| { | |
| "best_metric": 0.46117889881134033, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.1085334418667752, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.000542667209333876, | |
| "grad_norm": 1.8354359865188599, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 1.8492, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.000542667209333876, | |
| "eval_loss": 3.3348822593688965, | |
| "eval_runtime": 109.0843, | |
| "eval_samples_per_second": 28.455, | |
| "eval_steps_per_second": 14.228, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.001085334418667752, | |
| "grad_norm": 3.162593364715576, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 2.3208, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.001628001628001628, | |
| "grad_norm": 3.748394727706909, | |
| "learning_rate": 1e-05, | |
| "loss": 2.4477, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.002170668837335504, | |
| "grad_norm": 4.363120079040527, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 2.4994, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.00271333604666938, | |
| "grad_norm": 4.346236705780029, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 2.6665, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.003256003256003256, | |
| "grad_norm": 4.28872537612915, | |
| "learning_rate": 2e-05, | |
| "loss": 2.5443, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.003798670465337132, | |
| "grad_norm": 4.303534984588623, | |
| "learning_rate": 2.3333333333333336e-05, | |
| "loss": 2.3335, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.004341337674671008, | |
| "grad_norm": 3.9452812671661377, | |
| "learning_rate": 2.6666666666666667e-05, | |
| "loss": 2.257, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.004884004884004884, | |
| "grad_norm": 4.098211288452148, | |
| "learning_rate": 3e-05, | |
| "loss": 2.0106, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.00542667209333876, | |
| "grad_norm": 4.80372953414917, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 1.6557, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.005969339302672636, | |
| "grad_norm": 5.134768486022949, | |
| "learning_rate": 3.6666666666666666e-05, | |
| "loss": 1.4699, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.006512006512006512, | |
| "grad_norm": 5.335248947143555, | |
| "learning_rate": 4e-05, | |
| "loss": 1.5766, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.007054673721340388, | |
| "grad_norm": 5.714169979095459, | |
| "learning_rate": 4.3333333333333334e-05, | |
| "loss": 1.0583, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.007597340930674264, | |
| "grad_norm": 4.347225189208984, | |
| "learning_rate": 4.666666666666667e-05, | |
| "loss": 0.6941, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.00814000814000814, | |
| "grad_norm": 2.579446792602539, | |
| "learning_rate": 5e-05, | |
| "loss": 0.3363, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.008682675349342016, | |
| "grad_norm": 2.0687601566314697, | |
| "learning_rate": 5.333333333333333e-05, | |
| "loss": 0.1997, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.009225342558675891, | |
| "grad_norm": 2.1691503524780273, | |
| "learning_rate": 5.666666666666667e-05, | |
| "loss": 0.3088, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.009768009768009768, | |
| "grad_norm": 11.303365707397461, | |
| "learning_rate": 6e-05, | |
| "loss": 1.2146, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.010310676977343645, | |
| "grad_norm": 5.85835599899292, | |
| "learning_rate": 6.333333333333333e-05, | |
| "loss": 0.7783, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.01085334418667752, | |
| "grad_norm": 3.3280720710754395, | |
| "learning_rate": 6.666666666666667e-05, | |
| "loss": 0.5964, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.011396011396011397, | |
| "grad_norm": 3.908247947692871, | |
| "learning_rate": 7e-05, | |
| "loss": 0.4214, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.011938678605345272, | |
| "grad_norm": 7.787004470825195, | |
| "learning_rate": 7.333333333333333e-05, | |
| "loss": 0.4716, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.012481345814679148, | |
| "grad_norm": 7.936584949493408, | |
| "learning_rate": 7.666666666666667e-05, | |
| "loss": 0.4301, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.013024013024013023, | |
| "grad_norm": 5.566396713256836, | |
| "learning_rate": 8e-05, | |
| "loss": 0.727, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0135666802333469, | |
| "grad_norm": 3.142136335372925, | |
