| { | |
| "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": 4.898977360288234e-06, | |
| "loss": 0.7672, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6.160712527409633e-06, | |
| "loss": 0.6426, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.85912902234906e-06, | |
| "loss": 0.6038, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.344547104469332e-06, | |
| "loss": 0.6007, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.716963756434345e-06, | |
| "loss": 0.5597, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.019180844200955e-06, | |
| "loss": 0.5361, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.27351214279797e-06, | |
| "loss": 0.52, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.49307723936858e-06, | |
| "loss": 0.5044, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.686247975778677e-06, | |
| "loss": 0.5196, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.858694625217149e-06, | |
| "loss": 0.4631, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.014436199608479e-06, | |
| "loss": 0.4645, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.156425255148058e-06, | |
| "loss": 0.4442, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.28689473531776e-06, | |
| "loss": 0.4742, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.407574351377137e-06, | |
| "loss": 0.4525, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.519831289296397e-06, | |
| "loss": 0.4513, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.624764935335318e-06, | |
| "loss": 0.4655, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.723272550712454e-06, | |
| "loss": 0.4471, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.816095971633122e-06, | |
| "loss": 0.4505, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.90385555539545e-06, | |
| "loss": 0.4212, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.987075336738768e-06, | |
| "loss": 0.3854, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.978947368421053e-06, | |
| "loss": 0.4108, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.95263157894737e-06, | |
| "loss": 0.4051, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.926315789473685e-06, | |
| "loss": 0.4071, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.9e-06, | |
| "loss": 0.3635, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.873684210526317e-06, | |
| "loss": 0.4016, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.847368421052632e-06, | |
| "loss": 0.3839, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.821052631578948e-06, | |
| "loss": 0.3805, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.794736842105263e-06, | |
| "loss": 0.3978, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.76842105263158e-06, | |
| "loss": 0.3772, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.742105263157897e-06, | |
| "loss": 0.3692, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.715789473684212e-06, | |
| "loss": 0.4012, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.689473684210527e-06, | |
| "loss": 0.3688, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.663157894736843e-06, | |
| "loss": 0.339, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.636842105263158e-06, | |
| "loss": 0.3346, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.610526315789475e-06, | |
| "loss": 0.3649, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.58421052631579e-06, | |
| "loss": 0.3608, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.557894736842107e-06, | |
| "loss": 0.3659, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.531578947368422e-06, | |
| "loss": 0.3482, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.505263157894738e-06, | |
| "loss": 0.3649, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.478947368421053e-06, | |
| "loss": 0.3701, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.452631578947368e-06, | |
| "loss": 0.3517, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.426315789473685e-06, | |
| "loss": 0.356, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.4e-06, | |
| "loss": 0.3206, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.373684210526316e-06, | |
| "loss": 0.3179, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.347368421052633e-06, | |
| "loss": 0.315, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.321052631578948e-06, | |
| "loss": 0.3161, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.294736842105265e-06, | |
| "loss": 0.3327, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.26842105263158e-06, | |
| "loss": 0.3365, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.242105263157896e-06, | |
| "loss": 0.3199, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.215789473684211e-06, | |
| "loss": 0.335, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.189473684210526e-06, | |
| "loss": 0.3206, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.163157894736843e-06, | |
| "loss": 0.3714, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.136842105263158e-06, | |
| "loss": 0.3191, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.110526315789475e-06, | |
| "loss": 0.338, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.08421052631579e-06, | |
| "loss": 0.3158, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.057894736842106e-06, | |
| "loss": 0.33, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.031578947368423e-06, | |
| "loss": 0.3031, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.005263157894738e-06, | |
| "loss": 0.3086, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.978947368421055e-06, | |
| "loss": 0.3125, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.95263157894737e-06, | |
| "loss": 0.3229, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 8.926315789473685e-06, | |
| "loss": 0.3319, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 0.3013, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 8.873684210526316e-06, | |
| "loss": 0.3065, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 8.847368421052633e-06, | |
| "loss": 0.3196, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 8.821052631578948e-06, | |
| "loss": 0.3177, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 8.794736842105264e-06, | |
| "loss": 0.3362, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 8.76842105263158e-06, | |
| "loss": 0.2915, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 8.742105263157894e-06, | |
| "loss": 0.2931, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 8.715789473684211e-06, | |
| "loss": 0.3116, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 8.689473684210526e-06, | |
| "loss": 0.301, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.663157894736843e-06, | |
| "loss": 0.304, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.63684210526316e-06, | |
| "loss": 0.2975, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.610526315789474e-06, | |
| "loss": 0.2888, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.584210526315791e-06, | |
| "loss": 0.2922, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.557894736842106e-06, | |
| "loss": 0.3138, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.531578947368423e-06, | |
| "loss": 0.2941, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.505263157894738e-06, | |
| "loss": 0.291, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.478947368421053e-06, | |
| "loss": 0.3047, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.45263157894737e-06, | |
| "loss": 0.3012, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.426315789473684e-06, | |
| "loss": 0.2927, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 0.2959, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.373684210526316e-06, | |
| "loss": 0.2956, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.347368421052633e-06, | |
| "loss": 0.3106, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.32105263157895e-06, | |
| "loss": 0.2887, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.294736842105264e-06, | |
| "loss": 0.2964, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.26842105263158e-06, | |
| "loss": 0.2827, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.242105263157896e-06, | |
| "loss": 0.2966, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.21578947368421e-06, | |
| "loss": 0.3052, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.189473684210527e-06, | |
| "loss": 0.2726, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.163157894736842e-06, | |
| "loss": 0.2703, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.136842105263159e-06, | |
| "loss": 0.2868, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.110526315789474e-06, | |
| "loss": 0.2641, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.08421052631579e-06, | |
| "loss": 0.2668, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.057894736842106e-06, | |
| "loss": 0.2481, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.03157894736842e-06, | |
| "loss": 0.26, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.005263157894737e-06, | |
| "loss": 0.3047, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 7.978947368421052e-06, | |
| "loss": 0.2758, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 7.952631578947369e-06, | |
| "loss": 0.2962, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 7.926315789473686e-06, | |
| "loss": 0.2682, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 7.9e-06, | |
| "loss": 0.2833, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 7.873684210526317e-06, | |
| "loss": 0.2741, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 7.847368421052632e-06, | |
| "loss": 0.3021, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 7.821052631578949e-06, | |
| "loss": 0.2806, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 7.794736842105264e-06, | |
| "loss": 0.2592, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 7.768421052631579e-06, | |
| "loss": 0.2953, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 7.742105263157896e-06, | |
| "loss": 0.2778, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 7.71578947368421e-06, | |
| "loss": 0.2854, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 7.689473684210527e-06, | |
| "loss": 0.2699, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 7.663157894736842e-06, | |
| "loss": 0.2754, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 7.636842105263159e-06, | |
| "loss": 0.2363, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 7.610526315789474e-06, | |
| "loss": 0.2489, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 7.58421052631579e-06, | |
| "loss": 0.2553, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 7.557894736842106e-06, | |
| "loss": 0.2782, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 7.531578947368422e-06, | |
| "loss": 0.2552, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 7.505263157894738e-06, | |
| "loss": 0.2726, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 7.478947368421054e-06, | |
| "loss": 0.2672, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 7.4526315789473695e-06, | |
| "loss": 0.2604, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 7.4263157894736845e-06, | |
| "loss": 0.2552, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 7.4e-06, | |
| "loss": 0.2747, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.373684210526316e-06, | |
| "loss": 0.2645, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.349473684210527e-06, | |
| "loss": 0.2602, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.323157894736843e-06, | |
| "loss": 0.2799, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.296842105263159e-06, | |
| "loss": 0.2854, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.2705263157894745e-06, | |
| "loss": 0.263, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.24421052631579e-06, | |
| "loss": 0.2339, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.217894736842106e-06, | |
| "loss": 0.2597, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.191578947368422e-06, | |
| "loss": 0.2518, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.165263157894738e-06, | |
| "loss": 0.2513, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.138947368421053e-06, | |
| "loss": 0.2841, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.1126315789473685e-06, | |
| "loss": 0.2603, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.086315789473684e-06, | |
| "loss": 0.2507, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.06e-06, | |
| "loss": 0.2664, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.033684210526316e-06, | |
| "loss": 0.2569, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.007368421052632e-06, | |
| "loss": 0.2557, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 6.9810526315789485e-06, | |
| "loss": 0.2793, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 6.954736842105264e-06, | |
