Upload task output 1
Browse files- adapter_config.json +5 -5
- adapter_model.safetensors +1 -1
- loss.txt +1 -1
- tokenizer.json +2 -2
- trainer_state.json +217 -1720
- training_args.bin +1 -1
adapter_config.json
CHANGED
|
@@ -29,13 +29,13 @@
|
|
| 29 |
"rank_pattern": {},
|
| 30 |
"revision": null,
|
| 31 |
"target_modules": [
|
| 32 |
-
"
|
| 33 |
-
"down_proj",
|
| 34 |
-
"up_proj",
|
| 35 |
"o_proj",
|
|
|
|
|
|
|
| 36 |
"k_proj",
|
| 37 |
-
"
|
| 38 |
-
"
|
| 39 |
],
|
| 40 |
"target_parameters": null,
|
| 41 |
"task_type": "CAUSAL_LM",
|
|
|
|
| 29 |
"rank_pattern": {},
|
| 30 |
"revision": null,
|
| 31 |
"target_modules": [
|
| 32 |
+
"gate_proj",
|
|
|
|
|
|
|
| 33 |
"o_proj",
|
| 34 |
+
"v_proj",
|
| 35 |
+
"q_proj",
|
| 36 |
"k_proj",
|
| 37 |
+
"up_proj",
|
| 38 |
+
"down_proj"
|
| 39 |
],
|
| 40 |
"target_parameters": null,
|
| 41 |
"task_type": "CAUSAL_LM",
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 319876032
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4ec359459fe517709f85e49817da65dc6bc2ef784dff0e2f3f9dd5a384abdc9d
|
| 3 |
size 319876032
|
loss.txt
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
|
|
|
|
| 1 |
+
10,18.063774490356444
|
tokenizer.json
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:ef304ea6a0a92c4275c959dd6e1a896bca7e36af0adb2ce25563e7c11bf3927f
|
| 3 |
+
size 3620829
|
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,
|
|
@@ -15,32 +15,32 @@
|
|
| 15 |
"clip_ratio/low_mean": 0.0,
|
| 16 |
"clip_ratio/low_min": 0.0,
|
| 17 |
"clip_ratio/region_mean": 0.0,
|
| 18 |
-
"completions/clipped_ratio": 0.
|
| 19 |
"completions/max_length": 16.0,
|
| 20 |
-
"completions/max_terminated_length":
|
| 21 |
-
"completions/mean_length":
|
| 22 |
-
"completions/mean_terminated_length":
|
| 23 |
-
"completions/min_length":
|
| 24 |
-
"completions/min_terminated_length":
|
| 25 |
-
"entropy": 0.
|
| 26 |
"epoch": 2e-05,
|
| 27 |
"frac_reward_zero_std": 0.0,
|
| 28 |
-
"grad_norm": 0.
|
| 29 |
"kl": 0.0,
|
| 30 |
"learning_rate": 0.0,
|
| 31 |
-
"loss": 0.
|
| 32 |
-
"num_tokens":
|
| 33 |
-
"reward": -
|
| 34 |
-
"reward_std":
|
| 35 |
-
"rewards/rollout_reward_func/mean": -
|
| 36 |
-
"rewards/rollout_reward_func/std":
|
| 37 |
-
"sampling/importance_sampling_ratio/max": 1.
|
| 38 |
-
"sampling/importance_sampling_ratio/mean": 1.
|
| 39 |
-
"sampling/importance_sampling_ratio/min": 0.
|
| 40 |
-
"sampling/sampling_logp_difference/max": 0.
|
| 41 |
-
"sampling/sampling_logp_difference/mean": 0.
|
| 42 |
"step": 1,
|
| 43 |
-
"step_time": 13.
|
| 44 |
},
|
| 45 |
{
|
| 46 |
"clip_ratio/high_max": 0.0,
|
|
@@ -48,302 +48,143 @@
|
|
| 48 |
"clip_ratio/low_mean": 0.0,
|
| 49 |
"clip_ratio/low_min": 0.0,
|
| 50 |
"clip_ratio/region_mean": 0.0,
|
| 51 |
-
"entropy": 0.
|
| 52 |
"epoch": 4e-05,
|
| 53 |
-
"grad_norm": 0.
|
| 54 |
"kl": 0.0,
|
| 55 |
-
"learning_rate": 8.
|
| 56 |
-
"loss": 0.
|
| 57 |
"step": 2,
|
| 58 |
-
"step_time": 3.
|
| 59 |
},
|
| 60 |
{
|
| 61 |
"clip_ratio/high_max": 0.0,
|
| 62 |
"clip_ratio/high_mean": 0.0,
|
| 63 |
-
"clip_ratio/low_mean": 0.
|
| 64 |
"clip_ratio/low_min": 0.0,
|
| 65 |
-
"clip_ratio/region_mean": 0.
|
| 66 |
"completions/clipped_ratio": 0.75,
|
| 67 |
"completions/max_length": 16.0,
|
| 68 |
-
"completions/max_terminated_length":
|
| 69 |
-
"completions/mean_length": 14.
|
| 70 |
-
"completions/mean_terminated_length":
|
| 71 |
-
"completions/min_length":
|
| 72 |
-
"completions/min_terminated_length":
|
| 73 |
-
"entropy": 0.
|
| 74 |
"epoch": 6e-05,
|
| 75 |
"frac_reward_zero_std": 0.0,
|
| 76 |
-
"grad_norm": 0.
|
| 77 |
-
"kl": 0.
|
| 78 |
-
"learning_rate": 1.
|
| 79 |
-
"loss": 0.
|
| 80 |
-
"num_tokens":
|
| 81 |
-
"reward": -
|
| 82 |
-
"reward_std":
|
| 83 |
-
"rewards/rollout_reward_func/mean": -
|
| 84 |
-
"rewards/rollout_reward_func/std":
|
| 85 |
-
"sampling/importance_sampling_ratio/max": 1.
|
| 86 |
-
"sampling/importance_sampling_ratio/mean": 0.
|
| 87 |
-
"sampling/importance_sampling_ratio/min": 0.
|
| 88 |
-
"sampling/sampling_logp_difference/max": 0.
|
| 89 |
-
"sampling/sampling_logp_difference/mean": 0.
|
| 90 |
"step": 3,
|
| 91 |
-
"step_time": 13.
|
| 92 |
},
|
| 93 |
{
|
| 94 |
-
"clip_ratio/high_max": 0.
|
| 95 |
-
"clip_ratio/high_mean": 0.
|
| 96 |
"clip_ratio/low_mean": 0.0,
|
| 97 |
"clip_ratio/low_min": 0.0,
|
| 98 |
-
"clip_ratio/region_mean": 0.
|
| 99 |
-
"entropy": 0.
|
| 100 |
"epoch": 8e-05,
|
| 101 |
-
"grad_norm": 0.
|
| 102 |
-
"kl": 0.
|
| 103 |
-
"learning_rate": 2.
|
| 104 |
-
"loss": 0.
|
| 105 |
"step": 4,
|
| 106 |
-
"step_time": 3.
|
| 107 |
},
|
| 108 |
{
|
| 109 |
-
"clip_ratio/high_max": 0.
|
| 110 |
-
"clip_ratio/high_mean": 0.
|
| 111 |
"clip_ratio/low_mean": 0.0,
|
| 112 |
"clip_ratio/low_min": 0.0,
|
| 113 |
-
"clip_ratio/region_mean": 0.
|
| 114 |
-
"completions/clipped_ratio": 0.
|
| 115 |
"completions/max_length": 16.0,
|
| 116 |
-
"completions/max_terminated_length":
|
| 117 |
-
"completions/mean_length":
|
| 118 |
-
"completions/mean_terminated_length":
|
| 119 |
"completions/min_length": 4.0,
|
| 120 |
"completions/min_terminated_length": 4.0,
|
| 121 |
-
"entropy": 0.
|
| 122 |
"epoch": 0.0001,
|
| 123 |
"frac_reward_zero_std": 0.0,
|
| 124 |
-
"grad_norm": 0.
|
| 125 |
-
"kl": 0.
|
| 126 |
-
"learning_rate": 3.
|
| 127 |
-
"loss":
|
| 128 |
-
"num_tokens":
|
| 129 |
-
"reward": -
|
| 130 |
-
"reward_std":
|
| 131 |
-
"rewards/rollout_reward_func/mean": -
|
| 132 |
-
"rewards/rollout_reward_func/std":
|
| 133 |
-
"sampling/importance_sampling_ratio/max": 1.
|
| 134 |
-
"sampling/importance_sampling_ratio/mean": 0.
|
| 135 |
-
"sampling/importance_sampling_ratio/min": 0.
|
| 136 |
-
"sampling/sampling_logp_difference/max": 0.
|
| 137 |
-
"sampling/sampling_logp_difference/mean": 0.
|
| 138 |
"step": 5,
|
| 139 |
-
"step_time":
|
| 140 |
},
|
| 141 |
{
|
| 142 |
-
"
|
| 143 |
-
"
|
| 144 |
-
"
|
| 145 |
-
"
|
| 146 |
-
"
|
| 147 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 148 |
"epoch": 0.00012,
|
| 149 |
-
"grad_norm": 0.
|
| 150 |
-
"kl": 0.
|
| 151 |
-
"learning_rate": 4.
|
| 152 |
-
"loss":
|
| 153 |
"step": 6,
|
| 154 |
-
"step_time": 3.
|
| 155 |
-
},
|
| 156 |
-
{
|
| 157 |
-
"clip_ratio/high_max": 0.008620689623057842,
|
| 158 |
-
"clip_ratio/high_mean": 0.0021551724057644606,
|
| 159 |
-
"clip_ratio/low_mean": 0.0,
|
| 160 |
-
"clip_ratio/low_min": 0.0,
|
| 161 |
-
"clip_ratio/region_mean": 0.0021551724057644606,
|
| 162 |
-
"completions/clipped_ratio": 0.6875,
|
| 163 |
-
"completions/max_length": 16.0,
|
| 164 |
-
"completions/max_terminated_length": 14.0,
|
| 165 |
-
"completions/mean_length": 13.6875,
|
| 166 |
-
"completions/mean_terminated_length": 8.600000381469727,
|
| 167 |
-
"completions/min_length": 4.0,
|
| 168 |
-
"completions/min_terminated_length": 4.0,
|
| 169 |
-
"entropy": 0.42659611254930496,
|
| 170 |
-
"epoch": 0.00014,
|
| 171 |
-
"frac_reward_zero_std": 0.0,
|
| 172 |
-
"grad_norm": 0.47957348823547363,
|
| 173 |
-
"kl": 0.00030913886985217687,
|
| 174 |
-
"learning_rate": 5.142857142857143e-07,
|
| 175 |
-
"loss": 0.1087,
|
| 176 |
-
"num_tokens": 209683.0,
|
| 177 |
-
"reward": -21.248210906982422,
|
| 178 |
-
"reward_std": 16.31087303161621,
|
| 179 |
-
"rewards/rollout_reward_func/mean": -21.248210906982422,
|
| 180 |
-
"rewards/rollout_reward_func/std": 16.31087303161621,
|
| 181 |
-
"sampling/importance_sampling_ratio/max": 1.2203210592269897,
|
| 182 |
-
"sampling/importance_sampling_ratio/mean": 0.9698781967163086,
|
| 183 |
-
"sampling/importance_sampling_ratio/min": 0.6302762627601624,
|
| 184 |
-
"sampling/sampling_logp_difference/max": 0.24562978744506836,
|
| 185 |
-
"sampling/sampling_logp_difference/mean": 0.011444764211773872,
|
| 186 |
-
"step": 7,
|
| 187 |
-
"step_time": 13.088727697000195
|
| 188 |
-
},
|
| 189 |
-
{
|
| 190 |
-
"clip_ratio/high_max": 0.008620689623057842,
|
| 191 |
-
"clip_ratio/high_mean": 0.004108297638595104,
|
| 192 |
-
"clip_ratio/low_mean": 0.0,
|
| 193 |
-
"clip_ratio/low_min": 0.0,
|
| 194 |
-
"clip_ratio/region_mean": 0.004108297638595104,
|
| 195 |
-
"entropy": 0.4278160482645035,
|
| 196 |
-
"epoch": 0.00016,
|
| 197 |
-
"grad_norm": 0.4772587716579437,
|
| 198 |
-
"kl": 0.000408434629207477,
|
| 199 |
-
"learning_rate": 6.000000000000001e-07,
|
| 200 |
-
"loss": 0.108,
|
| 201 |
-
"step": 8,
|
| 202 |
-
"step_time": 4.637348582999948
|
| 203 |
-
},
|
| 204 |
-
{
|
| 205 |
-
"clip_ratio/high_max": 0.008064515888690948,
|
| 206 |
-
"clip_ratio/high_mean": 0.002016128972172737,
|
| 207 |
-
"clip_ratio/low_mean": 0.0,
|
| 208 |
-
"clip_ratio/low_min": 0.0,
|
| 209 |
-
"clip_ratio/region_mean": 0.002016128972172737,
|
| 210 |
-
"completions/clipped_ratio": 0.84375,
|
| 211 |
-
"completions/max_length": 16.0,
|
| 212 |
-
"completions/max_terminated_length": 15.0,
|
| 213 |
-
"completions/mean_length": 15.34375,
|
| 214 |
-
"completions/mean_terminated_length": 11.800000190734863,
|
| 215 |
-
"completions/min_length": 6.0,
|
| 216 |
-
"completions/min_terminated_length": 6.0,
|
| 217 |
-
"entropy": 0.39533767104148865,
|
| 218 |
-
"epoch": 0.00018,
|
| 219 |
-
"frac_reward_zero_std": 0.0,
|
| 220 |
-
"grad_norm": 0.4479968845844269,
|
| 221 |
-
"kl": 0.0003632318548625335,
|
| 222 |
-
"learning_rate": 6.857142857142857e-07,
|
| 223 |
-
"loss": 0.0331,
|
| 224 |
-
"num_tokens": 262786.0,
|
| 225 |
-
"reward": -20.36313247680664,
|
| 226 |
-
"reward_std": 12.939953804016113,
|
| 227 |
-
"rewards/rollout_reward_func/mean": -20.36313247680664,
|
| 228 |
-
"rewards/rollout_reward_func/std": 12.939953804016113,
|
| 229 |
-
"sampling/importance_sampling_ratio/max": 1.270872712135315,
|
| 230 |
-
"sampling/importance_sampling_ratio/mean": 1.0103130340576172,
|
| 231 |
-
"sampling/importance_sampling_ratio/min": 0.8130013346672058,
|
| 232 |
-
"sampling/sampling_logp_difference/max": 0.18152618408203125,
|
| 233 |
-
"sampling/sampling_logp_difference/mean": 0.010795784182846546,
|
| 234 |
-
"step": 9,
|
| 235 |
-
"step_time": 12.636375802000202
|
| 236 |
-
},
|
| 237 |
-
{
|
| 238 |
-
"clip_ratio/high_max": 0.0,
|
| 239 |
-
"clip_ratio/high_mean": 0.0,
|
| 240 |
-
"clip_ratio/low_mean": 0.0,
|
| 241 |
-
"clip_ratio/low_min": 0.0,
|
| 242 |
-
"clip_ratio/region_mean": 0.0,
|
| 243 |
-
"entropy": 0.3951461538672447,
|
| 244 |
-
"epoch": 0.0002,
|
| 245 |
-
"grad_norm": 0.4540300667285919,
|
| 246 |
-
"kl": 0.00028537910839077085,
|
| 247 |
-
"learning_rate": 7.714285714285714e-07,
|
| 248 |
-
"loss": 0.0341,
|
| 249 |
-
"step": 10,
|
| 250 |
-
"step_time": 3.5246556089996375
|
| 251 |
-
},
|
| 252 |
-
{
|
| 253 |
-
"clip_ratio/high_max": 0.0078125,
|
| 254 |
-
"clip_ratio/high_mean": 0.001953125,
|
| 255 |
-
"clip_ratio/low_mean": 0.0,
|
| 256 |
-
"clip_ratio/low_min": 0.0,
|
| 257 |
-
"clip_ratio/region_mean": 0.001953125,
|
| 258 |
-
"completions/clipped_ratio": 0.84375,
|
| 259 |
-
"completions/max_length": 16.0,
|
| 260 |
-
"completions/max_terminated_length": 16.0,
|
| 261 |
-
"completions/mean_length": 15.09375,
|
| 262 |
-
"completions/mean_terminated_length": 10.199999809265137,
|
| 263 |
-
"completions/min_length": 4.0,
|
| 264 |
-
"completions/min_terminated_length": 4.0,
|
| 265 |
-
"entropy": 0.44926057010889053,
|
| 266 |
-
"epoch": 0.00022,
|
| 267 |
-
"frac_reward_zero_std": 0.0,
|
| 268 |
-
"grad_norm": 0.6062055826187134,
|
| 269 |
-
"kl": 0.0004280444991309196,
|
| 270 |
-
"learning_rate": 8.571428571428571e-07,
|
| 271 |
-
"loss": 0.1022,
|
| 272 |
-
"num_tokens": 316291.0,
|
| 273 |
-
"reward": -22.426130294799805,
|
| 274 |
-
"reward_std": 11.176600456237793,
|
| 275 |
-
"rewards/rollout_reward_func/mean": -22.426130294799805,
|
| 276 |
-
"rewards/rollout_reward_func/std": 11.17660140991211,
|
| 277 |
-
"sampling/importance_sampling_ratio/max": 1.3410085439682007,
|
| 278 |
-
"sampling/importance_sampling_ratio/mean": 1.018998384475708,
|
| 279 |
-
"sampling/importance_sampling_ratio/min": 0.8014979362487793,
|
| 280 |
-
"sampling/sampling_logp_difference/max": 0.26187610626220703,
|
| 281 |
-
"sampling/sampling_logp_difference/mean": 0.011133530177175999,
|
| 282 |
-
"step": 11,
|
| 283 |
-
"step_time": 13.926207335999152
|
| 284 |
-
},
|
| 285 |
-
{
|
| 286 |
-
"clip_ratio/high_max": 0.0,
|
| 287 |
-
"clip_ratio/high_mean": 0.0,
|
| 288 |
-
"clip_ratio/low_mean": 0.0,
|
| 289 |
-
"clip_ratio/low_min": 0.0,
|
| 290 |
-
"clip_ratio/region_mean": 0.0,
|
| 291 |
-
"entropy": 0.4503633603453636,
|
| 292 |
-
"epoch": 0.00024,
|
| 293 |
-
"grad_norm": 0.6362693905830383,
|
| 294 |
-
"kl": 0.0003500747407088056,
|
| 295 |
-
"learning_rate": 9.428571428571429e-07,
|
| 296 |
-
"loss": 0.1016,
|
| 297 |
-
"step": 12,
|
| 298 |
-
"step_time": 3.686019106999538
|
| 299 |
-
},
|
| 300 |
-
{
|
| 301 |
-
"clip_ratio/high_max": 0.0,
|
| 302 |
-
"clip_ratio/high_mean": 0.0,
|
| 303 |
-
"clip_ratio/low_mean": 0.0,
|
| 304 |
-
"clip_ratio/low_min": 0.0,
|
| 305 |
-
"clip_ratio/region_mean": 0.0,
|
| 306 |
-
"completions/clipped_ratio": 0.84375,
|
| 307 |
-
"completions/max_length": 16.0,
|
| 308 |
-
"completions/max_terminated_length": 15.0,
|
| 309 |
-
"completions/mean_length": 15.03125,
|
| 310 |
-
"completions/mean_terminated_length": 9.800000190734863,
|
| 311 |
-
"completions/min_length": 3.0,
|
| 312 |
-
"completions/min_terminated_length": 3.0,
|
| 313 |
-
"entropy": 0.47256265580654144,
|
| 314 |
-
"epoch": 0.00026,
|
| 315 |
-
"frac_reward_zero_std": 0.0,
|
| 316 |
-
"grad_norm": 0.3631357252597809,
|
| 317 |
-
"kl": 0.0003765556830330752,
|
| 318 |
-
"learning_rate": 1.0285714285714286e-06,
|
| 319 |
-
"loss": 0.0954,
|
| 320 |
-
"num_tokens": 367798.0,
|
| 321 |
-
"reward": -21.7075138092041,
|
| 322 |
-
"reward_std": 14.198619842529297,
|
| 323 |
-
"rewards/rollout_reward_func/mean": -21.7075138092041,
|
| 324 |
-
"rewards/rollout_reward_func/std": 14.198619842529297,
|
| 325 |
-
"sampling/importance_sampling_ratio/max": 1.2773635387420654,
|
| 326 |
-
"sampling/importance_sampling_ratio/mean": 0.995408833026886,
|
| 327 |
-
"sampling/importance_sampling_ratio/min": 0.778171181678772,
|
| 328 |
-
"sampling/sampling_logp_difference/max": 0.18239641189575195,
|
| 329 |
-
"sampling/sampling_logp_difference/mean": 0.010933874174952507,
|
| 330 |
-
"step": 13,
|
| 331 |
-
"step_time": 12.429970339999727
|
| 332 |
-
},
|
| 333 |
-
{
|
| 334 |
-
"clip_ratio/high_max": 0.0,
|
| 335 |
-
"clip_ratio/high_mean": 0.0,
|
| 336 |
-
"clip_ratio/low_mean": 0.0,
|
| 337 |
-
"clip_ratio/low_min": 0.0,
|
| 338 |
-
"clip_ratio/region_mean": 0.0,
|
| 339 |
-
"entropy": 0.4725050553679466,
|
| 340 |
-
"epoch": 0.00028,
|
| 341 |
-
"grad_norm": 0.3502637445926666,
|
| 342 |
-
"kl": 0.00048724883527029306,
|
| 343 |
-
"learning_rate": 1.1142857142857143e-06,
|
| 344 |
-
"loss": 0.0956,
|
| 345 |
-
"step": 14,
|
| 346 |
-
"step_time": 3.513378806000219
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"clip_ratio/high_max": 0.0,
|
|
@@ -353,333 +194,30 @@
|
|
| 353 |
"clip_ratio/region_mean": 0.0,
|
| 354 |
"completions/clipped_ratio": 0.78125,
|
| 355 |
"completions/max_length": 16.0,
|
| 356 |
-
"completions/max_terminated_length":
|
| 357 |
"completions/mean_length": 14.6875,
|
| 358 |
"completions/mean_terminated_length": 10.0,
|
| 359 |
-
"completions/min_length": 4.0,
|
| 360 |
-
"completions/min_terminated_length": 4.0,
|
| 361 |
-
"entropy": 0.48245932161808014,
|
| 362 |
-
"epoch": 0.0003,
|
| 363 |
-
"frac_reward_zero_std": 0.0,
|
| 364 |
-
"grad_norm": 0.5791538953781128,
|
| 365 |
-
"kl": 0.00043258603545837104,
|
| 366 |
-
"learning_rate": 1.2000000000000002e-06,
|
| 367 |
-
"loss": 0.0699,
|
| 368 |
-
"num_tokens": 422015.0,
|
| 369 |
-
"reward": -18.020889282226562,
|
| 370 |
-
"reward_std": 14.099430084228516,
|
| 371 |
-
"rewards/rollout_reward_func/mean": -18.020889282226562,
|
| 372 |
-
"rewards/rollout_reward_func/std": 14.0994291305542,
|
| 373 |
-
"sampling/importance_sampling_ratio/max": 1.131913423538208,
|
| 374 |
-
"sampling/importance_sampling_ratio/mean": 0.9780954122543335,
|
| 375 |
-
"sampling/importance_sampling_ratio/min": 0.7532981038093567,
|
| 376 |
-
"sampling/sampling_logp_difference/max": 0.2378932237625122,
|
| 377 |
-
"sampling/sampling_logp_difference/mean": 0.01171853393316269,
|
| 378 |
-
"step": 15,
|
| 379 |
-
"step_time": 12.757522889000484
|
| 380 |
-
},
|
| 381 |
-
{
|
| 382 |
-
"clip_ratio/high_max": 0.012500000186264515,
|
| 383 |
-
"clip_ratio/high_mean": 0.0031250000465661287,
|
| 384 |
-
"clip_ratio/low_mean": 0.0,
|
| 385 |
-
"clip_ratio/low_min": 0.0,
|
| 386 |
-
"clip_ratio/region_mean": 0.0031250000465661287,
|
| 387 |
-
"entropy": 0.48212701082229614,
|
| 388 |
-
"epoch": 0.00032,
|
| 389 |
-
"grad_norm": 0.5567578077316284,
|
| 390 |
-
"kl": 0.0003868155945383478,
|
| 391 |
-
"learning_rate": 1.2857142857142856e-06,
|
| 392 |
-
"loss": 0.0683,
|
| 393 |
-
"step": 16,
|
| 394 |
-
"step_time": 3.6391623570002594
|
| 395 |
-
},
|
| 396 |
-
{
|
| 397 |
-
"clip_ratio/high_max": 0.0,
|
| 398 |
-
"clip_ratio/high_mean": 0.0,
|
| 399 |
-
"clip_ratio/low_mean": 0.0,
|
| 400 |
-
"clip_ratio/low_min": 0.0,
|
| 401 |
-
"clip_ratio/region_mean": 0.0,
|
| 402 |
-
"completions/clipped_ratio": 0.6875,
|
| 403 |
-
"completions/max_length": 16.0,
|
| 404 |
-
"completions/max_terminated_length": 15.0,
|
| 405 |
-
"completions/mean_length": 14.125,
|
| 406 |
-
"completions/mean_terminated_length": 10.0,
|
| 407 |
"completions/min_length": 3.0,
|
| 408 |
"completions/min_terminated_length": 3.0,
|
| 409 |
-
"entropy": 0.
|
| 410 |
-
"epoch": 0.
|
| 411 |
-
"frac_reward_zero_std": 0.0,
|
| 412 |
-
"grad_norm": 0.46128952503204346,
|
| 413 |
-
"kl": 0.00022623784388997592,
|
| 414 |
-
"learning_rate": 1.3714285714285715e-06,
|
| 415 |
-
"loss": 0.093,
|
| 416 |
-
"num_tokens": 476133.0,
|
| 417 |
-
"reward": -17.307281494140625,
|
| 418 |
-
"reward_std": 17.012971878051758,
|
| 419 |
-
"rewards/rollout_reward_func/mean": -17.307281494140625,
|
| 420 |
-
"rewards/rollout_reward_func/std": 17.012971878051758,
|
| 421 |
-
"sampling/importance_sampling_ratio/max": 1.3407179117202759,
|
| 422 |
-
"sampling/importance_sampling_ratio/mean": 1.0220906734466553,
|
| 423 |
-
"sampling/importance_sampling_ratio/min": 0.7680003643035889,
|
| 424 |
-
"sampling/sampling_logp_difference/max": 0.22670793533325195,
|
| 425 |
-
"sampling/sampling_logp_difference/mean": 0.01049777027219534,
|
| 426 |
-
"step": 17,
|
| 427 |
-
"step_time": 14.12015571899974
|
| 428 |
-
},
|
| 429 |
-
{
|
| 430 |
-
"clip_ratio/high_max": 0.0,
|
| 431 |
-
"clip_ratio/high_mean": 0.0,
|
| 432 |
-
"clip_ratio/low_mean": 0.0,
|
| 433 |
-
"clip_ratio/low_min": 0.0,
|
| 434 |
-
"clip_ratio/region_mean": 0.0,
|
| 435 |
-
"entropy": 0.3973981589078903,
|
| 436 |
-
"epoch": 0.00036,
|
| 437 |
-
"grad_norm": 0.4621075987815857,
|
| 438 |
-
"kl": 0.0002368360073887743,
|
| 439 |
-
"learning_rate": 1.4571428571428571e-06,
|
| 440 |
-
"loss": 0.0935,
|
| 441 |
-
"step": 18,
|
| 442 |
-
"step_time": 4.419824714000242
|
| 443 |
-
},
|
| 444 |
-
{
|
| 445 |
-
"clip_ratio/high_max": 0.0,
|
| 446 |
-
"clip_ratio/high_mean": 0.0,
|
| 447 |
-
"clip_ratio/low_mean": 0.0,
|
| 448 |
-
"clip_ratio/low_min": 0.0,
|
| 449 |
-
"clip_ratio/region_mean": 0.0,
|
| 450 |
-
"completions/clipped_ratio": 0.75,
|
| 451 |
-
"completions/max_length": 16.0,
|
| 452 |
-
"completions/max_terminated_length": 16.0,
|
| 453 |
-
"completions/mean_length": 14.53125,
|
| 454 |
-
"completions/mean_terminated_length": 10.125,
|
| 455 |
-
"completions/min_length": 4.0,
|
| 456 |
-
"completions/min_terminated_length": 4.0,
|
| 457 |
-
"entropy": 0.4408823549747467,
|
| 458 |
-
"epoch": 0.00038,
|
| 459 |
-
"frac_reward_zero_std": 0.0,
|
| 460 |
-
"grad_norm": 0.49735671281814575,
|
| 461 |
-
"kl": 0.00043844833271577954,
|
| 462 |
-
"learning_rate": 1.5428571428571428e-06,
|
| 463 |
-
"loss": 0.0479,
|
| 464 |
-
"num_tokens": 528759.0,
|
| 465 |
-
"reward": -20.804988861083984,
|
| 466 |
-
"reward_std": 14.42788314819336,
|
| 467 |
-
"rewards/rollout_reward_func/mean": -20.804988861083984,
|
| 468 |
-
"rewards/rollout_reward_func/std": 14.42788314819336,
|
| 469 |
-
"sampling/importance_sampling_ratio/max": 1.2070324420928955,
|
| 470 |
-
"sampling/importance_sampling_ratio/mean": 0.9982270002365112,
|
| 471 |
-
"sampling/importance_sampling_ratio/min": 0.8168321251869202,
|
| 472 |
-
"sampling/sampling_logp_difference/max": 0.16861987113952637,
|
| 473 |
-
"sampling/sampling_logp_difference/mean": 0.010207044892013073,
|
| 474 |
-
"step": 19,
|
| 475 |
-
"step_time": 12.619404869999471
|
| 476 |
-
},
|
| 477 |
-
{
|
| 478 |
-
"clip_ratio/high_max": 0.0,
|
| 479 |
-
"clip_ratio/high_mean": 0.0,
|
| 480 |
-
"clip_ratio/low_mean": 0.0,
|
| 481 |
-
"clip_ratio/low_min": 0.0,
|
| 482 |
-
"clip_ratio/region_mean": 0.0,
|
| 483 |
-
"entropy": 0.44093089550733566,
|
| 484 |
-
"epoch": 0.0004,
|
| 485 |
-
"grad_norm": 0.4937126040458679,
|
| 486 |
-
"kl": 0.00047725086915306747,
|
| 487 |
-
"learning_rate": 1.6285714285714284e-06,
|
| 488 |
-
"loss": 0.0483,
|
| 489 |
-
"step": 20,
|
| 490 |
-
"step_time": 3.610557893000532
|
| 491 |
-
},
|
| 492 |
-
{
|
| 493 |
-
"clip_ratio/high_max": 0.0,
|
| 494 |
-
"clip_ratio/high_mean": 0.0,
|
| 495 |
-
"clip_ratio/low_mean": 0.0,
|
| 496 |
-
"clip_ratio/low_min": 0.0,
|
| 497 |
-
"clip_ratio/region_mean": 0.0,
|
| 498 |
-
"completions/clipped_ratio": 0.75,
|
| 499 |
-
"completions/max_length": 16.0,
|
| 500 |
-
"completions/max_terminated_length": 16.0,
|
| 501 |
-
"completions/mean_length": 14.78125,
|
| 502 |
-
"completions/mean_terminated_length": 11.125,
|
| 503 |
-
"completions/min_length": 4.0,
|
| 504 |
-
"completions/min_terminated_length": 4.0,
|
| 505 |
-
"entropy": 0.4765045493841171,
|
| 506 |
-
"epoch": 0.00042,
|
| 507 |
-
"frac_reward_zero_std": 0.0,
|
| 508 |
-
"grad_norm": 0.4461357593536377,
|
| 509 |
-
"kl": 0.00039704455411992967,
|
| 510 |
-
"learning_rate": 1.7142857142857143e-06,
|
| 511 |
-
"loss": 0.0319,
|
| 512 |
-
"num_tokens": 582547.0,
|
| 513 |
-
"reward": -21.366497039794922,
|
| 514 |
-
"reward_std": 13.540549278259277,
|
| 515 |
-
"rewards/rollout_reward_func/mean": -21.366497039794922,
|
| 516 |
-
"rewards/rollout_reward_func/std": 13.540549278259277,
|
| 517 |
-
"sampling/importance_sampling_ratio/max": 1.3568636178970337,
|
| 518 |
-
"sampling/importance_sampling_ratio/mean": 0.9794997572898865,
|
| 519 |
-
"sampling/importance_sampling_ratio/min": 0.7170642018318176,
|
| 520 |
-
"sampling/sampling_logp_difference/max": 0.17396187782287598,
|
| 521 |
-
"sampling/sampling_logp_difference/mean": 0.011949114501476288,
|
| 522 |
-
"step": 21,
|
| 523 |
-
"step_time": 13.120537562000209
|
| 524 |
-
},
|
| 525 |
-
{
|
| 526 |
-
"clip_ratio/high_max": 0.0,
|
| 527 |
-
"clip_ratio/high_mean": 0.0,
|
| 528 |
-
"clip_ratio/low_mean": 0.0,
|
| 529 |
-
"clip_ratio/low_min": 0.0,
|
| 530 |
-
"clip_ratio/region_mean": 0.0,
|
| 531 |
-
"entropy": 0.4768811762332916,
|
| 532 |
-
"epoch": 0.00044,
|
| 533 |
-
"grad_norm": 0.42503097653388977,
|
| 534 |
-
"kl": 0.0004109822984901257,
|
| 535 |
-
"learning_rate": 1.8e-06,
|
| 536 |
-
"loss": 0.0334,
|
| 537 |
-
"step": 22,
|
| 538 |
-
"step_time": 3.706944928000212
|
| 539 |
-
},
|
| 540 |
-
{
|
| 541 |
-
"clip_ratio/high_max": 0.0,
|
| 542 |
-
"clip_ratio/high_mean": 0.0,
|
| 543 |
-
"clip_ratio/low_mean": 0.0,
|
| 544 |
-
"clip_ratio/low_min": 0.0,
|
| 545 |
-
"clip_ratio/region_mean": 0.0,
|
| 546 |
-
"completions/clipped_ratio": 0.8125,
|
| 547 |
-
"completions/max_length": 16.0,
|
| 548 |
-
"completions/max_terminated_length": 15.0,
|
| 549 |
-
"completions/mean_length": 14.84375,
|
| 550 |
-
"completions/mean_terminated_length": 9.833333969116211,
|
| 551 |
-
"completions/min_length": 4.0,
|
| 552 |
-
"completions/min_terminated_length": 4.0,
|
| 553 |
-
"entropy": 0.5320451036095619,
|
| 554 |
-
"epoch": 0.00046,
|
| 555 |
-
"frac_reward_zero_std": 0.0,
|
| 556 |
-
"grad_norm": 0.4428236186504364,
|
| 557 |
-
"kl": 0.0005832012830069289,
|
| 558 |
-
"learning_rate": 1.8857142857142858e-06,
|
| 559 |
-
"loss": 0.092,
|
| 560 |
-
"num_tokens": 634760.0,
|
| 561 |
-
"reward": -19.927682876586914,
|
| 562 |
-
"reward_std": 16.771766662597656,
|
| 563 |
-
"rewards/rollout_reward_func/mean": -19.927682876586914,
|
| 564 |
-
"rewards/rollout_reward_func/std": 16.771766662597656,
|
| 565 |
-
"sampling/importance_sampling_ratio/max": 1.4887969493865967,
|
| 566 |
-
"sampling/importance_sampling_ratio/mean": 1.030691146850586,
|
| 567 |
-
"sampling/importance_sampling_ratio/min": 0.8164731860160828,
|
| 568 |
-
"sampling/sampling_logp_difference/max": 0.2149658203125,
|
| 569 |
-
"sampling/sampling_logp_difference/mean": 0.013042854145169258,
|
| 570 |
-
"step": 23,
|
| 571 |
-
"step_time": 12.996803623999767
|
| 572 |
-
},
|
| 573 |
-
{
|
| 574 |
-
"clip_ratio/high_max": 0.0,
|
| 575 |
-
"clip_ratio/high_mean": 0.0,
|
| 576 |
-
"clip_ratio/low_mean": 0.0,
|
| 577 |
-
"clip_ratio/low_min": 0.0,
|
| 578 |
-
"clip_ratio/region_mean": 0.0,
|
| 579 |
-
"entropy": 0.5324564427137375,
|
| 580 |
-
"epoch": 0.00048,
|
| 581 |
-
"grad_norm": 0.438973069190979,
|
| 582 |
-
"kl": 0.0006122617051005363,
|
| 583 |
-
"learning_rate": 1.9714285714285714e-06,
|
| 584 |
-
"loss": 0.0906,
|
| 585 |
-
"step": 24,
|
| 586 |
-
"step_time": 3.5907621789997393
|
| 587 |
-
},
|
| 588 |
-
{
|
| 589 |
-
"clip_ratio/high_max": 0.0,
|
| 590 |
-
"clip_ratio/high_mean": 0.0,
|
| 591 |
-
"clip_ratio/low_mean": 0.0,
|
| 592 |
-
"clip_ratio/low_min": 0.0,
|
| 593 |
-
"clip_ratio/region_mean": 0.0,
|
| 594 |
-
"completions/clipped_ratio": 0.78125,
|
| 595 |
-
"completions/max_length": 16.0,
|
| 596 |
-
"completions/max_terminated_length": 16.0,
|
| 597 |
-
"completions/mean_length": 14.5625,
|
| 598 |
-
"completions/mean_terminated_length": 9.428571701049805,
|
| 599 |
-
"completions/min_length": 4.0,
|
| 600 |
-
"completions/min_terminated_length": 4.0,
|
| 601 |
-
"entropy": 0.4107111692428589,
|
| 602 |
-
"epoch": 0.0005,
|
| 603 |
-
"frac_reward_zero_std": 0.0,
|
| 604 |
-
"grad_norm": 0.48443400859832764,
|
| 605 |
-
"kl": 0.0005294791480991989,
|
| 606 |
-
"learning_rate": 2.0571428571428573e-06,
|
| 607 |
-
"loss": 0.0265,
|
| 608 |
-
"num_tokens": 686641.0,
|
| 609 |
-
"reward": -24.298006057739258,
|
| 610 |
-
"reward_std": 11.884275436401367,
|
| 611 |
-
"rewards/rollout_reward_func/mean": -24.298006057739258,
|
| 612 |
-
"rewards/rollout_reward_func/std": 11.88427448272705,
|
| 613 |
-
"sampling/importance_sampling_ratio/max": 1.3545169830322266,
|
| 614 |
-
"sampling/importance_sampling_ratio/mean": 1.0202080011367798,
|
| 615 |
-
"sampling/importance_sampling_ratio/min": 0.8227697610855103,
|
| 616 |
-
"sampling/sampling_logp_difference/max": 0.19383776187896729,
|
| 617 |
-
"sampling/sampling_logp_difference/mean": 0.010375281795859337,
|
| 618 |
-
"step": 25,
|
| 619 |
-
"step_time": 11.739334979000432
|
| 620 |
-
},
|
| 621 |
-
{
|
| 622 |
-
"clip_ratio/high_max": 0.0,
|
| 623 |
-
"clip_ratio/high_mean": 0.0,
|
| 624 |
-
"clip_ratio/low_mean": 0.0,
|
| 625 |
-
"clip_ratio/low_min": 0.0,
|
| 626 |
-
"clip_ratio/region_mean": 0.0,
|
| 627 |
-
"entropy": 0.41023270785808563,
|
| 628 |
-
"epoch": 0.00052,
|
| 629 |
-
"grad_norm": 0.4764305651187897,
|
| 630 |
-
"kl": 0.0004724617610918358,
|
| 631 |
-
"learning_rate": 2.142857142857143e-06,
|
| 632 |
-
"loss": 0.0261,
|
| 633 |
-
"step": 26,
|
| 634 |
-
"step_time": 3.5787725539998974
|
| 635 |
-
},
|
| 636 |
-
{
|
| 637 |
-
"clip_ratio/high_max": 0.0,
|
| 638 |
-
"clip_ratio/high_mean": 0.0,
|
| 639 |
-
"clip_ratio/low_mean": 0.0,
|
| 640 |
-
"clip_ratio/low_min": 0.0,
|
| 641 |
-
"clip_ratio/region_mean": 0.0,
|
| 642 |
-
"completions/clipped_ratio": 0.71875,
|
| 643 |
-
"completions/max_length": 16.0,
|
| 644 |
-
"completions/max_terminated_length": 16.0,
|
| 645 |
-
"completions/mean_length": 14.09375,
|
| 646 |
-
"completions/mean_terminated_length": 9.222222328186035,
|
| 647 |
-
"completions/min_length": 4.0,
|
| 648 |
-
"completions/min_terminated_length": 4.0,
|
| 649 |
-
"entropy": 0.42711327224969864,
|
| 650 |
-
"epoch": 0.00054,
|
| 651 |
"frac_reward_zero_std": 0.0,
|
| 652 |
-
"grad_norm": 0.
|
| 653 |
-
"kl": 0.
|
| 654 |
-
"learning_rate":
|
| 655 |
-
"loss": 0.
|
| 656 |
-
"num_tokens":
|
| 657 |
-
"reward": -21.
|
| 658 |
-
"reward_std":
|
| 659 |
-
"rewards/rollout_reward_func/mean": -21.
|
| 660 |
-
"rewards/rollout_reward_func/std":
|
| 661 |
-
"sampling/importance_sampling_ratio/max": 1.
|
| 662 |
-
"sampling/importance_sampling_ratio/mean": 0.
|
| 663 |
-
"sampling/importance_sampling_ratio/min": 0.
|
| 664 |
-
"sampling/sampling_logp_difference/max": 0.
|
| 665 |
-
"sampling/sampling_logp_difference/mean": 0.
|
| 666 |
-
"step":
|
| 667 |
-
"step_time": 13.
|
| 668 |
-
},
|
| 669 |
-
{
|
| 670 |
-
"clip_ratio/high_max": 0.0078125,
|
| 671 |
-
"clip_ratio/high_mean": 0.001953125,
|
| 672 |
-
"clip_ratio/low_mean": 0.0,
|
| 673 |
-
"clip_ratio/low_min": 0.0,
|
| 674 |
-
"clip_ratio/region_mean": 0.001953125,
|
| 675 |
-
"entropy": 0.4274953231215477,
|
| 676 |
-
"epoch": 0.00056,
|
| 677 |
-
"grad_norm": 0.5122133493423462,
|
| 678 |
-
"kl": 0.0005963902549410705,
|
| 679 |
-
"learning_rate": 2.3142857142857145e-06,
|
| 680 |
-
"loss": 0.1015,
|
| 681 |
-
"step": 28,
|
| 682 |
-
"step_time": 3.6348850080003103
|
| 683 |
},
|
| 684 |
{
|
| 685 |
"clip_ratio/high_max": 0.0,
|
|
@@ -687,47 +225,14 @@
|
|
| 687 |
"clip_ratio/low_mean": 0.001953125,
|
| 688 |
"clip_ratio/low_min": 0.0,
|
| 689 |
"clip_ratio/region_mean": 0.001953125,
|
| 690 |
-
"
|
| 691 |
-
"
|
| 692 |
-
"
|
| 693 |
-
"
|
| 694 |
-
"
|
| 695 |
-
"
|
| 696 |
-
"
|
| 697 |
-
"
|
| 698 |
-
"epoch": 0.00058,
|
| 699 |
-
"frac_reward_zero_std": 0.0,
|
| 700 |
-
"grad_norm": 0.47282618284225464,
|
| 701 |
-
"kl": 0.000547109026229009,
|
| 702 |
-
"learning_rate": 2.4000000000000003e-06,
|
| 703 |
-
"loss": 0.0551,
|
| 704 |
-
"num_tokens": 791760.0,
|
| 705 |
-
"reward": -22.874284744262695,
|
| 706 |
-
"reward_std": 16.014789581298828,
|
| 707 |
-
"rewards/rollout_reward_func/mean": -22.874284744262695,
|
| 708 |
-
"rewards/rollout_reward_func/std": 16.014787673950195,
|
| 709 |
-
"sampling/importance_sampling_ratio/max": 1.445327639579773,
|
| 710 |
-
"sampling/importance_sampling_ratio/mean": 0.9864210486412048,
|
| 711 |
-
"sampling/importance_sampling_ratio/min": 0.6746537685394287,
|
| 712 |
-
"sampling/sampling_logp_difference/max": 0.24254560470581055,
|
| 713 |
-
"sampling/sampling_logp_difference/mean": 0.01340758427977562,
|
| 714 |
-
"step": 29,
|
| 715 |
-
"step_time": 12.450915582999642
|
| 716 |
-
},
|
| 717 |
-
{
|
| 718 |
-
"clip_ratio/high_max": 0.0,
|
| 719 |
-
"clip_ratio/high_mean": 0.0,
|
| 720 |
-
"clip_ratio/low_mean": 0.0024038462433964014,
|
| 721 |
-
"clip_ratio/low_min": 0.0,
|
| 722 |
-
"clip_ratio/region_mean": 0.0024038462433964014,
|
| 723 |
-
"entropy": 0.44682736694812775,
|
| 724 |
-
"epoch": 0.0006,
|
| 725 |
-
"grad_norm": 0.47560998797416687,
|
| 726 |
-
"kl": 0.0005742600187659264,
|
| 727 |
-
"learning_rate": 2.4857142857142858e-06,
|
| 728 |
-
"loss": 0.0575,
|
| 729 |
-
"step": 30,
|
| 730 |
-
"step_time": 3.652044981000472
|
| 731 |
},
|
| 732 |
{
|
| 733 |
"clip_ratio/high_max": 0.0,
|
|
@@ -735,32 +240,32 @@
|
|
| 735 |
"clip_ratio/low_mean": 0.0,
|
| 736 |
"clip_ratio/low_min": 0.0,
|
| 737 |
"clip_ratio/region_mean": 0.0,
|
| 738 |
-
"completions/clipped_ratio": 0.
|
| 739 |
"completions/max_length": 16.0,
|
| 740 |
-
"completions/max_terminated_length":
|
| 741 |
-
"completions/mean_length":
|
| 742 |
-
"completions/mean_terminated_length":
|
| 743 |
-
"completions/min_length":
|
| 744 |
-
"completions/min_terminated_length":
|
| 745 |
-
"entropy": 0.
|
| 746 |
-
"epoch": 0.
|
| 747 |
"frac_reward_zero_std": 0.0,
|
| 748 |
-
"grad_norm": 0.
|
| 749 |
-
"kl": 0.
|
| 750 |
-
"learning_rate":
|
| 751 |
-
"loss": 0.
|
| 752 |
-
"num_tokens":
|
| 753 |
-
"reward": -
|
| 754 |
-
"reward_std":
|
| 755 |
-
"rewards/rollout_reward_func/mean": -
|
| 756 |
-
"rewards/rollout_reward_func/std":
|
| 757 |
-
"sampling/importance_sampling_ratio/max": 1.
|
| 758 |
-
"sampling/importance_sampling_ratio/mean":
|
| 759 |
-
"sampling/importance_sampling_ratio/min": 0.
|
| 760 |
-
"sampling/sampling_logp_difference/max": 0.
|
| 761 |
-
"sampling/sampling_logp_difference/mean": 0.
|
| 762 |
-
"step":
|
| 763 |
-
"step_time": 12.
|
| 764 |
},
|
| 765 |
{
|
| 766 |
"clip_ratio/high_max": 0.0,
|
|
@@ -768,1060 +273,52 @@
|
|
| 768 |
"clip_ratio/low_mean": 0.0,
|
| 769 |
"clip_ratio/low_min": 0.0,
|
| 770 |
"clip_ratio/region_mean": 0.0,
|
| 771 |
-
"entropy": 0.
|
| 772 |
-
"epoch": 0.
|
| 773 |
-
"grad_norm": 0.
|
| 774 |
-
"kl": 0.
|
| 775 |
-
"learning_rate":
|
| 776 |
-
"loss": 0.
|
| 777 |
-
"step":
|
| 778 |
-
"step_time": 3.
|
| 779 |
},
|
| 780 |
{
|
| 781 |
-
"
|
| 782 |
-
"
|
| 783 |
-
"
|
| 784 |
-
"
|
| 785 |
-
"
|
| 786 |
-
"
|
| 787 |
-
"
|
| 788 |
-
"
|
| 789 |
-
"
|
| 790 |
-
"
|
| 791 |
-
"
|
| 792 |
-
"
|
| 793 |
-
"
|
| 794 |
-
"
|
| 795 |
-
"
|
| 796 |
-
"
|
| 797 |
-
"
|
| 798 |
-
"
|
| 799 |
-
"
|
| 800 |
-
"
|
| 801 |
-
"
|
| 802 |
-
"
|
| 803 |
-
"
|
| 804 |
-
"
|
| 805 |
-
"
|
| 806 |
-
"
|
| 807 |
-
"
|
| 808 |
-
"
|
| 809 |
-
"
|
| 810 |
-
"
|
| 811 |
-
"
|
| 812 |
-
},
|
| 813 |
-
{
|
| 814 |
-
"clip_ratio/high_max": 0.0,
|
| 815 |
-
"clip_ratio/high_mean": 0.0,
|
| 816 |
-
"clip_ratio/low_mean": 0.001953125,
|
| 817 |
-
"clip_ratio/low_min": 0.0,
|
| 818 |
-
"clip_ratio/region_mean": 0.001953125,
|
| 819 |
-
"entropy": 0.45316264033317566,
|
| 820 |
-
"epoch": 0.00068,
|
| 821 |
-
"grad_norm": 0.47820040583610535,
|
| 822 |
-
"kl": 0.0006169998086988926,
|
| 823 |
-
"learning_rate": 2.8285714285714288e-06,
|
| 824 |
-
"loss": 0.0429,
|
| 825 |
-
"step": 34,
|
| 826 |
-
"step_time": 3.612241633999929
|
| 827 |
-
},
|
| 828 |
-
{
|
| 829 |
-
"clip_ratio/high_max": 0.0,
|
| 830 |
-
"clip_ratio/high_mean": 0.0,
|
| 831 |
-
"clip_ratio/low_mean": 0.0,
|
| 832 |
-
"clip_ratio/low_min": 0.0,
|
| 833 |
-
"clip_ratio/region_mean": 0.0,
|
| 834 |
-
"completions/clipped_ratio": 0.84375,
|
| 835 |
-
"completions/max_length": 16.0,
|
| 836 |
-
"completions/max_terminated_length": 16.0,
|
| 837 |
-
"completions/mean_length": 15.09375,
|
| 838 |
-
"completions/mean_terminated_length": 10.199999809265137,
|
| 839 |
-
"completions/min_length": 4.0,
|
| 840 |
-
"completions/min_terminated_length": 4.0,
|
| 841 |
-
"entropy": 0.42138345539569855,
|
| 842 |
-
"epoch": 0.0007,
|
| 843 |
-
"frac_reward_zero_std": 0.0,
|
| 844 |
-
"grad_norm": 0.46025797724723816,
|
| 845 |
-
"kl": 0.0004996234347345307,
|
| 846 |
-
"learning_rate": 2.9142857142857142e-06,
|
| 847 |
-
"loss": 0.053,
|
| 848 |
-
"num_tokens": 948802.0,
|
| 849 |
-
"reward": -24.316551208496094,
|
| 850 |
-
"reward_std": 12.835836410522461,
|
| 851 |
-
"rewards/rollout_reward_func/mean": -24.316551208496094,
|
| 852 |
-
"rewards/rollout_reward_func/std": 12.835835456848145,
|
| 853 |
-
"sampling/importance_sampling_ratio/max": 1.328806757926941,
|
| 854 |
-
"sampling/importance_sampling_ratio/mean": 1.0274875164031982,
|
| 855 |
-
"sampling/importance_sampling_ratio/min": 0.894749641418457,
|
| 856 |
-
"sampling/sampling_logp_difference/max": 0.18797016143798828,
|
| 857 |
-
"sampling/sampling_logp_difference/mean": 0.011560991406440735,
|
| 858 |
-
"step": 35,
|
| 859 |
-
"step_time": 13.08812723400024
|
| 860 |
-
},
|
| 861 |
-
{
|
| 862 |
-
"clip_ratio/high_max": 0.0,
|
| 863 |
-
"clip_ratio/high_mean": 0.0,
|
| 864 |
-
"clip_ratio/low_mean": 0.001953125,
|
| 865 |
-
"clip_ratio/low_min": 0.0,
|
| 866 |
-
"clip_ratio/region_mean": 0.001953125,
|
| 867 |
-
"entropy": 0.42126451432704926,
|
| 868 |
-
"epoch": 0.00072,
|
| 869 |
-
"grad_norm": 0.43526744842529297,
|
| 870 |
-
"kl": 0.000492801918881014,
|
| 871 |
-
"learning_rate": 3e-06,
|
| 872 |
-
"loss": 0.054,
|
| 873 |
-
"step": 36,
|
| 874 |
-
"step_time": 4.318984840999747
|
| 875 |
-
},
|
| 876 |
-
{
|
| 877 |
-
"clip_ratio/high_max": 0.0,
|
| 878 |
-
"clip_ratio/high_mean": 0.0,
|
| 879 |
-
"clip_ratio/low_mean": 0.0,
|
| 880 |
-
"clip_ratio/low_min": 0.0,
|
| 881 |
-
"clip_ratio/region_mean": 0.0,
|
| 882 |
-
"completions/clipped_ratio": 0.75,
|
| 883 |
-
"completions/max_length": 16.0,
|
| 884 |
-
"completions/max_terminated_length": 15.0,
|
| 885 |
-
"completions/mean_length": 14.53125,
|
| 886 |
-
"completions/mean_terminated_length": 10.125,
|
| 887 |
-
"completions/min_length": 4.0,
|
| 888 |
-
"completions/min_terminated_length": 4.0,
|
| 889 |
-
"entropy": 0.4301793724298477,
|
| 890 |
-
"epoch": 0.00074,
|
| 891 |
-
"frac_reward_zero_std": 0.0,
|
| 892 |
-
"grad_norm": 0.3941028416156769,
|
| 893 |
-
"kl": 0.0006222588490345515,
|
| 894 |
-
"learning_rate": 2.999999999444446e-06,
|
| 895 |
-
"loss": 0.058,
|
| 896 |
-
"num_tokens": 1001889.0,
|
| 897 |
-
"reward": -21.4174861907959,
|
| 898 |
-
"reward_std": 12.194807052612305,
|
| 899 |
-
"rewards/rollout_reward_func/mean": -21.4174861907959,
|
| 900 |
-
"rewards/rollout_reward_func/std": 12.194808006286621,
|
| 901 |
-
"sampling/importance_sampling_ratio/max": 1.2621595859527588,
|
| 902 |
-
"sampling/importance_sampling_ratio/mean": 1.006618618965149,
|
| 903 |
-
"sampling/importance_sampling_ratio/min": 0.8662732243537903,
|
| 904 |
-
"sampling/sampling_logp_difference/max": 0.19760990142822266,
|
| 905 |
-
"sampling/sampling_logp_difference/mean": 0.00977108720690012,
|
| 906 |
-
"step": 37,
|
| 907 |
-
"step_time": 12.948201442999562
|
| 908 |
-
},
|
| 909 |
-
{
|
| 910 |
-
"clip_ratio/high_max": 0.0,
|
| 911 |
-
"clip_ratio/high_mean": 0.0,
|
| 912 |
-
"clip_ratio/low_mean": 0.0,
|
| 913 |
-
"clip_ratio/low_min": 0.0,
|
| 914 |
-
"clip_ratio/region_mean": 0.0,
|
| 915 |
-
"entropy": 0.43141717463731766,
|
| 916 |
-
"epoch": 0.00076,
|
| 917 |
-
"grad_norm": 0.39947450160980225,
|
| 918 |
-
"kl": 0.000883200435055187,
|
| 919 |
-
"learning_rate": 2.9999999977777837e-06,
|
| 920 |
-
"loss": 0.0575,
|
| 921 |
-
"step": 38,
|
| 922 |
-
"step_time": 3.602961716000209
|
| 923 |
-
},
|
| 924 |
-
{
|
| 925 |
-
"clip_ratio/high_max": 0.0,
|
| 926 |
-
"clip_ratio/high_mean": 0.0,
|
| 927 |
-
"clip_ratio/low_mean": 0.0,
|
| 928 |
-
"clip_ratio/low_min": 0.0,
|
| 929 |
-
"clip_ratio/region_mean": 0.0,
|
| 930 |
-
"completions/clipped_ratio": 0.71875,
|
| 931 |
-
"completions/max_length": 16.0,
|
| 932 |
-
"completions/max_terminated_length": 15.0,
|
| 933 |
-
"completions/mean_length": 13.625,
|
| 934 |
-
"completions/mean_terminated_length": 7.555555820465088,
|
| 935 |
-
"completions/min_length": 3.0,
|
| 936 |
-
"completions/min_terminated_length": 3.0,
|
| 937 |
-
"entropy": 0.48835278302431107,
|
| 938 |
-
"epoch": 0.00078,
|
| 939 |
-
"frac_reward_zero_std": 0.125,
|
| 940 |
-
"grad_norm": 0.39885786175727844,
|
| 941 |
-
"kl": 0.0009218690829584375,
|
| 942 |
-
"learning_rate": 2.9999999950000137e-06,
|
| 943 |
-
"loss": 0.0949,
|
| 944 |
-
"num_tokens": 1056411.0,
|
| 945 |
-
"reward": -22.038280487060547,
|
| 946 |
-
"reward_std": 14.705154418945312,
|
| 947 |
-
"rewards/rollout_reward_func/mean": -22.038280487060547,
|
| 948 |
-
"rewards/rollout_reward_func/std": 14.705153465270996,
|
| 949 |
-
"sampling/importance_sampling_ratio/max": 1.3119615316390991,
|
| 950 |
-
"sampling/importance_sampling_ratio/mean": 1.0176267623901367,
|
| 951 |
-
"sampling/importance_sampling_ratio/min": 0.7629126310348511,
|
| 952 |
-
"sampling/sampling_logp_difference/max": 0.12403631210327148,
|
| 953 |
-
"sampling/sampling_logp_difference/mean": 0.01314721629023552,
|
| 954 |
-
"step": 39,
|
| 955 |
-
"step_time": 12.314849860000322
|
| 956 |
-
},
|
| 957 |
-
{
|
| 958 |
-
"clip_ratio/high_max": 0.0,
|
| 959 |
-
"clip_ratio/high_mean": 0.0,
|
| 960 |
-
"clip_ratio/low_mean": 0.0,
|
| 961 |
-
"clip_ratio/low_min": 0.0,
|
| 962 |
-
"clip_ratio/region_mean": 0.0,
|
| 963 |
-
"entropy": 0.4897266551852226,
|
| 964 |
-
"epoch": 0.0008,
|
| 965 |
-
"grad_norm": 0.3934723734855652,
|
| 966 |
-
"kl": 0.0010027830867329612,
|
| 967 |
-
"learning_rate": 2.999999991111135e-06,
|
| 968 |
-
"loss": 0.095,
|
| 969 |
-
"step": 40,
|
| 970 |
-
"step_time": 3.6701166389993887
|
| 971 |
-
},
|
| 972 |
-
{
|
| 973 |
-
"clip_ratio/high_max": 0.0,
|
| 974 |
-
"clip_ratio/high_mean": 0.0,
|
| 975 |
-
"clip_ratio/low_mean": 0.0,
|
| 976 |
-
"clip_ratio/low_min": 0.0,
|
| 977 |
-
"clip_ratio/region_mean": 0.0,
|
| 978 |
-
"completions/clipped_ratio": 0.6875,
|
| 979 |
-
"completions/max_length": 16.0,
|
| 980 |
-
"completions/max_terminated_length": 16.0,
|
| 981 |
-
"completions/mean_length": 14.28125,
|
| 982 |
-
"completions/mean_terminated_length": 10.5,
|
| 983 |
-
"completions/min_length": 3.0,
|
| 984 |
-
"completions/min_terminated_length": 3.0,
|
| 985 |
-
"entropy": 0.4647715613245964,
|
| 986 |
-
"epoch": 0.00082,
|
| 987 |
-
"frac_reward_zero_std": 0.0,
|
| 988 |
-
"grad_norm": 0.4407760500907898,
|
| 989 |
-
"kl": 0.000995866059383843,
|
| 990 |
-
"learning_rate": 2.9999999861111487e-06,
|
| 991 |
-
"loss": 0.0416,
|
| 992 |
-
"num_tokens": 1109059.0,
|
| 993 |
-
"reward": -20.27096939086914,
|
| 994 |
-
"reward_std": 16.170513153076172,
|
| 995 |
-
"rewards/rollout_reward_func/mean": -20.27096939086914,
|
| 996 |
-
"rewards/rollout_reward_func/std": 16.170513153076172,
|
| 997 |
-
"sampling/importance_sampling_ratio/max": 1.238898515701294,
|
| 998 |
-
"sampling/importance_sampling_ratio/mean": 0.9764783382415771,
|
| 999 |
-
"sampling/importance_sampling_ratio/min": 0.7559868693351746,
|
| 1000 |
-
"sampling/sampling_logp_difference/max": 0.24738311767578125,
|
| 1001 |
-
"sampling/sampling_logp_difference/mean": 0.011778589338064194,
|
| 1002 |
-
"step": 41,
|
| 1003 |
-
"step_time": 12.945746578000126
|
| 1004 |
-
},
|
| 1005 |
-
{
|
| 1006 |
-
"clip_ratio/high_max": 0.0,
|
| 1007 |
-
"clip_ratio/high_mean": 0.0,
|
| 1008 |
-
"clip_ratio/low_mean": 0.0,
|
| 1009 |
-
"clip_ratio/low_min": 0.0,
|
| 1010 |
-
"clip_ratio/region_mean": 0.0,
|
| 1011 |
-
"entropy": 0.46506941318511963,
|
| 1012 |
-
"epoch": 0.00084,
|
| 1013 |
-
"grad_norm": 0.41802552342414856,
|
| 1014 |
-
"kl": 0.0008689216920174658,
|
| 1015 |
-
"learning_rate": 2.999999980000054e-06,
|
| 1016 |
-
"loss": 0.0416,
|
| 1017 |
-
"step": 42,
|
| 1018 |
-
"step_time": 3.5847704960001465
|
| 1019 |
-
},
|
| 1020 |
-
{
|
| 1021 |
-
"clip_ratio/high_max": 0.0,
|
| 1022 |
-
"clip_ratio/high_mean": 0.0,
|
| 1023 |
-
"clip_ratio/low_mean": 0.0,
|
| 1024 |
-
"clip_ratio/low_min": 0.0,
|
| 1025 |
-
"clip_ratio/region_mean": 0.0,
|
| 1026 |
-
"completions/clipped_ratio": 0.8125,
|
| 1027 |
-
"completions/max_length": 16.0,
|
| 1028 |
-
"completions/max_terminated_length": 16.0,
|
| 1029 |
-
"completions/mean_length": 14.90625,
|
| 1030 |
-
"completions/mean_terminated_length": 10.166666984558105,
|
| 1031 |
-
"completions/min_length": 3.0,
|
| 1032 |
-
"completions/min_terminated_length": 3.0,
|
| 1033 |
-
"entropy": 0.40763208270072937,
|
| 1034 |
-
"epoch": 0.00086,
|
| 1035 |
-
"frac_reward_zero_std": 0.0,
|
| 1036 |
-
"grad_norm": 0.38299527764320374,
|
| 1037 |
-
"kl": 0.0007383096963167191,
|
| 1038 |
-
"learning_rate": 2.999999972777851e-06,
|
| 1039 |
-
"loss": 0.0721,
|
| 1040 |
-
"num_tokens": 1163200.0,
|
| 1041 |
-
"reward": -18.545978546142578,
|
| 1042 |
-
"reward_std": 14.86248779296875,
|
| 1043 |
-
"rewards/rollout_reward_func/mean": -18.545978546142578,
|
| 1044 |
-
"rewards/rollout_reward_func/std": 14.862486839294434,
|
| 1045 |
-
"sampling/importance_sampling_ratio/max": 1.201629400253296,
|
| 1046 |
-
"sampling/importance_sampling_ratio/mean": 1.003993272781372,
|
| 1047 |
-
"sampling/importance_sampling_ratio/min": 0.8553942441940308,
|
| 1048 |
-
"sampling/sampling_logp_difference/max": 0.14433181285858154,
|
| 1049 |
-
"sampling/sampling_logp_difference/mean": 0.010116634890437126,
|
| 1050 |
-
"step": 43,
|
| 1051 |
-
"step_time": 13.120121834000201
|
| 1052 |
-
},
|
| 1053 |
-
{
|
| 1054 |
-
"clip_ratio/high_max": 0.0,
|
| 1055 |
-
"clip_ratio/high_mean": 0.0,
|
| 1056 |
-
"clip_ratio/low_mean": 0.003969253972172737,
|
| 1057 |
-
"clip_ratio/low_min": 0.0,
|
| 1058 |
-
"clip_ratio/region_mean": 0.003969253972172737,
|
| 1059 |
-
"entropy": 0.4081807807087898,
|
| 1060 |
-
"epoch": 0.00088,
|
| 1061 |
-
"grad_norm": 0.38337594270706177,
|
| 1062 |
-
"kl": 0.0009075161142391153,
|
| 1063 |
-
"learning_rate": 2.99999996444454e-06,
|
| 1064 |
-
"loss": 0.071,
|
| 1065 |
-
"step": 44,
|
| 1066 |
-
"step_time": 3.613092282999787
|
| 1067 |
-
},
|
| 1068 |
-
{
|
| 1069 |
-
"clip_ratio/high_max": 0.0,
|
| 1070 |
-
"clip_ratio/high_mean": 0.0,
|
| 1071 |
-
"clip_ratio/low_mean": 0.0,
|
| 1072 |
-
"clip_ratio/low_min": 0.0,
|
| 1073 |
-
"clip_ratio/region_mean": 0.0,
|
| 1074 |
-
"completions/clipped_ratio": 0.78125,
|
| 1075 |
-
"completions/max_length": 16.0,
|
| 1076 |
-
"completions/max_terminated_length": 16.0,
|
| 1077 |
-
"completions/mean_length": 14.75,
|
| 1078 |
-
"completions/mean_terminated_length": 10.285715103149414,
|
| 1079 |
-
"completions/min_length": 4.0,
|
| 1080 |
-
"completions/min_terminated_length": 4.0,
|
| 1081 |
-
"entropy": 0.439114011824131,
|
| 1082 |
-
"epoch": 0.0009,
|
| 1083 |
-
"frac_reward_zero_std": 0.0,
|
| 1084 |
-
"grad_norm": 0.45874863862991333,
|
| 1085 |
-
"kl": 0.001827418542234227,
|
| 1086 |
-
"learning_rate": 2.9999999550001207e-06,
|
| 1087 |
-
"loss": 0.0394,
|
| 1088 |
-
"num_tokens": 1218086.0,
|
| 1089 |
-
"reward": -17.6557559967041,
|
| 1090 |
-
"reward_std": 14.141748428344727,
|
| 1091 |
-
"rewards/rollout_reward_func/mean": -17.6557559967041,
|
| 1092 |
-
"rewards/rollout_reward_func/std": 14.14174747467041,
|
| 1093 |
-
"sampling/importance_sampling_ratio/max": 1.4917187690734863,
|
| 1094 |
-
"sampling/importance_sampling_ratio/mean": 0.9951067566871643,
|
| 1095 |
-
"sampling/importance_sampling_ratio/min": 0.7373844385147095,
|
| 1096 |
-
"sampling/sampling_logp_difference/max": 0.22326016426086426,
|
| 1097 |
-
"sampling/sampling_logp_difference/mean": 0.013849505223333836,
|
| 1098 |
-
"step": 45,
|
| 1099 |
-
"step_time": 12.33521299799986
|
| 1100 |
-
},
|
| 1101 |
-
{
|
| 1102 |
-
"clip_ratio/high_max": 0.0,
|
| 1103 |
-
"clip_ratio/high_mean": 0.0,
|
| 1104 |
-
"clip_ratio/low_mean": 0.0,
|
| 1105 |
-
"clip_ratio/low_min": 0.0,
|
| 1106 |
-
"clip_ratio/region_mean": 0.0,
|
| 1107 |
-
"entropy": 0.4387468546628952,
|
| 1108 |
-
"epoch": 0.00092,
|
| 1109 |
-
"grad_norm": 0.4409801959991455,
|
| 1110 |
-
"kl": 0.0015551562537439167,
|
| 1111 |
-
"learning_rate": 2.999999944444594e-06,
|
| 1112 |
-
"loss": 0.04,
|
| 1113 |
-
"step": 46,
|
| 1114 |
-
"step_time": 4.31953496000051
|
| 1115 |
-
},
|
| 1116 |
-
{
|
| 1117 |
-
"clip_ratio/high_max": 0.0,
|
| 1118 |
-
"clip_ratio/high_mean": 0.0,
|
| 1119 |
-
"clip_ratio/low_mean": 0.0,
|
| 1120 |
-
"clip_ratio/low_min": 0.0,
|
| 1121 |
-
"clip_ratio/region_mean": 0.0,
|
| 1122 |
-
"completions/clipped_ratio": 0.71875,
|
| 1123 |
-
"completions/max_length": 16.0,
|
| 1124 |
-
"completions/max_terminated_length": 14.0,
|
| 1125 |
-
"completions/mean_length": 14.0,
|
| 1126 |
-
"completions/mean_terminated_length": 8.88888931274414,
|
| 1127 |
-
"completions/min_length": 4.0,
|
| 1128 |
-
"completions/min_terminated_length": 4.0,
|
| 1129 |
-
"entropy": 0.4944975897669792,
|
| 1130 |
-
"epoch": 0.00094,
|
| 1131 |
-
"frac_reward_zero_std": 0.0,
|
| 1132 |
-
"grad_norm": 0.6308720707893372,
|
| 1133 |
-
"kl": 0.0014178274941514246,
|
| 1134 |
-
"learning_rate": 2.999999932777959e-06,
|
| 1135 |
-
"loss": 0.037,
|
| 1136 |
-
"num_tokens": 1271722.0,
|
| 1137 |
-
"reward": -16.693885803222656,
|
| 1138 |
-
"reward_std": 18.024343490600586,
|
| 1139 |
-
"rewards/rollout_reward_func/mean": -16.693885803222656,
|
| 1140 |
-
"rewards/rollout_reward_func/std": 18.02434539794922,
|
| 1141 |
-
"sampling/importance_sampling_ratio/max": 1.36251699924469,
|
| 1142 |
-
"sampling/importance_sampling_ratio/mean": 0.9912035465240479,
|
| 1143 |
-
"sampling/importance_sampling_ratio/min": 0.8056352734565735,
|
| 1144 |
-
"sampling/sampling_logp_difference/max": 0.24249553680419922,
|
| 1145 |
-
"sampling/sampling_logp_difference/mean": 0.011305240914225578,
|
| 1146 |
-
"step": 47,
|
| 1147 |
-
"step_time": 12.623038148999512
|
| 1148 |
-
},
|
| 1149 |
-
{
|
| 1150 |
-
"clip_ratio/high_max": 0.0,
|
| 1151 |
-
"clip_ratio/high_mean": 0.0,
|
| 1152 |
-
"clip_ratio/low_mean": 0.0,
|
| 1153 |
-
"clip_ratio/low_min": 0.0,
|
| 1154 |
-
"clip_ratio/region_mean": 0.0,
|
| 1155 |
-
"entropy": 0.4939929470419884,
|
| 1156 |
-
"epoch": 0.00096,
|
| 1157 |
-
"grad_norm": 0.6490526795387268,
|
| 1158 |
-
"kl": 0.0013543795867008157,
|
| 1159 |
-
"learning_rate": 2.9999999200002154e-06,
|
| 1160 |
-
"loss": 0.0363,
|
| 1161 |
-
"step": 48,
|
| 1162 |
-
"step_time": 3.6226852679997137
|
| 1163 |
-
},
|
| 1164 |
-
{
|
| 1165 |
-
"clip_ratio/high_max": 0.0,
|
| 1166 |
-
"clip_ratio/high_mean": 0.0,
|
| 1167 |
-
"clip_ratio/low_mean": 0.001953125,
|
| 1168 |
-
"clip_ratio/low_min": 0.0,
|
| 1169 |
-
"clip_ratio/region_mean": 0.001953125,
|
| 1170 |
-
"completions/clipped_ratio": 0.8125,
|
| 1171 |
-
"completions/max_length": 16.0,
|
| 1172 |
-
"completions/max_terminated_length": 16.0,
|
| 1173 |
-
"completions/mean_length": 15.0,
|
| 1174 |
-
"completions/mean_terminated_length": 10.666666984558105,
|
| 1175 |
-
"completions/min_length": 4.0,
|
| 1176 |
-
"completions/min_terminated_length": 4.0,
|
| 1177 |
-
"entropy": 0.419900543987751,
|
| 1178 |
-
"epoch": 0.00098,
|
| 1179 |
-
"frac_reward_zero_std": 0.0,
|
| 1180 |
-
"grad_norm": 0.4623699188232422,
|
| 1181 |
-
"kl": 0.0007145817362470552,
|
| 1182 |
-
"learning_rate": 2.999999906111364e-06,
|
| 1183 |
-
"loss": 0.1004,
|
| 1184 |
-
"num_tokens": 1325693.0,
|
| 1185 |
-
"reward": -20.183937072753906,
|
| 1186 |
-
"reward_std": 13.987285614013672,
|
| 1187 |
-
"rewards/rollout_reward_func/mean": -20.183937072753906,
|
| 1188 |
-
"rewards/rollout_reward_func/std": 13.987285614013672,
|
| 1189 |
-
"sampling/importance_sampling_ratio/max": 1.3713921308517456,
|
| 1190 |
-
"sampling/importance_sampling_ratio/mean": 1.0176022052764893,
|
| 1191 |
-
"sampling/importance_sampling_ratio/min": 0.8512320518493652,
|
| 1192 |
-
"sampling/sampling_logp_difference/max": 0.18564534187316895,
|
| 1193 |
-
"sampling/sampling_logp_difference/mean": 0.012047262862324715,
|
| 1194 |
-
"step": 49,
|
| 1195 |
-
"step_time": 12.366732422000041
|
| 1196 |
-
},
|
| 1197 |
-
{
|
| 1198 |
-
"clip_ratio/high_max": 0.012500000186264515,
|
| 1199 |
-
"clip_ratio/high_mean": 0.0031250000465661287,
|
| 1200 |
-
"clip_ratio/low_mean": 0.0,
|
| 1201 |
-
"clip_ratio/low_min": 0.0,
|
| 1202 |
-
"clip_ratio/region_mean": 0.0031250000465661287,
|
| 1203 |
-
"entropy": 0.42126236110925674,
|
| 1204 |
-
"epoch": 0.001,
|
| 1205 |
-
"grad_norm": 0.42615005373954773,
|
| 1206 |
-
"kl": 0.0008786674443399534,
|
| 1207 |
-
"learning_rate": 2.9999998911114045e-06,
|
| 1208 |
-
"loss": 0.0997,
|
| 1209 |
-
"step": 50,
|
| 1210 |
-
"step_time": 3.6702840139996624
|
| 1211 |
-
},
|
| 1212 |
-
{
|
| 1213 |
-
"clip_ratio/high_max": 0.0,
|
| 1214 |
-
"clip_ratio/high_mean": 0.0,
|
| 1215 |
-
"clip_ratio/low_mean": 0.0,
|
| 1216 |
-
"clip_ratio/low_min": 0.0,
|
| 1217 |
-
"clip_ratio/region_mean": 0.0,
|
| 1218 |
-
"completions/clipped_ratio": 0.84375,
|
| 1219 |
-
"completions/max_length": 16.0,
|
| 1220 |
-
"completions/max_terminated_length": 16.0,
|
| 1221 |
-
"completions/mean_length": 15.0625,
|
| 1222 |
-
"completions/mean_terminated_length": 10.0,
|
| 1223 |
-
"completions/min_length": 3.0,
|
| 1224 |
-
"completions/min_terminated_length": 3.0,
|
| 1225 |
-
"entropy": 0.45311274379491806,
|
| 1226 |
-
"epoch": 0.00102,
|
| 1227 |
-
"frac_reward_zero_std": 0.0,
|
| 1228 |
-
"grad_norm": 0.5940121412277222,
|
| 1229 |
-
"kl": 0.0008269677782664075,
|
| 1230 |
-
"learning_rate": 2.9999998750003373e-06,
|
| 1231 |
-
"loss": 0.0528,
|
| 1232 |
-
"num_tokens": 1378085.0,
|
| 1233 |
-
"reward": -21.42279815673828,
|
| 1234 |
-
"reward_std": 15.290786743164062,
|
| 1235 |
-
"rewards/rollout_reward_func/mean": -21.42279815673828,
|
| 1236 |
-
"rewards/rollout_reward_func/std": 15.290786743164062,
|
| 1237 |
-
"sampling/importance_sampling_ratio/max": 1.2291712760925293,
|
| 1238 |
-
"sampling/importance_sampling_ratio/mean": 1.0124390125274658,
|
| 1239 |
-
"sampling/importance_sampling_ratio/min": 0.8846192359924316,
|
| 1240 |
-
"sampling/sampling_logp_difference/max": 0.21129131317138672,
|
| 1241 |
-
"sampling/sampling_logp_difference/mean": 0.009809674695134163,
|
| 1242 |
-
"step": 51,
|
| 1243 |
-
"step_time": 13.111581994000062
|
| 1244 |
-
},
|
| 1245 |
-
{
|
| 1246 |
-
"clip_ratio/high_max": 0.0,
|
| 1247 |
-
"clip_ratio/high_mean": 0.0,
|
| 1248 |
-
"clip_ratio/low_mean": 0.0,
|
| 1249 |
-
"clip_ratio/low_min": 0.0,
|
| 1250 |
-
"clip_ratio/region_mean": 0.0,
|
| 1251 |
-
"entropy": 0.45369481295347214,
|
| 1252 |
-
"epoch": 0.00104,
|
| 1253 |
-
"grad_norm": 0.5907822251319885,
|
| 1254 |
-
"kl": 0.0009798892569961026,
|
| 1255 |
-
"learning_rate": 2.999999857778162e-06,
|
| 1256 |
-
"loss": 0.0518,
|
| 1257 |
-
"step": 52,
|
| 1258 |
-
"step_time": 3.608162464000088
|
| 1259 |
-
},
|
| 1260 |
-
{
|
| 1261 |
-
"clip_ratio/high_max": 0.01315789483487606,
|
| 1262 |
-
"clip_ratio/high_mean": 0.003289473708719015,
|
| 1263 |
-
"clip_ratio/low_mean": 0.0,
|
| 1264 |
-
"clip_ratio/low_min": 0.0,
|
| 1265 |
-
"clip_ratio/region_mean": 0.003289473708719015,
|
| 1266 |
-
"completions/clipped_ratio": 0.5625,
|
| 1267 |
-
"completions/max_length": 16.0,
|
| 1268 |
-
"completions/max_terminated_length": 14.0,
|
| 1269 |
-
"completions/mean_length": 12.28125,
|
| 1270 |
-
"completions/mean_terminated_length": 7.500000476837158,
|
| 1271 |
-
"completions/min_length": 3.0,
|
| 1272 |
-
"completions/min_terminated_length": 3.0,
|
| 1273 |
-
"entropy": 0.443049892783165,
|
| 1274 |
-
"epoch": 0.00106,
|
| 1275 |
-
"frac_reward_zero_std": 0.0,
|
| 1276 |
-
"grad_norm": 0.45799383521080017,
|
| 1277 |
-
"kl": 0.0016540068900212646,
|
| 1278 |
-
"learning_rate": 2.9999998394448782e-06,
|
| 1279 |
-
"loss": 0.1527,
|
| 1280 |
-
"num_tokens": 1429659.0,
|
| 1281 |
-
"reward": -19.423545837402344,
|
| 1282 |
-
"reward_std": 16.36824607849121,
|
| 1283 |
-
"rewards/rollout_reward_func/mean": -19.423545837402344,
|
| 1284 |
-
"rewards/rollout_reward_func/std": 16.36824607849121,
|
| 1285 |
-
"sampling/importance_sampling_ratio/max": 1.430258870124817,
|
| 1286 |
-
"sampling/importance_sampling_ratio/mean": 1.0030834674835205,
|
| 1287 |
-
"sampling/importance_sampling_ratio/min": 0.7807595729827881,
|
| 1288 |
-
"sampling/sampling_logp_difference/max": 0.17966920137405396,
|
| 1289 |
-
"sampling/sampling_logp_difference/mean": 0.015100691467523575,
|
| 1290 |
-
"step": 53,
|
| 1291 |
-
"step_time": 12.377532298999995
|
| 1292 |
-
},
|
| 1293 |
-
{
|
| 1294 |
-
"clip_ratio/high_max": 0.0,
|
| 1295 |
-
"clip_ratio/high_mean": 0.0,
|
| 1296 |
-
"clip_ratio/low_mean": 0.001953125,
|
| 1297 |
-
"clip_ratio/low_min": 0.0,
|
| 1298 |
-
"clip_ratio/region_mean": 0.001953125,
|
| 1299 |
-
"entropy": 0.4446665346622467,
|
| 1300 |
-
"epoch": 0.00108,
|
| 1301 |
-
"grad_norm": 0.47425946593284607,
|
| 1302 |
-
"kl": 0.001763768566888757,
|
| 1303 |
-
"learning_rate": 2.9999998200004873e-06,
|
| 1304 |
-
"loss": 0.1528,
|
| 1305 |
-
"step": 54,
|
| 1306 |
-
"step_time": 3.616540865999923
|
| 1307 |
-
},
|
| 1308 |
-
{
|
| 1309 |
-
"clip_ratio/high_max": 0.0,
|
| 1310 |
-
"clip_ratio/high_mean": 0.0,
|
| 1311 |
-
"clip_ratio/low_mean": 0.0,
|
| 1312 |
-
"clip_ratio/low_min": 0.0,
|
| 1313 |
-
"clip_ratio/region_mean": 0.0,
|
| 1314 |
-
"completions/clipped_ratio": 0.71875,
|
| 1315 |
-
"completions/max_length": 16.0,
|
| 1316 |
-
"completions/max_terminated_length": 11.0,
|
| 1317 |
-
"completions/mean_length": 13.375,
|
| 1318 |
-
"completions/mean_terminated_length": 6.666666507720947,
|
| 1319 |
-
"completions/min_length": 3.0,
|
| 1320 |
-
"completions/min_terminated_length": 3.0,
|
| 1321 |
-
"entropy": 0.5619853883981705,
|
| 1322 |
-
"epoch": 0.0011,
|
| 1323 |
-
"frac_reward_zero_std": 0.0,
|
| 1324 |
-
"grad_norm": 0.6419393420219421,
|
| 1325 |
-
"kl": 0.00159555982099846,
|
| 1326 |
-
"learning_rate": 2.9999997994449878e-06,
|
| 1327 |
-
"loss": 0.0778,
|
| 1328 |
-
"num_tokens": 1483948.0,
|
| 1329 |
-
"reward": -20.401737213134766,
|
| 1330 |
-
"reward_std": 16.437597274780273,
|
| 1331 |
-
"rewards/rollout_reward_func/mean": -20.401737213134766,
|
| 1332 |
-
"rewards/rollout_reward_func/std": 16.437597274780273,
|
| 1333 |
-
"sampling/importance_sampling_ratio/max": 1.144453525543213,
|
| 1334 |
-
"sampling/importance_sampling_ratio/mean": 0.9405947923660278,
|
| 1335 |
-
"sampling/importance_sampling_ratio/min": 0.610896110534668,
|
| 1336 |
-
"sampling/sampling_logp_difference/max": 0.264723539352417,
|
| 1337 |
-
"sampling/sampling_logp_difference/mean": 0.016143541783094406,
|
| 1338 |
-
"step": 55,
|
| 1339 |
-
"step_time": 14.737511243999961
|
| 1340 |
-
},
|
| 1341 |
-
{
|
| 1342 |
-
"clip_ratio/high_max": 0.0,
|
| 1343 |
-
"clip_ratio/high_mean": 0.0,
|
| 1344 |
-
"clip_ratio/low_mean": 0.0,
|
| 1345 |
-
"clip_ratio/low_min": 0.0,
|
| 1346 |
-
"clip_ratio/region_mean": 0.0,
|
| 1347 |
-
"entropy": 0.5617516338825226,
|
| 1348 |
-
"epoch": 0.00112,
|
| 1349 |
-
"grad_norm": 0.6307651996612549,
|
| 1350 |
-
"kl": 0.0013526002294383943,
|
| 1351 |
-
"learning_rate": 2.9999997777783805e-06,
|
| 1352 |
-
"loss": 0.0769,
|
| 1353 |
-
"step": 56,
|
| 1354 |
-
"step_time": 3.6947174689994426
|
| 1355 |
-
},
|
| 1356 |
-
{
|
| 1357 |
-
"clip_ratio/high_max": 0.0,
|
| 1358 |
-
"clip_ratio/high_mean": 0.0,
|
| 1359 |
-
"clip_ratio/low_mean": 0.0,
|
| 1360 |
-
"clip_ratio/low_min": 0.0,
|
| 1361 |
-
"clip_ratio/region_mean": 0.0,
|
| 1362 |
-
"completions/clipped_ratio": 0.65625,
|
| 1363 |
-
"completions/max_length": 16.0,
|
| 1364 |
-
"completions/max_terminated_length": 16.0,
|
| 1365 |
-
"completions/mean_length": 13.96875,
|
| 1366 |
-
"completions/mean_terminated_length": 10.090909004211426,
|
| 1367 |
-
"completions/min_length": 3.0,
|
| 1368 |
-
"completions/min_terminated_length": 3.0,
|
| 1369 |
-
"entropy": 0.44592302292585373,
|
| 1370 |
-
"epoch": 0.00114,
|
| 1371 |
-
"frac_reward_zero_std": 0.0,
|
| 1372 |
-
"grad_norm": 0.4100278317928314,
|
| 1373 |
-
"kl": 0.0011417547648306936,
|
| 1374 |
-
"learning_rate": 2.999999755000665e-06,
|
| 1375 |
-
"loss": 0.1278,
|
| 1376 |
-
"num_tokens": 1537342.0,
|
| 1377 |
-
"reward": -19.665666580200195,
|
| 1378 |
-
"reward_std": 15.70829963684082,
|
| 1379 |
-
"rewards/rollout_reward_func/mean": -19.665666580200195,
|
| 1380 |
-
"rewards/rollout_reward_func/std": 15.708298683166504,
|
| 1381 |
-
"sampling/importance_sampling_ratio/max": 1.1984679698944092,
|
| 1382 |
-
"sampling/importance_sampling_ratio/mean": 1.0152921676635742,
|
| 1383 |
-
"sampling/importance_sampling_ratio/min": 0.6651595830917358,
|
| 1384 |
-
"sampling/sampling_logp_difference/max": 0.22649931907653809,
|
| 1385 |
-
"sampling/sampling_logp_difference/mean": 0.011754296720027924,
|
| 1386 |
-
"step": 57,
|
| 1387 |
-
"step_time": 13.501649334999911
|
| 1388 |
-
},
|
| 1389 |
-
{
|
| 1390 |
-
"clip_ratio/high_max": 0.0,
|
| 1391 |
-
"clip_ratio/high_mean": 0.0,
|
| 1392 |
-
"clip_ratio/low_mean": 0.0,
|
| 1393 |
-
"clip_ratio/low_min": 0.0,
|
| 1394 |
-
"clip_ratio/region_mean": 0.0,
|
| 1395 |
-
"entropy": 0.4463474452495575,
|
| 1396 |
-
"epoch": 0.00116,
|
| 1397 |
-
"grad_norm": 0.40949589014053345,
|
| 1398 |
-
"kl": 0.0009677642083261162,
|
| 1399 |
-
"learning_rate": 2.9999997311118423e-06,
|
| 1400 |
-
"loss": 0.1279,
|
| 1401 |
-
"step": 58,
|
| 1402 |
-
"step_time": 3.6741236200000458
|
| 1403 |
-
},
|
| 1404 |
-
{
|
| 1405 |
-
"clip_ratio/high_max": 0.0,
|
| 1406 |
-
"clip_ratio/high_mean": 0.0,
|
| 1407 |
-
"clip_ratio/low_mean": 0.0,
|
| 1408 |
-
"clip_ratio/low_min": 0.0,
|
| 1409 |
-
"clip_ratio/region_mean": 0.0,
|
| 1410 |
-
"completions/clipped_ratio": 0.8125,
|
| 1411 |
-
"completions/max_length": 16.0,
|
| 1412 |
-
"completions/max_terminated_length": 13.0,
|
| 1413 |
-
"completions/mean_length": 14.46875,
|
| 1414 |
-
"completions/mean_terminated_length": 7.833333492279053,
|
| 1415 |
-
"completions/min_length": 3.0,
|
| 1416 |
-
"completions/min_terminated_length": 3.0,
|
| 1417 |
-
"entropy": 0.42436008155345917,
|
| 1418 |
-
"epoch": 0.00118,
|
| 1419 |
-
"frac_reward_zero_std": 0.0,
|
| 1420 |
-
"grad_norm": 0.43110400438308716,
|
| 1421 |
-
"kl": 0.0018153611745219678,
|
| 1422 |
-
"learning_rate": 2.9999997061119113e-06,
|
| 1423 |
-
"loss": 0.0759,
|
| 1424 |
-
"num_tokens": 1590549.0,
|
| 1425 |
-
"reward": -23.231876373291016,
|
| 1426 |
-
"reward_std": 14.007892608642578,
|
| 1427 |
-
"rewards/rollout_reward_func/mean": -23.231876373291016,
|
| 1428 |
-
"rewards/rollout_reward_func/std": 14.007891654968262,
|
| 1429 |
-
"sampling/importance_sampling_ratio/max": 1.177204966545105,
|
| 1430 |
-
"sampling/importance_sampling_ratio/mean": 0.9738996624946594,
|
| 1431 |
-
"sampling/importance_sampling_ratio/min": 0.7864365577697754,
|
| 1432 |
-
"sampling/sampling_logp_difference/max": 0.11908435821533203,
|
| 1433 |
-
"sampling/sampling_logp_difference/mean": 0.010427574627101421,
|
| 1434 |
-
"step": 59,
|
| 1435 |
-
"step_time": 13.387355422999917
|
| 1436 |
-
},
|
| 1437 |
-
{
|
| 1438 |
-
"clip_ratio/high_max": 0.0,
|
| 1439 |
-
"clip_ratio/high_mean": 0.0,
|
| 1440 |
-
"clip_ratio/low_mean": 0.0,
|
| 1441 |
-
"clip_ratio/low_min": 0.0,
|
| 1442 |
-
"clip_ratio/region_mean": 0.0,
|
| 1443 |
-
"entropy": 0.42374272644519806,
|
| 1444 |
-
"epoch": 0.0012,
|
| 1445 |
-
"grad_norm": 0.4322911202907562,
|
| 1446 |
-
"kl": 0.0018385612638667226,
|
| 1447 |
-
"learning_rate": 2.999999680000872e-06,
|
| 1448 |
-
"loss": 0.0772,
|
| 1449 |
-
"step": 60,
|
| 1450 |
-
"step_time": 3.6944392599998537
|
| 1451 |
-
},
|
| 1452 |
-
{
|
| 1453 |
-
"clip_ratio/high_max": 0.0078125,
|
| 1454 |
-
"clip_ratio/high_mean": 0.001953125,
|
| 1455 |
-
"clip_ratio/low_mean": 0.0,
|
| 1456 |
-
"clip_ratio/low_min": 0.0,
|
| 1457 |
-
"clip_ratio/region_mean": 0.001953125,
|
| 1458 |
-
"completions/clipped_ratio": 0.71875,
|
| 1459 |
-
"completions/max_length": 16.0,
|
| 1460 |
-
"completions/max_terminated_length": 16.0,
|
| 1461 |
-
"completions/mean_length": 14.25,
|
| 1462 |
-
"completions/mean_terminated_length": 9.777777671813965,
|
| 1463 |
-
"completions/min_length": 4.0,
|
| 1464 |
-
"completions/min_terminated_length": 4.0,
|
| 1465 |
-
"entropy": 0.4840910881757736,
|
| 1466 |
-
"epoch": 0.00122,
|
| 1467 |
-
"frac_reward_zero_std": 0.125,
|
| 1468 |
-
"grad_norm": 0.42527732253074646,
|
| 1469 |
-
"kl": 0.001206837929203175,
|
| 1470 |
-
"learning_rate": 2.9999996527787258e-06,
|
| 1471 |
-
"loss": -0.0195,
|
| 1472 |
-
"num_tokens": 1644232.0,
|
| 1473 |
-
"reward": -20.776317596435547,
|
| 1474 |
-
"reward_std": 15.025127410888672,
|
| 1475 |
-
"rewards/rollout_reward_func/mean": -20.776317596435547,
|
| 1476 |
-
"rewards/rollout_reward_func/std": 15.025128364562988,
|
| 1477 |
-
"sampling/importance_sampling_ratio/max": 1.7264726161956787,
|
| 1478 |
-
"sampling/importance_sampling_ratio/mean": 1.053358554840088,
|
| 1479 |
-
"sampling/importance_sampling_ratio/min": 0.8105581402778625,
|
| 1480 |
-
"sampling/sampling_logp_difference/max": 0.23974990844726562,
|
| 1481 |
-
"sampling/sampling_logp_difference/mean": 0.014413169585168362,
|
| 1482 |
-
"step": 61,
|
| 1483 |
-
"step_time": 12.136578666999867
|
| 1484 |
-
},
|
| 1485 |
-
{
|
| 1486 |
-
"clip_ratio/high_max": 0.0,
|
| 1487 |
-
"clip_ratio/high_mean": 0.0,
|
| 1488 |
-
"clip_ratio/low_mean": 0.0,
|
| 1489 |
-
"clip_ratio/low_min": 0.0,
|
| 1490 |
-
"clip_ratio/region_mean": 0.0,
|
| 1491 |
-
"entropy": 0.4840751215815544,
|
| 1492 |
-
"epoch": 0.00124,
|
| 1493 |
-
"grad_norm": 0.4295765161514282,
|
| 1494 |
-
"kl": 0.0011497176892589778,
|
| 1495 |
-
"learning_rate": 2.9999996244454716e-06,
|
| 1496 |
-
"loss": -0.0186,
|
| 1497 |
-
"step": 62,
|
| 1498 |
-
"step_time": 3.65384825800038
|
| 1499 |
-
},
|
| 1500 |
-
{
|
| 1501 |
-
"clip_ratio/high_max": 0.0,
|
| 1502 |
-
"clip_ratio/high_mean": 0.0,
|
| 1503 |
-
"clip_ratio/low_mean": 0.0,
|
| 1504 |
-
"clip_ratio/low_min": 0.0,
|
| 1505 |
-
"clip_ratio/region_mean": 0.0,
|
| 1506 |
-
"completions/clipped_ratio": 0.5625,
|
| 1507 |
-
"completions/max_length": 16.0,
|
| 1508 |
-
"completions/max_terminated_length": 15.0,
|
| 1509 |
-
"completions/mean_length": 13.875,
|
| 1510 |
-
"completions/mean_terminated_length": 11.142857551574707,
|
| 1511 |
-
"completions/min_length": 3.0,
|
| 1512 |
-
"completions/min_terminated_length": 3.0,
|
| 1513 |
-
"entropy": 0.3919781446456909,
|
| 1514 |
-
"epoch": 0.00126,
|
| 1515 |
-
"frac_reward_zero_std": 0.0,
|
| 1516 |
-
"grad_norm": 0.4190310835838318,
|
| 1517 |
-
"kl": 0.00048243661876767874,
|
| 1518 |
-
"learning_rate": 2.999999595001109e-06,
|
| 1519 |
-
"loss": 0.1343,
|
| 1520 |
-
"num_tokens": 1699341.0,
|
| 1521 |
-
"reward": -14.78390884399414,
|
| 1522 |
-
"reward_std": 15.05773639678955,
|
| 1523 |
-
"rewards/rollout_reward_func/mean": -14.78390884399414,
|
| 1524 |
-
"rewards/rollout_reward_func/std": 15.05773639678955,
|
| 1525 |
-
"sampling/importance_sampling_ratio/max": 1.1515072584152222,
|
| 1526 |
-
"sampling/importance_sampling_ratio/mean": 0.9987464547157288,
|
| 1527 |
-
"sampling/importance_sampling_ratio/min": 0.8501996397972107,
|
| 1528 |
-
"sampling/sampling_logp_difference/max": 0.20190739631652832,
|
| 1529 |
-
"sampling/sampling_logp_difference/mean": 0.008220545016229153,
|
| 1530 |
-
"step": 63,
|
| 1531 |
-
"step_time": 12.735009469000488
|
| 1532 |
-
},
|
| 1533 |
-
{
|
| 1534 |
-
"clip_ratio/high_max": 0.0,
|
| 1535 |
-
"clip_ratio/high_mean": 0.0,
|
| 1536 |
-
"clip_ratio/low_mean": 0.0,
|
| 1537 |
-
"clip_ratio/low_min": 0.0,
|
| 1538 |
-
"clip_ratio/region_mean": 0.0,
|
| 1539 |
-
"entropy": 0.3920908346772194,
|
| 1540 |
-
"epoch": 0.00128,
|
| 1541 |
-
"grad_norm": 0.41324248909950256,
|
| 1542 |
-
"kl": 0.000511926511535421,
|
| 1543 |
-
"learning_rate": 2.9999995644456395e-06,
|
| 1544 |
-
"loss": 0.1353,
|
| 1545 |
-
"step": 64,
|
| 1546 |
-
"step_time": 4.1289103880003495
|
| 1547 |
-
},
|
| 1548 |
-
{
|
| 1549 |
-
"clip_ratio/high_max": 0.0,
|
| 1550 |
-
"clip_ratio/high_mean": 0.0,
|
| 1551 |
-
"clip_ratio/low_mean": 0.0,
|
| 1552 |
-
"clip_ratio/low_min": 0.0,
|
| 1553 |
-
"clip_ratio/region_mean": 0.0,
|
| 1554 |
-
"completions/clipped_ratio": 0.875,
|
| 1555 |
-
"completions/max_length": 16.0,
|
| 1556 |
-
"completions/max_terminated_length": 6.0,
|
| 1557 |
-
"completions/mean_length": 14.5625,
|
| 1558 |
-
"completions/mean_terminated_length": 4.5,
|
| 1559 |
-
"completions/min_length": 4.0,
|
| 1560 |
-
"completions/min_terminated_length": 4.0,
|
| 1561 |
-
"entropy": 0.4723147228360176,
|
| 1562 |
-
"epoch": 0.0013,
|
| 1563 |
-
"frac_reward_zero_std": 0.0,
|
| 1564 |
-
"grad_norm": 0.521558403968811,
|
| 1565 |
-
"kl": 0.0010458012475282885,
|
| 1566 |
-
"learning_rate": 2.9999995327790616e-06,
|
| 1567 |
-
"loss": 0.1397,
|
| 1568 |
-
"num_tokens": 1752125.0,
|
| 1569 |
-
"reward": -23.402069091796875,
|
| 1570 |
-
"reward_std": 12.324657440185547,
|
| 1571 |
-
"rewards/rollout_reward_func/mean": -23.402069091796875,
|
| 1572 |
-
"rewards/rollout_reward_func/std": 12.324657440185547,
|
| 1573 |
-
"sampling/importance_sampling_ratio/max": 1.294718623161316,
|
| 1574 |
-
"sampling/importance_sampling_ratio/mean": 1.0054374933242798,
|
| 1575 |
-
"sampling/importance_sampling_ratio/min": 0.7819744944572449,
|
| 1576 |
-
"sampling/sampling_logp_difference/max": 0.2261056900024414,
|
| 1577 |
-
"sampling/sampling_logp_difference/mean": 0.013052692636847496,
|
| 1578 |
-
"step": 65,
|
| 1579 |
-
"step_time": 12.858984567999869
|
| 1580 |
-
},
|
| 1581 |
-
{
|
| 1582 |
-
"clip_ratio/high_max": 0.0,
|
| 1583 |
-
"clip_ratio/high_mean": 0.0,
|
| 1584 |
-
"clip_ratio/low_mean": 0.0,
|
| 1585 |
-
"clip_ratio/low_min": 0.0,
|
| 1586 |
-
"clip_ratio/region_mean": 0.0,
|
| 1587 |
-
"entropy": 0.4741643965244293,
|
| 1588 |
-
"epoch": 0.00132,
|
| 1589 |
-
"grad_norm": 0.5274317860603333,
|
| 1590 |
-
"kl": 0.001182250547572039,
|
| 1591 |
-
"learning_rate": 2.9999995000013764e-06,
|
| 1592 |
-
"loss": 0.1407,
|
| 1593 |
-
"step": 66,
|
| 1594 |
-
"step_time": 3.581743365999955
|
| 1595 |
-
},
|
| 1596 |
-
{
|
| 1597 |
-
"clip_ratio/high_max": 0.0,
|
| 1598 |
-
"clip_ratio/high_mean": 0.0,
|
| 1599 |
-
"clip_ratio/low_mean": 0.0,
|
| 1600 |
-
"clip_ratio/low_min": 0.0,
|
| 1601 |
-
"clip_ratio/region_mean": 0.0,
|
| 1602 |
-
"completions/clipped_ratio": 0.75,
|
| 1603 |
-
"completions/max_length": 16.0,
|
| 1604 |
-
"completions/max_terminated_length": 16.0,
|
| 1605 |
-
"completions/mean_length": 14.34375,
|
| 1606 |
-
"completions/mean_terminated_length": 9.375,
|
| 1607 |
-
"completions/min_length": 3.0,
|
| 1608 |
-
"completions/min_terminated_length": 3.0,
|
| 1609 |
-
"entropy": 0.47390806674957275,
|
| 1610 |
-
"epoch": 0.00134,
|
| 1611 |
-
"frac_reward_zero_std": 0.0,
|
| 1612 |
-
"grad_norm": 0.5683096051216125,
|
| 1613 |
-
"kl": 0.0013981750817038119,
|
| 1614 |
-
"learning_rate": 2.9999994661125834e-06,
|
| 1615 |
-
"loss": 0.1355,
|
| 1616 |
-
"num_tokens": 1806198.0,
|
| 1617 |
-
"reward": -19.801048278808594,
|
| 1618 |
-
"reward_std": 16.17534637451172,
|
| 1619 |
-
"rewards/rollout_reward_func/mean": -19.801048278808594,
|
| 1620 |
-
"rewards/rollout_reward_func/std": 16.17534637451172,
|
| 1621 |
-
"sampling/importance_sampling_ratio/max": 1.4784481525421143,
|
| 1622 |
-
"sampling/importance_sampling_ratio/mean": 1.0331366062164307,
|
| 1623 |
-
"sampling/importance_sampling_ratio/min": 0.8605573177337646,
|
| 1624 |
-
"sampling/sampling_logp_difference/max": 0.20436906814575195,
|
| 1625 |
-
"sampling/sampling_logp_difference/mean": 0.01395871490240097,
|
| 1626 |
-
"step": 67,
|
| 1627 |
-
"step_time": 12.371656173999781
|
| 1628 |
-
},
|
| 1629 |
-
{
|
| 1630 |
-
"clip_ratio/high_max": 0.0,
|
| 1631 |
-
"clip_ratio/high_mean": 0.0,
|
| 1632 |
-
"clip_ratio/low_mean": 0.0,
|
| 1633 |
-
"clip_ratio/low_min": 0.0,
|
| 1634 |
-
"clip_ratio/region_mean": 0.0,
|
| 1635 |
-
"entropy": 0.47276531904935837,
|
| 1636 |
-
"epoch": 0.00136,
|
| 1637 |
-
"grad_norm": 0.5657866597175598,
|
| 1638 |
-
"kl": 0.0015692545857746154,
|
| 1639 |
-
"learning_rate": 2.999999431112683e-06,
|
| 1640 |
-
"loss": 0.1327,
|
| 1641 |
-
"step": 68,
|
| 1642 |
-
"step_time": 3.63474406000023
|
| 1643 |
-
},
|
| 1644 |
-
{
|
| 1645 |
-
"clip_ratio/high_max": 0.0,
|
| 1646 |
-
"clip_ratio/high_mean": 0.0,
|
| 1647 |
-
"clip_ratio/low_mean": 0.0,
|
| 1648 |
-
"clip_ratio/low_min": 0.0,
|
| 1649 |
-
"clip_ratio/region_mean": 0.0,
|
| 1650 |
-
"completions/clipped_ratio": 0.65625,
|
| 1651 |
-
"completions/max_length": 16.0,
|
| 1652 |
-
"completions/max_terminated_length": 15.0,
|
| 1653 |
-
"completions/mean_length": 13.21875,
|
| 1654 |
-
"completions/mean_terminated_length": 7.909090995788574,
|
| 1655 |
-
"completions/min_length": 3.0,
|
| 1656 |
-
"completions/min_terminated_length": 3.0,
|
| 1657 |
-
"entropy": 0.4877706617116928,
|
| 1658 |
-
"epoch": 0.00138,
|
| 1659 |
-
"frac_reward_zero_std": 0.0,
|
| 1660 |
-
"grad_norm": 0.5023260712623596,
|
| 1661 |
-
"kl": 0.0024651199055369943,
|
| 1662 |
-
"learning_rate": 2.999999395001675e-06,
|
| 1663 |
-
"loss": 0.0828,
|
| 1664 |
-
"num_tokens": 1858004.0,
|
| 1665 |
-
"reward": -18.188522338867188,
|
| 1666 |
-
"reward_std": 19.19774627685547,
|
| 1667 |
-
"rewards/rollout_reward_func/mean": -18.188522338867188,
|
| 1668 |
-
"rewards/rollout_reward_func/std": 19.197744369506836,
|
| 1669 |
-
"sampling/importance_sampling_ratio/max": 1.3357534408569336,
|
| 1670 |
-
"sampling/importance_sampling_ratio/mean": 0.9936745166778564,
|
| 1671 |
-
"sampling/importance_sampling_ratio/min": 0.7666462063789368,
|
| 1672 |
-
"sampling/sampling_logp_difference/max": 0.20173311233520508,
|
| 1673 |
-
"sampling/sampling_logp_difference/mean": 0.016002200543880463,
|
| 1674 |
-
"step": 69,
|
| 1675 |
-
"step_time": 12.922190936000334
|
| 1676 |
-
},
|
| 1677 |
-
{
|
| 1678 |
-
"clip_ratio/high_max": 0.0,
|
| 1679 |
-
"clip_ratio/high_mean": 0.0,
|
| 1680 |
-
"clip_ratio/low_mean": 0.0,
|
| 1681 |
-
"clip_ratio/low_min": 0.0,
|
| 1682 |
-
"clip_ratio/region_mean": 0.0,
|
| 1683 |
-
"entropy": 0.48947182297706604,
|
| 1684 |
-
"epoch": 0.0014,
|
| 1685 |
-
"grad_norm": 0.5027741193771362,
|
| 1686 |
-
"kl": 0.002827476884704083,
|
| 1687 |
-
"learning_rate": 2.9999993577795593e-06,
|
| 1688 |
-
"loss": 0.0816,
|
| 1689 |
-
"step": 70,
|
| 1690 |
-
"step_time": 3.6084864500003277
|
| 1691 |
-
},
|
| 1692 |
-
{
|
| 1693 |
-
"clip_ratio/high_max": 0.0,
|
| 1694 |
-
"clip_ratio/high_mean": 0.0,
|
| 1695 |
-
"clip_ratio/low_mean": 0.0,
|
| 1696 |
-
"clip_ratio/low_min": 0.0,
|
| 1697 |
-
"clip_ratio/region_mean": 0.0,
|
| 1698 |
-
"completions/clipped_ratio": 0.65625,
|
| 1699 |
-
"completions/max_length": 16.0,
|
| 1700 |
-
"completions/max_terminated_length": 16.0,
|
| 1701 |
-
"completions/mean_length": 13.15625,
|
| 1702 |
-
"completions/mean_terminated_length": 7.727272987365723,
|
| 1703 |
-
"completions/min_length": 4.0,
|
| 1704 |
-
"completions/min_terminated_length": 4.0,
|
| 1705 |
-
"entropy": 0.4982185810804367,
|
| 1706 |
-
"epoch": 0.00142,
|
| 1707 |
-
"frac_reward_zero_std": 0.0,
|
| 1708 |
-
"grad_norm": 0.5128209590911865,
|
| 1709 |
-
"kl": 0.004301712120650336,
|
| 1710 |
-
"learning_rate": 2.999999319446336e-06,
|
| 1711 |
-
"loss": 0.1722,
|
| 1712 |
-
"num_tokens": 1912265.0,
|
| 1713 |
-
"reward": -19.690006256103516,
|
| 1714 |
-
"reward_std": 15.063920974731445,
|
| 1715 |
-
"rewards/rollout_reward_func/mean": -19.690006256103516,
|
| 1716 |
-
"rewards/rollout_reward_func/std": 15.063920974731445,
|
| 1717 |
-
"sampling/importance_sampling_ratio/max": 1.3594207763671875,
|
| 1718 |
-
"sampling/importance_sampling_ratio/mean": 1.0231478214263916,
|
| 1719 |
-
"sampling/importance_sampling_ratio/min": 0.6679673790931702,
|
| 1720 |
-
"sampling/sampling_logp_difference/max": 0.2407078742980957,
|
| 1721 |
-
"sampling/sampling_logp_difference/mean": 0.013544058427214622,
|
| 1722 |
-
"step": 71,
|
| 1723 |
-
"step_time": 12.93631576200005
|
| 1724 |
-
},
|
| 1725 |
-
{
|
| 1726 |
-
"clip_ratio/high_max": 0.0,
|
| 1727 |
-
"clip_ratio/high_mean": 0.0,
|
| 1728 |
-
"clip_ratio/low_mean": 0.0,
|
| 1729 |
-
"clip_ratio/low_min": 0.0,
|
| 1730 |
-
"clip_ratio/region_mean": 0.0,
|
| 1731 |
-
"entropy": 0.49808724224567413,
|
| 1732 |
-
"epoch": 0.00144,
|
| 1733 |
-
"grad_norm": 0.4984101355075836,
|
| 1734 |
-
"kl": 0.005028701314586215,
|
| 1735 |
-
"learning_rate": 2.9999992800020054e-06,
|
| 1736 |
-
"loss": 0.1724,
|
| 1737 |
-
"step": 72,
|
| 1738 |
-
"step_time": 3.6676567169997725
|
| 1739 |
-
},
|
| 1740 |
-
{
|
| 1741 |
-
"clip_ratio/high_max": 0.0,
|
| 1742 |
-
"clip_ratio/high_mean": 0.0,
|
| 1743 |
-
"clip_ratio/low_mean": 0.0,
|
| 1744 |
-
"clip_ratio/low_min": 0.0,
|
| 1745 |
-
"clip_ratio/region_mean": 0.0,
|
| 1746 |
-
"completions/clipped_ratio": 0.71875,
|
| 1747 |
-
"completions/max_length": 16.0,
|
| 1748 |
-
"completions/max_terminated_length": 14.0,
|
| 1749 |
-
"completions/mean_length": 14.09375,
|
| 1750 |
-
"completions/mean_terminated_length": 9.222222328186035,
|
| 1751 |
-
"completions/min_length": 3.0,
|
| 1752 |
-
"completions/min_terminated_length": 3.0,
|
| 1753 |
-
"entropy": 0.4910426437854767,
|
| 1754 |
-
"epoch": 0.00146,
|
| 1755 |
-
"frac_reward_zero_std": 0.0,
|
| 1756 |
-
"grad_norm": 0.5857475996017456,
|
| 1757 |
-
"kl": 0.0031242658151313663,
|
| 1758 |
-
"learning_rate": 2.9999992394465676e-06,
|
| 1759 |
-
"loss": 0.0632,
|
| 1760 |
-
"num_tokens": 1965150.0,
|
| 1761 |
-
"reward": -18.6701717376709,
|
| 1762 |
-
"reward_std": 15.390301704406738,
|
| 1763 |
-
"rewards/rollout_reward_func/mean": -18.6701717376709,
|
| 1764 |
-
"rewards/rollout_reward_func/std": 15.390301704406738,
|
| 1765 |
-
"sampling/importance_sampling_ratio/max": 1.6061283349990845,
|
| 1766 |
-
"sampling/importance_sampling_ratio/mean": 1.0126081705093384,
|
| 1767 |
-
"sampling/importance_sampling_ratio/min": 0.7513319849967957,
|
| 1768 |
-
"sampling/sampling_logp_difference/max": 0.38132286071777344,
|
| 1769 |
-
"sampling/sampling_logp_difference/mean": 0.013925802893936634,
|
| 1770 |
-
"step": 73,
|
| 1771 |
-
"step_time": 13.05256272500037
|
| 1772 |
-
},
|
| 1773 |
-
{
|
| 1774 |
-
"clip_ratio/high_max": 0.0,
|
| 1775 |
-
"clip_ratio/high_mean": 0.0,
|
| 1776 |
-
"clip_ratio/low_mean": 0.0,
|
| 1777 |
-
"clip_ratio/low_min": 0.0,
|
| 1778 |
-
"clip_ratio/region_mean": 0.0,
|
| 1779 |
-
"entropy": 0.4915947765111923,
|
| 1780 |
-
"epoch": 0.00148,
|
| 1781 |
-
"grad_norm": 0.5674751996994019,
|
| 1782 |
-
"kl": 0.003559437900548801,
|
| 1783 |
-
"learning_rate": 2.9999991977800225e-06,
|
| 1784 |
-
"loss": 0.0631,
|
| 1785 |
-
"step": 74,
|
| 1786 |
-
"step_time": 4.736504551000053
|
| 1787 |
-
},
|
| 1788 |
-
{
|
| 1789 |
-
"clip_ratio/high_max": 0.0,
|
| 1790 |
-
"clip_ratio/high_mean": 0.0,
|
| 1791 |
-
"clip_ratio/low_mean": 0.0,
|
| 1792 |
-
"clip_ratio/low_min": 0.0,
|
| 1793 |
-
"clip_ratio/region_mean": 0.0,
|
| 1794 |
-
"completions/clipped_ratio": 0.8125,
|
| 1795 |
-
"completions/max_length": 16.0,
|
| 1796 |
-
"completions/max_terminated_length": 14.0,
|
| 1797 |
-
"completions/mean_length": 14.46875,
|
| 1798 |
-
"completions/mean_terminated_length": 7.833333492279053,
|
| 1799 |
-
"completions/min_length": 4.0,
|
| 1800 |
-
"completions/min_terminated_length": 4.0,
|
| 1801 |
-
"entropy": 0.4485122114419937,
|
| 1802 |
-
"epoch": 0.0015,
|
| 1803 |
-
"frac_reward_zero_std": 0.0,
|
| 1804 |
-
"grad_norm": 0.5359442830085754,
|
| 1805 |
-
"kl": 0.0026965075376210734,
|
| 1806 |
-
"learning_rate": 2.9999991550023697e-06,
|
| 1807 |
-
"loss": 0.0904,
|
| 1808 |
-
"num_tokens": 2018404.0,
|
| 1809 |
-
"reward": -22.368053436279297,
|
| 1810 |
-
"reward_std": 13.008703231811523,
|
| 1811 |
-
"rewards/rollout_reward_func/mean": -22.368053436279297,
|
| 1812 |
-
"rewards/rollout_reward_func/std": 13.008703231811523,
|
| 1813 |
-
"sampling/importance_sampling_ratio/max": 1.2978986501693726,
|
| 1814 |
-
"sampling/importance_sampling_ratio/mean": 1.035149335861206,
|
| 1815 |
-
"sampling/importance_sampling_ratio/min": 0.8059366941452026,
|
| 1816 |
-
"sampling/sampling_logp_difference/max": 0.1702725887298584,
|
| 1817 |
-
"sampling/sampling_logp_difference/mean": 0.009985161945223808,
|
| 1818 |
-
"step": 75,
|
| 1819 |
-
"step_time": 13.661403129999599
|
| 1820 |
}
|
| 1821 |
],
|
| 1822 |
"logging_steps": 1.0,
|
| 1823 |
"max_steps": 100000,
|
| 1824 |
-
"num_input_tokens_seen":
|
| 1825 |
"num_train_epochs": 2,
|
| 1826 |
"save_steps": 500,
|
| 1827 |
"stateful_callbacks": {
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.0002,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 10,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 15 |
"clip_ratio/low_mean": 0.0,
|
| 16 |
"clip_ratio/low_min": 0.0,
|
| 17 |
"clip_ratio/region_mean": 0.0,
|
| 18 |
+
"completions/clipped_ratio": 0.90625,
|
| 19 |
"completions/max_length": 16.0,
|
| 20 |
+
"completions/max_terminated_length": 16.0,
|
| 21 |
+
"completions/mean_length": 15.59375,
|
| 22 |
+
"completions/mean_terminated_length": 11.666666984558105,
|
| 23 |
+
"completions/min_length": 9.0,
|
| 24 |
+
"completions/min_terminated_length": 9.0,
|
| 25 |
+
"entropy": 0.45986470580101013,
|
| 26 |
"epoch": 2e-05,
|
| 27 |
"frac_reward_zero_std": 0.0,
|
| 28 |
+
"grad_norm": 0.5555855631828308,
|
| 29 |
"kl": 0.0,
|
| 30 |
"learning_rate": 0.0,
|
| 31 |
+
"loss": -0.0036,
|
| 32 |
+
"num_tokens": 53302.0,
|
| 33 |
+
"reward": -20.565004348754883,
|
| 34 |
+
"reward_std": 12.108231544494629,
|
| 35 |
+
"rewards/rollout_reward_func/mean": -20.565004348754883,
|
| 36 |
+
"rewards/rollout_reward_func/std": 12.108230590820312,
|
| 37 |
+
"sampling/importance_sampling_ratio/max": 1.2069631814956665,
|
| 38 |
+
"sampling/importance_sampling_ratio/mean": 1.0164316892623901,
|
| 39 |
+
"sampling/importance_sampling_ratio/min": 0.691445529460907,
|
| 40 |
+
"sampling/sampling_logp_difference/max": 0.14298677444458008,
|
| 41 |
+
"sampling/sampling_logp_difference/mean": 0.01139700785279274,
|
| 42 |
"step": 1,
|
| 43 |
+
"step_time": 13.593145795000055
|
| 44 |
},
|
| 45 |
{
|
| 46 |
"clip_ratio/high_max": 0.0,
|
|
|
|
| 48 |
"clip_ratio/low_mean": 0.0,
|
| 49 |
"clip_ratio/low_min": 0.0,
|
| 50 |
"clip_ratio/region_mean": 0.0,
|
| 51 |
+
"entropy": 0.45986470580101013,
|
| 52 |
"epoch": 4e-05,
|
| 53 |
+
"grad_norm": 0.5478876829147339,
|
| 54 |
"kl": 0.0,
|
| 55 |
+
"learning_rate": 8.571428571428574e-08,
|
| 56 |
+
"loss": -0.0036,
|
| 57 |
"step": 2,
|
| 58 |
+
"step_time": 3.5665654679996805
|
| 59 |
},
|
| 60 |
{
|
| 61 |
"clip_ratio/high_max": 0.0,
|
| 62 |
"clip_ratio/high_mean": 0.0,
|
| 63 |
+
"clip_ratio/low_mean": 0.0031250000465661287,
|
| 64 |
"clip_ratio/low_min": 0.0,
|
| 65 |
+
"clip_ratio/region_mean": 0.0031250000465661287,
|
| 66 |
"completions/clipped_ratio": 0.75,
|
| 67 |
"completions/max_length": 16.0,
|
| 68 |
+
"completions/max_terminated_length": 16.0,
|
| 69 |
+
"completions/mean_length": 14.4375,
|
| 70 |
+
"completions/mean_terminated_length": 9.75,
|
| 71 |
+
"completions/min_length": 4.0,
|
| 72 |
+
"completions/min_terminated_length": 4.0,
|
| 73 |
+
"entropy": 0.4573390781879425,
|
| 74 |
"epoch": 6e-05,
|
| 75 |
"frac_reward_zero_std": 0.0,
|
| 76 |
+
"grad_norm": 0.4110209047794342,
|
| 77 |
+
"kl": 0.00046882770402589813,
|
| 78 |
+
"learning_rate": 1.7142857142857149e-07,
|
| 79 |
+
"loss": 0.1188,
|
| 80 |
+
"num_tokens": 106349.0,
|
| 81 |
+
"reward": -22.559879302978516,
|
| 82 |
+
"reward_std": 14.563770294189453,
|
| 83 |
+
"rewards/rollout_reward_func/mean": -22.559879302978516,
|
| 84 |
+
"rewards/rollout_reward_func/std": 14.563770294189453,
|
| 85 |
+
"sampling/importance_sampling_ratio/max": 1.2753626108169556,
|
| 86 |
+
"sampling/importance_sampling_ratio/mean": 0.9817989468574524,
|
| 87 |
+
"sampling/importance_sampling_ratio/min": 0.8001569509506226,
|
| 88 |
+
"sampling/sampling_logp_difference/max": 0.2265012264251709,
|
| 89 |
+
"sampling/sampling_logp_difference/mean": 0.012659025378525257,
|
| 90 |
"step": 3,
|
| 91 |
+
"step_time": 13.572802005000085
|
| 92 |
},
|
| 93 |
{
|
| 94 |
+
"clip_ratio/high_max": 0.0078125,
|
| 95 |
+
"clip_ratio/high_mean": 0.001953125,
|
| 96 |
"clip_ratio/low_mean": 0.0,
|
| 97 |
"clip_ratio/low_min": 0.0,
|
| 98 |
+
"clip_ratio/region_mean": 0.001953125,
|
| 99 |
+
"entropy": 0.45671091973781586,
|
| 100 |
"epoch": 8e-05,
|
| 101 |
+
"grad_norm": 0.3884861469268799,
|
| 102 |
+
"kl": 0.00040267666918225586,
|
| 103 |
+
"learning_rate": 2.571428571428572e-07,
|
| 104 |
+
"loss": 0.1171,
|
| 105 |
"step": 4,
|
| 106 |
+
"step_time": 3.672753089998878
|
| 107 |
},
|
| 108 |
{
|
| 109 |
+
"clip_ratio/high_max": 0.0,
|
| 110 |
+
"clip_ratio/high_mean": 0.0,
|
| 111 |
"clip_ratio/low_mean": 0.0,
|
| 112 |
"clip_ratio/low_min": 0.0,
|
| 113 |
+
"clip_ratio/region_mean": 0.0,
|
| 114 |
+
"completions/clipped_ratio": 0.78125,
|
| 115 |
"completions/max_length": 16.0,
|
| 116 |
+
"completions/max_terminated_length": 13.0,
|
| 117 |
+
"completions/mean_length": 14.09375,
|
| 118 |
+
"completions/mean_terminated_length": 7.285714626312256,
|
| 119 |
"completions/min_length": 4.0,
|
| 120 |
"completions/min_terminated_length": 4.0,
|
| 121 |
+
"entropy": 0.4637308567762375,
|
| 122 |
"epoch": 0.0001,
|
| 123 |
"frac_reward_zero_std": 0.0,
|
| 124 |
+
"grad_norm": 0.5923936367034912,
|
| 125 |
+
"kl": 0.00045897291784058325,
|
| 126 |
+
"learning_rate": 3.4285714285714297e-07,
|
| 127 |
+
"loss": 0.1461,
|
| 128 |
+
"num_tokens": 158201.0,
|
| 129 |
+
"reward": -23.239702224731445,
|
| 130 |
+
"reward_std": 11.33220386505127,
|
| 131 |
+
"rewards/rollout_reward_func/mean": -23.239702224731445,
|
| 132 |
+
"rewards/rollout_reward_func/std": 11.332201957702637,
|
| 133 |
+
"sampling/importance_sampling_ratio/max": 1.179511308670044,
|
| 134 |
+
"sampling/importance_sampling_ratio/mean": 0.9810092449188232,
|
| 135 |
+
"sampling/importance_sampling_ratio/min": 0.7553079128265381,
|
| 136 |
+
"sampling/sampling_logp_difference/max": 0.19478893280029297,
|
| 137 |
+
"sampling/sampling_logp_difference/mean": 0.012187635526061058,
|
| 138 |
"step": 5,
|
| 139 |
+
"step_time": 12.394922012998904
|
| 140 |
},
|
| 141 |
{
|
| 142 |
+
"epoch": 0.0001,
|
| 143 |
+
"eval_clip_ratio/high_max": 0.0,
|
| 144 |
+
"eval_clip_ratio/high_mean": 0.0,
|
| 145 |
+
"eval_clip_ratio/low_mean": 0.0,
|
| 146 |
+
"eval_clip_ratio/low_min": 0.0,
|
| 147 |
+
"eval_clip_ratio/region_mean": 0.0,
|
| 148 |
+
"eval_completions/clipped_ratio": 0.85,
|
| 149 |
+
"eval_completions/max_length": 16.0,
|
| 150 |
+
"eval_completions/max_terminated_length": 5.4,
|
| 151 |
+
"eval_completions/mean_length": 15.225,
|
| 152 |
+
"eval_completions/mean_terminated_length": 5.15,
|
| 153 |
+
"eval_completions/min_length": 12.9,
|
| 154 |
+
"eval_completions/min_terminated_length": 4.9,
|
| 155 |
+
"eval_entropy": 0.45741056799888613,
|
| 156 |
+
"eval_frac_reward_zero_std": 0.0,
|
| 157 |
+
"eval_kl": 0.00042195062269456685,
|
| 158 |
+
"eval_loss": 0.004154330585151911,
|
| 159 |
+
"eval_num_tokens": 158201.0,
|
| 160 |
+
"eval_reward": -25.101631736755373,
|
| 161 |
+
"eval_reward_std": 13.229044437408447,
|
| 162 |
+
"eval_rewards/rollout_reward_func/mean": -25.101631736755373,
|
| 163 |
+
"eval_rewards/rollout_reward_func/std": 13.2290442943573,
|
| 164 |
+
"eval_runtime": 13.4386,
|
| 165 |
+
"eval_samples_per_second": 0.744,
|
| 166 |
+
"eval_sampling/importance_sampling_ratio/max": 1.112263822555542,
|
| 167 |
+
"eval_sampling/importance_sampling_ratio/mean": 0.9942494869232178,
|
| 168 |
+
"eval_sampling/importance_sampling_ratio/min": 0.8756316006183624,
|
| 169 |
+
"eval_sampling/sampling_logp_difference/max": 0.1471207022666931,
|
| 170 |
+
"eval_sampling/sampling_logp_difference/mean": 0.013286880496889353,
|
| 171 |
+
"eval_steps_per_second": 0.223,
|
| 172 |
+
"step": 5
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"clip_ratio/high_max": 0.0,
|
| 176 |
+
"clip_ratio/high_mean": 0.0,
|
| 177 |
+
"clip_ratio/low_mean": 0.0,
|
| 178 |
+
"clip_ratio/low_min": 0.0,
|
| 179 |
+
"clip_ratio/region_mean": 0.0,
|
| 180 |
+
"entropy": 0.46300478279590607,
|
| 181 |
"epoch": 0.00012,
|
| 182 |
+
"grad_norm": 0.5754485726356506,
|
| 183 |
+
"kl": 0.0004794619817403145,
|
| 184 |
+
"learning_rate": 4.2857142857142867e-07,
|
| 185 |
+
"loss": 0.1461,
|
| 186 |
"step": 6,
|
| 187 |
+
"step_time": 3.5247847910004566
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 188 |
},
|
| 189 |
{
|
| 190 |
"clip_ratio/high_max": 0.0,
|
|
|
|
| 194 |
"clip_ratio/region_mean": 0.0,
|
| 195 |
"completions/clipped_ratio": 0.78125,
|
| 196 |
"completions/max_length": 16.0,
|
| 197 |
+
"completions/max_terminated_length": 14.0,
|
| 198 |
"completions/mean_length": 14.6875,
|
| 199 |
"completions/mean_terminated_length": 10.0,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 200 |
"completions/min_length": 3.0,
|
| 201 |
"completions/min_terminated_length": 3.0,
|
| 202 |
+
"entropy": 0.4924817681312561,
|
| 203 |
+
"epoch": 0.00014,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 204 |
"frac_reward_zero_std": 0.0,
|
| 205 |
+
"grad_norm": 0.5342097282409668,
|
| 206 |
+
"kl": 0.00047566716602887027,
|
| 207 |
+
"learning_rate": 5.142857142857144e-07,
|
| 208 |
+
"loss": 0.0788,
|
| 209 |
+
"num_tokens": 211022.0,
|
| 210 |
+
"reward": -21.68277359008789,
|
| 211 |
+
"reward_std": 14.98656940460205,
|
| 212 |
+
"rewards/rollout_reward_func/mean": -21.68277359008789,
|
| 213 |
+
"rewards/rollout_reward_func/std": 14.986568450927734,
|
| 214 |
+
"sampling/importance_sampling_ratio/max": 1.2211930751800537,
|
| 215 |
+
"sampling/importance_sampling_ratio/mean": 0.9879010319709778,
|
| 216 |
+
"sampling/importance_sampling_ratio/min": 0.7968342304229736,
|
| 217 |
+
"sampling/sampling_logp_difference/max": 0.2077500820159912,
|
| 218 |
+
"sampling/sampling_logp_difference/mean": 0.015078969299793243,
|
| 219 |
+
"step": 7,
|
| 220 |
+
"step_time": 13.125986247000583
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"clip_ratio/high_max": 0.0,
|
|
|
|
| 225 |
"clip_ratio/low_mean": 0.001953125,
|
| 226 |
"clip_ratio/low_min": 0.0,
|
| 227 |
"clip_ratio/region_mean": 0.001953125,
|
| 228 |
+
"entropy": 0.4939192831516266,
|
| 229 |
+
"epoch": 0.00016,
|
| 230 |
+
"grad_norm": 0.5269240140914917,
|
| 231 |
+
"kl": 0.0005614961846731603,
|
| 232 |
+
"learning_rate": 6.000000000000002e-07,
|
| 233 |
+
"loss": 0.078,
|
| 234 |
+
"step": 8,
|
| 235 |
+
"step_time": 4.57110312200075
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 236 |
},
|
| 237 |
{
|
| 238 |
"clip_ratio/high_max": 0.0,
|
|
|
|
| 240 |
"clip_ratio/low_mean": 0.0,
|
| 241 |
"clip_ratio/low_min": 0.0,
|
| 242 |
"clip_ratio/region_mean": 0.0,
|
| 243 |
+
"completions/clipped_ratio": 0.84375,
|
| 244 |
"completions/max_length": 16.0,
|
| 245 |
+
"completions/max_terminated_length": 16.0,
|
| 246 |
+
"completions/mean_length": 15.53125,
|
| 247 |
+
"completions/mean_terminated_length": 13.0,
|
| 248 |
+
"completions/min_length": 3.0,
|
| 249 |
+
"completions/min_terminated_length": 3.0,
|
| 250 |
+
"entropy": 0.4650999382138252,
|
| 251 |
+
"epoch": 0.00018,
|
| 252 |
"frac_reward_zero_std": 0.0,
|
| 253 |
+
"grad_norm": 0.4413743019104004,
|
| 254 |
+
"kl": 0.0005437940853880718,
|
| 255 |
+
"learning_rate": 6.857142857142859e-07,
|
| 256 |
+
"loss": 0.0419,
|
| 257 |
+
"num_tokens": 262651.0,
|
| 258 |
+
"reward": -24.077335357666016,
|
| 259 |
+
"reward_std": 11.67958927154541,
|
| 260 |
+
"rewards/rollout_reward_func/mean": -24.077335357666016,
|
| 261 |
+
"rewards/rollout_reward_func/std": 11.67958927154541,
|
| 262 |
+
"sampling/importance_sampling_ratio/max": 1.2573319673538208,
|
| 263 |
+
"sampling/importance_sampling_ratio/mean": 1.0008736848831177,
|
| 264 |
+
"sampling/importance_sampling_ratio/min": 0.8466115593910217,
|
| 265 |
+
"sampling/sampling_logp_difference/max": 0.15867352485656738,
|
| 266 |
+
"sampling/sampling_logp_difference/mean": 0.011756965890526772,
|
| 267 |
+
"step": 9,
|
| 268 |
+
"step_time": 12.814324255000429
|
| 269 |
},
|
| 270 |
{
|
| 271 |
"clip_ratio/high_max": 0.0,
|
|
|
|
| 273 |
"clip_ratio/low_mean": 0.0,
|
| 274 |
"clip_ratio/low_min": 0.0,
|
| 275 |
"clip_ratio/region_mean": 0.0,
|
| 276 |
+
"entropy": 0.46368078887462616,
|
| 277 |
+
"epoch": 0.0002,
|
| 278 |
+
"grad_norm": 0.4458825886249542,
|
| 279 |
+
"kl": 0.0005188173963688314,
|
| 280 |
+
"learning_rate": 7.714285714285716e-07,
|
| 281 |
+
"loss": 0.0428,
|
| 282 |
+
"step": 10,
|
| 283 |
+
"step_time": 3.553372275999209
|
| 284 |
},
|
| 285 |
{
|
| 286 |
+
"epoch": 0.0002,
|
| 287 |
+
"eval_clip_ratio/high_max": 0.0,
|
| 288 |
+
"eval_clip_ratio/high_mean": 0.0,
|
| 289 |
+
"eval_clip_ratio/low_mean": 0.0,
|
| 290 |
+
"eval_clip_ratio/low_min": 0.0,
|
| 291 |
+
"eval_clip_ratio/region_mean": 0.0,
|
| 292 |
+
"eval_completions/clipped_ratio": 0.7,
|
| 293 |
+
"eval_completions/max_length": 16.0,
|
| 294 |
+
"eval_completions/max_terminated_length": 9.8,
|
| 295 |
+
"eval_completions/mean_length": 14.7,
|
| 296 |
+
"eval_completions/mean_terminated_length": 9.16666669845581,
|
| 297 |
+
"eval_completions/min_length": 11.8,
|
| 298 |
+
"eval_completions/min_terminated_length": 8.6,
|
| 299 |
+
"eval_entropy": 0.40578661262989046,
|
| 300 |
+
"eval_frac_reward_zero_std": 0.0,
|
| 301 |
+
"eval_kl": 0.00042531737562967464,
|
| 302 |
+
"eval_loss": 0.02340291067957878,
|
| 303 |
+
"eval_num_tokens": 262651.0,
|
| 304 |
+
"eval_reward": -18.063774490356444,
|
| 305 |
+
"eval_reward_std": 14.382386589050293,
|
| 306 |
+
"eval_rewards/rollout_reward_func/mean": -18.063774490356444,
|
| 307 |
+
"eval_rewards/rollout_reward_func/std": 14.382386589050293,
|
| 308 |
+
"eval_runtime": 13.98,
|
| 309 |
+
"eval_samples_per_second": 0.715,
|
| 310 |
+
"eval_sampling/importance_sampling_ratio/max": 1.1354158759117126,
|
| 311 |
+
"eval_sampling/importance_sampling_ratio/mean": 1.0028302788734436,
|
| 312 |
+
"eval_sampling/importance_sampling_ratio/min": 0.9048627257347107,
|
| 313 |
+
"eval_sampling/sampling_logp_difference/max": 0.13887375593185425,
|
| 314 |
+
"eval_sampling/sampling_logp_difference/mean": 0.011902038706466556,
|
| 315 |
+
"eval_steps_per_second": 0.215,
|
| 316 |
+
"step": 10
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 317 |
}
|
| 318 |
],
|
| 319 |
"logging_steps": 1.0,
|
| 320 |
"max_steps": 100000,
|
| 321 |
+
"num_input_tokens_seen": 262651,
|
| 322 |
"num_train_epochs": 2,
|
| 323 |
"save_steps": 500,
|
| 324 |
"stateful_callbacks": {
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 8081
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6ec21def335d91e767cfc8442100434c1b8f711cd3db15cf69c2a4331fa4d590
|
| 3 |
size 8081
|