sujithatz/finbot-quen2.5-merged_adapter
Browse files- README.md +794 -202
- adapter_config.json +2 -7
- adapter_model.safetensors +2 -2
- all_results.json +13 -0
- eval_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +2997 -0
- training_args.bin +1 -1
README.md
CHANGED
|
@@ -18,7 +18,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 18 |
|
| 19 |
This model is a fine-tuned version of [Qwen/Qwen2.5-0.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-0.5B-Instruct) on an unknown dataset.
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
-
- Loss: 0.
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
|
@@ -37,216 +37,808 @@ More information needed
|
|
| 37 |
### Training hyperparameters
|
| 38 |
|
| 39 |
The following hyperparameters were used during training:
|
| 40 |
-
- learning_rate:
|
| 41 |
- train_batch_size: 4
|
| 42 |
- eval_batch_size: 4
|
| 43 |
- seed: 0
|
| 44 |
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
|
| 45 |
- lr_scheduler_type: cosine
|
| 46 |
- lr_scheduler_warmup_ratio: 0.01
|
| 47 |
-
- num_epochs:
|
| 48 |
|
| 49 |
### Training results
|
| 50 |
|
| 51 |
-
| Training Loss | Epoch
|
| 52 |
-
|
| 53 |
-
| 3.
|
| 54 |
-
|
|
| 55 |
-
|
|
| 56 |
-
|
|
| 57 |
-
|
|
| 58 |
-
|
|
| 59 |
-
|
|
| 60 |
-
|
|
| 61 |
-
|
|
| 62 |
-
|
|
| 63 |
-
|
|
| 64 |
-
|
|
| 65 |
-
|
|
| 66 |
-
|
|
| 67 |
-
|
|
| 68 |
-
|
|
| 69 |
-
|
|
| 70 |
-
|
|
| 71 |
-
|
|
| 72 |
-
|
|
| 73 |
-
|
|
| 74 |
-
|
|
| 75 |
-
|
|
| 76 |
-
|
|
| 77 |
-
|
|
| 78 |
-
|
|
| 79 |
-
|
|
| 80 |
-
|
|
| 81 |
-
|
|
| 82 |
-
|
|
| 83 |
-
|
|
| 84 |
-
|
|
| 85 |
-
|
|
| 86 |
-
|
|
| 87 |
-
|
|
| 88 |
-
|
|
| 89 |
-
|
|
| 90 |
-
|
|
| 91 |
-
|
|
| 92 |
-
|
|
| 93 |
-
|
|
| 94 |
-
|
|
| 95 |
-
|
|
| 96 |
-
|
|
| 97 |
-
|
|
| 98 |
-
|
|
| 99 |
-
|
|
| 100 |
-
|
|
| 101 |
-
|
|
| 102 |
-
|
|
| 103 |
-
|
|
| 104 |
-
|
|
| 105 |
-
|
|
| 106 |
-
|
|
| 107 |
-
|
|
| 108 |
-
|
|
| 109 |
-
|
|
| 110 |
-
|
|
| 111 |
-
|
|
| 112 |
-
|
|
| 113 |
-
|
|
| 114 |
-
| 0.
|
| 115 |
-
|
|
| 116 |
-
|
|
| 117 |
-
|
|
| 118 |
-
|
|
| 119 |
-
|
|
| 120 |
-
|
|
| 121 |
-
|
|
| 122 |
-
| 0.
|
| 123 |
-
|
|
| 124 |
-
|
|
| 125 |
-
|
|
| 126 |
-
|
|
| 127 |
-
|
|
| 128 |
-
|
|
| 129 |
-
|
|
| 130 |
-
|
|
| 131 |
-
|
|
| 132 |
-
|
|
| 133 |
-
|
|
| 134 |
-
|
|
| 135 |
-
| 0.
|
| 136 |
-
| 0.
|
| 137 |
-
|
|
| 138 |
-
|
|
| 139 |
-
|
|
| 140 |
-
|
|
| 141 |
-
| 0.
|
| 142 |
-
| 0.
|
| 143 |
-
|
|
| 144 |
-
|
|
| 145 |
-
|
|
| 146 |
-
| 0.
|
| 147 |
-
|
|
| 148 |
-
| 0.
|
| 149 |
-
| 0.
|
| 150 |
-
| 0.
|
| 151 |
-
| 0.
|
| 152 |
-
|
|
| 153 |
-
| 0.
|
| 154 |
-
| 0.
|
| 155 |
-
| 0.
|
| 156 |
-
| 0.
|
| 157 |
-
| 0.
|
| 158 |
-
| 0.
|
| 159 |
-
| 0.
|
| 160 |
-
| 0.
|
| 161 |
-
| 0.
|
| 162 |
-
| 0.
|
| 163 |
-
|
|
| 164 |
-
| 0.
|
| 165 |
-
|
|
| 166 |
-
| 0.
|
| 167 |
-
| 0.
|
| 168 |
-
|
|
| 169 |
-
| 0.
|
| 170 |
-
| 0.
|
| 171 |
-
| 0.
|
| 172 |
-
|
|
| 173 |
-
| 0.
|
| 174 |
-
| 0.
|
| 175 |
-
| 0.
|
| 176 |
-
|
|
| 177 |
-
| 0.
|
| 178 |
-
|
|
| 179 |
-
| 0.
|
| 180 |
-
| 0.
|
| 181 |
-
| 0.
|
| 182 |
-
|
|
| 183 |
-
| 0.
|
| 184 |
-
| 0.
|
| 185 |
-
| 0.
|
| 186 |
-
|
|
| 187 |
-
| 0.
|
| 188 |
-
|
|
| 189 |
-
| 0.
|
| 190 |
-
| 0.
|
| 191 |
-
|
|
| 192 |
-
| 0.
|
| 193 |
-
| 0.
|
| 194 |
-
| 0.
|
| 195 |
-
| 0.
|
| 196 |
-
| 0.
|
| 197 |
-
| 0.
|
| 198 |
-
| 0.
|
| 199 |
-
|
|
| 200 |
-
| 0.
|
| 201 |
-
| 0.
|
| 202 |
-
|
|
| 203 |
-
| 0.
|
| 204 |
-
|
|
| 205 |
-
| 0.
|
| 206 |
-
| 0.
|
| 207 |
-
| 0.
|
| 208 |
-
| 0.
|
| 209 |
-
| 0.
|
| 210 |
-
| 0.
|
| 211 |
-
| 0.
|
| 212 |
-
| 0.
|
| 213 |
-
| 0.
|
| 214 |
-
| 0.
|
| 215 |
-
| 0.
|
| 216 |
-
| 0.
|
| 217 |
-
| 0.
|
| 218 |
-
| 0.
|
| 219 |
-
| 0.
|
| 220 |
-
| 0.
|
| 221 |
-
| 0.
|
| 222 |
-
| 0.
|
| 223 |
-
| 0.
|
| 224 |
-
| 0.
|
| 225 |
-
| 0.
|
| 226 |
-
|
|
| 227 |
-
| 0.
|
| 228 |
-
| 0.
|
| 229 |
-
| 0.
|
| 230 |
-
| 0.
|
| 231 |
-
| 0.
|
| 232 |
-
| 0.
|
| 233 |
-
| 0.
|
| 234 |
-
|
|
| 235 |
-
| 0.
|
| 236 |
-
| 0.
|
| 237 |
-
| 0.
|
| 238 |
-
| 0.
|
| 239 |
-
| 0.
|
| 240 |
-
| 0.
|
| 241 |
-
| 0.
|
| 242 |
-
| 0.
|
| 243 |
-
| 0.
|
| 244 |
-
| 0.
|
| 245 |
-
|
|
| 246 |
-
| 0.
|
| 247 |
-
| 0.
|
| 248 |
-
| 0.
|
| 249 |
-
| 0.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 250 |
|
| 251 |
|
| 252 |
### Framework versions
|
|
|
|
| 18 |
|
| 19 |
This model is a fine-tuned version of [Qwen/Qwen2.5-0.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-0.5B-Instruct) on an unknown dataset.
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.7387
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
|
|
|
| 37 |
### Training hyperparameters
|
| 38 |
|
| 39 |
The following hyperparameters were used during training:
|
| 40 |
+
- learning_rate: 4e-05
|
| 41 |
- train_batch_size: 4
|
| 42 |
- eval_batch_size: 4
|
| 43 |
- seed: 0
|
| 44 |
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
|
| 45 |
- lr_scheduler_type: cosine
|
| 46 |
- lr_scheduler_warmup_ratio: 0.01
|
| 47 |
+
- num_epochs: 32
|
| 48 |
|
| 49 |
### Training results
|
| 50 |
|
| 51 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
| 52 |
+
|:-------------:|:-------:|:----:|:---------------:|
|
| 53 |
+
| 3.1208 | 0.0405 | 3 | 3.2027 |
|
| 54 |
+
| 3.1051 | 0.0811 | 6 | 3.1913 |
|
| 55 |
+
| 3.1475 | 0.1216 | 9 | 3.1761 |
|
| 56 |
+
| 2.9958 | 0.1622 | 12 | 3.1458 |
|
| 57 |
+
| 2.8187 | 0.2027 | 15 | 3.1096 |
|
| 58 |
+
| 2.8758 | 0.2432 | 18 | 3.0628 |
|
| 59 |
+
| 2.8679 | 0.2838 | 21 | 3.0088 |
|
| 60 |
+
| 2.8846 | 0.3243 | 24 | 2.9546 |
|
| 61 |
+
| 2.7974 | 0.3649 | 27 | 2.8958 |
|
| 62 |
+
| 2.8269 | 0.4054 | 30 | 2.8389 |
|
| 63 |
+
| 2.7379 | 0.4459 | 33 | 2.7776 |
|
| 64 |
+
| 2.6123 | 0.4865 | 36 | 2.7106 |
|
| 65 |
+
| 2.5641 | 0.5270 | 39 | 2.6495 |
|
| 66 |
+
| 2.5012 | 0.5676 | 42 | 2.5868 |
|
| 67 |
+
| 2.4825 | 0.6081 | 45 | 2.5148 |
|
| 68 |
+
| 2.336 | 0.6486 | 48 | 2.4660 |
|
| 69 |
+
| 2.4145 | 0.6892 | 51 | 2.4062 |
|
| 70 |
+
| 2.3906 | 0.7297 | 54 | 2.3530 |
|
| 71 |
+
| 2.2689 | 0.7703 | 57 | 2.2990 |
|
| 72 |
+
| 2.2372 | 0.8108 | 60 | 2.2181 |
|
| 73 |
+
| 2.2054 | 0.8514 | 63 | 2.1602 |
|
| 74 |
+
| 2.1776 | 0.8919 | 66 | 2.0944 |
|
| 75 |
+
| 1.9452 | 0.9324 | 69 | 2.0230 |
|
| 76 |
+
| 2.0305 | 0.9730 | 72 | 1.9619 |
|
| 77 |
+
| 1.8456 | 1.0135 | 75 | 1.8966 |
|
| 78 |
+
| 1.8186 | 1.0541 | 78 | 1.8361 |
|
| 79 |
+
| 1.7897 | 1.0946 | 81 | 1.7808 |
|
| 80 |
+
| 1.7574 | 1.1351 | 84 | 1.7257 |
|
| 81 |
+
| 1.7225 | 1.1757 | 87 | 1.6673 |
|
| 82 |
+
| 1.5421 | 1.2162 | 90 | 1.6174 |
|
| 83 |
+
| 1.5612 | 1.2568 | 93 | 1.5772 |
|
| 84 |
+
| 1.5226 | 1.2973 | 96 | 1.5207 |
|
| 85 |
+
| 1.4257 | 1.3378 | 99 | 1.4864 |
|
| 86 |
+
| 1.5004 | 1.3784 | 102 | 1.4500 |
|
| 87 |
+
| 1.3826 | 1.4189 | 105 | 1.4225 |
|
| 88 |
+
| 1.4051 | 1.4595 | 108 | 1.3967 |
|
| 89 |
+
| 1.4726 | 1.5 | 111 | 1.3684 |
|
| 90 |
+
| 1.2813 | 1.5405 | 114 | 1.3477 |
|
| 91 |
+
| 1.3158 | 1.5811 | 117 | 1.3245 |
|
| 92 |
+
| 1.2704 | 1.6216 | 120 | 1.3132 |
|
| 93 |
+
| 1.2899 | 1.6622 | 123 | 1.2953 |
|
| 94 |
+
| 1.5335 | 1.7027 | 126 | 1.2824 |
|
| 95 |
+
| 1.2311 | 1.7432 | 129 | 1.2727 |
|
| 96 |
+
| 1.2331 | 1.7838 | 132 | 1.2568 |
|
| 97 |
+
| 1.2152 | 1.8243 | 135 | 1.2386 |
|
| 98 |
+
| 1.2849 | 1.8649 | 138 | 1.2285 |
|
| 99 |
+
| 1.1681 | 1.9054 | 141 | 1.2190 |
|
| 100 |
+
| 1.2107 | 1.9459 | 144 | 1.2137 |
|
| 101 |
+
| 1.122 | 1.9865 | 147 | 1.2030 |
|
| 102 |
+
| 1.2358 | 2.0270 | 150 | 1.1939 |
|
| 103 |
+
| 1.444 | 2.0676 | 153 | 1.1884 |
|
| 104 |
+
| 1.2996 | 2.1081 | 156 | 1.1801 |
|
| 105 |
+
| 1.1422 | 2.1486 | 159 | 1.1748 |
|
| 106 |
+
| 1.1755 | 2.1892 | 162 | 1.1672 |
|
| 107 |
+
| 1.0404 | 2.2297 | 165 | 1.1653 |
|
| 108 |
+
| 1.1468 | 2.2703 | 168 | 1.1548 |
|
| 109 |
+
| 1.1091 | 2.3108 | 171 | 1.1488 |
|
| 110 |
+
| 1.1161 | 2.3514 | 174 | 1.1533 |
|
| 111 |
+
| 1.0319 | 2.3919 | 177 | 1.1421 |
|
| 112 |
+
| 1.0984 | 2.4324 | 180 | 1.1380 |
|
| 113 |
+
| 1.1694 | 2.4730 | 183 | 1.1273 |
|
| 114 |
+
| 0.9938 | 2.5135 | 186 | 1.1238 |
|
| 115 |
+
| 1.1249 | 2.5541 | 189 | 1.1151 |
|
| 116 |
+
| 1.2545 | 2.5946 | 192 | 1.1133 |
|
| 117 |
+
| 1.1458 | 2.6351 | 195 | 1.1104 |
|
| 118 |
+
| 1.1756 | 2.6757 | 198 | 1.1031 |
|
| 119 |
+
| 1.0413 | 2.7162 | 201 | 1.0978 |
|
| 120 |
+
| 1.0521 | 2.7568 | 204 | 1.0915 |
|
| 121 |
+
| 1.0209 | 2.7973 | 207 | 1.0917 |
|
| 122 |
+
| 0.9818 | 2.8378 | 210 | 1.0888 |
|
| 123 |
+
| 1.1486 | 2.8784 | 213 | 1.0861 |
|
| 124 |
+
| 1.0615 | 2.9189 | 216 | 1.0838 |
|
| 125 |
+
| 1.0094 | 2.9595 | 219 | 1.0781 |
|
| 126 |
+
| 1.027 | 3.0 | 222 | 1.0788 |
|
| 127 |
+
| 1.0367 | 3.0405 | 225 | 1.0744 |
|
| 128 |
+
| 1.0732 | 3.0811 | 228 | 1.0692 |
|
| 129 |
+
| 1.2963 | 3.1216 | 231 | 1.0681 |
|
| 130 |
+
| 1.0756 | 3.1622 | 234 | 1.0666 |
|
| 131 |
+
| 1.281 | 3.2027 | 237 | 1.0622 |
|
| 132 |
+
| 1.0219 | 3.2432 | 240 | 1.0607 |
|
| 133 |
+
| 1.0859 | 3.2838 | 243 | 1.0617 |
|
| 134 |
+
| 1.0362 | 3.3243 | 246 | 1.0554 |
|
| 135 |
+
| 0.9374 | 3.3649 | 249 | 1.0506 |
|
| 136 |
+
| 0.9561 | 3.4054 | 252 | 1.0439 |
|
| 137 |
+
| 1.0258 | 3.4459 | 255 | 1.0428 |
|
| 138 |
+
| 1.0305 | 3.4865 | 258 | 1.0417 |
|
| 139 |
+
| 1.0936 | 3.5270 | 261 | 1.0401 |
|
| 140 |
+
| 1.0279 | 3.5676 | 264 | 1.0381 |
|
| 141 |
+
| 0.9374 | 3.6081 | 267 | 1.0337 |
|
| 142 |
+
| 0.9181 | 3.6486 | 270 | 1.0314 |
|
| 143 |
+
| 1.1563 | 3.6892 | 273 | 1.0308 |
|
| 144 |
+
| 1.0048 | 3.7297 | 276 | 1.0329 |
|
| 145 |
+
| 1.0085 | 3.7703 | 279 | 1.0318 |
|
| 146 |
+
| 0.9796 | 3.8108 | 282 | 1.0305 |
|
| 147 |
+
| 1.0081 | 3.8514 | 285 | 1.0257 |
|
| 148 |
+
| 0.9767 | 3.8919 | 288 | 1.0221 |
|
| 149 |
+
| 0.9264 | 3.9324 | 291 | 1.0196 |
|
| 150 |
+
| 0.8994 | 3.9730 | 294 | 1.0177 |
|
| 151 |
+
| 0.8931 | 4.0135 | 297 | 1.0175 |
|
| 152 |
+
| 1.0456 | 4.0541 | 300 | 1.0140 |
|
| 153 |
+
| 0.9918 | 4.0946 | 303 | 1.0127 |
|
| 154 |
+
| 0.9371 | 4.1351 | 306 | 1.0089 |
|
| 155 |
+
| 0.9584 | 4.1757 | 309 | 1.0143 |
|
| 156 |
+
| 0.869 | 4.2162 | 312 | 1.0034 |
|
| 157 |
+
| 0.9225 | 4.2568 | 315 | 0.9995 |
|
| 158 |
+
| 0.9197 | 4.2973 | 318 | 1.0017 |
|
| 159 |
+
| 0.9189 | 4.3378 | 321 | 0.9999 |
|
| 160 |
+
| 0.9804 | 4.3784 | 324 | 0.9972 |
|
| 161 |
+
| 0.875 | 4.4189 | 327 | 0.9972 |
|
| 162 |
+
| 0.9624 | 4.4595 | 330 | 0.9968 |
|
| 163 |
+
| 1.0302 | 4.5 | 333 | 0.9937 |
|
| 164 |
+
| 0.9522 | 4.5405 | 336 | 0.9913 |
|
| 165 |
+
| 1.1925 | 4.5811 | 339 | 0.9910 |
|
| 166 |
+
| 0.9591 | 4.6216 | 342 | 0.9883 |
|
| 167 |
+
| 0.9859 | 4.6622 | 345 | 0.9852 |
|
| 168 |
+
| 1.1778 | 4.7027 | 348 | 0.9862 |
|
| 169 |
+
| 0.9507 | 4.7432 | 351 | 0.9867 |
|
| 170 |
+
| 0.9281 | 4.7838 | 354 | 0.9822 |
|
| 171 |
+
| 0.8874 | 4.8243 | 357 | 0.9779 |
|
| 172 |
+
| 1.0883 | 4.8649 | 360 | 0.9748 |
|
| 173 |
+
| 0.8251 | 4.9054 | 363 | 0.9764 |
|
| 174 |
+
| 0.9055 | 4.9459 | 366 | 0.9728 |
|
| 175 |
+
| 0.8919 | 4.9865 | 369 | 0.9768 |
|
| 176 |
+
| 1.0278 | 5.0270 | 372 | 0.9694 |
|
| 177 |
+
| 0.8941 | 5.0676 | 375 | 0.9679 |
|
| 178 |
+
| 1.0327 | 5.1081 | 378 | 0.9690 |
|
| 179 |
+
| 0.9175 | 5.1486 | 381 | 0.9655 |
|
| 180 |
+
| 0.8924 | 5.1892 | 384 | 0.9606 |
|
| 181 |
+
| 0.8728 | 5.2297 | 387 | 0.9610 |
|
| 182 |
+
| 1.1136 | 5.2703 | 390 | 0.9620 |
|
| 183 |
+
| 0.9395 | 5.3108 | 393 | 0.9580 |
|
| 184 |
+
| 0.9548 | 5.3514 | 396 | 0.9563 |
|
| 185 |
+
| 0.8094 | 5.3919 | 399 | 0.9535 |
|
| 186 |
+
| 1.1105 | 5.4324 | 402 | 0.9547 |
|
| 187 |
+
| 0.7559 | 5.4730 | 405 | 0.9552 |
|
| 188 |
+
| 1.0868 | 5.5135 | 408 | 0.9564 |
|
| 189 |
+
| 0.8924 | 5.5541 | 411 | 0.9534 |
|
| 190 |
+
| 0.7555 | 5.5946 | 414 | 0.9531 |
|
| 191 |
+
| 1.1301 | 5.6351 | 417 | 0.9530 |
|
| 192 |
+
| 0.853 | 5.6757 | 420 | 0.9504 |
|
| 193 |
+
| 0.8389 | 5.7162 | 423 | 0.9505 |
|
| 194 |
+
| 0.9227 | 5.7568 | 426 | 0.9475 |
|
| 195 |
+
| 0.8284 | 5.7973 | 429 | 0.9471 |
|
| 196 |
+
| 0.8861 | 5.8378 | 432 | 0.9462 |
|
| 197 |
+
| 0.8685 | 5.8784 | 435 | 0.9439 |
|
| 198 |
+
| 0.9224 | 5.9189 | 438 | 0.9397 |
|
| 199 |
+
| 1.0131 | 5.9595 | 441 | 0.9377 |
|
| 200 |
+
| 0.8987 | 6.0 | 444 | 0.9383 |
|
| 201 |
+
| 0.8799 | 6.0405 | 447 | 0.9382 |
|
| 202 |
+
| 1.0713 | 6.0811 | 450 | 0.9383 |
|
| 203 |
+
| 0.8875 | 6.1216 | 453 | 0.9358 |
|
| 204 |
+
| 1.1155 | 6.1622 | 456 | 0.9374 |
|
| 205 |
+
| 0.9545 | 6.2027 | 459 | 0.9332 |
|
| 206 |
+
| 0.856 | 6.2432 | 462 | 0.9301 |
|
| 207 |
+
| 0.8397 | 6.2838 | 465 | 0.9293 |
|
| 208 |
+
| 0.8901 | 6.3243 | 468 | 0.9270 |
|
| 209 |
+
| 0.8582 | 6.3649 | 471 | 0.9296 |
|
| 210 |
+
| 0.8374 | 6.4054 | 474 | 0.9270 |
|
| 211 |
+
| 0.8233 | 6.4459 | 477 | 0.9228 |
|
| 212 |
+
| 0.8611 | 6.4865 | 480 | 0.9205 |
|
| 213 |
+
| 0.9882 | 6.5270 | 483 | 0.9222 |
|
| 214 |
+
| 0.878 | 6.5676 | 486 | 0.9231 |
|
| 215 |
+
| 0.8198 | 6.6081 | 489 | 0.9216 |
|
| 216 |
+
| 0.8485 | 6.6486 | 492 | 0.9220 |
|
| 217 |
+
| 0.7551 | 6.6892 | 495 | 0.9232 |
|
| 218 |
+
| 0.9539 | 6.7297 | 498 | 0.9212 |
|
| 219 |
+
| 0.9046 | 6.7703 | 501 | 0.9219 |
|
| 220 |
+
| 0.8941 | 6.8108 | 504 | 0.9213 |
|
| 221 |
+
| 0.8578 | 6.8514 | 507 | 0.9151 |
|
| 222 |
+
| 0.8106 | 6.8919 | 510 | 0.9157 |
|
| 223 |
+
| 0.7412 | 6.9324 | 513 | 0.9164 |
|
| 224 |
+
| 0.9782 | 6.9730 | 516 | 0.9154 |
|
| 225 |
+
| 0.8391 | 7.0135 | 519 | 0.9178 |
|
| 226 |
+
| 1.0248 | 7.0541 | 522 | 0.9198 |
|
| 227 |
+
| 0.7815 | 7.0946 | 525 | 0.9139 |
|
| 228 |
+
| 0.7832 | 7.1351 | 528 | 0.9081 |
|
| 229 |
+
| 0.7286 | 7.1757 | 531 | 0.9074 |
|
| 230 |
+
| 0.7792 | 7.2162 | 534 | 0.9132 |
|
| 231 |
+
| 0.8675 | 7.2568 | 537 | 0.9078 |
|
| 232 |
+
| 0.8072 | 7.2973 | 540 | 0.9065 |
|
| 233 |
+
| 0.9848 | 7.3378 | 543 | 0.9038 |
|
| 234 |
+
| 1.1688 | 7.3784 | 546 | 0.9072 |
|
| 235 |
+
| 0.7637 | 7.4189 | 549 | 0.9032 |
|
| 236 |
+
| 0.8637 | 7.4595 | 552 | 0.8992 |
|
| 237 |
+
| 0.873 | 7.5 | 555 | 0.9019 |
|
| 238 |
+
| 0.8521 | 7.5405 | 558 | 0.9017 |
|
| 239 |
+
| 0.8397 | 7.5811 | 561 | 0.9028 |
|
| 240 |
+
| 0.8654 | 7.6216 | 564 | 0.8940 |
|
| 241 |
+
| 0.7825 | 7.6622 | 567 | 0.8960 |
|
| 242 |
+
| 0.929 | 7.7027 | 570 | 0.8933 |
|
| 243 |
+
| 0.8989 | 7.7432 | 573 | 0.8914 |
|
| 244 |
+
| 0.757 | 7.7838 | 576 | 0.8910 |
|
| 245 |
+
| 1.0276 | 7.8243 | 579 | 0.8913 |
|
| 246 |
+
| 0.6477 | 7.8649 | 582 | 0.8910 |
|
| 247 |
+
| 0.7152 | 7.9054 | 585 | 0.8939 |
|
| 248 |
+
| 0.9926 | 7.9459 | 588 | 0.8940 |
|
| 249 |
+
| 0.9186 | 7.9865 | 591 | 0.8911 |
|
| 250 |
+
| 0.7522 | 8.0270 | 594 | 0.8887 |
|
| 251 |
+
| 0.741 | 8.0676 | 597 | 0.8860 |
|
| 252 |
+
| 0.704 | 8.1081 | 600 | 0.8850 |
|
