| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.972136222910217, | |
| "eval_steps": 500, | |
| "global_step": 240, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0, | |
| "loss": 0.416, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0, | |
| "loss": 0.3341, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0, | |
| "loss": 0.3503, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0, | |
| "loss": 0.3645, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0, | |
| "loss": 0.4011, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0, | |
| "loss": 0.425, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0, | |
| "loss": 0.3697, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0, | |
| "loss": 0.3321, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0, | |
| "loss": 0.369, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0, | |
| "loss": 0.3766, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0, | |
| "loss": 0.3411, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0, | |
| "loss": 0.3398, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0, | |
| "loss": 0.3709, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0, | |
| "loss": 0.3917, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0, | |
| "loss": 0.3273, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0, | |
| "loss": 0.3648, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0, | |
| "loss": 0.3019, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0, | |
| "loss": 0.3749, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 5.578858913022596e-07, | |
| "loss": 0.267, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 5.578858913022596e-07, | |
| "loss": 0.3449, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.842282173954807e-07, | |
| "loss": 0.2667, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.1157717826045193e-06, | |
| "loss": 0.2683, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.295370924755994e-06, | |
| "loss": 0.2791, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.295370924755994e-06, | |
| "loss": 0.326, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.4421141086977403e-06, | |
| "loss": 0.3044, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.566183702893892e-06, | |
| "loss": 0.262, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.566183702893892e-06, | |
| "loss": 0.3506, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.6736576739067789e-06, | |
| "loss": 0.2547, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.7684564347909613e-06, | |
| "loss": 0.2664, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.8532568160582538e-06, | |
| "loss": 0.2599, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.9299680919626874e-06, | |
| "loss": 0.2514, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2e-06, | |
| "loss": 0.3075, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2e-06, | |
| "loss": 0.2747, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.9912280701754384e-06, | |
| "loss": 0.2548, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.982456140350877e-06, | |
| "loss": 0.3014, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.9736842105263157e-06, | |
| "loss": 0.2726, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.964912280701754e-06, | |
| "loss": 0.2959, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.956140350877193e-06, | |
| "loss": 0.2727, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.9473684210526315e-06, | |
| "loss": 0.2661, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.93859649122807e-06, | |
| "loss": 0.2807, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.929824561403509e-06, | |
| "loss": 0.3189, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.9210526315789474e-06, | |
| "loss": 0.2555, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.912280701754386e-06, | |
| "loss": 0.268, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.9035087719298243e-06, | |
| "loss": 0.2828, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.9035087719298243e-06, | |
| "loss": 0.2678, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.894736842105263e-06, | |
| "loss": 0.3396, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.8859649122807019e-06, | |
| "loss": 0.2819, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.8771929824561403e-06, | |
| "loss": 0.2698, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.868421052631579e-06, | |
| "loss": 0.2874, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.8596491228070175e-06, | |
| "loss": 0.2741, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.850877192982456e-06, | |
| "loss": 0.2739, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.8421052631578946e-06, | |
| "loss": 0.2775, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.833333333333333e-06, | |
| "loss": 0.255, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.8245614035087718e-06, | |
| "loss": 0.2734, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.8157894736842106e-06, | |
| "loss": 0.282, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.807017543859649e-06, | |
| "loss": 0.2776, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.7982456140350878e-06, | |
| "loss": 0.3005, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.7894736842105262e-06, | |
| "loss": 0.2459, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.7807017543859647e-06, | |
| "loss": 0.2693, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.7719298245614034e-06, | |
| "loss": 0.2477, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.7631578947368419e-06, | |
| "loss": 0.2351, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.7543859649122805e-06, | |
| "loss": 0.2539, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.7456140350877194e-06, | |
| "loss": 0.3251, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.7368421052631579e-06, | |
| "loss": 0.2607, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.7280701754385966e-06, | |
| "loss": 0.3042, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.719298245614035e-06, | |
| "loss": 0.2845, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.7105263157894735e-06, | |
| "loss": 0.2864, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.7017543859649122e-06, | |
| "loss": 0.3151, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.6929824561403506e-06, | |
| "loss": 0.2837, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.6842105263157893e-06, | |
| "loss": 0.209, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.6754385964912282e-06, | |
| "loss": 0.2971, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.2583, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.6578947368421053e-06, | |
| "loss": 0.2502, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.6491228070175438e-06, | |
| "loss": 0.2788, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.6403508771929823e-06, | |
| "loss": 0.2665, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.631578947368421e-06, | |
| "loss": 0.2678, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.6228070175438594e-06, | |
| "loss": 0.2825, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.614035087719298e-06, | |
| "loss": 0.2598, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.605263157894737e-06, | |
| "loss": 0.2859, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.5964912280701754e-06, | |
| "loss": 0.2537, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.5877192982456141e-06, | |
| "loss": 0.2606, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.5789473684210526e-06, | |
| "loss": 0.2414, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.570175438596491e-06, | |
| "loss": 0.2067, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.570175438596491e-06, | |
| "loss": 0.2452, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.5614035087719297e-06, | |
| "loss": 0.2199, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.5526315789473682e-06, | |
| "loss": 0.2369, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.5438596491228069e-06, | |
| "loss": 0.2181, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.5350877192982458e-06, | |
| "loss": 0.2451, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.5263157894736842e-06, | |
| "loss": 0.2935, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.517543859649123e-06, | |
| "loss": 0.1888, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.5087719298245614e-06, | |
| "loss": 0.2862, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.2385, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.4912280701754385e-06, | |
| "loss": 0.2121, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.482456140350877e-06, | |
| "loss": 0.2498, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.4736842105263156e-06, | |
| "loss": 0.2434, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.4649122807017541e-06, | |
| "loss": 0.2384, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.456140350877193e-06, | |
| "loss": 0.2106, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.4473684210526317e-06, | |
| "loss": 0.2242, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.4385964912280701e-06, | |
| "loss": 0.2526, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.4298245614035088e-06, | |
| "loss": 0.2346, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.4210526315789473e-06, | |
| "loss": 0.2705, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.4122807017543857e-06, | |
| "loss": 0.243, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.4035087719298244e-06, | |
| "loss": 0.2579, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.3947368421052629e-06, | |
| "loss": 0.2334, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.3859649122807018e-06, | |
| "loss": 0.245, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.3771929824561405e-06, | |
| "loss": 0.2699, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.368421052631579e-06, | |
| "loss": 0.2494, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.3596491228070176e-06, | |
| "loss": 0.2337, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.350877192982456e-06, | |
| "loss": 0.2357, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.3421052631578945e-06, | |
| "loss": 0.2686, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.3333333333333332e-06, | |
| "loss": 0.3018, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.3245614035087717e-06, | |
| "loss": 0.2688, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.3157894736842106e-06, | |
| "loss": 0.2339, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.3070175438596492e-06, | |
| "loss": 0.2525, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.2982456140350877e-06, | |
| "loss": 0.2433, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.2894736842105264e-06, | |
| "loss": 0.2509, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.2807017543859648e-06, | |
| "loss": 0.2669, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.2719298245614033e-06, | |
| "loss": 0.2289, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.263157894736842e-06, | |
| "loss": 0.2775, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.2543859649122804e-06, | |
| "loss": 0.2583, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.2456140350877193e-06, | |
| "loss": 0.2357, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.236842105263158e-06, | |
| "loss": 0.2591, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.2280701754385965e-06, | |
| "loss": 0.226, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.2192982456140352e-06, | |
| "loss": 0.2259, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.2105263157894736e-06, | |
| "loss": 0.2551, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1.201754385964912e-06, | |
| "loss": 0.2768, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1.1929824561403508e-06, | |
| "loss": 0.255, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 1.1842105263157892e-06, | |
| "loss": 0.2876, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.1754385964912281e-06, | |
| "loss": 0.2544, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.1666666666666668e-06, | |
| "loss": 0.3147, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.1578947368421053e-06, | |
| "loss": 0.297, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.149122807017544e-06, | |
| "loss": 0.1879, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.1403508771929824e-06, | |
| "loss": 0.2801, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.1315789473684209e-06, | |
| "loss": 0.2638, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.1228070175438595e-06, | |
| "loss": 0.3002, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.114035087719298e-06, | |
| "loss": 0.249, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.1052631578947369e-06, | |
| "loss": 0.2754, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.0964912280701756e-06, | |
| "loss": 0.2535, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.087719298245614e-06, | |
| "loss": 0.2581, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.0789473684210527e-06, | |
| "loss": 0.2437, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1.0701754385964912e-06, | |
| "loss": 0.2423, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 1.0614035087719296e-06, | |
| "loss": 0.2338, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1.0526315789473683e-06, | |
| "loss": 0.2777, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 1.0438596491228068e-06, | |
| "loss": 0.2216, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.0350877192982457e-06, | |
| "loss": 0.2526, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.0263157894736843e-06, | |
| "loss": 0.3199, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.0175438596491228e-06, | |
| "loss": 0.2716, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.0087719298245615e-06, | |
| "loss": 0.2898, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.0087719298245615e-06, | |
| "loss": 0.2678, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1e-06, | |
| "loss": 0.2756, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 9.912280701754384e-07, | |
| "loss": 0.2611, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 9.82456140350877e-07, | |
| "loss": 0.2686, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 9.736842105263158e-07, | |
| "loss": 0.2645, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 9.649122807017545e-07, | |
| "loss": 0.1975, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 9.56140350877193e-07, | |
| "loss": 0.2469, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 9.473684210526315e-07, | |
| "loss": 0.2348, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 9.385964912280702e-07, | |
| "loss": 0.2585, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 9.298245614035087e-07, | |
| "loss": 0.2635, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 9.210526315789473e-07, | |
| "loss": 0.233, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 9.122807017543859e-07, | |
| "loss": 0.2699, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 9.035087719298246e-07, | |
| "loss": 0.2968, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.947368421052631e-07, | |
| "loss": 0.2544, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.859649122807017e-07, | |
| "loss": 0.2103, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.771929824561403e-07, | |
| "loss": 0.1874, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 8.684210526315789e-07, | |
| "loss": 0.2223, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 8.596491228070175e-07, | |
| "loss": 0.201, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 8.508771929824561e-07, | |
| "loss": 0.1785, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 8.421052631578947e-07, | |
| "loss": 0.2338, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 0.1865, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 8.245614035087719e-07, | |
| "loss": 0.2021, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 8.157894736842105e-07, | |
| "loss": 0.2318, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 8.07017543859649e-07, | |
| "loss": 0.2243, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 7.982456140350877e-07, | |
| "loss": 0.1886, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 7.894736842105263e-07, | |
| "loss": 0.2154, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 7.807017543859649e-07, | |
| "loss": 0.2014, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 7.719298245614034e-07, | |
| "loss": 0.1973, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 7.631578947368421e-07, | |
| "loss": 0.225, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 7.543859649122807e-07, | |
| "loss": 0.2228, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 7.456140350877193e-07, | |
| "loss": 0.2187, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 7.368421052631578e-07, | |
| "loss": 0.2125, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 7.280701754385965e-07, | |
| "loss": 0.2051, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 7.192982456140351e-07, | |
| "loss": 0.1871, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 7.105263157894736e-07, | |
| "loss": 0.1799, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 7.017543859649122e-07, | |
| "loss": 0.2057, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 6.929824561403509e-07, | |
| "loss": 0.2006, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 6.842105263157895e-07, | |
| "loss": 0.2205, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 6.75438596491228e-07, | |
| "loss": 0.1811, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 6.666666666666666e-07, | |
| "loss": 0.2211, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 6.578947368421053e-07, | |
| "loss": 0.2089, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 6.491228070175438e-07, | |
| "loss": 0.2244, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 6.403508771929824e-07, | |
| "loss": 0.2272, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 6.31578947368421e-07, | |
| "loss": 0.2152, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 6.228070175438597e-07, | |
| "loss": 0.2262, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 6.140350877192982e-07, | |
| "loss": 0.1981, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 6.052631578947368e-07, | |
| "loss": 0.1936, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 5.964912280701754e-07, | |
| "loss": 0.1912, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 5.877192982456141e-07, | |
| "loss": 0.2036, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 5.789473684210526e-07, | |
| "loss": 0.1765, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 5.701754385964912e-07, | |
| "loss": 0.2102, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 5.614035087719298e-07, | |
| "loss": 0.2136, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 5.526315789473684e-07, | |
| "loss": 0.2049, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 5.43859649122807e-07, | |
| "loss": 0.2316, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 5.350877192982456e-07, | |
| "loss": 0.2, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 5.263157894736842e-07, | |
| "loss": 0.2129, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 5.175438596491228e-07, | |
| "loss": 0.2697, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 5.087719298245614e-07, | |
| "loss": 0.1862, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 5e-07, | |
| "loss": 0.1976, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 4.912280701754385e-07, | |
| "loss": 0.2277, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.824561403508772e-07, | |
| "loss": 0.2158, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.7368421052631574e-07, | |
| "loss": 0.2192, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.6491228070175437e-07, | |
| "loss": 0.2186, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 4.5614035087719294e-07, | |
| "loss": 0.2744, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 4.4736842105263156e-07, | |
| "loss": 0.199, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 4.3859649122807013e-07, | |
| "loss": 0.1962, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.2982456140350876e-07, | |
| "loss": 0.2287, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 4.2105263157894733e-07, | |
| "loss": 0.2447, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 4.1228070175438595e-07, | |
| "loss": 0.2288, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.035087719298245e-07, | |
| "loss": 0.2167, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 3.9473684210526315e-07, | |
| "loss": 0.2, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 3.859649122807017e-07, | |
| "loss": 0.2161, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 3.7719298245614034e-07, | |
| "loss": 0.2171, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 3.684210526315789e-07, | |
| "loss": 0.235, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 3.5964912280701754e-07, | |
| "loss": 0.2001, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.508771929824561e-07, | |
| "loss": 0.2502, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.4210526315789473e-07, | |
| "loss": 0.2161, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.333333333333333e-07, | |
| "loss": 0.247, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.245614035087719e-07, | |
| "loss": 0.2014, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 3.157894736842105e-07, | |
| "loss": 0.2266, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 3.070175438596491e-07, | |
| "loss": 0.2247, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.982456140350877e-07, | |
| "loss": 0.2291, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.894736842105263e-07, | |
| "loss": 0.2512, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.807017543859649e-07, | |
| "loss": 0.215, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 2.719298245614035e-07, | |
| "loss": 0.2, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 2.631578947368421e-07, | |
| "loss": 0.2471, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 2.543859649122807e-07, | |
| "loss": 0.2177, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 2.456140350877193e-07, | |
| "loss": 0.2436, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 2.3684210526315787e-07, | |
| "loss": 0.2244, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 2.2807017543859647e-07, | |
| "loss": 0.2251, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 2.1929824561403507e-07, | |
| "loss": 0.2408, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 2.1052631578947366e-07, | |
| "loss": 0.2238, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "step": 240, | |
| "total_flos": 5.43377784429783e+21, | |
| "train_loss": 0.2554052925979098, | |
| "train_runtime": 4871.4651, | |
| "train_samples_per_second": 11.139, | |
| "train_steps_per_second": 0.049 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 240, | |
| "num_train_epochs": 3, | |
| "save_steps": 1000, | |
| "total_flos": 5.43377784429783e+21, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |