| { | |
| "best_metric": 1.8002580404281616, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.005402047375955487, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 2.7010236879777435e-05, | |
| "grad_norm": 2.8267626762390137, | |
| "learning_rate": 1.0018000000000001e-05, | |
| "loss": 2.4272, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 2.7010236879777435e-05, | |
| "eval_loss": 2.406503200531006, | |
| "eval_runtime": 323.2701, | |
| "eval_samples_per_second": 48.223, | |
| "eval_steps_per_second": 12.058, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 5.402047375955487e-05, | |
| "grad_norm": 4.365167140960693, | |
| "learning_rate": 2.0036000000000003e-05, | |
| "loss": 2.3437, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 8.103071063933231e-05, | |
| "grad_norm": 4.1403656005859375, | |
| "learning_rate": 3.0054e-05, | |
| "loss": 1.9492, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00010804094751910974, | |
| "grad_norm": 4.30712890625, | |
| "learning_rate": 4.0072000000000005e-05, | |
| "loss": 2.1199, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.00013505118439888718, | |
| "grad_norm": 3.849486827850342, | |
| "learning_rate": 5.009e-05, | |
| "loss": 2.1014, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.00016206142127866461, | |
| "grad_norm": 4.485018253326416, | |
| "learning_rate": 6.0108e-05, | |
| "loss": 2.5438, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.00018907165815844205, | |
| "grad_norm": 4.500787734985352, | |
| "learning_rate": 7.0126e-05, | |
| "loss": 3.0282, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.00021608189503821948, | |
| "grad_norm": 4.6447014808654785, | |
| "learning_rate": 8.014400000000001e-05, | |
| "loss": 2.5221, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0002430921319179969, | |
| "grad_norm": 4.513657569885254, | |
| "learning_rate": 9.016200000000001e-05, | |
| "loss": 2.6626, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.00027010236879777437, | |
| "grad_norm": 4.137198448181152, | |
| "learning_rate": 0.00010018, | |
| "loss": 2.2971, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.00029711260567755177, | |
| "grad_norm": 4.193837642669678, | |
| "learning_rate": 9.965273684210526e-05, | |
| "loss": 2.3521, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.00032412284255732923, | |
| "grad_norm": 4.4173583984375, | |
| "learning_rate": 9.912547368421053e-05, | |
| "loss": 2.3232, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0003511330794371067, | |
| "grad_norm": 3.6425013542175293, | |
| "learning_rate": 9.859821052631579e-05, | |
| "loss": 2.1981, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0003781433163168841, | |
| "grad_norm": 4.589376449584961, | |
| "learning_rate": 9.807094736842106e-05, | |
| "loss": 2.7546, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.00040515355319666155, | |
| "grad_norm": 3.747391939163208, | |
| "learning_rate": 9.754368421052633e-05, | |
| "loss": 2.0203, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.00043216379007643895, | |
| "grad_norm": 4.8549346923828125, | |
| "learning_rate": 9.701642105263158e-05, | |
| "loss": 2.7234, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0004591740269562164, | |
| "grad_norm": 6.589162826538086, | |
| "learning_rate": 9.648915789473685e-05, | |
| "loss": 3.0366, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0004861842638359938, | |
| "grad_norm": 4.530044078826904, | |
| "learning_rate": 9.596189473684211e-05, | |
| "loss": 2.2864, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0005131945007157713, | |
| "grad_norm": 5.117183208465576, | |
| "learning_rate": 9.543463157894737e-05, | |
| "loss": 2.5922, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0005402047375955487, | |
| "grad_norm": 4.832082748413086, | |
| "learning_rate": 9.490736842105264e-05, | |
| "loss": 2.4913, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0005672149744753262, | |
| "grad_norm": 3.8494045734405518, | |
| "learning_rate": 9.43801052631579e-05, | |
| "loss": 1.9371, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0005942252113551035, | |
| "grad_norm": 3.686887264251709, | |
| "learning_rate": 9.385284210526316e-05, | |
| "loss": 2.0102, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.000621235448234881, | |
| "grad_norm": 4.124686241149902, | |
| "learning_rate": 9.332557894736843e-05, | |
| "loss": 2.0297, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0006482456851146585, | |
| "grad_norm": 3.525970220565796, | |
| "learning_rate": 9.279831578947369e-05, | |
| "loss": 1.8261, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0006752559219944359, | |
| "grad_norm": 2.8308498859405518, | |
| "learning_rate": 9.227105263157896e-05, | |
| "loss": 1.3451, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0007022661588742134, | |
| "grad_norm": 4.149746894836426, | |
| "learning_rate": 9.174378947368421e-05, | |
| "loss": 2.3011, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0007292763957539907, | |
| "grad_norm": 4.0569634437561035, | |
| "learning_rate": 9.121652631578948e-05, | |
| "loss": 2.2933, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0007562866326337682, | |
| "grad_norm": 3.2820467948913574, | |
| "learning_rate": 9.068926315789475e-05, | |
| "loss": 1.6895, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0007832968695135456, | |
| "grad_norm": 4.019529819488525, | |
| "learning_rate": 9.016200000000001e-05, | |
| "loss": 1.8502, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0008103071063933231, | |
| "grad_norm": 3.2498834133148193, | |
| "learning_rate": 8.963473684210526e-05, | |
| "loss": 1.585, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0008373173432731006, | |
| "grad_norm": 4.720952987670898, | |
| "learning_rate": 8.910747368421053e-05, | |
| "loss": 2.1338, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0008643275801528779, | |
| "grad_norm": 3.4808425903320312, | |
| "learning_rate": 8.858021052631579e-05, | |
| "loss": 1.5556, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0008913378170326554, | |
| "grad_norm": 3.387439012527466, | |
| "learning_rate": 8.805294736842106e-05, | |
| "loss": 1.7884, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0009183480539124328, | |
| "grad_norm": 3.980246067047119, | |
| "learning_rate": 8.752568421052633e-05, | |
| "loss": 2.0877, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0009453582907922103, | |
| "grad_norm": 3.564471483230591, | |
| "learning_rate": 8.699842105263159e-05, | |
| "loss": 1.8872, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0009723685276719876, | |
| "grad_norm": 3.5869364738464355, | |
| "learning_rate": 8.647115789473686e-05, | |
| "loss": 1.6645, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0009993787645517652, | |
| "grad_norm": 4.11824893951416, | |
| "learning_rate": 8.594389473684211e-05, | |
| "loss": 1.9814, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0010263890014315425, | |
| "grad_norm": 3.116626024246216, | |
| "learning_rate": 8.541663157894737e-05, | |
| "loss": 1.4822, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.00105339923831132, | |
| "grad_norm": 3.8870787620544434, | |
| "learning_rate": 8.488936842105264e-05, | |
| "loss": 2.2419, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0010804094751910975, | |
| "grad_norm": 3.9773778915405273, | |
| "learning_rate": 8.43621052631579e-05, | |
| "loss": 1.9448, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0011074197120708748, | |
| "grad_norm": 4.536470413208008, | |
| "learning_rate": 8.383484210526316e-05, | |
| "loss": 2.0608, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0011344299489506524, | |
| "grad_norm": 3.5629429817199707, | |
| "learning_rate": 8.330757894736843e-05, | |
| "loss": 1.7975, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0011614401858304297, | |
| "grad_norm": 3.6513662338256836, | |
| "learning_rate": 8.278031578947369e-05, | |
| "loss": 1.9062, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.001188450422710207, | |
| "grad_norm": 3.534175157546997, | |
| "learning_rate": 8.225305263157896e-05, | |
| "loss": 1.8835, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0012154606595899846, | |
| "grad_norm": 3.87431001663208, | |
| "learning_rate": 8.172578947368422e-05, | |
| "loss": 1.7802, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.001242470896469762, | |
| "grad_norm": 3.6539125442504883, | |
| "learning_rate": 8.119852631578947e-05, | |
| "loss": 1.9138, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0012694811333495396, | |
| "grad_norm": 3.291635036468506, | |
| "learning_rate": 8.067126315789474e-05, | |
| "loss": 1.6632, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.001296491370229317, | |
| "grad_norm": 4.104748249053955, | |
| "learning_rate": 8.014400000000001e-05, | |
| "loss": 1.8235, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0013235016071090943, | |
| "grad_norm": 3.2002694606781006, | |
| "learning_rate": 7.961673684210527e-05, | |
| "loss": 1.3527, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0013505118439888718, | |
| "grad_norm": 3.7805356979370117, | |
| "learning_rate": 7.908947368421054e-05, | |