| 253 |
+
| 0.6462 | 8.1486 | 603 | 0.8892 |
|
| 254 |
+
| 0.8187 | 8.1892 | 606 | 0.8826 |
|
| 255 |
+
| 0.8307 | 8.2297 | 609 | 0.8789 |
|
| 256 |
+
| 0.851 | 8.2703 | 612 | 0.8809 |
|
| 257 |
+
| 1.031 | 8.3108 | 615 | 0.8828 |
|
| 258 |
+
| 1.0176 | 8.3514 | 618 | 0.8813 |
|
| 259 |
+
| 0.7549 | 8.3919 | 621 | 0.8778 |
|
| 260 |
+
| 0.8189 | 8.4324 | 624 | 0.8763 |
|
| 261 |
+
| 1.1007 | 8.4730 | 627 | 0.8750 |
|
| 262 |
+
| 0.8216 | 8.5135 | 630 | 0.8762 |
|
| 263 |
+
| 0.715 | 8.5541 | 633 | 0.8743 |
|
| 264 |
+
| 0.8203 | 8.5946 | 636 | 0.8763 |
|
| 265 |
+
| 0.7866 | 8.6351 | 639 | 0.8748 |
|
| 266 |
+
| 0.9153 | 8.6757 | 642 | 0.8728 |
|
| 267 |
+
| 0.9688 | 8.7162 | 645 | 0.8685 |
|
| 268 |
+
| 0.6317 | 8.7568 | 648 | 0.8710 |
|
| 269 |
+
| 0.7267 | 8.7973 | 651 | 0.8707 |
|
| 270 |
+
| 0.8292 | 8.8378 | 654 | 0.8691 |
|
| 271 |
+
| 0.8516 | 8.8784 | 657 | 0.8648 |
|
| 272 |
+
| 0.7637 | 8.9189 | 660 | 0.8676 |
|
| 273 |
+
| 0.7873 | 8.9595 | 663 | 0.8703 |
|
| 274 |
+
| 0.9375 | 9.0 | 666 | 0.8711 |
|
| 275 |
+
| 0.7844 | 9.0405 | 669 | 0.8634 |
|
| 276 |
+
| 0.7969 | 9.0811 | 672 | 0.8639 |
|
| 277 |
+
| 1.0146 | 9.1216 | 675 | 0.8647 |
|
| 278 |
+
| 0.9386 | 9.1622 | 678 | 0.8636 |
|
| 279 |
+
| 0.7381 | 9.2027 | 681 | 0.8611 |
|
| 280 |
+
| 0.7708 | 9.2432 | 684 | 0.8619 |
|
| 281 |
+
| 0.8501 | 9.2838 | 687 | 0.8636 |
|
| 282 |
+
| 1.0895 | 9.3243 | 690 | 0.8624 |
|
| 283 |
+
| 0.7784 | 9.3649 | 693 | 0.8624 |
|
| 284 |
+
| 0.7794 | 9.4054 | 696 | 0.8637 |
|
| 285 |
+
| 0.6984 | 9.4459 | 699 | 0.8590 |
|
| 286 |
+
| 0.9442 | 9.4865 | 702 | 0.8558 |
|
| 287 |
+
| 0.8026 | 9.5270 | 705 | 0.8558 |
|
| 288 |
+
| 0.8767 | 9.5676 | 708 | 0.8536 |
|
| 289 |
+
| 0.7298 | 9.6081 | 711 | 0.8511 |
|
| 290 |
+
| 0.6928 | 9.6486 | 714 | 0.8524 |
|
| 291 |
+
| 0.7068 | 9.6892 | 717 | 0.8532 |
|
| 292 |
+
| 0.7159 | 9.7297 | 720 | 0.8539 |
|
| 293 |
+
| 0.627 | 9.7703 | 723 | 0.8534 |
|
| 294 |
+
| 0.7097 | 9.8108 | 726 | 0.8557 |
|
| 295 |
+
| 0.6333 | 9.8514 | 729 | 0.8602 |
|
| 296 |
+
| 0.858 | 9.8919 | 732 | 0.8563 |
|
| 297 |
+
| 0.8599 | 9.9324 | 735 | 0.8547 |
|
| 298 |
+
| 0.8943 | 9.9730 | 738 | 0.8533 |
|
| 299 |
+
| 0.6119 | 10.0135 | 741 | 0.8536 |
|
| 300 |
+
| 0.7747 | 10.0541 | 744 | 0.8496 |
|
| 301 |
+
| 0.6964 | 10.0946 | 747 | 0.8481 |
|
| 302 |
+
| 0.8405 | 10.1351 | 750 | 0.8477 |
|
| 303 |
+
| 0.7406 | 10.1757 | 753 | 0.8468 |
|
| 304 |
+
| 0.8602 | 10.2162 | 756 | 0.8464 |
|
| 305 |
+
| 0.7174 | 10.2568 | 759 | 0.8424 |
|
| 306 |
+
| 0.8197 | 10.2973 | 762 | 0.8421 |
|
| 307 |
+
| 0.9546 | 10.3378 | 765 | 0.8408 |
|
| 308 |
+
| 0.8244 | 10.3784 | 768 | 0.8390 |
|
| 309 |
+
| 0.8684 | 10.4189 | 771 | 0.8381 |
|
| 310 |
+
| 0.6375 | 10.4595 | 774 | 0.8381 |
|
| 311 |
+
| 1.0448 | 10.5 | 777 | 0.8379 |
|
| 312 |
+
| 0.7016 | 10.5405 | 780 | 0.8392 |
|
| 313 |
+
| 0.7816 | 10.5811 | 783 | 0.8380 |
|
| 314 |
+
| 0.7377 | 10.6216 | 786 | 0.8390 |
|
| 315 |
+
| 0.8123 | 10.6622 | 789 | 0.8397 |
|
| 316 |
+
| 0.7813 | 10.7027 | 792 | 0.8403 |
|
| 317 |
+
| 0.7057 | 10.7432 | 795 | 0.8377 |
|
| 318 |
+
| 0.8193 | 10.7838 | 798 | 0.8391 |
|
| 319 |
+
| 0.641 | 10.8243 | 801 | 0.8395 |
|
| 320 |
+
| 0.7123 | 10.8649 | 804 | 0.8363 |
|
| 321 |
+
| 0.895 | 10.9054 | 807 | 0.8356 |
|
| 322 |
+
| 0.7314 | 10.9459 | 810 | 0.8318 |
|
| 323 |
+
| 0.6816 | 10.9865 | 813 | 0.8328 |
|
| 324 |
+
| 0.7839 | 11.0270 | 816 | 0.8333 |
|
| 325 |
+
| 0.6903 | 11.0676 | 819 | 0.8403 |
|
| 326 |
+
| 0.8589 | 11.1081 | 822 | 0.8388 |
|
| 327 |
+
| 0.7616 | 11.1486 | 825 | 0.8345 |
|
| 328 |
+
| 0.8326 | 11.1892 | 828 | 0.8337 |
|
| 329 |
+
| 0.6695 | 11.2297 | 831 | 0.8311 |
|
| 330 |
+
| 0.6722 | 11.2703 | 834 | 0.8330 |
|
| 331 |
+
| 0.6469 | 11.3108 | 837 | 0.8343 |
|
| 332 |
+
| 0.7524 | 11.3514 | 840 | 0.8335 |
|
| 333 |
+
| 0.9929 | 11.3919 | 843 | 0.8263 |
|
| 334 |
+
| 0.9125 | 11.4324 | 846 | 0.8253 |
|
| 335 |
+
| 0.7264 | 11.4730 | 849 | 0.8281 |
|
| 336 |
+
| 0.7659 | 11.5135 | 852 | 0.8258 |
|
| 337 |
+
| 0.5996 | 11.5541 | 855 | 0.8266 |
|
| 338 |
+
| 0.9027 | 11.5946 | 858 | 0.8246 |
|
| 339 |
+
| 0.7925 | 11.6351 | 861 | 0.8287 |
|
| 340 |
+
| 0.7139 | 11.6757 | 864 | 0.8277 |
|
| 341 |
+
| 0.8313 | 11.7162 | 867 | 0.8277 |
|
| 342 |
+
| 0.8107 | 11.7568 | 870 | 0.8232 |
|
| 343 |
+
| 0.6551 | 11.7973 | 873 | 0.8238 |
|
| 344 |
+
| 0.7706 | 11.8378 | 876 | 0.8243 |
|
| 345 |
+
| 0.8224 | 11.8784 | 879 | 0.8218 |
|
| 346 |
+
| 0.5772 | 11.9189 | 882 | 0.8190 |
|
| 347 |
+
| 0.7463 | 11.9595 | 885 | 0.8195 |
|
| 348 |
+
| 0.7346 | 12.0 | 888 | 0.8215 |
|
| 349 |
+
| 0.7386 | 12.0405 | 891 | 0.8193 |
|
| 350 |
+
| 0.6977 | 12.0811 | 894 | 0.8177 |
|
| 351 |
+
| 0.7275 | 12.1216 | 897 | 0.8163 |
|
| 352 |
+
| 0.834 | 12.1622 | 900 | 0.8178 |
|
| 353 |
+
| 0.6453 | 12.2027 | 903 | 0.8172 |
|
| 354 |
+
| 0.6808 | 12.2432 | 906 | 0.8179 |
|
| 355 |
+
| 0.8235 | 12.2838 | 909 | 0.8196 |
|
| 356 |
+
| 0.6978 | 12.3243 | 912 | 0.8143 |
|
| 357 |
+
| 0.6775 | 12.3649 | 915 | 0.8141 |
|
| 358 |
+
| 0.7675 | 12.4054 | 918 | 0.8159 |
|
| 359 |
+
| 0.7184 | 12.4459 | 921 | 0.8149 |
|
| 360 |
+
| 0.8134 | 12.4865 | 924 | 0.8155 |
|
| 361 |
+
| 0.7627 | 12.5270 | 927 | 0.8142 |
|
| 362 |
+
| 0.6658 | 12.5676 | 930 | 0.8138 |
|
| 363 |
+
| 0.7203 | 12.6081 | 933 | 0.8094 |
|
| 364 |
+
| 0.8712 | 12.6486 | 936 | 0.8102 |
|
| 365 |
+
| 0.7376 | 12.6892 | 939 | 0.8110 |
|
| 366 |
+
| 0.6303 | 12.7297 | 942 | 0.8112 |
|
| 367 |
+
| 0.8266 | 12.7703 | 945 | 0.8112 |
|
| 368 |
+
| 0.6992 | 12.8108 | 948 | 0.8112 |
|
| 369 |
+
| 0.6973 | 12.8514 | 951 | 0.8204 |
|
| 370 |
+
| 0.7004 | 12.8919 | 954 | 0.8187 |
|
| 371 |
+
| 0.8472 | 12.9324 | 957 | 0.8104 |
|
| 372 |
+
| 0.6852 | 12.9730 | 960 | 0.8076 |
|
| 373 |
+
| 0.7859 | 13.0135 | 963 | 0.8034 |
|
| 374 |
+
| 0.748 | 13.0541 | 966 | 0.8005 |
|
| 375 |
+
| 0.647 | 13.0946 | 969 | 0.8017 |
|
| 376 |
+
| 0.5999 | 13.1351 | 972 | 0.8068 |
|
| 377 |
+
| 0.8251 | 13.1757 | 975 | 0.8085 |
|
| 378 |
+
| 0.751 | 13.2162 | 978 | 0.8019 |
|
| 379 |
+
| 0.8143 | 13.2568 | 981 | 0.8029 |
|
| 380 |
+
| 0.8602 | 13.2973 | 984 | 0.8021 |
|
| 381 |
+
| 0.7271 | 13.3378 | 987 | 0.8047 |
|
| 382 |
+
| 0.6583 | 13.3784 | 990 | 0.8032 |
|
| 383 |
+
| 0.7526 | 13.4189 | 993 | 0.8040 |
|
| 384 |
+
| 0.9622 | 13.4595 | 996 | 0.8047 |
|
| 385 |
+
| 0.7012 | 13.5 | 999 | 0.8016 |
|
| 386 |
+
| 0.7386 | 13.5405 | 1002 | 0.8023 |
|
| 387 |
+
| 0.7437 | 13.5811 | 1005 | 0.7979 |
|
| 388 |
+
| 0.7107 | 13.6216 | 1008 | 0.7971 |
|
| 389 |
+
| 0.5084 | 13.6622 | 1011 | 0.7955 |
|
| 390 |
+
| 0.7172 | 13.7027 | 1014 | 0.7958 |
|
| 391 |
+
| 0.6663 | 13.7432 | 1017 | 0.7957 |
|
| 392 |
+
| 0.6369 | 13.7838 | 1020 | 0.7937 |
|
| 393 |
+
| 0.6689 | 13.8243 | 1023 | 0.7912 |
|
| 394 |
+
| 0.7544 | 13.8649 | 1026 | 0.7928 |
|
| 395 |
+
| 0.7894 | 13.9054 | 1029 | 0.7913 |
|
| 396 |
+
| 0.7585 | 13.9459 | 1032 | 0.7911 |
|
| 397 |
+
| 0.7564 | 13.9865 | 1035 | 0.7915 |
|
| 398 |
+
| 0.7037 | 14.0270 | 1038 | 0.7942 |
|
| 399 |
+
| 0.6665 | 14.0676 | 1041 | 0.7924 |
|
| 400 |
+
| 0.8052 | 14.1081 | 1044 | 0.7961 |
|
| 401 |
+
| 0.7059 | 14.1486 | 1047 | 0.7966 |
|
| 402 |
+
| 0.9295 | 14.1892 | 1050 | 0.7922 |
|
| 403 |
+
| 0.7731 | 14.2297 | 1053 | 0.7905 |
|
| 404 |
+
| 0.7104 | 14.2703 | 1056 | 0.7953 |
|
| 405 |
+
| 0.7656 | 14.3108 | 1059 | 0.7908 |
|
| 406 |
+
| 0.6772 | 14.3514 | 1062 | 0.7932 |
|
| 407 |
+
| 0.564 | 14.3919 | 1065 | 0.7949 |
|
| 408 |
+
| 0.7368 | 14.4324 | 1068 | 0.7923 |
|
| 409 |
+
| 0.7671 | 14.4730 | 1071 | 0.7978 |
|
| 410 |
+
| 0.6184 | 14.5135 | 1074 | 0.7965 |
|
| 411 |
+
| 0.7065 | 14.5541 | 1077 | 0.7918 |
|
| 412 |
+
| 0.611 | 14.5946 | 1080 | 0.7901 |
|
| 413 |
+
| 0.5999 | 14.6351 | 1083 | 0.7906 |
|
| 414 |
+
| 0.7597 | 14.6757 | 1086 | 0.7887 |
|
| 415 |
+
| 0.5979 | 14.7162 | 1089 | 0.7868 |
|
| 416 |
+
| 0.6442 | 14.7568 | 1092 | 0.7880 |
|
| 417 |
+
| 0.7898 | 14.7973 | 1095 | 0.7863 |
|
| 418 |
+
| 0.7596 | 14.8378 | 1098 | 0.7869 |
|
| 419 |
+
| 0.7963 | 14.8784 | 1101 | 0.7873 |
|
| 420 |
+
| 0.7001 | 14.9189 | 1104 | 0.7863 |
|
| 421 |
+
| 0.613 | 14.9595 | 1107 | 0.7827 |
|
| 422 |
+
| 0.6938 | 15.0 | 1110 | 0.7820 |
|
| 423 |
+
| 0.5834 | 15.0405 | 1113 | 0.7846 |
|
| 424 |
+
| 0.8899 | 15.0811 | 1116 | 0.7872 |
|
| 425 |
+
| 0.7885 | 15.1216 | 1119 | 0.7838 |
|
| 426 |
+
| 0.7913 | 15.1622 | 1122 | 0.7838 |
|
| 427 |
+
| 0.6819 | 15.2027 | 1125 | 0.7847 |
|
| 428 |
+
| 0.627 | 15.2432 | 1128 | 0.7884 |
|
| 429 |
+
| 0.7899 | 15.2838 | 1131 | 0.7823 |
|
| 430 |
+
| 0.8729 | 15.3243 | 1134 | 0.7810 |
|
| 431 |
+
| 0.5565 | 15.3649 | 1137 | 0.7767 |
|
| 432 |
+
| 0.65 | 15.4054 | 1140 | 0.7798 |
|
| 433 |
+
| 0.8706 | 15.4459 | 1143 | 0.7759 |
|
| 434 |
+
| 0.6454 | 15.4865 | 1146 | 0.7752 |
|
| 435 |
+
| 0.4782 | 15.5270 | 1149 | 0.7768 |
|
| 436 |
+
| 0.5503 | 15.5676 | 1152 | 0.7770 |
|
| 437 |
+
| 0.6918 | 15.6081 | 1155 | 0.7785 |
|
| 438 |
+
| 0.6653 | 15.6486 | 1158 | 0.7786 |
|
| 439 |
+
| 0.9263 | 15.6892 | 1161 | 0.7767 |
|
| 440 |
+
| 0.555 | 15.7297 | 1164 | 0.7771 |
|
| 441 |
+
| 0.6206 | 15.7703 | 1167 | 0.7828 |
|
| 442 |
+
| 0.8373 | 15.8108 | 1170 | 0.7792 |
|
| 443 |
+
| 0.6903 | 15.8514 | 1173 | 0.7757 |
|
| 444 |
+
| 0.6952 | 15.8919 | 1176 | 0.7767 |
|
| 445 |
+
| 0.6991 | 15.9324 | 1179 | 0.7761 |
|
| 446 |
+
| 0.6659 | 15.9730 | 1182 | 0.7763 |
|
| 447 |
+
| 0.6286 | 16.0135 | 1185 | 0.7761 |
|
| 448 |
+
| 0.5775 | 16.0541 | 1188 | 0.7767 |
|
| 449 |
+
| 0.6576 | 16.0946 | 1191 | 0.7734 |
|
| 450 |
+
| 0.6216 | 16.1351 | 1194 | 0.7765 |
|
| 451 |
+
| 0.6129 | 16.1757 | 1197 | 0.7764 |
|
| 452 |
+
| 0.7025 | 16.2162 | 1200 | 0.7759 |
|
| 453 |
+
| 0.8607 | 16.2568 | 1203 | 0.7742 |
|
| 454 |
+
| 0.4131 | 16.2973 | 1206 | 0.7718 |
|
| 455 |
+
| 0.6074 | 16.3378 | 1209 | 0.7712 |
|
| 456 |
+
| 0.6138 | 16.3784 | 1212 | 0.7735 |
|
| 457 |
+
| 0.6341 | 16.4189 | 1215 | 0.7719 |
|
| 458 |
+
| 0.6118 | 16.4595 | 1218 | 0.7723 |
|
| 459 |
+
| 0.7456 | 16.5 | 1221 | 0.7731 |
|
| 460 |
+
| 0.7303 | 16.5405 | 1224 | 0.7744 |
|
| 461 |
+
| 0.9041 | 16.5811 | 1227 | 0.7728 |
|
| 462 |
+
| 0.6858 | 16.6216 | 1230 | 0.7696 |
|
| 463 |
+
| 0.7813 | 16.6622 | 1233 | 0.7709 |
|
| 464 |
+
| 0.6222 | 16.7027 | 1236 | 0.7718 |
|
| 465 |
+
| 0.756 | 16.7432 | 1239 | 0.7699 |
|
| 466 |
+
| 0.6971 | 16.7838 | 1242 | 0.7691 |
|
| 467 |
+
| 0.685 | 16.8243 | 1245 | 0.7671 |
|
| 468 |
+
| 0.8077 | 16.8649 | 1248 | 0.7700 |
|
| 469 |
+
| 0.5113 | 16.9054 | 1251 | 0.7725 |
|
| 470 |
+
| 0.863 | 16.9459 | 1254 | 0.7723 |
|
| 471 |
+
| 0.5907 | 16.9865 | 1257 | 0.7732 |
|
| 472 |
+
| 0.7326 | 17.0270 | 1260 | 0.7705 |
|
| 473 |
+
| 0.6305 | 17.0676 | 1263 | 0.7704 |
|
| 474 |
+
| 0.5333 | 17.1081 | 1266 | 0.7696 |
|
| 475 |
+
| 0.6263 | 17.1486 | 1269 | 0.7736 |
|
| 476 |
+
| 0.6031 | 17.1892 | 1272 | 0.7688 |
|
| 477 |
+
| 0.7785 | 17.2297 | 1275 | 0.7714 |
|
| 478 |
+
| 0.6821 | 17.2703 | 1278 | 0.7698 |
|
| 479 |
+
| 0.7371 | 17.3108 | 1281 | 0.7650 |
|
| 480 |
+
| 0.53 | 17.3514 | 1284 | 0.7681 |
|
| 481 |
+
| 0.7413 | 17.3919 | 1287 | 0.7674 |
|
| 482 |
+
| 0.5742 | 17.4324 | 1290 | 0.7650 |
|
| 483 |
+
| 0.5972 | 17.4730 | 1293 | 0.7647 |
|
| 484 |
+
| 0.6501 | 17.5135 | 1296 | 0.7638 |
|
| 485 |
+
| 0.7033 | 17.5541 | 1299 | 0.7623 |
|
| 486 |
+
| 0.9062 | 17.5946 | 1302 | 0.7605 |
|
| 487 |
+
| 0.6346 | 17.6351 | 1305 | 0.7623 |
|
| 488 |
+
| 0.6128 | 17.6757 | 1308 | 0.7610 |
|
| 489 |
+
| 0.5238 | 17.7162 | 1311 | 0.7649 |
|
| 490 |
+
| 0.7174 | 17.7568 | 1314 | 0.7638 |
|
| 491 |
+
| 0.5736 | 17.7973 | 1317 | 0.7613 |
|
| 492 |
+
| 0.6981 | 17.8378 | 1320 | 0.7663 |
|
| 493 |
+
| 0.7606 | 17.8784 | 1323 | 0.7639 |
|
| 494 |
+
| 0.8921 | 17.9189 | 1326 | 0.7627 |
|
| 495 |
+
| 0.8137 | 17.9595 | 1329 | 0.7596 |
|
| 496 |
+
| 0.5805 | 18.0 | 1332 | 0.7611 |
|
| 497 |
+
| 0.7555 | 18.0405 | 1335 | 0.7630 |
|
| 498 |
+
| 0.7198 | 18.0811 | 1338 | 0.7657 |
|
| 499 |
+
| 0.6985 | 18.1216 | 1341 | 0.7668 |
|
| 500 |
+
| 0.751 | 18.1622 | 1344 | 0.7647 |
|
| 501 |
+
| 0.6312 | 18.2027 | 1347 | 0.7666 |
|
| 502 |
+
| 0.5514 | 18.2432 | 1350 | 0.7633 |
|
| 503 |
+
| 0.7886 | 18.2838 | 1353 | 0.7598 |
|
| 504 |
+
| 0.6188 | 18.3243 | 1356 | 0.7591 |
|
| 505 |
+
| 0.9161 | 18.3649 | 1359 | 0.7609 |
|
| 506 |
+
| 0.7103 | 18.4054 | 1362 | 0.7597 |
|
| 507 |
+
| 0.5801 | 18.4459 | 1365 | 0.7600 |
|
| 508 |
+
| 0.5811 | 18.4865 | 1368 | 0.7568 |
|
| 509 |
+
| 0.7084 | 18.5270 | 1371 | 0.7583 |
|
| 510 |
+
| 0.6469 | 18.5676 | 1374 | 0.7567 |
|
| 511 |
+
| 0.4795 | 18.6081 | 1377 | 0.7577 |
|
| 512 |
+
| 0.5333 | 18.6486 | 1380 | 0.7608 |
|
| 513 |
+
| 0.6912 | 18.6892 | 1383 | 0.7599 |
|
| 514 |
+
| 0.5693 | 18.7297 | 1386 | 0.7605 |
|
| 515 |
+
| 0.765 | 18.7703 | 1389 | 0.7586 |
|
| 516 |
+
| 0.6908 | 18.8108 | 1392 | 0.7576 |
|
| 517 |
+
| 0.554 | 18.8514 | 1395 | 0.7588 |
|
| 518 |
+
| 0.5517 | 18.8919 | 1398 | 0.7593 |
|
| 519 |
+
| 0.6006 | 18.9324 | 1401 | 0.7624 |
|
| 520 |
+
| 0.8661 | 18.9730 | 1404 | 0.7610 |
|
| 521 |
+
| 0.5183 | 19.0135 | 1407 | 0.7586 |
|
| 522 |
+
| 0.7951 | 19.0541 | 1410 | 0.7585 |
|
| 523 |
+
| 0.6838 | 19.0946 | 1413 | 0.7561 |
|
| 524 |
+
| 0.7278 | 19.1351 | 1416 | 0.7545 |
|
| 525 |
+
| 0.6053 | 19.1757 | 1419 | 0.7541 |
|
| 526 |
+
| 0.5119 | 19.2162 | 1422 | 0.7537 |
|
| 527 |
+
| 0.5743 | 19.2568 | 1425 | 0.7566 |
|
| 528 |
+
| 0.8562 | 19.2973 | 1428 | 0.7532 |
|
| 529 |
+
| 0.4632 | 19.3378 | 1431 | 0.7539 |
|
| 530 |
+
| 0.8029 | 19.3784 | 1434 | 0.7568 |
|
| 531 |
+
| 0.8046 | 19.4189 | 1437 | 0.7553 |
|
| 532 |
+
| 0.6308 | 19.4595 | 1440 | 0.7547 |
|
| 533 |
+
| 0.5366 | 19.5 | 1443 | 0.7549 |
|
| 534 |
+
| 0.5249 | 19.5405 | 1446 | 0.7557 |
|
| 535 |
+
| 0.7144 | 19.5811 | 1449 | 0.7550 |
|
| 536 |
+
| 0.7054 | 19.6216 | 1452 | 0.7563 |
|
| 537 |
+
| 0.6575 | 19.6622 | 1455 | 0.7541 |
|
| 538 |
+
| 0.5057 | 19.7027 | 1458 | 0.7544 |
|
| 539 |
+
| 0.6339 | 19.7432 | 1461 | 0.7568 |
|
| 540 |
+
| 0.5965 | 19.7838 | 1464 | 0.7566 |
|
| 541 |
+
| 1.0225 | 19.8243 | 1467 | 0.7582 |
|
| 542 |
+
| 0.5466 | 19.8649 | 1470 | 0.7534 |
|
| 543 |
+
| 0.5946 | 19.9054 | 1473 | 0.7574 |
|
| 544 |
+
| 0.7015 | 19.9459 | 1476 | 0.7558 |
|
| 545 |
+
| 0.6772 | 19.9865 | 1479 | 0.7527 |
|
| 546 |
+
| 0.628 | 20.0270 | 1482 | 0.7520 |
|
| 547 |
+
| 0.8432 | 20.0676 | 1485 | 0.7546 |
|
| 548 |
+
| 0.5012 | 20.1081 | 1488 | 0.7524 |
|
| 549 |
+
| 0.7843 | 20.1486 | 1491 | 0.7560 |
|
| 550 |
+
| 0.6263 | 20.1892 | 1494 | 0.7533 |
|
| 551 |
+
| 0.5751 | 20.2297 | 1497 | 0.7530 |
|
| 552 |
+
| 0.5538 | 20.2703 | 1500 | 0.7574 |
|
| 553 |
+
| 0.4866 | 20.3108 | 1503 | 0.7526 |
|
| 554 |
+
| 0.5303 | 20.3514 | 1506 | 0.7558 |
|
| 555 |
+
| 0.8348 | 20.3919 | 1509 | 0.7519 |
|
| 556 |
+
| 0.6568 | 20.4324 | 1512 | 0.7540 |
|
| 557 |
+
| 0.587 | 20.4730 | 1515 | 0.7505 |
|
| 558 |
+
| 0.5837 | 20.5135 | 1518 | 0.7523 |
|
| 559 |
+
| 0.7225 | 20.5541 | 1521 | 0.7493 |
|
| 560 |
+
| 0.6323 | 20.5946 | 1524 | 0.7533 |
|
| 561 |
+
| 0.7393 | 20.6351 | 1527 | 0.7542 |
|
| 562 |
+
| 0.9057 | 20.6757 | 1530 | 0.7537 |
|
| 563 |
+
| 0.5939 | 20.7162 | 1533 | 0.7554 |
|
| 564 |
+
| 0.7613 | 20.7568 | 1536 | 0.7528 |
|
| 565 |
+
| 0.4678 | 20.7973 | 1539 | 0.7524 |
|
| 566 |
+
| 0.6316 | 20.8378 | 1542 | 0.7501 |
|
| 567 |
+
| 0.5362 | 20.8784 | 1545 | 0.7522 |
|
| 568 |
+
| 0.653 | 20.9189 | 1548 | 0.7495 |
|
| 569 |
+
| 0.8229 | 20.9595 | 1551 | 0.7499 |
|
| 570 |
+
| 0.5539 | 21.0 | 1554 | 0.7486 |
|
| 571 |
+
| 0.586 | 21.0405 | 1557 | 0.7498 |
|
| 572 |
+
| 0.5707 | 21.0811 | 1560 | 0.7505 |
|
| 573 |
+
| 0.6017 | 21.1216 | 1563 | 0.7497 |
|
| 574 |
+
| 0.5576 | 21.1622 | 1566 | 0.7489 |
|
| 575 |
+
| 0.7408 | 21.2027 | 1569 | 0.7503 |
|
| 576 |
+
| 0.542 | 21.2432 | 1572 | 0.7490 |
|
| 577 |
+
| 0.576 | 21.2838 | 1575 | 0.7478 |
|
| 578 |
+
| 0.5323 | 21.3243 | 1578 | 0.7525 |
|
| 579 |
+
| 0.6224 | 21.3649 | 1581 | 0.7499 |
|
| 580 |
+
| 0.5921 | 21.4054 | 1584 | 0.7521 |
|
| 581 |
+
| 0.5888 | 21.4459 | 1587 | 0.7480 |
|
| 582 |
+
| 0.493 | 21.4865 | 1590 | 0.7492 |
|
| 583 |
+
| 0.6878 | 21.5270 | 1593 | 0.7491 |
|
| 584 |
+
| 0.5939 | 21.5676 | 1596 | 0.7487 |
|
| 585 |
+
| 0.6568 | 21.6081 | 1599 | 0.7464 |
|
| 586 |
+
| 0.7762 | 21.6486 | 1602 | 0.7491 |
|
| 587 |
+
| 0.7653 | 21.6892 | 1605 | 0.7495 |
|
| 588 |
+
| 1.048 | 21.7297 | 1608 | 0.7442 |
|
| 589 |
+
| 0.4252 | 21.7703 | 1611 | 0.7448 |
|
| 590 |
+
| 0.7077 | 21.8108 | 1614 | 0.7462 |
|
| 591 |
+
| 0.7929 | 21.8514 | 1617 | 0.7441 |
|
| 592 |
+
| 0.7359 | 21.8919 | 1620 | 0.7431 |
|
| 593 |
+
| 0.579 | 21.9324 | 1623 | 0.7404 |
|
| 594 |
+
| 0.5986 | 21.9730 | 1626 | 0.7424 |
|
| 595 |
+
| 0.6294 | 22.0135 | 1629 | 0.7483 |
|
| 596 |
+
| 0.6028 | 22.0541 | 1632 | 0.7487 |
|
| 597 |
+
| 0.71 | 22.0946 | 1635 | 0.7486 |
|
| 598 |
+
| 0.7272 | 22.1351 | 1638 | 0.7474 |
|
| 599 |
+
| 0.7421 | 22.1757 | 1641 | 0.7484 |
|
| 600 |
+
| 0.6425 | 22.2162 | 1644 | 0.7485 |
|
| 601 |
+
| 0.7581 | 22.2568 | 1647 | 0.7454 |
|
| 602 |
+
| 0.6693 | 22.2973 | 1650 | 0.7461 |
|
| 603 |
+
| 0.7514 | 22.3378 | 1653 | 0.7439 |
|
| 604 |
+
| 0.4701 | 22.3784 | 1656 | 0.7446 |
|
| 605 |
+
| 0.5123 | 22.4189 | 1659 | 0.7474 |
|
| 606 |
+
| 0.5777 | 22.4595 | 1662 | 0.7448 |
|
| 607 |
+
| 0.611 | 22.5 | 1665 | 0.7442 |
|
| 608 |
+
| 0.5888 | 22.5405 | 1668 | 0.7456 |
|
| 609 |
+
| 0.5861 | 22.5811 | 1671 | 0.7462 |
|
| 610 |
+
| 0.7648 | 22.6216 | 1674 | 0.7442 |
|
| 611 |
+
| 0.5827 | 22.6622 | 1677 | 0.7464 |
|
| 612 |
+
| 0.7285 | 22.7027 | 1680 | 0.7477 |
|
| 613 |
+
| 0.5688 | 22.7432 | 1683 | 0.7465 |
|
| 614 |
+
| 0.65 | 22.7838 | 1686 | 0.7466 |
|
| 615 |
+
| 0.4758 | 22.8243 | 1689 | 0.7461 |
|
| 616 |
+
| 0.6542 | 22.8649 | 1692 | 0.7446 |
|
| 617 |
+
| 0.5368 | 22.9054 | 1695 | 0.7479 |
|
| 618 |
+
| 0.8234 | 22.9459 | 1698 | 0.7471 |
|
| 619 |
+
| 0.5605 | 22.9865 | 1701 | 0.7457 |
|
| 620 |
+
| 0.6373 | 23.0270 | 1704 | 0.7468 |
|
| 621 |
+
| 0.5194 | 23.0676 | 1707 | 0.7462 |
|
| 622 |
+
| 0.7406 | 23.1081 | 1710 | 0.7447 |
|
| 623 |
+
| 0.5238 | 23.1486 | 1713 | 0.7434 |
|
| 624 |
+
| 0.6407 | 23.1892 | 1716 | 0.7431 |
|
| 625 |
+
| 0.4933 | 23.2297 | 1719 | 0.7436 |
|
| 626 |
+
| 0.8002 | 23.2703 | 1722 | 0.7450 |
|
| 627 |
+
| 0.6489 | 23.3108 | 1725 | 0.7458 |
|
| 628 |
+
| 0.6325 | 23.3514 | 1728 | 0.7433 |
|
| 629 |
+
| 0.5101 | 23.3919 | 1731 | 0.7440 |
|
| 630 |
+
| 0.6503 | 23.4324 | 1734 | 0.7442 |
|
| 631 |
+
| 0.8555 | 23.4730 | 1737 | 0.7455 |
|
| 632 |
+
| 0.4222 | 23.5135 | 1740 | 0.7447 |
|
| 633 |
+
| 0.843 | 23.5541 | 1743 | 0.7446 |
|
| 634 |
+
| 0.8313 | 23.5946 | 1746 | 0.7441 |
|
| 635 |
+
| 0.7178 | 23.6351 | 1749 | 0.7439 |
|
| 636 |
+
| 0.6409 | 23.6757 | 1752 | 0.7431 |
|
| 637 |
+
| 0.6184 | 23.7162 | 1755 | 0.7437 |
|
| 638 |
+
| 0.5277 | 23.7568 | 1758 | 0.7432 |
|
| 639 |
+
| 0.5987 | 23.7973 | 1761 | 0.7445 |
|
| 640 |
+
| 0.7011 | 23.8378 | 1764 | 0.7407 |
|
| 641 |
+
| 0.6693 | 23.8784 | 1767 | 0.7429 |
|
| 642 |
+
| 0.6101 | 23.9189 | 1770 | 0.7423 |
|
| 643 |
+
| 0.5883 | 23.9595 | 1773 | 0.7437 |
|
| 644 |
+
| 0.4803 | 24.0 | 1776 | 0.7473 |
|
| 645 |
+
| 0.6667 | 24.0405 | 1779 | 0.7445 |
|
| 646 |
+
| 0.6156 | 24.0811 | 1782 | 0.7445 |
|
| 647 |
+
| 0.7909 | 24.1216 | 1785 | 0.7416 |
|
| 648 |
+
| 0.449 | 24.1622 | 1788 | 0.7397 |
|
| 649 |
+
| 0.6064 | 24.2027 | 1791 | 0.7412 |
|
| 650 |
+
| 0.5522 | 24.2432 | 1794 | 0.7404 |
|
| 651 |
+
| 0.5244 | 24.2838 | 1797 | 0.7440 |
|
| 652 |
+
| 0.6211 | 24.3243 | 1800 | 0.7421 |
|
| 653 |
+
| 0.572 | 24.3649 | 1803 | 0.7422 |
|
| 654 |
+
| 0.7 | 24.4054 | 1806 | 0.7425 |
|
| 655 |
+
| 0.6095 | 24.4459 | 1809 | 0.7447 |
|
| 656 |
+
| 0.9109 | 24.4865 | 1812 | 0.7447 |
|
| 657 |
+
| 0.5852 | 24.5270 | 1815 | 0.7445 |
|
| 658 |
+
| 0.8397 | 24.5676 | 1818 | 0.7408 |
|
| 659 |
+
| 0.6186 | 24.6081 | 1821 | 0.7393 |
|
| 660 |
+
| 0.7605 | 24.6486 | 1824 | 0.7419 |
|
| 661 |
+
| 0.4131 | 24.6892 | 1827 | 0.7393 |
|
| 662 |
+
| 0.8149 | 24.7297 | 1830 | 0.7434 |
|
| 663 |
+
| 0.5877 | 24.7703 | 1833 | 0.7435 |
|
| 664 |
+
| 0.5594 | 24.8108 | 1836 | 0.7420 |
|
| 665 |
+
| 0.6607 | 24.8514 | 1839 | 0.7433 |
|
| 666 |
+
| 0.5369 | 24.8919 | 1842 | 0.7437 |
|
| 667 |
+
| 0.5268 | 24.9324 | 1845 | 0.7410 |
|
| 668 |
+
| 0.5273 | 24.9730 | 1848 | 0.7440 |
|
| 669 |
+
| 0.6146 | 25.0135 | 1851 | 0.7427 |
|
| 670 |
+
| 0.8054 | 25.0541 | 1854 | 0.7440 |
|
| 671 |
+
| 0.8785 | 25.0946 | 1857 | 0.7451 |
|
| 672 |
+
| 0.7208 | 25.1351 | 1860 | 0.7450 |
|
| 673 |
+
| 0.6592 | 25.1757 | 1863 | 0.7434 |
|
| 674 |
+
| 0.7174 | 25.2162 | 1866 | 0.7430 |
|
| 675 |
+
| 0.5552 | 25.2568 | 1869 | 0.7398 |
|
| 676 |
+
| 0.7374 | 25.2973 | 1872 | 0.7411 |
|
| 677 |
+
| 0.6694 | 25.3378 | 1875 | 0.7400 |
|
| 678 |
+
| 0.5145 | 25.3784 | 1878 | 0.7406 |
|
| 679 |
+
| 0.6904 | 25.4189 | 1881 | 0.7405 |
|
| 680 |
+
| 0.6441 | 25.4595 | 1884 | 0.7425 |
|
| 681 |
+
| 0.4699 | 25.5 | 1887 | 0.7410 |
|
| 682 |
+
| 0.6629 | 25.5405 | 1890 | 0.7400 |
|
| 683 |
+
| 0.5895 | 25.5811 | 1893 | 0.7434 |
|
| 684 |
+
| 0.6364 | 25.6216 | 1896 | 0.7407 |
|
| 685 |
+
| 0.7567 | 25.6622 | 1899 | 0.7440 |
|
| 686 |
+
| 0.7204 | 25.7027 | 1902 | 0.7409 |
|
| 687 |
+
| 0.5391 | 25.7432 | 1905 | 0.7427 |
|
| 688 |
+
| 0.5649 | 25.7838 | 1908 | 0.7411 |
|
| 689 |
+
| 0.7254 | 25.8243 | 1911 | 0.7401 |
|
| 690 |
+
| 0.5305 | 25.8649 | 1914 | 0.7390 |
|
| 691 |
+
| 0.523 | 25.9054 | 1917 | 0.7389 |
|
| 692 |
+
| 0.4508 | 25.9459 | 1920 | 0.7394 |
|
| 693 |
+
| 0.4191 | 25.9865 | 1923 | 0.7414 |
|
| 694 |
+
| 0.5966 | 26.0270 | 1926 | 0.7438 |
|
| 695 |
+
| 0.6259 | 26.0676 | 1929 | 0.7404 |
|
| 696 |
+
| 0.5717 | 26.1081 | 1932 | 0.7429 |
|
| 697 |
+
| 0.5566 | 26.1486 | 1935 | 0.7413 |
|
| 698 |
+
| 0.612 | 26.1892 | 1938 | 0.7405 |
|
| 699 |
+
| 0.7154 | 26.2297 | 1941 | 0.7419 |
|
| 700 |
+
| 0.7489 | 26.2703 | 1944 | 0.7417 |
|
| 701 |
+
| 0.5922 | 26.3108 | 1947 | 0.7421 |
|
| 702 |
+
| 0.6247 | 26.3514 | 1950 | 0.7420 |
|
| 703 |
+
| 0.5243 | 26.3919 | 1953 | 0.7407 |
|
| 704 |
+
| 1.0104 | 26.4324 | 1956 | 0.7427 |
|
| 705 |
+
| 0.516 | 26.4730 | 1959 | 0.7419 |
|
| 706 |
+
| 0.5302 | 26.5135 | 1962 | 0.7393 |
|
| 707 |
+
| 0.742 | 26.5541 | 1965 | 0.7415 |
|
| 708 |
+
| 0.5358 | 26.5946 | 1968 | 0.7421 |
|
| 709 |
+
| 0.4858 | 26.6351 | 1971 | 0.7450 |
|
| 710 |
+
| 0.6545 | 26.6757 | 1974 | 0.7425 |
|
| 711 |
+
| 0.6161 | 26.7162 | 1977 | 0.7417 |
|
| 712 |
+
| 0.7481 | 26.7568 | 1980 | 0.7404 |
|
| 713 |
+
| 0.5783 | 26.7973 | 1983 | 0.7436 |
|
| 714 |
+
| 0.457 | 26.8378 | 1986 | 0.7422 |
|
| 715 |
+
| 0.5594 | 26.8784 | 1989 | 0.7424 |
|
| 716 |
+
| 0.609 | 26.9189 | 1992 | 0.7435 |
|
| 717 |
+
| 0.6018 | 26.9595 | 1995 | 0.7434 |
|
| 718 |
+
| 0.7986 | 27.0 | 1998 | 0.7404 |
|
| 719 |
+
| 0.7093 | 27.0405 | 2001 | 0.7419 |
|
| 720 |
+
| 0.8197 | 27.0811 | 2004 | 0.7427 |
|
| 721 |
+
| 0.5748 | 27.1216 | 2007 | 0.7434 |
|
| 722 |
+
| 0.446 | 27.1622 | 2010 | 0.7386 |
|
| 723 |
+
| 0.5958 | 27.2027 | 2013 | 0.7387 |
|
| 724 |
+
| 0.7049 | 27.2432 | 2016 | 0.7400 |
|
| 725 |
+
| 0.6765 | 27.2838 | 2019 | 0.7419 |
|
| 726 |
+
| 0.5303 | 27.3243 | 2022 | 0.7398 |
|
| 727 |
+
| 0.7938 | 27.3649 | 2025 | 0.7409 |
|
| 728 |
+
| 0.5543 | 27.4054 | 2028 | 0.7388 |
|
| 729 |
+
| 0.9195 | 27.4459 | 2031 | 0.7379 |
|
| 730 |
+
| 0.6682 | 27.4865 | 2034 | 0.7404 |
|
| 731 |
+
| 0.4695 | 27.5270 | 2037 | 0.7394 |
|
| 732 |
+
| 0.5242 | 27.5676 | 2040 | 0.7410 |
|
| 733 |
+
| 0.7289 | 27.6081 | 2043 | 0.7399 |
|
| 734 |
+
| 0.5307 | 27.6486 | 2046 | 0.7396 |
|
| 735 |
+
| 0.5228 | 27.6892 | 2049 | 0.7408 |
|
| 736 |
+
| 0.5756 | 27.7297 | 2052 | 0.7400 |
|
| 737 |
+
| 0.5873 | 27.7703 | 2055 | 0.7404 |
|
| 738 |
+
| 0.5304 | 27.8108 | 2058 | 0.7376 |
|
| 739 |
+
| 0.6231 | 27.8514 | 2061 | 0.7370 |
|
| 740 |
+
| 0.6331 | 27.8919 | 2064 | 0.7389 |
|
| 741 |
+
| 0.6596 | 27.9324 | 2067 | 0.7410 |
|
| 742 |
+
| 0.6554 | 27.9730 | 2070 | 0.7415 |
|
| 743 |
+
| 0.4172 | 28.0135 | 2073 | 0.7400 |
|
| 744 |
+
| 0.8645 | 28.0541 | 2076 | 0.7394 |
|
| 745 |
+
| 0.5343 | 28.0946 | 2079 | 0.7404 |
|
| 746 |
+
| 0.6244 | 28.1351 | 2082 | 0.7403 |
|
| 747 |
+
| 0.7855 | 28.1757 | 2085 | 0.7382 |
|
| 748 |
+
| 0.5522 | 28.2162 | 2088 | 0.7408 |
|
| 749 |
+
| 0.5947 | 28.2568 | 2091 | 0.7402 |
|
| 750 |
+
| 0.7008 | 28.2973 | 2094 | 0.7388 |
|
| 751 |
+
| 0.7432 | 28.3378 | 2097 | 0.7404 |
|
| 752 |
+
| 0.3971 | 28.3784 | 2100 | 0.7404 |
|
| 753 |
+
| 0.5977 | 28.4189 | 2103 | 0.7413 |
|
| 754 |
+
| 0.6171 | 28.4595 | 2106 | 0.7421 |
|
| 755 |
+
| 0.63 | 28.5 | 2109 | 0.7418 |
|
| 756 |
+
| 0.6669 | 28.5405 | 2112 | 0.7402 |
|
| 757 |
+
| 0.739 | 28.5811 | 2115 | 0.7413 |
|
| 758 |
+
| 0.7795 | 28.6216 | 2118 | 0.7426 |
|
| 759 |
+
| 0.49 | 28.6622 | 2121 | 0.7389 |
|
| 760 |
+
| 0.5843 | 28.7027 | 2124 | 0.7412 |
|
| 761 |
+
| 0.3649 | 28.7432 | 2127 | 0.7419 |
|
| 762 |
+
| 0.5686 | 28.7838 | 2130 | 0.7379 |
|
| 763 |
+
| 0.8602 | 28.8243 | 2133 | 0.7396 |
|
| 764 |
+
| 0.6258 | 28.8649 | 2136 | 0.7380 |
|
| 765 |
+
| 0.5918 | 28.9054 | 2139 | 0.7397 |
|
| 766 |
+
| 0.5472 | 28.9459 | 2142 | 0.7382 |
|
| 767 |
+
| 0.5999 | 28.9865 | 2145 | 0.7389 |
|
| 768 |
+
| 0.637 | 29.0270 | 2148 | 0.7395 |
|
| 769 |
+
| 0.7309 | 29.0676 | 2151 | 0.7386 |
|
| 770 |
+
| 0.5701 | 29.1081 | 2154 | 0.7396 |
|
| 771 |
+
| 0.5048 | 29.1486 | 2157 | 0.7398 |
|
| 772 |
+
| 0.4754 | 29.1892 | 2160 | 0.7370 |
|
| 773 |
+
| 0.5792 | 29.2297 | 2163 | 0.7381 |
|
| 774 |
+
| 0.7897 | 29.2703 | 2166 | 0.7387 |
|
| 775 |
+
| 0.6935 | 29.3108 | 2169 | 0.7407 |
|
| 776 |
+
| 0.7357 | 29.3514 | 2172 | 0.7412 |
|
| 777 |
+
| 0.5114 | 29.3919 | 2175 | 0.7389 |
|
| 778 |
+
| 0.7875 | 29.4324 | 2178 | 0.7399 |
|
| 779 |
+
| 0.5673 | 29.4730 | 2181 | 0.7385 |
|
| 780 |
+
| 0.6423 | 29.5135 | 2184 | 0.7382 |
|
| 781 |
+
| 0.7393 | 29.5541 | 2187 | 0.7396 |
|
| 782 |
+
| 0.5183 | 29.5946 | 2190 | 0.7395 |
|
| 783 |
+
| 0.5972 | 29.6351 | 2193 | 0.7389 |
|
| 784 |
+
| 0.6461 | 29.6757 | 2196 | 0.7398 |
|
| 785 |
+
| 0.427 | 29.7162 | 2199 | 0.7396 |
|
| 786 |
+
| 0.7357 | 29.7568 | 2202 | 0.7381 |
|
| 787 |
+
| 0.603 | 29.7973 | 2205 | 0.7394 |
|
| 788 |
+
| 0.6142 | 29.8378 | 2208 | 0.7406 |
|
| 789 |
+
| 0.4877 | 29.8784 | 2211 | 0.7408 |
|
| 790 |
+
| 0.7774 | 29.9189 | 2214 | 0.7401 |
|
| 791 |
+
| 0.6909 | 29.9595 | 2217 | 0.7413 |
|
| 792 |
+
| 0.5127 | 30.0 | 2220 | 0.7398 |
|
| 793 |
+
| 0.4667 | 30.0405 | 2223 | 0.7414 |
|
| 794 |
+
| 0.5583 | 30.0811 | 2226 | 0.7412 |
|
| 795 |
+
| 0.6727 | 30.1216 | 2229 | 0.7378 |
|
| 796 |
+
| 0.5809 | 30.1622 | 2232 | 0.7419 |
|
| 797 |
+
| 0.6331 | 30.2027 | 2235 | 0.7411 |
|
| 798 |
+
| 0.5906 | 30.2432 | 2238 | 0.7416 |
|
| 799 |
+
| 0.7849 | 30.2838 | 2241 | 0.7441 |
|
| 800 |
+
| 0.5856 | 30.3243 | 2244 | 0.7413 |
|
| 801 |
+
| 0.5886 | 30.3649 | 2247 | 0.7389 |
|
| 802 |
+
| 0.5763 | 30.4054 | 2250 | 0.7402 |
|
| 803 |
+
| 0.5239 | 30.4459 | 2253 | 0.7397 |
|
| 804 |
+
| 0.614 | 30.4865 | 2256 | 0.7412 |
|
| 805 |
+
| 0.6579 | 30.5270 | 2259 | 0.7418 |
|
| 806 |
+
| 0.7466 | 30.5676 | 2262 | 0.7408 |
|
| 807 |
+
| 0.7607 | 30.6081 | 2265 | 0.7414 |
|
| 808 |
+
| 0.6251 | 30.6486 | 2268 | 0.7414 |
|
| 809 |
+
| 0.9175 | 30.6892 | 2271 | 0.7396 |
|
| 810 |
+
| 0.5611 | 30.7297 | 2274 | 0.7390 |
|
| 811 |
+
| 0.7051 | 30.7703 | 2277 | 0.7403 |
|
| 812 |
+
| 0.6331 | 30.8108 | 2280 | 0.7381 |
|
| 813 |
+
| 0.5536 | 30.8514 | 2283 | 0.7380 |
|
| 814 |
+
| 0.6293 | 30.8919 | 2286 | 0.7393 |
|
| 815 |
+
| 0.4456 | 30.9324 | 2289 | 0.7400 |
|
| 816 |
+
| 0.5521 | 30.9730 | 2292 | 0.7393 |
|
| 817 |
+
| 0.5012 | 31.0135 | 2295 | 0.7403 |
|
| 818 |
+
| 0.5571 | 31.0541 | 2298 | 0.7413 |
|
| 819 |
+
| 0.6326 | 31.0946 | 2301 | 0.7420 |
|
| 820 |
+
| 0.7587 | 31.1351 | 2304 | 0.7394 |
|
| 821 |
+
| 0.6534 | 31.1757 | 2307 | 0.7400 |
|
| 822 |
+
| 0.5276 | 31.2162 | 2310 | 0.7402 |
|
| 823 |
+
| 0.6514 | 31.2568 | 2313 | 0.7392 |
|
| 824 |
+
| 0.8247 | 31.2973 | 2316 | 0.7392 |
|
| 825 |
+
| 0.7972 | 31.3378 | 2319 | 0.7409 |
|
| 826 |
+
| 0.4689 | 31.3784 | 2322 | 0.7420 |
|
| 827 |
+
| 0.5428 | 31.4189 | 2325 | 0.7398 |
|
| 828 |
+
| 0.7894 | 31.4595 | 2328 | 0.7409 |
|
| 829 |
+
| 0.5278 | 31.5 | 2331 | 0.7411 |
|
| 830 |
+
| 0.5384 | 31.5405 | 2334 | 0.7419 |
|
| 831 |
+
| 0.5847 | 31.5811 | 2337 | 0.7414 |
|
| 832 |
+
| 0.5634 | 31.6216 | 2340 | 0.7415 |
|
| 833 |
+
| 0.6 | 31.6622 | 2343 | 0.7418 |
|
| 834 |
+
| 0.7443 | 31.7027 | 2346 | 0.7425 |
|
| 835 |
+
| 0.5771 | 31.7432 | 2349 | 0.7392 |
|
| 836 |
+
| 0.6404 | 31.7838 | 2352 | 0.7381 |
|
| 837 |
+
| 0.5441 | 31.8243 | 2355 | 0.7406 |
|
| 838 |
+
| 0.5123 | 31.8649 | 2358 | 0.7375 |
|
| 839 |
+
| 0.7296 | 31.9054 | 2361 | 0.7383 |
|
| 840 |
+
| 0.4944 | 31.9459 | 2364 | 0.7399 |
|
| 841 |
+
| 0.7002 | 31.9865 | 2367 | 0.7387 |
|
| 842 |
|
| 843 |
|
| 844 |
### Framework versions
|
adapter_config.json
CHANGED
|
@@ -11,7 +11,7 @@
|
|
| 11 |
"layers_to_transform": null,
|
| 12 |
"loftq_config": {},
|
| 13 |
"lora_alpha": 32,
|
| 14 |
-
"lora_dropout": 0.
|
| 15 |
"megatron_config": null,
|
| 16 |
"megatron_core": "megatron.core",
|
| 17 |
"modules_to_save": null,
|
|
@@ -20,13 +20,8 @@
|
|
| 20 |
"rank_pattern": {},
|
| 21 |
"revision": null,
|
| 22 |
"target_modules": [
|
| 23 |
-
"down_proj",
|
| 24 |
-
"gate_proj",
|
| 25 |
-
"o_proj",
|
| 26 |
-
"up_proj",
|
| 27 |
"q_proj",
|
| 28 |
-
"v_proj"
|
| 29 |
-
"k_proj"
|
| 30 |
],
|
| 31 |
"task_type": "CAUSAL_LM",
|
| 32 |
"use_dora": false,
|
|
|
|
| 11 |
"layers_to_transform": null,
|
| 12 |
"loftq_config": {},
|
| 13 |
"lora_alpha": 32,
|
| 14 |
+
"lora_dropout": 0.01,
|
| 15 |
"megatron_config": null,
|
| 16 |
"megatron_core": "megatron.core",
|
| 17 |
"modules_to_save": null,
|
|
|
|
| 20 |
"rank_pattern": {},
|
| 21 |
"revision": null,
|
| 22 |
"target_modules": [
|
|
|
|
|
|
|
|
|
|
|
|
|
| 23 |
"q_proj",
|
| 24 |
+
"v_proj"
|
|
|
|
| 25 |
],
|
| 26 |
"task_type": "CAUSAL_LM",
|
| 27 |
"use_dora": false,
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:476a889673f1c2afe73af44a62f4f650acafa30b0fadec371509f3c66d8a583b
|
| 3 |
+
size 4338000
|
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 8.0,
|
| 3 |
+
"eval_loss": 0.7665330171585083,
|
| 4 |
+
"eval_runtime": 1.0749,
|
| 5 |
+
"eval_samples": 2,
|
| 6 |
+
"eval_samples_per_second": 14.885,
|
| 7 |
+
"eval_steps_per_second": 3.721,
|
| 8 |
+
"total_flos": 1188976147968000.0,
|
| 9 |
+
"train_loss": 0.4161318518926163,
|
| 10 |
+
"train_runtime": 741.424,
|
| 11 |
+
"train_samples_per_second": 3.194,
|
| 12 |
+
"train_steps_per_second": 0.798
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 8.0,
|
| 3 |
+
"eval_loss": 0.7665330171585083,
|
| 4 |
+
"eval_runtime": 1.0749,
|
| 5 |
+
"eval_samples": 2,
|
| 6 |
+
"eval_samples_per_second": 14.885,
|
| 7 |
+
"eval_steps_per_second": 3.721
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 8.0,
|
| 3 |
+
"total_flos": 1188976147968000.0,
|
| 4 |
+
"train_loss": 0.4161318518926163,
|
| 5 |
+
"train_runtime": 741.424,
|
| 6 |
+
"train_samples_per_second": 3.194,
|
| 7 |
+
"train_steps_per_second": 0.798
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2997 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 8.0,
|
| 5 |
+
"eval_steps": 3,
|
| 6 |
+
"global_step": 592,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.04054054054054054,
|
| 13 |
+
"grad_norm": 5.807250022888184,
|
| 14 |
+
"learning_rate": 5e-05,
|
| 15 |
+
"loss": 3.1119,
|
| 16 |
+
"step": 3
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.04054054054054054,
|
| 20 |
+
"eval_loss": 3.1016640663146973,
|
| 21 |
+
"eval_runtime": 1.0551,
|
| 22 |
+
"eval_samples_per_second": 15.164,
|
| 23 |
+
"eval_steps_per_second": 3.791,
|
| 24 |
+
"step": 3
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.08108108108108109,
|
| 28 |
+
"grad_norm": 4.004100322723389,
|
| 29 |
+
"learning_rate": 0.0001,
|
| 30 |
+
"loss": 2.8734,
|
| 31 |
+
"step": 6
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.08108108108108109,
|
| 35 |
+
"eval_loss": 2.6094236373901367,
|
| 36 |
+
"eval_runtime": 1.0592,
|
| 37 |
+
"eval_samples_per_second": 15.106,
|
| 38 |
+
"eval_steps_per_second": 3.777,
|
| 39 |
+
"step": 6
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.12162162162162163,
|
| 43 |
+
"grad_norm": 3.935053586959839,
|
| 44 |
+
"learning_rate": 9.999353337510526e-05,
|
| 45 |
+
"loss": 2.4188,
|
| 46 |
+
"step": 9
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.12162162162162163,
|
| 50 |
+
"eval_loss": 2.1545872688293457,
|
| 51 |
+
"eval_runtime": 1.0511,
|
| 52 |
+
"eval_samples_per_second": 15.222,
|
| 53 |
+
"eval_steps_per_second": 3.805,
|
| 54 |
+
"step": 9
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.16216216216216217,
|
| 58 |
+
"grad_norm": 5.741048812866211,
|
| 59 |
+
"learning_rate": 9.997413517311055e-05,
|
| 60 |
+
"loss": 1.9335,
|
| 61 |
+
"step": 12
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.16216216216216217,
|
| 65 |
+
"eval_loss": 1.786160945892334,
|
| 66 |
+
"eval_runtime": 1.0532,
|
| 67 |
+
"eval_samples_per_second": 15.192,
|
| 68 |
+
"eval_steps_per_second": 3.798,
|
| 69 |
+
"step": 12
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.20270270270270271,
|
| 73 |
+
"grad_norm": 4.155601978302002,
|
| 74 |
+
"learning_rate": 9.99418104116517e-05,
|
| 75 |
+
"loss": 1.5361,
|
| 76 |
+
"step": 15
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.20270270270270271,
|
| 80 |
+
"eval_loss": 1.4731855392456055,
|
| 81 |
+
"eval_runtime": 1.0511,
|
| 82 |
+
"eval_samples_per_second": 15.222,
|
| 83 |
+
"eval_steps_per_second": 3.805,
|
| 84 |
+
"step": 15
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.24324324324324326,
|
| 88 |
+
"grad_norm": 2.4831109046936035,
|
| 89 |
+
"learning_rate": 9.989656745201298e-05,
|
| 90 |
+
"loss": 1.314,
|
| 91 |
+
"step": 18
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 0.24324324324324326,
|
| 95 |
+
"eval_loss": 1.2790606021881104,
|
| 96 |
+
"eval_runtime": 1.0553,
|
| 97 |
+
"eval_samples_per_second": 15.161,
|
| 98 |
+
"eval_steps_per_second": 3.79,
|
| 99 |
+
"step": 18
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.28378378378378377,
|
| 103 |
+
"grad_norm": 1.9509971141815186,
|
| 104 |
+
"learning_rate": 9.983841799696438e-05,
|
| 105 |
+
"loss": 1.1747,
|
| 106 |
+
"step": 21
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 0.28378378378378377,
|
| 110 |
+
"eval_loss": 1.1653475761413574,
|
| 111 |
+
"eval_runtime": 1.0512,
|
| 112 |
+
"eval_samples_per_second": 15.221,
|
| 113 |
+
"eval_steps_per_second": 3.805,
|
| 114 |
+
"step": 21
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.32432432432432434,
|
| 118 |
+
"grad_norm": 2.245741367340088,
|
| 119 |
+
"learning_rate": 9.976737708773445e-05,
|
| 120 |
+
"loss": 1.1407,
|
| 121 |
+
"step": 24
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.32432432432432434,
|
| 125 |
+
"eval_loss": 1.110356092453003,
|
| 126 |
+
"eval_runtime": 1.0534,
|
| 127 |
+
"eval_samples_per_second": 15.188,
|
| 128 |
+
"eval_steps_per_second": 3.797,
|
| 129 |
+
"step": 24
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.36486486486486486,
|
| 133 |
+
"grad_norm": 2.0690531730651855,
|
| 134 |
+
"learning_rate": 9.968346310011964e-05,
|
| 135 |
+
"loss": 1.1734,
|
| 136 |
+
"step": 27
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.36486486486486486,
|
| 140 |
+
"eval_loss": 1.088733434677124,
|
| 141 |
+
"eval_runtime": 1.0508,
|
| 142 |
+
"eval_samples_per_second": 15.226,
|
| 143 |
+
"eval_steps_per_second": 3.806,
|
| 144 |
+
"step": 27
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.40540540540540543,
|
| 148 |
+
"grad_norm": 1.8963656425476074,
|
| 149 |
+
"learning_rate": 9.958669773973123e-05,
|
| 150 |
+
"loss": 1.0495,
|
| 151 |
+
"step": 30
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 0.40540540540540543,
|
| 155 |
+
"eval_loss": 1.0401344299316406,
|
| 156 |
+
"eval_runtime": 1.0528,
|
| 157 |
+
"eval_samples_per_second": 15.197,
|
| 158 |
+
"eval_steps_per_second": 3.799,
|
| 159 |
+
"step": 30
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.44594594594594594,
|
| 163 |
+
"grad_norm": 1.753909945487976,
|
| 164 |
+
"learning_rate": 9.947710603638078e-05,
|
| 165 |
+
"loss": 1.0401,
|
| 166 |
+
"step": 33
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 0.44594594594594594,
|
| 170 |
+
"eval_loss": 0.990611732006073,
|
| 171 |
+
"eval_runtime": 1.0507,
|
| 172 |
+
"eval_samples_per_second": 15.227,
|
| 173 |
+
"eval_steps_per_second": 3.807,
|
| 174 |
+
"step": 33
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.4864864864864865,
|
| 178 |
+
"grad_norm": 2.1073760986328125,
|
| 179 |
+
"learning_rate": 9.935471633760573e-05,
|
| 180 |
+
"loss": 1.0623,
|
| 181 |
+
"step": 36
|
| 182 |
+
},
|
| 183 |
+
{
|
| 184 |
+
"epoch": 0.4864864864864865,
|
| 185 |
+
"eval_loss": 0.9593618512153625,
|
| 186 |
+
"eval_runtime": 1.0535,
|
| 187 |
+
"eval_samples_per_second": 15.188,
|
| 188 |
+
"eval_steps_per_second": 3.797,
|
| 189 |
+
"step": 36
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.527027027027027,
|
| 193 |
+
"grad_norm": 1.5675249099731445,
|
| 194 |
+
"learning_rate": 9.921956030133701e-05,
|
| 195 |
+
"loss": 0.8152,
|
| 196 |
+
"step": 39
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 0.527027027027027,
|
| 200 |
+
"eval_loss": 0.9366932511329651,
|
| 201 |
+
"eval_runtime": 1.0514,
|
| 202 |
+
"eval_samples_per_second": 15.218,
|
| 203 |
+
"eval_steps_per_second": 3.805,
|
| 204 |
+
"step": 39
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.5675675675675675,
|
| 208 |
+
"grad_norm": 2.219888210296631,
|
| 209 |
+
"learning_rate": 9.907167288771019e-05,
|
| 210 |
+
"loss": 0.9261,
|
| 211 |
+
"step": 42
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 0.5675675675675675,
|
| 215 |
+
"eval_loss": 0.9247606992721558,
|
| 216 |
+
"eval_runtime": 1.0532,
|
| 217 |
+
"eval_samples_per_second": 15.192,
|
| 218 |
+
"eval_steps_per_second": 3.798,
|
| 219 |
+
"step": 42
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.6081081081081081,
|
| 223 |
+
"grad_norm": 1.6866446733474731,
|
| 224 |
+
"learning_rate": 9.891109235002249e-05,
|
| 225 |
+
"loss": 0.9469,
|
| 226 |
+
"step": 45
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.6081081081081081,
|
| 230 |
+
"eval_loss": 0.9134540557861328,
|
| 231 |
+
"eval_runtime": 1.0562,
|
| 232 |
+
"eval_samples_per_second": 15.149,
|
| 233 |
+
"eval_steps_per_second": 3.787,
|
| 234 |
+
"step": 45
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.6486486486486487,
|
| 238 |
+
"grad_norm": 1.7272800207138062,
|
| 239 |
+
"learning_rate": 9.8737860224838e-05,
|
| 240 |
+
"loss": 0.8381,
|
| 241 |
+
"step": 48
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.6486486486486487,
|
| 245 |
+
"eval_loss": 0.8871217370033264,
|
| 246 |
+
"eval_runtime": 1.0527,
|
| 247 |
+
"eval_samples_per_second": 15.199,
|
| 248 |
+
"eval_steps_per_second": 3.8,
|
| 249 |
+
"step": 48
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.6891891891891891,
|
| 253 |
+
"grad_norm": 2.6152303218841553,
|
| 254 |
+
"learning_rate": 9.855202132124365e-05,
|
| 255 |
+
"loss": 0.8456,
|
| 256 |
+
"step": 51
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 0.6891891891891891,
|
| 260 |
+
"eval_loss": 0.8553087711334229,
|
| 261 |
+
"eval_runtime": 1.0521,
|
| 262 |
+
"eval_samples_per_second": 15.208,
|
| 263 |
+
"eval_steps_per_second": 3.802,
|
| 264 |
+
"step": 51
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.7297297297297297,
|
| 268 |
+
"grad_norm": 1.8282960653305054,
|
| 269 |
+
"learning_rate": 9.835362370925868e-05,
|
| 270 |
+
"loss": 0.908,
|
| 271 |
+
"step": 54
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 0.7297297297297297,
|
| 275 |
+
"eval_loss": 0.8271682858467102,
|
| 276 |
+
"eval_runtime": 1.052,
|
| 277 |
+
"eval_samples_per_second": 15.21,
|
| 278 |
+
"eval_steps_per_second": 3.802,
|
| 279 |
+
"step": 54
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.7702702702702703,
|
| 283 |
+
"grad_norm": 2.466750383377075,
|
| 284 |
+
"learning_rate": 9.814271870740054e-05,
|
| 285 |
+
"loss": 0.999,
|
| 286 |
+
"step": 57
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 0.7702702702702703,
|
| 290 |
+
"eval_loss": 0.8151593208312988,
|
| 291 |
+
"eval_runtime": 1.0549,
|
| 292 |
+
"eval_samples_per_second": 15.167,
|
| 293 |
+
"eval_steps_per_second": 3.792,
|
| 294 |
+
"step": 57
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.8108108108108109,
|
| 298 |
+
"grad_norm": 1.8908120393753052,
|
| 299 |
+
"learning_rate": 9.791936086941064e-05,
|
| 300 |
+
"loss": 0.897,
|
| 301 |
+
"step": 60
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.8108108108108109,
|
| 305 |
+
"eval_loss": 0.8052847981452942,
|
| 306 |
+
"eval_runtime": 1.0512,
|
| 307 |
+
"eval_samples_per_second": 15.22,
|
| 308 |
+
"eval_steps_per_second": 3.805,
|
| 309 |
+
"step": 60
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.8513513513513513,
|
| 313 |
+
"grad_norm": 1.9563689231872559,
|
| 314 |
+
"learning_rate": 9.768360797014324e-05,
|
| 315 |
+
"loss": 0.8747,
|
| 316 |
+
"step": 63
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 0.8513513513513513,
|
| 320 |
+
"eval_loss": 0.7914941906929016,
|
| 321 |
+
"eval_runtime": 1.0519,
|
| 322 |
+
"eval_samples_per_second": 15.21,
|
| 323 |
+
"eval_steps_per_second": 3.803,
|
| 324 |
+
"step": 63
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.8918918918918919,
|
| 328 |
+
"grad_norm": 1.9292480945587158,
|
| 329 |
+
"learning_rate": 9.7435520990621e-05,
|
| 330 |
+
"loss": 1.0646,
|
| 331 |
+
"step": 66
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.8918918918918919,
|
| 335 |
+
"eval_loss": 0.7872657179832458,
|
| 336 |
+
"eval_runtime": 1.0526,
|
| 337 |
+
"eval_samples_per_second": 15.201,
|
| 338 |
+
"eval_steps_per_second": 3.8,
|
| 339 |
+
"step": 66
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.9324324324324325,
|
| 343 |
+
"grad_norm": 1.7248555421829224,
|
| 344 |
+
"learning_rate": 9.717516410226145e-05,
|
| 345 |
+
"loss": 0.6771,
|
| 346 |
+
"step": 69
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.9324324324324325,
|
| 350 |
+
"eval_loss": 0.7814666628837585,
|
| 351 |
+
"eval_runtime": 1.0522,
|
| 352 |
+
"eval_samples_per_second": 15.207,
|
| 353 |
+
"eval_steps_per_second": 3.802,
|
| 354 |
+
"step": 69
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.972972972972973,
|
| 358 |
+
"grad_norm": 2.171896457672119,
|
| 359 |
+
"learning_rate": 9.690260465027801e-05,
|
| 360 |
+
"loss": 0.9386,
|
| 361 |
+
"step": 72
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.972972972972973,
|
| 365 |
+
"eval_loss": 0.7634860873222351,
|
| 366 |
+
"eval_runtime": 1.0498,
|
| 367 |
+
"eval_samples_per_second": 15.241,
|
| 368 |
+
"eval_steps_per_second": 3.81,
|
| 369 |
+
"step": 72
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 1.0135135135135136,
|
| 373 |
+
"grad_norm": 1.625179409980774,
|
| 374 |
+
"learning_rate": 9.661791313626018e-05,
|
| 375 |
+
"loss": 0.6348,
|
| 376 |
+
"step": 75
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 1.0135135135135136,
|
| 380 |
+
"eval_loss": 0.75515216588974,
|
| 381 |
+
"eval_runtime": 1.0536,
|
| 382 |
+
"eval_samples_per_second": 15.186,
|
| 383 |
+
"eval_steps_per_second": 3.796,
|
| 384 |
+
"step": 75
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 1.054054054054054,
|
| 388 |
+
"grad_norm": 1.4293404817581177,
|
| 389 |
+
"learning_rate": 9.632116319993725e-05,
|
| 390 |
+
"loss": 0.5763,
|
| 391 |
+
"step": 78
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 1.054054054054054,
|
| 395 |
+
"eval_loss": 0.7473800182342529,
|
| 396 |
+
"eval_runtime": 1.0524,
|
| 397 |
+
"eval_samples_per_second": 15.203,
|
| 398 |
+
"eval_steps_per_second": 3.801,
|
| 399 |
+
"step": 78
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 1.0945945945945945,
|
| 403 |
+
"grad_norm": 1.9279707670211792,
|
| 404 |
+
"learning_rate": 9.601243160013023e-05,
|
| 405 |
+
"loss": 0.7059,
|
| 406 |
+
"step": 81
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 1.0945945945945945,
|
| 410 |
+
"eval_loss": 0.7430617213249207,
|
| 411 |
+
"eval_runtime": 1.0539,
|
| 412 |
+
"eval_samples_per_second": 15.181,
|
| 413 |
+
"eval_steps_per_second": 3.795,
|
| 414 |
+
"step": 81
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 1.135135135135135,
|
| 418 |
+
"grad_norm": 1.7644144296646118,
|
| 419 |
+
"learning_rate": 9.56917981948971e-05,
|
| 420 |
+
"loss": 0.6111,
|
| 421 |
+
"step": 84
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 1.135135135135135,
|
| 425 |
+
"eval_loss": 0.7393875122070312,
|
| 426 |
+
"eval_runtime": 1.0525,
|
| 427 |
+
"eval_samples_per_second": 15.202,
|
| 428 |
+
"eval_steps_per_second": 3.8,
|
| 429 |
+
"step": 84
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 1.1756756756756757,
|
| 433 |
+
"grad_norm": 1.4910467863082886,
|
| 434 |
+
"learning_rate": 9.535934592087627e-05,
|
| 435 |
+
"loss": 0.6937,
|
| 436 |
+
"step": 87
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 1.1756756756756757,
|
| 440 |
+
"eval_loss": 0.7415614724159241,
|
| 441 |
+
"eval_runtime": 1.0533,
|
| 442 |
+
"eval_samples_per_second": 15.191,
|
| 443 |
+
"eval_steps_per_second": 3.798,
|
| 444 |
+
"step": 87
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.2162162162162162,
|
| 448 |
+
"grad_norm": 1.989018440246582,
|
| 449 |
+
"learning_rate": 9.50151607718338e-05,
|
| 450 |
+
"loss": 0.6408,
|
| 451 |
+
"step": 90
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 1.2162162162162162,
|
| 455 |
+
"eval_loss": 0.7331891059875488,
|
| 456 |
+
"eval_runtime": 1.0504,
|
| 457 |
+
"eval_samples_per_second": 15.232,
|
| 458 |
+
"eval_steps_per_second": 3.808,
|
| 459 |
+
"step": 90
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.2567567567567568,
|
| 463 |
+
"grad_norm": 1.5546590089797974,
|
| 464 |
+
"learning_rate": 9.465933177641982e-05,
|
| 465 |
+
"loss": 0.5931,
|
| 466 |
+
"step": 93
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 1.2567567567567568,
|
| 470 |
+
"eval_loss": 0.7319458723068237,
|
| 471 |
+
"eval_runtime": 1.0532,
|
| 472 |
+
"eval_samples_per_second": 15.191,
|
| 473 |
+
"eval_steps_per_second": 3.798,
|
| 474 |
+
"step": 93
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 1.2972972972972974,
|
| 478 |
+
"grad_norm": 2.128746271133423,
|
| 479 |
+
"learning_rate": 9.429195097513993e-05,
|
| 480 |
+
"loss": 0.5792,
|
| 481 |
+
"step": 96
|
| 482 |
+
},
|
| 483 |
+
{
|
| 484 |
+
"epoch": 1.2972972972972974,
|
| 485 |
+
"eval_loss": 0.7179479598999023,
|
| 486 |
+
"eval_runtime": 1.0504,
|
| 487 |
+
"eval_samples_per_second": 15.232,
|
| 488 |
+
"eval_steps_per_second": 3.808,
|
| 489 |
+
"step": 96
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 1.3378378378378377,
|
| 493 |
+
"grad_norm": 2.069204092025757,
|
| 494 |
+
"learning_rate": 9.391311339654753e-05,
|
| 495 |
+
"loss": 0.5502,
|
| 496 |
+
"step": 99
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 1.3378378378378377,
|
| 500 |
+
"eval_loss": 0.7083268165588379,
|
| 501 |
+
"eval_runtime": 1.0531,
|
| 502 |
+
"eval_samples_per_second": 15.193,
|
| 503 |
+
"eval_steps_per_second": 3.798,
|
| 504 |
+
"step": 99
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 1.3783783783783785,
|
| 508 |
+
"grad_norm": 2.069469928741455,
|
| 509 |
+
"learning_rate": 9.352291703266331e-05,
|
| 510 |
+
"loss": 0.7356,
|
| 511 |
+
"step": 102
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 1.3783783783783785,
|
| 515 |
+
"eval_loss": 0.7048563957214355,
|
| 516 |
+
"eval_runtime": 1.0519,
|
| 517 |
+
"eval_samples_per_second": 15.21,
|
| 518 |
+
"eval_steps_per_second": 3.803,
|
| 519 |
+
"step": 102
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 1.4189189189189189,
|
| 523 |
+
"grad_norm": 1.507051706314087,
|
| 524 |
+
"learning_rate": 9.31214628136281e-05,
|
| 525 |
+
"loss": 0.5204,
|
| 526 |
+
"step": 105
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 1.4189189189189189,
|
| 530 |
+
"eval_loss": 0.6983195543289185,
|
| 531 |
+
"eval_runtime": 1.0543,
|
| 532 |
+
"eval_samples_per_second": 15.176,
|
| 533 |
+
"eval_steps_per_second": 3.794,
|
| 534 |
+
"step": 105
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 1.4594594594594594,
|
| 538 |
+
"grad_norm": 1.918865442276001,
|
| 539 |
+
"learning_rate": 9.270885458159575e-05,
|
| 540 |
+
"loss": 0.6132,
|
| 541 |
+
"step": 108
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 1.4594594594594594,
|
| 545 |
+
"eval_loss": 0.6857842803001404,
|
| 546 |
+
"eval_runtime": 1.0525,
|
| 547 |
+
"eval_samples_per_second": 15.202,
|
| 548 |
+
"eval_steps_per_second": 3.8,
|
| 549 |
+
"step": 108
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.5,
|
| 553 |
+
"grad_norm": 2.062997341156006,
|
| 554 |
+
"learning_rate": 9.228519906387288e-05,
|
| 555 |
+
"loss": 0.7527,
|
| 556 |
+
"step": 111
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 1.5,
|
| 560 |
+
"eval_loss": 0.6743776798248291,
|
| 561 |
+
"eval_runtime": 1.0512,
|
| 562 |
+
"eval_samples_per_second": 15.221,
|
| 563 |
+
"eval_steps_per_second": 3.805,
|
| 564 |
+
"step": 111
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.5405405405405406,
|
| 568 |
+
"grad_norm": 1.8099018335342407,
|
| 569 |
+
"learning_rate": 9.185060584531217e-05,
|
| 570 |
+
"loss": 0.6798,
|
| 571 |
+
"step": 114
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.5405405405405406,
|
| 575 |
+
"eval_loss": 0.6715844869613647,
|
| 576 |
+
"eval_runtime": 1.0529,
|
| 577 |
+
"eval_samples_per_second": 15.196,
|
| 578 |
+
"eval_steps_per_second": 3.799,
|
| 579 |
+
"step": 114
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 1.5810810810810811,
|
| 583 |
+
"grad_norm": 2.0540611743927,
|
| 584 |
+
"learning_rate": 9.140518733996672e-05,
|
| 585 |
+
"loss": 0.7266,
|
| 586 |
+
"step": 117
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 1.5810810810810811,
|
| 590 |
+
"eval_loss": 0.6656138896942139,
|
| 591 |
+
"eval_runtime": 1.0523,
|
| 592 |
+
"eval_samples_per_second": 15.204,
|
| 593 |
+
"eval_steps_per_second": 3.801,
|
| 594 |
+
"step": 117
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 1.6216216216216215,
|
| 598 |
+
"grad_norm": 2.3945634365081787,
|
| 599 |
+
"learning_rate": 9.094905876201229e-05,
|
| 600 |
+
"loss": 0.5347,
|
| 601 |
+
"step": 120
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 1.6216216216216215,
|
| 605 |
+
"eval_loss": 0.6710730791091919,
|
| 606 |
+
"eval_runtime": 1.053,
|
| 607 |
+
"eval_samples_per_second": 15.195,
|
| 608 |
+
"eval_steps_per_second": 3.799,
|
| 609 |
+
"step": 120
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 1.6621621621621623,
|
| 613 |
+
"grad_norm": 2.006612777709961,
|
| 614 |
+
"learning_rate": 9.048233809594561e-05,
|
| 615 |
+
"loss": 0.6522,
|
| 616 |
+
"step": 123
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 1.6621621621621623,
|
| 620 |
+
"eval_loss": 0.6679877042770386,
|
| 621 |
+
"eval_runtime": 1.0519,
|
| 622 |
+
"eval_samples_per_second": 15.211,
|
| 623 |
+
"eval_steps_per_second": 3.803,
|
| 624 |
+
"step": 123
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 1.7027027027027026,
|
| 628 |
+
"grad_norm": 1.751696228981018,
|
| 629 |
+
"learning_rate": 9.000514606606581e-05,
|
| 630 |
+
"loss": 0.8567,
|
| 631 |
+
"step": 126
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 1.7027027027027026,
|
| 635 |
+
"eval_loss": 0.6558159589767456,
|
| 636 |
+
"eval_runtime": 1.0531,
|
| 637 |
+
"eval_samples_per_second": 15.193,
|
| 638 |
+
"eval_steps_per_second": 3.798,
|
| 639 |
+
"step": 126
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 1.7432432432432432,
|
| 643 |
+
"grad_norm": 1.5286139249801636,
|
| 644 |
+
"learning_rate": 8.951760610524724e-05,
|
| 645 |
+
"loss": 0.5204,
|
| 646 |
+
"step": 129
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 1.7432432432432432,
|
| 650 |
+
"eval_loss": 0.6488269567489624,
|
| 651 |
+
"eval_runtime": 1.0516,
|
| 652 |
+
"eval_samples_per_second": 15.215,
|
| 653 |
+
"eval_steps_per_second": 3.804,
|
| 654 |
+
"step": 129
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 1.7837837837837838,
|
| 658 |
+
"grad_norm": 2.1092898845672607,
|
| 659 |
+
"learning_rate": 8.901984432301185e-05,
|
| 660 |
+
"loss": 0.6443,
|
| 661 |
+
"step": 132
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 1.7837837837837838,
|
| 665 |
+
"eval_loss": 0.6392868161201477,
|
| 666 |
+
"eval_runtime": 1.053,
|
| 667 |
+
"eval_samples_per_second": 15.195,
|
| 668 |
+
"eval_steps_per_second": 3.799,
|
| 669 |
+
"step": 132
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 1.8243243243243243,
|
| 673 |
+
"grad_norm": 1.7279053926467896,
|
| 674 |
+
"learning_rate": 8.851198947290894e-05,
|
| 675 |
+
"loss": 0.5436,
|
| 676 |
+
"step": 135
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 1.8243243243243243,
|
| 680 |
+
"eval_loss": 0.6321672201156616,
|
| 681 |
+
"eval_runtime": 1.0499,
|
| 682 |
+
"eval_samples_per_second": 15.239,
|
| 683 |
+
"eval_steps_per_second": 3.81,
|
| 684 |
+
"step": 135
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 1.864864864864865,
|
| 688 |
+
"grad_norm": 2.6842877864837646,
|
| 689 |
+
"learning_rate": 8.799417291921117e-05,
|
| 690 |
+
"loss": 0.6054,
|
| 691 |
+
"step": 138
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 1.864864864864865,
|
| 695 |
+
"eval_loss": 0.6346270442008972,
|
| 696 |
+
"eval_runtime": 1.0528,
|
| 697 |
+
"eval_samples_per_second": 15.198,
|
| 698 |
+
"eval_steps_per_second": 3.799,
|
| 699 |
+
"step": 138
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.9054054054054053,
|
| 703 |
+
"grad_norm": 1.9958398342132568,
|
| 704 |
+
"learning_rate": 8.746652860293523e-05,
|
| 705 |
+
"loss": 0.4488,
|
| 706 |
+
"step": 141
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.9054054054054053,
|
| 710 |
+
"eval_loss": 0.6389164924621582,
|
| 711 |
+
"eval_runtime": 1.0505,
|
| 712 |
+
"eval_samples_per_second": 15.231,
|
| 713 |
+
"eval_steps_per_second": 3.808,
|
| 714 |
+
"step": 141
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.945945945945946,
|
| 718 |
+
"grad_norm": 2.0705783367156982,
|
| 719 |
+
"learning_rate": 8.692919300719595e-05,
|
| 720 |
+
"loss": 0.7171,
|
| 721 |
+
"step": 144
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 1.945945945945946,
|
| 725 |
+
"eval_loss": 0.632194995880127,
|
| 726 |
+
"eval_runtime": 1.0537,
|
| 727 |
+
"eval_samples_per_second": 15.184,
|
| 728 |
+
"eval_steps_per_second": 3.796,
|
| 729 |
+
"step": 144
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 1.9864864864864864,
|
| 733 |
+
"grad_norm": 2.0737218856811523,
|
| 734 |
+
"learning_rate": 8.638230512190298e-05,
|
| 735 |
+
"loss": 0.5383,
|
| 736 |
+
"step": 147
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 1.9864864864864864,
|
| 740 |
+
"eval_loss": 0.6272808313369751,
|
| 741 |
+
"eval_runtime": 1.0507,
|
| 742 |
+
"eval_samples_per_second": 15.228,
|
| 743 |
+
"eval_steps_per_second": 3.807,
|
| 744 |
+
"step": 147
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 2.027027027027027,
|
| 748 |
+
"grad_norm": 1.6119190454483032,
|
| 749 |
+
"learning_rate": 8.58260064078088e-05,
|
| 750 |
+
"loss": 0.4812,
|
| 751 |
+
"step": 150
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 2.027027027027027,
|
| 755 |
+
"eval_loss": 0.6234598755836487,
|
| 756 |
+
"eval_runtime": 1.0541,
|
| 757 |
+
"eval_samples_per_second": 15.179,
|
| 758 |
+
"eval_steps_per_second": 3.795,
|
| 759 |
+
"step": 150
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 2.0675675675675675,
|
| 763 |
+
"grad_norm": 2.104738712310791,
|
| 764 |
+
"learning_rate": 8.526044075991802e-05,
|
| 765 |
+
"loss": 0.7911,
|
| 766 |
+
"step": 153
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 2.0675675675675675,
|
| 770 |
+
"eval_loss": 0.6295649409294128,
|
| 771 |
+
"eval_runtime": 1.0504,
|
| 772 |
+
"eval_samples_per_second": 15.232,
|
| 773 |
+
"eval_steps_per_second": 3.808,
|
| 774 |
+
"step": 153
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 2.108108108108108,
|
| 778 |
+
"grad_norm": 2.041696786880493,
|
| 779 |
+
"learning_rate": 8.468575447026651e-05,
|
| 780 |
+
"loss": 0.514,
|
| 781 |
+
"step": 156
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 2.108108108108108,
|
| 785 |
+
"eval_loss": 0.6444165706634521,
|
| 786 |
+
"eval_runtime": 1.0539,
|
| 787 |
+
"eval_samples_per_second": 15.182,
|
| 788 |
+
"eval_steps_per_second": 3.795,
|
| 789 |
+
"step": 156
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 2.1486486486486487,
|
| 793 |
+
"grad_norm": 1.7887616157531738,
|
| 794 |
+
"learning_rate": 8.410209619008101e-05,
|
| 795 |
+
"loss": 0.4481,
|
| 796 |
+
"step": 159
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 2.1486486486486487,
|
| 800 |
+
"eval_loss": 0.6452795267105103,
|
| 801 |
+
"eval_runtime": 1.0508,
|
| 802 |
+
"eval_samples_per_second": 15.227,
|
| 803 |
+
"eval_steps_per_second": 3.807,
|
| 804 |
+
"step": 159
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 2.189189189189189,
|
| 808 |
+
"grad_norm": 2.2852938175201416,
|
| 809 |
+
"learning_rate": 8.350961689132808e-05,
|
| 810 |
+
"loss": 0.3983,
|
| 811 |
+
"step": 162
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 2.189189189189189,
|
| 815 |
+
"eval_loss": 0.6356573104858398,
|
| 816 |
+
"eval_runtime": 1.0538,
|
| 817 |
+
"eval_samples_per_second": 15.183,
|
| 818 |
+
"eval_steps_per_second": 3.796,
|
| 819 |
+
"step": 162
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 2.22972972972973,
|
| 823 |
+
"grad_norm": 1.3814259767532349,
|
| 824 |
+
"learning_rate": 8.290846982766305e-05,
|
| 825 |
+
"loss": 0.2386,
|
| 826 |
+
"step": 165
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 2.22972972972973,
|
| 830 |
+
"eval_loss": 0.632733166217804,
|
| 831 |
+
"eval_runtime": 1.053,
|
| 832 |
+
"eval_samples_per_second": 15.195,
|
| 833 |
+
"eval_steps_per_second": 3.799,
|
| 834 |
+
"step": 165
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 2.27027027027027,
|
| 838 |
+
"grad_norm": 2.624509572982788,
|
| 839 |
+
"learning_rate": 8.22988104947886e-05,
|
| 840 |
+
"loss": 0.4447,
|
| 841 |
+
"step": 168
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 2.27027027027027,
|
| 845 |
+
"eval_loss": 0.6358802318572998,
|
| 846 |
+
"eval_runtime": 1.0518,
|
| 847 |
+
"eval_samples_per_second": 15.212,
|
| 848 |
+
"eval_steps_per_second": 3.803,
|
| 849 |
+
"step": 168
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 2.310810810810811,
|
| 853 |
+
"grad_norm": 2.1006217002868652,
|
| 854 |
+
"learning_rate": 8.168079659023349e-05,
|
| 855 |
+
"loss": 0.4302,
|
| 856 |
+
"step": 171
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 2.310810810810811,
|
| 860 |
+
"eval_loss": 0.6386667490005493,
|
| 861 |
+
"eval_runtime": 1.0534,
|
| 862 |
+
"eval_samples_per_second": 15.188,
|
| 863 |
+
"eval_steps_per_second": 3.797,
|
| 864 |
+
"step": 171
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 2.3513513513513513,
|
| 868 |
+
"grad_norm": 2.631301164627075,
|
| 869 |
+
"learning_rate": 8.105458797256178e-05,
|
| 870 |
+
"loss": 0.4514,
|
| 871 |
+
"step": 174
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 2.3513513513513513,
|
| 875 |
+
"eval_loss": 0.6402238607406616,
|
| 876 |
+
"eval_runtime": 1.0545,
|
| 877 |
+
"eval_samples_per_second": 15.174,
|
| 878 |
+
"eval_steps_per_second": 3.793,
|
| 879 |
+
"step": 174
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 2.391891891891892,
|
| 883 |
+
"grad_norm": 1.4005826711654663,
|
| 884 |
+
"learning_rate": 8.04203466200229e-05,
|
| 885 |
+
"loss": 0.2813,
|
| 886 |
+
"step": 177
|
| 887 |
+
},
|
| 888 |
+
{
|
| 889 |
+
"epoch": 2.391891891891892,
|
| 890 |
+
"eval_loss": 0.6313220262527466,
|
| 891 |
+
"eval_runtime": 1.0541,
|
| 892 |
+
"eval_samples_per_second": 15.178,
|
| 893 |
+
"eval_steps_per_second": 3.795,
|
| 894 |
+
"step": 177
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 2.4324324324324325,
|
| 898 |
+
"grad_norm": 2.4380390644073486,
|
| 899 |
+
"learning_rate": 7.977823658865364e-05,
|
| 900 |
+
"loss": 0.4747,
|
| 901 |
+
"step": 180
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 2.4324324324324325,
|
| 905 |
+
"eval_loss": 0.6258513927459717,
|
| 906 |
+
"eval_runtime": 1.0533,
|
| 907 |
+
"eval_samples_per_second": 15.191,
|
| 908 |
+
"eval_steps_per_second": 3.798,
|
| 909 |
+
"step": 180
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.472972972972973,
|
| 913 |
+
"grad_norm": 2.3655426502227783,
|
| 914 |
+
"learning_rate": 7.912842396984254e-05,
|
| 915 |
+
"loss": 0.547,
|
| 916 |
+
"step": 183
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 2.472972972972973,
|
| 920 |
+
"eval_loss": 0.6256988048553467,
|
| 921 |
+
"eval_runtime": 1.053,
|
| 922 |
+
"eval_samples_per_second": 15.195,
|
| 923 |
+
"eval_steps_per_second": 3.799,
|
| 924 |
+
"step": 183
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 2.5135135135135136,
|
| 928 |
+
"grad_norm": 1.9949471950531006,
|
| 929 |
+
"learning_rate": 7.847107684736792e-05,
|
| 930 |
+
"loss": 0.3154,
|
| 931 |
+
"step": 186
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 2.5135135135135136,
|
| 935 |
+
"eval_loss": 0.6247289776802063,
|
| 936 |
+
"eval_runtime": 1.0523,
|
| 937 |
+
"eval_samples_per_second": 15.205,
|
| 938 |
+
"eval_steps_per_second": 3.801,
|
| 939 |
+
"step": 186
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 2.554054054054054,
|
| 943 |
+
"grad_norm": 3.2453622817993164,
|
| 944 |
+
"learning_rate": 7.780636525392046e-05,
|
| 945 |
+
"loss": 0.5583,
|
| 946 |
+
"step": 189
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 2.554054054054054,
|
| 950 |
+
"eval_loss": 0.6129618883132935,
|
| 951 |
+
"eval_runtime": 1.0519,
|
| 952 |
+
"eval_samples_per_second": 15.21,
|
| 953 |
+
"eval_steps_per_second": 3.803,
|
| 954 |
+
"step": 189
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 2.5945945945945947,
|
| 958 |
+
"grad_norm": 2.022986888885498,
|
| 959 |
+
"learning_rate": 7.713446112712169e-05,
|
| 960 |
+
"loss": 0.5726,
|
| 961 |
+
"step": 192
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 2.5945945945945947,
|
| 965 |
+
"eval_loss": 0.6086827516555786,
|
| 966 |
+
"eval_runtime": 1.0543,
|
| 967 |
+
"eval_samples_per_second": 15.175,
|
| 968 |
+
"eval_steps_per_second": 3.794,
|
| 969 |
+
"step": 192
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 2.635135135135135,
|
| 973 |
+
"grad_norm": 2.429865598678589,
|
| 974 |
+
"learning_rate": 7.645553826504969e-05,
|
| 975 |
+
"loss": 0.4701,
|
| 976 |
+
"step": 195
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 2.635135135135135,
|
| 980 |
+
"eval_loss": 0.6085944175720215,
|
| 981 |
+
"eval_runtime": 1.0521,
|
| 982 |
+
"eval_samples_per_second": 15.208,
|
| 983 |
+
"eval_steps_per_second": 3.802,
|
| 984 |
+
"step": 195
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 2.6756756756756754,
|
| 988 |
+
"grad_norm": 1.991803526878357,
|
| 989 |
+
"learning_rate": 7.576977228128376e-05,
|
| 990 |
+
"loss": 0.4866,
|
| 991 |
+
"step": 198
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 2.6756756756756754,
|
| 995 |
+
"eval_loss": 0.6133272647857666,
|
| 996 |
+
"eval_runtime": 1.0535,
|
| 997 |
+
"eval_samples_per_second": 15.187,
|
| 998 |
+
"eval_steps_per_second": 3.797,
|
| 999 |
+
"step": 198
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 2.7162162162162162,
|
| 1003 |
+
"grad_norm": 2.537832021713257,
|
| 1004 |
+
"learning_rate": 7.50773405594792e-05,
|
| 1005 |
+
"loss": 0.4015,
|
| 1006 |
+
"step": 201
|
| 1007 |
+
},
|
| 1008 |
+
{
|
| 1009 |
+
"epoch": 2.7162162162162162,
|
| 1010 |
+
"eval_loss": 0.6213403940200806,
|
| 1011 |
+
"eval_runtime": 1.0524,
|
| 1012 |
+
"eval_samples_per_second": 15.203,
|
| 1013 |
+
"eval_steps_per_second": 3.801,
|
| 1014 |
+
"step": 201
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 2.756756756756757,
|
| 1018 |
+
"grad_norm": 1.758016586303711,
|
| 1019 |
+
"learning_rate": 7.437842220748441e-05,
|
| 1020 |
+
"loss": 0.4277,
|
| 1021 |
+
"step": 204
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 2.756756756756757,
|
| 1025 |
+
"eval_loss": 0.623763382434845,
|
| 1026 |
+
"eval_runtime": 1.0527,
|
| 1027 |
+
"eval_samples_per_second": 15.198,
|
| 1028 |
+
"eval_steps_per_second": 3.8,
|
| 1029 |
+
"step": 204
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 2.7972972972972974,
|
| 1033 |
+
"grad_norm": 1.8930737972259521,
|
| 1034 |
+
"learning_rate": 7.367319801101196e-05,
|
| 1035 |
+
"loss": 0.3157,
|
| 1036 |
+
"step": 207
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 2.7972972972972974,
|
| 1040 |
+
"eval_loss": 0.6248853206634521,
|
| 1041 |
+
"eval_runtime": 1.0562,
|
| 1042 |
+
"eval_samples_per_second": 15.149,
|
| 1043 |
+
"eval_steps_per_second": 3.787,
|
| 1044 |
+
"step": 207
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 2.8378378378378377,
|
| 1048 |
+
"grad_norm": 2.071988105773926,
|
| 1049 |
+
"learning_rate": 7.296185038687566e-05,
|
| 1050 |
+
"loss": 0.3883,
|
| 1051 |
+
"step": 210
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 2.8378378378378377,
|
| 1055 |
+
"eval_loss": 0.6209710240364075,
|
| 1056 |
+
"eval_runtime": 1.0518,
|
| 1057 |
+
"eval_samples_per_second": 15.212,
|
| 1058 |
+
"eval_steps_per_second": 3.803,
|
| 1059 |
+
"step": 210
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 2.8783783783783785,
|
| 1063 |
+
"grad_norm": 1.579237937927246,
|
| 1064 |
+
"learning_rate": 7.224456333580573e-05,
|
| 1065 |
+
"loss": 0.5436,
|
| 1066 |
+
"step": 213
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 2.8783783783783785,
|
| 1070 |
+
"eval_loss": 0.6127223968505859,
|
| 1071 |
+
"eval_runtime": 1.0524,
|
| 1072 |
+
"eval_samples_per_second": 15.204,
|
| 1073 |
+
"eval_steps_per_second": 3.801,
|
| 1074 |
+
"step": 213
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 2.918918918918919,
|
| 1078 |
+
"grad_norm": 2.4129927158355713,
|
| 1079 |
+
"learning_rate": 7.152152239485419e-05,
|
| 1080 |
+
"loss": 0.526,
|
| 1081 |
+
"step": 216
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 2.918918918918919,
|
| 1085 |
+
"eval_loss": 0.6055560111999512,
|
| 1086 |
+
"eval_runtime": 1.0502,
|
| 1087 |
+
"eval_samples_per_second": 15.236,
|
| 1088 |
+
"eval_steps_per_second": 3.809,
|
| 1089 |
+
"step": 216
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 2.9594594594594597,
|
| 1093 |
+
"grad_norm": 2.252251148223877,
|
| 1094 |
+
"learning_rate": 7.079291458940301e-05,
|
| 1095 |
+
"loss": 0.4465,
|
| 1096 |
+
"step": 219
|
| 1097 |
+
},
|
| 1098 |
+
{
|
| 1099 |
+
"epoch": 2.9594594594594597,
|
| 1100 |
+
"eval_loss": 0.5982283353805542,
|
| 1101 |
+
"eval_runtime": 1.0529,
|
| 1102 |
+
"eval_samples_per_second": 15.197,
|
| 1103 |
+
"eval_steps_per_second": 3.799,
|
| 1104 |
+
"step": 219
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 3.0,
|
| 1108 |
+
"grad_norm": 1.9773114919662476,
|
| 1109 |
+
"learning_rate": 7.005892838478711e-05,
|
| 1110 |
+
"loss": 0.3692,
|
| 1111 |
+
"step": 222
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"epoch": 3.0,
|
| 1115 |
+
"eval_loss": 0.5916565656661987,
|
| 1116 |
+
"eval_runtime": 1.0501,
|
| 1117 |
+
"eval_samples_per_second": 15.237,
|
| 1118 |
+
"eval_steps_per_second": 3.809,
|
| 1119 |
+
"step": 222
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 3.0405405405405403,
|
| 1123 |
+
"grad_norm": 1.1434626579284668,
|
| 1124 |
+
"learning_rate": 6.931975363754502e-05,
|
| 1125 |
+
"loss": 0.3022,
|
| 1126 |
+
"step": 225
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 3.0405405405405403,
|
| 1130 |
+
"eval_loss": 0.5955583453178406,
|
| 1131 |
+
"eval_runtime": 1.0535,
|
| 1132 |
+
"eval_samples_per_second": 15.187,
|
| 1133 |
+
"eval_steps_per_second": 3.797,
|
| 1134 |
+
"step": 225
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 3.081081081081081,
|
| 1138 |
+
"grad_norm": 1.9162238836288452,
|
| 1139 |
+
"learning_rate": 6.85755815463096e-05,
|
| 1140 |
+
"loss": 0.2875,
|
| 1141 |
+
"step": 228
|
| 1142 |
+
},
|
| 1143 |
+
{
|
| 1144 |
+
"epoch": 3.081081081081081,
|
| 1145 |
+
"eval_loss": 0.6152929067611694,
|
| 1146 |
+
"eval_runtime": 1.0516,
|
| 1147 |
+
"eval_samples_per_second": 15.215,
|
| 1148 |
+
"eval_steps_per_second": 3.804,
|
| 1149 |
+
"step": 228
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 3.1216216216216215,
|
| 1153 |
+
"grad_norm": 2.688631057739258,
|
| 1154 |
+
"learning_rate": 6.782660460235174e-05,
|
| 1155 |
+
"loss": 0.5544,
|
| 1156 |
+
"step": 231
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 3.1216216216216215,
|
| 1160 |
+
"eval_loss": 0.6343094110488892,
|
| 1161 |
+
"eval_runtime": 1.052,
|
| 1162 |
+
"eval_samples_per_second": 15.21,
|
| 1163 |
+
"eval_steps_per_second": 3.802,
|
| 1164 |
+
"step": 231
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 3.1621621621621623,
|
| 1168 |
+
"grad_norm": 2.58313250541687,
|
| 1169 |
+
"learning_rate": 6.707301653978945e-05,
|
| 1170 |
+
"loss": 0.4159,
|
| 1171 |
+
"step": 234
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 3.1621621621621623,
|
| 1175 |
+
"eval_loss": 0.6369538307189941,
|
| 1176 |
+
"eval_runtime": 1.0524,
|
| 1177 |
+
"eval_samples_per_second": 15.203,
|
| 1178 |
+
"eval_steps_per_second": 3.801,
|
| 1179 |
+
"step": 234
