| { | |
| "best_metric": 1.0460100173950195, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 1.322314049586777, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.006611570247933884, | |
| "grad_norm": 0.9164348840713501, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 4.3726, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.006611570247933884, | |
| "eval_loss": 1.6257771253585815, | |
| "eval_runtime": 14.0709, | |
| "eval_samples_per_second": 18.123, | |
| "eval_steps_per_second": 9.097, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.013223140495867768, | |
| "grad_norm": 0.8439459800720215, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 4.5082, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.019834710743801654, | |
| "grad_norm": 0.8232372403144836, | |
| "learning_rate": 1e-05, | |
| "loss": 4.7082, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.026446280991735537, | |
| "grad_norm": 1.0905650854110718, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 4.778, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.03305785123966942, | |
| "grad_norm": 0.9605323672294617, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 5.4669, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.03966942148760331, | |
| "grad_norm": 1.00596022605896, | |
| "learning_rate": 2e-05, | |
| "loss": 4.9968, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.04628099173553719, | |
| "grad_norm": 1.1926850080490112, | |
| "learning_rate": 2.3333333333333336e-05, | |
| "loss": 4.9989, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.05289256198347107, | |
| "grad_norm": 1.3035999536514282, | |
| "learning_rate": 2.6666666666666667e-05, | |
| "loss": 5.3719, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.05950413223140496, | |
| "grad_norm": 1.09066641330719, | |
| "learning_rate": 3e-05, | |
| "loss": 5.2938, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.06611570247933884, | |
| "grad_norm": 1.2595545053482056, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 5.4152, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.07272727272727272, | |
| "grad_norm": 1.241004228591919, | |
| "learning_rate": 3.6666666666666666e-05, | |
| "loss": 5.6351, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.07933884297520662, | |
| "grad_norm": 1.1945130825042725, | |
| "learning_rate": 4e-05, | |
| "loss": 5.4813, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0859504132231405, | |
| "grad_norm": 1.2076464891433716, | |
| "learning_rate": 4.3333333333333334e-05, | |
| "loss": 5.0437, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.09256198347107437, | |
| "grad_norm": 1.3075780868530273, | |
| "learning_rate": 4.666666666666667e-05, | |
| "loss": 5.4353, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.09917355371900827, | |
| "grad_norm": 1.3237282037734985, | |
| "learning_rate": 5e-05, | |
| "loss": 5.5073, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.10578512396694215, | |
| "grad_norm": 1.3928046226501465, | |
| "learning_rate": 5.333333333333333e-05, | |
| "loss": 5.7509, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.11239669421487604, | |
| "grad_norm": 1.4480962753295898, | |
| "learning_rate": 5.666666666666667e-05, | |
| "loss": 5.6378, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.11900826446280992, | |
| "grad_norm": 1.5458139181137085, | |
| "learning_rate": 6e-05, | |
| "loss": 5.4339, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.1256198347107438, | |
| "grad_norm": 1.5952441692352295, | |
| "learning_rate": 6.333333333333333e-05, | |
| "loss": 5.2634, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.1322314049586777, | |
| "grad_norm": 1.6097549200057983, | |
| "learning_rate": 6.666666666666667e-05, | |
| "loss": 5.4341, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.13884297520661157, | |
| "grad_norm": 1.6773674488067627, | |
| "learning_rate": 7e-05, | |
| "loss": 5.2169, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.14545454545454545, | |
| "grad_norm": 1.8317288160324097, | |
| "learning_rate": 7.333333333333333e-05, | |
| "loss": 5.3538, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.15206611570247933, | |
| "grad_norm": 1.7912957668304443, | |
| "learning_rate": 7.666666666666667e-05, | |
| "loss": 5.867, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.15867768595041323, | |
| "grad_norm": 1.8667272329330444, | |
| "learning_rate": 8e-05, | |
| "loss": 5.2434, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.1652892561983471, | |
| "grad_norm": 1.8276578187942505, | |
| "learning_rate": 8.333333333333334e-05, | |
| "loss": 5.4768, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.171900826446281, | |
| "grad_norm": 2.0258138179779053, | |
| "learning_rate": 8.666666666666667e-05, | |
| "loss": 5.7999, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.17851239669421487, | |
| "grad_norm": 2.286808967590332, | |
| "learning_rate": 9e-05, | |
| "loss": 4.9834, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.18512396694214875, | |
| "grad_norm": 1.9957102537155151, | |
| "learning_rate": 9.333333333333334e-05, | |
| "loss": 5.8383, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.19173553719008266, | |
| "grad_norm": 2.4122209548950195, | |
| "learning_rate": 9.666666666666667e-05, | |
| "loss": 5.8001, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.19834710743801653, | |
| "grad_norm": 2.4733917713165283, | |
| "learning_rate": 0.0001, | |
| "loss": 6.1634, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.2049586776859504, | |
| "grad_norm": 2.4017982482910156, | |
| "learning_rate": 9.999146252290264e-05, | |
| "loss": 6.1423, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.2115702479338843, | |
| "grad_norm": 2.3116109371185303, | |
| "learning_rate": 9.996585300715116e-05, | |
| "loss": 5.8569, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.21818181818181817, | |
| "grad_norm": 2.673358678817749, | |
| "learning_rate": 9.99231801983717e-05, | |
| "loss": 5.7446, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.22479338842975208, | |
| "grad_norm": 4.113661289215088, | |
| "learning_rate": 9.986345866928941e-05, | |
| "loss": 5.0621, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.23140495867768596, | |
| "grad_norm": 2.5697896480560303, | |
| "learning_rate": 9.978670881475172e-05, | |
| "loss": 5.4427, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.23801652892561984, | |
| "grad_norm": 3.4522359371185303, | |
| "learning_rate": 9.96929568447637e-05, | |
| "loss": 6.1919, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.24462809917355371, | |
| "grad_norm": 4.006448268890381, | |
| "learning_rate": 9.958223477553714e-05, | |
| "loss": 5.7438, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.2512396694214876, | |
| "grad_norm": 1.7176642417907715, | |
| "learning_rate": 9.94545804185573e-05, | |
| "loss": 4.2525, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.2578512396694215, | |
| "grad_norm": 1.2814679145812988, | |
| "learning_rate": 9.931003736767013e-05, | |
| "loss": 4.1325, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.2644628099173554, | |
| "grad_norm": 1.0523453950881958, | |
| "learning_rate": 9.91486549841951e-05, | |
| "loss": 3.9787, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.27107438016528923, | |
| "grad_norm": 0.9023359417915344, | |
| "learning_rate": 9.89704883800683e-05, | |
| "loss": 4.4185, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.27768595041322314, | |
| "grad_norm": 0.7565967440605164, | |
| "learning_rate": 9.877559839902184e-05, | |
| "loss": 4.6076, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.28429752066115704, | |
| "grad_norm": 0.740451455116272, | |
| "learning_rate": 9.85640515958057e-05, | |
| "loss": 4.4481, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.2909090909090909, | |
| "grad_norm": 0.7501481175422668, | |
| "learning_rate": 9.833592021345937e-05, | |
| "loss": 4.7943, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.2975206611570248, | |
| "grad_norm": 0.8105722665786743, | |
| "learning_rate": 9.809128215864097e-05, | |
| "loss": 4.4628, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.30413223140495865, | |
| "grad_norm": 1.0324733257293701, | |
| "learning_rate": 9.783022097502204e-05, | |
| "loss": 4.873, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.31074380165289256, | |
| "grad_norm": 0.9048627018928528, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 4.4168, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.31735537190082647, | |
| "grad_norm": 0.9578453302383423, | |
| "learning_rate": 9.725919140804099e-05, | |
| "loss": 4.8329, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.3239669421487603, | |
| "grad_norm": 0.9514991641044617, | |
| "learning_rate": 9.694941803075283e-05, | |
| "loss": 5.0063, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.3305785123966942, | |
| "grad_norm": 0.9336065649986267, | |
| "learning_rate": 9.662361147021779e-05, | |
| "loss": 4.6913, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.3305785123966942, | |
| "eval_loss": 1.1803337335586548, | |
| "eval_runtime": 14.155, | |
| "eval_samples_per_second": 18.015, | |
| "eval_steps_per_second": 9.043, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.3371900826446281, | |
| "grad_norm": 0.9709968566894531, | |
| "learning_rate": 9.628188298907782e-05, | |
| "loss": 4.9291, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.343801652892562, | |
| "grad_norm": 0.9347379803657532, | |
| "learning_rate": 9.592434928729616e-05, | |
| "loss": 4.3365, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.3504132231404959, | |
| "grad_norm": 1.0081912279129028, | |
| "learning_rate": 9.555113246230442e-05, | |
| "loss": 5.2088, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.35702479338842974, | |
| "grad_norm": 1.0832982063293457, | |
| "learning_rate": 9.516235996730645e-05, | |
| "loss": 5.0906, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.36363636363636365, | |
| "grad_norm": 1.1196680068969727, | |
| "learning_rate": 9.475816456775313e-05, | |
| "loss": 4.6937, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.3702479338842975, | |
| "grad_norm": 1.208888292312622, | |
| "learning_rate": 9.43386842960031e-05, | |
| "loss": 4.599, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.3768595041322314, | |
| "grad_norm": 1.2675275802612305, | |
| "learning_rate": 9.39040624041849e-05, | |
| "loss": 4.9299, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.3834710743801653, | |
| "grad_norm": 1.3029289245605469, | |
| "learning_rate": 9.345444731527642e-05, | |
| "loss": 4.7858, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.39008264462809916, | |
| "grad_norm": 1.3226304054260254, | |
| "learning_rate": 9.298999257241863e-05, | |
| "loss": 4.9087, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.39669421487603307, | |
| "grad_norm": 1.4871385097503662, | |
| "learning_rate": 9.251085678648072e-05, | |
| "loss": 4.8845, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.4033057851239669, | |
| "grad_norm": 1.5894980430603027, | |
| "learning_rate": 9.201720358189464e-05, | |
| "loss": 5.3603, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.4099173553719008, | |
| "grad_norm": 1.7288110256195068, | |
| "learning_rate": 9.150920154077754e-05, | |
| "loss": 5.3581, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.41652892561983473, | |
| "grad_norm": 1.6703156232833862, | |
| "learning_rate": 9.098702414536107e-05, | |
| "loss": 4.4855, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.4231404958677686, | |
| "grad_norm": 1.7052879333496094, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 4.1406, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.4297520661157025, | |
| "grad_norm": 1.659575343132019, | |
| "learning_rate": 8.9900861364012e-05, | |
| "loss": 5.162, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.43636363636363634, | |
| "grad_norm": 1.8146637678146362, | |
| "learning_rate": 8.933724690167417e-05, | |
| "loss": 5.2744, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.44297520661157025, | |
| "grad_norm": 1.932379126548767, | |
| "learning_rate": 8.876019880555649e-05, | |
| "loss": 5.659, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.44958677685950416, | |
| "grad_norm": 2.114957332611084, | |
| "learning_rate": 8.816991413705516e-05, | |
| "loss": 5.7257, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.456198347107438, | |
| "grad_norm": 2.27786922454834, | |
| "learning_rate": 8.756659447784368e-05, | |
| "loss": 6.0124, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.4628099173553719, | |
| "grad_norm": 2.2900562286376953, | |
| "learning_rate": 8.695044586103296e-05, | |
| "loss": 4.4817, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.46942148760330576, | |
| "grad_norm": 2.6403229236602783, | |
| "learning_rate": 8.632167870081121e-05, | |
| "loss": 4.2367, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.47603305785123967, | |
| "grad_norm": 2.5031816959381104, | |
| "learning_rate": 8.568050772058762e-05, | |
| "loss": 4.8966, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.4826446280991736, | |
| "grad_norm": 3.2316620349884033, | |
| "learning_rate": 8.502715187966455e-05, | |
| "loss": 5.2742, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.48925619834710743, | |
| "grad_norm": 4.309781074523926, | |
| "learning_rate": 8.436183429846313e-05, | |
| "loss": 4.9375, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.49586776859504134, | |
| "grad_norm": 1.4185460805892944, | |
| "learning_rate": 8.368478218232787e-05, | |
| "loss": 3.5337, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.5024793388429752, | |
| "grad_norm": 1.15443754196167, | |
| "learning_rate": 8.299622674393614e-05, | |
| "loss": 3.638, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.509090909090909, | |
| "grad_norm": 0.8915994763374329, | |
| "learning_rate": 8.229640312433937e-05, | |
| "loss": 3.9779, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.515702479338843, | |
| "grad_norm": 0.8362914323806763, | |
| "learning_rate": 8.158555031266254e-05, | |
| "loss": 4.2846, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.5223140495867769, | |
| "grad_norm": 0.7924783825874329, | |
| "learning_rate": 8.086391106448965e-05, | |
| "loss": 4.506, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.5289256198347108, | |
| "grad_norm": 0.7508944869041443, | |
| "learning_rate": 8.013173181896283e-05, | |
| "loss": 4.4354, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.5355371900826447, | |
| "grad_norm": 0.7106804251670837, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 4.4622, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.5421487603305785, | |
| "grad_norm": 0.7450253367424011, | |
| "learning_rate": 7.863675700402526e-05, | |
| "loss": 4.2489, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.5487603305785124, | |
| "grad_norm": 0.8078528046607971, | |
| "learning_rate": 7.787447196714427e-05, | |
| "loss": 4.4358, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.5553719008264463, | |
| "grad_norm": 0.8072985410690308, | |
| "learning_rate": 7.710266782362247e-05, | |
| "loss": 4.5949, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.5619834710743802, | |
| "grad_norm": 0.9731318950653076, | |
| "learning_rate": 7.63216081438678e-05, | |
| "loss": 4.6721, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.5685950413223141, | |
| "grad_norm": 0.9238491058349609, | |
| "learning_rate": 7.553155965904535e-05, | |
| "loss": 4.4037, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.5752066115702479, | |
| "grad_norm": 0.9422857761383057, | |
| "learning_rate": 7.473279216998895e-05, | |
| "loss": 4.1799, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.5818181818181818, | |
| "grad_norm": 1.0530166625976562, | |
| "learning_rate": 7.392557845506432e-05, | |
| "loss": 4.4404, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.5884297520661157, | |
| "grad_norm": 1.1109483242034912, | |
| "learning_rate": 7.311019417701566e-05, | |
| "loss": 4.4643, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.5950413223140496, | |
| "grad_norm": 1.0867807865142822, | |
| "learning_rate": 7.228691778882693e-05, | |
| "loss": 4.8055, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.6016528925619835, | |
| "grad_norm": 1.1179331541061401, | |
| "learning_rate": 7.145603043863045e-05, | |
| "loss": 4.6242, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.6082644628099173, | |
| "grad_norm": 1.1954206228256226, | |
| "learning_rate": 7.061781587369519e-05, | |
| "loss": 4.6997, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.6148760330578512, | |
| "grad_norm": 1.1525213718414307, | |
| "learning_rate": 6.977256034352712e-05, | |
| "loss": 4.7898, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.6214876033057851, | |
| "grad_norm": 1.2016812562942505, | |
| "learning_rate": 6.892055250211552e-05, | |
| "loss": 4.5403, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.628099173553719, | |
| "grad_norm": 1.323207974433899, | |
| "learning_rate": 6.806208330935766e-05, | |
| "loss": 4.7135, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.6347107438016529, | |
| "grad_norm": 1.2732231616973877, | |
| "learning_rate": 6.719744593169641e-05, | |
| "loss": 4.869, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.6413223140495867, | |
| "grad_norm": 1.5098425149917603, | |
| "learning_rate": 6.632693564200416e-05, | |
| "loss": 5.4372, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.6479338842975206, | |
| "grad_norm": 1.40667724609375, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 5.2639, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.6545454545454545, | |
| "grad_norm": 1.5204859972000122, | |
| "learning_rate": 6.456948734446624e-05, | |
| "loss": 4.8975, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.6611570247933884, | |
| "grad_norm": 1.6273950338363647, | |
| "learning_rate": 6.368314950360415e-05, | |
| "loss": 4.8439, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.6611570247933884, | |
| "eval_loss": 1.090095043182373, | |
| "eval_runtime": 14.1768, | |
| "eval_samples_per_second": 17.987, | |
| "eval_steps_per_second": 9.029, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.6677685950413224, | |
| "grad_norm": 1.766390323638916, | |
| "learning_rate": 6.279213887972179e-05, | |
| "loss": 4.463, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.6743801652892562, | |
| "grad_norm": 1.9048361778259277, | |
| "learning_rate": 6.189675975213094e-05, | |
| "loss": 5.3703, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.6809917355371901, | |
| "grad_norm": 1.8197228908538818, | |
| "learning_rate": 6.099731789198344e-05, | |
| "loss": 4.9689, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.687603305785124, | |
| "grad_norm": 2.086484432220459, | |
| "learning_rate": 6.009412045785051e-05, | |
| "loss": 5.6172, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.6942148760330579, | |
| "grad_norm": 2.267761707305908, | |
| "learning_rate": 5.918747589082853e-05, | |
| "loss": 4.8357, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.7008264462809918, | |
| "grad_norm": 2.259352445602417, | |
| "learning_rate": 5.82776938092065e-05, | |
| "loss": 5.1136, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.7074380165289256, | |
| "grad_norm": 2.6360812187194824, | |
| "learning_rate": 5.736508490273188e-05, | |
| "loss": 5.1514, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.7140495867768595, | |
| "grad_norm": 2.8661835193634033, | |
| "learning_rate": 5.644996082651017e-05, | |
| "loss": 4.3032, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.7206611570247934, | |
| "grad_norm": 2.702547073364258, | |
| "learning_rate": 5.553263409457504e-05, | |
| "loss": 4.8666, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.7272727272727273, | |
| "grad_norm": 2.970484495162964, | |
| "learning_rate": 5.4613417973165106e-05, | |
| "loss": 4.8789, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.7338842975206612, | |
| "grad_norm": 3.8799359798431396, | |
| "learning_rate": 5.3692626373743706e-05, | |
| "loss": 5.1477, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.740495867768595, | |
| "grad_norm": 0.6784772276878357, | |
| "learning_rate": 5.27705737457985e-05, | |
| "loss": 3.4236, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.7471074380165289, | |
| "grad_norm": 0.7395262718200684, | |
| "learning_rate": 5.184757496945726e-05, | |
| "loss": 3.885, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.7537190082644628, | |
| "grad_norm": 0.7164163589477539, | |
| "learning_rate": 5.092394524795649e-05, | |
| "loss": 3.8037, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.7603305785123967, | |
| "grad_norm": 0.7306711077690125, | |
| "learning_rate": 5e-05, | |
| "loss": 3.8429, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.7669421487603306, | |
| "grad_norm": 0.7196017503738403, | |
| "learning_rate": 4.907605475204352e-05, | |
| "loss": 4.1837, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.7735537190082644, | |
| "grad_norm": 0.7127442359924316, | |
| "learning_rate": 4.8152425030542766e-05, | |
| "loss": 3.8733, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.7801652892561983, | |
| "grad_norm": 0.7182602882385254, | |
| "learning_rate": 4.72294262542015e-05, | |
| "loss": 3.9125, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.7867768595041322, | |
| "grad_norm": 0.7722130417823792, | |
| "learning_rate": 4.6307373626256306e-05, | |
| "loss": 4.1602, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.7933884297520661, | |
| "grad_norm": 0.8073102831840515, | |
| "learning_rate": 4.5386582026834906e-05, | |
| "loss": 4.1581, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.8239835500717163, | |
| "learning_rate": 4.446736590542497e-05, | |
| "loss": 4.1797, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.8066115702479338, | |
| "grad_norm": 0.8800137639045715, | |
| "learning_rate": 4.3550039173489845e-05, | |
| "loss": 4.2168, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.8132231404958677, | |
| "grad_norm": 0.8760910034179688, | |
| "learning_rate": 4.2634915097268115e-05, | |
| "loss": 4.4288, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.8198347107438017, | |
| "grad_norm": 0.9671458601951599, | |
| "learning_rate": 4.1722306190793495e-05, | |
| "loss": 4.2465, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.8264462809917356, | |
| "grad_norm": 0.9589430689811707, | |
| "learning_rate": 4.0812524109171476e-05, | |
| "loss": 4.6314, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.8330578512396695, | |
| "grad_norm": 0.9631523489952087, | |
| "learning_rate": 3.99058795421495e-05, | |
| "loss": 3.8995, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.8396694214876033, | |
| "grad_norm": 1.1293405294418335, | |
| "learning_rate": 3.9002682108016585e-05, | |
| "loss": 4.4208, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.8462809917355372, | |
| "grad_norm": 1.1248273849487305, | |
| "learning_rate": 3.8103240247869075e-05, | |
| "loss": 4.797, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.8528925619834711, | |
| "grad_norm": 1.1502659320831299, | |
| "learning_rate": 3.720786112027822e-05, | |
| "loss": 4.5282, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.859504132231405, | |
| "grad_norm": 1.3296188116073608, | |
| "learning_rate": 3.631685049639586e-05, | |
| "loss": 4.5665, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.8661157024793389, | |
| "grad_norm": 1.222717523574829, | |
| "learning_rate": 3.543051265553377e-05, | |
| "loss": 4.4934, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.8727272727272727, | |
| "grad_norm": 1.3489303588867188, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 4.9627, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.8793388429752066, | |
| "grad_norm": 1.3762129545211792, | |
| "learning_rate": 3.367306435799584e-05, | |
| "loss": 4.9849, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.8859504132231405, | |
| "grad_norm": 1.4506964683532715, | |
| "learning_rate": 3.2802554068303596e-05, | |
| "loss": 4.5126, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.8925619834710744, | |
| "grad_norm": 1.541321039199829, | |
| "learning_rate": 3.1937916690642356e-05, | |
| "loss": 4.6768, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.8991735537190083, | |
| "grad_norm": 1.6167188882827759, | |
| "learning_rate": 3.107944749788449e-05, | |
| "loss": 4.4696, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.9057851239669421, | |
| "grad_norm": 1.765872597694397, | |
| "learning_rate": 3.0227439656472877e-05, | |
| "loss": 4.9427, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.912396694214876, | |
| "grad_norm": 1.81801438331604, | |
| "learning_rate": 2.9382184126304834e-05, | |
| "loss": 5.1971, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.9190082644628099, | |
| "grad_norm": 1.7761266231536865, | |
| "learning_rate": 2.8543969561369556e-05, | |
| "loss": 3.5547, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.9256198347107438, | |
| "grad_norm": 1.9924731254577637, | |
| "learning_rate": 2.771308221117309e-05, | |
| "loss": 5.4103, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.9322314049586777, | |
| "grad_norm": 2.269597291946411, | |
| "learning_rate": 2.688980582298435e-05, | |
| "loss": 5.2679, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.9388429752066115, | |
| "grad_norm": 2.1679582595825195, | |
| "learning_rate": 2.607442154493568e-05, | |
| "loss": 5.2392, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.9454545454545454, | |
| "grad_norm": 2.2235004901885986, | |
| "learning_rate": 2.5267207830011068e-05, | |
| "loss": 5.3694, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.9520661157024793, | |
| "grad_norm": 2.366504430770874, | |
| "learning_rate": 2.446844034095466e-05, | |
| "loss": 4.5467, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.9586776859504132, | |
| "grad_norm": 2.4724178314208984, | |
| "learning_rate": 2.3678391856132204e-05, | |
| "loss": 4.0637, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.9652892561983472, | |
| "grad_norm": 2.8730733394622803, | |
| "learning_rate": 2.2897332176377528e-05, | |
| "loss": 4.465, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.971900826446281, | |
| "grad_norm": 3.2861361503601074, | |
| "learning_rate": 2.2125528032855724e-05, | |
| "loss": 5.3115, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.9785123966942149, | |
| "grad_norm": 3.9318833351135254, | |
| "learning_rate": 2.136324299597474e-05, | |
| "loss": 5.3298, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.9851239669421488, | |
| "grad_norm": 0.7852454781532288, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 3.8637, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.9917355371900827, | |
| "grad_norm": 1.412460446357727, | |
| "learning_rate": 1.9868268181037185e-05, | |
| "loss": 4.5326, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.9917355371900827, | |
| "eval_loss": 1.0575369596481323, | |
| "eval_runtime": 14.1494, | |
| "eval_samples_per_second": 18.022, | |
| "eval_steps_per_second": 9.046, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.9983471074380166, | |
| "grad_norm": 2.1073575019836426, | |
| "learning_rate": 1.9136088935510362e-05, | |
| "loss": 4.4085, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.0049586776859505, | |
| "grad_norm": 2.898435115814209, | |
| "learning_rate": 1.8414449687337464e-05, | |
| "loss": 3.9522, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.0115702479338844, | |
| "grad_norm": 0.6054397821426392, | |
| "learning_rate": 1.7703596875660645e-05, | |
| "loss": 3.8139, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.018181818181818, | |
| "grad_norm": 0.6304721236228943, | |
| "learning_rate": 1.700377325606388e-05, | |
| "loss": 3.5105, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.024793388429752, | |
| "grad_norm": 0.6456805467605591, | |
| "learning_rate": 1.631521781767214e-05, | |
| "loss": 3.4875, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.031404958677686, | |
| "grad_norm": 0.6970878839492798, | |
| "learning_rate": 1.5638165701536868e-05, | |
| "loss": 3.9056, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.0380165289256198, | |
| "grad_norm": 0.7542885541915894, | |
| "learning_rate": 1.4972848120335453e-05, | |
| "loss": 3.6381, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.0446280991735537, | |
| "grad_norm": 0.8014428615570068, | |
| "learning_rate": 1.4319492279412388e-05, | |
| "loss": 4.0426, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.0512396694214876, | |
| "grad_norm": 0.7932935953140259, | |
| "learning_rate": 1.3678321299188801e-05, | |
| "loss": 4.1678, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.0578512396694215, | |
| "grad_norm": 0.8434247970581055, | |
| "learning_rate": 1.3049554138967051e-05, | |
| "loss": 3.845, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.0644628099173554, | |
| "grad_norm": 1.026289701461792, | |
| "learning_rate": 1.2433405522156332e-05, | |
| "loss": 4.0639, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.0710743801652893, | |
| "grad_norm": 0.9154835939407349, | |
| "learning_rate": 1.183008586294485e-05, | |
| "loss": 4.1187, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.0776859504132232, | |
| "grad_norm": 1.0376970767974854, | |
| "learning_rate": 1.1239801194443506e-05, | |
| "loss": 3.7358, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.084297520661157, | |
| "grad_norm": 0.9799293875694275, | |
| "learning_rate": 1.066275309832584e-05, | |
| "loss": 3.7865, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.0909090909090908, | |
| "grad_norm": 1.3073195219039917, | |
| "learning_rate": 1.0099138635988026e-05, | |
| "loss": 3.8564, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.0975206611570247, | |
| "grad_norm": 1.0965800285339355, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 4.1914, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.1041322314049586, | |
| "grad_norm": 1.167807936668396, | |
| "learning_rate": 9.012975854638949e-06, | |
| "loss": 4.4906, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.1107438016528925, | |
| "grad_norm": 1.1979173421859741, | |
| "learning_rate": 8.490798459222476e-06, | |
| "loss": 4.169, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.1173553719008265, | |
| "grad_norm": 1.294745922088623, | |
| "learning_rate": 7.982796418105371e-06, | |
| "loss": 4.7301, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.1239669421487604, | |
| "grad_norm": 1.308195948600769, | |
| "learning_rate": 7.489143213519301e-06, | |
| "loss": 4.0708, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.1305785123966943, | |
| "grad_norm": 1.4223898649215698, | |
| "learning_rate": 7.010007427581378e-06, | |
| "loss": 4.5165, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.1371900826446282, | |
| "grad_norm": 1.3016583919525146, | |
| "learning_rate": 6.5455526847235825e-06, | |
| "loss": 4.0952, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.143801652892562, | |
| "grad_norm": 1.5517874956130981, | |
| "learning_rate": 6.0959375958151045e-06, | |
| "loss": 4.2077, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.1504132231404958, | |
| "grad_norm": 1.4684065580368042, | |
| "learning_rate": 5.6613157039969055e-06, | |
| "loss": 4.1166, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.1570247933884297, | |
| "grad_norm": 1.5590767860412598, | |
| "learning_rate": 5.241835432246889e-06, | |
| "loss": 4.4264, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.1636363636363636, | |
| "grad_norm": 1.6318552494049072, | |
| "learning_rate": 4.837640032693558e-06, | |
| "loss": 4.4621, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.1702479338842975, | |
| "grad_norm": 1.6523213386535645, | |
| "learning_rate": 4.448867537695578e-06, | |
| "loss": 4.2173, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.1768595041322314, | |
| "grad_norm": 1.782486081123352, | |
| "learning_rate": 4.075650712703849e-06, | |
| "loss": 3.9855, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.1834710743801653, | |
| "grad_norm": 2.0244011878967285, | |
| "learning_rate": 3.71811701092219e-06, | |
| "loss": 3.1813, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.1900826446280992, | |
| "grad_norm": 1.8634884357452393, | |
| "learning_rate": 3.376388529782215e-06, | |
| "loss": 4.2433, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.1966942148760331, | |
| "grad_norm": 3.6111536026000977, | |
| "learning_rate": 3.0505819692471792e-06, | |
| "loss": 4.7134, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.203305785123967, | |
| "grad_norm": 2.2132115364074707, | |
| "learning_rate": 2.7408085919590264e-06, | |
| "loss": 5.0016, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.2099173553719007, | |
| "grad_norm": 2.191588878631592, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 4.8401, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.2165289256198348, | |
| "grad_norm": 2.5043983459472656, | |
| "learning_rate": 2.1697790249779636e-06, | |
| "loss": 4.4028, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.2231404958677685, | |
| "grad_norm": 2.4573326110839844, | |
| "learning_rate": 1.908717841359048e-06, | |
| "loss": 3.3899, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.2297520661157024, | |
| "grad_norm": 2.752991199493408, | |
| "learning_rate": 1.6640797865406288e-06, | |
| "loss": 3.9257, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.2363636363636363, | |
| "grad_norm": 3.2671091556549072, | |
| "learning_rate": 1.4359484041943038e-06, | |
| "loss": 4.4646, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.2429752066115702, | |
| "grad_norm": 3.656493663787842, | |
| "learning_rate": 1.2244016009781701e-06, | |
| "loss": 3.8991, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.2495867768595041, | |
| "grad_norm": 2.3350541591644287, | |
| "learning_rate": 1.0295116199317057e-06, | |
| "loss": 3.9637, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.256198347107438, | |
| "grad_norm": 0.6289302706718445, | |
| "learning_rate": 8.513450158049108e-07, | |
| "loss": 3.5885, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.262809917355372, | |
| "grad_norm": 0.6117539405822754, | |
| "learning_rate": 6.899626323298713e-07, | |
| "loss": 2.8164, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.2694214876033059, | |
| "grad_norm": 0.6665360331535339, | |
| "learning_rate": 5.454195814427021e-07, | |
| "loss": 3.8419, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.2760330578512398, | |
| "grad_norm": 0.6839153170585632, | |
| "learning_rate": 4.177652244628627e-07, | |
| "loss": 4.0613, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.2826446280991735, | |
| "grad_norm": 0.8262344598770142, | |
| "learning_rate": 3.0704315523631953e-07, | |
| "loss": 3.8355, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.2892561983471074, | |
| "grad_norm": 0.8175384402275085, | |
| "learning_rate": 2.1329118524827662e-07, | |
| "loss": 4.2571, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.2958677685950413, | |
| "grad_norm": 0.8560000658035278, | |
| "learning_rate": 1.3654133071059893e-07, | |
| "loss": 3.9339, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.3024793388429752, | |
| "grad_norm": 0.8736144304275513, | |
| "learning_rate": 7.681980162830282e-08, | |
| "loss": 4.096, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.309090909090909, | |
| "grad_norm": 0.9289442300796509, | |
| "learning_rate": 3.4146992848854695e-08, | |
| "loss": 3.9761, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.315702479338843, | |
| "grad_norm": 0.9881563186645508, | |
| "learning_rate": 8.537477097364522e-09, | |
| "loss": 3.987, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.322314049586777, | |
| "grad_norm": 1.0067229270935059, | |
| "learning_rate": 0.0, | |
| "loss": 4.4433, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.322314049586777, | |
| "eval_loss": 1.0460100173950195, | |
| "eval_runtime": 14.1299, | |
| "eval_samples_per_second": 18.047, | |
| "eval_steps_per_second": 9.059, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "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": 1.6229100552192e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |