| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.11470520761642579, |
| "eval_steps": 5, |
| "global_step": 500, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0011470520761642578, |
| "eval_loss": 2.755366325378418, |
| "eval_runtime": 14.0088, |
| "eval_samples_per_second": 469.989, |
| "eval_steps_per_second": 3.712, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.0022941041523285156, |
| "eval_loss": 2.750617027282715, |
| "eval_runtime": 14.0654, |
| "eval_samples_per_second": 468.099, |
| "eval_steps_per_second": 3.697, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.0034411562284927736, |
| "eval_loss": 2.7424304485321045, |
| "eval_runtime": 14.1186, |
| "eval_samples_per_second": 466.336, |
| "eval_steps_per_second": 3.683, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.004588208304657031, |
| "eval_loss": 2.730889081954956, |
| "eval_runtime": 14.2451, |
| "eval_samples_per_second": 462.193, |
| "eval_steps_per_second": 3.65, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.005735260380821289, |
| "eval_loss": 2.7159576416015625, |
| "eval_runtime": 14.245, |
| "eval_samples_per_second": 462.199, |
| "eval_steps_per_second": 3.65, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.006882312456985547, |
| "eval_loss": 2.6975343227386475, |
| "eval_runtime": 14.3448, |
| "eval_samples_per_second": 458.981, |
| "eval_steps_per_second": 3.625, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.008029364533149804, |
| "eval_loss": 2.675693988800049, |
| "eval_runtime": 16.965, |
| "eval_samples_per_second": 388.094, |
| "eval_steps_per_second": 3.065, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.009176416609314062, |
| "eval_loss": 2.650242328643799, |
| "eval_runtime": 14.2414, |
| "eval_samples_per_second": 462.315, |
| "eval_steps_per_second": 3.651, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.01032346868547832, |
| "eval_loss": 2.6213560104370117, |
| "eval_runtime": 14.4168, |
| "eval_samples_per_second": 456.689, |
| "eval_steps_per_second": 3.607, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.011470520761642579, |
| "eval_loss": 2.5892951488494873, |
| "eval_runtime": 14.3332, |
| "eval_samples_per_second": 459.353, |
| "eval_steps_per_second": 3.628, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.012617572837806837, |
| "eval_loss": 2.5538384914398193, |
| "eval_runtime": 14.4067, |
| "eval_samples_per_second": 457.01, |
| "eval_steps_per_second": 3.609, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.013764624913971095, |
| "eval_loss": 2.51452374458313, |
| "eval_runtime": 14.4948, |
| "eval_samples_per_second": 454.233, |
| "eval_steps_per_second": 3.588, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.014911676990135353, |
| "eval_loss": 2.4726295471191406, |
| "eval_runtime": 14.3162, |
| "eval_samples_per_second": 459.898, |
| "eval_steps_per_second": 3.632, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.01605872906629961, |
| "eval_loss": 2.428196668624878, |
| "eval_runtime": 14.3204, |
| "eval_samples_per_second": 459.763, |
| "eval_steps_per_second": 3.631, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.017205781142463867, |
| "eval_loss": 2.379453659057617, |
| "eval_runtime": 14.3796, |
| "eval_samples_per_second": 457.872, |
| "eval_steps_per_second": 3.616, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.018352833218628125, |
| "eval_loss": 2.327221393585205, |
| "eval_runtime": 14.3415, |
| "eval_samples_per_second": 459.089, |
| "eval_steps_per_second": 3.626, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.019499885294792383, |
| "eval_loss": 2.2712388038635254, |
| "eval_runtime": 14.462, |
| "eval_samples_per_second": 455.263, |
| "eval_steps_per_second": 3.596, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.02064693737095664, |
| "eval_loss": 2.2120068073272705, |
| "eval_runtime": 14.334, |
| "eval_samples_per_second": 459.326, |
| "eval_steps_per_second": 3.628, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.0217939894471209, |
| "eval_loss": 2.1501331329345703, |
| "eval_runtime": 14.3506, |
| "eval_samples_per_second": 458.796, |
| "eval_steps_per_second": 3.624, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.022941041523285157, |
| "grad_norm": 19.895355224609375, |
| "learning_rate": 7.645259938837921e-07, |
| "loss": 3.6197, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.022941041523285157, |
| "eval_loss": 2.086596727371216, |
| "eval_runtime": 16.6465, |
| "eval_samples_per_second": 395.519, |
| "eval_steps_per_second": 3.124, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.024088093599449415, |
| "eval_loss": 2.0223236083984375, |
| "eval_runtime": 14.471, |
| "eval_samples_per_second": 454.979, |
| "eval_steps_per_second": 3.593, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.025235145675613673, |
| "eval_loss": 1.9571231603622437, |
| "eval_runtime": 14.3783, |
| "eval_samples_per_second": 457.913, |
| "eval_steps_per_second": 3.617, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.02638219775177793, |
| "eval_loss": 1.890655517578125, |
| "eval_runtime": 14.288, |
| "eval_samples_per_second": 460.805, |
| "eval_steps_per_second": 3.639, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.02752924982794219, |
| "eval_loss": 1.823920488357544, |
| "eval_runtime": 14.3764, |
| "eval_samples_per_second": 457.972, |
| "eval_steps_per_second": 3.617, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.028676301904106447, |
| "eval_loss": 1.758331298828125, |
| "eval_runtime": 14.2666, |
| "eval_samples_per_second": 461.499, |
| "eval_steps_per_second": 3.645, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.029823353980270705, |
| "eval_loss": 1.6937522888183594, |
| "eval_runtime": 14.2522, |
| "eval_samples_per_second": 461.963, |
| "eval_steps_per_second": 3.649, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.030970406056434963, |
| "eval_loss": 1.6316019296646118, |
| "eval_runtime": 16.3104, |
| "eval_samples_per_second": 403.67, |
| "eval_steps_per_second": 3.188, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.03211745813259922, |
| "eval_loss": 1.571895718574524, |
| "eval_runtime": 14.2422, |
| "eval_samples_per_second": 462.288, |
| "eval_steps_per_second": 3.651, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.033264510208763476, |
| "eval_loss": 1.5148202180862427, |
| "eval_runtime": 14.3624, |
| "eval_samples_per_second": 458.418, |
| "eval_steps_per_second": 3.621, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.034411562284927734, |
| "eval_loss": 1.459762454032898, |
| "eval_runtime": 14.3035, |
| "eval_samples_per_second": 460.308, |
| "eval_steps_per_second": 3.635, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.03555861436109199, |
| "eval_loss": 1.4081143140792847, |
| "eval_runtime": 14.2988, |
| "eval_samples_per_second": 460.46, |
| "eval_steps_per_second": 3.637, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.03670566643725625, |
| "eval_loss": 1.3612124919891357, |
| "eval_runtime": 14.2569, |
| "eval_samples_per_second": 461.812, |
| "eval_steps_per_second": 3.647, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.03785271851342051, |
| "eval_loss": 1.318212866783142, |
| "eval_runtime": 14.282, |
| "eval_samples_per_second": 460.999, |
| "eval_steps_per_second": 3.641, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.038999770589584766, |
| "eval_loss": 1.2802687883377075, |
| "eval_runtime": 14.4058, |
| "eval_samples_per_second": 457.038, |
| "eval_steps_per_second": 3.61, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.040146822665749024, |
| "eval_loss": 1.246294617652893, |
| "eval_runtime": 14.2804, |
| "eval_samples_per_second": 461.051, |
| "eval_steps_per_second": 3.641, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.04129387474191328, |
| "eval_loss": 1.2160167694091797, |
| "eval_runtime": 14.3449, |
| "eval_samples_per_second": 458.977, |
| "eval_steps_per_second": 3.625, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.04244092681807754, |
| "eval_loss": 1.189509630203247, |
| "eval_runtime": 14.2361, |
| "eval_samples_per_second": 462.486, |
| "eval_steps_per_second": 3.653, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.0435879788942418, |
| "eval_loss": 1.1653709411621094, |
| "eval_runtime": 14.3748, |
| "eval_samples_per_second": 458.025, |
| "eval_steps_per_second": 3.617, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.044735030970406056, |
| "eval_loss": 1.143513798713684, |
| "eval_runtime": 14.25, |
| "eval_samples_per_second": 462.035, |
| "eval_steps_per_second": 3.649, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.045882083046570314, |
| "grad_norm": 8.192963600158691, |
| "learning_rate": 1.5290519877675841e-06, |
| "loss": 2.292, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.045882083046570314, |
| "eval_loss": 1.1239805221557617, |
| "eval_runtime": 16.3012, |
| "eval_samples_per_second": 403.896, |
| "eval_steps_per_second": 3.19, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.04702913512273457, |
| "eval_loss": 1.1064891815185547, |
| "eval_runtime": 14.2673, |
| "eval_samples_per_second": 461.476, |
| "eval_steps_per_second": 3.645, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.04817618719889883, |
| "eval_loss": 1.0907284021377563, |
| "eval_runtime": 14.2601, |
| "eval_samples_per_second": 461.707, |
| "eval_steps_per_second": 3.647, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.04932323927506309, |
| "eval_loss": 1.076059341430664, |
| "eval_runtime": 14.2845, |
| "eval_samples_per_second": 460.919, |
| "eval_steps_per_second": 3.64, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.050470291351227346, |
| "eval_loss": 1.062280535697937, |
| "eval_runtime": 14.217, |
| "eval_samples_per_second": 463.109, |
| "eval_steps_per_second": 3.658, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.051617343427391604, |
| "eval_loss": 1.0492550134658813, |
| "eval_runtime": 14.2233, |
| "eval_samples_per_second": 462.901, |
| "eval_steps_per_second": 3.656, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.05276439550355586, |
| "eval_loss": 1.0374095439910889, |
| "eval_runtime": 14.5693, |
| "eval_samples_per_second": 451.909, |
| "eval_steps_per_second": 3.569, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.05391144757972012, |
| "eval_loss": 1.0259910821914673, |
| "eval_runtime": 15.9876, |
| "eval_samples_per_second": 411.819, |
| "eval_steps_per_second": 3.253, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.05505849965588438, |
| "eval_loss": 1.0147359371185303, |
| "eval_runtime": 14.2101, |
| "eval_samples_per_second": 463.332, |
| "eval_steps_per_second": 3.659, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.056205551732048636, |
| "eval_loss": 1.0042893886566162, |
| "eval_runtime": 14.2547, |
| "eval_samples_per_second": 461.882, |
| "eval_steps_per_second": 3.648, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.057352603808212894, |
| "eval_loss": 0.9941452741622925, |
| "eval_runtime": 14.2125, |
| "eval_samples_per_second": 463.253, |
| "eval_steps_per_second": 3.659, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.05849965588437715, |
| "eval_loss": 0.9848644733428955, |
| "eval_runtime": 14.2628, |
| "eval_samples_per_second": 461.621, |
| "eval_steps_per_second": 3.646, |
| "step": 255 |
| }, |
| { |
| "epoch": 0.05964670796054141, |
| "eval_loss": 0.9763049483299255, |
| "eval_runtime": 14.2014, |
| "eval_samples_per_second": 463.617, |
| "eval_steps_per_second": 3.662, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.06079376003670567, |
| "eval_loss": 0.9682185649871826, |
| "eval_runtime": 16.4598, |
| "eval_samples_per_second": 400.004, |
| "eval_steps_per_second": 3.159, |
| "step": 265 |
| }, |
| { |
| "epoch": 0.06194081211286993, |
| "eval_loss": 0.9602033495903015, |
| "eval_runtime": 14.3432, |
| "eval_samples_per_second": 459.031, |
| "eval_steps_per_second": 3.625, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.06308786418903418, |
| "eval_loss": 0.952538251876831, |
| "eval_runtime": 14.2143, |
| "eval_samples_per_second": 463.195, |
| "eval_steps_per_second": 3.658, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.06423491626519844, |
| "eval_loss": 0.9450673460960388, |
| "eval_runtime": 14.2195, |
| "eval_samples_per_second": 463.025, |
| "eval_steps_per_second": 3.657, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.0653819683413627, |
| "eval_loss": 0.937529981136322, |
| "eval_runtime": 14.2726, |
| "eval_samples_per_second": 461.303, |
| "eval_steps_per_second": 3.643, |
| "step": 285 |
| }, |
| { |
| "epoch": 0.06652902041752695, |
| "eval_loss": 0.930277407169342, |
| "eval_runtime": 14.5409, |
| "eval_samples_per_second": 452.791, |
| "eval_steps_per_second": 3.576, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.06767607249369122, |
| "eval_loss": 0.9230740666389465, |
| "eval_runtime": 14.2552, |
| "eval_samples_per_second": 461.868, |
| "eval_steps_per_second": 3.648, |
| "step": 295 |
| }, |
| { |
| "epoch": 0.06882312456985547, |
| "grad_norm": 7.243612766265869, |
| "learning_rate": 2.2935779816513764e-06, |
| "loss": 1.5711, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.06882312456985547, |
| "eval_loss": 0.915981650352478, |
| "eval_runtime": 16.3997, |
| "eval_samples_per_second": 401.471, |
| "eval_steps_per_second": 3.171, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.06997017664601973, |
| "eval_loss": 0.9087598323822021, |
| "eval_runtime": 14.1993, |
| "eval_samples_per_second": 463.685, |
| "eval_steps_per_second": 3.662, |
| "step": 305 |
| }, |
| { |
| "epoch": 0.07111722872218398, |
| "eval_loss": 0.9022247791290283, |
| "eval_runtime": 14.3064, |
| "eval_samples_per_second": 460.214, |
| "eval_steps_per_second": 3.635, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.07226428079834825, |
| "eval_loss": 0.8950537443161011, |
| "eval_runtime": 14.2087, |
| "eval_samples_per_second": 463.377, |
| "eval_steps_per_second": 3.66, |
| "step": 315 |
| }, |
| { |
| "epoch": 0.0734113328745125, |
| "eval_loss": 0.8874984383583069, |
| "eval_runtime": 14.2295, |
| "eval_samples_per_second": 462.699, |
| "eval_steps_per_second": 3.654, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.07455838495067676, |
| "eval_loss": 0.8809635043144226, |
| "eval_runtime": 14.2266, |
| "eval_samples_per_second": 462.796, |
| "eval_steps_per_second": 3.655, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.07570543702684102, |
| "eval_loss": 0.8745627403259277, |
| "eval_runtime": 14.2942, |
| "eval_samples_per_second": 460.608, |
| "eval_steps_per_second": 3.638, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.07685248910300528, |
| "eval_loss": 0.8683872818946838, |
| "eval_runtime": 14.1413, |
| "eval_samples_per_second": 465.586, |
| "eval_steps_per_second": 3.677, |
| "step": 335 |
| }, |
| { |
| "epoch": 0.07799954117916953, |
| "eval_loss": 0.8624699115753174, |
| "eval_runtime": 14.1694, |
| "eval_samples_per_second": 464.662, |
| "eval_steps_per_second": 3.67, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.0791465932553338, |
| "eval_loss": 0.8569262027740479, |
| "eval_runtime": 14.2132, |
| "eval_samples_per_second": 463.231, |
| "eval_steps_per_second": 3.659, |
| "step": 345 |
| }, |
| { |
| "epoch": 0.08029364533149805, |
| "eval_loss": 0.8515614867210388, |
| "eval_runtime": 14.3011, |
| "eval_samples_per_second": 460.383, |
| "eval_steps_per_second": 3.636, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.08144069740766231, |
| "eval_loss": 0.8465690612792969, |
| "eval_runtime": 14.2483, |
| "eval_samples_per_second": 462.09, |
| "eval_steps_per_second": 3.65, |
| "step": 355 |
| }, |
| { |
| "epoch": 0.08258774948382656, |
| "eval_loss": 0.8418980836868286, |
| "eval_runtime": 14.1978, |
| "eval_samples_per_second": 463.735, |
| "eval_steps_per_second": 3.663, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.08373480155999083, |
| "eval_loss": 0.836972177028656, |
| "eval_runtime": 16.4774, |
| "eval_samples_per_second": 399.577, |
| "eval_steps_per_second": 3.156, |
| "step": 365 |
| }, |
| { |
| "epoch": 0.08488185363615508, |
| "eval_loss": 0.8320812582969666, |
| "eval_runtime": 14.3098, |
| "eval_samples_per_second": 460.103, |
| "eval_steps_per_second": 3.634, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.08602890571231935, |
| "eval_loss": 0.8273819088935852, |
| "eval_runtime": 14.2755, |
| "eval_samples_per_second": 461.209, |
| "eval_steps_per_second": 3.643, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.0871759577884836, |
| "eval_loss": 0.8223117589950562, |
| "eval_runtime": 14.1567, |
| "eval_samples_per_second": 465.081, |
| "eval_steps_per_second": 3.673, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.08832300986464786, |
| "eval_loss": 0.8169983625411987, |
| "eval_runtime": 14.1912, |
| "eval_samples_per_second": 463.948, |
| "eval_steps_per_second": 3.664, |
| "step": 385 |
| }, |
| { |
| "epoch": 0.08947006194081211, |
| "eval_loss": 0.811504602432251, |
| "eval_runtime": 14.1791, |
| "eval_samples_per_second": 464.347, |
| "eval_steps_per_second": 3.667, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.09061711401697638, |
| "eval_loss": 0.8054670095443726, |
| "eval_runtime": 14.2016, |
| "eval_samples_per_second": 463.611, |
| "eval_steps_per_second": 3.662, |
| "step": 395 |
| }, |
| { |
| "epoch": 0.09176416609314063, |
| "grad_norm": 5.852241516113281, |
| "learning_rate": 3.0581039755351682e-06, |
| "loss": 1.3859, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.09176416609314063, |
| "eval_loss": 0.8003625273704529, |
| "eval_runtime": 16.3667, |
| "eval_samples_per_second": 402.281, |
| "eval_steps_per_second": 3.177, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.0929112181693049, |
| "eval_loss": 0.7954948544502258, |
| "eval_runtime": 14.1443, |
| "eval_samples_per_second": 465.489, |
| "eval_steps_per_second": 3.676, |
| "step": 405 |
| }, |
| { |
| "epoch": 0.09405827024546914, |
| "eval_loss": 0.7906287312507629, |
| "eval_runtime": 14.2485, |
| "eval_samples_per_second": 462.082, |
| "eval_steps_per_second": 3.649, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.0952053223216334, |
| "eval_loss": 0.7856321930885315, |
| "eval_runtime": 14.2745, |
| "eval_samples_per_second": 461.242, |
| "eval_steps_per_second": 3.643, |
| "step": 415 |
| }, |
| { |
| "epoch": 0.09635237439779766, |
| "eval_loss": 0.7809005379676819, |
| "eval_runtime": 14.2039, |
| "eval_samples_per_second": 463.536, |
| "eval_steps_per_second": 3.661, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.09749942647396191, |
| "eval_loss": 0.7759343981742859, |
| "eval_runtime": 14.2261, |
| "eval_samples_per_second": 462.813, |
| "eval_steps_per_second": 3.655, |
| "step": 425 |
| }, |
| { |
| "epoch": 0.09864647855012618, |
| "eval_loss": 0.7706864476203918, |
| "eval_runtime": 16.5161, |
| "eval_samples_per_second": 398.642, |
| "eval_steps_per_second": 3.148, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.09979353062629043, |
| "eval_loss": 0.765408992767334, |
| "eval_runtime": 14.193, |
| "eval_samples_per_second": 463.89, |
| "eval_steps_per_second": 3.664, |
| "step": 435 |
| }, |
| { |
| "epoch": 0.10094058270245469, |
| "eval_loss": 0.7602970600128174, |
| "eval_runtime": 14.2387, |
| "eval_samples_per_second": 462.402, |
| "eval_steps_per_second": 3.652, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.10208763477861894, |
| "eval_loss": 0.7555930018424988, |
| "eval_runtime": 14.1713, |
| "eval_samples_per_second": 464.601, |
| "eval_steps_per_second": 3.669, |
| "step": 445 |
| }, |
| { |
| "epoch": 0.10323468685478321, |
| "eval_loss": 0.7515881061553955, |
| "eval_runtime": 14.2211, |
| "eval_samples_per_second": 462.973, |
| "eval_steps_per_second": 3.657, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.10438173893094746, |
| "eval_loss": 0.7474184036254883, |
| "eval_runtime": 14.1814, |
| "eval_samples_per_second": 464.27, |
| "eval_steps_per_second": 3.667, |
| "step": 455 |
| }, |
| { |
| "epoch": 0.10552879100711172, |
| "eval_loss": 0.7430188059806824, |
| "eval_runtime": 14.2099, |
| "eval_samples_per_second": 463.34, |
| "eval_steps_per_second": 3.659, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.10667584308327598, |
| "eval_loss": 0.7378737330436707, |
| "eval_runtime": 16.2296, |
| "eval_samples_per_second": 405.68, |
| "eval_steps_per_second": 3.204, |
| "step": 465 |
| }, |
| { |
| "epoch": 0.10782289515944024, |
| "eval_loss": 0.7332107424736023, |
| "eval_runtime": 14.1689, |
| "eval_samples_per_second": 464.68, |
| "eval_steps_per_second": 3.67, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.10896994723560449, |
| "eval_loss": 0.7287834286689758, |
| "eval_runtime": 14.4646, |
| "eval_samples_per_second": 455.18, |
| "eval_steps_per_second": 3.595, |
| "step": 475 |
| }, |
| { |
| "epoch": 0.11011699931176876, |
| "eval_loss": 0.7247486710548401, |
| "eval_runtime": 14.1683, |
| "eval_samples_per_second": 464.699, |
| "eval_steps_per_second": 3.67, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.11126405138793301, |
| "eval_loss": 0.7211238741874695, |
| "eval_runtime": 14.235, |
| "eval_samples_per_second": 462.523, |
| "eval_steps_per_second": 3.653, |
| "step": 485 |
| }, |
| { |
| "epoch": 0.11241110346409727, |
| "eval_loss": 0.7170650959014893, |
| "eval_runtime": 14.2367, |
| "eval_samples_per_second": 462.466, |
| "eval_steps_per_second": 3.653, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.11355815554026152, |
| "eval_loss": 0.7122666835784912, |
| "eval_runtime": 14.2574, |
| "eval_samples_per_second": 461.796, |
| "eval_steps_per_second": 3.647, |
| "step": 495 |
| }, |
| { |
| "epoch": 0.11470520761642579, |
| "grad_norm": 5.7844438552856445, |
| "learning_rate": 3.8226299694189605e-06, |
| "loss": 1.2632, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.11470520761642579, |
| "eval_loss": 0.7078971862792969, |
| "eval_runtime": 14.238, |
| "eval_samples_per_second": 462.423, |
| "eval_steps_per_second": 3.652, |
| "step": 500 |
| } |
| ], |
| "logging_steps": 100, |
| "max_steps": 13077, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 5, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 0.0, |
| "train_batch_size": 128, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