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 3.2027027027027026,
|
| 1183 |
+
"grad_norm": 2.2415409088134766,
|
| 1184 |
+
"learning_rate": 6.63150122854758e-05,
|
| 1185 |
+
"loss": 0.4963,
|
| 1186 |
+
"step": 237
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 3.2027027027027026,
|
| 1190 |
+
"eval_loss": 0.6289186477661133,
|
| 1191 |
+
"eval_runtime": 1.0528,
|
| 1192 |
+
"eval_samples_per_second": 15.198,
|
| 1193 |
+
"eval_steps_per_second": 3.799,
|
| 1194 |
+
"step": 237
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 3.2432432432432434,
|
| 1198 |
+
"grad_norm": 2.974931240081787,
|
| 1199 |
+
"learning_rate": 6.5552787908578e-05,
|
| 1200 |
+
"loss": 0.3248,
|
| 1201 |
+
"step": 240
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 3.2432432432432434,
|
| 1205 |
+
"eval_loss": 0.6189987659454346,
|
| 1206 |
+
"eval_runtime": 1.0515,
|
| 1207 |
+
"eval_samples_per_second": 15.217,
|
| 1208 |
+
"eval_steps_per_second": 3.804,
|
| 1209 |
+
"step": 240
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 3.2837837837837838,
|
| 1213 |
+
"grad_norm": 2.0078535079956055,
|
| 1214 |
+
"learning_rate": 6.478654056986131e-05,
|
| 1215 |
+
"loss": 0.349,
|
| 1216 |
+
"step": 243
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 3.2837837837837838,
|
| 1220 |
+
"eval_loss": 0.6110680103302002,
|
| 1221 |
+
"eval_runtime": 1.0532,
|
| 1222 |
+
"eval_samples_per_second": 15.192,
|
| 1223 |
+
"eval_steps_per_second": 3.798,
|
| 1224 |
+
"step": 243
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 3.3243243243243246,
|
| 1228 |
+
"grad_norm": 2.6236143112182617,
|
| 1229 |
+
"learning_rate": 6.401646847069039e-05,
|
| 1230 |
+
"loss": 0.3107,
|
| 1231 |
+
"step": 246
|
| 1232 |
+
},
|
| 1233 |
+
{
|
| 1234 |
+
"epoch": 3.3243243243243246,
|
| 1235 |
+
"eval_loss": 0.6120755672454834,
|
| 1236 |
+
"eval_runtime": 1.0508,
|
| 1237 |
+
"eval_samples_per_second": 15.227,
|
| 1238 |
+
"eval_steps_per_second": 3.807,
|
| 1239 |
+
"step": 246
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 3.364864864864865,
|
| 1243 |
+
"grad_norm": 1.75555419921875,
|
| 1244 |
+
"learning_rate": 6.32427708017615e-05,
|
| 1245 |
+
"loss": 0.2219,
|
| 1246 |
+
"step": 249
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 3.364864864864865,
|
| 1250 |
+
"eval_loss": 0.6196171641349792,
|
| 1251 |
+
"eval_runtime": 1.0523,
|
| 1252 |
+
"eval_samples_per_second": 15.204,
|
| 1253 |
+
"eval_steps_per_second": 3.801,
|
| 1254 |
+
"step": 249
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 3.4054054054054053,
|
| 1258 |
+
"grad_norm": 3.003138303756714,
|
| 1259 |
+
"learning_rate": 6.246564769157894e-05,
|
| 1260 |
+
"loss": 0.251,
|
| 1261 |
+
"step": 252
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 3.4054054054054053,
|
| 1265 |
+
"eval_loss": 0.6273298263549805,
|
| 1266 |
+
"eval_runtime": 1.0546,
|
| 1267 |
+
"eval_samples_per_second": 15.171,
|
| 1268 |
+
"eval_steps_per_second": 3.793,
|
| 1269 |
+
"step": 252
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 3.445945945945946,
|
| 1273 |
+
"grad_norm": 2.2066917419433594,
|
| 1274 |
+
"learning_rate": 6.168530015468872e-05,
|
| 1275 |
+
"loss": 0.3366,
|
| 1276 |
+
"step": 255
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 3.445945945945946,
|
| 1280 |
+
"eval_loss": 0.6258885860443115,
|
| 1281 |
+
"eval_runtime": 1.0514,
|
| 1282 |
+
"eval_samples_per_second": 15.217,
|
| 1283 |
+
"eval_steps_per_second": 3.804,
|
| 1284 |
+
"step": 255
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 3.4864864864864864,
|
| 1288 |
+
"grad_norm": 1.7121000289916992,
|
| 1289 |
+
"learning_rate": 6.0901930039683184e-05,
|
| 1290 |
+
"loss": 0.3182,
|
| 1291 |
+
"step": 258
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 3.4864864864864864,
|
| 1295 |
+
"eval_loss": 0.6243223547935486,
|
| 1296 |
+
"eval_runtime": 1.0739,
|
| 1297 |
+
"eval_samples_per_second": 14.898,
|
| 1298 |
+
"eval_steps_per_second": 3.725,
|
| 1299 |
+
"step": 258
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 3.527027027027027,
|
| 1303 |
+
"grad_norm": 2.7600913047790527,
|
| 1304 |
+
"learning_rate": 6.011573997698985e-05,
|
| 1305 |
+
"loss": 0.4133,
|
| 1306 |
+
"step": 261
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 3.527027027027027,
|
| 1310 |
+
"eval_loss": 0.6259996294975281,
|
| 1311 |
+
"eval_runtime": 1.0561,
|
| 1312 |
+
"eval_samples_per_second": 15.151,
|
| 1313 |
+
"eval_steps_per_second": 3.788,
|
| 1314 |
+
"step": 261
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 3.5675675675675675,
|
| 1318 |
+
"grad_norm": 2.611302614212036,
|
| 1319 |
+
"learning_rate": 5.9326933326457956e-05,
|
| 1320 |
+
"loss": 0.3297,
|
| 1321 |
+
"step": 264
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 3.5675675675675675,
|
| 1325 |
+
"eval_loss": 0.6303350925445557,
|
| 1326 |
+
"eval_runtime": 1.0534,
|
| 1327 |
+
"eval_samples_per_second": 15.189,
|
| 1328 |
+
"eval_steps_per_second": 3.797,
|
| 1329 |
+
"step": 264
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 3.608108108108108,
|
| 1333 |
+
"grad_norm": 1.6527258157730103,
|
| 1334 |
+
"learning_rate": 5.8535714124756434e-05,
|
| 1335 |
+
"loss": 0.2276,
|
| 1336 |
+
"step": 267
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 3.608108108108108,
|
| 1340 |
+
"eval_loss": 0.6364917159080505,
|
| 1341 |
+
"eval_runtime": 1.052,
|
| 1342 |
+
"eval_samples_per_second": 15.209,
|
| 1343 |
+
"eval_steps_per_second": 3.802,
|
| 1344 |
+
"step": 267
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 3.6486486486486487,
|
| 1348 |
+
"grad_norm": 1.1108059883117676,
|
| 1349 |
+
"learning_rate": 5.774228703259678e-05,
|
| 1350 |
+
"loss": 0.1842,
|
| 1351 |
+
"step": 270
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 3.6486486486486487,
|
| 1355 |
+
"eval_loss": 0.6382502317428589,
|
| 1356 |
+
"eval_runtime": 1.0549,
|
| 1357 |
+
"eval_samples_per_second": 15.168,
|
| 1358 |
+
"eval_steps_per_second": 3.792,
|
| 1359 |
+
"step": 270
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 3.689189189189189,
|
| 1363 |
+
"grad_norm": 2.822380781173706,
|
| 1364 |
+
"learning_rate": 5.694685728179442e-05,
|
| 1365 |
+
"loss": 0.4961,
|
| 1366 |
+
"step": 273
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 3.689189189189189,
|
| 1370 |
+
"eval_loss": 0.6313918828964233,
|
| 1371 |
+
"eval_runtime": 1.0523,
|
| 1372 |
+
"eval_samples_per_second": 15.205,
|
| 1373 |
+
"eval_steps_per_second": 3.801,
|
| 1374 |
+
"step": 273
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 3.72972972972973,
|
| 1378 |
+
"grad_norm": 2.4894397258758545,
|
| 1379 |
+
"learning_rate": 5.6149630622182526e-05,
|
| 1380 |
+
"loss": 0.3785,
|
| 1381 |
+
"step": 276
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 3.72972972972973,
|
| 1385 |
+
"eval_loss": 0.6239753365516663,
|
| 1386 |
+
"eval_runtime": 1.053,
|
| 1387 |
+
"eval_samples_per_second": 15.195,
|
| 1388 |
+
"eval_steps_per_second": 3.799,
|
| 1389 |
+
"step": 276
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 3.77027027027027,
|
| 1393 |
+
"grad_norm": 2.1039986610412598,
|
| 1394 |
+
"learning_rate": 5.535081326839165e-05,
|
| 1395 |
+
"loss": 0.2834,
|
| 1396 |
+
"step": 279
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 3.77027027027027,
|
| 1400 |
+
"eval_loss": 0.6189073920249939,
|
| 1401 |
+
"eval_runtime": 1.0515,
|
| 1402 |
+
"eval_samples_per_second": 15.217,
|
| 1403 |
+
"eval_steps_per_second": 3.804,
|
| 1404 |
+
"step": 279
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 3.810810810810811,
|
| 1408 |
+
"grad_norm": 2.7096340656280518,
|
| 1409 |
+
"learning_rate": 5.455061184650921e-05,
|
| 1410 |
+
"loss": 0.3397,
|
| 1411 |
+
"step": 282
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 3.810810810810811,
|
| 1415 |
+
"eval_loss": 0.6138538122177124,
|
| 1416 |
+
"eval_runtime": 1.0521,
|
| 1417 |
+
"eval_samples_per_second": 15.208,
|
| 1418 |
+
"eval_steps_per_second": 3.802,
|
| 1419 |
+
"step": 282
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 3.8513513513513513,
|
| 1423 |
+
"grad_norm": 2.030907154083252,
|
| 1424 |
+
"learning_rate": 5.3749233340632674e-05,
|
| 1425 |
+
"loss": 0.2795,
|
| 1426 |
+
"step": 285
|
| 1427 |
+
},
|
| 1428 |
+
{
|
| 1429 |
+
"epoch": 3.8513513513513513,
|
| 1430 |
+
"eval_loss": 0.6104437708854675,
|
| 1431 |
+
"eval_runtime": 1.0581,
|
| 1432 |
+
"eval_samples_per_second": 15.122,
|
| 1433 |
+
"eval_steps_per_second": 3.78,
|
| 1434 |
+
"step": 285
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 3.891891891891892,
|
| 1438 |
+
"grad_norm": 2.061206340789795,
|
| 1439 |
+
"learning_rate": 5.2946885039329866e-05,
|
| 1440 |
+
"loss": 0.3114,
|
| 1441 |
+
"step": 288
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 3.891891891891892,
|
| 1445 |
+
"eval_loss": 0.6077687740325928,
|
| 1446 |
+
"eval_runtime": 1.0527,
|
| 1447 |
+
"eval_samples_per_second": 15.199,
|
| 1448 |
+
"eval_steps_per_second": 3.8,
|
| 1449 |
+
"step": 288
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 3.9324324324324325,
|
| 1453 |
+
"grad_norm": 2.062087059020996,
|
| 1454 |
+
"learning_rate": 5.2143774482020744e-05,
|
| 1455 |
+
"loss": 0.2395,
|
| 1456 |
+
"step": 291
|
| 1457 |
+
},
|
| 1458 |
+
{
|
| 1459 |
+
"epoch": 3.9324324324324325,
|
| 1460 |
+
"eval_loss": 0.6111433506011963,
|
| 1461 |
+
"eval_runtime": 1.0517,
|
| 1462 |
+
"eval_samples_per_second": 15.214,
|
| 1463 |
+
"eval_steps_per_second": 3.804,
|
| 1464 |
+
"step": 291
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 3.972972972972973,
|
| 1468 |
+
"grad_norm": 1.6344010829925537,
|
| 1469 |
+
"learning_rate": 5.134010940529429e-05,
|
| 1470 |
+
"loss": 0.1948,
|
| 1471 |
+
"step": 294
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 3.972972972972973,
|
| 1475 |
+
"eval_loss": 0.6142452955245972,
|
| 1476 |
+
"eval_runtime": 1.0529,
|
| 1477 |
+
"eval_samples_per_second": 15.196,
|
| 1478 |
+
"eval_steps_per_second": 3.799,
|
| 1479 |
+
"step": 294
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 4.013513513513513,
|
| 1483 |
+
"grad_norm": 1.9017384052276611,
|
| 1484 |
+
"learning_rate": 5.053609768917413e-05,
|
| 1485 |
+
"loss": 0.2284,
|
| 1486 |
+
"step": 297
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 4.013513513513513,
|
| 1490 |
+
"eval_loss": 0.6194114685058594,
|
| 1491 |
+
"eval_runtime": 1.0515,
|
| 1492 |
+
"eval_samples_per_second": 15.217,
|
| 1493 |
+
"eval_steps_per_second": 3.804,
|
| 1494 |
+
"step": 297
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 4.054054054054054,
|
| 1498 |
+
"grad_norm": 2.1609394550323486,
|
| 1499 |
+
"learning_rate": 4.973194730334748e-05,
|
| 1500 |
+
"loss": 0.2638,
|
| 1501 |
+
"step": 300
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 4.054054054054054,
|
| 1505 |
+
"eval_loss": 0.6303145885467529,
|
| 1506 |
+
"eval_runtime": 1.053,
|
| 1507 |
+
"eval_samples_per_second": 15.194,
|
| 1508 |
+
"eval_steps_per_second": 3.798,
|
| 1509 |
+
"step": 300
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 4.094594594594595,
|
| 1513 |
+
"grad_norm": 1.5275555849075317,
|
| 1514 |
+
"learning_rate": 4.892786625337047e-05,
|
| 1515 |
+
"loss": 0.252,
|
| 1516 |
+
"step": 303
|
| 1517 |
+
},
|
| 1518 |
+
{
|
| 1519 |
+
"epoch": 4.094594594594595,
|
| 1520 |
+
"eval_loss": 0.6517325639724731,
|
| 1521 |
+
"eval_runtime": 1.051,
|
| 1522 |
+
"eval_samples_per_second": 15.224,
|
| 1523 |
+
"eval_steps_per_second": 3.806,
|
| 1524 |
+
"step": 303
|
| 1525 |
+
},
|
| 1526 |
+
{
|
| 1527 |
+
"epoch": 4.135135135135135,
|
| 1528 |
+
"grad_norm": 2.807483434677124,
|
| 1529 |
+
"learning_rate": 4.8124062526864534e-05,
|
| 1530 |
+
"loss": 0.183,
|
| 1531 |
+
"step": 306
|
| 1532 |
+
},
|
| 1533 |
+
{
|
| 1534 |
+
"epoch": 4.135135135135135,
|
| 1535 |
+
"eval_loss": 0.6644703149795532,
|
| 1536 |
+
"eval_runtime": 1.0531,
|
| 1537 |
+
"eval_samples_per_second": 15.193,
|
| 1538 |
+
"eval_steps_per_second": 3.798,
|
| 1539 |
+
"step": 306
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 4.175675675675675,
|
| 1543 |
+
"grad_norm": 2.6279256343841553,
|
| 1544 |
+
"learning_rate": 4.7320744039717154e-05,
|
| 1545 |
+
"loss": 0.2415,
|
| 1546 |
+
"step": 309
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 4.175675675675675,
|
| 1550 |
+
"eval_loss": 0.6603893041610718,
|
| 1551 |
+
"eval_runtime": 1.0531,
|
| 1552 |
+
"eval_samples_per_second": 15.193,
|
| 1553 |
+
"eval_steps_per_second": 3.798,
|
| 1554 |
+
"step": 309
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 4.216216216216216,
|
| 1558 |
+
"grad_norm": 0.42106354236602783,
|
| 1559 |
+
"learning_rate": 4.651811858230149e-05,
|
| 1560 |
+
"loss": 0.1791,
|
| 1561 |
+
"step": 312
|
| 1562 |
+
},
|
| 1563 |
+
{
|
| 1564 |
+
"epoch": 4.216216216216216,
|
| 1565 |
+
"eval_loss": 0.652984082698822,
|
| 1566 |
+
"eval_runtime": 1.053,
|
| 1567 |
+
"eval_samples_per_second": 15.195,
|
| 1568 |
+
"eval_steps_per_second": 3.799,
|
| 1569 |
+
"step": 312
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 4.256756756756757,
|
| 1573 |
+
"grad_norm": 2.064615249633789,
|
| 1574 |
+
"learning_rate": 4.571639376572806e-05,
|
| 1575 |
+
"loss": 0.2013,
|
| 1576 |
+
"step": 315
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 4.256756756756757,
|
| 1580 |
+
"eval_loss": 0.6488903760910034,
|
| 1581 |
+
"eval_runtime": 1.0505,
|
| 1582 |
+
"eval_samples_per_second": 15.23,
|
| 1583 |
+
"eval_steps_per_second": 3.808,
|
| 1584 |
+
"step": 315
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 4.297297297297297,
|
| 1588 |
+
"grad_norm": 2.4248170852661133,
|
| 1589 |
+
"learning_rate": 4.491577696814318e-05,
|
| 1590 |
+
"loss": 0.1827,
|
| 1591 |
+
"step": 318
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 4.297297297297297,
|
| 1595 |
+
"eval_loss": 0.653176486492157,
|
| 1596 |
+
"eval_runtime": 1.0536,
|
| 1597 |
+
"eval_samples_per_second": 15.186,
|
| 1598 |
+
"eval_steps_per_second": 3.797,
|
| 1599 |
+
"step": 318
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 4.337837837837838,
|
| 1603 |
+
"grad_norm": 2.055769443511963,
|
| 1604 |
+
"learning_rate": 4.411647528108743e-05,
|
| 1605 |
+
"loss": 0.1792,
|
| 1606 |
+
"step": 321
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 4.337837837837838,
|
| 1610 |
+
"eval_loss": 0.6584765315055847,
|
| 1611 |
+
"eval_runtime": 1.052,
|
| 1612 |
+
"eval_samples_per_second": 15.209,
|
| 1613 |
+
"eval_steps_per_second": 3.802,
|
| 1614 |
+
"step": 321
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 4.378378378378378,
|
| 1618 |
+
"grad_norm": 3.4611449241638184,
|
| 1619 |
+
"learning_rate": 4.331869545592834e-05,
|
| 1620 |
+
"loss": 0.2568,
|
| 1621 |
+
"step": 324
|
| 1622 |
+
},
|
| 1623 |
+
{
|
| 1624 |
+
"epoch": 4.378378378378378,
|
| 1625 |
+
"eval_loss": 0.6628451347351074,
|
| 1626 |
+
"eval_runtime": 1.055,
|
| 1627 |
+
"eval_samples_per_second": 15.166,
|
| 1628 |
+
"eval_steps_per_second": 3.791,
|
| 1629 |
+
"step": 324
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 4.418918918918919,
|
| 1633 |
+
"grad_norm": 1.6108025312423706,
|
| 1634 |
+
"learning_rate": 4.252264385038098e-05,
|
| 1635 |
+
"loss": 0.1682,
|
| 1636 |
+
"step": 327
|
| 1637 |
+
},
|
| 1638 |
+
{
|
| 1639 |
+
"epoch": 4.418918918918919,
|
| 1640 |
+
"eval_loss": 0.66502845287323,
|
| 1641 |
+
"eval_runtime": 1.0508,
|
| 1642 |
+
"eval_samples_per_second": 15.227,
|
| 1643 |
+
"eval_steps_per_second": 3.807,
|
| 1644 |
+
"step": 327
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"epoch": 4.45945945945946,
|
| 1648 |
+
"grad_norm": 1.828131914138794,
|
| 1649 |
+
"learning_rate": 4.1728526375130614e-05,
|
| 1650 |
+
"loss": 0.25,
|
| 1651 |
+
"step": 330
|
| 1652 |
+
},
|
| 1653 |
+
{
|
| 1654 |
+
"epoch": 4.45945945945946,
|
| 1655 |
+
"eval_loss": 0.6729562282562256,
|
| 1656 |
+
"eval_runtime": 1.0534,
|
| 1657 |
+
"eval_samples_per_second": 15.189,
|
| 1658 |
+
"eval_steps_per_second": 3.797,
|
| 1659 |
+
"step": 330
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 4.5,
|
| 1663 |
+
"grad_norm": 2.5057499408721924,
|
| 1664 |
+
"learning_rate": 4.093654844057059e-05,
|
| 1665 |
+
"loss": 0.2664,
|
| 1666 |
+
"step": 333
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 4.5,
|
| 1670 |
+
"eval_loss": 0.6741403937339783,
|
| 1671 |
+
"eval_runtime": 1.052,
|
| 1672 |
+
"eval_samples_per_second": 15.209,
|
| 1673 |
+
"eval_steps_per_second": 3.802,
|
| 1674 |
+
"step": 333
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 4.54054054054054,
|
| 1678 |
+
"grad_norm": 1.6008535623550415,
|
| 1679 |
+
"learning_rate": 4.014691490367e-05,
|
| 1680 |
+
"loss": 0.2316,
|
| 1681 |
+
"step": 336
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 4.54054054054054,
|
| 1685 |
+
"eval_loss": 0.6773088574409485,
|
| 1686 |
+
"eval_runtime": 1.053,
|
| 1687 |
+
"eval_samples_per_second": 15.194,
|
| 1688 |
+
"eval_steps_per_second": 3.799,
|
| 1689 |
+
"step": 336
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 4.581081081081081,
|
| 1693 |
+
"grad_norm": 2.551591157913208,
|
| 1694 |
+
"learning_rate": 3.935983001498439e-05,
|
| 1695 |
+
"loss": 0.3467,
|
| 1696 |
+
"step": 339
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 4.581081081081081,
|
| 1700 |
+
"eval_loss": 0.6705477237701416,
|
| 1701 |
+
"eval_runtime": 1.0509,
|
| 1702 |
+
"eval_samples_per_second": 15.226,
|
| 1703 |
+
"eval_steps_per_second": 3.806,
|
| 1704 |
+
"step": 339
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 4.621621621621622,
|
| 1708 |
+
"grad_norm": 2.130202054977417,
|
| 1709 |
+
"learning_rate": 3.857549736582316e-05,
|
| 1710 |
+
"loss": 0.2426,
|
| 1711 |
+
"step": 342
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 4.621621621621622,
|
| 1715 |
+
"eval_loss": 0.6681296825408936,
|
| 1716 |
+
"eval_runtime": 1.0529,
|
| 1717 |
+
"eval_samples_per_second": 15.196,
|
| 1718 |
+
"eval_steps_per_second": 3.799,
|
| 1719 |
+
"step": 342
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 4.662162162162162,
|
| 1723 |
+
"grad_norm": 2.043670415878296,
|
| 1724 |
+
"learning_rate": 3.7794119835587685e-05,
|
| 1725 |
+
"loss": 0.2421,
|
| 1726 |
+
"step": 345
|
| 1727 |
+
},
|
| 1728 |
+
{
|
| 1729 |
+
"epoch": 4.662162162162162,
|
| 1730 |
+
"eval_loss": 0.6622060537338257,
|
| 1731 |
+
"eval_runtime": 1.0519,
|
| 1732 |
+
"eval_samples_per_second": 15.21,
|
| 1733 |
+
"eval_steps_per_second": 3.803,
|
| 1734 |
+
"step": 345
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 4.702702702702703,
|
| 1738 |
+
"grad_norm": 1.9365885257720947,
|
| 1739 |
+
"learning_rate": 3.701589953929354e-05,
|
| 1740 |
+
"loss": 0.4063,
|
| 1741 |
+
"step": 348
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 4.702702702702703,
|
| 1745 |
+
"eval_loss": 0.6608781814575195,
|
| 1746 |
+
"eval_runtime": 1.0528,
|
| 1747 |
+
"eval_samples_per_second": 15.197,
|
| 1748 |
+
"eval_steps_per_second": 3.799,
|
| 1749 |
+
"step": 348
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 4.743243243243243,
|
| 1753 |
+
"grad_norm": 2.596634864807129,
|
| 1754 |
+
"learning_rate": 3.62410377752904e-05,
|
| 1755 |
+
"loss": 0.2255,
|
| 1756 |
+
"step": 351
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"epoch": 4.743243243243243,
|
| 1760 |
+
"eval_loss": 0.6569182276725769,
|
| 1761 |
+
"eval_runtime": 1.0522,
|
| 1762 |
+
"eval_samples_per_second": 15.206,
|
| 1763 |
+
"eval_steps_per_second": 3.802,
|
| 1764 |
+
"step": 351
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 4.783783783783784,
|
| 1768 |
+
"grad_norm": 2.039332628250122,
|
| 1769 |
+
"learning_rate": 3.546973497319319e-05,
|
| 1770 |
+
"loss": 0.1933,
|
| 1771 |
+
"step": 354
|
| 1772 |
+
},
|
| 1773 |
+
{
|
| 1774 |
+
"epoch": 4.783783783783784,
|
| 1775 |
+
"eval_loss": 0.6534222364425659,
|
| 1776 |
+
"eval_runtime": 1.0498,
|
| 1777 |
+
"eval_samples_per_second": 15.241,
|
| 1778 |
+
"eval_steps_per_second": 3.81,
|
| 1779 |
+
"step": 354
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 4.824324324324325,
|
| 1783 |
+
"grad_norm": 1.994629144668579,
|
| 1784 |
+
"learning_rate": 3.4702190642037944e-05,
|
| 1785 |
+
"loss": 0.1975,
|
| 1786 |
+
"step": 357
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 4.824324324324325,
|
| 1790 |
+
"eval_loss": 0.649687647819519,
|
| 1791 |
+
"eval_runtime": 1.0523,
|
| 1792 |
+
"eval_samples_per_second": 15.204,
|
| 1793 |
+
"eval_steps_per_second": 3.801,
|
| 1794 |
+
"step": 357
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 4.864864864864865,
|
| 1798 |
+
"grad_norm": 2.154684543609619,
|
| 1799 |
+
"learning_rate": 3.393860331867589e-05,
|
| 1800 |
+
"loss": 0.3065,
|
| 1801 |
+
"step": 360
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 4.864864864864865,
|
| 1805 |
+
"eval_loss": 0.6491411924362183,
|
| 1806 |
+
"eval_runtime": 1.0519,
|
| 1807 |
+
"eval_samples_per_second": 15.21,
|
| 1808 |
+
"eval_steps_per_second": 3.803,
|
| 1809 |
+
"step": 360
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 4.905405405405405,
|
| 1813 |
+
"grad_norm": 1.61858069896698,
|
| 1814 |
+
"learning_rate": 3.317917051641877e-05,
|
| 1815 |
+
"loss": 0.1641,
|
| 1816 |
+
"step": 363
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 4.905405405405405,
|
| 1820 |
+
"eval_loss": 0.651297926902771,
|
| 1821 |
+
"eval_runtime": 1.0521,
|
| 1822 |
+
"eval_samples_per_second": 15.208,
|
| 1823 |
+
"eval_steps_per_second": 3.802,
|
| 1824 |
+
"step": 363
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 4.945945945945946,
|
| 1828 |
+
"grad_norm": 2.7362637519836426,
|
| 1829 |
+
"learning_rate": 3.242408867394919e-05,
|
| 1830 |
+
"loss": 0.2032,
|
| 1831 |
+
"step": 366
|
| 1832 |
+
},
|
| 1833 |
+
{
|
| 1834 |
+
"epoch": 4.945945945945946,
|
| 1835 |
+
"eval_loss": 0.6552869081497192,
|
| 1836 |
+
"eval_runtime": 1.0506,
|
| 1837 |
+
"eval_samples_per_second": 15.229,
|
| 1838 |
+
"eval_steps_per_second": 3.807,
|
| 1839 |
+
"step": 366
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 4.986486486486487,
|
| 1843 |
+
"grad_norm": 2.0567097663879395,
|
| 1844 |
+
"learning_rate": 3.167355310450877e-05,
|
| 1845 |
+
"loss": 0.1886,
|
| 1846 |
+
"step": 369
|
| 1847 |
+
},
|
| 1848 |
+
{
|
| 1849 |
+
"epoch": 4.986486486486487,
|
| 1850 |
+
"eval_loss": 0.6590157747268677,
|
| 1851 |
+
"eval_runtime": 1.0528,
|
| 1852 |
+
"eval_samples_per_second": 15.197,
|
| 1853 |
+
"eval_steps_per_second": 3.799,
|
| 1854 |
+
"step": 369
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 5.027027027027027,
|
| 1858 |
+
"grad_norm": 1.5418853759765625,
|
| 1859 |
+
"learning_rate": 3.092775794537741e-05,
|
| 1860 |
+
"loss": 0.2539,
|
| 1861 |
+
"step": 372
|
| 1862 |
+
},
|
| 1863 |
+
{
|
| 1864 |
+
"epoch": 5.027027027027027,
|
| 1865 |
+
"eval_loss": 0.6676727533340454,
|
| 1866 |
+
"eval_runtime": 1.0516,
|
| 1867 |
+
"eval_samples_per_second": 15.215,
|
| 1868 |
+
"eval_steps_per_second": 3.804,
|
| 1869 |
+
"step": 372
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 5.0675675675675675,
|
| 1873 |
+
"grad_norm": 1.229972004890442,
|
| 1874 |
+
"learning_rate": 3.0186896107656803e-05,
|
| 1875 |
+
"loss": 0.1464,
|
| 1876 |
+
"step": 375
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 5.0675675675675675,
|
| 1880 |
+
"eval_loss": 0.687861979007721,
|
| 1881 |
+
"eval_runtime": 1.0539,
|
| 1882 |
+
"eval_samples_per_second": 15.182,
|
| 1883 |
+
"eval_steps_per_second": 3.796,
|
| 1884 |
+
"step": 375
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 5.108108108108108,
|
| 1888 |
+
"grad_norm": 2.421496868133545,
|
| 1889 |
+
"learning_rate": 2.9451159226371095e-05,
|
| 1890 |
+
"loss": 0.2295,
|
| 1891 |
+
"step": 378
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 5.108108108108108,
|
| 1895 |
+
"eval_loss": 0.7066453695297241,
|
| 1896 |
+
"eval_runtime": 1.0503,
|
| 1897 |
+
"eval_samples_per_second": 15.233,
|
| 1898 |
+
"eval_steps_per_second": 3.808,
|
| 1899 |
+
"step": 378
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 5.148648648648648,
|
| 1903 |
+
"grad_norm": 2.3475804328918457,
|
| 1904 |
+
"learning_rate": 2.8720737610897575e-05,
|
| 1905 |
+
"loss": 0.1438,
|
| 1906 |
+
"step": 381
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"epoch": 5.148648648648648,
|
| 1910 |
+
"eval_loss": 0.7166962623596191,
|
| 1911 |
+
"eval_runtime": 1.0534,
|
| 1912 |
+
"eval_samples_per_second": 15.189,
|
| 1913 |
+
"eval_steps_per_second": 3.797,
|
| 1914 |
+
"step": 381
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 5.1891891891891895,
|
| 1918 |
+
"grad_norm": 2.2746946811676025,
|
| 1919 |
+
"learning_rate": 2.799582019574033e-05,
|
| 1920 |
+
"loss": 0.1603,
|
| 1921 |
+
"step": 384
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 5.1891891891891895,
|
| 1925 |
+
"eval_loss": 0.7134541273117065,
|
| 1926 |
+
"eval_runtime": 1.0519,
|
| 1927 |
+
"eval_samples_per_second": 15.211,
|
| 1928 |
+
"eval_steps_per_second": 3.803,
|
| 1929 |
+
"step": 384
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 5.22972972972973,
|
| 1933 |
+
"grad_norm": 1.2550048828125,
|
| 1934 |
+
"learning_rate": 2.7276594491659525e-05,
|
| 1935 |
+
"loss": 0.1379,
|
| 1936 |
+
"step": 387
|
| 1937 |
+
},
|
| 1938 |
+
{
|
| 1939 |
+
"epoch": 5.22972972972973,
|
| 1940 |
+
"eval_loss": 0.7095359563827515,
|
| 1941 |
+
"eval_runtime": 1.0543,
|
| 1942 |
+
"eval_samples_per_second": 15.176,
|
| 1943 |
+
"eval_steps_per_second": 3.794,
|
| 1944 |
+
"step": 387
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 5.27027027027027,
|
| 1948 |
+
"grad_norm": 1.7738205194473267,
|
| 1949 |
+
"learning_rate": 2.656324653716884e-05,
|
| 1950 |
+
"loss": 0.2783,
|
| 1951 |
+
"step": 390
|
| 1952 |
+
},
|
| 1953 |
+
{
|
| 1954 |
+
"epoch": 5.27027027027027,
|
| 1955 |
+
"eval_loss": 0.7103461623191833,
|
| 1956 |
+
"eval_runtime": 1.0515,
|
| 1957 |
+
"eval_samples_per_second": 15.216,
|
| 1958 |
+
"eval_steps_per_second": 3.804,
|
| 1959 |
+
"step": 390
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 5.3108108108108105,
|
| 1963 |
+
"grad_norm": 2.2887580394744873,
|
| 1964 |
+
"learning_rate": 2.5855960850413935e-05,
|
| 1965 |
+
"loss": 0.1575,
|
| 1966 |
+
"step": 393
|
| 1967 |
+
},
|
| 1968 |
+
{
|
| 1969 |
+
"epoch": 5.3108108108108105,
|
| 1970 |
+
"eval_loss": 0.7042403817176819,
|
| 1971 |
+
"eval_runtime": 1.0523,
|
| 1972 |
+
"eval_samples_per_second": 15.204,
|
| 1973 |
+
"eval_steps_per_second": 3.801,
|
| 1974 |
+
"step": 393
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 5.351351351351352,
|
| 1978 |
+
"grad_norm": 2.6281135082244873,
|
| 1979 |
+
"learning_rate": 2.5154920381444025e-05,
|
| 1980 |
+
"loss": 0.1743,
|
| 1981 |
+
"step": 396
|
| 1982 |
+
},
|
| 1983 |
+
{
|
| 1984 |
+
"epoch": 5.351351351351352,
|
| 1985 |
+
"eval_loss": 0.7114053964614868,
|
| 1986 |
+
"eval_runtime": 1.0527,
|
| 1987 |
+
"eval_samples_per_second": 15.199,
|
| 1988 |
+
"eval_steps_per_second": 3.8,
|
| 1989 |
+
"step": 396
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 5.391891891891892,
|
| 1993 |
+
"grad_norm": 1.8125991821289062,
|
| 1994 |
+
"learning_rate": 2.4460306464889022e-05,
|
| 1995 |
+
"loss": 0.1168,
|
| 1996 |
+
"step": 399
|
| 1997 |
+
},
|
| 1998 |
+
{
|
| 1999 |
+
"epoch": 5.391891891891892,
|
| 2000 |
+
"eval_loss": 0.7083012461662292,
|
| 2001 |
+
"eval_runtime": 1.0506,
|
| 2002 |
+
"eval_samples_per_second": 15.23,
|
| 2003 |
+
"eval_steps_per_second": 3.807,
|
| 2004 |
+
"step": 399
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 5.4324324324324325,
|
| 2008 |
+
"grad_norm": 2.5157058238983154,
|
| 2009 |
+
"learning_rate": 2.3772298773054757e-05,
|
| 2010 |
+
"loss": 0.284,
|
| 2011 |
+
"step": 402
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 5.4324324324324325,
|
| 2015 |
+
"eval_loss": 0.7072416543960571,
|
| 2016 |
+
"eval_runtime": 1.0524,
|
| 2017 |
+
"eval_samples_per_second": 15.204,
|
| 2018 |
+
"eval_steps_per_second": 3.801,
|
| 2019 |
+
"step": 402
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 5.472972972972973,
|
| 2023 |
+
"grad_norm": 0.8739199042320251,
|
| 2024 |
+
"learning_rate": 2.309107526944792e-05,
|
| 2025 |
+
"loss": 0.1013,
|
| 2026 |
+
"step": 405
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 5.472972972972973,
|
| 2030 |
+
"eval_loss": 0.7062889933586121,
|
| 2031 |
+
"eval_runtime": 1.051,
|
| 2032 |
+
"eval_samples_per_second": 15.223,
|
| 2033 |
+
"eval_steps_per_second": 3.806,
|
| 2034 |
+
"step": 405
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 5.513513513513513,
|
| 2038 |
+
"grad_norm": 2.2809295654296875,
|
| 2039 |
+
"learning_rate": 2.2416812162743223e-05,
|
| 2040 |
+
"loss": 0.2612,
|
| 2041 |
+
"step": 408
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 5.513513513513513,
|
| 2045 |
+
"eval_loss": 0.70506751537323,
|
| 2046 |
+
"eval_runtime": 1.053,
|
| 2047 |
+
"eval_samples_per_second": 15.195,
|
| 2048 |
+
"eval_steps_per_second": 3.799,
|
| 2049 |
+
"step": 408
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 5.554054054054054,
|
| 2053 |
+
"grad_norm": 2.2030365467071533,
|
| 2054 |
+
"learning_rate": 2.17496838612043e-05,
|
| 2055 |
+
"loss": 0.1343,
|
| 2056 |
+
"step": 411
|
| 2057 |
+
},
|
| 2058 |
+
{
|
| 2059 |
+
"epoch": 5.554054054054054,
|
| 2060 |
+
"eval_loss": 0.7102519273757935,
|
| 2061 |
+
"eval_runtime": 1.0534,
|
| 2062 |
+
"eval_samples_per_second": 15.188,
|
| 2063 |
+
"eval_steps_per_second": 3.797,
|
| 2064 |
+
"step": 411
|
| 2065 |
+
},
|
| 2066 |
+
{
|
| 2067 |
+
"epoch": 5.594594594594595,
|
| 2068 |
+
"grad_norm": 1.4592159986495972,
|
| 2069 |
+
"learning_rate": 2.1089862927570475e-05,
|
| 2070 |
+
"loss": 0.1009,
|
| 2071 |
+
"step": 414
|
| 2072 |
+
},
|
| 2073 |
+
{
|
| 2074 |
+
"epoch": 5.594594594594595,
|
| 2075 |
+
"eval_loss": 0.7105306386947632,
|
| 2076 |
+
"eval_runtime": 1.0533,
|
| 2077 |
+
"eval_samples_per_second": 15.19,
|
| 2078 |
+
"eval_steps_per_second": 3.797,
|
| 2079 |
+
"step": 414
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 5.635135135135135,
|
| 2083 |
+
"grad_norm": 2.2018954753875732,
|
| 2084 |
+
"learning_rate": 2.0437520034420776e-05,
|
| 2085 |
+
"loss": 0.3127,
|
| 2086 |
+
"step": 417
|
| 2087 |
+
},
|
| 2088 |
+
{
|
| 2089 |
+
"epoch": 5.635135135135135,
|
| 2090 |
+
"eval_loss": 0.7089606523513794,
|
| 2091 |
+
"eval_runtime": 1.0533,
|
| 2092 |
+
"eval_samples_per_second": 15.191,
|
| 2093 |
+
"eval_steps_per_second": 3.798,
|
| 2094 |
+
"step": 417
|
| 2095 |
+
},
|
| 2096 |
+
{
|
| 2097 |
+
"epoch": 5.675675675675675,
|
| 2098 |
+
"grad_norm": 1.8359624147415161,
|
| 2099 |
+
"learning_rate": 1.979282392002691e-05,
|
| 2100 |
+
"loss": 0.1355,
|
| 2101 |
+
"step": 420
|
| 2102 |
+
},
|
| 2103 |
+
{
|
| 2104 |
+
"epoch": 5.675675675675675,
|
| 2105 |
+
"eval_loss": 0.7059516906738281,
|
| 2106 |
+
"eval_runtime": 1.0526,
|
| 2107 |
+
"eval_samples_per_second": 15.201,
|
| 2108 |
+
"eval_steps_per_second": 3.8,
|
| 2109 |
+
"step": 420
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 5.716216216216216,
|
| 2113 |
+
"grad_norm": 2.3145079612731934,
|
| 2114 |
+
"learning_rate": 1.9155941344706546e-05,
|
| 2115 |
+
"loss": 0.1345,
|
| 2116 |
+
"step": 423
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 5.716216216216216,
|
| 2120 |
+
"eval_loss": 0.705683171749115,
|
| 2121 |
+
"eval_runtime": 1.0519,
|
| 2122 |
+
"eval_samples_per_second": 15.21,
|
| 2123 |
+
"eval_steps_per_second": 3.802,
|
| 2124 |
+
"step": 423
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 5.756756756756757,
|
| 2128 |
+
"grad_norm": 1.7434961795806885,
|
| 2129 |
+
"learning_rate": 1.852703704768842e-05,
|
| 2130 |
+
"loss": 0.1865,
|
| 2131 |
+
"step": 426
|
| 2132 |
+
},
|
| 2133 |
+
{
|
| 2134 |
+
"epoch": 5.756756756756757,
|
| 2135 |
+
"eval_loss": 0.7038547396659851,
|
| 2136 |
+
"eval_runtime": 1.0535,
|
| 2137 |
+
"eval_samples_per_second": 15.188,
|
| 2138 |
+
"eval_steps_per_second": 3.797,
|
| 2139 |
+
"step": 426
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 5.797297297297297,
|
| 2143 |
+
"grad_norm": 1.5850327014923096,
|
| 2144 |
+
"learning_rate": 1.7906273704499845e-05,
|
| 2145 |
+
"loss": 0.119,
|
| 2146 |
+
"step": 429
|
| 2147 |
+
},
|
| 2148 |
+
{
|
| 2149 |
+
"epoch": 5.797297297297297,
|
| 2150 |
+
"eval_loss": 0.7066537737846375,
|
| 2151 |
+
"eval_runtime": 1.0521,
|
| 2152 |
+
"eval_samples_per_second": 15.208,
|
| 2153 |
+
"eval_steps_per_second": 3.802,
|
| 2154 |
+
"step": 429
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"epoch": 5.837837837837838,
|
| 2158 |
+
"grad_norm": 1.599552035331726,
|
| 2159 |
+
"learning_rate": 1.7293811884888344e-05,
|
| 2160 |
+
"loss": 0.149,
|
| 2161 |
+
"step": 432
|
| 2162 |
+
},
|
| 2163 |
+
{
|
| 2164 |
+
"epoch": 5.837837837837838,
|
| 2165 |
+
"eval_loss": 0.7120293974876404,
|
| 2166 |
+
"eval_runtime": 1.0536,
|
| 2167 |
+
"eval_samples_per_second": 15.185,
|
| 2168 |
+
"eval_steps_per_second": 3.796,
|
| 2169 |
+
"step": 432
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 5.878378378378378,
|
| 2173 |
+
"grad_norm": 1.8353303670883179,
|
| 2174 |
+
"learning_rate": 1.6689810011287932e-05,
|
| 2175 |
+
"loss": 0.1748,
|
| 2176 |
+
"step": 435
|
| 2177 |
+
},
|
| 2178 |
+
{
|
| 2179 |
+
"epoch": 5.878378378378378,
|
| 2180 |
+
"eval_loss": 0.7123138308525085,
|
| 2181 |
+
"eval_runtime": 1.0524,
|
| 2182 |
+
"eval_samples_per_second": 15.203,
|
| 2183 |
+
"eval_steps_per_second": 3.801,
|
| 2184 |
+
"step": 435
|
| 2185 |
+
},
|
| 2186 |
+
{
|
| 2187 |
+
"epoch": 5.918918918918919,
|
| 2188 |
+
"grad_norm": 1.4937026500701904,
|
| 2189 |
+
"learning_rate": 1.6094424317840723e-05,
|
| 2190 |
+
"loss": 0.1781,
|
| 2191 |
+
"step": 438
|
| 2192 |
+
},
|
| 2193 |
+
{
|
| 2194 |
+
"epoch": 5.918918918918919,
|
| 2195 |
+
"eval_loss": 0.7113088965415955,
|
| 2196 |
+
"eval_runtime": 1.0528,
|
| 2197 |
+
"eval_samples_per_second": 15.198,
|
| 2198 |
+
"eval_steps_per_second": 3.799,
|
| 2199 |
+
"step": 438
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 5.95945945945946,
|
| 2203 |
+
"grad_norm": 2.0092716217041016,
|
| 2204 |
+
"learning_rate": 1.550780880998456e-05,
|
| 2205 |
+
"loss": 0.2075,
|
| 2206 |
+
"step": 441
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 5.95945945945946,
|
| 2210 |
+
"eval_loss": 0.7117879390716553,
|
| 2211 |
+
"eval_runtime": 1.0532,
|
| 2212 |
+
"eval_samples_per_second": 15.192,
|
| 2213 |
+
"eval_steps_per_second": 3.798,
|
| 2214 |
+
"step": 441
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 6.0,
|
| 2218 |
+
"grad_norm": 2.762338161468506,
|
| 2219 |
+
"learning_rate": 1.4930115224617353e-05,
|
| 2220 |
+
"loss": 0.1591,
|
| 2221 |
+
"step": 444
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 6.0,
|
| 2225 |
+
"eval_loss": 0.7111848592758179,
|
| 2226 |
+
"eval_runtime": 1.0522,
|
| 2227 |
+
"eval_samples_per_second": 15.206,
|
| 2228 |
+
"eval_steps_per_second": 3.801,
|
| 2229 |
+
"step": 444
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 6.04054054054054,
|
| 2233 |
+
"grad_norm": 1.825244665145874,
|
| 2234 |
+
"learning_rate": 1.436149299084789e-05,
|
| 2235 |
+
"loss": 0.1224,
|
| 2236 |
+
"step": 447
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"epoch": 6.04054054054054,
|
| 2240 |
+
"eval_loss": 0.7117843627929688,
|
| 2241 |
+
"eval_runtime": 1.0529,
|
| 2242 |
+
"eval_samples_per_second": 15.195,
|
| 2243 |
+
"eval_steps_per_second": 3.799,
|
| 2244 |
+
"step": 447
|
| 2245 |
+
},
|
| 2246 |
+
{
|
| 2247 |
+
"epoch": 6.081081081081081,
|
| 2248 |
+
"grad_norm": 0.9274085760116577,
|
| 2249 |
+
"learning_rate": 1.380208919134392e-05,
|
| 2250 |
+
"loss": 0.2234,
|
| 2251 |
+
"step": 450
|
| 2252 |
+
},
|
| 2253 |
+
{
|
| 2254 |
+
"epoch": 6.081081081081081,
|
| 2255 |
+
"eval_loss": 0.7170644402503967,
|
| 2256 |
+
"eval_runtime": 1.0513,
|
| 2257 |
+
"eval_samples_per_second": 15.219,
|
| 2258 |
+
"eval_steps_per_second": 3.805,
|
| 2259 |
+
"step": 450
|
| 2260 |
+
},
|
| 2261 |
+
{
|
| 2262 |
+
"epoch": 6.121621621621622,
|
| 2263 |
+
"grad_norm": 1.5220532417297363,
|
| 2264 |
+
"learning_rate": 1.3252048524286842e-05,
|
| 2265 |
+
"loss": 0.1165,
|
| 2266 |
+
"step": 453
|
| 2267 |
+
},
|
| 2268 |
+
{
|
| 2269 |
+
"epoch": 6.121621621621622,
|
| 2270 |
+
"eval_loss": 0.7227377891540527,
|
| 2271 |
+
"eval_runtime": 1.0532,
|
| 2272 |
+
"eval_samples_per_second": 15.191,
|
| 2273 |
+
"eval_steps_per_second": 3.798,
|
| 2274 |
+
"step": 453
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 6.162162162162162,
|
| 2278 |
+
"grad_norm": 1.669662594795227,
|
| 2279 |
+
"learning_rate": 1.271151326594352e-05,
|
| 2280 |
+
"loss": 0.2518,
|
| 2281 |
+
"step": 456
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 6.162162162162162,
|
| 2285 |
+
"eval_loss": 0.7325636148452759,
|
| 2286 |
+
"eval_runtime": 1.0523,
|
| 2287 |
+
"eval_samples_per_second": 15.205,
|
| 2288 |
+
"eval_steps_per_second": 3.801,
|
| 2289 |
+
"step": 456
|
| 2290 |
+
},
|
| 2291 |
+
{
|
| 2292 |
+
"epoch": 6.202702702702703,
|
| 2293 |
+
"grad_norm": 1.6538748741149902,
|
| 2294 |
+
"learning_rate": 1.2180623233864253e-05,
|
| 2295 |
+
"loss": 0.1288,
|
| 2296 |
+
"step": 459
|
| 2297 |
+
},
|
| 2298 |
+
{
|
| 2299 |
+
"epoch": 6.202702702702703,
|
| 2300 |
+
"eval_loss": 0.7430564165115356,
|
| 2301 |
+
"eval_runtime": 1.0597,
|
| 2302 |
+
"eval_samples_per_second": 15.099,
|
| 2303 |
+
"eval_steps_per_second": 3.775,
|
| 2304 |
+
"step": 459
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 6.243243243243243,
|
| 2308 |
+
"grad_norm": 1.5836577415466309,
|
| 2309 |
+
"learning_rate": 1.1659515750716955e-05,
|
| 2310 |
+
"loss": 0.1176,
|
| 2311 |
+
"step": 462
|
| 2312 |
+
},
|
| 2313 |
+
{
|
| 2314 |
+
"epoch": 6.243243243243243,
|
| 2315 |
+
"eval_loss": 0.7481391429901123,
|
| 2316 |
+
"eval_runtime": 1.0512,
|
| 2317 |
+
"eval_samples_per_second": 15.221,
|
| 2318 |
+
"eval_steps_per_second": 3.805,
|
| 2319 |
+
"step": 462
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 6.283783783783784,
|
| 2323 |
+
"grad_norm": 1.0982418060302734,
|
| 2324 |
+
"learning_rate": 1.1148325608766585e-05,
|
| 2325 |
+
"loss": 0.1231,
|
| 2326 |
+
"step": 465
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"epoch": 6.283783783783784,
|
| 2330 |
+
"eval_loss": 0.7511347532272339,
|
| 2331 |
+
"eval_runtime": 1.0552,
|
| 2332 |
+
"eval_samples_per_second": 15.163,
|
| 2333 |
+
"eval_steps_per_second": 3.791,
|
| 2334 |
+
"step": 465
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 6.324324324324325,
|
| 2338 |
+
"grad_norm": 1.9232176542282104,
|
| 2339 |
+
"learning_rate": 1.0647185035009038e-05,
|
| 2340 |
+
"loss": 0.146,
|
| 2341 |
+
"step": 468
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 6.324324324324325,
|
| 2345 |
+
"eval_loss": 0.7529792785644531,
|
| 2346 |
+
"eval_runtime": 1.0535,
|
| 2347 |
+
"eval_samples_per_second": 15.188,
|
| 2348 |
+
"eval_steps_per_second": 3.797,
|
| 2349 |
+
"step": 468
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 6.364864864864865,
|
| 2353 |
+
"grad_norm": 2.5786333084106445,
|
| 2354 |
+
"learning_rate": 1.0156223656968694e-05,
|
| 2355 |
+
"loss": 0.1169,
|
| 2356 |
+
"step": 471
|
| 2357 |
+
},
|
| 2358 |
+
{
|
| 2359 |
+
"epoch": 6.364864864864865,
|
| 2360 |
+
"eval_loss": 0.7518468499183655,
|
| 2361 |
+
"eval_runtime": 1.0523,
|
| 2362 |
+
"eval_samples_per_second": 15.205,
|
| 2363 |
+
"eval_steps_per_second": 3.801,
|
| 2364 |
+
"step": 471
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"epoch": 6.405405405405405,
|
| 2368 |
+
"grad_norm": 1.4718759059906006,
|
| 2369 |
+
"learning_rate": 9.675568469168388e-06,
|
| 2370 |
+
"loss": 0.1048,
|
| 2371 |
+
"step": 474
|
| 2372 |
+
},
|
| 2373 |
+
{
|
| 2374 |
+
"epoch": 6.405405405405405,
|
| 2375 |
+
"eval_loss": 0.7540909051895142,
|
| 2376 |
+
"eval_runtime": 1.049,
|
| 2377 |
+
"eval_samples_per_second": 15.253,
|
| 2378 |
+
"eval_steps_per_second": 3.813,
|
| 2379 |
+
"step": 474
|
| 2380 |
+
},
|
| 2381 |
+
{
|
| 2382 |
+
"epoch": 6.445945945945946,
|
| 2383 |
+
"grad_norm": 1.3492368459701538,
|
| 2384 |
+
"learning_rate": 9.205343800280219e-06,
|
| 2385 |
+
"loss": 0.1092,
|
| 2386 |
+
"step": 477
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"epoch": 6.445945945945946,
|
| 2390 |
+
"eval_loss": 0.750686764717102,
|
| 2391 |
+
"eval_runtime": 1.0533,
|
| 2392 |
+
"eval_samples_per_second": 15.19,
|
| 2393 |
+
"eval_steps_per_second": 3.798,
|
| 2394 |
+
"step": 477
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"epoch": 6.486486486486487,
|
| 2398 |
+
"grad_norm": 2.10587739944458,
|
| 2399 |
+
"learning_rate": 8.745671280966177e-06,
|
| 2400 |
+
"loss": 0.1458,
|
| 2401 |
+
"step": 480
|
| 2402 |
+
},
|
| 2403 |
+
{
|
| 2404 |
+
"epoch": 6.486486486486487,
|
| 2405 |
+
"eval_loss": 0.7518497705459595,
|
| 2406 |
+
"eval_runtime": 1.0499,
|
| 2407 |
+
"eval_samples_per_second": 15.239,
|
| 2408 |
+
"eval_steps_per_second": 3.81,
|
| 2409 |
+
"step": 480
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 6.527027027027027,
|
| 2413 |
+
"grad_norm": 0.8871177434921265,
|
| 2414 |
+
"learning_rate": 8.296669812416547e-06,
|
| 2415 |
+
"loss": 0.2177,
|
| 2416 |
+
"step": 483
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 6.527027027027027,
|
| 2420 |
+
"eval_loss": 0.7509324550628662,
|
| 2421 |
+
"eval_runtime": 1.0528,
|
| 2422 |
+
"eval_samples_per_second": 15.198,
|
| 2423 |
+
"eval_steps_per_second": 3.8,
|
| 2424 |
+
"step": 483
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 6.5675675675675675,
|
| 2428 |
+
"grad_norm": 1.299116611480713,
|
| 2429 |
+
"learning_rate": 7.858455535594306e-06,
|
| 2430 |
+
"loss": 0.1585,
|
| 2431 |
+
"step": 486
|
| 2432 |
+
},
|
| 2433 |
+
{
|
| 2434 |
+
"epoch": 6.5675675675675675,
|
| 2435 |
+
"eval_loss": 0.7509753108024597,
|
| 2436 |
+
"eval_runtime": 1.0507,
|
| 2437 |
+
"eval_samples_per_second": 15.228,
|
| 2438 |
+
"eval_steps_per_second": 3.807,
|
| 2439 |
+
"step": 486
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 6.608108108108108,
|
| 2443 |
+
"grad_norm": 1.8996071815490723,
|
| 2444 |
+
"learning_rate": 7.431141801193508e-06,
|
| 2445 |
+
"loss": 0.1337,
|
| 2446 |
+
"step": 489
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 6.608108108108108,
|
| 2450 |
+
"eval_loss": 0.7546273469924927,
|
| 2451 |
+
"eval_runtime": 1.0538,
|
| 2452 |
+
"eval_samples_per_second": 15.183,
|
| 2453 |
+
"eval_steps_per_second": 3.796,
|
| 2454 |
+
"step": 489
|
| 2455 |
+
},
|
| 2456 |
+
{
|
| 2457 |
+
"epoch": 6.648648648648649,
|
| 2458 |
+
"grad_norm": 2.193199634552002,
|
| 2459 |
+
"learning_rate": 7.014839140319485e-06,
|
| 2460 |
+
"loss": 0.122,
|
| 2461 |
+
"step": 492
|
| 2462 |
+
},
|
| 2463 |
+
{
|
| 2464 |
+
"epoch": 6.648648648648649,
|
| 2465 |
+
"eval_loss": 0.7523775100708008,
|
| 2466 |
+
"eval_runtime": 1.0517,
|
| 2467 |
+
"eval_samples_per_second": 15.213,
|
| 2468 |
+
"eval_steps_per_second": 3.803,
|
| 2469 |
+
"step": 492
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 6.6891891891891895,
|
| 2473 |
+
"grad_norm": 1.310517430305481,
|
| 2474 |
+
"learning_rate": 6.609655235898227e-06,
|
| 2475 |
+
"loss": 0.0793,
|
| 2476 |
+
"step": 495
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 6.6891891891891895,
|
| 2480 |
+
"eval_loss": 0.7553800344467163,
|
| 2481 |
+
"eval_runtime": 1.0524,
|
| 2482 |
+
"eval_samples_per_second": 15.203,
|
| 2483 |
+
"eval_steps_per_second": 3.801,
|
| 2484 |
+
"step": 495
|
| 2485 |
+
},
|
| 2486 |
+
{
|
| 2487 |
+
"epoch": 6.72972972972973,
|
| 2488 |
+
"grad_norm": 1.7615861892700195,
|
| 2489 |
+
"learning_rate": 6.215694894822699e-06,
|
| 2490 |
+
"loss": 0.1544,
|
| 2491 |
+
"step": 498
|
| 2492 |
+
},
|
| 2493 |
+
{
|
| 2494 |
+
"epoch": 6.72972972972973,
|
| 2495 |
+
"eval_loss": 0.7521288394927979,
|
| 2496 |
+
"eval_runtime": 1.0505,
|
| 2497 |
+
"eval_samples_per_second": 15.231,
|
| 2498 |
+
"eval_steps_per_second": 3.808,
|
| 2499 |
+
"step": 498
|
| 2500 |
+
},
|
| 2501 |
+
{
|
| 2502 |
+
"epoch": 6.77027027027027,
|
| 2503 |
+
"grad_norm": 1.4952490329742432,
|
| 2504 |
+
"learning_rate": 5.83306002084284e-06,
|
| 2505 |
+
"loss": 0.1387,
|
| 2506 |
+
"step": 501
|
| 2507 |
+
},
|
| 2508 |
+
{
|
| 2509 |
+
"epoch": 6.77027027027027,
|
| 2510 |
+
"eval_loss": 0.7528640627861023,
|
| 2511 |
+
"eval_runtime": 1.052,
|
| 2512 |
+
"eval_samples_per_second": 15.209,
|
| 2513 |
+
"eval_steps_per_second": 3.802,
|
| 2514 |
+
"step": 501
|
| 2515 |
+
},
|
| 2516 |
+
{
|
| 2517 |
+
"epoch": 6.8108108108108105,
|
| 2518 |
+
"grad_norm": 1.7409045696258545,
|
| 2519 |
+
"learning_rate": 5.461849588206724e-06,
|
| 2520 |
+
"loss": 0.1253,
|
| 2521 |
+
"step": 504
|
| 2522 |
+
},
|
| 2523 |
+
{
|
| 2524 |
+
"epoch": 6.8108108108108105,
|
| 2525 |
+
"eval_loss": 0.7528926134109497,
|
| 2526 |
+
"eval_runtime": 1.059,
|
| 2527 |
+
"eval_samples_per_second": 15.108,
|
| 2528 |
+
"eval_steps_per_second": 3.777,
|
| 2529 |
+
"step": 504
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 6.851351351351351,
|
| 2533 |
+
"grad_norm": 0.7362686395645142,
|
| 2534 |
+
"learning_rate": 5.102159616059365e-06,
|
| 2535 |
+
"loss": 0.1296,
|
| 2536 |
+
"step": 507
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 6.851351351351351,
|
| 2540 |
+
"eval_loss": 0.7542049884796143,
|
| 2541 |
+
"eval_runtime": 1.0521,
|
| 2542 |
+
"eval_samples_per_second": 15.207,
|
| 2543 |
+
"eval_steps_per_second": 3.802,
|
| 2544 |
+
"step": 507
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 6.891891891891892,
|
| 2548 |
+
"grad_norm": 0.806505560874939,
|
| 2549 |
+
"learning_rate": 4.754083143605869e-06,
|
| 2550 |
+
"loss": 0.1094,
|
| 2551 |
+
"step": 510
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 6.891891891891892,
|
| 2555 |
+
"eval_loss": 0.7515612840652466,
|
| 2556 |
+
"eval_runtime": 1.0559,
|
| 2557 |
+
"eval_samples_per_second": 15.152,
|
| 2558 |
+
"eval_steps_per_second": 3.788,
|
| 2559 |
+
"step": 510
|
| 2560 |
+
},
|
| 2561 |
+
{
|
| 2562 |
+
"epoch": 6.9324324324324325,
|
| 2563 |
+
"grad_norm": 1.5709373950958252,
|
| 2564 |
+
"learning_rate": 4.417710206045533e-06,
|
| 2565 |
+
"loss": 0.1009,
|
| 2566 |
+
"step": 513
|
| 2567 |
+
},
|
| 2568 |
+
{
|
| 2569 |
+
"epoch": 6.9324324324324325,
|
| 2570 |
+
"eval_loss": 0.751240611076355,
|
| 2571 |
+
"eval_runtime": 1.0523,
|
| 2572 |
+
"eval_samples_per_second": 15.205,
|
| 2573 |
+
"eval_steps_per_second": 3.801,
|
| 2574 |
+
"step": 513
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 6.972972972972973,
|
| 2578 |
+
"grad_norm": 1.2641761302947998,
|
| 2579 |
+
"learning_rate": 4.093127811282821e-06,
|
| 2580 |
+
"loss": 0.1871,
|
| 2581 |
+
"step": 516
|
| 2582 |
+
},
|
| 2583 |
+
{
|
| 2584 |
+
"epoch": 6.972972972972973,
|
| 2585 |
+
"eval_loss": 0.7525576949119568,
|
| 2586 |
+
"eval_runtime": 1.0539,
|
| 2587 |
+
"eval_samples_per_second": 15.181,
|
| 2588 |
+
"eval_steps_per_second": 3.795,
|
| 2589 |
+
"step": 516
|
| 2590 |
+
},
|
| 2591 |
+
{
|
| 2592 |
+
"epoch": 7.013513513513513,
|
| 2593 |
+
"grad_norm": 0.9734938144683838,
|
| 2594 |
+
"learning_rate": 3.7804199174215183e-06,
|
| 2595 |
+
"loss": 0.1017,
|
| 2596 |
+
"step": 519
|
| 2597 |
+
},
|
| 2598 |
+
{
|
| 2599 |
+
"epoch": 7.013513513513513,
|
| 2600 |
+
"eval_loss": 0.7537960410118103,
|
| 2601 |
+
"eval_runtime": 1.0511,
|
| 2602 |
+
"eval_samples_per_second": 15.222,
|
| 2603 |
+
"eval_steps_per_second": 3.805,
|
| 2604 |
+
"step": 519
|
| 2605 |
+
},
|
| 2606 |
+
{
|
| 2607 |
+
"epoch": 7.054054054054054,
|
| 2608 |
+
"grad_norm": 1.4745818376541138,
|
| 2609 |
+
"learning_rate": 3.479667411047677e-06,
|
| 2610 |
+
"loss": 0.1536,
|
| 2611 |
+
"step": 522
|
| 2612 |
+
},
|
| 2613 |
+
{
|
| 2614 |
+
"epoch": 7.054054054054054,
|
| 2615 |
+
"eval_loss": 0.7529079914093018,
|
| 2616 |
+
"eval_runtime": 1.0543,
|
| 2617 |
+
"eval_samples_per_second": 15.176,
|
| 2618 |
+
"eval_steps_per_second": 3.794,
|
| 2619 |
+
"step": 522
|
| 2620 |
+
},
|
| 2621 |
+
{
|
| 2622 |
+
"epoch": 7.094594594594595,
|
| 2623 |
+
"grad_norm": 1.0725492238998413,
|
| 2624 |
+
"learning_rate": 3.1909480863070884e-06,
|
| 2625 |
+
"loss": 0.0886,
|
| 2626 |
+
"step": 525
|
| 2627 |
+
},
|
| 2628 |
+
{
|
| 2629 |
+
"epoch": 7.094594594594595,
|
| 2630 |
+
"eval_loss": 0.7565038204193115,
|
| 2631 |
+
"eval_runtime": 1.0511,
|
| 2632 |
+
"eval_samples_per_second": 15.222,
|
| 2633 |
+
"eval_steps_per_second": 3.806,
|
| 2634 |
+
"step": 525
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 7.135135135135135,
|
| 2638 |
+
"grad_norm": 1.1345540285110474,
|
| 2639 |
+
"learning_rate": 2.9143366247826598e-06,
|
| 2640 |
+
"loss": 0.0983,
|
| 2641 |
+
"step": 528
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 7.135135135135135,
|
| 2645 |
+
"eval_loss": 0.7576066255569458,
|
| 2646 |
+
"eval_runtime": 1.0528,
|
| 2647 |
+
"eval_samples_per_second": 15.198,
|
| 2648 |
+
"eval_steps_per_second": 3.799,
|
| 2649 |
+
"step": 528
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 7.175675675675675,
|
| 2653 |
+
"grad_norm": 1.122189998626709,
|
| 2654 |
+
"learning_rate": 2.6499045761769315e-06,
|
| 2655 |
+
"loss": 0.084,
|
| 2656 |
+
"step": 531
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 7.175675675675675,
|
| 2660 |
+
"eval_loss": 0.758578896522522,
|
| 2661 |
+
"eval_runtime": 1.0508,
|
| 2662 |
+
"eval_samples_per_second": 15.227,
|
| 2663 |
+
"eval_steps_per_second": 3.807,
|
| 2664 |
+
"step": 531
|
| 2665 |
+
},
|
| 2666 |
+
{
|
| 2667 |
+
"epoch": 7.216216216216216,
|
| 2668 |
+
"grad_norm": 1.6193064451217651,
|
| 2669 |
+
"learning_rate": 2.397720339804649e-06,
|
| 2670 |
+
"loss": 0.099,
|
| 2671 |
+
"step": 534
|
| 2672 |
+
},
|
| 2673 |
+
{
|
| 2674 |
+
"epoch": 7.216216216216216,
|
| 2675 |
+
"eval_loss": 0.7563527822494507,
|
| 2676 |
+
"eval_runtime": 1.0563,
|
| 2677 |
+
"eval_samples_per_second": 15.147,
|
| 2678 |
+
"eval_steps_per_second": 3.787,
|
| 2679 |
+
"step": 534
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 7.256756756756757,
|
| 2683 |
+
"grad_norm": 1.373356580734253,
|
| 2684 |
+
"learning_rate": 2.1578491469002373e-06,
|
| 2685 |
+
"loss": 0.1089,
|
| 2686 |
+
"step": 537
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 7.256756756756757,
|
| 2690 |
+
"eval_loss": 0.7592064142227173,
|
| 2691 |
+
"eval_runtime": 1.0528,
|
| 2692 |
+
"eval_samples_per_second": 15.197,
|
| 2693 |
+
"eval_steps_per_second": 3.799,
|
| 2694 |
+
"step": 537
|
| 2695 |
+
},
|
| 2696 |
+
{
|
| 2697 |
+
"epoch": 7.297297297297297,
|
| 2698 |
+
"grad_norm": 1.1875869035720825,
|
| 2699 |
+
"learning_rate": 1.9303530437448035e-06,
|
| 2700 |
+
"loss": 0.1145,
|
| 2701 |
+
"step": 540
|
| 2702 |
+
},
|
| 2703 |
+
{
|
| 2704 |
+
"epoch": 7.297297297297297,
|
| 2705 |
+
"eval_loss": 0.7611518502235413,
|
| 2706 |
+
"eval_runtime": 1.0529,
|
| 2707 |
+
"eval_samples_per_second": 15.196,
|
| 2708 |
+
"eval_steps_per_second": 3.799,
|
| 2709 |
+
"step": 540
|
| 2710 |
+
},
|
| 2711 |
+
{
|
| 2712 |
+
"epoch": 7.337837837837838,
|
| 2713 |
+
"grad_norm": 1.8787821531295776,
|
| 2714 |
+
"learning_rate": 1.7152908756169262e-06,
|
| 2715 |
+
"loss": 0.1823,
|
| 2716 |
+
"step": 543
|
| 2717 |
+
},
|
| 2718 |
+
{
|
| 2719 |
+
"epoch": 7.337837837837838,
|
| 2720 |
+
"eval_loss": 0.7614726424217224,
|
| 2721 |
+
"eval_runtime": 1.0548,
|
| 2722 |
+
"eval_samples_per_second": 15.168,
|
| 2723 |
+
"eval_steps_per_second": 3.792,
|
| 2724 |
+
"step": 543
|
| 2725 |
+
},
|
| 2726 |
+
{
|
| 2727 |
+
"epoch": 7.378378378378378,
|
| 2728 |
+
"grad_norm": 1.9469506740570068,
|
| 2729 |
+
"learning_rate": 1.5127182715714006e-06,
|
| 2730 |
+
"loss": 0.2784,
|
| 2731 |
+
"step": 546
|
| 2732 |
+
},
|
| 2733 |
+
{
|
| 2734 |
+
"epoch": 7.378378378378378,
|
| 2735 |
+
"eval_loss": 0.7602246999740601,
|
| 2736 |
+
"eval_runtime": 1.053,
|
| 2737 |
+
"eval_samples_per_second": 15.194,
|
| 2738 |
+
"eval_steps_per_second": 3.799,
|
| 2739 |
+
"step": 546
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 7.418918918918919,
|
| 2743 |
+
"grad_norm": 1.6328327655792236,
|
| 2744 |
+
"learning_rate": 1.3226876300500123e-06,
|
| 2745 |
+
"loss": 0.0887,
|
| 2746 |
+
"step": 549
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 7.418918918918919,
|
| 2750 |
+
"eval_loss": 0.7616763114929199,
|
| 2751 |
+
"eval_runtime": 1.0504,
|
| 2752 |
+
"eval_samples_per_second": 15.232,
|
| 2753 |
+
"eval_steps_per_second": 3.808,
|
| 2754 |
+
"step": 549
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 7.45945945945946,
|
| 2758 |
+
"grad_norm": 1.5713064670562744,
|
| 2759 |
+
"learning_rate": 1.1452481053278396e-06,
|
| 2760 |
+
"loss": 0.1133,
|
| 2761 |
+
"step": 552
|
| 2762 |
+
},
|
| 2763 |
+
{
|
| 2764 |
+
"epoch": 7.45945945945946,
|
| 2765 |
+
"eval_loss": 0.7640103101730347,
|
| 2766 |
+
"eval_runtime": 1.053,
|
| 2767 |
+
"eval_samples_per_second": 15.195,
|
| 2768 |
+
"eval_steps_per_second": 3.799,
|
| 2769 |
+
"step": 552
|
| 2770 |
+
},
|
| 2771 |
+
{
|
| 2772 |
+
"epoch": 7.5,
|
| 2773 |
+
"grad_norm": 1.5901539325714111,
|
| 2774 |
+
"learning_rate": 9.804455947988067e-07,
|
| 2775 |
+
"loss": 0.1207,
|
| 2776 |
+
"step": 555
|
| 2777 |
+
},
|
| 2778 |
+
{
|
| 2779 |
+
"epoch": 7.5,
|
| 2780 |
+
"eval_loss": 0.7629836797714233,
|
| 2781 |
+
"eval_runtime": 1.0516,
|
| 2782 |
+
"eval_samples_per_second": 15.216,
|
| 2783 |
+
"eval_steps_per_second": 3.804,
|
| 2784 |
+
"step": 555
|
| 2785 |
+
},
|
| 2786 |
+
{
|
| 2787 |
+
"epoch": 7.54054054054054,
|
| 2788 |
+
"grad_norm": 1.5648808479309082,
|
| 2789 |
+
"learning_rate": 8.283227271035976e-07,
|
| 2790 |
+
"loss": 0.0954,
|
| 2791 |
+
"step": 558
|
| 2792 |
+
},
|
| 2793 |
+
{
|
| 2794 |
+
"epoch": 7.54054054054054,
|
| 2795 |
+
"eval_loss": 0.7643275260925293,
|
| 2796 |
+
"eval_runtime": 1.0548,
|
| 2797 |
+
"eval_samples_per_second": 15.169,
|
| 2798 |
+
"eval_steps_per_second": 3.792,
|
| 2799 |
+
"step": 558
|
| 2800 |
+
},
|
| 2801 |
+
{
|
| 2802 |
+
"epoch": 7.581081081081081,
|
| 2803 |
+
"grad_norm": 1.6403340101242065,
|
| 2804 |
+
"learning_rate": 6.889188511031542e-07,
|
| 2805 |
+
"loss": 0.1135,
|
| 2806 |
+
"step": 561
|
| 2807 |
+
},
|
| 2808 |
+
{
|
| 2809 |
+
"epoch": 7.581081081081081,
|
| 2810 |
+
"eval_loss": 0.7628697156906128,
|
| 2811 |
+
"eval_runtime": 1.0531,
|
| 2812 |
+
"eval_samples_per_second": 15.194,
|
| 2813 |
+
"eval_steps_per_second": 3.798,
|
| 2814 |
+
"step": 561
|
| 2815 |
+
},
|
| 2816 |
+
{
|
| 2817 |
+
"epoch": 7.621621621621622,
|
| 2818 |
+
"grad_norm": 1.393983244895935,
|
| 2819 |
+
"learning_rate": 5.622700257004676e-07,
|
| 2820 |
+
"loss": 0.096,
|
| 2821 |
+
"step": 564
|
| 2822 |
+
},
|
| 2823 |
+
{
|
| 2824 |
+
"epoch": 7.621621621621622,
|
| 2825 |
+
"eval_loss": 0.7637063264846802,
|
| 2826 |
+
"eval_runtime": 1.0544,
|
| 2827 |
+
"eval_samples_per_second": 15.174,
|
| 2828 |
+
"eval_steps_per_second": 3.793,
|
| 2829 |
+
"step": 564
|
| 2830 |
+
},
|
| 2831 |
+
{
|
| 2832 |
+
"epoch": 7.662162162162162,
|
| 2833 |
+
"grad_norm": 1.2016361951828003,
|
| 2834 |
+
"learning_rate": 4.484090105134231e-07,
|
| 2835 |
+
"loss": 0.1088,
|
| 2836 |
+
"step": 567
|
| 2837 |
+
},
|
| 2838 |
+
{
|
| 2839 |
+
"epoch": 7.662162162162162,
|
| 2840 |
+
"eval_loss": 0.7655338048934937,
|
| 2841 |
+
"eval_runtime": 1.0534,
|
| 2842 |
+
"eval_samples_per_second": 15.189,
|
| 2843 |
+
"eval_steps_per_second": 3.797,
|
| 2844 |
+
"step": 567
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 7.702702702702703,
|
| 2848 |
+
"grad_norm": 1.1388864517211914,
|
| 2849 |
+
"learning_rate": 3.4736525740104444e-07,
|
| 2850 |
+
"loss": 0.1628,
|
| 2851 |
+
"step": 570
|
| 2852 |
+
},
|
| 2853 |
+
{
|
| 2854 |
+
"epoch": 7.702702702702703,
|
| 2855 |
+
"eval_loss": 0.7655097842216492,
|
| 2856 |
+
"eval_runtime": 1.053,
|
| 2857 |
+
"eval_samples_per_second": 15.195,
|
| 2858 |
+
"eval_steps_per_second": 3.799,
|
| 2859 |
+
"step": 570
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 7.743243243243243,
|
| 2863 |
+
"grad_norm": 1.9650497436523438,
|
| 2864 |
+
"learning_rate": 2.591649028453047e-07,
|
| 2865 |
+
"loss": 0.1431,
|
| 2866 |
+
"step": 573
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 7.743243243243243,
|
| 2870 |
+
"eval_loss": 0.7649960517883301,
|
| 2871 |
+
"eval_runtime": 1.0519,
|
| 2872 |
+
"eval_samples_per_second": 15.211,
|
| 2873 |
+
"eval_steps_per_second": 3.803,
|
| 2874 |
+
"step": 573
|
| 2875 |
+
},
|
| 2876 |
+
{
|
| 2877 |
+
"epoch": 7.783783783783784,
|
| 2878 |
+
"grad_norm": 1.7549225091934204,
|
| 2879 |
+
"learning_rate": 1.8383076119053432e-07,
|
| 2880 |
+
"loss": 0.1034,
|
| 2881 |
+
"step": 576
|
| 2882 |
+
},
|
| 2883 |
+
{
|
| 2884 |
+
"epoch": 7.783783783783784,
|
| 2885 |
+
"eval_loss": 0.763870358467102,
|
| 2886 |
+
"eval_runtime": 1.0529,
|
| 2887 |
+
"eval_samples_per_second": 15.196,
|
| 2888 |
+
"eval_steps_per_second": 3.799,
|
| 2889 |
+
"step": 576
|
| 2890 |
+
},
|
| 2891 |
+
{
|
| 2892 |
+
"epoch": 7.824324324324325,
|
| 2893 |
+
"grad_norm": 1.7549595832824707,
|
| 2894 |
+
"learning_rate": 1.2138231874217475e-07,
|
| 2895 |
+
"loss": 0.181,
|
| 2896 |
+
"step": 579
|
| 2897 |
+
},
|
| 2898 |
+
{
|
| 2899 |
+
"epoch": 7.824324324324325,
|
| 2900 |
+
"eval_loss": 0.7637079358100891,
|
| 2901 |
+
"eval_runtime": 1.0546,
|
| 2902 |
+
"eval_samples_per_second": 15.172,
|
| 2903 |
+
"eval_steps_per_second": 3.793,
|
| 2904 |
+
"step": 579
|
| 2905 |
+
},
|
| 2906 |
+
{
|
| 2907 |
+
"epoch": 7.864864864864865,
|
| 2908 |
+
"grad_norm": 1.3891515731811523,
|
| 2909 |
+
"learning_rate": 7.183572872632715e-08,
|
| 2910 |
+
"loss": 0.062,
|
| 2911 |
+
"step": 582
|
| 2912 |
+
},
|
| 2913 |
+
{
|
| 2914 |
+
"epoch": 7.864864864864865,
|
| 2915 |
+
"eval_loss": 0.7649126052856445,
|
| 2916 |
+
"eval_runtime": 1.0509,
|
| 2917 |
+
"eval_samples_per_second": 15.225,
|
| 2918 |
+
"eval_steps_per_second": 3.806,
|
| 2919 |
+
"step": 582
|
| 2920 |
+
},
|
| 2921 |
+
{
|
| 2922 |
+
"epoch": 7.905405405405405,
|
| 2923 |
+
"grad_norm": 1.0669249296188354,
|
| 2924 |
+
"learning_rate": 3.5203807111489074e-08,
|
| 2925 |
+
"loss": 0.0769,
|
| 2926 |
+
"step": 585
|
| 2927 |
+
},
|
| 2928 |
+
{
|
| 2929 |
+
"epoch": 7.905405405405405,
|
| 2930 |
+
"eval_loss": 0.7653980255126953,
|
| 2931 |
+
"eval_runtime": 1.0536,
|
| 2932 |
+
"eval_samples_per_second": 15.185,
|
| 2933 |
+
"eval_steps_per_second": 3.796,
|
| 2934 |
+
"step": 585
|
| 2935 |
+
},
|
| 2936 |
+
{
|
| 2937 |
+
"epoch": 7.945945945945946,
|
| 2938 |
+
"grad_norm": 2.3302104473114014,
|
| 2939 |
+
"learning_rate": 1.1496029293511789e-08,
|
| 2940 |
+
"loss": 0.1951,
|
| 2941 |
+
"step": 588
|
| 2942 |
+
},
|
| 2943 |
+
{
|
| 2944 |
+
"epoch": 7.945945945945946,
|
| 2945 |
+
"eval_loss": 0.7646524906158447,
|
| 2946 |
+
"eval_runtime": 1.0566,
|
| 2947 |
+
"eval_samples_per_second": 15.143,
|
| 2948 |
+
"eval_steps_per_second": 3.786,
|
| 2949 |
+
"step": 588
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 7.986486486486487,
|
| 2953 |
+
"grad_norm": 1.9744952917099,
|
| 2954 |
+
"learning_rate": 7.185276446441958e-10,
|
| 2955 |
+
"loss": 0.1175,
|
| 2956 |
+
"step": 591
|
| 2957 |
+
},
|
| 2958 |
+
{
|
| 2959 |
+
"epoch": 7.986486486486487,
|
| 2960 |
+
"eval_loss": 0.765015721321106,
|
| 2961 |
+
"eval_runtime": 1.0522,
|
| 2962 |
+
"eval_samples_per_second": 15.206,
|
| 2963 |
+
"eval_steps_per_second": 3.801,
|
| 2964 |
+
"step": 591
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 8.0,
|
| 2968 |
+
"step": 592,
|
| 2969 |
+
"total_flos": 1188976147968000.0,
|
| 2970 |
+
"train_loss": 0.4161318518926163,
|
| 2971 |
+
"train_runtime": 741.424,
|
| 2972 |
+
"train_samples_per_second": 3.194,
|
| 2973 |
+
"train_steps_per_second": 0.798
|
| 2974 |
+
}
|
| 2975 |
+
],
|
| 2976 |
+
"logging_steps": 3,
|
| 2977 |
+
"max_steps": 592,
|
| 2978 |
+
"num_input_tokens_seen": 0,
|
| 2979 |
+
"num_train_epochs": 8,
|
| 2980 |
+
"save_steps": 50,
|
| 2981 |
+
"stateful_callbacks": {
|
| 2982 |
+
"TrainerControl": {
|
| 2983 |
+
"args": {
|
| 2984 |
+
"should_epoch_stop": false,
|
| 2985 |
+
"should_evaluate": false,
|
| 2986 |
+
"should_log": false,
|
| 2987 |
+
"should_save": true,
|
| 2988 |
+
"should_training_stop": true
|
| 2989 |
+
},
|
| 2990 |
+
"attributes": {}
|
| 2991 |
+
}
|
| 2992 |
+
},
|
| 2993 |
+
"total_flos": 1188976147968000.0,
|
| 2994 |
+
"train_batch_size": 4,
|
| 2995 |
+
"trial_name": null,
|
| 2996 |
+
"trial_params": null
|
| 2997 |
+
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5496
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ada886b0a7db7fa910beb2db1b673de82468f97a7444c0403b79df92cd914bcd
|
| 3 |
size 5496
|