|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 0.8229775327133569, |
|
"eval_steps": 500, |
|
"global_step": 10000, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 5.0453611334320685e-06, |
|
"loss": 0.2826, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 6.229195710491767e-06, |
|
"loss": 0.2538, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 6.903829450223392e-06, |
|
"loss": 0.2949, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 7.377725845391017e-06, |
|
"loss": 0.3298, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 7.743343231239583e-06, |
|
"loss": 0.2927, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 8.041073861170494e-06, |
|
"loss": 0.3015, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 8.292222957399574e-06, |
|
"loss": 0.2914, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 8.509413541357755e-06, |
|
"loss": 0.3043, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 8.700744577655557e-06, |
|
"loss": 0.3396, |
|
"step": 225 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 8.871723942761204e-06, |
|
"loss": 0.2985, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 9.026267958246849e-06, |
|
"loss": 0.2895, |
|
"step": 275 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 9.16726106663399e-06, |
|
"loss": 0.2732, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 9.296889251455016e-06, |
|
"loss": 0.3234, |
|
"step": 325 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 9.416848797368692e-06, |
|
"loss": 0.3023, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 9.528482449516371e-06, |
|
"loss": 0.3097, |
|
"step": 375 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 9.632871309784314e-06, |
|
"loss": 0.3526, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 9.73089868785391e-06, |
|
"loss": 0.3205, |
|
"step": 425 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 9.823295589572114e-06, |
|
"loss": 0.3256, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 9.910673836465484e-06, |
|
"loss": 0.3189, |
|
"step": 475 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 9.993550644973805e-06, |
|
"loss": 0.2663, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 9.976842105263158e-06, |
|
"loss": 0.2888, |
|
"step": 525 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.950526315789475e-06, |
|
"loss": 0.2944, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.92421052631579e-06, |
|
"loss": 0.3057, |
|
"step": 575 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.897894736842107e-06, |
|
"loss": 0.2682, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.871578947368422e-06, |
|
"loss": 0.3085, |
|
"step": 625 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 9.845263157894738e-06, |
|
"loss": 0.2925, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.818947368421053e-06, |
|
"loss": 0.2887, |
|
"step": 675 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.79263157894737e-06, |
|
"loss": 0.3216, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.766315789473685e-06, |
|
"loss": 0.2931, |
|
"step": 725 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.74e-06, |
|
"loss": 0.2842, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 9.713684210526317e-06, |
|
"loss": 0.3191, |
|
"step": 775 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.687368421052632e-06, |
|
"loss": 0.2843, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.661052631578948e-06, |
|
"loss": 0.2623, |
|
"step": 825 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.634736842105265e-06, |
|
"loss": 0.2664, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.60842105263158e-06, |
|
"loss": 0.2963, |
|
"step": 875 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9.582105263157897e-06, |
|
"loss": 0.2996, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.555789473684211e-06, |
|
"loss": 0.304, |
|
"step": 925 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.529473684210528e-06, |
|
"loss": 0.2876, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.503157894736843e-06, |
|
"loss": 0.3041, |
|
"step": 975 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.476842105263158e-06, |
|
"loss": 0.3216, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.452631578947368e-06, |
|
"loss": 0.298, |
|
"step": 1025 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.426315789473685e-06, |
|
"loss": 0.3062, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.4e-06, |
|
"loss": 0.251, |
|
"step": 1075 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.373684210526316e-06, |
|
"loss": 0.2677, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.347368421052633e-06, |
|
"loss": 0.2588, |
|
"step": 1125 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.321052631578948e-06, |
|
"loss": 0.2658, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.294736842105265e-06, |
|
"loss": 0.2841, |
|
"step": 1175 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.26842105263158e-06, |
|
"loss": 0.2903, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.242105263157896e-06, |
|
"loss": 0.2688, |
|
"step": 1225 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.215789473684211e-06, |
|
"loss": 0.2839, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.189473684210526e-06, |
|
"loss": 0.2926, |
|
"step": 1275 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.163157894736843e-06, |
|
"loss": 0.3417, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.136842105263158e-06, |
|
"loss": 0.2811, |
|
"step": 1325 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.110526315789475e-06, |
|
"loss": 0.292, |
|
