{ "best_metric": 0.22539588809013367, "best_model_checkpoint": "./vit-mae-demo/checkpoint-69345", "epoch": 10.0, "global_step": 77050, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.2165844796262652e-08, "loss": 1.7322, "step": 10 }, { "epoch": 0.0, "learning_rate": 2.4331689592525304e-08, "loss": 1.6259, "step": 20 }, { "epoch": 0.0, "learning_rate": 3.6497534388787955e-08, "loss": 1.6849, "step": 30 }, { "epoch": 0.01, "learning_rate": 4.866337918505061e-08, "loss": 1.6124, "step": 40 }, { "epoch": 0.01, "learning_rate": 6.082922398131326e-08, "loss": 1.6617, "step": 50 }, { "epoch": 0.01, "learning_rate": 7.299506877757591e-08, "loss": 1.5997, "step": 60 }, { "epoch": 0.01, "learning_rate": 8.516091357383856e-08, "loss": 1.7401, "step": 70 }, { "epoch": 0.01, "learning_rate": 9.732675837010122e-08, "loss": 1.6155, "step": 80 }, { "epoch": 0.01, "learning_rate": 1.0949260316636386e-07, "loss": 1.5657, "step": 90 }, { "epoch": 0.01, "learning_rate": 1.2165844796262652e-07, "loss": 1.7435, "step": 100 }, { "epoch": 0.01, "learning_rate": 1.3382429275888916e-07, "loss": 1.7503, "step": 110 }, { "epoch": 0.02, "learning_rate": 1.4599013755515182e-07, "loss": 1.6788, "step": 120 }, { "epoch": 0.02, "learning_rate": 1.5815598235141445e-07, "loss": 1.5959, "step": 130 }, { "epoch": 0.02, "learning_rate": 1.703218271476771e-07, "loss": 1.6907, "step": 140 }, { "epoch": 0.02, "learning_rate": 1.8248767194393977e-07, "loss": 1.6337, "step": 150 }, { "epoch": 0.02, "learning_rate": 1.9465351674020243e-07, "loss": 1.6399, "step": 160 }, { "epoch": 0.02, "learning_rate": 2.0681936153646507e-07, "loss": 1.6009, "step": 170 }, { "epoch": 0.02, "learning_rate": 2.1898520633272773e-07, "loss": 1.4907, "step": 180 }, { "epoch": 0.02, "learning_rate": 2.311510511289904e-07, "loss": 1.4986, "step": 190 }, { "epoch": 0.03, "learning_rate": 2.4331689592525305e-07, "loss": 1.5683, "step": 200 }, { "epoch": 0.03, "learning_rate": 2.554827407215157e-07, "loss": 1.5328, "step": 210 }, { "epoch": 0.03, "learning_rate": 2.676485855177783e-07, "loss": 1.4055, "step": 220 }, { "epoch": 0.03, "learning_rate": 2.79814430314041e-07, "loss": 1.4164, "step": 230 }, { "epoch": 0.03, "learning_rate": 2.9198027511030364e-07, "loss": 1.3989, "step": 240 }, { "epoch": 0.03, "learning_rate": 3.0414611990656627e-07, "loss": 1.4021, "step": 250 }, { "epoch": 0.03, "learning_rate": 3.163119647028289e-07, "loss": 1.3266, "step": 260 }, { "epoch": 0.04, "learning_rate": 3.284778094990916e-07, "loss": 1.1322, "step": 270 }, { "epoch": 0.04, "learning_rate": 3.406436542953542e-07, "loss": 1.081, "step": 280 }, { "epoch": 0.04, "learning_rate": 3.528094990916169e-07, "loss": 0.9684, "step": 290 }, { "epoch": 0.04, "learning_rate": 3.6497534388787955e-07, "loss": 0.944, "step": 300 }, { "epoch": 0.04, "learning_rate": 3.7714118868414223e-07, "loss": 0.8615, "step": 310 }, { "epoch": 0.04, "learning_rate": 3.8930703348040487e-07, "loss": 0.8587, "step": 320 }, { "epoch": 0.04, "learning_rate": 4.0147287827666745e-07, "loss": 0.8252, "step": 330 }, { "epoch": 0.04, "learning_rate": 4.1363872307293014e-07, "loss": 0.7926, "step": 340 }, { "epoch": 0.05, "learning_rate": 4.2580456786919277e-07, "loss": 0.7762, "step": 350 }, { "epoch": 0.05, "learning_rate": 4.3797041266545546e-07, "loss": 0.7363, "step": 360 }, { "epoch": 0.05, "learning_rate": 4.501362574617181e-07, "loss": 0.719, "step": 370 }, { "epoch": 0.05, "learning_rate": 4.623021022579808e-07, "loss": 0.7276, "step": 380 }, { "epoch": 0.05, "learning_rate": 4.744679470542434e-07, "loss": 0.6823, "step": 390 }, { "epoch": 0.05, "learning_rate": 4.866337918505061e-07, "loss": 0.6613, "step": 400 }, { "epoch": 0.05, "learning_rate": 4.987996366467688e-07, "loss": 0.6724, "step": 410 }, { "epoch": 0.05, "learning_rate": 5.109654814430314e-07, "loss": 0.6752, "step": 420 }, { "epoch": 0.06, "learning_rate": 5.231313262392939e-07, "loss": 0.6083, "step": 430 }, { "epoch": 0.06, "learning_rate": 5.352971710355566e-07, "loss": 0.6163, "step": 440 }, { "epoch": 0.06, "learning_rate": 5.474630158318193e-07, "loss": 0.5894, "step": 450 }, { "epoch": 0.06, "learning_rate": 5.59628860628082e-07, "loss": 0.6046, "step": 460 }, { "epoch": 0.06, "learning_rate": 5.717947054243446e-07, "loss": 0.5486, "step": 470 }, { "epoch": 0.06, "learning_rate": 5.839605502206073e-07, "loss": 0.5622, "step": 480 }, { "epoch": 0.06, "learning_rate": 5.9612639501687e-07, "loss": 0.5639, "step": 490 }, { "epoch": 0.06, "learning_rate": 6.082922398131325e-07, "loss": 0.575, "step": 500 }, { "epoch": 0.07, "learning_rate": 6.204580846093952e-07, "loss": 0.5766, "step": 510 }, { "epoch": 0.07, "learning_rate": 6.326239294056578e-07, "loss": 0.5791, "step": 520 }, { "epoch": 0.07, "learning_rate": 6.447897742019206e-07, "loss": 0.56, "step": 530 }, { "epoch": 0.07, "learning_rate": 6.569556189981832e-07, "loss": 0.5412, "step": 540 }, { "epoch": 0.07, "learning_rate": 6.691214637944459e-07, "loss": 0.5453, "step": 550 }, { "epoch": 0.07, "learning_rate": 6.812873085907085e-07, "loss": 0.5484, "step": 560 }, { "epoch": 0.07, "learning_rate": 6.93453153386971e-07, "loss": 0.5314, "step": 570 }, { "epoch": 0.08, "learning_rate": 7.056189981832338e-07, "loss": 0.5343, "step": 580 }, { "epoch": 0.08, "learning_rate": 7.177848429794964e-07, "loss": 0.5481, "step": 590 }, { "epoch": 0.08, "learning_rate": 7.299506877757591e-07, "loss": 0.5392, "step": 600 }, { "epoch": 0.08, "learning_rate": 7.421165325720217e-07, "loss": 0.5056, "step": 610 }, { "epoch": 0.08, "learning_rate": 7.542823773682845e-07, "loss": 0.5147, "step": 620 }, { "epoch": 0.08, "learning_rate": 7.66448222164547e-07, "loss": 0.5131, "step": 630 }, { "epoch": 0.08, "learning_rate": 7.786140669608097e-07, "loss": 0.5294, "step": 640 }, { "epoch": 0.08, "learning_rate": 7.907799117570723e-07, "loss": 0.5048, "step": 650 }, { "epoch": 0.09, "learning_rate": 8.029457565533349e-07, "loss": 0.5148, "step": 660 }, { "epoch": 0.09, "learning_rate": 8.151116013495977e-07, "loss": 0.5106, "step": 670 }, { "epoch": 0.09, "learning_rate": 8.272774461458603e-07, "loss": 0.485, "step": 680 }, { "epoch": 0.09, "learning_rate": 8.39443290942123e-07, "loss": 0.499, "step": 690 }, { "epoch": 0.09, "learning_rate": 8.516091357383855e-07, "loss": 0.5061, "step": 700 }, { "epoch": 0.09, "learning_rate": 8.637749805346483e-07, "loss": 0.4786, "step": 710 }, { "epoch": 0.09, "learning_rate": 8.759408253309109e-07, "loss": 0.478, "step": 720 }, { "epoch": 0.09, "learning_rate": 8.881066701271736e-07, "loss": 0.4846, "step": 730 }, { "epoch": 0.1, "learning_rate": 9.002725149234362e-07, "loss": 0.4568, "step": 740 }, { "epoch": 0.1, "learning_rate": 9.124383597196988e-07, "loss": 0.4594, "step": 750 }, { "epoch": 0.1, "learning_rate": 9.246042045159616e-07, "loss": 0.4705, "step": 760 }, { "epoch": 0.1, "learning_rate": 9.367700493122241e-07, "loss": 0.4641, "step": 770 }, { "epoch": 0.1, "learning_rate": 9.489358941084868e-07, "loss": 0.442, "step": 780 }, { "epoch": 0.1, "learning_rate": 9.611017389047495e-07, "loss": 0.4849, "step": 790 }, { "epoch": 0.1, "learning_rate": 9.732675837010122e-07, "loss": 0.4622, "step": 800 }, { "epoch": 0.11, "learning_rate": 9.854334284972747e-07, "loss": 0.4543, "step": 810 }, { "epoch": 0.11, "learning_rate": 9.975992732935376e-07, "loss": 0.4427, "step": 820 }, { "epoch": 0.11, "learning_rate": 1.0097651180898e-06, "loss": 0.4411, "step": 830 }, { "epoch": 0.11, "learning_rate": 1.0219309628860627e-06, "loss": 0.43, "step": 840 }, { "epoch": 0.11, "learning_rate": 1.0340968076823254e-06, "loss": 0.4252, "step": 850 }, { "epoch": 0.11, "learning_rate": 1.0462626524785879e-06, "loss": 0.4124, "step": 860 }, { "epoch": 0.11, "learning_rate": 1.0584284972748508e-06, "loss": 0.4262, "step": 870 }, { "epoch": 0.11, "learning_rate": 1.0705943420711133e-06, "loss": 0.4206, "step": 880 }, { "epoch": 0.12, "learning_rate": 1.082760186867376e-06, "loss": 0.4337, "step": 890 }, { "epoch": 0.12, "learning_rate": 1.0949260316636386e-06, "loss": 0.4235, "step": 900 }, { "epoch": 0.12, "learning_rate": 1.1070918764599013e-06, "loss": 0.4429, "step": 910 }, { "epoch": 0.12, "learning_rate": 1.119257721256164e-06, "loss": 0.4163, "step": 920 }, { "epoch": 0.12, "learning_rate": 1.1314235660524267e-06, "loss": 0.4301, "step": 930 }, { "epoch": 0.12, "learning_rate": 1.1435894108486892e-06, "loss": 0.3982, "step": 940 }, { "epoch": 0.12, "learning_rate": 1.1557552556449519e-06, "loss": 0.3951, "step": 950 }, { "epoch": 0.12, "learning_rate": 1.1679211004412146e-06, "loss": 0.3963, "step": 960 }, { "epoch": 0.13, "learning_rate": 1.1800869452374772e-06, "loss": 0.4202, "step": 970 }, { "epoch": 0.13, "learning_rate": 1.19225279003374e-06, "loss": 0.3933, "step": 980 }, { "epoch": 0.13, "learning_rate": 1.2044186348300026e-06, "loss": 0.4028, "step": 990 }, { "epoch": 0.13, "learning_rate": 1.216584479626265e-06, "loss": 0.3852, "step": 1000 }, { "epoch": 0.13, "learning_rate": 1.2287503244225278e-06, "loss": 0.3905, "step": 1010 }, { "epoch": 0.13, "learning_rate": 1.2409161692187905e-06, "loss": 0.3803, "step": 1020 }, { "epoch": 0.13, "learning_rate": 1.2530820140150531e-06, "loss": 0.4151, "step": 1030 }, { "epoch": 0.13, "learning_rate": 1.2652478588113156e-06, "loss": 0.4012, "step": 1040 }, { "epoch": 0.14, "learning_rate": 1.2774137036075783e-06, "loss": 0.4081, "step": 1050 }, { "epoch": 0.14, "learning_rate": 1.2895795484038412e-06, "loss": 0.3819, "step": 1060 }, { "epoch": 0.14, "learning_rate": 1.3017453932001037e-06, "loss": 0.3888, "step": 1070 }, { "epoch": 0.14, "learning_rate": 1.3139112379963664e-06, "loss": 0.3658, "step": 1080 }, { "epoch": 0.14, "learning_rate": 1.326077082792629e-06, "loss": 0.3739, "step": 1090 }, { "epoch": 0.14, "learning_rate": 1.3382429275888917e-06, "loss": 0.3919, "step": 1100 }, { "epoch": 0.14, "learning_rate": 1.3504087723851542e-06, "loss": 0.3838, "step": 1110 }, { "epoch": 0.15, "learning_rate": 1.362574617181417e-06, "loss": 0.3737, "step": 1120 }, { "epoch": 0.15, "learning_rate": 1.3747404619776796e-06, "loss": 0.3719, "step": 1130 }, { "epoch": 0.15, "learning_rate": 1.386906306773942e-06, "loss": 0.3655, "step": 1140 }, { "epoch": 0.15, "learning_rate": 1.399072151570205e-06, "loss": 0.3825, "step": 1150 }, { "epoch": 0.15, "learning_rate": 1.4112379963664677e-06, "loss": 0.3889, "step": 1160 }, { "epoch": 0.15, "learning_rate": 1.4234038411627303e-06, "loss": 0.3692, "step": 1170 }, { "epoch": 0.15, "learning_rate": 1.4355696859589928e-06, "loss": 0.3772, "step": 1180 }, { "epoch": 0.15, "learning_rate": 1.4477355307552555e-06, "loss": 0.3596, "step": 1190 }, { "epoch": 0.16, "learning_rate": 1.4599013755515182e-06, "loss": 0.3695, "step": 1200 }, { "epoch": 0.16, "learning_rate": 1.4720672203477809e-06, "loss": 0.3405, "step": 1210 }, { "epoch": 0.16, "learning_rate": 1.4842330651440434e-06, "loss": 0.3501, "step": 1220 }, { "epoch": 0.16, "learning_rate": 1.496398909940306e-06, "loss": 0.3586, "step": 1230 }, { "epoch": 0.16, "learning_rate": 1.508564754736569e-06, "loss": 0.3521, "step": 1240 }, { "epoch": 0.16, "learning_rate": 1.5207305995328314e-06, "loss": 0.3674, "step": 1250 }, { "epoch": 0.16, "learning_rate": 1.532896444329094e-06, "loss": 0.367, "step": 1260 }, { "epoch": 0.16, "learning_rate": 1.5450622891253568e-06, "loss": 0.372, "step": 1270 }, { "epoch": 0.17, "learning_rate": 1.5572281339216195e-06, "loss": 0.3547, "step": 1280 }, { "epoch": 0.17, "learning_rate": 1.569393978717882e-06, "loss": 0.3524, "step": 1290 }, { "epoch": 0.17, "learning_rate": 1.5815598235141446e-06, "loss": 0.3573, "step": 1300 }, { "epoch": 0.17, "learning_rate": 1.5937256683104073e-06, "loss": 0.3578, "step": 1310 }, { "epoch": 0.17, "learning_rate": 1.6058915131066698e-06, "loss": 0.3601, "step": 1320 }, { "epoch": 0.17, "learning_rate": 1.6180573579029325e-06, "loss": 0.3671, "step": 1330 }, { "epoch": 0.17, "learning_rate": 1.6302232026991954e-06, "loss": 0.3526, "step": 1340 }, { "epoch": 0.18, "learning_rate": 1.642389047495458e-06, "loss": 0.3567, "step": 1350 }, { "epoch": 0.18, "learning_rate": 1.6545548922917205e-06, "loss": 0.3609, "step": 1360 }, { "epoch": 0.18, "learning_rate": 1.6667207370879832e-06, "loss": 0.3527, "step": 1370 }, { "epoch": 0.18, "learning_rate": 1.678886581884246e-06, "loss": 0.37, "step": 1380 }, { "epoch": 0.18, "learning_rate": 1.6910524266805086e-06, "loss": 0.3359, "step": 1390 }, { "epoch": 0.18, "learning_rate": 1.703218271476771e-06, "loss": 0.3317, "step": 1400 }, { "epoch": 0.18, "learning_rate": 1.7153841162730338e-06, "loss": 0.357, "step": 1410 }, { "epoch": 0.18, "learning_rate": 1.7275499610692967e-06, "loss": 0.3508, "step": 1420 }, { "epoch": 0.19, "learning_rate": 1.7397158058655591e-06, "loss": 0.3426, "step": 1430 }, { "epoch": 0.19, "learning_rate": 1.7518816506618218e-06, "loss": 0.3401, "step": 1440 }, { "epoch": 0.19, "learning_rate": 1.7640474954580845e-06, "loss": 0.3296, "step": 1450 }, { "epoch": 0.19, "learning_rate": 1.7762133402543472e-06, "loss": 0.3513, "step": 1460 }, { "epoch": 0.19, "learning_rate": 1.7883791850506097e-06, "loss": 0.3492, "step": 1470 }, { "epoch": 0.19, "learning_rate": 1.8005450298468724e-06, "loss": 0.3207, "step": 1480 }, { "epoch": 0.19, "learning_rate": 1.812710874643135e-06, "loss": 0.3185, "step": 1490 }, { "epoch": 0.19, "learning_rate": 1.8248767194393975e-06, "loss": 0.3367, "step": 1500 }, { "epoch": 0.2, "learning_rate": 1.8370425642356602e-06, "loss": 0.3397, "step": 1510 }, { "epoch": 0.2, "learning_rate": 1.8492084090319231e-06, "loss": 0.3392, "step": 1520 }, { "epoch": 0.2, "learning_rate": 1.8613742538281858e-06, "loss": 0.3378, "step": 1530 }, { "epoch": 0.2, "learning_rate": 1.8735400986244483e-06, "loss": 0.3283, "step": 1540 }, { "epoch": 0.2, "learning_rate": 1.885705943420711e-06, "loss": 0.3328, "step": 1550 }, { "epoch": 0.2, "learning_rate": 1.8978717882169736e-06, "loss": 0.3308, "step": 1560 }, { "epoch": 0.2, "learning_rate": 1.9100376330132365e-06, "loss": 0.3166, "step": 1570 }, { "epoch": 0.21, "learning_rate": 1.922203477809499e-06, "loss": 0.3261, "step": 1580 }, { "epoch": 0.21, "learning_rate": 1.9343693226057615e-06, "loss": 0.3384, "step": 1590 }, { "epoch": 0.21, "learning_rate": 1.9465351674020244e-06, "loss": 0.3184, "step": 1600 }, { "epoch": 0.21, "learning_rate": 1.958701012198287e-06, "loss": 0.3265, "step": 1610 }, { "epoch": 0.21, "learning_rate": 1.9708668569945493e-06, "loss": 0.3195, "step": 1620 }, { "epoch": 0.21, "learning_rate": 1.9830327017908122e-06, "loss": 0.3211, "step": 1630 }, { "epoch": 0.21, "learning_rate": 1.995198546587075e-06, "loss": 0.3401, "step": 1640 }, { "epoch": 0.21, "learning_rate": 2.007364391383337e-06, "loss": 0.3358, "step": 1650 }, { "epoch": 0.22, "learning_rate": 2.0195302361796e-06, "loss": 0.3337, "step": 1660 }, { "epoch": 0.22, "learning_rate": 2.031696080975863e-06, "loss": 0.307, "step": 1670 }, { "epoch": 0.22, "learning_rate": 2.0438619257721255e-06, "loss": 0.3257, "step": 1680 }, { "epoch": 0.22, "learning_rate": 2.056027770568388e-06, "loss": 0.3018, "step": 1690 }, { "epoch": 0.22, "learning_rate": 2.068193615364651e-06, "loss": 0.3279, "step": 1700 }, { "epoch": 0.22, "learning_rate": 2.0803594601609133e-06, "loss": 0.3134, "step": 1710 }, { "epoch": 0.22, "learning_rate": 2.0925253049571758e-06, "loss": 0.3233, "step": 1720 }, { "epoch": 0.22, "learning_rate": 2.1046911497534387e-06, "loss": 0.3276, "step": 1730 }, { "epoch": 0.23, "learning_rate": 2.1168569945497016e-06, "loss": 0.3147, "step": 1740 }, { "epoch": 0.23, "learning_rate": 2.129022839345964e-06, "loss": 0.3289, "step": 1750 }, { "epoch": 0.23, "learning_rate": 2.1411886841422265e-06, "loss": 0.3223, "step": 1760 }, { "epoch": 0.23, "learning_rate": 2.1533545289384894e-06, "loss": 0.3142, "step": 1770 }, { "epoch": 0.23, "learning_rate": 2.165520373734752e-06, "loss": 0.3249, "step": 1780 }, { "epoch": 0.23, "learning_rate": 2.1776862185310144e-06, "loss": 0.3206, "step": 1790 }, { "epoch": 0.23, "learning_rate": 2.1898520633272773e-06, "loss": 0.3023, "step": 1800 }, { "epoch": 0.23, "learning_rate": 2.2020179081235398e-06, "loss": 0.3175, "step": 1810 }, { "epoch": 0.24, "learning_rate": 2.2141837529198027e-06, "loss": 0.3074, "step": 1820 }, { "epoch": 0.24, "learning_rate": 2.226349597716065e-06, "loss": 0.2978, "step": 1830 }, { "epoch": 0.24, "learning_rate": 2.238515442512328e-06, "loss": 0.2986, "step": 1840 }, { "epoch": 0.24, "learning_rate": 2.2506812873085905e-06, "loss": 0.314, "step": 1850 }, { "epoch": 0.24, "learning_rate": 2.2628471321048534e-06, "loss": 0.3182, "step": 1860 }, { "epoch": 0.24, "learning_rate": 2.275012976901116e-06, "loss": 0.3048, "step": 1870 }, { "epoch": 0.24, "learning_rate": 2.2871788216973784e-06, "loss": 0.2938, "step": 1880 }, { "epoch": 0.25, "learning_rate": 2.2993446664936413e-06, "loss": 0.3081, "step": 1890 }, { "epoch": 0.25, "learning_rate": 2.3115105112899037e-06, "loss": 0.3152, "step": 1900 }, { "epoch": 0.25, "learning_rate": 2.323676356086166e-06, "loss": 0.3088, "step": 1910 }, { "epoch": 0.25, "learning_rate": 2.335842200882429e-06, "loss": 0.3165, "step": 1920 }, { "epoch": 0.25, "learning_rate": 2.3480080456786916e-06, "loss": 0.3125, "step": 1930 }, { "epoch": 0.25, "learning_rate": 2.3601738904749545e-06, "loss": 0.3006, "step": 1940 }, { "epoch": 0.25, "learning_rate": 2.372339735271217e-06, "loss": 0.2994, "step": 1950 }, { "epoch": 0.25, "learning_rate": 2.38450558006748e-06, "loss": 0.303, "step": 1960 }, { "epoch": 0.26, "learning_rate": 2.3966714248637423e-06, "loss": 0.2899, "step": 1970 }, { "epoch": 0.26, "learning_rate": 2.4088372696600052e-06, "loss": 0.2953, "step": 1980 }, { "epoch": 0.26, "learning_rate": 2.4210031144562673e-06, "loss": 0.302, "step": 1990 }, { "epoch": 0.26, "learning_rate": 2.43316895925253e-06, "loss": 0.2988, "step": 2000 }, { "epoch": 0.26, "learning_rate": 2.4453348040487926e-06, "loss": 0.2828, "step": 2010 }, { "epoch": 0.26, "learning_rate": 2.4575006488450555e-06, "loss": 0.3016, "step": 2020 }, { "epoch": 0.26, "learning_rate": 2.4696664936413184e-06, "loss": 0.2975, "step": 2030 }, { "epoch": 0.26, "learning_rate": 2.481832338437581e-06, "loss": 0.2861, "step": 2040 }, { "epoch": 0.27, "learning_rate": 2.493998183233844e-06, "loss": 0.2819, "step": 2050 }, { "epoch": 0.27, "learning_rate": 2.5061640280301063e-06, "loss": 0.2852, "step": 2060 }, { "epoch": 0.27, "learning_rate": 2.5183298728263688e-06, "loss": 0.3012, "step": 2070 }, { "epoch": 0.27, "learning_rate": 2.5304957176226312e-06, "loss": 0.2935, "step": 2080 }, { "epoch": 0.27, "learning_rate": 2.542661562418894e-06, "loss": 0.2899, "step": 2090 }, { "epoch": 0.27, "learning_rate": 2.5548274072151566e-06, "loss": 0.301, "step": 2100 }, { "epoch": 0.27, "learning_rate": 2.5669932520114195e-06, "loss": 0.2887, "step": 2110 }, { "epoch": 0.28, "learning_rate": 2.5791590968076824e-06, "loss": 0.281, "step": 2120 }, { "epoch": 0.28, "learning_rate": 2.591324941603945e-06, "loss": 0.2914, "step": 2130 }, { "epoch": 0.28, "learning_rate": 2.6034907864002074e-06, "loss": 0.2945, "step": 2140 }, { "epoch": 0.28, "learning_rate": 2.61565663119647e-06, "loss": 0.2962, "step": 2150 }, { "epoch": 0.28, "learning_rate": 2.6278224759927327e-06, "loss": 0.2994, "step": 2160 }, { "epoch": 0.28, "learning_rate": 2.6399883207889952e-06, "loss": 0.3005, "step": 2170 }, { "epoch": 0.28, "learning_rate": 2.652154165585258e-06, "loss": 0.3009, "step": 2180 }, { "epoch": 0.28, "learning_rate": 2.664320010381521e-06, "loss": 0.294, "step": 2190 }, { "epoch": 0.29, "learning_rate": 2.6764858551777835e-06, "loss": 0.299, "step": 2200 }, { "epoch": 0.29, "learning_rate": 2.6886516999740455e-06, "loss": 0.2982, "step": 2210 }, { "epoch": 0.29, "learning_rate": 2.7008175447703084e-06, "loss": 0.2988, "step": 2220 }, { "epoch": 0.29, "learning_rate": 2.7129833895665713e-06, "loss": 0.2782, "step": 2230 }, { "epoch": 0.29, "learning_rate": 2.725149234362834e-06, "loss": 0.3055, "step": 2240 }, { "epoch": 0.29, "learning_rate": 2.7373150791590967e-06, "loss": 0.2867, "step": 2250 }, { "epoch": 0.29, "learning_rate": 2.749480923955359e-06, "loss": 0.3086, "step": 2260 }, { "epoch": 0.29, "learning_rate": 2.761646768751622e-06, "loss": 0.3043, "step": 2270 }, { "epoch": 0.3, "learning_rate": 2.773812613547884e-06, "loss": 0.3107, "step": 2280 }, { "epoch": 0.3, "learning_rate": 2.785978458344147e-06, "loss": 0.2918, "step": 2290 }, { "epoch": 0.3, "learning_rate": 2.79814430314041e-06, "loss": 0.2954, "step": 2300 }, { "epoch": 0.3, "learning_rate": 2.8103101479366724e-06, "loss": 0.299, "step": 2310 }, { "epoch": 0.3, "learning_rate": 2.8224759927329353e-06, "loss": 0.284, "step": 2320 }, { "epoch": 0.3, "learning_rate": 2.8346418375291978e-06, "loss": 0.2941, "step": 2330 }, { "epoch": 0.3, "learning_rate": 2.8468076823254607e-06, "loss": 0.291, "step": 2340 }, { "epoch": 0.3, "learning_rate": 2.8589735271217227e-06, "loss": 0.2841, "step": 2350 }, { "epoch": 0.31, "learning_rate": 2.8711393719179856e-06, "loss": 0.2818, "step": 2360 }, { "epoch": 0.31, "learning_rate": 2.883305216714248e-06, "loss": 0.2786, "step": 2370 }, { "epoch": 0.31, "learning_rate": 2.895471061510511e-06, "loss": 0.2985, "step": 2380 }, { "epoch": 0.31, "learning_rate": 2.907636906306774e-06, "loss": 0.2605, "step": 2390 }, { "epoch": 0.31, "learning_rate": 2.9198027511030364e-06, "loss": 0.279, "step": 2400 }, { "epoch": 0.31, "learning_rate": 2.9319685958992993e-06, "loss": 0.2727, "step": 2410 }, { "epoch": 0.31, "learning_rate": 2.9441344406955618e-06, "loss": 0.2943, "step": 2420 }, { "epoch": 0.32, "learning_rate": 2.9563002854918242e-06, "loss": 0.291, "step": 2430 }, { "epoch": 0.32, "learning_rate": 2.9684661302880867e-06, "loss": 0.2786, "step": 2440 }, { "epoch": 0.32, "learning_rate": 2.9806319750843496e-06, "loss": 0.2737, "step": 2450 }, { "epoch": 0.32, "learning_rate": 2.992797819880612e-06, "loss": 0.3078, "step": 2460 }, { "epoch": 0.32, "learning_rate": 3.004963664676875e-06, "loss": 0.2861, "step": 2470 }, { "epoch": 0.32, "learning_rate": 3.017129509473138e-06, "loss": 0.2963, "step": 2480 }, { "epoch": 0.32, "learning_rate": 3.0292953542694003e-06, "loss": 0.2951, "step": 2490 }, { "epoch": 0.32, "learning_rate": 3.041461199065663e-06, "loss": 0.282, "step": 2500 }, { "epoch": 0.33, "learning_rate": 3.0536270438619253e-06, "loss": 0.2949, "step": 2510 }, { "epoch": 0.33, "learning_rate": 3.065792888658188e-06, "loss": 0.2726, "step": 2520 }, { "epoch": 0.33, "learning_rate": 3.0779587334544507e-06, "loss": 0.2711, "step": 2530 }, { "epoch": 0.33, "learning_rate": 3.0901245782507136e-06, "loss": 0.2737, "step": 2540 }, { "epoch": 0.33, "learning_rate": 3.102290423046976e-06, "loss": 0.2751, "step": 2550 }, { "epoch": 0.33, "learning_rate": 3.114456267843239e-06, "loss": 0.283, "step": 2560 }, { "epoch": 0.33, "learning_rate": 3.126622112639501e-06, "loss": 0.2712, "step": 2570 }, { "epoch": 0.33, "learning_rate": 3.138787957435764e-06, "loss": 0.2908, "step": 2580 }, { "epoch": 0.34, "learning_rate": 3.150953802232027e-06, "loss": 0.2836, "step": 2590 }, { "epoch": 0.34, "learning_rate": 3.1631196470282893e-06, "loss": 0.2728, "step": 2600 }, { "epoch": 0.34, "learning_rate": 3.175285491824552e-06, "loss": 0.2671, "step": 2610 }, { "epoch": 0.34, "learning_rate": 3.1874513366208146e-06, "loss": 0.2863, "step": 2620 }, { "epoch": 0.34, "learning_rate": 3.1996171814170775e-06, "loss": 0.2922, "step": 2630 }, { "epoch": 0.34, "learning_rate": 3.2117830262133396e-06, "loss": 0.2768, "step": 2640 }, { "epoch": 0.34, "learning_rate": 3.2239488710096025e-06, "loss": 0.2825, "step": 2650 }, { "epoch": 0.35, "learning_rate": 3.236114715805865e-06, "loss": 0.269, "step": 2660 }, { "epoch": 0.35, "learning_rate": 3.248280560602128e-06, "loss": 0.2775, "step": 2670 }, { "epoch": 0.35, "learning_rate": 3.2604464053983908e-06, "loss": 0.2625, "step": 2680 }, { "epoch": 0.35, "learning_rate": 3.2726122501946532e-06, "loss": 0.2833, "step": 2690 }, { "epoch": 0.35, "learning_rate": 3.284778094990916e-06, "loss": 0.2644, "step": 2700 }, { "epoch": 0.35, "learning_rate": 3.2969439397871786e-06, "loss": 0.2676, "step": 2710 }, { "epoch": 0.35, "learning_rate": 3.309109784583441e-06, "loss": 0.2816, "step": 2720 }, { "epoch": 0.35, "learning_rate": 3.3212756293797036e-06, "loss": 0.2995, "step": 2730 }, { "epoch": 0.36, "learning_rate": 3.3334414741759665e-06, "loss": 0.2765, "step": 2740 }, { "epoch": 0.36, "learning_rate": 3.3456073189722294e-06, "loss": 0.2821, "step": 2750 }, { "epoch": 0.36, "learning_rate": 3.357773163768492e-06, "loss": 0.2869, "step": 2760 }, { "epoch": 0.36, "learning_rate": 3.3699390085647547e-06, "loss": 0.2864, "step": 2770 }, { "epoch": 0.36, "learning_rate": 3.382104853361017e-06, "loss": 0.2888, "step": 2780 }, { "epoch": 0.36, "learning_rate": 3.3942706981572797e-06, "loss": 0.2767, "step": 2790 }, { "epoch": 0.36, "learning_rate": 3.406436542953542e-06, "loss": 0.2843, "step": 2800 }, { "epoch": 0.36, "learning_rate": 3.418602387749805e-06, "loss": 0.2565, "step": 2810 }, { "epoch": 0.37, "learning_rate": 3.4307682325460675e-06, "loss": 0.2696, "step": 2820 }, { "epoch": 0.37, "learning_rate": 3.4429340773423304e-06, "loss": 0.2721, "step": 2830 }, { "epoch": 0.37, "learning_rate": 3.4550999221385933e-06, "loss": 0.2799, "step": 2840 }, { "epoch": 0.37, "learning_rate": 3.467265766934856e-06, "loss": 0.2737, "step": 2850 }, { "epoch": 0.37, "learning_rate": 3.4794316117311183e-06, "loss": 0.27, "step": 2860 }, { "epoch": 0.37, "learning_rate": 3.4915974565273808e-06, "loss": 0.2831, "step": 2870 }, { "epoch": 0.37, "learning_rate": 3.5037633013236437e-06, "loss": 0.266, "step": 2880 }, { "epoch": 0.38, "learning_rate": 3.515929146119906e-06, "loss": 0.2909, "step": 2890 }, { "epoch": 0.38, "learning_rate": 3.528094990916169e-06, "loss": 0.2722, "step": 2900 }, { "epoch": 0.38, "learning_rate": 3.5402608357124315e-06, "loss": 0.275, "step": 2910 }, { "epoch": 0.38, "learning_rate": 3.5524266805086944e-06, "loss": 0.2444, "step": 2920 }, { "epoch": 0.38, "learning_rate": 3.5645925253049565e-06, "loss": 0.2829, "step": 2930 }, { "epoch": 0.38, "learning_rate": 3.5767583701012193e-06, "loss": 0.2322, "step": 2940 }, { "epoch": 0.38, "learning_rate": 3.5889242148974822e-06, "loss": 0.2697, "step": 2950 }, { "epoch": 0.38, "learning_rate": 3.6010900596937447e-06, "loss": 0.2598, "step": 2960 }, { "epoch": 0.39, "learning_rate": 3.6132559044900076e-06, "loss": 0.2838, "step": 2970 }, { "epoch": 0.39, "learning_rate": 3.62542174928627e-06, "loss": 0.2592, "step": 2980 }, { "epoch": 0.39, "learning_rate": 3.637587594082533e-06, "loss": 0.2633, "step": 2990 }, { "epoch": 0.39, "learning_rate": 3.649753438878795e-06, "loss": 0.2659, "step": 3000 }, { "epoch": 0.39, "learning_rate": 3.661919283675058e-06, "loss": 0.2817, "step": 3010 }, { "epoch": 0.39, "learning_rate": 3.6740851284713204e-06, "loss": 0.2807, "step": 3020 }, { "epoch": 0.39, "learning_rate": 3.6862509732675833e-06, "loss": 0.2746, "step": 3030 }, { "epoch": 0.39, "learning_rate": 3.6984168180638462e-06, "loss": 0.2784, "step": 3040 }, { "epoch": 0.4, "learning_rate": 3.7105826628601087e-06, "loss": 0.266, "step": 3050 }, { "epoch": 0.4, "learning_rate": 3.7227485076563716e-06, "loss": 0.2736, "step": 3060 }, { "epoch": 0.4, "learning_rate": 3.734914352452634e-06, "loss": 0.2724, "step": 3070 }, { "epoch": 0.4, "learning_rate": 3.7470801972488965e-06, "loss": 0.2551, "step": 3080 }, { "epoch": 0.4, "learning_rate": 3.759246042045159e-06, "loss": 0.2706, "step": 3090 }, { "epoch": 0.4, "learning_rate": 3.771411886841422e-06, "loss": 0.2619, "step": 3100 }, { "epoch": 0.4, "learning_rate": 3.7835777316376844e-06, "loss": 0.2736, "step": 3110 }, { "epoch": 0.4, "learning_rate": 3.7957435764339473e-06, "loss": 0.2643, "step": 3120 }, { "epoch": 0.41, "learning_rate": 3.80790942123021e-06, "loss": 0.256, "step": 3130 }, { "epoch": 0.41, "learning_rate": 3.820075266026473e-06, "loss": 0.2767, "step": 3140 }, { "epoch": 0.41, "learning_rate": 3.832241110822735e-06, "loss": 0.2808, "step": 3150 }, { "epoch": 0.41, "learning_rate": 3.844406955618998e-06, "loss": 0.2673, "step": 3160 }, { "epoch": 0.41, "learning_rate": 3.8565728004152605e-06, "loss": 0.2686, "step": 3170 }, { "epoch": 0.41, "learning_rate": 3.868738645211523e-06, "loss": 0.2845, "step": 3180 }, { "epoch": 0.41, "learning_rate": 3.8809044900077855e-06, "loss": 0.2796, "step": 3190 }, { "epoch": 0.42, "learning_rate": 3.893070334804049e-06, "loss": 0.2749, "step": 3200 }, { "epoch": 0.42, "learning_rate": 3.905236179600311e-06, "loss": 0.2513, "step": 3210 }, { "epoch": 0.42, "learning_rate": 3.917402024396574e-06, "loss": 0.2717, "step": 3220 }, { "epoch": 0.42, "learning_rate": 3.929567869192836e-06, "loss": 0.2684, "step": 3230 }, { "epoch": 0.42, "learning_rate": 3.941733713989099e-06, "loss": 0.2682, "step": 3240 }, { "epoch": 0.42, "learning_rate": 3.953899558785362e-06, "loss": 0.2574, "step": 3250 }, { "epoch": 0.42, "learning_rate": 3.9660654035816245e-06, "loss": 0.2821, "step": 3260 }, { "epoch": 0.42, "learning_rate": 3.978231248377887e-06, "loss": 0.2689, "step": 3270 }, { "epoch": 0.43, "learning_rate": 3.99039709317415e-06, "loss": 0.2664, "step": 3280 }, { "epoch": 0.43, "learning_rate": 4.002562937970412e-06, "loss": 0.2575, "step": 3290 }, { "epoch": 0.43, "learning_rate": 4.014728782766674e-06, "loss": 0.2667, "step": 3300 }, { "epoch": 0.43, "learning_rate": 4.026894627562938e-06, "loss": 0.2564, "step": 3310 }, { "epoch": 0.43, "learning_rate": 4.0390604723592e-06, "loss": 0.2674, "step": 3320 }, { "epoch": 0.43, "learning_rate": 4.051226317155463e-06, "loss": 0.2537, "step": 3330 }, { "epoch": 0.43, "learning_rate": 4.063392161951726e-06, "loss": 0.2625, "step": 3340 }, { "epoch": 0.43, "learning_rate": 4.0755580067479885e-06, "loss": 0.2639, "step": 3350 }, { "epoch": 0.44, "learning_rate": 4.087723851544251e-06, "loss": 0.2641, "step": 3360 }, { "epoch": 0.44, "learning_rate": 4.099889696340513e-06, "loss": 0.2742, "step": 3370 }, { "epoch": 0.44, "learning_rate": 4.112055541136776e-06, "loss": 0.2622, "step": 3380 }, { "epoch": 0.44, "learning_rate": 4.124221385933038e-06, "loss": 0.2546, "step": 3390 }, { "epoch": 0.44, "learning_rate": 4.136387230729302e-06, "loss": 0.2676, "step": 3400 }, { "epoch": 0.44, "learning_rate": 4.148553075525564e-06, "loss": 0.2801, "step": 3410 }, { "epoch": 0.44, "learning_rate": 4.160718920321827e-06, "loss": 0.2611, "step": 3420 }, { "epoch": 0.45, "learning_rate": 4.17288476511809e-06, "loss": 0.2517, "step": 3430 }, { "epoch": 0.45, "learning_rate": 4.1850506099143516e-06, "loss": 0.252, "step": 3440 }, { "epoch": 0.45, "learning_rate": 4.197216454710615e-06, "loss": 0.2786, "step": 3450 }, { "epoch": 0.45, "learning_rate": 4.209382299506877e-06, "loss": 0.2727, "step": 3460 }, { "epoch": 0.45, "learning_rate": 4.22154814430314e-06, "loss": 0.2667, "step": 3470 }, { "epoch": 0.45, "learning_rate": 4.233713989099403e-06, "loss": 0.2639, "step": 3480 }, { "epoch": 0.45, "learning_rate": 4.245879833895666e-06, "loss": 0.2619, "step": 3490 }, { "epoch": 0.45, "learning_rate": 4.258045678691928e-06, "loss": 0.2663, "step": 3500 }, { "epoch": 0.46, "learning_rate": 4.270211523488191e-06, "loss": 0.267, "step": 3510 }, { "epoch": 0.46, "learning_rate": 4.282377368284453e-06, "loss": 0.2548, "step": 3520 }, { "epoch": 0.46, "learning_rate": 4.2945432130807155e-06, "loss": 0.2707, "step": 3530 }, { "epoch": 0.46, "learning_rate": 4.306709057876979e-06, "loss": 0.269, "step": 3540 }, { "epoch": 0.46, "learning_rate": 4.318874902673241e-06, "loss": 0.2336, "step": 3550 }, { "epoch": 0.46, "learning_rate": 4.331040747469504e-06, "loss": 0.2545, "step": 3560 }, { "epoch": 0.46, "learning_rate": 4.343206592265767e-06, "loss": 0.2723, "step": 3570 }, { "epoch": 0.46, "learning_rate": 4.355372437062029e-06, "loss": 0.2787, "step": 3580 }, { "epoch": 0.47, "learning_rate": 4.367538281858292e-06, "loss": 0.269, "step": 3590 }, { "epoch": 0.47, "learning_rate": 4.3797041266545546e-06, "loss": 0.2574, "step": 3600 }, { "epoch": 0.47, "learning_rate": 4.391869971450817e-06, "loss": 0.2436, "step": 3610 }, { "epoch": 0.47, "learning_rate": 4.4040358162470795e-06, "loss": 0.2706, "step": 3620 }, { "epoch": 0.47, "learning_rate": 4.416201661043343e-06, "loss": 0.265, "step": 3630 }, { "epoch": 0.47, "learning_rate": 4.428367505839605e-06, "loss": 0.2582, "step": 3640 }, { "epoch": 0.47, "learning_rate": 4.440533350635868e-06, "loss": 0.2706, "step": 3650 }, { "epoch": 0.48, "learning_rate": 4.45269919543213e-06, "loss": 0.2724, "step": 3660 }, { "epoch": 0.48, "learning_rate": 4.464865040228393e-06, "loss": 0.2607, "step": 3670 }, { "epoch": 0.48, "learning_rate": 4.477030885024656e-06, "loss": 0.2631, "step": 3680 }, { "epoch": 0.48, "learning_rate": 4.4891967298209185e-06, "loss": 0.2676, "step": 3690 }, { "epoch": 0.48, "learning_rate": 4.501362574617181e-06, "loss": 0.249, "step": 3700 }, { "epoch": 0.48, "learning_rate": 4.5135284194134435e-06, "loss": 0.2561, "step": 3710 }, { "epoch": 0.48, "learning_rate": 4.525694264209707e-06, "loss": 0.2618, "step": 3720 }, { "epoch": 0.48, "learning_rate": 4.5378601090059684e-06, "loss": 0.2732, "step": 3730 }, { "epoch": 0.49, "learning_rate": 4.550025953802232e-06, "loss": 0.2557, "step": 3740 }, { "epoch": 0.49, "learning_rate": 4.562191798598494e-06, "loss": 0.263, "step": 3750 }, { "epoch": 0.49, "learning_rate": 4.574357643394757e-06, "loss": 0.2599, "step": 3760 }, { "epoch": 0.49, "learning_rate": 4.58652348819102e-06, "loss": 0.2489, "step": 3770 }, { "epoch": 0.49, "learning_rate": 4.5986893329872825e-06, "loss": 0.2806, "step": 3780 }, { "epoch": 0.49, "learning_rate": 4.610855177783545e-06, "loss": 0.2504, "step": 3790 }, { "epoch": 0.49, "learning_rate": 4.6230210225798075e-06, "loss": 0.2742, "step": 3800 }, { "epoch": 0.49, "learning_rate": 4.63518686737607e-06, "loss": 0.2615, "step": 3810 }, { "epoch": 0.5, "learning_rate": 4.647352712172332e-06, "loss": 0.2617, "step": 3820 }, { "epoch": 0.5, "learning_rate": 4.659518556968596e-06, "loss": 0.2495, "step": 3830 }, { "epoch": 0.5, "learning_rate": 4.671684401764858e-06, "loss": 0.2526, "step": 3840 }, { "epoch": 0.5, "learning_rate": 4.683850246561121e-06, "loss": 0.2415, "step": 3850 }, { "epoch": 0.5, "learning_rate": 4.68749989422317e-06, "loss": 0.2485, "step": 3860 }, { "epoch": 0.5, "learning_rate": 4.687499376132596e-06, "loss": 0.2678, "step": 3870 }, { "epoch": 0.5, "learning_rate": 4.687498426299977e-06, "loss": 0.2463, "step": 3880 }, { "epoch": 0.5, "learning_rate": 4.687497044725488e-06, "loss": 0.265, "step": 3890 }, { "epoch": 0.51, "learning_rate": 4.687495231409383e-06, "loss": 0.2719, "step": 3900 }, { "epoch": 0.51, "learning_rate": 4.687492986351995e-06, "loss": 0.2564, "step": 3910 }, { "epoch": 0.51, "learning_rate": 4.687490309553739e-06, "loss": 0.2483, "step": 3920 }, { "epoch": 0.51, "learning_rate": 4.687487201015109e-06, "loss": 0.2488, "step": 3930 }, { "epoch": 0.51, "learning_rate": 4.687483660736675e-06, "loss": 0.2584, "step": 3940 }, { "epoch": 0.51, "learning_rate": 4.687479688719092e-06, "loss": 0.2625, "step": 3950 }, { "epoch": 0.51, "learning_rate": 4.6874752849630885e-06, "loss": 0.2705, "step": 3960 }, { "epoch": 0.52, "learning_rate": 4.6874704494694785e-06, "loss": 0.2582, "step": 3970 }, { "epoch": 0.52, "learning_rate": 4.68746518223915e-06, "loss": 0.2656, "step": 3980 }, { "epoch": 0.52, "learning_rate": 4.687459483273077e-06, "loss": 0.2676, "step": 3990 }, { "epoch": 0.52, "learning_rate": 4.687453352572306e-06, "loss": 0.2476, "step": 4000 }, { "epoch": 0.52, "learning_rate": 4.687446790137968e-06, "loss": 0.2599, "step": 4010 }, { "epoch": 0.52, "learning_rate": 4.687439795971271e-06, "loss": 0.2548, "step": 4020 }, { "epoch": 0.52, "learning_rate": 4.687432370073504e-06, "loss": 0.253, "step": 4030 }, { "epoch": 0.52, "learning_rate": 4.687424512446036e-06, "loss": 0.2666, "step": 4040 }, { "epoch": 0.53, "learning_rate": 4.687416223090312e-06, "loss": 0.257, "step": 4050 }, { "epoch": 0.53, "learning_rate": 4.68740750200786e-06, "loss": 0.2616, "step": 4060 }, { "epoch": 0.53, "learning_rate": 4.687398349200288e-06, "loss": 0.2737, "step": 4070 }, { "epoch": 0.53, "learning_rate": 4.687388764669279e-06, "loss": 0.2563, "step": 4080 }, { "epoch": 0.53, "learning_rate": 4.687378748416601e-06, "loss": 0.2738, "step": 4090 }, { "epoch": 0.53, "learning_rate": 4.687368300444099e-06, "loss": 0.2546, "step": 4100 }, { "epoch": 0.53, "learning_rate": 4.687357420753696e-06, "loss": 0.2597, "step": 4110 }, { "epoch": 0.53, "learning_rate": 4.687346109347398e-06, "loss": 0.2448, "step": 4120 }, { "epoch": 0.54, "learning_rate": 4.687334366227287e-06, "loss": 0.2663, "step": 4130 }, { "epoch": 0.54, "learning_rate": 4.6873221913955285e-06, "loss": 0.2443, "step": 4140 }, { "epoch": 0.54, "learning_rate": 4.687309584854363e-06, "loss": 0.2609, "step": 4150 }, { "epoch": 0.54, "learning_rate": 4.687296546606113e-06, "loss": 0.2481, "step": 4160 }, { "epoch": 0.54, "learning_rate": 4.687283076653182e-06, "loss": 0.2519, "step": 4170 }, { "epoch": 0.54, "learning_rate": 4.687269174998049e-06, "loss": 0.2631, "step": 4180 }, { "epoch": 0.54, "learning_rate": 4.687254841643276e-06, "loss": 0.2638, "step": 4190 }, { "epoch": 0.55, "learning_rate": 4.687240076591504e-06, "loss": 0.2405, "step": 4200 }, { "epoch": 0.55, "learning_rate": 4.6872248798454515e-06, "loss": 0.2657, "step": 4210 }, { "epoch": 0.55, "learning_rate": 4.687209251407919e-06, "loss": 0.2507, "step": 4220 }, { "epoch": 0.55, "learning_rate": 4.687193191281784e-06, "loss": 0.2568, "step": 4230 }, { "epoch": 0.55, "learning_rate": 4.687176699470007e-06, "loss": 0.2462, "step": 4240 }, { "epoch": 0.55, "learning_rate": 4.6871597759756245e-06, "loss": 0.2668, "step": 4250 }, { "epoch": 0.55, "learning_rate": 4.687142420801754e-06, "loss": 0.2669, "step": 4260 }, { "epoch": 0.55, "learning_rate": 4.687124633951593e-06, "loss": 0.2379, "step": 4270 }, { "epoch": 0.56, "learning_rate": 4.687106415428418e-06, "loss": 0.2677, "step": 4280 }, { "epoch": 0.56, "learning_rate": 4.687087765235584e-06, "loss": 0.2428, "step": 4290 }, { "epoch": 0.56, "learning_rate": 4.687068683376529e-06, "loss": 0.2608, "step": 4300 }, { "epoch": 0.56, "learning_rate": 4.687049169854766e-06, "loss": 0.2533, "step": 4310 }, { "epoch": 0.56, "learning_rate": 4.687029224673889e-06, "loss": 0.2572, "step": 4320 }, { "epoch": 0.56, "learning_rate": 4.687008847837574e-06, "loss": 0.2607, "step": 4330 }, { "epoch": 0.56, "learning_rate": 4.686988039349573e-06, "loss": 0.2443, "step": 4340 }, { "epoch": 0.56, "learning_rate": 4.6869667992137205e-06, "loss": 0.2719, "step": 4350 }, { "epoch": 0.57, "learning_rate": 4.686945127433929e-06, "loss": 0.2679, "step": 4360 }, { "epoch": 0.57, "learning_rate": 4.686923024014189e-06, "loss": 0.251, "step": 4370 }, { "epoch": 0.57, "learning_rate": 4.6869004889585745e-06, "loss": 0.2544, "step": 4380 }, { "epoch": 0.57, "learning_rate": 4.6868775222712356e-06, "loss": 0.2607, "step": 4390 }, { "epoch": 0.57, "learning_rate": 4.686854123956402e-06, "loss": 0.2682, "step": 4400 }, { "epoch": 0.57, "learning_rate": 4.686830294018386e-06, "loss": 0.2573, "step": 4410 }, { "epoch": 0.57, "learning_rate": 4.686806032461576e-06, "loss": 0.2615, "step": 4420 }, { "epoch": 0.57, "learning_rate": 4.686781339290441e-06, "loss": 0.2745, "step": 4430 }, { "epoch": 0.58, "learning_rate": 4.68675621450953e-06, "loss": 0.2646, "step": 4440 }, { "epoch": 0.58, "learning_rate": 4.6867306581234716e-06, "loss": 0.2539, "step": 4450 }, { "epoch": 0.58, "learning_rate": 4.6867046701369734e-06, "loss": 0.2716, "step": 4460 }, { "epoch": 0.58, "learning_rate": 4.686678250554823e-06, "loss": 0.2622, "step": 4470 }, { "epoch": 0.58, "learning_rate": 4.686651399381886e-06, "loss": 0.2572, "step": 4480 }, { "epoch": 0.58, "learning_rate": 4.68662411662311e-06, "loss": 0.2476, "step": 4490 }, { "epoch": 0.58, "learning_rate": 4.6865964022835185e-06, "loss": 0.2679, "step": 4500 }, { "epoch": 0.59, "learning_rate": 4.68656825636822e-06, "loss": 0.2544, "step": 4510 }, { "epoch": 0.59, "learning_rate": 4.686539678882397e-06, "loss": 0.2533, "step": 4520 }, { "epoch": 0.59, "learning_rate": 4.686510669831315e-06, "loss": 0.2638, "step": 4530 }, { "epoch": 0.59, "learning_rate": 4.686481229220316e-06, "loss": 0.2631, "step": 4540 }, { "epoch": 0.59, "learning_rate": 4.6864513570548255e-06, "loss": 0.2573, "step": 4550 }, { "epoch": 0.59, "learning_rate": 4.686421053340344e-06, "loss": 0.254, "step": 4560 }, { "epoch": 0.59, "learning_rate": 4.6863903180824565e-06, "loss": 0.2505, "step": 4570 }, { "epoch": 0.59, "learning_rate": 4.686359151286823e-06, "loss": 0.2384, "step": 4580 }, { "epoch": 0.6, "learning_rate": 4.686327552959184e-06, "loss": 0.2541, "step": 4590 }, { "epoch": 0.6, "learning_rate": 4.68629552310536e-06, "loss": 0.2471, "step": 4600 }, { "epoch": 0.6, "learning_rate": 4.686263061731254e-06, "loss": 0.2589, "step": 4610 }, { "epoch": 0.6, "learning_rate": 4.686230168842844e-06, "loss": 0.2411, "step": 4620 }, { "epoch": 0.6, "learning_rate": 4.686196844446188e-06, "loss": 0.2746, "step": 4630 }, { "epoch": 0.6, "learning_rate": 4.686163088547427e-06, "loss": 0.2676, "step": 4640 }, { "epoch": 0.6, "learning_rate": 4.686128901152778e-06, "loss": 0.2727, "step": 4650 }, { "epoch": 0.6, "learning_rate": 4.686094282268539e-06, "loss": 0.2555, "step": 4660 }, { "epoch": 0.61, "learning_rate": 4.686059231901086e-06, "loss": 0.2387, "step": 4670 }, { "epoch": 0.61, "learning_rate": 4.686023750056878e-06, "loss": 0.2384, "step": 4680 }, { "epoch": 0.61, "learning_rate": 4.685987836742448e-06, "loss": 0.2612, "step": 4690 }, { "epoch": 0.61, "learning_rate": 4.685951491964414e-06, "loss": 0.2403, "step": 4700 }, { "epoch": 0.61, "learning_rate": 4.68591471572947e-06, "loss": 0.2463, "step": 4710 }, { "epoch": 0.61, "learning_rate": 4.685877508044391e-06, "loss": 0.2612, "step": 4720 }, { "epoch": 0.61, "learning_rate": 4.685839868916032e-06, "loss": 0.2685, "step": 4730 }, { "epoch": 0.62, "learning_rate": 4.685801798351324e-06, "loss": 0.2481, "step": 4740 }, { "epoch": 0.62, "learning_rate": 4.685763296357282e-06, "loss": 0.2585, "step": 4750 }, { "epoch": 0.62, "learning_rate": 4.685724362940997e-06, "loss": 0.2584, "step": 4760 }, { "epoch": 0.62, "learning_rate": 4.685684998109642e-06, "loss": 0.2564, "step": 4770 }, { "epoch": 0.62, "learning_rate": 4.685645201870469e-06, "loss": 0.2545, "step": 4780 }, { "epoch": 0.62, "learning_rate": 4.685604974230807e-06, "loss": 0.2445, "step": 4790 }, { "epoch": 0.62, "learning_rate": 4.6855643151980676e-06, "loss": 0.2574, "step": 4800 }, { "epoch": 0.62, "learning_rate": 4.68552322477974e-06, "loss": 0.2459, "step": 4810 }, { "epoch": 0.63, "learning_rate": 4.685481702983395e-06, "loss": 0.2588, "step": 4820 }, { "epoch": 0.63, "learning_rate": 4.68543974981668e-06, "loss": 0.2659, "step": 4830 }, { "epoch": 0.63, "learning_rate": 4.685397365287321e-06, "loss": 0.257, "step": 4840 }, { "epoch": 0.63, "learning_rate": 4.68535454940313e-06, "loss": 0.2615, "step": 4850 }, { "epoch": 0.63, "learning_rate": 4.685311302171992e-06, "loss": 0.2544, "step": 4860 }, { "epoch": 0.63, "learning_rate": 4.6852676236018735e-06, "loss": 0.2534, "step": 4870 }, { "epoch": 0.63, "learning_rate": 4.68522351370082e-06, "loss": 0.2477, "step": 4880 }, { "epoch": 0.63, "learning_rate": 4.685178972476958e-06, "loss": 0.2519, "step": 4890 }, { "epoch": 0.64, "learning_rate": 4.6851339999384935e-06, "loss": 0.2628, "step": 4900 }, { "epoch": 0.64, "learning_rate": 4.685088596093707e-06, "loss": 0.2353, "step": 4910 }, { "epoch": 0.64, "learning_rate": 4.6850427609509666e-06, "loss": 0.2513, "step": 4920 }, { "epoch": 0.64, "learning_rate": 4.684996494518713e-06, "loss": 0.25, "step": 4930 }, { "epoch": 0.64, "learning_rate": 4.68494979680547e-06, "loss": 0.2465, "step": 4940 }, { "epoch": 0.64, "learning_rate": 4.68490266781984e-06, "loss": 0.2517, "step": 4950 }, { "epoch": 0.64, "learning_rate": 4.684855107570504e-06, "loss": 0.2527, "step": 4960 }, { "epoch": 0.65, "learning_rate": 4.684807116066223e-06, "loss": 0.2517, "step": 4970 }, { "epoch": 0.65, "learning_rate": 4.684758693315839e-06, "loss": 0.242, "step": 4980 }, { "epoch": 0.65, "learning_rate": 4.6847098393282685e-06, "loss": 0.2511, "step": 4990 }, { "epoch": 0.65, "learning_rate": 4.684660554112515e-06, "loss": 0.2695, "step": 5000 }, { "epoch": 0.65, "learning_rate": 4.6846108376776546e-06, "loss": 0.2741, "step": 5010 }, { "epoch": 0.65, "learning_rate": 4.684560690032846e-06, "loss": 0.2427, "step": 5020 }, { "epoch": 0.65, "learning_rate": 4.684510111187328e-06, "loss": 0.2369, "step": 5030 }, { "epoch": 0.65, "learning_rate": 4.6844591011504165e-06, "loss": 0.2529, "step": 5040 }, { "epoch": 0.66, "learning_rate": 4.68440765993151e-06, "loss": 0.2534, "step": 5050 }, { "epoch": 0.66, "learning_rate": 4.684355787540081e-06, "loss": 0.2457, "step": 5060 }, { "epoch": 0.66, "learning_rate": 4.684303483985688e-06, "loss": 0.2503, "step": 5070 }, { "epoch": 0.66, "learning_rate": 4.684250749277965e-06, "loss": 0.24, "step": 5080 }, { "epoch": 0.66, "learning_rate": 4.684197583426625e-06, "loss": 0.251, "step": 5090 }, { "epoch": 0.66, "learning_rate": 4.684143986441464e-06, "loss": 0.2551, "step": 5100 }, { "epoch": 0.66, "learning_rate": 4.684089958332354e-06, "loss": 0.2355, "step": 5110 }, { "epoch": 0.66, "learning_rate": 4.6840354991092474e-06, "loss": 0.2551, "step": 5120 }, { "epoch": 0.67, "learning_rate": 4.683980608782175e-06, "loss": 0.2478, "step": 5130 }, { "epoch": 0.67, "learning_rate": 4.6839252873612495e-06, "loss": 0.2537, "step": 5140 }, { "epoch": 0.67, "learning_rate": 4.683869534856663e-06, "loss": 0.2404, "step": 5150 }, { "epoch": 0.67, "learning_rate": 4.683813351278682e-06, "loss": 0.2558, "step": 5160 }, { "epoch": 0.67, "learning_rate": 4.683756736637659e-06, "loss": 0.2527, "step": 5170 }, { "epoch": 0.67, "learning_rate": 4.6836996909440225e-06, "loss": 0.2496, "step": 5180 }, { "epoch": 0.67, "learning_rate": 4.683642214208281e-06, "loss": 0.2312, "step": 5190 }, { "epoch": 0.67, "learning_rate": 4.68358430644102e-06, "loss": 0.2411, "step": 5200 }, { "epoch": 0.68, "learning_rate": 4.68352596765291e-06, "loss": 0.2558, "step": 5210 }, { "epoch": 0.68, "learning_rate": 4.6834671978546965e-06, "loss": 0.2381, "step": 5220 }, { "epoch": 0.68, "learning_rate": 4.6834079970572035e-06, "loss": 0.2468, "step": 5230 }, { "epoch": 0.68, "learning_rate": 4.683348365271339e-06, "loss": 0.2542, "step": 5240 }, { "epoch": 0.68, "learning_rate": 4.683288302508087e-06, "loss": 0.2492, "step": 5250 }, { "epoch": 0.68, "learning_rate": 4.68322780877851e-06, "loss": 0.2554, "step": 5260 }, { "epoch": 0.68, "learning_rate": 4.683166884093755e-06, "loss": 0.2511, "step": 5270 }, { "epoch": 0.69, "learning_rate": 4.683105528465041e-06, "loss": 0.2539, "step": 5280 }, { "epoch": 0.69, "learning_rate": 4.683043741903673e-06, "loss": 0.2481, "step": 5290 }, { "epoch": 0.69, "learning_rate": 4.682981524421032e-06, "loss": 0.258, "step": 5300 }, { "epoch": 0.69, "learning_rate": 4.682918876028579e-06, "loss": 0.2522, "step": 5310 }, { "epoch": 0.69, "learning_rate": 4.6828557967378545e-06, "loss": 0.2464, "step": 5320 }, { "epoch": 0.69, "learning_rate": 4.682792286560478e-06, "loss": 0.2535, "step": 5330 }, { "epoch": 0.69, "learning_rate": 4.682728345508149e-06, "loss": 0.2548, "step": 5340 }, { "epoch": 0.69, "learning_rate": 4.682663973592646e-06, "loss": 0.2593, "step": 5350 }, { "epoch": 0.7, "learning_rate": 4.682599170825827e-06, "loss": 0.2493, "step": 5360 }, { "epoch": 0.7, "learning_rate": 4.682533937219629e-06, "loss": 0.2502, "step": 5370 }, { "epoch": 0.7, "learning_rate": 4.68246827278607e-06, "loss": 0.239, "step": 5380 }, { "epoch": 0.7, "learning_rate": 4.682402177537245e-06, "loss": 0.2634, "step": 5390 }, { "epoch": 0.7, "learning_rate": 4.682335651485329e-06, "loss": 0.248, "step": 5400 }, { "epoch": 0.7, "learning_rate": 4.682268694642577e-06, "loss": 0.2346, "step": 5410 }, { "epoch": 0.7, "learning_rate": 4.682201307021324e-06, "loss": 0.2635, "step": 5420 }, { "epoch": 0.7, "learning_rate": 4.682133488633982e-06, "loss": 0.2479, "step": 5430 }, { "epoch": 0.71, "learning_rate": 4.682065239493045e-06, "loss": 0.2366, "step": 5440 }, { "epoch": 0.71, "learning_rate": 4.681996559611085e-06, "loss": 0.2713, "step": 5450 }, { "epoch": 0.71, "learning_rate": 4.681927449000753e-06, "loss": 0.2401, "step": 5460 }, { "epoch": 0.71, "learning_rate": 4.68185790767478e-06, "loss": 0.2448, "step": 5470 }, { "epoch": 0.71, "learning_rate": 4.681787935645978e-06, "loss": 0.2346, "step": 5480 }, { "epoch": 0.71, "learning_rate": 4.6817175329272335e-06, "loss": 0.2488, "step": 5490 }, { "epoch": 0.71, "learning_rate": 4.681646699531517e-06, "loss": 0.2509, "step": 5500 }, { "epoch": 0.72, "learning_rate": 4.681575435471877e-06, "loss": 0.2538, "step": 5510 }, { "epoch": 0.72, "learning_rate": 4.68150374076144e-06, "loss": 0.259, "step": 5520 }, { "epoch": 0.72, "learning_rate": 4.681431615413414e-06, "loss": 0.257, "step": 5530 }, { "epoch": 0.72, "learning_rate": 4.681359059441084e-06, "loss": 0.2541, "step": 5540 }, { "epoch": 0.72, "learning_rate": 4.681286072857818e-06, "loss": 0.2455, "step": 5550 }, { "epoch": 0.72, "learning_rate": 4.681212655677057e-06, "loss": 0.2526, "step": 5560 }, { "epoch": 0.72, "learning_rate": 4.681138807912329e-06, "loss": 0.2499, "step": 5570 }, { "epoch": 0.72, "learning_rate": 4.681064529577235e-06, "loss": 0.25, "step": 5580 }, { "epoch": 0.73, "learning_rate": 4.680989820685458e-06, "loss": 0.2558, "step": 5590 }, { "epoch": 0.73, "learning_rate": 4.680914681250761e-06, "loss": 0.2483, "step": 5600 }, { "epoch": 0.73, "learning_rate": 4.680839111286985e-06, "loss": 0.2511, "step": 5610 }, { "epoch": 0.73, "learning_rate": 4.680763110808052e-06, "loss": 0.2421, "step": 5620 }, { "epoch": 0.73, "learning_rate": 4.68068667982796e-06, "loss": 0.2548, "step": 5630 }, { "epoch": 0.73, "learning_rate": 4.68060981836079e-06, "loss": 0.2514, "step": 5640 }, { "epoch": 0.73, "learning_rate": 4.680532526420699e-06, "loss": 0.2494, "step": 5650 }, { "epoch": 0.73, "learning_rate": 4.680454804021926e-06, "loss": 0.2451, "step": 5660 }, { "epoch": 0.74, "learning_rate": 4.6803766511787886e-06, "loss": 0.2537, "step": 5670 }, { "epoch": 0.74, "learning_rate": 4.680298067905683e-06, "loss": 0.2448, "step": 5680 }, { "epoch": 0.74, "learning_rate": 4.680219054217084e-06, "loss": 0.2576, "step": 5690 }, { "epoch": 0.74, "learning_rate": 4.6801396101275485e-06, "loss": 0.2455, "step": 5700 }, { "epoch": 0.74, "learning_rate": 4.680059735651709e-06, "loss": 0.2602, "step": 5710 }, { "epoch": 0.74, "learning_rate": 4.679979430804281e-06, "loss": 0.2607, "step": 5720 }, { "epoch": 0.74, "learning_rate": 4.679898695600057e-06, "loss": 0.2522, "step": 5730 }, { "epoch": 0.74, "learning_rate": 4.679817530053908e-06, "loss": 0.2439, "step": 5740 }, { "epoch": 0.75, "learning_rate": 4.679735934180786e-06, "loss": 0.2583, "step": 5750 }, { "epoch": 0.75, "learning_rate": 4.679653907995724e-06, "loss": 0.2556, "step": 5760 }, { "epoch": 0.75, "learning_rate": 4.679571451513828e-06, "loss": 0.2693, "step": 5770 }, { "epoch": 0.75, "learning_rate": 4.67948856475029e-06, "loss": 0.2487, "step": 5780 }, { "epoch": 0.75, "learning_rate": 4.679405247720379e-06, "loss": 0.246, "step": 5790 }, { "epoch": 0.75, "learning_rate": 4.679321500439441e-06, "loss": 0.2643, "step": 5800 }, { "epoch": 0.75, "learning_rate": 4.679237322922904e-06, "loss": 0.2537, "step": 5810 }, { "epoch": 0.76, "learning_rate": 4.679152715186275e-06, "loss": 0.2345, "step": 5820 }, { "epoch": 0.76, "learning_rate": 4.679067677245138e-06, "loss": 0.2485, "step": 5830 }, { "epoch": 0.76, "learning_rate": 4.678982209115159e-06, "loss": 0.2633, "step": 5840 }, { "epoch": 0.76, "learning_rate": 4.678896310812082e-06, "loss": 0.2415, "step": 5850 }, { "epoch": 0.76, "learning_rate": 4.6788099823517304e-06, "loss": 0.2594, "step": 5860 }, { "epoch": 0.76, "learning_rate": 4.678723223750006e-06, "loss": 0.2482, "step": 5870 }, { "epoch": 0.76, "learning_rate": 4.678636035022892e-06, "loss": 0.2541, "step": 5880 }, { "epoch": 0.76, "learning_rate": 4.678548416186448e-06, "loss": 0.2595, "step": 5890 }, { "epoch": 0.77, "learning_rate": 4.678460367256814e-06, "loss": 0.2477, "step": 5900 }, { "epoch": 0.77, "learning_rate": 4.6783718882502115e-06, "loss": 0.2516, "step": 5910 }, { "epoch": 0.77, "learning_rate": 4.678282979182937e-06, "loss": 0.2438, "step": 5920 }, { "epoch": 0.77, "learning_rate": 4.67819364007137e-06, "loss": 0.2497, "step": 5930 }, { "epoch": 0.77, "learning_rate": 4.678103870931967e-06, "loss": 0.2481, "step": 5940 }, { "epoch": 0.77, "learning_rate": 4.678013671781264e-06, "loss": 0.2548, "step": 5950 }, { "epoch": 0.77, "learning_rate": 4.677923042635878e-06, "loss": 0.2403, "step": 5960 }, { "epoch": 0.77, "learning_rate": 4.6778319835125015e-06, "loss": 0.2313, "step": 5970 }, { "epoch": 0.78, "learning_rate": 4.6777404944279116e-06, "loss": 0.2399, "step": 5980 }, { "epoch": 0.78, "learning_rate": 4.677648575398959e-06, "loss": 0.2659, "step": 5990 }, { "epoch": 0.78, "learning_rate": 4.6775562264425766e-06, "loss": 0.2529, "step": 6000 }, { "epoch": 0.78, "learning_rate": 4.677463447575777e-06, "loss": 0.2629, "step": 6010 }, { "epoch": 0.78, "learning_rate": 4.677370238815649e-06, "loss": 0.242, "step": 6020 }, { "epoch": 0.78, "learning_rate": 4.677276600179364e-06, "loss": 0.2465, "step": 6030 }, { "epoch": 0.78, "learning_rate": 4.677182531684172e-06, "loss": 0.2455, "step": 6040 }, { "epoch": 0.79, "learning_rate": 4.6770880333474e-06, "loss": 0.2537, "step": 6050 }, { "epoch": 0.79, "learning_rate": 4.676993105186456e-06, "loss": 0.2501, "step": 6060 }, { "epoch": 0.79, "learning_rate": 4.676897747218826e-06, "loss": 0.2583, "step": 6070 }, { "epoch": 0.79, "learning_rate": 4.6768019594620775e-06, "loss": 0.2567, "step": 6080 }, { "epoch": 0.79, "learning_rate": 4.676705741933854e-06, "loss": 0.2474, "step": 6090 }, { "epoch": 0.79, "learning_rate": 4.67660909465188e-06, "loss": 0.2497, "step": 6100 }, { "epoch": 0.79, "learning_rate": 4.67651201763396e-06, "loss": 0.2644, "step": 6110 }, { "epoch": 0.79, "learning_rate": 4.6764145108979745e-06, "loss": 0.2477, "step": 6120 }, { "epoch": 0.8, "learning_rate": 4.676316574461888e-06, "loss": 0.2371, "step": 6130 }, { "epoch": 0.8, "learning_rate": 4.676218208343739e-06, "loss": 0.264, "step": 6140 }, { "epoch": 0.8, "learning_rate": 4.676119412561649e-06, "loss": 0.2434, "step": 6150 }, { "epoch": 0.8, "learning_rate": 4.676020187133816e-06, "loss": 0.2389, "step": 6160 }, { "epoch": 0.8, "learning_rate": 4.675920532078518e-06, "loss": 0.2491, "step": 6170 }, { "epoch": 0.8, "learning_rate": 4.675820447414115e-06, "loss": 0.2484, "step": 6180 }, { "epoch": 0.8, "learning_rate": 4.675719933159041e-06, "loss": 0.2527, "step": 6190 }, { "epoch": 0.8, "learning_rate": 4.675618989331812e-06, "loss": 0.2393, "step": 6200 }, { "epoch": 0.81, "learning_rate": 4.6755176159510245e-06, "loss": 0.2487, "step": 6210 }, { "epoch": 0.81, "learning_rate": 4.675415813035351e-06, "loss": 0.2401, "step": 6220 }, { "epoch": 0.81, "learning_rate": 4.675313580603545e-06, "loss": 0.2323, "step": 6230 }, { "epoch": 0.81, "learning_rate": 4.6752109186744396e-06, "loss": 0.23, "step": 6240 }, { "epoch": 0.81, "learning_rate": 4.675107827266944e-06, "loss": 0.2393, "step": 6250 }, { "epoch": 0.81, "learning_rate": 4.675004306400051e-06, "loss": 0.2539, "step": 6260 }, { "epoch": 0.81, "learning_rate": 4.67490035609283e-06, "loss": 0.2514, "step": 6270 }, { "epoch": 0.82, "learning_rate": 4.674795976364428e-06, "loss": 0.267, "step": 6280 }, { "epoch": 0.82, "learning_rate": 4.6746911672340735e-06, "loss": 0.2491, "step": 6290 }, { "epoch": 0.82, "learning_rate": 4.674585928721074e-06, "loss": 0.2345, "step": 6300 }, { "epoch": 0.82, "learning_rate": 4.674480260844814e-06, "loss": 0.2525, "step": 6310 }, { "epoch": 0.82, "learning_rate": 4.674374163624761e-06, "loss": 0.2453, "step": 6320 }, { "epoch": 0.82, "learning_rate": 4.674267637080458e-06, "loss": 0.2314, "step": 6330 }, { "epoch": 0.82, "learning_rate": 4.674160681231527e-06, "loss": 0.2421, "step": 6340 }, { "epoch": 0.82, "learning_rate": 4.674053296097672e-06, "loss": 0.2515, "step": 6350 }, { "epoch": 0.83, "learning_rate": 4.673945481698674e-06, "loss": 0.2492, "step": 6360 }, { "epoch": 0.83, "learning_rate": 4.673837238054392e-06, "loss": 0.2416, "step": 6370 }, { "epoch": 0.83, "learning_rate": 4.6737285651847675e-06, "loss": 0.2496, "step": 6380 }, { "epoch": 0.83, "learning_rate": 4.67361946310982e-06, "loss": 0.2346, "step": 6390 }, { "epoch": 0.83, "learning_rate": 4.673509931849644e-06, "loss": 0.2548, "step": 6400 }, { "epoch": 0.83, "learning_rate": 4.673399971424419e-06, "loss": 0.2306, "step": 6410 }, { "epoch": 0.83, "learning_rate": 4.673289581854399e-06, "loss": 0.2302, "step": 6420 }, { "epoch": 0.83, "learning_rate": 4.67317876315992e-06, "loss": 0.2324, "step": 6430 }, { "epoch": 0.84, "learning_rate": 4.6730675153613944e-06, "loss": 0.2478, "step": 6440 }, { "epoch": 0.84, "learning_rate": 4.672955838479318e-06, "loss": 0.2395, "step": 6450 }, { "epoch": 0.84, "learning_rate": 4.67284373253426e-06, "loss": 0.2543, "step": 6460 }, { "epoch": 0.84, "learning_rate": 4.672731197546873e-06, "loss": 0.2533, "step": 6470 }, { "epoch": 0.84, "learning_rate": 4.672618233537887e-06, "loss": 0.2676, "step": 6480 }, { "epoch": 0.84, "learning_rate": 4.67250484052811e-06, "loss": 0.247, "step": 6490 }, { "epoch": 0.84, "learning_rate": 4.672391018538431e-06, "loss": 0.2669, "step": 6500 }, { "epoch": 0.84, "learning_rate": 4.672276767589816e-06, "loss": 0.2468, "step": 6510 }, { "epoch": 0.85, "learning_rate": 4.672162087703314e-06, "loss": 0.2428, "step": 6520 }, { "epoch": 0.85, "learning_rate": 4.6720469789000465e-06, "loss": 0.2488, "step": 6530 }, { "epoch": 0.85, "learning_rate": 4.6719314412012204e-06, "loss": 0.2499, "step": 6540 }, { "epoch": 0.85, "learning_rate": 4.671815474628118e-06, "loss": 0.2597, "step": 6550 }, { "epoch": 0.85, "learning_rate": 4.671699079202101e-06, "loss": 0.2482, "step": 6560 }, { "epoch": 0.85, "learning_rate": 4.671582254944612e-06, "loss": 0.2402, "step": 6570 }, { "epoch": 0.85, "learning_rate": 4.671465001877169e-06, "loss": 0.2517, "step": 6580 }, { "epoch": 0.86, "learning_rate": 4.671347320021373e-06, "loss": 0.2487, "step": 6590 }, { "epoch": 0.86, "learning_rate": 4.671229209398902e-06, "loss": 0.2553, "step": 6600 }, { "epoch": 0.86, "learning_rate": 4.671110670031513e-06, "loss": 0.2509, "step": 6610 }, { "epoch": 0.86, "learning_rate": 4.670991701941041e-06, "loss": 0.2588, "step": 6620 }, { "epoch": 0.86, "learning_rate": 4.6708723051494025e-06, "loss": 0.2492, "step": 6630 }, { "epoch": 0.86, "learning_rate": 4.6707524796785915e-06, "loss": 0.2471, "step": 6640 }, { "epoch": 0.86, "learning_rate": 4.67063222555068e-06, "loss": 0.2562, "step": 6650 }, { "epoch": 0.86, "learning_rate": 4.670511542787821e-06, "loss": 0.2549, "step": 6660 }, { "epoch": 0.87, "learning_rate": 4.670390431412246e-06, "loss": 0.2534, "step": 6670 }, { "epoch": 0.87, "learning_rate": 4.670268891446263e-06, "loss": 0.25, "step": 6680 }, { "epoch": 0.87, "learning_rate": 4.670146922912262e-06, "loss": 0.2632, "step": 6690 }, { "epoch": 0.87, "learning_rate": 4.670024525832711e-06, "loss": 0.252, "step": 6700 }, { "epoch": 0.87, "learning_rate": 4.669901700230156e-06, "loss": 0.2282, "step": 6710 }, { "epoch": 0.87, "learning_rate": 4.669778446127224e-06, "loss": 0.2489, "step": 6720 }, { "epoch": 0.87, "learning_rate": 4.669654763546619e-06, "loss": 0.2374, "step": 6730 }, { "epoch": 0.87, "learning_rate": 4.669530652511125e-06, "loss": 0.2406, "step": 6740 }, { "epoch": 0.88, "learning_rate": 4.669406113043602e-06, "loss": 0.2541, "step": 6750 }, { "epoch": 0.88, "learning_rate": 4.669281145166995e-06, "loss": 0.2472, "step": 6760 }, { "epoch": 0.88, "learning_rate": 4.669155748904323e-06, "loss": 0.2535, "step": 6770 }, { "epoch": 0.88, "learning_rate": 4.669029924278683e-06, "loss": 0.2474, "step": 6780 }, { "epoch": 0.88, "learning_rate": 4.668903671313257e-06, "loss": 0.2583, "step": 6790 }, { "epoch": 0.88, "learning_rate": 4.668776990031299e-06, "loss": 0.2315, "step": 6800 }, { "epoch": 0.88, "learning_rate": 4.668649880456146e-06, "loss": 0.2629, "step": 6810 }, { "epoch": 0.89, "learning_rate": 4.668522342611214e-06, "loss": 0.2399, "step": 6820 }, { "epoch": 0.89, "learning_rate": 4.668394376519995e-06, "loss": 0.2691, "step": 6830 }, { "epoch": 0.89, "learning_rate": 4.668265982206062e-06, "loss": 0.2465, "step": 6840 }, { "epoch": 0.89, "learning_rate": 4.668137159693068e-06, "loss": 0.2424, "step": 6850 }, { "epoch": 0.89, "learning_rate": 4.6680079090047415e-06, "loss": 0.2207, "step": 6860 }, { "epoch": 0.89, "learning_rate": 4.667878230164892e-06, "loss": 0.2607, "step": 6870 }, { "epoch": 0.89, "learning_rate": 4.667748123197409e-06, "loss": 0.2421, "step": 6880 }, { "epoch": 0.89, "learning_rate": 4.667617588126259e-06, "loss": 0.2484, "step": 6890 }, { "epoch": 0.9, "learning_rate": 4.667486624975486e-06, "loss": 0.2409, "step": 6900 }, { "epoch": 0.9, "learning_rate": 4.667355233769217e-06, "loss": 0.251, "step": 6910 }, { "epoch": 0.9, "learning_rate": 4.667223414531655e-06, "loss": 0.257, "step": 6920 }, { "epoch": 0.9, "learning_rate": 4.667091167287082e-06, "loss": 0.2565, "step": 6930 }, { "epoch": 0.9, "learning_rate": 4.666958492059859e-06, "loss": 0.2354, "step": 6940 }, { "epoch": 0.9, "learning_rate": 4.666825388874428e-06, "loss": 0.2284, "step": 6950 }, { "epoch": 0.9, "learning_rate": 4.666691857755305e-06, "loss": 0.2474, "step": 6960 }, { "epoch": 0.9, "learning_rate": 4.66655789872709e-06, "loss": 0.2499, "step": 6970 }, { "epoch": 0.91, "learning_rate": 4.666423511814459e-06, "loss": 0.2421, "step": 6980 }, { "epoch": 0.91, "learning_rate": 4.6662886970421675e-06, "loss": 0.246, "step": 6990 }, { "epoch": 0.91, "learning_rate": 4.666153454435049e-06, "loss": 0.2502, "step": 7000 }, { "epoch": 0.91, "learning_rate": 4.666017784018017e-06, "loss": 0.2368, "step": 7010 }, { "epoch": 0.91, "learning_rate": 4.665881685816064e-06, "loss": 0.2429, "step": 7020 }, { "epoch": 0.91, "learning_rate": 4.665745159854259e-06, "loss": 0.2534, "step": 7030 }, { "epoch": 0.91, "learning_rate": 4.665608206157754e-06, "loss": 0.253, "step": 7040 }, { "epoch": 0.91, "learning_rate": 4.6654708247517744e-06, "loss": 0.2516, "step": 7050 }, { "epoch": 0.92, "learning_rate": 4.66533301566163e-06, "loss": 0.2325, "step": 7060 }, { "epoch": 0.92, "learning_rate": 4.665194778912704e-06, "loss": 0.236, "step": 7070 }, { "epoch": 0.92, "learning_rate": 4.665056114530462e-06, "loss": 0.2361, "step": 7080 }, { "epoch": 0.92, "learning_rate": 4.664917022540449e-06, "loss": 0.2527, "step": 7090 }, { "epoch": 0.92, "learning_rate": 4.664777502968284e-06, "loss": 0.2466, "step": 7100 }, { "epoch": 0.92, "learning_rate": 4.66463755583967e-06, "loss": 0.2408, "step": 7110 }, { "epoch": 0.92, "learning_rate": 4.6644971811803865e-06, "loss": 0.2524, "step": 7120 }, { "epoch": 0.93, "learning_rate": 4.664356379016292e-06, "loss": 0.2309, "step": 7130 }, { "epoch": 0.93, "learning_rate": 4.664215149373323e-06, "loss": 0.2457, "step": 7140 }, { "epoch": 0.93, "learning_rate": 4.6640734922774955e-06, "loss": 0.2441, "step": 7150 }, { "epoch": 0.93, "learning_rate": 4.663931407754904e-06, "loss": 0.2593, "step": 7160 }, { "epoch": 0.93, "learning_rate": 4.663788895831723e-06, "loss": 0.2375, "step": 7170 }, { "epoch": 0.93, "learning_rate": 4.663645956534203e-06, "loss": 0.2502, "step": 7180 }, { "epoch": 0.93, "learning_rate": 4.663502589888677e-06, "loss": 0.2584, "step": 7190 }, { "epoch": 0.93, "learning_rate": 4.663358795921553e-06, "loss": 0.239, "step": 7200 }, { "epoch": 0.94, "learning_rate": 4.663214574659319e-06, "loss": 0.2461, "step": 7210 }, { "epoch": 0.94, "learning_rate": 4.663069926128543e-06, "loss": 0.2384, "step": 7220 }, { "epoch": 0.94, "learning_rate": 4.66292485035587e-06, "loss": 0.2373, "step": 7230 }, { "epoch": 0.94, "learning_rate": 4.662779347368024e-06, "loss": 0.2448, "step": 7240 }, { "epoch": 0.94, "learning_rate": 4.66263341719181e-06, "loss": 0.2377, "step": 7250 }, { "epoch": 0.94, "learning_rate": 4.662487059854108e-06, "loss": 0.2511, "step": 7260 }, { "epoch": 0.94, "learning_rate": 4.662340275381879e-06, "loss": 0.2384, "step": 7270 }, { "epoch": 0.94, "learning_rate": 4.662193063802163e-06, "loss": 0.2476, "step": 7280 }, { "epoch": 0.95, "learning_rate": 4.662045425142076e-06, "loss": 0.2343, "step": 7290 }, { "epoch": 0.95, "learning_rate": 4.661897359428816e-06, "loss": 0.2329, "step": 7300 }, { "epoch": 0.95, "learning_rate": 4.661748866689658e-06, "loss": 0.2631, "step": 7310 }, { "epoch": 0.95, "learning_rate": 4.661599946951955e-06, "loss": 0.2558, "step": 7320 }, { "epoch": 0.95, "learning_rate": 4.6614506002431405e-06, "loss": 0.2576, "step": 7330 }, { "epoch": 0.95, "learning_rate": 4.661300826590726e-06, "loss": 0.2485, "step": 7340 }, { "epoch": 0.95, "learning_rate": 4.6611506260223005e-06, "loss": 0.2465, "step": 7350 }, { "epoch": 0.96, "learning_rate": 4.660999998565532e-06, "loss": 0.2479, "step": 7360 }, { "epoch": 0.96, "learning_rate": 4.660848944248168e-06, "loss": 0.2446, "step": 7370 }, { "epoch": 0.96, "learning_rate": 4.660697463098034e-06, "loss": 0.24, "step": 7380 }, { "epoch": 0.96, "learning_rate": 4.660545555143035e-06, "loss": 0.2383, "step": 7390 }, { "epoch": 0.96, "learning_rate": 4.660393220411154e-06, "loss": 0.2531, "step": 7400 }, { "epoch": 0.96, "learning_rate": 4.660240458930452e-06, "loss": 0.2493, "step": 7410 }, { "epoch": 0.96, "learning_rate": 4.660087270729069e-06, "loss": 0.2483, "step": 7420 }, { "epoch": 0.96, "learning_rate": 4.659933655835224e-06, "loss": 0.2343, "step": 7430 }, { "epoch": 0.97, "learning_rate": 4.659779614277216e-06, "loss": 0.2423, "step": 7440 }, { "epoch": 0.97, "learning_rate": 4.659625146083418e-06, "loss": 0.2399, "step": 7450 }, { "epoch": 0.97, "learning_rate": 4.659470251282287e-06, "loss": 0.2384, "step": 7460 }, { "epoch": 0.97, "learning_rate": 4.659314929902355e-06, "loss": 0.25, "step": 7470 }, { "epoch": 0.97, "learning_rate": 4.659159181972235e-06, "loss": 0.2398, "step": 7480 }, { "epoch": 0.97, "learning_rate": 4.6590030075206155e-06, "loss": 0.2331, "step": 7490 }, { "epoch": 0.97, "learning_rate": 4.658846406576266e-06, "loss": 0.2466, "step": 7500 }, { "epoch": 0.97, "learning_rate": 4.658689379168035e-06, "loss": 0.2571, "step": 7510 }, { "epoch": 0.98, "learning_rate": 4.658531925324847e-06, "loss": 0.2615, "step": 7520 }, { "epoch": 0.98, "learning_rate": 4.6583740450757086e-06, "loss": 0.2342, "step": 7530 }, { "epoch": 0.98, "learning_rate": 4.658215738449701e-06, "loss": 0.2431, "step": 7540 }, { "epoch": 0.98, "learning_rate": 4.658057005475986e-06, "loss": 0.2499, "step": 7550 }, { "epoch": 0.98, "learning_rate": 4.657897846183805e-06, "loss": 0.2475, "step": 7560 }, { "epoch": 0.98, "learning_rate": 4.657738260602476e-06, "loss": 0.2399, "step": 7570 }, { "epoch": 0.98, "learning_rate": 4.657578248761396e-06, "loss": 0.2419, "step": 7580 }, { "epoch": 0.99, "learning_rate": 4.657417810690041e-06, "loss": 0.2459, "step": 7590 }, { "epoch": 0.99, "learning_rate": 4.6572569464179665e-06, "loss": 0.241, "step": 7600 }, { "epoch": 0.99, "learning_rate": 4.657095655974803e-06, "loss": 0.2396, "step": 7610 }, { "epoch": 0.99, "learning_rate": 4.656933939390263e-06, "loss": 0.2506, "step": 7620 }, { "epoch": 0.99, "learning_rate": 4.656771796694137e-06, "loss": 0.2467, "step": 7630 }, { "epoch": 0.99, "learning_rate": 4.6566092279162924e-06, "loss": 0.2491, "step": 7640 }, { "epoch": 0.99, "learning_rate": 4.6564462330866755e-06, "loss": 0.2478, "step": 7650 }, { "epoch": 0.99, "learning_rate": 4.656282812235313e-06, "loss": 0.2321, "step": 7660 }, { "epoch": 1.0, "learning_rate": 4.656118965392309e-06, "loss": 0.2291, "step": 7670 }, { "epoch": 1.0, "learning_rate": 4.655954692587843e-06, "loss": 0.2396, "step": 7680 }, { "epoch": 1.0, "learning_rate": 4.655789993852178e-06, "loss": 0.2322, "step": 7690 }, { "epoch": 1.0, "learning_rate": 4.655624869215652e-06, "loss": 0.249, "step": 7700 }, { "epoch": 1.0, "eval_loss": 0.24451367557048798, "eval_runtime": 60.412, "eval_samples_per_second": 180.047, "eval_steps_per_second": 22.512, "step": 7705 }, { "epoch": 1.0, "learning_rate": 4.655459318708684e-06, "loss": 0.2264, "step": 7710 }, { "epoch": 1.0, "learning_rate": 4.655293342361769e-06, "loss": 0.254, "step": 7720 }, { "epoch": 1.0, "learning_rate": 4.655126940205481e-06, "loss": 0.2369, "step": 7730 }, { "epoch": 1.0, "learning_rate": 4.654960112270474e-06, "loss": 0.2452, "step": 7740 }, { "epoch": 1.01, "learning_rate": 4.654792858587479e-06, "loss": 0.2479, "step": 7750 }, { "epoch": 1.01, "learning_rate": 4.654625179187306e-06, "loss": 0.2627, "step": 7760 }, { "epoch": 1.01, "learning_rate": 4.654457074100843e-06, "loss": 0.2434, "step": 7770 }, { "epoch": 1.01, "learning_rate": 4.6542885433590565e-06, "loss": 0.2318, "step": 7780 }, { "epoch": 1.01, "learning_rate": 4.654119586992992e-06, "loss": 0.2374, "step": 7790 }, { "epoch": 1.01, "learning_rate": 4.653950205033772e-06, "loss": 0.2471, "step": 7800 }, { "epoch": 1.01, "learning_rate": 4.653780397512599e-06, "loss": 0.2477, "step": 7810 }, { "epoch": 1.01, "learning_rate": 4.653610164460754e-06, "loss": 0.2425, "step": 7820 }, { "epoch": 1.02, "learning_rate": 4.653439505909595e-06, "loss": 0.2442, "step": 7830 }, { "epoch": 1.02, "learning_rate": 4.653268421890557e-06, "loss": 0.243, "step": 7840 }, { "epoch": 1.02, "learning_rate": 4.65309691243516e-06, "loss": 0.2409, "step": 7850 }, { "epoch": 1.02, "learning_rate": 4.6529249775749925e-06, "loss": 0.22, "step": 7860 }, { "epoch": 1.02, "learning_rate": 4.6527526173417305e-06, "loss": 0.2217, "step": 7870 }, { "epoch": 1.02, "learning_rate": 4.652579831767123e-06, "loss": 0.2248, "step": 7880 }, { "epoch": 1.02, "learning_rate": 4.652406620882999e-06, "loss": 0.2418, "step": 7890 }, { "epoch": 1.03, "learning_rate": 4.652232984721265e-06, "loss": 0.2445, "step": 7900 }, { "epoch": 1.03, "learning_rate": 4.652058923313907e-06, "loss": 0.2716, "step": 7910 }, { "epoch": 1.03, "learning_rate": 4.65188443669299e-06, "loss": 0.244, "step": 7920 }, { "epoch": 1.03, "learning_rate": 4.651709524890654e-06, "loss": 0.2224, "step": 7930 }, { "epoch": 1.03, "learning_rate": 4.651534187939122e-06, "loss": 0.2558, "step": 7940 }, { "epoch": 1.03, "learning_rate": 4.651358425870691e-06, "loss": 0.2451, "step": 7950 }, { "epoch": 1.03, "learning_rate": 4.651182238717738e-06, "loss": 0.2558, "step": 7960 }, { "epoch": 1.03, "learning_rate": 4.65100562651272e-06, "loss": 0.2362, "step": 7970 }, { "epoch": 1.04, "learning_rate": 4.650828589288169e-06, "loss": 0.2326, "step": 7980 }, { "epoch": 1.04, "learning_rate": 4.650651127076698e-06, "loss": 0.2307, "step": 7990 }, { "epoch": 1.04, "learning_rate": 4.6504732399109964e-06, "loss": 0.2423, "step": 8000 }, { "epoch": 1.04, "learning_rate": 4.6502949278238345e-06, "loss": 0.2517, "step": 8010 }, { "epoch": 1.04, "learning_rate": 4.650116190848058e-06, "loss": 0.2523, "step": 8020 }, { "epoch": 1.04, "learning_rate": 4.649937029016592e-06, "loss": 0.2457, "step": 8030 }, { "epoch": 1.04, "learning_rate": 4.64975744236244e-06, "loss": 0.2551, "step": 8040 }, { "epoch": 1.04, "learning_rate": 4.649577430918684e-06, "loss": 0.2345, "step": 8050 }, { "epoch": 1.05, "learning_rate": 4.649396994718484e-06, "loss": 0.2363, "step": 8060 }, { "epoch": 1.05, "learning_rate": 4.649216133795077e-06, "loss": 0.2458, "step": 8070 }, { "epoch": 1.05, "learning_rate": 4.649034848181781e-06, "loss": 0.2348, "step": 8080 }, { "epoch": 1.05, "learning_rate": 4.648853137911989e-06, "loss": 0.2423, "step": 8090 }, { "epoch": 1.05, "learning_rate": 4.648671003019175e-06, "loss": 0.2254, "step": 8100 }, { "epoch": 1.05, "learning_rate": 4.6484884435368895e-06, "loss": 0.2346, "step": 8110 }, { "epoch": 1.05, "learning_rate": 4.648305459498763e-06, "loss": 0.2476, "step": 8120 }, { "epoch": 1.06, "learning_rate": 4.648122050938501e-06, "loss": 0.2296, "step": 8130 }, { "epoch": 1.06, "learning_rate": 4.647938217889891e-06, "loss": 0.2487, "step": 8140 }, { "epoch": 1.06, "learning_rate": 4.647753960386795e-06, "loss": 0.2367, "step": 8150 }, { "epoch": 1.06, "learning_rate": 4.647569278463157e-06, "loss": 0.2433, "step": 8160 }, { "epoch": 1.06, "learning_rate": 4.647384172152996e-06, "loss": 0.2414, "step": 8170 }, { "epoch": 1.06, "learning_rate": 4.6471986414904115e-06, "loss": 0.2382, "step": 8180 }, { "epoch": 1.06, "learning_rate": 4.647012686509579e-06, "loss": 0.2254, "step": 8190 }, { "epoch": 1.06, "learning_rate": 4.6468263072447535e-06, "loss": 0.2679, "step": 8200 }, { "epoch": 1.07, "learning_rate": 4.646639503730267e-06, "loss": 0.2291, "step": 8210 }, { "epoch": 1.07, "learning_rate": 4.646452276000534e-06, "loss": 0.2486, "step": 8220 }, { "epoch": 1.07, "learning_rate": 4.64626462409004e-06, "loss": 0.2392, "step": 8230 }, { "epoch": 1.07, "learning_rate": 4.646076548033354e-06, "loss": 0.2402, "step": 8240 }, { "epoch": 1.07, "learning_rate": 4.6458880478651216e-06, "loss": 0.2448, "step": 8250 }, { "epoch": 1.07, "learning_rate": 4.645699123620066e-06, "loss": 0.2497, "step": 8260 }, { "epoch": 1.07, "learning_rate": 4.645509775332988e-06, "loss": 0.2467, "step": 8270 }, { "epoch": 1.07, "learning_rate": 4.64532000303877e-06, "loss": 0.2441, "step": 8280 }, { "epoch": 1.08, "learning_rate": 4.6451298067723675e-06, "loss": 0.2253, "step": 8290 }, { "epoch": 1.08, "learning_rate": 4.644939186568818e-06, "loss": 0.2565, "step": 8300 }, { "epoch": 1.08, "learning_rate": 4.644748142463235e-06, "loss": 0.2443, "step": 8310 }, { "epoch": 1.08, "learning_rate": 4.6445566744908096e-06, "loss": 0.2462, "step": 8320 }, { "epoch": 1.08, "learning_rate": 4.644364782686815e-06, "loss": 0.247, "step": 8330 }, { "epoch": 1.08, "learning_rate": 4.644172467086597e-06, "loss": 0.2437, "step": 8340 }, { "epoch": 1.08, "learning_rate": 4.6439797277255834e-06, "loss": 0.2448, "step": 8350 }, { "epoch": 1.09, "learning_rate": 4.643786564639278e-06, "loss": 0.2472, "step": 8360 }, { "epoch": 1.09, "learning_rate": 4.643592977863264e-06, "loss": 0.2447, "step": 8370 }, { "epoch": 1.09, "learning_rate": 4.643398967433202e-06, "loss": 0.2461, "step": 8380 }, { "epoch": 1.09, "learning_rate": 4.643204533384829e-06, "loss": 0.2468, "step": 8390 }, { "epoch": 1.09, "learning_rate": 4.643009675753964e-06, "loss": 0.2494, "step": 8400 }, { "epoch": 1.09, "learning_rate": 4.642814394576501e-06, "loss": 0.2517, "step": 8410 }, { "epoch": 1.09, "learning_rate": 4.642618689888412e-06, "loss": 0.2473, "step": 8420 }, { "epoch": 1.09, "learning_rate": 4.642422561725749e-06, "loss": 0.246, "step": 8430 }, { "epoch": 1.1, "learning_rate": 4.642226010124639e-06, "loss": 0.2418, "step": 8440 }, { "epoch": 1.1, "learning_rate": 4.64202903512129e-06, "loss": 0.2549, "step": 8450 }, { "epoch": 1.1, "learning_rate": 4.6418316367519875e-06, "loss": 0.256, "step": 8460 }, { "epoch": 1.1, "learning_rate": 4.641633815053092e-06, "loss": 0.2398, "step": 8470 }, { "epoch": 1.1, "learning_rate": 4.641435570061046e-06, "loss": 0.2352, "step": 8480 }, { "epoch": 1.1, "learning_rate": 4.641236901812369e-06, "loss": 0.2411, "step": 8490 }, { "epoch": 1.1, "learning_rate": 4.641037810343655e-06, "loss": 0.2413, "step": 8500 }, { "epoch": 1.1, "learning_rate": 4.640838295691581e-06, "loss": 0.2393, "step": 8510 }, { "epoch": 1.11, "learning_rate": 4.6406383578928974e-06, "loss": 0.2325, "step": 8520 }, { "epoch": 1.11, "learning_rate": 4.640437996984437e-06, "loss": 0.2398, "step": 8530 }, { "epoch": 1.11, "learning_rate": 4.640237213003106e-06, "loss": 0.236, "step": 8540 }, { "epoch": 1.11, "learning_rate": 4.640036005985894e-06, "loss": 0.2547, "step": 8550 }, { "epoch": 1.11, "learning_rate": 4.639834375969863e-06, "loss": 0.2326, "step": 8560 }, { "epoch": 1.11, "learning_rate": 4.639632322992155e-06, "loss": 0.2439, "step": 8570 }, { "epoch": 1.11, "learning_rate": 4.639429847089991e-06, "loss": 0.2363, "step": 8580 }, { "epoch": 1.11, "learning_rate": 4.639226948300669e-06, "loss": 0.2389, "step": 8590 }, { "epoch": 1.12, "learning_rate": 4.639023626661565e-06, "loss": 0.2393, "step": 8600 }, { "epoch": 1.12, "learning_rate": 4.638819882210133e-06, "loss": 0.2487, "step": 8610 }, { "epoch": 1.12, "learning_rate": 4.638615714983904e-06, "loss": 0.2453, "step": 8620 }, { "epoch": 1.12, "learning_rate": 4.638411125020489e-06, "loss": 0.2497, "step": 8630 }, { "epoch": 1.12, "learning_rate": 4.638206112357575e-06, "loss": 0.2518, "step": 8640 }, { "epoch": 1.12, "learning_rate": 4.638000677032926e-06, "loss": 0.2421, "step": 8650 }, { "epoch": 1.12, "learning_rate": 4.637794819084387e-06, "loss": 0.2373, "step": 8660 }, { "epoch": 1.13, "learning_rate": 4.637588538549878e-06, "loss": 0.2473, "step": 8670 }, { "epoch": 1.13, "learning_rate": 4.637381835467399e-06, "loss": 0.2415, "step": 8680 }, { "epoch": 1.13, "learning_rate": 4.6371747098750264e-06, "loss": 0.2501, "step": 8690 }, { "epoch": 1.13, "learning_rate": 4.636967161810913e-06, "loss": 0.2516, "step": 8700 }, { "epoch": 1.13, "learning_rate": 4.636759191313295e-06, "loss": 0.2346, "step": 8710 }, { "epoch": 1.13, "learning_rate": 4.636550798420479e-06, "loss": 0.2726, "step": 8720 }, { "epoch": 1.13, "learning_rate": 4.6363419831708545e-06, "loss": 0.2516, "step": 8730 }, { "epoch": 1.13, "learning_rate": 4.636132745602888e-06, "loss": 0.2542, "step": 8740 }, { "epoch": 1.14, "learning_rate": 4.635923085755122e-06, "loss": 0.2442, "step": 8750 }, { "epoch": 1.14, "learning_rate": 4.635713003666178e-06, "loss": 0.2346, "step": 8760 }, { "epoch": 1.14, "learning_rate": 4.635502499374757e-06, "loss": 0.2414, "step": 8770 }, { "epoch": 1.14, "learning_rate": 4.635291572919634e-06, "loss": 0.2426, "step": 8780 }, { "epoch": 1.14, "learning_rate": 4.635080224339663e-06, "loss": 0.2484, "step": 8790 }, { "epoch": 1.14, "learning_rate": 4.63486845367378e-06, "loss": 0.2424, "step": 8800 }, { "epoch": 1.14, "learning_rate": 4.634656260960992e-06, "loss": 0.2398, "step": 8810 }, { "epoch": 1.14, "learning_rate": 4.634443646240388e-06, "loss": 0.2393, "step": 8820 }, { "epoch": 1.15, "learning_rate": 4.6342306095511355e-06, "loss": 0.2332, "step": 8830 }, { "epoch": 1.15, "learning_rate": 4.6340171509324755e-06, "loss": 0.22, "step": 8840 }, { "epoch": 1.15, "learning_rate": 4.63380327042373e-06, "loss": 0.2403, "step": 8850 }, { "epoch": 1.15, "learning_rate": 4.633588968064298e-06, "loss": 0.2439, "step": 8860 }, { "epoch": 1.15, "learning_rate": 4.6333742438936556e-06, "loss": 0.2479, "step": 8870 }, { "epoch": 1.15, "learning_rate": 4.633159097951359e-06, "loss": 0.2459, "step": 8880 }, { "epoch": 1.15, "learning_rate": 4.6329435302770385e-06, "loss": 0.2418, "step": 8890 }, { "epoch": 1.16, "learning_rate": 4.632727540910405e-06, "loss": 0.2338, "step": 8900 }, { "epoch": 1.16, "learning_rate": 4.632511129891244e-06, "loss": 0.2328, "step": 8910 }, { "epoch": 1.16, "learning_rate": 4.632294297259422e-06, "loss": 0.2465, "step": 8920 }, { "epoch": 1.16, "learning_rate": 4.632077043054882e-06, "loss": 0.2342, "step": 8930 }, { "epoch": 1.16, "learning_rate": 4.6318593673176434e-06, "loss": 0.2358, "step": 8940 }, { "epoch": 1.16, "learning_rate": 4.631641270087806e-06, "loss": 0.2432, "step": 8950 }, { "epoch": 1.16, "learning_rate": 4.631422751405543e-06, "loss": 0.2391, "step": 8960 }, { "epoch": 1.16, "learning_rate": 4.6312038113111095e-06, "loss": 0.2336, "step": 8970 }, { "epoch": 1.17, "learning_rate": 4.630984449844836e-06, "loss": 0.251, "step": 8980 }, { "epoch": 1.17, "learning_rate": 4.63076466704713e-06, "loss": 0.2328, "step": 8990 }, { "epoch": 1.17, "learning_rate": 4.630544462958479e-06, "loss": 0.2387, "step": 9000 }, { "epoch": 1.17, "learning_rate": 4.630323837619447e-06, "loss": 0.2551, "step": 9010 }, { "epoch": 1.17, "learning_rate": 4.630102791070675e-06, "loss": 0.2341, "step": 9020 }, { "epoch": 1.17, "learning_rate": 4.629881323352882e-06, "loss": 0.2519, "step": 9030 }, { "epoch": 1.17, "learning_rate": 4.629659434506864e-06, "loss": 0.2329, "step": 9040 }, { "epoch": 1.17, "learning_rate": 4.629437124573496e-06, "loss": 0.2514, "step": 9050 }, { "epoch": 1.18, "learning_rate": 4.629214393593728e-06, "loss": 0.242, "step": 9060 }, { "epoch": 1.18, "learning_rate": 4.6289912416085914e-06, "loss": 0.2365, "step": 9070 }, { "epoch": 1.18, "learning_rate": 4.628767668659192e-06, "loss": 0.2434, "step": 9080 }, { "epoch": 1.18, "learning_rate": 4.6285436747867145e-06, "loss": 0.2366, "step": 9090 }, { "epoch": 1.18, "learning_rate": 4.62831926003242e-06, "loss": 0.2364, "step": 9100 }, { "epoch": 1.18, "learning_rate": 4.62809442443765e-06, "loss": 0.2382, "step": 9110 }, { "epoch": 1.18, "learning_rate": 4.627869168043819e-06, "loss": 0.2345, "step": 9120 }, { "epoch": 1.18, "learning_rate": 4.6276434908924224e-06, "loss": 0.2384, "step": 9130 }, { "epoch": 1.19, "learning_rate": 4.627417393025033e-06, "loss": 0.2353, "step": 9140 }, { "epoch": 1.19, "learning_rate": 4.627190874483299e-06, "loss": 0.2451, "step": 9150 }, { "epoch": 1.19, "learning_rate": 4.626963935308948e-06, "loss": 0.2289, "step": 9160 }, { "epoch": 1.19, "learning_rate": 4.6267365755437845e-06, "loss": 0.2361, "step": 9170 }, { "epoch": 1.19, "learning_rate": 4.62650879522969e-06, "loss": 0.2519, "step": 9180 }, { "epoch": 1.19, "learning_rate": 4.626280594408625e-06, "loss": 0.2502, "step": 9190 }, { "epoch": 1.19, "learning_rate": 4.626051973122626e-06, "loss": 0.2402, "step": 9200 }, { "epoch": 1.2, "learning_rate": 4.625822931413806e-06, "loss": 0.2332, "step": 9210 }, { "epoch": 1.2, "learning_rate": 4.625593469324358e-06, "loss": 0.2535, "step": 9220 }, { "epoch": 1.2, "learning_rate": 4.625363586896551e-06, "loss": 0.2385, "step": 9230 }, { "epoch": 1.2, "learning_rate": 4.625133284172731e-06, "loss": 0.2432, "step": 9240 }, { "epoch": 1.2, "learning_rate": 4.624902561195324e-06, "loss": 0.2478, "step": 9250 }, { "epoch": 1.2, "learning_rate": 4.624671418006829e-06, "loss": 0.2486, "step": 9260 }, { "epoch": 1.2, "learning_rate": 4.624439854649826e-06, "loss": 0.2349, "step": 9270 }, { "epoch": 1.2, "learning_rate": 4.624207871166972e-06, "loss": 0.2525, "step": 9280 }, { "epoch": 1.21, "learning_rate": 4.623975467601e-06, "loss": 0.2592, "step": 9290 }, { "epoch": 1.21, "learning_rate": 4.623742643994721e-06, "loss": 0.2446, "step": 9300 }, { "epoch": 1.21, "learning_rate": 4.623509400391023e-06, "loss": 0.2455, "step": 9310 }, { "epoch": 1.21, "learning_rate": 4.623275736832873e-06, "loss": 0.2469, "step": 9320 }, { "epoch": 1.21, "learning_rate": 4.623041653363312e-06, "loss": 0.2338, "step": 9330 }, { "epoch": 1.21, "learning_rate": 4.6228071500254626e-06, "loss": 0.2533, "step": 9340 }, { "epoch": 1.21, "learning_rate": 4.622572226862522e-06, "loss": 0.2445, "step": 9350 }, { "epoch": 1.21, "learning_rate": 4.622336883917765e-06, "loss": 0.2236, "step": 9360 }, { "epoch": 1.22, "learning_rate": 4.6221011212345454e-06, "loss": 0.2521, "step": 9370 }, { "epoch": 1.22, "learning_rate": 4.621864938856292e-06, "loss": 0.2626, "step": 9380 }, { "epoch": 1.22, "learning_rate": 4.621628336826512e-06, "loss": 0.2386, "step": 9390 }, { "epoch": 1.22, "learning_rate": 4.62139131518879e-06, "loss": 0.2388, "step": 9400 }, { "epoch": 1.22, "learning_rate": 4.621153873986788e-06, "loss": 0.2414, "step": 9410 }, { "epoch": 1.22, "learning_rate": 4.620916013264244e-06, "loss": 0.2441, "step": 9420 }, { "epoch": 1.22, "learning_rate": 4.620677733064975e-06, "loss": 0.2304, "step": 9430 }, { "epoch": 1.23, "learning_rate": 4.620439033432876e-06, "loss": 0.2314, "step": 9440 }, { "epoch": 1.23, "learning_rate": 4.620199914411915e-06, "loss": 0.2407, "step": 9450 }, { "epoch": 1.23, "learning_rate": 4.619960376046142e-06, "loss": 0.2604, "step": 9460 }, { "epoch": 1.23, "learning_rate": 4.619720418379683e-06, "loss": 0.2458, "step": 9470 }, { "epoch": 1.23, "learning_rate": 4.619480041456739e-06, "loss": 0.2589, "step": 9480 }, { "epoch": 1.23, "learning_rate": 4.619239245321589e-06, "loss": 0.23, "step": 9490 }, { "epoch": 1.23, "learning_rate": 4.618998030018593e-06, "loss": 0.2321, "step": 9500 }, { "epoch": 1.23, "learning_rate": 4.618756395592184e-06, "loss": 0.2373, "step": 9510 }, { "epoch": 1.24, "learning_rate": 4.618514342086873e-06, "loss": 0.237, "step": 9520 }, { "epoch": 1.24, "learning_rate": 4.618271869547248e-06, "loss": 0.243, "step": 9530 }, { "epoch": 1.24, "learning_rate": 4.618028978017977e-06, "loss": 0.2475, "step": 9540 }, { "epoch": 1.24, "learning_rate": 4.617785667543801e-06, "loss": 0.2288, "step": 9550 }, { "epoch": 1.24, "learning_rate": 4.617541938169541e-06, "loss": 0.2396, "step": 9560 }, { "epoch": 1.24, "learning_rate": 4.617297789940095e-06, "loss": 0.25, "step": 9570 }, { "epoch": 1.24, "learning_rate": 4.617053222900436e-06, "loss": 0.2535, "step": 9580 }, { "epoch": 1.24, "learning_rate": 4.616808237095618e-06, "loss": 0.2524, "step": 9590 }, { "epoch": 1.25, "learning_rate": 4.616562832570767e-06, "loss": 0.23, "step": 9600 }, { "epoch": 1.25, "learning_rate": 4.616317009371091e-06, "loss": 0.2387, "step": 9610 }, { "epoch": 1.25, "learning_rate": 4.616070767541873e-06, "loss": 0.2412, "step": 9620 }, { "epoch": 1.25, "learning_rate": 4.615824107128472e-06, "loss": 0.2552, "step": 9630 }, { "epoch": 1.25, "learning_rate": 4.615577028176327e-06, "loss": 0.2391, "step": 9640 }, { "epoch": 1.25, "learning_rate": 4.615329530730951e-06, "loss": 0.2451, "step": 9650 }, { "epoch": 1.25, "learning_rate": 4.6150816148379355e-06, "loss": 0.2464, "step": 9660 }, { "epoch": 1.26, "learning_rate": 4.61483328054295e-06, "loss": 0.2481, "step": 9670 }, { "epoch": 1.26, "learning_rate": 4.6145845278917395e-06, "loss": 0.23, "step": 9680 }, { "epoch": 1.26, "learning_rate": 4.614335356930127e-06, "loss": 0.2392, "step": 9690 }, { "epoch": 1.26, "learning_rate": 4.6140857677040124e-06, "loss": 0.2345, "step": 9700 }, { "epoch": 1.26, "learning_rate": 4.613835760259372e-06, "loss": 0.2336, "step": 9710 }, { "epoch": 1.26, "learning_rate": 4.613585334642261e-06, "loss": 0.2381, "step": 9720 }, { "epoch": 1.26, "learning_rate": 4.613334490898808e-06, "loss": 0.2563, "step": 9730 }, { "epoch": 1.26, "learning_rate": 4.6130832290752224e-06, "loss": 0.2431, "step": 9740 }, { "epoch": 1.27, "learning_rate": 4.612831549217789e-06, "loss": 