{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.999072786277237, "global_step": 2156, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 4.999734597774032e-05, "loss": 1.7592, "step": 10 }, { "epoch": 0.02, "learning_rate": 4.998938447446803e-05, "loss": 1.6546, "step": 20 }, { "epoch": 0.03, "learning_rate": 4.997611718058365e-05, "loss": 1.5632, "step": 30 }, { "epoch": 0.04, "learning_rate": 4.9957546913022665e-05, "loss": 1.5245, "step": 40 }, { "epoch": 0.05, "learning_rate": 4.993367761465736e-05, "loss": 1.5049, "step": 50 }, { "epoch": 0.06, "learning_rate": 4.9904514353459654e-05, "loss": 1.492, "step": 60 }, { "epoch": 0.06, "learning_rate": 4.9870063321425105e-05, "loss": 1.4803, "step": 70 }, { "epoch": 0.07, "learning_rate": 4.983033183325818e-05, "loss": 1.4827, "step": 80 }, { "epoch": 0.08, "learning_rate": 4.97853283248192e-05, "loss": 1.4866, "step": 90 }, { "epoch": 0.09, "learning_rate": 4.973506235133323e-05, "loss": 1.486, "step": 100 }, { "epoch": 0.1, "learning_rate": 4.967954458536126e-05, "loss": 1.4822, "step": 110 }, { "epoch": 0.11, "learning_rate": 4.9618786814534226e-05, "loss": 1.4801, "step": 120 }, { "epoch": 0.12, "learning_rate": 4.955280193905022e-05, "loss": 1.4764, "step": 130 }, { "epoch": 0.13, "learning_rate": 4.948160396893553e-05, "loss": 1.4662, "step": 140 }, { "epoch": 0.14, "learning_rate": 4.9405208021069946e-05, "loss": 1.4627, "step": 150 }, { "epoch": 0.15, "learning_rate": 4.9323630315977156e-05, "loss": 1.4543, "step": 160 }, { "epoch": 0.16, "learning_rate": 4.9236888174380784e-05, "loss": 1.4508, "step": 170 }, { "epoch": 0.17, "learning_rate": 4.91450000135268e-05, "loss": 1.464, "step": 180 }, { "epoch": 0.18, "learning_rate": 4.9047985343273154e-05, "loss": 1.4705, "step": 190 }, { "epoch": 0.19, "learning_rate": 4.894586476194739e-05, "loss": 1.4588, "step": 200 }, { "epoch": 0.19, "learning_rate": 4.883865995197319e-05, "loss": 1.4532, "step": 210 }, { "epoch": 0.2, "learning_rate": 4.8726393675266716e-05, "loss": 1.4629, "step": 220 }, { "epoch": 0.21, "learning_rate": 4.860908976840376e-05, "loss": 1.4429, "step": 230 }, { "epoch": 0.22, "learning_rate": 4.848677313755872e-05, "loss": 1.441, "step": 240 }, { "epoch": 0.23, "learning_rate": 4.835946975321647e-05, "loss": 1.4496, "step": 250 }, { "epoch": 0.24, "learning_rate": 4.822720664465827e-05, "loss": 1.4459, "step": 260 }, { "epoch": 0.25, "learning_rate": 4.809001189422287e-05, "loss": 1.4458, "step": 270 }, { "epoch": 0.26, "learning_rate": 4.794791463134399e-05, "loss": 1.4484, "step": 280 }, { "epoch": 0.27, "learning_rate": 4.780094502636552e-05, "loss": 1.4541, "step": 290 }, { "epoch": 0.28, "learning_rate": 4.764913428413572e-05, "loss": 1.437, "step": 300 }, { "epoch": 0.29, "learning_rate": 4.7492514637381727e-05, "loss": 1.4368, "step": 310 }, { "epoch": 0.3, "learning_rate": 4.733111933986583e-05, "loss": 1.4449, "step": 320 }, { "epoch": 0.31, "learning_rate": 4.716498265932501e-05, "loss": 1.4256, "step": 330 }, { "epoch": 0.32, "learning_rate": 4.699413987019512e-05, "loss": 1.4378, "step": 340 }, { "epoch": 0.32, "learning_rate": 4.681862724612141e-05, "loss": 1.454, "step": 350 }, { "epoch": 0.33, "learning_rate": 4.663848205225674e-05, "loss": 1.4207, "step": 360 }, { "epoch": 0.34, "learning_rate": 4.645374253734949e-05, "loss": 1.4286, "step": 370 }, { "epoch": 0.35, "learning_rate": 4.626444792562244e-05, "loss": 1.4244, "step": 380 }, { "epoch": 0.36, "learning_rate": 4.607063840844463e-05, "loss": 1.4342, "step": 390 }, { "epoch": 0.37, "learning_rate": 4.587235513579791e-05, "loss": 1.4241, "step": 400 }, { "epoch": 0.38, "learning_rate": 4.5669640207539786e-05, "loss": 1.443, "step": 410 }, { "epoch": 0.39, "learning_rate": 4.546253666446484e-05, "loss": 1.4281, "step": 420 }, { "epoch": 0.4, "learning_rate": 4.525108847916614e-05, "loss": 1.4193, "step": 430 }, { "epoch": 0.41, "learning_rate": 4.503534054669892e-05, "loss": 1.4154, "step": 440 }, { "epoch": 0.42, "learning_rate": 4.481533867504841e-05, "loss": 1.4273, "step": 450 }, { "epoch": 0.43, "learning_rate": 4.4591129575403765e-05, "loss": 1.4329, "step": 460 }, { "epoch": 0.44, "learning_rate": 4.43627608522403e-05, "loss": 1.4102, "step": 470 }, { "epoch": 0.45, "learning_rate": 4.4130280993211974e-05, "loss": 1.4244, "step": 480 }, { "epoch": 0.45, "learning_rate": 4.389373935885646e-05, "loss": 1.4407, "step": 490 }, { "epoch": 0.46, "learning_rate": 4.365318617211479e-05, "loss": 1.4282, "step": 500 }, { "epoch": 0.47, "learning_rate": 4.340867250766794e-05, "loss": 1.4269, "step": 510 }, { "epoch": 0.48, "learning_rate": 4.316025028109258e-05, "loss": 1.43, "step": 520 }, { "epoch": 0.49, "learning_rate": 4.2907972237838225e-05, "loss": 1.4271, "step": 530 }, { "epoch": 0.5, "learning_rate": 4.2651891942028274e-05, "loss": 1.4321, "step": 540 }, { "epoch": 0.51, "learning_rate": 4.239206376508717e-05, "loss": 1.4083, "step": 550 }, { "epoch": 0.52, "learning_rate": 4.212854287419611e-05, "loss": 1.4168, "step": 560 }, { "epoch": 0.53, "learning_rate": 4.1861385220579934e-05, "loss": 1.4144, "step": 570 }, { "epoch": 0.54, "learning_rate": 4.1590647527627404e-05, "loss": 1.4106, "step": 580 }, { "epoch": 0.55, "learning_rate": 4.131638727884762e-05, "loss": 1.4202, "step": 590 }, { "epoch": 0.56, "learning_rate": 4.103866270566498e-05, "loss": 1.417, "step": 600 }, { "epoch": 0.57, "learning_rate": 4.075753277505544e-05, "loss": 1.4324, "step": 610 }, { "epoch": 0.57, "learning_rate": 4.0473057177026484e-05, "loss": 1.4147, "step": 620 }, { "epoch": 0.58, "learning_rate": 4.018529631194369e-05, "loss": 1.4269, "step": 630 }, { "epoch": 0.59, "learning_rate": 3.989431127770635e-05, "loss": 1.4133, "step": 640 }, { "epoch": 0.6, "learning_rate": 3.960016385677513e-05, "loss": 1.4173, "step": 650 }, { "epoch": 0.61, "learning_rate": 3.9302916503054246e-05, "loss": 1.4007, "step": 660 }, { "epoch": 0.62, "learning_rate": 3.9002632328631164e-05, "loss": 1.4117, "step": 670 }, { "epoch": 0.63, "learning_rate": 3.8699375090376534e-05, "loss": 1.4247, "step": 680 }, { "epoch": 0.64, "learning_rate": 3.8393209176407223e-05, "loss": 1.4312, "step": 690 }, { "epoch": 0.65, "learning_rate": 3.8084199592415305e-05, "loss": 1.4136, "step": 700 }, { "epoch": 0.66, "learning_rate": 3.777241194786591e-05, "loss": 1.4132, "step": 710 }, { "epoch": 0.67, "learning_rate": 3.745791244206697e-05, "loss": 1.4145, "step": 720 }, { "epoch": 0.68, "learning_rate": 3.714076785011359e-05, "loss": 1.3951, "step": 730 }, { "epoch": 0.69, "learning_rate": 3.682104550871031e-05, "loss": 1.4139, "step": 740 }, { "epoch": 0.7, "learning_rate": 3.649881330187401e-05, "loss": 1.4078, "step": 750 }, { "epoch": 0.7, "learning_rate": 3.617413964652067e-05, "loss": 1.4048, "step": 760 }, { "epoch": 0.71, "learning_rate": 3.5847093477938956e-05, "loss": 1.4115, "step": 770 }, { "epoch": 0.72, "learning_rate": 3.551774423515378e-05, "loss": 1.4018, "step": 780 }, { "epoch": 0.73, "learning_rate": 3.518616184618288e-05, "loss": 1.4091, "step": 790 }, { "epoch": 0.74, "learning_rate": 3.4852416713189526e-05, "loss": 1.4135, "step": 800 }, { "epoch": 0.75, "learning_rate": 3.4516579697534705e-05, "loss": 1.3977, "step": 810 }, { "epoch": 0.76, "learning_rate": 3.417872210473162e-05, "loss": 1.4123, "step": 820 }, { "epoch": 0.77, "learning_rate": 3.3838915669306034e-05, "loss": 1.4096, "step": 830 }, { "epoch": 0.78, "learning_rate": 3.349723253956542e-05, "loss": 1.4031, "step": 840 }, { "epoch": 0.79, "learning_rate": 3.315374526228036e-05, "loss": 1.4104, "step": 850 }, { "epoch": 0.8, "learning_rate": 3.2808526767281225e-05, "loss": 1.3972, "step": 860 }, { "epoch": 0.81, "learning_rate": 3.246165035197364e-05, "loss": 1.3987, "step": 870 }, { "epoch": 0.82, "learning_rate": 3.211318966577581e-05, "loss": 1.4197, "step": 880 }, { "epoch": 0.83, "learning_rate": 3.176321869448116e-05, "loss": 1.4169, "step": 890 }, { "epoch": 0.83, "learning_rate": 3.1411811744549536e-05, "loss": 1.4055, "step": 900 }, { "epoch": 0.84, "learning_rate": 3.105904342733032e-05, "loss": 1.402, "step": 910 }, { "epoch": 0.85, "learning_rate": 3.070498864322081e-05, "loss": 1.4148, "step": 920 }, { "epoch": 0.86, "learning_rate": 3.034972256576328e-05, "loss": 1.4007, "step": 930 }, { "epoch": 0.87, "learning_rate": 2.999332062568395e-05, "loss": 1.3926, "step": 940 }, { "epoch": 0.88, "learning_rate": 2.9635858494877384e-05, "loss": 1.4099, "step": 950 }, { "epoch": 0.89, "learning_rate": 2.9277412070339782e-05, "loss": 1.401, "step": 960 }, { "epoch": 0.9, "learning_rate": 2.891805745805429e-05, "loss": 1.3995, "step": 970 }, { "epoch": 0.91, "learning_rate": 2.8557870956832132e-05, "loss": 1.4108, "step": 980 }, { "epoch": 0.92, "learning_rate": 2.8196929042112652e-05, "loss": 1.4006, "step": 990 }, { "epoch": 0.93, "learning_rate": 2.783530834972594e-05, "loss": 1.4045, "step": 1000 }, { "epoch": 0.94, "learning_rate": 2.7473085659621377e-05, "loss": 1.3907, "step": 1010 }, { "epoch": 0.95, "learning_rate": 2.711033787956555e-05, "loss": 1.3872, "step": 1020 }, { "epoch": 0.96, "learning_rate": 2.6747142028813105e-05, "loss": 1.405, "step": 1030 }, { "epoch": 0.96, "learning_rate": 2.638357522175383e-05, "loss": 1.3952, "step": 1040 }, { "epoch": 0.97, "learning_rate": 2.6019714651539646e-05, "loss": 1.3901, "step": 1050 }, { "epoch": 0.98, "learning_rate": 2.565563757369475e-05, "loss": 1.4027, "step": 1060 }, { "epoch": 0.99, "learning_rate": 2.529142128971268e-05, "loss": 1.3909, "step": 1070 }, { "epoch": 1.0, "learning_rate": 2.492714313064342e-05, "loss": 1.4039, "step": 1080 }, { "epoch": 1.01, "learning_rate": 2.45628804406744e-05, "loss": 1.3964, "step": 1090 }, { "epoch": 1.02, "learning_rate": 2.419871056070862e-05, "loss": 1.4076, "step": 1100 }, { "epoch": 1.03, "learning_rate": 2.3834710811943514e-05, "loss": 1.3891, "step": 1110 }, { "epoch": 1.04, "learning_rate": 2.3470958479453938e-05, "loss": 1.4004, "step": 1120 }, { "epoch": 1.05, "learning_rate": 2.3107530795782877e-05, "loss": 1.3879, "step": 1130 }, { "epoch": 1.06, "learning_rate": 2.2744504924543313e-05, "loss": 1.3879, "step": 1140 }, { "epoch": 1.07, "learning_rate": 2.23819579440347e-05, "loss": 1.3851, "step": 1150 }, { "epoch": 1.08, "learning_rate": 2.2019966830877545e-05, "loss": 1.3846, "step": 1160 }, { "epoch": 1.08, "learning_rate": 2.1658608443669635e-05, "loss": 1.3936, "step": 1170 }, { "epoch": 1.09, "learning_rate": 2.1297959506667224e-05, "loss": 1.3841, "step": 1180 }, { "epoch": 1.1, "learning_rate": 2.0938096593494855e-05, "loss": 1.3969, "step": 1190 }, { "epoch": 1.11, "learning_rate": 2.057909611088709e-05, "loss": 1.3953, "step": 1200 }, { "epoch": 1.12, "learning_rate": 2.02210342824657e-05, "loss": 1.3951, "step": 1210 }, { "epoch": 1.13, "learning_rate": 1.9863987132555706e-05, "loss": 1.4012, "step": 1220 }, { "epoch": 1.14, "learning_rate": 1.9508030470043806e-05, "loss": 1.3998, "step": 1230 }, { "epoch": 1.15, "learning_rate": 1.915323987228247e-05, "loss": 1.3933, "step": 1240 }, { "epoch": 1.16, "learning_rate": 1.8799690669043212e-05, "loss": 1.3977, "step": 1250 }, { "epoch": 1.17, "learning_rate": 1.8447457926522454e-05, "loss": 1.3818, "step": 1260 }, { "epoch": 1.18, "learning_rate": 1.8096616431403325e-05, "loss": 1.4032, "step": 1270 }, { "epoch": 1.19, "learning_rate": 1.7747240674976857e-05, "loss": 1.3853, "step": 1280 }, { "epoch": 1.2, "learning_rate": 1.7399404837325796e-05, "loss": 1.3988, "step": 1290 }, { "epoch": 1.21, "learning_rate": 1.7053182771574633e-05, "loss": 1.3963, "step": 1300 }, { "epoch": 1.21, "learning_rate": 1.6708647988208887e-05, "loss": 1.3881, "step": 1310 }, { "epoch": 1.22, "learning_rate": 1.6365873639467315e-05, "loss": 1.3804, "step": 1320 }, { "epoch": 1.23, "learning_rate": 1.602493250381003e-05, "loss": 1.3952, "step": 1330 }, { "epoch": 1.24, "learning_rate": 1.5685896970466123e-05, "loss": 1.4015, "step": 1340 }, { "epoch": 1.25, "learning_rate": 1.534883902406375e-05, "loss": 1.3797, "step": 1350 }, { "epoch": 1.26, "learning_rate": 1.5013830229346326e-05, "loss": 1.3931, "step": 1360 }, { "epoch": 1.27, "learning_rate": 1.4680941715977722e-05, "loss": 1.3946, "step": 1370 }, { "epoch": 1.28, "learning_rate": 1.4350244163439892e-05, "loss": 1.3851, "step": 1380 }, { "epoch": 1.29, "learning_rate": 1.4021807786026108e-05, "loss": 1.3891, "step": 1390 }, { "epoch": 1.3, "learning_rate": 1.3695702317932862e-05, "loss": 1.3958, "step": 1400 }, { "epoch": 1.31, "learning_rate": 1.337199699845387e-05, "loss": 1.3909, "step": 1410 }, { "epoch": 1.32, "learning_rate": 1.3050760557279023e-05, "loss": 1.3785, "step": 1420 }, { "epoch": 1.33, "learning_rate": 1.2732061199901562e-05, "loss": 1.3882, "step": 1430 }, { "epoch": 1.34, "learning_rate": 1.2415966593136547e-05, "loss": 1.3872, "step": 1440 }, { "epoch": 1.34, "learning_rate": 1.2102543850753808e-05, "loss": 1.4032, "step": 1450 }, { "epoch": 1.35, "learning_rate": 1.1791859519228138e-05, "loss": 1.3928, "step": 1460 }, { "epoch": 1.36, "learning_rate": 1.148397956361007e-05, "loss": 1.3785, "step": 1470 }, { "epoch": 1.37, "learning_rate": 1.1178969353520018e-05, "loss": 1.3797, "step": 1480 }, { "epoch": 1.38, "learning_rate": 1.087689364926897e-05, "loss": 1.3966, "step": 1490 }, { "epoch": 1.39, "learning_rate": 1.0577816588108378e-05, "loss": 1.3849, "step": 1500 }, { "epoch": 1.4, "learning_rate": 1.0281801670612449e-05, "loss": 1.3732, "step": 1510 }, { "epoch": 1.41, "learning_rate": 9.988911747195603e-06, "loss": 1.4018, "step": 1520 }, { "epoch": 1.42, "learning_rate": 9.699209004767953e-06, "loss": 1.3902, "step": 1530 }, { "epoch": 1.43, "learning_rate": 9.412754953531663e-06, "loss": 1.3991, "step": 1540 }, { "epoch": 1.44, "learning_rate": 9.129610413921005e-06, "loss": 1.3824, "step": 1550 }, { "epoch": 1.45, "learning_rate": 8.849835503688847e-06, "loss": 1.3904, "step": 1560 }, { "epoch": 1.46, "learning_rate": 8.57348962514234e-06, "loss": 1.3937, "step": 1570 }, { "epoch": 1.46, "learning_rate": 8.30063145253053e-06, "loss": 1.39, "step": 1580 }, { "epoch": 1.47, "learning_rate": 8.031318919586523e-06, "loss": 1.3773, "step": 1590 }, { "epoch": 1.48, "learning_rate": 7.7656092072269e-06, "loss": 1.3875, "step": 1600 }, { "epoch": 1.49, "learning_rate": 7.503558731410959e-06, "loss": 1.3859, "step": 1610 }, { "epoch": 1.5, "learning_rate": 7.245223131162376e-06, "loss": 1.3716, "step": 1620 }, { "epoch": 1.51, "learning_rate": 6.9906572567558285e-06, "loss": 1.3846, "step": 1630 }, { "epoch": 1.52, "learning_rate": 6.739915158071106e-06, "loss": 1.3741, "step": 1640 }, { "epoch": 1.53, "learning_rate": 6.493050073117116e-06, "loss": 1.3827, "step": 1650 }, { "epoch": 1.54, "learning_rate": 6.250114416728298e-06, "loss": 1.3951, "step": 1660 }, { "epoch": 1.55, "learning_rate": 6.011159769435823e-06, "loss": 1.3875, "step": 1670 }, { "epoch": 1.56, "learning_rate": 5.776236866515947e-06, "loss": 1.3783, "step": 1680 }, { "epoch": 1.57, "learning_rate": 5.545395587217763e-06, "loss": 1.3888, "step": 1690 }, { "epoch": 1.58, "learning_rate": 5.318684944172752e-06, "loss": 1.4006, "step": 1700 }, { "epoch": 1.59, "learning_rate": 5.096153072988366e-06, "loss": 1.4041, "step": 1710 }, { "epoch": 1.59, "learning_rate": 4.8778472220277874e-06, "loss": 1.3865, "step": 1720 }, { "epoch": 1.6, "learning_rate": 4.6638137423780395e-06, "loss": 1.3789, "step": 1730 }, { "epoch": 1.61, "learning_rate": 4.454098078008667e-06, "loss": 1.3861, "step": 1740 }, { "epoch": 1.62, "learning_rate": 4.248744756122986e-06, "loss": 1.3803, "step": 1750 }, { "epoch": 1.63, "learning_rate": 4.047797377703985e-06, "loss": 1.3695, "step": 1760 }, { "epoch": 1.64, "learning_rate": 3.851298608256892e-06, "loss": 1.3937, "step": 1770 }, { "epoch": 1.65, "learning_rate": 3.6592901687503566e-06, "loss": 1.3865, "step": 1780 }, { "epoch": 1.66, "learning_rate": 3.471812826758178e-06, "loss": 1.3799, "step": 1790 }, { "epoch": 1.67, "learning_rate": 3.288906387803464e-06, "loss": 1.3911, "step": 1800 }, { "epoch": 1.68, "learning_rate": 3.1106096869070483e-06, "loss": 1.3801, "step": 1810 }, { "epoch": 1.69, "learning_rate": 2.9369605803419715e-06, "loss": 1.382, "step": 1820 }, { "epoch": 1.7, "learning_rate": 2.767995937595766e-06, "loss": 1.3898, "step": 1830 }, { "epoch": 1.71, "learning_rate": 2.6037516335422728e-06, "loss": 1.3752, "step": 1840 }, { "epoch": 1.72, "learning_rate": 2.4442625408246074e-06, "loss": 1.3855, "step": 1850 }, { "epoch": 1.72, "learning_rate": 2.289562522450947e-06, "loss": 1.3665, "step": 1860 }, { "epoch": 1.73, "learning_rate": 2.1396844246046903e-06, "loss": 1.3747, "step": 1870 }, { "epoch": 1.74, "learning_rate": 1.9946600696704592e-06, "loss": 1.3861, "step": 1880 }, { "epoch": 1.75, "learning_rate": 1.8545202494775509e-06, "loss": 1.3844, "step": 1890 }, { "epoch": 1.76, "learning_rate": 1.7192947187621434e-06, "loss": 1.3796, "step": 1900 }, { "epoch": 1.77, "learning_rate": 1.5890121888497366e-06, "loss": 1.3963, "step": 1910 }, { "epoch": 1.78, "learning_rate": 1.463700321559075e-06, "loss": 1.385, "step": 1920 }, { "epoch": 1.79, "learning_rate": 1.3433857233289714e-06, "loss": 1.3811, "step": 1930 }, { "epoch": 1.8, "learning_rate": 1.2280939395691859e-06, "loss": 1.384, "step": 1940 }, { "epoch": 1.81, "learning_rate": 1.1178494492365465e-06, "loss": 1.3821, "step": 1950 }, { "epoch": 1.82, "learning_rate": 1.0126756596375686e-06, "loss": 1.3978, "step": 1960 }, { "epoch": 1.83, "learning_rate": 9.125949014585383e-07, "loss": 1.3724, "step": 1970 }, { "epoch": 1.84, "learning_rate": 8.176284240242638e-07, "loss": 1.3851, "step": 1980 }, { "epoch": 1.85, "learning_rate": 7.277963907863478e-07, "loss": 1.3793, "step": 1990 }, { "epoch": 1.85, "learning_rate": 6.431178750420513e-07, "loss": 1.3946, "step": 2000 }, { "epoch": 1.86, "learning_rate": 5.636108558846359e-07, "loss": 1.3794, "step": 2010 }, { "epoch": 1.87, "learning_rate": 4.892922143859918e-07, "loss": 1.3886, "step": 2020 }, { "epoch": 1.88, "learning_rate": 4.201777300124249e-07, "loss": 1.3794, "step": 2030 }, { "epoch": 1.89, "learning_rate": 3.562820772743392e-07, "loss": 1.3828, "step": 2040 }, { "epoch": 1.9, "learning_rate": 2.9761882261050143e-07, "loss": 1.374, "step": 2050 }, { "epoch": 1.91, "learning_rate": 2.4420042150761215e-07, "loss": 1.3804, "step": 2060 }, { "epoch": 1.92, "learning_rate": 1.9603821585572058e-07, "loss": 1.3772, "step": 2070 }, { "epoch": 1.93, "learning_rate": 1.531424315400931e-07, "loss": 1.3913, "step": 2080 }, { "epoch": 1.94, "learning_rate": 1.1552217627004425e-07, "loss": 1.3859, "step": 2090 }, { "epoch": 1.95, "learning_rate": 8.318543764516961e-08, "loss": 1.4035, "step": 2100 }, { "epoch": 1.96, "learning_rate": 5.613908145939428e-08, "loss": 1.4047, "step": 2110 }, { "epoch": 1.97, "learning_rate": 3.438885024322769e-08, "loss": 1.3886, "step": 2120 }, { "epoch": 1.97, "learning_rate": 1.7939362044494534e-08, "loss": 1.3874, "step": 2130 }, { "epoch": 1.98, "learning_rate": 6.794109447824659e-09, "loss": 1.3852, "step": 2140 }, { "epoch": 1.99, "learning_rate": 9.554588330934478e-10, "loss": 1.3859, "step": 2150 }, { "epoch": 2.0, "step": 2156, "total_flos": 4.637442392914395e+18, "train_loss": 1.4111853902105498, "train_runtime": 51425.1791, "train_samples_per_second": 2.684, "train_steps_per_second": 0.042 } ], "max_steps": 2156, "num_train_epochs": 2, "total_flos": 4.637442392914395e+18, "trial_name": null, "trial_params": null }