|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 2.187168922895821, |
|
"global_step": 13005, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 9.986209216279854e-06, |
|
"loss": 1.5736, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 9.974100235452405e-06, |
|
"loss": 2.2457, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.961823074335689e-06, |
|
"loss": 1.5859, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.948705011772622e-06, |
|
"loss": 1.728, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.938109653548606e-06, |
|
"loss": 1.7661, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.927009754456779e-06, |
|
"loss": 2.3678209948440125e+23, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.910191725529768e-06, |
|
"loss": 0.0, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.893373696602759e-06, |
|
"loss": 0.0, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.876555667675748e-06, |
|
"loss": 0.0, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.85973763874874e-06, |
|
"loss": 0.0, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.84291960982173e-06, |
|
"loss": 0.0, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.826101580894721e-06, |
|
"loss": 0.0, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.80928355196771e-06, |
|
"loss": 0.0, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 9.792465523040701e-06, |
|
"loss": 0.0, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 9.77564749411369e-06, |
|
"loss": 0.0, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 9.758829465186681e-06, |
|
"loss": 0.0, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 9.74201143625967e-06, |
|
"loss": 0.0, |
|
"step": 1700 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 9.725193407332661e-06, |
|
"loss": 0.0, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 9.70837537840565e-06, |
|
"loss": 0.0, |
|
"step": 1900 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 9.691557349478641e-06, |
|
"loss": 0.0, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 9.674739320551632e-06, |
|
"loss": 0.0, |
|
"step": 2100 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 9.657921291624623e-06, |
|
"loss": 0.0, |
|
"step": 2200 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 9.641103262697614e-06, |
|
"loss": 0.0, |
|
"step": 2300 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 9.624285233770603e-06, |
|
"loss": 0.0, |
|
"step": 2400 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 9.607467204843594e-06, |
|
"loss": 0.0, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 9.590649175916583e-06, |
|
"loss": 0.0, |
|
"step": 2600 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 9.573831146989574e-06, |
|
"loss": 0.0, |
|
"step": 2700 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 9.557013118062563e-06, |
|
"loss": 0.0, |
|
"step": 2800 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 9.540195089135554e-06, |
|
"loss": 0.0, |
|
"step": 2900 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 9.523377060208543e-06, |
|
"loss": 0.0, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 9.506559031281534e-06, |
|
"loss": 0.0, |
|
"step": 3100 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.489741002354525e-06, |
|
"loss": 0.0, |
|
"step": 3200 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.472922973427516e-06, |
|
"loss": 0.0, |
|
"step": 3300 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 9.456104944500505e-06, |
|
"loss": 0.0, |
|
"step": 3400 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 9.439286915573496e-06, |
|
"loss": 0.0, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 9.422468886646485e-06, |
|
"loss": 0.0, |
|
"step": 3600 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 9.405650857719476e-06, |
|
"loss": 0.0, |
|
"step": 3700 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 9.388832828792467e-06, |
|
"loss": 0.0, |
|
"step": 3800 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 9.372014799865456e-06, |
|
"loss": 0.0, |
|
"step": 3900 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 9.355196770938447e-06, |
|
"loss": 0.0, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 9.338378742011436e-06, |
|
"loss": 0.0, |
|
"step": 4100 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 9.321560713084427e-06, |
|
"loss": 0.0, |
|
"step": 4200 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 9.304742684157418e-06, |
|
"loss": 0.0, |
|
"step": 4300 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 9.287924655230409e-06, |
|
"loss": 0.0, |
|
"step": 4400 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 9.271106626303398e-06, |
|
"loss": 0.0, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 9.254288597376389e-06, |
|
"loss": 0.0, |
|
"step": 4600 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 9.237470568449378e-06, |
|
"loss": 0.0, |
|
"step": 4700 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 9.220652539522369e-06, |
|
"loss": 0.0, |
|
"step": 4800 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 9.203834510595358e-06, |
|
"loss": 0.0, |
|
"step": 4900 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 9.187016481668349e-06, |
|
"loss": 0.0, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 9.170198452741338e-06, |
|
"loss": 0.0, |
|
"step": 5100 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 9.153380423814329e-06, |
|
"loss": 0.0, |
|
"step": 5200 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 9.13656239488732e-06, |
|
"loss": 0.0, |
|
"step": 5300 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 9.119744365960311e-06, |
|
"loss": 0.0, |
|
"step": 5400 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 9.102926337033302e-06, |
|
"loss": 0.0, |
|
"step": 5500 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 9.086108308106291e-06, |
|
"loss": 0.0, |
|
"step": 5600 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 9.069290279179282e-06, |
|
"loss": 0.0, |
|
"step": 5700 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 9.052472250252271e-06, |
|
"loss": 0.0, |
|
"step": 5800 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 9.035654221325262e-06, |
|
"loss": 0.0, |
|
"step": 5900 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 9.018836192398251e-06, |
|
"loss": 0.0, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 9.002018163471242e-06, |
|
"loss": 0.0, |
|
"step": 6100 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 8.985200134544231e-06, |
|
"loss": 0.0, |
|
"step": 6200 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 8.968382105617222e-06, |
|
"loss": 0.0, |
|
"step": 6300 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 8.951564076690213e-06, |
|
"loss": 0.0, |
|
"step": 6400 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 8.934746047763204e-06, |
|
"loss": 0.0, |
|
"step": 6500 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 8.917928018836193e-06, |
|
"loss": 0.0, |
|
"step": 6600 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 8.901109989909184e-06, |
|
"loss": 0.0, |
|
"step": 6700 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 8.884291960982173e-06, |
|
"loss": 0.0, |
|
"step": 6800 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 8.867473932055164e-06, |
|
"loss": 0.0, |
|
"step": 6900 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 8.850655903128153e-06, |
|
"loss": 0.0, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 8.833837874201144e-06, |
|
"loss": 0.0, |
|
"step": 7100 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 8.817019845274135e-06, |
|
"loss": 0.0, |
|
"step": 7200 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 8.800201816347124e-06, |
|
"loss": 0.0, |
|
"step": 7300 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 8.783383787420115e-06, |
|
"loss": 0.0, |
|
"step": 7400 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 8.766565758493106e-06, |
|
"loss": 0.0, |
|
"step": 7500 |
|
}, |
|
{ |
|
"epoch": 1.28, |
|
"learning_rate": 8.749747729566097e-06, |
|
"loss": 0.0, |
|
"step": 7600 |
|
}, |
|
{ |
|
"epoch": 1.29, |
|
"learning_rate": 8.732929700639086e-06, |
|
"loss": 0.0, |
|
"step": 7700 |
|
}, |
|
{ |
|
"epoch": 1.31, |
|
"learning_rate": 8.716111671712077e-06, |
|
"loss": 0.0, |
|
"step": 7800 |
|
}, |
|
{ |
|
"epoch": 1.33, |
|
"learning_rate": 8.699293642785066e-06, |
|
"loss": 0.0, |
|
"step": 7900 |
|
}, |
|
{ |
|
"epoch": 1.35, |
|
"learning_rate": 8.682475613858057e-06, |
|
"loss": 0.0, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 1.36, |
|
"learning_rate": 8.665657584931046e-06, |
|
"loss": 0.0, |
|
"step": 8100 |
|
}, |
|
{ |
|
"epoch": 1.38, |
|
"learning_rate": 8.648839556004037e-06, |
|
"loss": 0.0, |
|
"step": 8200 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"learning_rate": 8.632021527077026e-06, |
|
"loss": 0.0, |
|
"step": 8300 |
|
}, |
|
{ |
|
"epoch": 1.41, |
|
"learning_rate": 8.615203498150017e-06, |
|
"loss": 0.0, |
|
"step": 8400 |
|
}, |
|
{ |
|
"epoch": 1.43, |
|
"learning_rate": 8.598385469223008e-06, |
|
"loss": 0.0, |
|
"step": 8500 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"learning_rate": 8.581567440295999e-06, |
|
"loss": 0.0, |
|
"step": 8600 |
|
}, |
|
{ |
|
"epoch": 1.46, |
|
"learning_rate": 8.564749411368988e-06, |
|
"loss": 0.0, |
|
"step": 8700 |
|
}, |
|
{ |
|
"epoch": 1.48, |
|
"learning_rate": 8.547931382441979e-06, |
|
"loss": 0.0, |
|
"step": 8800 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"learning_rate": 8.53111335351497e-06, |
|
"loss": 0.0, |
|
"step": 8900 |
|
}, |
|
{ |
|
"epoch": 1.51, |
|
"learning_rate": 8.514295324587959e-06, |
|
"loss": 0.0, |
|
"step": 9000 |
|
}, |
|
{ |
|
"epoch": 1.53, |
|
"learning_rate": 8.49747729566095e-06, |
|
"loss": 0.0, |
|
"step": 9100 |
|
}, |
|
{ |
|
"epoch": 1.55, |
|
"learning_rate": 8.480659266733939e-06, |
|
"loss": 0.0, |
|
"step": 9200 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"learning_rate": 8.46384123780693e-06, |
|
"loss": 0.0, |
|
"step": 9300 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"learning_rate": 8.447023208879919e-06, |
|
"loss": 0.0, |
|
"step": 9400 |
|
}, |
|
{ |
|
"epoch": 1.6, |
|
"learning_rate": 8.43020517995291e-06, |
|
"loss": 0.0, |
|
"step": 9500 |
|
}, |
|
{ |
|
"epoch": 1.61, |
|
"learning_rate": 8.413387151025901e-06, |
|
"loss": 0.0, |
|
"step": 9600 |
|
}, |
|
{ |
|
"epoch": 1.63, |
|
"learning_rate": 8.396569122098892e-06, |
|
"loss": 0.0, |
|
"step": 9700 |
|
}, |
|
{ |
|
"epoch": 1.65, |
|
"learning_rate": 8.379751093171881e-06, |
|
"loss": 0.0, |
|
"step": 9800 |
|
}, |
|
{ |
|
"epoch": 1.66, |
|
"learning_rate": 8.362933064244872e-06, |
|
"loss": 0.0, |
|
"step": 9900 |
|
}, |
|
{ |
|
"epoch": 1.68, |
|
"learning_rate": 8.346115035317861e-06, |
|
"loss": 0.0, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 1.7, |
|
"learning_rate": 8.329297006390852e-06, |
|
"loss": 0.0, |
|
"step": 10100 |
|
}, |
|
{ |
|
"epoch": 1.72, |
|
"learning_rate": 8.312478977463841e-06, |
|
"loss": 0.0, |
|
"step": 10200 |
|
}, |
|
{ |
|
"epoch": 1.73, |
|
"learning_rate": 8.295660948536832e-06, |
|
"loss": 0.0, |
|
"step": 10300 |
|
}, |
|
{ |
|
"epoch": 1.75, |
|
"learning_rate": 8.278842919609821e-06, |
|
"loss": 0.0, |
|
"step": 10400 |
|
}, |
|
{ |
|
"epoch": 1.77, |
|
"learning_rate": 8.262024890682812e-06, |
|
"loss": 0.0, |
|
"step": 10500 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 8.245206861755803e-06, |
|
"loss": 0.0, |
|
"step": 10600 |
|
}, |
|
{ |
|
"epoch": 1.8, |
|
"learning_rate": 8.228388832828794e-06, |
|
"loss": 0.0, |
|
"step": 10700 |
|
}, |
|
{ |
|
"epoch": 1.82, |
|
"learning_rate": 8.211570803901785e-06, |
|
"loss": 0.0, |
|
"step": 10800 |
|
}, |
|
{ |
|
"epoch": 1.83, |
|
"learning_rate": 8.194752774974774e-06, |
|
"loss": 0.0, |
|
"step": 10900 |
|
}, |
|
{ |
|
"epoch": 1.85, |
|
"learning_rate": 8.177934746047765e-06, |
|
"loss": 0.0, |
|
"step": 11000 |
|
}, |
|
{ |
|
"epoch": 1.87, |
|
"learning_rate": 8.161116717120754e-06, |
|
"loss": 0.0, |
|
"step": 11100 |
|
}, |
|
{ |
|
"epoch": 1.88, |
|
"learning_rate": 8.144298688193745e-06, |
|
"loss": 0.0, |
|
"step": 11200 |
|
}, |
|
{ |
|
"epoch": 1.9, |
|
"learning_rate": 8.127480659266734e-06, |
|
"loss": 0.0, |
|
"step": 11300 |
|
}, |
|
{ |
|
"epoch": 1.92, |
|
"learning_rate": 8.110662630339725e-06, |
|
"loss": 0.0, |
|
"step": 11400 |
|
}, |
|
{ |
|
"epoch": 1.93, |
|
"learning_rate": 8.093844601412714e-06, |
|
"loss": 0.0, |
|
"step": 11500 |
|
}, |
|
{ |
|
"epoch": 1.95, |
|
"learning_rate": 8.077026572485705e-06, |
|
"loss": 0.0, |
|
"step": 11600 |
|
}, |
|
{ |
|
"epoch": 1.97, |
|
"learning_rate": 8.060208543558696e-06, |
|
"loss": 0.0, |
|
"step": 11700 |
|
}, |
|
{ |
|
"epoch": 1.98, |
|
"learning_rate": 8.043390514631687e-06, |
|
"loss": 0.0, |
|
"step": 11800 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 8.026572485704676e-06, |
|
"loss": 0.0, |
|
"step": 11900 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 8.009754456777667e-06, |
|
"loss": 0.0, |
|
"step": 12000 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 7.992936427850656e-06, |
|
"loss": 0.0, |
|
"step": 12100 |
|
}, |
|
{ |
|
"epoch": 2.05, |
|
"learning_rate": 7.976118398923647e-06, |
|
"loss": 0.0, |
|
"step": 12200 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 7.959300369996638e-06, |
|
"loss": 0.0, |
|
"step": 12300 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 7.942482341069627e-06, |
|
"loss": 0.0, |
|
"step": 12400 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 7.925664312142618e-06, |
|
"loss": 0.0, |
|
"step": 12500 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 7.908846283215607e-06, |
|
"loss": 0.0, |
|
"step": 12600 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 7.892028254288598e-06, |
|
"loss": 0.0, |
|
"step": 12700 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 7.875210225361589e-06, |
|
"loss": 0.0, |
|
"step": 12800 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 7.85839219643458e-06, |
|
"loss": 0.0, |
|
"step": 12900 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 7.841574167507569e-06, |
|
"loss": 0.0, |
|
"step": 13000 |
|
} |
|
], |
|
"max_steps": 59460, |
|
"num_train_epochs": 10, |
|
"total_flos": 1.314423778738176e+18, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|