| "loss": 1.8133, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0013505118439888718, | |
| "eval_loss": 1.8875995874404907, | |
| "eval_runtime": 323.1313, | |
| "eval_samples_per_second": 48.244, | |
| "eval_steps_per_second": 12.063, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0013775220808686492, | |
| "grad_norm": 3.0262796878814697, | |
| "learning_rate": 7.856221052631579e-05, | |
| "loss": 2.1624, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.0014045323177484267, | |
| "grad_norm": 2.7900643348693848, | |
| "learning_rate": 7.803494736842106e-05, | |
| "loss": 1.9421, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.001431542554628204, | |
| "grad_norm": 2.8437888622283936, | |
| "learning_rate": 7.750768421052632e-05, | |
| "loss": 2.1949, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.0014585527915079814, | |
| "grad_norm": 2.974017858505249, | |
| "learning_rate": 7.698042105263157e-05, | |
| "loss": 2.146, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.001485563028387759, | |
| "grad_norm": 2.7476348876953125, | |
| "learning_rate": 7.645315789473686e-05, | |
| "loss": 2.0717, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0015125732652675364, | |
| "grad_norm": 2.9807491302490234, | |
| "learning_rate": 7.592589473684211e-05, | |
| "loss": 1.9227, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.001539583502147314, | |
| "grad_norm": 3.2775368690490723, | |
| "learning_rate": 7.539863157894737e-05, | |
| "loss": 2.129, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0015665937390270913, | |
| "grad_norm": 3.4354827404022217, | |
| "learning_rate": 7.487136842105264e-05, | |
| "loss": 2.3402, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0015936039759068686, | |
| "grad_norm": 3.4880383014678955, | |
| "learning_rate": 7.43441052631579e-05, | |
| "loss": 2.5008, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.0016206142127866462, | |
| "grad_norm": 3.056316375732422, | |
| "learning_rate": 7.381684210526315e-05, | |
| "loss": 1.8598, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0016476244496664235, | |
| "grad_norm": 4.000846862792969, | |
| "learning_rate": 7.328957894736844e-05, | |
| "loss": 2.1936, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0016746346865462011, | |
| "grad_norm": 3.282672643661499, | |
| "learning_rate": 7.276231578947369e-05, | |
| "loss": 2.2763, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.0017016449234259785, | |
| "grad_norm": 3.3345065116882324, | |
| "learning_rate": 7.223505263157895e-05, | |
| "loss": 2.0944, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0017286551603057558, | |
| "grad_norm": 3.6332178115844727, | |
| "learning_rate": 7.170778947368422e-05, | |
| "loss": 1.998, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.0017556653971855334, | |
| "grad_norm": 3.5125057697296143, | |
| "learning_rate": 7.118052631578947e-05, | |
| "loss": 2.0625, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.0017826756340653107, | |
| "grad_norm": 3.124108076095581, | |
| "learning_rate": 7.065326315789474e-05, | |
| "loss": 1.7761, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.001809685870945088, | |
| "grad_norm": 4.27671480178833, | |
| "learning_rate": 7.0126e-05, | |
| "loss": 2.4772, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.0018366961078248656, | |
| "grad_norm": 3.6868836879730225, | |
| "learning_rate": 6.959873684210527e-05, | |
| "loss": 2.1861, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.001863706344704643, | |
| "grad_norm": 3.100480556488037, | |
| "learning_rate": 6.907147368421054e-05, | |
| "loss": 1.8642, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.0018907165815844206, | |
| "grad_norm": 3.209296226501465, | |
| "learning_rate": 6.85442105263158e-05, | |
| "loss": 1.7952, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.001917726818464198, | |
| "grad_norm": 2.694065570831299, | |
| "learning_rate": 6.801694736842105e-05, | |
| "loss": 1.6004, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.0019447370553439753, | |
| "grad_norm": 3.056002616882324, | |
| "learning_rate": 6.748968421052632e-05, | |
| "loss": 1.9669, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.001971747292223753, | |
| "grad_norm": 3.0243594646453857, | |
| "learning_rate": 6.696242105263158e-05, | |
| "loss": 1.6867, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.0019987575291035304, | |
| "grad_norm": 2.582789897918701, | |
| "learning_rate": 6.643515789473685e-05, | |
| "loss": 1.485, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.0020257677659833075, | |
| "grad_norm": 3.274164915084839, | |
| "learning_rate": 6.590789473684212e-05, | |
| "loss": 1.8415, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.002052778002863085, | |
| "grad_norm": 2.930992603302002, | |
| "learning_rate": 6.538063157894737e-05, | |
| "loss": 1.5259, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0020797882397428627, | |
| "grad_norm": 2.674891710281372, | |
| "learning_rate": 6.485336842105264e-05, | |
| "loss": 1.5149, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.00210679847662264, | |
| "grad_norm": 3.3651278018951416, | |
| "learning_rate": 6.43261052631579e-05, | |
| "loss": 1.6893, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.0021338087135024174, | |
| "grad_norm": 3.8227052688598633, | |
| "learning_rate": 6.379884210526315e-05, | |
| "loss": 1.8186, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.002160818950382195, | |
| "grad_norm": 2.7289936542510986, | |
| "learning_rate": 6.327157894736842e-05, | |
| "loss": 1.464, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0021878291872619725, | |
| "grad_norm": 4.077561378479004, | |
| "learning_rate": 6.274431578947368e-05, | |
| "loss": 1.6729, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.0022148394241417496, | |
| "grad_norm": 3.349513053894043, | |
| "learning_rate": 6.221705263157895e-05, | |
| "loss": 1.8011, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.002241849661021527, | |
| "grad_norm": 3.1440014839172363, | |
| "learning_rate": 6.168978947368422e-05, | |
| "loss": 1.6098, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.0022688598979013048, | |
| "grad_norm": 3.1661434173583984, | |
| "learning_rate": 6.116252631578948e-05, | |
| "loss": 1.805, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.002295870134781082, | |
| "grad_norm": 3.5450026988983154, | |
| "learning_rate": 6.063526315789474e-05, | |
| "loss": 1.7791, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.0023228803716608595, | |
| "grad_norm": 3.2813570499420166, | |
| "learning_rate": 6.0108e-05, | |
| "loss": 1.8894, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.002349890608540637, | |
| "grad_norm": 2.9585013389587402, | |
| "learning_rate": 5.9580736842105264e-05, | |
| "loss": 1.547, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.002376900845420414, | |
| "grad_norm": 3.132396697998047, | |
| "learning_rate": 5.905347368421053e-05, | |
| "loss": 1.5077, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.0024039110823001917, | |
| "grad_norm": 3.426370859146118, | |
| "learning_rate": 5.85262105263158e-05, | |
| "loss": 1.6307, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.0024309213191799693, | |
| "grad_norm": 3.346588611602783, | |
| "learning_rate": 5.799894736842106e-05, | |
| "loss": 1.8479, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0024579315560597464, | |
| "grad_norm": 3.3509597778320312, | |
| "learning_rate": 5.747168421052632e-05, | |
| "loss": 1.7515, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.002484941792939524, | |
| "grad_norm": 3.85239839553833, | |
| "learning_rate": 5.694442105263158e-05, | |
| "loss": 2.0091, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.0025119520298193016, | |
| "grad_norm": 3.2585794925689697, | |
| "learning_rate": 5.641715789473684e-05, | |
| "loss": 1.3418, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.002538962266699079, | |
| "grad_norm": 3.9172606468200684, | |
| "learning_rate": 5.5889894736842104e-05, | |
| "loss": 1.708, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.0025659725035788563, | |
| "grad_norm": 3.717564821243286, | |
| "learning_rate": 5.5362631578947374e-05, | |
| "loss": 2.1761, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.002592982740458634, | |
| "grad_norm": 3.911463499069214, | |
| "learning_rate": 5.483536842105264e-05, | |
| "loss": 1.8531, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.0026199929773384114, | |
| "grad_norm": 3.121046304702759, | |
| "learning_rate": 5.43081052631579e-05, | |
| "loss": 1.2697, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.0026470032142181885, | |
| "grad_norm": 3.0859768390655518, | |
| "learning_rate": 5.378084210526316e-05, | |
| "loss": 1.4625, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.002674013451097966, | |
| "grad_norm": 5.068177223205566, | |
| "learning_rate": 5.3253578947368426e-05, | |
| "loss": 1.7805, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.0027010236879777437, | |
| "grad_norm": 4.796220302581787, | |
| "learning_rate": 5.272631578947368e-05, | |
| "loss": 2.0215, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0027010236879777437, | |
| "eval_loss": 1.84506356716156, | |
| "eval_runtime": 324.3562, | |
| "eval_samples_per_second": 48.061, | |
| "eval_steps_per_second": 12.018, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.002728033924857521, | |
| "grad_norm": 2.6804707050323486, | |
| "learning_rate": 5.2199052631578945e-05, | |
| "loss": 1.7513, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.0027550441617372984, | |
| "grad_norm": 2.743324041366577, | |
| "learning_rate": 5.167178947368422e-05, | |
| "loss": 1.5835, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.002782054398617076, | |
| "grad_norm": 2.639810800552368, | |
| "learning_rate": 5.114452631578948e-05, | |
| "loss": 1.7481, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.0028090646354968535, | |
| "grad_norm": 2.4762966632843018, | |
| "learning_rate": 5.061726315789474e-05, | |
| "loss": 1.7123, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.0028360748723766306, | |
| "grad_norm": 3.050942897796631, | |
| "learning_rate": 5.009e-05, | |
| "loss": 2.2438, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.002863085109256408, | |
| "grad_norm": 2.996593952178955, | |
| "learning_rate": 4.9562736842105266e-05, | |
| "loss": 2.2312, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.0028900953461361858, | |
| "grad_norm": 3.17206072807312, | |
| "learning_rate": 4.903547368421053e-05, | |
| "loss": 2.3683, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.002917105583015963, | |
| "grad_norm": 3.2420525550842285, | |
| "learning_rate": 4.850821052631579e-05, | |
| "loss": 2.261, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.0029441158198957405, | |
| "grad_norm": 3.376255989074707, | |
| "learning_rate": 4.7980947368421055e-05, | |
| "loss": 2.0114, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.002971126056775518, | |
| "grad_norm": 3.353220224380493, | |
| "learning_rate": 4.745368421052632e-05, | |
| "loss": 2.208, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.002998136293655295, | |
| "grad_norm": 3.3891849517822266, | |
| "learning_rate": 4.692642105263158e-05, | |
| "loss": 2.4548, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.0030251465305350727, | |
| "grad_norm": 3.2557313442230225, | |
| "learning_rate": 4.6399157894736844e-05, | |
| "loss": 2.1374, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.0030521567674148503, | |
| "grad_norm": 3.2548961639404297, | |
| "learning_rate": 4.5871894736842107e-05, | |
| "loss": 2.4684, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.003079167004294628, | |
| "grad_norm": 3.8495306968688965, | |
| "learning_rate": 4.5344631578947376e-05, | |
| "loss": 2.2422, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.003106177241174405, | |
| "grad_norm": 4.851099014282227, | |
| "learning_rate": 4.481736842105263e-05, | |
| "loss": 2.7144, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.0031331874780541826, | |
| "grad_norm": 3.6053466796875, | |
| "learning_rate": 4.4290105263157895e-05, | |
| "loss": 2.0815, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.00316019771493396, | |
| "grad_norm": 3.7665255069732666, | |
| "learning_rate": 4.3762842105263165e-05, | |
| "loss": 2.2639, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.0031872079518137373, | |
| "grad_norm": 2.965463638305664, | |
| "learning_rate": 4.323557894736843e-05, | |
| "loss": 1.6309, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.003214218188693515, | |
| "grad_norm": 2.783212661743164, | |
| "learning_rate": 4.2708315789473684e-05, | |
| "loss": 1.783, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.0032412284255732924, | |
| "grad_norm": 2.667881965637207, | |
| "learning_rate": 4.218105263157895e-05, | |
| "loss": 1.5607, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.0032682386624530695, | |
| "grad_norm": 2.851439952850342, | |
| "learning_rate": 4.1653789473684217e-05, | |
| "loss": 1.8242, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.003295248899332847, | |
| "grad_norm": 2.810152530670166, | |
| "learning_rate": 4.112652631578948e-05, | |
| "loss": 1.8485, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.0033222591362126247, | |
| "grad_norm": 2.722660779953003, | |
| "learning_rate": 4.0599263157894736e-05, | |
| "loss": 1.7689, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.0033492693730924022, | |
| "grad_norm": 2.6506547927856445, | |
| "learning_rate": 4.0072000000000005e-05, | |
| "loss": 1.5785, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.0033762796099721794, | |
| "grad_norm": 2.728020668029785, | |
| "learning_rate": 3.954473684210527e-05, | |
| "loss": 1.8581, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.003403289846851957, | |
| "grad_norm": 2.982895851135254, | |
| "learning_rate": 3.901747368421053e-05, | |
| "loss": 2.2773, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.0034303000837317345, | |
| "grad_norm": 2.847287178039551, | |
| "learning_rate": 3.849021052631579e-05, | |
| "loss": 1.5742, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.0034573103206115116, | |
| "grad_norm": 2.692082643508911, | |
| "learning_rate": 3.796294736842106e-05, | |
| "loss": 1.6871, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.003484320557491289, | |
| "grad_norm": 3.2670037746429443, | |
| "learning_rate": 3.743568421052632e-05, | |
| "loss": 1.8359, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.0035113307943710668, | |
| "grad_norm": 2.5639259815216064, | |
| "learning_rate": 3.6908421052631576e-05, | |
| "loss": 1.4626, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.003538341031250844, | |
| "grad_norm": 3.087303400039673, | |
| "learning_rate": 3.6381157894736846e-05, | |
| "loss": 1.8146, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.0035653512681306215, | |
| "grad_norm": 3.2488865852355957, | |
| "learning_rate": 3.585389473684211e-05, | |
| "loss": 1.8134, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.003592361505010399, | |
| "grad_norm": 3.2120273113250732, | |
| "learning_rate": 3.532663157894737e-05, | |
| "loss": 1.484, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.003619371741890176, | |
| "grad_norm": 2.597536087036133, | |
| "learning_rate": 3.4799368421052634e-05, | |
| "loss": 1.4691, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.0036463819787699537, | |
| "grad_norm": 3.0587146282196045, | |
| "learning_rate": 3.42721052631579e-05, | |
| "loss": 1.4249, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.0036733922156497313, | |
| "grad_norm": 3.46545672416687, | |
| "learning_rate": 3.374484210526316e-05, | |
| "loss": 1.5226, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.003700402452529509, | |
| "grad_norm": 2.8155453205108643, | |
| "learning_rate": 3.321757894736842e-05, | |
| "loss": 1.7263, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.003727412689409286, | |
| "grad_norm": 3.027614116668701, | |
| "learning_rate": 3.2690315789473686e-05, | |
| "loss": 1.8355, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.0037544229262890636, | |
| "grad_norm": 3.7238800525665283, | |
| "learning_rate": 3.216305263157895e-05, | |
| "loss": 1.4877, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.003781433163168841, | |
| "grad_norm": 3.3556723594665527, | |
| "learning_rate": 3.163578947368421e-05, | |
| "loss": 1.8014, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.0038084434000486183, | |
| "grad_norm": 2.854469060897827, | |
| "learning_rate": 3.1108526315789475e-05, | |
| "loss": 1.5076, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.003835453636928396, | |
| "grad_norm": 2.8921852111816406, | |
| "learning_rate": 3.058126315789474e-05, | |
| "loss": 1.7795, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.0038624638738081734, | |
| "grad_norm": 3.3138952255249023, | |
| "learning_rate": 3.0054e-05, | |
| "loss": 1.6566, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.0038894741106879505, | |
| "grad_norm": 3.637906551361084, | |
| "learning_rate": 2.9526736842105263e-05, | |
| "loss": 1.8738, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.0039164843475677285, | |
| "grad_norm": 3.2864222526550293, | |
| "learning_rate": 2.899947368421053e-05, | |
| "loss": 1.649, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.003943494584447506, | |
| "grad_norm": 2.840507984161377, | |
| "learning_rate": 2.847221052631579e-05, | |
| "loss": 1.4216, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.003970504821327283, | |
| "grad_norm": 2.9531781673431396, | |
| "learning_rate": 2.7944947368421052e-05, | |
| "loss": 1.5841, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.003997515058207061, | |
| "grad_norm": 3.016000509262085, | |
| "learning_rate": 2.741768421052632e-05, | |
| "loss": 1.3651, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.004024525295086838, | |
| "grad_norm": 3.315701961517334, | |
| "learning_rate": 2.689042105263158e-05, | |
| "loss": 1.4405, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.004051535531966615, | |
| "grad_norm": 3.887491226196289, | |
| "learning_rate": 2.636315789473684e-05, | |
| "loss": 1.8023, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.004051535531966615, | |
| "eval_loss": 1.8144692182540894, | |
| "eval_runtime": 321.8975, | |
| "eval_samples_per_second": 48.428, | |
| "eval_steps_per_second": 12.109, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.004078545768846393, | |
| "grad_norm": 2.3892409801483154, | |
| "learning_rate": 2.583589473684211e-05, | |
| "loss": 1.8418, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.00410555600572617, | |
| "grad_norm": 2.7030460834503174, | |
| "learning_rate": 2.530863157894737e-05, | |
| "loss": 1.9883, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.004132566242605947, | |
| "grad_norm": 2.809255838394165, | |
| "learning_rate": 2.4781368421052633e-05, | |
| "loss": 1.994, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.004159576479485725, | |
| "grad_norm": 2.7749948501586914, | |
| "learning_rate": 2.4254105263157896e-05, | |
| "loss": 1.7347, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.0041865867163655025, | |
| "grad_norm": 2.7523014545440674, | |
| "learning_rate": 2.372684210526316e-05, | |
| "loss": 1.4725, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.00421359695324528, | |
| "grad_norm": 3.1755049228668213, | |
| "learning_rate": 2.3199578947368422e-05, | |
| "loss": 1.9171, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.004240607190125058, | |
| "grad_norm": 3.1159286499023438, | |
| "learning_rate": 2.2672315789473688e-05, | |
| "loss": 2.0677, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.004267617427004835, | |
| "grad_norm": 2.927952527999878, | |
| "learning_rate": 2.2145052631578948e-05, | |
| "loss": 1.6294, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.004294627663884612, | |
| "grad_norm": 2.920624256134033, | |
| "learning_rate": 2.1617789473684214e-05, | |
| "loss": 1.9123, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.00432163790076439, | |
| "grad_norm": 3.3584253787994385, | |
| "learning_rate": 2.1090526315789473e-05, | |
| "loss": 2.0167, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.004348648137644167, | |
| "grad_norm": 4.036590099334717, | |
| "learning_rate": 2.056326315789474e-05, | |
| "loss": 2.3696, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.004375658374523945, | |
| "grad_norm": 3.8149144649505615, | |
| "learning_rate": 2.0036000000000003e-05, | |
| "loss": 2.3095, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.004402668611403722, | |
| "grad_norm": 3.711400032043457, | |
| "learning_rate": 1.9508736842105266e-05, | |
| "loss": 1.9094, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.004429678848283499, | |
| "grad_norm": 3.479764699935913, | |
| "learning_rate": 1.898147368421053e-05, | |
| "loss": 2.1776, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.004456689085163277, | |
| "grad_norm": 4.012089729309082, | |
| "learning_rate": 1.8454210526315788e-05, | |
| "loss": 1.9562, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.004483699322043054, | |
| "grad_norm": 5.2212815284729, | |
| "learning_rate": 1.7926947368421054e-05, | |
| "loss": 2.3301, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.0045107095589228315, | |
| "grad_norm": 4.928096294403076, | |
| "learning_rate": 1.7399684210526317e-05, | |
| "loss": 2.0416, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.0045377197958026095, | |
| "grad_norm": 3.51525616645813, | |
| "learning_rate": 1.687242105263158e-05, | |
| "loss": 2.1433, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.004564730032682387, | |
| "grad_norm": 2.5053954124450684, | |
| "learning_rate": 1.6345157894736843e-05, | |
| "loss": 1.2284, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.004591740269562164, | |
| "grad_norm": 3.1079280376434326, | |
| "learning_rate": 1.5817894736842106e-05, | |
| "loss": 1.8039, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.004618750506441942, | |
| "grad_norm": 2.9641072750091553, | |
| "learning_rate": 1.529063157894737e-05, | |
| "loss": 1.7875, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.004645760743321719, | |
| "grad_norm": 2.7184035778045654, | |
| "learning_rate": 1.4763368421052632e-05, | |
| "loss": 1.7067, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.004672770980201496, | |
| "grad_norm": 2.5093767642974854, | |
| "learning_rate": 1.4236105263157895e-05, | |
| "loss": 1.5895, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.004699781217081274, | |
| "grad_norm": 3.6973133087158203, | |
| "learning_rate": 1.370884210526316e-05, | |
| "loss": 1.8151, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.004726791453961051, | |
| "grad_norm": 2.5547432899475098, | |
| "learning_rate": 1.318157894736842e-05, | |
| "loss": 1.5454, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.004753801690840828, | |
| "grad_norm": 3.5522589683532715, | |
| "learning_rate": 1.2654315789473685e-05, | |
| "loss": 1.8471, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.004780811927720606, | |
| "grad_norm": 2.897728681564331, | |
| "learning_rate": 1.2127052631578948e-05, | |
| "loss": 1.7892, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.0048078221646003835, | |
| "grad_norm": 3.0070700645446777, | |
| "learning_rate": 1.1599789473684211e-05, | |
| "loss": 1.7458, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.004834832401480161, | |
| "grad_norm": 3.040104389190674, | |
| "learning_rate": 1.1072526315789474e-05, | |
| "loss": 1.6547, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.004861842638359939, | |
| "grad_norm": 2.3258042335510254, | |
| "learning_rate": 1.0545263157894737e-05, | |
| "loss": 1.2709, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.004888852875239716, | |
| "grad_norm": 2.8469483852386475, | |
| "learning_rate": 1.0018000000000001e-05, | |
| "loss": 1.6427, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.004915863112119493, | |
| "grad_norm": 3.239359140396118, | |
| "learning_rate": 9.490736842105264e-06, | |
| "loss": 1.6716, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.004942873348999271, | |
| "grad_norm": 2.6865952014923096, | |
| "learning_rate": 8.963473684210527e-06, | |
| "loss": 1.693, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.004969883585879048, | |
| "grad_norm": 3.1464898586273193, | |
| "learning_rate": 8.43621052631579e-06, | |
| "loss": 1.9052, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.004996893822758826, | |
| "grad_norm": 2.902247905731201, | |
| "learning_rate": 7.908947368421053e-06, | |
| "loss": 1.8767, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.005023904059638603, | |
| "grad_norm": 2.765345573425293, | |
| "learning_rate": 7.381684210526316e-06, | |
| "loss": 1.4926, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.00505091429651838, | |
| "grad_norm": 3.451422691345215, | |
| "learning_rate": 6.85442105263158e-06, | |
| "loss": 1.9693, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.005077924533398158, | |
| "grad_norm": 2.784844398498535, | |
| "learning_rate": 6.3271578947368425e-06, | |
| "loss": 1.4707, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.005104934770277935, | |
| "grad_norm": 3.7692813873291016, | |
| "learning_rate": 5.7998947368421054e-06, | |
| "loss": 2.1994, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.0051319450071577125, | |
| "grad_norm": 3.0001394748687744, | |
| "learning_rate": 5.272631578947368e-06, | |
| "loss": 1.5634, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.0051589552440374905, | |
| "grad_norm": 2.9124486446380615, | |
| "learning_rate": 4.745368421052632e-06, | |
| "loss": 1.5113, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.005185965480917268, | |
| "grad_norm": 2.946713924407959, | |
| "learning_rate": 4.218105263157895e-06, | |
| "loss": 1.6126, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.005212975717797045, | |
| "grad_norm": 3.2477729320526123, | |
| "learning_rate": 3.690842105263158e-06, | |
| "loss": 1.362, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.005239985954676823, | |
| "grad_norm": 3.283214807510376, | |
| "learning_rate": 3.1635789473684213e-06, | |
| "loss": 1.6263, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.0052669961915566, | |
| "grad_norm": 2.9746735095977783, | |
| "learning_rate": 2.636315789473684e-06, | |
| "loss": 1.5695, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.005294006428436377, | |
| "grad_norm": 3.2501778602600098, | |
| "learning_rate": 2.1090526315789475e-06, | |
| "loss": 1.733, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.005321016665316155, | |
| "grad_norm": 2.839373826980591, | |
| "learning_rate": 1.5817894736842106e-06, | |
| "loss": 1.6794, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.005348026902195932, | |
| "grad_norm": 3.3260462284088135, | |
| "learning_rate": 1.0545263157894738e-06, | |
| "loss": 1.5539, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.005375037139075709, | |
| "grad_norm": 2.837555408477783, | |
| "learning_rate": 5.272631578947369e-07, | |
| "loss": 1.4777, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.005402047375955487, | |
| "grad_norm": 2.7860605716705322, | |
| "learning_rate": 0.0, | |
| "loss": 1.6072, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.005402047375955487, | |
| "eval_loss": 1.8002580404281616, | |
| "eval_runtime": 323.0678, | |
| "eval_samples_per_second": 48.253, | |
| "eval_steps_per_second": 12.066, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5332652877938688.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |