Upload folder using huggingface_hub
Browse files- adapter_config.json +5 -5
- adapter_model.safetensors +1 -1
- optimizer.pt +2 -2
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +394 -4
- training_args.bin +1 -1
adapter_config.json
CHANGED
|
@@ -29,13 +29,13 @@
|
|
| 29 |
"rank_pattern": {},
|
| 30 |
"revision": null,
|
| 31 |
"target_modules": [
|
| 32 |
-
"
|
| 33 |
"v_proj",
|
| 34 |
-
"
|
| 35 |
"k_proj",
|
| 36 |
-
"
|
| 37 |
-
"
|
| 38 |
-
"
|
| 39 |
],
|
| 40 |
"task_type": "CAUSAL_LM",
|
| 41 |
"trainable_token_indices": null,
|
|
|
|
| 29 |
"rank_pattern": {},
|
| 30 |
"revision": null,
|
| 31 |
"target_modules": [
|
| 32 |
+
"up_proj",
|
| 33 |
"v_proj",
|
| 34 |
+
"q_proj",
|
| 35 |
"k_proj",
|
| 36 |
+
"gate_proj",
|
| 37 |
+
"down_proj",
|
| 38 |
+
"o_proj"
|
| 39 |
],
|
| 40 |
"task_type": "CAUSAL_LM",
|
| 41 |
"trainable_token_indices": null,
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 262406656
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:aac9ea0e9492c5121969cb8cd1243a9cbd537758d4beee2498d08aa1bfc7fb3b
|
| 3 |
size 262406656
|
optimizer.pt
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 sha256:4c1849c8173db31fb56cd75df79edfbc60942ddcc33bec2fbe2f50157adaf37e
|
| 3 |
+
size 122871883
|
rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14645
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:81888abba6fcb84e8644e14930eab357fa25630389fc14dcfeecc975e067404d
|
| 3 |
size 14645
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2fb49e328593b94208faae32bb0c92c92ac86ee6160811712359ffbd888f082c
|
| 3 |
size 1465
|
trainer_state.json
CHANGED
|
@@ -2,9 +2,9 @@
|
|
| 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,
|
|
@@ -138,11 +138,401 @@
|
|
| 138 |
"rewards/quality_reward_func/mean": 0.020000000298023225,
|
| 139 |
"rewards/quality_reward_func/std": 0.04000000059604645,
|
| 140 |
"step": 50
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 141 |
}
|
| 142 |
],
|
| 143 |
"logging_steps": 10,
|
| 144 |
-
"max_steps":
|
| 145 |
-
"num_input_tokens_seen":
|
| 146 |
"num_train_epochs": 1,
|
| 147 |
"save_steps": 50,
|
| 148 |
"stateful_callbacks": {
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.013751375137513752,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 200,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 138 |
"rewards/quality_reward_func/mean": 0.020000000298023225,
|
| 139 |
"rewards/quality_reward_func/std": 0.04000000059604645,
|
| 140 |
"step": 50
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"completion_length": 1.0,
|
| 144 |
+
"completions/clipped_ratio": 0.0,
|
| 145 |
+
"completions/max_length": 1.0,
|
| 146 |
+
"completions/max_terminated_length": 1.0,
|
| 147 |
+
"completions/mean_length": 1.0,
|
| 148 |
+
"completions/mean_terminated_length": 1.0,
|
| 149 |
+
"completions/min_length": 1.0,
|
| 150 |
+
"completions/min_terminated_length": 1.0,
|
| 151 |
+
"epoch": 0.004125412541254125,
|
| 152 |
+
"frac_reward_zero_std": 1.0,
|
| 153 |
+
"grad_norm": 0.0,
|
| 154 |
+
"kl": 1.6421079635620118e-06,
|
| 155 |
+
"learning_rate": 1.9666666666666668e-06,
|
| 156 |
+
"loss": 0.0,
|
| 157 |
+
"num_tokens": 79438.0,
|
| 158 |
+
"reward": 0.0,
|
| 159 |
+
"reward_std": 0.0,
|
| 160 |
+
"rewards/coherence_reward_func/mean": 0.0,
|
| 161 |
+
"rewards/coherence_reward_func/std": 0.0,
|
| 162 |
+
"rewards/formatting_reward_func/mean": 0.0,
|
| 163 |
+
"rewards/formatting_reward_func/std": 0.0,
|
| 164 |
+
"rewards/quality_reward_func/mean": 0.0,
|
| 165 |
+
"rewards/quality_reward_func/std": 0.0,
|
| 166 |
+
"step": 60
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"completion_length": 1.6,
|
| 170 |
+
"completions/clipped_ratio": 0.0,
|
| 171 |
+
"completions/max_length": 1.6,
|
| 172 |
+
"completions/max_terminated_length": 1.6,
|
| 173 |
+
"completions/mean_length": 1.6,
|
| 174 |
+
"completions/mean_terminated_length": 1.6,
|
| 175 |
+
"completions/min_length": 1.6,
|
| 176 |
+
"completions/min_terminated_length": 1.6,
|
| 177 |
+
"epoch": 0.004812981298129813,
|
| 178 |
+
"frac_reward_zero_std": 1.0,
|
| 179 |
+
"grad_norm": 0.0,
|
| 180 |
+
"kl": 1.8922347294392238e-06,
|
| 181 |
+
"learning_rate": 2.3000000000000004e-06,
|
| 182 |
+
"loss": 0.0,
|
| 183 |
+
"num_tokens": 92538.0,
|
| 184 |
+
"reward": 0.1100000023841858,
|
| 185 |
+
"reward_std": 0.0,
|
| 186 |
+
"rewards/coherence_reward_func/mean": 0.06000000238418579,
|
| 187 |
+
"rewards/coherence_reward_func/std": 0.0,
|
| 188 |
+
"rewards/formatting_reward_func/mean": 0.05,
|
| 189 |
+
"rewards/formatting_reward_func/std": 0.0,
|
| 190 |
+
"rewards/quality_reward_func/mean": 0.0,
|
| 191 |
+
"rewards/quality_reward_func/std": 0.0,
|
| 192 |
+
"step": 70
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"completion_length": 1.0,
|
| 196 |
+
"completions/clipped_ratio": 0.0,
|
| 197 |
+
"completions/max_length": 1.0,
|
| 198 |
+
"completions/max_terminated_length": 1.0,
|
| 199 |
+
"completions/mean_length": 1.0,
|
| 200 |
+
"completions/mean_terminated_length": 1.0,
|
| 201 |
+
"completions/min_length": 1.0,
|
| 202 |
+
"completions/min_terminated_length": 1.0,
|
| 203 |
+
"epoch": 0.005500550055005501,
|
| 204 |
+
"frac_reward_zero_std": 1.0,
|
| 205 |
+
"grad_norm": 0.0,
|
| 206 |
+
"kl": 1.56402587890625e-05,
|
| 207 |
+
"learning_rate": 2.6333333333333332e-06,
|
| 208 |
+
"loss": 0.0,
|
| 209 |
+
"num_tokens": 106926.0,
|
| 210 |
+
"reward": 0.0,
|
| 211 |
+
"reward_std": 0.0,
|
| 212 |
+
"rewards/coherence_reward_func/mean": 0.0,
|
| 213 |
+
"rewards/coherence_reward_func/std": 0.0,
|
| 214 |
+
"rewards/formatting_reward_func/mean": 0.0,
|
| 215 |
+
"rewards/formatting_reward_func/std": 0.0,
|
| 216 |
+
"rewards/quality_reward_func/mean": 0.0,
|
| 217 |
+
"rewards/quality_reward_func/std": 0.0,
|
| 218 |
+
"step": 80