| "learning_rate": 8.333333333333334e-05, | |
| "loss": 0.5904, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.014109347442680775, | |
| "grad_norm": 2.2857656478881836, | |
| "learning_rate": 8.666666666666667e-05, | |
| "loss": 0.401, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.014652014652014652, | |
| "grad_norm": 1.4375312328338623, | |
| "learning_rate": 9e-05, | |
| "loss": 0.2759, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.015194681861348529, | |
| "grad_norm": 1.1307240724563599, | |
| "learning_rate": 9.333333333333334e-05, | |
| "loss": 0.2203, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.015737349070682406, | |
| "grad_norm": 1.402246356010437, | |
| "learning_rate": 9.666666666666667e-05, | |
| "loss": 0.3203, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.01628001628001628, | |
| "grad_norm": 5.896515846252441, | |
| "learning_rate": 0.0001, | |
| "loss": 0.621, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.016822683489350156, | |
| "grad_norm": 3.45434308052063, | |
| "learning_rate": 9.999146252290264e-05, | |
| "loss": 0.4411, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.017365350698684032, | |
| "grad_norm": 2.1787056922912598, | |
| "learning_rate": 9.996585300715116e-05, | |
| "loss": 0.302, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.01790801790801791, | |
| "grad_norm": 1.774258017539978, | |
| "learning_rate": 9.99231801983717e-05, | |
| "loss": 0.1932, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.018450685117351782, | |
| "grad_norm": 2.169252634048462, | |
| "learning_rate": 9.986345866928941e-05, | |
| "loss": 0.3516, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.01899335232668566, | |
| "grad_norm": 2.699979543685913, | |
| "learning_rate": 9.978670881475172e-05, | |
| "loss": 0.5089, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.019536019536019536, | |
| "grad_norm": 2.209733724594116, | |
| "learning_rate": 9.96929568447637e-05, | |
| "loss": 0.3997, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.020078686745353413, | |
| "grad_norm": 2.640620470046997, | |
| "learning_rate": 9.958223477553714e-05, | |
| "loss": 0.5742, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.02062135395468729, | |
| "grad_norm": 2.7024731636047363, | |
| "learning_rate": 9.94545804185573e-05, | |
| "loss": 0.4846, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.021164021164021163, | |
| "grad_norm": 3.163442850112915, | |
| "learning_rate": 9.931003736767013e-05, | |
| "loss": 0.4393, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.02170668837335504, | |
| "grad_norm": 2.5975422859191895, | |
| "learning_rate": 9.91486549841951e-05, | |
| "loss": 0.3196, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.022249355582688916, | |
| "grad_norm": 3.849959135055542, | |
| "learning_rate": 9.89704883800683e-05, | |
| "loss": 0.2137, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.022792022792022793, | |
| "grad_norm": 1.7151565551757812, | |
| "learning_rate": 9.877559839902184e-05, | |
| "loss": 0.1716, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.023334690001356666, | |
| "grad_norm": 2.8859987258911133, | |
| "learning_rate": 9.85640515958057e-05, | |
| "loss": 0.434, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.023877357210690543, | |
| "grad_norm": 6.60436487197876, | |
| "learning_rate": 9.833592021345937e-05, | |
| "loss": 0.529, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.02442002442002442, | |
| "grad_norm": 3.7714180946350098, | |
| "learning_rate": 9.809128215864097e-05, | |
| "loss": 0.1852, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.024962691629358297, | |
| "grad_norm": 1.2581069469451904, | |
| "learning_rate": 9.783022097502204e-05, | |
| "loss": 0.0727, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.025505358838692174, | |
| "grad_norm": 0.7298163175582886, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.0407, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.026048026048026047, | |
| "grad_norm": 0.47441840171813965, | |
| "learning_rate": 9.725919140804099e-05, | |
| "loss": 0.0686, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.026590693257359924, | |
| "grad_norm": 0.4721538722515106, | |
| "learning_rate": 9.694941803075283e-05, | |
| "loss": 0.0462, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0271333604666938, | |
| "grad_norm": 3.0532617568969727, | |
| "learning_rate": 9.662361147021779e-05, | |
| "loss": 0.3636, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0271333604666938, | |
| "eval_loss": 1.24636971950531, | |
| "eval_runtime": 109.441, | |
| "eval_samples_per_second": 28.362, | |
| "eval_steps_per_second": 14.181, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.027676027676027677, | |
| "grad_norm": 3.458024501800537, | |
| "learning_rate": 9.628188298907782e-05, | |
| "loss": 1.8542, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.02821869488536155, | |
| "grad_norm": 3.7774367332458496, | |
| "learning_rate": 9.592434928729616e-05, | |
| "loss": 1.969, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.028761362094695427, | |
| "grad_norm": 3.341789484024048, | |
| "learning_rate": 9.555113246230442e-05, | |
| "loss": 1.9199, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.029304029304029304, | |
| "grad_norm": 2.504328489303589, | |
| "learning_rate": 9.516235996730645e-05, | |
| "loss": 1.7206, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.02984669651336318, | |
| "grad_norm": 2.0271387100219727, | |
| "learning_rate": 9.475816456775313e-05, | |
| "loss": 1.3525, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.030389363722697058, | |
| "grad_norm": 1.4228510856628418, | |
| "learning_rate": 9.43386842960031e-05, | |
| "loss": 1.2227, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.03093203093203093, | |
| "grad_norm": 1.3967491388320923, | |
| "learning_rate": 9.39040624041849e-05, | |
| "loss": 1.138, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.03147469814136481, | |
| "grad_norm": 0.9970569014549255, | |
| "learning_rate": 9.345444731527642e-05, | |
| "loss": 1.0437, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.032017365350698684, | |
| "grad_norm": 0.9406002163887024, | |
| "learning_rate": 9.298999257241863e-05, | |
| "loss": 0.9546, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.03256003256003256, | |
| "grad_norm": 1.0668526887893677, | |
| "learning_rate": 9.251085678648072e-05, | |
| "loss": 0.8642, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.03310269976936644, | |
| "grad_norm": 0.9564623832702637, | |
| "learning_rate": 9.201720358189464e-05, | |
| "loss": 0.4615, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.03364536697870031, | |
| "grad_norm": 0.7209700345993042, | |
| "learning_rate": 9.150920154077754e-05, | |
| "loss": 0.536, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.03418803418803419, | |
| "grad_norm": 0.9375424385070801, | |
| "learning_rate": 9.098702414536107e-05, | |
| "loss": 0.7291, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.034730701397368065, | |
| "grad_norm": 0.9395576119422913, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.4634, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.03527336860670194, | |
| "grad_norm": 0.6775506734848022, | |
| "learning_rate": 8.9900861364012e-05, | |
| "loss": 0.333, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.03581603581603582, | |
| "grad_norm": 0.7822219133377075, | |
| "learning_rate": 8.933724690167417e-05, | |
| "loss": 0.197, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.03635870302536969, | |
| "grad_norm": 0.6001350283622742, | |
| "learning_rate": 8.876019880555649e-05, | |
| "loss": 0.2904, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.036901370234703565, | |
| "grad_norm": 0.35400617122650146, | |
| "learning_rate": 8.816991413705516e-05, | |
| "loss": 0.1933, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.037444037444037445, | |
| "grad_norm": 1.2720855474472046, | |
| "learning_rate": 8.756659447784368e-05, | |
| "loss": 0.7092, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.03798670465337132, | |
| "grad_norm": 1.343387246131897, | |
| "learning_rate": 8.695044586103296e-05, | |
| "loss": 0.4298, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0385293718627052, | |
| "grad_norm": 1.2000885009765625, | |
| "learning_rate": 8.632167870081121e-05, | |
| "loss": 0.2887, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.03907203907203907, | |
| "grad_norm": 0.7588382959365845, | |
| "learning_rate": 8.568050772058762e-05, | |
| "loss": 0.4491, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.039614706281372945, | |
| "grad_norm": 0.8610361218452454, | |
| "learning_rate": 8.502715187966455e-05, | |
| "loss": 0.247, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.040157373490706826, | |
| "grad_norm": 1.5872234106063843, | |
| "learning_rate": 8.436183429846313e-05, | |
| "loss": 0.2978, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.0407000407000407, | |
| "grad_norm": 1.328898310661316, | |
| "learning_rate": 8.368478218232787e-05, | |
| "loss": 0.4512, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.04124270790937458, | |
| "grad_norm": 1.1457730531692505, | |
| "learning_rate": 8.299622674393614e-05, | |
| "loss": 0.5961, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.04178537511870845, | |
| "grad_norm": 1.1928189992904663, | |
| "learning_rate": 8.229640312433937e-05, | |
| "loss": 0.2556, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.042328042328042326, | |
| "grad_norm": 0.9460683465003967, | |
| "learning_rate": 8.158555031266254e-05, | |
| "loss": 0.2041, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.042870709537376206, | |
| "grad_norm": 0.7704111337661743, | |
| "learning_rate": 8.086391106448965e-05, | |
| "loss": 0.2108, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.04341337674671008, | |
| "grad_norm": 0.714647114276886, | |
| "learning_rate": 8.013173181896283e-05, | |
| "loss": 0.2097, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.04395604395604396, | |
| "grad_norm": 1.409079670906067, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.2898, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.04449871116537783, | |
| "grad_norm": 1.393788456916809, | |
| "learning_rate": 7.863675700402526e-05, | |
| "loss": 0.3064, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.045041378374711706, | |
| "grad_norm": 1.120473861694336, | |
| "learning_rate": 7.787447196714427e-05, | |
| "loss": 0.218, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.045584045584045586, | |
| "grad_norm": 0.7864734530448914, | |
| "learning_rate": 7.710266782362247e-05, | |
| "loss": 0.1783, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.04612671279337946, | |
| "grad_norm": 1.4955497980117798, | |
| "learning_rate": 7.63216081438678e-05, | |
| "loss": 0.3466, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.04666938000271333, | |
| "grad_norm": 1.4786607027053833, | |
| "learning_rate": 7.553155965904535e-05, | |
| "loss": 0.3539, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.04721204721204721, | |
| "grad_norm": 1.4631497859954834, | |
| "learning_rate": 7.473279216998895e-05, | |
| "loss": 0.3824, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.047754714421381086, | |
| "grad_norm": 1.5295202732086182, | |
| "learning_rate": 7.392557845506432e-05, | |
| "loss": 0.3619, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.04829738163071497, | |
| "grad_norm": 0.977811336517334, | |
| "learning_rate": 7.311019417701566e-05, | |
| "loss": 0.3502, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.04884004884004884, | |
| "grad_norm": 1.178490400314331, | |
| "learning_rate": 7.228691778882693e-05, | |
| "loss": 0.3345, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.04938271604938271, | |
| "grad_norm": 0.8995553255081177, | |
| "learning_rate": 7.145603043863045e-05, | |
| "loss": 0.3611, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.049925383258716594, | |
| "grad_norm": 0.9408777356147766, | |
| "learning_rate": 7.061781587369519e-05, | |
| "loss": 0.1358, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.05046805046805047, | |
| "grad_norm": 0.9388291835784912, | |
| "learning_rate": 6.977256034352712e-05, | |
| "loss": 0.0947, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.05101071767738435, | |
| "grad_norm": 1.260951280593872, | |
| "learning_rate": 6.892055250211552e-05, | |
| "loss": 0.3262, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.05155338488671822, | |
| "grad_norm": 2.3727824687957764, | |
| "learning_rate": 6.806208330935766e-05, | |
| "loss": 0.2905, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.052096052096052094, | |
| "grad_norm": 1.8929994106292725, | |
| "learning_rate": 6.719744593169641e-05, | |
| "loss": 0.0723, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.052638719305385974, | |
| "grad_norm": 0.7458711266517639, | |
| "learning_rate": 6.632693564200416e-05, | |
| "loss": 0.0405, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.05318138651471985, | |
| "grad_norm": 0.3814379572868347, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.028, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.05372405372405373, | |
| "grad_norm": 0.41902849078178406, | |
| "learning_rate": 6.456948734446624e-05, | |
| "loss": 0.0304, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.0542667209333876, | |
| "grad_norm": 2.0934085845947266, | |
| "learning_rate": 6.368314950360415e-05, | |
| "loss": 0.365, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0542667209333876, | |
| "eval_loss": 1.2894482612609863, | |
| "eval_runtime": 109.2762, | |
| "eval_samples_per_second": 28.405, | |
| "eval_steps_per_second": 14.203, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.054809388142721474, | |
| "grad_norm": 4.360636234283447, | |
| "learning_rate": 6.279213887972179e-05, | |
| "loss": 2.1731, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.055352055352055354, | |
| "grad_norm": 4.3935866355896, | |
| "learning_rate": 6.189675975213094e-05, | |
| "loss": 2.3281, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.05589472256138923, | |
| "grad_norm": 3.9898064136505127, | |
| "learning_rate": 6.099731789198344e-05, | |
| "loss": 2.1225, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.0564373897707231, | |
| "grad_norm": 3.203814744949341, | |
| "learning_rate": 6.009412045785051e-05, | |
| "loss": 2.0483, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.05698005698005698, | |
| "grad_norm": 2.7947652339935303, | |
| "learning_rate": 5.918747589082853e-05, | |
| "loss": 1.773, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.057522724189390854, | |
| "grad_norm": 2.35221529006958, | |
| "learning_rate": 5.82776938092065e-05, | |
| "loss": 1.4951, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.058065391398724735, | |
| "grad_norm": 1.9279205799102783, | |
| "learning_rate": 5.736508490273188e-05, | |
| "loss": 1.4358, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.05860805860805861, | |
| "grad_norm": 1.6778484582901, | |
| "learning_rate": 5.644996082651017e-05, | |
| "loss": 1.1536, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.05915072581739248, | |
| "grad_norm": 1.3214651346206665, | |
| "learning_rate": 5.553263409457504e-05, | |
| "loss": 1.0911, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.05969339302672636, | |
| "grad_norm": 1.1080145835876465, | |
| "learning_rate": 5.4613417973165106e-05, | |
| "loss": 0.9185, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.060236060236060235, | |
| "grad_norm": 1.219183325767517, | |
| "learning_rate": 5.3692626373743706e-05, | |
| "loss": 0.4164, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.060778727445394115, | |
| "grad_norm": 0.7684482932090759, | |
| "learning_rate": 5.27705737457985e-05, | |
| "loss": 0.578, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.06132139465472799, | |
| "grad_norm": 0.7245362401008606, | |
| "learning_rate": 5.184757496945726e-05, | |
| "loss": 0.6867, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.06186406186406186, | |
| "grad_norm": 0.7537386417388916, | |
| "learning_rate": 5.092394524795649e-05, | |
| "loss": 0.3152, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.06240672907339574, | |
| "grad_norm": 0.7625346779823303, | |
| "learning_rate": 5e-05, | |
| "loss": 0.2146, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.06294939628272962, | |
| "grad_norm": 0.7598631978034973, | |
| "learning_rate": 4.907605475204352e-05, | |
| "loss": 0.1776, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.06349206349206349, | |
| "grad_norm": 0.6567097902297974, | |
| "learning_rate": 4.8152425030542766e-05, | |
| "loss": 0.2142, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.06403473070139737, | |
| "grad_norm": 0.6322920918464661, | |
| "learning_rate": 4.72294262542015e-05, | |
| "loss": 0.5704, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.06457739791073125, | |
| "grad_norm": 0.7434191703796387, | |
| "learning_rate": 4.6307373626256306e-05, | |
| "loss": 0.4056, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.06512006512006512, | |
| "grad_norm": 0.7763521671295166, | |
| "learning_rate": 4.5386582026834906e-05, | |
| "loss": 0.411, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.065662732329399, | |
| "grad_norm": 0.6602493524551392, | |
| "learning_rate": 4.446736590542497e-05, | |
| "loss": 0.3652, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.06620539953873288, | |
| "grad_norm": 0.6829279065132141, | |
| "learning_rate": 4.3550039173489845e-05, | |
| "loss": 0.1626, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.06674806674806674, | |
| "grad_norm": 0.7579967379570007, | |
| "learning_rate": 4.2634915097268115e-05, | |
| "loss": 0.1324, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.06729073395740062, | |
| "grad_norm": 0.6911221742630005, | |
| "learning_rate": 4.1722306190793495e-05, | |
| "loss": 0.3195, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.0678334011667345, | |
| "grad_norm": 0.6620950102806091, | |
| "learning_rate": 4.0812524109171476e-05, | |
| "loss": 0.4545, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.06837606837606838, | |
| "grad_norm": 0.7261799573898315, | |
