Model save
Browse files- README.md +1 -1
- all_results.json +5 -5
- train_results.json +5 -5
- trainer_state.json +327 -151
README.md
CHANGED
|
@@ -27,7 +27,7 @@ print(output["generated_text"])
|
|
| 27 |
|
| 28 |
## Training procedure
|
| 29 |
|
| 30 |
-
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/yeokch/stream-of-search-train/runs/
|
| 31 |
|
| 32 |
|
| 33 |
This model was trained with SFT.
|
|
|
|
| 27 |
|
| 28 |
## Training procedure
|
| 29 |
|
| 30 |
+
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/yeokch/stream-of-search-train/runs/1bx3j201)
|
| 31 |
|
| 32 |
|
| 33 |
This model was trained with SFT.
|
all_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"total_flos":
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
-
"train_samples":
|
| 6 |
-
"train_samples_per_second": 0.
|
| 7 |
"train_steps_per_second": 0.028
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"total_flos": 2.5792410000818176e+17,
|
| 3 |
+
"train_loss": 0.5621715438993353,
|
| 4 |
+
"train_runtime": 8702.0895,
|
| 5 |
+
"train_samples": 6400,
|
| 6 |
+
"train_samples_per_second": 0.456,
|
| 7 |
"train_steps_per_second": 0.028
|
| 8 |
}
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"total_flos":
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
-
"train_samples":
|
| 6 |
-
"train_samples_per_second": 0.
|
| 7 |
"train_steps_per_second": 0.028
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"total_flos": 2.5792410000818176e+17,
|
| 3 |
+
"train_loss": 0.5621715438993353,
|
| 4 |
+
"train_runtime": 8702.0895,
|
| 5 |
+
"train_samples": 6400,
|
| 6 |
+
"train_samples_per_second": 0.456,
|
| 7 |
"train_steps_per_second": 0.028
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -2,250 +2,426 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 500,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
-
"epoch": 0.
|
| 14 |
-
"grad_norm": 0.
|
| 15 |
-
"learning_rate":
|
| 16 |
-
"loss": 0.
|
| 17 |
-
"mean_token_accuracy": 0.
|
| 18 |
"step": 1
|
| 19 |
},
|
| 20 |
{
|
| 21 |
-
"epoch": 0.
|
| 22 |
-
"grad_norm": 0.
|
| 23 |
-
"learning_rate":
|
| 24 |
-
"loss":
|
| 25 |
-
"mean_token_accuracy": 0.
|
| 26 |
"step": 5
|
| 27 |
},
|
| 28 |
{
|
| 29 |
-
"epoch": 0.
|
| 30 |
-
"grad_norm": 0.
|
| 31 |
-
"learning_rate":
|
| 32 |
-
"loss": 0.
|
| 33 |
-
"mean_token_accuracy": 0.
|
| 34 |
"step": 10
|
| 35 |
},
|
| 36 |
{
|
| 37 |
-
"epoch": 0.
|
| 38 |
-
"grad_norm": 0.
|
| 39 |
-
"learning_rate": 0.
|
| 40 |
-
"loss": 0.
|
| 41 |
-
"mean_token_accuracy": 0.
|
| 42 |
"step": 15
|
| 43 |
},
|
| 44 |
{
|
| 45 |
-
"epoch": 0.
|
| 46 |
-
"grad_norm": 0.
|
| 47 |
-
"learning_rate": 0.
|
| 48 |
-
"loss": 0.
|
| 49 |
-
"mean_token_accuracy": 0.
|
| 50 |
"step": 20
|
| 51 |
},
|
| 52 |
{
|
| 53 |
-
"epoch": 0.
|
| 54 |
-
"grad_norm": 0.
|
| 55 |
-
"learning_rate": 0.
|
| 56 |
-
"loss": 0.
|
| 57 |
-
"mean_token_accuracy": 0.
|
| 58 |
"step": 25
|
| 59 |
},
|
| 60 |
{
|
| 61 |
-
"epoch": 0.
|
| 62 |
-
"grad_norm": 0.
|
| 63 |
-
"learning_rate": 0.
|
| 64 |
-
"loss": 0.
|
| 65 |
-
"mean_token_accuracy": 0.
|
| 66 |
"step": 30
|
| 67 |
},
|
| 68 |
{
|
| 69 |
-
"epoch": 0.
|
| 70 |
-
"grad_norm": 0.
|
| 71 |
-
"learning_rate": 0.
|
| 72 |
-
"loss": 0.
|
| 73 |
-
"mean_token_accuracy": 0.
|
| 74 |
"step": 35
|
| 75 |
},
|
| 76 |
{
|
| 77 |
-
"epoch": 0.
|
| 78 |
-
"grad_norm": 0.
|
| 79 |
-
"learning_rate": 0.
|
| 80 |
-
"loss": 0.
|
| 81 |
-
"mean_token_accuracy": 0.
|
| 82 |
"step": 40
|
| 83 |
},
|
| 84 |
{
|
| 85 |
-
"epoch": 0.
|
| 86 |
-
"grad_norm": 0.
|
| 87 |
-
"learning_rate": 0.
|
| 88 |
-
"loss": 0.
|
| 89 |
-
"mean_token_accuracy": 0.
|
| 90 |
"step": 45
|
| 91 |
},
|
| 92 |
{
|
| 93 |
-
"epoch": 0.
|
| 94 |
-
"grad_norm": 0.
|
| 95 |
-
"learning_rate": 0.
|
| 96 |
-
"loss": 0.
|
| 97 |
-
"mean_token_accuracy": 0.
|
| 98 |
"step": 50
|
| 99 |
},
|
| 100 |
{
|
| 101 |
-
"epoch": 0.
|
| 102 |
-
"grad_norm": 0.
|
| 103 |
-
"learning_rate": 0.
|
| 104 |
-
"loss": 0.
|
| 105 |
-
"mean_token_accuracy": 0.
|
| 106 |
"step": 55
|
| 107 |
},
|
| 108 |
{
|
| 109 |
-
"epoch": 0.
|
| 110 |
-
"grad_norm": 0.
|
| 111 |
-
"learning_rate": 0.
|
| 112 |
-
"loss": 0.
|
| 113 |
-
"mean_token_accuracy": 0.
|
| 114 |
"step": 60
|
| 115 |
},
|
| 116 |
{
|
| 117 |
-
"epoch": 0.
|
| 118 |
-
"grad_norm": 0.
|
| 119 |
-
"learning_rate": 0.
|
| 120 |
-
"loss": 0.
|
| 121 |
-
"mean_token_accuracy": 0.
|
| 122 |
"step": 65
|
| 123 |
},
|
| 124 |
{
|
| 125 |
-
"epoch": 0.
|
| 126 |
-
"grad_norm": 0.
|
| 127 |
-
"learning_rate": 0.
|
| 128 |
-
"loss": 0.
|
| 129 |
-
"mean_token_accuracy": 0.
|
| 130 |
"step": 70
|
| 131 |
},
|
| 132 |
{
|
| 133 |
-
"epoch": 0.
|
| 134 |
-
"grad_norm": 0.
|
| 135 |
-
"learning_rate": 0.
|
| 136 |
-
"loss": 0.
|
| 137 |
-
"mean_token_accuracy": 0.
|
| 138 |
"step": 75
|
| 139 |
},
|
| 140 |
{
|
| 141 |
-
"epoch": 0.
|
| 142 |
-
"grad_norm": 0.
|
| 143 |
-
"learning_rate":
|
| 144 |
-
"loss": 0.
|
| 145 |
-
"mean_token_accuracy": 0.
|
| 146 |
"step": 80
|
| 147 |
},
|
| 148 |
{
|
| 149 |
-
"epoch": 0.
|
| 150 |
-
"grad_norm": 0.
|
| 151 |
-
"learning_rate":
|
| 152 |
-
"loss": 0.
|
| 153 |
-
"mean_token_accuracy": 0.
|
| 154 |
"step": 85
|
| 155 |
},
|
| 156 |
{
|
| 157 |
-
"epoch": 0.
|
| 158 |
-
"grad_norm": 0.
|
| 159 |
-
"learning_rate":
|
| 160 |
-
"loss": 0.
|
| 161 |
-
"mean_token_accuracy": 0.
|
| 162 |
"step": 90
|
| 163 |
},
|
| 164 |
{
|
| 165 |
-
"epoch": 0.
|
| 166 |
-
"grad_norm": 0.
|
| 167 |
-
"learning_rate":
|
| 168 |
-
"loss": 0.
|
| 169 |
-
"mean_token_accuracy": 0.
|
| 170 |
"step": 95
|
| 171 |
},
|
| 172 |
{
|
| 173 |
-
"epoch": 0.
|
| 174 |
-
"grad_norm": 0.
|
| 175 |
-
"learning_rate":
|
| 176 |
-
"loss": 0.
|
| 177 |
-
"mean_token_accuracy": 0.
|
| 178 |
"step": 100
|
| 179 |
},
|
| 180 |
{
|
| 181 |
-
"epoch": 0.
|
| 182 |
-
"grad_norm": 0.
|
| 183 |
-
"learning_rate":
|
| 184 |
-
"loss": 0.
|
| 185 |
-
"mean_token_accuracy": 0.
|
| 186 |
"step": 105
|
| 187 |
},
|
| 188 |
{
|
| 189 |
-
"epoch": 0.
|
| 190 |
-
"grad_norm": 0.
|
| 191 |
-
"learning_rate":
|
| 192 |
-
"loss": 0.
|
| 193 |
-
"mean_token_accuracy": 0.
|
| 194 |
"step": 110
|
| 195 |
},
|
| 196 |
{
|
| 197 |
-
"epoch": 0.
|
| 198 |
-
"grad_norm": 0.
|
| 199 |
-
"learning_rate":
|
| 200 |
-
"loss": 0.
|
| 201 |
-
"mean_token_accuracy": 0.
|
| 202 |
"step": 115
|
| 203 |
},
|
| 204 |
{
|
| 205 |
-
"epoch": 0.
|
| 206 |
-
"grad_norm": 0.
|
| 207 |
-
"learning_rate":
|
| 208 |
-
"loss": 0.
|
| 209 |
-
"mean_token_accuracy": 0.
|
| 210 |
"step": 120
|
| 211 |
},
|
| 212 |
{
|
| 213 |
-
"epoch": 0.
|
| 214 |
-
"grad_norm": 0.
|
| 215 |
-
"learning_rate":
|
| 216 |
-
"loss": 0.
|
| 217 |
-
"mean_token_accuracy": 0.
|
| 218 |
"step": 125
|
| 219 |
},
|
| 220 |
{
|
| 221 |
-
"epoch": 0.
|
| 222 |
-
"grad_norm": 0.
|
| 223 |
-
"learning_rate":
|
| 224 |
-
"loss": 0.
|
| 225 |
-
"mean_token_accuracy": 0.
|
| 226 |
"step": 130
|
| 227 |
},
|
| 228 |
{
|
| 229 |
-
"epoch": 0.
|
| 230 |
-
"grad_norm": 0.
|
| 231 |
-
"learning_rate":
|
| 232 |
-
"loss": 0.
|
| 233 |
-
"mean_token_accuracy": 0.
|
| 234 |
"step": 135
|
| 235 |
},
|
| 236 |
{
|
| 237 |
-
"epoch": 0.
|
| 238 |
-
"
|
| 239 |
-
"
|
| 240 |
-
"
|
| 241 |
-
"
|
| 242 |
-
"
|
| 243 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 244 |
"train_steps_per_second": 0.028
|
| 245 |
}
|
| 246 |
],
|
| 247 |
"logging_steps": 5,
|
| 248 |
-
"max_steps":
|
| 249 |
"num_input_tokens_seen": 0,
|
| 250 |
"num_train_epochs": 1,
|
| 251 |
"save_steps": 100,
|
|
@@ -261,7 +437,7 @@
|
|
| 261 |
"attributes": {}
|
| 262 |
}
|
| 263 |
},
|
| 264 |
-
"total_flos":
|
| 265 |
"train_batch_size": 1,
|
| 266 |
"trial_name": null,
|
| 267 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.9964699949571356,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 247,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
+
"epoch": 0.004034291477559254,
|
| 14 |
+
"grad_norm": 0.10618364810943604,
|
| 15 |
+
"learning_rate": 8.000000000000001e-06,
|
| 16 |
+
"loss": 0.8761,
|
| 17 |
+
"mean_token_accuracy": 0.7515642121434212,
|
| 18 |
"step": 1
|
| 19 |
},
|
| 20 |
{
|
| 21 |
+
"epoch": 0.020171457387796268,
|
| 22 |
+
"grad_norm": 0.1208970844745636,
|
| 23 |
+
"learning_rate": 4e-05,
|
| 24 |
+
"loss": 0.9213,
|
| 25 |
+
"mean_token_accuracy": 0.7374346107244492,
|
| 26 |
"step": 5
|
| 27 |
},
|
| 28 |
{
|
| 29 |
+
"epoch": 0.040342914775592535,
|
| 30 |
+
"grad_norm": 0.09196259081363678,
|
| 31 |
+
"learning_rate": 8e-05,
|
| 32 |
+
"loss": 0.8914,
|
| 33 |
+
"mean_token_accuracy": 0.7460352770984173,
|
| 34 |
"step": 10
|
| 35 |
},
|
| 36 |
{
|
| 37 |
+
"epoch": 0.060514372163388806,
|
| 38 |
+
"grad_norm": 0.10933861136436462,
|
| 39 |
+
"learning_rate": 0.00012,
|
| 40 |
+
"loss": 0.8602,
|
| 41 |
+
"mean_token_accuracy": 0.7503265753388405,
|
| 42 |
"step": 15
|
| 43 |
},
|
| 44 |
{
|
| 45 |
+
"epoch": 0.08068582955118507,
|
| 46 |
+
"grad_norm": 0.11680758744478226,
|
| 47 |
+
"learning_rate": 0.00016,
|
| 48 |
+
"loss": 0.8439,
|
| 49 |
+
"mean_token_accuracy": 0.7499923676252365,
|
| 50 |
"step": 20
|
| 51 |
},
|
| 52 |
{
|
| 53 |
+
"epoch": 0.10085728693898134,
|
| 54 |
+
"grad_norm": 0.0951286181807518,
|
| 55 |
+
"learning_rate": 0.0002,
|
| 56 |
+
"loss": 0.7772,
|
| 57 |
+
"mean_token_accuracy": 0.7655725114047527,
|
| 58 |
"step": 25
|
| 59 |
},
|
| 60 |
{
|
| 61 |
+
"epoch": 0.12102874432677761,
|
| 62 |
+
"grad_norm": 0.08053945004940033,
|
| 63 |
+
"learning_rate": 0.00019974977965945,
|
| 64 |
+
"loss": 0.7238,
|
| 65 |
+
"mean_token_accuracy": 0.7779346205294132,
|
| 66 |
"step": 30
|
| 67 |
},
|
| 68 |
{
|
| 69 |
+
"epoch": 0.14120020171457387,
|
| 70 |
+
"grad_norm": 0.08780888468027115,
|
| 71 |
+
"learning_rate": 0.00019900037084217637,
|
| 72 |
+
"loss": 0.6863,
|
| 73 |
+
"mean_token_accuracy": 0.7845150113105774,
|
| 74 |
"step": 35
|
| 75 |
},
|
| 76 |
{
|
| 77 |
+
"epoch": 0.16137165910237014,
|
| 78 |
+
"grad_norm": 0.08427116274833679,
|
| 79 |
+
"learning_rate": 0.00019775552389476864,
|
| 80 |
+
"loss": 0.6315,
|
| 81 |
+
"mean_token_accuracy": 0.7988936014473438,
|
| 82 |
"step": 40
|
| 83 |
},
|
| 84 |
{
|
| 85 |
+
"epoch": 0.1815431164901664,
|
| 86 |
+
"grad_norm": 0.08954589813947678,
|
| 87 |
+
"learning_rate": 0.00019602146853776894,
|
| 88 |
+
"loss": 0.617,
|
| 89 |
+
"mean_token_accuracy": 0.8002350099384785,
|
| 90 |
"step": 45
|
| 91 |
},
|
| 92 |
{
|
| 93 |
+
"epoch": 0.20171457387796268,
|
| 94 |
+
"grad_norm": 0.09293227642774582,
|
| 95 |
+
"learning_rate": 0.0001938068826896166,
|
| 96 |
+
"loss": 0.5883,
|
| 97 |
+
"mean_token_accuracy": 0.8079691044986248,
|
| 98 |
"step": 50
|
| 99 |
},
|
| 100 |
{
|
| 101 |
+
"epoch": 0.22188603126575895,
|
| 102 |
+
"grad_norm": 0.09921843558549881,
|
| 103 |
+
"learning_rate": 0.0001911228490388136,
|
| 104 |
+
"loss": 0.5817,
|
| 105 |
+
"mean_token_accuracy": 0.8088557526469231,
|
| 106 |
"step": 55
|
| 107 |
},
|
| 108 |
{
|
| 109 |
+
"epoch": 0.24205748865355523,
|
| 110 |
+
"grad_norm": 0.11710495501756668,
|
| 111 |
+
"learning_rate": 0.00018798279958164295,
|
| 112 |
+
"loss": 0.5737,
|
| 113 |
+
"mean_token_accuracy": 0.809910261631012,
|
| 114 |
"step": 60
|
| 115 |
},
|
| 116 |
{
|
| 117 |
+
"epoch": 0.2622289460413515,
|
| 118 |
+
"grad_norm": 0.10755322873592377,
|
| 119 |
+
"learning_rate": 0.00018440244840299506,
|
| 120 |
+
"loss": 0.567,
|
| 121 |
+
"mean_token_accuracy": 0.8104779615998268,
|
| 122 |
"step": 65
|
| 123 |
},
|
| 124 |
{
|
| 125 |
+
"epoch": 0.28240040342914774,
|
| 126 |
+
"grad_norm": 0.12730592489242554,
|
| 127 |
+
"learning_rate": 0.00018039971303669407,
|
| 128 |
+
"loss": 0.5546,
|
| 129 |
+
"mean_token_accuracy": 0.8143862128257752,
|
| 130 |
"step": 70
|
| 131 |
},
|
| 132 |
{
|
| 133 |
+
"epoch": 0.30257186081694404,
|
| 134 |
+
"grad_norm": 0.12127107381820679,
|
| 135 |
+
"learning_rate": 0.00017599462479886974,
|
| 136 |
+
"loss": 0.5522,
|
| 137 |
+
"mean_token_accuracy": 0.8144258864223957,
|
| 138 |
"step": 75
|
| 139 |
},
|
| 140 |
{
|
| 141 |
+
"epoch": 0.3227433182047403,
|
| 142 |
+
"grad_norm": 0.11108188331127167,
|
| 143 |
+
"learning_rate": 0.00017120922854310257,
|
| 144 |
+
"loss": 0.5305,
|
| 145 |
+
"mean_token_accuracy": 0.8214686810970306,
|
| 146 |
"step": 80
|
| 147 |
},
|
| 148 |
{
|
| 149 |
+
"epoch": 0.3429147755925366,
|
| 150 |
+
"grad_norm": 0.11264903098344803,
|
| 151 |
+
"learning_rate": 0.00016606747233900815,
|
| 152 |
+
"loss": 0.5398,
|
| 153 |
+
"mean_token_accuracy": 0.8181388042867184,
|
| 154 |
"step": 85
|
| 155 |
},
|
| 156 |
{
|
| 157 |
+
"epoch": 0.3630862329803328,
|
| 158 |
+
"grad_norm": 0.1377757340669632,
|
| 159 |
+
"learning_rate": 0.00016059508762635482,
|
| 160 |
+
"loss": 0.532,
|
| 161 |
+
"mean_token_accuracy": 0.8195081010460854,
|
| 162 |
"step": 90
|
| 163 |
},
|
| 164 |
{
|
| 165 |
+
"epoch": 0.3832576903681291,
|
| 166 |
+
"grad_norm": 0.12057465314865112,
|
| 167 |
+
"learning_rate": 0.00015481946044447099,
|
| 168 |
+
"loss": 0.5238,
|
| 169 |
+
"mean_token_accuracy": 0.8215388789772987,
|
| 170 |
"step": 95
|
| 171 |
},
|
| 172 |
{
|
| 173 |
+
"epoch": 0.40342914775592537,
|
| 174 |
+
"grad_norm": 0.11411843448877335,
|
| 175 |
+
"learning_rate": 0.00014876949438136347,
|
| 176 |
+
"loss": 0.5285,
|
| 177 |
+
"mean_token_accuracy": 0.819812896847725,
|
| 178 |
"step": 100
|
| 179 |
},
|
| 180 |
{
|
| 181 |
+
"epoch": 0.4236006051437216,
|
| 182 |
+
"grad_norm": 0.12218334525823593,
|
| 183 |
+
"learning_rate": 0.0001424754659284048,
|
| 184 |
+
"loss": 0.5178,
|
| 185 |
+
"mean_token_accuracy": 0.8232251830399037,
|
| 186 |
"step": 105
|
| 187 |
},
|
| 188 |
{
|
| 189 |
+
"epoch": 0.4437720625315179,
|
| 190 |
+
"grad_norm": 0.12559793889522552,
|
| 191 |
+
"learning_rate": 0.0001359688729644536,
|
| 192 |
+
"loss": 0.5269,
|
| 193 |
+
"mean_token_accuracy": 0.8201104834675789,
|
| 194 |
"step": 110
|
| 195 |
},
|
| 196 |
{
|
| 197 |
+
"epoch": 0.46394351991931415,
|
| 198 |
+
"grad_norm": 0.14769871532917023,
|
| 199 |
+
"learning_rate": 0.00012928227712765504,
|
| 200 |
+
"loss": 0.5203,
|
| 201 |
+
"mean_token_accuracy": 0.8221889816224575,
|
| 202 |
"step": 115
|
| 203 |
},
|
| 204 |
{
|
| 205 |
+
"epoch": 0.48411497730711045,
|
| 206 |
+
"grad_norm": 0.1480661779642105,
|
| 207 |
+
"learning_rate": 0.00012244914086375724,
|
| 208 |
+
"loss": 0.523,
|
| 209 |
+
"mean_token_accuracy": 0.8210825860500336,
|
| 210 |
"step": 120
|
| 211 |
},
|
| 212 |
{
|
| 213 |
+
"epoch": 0.5042864346949067,
|
| 214 |
+
"grad_norm": 0.14564430713653564,
|
| 215 |
+
"learning_rate": 0.00011550365996641979,
|
| 216 |
+
"loss": 0.5074,
|
| 217 |
+
"mean_token_accuracy": 0.8264268651604653,
|
| 218 |
"step": 125
|
| 219 |
},
|
| 220 |
{
|
| 221 |
+
"epoch": 0.524457892082703,
|
| 222 |
+
"grad_norm": 0.13270238041877747,
|
| 223 |
+
"learning_rate": 0.00010848059244755093,
|
| 224 |
+
"loss": 0.5092,
|
| 225 |
+
"mean_token_accuracy": 0.8253403089940547,
|
| 226 |
"step": 130
|
| 227 |
},
|
| 228 |
{
|
| 229 |
+
"epoch": 0.5446293494704992,
|
| 230 |
+
"grad_norm": 0.11992616951465607,
|
| 231 |
+
"learning_rate": 0.00010141508459407623,
|
| 232 |
+
"loss": 0.5012,
|
| 233 |
+
"mean_token_accuracy": 0.8276812911033631,
|
| 234 |
"step": 135
|
| 235 |
},
|
| 236 |
{
|
| 237 |
+
"epoch": 0.5648008068582955,
|
| 238 |
+
"grad_norm": 0.14025390148162842,
|
| 239 |
+
"learning_rate": 9.434249508162076e-05,
|
| 240 |
+
"loss": 0.5128,
|
| 241 |
+
"mean_token_accuracy": 0.8234357766807079,
|
| 242 |
+
"step": 140
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.5849722642460918,
|
| 246 |
+
"grad_norm": 0.11215393245220184,
|
| 247 |
+
"learning_rate": 8.729821802531212e-05,
|
| 248 |
+
"loss": 0.5133,
|
| 249 |
+
"mean_token_accuracy": 0.8236173823475837,
|
| 250 |
+
"step": 145
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.6051437216338881,
|
| 254 |
+
"grad_norm": 0.11705436557531357,
|
| 255 |
+
"learning_rate": 8.031750585322947e-05,
|
| 256 |
+
"loss": 0.5053,
|
| 257 |
+
"mean_token_accuracy": 0.8257141962647438,
|
| 258 |
+
"step": 150
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 0.6253151790216843,
|
| 262 |
+
"grad_norm": 0.12025938928127289,
|
| 263 |
+
"learning_rate": 7.343529288891239e-05,
|
| 264 |
+
"loss": 0.4955,
|
| 265 |
+
"mean_token_accuracy": 0.8292319044470787,
|
| 266 |
+
"step": 155
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.6454866364094806,
|
| 270 |
+
"grad_norm": 0.11226729303598404,
|
| 271 |
+
"learning_rate": 6.668602052579424e-05,
|
| 272 |
+
"loss": 0.4939,
|
| 273 |
+
"mean_token_accuracy": 0.8295842953026294,
|
| 274 |
+
"step": 160
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 0.6656580937972768,
|
| 278 |
+
"grad_norm": 0.10537487268447876,
|
| 279 |
+
"learning_rate": 6.010346486845837e-05,
|
| 280 |
+
"loss": 0.4907,
|
| 281 |
+
"mean_token_accuracy": 0.8306830637156963,
|
| 282 |
+
"step": 165
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.6858295511850732,
|
| 286 |
+
"grad_norm": 0.1326971799135208,
|
| 287 |
+
"learning_rate": 5.372056770327013e-05,
|
| 288 |
+
"loss": 0.4994,
|
| 289 |
+
"mean_token_accuracy": 0.8274142310023308,
|
| 290 |
+
"step": 170
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.7060010085728694,
|
| 294 |
+
"grad_norm": 0.0996076837182045,
|
| 295 |
+
"learning_rate": 4.756927164427685e-05,
|
| 296 |
+
"loss": 0.4973,
|
| 297 |
+
"mean_token_accuracy": 0.8285726100206375,
|
| 298 |
+
"step": 175
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.7261724659606656,
|
| 302 |
+
"grad_norm": 0.10915440320968628,
|
| 303 |
+
"learning_rate": 4.168036027937267e-05,
|
| 304 |
+
"loss": 0.4987,
|
| 305 |
+
"mean_token_accuracy": 0.8276034623384476,
|
| 306 |
+
"step": 180
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 0.7463439233484619,
|
| 310 |
+
"grad_norm": 0.10033515840768814,
|
| 311 |
+
"learning_rate": 3.6083304116701535e-05,
|
| 312 |
+
"loss": 0.4946,
|
| 313 |
+
"mean_token_accuracy": 0.8296850137412548,
|
| 314 |
+
"step": 185
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.7665153807362582,
|
| 318 |
+
"grad_norm": 0.10720834881067276,
|
| 319 |
+
"learning_rate": 3.080611310224539e-05,
|
| 320 |
+
"loss": 0.4952,
|
| 321 |
+
"mean_token_accuracy": 0.8289067208766937,
|
| 322 |
+
"step": 190
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 0.7866868381240545,
|
| 326 |
+
"grad_norm": 0.11353600770235062,
|
| 327 |
+
"learning_rate": 2.587519644666001e-05,
|
| 328 |
+
"loss": 0.4978,
|
| 329 |
+
"mean_token_accuracy": 0.8272387310862541,
|
| 330 |
+
"step": 195
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.8068582955118507,
|
| 334 |
+
"grad_norm": 0.09966889768838882,
|
| 335 |
+
"learning_rate": 2.1315230462840985e-05,
|
| 336 |
+
"loss": 0.4993,
|
| 337 |
+
"mean_token_accuracy": 0.8278293192386628,
|
| 338 |
+
"step": 200
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.827029752899647,
|
| 342 |
+
"grad_norm": 0.10758362710475922,
|
| 343 |
+
"learning_rate": 1.7149035075615794e-05,
|
| 344 |
+
"loss": 0.5055,
|
| 345 |
+
"mean_token_accuracy": 0.8247546002268791,
|
| 346 |
+
"step": 205
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.8472012102874432,
|
| 350 |
+
"grad_norm": 0.09960389137268066,
|
| 351 |
+
"learning_rate": 1.339745962155613e-05,
|
| 352 |
+
"loss": 0.4822,
|
| 353 |
+
"mean_token_accuracy": 0.8327905587852001,
|
| 354 |
+
"step": 210
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.8673726676752396,
|
| 358 |
+
"grad_norm": 0.10244999825954437,
|
| 359 |
+
"learning_rate": 1.0079278510416313e-05,
|
| 360 |
+
"loss": 0.4894,
|
| 361 |
+
"mean_token_accuracy": 0.8305548712611198,
|
| 362 |
+
"step": 215
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.8875441250630358,
|
| 366 |
+
"grad_norm": 0.09608753025531769,
|
| 367 |
+
"learning_rate": 7.211097270349066e-06,
|
| 368 |
+
"loss": 0.4998,
|
| 369 |
+
"mean_token_accuracy": 0.8276750639081001,
|
| 370 |
+
"step": 220
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.9077155824508321,
|
| 374 |
+
"grad_norm": 0.09114421159029007,
|
| 375 |
+
"learning_rate": 4.807269447087348e-06,
|
| 376 |
+
"loss": 0.4985,
|
| 377 |
+
"mean_token_accuracy": 0.827449332177639,
|
| 378 |
+
"step": 225
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 0.9278870398386283,
|
| 382 |
+
"grad_norm": 0.09354618191719055,
|
| 383 |
+
"learning_rate": 2.8798247729623806e-06,
|
| 384 |
+
"loss": 0.493,
|
| 385 |
+
"mean_token_accuracy": 0.8297582663595676,
|
| 386 |
+
"step": 230
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.9480584972264247,
|
| 390 |
+
"grad_norm": 0.09092767536640167,
|
| 391 |
+
"learning_rate": 1.4384089652291543e-06,
|
| 392 |
+
"loss": 0.4887,
|
| 393 |
+
"mean_token_accuracy": 0.8308509282767773,
|
| 394 |
+
"step": 235
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.9682299546142209,
|
| 398 |
+
"grad_norm": 0.09328487515449524,
|
| 399 |
+
"learning_rate": 4.902354549733978e-07,
|
| 400 |
+
"loss": 0.4923,
|
| 401 |
+
"mean_token_accuracy": 0.8295751377940178,
|
| 402 |
+
"step": 240
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.9884014120020171,
|
| 406 |
+
"grad_norm": 0.09508884698152542,
|
| 407 |
+
"learning_rate": 4.0049288167842705e-08,
|
| 408 |
+
"loss": 0.5046,
|
| 409 |
+
"mean_token_accuracy": 0.825819493830204,
|
| 410 |
+
"step": 245
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.9964699949571356,
|
| 414 |
+
"mean_token_accuracy": 0.825654674321413,
|
| 415 |
+
"step": 247,
|
| 416 |
+
"total_flos": 2.5792410000818176e+17,
|
| 417 |
+
"train_loss": 0.5621715438993353,
|
| 418 |
+
"train_runtime": 8702.0895,
|
| 419 |
+
"train_samples_per_second": 0.456,
|
| 420 |
"train_steps_per_second": 0.028
|
| 421 |
}
|
| 422 |
],
|
| 423 |
"logging_steps": 5,
|
| 424 |
+
"max_steps": 247,
|
| 425 |
"num_input_tokens_seen": 0,
|
| 426 |
"num_train_epochs": 1,
|
| 427 |
"save_steps": 100,
|
|
|
|
| 437 |
"attributes": {}
|
| 438 |
}
|
| 439 |
},
|
| 440 |
+
"total_flos": 2.5792410000818176e+17,
|
| 441 |
"train_batch_size": 1,
|
| 442 |
"trial_name": null,
|
| 443 |
"trial_params": null
|