| "loss": 0.2613, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 6.92842105263158e-06, | |
| "loss": 0.2445, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 6.902105263157896e-06, | |
| "loss": 0.246, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 6.875789473684211e-06, | |
| "loss": 0.2433, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 6.849473684210527e-06, | |
| "loss": 0.2414, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 6.8231578947368425e-06, | |
| "loss": 0.2517, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 6.796842105263158e-06, | |
| "loss": 0.2485, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 6.770526315789474e-06, | |
| "loss": 0.2435, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 6.74421052631579e-06, | |
| "loss": 0.2639, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 6.717894736842106e-06, | |
| "loss": 0.2576, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 6.691578947368421e-06, | |
| "loss": 0.2748, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 6.665263157894737e-06, | |
| "loss": 0.2598, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 6.638947368421054e-06, | |
| "loss": 0.2483, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 6.612631578947369e-06, | |
| "loss": 0.2594, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 6.586315789473685e-06, | |
| "loss": 0.2531, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 6.560000000000001e-06, | |
| "loss": 0.2561, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 6.5336842105263165e-06, | |
| "loss": 0.2556, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 6.507368421052632e-06, | |
| "loss": 0.2339, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 6.481052631578948e-06, | |
| "loss": 0.2439, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 6.454736842105264e-06, | |
| "loss": 0.2423, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 6.428421052631579e-06, | |
| "loss": 0.2478, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 6.402105263157895e-06, | |
| "loss": 0.2521, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 6.375789473684211e-06, | |
| "loss": 0.253, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 6.349473684210526e-06, | |
| "loss": 0.2156, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 6.323157894736842e-06, | |
| "loss": 0.2237, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 6.298947368421053e-06, | |
| "loss": 0.2585, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 6.272631578947369e-06, | |
| "loss": 0.234, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.246315789473685e-06, | |
| "loss": 0.2213, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.220000000000001e-06, | |
| "loss": 0.2575, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.193684210526316e-06, | |
| "loss": 0.2551, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.167368421052632e-06, | |
| "loss": 0.2657, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.141052631578947e-06, | |
| "loss": 0.2549, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.114736842105263e-06, | |
| "loss": 0.2401, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.088421052631579e-06, | |
| "loss": 0.2324, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.062105263157895e-06, | |
| "loss": 0.2244, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.035789473684211e-06, | |
| "loss": 0.261, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.009473684210527e-06, | |
| "loss": 0.2734, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 5.983157894736843e-06, | |
| "loss": 0.234, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 5.956842105263159e-06, | |
| "loss": 0.2344, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 5.930526315789475e-06, | |
| "loss": 0.2361, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 5.90421052631579e-06, | |
| "loss": 0.2513, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 5.877894736842105e-06, | |
| "loss": 0.2383, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 5.851578947368421e-06, | |
| "loss": 0.2277, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 5.825263157894737e-06, | |
| "loss": 0.2483, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 5.798947368421053e-06, | |
| "loss": 0.2257, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 5.772631578947369e-06, | |
| "loss": 0.2494, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 5.7463157894736845e-06, | |
| "loss": 0.2394, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 5.72e-06, | |
| "loss": 0.2278, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 5.693684210526316e-06, | |
| "loss": 0.244, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 5.667368421052633e-06, | |
| "loss": 0.2342, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 5.641052631578949e-06, | |
| "loss": 0.2382, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 5.6147368421052636e-06, | |
| "loss": 0.2195, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 5.588421052631579e-06, | |
| "loss": 0.2508, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 5.562105263157895e-06, | |
| "loss": 0.2454, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 5.535789473684211e-06, | |
| "loss": 0.2348, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 5.509473684210527e-06, | |
| "loss": 0.2245, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5.483157894736843e-06, | |
| "loss": 0.2439, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5.4568421052631585e-06, | |
| "loss": 0.2179, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5.4305263157894734e-06, | |
| "loss": 0.2581, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5.404210526315789e-06, | |
| "loss": 0.2344, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5.377894736842105e-06, | |
| "loss": 0.2453, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.351578947368421e-06, | |
| "loss": 0.2331, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.3252631578947376e-06, | |
| "loss": 0.2507, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.298947368421053e-06, | |
| "loss": 0.2219, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.272631578947369e-06, | |
| "loss": 0.2467, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.248421052631579e-06, | |
| "loss": 0.2757, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.222105263157895e-06, | |
| "loss": 0.229, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.195789473684211e-06, | |
| "loss": 0.2263, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.169473684210527e-06, | |
| "loss": 0.2482, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.1431578947368425e-06, | |
| "loss": 0.2807, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.1168421052631575e-06, | |
| "loss": 0.2472, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.090526315789475e-06, | |
| "loss": 0.2269, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.06421052631579e-06, | |
| "loss": 0.2438, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.037894736842106e-06, | |
| "loss": 0.2519, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.011578947368422e-06, | |
| "loss": 0.2469, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.9852631578947374e-06, | |
| "loss": 0.2423, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.958947368421053e-06, | |
| "loss": 0.2347, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.932631578947369e-06, | |
| "loss": 0.2456, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.906315789473685e-06, | |
| "loss": 0.2295, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.880000000000001e-06, | |
| "loss": 0.2164, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.853684210526316e-06, | |
| "loss": 0.2375, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.8273684210526315e-06, | |
| "loss": 0.2364, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.801052631578948e-06, | |
| "loss": 0.2226, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.774736842105264e-06, | |
| "loss": 0.2427, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.748421052631579e-06, | |
| "loss": 0.2342, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.722105263157895e-06, | |
| "loss": 0.2473, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.695789473684211e-06, | |
| "loss": 0.2298, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.669473684210526e-06, | |
| "loss": 0.2096, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.643157894736843e-06, | |
| "loss": 0.2454, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.616842105263158e-06, | |
| "loss": 0.2616, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.590526315789474e-06, | |
| "loss": 0.2317, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.56421052631579e-06, | |
| "loss": 0.2464, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.5378947368421055e-06, | |
| "loss": 0.2347, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.511578947368421e-06, | |
| "loss": 0.212, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.485263157894737e-06, | |
| "loss": 0.2099, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.458947368421053e-06, | |
| "loss": 0.2454, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.432631578947369e-06, | |
| "loss": 0.2292, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.406315789473685e-06, | |
| "loss": 0.2209, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.38e-06, | |
| "loss": 0.2332, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.353684210526316e-06, | |
| "loss": 0.2123, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.327368421052632e-06, | |
| "loss": 0.2417, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.301052631578948e-06, | |
| "loss": 0.2335, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.274736842105264e-06, | |
| "loss": 0.2406, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.2484210526315795e-06, | |
| "loss": 0.2321, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.222105263157895e-06, | |
| "loss": 0.22, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.197894736842106e-06, | |
| "loss": 0.2177, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.171578947368421e-06, | |
| "loss": 0.2122, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.145263157894737e-06, | |
| "loss": 0.2117, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.118947368421053e-06, | |
| "loss": 0.2166, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.092631578947369e-06, | |
| "loss": 0.2198, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.0663157894736845e-06, | |
| "loss": 0.2439, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.04e-06, | |
| "loss": 0.2462, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.013684210526316e-06, | |
| "loss": 0.2291, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 3.987368421052632e-06, | |
| "loss": 0.2311, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 3.961052631578948e-06, | |
| "loss": 0.2313, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 3.9347368421052636e-06, | |
| "loss": 0.2286, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 3.908421052631579e-06, | |
| "loss": 0.2425, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 3.882105263157895e-06, | |
| "loss": 0.2236, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 3.855789473684211e-06, | |
| "loss": 0.2064, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 3.829473684210527e-06, | |
| "loss": 0.2095, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 3.8031578947368426e-06, | |
| "loss": 0.2235, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 3.776842105263158e-06, | |
| "loss": 0.2176, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 3.750526315789474e-06, | |
| "loss": 0.2062, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 3.7242105263157897e-06, | |
| "loss": 0.2173, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 3.6978947368421055e-06, | |
| "loss": 0.201, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 3.6715789473684217e-06, | |
| "loss": 0.1935, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 3.645263157894737e-06, | |
| "loss": 0.2367, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 3.618947368421053e-06, | |
| "loss": 0.202, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 3.5926315789473688e-06, | |
| "loss": 0.2272, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 3.5663157894736846e-06, | |
| "loss": 0.1942, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 3.54e-06, | |
| "loss": 0.2303, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 3.513684210526316e-06, | |
| "loss": 0.2076, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 3.487368421052632e-06, | |
| "loss": 0.212, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 3.461052631578948e-06, | |
| "loss": 0.225, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 3.4347368421052637e-06, | |
| "loss": 0.2079, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 3.408421052631579e-06, | |
| "loss": 0.1967, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 3.382105263157895e-06, | |
| "loss": 0.1922, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 3.3557894736842107e-06, | |
| "loss": 0.2433, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 3.329473684210527e-06, | |
| "loss": 0.2058, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 3.3031578947368424e-06, | |
| "loss": 0.2173, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 3.276842105263158e-06, | |
| "loss": 0.1963, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 3.250526315789474e-06, | |
| "loss": 0.2353, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 3.22421052631579e-06, | |
| "loss": 0.2037, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 3.197894736842105e-06, | |
| "loss": 0.21, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.171578947368421e-06, | |
| "loss": 0.2268, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.147368421052632e-06, | |
| "loss": 0.2142, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.1210526315789473e-06, | |
| "loss": 0.2315, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.094736842105263e-06, | |
| "loss": 0.2164, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.0684210526315794e-06, | |
| "loss": 0.2052, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.042105263157895e-06, | |
| "loss": 0.2121, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.015789473684211e-06, | |
| "loss": 0.2037, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.9894736842105264e-06, | |
| "loss": 0.2217, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.9631578947368422e-06, | |
| "loss": 0.2051, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.936842105263158e-06, | |
| "loss": 0.2208, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.9105263157894743e-06, | |
| "loss": 0.2253, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.88421052631579e-06, | |
| "loss": 0.2211, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.8578947368421055e-06, | |
| "loss": 0.2129, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.8315789473684213e-06, | |
| "loss": 0.2176, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.805263157894737e-06, | |
| "loss": 0.1965, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.7789473684210525e-06, | |
| "loss": 0.192, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.7526315789473683e-06, | |
| "loss": 0.2126, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.7263157894736846e-06, | |
| "loss": 0.2365, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.7000000000000004e-06, | |
| "loss": 0.2306, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.6736842105263162e-06, | |
| "loss": 0.2193, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.6473684210526316e-06, | |
| "loss": 0.2388, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.6210526315789474e-06, | |
| "loss": 0.222, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.5947368421052633e-06, | |
| "loss": 0.2027, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.568421052631579e-06, | |
| "loss": 0.2414, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.5421052631578953e-06, | |
| "loss": 0.2122, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.5157894736842107e-06, | |
| "loss": 0.1938, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.4894736842105265e-06, | |
| "loss": 0.2222, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.4631578947368424e-06, | |
| "loss": 0.21, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.436842105263158e-06, | |
| "loss": 0.1995, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.410526315789474e-06, | |
| "loss": 0.1985, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.38421052631579e-06, | |
| "loss": 0.215, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.357894736842105e-06, | |
| "loss": 0.2197, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.331578947368421e-06, | |
| "loss": 0.209, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.3052631578947373e-06, | |
| "loss": 0.2226, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.2789473684210527e-06, | |
| "loss": 0.1847, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.2526315789473685e-06, | |
| "loss": 0.2253, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.2263157894736843e-06, | |
| "loss": 0.2128, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.2e-06, | |
| "loss": 0.2139, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.173684210526316e-06, | |
| "loss": 0.2059, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.1473684210526317e-06, | |
| "loss": 0.2081, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.1210526315789476e-06, | |
| "loss": 0.1793, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.096842105263158e-06, | |
| "loss": 0.203, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.070526315789474e-06, | |
| "loss": 0.2157, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.0442105263157897e-06, | |
| "loss": 0.2297, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.0178947368421055e-06, | |
| "loss": 0.223, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.9915789473684213e-06, | |
| "loss": 0.2374, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.965263157894737e-06, | |
| "loss": 0.2039, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.938947368421053e-06, | |
| "loss": 0.2209, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.9126315789473683e-06, | |
| "loss": 0.2024, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.8863157894736844e-06, | |
| "loss": 0.2127, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.8600000000000002e-06, | |
| "loss": 0.2103, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.8336842105263158e-06, | |
| "loss": 0.1989, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.8073684210526318e-06, | |
| "loss": 0.2192, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.7810526315789474e-06, | |
| "loss": 0.2317, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.7547368421052633e-06, | |
| "loss": 0.2245, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.7284210526315793e-06, | |
| "loss": 0.2216, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.7021052631578949e-06, | |
| "loss": 0.2392, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.6757894736842107e-06, | |
| "loss": 0.2088, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.6494736842105263e-06, | |
| "loss": 0.222, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.6231578947368423e-06, | |
| "loss": 0.207, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.596842105263158e-06, | |
| "loss": 0.203, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.5705263157894738e-06, | |
| "loss": 0.218, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.5442105263157898e-06, | |
| "loss": 0.1949, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.5178947368421054e-06, | |
| "loss": 0.2015, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4915789473684212e-06, | |
| "loss": 0.1839, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4652631578947368e-06, | |
| "loss": 0.2114, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4389473684210529e-06, | |
| "loss": 0.2247, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4126315789473685e-06, | |
| "loss": 0.2124, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.3863157894736843e-06, | |
| "loss": 0.2116, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.3600000000000001e-06, | |
| "loss": 0.2212, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.333684210526316e-06, | |
| "loss": 0.2214, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.3073684210526315e-06, | |
| "loss": 0.1966, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.2810526315789476e-06, | |
| "loss": 0.205, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.2547368421052634e-06, | |
| "loss": 0.2015, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.228421052631579e-06, | |
| "loss": 0.2347, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.2021052631578948e-06, | |
| "loss": 0.21, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.1757894736842106e-06, | |
| "loss": 0.1891, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.1494736842105264e-06, | |
| "loss": 0.2117, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.1231578947368423e-06, | |
| "loss": 0.2053, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.0968421052631579e-06, | |
| "loss": 0.1985, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.070526315789474e-06, | |
| "loss": 0.2126, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.0463157894736844e-06, | |
| "loss": 0.2417, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.02e-06, | |
| "loss": 0.1908, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.936842105263158e-07, | |
| "loss": 0.2168, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.673684210526316e-07, | |
| "loss": 0.1841, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.410526315789474e-07, | |
| "loss": 0.2246, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.147368421052632e-07, | |
| "loss": 0.2058, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.88421052631579e-07, | |
| "loss": 0.1996, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.621052631578948e-07, | |
| "loss": 0.211, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.357894736842106e-07, | |
| "loss": 0.2069, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.094736842105263e-07, | |
| "loss": 0.2186, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.831578947368422e-07, | |
| "loss": 0.1951, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.56842105263158e-07, | |
| "loss": 0.2041, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.305263157894738e-07, | |
| "loss": 0.2146, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.042105263157896e-07, | |
| "loss": 0.204, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 6.778947368421053e-07, | |
| "loss": 0.2424, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 6.515789473684211e-07, | |
| "loss": 0.2188, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.252631578947368e-07, | |
| "loss": 0.222, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 5.989473684210526e-07, | |
| "loss": 0.2161, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 5.726315789473685e-07, | |
| "loss": 0.219, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 5.463157894736843e-07, | |
| "loss": 0.1896, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 5.2e-07, | |
| "loss": 0.1996, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.936842105263158e-07, | |
| "loss": 0.2167, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.6736842105263163e-07, | |
| "loss": 0.1939, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.410526315789474e-07, | |
| "loss": 0.2089, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.1473684210526317e-07, | |
| "loss": 0.1889, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.88421052631579e-07, | |
| "loss": 0.2005, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.6210526315789475e-07, | |
| "loss": 0.2141, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.3578947368421057e-07, | |
| "loss": 0.2137, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.0947368421052633e-07, | |
| "loss": 0.2245, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.8315789473684215e-07, | |
| "loss": 0.2018, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.568421052631579e-07, | |
| "loss": 0.2241, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.305263157894737e-07, | |
| "loss": 0.216, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.042105263157895e-07, | |
| "loss": 0.2269, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.7789473684210527e-07, | |
| "loss": 0.209, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.5157894736842106e-07, | |
| "loss": 0.1998, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.2526315789473685e-07, | |
| "loss": 0.2129, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 9.894736842105264e-08, | |
| "loss": 0.1995, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.263157894736842e-08, | |
| "loss": 0.1983, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.631578947368422e-08, | |
| "loss": 0.2042, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2e-08, | |
| "loss": 0.2059, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "step": 10000, | |
| "total_flos": 4.617366439093862e+19, | |
| "train_loss": 0.26186515502929686, | |
| "train_runtime": 11850.3169, | |
| "train_samples_per_second": 13.502, | |
| "train_steps_per_second": 0.844 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 10000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 2000, | |
| "total_flos": 4.617366439093862e+19, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |