File size: 13,551 Bytes
2f6419b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 |
{
"best_metric": 0.4326101839542389,
"best_model_checkpoint": "./mistral/19-04-24-Weni-WeniGPT-Agents-Mistral-1.0.6-SFT-1.0.4-DPO_Experiment on DPO with other hyperparameters and best SFT model of WeniGPT-2_max_steps-732_batch_4_2024-04-19_ppid_9/checkpoint-180",
"epoch": 2.926829268292683,
"eval_steps": 30,
"global_step": 180,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.16,
"grad_norm": 38.021175384521484,
"learning_rate": 1.590909090909091e-06,
"logits/chosen": -1.7808929681777954,
"logits/rejected": -1.8136215209960938,
"logps/chosen": -52.2663688659668,
"logps/rejected": -88.76155853271484,
"loss": 0.6927,
"rewards/accuracies": 0.20000000298023224,
"rewards/chosen": 0.0008833981119096279,
"rewards/margins": 0.0009520913590677083,
"rewards/rejected": -6.869318895041943e-05,
"step": 10
},
{
"epoch": 0.33,
"grad_norm": 19.505126953125,
"learning_rate": 3.863636363636364e-06,
"logits/chosen": -1.865220308303833,
"logits/rejected": -1.880090355873108,
"logps/chosen": -34.3599967956543,
"logps/rejected": -53.656646728515625,
"loss": 0.6846,
"rewards/accuracies": 0.22499999403953552,
"rewards/chosen": 0.019346294924616814,
"rewards/margins": 0.017682768404483795,
"rewards/rejected": 0.0016635276842862368,
"step": 20
},
{
"epoch": 0.49,
"grad_norm": 16.67194938659668,
"learning_rate": 4.964788732394366e-06,
"logits/chosen": -1.7692855596542358,
"logits/rejected": -1.7932851314544678,
"logps/chosen": -38.679237365722656,
"logps/rejected": -79.23713684082031,
"loss": 0.665,
"rewards/accuracies": 0.375,
"rewards/chosen": 0.07158979028463364,
"rewards/margins": 0.061105482280254364,
"rewards/rejected": 0.010484304279088974,
"step": 30
},
{
"epoch": 0.49,
"eval_logits/chosen": -1.7801028490066528,
"eval_logits/rejected": -1.7981160879135132,
"eval_logps/chosen": -54.71597671508789,
"eval_logps/rejected": -95.75775909423828,
"eval_loss": 0.6322354674339294,
"eval_rewards/accuracies": 0.4285714328289032,
"eval_rewards/chosen": 0.14569346606731415,
"eval_rewards/margins": 0.13385750353336334,
"eval_rewards/rejected": 0.011835969984531403,
"eval_runtime": 8.5124,
"eval_samples_per_second": 3.289,
"eval_steps_per_second": 1.645,
"step": 30
},
{
"epoch": 0.65,
"grad_norm": 22.69721031188965,
"learning_rate": 4.894366197183099e-06,
"logits/chosen": -1.8458446264266968,
"logits/rejected": -1.8594070672988892,
"logps/chosen": -39.180992126464844,
"logps/rejected": -57.04741287231445,
"loss": 0.6164,
"rewards/accuracies": 0.3499999940395355,
"rewards/chosen": 0.19042351841926575,
"rewards/margins": 0.18086455762386322,
"rewards/rejected": 0.009558964520692825,
"step": 40
},
{
"epoch": 0.81,
"grad_norm": 0.0,
"learning_rate": 4.823943661971832e-06,
"logits/chosen": -1.8411815166473389,
"logits/rejected": -1.8659855127334595,
"logps/chosen": -31.194744110107422,
"logps/rejected": -58.17326736450195,
"loss": 0.6138,
"rewards/accuracies": 0.30000001192092896,
"rewards/chosen": 0.19823171198368073,
"rewards/margins": 0.19669394195079803,
"rewards/rejected": 0.0015377718955278397,
"step": 50
},
{
"epoch": 0.98,
"grad_norm": 10.77029800415039,
"learning_rate": 4.753521126760564e-06,
"logits/chosen": -1.816936731338501,
"logits/rejected": -1.8421777486801147,
"logps/chosen": -50.517478942871094,
"logps/rejected": -80.6562728881836,
"loss": 0.5445,
"rewards/accuracies": 0.42500001192092896,
"rewards/chosen": 0.4567543864250183,
"rewards/margins": 0.42180705070495605,
"rewards/rejected": 0.03494739532470703,
"step": 60
},
{
"epoch": 0.98,
"eval_logits/chosen": -1.784444808959961,
"eval_logits/rejected": -1.8030898571014404,
"eval_logps/chosen": -53.55432891845703,
"eval_logps/rejected": -95.73511505126953,
"eval_loss": 0.527186930179596,
"eval_rewards/accuracies": 0.4285714328289032,
"eval_rewards/chosen": 0.4941868484020233,
"eval_rewards/margins": 0.47555941343307495,
"eval_rewards/rejected": 0.018627464771270752,
"eval_runtime": 8.5149,
"eval_samples_per_second": 3.288,
"eval_steps_per_second": 1.644,
"step": 60
},
{
"epoch": 1.14,
"grad_norm": 8.583382606506348,
"learning_rate": 4.683098591549296e-06,
"logits/chosen": -1.7579460144042969,
"logits/rejected": -1.776581048965454,
"logps/chosen": -54.420066833496094,
"logps/rejected": -79.0608901977539,
"loss": 0.451,
"rewards/accuracies": 0.4749999940395355,
"rewards/chosen": 0.8353117108345032,
"rewards/margins": 0.8680375814437866,
"rewards/rejected": -0.032725896686315536,
"step": 70
},
{
"epoch": 1.3,
"grad_norm": 2.3610572814941406,
"learning_rate": 4.612676056338028e-06,
"logits/chosen": -1.8833833932876587,
"logits/rejected": -1.9092212915420532,
"logps/chosen": -25.5800724029541,
"logps/rejected": -64.96099853515625,
"loss": 0.5302,
"rewards/accuracies": 0.30000001192092896,
"rewards/chosen": 0.5431563258171082,
"rewards/margins": 0.6763826012611389,
"rewards/rejected": -0.13322624564170837,
"step": 80
},
{
"epoch": 1.46,
"grad_norm": 3.739367723464966,
"learning_rate": 4.542253521126761e-06,
"logits/chosen": -1.8194096088409424,
"logits/rejected": -1.8343127965927124,
"logps/chosen": -41.80801773071289,
"logps/rejected": -63.005592346191406,
"loss": 0.4616,
"rewards/accuracies": 0.4749999940395355,
"rewards/chosen": 0.830731213092804,
"rewards/margins": 0.9403120875358582,
"rewards/rejected": -0.109580859541893,
"step": 90
},
{
"epoch": 1.46,
"eval_logits/chosen": -1.7902206182479858,
"eval_logits/rejected": -1.8099161386489868,
"eval_logps/chosen": -52.01105880737305,
"eval_logps/rejected": -95.83198547363281,
"eval_loss": 0.46216800808906555,
"eval_rewards/accuracies": 0.4285714328289032,
"eval_rewards/chosen": 0.957168459892273,
"eval_rewards/margins": 0.9676021337509155,
"eval_rewards/rejected": -0.010433738119900227,
"eval_runtime": 8.5144,
"eval_samples_per_second": 3.289,
"eval_steps_per_second": 1.644,
"step": 90
},
{
"epoch": 1.63,
"grad_norm": 0.5382364988327026,
"learning_rate": 4.471830985915494e-06,
"logits/chosen": -1.845868468284607,
"logits/rejected": -1.8715919256210327,
"logps/chosen": -40.285335540771484,
"logps/rejected": -73.16102600097656,
"loss": 0.4611,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": 1.0631928443908691,
"rewards/margins": 1.3067411184310913,
"rewards/rejected": -0.24354824423789978,
"step": 100
},
{
"epoch": 1.79,
"grad_norm": 12.538246154785156,
"learning_rate": 4.401408450704226e-06,
"logits/chosen": -1.8342373371124268,
"logits/rejected": -1.862433671951294,
"logps/chosen": -29.438589096069336,
"logps/rejected": -62.31974411010742,
"loss": 0.5266,
"rewards/accuracies": 0.30000001192092896,
"rewards/chosen": 0.7012811899185181,
"rewards/margins": 0.8140069246292114,
"rewards/rejected": -0.11272567510604858,
"step": 110
},
{
"epoch": 1.95,
"grad_norm": 2.3604652881622314,
"learning_rate": 4.3309859154929575e-06,
"logits/chosen": -1.8481504917144775,
"logits/rejected": -1.8528194427490234,
"logps/chosen": -35.794334411621094,
"logps/rejected": -56.22735595703125,
"loss": 0.5327,
"rewards/accuracies": 0.32499998807907104,
"rewards/chosen": 1.078013300895691,
"rewards/margins": 0.9431617856025696,
"rewards/rejected": 0.1348516196012497,
"step": 120
},
{
"epoch": 1.95,
"eval_logits/chosen": -1.7962983846664429,
"eval_logits/rejected": -1.8171442747116089,
"eval_logps/chosen": -50.74287796020508,
"eval_logps/rejected": -96.0649185180664,
"eval_loss": 0.4424423575401306,
"eval_rewards/accuracies": 0.4285714328289032,
"eval_rewards/chosen": 1.337624192237854,
"eval_rewards/margins": 1.4179356098175049,
"eval_rewards/rejected": -0.08031141757965088,
"eval_runtime": 8.5188,
"eval_samples_per_second": 3.287,
"eval_steps_per_second": 1.643,
"step": 120
},
{
"epoch": 2.11,
"grad_norm": 5.3829216957092285,
"learning_rate": 4.26056338028169e-06,
"logits/chosen": -1.8153988122940063,
"logits/rejected": -1.853118658065796,
"logps/chosen": -44.251441955566406,
"logps/rejected": -85.38258361816406,
"loss": 0.4011,
"rewards/accuracies": 0.4749999940395355,
"rewards/chosen": 1.5001299381256104,
"rewards/margins": 1.9703232049942017,
"rewards/rejected": -0.4701933264732361,
"step": 130
},
{
"epoch": 2.28,
"grad_norm": 0.6282079815864563,
"learning_rate": 4.190140845070423e-06,
"logits/chosen": -1.8527545928955078,
"logits/rejected": -1.8757705688476562,
"logps/chosen": -35.6223258972168,
"logps/rejected": -80.99830627441406,
"loss": 0.4423,
"rewards/accuracies": 0.375,
"rewards/chosen": 1.484961748123169,
"rewards/margins": 1.8964126110076904,
"rewards/rejected": -0.41145071387290955,
"step": 140
},
{
"epoch": 2.44,
"grad_norm": 0.0,
"learning_rate": 4.119718309859156e-06,
"logits/chosen": -1.9129869937896729,
"logits/rejected": -1.927869439125061,
"logps/chosen": -24.400936126708984,
"logps/rejected": -50.861839294433594,
"loss": 0.5459,
"rewards/accuracies": 0.25,
"rewards/chosen": 0.8408910036087036,
"rewards/margins": 1.1376488208770752,
"rewards/rejected": -0.2967577874660492,
"step": 150
},
{
"epoch": 2.44,
"eval_logits/chosen": -1.8025389909744263,
"eval_logits/rejected": -1.824258804321289,
"eval_logps/chosen": -49.72334671020508,
"eval_logps/rejected": -96.41252899169922,
"eval_loss": 0.4335039258003235,
"eval_rewards/accuracies": 0.4285714328289032,
"eval_rewards/chosen": 1.6434847116470337,
"eval_rewards/margins": 1.828076720237732,
"eval_rewards/rejected": -0.1845920830965042,
"eval_runtime": 8.5149,
"eval_samples_per_second": 3.288,
"eval_steps_per_second": 1.644,
"step": 150
},
{
"epoch": 2.6,
"grad_norm": 0.7459288239479065,
"learning_rate": 4.0492957746478875e-06,
"logits/chosen": -1.786231279373169,
"logits/rejected": -1.8223682641983032,
"logps/chosen": -40.68749237060547,
"logps/rejected": -78.1619873046875,
"loss": 0.4348,
"rewards/accuracies": 0.44999998807907104,
"rewards/chosen": 1.9040056467056274,
"rewards/margins": 2.482433795928955,
"rewards/rejected": -0.5784280896186829,
"step": 160
},
{
"epoch": 2.76,
"grad_norm": 1.4669339656829834,
"learning_rate": 3.97887323943662e-06,
"logits/chosen": -1.8449195623397827,
"logits/rejected": -1.871626615524292,
"logps/chosen": -46.335182189941406,
"logps/rejected": -86.10281372070312,
"loss": 0.4232,
"rewards/accuracies": 0.4000000059604645,
"rewards/chosen": 1.8989893198013306,
"rewards/margins": 2.384274482727051,
"rewards/rejected": -0.4852851927280426,
"step": 170
},
{
"epoch": 2.93,
"grad_norm": 0.0301869735121727,
"learning_rate": 3.908450704225352e-06,
"logits/chosen": -1.8375946283340454,
"logits/rejected": -1.8527040481567383,
"logps/chosen": -39.01136016845703,
"logps/rejected": -65.78593444824219,
"loss": 0.4055,
"rewards/accuracies": 0.42500001192092896,
"rewards/chosen": 1.9841420650482178,
"rewards/margins": 2.3045599460601807,
"rewards/rejected": -0.32041770219802856,
"step": 180
},
{
"epoch": 2.93,
"eval_logits/chosen": -1.8074016571044922,
"eval_logits/rejected": -1.830051064491272,
"eval_logps/chosen": -48.993629455566406,
"eval_logps/rejected": -96.92731475830078,
"eval_loss": 0.4326101839542389,
"eval_rewards/accuracies": 0.4285714328289032,
"eval_rewards/chosen": 1.8623981475830078,
"eval_rewards/margins": 2.201425075531006,
"eval_rewards/rejected": -0.3390272557735443,
"eval_runtime": 8.5144,
"eval_samples_per_second": 3.289,
"eval_steps_per_second": 1.644,
"step": 180
}
],
"logging_steps": 10,
"max_steps": 732,
"num_input_tokens_seen": 0,
"num_train_epochs": 12,
"save_steps": 90,
"total_flos": 0.0,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}
|