| "learning_rate": 3.99058795421495e-05, | |
| "loss": 0.208, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.06891873558540225, | |
| "grad_norm": 0.6014336943626404, | |
| "learning_rate": 3.9002682108016585e-05, | |
| "loss": 0.2196, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.06946140279473613, | |
| "grad_norm": 0.7216453552246094, | |
| "learning_rate": 3.8103240247869075e-05, | |
| "loss": 0.146, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.07000407000407001, | |
| "grad_norm": 0.7673162817955017, | |
| "learning_rate": 3.720786112027822e-05, | |
| "loss": 0.245, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.07054673721340388, | |
| "grad_norm": 1.1044118404388428, | |
| "learning_rate": 3.631685049639586e-05, | |
| "loss": 0.3419, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.07108940442273776, | |
| "grad_norm": 1.402040719985962, | |
| "learning_rate": 3.543051265553377e-05, | |
| "loss": 0.2472, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.07163207163207164, | |
| "grad_norm": 1.219019889831543, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.1868, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.0721747388414055, | |
| "grad_norm": 0.914679229259491, | |
| "learning_rate": 3.367306435799584e-05, | |
| "loss": 0.1656, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.07271740605073938, | |
| "grad_norm": 0.9330466985702515, | |
| "learning_rate": 3.2802554068303596e-05, | |
| "loss": 0.2796, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.07326007326007326, | |
| "grad_norm": 1.7541531324386597, | |
| "learning_rate": 3.1937916690642356e-05, | |
| "loss": 0.263, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.07380274046940713, | |
| "grad_norm": 1.3093783855438232, | |
| "learning_rate": 3.107944749788449e-05, | |
| "loss": 0.2625, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.07434540767874101, | |
| "grad_norm": 1.9325745105743408, | |
| "learning_rate": 3.0227439656472877e-05, | |
| "loss": 0.3429, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.07488807488807489, | |
| "grad_norm": 1.6673192977905273, | |
| "learning_rate": 2.9382184126304834e-05, | |
| "loss": 0.2903, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.07543074209740877, | |
| "grad_norm": 0.6153419613838196, | |
| "learning_rate": 2.8543969561369556e-05, | |
| "loss": 0.3469, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.07597340930674264, | |
| "grad_norm": 0.7528604865074158, | |
| "learning_rate": 2.771308221117309e-05, | |
| "loss": 0.2047, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.07651607651607652, | |
| "grad_norm": 0.7970343232154846, | |
| "learning_rate": 2.688980582298435e-05, | |
| "loss": 0.1811, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.0770587437254104, | |
| "grad_norm": 1.1138691902160645, | |
| "learning_rate": 2.607442154493568e-05, | |
| "loss": 0.1813, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.07760141093474426, | |
| "grad_norm": 1.0628737211227417, | |
| "learning_rate": 2.5267207830011068e-05, | |
| "loss": 0.3945, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.07814407814407814, | |
| "grad_norm": 2.7330329418182373, | |
| "learning_rate": 2.446844034095466e-05, | |
| "loss": 0.3657, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.07868674535341202, | |
| "grad_norm": 2.918161153793335, | |
| "learning_rate": 2.3678391856132204e-05, | |
| "loss": 0.2343, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.07922941256274589, | |
| "grad_norm": 1.6331360340118408, | |
| "learning_rate": 2.2897332176377528e-05, | |
| "loss": 0.122, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.07977207977207977, | |
| "grad_norm": 1.2085009813308716, | |
| "learning_rate": 2.2125528032855724e-05, | |
| "loss": 0.091, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.08031474698141365, | |
| "grad_norm": 0.8547497391700745, | |
| "learning_rate": 2.136324299597474e-05, | |
| "loss": 0.0648, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.08085741419074752, | |
| "grad_norm": 0.6054078340530396, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.0454, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.0814000814000814, | |
| "grad_norm": 2.0531108379364014, | |
| "learning_rate": 1.9868268181037185e-05, | |
| "loss": 0.3661, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.0814000814000814, | |
| "eval_loss": 0.7246484756469727, | |
| "eval_runtime": 109.2416, | |
| "eval_samples_per_second": 28.414, | |
| "eval_steps_per_second": 14.207, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08194274860941528, | |
| "grad_norm": 2.4585351943969727, | |
| "learning_rate": 1.9136088935510362e-05, | |
| "loss": 1.6484, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.08248541581874916, | |
| "grad_norm": 2.7452824115753174, | |
| "learning_rate": 1.8414449687337464e-05, | |
| "loss": 1.7274, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.08302808302808302, | |
| "grad_norm": 2.7809834480285645, | |
| "learning_rate": 1.7703596875660645e-05, | |
| "loss": 1.6597, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.0835707502374169, | |
| "grad_norm": 2.423109769821167, | |
| "learning_rate": 1.700377325606388e-05, | |
| "loss": 1.6053, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.08411341744675079, | |
| "grad_norm": 2.2255589962005615, | |
| "learning_rate": 1.631521781767214e-05, | |
| "loss": 1.4753, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.08465608465608465, | |
| "grad_norm": 1.9394181966781616, | |
| "learning_rate": 1.5638165701536868e-05, | |
| "loss": 1.2572, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.08519875186541853, | |
| "grad_norm": 1.7988132238388062, | |
| "learning_rate": 1.4972848120335453e-05, | |
| "loss": 1.3386, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.08574141907475241, | |
| "grad_norm": 1.8108716011047363, | |
| "learning_rate": 1.4319492279412388e-05, | |
| "loss": 1.042, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.08628408628408628, | |
| "grad_norm": 1.5393013954162598, | |
| "learning_rate": 1.3678321299188801e-05, | |
| "loss": 1.0092, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.08682675349342016, | |
| "grad_norm": 1.369483232498169, | |
| "learning_rate": 1.3049554138967051e-05, | |
| "loss": 0.9053, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.08736942070275404, | |
| "grad_norm": 1.5199528932571411, | |
| "learning_rate": 1.2433405522156332e-05, | |
| "loss": 0.4746, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.08791208791208792, | |
| "grad_norm": 1.314212679862976, | |
| "learning_rate": 1.183008586294485e-05, | |
| "loss": 0.4789, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.08845475512142179, | |
| "grad_norm": 1.0513840913772583, | |
| "learning_rate": 1.1239801194443506e-05, | |
| "loss": 0.7773, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.08899742233075567, | |
| "grad_norm": 0.9221037030220032, | |
| "learning_rate": 1.066275309832584e-05, | |
| "loss": 0.613, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.08954008954008955, | |
| "grad_norm": 0.8674269914627075, | |
| "learning_rate": 1.0099138635988026e-05, | |
| "loss": 0.3583, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.09008275674942341, | |
| "grad_norm": 0.7268597483634949, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.2715, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.09062542395875729, | |
| "grad_norm": 1.0291173458099365, | |
| "learning_rate": 9.012975854638949e-06, | |
| "loss": 0.1534, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.09116809116809117, | |
| "grad_norm": 0.6670011878013611, | |
| "learning_rate": 8.490798459222476e-06, | |
| "loss": 0.4197, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.09171075837742504, | |
| "grad_norm": 0.7457062602043152, | |
| "learning_rate": 7.982796418105371e-06, | |
| "loss": 0.4144, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.09225342558675892, | |
| "grad_norm": 0.6587404012680054, | |
| "learning_rate": 7.489143213519301e-06, | |
| "loss": 0.5097, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.0927960927960928, | |
| "grad_norm": 0.794647753238678, | |
| "learning_rate": 7.010007427581378e-06, | |
| "loss": 0.3255, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.09333876000542667, | |
| "grad_norm": 0.5329033732414246, | |
| "learning_rate": 6.5455526847235825e-06, | |
| "loss": 0.255, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.09388142721476055, | |
| "grad_norm": 0.5956745743751526, | |
| "learning_rate": 6.0959375958151045e-06, | |
| "loss": 0.1974, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.09442409442409443, | |
| "grad_norm": 0.5087482333183289, | |
| "learning_rate": 5.6613157039969055e-06, | |
| "loss": 0.318, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.0949667616334283, | |
| "grad_norm": 0.6372755169868469, | |
| "learning_rate": 5.241835432246889e-06, | |
| "loss": 0.5034, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.09550942884276217, | |
| "grad_norm": 0.6690561175346375, | |
| "learning_rate": 4.837640032693558e-06, | |
| "loss": 0.2968, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.09605209605209605, | |
| "grad_norm": 0.7963606119155884, | |
| "learning_rate": 4.448867537695578e-06, | |
| "loss": 0.2768, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.09659476326142993, | |
| "grad_norm": 0.7704768776893616, | |
| "learning_rate": 4.075650712703849e-06, | |
| "loss": 0.3783, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.0971374304707638, | |
| "grad_norm": 0.7436457276344299, | |
| "learning_rate": 3.71811701092219e-06, | |
| "loss": 0.3566, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.09768009768009768, | |
| "grad_norm": 1.0820454359054565, | |
| "learning_rate": 3.376388529782215e-06, | |
| "loss": 0.2758, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.09822276488943156, | |
| "grad_norm": 1.1378484964370728, | |
| "learning_rate": 3.0505819692471792e-06, | |
| "loss": 0.2739, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.09876543209876543, | |
| "grad_norm": 1.144866704940796, | |
| "learning_rate": 2.7408085919590264e-06, | |
| "loss": 0.3067, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.0993080993080993, | |
| "grad_norm": 1.2071819305419922, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.2637, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.09985076651743319, | |
| "grad_norm": 0.7118948698043823, | |
| "learning_rate": 2.1697790249779636e-06, | |
| "loss": 0.2811, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.10039343372676705, | |
| "grad_norm": 0.7794045805931091, | |
| "learning_rate": 1.908717841359048e-06, | |
| "loss": 0.2323, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.10093610093610093, | |
| "grad_norm": 0.7099078893661499, | |
| "learning_rate": 1.6640797865406288e-06, | |
| "loss": 0.2769, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.10147876814543481, | |
| "grad_norm": 0.579207181930542, | |
| "learning_rate": 1.4359484041943038e-06, | |
| "loss": 0.2418, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.1020214353547687, | |
| "grad_norm": 0.9049748778343201, | |
| "learning_rate": 1.2244016009781701e-06, | |
| "loss": 0.2772, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.10256410256410256, | |
| "grad_norm": 0.7932896614074707, | |
| "learning_rate": 1.0295116199317057e-06, | |
| "loss": 0.3465, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.10310676977343644, | |
| "grad_norm": 1.252280592918396, | |
| "learning_rate": 8.513450158049108e-07, | |
| "loss": 0.267, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.10364943698277032, | |
| "grad_norm": 0.9156208038330078, | |
| "learning_rate": 6.899626323298713e-07, | |
| "loss": 0.2506, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.10419210419210419, | |
| "grad_norm": 1.0016757249832153, | |
| "learning_rate": 5.454195814427021e-07, | |
| "loss": 0.1856, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.10473477140143807, | |
| "grad_norm": 1.0268343687057495, | |
| "learning_rate": 4.177652244628627e-07, | |
| "loss": 0.2485, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.10527743861077195, | |
| "grad_norm": 1.2182773351669312, | |
| "learning_rate": 3.0704315523631953e-07, | |
| "loss": 0.3693, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10582010582010581, | |
| "grad_norm": 1.4859917163848877, | |
| "learning_rate": 2.1329118524827662e-07, | |
| "loss": 0.34, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.1063627730294397, | |
| "grad_norm": 1.299428939819336, | |
| "learning_rate": 1.3654133071059893e-07, | |
| "loss": 0.1563, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.10690544023877357, | |
| "grad_norm": 1.2561206817626953, | |
| "learning_rate": 7.681980162830282e-08, | |
| "loss": 0.1746, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.10744810744810745, | |
| "grad_norm": 1.2994426488876343, | |
| "learning_rate": 3.4146992848854695e-08, | |
| "loss": 0.1565, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.10799077465744132, | |
| "grad_norm": 1.261816382408142, | |
| "learning_rate": 8.537477097364522e-09, | |
| "loss": 0.1866, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.1085334418667752, | |
| "grad_norm": 1.7563239336013794, | |
| "learning_rate": 0.0, | |
| "loss": 0.4114, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.1085334418667752, | |
| "eval_loss": 0.46117889881134033, | |
| "eval_runtime": 109.2594, | |
| "eval_samples_per_second": 28.409, | |
| "eval_steps_per_second": 14.205, | |
| "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": 5.449698231975936e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |