{ "best_metric": null, "best_model_checkpoint": null, "epoch": 4.99752842313396, "eval_steps": 500, "global_step": 5055, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 5.928853754940712e-07, "loss": 2.4758, "step": 10 }, { "epoch": 0.02, "learning_rate": 1.1857707509881424e-06, "loss": 2.4692, "step": 20 }, { "epoch": 0.03, "learning_rate": 1.7786561264822134e-06, "loss": 2.4516, "step": 30 }, { "epoch": 0.04, "learning_rate": 2.371541501976285e-06, "loss": 2.4223, "step": 40 }, { "epoch": 0.05, "learning_rate": 2.9644268774703556e-06, "loss": 2.3921, "step": 50 }, { "epoch": 0.06, "learning_rate": 3.557312252964427e-06, "loss": 2.3448, "step": 60 }, { "epoch": 0.07, "learning_rate": 4.150197628458498e-06, "loss": 2.2757, "step": 70 }, { "epoch": 0.08, "learning_rate": 4.74308300395257e-06, "loss": 2.1523, "step": 80 }, { "epoch": 0.09, "learning_rate": 5.3359683794466405e-06, "loss": 2.0274, "step": 90 }, { "epoch": 0.1, "learning_rate": 5.928853754940711e-06, "loss": 1.8621, "step": 100 }, { "epoch": 0.11, "learning_rate": 6.521739130434783e-06, "loss": 1.7931, "step": 110 }, { "epoch": 0.12, "learning_rate": 7.114624505928854e-06, "loss": 1.7812, "step": 120 }, { "epoch": 0.13, "learning_rate": 7.707509881422925e-06, "loss": 1.6568, "step": 130 }, { "epoch": 0.14, "learning_rate": 8.300395256916996e-06, "loss": 1.6569, "step": 140 }, { "epoch": 0.15, "learning_rate": 8.893280632411067e-06, "loss": 1.6238, "step": 150 }, { "epoch": 0.16, "learning_rate": 9.48616600790514e-06, "loss": 1.5771, "step": 160 }, { "epoch": 0.17, "learning_rate": 1.007905138339921e-05, "loss": 1.5855, "step": 170 }, { "epoch": 0.18, "learning_rate": 1.0671936758893281e-05, "loss": 1.6522, "step": 180 }, { "epoch": 0.19, "learning_rate": 1.1264822134387352e-05, "loss": 1.5264, "step": 190 }, { "epoch": 0.2, "learning_rate": 1.1857707509881423e-05, "loss": 1.5091, "step": 200 }, { "epoch": 0.21, "learning_rate": 1.2450592885375493e-05, "loss": 1.5105, "step": 210 }, { "epoch": 0.22, "learning_rate": 1.3043478260869566e-05, "loss": 1.4426, "step": 220 }, { "epoch": 0.23, "learning_rate": 1.3636363636363637e-05, "loss": 1.4732, "step": 230 }, { "epoch": 0.24, "learning_rate": 1.4229249011857707e-05, "loss": 1.47, "step": 240 }, { "epoch": 0.25, "learning_rate": 1.4822134387351778e-05, "loss": 1.5379, "step": 250 }, { "epoch": 0.26, "learning_rate": 1.541501976284585e-05, "loss": 1.4277, "step": 260 }, { "epoch": 0.27, "learning_rate": 1.600790513833992e-05, "loss": 1.4962, "step": 270 }, { "epoch": 0.28, "learning_rate": 1.6600790513833992e-05, "loss": 1.451, "step": 280 }, { "epoch": 0.29, "learning_rate": 1.7193675889328063e-05, "loss": 1.3834, "step": 290 }, { "epoch": 0.3, "learning_rate": 1.7786561264822134e-05, "loss": 1.4241, "step": 300 }, { "epoch": 0.31, "learning_rate": 1.8379446640316208e-05, "loss": 1.3071, "step": 310 }, { "epoch": 0.32, "learning_rate": 1.897233201581028e-05, "loss": 1.3175, "step": 320 }, { "epoch": 0.33, "learning_rate": 1.956521739130435e-05, "loss": 1.2505, "step": 330 }, { "epoch": 0.34, "learning_rate": 2.015810276679842e-05, "loss": 1.1959, "step": 340 }, { "epoch": 0.35, "learning_rate": 2.075098814229249e-05, "loss": 1.2754, "step": 350 }, { "epoch": 0.36, "learning_rate": 2.1343873517786562e-05, "loss": 1.2427, "step": 360 }, { "epoch": 0.37, "learning_rate": 2.1936758893280633e-05, "loss": 1.1294, "step": 370 }, { "epoch": 0.38, "learning_rate": 2.2529644268774703e-05, "loss": 1.0736, "step": 380 }, { "epoch": 0.39, "learning_rate": 2.3122529644268774e-05, "loss": 1.0212, "step": 390 }, { "epoch": 0.4, "learning_rate": 2.3715415019762845e-05, "loss": 0.8713, "step": 400 }, { "epoch": 0.41, "learning_rate": 2.4308300395256916e-05, "loss": 0.9362, "step": 410 }, { "epoch": 0.42, "learning_rate": 2.4901185770750987e-05, "loss": 0.913, "step": 420 }, { "epoch": 0.43, "learning_rate": 2.549407114624506e-05, "loss": 0.9135, "step": 430 }, { "epoch": 0.43, "learning_rate": 2.608695652173913e-05, "loss": 0.7702, "step": 440 }, { "epoch": 0.44, "learning_rate": 2.6679841897233202e-05, "loss": 0.9104, "step": 450 }, { "epoch": 0.45, "learning_rate": 2.7272727272727273e-05, "loss": 0.8127, "step": 460 }, { "epoch": 0.46, "learning_rate": 2.7865612648221344e-05, "loss": 0.7761, "step": 470 }, { "epoch": 0.47, "learning_rate": 2.8458498023715415e-05, "loss": 0.7382, "step": 480 }, { "epoch": 0.48, "learning_rate": 2.9051383399209485e-05, "loss": 0.7763, "step": 490 }, { "epoch": 0.49, "learning_rate": 2.9644268774703556e-05, "loss": 0.6891, "step": 500 }, { "epoch": 0.5, "learning_rate": 2.9973620575950757e-05, "loss": 0.6791, "step": 510 }, { "epoch": 0.51, "learning_rate": 2.9907672015827656e-05, "loss": 0.7043, "step": 520 }, { "epoch": 0.52, "learning_rate": 2.984172345570455e-05, "loss": 0.5995, "step": 530 }, { "epoch": 0.53, "learning_rate": 2.9775774895581446e-05, "loss": 0.6426, "step": 540 }, { "epoch": 0.54, "learning_rate": 2.970982633545834e-05, "loss": 0.6375, "step": 550 }, { "epoch": 0.55, "learning_rate": 2.964387777533524e-05, "loss": 0.5358, "step": 560 }, { "epoch": 0.56, "learning_rate": 2.9577929215212135e-05, "loss": 0.5507, "step": 570 }, { "epoch": 0.57, "learning_rate": 2.951198065508903e-05, "loss": 0.5308, "step": 580 }, { "epoch": 0.58, "learning_rate": 2.9446032094965925e-05, "loss": 0.6004, "step": 590 }, { "epoch": 0.59, "learning_rate": 2.938008353484282e-05, "loss": 0.5332, "step": 600 }, { "epoch": 0.6, "learning_rate": 2.931413497471972e-05, "loss": 0.4689, "step": 610 }, { "epoch": 0.61, "learning_rate": 2.925478127060893e-05, "loss": 0.4989, "step": 620 }, { "epoch": 0.62, "learning_rate": 2.9188832710485824e-05, "loss": 0.4521, "step": 630 }, { "epoch": 0.63, "learning_rate": 2.912288415036272e-05, "loss": 0.4725, "step": 640 }, { "epoch": 0.64, "learning_rate": 2.9056935590239614e-05, "loss": 0.4091, "step": 650 }, { "epoch": 0.65, "learning_rate": 2.8990987030116512e-05, "loss": 0.4156, "step": 660 }, { "epoch": 0.66, "learning_rate": 2.8925038469993408e-05, "loss": 0.4098, "step": 670 }, { "epoch": 0.67, "learning_rate": 2.8859089909870303e-05, "loss": 0.3777, "step": 680 }, { "epoch": 0.68, "learning_rate": 2.8793141349747198e-05, "loss": 0.3831, "step": 690 }, { "epoch": 0.69, "learning_rate": 2.8727192789624093e-05, "loss": 0.3946, "step": 700 }, { "epoch": 0.7, "learning_rate": 2.866124422950099e-05, "loss": 0.3566, "step": 710 }, { "epoch": 0.71, "learning_rate": 2.8595295669377887e-05, "loss": 0.353, "step": 720 }, { "epoch": 0.72, "learning_rate": 2.8529347109254782e-05, "loss": 0.3432, "step": 730 }, { "epoch": 0.73, "learning_rate": 2.8463398549131677e-05, "loss": 0.3519, "step": 740 }, { "epoch": 0.74, "learning_rate": 2.8397449989008572e-05, "loss": 0.3569, "step": 750 }, { "epoch": 0.75, "learning_rate": 2.833150142888547e-05, "loss": 0.2985, "step": 760 }, { "epoch": 0.76, "learning_rate": 2.8265552868762366e-05, "loss": 0.3424, "step": 770 }, { "epoch": 0.77, "learning_rate": 2.819960430863926e-05, "loss": 0.3473, "step": 780 }, { "epoch": 0.78, "learning_rate": 2.8133655748516156e-05, "loss": 0.3691, "step": 790 }, { "epoch": 0.79, "learning_rate": 2.806770718839305e-05, "loss": 0.3621, "step": 800 }, { "epoch": 0.8, "learning_rate": 2.800175862826995e-05, "loss": 0.3124, "step": 810 }, { "epoch": 0.81, "learning_rate": 2.7935810068146845e-05, "loss": 0.3259, "step": 820 }, { "epoch": 0.82, "learning_rate": 2.786986150802374e-05, "loss": 0.3355, "step": 830 }, { "epoch": 0.83, "learning_rate": 2.780391294790064e-05, "loss": 0.2992, "step": 840 }, { "epoch": 0.84, "learning_rate": 2.7737964387777537e-05, "loss": 0.2576, "step": 850 }, { "epoch": 0.85, "learning_rate": 2.7672015827654432e-05, "loss": 0.3144, "step": 860 }, { "epoch": 0.86, "learning_rate": 2.7606067267531327e-05, "loss": 0.3806, "step": 870 }, { "epoch": 0.87, "learning_rate": 2.7540118707408223e-05, "loss": 0.3299, "step": 880 }, { "epoch": 0.88, "learning_rate": 2.7474170147285118e-05, "loss": 0.3396, "step": 890 }, { "epoch": 0.89, "learning_rate": 2.7408221587162016e-05, "loss": 0.2678, "step": 900 }, { "epoch": 0.9, "learning_rate": 2.734227302703891e-05, "loss": 0.2421, "step": 910 }, { "epoch": 0.91, "learning_rate": 2.7276324466915807e-05, "loss": 0.2265, "step": 920 }, { "epoch": 0.92, "learning_rate": 2.7210375906792702e-05, "loss": 0.3316, "step": 930 }, { "epoch": 0.93, "learning_rate": 2.7144427346669597e-05, "loss": 0.285, "step": 940 }, { "epoch": 0.94, "learning_rate": 2.7078478786546495e-05, "loss": 0.2601, "step": 950 }, { "epoch": 0.95, "learning_rate": 2.701253022642339e-05, "loss": 0.3122, "step": 960 }, { "epoch": 0.96, "learning_rate": 2.6946581666300286e-05, "loss": 0.2348, "step": 970 }, { "epoch": 0.97, "learning_rate": 2.688063310617718e-05, "loss": 0.2589, "step": 980 }, { "epoch": 0.98, "learning_rate": 2.681468454605408e-05, "loss": 0.262, "step": 990 }, { "epoch": 0.99, "learning_rate": 2.6748735985930975e-05, "loss": 0.3178, "step": 1000 }, { "epoch": 1.0, "learning_rate": 2.668278742580787e-05, "loss": 0.2861, "step": 1010 }, { "epoch": 1.01, "learning_rate": 2.6616838865684765e-05, "loss": 0.2352, "step": 1020 }, { "epoch": 1.02, "learning_rate": 2.655089030556166e-05, "loss": 0.2572, "step": 1030 }, { "epoch": 1.03, "learning_rate": 2.648494174543856e-05, "loss": 0.19, "step": 1040 }, { "epoch": 1.04, "learning_rate": 2.6418993185315457e-05, "loss": 0.2169, "step": 1050 }, { "epoch": 1.05, "learning_rate": 2.6353044625192352e-05, "loss": 0.2386, "step": 1060 }, { "epoch": 1.06, "learning_rate": 2.6287096065069247e-05, "loss": 0.2598, "step": 1070 }, { "epoch": 1.07, "learning_rate": 2.6221147504946143e-05, "loss": 0.2, "step": 1080 }, { "epoch": 1.08, "learning_rate": 2.615519894482304e-05, "loss": 0.2713, "step": 1090 }, { "epoch": 1.09, "learning_rate": 2.6089250384699936e-05, "loss": 0.2532, "step": 1100 }, { "epoch": 1.1, "learning_rate": 2.602330182457683e-05, "loss": 0.2501, "step": 1110 }, { "epoch": 1.11, "learning_rate": 2.5957353264453726e-05, "loss": 0.2721, "step": 1120 }, { "epoch": 1.12, "learning_rate": 2.589140470433062e-05, "loss": 0.2282, "step": 1130 }, { "epoch": 1.13, "learning_rate": 2.582545614420752e-05, "loss": 0.2824, "step": 1140 }, { "epoch": 1.14, "learning_rate": 2.5759507584084415e-05, "loss": 0.231, "step": 1150 }, { "epoch": 1.15, "learning_rate": 2.569355902396131e-05, "loss": 0.2715, "step": 1160 }, { "epoch": 1.16, "learning_rate": 2.5627610463838206e-05, "loss": 0.2291, "step": 1170 }, { "epoch": 1.17, "learning_rate": 2.5561661903715104e-05, "loss": 0.222, "step": 1180 }, { "epoch": 1.18, "learning_rate": 2.5495713343592e-05, "loss": 0.215, "step": 1190 }, { "epoch": 1.19, "learning_rate": 2.5429764783468894e-05, "loss": 0.2577, "step": 1200 }, { "epoch": 1.2, "learning_rate": 2.536381622334579e-05, "loss": 0.182, "step": 1210 }, { "epoch": 1.21, "learning_rate": 2.5297867663222685e-05, "loss": 0.1838, "step": 1220 }, { "epoch": 1.22, "learning_rate": 2.5231919103099583e-05, "loss": 0.2429, "step": 1230 }, { "epoch": 1.23, "learning_rate": 2.516597054297648e-05, "loss": 0.205, "step": 1240 }, { "epoch": 1.24, "learning_rate": 2.5100021982853374e-05, "loss": 0.198, "step": 1250 }, { "epoch": 1.25, "learning_rate": 2.503407342273027e-05, "loss": 0.194, "step": 1260 }, { "epoch": 1.26, "learning_rate": 2.4968124862607167e-05, "loss": 0.198, "step": 1270 }, { "epoch": 1.27, "learning_rate": 2.4902176302484066e-05, "loss": 0.1694, "step": 1280 }, { "epoch": 1.28, "learning_rate": 2.483622774236096e-05, "loss": 0.2266, "step": 1290 }, { "epoch": 1.29, "learning_rate": 2.4770279182237856e-05, "loss": 0.218, "step": 1300 }, { "epoch": 1.3, "learning_rate": 2.470433062211475e-05, "loss": 0.2122, "step": 1310 }, { "epoch": 1.3, "learning_rate": 2.4638382061991646e-05, "loss": 0.2208, "step": 1320 }, { "epoch": 1.31, "learning_rate": 2.4572433501868545e-05, "loss": 0.2528, "step": 1330 }, { "epoch": 1.32, "learning_rate": 2.450648494174544e-05, "loss": 0.2059, "step": 1340 }, { "epoch": 1.33, "learning_rate": 2.4440536381622335e-05, "loss": 0.1788, "step": 1350 }, { "epoch": 1.34, "learning_rate": 2.437458782149923e-05, "loss": 0.2225, "step": 1360 }, { "epoch": 1.35, "learning_rate": 2.430863926137613e-05, "loss": 0.2645, "step": 1370 }, { "epoch": 1.36, "learning_rate": 2.4242690701253024e-05, "loss": 0.1995, "step": 1380 }, { "epoch": 1.37, "learning_rate": 2.417674214112992e-05, "loss": 0.2409, "step": 1390 }, { "epoch": 1.38, "learning_rate": 2.4110793581006814e-05, "loss": 0.2101, "step": 1400 }, { "epoch": 1.39, "learning_rate": 2.404484502088371e-05, "loss": 0.2105, "step": 1410 }, { "epoch": 1.4, "learning_rate": 2.3978896460760608e-05, "loss": 0.1934, "step": 1420 }, { "epoch": 1.41, "learning_rate": 2.3912947900637503e-05, "loss": 0.1865, "step": 1430 }, { "epoch": 1.42, "learning_rate": 2.3846999340514398e-05, "loss": 0.23, "step": 1440 }, { "epoch": 1.43, "learning_rate": 2.3781050780391293e-05, "loss": 0.2255, "step": 1450 }, { "epoch": 1.44, "learning_rate": 2.371510222026819e-05, "loss": 0.2088, "step": 1460 }, { "epoch": 1.45, "learning_rate": 2.3649153660145087e-05, "loss": 0.1997, "step": 1470 }, { "epoch": 1.46, "learning_rate": 2.3583205100021982e-05, "loss": 0.1537, "step": 1480 }, { "epoch": 1.47, "learning_rate": 2.351725653989888e-05, "loss": 0.1836, "step": 1490 }, { "epoch": 1.48, "learning_rate": 2.3451307979775776e-05, "loss": 0.1849, "step": 1500 }, { "epoch": 1.49, "learning_rate": 2.338535941965267e-05, "loss": 0.228, "step": 1510 }, { "epoch": 1.5, "learning_rate": 2.331941085952957e-05, "loss": 0.1855, "step": 1520 }, { "epoch": 1.51, "learning_rate": 2.3253462299406465e-05, "loss": 0.2543, "step": 1530 }, { "epoch": 1.52, "learning_rate": 2.318751373928336e-05, "loss": 0.1879, "step": 1540 }, { "epoch": 1.53, "learning_rate": 2.3121565179160255e-05, "loss": 0.1946, "step": 1550 }, { "epoch": 1.54, "learning_rate": 2.3055616619037154e-05, "loss": 0.2283, "step": 1560 }, { "epoch": 1.55, "learning_rate": 2.298966805891405e-05, "loss": 0.1658, "step": 1570 }, { "epoch": 1.56, "learning_rate": 2.2923719498790944e-05, "loss": 0.2048, "step": 1580 }, { "epoch": 1.57, "learning_rate": 2.285777093866784e-05, "loss": 0.188, "step": 1590 }, { "epoch": 1.58, "learning_rate": 2.2791822378544734e-05, "loss": 0.2034, "step": 1600 }, { "epoch": 1.59, "learning_rate": 2.2725873818421633e-05, "loss": 0.1534, "step": 1610 }, { "epoch": 1.6, "learning_rate": 2.2659925258298528e-05, "loss": 0.1695, "step": 1620 }, { "epoch": 1.61, "learning_rate": 2.2593976698175423e-05, "loss": 0.1791, "step": 1630 }, { "epoch": 1.62, "learning_rate": 2.2528028138052318e-05, "loss": 0.1581, "step": 1640 }, { "epoch": 1.63, "learning_rate": 2.2462079577929213e-05, "loss": 0.1907, "step": 1650 }, { "epoch": 1.64, "learning_rate": 2.2396131017806112e-05, "loss": 0.2046, "step": 1660 }, { "epoch": 1.65, "learning_rate": 2.2330182457683007e-05, "loss": 0.1987, "step": 1670 }, { "epoch": 1.66, "learning_rate": 2.2264233897559902e-05, "loss": 0.1496, "step": 1680 }, { "epoch": 1.67, "learning_rate": 2.2198285337436797e-05, "loss": 0.183, "step": 1690 }, { "epoch": 1.68, "learning_rate": 2.2132336777313696e-05, "loss": 0.174, "step": 1700 }, { "epoch": 1.69, "learning_rate": 2.2066388217190594e-05, "loss": 0.2172, "step": 1710 }, { "epoch": 1.7, "learning_rate": 2.200043965706749e-05, "loss": 0.207, "step": 1720 }, { "epoch": 1.71, "learning_rate": 2.1934491096944385e-05, "loss": 0.1835, "step": 1730 }, { "epoch": 1.72, "learning_rate": 2.186854253682128e-05, "loss": 0.2178, "step": 1740 }, { "epoch": 1.73, "learning_rate": 2.180259397669818e-05, "loss": 0.1926, "step": 1750 }, { "epoch": 1.74, "learning_rate": 2.1736645416575074e-05, "loss": 0.2069, "step": 1760 }, { "epoch": 1.75, "learning_rate": 2.167069685645197e-05, "loss": 0.1593, "step": 1770 }, { "epoch": 1.76, "learning_rate": 2.1604748296328864e-05, "loss": 0.1436, "step": 1780 }, { "epoch": 1.77, "learning_rate": 2.153879973620576e-05, "loss": 0.2202, "step": 1790 }, { "epoch": 1.78, "learning_rate": 2.1472851176082657e-05, "loss": 0.1904, "step": 1800 }, { "epoch": 1.79, "learning_rate": 2.1406902615959553e-05, "loss": 0.1755, "step": 1810 }, { "epoch": 1.8, "learning_rate": 2.1340954055836448e-05, "loss": 0.201, "step": 1820 }, { "epoch": 1.81, "learning_rate": 2.1275005495713343e-05, "loss": 0.2124, "step": 1830 }, { "epoch": 1.82, "learning_rate": 2.1209056935590238e-05, "loss": 0.2169, "step": 1840 }, { "epoch": 1.83, "learning_rate": 2.1143108375467137e-05, "loss": 0.1813, "step": 1850 }, { "epoch": 1.84, "learning_rate": 2.1077159815344032e-05, "loss": 0.1809, "step": 1860 }, { "epoch": 1.85, "learning_rate": 2.1011211255220927e-05, "loss": 0.2056, "step": 1870 }, { "epoch": 1.86, "learning_rate": 2.0945262695097822e-05, "loss": 0.1828, "step": 1880 }, { "epoch": 1.87, "learning_rate": 2.087931413497472e-05, "loss": 0.185, "step": 1890 }, { "epoch": 1.88, "learning_rate": 2.0813365574851616e-05, "loss": 0.2169, "step": 1900 }, { "epoch": 1.89, "learning_rate": 2.074741701472851e-05, "loss": 0.1956, "step": 1910 }, { "epoch": 1.9, "learning_rate": 2.068146845460541e-05, "loss": 0.1624, "step": 1920 }, { "epoch": 1.91, "learning_rate": 2.0615519894482305e-05, "loss": 0.1974, "step": 1930 }, { "epoch": 1.92, "learning_rate": 2.0549571334359203e-05, "loss": 0.1938, "step": 1940 }, { "epoch": 1.93, "learning_rate": 2.0483622774236098e-05, "loss": 0.1856, "step": 1950 }, { "epoch": 1.94, "learning_rate": 2.0417674214112993e-05, "loss": 0.1841, "step": 1960 }, { "epoch": 1.95, "learning_rate": 2.035172565398989e-05, "loss": 0.2066, "step": 1970 }, { "epoch": 1.96, "learning_rate": 2.0285777093866784e-05, "loss": 0.162, "step": 1980 }, { "epoch": 1.97, "learning_rate": 2.0219828533743682e-05, "loss": 0.2127, "step": 1990 }, { "epoch": 1.98, "learning_rate": 2.0153879973620577e-05, "loss": 0.18, "step": 2000 }, { "epoch": 1.99, "learning_rate": 2.0087931413497473e-05, "loss": 0.1728, "step": 2010 }, { "epoch": 2.0, "learning_rate": 2.0021982853374368e-05, "loss": 0.222, "step": 2020 }, { "epoch": 2.01, "learning_rate": 1.9956034293251263e-05, "loss": 0.1823, "step": 2030 }, { "epoch": 2.02, "learning_rate": 1.989008573312816e-05, "loss": 0.2037, "step": 2040 }, { "epoch": 2.03, "learning_rate": 1.9824137173005056e-05, "loss": 0.1361, "step": 2050 }, { "epoch": 2.04, "learning_rate": 1.975818861288195e-05, "loss": 0.0991, "step": 2060 }, { "epoch": 2.05, "learning_rate": 1.9692240052758847e-05, "loss": 0.1545, "step": 2070 }, { "epoch": 2.06, "learning_rate": 1.9626291492635745e-05, "loss": 0.1691, "step": 2080 }, { "epoch": 2.07, "learning_rate": 1.956034293251264e-05, "loss": 0.1252, "step": 2090 }, { "epoch": 2.08, "learning_rate": 1.9494394372389536e-05, "loss": 0.1804, "step": 2100 }, { "epoch": 2.09, "learning_rate": 1.942844581226643e-05, "loss": 0.1545, "step": 2110 }, { "epoch": 2.1, "learning_rate": 1.9362497252143326e-05, "loss": 0.1415, "step": 2120 }, { "epoch": 2.11, "learning_rate": 1.9296548692020224e-05, "loss": 0.1691, "step": 2130 }, { "epoch": 2.12, "learning_rate": 1.9230600131897123e-05, "loss": 0.1567, "step": 2140 }, { "epoch": 2.13, "learning_rate": 1.9164651571774018e-05, "loss": 0.1768, "step": 2150 }, { "epoch": 2.14, "learning_rate": 1.9098703011650913e-05, "loss": 0.1858, "step": 2160 }, { "epoch": 2.15, "learning_rate": 1.903275445152781e-05, "loss": 0.2029, "step": 2170 }, { "epoch": 2.16, "learning_rate": 1.8966805891404707e-05, "loss": 0.1705, "step": 2180 }, { "epoch": 2.17, "learning_rate": 1.8900857331281602e-05, "loss": 0.1944, "step": 2190 }, { "epoch": 2.17, "learning_rate": 1.8834908771158497e-05, "loss": 0.1758, "step": 2200 }, { "epoch": 2.18, "learning_rate": 1.8768960211035392e-05, "loss": 0.1902, "step": 2210 }, { "epoch": 2.19, "learning_rate": 1.8703011650912288e-05, "loss": 0.1707, "step": 2220 }, { "epoch": 2.2, "learning_rate": 1.8637063090789186e-05, "loss": 0.1753, "step": 2230 }, { "epoch": 2.21, "learning_rate": 1.857111453066608e-05, "loss": 0.1855, "step": 2240 }, { "epoch": 2.22, "learning_rate": 1.8505165970542976e-05, "loss": 0.1406, "step": 2250 }, { "epoch": 2.23, "learning_rate": 1.843921741041987e-05, "loss": 0.1541, "step": 2260 }, { "epoch": 2.24, "learning_rate": 1.837326885029677e-05, "loss": 0.1276, "step": 2270 }, { "epoch": 2.25, "learning_rate": 1.8307320290173665e-05, "loss": 0.1751, "step": 2280 }, { "epoch": 2.26, "learning_rate": 1.824137173005056e-05, "loss": 0.2045, "step": 2290 }, { "epoch": 2.27, "learning_rate": 1.8175423169927456e-05, "loss": 0.1515, "step": 2300 }, { "epoch": 2.28, "learning_rate": 1.810947460980435e-05, "loss": 0.1639, "step": 2310 }, { "epoch": 2.29, "learning_rate": 1.804352604968125e-05, "loss": 0.1511, "step": 2320 }, { "epoch": 2.3, "learning_rate": 1.7977577489558144e-05, "loss": 0.1806, "step": 2330 }, { "epoch": 2.31, "learning_rate": 1.791162892943504e-05, "loss": 0.1593, "step": 2340 }, { "epoch": 2.32, "learning_rate": 1.7845680369311938e-05, "loss": 0.1901, "step": 2350 }, { "epoch": 2.33, "learning_rate": 1.7779731809188833e-05, "loss": 0.1391, "step": 2360 }, { "epoch": 2.34, "learning_rate": 1.7713783249065732e-05, "loss": 0.1668, "step": 2370 }, { "epoch": 2.35, "learning_rate": 1.7647834688942627e-05, "loss": 0.1809, "step": 2380 }, { "epoch": 2.36, "learning_rate": 1.7581886128819522e-05, "loss": 0.1652, "step": 2390 }, { "epoch": 2.37, "learning_rate": 1.7515937568696417e-05, "loss": 0.1266, "step": 2400 }, { "epoch": 2.38, "learning_rate": 1.7449989008573316e-05, "loss": 0.203, "step": 2410 }, { "epoch": 2.39, "learning_rate": 1.738404044845021e-05, "loss": 0.1855, "step": 2420 }, { "epoch": 2.4, "learning_rate": 1.7318091888327106e-05, "loss": 0.1426, "step": 2430 }, { "epoch": 2.41, "learning_rate": 1.7252143328204e-05, "loss": 0.1321, "step": 2440 }, { "epoch": 2.42, "learning_rate": 1.7186194768080896e-05, "loss": 0.161, "step": 2450 }, { "epoch": 2.43, "learning_rate": 1.7120246207957795e-05, "loss": 0.1757, "step": 2460 }, { "epoch": 2.44, "learning_rate": 1.705429764783469e-05, "loss": 0.1557, "step": 2470 }, { "epoch": 2.45, "learning_rate": 1.6988349087711585e-05, "loss": 0.1551, "step": 2480 }, { "epoch": 2.46, "learning_rate": 1.692240052758848e-05, "loss": 0.1265, "step": 2490 }, { "epoch": 2.47, "learning_rate": 1.6856451967465375e-05, "loss": 0.189, "step": 2500 }, { "epoch": 2.48, "learning_rate": 1.6790503407342274e-05, "loss": 0.228, "step": 2510 }, { "epoch": 2.49, "learning_rate": 1.672455484721917e-05, "loss": 0.155, "step": 2520 }, { "epoch": 2.5, "learning_rate": 1.6658606287096064e-05, "loss": 0.1437, "step": 2530 }, { "epoch": 2.51, "learning_rate": 1.659265772697296e-05, "loss": 0.1344, "step": 2540 }, { "epoch": 2.52, "learning_rate": 1.6526709166849855e-05, "loss": 0.2003, "step": 2550 }, { "epoch": 2.53, "learning_rate": 1.6460760606726753e-05, "loss": 0.1136, "step": 2560 }, { "epoch": 2.54, "learning_rate": 1.639481204660365e-05, "loss": 0.1827, "step": 2570 }, { "epoch": 2.55, "learning_rate": 1.6328863486480547e-05, "loss": 0.1626, "step": 2580 }, { "epoch": 2.56, "learning_rate": 1.6262914926357442e-05, "loss": 0.173, "step": 2590 }, { "epoch": 2.57, "learning_rate": 1.619696636623434e-05, "loss": 0.1338, "step": 2600 }, { "epoch": 2.58, "learning_rate": 1.6131017806111236e-05, "loss": 0.1171, "step": 2610 }, { "epoch": 2.59, "learning_rate": 1.606506924598813e-05, "loss": 0.1918, "step": 2620 }, { "epoch": 2.6, "learning_rate": 1.5999120685865026e-05, "loss": 0.1404, "step": 2630 }, { "epoch": 2.61, "learning_rate": 1.593317212574192e-05, "loss": 0.1228, "step": 2640 }, { "epoch": 2.62, "learning_rate": 1.586722356561882e-05, "loss": 0.1597, "step": 2650 }, { "epoch": 2.63, "learning_rate": 1.5801275005495715e-05, "loss": 0.1406, "step": 2660 }, { "epoch": 2.64, "learning_rate": 1.573532644537261e-05, "loss": 0.2154, "step": 2670 }, { "epoch": 2.65, "learning_rate": 1.5669377885249505e-05, "loss": 0.158, "step": 2680 }, { "epoch": 2.66, "learning_rate": 1.56034293251264e-05, "loss": 0.138, "step": 2690 }, { "epoch": 2.67, "learning_rate": 1.55374807650033e-05, "loss": 0.1273, "step": 2700 }, { "epoch": 2.68, "learning_rate": 1.5471532204880194e-05, "loss": 0.1851, "step": 2710 }, { "epoch": 2.69, "learning_rate": 1.540558364475709e-05, "loss": 0.1218, "step": 2720 }, { "epoch": 2.7, "learning_rate": 1.5339635084633984e-05, "loss": 0.1951, "step": 2730 }, { "epoch": 2.71, "learning_rate": 1.527368652451088e-05, "loss": 0.1701, "step": 2740 }, { "epoch": 2.72, "learning_rate": 1.520773796438778e-05, "loss": 0.1441, "step": 2750 }, { "epoch": 2.73, "learning_rate": 1.5141789404264675e-05, "loss": 0.1751, "step": 2760 }, { "epoch": 2.74, "learning_rate": 1.507584084414157e-05, "loss": 0.1279, "step": 2770 }, { "epoch": 2.75, "learning_rate": 1.5009892284018465e-05, "loss": 0.1323, "step": 2780 }, { "epoch": 2.76, "learning_rate": 1.4943943723895362e-05, "loss": 0.1589, "step": 2790 }, { "epoch": 2.77, "learning_rate": 1.4877995163772257e-05, "loss": 0.15, "step": 2800 }, { "epoch": 2.78, "learning_rate": 1.4812046603649154e-05, "loss": 0.1367, "step": 2810 }, { "epoch": 2.79, "learning_rate": 1.4746098043526049e-05, "loss": 0.1074, "step": 2820 }, { "epoch": 2.8, "learning_rate": 1.4680149483402946e-05, "loss": 0.1634, "step": 2830 }, { "epoch": 2.81, "learning_rate": 1.4614200923279843e-05, "loss": 0.203, "step": 2840 }, { "epoch": 2.82, "learning_rate": 1.454825236315674e-05, "loss": 0.0856, "step": 2850 }, { "epoch": 2.83, "learning_rate": 1.4482303803033635e-05, "loss": 0.153, "step": 2860 }, { "epoch": 2.84, "learning_rate": 1.441635524291053e-05, "loss": 0.1528, "step": 2870 }, { "epoch": 2.85, "learning_rate": 1.4350406682787427e-05, "loss": 0.1451, "step": 2880 }, { "epoch": 2.86, "learning_rate": 1.4284458122664322e-05, "loss": 0.1376, "step": 2890 }, { "epoch": 2.87, "learning_rate": 1.4218509562541219e-05, "loss": 0.1334, "step": 2900 }, { "epoch": 2.88, "learning_rate": 1.4152561002418114e-05, "loss": 0.1517, "step": 2910 }, { "epoch": 2.89, "learning_rate": 1.408661244229501e-05, "loss": 0.1556, "step": 2920 }, { "epoch": 2.9, "learning_rate": 1.4020663882171906e-05, "loss": 0.1306, "step": 2930 }, { "epoch": 2.91, "learning_rate": 1.39547153220488e-05, "loss": 0.1398, "step": 2940 }, { "epoch": 2.92, "learning_rate": 1.38887667619257e-05, "loss": 0.1071, "step": 2950 }, { "epoch": 2.93, "learning_rate": 1.3822818201802595e-05, "loss": 0.1802, "step": 2960 }, { "epoch": 2.94, "learning_rate": 1.3756869641679491e-05, "loss": 0.1312, "step": 2970 }, { "epoch": 2.95, "learning_rate": 1.3690921081556387e-05, "loss": 0.1135, "step": 2980 }, { "epoch": 2.96, "learning_rate": 1.3624972521433283e-05, "loss": 0.1499, "step": 2990 }, { "epoch": 2.97, "learning_rate": 1.3559023961310178e-05, "loss": 0.1589, "step": 3000 }, { "epoch": 2.98, "learning_rate": 1.3493075401187074e-05, "loss": 0.2078, "step": 3010 }, { "epoch": 2.99, "learning_rate": 1.342712684106397e-05, "loss": 0.149, "step": 3020 }, { "epoch": 3.0, "learning_rate": 1.3361178280940866e-05, "loss": 0.1353, "step": 3030 }, { "epoch": 3.01, "learning_rate": 1.3295229720817762e-05, "loss": 0.1641, "step": 3040 }, { "epoch": 3.02, "learning_rate": 1.3229281160694658e-05, "loss": 0.1325, "step": 3050 }, { "epoch": 3.03, "learning_rate": 1.3163332600571554e-05, "loss": 0.1777, "step": 3060 }, { "epoch": 3.04, "learning_rate": 1.3097384040448451e-05, "loss": 0.1321, "step": 3070 }, { "epoch": 3.04, "learning_rate": 1.3031435480325346e-05, "loss": 0.1089, "step": 3080 }, { "epoch": 3.05, "learning_rate": 1.2965486920202243e-05, "loss": 0.1136, "step": 3090 }, { "epoch": 3.06, "learning_rate": 1.2899538360079138e-05, "loss": 0.1197, "step": 3100 }, { "epoch": 3.07, "learning_rate": 1.2833589799956035e-05, "loss": 0.1227, "step": 3110 }, { "epoch": 3.08, "learning_rate": 1.276764123983293e-05, "loss": 0.1675, "step": 3120 }, { "epoch": 3.09, "learning_rate": 1.2701692679709826e-05, "loss": 0.1355, "step": 3130 }, { "epoch": 3.1, "learning_rate": 1.2635744119586722e-05, "loss": 0.1356, "step": 3140 }, { "epoch": 3.11, "learning_rate": 1.2569795559463618e-05, "loss": 0.153, "step": 3150 }, { "epoch": 3.12, "learning_rate": 1.2503846999340514e-05, "loss": 0.1471, "step": 3160 }, { "epoch": 3.13, "learning_rate": 1.2437898439217411e-05, "loss": 0.1055, "step": 3170 }, { "epoch": 3.14, "learning_rate": 1.2371949879094308e-05, "loss": 0.1172, "step": 3180 }, { "epoch": 3.15, "learning_rate": 1.2306001318971203e-05, "loss": 0.1238, "step": 3190 }, { "epoch": 3.16, "learning_rate": 1.2240052758848098e-05, "loss": 0.1822, "step": 3200 }, { "epoch": 3.17, "learning_rate": 1.2174104198724995e-05, "loss": 0.1395, "step": 3210 }, { "epoch": 3.18, "learning_rate": 1.210815563860189e-05, "loss": 0.1473, "step": 3220 }, { "epoch": 3.19, "learning_rate": 1.2042207078478787e-05, "loss": 0.0996, "step": 3230 }, { "epoch": 3.2, "learning_rate": 1.1976258518355682e-05, "loss": 0.1145, "step": 3240 }, { "epoch": 3.21, "learning_rate": 1.1910309958232578e-05, "loss": 0.141, "step": 3250 }, { "epoch": 3.22, "learning_rate": 1.1844361398109474e-05, "loss": 0.1335, "step": 3260 }, { "epoch": 3.23, "learning_rate": 1.1778412837986371e-05, "loss": 0.1007, "step": 3270 }, { "epoch": 3.24, "learning_rate": 1.1712464277863268e-05, "loss": 0.1436, "step": 3280 }, { "epoch": 3.25, "learning_rate": 1.1646515717740163e-05, "loss": 0.1327, "step": 3290 }, { "epoch": 3.26, "learning_rate": 1.158056715761706e-05, "loss": 0.1116, "step": 3300 }, { "epoch": 3.27, "learning_rate": 1.1514618597493955e-05, "loss": 0.1609, "step": 3310 }, { "epoch": 3.28, "learning_rate": 1.144867003737085e-05, "loss": 0.1597, "step": 3320 }, { "epoch": 3.29, "learning_rate": 1.1382721477247747e-05, "loss": 0.1425, "step": 3330 }, { "epoch": 3.3, "learning_rate": 1.1316772917124642e-05, "loss": 0.1429, "step": 3340 }, { "epoch": 3.31, "learning_rate": 1.1250824357001539e-05, "loss": 0.1177, "step": 3350 }, { "epoch": 3.32, "learning_rate": 1.1184875796878434e-05, "loss": 0.1242, "step": 3360 }, { "epoch": 3.33, "learning_rate": 1.1118927236755331e-05, "loss": 0.0877, "step": 3370 }, { "epoch": 3.34, "learning_rate": 1.1052978676632228e-05, "loss": 0.1758, "step": 3380 }, { "epoch": 3.35, "learning_rate": 1.0987030116509123e-05, "loss": 0.1445, "step": 3390 }, { "epoch": 3.36, "learning_rate": 1.092108155638602e-05, "loss": 0.1311, "step": 3400 }, { "epoch": 3.37, "learning_rate": 1.0855132996262915e-05, "loss": 0.1278, "step": 3410 }, { "epoch": 3.38, "learning_rate": 1.0789184436139812e-05, "loss": 0.1519, "step": 3420 }, { "epoch": 3.39, "learning_rate": 1.0723235876016707e-05, "loss": 0.106, "step": 3430 }, { "epoch": 3.4, "learning_rate": 1.0657287315893604e-05, "loss": 0.1592, "step": 3440 }, { "epoch": 3.41, "learning_rate": 1.0591338755770499e-05, "loss": 0.1479, "step": 3450 }, { "epoch": 3.42, "learning_rate": 1.0525390195647394e-05, "loss": 0.1133, "step": 3460 }, { "epoch": 3.43, "learning_rate": 1.0459441635524291e-05, "loss": 0.1484, "step": 3470 }, { "epoch": 3.44, "learning_rate": 1.0393493075401186e-05, "loss": 0.1155, "step": 3480 }, { "epoch": 3.45, "learning_rate": 1.0327544515278085e-05, "loss": 0.1557, "step": 3490 }, { "epoch": 3.46, "learning_rate": 1.026159595515498e-05, "loss": 0.1186, "step": 3500 }, { "epoch": 3.47, "learning_rate": 1.0195647395031875e-05, "loss": 0.093, "step": 3510 }, { "epoch": 3.48, "learning_rate": 1.0129698834908772e-05, "loss": 0.1454, "step": 3520 }, { "epoch": 3.49, "learning_rate": 1.0063750274785667e-05, "loss": 0.1354, "step": 3530 }, { "epoch": 3.5, "learning_rate": 9.997801714662564e-06, "loss": 0.1405, "step": 3540 }, { "epoch": 3.51, "learning_rate": 9.931853154539459e-06, "loss": 0.0923, "step": 3550 }, { "epoch": 3.52, "learning_rate": 9.865904594416356e-06, "loss": 0.1147, "step": 3560 }, { "epoch": 3.53, "learning_rate": 9.799956034293251e-06, "loss": 0.1586, "step": 3570 }, { "epoch": 3.54, "learning_rate": 9.734007474170146e-06, "loss": 0.1141, "step": 3580 }, { "epoch": 3.55, "learning_rate": 9.668058914047043e-06, "loss": 0.1491, "step": 3590 }, { "epoch": 3.56, "learning_rate": 9.60211035392394e-06, "loss": 0.1538, "step": 3600 }, { "epoch": 3.57, "learning_rate": 9.536161793800837e-06, "loss": 0.168, "step": 3610 }, { "epoch": 3.58, "learning_rate": 9.470213233677732e-06, "loss": 0.1334, "step": 3620 }, { "epoch": 3.59, "learning_rate": 9.404264673554629e-06, "loss": 0.1234, "step": 3630 }, { "epoch": 3.6, "learning_rate": 9.338316113431524e-06, "loss": 0.1063, "step": 3640 }, { "epoch": 3.61, "learning_rate": 9.272367553308419e-06, "loss": 0.193, "step": 3650 }, { "epoch": 3.62, "learning_rate": 9.206418993185316e-06, "loss": 0.1331, "step": 3660 }, { "epoch": 3.63, "learning_rate": 9.140470433062211e-06, "loss": 0.1483, "step": 3670 }, { "epoch": 3.64, "learning_rate": 9.074521872939108e-06, "loss": 0.1115, "step": 3680 }, { "epoch": 3.65, "learning_rate": 9.008573312816003e-06, "loss": 0.1318, "step": 3690 }, { "epoch": 3.66, "learning_rate": 8.9426247526929e-06, "loss": 0.1039, "step": 3700 }, { "epoch": 3.67, "learning_rate": 8.876676192569797e-06, "loss": 0.1405, "step": 3710 }, { "epoch": 3.68, "learning_rate": 8.810727632446692e-06, "loss": 0.1743, "step": 3720 }, { "epoch": 3.69, "learning_rate": 8.744779072323589e-06, "loss": 0.1126, "step": 3730 }, { "epoch": 3.7, "learning_rate": 8.678830512200484e-06, "loss": 0.1218, "step": 3740 }, { "epoch": 3.71, "learning_rate": 8.61288195207738e-06, "loss": 0.0908, "step": 3750 }, { "epoch": 3.72, "learning_rate": 8.546933391954276e-06, "loss": 0.123, "step": 3760 }, { "epoch": 3.73, "learning_rate": 8.480984831831171e-06, "loss": 0.133, "step": 3770 }, { "epoch": 3.74, "learning_rate": 8.415036271708068e-06, "loss": 0.1224, "step": 3780 }, { "epoch": 3.75, "learning_rate": 8.349087711584963e-06, "loss": 0.127, "step": 3790 }, { "epoch": 3.76, "learning_rate": 8.28313915146186e-06, "loss": 0.1376, "step": 3800 }, { "epoch": 3.77, "learning_rate": 8.217190591338755e-06, "loss": 0.1333, "step": 3810 }, { "epoch": 3.78, "learning_rate": 8.151242031215653e-06, "loss": 0.1426, "step": 3820 }, { "epoch": 3.79, "learning_rate": 8.085293471092549e-06, "loss": 0.115, "step": 3830 }, { "epoch": 3.8, "learning_rate": 8.019344910969444e-06, "loss": 0.1527, "step": 3840 }, { "epoch": 3.81, "learning_rate": 7.95339635084634e-06, "loss": 0.1254, "step": 3850 }, { "epoch": 3.82, "learning_rate": 7.887447790723236e-06, "loss": 0.0964, "step": 3860 }, { "epoch": 3.83, "learning_rate": 7.821499230600133e-06, "loss": 0.1546, "step": 3870 }, { "epoch": 3.84, "learning_rate": 7.755550670477028e-06, "loss": 0.1098, "step": 3880 }, { "epoch": 3.85, "learning_rate": 7.689602110353925e-06, "loss": 0.1356, "step": 3890 }, { "epoch": 3.86, "learning_rate": 7.6236535502308205e-06, "loss": 0.1723, "step": 3900 }, { "epoch": 3.87, "learning_rate": 7.557704990107716e-06, "loss": 0.1018, "step": 3910 }, { "epoch": 3.88, "learning_rate": 7.4917564299846125e-06, "loss": 0.1349, "step": 3920 }, { "epoch": 3.89, "learning_rate": 7.425807869861508e-06, "loss": 0.1295, "step": 3930 }, { "epoch": 3.9, "learning_rate": 7.359859309738404e-06, "loss": 0.1214, "step": 3940 }, { "epoch": 3.91, "learning_rate": 7.2939107496153005e-06, "loss": 0.1084, "step": 3950 }, { "epoch": 3.91, "learning_rate": 7.2279621894921965e-06, "loss": 0.0951, "step": 3960 }, { "epoch": 3.92, "learning_rate": 7.1620136293690925e-06, "loss": 0.1372, "step": 3970 }, { "epoch": 3.93, "learning_rate": 7.0960650692459885e-06, "loss": 0.1053, "step": 3980 }, { "epoch": 3.94, "learning_rate": 7.0301165091228845e-06, "loss": 0.1674, "step": 3990 }, { "epoch": 3.95, "learning_rate": 6.9641679489997805e-06, "loss": 0.1158, "step": 4000 }, { "epoch": 3.96, "learning_rate": 6.8982193888766765e-06, "loss": 0.0978, "step": 4010 }, { "epoch": 3.97, "learning_rate": 6.8322708287535724e-06, "loss": 0.15, "step": 4020 }, { "epoch": 3.98, "learning_rate": 6.7663222686304684e-06, "loss": 0.116, "step": 4030 }, { "epoch": 3.99, "learning_rate": 6.7003737085073644e-06, "loss": 0.1613, "step": 4040 }, { "epoch": 4.0, "learning_rate": 6.6344251483842604e-06, "loss": 0.1803, "step": 4050 }, { "epoch": 4.01, "learning_rate": 6.5684765882611564e-06, "loss": 0.1564, "step": 4060 }, { "epoch": 4.02, "learning_rate": 6.502528028138052e-06, "loss": 0.0706, "step": 4070 }, { "epoch": 4.03, "learning_rate": 6.436579468014948e-06, "loss": 0.0968, "step": 4080 }, { "epoch": 4.04, "learning_rate": 6.370630907891844e-06, "loss": 0.1162, "step": 4090 }, { "epoch": 4.05, "learning_rate": 6.30468234776874e-06, "loss": 0.1376, "step": 4100 }, { "epoch": 4.06, "learning_rate": 6.238733787645637e-06, "loss": 0.1332, "step": 4110 }, { "epoch": 4.07, "learning_rate": 6.172785227522533e-06, "loss": 0.1128, "step": 4120 }, { "epoch": 4.08, "learning_rate": 6.106836667399428e-06, "loss": 0.097, "step": 4130 }, { "epoch": 4.09, "learning_rate": 6.040888107276324e-06, "loss": 0.1306, "step": 4140 }, { "epoch": 4.1, "learning_rate": 5.97493954715322e-06, "loss": 0.0871, "step": 4150 }, { "epoch": 4.11, "learning_rate": 5.908990987030116e-06, "loss": 0.1372, "step": 4160 }, { "epoch": 4.12, "learning_rate": 5.843042426907013e-06, "loss": 0.0942, "step": 4170 }, { "epoch": 4.13, "learning_rate": 5.777093866783909e-06, "loss": 0.1473, "step": 4180 }, { "epoch": 4.14, "learning_rate": 5.711145306660804e-06, "loss": 0.1038, "step": 4190 }, { "epoch": 4.15, "learning_rate": 5.6451967465377e-06, "loss": 0.1069, "step": 4200 }, { "epoch": 4.16, "learning_rate": 5.579248186414596e-06, "loss": 0.0944, "step": 4210 }, { "epoch": 4.17, "learning_rate": 5.513299626291493e-06, "loss": 0.0887, "step": 4220 }, { "epoch": 4.18, "learning_rate": 5.447351066168389e-06, "loss": 0.1262, "step": 4230 }, { "epoch": 4.19, "learning_rate": 5.381402506045285e-06, "loss": 0.1578, "step": 4240 }, { "epoch": 4.2, "learning_rate": 5.31545394592218e-06, "loss": 0.1421, "step": 4250 }, { "epoch": 4.21, "learning_rate": 5.249505385799076e-06, "loss": 0.1146, "step": 4260 }, { "epoch": 4.22, "learning_rate": 5.183556825675972e-06, "loss": 0.1448, "step": 4270 }, { "epoch": 4.23, "learning_rate": 5.117608265552869e-06, "loss": 0.1439, "step": 4280 }, { "epoch": 4.24, "learning_rate": 5.051659705429765e-06, "loss": 0.1047, "step": 4290 }, { "epoch": 4.25, "learning_rate": 4.985711145306661e-06, "loss": 0.1303, "step": 4300 }, { "epoch": 4.26, "learning_rate": 4.919762585183557e-06, "loss": 0.0992, "step": 4310 }, { "epoch": 4.27, "learning_rate": 4.853814025060452e-06, "loss": 0.0815, "step": 4320 }, { "epoch": 4.28, "learning_rate": 4.787865464937349e-06, "loss": 0.1775, "step": 4330 }, { "epoch": 4.29, "learning_rate": 4.721916904814245e-06, "loss": 0.125, "step": 4340 }, { "epoch": 4.3, "learning_rate": 4.655968344691141e-06, "loss": 0.127, "step": 4350 }, { "epoch": 4.31, "learning_rate": 4.590019784568037e-06, "loss": 0.103, "step": 4360 }, { "epoch": 4.32, "learning_rate": 4.524071224444933e-06, "loss": 0.0925, "step": 4370 }, { "epoch": 4.33, "learning_rate": 4.458122664321829e-06, "loss": 0.1033, "step": 4380 }, { "epoch": 4.34, "learning_rate": 4.392174104198725e-06, "loss": 0.1294, "step": 4390 }, { "epoch": 4.35, "learning_rate": 4.326225544075621e-06, "loss": 0.1291, "step": 4400 }, { "epoch": 4.36, "learning_rate": 4.260276983952517e-06, "loss": 0.0933, "step": 4410 }, { "epoch": 4.37, "learning_rate": 4.194328423829413e-06, "loss": 0.1077, "step": 4420 }, { "epoch": 4.38, "learning_rate": 4.128379863706309e-06, "loss": 0.1154, "step": 4430 }, { "epoch": 4.39, "learning_rate": 4.062431303583206e-06, "loss": 0.1589, "step": 4440 }, { "epoch": 4.4, "learning_rate": 3.996482743460101e-06, "loss": 0.1659, "step": 4450 }, { "epoch": 4.41, "learning_rate": 3.930534183336997e-06, "loss": 0.1109, "step": 4460 }, { "epoch": 4.42, "learning_rate": 3.864585623213893e-06, "loss": 0.1058, "step": 4470 }, { "epoch": 4.43, "learning_rate": 3.7986370630907895e-06, "loss": 0.1291, "step": 4480 }, { "epoch": 4.44, "learning_rate": 3.7326885029676855e-06, "loss": 0.1205, "step": 4490 }, { "epoch": 4.45, "learning_rate": 3.666739942844581e-06, "loss": 0.099, "step": 4500 }, { "epoch": 4.46, "learning_rate": 3.6007913827214774e-06, "loss": 0.0999, "step": 4510 }, { "epoch": 4.47, "learning_rate": 3.5348428225983734e-06, "loss": 0.119, "step": 4520 }, { "epoch": 4.48, "learning_rate": 3.4688942624752694e-06, "loss": 0.1237, "step": 4530 }, { "epoch": 4.49, "learning_rate": 3.4029457023521654e-06, "loss": 0.1317, "step": 4540 }, { "epoch": 4.5, "learning_rate": 3.3369971422290614e-06, "loss": 0.1166, "step": 4550 }, { "epoch": 4.51, "learning_rate": 3.2710485821059574e-06, "loss": 0.1161, "step": 4560 }, { "epoch": 4.52, "learning_rate": 3.2051000219828534e-06, "loss": 0.1129, "step": 4570 }, { "epoch": 4.53, "learning_rate": 3.1391514618597494e-06, "loss": 0.1614, "step": 4580 }, { "epoch": 4.54, "learning_rate": 3.0732029017366454e-06, "loss": 0.1439, "step": 4590 }, { "epoch": 4.55, "learning_rate": 3.0072543416135414e-06, "loss": 0.103, "step": 4600 }, { "epoch": 4.56, "learning_rate": 2.941305781490438e-06, "loss": 0.1322, "step": 4610 }, { "epoch": 4.57, "learning_rate": 2.875357221367334e-06, "loss": 0.1276, "step": 4620 }, { "epoch": 4.58, "learning_rate": 2.8094086612442294e-06, "loss": 0.0867, "step": 4630 }, { "epoch": 4.59, "learning_rate": 2.743460101121126e-06, "loss": 0.1123, "step": 4640 }, { "epoch": 4.6, "learning_rate": 2.6775115409980218e-06, "loss": 0.0828, "step": 4650 }, { "epoch": 4.61, "learning_rate": 2.6181578368872278e-06, "loss": 0.1149, "step": 4660 }, { "epoch": 4.62, "learning_rate": 2.552209276764124e-06, "loss": 0.1199, "step": 4670 }, { "epoch": 4.63, "learning_rate": 2.48626071664102e-06, "loss": 0.1175, "step": 4680 }, { "epoch": 4.64, "learning_rate": 2.4203121565179157e-06, "loss": 0.1146, "step": 4690 }, { "epoch": 4.65, "learning_rate": 2.354363596394812e-06, "loss": 0.1068, "step": 4700 }, { "epoch": 4.66, "learning_rate": 2.288415036271708e-06, "loss": 0.0809, "step": 4710 }, { "epoch": 4.67, "learning_rate": 2.222466476148604e-06, "loss": 0.133, "step": 4720 }, { "epoch": 4.68, "learning_rate": 2.1565179160255e-06, "loss": 0.1401, "step": 4730 }, { "epoch": 4.69, "learning_rate": 2.090569355902396e-06, "loss": 0.0955, "step": 4740 }, { "epoch": 4.7, "learning_rate": 2.0246207957792926e-06, "loss": 0.1238, "step": 4750 }, { "epoch": 4.71, "learning_rate": 1.958672235656188e-06, "loss": 0.0778, "step": 4760 }, { "epoch": 4.72, "learning_rate": 1.8927236755330843e-06, "loss": 0.105, "step": 4770 }, { "epoch": 4.73, "learning_rate": 1.8267751154099803e-06, "loss": 0.1376, "step": 4780 }, { "epoch": 4.74, "learning_rate": 1.7608265552868763e-06, "loss": 0.1165, "step": 4790 }, { "epoch": 4.75, "learning_rate": 1.6948779951637723e-06, "loss": 0.1042, "step": 4800 }, { "epoch": 4.76, "learning_rate": 1.6289294350406683e-06, "loss": 0.0998, "step": 4810 }, { "epoch": 4.77, "learning_rate": 1.5629808749175643e-06, "loss": 0.1001, "step": 4820 }, { "epoch": 4.78, "learning_rate": 1.4970323147944603e-06, "loss": 0.139, "step": 4830 }, { "epoch": 4.78, "learning_rate": 1.4310837546713563e-06, "loss": 0.1461, "step": 4840 }, { "epoch": 4.79, "learning_rate": 1.3651351945482525e-06, "loss": 0.1185, "step": 4850 }, { "epoch": 4.8, "learning_rate": 1.2991866344251483e-06, "loss": 0.1626, "step": 4860 }, { "epoch": 4.81, "learning_rate": 1.2332380743020445e-06, "loss": 0.1287, "step": 4870 }, { "epoch": 4.82, "learning_rate": 1.1672895141789405e-06, "loss": 0.0977, "step": 4880 }, { "epoch": 4.83, "learning_rate": 1.1013409540558365e-06, "loss": 0.1002, "step": 4890 }, { "epoch": 4.84, "learning_rate": 1.0353923939327325e-06, "loss": 0.189, "step": 4900 }, { "epoch": 4.85, "learning_rate": 9.694438338096285e-07, "loss": 0.13, "step": 4910 }, { "epoch": 4.86, "learning_rate": 9.034952736865246e-07, "loss": 0.1346, "step": 4920 }, { "epoch": 4.87, "learning_rate": 8.375467135634206e-07, "loss": 0.1269, "step": 4930 }, { "epoch": 4.88, "learning_rate": 7.715981534403167e-07, "loss": 0.1316, "step": 4940 }, { "epoch": 4.89, "learning_rate": 7.056495933172125e-07, "loss": 0.1173, "step": 4950 }, { "epoch": 4.9, "learning_rate": 6.397010331941086e-07, "loss": 0.1105, "step": 4960 }, { "epoch": 4.91, "learning_rate": 5.737524730710046e-07, "loss": 0.1184, "step": 4970 }, { "epoch": 4.92, "learning_rate": 5.078039129479007e-07, "loss": 0.1432, "step": 4980 }, { "epoch": 4.93, "learning_rate": 4.418553528247967e-07, "loss": 0.0963, "step": 4990 }, { "epoch": 4.94, "learning_rate": 3.759067927016927e-07, "loss": 0.0694, "step": 5000 }, { "epoch": 4.95, "learning_rate": 3.099582325785887e-07, "loss": 0.0947, "step": 5010 }, { "epoch": 4.96, "learning_rate": 2.440096724554847e-07, "loss": 0.0612, "step": 5020 }, { "epoch": 4.97, "learning_rate": 1.7806111233238074e-07, "loss": 0.1119, "step": 5030 }, { "epoch": 4.98, "learning_rate": 1.1211255220927677e-07, "loss": 0.0999, "step": 5040 }, { "epoch": 4.99, "learning_rate": 4.6163992086172784e-08, "loss": 0.065, "step": 5050 }, { "epoch": 5.0, "step": 5055, "total_flos": 2.93687221089024e+18, "train_loss": 0.30798943754355584, "train_runtime": 3609.9688, "train_samples_per_second": 89.65, "train_steps_per_second": 1.4 } ], "logging_steps": 10, "max_steps": 5055, "num_train_epochs": 5, "save_steps": 200, "total_flos": 2.93687221089024e+18, "trial_name": null, "trial_params": null }