|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 1.0, |
|
"global_step": 1229, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 5.405405405405406e-05, |
|
"loss": 1.1849, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 9.99984371133188e-05, |
|
"loss": 1.049, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.990816464160552e-05, |
|
"loss": 1.0047, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.967925545267647e-05, |
|
"loss": 1.0005, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.931234541881273e-05, |
|
"loss": 0.9733, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.880845375606788e-05, |
|
"loss": 0.9806, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.81689801930526e-05, |
|
"loss": 0.9689, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.739570108271755e-05, |
|
"loss": 0.9475, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.649076446793543e-05, |
|
"loss": 0.9538, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.545668411458928e-05, |
|
"loss": 0.946, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.42963325287421e-05, |
|
"loss": 1.2086, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.301293297728486e-05, |
|
"loss": 0.956, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.161005053422832e-05, |
|
"loss": 0.9414, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.009158217751071e-05, |
|
"loss": 0.9282, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 8.846174596383044e-05, |
|
"loss": 0.9125, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 8.672506931157457e-05, |
|
"loss": 0.9019, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 8.488637642439121e-05, |
|
"loss": 0.9036, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 8.295077489034092e-05, |
|
"loss": 0.8963, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 8.092364149385254e-05, |
|
"loss": 0.8834, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 7.881060727989571e-05, |
|
"loss": 0.8846, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 7.661754191185898e-05, |
|
"loss": 0.8722, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 7.435053736658517e-05, |
|
"loss": 0.8661, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 7.201589101185622e-05, |
|
"loss": 0.8648, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 6.962008811333567e-05, |
|
"loss": 0.8414, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 6.716978381956145e-05, |
|
"loss": 0.8499, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 6.467178467503203e-05, |
|
"loss": 0.8432, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 6.213302971273911e-05, |
|
"loss": 0.8425, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 5.956057117866922e-05, |
|
"loss": 0.8231, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 5.696155494181792e-05, |
|
"loss": 0.8331, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 5.434320064413464e-05, |
|
"loss": 0.8198, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 5.1712781645538485e-05, |
|
"loss": 0.8124, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 4.9077604819714065e-05, |
|
"loss": 0.8078, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 4.644499025681171e-05, |
|
"loss": 0.816, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 4.382225092943455e-05, |
|
"loss": 0.7996, |
|
"step": 680 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 4.121667237839683e-05, |
|
"loss": 0.8034, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 3.863549247468331e-05, |
|
"loss": 0.7956, |
|
"step": 720 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 3.6085881313827754e-05, |
|
"loss": 0.7855, |
|
"step": 740 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 3.357492129856037e-05, |
|
"loss": 0.7896, |
|
"step": 760 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 3.110958746505165e-05, |
|
"loss": 0.7968, |
|
"step": 780 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 2.8696728107402794e-05, |
|
"loss": 0.7792, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 2.6343045754205053e-05, |
|
"loss": 0.7691, |
|
"step": 820 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 2.4055078550011774e-05, |
|
"loss": 0.7663, |
|
"step": 840 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 2.183918209344215e-05, |
|
"loss": 0.7622, |
|
"step": 860 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 1.970151178236821e-05, |
|
"loss": 0.7635, |
|
"step": 880 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 1.764800571522622e-05, |
|
"loss": 0.7565, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 1.568436819595077e-05, |
|
"loss": 0.7531, |
|
"step": 920 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 1.3816053888351405e-05, |
|
"loss": 0.7566, |
|
"step": 940 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 1.2048252663949039e-05, |
|
"loss": 0.7593, |
|
"step": 960 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 1.038587518536172e-05, |
|
"loss": 0.7548, |
|
"step": 980 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 8.833539265287217e-06, |
|
"loss": 0.7468, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 7.395557038974749e-06, |
|
"loss": 0.7507, |
|
"step": 1020 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 6.075922985818683e-06, |
|
"loss": 0.7552, |
|
"step": 1040 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 4.8783028333481525e-06, |
|
"loss": 0.7534, |
|
"step": 1060 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 3.8060233744356633e-06, |
|
"loss": 0.7528, |
|
"step": 1080 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 2.8620632260107195e-06, |
|
"loss": 0.7431, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 2.0490445549492197e-06, |
|
"loss": 0.7524, |
|
"step": 1120 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 1.3692257941228193e-06, |
|
"loss": 0.7412, |
|
"step": 1140 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 8.244953688418822e-07, |
|
"loss": 0.7275, |
|
"step": 1160 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 4.1636645111886583e-07, |
|
"loss": 0.7526, |
|
"step": 1180 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 1.4597275632398655e-07, |
|
"loss": 0.7362, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 1.4065393909407087e-08, |
|
"loss": 0.7286, |
|
"step": 1220 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"step": 1229, |
|
"total_flos": 257039699804160.0, |
|
"train_loss": 0.845892040975895, |
|
"train_runtime": 30624.5115, |
|
"train_samples_per_second": 2.567, |
|
"train_steps_per_second": 0.04 |
|
} |
|
], |
|
"max_steps": 1229, |
|
"num_train_epochs": 1, |
|
"total_flos": 257039699804160.0, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|