"step": 1350 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.08421052631579e-06, |
|
"loss": 0.2721, |
|
"step": 1375 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.057894736842106e-06, |
|
"loss": 0.2875, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.031578947368423e-06, |
|
"loss": 0.2684, |
|
"step": 1425 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.005263157894738e-06, |
|
"loss": 0.2691, |
|
"step": 1450 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 8.978947368421055e-06, |
|
"loss": 0.2733, |
|
"step": 1475 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 8.95263157894737e-06, |
|
"loss": 0.2866, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 8.926315789473685e-06, |
|
"loss": 0.3006, |
|
"step": 1525 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 8.900000000000001e-06, |
|
"loss": 0.2611, |
|
"step": 1550 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 8.873684210526316e-06, |
|
"loss": 0.2726, |
|
"step": 1575 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 8.847368421052633e-06, |
|
"loss": 0.2812, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 8.821052631578948e-06, |
|
"loss": 0.2998, |
|
"step": 1625 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 8.794736842105264e-06, |
|
"loss": 0.2951, |
|
"step": 1650 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 8.76842105263158e-06, |
|
"loss": 0.2693, |
|
"step": 1675 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 8.742105263157894e-06, |
|
"loss": 0.2616, |
|
"step": 1700 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 8.715789473684211e-06, |
|
"loss": 0.2723, |
|
"step": 1725 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 8.689473684210526e-06, |
|
"loss": 0.2738, |
|
"step": 1750 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 8.663157894736843e-06, |
|
"loss": 0.2789, |
|
"step": 1775 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 8.63684210526316e-06, |
|
"loss": 0.2665, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 8.610526315789474e-06, |
|
"loss": 0.2552, |
|
"step": 1825 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 8.584210526315791e-06, |
|
"loss": 0.2662, |
|
"step": 1850 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 8.557894736842106e-06, |
|
"loss": 0.3013, |
|
"step": 1875 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 8.531578947368423e-06, |
|
"loss": 0.2706, |
|
"step": 1900 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 8.505263157894738e-06, |
|
"loss": 0.2687, |
|
"step": 1925 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 8.478947368421053e-06, |
|
"loss": 0.2866, |
|
"step": 1950 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 8.45263157894737e-06, |
|
"loss": 0.2674, |
|
"step": 1975 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 8.426315789473684e-06, |
|
"loss": 0.2701, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 8.402105263157896e-06, |
|
"loss": 0.2691, |
|
"step": 2025 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 8.375789473684211e-06, |
|
"loss": 0.277, |
|
"step": 2050 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 8.349473684210528e-06, |
|
"loss": 0.2976, |
|
"step": 2075 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 8.323157894736843e-06, |
|
"loss": 0.2672, |
|
"step": 2100 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 8.29684210526316e-06, |
|
"loss": 0.2755, |
|
"step": 2125 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 8.270526315789474e-06, |
|
"loss": 0.254, |
|
"step": 2150 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 8.244210526315791e-06, |
|
"loss": 0.2853, |
|
"step": 2175 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 8.217894736842106e-06, |
|
"loss": 0.2965, |
|
"step": 2200 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 8.19157894736842e-06, |
|
"loss": 0.2498, |
|
"step": 2225 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 8.165263157894737e-06, |
|
"loss": 0.2461, |
|
"step": 2250 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 8.138947368421052e-06, |
|
"loss": 0.2716, |
|
"step": 2275 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 8.112631578947369e-06, |
|
"loss": 0.2376, |
|
"step": 2300 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 8.086315789473684e-06, |
|
"loss": 0.2556, |
|
"step": 2325 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 8.06e-06, |
|
"loss": 0.2181, |
|
"step": 2350 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 8.033684210526317e-06, |
|
"loss": 0.235, |
|
"step": 2375 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 8.007368421052632e-06, |
|
"loss": 0.2953, |
|
"step": 2400 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 7.981052631578949e-06, |
|
"loss": 0.2603, |
|
"step": 2425 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 7.954736842105264e-06, |
|
"loss": 0.2882, |
|
"step": 2450 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 7.928421052631579e-06, |
|
"loss": 0.2531, |
|
"step": 2475 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 7.902105263157896e-06, |
|
"loss": 0.269, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 7.87578947368421e-06, |
|
"loss": 0.2562, |
|
"step": 2525 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 7.849473684210527e-06, |
|
"loss": 0.2944, |
|
"step": 2550 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 7.823157894736842e-06, |
|
"loss": 0.2774, |
|
"step": 2575 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 7.796842105263159e-06, |
|
"loss": 0.2485, |
|
"step": 2600 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 7.770526315789474e-06, |
|
"loss": 0.2928, |
|
"step": 2625 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 7.744210526315789e-06, |
|
"loss": 0.2669, |
|
"step": 2650 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 7.717894736842107e-06, |
|
"loss": 0.2786, |
|
"step": 2675 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 7.691578947368422e-06, |
|
"loss": 0.254, |
|
"step": 2700 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 7.665263157894737e-06, |
|
"loss": 0.2564, |
|
"step": 2725 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 7.638947368421054e-06, |
|
"loss": 0.2158, |
|
"step": 2750 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 7.61263157894737e-06, |
|
"loss": 0.2322, |
|
"step": 2775 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 7.586315789473685e-06, |
|
"loss": 0.2353, |
|
"step": 2800 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 7.5600000000000005e-06, |
|
"loss": 0.2678, |
|
"step": 2825 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 7.533684210526316e-06, |
|
"loss": 0.2437, |
|
"step": 2850 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 7.507368421052632e-06, |
|
"loss": 0.2687, |
|
"step": 2875 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 7.481052631578948e-06, |
|
"loss": 0.2554, |
|
"step": 2900 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 7.454736842105264e-06, |
|
"loss": 0.2387, |
|
"step": 2925 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 7.4284210526315796e-06, |
|
"loss": 0.2348, |
|
"step": 2950 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 7.4021052631578945e-06, |
|
"loss": 0.2736, |
|
"step": 2975 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 7.37578947368421e-06, |
|
"loss": 0.2568, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 7.351578947368422e-06, |
|
"loss": 0.2451, |
|
"step": 3025 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 7.325263157894738e-06, |
|
"loss": 0.2792, |
|
"step": 3050 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 7.298947368421053e-06, |
|
"loss": 0.2817, |
|
"step": 3075 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 7.272631578947369e-06, |
|
"loss": 0.2562, |
|
"step": 3100 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 7.2463157894736845e-06, |
|
"loss": 0.2131, |
|
"step": 3125 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 7.22e-06, |
|
"loss": 0.2523, |
|
"step": 3150 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 7.193684210526316e-06, |
|
"loss": 0.2391, |
|
"step": 3175 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 7.167368421052632e-06, |
|
"loss": 0.2337, |
|
"step": 3200 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 7.141052631578948e-06, |
|
"loss": 0.2807, |
|
"step": 3225 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 7.1147368421052645e-06, |
|
"loss": 0.2541, |
|
"step": 3250 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 7.08842105263158e-06, |
|
"loss": 0.2365, |
|
"step": 3275 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 7.062105263157896e-06, |
|
"loss": 0.2566, |
|
"step": 3300 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 7.035789473684211e-06, |
|
"loss": 0.2538, |
|
"step": 3325 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 7.009473684210527e-06, |
|
"loss": 0.2496, |
|
"step": 3350 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 6.983157894736843e-06, |
|
"loss": 0.2757, |
|
"step": 3375 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 6.9568421052631585e-06, |
|
"loss": 0.2499, |
|
"step": 3400 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 6.930526315789474e-06, |
|
"loss": 0.2308, |
|
"step": 3425 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 6.90421052631579e-06, |
|
"loss": 0.2389, |
|
"step": 3450 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 6.877894736842106e-06, |
|
"loss": 0.2328, |
|
"step": 3475 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 6.851578947368421e-06, |
|
"loss": 0.2319, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 6.825263157894737e-06, |
|
"loss": 0.2441, |
|
"step": 3525 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 6.798947368421053e-06, |
|
"loss": 0.2423, |
|
"step": 3550 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 6.772631578947368e-06, |
|
"loss": 0.2347, |
|
"step": 3575 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 6.746315789473685e-06, |
|
"loss": 0.2621, |
|
"step": 3600 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 6.720000000000001e-06, |
|
"loss": 0.2517, |
|
"step": 3625 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 6.693684210526317e-06, |
|
"loss": 0.2801, |
|
"step": 3650 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 6.6673684210526325e-06, |
|
"loss": 0.2527, |
|
"step": 3675 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 6.641052631578948e-06, |
|
"loss": 0.2362, |
|
"step": 3700 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 6.614736842105264e-06, |
|
"loss": 0.2583, |
|
"step": 3725 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 6.588421052631579e-06, |
|
"loss": 0.2538, |
|
"step": 3750 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 6.562105263157895e-06, |
|
"loss": 0.2502, |
|
"step": 3775 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 6.535789473684211e-06, |
|
"loss": 0.2476, |
|
"step": 3800 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 6.509473684210527e-06, |
|
"loss": 0.2307, |
|
"step": 3825 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 6.483157894736842e-06, |
|
"loss": 0.2396, |
|
"step": 3850 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 6.456842105263158e-06, |
|
"loss": 0.2271, |
|
"step": 3875 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 6.430526315789474e-06, |
|
"loss": 0.243, |
|
"step": 3900 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 6.404210526315791e-06, |
|
"loss": 0.2468, |
|
"step": 3925 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 6.3778947368421065e-06, |
|
"loss": 0.2442, |
|
"step": 3950 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 6.351578947368422e-06, |
|
"loss": 0.2055, |
|
"step": 3975 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 6.325263157894737e-06, |
|
"loss": 0.2169, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 6.301052631578947e-06, |
|
"loss": 0.2578, |
|
"step": 4025 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 6.274736842105263e-06, |
|
"loss": 0.232, |
|
"step": 4050 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 6.248421052631579e-06, |
|
"loss": 0.2115, |
|
"step": 4075 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 6.222105263157895e-06, |
|
"loss": 0.2627, |
|
"step": 4100 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 6.195789473684211e-06, |
|
"loss": 0.2435, |
|
"step": 4125 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 6.1694736842105265e-06, |
|
"loss": 0.2651, |
|
"step": 4150 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 6.143157894736843e-06, |
|
"loss": 0.2516, |
|
"step": 4175 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 6.116842105263159e-06, |
|
"loss": 0.2383, |
|
"step": 4200 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 6.090526315789475e-06, |
|
"loss": 0.2199, |
|
"step": 4225 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 6.0642105263157906e-06, |
|
"loss": 0.2273, |
|
"step": 4250 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 6.0378947368421055e-06, |
|
"loss": 0.268, |
|
"step": 4275 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 6.011578947368421e-06, |
|
"loss": 0.2738, |
|
"step": 4300 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 5.985263157894737e-06, |
|
"loss": 0.2265, |
|
"step": 4325 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 5.9600000000000005e-06, |
|
"loss": 0.2412, |
|
"step": 4350 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 5.933684210526316e-06, |
|
"loss": 0.2274, |
|
"step": 4375 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 5.907368421052631e-06, |
|
"loss": 0.2482, |
|
"step": 4400 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 5.881052631578947e-06, |
|
"loss": 0.2313, |
|
"step": 4425 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 5.854736842105264e-06, |
|
"loss": 0.2243, |
|
"step": 4450 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 5.82842105263158e-06, |
|
"loss": 0.2475, |
|
"step": 4475 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 5.8021052631578954e-06, |
|
"loss": 0.2197, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 5.775789473684211e-06, |
|
"loss": 0.2484, |
|
"step": 4525 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 5.749473684210527e-06, |
|
"loss": 0.2398, |
|
"step": 4550 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 5.723157894736843e-06, |
|
"loss": 0.2228, |
|
"step": 4575 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 5.696842105263159e-06, |
|
"loss": 0.2405, |
|
"step": 4600 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 5.670526315789474e-06, |
|
"loss": 0.2356, |
|
"step": 4625 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 5.6442105263157895e-06, |
|
"loss": 0.2354, |
|
"step": 4650 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 5.617894736842105e-06, |
|
"loss": 0.2142, |
|
"step": 4675 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 5.591578947368421e-06, |
|
"loss": 0.2492, |
|
"step": 4700 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 5.565263157894737e-06, |
|
"loss": 0.2386, |
|
"step": 4725 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 5.538947368421053e-06, |
|
"loss": 0.2357, |
|
"step": 4750 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 5.512631578947369e-06, |
|
"loss": 0.2114, |
|
"step": 4775 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 5.486315789473685e-06, |
|
"loss": 0.2387, |
|
"step": 4800 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 5.460000000000001e-06, |
|
"loss": 0.2135, |
|
"step": 4825 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 5.433684210526317e-06, |
|
"loss": 0.271, |
|
"step": 4850 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 5.407368421052632e-06, |
|
"loss": 0.2306, |
|
"step": 4875 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 5.381052631578948e-06, |
|
"loss": 0.24, |
|
"step": 4900 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 5.3547368421052635e-06, |
|
"loss": 0.2306, |
|
"step": 4925 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 5.328421052631579e-06, |
|
"loss": 0.253, |
|
"step": 4950 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 5.302105263157895e-06, |
|
"loss": 0.2143, |
|
"step": 4975 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 5.275789473684211e-06, |
|
"loss": 0.2534, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 5.249473684210527e-06, |
|
"loss": 0.287, |
|
"step": 5025 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 5.223157894736843e-06, |
|
"loss": 0.232, |
|
"step": 5050 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 5.196842105263158e-06, |
|
"loss": 0.2135, |
|
"step": 5075 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 5.170526315789473e-06, |
|
"loss": 0.2544, |
|
"step": 5100 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 5.14421052631579e-06, |
|
"loss": 0.2916, |
|
"step": 5125 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 5.117894736842106e-06, |
|
"loss": 0.2373, |
|
"step": 5150 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 5.091578947368422e-06, |
|
"loss": 0.2238, |
|
"step": 5175 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 5.0652631578947375e-06, |
|
"loss": 0.2522, |
|
"step": 5200 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 5.038947368421053e-06, |
|
"loss": 0.2523, |
|
"step": 5225 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 5.012631578947369e-06, |
|
"loss": 0.2504, |
|
"step": 5250 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 4.986315789473685e-06, |
|
"loss": 0.2398, |
|
"step": 5275 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 4.960000000000001e-06, |
|
"loss": 0.2351, |
|
"step": 5300 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 4.933684210526316e-06, |
|
"loss": 0.2461, |
|
"step": 5325 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 4.907368421052632e-06, |
|
"loss": 0.225, |
|
"step": 5350 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 4.881052631578948e-06, |
|
"loss": 0.219, |
|
"step": 5375 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 4.854736842105264e-06, |
|
"loss": 0.2384, |
|
"step": 5400 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 4.828421052631579e-06, |
|
"loss": 0.2348, |
|
"step": 5425 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 4.802105263157895e-06, |
|
"loss": 0.2215, |
|
"step": 5450 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 4.775789473684211e-06, |
|
"loss": 0.2413, |
|
"step": 5475 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 4.7494736842105265e-06, |
|
"loss": 0.2252, |
|
"step": 5500 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 4.723157894736842e-06, |
|
"loss": 0.2549, |
|
"step": 5525 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 4.696842105263158e-06, |
|
"loss": 0.2257, |
|
"step": 5550 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 4.670526315789474e-06, |
|
"loss": 0.2006, |
|
"step": 5575 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 4.64421052631579e-06, |
|
"loss": 0.2441, |
|
"step": 5600 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 4.617894736842106e-06, |
|
"loss": 0.2728, |
|
"step": 5625 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 4.591578947368421e-06, |
|
"loss": 0.2353, |
|
"step": 5650 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 4.565263157894737e-06, |
|
"loss": 0.2535, |
|
"step": 5675 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 4.538947368421053e-06, |
|
"loss": 0.2475, |
|
"step": 5700 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 4.512631578947369e-06, |
|
"loss": 0.2029, |
|
"step": 5725 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 4.486315789473685e-06, |
|
"loss": 0.2016, |
|
"step": 5750 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 4.4600000000000005e-06, |
|
"loss": 0.2583, |
|
"step": 5775 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 4.433684210526316e-06, |
|
"loss": 0.2271, |
|
"step": 5800 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 4.407368421052632e-06, |
|
"loss": 0.2197, |
|
"step": 5825 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 4.381052631578948e-06, |
|
"loss": 0.233, |
|
"step": 5850 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 4.354736842105263e-06, |
|
"loss": 0.2042, |
|
"step": 5875 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 4.32842105263158e-06, |
|
"loss": 0.2533, |
|
"step": 5900 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 4.302105263157895e-06, |
|
"loss": 0.2425, |
|
"step": 5925 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 4.275789473684211e-06, |
|
"loss": 0.241, |
|
"step": 5950 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 4.249473684210527e-06, |
|
"loss": 0.2311, |
|
"step": 5975 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 4.223157894736842e-06, |
|
"loss": 0.213, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 4.196842105263158e-06, |
|
"loss": 0.2157, |
|
"step": 6025 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 4.170526315789474e-06, |
|
"loss": 0.2097, |
|
"step": 6050 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 4.14421052631579e-06, |
|
"loss": 0.2059, |
|
"step": 6075 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 4.117894736842106e-06, |
|
"loss": 0.2205, |
|
"step": 6100 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 4.091578947368421e-06, |
|
"loss": 0.2134, |
|
"step": 6125 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 4.065263157894737e-06, |
|
"loss": 0.2489, |
|
"step": 6150 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 4.038947368421053e-06, |
|
"loss": 0.2418, |
|
"step": 6175 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 4.0126315789473686e-06, |
|
"loss": 0.232, |
|
"step": 6200 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 3.986315789473684e-06, |
|
"loss": 0.2258, |
|
"step": 6225 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 3.96e-06, |
|
"loss": 0.2343, |
|
"step": 6250 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 3.933684210526316e-06, |
|
"loss": 0.2288, |
|
"step": 6275 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 3.907368421052632e-06, |
|
"loss": 0.2417, |
|
"step": 6300 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 3.881052631578948e-06, |
|
"loss": 0.2252, |
|
"step": 6325 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 3.8568421052631585e-06, |
|
"loss": 0.2039, |
|
"step": 6350 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 3.830526315789474e-06, |
|
"loss": 0.2081, |
|
"step": 6375 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 3.8042105263157898e-06, |
|
"loss": 0.2244, |
|
"step": 6400 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 3.7778947368421056e-06, |
|
"loss": 0.2121, |
|
"step": 6425 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 3.751578947368421e-06, |
|
"loss": 0.1981, |
|
"step": 6450 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 3.7252631578947372e-06, |
|
"loss": 0.2209, |
|
"step": 6475 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 3.698947368421053e-06, |
|
"loss": 0.1973, |
|
"step": 6500 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 3.672631578947369e-06, |
|
"loss": 0.184, |
|
"step": 6525 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 3.6463157894736847e-06, |
|
"loss": 0.2522, |
|
"step": 6550 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 3.62e-06, |
|
"loss": 0.1923, |
|
"step": 6575 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 3.593684210526316e-06, |
|
"loss": 0.2284, |
|
"step": 6600 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 3.567368421052632e-06, |
|
"loss": 0.1941, |
|
"step": 6625 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 3.541052631578948e-06, |
|
"loss": 0.2316, |
|
"step": 6650 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 3.5147368421052638e-06, |
|
"loss": 0.2039, |
|
"step": 6675 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 3.488421052631579e-06, |
|
"loss": 0.2123, |
|
"step": 6700 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 3.462105263157895e-06, |
|
"loss": 0.2254, |
|
"step": 6725 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 3.435789473684211e-06, |
|
"loss": 0.2041, |
|
"step": 6750 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 3.409473684210526e-06, |
|
"loss": 0.1916, |
|
"step": 6775 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 3.3831578947368424e-06, |
|
"loss": 0.1907, |
|
"step": 6800 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 3.3568421052631583e-06, |
|
"loss": 0.2589, |
|
"step": 6825 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 3.330526315789474e-06, |
|
"loss": 0.2016, |
|
"step": 6850 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 3.30421052631579e-06, |
|
"loss": 0.2198, |
|
"step": 6875 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 3.2778947368421053e-06, |
|
"loss": 0.1903, |
|
"step": 6900 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 3.251578947368421e-06, |
|
"loss": 0.2355, |
|
"step": 6925 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 3.225263157894737e-06, |
|
"loss": 0.2013, |
|
"step": 6950 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 3.198947368421053e-06, |
|
"loss": 0.2083, |
|
"step": 6975 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 3.172631578947369e-06, |
|
"loss": 0.2311, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 3.1463157894736844e-06, |
|
"loss": 0.2072, |
|
"step": 7025 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 3.12e-06, |
|
"loss": 0.2302, |
|
"step": 7050 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 3.093684210526316e-06, |
|
"loss": 0.2152, |
|
"step": 7075 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 3.067368421052632e-06, |
|
"loss": 0.1985, |
|
"step": 7100 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 3.0410526315789472e-06, |
|
"loss": 0.2066, |
|
"step": 7125 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 3.0147368421052635e-06, |
|
"loss": 0.1944, |
|
"step": 7150 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 2.9884210526315793e-06, |
|
"loss": 0.2183, |
|
"step": 7175 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 2.962105263157895e-06, |
|
"loss": 0.1926, |
|
"step": 7200 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 2.935789473684211e-06, |
|
"loss": 0.222, |
|
"step": 7225 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 2.9094736842105263e-06, |
|
"loss": 0.2211, |
|
"step": 7250 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 2.883157894736842e-06, |
|
"loss": 0.2201, |
|
"step": 7275 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 2.856842105263158e-06, |
|
"loss": 0.2165, |
|
"step": 7300 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 2.830526315789474e-06, |
|
"loss": 0.2177, |
|
"step": 7325 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 2.8063157894736842e-06, |
|
"loss": 0.1929, |
|
"step": 7350 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 2.7800000000000005e-06, |
|
"loss": 0.1903, |
|
"step": 7375 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 2.7536842105263163e-06, |
|
"loss": 0.2097, |
|
"step": 7400 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 2.7273684210526317e-06, |
|
"loss": 0.2335, |
|
"step": 7425 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 2.7010526315789475e-06, |
|
"loss": 0.2315, |
|
"step": 7450 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 2.6747368421052633e-06, |
|
"loss": 0.2249, |
|
"step": 7475 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 2.648421052631579e-06, |
|
"loss": 0.2397, |
|
"step": 7500 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 2.6221052631578946e-06, |
|
"loss": 0.2241, |
|
"step": 7525 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 2.595789473684211e-06, |
|
"loss": 0.1969, |
|
"step": 7550 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 2.5694736842105266e-06, |
|
"loss": 0.2475, |
|
"step": 7575 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 2.5431578947368424e-06, |
|
"loss": 0.2122, |
|
"step": 7600 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 2.5168421052631583e-06, |
|
"loss": 0.1849, |
|
"step": 7625 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 2.4905263157894736e-06, |
|
"loss": 0.2259, |
|
"step": 7650 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 2.46421052631579e-06, |
|
"loss": 0.2062, |
|
"step": 7675 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 2.4378947368421053e-06, |
|
"loss": 0.1968, |
|
"step": 7700 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 2.411578947368421e-06, |
|
"loss": 0.2024, |
|
"step": 7725 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 2.3852631578947373e-06, |
|
"loss": 0.2169, |
|
"step": 7750 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 2.3589473684210527e-06, |
|
"loss": 0.2229, |
|
"step": 7775 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 2.3326315789473686e-06, |
|
"loss": 0.2013, |
|
"step": 7800 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 2.3063157894736844e-06, |
|
"loss": 0.2276, |
|
"step": 7825 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 2.28e-06, |
|
"loss": 0.1745, |
|
"step": 7850 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 2.253684210526316e-06, |
|
"loss": 0.2252, |
|
"step": 7875 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 2.227368421052632e-06, |
|
"loss": 0.2135, |
|
"step": 7900 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 2.2010526315789476e-06, |
|
"loss": 0.2181, |
|
"step": 7925 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 2.1747368421052635e-06, |
|
"loss": 0.21, |
|
"step": 7950 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 2.148421052631579e-06, |
|
"loss": 0.2151, |
|
"step": 7975 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 2.1221052631578947e-06, |
|
"loss": 0.1712, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 2.095789473684211e-06, |
|
"loss": 0.2024, |
|
"step": 8025 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 2.0694736842105263e-06, |
|
"loss": 0.2238, |
|
"step": 8050 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 2.043157894736842e-06, |
|
"loss": 0.2428, |
|
"step": 8075 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 2.016842105263158e-06, |
|
"loss": 0.22, |
|
"step": 8100 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 1.9905263157894738e-06, |
|
"loss": 0.2441, |
|
"step": 8125 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 1.9642105263157896e-06, |
|
"loss": 0.1981, |
|
"step": 8150 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 1.9378947368421054e-06, |
|
"loss": 0.2208, |
|
"step": 8175 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 1.9115789473684212e-06, |
|
"loss": 0.2001, |
|
"step": 8200 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 1.885263157894737e-06, |
|
"loss": 0.2143, |
|
"step": 8225 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 1.8589473684210527e-06, |
|
"loss": 0.2097, |
|
"step": 8250 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 1.8326315789473687e-06, |
|
"loss": 0.2023, |
|
"step": 8275 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 1.8063157894736843e-06, |
|
"loss": 0.2233, |
|
"step": 8300 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 1.7800000000000001e-06, |
|
"loss": 0.2349, |
|
"step": 8325 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 1.7557894736842108e-06, |
|
"loss": 0.2292, |
|
"step": 8350 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 1.7294736842105264e-06, |
|
"loss": 0.2374, |
|
"step": 8375 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 1.7031578947368422e-06, |
|
"loss": 0.2459, |
|
"step": 8400 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 1.676842105263158e-06, |
|
"loss": 0.2072, |
|
"step": 8425 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 1.6505263157894739e-06, |
|
"loss": 0.2321, |
|
"step": 8450 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 1.6242105263157895e-06, |
|
"loss": 0.207, |
|
"step": 8475 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 1.5978947368421055e-06, |
|
"loss": 0.2005, |
|
"step": 8500 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 1.5715789473684213e-06, |
|
"loss": 0.2216, |
|
"step": 8525 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 1.545263157894737e-06, |
|
"loss": 0.1913, |
|
"step": 8550 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 1.518947368421053e-06, |
|
"loss": 0.2017, |
|
"step": 8575 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 1.4926315789473686e-06, |
|
"loss": 0.1724, |
|
"step": 8600 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 1.4663157894736844e-06, |
|
"loss": 0.2071, |
|
"step": 8625 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 1.44e-06, |
|
"loss": 0.2295, |
|
"step": 8650 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 1.413684210526316e-06, |
|
"loss": 0.2117, |
|
"step": 8675 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 1.3873684210526316e-06, |
|
"loss": 0.2116, |
|
"step": 8700 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 1.3610526315789474e-06, |
|
"loss": 0.2154, |
|
"step": 8725 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 1.3347368421052635e-06, |
|
"loss": 0.2274, |
|
"step": 8750 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 1.308421052631579e-06, |
|
"loss": 0.1889, |
|
"step": 8775 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 1.2821052631578949e-06, |
|
"loss": 0.1956, |
|
"step": 8800 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 1.2557894736842105e-06, |
|
"loss": 0.2036, |
|
"step": 8825 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 1.2294736842105263e-06, |
|
"loss": 0.2432, |
|
"step": 8850 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 1.2031578947368421e-06, |
|
"loss": 0.2063, |
|
"step": 8875 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 1.176842105263158e-06, |
|
"loss": 0.1829, |
|
"step": 8900 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 1.1505263157894738e-06, |
|
"loss": 0.214, |
|
"step": 8925 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 1.1242105263157896e-06, |
|
"loss": 0.2065, |
|
"step": 8950 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 1.0978947368421052e-06, |
|
"loss": 0.1974, |
|
"step": 8975 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 1.0715789473684212e-06, |
|
"loss": 0.2128, |
|
"step": 9000 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 1.045263157894737e-06, |
|
"loss": 0.2462, |
|
"step": 9025 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 1.0189473684210527e-06, |
|
"loss": 0.188, |
|
"step": 9050 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 9.936842105263158e-07, |
|
"loss": 0.2401, |
|
"step": 9075 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 9.673684210526316e-07, |
|
"loss": 0.1751, |
|
"step": 9100 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 9.410526315789474e-07, |
|
"loss": 0.2274, |
|
"step": 9125 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 9.147368421052632e-07, |
|
"loss": 0.205, |
|
"step": 9150 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 8.88421052631579e-07, |
|
"loss": 0.2005, |
|
"step": 9175 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 8.621052631578948e-07, |
|
"loss": 0.2095, |
|
"step": 9200 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 8.357894736842106e-07, |
|
"loss": 0.2118, |
|
"step": 9225 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 8.094736842105263e-07, |
|
"loss": 0.2223, |
|
"step": 9250 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 7.831578947368422e-07, |
|
"loss": 0.1951, |
|
"step": 9275 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 7.56842105263158e-07, |
|
"loss": 0.2063, |
|
"step": 9300 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 7.305263157894738e-07, |
|
"loss": 0.2202, |
|
"step": 9325 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 7.042105263157896e-07, |
|
"loss": 0.2056, |
|
"step": 9350 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 6.778947368421053e-07, |
|
"loss": 0.2485, |
|
"step": 9375 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 6.515789473684211e-07, |
|
"loss": 0.2224, |
|
"step": 9400 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 6.252631578947368e-07, |
|
"loss": 0.2212, |
|
"step": 9425 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 5.989473684210526e-07, |
|
"loss": 0.2204, |
|
"step": 9450 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 5.726315789473685e-07, |
|
"loss": 0.2251, |
|
"step": 9475 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 5.463157894736843e-07, |
|
"loss": 0.1869, |
|
"step": 9500 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 5.2e-07, |
|
"loss": 0.1992, |
|
"step": 9525 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 4.936842105263158e-07, |
|
"loss": 0.2253, |
|
"step": 9550 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 4.6736842105263163e-07, |
|
"loss": 0.1934, |
|
"step": 9575 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 4.410526315789474e-07, |
|
"loss": 0.2083, |
|
"step": 9600 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 4.1473684210526317e-07, |
|
"loss": 0.1822, |
|
"step": 9625 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 3.88421052631579e-07, |
|
"loss": 0.2022, |
|
"step": 9650 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 3.6210526315789475e-07, |
|
"loss": 0.215, |
|
"step": 9675 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 3.3578947368421057e-07, |
|
"loss": 0.2159, |
|
"step": 9700 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 3.0947368421052633e-07, |
|
"loss": 0.2243, |
|
"step": 9725 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.8315789473684215e-07, |
|
"loss": 0.2008, |
|
"step": 9750 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.568421052631579e-07, |
|
"loss": 0.2249, |
|
"step": 9775 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 2.305263157894737e-07, |
|
"loss": 0.2154, |
|
"step": 9800 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 2.042105263157895e-07, |
|
"loss": 0.2355, |
|
"step": 9825 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 1.7789473684210527e-07, |
|
"loss": 0.2024, |
|
"step": 9850 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 1.5157894736842106e-07, |
|
"loss": 0.1999, |
|
"step": 9875 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 1.2526315789473685e-07, |
|
"loss": 0.2139, |
|
"step": 9900 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 9.894736842105264e-08, |
|
"loss": 0.1962, |
|
"step": 9925 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 7.263157894736842e-08, |
|
"loss": 0.1945, |
|
"step": 9950 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 4.631578947368422e-08, |
|
"loss": 0.2174, |
|
"step": 9975 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 2e-08, |
|
"loss": 0.2085, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"step": 10000, |
|
"total_flos": 8.590715932450488e+19, |
|
"train_loss": 0.241443017578125, |
|
"train_runtime": 25061.6522, |
|
"train_samples_per_second": 6.384, |
|
"train_steps_per_second": 0.399 |
|
} |
|
], |
|
"logging_steps": 25, |
|
"max_steps": 10000, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 1, |
|
"save_steps": 2000, |
|
"total_flos": 8.590715932450488e+19, |
|
"train_batch_size": 16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|