0.247, "step": 9750 }, { "epoch": 1.27, "learning_rate": 4.6125794513728705e-06, "loss": 0.2436, "step": 9760 }, { "epoch": 1.27, "learning_rate": 4.612326935586905e-06, "loss": 0.2488, "step": 9770 }, { "epoch": 1.27, "learning_rate": 4.612074001906408e-06, "loss": 0.2284, "step": 9780 }, { "epoch": 1.27, "learning_rate": 4.611820650377972e-06, "loss": 0.2601, "step": 9790 }, { "epoch": 1.27, "learning_rate": 4.611566881048269e-06, "loss": 0.2461, "step": 9800 }, { "epoch": 1.27, "learning_rate": 4.611312693964044e-06, "loss": 0.2299, "step": 9810 }, { "epoch": 1.27, "learning_rate": 4.611058089172122e-06, "loss": 0.2273, "step": 9820 }, { "epoch": 1.28, "learning_rate": 4.610803066719403e-06, "loss": 0.2451, "step": 9830 }, { "epoch": 1.28, "learning_rate": 4.610547626652864e-06, "loss": 0.2498, "step": 9840 }, { "epoch": 1.28, "learning_rate": 4.610291769019561e-06, "loss": 0.2258, "step": 9850 }, { "epoch": 1.28, "learning_rate": 4.6100354938666255e-06, "loss": 0.2487, "step": 9860 }, { "epoch": 1.28, "learning_rate": 4.6097788012412654e-06, "loss": 0.2584, "step": 9870 }, { "epoch": 1.28, "learning_rate": 4.609521691190765e-06, "loss": 0.2379, "step": 9880 }, { "epoch": 1.28, "learning_rate": 4.609264163762489e-06, "loss": 0.2402, "step": 9890 }, { "epoch": 1.28, "learning_rate": 4.6090062190038735e-06, "loss": 0.254, "step": 9900 }, { "epoch": 1.29, "learning_rate": 4.608747856962438e-06, "loss": 0.2589, "step": 9910 }, { "epoch": 1.29, "learning_rate": 4.608489077685772e-06, "loss": 0.2481, "step": 9920 }, { "epoch": 1.29, "learning_rate": 4.608229881221548e-06, "loss": 0.2513, "step": 9930 }, { "epoch": 1.29, "learning_rate": 4.60797026761751e-06, "loss": 0.2539, "step": 9940 }, { "epoch": 1.29, "learning_rate": 4.607710236921484e-06, "loss": 0.2466, "step": 9950 }, { "epoch": 1.29, "learning_rate": 4.607449789181368e-06, "loss": 0.2362, "step": 9960 }, { "epoch": 1.29, "learning_rate": 4.607188924445141e-06, "loss": 0.2321, "step": 9970 }, { "epoch": 1.3, "learning_rate": 4.606927642760855e-06, "loss": 0.2403, "step": 9980 }, { "epoch": 1.3, "learning_rate": 4.6066659441766415e-06, "loss": 0.2235, "step": 9990 }, { "epoch": 1.3, "learning_rate": 4.6064038287407095e-06, "loss": 0.2398, "step": 10000 }, { "epoch": 1.3, "learning_rate": 4.606141296501341e-06, "loss": 0.2348, "step": 10010 }, { "epoch": 1.3, "learning_rate": 4.605878347506897e-06, "loss": 0.2253, "step": 10020 }, { "epoch": 1.3, "learning_rate": 4.6056149818058185e-06, "loss": 0.2242, "step": 10030 }, { "epoch": 1.3, "learning_rate": 4.6053511994466164e-06, "loss": 0.2411, "step": 10040 }, { "epoch": 1.3, "learning_rate": 4.605087000477884e-06, "loss": 0.2373, "step": 10050 }, { "epoch": 1.31, "learning_rate": 4.604822384948289e-06, "loss": 0.2395, "step": 10060 }, { "epoch": 1.31, "learning_rate": 4.604557352906576e-06, "loss": 0.2373, "step": 10070 }, { "epoch": 1.31, "learning_rate": 4.604291904401567e-06, "loss": 0.2462, "step": 10080 }, { "epoch": 1.31, "learning_rate": 4.604026039482159e-06, "loss": 0.2408, "step": 10090 }, { "epoch": 1.31, "learning_rate": 4.603759758197329e-06, "loss": 0.2524, "step": 10100 }, { "epoch": 1.31, "learning_rate": 4.603493060596127e-06, "loss": 0.2397, "step": 10110 }, { "epoch": 1.31, "learning_rate": 4.603225946727683e-06, "loss": 0.2287, "step": 10120 }, { "epoch": 1.31, "learning_rate": 4.6029584166411994e-06, "loss": 0.2299, "step": 10130 }, { "epoch": 1.32, "learning_rate": 4.602690470385961e-06, "loss": 0.2397, "step": 10140 }, { "epoch": 1.32, "learning_rate": 4.602422108011324e-06, "loss": 0.2374, "step": 10150 }, { "epoch": 1.32, "learning_rate": 4.6021533295667245e-06, "loss": 0.2365, "step": 10160 }, { "epoch": 1.32, "learning_rate": 4.601884135101674e-06, "loss": 0.2359, "step": 10170 }, { "epoch": 1.32, "learning_rate": 4.60161452466576e-06, "loss": 0.2461, "step": 10180 }, { "epoch": 1.32, "learning_rate": 4.601344498308648e-06, "loss": 0.2354, "step": 10190 }, { "epoch": 1.32, "learning_rate": 4.60107405608008e-06, "loss": 0.2247, "step": 10200 }, { "epoch": 1.33, "learning_rate": 4.600803198029874e-06, "loss": 0.2329, "step": 10210 }, { "epoch": 1.33, "learning_rate": 4.600531924207924e-06, "loss": 0.2327, "step": 10220 }, { "epoch": 1.33, "learning_rate": 4.600260234664203e-06, "loss": 0.2363, "step": 10230 }, { "epoch": 1.33, "learning_rate": 4.599988129448756e-06, "loss": 0.2405, "step": 10240 }, { "epoch": 1.33, "learning_rate": 4.599715608611711e-06, "loss": 0.2456, "step": 10250 }, { "epoch": 1.33, "learning_rate": 4.599442672203266e-06, "loss": 0.2578, "step": 10260 }, { "epoch": 1.33, "learning_rate": 4.5991693202737005e-06, "loss": 0.2507, "step": 10270 }, { "epoch": 1.33, "learning_rate": 4.598895552873369e-06, "loss": 0.2414, "step": 10280 }, { "epoch": 1.34, "learning_rate": 4.5986213700527e-06, "loss": 0.2409, "step": 10290 }, { "epoch": 1.34, "learning_rate": 4.598346771862203e-06, "loss": 0.236, "step": 10300 }, { "epoch": 1.34, "learning_rate": 4.59807175835246e-06, "loss": 0.2587, "step": 10310 }, { "epoch": 1.34, "learning_rate": 4.597796329574133e-06, "loss": 0.2355, "step": 10320 }, { "epoch": 1.34, "learning_rate": 4.597520485577957e-06, "loss": 0.2368, "step": 10330 }, { "epoch": 1.34, "learning_rate": 4.597244226414746e-06, "loss": 0.2446, "step": 10340 }, { "epoch": 1.34, "learning_rate": 4.59696755213539e-06, "loss": 0.232, "step": 10350 }, { "epoch": 1.34, "learning_rate": 4.596690462790854e-06, "loss": 0.2474, "step": 10360 }, { "epoch": 1.35, "learning_rate": 4.596412958432183e-06, "loss": 0.2411, "step": 10370 }, { "epoch": 1.35, "learning_rate": 4.596135039110493e-06, "loss": 0.2452, "step": 10380 }, { "epoch": 1.35, "learning_rate": 4.595856704876982e-06, "loss": 0.2516, "step": 10390 }, { "epoch": 1.35, "learning_rate": 4.5955779557829205e-06, "loss": 0.2265, "step": 10400 }, { "epoch": 1.35, "learning_rate": 4.5952987918796575e-06, "loss": 0.2268, "step": 10410 }, { "epoch": 1.35, "learning_rate": 4.595019213218618e-06, "loss": 0.2531, "step": 10420 }, { "epoch": 1.35, "learning_rate": 4.594739219851302e-06, "loss": 0.2522, "step": 10430 }, { "epoch": 1.35, "learning_rate": 4.594458811829289e-06, "loss": 0.2336, "step": 10440 }, { "epoch": 1.36, "learning_rate": 4.59417798920423e-06, "loss": 0.2399, "step": 10450 }, { "epoch": 1.36, "learning_rate": 4.593896752027859e-06, "loss": 0.256, "step": 10460 }, { "epoch": 1.36, "learning_rate": 4.59361510035198e-06, "loss": 0.2482, "step": 10470 }, { "epoch": 1.36, "learning_rate": 4.593333034228477e-06, "loss": 0.2395, "step": 10480 }, { "epoch": 1.36, "learning_rate": 4.593050553709309e-06, "loss": 0.2441, "step": 10490 }, { "epoch": 1.36, "learning_rate": 4.592767658846513e-06, "loss": 0.2383, "step": 10500 }, { "epoch": 1.36, "learning_rate": 4.5924843496922e-06, "loss": 0.2442, "step": 10510 }, { "epoch": 1.37, "learning_rate": 4.592200626298558e-06, "loss": 0.2298, "step": 10520 }, { "epoch": 1.37, "learning_rate": 4.5919164887178525e-06, "loss": 0.2457, "step": 10530 }, { "epoch": 1.37, "learning_rate": 4.591631937002424e-06, "loss": 0.2366, "step": 10540 }, { "epoch": 1.37, "learning_rate": 4.5913469712046895e-06, "loss": 0.2484, "step": 10550 }, { "epoch": 1.37, "learning_rate": 4.591061591377143e-06, "loss": 0.2414, "step": 10560 }, { "epoch": 1.37, "learning_rate": 4.590775797572355e-06, "loss": 0.2483, "step": 10570 }, { "epoch": 1.37, "learning_rate": 4.590489589842971e-06, "loss": 0.2498, "step": 10580 }, { "epoch": 1.37, "learning_rate": 4.590202968241713e-06, "loss": 0.2592, "step": 10590 }, { "epoch": 1.38, "learning_rate": 4.5899159328213785e-06, "loss": 0.2359, "step": 10600 }, { "epoch": 1.38, "learning_rate": 4.589628483634845e-06, "loss": 0.2422, "step": 10610 }, { "epoch": 1.38, "learning_rate": 4.589340620735061e-06, "loss": 0.2459, "step": 10620 }, { "epoch": 1.38, "learning_rate": 4.589052344175055e-06, "loss": 0.2429, "step": 10630 }, { "epoch": 1.38, "learning_rate": 4.588763654007931e-06, "loss": 0.2286, "step": 10640 }, { "epoch": 1.38, "learning_rate": 4.588474550286868e-06, "loss": 0.2465, "step": 10650 }, { "epoch": 1.38, "learning_rate": 4.5881850330651205e-06, "loss": 0.2228, "step": 10660 }, { "epoch": 1.38, "learning_rate": 4.587895102396021e-06, "loss": 0.2352, "step": 10670 }, { "epoch": 1.39, "learning_rate": 4.58760475833298e-06, "loss": 0.2526, "step": 10680 }, { "epoch": 1.39, "learning_rate": 4.587314000929479e-06, "loss": 0.2402, "step": 10690 }, { "epoch": 1.39, "learning_rate": 4.58702283023908e-06, "loss": 0.2382, "step": 10700 }, { "epoch": 1.39, "learning_rate": 4.586731246315418e-06, "loss": 0.2437, "step": 10710 }, { "epoch": 1.39, "learning_rate": 4.5864392492122065e-06, "loss": 0.2525, "step": 10720 }, { "epoch": 1.39, "learning_rate": 4.586146838983235e-06, "loss": 0.2445, "step": 10730 }, { "epoch": 1.39, "learning_rate": 4.585854015682367e-06, "loss": 0.2409, "step": 10740 }, { "epoch": 1.4, "learning_rate": 4.585560779363546e-06, "loss": 0.2396, "step": 10750 }, { "epoch": 1.4, "learning_rate": 4.585267130080786e-06, "loss": 0.235, "step": 10760 }, { "epoch": 1.4, "learning_rate": 4.584973067888182e-06, "loss": 0.246, "step": 10770 }, { "epoch": 1.4, "learning_rate": 4.5846785928399015e-06, "loss": 0.253, "step": 10780 }, { "epoch": 1.4, "learning_rate": 4.5843837049901915e-06, "loss": 0.2449, "step": 10790 }, { "epoch": 1.4, "learning_rate": 4.584088404393373e-06, "loss": 0.2393, "step": 10800 }, { "epoch": 1.4, "learning_rate": 4.583792691103843e-06, "loss": 0.2337, "step": 10810 }, { "epoch": 1.4, "learning_rate": 4.583496565176075e-06, "loss": 0.2265, "step": 10820 }, { "epoch": 1.41, "learning_rate": 4.583200026664617e-06, "loss": 0.2488, "step": 10830 }, { "epoch": 1.41, "learning_rate": 4.5829030756240964e-06, "loss": 0.2267, "step": 10840 }, { "epoch": 1.41, "learning_rate": 4.582605712109214e-06, "loss": 0.2355, "step": 10850 }, { "epoch": 1.41, "learning_rate": 4.582307936174746e-06, "loss": 0.2234, "step": 10860 }, { "epoch": 1.41, "learning_rate": 4.582009747875547e-06, "loss": 0.2587, "step": 10870 }, { "epoch": 1.41, "learning_rate": 4.581711147266546e-06, "loss": 0.2514, "step": 10880 }, { "epoch": 1.41, "learning_rate": 4.581412134402747e-06, "loss": 0.233, "step": 10890 }, { "epoch": 1.41, "learning_rate": 4.581112709339233e-06, "loss": 0.2291, "step": 10900 }, { "epoch": 1.42, "learning_rate": 4.58081287213116e-06, "loss": 0.2643, "step": 10910 }, { "epoch": 1.42, "learning_rate": 4.5805126228337615e-06, "loss": 0.2436, "step": 10920 }, { "epoch": 1.42, "learning_rate": 4.580211961502346e-06, "loss": 0.2254, "step": 10930 }, { "epoch": 1.42, "learning_rate": 4.579910888192298e-06, "loss": 0.2415, "step": 10940 }, { "epoch": 1.42, "learning_rate": 4.5796094029590795e-06, "loss": 0.2395, "step": 10950 }, { "epoch": 1.42, "learning_rate": 4.579307505858226e-06, "loss": 0.2394, "step": 10960 }, { "epoch": 1.42, "learning_rate": 4.57900519694535e-06, "loss": 0.2475, "step": 10970 }, { "epoch": 1.43, "learning_rate": 4.5787024762761395e-06, "loss": 0.2443, "step": 10980 }, { "epoch": 1.43, "learning_rate": 4.57839934390636e-06, "loss": 0.2385, "step": 10990 }, { "epoch": 1.43, "learning_rate": 4.578095799891851e-06, "loss": 0.2277, "step": 11000 }, { "epoch": 1.43, "learning_rate": 4.5777918442885275e-06, "loss": 0.2142, "step": 11010 }, { "epoch": 1.43, "learning_rate": 4.577487477152382e-06, "loss": 0.2425, "step": 11020 }, { "epoch": 1.43, "learning_rate": 4.577182698539482e-06, "loss": 0.2449, "step": 11030 }, { "epoch": 1.43, "learning_rate": 4.5768775085059695e-06, "loss": 0.23, "step": 11040 }, { "epoch": 1.43, "learning_rate": 4.5765719071080655e-06, "loss": 0.233, "step": 11050 }, { "epoch": 1.44, "learning_rate": 4.5762658944020635e-06, "loss": 0.2362, "step": 11060 }, { "epoch": 1.44, "learning_rate": 4.575959470444335e-06, "loss": 0.2395, "step": 11070 }, { "epoch": 1.44, "learning_rate": 4.5756526352913254e-06, "loss": 0.2469, "step": 11080 }, { "epoch": 1.44, "learning_rate": 4.575345388999557e-06, "loss": 0.231, "step": 11090 }, { "epoch": 1.44, "learning_rate": 4.575037731625628e-06, "loss": 0.2217, "step": 11100 }, { "epoch": 1.44, "learning_rate": 4.574729663226212e-06, "loss": 0.2433, "step": 11110 }, { "epoch": 1.44, "learning_rate": 4.574421183858057e-06, "loss": 0.2419, "step": 11120 }, { "epoch": 1.44, "learning_rate": 4.57411229357799e-06, "loss": 0.2268, "step": 11130 }, { "epoch": 1.45, "learning_rate": 4.573802992442912e-06, "loss": 0.2521, "step": 11140 }, { "epoch": 1.45, "learning_rate": 4.573493280509797e-06, "loss": 0.2343, "step": 11150 }, { "epoch": 1.45, "learning_rate": 4.573183157835697e-06, "loss": 0.2304, "step": 11160 }, { "epoch": 1.45, "learning_rate": 4.5728726244777415e-06, "loss": 0.2462, "step": 11170 }, { "epoch": 1.45, "learning_rate": 4.572561680493133e-06, "loss": 0.2244, "step": 11180 }, { "epoch": 1.45, "learning_rate": 4.5722503259391515e-06, "loss": 0.2374, "step": 11190 }, { "epoch": 1.45, "learning_rate": 4.57193856087315e-06, "loss": 0.2375, "step": 11200 }, { "epoch": 1.45, "learning_rate": 4.57162638535256e-06, "loss": 0.2374, "step": 11210 }, { "epoch": 1.46, "learning_rate": 4.571313799434887e-06, "loss": 0.2358, "step": 11220 }, { "epoch": 1.46, "learning_rate": 4.571000803177712e-06, "loss": 0.2346, "step": 11230 }, { "epoch": 1.46, "learning_rate": 4.570687396638692e-06, "loss": 0.2242, "step": 11240 }, { "epoch": 1.46, "learning_rate": 4.5703735798755605e-06, "loss": 0.2348, "step": 11250 }, { "epoch": 1.46, "learning_rate": 4.570059352946125e-06, "loss": 0.2612, "step": 11260 }, { "epoch": 1.46, "learning_rate": 4.569744715908269e-06, "loss": 0.2303, "step": 11270 }, { "epoch": 1.46, "learning_rate": 4.569429668819953e-06, "loss": 0.2403, "step": 11280 }, { "epoch": 1.47, "learning_rate": 4.569114211739211e-06, "loss": 0.2509, "step": 11290 }, { "epoch": 1.47, "learning_rate": 4.568798344724152e-06, "loss": 0.2345, "step": 11300 }, { "epoch": 1.47, "learning_rate": 4.568482067832964e-06, "loss": 0.2383, "step": 11310 }, { "epoch": 1.47, "learning_rate": 4.568165381123908e-06, "loss": 0.2377, "step": 11320 }, { "epoch": 1.47, "learning_rate": 4.56784828465532e-06, "loss": 0.2402, "step": 11330 }, { "epoch": 1.47, "learning_rate": 4.567530778485613e-06, "loss": 0.2447, "step": 11340 }, { "epoch": 1.47, "learning_rate": 4.567212862673274e-06, "loss": 0.2396, "step": 11350 }, { "epoch": 1.47, "learning_rate": 4.566894537276867e-06, "loss": 0.2476, "step": 11360 }, { "epoch": 1.48, "learning_rate": 4.566575802355031e-06, "loss": 0.2473, "step": 11370 }, { "epoch": 1.48, "learning_rate": 4.5662566579664794e-06, "loss": 0.2322, "step": 11380 }, { "epoch": 1.48, "learning_rate": 4.565937104170002e-06, "loss": 0.2397, "step": 11390 }, { "epoch": 1.48, "learning_rate": 4.565617141024463e-06, "loss": 0.2222, "step": 11400 }, { "epoch": 1.48, "learning_rate": 4.565296768588806e-06, "loss": 0.243, "step": 11410 }, { "epoch": 1.48, "learning_rate": 4.564975986922042e-06, "loss": 0.2405, "step": 11420 }, { "epoch": 1.48, "learning_rate": 4.564654796083266e-06, "loss": 0.2562, "step": 11430 }, { "epoch": 1.48, "learning_rate": 4.564333196131643e-06, "loss": 0.2405, "step": 11440 }, { "epoch": 1.49, "learning_rate": 4.564011187126415e-06, "loss": 0.2493, "step": 11450 }, { "epoch": 1.49, "learning_rate": 4.5636887691269e-06, "loss": 0.2539, "step": 11460 }, { "epoch": 1.49, "learning_rate": 4.5633659421924905e-06, "loss": 0.2423, "step": 11470 }, { "epoch": 1.49, "learning_rate": 4.563042706382653e-06, "loss": 0.2325, "step": 11480 }, { "epoch": 1.49, "learning_rate": 4.562719061756932e-06, "loss": 0.236, "step": 11490 }, { "epoch": 1.49, "learning_rate": 4.562395008374945e-06, "loss": 0.2662, "step": 11500 }, { "epoch": 1.49, "learning_rate": 4.562070546296388e-06, "loss": 0.2456, "step": 11510 }, { "epoch": 1.5, "learning_rate": 4.561745675581029e-06, "loss": 0.239, "step": 11520 }, { "epoch": 1.5, "learning_rate": 4.561420396288711e-06, "loss": 0.2443, "step": 11530 }, { "epoch": 1.5, "learning_rate": 4.561094708479356e-06, "loss": 0.2242, "step": 11540 }, { "epoch": 1.5, "learning_rate": 4.560768612212957e-06, "loss": 0.2295, "step": 11550 }, { "epoch": 1.5, "learning_rate": 4.560442107549586e-06, "loss": 0.2378, "step": 11560 }, { "epoch": 1.5, "learning_rate": 4.560115194549388e-06, "loss": 0.2331, "step": 11570 }, { "epoch": 1.5, "learning_rate": 4.559787873272582e-06, "loss": 0.2254, "step": 11580 }, { "epoch": 1.5, "learning_rate": 4.559460143779465e-06, "loss": 0.2397, "step": 11590 }, { "epoch": 1.51, "learning_rate": 4.5591320061304096e-06, "loss": 0.2333, "step": 11600 }, { "epoch": 1.51, "learning_rate": 4.558803460385859e-06, "loss": 0.2493, "step": 11610 }, { "epoch": 1.51, "learning_rate": 4.558474506606337e-06, "loss": 0.2434, "step": 11620 }, { "epoch": 1.51, "learning_rate": 4.558145144852439e-06, "loss": 0.2406, "step": 11630 }, { "epoch": 1.51, "learning_rate": 4.557815375184836e-06, "loss": 0.2268, "step": 11640 }, { "epoch": 1.51, "learning_rate": 4.5574851976642775e-06, "loss": 0.2364, "step": 11650 }, { "epoch": 1.51, "learning_rate": 4.557154612351583e-06, "loss": 0.25, "step": 11660 }, { "epoch": 1.51, "learning_rate": 4.55682361930765e-06, "loss": 0.2348, "step": 11670 }, { "epoch": 1.52, "learning_rate": 4.556492218593453e-06, "loss": 0.2391, "step": 11680 }, { "epoch": 1.52, "learning_rate": 4.5561604102700365e-06, "loss": 0.2432, "step": 11690 }, { "epoch": 1.52, "learning_rate": 4.555828194398524e-06, "loss": 0.2329, "step": 11700 }, { "epoch": 1.52, "learning_rate": 4.555495571040112e-06, "loss": 0.2246, "step": 11710 }, { "epoch": 1.52, "learning_rate": 4.555162540256076e-06, "loss": 0.2445, "step": 11720 }, { "epoch": 1.52, "learning_rate": 4.55482910210776e-06, "loss": 0.2425, "step": 11730 }, { "epoch": 1.52, "learning_rate": 4.5544952566565885e-06, "loss": 0.2304, "step": 11740 }, { "epoch": 1.52, "learning_rate": 4.5541610039640595e-06, "loss": 0.2538, "step": 11750 }, { "epoch": 1.53, "learning_rate": 4.553826344091744e-06, "loss": 0.2278, "step": 11760 }, { "epoch": 1.53, "learning_rate": 4.553491277101292e-06, "loss": 0.2297, "step": 11770 }, { "epoch": 1.53, "learning_rate": 4.553155803054423e-06, "loss": 0.225, "step": 11780 }, { "epoch": 1.53, "learning_rate": 4.552819922012938e-06, "loss": 0.2537, "step": 11790 }, { "epoch": 1.53, "learning_rate": 4.552483634038709e-06, "loss": 0.2392, "step": 11800 }, { "epoch": 1.53, "learning_rate": 4.552146939193682e-06, "loss": 0.2404, "step": 11810 }, { "epoch": 1.53, "learning_rate": 4.55180983753988e-06, "loss": 0.2283, "step": 11820 }, { "epoch": 1.54, "learning_rate": 4.5514723291394016e-06, "loss": 0.2265, "step": 11830 }, { "epoch": 1.54, "learning_rate": 4.5511344140544185e-06, "loss": 0.2322, "step": 11840 }, { "epoch": 1.54, "learning_rate": 4.550796092347178e-06, "loss": 0.2333, "step": 11850 }, { "epoch": 1.54, "learning_rate": 4.550457364080002e-06, "loss": 0.2354, "step": 11860 }, { "epoch": 1.54, "learning_rate": 4.550118229315288e-06, "loss": 0.2431, "step": 11870 }, { "epoch": 1.54, "learning_rate": 4.549778688115509e-06, "loss": 0.2474, "step": 11880 }, { "epoch": 1.54, "learning_rate": 4.54943874054321e-06, "loss": 0.2498, "step": 11890 }, { "epoch": 1.54, "learning_rate": 4.5490983866610136e-06, "loss": 0.2341, "step": 11900 }, { "epoch": 1.55, "learning_rate": 4.548757626531617e-06, "loss": 0.232, "step": 11910 }, { "epoch": 1.55, "learning_rate": 4.54841646021779e-06, "loss": 0.2353, "step": 11920 }, { "epoch": 1.55, "learning_rate": 4.548074887782381e-06, "loss": 0.2384, "step": 11930 }, { "epoch": 1.55, "learning_rate": 4.54773290928831e-06, "loss": 0.2367, "step": 11940 }, { "epoch": 1.55, "learning_rate": 4.547390524798572e-06, "loss": 0.2364, "step": 11950 }, { "epoch": 1.55, "learning_rate": 4.54704773437624e-06, "loss": 0.24, "step": 11960 }, { "epoch": 1.55, "learning_rate": 4.546704538084456e-06, "loss": 0.2451, "step": 11970 }, { "epoch": 1.55, "learning_rate": 4.546360935986443e-06, "loss": 0.2393, "step": 11980 }, { "epoch": 1.56, "learning_rate": 4.546016928145495e-06, "loss": 0.2476, "step": 11990 }, { "epoch": 1.56, "learning_rate": 4.545672514624982e-06, "loss": 0.2458, "step": 12000 }, { "epoch": 1.56, "learning_rate": 4.545327695488347e-06, "loss": 0.2408, "step": 12010 }, { "epoch": 1.56, "learning_rate": 4.544982470799111e-06, "loss": 0.2323, "step": 12020 }, { "epoch": 1.56, "learning_rate": 4.544636840620866e-06, "loss": 0.2135, "step": 12030 }, { "epoch": 1.56, "learning_rate": 4.544290805017283e-06, "loss": 0.2356, "step": 12040 }, { "epoch": 1.56, "learning_rate": 4.543944364052104e-06, "loss": 0.2295, "step": 12050 }, { "epoch": 1.57, "learning_rate": 4.543597517789145e-06, "loss": 0.2408, "step": 12060 }, { "epoch": 1.57, "learning_rate": 4.543250266292302e-06, "loss": 0.2357, "step": 12070 }, { "epoch": 1.57, "learning_rate": 4.5429026096255395e-06, "loss": 0.2259, "step": 12080 }, { "epoch": 1.57, "learning_rate": 4.5425545478529e-06, "loss": 0.2474, "step": 12090 }, { "epoch": 1.57, "learning_rate": 4.542206081038501e-06, "loss": 0.2502, "step": 12100 }, { "epoch": 1.57, "learning_rate": 4.541857209246532e-06, "loss": 0.2469, "step": 12110 }, { "epoch": 1.57, "learning_rate": 4.54150793254126e-06, "loss": 0.2383, "step": 12120 }, { "epoch": 1.57, "learning_rate": 4.541158250987024e-06, "loss": 0.2253, "step": 12130 }, { "epoch": 1.58, "learning_rate": 4.5408081646482404e-06, "loss": 0.2397, "step": 12140 }, { "epoch": 1.58, "learning_rate": 4.540457673589398e-06, "loss": 0.229, "step": 12150 }, { "epoch": 1.58, "learning_rate": 4.540106777875059e-06, "loss": 0.2514, "step": 12160 }, { "epoch": 1.58, "learning_rate": 4.539755477569865e-06, "loss": 0.22, "step": 12170 }, { "epoch": 1.58, "learning_rate": 4.539403772738526e-06, "loss": 0.2494, "step": 12180 }, { "epoch": 1.58, "learning_rate": 4.5390516634458315e-06, "loss": 0.2305, "step": 12190 }, { "epoch": 1.58, "learning_rate": 4.538699149756644e-06, "loss": 0.2402, "step": 12200 }, { "epoch": 1.58, "learning_rate": 4.538346231735898e-06, "loss": 0.2416, "step": 12210 }, { "epoch": 1.59, "learning_rate": 4.537992909448606e-06, "loss": 0.2383, "step": 12220 }, { "epoch": 1.59, "learning_rate": 4.537639182959853e-06, "loss": 0.242, "step": 12230 }, { "epoch": 1.59, "learning_rate": 4.537285052334798e-06, "loss": 0.2582, "step": 12240 }, { "epoch": 1.59, "learning_rate": 4.5369305176386774e-06, "loss": 0.2402, "step": 12250 }, { "epoch": 1.59, "learning_rate": 4.536575578936799e-06, "loss": 0.2437, "step": 12260 }, { "epoch": 1.59, "learning_rate": 4.536220236294546e-06, "loss": 0.2299, "step": 12270 }, { "epoch": 1.59, "learning_rate": 4.535864489777377e-06, "loss": 0.2385, "step": 12280 }, { "epoch": 1.6, "learning_rate": 4.535508339450822e-06, "loss": 0.248, "step": 12290 }, { "epoch": 1.6, "learning_rate": 4.535151785380489e-06, "loss": 0.2424, "step": 12300 }, { "epoch": 1.6, "learning_rate": 4.534794827632059e-06, "loss": 0.2281, "step": 12310 }, { "epoch": 1.6, "learning_rate": 4.534437466271286e-06, "loss": 0.246, "step": 12320 }, { "epoch": 1.6, "learning_rate": 4.534079701364001e-06, "loss": 0.2515, "step": 12330 }, { "epoch": 1.6, "learning_rate": 4.5337215329761064e-06, "loss": 0.2456, "step": 12340 }, { "epoch": 1.6, "learning_rate": 4.533362961173581e-06, "loss": 0.2346, "step": 12350 }, { "epoch": 1.6, "learning_rate": 4.533003986022478e-06, "loss": 0.2239, "step": 12360 }, { "epoch": 1.61, "learning_rate": 4.5326446075889235e-06, "loss": 0.2342, "step": 12370 }, { "epoch": 1.61, "learning_rate": 4.532284825939118e-06, "loss": 0.2439, "step": 12380 }, { "epoch": 1.61, "learning_rate": 4.531924641139338e-06, "loss": 0.2359, "step": 12390 }, { "epoch": 1.61, "learning_rate": 4.5315640532559334e-06, "loss": 0.2389, "step": 12400 }, { "epoch": 1.61, "learning_rate": 4.531203062355326e-06, "loss": 0.2256, "step": 12410 }, { "epoch": 1.61, "learning_rate": 4.530841668504017e-06, "loss": 0.2509, "step": 12420 }, { "epoch": 1.61, "learning_rate": 4.530479871768576e-06, "loss": 0.2332, "step": 12430 }, { "epoch": 1.61, "learning_rate": 4.530117672215651e-06, "loss": 0.2352, "step": 12440 }, { "epoch": 1.62, "learning_rate": 4.529755069911962e-06, "loss": 0.2479, "step": 12450 }, { "epoch": 1.62, "learning_rate": 4.529392064924305e-06, "loss": 0.227, "step": 12460 }, { "epoch": 1.62, "learning_rate": 4.529028657319548e-06, "loss": 0.2527, "step": 12470 }, { "epoch": 1.62, "learning_rate": 4.528664847164635e-06, "loss": 0.2455, "step": 12480 }, { "epoch": 1.62, "learning_rate": 4.528300634526583e-06, "loss": 0.2445, "step": 12490 }, { "epoch": 1.62, "learning_rate": 4.527936019472484e-06, "loss": 0.2449, "step": 12500 }, { "epoch": 1.62, "learning_rate": 4.527571002069504e-06, "loss": 0.2518, "step": 12510 }, { "epoch": 1.62, "learning_rate": 4.527205582384883e-06, "loss": 0.2248, "step": 12520 }, { "epoch": 1.63, "learning_rate": 4.526839760485933e-06, "loss": 0.2542, "step": 12530 }, { "epoch": 1.63, "learning_rate": 4.526473536440046e-06, "loss": 0.2316, "step": 12540 }, { "epoch": 1.63, "learning_rate": 4.526106910314679e-06, "loss": 0.2368, "step": 12550 }, { "epoch": 1.63, "learning_rate": 4.525739882177372e-06, "loss": 0.2322, "step": 12560 }, { "epoch": 1.63, "learning_rate": 4.525372452095734e-06, "loss": 0.2435, "step": 12570 }, { "epoch": 1.63, "learning_rate": 4.52500462013745e-06, "loss": 0.2326, "step": 12580 }, { "epoch": 1.63, "learning_rate": 4.5246363863702766e-06, "loss": 0.2405, "step": 12590 }, { "epoch": 1.64, "learning_rate": 4.524267750862048e-06, "loss": 0.2372, "step": 12600 }, { "epoch": 1.64, "learning_rate": 4.523898713680669e-06, "loss": 0.2256, "step": 12610 }, { "epoch": 1.64, "learning_rate": 4.5235292748941204e-06, "loss": 0.2413, "step": 12620 }, { "epoch": 1.64, "learning_rate": 4.523159434570458e-06, "loss": 0.2298, "step": 12630 }, { "epoch": 1.64, "learning_rate": 4.522789192777809e-06, "loss": 0.2437, "step": 12640 }, { "epoch": 1.64, "learning_rate": 4.522418549584374e-06, "loss": 0.2265, "step": 12650 }, { "epoch": 1.64, "learning_rate": 4.5220475050584325e-06, "loss": 0.2259, "step": 12660 }, { "epoch": 1.64, "learning_rate": 4.5216760592683315e-06, "loss": 0.2358, "step": 12670 }, { "epoch": 1.65, "learning_rate": 4.521304212282498e-06, "loss": 0.2443, "step": 12680 }, { "epoch": 1.65, "learning_rate": 4.520931964169427e-06, "loss": 0.2471, "step": 12690 }, { "epoch": 1.65, "learning_rate": 4.520559314997691e-06, "loss": 0.2289, "step": 12700 }, { "epoch": 1.65, "learning_rate": 4.520186264835937e-06, "loss": 0.2309, "step": 12710 }, { "epoch": 1.65, "learning_rate": 4.5198128137528845e-06, "loss": 0.2228, "step": 12720 }, { "epoch": 1.65, "learning_rate": 4.5194389618173255e-06, "loss": 0.2374, "step": 12730 }, { "epoch": 1.65, "learning_rate": 4.519064709098129e-06, "loss": 0.2512, "step": 12740 }, { "epoch": 1.65, "learning_rate": 4.5186900556642336e-06, "loss": 0.2283, "step": 12750 }, { "epoch": 1.66, "learning_rate": 4.518315001584657e-06, "loss": 0.2365, "step": 12760 }, { "epoch": 1.66, "learning_rate": 4.517939546928487e-06, "loss": 0.2262, "step": 12770 }, { "epoch": 1.66, "learning_rate": 4.517563691764885e-06, "loss": 0.2429, "step": 12780 }, { "epoch": 1.66, "learning_rate": 4.517187436163087e-06, "loss": 0.2224, "step": 12790 }, { "epoch": 1.66, "learning_rate": 4.516810780192405e-06, "loss": 0.2374, "step": 12800 }, { "epoch": 1.66, "learning_rate": 4.5164337239222225e-06, "loss": 0.2408, "step": 12810 }, { "epoch": 1.66, "learning_rate": 4.516056267421995e-06, "loss": 0.2427, "step": 12820 }, { "epoch": 1.67, "learning_rate": 4.5156784107612565e-06, "loss": 0.2528, "step": 12830 }, { "epoch": 1.67, "learning_rate": 4.51530015400961e-06, "loss": 0.2223, "step": 12840 }, { "epoch": 1.67, "learning_rate": 4.514921497236735e-06, "loss": 0.2544, "step": 12850 }, { "epoch": 1.67, "learning_rate": 4.514542440512383e-06, "loss": 0.2341, "step": 12860 }, { "epoch": 1.67, "learning_rate": 4.514162983906381e-06, "loss": 0.235, "step": 12870 }, { "epoch": 1.67, "learning_rate": 4.513783127488629e-06, "loss": 0.2464, "step": 12880 }, { "epoch": 1.67, "learning_rate": 4.513402871329099e-06, "loss": 0.2424, "step": 12890 }, { "epoch": 1.67, "learning_rate": 4.51302221549784e-06, "loss": 0.2414, "step": 12900 }, { "epoch": 1.68, "learning_rate": 4.512641160064969e-06, "loss": 0.227, "step": 12910 }, { "epoch": 1.68, "learning_rate": 4.512259705100684e-06, "loss": 0.2459, "step": 12920 }, { "epoch": 1.68, "learning_rate": 4.511877850675251e-06, "loss": 0.2416, "step": 12930 }, { "epoch": 1.68, "learning_rate": 4.511495596859013e-06, "loss": 0.229, "step": 12940 }, { "epoch": 1.68, "learning_rate": 4.511112943722382e-06, "loss": 0.2446, "step": 12950 }, { "epoch": 1.68, "learning_rate": 4.5107298913358485e-06, "loss": 0.2345, "step": 12960 }, { "epoch": 1.68, "learning_rate": 4.510346439769975e-06, "loss": 0.2334, "step": 12970 }, { "epoch": 1.68, "learning_rate": 4.509962589095396e-06, "loss": 0.2372, "step": 12980 }, { "epoch": 1.69, "learning_rate": 4.509578339382822e-06, "loss": 0.2407, "step": 12990 }, { "epoch": 1.69, "learning_rate": 4.509193690703034e-06, "loss": 0.2559, "step": 13000 }, { "epoch": 1.69, "learning_rate": 4.508808643126888e-06, "loss": 0.2422, "step": 13010 }, { "epoch": 1.69, "learning_rate": 4.508423196725315e-06, "loss": 0.2375, "step": 13020 }, { "epoch": 1.69, "learning_rate": 4.5080373515693175e-06, "loss": 0.2393, "step": 13030 }, { "epoch": 1.69, "learning_rate": 4.507651107729972e-06, "loss": 0.2353, "step": 13040 }, { "epoch": 1.69, "learning_rate": 4.507264465278429e-06, "loss": 0.2474, "step": 13050 }, { "epoch": 1.7, "learning_rate": 4.506877424285911e-06, "loss": 0.2423, "step": 13060 }, { "epoch": 1.7, "learning_rate": 4.506489984823715e-06, "loss": 0.2267, "step": 13070 }, { "epoch": 1.7, "learning_rate": 4.506102146963212e-06, "loss": 0.243, "step": 13080 }, { "epoch": 1.7, "learning_rate": 4.5057139107758455e-06, "loss": 0.2425, "step": 13090 }, { "epoch": 1.7, "learning_rate": 4.505325276333131e-06, "loss": 0.2319, "step": 13100 }, { "epoch": 1.7, "learning_rate": 4.50493624370666e-06, "loss": 0.2304, "step": 13110 }, { "epoch": 1.7, "learning_rate": 4.504546812968097e-06, "loss": 0.2302, "step": 13120 }, { "epoch": 1.7, "learning_rate": 4.504156984189177e-06, "loss": 0.2489, "step": 13130 }, { "epoch": 1.71, "learning_rate": 4.503766757441713e-06, "loss": 0.2355, "step": 13140 }, { "epoch": 1.71, "learning_rate": 4.503376132797585e-06, "loss": 0.2322, "step": 13150 }, { "epoch": 1.71, "learning_rate": 4.502985110328753e-06, "loss": 0.2388, "step": 13160 }, { "epoch": 1.71, "learning_rate": 4.502593690107247e-06, "loss": 0.225, "step": 13170 }, { "epoch": 1.71, "learning_rate": 4.502201872205169e-06, "loss": 0.2432, "step": 13180 }, { "epoch": 1.71, "learning_rate": 4.501809656694698e-06, "loss": 0.2516, "step": 13190 }, { "epoch": 1.71, "learning_rate": 4.501417043648081e-06, "loss": 0.2375, "step": 13200 }, { "epoch": 1.71, "learning_rate": 4.501024033137643e-06, "loss": 0.2258, "step": 13210 }, { "epoch": 1.72, "learning_rate": 4.500630625235781e-06, "loss": 0.2352, "step": 13220 }, { "epoch": 1.72, "learning_rate": 4.500236820014964e-06, "loss": 0.236, "step": 13230 }, { "epoch": 1.72, "learning_rate": 4.499842617547734e-06, "loss": 0.2432, "step": 13240 }, { "epoch": 1.72, "learning_rate": 4.4994480179067095e-06, "loss": 0.2401, "step": 13250 }, { "epoch": 1.72, "learning_rate": 4.499053021164577e-06, "loss": 0.2339, "step": 13260 }, { "epoch": 1.72, "learning_rate": 4.4986576273941e-06, "loss": 0.2558, "step": 13270 }, { "epoch": 1.72, "learning_rate": 4.498261836668114e-06, "loss": 0.2277, "step": 13280 }, { "epoch": 1.72, "learning_rate": 4.497865649059528e-06, "loss": 0.2637, "step": 13290 }, { "epoch": 1.73, "learning_rate": 4.497469064641323e-06, "loss": 0.2209, "step": 13300 }, { "epoch": 1.73, "learning_rate": 4.497072083486554e-06, "loss": 0.223, "step": 13310 }, { "epoch": 1.73, "learning_rate": 4.4966747056683485e-06, "loss": 0.2515, "step": 13320 }, { "epoch": 1.73, "learning_rate": 4.4962769312599086e-06, "loss": 0.2397, "step": 13330 }, { "epoch": 1.73, "learning_rate": 4.4958787603345065e-06, "loss": 0.2447, "step": 13340 }, { "epoch": 1.73, "learning_rate": 4.495480192965492e-06, "loss": 0.2432, "step": 13350 }, { "epoch": 1.73, "learning_rate": 4.495081229226283e-06, "loss": 0.2323, "step": 13360 }, { "epoch": 1.74, "learning_rate": 4.494681869190373e-06, "loss": 0.2421, "step": 13370 }, { "epoch": 1.74, "learning_rate": 4.494282112931328e-06, "loss": 0.2441, "step": 13380 }, { "epoch": 1.74, "learning_rate": 4.493881960522787e-06, "loss": 0.2456, "step": 13390 }, { "epoch": 1.74, "learning_rate": 4.493481412038463e-06, "loss": 0.2383, "step": 13400 }, { "epoch": 1.74, "learning_rate": 4.493080467552142e-06, "loss": 0.2398, "step": 13410 }, { "epoch": 1.74, "learning_rate": 4.492679127137678e-06, "loss": 0.2354, "step": 13420 }, { "epoch": 1.74, "learning_rate": 4.492277390869006e-06, "loss": 0.2484, "step": 13430 }, { "epoch": 1.74, "learning_rate": 4.491875258820127e-06, "loss": 0.2455, "step": 13440 }, { "epoch": 1.75, "learning_rate": 4.49147273106512e-06, "loss": 0.2304, "step": 13450 }, { "epoch": 1.75, "learning_rate": 4.491069807678133e-06, "loss": 0.2442, "step": 13460 }, { "epoch": 1.75, "learning_rate": 4.4906664887333895e-06, "loss": 0.247, "step": 13470 }, { "epoch": 1.75, "learning_rate": 4.490262774305185e-06, "loss": 0.2295, "step": 13480 }, { "epoch": 1.75, "learning_rate": 4.489858664467886e-06, "loss": 0.2551, "step": 13490 }, { "epoch": 1.75, "learning_rate": 4.489454159295935e-06, "loss": 0.2294, "step": 13500 }, { "epoch": 1.75, "learning_rate": 4.489049258863846e-06, "loss": 0.2322, "step": 13510 }, { "epoch": 1.75, "learning_rate": 4.488643963246205e-06, "loss": 0.2397, "step": 13520 }, { "epoch": 1.76, "learning_rate": 4.488238272517672e-06, "loss": 0.2337, "step": 13530 }, { "epoch": 1.76, "learning_rate": 4.487832186752979e-06, "loss": 0.2331, "step": 13540 }, { "epoch": 1.76, "learning_rate": 4.4874257060269326e-06, "loss": 0.2393, "step": 13550 }, { "epoch": 1.76, "learning_rate": 4.487018830414408e-06, "loss": 0.2324, "step": 13560 }, { "epoch": 1.76, "learning_rate": 4.486611559990357e-06, "loss": 0.2342, "step": 13570 }, { "epoch": 1.76, "learning_rate": 4.486203894829804e-06, "loss": 0.2342, "step": 13580 }, { "epoch": 1.76, "learning_rate": 4.485795835007842e-06, "loss": 0.2351, "step": 13590 }, { "epoch": 1.77, "learning_rate": 4.485387380599642e-06, "loss": 0.2373, "step": 13600 }, { "epoch": 1.77, "learning_rate": 4.4849785316804465e-06, "loss": 0.2323, "step": 13610 }, { "epoch": 1.77, "learning_rate": 4.484569288325566e-06, "loss": 0.2424, "step": 13620 }, { "epoch": 1.77, "learning_rate": 4.484159650610391e-06, "loss": 0.241, "step": 13630 }, { "epoch": 1.77, "learning_rate": 4.483749618610378e-06, "loss": 0.2326, "step": 13640 }, { "epoch": 1.77, "learning_rate": 4.48333919240106e-06, "loss": 0.2446, "step": 13650 }, { "epoch": 1.77, "learning_rate": 4.482928372058041e-06, "loss": 0.2332, "step": 13660 }, { "epoch": 1.77, "learning_rate": 4.482517157657e-06, "loss": 0.2372, "step": 13670 }, { "epoch": 1.78, "learning_rate": 4.482105549273686e-06, "loss": 0.243, "step": 13680 }, { "epoch": 1.78, "learning_rate": 4.48169354698392e-06, "loss": 0.2371, "step": 13690 }, { "epoch": 1.78, "learning_rate": 4.481281150863599e-06, "loss": 0.2284, "step": 13700 }, { "epoch": 1.78, "learning_rate": 4.480868360988688e-06, "loss": 0.2262, "step": 13710 }, { "epoch": 1.78, "learning_rate": 4.480455177435229e-06, "loss": 0.2309, "step": 13720 }, { "epoch": 1.78, "learning_rate": 4.480041600279335e-06, "loss": 0.2237, "step": 13730 }, { "epoch": 1.78, "learning_rate": 4.479627629597189e-06, "loss": 0.2283, "step": 13740 }, { "epoch": 1.78, "learning_rate": 4.47921326546505e-06, "loss": 0.2408, "step": 13750 }, { "epoch": 1.79, "learning_rate": 4.478798507959248e-06, "loss": 0.2383, "step": 13760 }, { "epoch": 1.79, "learning_rate": 4.478383357156184e-06, "loss": 0.2303, "step": 13770 }, { "epoch": 1.79, "learning_rate": 4.477967813132335e-06, "loss": 0.2345, "step": 13780 }, { "epoch": 1.79, "learning_rate": 4.477551875964247e-06, "loss": 0.2249, "step": 13790 }, { "epoch": 1.79, "learning_rate": 4.477135545728541e-06, "loss": 0.24, "step": 13800 }, { "epoch": 1.79, "learning_rate": 4.476718822501907e-06, "loss": 0.2308, "step": 13810 }, { "epoch": 1.79, "learning_rate": 4.476301706361111e-06, "loss": 0.2276, "step": 13820 }, { "epoch": 1.79, "learning_rate": 4.475884197382991e-06, "loss": 0.2231, "step": 13830 }, { "epoch": 1.8, "learning_rate": 4.475466295644454e-06, "loss": 0.2508, "step": 13840 }, { "epoch": 1.8, "learning_rate": 4.475048001222484e-06, "loss": 0.2403, "step": 13850 }, { "epoch": 1.8, "learning_rate": 4.474629314194133e-06, "loss": 0.2305, "step": 13860 }, { "epoch": 1.8, "learning_rate": 4.474210234636528e-06, "loss": 0.237, "step": 13870 }, { "epoch": 1.8, "learning_rate": 4.473790762626869e-06, "loss": 0.237, "step": 13880 }, { "epoch": 1.8, "learning_rate": 4.473370898242425e-06, "loss": 0.2285, "step": 13890 }, { "epoch": 1.8, "learning_rate": 4.4729506415605395e-06, "loss": 0.235, "step": 13900 }, { "epoch": 1.81, "learning_rate": 4.47252999265863e-06, "loss": 0.2328, "step": 13910 }, { "epoch": 1.81, "learning_rate": 4.472108951614182e-06, "loss": 0.23, "step": 13920 }, { "epoch": 1.81, "learning_rate": 4.471687518504756e-06, "loss": 0.2277, "step": 13930 }, { "epoch": 1.81, "learning_rate": 4.471265693407984e-06, "loss": 0.2306, "step": 13940 }, { "epoch": 1.81, "learning_rate": 4.470843476401572e-06, "loss": 0.2401, "step": 13950 }, { "epoch": 1.81, "learning_rate": 4.470420867563294e-06, "loss": 0.2461, "step": 13960 }, { "epoch": 1.81, "learning_rate": 4.4699978669710005e-06, "loss": 0.2316, "step": 13970 }, { "epoch": 1.81, "learning_rate": 4.469574474702612e-06, "loss": 0.2262, "step": 13980 }, { "epoch": 1.82, "learning_rate": 4.469150690836122e-06, "loss": 0.2311, "step": 13990 }, { "epoch": 1.82, "learning_rate": 4.468726515449596e-06, "loss": 0.2279, "step": 14000 }, { "epoch": 1.82, "learning_rate": 4.468301948621169e-06, "loss": 0.2481, "step": 14010 }, { "epoch": 1.82, "learning_rate": 4.467876990429053e-06, "loss": 0.2499, "step": 14020 }, { "epoch": 1.82, "learning_rate": 4.467451640951529e-06, "loss": 0.2328, "step": 14030 }, { "epoch": 1.82, "learning_rate": 4.46702590026695e-06, "loss": 0.2363, "step": 14040 }, { "epoch": 1.82, "learning_rate": 4.466599768453742e-06, "loss": 0.2366, "step": 14050 }, { "epoch": 1.82, "learning_rate": 4.466173245590403e-06, "loss": 0.2437, "step": 14060 }, { "epoch": 1.83, "learning_rate": 4.4657463317555025e-06, "loss": 0.2532, "step": 14070 }, { "epoch": 1.83, "learning_rate": 4.465319027027682e-06, "loss": 0.2512, "step": 14080 }, { "epoch": 1.83, "learning_rate": 4.464891331485656e-06, "loss": 0.2542, "step": 14090 }, { "epoch": 1.83, "learning_rate": 4.46446324520821e-06, "loss": 0.2407, "step": 14100 }, { "epoch": 1.83, "learning_rate": 4.464034768274201e-06, "loss": 0.2329, "step": 14110 }, { "epoch": 1.83, "learning_rate": 4.46360590076256e-06, "loss": 0.23, "step": 14120 }, { "epoch": 1.83, "learning_rate": 4.463176642752288e-06, "loss": 0.2318, "step": 14130 }, { "epoch": 1.84, "learning_rate": 4.4627469943224575e-06, "loss": 0.2423, "step": 14140 }, { "epoch": 1.84, "learning_rate": 4.4623169555522174e-06, "loss": 0.235, "step": 14150 }, { "epoch": 1.84, "learning_rate": 4.461886526520781e-06, "loss": 0.2376, "step": 14160 }, { "epoch": 1.84, "learning_rate": 4.461455707307441e-06, "loss": 0.2262, "step": 14170 }, { "epoch": 1.84, "learning_rate": 4.461024497991556e-06, "loss": 0.257, "step": 14180 }, { "epoch": 1.84, "learning_rate": 4.460592898652561e-06, "loss": 0.2285, "step": 14190 }, { "epoch": 1.84, "learning_rate": 4.46016090936996e-06, "loss": 0.237, "step": 14200 }, { "epoch": 1.84, "learning_rate": 4.45972853022333e-06, "loss": 0.2228, "step": 14210 }, { "epoch": 1.85, "learning_rate": 4.459295761292319e-06, "loss": 0.2286, "step": 14220 }, { "epoch": 1.85, "learning_rate": 4.458862602656648e-06, "loss": 0.2312, "step": 14230 }, { "epoch": 1.85, "learning_rate": 4.45842905439611e-06, "loss": 0.2508, "step": 14240 }, { "epoch": 1.85, "learning_rate": 4.457995116590567e-06, "loss": 0.247, "step": 14250 }, { "epoch": 1.85, "learning_rate": 4.457560789319955e-06, "loss": 0.2223, "step": 14260 }, { "epoch": 1.85, "learning_rate": 4.457126072664283e-06, "loss": 0.2405, "step": 14270 }, { "epoch": 1.85, "learning_rate": 4.456690966703628e-06, "loss": 0.2463, "step": 14280 }, { "epoch": 1.85, "learning_rate": 4.456255471518143e-06, "loss": 0.2308, "step": 14290 }, { "epoch": 1.86, "learning_rate": 4.4558195871880495e-06, "loss": 0.2417, "step": 14300 }, { "epoch": 1.86, "learning_rate": 4.4553833137936416e-06, "loss": 0.2407, "step": 14310 }, { "epoch": 1.86, "learning_rate": 4.454946651415285e-06, "loss": 0.2307, "step": 14320 }, { "epoch": 1.86, "learning_rate": 4.454509600133418e-06, "loss": 0.238, "step": 14330 }, { "epoch": 1.86, "learning_rate": 4.45407216002855e-06, "loss": 0.2445, "step": 14340 }, { "epoch": 1.86, "learning_rate": 4.453634331181261e-06, "loss": 0.2344, "step": 14350 }, { "epoch": 1.86, "learning_rate": 4.453196113672203e-06, "loss": 0.2425, "step": 14360 }, { "epoch": 1.87, "learning_rate": 4.452757507582102e-06, "loss": 0.2375, "step": 14370 }, { "epoch": 1.87, "learning_rate": 4.452318512991752e-06, "loss": 0.2292, "step": 14380 }, { "epoch": 1.87, "learning_rate": 4.451879129982021e-06, "loss": 0.2472, "step": 14390 }, { "epoch": 1.87, "learning_rate": 4.4514393586338476e-06, "loss": 0.2477, "step": 14400 }, { "epoch": 1.87, "learning_rate": 4.4509991990282405e-06, "loss": 0.2364, "step": 14410 }, { "epoch": 1.87, "learning_rate": 4.450558651246284e-06, "loss": 0.2209, "step": 14420 }, { "epoch": 1.87, "learning_rate": 4.45011771536913e-06, "loss": 0.2463, "step": 14430 }, { "epoch": 1.87, "learning_rate": 4.449676391478004e-06, "loss": 0.2401, "step": 14440 }, { "epoch": 1.88, "learning_rate": 4.449234679654201e-06, "loss": 0.2405, "step": 14450 }, { "epoch": 1.88, "learning_rate": 4.4487925799790896e-06, "loss": 0.2406, "step": 14460 }, { "epoch": 1.88, "learning_rate": 4.44835009253411e-06, "loss": 0.2247, "step": 14470 }, { "epoch": 1.88, "learning_rate": 4.447907217400771e-06, "loss": 0.2324, "step": 14480 }, { "epoch": 1.88, "learning_rate": 4.447463954660655e-06, "loss": 0.2277, "step": 14490 }, { "epoch": 1.88, "learning_rate": 4.447020304395416e-06, "loss": 0.2374, "step": 14500 }, { "epoch": 1.88, "learning_rate": 4.4465762666867786e-06, "loss": 0.2404, "step": 14510 }, { "epoch": 1.88, "learning_rate": 4.4461318416165395e-06, "loss": 0.237, "step": 14520 }, { "epoch": 1.89, "learning_rate": 4.445687029266565e-06, "loss": 0.2462, "step": 14530 }, { "epoch": 1.89, "learning_rate": 4.445241829718795e-06, "loss": 0.2406, "step": 14540 }, { "epoch": 1.89, "learning_rate": 4.4447962430552395e-06, "loss": 0.2356, "step": 14550 }, { "epoch": 1.89, "learning_rate": 4.44435026935798e-06, "loss": 0.2409, "step": 14560 }, { "epoch": 1.89, "learning_rate": 4.443903908709168e-06, "loss": 0.2344, "step": 14570 }, { "epoch": 1.89, "learning_rate": 4.443457161191031e-06, "loss": 0.2343, "step": 14580 }, { "epoch": 1.89, "learning_rate": 4.443010026885861e-06, "loss": 0.2502, "step": 14590 }, { "epoch": 1.89, "learning_rate": 4.4425625058760255e-06, "loss": 0.2416, "step": 14600 }, { "epoch": 1.9, "learning_rate": 4.4421145982439624e-06, "loss": 0.2397, "step": 14610 }, { "epoch": 1.9, "learning_rate": 4.4416663040721815e-06, "loss": 0.2212, "step": 14620 }, { "epoch": 1.9, "learning_rate": 4.4412176234432625e-06, "loss": 0.2393, "step": 14630 }, { "epoch": 1.9, "learning_rate": 4.440768556439857e-06, "loss": 0.2443, "step": 14640 }, { "epoch": 1.9, "learning_rate": 4.440319103144687e-06, "loss": 0.238, "step": 14650 }, { "epoch": 1.9, "learning_rate": 4.439869263640547e-06, "loss": 0.2281, "step": 14660 }, { "epoch": 1.9, "learning_rate": 4.439419038010302e-06, "loss": 0.2227, "step": 14670 }, { "epoch": 1.91, "learning_rate": 4.438968426336888e-06, "loss": 0.2394, "step": 14680 }, { "epoch": 1.91, "learning_rate": 4.438517428703311e-06, "loss": 0.2411, "step": 14690 }, { "epoch": 1.91, "learning_rate": 4.438066045192651e-06, "loss": 0.23, "step": 14700 }, { "epoch": 1.91, "learning_rate": 4.437614275888056e-06, "loss": 0.2371, "step": 14710 }, { "epoch": 1.91, "learning_rate": 4.437162120872748e-06, "loss": 0.2436, "step": 14720 }, { "epoch": 1.91, "learning_rate": 4.4367095802300176e-06, "loss": 0.2492, "step": 14730 }, { "epoch": 1.91, "learning_rate": 4.436256654043226e-06, "loss": 0.2367, "step": 14740 }, { "epoch": 1.91, "learning_rate": 4.435803342395808e-06, "loss": 0.2374, "step": 14750 }, { "epoch": 1.92, "learning_rate": 4.435349645371269e-06, "loss": 0.2425, "step": 14760 }, { "epoch": 1.92, "learning_rate": 4.4348955630531824e-06, "loss": 0.2337, "step": 14770 }, { "epoch": 1.92, "learning_rate": 4.434441095525196e-06, "loss": 0.2439, "step": 14780 }, { "epoch": 1.92, "learning_rate": 4.433986242871028e-06, "loss": 0.2305, "step": 14790 }, { "epoch": 1.92, "learning_rate": 4.433531005174466e-06, "loss": 0.2366, "step": 14800 }, { "epoch": 1.92, "learning_rate": 4.433075382519368e-06, "loss": 0.2305, "step": 14810 }, { "epoch": 1.92, "learning_rate": 4.4326193749896656e-06, "loss": 0.2441, "step": 14820 }, { "epoch": 1.92, "learning_rate": 4.432162982669361e-06, "loss": 0.236, "step": 14830 }, { "epoch": 1.93, "learning_rate": 4.431706205642524e-06, "loss": 0.2349, "step": 14840 }, { "epoch": 1.93, "learning_rate": 4.431249043993299e-06, "loss": 0.2375, "step": 14850 }, { "epoch": 1.93, "learning_rate": 4.430791497805899e-06, "loss": 0.2294, "step": 14860 }, { "epoch": 1.93, "learning_rate": 4.43033356716461e-06, "loss": 0.2266, "step": 14870 }, { "epoch": 1.93, "learning_rate": 4.429875252153785e-06, "loss": 0.2541, "step": 14880 }, { "epoch": 1.93, "learning_rate": 4.429416552857852e-06, "loss": 0.2452, "step": 14890 }, { "epoch": 1.93, "learning_rate": 4.428957469361307e-06, "loss": 0.2346, "step": 14900 }, { "epoch": 1.94, "learning_rate": 4.4284980017487186e-06, "loss": 0.2278, "step": 14910 }, { "epoch": 1.94, "learning_rate": 4.428038150104725e-06, "loss": 0.2362, "step": 14920 }, { "epoch": 1.94, "learning_rate": 4.427577914514035e-06, "loss": 0.2179, "step": 14930 }, { "epoch": 1.94, "learning_rate": 4.4271172950614295e-06, "loss": 0.2302, "step": 14940 }, { "epoch": 1.94, "learning_rate": 4.426656291831759e-06, "loss": 0.2303, "step": 14950 }, { "epoch": 1.94, "learning_rate": 4.426194904909943e-06, "loss": 0.2239, "step": 14960 }, { "epoch": 1.94, "learning_rate": 4.425733134380977e-06, "loss": 0.2306, "step": 14970 }, { "epoch": 1.94, "learning_rate": 4.425270980329921e-06, "loss": 0.2284, "step": 14980 }, { "epoch": 1.95, "learning_rate": 4.42480844284191e-06, "loss": 0.251, "step": 14990 }, { "epoch": 1.95, "learning_rate": 4.424345522002147e-06, "loss": 0.2432, "step": 15000 }, { "epoch": 1.95, "learning_rate": 4.423882217895907e-06, "loss": 0.2458, "step": 15010 }, { "epoch": 1.95, "learning_rate": 4.4234185306085355e-06, "loss": 0.2374, "step": 15020 }, { "epoch": 1.95, "learning_rate": 4.422954460225447e-06, "loss": 0.2465, "step": 15030 }, { "epoch": 1.95, "learning_rate": 4.42249000683213e-06, "loss": 0.2458, "step": 15040 }, { "epoch": 1.95, "learning_rate": 4.42202517051414e-06, "loss": 0.231, "step": 15050 }, { "epoch": 1.95, "learning_rate": 4.421559951357105e-06, "loss": 0.2247, "step": 15060 }, { "epoch": 1.96, "learning_rate": 4.421094349446723e-06, "loss": 0.2345, "step": 15070 }, { "epoch": 1.96, "learning_rate": 4.420628364868762e-06, "loss": 0.2303, "step": 15080 }, { "epoch": 1.96, "learning_rate": 4.420161997709063e-06, "loss": 0.2406, "step": 15090 }, { "epoch": 1.96, "learning_rate": 4.419695248053532e-06, "loss": 0.2413, "step": 15100 }, { "epoch": 1.96, "learning_rate": 4.4192281159881524e-06, "loss": 0.2575, "step": 15110 }, { "epoch": 1.96, "learning_rate": 4.418760601598972e-06, "loss": 0.2523, "step": 15120 }, { "epoch": 1.96, "learning_rate": 4.418292704972114e-06, "loss": 0.2279, "step": 15130 }, { "epoch": 1.96, "learning_rate": 4.417824426193767e-06, "loss": 0.2268, "step": 15140 }, { "epoch": 1.97, "learning_rate": 4.4173557653501945e-06, "loss": 0.2505, "step": 15150 }, { "epoch": 1.97, "learning_rate": 4.416886722527728e-06, "loss": 0.2249, "step": 15160 }, { "epoch": 1.97, "learning_rate": 4.41641729781277e-06, "loss": 0.2491, "step": 15170 }, { "epoch": 1.97, "learning_rate": 4.415947491291793e-06, "loss": 0.2411, "step": 15180 }, { "epoch": 1.97, "learning_rate": 4.415477303051339e-06, "loss": 0.239, "step": 15190 }, { "epoch": 1.97, "learning_rate": 4.415006733178024e-06, "loss": 0.228, "step": 15200 }, { "epoch": 1.97, "learning_rate": 4.414535781758531e-06, "loss": 0.2452, "step": 15210 }, { "epoch": 1.98, "learning_rate": 4.414064448879611e-06, "loss": 0.2431, "step": 15220 }, { "epoch": 1.98, "learning_rate": 4.4135927346280915e-06, "loss": 0.2243, "step": 15230 }, { "epoch": 1.98, "learning_rate": 4.413120639090866e-06, "loss": 0.2366, "step": 15240 }, { "epoch": 1.98, "learning_rate": 4.412648162354898e-06, "loss": 0.2247, "step": 15250 }, { "epoch": 1.98, "learning_rate": 4.412175304507224e-06, "loss": 0.2361, "step": 15260 }, { "epoch": 1.98, "learning_rate": 4.411702065634948e-06, "loss": 0.2433, "step": 15270 }, { "epoch": 1.98, "learning_rate": 4.411228445825247e-06, "loss": 0.2486, "step": 15280 }, { "epoch": 1.98, "learning_rate": 4.410754445165365e-06, "loss": 0.243, "step": 15290 }, { "epoch": 1.99, "learning_rate": 4.4102800637426185e-06, "loss": 0.2257, "step": 15300 }, { "epoch": 1.99, "learning_rate": 4.409805301644393e-06, "loss": 0.2277, "step": 15310 }, { "epoch": 1.99, "learning_rate": 4.409330158958144e-06, "loss": 0.2365, "step": 15320 }, { "epoch": 1.99, "learning_rate": 4.408854635771397e-06, "loss": 0.2401, "step": 15330 }, { "epoch": 1.99, "learning_rate": 4.408378732171751e-06, "loss": 0.2469, "step": 15340 }, { "epoch": 1.99, "learning_rate": 4.407902448246868e-06, "loss": 0.2429, "step": 15350 }, { "epoch": 1.99, "learning_rate": 4.407425784084487e-06, "loss": 0.2389, "step": 15360 }, { "epoch": 1.99, "learning_rate": 4.406948739772414e-06, "loss": 0.2366, "step": 15370 }, { "epoch": 2.0, "learning_rate": 4.406471315398526e-06, "loss": 0.2372, "step": 15380 }, { "epoch": 2.0, "learning_rate": 4.4059935110507675e-06, "loss": 0.2472, "step": 15390 }, { "epoch": 2.0, "learning_rate": 4.405515326817155e-06, "loss": 0.2405, "step": 15400 }, { "epoch": 2.0, "learning_rate": 4.405036762785777e-06, "loss": 0.2269, "step": 15410 }, { "epoch": 2.0, "eval_loss": 0.23726505041122437, "eval_runtime": 59.0951, "eval_samples_per_second": 184.059, "eval_steps_per_second": 23.014, "step": 15410 }, { "epoch": 2.0, "learning_rate": 4.4045578190447866e-06, "loss": 0.2392, "step": 15420 }, { "epoch": 2.0, "learning_rate": 4.404078495682413e-06, "loss": 0.2389, "step": 15430 }, { "epoch": 2.0, "learning_rate": 4.4035987927869516e-06, "loss": 0.2232, "step": 15440 }, { "epoch": 2.01, "learning_rate": 4.403118710446766e-06, "loss": 0.2368, "step": 15450 }, { "epoch": 2.01, "learning_rate": 4.402638248750296e-06, "loss": 0.237, "step": 15460 }, { "epoch": 2.01, "learning_rate": 4.402157407786044e-06, "loss": 0.2526, "step": 15470 }, { "epoch": 2.01, "learning_rate": 4.401676187642589e-06, "loss": 0.2276, "step": 15480 }, { "epoch": 2.01, "learning_rate": 4.401194588408573e-06, "loss": 0.2359, "step": 15490 }, { "epoch": 2.01, "learning_rate": 4.400712610172714e-06, "loss": 0.2393, "step": 15500 }, { "epoch": 2.01, "learning_rate": 4.400230253023797e-06, "loss": 0.2466, "step": 15510 }, { "epoch": 2.01, "learning_rate": 4.399747517050675e-06, "loss": 0.2305, "step": 15520 }, { "epoch": 2.02, "learning_rate": 4.399264402342276e-06, "loss": 0.2157, "step": 15530 }, { "epoch": 2.02, "learning_rate": 4.398780908987591e-06, "loss": 0.2301, "step": 15540 }, { "epoch": 2.02, "learning_rate": 4.398297037075687e-06, "loss": 0.234, "step": 15550 }, { "epoch": 2.02, "learning_rate": 4.397812786695697e-06, "loss": 0.2264, "step": 15560 }, { "epoch": 2.02, "learning_rate": 4.397328157936825e-06, "loss": 0.2194, "step": 15570 }, { "epoch": 2.02, "learning_rate": 4.3968431508883445e-06, "loss": 0.2188, "step": 15580 }, { "epoch": 2.02, "learning_rate": 4.396357765639598e-06, "loss": 0.241, "step": 15590 }, { "epoch": 2.02, "learning_rate": 4.395872002279999e-06, "loss": 0.242, "step": 15600 }, { "epoch": 2.03, "learning_rate": 4.395385860899029e-06, "loss": 0.2545, "step": 15610 }, { "epoch": 2.03, "learning_rate": 4.394899341586241e-06, "loss": 0.2401, "step": 15620 }, { "epoch": 2.03, "learning_rate": 4.394412444431257e-06, "loss": 0.2295, "step": 15630 }, { "epoch": 2.03, "learning_rate": 4.3939251695237684e-06, "loss": 0.2378, "step": 15640 }, { "epoch": 2.03, "learning_rate": 4.393437516953535e-06, "loss": 0.2227, "step": 15650 }, { "epoch": 2.03, "learning_rate": 4.392949486810388e-06, "loss": 0.2276, "step": 15660 }, { "epoch": 2.03, "learning_rate": 4.392461079184226e-06, "loss": 0.2314, "step": 15670 }, { "epoch": 2.04, "learning_rate": 4.391972294165022e-06, "loss": 0.2301, "step": 15680 }, { "epoch": 2.04, "learning_rate": 4.3914831318428115e-06, "loss": 0.2313, "step": 15690 }, { "epoch": 2.04, "learning_rate": 4.390993592307705e-06, "loss": 0.2287, "step": 15700 }, { "epoch": 2.04, "learning_rate": 4.39050367564988e-06, "loss": 0.2456, "step": 15710 }, { "epoch": 2.04, "learning_rate": 4.390013381959585e-06, "loss": 0.2403, "step": 15720 }, { "epoch": 2.04, "learning_rate": 4.389522711327134e-06, "loss": 0.2564, "step": 15730 }, { "epoch": 2.04, "learning_rate": 4.389031663842917e-06, "loss": 0.2461, "step": 15740 }, { "epoch": 2.04, "learning_rate": 4.388540239597388e-06, "loss": 0.2278, "step": 15750 }, { "epoch": 2.05, "learning_rate": 4.388048438681073e-06, "loss": 0.2424, "step": 15760 }, { "epoch": 2.05, "learning_rate": 4.3875562611845666e-06, "loss": 0.2348, "step": 15770 }, { "epoch": 2.05, "learning_rate": 4.3870637071985315e-06, "loss": 0.2334, "step": 15780 }, { "epoch": 2.05, "learning_rate": 4.3865707768137026e-06, "loss": 0.234, "step": 15790 }, { "epoch": 2.05, "learning_rate": 4.386077470120882e-06, "loss": 0.2436, "step": 15800 }, { "epoch": 2.05, "learning_rate": 4.385583787210942e-06, "loss": 0.2299, "step": 15810 }, { "epoch": 2.05, "learning_rate": 4.385089728174823e-06, "loss": 0.2338, "step": 15820 }, { "epoch": 2.05, "learning_rate": 4.384595293103537e-06, "loss": 0.2439, "step": 15830 }, { "epoch": 2.06, "learning_rate": 4.384100482088163e-06, "loss": 0.2269, "step": 15840 }, { "epoch": 2.06, "learning_rate": 4.383605295219851e-06, "loss": 0.2375, "step": 15850 }, { "epoch": 2.06, "learning_rate": 4.383109732589818e-06, "loss": 0.2276, "step": 15860 }, { "epoch": 2.06, "learning_rate": 4.382613794289352e-06, "loss": 0.2434, "step": 15870 }, { "epoch": 2.06, "learning_rate": 4.3821174804098105e-06, "loss": 0.2426, "step": 15880 }, { "epoch": 2.06, "learning_rate": 4.381620791042619e-06, "loss": 0.2213, "step": 15890 }, { "epoch": 2.06, "learning_rate": 4.381123726279273e-06, "loss": 0.2439, "step": 15900 }, { "epoch": 2.06, "learning_rate": 4.380626286211336e-06, "loss": 0.2306, "step": 15910 }, { "epoch": 2.07, "learning_rate": 4.380128470930442e-06, "loss": 0.2275, "step": 15920 }, { "epoch": 2.07, "learning_rate": 4.379630280528294e-06, "loss": 0.2361, "step": 15930 }, { "epoch": 2.07, "learning_rate": 4.379131715096663e-06, "loss": 0.2351, "step": 15940 }, { "epoch": 2.07, "learning_rate": 4.378632774727388e-06, "loss": 0.2461, "step": 15950 }, { "epoch": 2.07, "learning_rate": 4.378133459512383e-06, "loss": 0.2381, "step": 15960 }, { "epoch": 2.07, "learning_rate": 4.3776337695436234e-06, "loss": 0.2251, "step": 15970 }, { "epoch": 2.07, "learning_rate": 4.377133704913158e-06, "loss": 0.241, "step": 15980 }, { "epoch": 2.08, "learning_rate": 4.376633265713105e-06, "loss": 0.2253, "step": 15990 }, { "epoch": 2.08, "learning_rate": 4.376132452035648e-06, "loss": 0.2281, "step": 16000 }, { "epoch": 2.08, "learning_rate": 4.375631263973043e-06, "loss": 0.2315, "step": 16010 }, { "epoch": 2.08, "learning_rate": 4.375129701617615e-06, "loss": 0.2486, "step": 16020 }, { "epoch": 2.08, "learning_rate": 4.374627765061755e-06, "loss": 0.2365, "step": 16030 }, { "epoch": 2.08, "learning_rate": 4.3741254543979255e-06, "loss": 0.2309, "step": 16040 }, { "epoch": 2.08, "learning_rate": 4.373622769718657e-06, "loss": 0.233, "step": 16050 }, { "epoch": 2.08, "learning_rate": 4.37311971111655e-06, "loss": 0.2378, "step": 16060 }, { "epoch": 2.09, "learning_rate": 4.372616278684271e-06, "loss": 0.2503, "step": 16070 }, { "epoch": 2.09, "learning_rate": 4.372112472514558e-06, "loss": 0.2275, "step": 16080 }, { "epoch": 2.09, "learning_rate": 4.371608292700218e-06, "loss": 0.2447, "step": 16090 }, { "epoch": 2.09, "learning_rate": 4.371103739334125e-06, "loss": 0.2492, "step": 16100 }, { "epoch": 2.09, "learning_rate": 4.370598812509224e-06, "loss": 0.2503, "step": 16110 }, { "epoch": 2.09, "learning_rate": 4.370093512318526e-06, "loss": 0.2284, "step": 16120 }, { "epoch": 2.09, "learning_rate": 4.369587838855113e-06, "loss": 0.2436, "step": 16130 }, { "epoch": 2.09, "learning_rate": 4.369081792212136e-06, "loss": 0.2249, "step": 16140 }, { "epoch": 2.1, "learning_rate": 4.368575372482812e-06, "loss": 0.2324, "step": 16150 }, { "epoch": 2.1, "learning_rate": 4.368068579760431e-06, "loss": 0.24, "step": 16160 }, { "epoch": 2.1, "learning_rate": 4.367561414138346e-06, "loss": 0.2269, "step": 16170 }, { "epoch": 2.1, "learning_rate": 4.367053875709985e-06, "loss": 0.2303, "step": 16180 }, { "epoch": 2.1, "learning_rate": 4.366545964568841e-06, "loss": 0.2499, "step": 16190 }, { "epoch": 2.1, "learning_rate": 4.366037680808475e-06, "loss": 0.2245, "step": 16200 }, { "epoch": 2.1, "learning_rate": 4.365529024522519e-06, "loss": 0.2396, "step": 16210 }, { "epoch": 2.11, "learning_rate": 4.3650199958046715e-06, "loss": 0.223, "step": 16220 }, { "epoch": 2.11, "learning_rate": 4.364510594748702e-06, "loss": 0.2356, "step": 16230 }, { "epoch": 2.11, "learning_rate": 4.364000821448448e-06, "loss": 0.2291, "step": 16240 }, { "epoch": 2.11, "learning_rate": 4.3634906759978116e-06, "loss": 0.2372, "step": 16250 }, { "epoch": 2.11, "learning_rate": 4.36298015849077e-06, "loss": 0.2371, "step": 16260 }, { "epoch": 2.11, "learning_rate": 4.362469269021363e-06, "loss": 0.2367, "step": 16270 }, { "epoch": 2.11, "learning_rate": 4.361958007683704e-06, "loss": 0.2592, "step": 16280 }, { "epoch": 2.11, "learning_rate": 4.361446374571971e-06, "loss": 0.2423, "step": 16290 }, { "epoch": 2.12, "learning_rate": 4.360934369780413e-06, "loss": 0.236, "step": 16300 }, { "epoch": 2.12, "learning_rate": 4.360421993403345e-06, "loss": 0.2287, "step": 16310 }, { "epoch": 2.12, "learning_rate": 4.359909245535152e-06, "loss": 0.2356, "step": 16320 }, { "epoch": 2.12, "learning_rate": 4.359396126270288e-06, "loss": 0.2448, "step": 16330 }, { "epoch": 2.12, "learning_rate": 4.358882635703274e-06, "loss": 0.2342, "step": 16340 }, { "epoch": 2.12, "learning_rate": 4.358368773928701e-06, "loss": 0.2465, "step": 16350 }, { "epoch": 2.12, "learning_rate": 4.357854541041227e-06, "loss": 0.2461, "step": 16360 }, { "epoch": 2.12, "learning_rate": 4.357339937135579e-06, "loss": 0.2386, "step": 16370 }, { "epoch": 2.13, "learning_rate": 4.3568249623065515e-06, "loss": 0.2421, "step": 16380 }, { "epoch": 2.13, "learning_rate": 4.356309616649009e-06, "loss": 0.2535, "step": 16390 }, { "epoch": 2.13, "learning_rate": 4.355793900257883e-06, "loss": 0.2425, "step": 16400 }, { "epoch": 2.13, "learning_rate": 4.355277813228172e-06, "loss": 0.2218, "step": 16410 }, { "epoch": 2.13, "learning_rate": 4.354761355654947e-06, "loss": 0.2234, "step": 16420 }, { "epoch": 2.13, "learning_rate": 4.3542445276333435e-06, "loss": 0.2429, "step": 16430 }, { "epoch": 2.13, "learning_rate": 4.353727329258566e-06, "loss": 0.2301, "step": 16440 }, { "epoch": 2.13, "learning_rate": 4.353209760625889e-06, "loss": 0.2434, "step": 16450 }, { "epoch": 2.14, "learning_rate": 4.352691821830651e-06, "loss": 0.2243, "step": 16460 }, { "epoch": 2.14, "learning_rate": 4.352173512968265e-06, "loss": 0.2314, "step": 16470 }, { "epoch": 2.14, "learning_rate": 4.351654834134206e-06, "loss": 0.237, "step": 16480 }, { "epoch": 2.14, "learning_rate": 4.351135785424021e-06, "loss": 0.2323, "step": 16490 }, { "epoch": 2.14, "learning_rate": 4.350616366933323e-06, "loss": 0.2458, "step": 16500 }, { "epoch": 2.14, "learning_rate": 4.350096578757796e-06, "loss": 0.2422, "step": 16510 }, { "epoch": 2.14, "learning_rate": 4.349576420993189e-06, "loss": 0.2366, "step": 16520 }, { "epoch": 2.15, "learning_rate": 4.349055893735319e-06, "loss": 0.2415, "step": 16530 }, { "epoch": 2.15, "learning_rate": 4.348534997080075e-06, "loss": 0.2241, "step": 16540 }, { "epoch": 2.15, "learning_rate": 4.348013731123408e-06, "loss": 0.2261, "step": 16550 }, { "epoch": 2.15, "learning_rate": 4.3474920959613435e-06, "loss": 0.2409, "step": 16560 }, { "epoch": 2.15, "learning_rate": 4.346970091689971e-06, "loss": 0.2358, "step": 16570 }, { "epoch": 2.15, "learning_rate": 4.3464477184054485e-06, "loss": 0.2278, "step": 16580 }, { "epoch": 2.15, "learning_rate": 4.345924976204001e-06, "loss": 0.2325, "step": 16590 }, { "epoch": 2.15, "learning_rate": 4.345401865181926e-06, "loss": 0.2461, "step": 16600 }, { "epoch": 2.16, "learning_rate": 4.344878385435584e-06, "loss": 0.2383, "step": 16610 }, { "epoch": 2.16, "learning_rate": 4.344354537061406e-06, "loss": 0.229, "step": 16620 }, { "epoch": 2.16, "learning_rate": 4.343830320155888e-06, "loss": 0.245, "step": 16630 }, { "epoch": 2.16, "learning_rate": 4.343305734815597e-06, "loss": 0.2364, "step": 16640 }, { "epoch": 2.16, "learning_rate": 4.342780781137168e-06, "loss": 0.2384, "step": 16650 }, { "epoch": 2.16, "learning_rate": 4.342255459217302e-06, "loss": 0.2375, "step": 16660 }, { "epoch": 2.16, "learning_rate": 4.3417297691527685e-06, "loss": 0.2376, "step": 16670 }, { "epoch": 2.16, "learning_rate": 4.341203711040406e-06, "loss": 0.2322, "step": 16680 }, { "epoch": 2.17, "learning_rate": 4.340677284977116e-06, "loss": 0.2579, "step": 16690 }, { "epoch": 2.17, "learning_rate": 4.3401504910598755e-06, "loss": 0.244, "step": 16700 }, { "epoch": 2.17, "learning_rate": 4.339623329385724e-06, "loss": 0.2221, "step": 16710 }, { "epoch": 2.17, "learning_rate": 4.3390958000517695e-06, "loss": 0.2298, "step": 16720 }, { "epoch": 2.17, "learning_rate": 4.338567903155188e-06, "loss": 0.2371, "step": 16730 }, { "epoch": 2.17, "learning_rate": 4.338039638793223e-06, "loss": 0.2298, "step": 16740 }, { "epoch": 2.17, "learning_rate": 4.337511007063188e-06, "loss": 0.2314, "step": 16750 }, { "epoch": 2.18, "learning_rate": 4.3369820080624595e-06, "loss": 0.2358, "step": 16760 }, { "epoch": 2.18, "learning_rate": 4.336452641888487e-06, "loss": 0.2401, "step": 16770 }, { "epoch": 2.18, "learning_rate": 4.335922908638784e-06, "loss": 0.2257, "step": 16780 }, { "epoch": 2.18, "learning_rate": 4.3353928084109315e-06, "loss": 0.2315, "step": 16790 }, { "epoch": 2.18, "learning_rate": 4.334862341302582e-06, "loss": 0.2342, "step": 16800 }, { "epoch": 2.18, "learning_rate": 4.33433150741145e-06, "loss": 0.2307, "step": 16810 }, { "epoch": 2.18, "learning_rate": 4.333800306835322e-06, "loss": 0.2256, "step": 16820 }, { "epoch": 2.18, "learning_rate": 4.333268739672049e-06, "loss": 0.2362, "step": 16830 }, { "epoch": 2.19, "learning_rate": 4.332736806019553e-06, "loss": 0.2329, "step": 16840 }, { "epoch": 2.19, "learning_rate": 4.3322045059758204e-06, "loss": 0.2245, "step": 16850 }, { "epoch": 2.19, "learning_rate": 4.331671839638906e-06, "loss": 0.2386, "step": 16860 }, { "epoch": 2.19, "learning_rate": 4.3311388071069334e-06, "loss": 0.2409, "step": 16870 }, { "epoch": 2.19, "learning_rate": 4.330605408478091e-06, "loss": 0.2293, "step": 16880 }, { "epoch": 2.19, "learning_rate": 4.330071643850637e-06, "loss": 0.24, "step": 16890 }, { "epoch": 2.19, "learning_rate": 4.3295375133228956e-06, "loss": 0.2491, "step": 16900 }, { "epoch": 2.19, "learning_rate": 4.32900301699326e-06, "loss": 0.2344, "step": 16910 }, { "epoch": 2.2, "learning_rate": 4.3284681549601885e-06, "loss": 0.2412, "step": 16920 }, { "epoch": 2.2, "learning_rate": 4.327932927322209e-06, "loss": 0.2418, "step": 16930 }, { "epoch": 2.2, "learning_rate": 4.327397334177915e-06, "loss": 0.2302, "step": 16940 }, { "epoch": 2.2, "learning_rate": 4.326861375625969e-06, "loss": 0.2323, "step": 16950 }, { "epoch": 2.2, "learning_rate": 4.326325051765099e-06, "loss": 0.2453, "step": 16960 }, { "epoch": 2.2, "learning_rate": 4.325788362694103e-06, "loss": 0.2458, "step": 16970 }, { "epoch": 2.2, "learning_rate": 4.325251308511842e-06, "loss": 0.2479, "step": 16980 }, { "epoch": 2.21, "learning_rate": 4.324713889317248e-06, "loss": 0.2279, "step": 16990 }, { "epoch": 2.21, "learning_rate": 4.324176105209319e-06, "loss": 0.2463, "step": 17000 }, { "epoch": 2.21, "learning_rate": 4.323637956287121e-06, "loss": 0.2423, "step": 17010 }, { "epoch": 2.21, "learning_rate": 4.323099442649784e-06, "loss": 0.2364, "step": 17020 }, { "epoch": 2.21, "learning_rate": 4.32256056439651e-06, "loss": 0.239, "step": 17030 }, { "epoch": 2.21, "learning_rate": 4.322021321626565e-06, "loss": 0.2347, "step": 17040 }, { "epoch": 2.21, "learning_rate": 4.321481714439282e-06, "loss": 0.2351, "step": 17050 }, { "epoch": 2.21, "learning_rate": 4.3209417429340635e-06, "loss": 0.2443, "step": 17060 }, { "epoch": 2.22, "learning_rate": 4.320401407210377e-06, "loss": 0.2387, "step": 17070 }, { "epoch": 2.22, "learning_rate": 4.319860707367757e-06, "loss": 0.2414, "step": 17080 }, { "epoch": 2.22, "learning_rate": 4.319319643505807e-06, "loss": 0.2336, "step": 17090 }, { "epoch": 2.22, "learning_rate": 4.318778215724195e-06, "loss": 0.2332, "step": 17100 }, { "epoch": 2.22, "learning_rate": 4.318236424122659e-06, "loss": 0.225, "step": 17110 }, { "epoch": 2.22, "learning_rate": 4.317694268801002e-06, "loss": 0.2291, "step": 17120 }, { "epoch": 2.22, "learning_rate": 4.317151749859094e-06, "loss": 0.2462, "step": 17130 }, { "epoch": 2.22, "learning_rate": 4.316608867396872e-06, "loss": 0.2559, "step": 17140 }, { "epoch": 2.23, "learning_rate": 4.316065621514341e-06, "loss": 0.2323, "step": 17150 }, { "epoch": 2.23, "learning_rate": 4.315522012311572e-06, "loss": 0.2331, "step": 17160 }, { "epoch": 2.23, "learning_rate": 4.314978039888704e-06, "loss": 0.247, "step": 17170 }, { "epoch": 2.23, "learning_rate": 4.314433704345942e-06, "loss": 0.2216, "step": 17180 }, { "epoch": 2.23, "learning_rate": 4.3138890057835564e-06, "loss": 0.2295, "step": 17190 }, { "epoch": 2.23, "learning_rate": 4.3133439443018885e-06, "loss": 0.233, "step": 17200 }, { "epoch": 2.23, "learning_rate": 4.312798520001342e-06, "loss": 0.2366, "step": 17210 }, { "epoch": 2.23, "learning_rate": 4.312252732982391e-06, "loss": 0.2386, "step": 17220 }, { "epoch": 2.24, "learning_rate": 4.311706583345575e-06, "loss": 0.2424, "step": 17230 }, { "epoch": 2.24, "learning_rate": 4.311160071191498e-06, "loss": 0.2447, "step": 17240 }, { "epoch": 2.24, "learning_rate": 4.310613196620836e-06, "loss": 0.2339, "step": 17250 }, { "epoch": 2.24, "learning_rate": 4.310065959734326e-06, "loss": 0.2323, "step": 17260 }, { "epoch": 2.24, "learning_rate": 4.309518360632777e-06, "loss": 0.2474, "step": 17270 }, { "epoch": 2.24, "learning_rate": 4.30897039941706e-06, "loss": 0.2464, "step": 17280 }, { "epoch": 2.24, "learning_rate": 4.308422076188117e-06, "loss": 0.2403, "step": 17290 }, { "epoch": 2.25, "learning_rate": 4.307873391046953e-06, "loss": 0.2204, "step": 17300 }, { "epoch": 2.25, "learning_rate": 4.307324344094642e-06, "loss": 0.2277, "step": 17310 }, { "epoch": 2.25, "learning_rate": 4.306774935432324e-06, "loss": 0.2295, "step": 17320 }, { "epoch": 2.25, "learning_rate": 4.3062251651612065e-06, "loss": 0.2466, "step": 17330 }, { "epoch": 2.25, "learning_rate": 4.30567503338256e-06, "loss": 0.2474, "step": 17340 }, { "epoch": 2.25, "learning_rate": 4.3051245401977265e-06, "loss": 0.2326, "step": 17350 }, { "epoch": 2.25, "learning_rate": 4.3045736857081115e-06, "loss": 0.2382, "step": 17360 }, { "epoch": 2.25, "learning_rate": 4.3040224700151875e-06, "loss": 0.2335, "step": 17370 }, { "epoch": 2.26, "learning_rate": 4.303470893220495e-06, "loss": 0.2388, "step": 17380 }, { "epoch": 2.26, "learning_rate": 4.302918955425639e-06, "loss": 0.2316, "step": 17390 }, { "epoch": 2.26, "learning_rate": 4.302366656732293e-06, "loss": 0.2395, "step": 17400 }, { "epoch": 2.26, "learning_rate": 4.301813997242195e-06, "loss": 0.237, "step": 17410 }, { "epoch": 2.26, "learning_rate": 4.301260977057151e-06, "loss": 0.2403, "step": 17420 }, { "epoch": 2.26, "learning_rate": 4.300707596279033e-06, "loss": 0.2377, "step": 17430 }, { "epoch": 2.26, "learning_rate": 4.300153855009778e-06, "loss": 0.2144, "step": 17440 }, { "epoch": 2.26, "learning_rate": 4.299599753351392e-06, "loss": 0.2454, "step": 17450 }, { "epoch": 2.27, "learning_rate": 4.2990452914059455e-06, "loss": 0.2409, "step": 17460 }, { "epoch": 2.27, "learning_rate": 4.298490469275577e-06, "loss": 0.2332, "step": 17470 }, { "epoch": 2.27, "learning_rate": 4.297935287062488e-06, "loss": 0.2299, "step": 17480 }, { "epoch": 2.27, "learning_rate": 4.29737974486895e-06, "loss": 0.2329, "step": 17490 }, { "epoch": 2.27, "learning_rate": 4.2968238427973e-06, "loss": 0.2284, "step": 17500 }, { "epoch": 2.27, "learning_rate": 4.296267580949939e-06, "loss": 0.2452, "step": 17510 }, { "epoch": 2.27, "learning_rate": 4.295710959429338e-06, "loss": 0.2377, "step": 17520 }, { "epoch": 2.28, "learning_rate": 4.295153978338031e-06, "loss": 0.24, "step": 17530 }, { "epoch": 2.28, "learning_rate": 4.29459663777862e-06, "loss": 0.24, "step": 17540 }, { "epoch": 2.28, "learning_rate": 4.294038937853772e-06, "loss": 0.2443, "step": 17550 }, { "epoch": 2.28, "learning_rate": 4.293480878666221e-06, "loss": 0.2363, "step": 17560 }, { "epoch": 2.28, "learning_rate": 4.292922460318769e-06, "loss": 0.2398, "step": 17570 }, { "epoch": 2.28, "learning_rate": 4.292363682914279e-06, "loss": 0.2511, "step": 17580 }, { "epoch": 2.28, "learning_rate": 4.291804546555685e-06, "loss": 0.2419, "step": 17590 }, { "epoch": 2.28, "learning_rate": 4.291245051345987e-06, "loss": 0.2393, "step": 17600 }, { "epoch": 2.29, "learning_rate": 4.2906851973882475e-06, "loss": 0.2395, "step": 17610 }, { "epoch": 2.29, "learning_rate": 4.2901249847855974e-06, "loss": 0.2459, "step": 17620 }, { "epoch": 2.29, "learning_rate": 4.2895644136412335e-06, "loss": 0.2367, "step": 17630 }, { "epoch": 2.29, "learning_rate": 4.28900348405842e-06, "loss": 0.245, "step": 17640 }, { "epoch": 2.29, "learning_rate": 4.288442196140484e-06, "loss": 0.2471, "step": 17650 }, { "epoch": 2.29, "learning_rate": 4.287880549990822e-06, "loss": 0.2403, "step": 17660 }, { "epoch": 2.29, "learning_rate": 4.287318545712894e-06, "loss": 0.2363, "step": 17670 }, { "epoch": 2.29, "learning_rate": 4.286756183410226e-06, "loss": 0.2228, "step": 17680 }, { "epoch": 2.3, "learning_rate": 4.286193463186412e-06, "loss": 0.2461, "step": 17690 }, { "epoch": 2.3, "learning_rate": 4.285630385145109e-06, "loss": 0.2249, "step": 17700 }, { "epoch": 2.3, "learning_rate": 4.2850669493900436e-06, "loss": 0.2428, "step": 17710 }, { "epoch": 2.3, "learning_rate": 4.284503156025005e-06, "loss": 0.2436, "step": 17720 }, { "epoch": 2.3, "learning_rate": 4.283939005153851e-06, "loss": 0.2399, "step": 17730 }, { "epoch": 2.3, "learning_rate": 4.283374496880502e-06, "loss": 0.237, "step": 17740 }, { "epoch": 2.3, "learning_rate": 4.2828096313089474e-06, "loss": 0.2387, "step": 17750 }, { "epoch": 2.3, "learning_rate": 4.28224440854324e-06, "loss": 0.2437, "step": 17760 }, { "epoch": 2.31, "learning_rate": 4.2816788286875e-06, "loss": 0.242, "step": 17770 }, { "epoch": 2.31, "learning_rate": 4.281112891845913e-06, "loss": 0.2503, "step": 17780 }, { "epoch": 2.31, "learning_rate": 4.280546598122731e-06, "loss": 0.2393, "step": 17790 }, { "epoch": 2.31, "learning_rate": 4.279979947622269e-06, "loss": 0.2271, "step": 17800 }, { "epoch": 2.31, "learning_rate": 4.279412940448911e-06, "loss": 0.2374, "step": 17810 }, { "epoch": 2.31, "learning_rate": 4.278845576707106e-06, "loss": 0.2305, "step": 17820 }, { "epoch": 2.31, "learning_rate": 4.278277856501367e-06, "loss": 0.2316, "step": 17830 }, { "epoch": 2.32, "learning_rate": 4.2777097799362735e-06, "loss": 0.2383, "step": 17840 }, { "epoch": 2.32, "learning_rate": 4.277141347116471e-06, "loss": 0.2448, "step": 17850 }, { "epoch": 2.32, "learning_rate": 4.276572558146671e-06, "loss": 0.2413, "step": 17860 }, { "epoch": 2.32, "learning_rate": 4.276003413131651e-06, "loss": 0.2098, "step": 17870 }, { "epoch": 2.32, "learning_rate": 4.275433912176251e-06, "loss": 0.2358, "step": 17880 }, { "epoch": 2.32, "learning_rate": 4.274864055385381e-06, "loss": 0.2341, "step": 17890 }, { "epoch": 2.32, "learning_rate": 4.274293842864012e-06, "loss": 0.2355, "step": 17900 }, { "epoch": 2.32, "learning_rate": 4.273723274717184e-06, "loss": 0.2211, "step": 17910 }, { "epoch": 2.33, "learning_rate": 4.2731523510500025e-06, "loss": 0.2338, "step": 17920 }, { "epoch": 2.33, "learning_rate": 4.272581071967635e-06, "loss": 0.2338, "step": 17930 }, { "epoch": 2.33, "learning_rate": 4.272009437575319e-06, "loss": 0.2434, "step": 17940 }, { "epoch": 2.33, "learning_rate": 4.271437447978355e-06, "loss": 0.2395, "step": 17950 }, { "epoch": 2.33, "learning_rate": 4.270865103282108e-06, "loss": 0.2378, "step": 17960 }, { "epoch": 2.33, "learning_rate": 4.27029240359201e-06, "loss": 0.2278, "step": 17970 }, { "epoch": 2.33, "learning_rate": 4.269719349013557e-06, "loss": 0.2445, "step": 17980 }, { "epoch": 2.33, "learning_rate": 4.269145939652315e-06, "loss": 0.2386, "step": 17990 }, { "epoch": 2.34, "learning_rate": 4.268572175613907e-06, "loss": 0.2271, "step": 18000 }, { "epoch": 2.34, "learning_rate": 4.267998057004029e-06, "loss": 0.2258, "step": 18010 }, { "epoch": 2.34, "learning_rate": 4.267423583928439e-06, "loss": 0.2402, "step": 18020 }, { "epoch": 2.34, "learning_rate": 4.266848756492961e-06, "loss": 0.2244, "step": 18030 }, { "epoch": 2.34, "learning_rate": 4.266273574803482e-06, "loss": 0.2313, "step": 18040 }, { "epoch": 2.34, "learning_rate": 4.2656980389659584e-06, "loss": 0.243, "step": 18050 }, { "epoch": 2.34, "learning_rate": 4.265122149086408e-06, "loss": 0.2372, "step": 18060 }, { "epoch": 2.35, "learning_rate": 4.264545905270918e-06, "loss": 0.2435, "step": 18070 }, { "epoch": 2.35, "learning_rate": 4.263969307625635e-06, "loss": 0.2355, "step": 18080 }, { "epoch": 2.35, "learning_rate": 4.263392356256776e-06, "loss": 0.2378, "step": 18090 }, { "epoch": 2.35, "learning_rate": 4.26281505127062e-06, "loss": 0.2356, "step": 18100 }, { "epoch": 2.35, "learning_rate": 4.2622373927735144e-06, "loss": 0.2414, "step": 18110 }, { "epoch": 2.35, "learning_rate": 4.261659380871869e-06, "loss": 0.2155, "step": 18120 }, { "epoch": 2.35, "learning_rate": 4.261081015672156e-06, "loss": 0.2307, "step": 18130 }, { "epoch": 2.35, "learning_rate": 4.260502297280921e-06, "loss": 0.2314, "step": 18140 }, { "epoch": 2.36, "learning_rate": 4.259923225804767e-06, "loss": 0.2292, "step": 18150 }, { "epoch": 2.36, "learning_rate": 4.259343801350365e-06, "loss": 0.2471, "step": 18160 }, { "epoch": 2.36, "learning_rate": 4.258764024024452e-06, "loss": 0.2472, "step": 18170 }, { "epoch": 2.36, "learning_rate": 4.258183893933828e-06, "loss": 0.2338, "step": 18180 }, { "epoch": 2.36, "learning_rate": 4.2576034111853565e-06, "loss": 0.2309, "step": 18190 }, { "epoch": 2.36, "learning_rate": 4.257022575885972e-06, "loss": 0.2316, "step": 18200 }, { "epoch": 2.36, "learning_rate": 4.256441388142668e-06, "loss": 0.2438, "step": 18210 }, { "epoch": 2.36, "learning_rate": 4.255859848062505e-06, "loss": 0.2221, "step": 18220 }, { "epoch": 2.37, "learning_rate": 4.25527795575261e-06, "loss": 0.2367, "step": 18230 }, { "epoch": 2.37, "learning_rate": 4.254695711320172e-06, "loss": 0.2154, "step": 18240 }, { "epoch": 2.37, "learning_rate": 4.254113114872447e-06, "loss": 0.2241, "step": 18250 }, { "epoch": 2.37, "learning_rate": 4.253530166516754e-06, "loss": 0.2393, "step": 18260 }, { "epoch": 2.37, "learning_rate": 4.252946866360479e-06, "loss": 0.2278, "step": 18270 }, { "epoch": 2.37, "learning_rate": 4.252363214511072e-06, "loss": 0.2533, "step": 18280 }, { "epoch": 2.37, "learning_rate": 4.2517792110760455e-06, "loss": 0.2401, "step": 18290 }, { "epoch": 2.38, "learning_rate": 4.251194856162981e-06, "loss": 0.2259, "step": 18300 }, { "epoch": 2.38, "learning_rate": 4.250610149879522e-06, "loss": 0.2497, "step": 18310 }, { "epoch": 2.38, "learning_rate": 4.250025092333377e-06, "loss": 0.2283, "step": 18320 }, { "epoch": 2.38, "learning_rate": 4.249439683632318e-06, "loss": 0.2348, "step": 18330 }, { "epoch": 2.38, "learning_rate": 4.2488539238841854e-06, "loss": 0.2361, "step": 18340 }, { "epoch": 2.38, "learning_rate": 4.248267813196881e-06, "loss": 0.2287, "step": 18350 }, { "epoch": 2.38, "learning_rate": 4.2476813516783715e-06, "loss": 0.2255, "step": 18360 }, { "epoch": 2.38, "learning_rate": 4.2470945394366905e-06, "loss": 0.2275, "step": 18370 }, { "epoch": 2.39, "learning_rate": 4.246507376579934e-06, "loss": 0.2293, "step": 18380 }, { "epoch": 2.39, "learning_rate": 4.245919863216264e-06, "loss": 0.2314, "step": 18390 }, { "epoch": 2.39, "learning_rate": 4.245331999453904e-06, "loss": 0.2413, "step": 18400 }, { "epoch": 2.39, "learning_rate": 4.2447437854011464e-06, "loss": 0.2294, "step": 18410 }, { "epoch": 2.39, "learning_rate": 4.2441552211663455e-06, "loss": 0.2485, "step": 18420 }, { "epoch": 2.39, "learning_rate": 4.243566306857921e-06, "loss": 0.2227, "step": 18430 }, { "epoch": 2.39, "learning_rate": 4.242977042584357e-06, "loss": 0.2416, "step": 18440 }, { "epoch": 2.39, "learning_rate": 4.242387428454201e-06, "loss": 0.2445, "step": 18450 }, { "epoch": 2.4, "learning_rate": 4.241797464576066e-06, "loss": 0.2401, "step": 18460 }, { "epoch": 2.4, "learning_rate": 4.241207151058631e-06, "loss": 0.2378, "step": 18470 }, { "epoch": 2.4, "learning_rate": 4.240616488010635e-06, "loss": 0.2464, "step": 18480 }, { "epoch": 2.4, "learning_rate": 4.240025475540885e-06, "loss": 0.2278, "step": 18490 }, { "epoch": 2.4, "learning_rate": 4.239434113758252e-06, "loss": 0.2477, "step": 18500 }, { "epoch": 2.4, "learning_rate": 4.238842402771671e-06, "loss": 0.2328, "step": 18510 }, { "epoch": 2.4, "learning_rate": 4.238250342690139e-06, "loss": 0.2355, "step": 18520 }, { "epoch": 2.4, "learning_rate": 4.2376579336227215e-06, "loss": 0.2432, "step": 18530 }, { "epoch": 2.41, "learning_rate": 4.237065175678546e-06, "loss": 0.2348, "step": 18540 }, { "epoch": 2.41, "learning_rate": 4.236472068966803e-06, "loss": 0.2375, "step": 18550 }, { "epoch": 2.41, "learning_rate": 4.235878613596751e-06, "loss": 0.237, "step": 18560 }, { "epoch": 2.41, "learning_rate": 4.235284809677708e-06, "loss": 0.2242, "step": 18570 }, { "epoch": 2.41, "learning_rate": 4.234690657319059e-06, "loss": 0.231, "step": 18580 }, { "epoch": 2.41, "learning_rate": 4.234096156630255e-06, "loss": 0.2383, "step": 18590 }, { "epoch": 2.41, "learning_rate": 4.233501307720806e-06, "loss": 0.2397, "step": 18600 }, { "epoch": 2.42, "learning_rate": 4.232906110700291e-06, "loss": 0.2305, "step": 18610 }, { "epoch": 2.42, "learning_rate": 4.232310565678351e-06, "loss": 0.2192, "step": 18620 }, { "epoch": 2.42, "learning_rate": 4.2317146727646915e-06, "loss": 0.2322, "step": 18630 }, { "epoch": 2.42, "learning_rate": 4.231118432069081e-06, "loss": 0.2276, "step": 18640 }, { "epoch": 2.42, "learning_rate": 4.230521843701353e-06, "loss": 0.2343, "step": 18650 }, { "epoch": 2.42, "learning_rate": 4.229924907771405e-06, "loss": 0.2381, "step": 18660 }, { "epoch": 2.42, "learning_rate": 4.2293276243892e-06, "loss": 0.2312, "step": 18670 }, { "epoch": 2.42, "learning_rate": 4.228729993664764e-06, "loss": 0.2435, "step": 18680 }, { "epoch": 2.43, "learning_rate": 4.228132015708183e-06, "loss": 0.2345, "step": 18690 }, { "epoch": 2.43, "learning_rate": 4.227533690629614e-06, "loss": 0.2245, "step": 18700 }, { "epoch": 2.43, "learning_rate": 4.226935018539273e-06, "loss": 0.2333, "step": 18710 }, { "epoch": 2.43, "learning_rate": 4.226335999547441e-06, "loss": 0.2338, "step": 18720 }, { "epoch": 2.43, "learning_rate": 4.2257366337644644e-06, "loss": 0.2316, "step": 18730 }, { "epoch": 2.43, "learning_rate": 4.225136921300751e-06, "loss": 0.2505, "step": 18740 }, { "epoch": 2.43, "learning_rate": 4.224536862266776e-06, "loss": 0.247, "step": 18750 }, { "epoch": 2.43, "learning_rate": 4.223936456773073e-06, "loss": 0.2333, "step": 18760 }, { "epoch": 2.44, "learning_rate": 4.223335704930246e-06, "loss": 0.2489, "step": 18770 }, { "epoch": 2.44, "learning_rate": 4.222734606848958e-06, "loss": 0.2353, "step": 18780 }, { "epoch": 2.44, "learning_rate": 4.222133162639936e-06, "loss": 0.2412, "step": 18790 }, { "epoch": 2.44, "learning_rate": 4.221531372413975e-06, "loss": 0.2364, "step": 18800 }, { "epoch": 2.44, "learning_rate": 4.2209292362819284e-06, "loss": 0.2364, "step": 18810 }, { "epoch": 2.44, "learning_rate": 4.2203267543547166e-06, "loss": 0.2295, "step": 18820 }, { "epoch": 2.44, "learning_rate": 4.219723926743323e-06, "loss": 0.2387, "step": 18830 }, { "epoch": 2.45, "learning_rate": 4.219120753558794e-06, "loss": 0.2373, "step": 18840 }, { "epoch": 2.45, "learning_rate": 4.21851723491224e-06, "loss": 0.2265, "step": 18850 }, { "epoch": 2.45, "learning_rate": 4.2179133709148344e-06, "loss": 0.2522, "step": 18860 }, { "epoch": 2.45, "learning_rate": 4.217309161677817e-06, "loss": 0.2491, "step": 18870 }, { "epoch": 2.45, "learning_rate": 4.2167046073124875e-06, "loss": 0.2376, "step": 18880 }, { "epoch": 2.45, "learning_rate": 4.2160997079302125e-06, "loss": 0.2294, "step": 18890 }, { "epoch": 2.45, "learning_rate": 4.215494463642419e-06, "loss": 0.2203, "step": 18900 }, { "epoch": 2.45, "learning_rate": 4.2148888745606e-06, "loss": 0.2381, "step": 18910 }, { "epoch": 2.46, "learning_rate": 4.214282940796309e-06, "loss": 0.2394, "step": 18920 }, { "epoch": 2.46, "learning_rate": 4.213676662461168e-06, "loss": 0.2418, "step": 18930 }, { "epoch": 2.46, "learning_rate": 4.213070039666858e-06, "loss": 0.2445, "step": 18940 }, { "epoch": 2.46, "learning_rate": 4.2124630725251255e-06, "loss": 0.2344, "step": 18950 }, { "epoch": 2.46, "learning_rate": 4.211855761147778e-06, "loss": 0.2376, "step": 18960 }, { "epoch": 2.46, "learning_rate": 4.211248105646692e-06, "loss": 0.2353, "step": 18970 }, { "epoch": 2.46, "learning_rate": 4.2106401061338e-06, "loss": 0.2291, "step": 18980 }, { "epoch": 2.46, "learning_rate": 4.210031762721104e-06, "loss": 0.2475, "step": 18990 }, { "epoch": 2.47, "learning_rate": 4.209423075520666e-06, "loss": 0.2168, "step": 19000 }, { "epoch": 2.47, "learning_rate": 4.2088140446446115e-06, "loss": 0.235, "step": 19010 }, { "epoch": 2.47, "learning_rate": 4.208204670205132e-06, "loss": 0.2261, "step": 19020 }, { "epoch": 2.47, "learning_rate": 4.207594952314479e-06, "loss": 0.2525, "step": 19030 }, { "epoch": 2.47, "learning_rate": 4.2069848910849685e-06, "loss": 0.2415, "step": 19040 }, { "epoch": 2.47, "learning_rate": 4.2063744866289805e-06, "loss": 0.2307, "step": 19050 }, { "epoch": 2.47, "learning_rate": 4.205763739058958e-06, "loss": 0.229, "step": 19060 }, { "epoch": 2.48, "learning_rate": 4.205152648487406e-06, "loss": 0.2524, "step": 19070 }, { "epoch": 2.48, "learning_rate": 4.204541215026893e-06, "loss": 0.2265, "step": 19080 }, { "epoch": 2.48, "learning_rate": 4.203929438790052e-06, "loss": 0.22, "step": 19090 }, { "epoch": 2.48, "learning_rate": 4.203317319889579e-06, "loss": 0.2436, "step": 19100 }, { "epoch": 2.48, "learning_rate": 4.20270485843823e-06, "loss": 0.2334, "step": 19110 }, { "epoch": 2.48, "learning_rate": 4.20209205454883e-06, "loss": 0.239, "step": 19120 }, { "epoch": 2.48, "learning_rate": 4.201478908334261e-06, "loss": 0.2355, "step": 19130 }, { "epoch": 2.48, "learning_rate": 4.20086541990747e-06, "loss": 0.2368, "step": 19140 }, { "epoch": 2.49, "learning_rate": 4.200251589381471e-06, "loss": 0.2397, "step": 19150 }, { "epoch": 2.49, "learning_rate": 4.199637416869335e-06, "loss": 0.2356, "step": 19160 }, { "epoch": 2.49, "learning_rate": 4.199022902484199e-06, "loss": 0.2441, "step": 19170 }, { "epoch": 2.49, "learning_rate": 4.198408046339263e-06, "loss": 0.2313, "step": 19180 }, { "epoch": 2.49, "learning_rate": 4.19779284854779e-06, "loss": 0.2257, "step": 19190 }, { "epoch": 2.49, "learning_rate": 4.197177309223105e-06, "loss": 0.2322, "step": 19200 }, { "epoch": 2.49, "learning_rate": 4.196561428478597e-06, "loss": 0.237, "step": 19210 }, { "epoch": 2.49, "learning_rate": 4.195945206427718e-06, "loss": 0.2338, "step": 19220 }, { "epoch": 2.5, "learning_rate": 4.195328643183981e-06, "loss": 0.2215, "step": 19230 }, { "epoch": 2.5, "learning_rate": 4.194711738860964e-06, "loss": 0.2229, "step": 19240 }, { "epoch": 2.5, "learning_rate": 4.194094493572306e-06, "loss": 0.2278, "step": 19250 }, { "epoch": 2.5, "learning_rate": 4.19347690743171e-06, "loss": 0.2303, "step": 19260 }, { "epoch": 2.5, "learning_rate": 4.192858980552942e-06, "loss": 0.2363, "step": 19270 }, { "epoch": 2.5, "learning_rate": 4.19224071304983e-06, "loss": 0.2313, "step": 19280 }, { "epoch": 2.5, "learning_rate": 4.191622105036266e-06, "loss": 0.2329, "step": 19290 }, { "epoch": 2.5, "learning_rate": 4.191003156626202e-06, "loss": 0.2313, "step": 19300 }, { "epoch": 2.51, "learning_rate": 4.190383867933655e-06, "loss": 0.2515, "step": 19310 }, { "epoch": 2.51, "learning_rate": 4.1897642390727064e-06, "loss": 0.2287, "step": 19320 }, { "epoch": 2.51, "learning_rate": 4.189144270157496e-06, "loss": 0.223, "step": 19330 }, { "epoch": 2.51, "learning_rate": 4.188523961302226e-06, "loss": 0.2338, "step": 19340 }, { "epoch": 2.51, "learning_rate": 4.1879033126211686e-06, "loss": 0.2329, "step": 19350 }, { "epoch": 2.51, "learning_rate": 4.18728232422865e-06, "loss": 0.2464, "step": 19360 }, { "epoch": 2.51, "learning_rate": 4.186660996239063e-06, "loss": 0.2291, "step": 19370 }, { "epoch": 2.52, "learning_rate": 4.186039328766863e-06, "loss": 0.2376, "step": 19380 }, { "epoch": 2.52, "learning_rate": 4.185417321926566e-06, "loss": 0.2372, "step": 19390 }, { "epoch": 2.52, "learning_rate": 4.184794975832755e-06, "loss": 0.2389, "step": 19400 }, { "epoch": 2.52, "learning_rate": 4.184172290600069e-06, "loss": 0.2276, "step": 19410 }, { "epoch": 2.52, "learning_rate": 4.183549266343215e-06, "loss": 0.2211, "step": 19420 }, { "epoch": 2.52, "learning_rate": 4.182925903176959e-06, "loss": 0.235, "step": 19430 }, { "epoch": 2.52, "learning_rate": 4.18230220121613e-06, "loss": 0.2354, "step": 19440 }, { "epoch": 2.52, "learning_rate": 4.181678160575623e-06, "loss": 0.236, "step": 19450 }, { "epoch": 2.53, "learning_rate": 4.18105378137039e-06, "loss": 0.2419, "step": 19460 }, { "epoch": 2.53, "learning_rate": 4.180429063715449e-06, "loss": 0.2489, "step": 19470 }, { "epoch": 2.53, "learning_rate": 4.1798040077258785e-06, "loss": 0.2375, "step": 19480 }, { "epoch": 2.53, "learning_rate": 4.179178613516821e-06, "loss": 0.2269, "step": 19490 }, { "epoch": 2.53, "learning_rate": 4.178552881203481e-06, "loss": 0.2382, "step": 19500 }, { "epoch": 2.53, "learning_rate": 4.177926810901122e-06, "loss": 0.2343, "step": 19510 }, { "epoch": 2.53, "learning_rate": 4.177300402725074e-06, "loss": 0.2326, "step": 19520 }, { "epoch": 2.53, "learning_rate": 4.176673656790728e-06, "loss": 0.2315, "step": 19530 }, { "epoch": 2.54, "learning_rate": 4.176046573213536e-06, "loss": 0.2226, "step": 19540 }, { "epoch": 2.54, "learning_rate": 4.175419152109013e-06, "loss": 0.2309, "step": 19550 }, { "epoch": 2.54, "learning_rate": 4.174791393592737e-06, "loss": 0.2296, "step": 19560 }, { "epoch": 2.54, "learning_rate": 4.1741632977803485e-06, "loss": 0.24, "step": 19570 }, { "epoch": 2.54, "learning_rate": 4.173534864787546e-06, "loss": 0.2278, "step": 19580 }, { "epoch": 2.54, "learning_rate": 4.172906094730096e-06, "loss": 0.2343, "step": 19590 }, { "epoch": 2.54, "learning_rate": 4.172276987723822e-06, "loss": 0.2295, "step": 19600 }, { "epoch": 2.55, "learning_rate": 4.171647543884613e-06, "loss": 0.2407, "step": 19610 }, { "epoch": 2.55, "learning_rate": 4.171017763328419e-06, "loss": 0.2266, "step": 19620 }, { "epoch": 2.55, "learning_rate": 4.170387646171251e-06, "loss": 0.2326, "step": 19630 }, { "epoch": 2.55, "learning_rate": 4.169757192529184e-06, "loss": 0.2258, "step": 19640 }, { "epoch": 2.55, "learning_rate": 4.169126402518352e-06, "loss": 0.2382, "step": 19650 }, { "epoch": 2.55, "learning_rate": 4.168495276254955e-06, "loss": 0.2351, "step": 19660 }, { "epoch": 2.55, "learning_rate": 4.167863813855251e-06, "loss": 0.2342, "step": 19670 }, { "epoch": 2.55, "learning_rate": 4.167232015435563e-06, "loss": 0.219, "step": 19680 }, { "epoch": 2.56, "learning_rate": 4.166599881112275e-06, "loss": 0.2312, "step": 19690 }, { "epoch": 2.56, "learning_rate": 4.1659674110018305e-06, "loss": 0.2364, "step": 19700 }, { "epoch": 2.56, "learning_rate": 4.165334605220738e-06, "loss": 0.2322, "step": 19710 }, { "epoch": 2.56, "learning_rate": 4.164701463885566e-06, "loss": 0.235, "step": 19720 }, { "epoch": 2.56, "learning_rate": 4.164067987112947e-06, "loss": 0.255, "step": 19730 }, { "epoch": 2.56, "learning_rate": 4.163434175019571e-06, "loss": 0.2334, "step": 19740 }, { "epoch": 2.56, "learning_rate": 4.162800027722196e-06, "loss": 0.2289, "step": 19750 }, { "epoch": 2.56, "learning_rate": 4.162165545337636e-06, "loss": 0.2308, "step": 19760 }, { "epoch": 2.57, "learning_rate": 4.161530727982769e-06, "loss": 0.2266, "step": 19770 }, { "epoch": 2.57, "learning_rate": 4.160895575774535e-06, "loss": 0.2293, "step": 19780 }, { "epoch": 2.57, "learning_rate": 4.160260088829936e-06, "loss": 0.2375, "step": 19790 }, { "epoch": 2.57, "learning_rate": 4.159624267266035e-06, "loss": 0.2421, "step": 19800 }, { "epoch": 2.57, "learning_rate": 4.158988111199956e-06, "loss": 0.2311, "step": 19810 }, { "epoch": 2.57, "learning_rate": 4.158351620748885e-06, "loss": 0.2245, "step": 19820 }, { "epoch": 2.57, "learning_rate": 4.15771479603007e-06, "loss": 0.2347, "step": 19830 }, { "epoch": 2.57, "learning_rate": 4.157077637160822e-06, "loss": 0.2269, "step": 19840 }, { "epoch": 2.58, "learning_rate": 4.1564401442585116e-06, "loss": 0.24, "step": 19850 }, { "epoch": 2.58, "learning_rate": 4.155802317440569e-06, "loss": 0.233, "step": 19860 }, { "epoch": 2.58, "learning_rate": 4.155164156824491e-06, "loss": 0.2327, "step": 19870 }, { "epoch": 2.58, "learning_rate": 4.154525662527832e-06, "loss": 0.2471, "step": 19880 }, { "epoch": 2.58, "learning_rate": 4.153886834668209e-06, "loss": 0.2445, "step": 19890 }, { "epoch": 2.58, "learning_rate": 4.153247673363301e-06, "loss": 0.2357, "step": 19900 }, { "epoch": 2.58, "learning_rate": 4.152608178730848e-06, "loss": 0.2279, "step": 19910 }, { "epoch": 2.59, "learning_rate": 4.151968350888649e-06, "loss": 0.2266, "step": 19920 }, { "epoch": 2.59, "learning_rate": 4.151328189954571e-06, "loss": 0.2388, "step": 19930 }, { "epoch": 2.59, "learning_rate": 4.150687696046534e-06, "loss": 0.2345, "step": 19940 }, { "epoch": 2.59, "learning_rate": 4.150046869282524e-06, "loss": 0.2395, "step": 19950 }, { "epoch": 2.59, "learning_rate": 4.149405709780591e-06, "loss": 0.2432, "step": 19960 }, { "epoch": 2.59, "learning_rate": 4.148764217658839e-06, "loss": 0.2329, "step": 19970 }, { "epoch": 2.59, "learning_rate": 4.148122393035439e-06, "loss": 0.2484, "step": 19980 }, { "epoch": 2.59, "learning_rate": 4.147480236028622e-06, "loss": 0.241, "step": 19990 }, { "epoch": 2.6, "learning_rate": 4.146837746756679e-06, "loss": 0.2242, "step": 20000 }, { "epoch": 2.6, "learning_rate": 4.146194925337963e-06, "loss": 0.2394, "step": 20010 }, { "epoch": 2.6, "learning_rate": 4.145551771890888e-06, "loss": 0.2395, "step": 20020 }, { "epoch": 2.6, "learning_rate": 4.1449082865339294e-06, "loss": 0.2151, "step": 20030 }, { "epoch": 2.6, "learning_rate": 4.1442644693856245e-06, "loss": 0.2287, "step": 20040 }, { "epoch": 2.6, "learning_rate": 4.143620320564569e-06, "loss": 0.2401, "step": 20050 }, { "epoch": 2.6, "learning_rate": 4.142975840189423e-06, "loss": 0.2422, "step": 20060 }, { "epoch": 2.6, "learning_rate": 4.142331028378906e-06, "loss": 0.2283, "step": 20070 }, { "epoch": 2.61, "learning_rate": 4.141685885251798e-06, "loss": 0.2215, "step": 20080 }, { "epoch": 2.61, "learning_rate": 4.141040410926942e-06, "loss": 0.2427, "step": 20090 }, { "epoch": 2.61, "learning_rate": 4.14039460552324e-06, "loss": 0.2166, "step": 20100 }, { "epoch": 2.61, "learning_rate": 4.139748469159656e-06, "loss": 0.2322, "step": 20110 }, { "epoch": 2.61, "learning_rate": 4.139102001955215e-06, "loss": 0.2335, "step": 20120 }, { "epoch": 2.61, "learning_rate": 4.138455204029003e-06, "loss": 0.222, "step": 20130 }, { "epoch": 2.61, "learning_rate": 4.1378080755001655e-06, "loss": 0.2371, "step": 20140 }, { "epoch": 2.62, "learning_rate": 4.137160616487911e-06, "loss": 0.2245, "step": 20150 }, { "epoch": 2.62, "learning_rate": 4.136512827111507e-06, "loss": 0.2357, "step": 20160 }, { "epoch": 2.62, "learning_rate": 4.1358647074902835e-06, "loss": 0.24, "step": 20170 }, { "epoch": 2.62, "learning_rate": 4.135216257743631e-06, "loss": 0.2348, "step": 20180 }, { "epoch": 2.62, "learning_rate": 4.134567477991e-06, "loss": 0.2414, "step": 20190 }, { "epoch": 2.62, "learning_rate": 4.133918368351902e-06, "loss": 0.2386, "step": 20200 }, { "epoch": 2.62, "learning_rate": 4.133268928945909e-06, "loss": 0.2184, "step": 20210 }, { "epoch": 2.62, "learning_rate": 4.132619159892655e-06, "loss": 0.2311, "step": 20220 }, { "epoch": 2.63, "learning_rate": 4.131969061311834e-06, "loss": 0.2314, "step": 20230 }, { "epoch": 2.63, "learning_rate": 4.1313186333232005e-06, "loss": 0.224, "step": 20240 }, { "epoch": 2.63, "learning_rate": 4.13066787604657e-06, "loss": 0.2448, "step": 20250 }, { "epoch": 2.63, "learning_rate": 4.130016789601818e-06, "loss": 0.2478, "step": 20260 }, { "epoch": 2.63, "learning_rate": 4.129365374108881e-06, "loss": 0.2312, "step": 20270 }, { "epoch": 2.63, "learning_rate": 4.128713629687757e-06, "loss": 0.2416, "step": 20280 }, { "epoch": 2.63, "learning_rate": 4.128061556458503e-06, "loss": 0.228, "step": 20290 }, { "epoch": 2.63, "learning_rate": 4.127409154541239e-06, "loss": 0.2249, "step": 20300 }, { "epoch": 2.64, "learning_rate": 4.126756424056142e-06, "loss": 0.2347, "step": 20310 }, { "epoch": 2.64, "learning_rate": 4.126103365123453e-06, "loss": 0.2163, "step": 20320 }, { "epoch": 2.64, "learning_rate": 4.12544997786347e-06, "loss": 0.2252, "step": 20330 }, { "epoch": 2.64, "learning_rate": 4.1247962623965555e-06, "loss": 0.2343, "step": 20340 }, { "epoch": 2.64, "learning_rate": 4.1241422188431295e-06, "loss": 0.2403, "step": 20350 }, { "epoch": 2.64, "learning_rate": 4.123487847323673e-06, "loss": 0.2352, "step": 20360 }, { "epoch": 2.64, "learning_rate": 4.1228331479587284e-06, "loss": 0.2347, "step": 20370 }, { "epoch": 2.65, "learning_rate": 4.122178120868899e-06, "loss": 0.2167, "step": 20380 }, { "epoch": 2.65, "learning_rate": 4.121522766174845e-06, "loss": 0.2373, "step": 20390 }, { "epoch": 2.65, "learning_rate": 4.120867083997291e-06, "loss": 0.2382, "step": 20400 }, { "epoch": 2.65, "learning_rate": 4.1202110744570184e-06, "loss": 0.2237, "step": 20410 }, { "epoch": 2.65, "learning_rate": 4.119554737674873e-06, "loss": 0.2292, "step": 20420 }, { "epoch": 2.65, "learning_rate": 4.118898073771756e-06, "loss": 0.2362, "step": 20430 }, { "epoch": 2.65, "learning_rate": 4.118241082868634e-06, "loss": 0.2297, "step": 20440 }, { "epoch": 2.65, "learning_rate": 4.117583765086529e-06, "loss": 0.2429, "step": 20450 }, { "epoch": 2.66, "learning_rate": 4.116926120546528e-06, "loss": 0.2432, "step": 20460 }, { "epoch": 2.66, "learning_rate": 4.116268149369775e-06, "loss": 0.2424, "step": 20470 }, { "epoch": 2.66, "learning_rate": 4.1156098516774725e-06, "loss": 0.2309, "step": 20480 }, { "epoch": 2.66, "learning_rate": 4.114951227590888e-06, "loss": 0.2224, "step": 20490 }, { "epoch": 2.66, "learning_rate": 4.114292277231346e-06, "loss": 0.2358, "step": 20500 }, { "epoch": 2.66, "learning_rate": 4.1136330007202314e-06, "loss": 0.2319, "step": 20510 }, { "epoch": 2.66, "learning_rate": 4.1129733981789905e-06, "loss": 0.227, "step": 20520 }, { "epoch": 2.66, "learning_rate": 4.112313469729127e-06, "loss": 0.2293, "step": 20530 }, { "epoch": 2.67, "learning_rate": 4.111653215492208e-06, "loss": 0.2357, "step": 20540 }, { "epoch": 2.67, "learning_rate": 4.1109926355898566e-06, "loss": 0.2275, "step": 20550 }, { "epoch": 2.67, "learning_rate": 4.110331730143762e-06, "loss": 0.231, "step": 20560 }, { "epoch": 2.67, "learning_rate": 4.109670499275666e-06, "loss": 0.2369, "step": 20570 }, { "epoch": 2.67, "learning_rate": 4.109008943107375e-06, "loss": 0.2496, "step": 20580 }, { "epoch": 2.67, "learning_rate": 4.1083470617607555e-06, "loss": 0.2313, "step": 20590 }, { "epoch": 2.67, "learning_rate": 4.107684855357731e-06, "loss": 0.2252, "step": 20600 }, { "epoch": 2.67, "learning_rate": 4.107022324020287e-06, "loss": 0.2325, "step": 20610 }, { "epoch": 2.68, "learning_rate": 4.106359467870469e-06, "loss": 0.218, "step": 20620 }, { "epoch": 2.68, "learning_rate": 4.105696287030382e-06, "loss": 0.2281, "step": 20630 }, { "epoch": 2.68, "learning_rate": 4.105032781622189e-06, "loss": 0.2205, "step": 20640 }, { "epoch": 2.68, "learning_rate": 4.1043689517681155e-06, "loss": 0.2314, "step": 20650 }, { "epoch": 2.68, "learning_rate": 4.103704797590445e-06, "loss": 0.2357, "step": 20660 }, { "epoch": 2.68, "learning_rate": 4.1030403192115225e-06, "loss": 0.2367, "step": 20670 }, { "epoch": 2.68, "learning_rate": 4.10237551675375e-06, "loss": 0.2315, "step": 20680 }, { "epoch": 2.69, "learning_rate": 4.101710390339591e-06, "loss": 0.2408, "step": 20690 }, { "epoch": 2.69, "learning_rate": 4.1010449400915695e-06, "loss": 0.2369, "step": 20700 }, { "epoch": 2.69, "learning_rate": 4.100379166132267e-06, "loss": 0.2472, "step": 20710 }, { "epoch": 2.69, "learning_rate": 4.099713068584327e-06, "loss": 0.2305, "step": 20720 }, { "epoch": 2.69, "learning_rate": 4.09904664757045e-06, "loss": 0.2414, "step": 20730 }, { "epoch": 2.69, "learning_rate": 4.098379903213397e-06, "loss": 0.221, "step": 20740 }, { "epoch": 2.69, "learning_rate": 4.097712835635991e-06, "loss": 0.2478, "step": 20750 }, { "epoch": 2.69, "learning_rate": 4.097045444961111e-06, "loss": 0.2288, "step": 20760 }, { "epoch": 2.7, "learning_rate": 4.096377731311698e-06, "loss": 0.2367, "step": 20770 }, { "epoch": 2.7, "learning_rate": 4.0957096948107505e-06, "loss": 0.2312, "step": 20780 }, { "epoch": 2.7, "learning_rate": 4.0950413355813275e-06, "loss": 0.2419, "step": 20790 }, { "epoch": 2.7, "learning_rate": 4.0943726537465485e-06, "loss": 0.2233, "step": 20800 }, { "epoch": 2.7, "learning_rate": 4.093703649429589e-06, "loss": 0.2357, "step": 20810 }, { "epoch": 2.7, "learning_rate": 4.09303432275369e-06, "loss": 0.2345, "step": 20820 }, { "epoch": 2.7, "learning_rate": 4.092364673842146e-06, "loss": 0.2324, "step": 20830 }, { "epoch": 2.7, "learning_rate": 4.091694702818312e-06, "loss": 0.2479, "step": 20840 }, { "epoch": 2.71, "learning_rate": 4.0910244098056065e-06, "loss": 0.2491, "step": 20850 }, { "epoch": 2.71, "learning_rate": 4.0903537949275e-06, "loss": 0.2369, "step": 20860 }, { "epoch": 2.71, "learning_rate": 4.08968285830753e-06, "loss": 0.2231, "step": 20870 }, { "epoch": 2.71, "learning_rate": 4.089011600069288e-06, "loss": 0.233, "step": 20880 }, { "epoch": 2.71, "learning_rate": 4.088340020336427e-06, "loss": 0.2501, "step": 20890 }, { "epoch": 2.71, "learning_rate": 4.087668119232659e-06, "loss": 0.2308, "step": 20900 }, { "epoch": 2.71, "learning_rate": 4.086995896881754e-06, "loss": 0.2437, "step": 20910 }, { "epoch": 2.72, "learning_rate": 4.086323353407542e-06, "loss": 0.2308, "step": 20920 }, { "epoch": 2.72, "learning_rate": 4.085650488933913e-06, "loss": 0.2388, "step": 20930 }, { "epoch": 2.72, "learning_rate": 4.0849773035848155e-06, "loss": 0.24, "step": 20940 }, { "epoch": 2.72, "learning_rate": 4.0843037974842565e-06, "loss": 0.2166, "step": 20950 }, { "epoch": 2.72, "learning_rate": 4.083629970756302e-06, "loss": 0.2427, "step": 20960 }, { "epoch": 2.72, "learning_rate": 4.082955823525079e-06, "loss": 0.2395, "step": 20970 }, { "epoch": 2.72, "learning_rate": 4.082281355914771e-06, "loss": 0.2185, "step": 20980 }, { "epoch": 2.72, "learning_rate": 4.081606568049623e-06, "loss": 0.2472, "step": 20990 }, { "epoch": 2.73, "learning_rate": 4.080931460053936e-06, "loss": 0.2364, "step": 21000 }, { "epoch": 2.73, "learning_rate": 4.080256032052072e-06, "loss": 0.2386, "step": 21010 }, { "epoch": 2.73, "learning_rate": 4.079580284168453e-06, "loss": 0.2221, "step": 21020 }, { "epoch": 2.73, "learning_rate": 4.078904216527557e-06, "loss": 0.2363, "step": 21030 }, { "epoch": 2.73, "learning_rate": 4.078227829253922e-06, "loss": 0.2275, "step": 21040 }, { "epoch": 2.73, "learning_rate": 4.077551122472147e-06, "loss": 0.2463, "step": 21050 }, { "epoch": 2.73, "learning_rate": 4.076874096306888e-06, "loss": 0.2204, "step": 21060 }, { "epoch": 2.73, "learning_rate": 4.076196750882858e-06, "loss": 0.2268, "step": 21070 }, { "epoch": 2.74, "learning_rate": 4.075519086324831e-06, "loss": 0.2314, "step": 21080 }, { "epoch": 2.74, "learning_rate": 4.074841102757643e-06, "loss": 0.2202, "step": 21090 }, { "epoch": 2.74, "learning_rate": 4.074162800306182e-06, "loss": 0.2336, "step": 21100 }, { "epoch": 2.74, "learning_rate": 4.0734841790954e-06, "loss": 0.2392, "step": 21110 }, { "epoch": 2.74, "learning_rate": 4.0728052392503036e-06, "loss": 0.2295, "step": 21120 }, { "epoch": 2.74, "learning_rate": 4.072125980895963e-06, "loss": 0.2544, "step": 21130 }, { "epoch": 2.74, "learning_rate": 4.071446404157502e-06, "loss": 0.2187, "step": 21140 }, { "epoch": 2.74, "learning_rate": 4.070766509160107e-06, "loss": 0.2202, "step": 21150 }, { "epoch": 2.75, "learning_rate": 4.07008629602902e-06, "loss": 0.2255, "step": 21160 }, { "epoch": 2.75, "learning_rate": 4.069405764889544e-06, "loss": 0.2267, "step": 21170 }, { "epoch": 2.75, "learning_rate": 4.068724915867039e-06, "loss": 0.2552, "step": 21180 }, { "epoch": 2.75, "learning_rate": 4.068043749086927e-06, "loss": 0.2363, "step": 21190 }, { "epoch": 2.75, "learning_rate": 4.067362264674683e-06, "loss": 0.2181, "step": 21200 }, { "epoch": 2.75, "learning_rate": 4.066680462755841e-06, "loss": 0.2333, "step": 21210 }, { "epoch": 2.75, "learning_rate": 4.065998343456001e-06, "loss": 0.2319, "step": 21220 }, { "epoch": 2.76, "learning_rate": 4.065315906900813e-06, "loss": 0.2271, "step": 21230 }, { "epoch": 2.76, "learning_rate": 4.064633153215989e-06, "loss": 0.2323, "step": 21240 }, { "epoch": 2.76, "learning_rate": 4.063950082527299e-06, "loss": 0.2468, "step": 21250 }, { "epoch": 2.76, "learning_rate": 4.063266694960572e-06, "loss": 0.2219, "step": 21260 }, { "epoch": 2.76, "learning_rate": 4.062582990641695e-06, "loss": 0.2467, "step": 21270 }, { "epoch": 2.76, "learning_rate": 4.061898969696612e-06, "loss": 0.2298, "step": 21280 }, { "epoch": 2.76, "learning_rate": 4.061214632251327e-06, "loss": 0.2426, "step": 21290 }, { "epoch": 2.76, "learning_rate": 4.0605299784319026e-06, "loss": 0.2409, "step": 21300 }, { "epoch": 2.77, "learning_rate": 4.059845008364458e-06, "loss": 0.2439, "step": 21310 }, { "epoch": 2.77, "learning_rate": 4.059159722175172e-06, "loss": 0.2436, "step": 21320 }, { "epoch": 2.77, "learning_rate": 4.058474119990281e-06, "loss": 0.2183, "step": 21330 }, { "epoch": 2.77, "learning_rate": 4.05778820193608e-06, "loss": 0.2342, "step": 21340 }, { "epoch": 2.77, "learning_rate": 4.057101968138921e-06, "loss": 0.2519, "step": 21350 }, { "epoch": 2.77, "learning_rate": 4.056415418725216e-06, "loss": 0.2404, "step": 21360 }, { "epoch": 2.77, "learning_rate": 4.0557285538214345e-06, "loss": 0.2376, "step": 21370 }, { "epoch": 2.77, "learning_rate": 4.055041373554103e-06, "loss": 0.2274, "step": 21380 }, { "epoch": 2.78, "learning_rate": 4.0543538780498075e-06, "loss": 0.2308, "step": 21390 }, { "epoch": 2.78, "learning_rate": 4.053666067435191e-06, "loss": 0.2327, "step": 21400 }, { "epoch": 2.78, "learning_rate": 4.052977941836956e-06, "loss": 0.2276, "step": 21410 }, { "epoch": 2.78, "learning_rate": 4.052289501381862e-06, "loss": 0.2323, "step": 21420 }, { "epoch": 2.78, "learning_rate": 4.051600746196725e-06, "loss": 0.2362, "step": 21430 }, { "epoch": 2.78, "learning_rate": 4.050911676408423e-06, "loss": 0.2246, "step": 21440 }, { "epoch": 2.78, "learning_rate": 4.050222292143888e-06, "loss": 0.2241, "step": 21450 }, { "epoch": 2.79, "learning_rate": 4.049532593530112e-06, "loss": 0.2278, "step": 21460 }, { "epoch": 2.79, "learning_rate": 4.048842580694144e-06, "loss": 0.221, "step": 21470 }, { "epoch": 2.79, "learning_rate": 4.048152253763091e-06, "loss": 0.23, "step": 21480 }, { "epoch": 2.79, "learning_rate": 4.04746161286412e-06, "loss": 0.2207, "step": 21490 }, { "epoch": 2.79, "learning_rate": 4.046770658124451e-06, "loss": 0.2211, "step": 21500 }, { "epoch": 2.79, "learning_rate": 4.046079389671367e-06, "loss": 0.2516, "step": 21510 }, { "epoch": 2.79, "learning_rate": 4.045387807632206e-06, "loss": 0.2264, "step": 21520 }, { "epoch": 2.79, "learning_rate": 4.044695912134364e-06, "loss": 0.241, "step": 21530 }, { "epoch": 2.8, "learning_rate": 4.044003703305295e-06, "loss": 0.2268, "step": 21540 }, { "epoch": 2.8, "learning_rate": 4.043311181272511e-06, "loss": 0.2298, "step": 21550 }, { "epoch": 2.8, "learning_rate": 4.042618346163582e-06, "loss": 0.2502, "step": 21560 }, { "epoch": 2.8, "learning_rate": 4.0419251981061344e-06, "loss": 0.2191, "step": 21570 }, { "epoch": 2.8, "learning_rate": 4.041231737227853e-06, "loss": 0.2267, "step": 21580 }, { "epoch": 2.8, "learning_rate": 4.0405379636564805e-06, "loss": 0.2468, "step": 21590 }, { "epoch": 2.8, "learning_rate": 4.039843877519817e-06, "loss": 0.2296, "step": 21600 }, { "epoch": 2.8, "learning_rate": 4.03914947894572e-06, "loss": 0.2332, "step": 21610 }, { "epoch": 2.81, "learning_rate": 4.038454768062104e-06, "loss": 0.2348, "step": 21620 }, { "epoch": 2.81, "learning_rate": 4.037759744996944e-06, "loss": 0.2339, "step": 21630 }, { "epoch": 2.81, "learning_rate": 4.037064409878266e-06, "loss": 0.227, "step": 21640 }, { "epoch": 2.81, "learning_rate": 4.036368762834162e-06, "loss": 0.2411, "step": 21650 }, { "epoch": 2.81, "learning_rate": 4.035672803992775e-06, "loss": 0.2406, "step": 21660 }, { "epoch": 2.81, "learning_rate": 4.034976533482308e-06, "loss": 0.2434, "step": 21670 }, { "epoch": 2.81, "learning_rate": 4.034279951431021e-06, "loss": 0.2401, "step": 21680 }, { "epoch": 2.82, "learning_rate": 4.03358305796723e-06, "loss": 0.2377, "step": 21690 }, { "epoch": 2.82, "learning_rate": 4.032885853219312e-06, "loss": 0.2567, "step": 21700 }, { "epoch": 2.82, "learning_rate": 4.032188337315698e-06, "loss": 0.2299, "step": 21710 }, { "epoch": 2.82, "learning_rate": 4.0314905103848765e-06, "loss": 0.223, "step": 21720 }, { "epoch": 2.82, "learning_rate": 4.030792372555396e-06, "loss": 0.2315, "step": 21730 }, { "epoch": 2.82, "learning_rate": 4.0300939239558596e-06, "loss": 0.2237, "step": 21740 }, { "epoch": 2.82, "learning_rate": 4.0293951647149286e-06, "loss": 0.2333, "step": 21750 }, { "epoch": 2.82, "learning_rate": 4.028696094961321e-06, "loss": 0.2311, "step": 21760 }, { "epoch": 2.83, "learning_rate": 4.027996714823813e-06, "loss": 0.23, "step": 21770 }, { "epoch": 2.83, "learning_rate": 4.027297024431238e-06, "loss": 0.2374, "step": 21780 }, { "epoch": 2.83, "learning_rate": 4.026597023912484e-06, "loss": 0.2344, "step": 21790 }, { "epoch": 2.83, "learning_rate": 4.025896713396499e-06, "loss": 0.2281, "step": 21800 }, { "epoch": 2.83, "learning_rate": 4.025196093012287e-06, "loss": 0.2228, "step": 21810 }, { "epoch": 2.83, "learning_rate": 4.0244951628889105e-06, "loss": 0.2336, "step": 21820 }, { "epoch": 2.83, "learning_rate": 4.023793923155488e-06, "loss": 0.2404, "step": 21830 }, { "epoch": 2.83, "learning_rate": 4.023092373941192e-06, "loss": 0.2385, "step": 21840 }, { "epoch": 2.84, "learning_rate": 4.022390515375258e-06, "loss": 0.2245, "step": 21850 }, { "epoch": 2.84, "learning_rate": 4.021688347586974e-06, "loss": 0.2469, "step": 21860 }, { "epoch": 2.84, "learning_rate": 4.020985870705686e-06, "loss": 0.2461, "step": 21870 }, { "epoch": 2.84, "learning_rate": 4.0202830848607976e-06, "loss": 0.23, "step": 21880 }, { "epoch": 2.84, "learning_rate": 4.01957999018177e-06, "loss": 0.2167, "step": 21890 }, { "epoch": 2.84, "learning_rate": 4.018876586798118e-06, "loss": 0.2162, "step": 21900 }, { "epoch": 2.84, "learning_rate": 4.0181728748394185e-06, "loss": 0.2301, "step": 21910 }, { "epoch": 2.84, "learning_rate": 4.0174688544352995e-06, "loss": 0.2386, "step": 21920 }, { "epoch": 2.85, "learning_rate": 4.016764525715449e-06, "loss": 0.2142, "step": 21930 }, { "epoch": 2.85, "learning_rate": 4.016059888809613e-06, "loss": 0.2292, "step": 21940 }, { "epoch": 2.85, "learning_rate": 4.015354943847592e-06, "loss": 0.2303, "step": 21950 }, { "epoch": 2.85, "learning_rate": 4.014649690959243e-06, "loss": 0.2274, "step": 21960 }, { "epoch": 2.85, "learning_rate": 4.0139441302744815e-06, "loss": 0.253, "step": 21970 }, { "epoch": 2.85, "learning_rate": 4.013238261923279e-06, "loss": 0.2358, "step": 21980 }, { "epoch": 2.85, "learning_rate": 4.0125320860356625e-06, "loss": 0.225, "step": 21990 }, { "epoch": 2.86, "learning_rate": 4.0118256027417186e-06, "loss": 0.229, "step": 22000 }, { "epoch": 2.86, "learning_rate": 4.0111188121715865e-06, "loss": 0.2393, "step": 22010 }, { "epoch": 2.86, "learning_rate": 4.010411714455464e-06, "loss": 0.2464, "step": 22020 }, { "epoch": 2.86, "learning_rate": 4.009704309723607e-06, "loss": 0.219, "step": 22030 }, { "epoch": 2.86, "learning_rate": 4.008996598106326e-06, "loss": 0.2251, "step": 22040 }, { "epoch": 2.86, "learning_rate": 4.008288579733988e-06, "loss": 0.2378, "step": 22050 }, { "epoch": 2.86, "learning_rate": 4.007580254737018e-06, "loss": 0.2321, "step": 22060 }, { "epoch": 2.86, "learning_rate": 4.006871623245896e-06, "loss": 0.2332, "step": 22070 }, { "epoch": 2.87, "learning_rate": 4.006162685391159e-06, "loss": 0.2347, "step": 22080 }, { "epoch": 2.87, "learning_rate": 4.0054534413034004e-06, "loss": 0.2159, "step": 22090 }, { "epoch": 2.87, "learning_rate": 4.00474389111327e-06, "loss": 0.2251, "step": 22100 }, { "epoch": 2.87, "learning_rate": 4.004034034951474e-06, "loss": 0.2385, "step": 22110 }, { "epoch": 2.87, "learning_rate": 4.003323872948776e-06, "loss": 0.2477, "step": 22120 }, { "epoch": 2.87, "learning_rate": 4.002613405235993e-06, "loss": 0.217, "step": 22130 }, { "epoch": 2.87, "learning_rate": 4.001902631944003e-06, "loss": 0.2589, "step": 22140 }, { "epoch": 2.87, "learning_rate": 4.001191553203734e-06, "loss": 0.2314, "step": 22150 }, { "epoch": 2.88, "learning_rate": 4.000480169146177e-06, "loss": 0.235, "step": 22160 }, { "epoch": 2.88, "learning_rate": 3.999768479902374e-06, "loss": 0.2427, "step": 22170 }, { "epoch": 2.88, "learning_rate": 3.999056485603426e-06, "loss": 0.221, "step": 22180 }, { "epoch": 2.88, "learning_rate": 3.99834418638049e-06, "loss": 0.2427, "step": 22190 }, { "epoch": 2.88, "learning_rate": 3.997631582364777e-06, "loss": 0.2196, "step": 22200 }, { "epoch": 2.88, "learning_rate": 3.996918673687559e-06, "loss": 0.2307, "step": 22210 }, { "epoch": 2.88, "learning_rate": 3.996205460480157e-06, "loss": 0.2472, "step": 22220 }, { "epoch": 2.89, "learning_rate": 3.995491942873953e-06, "loss": 0.2329, "step": 22230 }, { "epoch": 2.89, "learning_rate": 3.994778121000387e-06, "loss": 0.2154, "step": 22240 }, { "epoch": 2.89, "learning_rate": 3.994063994990949e-06, "loss": 0.224, "step": 22250 }, { "epoch": 2.89, "learning_rate": 3.993349564977189e-06, "loss": 0.2159, "step": 22260 }, { "epoch": 2.89, "learning_rate": 3.9926348310907125e-06, "loss": 0.238, "step": 22270 }, { "epoch": 2.89, "learning_rate": 3.99191979346318e-06, "loss": 0.2419, "step": 22280 }, { "epoch": 2.89, "learning_rate": 3.991204452226309e-06, "loss": 0.2325, "step": 22290 }, { "epoch": 2.89, "learning_rate": 3.990488807511873e-06, "loss": 0.2407, "step": 22300 }, { "epoch": 2.9, "learning_rate": 3.989772859451699e-06, "loss": 0.2381, "step": 22310 }, { "epoch": 2.9, "learning_rate": 3.989056608177675e-06, "loss": 0.2258, "step": 22320 }, { "epoch": 2.9, "learning_rate": 3.988340053821739e-06, "loss": 0.2373, "step": 22330 }, { "epoch": 2.9, "learning_rate": 3.9876231965158876e-06, "loss": 0.2386, "step": 22340 }, { "epoch": 2.9, "learning_rate": 3.986906036392174e-06, "loss": 0.237, "step": 22350 }, { "epoch": 2.9, "learning_rate": 3.986188573582706e-06, "loss": 0.2386, "step": 22360 }, { "epoch": 2.9, "learning_rate": 3.985470808219648e-06, "loss": 0.2284, "step": 22370 }, { "epoch": 2.9, "learning_rate": 3.984752740435218e-06, "loss": 0.2505, "step": 22380 }, { "epoch": 2.91, "learning_rate": 3.984034370361693e-06, "loss": 0.2376, "step": 22390 }, { "epoch": 2.91, "learning_rate": 3.983315698131403e-06, "loss": 0.2418, "step": 22400 }, { "epoch": 2.91, "learning_rate": 3.982596723876734e-06, "loss": 0.2418, "step": 22410 }, { "epoch": 2.91, "learning_rate": 3.98187744773013e-06, "loss": 0.228, "step": 22420 }, { "epoch": 2.91, "learning_rate": 3.9811578698240865e-06, "loss": 0.2245, "step": 22430 }, { "epoch": 2.91, "learning_rate": 3.980437990291159e-06, "loss": 0.2261, "step": 22440 }, { "epoch": 2.91, "learning_rate": 3.9797178092639565e-06, "loss": 0.2309, "step": 22450 }, { "epoch": 2.91, "learning_rate": 3.978997326875142e-06, "loss": 0.2309, "step": 22460 }, { "epoch": 2.92, "learning_rate": 3.978276543257436e-06, "loss": 0.2423, "step": 22470 }, { "epoch": 2.92, "learning_rate": 3.977555458543615e-06, "loss": 0.2334, "step": 22480 }, { "epoch": 2.92, "learning_rate": 3.97683407286651e-06, "loss": 0.2443, "step": 22490 }, { "epoch": 2.92, "learning_rate": 3.976112386359006e-06, "loss": 0.2246, "step": 22500 }, { "epoch": 2.92, "learning_rate": 3.975390399154046e-06, "loss": 0.246, "step": 22510 }, { "epoch": 2.92, "learning_rate": 3.974668111384626e-06, "loss": 0.2354, "step": 22520 }, { "epoch": 2.92, "learning_rate": 3.973945523183801e-06, "loss": 0.2501, "step": 22530 }, { "epoch": 2.93, "learning_rate": 3.973222634684677e-06, "loss": 0.2478, "step": 22540 }, { "epoch": 2.93, "learning_rate": 3.972499446020418e-06, "loss": 0.2438, "step": 22550 }, { "epoch": 2.93, "learning_rate": 3.971775957324243e-06, "loss": 0.2362, "step": 22560 }, { "epoch": 2.93, "learning_rate": 3.971052168729424e-06, "loss": 0.2307, "step": 22570 }, { "epoch": 2.93, "learning_rate": 3.970328080369292e-06, "loss": 0.2224, "step": 22580 }, { "epoch": 2.93, "learning_rate": 3.969603692377231e-06, "loss": 0.2313, "step": 22590 }, { "epoch": 2.93, "learning_rate": 3.968879004886679e-06, "loss": 0.2244, "step": 22600 }, { "epoch": 2.93, "learning_rate": 3.968154018031133e-06, "loss": 0.2493, "step": 22610 }, { "epoch": 2.94, "learning_rate": 3.967428731944141e-06, "loss": 0.2274, "step": 22620 }, { "epoch": 2.94, "learning_rate": 3.966703146759309e-06, "loss": 0.2498, "step": 22630 }, { "epoch": 2.94, "learning_rate": 3.965977262610296e-06, "loss": 0.2412, "step": 22640 }, { "epoch": 2.94, "learning_rate": 3.9652510796308185e-06, "loss": 0.2246, "step": 22650 }, { "epoch": 2.94, "learning_rate": 3.964524597954646e-06, "loss": 0.223, "step": 22660 }, { "epoch": 2.94, "learning_rate": 3.963797817715603e-06, "loss": 0.2279, "step": 22670 }, { "epoch": 2.94, "learning_rate": 3.96307073904757e-06, "loss": 0.2155, "step": 22680 }, { "epoch": 2.94, "learning_rate": 3.962343362084483e-06, "loss": 0.2454, "step": 22690 }, { "epoch": 2.95, "learning_rate": 3.961615686960332e-06, "loss": 0.2356, "step": 22700 }, { "epoch": 2.95, "learning_rate": 3.960887713809161e-06, "loss": 0.2495, "step": 22710 }, { "epoch": 2.95, "learning_rate": 3.9601594427650705e-06, "loss": 0.2356, "step": 22720 }, { "epoch": 2.95, "learning_rate": 3.959430873962215e-06, "loss": 0.2334, "step": 22730 }, { "epoch": 2.95, "learning_rate": 3.958702007534804e-06, "loss": 0.2332, "step": 22740 }, { "epoch": 2.95, "learning_rate": 3.957972843617103e-06, "loss": 0.2267, "step": 22750 }, { "epoch": 2.95, "learning_rate": 3.957243382343431e-06, "loss": 0.2128, "step": 22760 }, { "epoch": 2.96, "learning_rate": 3.956513623848161e-06, "loss": 0.2263, "step": 22770 }, { "epoch": 2.96, "learning_rate": 3.9557835682657235e-06, "loss": 0.2362, "step": 22780 }, { "epoch": 2.96, "learning_rate": 3.9550532157306005e-06, "loss": 0.2403, "step": 22790 }, { "epoch": 2.96, "learning_rate": 3.954322566377331e-06, "loss": 0.2443, "step": 22800 }, { "epoch": 2.96, "learning_rate": 3.9535916203405075e-06, "loss": 0.2306, "step": 22810 }, { "epoch": 2.96, "learning_rate": 3.952860377754778e-06, "loss": 0.2334, "step": 22820 }, { "epoch": 2.96, "learning_rate": 3.952128838754844e-06, "loss": 0.2424, "step": 22830 }, { "epoch": 2.96, "learning_rate": 3.951397003475463e-06, "loss": 0.2407, "step": 22840 }, { "epoch": 2.97, "learning_rate": 3.950664872051447e-06, "loss": 0.2216, "step": 22850 }, { "epoch": 2.97, "learning_rate": 3.949932444617659e-06, "loss": 0.2417, "step": 22860 }, { "epoch": 2.97, "learning_rate": 3.949199721309023e-06, "loss": 0.2351, "step": 22870 }, { "epoch": 2.97, "learning_rate": 3.948466702260512e-06, "loss": 0.2447, "step": 22880 }, { "epoch": 2.97, "learning_rate": 3.947733387607156e-06, "loss": 0.2375, "step": 22890 }, { "epoch": 2.97, "learning_rate": 3.946999777484039e-06, "loss": 0.246, "step": 22900 }, { "epoch": 2.97, "learning_rate": 3.946265872026298e-06, "loss": 0.2433, "step": 22910 }, { "epoch": 2.97, "learning_rate": 3.9455316713691285e-06, "loss": 0.2246, "step": 22920 }, { "epoch": 2.98, "learning_rate": 3.944797175647775e-06, "loss": 0.2282, "step": 22930 }, { "epoch": 2.98, "learning_rate": 3.944062384997539e-06, "loss": 0.2333, "step": 22940 }, { "epoch": 2.98, "learning_rate": 3.943327299553778e-06, "loss": 0.2525, "step": 22950 }, { "epoch": 2.98, "learning_rate": 3.942591919451901e-06, "loss": 0.2298, "step": 22960 }, { "epoch": 2.98, "learning_rate": 3.941856244827372e-06, "loss": 0.245, "step": 22970 }, { "epoch": 2.98, "learning_rate": 3.941120275815712e-06, "loss": 0.2177, "step": 22980 }, { "epoch": 2.98, "learning_rate": 3.9403840125524895e-06, "loss": 0.2384, "step": 22990 }, { "epoch": 2.99, "learning_rate": 3.939647455173335e-06, "loss": 0.2248, "step": 23000 }, { "epoch": 2.99, "learning_rate": 3.93891060381393e-06, "loss": 0.2423, "step": 23010 }, { "epoch": 2.99, "learning_rate": 3.938173458610007e-06, "loss": 0.2236, "step": 23020 }, { "epoch": 2.99, "learning_rate": 3.937436019697358e-06, "loss": 0.2278, "step": 23030 }, { "epoch": 2.99, "learning_rate": 3.936698287211826e-06, "loss": 0.233, "step": 23040 }, { "epoch": 2.99, "learning_rate": 3.935960261289307e-06, "loss": 0.235, "step": 23050 }, { "epoch": 2.99, "learning_rate": 3.935221942065756e-06, "loss": 0.239, "step": 23060 }, { "epoch": 2.99, "learning_rate": 3.934483329677177e-06, "loss": 0.2414, "step": 23070 }, { "epoch": 3.0, "learning_rate": 3.933744424259629e-06, "loss": 0.2413, "step": 23080 }, { "epoch": 3.0, "learning_rate": 3.933005225949226e-06, "loss": 0.2396, "step": 23090 }, { "epoch": 3.0, "learning_rate": 3.9322657348821366e-06, "loss": 0.2182, "step": 23100 }, { "epoch": 3.0, "learning_rate": 3.931525951194582e-06, "loss": 0.2401, "step": 23110 }, { "epoch": 3.0, "eval_loss": 0.2334161400794983, "eval_runtime": 56.2221, "eval_samples_per_second": 193.465, "eval_steps_per_second": 24.19, "step": 23115 }, { "epoch": 3.0, "learning_rate": 3.930785875022839e-06, "loss": 0.2254, "step": 23120 }, { "epoch": 3.0, "learning_rate": 3.930045506503234e-06, "loss": 0.2291, "step": 23130 }, { "epoch": 3.0, "learning_rate": 3.929304845772153e-06, "loss": 0.2282, "step": 23140 }, { "epoch": 3.0, "learning_rate": 3.928563892966033e-06, "loss": 0.2282, "step": 23150 }, { "epoch": 3.01, "learning_rate": 3.9278226482213635e-06, "loss": 0.2366, "step": 23160 }, { "epoch": 3.01, "learning_rate": 3.9270811116746895e-06, "loss": 0.2227, "step": 23170 }, { "epoch": 3.01, "learning_rate": 3.92633928346261e-06, "loss": 0.2258, "step": 23180 }, { "epoch": 3.01, "learning_rate": 3.925597163721777e-06, "loss": 0.225, "step": 23190 }, { "epoch": 3.01, "learning_rate": 3.924854752588897e-06, "loss": 0.2398, "step": 23200 }, { "epoch": 3.01, "learning_rate": 3.924112050200727e-06, "loss": 0.2376, "step": 23210 }, { "epoch": 3.01, "learning_rate": 3.923369056694083e-06, "loss": 0.235, "step": 23220 }, { "epoch": 3.01, "learning_rate": 3.922625772205831e-06, "loss": 0.2326, "step": 23230 }, { "epoch": 3.02, "learning_rate": 3.921882196872892e-06, "loss": 0.2402, "step": 23240 }, { "epoch": 3.02, "learning_rate": 3.921138330832238e-06, "loss": 0.2309, "step": 23250 }, { "epoch": 3.02, "learning_rate": 3.920394174220899e-06, "loss": 0.2294, "step": 23260 }, { "epoch": 3.02, "learning_rate": 3.919649727175954e-06, "loss": 0.2425, "step": 23270 }, { "epoch": 3.02, "learning_rate": 3.918904989834539e-06, "loss": 0.244, "step": 23280 }, { "epoch": 3.02, "learning_rate": 3.918159962333841e-06, "loss": 0.2328, "step": 23290 }, { "epoch": 3.02, "learning_rate": 3.917414644811103e-06, "loss": 0.2369, "step": 23300 }, { "epoch": 3.03, "learning_rate": 3.916669037403618e-06, "loss": 0.2398, "step": 23310 }, { "epoch": 3.03, "learning_rate": 3.915923140248736e-06, "loss": 0.239, "step": 23320 }, { "epoch": 3.03, "learning_rate": 3.915176953483858e-06, "loss": 0.224, "step": 23330 }, { "epoch": 3.03, "learning_rate": 3.914430477246439e-06, "loss": 0.2287, "step": 23340 }, { "epoch": 3.03, "learning_rate": 3.913683711673988e-06, "loss": 0.2387, "step": 23350 }, { "epoch": 3.03, "learning_rate": 3.912936656904066e-06, "loss": 0.2201, "step": 23360 }, { "epoch": 3.03, "learning_rate": 3.912189313074287e-06, "loss": 0.2292, "step": 23370 }, { "epoch": 3.03, "learning_rate": 3.911441680322321e-06, "loss": 0.2284, "step": 23380 }, { "epoch": 3.04, "learning_rate": 3.910693758785889e-06, "loss": 0.2277, "step": 23390 }, { "epoch": 3.04, "learning_rate": 3.909945548602765e-06, "loss": 0.2299, "step": 23400 }, { "epoch": 3.04, "learning_rate": 3.909197049910778e-06, "loss": 0.2244, "step": 23410 }, { "epoch": 3.04, "learning_rate": 3.908448262847807e-06, "loss": 0.2259, "step": 23420 }, { "epoch": 3.04, "learning_rate": 3.907699187551787e-06, "loss": 0.2257, "step": 23430 }, { "epoch": 3.04, "learning_rate": 3.906949824160706e-06, "loss": 0.2262, "step": 23440 }, { "epoch": 3.04, "learning_rate": 3.906200172812604e-06, "loss": 0.2247, "step": 23450 }, { "epoch": 3.04, "learning_rate": 3.9054502336455724e-06, "loss": 0.2333, "step": 23460 }, { "epoch": 3.05, "learning_rate": 3.90470000679776e-06, "loss": 0.2288, "step": 23470 }, { "epoch": 3.05, "learning_rate": 3.903949492407364e-06, "loss": 0.2216, "step": 23480 }, { "epoch": 3.05, "learning_rate": 3.903198690612638e-06, "loss": 0.2458, "step": 23490 }, { "epoch": 3.05, "learning_rate": 3.902447601551887e-06, "loss": 0.2412, "step": 23500 }, { "epoch": 3.05, "learning_rate": 3.901696225363468e-06, "loss": 0.2309, "step": 23510 }, { "epoch": 3.05, "learning_rate": 3.900944562185793e-06, "loss": 0.2268, "step": 23520 }, { "epoch": 3.05, "learning_rate": 3.9001926121573265e-06, "loss": 0.2403, "step": 23530 }, { "epoch": 3.06, "learning_rate": 3.899440375416584e-06, "loss": 0.2362, "step": 23540 }, { "epoch": 3.06, "learning_rate": 3.898687852102137e-06, "loss": 0.2355, "step": 23550 }, { "epoch": 3.06, "learning_rate": 3.897935042352604e-06, "loss": 0.2291, "step": 23560 }, { "epoch": 3.06, "learning_rate": 3.897181946306664e-06, "loss": 0.2347, "step": 23570 }, { "epoch": 3.06, "learning_rate": 3.896428564103042e-06, "loss": 0.2333, "step": 23580 }, { "epoch": 3.06, "learning_rate": 3.89567489588052e-06, "loss": 0.2308, "step": 23590 }, { "epoch": 3.06, "learning_rate": 3.894920941777931e-06, "loss": 0.2113, "step": 23600 }, { "epoch": 3.06, "learning_rate": 3.8941667019341614e-06, "loss": 0.2347, "step": 23610 }, { "epoch": 3.07, "learning_rate": 3.8934121764881485e-06, "loss": 0.2372, "step": 23620 }, { "epoch": 3.07, "learning_rate": 3.892657365578884e-06, "loss": 0.2341, "step": 23630 }, { "epoch": 3.07, "learning_rate": 3.891902269345412e-06, "loss": 0.2187, "step": 23640 }, { "epoch": 3.07, "learning_rate": 3.891146887926827e-06, "loss": 0.248, "step": 23650 }, { "epoch": 3.07, "learning_rate": 3.890391221462281e-06, "loss": 0.224, "step": 23660 }, { "epoch": 3.07, "learning_rate": 3.8896352700909724e-06, "loss": 0.2355, "step": 23670 }, { "epoch": 3.07, "learning_rate": 3.888879033952156e-06, "loss": 0.2436, "step": 23680 }, { "epoch": 3.07, "learning_rate": 3.888122513185139e-06, "loss": 0.2141, "step": 23690 }, { "epoch": 3.08, "learning_rate": 3.8873657079292795e-06, "loss": 0.223, "step": 23700 }, { "epoch": 3.08, "learning_rate": 3.8866086183239875e-06, "loss": 0.2354, "step": 23710 }, { "epoch": 3.08, "learning_rate": 3.885851244508727e-06, "loss": 0.2183, "step": 23720 }, { "epoch": 3.08, "learning_rate": 3.885093586623015e-06, "loss": 0.2295, "step": 23730 }, { "epoch": 3.08, "learning_rate": 3.884335644806419e-06, "loss": 0.2198, "step": 23740 }, { "epoch": 3.08, "learning_rate": 3.883577419198558e-06, "loss": 0.2372, "step": 23750 }, { "epoch": 3.08, "learning_rate": 3.882818909939107e-06, "loss": 0.2265, "step": 23760 }, { "epoch": 3.09, "learning_rate": 3.882060117167789e-06, "loss": 0.2364, "step": 23770 }, { "epoch": 3.09, "learning_rate": 3.881301041024383e-06, "loss": 0.2468, "step": 23780 }, { "epoch": 3.09, "learning_rate": 3.880541681648717e-06, "loss": 0.2487, "step": 23790 }, { "epoch": 3.09, "learning_rate": 3.879782039180673e-06, "loss": 0.2324, "step": 23800 }, { "epoch": 3.09, "learning_rate": 3.8790221137601855e-06, "loss": 0.2133, "step": 23810 }, { "epoch": 3.09, "learning_rate": 3.878261905527239e-06, "loss": 0.2311, "step": 23820 }, { "epoch": 3.09, "learning_rate": 3.877501414621873e-06, "loss": 0.231, "step": 23830 }, { "epoch": 3.09, "learning_rate": 3.876740641184176e-06, "loss": 0.2465, "step": 23840 }, { "epoch": 3.1, "learning_rate": 3.875979585354291e-06, "loss": 0.2252, "step": 23850 }, { "epoch": 3.1, "learning_rate": 3.875218247272411e-06, "loss": 0.2265, "step": 23860 }, { "epoch": 3.1, "learning_rate": 3.8744566270787845e-06, "loss": 0.2433, "step": 23870 }, { "epoch": 3.1, "learning_rate": 3.873694724913707e-06, "loss": 0.2478, "step": 23880 }, { "epoch": 3.1, "learning_rate": 3.8729325409175295e-06, "loss": 0.2391, "step": 23890 }, { "epoch": 3.1, "learning_rate": 3.872170075230654e-06, "loss": 0.231, "step": 23900 }, { "epoch": 3.1, "learning_rate": 3.871407327993534e-06, "loss": 0.2214, "step": 23910 }, { "epoch": 3.1, "learning_rate": 3.870644299346675e-06, "loss": 0.2325, "step": 23920 }, { "epoch": 3.11, "learning_rate": 3.869880989430635e-06, "loss": 0.2261, "step": 23930 }, { "epoch": 3.11, "learning_rate": 3.869117398386024e-06, "loss": 0.2261, "step": 23940 }, { "epoch": 3.11, "learning_rate": 3.868353526353502e-06, "loss": 0.2342, "step": 23950 }, { "epoch": 3.11, "learning_rate": 3.8675893734737814e-06, "loss": 0.2323, "step": 23960 }, { "epoch": 3.11, "learning_rate": 3.866824939887628e-06, "loss": 0.2284, "step": 23970 }, { "epoch": 3.11, "learning_rate": 3.866060225735857e-06, "loss": 0.2341, "step": 23980 }, { "epoch": 3.11, "learning_rate": 3.865295231159337e-06, "loss": 0.2292, "step": 23990 }, { "epoch": 3.11, "learning_rate": 3.864529956298988e-06, "loss": 0.23, "step": 24000 }, { "epoch": 3.12, "learning_rate": 3.863764401295781e-06, "loss": 0.2313, "step": 24010 }, { "epoch": 3.12, "learning_rate": 3.8629985662907375e-06, "loss": 0.2287, "step": 24020 }, { "epoch": 3.12, "learning_rate": 3.8622324514249345e-06, "loss": 0.2363, "step": 24030 }, { "epoch": 3.12, "learning_rate": 3.861466056839496e-06, "loss": 0.2244, "step": 24040 }, { "epoch": 3.12, "learning_rate": 3.860699382675601e-06, "loss": 0.218, "step": 24050 }, { "epoch": 3.12, "learning_rate": 3.859932429074477e-06, "loss": 0.222, "step": 24060 }, { "epoch": 3.12, "learning_rate": 3.859165196177406e-06, "loss": 0.2475, "step": 24070 }, { "epoch": 3.13, "learning_rate": 3.858397684125718e-06, "loss": 0.2353, "step": 24080 }, { "epoch": 3.13, "learning_rate": 3.857629893060799e-06, "loss": 0.227, "step": 24090 }, { "epoch": 3.13, "learning_rate": 3.856861823124082e-06, "loss": 0.2315, "step": 24100 }, { "epoch": 3.13, "learning_rate": 3.856093474457054e-06, "loss": 0.2226, "step": 24110 }, { "epoch": 3.13, "learning_rate": 3.855324847201252e-06, "loss": 0.2319, "step": 24120 }, { "epoch": 3.13, "learning_rate": 3.854555941498264e-06, "loss": 0.2484, "step": 24130 }, { "epoch": 3.13, "learning_rate": 3.853786757489732e-06, "loss": 0.2479, "step": 24140 }, { "epoch": 3.13, "learning_rate": 3.853017295317346e-06, "loss": 0.2368, "step": 24150 }, { "epoch": 3.14, "learning_rate": 3.8522475551228485e-06, "loss": 0.2322, "step": 24160 }, { "epoch": 3.14, "learning_rate": 3.851477537048035e-06, "loss": 0.2183, "step": 24170 }, { "epoch": 3.14, "learning_rate": 3.850707241234749e-06, "loss": 0.2183, "step": 24180 }, { "epoch": 3.14, "learning_rate": 3.849936667824887e-06, "loss": 0.2222, "step": 24190 }, { "epoch": 3.14, "learning_rate": 3.849165816960396e-06, "loss": 0.2312, "step": 24200 }, { "epoch": 3.14, "learning_rate": 3.848394688783275e-06, "loss": 0.2346, "step": 24210 }, { "epoch": 3.14, "learning_rate": 3.847623283435573e-06, "loss": 0.2354, "step": 24220 }, { "epoch": 3.14, "learning_rate": 3.846851601059391e-06, "loss": 0.2324, "step": 24230 }, { "epoch": 3.15, "learning_rate": 3.846079641796881e-06, "loss": 0.2319, "step": 24240 }, { "epoch": 3.15, "learning_rate": 3.845307405790244e-06, "loss": 0.2321, "step": 24250 }, { "epoch": 3.15, "learning_rate": 3.844534893181736e-06, "loss": 0.2372, "step": 24260 }, { "epoch": 3.15, "learning_rate": 3.843762104113659e-06, "loss": 0.2297, "step": 24270 }, { "epoch": 3.15, "learning_rate": 3.842989038728371e-06, "loss": 0.2421, "step": 24280 }, { "epoch": 3.15, "learning_rate": 3.8422156971682766e-06, "loss": 0.2298, "step": 24290 }, { "epoch": 3.15, "learning_rate": 3.8414420795758325e-06, "loss": 0.2236, "step": 24300 }, { "epoch": 3.16, "learning_rate": 3.840668186093549e-06, "loss": 0.2403, "step": 24310 }, { "epoch": 3.16, "learning_rate": 3.839894016863983e-06, "loss": 0.2427, "step": 24320 }, { "epoch": 3.16, "learning_rate": 3.839119572029745e-06, "loss": 0.2538, "step": 24330 }, { "epoch": 3.16, "learning_rate": 3.838344851733496e-06, "loss": 0.2332, "step": 24340 }, { "epoch": 3.16, "learning_rate": 3.837569856117946e-06, "loss": 0.2374, "step": 24350 }, { "epoch": 3.16, "learning_rate": 3.836794585325858e-06, "loss": 0.2275, "step": 24360 }, { "epoch": 3.16, "learning_rate": 3.8360190395000445e-06, "loss": 0.2371, "step": 24370 }, { "epoch": 3.16, "learning_rate": 3.835243218783368e-06, "loss": 0.2264, "step": 24380 }, { "epoch": 3.17, "learning_rate": 3.834467123318743e-06, "loss": 0.2257, "step": 24390 }, { "epoch": 3.17, "learning_rate": 3.8336907532491345e-06, "loss": 0.2389, "step": 24400 }, { "epoch": 3.17, "learning_rate": 3.832914108717557e-06, "loss": 0.2257, "step": 24410 }, { "epoch": 3.17, "learning_rate": 3.832137189867076e-06, "loss": 0.2314, "step": 24420 }, { "epoch": 3.17, "learning_rate": 3.831359996840808e-06, "loss": 0.2348, "step": 24430 }, { "epoch": 3.17, "learning_rate": 3.83058252978192e-06, "loss": 0.245, "step": 24440 }, { "epoch": 3.17, "learning_rate": 3.829804788833629e-06, "loss": 0.2294, "step": 24450 }, { "epoch": 3.17, "learning_rate": 3.829026774139202e-06, "loss": 0.2256, "step": 24460 }, { "epoch": 3.18, "learning_rate": 3.828248485841958e-06, "loss": 0.2153, "step": 24470 }, { "epoch": 3.18, "learning_rate": 3.827469924085265e-06, "loss": 0.2385, "step": 24480 }, { "epoch": 3.18, "learning_rate": 3.826691089012542e-06, "loss": 0.2188, "step": 24490 }, { "epoch": 3.18, "learning_rate": 3.8259119807672585e-06, "loss": 0.24, "step": 24500 }, { "epoch": 3.18, "learning_rate": 3.825132599492933e-06, "loss": 0.22, "step": 24510 }, { "epoch": 3.18, "learning_rate": 3.824352945333136e-06, "loss": 0.246, "step": 24520 }, { "epoch": 3.18, "learning_rate": 3.823573018431488e-06, "loss": 0.2385, "step": 24530 }, { "epoch": 3.18, "learning_rate": 3.822792818931659e-06, "loss": 0.2244, "step": 24540 }, { "epoch": 3.19, "learning_rate": 3.82201234697737e-06, "loss": 0.235, "step": 24550 }, { "epoch": 3.19, "learning_rate": 3.82123160271239e-06, "loss": 0.2344, "step": 24560 }, { "epoch": 3.19, "learning_rate": 3.820450586280542e-06, "loss": 0.2552, "step": 24570 }, { "epoch": 3.19, "learning_rate": 3.819669297825695e-06, "loss": 0.2392, "step": 24580 }, { "epoch": 3.19, "learning_rate": 3.818887737491772e-06, "loss": 0.2501, "step": 24590 }, { "epoch": 3.19, "learning_rate": 3.818105905422744e-06, "loss": 0.2262, "step": 24600 }, { "epoch": 3.19, "learning_rate": 3.817323801762631e-06, "loss": 0.2416, "step": 24610 }, { "epoch": 3.2, "learning_rate": 3.816541426655505e-06, "loss": 0.2327, "step": 24620 }, { "epoch": 3.2, "learning_rate": 3.815758780245488e-06, "loss": 0.2401, "step": 24630 }, { "epoch": 3.2, "learning_rate": 3.81497586267675e-06, "loss": 0.2442, "step": 24640 }, { "epoch": 3.2, "learning_rate": 3.814192674093513e-06, "loss": 0.2369, "step": 24650 }, { "epoch": 3.2, "learning_rate": 3.8134092146400485e-06, "loss": 0.244, "step": 24660 }, { "epoch": 3.2, "learning_rate": 3.812625484460677e-06, "loss": 0.2432, "step": 24670 }, { "epoch": 3.2, "learning_rate": 3.811841483699769e-06, "loss": 0.2408, "step": 24680 }, { "epoch": 3.2, "learning_rate": 3.8110572125017464e-06, "loss": 0.227, "step": 24690 }, { "epoch": 3.21, "learning_rate": 3.8102726710110786e-06, "loss": 0.2355, "step": 24700 }, { "epoch": 3.21, "learning_rate": 3.809487859372287e-06, "loss": 0.226, "step": 24710 }, { "epoch": 3.21, "learning_rate": 3.8087027777299407e-06, "loss": 0.231, "step": 24720 }, { "epoch": 3.21, "learning_rate": 3.8079174262286607e-06, "loss": 0.2379, "step": 24730 }, { "epoch": 3.21, "learning_rate": 3.807131805013116e-06, "loss": 0.2351, "step": 24740 }, { "epoch": 3.21, "learning_rate": 3.8063459142280253e-06, "loss": 0.2366, "step": 24750 }, { "epoch": 3.21, "learning_rate": 3.80555975401816e-06, "loss": 0.2306, "step": 24760 }, { "epoch": 3.21, "learning_rate": 3.8047733245283353e-06, "loss": 0.2352, "step": 24770 }, { "epoch": 3.22, "learning_rate": 3.803986625903421e-06, "loss": 0.2288, "step": 24780 }, { "epoch": 3.22, "learning_rate": 3.803199658288335e-06, "loss": 0.2412, "step": 24790 }, { "epoch": 3.22, "learning_rate": 3.802412421828044e-06, "loss": 0.2467, "step": 24800 }, { "epoch": 3.22, "learning_rate": 3.801624916667566e-06, "loss": 0.2326, "step": 24810 }, { "epoch": 3.22, "learning_rate": 3.8008371429519663e-06, "loss": 0.2492, "step": 24820 }, { "epoch": 3.22, "learning_rate": 3.8000491008263605e-06, "loss": 0.2361, "step": 24830 }, { "epoch": 3.22, "learning_rate": 3.7992607904359145e-06, "loss": 0.249, "step": 24840 }, { "epoch": 3.23, "learning_rate": 3.798472211925842e-06, "loss": 0.2435, "step": 24850 }, { "epoch": 3.23, "learning_rate": 3.797683365441408e-06, "loss": 0.2251, "step": 24860 }, { "epoch": 3.23, "learning_rate": 3.796894251127926e-06, "loss": 0.2293, "step": 24870 }, { "epoch": 3.23, "learning_rate": 3.7961048691307583e-06, "loss": 0.2281, "step": 24880 }, { "epoch": 3.23, "learning_rate": 3.795315219595317e-06, "loss": 0.2243, "step": 24890 }, { "epoch": 3.23, "learning_rate": 3.7945253026670625e-06, "loss": 0.2251, "step": 24900 }, { "epoch": 3.23, "learning_rate": 3.793735118491507e-06, "loss": 0.2389, "step": 24910 }, { "epoch": 3.23, "learning_rate": 3.7929446672142096e-06, "loss": 0.2197, "step": 24920 }, { "epoch": 3.24, "learning_rate": 3.7921539489807796e-06, "loss": 0.2276, "step": 24930 }, { "epoch": 3.24, "learning_rate": 3.791362963936875e-06, "loss": 0.2282, "step": 24940 }, { "epoch": 3.24, "learning_rate": 3.790571712228202e-06, "loss": 0.2299, "step": 24950 }, { "epoch": 3.24, "learning_rate": 3.7897801940005197e-06, "loss": 0.2308, "step": 24960 }, { "epoch": 3.24, "learning_rate": 3.7889884093996313e-06, "loss": 0.2478, "step": 24970 }, { "epoch": 3.24, "learning_rate": 3.7881963585713925e-06, "loss": 0.2301, "step": 24980 }, { "epoch": 3.24, "learning_rate": 3.7874040416617065e-06, "loss": 0.224, "step": 24990 }, { "epoch": 3.24, "learning_rate": 3.786611458816526e-06, "loss": 0.2389, "step": 25000 }, { "epoch": 3.25, "learning_rate": 3.785818610181853e-06, "loss": 0.2314, "step": 25010 }, { "epoch": 3.25, "learning_rate": 3.7850254959037384e-06, "loss": 0.2404, "step": 25020 }, { "epoch": 3.25, "learning_rate": 3.784232116128281e-06, "loss": 0.2329, "step": 25030 }, { "epoch": 3.25, "learning_rate": 3.7834384710016297e-06, "loss": 0.2238, "step": 25040 }, { "epoch": 3.25, "learning_rate": 3.7826445606699817e-06, "loss": 0.226, "step": 25050 }, { "epoch": 3.25, "learning_rate": 3.781850385279584e-06, "loss": 0.2134, "step": 25060 }, { "epoch": 3.25, "learning_rate": 3.7810559449767306e-06, "loss": 0.2342, "step": 25070 }, { "epoch": 3.26, "learning_rate": 3.7802612399077654e-06, "loss": 0.2282, "step": 25080 }, { "epoch": 3.26, "learning_rate": 3.7794662702190823e-06, "loss": 0.2253, "step": 25090 }, { "epoch": 3.26, "learning_rate": 3.7786710360571204e-06, "loss": 0.2308, "step": 25100 }, { "epoch": 3.26, "learning_rate": 3.7778755375683718e-06, "loss": 0.2348, "step": 25110 }, { "epoch": 3.26, "learning_rate": 3.777079774899375e-06, "loss": 0.2251, "step": 25120 }, { "epoch": 3.26, "learning_rate": 3.776283748196716e-06, "loss": 0.2231, "step": 25130 }, { "epoch": 3.26, "learning_rate": 3.7754874576070323e-06, "loss": 0.2325, "step": 25140 }, { "epoch": 3.26, "learning_rate": 3.774690903277008e-06, "loss": 0.2166, "step": 25150 }, { "epoch": 3.27, "learning_rate": 3.7738940853533767e-06, "loss": 0.2263, "step": 25160 }, { "epoch": 3.27, "learning_rate": 3.7730970039829196e-06, "loss": 0.2231, "step": 25170 }, { "epoch": 3.27, "learning_rate": 3.772299659312468e-06, "loss": 0.2485, "step": 25180 }, { "epoch": 3.27, "learning_rate": 3.7715020514889e-06, "loss": 0.2298, "step": 25190 }, { "epoch": 3.27, "learning_rate": 3.7707041806591425e-06, "loss": 0.2274, "step": 25200 }, { "epoch": 3.27, "learning_rate": 3.7699060469701727e-06, "loss": 0.2374, "step": 25210 }, { "epoch": 3.27, "learning_rate": 3.7691076505690135e-06, "loss": 0.246, "step": 25220 }, { "epoch": 3.27, "learning_rate": 3.768308991602739e-06, "loss": 0.241, "step": 25230 }, { "epoch": 3.28, "learning_rate": 3.767510070218468e-06, "loss": 0.2423, "step": 25240 }, { "epoch": 3.28, "learning_rate": 3.766710886563371e-06, "loss": 0.2455, "step": 25250 }, { "epoch": 3.28, "learning_rate": 3.7659114407846666e-06, "loss": 0.2382, "step": 25260 }, { "epoch": 3.28, "learning_rate": 3.765111733029619e-06, "loss": 0.2432, "step": 25270 }, { "epoch": 3.28, "learning_rate": 3.764311763445543e-06, "loss": 0.2276, "step": 25280 }, { "epoch": 3.28, "learning_rate": 3.763511532179801e-06, "loss": 0.2323, "step": 25290 }, { "epoch": 3.28, "learning_rate": 3.7627110393798036e-06, "loss": 0.2474, "step": 25300 }, { "epoch": 3.28, "learning_rate": 3.76191028519301e-06, "loss": 0.2211, "step": 25310 }, { "epoch": 3.29, "learning_rate": 3.7611092697669264e-06, "loss": 0.2304, "step": 25320 }, { "epoch": 3.29, "learning_rate": 3.760307993249108e-06, "loss": 0.2353, "step": 25330 }, { "epoch": 3.29, "learning_rate": 3.7595064557871574e-06, "loss": 0.2312, "step": 25340 }, { "epoch": 3.29, "learning_rate": 3.758704657528728e-06, "loss": 0.2408, "step": 25350 }, { "epoch": 3.29, "learning_rate": 3.7579025986215167e-06, "loss": 0.2266, "step": 25360 }, { "epoch": 3.29, "learning_rate": 3.7571002792132717e-06, "loss": 0.2378, "step": 25370 }, { "epoch": 3.29, "learning_rate": 3.7562976994517876e-06, "loss": 0.2301, "step": 25380 }, { "epoch": 3.3, "learning_rate": 3.755494859484909e-06, "loss": 0.228, "step": 25390 }, { "epoch": 3.3, "learning_rate": 3.754691759460526e-06, "loss": 0.2281, "step": 25400 }, { "epoch": 3.3, "learning_rate": 3.7538883995265772e-06, "loss": 0.2259, "step": 25410 }, { "epoch": 3.3, "learning_rate": 3.753084779831051e-06, "loss": 0.2336, "step": 25420 }, { "epoch": 3.3, "learning_rate": 3.752280900521981e-06, "loss": 0.2361, "step": 25430 }, { "epoch": 3.3, "learning_rate": 3.75147676174745e-06, "loss": 0.2267, "step": 25440 }, { "epoch": 3.3, "learning_rate": 3.7506723636555884e-06, "loss": 0.2494, "step": 25450 }, { "epoch": 3.3, "learning_rate": 3.7498677063945753e-06, "loss": 0.2358, "step": 25460 }, { "epoch": 3.31, "learning_rate": 3.7490627901126348e-06, "loss": 0.2208, "step": 25470 }, { "epoch": 3.31, "learning_rate": 3.748257614958042e-06, "loss": 0.2464, "step": 25480 }, { "epoch": 3.31, "learning_rate": 3.7474521810791177e-06, "loss": 0.2375, "step": 25490 }, { "epoch": 3.31, "learning_rate": 3.7466464886242306e-06, "loss": 0.232, "step": 25500 }, { "epoch": 3.31, "learning_rate": 3.7458405377417978e-06, "loss": 0.241, "step": 25510 }, { "epoch": 3.31, "learning_rate": 3.745034328580283e-06, "loss": 0.2256, "step": 25520 }, { "epoch": 3.31, "learning_rate": 3.744227861288199e-06, "loss": 0.2281, "step": 25530 }, { "epoch": 3.31, "learning_rate": 3.743421136014104e-06, "loss": 0.2326, "step": 25540 }, { "epoch": 3.32, "learning_rate": 3.7426141529066053e-06, "loss": 0.2375, "step": 25550 }, { "epoch": 3.32, "learning_rate": 3.7418069121143566e-06, "loss": 0.2387, "step": 25560 }, { "epoch": 3.32, "learning_rate": 3.7409994137860615e-06, "loss": 0.2309, "step": 25570 }, { "epoch": 3.32, "learning_rate": 3.7401916580704674e-06, "loss": 0.222, "step": 25580 }, { "epoch": 3.32, "learning_rate": 3.7393836451163713e-06, "loss": 0.2415, "step": 25590 }, { "epoch": 3.32, "learning_rate": 3.7385753750726176e-06, "loss": 0.2384, "step": 25600 }, { "epoch": 3.32, "learning_rate": 3.7377668480880982e-06, "loss": 0.2367, "step": 25610 }, { "epoch": 3.33, "learning_rate": 3.736958064311751e-06, "loss": 0.2198, "step": 25620 }, { "epoch": 3.33, "learning_rate": 3.7361490238925633e-06, "loss": 0.2376, "step": 25630 }, { "epoch": 3.33, "learning_rate": 3.735339726979566e-06, "loss": 0.2311, "step": 25640 }, { "epoch": 3.33, "learning_rate": 3.7345301737218424e-06, "loss": 0.2504, "step": 25650 }, { "epoch": 3.33, "learning_rate": 3.733720364268519e-06, "loss": 0.236, "step": 25660 }, { "epoch": 3.33, "learning_rate": 3.7329102987687707e-06, "loss": 0.2283, "step": 25670 }, { "epoch": 3.33, "learning_rate": 3.73209997737182e-06, "loss": 0.2298, "step": 25680 }, { "epoch": 3.33, "learning_rate": 3.7312894002269356e-06, "loss": 0.2254, "step": 25690 }, { "epoch": 3.34, "learning_rate": 3.7304785674834343e-06, "loss": 0.2142, "step": 25700 }, { "epoch": 3.34, "learning_rate": 3.7296674792906805e-06, "loss": 0.2357, "step": 25710 }, { "epoch": 3.34, "learning_rate": 3.7288561357980836e-06, "loss": 0.2184, "step": 25720 }, { "epoch": 3.34, "learning_rate": 3.728044537155101e-06, "loss": 0.224, "step": 25730 }, { "epoch": 3.34, "learning_rate": 3.727232683511238e-06, "loss": 0.228, "step": 25740 }, { "epoch": 3.34, "learning_rate": 3.7264205750160456e-06, "loss": 0.2378, "step": 25750 }, { "epoch": 3.34, "learning_rate": 3.7256082118191234e-06, "loss": 0.2397, "step": 25760 }, { "epoch": 3.34, "learning_rate": 3.724795594070115e-06, "loss": 0.2322, "step": 25770 }, { "epoch": 3.35, "learning_rate": 3.7239827219187145e-06, "loss": 0.2267, "step": 25780 }, { "epoch": 3.35, "learning_rate": 3.72316959551466e-06, "loss": 0.2207, "step": 25790 }, { "epoch": 3.35, "learning_rate": 3.7223562150077383e-06, "loss": 0.2272, "step": 25800 }, { "epoch": 3.35, "learning_rate": 3.721542580547781e-06, "loss": 0.2192, "step": 25810 }, { "epoch": 3.35, "learning_rate": 3.7207286922846684e-06, "loss": 0.2443, "step": 25820 }, { "epoch": 3.35, "learning_rate": 3.719914550368327e-06, "loss": 0.2371, "step": 25830 }, { "epoch": 3.35, "learning_rate": 3.7191001549487303e-06, "loss": 0.2415, "step": 25840 }, { "epoch": 3.35, "learning_rate": 3.7182855061758963e-06, "loss": 0.2508, "step": 25850 }, { "epoch": 3.36, "learning_rate": 3.717470604199894e-06, "loss": 0.2313, "step": 25860 }, { "epoch": 3.36, "learning_rate": 3.716655449170834e-06, "loss": 0.2275, "step": 25870 }, { "epoch": 3.36, "learning_rate": 3.7158400412388776e-06, "loss": 0.2312, "step": 25880 }, { "epoch": 3.36, "learning_rate": 3.71502438055423e-06, "loss": 0.2457, "step": 25890 }, { "epoch": 3.36, "learning_rate": 3.7142084672671456e-06, "loss": 0.2314, "step": 25900 }, { "epoch": 3.36, "learning_rate": 3.713392301527922e-06, "loss": 0.2369, "step": 25910 }, { "epoch": 3.36, "learning_rate": 3.7125758834869064e-06, "loss": 0.2313, "step": 25920 }, { "epoch": 3.37, "learning_rate": 3.711759213294491e-06, "loss": 0.2262, "step": 25930 }, { "epoch": 3.37, "learning_rate": 3.710942291101114e-06, "loss": 0.242, "step": 25940 }, { "epoch": 3.37, "learning_rate": 3.71012511705726e-06, "loss": 0.2329, "step": 25950 }, { "epoch": 3.37, "learning_rate": 3.709307691313462e-06, "loss": 0.2348, "step": 25960 }, { "epoch": 3.37, "learning_rate": 3.708490014020298e-06, "loss": 0.2328, "step": 25970 }, { "epoch": 3.37, "learning_rate": 3.7076720853283917e-06, "loss": 0.2215, "step": 25980 }, { "epoch": 3.37, "learning_rate": 3.7068539053884137e-06, "loss": 0.239, "step": 25990 }, { "epoch": 3.37, "learning_rate": 3.706035474351081e-06, "loss": 0.2439, "step": 26000 }, { "epoch": 3.38, "learning_rate": 3.705216792367157e-06, "loss": 0.2328, "step": 26010 }, { "epoch": 3.38, "learning_rate": 3.704397859587451e-06, "loss": 0.2259, "step": 26020 }, { "epoch": 3.38, "learning_rate": 3.703578676162818e-06, "loss": 0.2238, "step": 26030 }, { "epoch": 3.38, "learning_rate": 3.702759242244161e-06, "loss": 0.2204, "step": 26040 }, { "epoch": 3.38, "learning_rate": 3.7019395579824272e-06, "loss": 0.2297, "step": 26050 }, { "epoch": 3.38, "learning_rate": 3.70111962352861e-06, "loss": 0.2242, "step": 26060 }, { "epoch": 3.38, "learning_rate": 3.700299439033751e-06, "loss": 0.2226, "step": 26070 }, { "epoch": 3.38, "learning_rate": 3.699479004648935e-06, "loss": 0.2231, "step": 26080 }, { "epoch": 3.39, "learning_rate": 3.6986583205252945e-06, "loss": 0.2346, "step": 26090 }, { "epoch": 3.39, "learning_rate": 3.697837386814009e-06, "loss": 0.2204, "step": 26100 }, { "epoch": 3.39, "learning_rate": 3.6970162036663006e-06, "loss": 0.2437, "step": 26110 }, { "epoch": 3.39, "learning_rate": 3.6961947712334415e-06, "loss": 0.2369, "step": 26120 }, { "epoch": 3.39, "learning_rate": 3.695373089666746e-06, "loss": 0.2238, "step": 26130 }, { "epoch": 3.39, "learning_rate": 3.694551159117577e-06, "loss": 0.2349, "step": 26140 }, { "epoch": 3.39, "learning_rate": 3.6937289797373423e-06, "loss": 0.2281, "step": 26150 }, { "epoch": 3.4, "learning_rate": 3.692906551677495e-06, "loss": 0.2245, "step": 26160 }, { "epoch": 3.4, "learning_rate": 3.6920838750895354e-06, "loss": 0.2274, "step": 26170 }, { "epoch": 3.4, "learning_rate": 3.691260950125008e-06, "loss": 0.2397, "step": 26180 }, { "epoch": 3.4, "learning_rate": 3.6904377769355044e-06, "loss": 0.2487, "step": 26190 }, { "epoch": 3.4, "learning_rate": 3.6896143556726605e-06, "loss": 0.2142, "step": 26200 }, { "epoch": 3.4, "learning_rate": 3.6887906864881595e-06, "loss": 0.2236, "step": 26210 }, { "epoch": 3.4, "learning_rate": 3.687966769533729e-06, "loss": 0.2269, "step": 26220 }, { "epoch": 3.4, "learning_rate": 3.6871426049611433e-06, "loss": 0.2286, "step": 26230 }, { "epoch": 3.41, "learning_rate": 3.6863181929222216e-06, "loss": 0.2252, "step": 26240 }, { "epoch": 3.41, "learning_rate": 3.6854935335688282e-06, "loss": 0.2401, "step": 26250 }, { "epoch": 3.41, "learning_rate": 3.684668627052874e-06, "loss": 0.2403, "step": 26260 }, { "epoch": 3.41, "learning_rate": 3.683843473526316e-06, "loss": 0.2301, "step": 26270 }, { "epoch": 3.41, "learning_rate": 3.6830180731411538e-06, "loss": 0.2334, "step": 26280 }, { "epoch": 3.41, "learning_rate": 3.682192426049435e-06, "loss": 0.2241, "step": 26290 }, { "epoch": 3.41, "learning_rate": 3.681366532403253e-06, "loss": 0.2374, "step": 26300 }, { "epoch": 3.41, "learning_rate": 3.6805403923547448e-06, "loss": 0.2367, "step": 26310 }, { "epoch": 3.42, "learning_rate": 3.6797140060560942e-06, "loss": 0.2421, "step": 26320 }, { "epoch": 3.42, "learning_rate": 3.6788873736595294e-06, "loss": 0.2318, "step": 26330 }, { "epoch": 3.42, "learning_rate": 3.678060495317325e-06, "loss": 0.242, "step": 26340 }, { "epoch": 3.42, "learning_rate": 3.677233371181799e-06, "loss": 0.2451, "step": 26350 }, { "epoch": 3.42, "learning_rate": 3.676406001405317e-06, "loss": 0.2022, "step": 26360 }, { "epoch": 3.42, "learning_rate": 3.675578386140288e-06, "loss": 0.2248, "step": 26370 }, { "epoch": 3.42, "learning_rate": 3.6747505255391678e-06, "loss": 0.2428, "step": 26380 }, { "epoch": 3.43, "learning_rate": 3.6739224197544563e-06, "loss": 0.2312, "step": 26390 }, { "epoch": 3.43, "learning_rate": 3.673094068938698e-06, "loss": 0.228, "step": 26400 }, { "epoch": 3.43, "learning_rate": 3.672265473244485e-06, "loss": 0.2414, "step": 26410 }, { "epoch": 3.43, "learning_rate": 3.6714366328244515e-06, "loss": 0.2355, "step": 26420 }, { "epoch": 3.43, "learning_rate": 3.670607547831279e-06, "loss": 0.2229, "step": 26430 }, { "epoch": 3.43, "learning_rate": 3.669778218417693e-06, "loss": 0.2279, "step": 26440 }, { "epoch": 3.43, "learning_rate": 3.6689486447364646e-06, "loss": 0.2354, "step": 26450 }, { "epoch": 3.43, "learning_rate": 3.668118826940409e-06, "loss": 0.2258, "step": 26460 }, { "epoch": 3.44, "learning_rate": 3.6672887651823863e-06, "loss": 0.2285, "step": 26470 }, { "epoch": 3.44, "learning_rate": 3.6664584596153044e-06, "loss": 0.2349, "step": 26480 }, { "epoch": 3.44, "learning_rate": 3.6656279103921114e-06, "loss": 0.2334, "step": 26490 }, { "epoch": 3.44, "learning_rate": 3.6647971176658044e-06, "loss": 0.2287, "step": 26500 }, { "epoch": 3.44, "learning_rate": 3.6639660815894224e-06, "loss": 0.2335, "step": 26510 }, { "epoch": 3.44, "learning_rate": 3.663134802316052e-06, "loss": 0.2401, "step": 26520 }, { "epoch": 3.44, "learning_rate": 3.6623032799988227e-06, "loss": 0.2259, "step": 26530 }, { "epoch": 3.44, "learning_rate": 3.6614715147909083e-06, "loss": 0.2305, "step": 26540 }, { "epoch": 3.45, "learning_rate": 3.6606395068455296e-06, "loss": 0.2287, "step": 26550 }, { "epoch": 3.45, "learning_rate": 3.65980725631595e-06, "loss": 0.2348, "step": 26560 }, { "epoch": 3.45, "learning_rate": 3.658974763355478e-06, "loss": 0.2382, "step": 26570 }, { "epoch": 3.45, "learning_rate": 3.6581420281174675e-06, "loss": 0.2441, "step": 26580 }, { "epoch": 3.45, "learning_rate": 3.6573090507553168e-06, "loss": 0.2543, "step": 26590 }, { "epoch": 3.45, "learning_rate": 3.6564758314224684e-06, "loss": 0.2325, "step": 26600 }, { "epoch": 3.45, "learning_rate": 3.655642370272411e-06, "loss": 0.2391, "step": 26610 }, { "epoch": 3.45, "learning_rate": 3.6548086674586738e-06, "loss": 0.2382, "step": 26620 }, { "epoch": 3.46, "learning_rate": 3.6539747231348353e-06, "loss": 0.2448, "step": 26630 }, { "epoch": 3.46, "learning_rate": 3.653140537454516e-06, "loss": 0.2281, "step": 26640 }, { "epoch": 3.46, "learning_rate": 3.6523061105713806e-06, "loss": 0.2216, "step": 26650 }, { "epoch": 3.46, "learning_rate": 3.6514714426391394e-06, "loss": 0.2271, "step": 26660 }, { "epoch": 3.46, "learning_rate": 3.6506365338115458e-06, "loss": 0.2348, "step": 26670 }, { "epoch": 3.46, "learning_rate": 3.6498013842423998e-06, "loss": 0.2237, "step": 26680 }, { "epoch": 3.46, "learning_rate": 3.648965994085543e-06, "loss": 0.2442, "step": 26690 }, { "epoch": 3.47, "learning_rate": 3.6481303634948637e-06, "loss": 0.2166, "step": 26700 }, { "epoch": 3.47, "learning_rate": 3.647294492624292e-06, "loss": 0.2369, "step": 26710 }, { "epoch": 3.47, "learning_rate": 3.646458381627804e-06, "loss": 0.2249, "step": 26720 }, { "epoch": 3.47, "learning_rate": 3.6456220306594215e-06, "loss": 0.2384, "step": 26730 }, { "epoch": 3.47, "learning_rate": 3.6447854398732063e-06, "loss": 0.2386, "step": 26740 }, { "epoch": 3.47, "learning_rate": 3.6439486094232683e-06, "loss": 0.2324, "step": 26750 }, { "epoch": 3.47, "learning_rate": 3.643111539463759e-06, "loss": 0.2184, "step": 26760 }, { "epoch": 3.47, "learning_rate": 3.6422742301488755e-06, "loss": 0.2243, "step": 26770 }, { "epoch": 3.48, "learning_rate": 3.641436681632859e-06, "loss": 0.2292, "step": 26780 }, { "epoch": 3.48, "learning_rate": 3.640598894069994e-06, "loss": 0.2317, "step": 26790 }, { "epoch": 3.48, "learning_rate": 3.639760867614609e-06, "loss": 0.2554, "step": 26800 }, { "epoch": 3.48, "learning_rate": 3.6389226024210775e-06, "loss": 0.2355, "step": 26810 }, { "epoch": 3.48, "learning_rate": 3.6380840986438154e-06, "loss": 0.2333, "step": 26820 }, { "epoch": 3.48, "learning_rate": 3.637245356437284e-06, "loss": 0.2323, "step": 26830 }, { "epoch": 3.48, "learning_rate": 3.6364063759559885e-06, "loss": 0.2394, "step": 26840 }, { "epoch": 3.48, "learning_rate": 3.6355671573544762e-06, "loss": 0.2272, "step": 26850 }, { "epoch": 3.49, "learning_rate": 3.6347277007873406e-06, "loss": 0.2313, "step": 26860 }, { "epoch": 3.49, "learning_rate": 3.6338880064092177e-06, "loss": 0.2306, "step": 26870 }, { "epoch": 3.49, "learning_rate": 3.633048074374788e-06, "loss": 0.2197, "step": 26880 }, { "epoch": 3.49, "learning_rate": 3.6322079048387744e-06, "loss": 0.233, "step": 26890 }, { "epoch": 3.49, "learning_rate": 3.6313674979559455e-06, "loss": 0.238, "step": 26900 }, { "epoch": 3.49, "learning_rate": 3.630526853881112e-06, "loss": 0.2351, "step": 26910 }, { "epoch": 3.49, "learning_rate": 3.6296859727691293e-06, "loss": 0.23, "step": 26920 }, { "epoch": 3.5, "learning_rate": 3.628844854774896e-06, "loss": 0.2371, "step": 26930 }, { "epoch": 3.5, "learning_rate": 3.628003500053354e-06, "loss": 0.2191, "step": 26940 }, { "epoch": 3.5, "learning_rate": 3.62716190875949e-06, "loss": 0.2259, "step": 26950 }, { "epoch": 3.5, "learning_rate": 3.6263200810483328e-06, "loss": 0.2343, "step": 26960 }, { "epoch": 3.5, "learning_rate": 3.6254780170749565e-06, "loss": 0.2487, "step": 26970 }, { "epoch": 3.5, "learning_rate": 3.624635716994476e-06, "loss": 0.2287, "step": 26980 }, { "epoch": 3.5, "learning_rate": 3.623793180962053e-06, "loss": 0.2267, "step": 26990 }, { "epoch": 3.5, "learning_rate": 3.6229504091328904e-06, "loss": 0.2249, "step": 27000 }, { "epoch": 3.51, "learning_rate": 3.6221074016622354e-06, "loss": 0.2193, "step": 27010 }, { "epoch": 3.51, "learning_rate": 3.6212641587053783e-06, "loss": 0.2309, "step": 27020 }, { "epoch": 3.51, "learning_rate": 3.6204206804176526e-06, "loss": 0.2397, "step": 27030 }, { "epoch": 3.51, "learning_rate": 3.619576966954436e-06, "loss": 0.2337, "step": 27040 }, { "epoch": 3.51, "learning_rate": 3.618733018471148e-06, "loss": 0.2292, "step": 27050 }, { "epoch": 3.51, "learning_rate": 3.6178888351232534e-06, "loss": 0.2313, "step": 27060 }, { "epoch": 3.51, "learning_rate": 3.6170444170662582e-06, "loss": 0.2371, "step": 27070 }, { "epoch": 3.51, "learning_rate": 3.616199764455714e-06, "loss": 0.2342, "step": 27080 }, { "epoch": 3.52, "learning_rate": 3.615354877447213e-06, "loss": 0.2191, "step": 27090 }, { "epoch": 3.52, "learning_rate": 3.6145097561963923e-06, "loss": 0.2372, "step": 27100 }, { "epoch": 3.52, "learning_rate": 3.613664400858931e-06, "loss": 0.2295, "step": 27110 }, { "epoch": 3.52, "learning_rate": 3.6128188115905536e-06, "loss": 0.2317, "step": 27120 }, { "epoch": 3.52, "learning_rate": 3.611972988547025e-06, "loss": 0.2266, "step": 27130 }, { "epoch": 3.52, "learning_rate": 3.6111269318841533e-06, "loss": 0.2417, "step": 27140 }, { "epoch": 3.52, "learning_rate": 3.610280641757793e-06, "loss": 0.2389, "step": 27150 }, { "epoch": 3.52, "learning_rate": 3.6094341183238372e-06, "loss": 0.2464, "step": 27160 }, { "epoch": 3.53, "learning_rate": 3.608587361738225e-06, "loss": 0.2202, "step": 27170 }, { "epoch": 3.53, "learning_rate": 3.607740372156937e-06, "loss": 0.2605, "step": 27180 }, { "epoch": 3.53, "learning_rate": 3.6068931497359964e-06, "loss": 0.2237, "step": 27190 }, { "epoch": 3.53, "learning_rate": 3.6060456946314718e-06, "loss": 0.246, "step": 27200 }, { "epoch": 3.53, "learning_rate": 3.6051980069994718e-06, "loss": 0.2263, "step": 27210 }, { "epoch": 3.53, "learning_rate": 3.6043500869961486e-06, "loss": 0.2366, "step": 27220 }, { "epoch": 3.53, "learning_rate": 3.603501934777698e-06, "loss": 0.2411, "step": 27230 }, { "epoch": 3.54, "learning_rate": 3.6026535505003586e-06, "loss": 0.2352, "step": 27240 }, { "epoch": 3.54, "learning_rate": 3.60180493432041e-06, "loss": 0.2127, "step": 27250 }, { "epoch": 3.54, "learning_rate": 3.6009560863941766e-06, "loss": 0.219, "step": 27260 }, { "epoch": 3.54, "learning_rate": 3.600107006878025e-06, "loss": 0.2292, "step": 27270 }, { "epoch": 3.54, "learning_rate": 3.5992576959283637e-06, "loss": 0.2242, "step": 27280 }, { "epoch": 3.54, "learning_rate": 3.598408153701644e-06, "loss": 0.2464, "step": 27290 }, { "epoch": 3.54, "learning_rate": 3.5975583803543608e-06, "loss": 0.2333, "step": 27300 }, { "epoch": 3.54, "learning_rate": 3.5967083760430506e-06, "loss": 0.231, "step": 27310 }, { "epoch": 3.55, "learning_rate": 3.5958581409242916e-06, "loss": 0.2277, "step": 27320 }, { "epoch": 3.55, "learning_rate": 3.595007675154707e-06, "loss": 0.2176, "step": 27330 }, { "epoch": 3.55, "learning_rate": 3.59415697889096e-06, "loss": 0.222, "step": 27340 }, { "epoch": 3.55, "learning_rate": 3.593306052289759e-06, "loss": 0.2355, "step": 27350 }, { "epoch": 3.55, "learning_rate": 3.592454895507851e-06, "loss": 0.2289, "step": 27360 }, { "epoch": 3.55, "learning_rate": 3.5916035087020286e-06, "loss": 0.2241, "step": 27370 }, { "epoch": 3.55, "learning_rate": 3.5907518920291265e-06, "loss": 0.2443, "step": 27380 }, { "epoch": 3.55, "learning_rate": 3.5899000456460204e-06, "loss": 0.2272, "step": 27390 }, { "epoch": 3.56, "learning_rate": 3.589047969709628e-06, "loss": 0.2223, "step": 27400 }, { "epoch": 3.56, "learning_rate": 3.5881956643769116e-06, "loss": 0.2343, "step": 27410 }, { "epoch": 3.56, "learning_rate": 3.5873431298048736e-06, "loss": 0.226, "step": 27420 }, { "epoch": 3.56, "learning_rate": 3.5864903661505596e-06, "loss": 0.232, "step": 27430 }, { "epoch": 3.56, "learning_rate": 3.5856373735710577e-06, "loss": 0.2262, "step": 27440 }, { "epoch": 3.56, "learning_rate": 3.584784152223497e-06, "loss": 0.2367, "step": 27450 }, { "epoch": 3.56, "learning_rate": 3.5839307022650488e-06, "loss": 0.2353, "step": 27460 }, { "epoch": 3.57, "learning_rate": 3.5830770238529277e-06, "loss": 0.2177, "step": 27470 }, { "epoch": 3.57, "learning_rate": 3.582223117144391e-06, "loss": 0.2239, "step": 27480 }, { "epoch": 3.57, "learning_rate": 3.5813689822967348e-06, "loss": 0.2266, "step": 27490 }, { "epoch": 3.57, "learning_rate": 3.5805146194673005e-06, "loss": 0.2272, "step": 27500 }, { "epoch": 3.57, "learning_rate": 3.5796600288134698e-06, "loss": 0.2362, "step": 27510 }, { "epoch": 3.57, "learning_rate": 3.5788052104926674e-06, "loss": 0.2371, "step": 27520 }, { "epoch": 3.57, "learning_rate": 3.577950164662358e-06, "loss": 0.2334, "step": 27530 }, { "epoch": 3.57, "learning_rate": 3.577094891480052e-06, "loss": 0.2248, "step": 27540 }, { "epoch": 3.58, "learning_rate": 3.576239391103297e-06, "loss": 0.2427, "step": 27550 }, { "epoch": 3.58, "learning_rate": 3.575383663689686e-06, "loss": 0.2392, "step": 27560 }, { "epoch": 3.58, "learning_rate": 3.5745277093968513e-06, "loss": 0.224, "step": 27570 }, { "epoch": 3.58, "learning_rate": 3.5736715283824695e-06, "loss": 0.2159, "step": 27580 }, { "epoch": 3.58, "learning_rate": 3.5728151208042575e-06, "loss": 0.2281, "step": 27590 }, { "epoch": 3.58, "learning_rate": 3.571958486819973e-06, "loss": 0.2253, "step": 27600 }, { "epoch": 3.58, "learning_rate": 3.5711016265874175e-06, "loss": 0.2358, "step": 27610 }, { "epoch": 3.58, "learning_rate": 3.570244540264434e-06, "loss": 0.2311, "step": 27620 }, { "epoch": 3.59, "learning_rate": 3.5693872280089048e-06, "loss": 0.2325, "step": 27630 }, { "epoch": 3.59, "learning_rate": 3.568529689978756e-06, "loss": 0.2239, "step": 27640 }, { "epoch": 3.59, "learning_rate": 3.5676719263319545e-06, "loss": 0.227, "step": 27650 }, { "epoch": 3.59, "learning_rate": 3.56681393722651e-06, "loss": 0.2181, "step": 27660 }, { "epoch": 3.59, "learning_rate": 3.565955722820471e-06, "loss": 0.2336, "step": 27670 }, { "epoch": 3.59, "learning_rate": 3.56509728327193e-06, "loss": 0.2321, "step": 27680 }, { "epoch": 3.59, "learning_rate": 3.564238618739021e-06, "loss": 0.2294, "step": 27690 }, { "epoch": 3.6, "learning_rate": 3.5633797293799173e-06, "loss": 0.2381, "step": 27700 }, { "epoch": 3.6, "learning_rate": 3.562520615352834e-06, "loss": 0.2327, "step": 27710 }, { "epoch": 3.6, "learning_rate": 3.5616612768160305e-06, "loss": 0.227, "step": 27720 }, { "epoch": 3.6, "learning_rate": 3.560801713927805e-06, "loss": 0.2431, "step": 27730 }, { "epoch": 3.6, "learning_rate": 3.5599419268464983e-06, "loss": 0.2418, "step": 27740 }, { "epoch": 3.6, "learning_rate": 3.5590819157304898e-06, "loss": 0.2227, "step": 27750 }, { "epoch": 3.6, "learning_rate": 3.5582216807382036e-06, "loss": 0.2329, "step": 27760 }, { "epoch": 3.6, "learning_rate": 3.5573612220281026e-06, "loss": 0.2226, "step": 27770 }, { "epoch": 3.61, "learning_rate": 3.556500539758693e-06, "loss": 0.2213, "step": 27780 }, { "epoch": 3.61, "learning_rate": 3.5556396340885195e-06, "loss": 0.2358, "step": 27790 }, { "epoch": 3.61, "learning_rate": 3.5547785051761715e-06, "loss": 0.231, "step": 27800 }, { "epoch": 3.61, "learning_rate": 3.5539171531802764e-06, "loss": 0.2269, "step": 27810 }, { "epoch": 3.61, "learning_rate": 3.553055578259504e-06, "loss": 0.23, "step": 27820 }, { "epoch": 3.61, "learning_rate": 3.5521937805725654e-06, "loss": 0.2231, "step": 27830 }, { "epoch": 3.61, "learning_rate": 3.5513317602782116e-06, "loss": 0.2211, "step": 27840 }, { "epoch": 3.61, "learning_rate": 3.5504695175352355e-06, "loss": 0.2248, "step": 27850 }, { "epoch": 3.62, "learning_rate": 3.5496070525024714e-06, "loss": 0.2284, "step": 27860 }, { "epoch": 3.62, "learning_rate": 3.548744365338794e-06, "loss": 0.2302, "step": 27870 }, { "epoch": 3.62, "learning_rate": 3.5478814562031177e-06, "loss": 0.2404, "step": 27880 }, { "epoch": 3.62, "learning_rate": 3.5470183252544e-06, "loss": 0.235, "step": 27890 }, { "epoch": 3.62, "learning_rate": 3.5461549726516383e-06, "loss": 0.232, "step": 27900 }, { "epoch": 3.62, "learning_rate": 3.5452913985538707e-06, "loss": 0.2316, "step": 27910 }, { "epoch": 3.62, "learning_rate": 3.5444276031201755e-06, "loss": 0.2461, "step": 27920 }, { "epoch": 3.62, "learning_rate": 3.5435635865096734e-06, "loss": 0.2358, "step": 27930 }, { "epoch": 3.63, "learning_rate": 3.5426993488815243e-06, "loss": 0.2384, "step": 27940 }, { "epoch": 3.63, "learning_rate": 3.5418348903949294e-06, "loss": 0.2329, "step": 27950 }, { "epoch": 3.63, "learning_rate": 3.5409702112091304e-06, "loss": 0.2307, "step": 27960 }, { "epoch": 3.63, "learning_rate": 3.540105311483411e-06, "loss": 0.2224, "step": 27970 }, { "epoch": 3.63, "learning_rate": 3.539240191377093e-06, "loss": 0.2354, "step": 27980 }, { "epoch": 3.63, "learning_rate": 3.5383748510495406e-06, "loss": 0.2297, "step": 27990 }, { "epoch": 3.63, "learning_rate": 3.5375092906601585e-06, "loss": 0.2194, "step": 28000 }, { "epoch": 3.64, "learning_rate": 3.536643510368391e-06, "loss": 0.2332, "step": 28010 }, { "epoch": 3.64, "learning_rate": 3.535777510333724e-06, "loss": 0.2341, "step": 28020 }, { "epoch": 3.64, "learning_rate": 3.534911290715683e-06, "loss": 0.2337, "step": 28030 }, { "epoch": 3.64, "learning_rate": 3.5340448516738347e-06, "loss": 0.2416, "step": 28040 }, { "epoch": 3.64, "learning_rate": 3.5331781933677854e-06, "loss": 0.2264, "step": 28050 }, { "epoch": 3.64, "learning_rate": 3.5323113159571816e-06, "loss": 0.2292, "step": 28060 }, { "epoch": 3.64, "learning_rate": 3.531444219601713e-06, "loss": 0.2379, "step": 28070 }, { "epoch": 3.64, "learning_rate": 3.5305769044611055e-06, "loss": 0.2355, "step": 28080 }, { "epoch": 3.65, "learning_rate": 3.529709370695127e-06, "loss": 0.232, "step": 28090 }, { "epoch": 3.65, "learning_rate": 3.528841618463587e-06, "loss": 0.2393, "step": 28100 }, { "epoch": 3.65, "learning_rate": 3.5279736479263336e-06, "loss": 0.24, "step": 28110 }, { "epoch": 3.65, "learning_rate": 3.5271054592432557e-06, "loss": 0.2236, "step": 28120 }, { "epoch": 3.65, "learning_rate": 3.5262370525742828e-06, "loss": 0.2241, "step": 28130 }, { "epoch": 3.65, "learning_rate": 3.525368428079383e-06, "loss": 0.2447, "step": 28140 }, { "epoch": 3.65, "learning_rate": 3.524499585918567e-06, "loss": 0.2441, "step": 28150 }, { "epoch": 3.65, "learning_rate": 3.523630526251882e-06, "loss": 0.2249, "step": 28160 }, { "epoch": 3.66, "learning_rate": 3.522761249239421e-06, "loss": 0.2303, "step": 28170 }, { "epoch": 3.66, "learning_rate": 3.52189175504131e-06, "loss": 0.2402, "step": 28180 }, { "epoch": 3.66, "learning_rate": 3.521022043817721e-06, "loss": 0.2288, "step": 28190 }, { "epoch": 3.66, "learning_rate": 3.520152115728861e-06, "loss": 0.2429, "step": 28200 }, { "epoch": 3.66, "learning_rate": 3.5192819709349824e-06, "loss": 0.2322, "step": 28210 }, { "epoch": 3.66, "learning_rate": 3.5184116095963723e-06, "loss": 0.2308, "step": 28220 }, { "epoch": 3.66, "learning_rate": 3.51754103187336e-06, "loss": 0.2286, "step": 28230 }, { "epoch": 3.67, "learning_rate": 3.5166702379263166e-06, "loss": 0.2408, "step": 28240 }, { "epoch": 3.67, "learning_rate": 3.515799227915649e-06, "loss": 0.2223, "step": 28250 }, { "epoch": 3.67, "learning_rate": 3.5149280020018073e-06, "loss": 0.2377, "step": 28260 }, { "epoch": 3.67, "learning_rate": 3.5140565603452785e-06, "loss": 0.2455, "step": 28270 }, { "epoch": 3.67, "learning_rate": 3.513184903106593e-06, "loss": 0.2289, "step": 28280 }, { "epoch": 3.67, "learning_rate": 3.5123130304463163e-06, "loss": 0.2283, "step": 28290 }, { "epoch": 3.67, "learning_rate": 3.511440942525058e-06, "loss": 0.2485, "step": 28300 }, { "epoch": 3.67, "learning_rate": 3.5105686395034647e-06, "loss": 0.2293, "step": 28310 }, { "epoch": 3.68, "learning_rate": 3.5096961215422233e-06, "loss": 0.2382, "step": 28320 }, { "epoch": 3.68, "learning_rate": 3.5088233888020596e-06, "loss": 0.2388, "step": 28330 }, { "epoch": 3.68, "learning_rate": 3.5079504414437413e-06, "loss": 0.2286, "step": 28340 }, { "epoch": 3.68, "learning_rate": 3.5070772796280734e-06, "loss": 0.2233, "step": 28350 }, { "epoch": 3.68, "learning_rate": 3.5062039035159e-06, "loss": 0.2164, "step": 28360 }, { "epoch": 3.68, "learning_rate": 3.5053303132681074e-06, "loss": 0.2266, "step": 28370 }, { "epoch": 3.68, "learning_rate": 3.504456509045618e-06, "loss": 0.2343, "step": 28380 }, { "epoch": 3.68, "learning_rate": 3.5035824910093965e-06, "loss": 0.2384, "step": 28390 }, { "epoch": 3.69, "learning_rate": 3.5027082593204448e-06, "loss": 0.2316, "step": 28400 }, { "epoch": 3.69, "learning_rate": 3.5018338141398055e-06, "loss": 0.2433, "step": 28410 }, { "epoch": 3.69, "learning_rate": 3.500959155628561e-06, "loss": 0.2404, "step": 28420 }, { "epoch": 3.69, "learning_rate": 3.5000842839478307e-06, "loss": 0.2216, "step": 28430 }, { "epoch": 3.69, "learning_rate": 3.4992091992587757e-06, "loss": 0.2143, "step": 28440 }, { "epoch": 3.69, "learning_rate": 3.498333901722595e-06, "loss": 0.2267, "step": 28450 }, { "epoch": 3.69, "learning_rate": 3.4974583915005265e-06, "loss": 0.2352, "step": 28460 }, { "epoch": 3.7, "learning_rate": 3.496582668753849e-06, "loss": 0.2405, "step": 28470 }, { "epoch": 3.7, "learning_rate": 3.495706733643879e-06, "loss": 0.2308, "step": 28480 }, { "epoch": 3.7, "learning_rate": 3.4948305863319723e-06, "loss": 0.221, "step": 28490 }, { "epoch": 3.7, "learning_rate": 3.4939542269795243e-06, "loss": 0.218, "step": 28500 }, { "epoch": 3.7, "learning_rate": 3.4930776557479686e-06, "loss": 0.2315, "step": 28510 }, { "epoch": 3.7, "learning_rate": 3.4922008727987796e-06, "loss": 0.2437, "step": 28520 }, { "epoch": 3.7, "learning_rate": 3.4913238782934676e-06, "loss": 0.2265, "step": 28530 }, { "epoch": 3.7, "learning_rate": 3.4904466723935853e-06, "loss": 0.2195, "step": 28540 }, { "epoch": 3.71, "learning_rate": 3.4895692552607222e-06, "loss": 0.232, "step": 28550 }, { "epoch": 3.71, "learning_rate": 3.4886916270565076e-06, "loss": 0.2292, "step": 28560 }, { "epoch": 3.71, "learning_rate": 3.4878137879426084e-06, "loss": 0.227, "step": 28570 }, { "epoch": 3.71, "learning_rate": 3.4869357380807327e-06, "loss": 0.2272, "step": 28580 }, { "epoch": 3.71, "learning_rate": 3.4860574776326254e-06, "loss": 0.2234, "step": 28590 }, { "epoch": 3.71, "learning_rate": 3.485179006760071e-06, "loss": 0.234, "step": 28600 }, { "epoch": 3.71, "learning_rate": 3.4843003256248923e-06, "loss": 0.248, "step": 28610 }, { "epoch": 3.71, "learning_rate": 3.4834214343889508e-06, "loss": 0.2239, "step": 28620 }, { "epoch": 3.72, "learning_rate": 3.4825423332141484e-06, "loss": 0.2172, "step": 28630 }, { "epoch": 3.72, "learning_rate": 3.481663022262423e-06, "loss": 0.2441, "step": 28640 }, { "epoch": 3.72, "learning_rate": 3.4807835016957532e-06, "loss": 0.2267, "step": 28650 }, { "epoch": 3.72, "learning_rate": 3.4799037716761552e-06, "loss": 0.2371, "step": 28660 }, { "epoch": 3.72, "learning_rate": 3.479023832365684e-06, "loss": 0.2261, "step": 28670 }, { "epoch": 3.72, "learning_rate": 3.4781436839264333e-06, "loss": 0.234, "step": 28680 }, { "epoch": 3.72, "learning_rate": 3.477263326520535e-06, "loss": 0.2317, "step": 28690 }, { "epoch": 3.72, "learning_rate": 3.4763827603101606e-06, "loss": 0.2153, "step": 28700 }, { "epoch": 3.73, "learning_rate": 3.4755019854575184e-06, "loss": 0.2341, "step": 28710 }, { "epoch": 3.73, "learning_rate": 3.474621002124855e-06, "loss": 0.2281, "step": 28720 }, { "epoch": 3.73, "learning_rate": 3.4737398104744592e-06, "loss": 0.247, "step": 28730 }, { "epoch": 3.73, "learning_rate": 3.472858410668652e-06, "loss": 0.2392, "step": 28740 }, { "epoch": 3.73, "learning_rate": 3.471976802869798e-06, "loss": 0.2196, "step": 28750 }, { "epoch": 3.73, "learning_rate": 3.4710949872402976e-06, "loss": 0.2373, "step": 28760 }, { "epoch": 3.73, "learning_rate": 3.4702129639425905e-06, "loss": 0.238, "step": 28770 }, { "epoch": 3.74, "learning_rate": 3.469330733139154e-06, "loss": 0.2407, "step": 28780 }, { "epoch": 3.74, "learning_rate": 3.4684482949925026e-06, "loss": 0.2323, "step": 28790 }, { "epoch": 3.74, "learning_rate": 3.467565649665193e-06, "loss": 0.2492, "step": 28800 }, { "epoch": 3.74, "learning_rate": 3.466682797319814e-06, "loss": 0.2415, "step": 28810 }, { "epoch": 3.74, "learning_rate": 3.4657997381189984e-06, "loss": 0.2371, "step": 28820 }, { "epoch": 3.74, "learning_rate": 3.4649164722254134e-06, "loss": 0.232, "step": 28830 }, { "epoch": 3.74, "learning_rate": 3.4640329998017656e-06, "loss": 0.228, "step": 28840 }, { "epoch": 3.74, "learning_rate": 3.463149321010799e-06, "loss": 0.2395, "step": 28850 }, { "epoch": 3.75, "learning_rate": 3.4622654360152964e-06, "loss": 0.2344, "step": 28860 }, { "epoch": 3.75, "learning_rate": 3.4613813449780788e-06, "loss": 0.223, "step": 28870 }, { "epoch": 3.75, "learning_rate": 3.4604970480620037e-06, "loss": 0.2331, "step": 28880 }, { "epoch": 3.75, "learning_rate": 3.459612545429968e-06, "loss": 0.2258, "step": 28890 }, { "epoch": 3.75, "learning_rate": 3.4587278372449057e-06, "loss": 0.2239, "step": 28900 }, { "epoch": 3.75, "learning_rate": 3.4578429236697885e-06, "loss": 0.2297, "step": 28910 }, { "epoch": 3.75, "learning_rate": 3.456957804867627e-06, "loss": 0.2303, "step": 28920 }, { "epoch": 3.75, "learning_rate": 3.456072481001469e-06, "loss": 0.229, "step": 28930 }, { "epoch": 3.76, "learning_rate": 3.4551869522344e-06, "loss": 0.2342, "step": 28940 }, { "epoch": 3.76, "learning_rate": 3.4543012187295415e-06, "loss": 0.2288, "step": 28950 }, { "epoch": 3.76, "learning_rate": 3.4534152806500565e-06, "loss": 0.2272, "step": 28960 }, { "epoch": 3.76, "learning_rate": 3.4525291381591435e-06, "loss": 0.2289, "step": 28970 }, { "epoch": 3.76, "learning_rate": 3.451642791420038e-06, "loss": 0.2263, "step": 28980 }, { "epoch": 3.76, "learning_rate": 3.4507562405960145e-06, "loss": 0.2408, "step": 28990 }, { "epoch": 3.76, "learning_rate": 3.4498694858503833e-06, "loss": 0.2185, "step": 29000 }, { "epoch": 3.77, "learning_rate": 3.4489825273464955e-06, "loss": 0.2477, "step": 29010 }, { "epoch": 3.77, "learning_rate": 3.448095365247736e-06, "loss": 0.2348, "step": 29020 }, { "epoch": 3.77, "learning_rate": 3.4472079997175298e-06, "loss": 0.2413, "step": 29030 }, { "epoch": 3.77, "learning_rate": 3.4463204309193384e-06, "loss": 0.2321, "step": 29040 }, { "epoch": 3.77, "learning_rate": 3.4454326590166608e-06, "loss": 0.2395, "step": 29050 }, { "epoch": 3.77, "learning_rate": 3.4445446841730324e-06, "loss": 0.2406, "step": 29060 }, { "epoch": 3.77, "learning_rate": 3.443656506552029e-06, "loss": 0.2453, "step": 29070 }, { "epoch": 3.77, "learning_rate": 3.442768126317261e-06, "loss": 0.2419, "step": 29080 }, { "epoch": 3.78, "learning_rate": 3.4418795436323754e-06, "loss": 0.2383, "step": 29090 }, { "epoch": 3.78, "learning_rate": 3.44099075866106e-06, "loss": 0.218, "step": 29100 }, { "epoch": 3.78, "learning_rate": 3.4401017715670374e-06, "loss": 0.2358, "step": 29110 }, { "epoch": 3.78, "learning_rate": 3.4392125825140674e-06, "loss": 0.2383, "step": 29120 }, { "epoch": 3.78, "learning_rate": 3.4383231916659476e-06, "loss": 0.2253, "step": 29130 }, { "epoch": 3.78, "learning_rate": 3.4374335991865128e-06, "loss": 0.2415, "step": 29140 }, { "epoch": 3.78, "learning_rate": 3.4365438052396354e-06, "loss": 0.2217, "step": 29150 }, { "epoch": 3.78, "learning_rate": 3.4356538099892236e-06, "loss": 0.2335, "step": 29160 }, { "epoch": 3.79, "learning_rate": 3.4347636135992235e-06, "loss": 0.2392, "step": 29170 }, { "epoch": 3.79, "learning_rate": 3.4338732162336186e-06, "loss": 0.241, "step": 29180 }, { "epoch": 3.79, "learning_rate": 3.4329826180564283e-06, "loss": 0.2256, "step": 29190 }, { "epoch": 3.79, "learning_rate": 3.43209181923171e-06, "loss": 0.2315, "step": 29200 }, { "epoch": 3.79, "learning_rate": 3.4312008199235577e-06, "loss": 0.2324, "step": 29210 }, { "epoch": 3.79, "learning_rate": 3.4303096202961026e-06, "loss": 0.2368, "step": 29220 }, { "epoch": 3.79, "learning_rate": 3.4294182205135126e-06, "loss": 0.2328, "step": 29230 }, { "epoch": 3.79, "learning_rate": 3.428526620739992e-06, "loss": 0.2223, "step": 29240 }, { "epoch": 3.8, "learning_rate": 3.427634821139783e-06, "loss": 0.226, "step": 29250 }, { "epoch": 3.8, "learning_rate": 3.4267428218771636e-06, "loss": 0.2505, "step": 29260 }, { "epoch": 3.8, "learning_rate": 3.4258506231164486e-06, "loss": 0.231, "step": 29270 }, { "epoch": 3.8, "learning_rate": 3.42495822502199e-06, "loss": 0.2401, "step": 29280 }, { "epoch": 3.8, "learning_rate": 3.424065627758178e-06, "loss": 0.2211, "step": 29290 }, { "epoch": 3.8, "learning_rate": 3.4231728314894363e-06, "loss": 0.2327, "step": 29300 }, { "epoch": 3.8, "learning_rate": 3.4222798363802267e-06, "loss": 0.2302, "step": 29310 }, { "epoch": 3.81, "learning_rate": 3.4213866425950495e-06, "loss": 0.243, "step": 29320 }, { "epoch": 3.81, "learning_rate": 3.420493250298438e-06, "loss": 0.2314, "step": 29330 }, { "epoch": 3.81, "learning_rate": 3.4195996596549652e-06, "loss": 0.2206, "step": 29340 }, { "epoch": 3.81, "learning_rate": 3.418705870829239e-06, "loss": 0.2298, "step": 29350 }, { "epoch": 3.81, "learning_rate": 3.417811883985904e-06, "loss": 0.2289, "step": 29360 }, { "epoch": 3.81, "learning_rate": 3.4169176992896423e-06, "loss": 0.2392, "step": 29370 }, { "epoch": 3.81, "learning_rate": 3.416023316905171e-06, "loss": 0.228, "step": 29380 }, { "epoch": 3.81, "learning_rate": 3.4151287369972443e-06, "loss": 0.2193, "step": 29390 }, { "epoch": 3.82, "learning_rate": 3.4142339597306526e-06, "loss": 0.2352, "step": 29400 }, { "epoch": 3.82, "learning_rate": 3.4133389852702237e-06, "loss": 0.2432, "step": 29410 }, { "epoch": 3.82, "learning_rate": 3.4124438137808193e-06, "loss": 0.2146, "step": 29420 }, { "epoch": 3.82, "learning_rate": 3.411548445427341e-06, "loss": 0.236, "step": 29430 }, { "epoch": 3.82, "learning_rate": 3.4106528803747223e-06, "loss": 0.2212, "step": 29440 }, { "epoch": 3.82, "learning_rate": 3.409757118787937e-06, "loss": 0.2237, "step": 29450 }, { "epoch": 3.82, "learning_rate": 3.408861160831993e-06, "loss": 0.2354, "step": 29460 }, { "epoch": 3.82, "learning_rate": 3.407965006671934e-06, "loss": 0.2345, "step": 29470 }, { "epoch": 3.83, "learning_rate": 3.4070686564728402e-06, "loss": 0.2403, "step": 29480 }, { "epoch": 3.83, "learning_rate": 3.40617211039983e-06, "loss": 0.2268, "step": 29490 }, { "epoch": 3.83, "learning_rate": 3.405275368618055e-06, "loss": 0.2209, "step": 29500 }, { "epoch": 3.83, "learning_rate": 3.4043784312927043e-06, "loss": 0.2416, "step": 29510 }, { "epoch": 3.83, "learning_rate": 3.403481298589002e-06, "loss": 0.239, "step": 29520 }, { "epoch": 3.83, "learning_rate": 3.40258397067221e-06, "loss": 0.2317, "step": 29530 }, { "epoch": 3.83, "learning_rate": 3.4016864477076245e-06, "loss": 0.24, "step": 29540 }, { "epoch": 3.84, "learning_rate": 3.4007887298605776e-06, "loss": 0.2353, "step": 29550 }, { "epoch": 3.84, "learning_rate": 3.399890817296439e-06, "loss": 0.2338, "step": 29560 }, { "epoch": 3.84, "learning_rate": 3.398992710180613e-06, "loss": 0.2264, "step": 29570 }, { "epoch": 3.84, "learning_rate": 3.3980944086785388e-06, "loss": 0.2355, "step": 29580 }, { "epoch": 3.84, "learning_rate": 3.397195912955694e-06, "loss": 0.2083, "step": 29590 }, { "epoch": 3.84, "learning_rate": 3.3962972231775895e-06, "loss": 0.2232, "step": 29600 }, { "epoch": 3.84, "learning_rate": 3.395398339509773e-06, "loss": 0.26, "step": 29610 }, { "epoch": 3.84, "learning_rate": 3.394499262117828e-06, "loss": 0.2391, "step": 29620 }, { "epoch": 3.85, "learning_rate": 3.393599991167374e-06, "loss": 0.2305, "step": 29630 }, { "epoch": 3.85, "learning_rate": 3.3927005268240654e-06, "loss": 0.2127, "step": 29640 }, { "epoch": 3.85, "learning_rate": 3.3918008692535923e-06, "loss": 0.2371, "step": 29650 }, { "epoch": 3.85, "learning_rate": 3.39090101862168e-06, "loss": 0.2382, "step": 29660 }, { "epoch": 3.85, "learning_rate": 3.390000975094092e-06, "loss": 0.2251, "step": 29670 }, { "epoch": 3.85, "learning_rate": 3.3891007388366227e-06, "loss": 0.2386, "step": 29680 }, { "epoch": 3.85, "learning_rate": 3.3882003100151065e-06, "loss": 0.2055, "step": 29690 }, { "epoch": 3.85, "learning_rate": 3.387299688795411e-06, "loss": 0.2143, "step": 29700 }, { "epoch": 3.86, "learning_rate": 3.386398875343439e-06, "loss": 0.228, "step": 29710 }, { "epoch": 3.86, "learning_rate": 3.3854978698251298e-06, "loss": 0.2325, "step": 29720 }, { "epoch": 3.86, "learning_rate": 3.3845966724064576e-06, "loss": 0.2397, "step": 29730 }, { "epoch": 3.86, "learning_rate": 3.3836952832534325e-06, "loss": 0.2448, "step": 29740 }, { "epoch": 3.86, "learning_rate": 3.3827937025320983e-06, "loss": 0.2405, "step": 29750 }, { "epoch": 3.86, "learning_rate": 3.3818919304085353e-06, "loss": 0.2283, "step": 29760 }, { "epoch": 3.86, "learning_rate": 3.3809899670488603e-06, "loss": 0.2294, "step": 29770 }, { "epoch": 3.87, "learning_rate": 3.3800878126192225e-06, "loss": 0.2384, "step": 29780 }, { "epoch": 3.87, "learning_rate": 3.379185467285808e-06, "loss": 0.2338, "step": 29790 }, { "epoch": 3.87, "learning_rate": 3.378282931214838e-06, "loss": 0.2294, "step": 29800 }, { "epoch": 3.87, "learning_rate": 3.3773802045725695e-06, "loss": 0.2301, "step": 29810 }, { "epoch": 3.87, "learning_rate": 3.3764772875252922e-06, "loss": 0.2266, "step": 29820 }, { "epoch": 3.87, "learning_rate": 3.3755741802393332e-06, "loss": 0.2334, "step": 29830 }, { "epoch": 3.87, "learning_rate": 3.3746708828810546e-06, "loss": 0.2476, "step": 29840 }, { "epoch": 3.87, "learning_rate": 3.373767395616852e-06, "loss": 0.2269, "step": 29850 }, { "epoch": 3.88, "learning_rate": 3.3728637186131564e-06, "loss": 0.2394, "step": 29860 }, { "epoch": 3.88, "learning_rate": 3.3719598520364344e-06, "loss": 0.2258, "step": 29870 }, { "epoch": 3.88, "learning_rate": 3.371055796053188e-06, "loss": 0.2327, "step": 29880 }, { "epoch": 3.88, "learning_rate": 3.3701515508299527e-06, "loss": 0.2311, "step": 29890 }, { "epoch": 3.88, "learning_rate": 3.3692471165332997e-06, "loss": 0.231, "step": 29900 }, { "epoch": 3.88, "learning_rate": 3.368342493329835e-06, "loss": 0.2396, "step": 29910 }, { "epoch": 3.88, "learning_rate": 3.3674376813861983e-06, "loss": 0.2194, "step": 29920 }, { "epoch": 3.88, "learning_rate": 3.366532680869066e-06, "loss": 0.229, "step": 29930 }, { "epoch": 3.89, "learning_rate": 3.3656274919451476e-06, "loss": 0.2301, "step": 29940 }, { "epoch": 3.89, "learning_rate": 3.364722114781189e-06, "loss": 0.2234, "step": 29950 }, { "epoch": 3.89, "learning_rate": 3.363816549543968e-06, "loss": 0.2402, "step": 29960 }, { "epoch": 3.89, "learning_rate": 3.3629107964003004e-06, "loss": 0.225, "step": 29970 }, { "epoch": 3.89, "learning_rate": 3.3620048555170345e-06, "loss": 0.2224, "step": 29980 }, { "epoch": 3.89, "learning_rate": 3.361098727061053e-06, "loss": 0.2249, "step": 29990 }, { "epoch": 3.89, "learning_rate": 3.3601924111992744e-06, "loss": 0.2231, "step": 30000 }, { "epoch": 3.89, "learning_rate": 3.359285908098651e-06, "loss": 0.2335, "step": 30010 }, { "epoch": 3.9, "learning_rate": 3.3583792179261698e-06, "loss": 0.2376, "step": 30020 }, { "epoch": 3.9, "learning_rate": 3.357472340848852e-06, "loss": 0.2392, "step": 30030 }, { "epoch": 3.9, "learning_rate": 3.3565652770337533e-06, "loss": 0.2341, "step": 30040 }, { "epoch": 3.9, "learning_rate": 3.355658026647965e-06, "loss": 0.2304, "step": 30050 }, { "epoch": 3.9, "learning_rate": 3.35475058985861e-06, "loss": 0.2388, "step": 30060 }, { "epoch": 3.9, "learning_rate": 3.353842966832848e-06, "loss": 0.231, "step": 30070 }, { "epoch": 3.9, "learning_rate": 3.352935157737873e-06, "loss": 0.2266, "step": 30080 }, { "epoch": 3.91, "learning_rate": 3.352027162740911e-06, "loss": 0.2353, "step": 30090 }, { "epoch": 3.91, "learning_rate": 3.3511189820092246e-06, "loss": 0.2324, "step": 30100 }, { "epoch": 3.91, "learning_rate": 3.35021061571011e-06, "loss": 0.2398, "step": 30110 }, { "epoch": 3.91, "learning_rate": 3.349302064010898e-06, "loss": 0.2148, "step": 30120 }, { "epoch": 3.91, "learning_rate": 3.3483933270789503e-06, "loss": 0.2448, "step": 30130 }, { "epoch": 3.91, "learning_rate": 3.3474844050816677e-06, "loss": 0.2304, "step": 30140 }, { "epoch": 3.91, "learning_rate": 3.346575298186482e-06, "loss": 0.2259, "step": 30150 }, { "epoch": 3.91, "learning_rate": 3.3456660065608596e-06, "loss": 0.2287, "step": 30160 }, { "epoch": 3.92, "learning_rate": 3.3447565303723013e-06, "loss": 0.2269, "step": 30170 }, { "epoch": 3.92, "learning_rate": 3.3438468697883418e-06, "loss": 0.2174, "step": 30180 }, { "epoch": 3.92, "learning_rate": 3.342937024976549e-06, "loss": 0.2264, "step": 30190 }, { "epoch": 3.92, "learning_rate": 3.342026996104526e-06, "loss": 0.2261, "step": 30200 }, { "epoch": 3.92, "learning_rate": 3.3411167833399097e-06, "loss": 0.24, "step": 30210 }, { "epoch": 3.92, "learning_rate": 3.340206386850369e-06, "loss": 0.2283, "step": 30220 }, { "epoch": 3.92, "learning_rate": 3.3392958068036096e-06, "loss": 0.2215, "step": 30230 }, { "epoch": 3.92, "learning_rate": 3.338385043367368e-06, "loss": 0.2294, "step": 30240 }, { "epoch": 3.93, "learning_rate": 3.337474096709417e-06, "loss": 0.2259, "step": 30250 }, { "epoch": 3.93, "learning_rate": 3.3365629669975622e-06, "loss": 0.2196, "step": 30260 }, { "epoch": 3.93, "learning_rate": 3.3356516543996414e-06, "loss": 0.2353, "step": 30270 }, { "epoch": 3.93, "learning_rate": 3.3347401590835286e-06, "loss": 0.2316, "step": 30280 }, { "epoch": 3.93, "learning_rate": 3.3338284812171297e-06, "loss": 0.2343, "step": 30290 }, { "epoch": 3.93, "learning_rate": 3.3329166209683857e-06, "loss": 0.2357, "step": 30300 }, { "epoch": 3.93, "learning_rate": 3.3320045785052695e-06, "loss": 0.2361, "step": 30310 }, { "epoch": 3.94, "learning_rate": 3.331092353995789e-06, "loss": 0.2311, "step": 30320 }, { "epoch": 3.94, "learning_rate": 3.3301799476079855e-06, "loss": 0.2292, "step": 30330 }, { "epoch": 3.94, "learning_rate": 3.3292673595099324e-06, "loss": 0.2348, "step": 30340 }, { "epoch": 3.94, "learning_rate": 3.3283545898697373e-06, "loss": 0.2265, "step": 30350 }, { "epoch": 3.94, "learning_rate": 3.327441638855543e-06, "loss": 0.2297, "step": 30360 }, { "epoch": 3.94, "learning_rate": 3.3265285066355226e-06, "loss": 0.2402, "step": 30370 }, { "epoch": 3.94, "learning_rate": 3.3256151933778855e-06, "loss": 0.2299, "step": 30380 }, { "epoch": 3.94, "learning_rate": 3.3247016992508717e-06, "loss": 0.2499, "step": 30390 }, { "epoch": 3.95, "learning_rate": 3.323788024422758e-06, "loss": 0.2433, "step": 30400 }, { "epoch": 3.95, "learning_rate": 3.32287416906185e-06, "loss": 0.2252, "step": 30410 }, { "epoch": 3.95, "learning_rate": 3.3219601333364904e-06, "loss": 0.2257, "step": 30420 }, { "epoch": 3.95, "learning_rate": 3.321045917415054e-06, "loss": 0.2341, "step": 30430 }, { "epoch": 3.95, "learning_rate": 3.320131521465948e-06, "loss": 0.2173, "step": 30440 }, { "epoch": 3.95, "learning_rate": 3.3192169456576125e-06, "loss": 0.2525, "step": 30450 }, { "epoch": 3.95, "learning_rate": 3.3183021901585224e-06, "loss": 0.2271, "step": 30460 }, { "epoch": 3.95, "learning_rate": 3.3173872551371853e-06, "loss": 0.2142, "step": 30470 }, { "epoch": 3.96, "learning_rate": 3.31647214076214e-06, "loss": 0.2296, "step": 30480 }, { "epoch": 3.96, "learning_rate": 3.3155568472019605e-06, "loss": 0.2241, "step": 30490 }, { "epoch": 3.96, "learning_rate": 3.3146413746252534e-06, "loss": 0.2301, "step": 30500 }, { "epoch": 3.96, "learning_rate": 3.3137257232006574e-06, "loss": 0.2149, "step": 30510 }, { "epoch": 3.96, "learning_rate": 3.312809893096844e-06, "loss": 0.2165, "step": 30520 }, { "epoch": 3.96, "learning_rate": 3.3118938844825194e-06, "loss": 0.2212, "step": 30530 }, { "epoch": 3.96, "learning_rate": 3.3109776975264213e-06, "loss": 0.2166, "step": 30540 }, { "epoch": 3.96, "learning_rate": 3.3100613323973196e-06, "loss": 0.2177, "step": 30550 }, { "epoch": 3.97, "learning_rate": 3.3091447892640184e-06, "loss": 0.2276, "step": 30560 }, { "epoch": 3.97, "learning_rate": 3.308228068295355e-06, "loss": 0.2385, "step": 30570 }, { "epoch": 3.97, "learning_rate": 3.3073111696601976e-06, "loss": 0.2426, "step": 30580 }, { "epoch": 3.97, "learning_rate": 3.3063940935274486e-06, "loss": 0.2312, "step": 30590 }, { "epoch": 3.97, "learning_rate": 3.3054768400660417e-06, "loss": 0.215, "step": 30600 }, { "epoch": 3.97, "learning_rate": 3.3045594094449453e-06, "loss": 0.2243, "step": 30610 }, { "epoch": 3.97, "learning_rate": 3.303641801833158e-06, "loss": 0.2286, "step": 30620 }, { "epoch": 3.98, "learning_rate": 3.3027240173997135e-06, "loss": 0.2359, "step": 30630 }, { "epoch": 3.98, "learning_rate": 3.301806056313676e-06, "loss": 0.2177, "step": 30640 }, { "epoch": 3.98, "learning_rate": 3.3008879187441443e-06, "loss": 0.2373, "step": 30650 }, { "epoch": 3.98, "learning_rate": 3.2999696048602467e-06, "loss": 0.2277, "step": 30660 }, { "epoch": 3.98, "learning_rate": 3.299051114831147e-06, "loss": 0.2364, "step": 30670 }, { "epoch": 3.98, "learning_rate": 3.2981324488260407e-06, "loss": 0.2371, "step": 30680 }, { "epoch": 3.98, "learning_rate": 3.2972136070141534e-06, "loss": 0.224, "step": 30690 }, { "epoch": 3.98, "learning_rate": 3.2962945895647468e-06, "loss": 0.2223, "step": 30700 }, { "epoch": 3.99, "learning_rate": 3.2953753966471124e-06, "loss": 0.2321, "step": 30710 }, { "epoch": 3.99, "learning_rate": 3.294456028430575e-06, "loss": 0.2291, "step": 30720 }, { "epoch": 3.99, "learning_rate": 3.2935364850844903e-06, "loss": 0.2324, "step": 30730 }, { "epoch": 3.99, "learning_rate": 3.2926167667782484e-06, "loss": 0.216, "step": 30740 }, { "epoch": 3.99, "learning_rate": 3.291696873681271e-06, "loss": 0.2189, "step": 30750 }, { "epoch": 3.99, "learning_rate": 3.2907768059630108e-06, "loss": 0.2312, "step": 30760 }, { "epoch": 3.99, "learning_rate": 3.289856563792953e-06, "loss": 0.2433, "step": 30770 }, { "epoch": 3.99, "learning_rate": 3.288936147340617e-06, "loss": 0.2251, "step": 30780 }, { "epoch": 4.0, "learning_rate": 3.288015556775551e-06, "loss": 0.2434, "step": 30790 }, { "epoch": 4.0, "learning_rate": 3.287094792267338e-06, "loss": 0.2437, "step": 30800 }, { "epoch": 4.0, "learning_rate": 3.2861738539855918e-06, "loss": 0.232, "step": 30810 }, { "epoch": 4.0, "learning_rate": 3.2852527420999584e-06, "loss": 0.2202, "step": 30820 }, { "epoch": 4.0, "eval_loss": 0.2305249571800232, "eval_runtime": 60.8144, "eval_samples_per_second": 178.856, "eval_steps_per_second": 22.363, "step": 30820 }, { "epoch": 4.0, "learning_rate": 3.284331456780115e-06, "loss": 0.2428, "step": 30830 }, { "epoch": 4.0, "learning_rate": 3.2834099981957733e-06, "loss": 0.2206, "step": 30840 }, { "epoch": 4.0, "learning_rate": 3.282488366516674e-06, "loss": 0.2279, "step": 30850 }, { "epoch": 4.01, "learning_rate": 3.2815665619125905e-06, "loss": 0.2437, "step": 30860 }, { "epoch": 4.01, "learning_rate": 3.2806445845533297e-06, "loss": 0.2251, "step": 30870 }, { "epoch": 4.01, "learning_rate": 3.279722434608728e-06, "loss": 0.2233, "step": 30880 }, { "epoch": 4.01, "learning_rate": 3.2788001122486552e-06, "loss": 0.2283, "step": 30890 }, { "epoch": 4.01, "learning_rate": 3.277877617643012e-06, "loss": 0.2408, "step": 30900 }, { "epoch": 4.01, "learning_rate": 3.276954950961731e-06, "loss": 0.2337, "step": 30910 }, { "epoch": 4.01, "learning_rate": 3.276032112374777e-06, "loss": 0.2354, "step": 30920 }, { "epoch": 4.01, "learning_rate": 3.275109102052146e-06, "loss": 0.2213, "step": 30930 }, { "epoch": 4.02, "learning_rate": 3.2741859201638654e-06, "loss": 0.2251, "step": 30940 }, { "epoch": 4.02, "learning_rate": 3.273262566879994e-06, "loss": 0.2118, "step": 30950 }, { "epoch": 4.02, "learning_rate": 3.2723390423706247e-06, "loss": 0.2295, "step": 30960 }, { "epoch": 4.02, "learning_rate": 3.271415346805878e-06, "loss": 0.2218, "step": 30970 }, { "epoch": 4.02, "learning_rate": 3.2704914803559087e-06, "loss": 0.2237, "step": 30980 }, { "epoch": 4.02, "learning_rate": 3.269567443190902e-06, "loss": 0.2278, "step": 30990 }, { "epoch": 4.02, "learning_rate": 3.2686432354810746e-06, "loss": 0.2389, "step": 31000 }, { "epoch": 4.02, "learning_rate": 3.2677188573966755e-06, "loss": 0.2317, "step": 31010 }, { "epoch": 4.03, "learning_rate": 3.266794309107983e-06, "loss": 0.2357, "step": 31020 }, { "epoch": 4.03, "learning_rate": 3.26586959078531e-06, "loss": 0.2299, "step": 31030 }, { "epoch": 4.03, "learning_rate": 3.264944702598997e-06, "loss": 0.2402, "step": 31040 }, { "epoch": 4.03, "learning_rate": 3.2640196447194187e-06, "loss": 0.2409, "step": 31050 }, { "epoch": 4.03, "learning_rate": 3.2630944173169806e-06, "loss": 0.2395, "step": 31060 }, { "epoch": 4.03, "learning_rate": 3.2621690205621175e-06, "loss": 0.2363, "step": 31070 }, { "epoch": 4.03, "learning_rate": 3.261243454625297e-06, "loss": 0.2113, "step": 31080 }, { "epoch": 4.04, "learning_rate": 3.260317719677018e-06, "loss": 0.2242, "step": 31090 }, { "epoch": 4.04, "learning_rate": 3.25939181588781e-06, "loss": 0.2503, "step": 31100 }, { "epoch": 4.04, "learning_rate": 3.2584657434282338e-06, "loss": 0.2148, "step": 31110 }, { "epoch": 4.04, "learning_rate": 3.257539502468881e-06, "loss": 0.2398, "step": 31120 }, { "epoch": 4.04, "learning_rate": 3.256613093180375e-06, "loss": 0.2277, "step": 31130 }, { "epoch": 4.04, "learning_rate": 3.2556865157333682e-06, "loss": 0.2254, "step": 31140 }, { "epoch": 4.04, "learning_rate": 3.254759770298547e-06, "loss": 0.218, "step": 31150 }, { "epoch": 4.04, "learning_rate": 3.2538328570466264e-06, "loss": 0.2262, "step": 31160 }, { "epoch": 4.05, "learning_rate": 3.2529057761483536e-06, "loss": 0.223, "step": 31170 }, { "epoch": 4.05, "learning_rate": 3.251978527774506e-06, "loss": 0.2189, "step": 31180 }, { "epoch": 4.05, "learning_rate": 3.251051112095891e-06, "loss": 0.2489, "step": 31190 }, { "epoch": 4.05, "learning_rate": 3.2501235292833496e-06, "loss": 0.2267, "step": 31200 }, { "epoch": 4.05, "learning_rate": 3.24919577950775e-06, "loss": 0.2319, "step": 31210 }, { "epoch": 4.05, "learning_rate": 3.2482678629399944e-06, "loss": 0.2351, "step": 31220 }, { "epoch": 4.05, "learning_rate": 3.247339779751013e-06, "loss": 0.2426, "step": 31230 }, { "epoch": 4.05, "learning_rate": 3.24641153011177e-06, "loss": 0.2273, "step": 31240 }, { "epoch": 4.06, "learning_rate": 3.245483114193256e-06, "loss": 0.2345, "step": 31250 }, { "epoch": 4.06, "learning_rate": 3.244554532166496e-06, "loss": 0.2436, "step": 31260 }, { "epoch": 4.06, "learning_rate": 3.243625784202543e-06, "loss": 0.2223, "step": 31270 }, { "epoch": 4.06, "learning_rate": 3.2426968704724828e-06, "loss": 0.2362, "step": 31280 }, { "epoch": 4.06, "learning_rate": 3.2417677911474292e-06, "loss": 0.233, "step": 31290 }, { "epoch": 4.06, "learning_rate": 3.2408385463985292e-06, "loss": 0.2286, "step": 31300 }, { "epoch": 4.06, "learning_rate": 3.239909136396959e-06, "loss": 0.2396, "step": 31310 }, { "epoch": 4.06, "learning_rate": 3.2389795613139237e-06, "loss": 0.2335, "step": 31320 }, { "epoch": 4.07, "learning_rate": 3.238049821320662e-06, "loss": 0.2228, "step": 31330 }, { "epoch": 4.07, "learning_rate": 3.2371199165884403e-06, "loss": 0.2215, "step": 31340 }, { "epoch": 4.07, "learning_rate": 3.2361898472885568e-06, "loss": 0.2189, "step": 31350 }, { "epoch": 4.07, "learning_rate": 3.2352596135923387e-06, "loss": 0.2344, "step": 31360 }, { "epoch": 4.07, "learning_rate": 3.2343292156711456e-06, "loss": 0.235, "step": 31370 }, { "epoch": 4.07, "learning_rate": 3.233398653696366e-06, "loss": 0.2207, "step": 31380 }, { "epoch": 4.07, "learning_rate": 3.232467927839417e-06, "loss": 0.2238, "step": 31390 }, { "epoch": 4.08, "learning_rate": 3.2315370382717495e-06, "loss": 0.2242, "step": 31400 }, { "epoch": 4.08, "learning_rate": 3.2306059851648424e-06, "loss": 0.2368, "step": 31410 }, { "epoch": 4.08, "learning_rate": 3.229674768690204e-06, "loss": 0.2354, "step": 31420 }, { "epoch": 4.08, "learning_rate": 3.2287433890193736e-06, "loss": 0.229, "step": 31430 }, { "epoch": 4.08, "learning_rate": 3.227811846323922e-06, "loss": 0.2276, "step": 31440 }, { "epoch": 4.08, "learning_rate": 3.2268801407754484e-06, "loss": 0.2299, "step": 31450 }, { "epoch": 4.08, "learning_rate": 3.2259482725455808e-06, "loss": 0.2315, "step": 31460 }, { "epoch": 4.08, "learning_rate": 3.2250162418059798e-06, "loss": 0.2254, "step": 31470 }, { "epoch": 4.09, "learning_rate": 3.2240840487283356e-06, "loss": 0.2406, "step": 31480 }, { "epoch": 4.09, "learning_rate": 3.223151693484365e-06, "loss": 0.2246, "step": 31490 }, { "epoch": 4.09, "learning_rate": 3.22221917624582e-06, "loss": 0.2418, "step": 31500 }, { "epoch": 4.09, "learning_rate": 3.221286497184477e-06, "loss": 0.2259, "step": 31510 }, { "epoch": 4.09, "learning_rate": 3.220353656472147e-06, "loss": 0.2205, "step": 31520 }, { "epoch": 4.09, "learning_rate": 3.2194206542806675e-06, "loss": 0.2413, "step": 31530 }, { "epoch": 4.09, "learning_rate": 3.2184874907819066e-06, "loss": 0.2237, "step": 31540 }, { "epoch": 4.09, "learning_rate": 3.2175541661477627e-06, "loss": 0.2271, "step": 31550 }, { "epoch": 4.1, "learning_rate": 3.2166206805501635e-06, "loss": 0.2376, "step": 31560 }, { "epoch": 4.1, "learning_rate": 3.2156870341610664e-06, "loss": 0.2224, "step": 31570 }, { "epoch": 4.1, "learning_rate": 3.214753227152458e-06, "loss": 0.2256, "step": 31580 }, { "epoch": 4.1, "learning_rate": 3.2138192596963557e-06, "loss": 0.2329, "step": 31590 }, { "epoch": 4.1, "learning_rate": 3.212885131964805e-06, "loss": 0.2391, "step": 31600 }, { "epoch": 4.1, "learning_rate": 3.2119508441298814e-06, "loss": 0.2236, "step": 31610 }, { "epoch": 4.1, "learning_rate": 3.211016396363691e-06, "loss": 0.244, "step": 31620 }, { "epoch": 4.11, "learning_rate": 3.210081788838367e-06, "loss": 0.2188, "step": 31630 }, { "epoch": 4.11, "learning_rate": 3.2091470217260746e-06, "loss": 0.2246, "step": 31640 }, { "epoch": 4.11, "learning_rate": 3.2082120951990074e-06, "loss": 0.2474, "step": 31650 }, { "epoch": 4.11, "learning_rate": 3.2072770094293865e-06, "loss": 0.2257, "step": 31660 }, { "epoch": 4.11, "learning_rate": 3.2063417645894663e-06, "loss": 0.227, "step": 31670 }, { "epoch": 4.11, "learning_rate": 3.205406360851526e-06, "loss": 0.2411, "step": 31680 }, { "epoch": 4.11, "learning_rate": 3.204470798387878e-06, "loss": 0.2169, "step": 31690 }, { "epoch": 4.11, "learning_rate": 3.2035350773708624e-06, "loss": 0.2384, "step": 31700 }, { "epoch": 4.12, "learning_rate": 3.2025991979728466e-06, "loss": 0.2244, "step": 31710 }, { "epoch": 4.12, "learning_rate": 3.2016631603662308e-06, "loss": 0.2404, "step": 31720 }, { "epoch": 4.12, "learning_rate": 3.200726964723441e-06, "loss": 0.2345, "step": 31730 }, { "epoch": 4.12, "learning_rate": 3.1997906112169346e-06, "loss": 0.2327, "step": 31740 }, { "epoch": 4.12, "learning_rate": 3.198854100019197e-06, "loss": 0.2288, "step": 31750 }, { "epoch": 4.12, "learning_rate": 3.197917431302744e-06, "loss": 0.2266, "step": 31760 }, { "epoch": 4.12, "learning_rate": 3.1969806052401176e-06, "loss": 0.2261, "step": 31770 }, { "epoch": 4.12, "learning_rate": 3.1960436220038913e-06, "loss": 0.2436, "step": 31780 }, { "epoch": 4.13, "learning_rate": 3.195106481766667e-06, "loss": 0.2212, "step": 31790 }, { "epoch": 4.13, "learning_rate": 3.194169184701075e-06, "loss": 0.2244, "step": 31800 }, { "epoch": 4.13, "learning_rate": 3.1932317309797748e-06, "loss": 0.2253, "step": 31810 }, { "epoch": 4.13, "learning_rate": 3.192294120775454e-06, "loss": 0.2243, "step": 31820 }, { "epoch": 4.13, "learning_rate": 3.1913563542608323e-06, "loss": 0.2297, "step": 31830 }, { "epoch": 4.13, "learning_rate": 3.190418431608653e-06, "loss": 0.2316, "step": 31840 }, { "epoch": 4.13, "learning_rate": 3.1894803529916914e-06, "loss": 0.2315, "step": 31850 }, { "epoch": 4.13, "learning_rate": 3.1885421185827516e-06, "loss": 0.2317, "step": 31860 }, { "epoch": 4.14, "learning_rate": 3.187603728554665e-06, "loss": 0.2216, "step": 31870 }, { "epoch": 4.14, "learning_rate": 3.1866651830802927e-06, "loss": 0.2345, "step": 31880 }, { "epoch": 4.14, "learning_rate": 3.1857264823325244e-06, "loss": 0.2317, "step": 31890 }, { "epoch": 4.14, "learning_rate": 3.184787626484279e-06, "loss": 0.236, "step": 31900 }, { "epoch": 4.14, "learning_rate": 3.1838486157085013e-06, "loss": 0.221, "step": 31910 }, { "epoch": 4.14, "learning_rate": 3.1829094501781677e-06, "loss": 0.2266, "step": 31920 }, { "epoch": 4.14, "learning_rate": 3.1819701300662817e-06, "loss": 0.2282, "step": 31930 }, { "epoch": 4.15, "learning_rate": 3.1810306555458747e-06, "loss": 0.2315, "step": 31940 }, { "epoch": 4.15, "learning_rate": 3.180091026790008e-06, "loss": 0.2205, "step": 31950 }, { "epoch": 4.15, "learning_rate": 3.1791512439717707e-06, "loss": 0.2253, "step": 31960 }, { "epoch": 4.15, "learning_rate": 3.1782113072642797e-06, "loss": 0.2331, "step": 31970 }, { "epoch": 4.15, "learning_rate": 3.1772712168406805e-06, "loss": 0.2445, "step": 31980 }, { "epoch": 4.15, "learning_rate": 3.1763309728741483e-06, "loss": 0.2443, "step": 31990 }, { "epoch": 4.15, "learning_rate": 3.1753905755378845e-06, "loss": 0.2189, "step": 32000 }, { "epoch": 4.15, "learning_rate": 3.1744500250051205e-06, "loss": 0.2521, "step": 32010 }, { "epoch": 4.16, "learning_rate": 3.1735093214491134e-06, "loss": 0.232, "step": 32020 }, { "epoch": 4.16, "learning_rate": 3.1725684650431514e-06, "loss": 0.2277, "step": 32030 }, { "epoch": 4.16, "learning_rate": 3.1716274559605506e-06, "loss": 0.2249, "step": 32040 }, { "epoch": 4.16, "learning_rate": 3.170686294374653e-06, "loss": 0.2376, "step": 32050 }, { "epoch": 4.16, "learning_rate": 3.1697449804588293e-06, "loss": 0.2371, "step": 32060 }, { "epoch": 4.16, "learning_rate": 3.168803514386481e-06, "loss": 0.2437, "step": 32070 }, { "epoch": 4.16, "learning_rate": 3.1678618963310332e-06, "loss": 0.2215, "step": 32080 }, { "epoch": 4.16, "learning_rate": 3.1669201264659427e-06, "loss": 0.2422, "step": 32090 }, { "epoch": 4.17, "learning_rate": 3.1659782049646934e-06, "loss": 0.2184, "step": 32100 }, { "epoch": 4.17, "learning_rate": 3.1650361320007958e-06, "loss": 0.2421, "step": 32110 }, { "epoch": 4.17, "learning_rate": 3.164093907747789e-06, "loss": 0.2232, "step": 32120 }, { "epoch": 4.17, "learning_rate": 3.1631515323792404e-06, "loss": 0.2287, "step": 32130 }, { "epoch": 4.17, "learning_rate": 3.162209006068746e-06, "loss": 0.2322, "step": 32140 }, { "epoch": 4.17, "learning_rate": 3.1612663289899265e-06, "loss": 0.2172, "step": 32150 }, { "epoch": 4.17, "learning_rate": 3.1603235013164338e-06, "loss": 0.2166, "step": 32160 }, { "epoch": 4.18, "learning_rate": 3.159380523221946e-06, "loss": 0.235, "step": 32170 }, { "epoch": 4.18, "learning_rate": 3.158437394880169e-06, "loss": 0.2223, "step": 32180 }, { "epoch": 4.18, "learning_rate": 3.1574941164648365e-06, "loss": 0.2375, "step": 32190 }, { "epoch": 4.18, "learning_rate": 3.156550688149709e-06, "loss": 0.2353, "step": 32200 }, { "epoch": 4.18, "learning_rate": 3.1556071101085774e-06, "loss": 0.2383, "step": 32210 }, { "epoch": 4.18, "learning_rate": 3.154663382515256e-06, "loss": 0.2368, "step": 32220 }, { "epoch": 4.18, "learning_rate": 3.15371950554359e-06, "loss": 0.2221, "step": 32230 }, { "epoch": 4.18, "learning_rate": 3.1527754793674503e-06, "loss": 0.2293, "step": 32240 }, { "epoch": 4.19, "learning_rate": 3.1518313041607377e-06, "loss": 0.2192, "step": 32250 }, { "epoch": 4.19, "learning_rate": 3.1508869800973766e-06, "loss": 0.2281, "step": 32260 }, { "epoch": 4.19, "learning_rate": 3.1499425073513216e-06, "loss": 0.2181, "step": 32270 }, { "epoch": 4.19, "learning_rate": 3.1489978860965546e-06, "loss": 0.2255, "step": 32280 }, { "epoch": 4.19, "learning_rate": 3.1480531165070836e-06, "loss": 0.232, "step": 32290 }, { "epoch": 4.19, "learning_rate": 3.147108198756945e-06, "loss": 0.2241, "step": 32300 }, { "epoch": 4.19, "learning_rate": 3.1461631330202014e-06, "loss": 0.2268, "step": 32310 }, { "epoch": 4.19, "learning_rate": 3.1452179194709442e-06, "loss": 0.2367, "step": 32320 }, { "epoch": 4.2, "learning_rate": 3.144272558283291e-06, "loss": 0.229, "step": 32330 }, { "epoch": 4.2, "learning_rate": 3.1433270496313866e-06, "loss": 0.2205, "step": 32340 }, { "epoch": 4.2, "learning_rate": 3.1423813936894034e-06, "loss": 0.252, "step": 32350 }, { "epoch": 4.2, "learning_rate": 3.14143559063154e-06, "loss": 0.2215, "step": 32360 }, { "epoch": 4.2, "learning_rate": 3.1404896406320225e-06, "loss": 0.2196, "step": 32370 }, { "epoch": 4.2, "learning_rate": 3.1395435438651055e-06, "loss": 0.2212, "step": 32380 }, { "epoch": 4.2, "learning_rate": 3.13859730050507e-06, "loss": 0.2328, "step": 32390 }, { "epoch": 4.21, "learning_rate": 3.1376509107262215e-06, "loss": 0.2243, "step": 32400 }, { "epoch": 4.21, "learning_rate": 3.1367043747028954e-06, "loss": 0.2348, "step": 32410 }, { "epoch": 4.21, "learning_rate": 3.1357576926094533e-06, "loss": 0.2355, "step": 32420 }, { "epoch": 4.21, "learning_rate": 3.134810864620284e-06, "loss": 0.2356, "step": 32430 }, { "epoch": 4.21, "learning_rate": 3.1338638909098013e-06, "loss": 0.2344, "step": 32440 }, { "epoch": 4.21, "learning_rate": 3.132916771652448e-06, "loss": 0.228, "step": 32450 }, { "epoch": 4.21, "learning_rate": 3.1319695070226923e-06, "loss": 0.2325, "step": 32460 }, { "epoch": 4.21, "learning_rate": 3.131022097195031e-06, "loss": 0.2282, "step": 32470 }, { "epoch": 4.22, "learning_rate": 3.130074542343986e-06, "loss": 0.2319, "step": 32480 }, { "epoch": 4.22, "learning_rate": 3.1291268426441063e-06, "loss": 0.2418, "step": 32490 }, { "epoch": 4.22, "learning_rate": 3.1281789982699666e-06, "loss": 0.2342, "step": 32500 }, { "epoch": 4.22, "learning_rate": 3.1272310093961705e-06, "loss": 0.2227, "step": 32510 }, { "epoch": 4.22, "learning_rate": 3.1262828761973474e-06, "loss": 0.2222, "step": 32520 }, { "epoch": 4.22, "learning_rate": 3.1253345988481513e-06, "loss": 0.2081, "step": 32530 }, { "epoch": 4.22, "learning_rate": 3.124386177523266e-06, "loss": 0.2352, "step": 32540 }, { "epoch": 4.22, "learning_rate": 3.1234376123973996e-06, "loss": 0.2402, "step": 32550 }, { "epoch": 4.23, "learning_rate": 3.1224889036452865e-06, "loss": 0.23, "step": 32560 }, { "epoch": 4.23, "learning_rate": 3.1215400514416894e-06, "loss": 0.2249, "step": 32570 }, { "epoch": 4.23, "learning_rate": 3.1205910559613956e-06, "loss": 0.2313, "step": 32580 }, { "epoch": 4.23, "learning_rate": 3.11964191737922e-06, "loss": 0.2282, "step": 32590 }, { "epoch": 4.23, "learning_rate": 3.1186926358700036e-06, "loss": 0.23, "step": 32600 }, { "epoch": 4.23, "learning_rate": 3.117743211608613e-06, "loss": 0.222, "step": 32610 }, { "epoch": 4.23, "learning_rate": 3.116793644769942e-06, "loss": 0.2211, "step": 32620 }, { "epoch": 4.23, "learning_rate": 3.11584393552891e-06, "loss": 0.215, "step": 32630 }, { "epoch": 4.24, "learning_rate": 3.114894084060463e-06, "loss": 0.2475, "step": 32640 }, { "epoch": 4.24, "learning_rate": 3.1139440905395727e-06, "loss": 0.2362, "step": 32650 }, { "epoch": 4.24, "learning_rate": 3.1129939551412392e-06, "loss": 0.2218, "step": 32660 }, { "epoch": 4.24, "learning_rate": 3.1120436780404846e-06, "loss": 0.2307, "step": 32670 }, { "epoch": 4.24, "learning_rate": 3.1110932594123606e-06, "loss": 0.2455, "step": 32680 }, { "epoch": 4.24, "learning_rate": 3.1101426994319445e-06, "loss": 0.2332, "step": 32690 }, { "epoch": 4.24, "learning_rate": 3.1091919982743372e-06, "loss": 0.2345, "step": 32700 }, { "epoch": 4.25, "learning_rate": 3.1082411561146687e-06, "loss": 0.2282, "step": 32710 }, { "epoch": 4.25, "learning_rate": 3.1072901731280927e-06, "loss": 0.2253, "step": 32720 }, { "epoch": 4.25, "learning_rate": 3.1063390494897908e-06, "loss": 0.21, "step": 32730 }, { "epoch": 4.25, "learning_rate": 3.1053877853749688e-06, "loss": 0.2248, "step": 32740 }, { "epoch": 4.25, "learning_rate": 3.104436380958859e-06, "loss": 0.2197, "step": 32750 }, { "epoch": 4.25, "learning_rate": 3.10348483641672e-06, "loss": 0.2401, "step": 32760 }, { "epoch": 4.25, "learning_rate": 3.1025331519238356e-06, "loss": 0.2375, "step": 32770 }, { "epoch": 4.25, "learning_rate": 3.1015813276555156e-06, "loss": 0.2299, "step": 32780 }, { "epoch": 4.26, "learning_rate": 3.1006293637870947e-06, "loss": 0.2247, "step": 32790 }, { "epoch": 4.26, "learning_rate": 3.0996772604939365e-06, "loss": 0.2279, "step": 32800 }, { "epoch": 4.26, "learning_rate": 3.098725017951425e-06, "loss": 0.2278, "step": 32810 }, { "epoch": 4.26, "learning_rate": 3.097772636334975e-06, "loss": 0.2353, "step": 32820 }, { "epoch": 4.26, "learning_rate": 3.0968201158200236e-06, "loss": 0.2213, "step": 32830 }, { "epoch": 4.26, "learning_rate": 3.095867456582035e-06, "loss": 0.2352, "step": 32840 }, { "epoch": 4.26, "learning_rate": 3.0949146587964984e-06, "loss": 0.2332, "step": 32850 }, { "epoch": 4.26, "learning_rate": 3.093961722638928e-06, "loss": 0.2259, "step": 32860 }, { "epoch": 4.27, "learning_rate": 3.093008648284866e-06, "loss": 0.2113, "step": 32870 }, { "epoch": 4.27, "learning_rate": 3.0920554359098767e-06, "loss": 0.2397, "step": 32880 }, { "epoch": 4.27, "learning_rate": 3.091102085689551e-06, "loss": 0.2381, "step": 32890 }, { "epoch": 4.27, "learning_rate": 3.090148597799508e-06, "loss": 0.2255, "step": 32900 }, { "epoch": 4.27, "learning_rate": 3.089194972415386e-06, "loss": 0.2382, "step": 32910 }, { "epoch": 4.27, "learning_rate": 3.088241209712855e-06, "loss": 0.2257, "step": 32920 }, { "epoch": 4.27, "learning_rate": 3.087287309867607e-06, "loss": 0.2351, "step": 32930 }, { "epoch": 4.28, "learning_rate": 3.08633327305536e-06, "loss": 0.2429, "step": 32940 }, { "epoch": 4.28, "learning_rate": 3.0853790994518563e-06, "loss": 0.2292, "step": 32950 }, { "epoch": 4.28, "learning_rate": 3.084424789232865e-06, "loss": 0.2407, "step": 32960 }, { "epoch": 4.28, "learning_rate": 3.08347034257418e-06, "loss": 0.2306, "step": 32970 }, { "epoch": 4.28, "learning_rate": 3.082515759651619e-06, "loss": 0.225, "step": 32980 }, { "epoch": 4.28, "learning_rate": 3.0815610406410248e-06, "loss": 0.2295, "step": 32990 }, { "epoch": 4.28, "learning_rate": 3.0806061857182688e-06, "loss": 0.2158, "step": 33000 }, { "epoch": 4.28, "learning_rate": 3.0796511950592424e-06, "loss": 0.2402, "step": 33010 }, { "epoch": 4.29, "learning_rate": 3.0786960688398658e-06, "loss": 0.2284, "step": 33020 }, { "epoch": 4.29, "learning_rate": 3.0777408072360826e-06, "loss": 0.2217, "step": 33030 }, { "epoch": 4.29, "learning_rate": 3.0767854104238614e-06, "loss": 0.2376, "step": 33040 }, { "epoch": 4.29, "learning_rate": 3.075829878579195e-06, "loss": 0.227, "step": 33050 }, { "epoch": 4.29, "learning_rate": 3.0748742118781027e-06, "loss": 0.2098, "step": 33060 }, { "epoch": 4.29, "learning_rate": 3.0739184104966285e-06, "loss": 0.2298, "step": 33070 }, { "epoch": 4.29, "learning_rate": 3.072962474610839e-06, "loss": 0.2324, "step": 33080 }, { "epoch": 4.29, "learning_rate": 3.072006404396829e-06, "loss": 0.2397, "step": 33090 }, { "epoch": 4.3, "learning_rate": 3.071050200030714e-06, "loss": 0.2094, "step": 33100 }, { "epoch": 4.3, "learning_rate": 3.070093861688639e-06, "loss": 0.2382, "step": 33110 }, { "epoch": 4.3, "learning_rate": 3.0691373895467686e-06, "loss": 0.2352, "step": 33120 }, { "epoch": 4.3, "learning_rate": 3.068180783781296e-06, "loss": 0.2193, "step": 33130 }, { "epoch": 4.3, "learning_rate": 3.067224044568437e-06, "loss": 0.2469, "step": 33140 }, { "epoch": 4.3, "learning_rate": 3.0662671720844328e-06, "loss": 0.2277, "step": 33150 }, { "epoch": 4.3, "learning_rate": 3.065310166505548e-06, "loss": 0.2462, "step": 33160 }, { "epoch": 4.3, "learning_rate": 3.0643530280080736e-06, "loss": 0.2174, "step": 33170 }, { "epoch": 4.31, "learning_rate": 3.0633957567683246e-06, "loss": 0.2342, "step": 33180 }, { "epoch": 4.31, "learning_rate": 3.0624383529626384e-06, "loss": 0.2262, "step": 33190 }, { "epoch": 4.31, "learning_rate": 3.061480816767379e-06, "loss": 0.2312, "step": 33200 }, { "epoch": 4.31, "learning_rate": 3.0605231483589343e-06, "loss": 0.2384, "step": 33210 }, { "epoch": 4.31, "learning_rate": 3.059565347913716e-06, "loss": 0.236, "step": 33220 }, { "epoch": 4.31, "learning_rate": 3.058607415608161e-06, "loss": 0.2296, "step": 33230 }, { "epoch": 4.31, "learning_rate": 3.0576493516187293e-06, "loss": 0.2259, "step": 33240 }, { "epoch": 4.32, "learning_rate": 3.0566911561219075e-06, "loss": 0.2243, "step": 33250 }, { "epoch": 4.32, "learning_rate": 3.0557328292942023e-06, "loss": 0.2322, "step": 33260 }, { "epoch": 4.32, "learning_rate": 3.0547743713121484e-06, "loss": 0.234, "step": 33270 }, { "epoch": 4.32, "learning_rate": 3.0538157823523045e-06, "loss": 0.2351, "step": 33280 }, { "epoch": 4.32, "learning_rate": 3.05285706259125e-06, "loss": 0.2332, "step": 33290 }, { "epoch": 4.32, "learning_rate": 3.0518982122055917e-06, "loss": 0.2192, "step": 33300 }, { "epoch": 4.32, "learning_rate": 3.0509392313719597e-06, "loss": 0.2279, "step": 33310 }, { "epoch": 4.32, "learning_rate": 3.049980120267008e-06, "loss": 0.2296, "step": 33320 }, { "epoch": 4.33, "learning_rate": 3.0490208790674136e-06, "loss": 0.2386, "step": 33330 }, { "epoch": 4.33, "learning_rate": 3.0480615079498787e-06, "loss": 0.2139, "step": 33340 }, { "epoch": 4.33, "learning_rate": 3.047102007091129e-06, "loss": 0.2322, "step": 33350 }, { "epoch": 4.33, "learning_rate": 3.0461423766679146e-06, "loss": 0.2302, "step": 33360 }, { "epoch": 4.33, "learning_rate": 3.0451826168570086e-06, "loss": 0.2275, "step": 33370 }, { "epoch": 4.33, "learning_rate": 3.044222727835208e-06, "loss": 0.2232, "step": 33380 }, { "epoch": 4.33, "learning_rate": 3.0432627097793346e-06, "loss": 0.2213, "step": 33390 }, { "epoch": 4.33, "learning_rate": 3.042302562866233e-06, "loss": 0.226, "step": 33400 }, { "epoch": 4.34, "learning_rate": 3.0413422872727715e-06, "loss": 0.2409, "step": 33410 }, { "epoch": 4.34, "learning_rate": 3.040381883175844e-06, "loss": 0.2249, "step": 33420 }, { "epoch": 4.34, "learning_rate": 3.039421350752364e-06, "loss": 0.2382, "step": 33430 }, { "epoch": 4.34, "learning_rate": 3.038460690179273e-06, "loss": 0.2229, "step": 33440 }, { "epoch": 4.34, "learning_rate": 3.0374999016335337e-06, "loss": 0.2207, "step": 33450 }, { "epoch": 4.34, "learning_rate": 3.0365389852921335e-06, "loss": 0.2322, "step": 33460 }, { "epoch": 4.34, "learning_rate": 3.0355779413320824e-06, "loss": 0.2372, "step": 33470 }, { "epoch": 4.35, "learning_rate": 3.0346167699304135e-06, "loss": 0.2341, "step": 33480 }, { "epoch": 4.35, "learning_rate": 3.033655471264186e-06, "loss": 0.2325, "step": 33490 }, { "epoch": 4.35, "learning_rate": 3.0326940455104785e-06, "loss": 0.237, "step": 33500 }, { "epoch": 4.35, "learning_rate": 3.0317324928463967e-06, "loss": 0.2202, "step": 33510 }, { "epoch": 4.35, "learning_rate": 3.0307708134490682e-06, "loss": 0.2227, "step": 33520 }, { "epoch": 4.35, "learning_rate": 3.0298090074956434e-06, "loss": 0.234, "step": 33530 }, { "epoch": 4.35, "learning_rate": 3.028847075163297e-06, "loss": 0.2266, "step": 33540 }, { "epoch": 4.35, "learning_rate": 3.0278850166292257e-06, "loss": 0.2307, "step": 33550 }, { "epoch": 4.36, "learning_rate": 3.0269228320706514e-06, "loss": 0.2222, "step": 33560 }, { "epoch": 4.36, "learning_rate": 3.0259605216648177e-06, "loss": 0.2297, "step": 33570 }, { "epoch": 4.36, "learning_rate": 3.024998085588992e-06, "loss": 0.2273, "step": 33580 }, { "epoch": 4.36, "learning_rate": 3.024035524020463e-06, "loss": 0.2354, "step": 33590 }, { "epoch": 4.36, "learning_rate": 3.023072837136547e-06, "loss": 0.2392, "step": 33600 }, { "epoch": 4.36, "learning_rate": 3.0221100251145782e-06, "loss": 0.2301, "step": 33610 }, { "epoch": 4.36, "learning_rate": 3.0211470881319176e-06, "loss": 0.2317, "step": 33620 }, { "epoch": 4.36, "learning_rate": 3.0201840263659472e-06, "loss": 0.2447, "step": 33630 }, { "epoch": 4.37, "learning_rate": 3.0192208399940717e-06, "loss": 0.2207, "step": 33640 }, { "epoch": 4.37, "learning_rate": 3.018257529193722e-06, "loss": 0.2399, "step": 33650 }, { "epoch": 4.37, "learning_rate": 3.0172940941423467e-06, "loss": 0.2357, "step": 33660 }, { "epoch": 4.37, "learning_rate": 3.0163305350174228e-06, "loss": 0.2138, "step": 33670 }, { "epoch": 4.37, "learning_rate": 3.015366851996446e-06, "loss": 0.2311, "step": 33680 }, { "epoch": 4.37, "learning_rate": 3.0144030452569362e-06, "loss": 0.222, "step": 33690 }, { "epoch": 4.37, "learning_rate": 3.013439114976437e-06, "loss": 0.2217, "step": 33700 }, { "epoch": 4.38, "learning_rate": 3.012475061332514e-06, "loss": 0.235, "step": 33710 }, { "epoch": 4.38, "learning_rate": 3.0115108845027547e-06, "loss": 0.2338, "step": 33720 }, { "epoch": 4.38, "learning_rate": 3.0105465846647707e-06, "loss": 0.2267, "step": 33730 }, { "epoch": 4.38, "learning_rate": 3.0095821619961962e-06, "loss": 0.2394, "step": 33740 }, { "epoch": 4.38, "learning_rate": 3.008617616674686e-06, "loss": 0.2347, "step": 33750 }, { "epoch": 4.38, "learning_rate": 3.00765294887792e-06, "loss": 0.2248, "step": 33760 }, { "epoch": 4.38, "learning_rate": 3.0066881587836e-06, "loss": 0.2346, "step": 33770 }, { "epoch": 4.38, "learning_rate": 3.005723246569449e-06, "loss": 0.2337, "step": 33780 }, { "epoch": 4.39, "learning_rate": 3.0047582124132136e-06, "loss": 0.2446, "step": 33790 }, { "epoch": 4.39, "learning_rate": 3.0037930564926637e-06, "loss": 0.2344, "step": 33800 }, { "epoch": 4.39, "learning_rate": 3.0028277789855902e-06, "loss": 0.2316, "step": 33810 }, { "epoch": 4.39, "learning_rate": 3.001862380069806e-06, "loss": 0.2204, "step": 33820 }, { "epoch": 4.39, "learning_rate": 3.000896859923148e-06, "loss": 0.2242, "step": 33830 }, { "epoch": 4.39, "learning_rate": 2.999931218723475e-06, "loss": 0.2281, "step": 33840 }, { "epoch": 4.39, "learning_rate": 2.998965456648667e-06, "loss": 0.2317, "step": 33850 }, { "epoch": 4.39, "learning_rate": 2.997999573876627e-06, "loss": 0.2394, "step": 33860 }, { "epoch": 4.4, "learning_rate": 2.997033570585281e-06, "loss": 0.2267, "step": 33870 }, { "epoch": 4.4, "learning_rate": 2.996067446952576e-06, "loss": 0.2193, "step": 33880 }, { "epoch": 4.4, "learning_rate": 2.9951012031564814e-06, "loss": 0.2306, "step": 33890 }, { "epoch": 4.4, "learning_rate": 2.9941348393749897e-06, "loss": 0.2231, "step": 33900 }, { "epoch": 4.4, "learning_rate": 2.993168355786114e-06, "loss": 0.2314, "step": 33910 }, { "epoch": 4.4, "learning_rate": 2.9922017525678906e-06, "loss": 0.2251, "step": 33920 }, { "epoch": 4.4, "learning_rate": 2.991235029898377e-06, "loss": 0.2362, "step": 33930 }, { "epoch": 4.4, "learning_rate": 2.990268187955654e-06, "loss": 0.2466, "step": 33940 }, { "epoch": 4.41, "learning_rate": 2.9893012269178228e-06, "loss": 0.2335, "step": 33950 }, { "epoch": 4.41, "learning_rate": 2.9883341469630072e-06, "loss": 0.2242, "step": 33960 }, { "epoch": 4.41, "learning_rate": 2.987366948269354e-06, "loss": 0.2353, "step": 33970 }, { "epoch": 4.41, "learning_rate": 2.98639963101503e-06, "loss": 0.2426, "step": 33980 }, { "epoch": 4.41, "learning_rate": 2.985432195378225e-06, "loss": 0.2421, "step": 33990 }, { "epoch": 4.41, "learning_rate": 2.98446464153715e-06, "loss": 0.2279, "step": 34000 }, { "epoch": 4.41, "learning_rate": 2.9834969696700384e-06, "loss": 0.2353, "step": 34010 }, { "epoch": 4.42, "learning_rate": 2.982529179955145e-06, "loss": 0.2165, "step": 34020 }, { "epoch": 4.42, "learning_rate": 2.981561272570746e-06, "loss": 0.2396, "step": 34030 }, { "epoch": 4.42, "learning_rate": 2.9805932476951405e-06, "loss": 0.2389, "step": 34040 }, { "epoch": 4.42, "learning_rate": 2.9796251055066474e-06, "loss": 0.2248, "step": 34050 }, { "epoch": 4.42, "learning_rate": 2.9786568461836083e-06, "loss": 0.2277, "step": 34060 }, { "epoch": 4.42, "learning_rate": 2.9776884699043863e-06, "loss": 0.2346, "step": 34070 }, { "epoch": 4.42, "learning_rate": 2.9767199768473665e-06, "loss": 0.2399, "step": 34080 }, { "epoch": 4.42, "learning_rate": 2.975751367190955e-06, "loss": 0.2246, "step": 34090 }, { "epoch": 4.43, "learning_rate": 2.974782641113578e-06, "loss": 0.226, "step": 34100 }, { "epoch": 4.43, "learning_rate": 2.9738137987936856e-06, "loss": 0.2311, "step": 34110 }, { "epoch": 4.43, "learning_rate": 2.972844840409749e-06, "loss": 0.2465, "step": 34120 }, { "epoch": 4.43, "learning_rate": 2.9718757661402583e-06, "loss": 0.235, "step": 34130 }, { "epoch": 4.43, "learning_rate": 2.9709065761637276e-06, "loss": 0.2281, "step": 34140 }, { "epoch": 4.43, "learning_rate": 2.969937270658692e-06, "loss": 0.2209, "step": 34150 }, { "epoch": 4.43, "learning_rate": 2.9689678498037055e-06, "loss": 0.2232, "step": 34160 }, { "epoch": 4.43, "learning_rate": 2.967998313777346e-06, "loss": 0.2191, "step": 34170 }, { "epoch": 4.44, "learning_rate": 2.9670286627582124e-06, "loss": 0.2233, "step": 34180 }, { "epoch": 4.44, "learning_rate": 2.966058896924924e-06, "loss": 0.2381, "step": 34190 }, { "epoch": 4.44, "learning_rate": 2.9650890164561202e-06, "loss": 0.2297, "step": 34200 }, { "epoch": 4.44, "learning_rate": 2.964119021530464e-06, "loss": 0.2292, "step": 34210 }, { "epoch": 4.44, "learning_rate": 2.963148912326637e-06, "loss": 0.2455, "step": 34220 }, { "epoch": 4.44, "learning_rate": 2.9621786890233437e-06, "loss": 0.2357, "step": 34230 }, { "epoch": 4.44, "learning_rate": 2.961208351799309e-06, "loss": 0.2447, "step": 34240 }, { "epoch": 4.45, "learning_rate": 2.960237900833278e-06, "loss": 0.2348, "step": 34250 }, { "epoch": 4.45, "learning_rate": 2.959267336304018e-06, "loss": 0.2331, "step": 34260 }, { "epoch": 4.45, "learning_rate": 2.958296658390317e-06, "loss": 0.231, "step": 34270 }, { "epoch": 4.45, "learning_rate": 2.9573258672709827e-06, "loss": 0.2296, "step": 34280 }, { "epoch": 4.45, "learning_rate": 2.956354963124846e-06, "loss": 0.2224, "step": 34290 }, { "epoch": 4.45, "learning_rate": 2.9553839461307553e-06, "loss": 0.2351, "step": 34300 }, { "epoch": 4.45, "learning_rate": 2.9544128164675825e-06, "loss": 0.2244, "step": 34310 }, { "epoch": 4.45, "learning_rate": 2.953441574314219e-06, "loss": 0.2159, "step": 34320 }, { "epoch": 4.46, "learning_rate": 2.9524702198495785e-06, "loss": 0.2178, "step": 34330 }, { "epoch": 4.46, "learning_rate": 2.9514987532525925e-06, "loss": 0.2342, "step": 34340 }, { "epoch": 4.46, "learning_rate": 2.950527174702216e-06, "loss": 0.2357, "step": 34350 }, { "epoch": 4.46, "learning_rate": 2.9495554843774243e-06, "loss": 0.2312, "step": 34360 }, { "epoch": 4.46, "learning_rate": 2.9485836824572097e-06, "loss": 0.2311, "step": 34370 }, { "epoch": 4.46, "learning_rate": 2.9476117691205906e-06, "loss": 0.2412, "step": 34380 }, { "epoch": 4.46, "learning_rate": 2.9466397445466006e-06, "loss": 0.2222, "step": 34390 }, { "epoch": 4.46, "learning_rate": 2.9456676089142995e-06, "loss": 0.2273, "step": 34400 }, { "epoch": 4.47, "learning_rate": 2.9446953624027615e-06, "loss": 0.2274, "step": 34410 }, { "epoch": 4.47, "learning_rate": 2.9437230051910853e-06, "loss": 0.2402, "step": 34420 }, { "epoch": 4.47, "learning_rate": 2.9427505374583897e-06, "loss": 0.2248, "step": 34430 }, { "epoch": 4.47, "learning_rate": 2.941777959383811e-06, "loss": 0.2261, "step": 34440 }, { "epoch": 4.47, "learning_rate": 2.9408052711465088e-06, "loss": 0.2276, "step": 34450 }, { "epoch": 4.47, "learning_rate": 2.939832472925663e-06, "loss": 0.2357, "step": 34460 }, { "epoch": 4.47, "learning_rate": 2.9388595649004708e-06, "loss": 0.2303, "step": 34470 }, { "epoch": 4.48, "learning_rate": 2.9378865472501533e-06, "loss": 0.2222, "step": 34480 }, { "epoch": 4.48, "learning_rate": 2.9369134201539484e-06, "loss": 0.2329, "step": 34490 }, { "epoch": 4.48, "learning_rate": 2.9359401837911176e-06, "loss": 0.231, "step": 34500 }, { "epoch": 4.48, "learning_rate": 2.9349668383409397e-06, "loss": 0.2357, "step": 34510 }, { "epoch": 4.48, "learning_rate": 2.9339933839827145e-06, "loss": 0.2427, "step": 34520 }, { "epoch": 4.48, "learning_rate": 2.9330198208957624e-06, "loss": 0.2238, "step": 34530 }, { "epoch": 4.48, "learning_rate": 2.9320461492594242e-06, "loss": 0.2255, "step": 34540 }, { "epoch": 4.48, "learning_rate": 2.9310723692530585e-06, "loss": 0.2316, "step": 34550 }, { "epoch": 4.49, "learning_rate": 2.9300984810560456e-06, "loss": 0.2367, "step": 34560 }, { "epoch": 4.49, "learning_rate": 2.9291244848477867e-06, "loss": 0.2119, "step": 34570 }, { "epoch": 4.49, "learning_rate": 2.9281503808077003e-06, "loss": 0.2407, "step": 34580 }, { "epoch": 4.49, "learning_rate": 2.9271761691152265e-06, "loss": 0.2293, "step": 34590 }, { "epoch": 4.49, "learning_rate": 2.9262018499498245e-06, "loss": 0.2166, "step": 34600 }, { "epoch": 4.49, "learning_rate": 2.9252274234909746e-06, "loss": 0.2345, "step": 34610 }, { "epoch": 4.49, "learning_rate": 2.9242528899181746e-06, "loss": 0.227, "step": 34620 }, { "epoch": 4.49, "learning_rate": 2.9232782494109446e-06, "loss": 0.2433, "step": 34630 }, { "epoch": 4.5, "learning_rate": 2.9223035021488226e-06, "loss": 0.2304, "step": 34640 }, { "epoch": 4.5, "learning_rate": 2.9213286483113655e-06, "loss": 0.2168, "step": 34650 }, { "epoch": 4.5, "learning_rate": 2.9203536880781538e-06, "loss": 0.2221, "step": 34660 }, { "epoch": 4.5, "learning_rate": 2.919378621628782e-06, "loss": 0.2257, "step": 34670 }, { "epoch": 4.5, "learning_rate": 2.9184034491428695e-06, "loss": 0.2298, "step": 34680 }, { "epoch": 4.5, "learning_rate": 2.9174281708000515e-06, "loss": 0.2234, "step": 34690 }, { "epoch": 4.5, "learning_rate": 2.916452786779984e-06, "loss": 0.2392, "step": 34700 }, { "epoch": 4.5, "learning_rate": 2.9154772972623435e-06, "loss": 0.2432, "step": 34710 }, { "epoch": 4.51, "learning_rate": 2.9145017024268235e-06, "loss": 0.2389, "step": 34720 }, { "epoch": 4.51, "learning_rate": 2.913526002453139e-06, "loss": 0.2098, "step": 34730 }, { "epoch": 4.51, "learning_rate": 2.912550197521024e-06, "loss": 0.2261, "step": 34740 }, { "epoch": 4.51, "learning_rate": 2.9115742878102313e-06, "loss": 0.237, "step": 34750 }, { "epoch": 4.51, "learning_rate": 2.910598273500533e-06, "loss": 0.2368, "step": 34760 }, { "epoch": 4.51, "learning_rate": 2.9096221547717205e-06, "loss": 0.2371, "step": 34770 }, { "epoch": 4.51, "learning_rate": 2.908645931803606e-06, "loss": 0.2484, "step": 34780 }, { "epoch": 4.52, "learning_rate": 2.907669604776017e-06, "loss": 0.2232, "step": 34790 }, { "epoch": 4.52, "learning_rate": 2.906693173868805e-06, "loss": 0.2201, "step": 34800 }, { "epoch": 4.52, "learning_rate": 2.905716639261837e-06, "loss": 0.2392, "step": 34810 }, { "epoch": 4.52, "learning_rate": 2.9047400011350014e-06, "loss": 0.2313, "step": 34820 }, { "epoch": 4.52, "learning_rate": 2.9037632596682034e-06, "loss": 0.2409, "step": 34830 }, { "epoch": 4.52, "learning_rate": 2.90278641504137e-06, "loss": 0.2364, "step": 34840 }, { "epoch": 4.52, "learning_rate": 2.9018094674344454e-06, "loss": 0.2093, "step": 34850 }, { "epoch": 4.52, "learning_rate": 2.9008324170273918e-06, "loss": 0.2323, "step": 34860 }, { "epoch": 4.53, "learning_rate": 2.8998552640001934e-06, "loss": 0.2162, "step": 34870 }, { "epoch": 4.53, "learning_rate": 2.8988780085328504e-06, "loss": 0.2278, "step": 34880 }, { "epoch": 4.53, "learning_rate": 2.8979006508053833e-06, "loss": 0.224, "step": 34890 }, { "epoch": 4.53, "learning_rate": 2.896923190997831e-06, "loss": 0.2512, "step": 34900 }, { "epoch": 4.53, "learning_rate": 2.8959456292902523e-06, "loss": 0.2307, "step": 34910 }, { "epoch": 4.53, "learning_rate": 2.894967965862723e-06, "loss": 0.2258, "step": 34920 }, { "epoch": 4.53, "learning_rate": 2.893990200895338e-06, "loss": 0.2381, "step": 34930 }, { "epoch": 4.53, "learning_rate": 2.8930123345682123e-06, "loss": 0.2193, "step": 34940 }, { "epoch": 4.54, "learning_rate": 2.8920343670614783e-06, "loss": 0.231, "step": 34950 }, { "epoch": 4.54, "learning_rate": 2.8910562985552875e-06, "loss": 0.2361, "step": 34960 }, { "epoch": 4.54, "learning_rate": 2.8900781292298096e-06, "loss": 0.2347, "step": 34970 }, { "epoch": 4.54, "learning_rate": 2.8890998592652336e-06, "loss": 0.2216, "step": 34980 }, { "epoch": 4.54, "learning_rate": 2.888121488841767e-06, "loss": 0.2419, "step": 34990 }, { "epoch": 4.54, "learning_rate": 2.8871430181396332e-06, "loss": 0.2316, "step": 35000 }, { "epoch": 4.54, "learning_rate": 2.8861644473390786e-06, "loss": 0.2157, "step": 35010 }, { "epoch": 4.55, "learning_rate": 2.8851857766203652e-06, "loss": 0.2458, "step": 35020 }, { "epoch": 4.55, "learning_rate": 2.884207006163773e-06, "loss": 0.2285, "step": 35030 }, { "epoch": 4.55, "learning_rate": 2.8832281361496027e-06, "loss": 0.2229, "step": 35040 }, { "epoch": 4.55, "learning_rate": 2.8822491667581707e-06, "loss": 0.2253, "step": 35050 }, { "epoch": 4.55, "learning_rate": 2.8812700981698136e-06, "loss": 0.2298, "step": 35060 }, { "epoch": 4.55, "learning_rate": 2.8802909305648855e-06, "loss": 0.2246, "step": 35070 }, { "epoch": 4.55, "learning_rate": 2.8793116641237584e-06, "loss": 0.2343, "step": 35080 }, { "epoch": 4.55, "learning_rate": 2.8783322990268243e-06, "loss": 0.2299, "step": 35090 }, { "epoch": 4.56, "learning_rate": 2.8773528354544902e-06, "loss": 0.213, "step": 35100 }, { "epoch": 4.56, "learning_rate": 2.876373273587184e-06, "loss": 0.2325, "step": 35110 }, { "epoch": 4.56, "learning_rate": 2.8753936136053508e-06, "loss": 0.238, "step": 35120 }, { "epoch": 4.56, "learning_rate": 2.8744138556894542e-06, "loss": 0.221, "step": 35130 }, { "epoch": 4.56, "learning_rate": 2.8734340000199743e-06, "loss": 0.2361, "step": 35140 }, { "epoch": 4.56, "learning_rate": 2.872454046777411e-06, "loss": 0.2256, "step": 35150 }, { "epoch": 4.56, "learning_rate": 2.8714739961422817e-06, "loss": 0.2192, "step": 35160 }, { "epoch": 4.56, "learning_rate": 2.8704938482951207e-06, "loss": 0.2223, "step": 35170 }, { "epoch": 4.57, "learning_rate": 2.8695136034164816e-06, "loss": 0.2362, "step": 35180 }, { "epoch": 4.57, "learning_rate": 2.8685332616869357e-06, "loss": 0.2226, "step": 35190 }, { "epoch": 4.57, "learning_rate": 2.867552823287071e-06, "loss": 0.2395, "step": 35200 }, { "epoch": 4.57, "learning_rate": 2.866572288397494e-06, "loss": 0.2315, "step": 35210 }, { "epoch": 4.57, "learning_rate": 2.8655916571988294e-06, "loss": 0.2304, "step": 35220 }, { "epoch": 4.57, "learning_rate": 2.86461092987172e-06, "loss": 0.2294, "step": 35230 }, { "epoch": 4.57, "learning_rate": 2.8636301065968234e-06, "loss": 0.2291, "step": 35240 }, { "epoch": 4.57, "learning_rate": 2.862649187554819e-06, "loss": 0.2358, "step": 35250 }, { "epoch": 4.58, "learning_rate": 2.8616681729264014e-06, "loss": 0.237, "step": 35260 }, { "epoch": 4.58, "learning_rate": 2.8606870628922834e-06, "loss": 0.2211, "step": 35270 }, { "epoch": 4.58, "learning_rate": 2.8597058576331943e-06, "loss": 0.2367, "step": 35280 }, { "epoch": 4.58, "learning_rate": 2.858724557329883e-06, "loss": 0.223, "step": 35290 }, { "epoch": 4.58, "learning_rate": 2.857743162163115e-06, "loss": 0.231, "step": 35300 }, { "epoch": 4.58, "learning_rate": 2.856761672313672e-06, "loss": 0.2098, "step": 35310 }, { "epoch": 4.58, "learning_rate": 2.855780087962355e-06, "loss": 0.2334, "step": 35320 }, { "epoch": 4.59, "learning_rate": 2.8547984092899807e-06, "loss": 0.2227, "step": 35330 }, { "epoch": 4.59, "learning_rate": 2.853816636477386e-06, "loss": 0.2319, "step": 35340 }, { "epoch": 4.59, "learning_rate": 2.852834769705421e-06, "loss": 0.23, "step": 35350 }, { "epoch": 4.59, "learning_rate": 2.851852809154957e-06, "loss": 0.2472, "step": 35360 }, { "epoch": 4.59, "learning_rate": 2.8508707550068813e-06, "loss": 0.2291, "step": 35370 }, { "epoch": 4.59, "learning_rate": 2.849888607442096e-06, "loss": 0.2256, "step": 35380 }, { "epoch": 4.59, "learning_rate": 2.8489063666415237e-06, "loss": 0.2301, "step": 35390 }, { "epoch": 4.59, "learning_rate": 2.847924032786103e-06, "loss": 0.2249, "step": 35400 }, { "epoch": 4.6, "learning_rate": 2.8469416060567903e-06, "loss": 0.2259, "step": 35410 }, { "epoch": 4.6, "learning_rate": 2.8459590866345572e-06, "loss": 0.2254, "step": 35420 }, { "epoch": 4.6, "learning_rate": 2.8449764747003936e-06, "loss": 0.2295, "step": 35430 }, { "epoch": 4.6, "learning_rate": 2.843993770435308e-06, "loss": 0.2317, "step": 35440 }, { "epoch": 4.6, "learning_rate": 2.8430109740203224e-06, "loss": 0.2279, "step": 35450 }, { "epoch": 4.6, "learning_rate": 2.842028085636478e-06, "loss": 0.2258, "step": 35460 }, { "epoch": 4.6, "learning_rate": 2.841045105464834e-06, "loss": 0.2157, "step": 35470 }, { "epoch": 4.6, "learning_rate": 2.840062033686465e-06, "loss": 0.2311, "step": 35480 }, { "epoch": 4.61, "learning_rate": 2.839078870482461e-06, "loss": 0.2128, "step": 35490 }, { "epoch": 4.61, "learning_rate": 2.838095616033932e-06, "loss": 0.2336, "step": 35500 }, { "epoch": 4.61, "learning_rate": 2.837112270522003e-06, "loss": 0.2356, "step": 35510 }, { "epoch": 4.61, "learning_rate": 2.836128834127815e-06, "loss": 0.2338, "step": 35520 }, { "epoch": 4.61, "learning_rate": 2.8351453070325285e-06, "loss": 0.227, "step": 35530 }, { "epoch": 4.61, "learning_rate": 2.8341616894173176e-06, "loss": 0.2202, "step": 35540 }, { "epoch": 4.61, "learning_rate": 2.833177981463376e-06, "loss": 0.2293, "step": 35550 }, { "epoch": 4.62, "learning_rate": 2.8321941833519104e-06, "loss": 0.2271, "step": 35560 }, { "epoch": 4.62, "learning_rate": 2.831210295264148e-06, "loss": 0.2285, "step": 35570 }, { "epoch": 4.62, "learning_rate": 2.8302263173813305e-06, "loss": 0.2218, "step": 35580 }, { "epoch": 4.62, "learning_rate": 2.8292422498847164e-06, "loss": 0.2293, "step": 35590 }, { "epoch": 4.62, "learning_rate": 2.82825809295558e-06, "loss": 0.2239, "step": 35600 }, { "epoch": 4.62, "learning_rate": 2.8272738467752132e-06, "loss": 0.2332, "step": 35610 }, { "epoch": 4.62, "learning_rate": 2.826289511524925e-06, "loss": 0.233, "step": 35620 }, { "epoch": 4.62, "learning_rate": 2.825305087386038e-06, "loss": 0.2266, "step": 35630 }, { "epoch": 4.63, "learning_rate": 2.8243205745398947e-06, "loss": 0.2298, "step": 35640 }, { "epoch": 4.63, "learning_rate": 2.8233359731678515e-06, "loss": 0.2268, "step": 35650 }, { "epoch": 4.63, "learning_rate": 2.8223512834512812e-06, "loss": 0.2308, "step": 35660 }, { "epoch": 4.63, "learning_rate": 2.821366505571574e-06, "loss": 0.2287, "step": 35670 }, { "epoch": 4.63, "learning_rate": 2.820381639710136e-06, "loss": 0.2271, "step": 35680 }, { "epoch": 4.63, "learning_rate": 2.81939668604839e-06, "loss": 0.2276, "step": 35690 }, { "epoch": 4.63, "learning_rate": 2.8184116447677726e-06, "loss": 0.2401, "step": 35700 }, { "epoch": 4.63, "learning_rate": 2.817426516049739e-06, "loss": 0.2281, "step": 35710 }, { "epoch": 4.64, "learning_rate": 2.81644130007576e-06, "loss": 0.2454, "step": 35720 }, { "epoch": 4.64, "learning_rate": 2.8154559970273227e-06, "loss": 0.2386, "step": 35730 }, { "epoch": 4.64, "learning_rate": 2.8144706070859282e-06, "loss": 0.2142, "step": 35740 }, { "epoch": 4.64, "learning_rate": 2.8134851304330963e-06, "loss": 0.2253, "step": 35750 }, { "epoch": 4.64, "learning_rate": 2.812499567250362e-06, "loss": 0.2349, "step": 35760 }, { "epoch": 4.64, "learning_rate": 2.811513917719275e-06, "loss": 0.2247, "step": 35770 }, { "epoch": 4.64, "learning_rate": 2.810528182021402e-06, "loss": 0.2312, "step": 35780 }, { "epoch": 4.65, "learning_rate": 2.8095423603383264e-06, "loss": 0.2171, "step": 35790 }, { "epoch": 4.65, "learning_rate": 2.8085564528516443e-06, "loss": 0.2239, "step": 35800 }, { "epoch": 4.65, "learning_rate": 2.8075704597429717e-06, "loss": 0.2333, "step": 35810 }, { "epoch": 4.65, "learning_rate": 2.806584381193938e-06, "loss": 0.2317, "step": 35820 }, { "epoch": 4.65, "learning_rate": 2.8055982173861865e-06, "loss": 0.223, "step": 35830 }, { "epoch": 4.65, "learning_rate": 2.8046119685013817e-06, "loss": 0.2213, "step": 35840 }, { "epoch": 4.65, "learning_rate": 2.8036256347211983e-06, "loss": 0.2377, "step": 35850 }, { "epoch": 4.65, "learning_rate": 2.802639216227331e-06, "loss": 0.2461, "step": 35860 }, { "epoch": 4.66, "learning_rate": 2.801652713201485e-06, "loss": 0.2172, "step": 35870 }, { "epoch": 4.66, "learning_rate": 2.8006661258253857e-06, "loss": 0.2377, "step": 35880 }, { "epoch": 4.66, "learning_rate": 2.7996794542807726e-06, "loss": 0.225, "step": 35890 }, { "epoch": 4.66, "learning_rate": 2.7986926987493992e-06, "loss": 0.229, "step": 35900 }, { "epoch": 4.66, "learning_rate": 2.797705859413037e-06, "loss": 0.2417, "step": 35910 }, { "epoch": 4.66, "learning_rate": 2.7967189364534713e-06, "loss": 0.2333, "step": 35920 }, { "epoch": 4.66, "learning_rate": 2.7957319300525035e-06, "loss": 0.2339, "step": 35930 }, { "epoch": 4.66, "learning_rate": 2.7947448403919494e-06, "loss": 0.2361, "step": 35940 }, { "epoch": 4.67, "learning_rate": 2.79375766765364e-06, "loss": 0.242, "step": 35950 }, { "epoch": 4.67, "learning_rate": 2.7927704120194246e-06, "loss": 0.2286, "step": 35960 }, { "epoch": 4.67, "learning_rate": 2.791783073671164e-06, "loss": 0.2315, "step": 35970 }, { "epoch": 4.67, "learning_rate": 2.7907956527907357e-06, "loss": 0.2279, "step": 35980 }, { "epoch": 4.67, "learning_rate": 2.7898081495600325e-06, "loss": 0.2355, "step": 35990 }, { "epoch": 4.67, "learning_rate": 2.788820564160964e-06, "loss": 0.2244, "step": 36000 }, { "epoch": 4.67, "learning_rate": 2.787832896775451e-06, "loss": 0.2268, "step": 36010 }, { "epoch": 4.67, "learning_rate": 2.786845147585433e-06, "loss": 0.2345, "step": 36020 }, { "epoch": 4.68, "learning_rate": 2.785857316772864e-06, "loss": 0.225, "step": 36030 }, { "epoch": 4.68, "learning_rate": 2.7848694045197095e-06, "loss": 0.2279, "step": 36040 }, { "epoch": 4.68, "learning_rate": 2.783881411007956e-06, "loss": 0.2343, "step": 36050 }, { "epoch": 4.68, "learning_rate": 2.7828933364195994e-06, "loss": 0.2243, "step": 36060 }, { "epoch": 4.68, "learning_rate": 2.7819051809366543e-06, "loss": 0.2299, "step": 36070 }, { "epoch": 4.68, "learning_rate": 2.7809169447411487e-06, "loss": 0.2346, "step": 36080 }, { "epoch": 4.68, "learning_rate": 2.779928628015125e-06, "loss": 0.2306, "step": 36090 }, { "epoch": 4.69, "learning_rate": 2.7789402309406413e-06, "loss": 0.2319, "step": 36100 }, { "epoch": 4.69, "learning_rate": 2.77795175369977e-06, "loss": 0.2293, "step": 36110 }, { "epoch": 4.69, "learning_rate": 2.776963196474599e-06, "loss": 0.2282, "step": 36120 }, { "epoch": 4.69, "learning_rate": 2.77597455944723e-06, "loss": 0.2219, "step": 36130 }, { "epoch": 4.69, "learning_rate": 2.77498584279978e-06, "loss": 0.2283, "step": 36140 }, { "epoch": 4.69, "learning_rate": 2.77399704671438e-06, "loss": 0.2236, "step": 36150 }, { "epoch": 4.69, "learning_rate": 2.7730081713731765e-06, "loss": 0.2429, "step": 36160 }, { "epoch": 4.69, "learning_rate": 2.772019216958331e-06, "loss": 0.2401, "step": 36170 }, { "epoch": 4.7, "learning_rate": 2.7710301836520173e-06, "loss": 0.2379, "step": 36180 }, { "epoch": 4.7, "learning_rate": 2.770041071636426e-06, "loss": 0.2218, "step": 36190 }, { "epoch": 4.7, "learning_rate": 2.7690518810937616e-06, "loss": 0.2433, "step": 36200 }, { "epoch": 4.7, "learning_rate": 2.768062612206243e-06, "loss": 0.2332, "step": 36210 }, { "epoch": 4.7, "learning_rate": 2.7670732651561023e-06, "loss": 0.2379, "step": 36220 }, { "epoch": 4.7, "learning_rate": 2.7660838401255874e-06, "loss": 0.2327, "step": 36230 }, { "epoch": 4.7, "learning_rate": 2.765094337296962e-06, "loss": 0.2297, "step": 36240 }, { "epoch": 4.7, "learning_rate": 2.7641047568525004e-06, "loss": 0.2242, "step": 36250 }, { "epoch": 4.71, "learning_rate": 2.763115098974494e-06, "loss": 0.225, "step": 36260 }, { "epoch": 4.71, "learning_rate": 2.7621253638452465e-06, "loss": 0.2349, "step": 36270 }, { "epoch": 4.71, "learning_rate": 2.76113555164708e-06, "loss": 0.221, "step": 36280 }, { "epoch": 4.71, "learning_rate": 2.7601456625623247e-06, "loss": 0.2387, "step": 36290 }, { "epoch": 4.71, "learning_rate": 2.7591556967733292e-06, "loss": 0.2346, "step": 36300 }, { "epoch": 4.71, "learning_rate": 2.7581656544624557e-06, "loss": 0.2167, "step": 36310 }, { "epoch": 4.71, "learning_rate": 2.7571755358120788e-06, "loss": 0.2422, "step": 36320 }, { "epoch": 4.72, "learning_rate": 2.7561853410045893e-06, "loss": 0.2339, "step": 36330 }, { "epoch": 4.72, "learning_rate": 2.7551950702223894e-06, "loss": 0.2095, "step": 36340 }, { "epoch": 4.72, "learning_rate": 2.754204723647899e-06, "loss": 0.2299, "step": 36350 }, { "epoch": 4.72, "learning_rate": 2.7532143014635484e-06, "loss": 0.2307, "step": 36360 }, { "epoch": 4.72, "learning_rate": 2.7522238038517833e-06, "loss": 0.23, "step": 36370 }, { "epoch": 4.72, "learning_rate": 2.751233230995064e-06, "loss": 0.215, "step": 36380 }, { "epoch": 4.72, "learning_rate": 2.7502425830758623e-06, "loss": 0.2344, "step": 36390 }, { "epoch": 4.72, "learning_rate": 2.7492518602766676e-06, "loss": 0.2253, "step": 36400 }, { "epoch": 4.73, "learning_rate": 2.7482610627799793e-06, "loss": 0.2296, "step": 36410 }, { "epoch": 4.73, "learning_rate": 2.7472701907683135e-06, "loss": 0.2441, "step": 36420 }, { "epoch": 4.73, "learning_rate": 2.7462792444241974e-06, "loss": 0.2316, "step": 36430 }, { "epoch": 4.73, "learning_rate": 2.7452882239301743e-06, "loss": 0.2256, "step": 36440 }, { "epoch": 4.73, "learning_rate": 2.7442971294688e-06, "loss": 0.228, "step": 36450 }, { "epoch": 4.73, "learning_rate": 2.7433059612226426e-06, "loss": 0.2233, "step": 36460 }, { "epoch": 4.73, "learning_rate": 2.742314719374287e-06, "loss": 0.2187, "step": 36470 }, { "epoch": 4.73, "learning_rate": 2.7413234041063286e-06, "loss": 0.2266, "step": 36480 }, { "epoch": 4.74, "learning_rate": 2.7403320156013786e-06, "loss": 0.2388, "step": 36490 }, { "epoch": 4.74, "learning_rate": 2.739340554042059e-06, "loss": 0.234, "step": 36500 }, { "epoch": 4.74, "learning_rate": 2.738349019611009e-06, "loss": 0.2311, "step": 36510 }, { "epoch": 4.74, "learning_rate": 2.7373574124908784e-06, "loss": 0.2273, "step": 36520 }, { "epoch": 4.74, "learning_rate": 2.7363657328643297e-06, "loss": 0.2231, "step": 36530 }, { "epoch": 4.74, "learning_rate": 2.7353739809140423e-06, "loss": 0.2277, "step": 36540 }, { "epoch": 4.74, "learning_rate": 2.734382156822706e-06, "loss": 0.2462, "step": 36550 }, { "epoch": 4.74, "learning_rate": 2.733390260773024e-06, "loss": 0.2287, "step": 36560 }, { "epoch": 4.75, "learning_rate": 2.7323982929477143e-06, "loss": 0.2197, "step": 36570 }, { "epoch": 4.75, "learning_rate": 2.7314062535295073e-06, "loss": 0.2287, "step": 36580 }, { "epoch": 4.75, "learning_rate": 2.7304141427011463e-06, "loss": 0.231, "step": 36590 }, { "epoch": 4.75, "learning_rate": 2.729421960645388e-06, "loss": 0.2458, "step": 36600 }, { "epoch": 4.75, "learning_rate": 2.728429707545002e-06, "loss": 0.2286, "step": 36610 }, { "epoch": 4.75, "learning_rate": 2.727437383582772e-06, "loss": 0.2341, "step": 36620 }, { "epoch": 4.75, "learning_rate": 2.7264449889414924e-06, "loss": 0.236, "step": 36630 }, { "epoch": 4.76, "learning_rate": 2.7254525238039747e-06, "loss": 0.227, "step": 36640 }, { "epoch": 4.76, "learning_rate": 2.724459988353039e-06, "loss": 0.2305, "step": 36650 }, { "epoch": 4.76, "learning_rate": 2.7234673827715203e-06, "loss": 0.2346, "step": 36660 }, { "epoch": 4.76, "learning_rate": 2.7224747072422675e-06, "loss": 0.2417, "step": 36670 }, { "epoch": 4.76, "learning_rate": 2.7214819619481405e-06, "loss": 0.2335, "step": 36680 }, { "epoch": 4.76, "learning_rate": 2.7204891470720136e-06, "loss": 0.2401, "step": 36690 }, { "epoch": 4.76, "learning_rate": 2.7194962627967727e-06, "loss": 0.2306, "step": 36700 }, { "epoch": 4.76, "learning_rate": 2.7185033093053168e-06, "loss": 0.2332, "step": 36710 }, { "epoch": 4.77, "learning_rate": 2.7175102867805583e-06, "loss": 0.237, "step": 36720 }, { "epoch": 4.77, "learning_rate": 2.7165171954054213e-06, "loss": 0.2118, "step": 36730 }, { "epoch": 4.77, "learning_rate": 2.715524035362844e-06, "loss": 0.2375, "step": 36740 }, { "epoch": 4.77, "learning_rate": 2.7145308068357756e-06, "loss": 0.2261, "step": 36750 }, { "epoch": 4.77, "learning_rate": 2.7135375100071797e-06, "loss": 0.225, "step": 36760 }, { "epoch": 4.77, "learning_rate": 2.7125441450600303e-06, "loss": 0.2299, "step": 36770 }, { "epoch": 4.77, "learning_rate": 2.7115507121773157e-06, "loss": 0.2228, "step": 36780 }, { "epoch": 4.77, "learning_rate": 2.710557211542037e-06, "loss": 0.2296, "step": 36790 }, { "epoch": 4.78, "learning_rate": 2.709563643337205e-06, "loss": 0.2276, "step": 36800 }, { "epoch": 4.78, "learning_rate": 2.7085700077458463e-06, "loss": 0.2304, "step": 36810 }, { "epoch": 4.78, "learning_rate": 2.7075763049509983e-06, "loss": 0.2522, "step": 36820 }, { "epoch": 4.78, "learning_rate": 2.706582535135711e-06, "loss": 0.2358, "step": 36830 }, { "epoch": 4.78, "learning_rate": 2.7055886984830466e-06, "loss": 0.2142, "step": 36840 }, { "epoch": 4.78, "learning_rate": 2.7045947951760797e-06, "loss": 0.2384, "step": 36850 }, { "epoch": 4.78, "learning_rate": 2.7036008253978973e-06, "loss": 0.2333, "step": 36860 }, { "epoch": 4.79, "learning_rate": 2.7026067893315983e-06, "loss": 0.2228, "step": 36870 }, { "epoch": 4.79, "learning_rate": 2.7016126871602943e-06, "loss": 0.2296, "step": 36880 }, { "epoch": 4.79, "learning_rate": 2.700618519067109e-06, "loss": 0.2213, "step": 36890 }, { "epoch": 4.79, "learning_rate": 2.6996242852351785e-06, "loss": 0.2417, "step": 36900 }, { "epoch": 4.79, "learning_rate": 2.6986299858476493e-06, "loss": 0.2162, "step": 36910 }, { "epoch": 4.79, "learning_rate": 2.697635621087682e-06, "loss": 0.2443, "step": 36920 }, { "epoch": 4.79, "learning_rate": 2.696641191138449e-06, "loss": 0.2233, "step": 36930 }, { "epoch": 4.79, "learning_rate": 2.6956466961831337e-06, "loss": 0.2418, "step": 36940 }, { "epoch": 4.8, "learning_rate": 2.694652136404931e-06, "loss": 0.2385, "step": 36950 }, { "epoch": 4.8, "learning_rate": 2.6936575119870507e-06, "loss": 0.227, "step": 36960 }, { "epoch": 4.8, "learning_rate": 2.692662823112712e-06, "loss": 0.2283, "step": 36970 }, { "epoch": 4.8, "learning_rate": 2.691668069965146e-06, "loss": 0.2274, "step": 36980 }, { "epoch": 4.8, "learning_rate": 2.690673252727596e-06, "loss": 0.228, "step": 36990 }, { "epoch": 4.8, "learning_rate": 2.6896783715833184e-06, "loss": 0.215, "step": 37000 }, { "epoch": 4.8, "learning_rate": 2.6886834267155785e-06, "loss": 0.2169, "step": 37010 }, { "epoch": 4.8, "learning_rate": 2.6876884183076566e-06, "loss": 0.2299, "step": 37020 }, { "epoch": 4.81, "learning_rate": 2.6866933465428425e-06, "loss": 0.239, "step": 37030 }, { "epoch": 4.81, "learning_rate": 2.685698211604439e-06, "loss": 0.2307, "step": 37040 }, { "epoch": 4.81, "learning_rate": 2.6847030136757584e-06, "loss": 0.2331, "step": 37050 }, { "epoch": 4.81, "learning_rate": 2.6837077529401274e-06, "loss": 0.2338, "step": 37060 }, { "epoch": 4.81, "learning_rate": 2.6827124295808836e-06, "loss": 0.2211, "step": 37070 }, { "epoch": 4.81, "learning_rate": 2.6817170437813737e-06, "loss": 0.2309, "step": 37080 }, { "epoch": 4.81, "learning_rate": 2.6807215957249587e-06, "loss": 0.2325, "step": 37090 }, { "epoch": 4.82, "learning_rate": 2.67972608559501e-06, "loss": 0.2536, "step": 37100 }, { "epoch": 4.82, "learning_rate": 2.678730513574911e-06, "loss": 0.2349, "step": 37110 }, { "epoch": 4.82, "learning_rate": 2.6777348798480543e-06, "loss": 0.2261, "step": 37120 }, { "epoch": 4.82, "learning_rate": 2.6767391845978473e-06, "loss": 0.2243, "step": 37130 }, { "epoch": 4.82, "learning_rate": 2.6757434280077068e-06, "loss": 0.225, "step": 37140 }, { "epoch": 4.82, "learning_rate": 2.6747476102610607e-06, "loss": 0.2376, "step": 37150 }, { "epoch": 4.82, "learning_rate": 2.673751731541348e-06, "loss": 0.2262, "step": 37160 }, { "epoch": 4.82, "learning_rate": 2.6727557920320214e-06, "loss": 0.2212, "step": 37170 }, { "epoch": 4.83, "learning_rate": 2.671759791916541e-06, "loss": 0.2298, "step": 37180 }, { "epoch": 4.83, "learning_rate": 2.67076373137838e-06, "loss": 0.2284, "step": 37190 }, { "epoch": 4.83, "learning_rate": 2.669767610601024e-06, "loss": 0.2207, "step": 37200 }, { "epoch": 4.83, "learning_rate": 2.6687714297679677e-06, "loss": 0.2235, "step": 37210 }, { "epoch": 4.83, "learning_rate": 2.6677751890627175e-06, "loss": 0.2237, "step": 37220 }, { "epoch": 4.83, "learning_rate": 2.6667788886687906e-06, "loss": 0.2146, "step": 37230 }, { "epoch": 4.83, "learning_rate": 2.6657825287697167e-06, "loss": 0.2274, "step": 37240 }, { "epoch": 4.83, "learning_rate": 2.664786109549034e-06, "loss": 0.2152, "step": 37250 }, { "epoch": 4.84, "learning_rate": 2.663789631190293e-06, "loss": 0.2254, "step": 37260 }, { "epoch": 4.84, "learning_rate": 2.6627930938770556e-06, "loss": 0.2134, "step": 37270 }, { "epoch": 4.84, "learning_rate": 2.6617964977928935e-06, "loss": 0.2403, "step": 37280 }, { "epoch": 4.84, "learning_rate": 2.6607998431213895e-06, "loss": 0.2073, "step": 37290 }, { "epoch": 4.84, "learning_rate": 2.6598031300461375e-06, "loss": 0.2228, "step": 37300 }, { "epoch": 4.84, "learning_rate": 2.658806358750743e-06, "loss": 0.2219, "step": 37310 }, { "epoch": 4.84, "learning_rate": 2.657809529418819e-06, "loss": 0.2205, "step": 37320 }, { "epoch": 4.84, "learning_rate": 2.6568126422339937e-06, "loss": 0.2231, "step": 37330 }, { "epoch": 4.85, "learning_rate": 2.655815697379902e-06, "loss": 0.219, "step": 37340 }, { "epoch": 4.85, "learning_rate": 2.6548186950401923e-06, "loss": 0.2233, "step": 37350 }, { "epoch": 4.85, "learning_rate": 2.653821635398522e-06, "loss": 0.2246, "step": 37360 }, { "epoch": 4.85, "learning_rate": 2.6528245186385583e-06, "loss": 0.2205, "step": 37370 }, { "epoch": 4.85, "learning_rate": 2.651827344943983e-06, "loss": 0.2394, "step": 37380 }, { "epoch": 4.85, "learning_rate": 2.6508301144984817e-06, "loss": 0.2298, "step": 37390 }, { "epoch": 4.85, "learning_rate": 2.6498328274857567e-06, "loss": 0.2255, "step": 37400 }, { "epoch": 4.86, "learning_rate": 2.6488354840895173e-06, "loss": 0.225, "step": 37410 }, { "epoch": 4.86, "learning_rate": 2.647838084493485e-06, "loss": 0.225, "step": 37420 }, { "epoch": 4.86, "learning_rate": 2.64684062888139e-06, "loss": 0.2238, "step": 37430 }, { "epoch": 4.86, "learning_rate": 2.645843117436974e-06, "loss": 0.233, "step": 37440 }, { "epoch": 4.86, "learning_rate": 2.6448455503439885e-06, "loss": 0.2269, "step": 37450 }, { "epoch": 4.86, "learning_rate": 2.6438479277861947e-06, "loss": 0.231, "step": 37460 }, { "epoch": 4.86, "learning_rate": 2.6428502499473644e-06, "loss": 0.229, "step": 37470 }, { "epoch": 4.86, "learning_rate": 2.6418525170112817e-06, "loss": 0.228, "step": 37480 }, { "epoch": 4.87, "learning_rate": 2.640854729161737e-06, "loss": 0.2286, "step": 37490 }, { "epoch": 4.87, "learning_rate": 2.6398568865825342e-06, "loss": 0.2263, "step": 37500 }, { "epoch": 4.87, "learning_rate": 2.6388589894574847e-06, "loss": 0.2263, "step": 37510 }, { "epoch": 4.87, "learning_rate": 2.6378610379704124e-06, "loss": 0.2172, "step": 37520 }, { "epoch": 4.87, "learning_rate": 2.636863032305149e-06, "loss": 0.2341, "step": 37530 }, { "epoch": 4.87, "learning_rate": 2.6358649726455365e-06, "loss": 0.2158, "step": 37540 }, { "epoch": 4.87, "learning_rate": 2.634866859175429e-06, "loss": 0.227, "step": 37550 }, { "epoch": 4.87, "learning_rate": 2.6338686920786883e-06, "loss": 0.2393, "step": 37560 }, { "epoch": 4.88, "learning_rate": 2.6328704715391864e-06, "loss": 0.2214, "step": 37570 }, { "epoch": 4.88, "learning_rate": 2.6318721977408057e-06, "loss": 0.2268, "step": 37580 }, { "epoch": 4.88, "learning_rate": 2.6308738708674388e-06, "loss": 0.2305, "step": 37590 }, { "epoch": 4.88, "learning_rate": 2.6298754911029867e-06, "loss": 0.2186, "step": 37600 }, { "epoch": 4.88, "learning_rate": 2.6288770586313605e-06, "loss": 0.2345, "step": 37610 }, { "epoch": 4.88, "learning_rate": 2.6278785736364825e-06, "loss": 0.2481, "step": 37620 }, { "epoch": 4.88, "learning_rate": 2.6268800363022842e-06, "loss": 0.2234, "step": 37630 }, { "epoch": 4.89, "learning_rate": 2.6258814468127044e-06, "loss": 0.2232, "step": 37640 }, { "epoch": 4.89, "learning_rate": 2.624882805351694e-06, "loss": 0.2414, "step": 37650 }, { "epoch": 4.89, "learning_rate": 2.6238841121032135e-06, "loss": 0.2256, "step": 37660 }, { "epoch": 4.89, "learning_rate": 2.6228853672512307e-06, "loss": 0.2288, "step": 37670 }, { "epoch": 4.89, "learning_rate": 2.6218865709797253e-06, "loss": 0.2248, "step": 37680 }, { "epoch": 4.89, "learning_rate": 2.620887723472685e-06, "loss": 0.2374, "step": 37690 }, { "epoch": 4.89, "learning_rate": 2.6198888249141084e-06, "loss": 0.2187, "step": 37700 }, { "epoch": 4.89, "learning_rate": 2.618889875488002e-06, "loss": 0.2337, "step": 37710 }, { "epoch": 4.9, "learning_rate": 2.6178908753783823e-06, "loss": 0.2268, "step": 37720 }, { "epoch": 4.9, "learning_rate": 2.6168918247692756e-06, "loss": 0.2174, "step": 37730 }, { "epoch": 4.9, "learning_rate": 2.6158927238447152e-06, "loss": 0.2325, "step": 37740 }, { "epoch": 4.9, "learning_rate": 2.614893572788748e-06, "loss": 0.2149, "step": 37750 }, { "epoch": 4.9, "learning_rate": 2.613894371785426e-06, "loss": 0.2191, "step": 37760 }, { "epoch": 4.9, "learning_rate": 2.612895121018812e-06, "loss": 0.2143, "step": 37770 }, { "epoch": 4.9, "learning_rate": 2.6118958206729787e-06, "loss": 0.2386, "step": 37780 }, { "epoch": 4.9, "learning_rate": 2.6108964709320073e-06, "loss": 0.228, "step": 37790 }, { "epoch": 4.91, "learning_rate": 2.6098970719799873e-06, "loss": 0.225, "step": 37800 }, { "epoch": 4.91, "learning_rate": 2.6088976240010185e-06, "loss": 0.2198, "step": 37810 }, { "epoch": 4.91, "learning_rate": 2.6078981271792086e-06, "loss": 0.2447, "step": 37820 }, { "epoch": 4.91, "learning_rate": 2.606898581698675e-06, "loss": 0.2169, "step": 37830 }, { "epoch": 4.91, "learning_rate": 2.6058989877435447e-06, "loss": 0.2287, "step": 37840 }, { "epoch": 4.91, "learning_rate": 2.604899345497952e-06, "loss": 0.2369, "step": 37850 }, { "epoch": 4.91, "learning_rate": 2.6038996551460415e-06, "loss": 0.222, "step": 37860 }, { "epoch": 4.91, "learning_rate": 2.6028999168719664e-06, "loss": 0.2263, "step": 37870 }, { "epoch": 4.92, "learning_rate": 2.601900130859888e-06, "loss": 0.227, "step": 37880 }, { "epoch": 4.92, "learning_rate": 2.6009002972939764e-06, "loss": 0.2383, "step": 37890 }, { "epoch": 4.92, "learning_rate": 2.599900416358412e-06, "loss": 0.2132, "step": 37900 }, { "epoch": 4.92, "learning_rate": 2.5989004882373826e-06, "loss": 0.2165, "step": 37910 }, { "epoch": 4.92, "learning_rate": 2.5979005131150845e-06, "loss": 0.2346, "step": 37920 }, { "epoch": 4.92, "learning_rate": 2.5969004911757228e-06, "loss": 0.2397, "step": 37930 }, { "epoch": 4.92, "learning_rate": 2.595900422603513e-06, "loss": 0.2338, "step": 37940 }, { "epoch": 4.93, "learning_rate": 2.5949003075826764e-06, "loss": 0.226, "step": 37950 }, { "epoch": 4.93, "learning_rate": 2.593900146297444e-06, "loss": 0.2419, "step": 37960 }, { "epoch": 4.93, "learning_rate": 2.592899938932056e-06, "loss": 0.2329, "step": 37970 }, { "epoch": 4.93, "learning_rate": 2.591899685670762e-06, "loss": 0.2361, "step": 37980 }, { "epoch": 4.93, "learning_rate": 2.590899386697816e-06, "loss": 0.2431, "step": 37990 }, { "epoch": 4.93, "learning_rate": 2.589899042197485e-06, "loss": 0.2338, "step": 38000 }, { "epoch": 4.93, "learning_rate": 2.5888986523540416e-06, "loss": 0.2186, "step": 38010 }, { "epoch": 4.93, "learning_rate": 2.5878982173517677e-06, "loss": 0.2294, "step": 38020 }, { "epoch": 4.94, "learning_rate": 2.586897737374953e-06, "loss": 0.2173, "step": 38030 }, { "epoch": 4.94, "learning_rate": 2.585897212607897e-06, "loss": 0.2167, "step": 38040 }, { "epoch": 4.94, "learning_rate": 2.584896643234906e-06, "loss": 0.2309, "step": 38050 }, { "epoch": 4.94, "learning_rate": 2.5838960294402943e-06, "loss": 0.2308, "step": 38060 }, { "epoch": 4.94, "learning_rate": 2.5828953714083856e-06, "loss": 0.2084, "step": 38070 }, { "epoch": 4.94, "learning_rate": 2.5818946693235105e-06, "loss": 0.2477, "step": 38080 }, { "epoch": 4.94, "learning_rate": 2.5808939233700088e-06, "loss": 0.2347, "step": 38090 }, { "epoch": 4.94, "learning_rate": 2.5798931337322274e-06, "loss": 0.2278, "step": 38100 }, { "epoch": 4.95, "learning_rate": 2.5788923005945227e-06, "loss": 0.2273, "step": 38110 }, { "epoch": 4.95, "learning_rate": 2.577891424141257e-06, "loss": 0.2433, "step": 38120 }, { "epoch": 4.95, "learning_rate": 2.576890504556802e-06, "loss": 0.2394, "step": 38130 }, { "epoch": 4.95, "learning_rate": 2.5758895420255378e-06, "loss": 0.2244, "step": 38140 }, { "epoch": 4.95, "learning_rate": 2.5748885367318517e-06, "loss": 0.2164, "step": 38150 }, { "epoch": 4.95, "learning_rate": 2.5738874888601378e-06, "loss": 0.2401, "step": 38160 }, { "epoch": 4.95, "learning_rate": 2.5728863985947996e-06, "loss": 0.2144, "step": 38170 }, { "epoch": 4.96, "learning_rate": 2.571885266120249e-06, "loss": 0.2305, "step": 38180 }, { "epoch": 4.96, "learning_rate": 2.570884091620903e-06, "loss": 0.2273, "step": 38190 }, { "epoch": 4.96, "learning_rate": 2.5698828752811882e-06, "loss": 0.2421, "step": 38200 }, { "epoch": 4.96, "learning_rate": 2.5688816172855393e-06, "loss": 0.2346, "step": 38210 }, { "epoch": 4.96, "learning_rate": 2.5678803178183982e-06, "loss": 0.2267, "step": 38220 }, { "epoch": 4.96, "learning_rate": 2.5668789770642134e-06, "loss": 0.2306, "step": 38230 }, { "epoch": 4.96, "learning_rate": 2.5658775952074426e-06, "loss": 0.2259, "step": 38240 }, { "epoch": 4.96, "learning_rate": 2.5648761724325503e-06, "loss": 0.2315, "step": 38250 }, { "epoch": 4.97, "learning_rate": 2.5638747089240078e-06, "loss": 0.2392, "step": 38260 }, { "epoch": 4.97, "learning_rate": 2.562873204866295e-06, "loss": 0.2327, "step": 38270 }, { "epoch": 4.97, "learning_rate": 2.5618716604439e-06, "loss": 0.2135, "step": 38280 }, { "epoch": 4.97, "learning_rate": 2.5608700758413166e-06, "loss": 0.2284, "step": 38290 }, { "epoch": 4.97, "learning_rate": 2.5598684512430458e-06, "loss": 0.2234, "step": 38300 }, { "epoch": 4.97, "learning_rate": 2.5588667868335975e-06, "loss": 0.2418, "step": 38310 }, { "epoch": 4.97, "learning_rate": 2.5578650827974894e-06, "loss": 0.2358, "step": 38320 }, { "epoch": 4.97, "learning_rate": 2.5568633393192435e-06, "loss": 0.2219, "step": 38330 }, { "epoch": 4.98, "learning_rate": 2.555861556583392e-06, "loss": 0.235, "step": 38340 }, { "epoch": 4.98, "learning_rate": 2.554859734774473e-06, "loss": 0.2359, "step": 38350 }, { "epoch": 4.98, "learning_rate": 2.5538578740770327e-06, "loss": 0.2293, "step": 38360 }, { "epoch": 4.98, "learning_rate": 2.5528559746756226e-06, "loss": 0.2348, "step": 38370 }, { "epoch": 4.98, "learning_rate": 2.551854036754804e-06, "loss": 0.2348, "step": 38380 }, { "epoch": 4.98, "learning_rate": 2.5508520604991427e-06, "loss": 0.2273, "step": 38390 }, { "epoch": 4.98, "learning_rate": 2.5498500460932136e-06, "loss": 0.2292, "step": 38400 }, { "epoch": 4.99, "learning_rate": 2.5488479937215967e-06, "loss": 0.2288, "step": 38410 }, { "epoch": 4.99, "learning_rate": 2.5478459035688815e-06, "loss": 0.2146, "step": 38420 }, { "epoch": 4.99, "learning_rate": 2.5468437758196628e-06, "loss": 0.2323, "step": 38430 }, { "epoch": 4.99, "learning_rate": 2.545841610658541e-06, "loss": 0.2359, "step": 38440 }, { "epoch": 4.99, "learning_rate": 2.544839408270126e-06, "loss": 0.218, "step": 38450 }, { "epoch": 4.99, "learning_rate": 2.543837168839034e-06, "loss": 0.2274, "step": 38460 }, { "epoch": 4.99, "learning_rate": 2.5428348925498866e-06, "loss": 0.2246, "step": 38470 }, { "epoch": 4.99, "learning_rate": 2.5418325795873132e-06, "loss": 0.2382, "step": 38480 }, { "epoch": 5.0, "learning_rate": 2.54083023013595e-06, "loss": 0.2239, "step": 38490 }, { "epoch": 5.0, "learning_rate": 2.5398278443804406e-06, "loss": 0.216, "step": 38500 }, { "epoch": 5.0, "learning_rate": 2.5388254225054333e-06, "loss": 0.2358, "step": 38510 }, { "epoch": 5.0, "learning_rate": 2.5378229646955847e-06, "loss": 0.2173, "step": 38520 }, { "epoch": 5.0, "eval_loss": 0.22834357619285583, "eval_runtime": 61.7471, "eval_samples_per_second": 176.154, "eval_steps_per_second": 22.025, "step": 38525 }, { "epoch": 5.0, "learning_rate": 2.5368204711355577e-06, "loss": 0.2192, "step": 38530 }, { "epoch": 5.0, "learning_rate": 2.5358179420100213e-06, "loss": 0.2123, "step": 38540 }, { "epoch": 5.0, "learning_rate": 2.5348153775036518e-06, "loss": 0.2111, "step": 38550 }, { "epoch": 5.0, "learning_rate": 2.5338127778011312e-06, "loss": 0.2339, "step": 38560 }, { "epoch": 5.01, "learning_rate": 2.5328101430871484e-06, "loss": 0.2376, "step": 38570 }, { "epoch": 5.01, "learning_rate": 2.531807473546399e-06, "loss": 0.2274, "step": 38580 }, { "epoch": 5.01, "learning_rate": 2.5308047693635845e-06, "loss": 0.2015, "step": 38590 }, { "epoch": 5.01, "learning_rate": 2.529802030723413e-06, "loss": 0.2368, "step": 38600 }, { "epoch": 5.01, "learning_rate": 2.5287992578105987e-06, "loss": 0.2266, "step": 38610 }, { "epoch": 5.01, "learning_rate": 2.5277964508098632e-06, "loss": 0.24, "step": 38620 }, { "epoch": 5.01, "learning_rate": 2.5267936099059328e-06, "loss": 0.2289, "step": 38630 }, { "epoch": 5.01, "learning_rate": 2.525790735283542e-06, "loss": 0.2288, "step": 38640 }, { "epoch": 5.02, "learning_rate": 2.524787827127429e-06, "loss": 0.2199, "step": 38650 }, { "epoch": 5.02, "learning_rate": 2.5237848856223397e-06, "loss": 0.2241, "step": 38660 }, { "epoch": 5.02, "learning_rate": 2.5227819109530265e-06, "loss": 0.225, "step": 38670 }, { "epoch": 5.02, "learning_rate": 2.521778903304247e-06, "loss": 0.2211, "step": 38680 }, { "epoch": 5.02, "learning_rate": 2.520775862860765e-06, "loss": 0.2282, "step": 38690 }, { "epoch": 5.02, "learning_rate": 2.5197727898073506e-06, "loss": 0.2393, "step": 38700 }, { "epoch": 5.02, "learning_rate": 2.5187696843287814e-06, "loss": 0.2344, "step": 38710 }, { "epoch": 5.03, "learning_rate": 2.517766546609837e-06, "loss": 0.2268, "step": 38720 }, { "epoch": 5.03, "learning_rate": 2.516763376835307e-06, "loss": 0.217, "step": 38730 }, { "epoch": 5.03, "learning_rate": 2.5157601751899846e-06, "loss": 0.2323, "step": 38740 }, { "epoch": 5.03, "learning_rate": 2.5147569418586704e-06, "loss": 0.2278, "step": 38750 }, { "epoch": 5.03, "learning_rate": 2.5137536770261686e-06, "loss": 0.2245, "step": 38760 }, { "epoch": 5.03, "learning_rate": 2.512750380877292e-06, "loss": 0.2364, "step": 38770 }, { "epoch": 5.03, "learning_rate": 2.5117470535968575e-06, "loss": 0.227, "step": 38780 }, { "epoch": 5.03, "learning_rate": 2.510743695369687e-06, "loss": 0.2145, "step": 38790 }, { "epoch": 5.04, "learning_rate": 2.5097403063806098e-06, "loss": 0.2175, "step": 38800 }, { "epoch": 5.04, "learning_rate": 2.5087368868144607e-06, "loss": 0.2196, "step": 38810 }, { "epoch": 5.04, "learning_rate": 2.507733436856079e-06, "loss": 0.2357, "step": 38820 }, { "epoch": 5.04, "learning_rate": 2.5067299566903097e-06, "loss": 0.2296, "step": 38830 }, { "epoch": 5.04, "learning_rate": 2.5057264465020047e-06, "loss": 0.2355, "step": 38840 }, { "epoch": 5.04, "learning_rate": 2.5047229064760213e-06, "loss": 0.2284, "step": 38850 }, { "epoch": 5.04, "learning_rate": 2.50371933679722e-06, "loss": 0.2338, "step": 38860 }, { "epoch": 5.04, "learning_rate": 2.502715737650469e-06, "loss": 0.2394, "step": 38870 }, { "epoch": 5.05, "learning_rate": 2.501712109220641e-06, "loss": 0.2205, "step": 38880 }, { "epoch": 5.05, "learning_rate": 2.500708451692615e-06, "loss": 0.2358, "step": 38890 }, { "epoch": 5.05, "learning_rate": 2.4997047652512742e-06, "loss": 0.2475, "step": 38900 }, { "epoch": 5.05, "learning_rate": 2.498701050081508e-06, "loss": 0.2379, "step": 38910 }, { "epoch": 5.05, "learning_rate": 2.4976973063682107e-06, "loss": 0.2371, "step": 38920 }, { "epoch": 5.05, "learning_rate": 2.4966935342962818e-06, "loss": 0.2242, "step": 38930 }, { "epoch": 5.05, "learning_rate": 2.4956897340506262e-06, "loss": 0.2249, "step": 38940 }, { "epoch": 5.06, "learning_rate": 2.494685905816155e-06, "loss": 0.2409, "step": 38950 }, { "epoch": 5.06, "learning_rate": 2.493682049777781e-06, "loss": 0.2241, "step": 38960 }, { "epoch": 5.06, "learning_rate": 2.492678166120426e-06, "loss": 0.2191, "step": 38970 }, { "epoch": 5.06, "learning_rate": 2.4916742550290152e-06, "loss": 0.2336, "step": 38980 }, { "epoch": 5.06, "learning_rate": 2.4906703166884803e-06, "loss": 0.2324, "step": 38990 }, { "epoch": 5.06, "learning_rate": 2.4896663512837544e-06, "loss": 0.242, "step": 39000 }, { "epoch": 5.06, "learning_rate": 2.4886623589997793e-06, "loss": 0.2303, "step": 39010 }, { "epoch": 5.06, "learning_rate": 2.4876583400215007e-06, "loss": 0.2226, "step": 39020 }, { "epoch": 5.07, "learning_rate": 2.486654294533868e-06, "loss": 0.2198, "step": 39030 }, { "epoch": 5.07, "learning_rate": 2.485650222721837e-06, "loss": 0.2425, "step": 39040 }, { "epoch": 5.07, "learning_rate": 2.484646124770368e-06, "loss": 0.2383, "step": 39050 }, { "epoch": 5.07, "learning_rate": 2.4836420008644248e-06, "loss": 0.2355, "step": 39060 }, { "epoch": 5.07, "learning_rate": 2.4826378511889776e-06, "loss": 0.2338, "step": 39070 }, { "epoch": 5.07, "learning_rate": 2.4816336759290016e-06, "loss": 0.2379, "step": 39080 }, { "epoch": 5.07, "learning_rate": 2.480629475269475e-06, "loss": 0.2345, "step": 39090 }, { "epoch": 5.07, "learning_rate": 2.4796252493953814e-06, "loss": 0.2271, "step": 39100 }, { "epoch": 5.08, "learning_rate": 2.4786209984917103e-06, "loss": 0.2405, "step": 39110 }, { "epoch": 5.08, "learning_rate": 2.477616722743454e-06, "loss": 0.2298, "step": 39120 }, { "epoch": 5.08, "learning_rate": 2.4766124223356106e-06, "loss": 0.2292, "step": 39130 }, { "epoch": 5.08, "learning_rate": 2.475608097453181e-06, "loss": 0.2235, "step": 39140 }, { "epoch": 5.08, "learning_rate": 2.474603748281173e-06, "loss": 0.2315, "step": 39150 }, { "epoch": 5.08, "learning_rate": 2.4735993750045984e-06, "loss": 0.2198, "step": 39160 }, { "epoch": 5.08, "learning_rate": 2.4725949778084715e-06, "loss": 0.2201, "step": 39170 }, { "epoch": 5.09, "learning_rate": 2.4715905568778127e-06, "loss": 0.2199, "step": 39180 }, { "epoch": 5.09, "learning_rate": 2.4705861123976465e-06, "loss": 0.228, "step": 39190 }, { "epoch": 5.09, "learning_rate": 2.469581644553001e-06, "loss": 0.2247, "step": 39200 }, { "epoch": 5.09, "learning_rate": 2.46857715352891e-06, "loss": 0.231, "step": 39210 }, { "epoch": 5.09, "learning_rate": 2.46757263951041e-06, "loss": 0.237, "step": 39220 }, { "epoch": 5.09, "learning_rate": 2.466568102682544e-06, "loss": 0.2346, "step": 39230 }, { "epoch": 5.09, "learning_rate": 2.4655635432303563e-06, "loss": 0.2214, "step": 39240 }, { "epoch": 5.09, "learning_rate": 2.4645589613388972e-06, "loss": 0.2345, "step": 39250 }, { "epoch": 5.1, "learning_rate": 2.463554357193221e-06, "loss": 0.236, "step": 39260 }, { "epoch": 5.1, "learning_rate": 2.4625497309783852e-06, "loss": 0.2259, "step": 39270 }, { "epoch": 5.1, "learning_rate": 2.4615450828794518e-06, "loss": 0.2185, "step": 39280 }, { "epoch": 5.1, "learning_rate": 2.4605404130814883e-06, "loss": 0.2335, "step": 39290 }, { "epoch": 5.1, "learning_rate": 2.4595357217695644e-06, "loss": 0.2447, "step": 39300 }, { "epoch": 5.1, "learning_rate": 2.4585310091287532e-06, "loss": 0.2182, "step": 39310 }, { "epoch": 5.1, "learning_rate": 2.457526275344134e-06, "loss": 0.2292, "step": 39320 }, { "epoch": 5.1, "learning_rate": 2.456521520600789e-06, "loss": 0.2391, "step": 39330 }, { "epoch": 5.11, "learning_rate": 2.455516745083803e-06, "loss": 0.2204, "step": 39340 }, { "epoch": 5.11, "learning_rate": 2.454511948978266e-06, "loss": 0.2371, "step": 39350 }, { "epoch": 5.11, "learning_rate": 2.4535071324692714e-06, "loss": 0.2272, "step": 39360 }, { "epoch": 5.11, "learning_rate": 2.4525022957419176e-06, "loss": 0.225, "step": 39370 }, { "epoch": 5.11, "learning_rate": 2.4514974389813046e-06, "loss": 0.2278, "step": 39380 }, { "epoch": 5.11, "learning_rate": 2.450492562372536e-06, "loss": 0.2302, "step": 39390 }, { "epoch": 5.11, "learning_rate": 2.4494876661007223e-06, "loss": 0.2212, "step": 39400 }, { "epoch": 5.11, "learning_rate": 2.4484827503509743e-06, "loss": 0.2257, "step": 39410 }, { "epoch": 5.12, "learning_rate": 2.447477815308407e-06, "loss": 0.2155, "step": 39420 }, { "epoch": 5.12, "learning_rate": 2.44647286115814e-06, "loss": 0.2322, "step": 39430 }, { "epoch": 5.12, "learning_rate": 2.4454678880852968e-06, "loss": 0.2326, "step": 39440 }, { "epoch": 5.12, "learning_rate": 2.4444628962750017e-06, "loss": 0.2261, "step": 39450 }, { "epoch": 5.12, "learning_rate": 2.4434578859123853e-06, "loss": 0.2404, "step": 39460 }, { "epoch": 5.12, "learning_rate": 2.4424528571825812e-06, "loss": 0.2489, "step": 39470 }, { "epoch": 5.12, "learning_rate": 2.4414478102707245e-06, "loss": 0.2193, "step": 39480 }, { "epoch": 5.13, "learning_rate": 2.440442745361955e-06, "loss": 0.2283, "step": 39490 }, { "epoch": 5.13, "learning_rate": 2.439437662641416e-06, "loss": 0.2464, "step": 39500 }, { "epoch": 5.13, "learning_rate": 2.4384325622942537e-06, "loss": 0.222, "step": 39510 }, { "epoch": 5.13, "learning_rate": 2.437427444505618e-06, "loss": 0.239, "step": 39520 }, { "epoch": 5.13, "learning_rate": 2.4364223094606614e-06, "loss": 0.2213, "step": 39530 }, { "epoch": 5.13, "learning_rate": 2.43541715734454e-06, "loss": 0.2205, "step": 39540 }, { "epoch": 5.13, "learning_rate": 2.434411988342412e-06, "loss": 0.2355, "step": 39550 }, { "epoch": 5.13, "learning_rate": 2.43340680263944e-06, "loss": 0.2349, "step": 39560 }, { "epoch": 5.14, "learning_rate": 2.4324016004207895e-06, "loss": 0.2254, "step": 39570 }, { "epoch": 5.14, "learning_rate": 2.431396381871629e-06, "loss": 0.223, "step": 39580 }, { "epoch": 5.14, "learning_rate": 2.4303911471771296e-06, "loss": 0.2301, "step": 39590 }, { "epoch": 5.14, "learning_rate": 2.4293858965224652e-06, "loss": 0.2229, "step": 39600 }, { "epoch": 5.14, "learning_rate": 2.4283806300928142e-06, "loss": 0.2446, "step": 39610 }, { "epoch": 5.14, "learning_rate": 2.4273753480733554e-06, "loss": 0.2322, "step": 39620 }, { "epoch": 5.14, "learning_rate": 2.4263700506492713e-06, "loss": 0.2128, "step": 39630 }, { "epoch": 5.14, "learning_rate": 2.4253647380057495e-06, "loss": 0.2238, "step": 39640 }, { "epoch": 5.15, "learning_rate": 2.424359410327978e-06, "loss": 0.2292, "step": 39650 }, { "epoch": 5.15, "learning_rate": 2.4233540678011465e-06, "loss": 0.2298, "step": 39660 }, { "epoch": 5.15, "learning_rate": 2.4223487106104515e-06, "loss": 0.2119, "step": 39670 }, { "epoch": 5.15, "learning_rate": 2.4213433389410895e-06, "loss": 0.2243, "step": 39680 }, { "epoch": 5.15, "learning_rate": 2.4203379529782586e-06, "loss": 0.2252, "step": 39690 }, { "epoch": 5.15, "learning_rate": 2.4193325529071623e-06, "loss": 0.2346, "step": 39700 }, { "epoch": 5.15, "learning_rate": 2.4183271389130043e-06, "loss": 0.2457, "step": 39710 }, { "epoch": 5.16, "learning_rate": 2.417321711180993e-06, "loss": 0.2318, "step": 39720 }, { "epoch": 5.16, "learning_rate": 2.416316269896337e-06, "loss": 0.2279, "step": 39730 }, { "epoch": 5.16, "learning_rate": 2.4153108152442495e-06, "loss": 0.2173, "step": 39740 }, { "epoch": 5.16, "learning_rate": 2.4143053474099452e-06, "loss": 0.241, "step": 39750 }, { "epoch": 5.16, "learning_rate": 2.4132998665786412e-06, "loss": 0.2325, "step": 39760 }, { "epoch": 5.16, "learning_rate": 2.4122943729355573e-06, "loss": 0.2245, "step": 39770 }, { "epoch": 5.16, "learning_rate": 2.4112888666659146e-06, "loss": 0.2291, "step": 39780 }, { "epoch": 5.16, "learning_rate": 2.4102833479549385e-06, "loss": 0.2098, "step": 39790 }, { "epoch": 5.17, "learning_rate": 2.409277816987855e-06, "loss": 0.2412, "step": 39800 }, { "epoch": 5.17, "learning_rate": 2.4082722739498932e-06, "loss": 0.2316, "step": 39810 }, { "epoch": 5.17, "learning_rate": 2.407266719026284e-06, "loss": 0.2259, "step": 39820 }, { "epoch": 5.17, "learning_rate": 2.406261152402261e-06, "loss": 0.2237, "step": 39830 }, { "epoch": 5.17, "learning_rate": 2.4052555742630584e-06, "loss": 0.2094, "step": 39840 }, { "epoch": 5.17, "learning_rate": 2.4042499847939153e-06, "loss": 0.2341, "step": 39850 }, { "epoch": 5.17, "learning_rate": 2.403244384180071e-06, "loss": 0.226, "step": 39860 }, { "epoch": 5.17, "learning_rate": 2.4022387726067666e-06, "loss": 0.2176, "step": 39870 }, { "epoch": 5.18, "learning_rate": 2.401233150259246e-06, "loss": 0.2225, "step": 39880 }, { "epoch": 5.18, "learning_rate": 2.4002275173227556e-06, "loss": 0.2233, "step": 39890 }, { "epoch": 5.18, "learning_rate": 2.399221873982541e-06, "loss": 0.2387, "step": 39900 }, { "epoch": 5.18, "learning_rate": 2.398216220423854e-06, "loss": 0.241, "step": 39910 }, { "epoch": 5.18, "learning_rate": 2.3972105568319455e-06, "loss": 0.2246, "step": 39920 }, { "epoch": 5.18, "learning_rate": 2.3962048833920683e-06, "loss": 0.2408, "step": 39930 }, { "epoch": 5.18, "learning_rate": 2.3951992002894773e-06, "loss": 0.2309, "step": 39940 }, { "epoch": 5.18, "learning_rate": 2.3941935077094296e-06, "loss": 0.2398, "step": 39950 }, { "epoch": 5.19, "learning_rate": 2.393187805837185e-06, "loss": 0.2198, "step": 39960 }, { "epoch": 5.19, "learning_rate": 2.392182094858001e-06, "loss": 0.2271, "step": 39970 }, { "epoch": 5.19, "learning_rate": 2.3911763749571427e-06, "loss": 0.2149, "step": 39980 }, { "epoch": 5.19, "learning_rate": 2.390170646319872e-06, "loss": 0.2212, "step": 39990 }, { "epoch": 5.19, "learning_rate": 2.389164909131455e-06, "loss": 0.2213, "step": 40000 }, { "epoch": 5.19, "learning_rate": 2.388159163577157e-06, "loss": 0.238, "step": 40010 }, { "epoch": 5.19, "learning_rate": 2.387153409842248e-06, "loss": 0.2311, "step": 40020 }, { "epoch": 5.2, "learning_rate": 2.3861476481119985e-06, "loss": 0.2282, "step": 40030 }, { "epoch": 5.2, "learning_rate": 2.385141878571677e-06, "loss": 0.2208, "step": 40040 }, { "epoch": 5.2, "learning_rate": 2.384136101406559e-06, "loss": 0.2278, "step": 40050 }, { "epoch": 5.2, "learning_rate": 2.383130316801918e-06, "loss": 0.2377, "step": 40060 }, { "epoch": 5.2, "learning_rate": 2.3821245249430285e-06, "loss": 0.2493, "step": 40070 }, { "epoch": 5.2, "learning_rate": 2.3811187260151686e-06, "loss": 0.2198, "step": 40080 }, { "epoch": 5.2, "learning_rate": 2.380112920203616e-06, "loss": 0.2445, "step": 40090 }, { "epoch": 5.2, "learning_rate": 2.3791071076936504e-06, "loss": 0.2375, "step": 40100 }, { "epoch": 5.21, "learning_rate": 2.3781012886705525e-06, "loss": 0.241, "step": 40110 }, { "epoch": 5.21, "learning_rate": 2.3770954633196036e-06, "loss": 0.2257, "step": 40120 }, { "epoch": 5.21, "learning_rate": 2.376089631826088e-06, "loss": 0.2311, "step": 40130 }, { "epoch": 5.21, "learning_rate": 2.375083794375289e-06, "loss": 0.2346, "step": 40140 }, { "epoch": 5.21, "learning_rate": 2.3740779511524914e-06, "loss": 0.2271, "step": 40150 }, { "epoch": 5.21, "learning_rate": 2.373072102342983e-06, "loss": 0.2346, "step": 40160 }, { "epoch": 5.21, "learning_rate": 2.3720662481320504e-06, "loss": 0.2205, "step": 40170 }, { "epoch": 5.21, "learning_rate": 2.3710603887049816e-06, "loss": 0.2301, "step": 40180 }, { "epoch": 5.22, "learning_rate": 2.370054524247067e-06, "loss": 0.2234, "step": 40190 }, { "epoch": 5.22, "learning_rate": 2.369048654943596e-06, "loss": 0.2276, "step": 40200 }, { "epoch": 5.22, "learning_rate": 2.36804278097986e-06, "loss": 0.2229, "step": 40210 }, { "epoch": 5.22, "learning_rate": 2.3670369025411518e-06, "loss": 0.2237, "step": 40220 }, { "epoch": 5.22, "learning_rate": 2.3660310198127627e-06, "loss": 0.2412, "step": 40230 }, { "epoch": 5.22, "learning_rate": 2.365025132979988e-06, "loss": 0.2327, "step": 40240 }, { "epoch": 5.22, "learning_rate": 2.3640192422281207e-06, "loss": 0.2204, "step": 40250 }, { "epoch": 5.23, "learning_rate": 2.363013347742456e-06, "loss": 0.2524, "step": 40260 }, { "epoch": 5.23, "learning_rate": 2.362007449708291e-06, "loss": 0.2275, "step": 40270 }, { "epoch": 5.23, "learning_rate": 2.3610015483109217e-06, "loss": 0.2137, "step": 40280 }, { "epoch": 5.23, "learning_rate": 2.359995643735644e-06, "loss": 0.2313, "step": 40290 }, { "epoch": 5.23, "learning_rate": 2.358989736167757e-06, "loss": 0.2149, "step": 40300 }, { "epoch": 5.23, "learning_rate": 2.3579838257925584e-06, "loss": 0.2339, "step": 40310 }, { "epoch": 5.23, "learning_rate": 2.3569779127953465e-06, "loss": 0.2267, "step": 40320 }, { "epoch": 5.23, "learning_rate": 2.3559719973614213e-06, "loss": 0.2271, "step": 40330 }, { "epoch": 5.24, "learning_rate": 2.354966079676082e-06, "loss": 0.2203, "step": 40340 }, { "epoch": 5.24, "learning_rate": 2.3539601599246286e-06, "loss": 0.2224, "step": 40350 }, { "epoch": 5.24, "learning_rate": 2.3529542382923623e-06, "loss": 0.2274, "step": 40360 }, { "epoch": 5.24, "learning_rate": 2.3519483149645825e-06, "loss": 0.2322, "step": 40370 }, { "epoch": 5.24, "learning_rate": 2.350942390126592e-06, "loss": 0.2136, "step": 40380 }, { "epoch": 5.24, "learning_rate": 2.3499364639636914e-06, "loss": 0.2534, "step": 40390 }, { "epoch": 5.24, "learning_rate": 2.348930536661182e-06, "loss": 0.2212, "step": 40400 }, { "epoch": 5.24, "learning_rate": 2.3479246084043667e-06, "loss": 0.2119, "step": 40410 }, { "epoch": 5.25, "learning_rate": 2.3469186793785454e-06, "loss": 0.2228, "step": 40420 }, { "epoch": 5.25, "learning_rate": 2.3459127497690223e-06, "loss": 0.2266, "step": 40430 }, { "epoch": 5.25, "learning_rate": 2.3449068197610994e-06, "loss": 0.2242, "step": 40440 }, { "epoch": 5.25, "learning_rate": 2.343900889540079e-06, "loss": 0.2198, "step": 40450 }, { "epoch": 5.25, "learning_rate": 2.342894959291262e-06, "loss": 0.2171, "step": 40460 }, { "epoch": 5.25, "learning_rate": 2.3418890291999534e-06, "loss": 0.2175, "step": 40470 }, { "epoch": 5.25, "learning_rate": 2.340883099451454e-06, "loss": 0.2293, "step": 40480 }, { "epoch": 5.26, "learning_rate": 2.3398771702310657e-06, "loss": 0.2337, "step": 40490 }, { "epoch": 5.26, "learning_rate": 2.3388712417240915e-06, "loss": 0.2354, "step": 40500 }, { "epoch": 5.26, "learning_rate": 2.337865314115834e-06, "loss": 0.2279, "step": 40510 }, { "epoch": 5.26, "learning_rate": 2.3368593875915934e-06, "loss": 0.2234, "step": 40520 }, { "epoch": 5.26, "learning_rate": 2.335853462336673e-06, "loss": 0.2116, "step": 40530 }, { "epoch": 5.26, "learning_rate": 2.3348475385363742e-06, "loss": 0.2343, "step": 40540 }, { "epoch": 5.26, "learning_rate": 2.333841616375997e-06, "loss": 0.229, "step": 40550 }, { "epoch": 5.26, "learning_rate": 2.332835696040843e-06, "loss": 0.2335, "step": 40560 }, { "epoch": 5.27, "learning_rate": 2.3318297777162127e-06, "loss": 0.2219, "step": 40570 }, { "epoch": 5.27, "learning_rate": 2.330823861587407e-06, "loss": 0.2314, "step": 40580 }, { "epoch": 5.27, "learning_rate": 2.329817947839725e-06, "loss": 0.2453, "step": 40590 }, { "epoch": 5.27, "learning_rate": 2.3288120366584657e-06, "loss": 0.2199, "step": 40600 }, { "epoch": 5.27, "learning_rate": 2.3278061282289295e-06, "loss": 0.2345, "step": 40610 }, { "epoch": 5.27, "learning_rate": 2.3268002227364126e-06, "loss": 0.2256, "step": 40620 }, { "epoch": 5.27, "learning_rate": 2.3257943203662142e-06, "loss": 0.2266, "step": 40630 }, { "epoch": 5.27, "learning_rate": 2.3247884213036312e-06, "loss": 0.2231, "step": 40640 }, { "epoch": 5.28, "learning_rate": 2.3237825257339614e-06, "loss": 0.2191, "step": 40650 }, { "epoch": 5.28, "learning_rate": 2.3227766338424986e-06, "loss": 0.2274, "step": 40660 }, { "epoch": 5.28, "learning_rate": 2.32177074581454e-06, "loss": 0.2345, "step": 40670 }, { "epoch": 5.28, "learning_rate": 2.3207648618353797e-06, "loss": 0.2109, "step": 40680 }, { "epoch": 5.28, "learning_rate": 2.319758982090311e-06, "loss": 0.2343, "step": 40690 }, { "epoch": 5.28, "learning_rate": 2.3187531067646273e-06, "loss": 0.2226, "step": 40700 }, { "epoch": 5.28, "learning_rate": 2.317747236043621e-06, "loss": 0.2288, "step": 40710 }, { "epoch": 5.28, "learning_rate": 2.3167413701125844e-06, "loss": 0.2328, "step": 40720 }, { "epoch": 5.29, "learning_rate": 2.3157355091568066e-06, "loss": 0.2385, "step": 40730 }, { "epoch": 5.29, "learning_rate": 2.3147296533615775e-06, "loss": 0.2348, "step": 40740 }, { "epoch": 5.29, "learning_rate": 2.313723802912187e-06, "loss": 0.2263, "step": 40750 }, { "epoch": 5.29, "learning_rate": 2.312717957993921e-06, "loss": 0.2367, "step": 40760 }, { "epoch": 5.29, "learning_rate": 2.3117121187920672e-06, "loss": 0.2284, "step": 40770 }, { "epoch": 5.29, "learning_rate": 2.3107062854919114e-06, "loss": 0.2322, "step": 40780 }, { "epoch": 5.29, "learning_rate": 2.309700458278738e-06, "loss": 0.2339, "step": 40790 }, { "epoch": 5.3, "learning_rate": 2.30869463733783e-06, "loss": 0.2344, "step": 40800 }, { "epoch": 5.3, "learning_rate": 2.30768882285447e-06, "loss": 0.2395, "step": 40810 }, { "epoch": 5.3, "learning_rate": 2.30668301501394e-06, "loss": 0.24, "step": 40820 }, { "epoch": 5.3, "learning_rate": 2.3056772140015174e-06, "loss": 0.2252, "step": 40830 }, { "epoch": 5.3, "learning_rate": 2.304671420002483e-06, "loss": 0.2212, "step": 40840 }, { "epoch": 5.3, "learning_rate": 2.3036656332021133e-06, "loss": 0.2506, "step": 40850 }, { "epoch": 5.3, "learning_rate": 2.3026598537856844e-06, "loss": 0.2244, "step": 40860 }, { "epoch": 5.3, "learning_rate": 2.3016540819384707e-06, "loss": 0.2206, "step": 40870 }, { "epoch": 5.31, "learning_rate": 2.3006483178457453e-06, "loss": 0.2328, "step": 40880 }, { "epoch": 5.31, "learning_rate": 2.299642561692781e-06, "loss": 0.2309, "step": 40890 }, { "epoch": 5.31, "learning_rate": 2.298636813664847e-06, "loss": 0.2266, "step": 40900 }, { "epoch": 5.31, "learning_rate": 2.297631073947212e-06, "loss": 0.2323, "step": 40910 }, { "epoch": 5.31, "learning_rate": 2.2966253427251445e-06, "loss": 0.2157, "step": 40920 }, { "epoch": 5.31, "learning_rate": 2.29561962018391e-06, "loss": 0.2473, "step": 40930 }, { "epoch": 5.31, "learning_rate": 2.2946139065087715e-06, "loss": 0.2365, "step": 40940 }, { "epoch": 5.31, "learning_rate": 2.293608201884992e-06, "loss": 0.2082, "step": 40950 }, { "epoch": 5.32, "learning_rate": 2.2926025064978333e-06, "loss": 0.2333, "step": 40960 }, { "epoch": 5.32, "learning_rate": 2.2915968205325534e-06, "loss": 0.2191, "step": 40970 }, { "epoch": 5.32, "learning_rate": 2.29059114417441e-06, "loss": 0.2526, "step": 40980 }, { "epoch": 5.32, "learning_rate": 2.2895854776086586e-06, "loss": 0.2272, "step": 40990 }, { "epoch": 5.32, "learning_rate": 2.2885798210205536e-06, "loss": 0.2386, "step": 41000 }, { "epoch": 5.32, "learning_rate": 2.2875741745953465e-06, "loss": 0.2263, "step": 41010 }, { "epoch": 5.32, "learning_rate": 2.2865685385182875e-06, "loss": 0.2259, "step": 41020 }, { "epoch": 5.33, "learning_rate": 2.2855629129746252e-06, "loss": 0.233, "step": 41030 }, { "epoch": 5.33, "learning_rate": 2.2845572981496055e-06, "loss": 0.2207, "step": 41040 }, { "epoch": 5.33, "learning_rate": 2.2835516942284722e-06, "loss": 0.2311, "step": 41050 }, { "epoch": 5.33, "learning_rate": 2.282546101396468e-06, "loss": 0.2337, "step": 41060 }, { "epoch": 5.33, "learning_rate": 2.2815405198388343e-06, "loss": 0.2347, "step": 41070 }, { "epoch": 5.33, "learning_rate": 2.2805349497408075e-06, "loss": 0.2188, "step": 41080 }, { "epoch": 5.33, "learning_rate": 2.2795293912876244e-06, "loss": 0.2306, "step": 41090 }, { "epoch": 5.33, "learning_rate": 2.2785238446645193e-06, "loss": 0.2038, "step": 41100 }, { "epoch": 5.34, "learning_rate": 2.2775183100567228e-06, "loss": 0.2321, "step": 41110 }, { "epoch": 5.34, "learning_rate": 2.2765127876494655e-06, "loss": 0.2227, "step": 41120 }, { "epoch": 5.34, "learning_rate": 2.2755072776279742e-06, "loss": 0.2287, "step": 41130 }, { "epoch": 5.34, "learning_rate": 2.2745017801774747e-06, "loss": 0.2379, "step": 41140 }, { "epoch": 5.34, "learning_rate": 2.273496295483188e-06, "loss": 0.2219, "step": 41150 }, { "epoch": 5.34, "learning_rate": 2.272490823730336e-06, "loss": 0.2192, "step": 41160 }, { "epoch": 5.34, "learning_rate": 2.271485365104136e-06, "loss": 0.2399, "step": 41170 }, { "epoch": 5.34, "learning_rate": 2.2704799197898034e-06, "loss": 0.2429, "step": 41180 }, { "epoch": 5.35, "learning_rate": 2.269474487972551e-06, "loss": 0.2357, "step": 41190 }, { "epoch": 5.35, "learning_rate": 2.26846906983759e-06, "loss": 0.22, "step": 41200 }, { "epoch": 5.35, "learning_rate": 2.267463665570129e-06, "loss": 0.2281, "step": 41210 }, { "epoch": 5.35, "learning_rate": 2.266458275355372e-06, "loss": 0.2313, "step": 41220 }, { "epoch": 5.35, "learning_rate": 2.265452899378523e-06, "loss": 0.2315, "step": 41230 }, { "epoch": 5.35, "learning_rate": 2.2644475378247825e-06, "loss": 0.23, "step": 41240 }, { "epoch": 5.35, "learning_rate": 2.2634421908793466e-06, "loss": 0.242, "step": 41250 }, { "epoch": 5.35, "learning_rate": 2.262436858727412e-06, "loss": 0.2428, "step": 41260 }, { "epoch": 5.36, "learning_rate": 2.2614315415541696e-06, "loss": 0.2141, "step": 41270 }, { "epoch": 5.36, "learning_rate": 2.2604262395448103e-06, "loss": 0.226, "step": 41280 }, { "epoch": 5.36, "learning_rate": 2.2594209528845193e-06, "loss": 0.2254, "step": 41290 }, { "epoch": 5.36, "learning_rate": 2.2584156817584814e-06, "loss": 0.237, "step": 41300 }, { "epoch": 5.36, "learning_rate": 2.2574104263518775e-06, "loss": 0.2399, "step": 41310 }, { "epoch": 5.36, "learning_rate": 2.256405186849885e-06, "loss": 0.2519, "step": 41320 }, { "epoch": 5.36, "learning_rate": 2.2553999634376797e-06, "loss": 0.219, "step": 41330 }, { "epoch": 5.37, "learning_rate": 2.2543947563004337e-06, "loss": 0.2224, "step": 41340 }, { "epoch": 5.37, "learning_rate": 2.253389565623316e-06, "loss": 0.2249, "step": 41350 }, { "epoch": 5.37, "learning_rate": 2.252384391591493e-06, "loss": 0.2344, "step": 41360 }, { "epoch": 5.37, "learning_rate": 2.2513792343901273e-06, "loss": 0.2263, "step": 41370 }, { "epoch": 5.37, "learning_rate": 2.2503740942043797e-06, "loss": 0.227, "step": 41380 }, { "epoch": 5.37, "learning_rate": 2.2493689712194063e-06, "loss": 0.2335, "step": 41390 }, { "epoch": 5.37, "learning_rate": 2.248363865620361e-06, "loss": 0.2112, "step": 41400 }, { "epoch": 5.37, "learning_rate": 2.247358777592395e-06, "loss": 0.2211, "step": 41410 }, { "epoch": 5.38, "learning_rate": 2.2463537073206546e-06, "loss": 0.2084, "step": 41420 }, { "epoch": 5.38, "learning_rate": 2.245348654990284e-06, "loss": 0.2345, "step": 41430 }, { "epoch": 5.38, "learning_rate": 2.244343620786424e-06, "loss": 0.2165, "step": 41440 }, { "epoch": 5.38, "learning_rate": 2.2433386048942127e-06, "loss": 0.2287, "step": 41450 }, { "epoch": 5.38, "learning_rate": 2.2423336074987827e-06, "loss": 0.2294, "step": 41460 }, { "epoch": 5.38, "learning_rate": 2.241328628785265e-06, "loss": 0.2277, "step": 41470 }, { "epoch": 5.38, "learning_rate": 2.2403236689387872e-06, "loss": 0.2302, "step": 41480 }, { "epoch": 5.38, "learning_rate": 2.239318728144473e-06, "loss": 0.2213, "step": 41490 }, { "epoch": 5.39, "learning_rate": 2.238313806587441e-06, "loss": 0.2366, "step": 41500 }, { "epoch": 5.39, "learning_rate": 2.23730890445281e-06, "loss": 0.2252, "step": 41510 }, { "epoch": 5.39, "learning_rate": 2.236304021925692e-06, "loss": 0.2185, "step": 41520 }, { "epoch": 5.39, "learning_rate": 2.235299159191196e-06, "loss": 0.2178, "step": 41530 }, { "epoch": 5.39, "learning_rate": 2.2342943164344284e-06, "loss": 0.2255, "step": 41540 }, { "epoch": 5.39, "learning_rate": 2.233289493840491e-06, "loss": 0.229, "step": 41550 }, { "epoch": 5.39, "learning_rate": 2.232284691594482e-06, "loss": 0.2243, "step": 41560 }, { "epoch": 5.4, "learning_rate": 2.231279909881496e-06, "loss": 0.2251, "step": 41570 }, { "epoch": 5.4, "learning_rate": 2.230275148886624e-06, "loss": 0.2271, "step": 41580 }, { "epoch": 5.4, "learning_rate": 2.2292704087949538e-06, "loss": 0.2439, "step": 41590 }, { "epoch": 5.4, "learning_rate": 2.228265689791567e-06, "loss": 0.2274, "step": 41600 }, { "epoch": 5.4, "learning_rate": 2.2272609920615435e-06, "loss": 0.2266, "step": 41610 }, { "epoch": 5.4, "learning_rate": 2.2262563157899592e-06, "loss": 0.237, "step": 41620 }, { "epoch": 5.4, "learning_rate": 2.225251661161885e-06, "loss": 0.2168, "step": 41630 }, { "epoch": 5.4, "learning_rate": 2.2242470283623884e-06, "loss": 0.2358, "step": 41640 }, { "epoch": 5.41, "learning_rate": 2.2232424175765327e-06, "loss": 0.2239, "step": 41650 }, { "epoch": 5.41, "learning_rate": 2.2222378289893778e-06, "loss": 0.228, "step": 41660 }, { "epoch": 5.41, "learning_rate": 2.221233262785978e-06, "loss": 0.2351, "step": 41670 }, { "epoch": 5.41, "learning_rate": 2.220228719151385e-06, "loss": 0.2371, "step": 41680 }, { "epoch": 5.41, "learning_rate": 2.2192241982706463e-06, "loss": 0.22, "step": 41690 }, { "epoch": 5.41, "learning_rate": 2.2182197003288035e-06, "loss": 0.2263, "step": 41700 }, { "epoch": 5.41, "learning_rate": 2.2172152255108957e-06, "loss": 0.2232, "step": 41710 }, { "epoch": 5.41, "learning_rate": 2.2162107740019575e-06, "loss": 0.2278, "step": 41720 }, { "epoch": 5.42, "learning_rate": 2.215206345987019e-06, "loss": 0.2375, "step": 41730 }, { "epoch": 5.42, "learning_rate": 2.214201941651105e-06, "loss": 0.2318, "step": 41740 }, { "epoch": 5.42, "learning_rate": 2.2131975611792375e-06, "loss": 0.2332, "step": 41750 }, { "epoch": 5.42, "learning_rate": 2.2121932047564335e-06, "loss": 0.2296, "step": 41760 }, { "epoch": 5.42, "learning_rate": 2.2111888725677047e-06, "loss": 0.2374, "step": 41770 }, { "epoch": 5.42, "learning_rate": 2.21018456479806e-06, "loss": 0.2267, "step": 41780 }, { "epoch": 5.42, "learning_rate": 2.2091802816325023e-06, "loss": 0.2198, "step": 41790 }, { "epoch": 5.43, "learning_rate": 2.208176023256031e-06, "loss": 0.2326, "step": 41800 }, { "epoch": 5.43, "learning_rate": 2.2071717898536405e-06, "loss": 0.2233, "step": 41810 }, { "epoch": 5.43, "learning_rate": 2.20616758161032e-06, "loss": 0.23, "step": 41820 }, { "epoch": 5.43, "learning_rate": 2.2051633987110563e-06, "loss": 0.2265, "step": 41830 }, { "epoch": 5.43, "learning_rate": 2.2041592413408277e-06, "loss": 0.2287, "step": 41840 }, { "epoch": 5.43, "learning_rate": 2.2031551096846115e-06, "loss": 0.2174, "step": 41850 }, { "epoch": 5.43, "learning_rate": 2.2021510039273784e-06, "loss": 0.2422, "step": 41860 }, { "epoch": 5.43, "learning_rate": 2.201146924254095e-06, "loss": 0.2407, "step": 41870 }, { "epoch": 5.44, "learning_rate": 2.200142870849722e-06, "loss": 0.2281, "step": 41880 }, { "epoch": 5.44, "learning_rate": 2.199138843899217e-06, "loss": 0.2224, "step": 41890 }, { "epoch": 5.44, "learning_rate": 2.1981348435875317e-06, "loss": 0.221, "step": 41900 }, { "epoch": 5.44, "learning_rate": 2.1971308700996124e-06, "loss": 0.2227, "step": 41910 }, { "epoch": 5.44, "learning_rate": 2.1961269236204008e-06, "loss": 0.2169, "step": 41920 }, { "epoch": 5.44, "learning_rate": 2.195123004334835e-06, "loss": 0.2428, "step": 41930 }, { "epoch": 5.44, "learning_rate": 2.1941191124278463e-06, "loss": 0.2319, "step": 41940 }, { "epoch": 5.44, "learning_rate": 2.1931152480843616e-06, "loss": 0.2186, "step": 41950 }, { "epoch": 5.45, "learning_rate": 2.1921114114893023e-06, "loss": 0.2339, "step": 41960 }, { "epoch": 5.45, "learning_rate": 2.1911076028275867e-06, "loss": 0.2322, "step": 41970 }, { "epoch": 5.45, "learning_rate": 2.190103822284124e-06, "loss": 0.2361, "step": 41980 }, { "epoch": 5.45, "learning_rate": 2.189100070043823e-06, "loss": 0.2126, "step": 41990 }, { "epoch": 5.45, "learning_rate": 2.188096346291583e-06, "loss": 0.2267, "step": 42000 }, { "epoch": 5.45, "learning_rate": 2.1870926512123015e-06, "loss": 0.2337, "step": 42010 }, { "epoch": 5.45, "learning_rate": 2.186088984990868e-06, "loss": 0.2394, "step": 42020 }, { "epoch": 5.45, "learning_rate": 2.185085347812168e-06, "loss": 0.2193, "step": 42030 }, { "epoch": 5.46, "learning_rate": 2.1840817398610826e-06, "loss": 0.2194, "step": 42040 }, { "epoch": 5.46, "learning_rate": 2.1830781613224847e-06, "loss": 0.2129, "step": 42050 }, { "epoch": 5.46, "learning_rate": 2.1820746123812445e-06, "loss": 0.2246, "step": 42060 }, { "epoch": 5.46, "learning_rate": 2.1810710932222257e-06, "loss": 0.2136, "step": 42070 }, { "epoch": 5.46, "learning_rate": 2.180067604030287e-06, "loss": 0.2248, "step": 42080 }, { "epoch": 5.46, "learning_rate": 2.1790641449902797e-06, "loss": 0.2108, "step": 42090 }, { "epoch": 5.46, "learning_rate": 2.178060716287052e-06, "loss": 0.2314, "step": 42100 }, { "epoch": 5.47, "learning_rate": 2.1770573181054457e-06, "loss": 0.2281, "step": 42110 }, { "epoch": 5.47, "learning_rate": 2.176053950630296e-06, "loss": 0.2209, "step": 42120 }, { "epoch": 5.47, "learning_rate": 2.175050614046434e-06, "loss": 0.2311, "step": 42130 }, { "epoch": 5.47, "learning_rate": 2.1740473085386832e-06, "loss": 0.2183, "step": 42140 }, { "epoch": 5.47, "learning_rate": 2.1730440342918643e-06, "loss": 0.2291, "step": 42150 }, { "epoch": 5.47, "learning_rate": 2.1720407914907886e-06, "loss": 0.2318, "step": 42160 }, { "epoch": 5.47, "learning_rate": 2.1710375803202644e-06, "loss": 0.2228, "step": 42170 }, { "epoch": 5.47, "learning_rate": 2.1700344009650935e-06, "loss": 0.2295, "step": 42180 }, { "epoch": 5.48, "learning_rate": 2.1690312536100705e-06, "loss": 0.2316, "step": 42190 }, { "epoch": 5.48, "learning_rate": 2.168028138439986e-06, "loss": 0.2351, "step": 42200 }, { "epoch": 5.48, "learning_rate": 2.167025055639623e-06, "loss": 0.2232, "step": 42210 }, { "epoch": 5.48, "learning_rate": 2.166022005393761e-06, "loss": 0.2423, "step": 42220 }, { "epoch": 5.48, "learning_rate": 2.16501898788717e-06, "loss": 0.2369, "step": 42230 }, { "epoch": 5.48, "learning_rate": 2.164016003304617e-06, "loss": 0.2317, "step": 42240 }, { "epoch": 5.48, "learning_rate": 2.1630130518308617e-06, "loss": 0.2377, "step": 42250 }, { "epoch": 5.48, "learning_rate": 2.162010133650657e-06, "loss": 0.2246, "step": 42260 }, { "epoch": 5.49, "learning_rate": 2.1610072489487508e-06, "loss": 0.2225, "step": 42270 }, { "epoch": 5.49, "learning_rate": 2.1600043979098856e-06, "loss": 0.225, "step": 42280 }, { "epoch": 5.49, "learning_rate": 2.159001580718795e-06, "loss": 0.23, "step": 42290 }, { "epoch": 5.49, "learning_rate": 2.1579987975602086e-06, "loss": 0.2164, "step": 42300 }, { "epoch": 5.49, "learning_rate": 2.1569960486188485e-06, "loss": 0.2326, "step": 42310 }, { "epoch": 5.49, "learning_rate": 2.1559933340794326e-06, "loss": 0.2295, "step": 42320 }, { "epoch": 5.49, "learning_rate": 2.1549906541266694e-06, "loss": 0.2292, "step": 42330 }, { "epoch": 5.5, "learning_rate": 2.1539880089452635e-06, "loss": 0.237, "step": 42340 }, { "epoch": 5.5, "learning_rate": 2.1529853987199117e-06, "loss": 0.2303, "step": 42350 }, { "epoch": 5.5, "learning_rate": 2.1519828236353047e-06, "loss": 0.215, "step": 42360 }, { "epoch": 5.5, "learning_rate": 2.150980283876127e-06, "loss": 0.22, "step": 42370 }, { "epoch": 5.5, "learning_rate": 2.1499777796270565e-06, "loss": 0.226, "step": 42380 }, { "epoch": 5.5, "learning_rate": 2.148975311072765e-06, "loss": 0.2346, "step": 42390 }, { "epoch": 5.5, "learning_rate": 2.147972878397916e-06, "loss": 0.2273, "step": 42400 }, { "epoch": 5.5, "learning_rate": 2.1469704817871684e-06, "loss": 0.2312, "step": 42410 }, { "epoch": 5.51, "learning_rate": 2.145968121425174e-06, "loss": 0.2263, "step": 42420 }, { "epoch": 5.51, "learning_rate": 2.1449657974965766e-06, "loss": 0.2483, "step": 42430 }, { "epoch": 5.51, "learning_rate": 2.143963510186015e-06, "loss": 0.217, "step": 42440 }, { "epoch": 5.51, "learning_rate": 2.1429612596781204e-06, "loss": 0.2247, "step": 42450 }, { "epoch": 5.51, "learning_rate": 2.1419590461575175e-06, "loss": 0.2349, "step": 42460 }, { "epoch": 5.51, "learning_rate": 2.1409568698088238e-06, "loss": 0.2327, "step": 42470 }, { "epoch": 5.51, "learning_rate": 2.13995473081665e-06, "loss": 0.2252, "step": 42480 }, { "epoch": 5.51, "learning_rate": 2.138952629365601e-06, "loss": 0.2396, "step": 42490 }, { "epoch": 5.52, "learning_rate": 2.137950565640273e-06, "loss": 0.2312, "step": 42500 }, { "epoch": 5.52, "learning_rate": 2.136948539825256e-06, "loss": 0.2218, "step": 42510 }, { "epoch": 5.52, "learning_rate": 2.1359465521051338e-06, "loss": 0.2314, "step": 42520 }, { "epoch": 5.52, "learning_rate": 2.1349446026644824e-06, "loss": 0.226, "step": 42530 }, { "epoch": 5.52, "learning_rate": 2.1339426916878708e-06, "loss": 0.2288, "step": 42540 }, { "epoch": 5.52, "learning_rate": 2.132940819359861e-06, "loss": 0.2284, "step": 42550 }, { "epoch": 5.52, "learning_rate": 2.1319389858650084e-06, "loss": 0.2233, "step": 42560 }, { "epoch": 5.52, "learning_rate": 2.13093719138786e-06, "loss": 0.2296, "step": 42570 }, { "epoch": 5.53, "learning_rate": 2.1299354361129564e-06, "loss": 0.2158, "step": 42580 }, { "epoch": 5.53, "learning_rate": 2.128933720224831e-06, "loss": 0.2189, "step": 42590 }, { "epoch": 5.53, "learning_rate": 2.1279320439080103e-06, "loss": 0.2341, "step": 42600 }, { "epoch": 5.53, "learning_rate": 2.1269304073470127e-06, "loss": 0.2322, "step": 42610 }, { "epoch": 5.53, "learning_rate": 2.1259288107263493e-06, "loss": 0.2419, "step": 42620 }, { "epoch": 5.53, "learning_rate": 2.1249272542305257e-06, "loss": 0.2207, "step": 42630 }, { "epoch": 5.53, "learning_rate": 2.123925738044037e-06, "loss": 0.2319, "step": 42640 }, { "epoch": 5.54, "learning_rate": 2.1229242623513727e-06, "loss": 0.2242, "step": 42650 }, { "epoch": 5.54, "learning_rate": 2.1219228273370152e-06, "loss": 0.224, "step": 42660 }, { "epoch": 5.54, "learning_rate": 2.1209214331854393e-06, "loss": 0.2248, "step": 42670 }, { "epoch": 5.54, "learning_rate": 2.11992008008111e-06, "loss": 0.2248, "step": 42680 }, { "epoch": 5.54, "learning_rate": 2.1189187682084883e-06, "loss": 0.2312, "step": 42690 }, { "epoch": 5.54, "learning_rate": 2.1179174977520254e-06, "loss": 0.2192, "step": 42700 }, { "epoch": 5.54, "learning_rate": 2.1169162688961644e-06, "loss": 0.2443, "step": 42710 }, { "epoch": 5.54, "learning_rate": 2.115915081825343e-06, "loss": 0.2355, "step": 42720 }, { "epoch": 5.55, "learning_rate": 2.1149139367239887e-06, "loss": 0.2266, "step": 42730 }, { "epoch": 5.55, "learning_rate": 2.1139128337765235e-06, "loss": 0.2391, "step": 42740 }, { "epoch": 5.55, "learning_rate": 2.1129117731673597e-06, "loss": 0.2213, "step": 42750 }, { "epoch": 5.55, "learning_rate": 2.111910755080903e-06, "loss": 0.2315, "step": 42760 }, { "epoch": 5.55, "learning_rate": 2.1109097797015512e-06, "loss": 0.2232, "step": 42770 }, { "epoch": 5.55, "learning_rate": 2.1099088472136936e-06, "loss": 0.2242, "step": 42780 }, { "epoch": 5.55, "learning_rate": 2.108907957801712e-06, "loss": 0.2099, "step": 42790 }, { "epoch": 5.55, "learning_rate": 2.10790711164998e-06, "loss": 0.2196, "step": 42800 }, { "epoch": 5.56, "learning_rate": 2.1069063089428643e-06, "loss": 0.2335, "step": 42810 }, { "epoch": 5.56, "learning_rate": 2.1059055498647217e-06, "loss": 0.2275, "step": 42820 }, { "epoch": 5.56, "learning_rate": 2.1049048345999025e-06, "loss": 0.2295, "step": 42830 }, { "epoch": 5.56, "learning_rate": 2.1039041633327487e-06, "loss": 0.2242, "step": 42840 }, { "epoch": 5.56, "learning_rate": 2.1029035362475936e-06, "loss": 0.2437, "step": 42850 }, { "epoch": 5.56, "learning_rate": 2.1019029535287626e-06, "loss": 0.2214, "step": 42860 }, { "epoch": 5.56, "learning_rate": 2.100902415360573e-06, "loss": 0.2165, "step": 42870 }, { "epoch": 5.57, "learning_rate": 2.099901921927335e-06, "loss": 0.2179, "step": 42880 }, { "epoch": 5.57, "learning_rate": 2.0989014734133484e-06, "loss": 0.231, "step": 42890 }, { "epoch": 5.57, "learning_rate": 2.0979010700029055e-06, "loss": 0.2324, "step": 42900 }, { "epoch": 5.57, "learning_rate": 2.096900711880292e-06, "loss": 0.2183, "step": 42910 }, { "epoch": 5.57, "learning_rate": 2.0959003992297822e-06, "loss": 0.2218, "step": 42920 }, { "epoch": 5.57, "learning_rate": 2.0949001322356447e-06, "loss": 0.2223, "step": 42930 }, { "epoch": 5.57, "learning_rate": 2.0938999110821382e-06, "loss": 0.2278, "step": 42940 }, { "epoch": 5.57, "learning_rate": 2.0928997359535146e-06, "loss": 0.2221, "step": 42950 }, { "epoch": 5.58, "learning_rate": 2.0918996070340145e-06, "loss": 0.2137, "step": 42960 }, { "epoch": 5.58, "learning_rate": 2.090899524507872e-06, "loss": 0.2152, "step": 42970 }, { "epoch": 5.58, "learning_rate": 2.089899488559314e-06, "loss": 0.2351, "step": 42980 }, { "epoch": 5.58, "learning_rate": 2.0888994993725547e-06, "loss": 0.2331, "step": 42990 }, { "epoch": 5.58, "learning_rate": 2.0878995571318033e-06, "loss": 0.2199, "step": 43000 }, { "epoch": 5.58, "learning_rate": 2.086899662021259e-06, "loss": 0.229, "step": 43010 }, { "epoch": 5.58, "learning_rate": 2.0858998142251127e-06, "loss": 0.2314, "step": 43020 }, { "epoch": 5.58, "learning_rate": 2.0849000139275465e-06, "loss": 0.2302, "step": 43030 }, { "epoch": 5.59, "learning_rate": 2.0839002613127323e-06, "loss": 0.2243, "step": 43040 }, { "epoch": 5.59, "learning_rate": 2.0829005565648366e-06, "loss": 0.247, "step": 43050 }, { "epoch": 5.59, "learning_rate": 2.081900899868013e-06, "loss": 0.2246, "step": 43060 }, { "epoch": 5.59, "learning_rate": 2.080901291406409e-06, "loss": 0.2189, "step": 43070 }, { "epoch": 5.59, "learning_rate": 2.0799017313641626e-06, "loss": 0.2315, "step": 43080 }, { "epoch": 5.59, "learning_rate": 2.0789022199254036e-06, "loss": 0.2351, "step": 43090 }, { "epoch": 5.59, "learning_rate": 2.0779027572742504e-06, "loss": 0.2332, "step": 43100 }, { "epoch": 5.6, "learning_rate": 2.0769033435948144e-06, "loss": 0.2366, "step": 43110 }, { "epoch": 5.6, "learning_rate": 2.0759039790711987e-06, "loss": 0.2141, "step": 43120 }, { "epoch": 5.6, "learning_rate": 2.074904663887495e-06, "loss": 0.2094, "step": 43130 }, { "epoch": 5.6, "learning_rate": 2.073905398227787e-06, "loss": 0.214, "step": 43140 }, { "epoch": 5.6, "learning_rate": 2.072906182276151e-06, "loss": 0.2289, "step": 43150 }, { "epoch": 5.6, "learning_rate": 2.071907016216651e-06, "loss": 0.2306, "step": 43160 }, { "epoch": 5.6, "learning_rate": 2.070907900233344e-06, "loss": 0.2278, "step": 43170 }, { "epoch": 5.6, "learning_rate": 2.0699088345102767e-06, "loss": 0.2224, "step": 43180 }, { "epoch": 5.61, "learning_rate": 2.0689098192314877e-06, "loss": 0.2291, "step": 43190 }, { "epoch": 5.61, "learning_rate": 2.0679108545810052e-06, "loss": 0.2259, "step": 43200 }, { "epoch": 5.61, "learning_rate": 2.066911940742848e-06, "loss": 0.2293, "step": 43210 }, { "epoch": 5.61, "learning_rate": 2.0659130779010274e-06, "loss": 0.2339, "step": 43220 }, { "epoch": 5.61, "learning_rate": 2.0649142662395424e-06, "loss": 0.2325, "step": 43230 }, { "epoch": 5.61, "learning_rate": 2.063915505942385e-06, "loss": 0.2329, "step": 43240 }, { "epoch": 5.61, "learning_rate": 2.0629167971935356e-06, "loss": 0.2457, "step": 43250 }, { "epoch": 5.61, "learning_rate": 2.0619181401769683e-06, "loss": 0.2353, "step": 43260 }, { "epoch": 5.62, "learning_rate": 2.060919535076644e-06, "loss": 0.2187, "step": 43270 }, { "epoch": 5.62, "learning_rate": 2.0599209820765164e-06, "loss": 0.2432, "step": 43280 }, { "epoch": 5.62, "learning_rate": 2.05892248136053e-06, "loss": 0.1966, "step": 43290 }, { "epoch": 5.62, "learning_rate": 2.057924033112616e-06, "loss": 0.2352, "step": 43300 }, { "epoch": 5.62, "learning_rate": 2.056925637516701e-06, "loss": 0.2214, "step": 43310 }, { "epoch": 5.62, "learning_rate": 2.055927294756698e-06, "loss": 0.2294, "step": 43320 }, { "epoch": 5.62, "learning_rate": 2.0549290050165126e-06, "loss": 0.2349, "step": 43330 }, { "epoch": 5.62, "learning_rate": 2.0539307684800387e-06, "loss": 0.2484, "step": 43340 }, { "epoch": 5.63, "learning_rate": 2.0529325853311626e-06, "loss": 0.2229, "step": 43350 }, { "epoch": 5.63, "learning_rate": 2.051934455753759e-06, "loss": 0.236, "step": 43360 }, { "epoch": 5.63, "learning_rate": 2.0509363799316935e-06, "loss": 0.228, "step": 43370 }, { "epoch": 5.63, "learning_rate": 2.049938358048821e-06, "loss": 0.2303, "step": 43380 }, { "epoch": 5.63, "learning_rate": 2.048940390288988e-06, "loss": 0.2221, "step": 43390 }, { "epoch": 5.63, "learning_rate": 2.0479424768360307e-06, "loss": 0.2193, "step": 43400 }, { "epoch": 5.63, "learning_rate": 2.0469446178737727e-06, "loss": 0.2384, "step": 43410 }, { "epoch": 5.64, "learning_rate": 2.045946813586031e-06, "loss": 0.2313, "step": 43420 }, { "epoch": 5.64, "learning_rate": 2.044949064156611e-06, "loss": 0.2261, "step": 43430 }, { "epoch": 5.64, "learning_rate": 2.0439513697693074e-06, "loss": 0.2204, "step": 43440 }, { "epoch": 5.64, "learning_rate": 2.0429537306079065e-06, "loss": 0.2248, "step": 43450 }, { "epoch": 5.64, "learning_rate": 2.0419561468561826e-06, "loss": 0.2247, "step": 43460 }, { "epoch": 5.64, "learning_rate": 2.0409586186979014e-06, "loss": 0.222, "step": 43470 }, { "epoch": 5.64, "learning_rate": 2.0399611463168164e-06, "loss": 0.2238, "step": 43480 }, { "epoch": 5.64, "learning_rate": 2.038963729896673e-06, "loss": 0.2262, "step": 43490 }, { "epoch": 5.65, "learning_rate": 2.037966369621205e-06, "loss": 0.2409, "step": 43500 }, { "epoch": 5.65, "learning_rate": 2.036969065674136e-06, "loss": 0.2132, "step": 43510 }, { "epoch": 5.65, "learning_rate": 2.0359718182391785e-06, "loss": 0.2264, "step": 43520 }, { "epoch": 5.65, "learning_rate": 2.034974627500037e-06, "loss": 0.2421, "step": 43530 }, { "epoch": 5.65, "learning_rate": 2.033977493640404e-06, "loss": 0.2403, "step": 43540 }, { "epoch": 5.65, "learning_rate": 2.0329804168439594e-06, "loss": 0.2213, "step": 43550 }, { "epoch": 5.65, "learning_rate": 2.031983397294377e-06, "loss": 0.2176, "step": 43560 }, { "epoch": 5.65, "learning_rate": 2.030986435175317e-06, "loss": 0.2359, "step": 43570 }, { "epoch": 5.66, "learning_rate": 2.029989530670429e-06, "loss": 0.2353, "step": 43580 }, { "epoch": 5.66, "learning_rate": 2.0289926839633532e-06, "loss": 0.2321, "step": 43590 }, { "epoch": 5.66, "learning_rate": 2.0279958952377197e-06, "loss": 0.2097, "step": 43600 }, { "epoch": 5.66, "learning_rate": 2.0269991646771463e-06, "loss": 0.2273, "step": 43610 }, { "epoch": 5.66, "learning_rate": 2.02600249246524e-06, "loss": 0.2326, "step": 43620 }, { "epoch": 5.66, "learning_rate": 2.0250058787855985e-06, "loss": 0.2393, "step": 43630 }, { "epoch": 5.66, "learning_rate": 2.0240093238218084e-06, "loss": 0.2215, "step": 43640 }, { "epoch": 5.67, "learning_rate": 2.023012827757444e-06, "loss": 0.2315, "step": 43650 }, { "epoch": 5.67, "learning_rate": 2.0220163907760705e-06, "loss": 0.2189, "step": 43660 }, { "epoch": 5.67, "learning_rate": 2.0210200130612416e-06, "loss": 0.2376, "step": 43670 }, { "epoch": 5.67, "learning_rate": 2.0200236947964997e-06, "loss": 0.2219, "step": 43680 }, { "epoch": 5.67, "learning_rate": 2.0190274361653765e-06, "loss": 0.232, "step": 43690 }, { "epoch": 5.67, "learning_rate": 2.0180312373513934e-06, "loss": 0.2192, "step": 43700 }, { "epoch": 5.67, "learning_rate": 2.01703509853806e-06, "loss": 0.2266, "step": 43710 }, { "epoch": 5.67, "learning_rate": 2.0160390199088744e-06, "loss": 0.2029, "step": 43720 }, { "epoch": 5.68, "learning_rate": 2.0150430016473244e-06, "loss": 0.2367, "step": 43730 }, { "epoch": 5.68, "learning_rate": 2.014047043936887e-06, "loss": 0.2079, "step": 43740 }, { "epoch": 5.68, "learning_rate": 2.0130511469610278e-06, "loss": 0.2295, "step": 43750 }, { "epoch": 5.68, "learning_rate": 2.0120553109031998e-06, "loss": 0.229, "step": 43760 }, { "epoch": 5.68, "learning_rate": 2.011059535946847e-06, "loss": 0.2257, "step": 43770 }, { "epoch": 5.68, "learning_rate": 2.010063822275401e-06, "loss": 0.2335, "step": 43780 }, { "epoch": 5.68, "learning_rate": 2.0090681700722815e-06, "loss": 0.2319, "step": 43790 }, { "epoch": 5.68, "learning_rate": 2.008072579520898e-06, "loss": 0.2284, "step": 43800 }, { "epoch": 5.69, "learning_rate": 2.0070770508046485e-06, "loss": 0.2289, "step": 43810 }, { "epoch": 5.69, "learning_rate": 2.006081584106919e-06, "loss": 0.2234, "step": 43820 }, { "epoch": 5.69, "learning_rate": 2.0050861796110843e-06, "loss": 0.2258, "step": 43830 }, { "epoch": 5.69, "learning_rate": 2.004090837500508e-06, "loss": 0.2305, "step": 43840 }, { "epoch": 5.69, "learning_rate": 2.0030955579585425e-06, "loss": 0.2168, "step": 43850 }, { "epoch": 5.69, "learning_rate": 2.0021003411685275e-06, "loss": 0.2316, "step": 43860 }, { "epoch": 5.69, "learning_rate": 2.001105187313792e-06, "loss": 0.2117, "step": 43870 }, { "epoch": 5.7, "learning_rate": 2.0001100965776535e-06, "loss": 0.2097, "step": 43880 }, { "epoch": 5.7, "learning_rate": 1.9991150691434183e-06, "loss": 0.2362, "step": 43890 }, { "epoch": 5.7, "learning_rate": 1.998120105194379e-06, "loss": 0.2393, "step": 43900 }, { "epoch": 5.7, "learning_rate": 1.997125204913819e-06, "loss": 0.2177, "step": 43910 }, { "epoch": 5.7, "learning_rate": 1.9961303684850084e-06, "loss": 0.2171, "step": 43920 }, { "epoch": 5.7, "learning_rate": 1.995135596091206e-06, "loss": 0.2285, "step": 43930 }, { "epoch": 5.7, "learning_rate": 1.994140887915659e-06, "loss": 0.2126, "step": 43940 }, { "epoch": 5.7, "learning_rate": 1.993146244141602e-06, "loss": 0.2534, "step": 43950 }, { "epoch": 5.71, "learning_rate": 1.99215166495226e-06, "loss": 0.2165, "step": 43960 }, { "epoch": 5.71, "learning_rate": 1.991157150530842e-06, "loss": 0.2129, "step": 43970 }, { "epoch": 5.71, "learning_rate": 1.9901627010605494e-06, "loss": 0.2314, "step": 43980 }, { "epoch": 5.71, "learning_rate": 1.989168316724569e-06, "loss": 0.2439, "step": 43990 }, { "epoch": 5.71, "learning_rate": 1.9881739977060764e-06, "loss": 0.2343, "step": 44000 }, { "epoch": 5.71, "learning_rate": 1.9871797441882344e-06, "loss": 0.2322, "step": 44010 }, { "epoch": 5.71, "learning_rate": 1.9861855563541956e-06, "loss": 0.2282, "step": 44020 }, { "epoch": 5.71, "learning_rate": 1.985191434387099e-06, "loss": 0.24, "step": 44030 }, { "epoch": 5.72, "learning_rate": 1.984197378470071e-06, "loss": 0.2184, "step": 44040 }, { "epoch": 5.72, "learning_rate": 1.9832033887862278e-06, "loss": 0.2292, "step": 44050 }, { "epoch": 5.72, "learning_rate": 1.982209465518672e-06, "loss": 0.2375, "step": 44060 }, { "epoch": 5.72, "learning_rate": 1.981215608850493e-06, "loss": 0.2266, "step": 44070 }, { "epoch": 5.72, "learning_rate": 1.9802218189647704e-06, "loss": 0.2429, "step": 44080 }, { "epoch": 5.72, "learning_rate": 1.9792280960445695e-06, "loss": 0.2213, "step": 44090 }, { "epoch": 5.72, "learning_rate": 1.978234440272945e-06, "loss": 0.2234, "step": 44100 }, { "epoch": 5.72, "learning_rate": 1.977240851832937e-06, "loss": 0.2243, "step": 44110 }, { "epoch": 5.73, "learning_rate": 1.976247330907575e-06, "loss": 0.2379, "step": 44120 }, { "epoch": 5.73, "learning_rate": 1.975253877679876e-06, "loss": 0.2273, "step": 44130 }, { "epoch": 5.73, "learning_rate": 1.974260492332843e-06, "loss": 0.2217, "step": 44140 }, { "epoch": 5.73, "learning_rate": 1.9732671750494674e-06, "loss": 0.2206, "step": 44150 }, { "epoch": 5.73, "learning_rate": 1.9722739260127294e-06, "loss": 0.2214, "step": 44160 }, { "epoch": 5.73, "learning_rate": 1.9712807454055945e-06, "loss": 0.2189, "step": 44170 }, { "epoch": 5.73, "learning_rate": 1.9702876334110163e-06, "loss": 0.2414, "step": 44180 }, { "epoch": 5.74, "learning_rate": 1.9692945902119365e-06, "loss": 0.2241, "step": 44190 }, { "epoch": 5.74, "learning_rate": 1.968301615991284e-06, "loss": 0.2344, "step": 44200 }, { "epoch": 5.74, "learning_rate": 1.9673087109319732e-06, "loss": 0.2149, "step": 44210 }, { "epoch": 5.74, "learning_rate": 1.9663158752169084e-06, "loss": 0.2354, "step": 44220 }, { "epoch": 5.74, "learning_rate": 1.965323109028979e-06, "loss": 0.2361, "step": 44230 }, { "epoch": 5.74, "learning_rate": 1.964330412551063e-06, "loss": 0.2307, "step": 44240 }, { "epoch": 5.74, "learning_rate": 1.9633377859660246e-06, "loss": 0.2332, "step": 44250 }, { "epoch": 5.74, "learning_rate": 1.962345229456716e-06, "loss": 0.2163, "step": 44260 }, { "epoch": 5.75, "learning_rate": 1.9613527432059757e-06, "loss": 0.2314, "step": 44270 }, { "epoch": 5.75, "learning_rate": 1.9603603273966296e-06, "loss": 0.2237, "step": 44280 }, { "epoch": 5.75, "learning_rate": 1.9593679822114905e-06, "loss": 0.2297, "step": 44290 }, { "epoch": 5.75, "learning_rate": 1.9583757078333587e-06, "loss": 0.216, "step": 44300 }, { "epoch": 5.75, "learning_rate": 1.957383504445021e-06, "loss": 0.2235, "step": 44310 }, { "epoch": 5.75, "learning_rate": 1.95639137222925e-06, "loss": 0.2324, "step": 44320 }, { "epoch": 5.75, "learning_rate": 1.955399311368808e-06, "loss": 0.2293, "step": 44330 }, { "epoch": 5.75, "learning_rate": 1.9544073220464424e-06, "loss": 0.2326, "step": 44340 }, { "epoch": 5.76, "learning_rate": 1.953415404444886e-06, "loss": 0.2211, "step": 44350 }, { "epoch": 5.76, "learning_rate": 1.9524235587468614e-06, "loss": 0.2317, "step": 44360 }, { "epoch": 5.76, "learning_rate": 1.9514317851350764e-06, "loss": 0.228, "step": 44370 }, { "epoch": 5.76, "learning_rate": 1.9504400837922246e-06, "loss": 0.2365, "step": 44380 }, { "epoch": 5.76, "learning_rate": 1.949448454900988e-06, "loss": 0.2298, "step": 44390 }, { "epoch": 5.76, "learning_rate": 1.948456898644034e-06, "loss": 0.2186, "step": 44400 }, { "epoch": 5.76, "learning_rate": 1.947465415204019e-06, "loss": 0.2326, "step": 44410 }, { "epoch": 5.77, "learning_rate": 1.9464740047635816e-06, "loss": 0.2297, "step": 44420 }, { "epoch": 5.77, "learning_rate": 1.945482667505351e-06, "loss": 0.2159, "step": 44430 }, { "epoch": 5.77, "learning_rate": 1.9444914036119415e-06, "loss": 0.2125, "step": 44440 }, { "epoch": 5.77, "learning_rate": 1.943500213265953e-06, "loss": 0.2253, "step": 44450 }, { "epoch": 5.77, "learning_rate": 1.942509096649973e-06, "loss": 0.2315, "step": 44460 }, { "epoch": 5.77, "learning_rate": 1.941518053946575e-06, "loss": 0.2261, "step": 44470 }, { "epoch": 5.77, "learning_rate": 1.94052708533832e-06, "loss": 0.2311, "step": 44480 }, { "epoch": 5.77, "learning_rate": 1.9395361910077524e-06, "loss": 0.2324, "step": 44490 }, { "epoch": 5.78, "learning_rate": 1.938545371137406e-06, "loss": 0.2304, "step": 44500 }, { "epoch": 5.78, "learning_rate": 1.9375546259098006e-06, "loss": 0.2312, "step": 44510 }, { "epoch": 5.78, "learning_rate": 1.9365639555074397e-06, "loss": 0.2081, "step": 44520 }, { "epoch": 5.78, "learning_rate": 1.9355733601128147e-06, "loss": 0.2265, "step": 44530 }, { "epoch": 5.78, "learning_rate": 1.9345828399084045e-06, "loss": 0.2285, "step": 44540 }, { "epoch": 5.78, "learning_rate": 1.933592395076672e-06, "loss": 0.238, "step": 44550 }, { "epoch": 5.78, "learning_rate": 1.9326020258000675e-06, "loss": 0.2091, "step": 44560 }, { "epoch": 5.78, "learning_rate": 1.931611732261026e-06, "loss": 0.2108, "step": 44570 }, { "epoch": 5.79, "learning_rate": 1.9306215146419706e-06, "loss": 0.2341, "step": 44580 }, { "epoch": 5.79, "learning_rate": 1.9296313731253083e-06, "loss": 0.2401, "step": 44590 }, { "epoch": 5.79, "learning_rate": 1.928641307893433e-06, "loss": 0.2252, "step": 44600 }, { "epoch": 5.79, "learning_rate": 1.9276513191287256e-06, "loss": 0.2313, "step": 44610 }, { "epoch": 5.79, "learning_rate": 1.926661407013552e-06, "loss": 0.2253, "step": 44620 }, { "epoch": 5.79, "learning_rate": 1.9256715717302626e-06, "loss": 0.2216, "step": 44630 }, { "epoch": 5.79, "learning_rate": 1.9246818134611954e-06, "loss": 0.2284, "step": 44640 }, { "epoch": 5.79, "learning_rate": 1.9236921323886748e-06, "loss": 0.212, "step": 44650 }, { "epoch": 5.8, "learning_rate": 1.9227025286950084e-06, "loss": 0.2424, "step": 44660 }, { "epoch": 5.8, "learning_rate": 1.921713002562492e-06, "loss": 0.2236, "step": 44670 }, { "epoch": 5.8, "learning_rate": 1.9207235541734056e-06, "loss": 0.2236, "step": 44680 }, { "epoch": 5.8, "learning_rate": 1.9197341837100165e-06, "loss": 0.2271, "step": 44690 }, { "epoch": 5.8, "learning_rate": 1.918744891354576e-06, "loss": 0.211, "step": 44700 }, { "epoch": 5.8, "learning_rate": 1.917755677289321e-06, "loss": 0.234, "step": 44710 }, { "epoch": 5.8, "learning_rate": 1.9167665416964763e-06, "loss": 0.2227, "step": 44720 }, { "epoch": 5.81, "learning_rate": 1.9157774847582486e-06, "loss": 0.241, "step": 44730 }, { "epoch": 5.81, "learning_rate": 1.9147885066568333e-06, "loss": 0.2152, "step": 44740 }, { "epoch": 5.81, "learning_rate": 1.9137996075744096e-06, "loss": 0.2219, "step": 44750 }, { "epoch": 5.81, "learning_rate": 1.912810787693143e-06, "loss": 0.223, "step": 44760 }, { "epoch": 5.81, "learning_rate": 1.9118220471951836e-06, "loss": 0.226, "step": 44770 }, { "epoch": 5.81, "learning_rate": 1.9108333862626672e-06, "loss": 0.2243, "step": 44780 }, { "epoch": 5.81, "learning_rate": 1.9098448050777154e-06, "loss": 0.2241, "step": 44790 }, { "epoch": 5.81, "learning_rate": 1.9088563038224343e-06, "loss": 0.2198, "step": 44800 }, { "epoch": 5.82, "learning_rate": 1.907867882678916e-06, "loss": 0.2227, "step": 44810 }, { "epoch": 5.82, "learning_rate": 1.9068795418292373e-06, "loss": 0.2309, "step": 44820 }, { "epoch": 5.82, "learning_rate": 1.9058912814554611e-06, "loss": 0.2106, "step": 44830 }, { "epoch": 5.82, "learning_rate": 1.9049031017396341e-06, "loss": 0.2107, "step": 44840 }, { "epoch": 5.82, "learning_rate": 1.9039150028637893e-06, "loss": 0.2322, "step": 44850 }, { "epoch": 5.82, "learning_rate": 1.9029269850099446e-06, "loss": 0.2316, "step": 44860 }, { "epoch": 5.82, "learning_rate": 1.9019390483601016e-06, "loss": 0.2119, "step": 44870 }, { "epoch": 5.82, "learning_rate": 1.900951193096249e-06, "loss": 0.2378, "step": 44880 }, { "epoch": 5.83, "learning_rate": 1.8999634194003599e-06, "loss": 0.2294, "step": 44890 }, { "epoch": 5.83, "learning_rate": 1.8989757274543918e-06, "loss": 0.23, "step": 44900 }, { "epoch": 5.83, "learning_rate": 1.8979881174402865e-06, "loss": 0.233, "step": 44910 }, { "epoch": 5.83, "learning_rate": 1.8970005895399728e-06, "loss": 0.2302, "step": 44920 }, { "epoch": 5.83, "learning_rate": 1.896013143935363e-06, "loss": 0.2286, "step": 44930 }, { "epoch": 5.83, "learning_rate": 1.8950257808083544e-06, "loss": 0.2222, "step": 44940 }, { "epoch": 5.83, "learning_rate": 1.8940385003408284e-06, "loss": 0.2271, "step": 44950 }, { "epoch": 5.84, "learning_rate": 1.893051302714653e-06, "loss": 0.2251, "step": 44960 }, { "epoch": 5.84, "learning_rate": 1.8920641881116798e-06, "loss": 0.2463, "step": 44970 }, { "epoch": 5.84, "learning_rate": 1.891077156713744e-06, "loss": 0.2316, "step": 44980 }, { "epoch": 5.84, "learning_rate": 1.8900902087026676e-06, "loss": 0.2232, "step": 44990 }, { "epoch": 5.84, "learning_rate": 1.8891033442602567e-06, "loss": 0.2099, "step": 45000 }, { "epoch": 5.84, "learning_rate": 1.8881165635683003e-06, "loss": 0.2288, "step": 45010 }, { "epoch": 5.84, "learning_rate": 1.887129866808574e-06, "loss": 0.2452, "step": 45020 }, { "epoch": 5.84, "learning_rate": 1.8861432541628372e-06, "loss": 0.2398, "step": 45030 }, { "epoch": 5.85, "learning_rate": 1.8851567258128342e-06, "loss": 0.2238, "step": 45040 }, { "epoch": 5.85, "learning_rate": 1.8841702819402925e-06, "loss": 0.2219, "step": 45050 }, { "epoch": 5.85, "learning_rate": 1.8831839227269253e-06, "loss": 0.2158, "step": 45060 }, { "epoch": 5.85, "learning_rate": 1.8821976483544305e-06, "loss": 0.2228, "step": 45070 }, { "epoch": 5.85, "learning_rate": 1.881211459004488e-06, "loss": 0.2262, "step": 45080 }, { "epoch": 5.85, "learning_rate": 1.8802253548587655e-06, "loss": 0.2357, "step": 45090 }, { "epoch": 5.85, "learning_rate": 1.8792393360989126e-06, "loss": 0.2232, "step": 45100 }, { "epoch": 5.85, "learning_rate": 1.8782534029065633e-06, "loss": 0.2456, "step": 45110 }, { "epoch": 5.86, "learning_rate": 1.8772675554633367e-06, "loss": 0.2141, "step": 45120 }, { "epoch": 5.86, "learning_rate": 1.8762817939508358e-06, "loss": 0.2301, "step": 45130 }, { "epoch": 5.86, "learning_rate": 1.8752961185506482e-06, "loss": 0.2344, "step": 45140 }, { "epoch": 5.86, "learning_rate": 1.8743105294443446e-06, "loss": 0.2208, "step": 45150 }, { "epoch": 5.86, "learning_rate": 1.87332502681348e-06, "loss": 0.2271, "step": 45160 }, { "epoch": 5.86, "learning_rate": 1.8723396108395947e-06, "loss": 0.2206, "step": 45170 }, { "epoch": 5.86, "learning_rate": 1.8713542817042116e-06, "loss": 0.2273, "step": 45180 }, { "epoch": 5.87, "learning_rate": 1.8703690395888378e-06, "loss": 0.221, "step": 45190 }, { "epoch": 5.87, "learning_rate": 1.8693838846749657e-06, "loss": 0.2202, "step": 45200 }, { "epoch": 5.87, "learning_rate": 1.8683988171440706e-06, "loss": 0.2189, "step": 45210 }, { "epoch": 5.87, "learning_rate": 1.8674138371776105e-06, "loss": 0.2362, "step": 45220 }, { "epoch": 5.87, "learning_rate": 1.86642894495703e-06, "loss": 0.2289, "step": 45230 }, { "epoch": 5.87, "learning_rate": 1.8654441406637555e-06, "loss": 0.23, "step": 45240 }, { "epoch": 5.87, "learning_rate": 1.8644594244791978e-06, "loss": 0.2413, "step": 45250 }, { "epoch": 5.87, "learning_rate": 1.863474796584751e-06, "loss": 0.2297, "step": 45260 }, { "epoch": 5.88, "learning_rate": 1.8624902571617942e-06, "loss": 0.2206, "step": 45270 }, { "epoch": 5.88, "learning_rate": 1.8615058063916894e-06, "loss": 0.2284, "step": 45280 }, { "epoch": 5.88, "learning_rate": 1.8605214444557815e-06, "loss": 0.2156, "step": 45290 }, { "epoch": 5.88, "learning_rate": 1.8595371715354002e-06, "loss": 0.223, "step": 45300 }, { "epoch": 5.88, "learning_rate": 1.858552987811859e-06, "loss": 0.2275, "step": 45310 }, { "epoch": 5.88, "learning_rate": 1.8575688934664531e-06, "loss": 0.2107, "step": 45320 }, { "epoch": 5.88, "learning_rate": 1.8565848886804635e-06, "loss": 0.239, "step": 45330 }, { "epoch": 5.88, "learning_rate": 1.8556009736351528e-06, "loss": 0.2241, "step": 45340 }, { "epoch": 5.89, "learning_rate": 1.8546171485117698e-06, "loss": 0.2326, "step": 45350 }, { "epoch": 5.89, "learning_rate": 1.8536334134915429e-06, "loss": 0.245, "step": 45360 }, { "epoch": 5.89, "learning_rate": 1.8526497687556868e-06, "loss": 0.2377, "step": 45370 }, { "epoch": 5.89, "learning_rate": 1.8516662144853988e-06, "loss": 0.2175, "step": 45380 }, { "epoch": 5.89, "learning_rate": 1.850682750861859e-06, "loss": 0.2423, "step": 45390 }, { "epoch": 5.89, "learning_rate": 1.8496993780662315e-06, "loss": 0.2223, "step": 45400 }, { "epoch": 5.89, "learning_rate": 1.8487160962796633e-06, "loss": 0.2345, "step": 45410 }, { "epoch": 5.89, "learning_rate": 1.8477329056832851e-06, "loss": 0.2392, "step": 45420 }, { "epoch": 5.9, "learning_rate": 1.8467498064582098e-06, "loss": 0.2322, "step": 45430 }, { "epoch": 5.9, "learning_rate": 1.8457667987855344e-06, "loss": 0.222, "step": 45440 }, { "epoch": 5.9, "learning_rate": 1.844783882846339e-06, "loss": 0.2299, "step": 45450 }, { "epoch": 5.9, "learning_rate": 1.843801058821686e-06, "loss": 0.2186, "step": 45460 }, { "epoch": 5.9, "learning_rate": 1.8428183268926215e-06, "loss": 0.2224, "step": 45470 }, { "epoch": 5.9, "learning_rate": 1.8418356872401748e-06, "loss": 0.2344, "step": 45480 }, { "epoch": 5.9, "learning_rate": 1.8408531400453585e-06, "loss": 0.2193, "step": 45490 }, { "epoch": 5.91, "learning_rate": 1.839870685489166e-06, "loss": 0.239, "step": 45500 }, { "epoch": 5.91, "learning_rate": 1.8388883237525765e-06, "loss": 0.2291, "step": 45510 }, { "epoch": 5.91, "learning_rate": 1.837906055016551e-06, "loss": 0.223, "step": 45520 }, { "epoch": 5.91, "learning_rate": 1.8369238794620324e-06, "loss": 0.2125, "step": 45530 }, { "epoch": 5.91, "learning_rate": 1.8359417972699473e-06, "loss": 0.2182, "step": 45540 }, { "epoch": 5.91, "learning_rate": 1.8349598086212054e-06, "loss": 0.2331, "step": 45550 }, { "epoch": 5.91, "learning_rate": 1.8339779136966994e-06, "loss": 0.2319, "step": 45560 }, { "epoch": 5.91, "learning_rate": 1.832996112677303e-06, "loss": 0.2283, "step": 45570 }, { "epoch": 5.92, "learning_rate": 1.8320144057438741e-06, "loss": 0.2121, "step": 45580 }, { "epoch": 5.92, "learning_rate": 1.8310327930772536e-06, "loss": 0.2187, "step": 45590 }, { "epoch": 5.92, "learning_rate": 1.8300512748582637e-06, "loss": 0.2161, "step": 45600 }, { "epoch": 5.92, "learning_rate": 1.8290698512677094e-06, "loss": 0.2269, "step": 45610 }, { "epoch": 5.92, "learning_rate": 1.8280885224863799e-06, "loss": 0.2184, "step": 45620 }, { "epoch": 5.92, "learning_rate": 1.8271072886950454e-06, "loss": 0.2312, "step": 45630 }, { "epoch": 5.92, "learning_rate": 1.8261261500744584e-06, "loss": 0.2135, "step": 45640 }, { "epoch": 5.92, "learning_rate": 1.8251451068053545e-06, "loss": 0.2382, "step": 45650 }, { "epoch": 5.93, "learning_rate": 1.824164159068453e-06, "loss": 0.2278, "step": 45660 }, { "epoch": 5.93, "learning_rate": 1.8231833070444524e-06, "loss": 0.2038, "step": 45670 }, { "epoch": 5.93, "learning_rate": 1.8222025509140363e-06, "loss": 0.2204, "step": 45680 }, { "epoch": 5.93, "learning_rate": 1.82122189085787e-06, "loss": 0.2263, "step": 45690 }, { "epoch": 5.93, "learning_rate": 1.8202413270566007e-06, "loss": 0.2248, "step": 45700 }, { "epoch": 5.93, "learning_rate": 1.8192608596908575e-06, "loss": 0.2252, "step": 45710 }, { "epoch": 5.93, "learning_rate": 1.8182804889412533e-06, "loss": 0.2205, "step": 45720 }, { "epoch": 5.94, "learning_rate": 1.8173002149883817e-06, "loss": 0.2328, "step": 45730 }, { "epoch": 5.94, "learning_rate": 1.8163200380128185e-06, "loss": 0.2247, "step": 45740 }, { "epoch": 5.94, "learning_rate": 1.8153399581951226e-06, "loss": 0.2222, "step": 45750 }, { "epoch": 5.94, "learning_rate": 1.814359975715834e-06, "loss": 0.2321, "step": 45760 }, { "epoch": 5.94, "learning_rate": 1.8133800907554766e-06, "loss": 0.2278, "step": 45770 }, { "epoch": 5.94, "learning_rate": 1.8124003034945533e-06, "loss": 0.2248, "step": 45780 }, { "epoch": 5.94, "learning_rate": 1.8114206141135513e-06, "loss": 0.2241, "step": 45790 }, { "epoch": 5.94, "learning_rate": 1.8104410227929402e-06, "loss": 0.2294, "step": 45800 }, { "epoch": 5.95, "learning_rate": 1.8094615297131688e-06, "loss": 0.2288, "step": 45810 }, { "epoch": 5.95, "learning_rate": 1.8084821350546705e-06, "loss": 0.2315, "step": 45820 }, { "epoch": 5.95, "learning_rate": 1.8075028389978592e-06, "loss": 0.2188, "step": 45830 }, { "epoch": 5.95, "learning_rate": 1.8065236417231318e-06, "loss": 0.2262, "step": 45840 }, { "epoch": 5.95, "learning_rate": 1.8055445434108652e-06, "loss": 0.2231, "step": 45850 }, { "epoch": 5.95, "learning_rate": 1.8045655442414194e-06, "loss": 0.2095, "step": 45860 }, { "epoch": 5.95, "learning_rate": 1.8035866443951367e-06, "loss": 0.2199, "step": 45870 }, { "epoch": 5.95, "learning_rate": 1.802607844052339e-06, "loss": 0.2351, "step": 45880 }, { "epoch": 5.96, "learning_rate": 1.8016291433933312e-06, "loss": 0.2249, "step": 45890 }, { "epoch": 5.96, "learning_rate": 1.8006505425984003e-06, "loss": 0.2323, "step": 45900 }, { "epoch": 5.96, "learning_rate": 1.7996720418478148e-06, "loss": 0.22, "step": 45910 }, { "epoch": 5.96, "learning_rate": 1.7986936413218231e-06, "loss": 0.2084, "step": 45920 }, { "epoch": 5.96, "learning_rate": 1.7977153412006566e-06, "loss": 0.2184, "step": 45930 }, { "epoch": 5.96, "learning_rate": 1.7967371416645291e-06, "loss": 0.2082, "step": 45940 }, { "epoch": 5.96, "learning_rate": 1.7957590428936334e-06, "loss": 0.2316, "step": 45950 }, { "epoch": 5.96, "learning_rate": 1.7947810450681454e-06, "loss": 0.237, "step": 45960 }, { "epoch": 5.97, "learning_rate": 1.7938031483682224e-06, "loss": 0.2213, "step": 45970 }, { "epoch": 5.97, "learning_rate": 1.792825352974003e-06, "loss": 0.228, "step": 45980 }, { "epoch": 5.97, "learning_rate": 1.7918476590656058e-06, "loss": 0.2245, "step": 45990 }, { "epoch": 5.97, "learning_rate": 1.7908700668231328e-06, "loss": 0.2207, "step": 46000 }, { "epoch": 5.97, "learning_rate": 1.7898925764266661e-06, "loss": 0.2353, "step": 46010 }, { "epoch": 5.97, "learning_rate": 1.7889151880562686e-06, "loss": 0.2166, "step": 46020 }, { "epoch": 5.97, "learning_rate": 1.7879379018919852e-06, "loss": 0.2291, "step": 46030 }, { "epoch": 5.98, "learning_rate": 1.7869607181138427e-06, "loss": 0.2209, "step": 46040 }, { "epoch": 5.98, "learning_rate": 1.7859836369018464e-06, "loss": 0.2421, "step": 46050 }, { "epoch": 5.98, "learning_rate": 1.785006658435986e-06, "loss": 0.2231, "step": 46060 }, { "epoch": 5.98, "learning_rate": 1.7840297828962297e-06, "loss": 0.2376, "step": 46070 }, { "epoch": 5.98, "learning_rate": 1.7830530104625288e-06, "loss": 0.2346, "step": 46080 }, { "epoch": 5.98, "learning_rate": 1.782076341314813e-06, "loss": 0.2224, "step": 46090 }, { "epoch": 5.98, "learning_rate": 1.7810997756329959e-06, "loss": 0.2125, "step": 46100 }, { "epoch": 5.98, "learning_rate": 1.78012331359697e-06, "loss": 0.2376, "step": 46110 }, { "epoch": 5.99, "learning_rate": 1.7791469553866096e-06, "loss": 0.2179, "step": 46120 }, { "epoch": 5.99, "learning_rate": 1.7781707011817691e-06, "loss": 0.2274, "step": 46130 }, { "epoch": 5.99, "learning_rate": 1.7771945511622847e-06, "loss": 0.2256, "step": 46140 }, { "epoch": 5.99, "learning_rate": 1.7762185055079736e-06, "loss": 0.2342, "step": 46150 }, { "epoch": 5.99, "learning_rate": 1.775242564398632e-06, "loss": 0.2364, "step": 46160 }, { "epoch": 5.99, "learning_rate": 1.7742667280140384e-06, "loss": 0.2099, "step": 46170 }, { "epoch": 5.99, "learning_rate": 1.7732909965339526e-06, "loss": 0.229, "step": 46180 }, { "epoch": 5.99, "learning_rate": 1.7723153701381121e-06, "loss": 0.2346, "step": 46190 }, { "epoch": 6.0, "learning_rate": 1.7713398490062387e-06, "loss": 0.2261, "step": 46200 }, { "epoch": 6.0, "learning_rate": 1.7703644333180326e-06, "loss": 0.2415, "step": 46210 }, { "epoch": 6.0, "learning_rate": 1.7693891232531752e-06, "loss": 0.2472, "step": 46220 }, { "epoch": 6.0, "learning_rate": 1.768413918991328e-06, "loss": 0.2347, "step": 46230 }, { "epoch": 6.0, "eval_loss": 0.22818733751773834, "eval_runtime": 60.6958, "eval_samples_per_second": 179.205, "eval_steps_per_second": 22.407, "step": 46230 }, { "epoch": 6.0, "learning_rate": 1.7674388207121334e-06, "loss": 0.2307, "step": 46240 }, { "epoch": 6.0, "learning_rate": 1.7664638285952152e-06, "loss": 0.2243, "step": 46250 }, { "epoch": 6.0, "learning_rate": 1.7654889428201754e-06, "loss": 0.2287, "step": 46260 }, { "epoch": 6.01, "learning_rate": 1.7645141635665982e-06, "loss": 0.231, "step": 46270 }, { "epoch": 6.01, "learning_rate": 1.7635394910140475e-06, "loss": 0.2317, "step": 46280 }, { "epoch": 6.01, "learning_rate": 1.7625649253420687e-06, "loss": 0.2299, "step": 46290 }, { "epoch": 6.01, "learning_rate": 1.7615904667301853e-06, "loss": 0.2227, "step": 46300 }, { "epoch": 6.01, "learning_rate": 1.7606161153579028e-06, "loss": 0.2379, "step": 46310 }, { "epoch": 6.01, "learning_rate": 1.7596418714047068e-06, "loss": 0.2289, "step": 46320 }, { "epoch": 6.01, "learning_rate": 1.758667735050062e-06, "loss": 0.2114, "step": 46330 }, { "epoch": 6.01, "learning_rate": 1.7576937064734145e-06, "loss": 0.2333, "step": 46340 }, { "epoch": 6.02, "learning_rate": 1.7567197858541899e-06, "loss": 0.2263, "step": 46350 }, { "epoch": 6.02, "learning_rate": 1.7557459733717945e-06, "loss": 0.2269, "step": 46360 }, { "epoch": 6.02, "learning_rate": 1.754772269205614e-06, "loss": 0.2266, "step": 46370 }, { "epoch": 6.02, "learning_rate": 1.7537986735350142e-06, "loss": 0.2229, "step": 46380 }, { "epoch": 6.02, "learning_rate": 1.7528251865393422e-06, "loss": 0.2295, "step": 46390 }, { "epoch": 6.02, "learning_rate": 1.7518518083979225e-06, "loss": 0.2294, "step": 46400 }, { "epoch": 6.02, "learning_rate": 1.7508785392900618e-06, "loss": 0.2262, "step": 46410 }, { "epoch": 6.02, "learning_rate": 1.749905379395046e-06, "loss": 0.2206, "step": 46420 }, { "epoch": 6.03, "learning_rate": 1.7489323288921413e-06, "loss": 0.2195, "step": 46430 }, { "epoch": 6.03, "learning_rate": 1.7479593879605927e-06, "loss": 0.2155, "step": 46440 }, { "epoch": 6.03, "learning_rate": 1.746986556779626e-06, "loss": 0.2282, "step": 46450 }, { "epoch": 6.03, "learning_rate": 1.7460138355284463e-06, "loss": 0.2401, "step": 46460 }, { "epoch": 6.03, "learning_rate": 1.7450412243862382e-06, "loss": 0.2323, "step": 46470 }, { "epoch": 6.03, "learning_rate": 1.7440687235321666e-06, "loss": 0.2335, "step": 46480 }, { "epoch": 6.03, "learning_rate": 1.7430963331453759e-06, "loss": 0.2266, "step": 46490 }, { "epoch": 6.04, "learning_rate": 1.742124053404991e-06, "loss": 0.2184, "step": 46500 }, { "epoch": 6.04, "learning_rate": 1.7411518844901141e-06, "loss": 0.2289, "step": 46510 }, { "epoch": 6.04, "learning_rate": 1.7401798265798294e-06, "loss": 0.2331, "step": 46520 }, { "epoch": 6.04, "learning_rate": 1.7392078798531995e-06, "loss": 0.2282, "step": 46530 }, { "epoch": 6.04, "learning_rate": 1.7382360444892662e-06, "loss": 0.2383, "step": 46540 }, { "epoch": 6.04, "learning_rate": 1.7372643206670516e-06, "loss": 0.2263, "step": 46550 }, { "epoch": 6.04, "learning_rate": 1.736292708565557e-06, "loss": 0.2228, "step": 46560 }, { "epoch": 6.04, "learning_rate": 1.7353212083637634e-06, "loss": 0.2355, "step": 46570 }, { "epoch": 6.05, "learning_rate": 1.7343498202406301e-06, "loss": 0.2217, "step": 46580 }, { "epoch": 6.05, "learning_rate": 1.7333785443750966e-06, "loss": 0.2328, "step": 46590 }, { "epoch": 6.05, "learning_rate": 1.7324073809460827e-06, "loss": 0.2408, "step": 46600 }, { "epoch": 6.05, "learning_rate": 1.731436330132485e-06, "loss": 0.2203, "step": 46610 }, { "epoch": 6.05, "learning_rate": 1.7304653921131812e-06, "loss": 0.204, "step": 46620 }, { "epoch": 6.05, "learning_rate": 1.7294945670670278e-06, "loss": 0.2401, "step": 46630 }, { "epoch": 6.05, "learning_rate": 1.7285238551728615e-06, "loss": 0.2258, "step": 46640 }, { "epoch": 6.05, "learning_rate": 1.7275532566094953e-06, "loss": 0.2166, "step": 46650 }, { "epoch": 6.06, "learning_rate": 1.7265827715557243e-06, "loss": 0.2279, "step": 46660 }, { "epoch": 6.06, "learning_rate": 1.7256124001903217e-06, "loss": 0.2358, "step": 46670 }, { "epoch": 6.06, "learning_rate": 1.7246421426920388e-06, "loss": 0.2204, "step": 46680 }, { "epoch": 6.06, "learning_rate": 1.723671999239607e-06, "loss": 0.2375, "step": 46690 }, { "epoch": 6.06, "learning_rate": 1.7227019700117365e-06, "loss": 0.227, "step": 46700 }, { "epoch": 6.06, "learning_rate": 1.7217320551871169e-06, "loss": 0.2246, "step": 46710 }, { "epoch": 6.06, "learning_rate": 1.7207622549444157e-06, "loss": 0.2149, "step": 46720 }, { "epoch": 6.06, "learning_rate": 1.719792569462279e-06, "loss": 0.2266, "step": 46730 }, { "epoch": 6.07, "learning_rate": 1.7188229989193344e-06, "loss": 0.229, "step": 46740 }, { "epoch": 6.07, "learning_rate": 1.7178535434941847e-06, "loss": 0.2218, "step": 46750 }, { "epoch": 6.07, "learning_rate": 1.7168842033654143e-06, "loss": 0.2346, "step": 46760 }, { "epoch": 6.07, "learning_rate": 1.7159149787115846e-06, "loss": 0.2468, "step": 46770 }, { "epoch": 6.07, "learning_rate": 1.7149458697112376e-06, "loss": 0.2339, "step": 46780 }, { "epoch": 6.07, "learning_rate": 1.7139768765428916e-06, "loss": 0.2362, "step": 46790 }, { "epoch": 6.07, "learning_rate": 1.7130079993850456e-06, "loss": 0.2327, "step": 46800 }, { "epoch": 6.08, "learning_rate": 1.7120392384161766e-06, "loss": 0.235, "step": 46810 }, { "epoch": 6.08, "learning_rate": 1.7110705938147391e-06, "loss": 0.2313, "step": 46820 }, { "epoch": 6.08, "learning_rate": 1.7101020657591678e-06, "loss": 0.2181, "step": 46830 }, { "epoch": 6.08, "learning_rate": 1.709133654427875e-06, "loss": 0.2132, "step": 46840 }, { "epoch": 6.08, "learning_rate": 1.7081653599992524e-06, "loss": 0.238, "step": 46850 }, { "epoch": 6.08, "learning_rate": 1.7071971826516687e-06, "loss": 0.2149, "step": 46860 }, { "epoch": 6.08, "learning_rate": 1.7062291225634718e-06, "loss": 0.2033, "step": 46870 }, { "epoch": 6.08, "learning_rate": 1.705261179912989e-06, "loss": 0.212, "step": 46880 }, { "epoch": 6.09, "learning_rate": 1.704293354878524e-06, "loss": 0.2319, "step": 46890 }, { "epoch": 6.09, "learning_rate": 1.7033256476383596e-06, "loss": 0.2186, "step": 46900 }, { "epoch": 6.09, "learning_rate": 1.7023580583707587e-06, "loss": 0.2314, "step": 46910 }, { "epoch": 6.09, "learning_rate": 1.7013905872539595e-06, "loss": 0.2363, "step": 46920 }, { "epoch": 6.09, "learning_rate": 1.70042323446618e-06, "loss": 0.2362, "step": 46930 }, { "epoch": 6.09, "learning_rate": 1.6994560001856167e-06, "loss": 0.2417, "step": 46940 }, { "epoch": 6.09, "learning_rate": 1.6984888845904443e-06, "loss": 0.2177, "step": 46950 }, { "epoch": 6.09, "learning_rate": 1.6975218878588136e-06, "loss": 0.2102, "step": 46960 }, { "epoch": 6.1, "learning_rate": 1.6965550101688563e-06, "loss": 0.2196, "step": 46970 }, { "epoch": 6.1, "learning_rate": 1.6955882516986808e-06, "loss": 0.2173, "step": 46980 }, { "epoch": 6.1, "learning_rate": 1.6946216126263733e-06, "loss": 0.24, "step": 46990 }, { "epoch": 6.1, "learning_rate": 1.6936550931299982e-06, "loss": 0.2156, "step": 47000 }, { "epoch": 6.1, "learning_rate": 1.6926886933875985e-06, "loss": 0.2205, "step": 47010 }, { "epoch": 6.1, "learning_rate": 1.691722413577195e-06, "loss": 0.2277, "step": 47020 }, { "epoch": 6.1, "learning_rate": 1.690756253876785e-06, "loss": 0.23, "step": 47030 }, { "epoch": 6.11, "learning_rate": 1.6897902144643458e-06, "loss": 0.2163, "step": 47040 }, { "epoch": 6.11, "learning_rate": 1.6888242955178312e-06, "loss": 0.232, "step": 47050 }, { "epoch": 6.11, "learning_rate": 1.687858497215173e-06, "loss": 0.2283, "step": 47060 }, { "epoch": 6.11, "learning_rate": 1.6868928197342806e-06, "loss": 0.2085, "step": 47070 }, { "epoch": 6.11, "learning_rate": 1.6859272632530415e-06, "loss": 0.2247, "step": 47080 }, { "epoch": 6.11, "learning_rate": 1.684961827949322e-06, "loss": 0.234, "step": 47090 }, { "epoch": 6.11, "learning_rate": 1.6839965140009634e-06, "loss": 0.2283, "step": 47100 }, { "epoch": 6.11, "learning_rate": 1.683031321585787e-06, "loss": 0.2352, "step": 47110 }, { "epoch": 6.12, "learning_rate": 1.6820662508815912e-06, "loss": 0.2267, "step": 47120 }, { "epoch": 6.12, "learning_rate": 1.6811013020661505e-06, "loss": 0.2449, "step": 47130 }, { "epoch": 6.12, "learning_rate": 1.6801364753172188e-06, "loss": 0.2313, "step": 47140 }, { "epoch": 6.12, "learning_rate": 1.6791717708125266e-06, "loss": 0.2249, "step": 47150 }, { "epoch": 6.12, "learning_rate": 1.6782071887297832e-06, "loss": 0.2253, "step": 47160 }, { "epoch": 6.12, "learning_rate": 1.6772427292466728e-06, "loss": 0.2209, "step": 47170 }, { "epoch": 6.12, "learning_rate": 1.6762783925408588e-06, "loss": 0.2243, "step": 47180 }, { "epoch": 6.12, "learning_rate": 1.6753141787899824e-06, "loss": 0.2382, "step": 47190 }, { "epoch": 6.13, "learning_rate": 1.6743500881716603e-06, "loss": 0.227, "step": 47200 }, { "epoch": 6.13, "learning_rate": 1.6733861208634882e-06, "loss": 0.2313, "step": 47210 }, { "epoch": 6.13, "learning_rate": 1.672422277043039e-06, "loss": 0.2405, "step": 47220 }, { "epoch": 6.13, "learning_rate": 1.6714585568878619e-06, "loss": 0.2341, "step": 47230 }, { "epoch": 6.13, "learning_rate": 1.6704949605754836e-06, "loss": 0.2352, "step": 47240 }, { "epoch": 6.13, "learning_rate": 1.6695314882834078e-06, "loss": 0.2333, "step": 47250 }, { "epoch": 6.13, "learning_rate": 1.668568140189117e-06, "loss": 0.2296, "step": 47260 }, { "epoch": 6.13, "learning_rate": 1.6676049164700673e-06, "loss": 0.2267, "step": 47270 }, { "epoch": 6.14, "learning_rate": 1.6666418173036975e-06, "loss": 0.2154, "step": 47280 }, { "epoch": 6.14, "learning_rate": 1.6656788428674171e-06, "loss": 0.2397, "step": 47290 }, { "epoch": 6.14, "learning_rate": 1.664715993338617e-06, "loss": 0.242, "step": 47300 }, { "epoch": 6.14, "learning_rate": 1.663753268894663e-06, "loss": 0.2332, "step": 47310 }, { "epoch": 6.14, "learning_rate": 1.6627906697129001e-06, "loss": 0.215, "step": 47320 }, { "epoch": 6.14, "learning_rate": 1.6618281959706464e-06, "loss": 0.2189, "step": 47330 }, { "epoch": 6.14, "learning_rate": 1.6608658478452005e-06, "loss": 0.2266, "step": 47340 }, { "epoch": 6.15, "learning_rate": 1.6599036255138372e-06, "loss": 0.2449, "step": 47350 }, { "epoch": 6.15, "learning_rate": 1.658941529153806e-06, "loss": 0.2266, "step": 47360 }, { "epoch": 6.15, "learning_rate": 1.6579795589423348e-06, "loss": 0.2215, "step": 47370 }, { "epoch": 6.15, "learning_rate": 1.6570177150566292e-06, "loss": 0.2354, "step": 47380 }, { "epoch": 6.15, "learning_rate": 1.6560559976738702e-06, "loss": 0.2254, "step": 47390 }, { "epoch": 6.15, "learning_rate": 1.6550944069712148e-06, "loss": 0.2389, "step": 47400 }, { "epoch": 6.15, "learning_rate": 1.6541329431257983e-06, "loss": 0.2313, "step": 47410 }, { "epoch": 6.15, "learning_rate": 1.6531716063147323e-06, "loss": 0.2084, "step": 47420 }, { "epoch": 6.16, "learning_rate": 1.6522103967151036e-06, "loss": 0.2254, "step": 47430 }, { "epoch": 6.16, "learning_rate": 1.651249314503977e-06, "loss": 0.2249, "step": 47440 }, { "epoch": 6.16, "learning_rate": 1.6502883598583943e-06, "loss": 0.2263, "step": 47450 }, { "epoch": 6.16, "learning_rate": 1.6493275329553727e-06, "loss": 0.2259, "step": 47460 }, { "epoch": 6.16, "learning_rate": 1.6483668339719047e-06, "loss": 0.2141, "step": 47470 }, { "epoch": 6.16, "learning_rate": 1.647406263084962e-06, "loss": 0.228, "step": 47480 }, { "epoch": 6.16, "learning_rate": 1.6464458204714914e-06, "loss": 0.2386, "step": 47490 }, { "epoch": 6.16, "learning_rate": 1.645485506308415e-06, "loss": 0.2225, "step": 47500 }, { "epoch": 6.17, "learning_rate": 1.6445253207726334e-06, "loss": 0.2236, "step": 47510 }, { "epoch": 6.17, "learning_rate": 1.6435652640410214e-06, "loss": 0.2174, "step": 47520 }, { "epoch": 6.17, "learning_rate": 1.6426053362904322e-06, "loss": 0.2395, "step": 47530 }, { "epoch": 6.17, "learning_rate": 1.6416455376976926e-06, "loss": 0.2323, "step": 47540 }, { "epoch": 6.17, "learning_rate": 1.6406858684396082e-06, "loss": 0.2307, "step": 47550 }, { "epoch": 6.17, "learning_rate": 1.6397263286929597e-06, "loss": 0.2184, "step": 47560 }, { "epoch": 6.17, "learning_rate": 1.6387669186345032e-06, "loss": 0.218, "step": 47570 }, { "epoch": 6.18, "learning_rate": 1.637807638440972e-06, "loss": 0.2131, "step": 47580 }, { "epoch": 6.18, "learning_rate": 1.6368484882890746e-06, "loss": 0.246, "step": 47590 }, { "epoch": 6.18, "learning_rate": 1.6358894683554972e-06, "loss": 0.2224, "step": 47600 }, { "epoch": 6.18, "learning_rate": 1.6349305788168997e-06, "loss": 0.2229, "step": 47610 }, { "epoch": 6.18, "learning_rate": 1.6339718198499195e-06, "loss": 0.2435, "step": 47620 }, { "epoch": 6.18, "learning_rate": 1.63301319163117e-06, "loss": 0.2223, "step": 47630 }, { "epoch": 6.18, "learning_rate": 1.6320546943372394e-06, "loss": 0.2244, "step": 47640 }, { "epoch": 6.18, "learning_rate": 1.6310963281446929e-06, "loss": 0.2429, "step": 47650 }, { "epoch": 6.19, "learning_rate": 1.630138093230071e-06, "loss": 0.2237, "step": 47660 }, { "epoch": 6.19, "learning_rate": 1.6291799897698906e-06, "loss": 0.2333, "step": 47670 }, { "epoch": 6.19, "learning_rate": 1.628222017940643e-06, "loss": 0.2171, "step": 47680 }, { "epoch": 6.19, "learning_rate": 1.6272641779187968e-06, "loss": 0.2181, "step": 47690 }, { "epoch": 6.19, "learning_rate": 1.626306469880796e-06, "loss": 0.2234, "step": 47700 }, { "epoch": 6.19, "learning_rate": 1.625348894003059e-06, "loss": 0.2434, "step": 47710 }, { "epoch": 6.19, "learning_rate": 1.6243914504619817e-06, "loss": 0.2266, "step": 47720 }, { "epoch": 6.19, "learning_rate": 1.623434139433934e-06, "loss": 0.2166, "step": 47730 }, { "epoch": 6.2, "learning_rate": 1.6224769610952636e-06, "loss": 0.2259, "step": 47740 }, { "epoch": 6.2, "learning_rate": 1.6215199156222905e-06, "loss": 0.2257, "step": 47750 }, { "epoch": 6.2, "learning_rate": 1.6205630031913131e-06, "loss": 0.2234, "step": 47760 }, { "epoch": 6.2, "learning_rate": 1.6196062239786044e-06, "loss": 0.2369, "step": 47770 }, { "epoch": 6.2, "learning_rate": 1.6186495781604119e-06, "loss": 0.213, "step": 47780 }, { "epoch": 6.2, "learning_rate": 1.6176930659129596e-06, "loss": 0.2096, "step": 47790 }, { "epoch": 6.2, "learning_rate": 1.6167366874124465e-06, "loss": 0.2269, "step": 47800 }, { "epoch": 6.21, "learning_rate": 1.6157804428350483e-06, "loss": 0.2232, "step": 47810 }, { "epoch": 6.21, "learning_rate": 1.6148243323569127e-06, "loss": 0.2303, "step": 47820 }, { "epoch": 6.21, "learning_rate": 1.6138683561541665e-06, "loss": 0.2113, "step": 47830 }, { "epoch": 6.21, "learning_rate": 1.6129125144029098e-06, "loss": 0.2315, "step": 47840 }, { "epoch": 6.21, "learning_rate": 1.6119568072792171e-06, "loss": 0.2159, "step": 47850 }, { "epoch": 6.21, "learning_rate": 1.61100123495914e-06, "loss": 0.2194, "step": 47860 }, { "epoch": 6.21, "learning_rate": 1.6100457976187049e-06, "loss": 0.2188, "step": 47870 }, { "epoch": 6.21, "learning_rate": 1.609090495433912e-06, "loss": 0.2302, "step": 47880 }, { "epoch": 6.22, "learning_rate": 1.608135328580738e-06, "loss": 0.2295, "step": 47890 }, { "epoch": 6.22, "learning_rate": 1.607180297235134e-06, "loss": 0.2258, "step": 47900 }, { "epoch": 6.22, "learning_rate": 1.606225401573027e-06, "loss": 0.2212, "step": 47910 }, { "epoch": 6.22, "learning_rate": 1.6052706417703172e-06, "loss": 0.2318, "step": 47920 }, { "epoch": 6.22, "learning_rate": 1.6043160180028815e-06, "loss": 0.2329, "step": 47930 }, { "epoch": 6.22, "learning_rate": 1.6033615304465714e-06, "loss": 0.2273, "step": 47940 }, { "epoch": 6.22, "learning_rate": 1.602407179277212e-06, "loss": 0.231, "step": 47950 }, { "epoch": 6.22, "learning_rate": 1.6014529646706056e-06, "loss": 0.2087, "step": 47960 }, { "epoch": 6.23, "learning_rate": 1.600498886802527e-06, "loss": 0.216, "step": 47970 }, { "epoch": 6.23, "learning_rate": 1.5995449458487283e-06, "loss": 0.23, "step": 47980 }, { "epoch": 6.23, "learning_rate": 1.598591141984933e-06, "loss": 0.2503, "step": 47990 }, { "epoch": 6.23, "learning_rate": 1.5976374753868425e-06, "loss": 0.2354, "step": 48000 }, { "epoch": 6.23, "learning_rate": 1.5966839462301322e-06, "loss": 0.2324, "step": 48010 }, { "epoch": 6.23, "learning_rate": 1.5957305546904503e-06, "loss": 0.2234, "step": 48020 }, { "epoch": 6.23, "learning_rate": 1.5947773009434217e-06, "loss": 0.2258, "step": 48030 }, { "epoch": 6.23, "learning_rate": 1.5938241851646455e-06, "loss": 0.2153, "step": 48040 }, { "epoch": 6.24, "learning_rate": 1.5928712075296953e-06, "loss": 0.2312, "step": 48050 }, { "epoch": 6.24, "learning_rate": 1.5919183682141186e-06, "loss": 0.2358, "step": 48060 }, { "epoch": 6.24, "learning_rate": 1.5909656673934375e-06, "loss": 0.217, "step": 48070 }, { "epoch": 6.24, "learning_rate": 1.5900131052431508e-06, "loss": 0.2234, "step": 48080 }, { "epoch": 6.24, "learning_rate": 1.5890606819387278e-06, "loss": 0.2323, "step": 48090 }, { "epoch": 6.24, "learning_rate": 1.588108397655615e-06, "loss": 0.2232, "step": 48100 }, { "epoch": 6.24, "learning_rate": 1.5871562525692336e-06, "loss": 0.2318, "step": 48110 }, { "epoch": 6.25, "learning_rate": 1.5862042468549778e-06, "loss": 0.2467, "step": 48120 }, { "epoch": 6.25, "learning_rate": 1.5852523806882163e-06, "loss": 0.2328, "step": 48130 }, { "epoch": 6.25, "learning_rate": 1.5843006542442922e-06, "loss": 0.2307, "step": 48140 }, { "epoch": 6.25, "learning_rate": 1.5833490676985239e-06, "loss": 0.231, "step": 48150 }, { "epoch": 6.25, "learning_rate": 1.582397621226202e-06, "loss": 0.2407, "step": 48160 }, { "epoch": 6.25, "learning_rate": 1.581446315002593e-06, "loss": 0.2253, "step": 48170 }, { "epoch": 6.25, "learning_rate": 1.5804951492029368e-06, "loss": 0.2294, "step": 48180 }, { "epoch": 6.25, "learning_rate": 1.5795441240024486e-06, "loss": 0.2238, "step": 48190 }, { "epoch": 6.26, "learning_rate": 1.5785932395763152e-06, "loss": 0.2425, "step": 48200 }, { "epoch": 6.26, "learning_rate": 1.5776424960997e-06, "loss": 0.2243, "step": 48210 }, { "epoch": 6.26, "learning_rate": 1.5766918937477395e-06, "loss": 0.2263, "step": 48220 }, { "epoch": 6.26, "learning_rate": 1.5757414326955434e-06, "loss": 0.2139, "step": 48230 }, { "epoch": 6.26, "learning_rate": 1.5747911131181967e-06, "loss": 0.2331, "step": 48240 }, { "epoch": 6.26, "learning_rate": 1.5738409351907576e-06, "loss": 0.2428, "step": 48250 }, { "epoch": 6.26, "learning_rate": 1.5728908990882589e-06, "loss": 0.2218, "step": 48260 }, { "epoch": 6.26, "learning_rate": 1.5719410049857058e-06, "loss": 0.2201, "step": 48270 }, { "epoch": 6.27, "learning_rate": 1.5709912530580787e-06, "loss": 0.2215, "step": 48280 }, { "epoch": 6.27, "learning_rate": 1.570041643480332e-06, "loss": 0.2251, "step": 48290 }, { "epoch": 6.27, "learning_rate": 1.5690921764273918e-06, "loss": 0.2272, "step": 48300 }, { "epoch": 6.27, "learning_rate": 1.5681428520741608e-06, "loss": 0.2198, "step": 48310 }, { "epoch": 6.27, "learning_rate": 1.5671936705955132e-06, "loss": 0.2316, "step": 48320 }, { "epoch": 6.27, "learning_rate": 1.5662446321662987e-06, "loss": 0.2297, "step": 48330 }, { "epoch": 6.27, "learning_rate": 1.5652957369613386e-06, "loss": 0.2303, "step": 48340 }, { "epoch": 6.28, "learning_rate": 1.5643469851554291e-06, "loss": 0.2198, "step": 48350 }, { "epoch": 6.28, "learning_rate": 1.5633983769233406e-06, "loss": 0.2334, "step": 48360 }, { "epoch": 6.28, "learning_rate": 1.562449912439815e-06, "loss": 0.2295, "step": 48370 }, { "epoch": 6.28, "learning_rate": 1.5615015918795693e-06, "loss": 0.2313, "step": 48380 }, { "epoch": 6.28, "learning_rate": 1.560553415417294e-06, "loss": 0.2358, "step": 48390 }, { "epoch": 6.28, "learning_rate": 1.5596053832276528e-06, "loss": 0.2337, "step": 48400 }, { "epoch": 6.28, "learning_rate": 1.558657495485282e-06, "loss": 0.232, "step": 48410 }, { "epoch": 6.28, "learning_rate": 1.5577097523647923e-06, "loss": 0.2302, "step": 48420 }, { "epoch": 6.29, "learning_rate": 1.556762154040768e-06, "loss": 0.2236, "step": 48430 }, { "epoch": 6.29, "learning_rate": 1.5558147006877653e-06, "loss": 0.2284, "step": 48440 }, { "epoch": 6.29, "learning_rate": 1.554867392480315e-06, "loss": 0.2225, "step": 48450 }, { "epoch": 6.29, "learning_rate": 1.5539202295929204e-06, "loss": 0.2305, "step": 48460 }, { "epoch": 6.29, "learning_rate": 1.5529732122000595e-06, "loss": 0.2265, "step": 48470 }, { "epoch": 6.29, "learning_rate": 1.552026340476181e-06, "loss": 0.2354, "step": 48480 }, { "epoch": 6.29, "learning_rate": 1.5510796145957083e-06, "loss": 0.2286, "step": 48490 }, { "epoch": 6.29, "learning_rate": 1.550133034733039e-06, "loss": 0.2187, "step": 48500 }, { "epoch": 6.3, "learning_rate": 1.5491866010625408e-06, "loss": 0.2276, "step": 48510 }, { "epoch": 6.3, "learning_rate": 1.5482403137585573e-06, "loss": 0.2253, "step": 48520 }, { "epoch": 6.3, "learning_rate": 1.5472941729954037e-06, "loss": 0.2163, "step": 48530 }, { "epoch": 6.3, "learning_rate": 1.5463481789473697e-06, "loss": 0.2122, "step": 48540 }, { "epoch": 6.3, "learning_rate": 1.545402331788715e-06, "loss": 0.2232, "step": 48550 }, { "epoch": 6.3, "learning_rate": 1.5444566316936754e-06, "loss": 0.2259, "step": 48560 }, { "epoch": 6.3, "learning_rate": 1.543511078836458e-06, "loss": 0.2255, "step": 48570 }, { "epoch": 6.3, "learning_rate": 1.5425656733912428e-06, "loss": 0.2296, "step": 48580 }, { "epoch": 6.31, "learning_rate": 1.541620415532183e-06, "loss": 0.2182, "step": 48590 }, { "epoch": 6.31, "learning_rate": 1.5406753054334048e-06, "loss": 0.2218, "step": 48600 }, { "epoch": 6.31, "learning_rate": 1.5397303432690072e-06, "loss": 0.2291, "step": 48610 }, { "epoch": 6.31, "learning_rate": 1.5387855292130607e-06, "loss": 0.2054, "step": 48620 }, { "epoch": 6.31, "learning_rate": 1.5378408634396101e-06, "loss": 0.2235, "step": 48630 }, { "epoch": 6.31, "learning_rate": 1.5368963461226725e-06, "loss": 0.234, "step": 48640 }, { "epoch": 6.31, "learning_rate": 1.535951977436237e-06, "loss": 0.2341, "step": 48650 }, { "epoch": 6.32, "learning_rate": 1.5350077575542655e-06, "loss": 0.2321, "step": 48660 }, { "epoch": 6.32, "learning_rate": 1.5340636866506933e-06, "loss": 0.2409, "step": 48670 }, { "epoch": 6.32, "learning_rate": 1.5331197648994282e-06, "loss": 0.2282, "step": 48680 }, { "epoch": 6.32, "learning_rate": 1.5321759924743487e-06, "loss": 0.2339, "step": 48690 }, { "epoch": 6.32, "learning_rate": 1.5312323695493075e-06, "loss": 0.2305, "step": 48700 }, { "epoch": 6.32, "learning_rate": 1.5302888962981303e-06, "loss": 0.2452, "step": 48710 }, { "epoch": 6.32, "learning_rate": 1.529345572894613e-06, "loss": 0.2399, "step": 48720 }, { "epoch": 6.32, "learning_rate": 1.5284023995125258e-06, "loss": 0.2346, "step": 48730 }, { "epoch": 6.33, "learning_rate": 1.5274593763256107e-06, "loss": 0.2296, "step": 48740 }, { "epoch": 6.33, "learning_rate": 1.5265165035075823e-06, "loss": 0.2259, "step": 48750 }, { "epoch": 6.33, "learning_rate": 1.5255737812321263e-06, "loss": 0.2283, "step": 48760 }, { "epoch": 6.33, "learning_rate": 1.524631209672902e-06, "loss": 0.2233, "step": 48770 }, { "epoch": 6.33, "learning_rate": 1.5236887890035417e-06, "loss": 0.2435, "step": 48780 }, { "epoch": 6.33, "learning_rate": 1.5227465193976463e-06, "loss": 0.2312, "step": 48790 }, { "epoch": 6.33, "learning_rate": 1.521804401028793e-06, "loss": 0.2161, "step": 48800 }, { "epoch": 6.33, "learning_rate": 1.5208624340705295e-06, "loss": 0.2226, "step": 48810 }, { "epoch": 6.34, "learning_rate": 1.5199206186963743e-06, "loss": 0.2208, "step": 48820 }, { "epoch": 6.34, "learning_rate": 1.5189789550798197e-06, "loss": 0.2153, "step": 48830 }, { "epoch": 6.34, "learning_rate": 1.5180374433943301e-06, "loss": 0.2083, "step": 48840 }, { "epoch": 6.34, "learning_rate": 1.5170960838133413e-06, "loss": 0.2181, "step": 48850 }, { "epoch": 6.34, "learning_rate": 1.5161548765102603e-06, "loss": 0.2399, "step": 48860 }, { "epoch": 6.34, "learning_rate": 1.5152138216584677e-06, "loss": 0.2377, "step": 48870 }, { "epoch": 6.34, "learning_rate": 1.514272919431315e-06, "loss": 0.2161, "step": 48880 }, { "epoch": 6.35, "learning_rate": 1.5133321700021254e-06, "loss": 0.2278, "step": 48890 }, { "epoch": 6.35, "learning_rate": 1.5123915735441947e-06, "loss": 0.228, "step": 48900 }, { "epoch": 6.35, "learning_rate": 1.51145113023079e-06, "loss": 0.2481, "step": 48910 }, { "epoch": 6.35, "learning_rate": 1.5105108402351509e-06, "loss": 0.2385, "step": 48920 }, { "epoch": 6.35, "learning_rate": 1.5095707037304874e-06, "loss": 0.2218, "step": 48930 }, { "epoch": 6.35, "learning_rate": 1.5086307208899825e-06, "loss": 0.2189, "step": 48940 }, { "epoch": 6.35, "learning_rate": 1.5076908918867908e-06, "loss": 0.2225, "step": 48950 }, { "epoch": 6.35, "learning_rate": 1.506751216894037e-06, "loss": 0.213, "step": 48960 }, { "epoch": 6.36, "learning_rate": 1.5058116960848195e-06, "loss": 0.2364, "step": 48970 }, { "epoch": 6.36, "learning_rate": 1.5048723296322073e-06, "loss": 0.2266, "step": 48980 }, { "epoch": 6.36, "learning_rate": 1.5039331177092413e-06, "loss": 0.2228, "step": 48990 }, { "epoch": 6.36, "learning_rate": 1.5029940604889331e-06, "loss": 0.2262, "step": 49000 }, { "epoch": 6.36, "learning_rate": 1.5020551581442668e-06, "loss": 0.2368, "step": 49010 }, { "epoch": 6.36, "learning_rate": 1.501116410848198e-06, "loss": 0.2339, "step": 49020 }, { "epoch": 6.36, "learning_rate": 1.5001778187736526e-06, "loss": 0.241, "step": 49030 }, { "epoch": 6.36, "learning_rate": 1.4992393820935286e-06, "loss": 0.2315, "step": 49040 }, { "epoch": 6.37, "learning_rate": 1.498301100980696e-06, "loss": 0.22, "step": 49050 }, { "epoch": 6.37, "learning_rate": 1.4973629756079957e-06, "loss": 0.2328, "step": 49060 }, { "epoch": 6.37, "learning_rate": 1.496425006148239e-06, "loss": 0.2283, "step": 49070 }, { "epoch": 6.37, "learning_rate": 1.4954871927742095e-06, "loss": 0.2353, "step": 49080 }, { "epoch": 6.37, "learning_rate": 1.4945495356586624e-06, "loss": 0.2288, "step": 49090 }, { "epoch": 6.37, "learning_rate": 1.4936120349743225e-06, "loss": 0.2338, "step": 49100 }, { "epoch": 6.37, "learning_rate": 1.4926746908938874e-06, "loss": 0.219, "step": 49110 }, { "epoch": 6.38, "learning_rate": 1.4917375035900248e-06, "loss": 0.2345, "step": 49120 }, { "epoch": 6.38, "learning_rate": 1.490800473235375e-06, "loss": 0.2262, "step": 49130 }, { "epoch": 6.38, "learning_rate": 1.489863600002547e-06, "loss": 0.2274, "step": 49140 }, { "epoch": 6.38, "learning_rate": 1.488926884064123e-06, "loss": 0.2259, "step": 49150 }, { "epoch": 6.38, "learning_rate": 1.487990325592656e-06, "loss": 0.2177, "step": 49160 }, { "epoch": 6.38, "learning_rate": 1.4870539247606678e-06, "loss": 0.237, "step": 49170 }, { "epoch": 6.38, "learning_rate": 1.486117681740654e-06, "loss": 0.2326, "step": 49180 }, { "epoch": 6.38, "learning_rate": 1.4851815967050793e-06, "loss": 0.2286, "step": 49190 }, { "epoch": 6.39, "learning_rate": 1.4842456698263809e-06, "loss": 0.2304, "step": 49200 }, { "epoch": 6.39, "learning_rate": 1.483309901276965e-06, "loss": 0.225, "step": 49210 }, { "epoch": 6.39, "learning_rate": 1.4823742912292092e-06, "loss": 0.2359, "step": 49220 }, { "epoch": 6.39, "learning_rate": 1.4814388398554633e-06, "loss": 0.2497, "step": 49230 }, { "epoch": 6.39, "learning_rate": 1.480503547328046e-06, "loss": 0.2243, "step": 49240 }, { "epoch": 6.39, "learning_rate": 1.4795684138192478e-06, "loss": 0.2189, "step": 49250 }, { "epoch": 6.39, "learning_rate": 1.4786334395013293e-06, "loss": 0.2032, "step": 49260 }, { "epoch": 6.39, "learning_rate": 1.477698624546523e-06, "loss": 0.22, "step": 49270 }, { "epoch": 6.4, "learning_rate": 1.47676396912703e-06, "loss": 0.2332, "step": 49280 }, { "epoch": 6.4, "learning_rate": 1.475829473415024e-06, "loss": 0.2295, "step": 49290 }, { "epoch": 6.4, "learning_rate": 1.4748951375826482e-06, "loss": 0.237, "step": 49300 }, { "epoch": 6.4, "learning_rate": 1.4739609618020165e-06, "loss": 0.2225, "step": 49310 }, { "epoch": 6.4, "learning_rate": 1.473026946245213e-06, "loss": 0.2289, "step": 49320 }, { "epoch": 6.4, "learning_rate": 1.4720930910842932e-06, "loss": 0.2339, "step": 49330 }, { "epoch": 6.4, "learning_rate": 1.4711593964912832e-06, "loss": 0.2107, "step": 49340 }, { "epoch": 6.4, "learning_rate": 1.4702258626381776e-06, "loss": 0.2245, "step": 49350 }, { "epoch": 6.41, "learning_rate": 1.4692924896969432e-06, "loss": 0.2357, "step": 49360 }, { "epoch": 6.41, "learning_rate": 1.4683592778395172e-06, "loss": 0.2258, "step": 49370 }, { "epoch": 6.41, "learning_rate": 1.4674262272378054e-06, "loss": 0.241, "step": 49380 }, { "epoch": 6.41, "learning_rate": 1.4664933380636857e-06, "loss": 0.24, "step": 49390 }, { "epoch": 6.41, "learning_rate": 1.4655606104890053e-06, "loss": 0.2294, "step": 49400 }, { "epoch": 6.41, "learning_rate": 1.4646280446855828e-06, "loss": 0.2204, "step": 49410 }, { "epoch": 6.41, "learning_rate": 1.4636956408252052e-06, "loss": 0.2134, "step": 49420 }, { "epoch": 6.42, "learning_rate": 1.462763399079631e-06, "loss": 0.231, "step": 49430 }, { "epoch": 6.42, "learning_rate": 1.4618313196205886e-06, "loss": 0.2126, "step": 49440 }, { "epoch": 6.42, "learning_rate": 1.4608994026197755e-06, "loss": 0.2328, "step": 49450 }, { "epoch": 6.42, "learning_rate": 1.459967648248861e-06, "loss": 0.2462, "step": 49460 }, { "epoch": 6.42, "learning_rate": 1.459036056679483e-06, "loss": 0.2219, "step": 49470 }, { "epoch": 6.42, "learning_rate": 1.4581046280832508e-06, "loss": 0.2198, "step": 49480 }, { "epoch": 6.42, "learning_rate": 1.4571733626317419e-06, "loss": 0.22, "step": 49490 }, { "epoch": 6.42, "learning_rate": 1.4562422604965052e-06, "loss": 0.2173, "step": 49500 }, { "epoch": 6.43, "learning_rate": 1.455311321849059e-06, "loss": 0.2365, "step": 49510 }, { "epoch": 6.43, "learning_rate": 1.4543805468608914e-06, "loss": 0.2226, "step": 49520 }, { "epoch": 6.43, "learning_rate": 1.4534499357034603e-06, "loss": 0.2311, "step": 49530 }, { "epoch": 6.43, "learning_rate": 1.4525194885481936e-06, "loss": 0.2319, "step": 49540 }, { "epoch": 6.43, "learning_rate": 1.4515892055664898e-06, "loss": 0.2457, "step": 49550 }, { "epoch": 6.43, "learning_rate": 1.450659086929715e-06, "loss": 0.2056, "step": 49560 }, { "epoch": 6.43, "learning_rate": 1.4497291328092069e-06, "loss": 0.2149, "step": 49570 }, { "epoch": 6.43, "learning_rate": 1.4487993433762728e-06, "loss": 0.2264, "step": 49580 }, { "epoch": 6.44, "learning_rate": 1.4478697188021885e-06, "loss": 0.22, "step": 49590 }, { "epoch": 6.44, "learning_rate": 1.4469402592582003e-06, "loss": 0.2184, "step": 49600 }, { "epoch": 6.44, "learning_rate": 1.4460109649155242e-06, "loss": 0.2358, "step": 49610 }, { "epoch": 6.44, "learning_rate": 1.4450818359453457e-06, "loss": 0.2316, "step": 49620 }, { "epoch": 6.44, "learning_rate": 1.4441528725188187e-06, "loss": 0.2162, "step": 49630 }, { "epoch": 6.44, "learning_rate": 1.4432240748070685e-06, "loss": 0.2463, "step": 49640 }, { "epoch": 6.44, "learning_rate": 1.4422954429811887e-06, "loss": 0.2249, "step": 49650 }, { "epoch": 6.45, "learning_rate": 1.4413669772122421e-06, "loss": 0.2421, "step": 49660 }, { "epoch": 6.45, "learning_rate": 1.4404386776712613e-06, "loss": 0.2198, "step": 49670 }, { "epoch": 6.45, "learning_rate": 1.4395105445292491e-06, "loss": 0.227, "step": 49680 }, { "epoch": 6.45, "learning_rate": 1.438582577957177e-06, "loss": 0.2283, "step": 49690 }, { "epoch": 6.45, "learning_rate": 1.4376547781259846e-06, "loss": 0.2282, "step": 49700 }, { "epoch": 6.45, "learning_rate": 1.4367271452065827e-06, "loss": 0.2305, "step": 49710 }, { "epoch": 6.45, "learning_rate": 1.4357996793698505e-06, "loss": 0.2214, "step": 49720 }, { "epoch": 6.45, "learning_rate": 1.4348723807866363e-06, "loss": 0.2278, "step": 49730 }, { "epoch": 6.46, "learning_rate": 1.4339452496277576e-06, "loss": 0.2319, "step": 49740 }, { "epoch": 6.46, "learning_rate": 1.433018286064002e-06, "loss": 0.2223, "step": 49750 }, { "epoch": 6.46, "learning_rate": 1.4320914902661243e-06, "loss": 0.22, "step": 49760 }, { "epoch": 6.46, "learning_rate": 1.4311648624048502e-06, "loss": 0.2314, "step": 49770 }, { "epoch": 6.46, "learning_rate": 1.4302384026508733e-06, "loss": 0.2221, "step": 49780 }, { "epoch": 6.46, "learning_rate": 1.4293121111748581e-06, "loss": 0.2318, "step": 49790 }, { "epoch": 6.46, "learning_rate": 1.4283859881474353e-06, "loss": 0.226, "step": 49800 }, { "epoch": 6.46, "learning_rate": 1.4274600337392064e-06, "loss": 0.2359, "step": 49810 }, { "epoch": 6.47, "learning_rate": 1.426534248120742e-06, "loss": 0.2199, "step": 49820 }, { "epoch": 6.47, "learning_rate": 1.4256086314625802e-06, "loss": 0.2445, "step": 49830 }, { "epoch": 6.47, "learning_rate": 1.4246831839352294e-06, "loss": 0.2278, "step": 49840 }, { "epoch": 6.47, "learning_rate": 1.4237579057091662e-06, "loss": 0.2269, "step": 49850 }, { "epoch": 6.47, "learning_rate": 1.4228327969548362e-06, "loss": 0.2115, "step": 49860 }, { "epoch": 6.47, "learning_rate": 1.421907857842653e-06, "loss": 0.2253, "step": 49870 }, { "epoch": 6.47, "learning_rate": 1.4209830885430005e-06, "loss": 0.23, "step": 49880 }, { "epoch": 6.48, "learning_rate": 1.4200584892262305e-06, "loss": 0.2304, "step": 49890 }, { "epoch": 6.48, "learning_rate": 1.4191340600626622e-06, "loss": 0.2376, "step": 49900 }, { "epoch": 6.48, "learning_rate": 1.4182098012225856e-06, "loss": 0.2286, "step": 49910 }, { "epoch": 6.48, "learning_rate": 1.4172857128762584e-06, "loss": 0.2213, "step": 49920 }, { "epoch": 6.48, "learning_rate": 1.4163617951939076e-06, "loss": 0.231, "step": 49930 }, { "epoch": 6.48, "learning_rate": 1.415438048345726e-06, "loss": 0.2186, "step": 49940 }, { "epoch": 6.48, "learning_rate": 1.4145144725018786e-06, "loss": 0.2358, "step": 49950 }, { "epoch": 6.48, "learning_rate": 1.4135910678324977e-06, "loss": 0.223, "step": 49960 }, { "epoch": 6.49, "learning_rate": 1.4126678345076822e-06, "loss": 0.2206, "step": 49970 }, { "epoch": 6.49, "learning_rate": 1.4117447726975015e-06, "loss": 0.2289, "step": 49980 }, { "epoch": 6.49, "learning_rate": 1.4108218825719934e-06, "loss": 0.2398, "step": 49990 }, { "epoch": 6.49, "learning_rate": 1.409899164301163e-06, "loss": 0.2438, "step": 50000 }, { "epoch": 6.49, "learning_rate": 1.408976618054984e-06, "loss": 0.2384, "step": 50010 }, { "epoch": 6.49, "learning_rate": 1.4080542440033989e-06, "loss": 0.2333, "step": 50020 }, { "epoch": 6.49, "learning_rate": 1.4071320423163185e-06, "loss": 0.2292, "step": 50030 }, { "epoch": 6.49, "learning_rate": 1.4062100131636207e-06, "loss": 0.2068, "step": 50040 }, { "epoch": 6.5, "learning_rate": 1.4052881567151532e-06, "loss": 0.2242, "step": 50050 }, { "epoch": 6.5, "learning_rate": 1.4043664731407308e-06, "loss": 0.2473, "step": 50060 }, { "epoch": 6.5, "learning_rate": 1.4034449626101376e-06, "loss": 0.2245, "step": 50070 }, { "epoch": 6.5, "learning_rate": 1.402523625293124e-06, "loss": 0.2301, "step": 50080 }, { "epoch": 6.5, "learning_rate": 1.4016024613594098e-06, "loss": 0.2183, "step": 50090 }, { "epoch": 6.5, "learning_rate": 1.4006814709786836e-06, "loss": 0.215, "step": 50100 }, { "epoch": 6.5, "learning_rate": 1.3997606543205993e-06, "loss": 0.2361, "step": 50110 }, { "epoch": 6.5, "learning_rate": 1.3988400115547814e-06, "loss": 0.2263, "step": 50120 }, { "epoch": 6.51, "learning_rate": 1.3979195428508219e-06, "loss": 0.222, "step": 50130 }, { "epoch": 6.51, "learning_rate": 1.3969992483782802e-06, "loss": 0.236, "step": 50140 }, { "epoch": 6.51, "learning_rate": 1.396079128306683e-06, "loss": 0.2375, "step": 50150 }, { "epoch": 6.51, "learning_rate": 1.3951591828055264e-06, "loss": 0.2409, "step": 50160 }, { "epoch": 6.51, "learning_rate": 1.3942394120442735e-06, "loss": 0.2303, "step": 50170 }, { "epoch": 6.51, "learning_rate": 1.3933198161923548e-06, "loss": 0.2323, "step": 50180 }, { "epoch": 6.51, "learning_rate": 1.392400395419169e-06, "loss": 0.2362, "step": 50190 }, { "epoch": 6.52, "learning_rate": 1.3914811498940827e-06, "loss": 0.2259, "step": 50200 }, { "epoch": 6.52, "learning_rate": 1.3905620797864308e-06, "loss": 0.2306, "step": 50210 }, { "epoch": 6.52, "learning_rate": 1.3896431852655142e-06, "loss": 0.2303, "step": 50220 }, { "epoch": 6.52, "learning_rate": 1.3887244665006028e-06, "loss": 0.2293, "step": 50230 }, { "epoch": 6.52, "learning_rate": 1.3878059236609346e-06, "loss": 0.2304, "step": 50240 }, { "epoch": 6.52, "learning_rate": 1.3868875569157131e-06, "loss": 0.2168, "step": 50250 }, { "epoch": 6.52, "learning_rate": 1.385969366434111e-06, "loss": 0.2262, "step": 50260 }, { "epoch": 6.52, "learning_rate": 1.3850513523852685e-06, "loss": 0.2215, "step": 50270 }, { "epoch": 6.53, "learning_rate": 1.3841335149382927e-06, "loss": 0.2179, "step": 50280 }, { "epoch": 6.53, "learning_rate": 1.3832158542622588e-06, "loss": 0.2054, "step": 50290 }, { "epoch": 6.53, "learning_rate": 1.3822983705262083e-06, "loss": 0.2463, "step": 50300 }, { "epoch": 6.53, "learning_rate": 1.3813810638991522e-06, "loss": 0.2228, "step": 50310 }, { "epoch": 6.53, "learning_rate": 1.3804639345500657e-06, "loss": 0.2288, "step": 50320 }, { "epoch": 6.53, "learning_rate": 1.3795469826478943e-06, "loss": 0.2353, "step": 50330 }, { "epoch": 6.53, "learning_rate": 1.3786302083615496e-06, "loss": 0.2339, "step": 50340 }, { "epoch": 6.53, "learning_rate": 1.3777136118599111e-06, "loss": 0.2122, "step": 50350 }, { "epoch": 6.54, "learning_rate": 1.376797193311824e-06, "loss": 0.232, "step": 50360 }, { "epoch": 6.54, "learning_rate": 1.3758809528861019e-06, "loss": 0.2312, "step": 50370 }, { "epoch": 6.54, "learning_rate": 1.3749648907515265e-06, "loss": 0.223, "step": 50380 }, { "epoch": 6.54, "learning_rate": 1.3740490070768444e-06, "loss": 0.2119, "step": 50390 }, { "epoch": 6.54, "learning_rate": 1.3731333020307707e-06, "loss": 0.2175, "step": 50400 }, { "epoch": 6.54, "learning_rate": 1.3722177757819882e-06, "loss": 0.2131, "step": 50410 }, { "epoch": 6.54, "learning_rate": 1.3713024284991455e-06, "loss": 0.2338, "step": 50420 }, { "epoch": 6.55, "learning_rate": 1.3703872603508582e-06, "loss": 0.2269, "step": 50430 }, { "epoch": 6.55, "learning_rate": 1.3694722715057103e-06, "loss": 0.2184, "step": 50440 }, { "epoch": 6.55, "learning_rate": 1.3685574621322517e-06, "loss": 0.2205, "step": 50450 }, { "epoch": 6.55, "learning_rate": 1.3676428323989985e-06, "loss": 0.2176, "step": 50460 }, { "epoch": 6.55, "learning_rate": 1.366728382474436e-06, "loss": 0.2078, "step": 50470 }, { "epoch": 6.55, "learning_rate": 1.3658141125270136e-06, "loss": 0.2224, "step": 50480 }, { "epoch": 6.55, "learning_rate": 1.3649000227251503e-06, "loss": 0.2204, "step": 50490 }, { "epoch": 6.55, "learning_rate": 1.3639861132372303e-06, "loss": 0.2097, "step": 50500 }, { "epoch": 6.56, "learning_rate": 1.3630723842316037e-06, "loss": 0.2331, "step": 50510 }, { "epoch": 6.56, "learning_rate": 1.36215883587659e-06, "loss": 0.2283, "step": 50520 }, { "epoch": 6.56, "learning_rate": 1.3612454683404723e-06, "loss": 0.2286, "step": 50530 }, { "epoch": 6.56, "learning_rate": 1.360332281791504e-06, "loss": 0.2254, "step": 50540 }, { "epoch": 6.56, "learning_rate": 1.3594192763979023e-06, "loss": 0.2233, "step": 50550 }, { "epoch": 6.56, "learning_rate": 1.3585064523278508e-06, "loss": 0.2263, "step": 50560 }, { "epoch": 6.56, "learning_rate": 1.3575938097495026e-06, "loss": 0.2259, "step": 50570 }, { "epoch": 6.56, "learning_rate": 1.3566813488309746e-06, "loss": 0.2355, "step": 50580 }, { "epoch": 6.57, "learning_rate": 1.3557690697403508e-06, "loss": 0.2314, "step": 50590 }, { "epoch": 6.57, "learning_rate": 1.3548569726456836e-06, "loss": 0.2193, "step": 50600 }, { "epoch": 6.57, "learning_rate": 1.3539450577149896e-06, "loss": 0.2044, "step": 50610 }, { "epoch": 6.57, "learning_rate": 1.3530333251162513e-06, "loss": 0.212, "step": 50620 }, { "epoch": 6.57, "learning_rate": 1.3521217750174216e-06, "loss": 0.241, "step": 50630 }, { "epoch": 6.57, "learning_rate": 1.3512104075864153e-06, "loss": 0.221, "step": 50640 }, { "epoch": 6.57, "learning_rate": 1.3502992229911149e-06, "loss": 0.2448, "step": 50650 }, { "epoch": 6.57, "learning_rate": 1.3493882213993717e-06, "loss": 0.2118, "step": 50660 }, { "epoch": 6.58, "learning_rate": 1.3484774029789993e-06, "loss": 0.2365, "step": 50670 }, { "epoch": 6.58, "learning_rate": 1.3475667678977817e-06, "loss": 0.2206, "step": 50680 }, { "epoch": 6.58, "learning_rate": 1.3466563163234655e-06, "loss": 0.2245, "step": 50690 }, { "epoch": 6.58, "learning_rate": 1.345746048423764e-06, "loss": 0.2236, "step": 50700 }, { "epoch": 6.58, "learning_rate": 1.3448359643663602e-06, "loss": 0.2179, "step": 50710 }, { "epoch": 6.58, "learning_rate": 1.343926064318899e-06, "loss": 0.2312, "step": 50720 }, { "epoch": 6.58, "learning_rate": 1.3430163484489928e-06, "loss": 0.2344, "step": 50730 }, { "epoch": 6.59, "learning_rate": 1.3421068169242216e-06, "loss": 0.2311, "step": 50740 }, { "epoch": 6.59, "learning_rate": 1.3411974699121298e-06, "loss": 0.225, "step": 50750 }, { "epoch": 6.59, "learning_rate": 1.3402883075802268e-06, "loss": 0.2344, "step": 50760 }, { "epoch": 6.59, "learning_rate": 1.3393793300959915e-06, "loss": 0.2219, "step": 50770 }, { "epoch": 6.59, "learning_rate": 1.3384705376268657e-06, "loss": 0.2364, "step": 50780 }, { "epoch": 6.59, "learning_rate": 1.3375619303402572e-06, "loss": 0.2438, "step": 50790 }, { "epoch": 6.59, "learning_rate": 1.3366535084035423e-06, "loss": 0.2454, "step": 50800 }, { "epoch": 6.59, "learning_rate": 1.3357452719840594e-06, "loss": 0.2278, "step": 50810 }, { "epoch": 6.6, "learning_rate": 1.3348372212491166e-06, "loss": 0.2251, "step": 50820 }, { "epoch": 6.6, "learning_rate": 1.3339293563659854e-06, "loss": 0.2109, "step": 50830 }, { "epoch": 6.6, "learning_rate": 1.333021677501902e-06, "loss": 0.2194, "step": 50840 }, { "epoch": 6.6, "learning_rate": 1.3321141848240723e-06, "loss": 0.2419, "step": 50850 }, { "epoch": 6.6, "learning_rate": 1.3312068784996645e-06, "loss": 0.2132, "step": 50860 }, { "epoch": 6.6, "learning_rate": 1.3302997586958121e-06, "loss": 0.2312, "step": 50870 }, { "epoch": 6.6, "learning_rate": 1.3293928255796181e-06, "loss": 0.2417, "step": 50880 }, { "epoch": 6.6, "learning_rate": 1.3284860793181473e-06, "loss": 0.2228, "step": 50890 }, { "epoch": 6.61, "learning_rate": 1.3275795200784308e-06, "loss": 0.2266, "step": 50900 }, { "epoch": 6.61, "learning_rate": 1.3266731480274675e-06, "loss": 0.2264, "step": 50910 }, { "epoch": 6.61, "learning_rate": 1.3257669633322192e-06, "loss": 0.2252, "step": 50920 }, { "epoch": 6.61, "learning_rate": 1.3248609661596137e-06, "loss": 0.2217, "step": 50930 }, { "epoch": 6.61, "learning_rate": 1.3239551566765459e-06, "loss": 0.2265, "step": 50940 }, { "epoch": 6.61, "learning_rate": 1.3230495350498734e-06, "loss": 0.2283, "step": 50950 }, { "epoch": 6.61, "learning_rate": 1.3221441014464232e-06, "loss": 0.2177, "step": 50960 }, { "epoch": 6.62, "learning_rate": 1.3212388560329831e-06, "loss": 0.2367, "step": 50970 }, { "epoch": 6.62, "learning_rate": 1.3203337989763083e-06, "loss": 0.2353, "step": 50980 }, { "epoch": 6.62, "learning_rate": 1.319428930443121e-06, "loss": 0.241, "step": 50990 }, { "epoch": 6.62, "learning_rate": 1.318524250600106e-06, "loss": 0.2264, "step": 51000 }, { "epoch": 6.62, "learning_rate": 1.3176197596139137e-06, "loss": 0.2296, "step": 51010 }, { "epoch": 6.62, "learning_rate": 1.3167154576511618e-06, "loss": 0.2198, "step": 51020 }, { "epoch": 6.62, "learning_rate": 1.3158113448784312e-06, "loss": 0.2281, "step": 51030 }, { "epoch": 6.62, "learning_rate": 1.3149074214622672e-06, "loss": 0.2192, "step": 51040 }, { "epoch": 6.63, "learning_rate": 1.3140036875691839e-06, "loss": 0.2192, "step": 51050 }, { "epoch": 6.63, "learning_rate": 1.3131001433656565e-06, "loss": 0.2251, "step": 51060 }, { "epoch": 6.63, "learning_rate": 1.3121967890181265e-06, "loss": 0.2097, "step": 51070 }, { "epoch": 6.63, "learning_rate": 1.3112936246930023e-06, "loss": 0.2255, "step": 51080 }, { "epoch": 6.63, "learning_rate": 1.3103906505566554e-06, "loss": 0.2275, "step": 51090 }, { "epoch": 6.63, "learning_rate": 1.3094878667754213e-06, "loss": 0.2323, "step": 51100 }, { "epoch": 6.63, "learning_rate": 1.3085852735156037e-06, "loss": 0.2137, "step": 51110 }, { "epoch": 6.63, "learning_rate": 1.3076828709434673e-06, "loss": 0.2267, "step": 51120 }, { "epoch": 6.64, "learning_rate": 1.306780659225246e-06, "loss": 0.2428, "step": 51130 }, { "epoch": 6.64, "learning_rate": 1.3058786385271343e-06, "loss": 0.2357, "step": 51140 }, { "epoch": 6.64, "learning_rate": 1.3049768090152939e-06, "loss": 0.2291, "step": 51150 }, { "epoch": 6.64, "learning_rate": 1.3040751708558515e-06, "loss": 0.2161, "step": 51160 }, { "epoch": 6.64, "learning_rate": 1.3031737242148972e-06, "loss": 0.2192, "step": 51170 }, { "epoch": 6.64, "learning_rate": 1.3022724692584856e-06, "loss": 0.2301, "step": 51180 }, { "epoch": 6.64, "learning_rate": 1.3013714061526391e-06, "loss": 0.2283, "step": 51190 }, { "epoch": 6.65, "learning_rate": 1.3004705350633412e-06, "loss": 0.2242, "step": 51200 }, { "epoch": 6.65, "learning_rate": 1.2995698561565407e-06, "loss": 0.2274, "step": 51210 }, { "epoch": 6.65, "learning_rate": 1.298669369598153e-06, "loss": 0.2209, "step": 51220 }, { "epoch": 6.65, "learning_rate": 1.2977690755540564e-06, "loss": 0.2127, "step": 51230 }, { "epoch": 6.65, "learning_rate": 1.296868974190093e-06, "loss": 0.2358, "step": 51240 }, { "epoch": 6.65, "learning_rate": 1.2959690656720718e-06, "loss": 0.2157, "step": 51250 }, { "epoch": 6.65, "learning_rate": 1.2950693501657636e-06, "loss": 0.2056, "step": 51260 }, { "epoch": 6.65, "learning_rate": 1.2941698278369064e-06, "loss": 0.2191, "step": 51270 }, { "epoch": 6.66, "learning_rate": 1.2932704988512009e-06, "loss": 0.2305, "step": 51280 }, { "epoch": 6.66, "learning_rate": 1.2923713633743112e-06, "loss": 0.2223, "step": 51290 }, { "epoch": 6.66, "learning_rate": 1.2914724215718688e-06, "loss": 0.2112, "step": 51300 }, { "epoch": 6.66, "learning_rate": 1.290573673609467e-06, "loss": 0.2119, "step": 51310 }, { "epoch": 6.66, "learning_rate": 1.289675119652663e-06, "loss": 0.235, "step": 51320 }, { "epoch": 6.66, "learning_rate": 1.2887767598669818e-06, "loss": 0.2229, "step": 51330 }, { "epoch": 6.66, "learning_rate": 1.2878785944179087e-06, "loss": 0.2394, "step": 51340 }, { "epoch": 6.66, "learning_rate": 1.2869806234708938e-06, "loss": 0.2179, "step": 51350 }, { "epoch": 6.67, "learning_rate": 1.286082847191355e-06, "loss": 0.2343, "step": 51360 }, { "epoch": 6.67, "learning_rate": 1.2851852657446695e-06, "loss": 0.2216, "step": 51370 }, { "epoch": 6.67, "learning_rate": 1.2842878792961808e-06, "loss": 0.2296, "step": 51380 }, { "epoch": 6.67, "learning_rate": 1.2833906880111982e-06, "loss": 0.2193, "step": 51390 }, { "epoch": 6.67, "learning_rate": 1.282493692054991e-06, "loss": 0.2103, "step": 51400 }, { "epoch": 6.67, "learning_rate": 1.281596891592797e-06, "loss": 0.2368, "step": 51410 }, { "epoch": 6.67, "learning_rate": 1.2807002867898149e-06, "loss": 0.2395, "step": 51420 }, { "epoch": 6.67, "learning_rate": 1.279803877811207e-06, "loss": 0.2234, "step": 51430 }, { "epoch": 6.68, "learning_rate": 1.2789076648221033e-06, "loss": 0.2384, "step": 51440 }, { "epoch": 6.68, "learning_rate": 1.2780116479875935e-06, "loss": 0.2135, "step": 51450 }, { "epoch": 6.68, "learning_rate": 1.2771158274727322e-06, "loss": 0.2104, "step": 51460 }, { "epoch": 6.68, "learning_rate": 1.2762202034425405e-06, "loss": 0.2142, "step": 51470 }, { "epoch": 6.68, "learning_rate": 1.2753247760620002e-06, "loss": 0.2289, "step": 51480 }, { "epoch": 6.68, "learning_rate": 1.2744295454960572e-06, "loss": 0.2235, "step": 51490 }, { "epoch": 6.68, "learning_rate": 1.2735345119096237e-06, "loss": 0.2351, "step": 51500 }, { "epoch": 6.69, "learning_rate": 1.2726396754675727e-06, "loss": 0.2287, "step": 51510 }, { "epoch": 6.69, "learning_rate": 1.2717450363347412e-06, "loss": 0.2322, "step": 51520 }, { "epoch": 6.69, "learning_rate": 1.2708505946759326e-06, "loss": 0.2256, "step": 51530 }, { "epoch": 6.69, "learning_rate": 1.2699563506559103e-06, "loss": 0.2324, "step": 51540 }, { "epoch": 6.69, "learning_rate": 1.2690623044394046e-06, "loss": 0.2299, "step": 51550 }, { "epoch": 6.69, "learning_rate": 1.2681684561911066e-06, "loss": 0.2267, "step": 51560 }, { "epoch": 6.69, "learning_rate": 1.2672748060756718e-06, "loss": 0.2202, "step": 51570 }, { "epoch": 6.69, "learning_rate": 1.2663813542577208e-06, "loss": 0.2185, "step": 51580 }, { "epoch": 6.7, "learning_rate": 1.2654881009018352e-06, "loss": 0.2215, "step": 51590 }, { "epoch": 6.7, "learning_rate": 1.264595046172561e-06, "loss": 0.2281, "step": 51600 }, { "epoch": 6.7, "learning_rate": 1.263702190234409e-06, "loss": 0.2419, "step": 51610 }, { "epoch": 6.7, "learning_rate": 1.2628095332518517e-06, "loss": 0.2184, "step": 51620 }, { "epoch": 6.7, "learning_rate": 1.261917075389324e-06, "loss": 0.2121, "step": 51630 }, { "epoch": 6.7, "learning_rate": 1.2610248168112283e-06, "loss": 0.2207, "step": 51640 }, { "epoch": 6.7, "learning_rate": 1.2601327576819258e-06, "loss": 0.2271, "step": 51650 }, { "epoch": 6.7, "learning_rate": 1.259240898165742e-06, "loss": 0.2186, "step": 51660 }, { "epoch": 6.71, "learning_rate": 1.2583492384269683e-06, "loss": 0.2231, "step": 51670 }, { "epoch": 6.71, "learning_rate": 1.2574577786298552e-06, "loss": 0.2236, "step": 51680 }, { "epoch": 6.71, "learning_rate": 1.2565665189386206e-06, "loss": 0.2296, "step": 51690 }, { "epoch": 6.71, "learning_rate": 1.2556754595174426e-06, "loss": 0.2401, "step": 51700 }, { "epoch": 6.71, "learning_rate": 1.254784600530462e-06, "loss": 0.2285, "step": 51710 }, { "epoch": 6.71, "learning_rate": 1.253893942141786e-06, "loss": 0.2336, "step": 51720 }, { "epoch": 6.71, "learning_rate": 1.2530034845154819e-06, "loss": 0.2193, "step": 51730 }, { "epoch": 6.72, "learning_rate": 1.2521132278155795e-06, "loss": 0.2154, "step": 51740 }, { "epoch": 6.72, "learning_rate": 1.2512231722060757e-06, "loss": 0.2246, "step": 51750 }, { "epoch": 6.72, "learning_rate": 1.250333317850926e-06, "loss": 0.2236, "step": 51760 }, { "epoch": 6.72, "learning_rate": 1.2494436649140496e-06, "loss": 0.2235, "step": 51770 }, { "epoch": 6.72, "learning_rate": 1.2485542135593315e-06, "loss": 0.2203, "step": 51780 }, { "epoch": 6.72, "learning_rate": 1.2476649639506164e-06, "loss": 0.2292, "step": 51790 }, { "epoch": 6.72, "learning_rate": 1.2467759162517123e-06, "loss": 0.2155, "step": 51800 }, { "epoch": 6.72, "learning_rate": 1.2458870706263924e-06, "loss": 0.2124, "step": 51810 }, { "epoch": 6.73, "learning_rate": 1.244998427238389e-06, "loss": 0.2364, "step": 51820 }, { "epoch": 6.73, "learning_rate": 1.2441099862514012e-06, "loss": 0.2274, "step": 51830 }, { "epoch": 6.73, "learning_rate": 1.2432217478290875e-06, "loss": 0.2075, "step": 51840 }, { "epoch": 6.73, "learning_rate": 1.2423337121350696e-06, "loss": 0.2182, "step": 51850 }, { "epoch": 6.73, "learning_rate": 1.241445879332934e-06, "loss": 0.2299, "step": 51860 }, { "epoch": 6.73, "learning_rate": 1.2405582495862278e-06, "loss": 0.2213, "step": 51870 }, { "epoch": 6.73, "learning_rate": 1.23967082305846e-06, "loss": 0.2295, "step": 51880 }, { "epoch": 6.73, "learning_rate": 1.2387835999131058e-06, "loss": 0.2188, "step": 51890 }, { "epoch": 6.74, "learning_rate": 1.2378965803135993e-06, "loss": 0.2188, "step": 51900 }, { "epoch": 6.74, "learning_rate": 1.2370097644233374e-06, "loss": 0.2094, "step": 51910 }, { "epoch": 6.74, "learning_rate": 1.2361231524056827e-06, "loss": 0.2332, "step": 51920 }, { "epoch": 6.74, "learning_rate": 1.2352367444239564e-06, "loss": 0.2215, "step": 51930 }, { "epoch": 6.74, "learning_rate": 1.2343505406414436e-06, "loss": 0.243, "step": 51940 }, { "epoch": 6.74, "learning_rate": 1.2334645412213928e-06, "loss": 0.2295, "step": 51950 }, { "epoch": 6.74, "learning_rate": 1.2325787463270128e-06, "loss": 0.2239, "step": 51960 }, { "epoch": 6.74, "learning_rate": 1.2316931561214774e-06, "loss": 0.2292, "step": 51970 }, { "epoch": 6.75, "learning_rate": 1.2308077707679204e-06, "loss": 0.234, "step": 51980 }, { "epoch": 6.75, "learning_rate": 1.2299225904294375e-06, "loss": 0.2402, "step": 51990 }, { "epoch": 6.75, "learning_rate": 1.2290376152690899e-06, "loss": 0.232, "step": 52000 }, { "epoch": 6.75, "learning_rate": 1.2281528454498978e-06, "loss": 0.2358, "step": 52010 }, { "epoch": 6.75, "learning_rate": 1.2272682811348433e-06, "loss": 0.2161, "step": 52020 }, { "epoch": 6.75, "learning_rate": 1.2263839224868747e-06, "loss": 0.2278, "step": 52030 }, { "epoch": 6.75, "learning_rate": 1.2254997696688984e-06, "loss": 0.2296, "step": 52040 }, { "epoch": 6.76, "learning_rate": 1.2246158228437826e-06, "loss": 0.2194, "step": 52050 }, { "epoch": 6.76, "learning_rate": 1.2237320821743619e-06, "loss": 0.2212, "step": 52060 }, { "epoch": 6.76, "learning_rate": 1.2228485478234289e-06, "loss": 0.2171, "step": 52070 }, { "epoch": 6.76, "learning_rate": 1.2219652199537382e-06, "loss": 0.2287, "step": 52080 }, { "epoch": 6.76, "learning_rate": 1.2210820987280101e-06, "loss": 0.2192, "step": 52090 }, { "epoch": 6.76, "learning_rate": 1.220199184308923e-06, "loss": 0.2319, "step": 52100 }, { "epoch": 6.76, "learning_rate": 1.2193164768591175e-06, "loss": 0.2261, "step": 52110 }, { "epoch": 6.76, "learning_rate": 1.2184339765411993e-06, "loss": 0.2227, "step": 52120 }, { "epoch": 6.77, "learning_rate": 1.2175516835177317e-06, "loss": 0.2333, "step": 52130 }, { "epoch": 6.77, "learning_rate": 1.2166695979512443e-06, "loss": 0.2184, "step": 52140 }, { "epoch": 6.77, "learning_rate": 1.2157877200042244e-06, "loss": 0.2205, "step": 52150 }, { "epoch": 6.77, "learning_rate": 1.2149060498391223e-06, "loss": 0.2292, "step": 52160 }, { "epoch": 6.77, "learning_rate": 1.2140245876183517e-06, "loss": 0.2215, "step": 52170 }, { "epoch": 6.77, "learning_rate": 1.2131433335042866e-06, "loss": 0.2306, "step": 52180 }, { "epoch": 6.77, "learning_rate": 1.2122622876592615e-06, "loss": 0.2344, "step": 52190 }, { "epoch": 6.77, "learning_rate": 1.2113814502455754e-06, "loss": 0.2301, "step": 52200 }, { "epoch": 6.78, "learning_rate": 1.210500821425487e-06, "loss": 0.2216, "step": 52210 }, { "epoch": 6.78, "learning_rate": 1.2096204013612157e-06, "loss": 0.2312, "step": 52220 }, { "epoch": 6.78, "learning_rate": 1.2087401902149455e-06, "loss": 0.227, "step": 52230 }, { "epoch": 6.78, "learning_rate": 1.207860188148819e-06, "loss": 0.2405, "step": 52240 }, { "epoch": 6.78, "learning_rate": 1.206980395324941e-06, "loss": 0.2235, "step": 52250 }, { "epoch": 6.78, "learning_rate": 1.2061008119053796e-06, "loss": 0.2183, "step": 52260 }, { "epoch": 6.78, "learning_rate": 1.2052214380521609e-06, "loss": 0.2359, "step": 52270 }, { "epoch": 6.79, "learning_rate": 1.2043422739272764e-06, "loss": 0.2348, "step": 52280 }, { "epoch": 6.79, "learning_rate": 1.2034633196926756e-06, "loss": 0.2222, "step": 52290 }, { "epoch": 6.79, "learning_rate": 1.2025845755102706e-06, "loss": 0.233, "step": 52300 }, { "epoch": 6.79, "learning_rate": 1.2017060415419358e-06, "loss": 0.2336, "step": 52310 }, { "epoch": 6.79, "learning_rate": 1.2008277179495056e-06, "loss": 0.2317, "step": 52320 }, { "epoch": 6.79, "learning_rate": 1.1999496048947746e-06, "loss": 0.2313, "step": 52330 }, { "epoch": 6.79, "learning_rate": 1.1990717025395021e-06, "loss": 0.227, "step": 52340 }, { "epoch": 6.79, "learning_rate": 1.1981940110454054e-06, "loss": 0.2412, "step": 52350 }, { "epoch": 6.8, "learning_rate": 1.1973165305741636e-06, "loss": 0.2124, "step": 52360 }, { "epoch": 6.8, "learning_rate": 1.1964392612874185e-06, "loss": 0.2259, "step": 52370 }, { "epoch": 6.8, "learning_rate": 1.1955622033467715e-06, "loss": 0.2196, "step": 52380 }, { "epoch": 6.8, "learning_rate": 1.1946853569137842e-06, "loss": 0.226, "step": 52390 }, { "epoch": 6.8, "learning_rate": 1.1938087221499822e-06, "loss": 0.2186, "step": 52400 }, { "epoch": 6.8, "learning_rate": 1.192932299216849e-06, "loss": 0.2191, "step": 52410 }, { "epoch": 6.8, "learning_rate": 1.1920560882758322e-06, "loss": 0.236, "step": 52420 }, { "epoch": 6.8, "learning_rate": 1.1911800894883374e-06, "loss": 0.233, "step": 52430 }, { "epoch": 6.81, "learning_rate": 1.1903043030157312e-06, "loss": 0.2123, "step": 52440 }, { "epoch": 6.81, "learning_rate": 1.1894287290193447e-06, "loss": 0.2215, "step": 52450 }, { "epoch": 6.81, "learning_rate": 1.1885533676604664e-06, "loss": 0.2396, "step": 52460 }, { "epoch": 6.81, "learning_rate": 1.187678219100345e-06, "loss": 0.2291, "step": 52470 }, { "epoch": 6.81, "learning_rate": 1.186803283500194e-06, "loss": 0.233, "step": 52480 }, { "epoch": 6.81, "learning_rate": 1.1859285610211844e-06, "loss": 0.227, "step": 52490 }, { "epoch": 6.81, "learning_rate": 1.1850540518244476e-06, "loss": 0.231, "step": 52500 }, { "epoch": 6.82, "learning_rate": 1.1841797560710788e-06, "loss": 0.2238, "step": 52510 }, { "epoch": 6.82, "learning_rate": 1.1833056739221316e-06, "loss": 0.2462, "step": 52520 }, { "epoch": 6.82, "learning_rate": 1.1824318055386191e-06, "loss": 0.2382, "step": 52530 }, { "epoch": 6.82, "learning_rate": 1.1815581510815188e-06, "loss": 0.2295, "step": 52540 }, { "epoch": 6.82, "learning_rate": 1.1806847107117643e-06, "loss": 0.2194, "step": 52550 }, { "epoch": 6.82, "learning_rate": 1.1798114845902545e-06, "loss": 0.2324, "step": 52560 }, { "epoch": 6.82, "learning_rate": 1.1789384728778449e-06, "loss": 0.2263, "step": 52570 }, { "epoch": 6.82, "learning_rate": 1.1780656757353524e-06, "loss": 0.2046, "step": 52580 }, { "epoch": 6.83, "learning_rate": 1.1771930933235568e-06, "loss": 0.2174, "step": 52590 }, { "epoch": 6.83, "learning_rate": 1.1763207258031952e-06, "loss": 0.222, "step": 52600 }, { "epoch": 6.83, "learning_rate": 1.175448573334966e-06, "loss": 0.2342, "step": 52610 }, { "epoch": 6.83, "learning_rate": 1.17457663607953e-06, "loss": 0.2188, "step": 52620 }, { "epoch": 6.83, "learning_rate": 1.1737049141975058e-06, "loss": 0.2285, "step": 52630 }, { "epoch": 6.83, "learning_rate": 1.1728334078494724e-06, "loss": 0.2268, "step": 52640 }, { "epoch": 6.83, "learning_rate": 1.1719621171959715e-06, "loss": 0.2302, "step": 52650 }, { "epoch": 6.83, "learning_rate": 1.171091042397503e-06, "loss": 0.2167, "step": 52660 }, { "epoch": 6.84, "learning_rate": 1.1702201836145267e-06, "loss": 0.216, "step": 52670 }, { "epoch": 6.84, "learning_rate": 1.1693495410074651e-06, "loss": 0.2418, "step": 52680 }, { "epoch": 6.84, "learning_rate": 1.1684791147366978e-06, "loss": 0.2358, "step": 52690 }, { "epoch": 6.84, "learning_rate": 1.1676089049625673e-06, "loss": 0.2165, "step": 52700 }, { "epoch": 6.84, "learning_rate": 1.1667389118453745e-06, "loss": 0.2264, "step": 52710 }, { "epoch": 6.84, "learning_rate": 1.1658691355453796e-06, "loss": 0.2095, "step": 52720 }, { "epoch": 6.84, "learning_rate": 1.164999576222806e-06, "loss": 0.2287, "step": 52730 }, { "epoch": 6.84, "learning_rate": 1.1641302340378342e-06, "loss": 0.2339, "step": 52740 }, { "epoch": 6.85, "learning_rate": 1.1632611091506052e-06, "loss": 0.2488, "step": 52750 }, { "epoch": 6.85, "learning_rate": 1.1623922017212217e-06, "loss": 0.2439, "step": 52760 }, { "epoch": 6.85, "learning_rate": 1.1615235119097448e-06, "loss": 0.2218, "step": 52770 }, { "epoch": 6.85, "learning_rate": 1.1606550398761944e-06, "loss": 0.2164, "step": 52780 }, { "epoch": 6.85, "learning_rate": 1.1597867857805541e-06, "loss": 0.2162, "step": 52790 }, { "epoch": 6.85, "learning_rate": 1.1589187497827637e-06, "loss": 0.216, "step": 52800 }, { "epoch": 6.85, "learning_rate": 1.1580509320427231e-06, "loss": 0.2313, "step": 52810 }, { "epoch": 6.86, "learning_rate": 1.157183332720295e-06, "loss": 0.2315, "step": 52820 }, { "epoch": 6.86, "learning_rate": 1.1563159519752983e-06, "loss": 0.2298, "step": 52830 }, { "epoch": 6.86, "learning_rate": 1.1554487899675145e-06, "loss": 0.2231, "step": 52840 }, { "epoch": 6.86, "learning_rate": 1.1545818468566827e-06, "loss": 0.2207, "step": 52850 }, { "epoch": 6.86, "learning_rate": 1.153715122802502e-06, "loss": 0.2224, "step": 52860 }, { "epoch": 6.86, "learning_rate": 1.1528486179646332e-06, "loss": 0.2295, "step": 52870 }, { "epoch": 6.86, "learning_rate": 1.151982332502694e-06, "loss": 0.229, "step": 52880 }, { "epoch": 6.86, "learning_rate": 1.151116266576262e-06, "loss": 0.2158, "step": 52890 }, { "epoch": 6.87, "learning_rate": 1.1502504203448775e-06, "loss": 0.2249, "step": 52900 }, { "epoch": 6.87, "learning_rate": 1.1493847939680364e-06, "loss": 0.2167, "step": 52910 }, { "epoch": 6.87, "learning_rate": 1.1485193876051953e-06, "loss": 0.235, "step": 52920 }, { "epoch": 6.87, "learning_rate": 1.1476542014157723e-06, "loss": 0.2373, "step": 52930 }, { "epoch": 6.87, "learning_rate": 1.1467892355591425e-06, "loss": 0.2336, "step": 52940 }, { "epoch": 6.87, "learning_rate": 1.1459244901946404e-06, "loss": 0.2154, "step": 52950 }, { "epoch": 6.87, "learning_rate": 1.1450599654815624e-06, "loss": 0.2199, "step": 52960 }, { "epoch": 6.87, "learning_rate": 1.1441956615791618e-06, "loss": 0.2321, "step": 52970 }, { "epoch": 6.88, "learning_rate": 1.1433315786466511e-06, "loss": 0.2201, "step": 52980 }, { "epoch": 6.88, "learning_rate": 1.1424677168432048e-06, "loss": 0.2308, "step": 52990 }, { "epoch": 6.88, "learning_rate": 1.1416040763279531e-06, "loss": 0.2158, "step": 53000 }, { "epoch": 6.88, "learning_rate": 1.1407406572599893e-06, "loss": 0.2283, "step": 53010 }, { "epoch": 6.88, "learning_rate": 1.1398774597983622e-06, "loss": 0.2322, "step": 53020 }, { "epoch": 6.88, "learning_rate": 1.1390144841020806e-06, "loss": 0.2278, "step": 53030 }, { "epoch": 6.88, "learning_rate": 1.1381517303301158e-06, "loss": 0.2304, "step": 53040 }, { "epoch": 6.89, "learning_rate": 1.1372891986413942e-06, "loss": 0.2249, "step": 53050 }, { "epoch": 6.89, "learning_rate": 1.1364268891948018e-06, "loss": 0.2264, "step": 53060 }, { "epoch": 6.89, "learning_rate": 1.1355648021491868e-06, "loss": 0.219, "step": 53070 }, { "epoch": 6.89, "learning_rate": 1.1347029376633529e-06, "loss": 0.2151, "step": 53080 }, { "epoch": 6.89, "learning_rate": 1.1338412958960635e-06, "loss": 0.2229, "step": 53090 }, { "epoch": 6.89, "learning_rate": 1.1329798770060433e-06, "loss": 0.2264, "step": 53100 }, { "epoch": 6.89, "learning_rate": 1.1321186811519734e-06, "loss": 0.2327, "step": 53110 }, { "epoch": 6.89, "learning_rate": 1.131257708492494e-06, "loss": 0.2277, "step": 53120 }, { "epoch": 6.9, "learning_rate": 1.1303969591862063e-06, "loss": 0.2272, "step": 53130 }, { "epoch": 6.9, "learning_rate": 1.1295364333916673e-06, "loss": 0.2289, "step": 53140 }, { "epoch": 6.9, "learning_rate": 1.1286761312673962e-06, "loss": 0.2226, "step": 53150 }, { "epoch": 6.9, "learning_rate": 1.1278160529718687e-06, "loss": 0.225, "step": 53160 }, { "epoch": 6.9, "learning_rate": 1.1269561986635185e-06, "loss": 0.2136, "step": 53170 }, { "epoch": 6.9, "learning_rate": 1.126096568500741e-06, "loss": 0.2028, "step": 53180 }, { "epoch": 6.9, "learning_rate": 1.1252371626418883e-06, "loss": 0.2322, "step": 53190 }, { "epoch": 6.9, "learning_rate": 1.12437798124527e-06, "loss": 0.2504, "step": 53200 }, { "epoch": 6.91, "learning_rate": 1.1235190244691583e-06, "loss": 0.2133, "step": 53210 }, { "epoch": 6.91, "learning_rate": 1.1226602924717804e-06, "loss": 0.2308, "step": 53220 }, { "epoch": 6.91, "learning_rate": 1.1218017854113226e-06, "loss": 0.2236, "step": 53230 }, { "epoch": 6.91, "learning_rate": 1.1209435034459318e-06, "loss": 0.2305, "step": 53240 }, { "epoch": 6.91, "learning_rate": 1.1200854467337118e-06, "loss": 0.2322, "step": 53250 }, { "epoch": 6.91, "learning_rate": 1.119227615432724e-06, "loss": 0.2268, "step": 53260 }, { "epoch": 6.91, "learning_rate": 1.1183700097009919e-06, "loss": 0.2232, "step": 53270 }, { "epoch": 6.91, "learning_rate": 1.1175126296964922e-06, "loss": 0.2425, "step": 53280 }, { "epoch": 6.92, "learning_rate": 1.116655475577165e-06, "loss": 0.2181, "step": 53290 }, { "epoch": 6.92, "learning_rate": 1.1157985475009062e-06, "loss": 0.2111, "step": 53300 }, { "epoch": 6.92, "learning_rate": 1.1149418456255694e-06, "loss": 0.2375, "step": 53310 }, { "epoch": 6.92, "learning_rate": 1.1140853701089694e-06, "loss": 0.2269, "step": 53320 }, { "epoch": 6.92, "learning_rate": 1.1132291211088766e-06, "loss": 0.2299, "step": 53330 }, { "epoch": 6.92, "learning_rate": 1.1123730987830197e-06, "loss": 0.2275, "step": 53340 }, { "epoch": 6.92, "learning_rate": 1.1115173032890882e-06, "loss": 0.2093, "step": 53350 }, { "epoch": 6.93, "learning_rate": 1.1106617347847279e-06, "loss": 0.2323, "step": 53360 }, { "epoch": 6.93, "learning_rate": 1.1098063934275413e-06, "loss": 0.236, "step": 53370 }, { "epoch": 6.93, "learning_rate": 1.1089512793750934e-06, "loss": 0.2336, "step": 53380 }, { "epoch": 6.93, "learning_rate": 1.1080963927849034e-06, "loss": 0.2365, "step": 53390 }, { "epoch": 6.93, "learning_rate": 1.1072417338144493e-06, "loss": 0.2348, "step": 53400 }, { "epoch": 6.93, "learning_rate": 1.1063873026211693e-06, "loss": 0.209, "step": 53410 }, { "epoch": 6.93, "learning_rate": 1.1055330993624566e-06, "loss": 0.2227, "step": 53420 }, { "epoch": 6.93, "learning_rate": 1.1046791241956657e-06, "loss": 0.2191, "step": 53430 }, { "epoch": 6.94, "learning_rate": 1.1038253772781065e-06, "loss": 0.2363, "step": 53440 }, { "epoch": 6.94, "learning_rate": 1.1029718587670468e-06, "loss": 0.2343, "step": 53450 }, { "epoch": 6.94, "learning_rate": 1.1021185688197153e-06, "loss": 0.2276, "step": 53460 }, { "epoch": 6.94, "learning_rate": 1.101265507593295e-06, "loss": 0.2402, "step": 53470 }, { "epoch": 6.94, "learning_rate": 1.1004126752449278e-06, "loss": 0.2238, "step": 53480 }, { "epoch": 6.94, "learning_rate": 1.0995600719317158e-06, "loss": 0.2209, "step": 53490 }, { "epoch": 6.94, "learning_rate": 1.0987076978107158e-06, "loss": 0.2414, "step": 53500 }, { "epoch": 6.94, "learning_rate": 1.0978555530389433e-06, "loss": 0.237, "step": 53510 }, { "epoch": 6.95, "learning_rate": 1.097003637773373e-06, "loss": 0.2225, "step": 53520 }, { "epoch": 6.95, "learning_rate": 1.0961519521709359e-06, "loss": 0.2399, "step": 53530 }, { "epoch": 6.95, "learning_rate": 1.0953004963885196e-06, "loss": 0.2214, "step": 53540 }, { "epoch": 6.95, "learning_rate": 1.0944492705829732e-06, "loss": 0.218, "step": 53550 }, { "epoch": 6.95, "learning_rate": 1.0935982749110986e-06, "loss": 0.2256, "step": 53560 }, { "epoch": 6.95, "learning_rate": 1.0927475095296594e-06, "loss": 0.2202, "step": 53570 }, { "epoch": 6.95, "learning_rate": 1.091896974595375e-06, "loss": 0.2207, "step": 53580 }, { "epoch": 6.96, "learning_rate": 1.0910466702649205e-06, "loss": 0.2269, "step": 53590 }, { "epoch": 6.96, "learning_rate": 1.090196596694933e-06, "loss": 0.2339, "step": 53600 }, { "epoch": 6.96, "learning_rate": 1.0893467540420034e-06, "loss": 0.2271, "step": 53610 }, { "epoch": 6.96, "learning_rate": 1.0884971424626804e-06, "loss": 0.2402, "step": 53620 }, { "epoch": 6.96, "learning_rate": 1.0876477621134726e-06, "loss": 0.2168, "step": 53630 }, { "epoch": 6.96, "learning_rate": 1.0867986131508433e-06, "loss": 0.2328, "step": 53640 }, { "epoch": 6.96, "learning_rate": 1.0859496957312137e-06, "loss": 0.231, "step": 53650 }, { "epoch": 6.96, "learning_rate": 1.0851010100109645e-06, "loss": 0.2225, "step": 53660 }, { "epoch": 6.97, "learning_rate": 1.0842525561464308e-06, "loss": 0.2126, "step": 53670 }, { "epoch": 6.97, "learning_rate": 1.0834043342939064e-06, "loss": 0.2402, "step": 53680 }, { "epoch": 6.97, "learning_rate": 1.082556344609643e-06, "loss": 0.246, "step": 53690 }, { "epoch": 6.97, "learning_rate": 1.081708587249847e-06, "loss": 0.2244, "step": 53700 }, { "epoch": 6.97, "learning_rate": 1.0808610623706865e-06, "loss": 0.2493, "step": 53710 }, { "epoch": 6.97, "learning_rate": 1.0800137701282823e-06, "loss": 0.2415, "step": 53720 }, { "epoch": 6.97, "learning_rate": 1.0791667106787134e-06, "loss": 0.2252, "step": 53730 }, { "epoch": 6.97, "learning_rate": 1.0783198841780188e-06, "loss": 0.2143, "step": 53740 }, { "epoch": 6.98, "learning_rate": 1.0774732907821908e-06, "loss": 0.2236, "step": 53750 }, { "epoch": 6.98, "learning_rate": 1.07662693064718e-06, "loss": 0.2308, "step": 53760 }, { "epoch": 6.98, "learning_rate": 1.075780803928896e-06, "loss": 0.2249, "step": 53770 }, { "epoch": 6.98, "learning_rate": 1.0749349107832032e-06, "loss": 0.2237, "step": 53780 }, { "epoch": 6.98, "learning_rate": 1.074089251365922e-06, "loss": 0.2393, "step": 53790 }, { "epoch": 6.98, "learning_rate": 1.0732438258328335e-06, "loss": 0.2328, "step": 53800 }, { "epoch": 6.98, "learning_rate": 1.072398634339673e-06, "loss": 0.2069, "step": 53810 }, { "epoch": 6.99, "learning_rate": 1.0715536770421315e-06, "loss": 0.2454, "step": 53820 }, { "epoch": 6.99, "learning_rate": 1.070708954095861e-06, "loss": 0.2323, "step": 53830 }, { "epoch": 6.99, "learning_rate": 1.0698644656564659e-06, "loss": 0.221, "step": 53840 }, { "epoch": 6.99, "learning_rate": 1.0690202118795114e-06, "loss": 0.2365, "step": 53850 }, { "epoch": 6.99, "learning_rate": 1.068176192920516e-06, "loss": 0.2245, "step": 53860 }, { "epoch": 6.99, "learning_rate": 1.0673324089349564e-06, "loss": 0.2301, "step": 53870 }, { "epoch": 6.99, "learning_rate": 1.0664888600782672e-06, "loss": 0.2212, "step": 53880 }, { "epoch": 6.99, "learning_rate": 1.0656455465058378e-06, "loss": 0.2203, "step": 53890 }, { "epoch": 7.0, "learning_rate": 1.0648024683730138e-06, "loss": 0.2187, "step": 53900 }, { "epoch": 7.0, "learning_rate": 1.063959625835101e-06, "loss": 0.2274, "step": 53910 }, { "epoch": 7.0, "learning_rate": 1.063117019047358e-06, "loss": 0.221, "step": 53920 }, { "epoch": 7.0, "learning_rate": 1.0622746481650009e-06, "loss": 0.2304, "step": 53930 }, { "epoch": 7.0, "eval_loss": 0.22682592272758484, "eval_runtime": 59.946, "eval_samples_per_second": 181.447, "eval_steps_per_second": 22.687, "step": 53935 }, { "epoch": 7.0, "learning_rate": 1.0614325133432042e-06, "loss": 0.2158, "step": 53940 }, { "epoch": 7.0, "learning_rate": 1.060590614737097e-06, "loss": 0.2442, "step": 53950 }, { "epoch": 7.0, "learning_rate": 1.0597489525017644e-06, "loss": 0.2375, "step": 53960 }, { "epoch": 7.0, "learning_rate": 1.0589075267922508e-06, "loss": 0.2318, "step": 53970 }, { "epoch": 7.01, "learning_rate": 1.058066337763554e-06, "loss": 0.2296, "step": 53980 }, { "epoch": 7.01, "learning_rate": 1.0572253855706292e-06, "loss": 0.2201, "step": 53990 }, { "epoch": 7.01, "learning_rate": 1.0563846703683893e-06, "loss": 0.221, "step": 54000 }, { "epoch": 7.01, "learning_rate": 1.0555441923117015e-06, "loss": 0.2361, "step": 54010 }, { "epoch": 7.01, "learning_rate": 1.054703951555391e-06, "loss": 0.2228, "step": 54020 }, { "epoch": 7.01, "learning_rate": 1.053863948254238e-06, "loss": 0.2192, "step": 54030 }, { "epoch": 7.01, "learning_rate": 1.0530241825629794e-06, "loss": 0.2454, "step": 54040 }, { "epoch": 7.01, "learning_rate": 1.052184654636309e-06, "loss": 0.2302, "step": 54050 }, { "epoch": 7.02, "learning_rate": 1.0513453646288762e-06, "loss": 0.2309, "step": 54060 }, { "epoch": 7.02, "learning_rate": 1.0505063126952855e-06, "loss": 0.245, "step": 54070 }, { "epoch": 7.02, "learning_rate": 1.0496674989901003e-06, "loss": 0.2278, "step": 54080 }, { "epoch": 7.02, "learning_rate": 1.0488289236678372e-06, "loss": 0.2339, "step": 54090 }, { "epoch": 7.02, "learning_rate": 1.0479905868829701e-06, "loss": 0.2506, "step": 54100 }, { "epoch": 7.02, "learning_rate": 1.0471524887899301e-06, "loss": 0.2207, "step": 54110 }, { "epoch": 7.02, "learning_rate": 1.046314629543103e-06, "loss": 0.209, "step": 54120 }, { "epoch": 7.03, "learning_rate": 1.0454770092968293e-06, "loss": 0.2163, "step": 54130 }, { "epoch": 7.03, "learning_rate": 1.0446396282054092e-06, "loss": 0.2126, "step": 54140 }, { "epoch": 7.03, "learning_rate": 1.0438024864230946e-06, "loss": 0.216, "step": 54150 }, { "epoch": 7.03, "learning_rate": 1.0429655841040975e-06, "loss": 0.2179, "step": 54160 }, { "epoch": 7.03, "learning_rate": 1.0421289214025824e-06, "loss": 0.2035, "step": 54170 }, { "epoch": 7.03, "learning_rate": 1.0412924984726704e-06, "loss": 0.2306, "step": 54180 }, { "epoch": 7.03, "learning_rate": 1.0404563154684404e-06, "loss": 0.2343, "step": 54190 }, { "epoch": 7.03, "learning_rate": 1.039620372543925e-06, "loss": 0.2303, "step": 54200 }, { "epoch": 7.04, "learning_rate": 1.0387846698531122e-06, "loss": 0.2238, "step": 54210 }, { "epoch": 7.04, "learning_rate": 1.0379492075499486e-06, "loss": 0.227, "step": 54220 }, { "epoch": 7.04, "learning_rate": 1.0371139857883342e-06, "loss": 0.2339, "step": 54230 }, { "epoch": 7.04, "learning_rate": 1.0362790047221235e-06, "loss": 0.2413, "step": 54240 }, { "epoch": 7.04, "learning_rate": 1.0354442645051308e-06, "loss": 0.211, "step": 54250 }, { "epoch": 7.04, "learning_rate": 1.0346097652911224e-06, "loss": 0.2286, "step": 54260 }, { "epoch": 7.04, "learning_rate": 1.0337755072338206e-06, "loss": 0.2247, "step": 54270 }, { "epoch": 7.04, "learning_rate": 1.032941490486906e-06, "loss": 0.2168, "step": 54280 }, { "epoch": 7.05, "learning_rate": 1.0321077152040108e-06, "loss": 0.206, "step": 54290 }, { "epoch": 7.05, "learning_rate": 1.0312741815387263e-06, "loss": 0.2348, "step": 54300 }, { "epoch": 7.05, "learning_rate": 1.0304408896445975e-06, "loss": 0.2262, "step": 54310 }, { "epoch": 7.05, "learning_rate": 1.0296078396751236e-06, "loss": 0.2175, "step": 54320 }, { "epoch": 7.05, "learning_rate": 1.0287750317837627e-06, "loss": 0.2211, "step": 54330 }, { "epoch": 7.05, "learning_rate": 1.0279424661239256e-06, "loss": 0.2284, "step": 54340 }, { "epoch": 7.05, "learning_rate": 1.027110142848978e-06, "loss": 0.2254, "step": 54350 }, { "epoch": 7.06, "learning_rate": 1.0262780621122442e-06, "loss": 0.2337, "step": 54360 }, { "epoch": 7.06, "learning_rate": 1.0254462240670009e-06, "loss": 0.2352, "step": 54370 }, { "epoch": 7.06, "learning_rate": 1.0246146288664798e-06, "loss": 0.2223, "step": 54380 }, { "epoch": 7.06, "learning_rate": 1.0237832766638711e-06, "loss": 0.2306, "step": 54390 }, { "epoch": 7.06, "learning_rate": 1.0229521676123174e-06, "loss": 0.2224, "step": 54400 }, { "epoch": 7.06, "learning_rate": 1.022121301864916e-06, "loss": 0.2252, "step": 54410 }, { "epoch": 7.06, "learning_rate": 1.021290679574723e-06, "loss": 0.2076, "step": 54420 }, { "epoch": 7.06, "learning_rate": 1.0204603008947448e-06, "loss": 0.2253, "step": 54430 }, { "epoch": 7.07, "learning_rate": 1.0196301659779478e-06, "loss": 0.2284, "step": 54440 }, { "epoch": 7.07, "learning_rate": 1.0188002749772501e-06, "loss": 0.2227, "step": 54450 }, { "epoch": 7.07, "learning_rate": 1.017970628045525e-06, "loss": 0.2283, "step": 54460 }, { "epoch": 7.07, "learning_rate": 1.0171412253356037e-06, "loss": 0.2138, "step": 54470 }, { "epoch": 7.07, "learning_rate": 1.0163120670002689e-06, "loss": 0.2168, "step": 54480 }, { "epoch": 7.07, "learning_rate": 1.01548315319226e-06, "loss": 0.2347, "step": 54490 }, { "epoch": 7.07, "learning_rate": 1.0146544840642723e-06, "loss": 0.2276, "step": 54500 }, { "epoch": 7.07, "learning_rate": 1.013826059768954e-06, "loss": 0.2312, "step": 54510 }, { "epoch": 7.08, "learning_rate": 1.0129978804589085e-06, "loss": 0.218, "step": 54520 }, { "epoch": 7.08, "learning_rate": 1.0121699462866966e-06, "loss": 0.2227, "step": 54530 }, { "epoch": 7.08, "learning_rate": 1.011342257404831e-06, "loss": 0.2346, "step": 54540 }, { "epoch": 7.08, "learning_rate": 1.010514813965779e-06, "loss": 0.2168, "step": 54550 }, { "epoch": 7.08, "learning_rate": 1.0096876161219662e-06, "loss": 0.2321, "step": 54560 }, { "epoch": 7.08, "learning_rate": 1.008860664025769e-06, "loss": 0.2325, "step": 54570 }, { "epoch": 7.08, "learning_rate": 1.0080339578295218e-06, "loss": 0.22, "step": 54580 }, { "epoch": 7.09, "learning_rate": 1.007207497685511e-06, "loss": 0.2271, "step": 54590 }, { "epoch": 7.09, "learning_rate": 1.0063812837459783e-06, "loss": 0.2192, "step": 54600 }, { "epoch": 7.09, "learning_rate": 1.005555316163122e-06, "loss": 0.221, "step": 54610 }, { "epoch": 7.09, "learning_rate": 1.0047295950890932e-06, "loss": 0.2365, "step": 54620 }, { "epoch": 7.09, "learning_rate": 1.0039041206759968e-06, "loss": 0.2053, "step": 54630 }, { "epoch": 7.09, "learning_rate": 1.0030788930758954e-06, "loss": 0.2326, "step": 54640 }, { "epoch": 7.09, "learning_rate": 1.0022539124408025e-06, "loss": 0.2338, "step": 54650 }, { "epoch": 7.09, "learning_rate": 1.0014291789226878e-06, "loss": 0.2198, "step": 54660 }, { "epoch": 7.1, "learning_rate": 1.0006046926734768e-06, "loss": 0.221, "step": 54670 }, { "epoch": 7.1, "learning_rate": 9.997804538450474e-07, "loss": 0.2265, "step": 54680 }, { "epoch": 7.1, "learning_rate": 9.989564625892316e-07, "loss": 0.2258, "step": 54690 }, { "epoch": 7.1, "learning_rate": 9.981327190578185e-07, "loss": 0.2298, "step": 54700 }, { "epoch": 7.1, "learning_rate": 9.973092234025485e-07, "loss": 0.2205, "step": 54710 }, { "epoch": 7.1, "learning_rate": 9.96485975775119e-07, "loss": 0.213, "step": 54720 }, { "epoch": 7.1, "learning_rate": 9.956629763271798e-07, "loss": 0.2296, "step": 54730 }, { "epoch": 7.1, "learning_rate": 9.948402252103351e-07, "loss": 0.2323, "step": 54740 }, { "epoch": 7.11, "learning_rate": 9.94017722576145e-07, "loss": 0.2174, "step": 54750 }, { "epoch": 7.11, "learning_rate": 9.93195468576122e-07, "loss": 0.2192, "step": 54760 }, { "epoch": 7.11, "learning_rate": 9.923734633617326e-07, "loss": 0.2248, "step": 54770 }, { "epoch": 7.11, "learning_rate": 9.915517070844005e-07, "loss": 0.2384, "step": 54780 }, { "epoch": 7.11, "learning_rate": 9.907301998955004e-07, "loss": 0.2285, "step": 54790 }, { "epoch": 7.11, "learning_rate": 9.899089419463609e-07, "loss": 0.2215, "step": 54800 }, { "epoch": 7.11, "learning_rate": 9.890879333882678e-07, "loss": 0.2433, "step": 54810 }, { "epoch": 7.11, "learning_rate": 9.882671743724585e-07, "loss": 0.2103, "step": 54820 }, { "epoch": 7.12, "learning_rate": 9.874466650501239e-07, "loss": 0.2219, "step": 54830 }, { "epoch": 7.12, "learning_rate": 9.866264055724117e-07, "loss": 0.2331, "step": 54840 }, { "epoch": 7.12, "learning_rate": 9.858063960904213e-07, "loss": 0.2303, "step": 54850 }, { "epoch": 7.12, "learning_rate": 9.849866367552054e-07, "loss": 0.2319, "step": 54860 }, { "epoch": 7.12, "learning_rate": 9.841671277177737e-07, "loss": 0.2294, "step": 54870 }, { "epoch": 7.12, "learning_rate": 9.833478691290862e-07, "loss": 0.2239, "step": 54880 }, { "epoch": 7.12, "learning_rate": 9.825288611400606e-07, "loss": 0.2309, "step": 54890 }, { "epoch": 7.13, "learning_rate": 9.817101039015651e-07, "loss": 0.2445, "step": 54900 }, { "epoch": 7.13, "learning_rate": 9.808915975644226e-07, "loss": 0.224, "step": 54910 }, { "epoch": 7.13, "learning_rate": 9.800733422794108e-07, "loss": 0.2308, "step": 54920 }, { "epoch": 7.13, "learning_rate": 9.792553381972607e-07, "loss": 0.2379, "step": 54930 }, { "epoch": 7.13, "learning_rate": 9.784375854686558e-07, "loss": 0.229, "step": 54940 }, { "epoch": 7.13, "learning_rate": 9.776200842442356e-07, "loss": 0.238, "step": 54950 }, { "epoch": 7.13, "learning_rate": 9.76802834674591e-07, "loss": 0.2293, "step": 54960 }, { "epoch": 7.13, "learning_rate": 9.759858369102675e-07, "loss": 0.2241, "step": 54970 }, { "epoch": 7.14, "learning_rate": 9.751690911017655e-07, "loss": 0.2188, "step": 54980 }, { "epoch": 7.14, "learning_rate": 9.743525973995366e-07, "loss": 0.2235, "step": 54990 }, { "epoch": 7.14, "learning_rate": 9.735363559539867e-07, "loss": 0.213, "step": 55000 }, { "epoch": 7.14, "learning_rate": 9.727203669154772e-07, "loss": 0.2309, "step": 55010 }, { "epoch": 7.14, "learning_rate": 9.719046304343196e-07, "loss": 0.2214, "step": 55020 }, { "epoch": 7.14, "learning_rate": 9.710891466607824e-07, "loss": 0.2404, "step": 55030 }, { "epoch": 7.14, "learning_rate": 9.70273915745085e-07, "loss": 0.2182, "step": 55040 }, { "epoch": 7.14, "learning_rate": 9.694589378374005e-07, "loss": 0.2248, "step": 55050 }, { "epoch": 7.15, "learning_rate": 9.686442130878575e-07, "loss": 0.248, "step": 55060 }, { "epoch": 7.15, "learning_rate": 9.678297416465354e-07, "loss": 0.2209, "step": 55070 }, { "epoch": 7.15, "learning_rate": 9.670155236634677e-07, "loss": 0.2335, "step": 55080 }, { "epoch": 7.15, "learning_rate": 9.662015592886424e-07, "loss": 0.228, "step": 55090 }, { "epoch": 7.15, "learning_rate": 9.653878486719998e-07, "loss": 0.2134, "step": 55100 }, { "epoch": 7.15, "learning_rate": 9.645743919634318e-07, "loss": 0.2239, "step": 55110 }, { "epoch": 7.15, "learning_rate": 9.637611893127878e-07, "loss": 0.2216, "step": 55120 }, { "epoch": 7.16, "learning_rate": 9.629482408698662e-07, "loss": 0.2356, "step": 55130 }, { "epoch": 7.16, "learning_rate": 9.621355467844203e-07, "loss": 0.2332, "step": 55140 }, { "epoch": 7.16, "learning_rate": 9.613231072061573e-07, "loss": 0.2369, "step": 55150 }, { "epoch": 7.16, "learning_rate": 9.605109222847354e-07, "loss": 0.2354, "step": 55160 }, { "epoch": 7.16, "learning_rate": 9.596989921697692e-07, "loss": 0.237, "step": 55170 }, { "epoch": 7.16, "learning_rate": 9.588873170108226e-07, "loss": 0.2402, "step": 55180 }, { "epoch": 7.16, "learning_rate": 9.580758969574145e-07, "loss": 0.2161, "step": 55190 }, { "epoch": 7.16, "learning_rate": 9.572647321590172e-07, "loss": 0.2181, "step": 55200 }, { "epoch": 7.17, "learning_rate": 9.564538227650555e-07, "loss": 0.2293, "step": 55210 }, { "epoch": 7.17, "learning_rate": 9.556431689249052e-07, "loss": 0.2248, "step": 55220 }, { "epoch": 7.17, "learning_rate": 9.548327707878993e-07, "loss": 0.2267, "step": 55230 }, { "epoch": 7.17, "learning_rate": 9.540226285033201e-07, "loss": 0.2286, "step": 55240 }, { "epoch": 7.17, "learning_rate": 9.532127422204029e-07, "loss": 0.2215, "step": 55250 }, { "epoch": 7.17, "learning_rate": 9.524031120883388e-07, "loss": 0.2413, "step": 55260 }, { "epoch": 7.17, "learning_rate": 9.515937382562689e-07, "loss": 0.2277, "step": 55270 }, { "epoch": 7.17, "learning_rate": 9.507846208732868e-07, "loss": 0.216, "step": 55280 }, { "epoch": 7.18, "learning_rate": 9.49975760088442e-07, "loss": 0.2182, "step": 55290 }, { "epoch": 7.18, "learning_rate": 9.49167156050733e-07, "loss": 0.2257, "step": 55300 }, { "epoch": 7.18, "learning_rate": 9.483588089091145e-07, "loss": 0.2286, "step": 55310 }, { "epoch": 7.18, "learning_rate": 9.475507188124909e-07, "loss": 0.2346, "step": 55320 }, { "epoch": 7.18, "learning_rate": 9.467428859097197e-07, "loss": 0.2296, "step": 55330 }, { "epoch": 7.18, "learning_rate": 9.459353103496141e-07, "loss": 0.2283, "step": 55340 }, { "epoch": 7.18, "learning_rate": 9.45127992280936e-07, "loss": 0.2409, "step": 55350 }, { "epoch": 7.18, "learning_rate": 9.44320931852401e-07, "loss": 0.2451, "step": 55360 }, { "epoch": 7.19, "learning_rate": 9.435141292126792e-07, "loss": 0.2236, "step": 55370 }, { "epoch": 7.19, "learning_rate": 9.427075845103909e-07, "loss": 0.2205, "step": 55380 }, { "epoch": 7.19, "learning_rate": 9.41901297894109e-07, "loss": 0.234, "step": 55390 }, { "epoch": 7.19, "learning_rate": 9.410952695123611e-07, "loss": 0.2177, "step": 55400 }, { "epoch": 7.19, "learning_rate": 9.402894995136248e-07, "loss": 0.222, "step": 55410 }, { "epoch": 7.19, "learning_rate": 9.394839880463304e-07, "loss": 0.2255, "step": 55420 }, { "epoch": 7.19, "learning_rate": 9.386787352588624e-07, "loss": 0.2156, "step": 55430 }, { "epoch": 7.2, "learning_rate": 9.378737412995549e-07, "loss": 0.2183, "step": 55440 }, { "epoch": 7.2, "learning_rate": 9.370690063166979e-07, "loss": 0.2247, "step": 55450 }, { "epoch": 7.2, "learning_rate": 9.362645304585303e-07, "loss": 0.2287, "step": 55460 }, { "epoch": 7.2, "learning_rate": 9.354603138732437e-07, "loss": 0.2397, "step": 55470 }, { "epoch": 7.2, "learning_rate": 9.34656356708985e-07, "loss": 0.2194, "step": 55480 }, { "epoch": 7.2, "learning_rate": 9.338526591138498e-07, "loss": 0.212, "step": 55490 }, { "epoch": 7.2, "learning_rate": 9.330492212358865e-07, "loss": 0.2494, "step": 55500 }, { "epoch": 7.2, "learning_rate": 9.322460432230985e-07, "loss": 0.2176, "step": 55510 }, { "epoch": 7.21, "learning_rate": 9.314431252234376e-07, "loss": 0.2323, "step": 55520 }, { "epoch": 7.21, "learning_rate": 9.306404673848094e-07, "loss": 0.2264, "step": 55530 }, { "epoch": 7.21, "learning_rate": 9.298380698550721e-07, "loss": 0.2485, "step": 55540 }, { "epoch": 7.21, "learning_rate": 9.290359327820355e-07, "loss": 0.2239, "step": 55550 }, { "epoch": 7.21, "learning_rate": 9.282340563134599e-07, "loss": 0.2315, "step": 55560 }, { "epoch": 7.21, "learning_rate": 9.274324405970606e-07, "loss": 0.2189, "step": 55570 }, { "epoch": 7.21, "learning_rate": 9.26631085780502e-07, "loss": 0.227, "step": 55580 }, { "epoch": 7.21, "learning_rate": 9.258299920114025e-07, "loss": 0.2327, "step": 55590 }, { "epoch": 7.22, "learning_rate": 9.250291594373317e-07, "loss": 0.2363, "step": 55600 }, { "epoch": 7.22, "learning_rate": 9.242285882058095e-07, "loss": 0.2167, "step": 55610 }, { "epoch": 7.22, "learning_rate": 9.23428278464311e-07, "loss": 0.2294, "step": 55620 }, { "epoch": 7.22, "learning_rate": 9.226282303602601e-07, "loss": 0.22, "step": 55630 }, { "epoch": 7.22, "learning_rate": 9.218284440410335e-07, "loss": 0.2366, "step": 55640 }, { "epoch": 7.22, "learning_rate": 9.210289196539607e-07, "loss": 0.2312, "step": 55650 }, { "epoch": 7.22, "learning_rate": 9.202296573463219e-07, "loss": 0.2308, "step": 55660 }, { "epoch": 7.23, "learning_rate": 9.194306572653482e-07, "loss": 0.2261, "step": 55670 }, { "epoch": 7.23, "learning_rate": 9.186319195582247e-07, "loss": 0.2344, "step": 55680 }, { "epoch": 7.23, "learning_rate": 9.178334443720864e-07, "loss": 0.2238, "step": 55690 }, { "epoch": 7.23, "learning_rate": 9.170352318540194e-07, "loss": 0.2241, "step": 55700 }, { "epoch": 7.23, "learning_rate": 9.16237282151064e-07, "loss": 0.2152, "step": 55710 }, { "epoch": 7.23, "learning_rate": 9.1543959541021e-07, "loss": 0.2251, "step": 55720 }, { "epoch": 7.23, "learning_rate": 9.146421717783985e-07, "loss": 0.2405, "step": 55730 }, { "epoch": 7.23, "learning_rate": 9.138450114025239e-07, "loss": 0.2225, "step": 55740 }, { "epoch": 7.24, "learning_rate": 9.1304811442943e-07, "loss": 0.2228, "step": 55750 }, { "epoch": 7.24, "learning_rate": 9.122514810059148e-07, "loss": 0.2428, "step": 55760 }, { "epoch": 7.24, "learning_rate": 9.114551112787251e-07, "loss": 0.2426, "step": 55770 }, { "epoch": 7.24, "learning_rate": 9.106590053945595e-07, "loss": 0.2407, "step": 55780 }, { "epoch": 7.24, "learning_rate": 9.098631635000691e-07, "loss": 0.2337, "step": 55790 }, { "epoch": 7.24, "learning_rate": 9.090675857418576e-07, "loss": 0.2317, "step": 55800 }, { "epoch": 7.24, "learning_rate": 9.082722722664765e-07, "loss": 0.2209, "step": 55810 }, { "epoch": 7.24, "learning_rate": 9.074772232204304e-07, "loss": 0.2207, "step": 55820 }, { "epoch": 7.25, "learning_rate": 9.066824387501769e-07, "loss": 0.2447, "step": 55830 }, { "epoch": 7.25, "learning_rate": 9.058879190021212e-07, "loss": 0.2259, "step": 55840 }, { "epoch": 7.25, "learning_rate": 9.050936641226234e-07, "loss": 0.2248, "step": 55850 }, { "epoch": 7.25, "learning_rate": 9.042996742579928e-07, "loss": 0.228, "step": 55860 }, { "epoch": 7.25, "learning_rate": 9.035059495544891e-07, "loss": 0.2257, "step": 55870 }, { "epoch": 7.25, "learning_rate": 9.027124901583262e-07, "loss": 0.2321, "step": 55880 }, { "epoch": 7.25, "learning_rate": 9.019192962156663e-07, "loss": 0.2234, "step": 55890 }, { "epoch": 7.26, "learning_rate": 9.011263678726227e-07, "loss": 0.2309, "step": 55900 }, { "epoch": 7.26, "learning_rate": 9.003337052752625e-07, "loss": 0.2187, "step": 55910 }, { "epoch": 7.26, "learning_rate": 8.995413085696012e-07, "loss": 0.2336, "step": 55920 }, { "epoch": 7.26, "learning_rate": 8.987491779016055e-07, "loss": 0.247, "step": 55930 }, { "epoch": 7.26, "learning_rate": 8.979573134171955e-07, "loss": 0.2192, "step": 55940 }, { "epoch": 7.26, "learning_rate": 8.971657152622392e-07, "loss": 0.2167, "step": 55950 }, { "epoch": 7.26, "learning_rate": 8.963743835825565e-07, "loss": 0.2313, "step": 55960 }, { "epoch": 7.26, "learning_rate": 8.955833185239202e-07, "loss": 0.224, "step": 55970 }, { "epoch": 7.27, "learning_rate": 8.947925202320506e-07, "loss": 0.2214, "step": 55980 }, { "epoch": 7.27, "learning_rate": 8.940019888526227e-07, "loss": 0.2139, "step": 55990 }, { "epoch": 7.27, "learning_rate": 8.932117245312589e-07, "loss": 0.2052, "step": 56000 }, { "epoch": 7.27, "learning_rate": 8.92421727413533e-07, "loss": 0.2287, "step": 56010 }, { "epoch": 7.27, "learning_rate": 8.916319976449726e-07, "loss": 0.2199, "step": 56020 }, { "epoch": 7.27, "learning_rate": 8.908425353710524e-07, "loss": 0.2318, "step": 56030 }, { "epoch": 7.27, "learning_rate": 8.900533407371986e-07, "loss": 0.2233, "step": 56040 }, { "epoch": 7.27, "learning_rate": 8.892644138887904e-07, "loss": 0.2292, "step": 56050 }, { "epoch": 7.28, "learning_rate": 8.884757549711551e-07, "loss": 0.2195, "step": 56060 }, { "epoch": 7.28, "learning_rate": 8.876873641295707e-07, "loss": 0.2249, "step": 56070 }, { "epoch": 7.28, "learning_rate": 8.868992415092687e-07, "loss": 0.2286, "step": 56080 }, { "epoch": 7.28, "learning_rate": 8.86111387255428e-07, "loss": 0.2349, "step": 56090 }, { "epoch": 7.28, "learning_rate": 8.853238015131782e-07, "loss": 0.2249, "step": 56100 }, { "epoch": 7.28, "learning_rate": 8.845364844276026e-07, "loss": 0.2292, "step": 56110 }, { "epoch": 7.28, "learning_rate": 8.83749436143731e-07, "loss": 0.2276, "step": 56120 }, { "epoch": 7.28, "learning_rate": 8.829626568065468e-07, "loss": 0.222, "step": 56130 }, { "epoch": 7.29, "learning_rate": 8.821761465609825e-07, "loss": 0.2233, "step": 56140 }, { "epoch": 7.29, "learning_rate": 8.813899055519201e-07, "loss": 0.2261, "step": 56150 }, { "epoch": 7.29, "learning_rate": 8.806039339241943e-07, "loss": 0.2332, "step": 56160 }, { "epoch": 7.29, "learning_rate": 8.798182318225883e-07, "loss": 0.2209, "step": 56170 }, { "epoch": 7.29, "learning_rate": 8.790327993918359e-07, "loss": 0.239, "step": 56180 }, { "epoch": 7.29, "learning_rate": 8.782476367766226e-07, "loss": 0.2183, "step": 56190 }, { "epoch": 7.29, "learning_rate": 8.774627441215829e-07, "loss": 0.2274, "step": 56200 }, { "epoch": 7.3, "learning_rate": 8.766781215713005e-07, "loss": 0.2297, "step": 56210 }, { "epoch": 7.3, "learning_rate": 8.758937692703128e-07, "loss": 0.2216, "step": 56220 }, { "epoch": 7.3, "learning_rate": 8.751096873631044e-07, "loss": 0.2212, "step": 56230 }, { "epoch": 7.3, "learning_rate": 8.7432587599411e-07, "loss": 0.2122, "step": 56240 }, { "epoch": 7.3, "learning_rate": 8.735423353077175e-07, "loss": 0.2201, "step": 56250 }, { "epoch": 7.3, "learning_rate": 8.72759065448261e-07, "loss": 0.2404, "step": 56260 }, { "epoch": 7.3, "learning_rate": 8.719760665600282e-07, "loss": 0.2292, "step": 56270 }, { "epoch": 7.3, "learning_rate": 8.711933387872545e-07, "loss": 0.2242, "step": 56280 }, { "epoch": 7.31, "learning_rate": 8.70410882274126e-07, "loss": 0.2207, "step": 56290 }, { "epoch": 7.31, "learning_rate": 8.696286971647798e-07, "loss": 0.2296, "step": 56300 }, { "epoch": 7.31, "learning_rate": 8.688467836033017e-07, "loss": 0.2239, "step": 56310 }, { "epoch": 7.31, "learning_rate": 8.680651417337272e-07, "loss": 0.2291, "step": 56320 }, { "epoch": 7.31, "learning_rate": 8.672837717000441e-07, "loss": 0.2244, "step": 56330 }, { "epoch": 7.31, "learning_rate": 8.665026736461878e-07, "loss": 0.2152, "step": 56340 }, { "epoch": 7.31, "learning_rate": 8.657218477160437e-07, "loss": 0.2347, "step": 56350 }, { "epoch": 7.31, "learning_rate": 8.649412940534493e-07, "loss": 0.2141, "step": 56360 }, { "epoch": 7.32, "learning_rate": 8.641610128021895e-07, "loss": 0.2088, "step": 56370 }, { "epoch": 7.32, "learning_rate": 8.633810041059987e-07, "loss": 0.2193, "step": 56380 }, { "epoch": 7.32, "learning_rate": 8.626012681085646e-07, "loss": 0.2368, "step": 56390 }, { "epoch": 7.32, "learning_rate": 8.618218049535205e-07, "loss": 0.2258, "step": 56400 }, { "epoch": 7.32, "learning_rate": 8.610426147844528e-07, "loss": 0.2207, "step": 56410 }, { "epoch": 7.32, "learning_rate": 8.602636977448951e-07, "loss": 0.2126, "step": 56420 }, { "epoch": 7.32, "learning_rate": 8.594850539783316e-07, "loss": 0.2261, "step": 56430 }, { "epoch": 7.33, "learning_rate": 8.58706683628197e-07, "loss": 0.2248, "step": 56440 }, { "epoch": 7.33, "learning_rate": 8.579285868378746e-07, "loss": 0.2318, "step": 56450 }, { "epoch": 7.33, "learning_rate": 8.571507637506968e-07, "loss": 0.219, "step": 56460 }, { "epoch": 7.33, "learning_rate": 8.563732145099479e-07, "loss": 0.2424, "step": 56470 }, { "epoch": 7.33, "learning_rate": 8.555959392588593e-07, "loss": 0.2365, "step": 56480 }, { "epoch": 7.33, "learning_rate": 8.548189381406123e-07, "loss": 0.226, "step": 56490 }, { "epoch": 7.33, "learning_rate": 8.540422112983396e-07, "loss": 0.2297, "step": 56500 }, { "epoch": 7.33, "learning_rate": 8.532657588751218e-07, "loss": 0.2148, "step": 56510 }, { "epoch": 7.34, "learning_rate": 8.524895810139875e-07, "loss": 0.2209, "step": 56520 }, { "epoch": 7.34, "learning_rate": 8.517136778579187e-07, "loss": 0.2249, "step": 56530 }, { "epoch": 7.34, "learning_rate": 8.509380495498428e-07, "loss": 0.2219, "step": 56540 }, { "epoch": 7.34, "learning_rate": 8.501626962326398e-07, "loss": 0.2369, "step": 56550 }, { "epoch": 7.34, "learning_rate": 8.493876180491366e-07, "loss": 0.2285, "step": 56560 }, { "epoch": 7.34, "learning_rate": 8.486128151421097e-07, "loss": 0.2262, "step": 56570 }, { "epoch": 7.34, "learning_rate": 8.478382876542871e-07, "loss": 0.2267, "step": 56580 }, { "epoch": 7.34, "learning_rate": 8.470640357283437e-07, "loss": 0.2305, "step": 56590 }, { "epoch": 7.35, "learning_rate": 8.462900595069037e-07, "loss": 0.2238, "step": 56600 }, { "epoch": 7.35, "learning_rate": 8.455163591325428e-07, "loss": 0.245, "step": 56610 }, { "epoch": 7.35, "learning_rate": 8.447429347477836e-07, "loss": 0.2211, "step": 56620 }, { "epoch": 7.35, "learning_rate": 8.439697864950976e-07, "loss": 0.225, "step": 56630 }, { "epoch": 7.35, "learning_rate": 8.431969145169081e-07, "loss": 0.2325, "step": 56640 }, { "epoch": 7.35, "learning_rate": 8.424243189555851e-07, "loss": 0.2387, "step": 56650 }, { "epoch": 7.35, "learning_rate": 8.416519999534478e-07, "loss": 0.2316, "step": 56660 }, { "epoch": 7.35, "learning_rate": 8.408799576527663e-07, "loss": 0.2196, "step": 56670 }, { "epoch": 7.36, "learning_rate": 8.40108192195758e-07, "loss": 0.2335, "step": 56680 }, { "epoch": 7.36, "learning_rate": 8.393367037245888e-07, "loss": 0.2276, "step": 56690 }, { "epoch": 7.36, "learning_rate": 8.385654923813762e-07, "loss": 0.2244, "step": 56700 }, { "epoch": 7.36, "learning_rate": 8.377945583081839e-07, "loss": 0.2304, "step": 56710 }, { "epoch": 7.36, "learning_rate": 8.370239016470266e-07, "loss": 0.2274, "step": 56720 }, { "epoch": 7.36, "learning_rate": 8.362535225398663e-07, "loss": 0.2067, "step": 56730 }, { "epoch": 7.36, "learning_rate": 8.354834211286141e-07, "loss": 0.2355, "step": 56740 }, { "epoch": 7.37, "learning_rate": 8.347135975551316e-07, "loss": 0.2238, "step": 56750 }, { "epoch": 7.37, "learning_rate": 8.339440519612274e-07, "loss": 0.236, "step": 56760 }, { "epoch": 7.37, "learning_rate": 8.331747844886584e-07, "loss": 0.2242, "step": 56770 }, { "epoch": 7.37, "learning_rate": 8.324057952791333e-07, "loss": 0.2333, "step": 56780 }, { "epoch": 7.37, "learning_rate": 8.316370844743064e-07, "loss": 0.2407, "step": 56790 }, { "epoch": 7.37, "learning_rate": 8.308686522157818e-07, "loss": 0.2441, "step": 56800 }, { "epoch": 7.37, "learning_rate": 8.301004986451132e-07, "loss": 0.2306, "step": 56810 }, { "epoch": 7.37, "learning_rate": 8.293326239038017e-07, "loss": 0.2337, "step": 56820 }, { "epoch": 7.38, "learning_rate": 8.28565028133297e-07, "loss": 0.2303, "step": 56830 }, { "epoch": 7.38, "learning_rate": 8.277977114749989e-07, "loss": 0.2278, "step": 56840 }, { "epoch": 7.38, "learning_rate": 8.270306740702538e-07, "loss": 0.2338, "step": 56850 }, { "epoch": 7.38, "learning_rate": 8.262639160603587e-07, "loss": 0.2188, "step": 56860 }, { "epoch": 7.38, "learning_rate": 8.254974375865577e-07, "loss": 0.2312, "step": 56870 }, { "epoch": 7.38, "learning_rate": 8.247312387900428e-07, "loss": 0.2266, "step": 56880 }, { "epoch": 7.38, "learning_rate": 8.239653198119572e-07, "loss": 0.2207, "step": 56890 }, { "epoch": 7.38, "learning_rate": 8.231996807933898e-07, "loss": 0.2281, "step": 56900 }, { "epoch": 7.39, "learning_rate": 8.224343218753782e-07, "loss": 0.2225, "step": 56910 }, { "epoch": 7.39, "learning_rate": 8.216692431989109e-07, "loss": 0.2427, "step": 56920 }, { "epoch": 7.39, "learning_rate": 8.209044449049223e-07, "loss": 0.233, "step": 56930 }, { "epoch": 7.39, "learning_rate": 8.201399271342949e-07, "loss": 0.2241, "step": 56940 }, { "epoch": 7.39, "learning_rate": 8.193756900278621e-07, "loss": 0.2457, "step": 56950 }, { "epoch": 7.39, "learning_rate": 8.186117337264035e-07, "loss": 0.2301, "step": 56960 }, { "epoch": 7.39, "learning_rate": 8.178480583706462e-07, "loss": 0.2331, "step": 56970 }, { "epoch": 7.4, "learning_rate": 8.170846641012689e-07, "loss": 0.2171, "step": 56980 }, { "epoch": 7.4, "learning_rate": 8.163215510588945e-07, "loss": 0.2309, "step": 56990 }, { "epoch": 7.4, "learning_rate": 8.155587193840978e-07, "loss": 0.2387, "step": 57000 }, { "epoch": 7.4, "learning_rate": 8.14796169217399e-07, "loss": 0.2346, "step": 57010 }, { "epoch": 7.4, "learning_rate": 8.140339006992672e-07, "loss": 0.2324, "step": 57020 }, { "epoch": 7.4, "learning_rate": 8.13271913970121e-07, "loss": 0.2202, "step": 57030 }, { "epoch": 7.4, "learning_rate": 8.125102091703254e-07, "loss": 0.2142, "step": 57040 }, { "epoch": 7.4, "learning_rate": 8.117487864401929e-07, "loss": 0.2346, "step": 57050 }, { "epoch": 7.41, "learning_rate": 8.109876459199872e-07, "loss": 0.2044, "step": 57060 }, { "epoch": 7.41, "learning_rate": 8.102267877499171e-07, "loss": 0.2133, "step": 57070 }, { "epoch": 7.41, "learning_rate": 8.094662120701391e-07, "loss": 0.2303, "step": 57080 }, { "epoch": 7.41, "learning_rate": 8.087059190207606e-07, "loss": 0.2227, "step": 57090 }, { "epoch": 7.41, "learning_rate": 8.079459087418349e-07, "loss": 0.23, "step": 57100 }, { "epoch": 7.41, "learning_rate": 8.071861813733619e-07, "loss": 0.2135, "step": 57110 }, { "epoch": 7.41, "learning_rate": 8.064267370552932e-07, "loss": 0.2242, "step": 57120 }, { "epoch": 7.41, "learning_rate": 8.056675759275238e-07, "loss": 0.2318, "step": 57130 }, { "epoch": 7.42, "learning_rate": 8.049086981299011e-07, "loss": 0.2293, "step": 57140 }, { "epoch": 7.42, "learning_rate": 8.041501038022167e-07, "loss": 0.2277, "step": 57150 }, { "epoch": 7.42, "learning_rate": 8.033917930842102e-07, "loss": 0.2228, "step": 57160 }, { "epoch": 7.42, "learning_rate": 8.026337661155723e-07, "loss": 0.2259, "step": 57170 }, { "epoch": 7.42, "learning_rate": 8.018760230359381e-07, "loss": 0.2292, "step": 57180 }, { "epoch": 7.42, "learning_rate": 8.011185639848903e-07, "loss": 0.2268, "step": 57190 }, { "epoch": 7.42, "learning_rate": 8.003613891019621e-07, "loss": 0.2221, "step": 57200 }, { "epoch": 7.43, "learning_rate": 7.996044985266323e-07, "loss": 0.2151, "step": 57210 }, { "epoch": 7.43, "learning_rate": 7.988478923983267e-07, "loss": 0.2364, "step": 57220 }, { "epoch": 7.43, "learning_rate": 7.980915708564209e-07, "loss": 0.2253, "step": 57230 }, { "epoch": 7.43, "learning_rate": 7.973355340402367e-07, "loss": 0.2219, "step": 57240 }, { "epoch": 7.43, "learning_rate": 7.965797820890425e-07, "loss": 0.2354, "step": 57250 }, { "epoch": 7.43, "learning_rate": 7.958243151420571e-07, "loss": 0.2167, "step": 57260 }, { "epoch": 7.43, "learning_rate": 7.950691333384433e-07, "loss": 0.2278, "step": 57270 }, { "epoch": 7.43, "learning_rate": 7.943142368173147e-07, "loss": 0.2101, "step": 57280 }, { "epoch": 7.44, "learning_rate": 7.935596257177303e-07, "loss": 0.2216, "step": 57290 }, { "epoch": 7.44, "learning_rate": 7.928053001786958e-07, "loss": 0.2311, "step": 57300 }, { "epoch": 7.44, "learning_rate": 7.920512603391676e-07, "loss": 0.2133, "step": 57310 }, { "epoch": 7.44, "learning_rate": 7.912975063380459e-07, "loss": 0.2028, "step": 57320 }, { "epoch": 7.44, "learning_rate": 7.905440383141794e-07, "loss": 0.2179, "step": 57330 }, { "epoch": 7.44, "learning_rate": 7.897908564063659e-07, "loss": 0.2225, "step": 57340 }, { "epoch": 7.44, "learning_rate": 7.890379607533483e-07, "loss": 0.2151, "step": 57350 }, { "epoch": 7.44, "learning_rate": 7.882853514938164e-07, "loss": 0.2218, "step": 57360 }, { "epoch": 7.45, "learning_rate": 7.875330287664105e-07, "loss": 0.2398, "step": 57370 }, { "epoch": 7.45, "learning_rate": 7.867809927097146e-07, "loss": 0.2215, "step": 57380 }, { "epoch": 7.45, "learning_rate": 7.860292434622608e-07, "loss": 0.2223, "step": 57390 }, { "epoch": 7.45, "learning_rate": 7.852777811625303e-07, "loss": 0.2167, "step": 57400 }, { "epoch": 7.45, "learning_rate": 7.845266059489484e-07, "loss": 0.2219, "step": 57410 }, { "epoch": 7.45, "learning_rate": 7.837757179598909e-07, "loss": 0.2174, "step": 57420 }, { "epoch": 7.45, "learning_rate": 7.830251173336773e-07, "loss": 0.228, "step": 57430 }, { "epoch": 7.45, "learning_rate": 7.822748042085759e-07, "loss": 0.228, "step": 57440 }, { "epoch": 7.46, "learning_rate": 7.815247787228029e-07, "loss": 0.2365, "step": 57450 }, { "epoch": 7.46, "learning_rate": 7.807750410145196e-07, "loss": 0.2395, "step": 57460 }, { "epoch": 7.46, "learning_rate": 7.800255912218348e-07, "loss": 0.2262, "step": 57470 }, { "epoch": 7.46, "learning_rate": 7.792764294828061e-07, "loss": 0.2323, "step": 57480 }, { "epoch": 7.46, "learning_rate": 7.785275559354355e-07, "loss": 0.2235, "step": 57490 }, { "epoch": 7.46, "learning_rate": 7.777789707176724e-07, "loss": 0.236, "step": 57500 }, { "epoch": 7.46, "learning_rate": 7.770306739674156e-07, "loss": 0.2179, "step": 57510 }, { "epoch": 7.47, "learning_rate": 7.762826658225074e-07, "loss": 0.2258, "step": 57520 }, { "epoch": 7.47, "learning_rate": 7.755349464207377e-07, "loss": 0.2297, "step": 57530 }, { "epoch": 7.47, "learning_rate": 7.747875158998459e-07, "loss": 0.2232, "step": 57540 }, { "epoch": 7.47, "learning_rate": 7.740403743975144e-07, "loss": 0.2307, "step": 57550 }, { "epoch": 7.47, "learning_rate": 7.732935220513755e-07, "loss": 0.2226, "step": 57560 }, { "epoch": 7.47, "learning_rate": 7.725469589990062e-07, "loss": 0.2425, "step": 57570 }, { "epoch": 7.47, "learning_rate": 7.718006853779301e-07, "loss": 0.2258, "step": 57580 }, { "epoch": 7.47, "learning_rate": 7.710547013256199e-07, "loss": 0.2299, "step": 57590 }, { "epoch": 7.48, "learning_rate": 7.703090069794924e-07, "loss": 0.2346, "step": 57600 }, { "epoch": 7.48, "learning_rate": 7.695636024769113e-07, "loss": 0.2276, "step": 57610 }, { "epoch": 7.48, "learning_rate": 7.68818487955189e-07, "loss": 0.2114, "step": 57620 }, { "epoch": 7.48, "learning_rate": 7.680736635515821e-07, "loss": 0.2433, "step": 57630 }, { "epoch": 7.48, "learning_rate": 7.673291294032942e-07, "loss": 0.2296, "step": 57640 }, { "epoch": 7.48, "learning_rate": 7.665848856474774e-07, "loss": 0.2218, "step": 57650 }, { "epoch": 7.48, "learning_rate": 7.658409324212282e-07, "loss": 0.2271, "step": 57660 }, { "epoch": 7.48, "learning_rate": 7.650972698615892e-07, "loss": 0.2291, "step": 57670 }, { "epoch": 7.49, "learning_rate": 7.64353898105552e-07, "loss": 0.2284, "step": 57680 }, { "epoch": 7.49, "learning_rate": 7.636108172900525e-07, "loss": 0.2248, "step": 57690 }, { "epoch": 7.49, "learning_rate": 7.628680275519728e-07, "loss": 0.2253, "step": 57700 }, { "epoch": 7.49, "learning_rate": 7.621255290281436e-07, "loss": 0.2342, "step": 57710 }, { "epoch": 7.49, "learning_rate": 7.613833218553391e-07, "loss": 0.2159, "step": 57720 }, { "epoch": 7.49, "learning_rate": 7.60641406170283e-07, "loss": 0.2228, "step": 57730 }, { "epoch": 7.49, "learning_rate": 7.598997821096426e-07, "loss": 0.2263, "step": 57740 }, { "epoch": 7.5, "learning_rate": 7.591584498100315e-07, "loss": 0.2321, "step": 57750 }, { "epoch": 7.5, "learning_rate": 7.584174094080124e-07, "loss": 0.2245, "step": 57760 }, { "epoch": 7.5, "learning_rate": 7.576766610400917e-07, "loss": 0.2321, "step": 57770 }, { "epoch": 7.5, "learning_rate": 7.569362048427212e-07, "loss": 0.2303, "step": 57780 }, { "epoch": 7.5, "learning_rate": 7.561960409523025e-07, "loss": 0.2163, "step": 57790 }, { "epoch": 7.5, "learning_rate": 7.554561695051801e-07, "loss": 0.2321, "step": 57800 }, { "epoch": 7.5, "learning_rate": 7.547165906376452e-07, "loss": 0.2251, "step": 57810 }, { "epoch": 7.5, "learning_rate": 7.539773044859371e-07, "loss": 0.2222, "step": 57820 }, { "epoch": 7.51, "learning_rate": 7.532383111862386e-07, "loss": 0.2382, "step": 57830 }, { "epoch": 7.51, "learning_rate": 7.524996108746793e-07, "loss": 0.234, "step": 57840 }, { "epoch": 7.51, "learning_rate": 7.517612036873363e-07, "loss": 0.2293, "step": 57850 }, { "epoch": 7.51, "learning_rate": 7.510230897602303e-07, "loss": 0.2382, "step": 57860 }, { "epoch": 7.51, "learning_rate": 7.502852692293308e-07, "loss": 0.2245, "step": 57870 }, { "epoch": 7.51, "learning_rate": 7.495477422305509e-07, "loss": 0.2253, "step": 57880 }, { "epoch": 7.51, "learning_rate": 7.488105088997496e-07, "loss": 0.2272, "step": 57890 }, { "epoch": 7.51, "learning_rate": 7.480735693727342e-07, "loss": 0.2284, "step": 57900 }, { "epoch": 7.52, "learning_rate": 7.473369237852559e-07, "loss": 0.2248, "step": 57910 }, { "epoch": 7.52, "learning_rate": 7.466005722730107e-07, "loss": 0.2215, "step": 57920 }, { "epoch": 7.52, "learning_rate": 7.458645149716441e-07, "loss": 0.2258, "step": 57930 }, { "epoch": 7.52, "learning_rate": 7.451287520167443e-07, "loss": 0.2252, "step": 57940 }, { "epoch": 7.52, "learning_rate": 7.44393283543845e-07, "loss": 0.2228, "step": 57950 }, { "epoch": 7.52, "learning_rate": 7.436581096884289e-07, "loss": 0.2196, "step": 57960 }, { "epoch": 7.52, "learning_rate": 7.429232305859214e-07, "loss": 0.2361, "step": 57970 }, { "epoch": 7.52, "learning_rate": 7.421886463716936e-07, "loss": 0.2232, "step": 57980 }, { "epoch": 7.53, "learning_rate": 7.41454357181065e-07, "loss": 0.2217, "step": 57990 }, { "epoch": 7.53, "learning_rate": 7.407203631492976e-07, "loss": 0.2423, "step": 58000 }, { "epoch": 7.53, "learning_rate": 7.399866644116015e-07, "loss": 0.2301, "step": 58010 }, { "epoch": 7.53, "learning_rate": 7.392532611031312e-07, "loss": 0.2298, "step": 58020 }, { "epoch": 7.53, "learning_rate": 7.385201533589857e-07, "loss": 0.2264, "step": 58030 }, { "epoch": 7.53, "learning_rate": 7.377873413142124e-07, "loss": 0.2248, "step": 58040 }, { "epoch": 7.53, "learning_rate": 7.370548251038016e-07, "loss": 0.1992, "step": 58050 }, { "epoch": 7.54, "learning_rate": 7.3632260486269e-07, "loss": 0.2267, "step": 58060 }, { "epoch": 7.54, "learning_rate": 7.355906807257607e-07, "loss": 0.2352, "step": 58070 }, { "epoch": 7.54, "learning_rate": 7.348590528278413e-07, "loss": 0.2177, "step": 58080 }, { "epoch": 7.54, "learning_rate": 7.341277213037036e-07, "loss": 0.2277, "step": 58090 }, { "epoch": 7.54, "learning_rate": 7.333966862880682e-07, "loss": 0.2474, "step": 58100 }, { "epoch": 7.54, "learning_rate": 7.326659479155979e-07, "loss": 0.2283, "step": 58110 }, { "epoch": 7.54, "learning_rate": 7.319355063209013e-07, "loss": 0.2396, "step": 58120 }, { "epoch": 7.54, "learning_rate": 7.312053616385348e-07, "loss": 0.2283, "step": 58130 }, { "epoch": 7.55, "learning_rate": 7.304755140029966e-07, "loss": 0.239, "step": 58140 }, { "epoch": 7.55, "learning_rate": 7.297459635487336e-07, "loss": 0.2284, "step": 58150 }, { "epoch": 7.55, "learning_rate": 7.290167104101358e-07, "loss": 0.2187, "step": 58160 }, { "epoch": 7.55, "learning_rate": 7.282877547215373e-07, "loss": 0.2249, "step": 58170 }, { "epoch": 7.55, "learning_rate": 7.275590966172214e-07, "loss": 0.2377, "step": 58180 }, { "epoch": 7.55, "learning_rate": 7.268307362314129e-07, "loss": 0.2194, "step": 58190 }, { "epoch": 7.55, "learning_rate": 7.261026736982825e-07, "loss": 0.2128, "step": 58200 }, { "epoch": 7.55, "learning_rate": 7.253749091519481e-07, "loss": 0.232, "step": 58210 }, { "epoch": 7.56, "learning_rate": 7.246474427264706e-07, "loss": 0.2291, "step": 58220 }, { "epoch": 7.56, "learning_rate": 7.239202745558552e-07, "loss": 0.2339, "step": 58230 }, { "epoch": 7.56, "learning_rate": 7.23193404774056e-07, "loss": 0.2312, "step": 58240 }, { "epoch": 7.56, "learning_rate": 7.224668335149682e-07, "loss": 0.2386, "step": 58250 }, { "epoch": 7.56, "learning_rate": 7.217405609124333e-07, "loss": 0.2332, "step": 58260 }, { "epoch": 7.56, "learning_rate": 7.210145871002387e-07, "loss": 0.2448, "step": 58270 }, { "epoch": 7.56, "learning_rate": 7.202889122121151e-07, "loss": 0.2173, "step": 58280 }, { "epoch": 7.57, "learning_rate": 7.195635363817406e-07, "loss": 0.2301, "step": 58290 }, { "epoch": 7.57, "learning_rate": 7.188384597427354e-07, "loss": 0.2352, "step": 58300 }, { "epoch": 7.57, "learning_rate": 7.181136824286657e-07, "loss": 0.2382, "step": 58310 }, { "epoch": 7.57, "learning_rate": 7.173892045730441e-07, "loss": 0.2244, "step": 58320 }, { "epoch": 7.57, "learning_rate": 7.166650263093256e-07, "loss": 0.2214, "step": 58330 }, { "epoch": 7.57, "learning_rate": 7.159411477709105e-07, "loss": 0.2271, "step": 58340 }, { "epoch": 7.57, "learning_rate": 7.152175690911459e-07, "loss": 0.2234, "step": 58350 }, { "epoch": 7.57, "learning_rate": 7.144942904033217e-07, "loss": 0.2258, "step": 58360 }, { "epoch": 7.58, "learning_rate": 7.137713118406721e-07, "loss": 0.227, "step": 58370 }, { "epoch": 7.58, "learning_rate": 7.130486335363784e-07, "loss": 0.2318, "step": 58380 }, { "epoch": 7.58, "learning_rate": 7.123262556235646e-07, "loss": 0.2249, "step": 58390 }, { "epoch": 7.58, "learning_rate": 7.116041782352991e-07, "loss": 0.2358, "step": 58400 }, { "epoch": 7.58, "learning_rate": 7.108824015045973e-07, "loss": 0.2175, "step": 58410 }, { "epoch": 7.58, "learning_rate": 7.101609255644162e-07, "loss": 0.2202, "step": 58420 }, { "epoch": 7.58, "learning_rate": 7.094397505476602e-07, "loss": 0.2438, "step": 58430 }, { "epoch": 7.58, "learning_rate": 7.087188765871761e-07, "loss": 0.2227, "step": 58440 }, { "epoch": 7.59, "learning_rate": 7.079983038157559e-07, "loss": 0.231, "step": 58450 }, { "epoch": 7.59, "learning_rate": 7.072780323661374e-07, "loss": 0.2377, "step": 58460 }, { "epoch": 7.59, "learning_rate": 7.06558062371001e-07, "loss": 0.2276, "step": 58470 }, { "epoch": 7.59, "learning_rate": 7.05838393962972e-07, "loss": 0.2273, "step": 58480 }, { "epoch": 7.59, "learning_rate": 7.051190272746215e-07, "loss": 0.2228, "step": 58490 }, { "epoch": 7.59, "learning_rate": 7.043999624384636e-07, "loss": 0.2325, "step": 58500 }, { "epoch": 7.59, "learning_rate": 7.036811995869562e-07, "loss": 0.2234, "step": 58510 }, { "epoch": 7.6, "learning_rate": 7.029627388525044e-07, "loss": 0.2361, "step": 58520 }, { "epoch": 7.6, "learning_rate": 7.022445803674549e-07, "loss": 0.2175, "step": 58530 }, { "epoch": 7.6, "learning_rate": 7.015267242640991e-07, "loss": 0.2422, "step": 58540 }, { "epoch": 7.6, "learning_rate": 7.008091706746744e-07, "loss": 0.2434, "step": 58550 }, { "epoch": 7.6, "learning_rate": 7.00091919731361e-07, "loss": 0.2224, "step": 58560 }, { "epoch": 7.6, "learning_rate": 6.993749715662828e-07, "loss": 0.2164, "step": 58570 }, { "epoch": 7.6, "learning_rate": 6.986583263115103e-07, "loss": 0.2297, "step": 58580 }, { "epoch": 7.6, "learning_rate": 6.979419840990553e-07, "loss": 0.2156, "step": 58590 }, { "epoch": 7.61, "learning_rate": 6.972259450608764e-07, "loss": 0.2328, "step": 58600 }, { "epoch": 7.61, "learning_rate": 6.965102093288746e-07, "loss": 0.2386, "step": 58610 }, { "epoch": 7.61, "learning_rate": 6.957947770348949e-07, "loss": 0.2307, "step": 58620 }, { "epoch": 7.61, "learning_rate": 6.950796483107285e-07, "loss": 0.2208, "step": 58630 }, { "epoch": 7.61, "learning_rate": 6.943648232881083e-07, "loss": 0.2281, "step": 58640 }, { "epoch": 7.61, "learning_rate": 6.936503020987118e-07, "loss": 0.2342, "step": 58650 }, { "epoch": 7.61, "learning_rate": 6.929360848741623e-07, "loss": 0.2303, "step": 58660 }, { "epoch": 7.61, "learning_rate": 6.92222171746025e-07, "loss": 0.2318, "step": 58670 }, { "epoch": 7.62, "learning_rate": 6.915085628458089e-07, "loss": 0.2053, "step": 58680 }, { "epoch": 7.62, "learning_rate": 6.907952583049697e-07, "loss": 0.2118, "step": 58690 }, { "epoch": 7.62, "learning_rate": 6.900822582549044e-07, "loss": 0.212, "step": 58700 }, { "epoch": 7.62, "learning_rate": 6.893695628269535e-07, "loss": 0.2365, "step": 58710 }, { "epoch": 7.62, "learning_rate": 6.886571721524049e-07, "loss": 0.2375, "step": 58720 }, { "epoch": 7.62, "learning_rate": 6.87945086362486e-07, "loss": 0.2392, "step": 58730 }, { "epoch": 7.62, "learning_rate": 6.87233305588372e-07, "loss": 0.2229, "step": 58740 }, { "epoch": 7.62, "learning_rate": 6.86521829961179e-07, "loss": 0.2327, "step": 58750 }, { "epoch": 7.63, "learning_rate": 6.858106596119672e-07, "loss": 0.2247, "step": 58760 }, { "epoch": 7.63, "learning_rate": 6.850997946717429e-07, "loss": 0.2371, "step": 58770 }, { "epoch": 7.63, "learning_rate": 6.843892352714536e-07, "loss": 0.2189, "step": 58780 }, { "epoch": 7.63, "learning_rate": 6.836789815419909e-07, "loss": 0.2292, "step": 58790 }, { "epoch": 7.63, "learning_rate": 6.829690336141919e-07, "loss": 0.22, "step": 58800 }, { "epoch": 7.63, "learning_rate": 6.822593916188357e-07, "loss": 0.2213, "step": 58810 }, { "epoch": 7.63, "learning_rate": 6.815500556866442e-07, "loss": 0.2265, "step": 58820 }, { "epoch": 7.64, "learning_rate": 6.808410259482862e-07, "loss": 0.2262, "step": 58830 }, { "epoch": 7.64, "learning_rate": 6.80132302534371e-07, "loss": 0.2293, "step": 58840 }, { "epoch": 7.64, "learning_rate": 6.794238855754518e-07, "loss": 0.2524, "step": 58850 }, { "epoch": 7.64, "learning_rate": 6.787157752020274e-07, "loss": 0.2339, "step": 58860 }, { "epoch": 7.64, "learning_rate": 6.780079715445379e-07, "loss": 0.207, "step": 58870 }, { "epoch": 7.64, "learning_rate": 6.773004747333686e-07, "loss": 0.2297, "step": 58880 }, { "epoch": 7.64, "learning_rate": 6.765932848988471e-07, "loss": 0.2125, "step": 58890 }, { "epoch": 7.64, "learning_rate": 6.758864021712442e-07, "loss": 0.24, "step": 58900 }, { "epoch": 7.65, "learning_rate": 6.751798266807759e-07, "loss": 0.2164, "step": 58910 }, { "epoch": 7.65, "learning_rate": 6.744735585576e-07, "loss": 0.24, "step": 58920 }, { "epoch": 7.65, "learning_rate": 6.737675979318171e-07, "loss": 0.2042, "step": 58930 }, { "epoch": 7.65, "learning_rate": 6.73061944933474e-07, "loss": 0.2276, "step": 58940 }, { "epoch": 7.65, "learning_rate": 6.723565996925582e-07, "loss": 0.22, "step": 58950 }, { "epoch": 7.65, "learning_rate": 6.716515623390008e-07, "loss": 0.2165, "step": 58960 }, { "epoch": 7.65, "learning_rate": 6.709468330026776e-07, "loss": 0.2361, "step": 58970 }, { "epoch": 7.65, "learning_rate": 6.702424118134064e-07, "loss": 0.2296, "step": 58980 }, { "epoch": 7.66, "learning_rate": 6.695382989009482e-07, "loss": 0.2367, "step": 58990 }, { "epoch": 7.66, "learning_rate": 6.688344943950085e-07, "loss": 0.2277, "step": 59000 }, { "epoch": 7.66, "learning_rate": 6.681309984252342e-07, "loss": 0.2164, "step": 59010 }, { "epoch": 7.66, "learning_rate": 6.674278111212173e-07, "loss": 0.2263, "step": 59020 }, { "epoch": 7.66, "learning_rate": 6.667249326124917e-07, "loss": 0.2242, "step": 59030 }, { "epoch": 7.66, "learning_rate": 6.660223630285332e-07, "loss": 0.2069, "step": 59040 }, { "epoch": 7.66, "learning_rate": 6.653201024987642e-07, "loss": 0.2263, "step": 59050 }, { "epoch": 7.67, "learning_rate": 6.646181511525473e-07, "loss": 0.2357, "step": 59060 }, { "epoch": 7.67, "learning_rate": 6.63916509119188e-07, "loss": 0.2363, "step": 59070 }, { "epoch": 7.67, "learning_rate": 6.632151765279371e-07, "loss": 0.223, "step": 59080 }, { "epoch": 7.67, "learning_rate": 6.625141535079868e-07, "loss": 0.2241, "step": 59090 }, { "epoch": 7.67, "learning_rate": 6.618134401884716e-07, "loss": 0.2232, "step": 59100 }, { "epoch": 7.67, "learning_rate": 6.611130366984714e-07, "loss": 0.2207, "step": 59110 }, { "epoch": 7.67, "learning_rate": 6.604129431670065e-07, "loss": 0.2211, "step": 59120 }, { "epoch": 7.67, "learning_rate": 6.597131597230406e-07, "loss": 0.2334, "step": 59130 }, { "epoch": 7.68, "learning_rate": 6.590136864954824e-07, "loss": 0.2276, "step": 59140 }, { "epoch": 7.68, "learning_rate": 6.583145236131801e-07, "loss": 0.2336, "step": 59150 }, { "epoch": 7.68, "learning_rate": 6.576156712049283e-07, "loss": 0.2234, "step": 59160 }, { "epoch": 7.68, "learning_rate": 6.569171293994617e-07, "loss": 0.2272, "step": 59170 }, { "epoch": 7.68, "learning_rate": 6.562188983254577e-07, "loss": 0.2271, "step": 59180 }, { "epoch": 7.68, "learning_rate": 6.555209781115393e-07, "loss": 0.2269, "step": 59190 }, { "epoch": 7.68, "learning_rate": 6.548233688862693e-07, "loss": 0.237, "step": 59200 }, { "epoch": 7.68, "learning_rate": 6.541260707781537e-07, "loss": 0.2159, "step": 59210 }, { "epoch": 7.69, "learning_rate": 6.534290839156436e-07, "loss": 0.2245, "step": 59220 }, { "epoch": 7.69, "learning_rate": 6.527324084271296e-07, "loss": 0.2235, "step": 59230 }, { "epoch": 7.69, "learning_rate": 6.52036044440946e-07, "loss": 0.2441, "step": 59240 }, { "epoch": 7.69, "learning_rate": 6.513399920853712e-07, "loss": 0.2322, "step": 59250 }, { "epoch": 7.69, "learning_rate": 6.506442514886242e-07, "loss": 0.2197, "step": 59260 }, { "epoch": 7.69, "learning_rate": 6.499488227788669e-07, "loss": 0.2341, "step": 59270 }, { "epoch": 7.69, "learning_rate": 6.492537060842056e-07, "loss": 0.2421, "step": 59280 }, { "epoch": 7.7, "learning_rate": 6.485589015326861e-07, "loss": 0.2223, "step": 59290 }, { "epoch": 7.7, "learning_rate": 6.478644092523001e-07, "loss": 0.2268, "step": 59300 }, { "epoch": 7.7, "learning_rate": 6.471702293709791e-07, "loss": 0.2374, "step": 59310 }, { "epoch": 7.7, "learning_rate": 6.464763620165972e-07, "loss": 0.2305, "step": 59320 }, { "epoch": 7.7, "learning_rate": 6.457828073169734e-07, "loss": 0.2219, "step": 59330 }, { "epoch": 7.7, "learning_rate": 6.450895653998665e-07, "loss": 0.2201, "step": 59340 }, { "epoch": 7.7, "learning_rate": 6.443966363929775e-07, "loss": 0.2222, "step": 59350 }, { "epoch": 7.7, "learning_rate": 6.43704020423953e-07, "loss": 0.2357, "step": 59360 }, { "epoch": 7.71, "learning_rate": 6.430117176203785e-07, "loss": 0.229, "step": 59370 }, { "epoch": 7.71, "learning_rate": 6.423197281097826e-07, "loss": 0.2317, "step": 59380 }, { "epoch": 7.71, "learning_rate": 6.41628052019638e-07, "loss": 0.2342, "step": 59390 }, { "epoch": 7.71, "learning_rate": 6.409366894773579e-07, "loss": 0.2183, "step": 59400 }, { "epoch": 7.71, "learning_rate": 6.402456406102974e-07, "loss": 0.2286, "step": 59410 }, { "epoch": 7.71, "learning_rate": 6.395549055457556e-07, "loss": 0.2324, "step": 59420 }, { "epoch": 7.71, "learning_rate": 6.388644844109727e-07, "loss": 0.2196, "step": 59430 }, { "epoch": 7.71, "learning_rate": 6.3817437733313e-07, "loss": 0.2417, "step": 59440 }, { "epoch": 7.72, "learning_rate": 6.374845844393538e-07, "loss": 0.2391, "step": 59450 }, { "epoch": 7.72, "learning_rate": 6.36795105856709e-07, "loss": 0.2229, "step": 59460 }, { "epoch": 7.72, "learning_rate": 6.361059417122066e-07, "loss": 0.2399, "step": 59470 }, { "epoch": 7.72, "learning_rate": 6.354170921327963e-07, "loss": 0.2271, "step": 59480 }, { "epoch": 7.72, "learning_rate": 6.347285572453705e-07, "loss": 0.2468, "step": 59490 }, { "epoch": 7.72, "learning_rate": 6.340403371767654e-07, "loss": 0.234, "step": 59500 }, { "epoch": 7.72, "learning_rate": 6.333524320537576e-07, "loss": 0.2407, "step": 59510 }, { "epoch": 7.72, "learning_rate": 6.326648420030652e-07, "loss": 0.2239, "step": 59520 }, { "epoch": 7.73, "learning_rate": 6.319775671513506e-07, "loss": 0.2196, "step": 59530 }, { "epoch": 7.73, "learning_rate": 6.312906076252162e-07, "loss": 0.2113, "step": 59540 }, { "epoch": 7.73, "learning_rate": 6.306039635512059e-07, "loss": 0.2376, "step": 59550 }, { "epoch": 7.73, "learning_rate": 6.299176350558077e-07, "loss": 0.2173, "step": 59560 }, { "epoch": 7.73, "learning_rate": 6.292316222654495e-07, "loss": 0.2209, "step": 59570 }, { "epoch": 7.73, "learning_rate": 6.285459253065013e-07, "loss": 0.2308, "step": 59580 }, { "epoch": 7.73, "learning_rate": 6.278605443052766e-07, "loss": 0.2395, "step": 59590 }, { "epoch": 7.74, "learning_rate": 6.271754793880279e-07, "loss": 0.2245, "step": 59600 }, { "epoch": 7.74, "learning_rate": 6.264907306809522e-07, "loss": 0.2337, "step": 59610 }, { "epoch": 7.74, "learning_rate": 6.258062983101869e-07, "loss": 0.2089, "step": 59620 }, { "epoch": 7.74, "learning_rate": 6.251221824018101e-07, "loss": 0.2344, "step": 59630 }, { "epoch": 7.74, "learning_rate": 6.244383830818444e-07, "loss": 0.2181, "step": 59640 }, { "epoch": 7.74, "learning_rate": 6.237549004762518e-07, "loss": 0.2287, "step": 59650 }, { "epoch": 7.74, "learning_rate": 6.230717347109359e-07, "loss": 0.2533, "step": 59660 }, { "epoch": 7.74, "learning_rate": 6.223888859117439e-07, "loss": 0.2261, "step": 59670 }, { "epoch": 7.75, "learning_rate": 6.21706354204463e-07, "loss": 0.237, "step": 59680 }, { "epoch": 7.75, "learning_rate": 6.210241397148212e-07, "loss": 0.238, "step": 59690 }, { "epoch": 7.75, "learning_rate": 6.20342242568491e-07, "loss": 0.2174, "step": 59700 }, { "epoch": 7.75, "learning_rate": 6.196606628910837e-07, "loss": 0.2356, "step": 59710 }, { "epoch": 7.75, "learning_rate": 6.189794008081527e-07, "loss": 0.218, "step": 59720 }, { "epoch": 7.75, "learning_rate": 6.182984564451945e-07, "loss": 0.2329, "step": 59730 }, { "epoch": 7.75, "learning_rate": 6.176178299276442e-07, "loss": 0.225, "step": 59740 }, { "epoch": 7.75, "learning_rate": 6.169375213808817e-07, "loss": 0.2371, "step": 59750 }, { "epoch": 7.76, "learning_rate": 6.16257530930226e-07, "loss": 0.2246, "step": 59760 }, { "epoch": 7.76, "learning_rate": 6.155778587009376e-07, "loss": 0.2215, "step": 59770 }, { "epoch": 7.76, "learning_rate": 6.148985048182194e-07, "loss": 0.2478, "step": 59780 }, { "epoch": 7.76, "learning_rate": 6.142194694072155e-07, "loss": 0.2384, "step": 59790 }, { "epoch": 7.76, "learning_rate": 6.1354075259301e-07, "loss": 0.236, "step": 59800 }, { "epoch": 7.76, "learning_rate": 6.128623545006305e-07, "loss": 0.2367, "step": 59810 }, { "epoch": 7.76, "learning_rate": 6.121842752550439e-07, "loss": 0.2152, "step": 59820 }, { "epoch": 7.77, "learning_rate": 6.11506514981159e-07, "loss": 0.2164, "step": 59830 }, { "epoch": 7.77, "learning_rate": 6.10829073803827e-07, "loss": 0.2177, "step": 59840 }, { "epoch": 7.77, "learning_rate": 6.101519518478386e-07, "loss": 0.2298, "step": 59850 }, { "epoch": 7.77, "learning_rate": 6.094751492379261e-07, "loss": 0.2296, "step": 59860 }, { "epoch": 7.77, "learning_rate": 6.087986660987644e-07, "loss": 0.2277, "step": 59870 }, { "epoch": 7.77, "learning_rate": 6.08122502554967e-07, "loss": 0.215, "step": 59880 }, { "epoch": 7.77, "learning_rate": 6.074466587310916e-07, "loss": 0.2318, "step": 59890 }, { "epoch": 7.77, "learning_rate": 6.067711347516347e-07, "loss": 0.2282, "step": 59900 }, { "epoch": 7.78, "learning_rate": 6.060959307410334e-07, "loss": 0.2189, "step": 59910 }, { "epoch": 7.78, "learning_rate": 6.054210468236689e-07, "loss": 0.2221, "step": 59920 }, { "epoch": 7.78, "learning_rate": 6.047464831238608e-07, "loss": 0.2164, "step": 59930 }, { "epoch": 7.78, "learning_rate": 6.040722397658699e-07, "loss": 0.2253, "step": 59940 }, { "epoch": 7.78, "learning_rate": 6.033983168738996e-07, "loss": 0.211, "step": 59950 }, { "epoch": 7.78, "learning_rate": 6.027247145720926e-07, "loss": 0.2414, "step": 59960 }, { "epoch": 7.78, "learning_rate": 6.020514329845326e-07, "loss": 0.2294, "step": 59970 }, { "epoch": 7.78, "learning_rate": 6.013784722352461e-07, "loss": 0.237, "step": 59980 }, { "epoch": 7.79, "learning_rate": 6.007058324481986e-07, "loss": 0.2337, "step": 59990 }, { "epoch": 7.79, "learning_rate": 6.000335137472963e-07, "loss": 0.2246, "step": 60000 }, { "epoch": 7.79, "learning_rate": 5.993615162563885e-07, "loss": 0.2168, "step": 60010 }, { "epoch": 7.79, "learning_rate": 5.986898400992623e-07, "loss": 0.2197, "step": 60020 }, { "epoch": 7.79, "learning_rate": 5.980184853996486e-07, "loss": 0.2259, "step": 60030 }, { "epoch": 7.79, "learning_rate": 5.97347452281217e-07, "loss": 0.2221, "step": 60040 }, { "epoch": 7.79, "learning_rate": 5.966767408675779e-07, "loss": 0.2194, "step": 60050 }, { "epoch": 7.79, "learning_rate": 5.960063512822843e-07, "loss": 0.2221, "step": 60060 }, { "epoch": 7.8, "learning_rate": 5.953362836488279e-07, "loss": 0.2235, "step": 60070 }, { "epoch": 7.8, "learning_rate": 5.946665380906412e-07, "loss": 0.2264, "step": 60080 }, { "epoch": 7.8, "learning_rate": 5.939971147310996e-07, "loss": 0.2225, "step": 60090 }, { "epoch": 7.8, "learning_rate": 5.933280136935167e-07, "loss": 0.2268, "step": 60100 }, { "epoch": 7.8, "learning_rate": 5.92659235101147e-07, "loss": 0.2361, "step": 60110 }, { "epoch": 7.8, "learning_rate": 5.919907790771873e-07, "loss": 0.232, "step": 60120 }, { "epoch": 7.8, "learning_rate": 5.913226457447736e-07, "loss": 0.211, "step": 60130 }, { "epoch": 7.81, "learning_rate": 5.906548352269816e-07, "loss": 0.2178, "step": 60140 }, { "epoch": 7.81, "learning_rate": 5.899873476468305e-07, "loss": 0.2066, "step": 60150 }, { "epoch": 7.81, "learning_rate": 5.893201831272765e-07, "loss": 0.2261, "step": 60160 }, { "epoch": 7.81, "learning_rate": 5.886533417912192e-07, "loss": 0.2295, "step": 60170 }, { "epoch": 7.81, "learning_rate": 5.879868237614973e-07, "loss": 0.2389, "step": 60180 }, { "epoch": 7.81, "learning_rate": 5.873206291608891e-07, "loss": 0.2313, "step": 60190 }, { "epoch": 7.81, "learning_rate": 5.866547581121154e-07, "loss": 0.2246, "step": 60200 }, { "epoch": 7.81, "learning_rate": 5.859892107378359e-07, "loss": 0.2244, "step": 60210 }, { "epoch": 7.82, "learning_rate": 5.853239871606506e-07, "loss": 0.2331, "step": 60220 }, { "epoch": 7.82, "learning_rate": 5.846590875031012e-07, "loss": 0.2292, "step": 60230 }, { "epoch": 7.82, "learning_rate": 5.839945118876682e-07, "loss": 0.2243, "step": 60240 }, { "epoch": 7.82, "learning_rate": 5.83330260436773e-07, "loss": 0.2206, "step": 60250 }, { "epoch": 7.82, "learning_rate": 5.826663332727783e-07, "loss": 0.2264, "step": 60260 }, { "epoch": 7.82, "learning_rate": 5.82002730517985e-07, "loss": 0.2243, "step": 60270 }, { "epoch": 7.82, "learning_rate": 5.813394522946356e-07, "loss": 0.2274, "step": 60280 }, { "epoch": 7.82, "learning_rate": 5.806764987249134e-07, "loss": 0.2192, "step": 60290 }, { "epoch": 7.83, "learning_rate": 5.800138699309396e-07, "loss": 0.2589, "step": 60300 }, { "epoch": 7.83, "learning_rate": 5.793515660347785e-07, "loss": 0.225, "step": 60310 }, { "epoch": 7.83, "learning_rate": 5.786895871584326e-07, "loss": 0.2247, "step": 60320 }, { "epoch": 7.83, "learning_rate": 5.780279334238443e-07, "loss": 0.2385, "step": 60330 }, { "epoch": 7.83, "learning_rate": 5.77366604952898e-07, "loss": 0.2244, "step": 60340 }, { "epoch": 7.83, "learning_rate": 5.767056018674165e-07, "loss": 0.237, "step": 60350 }, { "epoch": 7.83, "learning_rate": 5.760449242891623e-07, "loss": 0.2178, "step": 60360 }, { "epoch": 7.84, "learning_rate": 5.753845723398405e-07, "loss": 0.2315, "step": 60370 }, { "epoch": 7.84, "learning_rate": 5.747245461410934e-07, "loss": 0.2183, "step": 60380 }, { "epoch": 7.84, "learning_rate": 5.740648458145043e-07, "loss": 0.2321, "step": 60390 }, { "epoch": 7.84, "learning_rate": 5.734054714815974e-07, "loss": 0.2274, "step": 60400 }, { "epoch": 7.84, "learning_rate": 5.727464232638357e-07, "loss": 0.2298, "step": 60410 }, { "epoch": 7.84, "learning_rate": 5.720877012826215e-07, "loss": 0.2262, "step": 60420 }, { "epoch": 7.84, "learning_rate": 5.714293056592997e-07, "loss": 0.209, "step": 60430 }, { "epoch": 7.84, "learning_rate": 5.707712365151525e-07, "loss": 0.225, "step": 60440 }, { "epoch": 7.85, "learning_rate": 5.701134939714018e-07, "loss": 0.2289, "step": 60450 }, { "epoch": 7.85, "learning_rate": 5.694560781492124e-07, "loss": 0.2299, "step": 60460 }, { "epoch": 7.85, "learning_rate": 5.687989891696849e-07, "loss": 0.2294, "step": 60470 }, { "epoch": 7.85, "learning_rate": 5.681422271538632e-07, "loss": 0.2315, "step": 60480 }, { "epoch": 7.85, "learning_rate": 5.674857922227288e-07, "loss": 0.2294, "step": 60490 }, { "epoch": 7.85, "learning_rate": 5.66829684497203e-07, "loss": 0.2107, "step": 60500 }, { "epoch": 7.85, "learning_rate": 5.661739040981484e-07, "loss": 0.2243, "step": 60510 }, { "epoch": 7.85, "learning_rate": 5.655184511463657e-07, "loss": 0.2276, "step": 60520 }, { "epoch": 7.86, "learning_rate": 5.648633257625958e-07, "loss": 0.2259, "step": 60530 }, { "epoch": 7.86, "learning_rate": 5.642085280675198e-07, "loss": 0.2276, "step": 60540 }, { "epoch": 7.86, "learning_rate": 5.63554058181758e-07, "loss": 0.2219, "step": 60550 }, { "epoch": 7.86, "learning_rate": 5.628999162258693e-07, "loss": 0.2384, "step": 60560 }, { "epoch": 7.86, "learning_rate": 5.622461023203544e-07, "loss": 0.2144, "step": 60570 }, { "epoch": 7.86, "learning_rate": 5.61592616585652e-07, "loss": 0.2216, "step": 60580 }, { "epoch": 7.86, "learning_rate": 5.609394591421401e-07, "loss": 0.227, "step": 60590 }, { "epoch": 7.87, "learning_rate": 5.602866301101375e-07, "loss": 0.2221, "step": 60600 }, { "epoch": 7.87, "learning_rate": 5.596341296099012e-07, "loss": 0.2289, "step": 60610 }, { "epoch": 7.87, "learning_rate": 5.589819577616294e-07, "loss": 0.2223, "step": 60620 }, { "epoch": 7.87, "learning_rate": 5.583301146854577e-07, "loss": 0.2208, "step": 60630 }, { "epoch": 7.87, "learning_rate": 5.576786005014621e-07, "loss": 0.2348, "step": 60640 }, { "epoch": 7.87, "learning_rate": 5.570274153296588e-07, "loss": 0.227, "step": 60650 }, { "epoch": 7.87, "learning_rate": 5.563765592900022e-07, "loss": 0.2303, "step": 60660 }, { "epoch": 7.87, "learning_rate": 5.557260325023855e-07, "loss": 0.2258, "step": 60670 }, { "epoch": 7.88, "learning_rate": 5.550758350866442e-07, "loss": 0.2146, "step": 60680 }, { "epoch": 7.88, "learning_rate": 5.544259671625498e-07, "loss": 0.2153, "step": 60690 }, { "epoch": 7.88, "learning_rate": 5.537764288498142e-07, "loss": 0.2213, "step": 60700 }, { "epoch": 7.88, "learning_rate": 5.531272202680902e-07, "loss": 0.2325, "step": 60710 }, { "epoch": 7.88, "learning_rate": 5.524783415369673e-07, "loss": 0.2263, "step": 60720 }, { "epoch": 7.88, "learning_rate": 5.518297927759757e-07, "loss": 0.2357, "step": 60730 }, { "epoch": 7.88, "learning_rate": 5.511815741045849e-07, "loss": 0.2362, "step": 60740 }, { "epoch": 7.88, "learning_rate": 5.505336856422029e-07, "loss": 0.2234, "step": 60750 }, { "epoch": 7.89, "learning_rate": 5.498861275081776e-07, "loss": 0.2237, "step": 60760 }, { "epoch": 7.89, "learning_rate": 5.492388998217959e-07, "loss": 0.228, "step": 60770 }, { "epoch": 7.89, "learning_rate": 5.485920027022821e-07, "loss": 0.2303, "step": 60780 }, { "epoch": 7.89, "learning_rate": 5.479454362688031e-07, "loss": 0.2195, "step": 60790 }, { "epoch": 7.89, "learning_rate": 5.472992006404617e-07, "loss": 0.2147, "step": 60800 }, { "epoch": 7.89, "learning_rate": 5.466532959363009e-07, "loss": 0.2313, "step": 60810 }, { "epoch": 7.89, "learning_rate": 5.460077222753037e-07, "loss": 0.218, "step": 60820 }, { "epoch": 7.89, "learning_rate": 5.453624797763908e-07, "loss": 0.2339, "step": 60830 }, { "epoch": 7.9, "learning_rate": 5.447175685584211e-07, "loss": 0.2171, "step": 60840 }, { "epoch": 7.9, "learning_rate": 5.440729887401958e-07, "loss": 0.2177, "step": 60850 }, { "epoch": 7.9, "learning_rate": 5.434287404404517e-07, "loss": 0.2312, "step": 60860 }, { "epoch": 7.9, "learning_rate": 5.427848237778654e-07, "loss": 0.2121, "step": 60870 }, { "epoch": 7.9, "learning_rate": 5.42141238871054e-07, "loss": 0.2337, "step": 60880 }, { "epoch": 7.9, "learning_rate": 5.41497985838571e-07, "loss": 0.2431, "step": 60890 }, { "epoch": 7.9, "learning_rate": 5.408550647989112e-07, "loss": 0.2286, "step": 60900 }, { "epoch": 7.91, "learning_rate": 5.402124758705064e-07, "loss": 0.2272, "step": 60910 }, { "epoch": 7.91, "learning_rate": 5.395702191717276e-07, "loss": 0.2175, "step": 60920 }, { "epoch": 7.91, "learning_rate": 5.389282948208859e-07, "loss": 0.2163, "step": 60930 }, { "epoch": 7.91, "learning_rate": 5.382867029362297e-07, "loss": 0.2231, "step": 60940 }, { "epoch": 7.91, "learning_rate": 5.376454436359457e-07, "loss": 0.2097, "step": 60950 }, { "epoch": 7.91, "learning_rate": 5.370045170381619e-07, "loss": 0.2224, "step": 60960 }, { "epoch": 7.91, "learning_rate": 5.363639232609425e-07, "loss": 0.2226, "step": 60970 }, { "epoch": 7.91, "learning_rate": 5.357236624222907e-07, "loss": 0.2135, "step": 60980 }, { "epoch": 7.92, "learning_rate": 5.350837346401504e-07, "loss": 0.2229, "step": 60990 }, { "epoch": 7.92, "learning_rate": 5.344441400324018e-07, "loss": 0.2144, "step": 61000 }, { "epoch": 7.92, "learning_rate": 5.33804878716864e-07, "loss": 0.2273, "step": 61010 }, { "epoch": 7.92, "learning_rate": 5.331659508112967e-07, "loss": 0.2368, "step": 61020 }, { "epoch": 7.92, "learning_rate": 5.325273564333955e-07, "loss": 0.2242, "step": 61030 }, { "epoch": 7.92, "learning_rate": 5.318890957007971e-07, "loss": 0.2293, "step": 61040 }, { "epoch": 7.92, "learning_rate": 5.312511687310752e-07, "loss": 0.2323, "step": 61050 }, { "epoch": 7.92, "learning_rate": 5.306135756417409e-07, "loss": 0.2279, "step": 61060 }, { "epoch": 7.93, "learning_rate": 5.299763165502472e-07, "loss": 0.2251, "step": 61070 }, { "epoch": 7.93, "learning_rate": 5.293393915739828e-07, "loss": 0.2185, "step": 61080 }, { "epoch": 7.93, "learning_rate": 5.287028008302748e-07, "loss": 0.2217, "step": 61090 }, { "epoch": 7.93, "learning_rate": 5.280665444363909e-07, "loss": 0.2111, "step": 61100 }, { "epoch": 7.93, "learning_rate": 5.274306225095354e-07, "loss": 0.2389, "step": 61110 }, { "epoch": 7.93, "learning_rate": 5.267950351668506e-07, "loss": 0.2126, "step": 61120 }, { "epoch": 7.93, "learning_rate": 5.261597825254198e-07, "loss": 0.2241, "step": 61130 }, { "epoch": 7.94, "learning_rate": 5.255248647022615e-07, "loss": 0.2235, "step": 61140 }, { "epoch": 7.94, "learning_rate": 5.248902818143339e-07, "loss": 0.2146, "step": 61150 }, { "epoch": 7.94, "learning_rate": 5.242560339785347e-07, "loss": 0.2185, "step": 61160 }, { "epoch": 7.94, "learning_rate": 5.236221213116969e-07, "loss": 0.2279, "step": 61170 }, { "epoch": 7.94, "learning_rate": 5.229885439305955e-07, "loss": 0.2302, "step": 61180 }, { "epoch": 7.94, "learning_rate": 5.223553019519408e-07, "loss": 0.2177, "step": 61190 }, { "epoch": 7.94, "learning_rate": 5.217223954923814e-07, "loss": 0.2123, "step": 61200 }, { "epoch": 7.94, "learning_rate": 5.210898246685069e-07, "loss": 0.2428, "step": 61210 }, { "epoch": 7.95, "learning_rate": 5.20457589596842e-07, "loss": 0.2226, "step": 61220 }, { "epoch": 7.95, "learning_rate": 5.198256903938503e-07, "loss": 0.2262, "step": 61230 }, { "epoch": 7.95, "learning_rate": 5.191941271759352e-07, "loss": 0.232, "step": 61240 }, { "epoch": 7.95, "learning_rate": 5.185629000594363e-07, "loss": 0.2319, "step": 61250 }, { "epoch": 7.95, "learning_rate": 5.179320091606311e-07, "loss": 0.2293, "step": 61260 }, { "epoch": 7.95, "learning_rate": 5.173014545957377e-07, "loss": 0.2262, "step": 61270 }, { "epoch": 7.95, "learning_rate": 5.166712364809097e-07, "loss": 0.2187, "step": 61280 }, { "epoch": 7.95, "learning_rate": 5.160413549322388e-07, "loss": 0.2327, "step": 61290 }, { "epoch": 7.96, "learning_rate": 5.15411810065757e-07, "loss": 0.2196, "step": 61300 }, { "epoch": 7.96, "learning_rate": 5.14782601997432e-07, "loss": 0.2234, "step": 61310 }, { "epoch": 7.96, "learning_rate": 5.141537308431694e-07, "loss": 0.2259, "step": 61320 }, { "epoch": 7.96, "learning_rate": 5.135251967188151e-07, "loss": 0.2181, "step": 61330 }, { "epoch": 7.96, "learning_rate": 5.1289699974015e-07, "loss": 0.2368, "step": 61340 }, { "epoch": 7.96, "learning_rate": 5.122691400228954e-07, "loss": 0.2223, "step": 61350 }, { "epoch": 7.96, "learning_rate": 5.116416176827088e-07, "loss": 0.2301, "step": 61360 }, { "epoch": 7.96, "learning_rate": 5.110144328351856e-07, "loss": 0.2299, "step": 61370 }, { "epoch": 7.97, "learning_rate": 5.103875855958603e-07, "loss": 0.2288, "step": 61380 }, { "epoch": 7.97, "learning_rate": 5.097610760802045e-07, "loss": 0.215, "step": 61390 }, { "epoch": 7.97, "learning_rate": 5.091349044036263e-07, "loss": 0.2281, "step": 61400 }, { "epoch": 7.97, "learning_rate": 5.085090706814745e-07, "loss": 0.2193, "step": 61410 }, { "epoch": 7.97, "learning_rate": 5.078835750290328e-07, "loss": 0.2403, "step": 61420 }, { "epoch": 7.97, "learning_rate": 5.072584175615233e-07, "loss": 0.2228, "step": 61430 }, { "epoch": 7.97, "learning_rate": 5.066335983941077e-07, "loss": 0.2325, "step": 61440 }, { "epoch": 7.98, "learning_rate": 5.06009117641883e-07, "loss": 0.2282, "step": 61450 }, { "epoch": 7.98, "learning_rate": 5.053849754198843e-07, "loss": 0.2361, "step": 61460 }, { "epoch": 7.98, "learning_rate": 5.047611718430863e-07, "loss": 0.2137, "step": 61470 }, { "epoch": 7.98, "learning_rate": 5.041377070263983e-07, "loss": 0.232, "step": 61480 }, { "epoch": 7.98, "learning_rate": 5.035145810846699e-07, "loss": 0.2333, "step": 61490 }, { "epoch": 7.98, "learning_rate": 5.02891794132687e-07, "loss": 0.2363, "step": 61500 }, { "epoch": 7.98, "learning_rate": 5.022693462851723e-07, "loss": 0.2316, "step": 61510 }, { "epoch": 7.98, "learning_rate": 5.016472376567879e-07, "loss": 0.2182, "step": 61520 }, { "epoch": 7.99, "learning_rate": 5.010254683621321e-07, "loss": 0.2311, "step": 61530 }, { "epoch": 7.99, "learning_rate": 5.004040385157403e-07, "loss": 0.2202, "step": 61540 }, { "epoch": 7.99, "learning_rate": 4.997829482320872e-07, "loss": 0.2214, "step": 61550 }, { "epoch": 7.99, "learning_rate": 4.991621976255834e-07, "loss": 0.2299, "step": 61560 }, { "epoch": 7.99, "learning_rate": 4.98541786810577e-07, "loss": 0.2196, "step": 61570 }, { "epoch": 7.99, "learning_rate": 4.979217159013545e-07, "loss": 0.2279, "step": 61580 }, { "epoch": 7.99, "learning_rate": 4.973019850121387e-07, "loss": 0.2151, "step": 61590 }, { "epoch": 7.99, "learning_rate": 4.966825942570902e-07, "loss": 0.2259, "step": 61600 }, { "epoch": 8.0, "learning_rate": 4.960635437503073e-07, "loss": 0.228, "step": 61610 }, { "epoch": 8.0, "learning_rate": 4.954448336058245e-07, "loss": 0.225, "step": 61620 }, { "epoch": 8.0, "learning_rate": 4.948264639376158e-07, "loss": 0.2375, "step": 61630 }, { "epoch": 8.0, "learning_rate": 4.942084348595901e-07, "loss": 0.2267, "step": 61640 }, { "epoch": 8.0, "eval_loss": 0.22624196112155914, "eval_runtime": 61.5665, "eval_samples_per_second": 176.671, "eval_steps_per_second": 22.09, "step": 61640 }, { "epoch": 8.0, "learning_rate": 4.935907464855943e-07, "loss": 0.2255, "step": 61650 }, { "epoch": 8.0, "learning_rate": 4.929733989294136e-07, "loss": 0.2038, "step": 61660 }, { "epoch": 8.0, "learning_rate": 4.923563923047693e-07, "loss": 0.2328, "step": 61670 }, { "epoch": 8.01, "learning_rate": 4.917397267253192e-07, "loss": 0.2231, "step": 61680 }, { "epoch": 8.01, "learning_rate": 4.911234023046609e-07, "loss": 0.2352, "step": 61690 }, { "epoch": 8.01, "learning_rate": 4.905074191563265e-07, "loss": 0.2364, "step": 61700 }, { "epoch": 8.01, "learning_rate": 4.898917773937858e-07, "loss": 0.2243, "step": 61710 }, { "epoch": 8.01, "learning_rate": 4.892764771304477e-07, "loss": 0.2241, "step": 61720 }, { "epoch": 8.01, "learning_rate": 4.886615184796555e-07, "loss": 0.2226, "step": 61730 }, { "epoch": 8.01, "learning_rate": 4.880469015546906e-07, "loss": 0.2311, "step": 61740 }, { "epoch": 8.01, "learning_rate": 4.874326264687725e-07, "loss": 0.2225, "step": 61750 }, { "epoch": 8.02, "learning_rate": 4.868186933350555e-07, "loss": 0.2271, "step": 61760 }, { "epoch": 8.02, "learning_rate": 4.862051022666337e-07, "loss": 0.2289, "step": 61770 }, { "epoch": 8.02, "learning_rate": 4.855918533765362e-07, "loss": 0.222, "step": 61780 }, { "epoch": 8.02, "learning_rate": 4.849789467777284e-07, "loss": 0.2137, "step": 61790 }, { "epoch": 8.02, "learning_rate": 4.843663825831157e-07, "loss": 0.2363, "step": 61800 }, { "epoch": 8.02, "learning_rate": 4.837541609055374e-07, "loss": 0.2289, "step": 61810 }, { "epoch": 8.02, "learning_rate": 4.831422818577707e-07, "loss": 0.2142, "step": 61820 }, { "epoch": 8.02, "learning_rate": 4.825307455525307e-07, "loss": 0.2288, "step": 61830 }, { "epoch": 8.03, "learning_rate": 4.819195521024679e-07, "loss": 0.2281, "step": 61840 }, { "epoch": 8.03, "learning_rate": 4.813087016201697e-07, "loss": 0.2327, "step": 61850 }, { "epoch": 8.03, "learning_rate": 4.806981942181624e-07, "loss": 0.2233, "step": 61860 }, { "epoch": 8.03, "learning_rate": 4.800880300089062e-07, "loss": 0.2225, "step": 61870 }, { "epoch": 8.03, "learning_rate": 4.794782091047998e-07, "loss": 0.2313, "step": 61880 }, { "epoch": 8.03, "learning_rate": 4.788687316181787e-07, "loss": 0.2347, "step": 61890 }, { "epoch": 8.03, "learning_rate": 4.782595976613138e-07, "loss": 0.2231, "step": 61900 }, { "epoch": 8.04, "learning_rate": 4.77650807346415e-07, "loss": 0.2287, "step": 61910 }, { "epoch": 8.04, "learning_rate": 4.77042360785627e-07, "loss": 0.2133, "step": 61920 }, { "epoch": 8.04, "learning_rate": 4.7643425809103087e-07, "loss": 0.2305, "step": 61930 }, { "epoch": 8.04, "learning_rate": 4.758264993746466e-07, "loss": 0.2232, "step": 61940 }, { "epoch": 8.04, "learning_rate": 4.752190847484285e-07, "loss": 0.2318, "step": 61950 }, { "epoch": 8.04, "learning_rate": 4.7461201432426823e-07, "loss": 0.2381, "step": 61960 }, { "epoch": 8.04, "learning_rate": 4.7400528821399525e-07, "loss": 0.2331, "step": 61970 }, { "epoch": 8.04, "learning_rate": 4.7339890652937404e-07, "loss": 0.2246, "step": 61980 }, { "epoch": 8.05, "learning_rate": 4.727928693821054e-07, "loss": 0.2218, "step": 61990 }, { "epoch": 8.05, "learning_rate": 4.7218717688382866e-07, "loss": 0.2316, "step": 62000 }, { "epoch": 8.05, "learning_rate": 4.715818291461181e-07, "loss": 0.2284, "step": 62010 }, { "epoch": 8.05, "learning_rate": 4.7097682628048376e-07, "loss": 0.2241, "step": 62020 }, { "epoch": 8.05, "learning_rate": 4.7037216839837476e-07, "loss": 0.2266, "step": 62030 }, { "epoch": 8.05, "learning_rate": 4.697678556111739e-07, "loss": 0.2235, "step": 62040 }, { "epoch": 8.05, "learning_rate": 4.691638880302025e-07, "loss": 0.228, "step": 62050 }, { "epoch": 8.05, "learning_rate": 4.685602657667174e-07, "loss": 0.2343, "step": 62060 }, { "epoch": 8.06, "learning_rate": 4.679569889319107e-07, "loss": 0.2237, "step": 62070 }, { "epoch": 8.06, "learning_rate": 4.6735405763691357e-07, "loss": 0.2267, "step": 62080 }, { "epoch": 8.06, "learning_rate": 4.6675147199279123e-07, "loss": 0.235, "step": 62090 }, { "epoch": 8.06, "learning_rate": 4.6614923211054505e-07, "loss": 0.2431, "step": 62100 }, { "epoch": 8.06, "learning_rate": 4.6554733810111547e-07, "loss": 0.222, "step": 62110 }, { "epoch": 8.06, "learning_rate": 4.6494579007537653e-07, "loss": 0.2288, "step": 62120 }, { "epoch": 8.06, "learning_rate": 4.643445881441386e-07, "loss": 0.2336, "step": 62130 }, { "epoch": 8.06, "learning_rate": 4.6374373241815055e-07, "loss": 0.2198, "step": 62140 }, { "epoch": 8.07, "learning_rate": 4.631432230080952e-07, "loss": 0.2395, "step": 62150 }, { "epoch": 8.07, "learning_rate": 4.6254306002459185e-07, "loss": 0.2324, "step": 62160 }, { "epoch": 8.07, "learning_rate": 4.619432435781977e-07, "loss": 0.2316, "step": 62170 }, { "epoch": 8.07, "learning_rate": 4.613437737794037e-07, "loss": 0.2087, "step": 62180 }, { "epoch": 8.07, "learning_rate": 4.607446507386396e-07, "loss": 0.2215, "step": 62190 }, { "epoch": 8.07, "learning_rate": 4.601458745662691e-07, "loss": 0.2287, "step": 62200 }, { "epoch": 8.07, "learning_rate": 4.595474453725918e-07, "loss": 0.2359, "step": 62210 }, { "epoch": 8.08, "learning_rate": 4.5894936326784604e-07, "loss": 0.2387, "step": 62220 }, { "epoch": 8.08, "learning_rate": 4.583516283622039e-07, "loss": 0.2238, "step": 62230 }, { "epoch": 8.08, "learning_rate": 4.577542407657733e-07, "loss": 0.2167, "step": 62240 }, { "epoch": 8.08, "learning_rate": 4.571572005886002e-07, "loss": 0.2269, "step": 62250 }, { "epoch": 8.08, "learning_rate": 4.565605079406647e-07, "loss": 0.2255, "step": 62260 }, { "epoch": 8.08, "learning_rate": 4.559641629318832e-07, "loss": 0.2182, "step": 62270 }, { "epoch": 8.08, "learning_rate": 4.553681656721093e-07, "loss": 0.2216, "step": 62280 }, { "epoch": 8.08, "learning_rate": 4.5477251627113144e-07, "loss": 0.2125, "step": 62290 }, { "epoch": 8.09, "learning_rate": 4.5417721483867343e-07, "loss": 0.2323, "step": 62300 }, { "epoch": 8.09, "learning_rate": 4.535822614843968e-07, "loss": 0.2165, "step": 62310 }, { "epoch": 8.09, "learning_rate": 4.5298765631789724e-07, "loss": 0.2119, "step": 62320 }, { "epoch": 8.09, "learning_rate": 4.523933994487066e-07, "loss": 0.2187, "step": 62330 }, { "epoch": 8.09, "learning_rate": 4.517994909862942e-07, "loss": 0.2264, "step": 62340 }, { "epoch": 8.09, "learning_rate": 4.512059310400622e-07, "loss": 0.2356, "step": 62350 }, { "epoch": 8.09, "learning_rate": 4.5061271971935194e-07, "loss": 0.2359, "step": 62360 }, { "epoch": 8.09, "learning_rate": 4.500198571334384e-07, "loss": 0.2274, "step": 62370 }, { "epoch": 8.1, "learning_rate": 4.4942734339153173e-07, "loss": 0.2225, "step": 62380 }, { "epoch": 8.1, "learning_rate": 4.488351786027802e-07, "loss": 0.215, "step": 62390 }, { "epoch": 8.1, "learning_rate": 4.4824336287626613e-07, "loss": 0.2318, "step": 62400 }, { "epoch": 8.1, "learning_rate": 4.4765189632100723e-07, "loss": 0.2235, "step": 62410 }, { "epoch": 8.1, "learning_rate": 4.470607790459583e-07, "loss": 0.2343, "step": 62420 }, { "epoch": 8.1, "learning_rate": 4.464700111600091e-07, "loss": 0.2311, "step": 62430 }, { "epoch": 8.1, "learning_rate": 4.4587959277198394e-07, "loss": 0.2231, "step": 62440 }, { "epoch": 8.11, "learning_rate": 4.452895239906451e-07, "loss": 0.2185, "step": 62450 }, { "epoch": 8.11, "learning_rate": 4.446998049246887e-07, "loss": 0.2308, "step": 62460 }, { "epoch": 8.11, "learning_rate": 4.441104356827461e-07, "loss": 0.2135, "step": 62470 }, { "epoch": 8.11, "learning_rate": 4.435214163733861e-07, "loss": 0.2321, "step": 62480 }, { "epoch": 8.11, "learning_rate": 4.429327471051108e-07, "loss": 0.2295, "step": 62490 }, { "epoch": 8.11, "learning_rate": 4.4234442798636044e-07, "loss": 0.2171, "step": 62500 }, { "epoch": 8.11, "learning_rate": 4.4175645912550834e-07, "loss": 0.2206, "step": 62510 }, { "epoch": 8.11, "learning_rate": 4.411688406308639e-07, "loss": 0.2337, "step": 62520 }, { "epoch": 8.12, "learning_rate": 4.405815726106733e-07, "loss": 0.2235, "step": 62530 }, { "epoch": 8.12, "learning_rate": 4.3999465517311676e-07, "loss": 0.2216, "step": 62540 }, { "epoch": 8.12, "learning_rate": 4.3940808842630955e-07, "loss": 0.235, "step": 62550 }, { "epoch": 8.12, "learning_rate": 4.3882187247830433e-07, "loss": 0.2184, "step": 62560 }, { "epoch": 8.12, "learning_rate": 4.382360074370874e-07, "loss": 0.2349, "step": 62570 }, { "epoch": 8.12, "learning_rate": 4.376504934105804e-07, "loss": 0.2164, "step": 62580 }, { "epoch": 8.12, "learning_rate": 4.3706533050664226e-07, "loss": 0.2217, "step": 62590 }, { "epoch": 8.12, "learning_rate": 4.3648051883306477e-07, "loss": 0.2247, "step": 62600 }, { "epoch": 8.13, "learning_rate": 4.3589605849757587e-07, "loss": 0.2265, "step": 62610 }, { "epoch": 8.13, "learning_rate": 4.353119496078399e-07, "loss": 0.2271, "step": 62620 }, { "epoch": 8.13, "learning_rate": 4.3472819227145445e-07, "loss": 0.2183, "step": 62630 }, { "epoch": 8.13, "learning_rate": 4.3414478659595485e-07, "loss": 0.2405, "step": 62640 }, { "epoch": 8.13, "learning_rate": 4.3356173268880927e-07, "loss": 0.2153, "step": 62650 }, { "epoch": 8.13, "learning_rate": 4.329790306574219e-07, "loss": 0.2298, "step": 62660 }, { "epoch": 8.13, "learning_rate": 4.3239668060913316e-07, "loss": 0.2398, "step": 62670 }, { "epoch": 8.13, "learning_rate": 4.318146826512172e-07, "loss": 0.226, "step": 62680 }, { "epoch": 8.14, "learning_rate": 4.3123303689088324e-07, "loss": 0.2247, "step": 62690 }, { "epoch": 8.14, "learning_rate": 4.3065174343527766e-07, "loss": 0.2214, "step": 62700 }, { "epoch": 8.14, "learning_rate": 4.3007080239147987e-07, "loss": 0.2223, "step": 62710 }, { "epoch": 8.14, "learning_rate": 4.2949021386650406e-07, "loss": 0.2218, "step": 62720 }, { "epoch": 8.14, "learning_rate": 4.2890997796730185e-07, "loss": 0.2221, "step": 62730 }, { "epoch": 8.14, "learning_rate": 4.283300948007581e-07, "loss": 0.2278, "step": 62740 }, { "epoch": 8.14, "learning_rate": 4.277505644736923e-07, "loss": 0.2386, "step": 62750 }, { "epoch": 8.15, "learning_rate": 4.2717138709286073e-07, "loss": 0.237, "step": 62760 }, { "epoch": 8.15, "learning_rate": 4.265925627649528e-07, "loss": 0.2293, "step": 62770 }, { "epoch": 8.15, "learning_rate": 4.260140915965946e-07, "loss": 0.217, "step": 62780 }, { "epoch": 8.15, "learning_rate": 4.25435973694346e-07, "loss": 0.2206, "step": 62790 }, { "epoch": 8.15, "learning_rate": 4.248582091647015e-07, "loss": 0.2307, "step": 62800 }, { "epoch": 8.15, "learning_rate": 4.2428079811409213e-07, "loss": 0.2133, "step": 62810 }, { "epoch": 8.15, "learning_rate": 4.2370374064888195e-07, "loss": 0.2261, "step": 62820 }, { "epoch": 8.15, "learning_rate": 4.231270368753704e-07, "loss": 0.231, "step": 62830 }, { "epoch": 8.16, "learning_rate": 4.2255068689979355e-07, "loss": 0.2225, "step": 62840 }, { "epoch": 8.16, "learning_rate": 4.2197469082831953e-07, "loss": 0.2116, "step": 62850 }, { "epoch": 8.16, "learning_rate": 4.213990487670527e-07, "loss": 0.2293, "step": 62860 }, { "epoch": 8.16, "learning_rate": 4.208237608220325e-07, "loss": 0.2197, "step": 62870 }, { "epoch": 8.16, "learning_rate": 4.20248827099233e-07, "loss": 0.2138, "step": 62880 }, { "epoch": 8.16, "learning_rate": 4.196742477045614e-07, "loss": 0.2367, "step": 62890 }, { "epoch": 8.16, "learning_rate": 4.1910002274386244e-07, "loss": 0.2308, "step": 62900 }, { "epoch": 8.16, "learning_rate": 4.185261523229131e-07, "loss": 0.2496, "step": 62910 }, { "epoch": 8.17, "learning_rate": 4.179526365474267e-07, "loss": 0.234, "step": 62920 }, { "epoch": 8.17, "learning_rate": 4.173794755230506e-07, "loss": 0.2132, "step": 62930 }, { "epoch": 8.17, "learning_rate": 4.168066693553659e-07, "loss": 0.222, "step": 62940 }, { "epoch": 8.17, "learning_rate": 4.1623421814989033e-07, "loss": 0.23, "step": 62950 }, { "epoch": 8.17, "learning_rate": 4.1566212201207443e-07, "loss": 0.2179, "step": 62960 }, { "epoch": 8.17, "learning_rate": 4.1509038104730393e-07, "loss": 0.2267, "step": 62970 }, { "epoch": 8.17, "learning_rate": 4.1451899536090003e-07, "loss": 0.2133, "step": 62980 }, { "epoch": 8.18, "learning_rate": 4.1394796505811684e-07, "loss": 0.2328, "step": 62990 }, { "epoch": 8.18, "learning_rate": 4.133772902441439e-07, "loss": 0.2319, "step": 63000 }, { "epoch": 8.18, "learning_rate": 4.12806971024106e-07, "loss": 0.228, "step": 63010 }, { "epoch": 8.18, "learning_rate": 4.12237007503061e-07, "loss": 0.237, "step": 63020 }, { "epoch": 8.18, "learning_rate": 4.1166739978600123e-07, "loss": 0.2164, "step": 63030 }, { "epoch": 8.18, "learning_rate": 4.1109814797785557e-07, "loss": 0.2216, "step": 63040 }, { "epoch": 8.18, "learning_rate": 4.1052925218348486e-07, "loss": 0.2378, "step": 63050 }, { "epoch": 8.18, "learning_rate": 4.0996071250768597e-07, "loss": 0.2242, "step": 63060 }, { "epoch": 8.19, "learning_rate": 4.0939252905518955e-07, "loss": 0.2394, "step": 63070 }, { "epoch": 8.19, "learning_rate": 4.0882470193065975e-07, "loss": 0.2269, "step": 63080 }, { "epoch": 8.19, "learning_rate": 4.082572312386972e-07, "loss": 0.2164, "step": 63090 }, { "epoch": 8.19, "learning_rate": 4.076901170838351e-07, "loss": 0.2236, "step": 63100 }, { "epoch": 8.19, "learning_rate": 4.071233595705413e-07, "loss": 0.2344, "step": 63110 }, { "epoch": 8.19, "learning_rate": 4.0655695880321883e-07, "loss": 0.219, "step": 63120 }, { "epoch": 8.19, "learning_rate": 4.059909148862043e-07, "loss": 0.2241, "step": 63130 }, { "epoch": 8.19, "learning_rate": 4.0542522792376756e-07, "loss": 0.2129, "step": 63140 }, { "epoch": 8.2, "learning_rate": 4.048598980201152e-07, "loss": 0.2221, "step": 63150 }, { "epoch": 8.2, "learning_rate": 4.0429492527938635e-07, "loss": 0.2249, "step": 63160 }, { "epoch": 8.2, "learning_rate": 4.037303098056536e-07, "loss": 0.2217, "step": 63170 }, { "epoch": 8.2, "learning_rate": 4.031660517029261e-07, "loss": 0.2267, "step": 63180 }, { "epoch": 8.2, "learning_rate": 4.0260215107514515e-07, "loss": 0.2139, "step": 63190 }, { "epoch": 8.2, "learning_rate": 4.0203860802618645e-07, "loss": 0.2235, "step": 63200 }, { "epoch": 8.2, "learning_rate": 4.0147542265986125e-07, "loss": 0.2283, "step": 63210 }, { "epoch": 8.21, "learning_rate": 4.009125950799129e-07, "loss": 0.2184, "step": 63220 }, { "epoch": 8.21, "learning_rate": 4.003501253900212e-07, "loss": 0.2334, "step": 63230 }, { "epoch": 8.21, "learning_rate": 3.9978801369379775e-07, "loss": 0.2266, "step": 63240 }, { "epoch": 8.21, "learning_rate": 3.992262600947885e-07, "loss": 0.2032, "step": 63250 }, { "epoch": 8.21, "learning_rate": 3.9866486469647566e-07, "loss": 0.2332, "step": 63260 }, { "epoch": 8.21, "learning_rate": 3.9810382760227315e-07, "loss": 0.2341, "step": 63270 }, { "epoch": 8.21, "learning_rate": 3.975431489155288e-07, "loss": 0.2297, "step": 63280 }, { "epoch": 8.21, "learning_rate": 3.9698282873952635e-07, "loss": 0.2143, "step": 63290 }, { "epoch": 8.22, "learning_rate": 3.964228671774822e-07, "loss": 0.2207, "step": 63300 }, { "epoch": 8.22, "learning_rate": 3.9586326433254596e-07, "loss": 0.2362, "step": 63310 }, { "epoch": 8.22, "learning_rate": 3.9530402030780327e-07, "loss": 0.2326, "step": 63320 }, { "epoch": 8.22, "learning_rate": 3.947451352062721e-07, "loss": 0.2383, "step": 63330 }, { "epoch": 8.22, "learning_rate": 3.9418660913090385e-07, "loss": 0.2161, "step": 63340 }, { "epoch": 8.22, "learning_rate": 3.936284421845858e-07, "loss": 0.2267, "step": 63350 }, { "epoch": 8.22, "learning_rate": 3.930706344701368e-07, "loss": 0.2296, "step": 63360 }, { "epoch": 8.22, "learning_rate": 3.9251318609031153e-07, "loss": 0.2405, "step": 63370 }, { "epoch": 8.23, "learning_rate": 3.9195609714779714e-07, "loss": 0.2261, "step": 63380 }, { "epoch": 8.23, "learning_rate": 3.9139936774521446e-07, "loss": 0.2373, "step": 63390 }, { "epoch": 8.23, "learning_rate": 3.9084299798511995e-07, "loss": 0.2211, "step": 63400 }, { "epoch": 8.23, "learning_rate": 3.9028698797000144e-07, "loss": 0.221, "step": 63410 }, { "epoch": 8.23, "learning_rate": 3.897313378022812e-07, "loss": 0.2252, "step": 63420 }, { "epoch": 8.23, "learning_rate": 3.891760475843166e-07, "loss": 0.2378, "step": 63430 }, { "epoch": 8.23, "learning_rate": 3.886211174183972e-07, "loss": 0.2256, "step": 63440 }, { "epoch": 8.23, "learning_rate": 3.8806654740674594e-07, "loss": 0.2198, "step": 63450 }, { "epoch": 8.24, "learning_rate": 3.875123376515217e-07, "loss": 0.2333, "step": 63460 }, { "epoch": 8.24, "learning_rate": 3.8695848825481423e-07, "loss": 0.2405, "step": 63470 }, { "epoch": 8.24, "learning_rate": 3.864049993186481e-07, "loss": 0.2238, "step": 63480 }, { "epoch": 8.24, "learning_rate": 3.858518709449824e-07, "loss": 0.2314, "step": 63490 }, { "epoch": 8.24, "learning_rate": 3.852991032357078e-07, "loss": 0.2314, "step": 63500 }, { "epoch": 8.24, "learning_rate": 3.847466962926508e-07, "loss": 0.231, "step": 63510 }, { "epoch": 8.24, "learning_rate": 3.8419465021756945e-07, "loss": 0.2244, "step": 63520 }, { "epoch": 8.25, "learning_rate": 3.836429651121561e-07, "loss": 0.1989, "step": 63530 }, { "epoch": 8.25, "learning_rate": 3.830916410780374e-07, "loss": 0.2323, "step": 63540 }, { "epoch": 8.25, "learning_rate": 3.8254067821677236e-07, "loss": 0.2219, "step": 63550 }, { "epoch": 8.25, "learning_rate": 3.819900766298531e-07, "loss": 0.2222, "step": 63560 }, { "epoch": 8.25, "learning_rate": 3.814398364187073e-07, "loss": 0.2477, "step": 63570 }, { "epoch": 8.25, "learning_rate": 3.808899576846937e-07, "loss": 0.2138, "step": 63580 }, { "epoch": 8.25, "learning_rate": 3.8034044052910547e-07, "loss": 0.2382, "step": 63590 }, { "epoch": 8.25, "learning_rate": 3.7979128505316984e-07, "loss": 0.2326, "step": 63600 }, { "epoch": 8.26, "learning_rate": 3.7924249135804653e-07, "loss": 0.2157, "step": 63610 }, { "epoch": 8.26, "learning_rate": 3.7869405954482786e-07, "loss": 0.2207, "step": 63620 }, { "epoch": 8.26, "learning_rate": 3.7814598971454195e-07, "loss": 0.2107, "step": 63630 }, { "epoch": 8.26, "learning_rate": 3.775982819681475e-07, "loss": 0.2371, "step": 63640 }, { "epoch": 8.26, "learning_rate": 3.770509364065386e-07, "loss": 0.2367, "step": 63650 }, { "epoch": 8.26, "learning_rate": 3.7650395313054166e-07, "loss": 0.2217, "step": 63660 }, { "epoch": 8.26, "learning_rate": 3.7595733224091533e-07, "loss": 0.2322, "step": 63670 }, { "epoch": 8.26, "learning_rate": 3.754110738383544e-07, "loss": 0.2382, "step": 63680 }, { "epoch": 8.27, "learning_rate": 3.7486517802348404e-07, "loss": 0.2312, "step": 63690 }, { "epoch": 8.27, "learning_rate": 3.7431964489686343e-07, "loss": 0.2354, "step": 63700 }, { "epoch": 8.27, "learning_rate": 3.737744745589864e-07, "loss": 0.2089, "step": 63710 }, { "epoch": 8.27, "learning_rate": 3.7322966711027784e-07, "loss": 0.2353, "step": 63720 }, { "epoch": 8.27, "learning_rate": 3.7268522265109677e-07, "loss": 0.223, "step": 63730 }, { "epoch": 8.27, "learning_rate": 3.7214114128173587e-07, "loss": 0.2316, "step": 63740 }, { "epoch": 8.27, "learning_rate": 3.7159742310242e-07, "loss": 0.2353, "step": 63750 }, { "epoch": 8.28, "learning_rate": 3.710540682133067e-07, "loss": 0.2202, "step": 63760 }, { "epoch": 8.28, "learning_rate": 3.705110767144891e-07, "loss": 0.2374, "step": 63770 }, { "epoch": 8.28, "learning_rate": 3.699684487059898e-07, "loss": 0.2215, "step": 63780 }, { "epoch": 8.28, "learning_rate": 3.69426184287768e-07, "loss": 0.2163, "step": 63790 }, { "epoch": 8.28, "learning_rate": 3.688842835597134e-07, "loss": 0.2341, "step": 63800 }, { "epoch": 8.28, "learning_rate": 3.683427466216489e-07, "loss": 0.2244, "step": 63810 }, { "epoch": 8.28, "learning_rate": 3.678015735733322e-07, "loss": 0.2345, "step": 63820 }, { "epoch": 8.28, "learning_rate": 3.672607645144522e-07, "loss": 0.2156, "step": 63830 }, { "epoch": 8.29, "learning_rate": 3.667203195446309e-07, "loss": 0.2355, "step": 63840 }, { "epoch": 8.29, "learning_rate": 3.661802387634248e-07, "loss": 0.2175, "step": 63850 }, { "epoch": 8.29, "learning_rate": 3.656405222703212e-07, "loss": 0.2317, "step": 63860 }, { "epoch": 8.29, "learning_rate": 3.6510117016474115e-07, "loss": 0.228, "step": 63870 }, { "epoch": 8.29, "learning_rate": 3.6456218254603985e-07, "loss": 0.2152, "step": 63880 }, { "epoch": 8.29, "learning_rate": 3.640235595135034e-07, "loss": 0.2348, "step": 63890 }, { "epoch": 8.29, "learning_rate": 3.6348530116635095e-07, "loss": 0.2164, "step": 63900 }, { "epoch": 8.29, "learning_rate": 3.629474076037365e-07, "loss": 0.2132, "step": 63910 }, { "epoch": 8.3, "learning_rate": 3.624098789247438e-07, "loss": 0.2275, "step": 63920 }, { "epoch": 8.3, "learning_rate": 3.618727152283926e-07, "loss": 0.224, "step": 63930 }, { "epoch": 8.3, "learning_rate": 3.613359166136329e-07, "loss": 0.2289, "step": 63940 }, { "epoch": 8.3, "learning_rate": 3.6079948317934814e-07, "loss": 0.2282, "step": 63950 }, { "epoch": 8.3, "learning_rate": 3.6026341502435566e-07, "loss": 0.2256, "step": 63960 }, { "epoch": 8.3, "learning_rate": 3.5972771224740394e-07, "loss": 0.2293, "step": 63970 }, { "epoch": 8.3, "learning_rate": 3.591923749471746e-07, "loss": 0.2182, "step": 63980 }, { "epoch": 8.3, "learning_rate": 3.586574032222827e-07, "loss": 0.2156, "step": 63990 }, { "epoch": 8.31, "learning_rate": 3.5812279717127495e-07, "loss": 0.2368, "step": 64000 }, { "epoch": 8.31, "learning_rate": 3.5758855689263087e-07, "loss": 0.2313, "step": 64010 }, { "epoch": 8.31, "learning_rate": 3.570546824847637e-07, "loss": 0.2283, "step": 64020 }, { "epoch": 8.31, "learning_rate": 3.565211740460177e-07, "loss": 0.2147, "step": 64030 }, { "epoch": 8.31, "learning_rate": 3.559880316746705e-07, "loss": 0.2283, "step": 64040 }, { "epoch": 8.31, "learning_rate": 3.5545525546893277e-07, "loss": 0.215, "step": 64050 }, { "epoch": 8.31, "learning_rate": 3.549228455269465e-07, "loss": 0.2296, "step": 64060 }, { "epoch": 8.32, "learning_rate": 3.5439080194678805e-07, "loss": 0.2343, "step": 64070 }, { "epoch": 8.32, "learning_rate": 3.538591248264642e-07, "loss": 0.2191, "step": 64080 }, { "epoch": 8.32, "learning_rate": 3.5332781426391483e-07, "loss": 0.2017, "step": 64090 }, { "epoch": 8.32, "learning_rate": 3.527968703570138e-07, "loss": 0.2184, "step": 64100 }, { "epoch": 8.32, "learning_rate": 3.5226629320356586e-07, "loss": 0.2244, "step": 64110 }, { "epoch": 8.32, "learning_rate": 3.517360829013077e-07, "loss": 0.2289, "step": 64120 }, { "epoch": 8.32, "learning_rate": 3.512062395479107e-07, "loss": 0.2259, "step": 64130 }, { "epoch": 8.32, "learning_rate": 3.5067676324097703e-07, "loss": 0.2244, "step": 64140 }, { "epoch": 8.33, "learning_rate": 3.5014765407804054e-07, "loss": 0.2224, "step": 64150 }, { "epoch": 8.33, "learning_rate": 3.496189121565694e-07, "loss": 0.2177, "step": 64160 }, { "epoch": 8.33, "learning_rate": 3.490905375739631e-07, "loss": 0.235, "step": 64170 }, { "epoch": 8.33, "learning_rate": 3.4856253042755254e-07, "loss": 0.2311, "step": 64180 }, { "epoch": 8.33, "learning_rate": 3.480348908146035e-07, "loss": 0.2365, "step": 64190 }, { "epoch": 8.33, "learning_rate": 3.4750761883231136e-07, "loss": 0.2289, "step": 64200 }, { "epoch": 8.33, "learning_rate": 3.4698071457780463e-07, "loss": 0.2282, "step": 64210 }, { "epoch": 8.33, "learning_rate": 3.4645417814814537e-07, "loss": 0.2032, "step": 64220 }, { "epoch": 8.34, "learning_rate": 3.459280096403257e-07, "loss": 0.225, "step": 64230 }, { "epoch": 8.34, "learning_rate": 3.454022091512723e-07, "loss": 0.2218, "step": 64240 }, { "epoch": 8.34, "learning_rate": 3.448767767778425e-07, "loss": 0.2331, "step": 64250 }, { "epoch": 8.34, "learning_rate": 3.443517126168251e-07, "loss": 0.2053, "step": 64260 }, { "epoch": 8.34, "learning_rate": 3.4382701676494376e-07, "loss": 0.212, "step": 64270 }, { "epoch": 8.34, "learning_rate": 3.433026893188522e-07, "loss": 0.2293, "step": 64280 }, { "epoch": 8.34, "learning_rate": 3.427787303751358e-07, "loss": 0.214, "step": 64290 }, { "epoch": 8.35, "learning_rate": 3.4225514003031377e-07, "loss": 0.217, "step": 64300 }, { "epoch": 8.35, "learning_rate": 3.417319183808374e-07, "loss": 0.2331, "step": 64310 }, { "epoch": 8.35, "learning_rate": 3.412090655230881e-07, "loss": 0.2275, "step": 64320 }, { "epoch": 8.35, "learning_rate": 3.4068658155338155e-07, "loss": 0.2162, "step": 64330 }, { "epoch": 8.35, "learning_rate": 3.401644665679643e-07, "loss": 0.2147, "step": 64340 }, { "epoch": 8.35, "learning_rate": 3.396427206630142e-07, "loss": 0.2356, "step": 64350 }, { "epoch": 8.35, "learning_rate": 3.3912134393464356e-07, "loss": 0.2433, "step": 64360 }, { "epoch": 8.35, "learning_rate": 3.386003364788942e-07, "loss": 0.2258, "step": 64370 }, { "epoch": 8.36, "learning_rate": 3.3807969839174076e-07, "loss": 0.2271, "step": 64380 }, { "epoch": 8.36, "learning_rate": 3.375594297690908e-07, "loss": 0.2272, "step": 64390 }, { "epoch": 8.36, "learning_rate": 3.3703953070678255e-07, "loss": 0.2351, "step": 64400 }, { "epoch": 8.36, "learning_rate": 3.3652000130058605e-07, "loss": 0.217, "step": 64410 }, { "epoch": 8.36, "learning_rate": 3.3600084164620517e-07, "loss": 0.2265, "step": 64420 }, { "epoch": 8.36, "learning_rate": 3.35482051839273e-07, "loss": 0.2338, "step": 64430 }, { "epoch": 8.36, "learning_rate": 3.3496363197535613e-07, "loss": 0.229, "step": 64440 }, { "epoch": 8.36, "learning_rate": 3.344455821499532e-07, "loss": 0.2198, "step": 64450 }, { "epoch": 8.37, "learning_rate": 3.339279024584935e-07, "loss": 0.2358, "step": 64460 }, { "epoch": 8.37, "learning_rate": 3.334105929963394e-07, "loss": 0.2206, "step": 64470 }, { "epoch": 8.37, "learning_rate": 3.328936538587841e-07, "loss": 0.238, "step": 64480 }, { "epoch": 8.37, "learning_rate": 3.3237708514105277e-07, "loss": 0.2203, "step": 64490 }, { "epoch": 8.37, "learning_rate": 3.3186088693830323e-07, "loss": 0.2125, "step": 64500 }, { "epoch": 8.37, "learning_rate": 3.313450593456239e-07, "loss": 0.227, "step": 64510 }, { "epoch": 8.37, "learning_rate": 3.308296024580346e-07, "loss": 0.2221, "step": 64520 }, { "epoch": 8.38, "learning_rate": 3.303145163704894e-07, "loss": 0.2275, "step": 64530 }, { "epoch": 8.38, "learning_rate": 3.2979980117787085e-07, "loss": 0.2298, "step": 64540 }, { "epoch": 8.38, "learning_rate": 3.292854569749947e-07, "loss": 0.219, "step": 64550 }, { "epoch": 8.38, "learning_rate": 3.2877148385660953e-07, "loss": 0.2173, "step": 64560 }, { "epoch": 8.38, "learning_rate": 3.28257881917393e-07, "loss": 0.2215, "step": 64570 }, { "epoch": 8.38, "learning_rate": 3.277446512519559e-07, "loss": 0.2255, "step": 64580 }, { "epoch": 8.38, "learning_rate": 3.272317919548414e-07, "loss": 0.2269, "step": 64590 }, { "epoch": 8.38, "learning_rate": 3.26719304120522e-07, "loss": 0.2307, "step": 64600 }, { "epoch": 8.39, "learning_rate": 3.2620718784340433e-07, "loss": 0.2164, "step": 64610 }, { "epoch": 8.39, "learning_rate": 3.256954432178245e-07, "loss": 0.234, "step": 64620 }, { "epoch": 8.39, "learning_rate": 3.251840703380508e-07, "loss": 0.2332, "step": 64630 }, { "epoch": 8.39, "learning_rate": 3.24673069298284e-07, "loss": 0.2088, "step": 64640 }, { "epoch": 8.39, "learning_rate": 3.2416244019265534e-07, "loss": 0.2139, "step": 64650 }, { "epoch": 8.39, "learning_rate": 3.236521831152272e-07, "loss": 0.2377, "step": 64660 }, { "epoch": 8.39, "learning_rate": 3.2314229815999504e-07, "loss": 0.2169, "step": 64670 }, { "epoch": 8.39, "learning_rate": 3.226327854208842e-07, "loss": 0.2255, "step": 64680 }, { "epoch": 8.4, "learning_rate": 3.221236449917514e-07, "loss": 0.2177, "step": 64690 }, { "epoch": 8.4, "learning_rate": 3.2161487696638665e-07, "loss": 0.2326, "step": 64700 }, { "epoch": 8.4, "learning_rate": 3.211064814385097e-07, "loss": 0.2245, "step": 64710 }, { "epoch": 8.4, "learning_rate": 3.205984585017711e-07, "loss": 0.2283, "step": 64720 }, { "epoch": 8.4, "learning_rate": 3.20090808249755e-07, "loss": 0.216, "step": 64730 }, { "epoch": 8.4, "learning_rate": 3.195835307759747e-07, "loss": 0.2391, "step": 64740 }, { "epoch": 8.4, "learning_rate": 3.19076626173877e-07, "loss": 0.2171, "step": 64750 }, { "epoch": 8.4, "learning_rate": 3.185700945368378e-07, "loss": 0.2219, "step": 64760 }, { "epoch": 8.41, "learning_rate": 3.1806393595816506e-07, "loss": 0.2206, "step": 64770 }, { "epoch": 8.41, "learning_rate": 3.175581505310991e-07, "loss": 0.2272, "step": 64780 }, { "epoch": 8.41, "learning_rate": 3.170527383488104e-07, "loss": 0.2266, "step": 64790 }, { "epoch": 8.41, "learning_rate": 3.165476995044003e-07, "loss": 0.2121, "step": 64800 }, { "epoch": 8.41, "learning_rate": 3.1604303409090295e-07, "loss": 0.2303, "step": 64810 }, { "epoch": 8.41, "learning_rate": 3.1553874220128226e-07, "loss": 0.2342, "step": 64820 }, { "epoch": 8.41, "learning_rate": 3.150348239284335e-07, "loss": 0.2371, "step": 64830 }, { "epoch": 8.42, "learning_rate": 3.145312793651842e-07, "loss": 0.2279, "step": 64840 }, { "epoch": 8.42, "learning_rate": 3.140281086042919e-07, "loss": 0.2395, "step": 64850 }, { "epoch": 8.42, "learning_rate": 3.135253117384453e-07, "loss": 0.2212, "step": 64860 }, { "epoch": 8.42, "learning_rate": 3.1302288886026553e-07, "loss": 0.2152, "step": 64870 }, { "epoch": 8.42, "learning_rate": 3.125208400623026e-07, "loss": 0.2211, "step": 64880 }, { "epoch": 8.42, "learning_rate": 3.120191654370405e-07, "loss": 0.2304, "step": 64890 }, { "epoch": 8.42, "learning_rate": 3.115178650768919e-07, "loss": 0.2346, "step": 64900 }, { "epoch": 8.42, "learning_rate": 3.1101693907420056e-07, "loss": 0.2326, "step": 64910 }, { "epoch": 8.43, "learning_rate": 3.1051638752124346e-07, "loss": 0.2242, "step": 64920 }, { "epoch": 8.43, "learning_rate": 3.100162105102268e-07, "loss": 0.2284, "step": 64930 }, { "epoch": 8.43, "learning_rate": 3.095164081332873e-07, "loss": 0.2365, "step": 64940 }, { "epoch": 8.43, "learning_rate": 3.090169804824944e-07, "loss": 0.2264, "step": 64950 }, { "epoch": 8.43, "learning_rate": 3.085179276498477e-07, "loss": 0.2289, "step": 64960 }, { "epoch": 8.43, "learning_rate": 3.080192497272771e-07, "loss": 0.2329, "step": 64970 }, { "epoch": 8.43, "learning_rate": 3.075209468066446e-07, "loss": 0.2125, "step": 64980 }, { "epoch": 8.43, "learning_rate": 3.070230189797427e-07, "loss": 0.2184, "step": 64990 }, { "epoch": 8.44, "learning_rate": 3.065254663382938e-07, "loss": 0.231, "step": 65000 }, { "epoch": 8.44, "learning_rate": 3.060282889739529e-07, "loss": 0.2081, "step": 65010 }, { "epoch": 8.44, "learning_rate": 3.0553148697830527e-07, "loss": 0.2294, "step": 65020 }, { "epoch": 8.44, "learning_rate": 3.050350604428653e-07, "loss": 0.2312, "step": 65030 }, { "epoch": 8.44, "learning_rate": 3.0453900945908154e-07, "loss": 0.2327, "step": 65040 }, { "epoch": 8.44, "learning_rate": 3.0404333411833006e-07, "loss": 0.2218, "step": 65050 }, { "epoch": 8.44, "learning_rate": 3.0354803451192067e-07, "loss": 0.228, "step": 65060 }, { "epoch": 8.45, "learning_rate": 3.030531107310913e-07, "loss": 0.2169, "step": 65070 }, { "epoch": 8.45, "learning_rate": 3.0255856286701197e-07, "loss": 0.2285, "step": 65080 }, { "epoch": 8.45, "learning_rate": 3.0206439101078415e-07, "loss": 0.2213, "step": 65090 }, { "epoch": 8.45, "learning_rate": 3.015705952534388e-07, "loss": 0.2187, "step": 65100 }, { "epoch": 8.45, "learning_rate": 3.0107717568593725e-07, "loss": 0.2305, "step": 65110 }, { "epoch": 8.45, "learning_rate": 3.0058413239917356e-07, "loss": 0.2409, "step": 65120 }, { "epoch": 8.45, "learning_rate": 3.000914654839706e-07, "loss": 0.2247, "step": 65130 }, { "epoch": 8.45, "learning_rate": 2.995991750310821e-07, "loss": 0.2207, "step": 65140 }, { "epoch": 8.46, "learning_rate": 2.991072611311941e-07, "loss": 0.2329, "step": 65150 }, { "epoch": 8.46, "learning_rate": 2.986157238749211e-07, "loss": 0.2361, "step": 65160 }, { "epoch": 8.46, "learning_rate": 2.9812456335280914e-07, "loss": 0.2211, "step": 65170 }, { "epoch": 8.46, "learning_rate": 2.9763377965533557e-07, "loss": 0.2284, "step": 65180 }, { "epoch": 8.46, "learning_rate": 2.971433728729068e-07, "loss": 0.2355, "step": 65190 }, { "epoch": 8.46, "learning_rate": 2.966533430958614e-07, "loss": 0.2307, "step": 65200 }, { "epoch": 8.46, "learning_rate": 2.9616369041446764e-07, "loss": 0.2282, "step": 65210 }, { "epoch": 8.46, "learning_rate": 2.956744149189236e-07, "loss": 0.2352, "step": 65220 }, { "epoch": 8.47, "learning_rate": 2.951855166993601e-07, "loss": 0.226, "step": 65230 }, { "epoch": 8.47, "learning_rate": 2.94696995845836e-07, "loss": 0.2248, "step": 65240 }, { "epoch": 8.47, "learning_rate": 2.942088524483419e-07, "loss": 0.2268, "step": 65250 }, { "epoch": 8.47, "learning_rate": 2.9372108659679907e-07, "loss": 0.2367, "step": 65260 }, { "epoch": 8.47, "learning_rate": 2.932336983810588e-07, "loss": 0.2259, "step": 65270 }, { "epoch": 8.47, "learning_rate": 2.927466878909022e-07, "loss": 0.2355, "step": 65280 }, { "epoch": 8.47, "learning_rate": 2.9226005521604257e-07, "loss": 0.2399, "step": 65290 }, { "epoch": 8.48, "learning_rate": 2.9177380044612155e-07, "loss": 0.2143, "step": 65300 }, { "epoch": 8.48, "learning_rate": 2.9128792367071223e-07, "loss": 0.2302, "step": 65310 }, { "epoch": 8.48, "learning_rate": 2.908024249793186e-07, "loss": 0.2432, "step": 65320 }, { "epoch": 8.48, "learning_rate": 2.9031730446137354e-07, "loss": 0.2234, "step": 65330 }, { "epoch": 8.48, "learning_rate": 2.898325622062422e-07, "loss": 0.2218, "step": 65340 }, { "epoch": 8.48, "learning_rate": 2.893481983032181e-07, "loss": 0.228, "step": 65350 }, { "epoch": 8.48, "learning_rate": 2.888642128415257e-07, "loss": 0.209, "step": 65360 }, { "epoch": 8.48, "learning_rate": 2.8838060591032067e-07, "loss": 0.2209, "step": 65370 }, { "epoch": 8.49, "learning_rate": 2.87897377598688e-07, "loss": 0.2176, "step": 65380 }, { "epoch": 8.49, "learning_rate": 2.8741452799564256e-07, "loss": 0.2275, "step": 65390 }, { "epoch": 8.49, "learning_rate": 2.869320571901311e-07, "loss": 0.2211, "step": 65400 }, { "epoch": 8.49, "learning_rate": 2.8644996527102906e-07, "loss": 0.2377, "step": 65410 }, { "epoch": 8.49, "learning_rate": 2.859682523271423e-07, "loss": 0.218, "step": 65420 }, { "epoch": 8.49, "learning_rate": 2.8548691844720766e-07, "loss": 0.2328, "step": 65430 }, { "epoch": 8.49, "learning_rate": 2.8500596371989145e-07, "loss": 0.2318, "step": 65440 }, { "epoch": 8.49, "learning_rate": 2.8452538823378995e-07, "loss": 0.2302, "step": 65450 }, { "epoch": 8.5, "learning_rate": 2.840451920774308e-07, "loss": 0.2213, "step": 65460 }, { "epoch": 8.5, "learning_rate": 2.8356537533926983e-07, "loss": 0.223, "step": 65470 }, { "epoch": 8.5, "learning_rate": 2.8308593810769543e-07, "loss": 0.2194, "step": 65480 }, { "epoch": 8.5, "learning_rate": 2.8260688047102393e-07, "loss": 0.2103, "step": 65490 }, { "epoch": 8.5, "learning_rate": 2.8212820251750197e-07, "loss": 0.2247, "step": 65500 }, { "epoch": 8.5, "learning_rate": 2.816499043353079e-07, "loss": 0.2284, "step": 65510 }, { "epoch": 8.5, "learning_rate": 2.8117198601254887e-07, "loss": 0.2309, "step": 65520 }, { "epoch": 8.5, "learning_rate": 2.80694447637261e-07, "loss": 0.2213, "step": 65530 }, { "epoch": 8.51, "learning_rate": 2.8021728929741324e-07, "loss": 0.2249, "step": 65540 }, { "epoch": 8.51, "learning_rate": 2.7974051108090217e-07, "loss": 0.2379, "step": 65550 }, { "epoch": 8.51, "learning_rate": 2.792641130755547e-07, "loss": 0.2223, "step": 65560 }, { "epoch": 8.51, "learning_rate": 2.78788095369129e-07, "loss": 0.2239, "step": 65570 }, { "epoch": 8.51, "learning_rate": 2.78312458049312e-07, "loss": 0.2386, "step": 65580 }, { "epoch": 8.51, "learning_rate": 2.7783720120372e-07, "loss": 0.2278, "step": 65590 }, { "epoch": 8.51, "learning_rate": 2.773623249199013e-07, "loss": 0.2369, "step": 65600 }, { "epoch": 8.52, "learning_rate": 2.768878292853317e-07, "loss": 0.2308, "step": 65610 }, { "epoch": 8.52, "learning_rate": 2.764137143874193e-07, "loss": 0.2172, "step": 65620 }, { "epoch": 8.52, "learning_rate": 2.759399803134999e-07, "loss": 0.2319, "step": 65630 }, { "epoch": 8.52, "learning_rate": 2.754666271508398e-07, "loss": 0.2288, "step": 65640 }, { "epoch": 8.52, "learning_rate": 2.7499365498663644e-07, "loss": 0.2173, "step": 65650 }, { "epoch": 8.52, "learning_rate": 2.7452106390801545e-07, "loss": 0.2207, "step": 65660 }, { "epoch": 8.52, "learning_rate": 2.740488540020324e-07, "loss": 0.2293, "step": 65670 }, { "epoch": 8.52, "learning_rate": 2.735770253556738e-07, "loss": 0.217, "step": 65680 }, { "epoch": 8.53, "learning_rate": 2.7310557805585493e-07, "loss": 0.2186, "step": 65690 }, { "epoch": 8.53, "learning_rate": 2.726345121894204e-07, "loss": 0.221, "step": 65700 }, { "epoch": 8.53, "learning_rate": 2.721638278431467e-07, "loss": 0.2345, "step": 65710 }, { "epoch": 8.53, "learning_rate": 2.7169352510373755e-07, "loss": 0.2265, "step": 65720 }, { "epoch": 8.53, "learning_rate": 2.712236040578271e-07, "loss": 0.2293, "step": 65730 }, { "epoch": 8.53, "learning_rate": 2.7075406479198077e-07, "loss": 0.2245, "step": 65740 }, { "epoch": 8.53, "learning_rate": 2.7028490739269095e-07, "loss": 0.2338, "step": 65750 }, { "epoch": 8.53, "learning_rate": 2.698161319463824e-07, "loss": 0.2227, "step": 65760 }, { "epoch": 8.54, "learning_rate": 2.6934773853940764e-07, "loss": 0.2129, "step": 65770 }, { "epoch": 8.54, "learning_rate": 2.688797272580489e-07, "loss": 0.2375, "step": 65780 }, { "epoch": 8.54, "learning_rate": 2.684120981885195e-07, "loss": 0.225, "step": 65790 }, { "epoch": 8.54, "learning_rate": 2.6794485141696104e-07, "loss": 0.2287, "step": 65800 }, { "epoch": 8.54, "learning_rate": 2.674779870294444e-07, "loss": 0.2371, "step": 65810 }, { "epoch": 8.54, "learning_rate": 2.6701150511197173e-07, "loss": 0.2174, "step": 65820 }, { "epoch": 8.54, "learning_rate": 2.6654540575047317e-07, "loss": 0.2241, "step": 65830 }, { "epoch": 8.55, "learning_rate": 2.660796890308085e-07, "loss": 0.2245, "step": 65840 }, { "epoch": 8.55, "learning_rate": 2.656143550387681e-07, "loss": 0.2161, "step": 65850 }, { "epoch": 8.55, "learning_rate": 2.6514940386007076e-07, "loss": 0.2449, "step": 65860 }, { "epoch": 8.55, "learning_rate": 2.6468483558036477e-07, "loss": 0.2219, "step": 65870 }, { "epoch": 8.55, "learning_rate": 2.642206502852289e-07, "loss": 0.2165, "step": 65880 }, { "epoch": 8.55, "learning_rate": 2.637568480601703e-07, "loss": 0.2197, "step": 65890 }, { "epoch": 8.55, "learning_rate": 2.632934289906263e-07, "loss": 0.2361, "step": 65900 }, { "epoch": 8.55, "learning_rate": 2.628303931619632e-07, "loss": 0.2144, "step": 65910 }, { "epoch": 8.56, "learning_rate": 2.623677406594764e-07, "loss": 0.2318, "step": 65920 }, { "epoch": 8.56, "learning_rate": 2.61905471568392e-07, "loss": 0.2313, "step": 65930 }, { "epoch": 8.56, "learning_rate": 2.6144358597386393e-07, "loss": 0.221, "step": 65940 }, { "epoch": 8.56, "learning_rate": 2.609820839609758e-07, "loss": 0.2153, "step": 65950 }, { "epoch": 8.56, "learning_rate": 2.6052096561474206e-07, "loss": 0.2358, "step": 65960 }, { "epoch": 8.56, "learning_rate": 2.600602310201042e-07, "loss": 0.2239, "step": 65970 }, { "epoch": 8.56, "learning_rate": 2.5959988026193447e-07, "loss": 0.2207, "step": 65980 }, { "epoch": 8.56, "learning_rate": 2.5913991342503444e-07, "loss": 0.22, "step": 65990 }, { "epoch": 8.57, "learning_rate": 2.5868033059413434e-07, "loss": 0.2062, "step": 66000 }, { "epoch": 8.57, "learning_rate": 2.582211318538933e-07, "loss": 0.2381, "step": 66010 }, { "epoch": 8.57, "learning_rate": 2.577623172889014e-07, "loss": 0.2267, "step": 66020 }, { "epoch": 8.57, "learning_rate": 2.573038869836765e-07, "loss": 0.2278, "step": 66030 }, { "epoch": 8.57, "learning_rate": 2.5684584102266525e-07, "loss": 0.2172, "step": 66040 }, { "epoch": 8.57, "learning_rate": 2.5638817949024567e-07, "loss": 0.2262, "step": 66050 }, { "epoch": 8.57, "learning_rate": 2.559309024707223e-07, "loss": 0.2235, "step": 66060 }, { "epoch": 8.57, "learning_rate": 2.5547401004833094e-07, "loss": 0.2233, "step": 66070 }, { "epoch": 8.58, "learning_rate": 2.550175023072358e-07, "loss": 0.2227, "step": 66080 }, { "epoch": 8.58, "learning_rate": 2.545613793315292e-07, "loss": 0.2229, "step": 66090 }, { "epoch": 8.58, "learning_rate": 2.5410564120523487e-07, "loss": 0.2235, "step": 66100 }, { "epoch": 8.58, "learning_rate": 2.5365028801230356e-07, "loss": 0.214, "step": 66110 }, { "epoch": 8.58, "learning_rate": 2.531953198366156e-07, "loss": 0.2257, "step": 66120 }, { "epoch": 8.58, "learning_rate": 2.5274073676198125e-07, "loss": 0.2407, "step": 66130 }, { "epoch": 8.58, "learning_rate": 2.5228653887213935e-07, "loss": 0.2231, "step": 66140 }, { "epoch": 8.59, "learning_rate": 2.5183272625075673e-07, "loss": 0.2129, "step": 66150 }, { "epoch": 8.59, "learning_rate": 2.5137929898143147e-07, "loss": 0.2371, "step": 66160 }, { "epoch": 8.59, "learning_rate": 2.5092625714768884e-07, "loss": 0.224, "step": 66170 }, { "epoch": 8.59, "learning_rate": 2.504736008329832e-07, "loss": 0.2362, "step": 66180 }, { "epoch": 8.59, "learning_rate": 2.5002133012069914e-07, "loss": 0.2334, "step": 66190 }, { "epoch": 8.59, "learning_rate": 2.495694450941485e-07, "loss": 0.2163, "step": 66200 }, { "epoch": 8.59, "learning_rate": 2.491179458365742e-07, "loss": 0.2316, "step": 66210 }, { "epoch": 8.59, "learning_rate": 2.4866683243114656e-07, "loss": 0.2161, "step": 66220 }, { "epoch": 8.6, "learning_rate": 2.4821610496096437e-07, "loss": 0.2348, "step": 66230 }, { "epoch": 8.6, "learning_rate": 2.4776576350905704e-07, "loss": 0.2192, "step": 66240 }, { "epoch": 8.6, "learning_rate": 2.473158081583817e-07, "loss": 0.2272, "step": 66250 }, { "epoch": 8.6, "learning_rate": 2.468662389918242e-07, "loss": 0.2243, "step": 66260 }, { "epoch": 8.6, "learning_rate": 2.4641705609220014e-07, "loss": 0.2463, "step": 66270 }, { "epoch": 8.6, "learning_rate": 2.459682595422536e-07, "loss": 0.237, "step": 66280 }, { "epoch": 8.6, "learning_rate": 2.4551984942465676e-07, "loss": 0.2321, "step": 66290 }, { "epoch": 8.6, "learning_rate": 2.4507182582201224e-07, "loss": 0.2336, "step": 66300 }, { "epoch": 8.61, "learning_rate": 2.446241888168496e-07, "loss": 0.2312, "step": 66310 }, { "epoch": 8.61, "learning_rate": 2.441769384916279e-07, "loss": 0.2185, "step": 66320 }, { "epoch": 8.61, "learning_rate": 2.437300749287362e-07, "loss": 0.2393, "step": 66330 }, { "epoch": 8.61, "learning_rate": 2.432835982104898e-07, "loss": 0.2284, "step": 66340 }, { "epoch": 8.61, "learning_rate": 2.428375084191356e-07, "loss": 0.2243, "step": 66350 }, { "epoch": 8.61, "learning_rate": 2.4239180563684725e-07, "loss": 0.2263, "step": 66360 }, { "epoch": 8.61, "learning_rate": 2.4194648994572694e-07, "loss": 0.2216, "step": 66370 }, { "epoch": 8.62, "learning_rate": 2.4150156142780737e-07, "loss": 0.2199, "step": 66380 }, { "epoch": 8.62, "learning_rate": 2.410570201650481e-07, "loss": 0.241, "step": 66390 }, { "epoch": 8.62, "learning_rate": 2.4061286623933816e-07, "loss": 0.245, "step": 66400 }, { "epoch": 8.62, "learning_rate": 2.401690997324956e-07, "loss": 0.227, "step": 66410 }, { "epoch": 8.62, "learning_rate": 2.397257207262661e-07, "loss": 0.2347, "step": 66420 }, { "epoch": 8.62, "learning_rate": 2.3928272930232444e-07, "loss": 0.2361, "step": 66430 }, { "epoch": 8.62, "learning_rate": 2.3884012554227496e-07, "loss": 0.2249, "step": 66440 }, { "epoch": 8.62, "learning_rate": 2.3839790952764875e-07, "loss": 0.2369, "step": 66450 }, { "epoch": 8.63, "learning_rate": 2.3795608133990643e-07, "loss": 0.233, "step": 66460 }, { "epoch": 8.63, "learning_rate": 2.3751464106043783e-07, "loss": 0.2258, "step": 66470 }, { "epoch": 8.63, "learning_rate": 2.3707358877055994e-07, "loss": 0.2093, "step": 66480 }, { "epoch": 8.63, "learning_rate": 2.3663292455152003e-07, "loss": 0.2431, "step": 66490 }, { "epoch": 8.63, "learning_rate": 2.3619264848449196e-07, "loss": 0.2356, "step": 66500 }, { "epoch": 8.63, "learning_rate": 2.3575276065057854e-07, "loss": 0.2268, "step": 66510 }, { "epoch": 8.63, "learning_rate": 2.353132611308129e-07, "loss": 0.2076, "step": 66520 }, { "epoch": 8.63, "learning_rate": 2.3487415000615447e-07, "loss": 0.2183, "step": 66530 }, { "epoch": 8.64, "learning_rate": 2.344354273574914e-07, "loss": 0.2341, "step": 66540 }, { "epoch": 8.64, "learning_rate": 2.3399709326564191e-07, "loss": 0.225, "step": 66550 }, { "epoch": 8.64, "learning_rate": 2.3355914781135074e-07, "loss": 0.2164, "step": 66560 }, { "epoch": 8.64, "learning_rate": 2.3312159107529166e-07, "loss": 0.2103, "step": 66570 }, { "epoch": 8.64, "learning_rate": 2.3268442313806763e-07, "loss": 0.2208, "step": 66580 }, { "epoch": 8.64, "learning_rate": 2.322476440802093e-07, "loss": 0.2202, "step": 66590 }, { "epoch": 8.64, "learning_rate": 2.3181125398217494e-07, "loss": 0.2154, "step": 66600 }, { "epoch": 8.65, "learning_rate": 2.3137525292435287e-07, "loss": 0.2311, "step": 66610 }, { "epoch": 8.65, "learning_rate": 2.3093964098705826e-07, "loss": 0.246, "step": 66620 }, { "epoch": 8.65, "learning_rate": 2.305044182505358e-07, "loss": 0.2345, "step": 66630 }, { "epoch": 8.65, "learning_rate": 2.3006958479495725e-07, "loss": 0.21, "step": 66640 }, { "epoch": 8.65, "learning_rate": 2.2963514070042317e-07, "loss": 0.2253, "step": 66650 }, { "epoch": 8.65, "learning_rate": 2.2920108604696325e-07, "loss": 0.229, "step": 66660 }, { "epoch": 8.65, "learning_rate": 2.287674209145344e-07, "loss": 0.2165, "step": 66670 }, { "epoch": 8.65, "learning_rate": 2.283341453830214e-07, "loss": 0.2302, "step": 66680 }, { "epoch": 8.66, "learning_rate": 2.2790125953223907e-07, "loss": 0.2283, "step": 66690 }, { "epoch": 8.66, "learning_rate": 2.2746876344192853e-07, "loss": 0.2268, "step": 66700 }, { "epoch": 8.66, "learning_rate": 2.2703665719175965e-07, "loss": 0.2305, "step": 66710 }, { "epoch": 8.66, "learning_rate": 2.2660494086133158e-07, "loss": 0.2271, "step": 66720 }, { "epoch": 8.66, "learning_rate": 2.2617361453017047e-07, "loss": 0.2245, "step": 66730 }, { "epoch": 8.66, "learning_rate": 2.2574267827773028e-07, "loss": 0.2299, "step": 66740 }, { "epoch": 8.66, "learning_rate": 2.253121321833946e-07, "loss": 0.2216, "step": 66750 }, { "epoch": 8.66, "learning_rate": 2.248819763264737e-07, "loss": 0.2217, "step": 66760 }, { "epoch": 8.67, "learning_rate": 2.244522107862073e-07, "loss": 0.2313, "step": 66770 }, { "epoch": 8.67, "learning_rate": 2.2402283564176203e-07, "loss": 0.2381, "step": 66780 }, { "epoch": 8.67, "learning_rate": 2.2359385097223272e-07, "loss": 0.2313, "step": 66790 }, { "epoch": 8.67, "learning_rate": 2.2316525685664365e-07, "loss": 0.234, "step": 66800 }, { "epoch": 8.67, "learning_rate": 2.2273705337394546e-07, "loss": 0.2342, "step": 66810 }, { "epoch": 8.67, "learning_rate": 2.2230924060301698e-07, "loss": 0.2392, "step": 66820 }, { "epoch": 8.67, "learning_rate": 2.2188181862266676e-07, "loss": 0.2338, "step": 66830 }, { "epoch": 8.67, "learning_rate": 2.2145478751162978e-07, "loss": 0.2348, "step": 66840 }, { "epoch": 8.68, "learning_rate": 2.2102814734856884e-07, "loss": 0.2122, "step": 66850 }, { "epoch": 8.68, "learning_rate": 2.206018982120761e-07, "loss": 0.2167, "step": 66860 }, { "epoch": 8.68, "learning_rate": 2.201760401806707e-07, "loss": 0.2261, "step": 66870 }, { "epoch": 8.68, "learning_rate": 2.1975057333279933e-07, "loss": 0.2217, "step": 66880 }, { "epoch": 8.68, "learning_rate": 2.1932549774683834e-07, "loss": 0.237, "step": 66890 }, { "epoch": 8.68, "learning_rate": 2.1890081350109018e-07, "loss": 0.2416, "step": 66900 }, { "epoch": 8.68, "learning_rate": 2.184765206737855e-07, "loss": 0.2246, "step": 66910 }, { "epoch": 8.69, "learning_rate": 2.1805261934308444e-07, "loss": 0.211, "step": 66920 }, { "epoch": 8.69, "learning_rate": 2.1762910958707273e-07, "loss": 0.2195, "step": 66930 }, { "epoch": 8.69, "learning_rate": 2.1720599148376607e-07, "loss": 0.2289, "step": 66940 }, { "epoch": 8.69, "learning_rate": 2.1678326511110655e-07, "loss": 0.2223, "step": 66950 }, { "epoch": 8.69, "learning_rate": 2.163609305469642e-07, "loss": 0.2258, "step": 66960 }, { "epoch": 8.69, "learning_rate": 2.159389878691382e-07, "loss": 0.2219, "step": 66970 }, { "epoch": 8.69, "learning_rate": 2.1551743715535413e-07, "loss": 0.2279, "step": 66980 }, { "epoch": 8.69, "learning_rate": 2.150962784832653e-07, "loss": 0.2269, "step": 66990 }, { "epoch": 8.7, "learning_rate": 2.146755119304541e-07, "loss": 0.2231, "step": 67000 }, { "epoch": 8.7, "learning_rate": 2.1425513757442995e-07, "loss": 0.2263, "step": 67010 }, { "epoch": 8.7, "learning_rate": 2.1383515549262906e-07, "loss": 0.2358, "step": 67020 }, { "epoch": 8.7, "learning_rate": 2.1341556576241766e-07, "loss": 0.2184, "step": 67030 }, { "epoch": 8.7, "learning_rate": 2.129963684610876e-07, "loss": 0.2268, "step": 67040 }, { "epoch": 8.7, "learning_rate": 2.1257756366585884e-07, "loss": 0.2296, "step": 67050 }, { "epoch": 8.7, "learning_rate": 2.1215915145388036e-07, "loss": 0.2237, "step": 67060 }, { "epoch": 8.7, "learning_rate": 2.1174113190222724e-07, "loss": 0.2208, "step": 67070 }, { "epoch": 8.71, "learning_rate": 2.1132350508790322e-07, "loss": 0.2206, "step": 67080 }, { "epoch": 8.71, "learning_rate": 2.1090627108783898e-07, "loss": 0.2197, "step": 67090 }, { "epoch": 8.71, "learning_rate": 2.1048942997889307e-07, "loss": 0.2373, "step": 67100 }, { "epoch": 8.71, "learning_rate": 2.1007298183785253e-07, "loss": 0.2272, "step": 67110 }, { "epoch": 8.71, "learning_rate": 2.096569267414307e-07, "loss": 0.2223, "step": 67120 }, { "epoch": 8.71, "learning_rate": 2.0924126476626864e-07, "loss": 0.2292, "step": 67130 }, { "epoch": 8.71, "learning_rate": 2.088259959889366e-07, "loss": 0.2179, "step": 67140 }, { "epoch": 8.72, "learning_rate": 2.0841112048593042e-07, "loss": 0.2302, "step": 67150 }, { "epoch": 8.72, "learning_rate": 2.0799663833367414e-07, "loss": 0.2086, "step": 67160 }, { "epoch": 8.72, "learning_rate": 2.0758254960852048e-07, "loss": 0.2355, "step": 67170 }, { "epoch": 8.72, "learning_rate": 2.0716885438674772e-07, "loss": 0.2301, "step": 67180 }, { "epoch": 8.72, "learning_rate": 2.0675555274456292e-07, "loss": 0.2221, "step": 67190 }, { "epoch": 8.72, "learning_rate": 2.0634264475810096e-07, "loss": 0.2287, "step": 67200 }, { "epoch": 8.72, "learning_rate": 2.0593013050342264e-07, "loss": 0.2267, "step": 67210 }, { "epoch": 8.72, "learning_rate": 2.0551801005651794e-07, "loss": 0.2225, "step": 67220 }, { "epoch": 8.73, "learning_rate": 2.051062834933035e-07, "loss": 0.221, "step": 67230 }, { "epoch": 8.73, "learning_rate": 2.0469495088962306e-07, "loss": 0.2136, "step": 67240 }, { "epoch": 8.73, "learning_rate": 2.0428401232124857e-07, "loss": 0.2315, "step": 67250 }, { "epoch": 8.73, "learning_rate": 2.0387346786387912e-07, "loss": 0.2025, "step": 67260 }, { "epoch": 8.73, "learning_rate": 2.0346331759314013e-07, "loss": 0.2262, "step": 67270 }, { "epoch": 8.73, "learning_rate": 2.030535615845868e-07, "loss": 0.2441, "step": 67280 }, { "epoch": 8.73, "learning_rate": 2.0264419991369963e-07, "loss": 0.2179, "step": 67290 }, { "epoch": 8.73, "learning_rate": 2.0223523265588655e-07, "loss": 0.2313, "step": 67300 }, { "epoch": 8.74, "learning_rate": 2.018266598864844e-07, "loss": 0.2264, "step": 67310 }, { "epoch": 8.74, "learning_rate": 2.0141848168075594e-07, "loss": 0.2187, "step": 67320 }, { "epoch": 8.74, "learning_rate": 2.0101069811389098e-07, "loss": 0.2398, "step": 67330 }, { "epoch": 8.74, "learning_rate": 2.0060330926100886e-07, "loss": 0.2437, "step": 67340 }, { "epoch": 8.74, "learning_rate": 2.0019631519715296e-07, "loss": 0.2104, "step": 67350 }, { "epoch": 8.74, "learning_rate": 1.9978971599729716e-07, "loss": 0.2346, "step": 67360 }, { "epoch": 8.74, "learning_rate": 1.9938351173634039e-07, "loss": 0.2101, "step": 67370 }, { "epoch": 8.74, "learning_rate": 1.9897770248910922e-07, "loss": 0.2376, "step": 67380 }, { "epoch": 8.75, "learning_rate": 1.9857228833035852e-07, "loss": 0.2212, "step": 67390 }, { "epoch": 8.75, "learning_rate": 1.9816726933476936e-07, "loss": 0.2141, "step": 67400 }, { "epoch": 8.75, "learning_rate": 1.9776264557694954e-07, "loss": 0.2282, "step": 67410 }, { "epoch": 8.75, "learning_rate": 1.9735841713143579e-07, "loss": 0.243, "step": 67420 }, { "epoch": 8.75, "learning_rate": 1.9695458407269094e-07, "loss": 0.2179, "step": 67430 }, { "epoch": 8.75, "learning_rate": 1.965511464751042e-07, "loss": 0.226, "step": 67440 }, { "epoch": 8.75, "learning_rate": 1.9614810441299367e-07, "loss": 0.2207, "step": 67450 }, { "epoch": 8.76, "learning_rate": 1.9574545796060365e-07, "loss": 0.2337, "step": 67460 }, { "epoch": 8.76, "learning_rate": 1.9534320719210502e-07, "loss": 0.2376, "step": 67470 }, { "epoch": 8.76, "learning_rate": 1.949413521815971e-07, "loss": 0.2294, "step": 67480 }, { "epoch": 8.76, "learning_rate": 1.9453989300310504e-07, "loss": 0.2227, "step": 67490 }, { "epoch": 8.76, "learning_rate": 1.9413882973058217e-07, "loss": 0.23, "step": 67500 }, { "epoch": 8.76, "learning_rate": 1.9373816243790798e-07, "loss": 0.2059, "step": 67510 }, { "epoch": 8.76, "learning_rate": 1.9333789119888925e-07, "loss": 0.2263, "step": 67520 }, { "epoch": 8.76, "learning_rate": 1.9293801608726055e-07, "loss": 0.2296, "step": 67530 }, { "epoch": 8.77, "learning_rate": 1.9253853717668246e-07, "loss": 0.2309, "step": 67540 }, { "epoch": 8.77, "learning_rate": 1.9213945454074274e-07, "loss": 0.2339, "step": 67550 }, { "epoch": 8.77, "learning_rate": 1.9174076825295678e-07, "loss": 0.2327, "step": 67560 }, { "epoch": 8.77, "learning_rate": 1.9134247838676663e-07, "loss": 0.2306, "step": 67570 }, { "epoch": 8.77, "learning_rate": 1.9094458501554096e-07, "loss": 0.2242, "step": 67580 }, { "epoch": 8.77, "learning_rate": 1.905470882125758e-07, "loss": 0.2198, "step": 67590 }, { "epoch": 8.77, "learning_rate": 1.9014998805109435e-07, "loss": 0.218, "step": 67600 }, { "epoch": 8.77, "learning_rate": 1.8975328460424569e-07, "loss": 0.2143, "step": 67610 }, { "epoch": 8.78, "learning_rate": 1.893569779451075e-07, "loss": 0.2397, "step": 67620 }, { "epoch": 8.78, "learning_rate": 1.8896106814668237e-07, "loss": 0.2226, "step": 67630 }, { "epoch": 8.78, "learning_rate": 1.8856555528190204e-07, "loss": 0.2221, "step": 67640 }, { "epoch": 8.78, "learning_rate": 1.881704394236231e-07, "loss": 0.2191, "step": 67650 }, { "epoch": 8.78, "learning_rate": 1.8777572064462973e-07, "loss": 0.2311, "step": 67660 }, { "epoch": 8.78, "learning_rate": 1.873813990176336e-07, "loss": 0.2387, "step": 67670 }, { "epoch": 8.78, "learning_rate": 1.8698747461527267e-07, "loss": 0.22, "step": 67680 }, { "epoch": 8.79, "learning_rate": 1.8659394751011133e-07, "loss": 0.2221, "step": 67690 }, { "epoch": 8.79, "learning_rate": 1.8620081777464155e-07, "loss": 0.2222, "step": 67700 }, { "epoch": 8.79, "learning_rate": 1.8580808548128175e-07, "loss": 0.2201, "step": 67710 }, { "epoch": 8.79, "learning_rate": 1.854157507023766e-07, "loss": 0.2125, "step": 67720 }, { "epoch": 8.79, "learning_rate": 1.8502381351019882e-07, "loss": 0.2421, "step": 67730 }, { "epoch": 8.79, "learning_rate": 1.8463227397694686e-07, "loss": 0.2306, "step": 67740 }, { "epoch": 8.79, "learning_rate": 1.8424113217474588e-07, "loss": 0.2264, "step": 67750 }, { "epoch": 8.79, "learning_rate": 1.8385038817564888e-07, "loss": 0.2297, "step": 67760 }, { "epoch": 8.8, "learning_rate": 1.8346004205163372e-07, "loss": 0.2252, "step": 67770 }, { "epoch": 8.8, "learning_rate": 1.8307009387460747e-07, "loss": 0.2095, "step": 67780 }, { "epoch": 8.8, "learning_rate": 1.826805437164015e-07, "loss": 0.2384, "step": 67790 }, { "epoch": 8.8, "learning_rate": 1.8229139164877447e-07, "loss": 0.222, "step": 67800 }, { "epoch": 8.8, "learning_rate": 1.8190263774341312e-07, "loss": 0.2074, "step": 67810 }, { "epoch": 8.8, "learning_rate": 1.815142820719294e-07, "loss": 0.2316, "step": 67820 }, { "epoch": 8.8, "learning_rate": 1.8112632470586195e-07, "loss": 0.2196, "step": 67830 }, { "epoch": 8.8, "learning_rate": 1.8073876571667702e-07, "loss": 0.2267, "step": 67840 }, { "epoch": 8.81, "learning_rate": 1.8035160517576621e-07, "loss": 0.2215, "step": 67850 }, { "epoch": 8.81, "learning_rate": 1.799648431544485e-07, "loss": 0.226, "step": 67860 }, { "epoch": 8.81, "learning_rate": 1.7957847972397003e-07, "loss": 0.2247, "step": 67870 }, { "epoch": 8.81, "learning_rate": 1.79192514955502e-07, "loss": 0.219, "step": 67880 }, { "epoch": 8.81, "learning_rate": 1.7880694892014273e-07, "loss": 0.214, "step": 67890 }, { "epoch": 8.81, "learning_rate": 1.7842178168891848e-07, "loss": 0.2192, "step": 67900 }, { "epoch": 8.81, "learning_rate": 1.7803701333278007e-07, "loss": 0.2267, "step": 67910 }, { "epoch": 8.82, "learning_rate": 1.7765264392260542e-07, "loss": 0.2289, "step": 67920 }, { "epoch": 8.82, "learning_rate": 1.7726867352920014e-07, "loss": 0.2321, "step": 67930 }, { "epoch": 8.82, "learning_rate": 1.7688510222329436e-07, "loss": 0.2321, "step": 67940 }, { "epoch": 8.82, "learning_rate": 1.7650193007554696e-07, "loss": 0.2162, "step": 67950 }, { "epoch": 8.82, "learning_rate": 1.7611915715654125e-07, "loss": 0.244, "step": 67960 }, { "epoch": 8.82, "learning_rate": 1.7573678353678756e-07, "loss": 0.238, "step": 67970 }, { "epoch": 8.82, "learning_rate": 1.7535480928672377e-07, "loss": 0.2389, "step": 67980 }, { "epoch": 8.82, "learning_rate": 1.749732344767129e-07, "loss": 0.2154, "step": 67990 }, { "epoch": 8.83, "learning_rate": 1.7459205917704456e-07, "loss": 0.2149, "step": 68000 }, { "epoch": 8.83, "learning_rate": 1.7421128345793573e-07, "loss": 0.2163, "step": 68010 }, { "epoch": 8.83, "learning_rate": 1.7383090738952877e-07, "loss": 0.2252, "step": 68020 }, { "epoch": 8.83, "learning_rate": 1.7345093104189234e-07, "loss": 0.2238, "step": 68030 }, { "epoch": 8.83, "learning_rate": 1.730713544850228e-07, "loss": 0.2185, "step": 68040 }, { "epoch": 8.83, "learning_rate": 1.7269217778884134e-07, "loss": 0.2411, "step": 68050 }, { "epoch": 8.83, "learning_rate": 1.7231340102319543e-07, "loss": 0.2236, "step": 68060 }, { "epoch": 8.83, "learning_rate": 1.7193502425786104e-07, "loss": 0.2219, "step": 68070 }, { "epoch": 8.84, "learning_rate": 1.7155704756253763e-07, "loss": 0.2323, "step": 68080 }, { "epoch": 8.84, "learning_rate": 1.7117947100685333e-07, "loss": 0.2311, "step": 68090 }, { "epoch": 8.84, "learning_rate": 1.7080229466036087e-07, "loss": 0.2151, "step": 68100 }, { "epoch": 8.84, "learning_rate": 1.7042551859253956e-07, "loss": 0.2184, "step": 68110 }, { "epoch": 8.84, "learning_rate": 1.7004914287279638e-07, "loss": 0.2272, "step": 68120 }, { "epoch": 8.84, "learning_rate": 1.6967316757046283e-07, "loss": 0.2363, "step": 68130 }, { "epoch": 8.84, "learning_rate": 1.6929759275479708e-07, "loss": 0.2338, "step": 68140 }, { "epoch": 8.84, "learning_rate": 1.689224184949844e-07, "loss": 0.2287, "step": 68150 }, { "epoch": 8.85, "learning_rate": 1.6854764486013537e-07, "loss": 0.2217, "step": 68160 }, { "epoch": 8.85, "learning_rate": 1.6817327191928648e-07, "loss": 0.2264, "step": 68170 }, { "epoch": 8.85, "learning_rate": 1.677992997414018e-07, "loss": 0.2265, "step": 68180 }, { "epoch": 8.85, "learning_rate": 1.6742572839537052e-07, "loss": 0.2274, "step": 68190 }, { "epoch": 8.85, "learning_rate": 1.6705255795000787e-07, "loss": 0.2331, "step": 68200 }, { "epoch": 8.85, "learning_rate": 1.6667978847405605e-07, "loss": 0.2215, "step": 68210 }, { "epoch": 8.85, "learning_rate": 1.663074200361822e-07, "loss": 0.2206, "step": 68220 }, { "epoch": 8.86, "learning_rate": 1.6593545270498124e-07, "loss": 0.2404, "step": 68230 }, { "epoch": 8.86, "learning_rate": 1.6556388654897285e-07, "loss": 0.2149, "step": 68240 }, { "epoch": 8.86, "learning_rate": 1.6519272163660274e-07, "loss": 0.2178, "step": 68250 }, { "epoch": 8.86, "learning_rate": 1.6482195803624414e-07, "loss": 0.2188, "step": 68260 }, { "epoch": 8.86, "learning_rate": 1.6445159581619474e-07, "loss": 0.2385, "step": 68270 }, { "epoch": 8.86, "learning_rate": 1.6408163504467884e-07, "loss": 0.2315, "step": 68280 }, { "epoch": 8.86, "learning_rate": 1.6371207578984743e-07, "loss": 0.2267, "step": 68290 }, { "epoch": 8.86, "learning_rate": 1.6334291811977708e-07, "loss": 0.2237, "step": 68300 }, { "epoch": 8.87, "learning_rate": 1.629741621024693e-07, "loss": 0.2375, "step": 68310 }, { "epoch": 8.87, "learning_rate": 1.6260580780585392e-07, "loss": 0.2221, "step": 68320 }, { "epoch": 8.87, "learning_rate": 1.6223785529778497e-07, "loss": 0.2273, "step": 68330 }, { "epoch": 8.87, "learning_rate": 1.6187030464604238e-07, "loss": 0.2315, "step": 68340 }, { "epoch": 8.87, "learning_rate": 1.6150315591833369e-07, "loss": 0.2328, "step": 68350 }, { "epoch": 8.87, "learning_rate": 1.6113640918229046e-07, "loss": 0.2313, "step": 68360 }, { "epoch": 8.87, "learning_rate": 1.6077006450547196e-07, "loss": 0.2242, "step": 68370 }, { "epoch": 8.87, "learning_rate": 1.6040412195536195e-07, "loss": 0.2177, "step": 68380 }, { "epoch": 8.88, "learning_rate": 1.6003858159937058e-07, "loss": 0.2214, "step": 68390 }, { "epoch": 8.88, "learning_rate": 1.5967344350483486e-07, "loss": 0.2127, "step": 68400 }, { "epoch": 8.88, "learning_rate": 1.5930870773901585e-07, "loss": 0.2407, "step": 68410 }, { "epoch": 8.88, "learning_rate": 1.5894437436910194e-07, "loss": 0.2355, "step": 68420 }, { "epoch": 8.88, "learning_rate": 1.5858044346220722e-07, "loss": 0.2265, "step": 68430 }, { "epoch": 8.88, "learning_rate": 1.5821691508537126e-07, "loss": 0.2304, "step": 68440 }, { "epoch": 8.88, "learning_rate": 1.5785378930555926e-07, "loss": 0.2269, "step": 68450 }, { "epoch": 8.89, "learning_rate": 1.5749106618966324e-07, "loss": 0.227, "step": 68460 }, { "epoch": 8.89, "learning_rate": 1.5712874580450012e-07, "loss": 0.2131, "step": 68470 }, { "epoch": 8.89, "learning_rate": 1.5676682821681233e-07, "loss": 0.2184, "step": 68480 }, { "epoch": 8.89, "learning_rate": 1.5640531349326995e-07, "loss": 0.2285, "step": 68490 }, { "epoch": 8.89, "learning_rate": 1.5604420170046637e-07, "loss": 0.23, "step": 68500 }, { "epoch": 8.89, "learning_rate": 1.5568349290492306e-07, "loss": 0.2214, "step": 68510 }, { "epoch": 8.89, "learning_rate": 1.553231871730856e-07, "loss": 0.2186, "step": 68520 }, { "epoch": 8.89, "learning_rate": 1.549632845713254e-07, "loss": 0.2258, "step": 68530 }, { "epoch": 8.9, "learning_rate": 1.5460378516594123e-07, "loss": 0.2255, "step": 68540 }, { "epoch": 8.9, "learning_rate": 1.5424468902315613e-07, "loss": 0.2181, "step": 68550 }, { "epoch": 8.9, "learning_rate": 1.5388599620911826e-07, "loss": 0.2323, "step": 68560 }, { "epoch": 8.9, "learning_rate": 1.5352770678990393e-07, "loss": 0.2232, "step": 68570 }, { "epoch": 8.9, "learning_rate": 1.5316982083151268e-07, "loss": 0.2195, "step": 68580 }, { "epoch": 8.9, "learning_rate": 1.5281233839987045e-07, "loss": 0.2259, "step": 68590 }, { "epoch": 8.9, "learning_rate": 1.524552595608297e-07, "loss": 0.2304, "step": 68600 }, { "epoch": 8.9, "learning_rate": 1.5209858438016783e-07, "loss": 0.2234, "step": 68610 }, { "epoch": 8.91, "learning_rate": 1.5174231292358774e-07, "loss": 0.2282, "step": 68620 }, { "epoch": 8.91, "learning_rate": 1.5138644525671843e-07, "loss": 0.2143, "step": 68630 }, { "epoch": 8.91, "learning_rate": 1.510309814451137e-07, "loss": 0.218, "step": 68640 }, { "epoch": 8.91, "learning_rate": 1.5067592155425429e-07, "loss": 0.2137, "step": 68650 }, { "epoch": 8.91, "learning_rate": 1.5032126564954562e-07, "loss": 0.2344, "step": 68660 }, { "epoch": 8.91, "learning_rate": 1.499670137963183e-07, "loss": 0.238, "step": 68670 }, { "epoch": 8.91, "learning_rate": 1.4961316605982977e-07, "loss": 0.2257, "step": 68680 }, { "epoch": 8.91, "learning_rate": 1.4925972250526222e-07, "loss": 0.2293, "step": 68690 }, { "epoch": 8.92, "learning_rate": 1.4890668319772273e-07, "loss": 0.2268, "step": 68700 }, { "epoch": 8.92, "learning_rate": 1.4855404820224543e-07, "loss": 0.2293, "step": 68710 }, { "epoch": 8.92, "learning_rate": 1.4820181758378907e-07, "loss": 0.2183, "step": 68720 }, { "epoch": 8.92, "learning_rate": 1.478499914072377e-07, "loss": 0.2293, "step": 68730 }, { "epoch": 8.92, "learning_rate": 1.4749856973740194e-07, "loss": 0.2181, "step": 68740 }, { "epoch": 8.92, "learning_rate": 1.4714755263901648e-07, "loss": 0.2187, "step": 68750 }, { "epoch": 8.92, "learning_rate": 1.467969401767421e-07, "loss": 0.2219, "step": 68760 }, { "epoch": 8.93, "learning_rate": 1.4644673241516591e-07, "loss": 0.2208, "step": 68770 }, { "epoch": 8.93, "learning_rate": 1.4609692941879884e-07, "loss": 0.2139, "step": 68780 }, { "epoch": 8.93, "learning_rate": 1.4574753125207842e-07, "loss": 0.2192, "step": 68790 }, { "epoch": 8.93, "learning_rate": 1.4539853797936742e-07, "loss": 0.2366, "step": 68800 }, { "epoch": 8.93, "learning_rate": 1.4504994966495357e-07, "loss": 0.2336, "step": 68810 }, { "epoch": 8.93, "learning_rate": 1.4470176637305106e-07, "loss": 0.2222, "step": 68820 }, { "epoch": 8.93, "learning_rate": 1.4435398816779795e-07, "loss": 0.2154, "step": 68830 }, { "epoch": 8.93, "learning_rate": 1.4400661511325832e-07, "loss": 0.215, "step": 68840 }, { "epoch": 8.94, "learning_rate": 1.4365964727342268e-07, "loss": 0.2296, "step": 68850 }, { "epoch": 8.94, "learning_rate": 1.4331308471220523e-07, "loss": 0.2196, "step": 68860 }, { "epoch": 8.94, "learning_rate": 1.429669274934461e-07, "loss": 0.2247, "step": 68870 }, { "epoch": 8.94, "learning_rate": 1.4262117568091172e-07, "loss": 0.234, "step": 68880 }, { "epoch": 8.94, "learning_rate": 1.422758293382925e-07, "loss": 0.2299, "step": 68890 }, { "epoch": 8.94, "learning_rate": 1.4193088852920455e-07, "loss": 0.2407, "step": 68900 }, { "epoch": 8.94, "learning_rate": 1.415863533171897e-07, "loss": 0.2279, "step": 68910 }, { "epoch": 8.94, "learning_rate": 1.412422237657149e-07, "loss": 0.2221, "step": 68920 }, { "epoch": 8.95, "learning_rate": 1.4089849993817163e-07, "loss": 0.2284, "step": 68930 }, { "epoch": 8.95, "learning_rate": 1.4055518189787803e-07, "loss": 0.2163, "step": 68940 }, { "epoch": 8.95, "learning_rate": 1.4021226970807614e-07, "loss": 0.2244, "step": 68950 }, { "epoch": 8.95, "learning_rate": 1.3986976343193424e-07, "loss": 0.2282, "step": 68960 }, { "epoch": 8.95, "learning_rate": 1.395276631325451e-07, "loss": 0.2373, "step": 68970 }, { "epoch": 8.95, "learning_rate": 1.3918596887292675e-07, "loss": 0.2238, "step": 68980 }, { "epoch": 8.95, "learning_rate": 1.3884468071602369e-07, "loss": 0.2177, "step": 68990 }, { "epoch": 8.96, "learning_rate": 1.3850379872470354e-07, "loss": 0.2258, "step": 69000 }, { "epoch": 8.96, "learning_rate": 1.3816332296176035e-07, "loss": 0.2264, "step": 69010 }, { "epoch": 8.96, "learning_rate": 1.3782325348991406e-07, "loss": 0.2233, "step": 69020 }, { "epoch": 8.96, "learning_rate": 1.3748359037180772e-07, "loss": 0.2357, "step": 69030 }, { "epoch": 8.96, "learning_rate": 1.3714433367001113e-07, "loss": 0.2252, "step": 69040 }, { "epoch": 8.96, "learning_rate": 1.3680548344701904e-07, "loss": 0.2247, "step": 69050 }, { "epoch": 8.96, "learning_rate": 1.3646703976525055e-07, "loss": 0.2072, "step": 69060 }, { "epoch": 8.96, "learning_rate": 1.361290026870506e-07, "loss": 0.2192, "step": 69070 }, { "epoch": 8.97, "learning_rate": 1.3579137227468913e-07, "loss": 0.2211, "step": 69080 }, { "epoch": 8.97, "learning_rate": 1.3545414859036044e-07, "loss": 0.2247, "step": 69090 }, { "epoch": 8.97, "learning_rate": 1.3511733169618566e-07, "loss": 0.2219, "step": 69100 }, { "epoch": 8.97, "learning_rate": 1.347809216542089e-07, "loss": 0.2195, "step": 69110 }, { "epoch": 8.97, "learning_rate": 1.3444491852640012e-07, "loss": 0.2229, "step": 69120 }, { "epoch": 8.97, "learning_rate": 1.3410932237465514e-07, "loss": 0.2285, "step": 69130 }, { "epoch": 8.97, "learning_rate": 1.3377413326079376e-07, "loss": 0.2148, "step": 69140 }, { "epoch": 8.97, "learning_rate": 1.3343935124656085e-07, "loss": 0.2152, "step": 69150 }, { "epoch": 8.98, "learning_rate": 1.3310497639362743e-07, "loss": 0.2202, "step": 69160 }, { "epoch": 8.98, "learning_rate": 1.327710087635882e-07, "loss": 0.2229, "step": 69170 }, { "epoch": 8.98, "learning_rate": 1.3243744841796274e-07, "loss": 0.2144, "step": 69180 }, { "epoch": 8.98, "learning_rate": 1.3210429541819744e-07, "loss": 0.2161, "step": 69190 }, { "epoch": 8.98, "learning_rate": 1.3177154982566143e-07, "loss": 0.2211, "step": 69200 }, { "epoch": 8.98, "learning_rate": 1.3143921170164997e-07, "loss": 0.2374, "step": 69210 }, { "epoch": 8.98, "learning_rate": 1.3110728110738363e-07, "loss": 0.226, "step": 69220 }, { "epoch": 8.99, "learning_rate": 1.3077575810400646e-07, "loss": 0.2303, "step": 69230 }, { "epoch": 8.99, "learning_rate": 1.3044464275258885e-07, "loss": 0.223, "step": 69240 }, { "epoch": 8.99, "learning_rate": 1.301139351141258e-07, "loss": 0.2178, "step": 69250 }, { "epoch": 8.99, "learning_rate": 1.2978363524953627e-07, "loss": 0.2117, "step": 69260 }, { "epoch": 8.99, "learning_rate": 1.2945374321966535e-07, "loss": 0.2158, "step": 69270 }, { "epoch": 8.99, "learning_rate": 1.291242590852821e-07, "loss": 0.2393, "step": 69280 }, { "epoch": 8.99, "learning_rate": 1.2879518290708074e-07, "loss": 0.2203, "step": 69290 }, { "epoch": 8.99, "learning_rate": 1.2846651474568097e-07, "loss": 0.2192, "step": 69300 }, { "epoch": 9.0, "learning_rate": 1.2813825466162628e-07, "loss": 0.2306, "step": 69310 }, { "epoch": 9.0, "learning_rate": 1.2781040271538524e-07, "loss": 0.2363, "step": 69320 }, { "epoch": 9.0, "learning_rate": 1.2748295896735224e-07, "loss": 0.2057, "step": 69330 }, { "epoch": 9.0, "learning_rate": 1.2715592347784494e-07, "loss": 0.2177, "step": 69340 }, { "epoch": 9.0, "eval_loss": 0.22539588809013367, "eval_runtime": 61.1031, "eval_samples_per_second": 178.01, "eval_steps_per_second": 22.257, "step": 69345 }, { "epoch": 9.0, "learning_rate": 1.2682929630710653e-07, "loss": 0.2338, "step": 69350 }, { "epoch": 9.0, "learning_rate": 1.2650307751530556e-07, "loss": 0.2208, "step": 69360 }, { "epoch": 9.0, "learning_rate": 1.2617726716253434e-07, "loss": 0.2302, "step": 69370 }, { "epoch": 9.0, "learning_rate": 1.258518653088107e-07, "loss": 0.2413, "step": 69380 }, { "epoch": 9.01, "learning_rate": 1.2552687201407684e-07, "loss": 0.2337, "step": 69390 }, { "epoch": 9.01, "learning_rate": 1.2520228733819918e-07, "loss": 0.2148, "step": 69400 }, { "epoch": 9.01, "learning_rate": 1.248781113409705e-07, "loss": 0.225, "step": 69410 }, { "epoch": 9.01, "learning_rate": 1.245543440821066e-07, "loss": 0.2364, "step": 69420 }, { "epoch": 9.01, "learning_rate": 1.2423098562124858e-07, "loss": 0.2407, "step": 69430 }, { "epoch": 9.01, "learning_rate": 1.2390803601796281e-07, "loss": 0.2157, "step": 69440 }, { "epoch": 9.01, "learning_rate": 1.2358549533173953e-07, "loss": 0.2353, "step": 69450 }, { "epoch": 9.01, "learning_rate": 1.2326336362199337e-07, "loss": 0.2272, "step": 69460 }, { "epoch": 9.02, "learning_rate": 1.2294164094806546e-07, "loss": 0.2255, "step": 69470 }, { "epoch": 9.02, "learning_rate": 1.2262032736921932e-07, "loss": 0.2332, "step": 69480 }, { "epoch": 9.02, "learning_rate": 1.2229942294464433e-07, "loss": 0.204, "step": 69490 }, { "epoch": 9.02, "learning_rate": 1.2197892773345465e-07, "loss": 0.2384, "step": 69500 }, { "epoch": 9.02, "learning_rate": 1.2165884179468795e-07, "loss": 0.2281, "step": 69510 }, { "epoch": 9.02, "learning_rate": 1.2133916518730825e-07, "loss": 0.2146, "step": 69520 }, { "epoch": 9.02, "learning_rate": 1.2101989797020287e-07, "loss": 0.2288, "step": 69530 }, { "epoch": 9.03, "learning_rate": 1.207010402021833e-07, "loss": 0.2259, "step": 69540 }, { "epoch": 9.03, "learning_rate": 1.2038259194198722e-07, "loss": 0.2089, "step": 69550 }, { "epoch": 9.03, "learning_rate": 1.200645532482755e-07, "loss": 0.2406, "step": 69560 }, { "epoch": 9.03, "learning_rate": 1.1974692417963412e-07, "loss": 0.2221, "step": 69570 }, { "epoch": 9.03, "learning_rate": 1.1942970479457354e-07, "loss": 0.2342, "step": 69580 }, { "epoch": 9.03, "learning_rate": 1.1911289515152905e-07, "loss": 0.2398, "step": 69590 }, { "epoch": 9.03, "learning_rate": 1.1879649530885946e-07, "loss": 0.2343, "step": 69600 }, { "epoch": 9.03, "learning_rate": 1.1848050532484963e-07, "loss": 0.24, "step": 69610 }, { "epoch": 9.04, "learning_rate": 1.1816492525770745e-07, "loss": 0.2154, "step": 69620 }, { "epoch": 9.04, "learning_rate": 1.1784975516556582e-07, "loss": 0.2215, "step": 69630 }, { "epoch": 9.04, "learning_rate": 1.1753499510648301e-07, "loss": 0.232, "step": 69640 }, { "epoch": 9.04, "learning_rate": 1.1722064513844023e-07, "loss": 0.2122, "step": 69650 }, { "epoch": 9.04, "learning_rate": 1.1690670531934403e-07, "loss": 0.2211, "step": 69660 }, { "epoch": 9.04, "learning_rate": 1.1659317570702547e-07, "loss": 0.2194, "step": 69670 }, { "epoch": 9.04, "learning_rate": 1.1628005635923943e-07, "loss": 0.2306, "step": 69680 }, { "epoch": 9.04, "learning_rate": 1.1596734733366631e-07, "loss": 0.2209, "step": 69690 }, { "epoch": 9.05, "learning_rate": 1.1565504868790978e-07, "loss": 0.2061, "step": 69700 }, { "epoch": 9.05, "learning_rate": 1.1534316047949803e-07, "loss": 0.2223, "step": 69710 }, { "epoch": 9.05, "learning_rate": 1.150316827658846e-07, "loss": 0.2257, "step": 69720 }, { "epoch": 9.05, "learning_rate": 1.1472061560444677e-07, "loss": 0.2138, "step": 69730 }, { "epoch": 9.05, "learning_rate": 1.144099590524858e-07, "loss": 0.2248, "step": 69740 }, { "epoch": 9.05, "learning_rate": 1.140997131672281e-07, "loss": 0.233, "step": 69750 }, { "epoch": 9.05, "learning_rate": 1.1378987800582376e-07, "loss": 0.251, "step": 69760 }, { "epoch": 9.06, "learning_rate": 1.1348045362534744e-07, "loss": 0.21, "step": 69770 }, { "epoch": 9.06, "learning_rate": 1.1317144008279886e-07, "loss": 0.2433, "step": 69780 }, { "epoch": 9.06, "learning_rate": 1.1286283743510097e-07, "loss": 0.2342, "step": 69790 }, { "epoch": 9.06, "learning_rate": 1.1255464573910102e-07, "loss": 0.2465, "step": 69800 }, { "epoch": 9.06, "learning_rate": 1.1224686505157181e-07, "loss": 0.2306, "step": 69810 }, { "epoch": 9.06, "learning_rate": 1.119394954292089e-07, "loss": 0.2378, "step": 69820 }, { "epoch": 9.06, "learning_rate": 1.1163253692863363e-07, "loss": 0.2196, "step": 69830 }, { "epoch": 9.06, "learning_rate": 1.113259896063901e-07, "loss": 0.2285, "step": 69840 }, { "epoch": 9.07, "learning_rate": 1.1101985351894749e-07, "loss": 0.2373, "step": 69850 }, { "epoch": 9.07, "learning_rate": 1.1071412872269972e-07, "loss": 0.2118, "step": 69860 }, { "epoch": 9.07, "learning_rate": 1.1040881527396372e-07, "loss": 0.2239, "step": 69870 }, { "epoch": 9.07, "learning_rate": 1.1010391322898126e-07, "loss": 0.2207, "step": 69880 }, { "epoch": 9.07, "learning_rate": 1.0979942264391908e-07, "loss": 0.2216, "step": 69890 }, { "epoch": 9.07, "learning_rate": 1.0949534357486645e-07, "loss": 0.2249, "step": 69900 }, { "epoch": 9.07, "learning_rate": 1.091916760778382e-07, "loss": 0.2224, "step": 69910 }, { "epoch": 9.07, "learning_rate": 1.0888842020877319e-07, "loss": 0.2283, "step": 69920 }, { "epoch": 9.08, "learning_rate": 1.08585576023534e-07, "loss": 0.2292, "step": 69930 }, { "epoch": 9.08, "learning_rate": 1.0828314357790702e-07, "loss": 0.2157, "step": 69940 }, { "epoch": 9.08, "learning_rate": 1.079811229276039e-07, "loss": 0.2418, "step": 69950 }, { "epoch": 9.08, "learning_rate": 1.076795141282596e-07, "loss": 0.2189, "step": 69960 }, { "epoch": 9.08, "learning_rate": 1.0737831723543382e-07, "loss": 0.2403, "step": 69970 }, { "epoch": 9.08, "learning_rate": 1.0707753230461005e-07, "loss": 0.2371, "step": 69980 }, { "epoch": 9.08, "learning_rate": 1.0677715939119503e-07, "loss": 0.215, "step": 69990 }, { "epoch": 9.09, "learning_rate": 1.0647719855052154e-07, "loss": 0.2233, "step": 70000 }, { "epoch": 9.09, "learning_rate": 1.061776498378449e-07, "loss": 0.2328, "step": 70010 }, { "epoch": 9.09, "learning_rate": 1.0587851330834465e-07, "loss": 0.2301, "step": 70020 }, { "epoch": 9.09, "learning_rate": 1.0557978901712542e-07, "loss": 0.2194, "step": 70030 }, { "epoch": 9.09, "learning_rate": 1.0528147701921452e-07, "loss": 0.2484, "step": 70040 }, { "epoch": 9.09, "learning_rate": 1.0498357736956438e-07, "loss": 0.2328, "step": 70050 }, { "epoch": 9.09, "learning_rate": 1.0468609012305113e-07, "loss": 0.219, "step": 70060 }, { "epoch": 9.09, "learning_rate": 1.043890153344747e-07, "loss": 0.228, "step": 70070 }, { "epoch": 9.1, "learning_rate": 1.0409235305855901e-07, "loss": 0.2263, "step": 70080 }, { "epoch": 9.1, "learning_rate": 1.0379610334995308e-07, "loss": 0.2152, "step": 70090 }, { "epoch": 9.1, "learning_rate": 1.0350026626322807e-07, "loss": 0.2247, "step": 70100 }, { "epoch": 9.1, "learning_rate": 1.0320484185288077e-07, "loss": 0.22, "step": 70110 }, { "epoch": 9.1, "learning_rate": 1.0290983017333119e-07, "loss": 0.2164, "step": 70120 }, { "epoch": 9.1, "learning_rate": 1.0261523127892283e-07, "loss": 0.2265, "step": 70130 }, { "epoch": 9.1, "learning_rate": 1.0232104522392478e-07, "loss": 0.2162, "step": 70140 }, { "epoch": 9.1, "learning_rate": 1.0202727206252859e-07, "loss": 0.2242, "step": 70150 }, { "epoch": 9.11, "learning_rate": 1.0173391184884958e-07, "loss": 0.2233, "step": 70160 }, { "epoch": 9.11, "learning_rate": 1.0144096463692861e-07, "loss": 0.231, "step": 70170 }, { "epoch": 9.11, "learning_rate": 1.011484304807293e-07, "loss": 0.224, "step": 70180 }, { "epoch": 9.11, "learning_rate": 1.008563094341385e-07, "loss": 0.2277, "step": 70190 }, { "epoch": 9.11, "learning_rate": 1.0056460155096887e-07, "loss": 0.2079, "step": 70200 }, { "epoch": 9.11, "learning_rate": 1.0027330688495558e-07, "loss": 0.2135, "step": 70210 }, { "epoch": 9.11, "learning_rate": 9.99824254897575e-08, "loss": 0.2327, "step": 70220 }, { "epoch": 9.11, "learning_rate": 9.969195741895862e-08, "loss": 0.2303, "step": 70230 }, { "epoch": 9.12, "learning_rate": 9.94019027260656e-08, "loss": 0.2243, "step": 70240 }, { "epoch": 9.12, "learning_rate": 9.911226146450992e-08, "loss": 0.2201, "step": 70250 }, { "epoch": 9.12, "learning_rate": 9.882303368764603e-08, "loss": 0.2202, "step": 70260 }, { "epoch": 9.12, "learning_rate": 9.853421944875217e-08, "loss": 0.2085, "step": 70270 }, { "epoch": 9.12, "learning_rate": 9.824581880103159e-08, "loss": 0.2081, "step": 70280 }, { "epoch": 9.12, "learning_rate": 9.795783179761029e-08, "loss": 0.2296, "step": 70290 }, { "epoch": 9.12, "learning_rate": 9.767025849153777e-08, "loss": 0.2256, "step": 70300 }, { "epoch": 9.13, "learning_rate": 9.738309893578882e-08, "loss": 0.2193, "step": 70310 }, { "epoch": 9.13, "learning_rate": 9.709635318326075e-08, "loss": 0.2241, "step": 70320 }, { "epoch": 9.13, "learning_rate": 9.681002128677455e-08, "loss": 0.2392, "step": 70330 }, { "epoch": 9.13, "learning_rate": 9.652410329907632e-08, "loss": 0.2419, "step": 70340 }, { "epoch": 9.13, "learning_rate": 9.623859927283435e-08, "loss": 0.2462, "step": 70350 }, { "epoch": 9.13, "learning_rate": 9.595350926064095e-08, "loss": 0.2291, "step": 70360 }, { "epoch": 9.13, "learning_rate": 9.566883331501373e-08, "loss": 0.2243, "step": 70370 }, { "epoch": 9.13, "learning_rate": 9.53845714883915e-08, "loss": 0.2587, "step": 70380 }, { "epoch": 9.14, "learning_rate": 9.510072383313938e-08, "loss": 0.2145, "step": 70390 }, { "epoch": 9.14, "learning_rate": 9.481729040154447e-08, "loss": 0.2171, "step": 70400 }, { "epoch": 9.14, "learning_rate": 9.453427124581732e-08, "loss": 0.2321, "step": 70410 }, { "epoch": 9.14, "learning_rate": 9.425166641809411e-08, "loss": 0.2329, "step": 70420 }, { "epoch": 9.14, "learning_rate": 9.396947597043268e-08, "loss": 0.2167, "step": 70430 }, { "epoch": 9.14, "learning_rate": 9.368769995481539e-08, "loss": 0.2317, "step": 70440 }, { "epoch": 9.14, "learning_rate": 9.340633842314839e-08, "loss": 0.2261, "step": 70450 }, { "epoch": 9.14, "learning_rate": 9.312539142726132e-08, "loss": 0.2246, "step": 70460 }, { "epoch": 9.15, "learning_rate": 9.284485901890705e-08, "loss": 0.2273, "step": 70470 }, { "epoch": 9.15, "learning_rate": 9.256474124976296e-08, "loss": 0.2177, "step": 70480 }, { "epoch": 9.15, "learning_rate": 9.228503817142949e-08, "loss": 0.2182, "step": 70490 }, { "epoch": 9.15, "learning_rate": 9.200574983543024e-08, "loss": 0.2261, "step": 70500 }, { "epoch": 9.15, "learning_rate": 9.172687629321338e-08, "loss": 0.2364, "step": 70510 }, { "epoch": 9.15, "learning_rate": 9.14484175961498e-08, "loss": 0.2266, "step": 70520 }, { "epoch": 9.15, "learning_rate": 9.117037379553467e-08, "loss": 0.2107, "step": 70530 }, { "epoch": 9.16, "learning_rate": 9.089274494258666e-08, "loss": 0.2166, "step": 70540 }, { "epoch": 9.16, "learning_rate": 9.06155310884474e-08, "loss": 0.2286, "step": 70550 }, { "epoch": 9.16, "learning_rate": 9.033873228418255e-08, "loss": 0.2179, "step": 70560 }, { "epoch": 9.16, "learning_rate": 9.006234858078129e-08, "loss": 0.2216, "step": 70570 }, { "epoch": 9.16, "learning_rate": 8.9786380029156e-08, "loss": 0.2144, "step": 70580 }, { "epoch": 9.16, "learning_rate": 8.951082668014311e-08, "loss": 0.2329, "step": 70590 }, { "epoch": 9.16, "learning_rate": 8.923568858450227e-08, "loss": 0.2362, "step": 70600 }, { "epoch": 9.16, "learning_rate": 8.896096579291637e-08, "loss": 0.2381, "step": 70610 }, { "epoch": 9.17, "learning_rate": 8.868665835599287e-08, "loss": 0.2244, "step": 70620 }, { "epoch": 9.17, "learning_rate": 8.841276632426111e-08, "loss": 0.226, "step": 70630 }, { "epoch": 9.17, "learning_rate": 8.813928974817476e-08, "loss": 0.2395, "step": 70640 }, { "epoch": 9.17, "learning_rate": 8.786622867811175e-08, "loss": 0.2322, "step": 70650 }, { "epoch": 9.17, "learning_rate": 8.759358316437195e-08, "loss": 0.2144, "step": 70660 }, { "epoch": 9.17, "learning_rate": 8.732135325717924e-08, "loss": 0.2305, "step": 70670 }, { "epoch": 9.17, "learning_rate": 8.704953900668179e-08, "loss": 0.2151, "step": 70680 }, { "epoch": 9.17, "learning_rate": 8.677814046294971e-08, "loss": 0.2132, "step": 70690 }, { "epoch": 9.18, "learning_rate": 8.650715767597816e-08, "loss": 0.2298, "step": 70700 }, { "epoch": 9.18, "learning_rate": 8.623659069568424e-08, "loss": 0.2285, "step": 70710 }, { "epoch": 9.18, "learning_rate": 8.596643957190907e-08, "loss": 0.2121, "step": 70720 }, { "epoch": 9.18, "learning_rate": 8.569670435441779e-08, "loss": 0.2365, "step": 70730 }, { "epoch": 9.18, "learning_rate": 8.5427385092898e-08, "loss": 0.25, "step": 70740 }, { "epoch": 9.18, "learning_rate": 8.515848183696053e-08, "loss": 0.2364, "step": 70750 }, { "epoch": 9.18, "learning_rate": 8.488999463614078e-08, "loss": 0.2277, "step": 70760 }, { "epoch": 9.18, "learning_rate": 8.462192353989655e-08, "loss": 0.2191, "step": 70770 }, { "epoch": 9.19, "learning_rate": 8.435426859760867e-08, "loss": 0.2256, "step": 70780 }, { "epoch": 9.19, "learning_rate": 8.408702985858275e-08, "loss": 0.2172, "step": 70790 }, { "epoch": 9.19, "learning_rate": 8.382020737204608e-08, "loss": 0.2246, "step": 70800 }, { "epoch": 9.19, "learning_rate": 8.355380118715022e-08, "loss": 0.2286, "step": 70810 }, { "epoch": 9.19, "learning_rate": 8.328781135297025e-08, "loss": 0.2308, "step": 70820 }, { "epoch": 9.19, "learning_rate": 8.302223791850368e-08, "loss": 0.2207, "step": 70830 }, { "epoch": 9.19, "learning_rate": 8.275708093267207e-08, "loss": 0.2179, "step": 70840 }, { "epoch": 9.2, "learning_rate": 8.24923404443202e-08, "loss": 0.2276, "step": 70850 }, { "epoch": 9.2, "learning_rate": 8.222801650221502e-08, "loss": 0.223, "step": 70860 }, { "epoch": 9.2, "learning_rate": 8.196410915504862e-08, "loss": 0.2265, "step": 70870 }, { "epoch": 9.2, "learning_rate": 8.170061845143493e-08, "loss": 0.2286, "step": 70880 }, { "epoch": 9.2, "learning_rate": 8.143754443991146e-08, "loss": 0.2245, "step": 70890 }, { "epoch": 9.2, "learning_rate": 8.117488716893917e-08, "loss": 0.2347, "step": 70900 }, { "epoch": 9.2, "learning_rate": 8.091264668690254e-08, "loss": 0.2245, "step": 70910 }, { "epoch": 9.2, "learning_rate": 8.065082304210824e-08, "loss": 0.2253, "step": 70920 }, { "epoch": 9.21, "learning_rate": 8.038941628278721e-08, "loss": 0.2195, "step": 70930 }, { "epoch": 9.21, "learning_rate": 8.012842645709312e-08, "loss": 0.2145, "step": 70940 }, { "epoch": 9.21, "learning_rate": 7.986785361310264e-08, "loss": 0.2212, "step": 70950 }, { "epoch": 9.21, "learning_rate": 7.960769779881641e-08, "loss": 0.2176, "step": 70960 }, { "epoch": 9.21, "learning_rate": 7.93479590621573e-08, "loss": 0.2241, "step": 70970 }, { "epoch": 9.21, "learning_rate": 7.908863745097194e-08, "loss": 0.2238, "step": 70980 }, { "epoch": 9.21, "learning_rate": 7.882973301302992e-08, "loss": 0.222, "step": 70990 }, { "epoch": 9.21, "learning_rate": 7.857124579602409e-08, "loss": 0.2271, "step": 71000 }, { "epoch": 9.22, "learning_rate": 7.831317584757026e-08, "loss": 0.2141, "step": 71010 }, { "epoch": 9.22, "learning_rate": 7.805552321520775e-08, "loss": 0.242, "step": 71020 }, { "epoch": 9.22, "learning_rate": 7.779828794639833e-08, "loss": 0.2292, "step": 71030 }, { "epoch": 9.22, "learning_rate": 7.75414700885278e-08, "loss": 0.2163, "step": 71040 }, { "epoch": 9.22, "learning_rate": 7.728506968890414e-08, "loss": 0.2413, "step": 71050 }, { "epoch": 9.22, "learning_rate": 7.702908679475885e-08, "loss": 0.2396, "step": 71060 }, { "epoch": 9.22, "learning_rate": 7.677352145324719e-08, "loss": 0.2164, "step": 71070 }, { "epoch": 9.23, "learning_rate": 7.651837371144605e-08, "loss": 0.2057, "step": 71080 }, { "epoch": 9.23, "learning_rate": 7.62636436163564e-08, "loss": 0.227, "step": 71090 }, { "epoch": 9.23, "learning_rate": 7.600933121490244e-08, "loss": 0.2339, "step": 71100 }, { "epoch": 9.23, "learning_rate": 7.575543655393076e-08, "loss": 0.2306, "step": 71110 }, { "epoch": 9.23, "learning_rate": 7.550195968021128e-08, "loss": 0.2304, "step": 71120 }, { "epoch": 9.23, "learning_rate": 7.524890064043712e-08, "loss": 0.2329, "step": 71130 }, { "epoch": 9.23, "learning_rate": 7.499625948122409e-08, "loss": 0.22, "step": 71140 }, { "epoch": 9.23, "learning_rate": 7.474403624911154e-08, "loss": 0.2245, "step": 71150 }, { "epoch": 9.24, "learning_rate": 7.4492230990561e-08, "loss": 0.2274, "step": 71160 }, { "epoch": 9.24, "learning_rate": 7.424084375195777e-08, "loss": 0.2191, "step": 71170 }, { "epoch": 9.24, "learning_rate": 7.398987457961037e-08, "loss": 0.2177, "step": 71180 }, { "epoch": 9.24, "learning_rate": 7.373932351974901e-08, "loss": 0.2334, "step": 71190 }, { "epoch": 9.24, "learning_rate": 7.348919061852792e-08, "loss": 0.2234, "step": 71200 }, { "epoch": 9.24, "learning_rate": 7.323947592202456e-08, "loss": 0.2213, "step": 71210 }, { "epoch": 9.24, "learning_rate": 7.299017947623859e-08, "loss": 0.2224, "step": 71220 }, { "epoch": 9.24, "learning_rate": 7.274130132709265e-08, "loss": 0.2227, "step": 71230 }, { "epoch": 9.25, "learning_rate": 7.249284152043289e-08, "loss": 0.2355, "step": 71240 }, { "epoch": 9.25, "learning_rate": 7.224480010202763e-08, "loss": 0.2141, "step": 71250 }, { "epoch": 9.25, "learning_rate": 7.199717711756924e-08, "loss": 0.2269, "step": 71260 }, { "epoch": 9.25, "learning_rate": 7.174997261267228e-08, "loss": 0.2168, "step": 71270 }, { "epoch": 9.25, "learning_rate": 7.150318663287349e-08, "loss": 0.2268, "step": 71280 }, { "epoch": 9.25, "learning_rate": 7.125681922363443e-08, "loss": 0.218, "step": 71290 }, { "epoch": 9.25, "learning_rate": 7.101087043033752e-08, "loss": 0.2308, "step": 71300 }, { "epoch": 9.26, "learning_rate": 7.076534029828926e-08, "loss": 0.2198, "step": 71310 }, { "epoch": 9.26, "learning_rate": 7.052022887271908e-08, "loss": 0.2275, "step": 71320 }, { "epoch": 9.26, "learning_rate": 7.027553619877862e-08, "loss": 0.2401, "step": 71330 }, { "epoch": 9.26, "learning_rate": 7.003126232154278e-08, "loss": 0.2343, "step": 71340 }, { "epoch": 9.26, "learning_rate": 6.97874072860094e-08, "loss": 0.2291, "step": 71350 }, { "epoch": 9.26, "learning_rate": 6.95439711370988e-08, "loss": 0.2182, "step": 71360 }, { "epoch": 9.26, "learning_rate": 6.930095391965429e-08, "loss": 0.2447, "step": 71370 }, { "epoch": 9.26, "learning_rate": 6.90583556784424e-08, "loss": 0.2329, "step": 71380 }, { "epoch": 9.27, "learning_rate": 6.881617645815157e-08, "loss": 0.2264, "step": 71390 }, { "epoch": 9.27, "learning_rate": 6.857441630339459e-08, "loss": 0.22, "step": 71400 }, { "epoch": 9.27, "learning_rate": 6.833307525870536e-08, "loss": 0.2502, "step": 71410 }, { "epoch": 9.27, "learning_rate": 6.809215336854102e-08, "loss": 0.2333, "step": 71420 }, { "epoch": 9.27, "learning_rate": 6.785165067728273e-08, "loss": 0.2176, "step": 71430 }, { "epoch": 9.27, "learning_rate": 6.761156722923308e-08, "loss": 0.2251, "step": 71440 }, { "epoch": 9.27, "learning_rate": 6.737190306861738e-08, "loss": 0.2367, "step": 71450 }, { "epoch": 9.27, "learning_rate": 6.713265823958466e-08, "loss": 0.2386, "step": 71460 }, { "epoch": 9.28, "learning_rate": 6.68938327862062e-08, "loss": 0.2276, "step": 71470 }, { "epoch": 9.28, "learning_rate": 6.665542675247598e-08, "loss": 0.2386, "step": 71480 }, { "epoch": 9.28, "learning_rate": 6.641744018231069e-08, "loss": 0.2175, "step": 71490 }, { "epoch": 9.28, "learning_rate": 6.617987311954972e-08, "loss": 0.2183, "step": 71500 }, { "epoch": 9.28, "learning_rate": 6.594272560795548e-08, "loss": 0.2279, "step": 71510 }, { "epoch": 9.28, "learning_rate": 6.570599769121281e-08, "loss": 0.2253, "step": 71520 }, { "epoch": 9.28, "learning_rate": 6.54696894129298e-08, "loss": 0.2166, "step": 71530 }, { "epoch": 9.28, "learning_rate": 6.523380081663596e-08, "loss": 0.2205, "step": 71540 }, { "epoch": 9.29, "learning_rate": 6.499833194578479e-08, "loss": 0.2177, "step": 71550 }, { "epoch": 9.29, "learning_rate": 6.476328284375203e-08, "loss": 0.2169, "step": 71560 }, { "epoch": 9.29, "learning_rate": 6.452865355383638e-08, "loss": 0.2092, "step": 71570 }, { "epoch": 9.29, "learning_rate": 6.429444411925844e-08, "loss": 0.2138, "step": 71580 }, { "epoch": 9.29, "learning_rate": 6.406065458316161e-08, "loss": 0.247, "step": 71590 }, { "epoch": 9.29, "learning_rate": 6.382728498861324e-08, "loss": 0.2218, "step": 71600 }, { "epoch": 9.29, "learning_rate": 6.35943353786016e-08, "loss": 0.2205, "step": 71610 }, { "epoch": 9.3, "learning_rate": 6.336180579603819e-08, "loss": 0.2092, "step": 71620 }, { "epoch": 9.3, "learning_rate": 6.312969628375804e-08, "loss": 0.2279, "step": 71630 }, { "epoch": 9.3, "learning_rate": 6.289800688451755e-08, "loss": 0.2166, "step": 71640 }, { "epoch": 9.3, "learning_rate": 6.266673764099613e-08, "loss": 0.2246, "step": 71650 }, { "epoch": 9.3, "learning_rate": 6.243588859579617e-08, "loss": 0.2277, "step": 71660 }, { "epoch": 9.3, "learning_rate": 6.220545979144223e-08, "loss": 0.245, "step": 71670 }, { "epoch": 9.3, "learning_rate": 6.197545127038135e-08, "loss": 0.2369, "step": 71680 }, { "epoch": 9.3, "learning_rate": 6.174586307498408e-08, "loss": 0.2214, "step": 71690 }, { "epoch": 9.31, "learning_rate": 6.151669524754208e-08, "loss": 0.2181, "step": 71700 }, { "epoch": 9.31, "learning_rate": 6.12879478302711e-08, "loss": 0.2342, "step": 71710 }, { "epoch": 9.31, "learning_rate": 6.105962086530798e-08, "loss": 0.24, "step": 71720 }, { "epoch": 9.31, "learning_rate": 6.083171439471334e-08, "loss": 0.2312, "step": 71730 }, { "epoch": 9.31, "learning_rate": 6.06042284604695e-08, "loss": 0.2293, "step": 71740 }, { "epoch": 9.31, "learning_rate": 6.0377163104482e-08, "loss": 0.216, "step": 71750 }, { "epoch": 9.31, "learning_rate": 6.015051836857805e-08, "loss": 0.2277, "step": 71760 }, { "epoch": 9.31, "learning_rate": 5.992429429450838e-08, "loss": 0.2212, "step": 71770 }, { "epoch": 9.32, "learning_rate": 5.969849092394537e-08, "loss": 0.2196, "step": 71780 }, { "epoch": 9.32, "learning_rate": 5.9473108298484406e-08, "loss": 0.2408, "step": 71790 }, { "epoch": 9.32, "learning_rate": 5.9248146459643305e-08, "loss": 0.2158, "step": 71800 }, { "epoch": 9.32, "learning_rate": 5.902360544886211e-08, "loss": 0.2203, "step": 71810 }, { "epoch": 9.32, "learning_rate": 5.879948530750356e-08, "loss": 0.2165, "step": 71820 }, { "epoch": 9.32, "learning_rate": 5.857578607685286e-08, "loss": 0.2203, "step": 71830 }, { "epoch": 9.32, "learning_rate": 5.8352507798117666e-08, "loss": 0.2279, "step": 71840 }, { "epoch": 9.33, "learning_rate": 5.8129650512427854e-08, "loss": 0.2189, "step": 71850 }, { "epoch": 9.33, "learning_rate": 5.790721426083651e-08, "loss": 0.2087, "step": 71860 }, { "epoch": 9.33, "learning_rate": 5.768519908431789e-08, "loss": 0.2275, "step": 71870 }, { "epoch": 9.33, "learning_rate": 5.7463605023770015e-08, "loss": 0.2245, "step": 71880 }, { "epoch": 9.33, "learning_rate": 5.724243212001257e-08, "loss": 0.2176, "step": 71890 }, { "epoch": 9.33, "learning_rate": 5.702168041378718e-08, "loss": 0.2153, "step": 71900 }, { "epoch": 9.33, "learning_rate": 5.6801349945759503e-08, "loss": 0.2288, "step": 71910 }, { "epoch": 9.33, "learning_rate": 5.658144075651608e-08, "loss": 0.2424, "step": 71920 }, { "epoch": 9.34, "learning_rate": 5.636195288656643e-08, "loss": 0.2202, "step": 71930 }, { "epoch": 9.34, "learning_rate": 5.614288637634254e-08, "loss": 0.2231, "step": 71940 }, { "epoch": 9.34, "learning_rate": 5.592424126619833e-08, "loss": 0.2196, "step": 71950 }, { "epoch": 9.34, "learning_rate": 5.570601759641068e-08, "loss": 0.2218, "step": 71960 }, { "epoch": 9.34, "learning_rate": 5.5488215407178696e-08, "loss": 0.2221, "step": 71970 }, { "epoch": 9.34, "learning_rate": 5.5270834738623416e-08, "loss": 0.2227, "step": 71980 }, { "epoch": 9.34, "learning_rate": 5.505387563078858e-08, "loss": 0.2121, "step": 71990 }, { "epoch": 9.34, "learning_rate": 5.4837338123640657e-08, "loss": 0.2328, "step": 72000 }, { "epoch": 9.35, "learning_rate": 5.462122225706727e-08, "loss": 0.2366, "step": 72010 }, { "epoch": 9.35, "learning_rate": 5.4405528070879544e-08, "loss": 0.2227, "step": 72020 }, { "epoch": 9.35, "learning_rate": 5.41902556048108e-08, "loss": 0.2158, "step": 72030 }, { "epoch": 9.35, "learning_rate": 5.3975404898515525e-08, "loss": 0.2269, "step": 72040 }, { "epoch": 9.35, "learning_rate": 5.376097599157246e-08, "loss": 0.2323, "step": 72050 }, { "epoch": 9.35, "learning_rate": 5.354696892348074e-08, "loss": 0.2354, "step": 72060 }, { "epoch": 9.35, "learning_rate": 5.333338373366275e-08, "loss": 0.2312, "step": 72070 }, { "epoch": 9.35, "learning_rate": 5.3120220461463046e-08, "loss": 0.2228, "step": 72080 }, { "epoch": 9.36, "learning_rate": 5.290747914614866e-08, "loss": 0.2263, "step": 72090 }, { "epoch": 9.36, "learning_rate": 5.269515982690856e-08, "loss": 0.2249, "step": 72100 }, { "epoch": 9.36, "learning_rate": 5.2483262542853905e-08, "loss": 0.2237, "step": 72110 }, { "epoch": 9.36, "learning_rate": 5.227178733301831e-08, "loss": 0.2332, "step": 72120 }, { "epoch": 9.36, "learning_rate": 5.206073423635786e-08, "loss": 0.2232, "step": 72130 }, { "epoch": 9.36, "learning_rate": 5.185010329175057e-08, "loss": 0.2204, "step": 72140 }, { "epoch": 9.36, "learning_rate": 5.1639894537996646e-08, "loss": 0.225, "step": 72150 }, { "epoch": 9.37, "learning_rate": 5.143010801381877e-08, "loss": 0.222, "step": 72160 }, { "epoch": 9.37, "learning_rate": 5.122074375786154e-08, "loss": 0.2199, "step": 72170 }, { "epoch": 9.37, "learning_rate": 5.101180180869177e-08, "loss": 0.2215, "step": 72180 }, { "epoch": 9.37, "learning_rate": 5.08032822047995e-08, "loss": 0.2292, "step": 72190 }, { "epoch": 9.37, "learning_rate": 5.059518498459515e-08, "loss": 0.2168, "step": 72200 }, { "epoch": 9.37, "learning_rate": 5.0387510186412904e-08, "loss": 0.2347, "step": 72210 }, { "epoch": 9.37, "learning_rate": 5.0180257848508355e-08, "loss": 0.2259, "step": 72220 }, { "epoch": 9.37, "learning_rate": 4.99734280090593e-08, "loss": 0.216, "step": 72230 }, { "epoch": 9.38, "learning_rate": 4.9767020706165986e-08, "loss": 0.2297, "step": 72240 }, { "epoch": 9.38, "learning_rate": 4.9561035977850864e-08, "loss": 0.2189, "step": 72250 }, { "epoch": 9.38, "learning_rate": 4.9355473862057806e-08, "loss": 0.2277, "step": 72260 }, { "epoch": 9.38, "learning_rate": 4.915033439665443e-08, "loss": 0.2141, "step": 72270 }, { "epoch": 9.38, "learning_rate": 4.894561761942849e-08, "loss": 0.2168, "step": 72280 }, { "epoch": 9.38, "learning_rate": 4.8741323568091213e-08, "loss": 0.2412, "step": 72290 }, { "epoch": 9.38, "learning_rate": 4.853745228027605e-08, "loss": 0.223, "step": 72300 }, { "epoch": 9.38, "learning_rate": 4.8334003793537316e-08, "loss": 0.2206, "step": 72310 }, { "epoch": 9.39, "learning_rate": 4.81309781453526e-08, "loss": 0.2232, "step": 72320 }, { "epoch": 9.39, "learning_rate": 4.792837537312168e-08, "loss": 0.2225, "step": 72330 }, { "epoch": 9.39, "learning_rate": 4.772619551416546e-08, "loss": 0.2219, "step": 72340 }, { "epoch": 9.39, "learning_rate": 4.7524438605727604e-08, "loss": 0.2395, "step": 72350 }, { "epoch": 9.39, "learning_rate": 4.732310468497395e-08, "loss": 0.2357, "step": 72360 }, { "epoch": 9.39, "learning_rate": 4.7122193788992015e-08, "loss": 0.231, "step": 72370 }, { "epoch": 9.39, "learning_rate": 4.692170595479152e-08, "loss": 0.2246, "step": 72380 }, { "epoch": 9.4, "learning_rate": 4.6721641219304646e-08, "loss": 0.234, "step": 72390 }, { "epoch": 9.4, "learning_rate": 4.652199961938498e-08, "loss": 0.2119, "step": 72400 }, { "epoch": 9.4, "learning_rate": 4.6322781191808576e-08, "loss": 0.236, "step": 72410 }, { "epoch": 9.4, "learning_rate": 4.612398597327394e-08, "loss": 0.2225, "step": 72420 }, { "epoch": 9.4, "learning_rate": 4.5925614000400216e-08, "loss": 0.2333, "step": 72430 }, { "epoch": 9.4, "learning_rate": 4.5727665309730316e-08, "loss": 0.2239, "step": 72440 }, { "epoch": 9.4, "learning_rate": 4.553013993772829e-08, "loss": 0.2418, "step": 72450 }, { "epoch": 9.4, "learning_rate": 4.533303792077963e-08, "loss": 0.2346, "step": 72460 }, { "epoch": 9.41, "learning_rate": 4.5136359295193033e-08, "loss": 0.227, "step": 72470 }, { "epoch": 9.41, "learning_rate": 4.494010409719865e-08, "loss": 0.2178, "step": 72480 }, { "epoch": 9.41, "learning_rate": 4.474427236294854e-08, "loss": 0.2177, "step": 72490 }, { "epoch": 9.41, "learning_rate": 4.454886412851699e-08, "loss": 0.2046, "step": 72500 }, { "epoch": 9.41, "learning_rate": 4.435387942989992e-08, "loss": 0.2118, "step": 72510 }, { "epoch": 9.41, "learning_rate": 4.415931830301549e-08, "loss": 0.2319, "step": 72520 }, { "epoch": 9.41, "learning_rate": 4.396518078370404e-08, "loss": 0.2156, "step": 72530 }, { "epoch": 9.41, "learning_rate": 4.377146690772758e-08, "loss": 0.2219, "step": 72540 }, { "epoch": 9.42, "learning_rate": 4.3578176710769815e-08, "loss": 0.2273, "step": 72550 }, { "epoch": 9.42, "learning_rate": 4.3385310228437157e-08, "loss": 0.2123, "step": 72560 }, { "epoch": 9.42, "learning_rate": 4.319286749625717e-08, "loss": 0.2268, "step": 72570 }, { "epoch": 9.42, "learning_rate": 4.300084854968015e-08, "loss": 0.2221, "step": 72580 }, { "epoch": 9.42, "learning_rate": 4.280925342407754e-08, "loss": 0.2322, "step": 72590 }, { "epoch": 9.42, "learning_rate": 4.2618082154742985e-08, "loss": 0.2377, "step": 72600 }, { "epoch": 9.42, "learning_rate": 4.2427334776892836e-08, "loss": 0.2219, "step": 72610 }, { "epoch": 9.43, "learning_rate": 4.223701132566384e-08, "loss": 0.231, "step": 72620 }, { "epoch": 9.43, "learning_rate": 4.2047111836115975e-08, "loss": 0.2231, "step": 72630 }, { "epoch": 9.43, "learning_rate": 4.185763634323037e-08, "loss": 0.2267, "step": 72640 }, { "epoch": 9.43, "learning_rate": 4.166858488191061e-08, "loss": 0.2369, "step": 72650 }, { "epoch": 9.43, "learning_rate": 4.1479957486981444e-08, "loss": 0.2244, "step": 72660 }, { "epoch": 9.43, "learning_rate": 4.129175419319009e-08, "loss": 0.2344, "step": 72670 }, { "epoch": 9.43, "learning_rate": 4.110397503520567e-08, "loss": 0.2212, "step": 72680 }, { "epoch": 9.43, "learning_rate": 4.0916620047618756e-08, "loss": 0.2123, "step": 72690 }, { "epoch": 9.44, "learning_rate": 4.0729689264942104e-08, "loss": 0.2211, "step": 72700 }, { "epoch": 9.44, "learning_rate": 4.054318272161015e-08, "loss": 0.2245, "step": 72710 }, { "epoch": 9.44, "learning_rate": 4.035710045197952e-08, "loss": 0.2337, "step": 72720 }, { "epoch": 9.44, "learning_rate": 4.0171442490328273e-08, "loss": 0.2292, "step": 72730 }, { "epoch": 9.44, "learning_rate": 3.9986208870856124e-08, "loss": 0.2141, "step": 72740 }, { "epoch": 9.44, "learning_rate": 3.9801399627685525e-08, "loss": 0.2153, "step": 72750 }, { "epoch": 9.44, "learning_rate": 3.9617014794859815e-08, "loss": 0.2194, "step": 72760 }, { "epoch": 9.44, "learning_rate": 3.943305440634453e-08, "loss": 0.2129, "step": 72770 }, { "epoch": 9.45, "learning_rate": 3.9249518496027155e-08, "loss": 0.2266, "step": 72780 }, { "epoch": 9.45, "learning_rate": 3.906640709771683e-08, "loss": 0.2318, "step": 72790 }, { "epoch": 9.45, "learning_rate": 3.8883720245144136e-08, "loss": 0.2359, "step": 72800 }, { "epoch": 9.45, "learning_rate": 3.870145797196236e-08, "loss": 0.2096, "step": 72810 }, { "epoch": 9.45, "learning_rate": 3.8519620311745945e-08, "loss": 0.2319, "step": 72820 }, { "epoch": 9.45, "learning_rate": 3.833820729799101e-08, "loss": 0.2289, "step": 72830 }, { "epoch": 9.45, "learning_rate": 3.8157218964115624e-08, "loss": 0.2002, "step": 72840 }, { "epoch": 9.45, "learning_rate": 3.797665534345977e-08, "loss": 0.2225, "step": 72850 }, { "epoch": 9.46, "learning_rate": 3.779651646928488e-08, "loss": 0.2323, "step": 72860 }, { "epoch": 9.46, "learning_rate": 3.761680237477482e-08, "loss": 0.2117, "step": 72870 }, { "epoch": 9.46, "learning_rate": 3.743751309303409e-08, "loss": 0.2332, "step": 72880 }, { "epoch": 9.46, "learning_rate": 3.725864865708993e-08, "loss": 0.2339, "step": 72890 }, { "epoch": 9.46, "learning_rate": 3.708020909989097e-08, "loss": 0.2252, "step": 72900 }, { "epoch": 9.46, "learning_rate": 3.690219445430754e-08, "loss": 0.2349, "step": 72910 }, { "epoch": 9.46, "learning_rate": 3.6724604753131383e-08, "loss": 0.2372, "step": 72920 }, { "epoch": 9.47, "learning_rate": 3.654744002907668e-08, "loss": 0.2363, "step": 72930 }, { "epoch": 9.47, "learning_rate": 3.637070031477854e-08, "loss": 0.2181, "step": 72940 }, { "epoch": 9.47, "learning_rate": 3.619438564279502e-08, "loss": 0.2173, "step": 72950 }, { "epoch": 9.47, "learning_rate": 3.6018496045604053e-08, "loss": 0.2304, "step": 72960 }, { "epoch": 9.47, "learning_rate": 3.584303155560681e-08, "loss": 0.2396, "step": 72970 }, { "epoch": 9.47, "learning_rate": 3.56679922051256e-08, "loss": 0.2213, "step": 72980 }, { "epoch": 9.47, "learning_rate": 3.549337802640443e-08, "loss": 0.227, "step": 72990 }, { "epoch": 9.47, "learning_rate": 3.531918905160846e-08, "loss": 0.2109, "step": 73000 }, { "epoch": 9.48, "learning_rate": 3.5145425312825806e-08, "loss": 0.2316, "step": 73010 }, { "epoch": 9.48, "learning_rate": 3.4972086842065e-08, "loss": 0.2144, "step": 73020 }, { "epoch": 9.48, "learning_rate": 3.479917367125648e-08, "loss": 0.2414, "step": 73030 }, { "epoch": 9.48, "learning_rate": 3.462668583225341e-08, "loss": 0.2194, "step": 73040 }, { "epoch": 9.48, "learning_rate": 3.445462335682908e-08, "loss": 0.2296, "step": 73050 }, { "epoch": 9.48, "learning_rate": 3.428298627667922e-08, "loss": 0.2132, "step": 73060 }, { "epoch": 9.48, "learning_rate": 3.411177462342152e-08, "loss": 0.2324, "step": 73070 }, { "epoch": 9.48, "learning_rate": 3.394098842859428e-08, "loss": 0.221, "step": 73080 }, { "epoch": 9.49, "learning_rate": 3.377062772365827e-08, "loss": 0.2233, "step": 73090 }, { "epoch": 9.49, "learning_rate": 3.3600692539995957e-08, "loss": 0.2319, "step": 73100 }, { "epoch": 9.49, "learning_rate": 3.3431182908910407e-08, "loss": 0.2187, "step": 73110 }, { "epoch": 9.49, "learning_rate": 3.32620988616277e-08, "loss": 0.2265, "step": 73120 }, { "epoch": 9.49, "learning_rate": 3.309344042929452e-08, "loss": 0.2177, "step": 73130 }, { "epoch": 9.49, "learning_rate": 3.292520764297925e-08, "loss": 0.2301, "step": 73140 }, { "epoch": 9.49, "learning_rate": 3.275740053367247e-08, "loss": 0.22, "step": 73150 }, { "epoch": 9.5, "learning_rate": 3.259001913228539e-08, "loss": 0.2314, "step": 73160 }, { "epoch": 9.5, "learning_rate": 3.242306346965167e-08, "loss": 0.2255, "step": 73170 }, { "epoch": 9.5, "learning_rate": 3.225653357652614e-08, "loss": 0.2299, "step": 73180 }, { "epoch": 9.5, "learning_rate": 3.209042948358557e-08, "loss": 0.2158, "step": 73190 }, { "epoch": 9.5, "learning_rate": 3.192475122142735e-08, "loss": 0.2083, "step": 73200 }, { "epoch": 9.5, "learning_rate": 3.1759498820571605e-08, "loss": 0.2435, "step": 73210 }, { "epoch": 9.5, "learning_rate": 3.1594672311459084e-08, "loss": 0.2173, "step": 73220 }, { "epoch": 9.5, "learning_rate": 3.143027172445274e-08, "loss": 0.2147, "step": 73230 }, { "epoch": 9.51, "learning_rate": 3.126629708983695e-08, "loss": 0.218, "step": 73240 }, { "epoch": 9.51, "learning_rate": 3.1102748437817225e-08, "loss": 0.2301, "step": 73250 }, { "epoch": 9.51, "learning_rate": 3.093962579852078e-08, "loss": 0.2114, "step": 73260 }, { "epoch": 9.51, "learning_rate": 3.077692920199675e-08, "loss": 0.2305, "step": 73270 }, { "epoch": 9.51, "learning_rate": 3.061465867821517e-08, "loss": 0.2286, "step": 73280 }, { "epoch": 9.51, "learning_rate": 3.0452814257068016e-08, "loss": 0.2303, "step": 73290 }, { "epoch": 9.51, "learning_rate": 3.029139596836894e-08, "loss": 0.2208, "step": 73300 }, { "epoch": 9.51, "learning_rate": 3.013040384185197e-08, "loss": 0.2169, "step": 73310 }, { "epoch": 9.52, "learning_rate": 2.996983790717437e-08, "loss": 0.2296, "step": 73320 }, { "epoch": 9.52, "learning_rate": 2.980969819391379e-08, "loss": 0.2252, "step": 73330 }, { "epoch": 9.52, "learning_rate": 2.964998473156902e-08, "loss": 0.2172, "step": 73340 }, { "epoch": 9.52, "learning_rate": 2.9490697549561324e-08, "loss": 0.2251, "step": 73350 }, { "epoch": 9.52, "learning_rate": 2.933183667723311e-08, "loss": 0.2302, "step": 73360 }, { "epoch": 9.52, "learning_rate": 2.9173402143847693e-08, "loss": 0.2243, "step": 73370 }, { "epoch": 9.52, "learning_rate": 2.9015393978590837e-08, "loss": 0.2317, "step": 73380 }, { "epoch": 9.52, "learning_rate": 2.885781221056869e-08, "loss": 0.2255, "step": 73390 }, { "epoch": 9.53, "learning_rate": 2.8700656868809856e-08, "loss": 0.2272, "step": 73400 }, { "epoch": 9.53, "learning_rate": 2.854392798226357e-08, "loss": 0.2157, "step": 73410 }, { "epoch": 9.53, "learning_rate": 2.838762557980101e-08, "loss": 0.2395, "step": 73420 }, { "epoch": 9.53, "learning_rate": 2.823174969021477e-08, "loss": 0.2179, "step": 73430 }, { "epoch": 9.53, "learning_rate": 2.8076300342218336e-08, "loss": 0.237, "step": 73440 }, { "epoch": 9.53, "learning_rate": 2.79212775644474e-08, "loss": 0.2311, "step": 73450 }, { "epoch": 9.53, "learning_rate": 2.7766681385458807e-08, "loss": 0.2262, "step": 73460 }, { "epoch": 9.54, "learning_rate": 2.7612511833730295e-08, "loss": 0.2327, "step": 73470 }, { "epoch": 9.54, "learning_rate": 2.7458768937661542e-08, "loss": 0.2252, "step": 73480 }, { "epoch": 9.54, "learning_rate": 2.7305452725573644e-08, "loss": 0.2196, "step": 73490 }, { "epoch": 9.54, "learning_rate": 2.7152563225709377e-08, "loss": 0.2197, "step": 73500 }, { "epoch": 9.54, "learning_rate": 2.7000100466231626e-08, "loss": 0.2279, "step": 73510 }, { "epoch": 9.54, "learning_rate": 2.684806447522626e-08, "loss": 0.2295, "step": 73520 }, { "epoch": 9.54, "learning_rate": 2.669645528069926e-08, "loss": 0.224, "step": 73530 }, { "epoch": 9.54, "learning_rate": 2.654527291057933e-08, "loss": 0.2168, "step": 73540 }, { "epoch": 9.55, "learning_rate": 2.6394517392715286e-08, "loss": 0.2321, "step": 73550 }, { "epoch": 9.55, "learning_rate": 2.624418875487762e-08, "loss": 0.2126, "step": 73560 }, { "epoch": 9.55, "learning_rate": 2.6094287024758764e-08, "loss": 0.2166, "step": 73570 }, { "epoch": 9.55, "learning_rate": 2.5944812229971786e-08, "loss": 0.2286, "step": 73580 }, { "epoch": 9.55, "learning_rate": 2.5795764398051694e-08, "loss": 0.2192, "step": 73590 }, { "epoch": 9.55, "learning_rate": 2.5647143556454652e-08, "loss": 0.2252, "step": 73600 }, { "epoch": 9.55, "learning_rate": 2.5498949732557714e-08, "loss": 0.2164, "step": 73610 }, { "epoch": 9.55, "learning_rate": 2.5351182953659878e-08, "loss": 0.219, "step": 73620 }, { "epoch": 9.56, "learning_rate": 2.52038432469813e-08, "loss": 0.2302, "step": 73630 }, { "epoch": 9.56, "learning_rate": 2.5056930639663544e-08, "loss": 0.2072, "step": 73640 }, { "epoch": 9.56, "learning_rate": 2.4910445158769082e-08, "loss": 0.2342, "step": 73650 }, { "epoch": 9.56, "learning_rate": 2.4764386831282317e-08, "loss": 0.2317, "step": 73660 }, { "epoch": 9.56, "learning_rate": 2.4618755684108292e-08, "loss": 0.2331, "step": 73670 }, { "epoch": 9.56, "learning_rate": 2.4473551744073982e-08, "loss": 0.2324, "step": 73680 }, { "epoch": 9.56, "learning_rate": 2.432877503792752e-08, "loss": 0.227, "step": 73690 }, { "epoch": 9.57, "learning_rate": 2.4184425592337684e-08, "loss": 0.2298, "step": 73700 }, { "epoch": 9.57, "learning_rate": 2.4040503433895695e-08, "loss": 0.2408, "step": 73710 }, { "epoch": 9.57, "learning_rate": 2.3897008589113167e-08, "loss": 0.2325, "step": 73720 }, { "epoch": 9.57, "learning_rate": 2.3753941084423377e-08, "loss": 0.2259, "step": 73730 }, { "epoch": 9.57, "learning_rate": 2.3611300946180505e-08, "loss": 0.2083, "step": 73740 }, { "epoch": 9.57, "learning_rate": 2.346908820066093e-08, "loss": 0.2262, "step": 73750 }, { "epoch": 9.57, "learning_rate": 2.3327302874060622e-08, "loss": 0.235, "step": 73760 }, { "epoch": 9.57, "learning_rate": 2.3185944992499052e-08, "loss": 0.2382, "step": 73770 }, { "epoch": 9.58, "learning_rate": 2.304501458201477e-08, "loss": 0.223, "step": 73780 }, { "epoch": 9.58, "learning_rate": 2.2904511668569036e-08, "loss": 0.2304, "step": 73790 }, { "epoch": 9.58, "learning_rate": 2.2764436278044016e-08, "loss": 0.2194, "step": 73800 }, { "epoch": 9.58, "learning_rate": 2.2624788436242503e-08, "loss": 0.2225, "step": 73810 }, { "epoch": 9.58, "learning_rate": 2.2485568168889752e-08, "loss": 0.2274, "step": 73820 }, { "epoch": 9.58, "learning_rate": 2.2346775501630877e-08, "loss": 0.2321, "step": 73830 }, { "epoch": 9.58, "learning_rate": 2.2208410460032925e-08, "loss": 0.2186, "step": 73840 }, { "epoch": 9.58, "learning_rate": 2.2070473069584362e-08, "loss": 0.2179, "step": 73850 }, { "epoch": 9.59, "learning_rate": 2.1932963355694807e-08, "loss": 0.2119, "step": 73860 }, { "epoch": 9.59, "learning_rate": 2.1795881343694266e-08, "loss": 0.2373, "step": 73870 }, { "epoch": 9.59, "learning_rate": 2.165922705883545e-08, "loss": 0.2275, "step": 73880 }, { "epoch": 9.59, "learning_rate": 2.152300052629093e-08, "loss": 0.2196, "step": 73890 }, { "epoch": 9.59, "learning_rate": 2.13872017711547e-08, "loss": 0.22, "step": 73900 }, { "epoch": 9.59, "learning_rate": 2.1251830818442944e-08, "loss": 0.2217, "step": 73910 }, { "epoch": 9.59, "learning_rate": 2.1116887693092223e-08, "loss": 0.2275, "step": 73920 }, { "epoch": 9.6, "learning_rate": 2.0982372419959744e-08, "loss": 0.2317, "step": 73930 }, { "epoch": 9.6, "learning_rate": 2.084828502382542e-08, "loss": 0.2263, "step": 73940 }, { "epoch": 9.6, "learning_rate": 2.0714625529388763e-08, "loss": 0.223, "step": 73950 }, { "epoch": 9.6, "learning_rate": 2.058139396127175e-08, "loss": 0.2082, "step": 73960 }, { "epoch": 9.6, "learning_rate": 2.044859034401672e-08, "loss": 0.2134, "step": 73970 }, { "epoch": 9.6, "learning_rate": 2.0316214702087442e-08, "loss": 0.205, "step": 73980 }, { "epoch": 9.6, "learning_rate": 2.0184267059868838e-08, "loss": 0.2312, "step": 73990 }, { "epoch": 9.6, "learning_rate": 2.0052747441666983e-08, "loss": 0.2256, "step": 74000 }, { "epoch": 9.61, "learning_rate": 1.9921655871709112e-08, "loss": 0.2228, "step": 74010 }, { "epoch": 9.61, "learning_rate": 1.979099237414362e-08, "loss": 0.2285, "step": 74020 }, { "epoch": 9.61, "learning_rate": 1.9660756973039793e-08, "loss": 0.2165, "step": 74030 }, { "epoch": 9.61, "learning_rate": 1.95309496923886e-08, "loss": 0.238, "step": 74040 }, { "epoch": 9.61, "learning_rate": 1.9401570556101895e-08, "loss": 0.2212, "step": 74050 }, { "epoch": 9.61, "learning_rate": 1.9272619588012443e-08, "loss": 0.2329, "step": 74060 }, { "epoch": 9.61, "learning_rate": 1.9144096811874158e-08, "loss": 0.2204, "step": 74070 }, { "epoch": 9.61, "learning_rate": 1.9016002251362374e-08, "loss": 0.2294, "step": 74080 }, { "epoch": 9.62, "learning_rate": 1.888833593007332e-08, "loss": 0.2239, "step": 74090 }, { "epoch": 9.62, "learning_rate": 1.8761097871524636e-08, "loss": 0.2278, "step": 74100 }, { "epoch": 9.62, "learning_rate": 1.8634288099154874e-08, "loss": 0.2221, "step": 74110 }, { "epoch": 9.62, "learning_rate": 1.8507906636322946e-08, "loss": 0.2141, "step": 74120 }, { "epoch": 9.62, "learning_rate": 1.838195350631049e-08, "loss": 0.2382, "step": 74130 }, { "epoch": 9.62, "learning_rate": 1.8256428732319003e-08, "loss": 0.2188, "step": 74140 }, { "epoch": 9.62, "learning_rate": 1.8131332337470872e-08, "loss": 0.2309, "step": 74150 }, { "epoch": 9.62, "learning_rate": 1.8006664344810946e-08, "loss": 0.2194, "step": 74160 }, { "epoch": 9.63, "learning_rate": 1.788242477730392e-08, "loss": 0.2214, "step": 74170 }, { "epoch": 9.63, "learning_rate": 1.7758613657835925e-08, "loss": 0.2349, "step": 74180 }, { "epoch": 9.63, "learning_rate": 1.7635231009214232e-08, "loss": 0.219, "step": 74190 }, { "epoch": 9.63, "learning_rate": 1.7512276854167278e-08, "loss": 0.2168, "step": 74200 }, { "epoch": 9.63, "learning_rate": 1.738975121534413e-08, "loss": 0.2201, "step": 74210 }, { "epoch": 9.63, "learning_rate": 1.7267654115315795e-08, "loss": 0.2343, "step": 74220 }, { "epoch": 9.63, "learning_rate": 1.714598557657314e-08, "loss": 0.2248, "step": 74230 }, { "epoch": 9.64, "learning_rate": 1.702474562152922e-08, "loss": 0.2368, "step": 74240 }, { "epoch": 9.64, "learning_rate": 1.6903934272517478e-08, "loss": 0.2229, "step": 74250 }, { "epoch": 9.64, "learning_rate": 1.6783551551792247e-08, "loss": 0.223, "step": 74260 }, { "epoch": 9.64, "learning_rate": 1.6663597481529797e-08, "loss": 0.2228, "step": 74270 }, { "epoch": 9.64, "learning_rate": 1.6544072083826773e-08, "loss": 0.2301, "step": 74280 }, { "epoch": 9.64, "learning_rate": 1.6424975380700465e-08, "loss": 0.2241, "step": 74290 }, { "epoch": 9.64, "learning_rate": 1.6306307394090088e-08, "loss": 0.2221, "step": 74300 }, { "epoch": 9.64, "learning_rate": 1.6188068145855243e-08, "loss": 0.2288, "step": 74310 }, { "epoch": 9.65, "learning_rate": 1.6070257657776945e-08, "loss": 0.2183, "step": 74320 }, { "epoch": 9.65, "learning_rate": 1.59528759515571e-08, "loss": 0.2298, "step": 74330 }, { "epoch": 9.65, "learning_rate": 1.583592304881852e-08, "loss": 0.2196, "step": 74340 }, { "epoch": 9.65, "learning_rate": 1.5719398971104912e-08, "loss": 0.2424, "step": 74350 }, { "epoch": 9.65, "learning_rate": 1.5603303739881653e-08, "loss": 0.2275, "step": 74360 }, { "epoch": 9.65, "learning_rate": 1.5487637376534242e-08, "loss": 0.2306, "step": 74370 }, { "epoch": 9.65, "learning_rate": 1.537239990236959e-08, "loss": 0.2354, "step": 74380 }, { "epoch": 9.65, "learning_rate": 1.5257591338615518e-08, "loss": 0.2296, "step": 74390 }, { "epoch": 9.66, "learning_rate": 1.5143211706421243e-08, "loss": 0.2194, "step": 74400 }, { "epoch": 9.66, "learning_rate": 1.5029261026856634e-08, "loss": 0.224, "step": 74410 }, { "epoch": 9.66, "learning_rate": 1.4915739320912195e-08, "loss": 0.2238, "step": 74420 }, { "epoch": 9.66, "learning_rate": 1.480264660949984e-08, "loss": 0.2286, "step": 74430 }, { "epoch": 9.66, "learning_rate": 1.4689982913452646e-08, "loss": 0.2185, "step": 74440 }, { "epoch": 9.66, "learning_rate": 1.4577748253524324e-08, "loss": 0.2314, "step": 74450 }, { "epoch": 9.66, "learning_rate": 1.446594265038922e-08, "loss": 0.2219, "step": 74460 }, { "epoch": 9.67, "learning_rate": 1.4354566124643359e-08, "loss": 0.2265, "step": 74470 }, { "epoch": 9.67, "learning_rate": 1.4243618696803666e-08, "loss": 0.2464, "step": 74480 }, { "epoch": 9.67, "learning_rate": 1.4133100387307174e-08, "loss": 0.2253, "step": 74490 }, { "epoch": 9.67, "learning_rate": 1.402301121651286e-08, "loss": 0.2295, "step": 74500 }, { "epoch": 9.67, "learning_rate": 1.3913351204700336e-08, "loss": 0.2116, "step": 74510 }, { "epoch": 9.67, "learning_rate": 1.3804120372069849e-08, "loss": 0.2209, "step": 74520 }, { "epoch": 9.67, "learning_rate": 1.3695318738742803e-08, "loss": 0.221, "step": 74530 }, { "epoch": 9.67, "learning_rate": 1.35869463247615e-08, "loss": 0.2326, "step": 74540 }, { "epoch": 9.68, "learning_rate": 1.3479003150089398e-08, "loss": 0.2474, "step": 74550 }, { "epoch": 9.68, "learning_rate": 1.337148923461059e-08, "loss": 0.2192, "step": 74560 }, { "epoch": 9.68, "learning_rate": 1.3264404598130329e-08, "loss": 0.2317, "step": 74570 }, { "epoch": 9.68, "learning_rate": 1.3157749260374502e-08, "loss": 0.2147, "step": 74580 }, { "epoch": 9.68, "learning_rate": 1.3051523240990413e-08, "loss": 0.2211, "step": 74590 }, { "epoch": 9.68, "learning_rate": 1.2945726559545483e-08, "loss": 0.218, "step": 74600 }, { "epoch": 9.68, "learning_rate": 1.284035923552881e-08, "loss": 0.2387, "step": 74610 }, { "epoch": 9.68, "learning_rate": 1.2735421288350127e-08, "loss": 0.2262, "step": 74620 }, { "epoch": 9.69, "learning_rate": 1.2630912737339807e-08, "loss": 0.2383, "step": 74630 }, { "epoch": 9.69, "learning_rate": 1.2526833601749896e-08, "loss": 0.2278, "step": 74640 }, { "epoch": 9.69, "learning_rate": 1.2423183900752298e-08, "loss": 0.2419, "step": 74650 }, { "epoch": 9.69, "learning_rate": 1.2319963653440335e-08, "loss": 0.2435, "step": 74660 }, { "epoch": 9.69, "learning_rate": 1.2217172878828483e-08, "loss": 0.2272, "step": 74670 }, { "epoch": 9.69, "learning_rate": 1.2114811595851596e-08, "loss": 0.2232, "step": 74680 }, { "epoch": 9.69, "learning_rate": 1.2012879823365944e-08, "loss": 0.2258, "step": 74690 }, { "epoch": 9.7, "learning_rate": 1.1911377580147916e-08, "loss": 0.2258, "step": 74700 }, { "epoch": 9.7, "learning_rate": 1.1810304884895833e-08, "loss": 0.2467, "step": 74710 }, { "epoch": 9.7, "learning_rate": 1.1709661756227877e-08, "loss": 0.2136, "step": 74720 }, { "epoch": 9.7, "learning_rate": 1.1609448212683386e-08, "loss": 0.2232, "step": 74730 }, { "epoch": 9.7, "learning_rate": 1.150966427272311e-08, "loss": 0.2215, "step": 74740 }, { "epoch": 9.7, "learning_rate": 1.141030995472818e-08, "loss": 0.2275, "step": 74750 }, { "epoch": 9.7, "learning_rate": 1.1311385277000367e-08, "loss": 0.2301, "step": 74760 }, { "epoch": 9.7, "learning_rate": 1.1212890257762848e-08, "loss": 0.2252, "step": 74770 }, { "epoch": 9.71, "learning_rate": 1.1114824915159186e-08, "loss": 0.229, "step": 74780 }, { "epoch": 9.71, "learning_rate": 1.1017189267254096e-08, "loss": 0.2045, "step": 74790 }, { "epoch": 9.71, "learning_rate": 1.0919983332033192e-08, "loss": 0.2286, "step": 74800 }, { "epoch": 9.71, "learning_rate": 1.0823207127402464e-08, "loss": 0.2271, "step": 74810 }, { "epoch": 9.71, "learning_rate": 1.0726860671189317e-08, "loss": 0.2146, "step": 74820 }, { "epoch": 9.71, "learning_rate": 1.0630943981141532e-08, "loss": 0.2249, "step": 74830 }, { "epoch": 9.71, "learning_rate": 1.0535457074928311e-08, "loss": 0.2268, "step": 74840 }, { "epoch": 9.71, "learning_rate": 1.0440399970138707e-08, "loss": 0.2287, "step": 74850 }, { "epoch": 9.72, "learning_rate": 1.0345772684283715e-08, "loss": 0.224, "step": 74860 }, { "epoch": 9.72, "learning_rate": 1.0251575234794183e-08, "loss": 0.2239, "step": 74870 }, { "epoch": 9.72, "learning_rate": 1.0157807639022633e-08, "loss": 0.2335, "step": 74880 }, { "epoch": 9.72, "learning_rate": 1.0064469914241712e-08, "loss": 0.2301, "step": 74890 }, { "epoch": 9.72, "learning_rate": 9.971562077645215e-09, "loss": 0.2093, "step": 74900 }, { "epoch": 9.72, "learning_rate": 9.879084146347839e-09, "loss": 0.2261, "step": 74910 }, { "epoch": 9.72, "learning_rate": 9.787036137384914e-09, "loss": 0.2187, "step": 74920 }, { "epoch": 9.72, "learning_rate": 9.695418067712409e-09, "loss": 0.2273, "step": 74930 }, { "epoch": 9.73, "learning_rate": 9.604229954207187e-09, "loss": 0.2264, "step": 74940 }, { "epoch": 9.73, "learning_rate": 9.51347181366753e-09, "loss": 0.2442, "step": 74950 }, { "epoch": 9.73, "learning_rate": 9.423143662811578e-09, "loss": 0.2297, "step": 74960 }, { "epoch": 9.73, "learning_rate": 9.33324551827888e-09, "loss": 0.2195, "step": 74970 }, { "epoch": 9.73, "learning_rate": 9.243777396629373e-09, "loss": 0.2271, "step": 74980 }, { "epoch": 9.73, "learning_rate": 9.154739314344139e-09, "loss": 0.2336, "step": 74990 }, { "epoch": 9.73, "learning_rate": 9.066131287824643e-09, "loss": 0.2168, "step": 75000 }, { "epoch": 9.74, "learning_rate": 8.977953333393767e-09, "loss": 0.2323, "step": 75010 }, { "epoch": 9.74, "learning_rate": 8.890205467294506e-09, "loss": 0.2222, "step": 75020 }, { "epoch": 9.74, "learning_rate": 8.802887705691014e-09, "loss": 0.2382, "step": 75030 }, { "epoch": 9.74, "learning_rate": 8.71600006466808e-09, "loss": 0.2359, "step": 75040 }, { "epoch": 9.74, "learning_rate": 8.629542560230872e-09, "loss": 0.2214, "step": 75050 }, { "epoch": 9.74, "learning_rate": 8.543515208306491e-09, "loss": 0.2122, "step": 75060 }, { "epoch": 9.74, "learning_rate": 8.457918024741376e-09, "loss": 0.2268, "step": 75070 }, { "epoch": 9.74, "learning_rate": 8.372751025303645e-09, "loss": 0.2391, "step": 75080 }, { "epoch": 9.75, "learning_rate": 8.288014225682048e-09, "loss": 0.2247, "step": 75090 }, { "epoch": 9.75, "learning_rate": 8.203707641485716e-09, "loss": 0.2347, "step": 75100 }, { "epoch": 9.75, "learning_rate": 8.119831288244931e-09, "loss": 0.2327, "step": 75110 }, { "epoch": 9.75, "learning_rate": 8.036385181410358e-09, "loss": 0.2176, "step": 75120 }, { "epoch": 9.75, "learning_rate": 7.953369336353554e-09, "loss": 0.2385, "step": 75130 }, { "epoch": 9.75, "learning_rate": 7.870783768367232e-09, "loss": 0.2216, "step": 75140 }, { "epoch": 9.75, "learning_rate": 7.788628492664227e-09, "loss": 0.2167, "step": 75150 }, { "epoch": 9.75, "learning_rate": 7.706903524378523e-09, "loss": 0.2203, "step": 75160 }, { "epoch": 9.76, "learning_rate": 7.625608878564487e-09, "loss": 0.2149, "step": 75170 }, { "epoch": 9.76, "learning_rate": 7.544744570197378e-09, "loss": 0.2124, "step": 75180 }, { "epoch": 9.76, "learning_rate": 7.464310614173352e-09, "loss": 0.2337, "step": 75190 }, { "epoch": 9.76, "learning_rate": 7.384307025309205e-09, "loss": 0.232, "step": 75200 }, { "epoch": 9.76, "learning_rate": 7.304733818342106e-09, "loss": 0.2156, "step": 75210 }, { "epoch": 9.76, "learning_rate": 7.2255910079303815e-09, "loss": 0.2178, "step": 75220 }, { "epoch": 9.76, "learning_rate": 7.146878608653254e-09, "loss": 0.2347, "step": 75230 }, { "epoch": 9.77, "learning_rate": 7.068596635009805e-09, "loss": 0.224, "step": 75240 }, { "epoch": 9.77, "learning_rate": 6.990745101420787e-09, "loss": 0.2332, "step": 75250 }, { "epoch": 9.77, "learning_rate": 6.9133240222270754e-09, "loss": 0.2326, "step": 75260 }, { "epoch": 9.77, "learning_rate": 6.8363334116901775e-09, "loss": 0.2233, "step": 75270 }, { "epoch": 9.77, "learning_rate": 6.759773283992759e-09, "loss": 0.2234, "step": 75280 }, { "epoch": 9.77, "learning_rate": 6.683643653238122e-09, "loss": 0.2116, "step": 75290 }, { "epoch": 9.77, "learning_rate": 6.607944533449684e-09, "loss": 0.2321, "step": 75300 }, { "epoch": 9.77, "learning_rate": 6.53267593857228e-09, "loss": 0.2325, "step": 75310 }, { "epoch": 9.78, "learning_rate": 6.457837882471121e-09, "loss": 0.2169, "step": 75320 }, { "epoch": 9.78, "learning_rate": 6.383430378932055e-09, "loss": 0.2229, "step": 75330 }, { "epoch": 9.78, "learning_rate": 6.309453441661564e-09, "loss": 0.2119, "step": 75340 }, { "epoch": 9.78, "learning_rate": 6.23590708428729e-09, "loss": 0.2376, "step": 75350 }, { "epoch": 9.78, "learning_rate": 6.162791320356731e-09, "loss": 0.2435, "step": 75360 }, { "epoch": 9.78, "learning_rate": 6.0901061633390585e-09, "loss": 0.2286, "step": 75370 }, { "epoch": 9.78, "learning_rate": 6.017851626623302e-09, "loss": 0.2315, "step": 75380 }, { "epoch": 9.78, "learning_rate": 5.946027723519389e-09, "loss": 0.2173, "step": 75390 }, { "epoch": 9.79, "learning_rate": 5.874634467258402e-09, "loss": 0.235, "step": 75400 }, { "epoch": 9.79, "learning_rate": 5.803671870991018e-09, "loss": 0.213, "step": 75410 }, { "epoch": 9.79, "learning_rate": 5.733139947789854e-09, "loss": 0.2286, "step": 75420 }, { "epoch": 9.79, "learning_rate": 5.663038710647643e-09, "loss": 0.2213, "step": 75430 }, { "epoch": 9.79, "learning_rate": 5.593368172476971e-09, "loss": 0.2314, "step": 75440 }, { "epoch": 9.79, "learning_rate": 5.524128346112884e-09, "loss": 0.2088, "step": 75450 }, { "epoch": 9.79, "learning_rate": 5.455319244309243e-09, "loss": 0.2124, "step": 75460 }, { "epoch": 9.79, "learning_rate": 5.386940879741585e-09, "loss": 0.2443, "step": 75470 }, { "epoch": 9.8, "learning_rate": 5.3189932650060835e-09, "loss": 0.2173, "step": 75480 }, { "epoch": 9.8, "learning_rate": 5.25147641261929e-09, "loss": 0.2212, "step": 75490 }, { "epoch": 9.8, "learning_rate": 5.184390335018392e-09, "loss": 0.2159, "step": 75500 }, { "epoch": 9.8, "learning_rate": 5.11773504456121e-09, "loss": 0.226, "step": 75510 }, { "epoch": 9.8, "learning_rate": 5.051510553526726e-09, "loss": 0.2108, "step": 75520 }, { "epoch": 9.8, "learning_rate": 4.9857168741137755e-09, "loss": 0.2101, "step": 75530 }, { "epoch": 9.8, "learning_rate": 4.92035401844235e-09, "loss": 0.237, "step": 75540 }, { "epoch": 9.81, "learning_rate": 4.8554219985528186e-09, "loss": 0.2254, "step": 75550 }, { "epoch": 9.81, "learning_rate": 4.790920826406447e-09, "loss": 0.2255, "step": 75560 }, { "epoch": 9.81, "learning_rate": 4.726850513884876e-09, "loss": 0.2138, "step": 75570 }, { "epoch": 9.81, "learning_rate": 4.663211072790645e-09, "loss": 0.2214, "step": 75580 }, { "epoch": 9.81, "learning_rate": 4.6000025148466676e-09, "loss": 0.2242, "step": 75590 }, { "epoch": 9.81, "learning_rate": 4.5372248516967535e-09, "loss": 0.215, "step": 75600 }, { "epoch": 9.81, "learning_rate": 4.474878094905092e-09, "loss": 0.2442, "step": 75610 }, { "epoch": 9.81, "learning_rate": 4.412962255956505e-09, "loss": 0.2343, "step": 75620 }, { "epoch": 9.82, "learning_rate": 4.351477346256453e-09, "loss": 0.2166, "step": 75630 }, { "epoch": 9.82, "learning_rate": 4.290423377131033e-09, "loss": 0.2357, "step": 75640 }, { "epoch": 9.82, "learning_rate": 4.229800359827236e-09, "loss": 0.2352, "step": 75650 }, { "epoch": 9.82, "learning_rate": 4.169608305512433e-09, "loss": 0.224, "step": 75660 }, { "epoch": 9.82, "learning_rate": 4.109847225274368e-09, "loss": 0.2122, "step": 75670 }, { "epoch": 9.82, "learning_rate": 4.050517130121684e-09, "loss": 0.2369, "step": 75680 }, { "epoch": 9.82, "learning_rate": 3.991618030983659e-09, "loss": 0.2104, "step": 75690 }, { "epoch": 9.82, "learning_rate": 3.9331499387102066e-09, "loss": 0.2299, "step": 75700 }, { "epoch": 9.83, "learning_rate": 3.87511286407162e-09, "loss": 0.232, "step": 75710 }, { "epoch": 9.83, "learning_rate": 3.817506817758563e-09, "loss": 0.2213, "step": 75720 }, { "epoch": 9.83, "learning_rate": 3.760331810383384e-09, "loss": 0.2214, "step": 75730 }, { "epoch": 9.83, "learning_rate": 3.703587852477499e-09, "loss": 0.2222, "step": 75740 }, { "epoch": 9.83, "learning_rate": 3.647274954494267e-09, "loss": 0.2228, "step": 75750 }, { "epoch": 9.83, "learning_rate": 3.5913931268068996e-09, "loss": 0.2155, "step": 75760 }, { "epoch": 9.83, "learning_rate": 3.535942379709507e-09, "loss": 0.2215, "step": 75770 }, { "epoch": 9.84, "learning_rate": 3.4809227234163136e-09, "loss": 0.2256, "step": 75780 }, { "epoch": 9.84, "learning_rate": 3.426334168062962e-09, "loss": 0.2374, "step": 75790 }, { "epoch": 9.84, "learning_rate": 3.3721767237049508e-09, "loss": 0.2307, "step": 75800 }, { "epoch": 9.84, "learning_rate": 3.318450400318674e-09, "loss": 0.2155, "step": 75810 }, { "epoch": 9.84, "learning_rate": 3.2651552078009034e-09, "loss": 0.2291, "step": 75820 }, { "epoch": 9.84, "learning_rate": 3.2122911559693057e-09, "loss": 0.2201, "step": 75830 }, { "epoch": 9.84, "learning_rate": 3.159858254561925e-09, "loss": 0.2289, "step": 75840 }, { "epoch": 9.84, "learning_rate": 3.107856513237441e-09, "loss": 0.2072, "step": 75850 }, { "epoch": 9.85, "learning_rate": 3.0562859415751708e-09, "loss": 0.2245, "step": 75860 }, { "epoch": 9.85, "learning_rate": 3.0051465490748066e-09, "loss": 0.2379, "step": 75870 }, { "epoch": 9.85, "learning_rate": 2.9544383451569376e-09, "loss": 0.2171, "step": 75880 }, { "epoch": 9.85, "learning_rate": 2.9041613391620096e-09, "loss": 0.2201, "step": 75890 }, { "epoch": 9.85, "learning_rate": 2.8543155403521445e-09, "loss": 0.2146, "step": 75900 }, { "epoch": 9.85, "learning_rate": 2.8049009579093203e-09, "loss": 0.2294, "step": 75910 }, { "epoch": 9.85, "learning_rate": 2.7559176009358914e-09, "loss": 0.2329, "step": 75920 }, { "epoch": 9.85, "learning_rate": 2.7073654784551083e-09, "loss": 0.2301, "step": 75930 }, { "epoch": 9.86, "learning_rate": 2.659244599411119e-09, "loss": 0.2051, "step": 75940 }, { "epoch": 9.86, "learning_rate": 2.6115549726681874e-09, "loss": 0.2308, "step": 75950 }, { "epoch": 9.86, "learning_rate": 2.564296607010952e-09, "loss": 0.2083, "step": 75960 }, { "epoch": 9.86, "learning_rate": 2.5174695111449504e-09, "loss": 0.2185, "step": 75970 }, { "epoch": 9.86, "learning_rate": 2.471073693696094e-09, "loss": 0.2344, "step": 75980 }, { "epoch": 9.86, "learning_rate": 2.4251091632114527e-09, "loss": 0.2228, "step": 75990 }, { "epoch": 9.86, "learning_rate": 2.3795759281576914e-09, "loss": 0.2144, "step": 76000 }, { "epoch": 9.87, "learning_rate": 2.334473996922631e-09, "loss": 0.2277, "step": 76010 }, { "epoch": 9.87, "learning_rate": 2.2898033778147305e-09, "loss": 0.2369, "step": 76020 }, { "epoch": 9.87, "learning_rate": 2.2455640790623037e-09, "loss": 0.2197, "step": 76030 }, { "epoch": 9.87, "learning_rate": 2.201756108814821e-09, "loss": 0.2398, "step": 76040 }, { "epoch": 9.87, "learning_rate": 2.1583794751423888e-09, "loss": 0.2312, "step": 76050 }, { "epoch": 9.87, "learning_rate": 2.1154341860349712e-09, "loss": 0.2262, "step": 76060 }, { "epoch": 9.87, "learning_rate": 2.072920249404208e-09, "loss": 0.2304, "step": 76070 }, { "epoch": 9.87, "learning_rate": 2.0308376730808137e-09, "loss": 0.2225, "step": 76080 }, { "epoch": 9.88, "learning_rate": 1.9891864648174418e-09, "loss": 0.2356, "step": 76090 }, { "epoch": 9.88, "learning_rate": 1.94796663228608e-09, "loss": 0.2277, "step": 76100 }, { "epoch": 9.88, "learning_rate": 1.9071781830803935e-09, "loss": 0.2207, "step": 76110 }, { "epoch": 9.88, "learning_rate": 1.8668211247139034e-09, "loss": 0.2121, "step": 76120 }, { "epoch": 9.88, "learning_rate": 1.826895464620507e-09, "loss": 0.2358, "step": 76130 }, { "epoch": 9.88, "learning_rate": 1.7874012101549984e-09, "loss": 0.2217, "step": 76140 }, { "epoch": 9.88, "learning_rate": 1.7483383685928075e-09, "loss": 0.217, "step": 76150 }, { "epoch": 9.88, "learning_rate": 1.7097069471294808e-09, "loss": 0.2424, "step": 76160 }, { "epoch": 9.89, "learning_rate": 1.6715069528814607e-09, "loss": 0.2254, "step": 76170 }, { "epoch": 9.89, "learning_rate": 1.6337383928855669e-09, "loss": 0.2195, "step": 76180 }, { "epoch": 9.89, "learning_rate": 1.5964012740989941e-09, "loss": 0.222, "step": 76190 }, { "epoch": 9.89, "learning_rate": 1.5594956033998344e-09, "loss": 0.2324, "step": 76200 }, { "epoch": 9.89, "learning_rate": 1.5230213875860355e-09, "loss": 0.2439, "step": 76210 }, { "epoch": 9.89, "learning_rate": 1.486978633377222e-09, "loss": 0.224, "step": 76220 }, { "epoch": 9.89, "learning_rate": 1.4513673474123542e-09, "loss": 0.2506, "step": 76230 }, { "epoch": 9.89, "learning_rate": 1.416187536251289e-09, "loss": 0.2323, "step": 76240 }, { "epoch": 9.9, "learning_rate": 1.38143920637478e-09, "loss": 0.2237, "step": 76250 }, { "epoch": 9.9, "learning_rate": 1.3471223641836963e-09, "loss": 0.2343, "step": 76260 }, { "epoch": 9.9, "learning_rate": 1.3132370159995437e-09, "loss": 0.2178, "step": 76270 }, { "epoch": 9.9, "learning_rate": 1.2797831680644645e-09, "loss": 0.2247, "step": 76280 }, { "epoch": 9.9, "learning_rate": 1.2467608265409775e-09, "loss": 0.2184, "step": 76290 }, { "epoch": 9.9, "learning_rate": 1.2141699975119767e-09, "loss": 0.2163, "step": 76300 }, { "epoch": 9.9, "learning_rate": 1.1820106869809935e-09, "loss": 0.216, "step": 76310 }, { "epoch": 9.91, "learning_rate": 1.1502829008721951e-09, "loss": 0.2437, "step": 76320 }, { "epoch": 9.91, "learning_rate": 1.1189866450301254e-09, "loss": 0.2261, "step": 76330 }, { "epoch": 9.91, "learning_rate": 1.088121925219964e-09, "loss": 0.227, "step": 76340 }, { "epoch": 9.91, "learning_rate": 1.057688747127007e-09, "loss": 0.2334, "step": 76350 }, { "epoch": 9.91, "learning_rate": 1.0276871163579676e-09, "loss": 0.2341, "step": 76360 }, { "epoch": 9.91, "learning_rate": 9.981170384386338e-10, "loss": 0.2087, "step": 76370 }, { "epoch": 9.91, "learning_rate": 9.689785188167313e-10, "loss": 0.2283, "step": 76380 }, { "epoch": 9.91, "learning_rate": 9.402715628598415e-10, "loss": 0.2257, "step": 76390 }, { "epoch": 9.92, "learning_rate": 9.119961758556616e-10, "loss": 0.2216, "step": 76400 }, { "epoch": 9.92, "learning_rate": 8.841523630130458e-10, "loss": 0.2238, "step": 76410 }, { "epoch": 9.92, "learning_rate": 8.567401294609644e-10, "loss": 0.2208, "step": 76420 }, { "epoch": 9.92, "learning_rate": 8.297594802492841e-10, "loss": 0.244, "step": 76430 }, { "epoch": 9.92, "learning_rate": 8.032104203479876e-10, "loss": 0.224, "step": 76440 }, { "epoch": 9.92, "learning_rate": 7.77092954647434e-10, "loss": 0.2382, "step": 76450 }, { "epoch": 9.92, "learning_rate": 7.51407087959139e-10, "loss": 0.232, "step": 76460 }, { "epoch": 9.92, "learning_rate": 7.261528250144741e-10, "loss": 0.2141, "step": 76470 }, { "epoch": 9.93, "learning_rate": 7.013301704657075e-10, "loss": 0.2446, "step": 76480 }, { "epoch": 9.93, "learning_rate": 6.769391288852233e-10, "loss": 0.2299, "step": 76490 }, { "epoch": 9.93, "learning_rate": 6.529797047660421e-10, "loss": 0.2323, "step": 76500 }, { "epoch": 9.93, "learning_rate": 6.294519025218207e-10, "loss": 0.2374, "step": 76510 }, { "epoch": 9.93, "learning_rate": 6.063557264865924e-10, "loss": 0.2301, "step": 76520 }, { "epoch": 9.93, "learning_rate": 5.836911809147665e-10, "loss": 0.22, "step": 76530 }, { "epoch": 9.93, "learning_rate": 5.614582699819091e-10, "loss": 0.2139, "step": 76540 }, { "epoch": 9.94, "learning_rate": 5.396569977829217e-10, "loss": 0.2239, "step": 76550 }, { "epoch": 9.94, "learning_rate": 5.182873683341228e-10, "loss": 0.2264, "step": 76560 }, { "epoch": 9.94, "learning_rate": 4.973493855722071e-10, "loss": 0.2209, "step": 76570 }, { "epoch": 9.94, "learning_rate": 4.768430533537252e-10, "loss": 0.2321, "step": 76580 }, { "epoch": 9.94, "learning_rate": 4.5676837545612426e-10, "loss": 0.2195, "step": 76590 }, { "epoch": 9.94, "learning_rate": 4.3712535557774777e-10, "loss": 0.2223, "step": 76600 }, { "epoch": 9.94, "learning_rate": 4.179139973370556e-10, "loss": 0.2217, "step": 76610 }, { "epoch": 9.94, "learning_rate": 3.991343042726233e-10, "loss": 0.2367, "step": 76620 }, { "epoch": 9.95, "learning_rate": 3.8078627984392333e-10, "loss": 0.2164, "step": 76630 }, { "epoch": 9.95, "learning_rate": 3.6286992743106426e-10, "loss": 0.2405, "step": 76640 }, { "epoch": 9.95, "learning_rate": 3.453852503342708e-10, "loss": 0.2277, "step": 76650 }, { "epoch": 9.95, "learning_rate": 3.28332251774404e-10, "loss": 0.2291, "step": 76660 }, { "epoch": 9.95, "learning_rate": 3.1171093489270116e-10, "loss": 0.2265, "step": 76670 }, { "epoch": 9.95, "learning_rate": 2.955213027512962e-10, "loss": 0.2156, "step": 76680 }, { "epoch": 9.95, "learning_rate": 2.7976335833191854e-10, "loss": 0.2159, "step": 76690 }, { "epoch": 9.95, "learning_rate": 2.644371045379748e-10, "loss": 0.2274, "step": 76700 }, { "epoch": 9.96, "learning_rate": 2.4954254419246727e-10, "loss": 0.2265, "step": 76710 }, { "epoch": 9.96, "learning_rate": 2.350796800387744e-10, "loss": 0.2259, "step": 76720 }, { "epoch": 9.96, "learning_rate": 2.210485147416917e-10, "loss": 0.2325, "step": 76730 }, { "epoch": 9.96, "learning_rate": 2.0744905088561012e-10, "loss": 0.2199, "step": 76740 }, { "epoch": 9.96, "learning_rate": 1.9428129097555715e-10, "loss": 0.2162, "step": 76750 }, { "epoch": 9.96, "learning_rate": 1.8154523743719659e-10, "loss": 0.221, "step": 76760 }, { "epoch": 9.96, "learning_rate": 1.692408926168287e-10, "loss": 0.2087, "step": 76770 }, { "epoch": 9.96, "learning_rate": 1.5736825878112993e-10, "loss": 0.2241, "step": 76780 }, { "epoch": 9.97, "learning_rate": 1.4592733811689267e-10, "loss": 0.231, "step": 76790 }, { "epoch": 9.97, "learning_rate": 1.3491813273154573e-10, "loss": 0.2363, "step": 76800 }, { "epoch": 9.97, "learning_rate": 1.2434064465341454e-10, "loss": 0.2296, "step": 76810 }, { "epoch": 9.97, "learning_rate": 1.141948758309405e-10, "loss": 0.2328, "step": 76820 }, { "epoch": 9.97, "learning_rate": 1.0448082813294123e-10, "loss": 0.2274, "step": 76830 }, { "epoch": 9.97, "learning_rate": 9.519850334861051e-11, "loss": 0.217, "step": 76840 }, { "epoch": 9.97, "learning_rate": 8.634790318855917e-11, "loss": 0.2383, "step": 76850 }, { "epoch": 9.98, "learning_rate": 7.792902928247316e-11, "loss": 0.242, "step": 76860 }, { "epoch": 9.98, "learning_rate": 6.994188318119528e-11, "loss": 0.229, "step": 76870 }, { "epoch": 9.98, "learning_rate": 6.238646635672515e-11, "loss": 0.2216, "step": 76880 }, { "epoch": 9.98, "learning_rate": 5.5262780200137494e-11, "loss": 0.2234, "step": 76890 }, { "epoch": 9.98, "learning_rate": 4.857082602392412e-11, "loss": 0.222, "step": 76900 }, { "epoch": 9.98, "learning_rate": 4.231060506095299e-11, "loss": 0.226, "step": 76910 }, { "epoch": 9.98, "learning_rate": 3.648211846420807e-11, "loss": 0.2424, "step": 76920 }, { "epoch": 9.98, "learning_rate": 3.108536730730971e-11, "loss": 0.2294, "step": 76930 }, { "epoch": 9.99, "learning_rate": 2.6120352584774877e-11, "loss": 0.2316, "step": 76940 }, { "epoch": 9.99, "learning_rate": 2.158707521071612e-11, "loss": 0.2264, "step": 76950 }, { "epoch": 9.99, "learning_rate": 1.7485536020662995e-11, "loss": 0.2327, "step": 76960 }, { "epoch": 9.99, "learning_rate": 1.3815735770000836e-11, "loss": 0.2248, "step": 76970 }, { "epoch": 9.99, "learning_rate": 1.0577675134491176e-11, "loss": 0.2305, "step": 76980 }, { "epoch": 9.99, "learning_rate": 7.771354710792155e-12, "loss": 0.2151, "step": 76990 }, { "epoch": 9.99, "learning_rate": 5.396775016198313e-12, "loss": 0.222, "step": 77000 }, { "epoch": 9.99, "learning_rate": 3.4539364875997566e-12, "loss": 0.2222, "step": 77010 }, { "epoch": 10.0, "learning_rate": 1.9428394830434125e-12, "loss": 0.2164, "step": 77020 }, { "epoch": 10.0, "learning_rate": 8.634842809523979e-13, "loss": 0.2357, "step": 77030 }, { "epoch": 10.0, "learning_rate": 2.1587108012602327e-13, "loss": 0.2313, "step": 77040 }, { "epoch": 10.0, "learning_rate": 0.0, "loss": 0.2175, "step": 77050 }, { "epoch": 10.0, "eval_loss": 0.22619213163852692, "eval_runtime": 60.289, "eval_samples_per_second": 180.414, "eval_steps_per_second": 22.558, "step": 77050 }, { "epoch": 10.0, "step": 77050, "total_flos": 6.229344692427817e+19, "train_loss": 0.2411320797999442, "train_runtime": 9763.9799, "train_samples_per_second": 63.123, "train_steps_per_second": 7.891 } ], "max_steps": 77050, "num_train_epochs": 10, "total_flos": 6.229344692427817e+19, "trial_name": null, "trial_params": null }