| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.04799472058073612, | |
| "eval_steps": 500, | |
| "global_step": 300, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9999964908081455e-05, | |
| "loss": 0.7285, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999985963242432e-05, | |
| "loss": 0.6712, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999968417332415e-05, | |
| "loss": 0.6081, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999943853127351e-05, | |
| "loss": 0.6383, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999912270696202e-05, | |
| "loss": 0.6456, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9998736701276295e-05, | |
| "loss": 0.6228, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.99982805153e-05, | |
| "loss": 0.6134, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9997754150313815e-05, | |
| "loss": 0.5975, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.999715760779541e-05, | |
| "loss": 0.6053, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9996490889419514e-05, | |
| "loss": 0.6064, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.999575399705783e-05, | |
| "loss": 0.5947, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.999494693277907e-05, | |
| "loss": 0.5839, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.999406969884897e-05, | |
| "loss": 0.6106, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.999312229773022e-05, | |
| "loss": 0.6146, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.99921047320825e-05, | |
| "loss": 0.5659, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9991017004762496e-05, | |
| "loss": 0.5682, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.998985911882384e-05, | |
| "loss": 0.6352, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.998863107751711e-05, | |
| "loss": 0.6018, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.998733288428987e-05, | |
| "loss": 0.6342, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9985964542786614e-05, | |
| "loss": 0.5886, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.998452605684874e-05, | |
| "loss": 0.6097, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.998301743051459e-05, | |
| "loss": 0.5687, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.998143866801942e-05, | |
| "loss": 0.5866, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.997978977379536e-05, | |
| "loss": 0.5612, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.997807075247146e-05, | |
| "loss": 0.6221, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.997628160887361e-05, | |
| "loss": 0.5728, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.997442234802456e-05, | |
| "loss": 0.6105, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.997249297514394e-05, | |
| "loss": 0.6161, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.997049349564814e-05, | |
| "loss": 0.6511, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.996842391515044e-05, | |
| "loss": 0.6108, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.996628423946087e-05, | |
| "loss": 0.5664, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.996407447458626e-05, | |
| "loss": 0.6127, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.99617946267302e-05, | |
| "loss": 0.6213, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.995944470229302e-05, | |
| "loss": 0.596, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9957024707871806e-05, | |
| "loss": 0.5731, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.995453465026032e-05, | |
| "loss": 0.5704, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.995197453644905e-05, | |
| "loss": 0.5767, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.994934437362513e-05, | |
| "loss": 0.6134, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9946644169172355e-05, | |
| "loss": 0.5919, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.994387393067117e-05, | |
| "loss": 0.6031, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.994103366589859e-05, | |
| "loss": 0.6236, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.993812338282826e-05, | |
| "loss": 0.6307, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.993514308963036e-05, | |
| "loss": 0.5618, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.993209279467164e-05, | |
| "loss": 0.6093, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.992897250651535e-05, | |
| "loss": 0.5744, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.992578223392124e-05, | |
| "loss": 0.5844, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.992252198584554e-05, | |
| "loss": 0.5358, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9919191771440905e-05, | |
| "loss": 0.6067, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.991579160005644e-05, | |
| "loss": 0.6147, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.991232148123761e-05, | |
| "loss": 0.6185, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.990878142472628e-05, | |
| "loss": 0.5797, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.990517144046064e-05, | |
| "loss": 0.58, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9901491538575185e-05, | |
| "loss": 0.6051, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9897741729400705e-05, | |
| "loss": 0.6199, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9893922023464236e-05, | |
| "loss": 0.6173, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.989003243148904e-05, | |
| "loss": 0.5907, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.988607296439458e-05, | |
| "loss": 0.5818, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.988204363329648e-05, | |
| "loss": 0.5767, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.987794444950651e-05, | |
| "loss": 0.579, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.987377542453251e-05, | |
| "loss": 0.6454, | |
| "step": 300 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 9375, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 100, | |
| "total_flos": 6.892911867632026e+16, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |