| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.8670649738610905, | |
| "eval_steps": 500, | |
| "global_step": 2500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.98755290017426e-05, | |
| "loss": 1.7891, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.975105800348519e-05, | |
| "loss": 1.7869, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.962658700522779e-05, | |
| "loss": 1.7868, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9502116006970375e-05, | |
| "loss": 1.7547, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.937764500871298e-05, | |
| "loss": 1.6772, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9253174010455565e-05, | |
| "loss": 1.7076, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.912870301219816e-05, | |
| "loss": 1.6948, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9004232013940755e-05, | |
| "loss": 1.7135, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.887976101568335e-05, | |
| "loss": 1.7007, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.8755290017425944e-05, | |
| "loss": 1.7247, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.863081901916854e-05, | |
| "loss": 1.719, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.850634802091113e-05, | |
| "loss": 1.7209, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.838187702265373e-05, | |
| "loss": 1.7354, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.825740602439632e-05, | |
| "loss": 1.7025, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.813293502613891e-05, | |
| "loss": 1.6664, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.800846402788151e-05, | |
| "loss": 1.6851, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.78839930296241e-05, | |
| "loss": 1.6749, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7759522031366697e-05, | |
| "loss": 1.6887, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.763505103310929e-05, | |
| "loss": 1.6571, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.751058003485188e-05, | |
| "loss": 1.7136, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.738610903659448e-05, | |
| "loss": 1.7098, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.726163803833707e-05, | |
| "loss": 1.7285, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.7137167040079664e-05, | |
| "loss": 1.7038, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.701269604182226e-05, | |
| "loss": 1.692, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6888225043564854e-05, | |
| "loss": 1.6825, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.676375404530744e-05, | |
| "loss": 1.7254, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.6639283047050044e-05, | |
| "loss": 1.6765, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.651481204879263e-05, | |
| "loss": 1.6796, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6390341050535227e-05, | |
| "loss": 1.7302, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.626587005227782e-05, | |
| "loss": 1.7469, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.6141399054020416e-05, | |
| "loss": 1.6345, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.601692805576301e-05, | |
| "loss": 1.6761, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5892457057505606e-05, | |
| "loss": 1.6602, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5767986059248194e-05, | |
| "loss": 1.6637, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.5643515060990796e-05, | |
| "loss": 1.6953, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5519044062733384e-05, | |
| "loss": 1.6829, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.539457306447598e-05, | |
| "loss": 1.657, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5270102066218574e-05, | |
| "loss": 1.647, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.514563106796117e-05, | |
| "loss": 1.6854, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.502116006970376e-05, | |
| "loss": 1.6737, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.489668907144636e-05, | |
| "loss": 1.6938, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.4772218073188946e-05, | |
| "loss": 1.6721, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.464774707493155e-05, | |
| "loss": 1.6683, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.4523276076674136e-05, | |
| "loss": 1.6229, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.439880507841673e-05, | |
| "loss": 1.6593, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.4274334080159326e-05, | |
| "loss": 1.6844, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.414986308190192e-05, | |
| "loss": 1.6754, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.4025392083644516e-05, | |
| "loss": 1.6747, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.390092108538711e-05, | |
| "loss": 1.6789, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.37764500871297e-05, | |
| "loss": 1.6956, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.36519790888723e-05, | |
| "loss": 1.689, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.352750809061489e-05, | |
| "loss": 1.673, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.340303709235748e-05, | |
| "loss": 1.6968, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.327856609410008e-05, | |
| "loss": 1.6569, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.315409509584267e-05, | |
| "loss": 1.6777, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.302962409758526e-05, | |
| "loss": 1.71, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.290515309932786e-05, | |
| "loss": 1.6943, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.278068210107045e-05, | |
| "loss": 1.6925, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.2656211102813046e-05, | |
| "loss": 1.6628, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.253174010455564e-05, | |
| "loss": 1.7123, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.2407269106298235e-05, | |
| "loss": 1.6772, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.228279810804083e-05, | |
| "loss": 1.6429, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.2158327109783425e-05, | |
| "loss": 1.6472, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.203385611152601e-05, | |
| "loss": 1.6825, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.1909385113268615e-05, | |
| "loss": 1.6825, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.17849141150112e-05, | |
| "loss": 1.6818, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.16604431167538e-05, | |
| "loss": 1.7325, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.153597211849639e-05, | |
| "loss": 1.7008, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.141150112023899e-05, | |
| "loss": 1.655, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.128703012198158e-05, | |
| "loss": 1.6718, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.116255912372418e-05, | |
| "loss": 1.6412, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.1038088125466765e-05, | |
| "loss": 1.7091, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.091361712720937e-05, | |
| "loss": 1.726, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.0789146128951955e-05, | |
| "loss": 1.6655, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.066467513069455e-05, | |
| "loss": 1.7242, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.0540204132437145e-05, | |
| "loss": 1.6714, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.041573313417974e-05, | |
| "loss": 1.6646, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.029126213592233e-05, | |
| "loss": 1.6809, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.016679113766493e-05, | |
| "loss": 1.6868, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.004232013940752e-05, | |
| "loss": 1.694, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.991784914115011e-05, | |
| "loss": 1.6801, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.979337814289271e-05, | |
| "loss": 1.6861, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.96689071446353e-05, | |
| "loss": 1.6646, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.95444361463779e-05, | |
| "loss": 1.6764, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.941996514812049e-05, | |
| "loss": 1.6821, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.929549414986308e-05, | |
| "loss": 1.6681, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.917102315160568e-05, | |
| "loss": 1.6847, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.904655215334827e-05, | |
| "loss": 1.6084, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.8922081155090865e-05, | |
| "loss": 1.6624, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.879761015683346e-05, | |
| "loss": 1.6619, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.8673139158576054e-05, | |
| "loss": 1.7024, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.854866816031865e-05, | |
| "loss": 1.7301, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.8424197162061244e-05, | |
| "loss": 1.6907, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.829972616380383e-05, | |
| "loss": 1.7046, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.8175255165546434e-05, | |
| "loss": 1.6661, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.805078416728902e-05, | |
| "loss": 1.6981, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.792631316903162e-05, | |
| "loss": 1.662, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.780184217077421e-05, | |
| "loss": 1.7073, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.7677371172516806e-05, | |
| "loss": 1.6488, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.7552900174259395e-05, | |
| "loss": 1.6708, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.7428429176001996e-05, | |
| "loss": 1.7045, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.7303958177744584e-05, | |
| "loss": 1.6842, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.717948717948718e-05, | |
| "loss": 1.6682, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.7055016181229774e-05, | |
| "loss": 1.636, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.693054518297237e-05, | |
| "loss": 1.6801, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.6806074184714964e-05, | |
| "loss": 1.6371, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.668160318645756e-05, | |
| "loss": 1.6755, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.655713218820015e-05, | |
| "loss": 1.7046, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.643266118994275e-05, | |
| "loss": 1.707, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.6308190191685337e-05, | |
| "loss": 1.6568, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.618371919342793e-05, | |
| "loss": 1.6447, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.6059248195170526e-05, | |
| "loss": 1.691, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.593477719691312e-05, | |
| "loss": 1.6747, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.5810306198655716e-05, | |
| "loss": 1.656, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.568583520039831e-05, | |
| "loss": 1.6485, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.55613642021409e-05, | |
| "loss": 1.642, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.54368932038835e-05, | |
| "loss": 1.6891, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.531242220562609e-05, | |
| "loss": 1.716, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.5187951207368684e-05, | |
| "loss": 1.6785, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.506348020911128e-05, | |
| "loss": 1.7053, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.493900921085387e-05, | |
| "loss": 1.6833, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.481453821259646e-05, | |
| "loss": 1.7336, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.469006721433906e-05, | |
| "loss": 1.6496, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.456559621608165e-05, | |
| "loss": 1.7014, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.4441125217824246e-05, | |
| "loss": 1.6502, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.431665421956684e-05, | |
| "loss": 1.6617, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.4192183221309436e-05, | |
| "loss": 1.7155, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.406771222305203e-05, | |
| "loss": 1.691, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.3943241224794625e-05, | |
| "loss": 1.6601, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.3818770226537214e-05, | |
| "loss": 1.6607, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.3694299228279815e-05, | |
| "loss": 1.7053, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.35698282300224e-05, | |
| "loss": 1.6681, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.3445357231765e-05, | |
| "loss": 1.6712, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.332088623350759e-05, | |
| "loss": 1.6601, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.319641523525019e-05, | |
| "loss": 1.6522, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.307194423699278e-05, | |
| "loss": 1.6621, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.294747323873538e-05, | |
| "loss": 1.6651, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.2823002240477966e-05, | |
| "loss": 1.6665, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.269853124222057e-05, | |
| "loss": 1.6811, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.2574060243963156e-05, | |
| "loss": 1.6632, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.244958924570575e-05, | |
| "loss": 1.6529, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.2325118247448345e-05, | |
| "loss": 1.6943, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.220064724919094e-05, | |
| "loss": 1.6932, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.2076176250933535e-05, | |
| "loss": 1.6711, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.195170525267613e-05, | |
| "loss": 1.6711, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.182723425441872e-05, | |
| "loss": 1.6678, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.170276325616132e-05, | |
| "loss": 1.7058, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.157829225790391e-05, | |
| "loss": 1.7017, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.14538212596465e-05, | |
| "loss": 1.6862, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.13293502613891e-05, | |
| "loss": 1.6425, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.120487926313169e-05, | |
| "loss": 1.6977, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.108040826487428e-05, | |
| "loss": 1.698, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.095593726661688e-05, | |
| "loss": 1.6718, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.083146626835947e-05, | |
| "loss": 1.7025, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.0706995270102065e-05, | |
| "loss": 1.665, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.058252427184466e-05, | |
| "loss": 1.7035, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.0458053273587255e-05, | |
| "loss": 1.6849, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.0333582275329846e-05, | |
| "loss": 1.6345, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.0209111277072444e-05, | |
| "loss": 1.6593, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.0084640278815036e-05, | |
| "loss": 1.6248, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 2.996016928055763e-05, | |
| "loss": 1.6827, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.9835698282300222e-05, | |
| "loss": 1.7054, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.971122728404282e-05, | |
| "loss": 1.6518, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.9586756285785412e-05, | |
| "loss": 1.6795, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.9462285287528007e-05, | |
| "loss": 1.6286, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.93378142892706e-05, | |
| "loss": 1.6599, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.9213343291013197e-05, | |
| "loss": 1.6902, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.9088872292755788e-05, | |
| "loss": 1.6551, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.8964401294498383e-05, | |
| "loss": 1.6493, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.8839930296240975e-05, | |
| "loss": 1.7094, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.8715459297983573e-05, | |
| "loss": 1.6714, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.859098829972616e-05, | |
| "loss": 1.6296, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.846651730146876e-05, | |
| "loss": 1.6882, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.834204630321135e-05, | |
| "loss": 1.6493, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.8217575304953945e-05, | |
| "loss": 1.6489, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.8093104306696544e-05, | |
| "loss": 1.6683, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.7968633308439135e-05, | |
| "loss": 1.6988, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.784416231018173e-05, | |
| "loss": 1.648, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.771969131192432e-05, | |
| "loss": 1.6655, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.759522031366692e-05, | |
| "loss": 1.6709, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.747074931540951e-05, | |
| "loss": 1.665, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.7346278317152106e-05, | |
| "loss": 1.6411, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.7221807318894698e-05, | |
| "loss": 1.6913, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.7097336320637296e-05, | |
| "loss": 1.6906, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.6972865322379887e-05, | |
| "loss": 1.6345, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.6848394324122482e-05, | |
| "loss": 1.6903, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.6723923325865074e-05, | |
| "loss": 1.6819, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.6599452327607672e-05, | |
| "loss": 1.6393, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.6474981329350264e-05, | |
| "loss": 1.6725, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.635051033109286e-05, | |
| "loss": 1.6736, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.622603933283545e-05, | |
| "loss": 1.6734, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.6101568334578048e-05, | |
| "loss": 1.6761, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.597709733632064e-05, | |
| "loss": 1.6842, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.5852626338063234e-05, | |
| "loss": 1.7064, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.5728155339805826e-05, | |
| "loss": 1.7036, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.5603684341548424e-05, | |
| "loss": 1.6563, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.5479213343291016e-05, | |
| "loss": 1.6796, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.535474234503361e-05, | |
| "loss": 1.6469, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.5230271346776202e-05, | |
| "loss": 1.6681, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.51058003485188e-05, | |
| "loss": 1.6632, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.498132935026139e-05, | |
| "loss": 1.6799, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.4856858352003983e-05, | |
| "loss": 1.6811, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.4732387353746578e-05, | |
| "loss": 1.6365, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.4607916355489173e-05, | |
| "loss": 1.6773, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.4483445357231764e-05, | |
| "loss": 1.6194, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.435897435897436e-05, | |
| "loss": 1.6865, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.4234503360716954e-05, | |
| "loss": 1.6552, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.411003236245955e-05, | |
| "loss": 1.6823, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.398556136420214e-05, | |
| "loss": 1.6611, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.3861090365944735e-05, | |
| "loss": 1.6662, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.373661936768733e-05, | |
| "loss": 1.6898, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.3612148369429922e-05, | |
| "loss": 1.6474, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.3487677371172517e-05, | |
| "loss": 1.6635, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.336320637291511e-05, | |
| "loss": 1.6274, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.3238735374657706e-05, | |
| "loss": 1.6794, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.3114264376400298e-05, | |
| "loss": 1.6411, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.2989793378142893e-05, | |
| "loss": 1.6793, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.2865322379885488e-05, | |
| "loss": 1.6772, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.2740851381628083e-05, | |
| "loss": 1.6425, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.2616380383370674e-05, | |
| "loss": 1.6614, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.249190938511327e-05, | |
| "loss": 1.65, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.2367438386855864e-05, | |
| "loss": 1.685, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.2242967388598455e-05, | |
| "loss": 1.6808, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.211849639034105e-05, | |
| "loss": 1.6612, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.1994025392083645e-05, | |
| "loss": 1.6631, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.186955439382624e-05, | |
| "loss": 1.6675, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.174508339556883e-05, | |
| "loss": 1.6532, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.1620612397311426e-05, | |
| "loss": 1.7217, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.149614139905402e-05, | |
| "loss": 1.6697, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.1371670400796616e-05, | |
| "loss": 1.634, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.1247199402539207e-05, | |
| "loss": 1.6968, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.1122728404281802e-05, | |
| "loss": 1.6925, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.0998257406024397e-05, | |
| "loss": 1.6455, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.0873786407766992e-05, | |
| "loss": 1.6365, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.0749315409509583e-05, | |
| "loss": 1.6813, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.062484441125218e-05, | |
| "loss": 1.6725, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.0500373412994773e-05, | |
| "loss": 1.6718, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.0375902414737365e-05, | |
| "loss": 1.6954, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.025143141647996e-05, | |
| "loss": 1.6719, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.0126960418222554e-05, | |
| "loss": 1.6549, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 2.000248941996515e-05, | |
| "loss": 1.6761, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.987801842170774e-05, | |
| "loss": 1.6582, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.9753547423450336e-05, | |
| "loss": 1.6202, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.962907642519293e-05, | |
| "loss": 1.6645, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.9504605426935525e-05, | |
| "loss": 1.6565, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.9380134428678117e-05, | |
| "loss": 1.6395, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.9255663430420712e-05, | |
| "loss": 1.6469, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.9131192432163307e-05, | |
| "loss": 1.6523, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.9006721433905898e-05, | |
| "loss": 1.6799, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.8882250435648493e-05, | |
| "loss": 1.6584, | |
| "step": 2500 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 4017, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 1.5800620548096e+18, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |