| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 4881, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00029877074370006146, | |
| "loss": 2.9685, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00029754148740012294, | |
| "loss": 2.2769, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00029631223110018436, | |
| "loss": 2.1295, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00029508297480024584, | |
| "loss": 2.0832, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00029385371850030727, | |
| "loss": 2.0765, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00029262446220036875, | |
| "loss": 2.0831, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00029139520590043023, | |
| "loss": 2.1462, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00029016594960049166, | |
| "loss": 2.1126, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00028893669330055314, | |
| "loss": 2.0809, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002877074370006146, | |
| "loss": 2.1573, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00028647818070067605, | |
| "loss": 2.1391, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00028524892440073753, | |
| "loss": 2.1265, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00028401966810079896, | |
| "loss": 2.0266, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00028279041180086044, | |
| "loss": 2.0723, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0002815611555009219, | |
| "loss": 2.1082, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0002803318992009834, | |
| "loss": 2.061, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0002791026429010449, | |
| "loss": 2.0605, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0002778733866011063, | |
| "loss": 2.0401, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00027664413030116774, | |
| "loss": 2.0441, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0002754148740012292, | |
| "loss": 2.0578, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0002741856177012907, | |
| "loss": 2.0634, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00027295636140135213, | |
| "loss": 1.9869, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0002717271051014136, | |
| "loss": 2.0628, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0002704978488014751, | |
| "loss": 2.0732, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0002692685925015366, | |
| "loss": 2.038, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.000268039336201598, | |
| "loss": 2.0333, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0002668100799016595, | |
| "loss": 1.97, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0002655808236017209, | |
| "loss": 2.0097, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0002643515673017824, | |
| "loss": 1.9377, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00026312231100184387, | |
| "loss": 2.059, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.00026189305470190535, | |
| "loss": 1.9423, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0002606637984019668, | |
| "loss": 2.0869, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00025943454210202826, | |
| "loss": 1.9231, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0002582052858020897, | |
| "loss": 2.0086, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00025697602950215117, | |
| "loss": 1.9623, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00025574677320221265, | |
| "loss": 1.9348, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0002545175169022741, | |
| "loss": 2.0155, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00025328826060233556, | |
| "loss": 1.9043, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00025205900430239704, | |
| "loss": 2.0102, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0002508297480024585, | |
| "loss": 1.9552, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00024960049170251995, | |
| "loss": 1.9779, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0002483712354025814, | |
| "loss": 1.981, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00024714197910264286, | |
| "loss": 1.9206, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00024591272280270434, | |
| "loss": 1.9573, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0002446834665027658, | |
| "loss": 2.0444, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00024345421020282728, | |
| "loss": 1.8819, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00024222495390288873, | |
| "loss": 1.9977, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00024099569760295018, | |
| "loss": 1.946, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00023976644130301167, | |
| "loss": 1.953, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00023853718500307312, | |
| "loss": 2.0032, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.0002373079287031346, | |
| "loss": 1.9271, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00023607867240319603, | |
| "loss": 2.0297, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0002348494161032575, | |
| "loss": 2.049, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.00023362015980331896, | |
| "loss": 1.8794, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.00023239090350338044, | |
| "loss": 1.9826, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00023116164720344193, | |
| "loss": 2.0297, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00022993239090350335, | |
| "loss": 1.9343, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0002287031346035648, | |
| "loss": 1.956, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.0002274738783036263, | |
| "loss": 1.9653, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00022624462200368777, | |
| "loss": 1.9609, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.00022501536570374922, | |
| "loss": 2.0, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00022378610940381065, | |
| "loss": 2.0268, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00022255685310387213, | |
| "loss": 2.0342, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00022132759680393361, | |
| "loss": 1.9875, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.00022009834050399507, | |
| "loss": 1.9526, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.00021886908420405652, | |
| "loss": 1.9678, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.00021763982790411798, | |
| "loss": 2.1014, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.00021641057160417946, | |
| "loss": 1.9, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0002151813153042409, | |
| "loss": 1.9648, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0002139520590043024, | |
| "loss": 2.0172, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.00021272280270436382, | |
| "loss": 1.9327, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0002114935464044253, | |
| "loss": 1.9365, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.00021026429010448676, | |
| "loss": 1.9678, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.00020903503380454824, | |
| "loss": 1.9248, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0002078057775046097, | |
| "loss": 1.9312, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.00020657652120467115, | |
| "loss": 1.9107, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.0002053472649047326, | |
| "loss": 1.9331, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.00020411800860479408, | |
| "loss": 1.9028, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.00020288875230485556, | |
| "loss": 1.9537, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.00020165949600491702, | |
| "loss": 1.9189, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.00020043023970497845, | |
| "loss": 1.9677, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.00019920098340503993, | |
| "loss": 1.8744, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0001979717271051014, | |
| "loss": 1.9705, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00019674247080516286, | |
| "loss": 1.9018, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00019551321450522434, | |
| "loss": 1.9697, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.00019428395820528577, | |
| "loss": 1.9494, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00019305470190534725, | |
| "loss": 1.9556, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.0001918254456054087, | |
| "loss": 2.0141, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.0001905961893054702, | |
| "loss": 1.9397, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.00018936693300553164, | |
| "loss": 1.9258, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.0001881376767055931, | |
| "loss": 1.8207, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00018690842040565455, | |
| "loss": 1.9226, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00018567916410571603, | |
| "loss": 1.9205, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00018444990780577749, | |
| "loss": 1.9983, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00018322065150583897, | |
| "loss": 1.9727, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.0001819913952059004, | |
| "loss": 1.9329, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00018076213890596188, | |
| "loss": 2.0107, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00017953288260602333, | |
| "loss": 1.9328, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.0001783036263060848, | |
| "loss": 1.8733, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.0001770743700061463, | |
| "loss": 1.9073, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00017584511370620772, | |
| "loss": 1.9431, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.0001746158574062692, | |
| "loss": 1.8844, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00017338660110633066, | |
| "loss": 1.8902, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.00017215734480639214, | |
| "loss": 1.8887, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.00017092808850645356, | |
| "loss": 1.8653, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.00016969883220651505, | |
| "loss": 1.9092, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 0.0001684695759065765, | |
| "loss": 1.9427, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.00016724031960663798, | |
| "loss": 2.0016, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 0.00016601106330669943, | |
| "loss": 1.9192, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.0001647818070067609, | |
| "loss": 1.8911, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.00016355255070682234, | |
| "loss": 1.9627, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 0.00016232329440688382, | |
| "loss": 2.022, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.00016109403810694528, | |
| "loss": 1.8829, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.00015986478180700676, | |
| "loss": 2.0244, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.0001586355255070682, | |
| "loss": 1.9201, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 0.00015740626920712967, | |
| "loss": 1.9803, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.00015617701290719112, | |
| "loss": 1.9369, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.0001549477566072526, | |
| "loss": 1.9151, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 0.00015371850030731409, | |
| "loss": 1.8436, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 0.0001524892440073755, | |
| "loss": 1.9259, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 0.00015125998770743697, | |
| "loss": 1.8486, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.00015003073140749845, | |
| "loss": 1.8839, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.00014880147510755993, | |
| "loss": 1.917, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 0.00014757221880762138, | |
| "loss": 1.9474, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 0.00014634296250768284, | |
| "loss": 1.9225, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 0.0001451137062077443, | |
| "loss": 1.9257, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.00014388444990780577, | |
| "loss": 1.9105, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 0.00014265519360786723, | |
| "loss": 1.9799, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.00014142593730792868, | |
| "loss": 1.9632, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 0.00014019668100799016, | |
| "loss": 1.888, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 0.00013896742470805162, | |
| "loss": 1.9576, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 0.00013773816840811307, | |
| "loss": 2.0318, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 0.00013650891210817455, | |
| "loss": 1.9347, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.000135279655808236, | |
| "loss": 1.8473, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 0.00013405039950829746, | |
| "loss": 1.9583, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.00013282114320835892, | |
| "loss": 1.8072, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.0001315918869084204, | |
| "loss": 1.8624, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.00013036263060848185, | |
| "loss": 1.9646, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 0.0001291333743085433, | |
| "loss": 1.9538, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.0001279041180086048, | |
| "loss": 1.8872, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.00012667486170866624, | |
| "loss": 1.9137, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 0.00012544560540872772, | |
| "loss": 1.9049, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.00012421634910878918, | |
| "loss": 1.9567, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 0.00012298709280885063, | |
| "loss": 1.9053, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 0.0001217578365089121, | |
| "loss": 1.8927, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.00012052858020897355, | |
| "loss": 1.9475, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 0.00011929932390903502, | |
| "loss": 2.0315, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.00011807006760909648, | |
| "loss": 1.8144, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 0.00011684081130915794, | |
| "loss": 1.8562, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 0.00011561155500921942, | |
| "loss": 1.925, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 0.00011438229870928088, | |
| "loss": 1.9984, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 0.00011315304240934235, | |
| "loss": 1.9431, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.0001119237861094038, | |
| "loss": 1.9025, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 0.00011069452980946527, | |
| "loss": 1.9075, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.00010946527350952672, | |
| "loss": 1.9387, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 0.00010823601720958819, | |
| "loss": 1.8919, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.00010700676090964966, | |
| "loss": 1.9138, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 0.00010577750460971111, | |
| "loss": 1.9591, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.00010454824830977258, | |
| "loss": 1.9794, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.00010331899200983403, | |
| "loss": 1.9314, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.0001020897357098955, | |
| "loss": 1.9484, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 0.00010086047940995697, | |
| "loss": 1.839, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 9.963122311001842e-05, | |
| "loss": 1.8986, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.840196681007989e-05, | |
| "loss": 1.9466, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 9.717271051014135e-05, | |
| "loss": 1.9425, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 9.594345421020281e-05, | |
| "loss": 1.8959, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 9.471419791026428e-05, | |
| "loss": 1.8946, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 9.348494161032574e-05, | |
| "loss": 1.974, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 9.225568531038722e-05, | |
| "loss": 1.8717, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 9.102642901044866e-05, | |
| "loss": 1.9481, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 8.979717271051014e-05, | |
| "loss": 1.931, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 8.856791641057161e-05, | |
| "loss": 1.9382, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 8.733866011063306e-05, | |
| "loss": 1.9577, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 8.610940381069453e-05, | |
| "loss": 1.8826, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 8.488014751075598e-05, | |
| "loss": 1.902, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 8.365089121081745e-05, | |
| "loss": 1.9147, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 8.24216349108789e-05, | |
| "loss": 1.8675, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 8.119237861094037e-05, | |
| "loss": 1.877, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 7.996312231100184e-05, | |
| "loss": 2.0258, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 7.87338660110633e-05, | |
| "loss": 1.9204, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 7.750460971112476e-05, | |
| "loss": 1.9073, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 7.627535341118622e-05, | |
| "loss": 1.8481, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 7.504609711124769e-05, | |
| "loss": 1.896, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 7.381684081130915e-05, | |
| "loss": 1.9285, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 7.258758451137061e-05, | |
| "loss": 1.8239, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 7.135832821143208e-05, | |
| "loss": 1.9289, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 7.012907191149354e-05, | |
| "loss": 1.8431, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 6.8899815611555e-05, | |
| "loss": 1.9239, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 6.767055931161647e-05, | |
| "loss": 1.9132, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 6.644130301167793e-05, | |
| "loss": 1.9019, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 6.521204671173939e-05, | |
| "loss": 1.9947, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 6.398279041180086e-05, | |
| "loss": 1.9146, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 6.275353411186232e-05, | |
| "loss": 1.8672, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 6.152427781192378e-05, | |
| "loss": 1.9189, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 6.0295021511985245e-05, | |
| "loss": 1.8431, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 5.9065765212046706e-05, | |
| "loss": 1.9629, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 5.7836508912108174e-05, | |
| "loss": 1.8209, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 5.6607252612169635e-05, | |
| "loss": 1.9104, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 5.5377996312231096e-05, | |
| "loss": 1.8563, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 5.414874001229256e-05, | |
| "loss": 1.9091, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 5.291948371235402e-05, | |
| "loss": 1.9038, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 5.169022741241548e-05, | |
| "loss": 1.9083, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 5.046097111247695e-05, | |
| "loss": 1.9037, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 4.9231714812538415e-05, | |
| "loss": 1.8808, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 4.8002458512599876e-05, | |
| "loss": 1.9638, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 4.677320221266134e-05, | |
| "loss": 1.9124, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 4.55439459127228e-05, | |
| "loss": 1.9119, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 4.4314689612784266e-05, | |
| "loss": 1.8596, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 4.3085433312845727e-05, | |
| "loss": 1.8779, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 4.185617701290719e-05, | |
| "loss": 1.9442, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.062692071296865e-05, | |
| "loss": 1.8899, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 3.939766441303011e-05, | |
| "loss": 1.8925, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 3.816840811309157e-05, | |
| "loss": 1.9306, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 3.693915181315304e-05, | |
| "loss": 1.988, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 3.5709895513214506e-05, | |
| "loss": 1.8516, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 3.448063921327597e-05, | |
| "loss": 1.9874, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 3.325138291333743e-05, | |
| "loss": 1.9059, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 3.202212661339889e-05, | |
| "loss": 1.9219, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 3.079287031346035e-05, | |
| "loss": 1.8807, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 2.9563614013521815e-05, | |
| "loss": 1.842, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 2.833435771358328e-05, | |
| "loss": 1.9281, | |
| "step": 4420 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 2.7105101413644744e-05, | |
| "loss": 1.8918, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 2.5875845113706208e-05, | |
| "loss": 1.8154, | |
| "step": 4460 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 2.464658881376767e-05, | |
| "loss": 1.8775, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 2.341733251382913e-05, | |
| "loss": 1.9259, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 2.2188076213890594e-05, | |
| "loss": 1.8731, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 2.095881991395206e-05, | |
| "loss": 1.916, | |
| "step": 4540 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 1.972956361401352e-05, | |
| "loss": 1.8511, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 1.850030731407498e-05, | |
| "loss": 1.8983, | |
| "step": 4580 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 1.7271051014136445e-05, | |
| "loss": 1.8668, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 1.604179471419791e-05, | |
| "loss": 1.8947, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 1.4812538414259372e-05, | |
| "loss": 1.9118, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 1.3583282114320835e-05, | |
| "loss": 1.9495, | |
| "step": 4660 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 1.2354025814382298e-05, | |
| "loss": 1.9175, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.112476951444376e-05, | |
| "loss": 1.9496, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 9.895513214505223e-06, | |
| "loss": 1.9861, | |
| "step": 4720 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 8.666256914566686e-06, | |
| "loss": 1.8718, | |
| "step": 4740 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 7.4370006146281495e-06, | |
| "loss": 1.88, | |
| "step": 4760 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 6.207744314689612e-06, | |
| "loss": 1.8669, | |
| "step": 4780 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.978488014751076e-06, | |
| "loss": 1.8853, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 3.749231714812538e-06, | |
| "loss": 1.958, | |
| "step": 4820 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 2.519975414874001e-06, | |
| "loss": 1.8813, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 1.2907191149354641e-06, | |
| "loss": 1.8783, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 6.146281499692685e-08, | |
| "loss": 1.9501, | |
| "step": 4880 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 1.9423085451126099, | |
| "eval_runtime": 128.9524, | |
| "eval_samples_per_second": 15.51, | |
| "eval_steps_per_second": 1.939, | |
| "step": 4881 | |
| } | |
| ], | |
| "max_steps": 4881, | |
| "num_train_epochs": 3, | |
| "total_flos": 4.564905941532672e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |