| { | |
| "best_metric": 11.5, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-50", | |
| "epoch": 1.8691588785046729, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.009345794392523364, | |
| "grad_norm": 2.5344741516164504e-05, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 46.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.009345794392523364, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 1.1265, | |
| "eval_samples_per_second": 160.679, | |
| "eval_steps_per_second": 80.783, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.018691588785046728, | |
| "grad_norm": 2.615838275232818e-05, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 46.0, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.028037383177570093, | |
| "grad_norm": 2.7344363843440078e-05, | |
| "learning_rate": 1e-05, | |
| "loss": 46.0, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.037383177570093455, | |
| "grad_norm": 2.452568332955707e-05, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 46.0, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.04672897196261682, | |
| "grad_norm": 3.134661528747529e-05, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 46.0, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.056074766355140186, | |
| "grad_norm": 3.4900876926258206e-05, | |
| "learning_rate": 2e-05, | |
| "loss": 46.0, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.06542056074766354, | |
| "grad_norm": 2.6106856239493936e-05, | |
| "learning_rate": 2.3333333333333336e-05, | |
| "loss": 46.0, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.07476635514018691, | |
| "grad_norm": 2.7752683308790438e-05, | |
| "learning_rate": 2.6666666666666667e-05, | |
| "loss": 46.0, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.08411214953271028, | |
| "grad_norm": 2.7367219445295632e-05, | |
| "learning_rate": 3e-05, | |
| "loss": 46.0, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.09345794392523364, | |
| "grad_norm": 2.9516531867557205e-05, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 46.0, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.102803738317757, | |
| "grad_norm": 3.180565909133293e-05, | |
| "learning_rate": 3.6666666666666666e-05, | |
| "loss": 46.0, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.11214953271028037, | |
| "grad_norm": 2.7306345145916566e-05, | |
| "learning_rate": 4e-05, | |
| "loss": 46.0, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.12149532710280374, | |
| "grad_norm": 3.706616917042993e-05, | |
| "learning_rate": 4.3333333333333334e-05, | |
| "loss": 46.0, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.1308411214953271, | |
| "grad_norm": 2.9705503038712777e-05, | |
| "learning_rate": 4.666666666666667e-05, | |
| "loss": 46.0, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.14018691588785046, | |
| "grad_norm": 3.355026638018899e-05, | |
| "learning_rate": 5e-05, | |
| "loss": 46.0, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.14953271028037382, | |
| "grad_norm": 3.9220551116159186e-05, | |
| "learning_rate": 5.333333333333333e-05, | |
| "loss": 46.0, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.1588785046728972, | |
| "grad_norm": 4.0627914131619036e-05, | |
| "learning_rate": 5.666666666666667e-05, | |
| "loss": 46.0, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.16822429906542055, | |
| "grad_norm": 3.929124432033859e-05, | |
| "learning_rate": 6e-05, | |
| "loss": 46.0, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.17757009345794392, | |
| "grad_norm": 2.4858816686901264e-05, | |
| "learning_rate": 6.333333333333333e-05, | |
| "loss": 46.0, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.18691588785046728, | |
| "grad_norm": 3.943190677091479e-05, | |
| "learning_rate": 6.666666666666667e-05, | |
| "loss": 46.0, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.19626168224299065, | |
| "grad_norm": 4.8376921768067405e-05, | |
| "learning_rate": 7e-05, | |
| "loss": 46.0, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.205607476635514, | |
| "grad_norm": 4.936111145070754e-05, | |
| "learning_rate": 7.333333333333333e-05, | |
| "loss": 46.0, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.21495327102803738, | |
| "grad_norm": 5.0592880143085495e-05, | |
| "learning_rate": 7.666666666666667e-05, | |
| "loss": 46.0, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.22429906542056074, | |
| "grad_norm": 4.4427615648601204e-05, | |
| "learning_rate": 8e-05, | |
| "loss": 46.0, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.2336448598130841, | |
| "grad_norm": 5.0494058086769655e-05, | |
| "learning_rate": 8.333333333333334e-05, | |
| "loss": 46.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.24299065420560748, | |
| "grad_norm": 3.8928865251364186e-05, | |
| "learning_rate": 8.666666666666667e-05, | |
| "loss": 46.0, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.2523364485981308, | |
| "grad_norm": 4.032924334751442e-05, | |
| "learning_rate": 9e-05, | |
| "loss": 46.0, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.2616822429906542, | |
| "grad_norm": 4.888198964181356e-05, | |
| "learning_rate": 9.333333333333334e-05, | |
| "loss": 46.0, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.27102803738317754, | |
| "grad_norm": 3.859275238937698e-05, | |
| "learning_rate": 9.666666666666667e-05, | |
| "loss": 46.0, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.2803738317757009, | |
| "grad_norm": 4.155114584136754e-05, | |
| "learning_rate": 0.0001, | |
| "loss": 46.0, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.2897196261682243, | |
| "grad_norm": 3.540813486324623e-05, | |
| "learning_rate": 9.999146252290264e-05, | |
| "loss": 46.0, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.29906542056074764, | |
| "grad_norm": 4.2967913032043725e-05, | |
| "learning_rate": 9.996585300715116e-05, | |
| "loss": 46.0, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.308411214953271, | |
| "grad_norm": 4.8990899813361466e-05, | |
| "learning_rate": 9.99231801983717e-05, | |
| "loss": 46.0, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.3177570093457944, | |
| "grad_norm": 4.408402310218662e-05, | |
| "learning_rate": 9.986345866928941e-05, | |
| "loss": 46.0, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.32710280373831774, | |
| "grad_norm": 4.686412648879923e-05, | |
| "learning_rate": 9.978670881475172e-05, | |
| "loss": 46.0, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.3364485981308411, | |
| "grad_norm": 5.995184983476065e-05, | |
| "learning_rate": 9.96929568447637e-05, | |
| "loss": 46.0, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.34579439252336447, | |
| "grad_norm": 6.438123091356829e-05, | |
| "learning_rate": 9.958223477553714e-05, | |
| "loss": 46.0, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.35514018691588783, | |
| "grad_norm": 6.780792318750173e-05, | |
| "learning_rate": 9.94545804185573e-05, | |
| "loss": 46.0, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.3644859813084112, | |
| "grad_norm": 6.843348819529638e-05, | |
| "learning_rate": 9.931003736767013e-05, | |
| "loss": 46.0, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.37383177570093457, | |
| "grad_norm": 0.00011038873344659805, | |
| "learning_rate": 9.91486549841951e-05, | |
| "loss": 46.0, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.38317757009345793, | |
| "grad_norm": 5.5483953474322334e-05, | |
| "learning_rate": 9.89704883800683e-05, | |
| "loss": 46.0, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.3925233644859813, | |
| "grad_norm": 8.126490865834057e-05, | |
| "learning_rate": 9.877559839902184e-05, | |
| "loss": 46.0, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.40186915887850466, | |
| "grad_norm": 6.556714652106166e-05, | |
| "learning_rate": 9.85640515958057e-05, | |
| "loss": 46.0, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.411214953271028, | |
| "grad_norm": 5.473397686728276e-05, | |
| "learning_rate": 9.833592021345937e-05, | |
| "loss": 46.0, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.4205607476635514, | |
| "grad_norm": 0.00011574680684134364, | |
| "learning_rate": 9.809128215864097e-05, | |
| "loss": 46.0, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.42990654205607476, | |
| "grad_norm": 9.40759273362346e-05, | |
| "learning_rate": 9.783022097502204e-05, | |
| "loss": 46.0, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.4392523364485981, | |
| "grad_norm": 0.00010287982877343893, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 46.0, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.4485981308411215, | |
| "grad_norm": 0.00010786245547933504, | |
| "learning_rate": 9.725919140804099e-05, | |
| "loss": 46.0, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.45794392523364486, | |
| "grad_norm": 0.00014578491391148418, | |
| "learning_rate": 9.694941803075283e-05, | |
| "loss": 46.0, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.4672897196261682, | |
| "grad_norm": 0.00013963892706669867, | |
| "learning_rate": 9.662361147021779e-05, | |
| "loss": 46.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.4672897196261682, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 1.1298, | |
| "eval_samples_per_second": 160.208, | |
| "eval_steps_per_second": 80.547, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.4766355140186916, | |
| "grad_norm": 9.963639604393393e-05, | |
| "learning_rate": 9.628188298907782e-05, | |
| "loss": 46.0, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.48598130841121495, | |
| "grad_norm": 0.00012790461187250912, | |
| "learning_rate": 9.592434928729616e-05, | |
| "loss": 46.0, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.4953271028037383, | |
| "grad_norm": 9.55718569457531e-05, | |
| "learning_rate": 9.555113246230442e-05, | |
| "loss": 46.0, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.5046728971962616, | |
| "grad_norm": 0.00014584721066057682, | |
| "learning_rate": 9.516235996730645e-05, | |
| "loss": 46.0, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.514018691588785, | |
| "grad_norm": 0.00011913034541066736, | |
| "learning_rate": 9.475816456775313e-05, | |
| "loss": 46.0, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.5233644859813084, | |
| "grad_norm": 9.888441854855046e-05, | |
| "learning_rate": 9.43386842960031e-05, | |
| "loss": 46.0, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.5327102803738317, | |
| "grad_norm": 0.00010833790292963386, | |
| "learning_rate": 9.39040624041849e-05, | |
| "loss": 46.0, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.5420560747663551, | |
| "grad_norm": 0.00012922610039822757, | |
| "learning_rate": 9.345444731527642e-05, | |
| "loss": 46.0, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.5514018691588785, | |
| "grad_norm": 0.0001300368458032608, | |
| "learning_rate": 9.298999257241863e-05, | |
| "loss": 46.0, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.5607476635514018, | |
| "grad_norm": 8.210381929529831e-05, | |
| "learning_rate": 9.251085678648072e-05, | |
| "loss": 46.0, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.5700934579439252, | |
| "grad_norm": 0.0001599452516529709, | |
| "learning_rate": 9.201720358189464e-05, | |
| "loss": 46.0, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.5794392523364486, | |
| "grad_norm": 0.0001269468484679237, | |
| "learning_rate": 9.150920154077754e-05, | |
| "loss": 46.0, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.5887850467289719, | |
| "grad_norm": 0.0001564062258694321, | |
| "learning_rate": 9.098702414536107e-05, | |
| "loss": 46.0, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.5981308411214953, | |
| "grad_norm": 0.00021565529459621757, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 46.0, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.6074766355140186, | |
| "grad_norm": 0.00012543653429020196, | |
| "learning_rate": 8.9900861364012e-05, | |
| "loss": 46.0, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.616822429906542, | |
| "grad_norm": 0.00016946744290180504, | |
| "learning_rate": 8.933724690167417e-05, | |
| "loss": 46.0, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.6261682242990654, | |
| "grad_norm": 0.00016576446068938822, | |
| "learning_rate": 8.876019880555649e-05, | |
| "loss": 46.0, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.6355140186915887, | |
| "grad_norm": 0.00015629635890945792, | |
| "learning_rate": 8.816991413705516e-05, | |
| "loss": 46.0, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.6448598130841121, | |
| "grad_norm": 0.00017187913181260228, | |
| "learning_rate": 8.756659447784368e-05, | |
| "loss": 46.0, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.6542056074766355, | |
| "grad_norm": 0.00017371830472256988, | |
| "learning_rate": 8.695044586103296e-05, | |
| "loss": 46.0, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.6635514018691588, | |
| "grad_norm": 0.00018266221741214395, | |
| "learning_rate": 8.632167870081121e-05, | |
| "loss": 46.0, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.6728971962616822, | |
| "grad_norm": 0.00013964249228592962, | |
| "learning_rate": 8.568050772058762e-05, | |
| "loss": 46.0, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.6822429906542056, | |
| "grad_norm": 0.00014923812705092132, | |
| "learning_rate": 8.502715187966455e-05, | |
| "loss": 46.0, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.6915887850467289, | |
| "grad_norm": 0.00016153555770870298, | |
| "learning_rate": 8.436183429846313e-05, | |
| "loss": 46.0, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.7009345794392523, | |
| "grad_norm": 0.00014358038606587797, | |
| "learning_rate": 8.368478218232787e-05, | |
| "loss": 46.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.7102803738317757, | |
| "grad_norm": 0.00019736040849238634, | |
| "learning_rate": 8.299622674393614e-05, | |
| "loss": 46.0, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.719626168224299, | |
| "grad_norm": 0.0003134732542093843, | |
| "learning_rate": 8.229640312433937e-05, | |
| "loss": 46.0, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.7289719626168224, | |
| "grad_norm": 0.0002348859270568937, | |
| "learning_rate": 8.158555031266254e-05, | |
| "loss": 46.0, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.7383177570093458, | |
| "grad_norm": 0.00017541946726851165, | |
| "learning_rate": 8.086391106448965e-05, | |
| "loss": 46.0, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.7476635514018691, | |
| "grad_norm": 0.00016056875756476074, | |
| "learning_rate": 8.013173181896283e-05, | |
| "loss": 46.0, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.7570093457943925, | |
| "grad_norm": 0.00022755710233468562, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 46.0, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.7663551401869159, | |
| "grad_norm": 0.0002692342677619308, | |
| "learning_rate": 7.863675700402526e-05, | |
| "loss": 46.0, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.7757009345794392, | |
| "grad_norm": 0.0002028964809142053, | |
| "learning_rate": 7.787447196714427e-05, | |
| "loss": 46.0, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.7850467289719626, | |
| "grad_norm": 0.000286884926026687, | |
| "learning_rate": 7.710266782362247e-05, | |
| "loss": 46.0, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.794392523364486, | |
| "grad_norm": 0.00020119328110013157, | |
| "learning_rate": 7.63216081438678e-05, | |
| "loss": 46.0, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.8037383177570093, | |
| "grad_norm": 0.00034160970244556665, | |
| "learning_rate": 7.553155965904535e-05, | |
| "loss": 46.0, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.8130841121495327, | |
| "grad_norm": 0.0002135804679710418, | |
| "learning_rate": 7.473279216998895e-05, | |
| "loss": 46.0, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.822429906542056, | |
| "grad_norm": 0.00017743604257702827, | |
| "learning_rate": 7.392557845506432e-05, | |
| "loss": 46.0, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.8317757009345794, | |
| "grad_norm": 0.0001699524582363665, | |
| "learning_rate": 7.311019417701566e-05, | |
| "loss": 46.0, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.8411214953271028, | |
| "grad_norm": 0.00032696506241336465, | |
| "learning_rate": 7.228691778882693e-05, | |
| "loss": 46.0, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.8504672897196262, | |
| "grad_norm": 0.00027825115830637515, | |
| "learning_rate": 7.145603043863045e-05, | |
| "loss": 46.0, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.8598130841121495, | |
| "grad_norm": 0.0001989601005334407, | |
| "learning_rate": 7.061781587369519e-05, | |
| "loss": 46.0, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.8691588785046729, | |
| "grad_norm": 0.0003293946501798928, | |
| "learning_rate": 6.977256034352712e-05, | |
| "loss": 46.0, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.8785046728971962, | |
| "grad_norm": 0.00025171003653667867, | |
| "learning_rate": 6.892055250211552e-05, | |
| "loss": 46.0, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.8878504672897196, | |
| "grad_norm": 0.00037094487925060093, | |
| "learning_rate": 6.806208330935766e-05, | |
| "loss": 46.0, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.897196261682243, | |
| "grad_norm": 0.00023440983204636723, | |
| "learning_rate": 6.719744593169641e-05, | |
| "loss": 46.0, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.9065420560747663, | |
| "grad_norm": 0.0004414333088789135, | |
| "learning_rate": 6.632693564200416e-05, | |
| "loss": 46.0, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.9158878504672897, | |
| "grad_norm": 0.00024191121337935328, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 46.0, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.9252336448598131, | |
| "grad_norm": 0.00048082045395858586, | |
| "learning_rate": 6.456948734446624e-05, | |
| "loss": 46.0, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.9345794392523364, | |
| "grad_norm": 0.00038313641562126577, | |
| "learning_rate": 6.368314950360415e-05, | |
| "loss": 46.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.9345794392523364, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 1.1284, | |
| "eval_samples_per_second": 160.407, | |
| "eval_steps_per_second": 80.647, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.9439252336448598, | |
| "grad_norm": 0.00041886104736477137, | |
| "learning_rate": 6.279213887972179e-05, | |
| "loss": 46.0, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.9532710280373832, | |
| "grad_norm": 0.00037539354525506496, | |
| "learning_rate": 6.189675975213094e-05, | |
| "loss": 46.0, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.9626168224299065, | |
| "grad_norm": 0.0005437719519250095, | |
| "learning_rate": 6.099731789198344e-05, | |
| "loss": 46.0, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.9719626168224299, | |
| "grad_norm": 0.0002524543961044401, | |
| "learning_rate": 6.009412045785051e-05, | |
| "loss": 46.0, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.9813084112149533, | |
| "grad_norm": 0.00037665668060071766, | |
| "learning_rate": 5.918747589082853e-05, | |
| "loss": 46.0, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.9906542056074766, | |
| "grad_norm": 0.0006277555949054658, | |
| "learning_rate": 5.82776938092065e-05, | |
| "loss": 46.0, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.0004002438799943775, | |
| "learning_rate": 5.736508490273188e-05, | |
| "loss": 46.0, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.0093457943925233, | |
| "grad_norm": 0.00047434697626158595, | |
| "learning_rate": 5.644996082651017e-05, | |
| "loss": 46.0, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.0186915887850467, | |
| "grad_norm": 0.0003720087406691164, | |
| "learning_rate": 5.553263409457504e-05, | |
| "loss": 46.0, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.02803738317757, | |
| "grad_norm": 0.00036402122350409627, | |
| "learning_rate": 5.4613417973165106e-05, | |
| "loss": 46.0, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.0373831775700935, | |
| "grad_norm": 0.0003647180856205523, | |
| "learning_rate": 5.3692626373743706e-05, | |
| "loss": 46.0, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.0467289719626167, | |
| "grad_norm": 0.0003529293753672391, | |
| "learning_rate": 5.27705737457985e-05, | |
| "loss": 46.0, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.0560747663551402, | |
| "grad_norm": 0.00038978244992904365, | |
| "learning_rate": 5.184757496945726e-05, | |
| "loss": 46.0, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.0654205607476634, | |
| "grad_norm": 0.00036455257213674486, | |
| "learning_rate": 5.092394524795649e-05, | |
| "loss": 46.0, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.074766355140187, | |
| "grad_norm": 0.0003454765537753701, | |
| "learning_rate": 5e-05, | |
| "loss": 46.0, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.0841121495327102, | |
| "grad_norm": 0.000328146357787773, | |
| "learning_rate": 4.907605475204352e-05, | |
| "loss": 46.0, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.0934579439252337, | |
| "grad_norm": 0.00017895177006721497, | |
| "learning_rate": 4.8152425030542766e-05, | |
| "loss": 46.0, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.102803738317757, | |
| "grad_norm": 0.000497806235216558, | |
| "learning_rate": 4.72294262542015e-05, | |
| "loss": 46.0, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.1121495327102804, | |
| "grad_norm": 0.000430753396358341, | |
| "learning_rate": 4.6307373626256306e-05, | |
| "loss": 46.0, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.1214953271028036, | |
| "grad_norm": 0.00042813917389139533, | |
| "learning_rate": 4.5386582026834906e-05, | |
| "loss": 46.0, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.1308411214953271, | |
| "grad_norm": 0.000581981847062707, | |
| "learning_rate": 4.446736590542497e-05, | |
| "loss": 46.0, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.1401869158878504, | |
| "grad_norm": 0.00040982503560371697, | |
| "learning_rate": 4.3550039173489845e-05, | |
| "loss": 46.0, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.1495327102803738, | |
| "grad_norm": 0.0004324937763158232, | |
| "learning_rate": 4.2634915097268115e-05, | |
| "loss": 46.0, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.158878504672897, | |
| "grad_norm": 0.0006235907203517854, | |
| "learning_rate": 4.1722306190793495e-05, | |
| "loss": 46.0, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.1682242990654206, | |
| "grad_norm": 0.000507604272570461, | |
| "learning_rate": 4.0812524109171476e-05, | |
| "loss": 46.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.1775700934579438, | |
| "grad_norm": 0.0005539056146517396, | |
| "learning_rate": 3.99058795421495e-05, | |
| "loss": 46.0, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.1869158878504673, | |
| "grad_norm": 0.0007270352798514068, | |
| "learning_rate": 3.9002682108016585e-05, | |
| "loss": 46.0, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.1962616822429906, | |
| "grad_norm": 0.0006474956753663719, | |
| "learning_rate": 3.8103240247869075e-05, | |
| "loss": 46.0, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.205607476635514, | |
| "grad_norm": 0.00030030321795493364, | |
| "learning_rate": 3.720786112027822e-05, | |
| "loss": 46.0, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.2149532710280373, | |
| "grad_norm": 0.0006375272641889751, | |
| "learning_rate": 3.631685049639586e-05, | |
| "loss": 46.0, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.2242990654205608, | |
| "grad_norm": 0.0006298018852248788, | |
| "learning_rate": 3.543051265553377e-05, | |
| "loss": 46.0, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.233644859813084, | |
| "grad_norm": 0.000947554421145469, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 46.0, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.2429906542056075, | |
| "grad_norm": 0.0003460047300904989, | |
| "learning_rate": 3.367306435799584e-05, | |
| "loss": 46.0, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.2523364485981308, | |
| "grad_norm": 0.00037807237822562456, | |
| "learning_rate": 3.2802554068303596e-05, | |
| "loss": 46.0, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.2616822429906542, | |
| "grad_norm": 0.00040374876698479056, | |
| "learning_rate": 3.1937916690642356e-05, | |
| "loss": 46.0, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.2710280373831775, | |
| "grad_norm": 0.000543224741704762, | |
| "learning_rate": 3.107944749788449e-05, | |
| "loss": 46.0, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.280373831775701, | |
| "grad_norm": 0.000598748738411814, | |
| "learning_rate": 3.0227439656472877e-05, | |
| "loss": 46.0, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.2897196261682242, | |
| "grad_norm": 0.0002985202008858323, | |
| "learning_rate": 2.9382184126304834e-05, | |
| "loss": 46.0, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.2990654205607477, | |
| "grad_norm": 0.00044871296267956495, | |
| "learning_rate": 2.8543969561369556e-05, | |
| "loss": 46.0, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 1.308411214953271, | |
| "grad_norm": 0.000572655932046473, | |
| "learning_rate": 2.771308221117309e-05, | |
| "loss": 46.0, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.3177570093457944, | |
| "grad_norm": 0.000479130627354607, | |
| "learning_rate": 2.688980582298435e-05, | |
| "loss": 46.0, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 1.3271028037383177, | |
| "grad_norm": 0.00036988427746109664, | |
| "learning_rate": 2.607442154493568e-05, | |
| "loss": 46.0, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.3364485981308412, | |
| "grad_norm": 0.00034764522570185363, | |
| "learning_rate": 2.5267207830011068e-05, | |
| "loss": 46.0, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.3457943925233644, | |
| "grad_norm": 0.0005145801696926355, | |
| "learning_rate": 2.446844034095466e-05, | |
| "loss": 46.0, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.355140186915888, | |
| "grad_norm": 0.00038314744597300887, | |
| "learning_rate": 2.3678391856132204e-05, | |
| "loss": 46.0, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.3644859813084111, | |
| "grad_norm": 0.0006724435952492058, | |
| "learning_rate": 2.2897332176377528e-05, | |
| "loss": 46.0, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.3738317757009346, | |
| "grad_norm": 0.0005267965025268495, | |
| "learning_rate": 2.2125528032855724e-05, | |
| "loss": 46.0, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.3831775700934579, | |
| "grad_norm": 0.0008383267559111118, | |
| "learning_rate": 2.136324299597474e-05, | |
| "loss": 46.0, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.3925233644859814, | |
| "grad_norm": 0.0005008934531360865, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 46.0, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.4018691588785046, | |
| "grad_norm": 0.0004917852929793298, | |
| "learning_rate": 1.9868268181037185e-05, | |
| "loss": 46.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.4018691588785046, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 1.1309, | |
| "eval_samples_per_second": 160.044, | |
| "eval_steps_per_second": 80.464, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.411214953271028, | |
| "grad_norm": 0.00040427924250252545, | |
| "learning_rate": 1.9136088935510362e-05, | |
| "loss": 46.0, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.4205607476635513, | |
| "grad_norm": 0.000727128644939512, | |
| "learning_rate": 1.8414449687337464e-05, | |
| "loss": 46.0, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.4299065420560748, | |
| "grad_norm": 0.0006024191388860345, | |
| "learning_rate": 1.7703596875660645e-05, | |
| "loss": 46.0, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.439252336448598, | |
| "grad_norm": 0.00046983369975350797, | |
| "learning_rate": 1.700377325606388e-05, | |
| "loss": 46.0, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.4485981308411215, | |
| "grad_norm": 0.0007036470924504101, | |
| "learning_rate": 1.631521781767214e-05, | |
| "loss": 46.0, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.4579439252336448, | |
| "grad_norm": 0.0006356867379508913, | |
| "learning_rate": 1.5638165701536868e-05, | |
| "loss": 46.0, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.4672897196261683, | |
| "grad_norm": 0.00041911471635103226, | |
| "learning_rate": 1.4972848120335453e-05, | |
| "loss": 46.0, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.4766355140186915, | |
| "grad_norm": 0.0004606099391821772, | |
| "learning_rate": 1.4319492279412388e-05, | |
| "loss": 46.0, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.485981308411215, | |
| "grad_norm": 0.0005386231350712478, | |
| "learning_rate": 1.3678321299188801e-05, | |
| "loss": 46.0, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.4953271028037383, | |
| "grad_norm": 0.0004815705760847777, | |
| "learning_rate": 1.3049554138967051e-05, | |
| "loss": 46.0, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.5046728971962615, | |
| "grad_norm": 0.000545655726455152, | |
| "learning_rate": 1.2433405522156332e-05, | |
| "loss": 46.0, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.514018691588785, | |
| "grad_norm": 0.00040522534982301295, | |
| "learning_rate": 1.183008586294485e-05, | |
| "loss": 46.0, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.5233644859813085, | |
| "grad_norm": 0.0005212983232922852, | |
| "learning_rate": 1.1239801194443506e-05, | |
| "loss": 46.0, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.5327102803738317, | |
| "grad_norm": 0.0006503637996502221, | |
| "learning_rate": 1.066275309832584e-05, | |
| "loss": 46.0, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.542056074766355, | |
| "grad_norm": 0.0006243106327019632, | |
| "learning_rate": 1.0099138635988026e-05, | |
| "loss": 46.0, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.5514018691588785, | |
| "grad_norm": 0.0005656758439727128, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 46.0, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.560747663551402, | |
| "grad_norm": 0.000596692378167063, | |
| "learning_rate": 9.012975854638949e-06, | |
| "loss": 46.0, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.5700934579439252, | |
| "grad_norm": 0.0006169018452055752, | |
| "learning_rate": 8.490798459222476e-06, | |
| "loss": 46.0, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.5794392523364484, | |
| "grad_norm": 0.00040590198477730155, | |
| "learning_rate": 7.982796418105371e-06, | |
| "loss": 46.0, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.588785046728972, | |
| "grad_norm": 0.000642999482806772, | |
| "learning_rate": 7.489143213519301e-06, | |
| "loss": 46.0, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.5981308411214954, | |
| "grad_norm": 0.00047177704982459545, | |
| "learning_rate": 7.010007427581378e-06, | |
| "loss": 46.0, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.6074766355140186, | |
| "grad_norm": 0.0006852488149888813, | |
| "learning_rate": 6.5455526847235825e-06, | |
| "loss": 46.0, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.616822429906542, | |
| "grad_norm": 0.0005702375201508403, | |
| "learning_rate": 6.0959375958151045e-06, | |
| "loss": 46.0, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.6261682242990654, | |
| "grad_norm": 0.0003300758544355631, | |
| "learning_rate": 5.6613157039969055e-06, | |
| "loss": 46.0, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.6355140186915889, | |
| "grad_norm": 0.0005260781035758555, | |
| "learning_rate": 5.241835432246889e-06, | |
| "loss": 46.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.644859813084112, | |
| "grad_norm": 0.0005789488204754889, | |
| "learning_rate": 4.837640032693558e-06, | |
| "loss": 46.0, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.6542056074766354, | |
