|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 2.998038536979347, |
|
"global_step": 1218, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 1.092896174863388e-06, |
|
"loss": 5.1703, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 2.185792349726776e-06, |
|
"loss": 4.4905, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 3.2786885245901638e-06, |
|
"loss": 3.712, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 4.371584699453552e-06, |
|
"loss": 2.9301, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 5.464480874316941e-06, |
|
"loss": 2.4493, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 6.5573770491803276e-06, |
|
"loss": 2.2009, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 7.650273224043716e-06, |
|
"loss": 2.1093, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 8.743169398907103e-06, |
|
"loss": 2.0401, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.836065573770493e-06, |
|
"loss": 1.9798, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.0928961748633882e-05, |
|
"loss": 1.9179, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.202185792349727e-05, |
|
"loss": 1.8746, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 1.3114754098360655e-05, |
|
"loss": 1.8008, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.4207650273224044e-05, |
|
"loss": 1.7745, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.5300546448087432e-05, |
|
"loss": 1.7164, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.639344262295082e-05, |
|
"loss": 1.6755, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 1.7486338797814207e-05, |
|
"loss": 1.613, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.85792349726776e-05, |
|
"loss": 1.5696, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.9672131147540985e-05, |
|
"loss": 1.5038, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.986473429951691e-05, |
|
"loss": 1.482, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.9671497584541064e-05, |
|
"loss": 1.4213, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.947826086956522e-05, |
|
"loss": 1.3555, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 1.9285024154589372e-05, |
|
"loss": 1.3546, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 1.909178743961353e-05, |
|
"loss": 1.3325, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 1.8898550724637685e-05, |
|
"loss": 1.2889, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 1.8705314009661837e-05, |
|
"loss": 1.2737, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 1.8512077294685994e-05, |
|
"loss": 1.2861, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 1.8318840579710146e-05, |
|
"loss": 1.2707, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 1.81256038647343e-05, |
|
"loss": 1.2544, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 1.7932367149758455e-05, |
|
"loss": 1.2397, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 1.773913043478261e-05, |
|
"loss": 1.2487, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 1.7545893719806767e-05, |
|
"loss": 1.2469, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 1.735265700483092e-05, |
|
"loss": 1.2192, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 1.7159420289855073e-05, |
|
"loss": 1.2422, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.696618357487923e-05, |
|
"loss": 1.23, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.677294685990338e-05, |
|
"loss": 1.2295, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 1.6579710144927538e-05, |
|
"loss": 1.224, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 1.6386473429951694e-05, |
|
"loss": 1.2363, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 1.6193236714975847e-05, |
|
"loss": 1.216, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 1.6000000000000003e-05, |
|
"loss": 1.205, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 1.5806763285024155e-05, |
|
"loss": 1.2222, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 1.5613526570048308e-05, |
|
"loss": 1.1912, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 1.5420289855072464e-05, |
|
"loss": 1.1828, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 1.5227053140096619e-05, |
|
"loss": 1.1682, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 1.5033816425120775e-05, |
|
"loss": 1.1822, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 1.484057971014493e-05, |
|
"loss": 1.166, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 1.4647342995169082e-05, |
|
"loss": 1.1727, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 1.4454106280193238e-05, |
|
"loss": 1.1796, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 1.4260869565217392e-05, |
|
"loss": 1.1568, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 1.4067632850241547e-05, |
|
"loss": 1.1816, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 1.3874396135265701e-05, |
|
"loss": 1.1614, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 1.3681159420289856e-05, |
|
"loss": 1.1636, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 1.28, |
|
"learning_rate": 1.3487922705314012e-05, |
|
"loss": 1.1602, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 1.3, |
|
"learning_rate": 1.3294685990338165e-05, |
|
"loss": 1.1624, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 1.33, |
|
"learning_rate": 1.3101449275362319e-05, |
|
"loss": 1.1523, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 1.35, |
|
"learning_rate": 1.2908212560386475e-05, |
|
"loss": 1.1672, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 1.38, |
|
"learning_rate": 1.271497584541063e-05, |
|
"loss": 1.1565, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"learning_rate": 1.2521739130434784e-05, |
|
"loss": 1.1585, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 1.43, |
|
"learning_rate": 1.2328502415458938e-05, |
|
"loss": 1.1564, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"learning_rate": 1.2135265700483093e-05, |
|
"loss": 1.1381, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 1.48, |
|
"learning_rate": 1.1942028985507247e-05, |
|
"loss": 1.17, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"learning_rate": 1.1748792270531402e-05, |
|
"loss": 1.1754, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 1.53, |
|
"learning_rate": 1.1555555555555556e-05, |
|
"loss": 1.1292, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 1.55, |
|
"learning_rate": 1.1362318840579712e-05, |
|
"loss": 1.1239, |
|
"step": 630 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"learning_rate": 1.1169082125603865e-05, |
|
"loss": 1.1878, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 1.6, |
|
"learning_rate": 1.0975845410628021e-05, |
|
"loss": 1.1506, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 1.62, |
|
"learning_rate": 1.0782608695652175e-05, |
|
"loss": 1.1536, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 1.65, |
|
"learning_rate": 1.0589371980676328e-05, |
|
"loss": 1.1449, |
|
"step": 670 |
|
}, |
|
{ |
|
"epoch": 1.67, |
|
"learning_rate": 1.0396135265700484e-05, |
|
"loss": 1.1287, |
|
"step": 680 |
|
}, |
|
{ |
|
"epoch": 1.7, |
|
"learning_rate": 1.0202898550724639e-05, |
|
"loss": 1.1366, |
|
"step": 690 |
|
}, |
|
{ |
|
"epoch": 1.72, |
|
"learning_rate": 1.0009661835748795e-05, |
|
"loss": 1.1468, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 1.75, |
|
"learning_rate": 9.816425120772947e-06, |
|
"loss": 1.1484, |
|
"step": 710 |
|
}, |
|
{ |
|
"epoch": 1.77, |
|
"learning_rate": 9.623188405797102e-06, |
|
"loss": 1.1549, |
|
"step": 720 |
|
}, |
|
{ |
|
"epoch": 1.8, |
|
"learning_rate": 9.429951690821256e-06, |
|
"loss": 1.1501, |
|
"step": 730 |
|
}, |
|
{ |
|
"epoch": 1.82, |
|
"learning_rate": 9.23671497584541e-06, |
|
"loss": 1.1433, |
|
"step": 740 |
|
}, |
|
{ |
|
"epoch": 1.85, |
|
"learning_rate": 9.043478260869565e-06, |
|
"loss": 1.1315, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 1.87, |
|
"learning_rate": 8.850241545893721e-06, |
|
"loss": 1.1615, |
|
"step": 760 |
|
}, |
|
{ |
|
"epoch": 1.9, |
|
"learning_rate": 8.657004830917876e-06, |
|
"loss": 1.1521, |
|
"step": 770 |
|
}, |
|
{ |
|
"epoch": 1.92, |
|
"learning_rate": 8.46376811594203e-06, |
|
"loss": 1.138, |
|
"step": 780 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"learning_rate": 8.270531400966184e-06, |
|
"loss": 1.1608, |
|
"step": 790 |
|
}, |
|
{ |
|
"epoch": 1.97, |
|
"learning_rate": 8.077294685990339e-06, |
|
"loss": 1.1655, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 1.99, |
|
"learning_rate": 7.884057971014493e-06, |
|
"loss": 1.1367, |
|
"step": 810 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 7.690821256038648e-06, |
|
"loss": 1.1201, |
|
"step": 820 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 7.497584541062802e-06, |
|
"loss": 1.1403, |
|
"step": 830 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 7.304347826086957e-06, |
|
"loss": 1.1166, |
|
"step": 840 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 7.111111111111112e-06, |
|
"loss": 1.1243, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 6.917874396135267e-06, |
|
"loss": 1.0965, |
|
"step": 860 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 6.724637681159421e-06, |
|
"loss": 1.1033, |
|
"step": 870 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 6.531400966183575e-06, |
|
"loss": 1.1162, |
|
"step": 880 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 6.33816425120773e-06, |
|
"loss": 1.1242, |
|
"step": 890 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 6.144927536231885e-06, |
|
"loss": 1.1151, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 2.24, |
|
"learning_rate": 5.951690821256038e-06, |
|
"loss": 1.1068, |
|
"step": 910 |
|
}, |
|
{ |
|
"epoch": 2.26, |
|
"learning_rate": 5.758454106280194e-06, |
|
"loss": 1.1132, |
|
"step": 920 |
|
}, |
|
{ |
|
"epoch": 2.29, |
|
"learning_rate": 5.565217391304348e-06, |
|
"loss": 1.1307, |
|
"step": 930 |
|
}, |
|
{ |
|
"epoch": 2.31, |
|
"learning_rate": 5.371980676328503e-06, |
|
"loss": 1.1174, |
|
"step": 940 |
|
}, |
|
{ |
|
"epoch": 2.34, |
|
"learning_rate": 5.178743961352657e-06, |
|
"loss": 1.1135, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 2.36, |
|
"learning_rate": 4.985507246376812e-06, |
|
"loss": 1.0946, |
|
"step": 960 |
|
}, |
|
{ |
|
"epoch": 2.39, |
|
"learning_rate": 4.7922705314009665e-06, |
|
"loss": 1.1089, |
|
"step": 970 |
|
}, |
|
{ |
|
"epoch": 2.41, |
|
"learning_rate": 4.599033816425121e-06, |
|
"loss": 1.116, |
|
"step": 980 |
|
}, |
|
{ |
|
"epoch": 2.44, |
|
"learning_rate": 4.405797101449275e-06, |
|
"loss": 1.1129, |
|
"step": 990 |
|
}, |
|
{ |
|
"epoch": 2.46, |
|
"learning_rate": 4.212560386473431e-06, |
|
"loss": 1.0921, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 2.49, |
|
"learning_rate": 4.019323671497585e-06, |
|
"loss": 1.0999, |
|
"step": 1010 |
|
}, |
|
{ |
|
"epoch": 2.51, |
|
"learning_rate": 3.8260869565217395e-06, |
|
"loss": 1.1141, |
|
"step": 1020 |
|
}, |
|
{ |
|
"epoch": 2.54, |
|
"learning_rate": 3.632850241545894e-06, |
|
"loss": 1.1274, |
|
"step": 1030 |
|
}, |
|
{ |
|
"epoch": 2.56, |
|
"learning_rate": 3.4396135265700487e-06, |
|
"loss": 1.1194, |
|
"step": 1040 |
|
}, |
|
{ |
|
"epoch": 2.58, |
|
"learning_rate": 3.2463768115942027e-06, |
|
"loss": 1.1065, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 2.61, |
|
"learning_rate": 3.0531400966183576e-06, |
|
"loss": 1.1107, |
|
"step": 1060 |
|
}, |
|
{ |
|
"epoch": 2.63, |
|
"learning_rate": 2.8599033816425124e-06, |
|
"loss": 1.112, |
|
"step": 1070 |
|
}, |
|
{ |
|
"epoch": 2.66, |
|
"learning_rate": 2.666666666666667e-06, |
|
"loss": 1.1252, |
|
"step": 1080 |
|
}, |
|
{ |
|
"epoch": 2.68, |
|
"learning_rate": 2.4734299516908212e-06, |
|
"loss": 1.1116, |
|
"step": 1090 |
|
}, |
|
{ |
|
"epoch": 2.71, |
|
"learning_rate": 2.280193236714976e-06, |
|
"loss": 1.0865, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 2.73, |
|
"learning_rate": 2.0869565217391305e-06, |
|
"loss": 1.1122, |
|
"step": 1110 |
|
}, |
|
{ |
|
"epoch": 2.76, |
|
"learning_rate": 1.8937198067632853e-06, |
|
"loss": 1.1225, |
|
"step": 1120 |
|
}, |
|
{ |
|
"epoch": 2.78, |
|
"learning_rate": 1.7004830917874398e-06, |
|
"loss": 1.1077, |
|
"step": 1130 |
|
}, |
|
{ |
|
"epoch": 2.81, |
|
"learning_rate": 1.5072463768115944e-06, |
|
"loss": 1.1104, |
|
"step": 1140 |
|
}, |
|
{ |
|
"epoch": 2.83, |
|
"learning_rate": 1.314009661835749e-06, |
|
"loss": 1.0929, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 2.86, |
|
"learning_rate": 1.1207729468599034e-06, |
|
"loss": 1.0986, |
|
"step": 1160 |
|
}, |
|
{ |
|
"epoch": 2.88, |
|
"learning_rate": 9.275362318840581e-07, |
|
"loss": 1.1137, |
|
"step": 1170 |
|
}, |
|
{ |
|
"epoch": 2.9, |
|
"learning_rate": 7.342995169082126e-07, |
|
"loss": 1.0868, |
|
"step": 1180 |
|
}, |
|
{ |
|
"epoch": 2.93, |
|
"learning_rate": 5.410628019323672e-07, |
|
"loss": 1.1027, |
|
"step": 1190 |
|
}, |
|
{ |
|
"epoch": 2.95, |
|
"learning_rate": 3.4782608695652175e-07, |
|
"loss": 1.1193, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 2.98, |
|
"learning_rate": 1.5458937198067633e-07, |
|
"loss": 1.1247, |
|
"step": 1210 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"step": 1218, |
|
"total_flos": 1.1227700242944e+16, |
|
"train_loss": 1.3416433044646565, |
|
"train_runtime": 4283.2723, |
|
"train_samples_per_second": 36.422, |
|
"train_steps_per_second": 0.284 |
|
} |
|
], |
|
"max_steps": 1218, |
|
"num_train_epochs": 3, |
|
"total_flos": 1.1227700242944e+16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|