| { | |
| "best_metric": 1.445023536682129, | |
| "best_model_checkpoint": "/home/co-ou1/rds/hpc-work/models/bart_large_gov/checkpoint-545", | |
| "epoch": 19.935868071461293, | |
| "eval_steps": 500, | |
| "global_step": 2720, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.996323529411765e-05, | |
| "loss": 2.9333, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.99264705882353e-05, | |
| "loss": 2.467, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9889705882352945e-05, | |
| "loss": 2.1631, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.985294117647059e-05, | |
| "loss": 1.9596, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.9816176470588236e-05, | |
| "loss": 1.9644, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.977941176470589e-05, | |
| "loss": 1.9089, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.9742647058823534e-05, | |
| "loss": 1.9869, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.970588235294118e-05, | |
| "loss": 1.9179, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9669117647058824e-05, | |
| "loss": 1.9178, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9632352941176476e-05, | |
| "loss": 1.8694, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.959558823529412e-05, | |
| "loss": 1.8608, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.955882352941177e-05, | |
| "loss": 1.8558, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.952205882352941e-05, | |
| "loss": 1.7973, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.9485294117647065e-05, | |
| "loss": 1.819, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.944852941176471e-05, | |
| "loss": 1.8119, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.9411764705882355e-05, | |
| "loss": 1.8298, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.937500000000001e-05, | |
| "loss": 1.722, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.933823529411765e-05, | |
| "loss": 1.7837, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.93014705882353e-05, | |
| "loss": 1.7793, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.9264705882352944e-05, | |
| "loss": 1.7033, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.9227941176470596e-05, | |
| "loss": 1.7449, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.9191176470588234e-05, | |
| "loss": 1.7904, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.915441176470588e-05, | |
| "loss": 1.7559, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.911764705882353e-05, | |
| "loss": 1.7203, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.908088235294118e-05, | |
| "loss": 1.7266, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.904411764705882e-05, | |
| "loss": 1.707, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.900735294117647e-05, | |
| "loss": 1.777, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.897058823529412e-05, | |
| "loss": 1.7624, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.8933823529411766e-05, | |
| "loss": 1.7569, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.889705882352941e-05, | |
| "loss": 1.7643, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.886029411764706e-05, | |
| "loss": 1.7315, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.882352941176471e-05, | |
| "loss": 1.774, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.8786764705882354e-05, | |
| "loss": 1.7839, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.875e-05, | |
| "loss": 1.6725, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.871323529411765e-05, | |
| "loss": 1.7551, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.86764705882353e-05, | |
| "loss": 1.7377, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.863970588235294e-05, | |
| "loss": 1.6739, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.860294117647059e-05, | |
| "loss": 1.7351, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.856617647058824e-05, | |
| "loss": 1.6912, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.8529411764705885e-05, | |
| "loss": 1.63, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.849264705882353e-05, | |
| "loss": 1.7231, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.8455882352941176e-05, | |
| "loss": 1.741, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.841911764705883e-05, | |
| "loss": 1.6945, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.838235294117647e-05, | |
| "loss": 1.7284, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.834558823529412e-05, | |
| "loss": 1.7246, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.830882352941177e-05, | |
| "loss": 1.6913, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.8272058823529416e-05, | |
| "loss": 1.6901, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.823529411764706e-05, | |
| "loss": 1.6237, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.819852941176471e-05, | |
| "loss": 1.6793, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.816176470588236e-05, | |
| "loss": 1.7057, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.8125000000000004e-05, | |
| "loss": 1.7349, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.808823529411765e-05, | |
| "loss": 1.7279, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.8051470588235295e-05, | |
| "loss": 1.6742, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.801470588235295e-05, | |
| "loss": 1.6748, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.797794117647059e-05, | |
| "loss": 1.6322, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.794117647058824e-05, | |
| "loss": 1.6596, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.790441176470588e-05, | |
| "loss": 1.6613, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.7867647058823535e-05, | |
| "loss": 1.639, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.783088235294118e-05, | |
| "loss": 1.6603, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.7794117647058826e-05, | |
| "loss": 1.5827, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.775735294117648e-05, | |
| "loss": 1.693, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.7720588235294124e-05, | |
| "loss": 1.6421, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.768382352941177e-05, | |
| "loss": 1.6872, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.7647058823529414e-05, | |
| "loss": 1.6206, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.761029411764706e-05, | |
| "loss": 1.6894, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.7573529411764705e-05, | |
| "loss": 1.6132, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.753676470588235e-05, | |
| "loss": 1.6449, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.75e-05, | |
| "loss": 1.6694, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_gen_len": 139.72530864197532, | |
| "eval_loss": 1.533786654472351, | |
| "eval_rouge1": 54.2061, | |
| "eval_rouge2": 29.3577, | |
| "eval_rougeL": 37.2911, | |
| "eval_rougeLsum": 49.8337, | |
| "eval_runtime": 275.7588, | |
| "eval_samples_per_second": 3.525, | |
| "eval_steps_per_second": 0.442, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.746323529411765e-05, | |
| "loss": 1.5791, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.742647058823529e-05, | |
| "loss": 1.5997, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 4.738970588235294e-05, | |
| "loss": 1.5539, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.735294117647059e-05, | |
| "loss": 1.6384, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.7316176470588236e-05, | |
| "loss": 1.5839, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.727941176470588e-05, | |
| "loss": 1.5375, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.7242647058823534e-05, | |
| "loss": 1.5505, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.720588235294118e-05, | |
| "loss": 1.5994, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 4.7169117647058824e-05, | |
| "loss": 1.6364, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.713235294117647e-05, | |
| "loss": 1.5387, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.709558823529412e-05, | |
| "loss": 1.5614, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.705882352941177e-05, | |
| "loss": 1.572, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.702205882352941e-05, | |
| "loss": 1.5236, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.698529411764706e-05, | |
| "loss": 1.5675, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.694852941176471e-05, | |
| "loss": 1.5518, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.6911764705882356e-05, | |
| "loss": 1.5171, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.6875e-05, | |
| "loss": 1.6097, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.6838235294117646e-05, | |
| "loss": 1.486, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4.68014705882353e-05, | |
| "loss": 1.5021, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 4.6764705882352944e-05, | |
| "loss": 1.5326, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.672794117647059e-05, | |
| "loss": 1.551, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.669117647058824e-05, | |
| "loss": 1.5579, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.665441176470589e-05, | |
| "loss": 1.5516, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.661764705882353e-05, | |
| "loss": 1.5211, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.658088235294118e-05, | |
| "loss": 1.4823, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.654411764705883e-05, | |
| "loss": 1.5625, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.6507352941176475e-05, | |
| "loss": 1.4965, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.647058823529412e-05, | |
| "loss": 1.5516, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.6433823529411766e-05, | |
| "loss": 1.5452, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.639705882352942e-05, | |
| "loss": 1.558, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.636029411764706e-05, | |
| "loss": 1.4985, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 4.632352941176471e-05, | |
| "loss": 1.5132, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.6286764705882354e-05, | |
| "loss": 1.5821, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.6250000000000006e-05, | |
| "loss": 1.5036, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.621323529411765e-05, | |
| "loss": 1.55, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.61764705882353e-05, | |
| "loss": 1.456, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.613970588235295e-05, | |
| "loss": 1.5497, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.6102941176470594e-05, | |
| "loss": 1.4507, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.606617647058823e-05, | |
| "loss": 1.498, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.6029411764705885e-05, | |
| "loss": 1.4755, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.599264705882353e-05, | |
| "loss": 1.5222, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.5955882352941176e-05, | |
| "loss": 1.5111, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.591911764705882e-05, | |
| "loss": 1.4812, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.588235294117647e-05, | |
| "loss": 1.5365, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.584558823529412e-05, | |
| "loss": 1.5188, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.5808823529411764e-05, | |
| "loss": 1.4869, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.577205882352941e-05, | |
| "loss": 1.5374, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.573529411764706e-05, | |
| "loss": 1.5932, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.569852941176471e-05, | |
| "loss": 1.5479, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.566176470588235e-05, | |
| "loss": 1.5238, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 4.5625e-05, | |
| "loss": 1.5053, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 4.558823529411765e-05, | |
| "loss": 1.5273, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.5551470588235295e-05, | |
| "loss": 1.5534, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.551470588235294e-05, | |
| "loss": 1.5814, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.547794117647059e-05, | |
| "loss": 1.517, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.544117647058824e-05, | |
| "loss": 1.521, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.5404411764705883e-05, | |
| "loss": 1.4901, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.536764705882353e-05, | |
| "loss": 1.5161, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.533088235294118e-05, | |
| "loss": 1.5796, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.5294117647058826e-05, | |
| "loss": 1.4371, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.525735294117647e-05, | |
| "loss": 1.5034, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 4.522058823529412e-05, | |
| "loss": 1.5708, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.518382352941177e-05, | |
| "loss": 1.4524, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 4.5147058823529415e-05, | |
| "loss": 1.4931, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.511029411764706e-05, | |
| "loss": 1.5326, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.507352941176471e-05, | |
| "loss": 1.5344, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 4.503676470588236e-05, | |
| "loss": 1.5018, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.5e-05, | |
| "loss": 1.5178, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "eval_gen_len": 128.94753086419752, | |
| "eval_loss": 1.4697880744934082, | |
| "eval_rouge1": 55.6621, | |
| "eval_rouge2": 30.6254, | |
| "eval_rougeL": 38.7491, | |
| "eval_rougeLsum": 51.2934, | |
| "eval_runtime": 225.1875, | |
| "eval_samples_per_second": 4.316, | |
| "eval_steps_per_second": 0.542, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.496323529411765e-05, | |
| "loss": 1.4707, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.49264705882353e-05, | |
| "loss": 1.4499, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 4.4889705882352946e-05, | |
| "loss": 1.4416, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 4.485294117647059e-05, | |
| "loss": 1.4304, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 4.4816176470588236e-05, | |
| "loss": 1.3961, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 4.477941176470589e-05, | |
| "loss": 1.4177, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 4.4742647058823534e-05, | |
| "loss": 1.426, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 4.470588235294118e-05, | |
| "loss": 1.411, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 4.4669117647058825e-05, | |
| "loss": 1.4288, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 4.463235294117648e-05, | |
| "loss": 1.4189, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.459558823529412e-05, | |
| "loss": 1.4396, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.455882352941177e-05, | |
| "loss": 1.448, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 4.452205882352941e-05, | |
| "loss": 1.4147, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.448529411764706e-05, | |
| "loss": 1.4604, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 4.4448529411764704e-05, | |
| "loss": 1.4261, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 4.4411764705882356e-05, | |
| "loss": 1.39, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 4.4375e-05, | |
| "loss": 1.4104, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 4.4338235294117647e-05, | |
| "loss": 1.4143, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 4.430147058823529e-05, | |
| "loss": 1.3803, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 4.4264705882352944e-05, | |
| "loss": 1.4426, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.422794117647059e-05, | |
| "loss": 1.4429, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.4191176470588235e-05, | |
| "loss": 1.4092, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 4.415441176470588e-05, | |
| "loss": 1.4195, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 4.411764705882353e-05, | |
| "loss": 1.4206, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 4.408088235294118e-05, | |
| "loss": 1.4465, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 4.404411764705882e-05, | |
| "loss": 1.4232, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 4.400735294117647e-05, | |
| "loss": 1.3775, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 4.397058823529412e-05, | |
| "loss": 1.4158, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 4.3933823529411766e-05, | |
| "loss": 1.4791, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 4.389705882352941e-05, | |
| "loss": 1.442, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 4.3860294117647063e-05, | |
| "loss": 1.4081, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 4.382352941176471e-05, | |
| "loss": 1.3624, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 4.3786764705882354e-05, | |
| "loss": 1.4339, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 4.375e-05, | |
| "loss": 1.4567, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 4.371323529411765e-05, | |
| "loss": 1.4509, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 4.36764705882353e-05, | |
| "loss": 1.4989, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 4.363970588235294e-05, | |
| "loss": 1.4039, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 4.360294117647059e-05, | |
| "loss": 1.4554, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 4.356617647058824e-05, | |
| "loss": 1.3791, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 4.3529411764705885e-05, | |
| "loss": 1.4329, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.349264705882353e-05, | |
| "loss": 1.414, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.345588235294118e-05, | |
| "loss": 1.4346, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 4.341911764705883e-05, | |
| "loss": 1.3559, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 4.3382352941176474e-05, | |
| "loss": 1.4023, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 4.334558823529412e-05, | |
| "loss": 1.3509, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 4.330882352941177e-05, | |
| "loss": 1.3932, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 4.3272058823529416e-05, | |
| "loss": 1.4956, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.323529411764706e-05, | |
| "loss": 1.4287, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.319852941176471e-05, | |
| "loss": 1.4228, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.316176470588236e-05, | |
| "loss": 1.3941, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.3125000000000005e-05, | |
| "loss": 1.4925, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.308823529411765e-05, | |
| "loss": 1.39, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.3051470588235295e-05, | |
| "loss": 1.3544, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.301470588235295e-05, | |
| "loss": 1.3676, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.297794117647059e-05, | |
| "loss": 1.4096, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.294117647058823e-05, | |
| "loss": 1.4409, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.2904411764705884e-05, | |
| "loss": 1.3851, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.286764705882353e-05, | |
| "loss": 1.4253, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.2830882352941174e-05, | |
| "loss": 1.3793, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 4.2794117647058827e-05, | |
| "loss": 1.4005, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 4.275735294117647e-05, | |
| "loss": 1.463, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 4.272058823529412e-05, | |
| "loss": 1.4231, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 4.268382352941176e-05, | |
| "loss": 1.4883, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.2647058823529415e-05, | |
| "loss": 1.4709, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.261029411764706e-05, | |
| "loss": 1.3951, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 4.2573529411764706e-05, | |
| "loss": 1.4843, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 4.253676470588235e-05, | |
| "loss": 1.4263, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 4.25e-05, | |
| "loss": 1.4208, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_gen_len": 129.56584362139918, | |
| "eval_loss": 1.4486558437347412, | |
| "eval_rouge1": 55.4905, | |
| "eval_rouge2": 30.4201, | |
| "eval_rougeL": 38.412, | |
| "eval_rougeLsum": 51.1108, | |
| "eval_runtime": 255.4888, | |
| "eval_samples_per_second": 3.804, | |
| "eval_steps_per_second": 0.478, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.246323529411765e-05, | |
| "loss": 1.3905, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.2426470588235294e-05, | |
| "loss": 1.3456, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 4.238970588235294e-05, | |
| "loss": 1.3687, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 4.235294117647059e-05, | |
| "loss": 1.326, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 4.231617647058824e-05, | |
| "loss": 1.3037, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 4.227941176470588e-05, | |
| "loss": 1.2895, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 4.2242647058823534e-05, | |
| "loss": 1.2996, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 4.220588235294118e-05, | |
| "loss": 1.3182, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 4.2169117647058825e-05, | |
| "loss": 1.3633, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 4.213235294117647e-05, | |
| "loss": 1.3328, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 4.209558823529412e-05, | |
| "loss": 1.4186, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 4.205882352941177e-05, | |
| "loss": 1.3802, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 4.202205882352941e-05, | |
| "loss": 1.3181, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 4.198529411764706e-05, | |
| "loss": 1.3821, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 4.194852941176471e-05, | |
| "loss": 1.3545, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 4.1911764705882356e-05, | |
| "loss": 1.3861, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 4.1875e-05, | |
| "loss": 1.2915, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 4.1838235294117654e-05, | |
| "loss": 1.3581, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 4.18014705882353e-05, | |
| "loss": 1.3574, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 4.1764705882352944e-05, | |
| "loss": 1.3463, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 4.172794117647059e-05, | |
| "loss": 1.3353, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 4.169117647058824e-05, | |
| "loss": 1.3506, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 4.165441176470589e-05, | |
| "loss": 1.3615, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 4.161764705882353e-05, | |
| "loss": 1.3299, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 4.158088235294118e-05, | |
| "loss": 1.3412, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 4.154411764705883e-05, | |
| "loss": 1.3079, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 4.1507352941176475e-05, | |
| "loss": 1.3701, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 4.147058823529412e-05, | |
| "loss": 1.3815, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 4.1433823529411766e-05, | |
| "loss": 1.4218, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 4.139705882352941e-05, | |
| "loss": 1.3304, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 4.136029411764706e-05, | |
| "loss": 1.3453, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 4.13235294117647e-05, | |
| "loss": 1.3061, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 4.1286764705882354e-05, | |
| "loss": 1.3804, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 4.125e-05, | |
| "loss": 1.3491, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 4.1213235294117645e-05, | |
| "loss": 1.357, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 4.11764705882353e-05, | |
| "loss": 1.3392, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 4.113970588235294e-05, | |
| "loss": 1.3016, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 4.110294117647059e-05, | |
| "loss": 1.3255, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 4.1066176470588233e-05, | |
| "loss": 1.2793, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 4.1029411764705886e-05, | |
| "loss": 1.3481, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 4.099264705882353e-05, | |
| "loss": 1.337, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 4.0955882352941176e-05, | |
| "loss": 1.3136, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 4.091911764705882e-05, | |
| "loss": 1.3408, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 4.0882352941176474e-05, | |
| "loss": 1.3054, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 4.084558823529412e-05, | |
| "loss": 1.3519, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 4.0808823529411765e-05, | |
| "loss": 1.3548, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 4.077205882352941e-05, | |
| "loss": 1.3478, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 4.073529411764706e-05, | |
| "loss": 1.325, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 4.069852941176471e-05, | |
| "loss": 1.3395, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 4.066176470588235e-05, | |
| "loss": 1.3672, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 4.0625000000000005e-05, | |
| "loss": 1.2985, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 4.058823529411765e-05, | |
| "loss": 1.3651, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 4.0551470588235296e-05, | |
| "loss": 1.3319, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 4.051470588235294e-05, | |
| "loss": 1.3808, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 4.047794117647059e-05, | |
| "loss": 1.3148, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 4.044117647058824e-05, | |
| "loss": 1.3408, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 4.0404411764705884e-05, | |
| "loss": 1.3475, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 4.036764705882353e-05, | |
| "loss": 1.2896, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 4.033088235294118e-05, | |
| "loss": 1.319, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 4.029411764705883e-05, | |
| "loss": 1.3878, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 4.025735294117647e-05, | |
| "loss": 1.3426, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 4.0220588235294124e-05, | |
| "loss": 1.3433, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 4.018382352941177e-05, | |
| "loss": 1.3111, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 4.0147058823529415e-05, | |
| "loss": 1.3679, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 4.011029411764706e-05, | |
| "loss": 1.342, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 4.007352941176471e-05, | |
| "loss": 1.3492, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 4.003676470588236e-05, | |
| "loss": 1.3343, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 4e-05, | |
| "loss": 1.3399, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "eval_gen_len": 128.50617283950618, | |
| "eval_loss": 1.445023536682129, | |
| "eval_rouge1": 56.2783, | |
| "eval_rouge2": 31.1387, | |
| "eval_rougeL": 39.2121, | |
| "eval_rougeLsum": 51.8068, | |
| "eval_runtime": 215.4293, | |
| "eval_samples_per_second": 4.512, | |
| "eval_steps_per_second": 0.566, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3.996323529411765e-05, | |
| "loss": 1.2389, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.99264705882353e-05, | |
| "loss": 1.3121, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 3.9889705882352946e-05, | |
| "loss": 1.1987, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 3.985294117647059e-05, | |
| "loss": 1.2706, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 3.981617647058824e-05, | |
| "loss": 1.3032, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.977941176470588e-05, | |
| "loss": 1.1786, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 3.974264705882353e-05, | |
| "loss": 1.1998, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.970588235294117e-05, | |
| "loss": 1.2755, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.9669117647058825e-05, | |
| "loss": 1.2423, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.963235294117647e-05, | |
| "loss": 1.2877, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 3.9595588235294116e-05, | |
| "loss": 1.2659, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 3.955882352941177e-05, | |
| "loss": 1.319, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 3.952205882352941e-05, | |
| "loss": 1.2898, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 3.948529411764706e-05, | |
| "loss": 1.298, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 3.9448529411764704e-05, | |
| "loss": 1.2663, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 3.9411764705882356e-05, | |
| "loss": 1.2646, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 3.9375e-05, | |
| "loss": 1.3008, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 3.933823529411765e-05, | |
| "loss": 1.2065, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 3.930147058823529e-05, | |
| "loss": 1.3354, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 3.9264705882352945e-05, | |
| "loss": 1.2798, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 3.922794117647059e-05, | |
| "loss": 1.2848, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 3.9191176470588235e-05, | |
| "loss": 1.3071, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 3.915441176470588e-05, | |
| "loss": 1.2356, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 3.911764705882353e-05, | |
| "loss": 1.2597, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 3.908088235294118e-05, | |
| "loss": 1.2158, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 3.9044117647058823e-05, | |
| "loss": 1.2774, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 3.9007352941176476e-05, | |
| "loss": 1.2969, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 3.897058823529412e-05, | |
| "loss": 1.2796, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 3.8933823529411766e-05, | |
| "loss": 1.222, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 3.889705882352941e-05, | |
| "loss": 1.2782, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 3.8860294117647064e-05, | |
| "loss": 1.2615, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 3.882352941176471e-05, | |
| "loss": 1.2746, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 3.8786764705882355e-05, | |
| "loss": 1.2141, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 3.875e-05, | |
| "loss": 1.333, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 3.871323529411765e-05, | |
| "loss": 1.2931, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 3.86764705882353e-05, | |
| "loss": 1.3037, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 3.863970588235294e-05, | |
| "loss": 1.2758, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 3.8602941176470595e-05, | |
| "loss": 1.2348, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 3.856617647058824e-05, | |
| "loss": 1.3094, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 3.8529411764705886e-05, | |
| "loss": 1.2596, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 3.849264705882353e-05, | |
| "loss": 1.2625, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 3.845588235294118e-05, | |
| "loss": 1.2935, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 3.841911764705883e-05, | |
| "loss": 1.2953, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 3.8382352941176474e-05, | |
| "loss": 1.3194, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 3.834558823529412e-05, | |
| "loss": 1.2109, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 3.830882352941177e-05, | |
| "loss": 1.2562, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 3.827205882352941e-05, | |
| "loss": 1.2706, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 3.8235294117647055e-05, | |
| "loss": 1.2471, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 3.819852941176471e-05, | |
| "loss": 1.2963, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 3.816176470588235e-05, | |
| "loss": 1.3085, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 3.8125e-05, | |
| "loss": 1.3405, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 3.8088235294117644e-05, | |
| "loss": 1.2403, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 3.8051470588235296e-05, | |
| "loss": 1.3096, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 3.801470588235294e-05, | |
| "loss": 1.2445, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 3.7977941176470587e-05, | |
| "loss": 1.2315, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 3.794117647058824e-05, | |
| "loss": 1.2387, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 3.7904411764705884e-05, | |
| "loss": 1.2679, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 3.786764705882353e-05, | |
| "loss": 1.2949, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 3.7830882352941175e-05, | |
| "loss": 1.2746, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 3.779411764705883e-05, | |
| "loss": 1.2897, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 3.775735294117647e-05, | |
| "loss": 1.2759, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 3.772058823529412e-05, | |
| "loss": 1.2655, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 3.768382352941176e-05, | |
| "loss": 1.2816, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 3.7647058823529415e-05, | |
| "loss": 1.2932, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 3.761029411764706e-05, | |
| "loss": 1.2306, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 3.7573529411764706e-05, | |
| "loss": 1.2941, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 3.753676470588235e-05, | |
| "loss": 1.2228, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 3.7500000000000003e-05, | |
| "loss": 1.2257, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3.746323529411765e-05, | |
| "loss": 1.2326, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_gen_len": 129.1358024691358, | |
| "eval_loss": 1.4477734565734863, | |
| "eval_rouge1": 56.0182, | |
| "eval_rouge2": 30.7104, | |
| "eval_rougeL": 38.8337, | |
| "eval_rougeLsum": 51.6162, | |
| "eval_runtime": 220.6236, | |
| "eval_samples_per_second": 4.406, | |
| "eval_steps_per_second": 0.553, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.7426470588235294e-05, | |
| "loss": 1.2188, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.7389705882352946e-05, | |
| "loss": 1.2168, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.735294117647059e-05, | |
| "loss": 1.2027, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 3.731617647058824e-05, | |
| "loss": 1.223, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 3.727941176470588e-05, | |
| "loss": 1.1577, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 3.7242647058823535e-05, | |
| "loss": 1.2015, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 3.720588235294118e-05, | |
| "loss": 1.15, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 3.7169117647058825e-05, | |
| "loss": 1.2414, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 3.713235294117647e-05, | |
| "loss": 1.2267, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 3.709558823529412e-05, | |
| "loss": 1.2259, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 5.16, | |
| "learning_rate": 3.705882352941177e-05, | |
| "loss": 1.2137, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 3.7022058823529414e-05, | |
| "loss": 1.2525, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 3.6985294117647066e-05, | |
| "loss": 1.2228, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "learning_rate": 3.694852941176471e-05, | |
| "loss": 1.2692, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 3.6911764705882356e-05, | |
| "loss": 1.2168, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 3.6875e-05, | |
| "loss": 1.2035, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 5.25, | |
| "learning_rate": 3.6838235294117654e-05, | |
| "loss": 1.1863, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 3.68014705882353e-05, | |
| "loss": 1.2229, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 3.6764705882352945e-05, | |
| "loss": 1.2149, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 3.672794117647059e-05, | |
| "loss": 1.218, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 3.6691176470588235e-05, | |
| "loss": 1.2226, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 5.32, | |
| "learning_rate": 3.665441176470588e-05, | |
| "loss": 1.207, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 5.34, | |
| "learning_rate": 3.6617647058823526e-05, | |
| "loss": 1.2022, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 3.658088235294118e-05, | |
| "loss": 1.1717, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 5.37, | |
| "learning_rate": 3.6544117647058824e-05, | |
| "loss": 1.1861, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 5.38, | |
| "learning_rate": 3.650735294117647e-05, | |
| "loss": 1.2465, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 3.6470588235294114e-05, | |
| "loss": 1.2025, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 3.6433823529411767e-05, | |
| "loss": 1.2396, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 3.639705882352941e-05, | |
| "loss": 1.2238, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 3.636029411764706e-05, | |
| "loss": 1.187, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "learning_rate": 3.632352941176471e-05, | |
| "loss": 1.2013, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 3.6286764705882355e-05, | |
| "loss": 1.2093, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 3.625e-05, | |
| "loss": 1.2343, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 5.5, | |
| "learning_rate": 3.6213235294117646e-05, | |
| "loss": 1.2589, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 3.61764705882353e-05, | |
| "loss": 1.2349, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 3.613970588235294e-05, | |
| "loss": 1.1512, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 5.54, | |
| "learning_rate": 3.610294117647059e-05, | |
| "loss": 1.2158, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 3.6066176470588234e-05, | |
| "loss": 1.1684, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 5.57, | |
| "learning_rate": 3.6029411764705886e-05, | |
| "loss": 1.255, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 3.599264705882353e-05, | |
| "loss": 1.2116, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 3.595588235294118e-05, | |
| "loss": 1.2747, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "learning_rate": 3.591911764705882e-05, | |
| "loss": 1.1318, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 3.5882352941176474e-05, | |
| "loss": 1.1944, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 3.584558823529412e-05, | |
| "loss": 1.2051, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 3.5808823529411765e-05, | |
| "loss": 1.2404, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 3.577205882352942e-05, | |
| "loss": 1.1943, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "learning_rate": 3.573529411764706e-05, | |
| "loss": 1.1969, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 3.569852941176471e-05, | |
| "loss": 1.1994, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 3.566176470588235e-05, | |
| "loss": 1.2162, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 5.73, | |
| "learning_rate": 3.5625000000000005e-05, | |
| "loss": 1.2033, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 3.558823529411765e-05, | |
| "loss": 1.1769, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 5.76, | |
| "learning_rate": 3.5551470588235296e-05, | |
| "loss": 1.1423, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 3.551470588235294e-05, | |
| "loss": 1.2333, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 3.5477941176470594e-05, | |
| "loss": 1.2148, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 5.8, | |
| "learning_rate": 3.544117647058824e-05, | |
| "loss": 1.1786, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 3.5404411764705884e-05, | |
| "loss": 1.1703, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 3.5367647058823536e-05, | |
| "loss": 1.1893, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 3.533088235294118e-05, | |
| "loss": 1.1783, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 3.529411764705883e-05, | |
| "loss": 1.1627, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "learning_rate": 3.525735294117647e-05, | |
| "loss": 1.2095, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 3.5220588235294125e-05, | |
| "loss": 1.1953, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 3.518382352941177e-05, | |
| "loss": 1.1971, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 5.92, | |
| "learning_rate": 3.514705882352941e-05, | |
| "loss": 1.2699, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 3.511029411764706e-05, | |
| "loss": 1.2029, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 3.5073529411764706e-05, | |
| "loss": 1.2145, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 3.503676470588235e-05, | |
| "loss": 1.2096, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "learning_rate": 3.5e-05, | |
| "loss": 1.2102, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 3.496323529411765e-05, | |
| "loss": 1.1784, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_gen_len": 128.7314814814815, | |
| "eval_loss": 1.4533237218856812, | |
| "eval_rouge1": 56.4333, | |
| "eval_rouge2": 31.4483, | |
| "eval_rougeL": 39.5546, | |
| "eval_rougeLsum": 52.1347, | |
| "eval_runtime": 216.1787, | |
| "eval_samples_per_second": 4.496, | |
| "eval_steps_per_second": 0.564, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 3.4926470588235294e-05, | |
| "loss": 1.1578, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 3.488970588235294e-05, | |
| "loss": 1.1257, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 3.4852941176470585e-05, | |
| "loss": 1.1946, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 3.481617647058824e-05, | |
| "loss": 1.1224, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 3.477941176470588e-05, | |
| "loss": 1.1055, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 3.474264705882353e-05, | |
| "loss": 1.1109, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 3.470588235294118e-05, | |
| "loss": 1.1532, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 3.4669117647058826e-05, | |
| "loss": 1.0739, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 3.463235294117647e-05, | |
| "loss": 1.1052, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 6.14, | |
| "learning_rate": 3.4595588235294116e-05, | |
| "loss": 1.1529, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 6.16, | |
| "learning_rate": 3.455882352941177e-05, | |
| "loss": 1.1235, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 6.17, | |
| "learning_rate": 3.4522058823529414e-05, | |
| "loss": 1.1584, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 6.19, | |
| "learning_rate": 3.448529411764706e-05, | |
| "loss": 1.1058, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 3.4448529411764705e-05, | |
| "loss": 1.2066, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 3.441176470588236e-05, | |
| "loss": 1.1805, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 6.23, | |
| "learning_rate": 3.4375e-05, | |
| "loss": 1.1599, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 3.433823529411765e-05, | |
| "loss": 1.1682, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 6.26, | |
| "learning_rate": 3.430147058823529e-05, | |
| "loss": 1.21, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 6.27, | |
| "learning_rate": 3.4264705882352945e-05, | |
| "loss": 1.1669, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "learning_rate": 3.422794117647059e-05, | |
| "loss": 1.1205, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 6.3, | |
| "learning_rate": 3.4191176470588236e-05, | |
| "loss": 1.1411, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "learning_rate": 3.415441176470589e-05, | |
| "loss": 1.1476, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 6.33, | |
| "learning_rate": 3.411764705882353e-05, | |
| "loss": 1.1084, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 6.35, | |
| "learning_rate": 3.408088235294118e-05, | |
| "loss": 1.1273, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "learning_rate": 3.4044117647058824e-05, | |
| "loss": 1.119, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 6.38, | |
| "learning_rate": 3.4007352941176476e-05, | |
| "loss": 1.1379, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 3.397058823529412e-05, | |
| "loss": 1.1741, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 6.41, | |
| "learning_rate": 3.393382352941177e-05, | |
| "loss": 1.1785, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 3.389705882352941e-05, | |
| "loss": 1.1535, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 6.44, | |
| "learning_rate": 3.3860294117647064e-05, | |
| "loss": 1.1566, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 6.45, | |
| "learning_rate": 3.382352941176471e-05, | |
| "loss": 1.1341, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 6.46, | |
| "learning_rate": 3.3786764705882355e-05, | |
| "loss": 1.1289, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 6.48, | |
| "learning_rate": 3.375000000000001e-05, | |
| "loss": 1.1759, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "learning_rate": 3.371323529411765e-05, | |
| "loss": 1.1628, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 6.51, | |
| "learning_rate": 3.36764705882353e-05, | |
| "loss": 1.1737, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 3.363970588235294e-05, | |
| "loss": 1.1817, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 6.54, | |
| "learning_rate": 3.3602941176470595e-05, | |
| "loss": 1.1518, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 6.55, | |
| "learning_rate": 3.3566176470588234e-05, | |
| "loss": 1.1865, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 6.57, | |
| "learning_rate": 3.352941176470588e-05, | |
| "loss": 1.1342, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 6.58, | |
| "learning_rate": 3.349264705882353e-05, | |
| "loss": 1.1437, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "learning_rate": 3.345588235294118e-05, | |
| "loss": 1.2004, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 6.61, | |
| "learning_rate": 3.341911764705882e-05, | |
| "loss": 1.148, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 6.63, | |
| "learning_rate": 3.338235294117647e-05, | |
| "loss": 1.1445, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 6.64, | |
| "learning_rate": 3.334558823529412e-05, | |
| "loss": 1.1467, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "learning_rate": 3.3308823529411765e-05, | |
| "loss": 1.1816, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 6.67, | |
| "learning_rate": 3.327205882352941e-05, | |
| "loss": 1.204, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 6.68, | |
| "learning_rate": 3.3235294117647056e-05, | |
| "loss": 1.1821, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 3.319852941176471e-05, | |
| "loss": 1.1906, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 3.3161764705882353e-05, | |
| "loss": 1.138, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 6.73, | |
| "learning_rate": 3.3125e-05, | |
| "loss": 1.1876, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 6.74, | |
| "learning_rate": 3.308823529411765e-05, | |
| "loss": 1.1551, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 6.76, | |
| "learning_rate": 3.3051470588235296e-05, | |
| "loss": 1.1642, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 6.77, | |
| "learning_rate": 3.301470588235294e-05, | |
| "loss": 1.1545, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 6.79, | |
| "learning_rate": 3.297794117647059e-05, | |
| "loss": 1.2166, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 6.8, | |
| "learning_rate": 3.294117647058824e-05, | |
| "loss": 1.1004, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 6.82, | |
| "learning_rate": 3.2904411764705885e-05, | |
| "loss": 1.1681, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "learning_rate": 3.286764705882353e-05, | |
| "loss": 1.1348, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 6.85, | |
| "learning_rate": 3.2830882352941175e-05, | |
| "loss": 1.1448, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 6.86, | |
| "learning_rate": 3.279411764705883e-05, | |
| "loss": 1.1787, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 6.87, | |
| "learning_rate": 3.275735294117647e-05, | |
| "loss": 1.1697, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 3.272058823529412e-05, | |
| "loss": 1.1161, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 6.9, | |
| "learning_rate": 3.2683823529411764e-05, | |
| "loss": 1.1297, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 6.92, | |
| "learning_rate": 3.2647058823529416e-05, | |
| "loss": 1.1287, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "learning_rate": 3.261029411764706e-05, | |
| "loss": 1.1702, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 6.95, | |
| "learning_rate": 3.2573529411764706e-05, | |
| "loss": 1.1213, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 6.96, | |
| "learning_rate": 3.253676470588236e-05, | |
| "loss": 1.1846, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 6.98, | |
| "learning_rate": 3.2500000000000004e-05, | |
| "loss": 1.0957, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 6.99, | |
| "learning_rate": 3.246323529411765e-05, | |
| "loss": 1.1739, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_gen_len": 128.82407407407408, | |
| "eval_loss": 1.4607478380203247, | |
| "eval_rouge1": 56.3636, | |
| "eval_rouge2": 31.1125, | |
| "eval_rougeL": 39.4055, | |
| "eval_rougeLsum": 51.9709, | |
| "eval_runtime": 217.708, | |
| "eval_samples_per_second": 4.465, | |
| "eval_steps_per_second": 0.56, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 3.2426470588235295e-05, | |
| "loss": 1.1437, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 3.238970588235295e-05, | |
| "loss": 1.054, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 3.235294117647059e-05, | |
| "loss": 1.0563, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 3.231617647058824e-05, | |
| "loss": 1.0716, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 3.227941176470588e-05, | |
| "loss": 1.1708, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 3.2242647058823535e-05, | |
| "loss": 1.1274, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 3.220588235294118e-05, | |
| "loss": 1.0927, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 3.2169117647058826e-05, | |
| "loss": 1.0811, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 7.12, | |
| "learning_rate": 3.213235294117647e-05, | |
| "loss": 1.0686, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "learning_rate": 3.209558823529412e-05, | |
| "loss": 1.0668, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 7.15, | |
| "learning_rate": 3.205882352941177e-05, | |
| "loss": 1.0802, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "learning_rate": 3.202205882352941e-05, | |
| "loss": 1.129, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 7.18, | |
| "learning_rate": 3.198529411764706e-05, | |
| "loss": 1.1028, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "learning_rate": 3.1948529411764705e-05, | |
| "loss": 1.069, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 7.21, | |
| "learning_rate": 3.191176470588235e-05, | |
| "loss": 1.0973, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 7.23, | |
| "learning_rate": 3.1875e-05, | |
| "loss": 1.0982, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 7.24, | |
| "learning_rate": 3.183823529411765e-05, | |
| "loss": 1.0913, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 7.26, | |
| "learning_rate": 3.180147058823529e-05, | |
| "loss": 1.072, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 7.27, | |
| "learning_rate": 3.176470588235294e-05, | |
| "loss": 1.1118, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 7.29, | |
| "learning_rate": 3.172794117647059e-05, | |
| "loss": 1.0822, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 7.3, | |
| "learning_rate": 3.1691176470588236e-05, | |
| "loss": 1.1018, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 3.165441176470588e-05, | |
| "loss": 1.0897, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 7.33, | |
| "learning_rate": 3.161764705882353e-05, | |
| "loss": 1.0972, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "learning_rate": 3.158088235294118e-05, | |
| "loss": 1.1716, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 7.36, | |
| "learning_rate": 3.1544117647058824e-05, | |
| "loss": 1.06, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 7.37, | |
| "learning_rate": 3.150735294117647e-05, | |
| "loss": 1.1046, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 7.39, | |
| "learning_rate": 3.147058823529412e-05, | |
| "loss": 1.0623, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 7.4, | |
| "learning_rate": 3.143382352941177e-05, | |
| "loss": 1.1041, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 7.42, | |
| "learning_rate": 3.139705882352941e-05, | |
| "loss": 1.0833, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 7.43, | |
| "learning_rate": 3.136029411764706e-05, | |
| "loss": 1.1318, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 7.45, | |
| "learning_rate": 3.132352941176471e-05, | |
| "loss": 1.0885, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "learning_rate": 3.1286764705882355e-05, | |
| "loss": 1.0923, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 7.48, | |
| "learning_rate": 3.125e-05, | |
| "loss": 1.0532, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 7.49, | |
| "learning_rate": 3.1213235294117646e-05, | |
| "loss": 1.1709, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "learning_rate": 3.11764705882353e-05, | |
| "loss": 1.0687, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 7.52, | |
| "learning_rate": 3.1139705882352944e-05, | |
| "loss": 1.0702, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 3.110294117647059e-05, | |
| "loss": 1.1208, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 3.1066176470588234e-05, | |
| "loss": 1.1507, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 7.56, | |
| "learning_rate": 3.1029411764705886e-05, | |
| "loss": 1.1507, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 7.58, | |
| "learning_rate": 3.099264705882353e-05, | |
| "loss": 1.1436, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 7.59, | |
| "learning_rate": 3.095588235294118e-05, | |
| "loss": 1.1445, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 7.61, | |
| "learning_rate": 3.091911764705883e-05, | |
| "loss": 1.0973, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 7.62, | |
| "learning_rate": 3.0882352941176475e-05, | |
| "loss": 1.0896, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 7.64, | |
| "learning_rate": 3.084558823529412e-05, | |
| "loss": 1.1257, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 7.65, | |
| "learning_rate": 3.0808823529411765e-05, | |
| "loss": 1.0905, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 7.67, | |
| "learning_rate": 3.077205882352942e-05, | |
| "loss": 1.0523, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "learning_rate": 3.073529411764706e-05, | |
| "loss": 1.1089, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 7.7, | |
| "learning_rate": 3.069852941176471e-05, | |
| "loss": 1.1549, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 7.71, | |
| "learning_rate": 3.0661764705882354e-05, | |
| "loss": 1.1035, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 7.73, | |
| "learning_rate": 3.0625000000000006e-05, | |
| "loss": 1.1345, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 7.74, | |
| "learning_rate": 3.058823529411765e-05, | |
| "loss": 1.1102, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 7.75, | |
| "learning_rate": 3.0551470588235297e-05, | |
| "loss": 1.0805, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 3.0514705882352945e-05, | |
| "loss": 1.0786, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 3.0477941176470594e-05, | |
| "loss": 1.1292, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 7.8, | |
| "learning_rate": 3.0441176470588233e-05, | |
| "loss": 1.0522, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "learning_rate": 3.040441176470588e-05, | |
| "loss": 1.1068, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 7.83, | |
| "learning_rate": 3.036764705882353e-05, | |
| "loss": 1.1305, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 7.84, | |
| "learning_rate": 3.0330882352941176e-05, | |
| "loss": 1.1219, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "learning_rate": 3.0294117647058824e-05, | |
| "loss": 1.1672, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 7.87, | |
| "learning_rate": 3.025735294117647e-05, | |
| "loss": 1.062, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 7.89, | |
| "learning_rate": 3.022058823529412e-05, | |
| "loss": 1.0983, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 7.9, | |
| "learning_rate": 3.0183823529411764e-05, | |
| "loss": 1.158, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "learning_rate": 3.0147058823529413e-05, | |
| "loss": 1.0447, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 7.93, | |
| "learning_rate": 3.0110294117647058e-05, | |
| "loss": 1.1328, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 7.95, | |
| "learning_rate": 3.0073529411764707e-05, | |
| "loss": 1.1709, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 7.96, | |
| "learning_rate": 3.0036764705882352e-05, | |
| "loss": 1.0755, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 7.97, | |
| "learning_rate": 3e-05, | |
| "loss": 1.1316, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 7.99, | |
| "learning_rate": 2.9963235294117646e-05, | |
| "loss": 1.1585, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_gen_len": 128.96399176954733, | |
| "eval_loss": 1.477362871170044, | |
| "eval_rouge1": 55.9356, | |
| "eval_rouge2": 30.7012, | |
| "eval_rougeL": 38.7824, | |
| "eval_rougeLsum": 51.5664, | |
| "eval_runtime": 219.0461, | |
| "eval_samples_per_second": 4.437, | |
| "eval_steps_per_second": 0.557, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 2.9926470588235295e-05, | |
| "loss": 1.064, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 2.988970588235294e-05, | |
| "loss": 1.0127, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 2.985294117647059e-05, | |
| "loss": 1.041, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 8.05, | |
| "learning_rate": 2.9816176470588238e-05, | |
| "loss": 1.0275, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 2.9779411764705883e-05, | |
| "loss": 1.0411, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 2.9742647058823532e-05, | |
| "loss": 0.9834, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 8.09, | |
| "learning_rate": 2.9705882352941177e-05, | |
| "loss": 1.062, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 8.11, | |
| "learning_rate": 2.9669117647058826e-05, | |
| "loss": 1.0085, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 8.12, | |
| "learning_rate": 2.963235294117647e-05, | |
| "loss": 1.0523, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 8.14, | |
| "learning_rate": 2.959558823529412e-05, | |
| "loss": 1.0417, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 8.15, | |
| "learning_rate": 2.9558823529411766e-05, | |
| "loss": 1.0556, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 8.16, | |
| "learning_rate": 2.9522058823529414e-05, | |
| "loss": 1.0554, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 2.948529411764706e-05, | |
| "loss": 1.0356, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 8.19, | |
| "learning_rate": 2.944852941176471e-05, | |
| "loss": 1.0252, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 8.21, | |
| "learning_rate": 2.9411764705882354e-05, | |
| "loss": 1.026, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 8.22, | |
| "learning_rate": 2.9375000000000003e-05, | |
| "loss": 1.0843, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 8.24, | |
| "learning_rate": 2.933823529411765e-05, | |
| "loss": 1.0634, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 8.25, | |
| "learning_rate": 2.9301470588235297e-05, | |
| "loss": 1.0619, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 8.27, | |
| "learning_rate": 2.9264705882352945e-05, | |
| "loss": 1.0327, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 8.28, | |
| "learning_rate": 2.922794117647059e-05, | |
| "loss": 1.0512, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 8.3, | |
| "learning_rate": 2.919117647058824e-05, | |
| "loss": 1.097, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 8.31, | |
| "learning_rate": 2.9154411764705885e-05, | |
| "loss": 1.0198, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 8.33, | |
| "learning_rate": 2.9117647058823534e-05, | |
| "loss": 1.046, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 8.34, | |
| "learning_rate": 2.908088235294118e-05, | |
| "loss": 1.0727, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 8.36, | |
| "learning_rate": 2.9044117647058828e-05, | |
| "loss": 1.0551, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "learning_rate": 2.9007352941176473e-05, | |
| "loss": 1.0817, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 8.38, | |
| "learning_rate": 2.8970588235294122e-05, | |
| "loss": 1.061, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 8.4, | |
| "learning_rate": 2.8933823529411767e-05, | |
| "loss": 1.0196, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 8.41, | |
| "learning_rate": 2.889705882352941e-05, | |
| "loss": 1.0931, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 8.43, | |
| "learning_rate": 2.8860294117647058e-05, | |
| "loss": 1.0846, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 8.44, | |
| "learning_rate": 2.8823529411764703e-05, | |
| "loss": 1.0732, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 8.46, | |
| "learning_rate": 2.8786764705882352e-05, | |
| "loss": 1.0944, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 2.8749999999999997e-05, | |
| "loss": 1.0691, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 8.49, | |
| "learning_rate": 2.8713235294117646e-05, | |
| "loss": 1.0971, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 8.5, | |
| "learning_rate": 2.8676470588235295e-05, | |
| "loss": 1.0711, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 2.863970588235294e-05, | |
| "loss": 1.1213, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 8.53, | |
| "learning_rate": 2.860294117647059e-05, | |
| "loss": 1.0351, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "learning_rate": 2.8566176470588234e-05, | |
| "loss": 1.0728, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 8.56, | |
| "learning_rate": 2.8529411764705883e-05, | |
| "loss": 1.0392, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 8.58, | |
| "learning_rate": 2.849264705882353e-05, | |
| "loss": 1.0651, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 8.59, | |
| "learning_rate": 2.8455882352941177e-05, | |
| "loss": 1.0739, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 8.6, | |
| "learning_rate": 2.8419117647058823e-05, | |
| "loss": 1.0238, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 8.62, | |
| "learning_rate": 2.838235294117647e-05, | |
| "loss": 1.0343, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 8.63, | |
| "learning_rate": 2.8345588235294117e-05, | |
| "loss": 1.085, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 8.65, | |
| "learning_rate": 2.8308823529411766e-05, | |
| "loss": 1.028, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 8.66, | |
| "learning_rate": 2.827205882352941e-05, | |
| "loss": 1.051, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 8.68, | |
| "learning_rate": 2.823529411764706e-05, | |
| "loss": 1.0861, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 8.69, | |
| "learning_rate": 2.819852941176471e-05, | |
| "loss": 1.0662, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "learning_rate": 2.8161764705882354e-05, | |
| "loss": 1.0152, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 8.72, | |
| "learning_rate": 2.8125000000000003e-05, | |
| "loss": 1.0921, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 8.74, | |
| "learning_rate": 2.8088235294117648e-05, | |
| "loss": 1.055, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 8.75, | |
| "learning_rate": 2.8051470588235297e-05, | |
| "loss": 1.091, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 8.77, | |
| "learning_rate": 2.8014705882352942e-05, | |
| "loss": 1.0633, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 8.78, | |
| "learning_rate": 2.797794117647059e-05, | |
| "loss": 1.0423, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 8.8, | |
| "learning_rate": 2.7941176470588236e-05, | |
| "loss": 1.0793, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 8.81, | |
| "learning_rate": 2.7904411764705885e-05, | |
| "loss": 1.0861, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 8.82, | |
| "learning_rate": 2.786764705882353e-05, | |
| "loss": 1.0503, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 8.84, | |
| "learning_rate": 2.783088235294118e-05, | |
| "loss": 1.1103, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 8.85, | |
| "learning_rate": 2.7794117647058824e-05, | |
| "loss": 1.0901, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 8.87, | |
| "learning_rate": 2.7757352941176473e-05, | |
| "loss": 1.0669, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 8.88, | |
| "learning_rate": 2.7720588235294122e-05, | |
| "loss": 1.04, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 8.9, | |
| "learning_rate": 2.7683823529411767e-05, | |
| "loss": 1.0737, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 8.91, | |
| "learning_rate": 2.7647058823529416e-05, | |
| "loss": 1.0484, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 8.93, | |
| "learning_rate": 2.761029411764706e-05, | |
| "loss": 1.0745, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 8.94, | |
| "learning_rate": 2.757352941176471e-05, | |
| "loss": 1.0958, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 8.96, | |
| "learning_rate": 2.7536764705882356e-05, | |
| "loss": 1.0877, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 8.97, | |
| "learning_rate": 2.7500000000000004e-05, | |
| "loss": 1.0696, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 8.99, | |
| "learning_rate": 2.746323529411765e-05, | |
| "loss": 1.0297, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 8.99, | |
| "eval_gen_len": 128.65534979423867, | |
| "eval_loss": 1.4938671588897705, | |
| "eval_rouge1": 56.7487, | |
| "eval_rouge2": 31.552, | |
| "eval_rougeL": 39.6461, | |
| "eval_rougeLsum": 52.411, | |
| "eval_runtime": 219.1861, | |
| "eval_samples_per_second": 4.435, | |
| "eval_steps_per_second": 0.557, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 2.74264705882353e-05, | |
| "loss": 1.0531, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 2.7389705882352944e-05, | |
| "loss": 1.0446, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 2.7352941176470593e-05, | |
| "loss": 1.0474, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 2.7316176470588235e-05, | |
| "loss": 0.9544, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 9.06, | |
| "learning_rate": 2.727941176470588e-05, | |
| "loss": 1.0093, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 9.07, | |
| "learning_rate": 2.724264705882353e-05, | |
| "loss": 0.9961, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 2.7205882352941174e-05, | |
| "loss": 1.0057, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 9.1, | |
| "learning_rate": 2.7169117647058823e-05, | |
| "loss": 1.0148, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 9.12, | |
| "learning_rate": 2.7132352941176468e-05, | |
| "loss": 0.99, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 9.13, | |
| "learning_rate": 2.7095588235294117e-05, | |
| "loss": 1.0108, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 2.7058823529411766e-05, | |
| "loss": 1.0202, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 9.16, | |
| "learning_rate": 2.702205882352941e-05, | |
| "loss": 1.0083, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 9.18, | |
| "learning_rate": 2.698529411764706e-05, | |
| "loss": 1.0384, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 9.19, | |
| "learning_rate": 2.6948529411764705e-05, | |
| "loss": 1.002, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 9.21, | |
| "learning_rate": 2.6911764705882354e-05, | |
| "loss": 1.0202, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "learning_rate": 2.6875e-05, | |
| "loss": 1.0125, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 9.23, | |
| "learning_rate": 2.6838235294117648e-05, | |
| "loss": 0.9531, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 9.25, | |
| "learning_rate": 2.6801470588235293e-05, | |
| "loss": 1.0564, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 9.26, | |
| "learning_rate": 2.6764705882352942e-05, | |
| "loss": 0.973, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 9.28, | |
| "learning_rate": 2.6727941176470588e-05, | |
| "loss": 0.9966, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 9.29, | |
| "learning_rate": 2.6691176470588236e-05, | |
| "loss": 1.0315, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 9.31, | |
| "learning_rate": 2.6654411764705882e-05, | |
| "loss": 1.0552, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 9.32, | |
| "learning_rate": 2.661764705882353e-05, | |
| "loss": 0.9671, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 9.34, | |
| "learning_rate": 2.658088235294118e-05, | |
| "loss": 0.9798, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 9.35, | |
| "learning_rate": 2.6544117647058825e-05, | |
| "loss": 0.9898, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 9.37, | |
| "learning_rate": 2.6507352941176473e-05, | |
| "loss": 1.0526, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 9.38, | |
| "learning_rate": 2.647058823529412e-05, | |
| "loss": 1.0261, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "learning_rate": 2.6433823529411767e-05, | |
| "loss": 0.9646, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 9.41, | |
| "learning_rate": 2.6397058823529413e-05, | |
| "loss": 0.9961, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 9.43, | |
| "learning_rate": 2.636029411764706e-05, | |
| "loss": 1.0317, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 9.44, | |
| "learning_rate": 2.6323529411764707e-05, | |
| "loss": 1.0495, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 9.45, | |
| "learning_rate": 2.6286764705882356e-05, | |
| "loss": 1.0679, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 9.47, | |
| "learning_rate": 2.625e-05, | |
| "loss": 1.0365, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 9.48, | |
| "learning_rate": 2.621323529411765e-05, | |
| "loss": 0.9912, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 9.5, | |
| "learning_rate": 2.6176470588235295e-05, | |
| "loss": 1.0174, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 9.51, | |
| "learning_rate": 2.6139705882352944e-05, | |
| "loss": 1.0361, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 9.53, | |
| "learning_rate": 2.6102941176470593e-05, | |
| "loss": 0.9918, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 9.54, | |
| "learning_rate": 2.6066176470588238e-05, | |
| "loss": 0.9941, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "learning_rate": 2.6029411764705887e-05, | |
| "loss": 1.0732, | |
| "step": 1304 | |
| }, | |
| { | |
| "epoch": 9.57, | |
| "learning_rate": 2.5992647058823532e-05, | |
| "loss": 1.0018, | |
| "step": 1306 | |
| }, | |
| { | |
| "epoch": 9.59, | |
| "learning_rate": 2.595588235294118e-05, | |
| "loss": 0.9767, | |
| "step": 1308 | |
| }, | |
| { | |
| "epoch": 9.6, | |
| "learning_rate": 2.5919117647058826e-05, | |
| "loss": 0.9916, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 9.62, | |
| "learning_rate": 2.5882352941176475e-05, | |
| "loss": 1.0272, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 9.63, | |
| "learning_rate": 2.584558823529412e-05, | |
| "loss": 1.0605, | |
| "step": 1314 | |
| }, | |
| { | |
| "epoch": 9.65, | |
| "learning_rate": 2.580882352941177e-05, | |
| "loss": 0.9939, | |
| "step": 1316 | |
| }, | |
| { | |
| "epoch": 9.66, | |
| "learning_rate": 2.577205882352941e-05, | |
| "loss": 1.0565, | |
| "step": 1318 | |
| }, | |
| { | |
| "epoch": 9.67, | |
| "learning_rate": 2.5735294117647057e-05, | |
| "loss": 1.0013, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 9.69, | |
| "learning_rate": 2.5698529411764705e-05, | |
| "loss": 1.0204, | |
| "step": 1322 | |
| }, | |
| { | |
| "epoch": 9.7, | |
| "learning_rate": 2.566176470588235e-05, | |
| "loss": 1.0265, | |
| "step": 1324 | |
| }, | |
| { | |
| "epoch": 9.72, | |
| "learning_rate": 2.5625e-05, | |
| "loss": 1.0473, | |
| "step": 1326 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "learning_rate": 2.5588235294117645e-05, | |
| "loss": 1.0479, | |
| "step": 1328 | |
| }, | |
| { | |
| "epoch": 9.75, | |
| "learning_rate": 2.5551470588235294e-05, | |
| "loss": 1.0774, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 9.76, | |
| "learning_rate": 2.551470588235294e-05, | |
| "loss": 0.9653, | |
| "step": 1332 | |
| }, | |
| { | |
| "epoch": 9.78, | |
| "learning_rate": 2.5477941176470588e-05, | |
| "loss": 1.0668, | |
| "step": 1334 | |
| }, | |
| { | |
| "epoch": 9.79, | |
| "learning_rate": 2.5441176470588236e-05, | |
| "loss": 1.0469, | |
| "step": 1336 | |
| }, | |
| { | |
| "epoch": 9.81, | |
| "learning_rate": 2.5404411764705882e-05, | |
| "loss": 1.0151, | |
| "step": 1338 | |
| }, | |
| { | |
| "epoch": 9.82, | |
| "learning_rate": 2.536764705882353e-05, | |
| "loss": 0.9794, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 9.84, | |
| "learning_rate": 2.5330882352941176e-05, | |
| "loss": 1.0478, | |
| "step": 1342 | |
| }, | |
| { | |
| "epoch": 9.85, | |
| "learning_rate": 2.5294117647058825e-05, | |
| "loss": 1.0012, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 9.87, | |
| "learning_rate": 2.525735294117647e-05, | |
| "loss": 0.9858, | |
| "step": 1346 | |
| }, | |
| { | |
| "epoch": 9.88, | |
| "learning_rate": 2.522058823529412e-05, | |
| "loss": 1.0248, | |
| "step": 1348 | |
| }, | |
| { | |
| "epoch": 9.89, | |
| "learning_rate": 2.5183823529411764e-05, | |
| "loss": 1.0285, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 9.91, | |
| "learning_rate": 2.5147058823529413e-05, | |
| "loss": 1.0664, | |
| "step": 1352 | |
| }, | |
| { | |
| "epoch": 9.92, | |
| "learning_rate": 2.511029411764706e-05, | |
| "loss": 1.0532, | |
| "step": 1354 | |
| }, | |
| { | |
| "epoch": 9.94, | |
| "learning_rate": 2.5073529411764707e-05, | |
| "loss": 1.0642, | |
| "step": 1356 | |
| }, | |
| { | |
| "epoch": 9.95, | |
| "learning_rate": 2.5036764705882352e-05, | |
| "loss": 1.0381, | |
| "step": 1358 | |
| }, | |
| { | |
| "epoch": 9.97, | |
| "learning_rate": 2.5e-05, | |
| "loss": 0.9934, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 9.98, | |
| "learning_rate": 2.496323529411765e-05, | |
| "loss": 1.0075, | |
| "step": 1362 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 2.4926470588235295e-05, | |
| "loss": 1.0085, | |
| "step": 1364 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_gen_len": 128.62654320987653, | |
| "eval_loss": 1.507472276687622, | |
| "eval_rouge1": 56.3918, | |
| "eval_rouge2": 31.2201, | |
| "eval_rougeL": 39.4213, | |
| "eval_rougeLsum": 51.9449, | |
| "eval_runtime": 217.0015, | |
| "eval_samples_per_second": 4.479, | |
| "eval_steps_per_second": 0.562, | |
| "step": 1364 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 2.4889705882352944e-05, | |
| "loss": 0.9832, | |
| "step": 1366 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 2.485294117647059e-05, | |
| "loss": 0.9834, | |
| "step": 1368 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "learning_rate": 2.4816176470588238e-05, | |
| "loss": 0.939, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 10.06, | |
| "learning_rate": 2.4779411764705884e-05, | |
| "loss": 0.9535, | |
| "step": 1372 | |
| }, | |
| { | |
| "epoch": 10.07, | |
| "learning_rate": 2.4742647058823532e-05, | |
| "loss": 0.941, | |
| "step": 1374 | |
| }, | |
| { | |
| "epoch": 10.09, | |
| "learning_rate": 2.4705882352941178e-05, | |
| "loss": 0.955, | |
| "step": 1376 | |
| }, | |
| { | |
| "epoch": 10.1, | |
| "learning_rate": 2.4669117647058826e-05, | |
| "loss": 0.9496, | |
| "step": 1378 | |
| }, | |
| { | |
| "epoch": 10.11, | |
| "learning_rate": 2.4632352941176472e-05, | |
| "loss": 0.9929, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 10.13, | |
| "learning_rate": 2.4595588235294117e-05, | |
| "loss": 1.0207, | |
| "step": 1382 | |
| }, | |
| { | |
| "epoch": 10.14, | |
| "learning_rate": 2.4558823529411766e-05, | |
| "loss": 0.9971, | |
| "step": 1384 | |
| }, | |
| { | |
| "epoch": 10.16, | |
| "learning_rate": 2.452205882352941e-05, | |
| "loss": 0.9831, | |
| "step": 1386 | |
| }, | |
| { | |
| "epoch": 10.17, | |
| "learning_rate": 2.448529411764706e-05, | |
| "loss": 0.933, | |
| "step": 1388 | |
| }, | |
| { | |
| "epoch": 10.19, | |
| "learning_rate": 2.4448529411764705e-05, | |
| "loss": 0.9764, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 10.2, | |
| "learning_rate": 2.4411764705882354e-05, | |
| "loss": 0.9694, | |
| "step": 1392 | |
| }, | |
| { | |
| "epoch": 10.22, | |
| "learning_rate": 2.4375e-05, | |
| "loss": 0.9751, | |
| "step": 1394 | |
| }, | |
| { | |
| "epoch": 10.23, | |
| "learning_rate": 2.433823529411765e-05, | |
| "loss": 0.9312, | |
| "step": 1396 | |
| }, | |
| { | |
| "epoch": 10.25, | |
| "learning_rate": 2.4301470588235294e-05, | |
| "loss": 0.9863, | |
| "step": 1398 | |
| }, | |
| { | |
| "epoch": 10.26, | |
| "learning_rate": 2.4264705882352942e-05, | |
| "loss": 1.0406, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 10.28, | |
| "learning_rate": 2.4227941176470588e-05, | |
| "loss": 0.95, | |
| "step": 1402 | |
| }, | |
| { | |
| "epoch": 10.29, | |
| "learning_rate": 2.4191176470588237e-05, | |
| "loss": 0.9429, | |
| "step": 1404 | |
| }, | |
| { | |
| "epoch": 10.31, | |
| "learning_rate": 2.4154411764705885e-05, | |
| "loss": 0.985, | |
| "step": 1406 | |
| }, | |
| { | |
| "epoch": 10.32, | |
| "learning_rate": 2.411764705882353e-05, | |
| "loss": 0.9571, | |
| "step": 1408 | |
| }, | |
| { | |
| "epoch": 10.33, | |
| "learning_rate": 2.408088235294118e-05, | |
| "loss": 1.0004, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 10.35, | |
| "learning_rate": 2.4044117647058825e-05, | |
| "loss": 0.9771, | |
| "step": 1412 | |
| }, | |
| { | |
| "epoch": 10.36, | |
| "learning_rate": 2.4007352941176474e-05, | |
| "loss": 1.0061, | |
| "step": 1414 | |
| }, | |
| { | |
| "epoch": 10.38, | |
| "learning_rate": 2.397058823529412e-05, | |
| "loss": 0.9816, | |
| "step": 1416 | |
| }, | |
| { | |
| "epoch": 10.39, | |
| "learning_rate": 2.3933823529411768e-05, | |
| "loss": 1.0354, | |
| "step": 1418 | |
| }, | |
| { | |
| "epoch": 10.41, | |
| "learning_rate": 2.3897058823529413e-05, | |
| "loss": 1.0185, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 10.42, | |
| "learning_rate": 2.3860294117647062e-05, | |
| "loss": 0.9801, | |
| "step": 1422 | |
| }, | |
| { | |
| "epoch": 10.44, | |
| "learning_rate": 2.3823529411764707e-05, | |
| "loss": 0.9537, | |
| "step": 1424 | |
| }, | |
| { | |
| "epoch": 10.45, | |
| "learning_rate": 2.3786764705882353e-05, | |
| "loss": 0.947, | |
| "step": 1426 | |
| }, | |
| { | |
| "epoch": 10.47, | |
| "learning_rate": 2.375e-05, | |
| "loss": 0.9349, | |
| "step": 1428 | |
| }, | |
| { | |
| "epoch": 10.48, | |
| "learning_rate": 2.3713235294117647e-05, | |
| "loss": 0.9977, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 10.5, | |
| "learning_rate": 2.3676470588235295e-05, | |
| "loss": 0.9688, | |
| "step": 1432 | |
| }, | |
| { | |
| "epoch": 10.51, | |
| "learning_rate": 2.363970588235294e-05, | |
| "loss": 0.9573, | |
| "step": 1434 | |
| }, | |
| { | |
| "epoch": 10.52, | |
| "learning_rate": 2.360294117647059e-05, | |
| "loss": 0.9805, | |
| "step": 1436 | |
| }, | |
| { | |
| "epoch": 10.54, | |
| "learning_rate": 2.3566176470588235e-05, | |
| "loss": 0.9624, | |
| "step": 1438 | |
| }, | |
| { | |
| "epoch": 10.55, | |
| "learning_rate": 2.3529411764705884e-05, | |
| "loss": 1.0043, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 10.57, | |
| "learning_rate": 2.349264705882353e-05, | |
| "loss": 1.0019, | |
| "step": 1442 | |
| }, | |
| { | |
| "epoch": 10.58, | |
| "learning_rate": 2.3455882352941178e-05, | |
| "loss": 0.9988, | |
| "step": 1444 | |
| }, | |
| { | |
| "epoch": 10.6, | |
| "learning_rate": 2.3419117647058823e-05, | |
| "loss": 0.9651, | |
| "step": 1446 | |
| }, | |
| { | |
| "epoch": 10.61, | |
| "learning_rate": 2.3382352941176472e-05, | |
| "loss": 0.9789, | |
| "step": 1448 | |
| }, | |
| { | |
| "epoch": 10.63, | |
| "learning_rate": 2.334558823529412e-05, | |
| "loss": 1.0235, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 10.64, | |
| "learning_rate": 2.3308823529411766e-05, | |
| "loss": 0.983, | |
| "step": 1452 | |
| }, | |
| { | |
| "epoch": 10.66, | |
| "learning_rate": 2.3272058823529415e-05, | |
| "loss": 0.9842, | |
| "step": 1454 | |
| }, | |
| { | |
| "epoch": 10.67, | |
| "learning_rate": 2.323529411764706e-05, | |
| "loss": 1.0099, | |
| "step": 1456 | |
| }, | |
| { | |
| "epoch": 10.69, | |
| "learning_rate": 2.319852941176471e-05, | |
| "loss": 0.9968, | |
| "step": 1458 | |
| }, | |
| { | |
| "epoch": 10.7, | |
| "learning_rate": 2.3161764705882354e-05, | |
| "loss": 0.9875, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 10.72, | |
| "learning_rate": 2.3125000000000003e-05, | |
| "loss": 0.9934, | |
| "step": 1462 | |
| }, | |
| { | |
| "epoch": 10.73, | |
| "learning_rate": 2.308823529411765e-05, | |
| "loss": 0.9903, | |
| "step": 1464 | |
| }, | |
| { | |
| "epoch": 10.74, | |
| "learning_rate": 2.3051470588235297e-05, | |
| "loss": 0.9859, | |
| "step": 1466 | |
| }, | |
| { | |
| "epoch": 10.76, | |
| "learning_rate": 2.3014705882352943e-05, | |
| "loss": 0.9909, | |
| "step": 1468 | |
| }, | |
| { | |
| "epoch": 10.77, | |
| "learning_rate": 2.2977941176470588e-05, | |
| "loss": 0.9937, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 10.79, | |
| "learning_rate": 2.2941176470588237e-05, | |
| "loss": 0.9585, | |
| "step": 1472 | |
| }, | |
| { | |
| "epoch": 10.8, | |
| "learning_rate": 2.2904411764705882e-05, | |
| "loss": 0.9539, | |
| "step": 1474 | |
| }, | |
| { | |
| "epoch": 10.82, | |
| "learning_rate": 2.286764705882353e-05, | |
| "loss": 1.0016, | |
| "step": 1476 | |
| }, | |
| { | |
| "epoch": 10.83, | |
| "learning_rate": 2.2830882352941176e-05, | |
| "loss": 0.9746, | |
| "step": 1478 | |
| }, | |
| { | |
| "epoch": 10.85, | |
| "learning_rate": 2.2794117647058825e-05, | |
| "loss": 0.9698, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 10.86, | |
| "learning_rate": 2.275735294117647e-05, | |
| "loss": 1.0281, | |
| "step": 1482 | |
| }, | |
| { | |
| "epoch": 10.88, | |
| "learning_rate": 2.272058823529412e-05, | |
| "loss": 0.9775, | |
| "step": 1484 | |
| }, | |
| { | |
| "epoch": 10.89, | |
| "learning_rate": 2.2683823529411764e-05, | |
| "loss": 0.9692, | |
| "step": 1486 | |
| }, | |
| { | |
| "epoch": 10.91, | |
| "learning_rate": 2.2647058823529413e-05, | |
| "loss": 0.9502, | |
| "step": 1488 | |
| }, | |
| { | |
| "epoch": 10.92, | |
| "learning_rate": 2.261029411764706e-05, | |
| "loss": 1.0112, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 10.94, | |
| "learning_rate": 2.2573529411764707e-05, | |
| "loss": 0.9891, | |
| "step": 1492 | |
| }, | |
| { | |
| "epoch": 10.95, | |
| "learning_rate": 2.2536764705882356e-05, | |
| "loss": 1.0143, | |
| "step": 1494 | |
| }, | |
| { | |
| "epoch": 10.96, | |
| "learning_rate": 2.25e-05, | |
| "loss": 1.0172, | |
| "step": 1496 | |
| }, | |
| { | |
| "epoch": 10.98, | |
| "learning_rate": 2.246323529411765e-05, | |
| "loss": 0.9478, | |
| "step": 1498 | |
| }, | |
| { | |
| "epoch": 10.99, | |
| "learning_rate": 2.2426470588235296e-05, | |
| "loss": 0.9738, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 10.99, | |
| "eval_gen_len": 128.8281893004115, | |
| "eval_loss": 1.5237066745758057, | |
| "eval_rouge1": 56.3041, | |
| "eval_rouge2": 30.9239, | |
| "eval_rougeL": 39.2625, | |
| "eval_rougeLsum": 51.8217, | |
| "eval_runtime": 217.7346, | |
| "eval_samples_per_second": 4.464, | |
| "eval_steps_per_second": 0.56, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 2.2389705882352944e-05, | |
| "loss": 0.9912, | |
| "step": 1502 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 2.235294117647059e-05, | |
| "loss": 0.9189, | |
| "step": 1504 | |
| }, | |
| { | |
| "epoch": 11.04, | |
| "learning_rate": 2.231617647058824e-05, | |
| "loss": 0.9605, | |
| "step": 1506 | |
| }, | |
| { | |
| "epoch": 11.05, | |
| "learning_rate": 2.2279411764705884e-05, | |
| "loss": 0.9208, | |
| "step": 1508 | |
| }, | |
| { | |
| "epoch": 11.07, | |
| "learning_rate": 2.224264705882353e-05, | |
| "loss": 0.9377, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 11.08, | |
| "learning_rate": 2.2205882352941178e-05, | |
| "loss": 0.912, | |
| "step": 1512 | |
| }, | |
| { | |
| "epoch": 11.1, | |
| "learning_rate": 2.2169117647058823e-05, | |
| "loss": 0.9114, | |
| "step": 1514 | |
| }, | |
| { | |
| "epoch": 11.11, | |
| "learning_rate": 2.2132352941176472e-05, | |
| "loss": 0.9191, | |
| "step": 1516 | |
| }, | |
| { | |
| "epoch": 11.13, | |
| "learning_rate": 2.2095588235294117e-05, | |
| "loss": 0.9428, | |
| "step": 1518 | |
| }, | |
| { | |
| "epoch": 11.14, | |
| "learning_rate": 2.2058823529411766e-05, | |
| "loss": 0.9373, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 11.16, | |
| "learning_rate": 2.202205882352941e-05, | |
| "loss": 0.9407, | |
| "step": 1522 | |
| }, | |
| { | |
| "epoch": 11.17, | |
| "learning_rate": 2.198529411764706e-05, | |
| "loss": 0.9159, | |
| "step": 1524 | |
| }, | |
| { | |
| "epoch": 11.18, | |
| "learning_rate": 2.1948529411764706e-05, | |
| "loss": 0.9831, | |
| "step": 1526 | |
| }, | |
| { | |
| "epoch": 11.2, | |
| "learning_rate": 2.1911764705882354e-05, | |
| "loss": 0.9569, | |
| "step": 1528 | |
| }, | |
| { | |
| "epoch": 11.21, | |
| "learning_rate": 2.1875e-05, | |
| "loss": 0.9624, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 11.23, | |
| "learning_rate": 2.183823529411765e-05, | |
| "loss": 0.9384, | |
| "step": 1532 | |
| }, | |
| { | |
| "epoch": 11.24, | |
| "learning_rate": 2.1801470588235294e-05, | |
| "loss": 0.9646, | |
| "step": 1534 | |
| }, | |
| { | |
| "epoch": 11.26, | |
| "learning_rate": 2.1764705882352943e-05, | |
| "loss": 0.9266, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 11.27, | |
| "learning_rate": 2.172794117647059e-05, | |
| "loss": 0.9742, | |
| "step": 1538 | |
| }, | |
| { | |
| "epoch": 11.29, | |
| "learning_rate": 2.1691176470588237e-05, | |
| "loss": 0.9294, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 11.3, | |
| "learning_rate": 2.1654411764705886e-05, | |
| "loss": 0.9703, | |
| "step": 1542 | |
| }, | |
| { | |
| "epoch": 11.32, | |
| "learning_rate": 2.161764705882353e-05, | |
| "loss": 0.9507, | |
| "step": 1544 | |
| }, | |
| { | |
| "epoch": 11.33, | |
| "learning_rate": 2.158088235294118e-05, | |
| "loss": 0.9248, | |
| "step": 1546 | |
| }, | |
| { | |
| "epoch": 11.35, | |
| "learning_rate": 2.1544117647058825e-05, | |
| "loss": 0.956, | |
| "step": 1548 | |
| }, | |
| { | |
| "epoch": 11.36, | |
| "learning_rate": 2.1507352941176474e-05, | |
| "loss": 0.945, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 11.38, | |
| "learning_rate": 2.1470588235294116e-05, | |
| "loss": 0.9612, | |
| "step": 1552 | |
| }, | |
| { | |
| "epoch": 11.39, | |
| "learning_rate": 2.1433823529411765e-05, | |
| "loss": 0.9711, | |
| "step": 1554 | |
| }, | |
| { | |
| "epoch": 11.4, | |
| "learning_rate": 2.1397058823529413e-05, | |
| "loss": 0.9508, | |
| "step": 1556 | |
| }, | |
| { | |
| "epoch": 11.42, | |
| "learning_rate": 2.136029411764706e-05, | |
| "loss": 1.001, | |
| "step": 1558 | |
| }, | |
| { | |
| "epoch": 11.43, | |
| "learning_rate": 2.1323529411764707e-05, | |
| "loss": 0.9397, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 11.45, | |
| "learning_rate": 2.1286764705882353e-05, | |
| "loss": 0.9023, | |
| "step": 1562 | |
| }, | |
| { | |
| "epoch": 11.46, | |
| "learning_rate": 2.125e-05, | |
| "loss": 0.9255, | |
| "step": 1564 | |
| }, | |
| { | |
| "epoch": 11.48, | |
| "learning_rate": 2.1213235294117647e-05, | |
| "loss": 0.9306, | |
| "step": 1566 | |
| }, | |
| { | |
| "epoch": 11.49, | |
| "learning_rate": 2.1176470588235296e-05, | |
| "loss": 0.9276, | |
| "step": 1568 | |
| }, | |
| { | |
| "epoch": 11.51, | |
| "learning_rate": 2.113970588235294e-05, | |
| "loss": 0.927, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 11.52, | |
| "learning_rate": 2.110294117647059e-05, | |
| "loss": 0.9508, | |
| "step": 1572 | |
| }, | |
| { | |
| "epoch": 11.54, | |
| "learning_rate": 2.1066176470588235e-05, | |
| "loss": 0.9379, | |
| "step": 1574 | |
| }, | |
| { | |
| "epoch": 11.55, | |
| "learning_rate": 2.1029411764705884e-05, | |
| "loss": 0.9608, | |
| "step": 1576 | |
| }, | |
| { | |
| "epoch": 11.57, | |
| "learning_rate": 2.099264705882353e-05, | |
| "loss": 0.9044, | |
| "step": 1578 | |
| }, | |
| { | |
| "epoch": 11.58, | |
| "learning_rate": 2.0955882352941178e-05, | |
| "loss": 0.9432, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 11.6, | |
| "learning_rate": 2.0919117647058827e-05, | |
| "loss": 0.9489, | |
| "step": 1582 | |
| }, | |
| { | |
| "epoch": 11.61, | |
| "learning_rate": 2.0882352941176472e-05, | |
| "loss": 0.9585, | |
| "step": 1584 | |
| }, | |
| { | |
| "epoch": 11.62, | |
| "learning_rate": 2.084558823529412e-05, | |
| "loss": 0.9307, | |
| "step": 1586 | |
| }, | |
| { | |
| "epoch": 11.64, | |
| "learning_rate": 2.0808823529411766e-05, | |
| "loss": 0.9143, | |
| "step": 1588 | |
| }, | |
| { | |
| "epoch": 11.65, | |
| "learning_rate": 2.0772058823529415e-05, | |
| "loss": 0.9661, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 11.67, | |
| "learning_rate": 2.073529411764706e-05, | |
| "loss": 0.9378, | |
| "step": 1592 | |
| }, | |
| { | |
| "epoch": 11.68, | |
| "learning_rate": 2.0698529411764706e-05, | |
| "loss": 0.9438, | |
| "step": 1594 | |
| }, | |
| { | |
| "epoch": 11.7, | |
| "learning_rate": 2.066176470588235e-05, | |
| "loss": 0.9832, | |
| "step": 1596 | |
| }, | |
| { | |
| "epoch": 11.71, | |
| "learning_rate": 2.0625e-05, | |
| "loss": 0.9139, | |
| "step": 1598 | |
| }, | |
| { | |
| "epoch": 11.73, | |
| "learning_rate": 2.058823529411765e-05, | |
| "loss": 0.973, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 11.74, | |
| "learning_rate": 2.0551470588235294e-05, | |
| "loss": 0.9838, | |
| "step": 1602 | |
| }, | |
| { | |
| "epoch": 11.76, | |
| "learning_rate": 2.0514705882352943e-05, | |
| "loss": 1.0008, | |
| "step": 1604 | |
| }, | |
| { | |
| "epoch": 11.77, | |
| "learning_rate": 2.0477941176470588e-05, | |
| "loss": 0.941, | |
| "step": 1606 | |
| }, | |
| { | |
| "epoch": 11.79, | |
| "learning_rate": 2.0441176470588237e-05, | |
| "loss": 0.9632, | |
| "step": 1608 | |
| }, | |
| { | |
| "epoch": 11.8, | |
| "learning_rate": 2.0404411764705882e-05, | |
| "loss": 0.9263, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 11.81, | |
| "learning_rate": 2.036764705882353e-05, | |
| "loss": 0.9247, | |
| "step": 1612 | |
| }, | |
| { | |
| "epoch": 11.83, | |
| "learning_rate": 2.0330882352941176e-05, | |
| "loss": 0.9677, | |
| "step": 1614 | |
| }, | |
| { | |
| "epoch": 11.84, | |
| "learning_rate": 2.0294117647058825e-05, | |
| "loss": 0.9709, | |
| "step": 1616 | |
| }, | |
| { | |
| "epoch": 11.86, | |
| "learning_rate": 2.025735294117647e-05, | |
| "loss": 0.9259, | |
| "step": 1618 | |
| }, | |
| { | |
| "epoch": 11.87, | |
| "learning_rate": 2.022058823529412e-05, | |
| "loss": 0.9373, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 11.89, | |
| "learning_rate": 2.0183823529411765e-05, | |
| "loss": 0.9283, | |
| "step": 1622 | |
| }, | |
| { | |
| "epoch": 11.9, | |
| "learning_rate": 2.0147058823529413e-05, | |
| "loss": 0.9833, | |
| "step": 1624 | |
| }, | |
| { | |
| "epoch": 11.92, | |
| "learning_rate": 2.0110294117647062e-05, | |
| "loss": 0.9238, | |
| "step": 1626 | |
| }, | |
| { | |
| "epoch": 11.93, | |
| "learning_rate": 2.0073529411764708e-05, | |
| "loss": 0.9651, | |
| "step": 1628 | |
| }, | |
| { | |
| "epoch": 11.95, | |
| "learning_rate": 2.0036764705882356e-05, | |
| "loss": 0.9671, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 11.96, | |
| "learning_rate": 2e-05, | |
| "loss": 1.0051, | |
| "step": 1632 | |
| }, | |
| { | |
| "epoch": 11.98, | |
| "learning_rate": 1.996323529411765e-05, | |
| "loss": 0.9691, | |
| "step": 1634 | |
| }, | |
| { | |
| "epoch": 11.99, | |
| "learning_rate": 1.9926470588235296e-05, | |
| "loss": 0.9583, | |
| "step": 1636 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_gen_len": 128.91358024691357, | |
| "eval_loss": 1.544446587562561, | |
| "eval_rouge1": 55.6539, | |
| "eval_rouge2": 30.2395, | |
| "eval_rougeL": 38.5901, | |
| "eval_rougeLsum": 51.1518, | |
| "eval_runtime": 218.8835, | |
| "eval_samples_per_second": 4.441, | |
| "eval_steps_per_second": 0.557, | |
| "step": 1637 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 1.988970588235294e-05, | |
| "loss": 0.978, | |
| "step": 1638 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 1.9852941176470586e-05, | |
| "loss": 0.9212, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 1.9816176470588235e-05, | |
| "loss": 0.9258, | |
| "step": 1642 | |
| }, | |
| { | |
| "epoch": 12.05, | |
| "learning_rate": 1.9779411764705884e-05, | |
| "loss": 0.8803, | |
| "step": 1644 | |
| }, | |
| { | |
| "epoch": 12.06, | |
| "learning_rate": 1.974264705882353e-05, | |
| "loss": 0.9238, | |
| "step": 1646 | |
| }, | |
| { | |
| "epoch": 12.08, | |
| "learning_rate": 1.9705882352941178e-05, | |
| "loss": 0.9472, | |
| "step": 1648 | |
| }, | |
| { | |
| "epoch": 12.09, | |
| "learning_rate": 1.9669117647058824e-05, | |
| "loss": 0.9415, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 12.11, | |
| "learning_rate": 1.9632352941176472e-05, | |
| "loss": 0.9185, | |
| "step": 1652 | |
| }, | |
| { | |
| "epoch": 12.12, | |
| "learning_rate": 1.9595588235294118e-05, | |
| "loss": 0.9, | |
| "step": 1654 | |
| }, | |
| { | |
| "epoch": 12.14, | |
| "learning_rate": 1.9558823529411766e-05, | |
| "loss": 0.8914, | |
| "step": 1656 | |
| }, | |
| { | |
| "epoch": 12.15, | |
| "learning_rate": 1.9522058823529412e-05, | |
| "loss": 0.8579, | |
| "step": 1658 | |
| }, | |
| { | |
| "epoch": 12.17, | |
| "learning_rate": 1.948529411764706e-05, | |
| "loss": 0.8696, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 12.18, | |
| "learning_rate": 1.9448529411764706e-05, | |
| "loss": 0.8479, | |
| "step": 1662 | |
| }, | |
| { | |
| "epoch": 12.2, | |
| "learning_rate": 1.9411764705882355e-05, | |
| "loss": 0.9053, | |
| "step": 1664 | |
| }, | |
| { | |
| "epoch": 12.21, | |
| "learning_rate": 1.9375e-05, | |
| "loss": 0.9217, | |
| "step": 1666 | |
| }, | |
| { | |
| "epoch": 12.23, | |
| "learning_rate": 1.933823529411765e-05, | |
| "loss": 0.8974, | |
| "step": 1668 | |
| }, | |
| { | |
| "epoch": 12.24, | |
| "learning_rate": 1.9301470588235298e-05, | |
| "loss": 0.922, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 12.25, | |
| "learning_rate": 1.9264705882352943e-05, | |
| "loss": 0.9335, | |
| "step": 1672 | |
| }, | |
| { | |
| "epoch": 12.27, | |
| "learning_rate": 1.922794117647059e-05, | |
| "loss": 0.9173, | |
| "step": 1674 | |
| }, | |
| { | |
| "epoch": 12.28, | |
| "learning_rate": 1.9191176470588237e-05, | |
| "loss": 0.951, | |
| "step": 1676 | |
| }, | |
| { | |
| "epoch": 12.3, | |
| "learning_rate": 1.9154411764705886e-05, | |
| "loss": 0.9103, | |
| "step": 1678 | |
| }, | |
| { | |
| "epoch": 12.31, | |
| "learning_rate": 1.9117647058823528e-05, | |
| "loss": 0.882, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 12.33, | |
| "learning_rate": 1.9080882352941176e-05, | |
| "loss": 0.9081, | |
| "step": 1682 | |
| }, | |
| { | |
| "epoch": 12.34, | |
| "learning_rate": 1.9044117647058822e-05, | |
| "loss": 0.9444, | |
| "step": 1684 | |
| }, | |
| { | |
| "epoch": 12.36, | |
| "learning_rate": 1.900735294117647e-05, | |
| "loss": 0.8887, | |
| "step": 1686 | |
| }, | |
| { | |
| "epoch": 12.37, | |
| "learning_rate": 1.897058823529412e-05, | |
| "loss": 0.912, | |
| "step": 1688 | |
| }, | |
| { | |
| "epoch": 12.39, | |
| "learning_rate": 1.8933823529411765e-05, | |
| "loss": 0.9049, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 12.4, | |
| "learning_rate": 1.8897058823529413e-05, | |
| "loss": 0.9131, | |
| "step": 1692 | |
| }, | |
| { | |
| "epoch": 12.42, | |
| "learning_rate": 1.886029411764706e-05, | |
| "loss": 0.8734, | |
| "step": 1694 | |
| }, | |
| { | |
| "epoch": 12.43, | |
| "learning_rate": 1.8823529411764708e-05, | |
| "loss": 0.8905, | |
| "step": 1696 | |
| }, | |
| { | |
| "epoch": 12.45, | |
| "learning_rate": 1.8786764705882353e-05, | |
| "loss": 0.9443, | |
| "step": 1698 | |
| }, | |
| { | |
| "epoch": 12.46, | |
| "learning_rate": 1.8750000000000002e-05, | |
| "loss": 0.9412, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 12.47, | |
| "learning_rate": 1.8713235294117647e-05, | |
| "loss": 0.8967, | |
| "step": 1702 | |
| }, | |
| { | |
| "epoch": 12.49, | |
| "learning_rate": 1.8676470588235296e-05, | |
| "loss": 0.9488, | |
| "step": 1704 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 1.863970588235294e-05, | |
| "loss": 0.8957, | |
| "step": 1706 | |
| }, | |
| { | |
| "epoch": 12.52, | |
| "learning_rate": 1.860294117647059e-05, | |
| "loss": 0.9519, | |
| "step": 1708 | |
| }, | |
| { | |
| "epoch": 12.53, | |
| "learning_rate": 1.8566176470588235e-05, | |
| "loss": 0.8753, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 12.55, | |
| "learning_rate": 1.8529411764705884e-05, | |
| "loss": 0.9227, | |
| "step": 1712 | |
| }, | |
| { | |
| "epoch": 12.56, | |
| "learning_rate": 1.8492647058823533e-05, | |
| "loss": 0.9583, | |
| "step": 1714 | |
| }, | |
| { | |
| "epoch": 12.58, | |
| "learning_rate": 1.8455882352941178e-05, | |
| "loss": 0.9345, | |
| "step": 1716 | |
| }, | |
| { | |
| "epoch": 12.59, | |
| "learning_rate": 1.8419117647058827e-05, | |
| "loss": 0.9291, | |
| "step": 1718 | |
| }, | |
| { | |
| "epoch": 12.61, | |
| "learning_rate": 1.8382352941176472e-05, | |
| "loss": 0.9343, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 12.62, | |
| "learning_rate": 1.8345588235294118e-05, | |
| "loss": 0.9206, | |
| "step": 1722 | |
| }, | |
| { | |
| "epoch": 12.64, | |
| "learning_rate": 1.8308823529411763e-05, | |
| "loss": 0.9619, | |
| "step": 1724 | |
| }, | |
| { | |
| "epoch": 12.65, | |
| "learning_rate": 1.8272058823529412e-05, | |
| "loss": 0.9643, | |
| "step": 1726 | |
| }, | |
| { | |
| "epoch": 12.67, | |
| "learning_rate": 1.8235294117647057e-05, | |
| "loss": 0.8768, | |
| "step": 1728 | |
| }, | |
| { | |
| "epoch": 12.68, | |
| "learning_rate": 1.8198529411764706e-05, | |
| "loss": 0.8937, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 12.69, | |
| "learning_rate": 1.8161764705882355e-05, | |
| "loss": 0.9184, | |
| "step": 1732 | |
| }, | |
| { | |
| "epoch": 12.71, | |
| "learning_rate": 1.8125e-05, | |
| "loss": 0.9213, | |
| "step": 1734 | |
| }, | |
| { | |
| "epoch": 12.72, | |
| "learning_rate": 1.808823529411765e-05, | |
| "loss": 0.9434, | |
| "step": 1736 | |
| }, | |
| { | |
| "epoch": 12.74, | |
| "learning_rate": 1.8051470588235294e-05, | |
| "loss": 0.952, | |
| "step": 1738 | |
| }, | |
| { | |
| "epoch": 12.75, | |
| "learning_rate": 1.8014705882352943e-05, | |
| "loss": 0.9266, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 12.77, | |
| "learning_rate": 1.797794117647059e-05, | |
| "loss": 0.8499, | |
| "step": 1742 | |
| }, | |
| { | |
| "epoch": 12.78, | |
| "learning_rate": 1.7941176470588237e-05, | |
| "loss": 0.9336, | |
| "step": 1744 | |
| }, | |
| { | |
| "epoch": 12.8, | |
| "learning_rate": 1.7904411764705882e-05, | |
| "loss": 0.9199, | |
| "step": 1746 | |
| }, | |
| { | |
| "epoch": 12.81, | |
| "learning_rate": 1.786764705882353e-05, | |
| "loss": 0.9746, | |
| "step": 1748 | |
| }, | |
| { | |
| "epoch": 12.83, | |
| "learning_rate": 1.7830882352941177e-05, | |
| "loss": 0.9535, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 12.84, | |
| "learning_rate": 1.7794117647058825e-05, | |
| "loss": 0.9126, | |
| "step": 1752 | |
| }, | |
| { | |
| "epoch": 12.86, | |
| "learning_rate": 1.775735294117647e-05, | |
| "loss": 0.9181, | |
| "step": 1754 | |
| }, | |
| { | |
| "epoch": 12.87, | |
| "learning_rate": 1.772058823529412e-05, | |
| "loss": 0.8901, | |
| "step": 1756 | |
| }, | |
| { | |
| "epoch": 12.89, | |
| "learning_rate": 1.7683823529411768e-05, | |
| "loss": 0.887, | |
| "step": 1758 | |
| }, | |
| { | |
| "epoch": 12.9, | |
| "learning_rate": 1.7647058823529414e-05, | |
| "loss": 0.9095, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 12.91, | |
| "learning_rate": 1.7610294117647062e-05, | |
| "loss": 0.8842, | |
| "step": 1762 | |
| }, | |
| { | |
| "epoch": 12.93, | |
| "learning_rate": 1.7573529411764704e-05, | |
| "loss": 0.9332, | |
| "step": 1764 | |
| }, | |
| { | |
| "epoch": 12.94, | |
| "learning_rate": 1.7536764705882353e-05, | |
| "loss": 0.93, | |
| "step": 1766 | |
| }, | |
| { | |
| "epoch": 12.96, | |
| "learning_rate": 1.75e-05, | |
| "loss": 0.9431, | |
| "step": 1768 | |
| }, | |
| { | |
| "epoch": 12.97, | |
| "learning_rate": 1.7463235294117647e-05, | |
| "loss": 0.8755, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 12.99, | |
| "learning_rate": 1.7426470588235293e-05, | |
| "loss": 0.9601, | |
| "step": 1772 | |
| }, | |
| { | |
| "epoch": 12.99, | |
| "eval_gen_len": 128.9783950617284, | |
| "eval_loss": 1.5515539646148682, | |
| "eval_rouge1": 55.9154, | |
| "eval_rouge2": 30.5471, | |
| "eval_rougeL": 38.8607, | |
| "eval_rougeLsum": 51.2856, | |
| "eval_runtime": 219.5635, | |
| "eval_samples_per_second": 4.427, | |
| "eval_steps_per_second": 0.556, | |
| "step": 1773 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 1.738970588235294e-05, | |
| "loss": 0.9193, | |
| "step": 1774 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 1.735294117647059e-05, | |
| "loss": 0.8611, | |
| "step": 1776 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 1.7316176470588235e-05, | |
| "loss": 0.9041, | |
| "step": 1778 | |
| }, | |
| { | |
| "epoch": 13.05, | |
| "learning_rate": 1.7279411764705884e-05, | |
| "loss": 0.8886, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 13.06, | |
| "learning_rate": 1.724264705882353e-05, | |
| "loss": 0.8797, | |
| "step": 1782 | |
| }, | |
| { | |
| "epoch": 13.08, | |
| "learning_rate": 1.720588235294118e-05, | |
| "loss": 0.9299, | |
| "step": 1784 | |
| }, | |
| { | |
| "epoch": 13.09, | |
| "learning_rate": 1.7169117647058824e-05, | |
| "loss": 0.8481, | |
| "step": 1786 | |
| }, | |
| { | |
| "epoch": 13.1, | |
| "learning_rate": 1.7132352941176472e-05, | |
| "loss": 0.9105, | |
| "step": 1788 | |
| }, | |
| { | |
| "epoch": 13.12, | |
| "learning_rate": 1.7095588235294118e-05, | |
| "loss": 0.8449, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 13.13, | |
| "learning_rate": 1.7058823529411767e-05, | |
| "loss": 0.9001, | |
| "step": 1792 | |
| }, | |
| { | |
| "epoch": 13.15, | |
| "learning_rate": 1.7022058823529412e-05, | |
| "loss": 0.8804, | |
| "step": 1794 | |
| }, | |
| { | |
| "epoch": 13.16, | |
| "learning_rate": 1.698529411764706e-05, | |
| "loss": 0.8577, | |
| "step": 1796 | |
| }, | |
| { | |
| "epoch": 13.18, | |
| "learning_rate": 1.6948529411764706e-05, | |
| "loss": 0.8812, | |
| "step": 1798 | |
| }, | |
| { | |
| "epoch": 13.19, | |
| "learning_rate": 1.6911764705882355e-05, | |
| "loss": 0.9037, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 13.21, | |
| "learning_rate": 1.6875000000000004e-05, | |
| "loss": 0.8882, | |
| "step": 1802 | |
| }, | |
| { | |
| "epoch": 13.22, | |
| "learning_rate": 1.683823529411765e-05, | |
| "loss": 0.883, | |
| "step": 1804 | |
| }, | |
| { | |
| "epoch": 13.24, | |
| "learning_rate": 1.6801470588235298e-05, | |
| "loss": 0.8434, | |
| "step": 1806 | |
| }, | |
| { | |
| "epoch": 13.25, | |
| "learning_rate": 1.676470588235294e-05, | |
| "loss": 0.9052, | |
| "step": 1808 | |
| }, | |
| { | |
| "epoch": 13.27, | |
| "learning_rate": 1.672794117647059e-05, | |
| "loss": 0.875, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 13.28, | |
| "learning_rate": 1.6691176470588234e-05, | |
| "loss": 0.8739, | |
| "step": 1812 | |
| }, | |
| { | |
| "epoch": 13.3, | |
| "learning_rate": 1.6654411764705883e-05, | |
| "loss": 0.8779, | |
| "step": 1814 | |
| }, | |
| { | |
| "epoch": 13.31, | |
| "learning_rate": 1.6617647058823528e-05, | |
| "loss": 0.8743, | |
| "step": 1816 | |
| }, | |
| { | |
| "epoch": 13.32, | |
| "learning_rate": 1.6580882352941177e-05, | |
| "loss": 0.8741, | |
| "step": 1818 | |
| }, | |
| { | |
| "epoch": 13.34, | |
| "learning_rate": 1.6544117647058825e-05, | |
| "loss": 0.8905, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 13.35, | |
| "learning_rate": 1.650735294117647e-05, | |
| "loss": 0.8617, | |
| "step": 1822 | |
| }, | |
| { | |
| "epoch": 13.37, | |
| "learning_rate": 1.647058823529412e-05, | |
| "loss": 0.8905, | |
| "step": 1824 | |
| }, | |
| { | |
| "epoch": 13.38, | |
| "learning_rate": 1.6433823529411765e-05, | |
| "loss": 0.8572, | |
| "step": 1826 | |
| }, | |
| { | |
| "epoch": 13.4, | |
| "learning_rate": 1.6397058823529414e-05, | |
| "loss": 0.8647, | |
| "step": 1828 | |
| }, | |
| { | |
| "epoch": 13.41, | |
| "learning_rate": 1.636029411764706e-05, | |
| "loss": 0.9023, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 13.43, | |
| "learning_rate": 1.6323529411764708e-05, | |
| "loss": 0.905, | |
| "step": 1832 | |
| }, | |
| { | |
| "epoch": 13.44, | |
| "learning_rate": 1.6286764705882353e-05, | |
| "loss": 0.9219, | |
| "step": 1834 | |
| }, | |
| { | |
| "epoch": 13.46, | |
| "learning_rate": 1.6250000000000002e-05, | |
| "loss": 0.9263, | |
| "step": 1836 | |
| }, | |
| { | |
| "epoch": 13.47, | |
| "learning_rate": 1.6213235294117647e-05, | |
| "loss": 0.8915, | |
| "step": 1838 | |
| }, | |
| { | |
| "epoch": 13.49, | |
| "learning_rate": 1.6176470588235296e-05, | |
| "loss": 0.9464, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 13.5, | |
| "learning_rate": 1.613970588235294e-05, | |
| "loss": 0.8873, | |
| "step": 1842 | |
| }, | |
| { | |
| "epoch": 13.52, | |
| "learning_rate": 1.610294117647059e-05, | |
| "loss": 0.8944, | |
| "step": 1844 | |
| }, | |
| { | |
| "epoch": 13.53, | |
| "learning_rate": 1.6066176470588236e-05, | |
| "loss": 0.8724, | |
| "step": 1846 | |
| }, | |
| { | |
| "epoch": 13.54, | |
| "learning_rate": 1.6029411764705884e-05, | |
| "loss": 0.9189, | |
| "step": 1848 | |
| }, | |
| { | |
| "epoch": 13.56, | |
| "learning_rate": 1.599264705882353e-05, | |
| "loss": 0.8386, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 13.57, | |
| "learning_rate": 1.5955882352941175e-05, | |
| "loss": 0.868, | |
| "step": 1852 | |
| }, | |
| { | |
| "epoch": 13.59, | |
| "learning_rate": 1.5919117647058824e-05, | |
| "loss": 0.8934, | |
| "step": 1854 | |
| }, | |
| { | |
| "epoch": 13.6, | |
| "learning_rate": 1.588235294117647e-05, | |
| "loss": 0.8914, | |
| "step": 1856 | |
| }, | |
| { | |
| "epoch": 13.62, | |
| "learning_rate": 1.5845588235294118e-05, | |
| "loss": 0.948, | |
| "step": 1858 | |
| }, | |
| { | |
| "epoch": 13.63, | |
| "learning_rate": 1.5808823529411763e-05, | |
| "loss": 0.9076, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 13.65, | |
| "learning_rate": 1.5772058823529412e-05, | |
| "loss": 0.8992, | |
| "step": 1862 | |
| }, | |
| { | |
| "epoch": 13.66, | |
| "learning_rate": 1.573529411764706e-05, | |
| "loss": 0.8784, | |
| "step": 1864 | |
| }, | |
| { | |
| "epoch": 13.68, | |
| "learning_rate": 1.5698529411764706e-05, | |
| "loss": 0.9055, | |
| "step": 1866 | |
| }, | |
| { | |
| "epoch": 13.69, | |
| "learning_rate": 1.5661764705882355e-05, | |
| "loss": 0.927, | |
| "step": 1868 | |
| }, | |
| { | |
| "epoch": 13.71, | |
| "learning_rate": 1.5625e-05, | |
| "loss": 0.8678, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 13.72, | |
| "learning_rate": 1.558823529411765e-05, | |
| "loss": 0.8993, | |
| "step": 1872 | |
| }, | |
| { | |
| "epoch": 13.74, | |
| "learning_rate": 1.5551470588235294e-05, | |
| "loss": 0.8649, | |
| "step": 1874 | |
| }, | |
| { | |
| "epoch": 13.75, | |
| "learning_rate": 1.5514705882352943e-05, | |
| "loss": 0.9001, | |
| "step": 1876 | |
| }, | |
| { | |
| "epoch": 13.76, | |
| "learning_rate": 1.547794117647059e-05, | |
| "loss": 0.8923, | |
| "step": 1878 | |
| }, | |
| { | |
| "epoch": 13.78, | |
| "learning_rate": 1.5441176470588237e-05, | |
| "loss": 0.9079, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 13.79, | |
| "learning_rate": 1.5404411764705883e-05, | |
| "loss": 0.8799, | |
| "step": 1882 | |
| }, | |
| { | |
| "epoch": 13.81, | |
| "learning_rate": 1.536764705882353e-05, | |
| "loss": 0.8619, | |
| "step": 1884 | |
| }, | |
| { | |
| "epoch": 13.82, | |
| "learning_rate": 1.5330882352941177e-05, | |
| "loss": 0.9564, | |
| "step": 1886 | |
| }, | |
| { | |
| "epoch": 13.84, | |
| "learning_rate": 1.5294117647058826e-05, | |
| "loss": 0.8944, | |
| "step": 1888 | |
| }, | |
| { | |
| "epoch": 13.85, | |
| "learning_rate": 1.5257352941176473e-05, | |
| "loss": 0.8868, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 13.87, | |
| "learning_rate": 1.5220588235294116e-05, | |
| "loss": 0.8411, | |
| "step": 1892 | |
| }, | |
| { | |
| "epoch": 13.88, | |
| "learning_rate": 1.5183823529411765e-05, | |
| "loss": 0.9226, | |
| "step": 1894 | |
| }, | |
| { | |
| "epoch": 13.9, | |
| "learning_rate": 1.5147058823529412e-05, | |
| "loss": 0.9408, | |
| "step": 1896 | |
| }, | |
| { | |
| "epoch": 13.91, | |
| "learning_rate": 1.511029411764706e-05, | |
| "loss": 0.8672, | |
| "step": 1898 | |
| }, | |
| { | |
| "epoch": 13.93, | |
| "learning_rate": 1.5073529411764706e-05, | |
| "loss": 0.9427, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 13.94, | |
| "learning_rate": 1.5036764705882353e-05, | |
| "loss": 0.9099, | |
| "step": 1902 | |
| }, | |
| { | |
| "epoch": 13.96, | |
| "learning_rate": 1.5e-05, | |
| "loss": 0.9027, | |
| "step": 1904 | |
| }, | |
| { | |
| "epoch": 13.97, | |
| "learning_rate": 1.4963235294117647e-05, | |
| "loss": 0.8715, | |
| "step": 1906 | |
| }, | |
| { | |
| "epoch": 13.98, | |
| "learning_rate": 1.4926470588235295e-05, | |
| "loss": 0.9116, | |
| "step": 1908 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 1.4889705882352942e-05, | |
| "loss": 0.8882, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_gen_len": 128.52057613168725, | |
| "eval_loss": 1.5736336708068848, | |
| "eval_rouge1": 56.3282, | |
| "eval_rouge2": 30.9807, | |
| "eval_rougeL": 39.2351, | |
| "eval_rougeLsum": 51.8022, | |
| "eval_runtime": 218.5676, | |
| "eval_samples_per_second": 4.447, | |
| "eval_steps_per_second": 0.558, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 1.4852941176470589e-05, | |
| "loss": 0.8246, | |
| "step": 1912 | |
| }, | |
| { | |
| "epoch": 14.03, | |
| "learning_rate": 1.4816176470588236e-05, | |
| "loss": 0.9006, | |
| "step": 1914 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "learning_rate": 1.4779411764705883e-05, | |
| "loss": 0.8617, | |
| "step": 1916 | |
| }, | |
| { | |
| "epoch": 14.06, | |
| "learning_rate": 1.474264705882353e-05, | |
| "loss": 0.9181, | |
| "step": 1918 | |
| }, | |
| { | |
| "epoch": 14.07, | |
| "learning_rate": 1.4705882352941177e-05, | |
| "loss": 0.848, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 14.09, | |
| "learning_rate": 1.4669117647058826e-05, | |
| "loss": 0.8767, | |
| "step": 1922 | |
| }, | |
| { | |
| "epoch": 14.1, | |
| "learning_rate": 1.4632352941176473e-05, | |
| "loss": 0.8644, | |
| "step": 1924 | |
| }, | |
| { | |
| "epoch": 14.12, | |
| "learning_rate": 1.459558823529412e-05, | |
| "loss": 0.8462, | |
| "step": 1926 | |
| }, | |
| { | |
| "epoch": 14.13, | |
| "learning_rate": 1.4558823529411767e-05, | |
| "loss": 0.8573, | |
| "step": 1928 | |
| }, | |
| { | |
| "epoch": 14.15, | |
| "learning_rate": 1.4522058823529414e-05, | |
| "loss": 0.8592, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 14.16, | |
| "learning_rate": 1.4485294117647061e-05, | |
| "loss": 0.8611, | |
| "step": 1932 | |
| }, | |
| { | |
| "epoch": 14.17, | |
| "learning_rate": 1.4448529411764705e-05, | |
| "loss": 0.8848, | |
| "step": 1934 | |
| }, | |
| { | |
| "epoch": 14.19, | |
| "learning_rate": 1.4411764705882352e-05, | |
| "loss": 0.8964, | |
| "step": 1936 | |
| }, | |
| { | |
| "epoch": 14.2, | |
| "learning_rate": 1.4374999999999999e-05, | |
| "loss": 0.8316, | |
| "step": 1938 | |
| }, | |
| { | |
| "epoch": 14.22, | |
| "learning_rate": 1.4338235294117647e-05, | |
| "loss": 0.8804, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 14.23, | |
| "learning_rate": 1.4301470588235295e-05, | |
| "loss": 0.8706, | |
| "step": 1942 | |
| }, | |
| { | |
| "epoch": 14.25, | |
| "learning_rate": 1.4264705882352942e-05, | |
| "loss": 0.8235, | |
| "step": 1944 | |
| }, | |
| { | |
| "epoch": 14.26, | |
| "learning_rate": 1.4227941176470589e-05, | |
| "loss": 0.8962, | |
| "step": 1946 | |
| }, | |
| { | |
| "epoch": 14.28, | |
| "learning_rate": 1.4191176470588236e-05, | |
| "loss": 0.8734, | |
| "step": 1948 | |
| }, | |
| { | |
| "epoch": 14.29, | |
| "learning_rate": 1.4154411764705883e-05, | |
| "loss": 0.8944, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 14.31, | |
| "learning_rate": 1.411764705882353e-05, | |
| "loss": 0.8675, | |
| "step": 1952 | |
| }, | |
| { | |
| "epoch": 14.32, | |
| "learning_rate": 1.4080882352941177e-05, | |
| "loss": 0.8344, | |
| "step": 1954 | |
| }, | |
| { | |
| "epoch": 14.34, | |
| "learning_rate": 1.4044117647058824e-05, | |
| "loss": 0.876, | |
| "step": 1956 | |
| }, | |
| { | |
| "epoch": 14.35, | |
| "learning_rate": 1.4007352941176471e-05, | |
| "loss": 0.8512, | |
| "step": 1958 | |
| }, | |
| { | |
| "epoch": 14.37, | |
| "learning_rate": 1.3970588235294118e-05, | |
| "loss": 0.8957, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 14.38, | |
| "learning_rate": 1.3933823529411765e-05, | |
| "loss": 0.8604, | |
| "step": 1962 | |
| }, | |
| { | |
| "epoch": 14.39, | |
| "learning_rate": 1.3897058823529412e-05, | |
| "loss": 0.8587, | |
| "step": 1964 | |
| }, | |
| { | |
| "epoch": 14.41, | |
| "learning_rate": 1.3860294117647061e-05, | |
| "loss": 0.8611, | |
| "step": 1966 | |
| }, | |
| { | |
| "epoch": 14.42, | |
| "learning_rate": 1.3823529411764708e-05, | |
| "loss": 0.8726, | |
| "step": 1968 | |
| }, | |
| { | |
| "epoch": 14.44, | |
| "learning_rate": 1.3786764705882355e-05, | |
| "loss": 0.872, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 14.45, | |
| "learning_rate": 1.3750000000000002e-05, | |
| "loss": 0.8704, | |
| "step": 1972 | |
| }, | |
| { | |
| "epoch": 14.47, | |
| "learning_rate": 1.371323529411765e-05, | |
| "loss": 0.8747, | |
| "step": 1974 | |
| }, | |
| { | |
| "epoch": 14.48, | |
| "learning_rate": 1.3676470588235296e-05, | |
| "loss": 0.8648, | |
| "step": 1976 | |
| }, | |
| { | |
| "epoch": 14.5, | |
| "learning_rate": 1.363970588235294e-05, | |
| "loss": 0.8985, | |
| "step": 1978 | |
| }, | |
| { | |
| "epoch": 14.51, | |
| "learning_rate": 1.3602941176470587e-05, | |
| "loss": 0.864, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 14.53, | |
| "learning_rate": 1.3566176470588234e-05, | |
| "loss": 0.8825, | |
| "step": 1982 | |
| }, | |
| { | |
| "epoch": 14.54, | |
| "learning_rate": 1.3529411764705883e-05, | |
| "loss": 0.8681, | |
| "step": 1984 | |
| }, | |
| { | |
| "epoch": 14.56, | |
| "learning_rate": 1.349264705882353e-05, | |
| "loss": 0.8663, | |
| "step": 1986 | |
| }, | |
| { | |
| "epoch": 14.57, | |
| "learning_rate": 1.3455882352941177e-05, | |
| "loss": 0.8551, | |
| "step": 1988 | |
| }, | |
| { | |
| "epoch": 14.59, | |
| "learning_rate": 1.3419117647058824e-05, | |
| "loss": 0.8468, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 14.6, | |
| "learning_rate": 1.3382352941176471e-05, | |
| "loss": 0.8105, | |
| "step": 1992 | |
| }, | |
| { | |
| "epoch": 14.61, | |
| "learning_rate": 1.3345588235294118e-05, | |
| "loss": 0.8792, | |
| "step": 1994 | |
| }, | |
| { | |
| "epoch": 14.63, | |
| "learning_rate": 1.3308823529411765e-05, | |
| "loss": 0.859, | |
| "step": 1996 | |
| }, | |
| { | |
| "epoch": 14.64, | |
| "learning_rate": 1.3272058823529412e-05, | |
| "loss": 0.8592, | |
| "step": 1998 | |
| }, | |
| { | |
| "epoch": 14.66, | |
| "learning_rate": 1.323529411764706e-05, | |
| "loss": 0.8974, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 14.67, | |
| "learning_rate": 1.3198529411764706e-05, | |
| "loss": 0.8388, | |
| "step": 2002 | |
| }, | |
| { | |
| "epoch": 14.69, | |
| "learning_rate": 1.3161764705882353e-05, | |
| "loss": 0.8662, | |
| "step": 2004 | |
| }, | |
| { | |
| "epoch": 14.7, | |
| "learning_rate": 1.3125e-05, | |
| "loss": 0.8543, | |
| "step": 2006 | |
| }, | |
| { | |
| "epoch": 14.72, | |
| "learning_rate": 1.3088235294117648e-05, | |
| "loss": 0.8434, | |
| "step": 2008 | |
| }, | |
| { | |
| "epoch": 14.73, | |
| "learning_rate": 1.3051470588235296e-05, | |
| "loss": 0.8778, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 14.75, | |
| "learning_rate": 1.3014705882352943e-05, | |
| "loss": 0.865, | |
| "step": 2012 | |
| }, | |
| { | |
| "epoch": 14.76, | |
| "learning_rate": 1.297794117647059e-05, | |
| "loss": 0.8724, | |
| "step": 2014 | |
| }, | |
| { | |
| "epoch": 14.78, | |
| "learning_rate": 1.2941176470588238e-05, | |
| "loss": 0.8574, | |
| "step": 2016 | |
| }, | |
| { | |
| "epoch": 14.79, | |
| "learning_rate": 1.2904411764705885e-05, | |
| "loss": 0.8757, | |
| "step": 2018 | |
| }, | |
| { | |
| "epoch": 14.81, | |
| "learning_rate": 1.2867647058823528e-05, | |
| "loss": 0.9153, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 14.82, | |
| "learning_rate": 1.2830882352941175e-05, | |
| "loss": 0.8359, | |
| "step": 2022 | |
| }, | |
| { | |
| "epoch": 14.83, | |
| "learning_rate": 1.2794117647058822e-05, | |
| "loss": 0.8779, | |
| "step": 2024 | |
| }, | |
| { | |
| "epoch": 14.85, | |
| "learning_rate": 1.275735294117647e-05, | |
| "loss": 0.8764, | |
| "step": 2026 | |
| }, | |
| { | |
| "epoch": 14.86, | |
| "learning_rate": 1.2720588235294118e-05, | |
| "loss": 0.8796, | |
| "step": 2028 | |
| }, | |
| { | |
| "epoch": 14.88, | |
| "learning_rate": 1.2683823529411765e-05, | |
| "loss": 0.8569, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 14.89, | |
| "learning_rate": 1.2647058823529412e-05, | |
| "loss": 0.8489, | |
| "step": 2032 | |
| }, | |
| { | |
| "epoch": 14.91, | |
| "learning_rate": 1.261029411764706e-05, | |
| "loss": 0.9111, | |
| "step": 2034 | |
| }, | |
| { | |
| "epoch": 14.92, | |
| "learning_rate": 1.2573529411764706e-05, | |
| "loss": 0.83, | |
| "step": 2036 | |
| }, | |
| { | |
| "epoch": 14.94, | |
| "learning_rate": 1.2536764705882354e-05, | |
| "loss": 0.9074, | |
| "step": 2038 | |
| }, | |
| { | |
| "epoch": 14.95, | |
| "learning_rate": 1.25e-05, | |
| "loss": 0.8712, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 14.97, | |
| "learning_rate": 1.2463235294117648e-05, | |
| "loss": 0.8541, | |
| "step": 2042 | |
| }, | |
| { | |
| "epoch": 14.98, | |
| "learning_rate": 1.2426470588235295e-05, | |
| "loss": 0.8579, | |
| "step": 2044 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 1.2389705882352942e-05, | |
| "loss": 0.851, | |
| "step": 2046 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_gen_len": 128.76234567901236, | |
| "eval_loss": 1.5890727043151855, | |
| "eval_rouge1": 56.0531, | |
| "eval_rouge2": 30.6748, | |
| "eval_rougeL": 38.8847, | |
| "eval_rougeLsum": 51.5739, | |
| "eval_runtime": 218.1756, | |
| "eval_samples_per_second": 4.455, | |
| "eval_steps_per_second": 0.559, | |
| "step": 2046 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 1.2352941176470589e-05, | |
| "loss": 0.8108, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 1.2316176470588236e-05, | |
| "loss": 0.9012, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 1.2279411764705883e-05, | |
| "loss": 0.796, | |
| "step": 2052 | |
| }, | |
| { | |
| "epoch": 15.05, | |
| "learning_rate": 1.224264705882353e-05, | |
| "loss": 0.8676, | |
| "step": 2054 | |
| }, | |
| { | |
| "epoch": 15.07, | |
| "learning_rate": 1.2205882352941177e-05, | |
| "loss": 0.8537, | |
| "step": 2056 | |
| }, | |
| { | |
| "epoch": 15.08, | |
| "learning_rate": 1.2169117647058824e-05, | |
| "loss": 0.8722, | |
| "step": 2058 | |
| }, | |
| { | |
| "epoch": 15.1, | |
| "learning_rate": 1.2132352941176471e-05, | |
| "loss": 0.872, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 15.11, | |
| "learning_rate": 1.2095588235294118e-05, | |
| "loss": 0.83, | |
| "step": 2062 | |
| }, | |
| { | |
| "epoch": 15.13, | |
| "learning_rate": 1.2058823529411765e-05, | |
| "loss": 0.8522, | |
| "step": 2064 | |
| }, | |
| { | |
| "epoch": 15.14, | |
| "learning_rate": 1.2022058823529412e-05, | |
| "loss": 0.8791, | |
| "step": 2066 | |
| }, | |
| { | |
| "epoch": 15.16, | |
| "learning_rate": 1.198529411764706e-05, | |
| "loss": 0.8428, | |
| "step": 2068 | |
| }, | |
| { | |
| "epoch": 15.17, | |
| "learning_rate": 1.1948529411764707e-05, | |
| "loss": 0.8357, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 15.19, | |
| "learning_rate": 1.1911764705882354e-05, | |
| "loss": 0.832, | |
| "step": 2072 | |
| }, | |
| { | |
| "epoch": 15.2, | |
| "learning_rate": 1.1875e-05, | |
| "loss": 0.8356, | |
| "step": 2074 | |
| }, | |
| { | |
| "epoch": 15.22, | |
| "learning_rate": 1.1838235294117648e-05, | |
| "loss": 0.7895, | |
| "step": 2076 | |
| }, | |
| { | |
| "epoch": 15.23, | |
| "learning_rate": 1.1801470588235295e-05, | |
| "loss": 0.8756, | |
| "step": 2078 | |
| }, | |
| { | |
| "epoch": 15.25, | |
| "learning_rate": 1.1764705882352942e-05, | |
| "loss": 0.8746, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 15.26, | |
| "learning_rate": 1.1727941176470589e-05, | |
| "loss": 0.8305, | |
| "step": 2082 | |
| }, | |
| { | |
| "epoch": 15.27, | |
| "learning_rate": 1.1691176470588236e-05, | |
| "loss": 0.8031, | |
| "step": 2084 | |
| }, | |
| { | |
| "epoch": 15.29, | |
| "learning_rate": 1.1654411764705883e-05, | |
| "loss": 0.8222, | |
| "step": 2086 | |
| }, | |
| { | |
| "epoch": 15.3, | |
| "learning_rate": 1.161764705882353e-05, | |
| "loss": 0.8424, | |
| "step": 2088 | |
| }, | |
| { | |
| "epoch": 15.32, | |
| "learning_rate": 1.1580882352941177e-05, | |
| "loss": 0.8262, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 15.33, | |
| "learning_rate": 1.1544117647058824e-05, | |
| "loss": 0.8239, | |
| "step": 2092 | |
| }, | |
| { | |
| "epoch": 15.35, | |
| "learning_rate": 1.1507352941176471e-05, | |
| "loss": 0.8205, | |
| "step": 2094 | |
| }, | |
| { | |
| "epoch": 15.36, | |
| "learning_rate": 1.1470588235294118e-05, | |
| "loss": 0.875, | |
| "step": 2096 | |
| }, | |
| { | |
| "epoch": 15.38, | |
| "learning_rate": 1.1433823529411765e-05, | |
| "loss": 0.9005, | |
| "step": 2098 | |
| }, | |
| { | |
| "epoch": 15.39, | |
| "learning_rate": 1.1397058823529412e-05, | |
| "loss": 0.8394, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 15.41, | |
| "learning_rate": 1.136029411764706e-05, | |
| "loss": 0.8192, | |
| "step": 2102 | |
| }, | |
| { | |
| "epoch": 15.42, | |
| "learning_rate": 1.1323529411764707e-05, | |
| "loss": 0.7979, | |
| "step": 2104 | |
| }, | |
| { | |
| "epoch": 15.44, | |
| "learning_rate": 1.1286764705882354e-05, | |
| "loss": 0.859, | |
| "step": 2106 | |
| }, | |
| { | |
| "epoch": 15.45, | |
| "learning_rate": 1.125e-05, | |
| "loss": 0.8507, | |
| "step": 2108 | |
| }, | |
| { | |
| "epoch": 15.46, | |
| "learning_rate": 1.1213235294117648e-05, | |
| "loss": 0.8197, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 15.48, | |
| "learning_rate": 1.1176470588235295e-05, | |
| "loss": 0.8413, | |
| "step": 2112 | |
| }, | |
| { | |
| "epoch": 15.49, | |
| "learning_rate": 1.1139705882352942e-05, | |
| "loss": 0.882, | |
| "step": 2114 | |
| }, | |
| { | |
| "epoch": 15.51, | |
| "learning_rate": 1.1102941176470589e-05, | |
| "loss": 0.8315, | |
| "step": 2116 | |
| }, | |
| { | |
| "epoch": 15.52, | |
| "learning_rate": 1.1066176470588236e-05, | |
| "loss": 0.8411, | |
| "step": 2118 | |
| }, | |
| { | |
| "epoch": 15.54, | |
| "learning_rate": 1.1029411764705883e-05, | |
| "loss": 0.8753, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 15.55, | |
| "learning_rate": 1.099264705882353e-05, | |
| "loss": 0.8606, | |
| "step": 2122 | |
| }, | |
| { | |
| "epoch": 15.57, | |
| "learning_rate": 1.0955882352941177e-05, | |
| "loss": 0.85, | |
| "step": 2124 | |
| }, | |
| { | |
| "epoch": 15.58, | |
| "learning_rate": 1.0919117647058824e-05, | |
| "loss": 0.8528, | |
| "step": 2126 | |
| }, | |
| { | |
| "epoch": 15.6, | |
| "learning_rate": 1.0882352941176471e-05, | |
| "loss": 0.867, | |
| "step": 2128 | |
| }, | |
| { | |
| "epoch": 15.61, | |
| "learning_rate": 1.0845588235294118e-05, | |
| "loss": 0.8622, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 15.63, | |
| "learning_rate": 1.0808823529411765e-05, | |
| "loss": 0.8359, | |
| "step": 2132 | |
| }, | |
| { | |
| "epoch": 15.64, | |
| "learning_rate": 1.0772058823529413e-05, | |
| "loss": 0.8579, | |
| "step": 2134 | |
| }, | |
| { | |
| "epoch": 15.66, | |
| "learning_rate": 1.0735294117647058e-05, | |
| "loss": 0.8362, | |
| "step": 2136 | |
| }, | |
| { | |
| "epoch": 15.67, | |
| "learning_rate": 1.0698529411764707e-05, | |
| "loss": 0.832, | |
| "step": 2138 | |
| }, | |
| { | |
| "epoch": 15.68, | |
| "learning_rate": 1.0661764705882354e-05, | |
| "loss": 0.8286, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 15.7, | |
| "learning_rate": 1.0625e-05, | |
| "loss": 0.8245, | |
| "step": 2142 | |
| }, | |
| { | |
| "epoch": 15.71, | |
| "learning_rate": 1.0588235294117648e-05, | |
| "loss": 0.894, | |
| "step": 2144 | |
| }, | |
| { | |
| "epoch": 15.73, | |
| "learning_rate": 1.0551470588235295e-05, | |
| "loss": 0.8512, | |
| "step": 2146 | |
| }, | |
| { | |
| "epoch": 15.74, | |
| "learning_rate": 1.0514705882352942e-05, | |
| "loss": 0.9047, | |
| "step": 2148 | |
| }, | |
| { | |
| "epoch": 15.76, | |
| "learning_rate": 1.0477941176470589e-05, | |
| "loss": 0.8459, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 15.77, | |
| "learning_rate": 1.0441176470588236e-05, | |
| "loss": 0.8563, | |
| "step": 2152 | |
| }, | |
| { | |
| "epoch": 15.79, | |
| "learning_rate": 1.0404411764705883e-05, | |
| "loss": 0.8453, | |
| "step": 2154 | |
| }, | |
| { | |
| "epoch": 15.8, | |
| "learning_rate": 1.036764705882353e-05, | |
| "loss": 0.8583, | |
| "step": 2156 | |
| }, | |
| { | |
| "epoch": 15.82, | |
| "learning_rate": 1.0330882352941176e-05, | |
| "loss": 0.8689, | |
| "step": 2158 | |
| }, | |
| { | |
| "epoch": 15.83, | |
| "learning_rate": 1.0294117647058824e-05, | |
| "loss": 0.8375, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 15.85, | |
| "learning_rate": 1.0257352941176471e-05, | |
| "loss": 0.8454, | |
| "step": 2162 | |
| }, | |
| { | |
| "epoch": 15.86, | |
| "learning_rate": 1.0220588235294118e-05, | |
| "loss": 0.8112, | |
| "step": 2164 | |
| }, | |
| { | |
| "epoch": 15.88, | |
| "learning_rate": 1.0183823529411766e-05, | |
| "loss": 0.8637, | |
| "step": 2166 | |
| }, | |
| { | |
| "epoch": 15.89, | |
| "learning_rate": 1.0147058823529413e-05, | |
| "loss": 0.8704, | |
| "step": 2168 | |
| }, | |
| { | |
| "epoch": 15.9, | |
| "learning_rate": 1.011029411764706e-05, | |
| "loss": 0.8203, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 15.92, | |
| "learning_rate": 1.0073529411764707e-05, | |
| "loss": 0.8463, | |
| "step": 2172 | |
| }, | |
| { | |
| "epoch": 15.93, | |
| "learning_rate": 1.0036764705882354e-05, | |
| "loss": 0.8285, | |
| "step": 2174 | |
| }, | |
| { | |
| "epoch": 15.95, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8072, | |
| "step": 2176 | |
| }, | |
| { | |
| "epoch": 15.96, | |
| "learning_rate": 9.963235294117648e-06, | |
| "loss": 0.8673, | |
| "step": 2178 | |
| }, | |
| { | |
| "epoch": 15.98, | |
| "learning_rate": 9.926470588235293e-06, | |
| "loss": 0.8373, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 15.99, | |
| "learning_rate": 9.889705882352942e-06, | |
| "loss": 0.8825, | |
| "step": 2182 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_gen_len": 128.7798353909465, | |
| "eval_loss": 1.5978367328643799, | |
| "eval_rouge1": 56.0084, | |
| "eval_rouge2": 30.7943, | |
| "eval_rougeL": 38.9692, | |
| "eval_rougeLsum": 51.5587, | |
| "eval_runtime": 220.9798, | |
| "eval_samples_per_second": 4.399, | |
| "eval_steps_per_second": 0.552, | |
| "step": 2183 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 9.852941176470589e-06, | |
| "loss": 0.8308, | |
| "step": 2184 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 9.816176470588236e-06, | |
| "loss": 0.8076, | |
| "step": 2186 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "learning_rate": 9.779411764705883e-06, | |
| "loss": 0.827, | |
| "step": 2188 | |
| }, | |
| { | |
| "epoch": 16.05, | |
| "learning_rate": 9.74264705882353e-06, | |
| "loss": 0.8192, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 16.07, | |
| "learning_rate": 9.705882352941177e-06, | |
| "loss": 0.8323, | |
| "step": 2192 | |
| }, | |
| { | |
| "epoch": 16.08, | |
| "learning_rate": 9.669117647058824e-06, | |
| "loss": 0.8062, | |
| "step": 2194 | |
| }, | |
| { | |
| "epoch": 16.1, | |
| "learning_rate": 9.632352941176471e-06, | |
| "loss": 0.8208, | |
| "step": 2196 | |
| }, | |
| { | |
| "epoch": 16.11, | |
| "learning_rate": 9.595588235294119e-06, | |
| "loss": 0.8459, | |
| "step": 2198 | |
| }, | |
| { | |
| "epoch": 16.12, | |
| "learning_rate": 9.558823529411764e-06, | |
| "loss": 0.8087, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 16.14, | |
| "learning_rate": 9.522058823529411e-06, | |
| "loss": 0.84, | |
| "step": 2202 | |
| }, | |
| { | |
| "epoch": 16.15, | |
| "learning_rate": 9.48529411764706e-06, | |
| "loss": 0.791, | |
| "step": 2204 | |
| }, | |
| { | |
| "epoch": 16.17, | |
| "learning_rate": 9.448529411764707e-06, | |
| "loss": 0.8096, | |
| "step": 2206 | |
| }, | |
| { | |
| "epoch": 16.18, | |
| "learning_rate": 9.411764705882354e-06, | |
| "loss": 0.837, | |
| "step": 2208 | |
| }, | |
| { | |
| "epoch": 16.2, | |
| "learning_rate": 9.375000000000001e-06, | |
| "loss": 0.85, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 16.21, | |
| "learning_rate": 9.338235294117648e-06, | |
| "loss": 0.8501, | |
| "step": 2212 | |
| }, | |
| { | |
| "epoch": 16.23, | |
| "learning_rate": 9.301470588235295e-06, | |
| "loss": 0.8556, | |
| "step": 2214 | |
| }, | |
| { | |
| "epoch": 16.24, | |
| "learning_rate": 9.264705882352942e-06, | |
| "loss": 0.8493, | |
| "step": 2216 | |
| }, | |
| { | |
| "epoch": 16.26, | |
| "learning_rate": 9.227941176470589e-06, | |
| "loss": 0.8204, | |
| "step": 2218 | |
| }, | |
| { | |
| "epoch": 16.27, | |
| "learning_rate": 9.191176470588236e-06, | |
| "loss": 0.8436, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 16.29, | |
| "learning_rate": 9.154411764705882e-06, | |
| "loss": 0.8219, | |
| "step": 2222 | |
| }, | |
| { | |
| "epoch": 16.3, | |
| "learning_rate": 9.117647058823529e-06, | |
| "loss": 0.8244, | |
| "step": 2224 | |
| }, | |
| { | |
| "epoch": 16.32, | |
| "learning_rate": 9.080882352941177e-06, | |
| "loss": 0.8164, | |
| "step": 2226 | |
| }, | |
| { | |
| "epoch": 16.33, | |
| "learning_rate": 9.044117647058824e-06, | |
| "loss": 0.8271, | |
| "step": 2228 | |
| }, | |
| { | |
| "epoch": 16.34, | |
| "learning_rate": 9.007352941176471e-06, | |
| "loss": 0.819, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 16.36, | |
| "learning_rate": 8.970588235294119e-06, | |
| "loss": 0.8254, | |
| "step": 2232 | |
| }, | |
| { | |
| "epoch": 16.37, | |
| "learning_rate": 8.933823529411766e-06, | |
| "loss": 0.8555, | |
| "step": 2234 | |
| }, | |
| { | |
| "epoch": 16.39, | |
| "learning_rate": 8.897058823529413e-06, | |
| "loss": 0.8445, | |
| "step": 2236 | |
| }, | |
| { | |
| "epoch": 16.4, | |
| "learning_rate": 8.86029411764706e-06, | |
| "loss": 0.8336, | |
| "step": 2238 | |
| }, | |
| { | |
| "epoch": 16.42, | |
| "learning_rate": 8.823529411764707e-06, | |
| "loss": 0.8301, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 16.43, | |
| "learning_rate": 8.786764705882352e-06, | |
| "loss": 0.9075, | |
| "step": 2242 | |
| }, | |
| { | |
| "epoch": 16.45, | |
| "learning_rate": 8.75e-06, | |
| "loss": 0.8457, | |
| "step": 2244 | |
| }, | |
| { | |
| "epoch": 16.46, | |
| "learning_rate": 8.713235294117646e-06, | |
| "loss": 0.818, | |
| "step": 2246 | |
| }, | |
| { | |
| "epoch": 16.48, | |
| "learning_rate": 8.676470588235295e-06, | |
| "loss": 0.8632, | |
| "step": 2248 | |
| }, | |
| { | |
| "epoch": 16.49, | |
| "learning_rate": 8.639705882352942e-06, | |
| "loss": 0.8443, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 16.51, | |
| "learning_rate": 8.60294117647059e-06, | |
| "loss": 0.8487, | |
| "step": 2252 | |
| }, | |
| { | |
| "epoch": 16.52, | |
| "learning_rate": 8.566176470588236e-06, | |
| "loss": 0.8031, | |
| "step": 2254 | |
| }, | |
| { | |
| "epoch": 16.54, | |
| "learning_rate": 8.529411764705883e-06, | |
| "loss": 0.7671, | |
| "step": 2256 | |
| }, | |
| { | |
| "epoch": 16.55, | |
| "learning_rate": 8.49264705882353e-06, | |
| "loss": 0.8483, | |
| "step": 2258 | |
| }, | |
| { | |
| "epoch": 16.56, | |
| "learning_rate": 8.455882352941177e-06, | |
| "loss": 0.8293, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 16.58, | |
| "learning_rate": 8.419117647058824e-06, | |
| "loss": 0.8185, | |
| "step": 2262 | |
| }, | |
| { | |
| "epoch": 16.59, | |
| "learning_rate": 8.38235294117647e-06, | |
| "loss": 0.7964, | |
| "step": 2264 | |
| }, | |
| { | |
| "epoch": 16.61, | |
| "learning_rate": 8.345588235294117e-06, | |
| "loss": 0.8519, | |
| "step": 2266 | |
| }, | |
| { | |
| "epoch": 16.62, | |
| "learning_rate": 8.308823529411764e-06, | |
| "loss": 0.7689, | |
| "step": 2268 | |
| }, | |
| { | |
| "epoch": 16.64, | |
| "learning_rate": 8.272058823529413e-06, | |
| "loss": 0.8181, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 16.65, | |
| "learning_rate": 8.23529411764706e-06, | |
| "loss": 0.8239, | |
| "step": 2272 | |
| }, | |
| { | |
| "epoch": 16.67, | |
| "learning_rate": 8.198529411764707e-06, | |
| "loss": 0.8342, | |
| "step": 2274 | |
| }, | |
| { | |
| "epoch": 16.68, | |
| "learning_rate": 8.161764705882354e-06, | |
| "loss": 0.8618, | |
| "step": 2276 | |
| }, | |
| { | |
| "epoch": 16.7, | |
| "learning_rate": 8.125000000000001e-06, | |
| "loss": 0.7793, | |
| "step": 2278 | |
| }, | |
| { | |
| "epoch": 16.71, | |
| "learning_rate": 8.088235294117648e-06, | |
| "loss": 0.8342, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 16.73, | |
| "learning_rate": 8.051470588235295e-06, | |
| "loss": 0.7744, | |
| "step": 2282 | |
| }, | |
| { | |
| "epoch": 16.74, | |
| "learning_rate": 8.014705882352942e-06, | |
| "loss": 0.8342, | |
| "step": 2284 | |
| }, | |
| { | |
| "epoch": 16.75, | |
| "learning_rate": 7.977941176470588e-06, | |
| "loss": 0.8492, | |
| "step": 2286 | |
| }, | |
| { | |
| "epoch": 16.77, | |
| "learning_rate": 7.941176470588235e-06, | |
| "loss": 0.8114, | |
| "step": 2288 | |
| }, | |
| { | |
| "epoch": 16.78, | |
| "learning_rate": 7.904411764705882e-06, | |
| "loss": 0.7961, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 16.8, | |
| "learning_rate": 7.86764705882353e-06, | |
| "loss": 0.8321, | |
| "step": 2292 | |
| }, | |
| { | |
| "epoch": 16.81, | |
| "learning_rate": 7.830882352941177e-06, | |
| "loss": 0.8185, | |
| "step": 2294 | |
| }, | |
| { | |
| "epoch": 16.83, | |
| "learning_rate": 7.794117647058825e-06, | |
| "loss": 0.8419, | |
| "step": 2296 | |
| }, | |
| { | |
| "epoch": 16.84, | |
| "learning_rate": 7.757352941176472e-06, | |
| "loss": 0.8241, | |
| "step": 2298 | |
| }, | |
| { | |
| "epoch": 16.86, | |
| "learning_rate": 7.720588235294119e-06, | |
| "loss": 0.8294, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 16.87, | |
| "learning_rate": 7.683823529411766e-06, | |
| "loss": 0.8553, | |
| "step": 2302 | |
| }, | |
| { | |
| "epoch": 16.89, | |
| "learning_rate": 7.647058823529413e-06, | |
| "loss": 0.7851, | |
| "step": 2304 | |
| }, | |
| { | |
| "epoch": 16.9, | |
| "learning_rate": 7.610294117647058e-06, | |
| "loss": 0.8508, | |
| "step": 2306 | |
| }, | |
| { | |
| "epoch": 16.92, | |
| "learning_rate": 7.573529411764706e-06, | |
| "loss": 0.8455, | |
| "step": 2308 | |
| }, | |
| { | |
| "epoch": 16.93, | |
| "learning_rate": 7.536764705882353e-06, | |
| "loss": 0.8724, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 16.95, | |
| "learning_rate": 7.5e-06, | |
| "loss": 0.8606, | |
| "step": 2312 | |
| }, | |
| { | |
| "epoch": 16.96, | |
| "learning_rate": 7.463235294117647e-06, | |
| "loss": 0.8677, | |
| "step": 2314 | |
| }, | |
| { | |
| "epoch": 16.97, | |
| "learning_rate": 7.426470588235294e-06, | |
| "loss": 0.8488, | |
| "step": 2316 | |
| }, | |
| { | |
| "epoch": 16.99, | |
| "learning_rate": 7.389705882352941e-06, | |
| "loss": 0.8169, | |
| "step": 2318 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_gen_len": 128.86316872427983, | |
| "eval_loss": 1.6075981855392456, | |
| "eval_rouge1": 55.8274, | |
| "eval_rouge2": 30.41, | |
| "eval_rougeL": 38.6258, | |
| "eval_rougeLsum": 51.3009, | |
| "eval_runtime": 219.3849, | |
| "eval_samples_per_second": 4.431, | |
| "eval_steps_per_second": 0.556, | |
| "step": 2319 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "learning_rate": 7.3529411764705884e-06, | |
| "loss": 0.7751, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 7.316176470588236e-06, | |
| "loss": 0.8066, | |
| "step": 2322 | |
| }, | |
| { | |
| "epoch": 17.03, | |
| "learning_rate": 7.279411764705883e-06, | |
| "loss": 0.8462, | |
| "step": 2324 | |
| }, | |
| { | |
| "epoch": 17.05, | |
| "learning_rate": 7.2426470588235305e-06, | |
| "loss": 0.8035, | |
| "step": 2326 | |
| }, | |
| { | |
| "epoch": 17.06, | |
| "learning_rate": 7.205882352941176e-06, | |
| "loss": 0.7691, | |
| "step": 2328 | |
| }, | |
| { | |
| "epoch": 17.08, | |
| "learning_rate": 7.169117647058824e-06, | |
| "loss": 0.7958, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 17.09, | |
| "learning_rate": 7.132352941176471e-06, | |
| "loss": 0.7804, | |
| "step": 2332 | |
| }, | |
| { | |
| "epoch": 17.11, | |
| "learning_rate": 7.095588235294118e-06, | |
| "loss": 0.8048, | |
| "step": 2334 | |
| }, | |
| { | |
| "epoch": 17.12, | |
| "learning_rate": 7.058823529411765e-06, | |
| "loss": 0.7775, | |
| "step": 2336 | |
| }, | |
| { | |
| "epoch": 17.14, | |
| "learning_rate": 7.022058823529412e-06, | |
| "loss": 0.8318, | |
| "step": 2338 | |
| }, | |
| { | |
| "epoch": 17.15, | |
| "learning_rate": 6.985294117647059e-06, | |
| "loss": 0.8566, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 17.17, | |
| "learning_rate": 6.948529411764706e-06, | |
| "loss": 0.797, | |
| "step": 2342 | |
| }, | |
| { | |
| "epoch": 17.18, | |
| "learning_rate": 6.911764705882354e-06, | |
| "loss": 0.8252, | |
| "step": 2344 | |
| }, | |
| { | |
| "epoch": 17.19, | |
| "learning_rate": 6.875000000000001e-06, | |
| "loss": 0.7782, | |
| "step": 2346 | |
| }, | |
| { | |
| "epoch": 17.21, | |
| "learning_rate": 6.838235294117648e-06, | |
| "loss": 0.8611, | |
| "step": 2348 | |
| }, | |
| { | |
| "epoch": 17.22, | |
| "learning_rate": 6.8014705882352935e-06, | |
| "loss": 0.8206, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 17.24, | |
| "learning_rate": 6.7647058823529414e-06, | |
| "loss": 0.829, | |
| "step": 2352 | |
| }, | |
| { | |
| "epoch": 17.25, | |
| "learning_rate": 6.7279411764705885e-06, | |
| "loss": 0.7828, | |
| "step": 2354 | |
| }, | |
| { | |
| "epoch": 17.27, | |
| "learning_rate": 6.6911764705882356e-06, | |
| "loss": 0.7837, | |
| "step": 2356 | |
| }, | |
| { | |
| "epoch": 17.28, | |
| "learning_rate": 6.654411764705883e-06, | |
| "loss": 0.8112, | |
| "step": 2358 | |
| }, | |
| { | |
| "epoch": 17.3, | |
| "learning_rate": 6.61764705882353e-06, | |
| "loss": 0.8087, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 17.31, | |
| "learning_rate": 6.580882352941177e-06, | |
| "loss": 0.7931, | |
| "step": 2362 | |
| }, | |
| { | |
| "epoch": 17.33, | |
| "learning_rate": 6.544117647058824e-06, | |
| "loss": 0.8214, | |
| "step": 2364 | |
| }, | |
| { | |
| "epoch": 17.34, | |
| "learning_rate": 6.507352941176472e-06, | |
| "loss": 0.7835, | |
| "step": 2366 | |
| }, | |
| { | |
| "epoch": 17.36, | |
| "learning_rate": 6.470588235294119e-06, | |
| "loss": 0.8415, | |
| "step": 2368 | |
| }, | |
| { | |
| "epoch": 17.37, | |
| "learning_rate": 6.433823529411764e-06, | |
| "loss": 0.8249, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 17.39, | |
| "learning_rate": 6.397058823529411e-06, | |
| "loss": 0.7932, | |
| "step": 2372 | |
| }, | |
| { | |
| "epoch": 17.4, | |
| "learning_rate": 6.360294117647059e-06, | |
| "loss": 0.844, | |
| "step": 2374 | |
| }, | |
| { | |
| "epoch": 17.41, | |
| "learning_rate": 6.323529411764706e-06, | |
| "loss": 0.8582, | |
| "step": 2376 | |
| }, | |
| { | |
| "epoch": 17.43, | |
| "learning_rate": 6.286764705882353e-06, | |
| "loss": 0.8629, | |
| "step": 2378 | |
| }, | |
| { | |
| "epoch": 17.44, | |
| "learning_rate": 6.25e-06, | |
| "loss": 0.828, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 17.46, | |
| "learning_rate": 6.213235294117647e-06, | |
| "loss": 0.8536, | |
| "step": 2382 | |
| }, | |
| { | |
| "epoch": 17.47, | |
| "learning_rate": 6.1764705882352944e-06, | |
| "loss": 0.79, | |
| "step": 2384 | |
| }, | |
| { | |
| "epoch": 17.49, | |
| "learning_rate": 6.1397058823529415e-06, | |
| "loss": 0.7974, | |
| "step": 2386 | |
| }, | |
| { | |
| "epoch": 17.5, | |
| "learning_rate": 6.1029411764705885e-06, | |
| "loss": 0.8383, | |
| "step": 2388 | |
| }, | |
| { | |
| "epoch": 17.52, | |
| "learning_rate": 6.066176470588236e-06, | |
| "loss": 0.8324, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 17.53, | |
| "learning_rate": 6.029411764705883e-06, | |
| "loss": 0.786, | |
| "step": 2392 | |
| }, | |
| { | |
| "epoch": 17.55, | |
| "learning_rate": 5.99264705882353e-06, | |
| "loss": 0.8319, | |
| "step": 2394 | |
| }, | |
| { | |
| "epoch": 17.56, | |
| "learning_rate": 5.955882352941177e-06, | |
| "loss": 0.8172, | |
| "step": 2396 | |
| }, | |
| { | |
| "epoch": 17.58, | |
| "learning_rate": 5.919117647058824e-06, | |
| "loss": 0.8089, | |
| "step": 2398 | |
| }, | |
| { | |
| "epoch": 17.59, | |
| "learning_rate": 5.882352941176471e-06, | |
| "loss": 0.7822, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 17.61, | |
| "learning_rate": 5.845588235294118e-06, | |
| "loss": 0.8161, | |
| "step": 2402 | |
| }, | |
| { | |
| "epoch": 17.62, | |
| "learning_rate": 5.808823529411765e-06, | |
| "loss": 0.8333, | |
| "step": 2404 | |
| }, | |
| { | |
| "epoch": 17.63, | |
| "learning_rate": 5.772058823529412e-06, | |
| "loss": 0.7785, | |
| "step": 2406 | |
| }, | |
| { | |
| "epoch": 17.65, | |
| "learning_rate": 5.735294117647059e-06, | |
| "loss": 0.7676, | |
| "step": 2408 | |
| }, | |
| { | |
| "epoch": 17.66, | |
| "learning_rate": 5.698529411764706e-06, | |
| "loss": 0.8038, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 17.68, | |
| "learning_rate": 5.661764705882353e-06, | |
| "loss": 0.8318, | |
| "step": 2412 | |
| }, | |
| { | |
| "epoch": 17.69, | |
| "learning_rate": 5.625e-06, | |
| "loss": 0.8082, | |
| "step": 2414 | |
| }, | |
| { | |
| "epoch": 17.71, | |
| "learning_rate": 5.588235294117647e-06, | |
| "loss": 0.8248, | |
| "step": 2416 | |
| }, | |
| { | |
| "epoch": 17.72, | |
| "learning_rate": 5.5514705882352945e-06, | |
| "loss": 0.7761, | |
| "step": 2418 | |
| }, | |
| { | |
| "epoch": 17.74, | |
| "learning_rate": 5.5147058823529415e-06, | |
| "loss": 0.8286, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 17.75, | |
| "learning_rate": 5.477941176470589e-06, | |
| "loss": 0.8185, | |
| "step": 2422 | |
| }, | |
| { | |
| "epoch": 17.77, | |
| "learning_rate": 5.441176470588236e-06, | |
| "loss": 0.8263, | |
| "step": 2424 | |
| }, | |
| { | |
| "epoch": 17.78, | |
| "learning_rate": 5.404411764705883e-06, | |
| "loss": 0.8006, | |
| "step": 2426 | |
| }, | |
| { | |
| "epoch": 17.8, | |
| "learning_rate": 5.367647058823529e-06, | |
| "loss": 0.8668, | |
| "step": 2428 | |
| }, | |
| { | |
| "epoch": 17.81, | |
| "learning_rate": 5.330882352941177e-06, | |
| "loss": 0.7969, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 17.83, | |
| "learning_rate": 5.294117647058824e-06, | |
| "loss": 0.8191, | |
| "step": 2432 | |
| }, | |
| { | |
| "epoch": 17.84, | |
| "learning_rate": 5.257352941176471e-06, | |
| "loss": 0.8315, | |
| "step": 2434 | |
| }, | |
| { | |
| "epoch": 17.85, | |
| "learning_rate": 5.220588235294118e-06, | |
| "loss": 0.7925, | |
| "step": 2436 | |
| }, | |
| { | |
| "epoch": 17.87, | |
| "learning_rate": 5.183823529411765e-06, | |
| "loss": 0.8377, | |
| "step": 2438 | |
| }, | |
| { | |
| "epoch": 17.88, | |
| "learning_rate": 5.147058823529412e-06, | |
| "loss": 0.8196, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 17.9, | |
| "learning_rate": 5.110294117647059e-06, | |
| "loss": 0.7612, | |
| "step": 2442 | |
| }, | |
| { | |
| "epoch": 17.91, | |
| "learning_rate": 5.073529411764706e-06, | |
| "loss": 0.8375, | |
| "step": 2444 | |
| }, | |
| { | |
| "epoch": 17.93, | |
| "learning_rate": 5.036764705882353e-06, | |
| "loss": 0.8251, | |
| "step": 2446 | |
| }, | |
| { | |
| "epoch": 17.94, | |
| "learning_rate": 5e-06, | |
| "loss": 0.8507, | |
| "step": 2448 | |
| }, | |
| { | |
| "epoch": 17.96, | |
| "learning_rate": 4.963235294117647e-06, | |
| "loss": 0.8701, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 17.97, | |
| "learning_rate": 4.9264705882352945e-06, | |
| "loss": 0.8294, | |
| "step": 2452 | |
| }, | |
| { | |
| "epoch": 17.99, | |
| "learning_rate": 4.889705882352942e-06, | |
| "loss": 0.8194, | |
| "step": 2454 | |
| }, | |
| { | |
| "epoch": 17.99, | |
| "eval_gen_len": 128.64609053497944, | |
| "eval_loss": 1.6177321672439575, | |
| "eval_rouge1": 56.3214, | |
| "eval_rouge2": 30.9896, | |
| "eval_rougeL": 39.4754, | |
| "eval_rougeLsum": 51.9525, | |
| "eval_runtime": 216.2607, | |
| "eval_samples_per_second": 4.495, | |
| "eval_steps_per_second": 0.564, | |
| "step": 2455 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "learning_rate": 4.852941176470589e-06, | |
| "loss": 0.8531, | |
| "step": 2456 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 4.816176470588236e-06, | |
| "loss": 0.7938, | |
| "step": 2458 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "learning_rate": 4.779411764705882e-06, | |
| "loss": 0.8304, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 18.04, | |
| "learning_rate": 4.74264705882353e-06, | |
| "loss": 0.7959, | |
| "step": 2462 | |
| }, | |
| { | |
| "epoch": 18.06, | |
| "learning_rate": 4.705882352941177e-06, | |
| "loss": 0.8089, | |
| "step": 2464 | |
| }, | |
| { | |
| "epoch": 18.07, | |
| "learning_rate": 4.669117647058824e-06, | |
| "loss": 0.8112, | |
| "step": 2466 | |
| }, | |
| { | |
| "epoch": 18.09, | |
| "learning_rate": 4.632352941176471e-06, | |
| "loss": 0.8237, | |
| "step": 2468 | |
| }, | |
| { | |
| "epoch": 18.1, | |
| "learning_rate": 4.595588235294118e-06, | |
| "loss": 0.8255, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 18.12, | |
| "learning_rate": 4.558823529411764e-06, | |
| "loss": 0.8026, | |
| "step": 2472 | |
| }, | |
| { | |
| "epoch": 18.13, | |
| "learning_rate": 4.522058823529412e-06, | |
| "loss": 0.8219, | |
| "step": 2474 | |
| }, | |
| { | |
| "epoch": 18.15, | |
| "learning_rate": 4.485294117647059e-06, | |
| "loss": 0.7687, | |
| "step": 2476 | |
| }, | |
| { | |
| "epoch": 18.16, | |
| "learning_rate": 4.448529411764706e-06, | |
| "loss": 0.7815, | |
| "step": 2478 | |
| }, | |
| { | |
| "epoch": 18.18, | |
| "learning_rate": 4.411764705882353e-06, | |
| "loss": 0.7863, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 18.19, | |
| "learning_rate": 4.375e-06, | |
| "loss": 0.805, | |
| "step": 2482 | |
| }, | |
| { | |
| "epoch": 18.21, | |
| "learning_rate": 4.3382352941176475e-06, | |
| "loss": 0.7961, | |
| "step": 2484 | |
| }, | |
| { | |
| "epoch": 18.22, | |
| "learning_rate": 4.301470588235295e-06, | |
| "loss": 0.8162, | |
| "step": 2486 | |
| }, | |
| { | |
| "epoch": 18.24, | |
| "learning_rate": 4.264705882352942e-06, | |
| "loss": 0.8307, | |
| "step": 2488 | |
| }, | |
| { | |
| "epoch": 18.25, | |
| "learning_rate": 4.227941176470589e-06, | |
| "loss": 0.8335, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 18.26, | |
| "learning_rate": 4.191176470588235e-06, | |
| "loss": 0.7927, | |
| "step": 2492 | |
| }, | |
| { | |
| "epoch": 18.28, | |
| "learning_rate": 4.154411764705882e-06, | |
| "loss": 0.7851, | |
| "step": 2494 | |
| }, | |
| { | |
| "epoch": 18.29, | |
| "learning_rate": 4.11764705882353e-06, | |
| "loss": 0.7844, | |
| "step": 2496 | |
| }, | |
| { | |
| "epoch": 18.31, | |
| "learning_rate": 4.080882352941177e-06, | |
| "loss": 0.8222, | |
| "step": 2498 | |
| }, | |
| { | |
| "epoch": 18.32, | |
| "learning_rate": 4.044117647058824e-06, | |
| "loss": 0.8013, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 18.34, | |
| "learning_rate": 4.007352941176471e-06, | |
| "loss": 0.8087, | |
| "step": 2502 | |
| }, | |
| { | |
| "epoch": 18.35, | |
| "learning_rate": 3.970588235294117e-06, | |
| "loss": 0.8239, | |
| "step": 2504 | |
| }, | |
| { | |
| "epoch": 18.37, | |
| "learning_rate": 3.933823529411765e-06, | |
| "loss": 0.8153, | |
| "step": 2506 | |
| }, | |
| { | |
| "epoch": 18.38, | |
| "learning_rate": 3.897058823529412e-06, | |
| "loss": 0.8534, | |
| "step": 2508 | |
| }, | |
| { | |
| "epoch": 18.4, | |
| "learning_rate": 3.860294117647059e-06, | |
| "loss": 0.7795, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 18.41, | |
| "learning_rate": 3.823529411764706e-06, | |
| "loss": 0.7732, | |
| "step": 2512 | |
| }, | |
| { | |
| "epoch": 18.43, | |
| "learning_rate": 3.786764705882353e-06, | |
| "loss": 0.8287, | |
| "step": 2514 | |
| }, | |
| { | |
| "epoch": 18.44, | |
| "learning_rate": 3.75e-06, | |
| "loss": 0.7886, | |
| "step": 2516 | |
| }, | |
| { | |
| "epoch": 18.46, | |
| "learning_rate": 3.713235294117647e-06, | |
| "loss": 0.7978, | |
| "step": 2518 | |
| }, | |
| { | |
| "epoch": 18.47, | |
| "learning_rate": 3.6764705882352942e-06, | |
| "loss": 0.8232, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 18.48, | |
| "learning_rate": 3.6397058823529417e-06, | |
| "loss": 0.7782, | |
| "step": 2522 | |
| }, | |
| { | |
| "epoch": 18.5, | |
| "learning_rate": 3.602941176470588e-06, | |
| "loss": 0.8201, | |
| "step": 2524 | |
| }, | |
| { | |
| "epoch": 18.51, | |
| "learning_rate": 3.5661764705882354e-06, | |
| "loss": 0.7897, | |
| "step": 2526 | |
| }, | |
| { | |
| "epoch": 18.53, | |
| "learning_rate": 3.5294117647058825e-06, | |
| "loss": 0.7692, | |
| "step": 2528 | |
| }, | |
| { | |
| "epoch": 18.54, | |
| "learning_rate": 3.4926470588235295e-06, | |
| "loss": 0.784, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 18.56, | |
| "learning_rate": 3.455882352941177e-06, | |
| "loss": 0.8474, | |
| "step": 2532 | |
| }, | |
| { | |
| "epoch": 18.57, | |
| "learning_rate": 3.419117647058824e-06, | |
| "loss": 0.7644, | |
| "step": 2534 | |
| }, | |
| { | |
| "epoch": 18.59, | |
| "learning_rate": 3.3823529411764707e-06, | |
| "loss": 0.7707, | |
| "step": 2536 | |
| }, | |
| { | |
| "epoch": 18.6, | |
| "learning_rate": 3.3455882352941178e-06, | |
| "loss": 0.7818, | |
| "step": 2538 | |
| }, | |
| { | |
| "epoch": 18.62, | |
| "learning_rate": 3.308823529411765e-06, | |
| "loss": 0.797, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 18.63, | |
| "learning_rate": 3.272058823529412e-06, | |
| "loss": 0.8142, | |
| "step": 2542 | |
| }, | |
| { | |
| "epoch": 18.65, | |
| "learning_rate": 3.2352941176470594e-06, | |
| "loss": 0.7797, | |
| "step": 2544 | |
| }, | |
| { | |
| "epoch": 18.66, | |
| "learning_rate": 3.1985294117647056e-06, | |
| "loss": 0.8282, | |
| "step": 2546 | |
| }, | |
| { | |
| "epoch": 18.68, | |
| "learning_rate": 3.161764705882353e-06, | |
| "loss": 0.7816, | |
| "step": 2548 | |
| }, | |
| { | |
| "epoch": 18.69, | |
| "learning_rate": 3.125e-06, | |
| "loss": 0.7697, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 18.7, | |
| "learning_rate": 3.0882352941176472e-06, | |
| "loss": 0.8033, | |
| "step": 2552 | |
| }, | |
| { | |
| "epoch": 18.72, | |
| "learning_rate": 3.0514705882352943e-06, | |
| "loss": 0.8221, | |
| "step": 2554 | |
| }, | |
| { | |
| "epoch": 18.73, | |
| "learning_rate": 3.0147058823529413e-06, | |
| "loss": 0.7657, | |
| "step": 2556 | |
| }, | |
| { | |
| "epoch": 18.75, | |
| "learning_rate": 2.9779411764705884e-06, | |
| "loss": 0.7999, | |
| "step": 2558 | |
| }, | |
| { | |
| "epoch": 18.76, | |
| "learning_rate": 2.9411764705882355e-06, | |
| "loss": 0.7996, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 18.78, | |
| "learning_rate": 2.9044117647058825e-06, | |
| "loss": 0.7903, | |
| "step": 2562 | |
| }, | |
| { | |
| "epoch": 18.79, | |
| "learning_rate": 2.8676470588235296e-06, | |
| "loss": 0.7895, | |
| "step": 2564 | |
| }, | |
| { | |
| "epoch": 18.81, | |
| "learning_rate": 2.8308823529411766e-06, | |
| "loss": 0.8348, | |
| "step": 2566 | |
| }, | |
| { | |
| "epoch": 18.82, | |
| "learning_rate": 2.7941176470588237e-06, | |
| "loss": 0.7819, | |
| "step": 2568 | |
| }, | |
| { | |
| "epoch": 18.84, | |
| "learning_rate": 2.7573529411764708e-06, | |
| "loss": 0.8005, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 18.85, | |
| "learning_rate": 2.720588235294118e-06, | |
| "loss": 0.7874, | |
| "step": 2572 | |
| }, | |
| { | |
| "epoch": 18.87, | |
| "learning_rate": 2.6838235294117645e-06, | |
| "loss": 0.7964, | |
| "step": 2574 | |
| }, | |
| { | |
| "epoch": 18.88, | |
| "learning_rate": 2.647058823529412e-06, | |
| "loss": 0.8508, | |
| "step": 2576 | |
| }, | |
| { | |
| "epoch": 18.9, | |
| "learning_rate": 2.610294117647059e-06, | |
| "loss": 0.8511, | |
| "step": 2578 | |
| }, | |
| { | |
| "epoch": 18.91, | |
| "learning_rate": 2.573529411764706e-06, | |
| "loss": 0.8308, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 18.92, | |
| "learning_rate": 2.536764705882353e-06, | |
| "loss": 0.8333, | |
| "step": 2582 | |
| }, | |
| { | |
| "epoch": 18.94, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.8121, | |
| "step": 2584 | |
| }, | |
| { | |
| "epoch": 18.95, | |
| "learning_rate": 2.4632352941176473e-06, | |
| "loss": 0.7601, | |
| "step": 2586 | |
| }, | |
| { | |
| "epoch": 18.97, | |
| "learning_rate": 2.4264705882352943e-06, | |
| "loss": 0.9098, | |
| "step": 2588 | |
| }, | |
| { | |
| "epoch": 18.98, | |
| "learning_rate": 2.389705882352941e-06, | |
| "loss": 0.8387, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "learning_rate": 2.3529411764705885e-06, | |
| "loss": 0.8441, | |
| "step": 2592 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_gen_len": 128.82407407407408, | |
| "eval_loss": 1.625985026359558, | |
| "eval_rouge1": 55.9842, | |
| "eval_rouge2": 30.6332, | |
| "eval_rougeL": 38.999, | |
| "eval_rougeLsum": 51.5685, | |
| "eval_runtime": 216.0054, | |
| "eval_samples_per_second": 4.5, | |
| "eval_steps_per_second": 0.565, | |
| "step": 2592 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 2.3161764705882355e-06, | |
| "loss": 0.7704, | |
| "step": 2594 | |
| }, | |
| { | |
| "epoch": 19.03, | |
| "learning_rate": 2.279411764705882e-06, | |
| "loss": 0.79, | |
| "step": 2596 | |
| }, | |
| { | |
| "epoch": 19.04, | |
| "learning_rate": 2.2426470588235296e-06, | |
| "loss": 0.8014, | |
| "step": 2598 | |
| }, | |
| { | |
| "epoch": 19.06, | |
| "learning_rate": 2.2058823529411767e-06, | |
| "loss": 0.8042, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 19.07, | |
| "learning_rate": 2.1691176470588238e-06, | |
| "loss": 0.7609, | |
| "step": 2602 | |
| }, | |
| { | |
| "epoch": 19.09, | |
| "learning_rate": 2.132352941176471e-06, | |
| "loss": 0.8418, | |
| "step": 2604 | |
| }, | |
| { | |
| "epoch": 19.1, | |
| "learning_rate": 2.0955882352941175e-06, | |
| "loss": 0.7901, | |
| "step": 2606 | |
| }, | |
| { | |
| "epoch": 19.11, | |
| "learning_rate": 2.058823529411765e-06, | |
| "loss": 0.7841, | |
| "step": 2608 | |
| }, | |
| { | |
| "epoch": 19.13, | |
| "learning_rate": 2.022058823529412e-06, | |
| "loss": 0.778, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 19.14, | |
| "learning_rate": 1.9852941176470586e-06, | |
| "loss": 0.8207, | |
| "step": 2612 | |
| }, | |
| { | |
| "epoch": 19.16, | |
| "learning_rate": 1.948529411764706e-06, | |
| "loss": 0.7564, | |
| "step": 2614 | |
| }, | |
| { | |
| "epoch": 19.17, | |
| "learning_rate": 1.911764705882353e-06, | |
| "loss": 0.7915, | |
| "step": 2616 | |
| }, | |
| { | |
| "epoch": 19.19, | |
| "learning_rate": 1.875e-06, | |
| "loss": 0.7916, | |
| "step": 2618 | |
| }, | |
| { | |
| "epoch": 19.2, | |
| "learning_rate": 1.8382352941176471e-06, | |
| "loss": 0.8082, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 19.22, | |
| "learning_rate": 1.801470588235294e-06, | |
| "loss": 0.8426, | |
| "step": 2622 | |
| }, | |
| { | |
| "epoch": 19.23, | |
| "learning_rate": 1.7647058823529412e-06, | |
| "loss": 0.8074, | |
| "step": 2624 | |
| }, | |
| { | |
| "epoch": 19.25, | |
| "learning_rate": 1.7279411764705885e-06, | |
| "loss": 0.8121, | |
| "step": 2626 | |
| }, | |
| { | |
| "epoch": 19.26, | |
| "learning_rate": 1.6911764705882354e-06, | |
| "loss": 0.8088, | |
| "step": 2628 | |
| }, | |
| { | |
| "epoch": 19.28, | |
| "learning_rate": 1.6544117647058824e-06, | |
| "loss": 0.816, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 19.29, | |
| "learning_rate": 1.6176470588235297e-06, | |
| "loss": 0.7726, | |
| "step": 2632 | |
| }, | |
| { | |
| "epoch": 19.31, | |
| "learning_rate": 1.5808823529411765e-06, | |
| "loss": 0.7635, | |
| "step": 2634 | |
| }, | |
| { | |
| "epoch": 19.32, | |
| "learning_rate": 1.5441176470588236e-06, | |
| "loss": 0.7938, | |
| "step": 2636 | |
| }, | |
| { | |
| "epoch": 19.33, | |
| "learning_rate": 1.5073529411764707e-06, | |
| "loss": 0.7696, | |
| "step": 2638 | |
| }, | |
| { | |
| "epoch": 19.35, | |
| "learning_rate": 1.4705882352941177e-06, | |
| "loss": 0.8152, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 19.36, | |
| "learning_rate": 1.4338235294117648e-06, | |
| "loss": 0.8182, | |
| "step": 2642 | |
| }, | |
| { | |
| "epoch": 19.38, | |
| "learning_rate": 1.3970588235294119e-06, | |
| "loss": 0.7999, | |
| "step": 2644 | |
| }, | |
| { | |
| "epoch": 19.39, | |
| "learning_rate": 1.360294117647059e-06, | |
| "loss": 0.8111, | |
| "step": 2646 | |
| }, | |
| { | |
| "epoch": 19.41, | |
| "learning_rate": 1.323529411764706e-06, | |
| "loss": 0.7841, | |
| "step": 2648 | |
| }, | |
| { | |
| "epoch": 19.42, | |
| "learning_rate": 1.286764705882353e-06, | |
| "loss": 0.7895, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 19.44, | |
| "learning_rate": 1.25e-06, | |
| "loss": 0.7895, | |
| "step": 2652 | |
| }, | |
| { | |
| "epoch": 19.45, | |
| "learning_rate": 1.2132352941176472e-06, | |
| "loss": 0.8028, | |
| "step": 2654 | |
| }, | |
| { | |
| "epoch": 19.47, | |
| "learning_rate": 1.1764705882352942e-06, | |
| "loss": 0.7947, | |
| "step": 2656 | |
| }, | |
| { | |
| "epoch": 19.48, | |
| "learning_rate": 1.139705882352941e-06, | |
| "loss": 0.7839, | |
| "step": 2658 | |
| }, | |
| { | |
| "epoch": 19.5, | |
| "learning_rate": 1.1029411764705884e-06, | |
| "loss": 0.8313, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 19.51, | |
| "learning_rate": 1.0661764705882354e-06, | |
| "loss": 0.803, | |
| "step": 2662 | |
| }, | |
| { | |
| "epoch": 19.53, | |
| "learning_rate": 1.0294117647058825e-06, | |
| "loss": 0.7748, | |
| "step": 2664 | |
| }, | |
| { | |
| "epoch": 19.54, | |
| "learning_rate": 9.926470588235293e-07, | |
| "loss": 0.7863, | |
| "step": 2666 | |
| }, | |
| { | |
| "epoch": 19.55, | |
| "learning_rate": 9.558823529411766e-07, | |
| "loss": 0.838, | |
| "step": 2668 | |
| }, | |
| { | |
| "epoch": 19.57, | |
| "learning_rate": 9.191176470588236e-07, | |
| "loss": 0.7913, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 19.58, | |
| "learning_rate": 8.823529411764706e-07, | |
| "loss": 0.8057, | |
| "step": 2672 | |
| }, | |
| { | |
| "epoch": 19.6, | |
| "learning_rate": 8.455882352941177e-07, | |
| "loss": 0.8014, | |
| "step": 2674 | |
| }, | |
| { | |
| "epoch": 19.61, | |
| "learning_rate": 8.088235294117648e-07, | |
| "loss": 0.8081, | |
| "step": 2676 | |
| }, | |
| { | |
| "epoch": 19.63, | |
| "learning_rate": 7.720588235294118e-07, | |
| "loss": 0.7946, | |
| "step": 2678 | |
| }, | |
| { | |
| "epoch": 19.64, | |
| "learning_rate": 7.352941176470589e-07, | |
| "loss": 0.7908, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 19.66, | |
| "learning_rate": 6.985294117647059e-07, | |
| "loss": 0.7736, | |
| "step": 2682 | |
| }, | |
| { | |
| "epoch": 19.67, | |
| "learning_rate": 6.61764705882353e-07, | |
| "loss": 0.7925, | |
| "step": 2684 | |
| }, | |
| { | |
| "epoch": 19.69, | |
| "learning_rate": 6.25e-07, | |
| "loss": 0.7959, | |
| "step": 2686 | |
| }, | |
| { | |
| "epoch": 19.7, | |
| "learning_rate": 5.882352941176471e-07, | |
| "loss": 0.7814, | |
| "step": 2688 | |
| }, | |
| { | |
| "epoch": 19.72, | |
| "learning_rate": 5.514705882352942e-07, | |
| "loss": 0.7872, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 19.73, | |
| "learning_rate": 5.147058823529412e-07, | |
| "loss": 0.804, | |
| "step": 2692 | |
| }, | |
| { | |
| "epoch": 19.75, | |
| "learning_rate": 4.779411764705883e-07, | |
| "loss": 0.7901, | |
| "step": 2694 | |
| }, | |
| { | |
| "epoch": 19.76, | |
| "learning_rate": 4.411764705882353e-07, | |
| "loss": 0.7893, | |
| "step": 2696 | |
| }, | |
| { | |
| "epoch": 19.77, | |
| "learning_rate": 4.044117647058824e-07, | |
| "loss": 0.7764, | |
| "step": 2698 | |
| }, | |
| { | |
| "epoch": 19.79, | |
| "learning_rate": 3.6764705882352943e-07, | |
| "loss": 0.8344, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 19.8, | |
| "learning_rate": 3.308823529411765e-07, | |
| "loss": 0.8093, | |
| "step": 2702 | |
| }, | |
| { | |
| "epoch": 19.82, | |
| "learning_rate": 2.9411764705882356e-07, | |
| "loss": 0.7541, | |
| "step": 2704 | |
| }, | |
| { | |
| "epoch": 19.83, | |
| "learning_rate": 2.573529411764706e-07, | |
| "loss": 0.8165, | |
| "step": 2706 | |
| }, | |
| { | |
| "epoch": 19.85, | |
| "learning_rate": 2.2058823529411765e-07, | |
| "loss": 0.8012, | |
| "step": 2708 | |
| }, | |
| { | |
| "epoch": 19.86, | |
| "learning_rate": 1.8382352941176472e-07, | |
| "loss": 0.7605, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 19.88, | |
| "learning_rate": 1.4705882352941178e-07, | |
| "loss": 0.8056, | |
| "step": 2712 | |
| }, | |
| { | |
| "epoch": 19.89, | |
| "learning_rate": 1.1029411764705883e-07, | |
| "loss": 0.8107, | |
| "step": 2714 | |
| }, | |
| { | |
| "epoch": 19.91, | |
| "learning_rate": 7.352941176470589e-08, | |
| "loss": 0.7975, | |
| "step": 2716 | |
| }, | |
| { | |
| "epoch": 19.92, | |
| "learning_rate": 3.6764705882352945e-08, | |
| "loss": 0.8409, | |
| "step": 2718 | |
| }, | |
| { | |
| "epoch": 19.94, | |
| "learning_rate": 0.0, | |
| "loss": 0.792, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 19.94, | |
| "eval_gen_len": 128.74074074074073, | |
| "eval_loss": 1.6327582597732544, | |
| "eval_rouge1": 55.8983, | |
| "eval_rouge2": 30.5018, | |
| "eval_rougeL": 38.7764, | |
| "eval_rougeLsum": 51.3611, | |
| "eval_runtime": 216.9358, | |
| "eval_samples_per_second": 4.481, | |
| "eval_steps_per_second": 0.562, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 19.94, | |
| "step": 2720, | |
| "total_flos": 7.21120699038892e+17, | |
| "train_loss": 1.0796797680504182, | |
| "train_runtime": 51892.12, | |
| "train_samples_per_second": 6.728, | |
| "train_steps_per_second": 0.052 | |
| } | |
| ], | |
| "logging_steps": 2, | |
| "max_steps": 2720, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 20, | |
| "save_steps": 500, | |
| "total_flos": 7.21120699038892e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |