Delete trainer_state.json
Browse files- trainer_state.json +0 -3034
trainer_state.json
DELETED
|
@@ -1,3034 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_global_step": null,
|
| 3 |
-
"best_metric": null,
|
| 4 |
-
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 0.192,
|
| 6 |
-
"eval_steps": 500,
|
| 7 |
-
"global_step": 240,
|
| 8 |
-
"is_hyper_param_search": false,
|
| 9 |
-
"is_local_process_zero": true,
|
| 10 |
-
"is_world_process_zero": true,
|
| 11 |
-
"log_history": [
|
| 12 |
-
{
|
| 13 |
-
"clip_ratio": 0.0,
|
| 14 |
-
"completion_length": 2761.8828125,
|
| 15 |
-
"epoch": 0.0008,
|
| 16 |
-
"grad_norm": 0.5506983381296895,
|
| 17 |
-
"kl": 0.0,
|
| 18 |
-
"learning_rate": 5e-07,
|
| 19 |
-
"loss": 0.0464,
|
| 20 |
-
"reward": 0.23066407296573743,
|
| 21 |
-
"reward_std": 0.45144409872591496,
|
| 22 |
-
"rewards/end_of_conversation_reward_func": 0.046093751094304025,
|
| 23 |
-
"rewards/end_rm_reward_func": 0.2392578125,
|
| 24 |
-
"rewards/length_reward_func": -0.0546875,
|
| 25 |
-
"rewards/thinking_reward_func": 0.0,
|
| 26 |
-
"step": 1
|
| 27 |
-
},
|
| 28 |
-
{
|
| 29 |
-
"clip_ratio": 0.004609759445884265,
|
| 30 |
-
"epoch": 0.0016,
|
| 31 |
-
"grad_norm": 0.5329509809633897,
|
| 32 |
-
"kl": 0.0012302398681640625,
|
| 33 |
-
"learning_rate": 5e-07,
|
| 34 |
-
"loss": 0.0463,
|
| 35 |
-
"step": 2
|
| 36 |
-
},
|
| 37 |
-
{
|
| 38 |
-
"clip_ratio": 0.0,
|
| 39 |
-
"completion_length": 3313.046875,
|
| 40 |
-
"epoch": 0.0024,
|
| 41 |
-
"grad_norm": 0.44037361030440025,
|
| 42 |
-
"kl": 0.0013537406921386719,
|
| 43 |
-
"learning_rate": 5e-07,
|
| 44 |
-
"loss": 0.0676,
|
| 45 |
-
"reward": 0.0886718796682544,
|
| 46 |
-
"reward_std": 0.45538751501590014,
|
| 47 |
-
"rewards/end_of_conversation_reward_func": 0.047656250768341124,
|
| 48 |
-
"rewards/end_rm_reward_func": 0.146484375,
|
| 49 |
-
"rewards/length_reward_func": -0.10546875,
|
| 50 |
-
"rewards/thinking_reward_func": 0.0,
|
| 51 |
-
"step": 3
|
| 52 |
-
},
|
| 53 |
-
{
|
| 54 |
-
"clip_ratio": 0.004155338363489136,
|
| 55 |
-
"epoch": 0.0032,
|
| 56 |
-
"grad_norm": 0.4238348103819464,
|
| 57 |
-
"kl": 0.00133514404296875,
|
| 58 |
-
"learning_rate": 5e-07,
|
| 59 |
-
"loss": 0.0677,
|
| 60 |
-
"step": 4
|
| 61 |
-
},
|
| 62 |
-
{
|
| 63 |
-
"clip_ratio": 0.0,
|
| 64 |
-
"completion_length": 3180.3046875,
|
| 65 |
-
"epoch": 0.004,
|
| 66 |
-
"grad_norm": 0.487685724080159,
|
| 67 |
-
"kl": 0.00140380859375,
|
| 68 |
-
"learning_rate": 5e-07,
|
| 69 |
-
"loss": 0.0617,
|
| 70 |
-
"reward": 0.12480469699949026,
|
| 71 |
-
"reward_std": 0.393667503958568,
|
| 72 |
-
"rewards/end_of_conversation_reward_func": 0.043750000768341124,
|
| 73 |
-
"rewards/end_rm_reward_func": 0.1279296875,
|
| 74 |
-
"rewards/length_reward_func": -0.046875,
|
| 75 |
-
"rewards/thinking_reward_func": 0.0,
|
| 76 |
-
"step": 5
|
| 77 |
-
},
|
| 78 |
-
{
|
| 79 |
-
"clip_ratio": 0.00448301769210957,
|
| 80 |
-
"epoch": 0.0048,
|
| 81 |
-
"grad_norm": 0.49115718476339393,
|
| 82 |
-
"kl": 0.001739501953125,
|
| 83 |
-
"learning_rate": 5e-07,
|
| 84 |
-
"loss": 0.0615,
|
| 85 |
-
"step": 6
|
| 86 |
-
},
|
| 87 |
-
{
|
| 88 |
-
"clip_ratio": 0.0,
|
| 89 |
-
"completion_length": 3680.03125,
|
| 90 |
-
"epoch": 0.0056,
|
| 91 |
-
"grad_norm": 0.36669507381078525,
|
| 92 |
-
"kl": 0.0017371177673339844,
|
| 93 |
-
"learning_rate": 5e-07,
|
| 94 |
-
"loss": 0.1115,
|
| 95 |
-
"reward": 0.12597656878642738,
|
| 96 |
-
"reward_std": 0.42063513631001115,
|
| 97 |
-
"rewards/end_of_conversation_reward_func": 0.035156251047737896,
|
| 98 |
-
"rewards/end_rm_reward_func": 0.1845703125,
|
| 99 |
-
"rewards/length_reward_func": -0.09375,
|
| 100 |
-
"rewards/thinking_reward_func": 0.0,
|
| 101 |
-
"step": 7
|
| 102 |
-
},
|
| 103 |
-
{
|
| 104 |
-
"clip_ratio": 0.004220222064759582,
|
| 105 |
-
"epoch": 0.0064,
|
| 106 |
-
"grad_norm": 0.36704470764422,
|
| 107 |
-
"kl": 0.001245260238647461,
|
| 108 |
-
"learning_rate": 5e-07,
|
| 109 |
-
"loss": 0.1116,
|
| 110 |
-
"step": 8
|
| 111 |
-
},
|
| 112 |
-
{
|
| 113 |
-
"clip_ratio": 0.0,
|
| 114 |
-
"completion_length": 3056.125,
|
| 115 |
-
"epoch": 0.0072,
|
| 116 |
-
"grad_norm": 0.3963636827663633,
|
| 117 |
-
"kl": 0.0019788742065429688,
|
| 118 |
-
"learning_rate": 5e-07,
|
| 119 |
-
"loss": 0.0823,
|
| 120 |
-
"reward": 0.259960952505935,
|
| 121 |
-
"reward_std": 0.4595574662089348,
|
| 122 |
-
"rewards/end_of_conversation_reward_func": 0.05000000080326572,
|
| 123 |
-
"rewards/end_rm_reward_func": 0.2685546875,
|
| 124 |
-
"rewards/length_reward_func": -0.05859375,
|
| 125 |
-
"rewards/thinking_reward_func": 0.0,
|
| 126 |
-
"step": 9
|
| 127 |
-
},
|
| 128 |
-
{
|
| 129 |
-
"clip_ratio": 0.004534602427156642,
|
| 130 |
-
"epoch": 0.008,
|
| 131 |
-
"grad_norm": 0.38789534112248125,
|
| 132 |
-
"kl": 0.001991748809814453,
|
| 133 |
-
"learning_rate": 5e-07,
|
| 134 |
-
"loss": 0.0822,
|
| 135 |
-
"step": 10
|
| 136 |
-
},
|
| 137 |
-
{
|
| 138 |
-
"clip_ratio": 0.0,
|
| 139 |
-
"completion_length": 3453.7265625,
|
| 140 |
-
"epoch": 0.0088,
|
| 141 |
-
"grad_norm": 0.38064460431215713,
|
| 142 |
-
"kl": 0.002022266387939453,
|
| 143 |
-
"learning_rate": 5e-07,
|
| 144 |
-
"loss": 0.0951,
|
| 145 |
-
"reward": 0.15000000572763383,
|
| 146 |
-
"reward_std": 0.4482696056365967,
|
| 147 |
-
"rewards/end_of_conversation_reward_func": 0.044531251245643944,
|
| 148 |
-
"rewards/end_rm_reward_func": 0.19140625,
|
| 149 |
-
"rewards/length_reward_func": -0.0859375,
|
| 150 |
-
"rewards/thinking_reward_func": 0.0,
|
| 151 |
-
"step": 11
|
| 152 |
-
},
|
| 153 |
-
{
|
| 154 |
-
"clip_ratio": 0.004099749305169098,
|
| 155 |
-
"epoch": 0.0096,
|
| 156 |
-
"grad_norm": 0.3751699254242308,
|
| 157 |
-
"kl": 0.0019767284393310547,
|
| 158 |
-
"learning_rate": 5e-07,
|
| 159 |
-
"loss": 0.0951,
|
| 160 |
-
"step": 12
|
| 161 |
-
},
|
| 162 |
-
{
|
| 163 |
-
"clip_ratio": 0.0,
|
| 164 |
-
"completion_length": 3352.5234375,
|
| 165 |
-
"epoch": 0.0104,
|
| 166 |
-
"grad_norm": 0.3498351189608962,
|
| 167 |
-
"kl": 0.0030684471130371094,
|
| 168 |
-
"learning_rate": 5e-07,
|
| 169 |
-
"loss": 0.1119,
|
| 170 |
-
"reward": 0.13613282376900315,
|
| 171 |
-
"reward_std": 0.4642677316442132,
|
| 172 |
-
"rewards/end_of_conversation_reward_func": 0.041406251140870154,
|
| 173 |
-
"rewards/end_rm_reward_func": 0.1650390625,
|
| 174 |
-
"rewards/length_reward_func": -0.0703125,
|
| 175 |
-
"rewards/thinking_reward_func": 0.0,
|
| 176 |
-
"step": 13
|
| 177 |
-
},
|
| 178 |
-
{
|
| 179 |
-
"clip_ratio": 0.0042787125712493435,
|
| 180 |
-
"epoch": 0.0112,
|
| 181 |
-
"grad_norm": 0.34790390559755574,
|
| 182 |
-
"kl": 0.0029201507568359375,
|
| 183 |
-
"learning_rate": 5e-07,
|
| 184 |
-
"loss": 0.112,
|
| 185 |
-
"step": 14
|
| 186 |
-
},
|
| 187 |
-
{
|
| 188 |
-
"clip_ratio": 0.0,
|
| 189 |
-
"completion_length": 2548.421875,
|
| 190 |
-
"epoch": 0.012,
|
| 191 |
-
"grad_norm": 0.5282889686332808,
|
| 192 |
-
"kl": 0.003574371337890625,
|
| 193 |
-
"learning_rate": 5e-07,
|
| 194 |
-
"loss": 0.0372,
|
| 195 |
-
"reward": 0.3990234467200935,
|
| 196 |
-
"reward_std": 0.3981876859907061,
|
| 197 |
-
"rewards/end_of_conversation_reward_func": 0.0523437510128133,
|
| 198 |
-
"rewards/end_rm_reward_func": 0.3779296875,
|
| 199 |
-
"rewards/length_reward_func": -0.03125,
|
| 200 |
-
"rewards/thinking_reward_func": 0.0,
|
| 201 |
-
"step": 15
|
| 202 |
-
},
|
| 203 |
-
{
|
| 204 |
-
"clip_ratio": 0.0044180305558256805,
|
| 205 |
-
"epoch": 0.0128,
|
| 206 |
-
"grad_norm": 0.5245952101164439,
|
| 207 |
-
"kl": 0.004303455352783203,
|
| 208 |
-
"learning_rate": 5e-07,
|
| 209 |
-
"loss": 0.0371,
|
| 210 |
-
"step": 16
|
| 211 |
-
},
|
| 212 |
-
{
|
| 213 |
-
"clip_ratio": 0.0,
|
| 214 |
-
"completion_length": 3264.7578125,
|
| 215 |
-
"epoch": 0.0136,
|
| 216 |
-
"grad_norm": 0.4930092476738244,
|
| 217 |
-
"kl": 0.0028238296508789062,
|
| 218 |
-
"learning_rate": 5e-07,
|
| 219 |
-
"loss": 0.0877,
|
| 220 |
-
"reward": 0.058398447930812836,
|
| 221 |
-
"reward_std": 0.43455274496227503,
|
| 222 |
-
"rewards/end_of_conversation_reward_func": 0.043750000884756446,
|
| 223 |
-
"rewards/end_rm_reward_func": 0.1083984375,
|
| 224 |
-
"rewards/length_reward_func": -0.09375,
|
| 225 |
-
"rewards/thinking_reward_func": 0.0,
|
| 226 |
-
"step": 17
|
| 227 |
-
},
|
| 228 |
-
{
|
| 229 |
-
"clip_ratio": 0.0050234832015121356,
|
| 230 |
-
"epoch": 0.0144,
|
| 231 |
-
"grad_norm": 0.4879552129621058,
|
| 232 |
-
"kl": 0.0030431747436523438,
|
| 233 |
-
"learning_rate": 5e-07,
|
| 234 |
-
"loss": 0.0877,
|
| 235 |
-
"step": 18
|
| 236 |
-
},
|
| 237 |
-
{
|
| 238 |
-
"clip_ratio": 0.0,
|
| 239 |
-
"completion_length": 3004.109375,
|
| 240 |
-
"epoch": 0.0152,
|
| 241 |
-
"grad_norm": 0.44669678095839005,
|
| 242 |
-
"kl": 0.0027985572814941406,
|
| 243 |
-
"learning_rate": 5e-07,
|
| 244 |
-
"loss": 0.0234,
|
| 245 |
-
"reward": 0.30312501173466444,
|
| 246 |
-
"reward_std": 0.38858885364606977,
|
| 247 |
-
"rewards/end_of_conversation_reward_func": 0.04140625096624717,
|
| 248 |
-
"rewards/end_rm_reward_func": 0.30078125,
|
| 249 |
-
"rewards/length_reward_func": -0.0390625,
|
| 250 |
-
"rewards/thinking_reward_func": 0.0,
|
| 251 |
-
"step": 19
|
| 252 |
-
},
|
| 253 |
-
{
|
| 254 |
-
"clip_ratio": 0.003934504988137633,
|
| 255 |
-
"epoch": 0.016,
|
| 256 |
-
"grad_norm": 0.44274866510741123,
|
| 257 |
-
"kl": 0.0028510093688964844,
|
| 258 |
-
"learning_rate": 5e-07,
|
| 259 |
-
"loss": 0.0232,
|
| 260 |
-
"step": 20
|
| 261 |
-
},
|
| 262 |
-
{
|
| 263 |
-
"clip_ratio": 0.0,
|
| 264 |
-
"completion_length": 3229.953125,
|
| 265 |
-
"epoch": 0.0168,
|
| 266 |
-
"grad_norm": 0.48924113242313605,
|
| 267 |
-
"kl": 0.0034546852111816406,
|
| 268 |
-
"learning_rate": 5e-07,
|
| 269 |
-
"loss": 0.0991,
|
| 270 |
-
"reward": 0.26152344699949026,
|
| 271 |
-
"reward_std": 0.45786723494529724,
|
| 272 |
-
"rewards/end_of_conversation_reward_func": 0.05546875047730282,
|
| 273 |
-
"rewards/end_rm_reward_func": 0.2724609375,
|
| 274 |
-
"rewards/length_reward_func": -0.06640625,
|
| 275 |
-
"rewards/thinking_reward_func": 0.0,
|
| 276 |
-
"step": 21
|
| 277 |
-
},
|
| 278 |
-
{
|
| 279 |
-
"clip_ratio": 0.004537505708867684,
|
| 280 |
-
"epoch": 0.0176,
|
| 281 |
-
"grad_norm": 0.4709410237868635,
|
| 282 |
-
"kl": 0.0032873153686523438,
|
| 283 |
-
"learning_rate": 5e-07,
|
| 284 |
-
"loss": 0.0992,
|
| 285 |
-
"step": 22
|
| 286 |
-
},
|
| 287 |
-
{
|
| 288 |
-
"clip_ratio": 0.0,
|
| 289 |
-
"completion_length": 3771.703125,
|
| 290 |
-
"epoch": 0.0184,
|
| 291 |
-
"grad_norm": 0.4520127965646581,
|
| 292 |
-
"kl": 0.003025531768798828,
|
| 293 |
-
"learning_rate": 5e-07,
|
| 294 |
-
"loss": 0.0923,
|
| 295 |
-
"reward": 0.1816406373400241,
|
| 296 |
-
"reward_std": 0.3960149073973298,
|
| 297 |
-
"rewards/end_of_conversation_reward_func": 0.04296875064028427,
|
| 298 |
-
"rewards/end_rm_reward_func": 0.236328125,
|
| 299 |
-
"rewards/length_reward_func": -0.09765625,
|
| 300 |
-
"rewards/thinking_reward_func": 0.0,
|
| 301 |
-
"step": 23
|
| 302 |
-
},
|
| 303 |
-
{
|
| 304 |
-
"clip_ratio": 0.004418392883962952,
|
| 305 |
-
"epoch": 0.0192,
|
| 306 |
-
"grad_norm": 0.44165197354566593,
|
| 307 |
-
"kl": 0.003177642822265625,
|
| 308 |
-
"learning_rate": 5e-07,
|
| 309 |
-
"loss": 0.0922,
|
| 310 |
-
"step": 24
|
| 311 |
-
},
|
| 312 |
-
{
|
| 313 |
-
"clip_ratio": 0.0,
|
| 314 |
-
"completion_length": 2827.5234375,
|
| 315 |
-
"epoch": 0.02,
|
| 316 |
-
"grad_norm": 0.4898734505712748,
|
| 317 |
-
"kl": 0.003976345062255859,
|
| 318 |
-
"learning_rate": 5e-07,
|
| 319 |
-
"loss": 0.0859,
|
| 320 |
-
"reward": 0.29980469786096364,
|
| 321 |
-
"reward_std": 0.4965838538482785,
|
| 322 |
-
"rewards/end_of_conversation_reward_func": 0.05078125186264515,
|
| 323 |
-
"rewards/end_rm_reward_func": 0.2919921875,
|
| 324 |
-
"rewards/length_reward_func": -0.04296875,
|
| 325 |
-
"rewards/thinking_reward_func": 0.0,
|
| 326 |
-
"step": 25
|
| 327 |
-
},
|
| 328 |
-
{
|
| 329 |
-
"clip_ratio": 0.004215578097500838,
|
| 330 |
-
"epoch": 0.0208,
|
| 331 |
-
"grad_norm": 0.4604932294168012,
|
| 332 |
-
"kl": 0.0045299530029296875,
|
| 333 |
-
"learning_rate": 5e-07,
|
| 334 |
-
"loss": 0.0857,
|
| 335 |
-
"step": 26
|
| 336 |
-
},
|
| 337 |
-
{
|
| 338 |
-
"clip_ratio": 0.0,
|
| 339 |
-
"completion_length": 2700.25,
|
| 340 |
-
"epoch": 0.0216,
|
| 341 |
-
"grad_norm": 0.5569547586903592,
|
| 342 |
-
"kl": 0.006651878356933594,
|
| 343 |
-
"learning_rate": 5e-07,
|
| 344 |
-
"loss": 0.0628,
|
| 345 |
-
"reward": 0.21445313654839993,
|
| 346 |
-
"reward_std": 0.4763609850779176,
|
| 347 |
-
"rewards/end_of_conversation_reward_func": 0.048437501303851604,
|
| 348 |
-
"rewards/end_rm_reward_func": 0.224609375,
|
| 349 |
-
"rewards/length_reward_func": -0.05859375,
|
| 350 |
-
"rewards/thinking_reward_func": 0.0,
|
| 351 |
-
"step": 27
|
| 352 |
-
},
|
| 353 |
-
{
|
| 354 |
-
"clip_ratio": 0.00526866911968682,
|
| 355 |
-
"epoch": 0.0224,
|
| 356 |
-
"grad_norm": 0.5363885121679981,
|
| 357 |
-
"kl": 0.008578300476074219,
|
| 358 |
-
"learning_rate": 5e-07,
|
| 359 |
-
"loss": 0.0626,
|
| 360 |
-
"step": 28
|
| 361 |
-
},
|
| 362 |
-
{
|
| 363 |
-
"clip_ratio": 0.0,
|
| 364 |
-
"completion_length": 3302.2109375,
|
| 365 |
-
"epoch": 0.0232,
|
| 366 |
-
"grad_norm": 0.461624948266616,
|
| 367 |
-
"kl": 0.18498802185058594,
|
| 368 |
-
"learning_rate": 5e-07,
|
| 369 |
-
"loss": 0.1148,
|
| 370 |
-
"reward": 0.18046875682193786,
|
| 371 |
-
"reward_std": 0.3992706071585417,
|
| 372 |
-
"rewards/end_of_conversation_reward_func": 0.043750000826548785,
|
| 373 |
-
"rewards/end_rm_reward_func": 0.21875,
|
| 374 |
-
"rewards/length_reward_func": -0.08203125,
|
| 375 |
-
"rewards/thinking_reward_func": 0.0,
|
| 376 |
-
"step": 29
|
| 377 |
-
},
|
| 378 |
-
{
|
| 379 |
-
"clip_ratio": 0.004589079937431961,
|
| 380 |
-
"epoch": 0.024,
|
| 381 |
-
"grad_norm": 0.47299298250006305,
|
| 382 |
-
"kl": 0.14548110961914062,
|
| 383 |
-
"learning_rate": 5e-07,
|
| 384 |
-
"loss": 0.1149,
|
| 385 |
-
"step": 30
|
| 386 |
-
},
|
| 387 |
-
{
|
| 388 |
-
"clip_ratio": 0.0,
|
| 389 |
-
"completion_length": 3118.6015625,
|
| 390 |
-
"epoch": 0.0248,
|
| 391 |
-
"grad_norm": 0.42713160024118774,
|
| 392 |
-
"kl": 0.008479118347167969,
|
| 393 |
-
"learning_rate": 5e-07,
|
| 394 |
-
"loss": 0.1257,
|
| 395 |
-
"reward": 0.34414063580334187,
|
| 396 |
-
"reward_std": 0.5011807959526777,
|
| 397 |
-
"rewards/end_of_conversation_reward_func": 0.05312500102445483,
|
| 398 |
-
"rewards/end_rm_reward_func": 0.361328125,
|
| 399 |
-
"rewards/length_reward_func": -0.0703125,
|
| 400 |
-
"rewards/thinking_reward_func": 0.0,
|
| 401 |
-
"step": 31
|
| 402 |
-
},
|
| 403 |
-
{
|
| 404 |
-
"clip_ratio": 0.004479929368244484,
|
| 405 |
-
"epoch": 0.0256,
|
| 406 |
-
"grad_norm": 0.44373075948513674,
|
| 407 |
-
"kl": 0.010387420654296875,
|
| 408 |
-
"learning_rate": 5e-07,
|
| 409 |
-
"loss": 0.1258,
|
| 410 |
-
"step": 32
|
| 411 |
-
},
|
| 412 |
-
{
|
| 413 |
-
"clip_ratio": 0.0,
|
| 414 |
-
"completion_length": 3374.3125,
|
| 415 |
-
"epoch": 0.0264,
|
| 416 |
-
"grad_norm": 0.4853755980068411,
|
| 417 |
-
"kl": 0.010019302368164062,
|
| 418 |
-
"learning_rate": 5e-07,
|
| 419 |
-
"loss": 0.0936,
|
| 420 |
-
"reward": 0.18574219779111445,
|
| 421 |
-
"reward_std": 0.39575440785847604,
|
| 422 |
-
"rewards/end_of_conversation_reward_func": 0.05000000144354999,
|
| 423 |
-
"rewards/end_rm_reward_func": 0.2138671875,
|
| 424 |
-
"rewards/length_reward_func": -0.078125,
|
| 425 |
-
"rewards/thinking_reward_func": 0.0,
|
| 426 |
-
"step": 33
|
| 427 |
-
},
|
| 428 |
-
{
|
| 429 |
-
"clip_ratio": 0.0049804064619820565,
|
| 430 |
-
"epoch": 0.0272,
|
| 431 |
-
"grad_norm": 0.48990044698626367,
|
| 432 |
-
"kl": 0.010984420776367188,
|
| 433 |
-
"learning_rate": 5e-07,
|
| 434 |
-
"loss": 0.0935,
|
| 435 |
-
"step": 34
|
| 436 |
-
},
|
| 437 |
-
{
|
| 438 |
-
"clip_ratio": 0.0,
|
| 439 |
-
"completion_length": 3179.3671875,
|
| 440 |
-
"epoch": 0.028,
|
| 441 |
-
"grad_norm": 0.5301584639571147,
|
| 442 |
-
"kl": 0.016666412353515625,
|
| 443 |
-
"learning_rate": 5e-07,
|
| 444 |
-
"loss": 0.0573,
|
| 445 |
-
"reward": 0.20410157088190317,
|
| 446 |
-
"reward_std": 0.5211825617589056,
|
| 447 |
-
"rewards/end_of_conversation_reward_func": 0.04687500069849193,
|
| 448 |
-
"rewards/end_rm_reward_func": 0.2197265625,
|
| 449 |
-
"rewards/length_reward_func": -0.0625,
|
| 450 |
-
"rewards/thinking_reward_func": 0.0,
|
| 451 |
-
"step": 35
|
| 452 |
-
},
|
| 453 |
-
{
|
| 454 |
-
"clip_ratio": 0.004679829871747643,
|
| 455 |
-
"epoch": 0.0288,
|
| 456 |
-
"grad_norm": 0.5145033225785893,
|
| 457 |
-
"kl": 0.021465301513671875,
|
| 458 |
-
"learning_rate": 5e-07,
|
| 459 |
-
"loss": 0.0571,
|
| 460 |
-
"step": 36
|
| 461 |
-
},
|
| 462 |
-
{
|
| 463 |
-
"clip_ratio": 0.0,
|
| 464 |
-
"completion_length": 3041.6875,
|
| 465 |
-
"epoch": 0.0296,
|
| 466 |
-
"grad_norm": 0.4531386948890339,
|
| 467 |
-
"kl": 0.014142990112304688,
|
| 468 |
-
"learning_rate": 5e-07,
|
| 469 |
-
"loss": 0.061,
|
| 470 |
-
"reward": 0.17304688156582415,
|
| 471 |
-
"reward_std": 0.42873420752584934,
|
| 472 |
-
"rewards/end_of_conversation_reward_func": 0.04609375062864274,
|
| 473 |
-
"rewards/end_rm_reward_func": 0.181640625,
|
| 474 |
-
"rewards/length_reward_func": -0.0546875,
|
| 475 |
-
"rewards/thinking_reward_func": 0.0,
|
| 476 |
-
"step": 37
|
| 477 |
-
},
|
| 478 |
-
{
|
| 479 |
-
"clip_ratio": 0.004625552362995222,
|
| 480 |
-
"epoch": 0.0304,
|
| 481 |
-
"grad_norm": 0.4389212247837138,
|
| 482 |
-
"kl": 0.01320648193359375,
|
| 483 |
-
"learning_rate": 5e-07,
|
| 484 |
-
"loss": 0.0608,
|
| 485 |
-
"step": 38
|
| 486 |
-
},
|
| 487 |
-
{
|
| 488 |
-
"clip_ratio": 0.0,
|
| 489 |
-
"completion_length": 2900.984375,
|
| 490 |
-
"epoch": 0.0312,
|
| 491 |
-
"grad_norm": 0.513108187473543,
|
| 492 |
-
"kl": 0.0476226806640625,
|
| 493 |
-
"learning_rate": 5e-07,
|
| 494 |
-
"loss": 0.0364,
|
| 495 |
-
"reward": 0.26054688123986125,
|
| 496 |
-
"reward_std": 0.4113442550878972,
|
| 497 |
-
"rewards/end_of_conversation_reward_func": 0.0515625016996637,
|
| 498 |
-
"rewards/end_rm_reward_func": 0.244140625,
|
| 499 |
-
"rewards/length_reward_func": -0.03515625,
|
| 500 |
-
"rewards/thinking_reward_func": 0.0,
|
| 501 |
-
"step": 39
|
| 502 |
-
},
|
| 503 |
-
{
|
| 504 |
-
"clip_ratio": 0.005045634257839993,
|
| 505 |
-
"epoch": 0.032,
|
| 506 |
-
"grad_norm": 0.4960768449341314,
|
| 507 |
-
"kl": 0.023647308349609375,
|
| 508 |
-
"learning_rate": 5e-07,
|
| 509 |
-
"loss": 0.0363,
|
| 510 |
-
"step": 40
|
| 511 |
-
},
|
| 512 |
-
{
|
| 513 |
-
"clip_ratio": 0.0,
|
| 514 |
-
"completion_length": 3094.5390625,
|
| 515 |
-
"epoch": 0.0328,
|
| 516 |
-
"grad_norm": 0.4111608127743996,
|
| 517 |
-
"kl": 0.023374557495117188,
|
| 518 |
-
"learning_rate": 5e-07,
|
| 519 |
-
"loss": 0.1087,
|
| 520 |
-
"reward": 0.18554687942378223,
|
| 521 |
-
"reward_std": 0.48099791165441275,
|
| 522 |
-
"rewards/end_of_conversation_reward_func": 0.03906250058207661,
|
| 523 |
-
"rewards/end_rm_reward_func": 0.205078125,
|
| 524 |
-
"rewards/length_reward_func": -0.05859375,
|
| 525 |
-
"rewards/thinking_reward_func": 0.0,
|
| 526 |
-
"step": 41
|
| 527 |
-
},
|
| 528 |
-
{
|
| 529 |
-
"clip_ratio": 0.004160198179306462,
|
| 530 |
-
"epoch": 0.0336,
|
| 531 |
-
"grad_norm": 0.4050446568608628,
|
| 532 |
-
"kl": 0.020130157470703125,
|
| 533 |
-
"learning_rate": 5e-07,
|
| 534 |
-
"loss": 0.1086,
|
| 535 |
-
"step": 42
|
| 536 |
-
},
|
| 537 |
-
{
|
| 538 |
-
"clip_ratio": 0.0,
|
| 539 |
-
"completion_length": 2732.453125,
|
| 540 |
-
"epoch": 0.0344,
|
| 541 |
-
"grad_norm": 0.5932767606982138,
|
| 542 |
-
"kl": 0.07580184936523438,
|
| 543 |
-
"learning_rate": 5e-07,
|
| 544 |
-
"loss": 0.0115,
|
| 545 |
-
"reward": 0.10800781659781933,
|
| 546 |
-
"reward_std": 0.4096634048037231,
|
| 547 |
-
"rewards/end_of_conversation_reward_func": 0.052343750663567334,
|
| 548 |
-
"rewards/end_rm_reward_func": 0.1103515625,
|
| 549 |
-
"rewards/length_reward_func": -0.0546875,
|
| 550 |
-
"rewards/thinking_reward_func": 0.0,
|
| 551 |
-
"step": 43
|
| 552 |
-
},
|
| 553 |
-
{
|
| 554 |
-
"clip_ratio": 0.005395003347075544,
|
| 555 |
-
"epoch": 0.0352,
|
| 556 |
-
"grad_norm": 0.562119953077685,
|
| 557 |
-
"kl": 0.1378173828125,
|
| 558 |
-
"learning_rate": 5e-07,
|
| 559 |
-
"loss": 0.0112,
|
| 560 |
-
"step": 44
|
| 561 |
-
},
|
| 562 |
-
{
|
| 563 |
-
"clip_ratio": 0.0,
|
| 564 |
-
"completion_length": 3149.03125,
|
| 565 |
-
"epoch": 0.036,
|
| 566 |
-
"grad_norm": 0.43738882704141613,
|
| 567 |
-
"kl": 0.07685470581054688,
|
| 568 |
-
"learning_rate": 5e-07,
|
| 569 |
-
"loss": 0.0961,
|
| 570 |
-
"reward": 0.18593750661239028,
|
| 571 |
-
"reward_std": 0.459837670205161,
|
| 572 |
-
"rewards/end_of_conversation_reward_func": 0.04140625079162419,
|
| 573 |
-
"rewards/end_rm_reward_func": 0.2109375,
|
| 574 |
-
"rewards/length_reward_func": -0.06640625,
|
| 575 |
-
"rewards/thinking_reward_func": 0.0,
|
| 576 |
-
"step": 45
|
| 577 |
-
},
|
| 578 |
-
{
|
| 579 |
-
"clip_ratio": 0.004886353664915077,
|
| 580 |
-
"epoch": 0.0368,
|
| 581 |
-
"grad_norm": 0.4246516268286063,
|
| 582 |
-
"kl": 0.07813262939453125,
|
| 583 |
-
"learning_rate": 5e-07,
|
| 584 |
-
"loss": 0.0961,
|
| 585 |
-
"step": 46
|
| 586 |
-
},
|
| 587 |
-
{
|
| 588 |
-
"clip_ratio": 0.0,
|
| 589 |
-
"completion_length": 2791.0625,
|
| 590 |
-
"epoch": 0.0376,
|
| 591 |
-
"grad_norm": 0.4757940646494544,
|
| 592 |
-
"kl": 0.041767120361328125,
|
| 593 |
-
"learning_rate": 5e-07,
|
| 594 |
-
"loss": 0.0631,
|
| 595 |
-
"reward": 0.2837890684604645,
|
| 596 |
-
"reward_std": 0.4126680698245764,
|
| 597 |
-
"rewards/end_of_conversation_reward_func": 0.056250001303851604,
|
| 598 |
-
"rewards/end_rm_reward_func": 0.2626953125,
|
| 599 |
-
"rewards/length_reward_func": -0.03515625,
|
| 600 |
-
"rewards/thinking_reward_func": 0.0,
|
| 601 |
-
"step": 47
|
| 602 |
-
},
|
| 603 |
-
{
|
| 604 |
-
"clip_ratio": 0.005135874525876716,
|
| 605 |
-
"epoch": 0.0384,
|
| 606 |
-
"grad_norm": 0.47547921032378754,
|
| 607 |
-
"kl": 0.09936141967773438,
|
| 608 |
-
"learning_rate": 5e-07,
|
| 609 |
-
"loss": 0.063,
|
| 610 |
-
"step": 48
|
| 611 |
-
},
|
| 612 |
-
{
|
| 613 |
-
"clip_ratio": 0.0,
|
| 614 |
-
"completion_length": 2754.609375,
|
| 615 |
-
"epoch": 0.0392,
|
| 616 |
-
"grad_norm": 0.49754042199838183,
|
| 617 |
-
"kl": 0.12404251098632812,
|
| 618 |
-
"learning_rate": 5e-07,
|
| 619 |
-
"loss": 0.0849,
|
| 620 |
-
"reward": 0.32207031920552254,
|
| 621 |
-
"reward_std": 0.46512654330581427,
|
| 622 |
-
"rewards/end_of_conversation_reward_func": 0.051562500884756446,
|
| 623 |
-
"rewards/end_rm_reward_func": 0.3017578125,
|
| 624 |
-
"rewards/length_reward_func": -0.03125,
|
| 625 |
-
"rewards/thinking_reward_func": 0.0,
|
| 626 |
-
"step": 49
|
| 627 |
-
},
|
| 628 |
-
{
|
| 629 |
-
"clip_ratio": 0.004882045017438941,
|
| 630 |
-
"epoch": 0.04,
|
| 631 |
-
"grad_norm": 0.4817430120792817,
|
| 632 |
-
"kl": 0.11900711059570312,
|
| 633 |
-
"learning_rate": 5e-07,
|
| 634 |
-
"loss": 0.0847,
|
| 635 |
-
"step": 50
|
| 636 |
-
},
|
| 637 |
-
{
|
| 638 |
-
"clip_ratio": 0.0,
|
| 639 |
-
"completion_length": 2970.8984375,
|
| 640 |
-
"epoch": 0.0408,
|
| 641 |
-
"grad_norm": 0.5071054759904158,
|
| 642 |
-
"kl": 0.057147979736328125,
|
| 643 |
-
"learning_rate": 5e-07,
|
| 644 |
-
"loss": 0.1121,
|
| 645 |
-
"reward": 0.2646484528668225,
|
| 646 |
-
"reward_std": 0.37708853278309107,
|
| 647 |
-
"rewards/end_of_conversation_reward_func": 0.05468750081490725,
|
| 648 |
-
"rewards/end_rm_reward_func": 0.2607421875,
|
| 649 |
-
"rewards/length_reward_func": -0.05078125,
|
| 650 |
-
"rewards/thinking_reward_func": 0.0,
|
| 651 |
-
"step": 51
|
| 652 |
-
},
|
| 653 |
-
{
|
| 654 |
-
"clip_ratio": 0.004119776640436612,
|
| 655 |
-
"epoch": 0.0416,
|
| 656 |
-
"grad_norm": 0.4894389235389238,
|
| 657 |
-
"kl": 0.05866241455078125,
|
| 658 |
-
"learning_rate": 5e-07,
|
| 659 |
-
"loss": 0.112,
|
| 660 |
-
"step": 52
|
| 661 |
-
},
|
| 662 |
-
{
|
| 663 |
-
"clip_ratio": 0.0,
|
| 664 |
-
"completion_length": 2454.84375,
|
| 665 |
-
"epoch": 0.0424,
|
| 666 |
-
"grad_norm": 0.6397704779379643,
|
| 667 |
-
"kl": 0.10573577880859375,
|
| 668 |
-
"learning_rate": 5e-07,
|
| 669 |
-
"loss": 0.0424,
|
| 670 |
-
"reward": 0.29101564083248377,
|
| 671 |
-
"reward_std": 0.3862592806108296,
|
| 672 |
-
"rewards/end_of_conversation_reward_func": 0.058593750989530236,
|
| 673 |
-
"rewards/end_rm_reward_func": 0.271484375,
|
| 674 |
-
"rewards/length_reward_func": -0.0390625,
|
| 675 |
-
"rewards/thinking_reward_func": 0.0,
|
| 676 |
-
"step": 53
|
| 677 |
-
},
|
| 678 |
-
{
|
| 679 |
-
"clip_ratio": 0.005806522516650148,
|
| 680 |
-
"epoch": 0.0432,
|
| 681 |
-
"grad_norm": 0.5935861300569137,
|
| 682 |
-
"kl": 0.22344207763671875,
|
| 683 |
-
"learning_rate": 5e-07,
|
| 684 |
-
"loss": 0.0421,
|
| 685 |
-
"step": 54
|
| 686 |
-
},
|
| 687 |
-
{
|
| 688 |
-
"clip_ratio": 0.0,
|
| 689 |
-
"completion_length": 2823.3515625,
|
| 690 |
-
"epoch": 0.044,
|
| 691 |
-
"grad_norm": 0.5249791154692802,
|
| 692 |
-
"kl": 0.09177398681640625,
|
| 693 |
-
"learning_rate": 5e-07,
|
| 694 |
-
"loss": 0.0814,
|
| 695 |
-
"reward": 0.26269532449077815,
|
| 696 |
-
"reward_std": 0.40016138833016157,
|
| 697 |
-
"rewards/end_of_conversation_reward_func": 0.05078125122236088,
|
| 698 |
-
"rewards/end_rm_reward_func": 0.2626953125,
|
| 699 |
-
"rewards/length_reward_func": -0.05078125,
|
| 700 |
-
"rewards/thinking_reward_func": 0.0,
|
| 701 |
-
"step": 55
|
| 702 |
-
},
|
| 703 |
-
{
|
| 704 |
-
"clip_ratio": 0.00533542905759532,
|
| 705 |
-
"epoch": 0.0448,
|
| 706 |
-
"grad_norm": 0.49819488661627054,
|
| 707 |
-
"kl": 0.0757293701171875,
|
| 708 |
-
"learning_rate": 5e-07,
|
| 709 |
-
"loss": 0.0811,
|
| 710 |
-
"step": 56
|
| 711 |
-
},
|
| 712 |
-
{
|
| 713 |
-
"clip_ratio": 0.0,
|
| 714 |
-
"completion_length": 2891.5859375,
|
| 715 |
-
"epoch": 0.0456,
|
| 716 |
-
"grad_norm": 0.5318285198540439,
|
| 717 |
-
"kl": 0.3561859130859375,
|
| 718 |
-
"learning_rate": 5e-07,
|
| 719 |
-
"loss": 0.0162,
|
| 720 |
-
"reward": 0.3619140787050128,
|
| 721 |
-
"reward_std": 0.45814486034214497,
|
| 722 |
-
"rewards/end_of_conversation_reward_func": 0.056250000721774995,
|
| 723 |
-
"rewards/end_rm_reward_func": 0.3486328125,
|
| 724 |
-
"rewards/length_reward_func": -0.04296875,
|
| 725 |
-
"rewards/thinking_reward_func": 0.0,
|
| 726 |
-
"step": 57
|
| 727 |
-
},
|
| 728 |
-
{
|
| 729 |
-
"clip_ratio": 0.005233535193838179,
|
| 730 |
-
"epoch": 0.0464,
|
| 731 |
-
"grad_norm": 0.5029588635123279,
|
| 732 |
-
"kl": 0.7606353759765625,
|
| 733 |
-
"learning_rate": 5e-07,
|
| 734 |
-
"loss": 0.0161,
|
| 735 |
-
"step": 58
|
| 736 |
-
},
|
| 737 |
-
{
|
| 738 |
-
"clip_ratio": 0.0,
|
| 739 |
-
"completion_length": 3584.8203125,
|
| 740 |
-
"epoch": 0.0472,
|
| 741 |
-
"grad_norm": 0.4934060163626245,
|
| 742 |
-
"kl": 0.31105804443359375,
|
| 743 |
-
"learning_rate": 5e-07,
|
| 744 |
-
"loss": 0.0458,
|
| 745 |
-
"reward": 0.20800781692378223,
|
| 746 |
-
"reward_std": 0.45996399596333504,
|
| 747 |
-
"rewards/end_of_conversation_reward_func": 0.03906250069849193,
|
| 748 |
-
"rewards/end_rm_reward_func": 0.2314453125,
|
| 749 |
-
"rewards/length_reward_func": -0.0625,
|
| 750 |
-
"rewards/thinking_reward_func": 0.0,
|
| 751 |
-
"step": 59
|
| 752 |
-
},
|
| 753 |
-
{
|
| 754 |
-
"clip_ratio": 0.004666702181566507,
|
| 755 |
-
"epoch": 0.048,
|
| 756 |
-
"grad_norm": 0.45620632413031426,
|
| 757 |
-
"kl": 0.1880645751953125,
|
| 758 |
-
"learning_rate": 5e-07,
|
| 759 |
-
"loss": 0.0457,
|
| 760 |
-
"step": 60
|
| 761 |
-
},
|
| 762 |
-
{
|
| 763 |
-
"clip_ratio": 0.0,
|
| 764 |
-
"completion_length": 2510.5703125,
|
| 765 |
-
"epoch": 0.0488,
|
| 766 |
-
"grad_norm": 0.62998979474741,
|
| 767 |
-
"kl": 0.158233642578125,
|
| 768 |
-
"learning_rate": 5e-07,
|
| 769 |
-
"loss": 0.0642,
|
| 770 |
-
"reward": 0.2816406446509063,
|
| 771 |
-
"reward_std": 0.35436337254941463,
|
| 772 |
-
"rewards/end_of_conversation_reward_func": 0.06093750131549314,
|
| 773 |
-
"rewards/end_rm_reward_func": 0.251953125,
|
| 774 |
-
"rewards/length_reward_func": -0.03125,
|
| 775 |
-
"rewards/thinking_reward_func": 0.0,
|
| 776 |
-
"step": 61
|
| 777 |
-
},
|
| 778 |
-
{
|
| 779 |
-
"clip_ratio": 0.006271788995945826,
|
| 780 |
-
"epoch": 0.0496,
|
| 781 |
-
"grad_norm": 0.6009428790886027,
|
| 782 |
-
"kl": 0.271392822265625,
|
| 783 |
-
"learning_rate": 5e-07,
|
| 784 |
-
"loss": 0.0639,
|
| 785 |
-
"step": 62
|
| 786 |
-
},
|
| 787 |
-
{
|
| 788 |
-
"clip_ratio": 0.0,
|
| 789 |
-
"completion_length": 2657.3671875,
|
| 790 |
-
"epoch": 0.0504,
|
| 791 |
-
"grad_norm": 0.6144033583238725,
|
| 792 |
-
"kl": 0.100860595703125,
|
| 793 |
-
"learning_rate": 5e-07,
|
| 794 |
-
"loss": 0.1188,
|
| 795 |
-
"reward": 0.266796879703179,
|
| 796 |
-
"reward_std": 0.39745712792500854,
|
| 797 |
-
"rewards/end_of_conversation_reward_func": 0.04609375091968104,
|
| 798 |
-
"rewards/end_rm_reward_func": 0.255859375,
|
| 799 |
-
"rewards/length_reward_func": -0.03515625,
|
| 800 |
-
"rewards/thinking_reward_func": 0.0,
|
| 801 |
-
"step": 63
|
| 802 |
-
},
|
| 803 |
-
{
|
| 804 |
-
"clip_ratio": 0.005829609275679104,
|
| 805 |
-
"epoch": 0.0512,
|
| 806 |
-
"grad_norm": 0.5803952639816394,
|
| 807 |
-
"kl": 0.1117401123046875,
|
| 808 |
-
"learning_rate": 5e-07,
|
| 809 |
-
"loss": 0.1186,
|
| 810 |
-
"step": 64
|
| 811 |
-
},
|
| 812 |
-
{
|
| 813 |
-
"clip_ratio": 0.0,
|
| 814 |
-
"completion_length": 2555.703125,
|
| 815 |
-
"epoch": 0.052,
|
| 816 |
-
"grad_norm": 0.607729281892395,
|
| 817 |
-
"kl": 1.57684326171875,
|
| 818 |
-
"learning_rate": 5e-07,
|
| 819 |
-
"loss": 0.0828,
|
| 820 |
-
"reward": 0.28125000261934474,
|
| 821 |
-
"reward_std": 0.4093571212142706,
|
| 822 |
-
"rewards/end_of_conversation_reward_func": 0.054687500349245965,
|
| 823 |
-
"rewards/end_rm_reward_func": 0.26171875,
|
| 824 |
-
"rewards/length_reward_func": -0.03515625,
|
| 825 |
-
"rewards/thinking_reward_func": 0.0,
|
| 826 |
-
"step": 65
|
| 827 |
-
},
|
| 828 |
-
{
|
| 829 |
-
"clip_ratio": 0.005639895083731972,
|
| 830 |
-
"epoch": 0.0528,
|
| 831 |
-
"grad_norm": 0.5870178045196847,
|
| 832 |
-
"kl": 3.4336700439453125,
|
| 833 |
-
"learning_rate": 5e-07,
|
| 834 |
-
"loss": 0.0823,
|
| 835 |
-
"step": 66
|
| 836 |
-
},
|
| 837 |
-
{
|
| 838 |
-
"clip_ratio": 0.0,
|
| 839 |
-
"completion_length": 3058.078125,
|
| 840 |
-
"epoch": 0.0536,
|
| 841 |
-
"grad_norm": 0.6392048858380123,
|
| 842 |
-
"kl": 0.2032470703125,
|
| 843 |
-
"learning_rate": 5e-07,
|
| 844 |
-
"loss": 0.0872,
|
| 845 |
-
"reward": 0.2697265713941306,
|
| 846 |
-
"reward_std": 0.46659763529896736,
|
| 847 |
-
"rewards/end_of_conversation_reward_func": 0.05000000074505806,
|
| 848 |
-
"rewards/end_rm_reward_func": 0.2705078125,
|
| 849 |
-
"rewards/length_reward_func": -0.05078125,
|
| 850 |
-
"rewards/thinking_reward_func": 0.0,
|
| 851 |
-
"step": 67
|
| 852 |
-
},
|
| 853 |
-
{
|
| 854 |
-
"clip_ratio": 0.005416913933004253,
|
| 855 |
-
"epoch": 0.0544,
|
| 856 |
-
"grad_norm": 0.5540548636439222,
|
| 857 |
-
"kl": 0.2025146484375,
|
| 858 |
-
"learning_rate": 5e-07,
|
| 859 |
-
"loss": 0.0868,
|
| 860 |
-
"step": 68
|
| 861 |
-
},
|
| 862 |
-
{
|
| 863 |
-
"clip_ratio": 0.0,
|
| 864 |
-
"completion_length": 3101.0234375,
|
| 865 |
-
"epoch": 0.0552,
|
| 866 |
-
"grad_norm": 0.5462471595428263,
|
| 867 |
-
"kl": 0.8668975830078125,
|
| 868 |
-
"learning_rate": 5e-07,
|
| 869 |
-
"loss": 0.1123,
|
| 870 |
-
"reward": 0.13945313333533704,
|
| 871 |
-
"reward_std": 0.4813323845155537,
|
| 872 |
-
"rewards/end_of_conversation_reward_func": 0.0515625006519258,
|
| 873 |
-
"rewards/end_rm_reward_func": 0.134765625,
|
| 874 |
-
"rewards/length_reward_func": -0.046875,
|
| 875 |
-
"rewards/thinking_reward_func": 0.0,
|
| 876 |
-
"step": 69
|
| 877 |
-
},
|
| 878 |
-
{
|
| 879 |
-
"clip_ratio": 0.005168676114408299,
|
| 880 |
-
"epoch": 0.056,
|
| 881 |
-
"grad_norm": 0.528066600936896,
|
| 882 |
-
"kl": 0.857147216796875,
|
| 883 |
-
"learning_rate": 5e-07,
|
| 884 |
-
"loss": 0.1122,
|
| 885 |
-
"step": 70
|
| 886 |
-
},
|
| 887 |
-
{
|
| 888 |
-
"clip_ratio": 0.0,
|
| 889 |
-
"completion_length": 2047.375,
|
| 890 |
-
"epoch": 0.0568,
|
| 891 |
-
"grad_norm": 0.691284897693145,
|
| 892 |
-
"kl": 15.969619750976562,
|
| 893 |
-
"learning_rate": 5e-07,
|
| 894 |
-
"loss": 0.0289,
|
| 895 |
-
"reward": 0.487109393812716,
|
| 896 |
-
"reward_std": 0.4644232392311096,
|
| 897 |
-
"rewards/end_of_conversation_reward_func": 0.07500000158324838,
|
| 898 |
-
"rewards/end_rm_reward_func": 0.427734375,
|
| 899 |
-
"rewards/length_reward_func": -0.015625,
|
| 900 |
-
"rewards/thinking_reward_func": 0.0,
|
| 901 |
-
"step": 71
|
| 902 |
-
},
|
| 903 |
-
{
|
| 904 |
-
"clip_ratio": 0.005814553995151073,
|
| 905 |
-
"epoch": 0.0576,
|
| 906 |
-
"grad_norm": 0.6404202865162691,
|
| 907 |
-
"kl": 30.811477661132812,
|
| 908 |
-
"learning_rate": 5e-07,
|
| 909 |
-
"loss": 0.0288,
|
| 910 |
-
"step": 72
|
| 911 |
-
},
|
| 912 |
-
{
|
| 913 |
-
"clip_ratio": 0.0,
|
| 914 |
-
"completion_length": 2905.2890625,
|
| 915 |
-
"epoch": 0.0584,
|
| 916 |
-
"grad_norm": 0.6445835293449337,
|
| 917 |
-
"kl": 1.801666259765625,
|
| 918 |
-
"learning_rate": 5e-07,
|
| 919 |
-
"loss": 0.0275,
|
| 920 |
-
"reward": 0.15878907404839993,
|
| 921 |
-
"reward_std": 0.3413613960146904,
|
| 922 |
-
"rewards/end_of_conversation_reward_func": 0.05625000089639798,
|
| 923 |
-
"rewards/end_rm_reward_func": 0.1337890625,
|
| 924 |
-
"rewards/length_reward_func": -0.03125,
|
| 925 |
-
"rewards/thinking_reward_func": 0.0,
|
| 926 |
-
"step": 73
|
| 927 |
-
},
|
| 928 |
-
{
|
| 929 |
-
"clip_ratio": 0.004967361441231333,
|
| 930 |
-
"epoch": 0.0592,
|
| 931 |
-
"grad_norm": 0.5958211572085673,
|
| 932 |
-
"kl": 1.816192626953125,
|
| 933 |
-
"learning_rate": 5e-07,
|
| 934 |
-
"loss": 0.027,
|
| 935 |
-
"step": 74
|
| 936 |
-
},
|
| 937 |
-
{
|
| 938 |
-
"clip_ratio": 0.0,
|
| 939 |
-
"completion_length": 3157.5546875,
|
| 940 |
-
"epoch": 0.06,
|
| 941 |
-
"grad_norm": 0.5787729026187358,
|
| 942 |
-
"kl": 1.14288330078125,
|
| 943 |
-
"learning_rate": 5e-07,
|
| 944 |
-
"loss": 0.0495,
|
| 945 |
-
"reward": 0.2250000073108822,
|
| 946 |
-
"reward_std": 0.38128331024199724,
|
| 947 |
-
"rewards/end_of_conversation_reward_func": 0.045312500617001206,
|
| 948 |
-
"rewards/end_rm_reward_func": 0.234375,
|
| 949 |
-
"rewards/length_reward_func": -0.0546875,
|
| 950 |
-
"rewards/thinking_reward_func": 0.0,
|
| 951 |
-
"step": 75
|
| 952 |
-
},
|
| 953 |
-
{
|
| 954 |
-
"clip_ratio": 0.0048050573241198435,
|
| 955 |
-
"epoch": 0.0608,
|
| 956 |
-
"grad_norm": 0.5568993189176473,
|
| 957 |
-
"kl": 0.77630615234375,
|
| 958 |
-
"learning_rate": 5e-07,
|
| 959 |
-
"loss": 0.0494,
|
| 960 |
-
"step": 76
|
| 961 |
-
},
|
| 962 |
-
{
|
| 963 |
-
"clip_ratio": 0.0,
|
| 964 |
-
"completion_length": 2372.6640625,
|
| 965 |
-
"epoch": 0.0616,
|
| 966 |
-
"grad_norm": 0.6173163766112336,
|
| 967 |
-
"kl": 0.433197021484375,
|
| 968 |
-
"learning_rate": 5e-07,
|
| 969 |
-
"loss": 0.085,
|
| 970 |
-
"reward": 0.21562500204890966,
|
| 971 |
-
"reward_std": 0.35906139435246587,
|
| 972 |
-
"rewards/end_of_conversation_reward_func": 0.055468750826548785,
|
| 973 |
-
"rewards/end_rm_reward_func": 0.1796875,
|
| 974 |
-
"rewards/length_reward_func": -0.01953125,
|
| 975 |
-
"rewards/thinking_reward_func": 0.0,
|
| 976 |
-
"step": 77
|
| 977 |
-
},
|
| 978 |
-
{
|
| 979 |
-
"clip_ratio": 0.005464794070576318,
|
| 980 |
-
"epoch": 0.0624,
|
| 981 |
-
"grad_norm": 0.5907021152198737,
|
| 982 |
-
"kl": 1.02618408203125,
|
| 983 |
-
"learning_rate": 5e-07,
|
| 984 |
-
"loss": 0.0846,
|
| 985 |
-
"step": 78
|
| 986 |
-
},
|
| 987 |
-
{
|
| 988 |
-
"clip_ratio": 0.0,
|
| 989 |
-
"completion_length": 2704.46875,
|
| 990 |
-
"epoch": 0.0632,
|
| 991 |
-
"grad_norm": 0.7611829534310949,
|
| 992 |
-
"kl": 3.341705322265625,
|
| 993 |
-
"learning_rate": 5e-07,
|
| 994 |
-
"loss": 0.0493,
|
| 995 |
-
"reward": 0.23417970072478056,
|
| 996 |
-
"reward_std": 0.36300004506483674,
|
| 997 |
-
"rewards/end_of_conversation_reward_func": 0.06328125088475645,
|
| 998 |
-
"rewards/end_rm_reward_func": 0.2255859375,
|
| 999 |
-
"rewards/length_reward_func": -0.0546875,
|
| 1000 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1001 |
-
"step": 79
|
| 1002 |
-
},
|
| 1003 |
-
{
|
| 1004 |
-
"clip_ratio": 0.005534523181268014,
|
| 1005 |
-
"epoch": 0.064,
|
| 1006 |
-
"grad_norm": 0.6276167653414357,
|
| 1007 |
-
"kl": 1.77569580078125,
|
| 1008 |
-
"learning_rate": 5e-07,
|
| 1009 |
-
"loss": 0.0488,
|
| 1010 |
-
"step": 80
|
| 1011 |
-
},
|
| 1012 |
-
{
|
| 1013 |
-
"clip_ratio": 0.0,
|
| 1014 |
-
"completion_length": 2530.2265625,
|
| 1015 |
-
"epoch": 0.0648,
|
| 1016 |
-
"grad_norm": 0.8681960156746225,
|
| 1017 |
-
"kl": 1.415252685546875,
|
| 1018 |
-
"learning_rate": 5e-07,
|
| 1019 |
-
"loss": 0.0355,
|
| 1020 |
-
"reward": 0.31796875852160156,
|
| 1021 |
-
"reward_std": 0.4489464135840535,
|
| 1022 |
-
"rewards/end_of_conversation_reward_func": 0.060156250721774995,
|
| 1023 |
-
"rewards/end_rm_reward_func": 0.28515625,
|
| 1024 |
-
"rewards/length_reward_func": -0.02734375,
|
| 1025 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1026 |
-
"step": 81
|
| 1027 |
-
},
|
| 1028 |
-
{
|
| 1029 |
-
"clip_ratio": 0.006155758979730308,
|
| 1030 |
-
"epoch": 0.0656,
|
| 1031 |
-
"grad_norm": 0.619758928092397,
|
| 1032 |
-
"kl": 3.17950439453125,
|
| 1033 |
-
"learning_rate": 5e-07,
|
| 1034 |
-
"loss": 0.035,
|
| 1035 |
-
"step": 82
|
| 1036 |
-
},
|
| 1037 |
-
{
|
| 1038 |
-
"clip_ratio": 0.0,
|
| 1039 |
-
"completion_length": 2703.9140625,
|
| 1040 |
-
"epoch": 0.0664,
|
| 1041 |
-
"grad_norm": 0.6148471519549782,
|
| 1042 |
-
"kl": 3.2836151123046875,
|
| 1043 |
-
"learning_rate": 5e-07,
|
| 1044 |
-
"loss": 0.1116,
|
| 1045 |
-
"reward": 0.2703125097323209,
|
| 1046 |
-
"reward_std": 0.4398716832511127,
|
| 1047 |
-
"rewards/end_of_conversation_reward_func": 0.05937500071013346,
|
| 1048 |
-
"rewards/end_rm_reward_func": 0.24609375,
|
| 1049 |
-
"rewards/length_reward_func": -0.03515625,
|
| 1050 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1051 |
-
"step": 83
|
| 1052 |
-
},
|
| 1053 |
-
{
|
| 1054 |
-
"clip_ratio": 0.005438056687125936,
|
| 1055 |
-
"epoch": 0.0672,
|
| 1056 |
-
"grad_norm": 0.6073114120206211,
|
| 1057 |
-
"kl": 2.31671142578125,
|
| 1058 |
-
"learning_rate": 5e-07,
|
| 1059 |
-
"loss": 0.1117,
|
| 1060 |
-
"step": 84
|
| 1061 |
-
},
|
| 1062 |
-
{
|
| 1063 |
-
"clip_ratio": 0.0,
|
| 1064 |
-
"completion_length": 2557.8515625,
|
| 1065 |
-
"epoch": 0.068,
|
| 1066 |
-
"grad_norm": 0.6280176622559991,
|
| 1067 |
-
"kl": 2.861480712890625,
|
| 1068 |
-
"learning_rate": 5e-07,
|
| 1069 |
-
"loss": 0.072,
|
| 1070 |
-
"reward": 0.22968750726431608,
|
| 1071 |
-
"reward_std": 0.3977765201125294,
|
| 1072 |
-
"rewards/end_of_conversation_reward_func": 0.05781250086147338,
|
| 1073 |
-
"rewards/end_rm_reward_func": 0.19921875,
|
| 1074 |
-
"rewards/length_reward_func": -0.02734375,
|
| 1075 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1076 |
-
"step": 85
|
| 1077 |
-
},
|
| 1078 |
-
{
|
| 1079 |
-
"clip_ratio": 0.00582106395449955,
|
| 1080 |
-
"epoch": 0.0688,
|
| 1081 |
-
"grad_norm": 0.5717462929165856,
|
| 1082 |
-
"kl": 4.08203125,
|
| 1083 |
-
"learning_rate": 5e-07,
|
| 1084 |
-
"loss": 0.0716,
|
| 1085 |
-
"step": 86
|
| 1086 |
-
},
|
| 1087 |
-
{
|
| 1088 |
-
"clip_ratio": 0.0,
|
| 1089 |
-
"completion_length": 2243.3359375,
|
| 1090 |
-
"epoch": 0.0696,
|
| 1091 |
-
"grad_norm": 183.6986339811298,
|
| 1092 |
-
"kl": 88.17166137695312,
|
| 1093 |
-
"learning_rate": 5e-07,
|
| 1094 |
-
"loss": 0.1291,
|
| 1095 |
-
"reward": 0.3546875163447112,
|
| 1096 |
-
"reward_std": 0.4497223272919655,
|
| 1097 |
-
"rewards/end_of_conversation_reward_func": 0.06562500132713467,
|
| 1098 |
-
"rewards/end_rm_reward_func": 0.3203125,
|
| 1099 |
-
"rewards/length_reward_func": -0.03125,
|
| 1100 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1101 |
-
"step": 87
|
| 1102 |
-
},
|
| 1103 |
-
{
|
| 1104 |
-
"clip_ratio": 0.00647925166413188,
|
| 1105 |
-
"epoch": 0.0704,
|
| 1106 |
-
"grad_norm": 20.29826374460904,
|
| 1107 |
-
"kl": 29.604248046875,
|
| 1108 |
-
"learning_rate": 5e-07,
|
| 1109 |
-
"loss": 0.0455,
|
| 1110 |
-
"step": 88
|
| 1111 |
-
},
|
| 1112 |
-
{
|
| 1113 |
-
"clip_ratio": 0.0,
|
| 1114 |
-
"completion_length": 2714.2265625,
|
| 1115 |
-
"epoch": 0.0712,
|
| 1116 |
-
"grad_norm": 0.5834283983979679,
|
| 1117 |
-
"kl": 4.820281982421875,
|
| 1118 |
-
"learning_rate": 5e-07,
|
| 1119 |
-
"loss": 0.0376,
|
| 1120 |
-
"reward": 0.3076171970460564,
|
| 1121 |
-
"reward_std": 0.40063817566260695,
|
| 1122 |
-
"rewards/end_of_conversation_reward_func": 0.06250000093132257,
|
| 1123 |
-
"rewards/end_rm_reward_func": 0.2802734375,
|
| 1124 |
-
"rewards/length_reward_func": -0.03515625,
|
| 1125 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1126 |
-
"step": 89
|
| 1127 |
-
},
|
| 1128 |
-
{
|
| 1129 |
-
"clip_ratio": 0.005377560810302384,
|
| 1130 |
-
"epoch": 0.072,
|
| 1131 |
-
"grad_norm": 0.5713867955563853,
|
| 1132 |
-
"kl": 4.233428955078125,
|
| 1133 |
-
"learning_rate": 5e-07,
|
| 1134 |
-
"loss": 0.0374,
|
| 1135 |
-
"step": 90
|
| 1136 |
-
},
|
| 1137 |
-
{
|
| 1138 |
-
"clip_ratio": 0.0,
|
| 1139 |
-
"completion_length": 2687.90625,
|
| 1140 |
-
"epoch": 0.0728,
|
| 1141 |
-
"grad_norm": 0.6405648178568916,
|
| 1142 |
-
"kl": 11.98095703125,
|
| 1143 |
-
"learning_rate": 5e-07,
|
| 1144 |
-
"loss": 0.0957,
|
| 1145 |
-
"reward": 0.18671875854488462,
|
| 1146 |
-
"reward_std": 0.38381645642220974,
|
| 1147 |
-
"rewards/end_of_conversation_reward_func": 0.053906251094304025,
|
| 1148 |
-
"rewards/end_rm_reward_func": 0.1640625,
|
| 1149 |
-
"rewards/length_reward_func": -0.03125,
|
| 1150 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1151 |
-
"step": 91
|
| 1152 |
-
},
|
| 1153 |
-
{
|
| 1154 |
-
"clip_ratio": 0.005748806172050536,
|
| 1155 |
-
"epoch": 0.0736,
|
| 1156 |
-
"grad_norm": 0.6406292246920093,
|
| 1157 |
-
"kl": 20.21795654296875,
|
| 1158 |
-
"learning_rate": 5e-07,
|
| 1159 |
-
"loss": 0.0954,
|
| 1160 |
-
"step": 92
|
| 1161 |
-
},
|
| 1162 |
-
{
|
| 1163 |
-
"clip_ratio": 0.0,
|
| 1164 |
-
"completion_length": 2914.953125,
|
| 1165 |
-
"epoch": 0.0744,
|
| 1166 |
-
"grad_norm": 0.5367785008596705,
|
| 1167 |
-
"kl": 4.4202880859375,
|
| 1168 |
-
"learning_rate": 5e-07,
|
| 1169 |
-
"loss": 0.0375,
|
| 1170 |
-
"reward": 0.22890625800937414,
|
| 1171 |
-
"reward_std": 0.4514218312688172,
|
| 1172 |
-
"rewards/end_of_conversation_reward_func": 0.053125001140870154,
|
| 1173 |
-
"rewards/end_rm_reward_func": 0.19921875,
|
| 1174 |
-
"rewards/length_reward_func": -0.0234375,
|
| 1175 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1176 |
-
"step": 93
|
| 1177 |
-
},
|
| 1178 |
-
{
|
| 1179 |
-
"clip_ratio": 0.0050481803773436695,
|
| 1180 |
-
"epoch": 0.0752,
|
| 1181 |
-
"grad_norm": 0.5190161687082562,
|
| 1182 |
-
"kl": 4.3549346923828125,
|
| 1183 |
-
"learning_rate": 5e-07,
|
| 1184 |
-
"loss": 0.0372,
|
| 1185 |
-
"step": 94
|
| 1186 |
-
},
|
| 1187 |
-
{
|
| 1188 |
-
"clip_ratio": 0.0,
|
| 1189 |
-
"completion_length": 3141.1796875,
|
| 1190 |
-
"epoch": 0.076,
|
| 1191 |
-
"grad_norm": 0.6792418921682883,
|
| 1192 |
-
"kl": 51.4798583984375,
|
| 1193 |
-
"learning_rate": 5e-07,
|
| 1194 |
-
"loss": 0.1449,
|
| 1195 |
-
"reward": 0.2333984500146471,
|
| 1196 |
-
"reward_std": 0.4384781029075384,
|
| 1197 |
-
"rewards/end_of_conversation_reward_func": 0.054687501047737896,
|
| 1198 |
-
"rewards/end_rm_reward_func": 0.2607421875,
|
| 1199 |
-
"rewards/length_reward_func": -0.08203125,
|
| 1200 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1201 |
-
"step": 95
|
| 1202 |
-
},
|
| 1203 |
-
{
|
| 1204 |
-
"clip_ratio": 0.004932425799779594,
|
| 1205 |
-
"epoch": 0.0768,
|
| 1206 |
-
"grad_norm": 0.5823614093199818,
|
| 1207 |
-
"kl": 39.2484130859375,
|
| 1208 |
-
"learning_rate": 5e-07,
|
| 1209 |
-
"loss": 0.1446,
|
| 1210 |
-
"step": 96
|
| 1211 |
-
},
|
| 1212 |
-
{
|
| 1213 |
-
"clip_ratio": 0.0,
|
| 1214 |
-
"completion_length": 2914.5,
|
| 1215 |
-
"epoch": 0.0776,
|
| 1216 |
-
"grad_norm": 0.5836376252053956,
|
| 1217 |
-
"kl": 3.556396484375,
|
| 1218 |
-
"learning_rate": 5e-07,
|
| 1219 |
-
"loss": 0.113,
|
| 1220 |
-
"reward": 0.12675781635334715,
|
| 1221 |
-
"reward_std": 0.3375853835605085,
|
| 1222 |
-
"rewards/end_of_conversation_reward_func": 0.05546875100117177,
|
| 1223 |
-
"rewards/end_rm_reward_func": 0.1181640625,
|
| 1224 |
-
"rewards/length_reward_func": -0.046875,
|
| 1225 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1226 |
-
"step": 97
|
| 1227 |
-
},
|
| 1228 |
-
{
|
| 1229 |
-
"clip_ratio": 0.005294336486258544,
|
| 1230 |
-
"epoch": 0.0784,
|
| 1231 |
-
"grad_norm": 0.600979136556631,
|
| 1232 |
-
"kl": 2.90380859375,
|
| 1233 |
-
"learning_rate": 5e-07,
|
| 1234 |
-
"loss": 0.1127,
|
| 1235 |
-
"step": 98
|
| 1236 |
-
},
|
| 1237 |
-
{
|
| 1238 |
-
"clip_ratio": 0.0,
|
| 1239 |
-
"completion_length": 2303.875,
|
| 1240 |
-
"epoch": 0.0792,
|
| 1241 |
-
"grad_norm": 1.206634850995924,
|
| 1242 |
-
"kl": 2.009918212890625,
|
| 1243 |
-
"learning_rate": 5e-07,
|
| 1244 |
-
"loss": 0.0546,
|
| 1245 |
-
"reward": 0.30605470086447895,
|
| 1246 |
-
"reward_std": 0.41795148770324886,
|
| 1247 |
-
"rewards/end_of_conversation_reward_func": 0.06875000114087015,
|
| 1248 |
-
"rewards/end_rm_reward_func": 0.2724609375,
|
| 1249 |
-
"rewards/length_reward_func": -0.03515625,
|
| 1250 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1251 |
-
"step": 99
|
| 1252 |
-
},
|
| 1253 |
-
{
|
| 1254 |
-
"clip_ratio": 0.006350497540552169,
|
| 1255 |
-
"epoch": 0.08,
|
| 1256 |
-
"grad_norm": 0.8269327675220272,
|
| 1257 |
-
"kl": 1.482818603515625,
|
| 1258 |
-
"learning_rate": 5e-07,
|
| 1259 |
-
"loss": 0.054,
|
| 1260 |
-
"step": 100
|
| 1261 |
-
},
|
| 1262 |
-
{
|
| 1263 |
-
"clip_ratio": 0.0,
|
| 1264 |
-
"completion_length": 1956.078125,
|
| 1265 |
-
"epoch": 0.0808,
|
| 1266 |
-
"grad_norm": 0.9861989653570116,
|
| 1267 |
-
"kl": 2.6380615234375,
|
| 1268 |
-
"learning_rate": 5e-07,
|
| 1269 |
-
"loss": 0.0302,
|
| 1270 |
-
"reward": 0.35351563640870154,
|
| 1271 |
-
"reward_std": 0.46613426599651575,
|
| 1272 |
-
"rewards/end_of_conversation_reward_func": 0.058593750931322575,
|
| 1273 |
-
"rewards/end_rm_reward_func": 0.310546875,
|
| 1274 |
-
"rewards/length_reward_func": -0.015625,
|
| 1275 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1276 |
-
"step": 101
|
| 1277 |
-
},
|
| 1278 |
-
{
|
| 1279 |
-
"clip_ratio": 0.007070519757689908,
|
| 1280 |
-
"epoch": 0.0816,
|
| 1281 |
-
"grad_norm": 0.7756715106427584,
|
| 1282 |
-
"kl": 1.98809814453125,
|
| 1283 |
-
"learning_rate": 5e-07,
|
| 1284 |
-
"loss": 0.0295,
|
| 1285 |
-
"step": 102
|
| 1286 |
-
},
|
| 1287 |
-
{
|
| 1288 |
-
"clip_ratio": 0.0,
|
| 1289 |
-
"completion_length": 2405.5546875,
|
| 1290 |
-
"epoch": 0.0824,
|
| 1291 |
-
"grad_norm": 0.8282537056915061,
|
| 1292 |
-
"kl": 22.3995361328125,
|
| 1293 |
-
"learning_rate": 5e-07,
|
| 1294 |
-
"loss": 0.0809,
|
| 1295 |
-
"reward": 0.2816406422643922,
|
| 1296 |
-
"reward_std": 0.4782843650318682,
|
| 1297 |
-
"rewards/end_of_conversation_reward_func": 0.06093750102445483,
|
| 1298 |
-
"rewards/end_rm_reward_func": 0.251953125,
|
| 1299 |
-
"rewards/length_reward_func": -0.03125,
|
| 1300 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1301 |
-
"step": 103
|
| 1302 |
-
},
|
| 1303 |
-
{
|
| 1304 |
-
"clip_ratio": 0.006699568155454472,
|
| 1305 |
-
"epoch": 0.0832,
|
| 1306 |
-
"grad_norm": 0.7403840647308474,
|
| 1307 |
-
"kl": 18.2386474609375,
|
| 1308 |
-
"learning_rate": 5e-07,
|
| 1309 |
-
"loss": 0.0804,
|
| 1310 |
-
"step": 104
|
| 1311 |
-
},
|
| 1312 |
-
{
|
| 1313 |
-
"clip_ratio": 0.0,
|
| 1314 |
-
"completion_length": 2866.75,
|
| 1315 |
-
"epoch": 0.084,
|
| 1316 |
-
"grad_norm": 0.6852221586130566,
|
| 1317 |
-
"kl": 157.30657958984375,
|
| 1318 |
-
"learning_rate": 5e-07,
|
| 1319 |
-
"loss": 0.1108,
|
| 1320 |
-
"reward": 0.13496094022411853,
|
| 1321 |
-
"reward_std": 0.354372289031744,
|
| 1322 |
-
"rewards/end_of_conversation_reward_func": 0.06171875010477379,
|
| 1323 |
-
"rewards/end_rm_reward_func": 0.1123046875,
|
| 1324 |
-
"rewards/length_reward_func": -0.0390625,
|
| 1325 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1326 |
-
"step": 105
|
| 1327 |
-
},
|
| 1328 |
-
{
|
| 1329 |
-
"clip_ratio": 0.004895409845630638,
|
| 1330 |
-
"epoch": 0.0848,
|
| 1331 |
-
"grad_norm": 0.6017523870744153,
|
| 1332 |
-
"kl": 252.56585693359375,
|
| 1333 |
-
"learning_rate": 5e-07,
|
| 1334 |
-
"loss": 0.1103,
|
| 1335 |
-
"step": 106
|
| 1336 |
-
},
|
| 1337 |
-
{
|
| 1338 |
-
"clip_ratio": 0.0,
|
| 1339 |
-
"completion_length": 2348.046875,
|
| 1340 |
-
"epoch": 0.0856,
|
| 1341 |
-
"grad_norm": 0.914456395912047,
|
| 1342 |
-
"kl": 21.73077392578125,
|
| 1343 |
-
"learning_rate": 5e-07,
|
| 1344 |
-
"loss": 0.0961,
|
| 1345 |
-
"reward": 0.3595703274477273,
|
| 1346 |
-
"reward_std": 0.4024982349947095,
|
| 1347 |
-
"rewards/end_of_conversation_reward_func": 0.06562500080326572,
|
| 1348 |
-
"rewards/end_rm_reward_func": 0.3095703125,
|
| 1349 |
-
"rewards/length_reward_func": -0.015625,
|
| 1350 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1351 |
-
"step": 107
|
| 1352 |
-
},
|
| 1353 |
-
{
|
| 1354 |
-
"clip_ratio": 0.0052587240061257035,
|
| 1355 |
-
"epoch": 0.0864,
|
| 1356 |
-
"grad_norm": 0.7438499135211943,
|
| 1357 |
-
"kl": 30.40380859375,
|
| 1358 |
-
"learning_rate": 5e-07,
|
| 1359 |
-
"loss": 0.0951,
|
| 1360 |
-
"step": 108
|
| 1361 |
-
},
|
| 1362 |
-
{
|
| 1363 |
-
"clip_ratio": 0.0,
|
| 1364 |
-
"completion_length": 2089.40625,
|
| 1365 |
-
"epoch": 0.0872,
|
| 1366 |
-
"grad_norm": 0.7458266319039674,
|
| 1367 |
-
"kl": 90.74044799804688,
|
| 1368 |
-
"learning_rate": 5e-07,
|
| 1369 |
-
"loss": 0.0166,
|
| 1370 |
-
"reward": 0.43183596106246114,
|
| 1371 |
-
"reward_std": 0.32141723250970244,
|
| 1372 |
-
"rewards/end_of_conversation_reward_func": 0.06953125190921128,
|
| 1373 |
-
"rewards/end_rm_reward_func": 0.3818359375,
|
| 1374 |
-
"rewards/length_reward_func": -0.01953125,
|
| 1375 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1376 |
-
"step": 109
|
| 1377 |
-
},
|
| 1378 |
-
{
|
| 1379 |
-
"clip_ratio": 0.00607385064358823,
|
| 1380 |
-
"epoch": 0.088,
|
| 1381 |
-
"grad_norm": 0.6925768537695978,
|
| 1382 |
-
"kl": 94.23519897460938,
|
| 1383 |
-
"learning_rate": 5e-07,
|
| 1384 |
-
"loss": 0.0162,
|
| 1385 |
-
"step": 110
|
| 1386 |
-
},
|
| 1387 |
-
{
|
| 1388 |
-
"clip_ratio": 0.0,
|
| 1389 |
-
"completion_length": 2093.875,
|
| 1390 |
-
"epoch": 0.0888,
|
| 1391 |
-
"grad_norm": 0.6947966762196172,
|
| 1392 |
-
"kl": 107.85791015625,
|
| 1393 |
-
"learning_rate": 5e-07,
|
| 1394 |
-
"loss": 0.053,
|
| 1395 |
-
"reward": 0.36738282535225153,
|
| 1396 |
-
"reward_std": 0.41466561891138554,
|
| 1397 |
-
"rewards/end_of_conversation_reward_func": 0.05781250121071935,
|
| 1398 |
-
"rewards/end_rm_reward_func": 0.3212890625,
|
| 1399 |
-
"rewards/length_reward_func": -0.01171875,
|
| 1400 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1401 |
-
"step": 111
|
| 1402 |
-
},
|
| 1403 |
-
{
|
| 1404 |
-
"clip_ratio": 0.006751221866579726,
|
| 1405 |
-
"epoch": 0.0896,
|
| 1406 |
-
"grad_norm": 0.6430098763794685,
|
| 1407 |
-
"kl": 92.2412109375,
|
| 1408 |
-
"learning_rate": 5e-07,
|
| 1409 |
-
"loss": 0.0524,
|
| 1410 |
-
"step": 112
|
| 1411 |
-
},
|
| 1412 |
-
{
|
| 1413 |
-
"clip_ratio": 0.0,
|
| 1414 |
-
"completion_length": 2857.453125,
|
| 1415 |
-
"epoch": 0.0904,
|
| 1416 |
-
"grad_norm": 5.39283424098446,
|
| 1417 |
-
"kl": 3.7830810546875,
|
| 1418 |
-
"learning_rate": 5e-07,
|
| 1419 |
-
"loss": 0.0634,
|
| 1420 |
-
"reward": 0.2548828301951289,
|
| 1421 |
-
"reward_std": 0.3852043147198856,
|
| 1422 |
-
"rewards/end_of_conversation_reward_func": 0.06640625046566129,
|
| 1423 |
-
"rewards/end_rm_reward_func": 0.2431640625,
|
| 1424 |
-
"rewards/length_reward_func": -0.0546875,
|
| 1425 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1426 |
-
"step": 113
|
| 1427 |
-
},
|
| 1428 |
-
{
|
| 1429 |
-
"clip_ratio": 0.005811518072732724,
|
| 1430 |
-
"epoch": 0.0912,
|
| 1431 |
-
"grad_norm": 2.422126311857217,
|
| 1432 |
-
"kl": 3.3790283203125,
|
| 1433 |
-
"learning_rate": 5e-07,
|
| 1434 |
-
"loss": 0.0607,
|
| 1435 |
-
"step": 114
|
| 1436 |
-
},
|
| 1437 |
-
{
|
| 1438 |
-
"clip_ratio": 0.0,
|
| 1439 |
-
"completion_length": 2221.609375,
|
| 1440 |
-
"epoch": 0.092,
|
| 1441 |
-
"grad_norm": 1.6598304873547154,
|
| 1442 |
-
"kl": 20.3868408203125,
|
| 1443 |
-
"learning_rate": 5e-07,
|
| 1444 |
-
"loss": 0.084,
|
| 1445 |
-
"reward": 0.39785158331505954,
|
| 1446 |
-
"reward_std": 0.4529835293069482,
|
| 1447 |
-
"rewards/end_of_conversation_reward_func": 0.07265625102445483,
|
| 1448 |
-
"rewards/end_rm_reward_func": 0.3330078125,
|
| 1449 |
-
"rewards/length_reward_func": -0.0078125,
|
| 1450 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1451 |
-
"step": 115
|
| 1452 |
-
},
|
| 1453 |
-
{
|
| 1454 |
-
"clip_ratio": 0.00665678849327378,
|
| 1455 |
-
"epoch": 0.0928,
|
| 1456 |
-
"grad_norm": 1.0235295266605744,
|
| 1457 |
-
"kl": 17.92303466796875,
|
| 1458 |
-
"learning_rate": 5e-07,
|
| 1459 |
-
"loss": 0.083,
|
| 1460 |
-
"step": 116
|
| 1461 |
-
},
|
| 1462 |
-
{
|
| 1463 |
-
"clip_ratio": 0.0,
|
| 1464 |
-
"completion_length": 2825.5546875,
|
| 1465 |
-
"epoch": 0.0936,
|
| 1466 |
-
"grad_norm": 0.6848549595945271,
|
| 1467 |
-
"kl": 1.4246826171875,
|
| 1468 |
-
"learning_rate": 5e-07,
|
| 1469 |
-
"loss": 0.0459,
|
| 1470 |
-
"reward": 0.3335937559604645,
|
| 1471 |
-
"reward_std": 0.3852117173373699,
|
| 1472 |
-
"rewards/end_of_conversation_reward_func": 0.064062500721775,
|
| 1473 |
-
"rewards/end_rm_reward_func": 0.3046875,
|
| 1474 |
-
"rewards/length_reward_func": -0.03515625,
|
| 1475 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1476 |
-
"step": 117
|
| 1477 |
-
},
|
| 1478 |
-
{
|
| 1479 |
-
"clip_ratio": 0.006251567567232996,
|
| 1480 |
-
"epoch": 0.0944,
|
| 1481 |
-
"grad_norm": 0.6767843027822119,
|
| 1482 |
-
"kl": 1.2928466796875,
|
| 1483 |
-
"learning_rate": 5e-07,
|
| 1484 |
-
"loss": 0.0453,
|
| 1485 |
-
"step": 118
|
| 1486 |
-
},
|
| 1487 |
-
{
|
| 1488 |
-
"clip_ratio": 0.0,
|
| 1489 |
-
"completion_length": 2318.1640625,
|
| 1490 |
-
"epoch": 0.0952,
|
| 1491 |
-
"grad_norm": 10.653439671553652,
|
| 1492 |
-
"kl": 23.6943359375,
|
| 1493 |
-
"learning_rate": 5e-07,
|
| 1494 |
-
"loss": 0.0204,
|
| 1495 |
-
"reward": 0.3726562592200935,
|
| 1496 |
-
"reward_std": 0.4175419304519892,
|
| 1497 |
-
"rewards/end_of_conversation_reward_func": 0.06406250066356733,
|
| 1498 |
-
"rewards/end_rm_reward_func": 0.33203125,
|
| 1499 |
-
"rewards/length_reward_func": -0.0234375,
|
| 1500 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1501 |
-
"step": 119
|
| 1502 |
-
},
|
| 1503 |
-
{
|
| 1504 |
-
"clip_ratio": 0.005671160441124812,
|
| 1505 |
-
"epoch": 0.096,
|
| 1506 |
-
"grad_norm": 2.3552929076073736,
|
| 1507 |
-
"kl": 25.97442626953125,
|
| 1508 |
-
"learning_rate": 5e-07,
|
| 1509 |
-
"loss": 0.0138,
|
| 1510 |
-
"step": 120
|
| 1511 |
-
},
|
| 1512 |
-
{
|
| 1513 |
-
"clip_ratio": 0.0,
|
| 1514 |
-
"completion_length": 1880.5078125,
|
| 1515 |
-
"epoch": 0.0968,
|
| 1516 |
-
"grad_norm": 0.7833349498274812,
|
| 1517 |
-
"kl": 6.0826416015625,
|
| 1518 |
-
"learning_rate": 5e-07,
|
| 1519 |
-
"loss": 0.0555,
|
| 1520 |
-
"reward": 0.4580078381113708,
|
| 1521 |
-
"reward_std": 0.47520614485256374,
|
| 1522 |
-
"rewards/end_of_conversation_reward_func": 0.07031250139698386,
|
| 1523 |
-
"rewards/end_rm_reward_func": 0.4111328125,
|
| 1524 |
-
"rewards/length_reward_func": -0.0234375,
|
| 1525 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1526 |
-
"step": 121
|
| 1527 |
-
},
|
| 1528 |
-
{
|
| 1529 |
-
"clip_ratio": 0.005595599315711297,
|
| 1530 |
-
"epoch": 0.0976,
|
| 1531 |
-
"grad_norm": 0.7882700293427872,
|
| 1532 |
-
"kl": 8.645355224609375,
|
| 1533 |
-
"learning_rate": 5e-07,
|
| 1534 |
-
"loss": 0.0553,
|
| 1535 |
-
"step": 122
|
| 1536 |
-
},
|
| 1537 |
-
{
|
| 1538 |
-
"clip_ratio": 0.0,
|
| 1539 |
-
"completion_length": 2476.6640625,
|
| 1540 |
-
"epoch": 0.0984,
|
| 1541 |
-
"grad_norm": 0.912901642164902,
|
| 1542 |
-
"kl": 1.38836669921875,
|
| 1543 |
-
"learning_rate": 5e-07,
|
| 1544 |
-
"loss": 0.0478,
|
| 1545 |
-
"reward": 0.3851562616182491,
|
| 1546 |
-
"reward_std": 0.4384188293479383,
|
| 1547 |
-
"rewards/end_of_conversation_reward_func": 0.06093750067520887,
|
| 1548 |
-
"rewards/end_rm_reward_func": 0.328125,
|
| 1549 |
-
"rewards/length_reward_func": -0.00390625,
|
| 1550 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1551 |
-
"step": 123
|
| 1552 |
-
},
|
| 1553 |
-
{
|
| 1554 |
-
"clip_ratio": 0.005667043791618198,
|
| 1555 |
-
"epoch": 0.0992,
|
| 1556 |
-
"grad_norm": 0.6713987475135793,
|
| 1557 |
-
"kl": 1.342529296875,
|
| 1558 |
-
"learning_rate": 5e-07,
|
| 1559 |
-
"loss": 0.047,
|
| 1560 |
-
"step": 124
|
| 1561 |
-
},
|
| 1562 |
-
{
|
| 1563 |
-
"clip_ratio": 0.0,
|
| 1564 |
-
"completion_length": 1897.9921875,
|
| 1565 |
-
"epoch": 0.1,
|
| 1566 |
-
"grad_norm": 1.8396052910399312,
|
| 1567 |
-
"kl": 52.3482666015625,
|
| 1568 |
-
"learning_rate": 5e-07,
|
| 1569 |
-
"loss": 0.0427,
|
| 1570 |
-
"reward": 0.36113282782025635,
|
| 1571 |
-
"reward_std": 0.4283226621337235,
|
| 1572 |
-
"rewards/end_of_conversation_reward_func": 0.06328125204890966,
|
| 1573 |
-
"rewards/end_rm_reward_func": 0.2978515625,
|
| 1574 |
-
"rewards/length_reward_func": 0.0,
|
| 1575 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1576 |
-
"step": 125
|
| 1577 |
-
},
|
| 1578 |
-
{
|
| 1579 |
-
"clip_ratio": 0.008028718992136419,
|
| 1580 |
-
"epoch": 0.1008,
|
| 1581 |
-
"grad_norm": 0.9744864161133155,
|
| 1582 |
-
"kl": 69.21954345703125,
|
| 1583 |
-
"learning_rate": 5e-07,
|
| 1584 |
-
"loss": 0.0419,
|
| 1585 |
-
"step": 126
|
| 1586 |
-
},
|
| 1587 |
-
{
|
| 1588 |
-
"clip_ratio": 0.0,
|
| 1589 |
-
"completion_length": 2345.0234375,
|
| 1590 |
-
"epoch": 0.1016,
|
| 1591 |
-
"grad_norm": 0.9724894359575696,
|
| 1592 |
-
"kl": 32.419189453125,
|
| 1593 |
-
"learning_rate": 5e-07,
|
| 1594 |
-
"loss": -0.0304,
|
| 1595 |
-
"reward": 0.3443359527736902,
|
| 1596 |
-
"reward_std": 0.4752398133277893,
|
| 1597 |
-
"rewards/end_of_conversation_reward_func": 0.06796875188592821,
|
| 1598 |
-
"rewards/end_rm_reward_func": 0.2841796875,
|
| 1599 |
-
"rewards/length_reward_func": -0.0078125,
|
| 1600 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1601 |
-
"step": 127
|
| 1602 |
-
},
|
| 1603 |
-
{
|
| 1604 |
-
"clip_ratio": 0.006735242568538524,
|
| 1605 |
-
"epoch": 0.1024,
|
| 1606 |
-
"grad_norm": 0.7654776812407493,
|
| 1607 |
-
"kl": 16.9310302734375,
|
| 1608 |
-
"learning_rate": 5e-07,
|
| 1609 |
-
"loss": -0.031,
|
| 1610 |
-
"step": 128
|
| 1611 |
-
},
|
| 1612 |
-
{
|
| 1613 |
-
"clip_ratio": 0.0,
|
| 1614 |
-
"completion_length": 2669.90625,
|
| 1615 |
-
"epoch": 0.1032,
|
| 1616 |
-
"grad_norm": 6.438509726974655,
|
| 1617 |
-
"kl": 2.64239501953125,
|
| 1618 |
-
"learning_rate": 5e-07,
|
| 1619 |
-
"loss": 0.0327,
|
| 1620 |
-
"reward": 0.21308594616129994,
|
| 1621 |
-
"reward_std": 0.44294028589501977,
|
| 1622 |
-
"rewards/end_of_conversation_reward_func": 0.0578125006868504,
|
| 1623 |
-
"rewards/end_rm_reward_func": 0.1787109375,
|
| 1624 |
-
"rewards/length_reward_func": -0.0234375,
|
| 1625 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1626 |
-
"step": 129
|
| 1627 |
-
},
|
| 1628 |
-
{
|
| 1629 |
-
"clip_ratio": 0.005368985250242986,
|
| 1630 |
-
"epoch": 0.104,
|
| 1631 |
-
"grad_norm": 0.8747379437194369,
|
| 1632 |
-
"kl": 1.28466796875,
|
| 1633 |
-
"learning_rate": 5e-07,
|
| 1634 |
-
"loss": 0.0294,
|
| 1635 |
-
"step": 130
|
| 1636 |
-
},
|
| 1637 |
-
{
|
| 1638 |
-
"clip_ratio": 0.0,
|
| 1639 |
-
"completion_length": 2500.625,
|
| 1640 |
-
"epoch": 0.1048,
|
| 1641 |
-
"grad_norm": 0.6477351309697729,
|
| 1642 |
-
"kl": 2.7420654296875,
|
| 1643 |
-
"learning_rate": 5e-07,
|
| 1644 |
-
"loss": -0.0209,
|
| 1645 |
-
"reward": 0.31210938445292413,
|
| 1646 |
-
"reward_std": 0.37651610001921654,
|
| 1647 |
-
"rewards/end_of_conversation_reward_func": 0.060156251420266926,
|
| 1648 |
-
"rewards/end_rm_reward_func": 0.279296875,
|
| 1649 |
-
"rewards/length_reward_func": -0.02734375,
|
| 1650 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1651 |
-
"step": 131
|
| 1652 |
-
},
|
| 1653 |
-
{
|
| 1654 |
-
"clip_ratio": 0.005497576406924054,
|
| 1655 |
-
"epoch": 0.1056,
|
| 1656 |
-
"grad_norm": 0.6479794324694829,
|
| 1657 |
-
"kl": 1.95208740234375,
|
| 1658 |
-
"learning_rate": 5e-07,
|
| 1659 |
-
"loss": -0.0211,
|
| 1660 |
-
"step": 132
|
| 1661 |
-
},
|
| 1662 |
-
{
|
| 1663 |
-
"clip_ratio": 0.0,
|
| 1664 |
-
"completion_length": 2211.4453125,
|
| 1665 |
-
"epoch": 0.1064,
|
| 1666 |
-
"grad_norm": 1.0082144435462406,
|
| 1667 |
-
"kl": 1.69384765625,
|
| 1668 |
-
"learning_rate": 5e-07,
|
| 1669 |
-
"loss": 0.0535,
|
| 1670 |
-
"reward": 0.3933593933470547,
|
| 1671 |
-
"reward_std": 0.4518305202946067,
|
| 1672 |
-
"rewards/end_of_conversation_reward_func": 0.06328125129221007,
|
| 1673 |
-
"rewards/end_rm_reward_func": 0.361328125,
|
| 1674 |
-
"rewards/length_reward_func": -0.03125,
|
| 1675 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1676 |
-
"step": 133
|
| 1677 |
-
},
|
| 1678 |
-
{
|
| 1679 |
-
"clip_ratio": 0.0054282144556054845,
|
| 1680 |
-
"epoch": 0.1072,
|
| 1681 |
-
"grad_norm": 0.7238908888250458,
|
| 1682 |
-
"kl": 1.4329833984375,
|
| 1683 |
-
"learning_rate": 5e-07,
|
| 1684 |
-
"loss": 0.0526,
|
| 1685 |
-
"step": 134
|
| 1686 |
-
},
|
| 1687 |
-
{
|
| 1688 |
-
"clip_ratio": 0.0,
|
| 1689 |
-
"completion_length": 2542.8046875,
|
| 1690 |
-
"epoch": 0.108,
|
| 1691 |
-
"grad_norm": 0.639586024156753,
|
| 1692 |
-
"kl": 125.3736572265625,
|
| 1693 |
-
"learning_rate": 5e-07,
|
| 1694 |
-
"loss": 0.0567,
|
| 1695 |
-
"reward": 0.27167970268055797,
|
| 1696 |
-
"reward_std": 0.4119821102358401,
|
| 1697 |
-
"rewards/end_of_conversation_reward_func": 0.06562500086147338,
|
| 1698 |
-
"rewards/end_rm_reward_func": 0.2255859375,
|
| 1699 |
-
"rewards/length_reward_func": -0.01953125,
|
| 1700 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1701 |
-
"step": 135
|
| 1702 |
-
},
|
| 1703 |
-
{
|
| 1704 |
-
"clip_ratio": 0.005144305920111947,
|
| 1705 |
-
"epoch": 0.1088,
|
| 1706 |
-
"grad_norm": 0.6324039246445893,
|
| 1707 |
-
"kl": 171.12823486328125,
|
| 1708 |
-
"learning_rate": 5e-07,
|
| 1709 |
-
"loss": 0.0565,
|
| 1710 |
-
"step": 136
|
| 1711 |
-
},
|
| 1712 |
-
{
|
| 1713 |
-
"clip_ratio": 0.0,
|
| 1714 |
-
"completion_length": 2629.3046875,
|
| 1715 |
-
"epoch": 0.1096,
|
| 1716 |
-
"grad_norm": 0.6701878690042156,
|
| 1717 |
-
"kl": 3.42401123046875,
|
| 1718 |
-
"learning_rate": 5e-07,
|
| 1719 |
-
"loss": 0.0713,
|
| 1720 |
-
"reward": 0.30917970347218215,
|
| 1721 |
-
"reward_std": 0.48418763000518084,
|
| 1722 |
-
"rewards/end_of_conversation_reward_func": 0.06015625176951289,
|
| 1723 |
-
"rewards/end_rm_reward_func": 0.2685546875,
|
| 1724 |
-
"rewards/length_reward_func": -0.01953125,
|
| 1725 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1726 |
-
"step": 137
|
| 1727 |
-
},
|
| 1728 |
-
{
|
| 1729 |
-
"clip_ratio": 0.004802481475053355,
|
| 1730 |
-
"epoch": 0.1104,
|
| 1731 |
-
"grad_norm": 0.6296621696156859,
|
| 1732 |
-
"kl": 4.0013427734375,
|
| 1733 |
-
"learning_rate": 5e-07,
|
| 1734 |
-
"loss": 0.0709,
|
| 1735 |
-
"step": 138
|
| 1736 |
-
},
|
| 1737 |
-
{
|
| 1738 |
-
"clip_ratio": 0.0,
|
| 1739 |
-
"completion_length": 2347.1875,
|
| 1740 |
-
"epoch": 0.1112,
|
| 1741 |
-
"grad_norm": 0.6949091471055229,
|
| 1742 |
-
"kl": 29.50958251953125,
|
| 1743 |
-
"learning_rate": 5e-07,
|
| 1744 |
-
"loss": 0.0567,
|
| 1745 |
-
"reward": 0.23105469765141606,
|
| 1746 |
-
"reward_std": 0.35395989241078496,
|
| 1747 |
-
"rewards/end_of_conversation_reward_func": 0.056250001885928214,
|
| 1748 |
-
"rewards/end_rm_reward_func": 0.1904296875,
|
| 1749 |
-
"rewards/length_reward_func": -0.015625,
|
| 1750 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1751 |
-
"step": 139
|
| 1752 |
-
},
|
| 1753 |
-
{
|
| 1754 |
-
"clip_ratio": 0.005910978390602395,
|
| 1755 |
-
"epoch": 0.112,
|
| 1756 |
-
"grad_norm": 0.649618448371434,
|
| 1757 |
-
"kl": 13.6500244140625,
|
| 1758 |
-
"learning_rate": 5e-07,
|
| 1759 |
-
"loss": 0.0561,
|
| 1760 |
-
"step": 140
|
| 1761 |
-
},
|
| 1762 |
-
{
|
| 1763 |
-
"clip_ratio": 0.0,
|
| 1764 |
-
"completion_length": 2259.3046875,
|
| 1765 |
-
"epoch": 0.1128,
|
| 1766 |
-
"grad_norm": 0.7642501957021136,
|
| 1767 |
-
"kl": 1.39849853515625,
|
| 1768 |
-
"learning_rate": 5e-07,
|
| 1769 |
-
"loss": 0.0365,
|
| 1770 |
-
"reward": 0.44687501853331923,
|
| 1771 |
-
"reward_std": 0.44018132728524506,
|
| 1772 |
-
"rewards/end_of_conversation_reward_func": 0.06796875060535967,
|
| 1773 |
-
"rewards/end_rm_reward_func": 0.390625,
|
| 1774 |
-
"rewards/length_reward_func": -0.01171875,
|
| 1775 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1776 |
-
"step": 141
|
| 1777 |
-
},
|
| 1778 |
-
{
|
| 1779 |
-
"clip_ratio": 0.006374053191393614,
|
| 1780 |
-
"epoch": 0.1136,
|
| 1781 |
-
"grad_norm": 0.6891960123007369,
|
| 1782 |
-
"kl": 1.7943115234375,
|
| 1783 |
-
"learning_rate": 5e-07,
|
| 1784 |
-
"loss": 0.0359,
|
| 1785 |
-
"step": 142
|
| 1786 |
-
},
|
| 1787 |
-
{
|
| 1788 |
-
"clip_ratio": 0.0,
|
| 1789 |
-
"completion_length": 2309.171875,
|
| 1790 |
-
"epoch": 0.1144,
|
| 1791 |
-
"grad_norm": 0.8952677516080388,
|
| 1792 |
-
"kl": 1.2454833984375,
|
| 1793 |
-
"learning_rate": 5e-07,
|
| 1794 |
-
"loss": 0.0781,
|
| 1795 |
-
"reward": 0.4085937717463821,
|
| 1796 |
-
"reward_std": 0.43159425538033247,
|
| 1797 |
-
"rewards/end_of_conversation_reward_func": 0.06875000125728548,
|
| 1798 |
-
"rewards/end_rm_reward_func": 0.34375,
|
| 1799 |
-
"rewards/length_reward_func": -0.00390625,
|
| 1800 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1801 |
-
"step": 143
|
| 1802 |
-
},
|
| 1803 |
-
{
|
| 1804 |
-
"clip_ratio": 0.005346491030650213,
|
| 1805 |
-
"epoch": 0.1152,
|
| 1806 |
-
"grad_norm": 0.8232427405557934,
|
| 1807 |
-
"kl": 2.2227783203125,
|
| 1808 |
-
"learning_rate": 5e-07,
|
| 1809 |
-
"loss": 0.0771,
|
| 1810 |
-
"step": 144
|
| 1811 |
-
},
|
| 1812 |
-
{
|
| 1813 |
-
"clip_ratio": 0.0,
|
| 1814 |
-
"completion_length": 2151.1484375,
|
| 1815 |
-
"epoch": 0.116,
|
| 1816 |
-
"grad_norm": 1.151846893342153,
|
| 1817 |
-
"kl": 2.817626953125,
|
| 1818 |
-
"learning_rate": 5e-07,
|
| 1819 |
-
"loss": 0.0217,
|
| 1820 |
-
"reward": 0.2933593824855052,
|
| 1821 |
-
"reward_std": 0.3206550294999033,
|
| 1822 |
-
"rewards/end_of_conversation_reward_func": 0.07265625079162419,
|
| 1823 |
-
"rewards/end_rm_reward_func": 0.228515625,
|
| 1824 |
-
"rewards/length_reward_func": -0.0078125,
|
| 1825 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1826 |
-
"step": 145
|
| 1827 |
-
},
|
| 1828 |
-
{
|
| 1829 |
-
"clip_ratio": 0.006764089310308918,
|
| 1830 |
-
"epoch": 0.1168,
|
| 1831 |
-
"grad_norm": 0.9623961306886136,
|
| 1832 |
-
"kl": 3.7879638671875,
|
| 1833 |
-
"learning_rate": 5e-07,
|
| 1834 |
-
"loss": 0.0207,
|
| 1835 |
-
"step": 146
|
| 1836 |
-
},
|
| 1837 |
-
{
|
| 1838 |
-
"clip_ratio": 0.0,
|
| 1839 |
-
"completion_length": 2613.7265625,
|
| 1840 |
-
"epoch": 0.1176,
|
| 1841 |
-
"grad_norm": 0.924531089984747,
|
| 1842 |
-
"kl": 0.8990478515625,
|
| 1843 |
-
"learning_rate": 5e-07,
|
| 1844 |
-
"loss": 0.0676,
|
| 1845 |
-
"reward": 0.39667969988659024,
|
| 1846 |
-
"reward_std": 0.4251254890114069,
|
| 1847 |
-
"rewards/end_of_conversation_reward_func": 0.06562500074505806,
|
| 1848 |
-
"rewards/end_rm_reward_func": 0.3466796875,
|
| 1849 |
-
"rewards/length_reward_func": -0.015625,
|
| 1850 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1851 |
-
"step": 147
|
| 1852 |
-
},
|
| 1853 |
-
{
|
| 1854 |
-
"clip_ratio": 0.005301513941958547,
|
| 1855 |
-
"epoch": 0.1184,
|
| 1856 |
-
"grad_norm": 0.8069427285929358,
|
| 1857 |
-
"kl": 1.39166259765625,
|
| 1858 |
-
"learning_rate": 5e-07,
|
| 1859 |
-
"loss": 0.0669,
|
| 1860 |
-
"step": 148
|
| 1861 |
-
},
|
| 1862 |
-
{
|
| 1863 |
-
"clip_ratio": 0.0,
|
| 1864 |
-
"completion_length": 2272.7734375,
|
| 1865 |
-
"epoch": 0.1192,
|
| 1866 |
-
"grad_norm": 0.9723370963767047,
|
| 1867 |
-
"kl": 0.997314453125,
|
| 1868 |
-
"learning_rate": 5e-07,
|
| 1869 |
-
"loss": 0.0187,
|
| 1870 |
-
"reward": 0.4199218891444616,
|
| 1871 |
-
"reward_std": 0.4015230783261359,
|
| 1872 |
-
"rewards/end_of_conversation_reward_func": 0.06640625093132257,
|
| 1873 |
-
"rewards/end_rm_reward_func": 0.365234375,
|
| 1874 |
-
"rewards/length_reward_func": -0.01171875,
|
| 1875 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1876 |
-
"step": 149
|
| 1877 |
-
},
|
| 1878 |
-
{
|
| 1879 |
-
"clip_ratio": 0.005973719438770786,
|
| 1880 |
-
"epoch": 0.12,
|
| 1881 |
-
"grad_norm": 0.8532859182657302,
|
| 1882 |
-
"kl": 0.9261474609375,
|
| 1883 |
-
"learning_rate": 5e-07,
|
| 1884 |
-
"loss": 0.0178,
|
| 1885 |
-
"step": 150
|
| 1886 |
-
},
|
| 1887 |
-
{
|
| 1888 |
-
"clip_ratio": 0.0,
|
| 1889 |
-
"completion_length": 2505.875,
|
| 1890 |
-
"epoch": 0.1208,
|
| 1891 |
-
"grad_norm": 0.8842796434094805,
|
| 1892 |
-
"kl": 9.147705078125,
|
| 1893 |
-
"learning_rate": 5e-07,
|
| 1894 |
-
"loss": 0.0431,
|
| 1895 |
-
"reward": 0.3224609545432031,
|
| 1896 |
-
"reward_std": 0.512933537364006,
|
| 1897 |
-
"rewards/end_of_conversation_reward_func": 0.06171875155996531,
|
| 1898 |
-
"rewards/end_rm_reward_func": 0.2724609375,
|
| 1899 |
-
"rewards/length_reward_func": -0.01171875,
|
| 1900 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1901 |
-
"step": 151
|
| 1902 |
-
},
|
| 1903 |
-
{
|
| 1904 |
-
"clip_ratio": 0.006512533509521745,
|
| 1905 |
-
"epoch": 0.1216,
|
| 1906 |
-
"grad_norm": 0.8332956238777746,
|
| 1907 |
-
"kl": 6.146240234375,
|
| 1908 |
-
"learning_rate": 5e-07,
|
| 1909 |
-
"loss": 0.0423,
|
| 1910 |
-
"step": 152
|
| 1911 |
-
},
|
| 1912 |
-
{
|
| 1913 |
-
"clip_ratio": 0.0,
|
| 1914 |
-
"completion_length": 2544.953125,
|
| 1915 |
-
"epoch": 0.1224,
|
| 1916 |
-
"grad_norm": 0.9874027538635356,
|
| 1917 |
-
"kl": 14.062255859375,
|
| 1918 |
-
"learning_rate": 5e-07,
|
| 1919 |
-
"loss": -0.0164,
|
| 1920 |
-
"reward": 0.17890625371364877,
|
| 1921 |
-
"reward_std": 0.379497186280787,
|
| 1922 |
-
"rewards/end_of_conversation_reward_func": 0.053906250395812094,
|
| 1923 |
-
"rewards/end_rm_reward_func": 0.13671875,
|
| 1924 |
-
"rewards/length_reward_func": -0.01171875,
|
| 1925 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1926 |
-
"step": 153
|
| 1927 |
-
},
|
| 1928 |
-
{
|
| 1929 |
-
"clip_ratio": 0.004760816096677445,
|
| 1930 |
-
"epoch": 0.1232,
|
| 1931 |
-
"grad_norm": 0.8194951383155811,
|
| 1932 |
-
"kl": 25.712646484375,
|
| 1933 |
-
"learning_rate": 5e-07,
|
| 1934 |
-
"loss": -0.017,
|
| 1935 |
-
"step": 154
|
| 1936 |
-
},
|
| 1937 |
-
{
|
| 1938 |
-
"clip_ratio": 0.0,
|
| 1939 |
-
"completion_length": 2325.8828125,
|
| 1940 |
-
"epoch": 0.124,
|
| 1941 |
-
"grad_norm": 0.8787405473781513,
|
| 1942 |
-
"kl": 2.60302734375,
|
| 1943 |
-
"learning_rate": 5e-07,
|
| 1944 |
-
"loss": 0.0489,
|
| 1945 |
-
"reward": 0.43007815058808774,
|
| 1946 |
-
"reward_std": 0.37591410893946886,
|
| 1947 |
-
"rewards/end_of_conversation_reward_func": 0.07656250160653144,
|
| 1948 |
-
"rewards/end_rm_reward_func": 0.373046875,
|
| 1949 |
-
"rewards/length_reward_func": -0.01953125,
|
| 1950 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1951 |
-
"step": 155
|
| 1952 |
-
},
|
| 1953 |
-
{
|
| 1954 |
-
"clip_ratio": 0.005131626487127505,
|
| 1955 |
-
"epoch": 0.1248,
|
| 1956 |
-
"grad_norm": 0.860882474096142,
|
| 1957 |
-
"kl": 2.1358642578125,
|
| 1958 |
-
"learning_rate": 5e-07,
|
| 1959 |
-
"loss": 0.0482,
|
| 1960 |
-
"step": 156
|
| 1961 |
-
},
|
| 1962 |
-
{
|
| 1963 |
-
"clip_ratio": 0.0,
|
| 1964 |
-
"completion_length": 1997.8984375,
|
| 1965 |
-
"epoch": 0.1256,
|
| 1966 |
-
"grad_norm": 4.174478136933485,
|
| 1967 |
-
"kl": 126.7437744140625,
|
| 1968 |
-
"learning_rate": 5e-07,
|
| 1969 |
-
"loss": 0.0147,
|
| 1970 |
-
"reward": 0.3750000196741894,
|
| 1971 |
-
"reward_std": 0.40887491311877966,
|
| 1972 |
-
"rewards/end_of_conversation_reward_func": 0.07421875069849193,
|
| 1973 |
-
"rewards/end_rm_reward_func": 0.3046875,
|
| 1974 |
-
"rewards/length_reward_func": -0.00390625,
|
| 1975 |
-
"rewards/thinking_reward_func": 0.0,
|
| 1976 |
-
"step": 157
|
| 1977 |
-
},
|
| 1978 |
-
{
|
| 1979 |
-
"clip_ratio": 0.006120361678767949,
|
| 1980 |
-
"epoch": 0.1264,
|
| 1981 |
-
"grad_norm": 1.4963821994530664,
|
| 1982 |
-
"kl": 273.1322021484375,
|
| 1983 |
-
"learning_rate": 5e-07,
|
| 1984 |
-
"loss": 0.0139,
|
| 1985 |
-
"step": 158
|
| 1986 |
-
},
|
| 1987 |
-
{
|
| 1988 |
-
"clip_ratio": 0.0,
|
| 1989 |
-
"completion_length": 2511.359375,
|
| 1990 |
-
"epoch": 0.1272,
|
| 1991 |
-
"grad_norm": 5.367528712708823,
|
| 1992 |
-
"kl": 5.2606201171875,
|
| 1993 |
-
"learning_rate": 5e-07,
|
| 1994 |
-
"loss": 0.0698,
|
| 1995 |
-
"reward": 0.3287109578377567,
|
| 1996 |
-
"reward_std": 0.3992913840338588,
|
| 1997 |
-
"rewards/end_of_conversation_reward_func": 0.06406250142026693,
|
| 1998 |
-
"rewards/end_rm_reward_func": 0.2724609375,
|
| 1999 |
-
"rewards/length_reward_func": -0.0078125,
|
| 2000 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2001 |
-
"step": 159
|
| 2002 |
-
},
|
| 2003 |
-
{
|
| 2004 |
-
"clip_ratio": 0.005198820639634505,
|
| 2005 |
-
"epoch": 0.128,
|
| 2006 |
-
"grad_norm": 2.8531678389715998,
|
| 2007 |
-
"kl": 4.1246337890625,
|
| 2008 |
-
"learning_rate": 5e-07,
|
| 2009 |
-
"loss": 0.0663,
|
| 2010 |
-
"step": 160
|
| 2011 |
-
},
|
| 2012 |
-
{
|
| 2013 |
-
"clip_ratio": 0.0,
|
| 2014 |
-
"completion_length": 2238.2734375,
|
| 2015 |
-
"epoch": 0.1288,
|
| 2016 |
-
"grad_norm": 1.2910611646405146,
|
| 2017 |
-
"kl": 2.57861328125,
|
| 2018 |
-
"learning_rate": 5e-07,
|
| 2019 |
-
"loss": 0.0318,
|
| 2020 |
-
"reward": 0.24882814101874828,
|
| 2021 |
-
"reward_std": 0.3937357231043279,
|
| 2022 |
-
"rewards/end_of_conversation_reward_func": 0.0710937506519258,
|
| 2023 |
-
"rewards/end_rm_reward_func": 0.185546875,
|
| 2024 |
-
"rewards/length_reward_func": -0.0078125,
|
| 2025 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2026 |
-
"step": 161
|
| 2027 |
-
},
|
| 2028 |
-
{
|
| 2029 |
-
"clip_ratio": 0.006061407417291775,
|
| 2030 |
-
"epoch": 0.1296,
|
| 2031 |
-
"grad_norm": 1.1788151032961487,
|
| 2032 |
-
"kl": 2.96826171875,
|
| 2033 |
-
"learning_rate": 5e-07,
|
| 2034 |
-
"loss": 0.0305,
|
| 2035 |
-
"step": 162
|
| 2036 |
-
},
|
| 2037 |
-
{
|
| 2038 |
-
"clip_ratio": 0.0,
|
| 2039 |
-
"completion_length": 1621.8671875,
|
| 2040 |
-
"epoch": 0.1304,
|
| 2041 |
-
"grad_norm": 1.082482183152501,
|
| 2042 |
-
"kl": 22.7098388671875,
|
| 2043 |
-
"learning_rate": 5e-07,
|
| 2044 |
-
"loss": 0.0309,
|
| 2045 |
-
"reward": 0.4742187724914402,
|
| 2046 |
-
"reward_std": 0.42131141386926174,
|
| 2047 |
-
"rewards/end_of_conversation_reward_func": 0.07968750153668225,
|
| 2048 |
-
"rewards/end_rm_reward_func": 0.39453125,
|
| 2049 |
-
"rewards/length_reward_func": 0.0,
|
| 2050 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2051 |
-
"step": 163
|
| 2052 |
-
},
|
| 2053 |
-
{
|
| 2054 |
-
"clip_ratio": 0.007301741745322943,
|
| 2055 |
-
"epoch": 0.1312,
|
| 2056 |
-
"grad_norm": 0.999311109978907,
|
| 2057 |
-
"kl": 28.77685546875,
|
| 2058 |
-
"learning_rate": 5e-07,
|
| 2059 |
-
"loss": 0.0297,
|
| 2060 |
-
"step": 164
|
| 2061 |
-
},
|
| 2062 |
-
{
|
| 2063 |
-
"clip_ratio": 0.0,
|
| 2064 |
-
"completion_length": 2164.078125,
|
| 2065 |
-
"epoch": 0.132,
|
| 2066 |
-
"grad_norm": 1.2711815845501575,
|
| 2067 |
-
"kl": 11.022705078125,
|
| 2068 |
-
"learning_rate": 5e-07,
|
| 2069 |
-
"loss": 0.007,
|
| 2070 |
-
"reward": 0.266796886222437,
|
| 2071 |
-
"reward_std": 0.38234878331422806,
|
| 2072 |
-
"rewards/end_of_conversation_reward_func": 0.06953125045401976,
|
| 2073 |
-
"rewards/end_rm_reward_func": 0.208984375,
|
| 2074 |
-
"rewards/length_reward_func": -0.01171875,
|
| 2075 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2076 |
-
"step": 165
|
| 2077 |
-
},
|
| 2078 |
-
{
|
| 2079 |
-
"clip_ratio": 0.005417483043856919,
|
| 2080 |
-
"epoch": 0.1328,
|
| 2081 |
-
"grad_norm": 1.0349824323527632,
|
| 2082 |
-
"kl": 56.939208984375,
|
| 2083 |
-
"learning_rate": 5e-07,
|
| 2084 |
-
"loss": 0.0058,
|
| 2085 |
-
"step": 166
|
| 2086 |
-
},
|
| 2087 |
-
{
|
| 2088 |
-
"clip_ratio": 0.0,
|
| 2089 |
-
"completion_length": 1862.1171875,
|
| 2090 |
-
"epoch": 0.1336,
|
| 2091 |
-
"grad_norm": 1.9132332469984104,
|
| 2092 |
-
"kl": 15.81640625,
|
| 2093 |
-
"learning_rate": 5e-07,
|
| 2094 |
-
"loss": 0.008,
|
| 2095 |
-
"reward": 0.20527344779111445,
|
| 2096 |
-
"reward_std": 0.31551461201161146,
|
| 2097 |
-
"rewards/end_of_conversation_reward_func": 0.07734375126892701,
|
| 2098 |
-
"rewards/end_rm_reward_func": 0.1318359375,
|
| 2099 |
-
"rewards/length_reward_func": -0.00390625,
|
| 2100 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2101 |
-
"step": 167
|
| 2102 |
-
},
|
| 2103 |
-
{
|
| 2104 |
-
"clip_ratio": 0.006755854556104168,
|
| 2105 |
-
"epoch": 0.1344,
|
| 2106 |
-
"grad_norm": 1.1876373433288951,
|
| 2107 |
-
"kl": 14.31689453125,
|
| 2108 |
-
"learning_rate": 5e-07,
|
| 2109 |
-
"loss": 0.0065,
|
| 2110 |
-
"step": 168
|
| 2111 |
-
},
|
| 2112 |
-
{
|
| 2113 |
-
"clip_ratio": 0.0,
|
| 2114 |
-
"completion_length": 1881.125,
|
| 2115 |
-
"epoch": 0.1352,
|
| 2116 |
-
"grad_norm": 1.1641180467318737,
|
| 2117 |
-
"kl": 314.045654296875,
|
| 2118 |
-
"learning_rate": 5e-07,
|
| 2119 |
-
"loss": 0.0178,
|
| 2120 |
-
"reward": 0.3947265843162313,
|
| 2121 |
-
"reward_std": 0.41996311163529754,
|
| 2122 |
-
"rewards/end_of_conversation_reward_func": 0.07343749981373549,
|
| 2123 |
-
"rewards/end_rm_reward_func": 0.3291015625,
|
| 2124 |
-
"rewards/length_reward_func": -0.0078125,
|
| 2125 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2126 |
-
"step": 169
|
| 2127 |
-
},
|
| 2128 |
-
{
|
| 2129 |
-
"clip_ratio": 0.0057207249192288145,
|
| 2130 |
-
"epoch": 0.136,
|
| 2131 |
-
"grad_norm": 1.0261348607435299,
|
| 2132 |
-
"kl": 365.5631103515625,
|
| 2133 |
-
"learning_rate": 5e-07,
|
| 2134 |
-
"loss": 0.0164,
|
| 2135 |
-
"step": 170
|
| 2136 |
-
},
|
| 2137 |
-
{
|
| 2138 |
-
"clip_ratio": 0.0,
|
| 2139 |
-
"completion_length": 2508.2734375,
|
| 2140 |
-
"epoch": 0.1368,
|
| 2141 |
-
"grad_norm": 1.8601217689473675,
|
| 2142 |
-
"kl": 4.70849609375,
|
| 2143 |
-
"learning_rate": 5e-07,
|
| 2144 |
-
"loss": 0.0189,
|
| 2145 |
-
"reward": 0.37148438789881766,
|
| 2146 |
-
"reward_std": 0.39709692588075995,
|
| 2147 |
-
"rewards/end_of_conversation_reward_func": 0.06874999997671694,
|
| 2148 |
-
"rewards/end_rm_reward_func": 0.310546875,
|
| 2149 |
-
"rewards/length_reward_func": -0.0078125,
|
| 2150 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2151 |
-
"step": 171
|
| 2152 |
-
},
|
| 2153 |
-
{
|
| 2154 |
-
"clip_ratio": 0.004716894996818155,
|
| 2155 |
-
"epoch": 0.1376,
|
| 2156 |
-
"grad_norm": 1.1874233704490673,
|
| 2157 |
-
"kl": 17.053955078125,
|
| 2158 |
-
"learning_rate": 5e-07,
|
| 2159 |
-
"loss": 0.017,
|
| 2160 |
-
"step": 172
|
| 2161 |
-
},
|
| 2162 |
-
{
|
| 2163 |
-
"clip_ratio": 0.0,
|
| 2164 |
-
"completion_length": 2176.171875,
|
| 2165 |
-
"epoch": 0.1384,
|
| 2166 |
-
"grad_norm": 2.060042420988078,
|
| 2167 |
-
"kl": 181.342529296875,
|
| 2168 |
-
"learning_rate": 5e-07,
|
| 2169 |
-
"loss": 0.0493,
|
| 2170 |
-
"reward": 0.30000001145526767,
|
| 2171 |
-
"reward_std": 0.35801823483780026,
|
| 2172 |
-
"rewards/end_of_conversation_reward_func": 0.06953125121071935,
|
| 2173 |
-
"rewards/end_rm_reward_func": 0.23046875,
|
| 2174 |
-
"rewards/length_reward_func": 0.0,
|
| 2175 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2176 |
-
"step": 173
|
| 2177 |
-
},
|
| 2178 |
-
{
|
| 2179 |
-
"clip_ratio": 0.012096417922293767,
|
| 2180 |
-
"epoch": 0.1392,
|
| 2181 |
-
"grad_norm": 2.4188316605382605,
|
| 2182 |
-
"kl": 385.28857421875,
|
| 2183 |
-
"learning_rate": 5e-07,
|
| 2184 |
-
"loss": 0.0476,
|
| 2185 |
-
"step": 174
|
| 2186 |
-
},
|
| 2187 |
-
{
|
| 2188 |
-
"clip_ratio": 0.0,
|
| 2189 |
-
"completion_length": 1939.6875,
|
| 2190 |
-
"epoch": 0.14,
|
| 2191 |
-
"grad_norm": 2.2213507537695856,
|
| 2192 |
-
"kl": 412.21533203125,
|
| 2193 |
-
"learning_rate": 5e-07,
|
| 2194 |
-
"loss": 0.0376,
|
| 2195 |
-
"reward": 0.5171875233063474,
|
| 2196 |
-
"reward_std": 0.4006890896707773,
|
| 2197 |
-
"rewards/end_of_conversation_reward_func": 0.0796875013038516,
|
| 2198 |
-
"rewards/end_rm_reward_func": 0.44140625,
|
| 2199 |
-
"rewards/length_reward_func": -0.00390625,
|
| 2200 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2201 |
-
"step": 175
|
| 2202 |
-
},
|
| 2203 |
-
{
|
| 2204 |
-
"clip_ratio": 0.008669883391121402,
|
| 2205 |
-
"epoch": 0.1408,
|
| 2206 |
-
"grad_norm": 2.0154021188215534,
|
| 2207 |
-
"kl": 349.34228515625,
|
| 2208 |
-
"learning_rate": 5e-07,
|
| 2209 |
-
"loss": 0.035,
|
| 2210 |
-
"step": 176
|
| 2211 |
-
},
|
| 2212 |
-
{
|
| 2213 |
-
"clip_ratio": 0.0,
|
| 2214 |
-
"completion_length": 2474.7265625,
|
| 2215 |
-
"epoch": 0.1416,
|
| 2216 |
-
"grad_norm": 2.9429100925925993,
|
| 2217 |
-
"kl": 36.90185546875,
|
| 2218 |
-
"learning_rate": 5e-07,
|
| 2219 |
-
"loss": 0.0301,
|
| 2220 |
-
"reward": 0.47812502505257726,
|
| 2221 |
-
"reward_std": 0.4000142142176628,
|
| 2222 |
-
"rewards/end_of_conversation_reward_func": 0.07578125200234354,
|
| 2223 |
-
"rewards/end_rm_reward_func": 0.4140625,
|
| 2224 |
-
"rewards/length_reward_func": -0.01171875,
|
| 2225 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2226 |
-
"step": 177
|
| 2227 |
-
},
|
| 2228 |
-
{
|
| 2229 |
-
"clip_ratio": 0.007487929397029802,
|
| 2230 |
-
"epoch": 0.1424,
|
| 2231 |
-
"grad_norm": 1.7072132896520176,
|
| 2232 |
-
"kl": 37.833984375,
|
| 2233 |
-
"learning_rate": 5e-07,
|
| 2234 |
-
"loss": 0.0267,
|
| 2235 |
-
"step": 178
|
| 2236 |
-
},
|
| 2237 |
-
{
|
| 2238 |
-
"clip_ratio": 0.0,
|
| 2239 |
-
"completion_length": 2216.28125,
|
| 2240 |
-
"epoch": 0.1432,
|
| 2241 |
-
"grad_norm": 3.4715981881060096,
|
| 2242 |
-
"kl": 153.06494140625,
|
| 2243 |
-
"learning_rate": 5e-07,
|
| 2244 |
-
"loss": 0.0629,
|
| 2245 |
-
"reward": 0.430859393440187,
|
| 2246 |
-
"reward_std": 0.4135680345352739,
|
| 2247 |
-
"rewards/end_of_conversation_reward_func": 0.06562500190921128,
|
| 2248 |
-
"rewards/end_rm_reward_func": 0.369140625,
|
| 2249 |
-
"rewards/length_reward_func": -0.00390625,
|
| 2250 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2251 |
-
"step": 179
|
| 2252 |
-
},
|
| 2253 |
-
{
|
| 2254 |
-
"clip_ratio": 0.007125947595341131,
|
| 2255 |
-
"epoch": 0.144,
|
| 2256 |
-
"grad_norm": 1.7642237159324998,
|
| 2257 |
-
"kl": 372.9677734375,
|
| 2258 |
-
"learning_rate": 5e-07,
|
| 2259 |
-
"loss": 0.0582,
|
| 2260 |
-
"step": 180
|
| 2261 |
-
},
|
| 2262 |
-
{
|
| 2263 |
-
"clip_ratio": 0.0,
|
| 2264 |
-
"completion_length": 2396.046875,
|
| 2265 |
-
"epoch": 0.1448,
|
| 2266 |
-
"grad_norm": 2.626090932602815,
|
| 2267 |
-
"kl": 134.4459228515625,
|
| 2268 |
-
"learning_rate": 5e-07,
|
| 2269 |
-
"loss": 0.0593,
|
| 2270 |
-
"reward": 0.3869140747701749,
|
| 2271 |
-
"reward_std": 0.3422806654125452,
|
| 2272 |
-
"rewards/end_of_conversation_reward_func": 0.06953125167638063,
|
| 2273 |
-
"rewards/end_rm_reward_func": 0.3212890625,
|
| 2274 |
-
"rewards/length_reward_func": -0.00390625,
|
| 2275 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2276 |
-
"step": 181
|
| 2277 |
-
},
|
| 2278 |
-
{
|
| 2279 |
-
"clip_ratio": 0.01481604718719609,
|
| 2280 |
-
"epoch": 0.1456,
|
| 2281 |
-
"grad_norm": 4.174800170660322,
|
| 2282 |
-
"kl": 131.9783935546875,
|
| 2283 |
-
"learning_rate": 5e-07,
|
| 2284 |
-
"loss": 0.0572,
|
| 2285 |
-
"step": 182
|
| 2286 |
-
},
|
| 2287 |
-
{
|
| 2288 |
-
"clip_ratio": 0.0,
|
| 2289 |
-
"completion_length": 2284.40625,
|
| 2290 |
-
"epoch": 0.1464,
|
| 2291 |
-
"grad_norm": 4.005039457486521,
|
| 2292 |
-
"kl": 1621.6591796875,
|
| 2293 |
-
"learning_rate": 5e-07,
|
| 2294 |
-
"loss": 0.0388,
|
| 2295 |
-
"reward": 0.3603515757713467,
|
| 2296 |
-
"reward_std": 0.4121582773514092,
|
| 2297 |
-
"rewards/end_of_conversation_reward_func": 0.07421875139698386,
|
| 2298 |
-
"rewards/end_rm_reward_func": 0.2861328125,
|
| 2299 |
-
"rewards/length_reward_func": 0.0,
|
| 2300 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2301 |
-
"step": 183
|
| 2302 |
-
},
|
| 2303 |
-
{
|
| 2304 |
-
"clip_ratio": 0.009662889962783083,
|
| 2305 |
-
"epoch": 0.1472,
|
| 2306 |
-
"grad_norm": 2.116076197062108,
|
| 2307 |
-
"kl": 1933.222412109375,
|
| 2308 |
-
"learning_rate": 5e-07,
|
| 2309 |
-
"loss": 0.0342,
|
| 2310 |
-
"step": 184
|
| 2311 |
-
},
|
| 2312 |
-
{
|
| 2313 |
-
"clip_ratio": 0.0,
|
| 2314 |
-
"completion_length": 2164.0546875,
|
| 2315 |
-
"epoch": 0.148,
|
| 2316 |
-
"grad_norm": 3.79272536544489,
|
| 2317 |
-
"kl": 8153.509033203125,
|
| 2318 |
-
"learning_rate": 5e-07,
|
| 2319 |
-
"loss": 0.0173,
|
| 2320 |
-
"reward": 0.24707032029982656,
|
| 2321 |
-
"reward_std": 0.33177149668335915,
|
| 2322 |
-
"rewards/end_of_conversation_reward_func": 0.07031250128056854,
|
| 2323 |
-
"rewards/end_rm_reward_func": 0.1767578125,
|
| 2324 |
-
"rewards/length_reward_func": 0.0,
|
| 2325 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2326 |
-
"step": 185
|
| 2327 |
-
},
|
| 2328 |
-
{
|
| 2329 |
-
"clip_ratio": 0.007355456036748365,
|
| 2330 |
-
"epoch": 0.1488,
|
| 2331 |
-
"grad_norm": 2.492559543240363,
|
| 2332 |
-
"kl": 19430.03955078125,
|
| 2333 |
-
"learning_rate": 5e-07,
|
| 2334 |
-
"loss": 0.0143,
|
| 2335 |
-
"step": 186
|
| 2336 |
-
},
|
| 2337 |
-
{
|
| 2338 |
-
"clip_ratio": 0.0,
|
| 2339 |
-
"completion_length": 2088.5234375,
|
| 2340 |
-
"epoch": 0.1496,
|
| 2341 |
-
"grad_norm": 3.215340142433475,
|
| 2342 |
-
"kl": 1088.0703125,
|
| 2343 |
-
"learning_rate": 5e-07,
|
| 2344 |
-
"loss": 0.0347,
|
| 2345 |
-
"reward": 0.32714844518341124,
|
| 2346 |
-
"reward_std": 0.3099647618364543,
|
| 2347 |
-
"rewards/end_of_conversation_reward_func": 0.07812500139698386,
|
| 2348 |
-
"rewards/end_rm_reward_func": 0.2490234375,
|
| 2349 |
-
"rewards/length_reward_func": 0.0,
|
| 2350 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2351 |
-
"step": 187
|
| 2352 |
-
},
|
| 2353 |
-
{
|
| 2354 |
-
"clip_ratio": 0.007153055747039616,
|
| 2355 |
-
"epoch": 0.1504,
|
| 2356 |
-
"grad_norm": 2.979896488060616,
|
| 2357 |
-
"kl": 884.30126953125,
|
| 2358 |
-
"learning_rate": 5e-07,
|
| 2359 |
-
"loss": 0.0304,
|
| 2360 |
-
"step": 188
|
| 2361 |
-
},
|
| 2362 |
-
{
|
| 2363 |
-
"clip_ratio": 0.0,
|
| 2364 |
-
"completion_length": 2295.28125,
|
| 2365 |
-
"epoch": 0.1512,
|
| 2366 |
-
"grad_norm": 4.27184297703827,
|
| 2367 |
-
"kl": 24.62060546875,
|
| 2368 |
-
"learning_rate": 5e-07,
|
| 2369 |
-
"loss": -0.0026,
|
| 2370 |
-
"reward": 0.48769533238373697,
|
| 2371 |
-
"reward_std": 0.4020336586982012,
|
| 2372 |
-
"rewards/end_of_conversation_reward_func": 0.07265625131549314,
|
| 2373 |
-
"rewards/end_rm_reward_func": 0.4150390625,
|
| 2374 |
-
"rewards/length_reward_func": 0.0,
|
| 2375 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2376 |
-
"step": 189
|
| 2377 |
-
},
|
| 2378 |
-
{
|
| 2379 |
-
"clip_ratio": 0.005939892347669229,
|
| 2380 |
-
"epoch": 0.152,
|
| 2381 |
-
"grad_norm": 3.3442192954819,
|
| 2382 |
-
"kl": 17.80908203125,
|
| 2383 |
-
"learning_rate": 5e-07,
|
| 2384 |
-
"loss": -0.0069,
|
| 2385 |
-
"step": 190
|
| 2386 |
-
},
|
| 2387 |
-
{
|
| 2388 |
-
"clip_ratio": 0.0,
|
| 2389 |
-
"completion_length": 2123.578125,
|
| 2390 |
-
"epoch": 0.1528,
|
| 2391 |
-
"grad_norm": 6.018069933911673,
|
| 2392 |
-
"kl": 6821.18798828125,
|
| 2393 |
-
"learning_rate": 5e-07,
|
| 2394 |
-
"loss": 0.0186,
|
| 2395 |
-
"reward": 0.44394533056765795,
|
| 2396 |
-
"reward_std": 0.4120704475790262,
|
| 2397 |
-
"rewards/end_of_conversation_reward_func": 0.07968750083819032,
|
| 2398 |
-
"rewards/end_rm_reward_func": 0.3642578125,
|
| 2399 |
-
"rewards/length_reward_func": 0.0,
|
| 2400 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2401 |
-
"step": 191
|
| 2402 |
-
},
|
| 2403 |
-
{
|
| 2404 |
-
"clip_ratio": 0.019422179379034787,
|
| 2405 |
-
"epoch": 0.1536,
|
| 2406 |
-
"grad_norm": 6.0000850540905555,
|
| 2407 |
-
"kl": 10920.389404296875,
|
| 2408 |
-
"learning_rate": 5e-07,
|
| 2409 |
-
"loss": 0.0159,
|
| 2410 |
-
"step": 192
|
| 2411 |
-
},
|
| 2412 |
-
{
|
| 2413 |
-
"clip_ratio": 0.0,
|
| 2414 |
-
"completion_length": 2315.28125,
|
| 2415 |
-
"epoch": 0.1544,
|
| 2416 |
-
"grad_norm": 5.51028975235637,
|
| 2417 |
-
"kl": 113.30224609375,
|
| 2418 |
-
"learning_rate": 5e-07,
|
| 2419 |
-
"loss": 0.0369,
|
| 2420 |
-
"reward": 0.41445315489545465,
|
| 2421 |
-
"reward_std": 0.40713366749696434,
|
| 2422 |
-
"rewards/end_of_conversation_reward_func": 0.08046875178115442,
|
| 2423 |
-
"rewards/end_rm_reward_func": 0.341796875,
|
| 2424 |
-
"rewards/length_reward_func": -0.0078125,
|
| 2425 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2426 |
-
"step": 193
|
| 2427 |
-
},
|
| 2428 |
-
{
|
| 2429 |
-
"clip_ratio": 0.010416857578093186,
|
| 2430 |
-
"epoch": 0.1552,
|
| 2431 |
-
"grad_norm": 4.641303725834707,
|
| 2432 |
-
"kl": 79.856689453125,
|
| 2433 |
-
"learning_rate": 5e-07,
|
| 2434 |
-
"loss": 0.0318,
|
| 2435 |
-
"step": 194
|
| 2436 |
-
},
|
| 2437 |
-
{
|
| 2438 |
-
"clip_ratio": 0.0,
|
| 2439 |
-
"completion_length": 2337.7265625,
|
| 2440 |
-
"epoch": 0.156,
|
| 2441 |
-
"grad_norm": 7.337511224807291,
|
| 2442 |
-
"kl": 121.304443359375,
|
| 2443 |
-
"learning_rate": 5e-07,
|
| 2444 |
-
"loss": 0.0442,
|
| 2445 |
-
"reward": 0.3572265817783773,
|
| 2446 |
-
"reward_std": 0.44290671218186617,
|
| 2447 |
-
"rewards/end_of_conversation_reward_func": 0.07890625100117177,
|
| 2448 |
-
"rewards/end_rm_reward_func": 0.2822265625,
|
| 2449 |
-
"rewards/length_reward_func": -0.00390625,
|
| 2450 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2451 |
-
"step": 195
|
| 2452 |
-
},
|
| 2453 |
-
{
|
| 2454 |
-
"clip_ratio": 0.01027121691731736,
|
| 2455 |
-
"epoch": 0.1568,
|
| 2456 |
-
"grad_norm": 5.231827919249724,
|
| 2457 |
-
"kl": 134.763427734375,
|
| 2458 |
-
"learning_rate": 5e-07,
|
| 2459 |
-
"loss": 0.0394,
|
| 2460 |
-
"step": 196
|
| 2461 |
-
},
|
| 2462 |
-
{
|
| 2463 |
-
"clip_ratio": 0.0,
|
| 2464 |
-
"completion_length": 2069.0546875,
|
| 2465 |
-
"epoch": 0.1576,
|
| 2466 |
-
"grad_norm": 7.402698135110819,
|
| 2467 |
-
"kl": 52.291015625,
|
| 2468 |
-
"learning_rate": 5e-07,
|
| 2469 |
-
"loss": 0.0311,
|
| 2470 |
-
"reward": 0.38632813934236765,
|
| 2471 |
-
"reward_std": 0.3869206882081926,
|
| 2472 |
-
"rewards/end_of_conversation_reward_func": 0.07187500083819032,
|
| 2473 |
-
"rewards/end_rm_reward_func": 0.318359375,
|
| 2474 |
-
"rewards/length_reward_func": -0.00390625,
|
| 2475 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2476 |
-
"step": 197
|
| 2477 |
-
},
|
| 2478 |
-
{
|
| 2479 |
-
"clip_ratio": 0.006990124413277954,
|
| 2480 |
-
"epoch": 0.1584,
|
| 2481 |
-
"grad_norm": 113.50707154622745,
|
| 2482 |
-
"kl": 21.02978515625,
|
| 2483 |
-
"learning_rate": 5e-07,
|
| 2484 |
-
"loss": 0.0278,
|
| 2485 |
-
"step": 198
|
| 2486 |
-
},
|
| 2487 |
-
{
|
| 2488 |
-
"clip_ratio": 0.0,
|
| 2489 |
-
"completion_length": 2180.6640625,
|
| 2490 |
-
"epoch": 0.1592,
|
| 2491 |
-
"grad_norm": 8.268979237999838,
|
| 2492 |
-
"kl": 146.75,
|
| 2493 |
-
"learning_rate": 5e-07,
|
| 2494 |
-
"loss": 0.0363,
|
| 2495 |
-
"reward": 0.5003906521014869,
|
| 2496 |
-
"reward_std": 0.4240156547166407,
|
| 2497 |
-
"rewards/end_of_conversation_reward_func": 0.0804687519557774,
|
| 2498 |
-
"rewards/end_rm_reward_func": 0.419921875,
|
| 2499 |
-
"rewards/length_reward_func": 0.0,
|
| 2500 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2501 |
-
"step": 199
|
| 2502 |
-
},
|
| 2503 |
-
{
|
| 2504 |
-
"clip_ratio": 0.0194309915532358,
|
| 2505 |
-
"epoch": 0.16,
|
| 2506 |
-
"grad_norm": 8.843678882129485,
|
| 2507 |
-
"kl": 351.8544921875,
|
| 2508 |
-
"learning_rate": 5e-07,
|
| 2509 |
-
"loss": 0.0337,
|
| 2510 |
-
"step": 200
|
| 2511 |
-
},
|
| 2512 |
-
{
|
| 2513 |
-
"clip_ratio": 0.0,
|
| 2514 |
-
"completion_length": 2056.3515625,
|
| 2515 |
-
"epoch": 0.1608,
|
| 2516 |
-
"grad_norm": 6.868955922992306,
|
| 2517 |
-
"kl": 53.767578125,
|
| 2518 |
-
"learning_rate": 5e-07,
|
| 2519 |
-
"loss": 0.036,
|
| 2520 |
-
"reward": 0.45488282945007086,
|
| 2521 |
-
"reward_std": 0.41126722283661366,
|
| 2522 |
-
"rewards/end_of_conversation_reward_func": 0.0867187506519258,
|
| 2523 |
-
"rewards/end_rm_reward_func": 0.3681640625,
|
| 2524 |
-
"rewards/length_reward_func": 0.0,
|
| 2525 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2526 |
-
"step": 201
|
| 2527 |
-
},
|
| 2528 |
-
{
|
| 2529 |
-
"clip_ratio": 0.004734704503789544,
|
| 2530 |
-
"epoch": 0.1616,
|
| 2531 |
-
"grad_norm": 4.719162650271088,
|
| 2532 |
-
"kl": 35.9853515625,
|
| 2533 |
-
"learning_rate": 5e-07,
|
| 2534 |
-
"loss": 0.0269,
|
| 2535 |
-
"step": 202
|
| 2536 |
-
},
|
| 2537 |
-
{
|
| 2538 |
-
"clip_ratio": 0.0,
|
| 2539 |
-
"completion_length": 2449.8671875,
|
| 2540 |
-
"epoch": 0.1624,
|
| 2541 |
-
"grad_norm": 9.848426982264245,
|
| 2542 |
-
"kl": 104.44140625,
|
| 2543 |
-
"learning_rate": 5e-07,
|
| 2544 |
-
"loss": 0.0108,
|
| 2545 |
-
"reward": 0.3212890766444616,
|
| 2546 |
-
"reward_std": 0.34017279790714383,
|
| 2547 |
-
"rewards/end_of_conversation_reward_func": 0.06640625069849193,
|
| 2548 |
-
"rewards/end_rm_reward_func": 0.2587890625,
|
| 2549 |
-
"rewards/length_reward_func": -0.00390625,
|
| 2550 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2551 |
-
"step": 203
|
| 2552 |
-
},
|
| 2553 |
-
{
|
| 2554 |
-
"clip_ratio": 0.012477430194849148,
|
| 2555 |
-
"epoch": 0.1632,
|
| 2556 |
-
"grad_norm": 5.484298870932829,
|
| 2557 |
-
"kl": 146.384765625,
|
| 2558 |
-
"learning_rate": 5e-07,
|
| 2559 |
-
"loss": 0.0006,
|
| 2560 |
-
"step": 204
|
| 2561 |
-
},
|
| 2562 |
-
{
|
| 2563 |
-
"clip_ratio": 0.0,
|
| 2564 |
-
"completion_length": 1957.1015625,
|
| 2565 |
-
"epoch": 0.164,
|
| 2566 |
-
"grad_norm": 11.459849219468296,
|
| 2567 |
-
"kl": 317.099609375,
|
| 2568 |
-
"learning_rate": 5e-07,
|
| 2569 |
-
"loss": 0.0305,
|
| 2570 |
-
"reward": 0.48828127491287887,
|
| 2571 |
-
"reward_std": 0.5096489517018199,
|
| 2572 |
-
"rewards/end_of_conversation_reward_func": 0.07812500046566129,
|
| 2573 |
-
"rewards/end_rm_reward_func": 0.41015625,
|
| 2574 |
-
"rewards/length_reward_func": 0.0,
|
| 2575 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2576 |
-
"step": 205
|
| 2577 |
-
},
|
| 2578 |
-
{
|
| 2579 |
-
"clip_ratio": 0.008903506706701592,
|
| 2580 |
-
"epoch": 0.1648,
|
| 2581 |
-
"grad_norm": 6.73218235011968,
|
| 2582 |
-
"kl": 414.3935546875,
|
| 2583 |
-
"learning_rate": 5e-07,
|
| 2584 |
-
"loss": 0.018,
|
| 2585 |
-
"step": 206
|
| 2586 |
-
},
|
| 2587 |
-
{
|
| 2588 |
-
"clip_ratio": 0.0,
|
| 2589 |
-
"completion_length": 1970.984375,
|
| 2590 |
-
"epoch": 0.1656,
|
| 2591 |
-
"grad_norm": 8.656738486892705,
|
| 2592 |
-
"kl": 246.80859375,
|
| 2593 |
-
"learning_rate": 5e-07,
|
| 2594 |
-
"loss": 0.0585,
|
| 2595 |
-
"reward": 0.5083984585944563,
|
| 2596 |
-
"reward_std": 0.4745393330231309,
|
| 2597 |
-
"rewards/end_of_conversation_reward_func": 0.07578125153668225,
|
| 2598 |
-
"rewards/end_rm_reward_func": 0.4365234375,
|
| 2599 |
-
"rewards/length_reward_func": -0.00390625,
|
| 2600 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2601 |
-
"step": 207
|
| 2602 |
-
},
|
| 2603 |
-
{
|
| 2604 |
-
"clip_ratio": 0.04178383277030662,
|
| 2605 |
-
"epoch": 0.1664,
|
| 2606 |
-
"grad_norm": 20.922629331798802,
|
| 2607 |
-
"kl": 279.943359375,
|
| 2608 |
-
"learning_rate": 5e-07,
|
| 2609 |
-
"loss": 0.0645,
|
| 2610 |
-
"step": 208
|
| 2611 |
-
},
|
| 2612 |
-
{
|
| 2613 |
-
"clip_ratio": 0.0,
|
| 2614 |
-
"completion_length": 1793.3828125,
|
| 2615 |
-
"epoch": 0.1672,
|
| 2616 |
-
"grad_norm": 10.449885777859176,
|
| 2617 |
-
"kl": 294.4990234375,
|
| 2618 |
-
"learning_rate": 5e-07,
|
| 2619 |
-
"loss": 0.0182,
|
| 2620 |
-
"reward": 0.5263672126457095,
|
| 2621 |
-
"reward_std": 0.4001456112600863,
|
| 2622 |
-
"rewards/end_of_conversation_reward_func": 0.08203125139698386,
|
| 2623 |
-
"rewards/end_rm_reward_func": 0.4443359375,
|
| 2624 |
-
"rewards/length_reward_func": 0.0,
|
| 2625 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2626 |
-
"step": 209
|
| 2627 |
-
},
|
| 2628 |
-
{
|
| 2629 |
-
"clip_ratio": 0.025160407996736467,
|
| 2630 |
-
"epoch": 0.168,
|
| 2631 |
-
"grad_norm": 17.92069739644721,
|
| 2632 |
-
"kl": 305.3837890625,
|
| 2633 |
-
"learning_rate": 5e-07,
|
| 2634 |
-
"loss": 0.0173,
|
| 2635 |
-
"step": 210
|
| 2636 |
-
},
|
| 2637 |
-
{
|
| 2638 |
-
"clip_ratio": 0.0,
|
| 2639 |
-
"completion_length": 2268.9609375,
|
| 2640 |
-
"epoch": 0.1688,
|
| 2641 |
-
"grad_norm": 73.96836941822615,
|
| 2642 |
-
"kl": 747.4951171875,
|
| 2643 |
-
"learning_rate": 5e-07,
|
| 2644 |
-
"loss": 0.1343,
|
| 2645 |
-
"reward": 0.36914064013399184,
|
| 2646 |
-
"reward_std": 0.3273033651057631,
|
| 2647 |
-
"rewards/end_of_conversation_reward_func": 0.07812500093132257,
|
| 2648 |
-
"rewards/end_rm_reward_func": 0.291015625,
|
| 2649 |
-
"rewards/length_reward_func": 0.0,
|
| 2650 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2651 |
-
"step": 211
|
| 2652 |
-
},
|
| 2653 |
-
{
|
| 2654 |
-
"clip_ratio": 0.012499632517574355,
|
| 2655 |
-
"epoch": 0.1696,
|
| 2656 |
-
"grad_norm": 18.974576313134882,
|
| 2657 |
-
"kl": 611.49267578125,
|
| 2658 |
-
"learning_rate": 5e-07,
|
| 2659 |
-
"loss": 0.0721,
|
| 2660 |
-
"step": 212
|
| 2661 |
-
},
|
| 2662 |
-
{
|
| 2663 |
-
"clip_ratio": 0.0,
|
| 2664 |
-
"completion_length": 2230.0,
|
| 2665 |
-
"epoch": 0.1704,
|
| 2666 |
-
"grad_norm": 6.681940389271876,
|
| 2667 |
-
"kl": 14805.2890625,
|
| 2668 |
-
"learning_rate": 5e-07,
|
| 2669 |
-
"loss": 0.0276,
|
| 2670 |
-
"reward": 0.3503906426485628,
|
| 2671 |
-
"reward_std": 0.4293831647373736,
|
| 2672 |
-
"rewards/end_of_conversation_reward_func": 0.06718750135041773,
|
| 2673 |
-
"rewards/end_rm_reward_func": 0.283203125,
|
| 2674 |
-
"rewards/length_reward_func": 0.0,
|
| 2675 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2676 |
-
"step": 213
|
| 2677 |
-
},
|
| 2678 |
-
{
|
| 2679 |
-
"clip_ratio": 0.036163119453703985,
|
| 2680 |
-
"epoch": 0.1712,
|
| 2681 |
-
"grad_norm": 20.214531495705373,
|
| 2682 |
-
"kl": 30367.22509765625,
|
| 2683 |
-
"learning_rate": 5e-07,
|
| 2684 |
-
"loss": 0.0295,
|
| 2685 |
-
"step": 214
|
| 2686 |
-
},
|
| 2687 |
-
{
|
| 2688 |
-
"clip_ratio": 0.0,
|
| 2689 |
-
"completion_length": 2377.1953125,
|
| 2690 |
-
"epoch": 0.172,
|
| 2691 |
-
"grad_norm": 7.270360214395911,
|
| 2692 |
-
"kl": 457.0078125,
|
| 2693 |
-
"learning_rate": 5e-07,
|
| 2694 |
-
"loss": 0.0059,
|
| 2695 |
-
"reward": 0.3910156455822289,
|
| 2696 |
-
"reward_std": 0.3999810148961842,
|
| 2697 |
-
"rewards/end_of_conversation_reward_func": 0.08046875172294676,
|
| 2698 |
-
"rewards/end_rm_reward_func": 0.310546875,
|
| 2699 |
-
"rewards/length_reward_func": 0.0,
|
| 2700 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2701 |
-
"step": 215
|
| 2702 |
-
},
|
| 2703 |
-
{
|
| 2704 |
-
"clip_ratio": 0.006911784948897548,
|
| 2705 |
-
"epoch": 0.1728,
|
| 2706 |
-
"grad_norm": 6.317486752283868,
|
| 2707 |
-
"kl": 325.4482421875,
|
| 2708 |
-
"learning_rate": 5e-07,
|
| 2709 |
-
"loss": -0.0004,
|
| 2710 |
-
"step": 216
|
| 2711 |
-
},
|
| 2712 |
-
{
|
| 2713 |
-
"clip_ratio": 0.0,
|
| 2714 |
-
"completion_length": 2922.046875,
|
| 2715 |
-
"epoch": 0.1736,
|
| 2716 |
-
"grad_norm": 6.82709648692313,
|
| 2717 |
-
"kl": 3166.01025390625,
|
| 2718 |
-
"learning_rate": 5e-07,
|
| 2719 |
-
"loss": 0.0348,
|
| 2720 |
-
"reward": 0.27285157423466444,
|
| 2721 |
-
"reward_std": 0.4200313645415008,
|
| 2722 |
-
"rewards/end_of_conversation_reward_func": 0.07265625067520887,
|
| 2723 |
-
"rewards/end_rm_reward_func": 0.2041015625,
|
| 2724 |
-
"rewards/length_reward_func": -0.00390625,
|
| 2725 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2726 |
-
"step": 217
|
| 2727 |
-
},
|
| 2728 |
-
{
|
| 2729 |
-
"clip_ratio": 0.00761759614397306,
|
| 2730 |
-
"epoch": 0.1744,
|
| 2731 |
-
"grad_norm": 6.808138897589715,
|
| 2732 |
-
"kl": 3725.60400390625,
|
| 2733 |
-
"learning_rate": 5e-07,
|
| 2734 |
-
"loss": 0.0255,
|
| 2735 |
-
"step": 218
|
| 2736 |
-
},
|
| 2737 |
-
{
|
| 2738 |
-
"clip_ratio": 0.0,
|
| 2739 |
-
"completion_length": 1952.4453125,
|
| 2740 |
-
"epoch": 0.1752,
|
| 2741 |
-
"grad_norm": 8.424241871323705,
|
| 2742 |
-
"kl": 1067.37109375,
|
| 2743 |
-
"learning_rate": 5e-07,
|
| 2744 |
-
"loss": 0.0463,
|
| 2745 |
-
"reward": 0.34335939423181117,
|
| 2746 |
-
"reward_std": 0.297838720260188,
|
| 2747 |
-
"rewards/end_of_conversation_reward_func": 0.07578125176951289,
|
| 2748 |
-
"rewards/end_rm_reward_func": 0.275390625,
|
| 2749 |
-
"rewards/length_reward_func": -0.0078125,
|
| 2750 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2751 |
-
"step": 219
|
| 2752 |
-
},
|
| 2753 |
-
{
|
| 2754 |
-
"clip_ratio": 0.010116227756952867,
|
| 2755 |
-
"epoch": 0.176,
|
| 2756 |
-
"grad_norm": 5.581257792673027,
|
| 2757 |
-
"kl": 643.158203125,
|
| 2758 |
-
"learning_rate": 5e-07,
|
| 2759 |
-
"loss": 0.0394,
|
| 2760 |
-
"step": 220
|
| 2761 |
-
},
|
| 2762 |
-
{
|
| 2763 |
-
"clip_ratio": 0.0,
|
| 2764 |
-
"completion_length": 2187.0078125,
|
| 2765 |
-
"epoch": 0.1768,
|
| 2766 |
-
"grad_norm": 8.701946640071931,
|
| 2767 |
-
"kl": 44.845703125,
|
| 2768 |
-
"learning_rate": 5e-07,
|
| 2769 |
-
"loss": 0.0026,
|
| 2770 |
-
"reward": 0.36250001611188054,
|
| 2771 |
-
"reward_std": 0.4365540761500597,
|
| 2772 |
-
"rewards/end_of_conversation_reward_func": 0.07734375179279596,
|
| 2773 |
-
"rewards/end_rm_reward_func": 0.28515625,
|
| 2774 |
-
"rewards/length_reward_func": 0.0,
|
| 2775 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2776 |
-
"step": 221
|
| 2777 |
-
},
|
| 2778 |
-
{
|
| 2779 |
-
"clip_ratio": 0.012115928897401318,
|
| 2780 |
-
"epoch": 0.1776,
|
| 2781 |
-
"grad_norm": 6.840735832043148,
|
| 2782 |
-
"kl": 76.4052734375,
|
| 2783 |
-
"learning_rate": 5e-07,
|
| 2784 |
-
"loss": -0.0039,
|
| 2785 |
-
"step": 222
|
| 2786 |
-
},
|
| 2787 |
-
{
|
| 2788 |
-
"clip_ratio": 0.0,
|
| 2789 |
-
"completion_length": 2262.546875,
|
| 2790 |
-
"epoch": 0.1784,
|
| 2791 |
-
"grad_norm": 11.037911784681958,
|
| 2792 |
-
"kl": 392.87890625,
|
| 2793 |
-
"learning_rate": 5e-07,
|
| 2794 |
-
"loss": 0.0408,
|
| 2795 |
-
"reward": 0.3087890681345016,
|
| 2796 |
-
"reward_std": 0.40431320294737816,
|
| 2797 |
-
"rewards/end_of_conversation_reward_func": 0.07343750074505806,
|
| 2798 |
-
"rewards/end_rm_reward_func": 0.2392578125,
|
| 2799 |
-
"rewards/length_reward_func": -0.00390625,
|
| 2800 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2801 |
-
"step": 223
|
| 2802 |
-
},
|
| 2803 |
-
{
|
| 2804 |
-
"clip_ratio": 0.020739365107147023,
|
| 2805 |
-
"epoch": 0.1792,
|
| 2806 |
-
"grad_norm": 10.349410625297242,
|
| 2807 |
-
"kl": 446.98828125,
|
| 2808 |
-
"learning_rate": 5e-07,
|
| 2809 |
-
"loss": 0.0332,
|
| 2810 |
-
"step": 224
|
| 2811 |
-
},
|
| 2812 |
-
{
|
| 2813 |
-
"clip_ratio": 0.0,
|
| 2814 |
-
"completion_length": 2353.828125,
|
| 2815 |
-
"epoch": 0.18,
|
| 2816 |
-
"grad_norm": 7.857636457593067,
|
| 2817 |
-
"kl": 49.89697265625,
|
| 2818 |
-
"learning_rate": 5e-07,
|
| 2819 |
-
"loss": 0.0002,
|
| 2820 |
-
"reward": 0.397070333128795,
|
| 2821 |
-
"reward_std": 0.3746516448445618,
|
| 2822 |
-
"rewards/end_of_conversation_reward_func": 0.08359375083819032,
|
| 2823 |
-
"rewards/end_rm_reward_func": 0.3134765625,
|
| 2824 |
-
"rewards/length_reward_func": 0.0,
|
| 2825 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2826 |
-
"step": 225
|
| 2827 |
-
},
|
| 2828 |
-
{
|
| 2829 |
-
"clip_ratio": 0.04557889769785106,
|
| 2830 |
-
"epoch": 0.1808,
|
| 2831 |
-
"grad_norm": 32.46158867190515,
|
| 2832 |
-
"kl": 143.626953125,
|
| 2833 |
-
"learning_rate": 5e-07,
|
| 2834 |
-
"loss": 0.008,
|
| 2835 |
-
"step": 226
|
| 2836 |
-
},
|
| 2837 |
-
{
|
| 2838 |
-
"clip_ratio": 0.0,
|
| 2839 |
-
"completion_length": 1967.0078125,
|
| 2840 |
-
"epoch": 0.1816,
|
| 2841 |
-
"grad_norm": 7.690726728971795,
|
| 2842 |
-
"kl": 168.8681640625,
|
| 2843 |
-
"learning_rate": 5e-07,
|
| 2844 |
-
"loss": 0.0427,
|
| 2845 |
-
"reward": 0.4964843970956281,
|
| 2846 |
-
"reward_std": 0.4297056058421731,
|
| 2847 |
-
"rewards/end_of_conversation_reward_func": 0.08046875172294676,
|
| 2848 |
-
"rewards/end_rm_reward_func": 0.416015625,
|
| 2849 |
-
"rewards/length_reward_func": 0.0,
|
| 2850 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2851 |
-
"step": 227
|
| 2852 |
-
},
|
| 2853 |
-
{
|
| 2854 |
-
"clip_ratio": 0.020050687453476712,
|
| 2855 |
-
"epoch": 0.1824,
|
| 2856 |
-
"grad_norm": 12.623134353539193,
|
| 2857 |
-
"kl": 229.0390625,
|
| 2858 |
-
"learning_rate": 5e-07,
|
| 2859 |
-
"loss": 0.0382,
|
| 2860 |
-
"step": 228
|
| 2861 |
-
},
|
| 2862 |
-
{
|
| 2863 |
-
"clip_ratio": 0.0,
|
| 2864 |
-
"completion_length": 2303.3203125,
|
| 2865 |
-
"epoch": 0.1832,
|
| 2866 |
-
"grad_norm": 8.985778261923576,
|
| 2867 |
-
"kl": 1521.31640625,
|
| 2868 |
-
"learning_rate": 5e-07,
|
| 2869 |
-
"loss": 0.0609,
|
| 2870 |
-
"reward": 0.34335938887670636,
|
| 2871 |
-
"reward_std": 0.4365329034626484,
|
| 2872 |
-
"rewards/end_of_conversation_reward_func": 0.07187500095460564,
|
| 2873 |
-
"rewards/end_rm_reward_func": 0.271484375,
|
| 2874 |
-
"rewards/length_reward_func": 0.0,
|
| 2875 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2876 |
-
"step": 229
|
| 2877 |
-
},
|
| 2878 |
-
{
|
| 2879 |
-
"clip_ratio": 0.046887944103218615,
|
| 2880 |
-
"epoch": 0.184,
|
| 2881 |
-
"grad_norm": 34.903609321208826,
|
| 2882 |
-
"kl": 1591.25390625,
|
| 2883 |
-
"learning_rate": 5e-07,
|
| 2884 |
-
"loss": 0.0655,
|
| 2885 |
-
"step": 230
|
| 2886 |
-
},
|
| 2887 |
-
{
|
| 2888 |
-
"clip_ratio": 0.0,
|
| 2889 |
-
"completion_length": 2103.6015625,
|
| 2890 |
-
"epoch": 0.1848,
|
| 2891 |
-
"grad_norm": 7.8152371982851045,
|
| 2892 |
-
"kl": 93.3984375,
|
| 2893 |
-
"learning_rate": 5e-07,
|
| 2894 |
-
"loss": 0.021,
|
| 2895 |
-
"reward": 0.41718751774169505,
|
| 2896 |
-
"reward_std": 0.4234715709462762,
|
| 2897 |
-
"rewards/end_of_conversation_reward_func": 0.07734375167638063,
|
| 2898 |
-
"rewards/end_rm_reward_func": 0.33984375,
|
| 2899 |
-
"rewards/length_reward_func": 0.0,
|
| 2900 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2901 |
-
"step": 231
|
| 2902 |
-
},
|
| 2903 |
-
{
|
| 2904 |
-
"clip_ratio": 0.01693671362590976,
|
| 2905 |
-
"epoch": 0.1856,
|
| 2906 |
-
"grad_norm": 15.100433483062815,
|
| 2907 |
-
"kl": 119.0732421875,
|
| 2908 |
-
"learning_rate": 5e-07,
|
| 2909 |
-
"loss": 0.0166,
|
| 2910 |
-
"step": 232
|
| 2911 |
-
},
|
| 2912 |
-
{
|
| 2913 |
-
"clip_ratio": 0.0,
|
| 2914 |
-
"completion_length": 2187.609375,
|
| 2915 |
-
"epoch": 0.1864,
|
| 2916 |
-
"grad_norm": 10.645873330116869,
|
| 2917 |
-
"kl": 176.8994140625,
|
| 2918 |
-
"learning_rate": 5e-07,
|
| 2919 |
-
"loss": 0.0302,
|
| 2920 |
-
"reward": 0.39726564241573215,
|
| 2921 |
-
"reward_std": 0.3991687847301364,
|
| 2922 |
-
"rewards/end_of_conversation_reward_func": 0.07109375053551048,
|
| 2923 |
-
"rewards/end_rm_reward_func": 0.333984375,
|
| 2924 |
-
"rewards/length_reward_func": -0.0078125,
|
| 2925 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2926 |
-
"step": 233
|
| 2927 |
-
},
|
| 2928 |
-
{
|
| 2929 |
-
"clip_ratio": 0.009701315342681482,
|
| 2930 |
-
"epoch": 0.1872,
|
| 2931 |
-
"grad_norm": 9.681021841986935,
|
| 2932 |
-
"kl": 173.751953125,
|
| 2933 |
-
"learning_rate": 5e-07,
|
| 2934 |
-
"loss": 0.0213,
|
| 2935 |
-
"step": 234
|
| 2936 |
-
},
|
| 2937 |
-
{
|
| 2938 |
-
"clip_ratio": 0.0,
|
| 2939 |
-
"completion_length": 2018.2421875,
|
| 2940 |
-
"epoch": 0.188,
|
| 2941 |
-
"grad_norm": 9.837481125291612,
|
| 2942 |
-
"kl": 250.2099609375,
|
| 2943 |
-
"learning_rate": 5e-07,
|
| 2944 |
-
"loss": 0.0135,
|
| 2945 |
-
"reward": 0.3800781453028321,
|
| 2946 |
-
"reward_std": 0.302262753713876,
|
| 2947 |
-
"rewards/end_of_conversation_reward_func": 0.08125000144354999,
|
| 2948 |
-
"rewards/end_rm_reward_func": 0.306640625,
|
| 2949 |
-
"rewards/length_reward_func": -0.0078125,
|
| 2950 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2951 |
-
"step": 235
|
| 2952 |
-
},
|
| 2953 |
-
{
|
| 2954 |
-
"clip_ratio": 0.02462595998076722,
|
| 2955 |
-
"epoch": 0.1888,
|
| 2956 |
-
"grad_norm": 18.237167770716425,
|
| 2957 |
-
"kl": 475.810546875,
|
| 2958 |
-
"learning_rate": 5e-07,
|
| 2959 |
-
"loss": 0.0139,
|
| 2960 |
-
"step": 236
|
| 2961 |
-
},
|
| 2962 |
-
{
|
| 2963 |
-
"clip_ratio": 0.0,
|
| 2964 |
-
"completion_length": 2196.796875,
|
| 2965 |
-
"epoch": 0.1896,
|
| 2966 |
-
"grad_norm": 10.707362677641337,
|
| 2967 |
-
"kl": 2000.958984375,
|
| 2968 |
-
"learning_rate": 5e-07,
|
| 2969 |
-
"loss": 0.0229,
|
| 2970 |
-
"reward": 0.38046876690350473,
|
| 2971 |
-
"reward_std": 0.30730303283780813,
|
| 2972 |
-
"rewards/end_of_conversation_reward_func": 0.07578125095460564,
|
| 2973 |
-
"rewards/end_rm_reward_func": 0.3046875,
|
| 2974 |
-
"rewards/length_reward_func": 0.0,
|
| 2975 |
-
"rewards/thinking_reward_func": 0.0,
|
| 2976 |
-
"step": 237
|
| 2977 |
-
},
|
| 2978 |
-
{
|
| 2979 |
-
"clip_ratio": 0.00941651189350523,
|
| 2980 |
-
"epoch": 0.1904,
|
| 2981 |
-
"grad_norm": 7.341564288985746,
|
| 2982 |
-
"kl": 3464.6708984375,
|
| 2983 |
-
"learning_rate": 5e-07,
|
| 2984 |
-
"loss": 0.0123,
|
| 2985 |
-
"step": 238
|
| 2986 |
-
},
|
| 2987 |
-
{
|
| 2988 |
-
"clip_ratio": 0.0,
|
| 2989 |
-
"completion_length": 1952.6328125,
|
| 2990 |
-
"epoch": 0.1912,
|
| 2991 |
-
"grad_norm": 9.746913854316412,
|
| 2992 |
-
"kl": 106.2119140625,
|
| 2993 |
-
"learning_rate": 5e-07,
|
| 2994 |
-
"loss": 0.0185,
|
| 2995 |
-
"reward": 0.37890626420266926,
|
| 2996 |
-
"reward_std": 0.44265066692605615,
|
| 2997 |
-
"rewards/end_of_conversation_reward_func": 0.07031250128056854,
|
| 2998 |
-
"rewards/end_rm_reward_func": 0.30859375,
|
| 2999 |
-
"rewards/length_reward_func": 0.0,
|
| 3000 |
-
"rewards/thinking_reward_func": 0.0,
|
| 3001 |
-
"step": 239
|
| 3002 |
-
},
|
| 3003 |
-
{
|
| 3004 |
-
"clip_ratio": 0.01394331140909344,
|
| 3005 |
-
"epoch": 0.192,
|
| 3006 |
-
"grad_norm": 10.146789021120865,
|
| 3007 |
-
"kl": 102.3330078125,
|
| 3008 |
-
"learning_rate": 5e-07,
|
| 3009 |
-
"loss": 0.0135,
|
| 3010 |
-
"step": 240
|
| 3011 |
-
}
|
| 3012 |
-
],
|
| 3013 |
-
"logging_steps": 1,
|
| 3014 |
-
"max_steps": 1250,
|
| 3015 |
-
"num_input_tokens_seen": 0,
|
| 3016 |
-
"num_train_epochs": 1,
|
| 3017 |
-
"save_steps": 10,
|
| 3018 |
-
"stateful_callbacks": {
|
| 3019 |
-
"TrainerControl": {
|
| 3020 |
-
"args": {
|
| 3021 |
-
"should_epoch_stop": false,
|
| 3022 |
-
"should_evaluate": false,
|
| 3023 |
-
"should_log": false,
|
| 3024 |
-
"should_save": true,
|
| 3025 |
-
"should_training_stop": false
|
| 3026 |
-
},
|
| 3027 |
-
"attributes": {}
|
| 3028 |
-
}
|
| 3029 |
-
},
|
| 3030 |
-
"total_flos": 0.0,
|
| 3031 |
-
"train_batch_size": 32,
|
| 3032 |
-
"trial_name": null,
|
| 3033 |
-
"trial_params": null
|
| 3034 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|