| "grad_norm": 0.0008496741647832096, | |
| "learning_rate": 4.448867537695578e-06, | |
| "loss": 46.0, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.6635514018691588, | |
| "grad_norm": 0.0009718951187096536, | |
| "learning_rate": 4.075650712703849e-06, | |
| "loss": 46.0, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.6728971962616823, | |
| "grad_norm": 0.00039575292612425983, | |
| "learning_rate": 3.71811701092219e-06, | |
| "loss": 46.0, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.6822429906542056, | |
| "grad_norm": 0.00040399827412329614, | |
| "learning_rate": 3.376388529782215e-06, | |
| "loss": 46.0, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.6915887850467288, | |
| "grad_norm": 0.0006307908333837986, | |
| "learning_rate": 3.0505819692471792e-06, | |
| "loss": 46.0, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.7009345794392523, | |
| "grad_norm": 0.0008413252071477473, | |
| "learning_rate": 2.7408085919590264e-06, | |
| "loss": 46.0, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.7102803738317758, | |
| "grad_norm": 0.0006041501183062792, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 46.0, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.719626168224299, | |
| "grad_norm": 0.0006907913484610617, | |
| "learning_rate": 2.1697790249779636e-06, | |
| "loss": 46.0, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.7289719626168223, | |
| "grad_norm": 0.0008346527465619147, | |
| "learning_rate": 1.908717841359048e-06, | |
| "loss": 46.0, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.7383177570093458, | |
| "grad_norm": 0.00033760842052288353, | |
| "learning_rate": 1.6640797865406288e-06, | |
| "loss": 46.0, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.7476635514018692, | |
| "grad_norm": 0.000893876887857914, | |
| "learning_rate": 1.4359484041943038e-06, | |
| "loss": 46.0, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.7570093457943925, | |
| "grad_norm": 0.000611237483099103, | |
| "learning_rate": 1.2244016009781701e-06, | |
| "loss": 46.0, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.7663551401869158, | |
| "grad_norm": 0.0004157586663495749, | |
| "learning_rate": 1.0295116199317057e-06, | |
| "loss": 46.0, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.7757009345794392, | |
| "grad_norm": 0.00046051564277149737, | |
| "learning_rate": 8.513450158049108e-07, | |
| "loss": 46.0, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.7850467289719627, | |
| "grad_norm": 0.0004915129975415766, | |
| "learning_rate": 6.899626323298713e-07, | |
| "loss": 46.0, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.794392523364486, | |
| "grad_norm": 0.0005064051947556436, | |
| "learning_rate": 5.454195814427021e-07, | |
| "loss": 46.0, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.8037383177570092, | |
| "grad_norm": 0.0005573919625021517, | |
| "learning_rate": 4.177652244628627e-07, | |
| "loss": 46.0, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.8130841121495327, | |
| "grad_norm": 0.0006023433525115252, | |
| "learning_rate": 3.0704315523631953e-07, | |
| "loss": 46.0, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.8224299065420562, | |
| "grad_norm": 0.0005069277249276638, | |
| "learning_rate": 2.1329118524827662e-07, | |
| "loss": 46.0, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.8317757009345794, | |
| "grad_norm": 0.0007788498187437654, | |
| "learning_rate": 1.3654133071059893e-07, | |
| "loss": 46.0, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.8411214953271027, | |
| "grad_norm": 0.0005983134033158422, | |
| "learning_rate": 7.681980162830282e-08, | |
| "loss": 46.0, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.8504672897196262, | |
| "grad_norm": 0.0005851113237440586, | |
| "learning_rate": 3.4146992848854695e-08, | |
| "loss": 46.0, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.8598130841121496, | |
| "grad_norm": 0.0003598025650717318, | |
| "learning_rate": 8.537477097364522e-09, | |
| "loss": 46.0, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.8691588785046729, | |
| "grad_norm": 0.0006664404063485563, | |
| "learning_rate": 0.0, | |
| "loss": 46.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.8691588785046729, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 1.1238, | |
| "eval_samples_per_second": 161.063, | |
| "eval_steps_per_second": 80.977, | |
| "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": 3 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 32158462771200.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |