| { | |
| "best_metric": 3.1339056491851807, | |
| "best_model_checkpoint": "output/drake/checkpoint-660", | |
| "epoch": 2.0, | |
| "global_step": 660, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001371432064088578, | |
| "loss": 4.0588, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00013697291967368176, | |
| "loss": 3.874, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000136689421753516, | |
| "loss": 3.7354, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001362931820613334, | |
| "loss": 3.678, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00013578485668678617, | |
| "loss": 3.4762, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00013516528730985944, | |
| "loss": 3.9075, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000134435499807226, | |
| "loss": 3.8826, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00013359670255361022, | |
| "loss": 3.6832, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00013265028442097337, | |
| "loss": 3.5663, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00013159781247883373, | |
| "loss": 3.875, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00013044102939952857, | |
| "loss": 3.6027, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00012918185057271526, | |
| "loss": 3.7132, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0001278223609338884, | |
| "loss": 3.5338, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00012636481151216485, | |
| "loss": 3.8949, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0001248116157030527, | |
| "loss": 3.5964, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0001231653452723755, | |
| "loss": 3.7202, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00012142872609796847, | |
| "loss": 3.6172, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00011960463365619783, | |
| "loss": 3.4145, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00011769608826077622, | |
| "loss": 3.555, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.000115706250061758, | |
| "loss": 3.4485, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00011363841381299484, | |
| "loss": 3.7157, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.000111496003416716, | |
| "loss": 3.6139, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0001092825662542656, | |
| "loss": 3.6335, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00010700176731238488, | |
| "loss": 3.8618, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00010465738311476428, | |
| "loss": 3.5889, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00010225329546891398, | |
| "loss": 3.5039, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.979348503870657e-05, | |
| "loss": 3.456, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.728202475323418e-05, | |
| "loss": 3.5593, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.472307306289397e-05, | |
| "loss": 3.6619, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.21208670538683e-05, | |
| "loss": 3.2545, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.94797154324004e-05, | |
| "loss": 3.5664, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.680399139048255e-05, | |
| "loss": 3.6659, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.409812536476914e-05, | |
| "loss": 3.4141, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.136659770070481e-05, | |
| "loss": 3.3853, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 7.86139312340139e-05, | |
| "loss": 3.5356, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 7.584468380183525e-05, | |
| "loss": 3.6393, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 7.306344069590158e-05, | |
| "loss": 3.7593, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 7.027480707026031e-05, | |
| "loss": 3.6784, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 6.748340031610621e-05, | |
| "loss": 3.4365, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 6.469384241635202e-05, | |
| "loss": 3.4692, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 6.191075229259615e-05, | |
| "loss": 3.5463, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5.9138738157159314e-05, | |
| "loss": 3.4183, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5.638238988285346e-05, | |
| "loss": 3.6126, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5.364627140311692e-05, | |
| "loss": 3.5828, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5.093491315510001e-05, | |
| "loss": 3.4454, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.825280457821314e-05, | |
| "loss": 3.5058, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.560438668055879e-05, | |
| "loss": 3.4436, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.299404468555545e-05, | |
| "loss": 3.3312, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.042610077092918e-05, | |
| "loss": 3.4519, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.7904806912095865e-05, | |
| "loss": 3.2724, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.543433784178337e-05, | |
| "loss": 3.708, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.30187841375518e-05, | |
| "loss": 3.3732, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.06621454486565e-05, | |
| "loss": 3.6682, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.8368323873469696e-05, | |
| "loss": 3.3303, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.614111749842566e-05, | |
| "loss": 3.4287, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.39842141091877e-05, | |
| "loss": 3.4274, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.1901185084450396e-05, | |
| "loss": 3.3246, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.9895479482486794e-05, | |
| "loss": 3.4282, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.7970418330233093e-05, | |
| "loss": 3.2375, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.61291891243658e-05, | |
| "loss": 3.581, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.4374840553477238e-05, | |
| "loss": 3.42, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.2710277450088041e-05, | |
| "loss": 3.4236, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.1138255980854819e-05, | |
| "loss": 3.3657, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.661379082937688e-06, | |
| "loss": 3.1107, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 8.282092154083231e-06, | |
| "loss": 3.3076, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.002679003559785e-06, | |
| "loss": 3.3086, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 5.825258070649407e-06, | |
| "loss": 3.3643, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.7517789169574735e-06, | |
| "loss": 3.348, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.7840189983485763e-06, | |
| "loss": 3.391, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.9235807218530457e-06, | |
| "loss": 3.2735, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.171888792417796e-06, | |
| "loss": 3.576, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.5301878538942795e-06, | |
| "loss": 3.2583, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 9.995404281699138e-07, | |
| "loss": 3.5956, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 5.808251558551173e-07, | |
| "loss": 3.7098, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.7473534143910847e-07, | |
| "loss": 3.3973, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.177780532344056e-08, | |
| "loss": 3.3842, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.2720446338799106e-09, | |
| "loss": 3.2786, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.0890007647780969e-05, | |
| "loss": 3.2655, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.2720089689346961e-05, | |
| "loss": 3.1689, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.4676757700644785e-05, | |
| "loss": 3.15, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.6755579199297876e-05, | |
| "loss": 3.3109, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.8951844985992176e-05, | |
| "loss": 3.3557, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.1260579812327114e-05, | |
| "loss": 3.2466, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.3676553651353433e-05, | |
| "loss": 3.4927, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.6194293545266464e-05, | |
| "loss": 3.5059, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.8808096003415798e-05, | |
| "loss": 3.3394, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 3.151203992254596e-05, | |
| "loss": 2.9776, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 3.429999999999997e-05, | |
| "loss": 3.1045, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.716566060949963e-05, | |
| "loss": 3.3412, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.0102530108070535e-05, | |
| "loss": 3.3821, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.3103955541701554e-05, | |
| "loss": 3.2771, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.6163137716424864e-05, | |
| "loss": 3.2889, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.927314660067792e-05, | |
| "loss": 3.1803, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.242693702405331e-05, | |
| "loss": 3.3752, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.561736463687583e-05, | |
| "loss": 3.4172, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.883720209445263e-05, | |
| "loss": 3.1247, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 6.207915542933309e-05, | |
| "loss": 3.1979, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 6.533588057449125e-05, | |
| "loss": 3.2403, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 6.859999999999999e-05, | |
| "loss": 3.3824, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 7.186411942550872e-05, | |
| "loss": 3.2565, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 7.512084457066689e-05, | |
| "loss": 3.1016, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 7.836279790554734e-05, | |
| "loss": 3.4194, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 8.158263536312414e-05, | |
| "loss": 3.1877, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 8.477306297594667e-05, | |
| "loss": 3.4791, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 8.792685339932205e-05, | |
| "loss": 3.2052, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.103686228357512e-05, | |
| "loss": 3.4702, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.409604445829843e-05, | |
| "loss": 3.152, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 9.709746989192944e-05, | |
| "loss": 3.2925, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.00010003433939050033, | |
| "loss": 3.2007, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.00010290000000000001, | |
| "loss": 3.444, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.00010568796007745401, | |
| "loss": 3.227, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.00010839190399658417, | |
| "loss": 3.133, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 0.00011100570645473351, | |
| "loss": 3.1161, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.00011352344634864656, | |
| "loss": 3.2043, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.00011593942018767285, | |
| "loss": 3.265, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.00011824815501400781, | |
| "loss": 3.3016, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 0.00012044442080070208, | |
| "loss": 3.3944, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.0001225232422993552, | |
| "loss": 3.4325, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.00012447991031065301, | |
| "loss": 3.3161, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.000126309992352219, | |
| "loss": 3.2818, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 0.00012800934269961248, | |
| "loss": 3.2362, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 0.00012957411177772773, | |
| "loss": 3.1107, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 0.00013100075488131993, | |
| "loss": 3.2821, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.00013228604020490257, | |
| "loss": 3.3709, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 0.00013342705616382626, | |
| "loss": 3.3293, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.00013442121798995453, | |
| "loss": 3.5064, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 0.00013526627358699495, | |
| "loss": 3.2024, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 0.00013596030863222166, | |
| "loss": 3.378, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.000136501750913032, | |
| "loss": 3.215, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.0001368893738885136, | |
| "loss": 3.2645, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.00013712229946795436, | |
| "loss": 3.2699, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.0001372, | |
| "loss": 2.9558, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 3.1339056491851807, | |
| "eval_runtime": 9.8661, | |
| "eval_samples_per_second": 45.611, | |
| "eval_steps_per_second": 5.777, | |
| "step": 660 | |
| } | |
| ], | |
| "max_steps": 660, | |
| "num_train_epochs": 2, | |
| "total_flos": 688765796352000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |