 0.77, "learning_rate": 1.5131059245960505e-05, "loss": 0.0012, "step": 7280 }, { "epoch": 0.77, "learning_rate": 1.5123877917414722e-05, "loss": 0.0012, "step": 7290 }, { "epoch": 0.77, "learning_rate": 1.5116696588868943e-05, "loss": 0.0013, "step": 7300 }, { "epoch": 0.77, "learning_rate": 1.5109515260323162e-05, "loss": 0.0014, "step": 7310 }, { "epoch": 0.77, "learning_rate": 1.510233393177738e-05, "loss": 0.0015, "step": 7320 }, { "epoch": 0.78, "learning_rate": 1.50951526032316e-05, "loss": 0.0012, "step": 7330 }, { "epoch": 0.78, "learning_rate": 1.5087971274685818e-05, "loss": 0.0011, "step": 7340 }, { "epoch": 0.78, "learning_rate": 1.5080789946140037e-05, "loss": 0.0016, "step": 7350 }, { "epoch": 0.78, "learning_rate": 1.5073608617594256e-05, "loss": 0.0014, "step": 7360 }, { "epoch": 0.78, "learning_rate": 1.5066427289048474e-05, "loss": 0.0013, "step": 7370 }, { "epoch": 0.78, "learning_rate": 1.5059245960502695e-05, "loss": 0.0013, "step": 7380 }, { "epoch": 0.78, "learning_rate": 1.5052064631956914e-05, "loss": 0.0011, "step": 7390 }, { "epoch": 0.78, "learning_rate": 1.5044883303411133e-05, "loss": 0.0011, "step": 7400 }, { "epoch": 0.78, "learning_rate": 1.5037701974865352e-05, "loss": 0.0011, "step": 7410 }, { "epoch": 0.79, "learning_rate": 1.503052064631957e-05, "loss": 0.0013, "step": 7420 }, { "epoch": 0.79, "learning_rate": 1.502333931777379e-05, "loss": 0.0012, "step": 7430 }, { "epoch": 0.79, "learning_rate": 1.5016157989228008e-05, "loss": 0.0014, "step": 7440 }, { "epoch": 0.79, "learning_rate": 1.5008976660682229e-05, "loss": 0.0012, "step": 7450 }, { "epoch": 0.79, "learning_rate": 1.5001795332136446e-05, "loss": 0.0013, "step": 7460 }, { "epoch": 0.79, "learning_rate": 1.4994614003590666e-05, "loss": 0.0013, "step": 7470 }, { "epoch": 0.79, "learning_rate": 1.4987432675044883e-05, "loss": 0.0015, "step": 7480 }, { "epoch": 0.79, "learning_rate": 1.4980251346499104e-05, "loss": 0.0013, "step": 7490 }, { "epoch": 0.79, "learning_rate": 1.4973070017953321e-05, "loss": 0.0014, "step": 7500 }, { "epoch": 0.79, "learning_rate": 1.4965888689407542e-05, "loss": 0.0014, "step": 7510 }, { "epoch": 0.8, "learning_rate": 1.4958707360861762e-05, "loss": 0.0013, "step": 7520 }, { "epoch": 0.8, "learning_rate": 1.495152603231598e-05, "loss": 0.0014, "step": 7530 }, { "epoch": 0.8, "learning_rate": 1.49443447037702e-05, "loss": 0.0012, "step": 7540 }, { "epoch": 0.8, "learning_rate": 1.4937163375224417e-05, "loss": 0.0013, "step": 7550 }, { "epoch": 0.8, "learning_rate": 1.4929982046678637e-05, "loss": 0.0013, "step": 7560 }, { "epoch": 0.8, "learning_rate": 1.4922800718132855e-05, "loss": 0.0011, "step": 7570 }, { "epoch": 0.8, "learning_rate": 1.4915619389587075e-05, "loss": 0.0015, "step": 7580 }, { "epoch": 0.8, "learning_rate": 1.4908438061041296e-05, "loss": 0.0011, "step": 7590 }, { "epoch": 0.8, "learning_rate": 1.4901256732495513e-05, "loss": 0.0011, "step": 7600 }, { "epoch": 0.81, "learning_rate": 1.4894075403949733e-05, "loss": 0.0011, "step": 7610 }, { "epoch": 0.81, "learning_rate": 1.488689407540395e-05, "loss": 0.0012, "step": 7620 }, { "epoch": 0.81, "learning_rate": 1.4879712746858171e-05, "loss": 0.0014, "step": 7630 }, { "epoch": 0.81, "learning_rate": 1.4872531418312388e-05, "loss": 0.0012, "step": 7640 }, { "epoch": 0.81, "learning_rate": 1.4865350089766609e-05, "loss": 0.0014, "step": 7650 }, { "epoch": 0.81, "learning_rate": 1.4858168761220827e-05, "loss": 0.0015, "step": 7660 }, { "epoch": 0.81, "learning_rate": 1.4850987432675046e-05, "loss": 0.0012, "step": 7670 }, { "epoch": 0.81, "learning_rate": 1.4843806104129265e-05, "loss": 0.0015, "step": 7680 }, { "epoch": 0.81, "learning_rate": 1.4836624775583484e-05, "loss": 0.0013, "step": 7690 }, { "epoch": 0.81, "learning_rate": 1.4829443447037703e-05, "loss": 0.0011, "step": 7700 }, { "epoch": 0.82, "learning_rate": 1.4822262118491922e-05, "loss": 0.0011, "step": 7710 }, { "epoch": 0.82, "learning_rate": 1.4815080789946142e-05, "loss": 0.0012, "step": 7720 }, { "epoch": 0.82, "learning_rate": 1.4807899461400361e-05, "loss": 0.0013, "step": 7730 }, { "epoch": 0.82, "learning_rate": 1.480071813285458e-05, "loss": 0.0012, "step": 7740 }, { "epoch": 0.82, "learning_rate": 1.4793536804308799e-05, "loss": 0.0013, "step": 7750 }, { "epoch": 0.82, "learning_rate": 1.4786355475763017e-05, "loss": 0.0011, "step": 7760 }, { "epoch": 0.82, "learning_rate": 1.4779174147217236e-05, "loss": 0.0009, "step": 7770 }, { "epoch": 0.82, "learning_rate": 1.4771992818671455e-05, "loss": 0.0013, "step": 7780 }, { "epoch": 0.82, "learning_rate": 1.4764811490125674e-05, "loss": 0.0011, "step": 7790 }, { "epoch": 0.83, "learning_rate": 1.4757630161579894e-05, "loss": 0.0016, "step": 7800 }, { "epoch": 0.83, "learning_rate": 1.4750448833034112e-05, "loss": 0.0015, "step": 7810 }, { "epoch": 0.83, "learning_rate": 1.4743267504488332e-05, "loss": 0.0013, "step": 7820 }, { "epoch": 0.83, "learning_rate": 1.473608617594255e-05, "loss": 0.0012, "step": 7830 }, { "epoch": 0.83, "learning_rate": 1.472890484739677e-05, "loss": 0.0014, "step": 7840 }, { "epoch": 0.83, "learning_rate": 1.4721723518850987e-05, "loss": 0.0011, "step": 7850 }, { "epoch": 0.83, "learning_rate": 1.4714542190305207e-05, "loss": 0.0013, "step": 7860 }, { "epoch": 0.83, "learning_rate": 1.4707360861759428e-05, "loss": 0.0013, "step": 7870 }, { "epoch": 0.83, "learning_rate": 1.4700179533213645e-05, "loss": 0.0012, "step": 7880 }, { "epoch": 0.83, "learning_rate": 1.4692998204667866e-05, "loss": 0.0013, "step": 7890 }, { "epoch": 0.84, "learning_rate": 1.4685816876122083e-05, "loss": 0.0014, "step": 7900 }, { "epoch": 0.84, "learning_rate": 1.4678635547576303e-05, "loss": 0.0011, "step": 7910 }, { "epoch": 0.84, "learning_rate": 1.467145421903052e-05, "loss": 0.0013, "step": 7920 }, { "epoch": 0.84, "learning_rate": 1.4664272890484741e-05, "loss": 0.0012, "step": 7930 }, { "epoch": 0.84, "learning_rate": 1.4657091561938961e-05, "loss": 0.0013, "step": 7940 }, { "epoch": 0.84, "learning_rate": 1.4649910233393179e-05, "loss": 0.0012, "step": 7950 }, { "epoch": 0.84, "learning_rate": 1.4642728904847399e-05, "loss": 0.0013, "step": 7960 }, { "epoch": 0.84, "learning_rate": 1.4635547576301616e-05, "loss": 0.0012, "step": 7970 }, { "epoch": 0.84, "learning_rate": 1.4628366247755837e-05, "loss": 0.0013, "step": 7980 }, { "epoch": 0.85, "learning_rate": 1.4621184919210054e-05, "loss": 0.0011, "step": 7990 }, { "epoch": 0.85, "learning_rate": 1.4614003590664274e-05, "loss": 0.0012, "step": 8000 }, { "epoch": 0.85, "learning_rate": 1.4606822262118493e-05, "loss": 0.0012, "step": 8010 }, { "epoch": 0.85, "learning_rate": 1.4599640933572712e-05, "loss": 0.0013, "step": 8020 }, { "epoch": 0.85, "learning_rate": 1.4592459605026933e-05, "loss": 0.0009, "step": 8030 }, { "epoch": 0.85, "learning_rate": 1.458527827648115e-05, "loss": 0.0012, "step": 8040 }, { "epoch": 0.85, "learning_rate": 1.457809694793537e-05, "loss": 0.0012, "step": 8050 }, { "epoch": 0.85, "learning_rate": 1.4570915619389587e-05, "loss": 0.0013, "step": 8060 }, { "epoch": 0.85, "learning_rate": 1.4563734290843808e-05, "loss": 0.0013, "step": 8070 }, { "epoch": 0.86, "learning_rate": 1.4556552962298027e-05, "loss": 0.0015, "step": 8080 }, { "epoch": 0.86, "learning_rate": 1.4549371633752246e-05, "loss": 0.0012, "step": 8090 }, { "epoch": 0.86, "learning_rate": 1.4542190305206464e-05, "loss": 0.0012, "step": 8100 }, { "epoch": 0.86, "learning_rate": 1.4535008976660683e-05, "loss": 0.0013, "step": 8110 }, { "epoch": 0.86, "learning_rate": 1.4527827648114902e-05, "loss": 0.0011, "step": 8120 }, { "epoch": 0.86, "learning_rate": 1.4520646319569121e-05, "loss": 0.0009, "step": 8130 }, { "epoch": 0.86, "learning_rate": 1.451346499102334e-05, "loss": 0.0012, "step": 8140 }, { "epoch": 0.86, "learning_rate": 1.450628366247756e-05, "loss": 0.0012, "step": 8150 }, { "epoch": 0.86, "learning_rate": 1.4499102333931777e-05, "loss": 0.0013, "step": 8160 }, { "epoch": 0.86, "learning_rate": 1.4491921005385998e-05, "loss": 0.0014, "step": 8170 }, { "epoch": 0.87, "learning_rate": 1.4484739676840217e-05, "loss": 0.0012, "step": 8180 }, { "epoch": 0.87, "learning_rate": 1.4477558348294436e-05, "loss": 0.0013, "step": 8190 }, { "epoch": 0.87, "learning_rate": 1.4470377019748654e-05, "loss": 0.0011, "step": 8200 }, { "epoch": 0.87, "learning_rate": 1.4463195691202873e-05, "loss": 0.0013, "step": 8210 }, { "epoch": 0.87, "learning_rate": 1.4456014362657094e-05, "loss": 0.0012, "step": 8220 }, { "epoch": 0.87, "learning_rate": 1.4448833034111311e-05, "loss": 0.0013, "step": 8230 }, { "epoch": 0.87, "learning_rate": 1.4441651705565531e-05, "loss": 0.0015, "step": 8240 }, { "epoch": 0.87, "learning_rate": 1.4434470377019749e-05, "loss": 0.0012, "step": 8250 }, { "epoch": 0.87, "learning_rate": 1.4427289048473969e-05, "loss": 0.0011, "step": 8260 }, { "epoch": 0.88, "learning_rate": 1.4420107719928186e-05, "loss": 0.0012, "step": 8270 }, { "epoch": 0.88, "learning_rate": 1.4412926391382407e-05, "loss": 0.0012, "step": 8280 }, { "epoch": 0.88, "learning_rate": 1.4405745062836627e-05, "loss": 0.0013, "step": 8290 }, { "epoch": 0.88, "learning_rate": 1.4398563734290844e-05, "loss": 0.0013, "step": 8300 }, { "epoch": 0.88, "learning_rate": 1.4391382405745065e-05, "loss": 0.0012, "step": 8310 }, { "epoch": 0.88, "learning_rate": 1.4384201077199282e-05, "loss": 0.0015, "step": 8320 }, { "epoch": 0.88, "learning_rate": 1.4377019748653503e-05, "loss": 0.0012, "step": 8330 }, { "epoch": 0.88, "learning_rate": 1.436983842010772e-05, "loss": 0.0015, "step": 8340 }, { "epoch": 0.88, "learning_rate": 1.436265709156194e-05, "loss": 0.0011, "step": 8350 }, { "epoch": 0.88, "learning_rate": 1.4355475763016161e-05, "loss": 0.0013, "step": 8360 }, { "epoch": 0.89, "learning_rate": 1.4348294434470378e-05, "loss": 0.0013, "step": 8370 }, { "epoch": 0.89, "learning_rate": 1.4341113105924599e-05, "loss": 0.0012, "step": 8380 }, { "epoch": 0.89, "learning_rate": 1.4333931777378816e-05, "loss": 0.0013, "step": 8390 }, { "epoch": 0.89, "learning_rate": 1.4326750448833036e-05, "loss": 0.0013, "step": 8400 }, { "epoch": 0.89, "learning_rate": 1.4319569120287253e-05, "loss": 0.001, "step": 8410 }, { "epoch": 0.89, "learning_rate": 1.4312387791741474e-05, "loss": 0.0012, "step": 8420 }, { "epoch": 0.89, "learning_rate": 1.4305206463195693e-05, "loss": 0.0015, "step": 8430 }, { "epoch": 0.89, "learning_rate": 1.4298025134649912e-05, "loss": 0.0012, "step": 8440 }, { "epoch": 0.89, "learning_rate": 1.429084380610413e-05, "loss": 0.0012, "step": 8450 }, { "epoch": 0.9, "learning_rate": 1.428366247755835e-05, "loss": 0.0013, "step": 8460 }, { "epoch": 0.9, "learning_rate": 1.4276481149012568e-05, "loss": 0.0011, "step": 8470 }, { "epoch": 0.9, "learning_rate": 1.4269299820466787e-05, "loss": 0.001, "step": 8480 }, { "epoch": 0.9, "learning_rate": 1.4262118491921006e-05, "loss": 0.0013, "step": 8490 }, { "epoch": 0.9, "learning_rate": 1.4254937163375226e-05, "loss": 0.001, "step": 8500 }, { "epoch": 0.9, "learning_rate": 1.4247755834829445e-05, "loss": 0.0013, "step": 8510 }, { "epoch": 0.9, "learning_rate": 1.4240574506283664e-05, "loss": 0.0011, "step": 8520 }, { "epoch": 0.9, "learning_rate": 1.4233393177737883e-05, "loss": 0.0014, "step": 8530 }, { "epoch": 0.9, "learning_rate": 1.4226211849192102e-05, "loss": 0.001, "step": 8540 }, { "epoch": 0.9, "learning_rate": 1.421903052064632e-05, "loss": 0.0013, "step": 8550 }, { "epoch": 0.91, "learning_rate": 1.421184919210054e-05, "loss": 0.0011, "step": 8560 }, { "epoch": 0.91, "learning_rate": 1.420466786355476e-05, "loss": 0.0012, "step": 8570 }, { "epoch": 0.91, "learning_rate": 1.4197486535008977e-05, "loss": 0.0012, "step": 8580 }, { "epoch": 0.91, "learning_rate": 1.4190305206463197e-05, "loss": 0.0012, "step": 8590 }, { "epoch": 0.91, "learning_rate": 1.4183123877917415e-05, "loss": 0.0014, "step": 8600 }, { "epoch": 0.91, "learning_rate": 1.4175942549371635e-05, "loss": 0.0012, "step": 8610 }, { "epoch": 0.91, "learning_rate": 1.4168761220825852e-05, "loss": 0.0012, "step": 8620 }, { "epoch": 0.91, "learning_rate": 1.4161579892280073e-05, "loss": 0.0015, "step": 8630 }, { "epoch": 0.91, "learning_rate": 1.4154398563734293e-05, "loss": 0.0013, "step": 8640 }, { "epoch": 0.92, "learning_rate": 1.414721723518851e-05, "loss": 0.0011, "step": 8650 }, { "epoch": 0.92, "learning_rate": 1.4140035906642731e-05, "loss": 0.0012, "step": 8660 }, { "epoch": 0.92, "learning_rate": 1.4132854578096948e-05, "loss": 0.0012, "step": 8670 }, { "epoch": 0.92, "learning_rate": 1.4125673249551169e-05, "loss": 0.0011, "step": 8680 }, { "epoch": 0.92, "learning_rate": 1.4118491921005386e-05, "loss": 0.0013, "step": 8690 }, { "epoch": 0.92, "learning_rate": 1.4111310592459606e-05, "loss": 0.0011, "step": 8700 }, { "epoch": 0.92, "learning_rate": 1.4104129263913827e-05, "loss": 0.0013, "step": 8710 }, { "epoch": 0.92, "learning_rate": 1.4096947935368044e-05, "loss": 0.0013, "step": 8720 }, { "epoch": 0.92, "learning_rate": 1.4089766606822264e-05, "loss": 0.0011, "step": 8730 }, { "epoch": 0.92, "learning_rate": 1.4082585278276482e-05, "loss": 0.0013, "step": 8740 }, { "epoch": 0.93, "learning_rate": 1.4075403949730702e-05, "loss": 0.0011, "step": 8750 }, { "epoch": 0.93, "learning_rate": 1.406822262118492e-05, "loss": 0.0012, "step": 8760 }, { "epoch": 0.93, "learning_rate": 1.406104129263914e-05, "loss": 0.0012, "step": 8770 }, { "epoch": 0.93, "learning_rate": 1.4053859964093359e-05, "loss": 0.0011, "step": 8780 }, { "epoch": 0.93, "learning_rate": 1.4046678635547577e-05, "loss": 0.0014, "step": 8790 }, { "epoch": 0.93, "learning_rate": 1.4039497307001796e-05, "loss": 0.0011, "step": 8800 }, { "epoch": 0.93, "learning_rate": 1.4032315978456015e-05, "loss": 0.0011, "step": 8810 }, { "epoch": 0.93, "learning_rate": 1.4025134649910236e-05, "loss": 0.0013, "step": 8820 }, { "epoch": 0.93, "learning_rate": 1.4017953321364453e-05, "loss": 0.001, "step": 8830 }, { "epoch": 0.94, "learning_rate": 1.4010771992818673e-05, "loss": 0.0013, "step": 8840 }, { "epoch": 0.94, "learning_rate": 1.4003590664272892e-05, "loss": 0.0013, "step": 8850 }, { "epoch": 0.94, "learning_rate": 1.3996409335727111e-05, "loss": 0.0013, "step": 8860 }, { "epoch": 0.94, "learning_rate": 1.398922800718133e-05, "loss": 0.0012, "step": 8870 }, { "epoch": 0.94, "learning_rate": 1.3982046678635549e-05, "loss": 0.0013, "step": 8880 }, { "epoch": 0.94, "learning_rate": 1.3974865350089767e-05, "loss": 0.0015, "step": 8890 }, { "epoch": 0.94, "learning_rate": 1.3967684021543986e-05, "loss": 0.0011, "step": 8900 }, { "epoch": 0.94, "learning_rate": 1.3960502692998205e-05, "loss": 0.001, "step": 8910 }, { "epoch": 0.94, "learning_rate": 1.3953321364452426e-05, "loss": 0.001, "step": 8920 }, { "epoch": 0.94, "learning_rate": 1.3946140035906643e-05, "loss": 0.0012, "step": 8930 }, { "epoch": 0.95, "learning_rate": 1.3938958707360863e-05, "loss": 0.0012, "step": 8940 }, { "epoch": 0.95, "learning_rate": 1.393177737881508e-05, "loss": 0.001, "step": 8950 }, { "epoch": 0.95, "learning_rate": 1.3924596050269301e-05, "loss": 0.0012, "step": 8960 }, { "epoch": 0.95, "learning_rate": 1.391741472172352e-05, "loss": 0.0014, "step": 8970 }, { "epoch": 0.95, "learning_rate": 1.3910233393177739e-05, "loss": 0.0014, "step": 8980 }, { "epoch": 0.95, "learning_rate": 1.3903052064631959e-05, "loss": 0.0011, "step": 8990 }, { "epoch": 0.95, "learning_rate": 1.3895870736086176e-05, "loss": 0.0014, "step": 9000 }, { "epoch": 0.95, "learning_rate": 1.3888689407540397e-05, "loss": 0.0013, "step": 9010 }, { "epoch": 0.95, "learning_rate": 1.3881508078994614e-05, "loss": 0.0013, "step": 9020 }, { "epoch": 0.96, "learning_rate": 1.3874326750448834e-05, "loss": 0.0015, "step": 9030 }, { "epoch": 0.96, "learning_rate": 1.3867145421903052e-05, "loss": 0.001, "step": 9040 }, { "epoch": 0.96, "learning_rate": 1.3859964093357272e-05, "loss": 0.001, "step": 9050 }, { "epoch": 0.96, "learning_rate": 1.3852782764811493e-05, "loss": 0.0011, "step": 9060 }, { "epoch": 0.96, "learning_rate": 1.384560143626571e-05, "loss": 0.0012, "step": 9070 }, { "epoch": 0.96, "learning_rate": 1.383842010771993e-05, "loss": 0.001, "step": 9080 }, { "epoch": 0.96, "learning_rate": 1.3831238779174147e-05, "loss": 0.0012, "step": 9090 }, { "epoch": 0.96, "learning_rate": 1.3824057450628368e-05, "loss": 0.0012, "step": 9100 }, { "epoch": 0.96, "learning_rate": 1.3816876122082585e-05, "loss": 0.0012, "step": 9110 }, { "epoch": 0.97, "learning_rate": 1.3809694793536806e-05, "loss": 0.0015, "step": 9120 }, { "epoch": 0.97, "learning_rate": 1.3802513464991026e-05, "loss": 0.0011, "step": 9130 }, { "epoch": 0.97, "learning_rate": 1.3795332136445243e-05, "loss": 0.0011, "step": 9140 }, { "epoch": 0.97, "learning_rate": 1.3788150807899464e-05, "loss": 0.0012, "step": 9150 }, { "epoch": 0.97, "learning_rate": 1.3780969479353681e-05, "loss": 0.0012, "step": 9160 }, { "epoch": 0.97, "learning_rate": 1.3773788150807901e-05, "loss": 0.0012, "step": 9170 }, { "epoch": 0.97, "learning_rate": 1.3766606822262119e-05, "loss": 0.0013, "step": 9180 }, { "epoch": 0.97, "learning_rate": 1.3759425493716339e-05, "loss": 0.0013, "step": 9190 }, { "epoch": 0.97, "learning_rate": 1.3752244165170558e-05, "loss": 0.0013, "step": 9200 }, { "epoch": 0.97, "learning_rate": 1.3745062836624777e-05, "loss": 0.0011, "step": 9210 }, { "epoch": 0.98, "learning_rate": 1.3737881508078996e-05, "loss": 0.01, "step": 9220 }, { "epoch": 0.98, "learning_rate": 1.3731418312387792e-05, "loss": 0.0657, "step": 9230 }, { "epoch": 0.98, "learning_rate": 1.3724236983842012e-05, "loss": 0.0811, "step": 9240 }, { "epoch": 0.98, "learning_rate": 1.3717055655296231e-05, "loss": 0.0016, "step": 9250 }, { "epoch": 0.98, "learning_rate": 1.370987432675045e-05, "loss": 0.0014, "step": 9260 }, { "epoch": 0.98, "learning_rate": 1.3702692998204669e-05, "loss": 0.0016, "step": 9270 }, { "epoch": 0.98, "learning_rate": 1.3695511669658888e-05, "loss": 0.0015, "step": 9280 }, { "epoch": 0.98, "learning_rate": 1.3688330341113108e-05, "loss": 0.0013, "step": 9290 }, { "epoch": 0.98, "learning_rate": 1.3681149012567325e-05, "loss": 0.0014, "step": 9300 }, { "epoch": 0.99, "learning_rate": 1.3673967684021546e-05, "loss": 0.0013, "step": 9310 }, { "epoch": 0.99, "learning_rate": 1.3666786355475765e-05, "loss": 0.0012, "step": 9320 }, { "epoch": 0.99, "learning_rate": 1.3659605026929984e-05, "loss": 0.0014, "step": 9330 }, { "epoch": 0.99, "learning_rate": 1.3652423698384202e-05, "loss": 0.0012, "step": 9340 }, { "epoch": 0.99, "learning_rate": 1.3645242369838421e-05, "loss": 0.0013, "step": 9350 }, { "epoch": 0.99, "learning_rate": 1.363806104129264e-05, "loss": 0.0011, "step": 9360 }, { "epoch": 0.99, "learning_rate": 1.3630879712746859e-05, "loss": 0.0011, "step": 9370 }, { "epoch": 0.99, "learning_rate": 1.3623698384201078e-05, "loss": 0.0013, "step": 9380 }, { "epoch": 0.99, "learning_rate": 1.3616517055655298e-05, "loss": 0.0011, "step": 9390 }, { "epoch": 0.99, "learning_rate": 1.3609335727109515e-05, "loss": 0.0012, "step": 9400 }, { "epoch": 1.0, "learning_rate": 1.3602154398563736e-05, "loss": 0.0011, "step": 9410 }, { "epoch": 1.0, "learning_rate": 1.3594973070017953e-05, "loss": 0.001, "step": 9420 }, { "epoch": 1.0, "learning_rate": 1.3587791741472174e-05, "loss": 0.0011, "step": 9430 }, { "epoch": 1.0, "learning_rate": 1.3580610412926392e-05, "loss": 0.0014, "step": 9440 }, { "epoch": 1.0, "learning_rate": 1.3573429084380611e-05, "loss": 0.0009, "step": 9450 }, { "epoch": 1.0, "eval_loss": 0.001209951937198639, "eval_runtime": 294.8047, "eval_samples_per_second": 170.961, "eval_steps_per_second": 21.37, "step": 9450 }, { "epoch": 1.0, "learning_rate": 1.3566247755834832e-05, "loss": 0.0009, "step": 9460 }, { "epoch": 1.0, "learning_rate": 1.3559066427289049e-05, "loss": 0.0011, "step": 9470 }, { "epoch": 1.0, "learning_rate": 1.355188509874327e-05, "loss": 0.0011, "step": 9480 }, { "epoch": 1.0, "learning_rate": 1.3544703770197487e-05, "loss": 0.001, "step": 9490 }, { "epoch": 1.01, "learning_rate": 1.3537522441651707e-05, "loss": 0.001, "step": 9500 }, { "epoch": 1.01, "learning_rate": 1.3530341113105924e-05, "loss": 0.0011, "step": 9510 }, { "epoch": 1.01, "learning_rate": 1.3523159784560145e-05, "loss": 0.0014, "step": 9520 }, { "epoch": 1.01, "learning_rate": 1.3515978456014365e-05, "loss": 0.0012, "step": 9530 }, { "epoch": 1.01, "learning_rate": 1.3508797127468582e-05, "loss": 0.0014, "step": 9540 }, { "epoch": 1.01, "learning_rate": 1.3501615798922803e-05, "loss": 0.0011, "step": 9550 }, { "epoch": 1.01, "learning_rate": 1.349443447037702e-05, "loss": 0.0011, "step": 9560 }, { "epoch": 1.01, "learning_rate": 1.348725314183124e-05, "loss": 0.0012, "step": 9570 }, { "epoch": 1.01, "learning_rate": 1.3480071813285458e-05, "loss": 0.0011, "step": 9580 }, { "epoch": 1.01, "learning_rate": 1.3472890484739678e-05, "loss": 0.0012, "step": 9590 }, { "epoch": 1.02, "learning_rate": 1.3465709156193897e-05, "loss": 0.0013, "step": 9600 }, { "epoch": 1.02, "learning_rate": 1.3458527827648116e-05, "loss": 0.0011, "step": 9610 }, { "epoch": 1.02, "learning_rate": 1.3451346499102336e-05, "loss": 0.0012, "step": 9620 }, { "epoch": 1.02, "learning_rate": 1.3444165170556554e-05, "loss": 0.0015, "step": 9630 }, { "epoch": 1.02, "learning_rate": 1.3436983842010774e-05, "loss": 0.0011, "step": 9640 }, { "epoch": 1.02, "learning_rate": 1.3429802513464991e-05, "loss": 0.0012, "step": 9650 }, { "epoch": 1.02, "learning_rate": 1.3422621184919212e-05, "loss": 0.0012, "step": 9660 }, { "epoch": 1.02, "learning_rate": 1.341543985637343e-05, "loss": 0.0012, "step": 9670 }, { "epoch": 1.02, "learning_rate": 1.340825852782765e-05, "loss": 0.0014, "step": 9680 }, { "epoch": 1.03, "learning_rate": 1.3401077199281868e-05, "loss": 0.0012, "step": 9690 }, { "epoch": 1.03, "learning_rate": 1.3393895870736087e-05, "loss": 0.0012, "step": 9700 }, { "epoch": 1.03, "learning_rate": 1.3386714542190306e-05, "loss": 0.0011, "step": 9710 }, { "epoch": 1.03, "learning_rate": 1.3379533213644525e-05, "loss": 0.0014, "step": 9720 }, { "epoch": 1.03, "learning_rate": 1.3372351885098744e-05, "loss": 0.0014, "step": 9730 }, { "epoch": 1.03, "learning_rate": 1.3365170556552964e-05, "loss": 0.0012, "step": 9740 }, { "epoch": 1.03, "learning_rate": 1.3357989228007181e-05, "loss": 0.0011, "step": 9750 }, { "epoch": 1.03, "learning_rate": 1.3350807899461402e-05, "loss": 0.0011, "step": 9760 }, { "epoch": 1.03, "learning_rate": 1.334362657091562e-05, "loss": 0.0012, "step": 9770 }, { "epoch": 1.03, "learning_rate": 1.333644524236984e-05, "loss": 0.0011, "step": 9780 }, { "epoch": 1.04, "learning_rate": 1.3329263913824058e-05, "loss": 0.0013, "step": 9790 }, { "epoch": 1.04, "learning_rate": 1.3322082585278277e-05, "loss": 0.0013, "step": 9800 }, { "epoch": 1.04, "learning_rate": 1.3314901256732498e-05, "loss": 0.0012, "step": 9810 }, { "epoch": 1.04, "learning_rate": 1.3307719928186715e-05, "loss": 0.001, "step": 9820 }, { "epoch": 1.04, "learning_rate": 1.3300538599640935e-05, "loss": 0.0014, "step": 9830 }, { "epoch": 1.04, "learning_rate": 1.3293357271095152e-05, "loss": 0.0011, "step": 9840 }, { "epoch": 1.04, "learning_rate": 1.3286175942549373e-05, "loss": 0.0015, "step": 9850 }, { "epoch": 1.04, "learning_rate": 1.327899461400359e-05, "loss": 0.0013, "step": 9860 }, { "epoch": 1.04, "learning_rate": 1.327181328545781e-05, "loss": 0.0013, "step": 9870 }, { "epoch": 1.05, "learning_rate": 1.3264631956912031e-05, "loss": 0.0012, "step": 9880 }, { "epoch": 1.05, "learning_rate": 1.3257450628366248e-05, "loss": 0.0013, "step": 9890 }, { "epoch": 1.05, "learning_rate": 1.3250269299820469e-05, "loss": 0.0013, "step": 9900 }, { "epoch": 1.05, "learning_rate": 1.3243087971274686e-05, "loss": 0.0011, "step": 9910 }, { "epoch": 1.05, "learning_rate": 1.3235906642728907e-05, "loss": 0.0012, "step": 9920 }, { "epoch": 1.05, "learning_rate": 1.3228725314183124e-05, "loss": 0.0014, "step": 9930 }, { "epoch": 1.05, "learning_rate": 1.3221543985637344e-05, "loss": 0.0011, "step": 9940 }, { "epoch": 1.05, "learning_rate": 1.3214362657091565e-05, "loss": 0.0012, "step": 9950 }, { "epoch": 1.05, "learning_rate": 1.3207181328545782e-05, "loss": 0.0015, "step": 9960 }, { "epoch": 1.06, "learning_rate": 1.3200000000000002e-05, "loss": 0.0009, "step": 9970 }, { "epoch": 1.06, "learning_rate": 1.319281867145422e-05, "loss": 0.0013, "step": 9980 }, { "epoch": 1.06, "learning_rate": 1.318563734290844e-05, "loss": 0.0013, "step": 9990 }, { "epoch": 1.06, "learning_rate": 1.3178456014362657e-05, "loss": 0.0013, "step": 10000 }, { "epoch": 1.06, "learning_rate": 1.3171274685816878e-05, "loss": 0.0014, "step": 10010 }, { "epoch": 1.06, "learning_rate": 1.3164093357271097e-05, "loss": 0.0011, "step": 10020 }, { "epoch": 1.06, "learning_rate": 1.3156912028725315e-05, "loss": 0.0013, "step": 10030 }, { "epoch": 1.06, "learning_rate": 1.3149730700179534e-05, "loss": 0.0014, "step": 10040 }, { "epoch": 1.06, "learning_rate": 1.3142549371633753e-05, "loss": 0.001, "step": 10050 }, { "epoch": 1.06, "learning_rate": 1.3135368043087972e-05, "loss": 0.0012, "step": 10060 }, { "epoch": 1.07, "learning_rate": 1.312818671454219e-05, "loss": 0.001, "step": 10070 }, { "epoch": 1.07, "learning_rate": 1.3121005385996411e-05, "loss": 0.0011, "step": 10080 }, { "epoch": 1.07, "learning_rate": 1.311382405745063e-05, "loss": 0.0012, "step": 10090 }, { "epoch": 1.07, "learning_rate": 1.3106642728904849e-05, "loss": 0.0013, "step": 10100 }, { "epoch": 1.07, "learning_rate": 1.3099461400359068e-05, "loss": 0.0013, "step": 10110 }, { "epoch": 1.07, "learning_rate": 1.3092280071813287e-05, "loss": 0.001, "step": 10120 }, { "epoch": 1.07, "learning_rate": 1.3085098743267505e-05, "loss": 0.0013, "step": 10130 }, { "epoch": 1.07, "learning_rate": 1.3077917414721724e-05, "loss": 0.0014, "step": 10140 }, { "epoch": 1.07, "learning_rate": 1.3070736086175943e-05, "loss": 0.0015, "step": 10150 }, { "epoch": 1.08, "learning_rate": 1.3063554757630164e-05, "loss": 0.0013, "step": 10160 }, { "epoch": 1.08, "learning_rate": 1.305637342908438e-05, "loss": 0.001, "step": 10170 }, { "epoch": 1.08, "learning_rate": 1.3049192100538601e-05, "loss": 0.0015, "step": 10180 }, { "epoch": 1.08, "learning_rate": 1.3042010771992818e-05, "loss": 0.0012, "step": 10190 }, { "epoch": 1.08, "learning_rate": 1.3034829443447039e-05, "loss": 0.0013, "step": 10200 }, { "epoch": 1.08, "learning_rate": 1.3027648114901256e-05, "loss": 0.0014, "step": 10210 }, { "epoch": 1.08, "learning_rate": 1.3020466786355477e-05, "loss": 0.001, "step": 10220 }, { "epoch": 1.08, "learning_rate": 1.3013285457809697e-05, "loss": 0.0014, "step": 10230 }, { "epoch": 1.08, "learning_rate": 1.3006104129263914e-05, "loss": 0.0011, "step": 10240 }, { "epoch": 1.08, "learning_rate": 1.2998922800718135e-05, "loss": 0.0014, "step": 10250 }, { "epoch": 1.09, "learning_rate": 1.2991741472172352e-05, "loss": 0.001, "step": 10260 }, { "epoch": 1.09, "learning_rate": 1.2984560143626572e-05, "loss": 0.0013, "step": 10270 }, { "epoch": 1.09, "learning_rate": 1.297737881508079e-05, "loss": 0.001, "step": 10280 }, { "epoch": 1.09, "learning_rate": 1.297019748653501e-05, "loss": 0.0013, "step": 10290 }, { "epoch": 1.09, "learning_rate": 1.296301615798923e-05, "loss": 0.0012, "step": 10300 }, { "epoch": 1.09, "learning_rate": 1.2955834829443448e-05, "loss": 0.0012, "step": 10310 }, { "epoch": 1.09, "learning_rate": 1.2948653500897668e-05, "loss": 0.0014, "step": 10320 }, { "epoch": 1.09, "learning_rate": 1.2941472172351885e-05, "loss": 0.0012, "step": 10330 }, { "epoch": 1.09, "learning_rate": 1.2934290843806106e-05, "loss": 0.0013, "step": 10340 }, { "epoch": 1.1, "learning_rate": 1.2927109515260323e-05, "loss": 0.0012, "step": 10350 }, { "epoch": 1.1, "learning_rate": 1.2919928186714544e-05, "loss": 0.0015, "step": 10360 }, { "epoch": 1.1, "learning_rate": 1.2912746858168762e-05, "loss": 0.0013, "step": 10370 }, { "epoch": 1.1, "learning_rate": 1.2905565529622981e-05, "loss": 0.0012, "step": 10380 }, { "epoch": 1.1, "learning_rate": 1.28983842010772e-05, "loss": 0.001, "step": 10390 }, { "epoch": 1.1, "learning_rate": 1.2891202872531419e-05, "loss": 0.0012, "step": 10400 }, { "epoch": 1.1, "learning_rate": 1.288402154398564e-05, "loss": 0.0013, "step": 10410 }, { "epoch": 1.1, "learning_rate": 1.2876840215439857e-05, "loss": 0.001, "step": 10420 }, { "epoch": 1.1, "learning_rate": 1.2869658886894077e-05, "loss": 0.0011, "step": 10430 }, { "epoch": 1.1, "learning_rate": 1.2862477558348296e-05, "loss": 0.0014, "step": 10440 }, { "epoch": 1.11, "learning_rate": 1.2855296229802515e-05, "loss": 0.0013, "step": 10450 }, { "epoch": 1.11, "learning_rate": 1.2848114901256734e-05, "loss": 0.0012, "step": 10460 }, { "epoch": 1.11, "learning_rate": 1.2840933572710952e-05, "loss": 0.0011, "step": 10470 }, { "epoch": 1.11, "learning_rate": 1.2833752244165171e-05, "loss": 0.0013, "step": 10480 }, { "epoch": 1.11, "learning_rate": 1.282657091561939e-05, "loss": 0.0011, "step": 10490 }, { "epoch": 1.11, "learning_rate": 1.2819389587073609e-05, "loss": 0.0012, "step": 10500 }, { "epoch": 1.11, "learning_rate": 1.281220825852783e-05, "loss": 0.0012, "step": 10510 }, { "epoch": 1.11, "learning_rate": 1.2805026929982047e-05, "loss": 0.0011, "step": 10520 }, { "epoch": 1.11, "learning_rate": 1.2797845601436267e-05, "loss": 0.0011, "step": 10530 }, { "epoch": 1.12, "learning_rate": 1.2790664272890484e-05, "loss": 0.0012, "step": 10540 }, { "epoch": 1.12, "learning_rate": 1.2783482944344705e-05, "loss": 0.001, "step": 10550 }, { "epoch": 1.12, "learning_rate": 1.2776301615798924e-05, "loss": 0.0012, "step": 10560 }, { "epoch": 1.12, "learning_rate": 1.2769120287253142e-05, "loss": 0.0012, "step": 10570 }, { "epoch": 1.12, "learning_rate": 1.2761938958707363e-05, "loss": 0.0013, "step": 10580 }, { "epoch": 1.12, "learning_rate": 1.275475763016158e-05, "loss": 0.0012, "step": 10590 }, { "epoch": 1.12, "learning_rate": 1.27475763016158e-05, "loss": 0.0011, "step": 10600 }, { "epoch": 1.12, "learning_rate": 1.2740394973070018e-05, "loss": 0.0011, "step": 10610 }, { "epoch": 1.12, "learning_rate": 1.2733213644524238e-05, "loss": 0.0014, "step": 10620 }, { "epoch": 1.12, "learning_rate": 1.2726032315978455e-05, "loss": 0.0013, "step": 10630 }, { "epoch": 1.13, "learning_rate": 1.2718850987432676e-05, "loss": 0.0013, "step": 10640 }, { "epoch": 1.13, "learning_rate": 1.2711669658886896e-05, "loss": 0.0012, "step": 10650 }, { "epoch": 1.13, "learning_rate": 1.2704488330341114e-05, "loss": 0.0012, "step": 10660 }, { "epoch": 1.13, "learning_rate": 1.2697307001795334e-05, "loss": 0.0015, "step": 10670 }, { "epoch": 1.13, "learning_rate": 1.2690125673249551e-05, "loss": 0.001, "step": 10680 }, { "epoch": 1.13, "learning_rate": 1.2682944344703772e-05, "loss": 0.0016, "step": 10690 }, { "epoch": 1.13, "learning_rate": 1.2675763016157989e-05, "loss": 0.0021, "step": 10700 }, { "epoch": 1.13, "learning_rate": 1.266858168761221e-05, "loss": 0.0014, "step": 10710 }, { "epoch": 1.13, "learning_rate": 1.266140035906643e-05, "loss": 0.0013, "step": 10720 }, { "epoch": 1.14, "learning_rate": 1.2654219030520647e-05, "loss": 0.001, "step": 10730 }, { "epoch": 1.14, "learning_rate": 1.2647037701974868e-05, "loss": 0.0011, "step": 10740 }, { "epoch": 1.14, "learning_rate": 1.2639856373429085e-05, "loss": 0.0013, "step": 10750 }, { "epoch": 1.14, "learning_rate": 1.2632675044883305e-05, "loss": 0.0014, "step": 10760 }, { "epoch": 1.14, "learning_rate": 1.2625493716337522e-05, "loss": 0.0014, "step": 10770 }, { "epoch": 1.14, "learning_rate": 1.2618312387791743e-05, "loss": 0.0013, "step": 10780 }, { "epoch": 1.14, "learning_rate": 1.2611131059245962e-05, "loss": 0.0012, "step": 10790 }, { "epoch": 1.14, "learning_rate": 1.260394973070018e-05, "loss": 0.0013, "step": 10800 }, { "epoch": 1.14, "learning_rate": 1.25967684021544e-05, "loss": 0.0012, "step": 10810 }, { "epoch": 1.14, "learning_rate": 1.2589587073608618e-05, "loss": 0.0012, "step": 10820 }, { "epoch": 1.15, "learning_rate": 1.2582405745062837e-05, "loss": 0.0013, "step": 10830 }, { "epoch": 1.15, "learning_rate": 1.2575224416517056e-05, "loss": 0.0011, "step": 10840 }, { "epoch": 1.15, "learning_rate": 1.2568043087971275e-05, "loss": 0.0013, "step": 10850 }, { "epoch": 1.15, "learning_rate": 1.2560861759425495e-05, "loss": 0.0011, "step": 10860 }, { "epoch": 1.15, "learning_rate": 1.2553680430879714e-05, "loss": 0.001, "step": 10870 }, { "epoch": 1.15, "learning_rate": 1.2546499102333933e-05, "loss": 0.0011, "step": 10880 }, { "epoch": 1.15, "learning_rate": 1.2539317773788152e-05, "loss": 0.0011, "step": 10890 }, { "epoch": 1.15, "learning_rate": 1.253213644524237e-05, "loss": 0.0011, "step": 10900 }, { "epoch": 1.15, "learning_rate": 1.252495511669659e-05, "loss": 0.0011, "step": 10910 }, { "epoch": 1.16, "learning_rate": 1.2517773788150808e-05, "loss": 0.0012, "step": 10920 }, { "epoch": 1.16, "learning_rate": 1.2510592459605029e-05, "loss": 0.0016, "step": 10930 }, { "epoch": 1.16, "learning_rate": 1.2503411131059246e-05, "loss": 0.0013, "step": 10940 }, { "epoch": 1.16, "learning_rate": 1.2496229802513467e-05, "loss": 0.0013, "step": 10950 }, { "epoch": 1.16, "learning_rate": 1.2489048473967684e-05, "loss": 0.0013, "step": 10960 }, { "epoch": 1.16, "learning_rate": 1.2481867145421904e-05, "loss": 0.0014, "step": 10970 }, { "epoch": 1.16, "learning_rate": 1.2474685816876121e-05, "loss": 0.0011, "step": 10980 }, { "epoch": 1.16, "learning_rate": 1.2467504488330342e-05, "loss": 0.0012, "step": 10990 }, { "epoch": 1.16, "learning_rate": 1.2460323159784562e-05, "loss": 0.0013, "step": 11000 }, { "epoch": 1.17, "learning_rate": 1.245314183123878e-05, "loss": 0.001, "step": 11010 }, { "epoch": 1.17, "learning_rate": 1.2445960502693e-05, "loss": 0.0013, "step": 11020 }, { "epoch": 1.17, "learning_rate": 1.2438779174147217e-05, "loss": 0.0011, "step": 11030 }, { "epoch": 1.17, "learning_rate": 1.2431597845601438e-05, "loss": 0.0011, "step": 11040 }, { "epoch": 1.17, "learning_rate": 1.2424416517055655e-05, "loss": 0.0012, "step": 11050 }, { "epoch": 1.17, "learning_rate": 1.2417235188509875e-05, "loss": 0.0012, "step": 11060 }, { "epoch": 1.17, "learning_rate": 1.2410053859964096e-05, "loss": 0.0011, "step": 11070 }, { "epoch": 1.17, "learning_rate": 1.2402872531418313e-05, "loss": 0.0012, "step": 11080 }, { "epoch": 1.17, "learning_rate": 1.2395691202872534e-05, "loss": 0.001, "step": 11090 }, { "epoch": 1.17, "learning_rate": 1.238850987432675e-05, "loss": 0.0012, "step": 11100 }, { "epoch": 1.18, "learning_rate": 1.2381328545780971e-05, "loss": 0.0014, "step": 11110 }, { "epoch": 1.18, "learning_rate": 1.2374147217235188e-05, "loss": 0.0011, "step": 11120 }, { "epoch": 1.18, "learning_rate": 1.2366965888689409e-05, "loss": 0.0012, "step": 11130 }, { "epoch": 1.18, "learning_rate": 1.2359784560143628e-05, "loss": 0.0013, "step": 11140 }, { "epoch": 1.18, "learning_rate": 1.2352603231597847e-05, "loss": 0.0012, "step": 11150 }, { "epoch": 1.18, "learning_rate": 1.2345421903052065e-05, "loss": 0.0013, "step": 11160 }, { "epoch": 1.18, "learning_rate": 1.2338240574506284e-05, "loss": 0.0012, "step": 11170 }, { "epoch": 1.18, "learning_rate": 1.2331059245960503e-05, "loss": 0.0011, "step": 11180 }, { "epoch": 1.18, "learning_rate": 1.2323877917414724e-05, "loss": 0.0014, "step": 11190 }, { "epoch": 1.19, "learning_rate": 1.2316696588868942e-05, "loss": 0.0015, "step": 11200 }, { "epoch": 1.19, "learning_rate": 1.2309515260323161e-05, "loss": 0.0014, "step": 11210 }, { "epoch": 1.19, "learning_rate": 1.230233393177738e-05, "loss": 0.0013, "step": 11220 }, { "epoch": 1.19, "learning_rate": 1.2295152603231599e-05, "loss": 0.0014, "step": 11230 }, { "epoch": 1.19, "learning_rate": 1.2287971274685818e-05, "loss": 0.0011, "step": 11240 }, { "epoch": 1.19, "learning_rate": 1.2280789946140037e-05, "loss": 0.0013, "step": 11250 }, { "epoch": 1.19, "learning_rate": 1.2273608617594257e-05, "loss": 0.001, "step": 11260 }, { "epoch": 1.19, "learning_rate": 1.2266427289048474e-05, "loss": 0.0014, "step": 11270 }, { "epoch": 1.19, "learning_rate": 1.2259245960502695e-05, "loss": 0.0012, "step": 11280 }, { "epoch": 1.19, "learning_rate": 1.2252064631956912e-05, "loss": 0.0013, "step": 11290 }, { "epoch": 1.2, "learning_rate": 1.2244883303411132e-05, "loss": 0.001, "step": 11300 }, { "epoch": 1.2, "learning_rate": 1.223770197486535e-05, "loss": 0.0013, "step": 11310 }, { "epoch": 1.2, "learning_rate": 1.223052064631957e-05, "loss": 0.0013, "step": 11320 }, { "epoch": 1.2, "learning_rate": 1.222333931777379e-05, "loss": 0.0012, "step": 11330 }, { "epoch": 1.2, "learning_rate": 1.2216157989228008e-05, "loss": 0.001, "step": 11340 }, { "epoch": 1.2, "learning_rate": 1.2208976660682228e-05, "loss": 0.0012, "step": 11350 }, { "epoch": 1.2, "learning_rate": 1.2201795332136445e-05, "loss": 0.0013, "step": 11360 }, { "epoch": 1.2, "learning_rate": 1.2194614003590666e-05, "loss": 0.0011, "step": 11370 }, { "epoch": 1.2, "learning_rate": 1.2187432675044883e-05, "loss": 0.0013, "step": 11380 }, { "epoch": 1.21, "learning_rate": 1.2180251346499104e-05, "loss": 0.0011, "step": 11390 }, { "epoch": 1.21, "learning_rate": 1.2173070017953324e-05, "loss": 0.0013, "step": 11400 }, { "epoch": 1.21, "learning_rate": 1.2165888689407541e-05, "loss": 0.0012, "step": 11410 }, { "epoch": 1.21, "learning_rate": 1.2158707360861762e-05, "loss": 0.0013, "step": 11420 }, { "epoch": 1.21, "learning_rate": 1.2151526032315979e-05, "loss": 0.0011, "step": 11430 }, { "epoch": 1.21, "learning_rate": 1.21443447037702e-05, "loss": 0.0012, "step": 11440 }, { "epoch": 1.21, "learning_rate": 1.2137163375224417e-05, "loss": 0.0011, "step": 11450 }, { "epoch": 1.21, "learning_rate": 1.2129982046678637e-05, "loss": 0.0014, "step": 11460 }, { "epoch": 1.21, "learning_rate": 1.2122800718132856e-05, "loss": 0.0014, "step": 11470 }, { "epoch": 1.21, "learning_rate": 1.2115619389587075e-05, "loss": 0.0012, "step": 11480 }, { "epoch": 1.22, "learning_rate": 1.2108438061041294e-05, "loss": 0.0013, "step": 11490 }, { "epoch": 1.22, "learning_rate": 1.2101256732495512e-05, "loss": 0.0012, "step": 11500 }, { "epoch": 1.22, "learning_rate": 1.2094075403949733e-05, "loss": 0.0013, "step": 11510 }, { "epoch": 1.22, "learning_rate": 1.208689407540395e-05, "loss": 0.0014, "step": 11520 }, { "epoch": 1.22, "learning_rate": 1.207971274685817e-05, "loss": 0.0013, "step": 11530 }, { "epoch": 1.22, "learning_rate": 1.207253141831239e-05, "loss": 0.0013, "step": 11540 }, { "epoch": 1.22, "learning_rate": 1.2065350089766608e-05, "loss": 0.0013, "step": 11550 }, { "epoch": 1.22, "learning_rate": 1.2058168761220827e-05, "loss": 0.0012, "step": 11560 }, { "epoch": 1.22, "learning_rate": 1.2050987432675046e-05, "loss": 0.001, "step": 11570 }, { "epoch": 1.23, "learning_rate": 1.2043806104129265e-05, "loss": 0.0013, "step": 11580 }, { "epoch": 1.23, "learning_rate": 1.2036624775583484e-05, "loss": 0.0013, "step": 11590 }, { "epoch": 1.23, "learning_rate": 1.2029443447037702e-05, "loss": 0.0014, "step": 11600 }, { "epoch": 1.23, "learning_rate": 1.2022262118491923e-05, "loss": 0.0012, "step": 11610 }, { "epoch": 1.23, "learning_rate": 1.201508078994614e-05, "loss": 0.0013, "step": 11620 }, { "epoch": 1.23, "learning_rate": 1.200789946140036e-05, "loss": 0.0011, "step": 11630 }, { "epoch": 1.23, "learning_rate": 1.2000718132854578e-05, "loss": 0.0012, "step": 11640 }, { "epoch": 1.23, "learning_rate": 1.1993536804308798e-05, "loss": 0.0011, "step": 11650 }, { "epoch": 1.23, "learning_rate": 1.1986355475763017e-05, "loss": 0.0013, "step": 11660 }, { "epoch": 1.23, "learning_rate": 1.1979174147217236e-05, "loss": 0.0012, "step": 11670 }, { "epoch": 1.24, "learning_rate": 1.1971992818671456e-05, "loss": 0.0012, "step": 11680 }, { "epoch": 1.24, "learning_rate": 1.1964811490125674e-05, "loss": 0.0012, "step": 11690 }, { "epoch": 1.24, "learning_rate": 1.1957630161579894e-05, "loss": 0.0014, "step": 11700 }, { "epoch": 1.24, "learning_rate": 1.1950448833034111e-05, "loss": 0.0012, "step": 11710 }, { "epoch": 1.24, "learning_rate": 1.1943267504488332e-05, "loss": 0.0013, "step": 11720 }, { "epoch": 1.24, "learning_rate": 1.1936086175942549e-05, "loss": 0.0012, "step": 11730 }, { "epoch": 1.24, "learning_rate": 1.192890484739677e-05, "loss": 0.0011, "step": 11740 }, { "epoch": 1.24, "learning_rate": 1.192172351885099e-05, "loss": 0.0015, "step": 11750 }, { "epoch": 1.24, "learning_rate": 1.1914542190305207e-05, "loss": 0.0013, "step": 11760 }, { "epoch": 1.25, "learning_rate": 1.1907360861759428e-05, "loss": 0.0013, "step": 11770 }, { "epoch": 1.25, "learning_rate": 1.1900179533213645e-05, "loss": 0.0013, "step": 11780 }, { "epoch": 1.25, "learning_rate": 1.1892998204667865e-05, "loss": 0.0012, "step": 11790 }, { "epoch": 1.25, "learning_rate": 1.1885816876122082e-05, "loss": 0.0014, "step": 11800 }, { "epoch": 1.25, "learning_rate": 1.1878635547576303e-05, "loss": 0.0012, "step": 11810 }, { "epoch": 1.25, "learning_rate": 1.1871454219030522e-05, "loss": 0.0012, "step": 11820 }, { "epoch": 1.25, "learning_rate": 1.186427289048474e-05, "loss": 0.0013, "step": 11830 }, { "epoch": 1.25, "learning_rate": 1.1857091561938961e-05, "loss": 0.0014, "step": 11840 }, { "epoch": 1.25, "learning_rate": 1.1849910233393178e-05, "loss": 0.0012, "step": 11850 }, { "epoch": 1.26, "learning_rate": 1.1842728904847399e-05, "loss": 0.0013, "step": 11860 }, { "epoch": 1.26, "learning_rate": 1.1835547576301616e-05, "loss": 0.0014, "step": 11870 }, { "epoch": 1.26, "learning_rate": 1.1828366247755836e-05, "loss": 0.0012, "step": 11880 }, { "epoch": 1.26, "learning_rate": 1.1821184919210055e-05, "loss": 0.0013, "step": 11890 }, { "epoch": 1.26, "learning_rate": 1.1814003590664274e-05, "loss": 0.0012, "step": 11900 }, { "epoch": 1.26, "learning_rate": 1.1806822262118493e-05, "loss": 0.0014, "step": 11910 }, { "epoch": 1.26, "learning_rate": 1.1799640933572712e-05, "loss": 0.0013, "step": 11920 }, { "epoch": 1.26, "learning_rate": 1.179245960502693e-05, "loss": 0.0012, "step": 11930 }, { "epoch": 1.26, "learning_rate": 1.178527827648115e-05, "loss": 0.001, "step": 11940 }, { "epoch": 1.26, "learning_rate": 1.1778096947935368e-05, "loss": 0.0012, "step": 11950 }, { "epoch": 1.27, "learning_rate": 1.1770915619389589e-05, "loss": 0.0012, "step": 11960 }, { "epoch": 1.27, "learning_rate": 1.1763734290843806e-05, "loss": 0.001, "step": 11970 }, { "epoch": 1.27, "learning_rate": 1.1756552962298026e-05, "loss": 0.0012, "step": 11980 }, { "epoch": 1.27, "learning_rate": 1.1749371633752245e-05, "loss": 0.0012, "step": 11990 }, { "epoch": 1.27, "learning_rate": 1.1742190305206464e-05, "loss": 0.0013, "step": 12000 }, { "epoch": 1.27, "learning_rate": 1.1735008976660683e-05, "loss": 0.0012, "step": 12010 }, { "epoch": 1.27, "learning_rate": 1.1727827648114902e-05, "loss": 0.0012, "step": 12020 }, { "epoch": 1.27, "learning_rate": 1.1720646319569122e-05, "loss": 0.0015, "step": 12030 }, { "epoch": 1.27, "learning_rate": 1.171346499102334e-05, "loss": 0.0013, "step": 12040 }, { "epoch": 1.28, "learning_rate": 1.170628366247756e-05, "loss": 0.0011, "step": 12050 }, { "epoch": 1.28, "learning_rate": 1.1699102333931777e-05, "loss": 0.0013, "step": 12060 }, { "epoch": 1.28, "learning_rate": 1.1691921005385998e-05, "loss": 0.0013, "step": 12070 }, { "epoch": 1.28, "learning_rate": 1.1684739676840215e-05, "loss": 0.0012, "step": 12080 }, { "epoch": 1.28, "learning_rate": 1.1677558348294435e-05, "loss": 0.0014, "step": 12090 }, { "epoch": 1.28, "learning_rate": 1.1670377019748656e-05, "loss": 0.0013, "step": 12100 }, { "epoch": 1.28, "learning_rate": 1.1663195691202873e-05, "loss": 0.0011, "step": 12110 }, { "epoch": 1.28, "learning_rate": 1.1656014362657094e-05, "loss": 0.0013, "step": 12120 }, { "epoch": 1.28, "learning_rate": 1.164883303411131e-05, "loss": 0.0013, "step": 12130 }, { "epoch": 1.28, "learning_rate": 1.1641651705565531e-05, "loss": 0.001, "step": 12140 }, { "epoch": 1.29, "learning_rate": 1.1634470377019748e-05, "loss": 0.0015, "step": 12150 }, { "epoch": 1.29, "learning_rate": 1.1627289048473969e-05, "loss": 0.0012, "step": 12160 }, { "epoch": 1.29, "learning_rate": 1.162010771992819e-05, "loss": 0.0013, "step": 12170 }, { "epoch": 1.29, "learning_rate": 1.1612926391382407e-05, "loss": 0.0014, "step": 12180 }, { "epoch": 1.29, "learning_rate": 1.1605745062836627e-05, "loss": 0.0011, "step": 12190 }, { "epoch": 1.29, "learning_rate": 1.1598563734290844e-05, "loss": 0.0014, "step": 12200 }, { "epoch": 1.29, "learning_rate": 1.1591382405745065e-05, "loss": 0.0011, "step": 12210 }, { "epoch": 1.29, "learning_rate": 1.1584201077199282e-05, "loss": 0.0013, "step": 12220 }, { "epoch": 1.29, "learning_rate": 1.1577019748653502e-05, "loss": 0.0013, "step": 12230 }, { "epoch": 1.3, "learning_rate": 1.1569838420107721e-05, "loss": 0.0015, "step": 12240 }, { "epoch": 1.3, "learning_rate": 1.156265709156194e-05, "loss": 0.0012, "step": 12250 }, { "epoch": 1.3, "learning_rate": 1.1555475763016159e-05, "loss": 0.001, "step": 12260 }, { "epoch": 1.3, "learning_rate": 1.1548294434470378e-05, "loss": 0.0013, "step": 12270 }, { "epoch": 1.3, "learning_rate": 1.1541113105924597e-05, "loss": 0.0012, "step": 12280 }, { "epoch": 1.3, "learning_rate": 1.1533931777378815e-05, "loss": 0.0011, "step": 12290 }, { "epoch": 1.3, "learning_rate": 1.1526750448833036e-05, "loss": 0.0011, "step": 12300 }, { "epoch": 1.3, "learning_rate": 1.1519569120287255e-05, "loss": 0.0012, "step": 12310 }, { "epoch": 1.3, "learning_rate": 1.1512387791741474e-05, "loss": 0.0013, "step": 12320 }, { "epoch": 1.3, "learning_rate": 1.1505206463195692e-05, "loss": 0.0012, "step": 12330 }, { "epoch": 1.31, "learning_rate": 1.1498025134649911e-05, "loss": 0.0011, "step": 12340 }, { "epoch": 1.31, "learning_rate": 1.149084380610413e-05, "loss": 0.0013, "step": 12350 }, { "epoch": 1.31, "learning_rate": 1.1483662477558349e-05, "loss": 0.0011, "step": 12360 }, { "epoch": 1.31, "learning_rate": 1.1476481149012568e-05, "loss": 0.0012, "step": 12370 }, { "epoch": 1.31, "learning_rate": 1.1469299820466788e-05, "loss": 0.0015, "step": 12380 }, { "epoch": 1.31, "learning_rate": 1.1462118491921005e-05, "loss": 0.0012, "step": 12390 }, { "epoch": 1.31, "learning_rate": 1.1454937163375226e-05, "loss": 0.0012, "step": 12400 }, { "epoch": 1.31, "learning_rate": 1.1447755834829443e-05, "loss": 0.0013, "step": 12410 }, { "epoch": 1.31, "learning_rate": 1.1440574506283664e-05, "loss": 0.0012, "step": 12420 }, { "epoch": 1.32, "learning_rate": 1.143339317773788e-05, "loss": 0.0012, "step": 12430 }, { "epoch": 1.32, "learning_rate": 1.1426211849192101e-05, "loss": 0.0012, "step": 12440 }, { "epoch": 1.32, "learning_rate": 1.1419030520646322e-05, "loss": 0.0013, "step": 12450 }, { "epoch": 1.32, "learning_rate": 1.1411849192100539e-05, "loss": 0.0012, "step": 12460 }, { "epoch": 1.32, "learning_rate": 1.140466786355476e-05, "loss": 0.0012, "step": 12470 }, { "epoch": 1.32, "learning_rate": 1.1397486535008977e-05, "loss": 0.0013, "step": 12480 }, { "epoch": 1.32, "learning_rate": 1.1390305206463197e-05, "loss": 0.0013, "step": 12490 }, { "epoch": 1.32, "learning_rate": 1.1383123877917414e-05, "loss": 0.0009, "step": 12500 }, { "epoch": 1.32, "learning_rate": 1.1375942549371635e-05, "loss": 0.0013, "step": 12510 }, { "epoch": 1.32, "learning_rate": 1.1368761220825855e-05, "loss": 0.0011, "step": 12520 }, { "epoch": 1.33, "learning_rate": 1.1361579892280072e-05, "loss": 0.0012, "step": 12530 }, { "epoch": 1.33, "learning_rate": 1.1354398563734293e-05, "loss": 0.0013, "step": 12540 }, { "epoch": 1.33, "learning_rate": 1.134721723518851e-05, "loss": 0.0013, "step": 12550 }, { "epoch": 1.33, "learning_rate": 1.134003590664273e-05, "loss": 0.001, "step": 12560 }, { "epoch": 1.33, "learning_rate": 1.1332854578096948e-05, "loss": 0.0012, "step": 12570 }, { "epoch": 1.33, "learning_rate": 1.1325673249551168e-05, "loss": 0.0013, "step": 12580 }, { "epoch": 1.33, "learning_rate": 1.1318491921005387e-05, "loss": 0.0012, "step": 12590 }, { "epoch": 1.33, "learning_rate": 1.1311310592459606e-05, "loss": 0.0012, "step": 12600 }, { "epoch": 1.33, "learning_rate": 1.1304129263913826e-05, "loss": 0.0014, "step": 12610 }, { "epoch": 1.34, "learning_rate": 1.1296947935368044e-05, "loss": 0.0013, "step": 12620 }, { "epoch": 1.34, "learning_rate": 1.1289766606822264e-05, "loss": 0.0015, "step": 12630 }, { "epoch": 1.34, "learning_rate": 1.1282585278276481e-05, "loss": 0.001, "step": 12640 }, { "epoch": 1.34, "learning_rate": 1.1275403949730702e-05, "loss": 0.0012, "step": 12650 }, { "epoch": 1.34, "learning_rate": 1.126822262118492e-05, "loss": 0.0015, "step": 12660 }, { "epoch": 1.34, "learning_rate": 1.126104129263914e-05, "loss": 0.0011, "step": 12670 }, { "epoch": 1.34, "learning_rate": 1.1253859964093358e-05, "loss": 0.001, "step": 12680 }, { "epoch": 1.34, "learning_rate": 1.1246678635547577e-05, "loss": 0.0012, "step": 12690 }, { "epoch": 1.34, "learning_rate": 1.1239497307001796e-05, "loss": 0.0014, "step": 12700 }, { "epoch": 1.34, "learning_rate": 1.1232315978456015e-05, "loss": 0.0011, "step": 12710 }, { "epoch": 1.35, "learning_rate": 1.1225134649910234e-05, "loss": 0.0013, "step": 12720 }, { "epoch": 1.35, "learning_rate": 1.1217953321364454e-05, "loss": 0.0013, "step": 12730 }, { "epoch": 1.35, "learning_rate": 1.1210771992818671e-05, "loss": 0.0012, "step": 12740 }, { "epoch": 1.35, "learning_rate": 1.1203590664272892e-05, "loss": 0.0012, "step": 12750 }, { "epoch": 1.35, "learning_rate": 1.1196409335727109e-05, "loss": 0.0013, "step": 12760 }, { "epoch": 1.35, "learning_rate": 1.118922800718133e-05, "loss": 0.0013, "step": 12770 }, { "epoch": 1.35, "learning_rate": 1.1182046678635548e-05, "loss": 0.0015, "step": 12780 }, { "epoch": 1.35, "learning_rate": 1.1174865350089767e-05, "loss": 0.0014, "step": 12790 }, { "epoch": 1.35, "learning_rate": 1.1167684021543988e-05, "loss": 0.0012, "step": 12800 }, { "epoch": 1.36, "learning_rate": 1.1160502692998205e-05, "loss": 0.001, "step": 12810 }, { "epoch": 1.36, "learning_rate": 1.1153321364452425e-05, "loss": 0.0013, "step": 12820 }, { "epoch": 1.36, "learning_rate": 1.1146140035906642e-05, "loss": 0.0015, "step": 12830 }, { "epoch": 1.36, "learning_rate": 1.1138958707360863e-05, "loss": 0.0013, "step": 12840 }, { "epoch": 1.36, "learning_rate": 1.113177737881508e-05, "loss": 0.0012, "step": 12850 }, { "epoch": 1.36, "learning_rate": 1.11245960502693e-05, "loss": 0.0012, "step": 12860 }, { "epoch": 1.36, "learning_rate": 1.1117414721723521e-05, "loss": 0.0012, "step": 12870 }, { "epoch": 1.36, "learning_rate": 1.1110233393177738e-05, "loss": 0.0012, "step": 12880 }, { "epoch": 1.36, "learning_rate": 1.1103052064631959e-05, "loss": 0.0014, "step": 12890 }, { "epoch": 1.37, "learning_rate": 1.1095870736086176e-05, "loss": 0.0012, "step": 12900 }, { "epoch": 1.37, "learning_rate": 1.1088689407540396e-05, "loss": 0.0015, "step": 12910 }, { "epoch": 1.37, "learning_rate": 1.1081508078994614e-05, "loss": 0.0011, "step": 12920 }, { "epoch": 1.37, "learning_rate": 1.1074326750448834e-05, "loss": 0.0012, "step": 12930 }, { "epoch": 1.37, "learning_rate": 1.1067145421903055e-05, "loss": 0.0012, "step": 12940 }, { "epoch": 1.37, "learning_rate": 1.1059964093357272e-05, "loss": 0.0011, "step": 12950 }, { "epoch": 1.37, "learning_rate": 1.1052782764811492e-05, "loss": 0.0011, "step": 12960 }, { "epoch": 1.37, "learning_rate": 1.104560143626571e-05, "loss": 0.0015, "step": 12970 }, { "epoch": 1.37, "learning_rate": 1.103842010771993e-05, "loss": 0.0011, "step": 12980 }, { "epoch": 1.37, "learning_rate": 1.1031238779174147e-05, "loss": 0.0012, "step": 12990 }, { "epoch": 1.38, "learning_rate": 1.1024057450628368e-05, "loss": 0.0014, "step": 13000 }, { "epoch": 1.38, "learning_rate": 1.1016876122082586e-05, "loss": 0.0015, "step": 13010 }, { "epoch": 1.38, "learning_rate": 1.1009694793536805e-05, "loss": 0.0009, "step": 13020 }, { "epoch": 1.38, "learning_rate": 1.1002513464991024e-05, "loss": 0.0014, "step": 13030 }, { "epoch": 1.38, "learning_rate": 1.0995332136445243e-05, "loss": 0.0013, "step": 13040 }, { "epoch": 1.38, "learning_rate": 1.0988150807899462e-05, "loss": 0.0012, "step": 13050 }, { "epoch": 1.38, "learning_rate": 1.098096947935368e-05, "loss": 0.0012, "step": 13060 }, { "epoch": 1.38, "learning_rate": 1.09737881508079e-05, "loss": 0.0014, "step": 13070 }, { "epoch": 1.38, "learning_rate": 1.096660682226212e-05, "loss": 0.0011, "step": 13080 }, { "epoch": 1.39, "learning_rate": 1.0959425493716339e-05, "loss": 0.0012, "step": 13090 }, { "epoch": 1.39, "learning_rate": 1.0952244165170558e-05, "loss": 0.0013, "step": 13100 }, { "epoch": 1.39, "learning_rate": 1.0945062836624776e-05, "loss": 0.0013, "step": 13110 }, { "epoch": 1.39, "learning_rate": 1.0937881508078995e-05, "loss": 0.0013, "step": 13120 }, { "epoch": 1.39, "learning_rate": 1.0930700179533214e-05, "loss": 0.0012, "step": 13130 }, { "epoch": 1.39, "learning_rate": 1.0923518850987433e-05, "loss": 0.0011, "step": 13140 }, { "epoch": 1.39, "learning_rate": 1.0916337522441654e-05, "loss": 0.0012, "step": 13150 }, { "epoch": 1.39, "learning_rate": 1.090915619389587e-05, "loss": 0.001, "step": 13160 }, { "epoch": 1.39, "learning_rate": 1.0901974865350091e-05, "loss": 0.0012, "step": 13170 }, { "epoch": 1.39, "learning_rate": 1.0894793536804308e-05, "loss": 0.0014, "step": 13180 }, { "epoch": 1.4, "learning_rate": 1.0887612208258529e-05, "loss": 0.0012, "step": 13190 }, { "epoch": 1.4, "learning_rate": 1.0880430879712746e-05, "loss": 0.001, "step": 13200 }, { "epoch": 1.4, "learning_rate": 1.0873249551166966e-05, "loss": 0.0011, "step": 13210 }, { "epoch": 1.4, "learning_rate": 1.0866068222621187e-05, "loss": 0.0015, "step": 13220 }, { "epoch": 1.4, "learning_rate": 1.0858886894075404e-05, "loss": 0.0012, "step": 13230 }, { "epoch": 1.4, "learning_rate": 1.0851705565529625e-05, "loss": 0.0012, "step": 13240 }, { "epoch": 1.4, "learning_rate": 1.0844524236983842e-05, "loss": 0.0012, "step": 13250 }, { "epoch": 1.4, "learning_rate": 1.0837342908438062e-05, "loss": 0.0012, "step": 13260 }, { "epoch": 1.4, "learning_rate": 1.083016157989228e-05, "loss": 0.0013, "step": 13270 }, { "epoch": 1.41, "learning_rate": 1.08229802513465e-05, "loss": 0.0014, "step": 13280 }, { "epoch": 1.41, "learning_rate": 1.081579892280072e-05, "loss": 0.0012, "step": 13290 }, { "epoch": 1.41, "learning_rate": 1.0808617594254938e-05, "loss": 0.0011, "step": 13300 }, { "epoch": 1.41, "learning_rate": 1.0801436265709158e-05, "loss": 0.0011, "step": 13310 }, { "epoch": 1.41, "learning_rate": 1.0794254937163375e-05, "loss": 0.0013, "step": 13320 }, { "epoch": 1.41, "learning_rate": 1.0787073608617596e-05, "loss": 0.0011, "step": 13330 }, { "epoch": 1.41, "learning_rate": 1.0779892280071813e-05, "loss": 0.0011, "step": 13340 }, { "epoch": 1.41, "learning_rate": 1.0772710951526034e-05, "loss": 0.0012, "step": 13350 }, { "epoch": 1.41, "learning_rate": 1.0765529622980252e-05, "loss": 0.0013, "step": 13360 }, { "epoch": 1.41, "learning_rate": 1.0758348294434471e-05, "loss": 0.0012, "step": 13370 }, { "epoch": 1.42, "learning_rate": 1.075116696588869e-05, "loss": 0.0012, "step": 13380 }, { "epoch": 1.42, "learning_rate": 1.0743985637342909e-05, "loss": 0.0013, "step": 13390 }, { "epoch": 1.42, "learning_rate": 1.073680430879713e-05, "loss": 0.0012, "step": 13400 }, { "epoch": 1.42, "learning_rate": 1.0729622980251347e-05, "loss": 0.0013, "step": 13410 }, { "epoch": 1.42, "learning_rate": 1.0722441651705567e-05, "loss": 0.0011, "step": 13420 }, { "epoch": 1.42, "learning_rate": 1.0715260323159786e-05, "loss": 0.0012, "step": 13430 }, { "epoch": 1.42, "learning_rate": 1.0708078994614005e-05, "loss": 0.0013, "step": 13440 }, { "epoch": 1.42, "learning_rate": 1.0700897666068224e-05, "loss": 0.0012, "step": 13450 }, { "epoch": 1.42, "learning_rate": 1.0693716337522442e-05, "loss": 0.0012, "step": 13460 }, { "epoch": 1.43, "learning_rate": 1.0686535008976661e-05, "loss": 0.0013, "step": 13470 }, { "epoch": 1.43, "learning_rate": 1.067935368043088e-05, "loss": 0.0011, "step": 13480 }, { "epoch": 1.43, "learning_rate": 1.0672172351885099e-05, "loss": 0.0013, "step": 13490 }, { "epoch": 1.43, "learning_rate": 1.066499102333932e-05, "loss": 0.0015, "step": 13500 }, { "epoch": 1.43, "learning_rate": 1.0657809694793537e-05, "loss": 0.0013, "step": 13510 }, { "epoch": 1.43, "learning_rate": 1.0650628366247757e-05, "loss": 0.0014, "step": 13520 }, { "epoch": 1.43, "learning_rate": 1.0643447037701974e-05, "loss": 0.0012, "step": 13530 }, { "epoch": 1.43, "learning_rate": 1.0636265709156195e-05, "loss": 0.0013, "step": 13540 }, { "epoch": 1.43, "learning_rate": 1.0629084380610412e-05, "loss": 0.0011, "step": 13550 }, { "epoch": 1.43, "learning_rate": 1.0621903052064632e-05, "loss": 0.0012, "step": 13560 }, { "epoch": 1.44, "learning_rate": 1.0614721723518853e-05, "loss": 0.0014, "step": 13570 }, { "epoch": 1.44, "learning_rate": 1.060754039497307e-05, "loss": 0.0013, "step": 13580 }, { "epoch": 1.44, "learning_rate": 1.060035906642729e-05, "loss": 0.0011, "step": 13590 }, { "epoch": 1.44, "learning_rate": 1.0593177737881508e-05, "loss": 0.0011, "step": 13600 }, { "epoch": 1.44, "learning_rate": 1.0585996409335728e-05, "loss": 0.001, "step": 13610 }, { "epoch": 1.44, "learning_rate": 1.0578815080789945e-05, "loss": 0.0013, "step": 13620 }, { "epoch": 1.44, "learning_rate": 1.0571633752244166e-05, "loss": 0.0012, "step": 13630 }, { "epoch": 1.44, "learning_rate": 1.0564452423698386e-05, "loss": 0.0013, "step": 13640 }, { "epoch": 1.44, "learning_rate": 1.0557271095152604e-05, "loss": 0.0012, "step": 13650 }, { "epoch": 1.45, "learning_rate": 1.0550089766606824e-05, "loss": 0.0013, "step": 13660 }, { "epoch": 1.45, "learning_rate": 1.0542908438061041e-05, "loss": 0.0013, "step": 13670 }, { "epoch": 1.45, "learning_rate": 1.0535727109515262e-05, "loss": 0.0013, "step": 13680 }, { "epoch": 1.45, "learning_rate": 1.052854578096948e-05, "loss": 0.0014, "step": 13690 }, { "epoch": 1.45, "learning_rate": 1.05213644524237e-05, "loss": 0.0012, "step": 13700 }, { "epoch": 1.45, "learning_rate": 1.0514183123877918e-05, "loss": 0.0011, "step": 13710 }, { "epoch": 1.45, "learning_rate": 1.0507001795332137e-05, "loss": 0.001, "step": 13720 }, { "epoch": 1.45, "learning_rate": 1.0499820466786358e-05, "loss": 0.0013, "step": 13730 }, { "epoch": 1.45, "learning_rate": 1.0492639138240575e-05, "loss": 0.0013, "step": 13740 }, { "epoch": 1.46, "learning_rate": 1.0485457809694795e-05, "loss": 0.0013, "step": 13750 }, { "epoch": 1.46, "learning_rate": 1.0478276481149014e-05, "loss": 0.0011, "step": 13760 }, { "epoch": 1.46, "learning_rate": 1.0471095152603233e-05, "loss": 0.001, "step": 13770 }, { "epoch": 1.46, "learning_rate": 1.0463913824057452e-05, "loss": 0.0011, "step": 13780 }, { "epoch": 1.46, "learning_rate": 1.045673249551167e-05, "loss": 0.0012, "step": 13790 }, { "epoch": 1.46, "learning_rate": 1.044955116696589e-05, "loss": 0.001, "step": 13800 }, { "epoch": 1.46, "learning_rate": 1.0442369838420108e-05, "loss": 0.001, "step": 13810 }, { "epoch": 1.46, "learning_rate": 1.0435188509874327e-05, "loss": 0.0012, "step": 13820 }, { "epoch": 1.46, "learning_rate": 1.0428007181328548e-05, "loss": 0.0012, "step": 13830 }, { "epoch": 1.46, "learning_rate": 1.0420825852782765e-05, "loss": 0.0011, "step": 13840 }, { "epoch": 1.47, "learning_rate": 1.0413644524236985e-05, "loss": 0.0011, "step": 13850 }, { "epoch": 1.47, "learning_rate": 1.0406463195691202e-05, "loss": 0.0012, "step": 13860 }, { "epoch": 1.47, "learning_rate": 1.0399281867145423e-05, "loss": 0.0011, "step": 13870 }, { "epoch": 1.47, "learning_rate": 1.0392100538599642e-05, "loss": 0.0012, "step": 13880 }, { "epoch": 1.47, "learning_rate": 1.038491921005386e-05, "loss": 0.001, "step": 13890 }, { "epoch": 1.47, "learning_rate": 1.0377737881508081e-05, "loss": 0.0008, "step": 13900 }, { "epoch": 1.47, "learning_rate": 1.0370556552962298e-05, "loss": 0.0012, "step": 13910 }, { "epoch": 1.47, "learning_rate": 1.0363375224416519e-05, "loss": 0.0012, "step": 13920 }, { "epoch": 1.47, "learning_rate": 1.0356193895870736e-05, "loss": 0.0012, "step": 13930 }, { "epoch": 1.48, "learning_rate": 1.0349012567324956e-05, "loss": 0.0016, "step": 13940 }, { "epoch": 1.48, "learning_rate": 1.0341831238779174e-05, "loss": 0.0013, "step": 13950 }, { "epoch": 1.48, "learning_rate": 1.0334649910233394e-05, "loss": 0.0013, "step": 13960 }, { "epoch": 1.48, "learning_rate": 1.0327468581687615e-05, "loss": 0.0014, "step": 13970 }, { "epoch": 1.48, "learning_rate": 1.0320287253141832e-05, "loss": 0.0011, "step": 13980 }, { "epoch": 1.48, "learning_rate": 1.0313105924596052e-05, "loss": 0.0012, "step": 13990 }, { "epoch": 1.48, "learning_rate": 1.030592459605027e-05, "loss": 0.0014, "step": 14000 }, { "epoch": 1.48, "learning_rate": 1.029874326750449e-05, "loss": 0.0011, "step": 14010 }, { "epoch": 1.48, "learning_rate": 1.0291561938958707e-05, "loss": 0.0013, "step": 14020 }, { "epoch": 1.48, "learning_rate": 1.0284380610412928e-05, "loss": 0.001, "step": 14030 }, { "epoch": 1.49, "learning_rate": 1.0277199281867148e-05, "loss": 0.0012, "step": 14040 }, { "epoch": 1.49, "learning_rate": 1.0270017953321365e-05, "loss": 0.001, "step": 14050 }, { "epoch": 1.49, "learning_rate": 1.0262836624775586e-05, "loss": 0.0011, "step": 14060 }, { "epoch": 1.49, "learning_rate": 1.0255655296229803e-05, "loss": 0.0013, "step": 14070 }, { "epoch": 1.49, "learning_rate": 1.0248473967684023e-05, "loss": 0.0015, "step": 14080 }, { "epoch": 1.49, "learning_rate": 1.024129263913824e-05, "loss": 0.0013, "step": 14090 }, { "epoch": 1.49, "learning_rate": 1.0234111310592461e-05, "loss": 0.0013, "step": 14100 }, { "epoch": 1.49, "learning_rate": 1.022692998204668e-05, "loss": 0.0011, "step": 14110 }, { "epoch": 1.49, "learning_rate": 1.0219748653500899e-05, "loss": 0.0012, "step": 14120 }, { "epoch": 1.5, "learning_rate": 1.0212567324955118e-05, "loss": 0.0013, "step": 14130 }, { "epoch": 1.5, "learning_rate": 1.0205385996409336e-05, "loss": 0.0011, "step": 14140 }, { "epoch": 1.5, "learning_rate": 1.0198204667863555e-05, "loss": 0.0014, "step": 14150 }, { "epoch": 1.5, "learning_rate": 1.0191023339317774e-05, "loss": 0.0013, "step": 14160 }, { "epoch": 1.5, "learning_rate": 1.0183842010771993e-05, "loss": 0.001, "step": 14170 }, { "epoch": 1.5, "learning_rate": 1.0176660682226213e-05, "loss": 0.0011, "step": 14180 }, { "epoch": 1.5, "learning_rate": 1.0169479353680432e-05, "loss": 0.0011, "step": 14190 }, { "epoch": 1.5, "learning_rate": 1.0162298025134651e-05, "loss": 0.0013, "step": 14200 }, { "epoch": 1.5, "learning_rate": 1.015511669658887e-05, "loss": 0.0014, "step": 14210 }, { "epoch": 1.5, "learning_rate": 1.0147935368043089e-05, "loss": 0.0012, "step": 14220 }, { "epoch": 1.51, "learning_rate": 1.0140754039497308e-05, "loss": 0.0015, "step": 14230 }, { "epoch": 1.51, "learning_rate": 1.0133572710951526e-05, "loss": 0.0014, "step": 14240 }, { "epoch": 1.51, "learning_rate": 1.0126391382405747e-05, "loss": 0.0014, "step": 14250 }, { "epoch": 1.51, "learning_rate": 1.0119210053859964e-05, "loss": 0.0011, "step": 14260 }, { "epoch": 1.51, "learning_rate": 1.0112028725314185e-05, "loss": 0.0012, "step": 14270 }, { "epoch": 1.51, "learning_rate": 1.0104847396768402e-05, "loss": 0.0013, "step": 14280 }, { "epoch": 1.51, "learning_rate": 1.0097666068222622e-05, "loss": 0.001, "step": 14290 }, { "epoch": 1.51, "learning_rate": 1.009048473967684e-05, "loss": 0.0013, "step": 14300 }, { "epoch": 1.51, "learning_rate": 1.008330341113106e-05, "loss": 0.001, "step": 14310 }, { "epoch": 1.52, "learning_rate": 1.007612208258528e-05, "loss": 0.0011, "step": 14320 }, { "epoch": 1.52, "learning_rate": 1.0068940754039498e-05, "loss": 0.001, "step": 14330 }, { "epoch": 1.52, "learning_rate": 1.0061759425493718e-05, "loss": 0.0012, "step": 14340 }, { "epoch": 1.52, "learning_rate": 1.0054578096947935e-05, "loss": 0.0011, "step": 14350 }, { "epoch": 1.52, "learning_rate": 1.0047396768402156e-05, "loss": 0.0012, "step": 14360 }, { "epoch": 1.52, "learning_rate": 1.0040215439856373e-05, "loss": 0.0009, "step": 14370 }, { "epoch": 1.52, "learning_rate": 1.0033034111310594e-05, "loss": 0.0012, "step": 14380 }, { "epoch": 1.52, "learning_rate": 1.0025852782764814e-05, "loss": 0.0012, "step": 14390 }, { "epoch": 1.52, "learning_rate": 1.0018671454219031e-05, "loss": 0.0011, "step": 14400 }, { "epoch": 1.52, "learning_rate": 1.0011490125673252e-05, "loss": 0.0013, "step": 14410 }, { "epoch": 1.53, "learning_rate": 1.0004308797127469e-05, "loss": 0.0011, "step": 14420 }, { "epoch": 1.53, "learning_rate": 9.99712746858169e-06, "loss": 0.0012, "step": 14430 }, { "epoch": 1.53, "learning_rate": 9.989946140035908e-06, "loss": 0.0016, "step": 14440 }, { "epoch": 1.53, "learning_rate": 9.982764811490127e-06, "loss": 0.0012, "step": 14450 }, { "epoch": 1.53, "learning_rate": 9.975583482944346e-06, "loss": 0.0012, "step": 14460 }, { "epoch": 1.53, "learning_rate": 9.968402154398565e-06, "loss": 0.0014, "step": 14470 }, { "epoch": 1.53, "learning_rate": 9.961220825852784e-06, "loss": 0.0012, "step": 14480 }, { "epoch": 1.53, "learning_rate": 9.954039497307002e-06, "loss": 0.0012, "step": 14490 }, { "epoch": 1.53, "learning_rate": 9.946858168761221e-06, "loss": 0.0013, "step": 14500 }, { "epoch": 1.54, "learning_rate": 9.939676840215442e-06, "loss": 0.0013, "step": 14510 }, { "epoch": 1.54, "learning_rate": 9.93249551166966e-06, "loss": 0.0016, "step": 14520 }, { "epoch": 1.54, "learning_rate": 9.92531418312388e-06, "loss": 0.001, "step": 14530 }, { "epoch": 1.54, "learning_rate": 9.918132854578098e-06, "loss": 0.0012, "step": 14540 }, { "epoch": 1.54, "learning_rate": 9.910951526032317e-06, "loss": 0.0009, "step": 14550 }, { "epoch": 1.54, "learning_rate": 9.903770197486536e-06, "loss": 0.0013, "step": 14560 }, { "epoch": 1.54, "learning_rate": 9.896588868940755e-06, "loss": 0.0013, "step": 14570 }, { "epoch": 1.54, "learning_rate": 9.889407540394974e-06, "loss": 0.0012, "step": 14580 }, { "epoch": 1.54, "learning_rate": 9.882226211849192e-06, "loss": 0.0014, "step": 14590 }, { "epoch": 1.54, "learning_rate": 9.875044883303411e-06, "loss": 0.0012, "step": 14600 }, { "epoch": 1.55, "learning_rate": 9.86786355475763e-06, "loss": 0.0016, "step": 14610 }, { "epoch": 1.55, "learning_rate": 9.860682226211849e-06, "loss": 0.0012, "step": 14620 }, { "epoch": 1.55, "learning_rate": 9.85350089766607e-06, "loss": 0.0012, "step": 14630 }, { "epoch": 1.55, "learning_rate": 9.846319569120288e-06, "loss": 0.0014, "step": 14640 }, { "epoch": 1.55, "learning_rate": 9.839138240574507e-06, "loss": 0.0013, "step": 14650 }, { "epoch": 1.55, "learning_rate": 9.831956912028726e-06, "loss": 0.0013, "step": 14660 }, { "epoch": 1.55, "learning_rate": 9.824775583482945e-06, "loss": 0.0013, "step": 14670 }, { "epoch": 1.55, "learning_rate": 9.817594254937164e-06, "loss": 0.0012, "step": 14680 }, { "epoch": 1.55, "learning_rate": 9.810412926391382e-06, "loss": 0.0012, "step": 14690 }, { "epoch": 1.56, "learning_rate": 9.803231597845603e-06, "loss": 0.0012, "step": 14700 }, { "epoch": 1.56, "learning_rate": 9.796050269299822e-06, "loss": 0.0011, "step": 14710 }, { "epoch": 1.56, "learning_rate": 9.78886894075404e-06, "loss": 0.001, "step": 14720 }, { "epoch": 1.56, "learning_rate": 9.78168761220826e-06, "loss": 0.0012, "step": 14730 }, { "epoch": 1.56, "learning_rate": 9.774506283662478e-06, "loss": 0.0013, "step": 14740 }, { "epoch": 1.56, "learning_rate": 9.767324955116697e-06, "loss": 0.0012, "step": 14750 }, { "epoch": 1.56, "learning_rate": 9.760143626570916e-06, "loss": 0.0014, "step": 14760 }, { "epoch": 1.56, "learning_rate": 9.752962298025136e-06, "loss": 0.0012, "step": 14770 }, { "epoch": 1.56, "learning_rate": 9.745780969479355e-06, "loss": 0.0011, "step": 14780 }, { "epoch": 1.57, "learning_rate": 9.738599640933574e-06, "loss": 0.0012, "step": 14790 }, { "epoch": 1.57, "learning_rate": 9.731418312387793e-06, "loss": 0.0013, "step": 14800 }, { "epoch": 1.57, "learning_rate": 9.724236983842012e-06, "loss": 0.0012, "step": 14810 }, { "epoch": 1.57, "learning_rate": 9.71705565529623e-06, "loss": 0.0011, "step": 14820 }, { "epoch": 1.57, "learning_rate": 9.70987432675045e-06, "loss": 0.0013, "step": 14830 }, { "epoch": 1.57, "learning_rate": 9.70269299820467e-06, "loss": 0.0011, "step": 14840 }, { "epoch": 1.57, "learning_rate": 9.695511669658889e-06, "loss": 0.0014, "step": 14850 }, { "epoch": 1.57, "learning_rate": 9.688330341113108e-06, "loss": 0.0013, "step": 14860 }, { "epoch": 1.57, "learning_rate": 9.681149012567326e-06, "loss": 0.0011, "step": 14870 }, { "epoch": 1.57, "learning_rate": 9.673967684021545e-06, "loss": 0.0011, "step": 14880 }, { "epoch": 1.58, "learning_rate": 9.666786355475764e-06, "loss": 0.0011, "step": 14890 }, { "epoch": 1.58, "learning_rate": 9.659605026929983e-06, "loss": 0.0013, "step": 14900 }, { "epoch": 1.58, "learning_rate": 9.652423698384202e-06, "loss": 0.0012, "step": 14910 }, { "epoch": 1.58, "learning_rate": 9.64524236983842e-06, "loss": 0.0012, "step": 14920 }, { "epoch": 1.58, "learning_rate": 9.63806104129264e-06, "loss": 0.0011, "step": 14930 }, { "epoch": 1.58, "learning_rate": 9.630879712746858e-06, "loss": 0.0012, "step": 14940 }, { "epoch": 1.58, "learning_rate": 9.623698384201077e-06, "loss": 0.0012, "step": 14950 }, { "epoch": 1.58, "learning_rate": 9.616517055655296e-06, "loss": 0.001, "step": 14960 }, { "epoch": 1.58, "learning_rate": 9.609335727109515e-06, "loss": 0.0009, "step": 14970 }, { "epoch": 1.59, "learning_rate": 9.602154398563735e-06, "loss": 0.0011, "step": 14980 }, { "epoch": 1.59, "learning_rate": 9.594973070017954e-06, "loss": 0.0011, "step": 14990 }, { "epoch": 1.59, "learning_rate": 9.587791741472173e-06, "loss": 0.001, "step": 15000 }, { "epoch": 1.59, "learning_rate": 9.580610412926392e-06, "loss": 0.0012, "step": 15010 }, { "epoch": 1.59, "learning_rate": 9.57342908438061e-06, "loss": 0.0013, "step": 15020 }, { "epoch": 1.59, "learning_rate": 9.56624775583483e-06, "loss": 0.0013, "step": 15030 }, { "epoch": 1.59, "learning_rate": 9.559066427289048e-06, "loss": 0.0011, "step": 15040 }, { "epoch": 1.59, "learning_rate": 9.551885098743269e-06, "loss": 0.0012, "step": 15050 }, { "epoch": 1.59, "learning_rate": 9.544703770197488e-06, "loss": 0.0012, "step": 15060 }, { "epoch": 1.59, "learning_rate": 9.537522441651706e-06, "loss": 0.0012, "step": 15070 }, { "epoch": 1.6, "learning_rate": 9.530341113105925e-06, "loss": 0.0012, "step": 15080 }, { "epoch": 1.6, "learning_rate": 9.523159784560144e-06, "loss": 0.0015, "step": 15090 }, { "epoch": 1.6, "learning_rate": 9.515978456014363e-06, "loss": 0.0012, "step": 15100 }, { "epoch": 1.6, "learning_rate": 9.508797127468582e-06, "loss": 0.0013, "step": 15110 }, { "epoch": 1.6, "learning_rate": 9.501615798922802e-06, "loss": 0.0012, "step": 15120 }, { "epoch": 1.6, "learning_rate": 9.494434470377021e-06, "loss": 0.0013, "step": 15130 }, { "epoch": 1.6, "learning_rate": 9.48725314183124e-06, "loss": 0.0015, "step": 15140 }, { "epoch": 1.6, "learning_rate": 9.480071813285459e-06, "loss": 0.0015, "step": 15150 }, { "epoch": 1.6, "learning_rate": 9.472890484739678e-06, "loss": 0.0012, "step": 15160 }, { "epoch": 1.61, "learning_rate": 9.465709156193896e-06, "loss": 0.0011, "step": 15170 }, { "epoch": 1.61, "learning_rate": 9.458527827648115e-06, "loss": 0.0012, "step": 15180 }, { "epoch": 1.61, "learning_rate": 9.451346499102336e-06, "loss": 0.0013, "step": 15190 }, { "epoch": 1.61, "learning_rate": 9.444165170556555e-06, "loss": 0.001, "step": 15200 }, { "epoch": 1.61, "learning_rate": 9.436983842010773e-06, "loss": 0.0012, "step": 15210 }, { "epoch": 1.61, "learning_rate": 9.429802513464992e-06, "loss": 0.0012, "step": 15220 }, { "epoch": 1.61, "learning_rate": 9.422621184919211e-06, "loss": 0.0012, "step": 15230 }, { "epoch": 1.61, "learning_rate": 9.41543985637343e-06, "loss": 0.0013, "step": 15240 }, { "epoch": 1.61, "learning_rate": 9.408258527827649e-06, "loss": 0.0015, "step": 15250 }, { "epoch": 1.61, "learning_rate": 9.401077199281868e-06, "loss": 0.0013, "step": 15260 }, { "epoch": 1.62, "learning_rate": 9.393895870736086e-06, "loss": 0.0013, "step": 15270 }, { "epoch": 1.62, "learning_rate": 9.386714542190305e-06, "loss": 0.0011, "step": 15280 }, { "epoch": 1.62, "learning_rate": 9.379533213644524e-06, "loss": 0.0013, "step": 15290 }, { "epoch": 1.62, "learning_rate": 9.372351885098745e-06, "loss": 0.0011, "step": 15300 }, { "epoch": 1.62, "learning_rate": 9.365170556552963e-06, "loss": 0.0013, "step": 15310 }, { "epoch": 1.62, "learning_rate": 9.357989228007182e-06, "loss": 0.0014, "step": 15320 }, { "epoch": 1.62, "learning_rate": 9.350807899461401e-06, "loss": 0.0013, "step": 15330 }, { "epoch": 1.62, "learning_rate": 9.34362657091562e-06, "loss": 0.0013, "step": 15340 }, { "epoch": 1.62, "learning_rate": 9.336445242369839e-06, "loss": 0.0011, "step": 15350 }, { "epoch": 1.63, "learning_rate": 9.329263913824058e-06, "loss": 0.0012, "step": 15360 }, { "epoch": 1.63, "learning_rate": 9.322082585278276e-06, "loss": 0.0012, "step": 15370 }, { "epoch": 1.63, "learning_rate": 9.314901256732495e-06, "loss": 0.0012, "step": 15380 }, { "epoch": 1.63, "learning_rate": 9.307719928186714e-06, "loss": 0.0014, "step": 15390 }, { "epoch": 1.63, "learning_rate": 9.300538599640935e-06, "loss": 0.0013, "step": 15400 }, { "epoch": 1.63, "learning_rate": 9.293357271095154e-06, "loss": 0.0016, "step": 15410 }, { "epoch": 1.63, "learning_rate": 9.286175942549372e-06, "loss": 0.0012, "step": 15420 }, { "epoch": 1.63, "learning_rate": 9.278994614003591e-06, "loss": 0.0011, "step": 15430 }, { "epoch": 1.63, "learning_rate": 9.27181328545781e-06, "loss": 0.001, "step": 15440 }, { "epoch": 1.63, "learning_rate": 9.264631956912029e-06, "loss": 0.001, "step": 15450 }, { "epoch": 1.64, "learning_rate": 9.257450628366248e-06, "loss": 0.0013, "step": 15460 }, { "epoch": 1.64, "learning_rate": 9.250269299820468e-06, "loss": 0.0012, "step": 15470 }, { "epoch": 1.64, "learning_rate": 9.243087971274687e-06, "loss": 0.0013, "step": 15480 }, { "epoch": 1.64, "learning_rate": 9.235906642728906e-06, "loss": 0.0011, "step": 15490 }, { "epoch": 1.64, "learning_rate": 9.228725314183125e-06, "loss": 0.001, "step": 15500 }, { "epoch": 1.64, "learning_rate": 9.221543985637344e-06, "loss": 0.0009, "step": 15510 }, { "epoch": 1.64, "learning_rate": 9.214362657091562e-06, "loss": 0.0011, "step": 15520 }, { "epoch": 1.64, "learning_rate": 9.207181328545781e-06, "loss": 0.001, "step": 15530 }, { "epoch": 1.64, "learning_rate": 9.200000000000002e-06, "loss": 0.0014, "step": 15540 }, { "epoch": 1.65, "learning_rate": 9.19281867145422e-06, "loss": 0.0011, "step": 15550 }, { "epoch": 1.65, "learning_rate": 9.18563734290844e-06, "loss": 0.0011, "step": 15560 }, { "epoch": 1.65, "learning_rate": 9.178456014362658e-06, "loss": 0.0012, "step": 15570 }, { "epoch": 1.65, "learning_rate": 9.171274685816877e-06, "loss": 0.0011, "step": 15580 }, { "epoch": 1.65, "learning_rate": 9.164093357271096e-06, "loss": 0.0013, "step": 15590 }, { "epoch": 1.65, "learning_rate": 9.156912028725315e-06, "loss": 0.0014, "step": 15600 }, { "epoch": 1.65, "learning_rate": 9.149730700179534e-06, "loss": 0.0013, "step": 15610 }, { "epoch": 1.65, "learning_rate": 9.142549371633754e-06, "loss": 0.0012, "step": 15620 }, { "epoch": 1.65, "learning_rate": 9.135368043087973e-06, "loss": 0.0013, "step": 15630 }, { "epoch": 1.66, "learning_rate": 9.128186714542192e-06, "loss": 0.0037, "step": 15640 }, { "epoch": 1.66, "learning_rate": 9.122441651705565e-06, "loss": 0.0024, "step": 15650 }, { "epoch": 1.66, "learning_rate": 9.115260323159786e-06, "loss": 0.0014, "step": 15660 }, { "epoch": 1.66, "learning_rate": 9.108078994614005e-06, "loss": 0.0014, "step": 15670 }, { "epoch": 1.66, "learning_rate": 9.100897666068224e-06, "loss": 0.0011, "step": 15680 }, { "epoch": 1.66, "learning_rate": 9.093716337522442e-06, "loss": 0.0013, "step": 15690 }, { "epoch": 1.66, "learning_rate": 9.086535008976661e-06, "loss": 0.0012, "step": 15700 }, { "epoch": 1.66, "learning_rate": 9.07935368043088e-06, "loss": 0.0011, "step": 15710 }, { "epoch": 1.66, "learning_rate": 9.072172351885099e-06, "loss": 0.0013, "step": 15720 }, { "epoch": 1.66, "learning_rate": 9.06499102333932e-06, "loss": 0.0011, "step": 15730 }, { "epoch": 1.67, "learning_rate": 9.057809694793538e-06, "loss": 0.0012, "step": 15740 }, { "epoch": 1.67, "learning_rate": 9.050628366247757e-06, "loss": 0.0013, "step": 15750 }, { "epoch": 1.67, "learning_rate": 9.043447037701976e-06, "loss": 0.001, "step": 15760 }, { "epoch": 1.67, "learning_rate": 9.036265709156195e-06, "loss": 0.0011, "step": 15770 }, { "epoch": 1.67, "learning_rate": 9.029084380610414e-06, "loss": 0.0012, "step": 15780 }, { "epoch": 1.67, "learning_rate": 9.021903052064632e-06, "loss": 0.0015, "step": 15790 }, { "epoch": 1.67, "learning_rate": 9.014721723518853e-06, "loss": 0.0012, "step": 15800 }, { "epoch": 1.67, "learning_rate": 9.007540394973072e-06, "loss": 0.0013, "step": 15810 }, { "epoch": 1.67, "learning_rate": 9.00035906642729e-06, "loss": 0.0014, "step": 15820 }, { "epoch": 1.68, "learning_rate": 8.99317773788151e-06, "loss": 0.0015, "step": 15830 }, { "epoch": 1.68, "learning_rate": 8.985996409335728e-06, "loss": 0.0009, "step": 15840 }, { "epoch": 1.68, "learning_rate": 8.978815080789947e-06, "loss": 0.0011, "step": 15850 }, { "epoch": 1.68, "learning_rate": 8.971633752244166e-06, "loss": 0.0012, "step": 15860 }, { "epoch": 1.68, "learning_rate": 8.964452423698385e-06, "loss": 0.0012, "step": 15870 }, { "epoch": 1.68, "learning_rate": 8.957271095152604e-06, "loss": 0.0012, "step": 15880 }, { "epoch": 1.68, "learning_rate": 8.950089766606822e-06, "loss": 0.0012, "step": 15890 }, { "epoch": 1.68, "learning_rate": 8.942908438061041e-06, "loss": 0.0013, "step": 15900 }, { "epoch": 1.68, "learning_rate": 8.93572710951526e-06, "loss": 0.0013, "step": 15910 }, { "epoch": 1.68, "learning_rate": 8.928545780969479e-06, "loss": 0.0013, "step": 15920 }, { "epoch": 1.69, "learning_rate": 8.9213644524237e-06, "loss": 0.0011, "step": 15930 }, { "epoch": 1.69, "learning_rate": 8.914183123877918e-06, "loss": 0.0011, "step": 15940 }, { "epoch": 1.69, "learning_rate": 8.907001795332137e-06, "loss": 0.0011, "step": 15950 }, { "epoch": 1.69, "learning_rate": 8.899820466786356e-06, "loss": 0.0013, "step": 15960 }, { "epoch": 1.69, "learning_rate": 8.892639138240575e-06, "loss": 0.001, "step": 15970 }, { "epoch": 1.69, "learning_rate": 8.885457809694794e-06, "loss": 0.0012, "step": 15980 }, { "epoch": 1.69, "learning_rate": 8.878276481149012e-06, "loss": 0.0012, "step": 15990 }, { "epoch": 1.69, "learning_rate": 8.871095152603231e-06, "loss": 0.0014, "step": 16000 }, { "epoch": 1.69, "learning_rate": 8.863913824057452e-06, "loss": 0.0012, "step": 16010 }, { "epoch": 1.7, "learning_rate": 8.85673249551167e-06, "loss": 0.0015, "step": 16020 }, { "epoch": 1.7, "learning_rate": 8.84955116696589e-06, "loss": 0.0012, "step": 16030 }, { "epoch": 1.7, "learning_rate": 8.842369838420108e-06, "loss": 0.0015, "step": 16040 }, { "epoch": 1.7, "learning_rate": 8.835188509874327e-06, "loss": 0.0012, "step": 16050 }, { "epoch": 1.7, "learning_rate": 8.828007181328546e-06, "loss": 0.0012, "step": 16060 }, { "epoch": 1.7, "learning_rate": 8.820825852782765e-06, "loss": 0.0013, "step": 16070 }, { "epoch": 1.7, "learning_rate": 8.813644524236985e-06, "loss": 0.0012, "step": 16080 }, { "epoch": 1.7, "learning_rate": 8.806463195691204e-06, "loss": 0.0012, "step": 16090 }, { "epoch": 1.7, "learning_rate": 8.799281867145423e-06, "loss": 0.0013, "step": 16100 }, { "epoch": 1.7, "learning_rate": 8.792100538599642e-06, "loss": 0.0012, "step": 16110 }, { "epoch": 1.71, "learning_rate": 8.78491921005386e-06, "loss": 0.0012, "step": 16120 }, { "epoch": 1.71, "learning_rate": 8.77773788150808e-06, "loss": 0.0013, "step": 16130 }, { "epoch": 1.71, "learning_rate": 8.770556552962298e-06, "loss": 0.0013, "step": 16140 }, { "epoch": 1.71, "learning_rate": 8.763375224416519e-06, "loss": 0.0011, "step": 16150 }, { "epoch": 1.71, "learning_rate": 8.756193895870738e-06, "loss": 0.0011, "step": 16160 }, { "epoch": 1.71, "learning_rate": 8.749012567324957e-06, "loss": 0.0012, "step": 16170 }, { "epoch": 1.71, "learning_rate": 8.741831238779175e-06, "loss": 0.0012, "step": 16180 }, { "epoch": 1.71, "learning_rate": 8.734649910233394e-06, "loss": 0.0011, "step": 16190 }, { "epoch": 1.71, "learning_rate": 8.727468581687613e-06, "loss": 0.0014, "step": 16200 }, { "epoch": 1.72, "learning_rate": 8.720287253141832e-06, "loss": 0.0011, "step": 16210 }, { "epoch": 1.72, "learning_rate": 8.71310592459605e-06, "loss": 0.0012, "step": 16220 }, { "epoch": 1.72, "learning_rate": 8.70592459605027e-06, "loss": 0.001, "step": 16230 }, { "epoch": 1.72, "learning_rate": 8.698743267504488e-06, "loss": 0.0011, "step": 16240 }, { "epoch": 1.72, "learning_rate": 8.691561938958709e-06, "loss": 0.0012, "step": 16250 }, { "epoch": 1.72, "learning_rate": 8.684380610412928e-06, "loss": 0.0012, "step": 16260 }, { "epoch": 1.72, "learning_rate": 8.677199281867147e-06, "loss": 0.0011, "step": 16270 }, { "epoch": 1.72, "learning_rate": 8.670017953321365e-06, "loss": 0.0015, "step": 16280 }, { "epoch": 1.72, "learning_rate": 8.662836624775584e-06, "loss": 0.0011, "step": 16290 }, { "epoch": 1.72, "learning_rate": 8.655655296229803e-06, "loss": 0.0012, "step": 16300 }, { "epoch": 1.73, "learning_rate": 8.648473967684022e-06, "loss": 0.0011, "step": 16310 }, { "epoch": 1.73, "learning_rate": 8.64129263913824e-06, "loss": 0.0008, "step": 16320 }, { "epoch": 1.73, "learning_rate": 8.63411131059246e-06, "loss": 0.0011, "step": 16330 }, { "epoch": 1.73, "learning_rate": 8.626929982046678e-06, "loss": 0.0012, "step": 16340 }, { "epoch": 1.73, "learning_rate": 8.619748653500897e-06, "loss": 0.0014, "step": 16350 }, { "epoch": 1.73, "learning_rate": 8.612567324955118e-06, "loss": 0.0014, "step": 16360 }, { "epoch": 1.73, "learning_rate": 8.605385996409337e-06, "loss": 0.0012, "step": 16370 }, { "epoch": 1.73, "learning_rate": 8.598204667863555e-06, "loss": 0.0014, "step": 16380 }, { "epoch": 1.73, "learning_rate": 8.591023339317774e-06, "loss": 0.0011, "step": 16390 }, { "epoch": 1.74, "learning_rate": 8.583842010771993e-06, "loss": 0.0012, "step": 16400 }, { "epoch": 1.74, "learning_rate": 8.576660682226212e-06, "loss": 0.0014, "step": 16410 }, { "epoch": 1.74, "learning_rate": 8.56947935368043e-06, "loss": 0.0012, "step": 16420 }, { "epoch": 1.74, "learning_rate": 8.562298025134651e-06, "loss": 0.0012, "step": 16430 }, { "epoch": 1.74, "learning_rate": 8.55511669658887e-06, "loss": 0.0011, "step": 16440 }, { "epoch": 1.74, "learning_rate": 8.547935368043089e-06, "loss": 0.0012, "step": 16450 }, { "epoch": 1.74, "learning_rate": 8.540754039497308e-06, "loss": 0.0013, "step": 16460 }, { "epoch": 1.74, "learning_rate": 8.533572710951527e-06, "loss": 0.001, "step": 16470 }, { "epoch": 1.74, "learning_rate": 8.526391382405745e-06, "loss": 0.0012, "step": 16480 }, { "epoch": 1.74, "learning_rate": 8.519210053859964e-06, "loss": 0.0012, "step": 16490 }, { "epoch": 1.75, "learning_rate": 8.512028725314185e-06, "loss": 0.0012, "step": 16500 }, { "epoch": 1.75, "learning_rate": 8.504847396768404e-06, "loss": 0.0013, "step": 16510 }, { "epoch": 1.75, "learning_rate": 8.497666068222622e-06, "loss": 0.0014, "step": 16520 }, { "epoch": 1.75, "learning_rate": 8.490484739676841e-06, "loss": 0.0014, "step": 16530 }, { "epoch": 1.75, "learning_rate": 8.48330341113106e-06, "loss": 0.0013, "step": 16540 }, { "epoch": 1.75, "learning_rate": 8.476122082585279e-06, "loss": 0.0016, "step": 16550 }, { "epoch": 1.75, "learning_rate": 8.468940754039498e-06, "loss": 0.001, "step": 16560 }, { "epoch": 1.75, "learning_rate": 8.461759425493718e-06, "loss": 0.0013, "step": 16570 }, { "epoch": 1.75, "learning_rate": 8.454578096947937e-06, "loss": 0.0011, "step": 16580 }, { "epoch": 1.76, "learning_rate": 8.447396768402156e-06, "loss": 0.001, "step": 16590 }, { "epoch": 1.76, "learning_rate": 8.440215439856375e-06, "loss": 0.0011, "step": 16600 }, { "epoch": 1.76, "learning_rate": 8.433034111310594e-06, "loss": 0.0015, "step": 16610 }, { "epoch": 1.76, "learning_rate": 8.425852782764812e-06, "loss": 0.0012, "step": 16620 }, { "epoch": 1.76, "learning_rate": 8.418671454219031e-06, "loss": 0.0014, "step": 16630 }, { "epoch": 1.76, "learning_rate": 8.41149012567325e-06, "loss": 0.001, "step": 16640 }, { "epoch": 1.76, "learning_rate": 8.404308797127469e-06, "loss": 0.0012, "step": 16650 }, { "epoch": 1.76, "learning_rate": 8.397127468581688e-06, "loss": 0.0012, "step": 16660 }, { "epoch": 1.76, "learning_rate": 8.389946140035907e-06, "loss": 0.001, "step": 16670 }, { "epoch": 1.77, "learning_rate": 8.382764811490125e-06, "loss": 0.0013, "step": 16680 }, { "epoch": 1.77, "learning_rate": 8.375583482944344e-06, "loss": 0.0014, "step": 16690 }, { "epoch": 1.77, "learning_rate": 8.368402154398565e-06, "loss": 0.0014, "step": 16700 }, { "epoch": 1.77, "learning_rate": 8.361220825852784e-06, "loss": 0.0013, "step": 16710 }, { "epoch": 1.77, "learning_rate": 8.354039497307002e-06, "loss": 0.0014, "step": 16720 }, { "epoch": 1.77, "learning_rate": 8.346858168761221e-06, "loss": 0.0012, "step": 16730 }, { "epoch": 1.77, "learning_rate": 8.33967684021544e-06, "loss": 0.0013, "step": 16740 }, { "epoch": 1.77, "learning_rate": 8.332495511669659e-06, "loss": 0.0013, "step": 16750 }, { "epoch": 1.77, "learning_rate": 8.325314183123878e-06, "loss": 0.001, "step": 16760 }, { "epoch": 1.77, "learning_rate": 8.318132854578098e-06, "loss": 0.0014, "step": 16770 }, { "epoch": 1.78, "learning_rate": 8.310951526032317e-06, "loss": 0.0013, "step": 16780 }, { "epoch": 1.78, "learning_rate": 8.303770197486536e-06, "loss": 0.0013, "step": 16790 }, { "epoch": 1.78, "learning_rate": 8.296588868940755e-06, "loss": 0.0015, "step": 16800 }, { "epoch": 1.78, "learning_rate": 8.289407540394974e-06, "loss": 0.0015, "step": 16810 }, { "epoch": 1.78, "learning_rate": 8.282226211849192e-06, "loss": 0.0016, "step": 16820 }, { "epoch": 1.78, "learning_rate": 8.275044883303411e-06, "loss": 0.0011, "step": 16830 }, { "epoch": 1.78, "learning_rate": 8.267863554757632e-06, "loss": 0.0013, "step": 16840 }, { "epoch": 1.78, "learning_rate": 8.26068222621185e-06, "loss": 0.0009, "step": 16850 }, { "epoch": 1.78, "learning_rate": 8.25350089766607e-06, "loss": 0.0012, "step": 16860 }, { "epoch": 1.79, "learning_rate": 8.246319569120288e-06, "loss": 0.001, "step": 16870 }, { "epoch": 1.79, "learning_rate": 8.239138240574507e-06, "loss": 0.0013, "step": 16880 }, { "epoch": 1.79, "learning_rate": 8.231956912028726e-06, "loss": 0.0011, "step": 16890 }, { "epoch": 1.79, "learning_rate": 8.224775583482945e-06, "loss": 0.0015, "step": 16900 }, { "epoch": 1.79, "learning_rate": 8.217594254937165e-06, "loss": 0.0013, "step": 16910 }, { "epoch": 1.79, "learning_rate": 8.210412926391384e-06, "loss": 0.0012, "step": 16920 }, { "epoch": 1.79, "learning_rate": 8.203231597845603e-06, "loss": 0.001, "step": 16930 }, { "epoch": 1.79, "learning_rate": 8.196050269299822e-06, "loss": 0.0013, "step": 16940 }, { "epoch": 1.79, "learning_rate": 8.18886894075404e-06, "loss": 0.0012, "step": 16950 }, { "epoch": 1.79, "learning_rate": 8.18168761220826e-06, "loss": 0.0012, "step": 16960 }, { "epoch": 1.8, "learning_rate": 8.174506283662478e-06, "loss": 0.0013, "step": 16970 }, { "epoch": 1.8, "learning_rate": 8.167324955116697e-06, "loss": 0.0011, "step": 16980 }, { "epoch": 1.8, "learning_rate": 8.160143626570916e-06, "loss": 0.0012, "step": 16990 }, { "epoch": 1.8, "learning_rate": 8.152962298025135e-06, "loss": 0.0015, "step": 17000 }, { "epoch": 1.8, "learning_rate": 8.145780969479354e-06, "loss": 0.0013, "step": 17010 }, { "epoch": 1.8, "learning_rate": 8.138599640933572e-06, "loss": 0.001, "step": 17020 }, { "epoch": 1.8, "learning_rate": 8.131418312387791e-06, "loss": 0.0014, "step": 17030 }, { "epoch": 1.8, "learning_rate": 8.124236983842012e-06, "loss": 0.0012, "step": 17040 }, { "epoch": 1.8, "learning_rate": 8.11705565529623e-06, "loss": 0.0012, "step": 17050 }, { "epoch": 1.81, "learning_rate": 8.10987432675045e-06, "loss": 0.001, "step": 17060 }, { "epoch": 1.81, "learning_rate": 8.102692998204668e-06, "loss": 0.0012, "step": 17070 }, { "epoch": 1.81, "learning_rate": 8.095511669658887e-06, "loss": 0.0012, "step": 17080 }, { "epoch": 1.81, "learning_rate": 8.088330341113106e-06, "loss": 0.0013, "step": 17090 }, { "epoch": 1.81, "learning_rate": 8.081149012567325e-06, "loss": 0.0011, "step": 17100 }, { "epoch": 1.81, "learning_rate": 8.073967684021544e-06, "loss": 0.0012, "step": 17110 }, { "epoch": 1.81, "learning_rate": 8.066786355475764e-06, "loss": 0.0013, "step": 17120 }, { "epoch": 1.81, "learning_rate": 8.059605026929983e-06, "loss": 0.0012, "step": 17130 }, { "epoch": 1.81, "learning_rate": 8.052423698384202e-06, "loss": 0.0013, "step": 17140 }, { "epoch": 1.81, "learning_rate": 8.04524236983842e-06, "loss": 0.0012, "step": 17150 }, { "epoch": 1.82, "learning_rate": 8.03806104129264e-06, "loss": 0.0015, "step": 17160 }, { "epoch": 1.82, "learning_rate": 8.030879712746858e-06, "loss": 0.0012, "step": 17170 }, { "epoch": 1.82, "learning_rate": 8.023698384201077e-06, "loss": 0.0012, "step": 17180 }, { "epoch": 1.82, "learning_rate": 8.016517055655298e-06, "loss": 0.0011, "step": 17190 }, { "epoch": 1.82, "learning_rate": 8.009335727109517e-06, "loss": 0.0013, "step": 17200 }, { "epoch": 1.82, "learning_rate": 8.002154398563735e-06, "loss": 0.001, "step": 17210 }, { "epoch": 1.82, "learning_rate": 7.994973070017954e-06, "loss": 0.0011, "step": 17220 }, { "epoch": 1.82, "learning_rate": 7.987791741472173e-06, "loss": 0.0012, "step": 17230 }, { "epoch": 1.82, "learning_rate": 7.980610412926392e-06, "loss": 0.001, "step": 17240 }, { "epoch": 1.83, "learning_rate": 7.97342908438061e-06, "loss": 0.0011, "step": 17250 }, { "epoch": 1.83, "learning_rate": 7.966247755834831e-06, "loss": 0.0014, "step": 17260 }, { "epoch": 1.83, "learning_rate": 7.95906642728905e-06, "loss": 0.0013, "step": 17270 }, { "epoch": 1.83, "learning_rate": 7.951885098743269e-06, "loss": 0.0013, "step": 17280 }, { "epoch": 1.83, "learning_rate": 7.944703770197488e-06, "loss": 0.0011, "step": 17290 }, { "epoch": 1.83, "learning_rate": 7.937522441651707e-06, "loss": 0.0012, "step": 17300 }, { "epoch": 1.83, "learning_rate": 7.930341113105925e-06, "loss": 0.0013, "step": 17310 }, { "epoch": 1.83, "learning_rate": 7.923159784560144e-06, "loss": 0.0013, "step": 17320 }, { "epoch": 1.83, "learning_rate": 7.915978456014363e-06, "loss": 0.0011, "step": 17330 }, { "epoch": 1.83, "learning_rate": 7.908797127468582e-06, "loss": 0.0013, "step": 17340 }, { "epoch": 1.84, "learning_rate": 7.9016157989228e-06, "loss": 0.0015, "step": 17350 }, { "epoch": 1.84, "learning_rate": 7.894434470377021e-06, "loss": 0.0012, "step": 17360 }, { "epoch": 1.84, "learning_rate": 7.88725314183124e-06, "loss": 0.0012, "step": 17370 }, { "epoch": 1.84, "learning_rate": 7.880071813285459e-06, "loss": 0.0011, "step": 17380 }, { "epoch": 1.84, "learning_rate": 7.872890484739678e-06, "loss": 0.001, "step": 17390 }, { "epoch": 1.84, "learning_rate": 7.865709156193897e-06, "loss": 0.0011, "step": 17400 }, { "epoch": 1.84, "learning_rate": 7.858527827648115e-06, "loss": 0.0013, "step": 17410 }, { "epoch": 1.84, "learning_rate": 7.851346499102334e-06, "loss": 0.0011, "step": 17420 }, { "epoch": 1.84, "learning_rate": 7.844165170556553e-06, "loss": 0.0014, "step": 17430 }, { "epoch": 1.85, "learning_rate": 7.836983842010772e-06, "loss": 0.0014, "step": 17440 }, { "epoch": 1.85, "learning_rate": 7.82980251346499e-06, "loss": 0.0011, "step": 17450 }, { "epoch": 1.85, "learning_rate": 7.82262118491921e-06, "loss": 0.0011, "step": 17460 }, { "epoch": 1.85, "learning_rate": 7.81543985637343e-06, "loss": 0.0013, "step": 17470 }, { "epoch": 1.85, "learning_rate": 7.808258527827649e-06, "loss": 0.0012, "step": 17480 }, { "epoch": 1.85, "learning_rate": 7.801077199281868e-06, "loss": 0.0012, "step": 17490 }, { "epoch": 1.85, "learning_rate": 7.793895870736087e-06, "loss": 0.0013, "step": 17500 }, { "epoch": 1.85, "learning_rate": 7.786714542190305e-06, "loss": 0.001, "step": 17510 }, { "epoch": 1.85, "learning_rate": 7.779533213644524e-06, "loss": 0.0011, "step": 17520 }, { "epoch": 1.86, "learning_rate": 7.772351885098743e-06, "loss": 0.0013, "step": 17530 }, { "epoch": 1.86, "learning_rate": 7.765170556552964e-06, "loss": 0.0014, "step": 17540 }, { "epoch": 1.86, "learning_rate": 7.757989228007182e-06, "loss": 0.0012, "step": 17550 }, { "epoch": 1.86, "learning_rate": 7.750807899461401e-06, "loss": 0.0013, "step": 17560 }, { "epoch": 1.86, "learning_rate": 7.74362657091562e-06, "loss": 0.0011, "step": 17570 }, { "epoch": 1.86, "learning_rate": 7.736445242369839e-06, "loss": 0.001, "step": 17580 }, { "epoch": 1.86, "learning_rate": 7.729263913824058e-06, "loss": 0.0012, "step": 17590 }, { "epoch": 1.86, "learning_rate": 7.722082585278277e-06, "loss": 0.001, "step": 17600 }, { "epoch": 1.86, "learning_rate": 7.714901256732497e-06, "loss": 0.0012, "step": 17610 }, { "epoch": 1.86, "learning_rate": 7.707719928186716e-06, "loss": 0.0012, "step": 17620 }, { "epoch": 1.87, "learning_rate": 7.700538599640935e-06, "loss": 0.0012, "step": 17630 }, { "epoch": 1.87, "learning_rate": 7.693357271095154e-06, "loss": 0.0012, "step": 17640 }, { "epoch": 1.87, "learning_rate": 7.686175942549372e-06, "loss": 0.0011, "step": 17650 }, { "epoch": 1.87, "learning_rate": 7.678994614003591e-06, "loss": 0.0013, "step": 17660 }, { "epoch": 1.87, "learning_rate": 7.67181328545781e-06, "loss": 0.0011, "step": 17670 }, { "epoch": 1.87, "learning_rate": 7.66463195691203e-06, "loss": 0.0011, "step": 17680 }, { "epoch": 1.87, "learning_rate": 7.65745062836625e-06, "loss": 0.0013, "step": 17690 }, { "epoch": 1.87, "learning_rate": 7.650269299820468e-06, "loss": 0.0012, "step": 17700 }, { "epoch": 1.87, "learning_rate": 7.643087971274687e-06, "loss": 0.0012, "step": 17710 }, { "epoch": 1.88, "learning_rate": 7.635906642728906e-06, "loss": 0.0013, "step": 17720 }, { "epoch": 1.88, "learning_rate": 7.628725314183124e-06, "loss": 0.0015, "step": 17730 }, { "epoch": 1.88, "learning_rate": 7.621543985637343e-06, "loss": 0.0009, "step": 17740 }, { "epoch": 1.88, "learning_rate": 7.614362657091563e-06, "loss": 0.001, "step": 17750 }, { "epoch": 1.88, "learning_rate": 7.607181328545782e-06, "loss": 0.0012, "step": 17760 }, { "epoch": 1.88, "learning_rate": 7.600000000000001e-06, "loss": 0.0013, "step": 17770 }, { "epoch": 1.88, "learning_rate": 7.59281867145422e-06, "loss": 0.0013, "step": 17780 }, { "epoch": 1.88, "learning_rate": 7.585637342908439e-06, "loss": 0.0011, "step": 17790 }, { "epoch": 1.88, "learning_rate": 7.578456014362657e-06, "loss": 0.0011, "step": 17800 }, { "epoch": 1.88, "learning_rate": 7.571274685816876e-06, "loss": 0.0014, "step": 17810 }, { "epoch": 1.89, "learning_rate": 7.564093357271096e-06, "loss": 0.0013, "step": 17820 }, { "epoch": 1.89, "learning_rate": 7.556912028725315e-06, "loss": 0.0014, "step": 17830 }, { "epoch": 1.89, "learning_rate": 7.549730700179534e-06, "loss": 0.0011, "step": 17840 }, { "epoch": 1.89, "learning_rate": 7.5425493716337524e-06, "loss": 0.0014, "step": 17850 }, { "epoch": 1.89, "learning_rate": 7.535368043087971e-06, "loss": 0.0013, "step": 17860 }, { "epoch": 1.89, "learning_rate": 7.528186714542191e-06, "loss": 0.0014, "step": 17870 }, { "epoch": 1.89, "learning_rate": 7.521005385996411e-06, "loss": 0.0013, "step": 17880 }, { "epoch": 1.89, "learning_rate": 7.5138240574506294e-06, "loss": 0.0015, "step": 17890 }, { "epoch": 1.89, "learning_rate": 7.506642728904848e-06, "loss": 0.0013, "step": 17900 }, { "epoch": 1.9, "learning_rate": 7.499461400359067e-06, "loss": 0.0011, "step": 17910 }, { "epoch": 1.9, "learning_rate": 7.492280071813286e-06, "loss": 0.0014, "step": 17920 }, { "epoch": 1.9, "learning_rate": 7.485098743267505e-06, "loss": 0.0012, "step": 17930 }, { "epoch": 1.9, "learning_rate": 7.477917414721724e-06, "loss": 0.0013, "step": 17940 }, { "epoch": 1.9, "learning_rate": 7.470736086175943e-06, "loss": 0.001, "step": 17950 }, { "epoch": 1.9, "learning_rate": 7.463554757630163e-06, "loss": 0.0015, "step": 17960 }, { "epoch": 1.9, "learning_rate": 7.456373429084382e-06, "loss": 0.0011, "step": 17970 }, { "epoch": 1.9, "learning_rate": 7.449192100538601e-06, "loss": 0.0013, "step": 17980 }, { "epoch": 1.9, "learning_rate": 7.4420107719928195e-06, "loss": 0.0011, "step": 17990 }, { "epoch": 1.9, "learning_rate": 7.434829443447038e-06, "loss": 0.0013, "step": 18000 }, { "epoch": 1.91, "learning_rate": 7.427648114901257e-06, "loss": 0.0014, "step": 18010 }, { "epoch": 1.91, "learning_rate": 7.420466786355477e-06, "loss": 0.0012, "step": 18020 }, { "epoch": 1.91, "learning_rate": 7.413285457809696e-06, "loss": 0.0012, "step": 18030 }, { "epoch": 1.91, "learning_rate": 7.4061041292639145e-06, "loss": 0.001, "step": 18040 }, { "epoch": 1.91, "learning_rate": 7.398922800718133e-06, "loss": 0.0013, "step": 18050 }, { "epoch": 1.91, "learning_rate": 7.391741472172352e-06, "loss": 0.0012, "step": 18060 }, { "epoch": 1.91, "learning_rate": 7.384560143626571e-06, "loss": 0.0014, "step": 18070 }, { "epoch": 1.91, "learning_rate": 7.37737881508079e-06, "loss": 0.0011, "step": 18080 }, { "epoch": 1.91, "learning_rate": 7.37019748653501e-06, "loss": 0.0011, "step": 18090 }, { "epoch": 1.92, "learning_rate": 7.363016157989229e-06, "loss": 0.0012, "step": 18100 }, { "epoch": 1.92, "learning_rate": 7.355834829443448e-06, "loss": 0.0011, "step": 18110 }, { "epoch": 1.92, "learning_rate": 7.348653500897667e-06, "loss": 0.0013, "step": 18120 }, { "epoch": 1.92, "learning_rate": 7.341472172351886e-06, "loss": 0.0012, "step": 18130 }, { "epoch": 1.92, "learning_rate": 7.3342908438061045e-06, "loss": 0.0013, "step": 18140 }, { "epoch": 1.92, "learning_rate": 7.327109515260323e-06, "loss": 0.0012, "step": 18150 }, { "epoch": 1.92, "learning_rate": 7.319928186714543e-06, "loss": 0.001, "step": 18160 }, { "epoch": 1.92, "learning_rate": 7.312746858168762e-06, "loss": 0.0012, "step": 18170 }, { "epoch": 1.92, "learning_rate": 7.305565529622981e-06, "loss": 0.0014, "step": 18180 }, { "epoch": 1.92, "learning_rate": 7.2983842010772e-06, "loss": 0.0011, "step": 18190 }, { "epoch": 1.93, "learning_rate": 7.291202872531419e-06, "loss": 0.0013, "step": 18200 }, { "epoch": 1.93, "learning_rate": 7.284021543985638e-06, "loss": 0.0014, "step": 18210 }, { "epoch": 1.93, "learning_rate": 7.276840215439857e-06, "loss": 0.0012, "step": 18220 }, { "epoch": 1.93, "learning_rate": 7.2696588868940765e-06, "loss": 0.0012, "step": 18230 }, { "epoch": 1.93, "learning_rate": 7.262477558348295e-06, "loss": 0.0012, "step": 18240 }, { "epoch": 1.93, "learning_rate": 7.255296229802514e-06, "loss": 0.0013, "step": 18250 }, { "epoch": 1.93, "learning_rate": 7.248114901256733e-06, "loss": 0.001, "step": 18260 }, { "epoch": 1.93, "learning_rate": 7.240933572710952e-06, "loss": 0.001, "step": 18270 }, { "epoch": 1.93, "learning_rate": 7.233752244165171e-06, "loss": 0.0011, "step": 18280 }, { "epoch": 1.94, "learning_rate": 7.2265709156193895e-06, "loss": 0.0012, "step": 18290 }, { "epoch": 1.94, "learning_rate": 7.21938958707361e-06, "loss": 0.001, "step": 18300 }, { "epoch": 1.94, "learning_rate": 7.212208258527829e-06, "loss": 0.001, "step": 18310 }, { "epoch": 1.94, "learning_rate": 7.205026929982048e-06, "loss": 0.0011, "step": 18320 }, { "epoch": 1.94, "learning_rate": 7.1978456014362665e-06, "loss": 0.0014, "step": 18330 }, { "epoch": 1.94, "learning_rate": 7.190664272890485e-06, "loss": 0.0012, "step": 18340 }, { "epoch": 1.94, "learning_rate": 7.183482944344704e-06, "loss": 0.0012, "step": 18350 }, { "epoch": 1.94, "learning_rate": 7.176301615798923e-06, "loss": 0.001, "step": 18360 }, { "epoch": 1.94, "learning_rate": 7.169120287253143e-06, "loss": 0.0013, "step": 18370 }, { "epoch": 1.94, "learning_rate": 7.1619389587073615e-06, "loss": 0.0013, "step": 18380 }, { "epoch": 1.95, "learning_rate": 7.15475763016158e-06, "loss": 0.0012, "step": 18390 }, { "epoch": 1.95, "learning_rate": 7.147576301615799e-06, "loss": 0.0012, "step": 18400 }, { "epoch": 1.95, "learning_rate": 7.140394973070018e-06, "loss": 0.0011, "step": 18410 }, { "epoch": 1.95, "learning_rate": 7.133213644524237e-06, "loss": 0.0011, "step": 18420 }, { "epoch": 1.95, "learning_rate": 7.1260323159784565e-06, "loss": 0.0012, "step": 18430 }, { "epoch": 1.95, "learning_rate": 7.118850987432676e-06, "loss": 0.0012, "step": 18440 }, { "epoch": 1.95, "learning_rate": 7.111669658886895e-06, "loss": 0.0011, "step": 18450 }, { "epoch": 1.95, "learning_rate": 7.104488330341114e-06, "loss": 0.0013, "step": 18460 }, { "epoch": 1.95, "learning_rate": 7.097307001795333e-06, "loss": 0.0011, "step": 18470 }, { "epoch": 1.96, "learning_rate": 7.0901256732495515e-06, "loss": 0.0011, "step": 18480 }, { "epoch": 1.96, "learning_rate": 7.08294434470377e-06, "loss": 0.0012, "step": 18490 }, { "epoch": 1.96, "learning_rate": 7.075763016157989e-06, "loss": 0.0012, "step": 18500 }, { "epoch": 1.96, "learning_rate": 7.06858168761221e-06, "loss": 0.0014, "step": 18510 }, { "epoch": 1.96, "learning_rate": 7.0614003590664285e-06, "loss": 0.0013, "step": 18520 }, { "epoch": 1.96, "learning_rate": 7.054219030520647e-06, "loss": 0.0013, "step": 18530 }, { "epoch": 1.96, "learning_rate": 7.047037701974866e-06, "loss": 0.0013, "step": 18540 }, { "epoch": 1.96, "learning_rate": 7.039856373429085e-06, "loss": 0.0012, "step": 18550 }, { "epoch": 1.96, "learning_rate": 7.032675044883304e-06, "loss": 0.0011, "step": 18560 }, { "epoch": 1.97, "learning_rate": 7.025493716337523e-06, "loss": 0.0012, "step": 18570 }, { "epoch": 1.97, "learning_rate": 7.018312387791742e-06, "loss": 0.0014, "step": 18580 }, { "epoch": 1.97, "learning_rate": 7.011131059245961e-06, "loss": 0.0014, "step": 18590 }, { "epoch": 1.97, "learning_rate": 7.00394973070018e-06, "loss": 0.0015, "step": 18600 }, { "epoch": 1.97, "learning_rate": 6.996768402154399e-06, "loss": 0.001, "step": 18610 }, { "epoch": 1.97, "learning_rate": 6.989587073608618e-06, "loss": 0.0014, "step": 18620 }, { "epoch": 1.97, "learning_rate": 6.9824057450628365e-06, "loss": 0.0014, "step": 18630 }, { "epoch": 1.97, "learning_rate": 6.975224416517055e-06, "loss": 0.0012, "step": 18640 }, { "epoch": 1.97, "learning_rate": 6.968043087971276e-06, "loss": 0.0013, "step": 18650 }, { "epoch": 1.97, "learning_rate": 6.960861759425495e-06, "loss": 0.0011, "step": 18660 }, { "epoch": 1.98, "learning_rate": 6.9536804308797136e-06, "loss": 0.0012, "step": 18670 }, { "epoch": 1.98, "learning_rate": 6.946499102333932e-06, "loss": 0.0012, "step": 18680 }, { "epoch": 1.98, "learning_rate": 6.939317773788151e-06, "loss": 0.0012, "step": 18690 }, { "epoch": 1.98, "learning_rate": 6.93213644524237e-06, "loss": 0.0013, "step": 18700 }, { "epoch": 1.98, "learning_rate": 6.924955116696589e-06, "loss": 0.0015, "step": 18710 }, { "epoch": 1.98, "learning_rate": 6.9177737881508086e-06, "loss": 0.0013, "step": 18720 }, { "epoch": 1.98, "learning_rate": 6.910592459605027e-06, "loss": 0.0012, "step": 18730 }, { "epoch": 1.98, "learning_rate": 6.903411131059246e-06, "loss": 0.0013, "step": 18740 }, { "epoch": 1.98, "learning_rate": 6.896229802513466e-06, "loss": 0.0013, "step": 18750 }, { "epoch": 1.99, "learning_rate": 6.889048473967685e-06, "loss": 0.0012, "step": 18760 }, { "epoch": 1.99, "learning_rate": 6.8818671454219036e-06, "loss": 0.0012, "step": 18770 }, { "epoch": 1.99, "learning_rate": 6.874685816876122e-06, "loss": 0.0013, "step": 18780 }, { "epoch": 1.99, "learning_rate": 6.867504488330342e-06, "loss": 0.0013, "step": 18790 }, { "epoch": 1.99, "learning_rate": 6.860323159784561e-06, "loss": 0.0012, "step": 18800 }, { "epoch": 1.99, "learning_rate": 6.85314183123878e-06, "loss": 0.0013, "step": 18810 }, { "epoch": 1.99, "learning_rate": 6.845960502692999e-06, "loss": 0.0013, "step": 18820 }, { "epoch": 1.99, "learning_rate": 6.838779174147217e-06, "loss": 0.0013, "step": 18830 }, { "epoch": 1.99, "learning_rate": 6.831597845601436e-06, "loss": 0.0012, "step": 18840 }, { "epoch": 1.99, "learning_rate": 6.824416517055655e-06, "loss": 0.001, "step": 18850 }, { "epoch": 2.0, "learning_rate": 6.817235188509876e-06, "loss": 0.0014, "step": 18860 }, { "epoch": 2.0, "learning_rate": 6.8100538599640944e-06, "loss": 0.0012, "step": 18870 }, { "epoch": 2.0, "learning_rate": 6.802872531418313e-06, "loss": 0.0011, "step": 18880 }, { "epoch": 2.0, "learning_rate": 6.795691202872532e-06, "loss": 0.0012, "step": 18890 }, { "epoch": 2.0, "learning_rate": 6.788509874326751e-06, "loss": 0.0013, "step": 18900 }, { "epoch": 2.0, "eval_loss": 0.0012332267360761762, "eval_runtime": 294.1507, "eval_samples_per_second": 171.341, "eval_steps_per_second": 21.418, "step": 18900 }, { "epoch": 2.0, "learning_rate": 6.78132854578097e-06, "loss": 0.0012, "step": 18910 }, { "epoch": 2.0, "learning_rate": 6.774147217235189e-06, "loss": 0.0013, "step": 18920 }, { "epoch": 2.0, "learning_rate": 6.766965888689408e-06, "loss": 0.0009, "step": 18930 }, { "epoch": 2.0, "learning_rate": 6.759784560143627e-06, "loss": 0.0011, "step": 18940 }, { "epoch": 2.01, "learning_rate": 6.752603231597846e-06, "loss": 0.0015, "step": 18950 }, { "epoch": 2.01, "learning_rate": 6.745421903052065e-06, "loss": 0.0014, "step": 18960 }, { "epoch": 2.01, "learning_rate": 6.738240574506284e-06, "loss": 0.0014, "step": 18970 }, { "epoch": 2.01, "learning_rate": 6.731059245960503e-06, "loss": 0.001, "step": 18980 }, { "epoch": 2.01, "learning_rate": 6.723877917414722e-06, "loss": 0.0015, "step": 18990 }, { "epoch": 2.01, "learning_rate": 6.716696588868942e-06, "loss": 0.0011, "step": 19000 }, { "epoch": 2.01, "learning_rate": 6.709515260323161e-06, "loss": 0.001, "step": 19010 }, { "epoch": 2.01, "learning_rate": 6.7023339317773794e-06, "loss": 0.0011, "step": 19020 }, { "epoch": 2.01, "learning_rate": 6.695152603231598e-06, "loss": 0.0012, "step": 19030 }, { "epoch": 2.01, "learning_rate": 6.687971274685817e-06, "loss": 0.0011, "step": 19040 }, { "epoch": 2.02, "learning_rate": 6.680789946140036e-06, "loss": 0.0014, "step": 19050 }, { "epoch": 2.02, "learning_rate": 6.673608617594255e-06, "loss": 0.0012, "step": 19060 }, { "epoch": 2.02, "learning_rate": 6.666427289048475e-06, "loss": 0.0016, "step": 19070 }, { "epoch": 2.02, "learning_rate": 6.659245960502694e-06, "loss": 0.0011, "step": 19080 }, { "epoch": 2.02, "learning_rate": 6.652064631956913e-06, "loss": 0.0012, "step": 19090 }, { "epoch": 2.02, "learning_rate": 6.644883303411132e-06, "loss": 0.0011, "step": 19100 }, { "epoch": 2.02, "learning_rate": 6.637701974865351e-06, "loss": 0.0015, "step": 19110 }, { "epoch": 2.02, "learning_rate": 6.6305206463195695e-06, "loss": 0.0011, "step": 19120 }, { "epoch": 2.02, "learning_rate": 6.623339317773789e-06, "loss": 0.0012, "step": 19130 }, { "epoch": 2.03, "learning_rate": 6.616157989228008e-06, "loss": 0.0013, "step": 19140 }, { "epoch": 2.03, "learning_rate": 6.608976660682227e-06, "loss": 0.0014, "step": 19150 }, { "epoch": 2.03, "learning_rate": 6.601795332136446e-06, "loss": 0.0012, "step": 19160 }, { "epoch": 2.03, "learning_rate": 6.5946140035906645e-06, "loss": 0.0012, "step": 19170 }, { "epoch": 2.03, "learning_rate": 6.587432675044883e-06, "loss": 0.0012, "step": 19180 }, { "epoch": 2.03, "learning_rate": 6.580251346499102e-06, "loss": 0.0012, "step": 19190 }, { "epoch": 2.03, "learning_rate": 6.573070017953323e-06, "loss": 0.0014, "step": 19200 }, { "epoch": 2.03, "learning_rate": 6.5658886894075415e-06, "loss": 0.0011, "step": 19210 }, { "epoch": 2.03, "learning_rate": 6.55870736086176e-06, "loss": 0.0012, "step": 19220 }, { "epoch": 2.03, "learning_rate": 6.551526032315979e-06, "loss": 0.0013, "step": 19230 }, { "epoch": 2.04, "learning_rate": 6.544344703770198e-06, "loss": 0.0014, "step": 19240 }, { "epoch": 2.04, "learning_rate": 6.537163375224417e-06, "loss": 0.0012, "step": 19250 }, { "epoch": 2.04, "learning_rate": 6.529982046678636e-06, "loss": 0.0013, "step": 19260 }, { "epoch": 2.04, "learning_rate": 6.522800718132855e-06, "loss": 0.0014, "step": 19270 }, { "epoch": 2.04, "learning_rate": 6.515619389587074e-06, "loss": 0.0012, "step": 19280 }, { "epoch": 2.04, "learning_rate": 6.508438061041293e-06, "loss": 0.0012, "step": 19290 }, { "epoch": 2.04, "learning_rate": 6.501256732495513e-06, "loss": 0.0013, "step": 19300 }, { "epoch": 2.04, "learning_rate": 6.4940754039497315e-06, "loss": 0.0012, "step": 19310 }, { "epoch": 2.04, "learning_rate": 6.48689407540395e-06, "loss": 0.0012, "step": 19320 }, { "epoch": 2.05, "learning_rate": 6.479712746858169e-06, "loss": 0.0015, "step": 19330 }, { "epoch": 2.05, "learning_rate": 6.472531418312389e-06, "loss": 0.0011, "step": 19340 }, { "epoch": 2.05, "learning_rate": 6.465350089766608e-06, "loss": 0.0012, "step": 19350 }, { "epoch": 2.05, "learning_rate": 6.4581687612208265e-06, "loss": 0.0011, "step": 19360 }, { "epoch": 2.05, "learning_rate": 6.450987432675045e-06, "loss": 0.0012, "step": 19370 }, { "epoch": 2.05, "learning_rate": 6.443806104129264e-06, "loss": 0.0011, "step": 19380 }, { "epoch": 2.05, "learning_rate": 6.436624775583483e-06, "loss": 0.0011, "step": 19390 }, { "epoch": 2.05, "learning_rate": 6.429443447037702e-06, "loss": 0.0013, "step": 19400 }, { "epoch": 2.05, "learning_rate": 6.422262118491922e-06, "loss": 0.0012, "step": 19410 }, { "epoch": 2.06, "learning_rate": 6.415080789946141e-06, "loss": 0.0011, "step": 19420 }, { "epoch": 2.06, "learning_rate": 6.40789946140036e-06, "loss": 0.0011, "step": 19430 }, { "epoch": 2.06, "learning_rate": 6.400718132854579e-06, "loss": 0.0008, "step": 19440 }, { "epoch": 2.06, "learning_rate": 6.393536804308798e-06, "loss": 0.0014, "step": 19450 }, { "epoch": 2.06, "learning_rate": 6.3863554757630165e-06, "loss": 0.0012, "step": 19460 }, { "epoch": 2.06, "learning_rate": 6.379174147217235e-06, "loss": 0.001, "step": 19470 }, { "epoch": 2.06, "learning_rate": 6.371992818671455e-06, "loss": 0.001, "step": 19480 }, { "epoch": 2.06, "learning_rate": 6.364811490125674e-06, "loss": 0.0012, "step": 19490 }, { "epoch": 2.06, "learning_rate": 6.357630161579893e-06, "loss": 0.0013, "step": 19500 }, { "epoch": 2.06, "learning_rate": 6.3504488330341115e-06, "loss": 0.001, "step": 19510 }, { "epoch": 2.07, "learning_rate": 6.34326750448833e-06, "loss": 0.001, "step": 19520 }, { "epoch": 2.07, "learning_rate": 6.33608617594255e-06, "loss": 0.0014, "step": 19530 }, { "epoch": 2.07, "learning_rate": 6.328904847396769e-06, "loss": 0.0013, "step": 19540 }, { "epoch": 2.07, "learning_rate": 6.3217235188509885e-06, "loss": 0.0015, "step": 19550 }, { "epoch": 2.07, "learning_rate": 6.314542190305207e-06, "loss": 0.0012, "step": 19560 }, { "epoch": 2.07, "learning_rate": 6.307360861759426e-06, "loss": 0.0012, "step": 19570 }, { "epoch": 2.07, "learning_rate": 6.300179533213645e-06, "loss": 0.0011, "step": 19580 }, { "epoch": 2.07, "learning_rate": 6.292998204667864e-06, "loss": 0.001, "step": 19590 }, { "epoch": 2.07, "learning_rate": 6.285816876122083e-06, "loss": 0.0013, "step": 19600 }, { "epoch": 2.08, "learning_rate": 6.2786355475763015e-06, "loss": 0.0012, "step": 19610 }, { "epoch": 2.08, "learning_rate": 6.271454219030522e-06, "loss": 0.0012, "step": 19620 }, { "epoch": 2.08, "learning_rate": 6.264272890484741e-06, "loss": 0.0012, "step": 19630 }, { "epoch": 2.08, "learning_rate": 6.25709156193896e-06, "loss": 0.0015, "step": 19640 }, { "epoch": 2.08, "learning_rate": 6.2499102333931785e-06, "loss": 0.0012, "step": 19650 }, { "epoch": 2.08, "learning_rate": 6.242728904847397e-06, "loss": 0.0012, "step": 19660 }, { "epoch": 2.08, "learning_rate": 6.235547576301616e-06, "loss": 0.0011, "step": 19670 }, { "epoch": 2.08, "learning_rate": 6.228366247755835e-06, "loss": 0.0012, "step": 19680 }, { "epoch": 2.08, "learning_rate": 6.221184919210055e-06, "loss": 0.001, "step": 19690 }, { "epoch": 2.08, "learning_rate": 6.2140035906642735e-06, "loss": 0.0013, "step": 19700 }, { "epoch": 2.09, "learning_rate": 6.206822262118492e-06, "loss": 0.001, "step": 19710 }, { "epoch": 2.09, "learning_rate": 6.199640933572711e-06, "loss": 0.0009, "step": 19720 }, { "epoch": 2.09, "learning_rate": 6.19245960502693e-06, "loss": 0.0013, "step": 19730 }, { "epoch": 2.09, "learning_rate": 6.185278276481149e-06, "loss": 0.0012, "step": 19740 }, { "epoch": 2.09, "learning_rate": 6.178096947935368e-06, "loss": 0.0014, "step": 19750 }, { "epoch": 2.09, "learning_rate": 6.170915619389588e-06, "loss": 0.0013, "step": 19760 }, { "epoch": 2.09, "learning_rate": 6.163734290843807e-06, "loss": 0.0011, "step": 19770 }, { "epoch": 2.09, "learning_rate": 6.156552962298026e-06, "loss": 0.0013, "step": 19780 }, { "epoch": 2.09, "learning_rate": 6.149371633752245e-06, "loss": 0.0012, "step": 19790 }, { "epoch": 2.1, "learning_rate": 6.1421903052064636e-06, "loss": 0.0012, "step": 19800 }, { "epoch": 2.1, "learning_rate": 6.135008976660682e-06, "loss": 0.0011, "step": 19810 }, { "epoch": 2.1, "learning_rate": 6.127827648114901e-06, "loss": 0.0011, "step": 19820 }, { "epoch": 2.1, "learning_rate": 6.120646319569121e-06, "loss": 0.0013, "step": 19830 }, { "epoch": 2.1, "learning_rate": 6.11346499102334e-06, "loss": 0.0012, "step": 19840 }, { "epoch": 2.1, "learning_rate": 6.106283662477559e-06, "loss": 0.0012, "step": 19850 }, { "epoch": 2.1, "learning_rate": 6.099102333931778e-06, "loss": 0.0014, "step": 19860 }, { "epoch": 2.1, "learning_rate": 6.091921005385997e-06, "loss": 0.0011, "step": 19870 }, { "epoch": 2.1, "learning_rate": 6.084739676840216e-06, "loss": 0.0013, "step": 19880 }, { "epoch": 2.1, "learning_rate": 6.077558348294435e-06, "loss": 0.001, "step": 19890 }, { "epoch": 2.11, "learning_rate": 6.070377019748654e-06, "loss": 0.0012, "step": 19900 }, { "epoch": 2.11, "learning_rate": 6.063195691202873e-06, "loss": 0.0012, "step": 19910 }, { "epoch": 2.11, "learning_rate": 6.056014362657092e-06, "loss": 0.0011, "step": 19920 }, { "epoch": 2.11, "learning_rate": 6.048833034111311e-06, "loss": 0.0012, "step": 19930 }, { "epoch": 2.11, "learning_rate": 6.04165170556553e-06, "loss": 0.0012, "step": 19940 }, { "epoch": 2.11, "learning_rate": 6.0344703770197486e-06, "loss": 0.0011, "step": 19950 }, { "epoch": 2.11, "learning_rate": 6.027289048473967e-06, "loss": 0.0014, "step": 19960 }, { "epoch": 2.11, "learning_rate": 6.020107719928188e-06, "loss": 0.0013, "step": 19970 }, { "epoch": 2.11, "learning_rate": 6.012926391382407e-06, "loss": 0.0011, "step": 19980 }, { "epoch": 2.12, "learning_rate": 6.005745062836626e-06, "loss": 0.0013, "step": 19990 }, { "epoch": 2.12, "learning_rate": 5.9985637342908444e-06, "loss": 0.0011, "step": 20000 }, { "epoch": 2.12, "learning_rate": 5.991382405745063e-06, "loss": 0.0014, "step": 20010 }, { "epoch": 2.12, "learning_rate": 5.984201077199282e-06, "loss": 0.0013, "step": 20020 }, { "epoch": 2.12, "learning_rate": 5.977019748653501e-06, "loss": 0.0011, "step": 20030 }, { "epoch": 2.12, "learning_rate": 5.969838420107721e-06, "loss": 0.0011, "step": 20040 }, { "epoch": 2.12, "learning_rate": 5.9626570915619394e-06, "loss": 0.0011, "step": 20050 }, { "epoch": 2.12, "learning_rate": 5.955475763016158e-06, "loss": 0.0012, "step": 20060 }, { "epoch": 2.12, "learning_rate": 5.948294434470377e-06, "loss": 0.0012, "step": 20070 }, { "epoch": 2.12, "learning_rate": 5.941113105924596e-06, "loss": 0.0016, "step": 20080 }, { "epoch": 2.13, "learning_rate": 5.933931777378816e-06, "loss": 0.0013, "step": 20090 }, { "epoch": 2.13, "learning_rate": 5.9267504488330344e-06, "loss": 0.0013, "step": 20100 }, { "epoch": 2.13, "learning_rate": 5.919569120287254e-06, "loss": 0.0011, "step": 20110 }, { "epoch": 2.13, "learning_rate": 5.912387791741473e-06, "loss": 0.0013, "step": 20120 }, { "epoch": 2.13, "learning_rate": 5.905206463195692e-06, "loss": 0.0015, "step": 20130 }, { "epoch": 2.13, "learning_rate": 5.898025134649911e-06, "loss": 0.0011, "step": 20140 }, { "epoch": 2.13, "learning_rate": 5.8908438061041294e-06, "loss": 0.0012, "step": 20150 }, { "epoch": 2.13, "learning_rate": 5.883662477558348e-06, "loss": 0.001, "step": 20160 }, { "epoch": 2.13, "learning_rate": 5.876481149012567e-06, "loss": 0.0011, "step": 20170 }, { "epoch": 2.14, "learning_rate": 5.869299820466788e-06, "loss": 0.0013, "step": 20180 }, { "epoch": 2.14, "learning_rate": 5.8621184919210065e-06, "loss": 0.0014, "step": 20190 }, { "epoch": 2.14, "learning_rate": 5.854937163375225e-06, "loss": 0.0014, "step": 20200 }, { "epoch": 2.14, "learning_rate": 5.847755834829444e-06, "loss": 0.0011, "step": 20210 }, { "epoch": 2.14, "learning_rate": 5.840574506283663e-06, "loss": 0.0014, "step": 20220 }, { "epoch": 2.14, "learning_rate": 5.833393177737882e-06, "loss": 0.0013, "step": 20230 }, { "epoch": 2.14, "learning_rate": 5.826211849192101e-06, "loss": 0.0012, "step": 20240 }, { "epoch": 2.14, "learning_rate": 5.81903052064632e-06, "loss": 0.0013, "step": 20250 }, { "epoch": 2.14, "learning_rate": 5.811849192100539e-06, "loss": 0.0011, "step": 20260 }, { "epoch": 2.14, "learning_rate": 5.804667863554758e-06, "loss": 0.0017, "step": 20270 }, { "epoch": 2.15, "learning_rate": 5.797486535008977e-06, "loss": 0.0012, "step": 20280 }, { "epoch": 2.15, "learning_rate": 5.790305206463196e-06, "loss": 0.0012, "step": 20290 }, { "epoch": 2.15, "learning_rate": 5.7831238779174145e-06, "loss": 0.0013, "step": 20300 }, { "epoch": 2.15, "learning_rate": 5.775942549371633e-06, "loss": 0.0012, "step": 20310 }, { "epoch": 2.15, "learning_rate": 5.768761220825854e-06, "loss": 0.0013, "step": 20320 }, { "epoch": 2.15, "learning_rate": 5.761579892280073e-06, "loss": 0.0012, "step": 20330 }, { "epoch": 2.15, "learning_rate": 5.7543985637342915e-06, "loss": 0.0014, "step": 20340 }, { "epoch": 2.15, "learning_rate": 5.74721723518851e-06, "loss": 0.0015, "step": 20350 }, { "epoch": 2.15, "learning_rate": 5.740035906642729e-06, "loss": 0.0012, "step": 20360 }, { "epoch": 2.16, "learning_rate": 5.732854578096948e-06, "loss": 0.0013, "step": 20370 }, { "epoch": 2.16, "learning_rate": 5.725673249551168e-06, "loss": 0.0012, "step": 20380 }, { "epoch": 2.16, "learning_rate": 5.7184919210053865e-06, "loss": 0.0014, "step": 20390 }, { "epoch": 2.16, "learning_rate": 5.711310592459605e-06, "loss": 0.0013, "step": 20400 }, { "epoch": 2.16, "learning_rate": 5.704129263913825e-06, "loss": 0.0011, "step": 20410 }, { "epoch": 2.16, "learning_rate": 5.696947935368044e-06, "loss": 0.0011, "step": 20420 }, { "epoch": 2.16, "learning_rate": 5.689766606822263e-06, "loss": 0.0009, "step": 20430 }, { "epoch": 2.16, "learning_rate": 5.6825852782764815e-06, "loss": 0.001, "step": 20440 }, { "epoch": 2.16, "learning_rate": 5.675403949730701e-06, "loss": 0.0013, "step": 20450 }, { "epoch": 2.17, "learning_rate": 5.66822262118492e-06, "loss": 0.0013, "step": 20460 }, { "epoch": 2.17, "learning_rate": 5.661041292639139e-06, "loss": 0.001, "step": 20470 }, { "epoch": 2.17, "learning_rate": 5.653859964093358e-06, "loss": 0.0012, "step": 20480 }, { "epoch": 2.17, "learning_rate": 5.6466786355475765e-06, "loss": 0.0014, "step": 20490 }, { "epoch": 2.17, "learning_rate": 5.639497307001795e-06, "loss": 0.001, "step": 20500 }, { "epoch": 2.17, "learning_rate": 5.632315978456014e-06, "loss": 0.0013, "step": 20510 }, { "epoch": 2.17, "learning_rate": 5.625134649910235e-06, "loss": 0.0012, "step": 20520 }, { "epoch": 2.17, "learning_rate": 5.6179533213644535e-06, "loss": 0.0013, "step": 20530 }, { "epoch": 2.17, "learning_rate": 5.610771992818672e-06, "loss": 0.0011, "step": 20540 }, { "epoch": 2.17, "learning_rate": 5.603590664272891e-06, "loss": 0.0012, "step": 20550 }, { "epoch": 2.18, "learning_rate": 5.59640933572711e-06, "loss": 0.0012, "step": 20560 }, { "epoch": 2.18, "learning_rate": 5.589228007181329e-06, "loss": 0.0013, "step": 20570 }, { "epoch": 2.18, "learning_rate": 5.582046678635548e-06, "loss": 0.0013, "step": 20580 }, { "epoch": 2.18, "learning_rate": 5.574865350089767e-06, "loss": 0.0011, "step": 20590 }, { "epoch": 2.18, "learning_rate": 5.567684021543986e-06, "loss": 0.0013, "step": 20600 }, { "epoch": 2.18, "learning_rate": 5.560502692998205e-06, "loss": 0.0012, "step": 20610 }, { "epoch": 2.18, "learning_rate": 5.553321364452424e-06, "loss": 0.0013, "step": 20620 }, { "epoch": 2.18, "learning_rate": 5.546140035906643e-06, "loss": 0.0014, "step": 20630 }, { "epoch": 2.18, "learning_rate": 5.538958707360862e-06, "loss": 0.0011, "step": 20640 }, { "epoch": 2.19, "learning_rate": 5.531777378815081e-06, "loss": 0.0014, "step": 20650 }, { "epoch": 2.19, "learning_rate": 5.524596050269301e-06, "loss": 0.0014, "step": 20660 }, { "epoch": 2.19, "learning_rate": 5.51741472172352e-06, "loss": 0.0014, "step": 20670 }, { "epoch": 2.19, "learning_rate": 5.5102333931777385e-06, "loss": 0.0013, "step": 20680 }, { "epoch": 2.19, "learning_rate": 5.503052064631957e-06, "loss": 0.0011, "step": 20690 }, { "epoch": 2.19, "learning_rate": 5.495870736086176e-06, "loss": 0.0011, "step": 20700 }, { "epoch": 2.19, "learning_rate": 5.488689407540395e-06, "loss": 0.0012, "step": 20710 }, { "epoch": 2.19, "learning_rate": 5.481508078994614e-06, "loss": 0.0012, "step": 20720 }, { "epoch": 2.19, "learning_rate": 5.474326750448834e-06, "loss": 0.0013, "step": 20730 }, { "epoch": 2.19, "learning_rate": 5.467145421903053e-06, "loss": 0.0012, "step": 20740 }, { "epoch": 2.2, "learning_rate": 5.459964093357272e-06, "loss": 0.0013, "step": 20750 }, { "epoch": 2.2, "learning_rate": 5.452782764811491e-06, "loss": 0.0012, "step": 20760 }, { "epoch": 2.2, "learning_rate": 5.44560143626571e-06, "loss": 0.0013, "step": 20770 }, { "epoch": 2.2, "learning_rate": 5.4384201077199285e-06, "loss": 0.0011, "step": 20780 }, { "epoch": 2.2, "learning_rate": 5.431238779174147e-06, "loss": 0.0013, "step": 20790 }, { "epoch": 2.2, "learning_rate": 5.424057450628367e-06, "loss": 0.0012, "step": 20800 }, { "epoch": 2.2, "learning_rate": 5.416876122082586e-06, "loss": 0.0013, "step": 20810 }, { "epoch": 2.2, "learning_rate": 5.409694793536805e-06, "loss": 0.0014, "step": 20820 }, { "epoch": 2.2, "learning_rate": 5.4025134649910235e-06, "loss": 0.0011, "step": 20830 }, { "epoch": 2.21, "learning_rate": 5.395332136445242e-06, "loss": 0.0013, "step": 20840 }, { "epoch": 2.21, "learning_rate": 5.388150807899461e-06, "loss": 0.0013, "step": 20850 }, { "epoch": 2.21, "learning_rate": 5.38096947935368e-06, "loss": 0.0012, "step": 20860 }, { "epoch": 2.21, "learning_rate": 5.3737881508079006e-06, "loss": 0.0012, "step": 20870 }, { "epoch": 2.21, "learning_rate": 5.366606822262119e-06, "loss": 0.0012, "step": 20880 }, { "epoch": 2.21, "learning_rate": 5.359425493716338e-06, "loss": 0.0014, "step": 20890 }, { "epoch": 2.21, "learning_rate": 5.352244165170557e-06, "loss": 0.0011, "step": 20900 }, { "epoch": 2.21, "learning_rate": 5.345062836624776e-06, "loss": 0.0013, "step": 20910 }, { "epoch": 2.21, "learning_rate": 5.337881508078995e-06, "loss": 0.0012, "step": 20920 }, { "epoch": 2.21, "learning_rate": 5.3307001795332136e-06, "loss": 0.0009, "step": 20930 }, { "epoch": 2.22, "learning_rate": 5.323518850987433e-06, "loss": 0.0011, "step": 20940 }, { "epoch": 2.22, "learning_rate": 5.316337522441652e-06, "loss": 0.0012, "step": 20950 }, { "epoch": 2.22, "learning_rate": 5.309156193895872e-06, "loss": 0.0013, "step": 20960 }, { "epoch": 2.22, "learning_rate": 5.301974865350091e-06, "loss": 0.0016, "step": 20970 }, { "epoch": 2.22, "learning_rate": 5.294793536804309e-06, "loss": 0.0012, "step": 20980 }, { "epoch": 2.22, "learning_rate": 5.287612208258528e-06, "loss": 0.0013, "step": 20990 }, { "epoch": 2.22, "learning_rate": 5.280430879712747e-06, "loss": 0.0013, "step": 21000 }, { "epoch": 2.22, "learning_rate": 5.273249551166967e-06, "loss": 0.0013, "step": 21010 }, { "epoch": 2.22, "learning_rate": 5.266068222621186e-06, "loss": 0.0014, "step": 21020 }, { "epoch": 2.23, "learning_rate": 5.258886894075404e-06, "loss": 0.0014, "step": 21030 }, { "epoch": 2.23, "learning_rate": 5.251705565529623e-06, "loss": 0.0012, "step": 21040 }, { "epoch": 2.23, "learning_rate": 5.244524236983842e-06, "loss": 0.0012, "step": 21050 }, { "epoch": 2.23, "learning_rate": 5.237342908438061e-06, "loss": 0.0011, "step": 21060 }, { "epoch": 2.23, "learning_rate": 5.23016157989228e-06, "loss": 0.0012, "step": 21070 }, { "epoch": 2.23, "learning_rate": 5.2229802513465e-06, "loss": 0.0013, "step": 21080 }, { "epoch": 2.23, "learning_rate": 5.215798922800719e-06, "loss": 0.0012, "step": 21090 }, { "epoch": 2.23, "learning_rate": 5.208617594254938e-06, "loss": 0.0012, "step": 21100 }, { "epoch": 2.23, "learning_rate": 5.201436265709157e-06, "loss": 0.0012, "step": 21110 }, { "epoch": 2.23, "learning_rate": 5.194254937163376e-06, "loss": 0.0011, "step": 21120 }, { "epoch": 2.24, "learning_rate": 5.1870736086175944e-06, "loss": 0.0012, "step": 21130 }, { "epoch": 2.24, "learning_rate": 5.179892280071813e-06, "loss": 0.0011, "step": 21140 }, { "epoch": 2.24, "learning_rate": 5.172710951526033e-06, "loss": 0.0013, "step": 21150 }, { "epoch": 2.24, "learning_rate": 5.165529622980252e-06, "loss": 0.0013, "step": 21160 }, { "epoch": 2.24, "learning_rate": 5.158348294434471e-06, "loss": 0.0012, "step": 21170 }, { "epoch": 2.24, "learning_rate": 5.1511669658886894e-06, "loss": 0.0013, "step": 21180 }, { "epoch": 2.24, "learning_rate": 5.143985637342908e-06, "loss": 0.0014, "step": 21190 }, { "epoch": 2.24, "learning_rate": 5.136804308797128e-06, "loss": 0.001, "step": 21200 }, { "epoch": 2.24, "learning_rate": 5.129622980251347e-06, "loss": 0.0013, "step": 21210 }, { "epoch": 2.25, "learning_rate": 5.1224416517055664e-06, "loss": 0.0013, "step": 21220 }, { "epoch": 2.25, "learning_rate": 5.115260323159785e-06, "loss": 0.0015, "step": 21230 }, { "epoch": 2.25, "learning_rate": 5.108078994614004e-06, "loss": 0.0013, "step": 21240 }, { "epoch": 2.25, "learning_rate": 5.100897666068223e-06, "loss": 0.0011, "step": 21250 }, { "epoch": 2.25, "learning_rate": 5.093716337522442e-06, "loss": 0.0011, "step": 21260 }, { "epoch": 2.25, "learning_rate": 5.086535008976661e-06, "loss": 0.0011, "step": 21270 }, { "epoch": 2.25, "learning_rate": 5.0793536804308794e-06, "loss": 0.0012, "step": 21280 }, { "epoch": 2.25, "learning_rate": 5.0721723518851e-06, "loss": 0.0014, "step": 21290 }, { "epoch": 2.25, "learning_rate": 5.064991023339319e-06, "loss": 0.001, "step": 21300 }, { "epoch": 2.26, "learning_rate": 5.057809694793538e-06, "loss": 0.001, "step": 21310 }, { "epoch": 2.26, "learning_rate": 5.0506283662477565e-06, "loss": 0.0013, "step": 21320 }, { "epoch": 2.26, "learning_rate": 5.043447037701975e-06, "loss": 0.001, "step": 21330 }, { "epoch": 2.26, "learning_rate": 5.036265709156194e-06, "loss": 0.0013, "step": 21340 }, { "epoch": 2.26, "learning_rate": 5.029084380610413e-06, "loss": 0.0011, "step": 21350 }, { "epoch": 2.26, "learning_rate": 5.021903052064633e-06, "loss": 0.0012, "step": 21360 }, { "epoch": 2.26, "learning_rate": 5.0147217235188515e-06, "loss": 0.0013, "step": 21370 }, { "epoch": 2.26, "learning_rate": 5.00754039497307e-06, "loss": 0.0011, "step": 21380 }, { "epoch": 2.26, "learning_rate": 5.000359066427289e-06, "loss": 0.0012, "step": 21390 }, { "epoch": 2.26, "learning_rate": 4.993177737881508e-06, "loss": 0.0014, "step": 21400 }, { "epoch": 2.27, "learning_rate": 4.985996409335728e-06, "loss": 0.0012, "step": 21410 }, { "epoch": 2.27, "learning_rate": 4.9788150807899465e-06, "loss": 0.0013, "step": 21420 }, { "epoch": 2.27, "learning_rate": 4.971633752244165e-06, "loss": 0.0013, "step": 21430 }, { "epoch": 2.27, "learning_rate": 4.964452423698385e-06, "loss": 0.0012, "step": 21440 }, { "epoch": 2.27, "learning_rate": 4.957271095152604e-06, "loss": 0.0012, "step": 21450 }, { "epoch": 2.27, "learning_rate": 4.950089766606823e-06, "loss": 0.0012, "step": 21460 }, { "epoch": 2.27, "learning_rate": 4.9429084380610415e-06, "loss": 0.0012, "step": 21470 }, { "epoch": 2.27, "learning_rate": 4.935727109515261e-06, "loss": 0.0013, "step": 21480 }, { "epoch": 2.27, "learning_rate": 4.92854578096948e-06, "loss": 0.0014, "step": 21490 }, { "epoch": 2.28, "learning_rate": 4.921364452423699e-06, "loss": 0.0013, "step": 21500 }, { "epoch": 2.28, "learning_rate": 4.914183123877918e-06, "loss": 0.0014, "step": 21510 }, { "epoch": 2.28, "learning_rate": 4.907001795332137e-06, "loss": 0.0014, "step": 21520 }, { "epoch": 2.28, "learning_rate": 4.899820466786356e-06, "loss": 0.0012, "step": 21530 }, { "epoch": 2.28, "learning_rate": 4.892639138240575e-06, "loss": 0.0014, "step": 21540 }, { "epoch": 2.28, "learning_rate": 4.885457809694794e-06, "loss": 0.0012, "step": 21550 }, { "epoch": 2.28, "learning_rate": 4.878276481149013e-06, "loss": 0.0012, "step": 21560 }, { "epoch": 2.28, "learning_rate": 4.8710951526032315e-06, "loss": 0.0013, "step": 21570 }, { "epoch": 2.28, "learning_rate": 4.863913824057451e-06, "loss": 0.0011, "step": 21580 }, { "epoch": 2.28, "learning_rate": 4.85673249551167e-06, "loss": 0.0014, "step": 21590 }, { "epoch": 2.29, "learning_rate": 4.849551166965889e-06, "loss": 0.0013, "step": 21600 }, { "epoch": 2.29, "learning_rate": 4.842369838420108e-06, "loss": 0.001, "step": 21610 }, { "epoch": 2.29, "learning_rate": 4.835188509874327e-06, "loss": 0.0009, "step": 21620 }, { "epoch": 2.29, "learning_rate": 4.828007181328546e-06, "loss": 0.0013, "step": 21630 }, { "epoch": 2.29, "learning_rate": 4.820825852782765e-06, "loss": 0.0013, "step": 21640 }, { "epoch": 2.29, "learning_rate": 4.814362657091562e-06, "loss": 0.0013, "step": 21650 }, { "epoch": 2.29, "learning_rate": 4.807181328545781e-06, "loss": 0.0014, "step": 21660 }, { "epoch": 2.29, "learning_rate": 4.800000000000001e-06, "loss": 0.0012, "step": 21670 }, { "epoch": 2.29, "learning_rate": 4.7928186714542195e-06, "loss": 0.0011, "step": 21680 }, { "epoch": 2.3, "learning_rate": 4.785637342908438e-06, "loss": 0.001, "step": 21690 }, { "epoch": 2.3, "learning_rate": 4.778456014362657e-06, "loss": 0.001, "step": 21700 }, { "epoch": 2.3, "learning_rate": 4.771274685816876e-06, "loss": 0.0013, "step": 21710 }, { "epoch": 2.3, "learning_rate": 4.764093357271096e-06, "loss": 0.0012, "step": 21720 }, { "epoch": 2.3, "learning_rate": 4.7569120287253145e-06, "loss": 0.0013, "step": 21730 }, { "epoch": 2.3, "learning_rate": 4.749730700179533e-06, "loss": 0.0012, "step": 21740 }, { "epoch": 2.3, "learning_rate": 4.742549371633753e-06, "loss": 0.0012, "step": 21750 }, { "epoch": 2.3, "learning_rate": 4.735368043087972e-06, "loss": 0.0011, "step": 21760 }, { "epoch": 2.3, "learning_rate": 4.728186714542191e-06, "loss": 0.0012, "step": 21770 }, { "epoch": 2.3, "learning_rate": 4.7210053859964095e-06, "loss": 0.0011, "step": 21780 }, { "epoch": 2.31, "learning_rate": 4.713824057450629e-06, "loss": 0.001, "step": 21790 }, { "epoch": 2.31, "learning_rate": 4.706642728904848e-06, "loss": 0.001, "step": 21800 }, { "epoch": 2.31, "learning_rate": 4.699461400359067e-06, "loss": 0.001, "step": 21810 }, { "epoch": 2.31, "learning_rate": 4.692280071813286e-06, "loss": 0.0014, "step": 21820 }, { "epoch": 2.31, "learning_rate": 4.685098743267505e-06, "loss": 0.0013, "step": 21830 }, { "epoch": 2.31, "learning_rate": 4.677917414721724e-06, "loss": 0.0013, "step": 21840 }, { "epoch": 2.31, "learning_rate": 4.670736086175943e-06, "loss": 0.0013, "step": 21850 }, { "epoch": 2.31, "learning_rate": 4.663554757630162e-06, "loss": 0.0011, "step": 21860 }, { "epoch": 2.31, "learning_rate": 4.656373429084381e-06, "loss": 0.0013, "step": 21870 }, { "epoch": 2.32, "learning_rate": 4.6491921005385995e-06, "loss": 0.001, "step": 21880 }, { "epoch": 2.32, "learning_rate": 4.642010771992819e-06, "loss": 0.0013, "step": 21890 }, { "epoch": 2.32, "learning_rate": 4.634829443447038e-06, "loss": 0.0013, "step": 21900 }, { "epoch": 2.32, "learning_rate": 4.627648114901257e-06, "loss": 0.0012, "step": 21910 }, { "epoch": 2.32, "learning_rate": 4.6204667863554765e-06, "loss": 0.0014, "step": 21920 }, { "epoch": 2.32, "learning_rate": 4.613285457809695e-06, "loss": 0.0013, "step": 21930 }, { "epoch": 2.32, "learning_rate": 4.606104129263914e-06, "loss": 0.0011, "step": 21940 }, { "epoch": 2.32, "learning_rate": 4.598922800718133e-06, "loss": 0.0011, "step": 21950 }, { "epoch": 2.32, "learning_rate": 4.591741472172353e-06, "loss": 0.0012, "step": 21960 }, { "epoch": 2.32, "learning_rate": 4.5845601436265715e-06, "loss": 0.0014, "step": 21970 }, { "epoch": 2.33, "learning_rate": 4.57737881508079e-06, "loss": 0.001, "step": 21980 }, { "epoch": 2.33, "learning_rate": 4.57019748653501e-06, "loss": 0.0012, "step": 21990 }, { "epoch": 2.33, "learning_rate": 4.563016157989229e-06, "loss": 0.0011, "step": 22000 }, { "epoch": 2.33, "learning_rate": 4.555834829443448e-06, "loss": 0.0014, "step": 22010 }, { "epoch": 2.33, "learning_rate": 4.5486535008976665e-06, "loss": 0.0012, "step": 22020 }, { "epoch": 2.33, "learning_rate": 4.541472172351885e-06, "loss": 0.0012, "step": 22030 }, { "epoch": 2.33, "learning_rate": 4.534290843806104e-06, "loss": 0.0012, "step": 22040 }, { "epoch": 2.33, "learning_rate": 4.527109515260323e-06, "loss": 0.0011, "step": 22050 }, { "epoch": 2.33, "learning_rate": 4.519928186714543e-06, "loss": 0.001, "step": 22060 }, { "epoch": 2.34, "learning_rate": 4.5127468581687615e-06, "loss": 0.0009, "step": 22070 }, { "epoch": 2.34, "learning_rate": 4.50556552962298e-06, "loss": 0.0013, "step": 22080 }, { "epoch": 2.34, "learning_rate": 4.498384201077199e-06, "loss": 0.0012, "step": 22090 }, { "epoch": 2.34, "learning_rate": 4.491202872531419e-06, "loss": 0.0011, "step": 22100 }, { "epoch": 2.34, "learning_rate": 4.484021543985638e-06, "loss": 0.0013, "step": 22110 }, { "epoch": 2.34, "learning_rate": 4.4768402154398565e-06, "loss": 0.0011, "step": 22120 }, { "epoch": 2.34, "learning_rate": 4.469658886894076e-06, "loss": 0.0011, "step": 22130 }, { "epoch": 2.34, "learning_rate": 4.462477558348295e-06, "loss": 0.0014, "step": 22140 }, { "epoch": 2.34, "learning_rate": 4.455296229802514e-06, "loss": 0.0009, "step": 22150 }, { "epoch": 2.34, "learning_rate": 4.448114901256733e-06, "loss": 0.0011, "step": 22160 }, { "epoch": 2.35, "learning_rate": 4.440933572710952e-06, "loss": 0.0013, "step": 22170 }, { "epoch": 2.35, "learning_rate": 4.433752244165171e-06, "loss": 0.0013, "step": 22180 }, { "epoch": 2.35, "learning_rate": 4.42657091561939e-06, "loss": 0.0014, "step": 22190 }, { "epoch": 2.35, "learning_rate": 4.419389587073609e-06, "loss": 0.0012, "step": 22200 }, { "epoch": 2.35, "learning_rate": 4.412208258527828e-06, "loss": 0.0013, "step": 22210 }, { "epoch": 2.35, "learning_rate": 4.4050269299820465e-06, "loss": 0.0012, "step": 22220 }, { "epoch": 2.35, "learning_rate": 4.397845601436266e-06, "loss": 0.001, "step": 22230 }, { "epoch": 2.35, "learning_rate": 4.390664272890485e-06, "loss": 0.0015, "step": 22240 }, { "epoch": 2.35, "learning_rate": 4.383482944344704e-06, "loss": 0.0012, "step": 22250 }, { "epoch": 2.36, "learning_rate": 4.376301615798923e-06, "loss": 0.0009, "step": 22260 }, { "epoch": 2.36, "learning_rate": 4.369120287253142e-06, "loss": 0.0013, "step": 22270 }, { "epoch": 2.36, "learning_rate": 4.361938958707361e-06, "loss": 0.0012, "step": 22280 }, { "epoch": 2.36, "learning_rate": 4.35475763016158e-06, "loss": 0.0015, "step": 22290 }, { "epoch": 2.36, "learning_rate": 4.347576301615799e-06, "loss": 0.0013, "step": 22300 }, { "epoch": 2.36, "learning_rate": 4.3403949730700186e-06, "loss": 0.0011, "step": 22310 }, { "epoch": 2.36, "learning_rate": 4.333213644524237e-06, "loss": 0.001, "step": 22320 }, { "epoch": 2.36, "learning_rate": 4.326032315978456e-06, "loss": 0.0013, "step": 22330 }, { "epoch": 2.36, "learning_rate": 4.318850987432676e-06, "loss": 0.0011, "step": 22340 }, { "epoch": 2.37, "learning_rate": 4.311669658886895e-06, "loss": 0.001, "step": 22350 }, { "epoch": 2.37, "learning_rate": 4.3044883303411136e-06, "loss": 0.0012, "step": 22360 }, { "epoch": 2.37, "learning_rate": 4.297307001795332e-06, "loss": 0.0012, "step": 22370 }, { "epoch": 2.37, "learning_rate": 4.290125673249551e-06, "loss": 0.0013, "step": 22380 }, { "epoch": 2.37, "learning_rate": 4.282944344703771e-06, "loss": 0.0013, "step": 22390 }, { "epoch": 2.37, "learning_rate": 4.27576301615799e-06, "loss": 0.001, "step": 22400 }, { "epoch": 2.37, "learning_rate": 4.2685816876122086e-06, "loss": 0.0013, "step": 22410 }, { "epoch": 2.37, "learning_rate": 4.261400359066427e-06, "loss": 0.001, "step": 22420 }, { "epoch": 2.37, "learning_rate": 4.254219030520646e-06, "loss": 0.0015, "step": 22430 }, { "epoch": 2.37, "learning_rate": 4.247037701974865e-06, "loss": 0.0011, "step": 22440 }, { "epoch": 2.38, "learning_rate": 4.239856373429085e-06, "loss": 0.0012, "step": 22450 }, { "epoch": 2.38, "learning_rate": 4.2326750448833036e-06, "loss": 0.0014, "step": 22460 }, { "epoch": 2.38, "learning_rate": 4.225493716337522e-06, "loss": 0.0011, "step": 22470 }, { "epoch": 2.38, "learning_rate": 4.218312387791742e-06, "loss": 0.0013, "step": 22480 }, { "epoch": 2.38, "learning_rate": 4.211131059245961e-06, "loss": 0.0013, "step": 22490 }, { "epoch": 2.38, "learning_rate": 4.20394973070018e-06, "loss": 0.001, "step": 22500 }, { "epoch": 2.38, "learning_rate": 4.1967684021543994e-06, "loss": 0.0015, "step": 22510 }, { "epoch": 2.38, "learning_rate": 4.189587073608618e-06, "loss": 0.0012, "step": 22520 }, { "epoch": 2.38, "learning_rate": 4.182405745062837e-06, "loss": 0.0015, "step": 22530 }, { "epoch": 2.39, "learning_rate": 4.175224416517056e-06, "loss": 0.0013, "step": 22540 }, { "epoch": 2.39, "learning_rate": 4.168043087971276e-06, "loss": 0.0012, "step": 22550 }, { "epoch": 2.39, "learning_rate": 4.1608617594254944e-06, "loss": 0.0013, "step": 22560 }, { "epoch": 2.39, "learning_rate": 4.153680430879713e-06, "loss": 0.0012, "step": 22570 }, { "epoch": 2.39, "learning_rate": 4.146499102333932e-06, "loss": 0.0012, "step": 22580 }, { "epoch": 2.39, "learning_rate": 4.139317773788151e-06, "loss": 0.0013, "step": 22590 }, { "epoch": 2.39, "learning_rate": 4.13213644524237e-06, "loss": 0.0012, "step": 22600 }, { "epoch": 2.39, "learning_rate": 4.124955116696589e-06, "loss": 0.0012, "step": 22610 }, { "epoch": 2.39, "learning_rate": 4.117773788150808e-06, "loss": 0.0011, "step": 22620 }, { "epoch": 2.39, "learning_rate": 4.110592459605027e-06, "loss": 0.0015, "step": 22630 }, { "epoch": 2.4, "learning_rate": 4.103411131059246e-06, "loss": 0.0014, "step": 22640 }, { "epoch": 2.4, "learning_rate": 4.096229802513466e-06, "loss": 0.0015, "step": 22650 }, { "epoch": 2.4, "learning_rate": 4.0890484739676844e-06, "loss": 0.0012, "step": 22660 }, { "epoch": 2.4, "learning_rate": 4.081867145421903e-06, "loss": 0.0012, "step": 22670 }, { "epoch": 2.4, "learning_rate": 4.074685816876122e-06, "loss": 0.0012, "step": 22680 }, { "epoch": 2.4, "learning_rate": 4.067504488330342e-06, "loss": 0.0012, "step": 22690 }, { "epoch": 2.4, "learning_rate": 4.060323159784561e-06, "loss": 0.001, "step": 22700 }, { "epoch": 2.4, "learning_rate": 4.0531418312387794e-06, "loss": 0.0011, "step": 22710 }, { "epoch": 2.4, "learning_rate": 4.045960502692999e-06, "loss": 0.0012, "step": 22720 }, { "epoch": 2.41, "learning_rate": 4.038779174147218e-06, "loss": 0.0012, "step": 22730 }, { "epoch": 2.41, "learning_rate": 4.031597845601437e-06, "loss": 0.0012, "step": 22740 }, { "epoch": 2.41, "learning_rate": 4.024416517055656e-06, "loss": 0.0012, "step": 22750 }, { "epoch": 2.41, "learning_rate": 4.0172351885098745e-06, "loss": 0.0012, "step": 22760 }, { "epoch": 2.41, "learning_rate": 4.010053859964093e-06, "loss": 0.0012, "step": 22770 }, { "epoch": 2.41, "learning_rate": 4.002872531418313e-06, "loss": 0.001, "step": 22780 }, { "epoch": 2.41, "learning_rate": 3.995691202872532e-06, "loss": 0.0012, "step": 22790 }, { "epoch": 2.41, "learning_rate": 3.988509874326751e-06, "loss": 0.0009, "step": 22800 }, { "epoch": 2.41, "learning_rate": 3.9813285457809695e-06, "loss": 0.0014, "step": 22810 }, { "epoch": 2.41, "learning_rate": 3.974147217235188e-06, "loss": 0.0014, "step": 22820 }, { "epoch": 2.42, "learning_rate": 3.966965888689408e-06, "loss": 0.001, "step": 22830 }, { "epoch": 2.42, "learning_rate": 3.959784560143627e-06, "loss": 0.0013, "step": 22840 }, { "epoch": 2.42, "learning_rate": 3.952603231597846e-06, "loss": 0.0009, "step": 22850 }, { "epoch": 2.42, "learning_rate": 3.945421903052065e-06, "loss": 0.0012, "step": 22860 }, { "epoch": 2.42, "learning_rate": 3.938240574506284e-06, "loss": 0.0012, "step": 22870 }, { "epoch": 2.42, "learning_rate": 3.931059245960503e-06, "loss": 0.0011, "step": 22880 }, { "epoch": 2.42, "learning_rate": 3.923877917414722e-06, "loss": 0.0013, "step": 22890 }, { "epoch": 2.42, "learning_rate": 3.9166965888689415e-06, "loss": 0.0011, "step": 22900 }, { "epoch": 2.42, "learning_rate": 3.90951526032316e-06, "loss": 0.0011, "step": 22910 }, { "epoch": 2.43, "learning_rate": 3.902333931777379e-06, "loss": 0.0011, "step": 22920 }, { "epoch": 2.43, "learning_rate": 3.895152603231598e-06, "loss": 0.0013, "step": 22930 }, { "epoch": 2.43, "learning_rate": 3.887971274685818e-06, "loss": 0.0013, "step": 22940 }, { "epoch": 2.43, "learning_rate": 3.8807899461400365e-06, "loss": 0.0013, "step": 22950 }, { "epoch": 2.43, "learning_rate": 3.873608617594255e-06, "loss": 0.0013, "step": 22960 }, { "epoch": 2.43, "learning_rate": 3.866427289048474e-06, "loss": 0.0012, "step": 22970 }, { "epoch": 2.43, "learning_rate": 3.859245960502693e-06, "loss": 0.0012, "step": 22980 }, { "epoch": 2.43, "learning_rate": 3.852064631956912e-06, "loss": 0.0014, "step": 22990 }, { "epoch": 2.43, "learning_rate": 3.8448833034111315e-06, "loss": 0.0015, "step": 23000 }, { "epoch": 2.43, "learning_rate": 3.83770197486535e-06, "loss": 0.0011, "step": 23010 }, { "epoch": 2.44, "learning_rate": 3.830520646319569e-06, "loss": 0.0013, "step": 23020 }, { "epoch": 2.44, "learning_rate": 3.823339317773788e-06, "loss": 0.0012, "step": 23030 }, { "epoch": 2.44, "learning_rate": 3.816157989228008e-06, "loss": 0.0012, "step": 23040 }, { "epoch": 2.44, "learning_rate": 3.8089766606822265e-06, "loss": 0.0015, "step": 23050 }, { "epoch": 2.44, "learning_rate": 3.8017953321364453e-06, "loss": 0.0012, "step": 23060 }, { "epoch": 2.44, "learning_rate": 3.7946140035906646e-06, "loss": 0.0012, "step": 23070 }, { "epoch": 2.44, "learning_rate": 3.787432675044884e-06, "loss": 0.0012, "step": 23080 }, { "epoch": 2.44, "learning_rate": 3.7802513464991027e-06, "loss": 0.0012, "step": 23090 }, { "epoch": 2.44, "learning_rate": 3.7730700179533215e-06, "loss": 0.0011, "step": 23100 }, { "epoch": 2.45, "learning_rate": 3.7658886894075408e-06, "loss": 0.0012, "step": 23110 }, { "epoch": 2.45, "learning_rate": 3.7587073608617596e-06, "loss": 0.0011, "step": 23120 }, { "epoch": 2.45, "learning_rate": 3.7515260323159784e-06, "loss": 0.0009, "step": 23130 }, { "epoch": 2.45, "learning_rate": 3.744344703770198e-06, "loss": 0.0015, "step": 23140 }, { "epoch": 2.45, "learning_rate": 3.737163375224417e-06, "loss": 0.0011, "step": 23150 }, { "epoch": 2.45, "learning_rate": 3.7299820466786358e-06, "loss": 0.0013, "step": 23160 }, { "epoch": 2.45, "learning_rate": 3.722800718132855e-06, "loss": 0.0014, "step": 23170 }, { "epoch": 2.45, "learning_rate": 3.715619389587074e-06, "loss": 0.0012, "step": 23180 }, { "epoch": 2.45, "learning_rate": 3.7084380610412927e-06, "loss": 0.0013, "step": 23190 }, { "epoch": 2.46, "learning_rate": 3.701256732495512e-06, "loss": 0.0013, "step": 23200 }, { "epoch": 2.46, "learning_rate": 3.694075403949731e-06, "loss": 0.0011, "step": 23210 }, { "epoch": 2.46, "learning_rate": 3.68689407540395e-06, "loss": 0.0013, "step": 23220 }, { "epoch": 2.46, "learning_rate": 3.679712746858169e-06, "loss": 0.0012, "step": 23230 }, { "epoch": 2.46, "learning_rate": 3.6725314183123885e-06, "loss": 0.001, "step": 23240 }, { "epoch": 2.46, "learning_rate": 3.6653500897666074e-06, "loss": 0.0009, "step": 23250 }, { "epoch": 2.46, "learning_rate": 3.658168761220826e-06, "loss": 0.0012, "step": 23260 }, { "epoch": 2.46, "learning_rate": 3.650987432675045e-06, "loss": 0.0011, "step": 23270 }, { "epoch": 2.46, "learning_rate": 3.6438061041292643e-06, "loss": 0.001, "step": 23280 }, { "epoch": 2.46, "learning_rate": 3.636624775583483e-06, "loss": 0.0012, "step": 23290 }, { "epoch": 2.47, "learning_rate": 3.629443447037702e-06, "loss": 0.0013, "step": 23300 }, { "epoch": 2.47, "learning_rate": 3.6222621184919216e-06, "loss": 0.0014, "step": 23310 }, { "epoch": 2.47, "learning_rate": 3.6150807899461405e-06, "loss": 0.0011, "step": 23320 }, { "epoch": 2.47, "learning_rate": 3.6078994614003593e-06, "loss": 0.0013, "step": 23330 }, { "epoch": 2.47, "learning_rate": 3.600718132854578e-06, "loss": 0.0013, "step": 23340 }, { "epoch": 2.47, "learning_rate": 3.5935368043087974e-06, "loss": 0.0011, "step": 23350 }, { "epoch": 2.47, "learning_rate": 3.5863554757630166e-06, "loss": 0.0013, "step": 23360 }, { "epoch": 2.47, "learning_rate": 3.5791741472172355e-06, "loss": 0.0012, "step": 23370 }, { "epoch": 2.47, "learning_rate": 3.5719928186714547e-06, "loss": 0.0013, "step": 23380 }, { "epoch": 2.48, "learning_rate": 3.5648114901256736e-06, "loss": 0.0012, "step": 23390 }, { "epoch": 2.48, "learning_rate": 3.5576301615798924e-06, "loss": 0.0012, "step": 23400 }, { "epoch": 2.48, "learning_rate": 3.5504488330341112e-06, "loss": 0.0012, "step": 23410 }, { "epoch": 2.48, "learning_rate": 3.543267504488331e-06, "loss": 0.0013, "step": 23420 }, { "epoch": 2.48, "learning_rate": 3.5360861759425497e-06, "loss": 0.0013, "step": 23430 }, { "epoch": 2.48, "learning_rate": 3.5289048473967686e-06, "loss": 0.0012, "step": 23440 }, { "epoch": 2.48, "learning_rate": 3.521723518850988e-06, "loss": 0.0011, "step": 23450 }, { "epoch": 2.48, "learning_rate": 3.5145421903052066e-06, "loss": 0.0012, "step": 23460 }, { "epoch": 2.48, "learning_rate": 3.507360861759426e-06, "loss": 0.0012, "step": 23470 }, { "epoch": 2.48, "learning_rate": 3.5001795332136447e-06, "loss": 0.0013, "step": 23480 }, { "epoch": 2.49, "learning_rate": 3.492998204667864e-06, "loss": 0.0012, "step": 23490 }, { "epoch": 2.49, "learning_rate": 3.485816876122083e-06, "loss": 0.001, "step": 23500 }, { "epoch": 2.49, "learning_rate": 3.4786355475763016e-06, "loss": 0.0013, "step": 23510 }, { "epoch": 2.49, "learning_rate": 3.4714542190305213e-06, "loss": 0.0013, "step": 23520 }, { "epoch": 2.49, "learning_rate": 3.46427289048474e-06, "loss": 0.001, "step": 23530 }, { "epoch": 2.49, "learning_rate": 3.457091561938959e-06, "loss": 0.0014, "step": 23540 }, { "epoch": 2.49, "learning_rate": 3.449910233393178e-06, "loss": 0.0012, "step": 23550 }, { "epoch": 2.49, "learning_rate": 3.442728904847397e-06, "loss": 0.0011, "step": 23560 }, { "epoch": 2.49, "learning_rate": 3.435547576301616e-06, "loss": 0.0011, "step": 23570 }, { "epoch": 2.5, "learning_rate": 3.4283662477558347e-06, "loss": 0.0013, "step": 23580 }, { "epoch": 2.5, "learning_rate": 3.4211849192100544e-06, "loss": 0.0013, "step": 23590 }, { "epoch": 2.5, "learning_rate": 3.4140035906642733e-06, "loss": 0.0011, "step": 23600 }, { "epoch": 2.5, "learning_rate": 3.406822262118492e-06, "loss": 0.0014, "step": 23610 }, { "epoch": 2.5, "learning_rate": 3.399640933572711e-06, "loss": 0.0014, "step": 23620 }, { "epoch": 2.5, "learning_rate": 3.3924596050269306e-06, "loss": 0.0013, "step": 23630 }, { "epoch": 2.5, "learning_rate": 3.3852782764811494e-06, "loss": 0.0011, "step": 23640 }, { "epoch": 2.5, "learning_rate": 3.3780969479353683e-06, "loss": 0.0011, "step": 23650 }, { "epoch": 2.5, "learning_rate": 3.3716337522441654e-06, "loss": 0.0011, "step": 23660 }, { "epoch": 2.5, "learning_rate": 3.3644524236983846e-06, "loss": 0.0009, "step": 23670 }, { "epoch": 2.51, "learning_rate": 3.3572710951526035e-06, "loss": 0.0012, "step": 23680 }, { "epoch": 2.51, "learning_rate": 3.3500897666068227e-06, "loss": 0.0014, "step": 23690 }, { "epoch": 2.51, "learning_rate": 3.3429084380610415e-06, "loss": 0.0012, "step": 23700 }, { "epoch": 2.51, "learning_rate": 3.3357271095152604e-06, "loss": 0.0012, "step": 23710 }, { "epoch": 2.51, "learning_rate": 3.32854578096948e-06, "loss": 0.0012, "step": 23720 }, { "epoch": 2.51, "learning_rate": 3.321364452423699e-06, "loss": 0.0014, "step": 23730 }, { "epoch": 2.51, "learning_rate": 3.3141831238779177e-06, "loss": 0.0011, "step": 23740 }, { "epoch": 2.51, "learning_rate": 3.3070017953321366e-06, "loss": 0.0013, "step": 23750 }, { "epoch": 2.51, "learning_rate": 3.299820466786356e-06, "loss": 0.0012, "step": 23760 }, { "epoch": 2.52, "learning_rate": 3.2926391382405746e-06, "loss": 0.0013, "step": 23770 }, { "epoch": 2.52, "learning_rate": 3.2854578096947935e-06, "loss": 0.0013, "step": 23780 }, { "epoch": 2.52, "learning_rate": 3.278276481149013e-06, "loss": 0.0013, "step": 23790 }, { "epoch": 2.52, "learning_rate": 3.271095152603232e-06, "loss": 0.0011, "step": 23800 }, { "epoch": 2.52, "learning_rate": 3.263913824057451e-06, "loss": 0.0012, "step": 23810 }, { "epoch": 2.52, "learning_rate": 3.2567324955116696e-06, "loss": 0.0012, "step": 23820 }, { "epoch": 2.52, "learning_rate": 3.2495511669658893e-06, "loss": 0.0013, "step": 23830 }, { "epoch": 2.52, "learning_rate": 3.242369838420108e-06, "loss": 0.0016, "step": 23840 }, { "epoch": 2.52, "learning_rate": 3.235188509874327e-06, "loss": 0.0011, "step": 23850 }, { "epoch": 2.52, "learning_rate": 3.2280071813285462e-06, "loss": 0.0013, "step": 23860 }, { "epoch": 2.53, "learning_rate": 3.220825852782765e-06, "loss": 0.0011, "step": 23870 }, { "epoch": 2.53, "learning_rate": 3.213644524236984e-06, "loss": 0.0011, "step": 23880 }, { "epoch": 2.53, "learning_rate": 3.2064631956912027e-06, "loss": 0.0012, "step": 23890 }, { "epoch": 2.53, "learning_rate": 3.1992818671454224e-06, "loss": 0.0015, "step": 23900 }, { "epoch": 2.53, "learning_rate": 3.1921005385996412e-06, "loss": 0.0012, "step": 23910 }, { "epoch": 2.53, "learning_rate": 3.18491921005386e-06, "loss": 0.0015, "step": 23920 }, { "epoch": 2.53, "learning_rate": 3.1777378815080793e-06, "loss": 0.0013, "step": 23930 }, { "epoch": 2.53, "learning_rate": 3.170556552962298e-06, "loss": 0.0012, "step": 23940 }, { "epoch": 2.53, "learning_rate": 3.1633752244165174e-06, "loss": 0.0011, "step": 23950 }, { "epoch": 2.54, "learning_rate": 3.1561938958707363e-06, "loss": 0.0012, "step": 23960 }, { "epoch": 2.54, "learning_rate": 3.1490125673249555e-06, "loss": 0.0012, "step": 23970 }, { "epoch": 2.54, "learning_rate": 3.1418312387791743e-06, "loss": 0.0013, "step": 23980 }, { "epoch": 2.54, "learning_rate": 3.134649910233393e-06, "loss": 0.0012, "step": 23990 }, { "epoch": 2.54, "learning_rate": 3.127468581687613e-06, "loss": 0.0013, "step": 24000 }, { "epoch": 2.54, "learning_rate": 3.1202872531418317e-06, "loss": 0.001, "step": 24010 }, { "epoch": 2.54, "learning_rate": 3.1131059245960505e-06, "loss": 0.0013, "step": 24020 }, { "epoch": 2.54, "learning_rate": 3.1059245960502693e-06, "loss": 0.0013, "step": 24030 }, { "epoch": 2.54, "learning_rate": 3.0987432675044886e-06, "loss": 0.0012, "step": 24040 }, { "epoch": 2.54, "learning_rate": 3.0915619389587074e-06, "loss": 0.0015, "step": 24050 }, { "epoch": 2.55, "learning_rate": 3.0843806104129263e-06, "loss": 0.0011, "step": 24060 }, { "epoch": 2.55, "learning_rate": 3.077199281867146e-06, "loss": 0.0011, "step": 24070 }, { "epoch": 2.55, "learning_rate": 3.0700179533213648e-06, "loss": 0.0013, "step": 24080 }, { "epoch": 2.55, "learning_rate": 3.0628366247755836e-06, "loss": 0.0013, "step": 24090 }, { "epoch": 2.55, "learning_rate": 3.0556552962298024e-06, "loss": 0.0011, "step": 24100 }, { "epoch": 2.55, "learning_rate": 3.048473967684022e-06, "loss": 0.0012, "step": 24110 }, { "epoch": 2.55, "learning_rate": 3.041292639138241e-06, "loss": 0.0014, "step": 24120 }, { "epoch": 2.55, "learning_rate": 3.0341113105924598e-06, "loss": 0.0013, "step": 24130 }, { "epoch": 2.55, "learning_rate": 3.026929982046679e-06, "loss": 0.0013, "step": 24140 }, { "epoch": 2.56, "learning_rate": 3.019748653500898e-06, "loss": 0.0013, "step": 24150 }, { "epoch": 2.56, "learning_rate": 3.0125673249551167e-06, "loss": 0.0013, "step": 24160 }, { "epoch": 2.56, "learning_rate": 3.0053859964093355e-06, "loss": 0.0013, "step": 24170 }, { "epoch": 2.56, "learning_rate": 2.998204667863555e-06, "loss": 0.0012, "step": 24180 }, { "epoch": 2.56, "learning_rate": 2.991023339317774e-06, "loss": 0.0015, "step": 24190 }, { "epoch": 2.56, "learning_rate": 2.983842010771993e-06, "loss": 0.001, "step": 24200 }, { "epoch": 2.56, "learning_rate": 2.976660682226212e-06, "loss": 0.0012, "step": 24210 }, { "epoch": 2.56, "learning_rate": 2.969479353680431e-06, "loss": 0.0015, "step": 24220 }, { "epoch": 2.56, "learning_rate": 2.9622980251346502e-06, "loss": 0.0013, "step": 24230 }, { "epoch": 2.57, "learning_rate": 2.955116696588869e-06, "loss": 0.0014, "step": 24240 }, { "epoch": 2.57, "learning_rate": 2.9479353680430883e-06, "loss": 0.0011, "step": 24250 }, { "epoch": 2.57, "learning_rate": 2.940754039497307e-06, "loss": 0.0013, "step": 24260 }, { "epoch": 2.57, "learning_rate": 2.933572710951526e-06, "loss": 0.0011, "step": 24270 }, { "epoch": 2.57, "learning_rate": 2.9263913824057456e-06, "loss": 0.0013, "step": 24280 }, { "epoch": 2.57, "learning_rate": 2.9192100538599645e-06, "loss": 0.0011, "step": 24290 }, { "epoch": 2.57, "learning_rate": 2.9120287253141833e-06, "loss": 0.001, "step": 24300 }, { "epoch": 2.57, "learning_rate": 2.9048473967684026e-06, "loss": 0.0013, "step": 24310 }, { "epoch": 2.57, "learning_rate": 2.8976660682226214e-06, "loss": 0.0013, "step": 24320 }, { "epoch": 2.57, "learning_rate": 2.8904847396768402e-06, "loss": 0.001, "step": 24330 }, { "epoch": 2.58, "learning_rate": 2.8833034111310595e-06, "loss": 0.0011, "step": 24340 }, { "epoch": 2.58, "learning_rate": 2.8761220825852787e-06, "loss": 0.0013, "step": 24350 }, { "epoch": 2.58, "learning_rate": 2.8689407540394976e-06, "loss": 0.0013, "step": 24360 }, { "epoch": 2.58, "learning_rate": 2.8617594254937164e-06, "loss": 0.0013, "step": 24370 }, { "epoch": 2.58, "learning_rate": 2.8545780969479357e-06, "loss": 0.0011, "step": 24380 }, { "epoch": 2.58, "learning_rate": 2.847396768402155e-06, "loss": 0.0013, "step": 24390 }, { "epoch": 2.58, "learning_rate": 2.8402154398563737e-06, "loss": 0.0012, "step": 24400 }, { "epoch": 2.58, "learning_rate": 2.8330341113105926e-06, "loss": 0.0012, "step": 24410 }, { "epoch": 2.58, "learning_rate": 2.825852782764812e-06, "loss": 0.0011, "step": 24420 }, { "epoch": 2.59, "learning_rate": 2.8186714542190307e-06, "loss": 0.0013, "step": 24430 }, { "epoch": 2.59, "learning_rate": 2.8114901256732495e-06, "loss": 0.0012, "step": 24440 }, { "epoch": 2.59, "learning_rate": 2.804308797127469e-06, "loss": 0.0013, "step": 24450 }, { "epoch": 2.59, "learning_rate": 2.797127468581688e-06, "loss": 0.0012, "step": 24460 }, { "epoch": 2.59, "learning_rate": 2.789946140035907e-06, "loss": 0.0012, "step": 24470 }, { "epoch": 2.59, "learning_rate": 2.7827648114901257e-06, "loss": 0.0014, "step": 24480 }, { "epoch": 2.59, "learning_rate": 2.775583482944345e-06, "loss": 0.0013, "step": 24490 }, { "epoch": 2.59, "learning_rate": 2.768402154398564e-06, "loss": 0.0012, "step": 24500 }, { "epoch": 2.59, "learning_rate": 2.761220825852783e-06, "loss": 0.0011, "step": 24510 }, { "epoch": 2.59, "learning_rate": 2.7540394973070023e-06, "loss": 0.0011, "step": 24520 }, { "epoch": 2.6, "learning_rate": 2.746858168761221e-06, "loss": 0.0011, "step": 24530 }, { "epoch": 2.6, "learning_rate": 2.73967684021544e-06, "loss": 0.0013, "step": 24540 }, { "epoch": 2.6, "learning_rate": 2.7324955116696588e-06, "loss": 0.001, "step": 24550 }, { "epoch": 2.6, "learning_rate": 2.7253141831238784e-06, "loss": 0.0012, "step": 24560 }, { "epoch": 2.6, "learning_rate": 2.7181328545780973e-06, "loss": 0.0011, "step": 24570 }, { "epoch": 2.6, "learning_rate": 2.710951526032316e-06, "loss": 0.0009, "step": 24580 }, { "epoch": 2.6, "learning_rate": 2.7037701974865354e-06, "loss": 0.0013, "step": 24590 }, { "epoch": 2.6, "learning_rate": 2.696588868940754e-06, "loss": 0.0012, "step": 24600 }, { "epoch": 2.6, "learning_rate": 2.689407540394973e-06, "loss": 0.0011, "step": 24610 }, { "epoch": 2.61, "learning_rate": 2.6822262118491923e-06, "loss": 0.0012, "step": 24620 }, { "epoch": 2.61, "learning_rate": 2.6750448833034115e-06, "loss": 0.0012, "step": 24630 }, { "epoch": 2.61, "learning_rate": 2.6678635547576304e-06, "loss": 0.0012, "step": 24640 }, { "epoch": 2.61, "learning_rate": 2.660682226211849e-06, "loss": 0.0011, "step": 24650 }, { "epoch": 2.61, "learning_rate": 2.653500897666069e-06, "loss": 0.0011, "step": 24660 }, { "epoch": 2.61, "learning_rate": 2.6463195691202877e-06, "loss": 0.0012, "step": 24670 }, { "epoch": 2.61, "learning_rate": 2.6391382405745065e-06, "loss": 0.0014, "step": 24680 }, { "epoch": 2.61, "learning_rate": 2.6319569120287254e-06, "loss": 0.0011, "step": 24690 }, { "epoch": 2.61, "learning_rate": 2.6247755834829446e-06, "loss": 0.0013, "step": 24700 }, { "epoch": 2.61, "learning_rate": 2.6175942549371634e-06, "loss": 0.0011, "step": 24710 }, { "epoch": 2.62, "learning_rate": 2.6104129263913823e-06, "loss": 0.0011, "step": 24720 }, { "epoch": 2.62, "learning_rate": 2.603231597845602e-06, "loss": 0.0011, "step": 24730 }, { "epoch": 2.62, "learning_rate": 2.596050269299821e-06, "loss": 0.0013, "step": 24740 }, { "epoch": 2.62, "learning_rate": 2.5888689407540396e-06, "loss": 0.0009, "step": 24750 }, { "epoch": 2.62, "learning_rate": 2.5816876122082585e-06, "loss": 0.0013, "step": 24760 }, { "epoch": 2.62, "learning_rate": 2.5745062836624777e-06, "loss": 0.0013, "step": 24770 }, { "epoch": 2.62, "learning_rate": 2.567324955116697e-06, "loss": 0.0012, "step": 24780 }, { "epoch": 2.62, "learning_rate": 2.560143626570916e-06, "loss": 0.0011, "step": 24790 }, { "epoch": 2.62, "learning_rate": 2.552962298025135e-06, "loss": 0.0011, "step": 24800 }, { "epoch": 2.63, "learning_rate": 2.545780969479354e-06, "loss": 0.0014, "step": 24810 }, { "epoch": 2.63, "learning_rate": 2.5385996409335727e-06, "loss": 0.0011, "step": 24820 }, { "epoch": 2.63, "learning_rate": 2.5314183123877915e-06, "loss": 0.0012, "step": 24830 }, { "epoch": 2.63, "learning_rate": 2.5242369838420112e-06, "loss": 0.0011, "step": 24840 }, { "epoch": 2.63, "learning_rate": 2.51705565529623e-06, "loss": 0.0012, "step": 24850 }, { "epoch": 2.63, "learning_rate": 2.509874326750449e-06, "loss": 0.0013, "step": 24860 }, { "epoch": 2.63, "learning_rate": 2.502692998204668e-06, "loss": 0.0011, "step": 24870 }, { "epoch": 2.63, "learning_rate": 2.495511669658887e-06, "loss": 0.0011, "step": 24880 }, { "epoch": 2.63, "learning_rate": 2.488330341113106e-06, "loss": 0.0011, "step": 24890 }, { "epoch": 2.63, "learning_rate": 2.481149012567325e-06, "loss": 0.0011, "step": 24900 }, { "epoch": 2.64, "learning_rate": 2.4739676840215443e-06, "loss": 0.0013, "step": 24910 }, { "epoch": 2.64, "learning_rate": 2.466786355475763e-06, "loss": 0.0012, "step": 24920 }, { "epoch": 2.64, "learning_rate": 2.4596050269299824e-06, "loss": 0.0013, "step": 24930 }, { "epoch": 2.64, "learning_rate": 2.4524236983842012e-06, "loss": 0.0012, "step": 24940 }, { "epoch": 2.64, "learning_rate": 2.4452423698384205e-06, "loss": 0.0013, "step": 24950 }, { "epoch": 2.64, "learning_rate": 2.4380610412926393e-06, "loss": 0.0013, "step": 24960 }, { "epoch": 2.64, "learning_rate": 2.430879712746858e-06, "loss": 0.0012, "step": 24970 }, { "epoch": 2.64, "learning_rate": 2.4236983842010774e-06, "loss": 0.0012, "step": 24980 }, { "epoch": 2.64, "learning_rate": 2.4165170556552962e-06, "loss": 0.0011, "step": 24990 }, { "epoch": 2.65, "learning_rate": 2.4093357271095155e-06, "loss": 0.0013, "step": 25000 } ], "max_steps": 28350, "num_train_epochs": 3, "total_flos": 4.3523506176e+17, "trial_name": null, "trial_params": null }