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"completion_length": 1.0,
|
| 222 |
+
"completions/clipped_ratio": 0.0,
|
| 223 |
+
"completions/max_length": 1.0,
|
| 224 |
+
"completions/max_terminated_length": 1.0,
|
| 225 |
+
"completions/mean_length": 1.0,
|
| 226 |
+
"completions/mean_terminated_length": 1.0,
|
| 227 |
+
"completions/min_length": 1.0,
|
| 228 |
+
"completions/min_terminated_length": 1.0,
|
| 229 |
+
"epoch": 0.006188118811881188,
|
| 230 |
+
"frac_reward_zero_std": 1.0,
|
| 231 |
+
"grad_norm": 0.0,
|
| 232 |
+
"kl": 2.9624998569488524e-05,
|
| 233 |
+
"learning_rate": 2.9666666666666673e-06,
|
| 234 |
+
"loss": 0.0,
|
| 235 |
+
"num_tokens": 120118.0,
|
| 236 |
+
"reward": 0.0,
|
| 237 |
+
"reward_std": 0.0,
|
| 238 |
+
"rewards/coherence_reward_func/mean": 0.0,
|
| 239 |
+
"rewards/coherence_reward_func/std": 0.0,
|
| 240 |
+
"rewards/formatting_reward_func/mean": 0.0,
|
| 241 |
+
"rewards/formatting_reward_func/std": 0.0,
|
| 242 |
+
"rewards/quality_reward_func/mean": 0.0,
|
| 243 |
+
"rewards/quality_reward_func/std": 0.0,
|
| 244 |
+
"step": 90
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"completion_length": 2.0,
|
| 248 |
+
"completions/clipped_ratio": 0.0,
|
| 249 |
+
"completions/max_length": 2.0,
|
| 250 |
+
"completions/max_terminated_length": 2.0,
|
| 251 |
+
"completions/mean_length": 1.25,
|
| 252 |
+
"completions/mean_terminated_length": 1.25,
|
| 253 |
+
"completions/min_length": 1.0,
|
| 254 |
+
"completions/min_terminated_length": 1.0,
|
| 255 |
+
"epoch": 0.006875687568756876,
|
| 256 |
+
"frac_reward_zero_std": 0.9,
|
| 257 |
+
"grad_norm": 0.0,
|
| 258 |
+
"kl": 0.0018734597397269681,
|
| 259 |
+
"learning_rate": 3.3000000000000006e-06,
|
| 260 |
+
"loss": 0.0,
|
| 261 |
+
"num_tokens": 134128.0,
|
| 262 |
+
"reward": 0.0899999976158142,
|
| 263 |
+
"reward_std": 0.1799999952316284,
|
| 264 |
+
"rewards/coherence_reward_func/mean": 0.032499998807907104,
|
| 265 |
+
"rewards/coherence_reward_func/std": 0.06499999761581421,
|
| 266 |
+
"rewards/formatting_reward_func/mean": 0.05,
|
| 267 |
+
"rewards/formatting_reward_func/std": 0.1,
|
| 268 |
+
"rewards/quality_reward_func/mean": 0.007500000298023224,
|
| 269 |
+
"rewards/quality_reward_func/std": 0.015000002086162567,
|
| 270 |
+
"step": 100
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"completion_length": 5.9,
|
| 274 |
+
"completions/clipped_ratio": 0.0,
|
| 275 |
+
"completions/max_length": 5.9,
|
| 276 |
+
"completions/max_terminated_length": 5.9,
|
| 277 |
+
"completions/mean_length": 3.5,
|
| 278 |
+
"completions/mean_terminated_length": 3.5,
|
| 279 |
+
"completions/min_length": 1.0,
|
| 280 |
+
"completions/min_terminated_length": 1.0,
|
| 281 |
+
"epoch": 0.007563256325632563,
|
| 282 |
+
"frac_reward_zero_std": 0.7,
|
| 283 |
+
"grad_norm": 0.5969054698944092,
|
| 284 |
+
"kl": 0.3142867418937385,
|
| 285 |
+
"learning_rate": 3.633333333333334e-06,
|
| 286 |
+
"loss": 0.0,
|
| 287 |
+
"num_tokens": 149936.0,
|
| 288 |
+
"reward": 0.6974999606609344,
|
| 289 |
+
"reward_std": 0.574999988079071,
|
| 290 |
+
"rewards/coherence_reward_func/mean": 0.21999999284744262,
|
| 291 |
+
"rewards/coherence_reward_func/std": 0.1799999952316284,
|
| 292 |
+
"rewards/formatting_reward_func/mean": 0.3375,
|
| 293 |
+
"rewards/formatting_reward_func/std": 0.275,
|
| 294 |
+
"rewards/quality_reward_func/mean": 0.14000000506639482,
|
| 295 |
+
"rewards/quality_reward_func/std": 0.12000000178813934,
|
| 296 |
+
"step": 110
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"completion_length": 10.3,
|
| 300 |
+
"completions/clipped_ratio": 0.0,
|
| 301 |
+
"completions/max_length": 10.3,
|
| 302 |
+
"completions/max_terminated_length": 10.3,
|
| 303 |
+
"completions/mean_length": 7.55,
|
| 304 |
+
"completions/mean_terminated_length": 7.55,
|
| 305 |
+
"completions/min_length": 6.0,
|
| 306 |
+
"completions/min_terminated_length": 6.0,
|
| 307 |
+
"epoch": 0.00825082508250825,
|
| 308 |
+
"frac_reward_zero_std": 0.7,
|
| 309 |
+
"grad_norm": 0.0026277885772287846,
|
| 310 |
+
"kl": 470.12245586041826,
|
| 311 |
+
"learning_rate": 3.966666666666667e-06,
|
| 312 |
+
"loss": 0.0119,
|
| 313 |
+
"num_tokens": 164358.0,
|
| 314 |
+
"reward": 1.6574999570846558,
|
| 315 |
+
"reward_std": 0.27903410643339155,
|
| 316 |
+
"rewards/coherence_reward_func/mean": 0.5399999856948853,
|
| 317 |
+
"rewards/coherence_reward_func/std": 0.09237603992223739,
|
| 318 |
+
"rewards/formatting_reward_func/mean": 0.8375,
|
| 319 |
+
"rewards/formatting_reward_func/std": 0.1404700517654419,
|
| 320 |
+
"rewards/quality_reward_func/mean": 0.2800000041723251,
|
| 321 |
+
"rewards/quality_reward_func/std": 0.046188023686408994,
|
| 322 |
+
"step": 120
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"completion_length": 18.6,
|
| 326 |
+
"completions/clipped_ratio": 0.0,
|
| 327 |
+
"completions/max_length": 18.6,
|
| 328 |
+
"completions/max_terminated_length": 18.6,
|
| 329 |
+
"completions/mean_length": 15.8,
|
| 330 |
+
"completions/mean_terminated_length": 15.8,
|
| 331 |
+
"completions/min_length": 11.9,
|
| 332 |
+
"completions/min_terminated_length": 11.9,
|
| 333 |
+
"epoch": 0.008938393839383938,
|
| 334 |
+
"frac_reward_zero_std": 0.8,
|
| 335 |
+
"grad_norm": 0.0,
|
| 336 |
+
"kl": 0.46882193982601167,
|
| 337 |
+
"learning_rate": 4.3e-06,
|
| 338 |
+
"loss": 0.0,
|
| 339 |
+
"num_tokens": 178634.0,
|
| 340 |
+
"reward": 3.5749999046325684,
|
| 341 |
+
"reward_std": 0.22999999523162842,
|
| 342 |
+
"rewards/coherence_reward_func/mean": 1.1374999582767487,
|
| 343 |
+
"rewards/coherence_reward_func/std": 0.06499999761581421,
|
| 344 |
+
"rewards/formatting_reward_func/mean": 1.75,
|
| 345 |
+
"rewards/formatting_reward_func/std": 0.1,
|
| 346 |
+
"rewards/quality_reward_func/mean": 0.6875000119209289,
|
| 347 |
+
"rewards/quality_reward_func/std": 0.06500000059604645,
|
| 348 |
+
"step": 130
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"completion_length": 20.6,
|
| 352 |
+
"completions/clipped_ratio": 0.0,
|
| 353 |
+
"completions/max_length": 20.6,
|
| 354 |
+
"completions/max_terminated_length": 20.6,
|
| 355 |
+
"completions/mean_length": 16.4,
|
| 356 |
+
"completions/mean_terminated_length": 16.4,
|
| 357 |
+
"completions/min_length": 11.8,
|
| 358 |
+
"completions/min_terminated_length": 11.8,
|
| 359 |
+
"epoch": 0.009625962596259627,
|
| 360 |
+
"frac_reward_zero_std": 0.7,
|
| 361 |
+
"grad_norm": 0.0,
|
| 362 |
+
"kl": 0.4925146855413914,
|
| 363 |
+
"learning_rate": 4.633333333333334e-06,
|
| 364 |
+
"loss": 0.0,
|
| 365 |
+
"num_tokens": 192926.0,
|
| 366 |
+
"reward": 3.2499999046325683,
|
| 367 |
+
"reward_std": 0.44126754999160767,
|
| 368 |
+
"rewards/coherence_reward_func/mean": 1.0424999654293061,
|
| 369 |
+
"rewards/coherence_reward_func/std": 0.14356523752212524,
|
| 370 |
+
"rewards/formatting_reward_func/mean": 1.6125,
|
| 371 |
+
"rewards/formatting_reward_func/std": 0.21160253882408142,
|
| 372 |
+
"rewards/quality_reward_func/mean": 0.595000010728836,
|
| 373 |
+
"rewards/quality_reward_func/std": 0.10350853204727173,
|
| 374 |
+
"step": 140
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"completion_length": 23.0,
|
| 378 |
+
"completions/clipped_ratio": 0.0,
|
| 379 |
+
"completions/max_length": 23.0,
|
| 380 |
+
"completions/max_terminated_length": 23.0,
|
| 381 |
+
"completions/mean_length": 17.825,
|
| 382 |
+
"completions/mean_terminated_length": 17.825,
|
| 383 |
+
"completions/min_length": 13.0,
|
| 384 |
+
"completions/min_terminated_length": 13.0,
|
| 385 |
+
"epoch": 0.010313531353135313,
|
| 386 |
+
"frac_reward_zero_std": 0.9,
|
| 387 |
+
"grad_norm": 0.0,
|
| 388 |
+
"kl": 1.0868607074022294,
|
| 389 |
+
"learning_rate": 4.966666666666667e-06,
|
| 390 |
+
"loss": 0.0,
|
| 391 |
+
"num_tokens": 206363.0,
|
| 392 |
+
"reward": 4.059999895095825,
|
| 393 |
+
"reward_std": 0.07999999523162842,
|
| 394 |
+
"rewards/coherence_reward_func/mean": 1.2924999475479126,
|
| 395 |
+
"rewards/coherence_reward_func/std": 0.01499999761581421,
|
| 396 |
+
"rewards/formatting_reward_func/mean": 1.9875,
|
| 397 |
+
"rewards/formatting_reward_func/std": 0.025,
|
| 398 |
+
"rewards/quality_reward_func/mean": 0.7800000131130218,
|
| 399 |
+
"rewards/quality_reward_func/std": 0.04000000059604645,
|
| 400 |
+
"step": 150
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"completion_length": 31.1,
|
| 404 |
+
"completions/clipped_ratio": 0.0,
|
| 405 |
+
"completions/max_length": 31.1,
|
| 406 |
+
"completions/max_terminated_length": 31.1,
|
| 407 |
+
"completions/mean_length": 20.3,
|
| 408 |
+
"completions/mean_terminated_length": 20.3,
|
| 409 |
+
"completions/min_length": 14.3,
|
| 410 |
+
"completions/min_terminated_length": 14.3,
|
| 411 |
+
"epoch": 0.011001100110011002,
|
| 412 |
+
"frac_reward_zero_std": 0.9,
|
| 413 |
+
"grad_norm": 0.0007889735861681402,
|
| 414 |
+
"kl": 0.7975522613618523,
|
| 415 |
+
"learning_rate": 4.999451708687114e-06,
|
| 416 |
+
"loss": 0.0,
|
| 417 |
+
"num_tokens": 219259.0,
|
| 418 |
+
"reward": 4.079999899864196,
|
| 419 |
+
"reward_std": 0.04000000059604645,
|
| 420 |
+
"rewards/coherence_reward_func/mean": 1.2924999475479126,
|
| 421 |
+
"rewards/coherence_reward_func/std": 0.01499999761581421,
|
| 422 |
+
"rewards/formatting_reward_func/mean": 1.9875,
|
| 423 |
+
"rewards/formatting_reward_func/std": 0.025,
|
| 424 |
+
"rewards/quality_reward_func/mean": 0.800000011920929,
|
| 425 |
+
"rewards/quality_reward_func/std": 0.0,
|
| 426 |
+
"step": 160
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"completion_length": 34.6,
|
| 430 |
+
"completions/clipped_ratio": 0.0,
|
| 431 |
+
"completions/max_length": 34.6,
|
| 432 |
+
"completions/max_terminated_length": 34.6,
|
| 433 |
+
"completions/mean_length": 23.1,
|
| 434 |
+
"completions/mean_terminated_length": 23.1,
|
| 435 |
+
"completions/min_length": 17.5,
|
| 436 |
+
"completions/min_terminated_length": 17.5,
|
| 437 |
+
"epoch": 0.011688668866886688,
|
| 438 |
+
"frac_reward_zero_std": 0.8,
|
| 439 |
+
"grad_norm": 0.419414222240448,
|
| 440 |
+
"kl": 0.7056910984218121,
|
| 441 |
+
"learning_rate": 4.9975566894538954e-06,
|
| 442 |
+
"loss": 0.0,
|
| 443 |
+
"num_tokens": 235091.0,
|
| 444 |
+
"reward": 4.077499866485596,
|
| 445 |
+
"reward_std": 0.03232050836086273,
|
| 446 |
+
"rewards/coherence_reward_func/mean": 1.2774999499320985,
|
| 447 |
+
"rewards/coherence_reward_func/std": 0.03232050389051437,
|
| 448 |
+
"rewards/formatting_reward_func/mean": 2.0,
|
| 449 |
+
"rewards/formatting_reward_func/std": 0.0,
|
| 450 |
+
"rewards/quality_reward_func/mean": 0.800000011920929,
|
| 451 |
+
"rewards/quality_reward_func/std": 0.0,
|
| 452 |
+
"step": 170
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"completion_length": 22.9,
|
| 456 |
+
"completions/clipped_ratio": 0.0,
|
| 457 |
+
"completions/max_length": 22.9,
|
| 458 |
+
"completions/max_terminated_length": 22.9,
|
| 459 |
+
"completions/mean_length": 18.05,
|
| 460 |
+
"completions/mean_terminated_length": 18.05,
|
| 461 |
+
"completions/min_length": 14.7,
|
| 462 |
+
"completions/min_terminated_length": 14.7,
|
| 463 |
+
"epoch": 0.012376237623762377,
|
| 464 |
+
"frac_reward_zero_std": 0.9,
|
| 465 |
+
"grad_norm": 0.0,
|
| 466 |
+
"kl": 1124.961197933089,
|
| 467 |
+
"learning_rate": 4.994309199213748e-06,
|
| 468 |
+
"loss": 0.0461,
|
| 469 |
+
"num_tokens": 247253.0,
|
| 470 |
+
"reward": 4.074999904632568,
|
| 471 |
+
"reward_std": 0.028867512941360474,
|
| 472 |
+
"rewards/coherence_reward_func/mean": 1.2999999523162842,
|
| 473 |
+
"rewards/coherence_reward_func/std": 0.0,
|
| 474 |
+
"rewards/formatting_reward_func/mean": 2.0,
|
| 475 |
+
"rewards/formatting_reward_func/std": 0.0,
|
| 476 |
+
"rewards/quality_reward_func/mean": 0.7750000119209289,
|
| 477 |
+
"rewards/quality_reward_func/std": 0.028867512941360474,
|
| 478 |
+
"step": 180
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"completion_length": 22.5,
|
| 482 |
+
"completions/clipped_ratio": 0.0,
|
| 483 |
+
"completions/max_length": 22.5,
|
| 484 |
+
"completions/max_terminated_length": 22.5,
|
| 485 |
+
"completions/mean_length": 20.05,
|
| 486 |
+
"completions/mean_terminated_length": 20.05,
|
| 487 |
+
"completions/min_length": 17.4,
|
| 488 |
+
"completions/min_terminated_length": 17.4,
|
| 489 |
+
"epoch": 0.013063806380638063,
|
| 490 |
+
"frac_reward_zero_std": 1.0,
|
| 491 |
+
"grad_norm": 0.0,
|
| 492 |
+
"kl": 0.693901395983994,
|
| 493 |
+
"learning_rate": 4.989710996539926e-06,
|
| 494 |
+
"loss": 0.0,
|
| 495 |
+
"num_tokens": 261675.0,
|
| 496 |
+
"reward": 4.099999904632568,
|
| 497 |
+
"reward_std": 0.0,
|
| 498 |
+
"rewards/coherence_reward_func/mean": 1.2999999523162842,
|
| 499 |
+
"rewards/coherence_reward_func/std": 0.0,
|
| 500 |
+
"rewards/formatting_reward_func/mean": 2.0,
|
| 501 |
+
"rewards/formatting_reward_func/std": 0.0,
|
| 502 |
+
"rewards/quality_reward_func/mean": 0.800000011920929,
|
| 503 |
+
"rewards/quality_reward_func/std": 0.0,
|
| 504 |
+
"step": 190
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"completion_length": 25.2,
|
| 508 |
+
"completions/clipped_ratio": 0.0,
|
| 509 |
+
"completions/max_length": 25.2,
|
| 510 |
+
"completions/max_terminated_length": 25.2,
|
| 511 |
+
"completions/mean_length": 21.575,
|
| 512 |
+
"completions/mean_terminated_length": 21.575,
|
| 513 |
+
"completions/min_length": 17.8,
|
| 514 |
+
"completions/min_terminated_length": 17.8,
|
| 515 |
+
"epoch": 0.013751375137513752,
|
| 516 |
+
"frac_reward_zero_std": 1.0,
|
| 517 |
+
"grad_norm": 0.0,
|
| 518 |
+
"kl": 0.7635734604671598,
|
| 519 |
+
"learning_rate": 4.983764571440296e-06,
|
| 520 |
+
"loss": 0.0,
|
| 521 |
+
"num_tokens": 277602.0,
|
| 522 |
+
"reward": 4.099999904632568,
|
| 523 |
+
"reward_std": 0.0,
|
| 524 |
+
"rewards/coherence_reward_func/mean": 1.2999999523162842,
|
| 525 |
+
"rewards/coherence_reward_func/std": 0.0,
|
| 526 |
+
"rewards/formatting_reward_func/mean": 2.0,
|
| 527 |
+
"rewards/formatting_reward_func/std": 0.0,
|
| 528 |
+
"rewards/quality_reward_func/mean": 0.800000011920929,
|
| 529 |
+
"rewards/quality_reward_func/std": 0.0,
|
| 530 |
+
"step": 200
|
| 531 |
}
|
| 532 |
],
|
| 533 |
"logging_steps": 10,
|
| 534 |
+
"max_steps": 1500,
|
| 535 |
+
"num_input_tokens_seen": 277602,
|
| 536 |
"num_train_epochs": 1,
|
| 537 |
"save_steps": 50,
|
| 538 |
"stateful_callbacks": {
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 7057
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:517ca74ddbfa64fa78f010d3585be0b071bd656ca06cca4c198fbdfb003305e6
|
| 3 |
size 7057
|