Delete trainer_state.json
Browse files- trainer_state.json +0 -1248
trainer_state.json
DELETED
|
@@ -1,1248 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": null,
|
| 3 |
-
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.8063872255489022,
|
| 5 |
-
"eval_steps": 500,
|
| 6 |
-
"global_step": 404,
|
| 7 |
-
"is_hyper_param_search": false,
|
| 8 |
-
"is_local_process_zero": true,
|
| 9 |
-
"is_world_process_zero": true,
|
| 10 |
-
"log_history": [
|
| 11 |
-
{
|
| 12 |
-
"epoch": 0.001996007984031936,
|
| 13 |
-
"grad_norm": 29.364113674215965,
|
| 14 |
-
"learning_rate": 1.5686274509803922e-08,
|
| 15 |
-
"logits/chosen": -1.6221545934677124,
|
| 16 |
-
"logits/rejected": -1.6521912813186646,
|
| 17 |
-
"logps/chosen": -0.940773606300354,
|
| 18 |
-
"logps/rejected": -0.851032555103302,
|
| 19 |
-
"loss": 0.6931,
|
| 20 |
-
"rewards/accuracies": 0.0,
|
| 21 |
-
"rewards/chosen": 0.0,
|
| 22 |
-
"rewards/margins": 0.0,
|
| 23 |
-
"rewards/rejected": 0.0,
|
| 24 |
-
"step": 1
|
| 25 |
-
},
|
| 26 |
-
{
|
| 27 |
-
"epoch": 0.00998003992015968,
|
| 28 |
-
"grad_norm": 31.405892041917173,
|
| 29 |
-
"learning_rate": 7.843137254901961e-08,
|
| 30 |
-
"logits/chosen": -1.5107966661453247,
|
| 31 |
-
"logits/rejected": -1.3356468677520752,
|
| 32 |
-
"logps/chosen": -0.7684795260429382,
|
| 33 |
-
"logps/rejected": -0.8418906331062317,
|
| 34 |
-
"loss": 0.692,
|
| 35 |
-
"rewards/accuracies": 0.515625,
|
| 36 |
-
"rewards/chosen": 0.0024527055211365223,
|
| 37 |
-
"rewards/margins": 0.004762639291584492,
|
| 38 |
-
"rewards/rejected": -0.0023099337704479694,
|
| 39 |
-
"step": 5
|
| 40 |
-
},
|
| 41 |
-
{
|
| 42 |
-
"epoch": 0.01996007984031936,
|
| 43 |
-
"grad_norm": 32.33090024343072,
|
| 44 |
-
"learning_rate": 1.5686274509803921e-07,
|
| 45 |
-
"logits/chosen": -1.4846872091293335,
|
| 46 |
-
"logits/rejected": -1.3766502141952515,
|
| 47 |
-
"logps/chosen": -0.8155986070632935,
|
| 48 |
-
"logps/rejected": -0.9123713374137878,
|
| 49 |
-
"loss": 0.6918,
|
| 50 |
-
"rewards/accuracies": 0.550000011920929,
|
| 51 |
-
"rewards/chosen": -0.003020394593477249,
|
| 52 |
-
"rewards/margins": 0.00013434886932373047,
|
| 53 |
-
"rewards/rejected": -0.0031547434628009796,
|
| 54 |
-
"step": 10
|
| 55 |
-
},
|
| 56 |
-
{
|
| 57 |
-
"epoch": 0.029940119760479042,
|
| 58 |
-
"grad_norm": 28.485164448197256,
|
| 59 |
-
"learning_rate": 2.352941176470588e-07,
|
| 60 |
-
"logits/chosen": -1.5661466121673584,
|
| 61 |
-
"logits/rejected": -1.494202971458435,
|
| 62 |
-
"logps/chosen": -0.9453551173210144,
|
| 63 |
-
"logps/rejected": -1.0322425365447998,
|
| 64 |
-
"loss": 0.6913,
|
| 65 |
-
"rewards/accuracies": 0.4375,
|
| 66 |
-
"rewards/chosen": -0.0035083256661891937,
|
| 67 |
-
"rewards/margins": -0.0018780697137117386,
|
| 68 |
-
"rewards/rejected": -0.0016302559524774551,
|
| 69 |
-
"step": 15
|
| 70 |
-
},
|
| 71 |
-
{
|
| 72 |
-
"epoch": 0.03992015968063872,
|
| 73 |
-
"grad_norm": 38.07958741824287,
|
| 74 |
-
"learning_rate": 3.1372549019607843e-07,
|
| 75 |
-
"logits/chosen": -1.5159441232681274,
|
| 76 |
-
"logits/rejected": -1.506949782371521,
|
| 77 |
-
"logps/chosen": -1.148013710975647,
|
| 78 |
-
"logps/rejected": -1.2202709913253784,
|
| 79 |
-
"loss": 0.6887,
|
| 80 |
-
"rewards/accuracies": 0.4625000059604645,
|
| 81 |
-
"rewards/chosen": -0.04793555662035942,
|
| 82 |
-
"rewards/margins": 0.017127588391304016,
|
| 83 |
-
"rewards/rejected": -0.06506314128637314,
|
| 84 |
-
"step": 20
|
| 85 |
-
},
|
| 86 |
-
{
|
| 87 |
-
"epoch": 0.0499001996007984,
|
| 88 |
-
"grad_norm": 27.572568095419346,
|
| 89 |
-
"learning_rate": 3.92156862745098e-07,
|
| 90 |
-
"logits/chosen": -1.4518249034881592,
|
| 91 |
-
"logits/rejected": -1.3797534704208374,
|
| 92 |
-
"logps/chosen": -0.9855057001113892,
|
| 93 |
-
"logps/rejected": -1.0200985670089722,
|
| 94 |
-
"loss": 0.691,
|
| 95 |
-
"rewards/accuracies": 0.5625,
|
| 96 |
-
"rewards/chosen": -0.07226995378732681,
|
| 97 |
-
"rewards/margins": 0.020743293687701225,
|
| 98 |
-
"rewards/rejected": -0.09301324933767319,
|
| 99 |
-
"step": 25
|
| 100 |
-
},
|
| 101 |
-
{
|
| 102 |
-
"epoch": 0.059880239520958084,
|
| 103 |
-
"grad_norm": 31.06009539067789,
|
| 104 |
-
"learning_rate": 4.705882352941176e-07,
|
| 105 |
-
"logits/chosen": -1.499815583229065,
|
| 106 |
-
"logits/rejected": -1.5191256999969482,
|
| 107 |
-
"logps/chosen": -0.9442195892333984,
|
| 108 |
-
"logps/rejected": -1.075582504272461,
|
| 109 |
-
"loss": 0.6833,
|
| 110 |
-
"rewards/accuracies": 0.612500011920929,
|
| 111 |
-
"rewards/chosen": -0.08427385240793228,
|
| 112 |
-
"rewards/margins": 0.05866638571023941,
|
| 113 |
-
"rewards/rejected": -0.14294025301933289,
|
| 114 |
-
"step": 30
|
| 115 |
-
},
|
| 116 |
-
{
|
| 117 |
-
"epoch": 0.06986027944111776,
|
| 118 |
-
"grad_norm": 30.966967781625176,
|
| 119 |
-
"learning_rate": 5.490196078431373e-07,
|
| 120 |
-
"logits/chosen": -1.4179936647415161,
|
| 121 |
-
"logits/rejected": -1.3696165084838867,
|
| 122 |
-
"logps/chosen": -0.8713253140449524,
|
| 123 |
-
"logps/rejected": -0.910132884979248,
|
| 124 |
-
"loss": 0.6843,
|
| 125 |
-
"rewards/accuracies": 0.7124999761581421,
|
| 126 |
-
"rewards/chosen": -0.03732147440314293,
|
| 127 |
-
"rewards/margins": 0.025124484673142433,
|
| 128 |
-
"rewards/rejected": -0.06244595721364021,
|
| 129 |
-
"step": 35
|
| 130 |
-
},
|
| 131 |
-
{
|
| 132 |
-
"epoch": 0.07984031936127745,
|
| 133 |
-
"grad_norm": 24.405429771065002,
|
| 134 |
-
"learning_rate": 6.274509803921569e-07,
|
| 135 |
-
"logits/chosen": -1.500732660293579,
|
| 136 |
-
"logits/rejected": -1.4325664043426514,
|
| 137 |
-
"logps/chosen": -0.9898351430892944,
|
| 138 |
-
"logps/rejected": -1.0182645320892334,
|
| 139 |
-
"loss": 0.6711,
|
| 140 |
-
"rewards/accuracies": 0.625,
|
| 141 |
-
"rewards/chosen": 0.010954104363918304,
|
| 142 |
-
"rewards/margins": 0.05013161152601242,
|
| 143 |
-
"rewards/rejected": -0.039177510887384415,
|
| 144 |
-
"step": 40
|
| 145 |
-
},
|
| 146 |
-
{
|
| 147 |
-
"epoch": 0.08982035928143713,
|
| 148 |
-
"grad_norm": 58.76778208087395,
|
| 149 |
-
"learning_rate": 7.058823529411765e-07,
|
| 150 |
-
"logits/chosen": -1.6222164630889893,
|
| 151 |
-
"logits/rejected": -1.4863214492797852,
|
| 152 |
-
"logps/chosen": -0.9788524508476257,
|
| 153 |
-
"logps/rejected": -1.1273555755615234,
|
| 154 |
-
"loss": 0.6701,
|
| 155 |
-
"rewards/accuracies": 0.5874999761581421,
|
| 156 |
-
"rewards/chosen": 0.04946264624595642,
|
| 157 |
-
"rewards/margins": 0.041922032833099365,
|
| 158 |
-
"rewards/rejected": 0.007540611084550619,
|
| 159 |
-
"step": 45
|
| 160 |
-
},
|
| 161 |
-
{
|
| 162 |
-
"epoch": 0.0998003992015968,
|
| 163 |
-
"grad_norm": 32.39288675133871,
|
| 164 |
-
"learning_rate": 7.84313725490196e-07,
|
| 165 |
-
"logits/chosen": -1.4457480907440186,
|
| 166 |
-
"logits/rejected": -1.4977794885635376,
|
| 167 |
-
"logps/chosen": -0.9463974237442017,
|
| 168 |
-
"logps/rejected": -1.2008764743804932,
|
| 169 |
-
"loss": 0.6473,
|
| 170 |
-
"rewards/accuracies": 0.7875000238418579,
|
| 171 |
-
"rewards/chosen": -0.026257414370775223,
|
| 172 |
-
"rewards/margins": 0.270720511674881,
|
| 173 |
-
"rewards/rejected": -0.2969779074192047,
|
| 174 |
-
"step": 50
|
| 175 |
-
},
|
| 176 |
-
{
|
| 177 |
-
"epoch": 0.10978043912175649,
|
| 178 |
-
"grad_norm": 25.562422423609192,
|
| 179 |
-
"learning_rate": 7.998440460161416e-07,
|
| 180 |
-
"logits/chosen": -1.6372053623199463,
|
| 181 |
-
"logits/rejected": -1.4895769357681274,
|
| 182 |
-
"logps/chosen": -0.9624001383781433,
|
| 183 |
-
"logps/rejected": -1.001164197921753,
|
| 184 |
-
"loss": 0.6452,
|
| 185 |
-
"rewards/accuracies": 0.762499988079071,
|
| 186 |
-
"rewards/chosen": -0.22816535830497742,
|
| 187 |
-
"rewards/margins": 0.20386116206645966,
|
| 188 |
-
"rewards/rejected": -0.4320264756679535,
|
| 189 |
-
"step": 55
|
| 190 |
-
},
|
| 191 |
-
{
|
| 192 |
-
"epoch": 0.11976047904191617,
|
| 193 |
-
"grad_norm": 24.2619415054151,
|
| 194 |
-
"learning_rate": 7.992106913713086e-07,
|
| 195 |
-
"logits/chosen": -1.4938465356826782,
|
| 196 |
-
"logits/rejected": -1.5313457250595093,
|
| 197 |
-
"logps/chosen": -1.0081576108932495,
|
| 198 |
-
"logps/rejected": -1.2644726037979126,
|
| 199 |
-
"loss": 0.6479,
|
| 200 |
-
"rewards/accuracies": 0.75,
|
| 201 |
-
"rewards/chosen": -0.19145138561725616,
|
| 202 |
-
"rewards/margins": 0.5550881624221802,
|
| 203 |
-
"rewards/rejected": -0.7465395331382751,
|
| 204 |
-
"step": 60
|
| 205 |
-
},
|
| 206 |
-
{
|
| 207 |
-
"epoch": 0.12974051896207583,
|
| 208 |
-
"grad_norm": 53.09152893491566,
|
| 209 |
-
"learning_rate": 7.980909599927324e-07,
|
| 210 |
-
"logits/chosen": -1.6245317459106445,
|
| 211 |
-
"logits/rejected": -1.7302849292755127,
|
| 212 |
-
"logps/chosen": -0.9516083002090454,
|
| 213 |
-
"logps/rejected": -1.0930800437927246,
|
| 214 |
-
"loss": 0.6259,
|
| 215 |
-
"rewards/accuracies": 0.7250000238418579,
|
| 216 |
-
"rewards/chosen": -0.033558934926986694,
|
| 217 |
-
"rewards/margins": 0.2880553603172302,
|
| 218 |
-
"rewards/rejected": -0.3216143250465393,
|
| 219 |
-
"step": 65
|
| 220 |
-
},
|
| 221 |
-
{
|
| 222 |
-
"epoch": 0.13972055888223553,
|
| 223 |
-
"grad_norm": 22.54639277386895,
|
| 224 |
-
"learning_rate": 7.964862161006166e-07,
|
| 225 |
-
"logits/chosen": -1.6116857528686523,
|
| 226 |
-
"logits/rejected": -1.5303642749786377,
|
| 227 |
-
"logps/chosen": -0.9262536764144897,
|
| 228 |
-
"logps/rejected": -1.1940580606460571,
|
| 229 |
-
"loss": 0.5986,
|
| 230 |
-
"rewards/accuracies": 0.8125,
|
| 231 |
-
"rewards/chosen": 0.052712906152009964,
|
| 232 |
-
"rewards/margins": 0.4166509509086609,
|
| 233 |
-
"rewards/rejected": -0.3639380633831024,
|
| 234 |
-
"step": 70
|
| 235 |
-
},
|
| 236 |
-
{
|
| 237 |
-
"epoch": 0.1497005988023952,
|
| 238 |
-
"grad_norm": 42.93936100609094,
|
| 239 |
-
"learning_rate": 7.94398414828202e-07,
|
| 240 |
-
"logits/chosen": -1.7100902795791626,
|
| 241 |
-
"logits/rejected": -1.6730142831802368,
|
| 242 |
-
"logps/chosen": -0.9657354354858398,
|
| 243 |
-
"logps/rejected": -0.9953187108039856,
|
| 244 |
-
"loss": 0.5873,
|
| 245 |
-
"rewards/accuracies": 0.7749999761581421,
|
| 246 |
-
"rewards/chosen": -0.0733981654047966,
|
| 247 |
-
"rewards/margins": 0.29382631182670593,
|
| 248 |
-
"rewards/rejected": -0.36722445487976074,
|
| 249 |
-
"step": 75
|
| 250 |
-
},
|
| 251 |
-
{
|
| 252 |
-
"epoch": 0.1596806387225549,
|
| 253 |
-
"grad_norm": 33.085627858420565,
|
| 254 |
-
"learning_rate": 7.918300998397376e-07,
|
| 255 |
-
"logits/chosen": -1.5861061811447144,
|
| 256 |
-
"logits/rejected": -1.5445070266723633,
|
| 257 |
-
"logps/chosen": -0.9483486413955688,
|
| 258 |
-
"logps/rejected": -1.051520824432373,
|
| 259 |
-
"loss": 0.5927,
|
| 260 |
-
"rewards/accuracies": 0.699999988079071,
|
| 261 |
-
"rewards/chosen": -0.18298718333244324,
|
| 262 |
-
"rewards/margins": 0.16031306982040405,
|
| 263 |
-
"rewards/rejected": -0.3433002233505249,
|
| 264 |
-
"step": 80
|
| 265 |
-
},
|
| 266 |
-
{
|
| 267 |
-
"epoch": 0.16966067864271456,
|
| 268 |
-
"grad_norm": 32.604735894329295,
|
| 269 |
-
"learning_rate": 7.887844002314185e-07,
|
| 270 |
-
"logits/chosen": -1.456621527671814,
|
| 271 |
-
"logits/rejected": -1.550982117652893,
|
| 272 |
-
"logps/chosen": -0.8949592709541321,
|
| 273 |
-
"logps/rejected": -1.0014525651931763,
|
| 274 |
-
"loss": 0.6049,
|
| 275 |
-
"rewards/accuracies": 0.75,
|
| 276 |
-
"rewards/chosen": -0.37570685148239136,
|
| 277 |
-
"rewards/margins": 0.4748932421207428,
|
| 278 |
-
"rewards/rejected": -0.8506000638008118,
|
| 279 |
-
"step": 85
|
| 280 |
-
},
|
| 281 |
-
{
|
| 282 |
-
"epoch": 0.17964071856287425,
|
| 283 |
-
"grad_norm": 50.600905445142125,
|
| 284 |
-
"learning_rate": 7.852650267190632e-07,
|
| 285 |
-
"logits/chosen": -1.6303730010986328,
|
| 286 |
-
"logits/rejected": -1.609903335571289,
|
| 287 |
-
"logps/chosen": -1.0100046396255493,
|
| 288 |
-
"logps/rejected": -1.0304863452911377,
|
| 289 |
-
"loss": 0.5861,
|
| 290 |
-
"rewards/accuracies": 0.7250000238418579,
|
| 291 |
-
"rewards/chosen": -0.46551984548568726,
|
| 292 |
-
"rewards/margins": 0.5207703709602356,
|
| 293 |
-
"rewards/rejected": -0.9862901568412781,
|
| 294 |
-
"step": 90
|
| 295 |
-
},
|
| 296 |
-
{
|
| 297 |
-
"epoch": 0.18962075848303392,
|
| 298 |
-
"grad_norm": 46.677854954895324,
|
| 299 |
-
"learning_rate": 7.812762671171788e-07,
|
| 300 |
-
"logits/chosen": -1.8154528141021729,
|
| 301 |
-
"logits/rejected": -1.7442315816879272,
|
| 302 |
-
"logps/chosen": -1.245895504951477,
|
| 303 |
-
"logps/rejected": -1.381966233253479,
|
| 304 |
-
"loss": 0.5881,
|
| 305 |
-
"rewards/accuracies": 0.8125,
|
| 306 |
-
"rewards/chosen": -0.5333404541015625,
|
| 307 |
-
"rewards/margins": 0.9066336750984192,
|
| 308 |
-
"rewards/rejected": -1.439974069595337,
|
| 309 |
-
"step": 95
|
| 310 |
-
},
|
| 311 |
-
{
|
| 312 |
-
"epoch": 0.1996007984031936,
|
| 313 |
-
"grad_norm": 45.70458018922122,
|
| 314 |
-
"learning_rate": 7.768229811149186e-07,
|
| 315 |
-
"logits/chosen": -1.6239620447158813,
|
| 316 |
-
"logits/rejected": -1.7077280282974243,
|
| 317 |
-
"logps/chosen": -1.0647861957550049,
|
| 318 |
-
"logps/rejected": -1.128450632095337,
|
| 319 |
-
"loss": 0.5655,
|
| 320 |
-
"rewards/accuracies": 0.7875000238418579,
|
| 321 |
-
"rewards/chosen": 0.02701367810368538,
|
| 322 |
-
"rewards/margins": 0.667129635810852,
|
| 323 |
-
"rewards/rejected": -0.6401158571243286,
|
| 324 |
-
"step": 100
|
| 325 |
-
},
|
| 326 |
-
{
|
| 327 |
-
"epoch": 0.20958083832335328,
|
| 328 |
-
"grad_norm": 68.02288800224188,
|
| 329 |
-
"learning_rate": 7.719105943553005e-07,
|
| 330 |
-
"logits/chosen": -1.665117859840393,
|
| 331 |
-
"logits/rejected": -1.7087205648422241,
|
| 332 |
-
"logps/chosen": -0.872536301612854,
|
| 333 |
-
"logps/rejected": -1.0017918348312378,
|
| 334 |
-
"loss": 0.5936,
|
| 335 |
-
"rewards/accuracies": 0.824999988079071,
|
| 336 |
-
"rewards/chosen": -0.5589737892150879,
|
| 337 |
-
"rewards/margins": 0.5613604784011841,
|
| 338 |
-
"rewards/rejected": -1.1203341484069824,
|
| 339 |
-
"step": 105
|
| 340 |
-
},
|
| 341 |
-
{
|
| 342 |
-
"epoch": 0.21956087824351297,
|
| 343 |
-
"grad_norm": 26.445396621363834,
|
| 344 |
-
"learning_rate": 7.665450918248958e-07,
|
| 345 |
-
"logits/chosen": -1.7527316808700562,
|
| 346 |
-
"logits/rejected": -1.701978087425232,
|
| 347 |
-
"logps/chosen": -0.9653258323669434,
|
| 348 |
-
"logps/rejected": -1.0634543895721436,
|
| 349 |
-
"loss": 0.5066,
|
| 350 |
-
"rewards/accuracies": 0.824999988079071,
|
| 351 |
-
"rewards/chosen": -0.627923846244812,
|
| 352 |
-
"rewards/margins": 0.6550946235656738,
|
| 353 |
-
"rewards/rejected": -1.2830184698104858,
|
| 354 |
-
"step": 110
|
| 355 |
-
},
|
| 356 |
-
{
|
| 357 |
-
"epoch": 0.22954091816367264,
|
| 358 |
-
"grad_norm": 30.90025286425145,
|
| 359 |
-
"learning_rate": 7.607330105620455e-07,
|
| 360 |
-
"logits/chosen": -1.7857640981674194,
|
| 361 |
-
"logits/rejected": -1.668609619140625,
|
| 362 |
-
"logps/chosen": -1.0989948511123657,
|
| 363 |
-
"logps/rejected": -1.1228173971176147,
|
| 364 |
-
"loss": 0.5595,
|
| 365 |
-
"rewards/accuracies": 0.7749999761581421,
|
| 366 |
-
"rewards/chosen": -0.6360177993774414,
|
| 367 |
-
"rewards/margins": 0.5694562196731567,
|
| 368 |
-
"rewards/rejected": -1.2054741382598877,
|
| 369 |
-
"step": 115
|
| 370 |
-
},
|
| 371 |
-
{
|
| 372 |
-
"epoch": 0.23952095808383234,
|
| 373 |
-
"grad_norm": 40.011128656457046,
|
| 374 |
-
"learning_rate": 7.544814316924858e-07,
|
| 375 |
-
"logits/chosen": -1.7470954656600952,
|
| 376 |
-
"logits/rejected": -1.745476484298706,
|
| 377 |
-
"logps/chosen": -1.1192394495010376,
|
| 378 |
-
"logps/rejected": -1.2689249515533447,
|
| 379 |
-
"loss": 0.5494,
|
| 380 |
-
"rewards/accuracies": 0.7250000238418579,
|
| 381 |
-
"rewards/chosen": -1.0388853549957275,
|
| 382 |
-
"rewards/margins": 0.6685038805007935,
|
| 383 |
-
"rewards/rejected": -1.7073894739151,
|
| 384 |
-
"step": 120
|
| 385 |
-
},
|
| 386 |
-
{
|
| 387 |
-
"epoch": 0.249500998003992,
|
| 388 |
-
"grad_norm": 21.89811631617477,
|
| 389 |
-
"learning_rate": 7.477979718020876e-07,
|
| 390 |
-
"logits/chosen": -1.5766489505767822,
|
| 391 |
-
"logits/rejected": -1.4870827198028564,
|
| 392 |
-
"logps/chosen": -1.0617902278900146,
|
| 393 |
-
"logps/rejected": -1.2029839754104614,
|
| 394 |
-
"loss": 0.4894,
|
| 395 |
-
"rewards/accuracies": 0.8500000238418579,
|
| 396 |
-
"rewards/chosen": -0.9320518374443054,
|
| 397 |
-
"rewards/margins": 0.8520485758781433,
|
| 398 |
-
"rewards/rejected": -1.7841002941131592,
|
| 399 |
-
"step": 125
|
| 400 |
-
},
|
| 401 |
-
{
|
| 402 |
-
"epoch": 0.25948103792415167,
|
| 403 |
-
"grad_norm": 32.50408891288284,
|
| 404 |
-
"learning_rate": 7.40690773657219e-07,
|
| 405 |
-
"logits/chosen": -1.6710205078125,
|
| 406 |
-
"logits/rejected": -1.6744377613067627,
|
| 407 |
-
"logps/chosen": -0.9914701581001282,
|
| 408 |
-
"logps/rejected": -1.1012578010559082,
|
| 409 |
-
"loss": 0.4908,
|
| 410 |
-
"rewards/accuracies": 0.8374999761581421,
|
| 411 |
-
"rewards/chosen": -0.8213608860969543,
|
| 412 |
-
"rewards/margins": 0.6687670946121216,
|
| 413 |
-
"rewards/rejected": -1.4901279211044312,
|
| 414 |
-
"step": 130
|
| 415 |
-
},
|
| 416 |
-
{
|
| 417 |
-
"epoch": 0.2694610778443114,
|
| 418 |
-
"grad_norm": 52.90692812097415,
|
| 419 |
-
"learning_rate": 7.331684962840398e-07,
|
| 420 |
-
"logits/chosen": -1.5417057275772095,
|
| 421 |
-
"logits/rejected": -1.563847303390503,
|
| 422 |
-
"logps/chosen": -0.9478904008865356,
|
| 423 |
-
"logps/rejected": -1.1853077411651611,
|
| 424 |
-
"loss": 0.4906,
|
| 425 |
-
"rewards/accuracies": 0.824999988079071,
|
| 426 |
-
"rewards/chosen": -0.5953829884529114,
|
| 427 |
-
"rewards/margins": 1.0272996425628662,
|
| 428 |
-
"rewards/rejected": -1.6226825714111328,
|
| 429 |
-
"step": 135
|
| 430 |
-
},
|
| 431 |
-
{
|
| 432 |
-
"epoch": 0.27944111776447106,
|
| 433 |
-
"grad_norm": 29.247151675001287,
|
| 434 |
-
"learning_rate": 7.25240304418811e-07,
|
| 435 |
-
"logits/chosen": -1.723270058631897,
|
| 436 |
-
"logits/rejected": -1.6206943988800049,
|
| 437 |
-
"logps/chosen": -1.0017354488372803,
|
| 438 |
-
"logps/rejected": -1.1501209735870361,
|
| 439 |
-
"loss": 0.4895,
|
| 440 |
-
"rewards/accuracies": 0.862500011920929,
|
| 441 |
-
"rewards/chosen": -0.7104414701461792,
|
| 442 |
-
"rewards/margins": 1.0606929063796997,
|
| 443 |
-
"rewards/rejected": -1.771134376525879,
|
| 444 |
-
"step": 140
|
| 445 |
-
},
|
| 446 |
-
{
|
| 447 |
-
"epoch": 0.2894211576846307,
|
| 448 |
-
"grad_norm": 23.202561617382436,
|
| 449 |
-
"learning_rate": 7.169158573420763e-07,
|
| 450 |
-
"logits/chosen": -1.7494615316390991,
|
| 451 |
-
"logits/rejected": -1.741146445274353,
|
| 452 |
-
"logps/chosen": -1.208905816078186,
|
| 453 |
-
"logps/rejected": -1.3130097389221191,
|
| 454 |
-
"loss": 0.4886,
|
| 455 |
-
"rewards/accuracies": 0.8125,
|
| 456 |
-
"rewards/chosen": -1.2182276248931885,
|
| 457 |
-
"rewards/margins": 1.148219347000122,
|
| 458 |
-
"rewards/rejected": -2.3664469718933105,
|
| 459 |
-
"step": 145
|
| 460 |
-
},
|
| 461 |
-
{
|
| 462 |
-
"epoch": 0.2994011976047904,
|
| 463 |
-
"grad_norm": 31.417521603405053,
|
| 464 |
-
"learning_rate": 7.082052971103157e-07,
|
| 465 |
-
"logits/chosen": -1.614556074142456,
|
| 466 |
-
"logits/rejected": -1.5841604471206665,
|
| 467 |
-
"logps/chosen": -1.156620979309082,
|
| 468 |
-
"logps/rejected": -1.2092293500900269,
|
| 469 |
-
"loss": 0.5008,
|
| 470 |
-
"rewards/accuracies": 0.800000011920929,
|
| 471 |
-
"rewards/chosen": -1.330456256866455,
|
| 472 |
-
"rewards/margins": 0.7478895783424377,
|
| 473 |
-
"rewards/rejected": -2.078345775604248,
|
| 474 |
-
"step": 150
|
| 475 |
-
},
|
| 476 |
-
{
|
| 477 |
-
"epoch": 0.3093812375249501,
|
| 478 |
-
"grad_norm": 22.72615529430511,
|
| 479 |
-
"learning_rate": 6.991192361994127e-07,
|
| 480 |
-
"logits/chosen": -1.748228669166565,
|
| 481 |
-
"logits/rejected": -1.6922461986541748,
|
| 482 |
-
"logps/chosen": -1.078489065170288,
|
| 483 |
-
"logps/rejected": -1.3014332056045532,
|
| 484 |
-
"loss": 0.4438,
|
| 485 |
-
"rewards/accuracies": 0.8500000238418579,
|
| 486 |
-
"rewards/chosen": -1.2569700479507446,
|
| 487 |
-
"rewards/margins": 1.386672854423523,
|
| 488 |
-
"rewards/rejected": -2.6436429023742676,
|
| 489 |
-
"step": 155
|
| 490 |
-
},
|
| 491 |
-
{
|
| 492 |
-
"epoch": 0.3193612774451098,
|
| 493 |
-
"grad_norm": 53.34962173479164,
|
| 494 |
-
"learning_rate": 6.896687445749871e-07,
|
| 495 |
-
"logits/chosen": -1.7363059520721436,
|
| 496 |
-
"logits/rejected": -1.7595733404159546,
|
| 497 |
-
"logps/chosen": -1.115309476852417,
|
| 498 |
-
"logps/rejected": -1.239806890487671,
|
| 499 |
-
"loss": 0.4622,
|
| 500 |
-
"rewards/accuracies": 0.8500000238418579,
|
| 501 |
-
"rewards/chosen": -1.4574905633926392,
|
| 502 |
-
"rewards/margins": 1.336918830871582,
|
| 503 |
-
"rewards/rejected": -2.7944092750549316,
|
| 504 |
-
"step": 160
|
| 505 |
-
},
|
| 506 |
-
{
|
| 507 |
-
"epoch": 0.32934131736526945,
|
| 508 |
-
"grad_norm": 40.83000749193078,
|
| 509 |
-
"learning_rate": 6.798653362053462e-07,
|
| 510 |
-
"logits/chosen": -1.650973916053772,
|
| 511 |
-
"logits/rejected": -1.6539020538330078,
|
| 512 |
-
"logps/chosen": -1.1761391162872314,
|
| 513 |
-
"logps/rejected": -1.4524037837982178,
|
| 514 |
-
"loss": 0.4558,
|
| 515 |
-
"rewards/accuracies": 0.862500011920929,
|
| 516 |
-
"rewards/chosen": -1.311657190322876,
|
| 517 |
-
"rewards/margins": 1.8300256729125977,
|
| 518 |
-
"rewards/rejected": -3.1416831016540527,
|
| 519 |
-
"step": 165
|
| 520 |
-
},
|
| 521 |
-
{
|
| 522 |
-
"epoch": 0.3393213572854291,
|
| 523 |
-
"grad_norm": 28.447642135314243,
|
| 524 |
-
"learning_rate": 6.697209550334893e-07,
|
| 525 |
-
"logits/chosen": -1.7830873727798462,
|
| 526 |
-
"logits/rejected": -1.8306477069854736,
|
| 527 |
-
"logps/chosen": -1.1625844240188599,
|
| 528 |
-
"logps/rejected": -1.4296667575836182,
|
| 529 |
-
"loss": 0.4682,
|
| 530 |
-
"rewards/accuracies": 0.762499988079071,
|
| 531 |
-
"rewards/chosen": -1.1936951875686646,
|
| 532 |
-
"rewards/margins": 1.3315410614013672,
|
| 533 |
-
"rewards/rejected": -2.525236129760742,
|
| 534 |
-
"step": 170
|
| 535 |
-
},
|
| 536 |
-
{
|
| 537 |
-
"epoch": 0.34930139720558884,
|
| 538 |
-
"grad_norm": 42.529786752880625,
|
| 539 |
-
"learning_rate": 6.592479604252524e-07,
|
| 540 |
-
"logits/chosen": -1.6931123733520508,
|
| 541 |
-
"logits/rejected": -1.6376163959503174,
|
| 542 |
-
"logps/chosen": -1.1715461015701294,
|
| 543 |
-
"logps/rejected": -1.3761537075042725,
|
| 544 |
-
"loss": 0.4751,
|
| 545 |
-
"rewards/accuracies": 0.762499988079071,
|
| 546 |
-
"rewards/chosen": -1.1568340063095093,
|
| 547 |
-
"rewards/margins": 1.0651991367340088,
|
| 548 |
-
"rewards/rejected": -2.2220332622528076,
|
| 549 |
-
"step": 175
|
| 550 |
-
},
|
| 551 |
-
{
|
| 552 |
-
"epoch": 0.3592814371257485,
|
| 553 |
-
"grad_norm": 49.01740947758452,
|
| 554 |
-
"learning_rate": 6.484591121113241e-07,
|
| 555 |
-
"logits/chosen": -1.7936656475067139,
|
| 556 |
-
"logits/rejected": -1.7206109762191772,
|
| 557 |
-
"logps/chosen": -1.047082781791687,
|
| 558 |
-
"logps/rejected": -1.2329199314117432,
|
| 559 |
-
"loss": 0.4627,
|
| 560 |
-
"rewards/accuracies": 0.8125,
|
| 561 |
-
"rewards/chosen": -1.0403250455856323,
|
| 562 |
-
"rewards/margins": 1.458751916885376,
|
| 563 |
-
"rewards/rejected": -2.499077081680298,
|
| 564 |
-
"step": 180
|
| 565 |
-
},
|
| 566 |
-
{
|
| 567 |
-
"epoch": 0.36926147704590817,
|
| 568 |
-
"grad_norm": 33.35075365946081,
|
| 569 |
-
"learning_rate": 6.373675546414806e-07,
|
| 570 |
-
"logits/chosen": -1.7752662897109985,
|
| 571 |
-
"logits/rejected": -1.8046382665634155,
|
| 572 |
-
"logps/chosen": -1.105947494506836,
|
| 573 |
-
"logps/rejected": -1.4365394115447998,
|
| 574 |
-
"loss": 0.4533,
|
| 575 |
-
"rewards/accuracies": 0.8125,
|
| 576 |
-
"rewards/chosen": -1.559922456741333,
|
| 577 |
-
"rewards/margins": 1.3131450414657593,
|
| 578 |
-
"rewards/rejected": -2.8730673789978027,
|
| 579 |
-
"step": 185
|
| 580 |
-
},
|
| 581 |
-
{
|
| 582 |
-
"epoch": 0.37924151696606784,
|
| 583 |
-
"grad_norm": 33.28309177685594,
|
| 584 |
-
"learning_rate": 6.259868013699751e-07,
|
| 585 |
-
"logits/chosen": -1.7205069065093994,
|
| 586 |
-
"logits/rejected": -1.6720272302627563,
|
| 587 |
-
"logps/chosen": -1.1314589977264404,
|
| 588 |
-
"logps/rejected": -1.2337143421173096,
|
| 589 |
-
"loss": 0.4049,
|
| 590 |
-
"rewards/accuracies": 0.800000011920929,
|
| 591 |
-
"rewards/chosen": -1.9102380275726318,
|
| 592 |
-
"rewards/margins": 1.2990434169769287,
|
| 593 |
-
"rewards/rejected": -3.2092814445495605,
|
| 594 |
-
"step": 190
|
| 595 |
-
},
|
| 596 |
-
{
|
| 597 |
-
"epoch": 0.38922155688622756,
|
| 598 |
-
"grad_norm": 22.448955038966368,
|
| 599 |
-
"learning_rate": 6.143307179915986e-07,
|
| 600 |
-
"logits/chosen": -1.63016676902771,
|
| 601 |
-
"logits/rejected": -1.6702827215194702,
|
| 602 |
-
"logps/chosen": -1.1162363290786743,
|
| 603 |
-
"logps/rejected": -1.364017128944397,
|
| 604 |
-
"loss": 0.3814,
|
| 605 |
-
"rewards/accuracies": 0.875,
|
| 606 |
-
"rewards/chosen": -1.8935344219207764,
|
| 607 |
-
"rewards/margins": 1.6944122314453125,
|
| 608 |
-
"rewards/rejected": -3.5879464149475098,
|
| 609 |
-
"step": 195
|
| 610 |
-
},
|
| 611 |
-
{
|
| 612 |
-
"epoch": 0.3992015968063872,
|
| 613 |
-
"grad_norm": 40.67962108560719,
|
| 614 |
-
"learning_rate": 6.024135056484655e-07,
|
| 615 |
-
"logits/chosen": -1.7706257104873657,
|
| 616 |
-
"logits/rejected": -1.7765319347381592,
|
| 617 |
-
"logps/chosen": -1.2081105709075928,
|
| 618 |
-
"logps/rejected": -1.5039583444595337,
|
| 619 |
-
"loss": 0.4225,
|
| 620 |
-
"rewards/accuracies": 0.875,
|
| 621 |
-
"rewards/chosen": -1.8676178455352783,
|
| 622 |
-
"rewards/margins": 2.3755626678466797,
|
| 623 |
-
"rewards/rejected": -4.243180274963379,
|
| 624 |
-
"step": 200
|
| 625 |
-
},
|
| 626 |
-
{
|
| 627 |
-
"epoch": 0.4091816367265469,
|
| 628 |
-
"grad_norm": 22.600562301516938,
|
| 629 |
-
"learning_rate": 5.902496836281101e-07,
|
| 630 |
-
"logits/chosen": -1.772716760635376,
|
| 631 |
-
"logits/rejected": -1.7146966457366943,
|
| 632 |
-
"logps/chosen": -1.1653116941452026,
|
| 633 |
-
"logps/rejected": -1.2957953214645386,
|
| 634 |
-
"loss": 0.353,
|
| 635 |
-
"rewards/accuracies": 0.8125,
|
| 636 |
-
"rewards/chosen": -1.8657941818237305,
|
| 637 |
-
"rewards/margins": 1.3124840259552002,
|
| 638 |
-
"rewards/rejected": -3.1782779693603516,
|
| 639 |
-
"step": 205
|
| 640 |
-
},
|
| 641 |
-
{
|
| 642 |
-
"epoch": 0.41916167664670656,
|
| 643 |
-
"grad_norm": 31.666375625024422,
|
| 644 |
-
"learning_rate": 5.778540716739709e-07,
|
| 645 |
-
"logits/chosen": -1.7554540634155273,
|
| 646 |
-
"logits/rejected": -1.8094005584716797,
|
| 647 |
-
"logps/chosen": -1.2131294012069702,
|
| 648 |
-
"logps/rejected": -1.54654860496521,
|
| 649 |
-
"loss": 0.3751,
|
| 650 |
-
"rewards/accuracies": 0.8999999761581421,
|
| 651 |
-
"rewards/chosen": -1.522384524345398,
|
| 652 |
-
"rewards/margins": 2.483349561691284,
|
| 653 |
-
"rewards/rejected": -4.005733966827393,
|
| 654 |
-
"step": 210
|
| 655 |
-
},
|
| 656 |
-
{
|
| 657 |
-
"epoch": 0.4291417165668663,
|
| 658 |
-
"grad_norm": 23.231350885180493,
|
| 659 |
-
"learning_rate": 5.652417719298168e-07,
|
| 660 |
-
"logits/chosen": -1.7632023096084595,
|
| 661 |
-
"logits/rejected": -1.6775318384170532,
|
| 662 |
-
"logps/chosen": -1.254577875137329,
|
| 663 |
-
"logps/rejected": -1.4464714527130127,
|
| 664 |
-
"loss": 0.42,
|
| 665 |
-
"rewards/accuracies": 0.8125,
|
| 666 |
-
"rewards/chosen": -2.3697409629821777,
|
| 667 |
-
"rewards/margins": 1.543129324913025,
|
| 668 |
-
"rewards/rejected": -3.912869930267334,
|
| 669 |
-
"step": 215
|
| 670 |
-
},
|
| 671 |
-
{
|
| 672 |
-
"epoch": 0.43912175648702595,
|
| 673 |
-
"grad_norm": 75.06138455980248,
|
| 674 |
-
"learning_rate": 5.524281505401096e-07,
|
| 675 |
-
"logits/chosen": -1.7130699157714844,
|
| 676 |
-
"logits/rejected": -1.7215654850006104,
|
| 677 |
-
"logps/chosen": -1.1106432676315308,
|
| 678 |
-
"logps/rejected": -1.389941692352295,
|
| 679 |
-
"loss": 0.4627,
|
| 680 |
-
"rewards/accuracies": 0.8125,
|
| 681 |
-
"rewards/chosen": -2.0083611011505127,
|
| 682 |
-
"rewards/margins": 1.8821561336517334,
|
| 683 |
-
"rewards/rejected": -3.890516996383667,
|
| 684 |
-
"step": 220
|
| 685 |
-
},
|
| 686 |
-
{
|
| 687 |
-
"epoch": 0.4491017964071856,
|
| 688 |
-
"grad_norm": 38.6593639633161,
|
| 689 |
-
"learning_rate": 5.394288189287261e-07,
|
| 690 |
-
"logits/chosen": -1.787519097328186,
|
| 691 |
-
"logits/rejected": -1.7786083221435547,
|
| 692 |
-
"logps/chosen": -1.1749248504638672,
|
| 693 |
-
"logps/rejected": -1.4376263618469238,
|
| 694 |
-
"loss": 0.3756,
|
| 695 |
-
"rewards/accuracies": 0.862500011920929,
|
| 696 |
-
"rewards/chosen": -2.5106921195983887,
|
| 697 |
-
"rewards/margins": 2.196367025375366,
|
| 698 |
-
"rewards/rejected": -4.707058906555176,
|
| 699 |
-
"step": 225
|
| 700 |
-
},
|
| 701 |
-
{
|
| 702 |
-
"epoch": 0.4590818363273453,
|
| 703 |
-
"grad_norm": 78.42260836111647,
|
| 704 |
-
"learning_rate": 5.26259614778841e-07,
|
| 705 |
-
"logits/chosen": -1.642748475074768,
|
| 706 |
-
"logits/rejected": -1.6440712213516235,
|
| 707 |
-
"logps/chosen": -1.1253970861434937,
|
| 708 |
-
"logps/rejected": -1.2666562795639038,
|
| 709 |
-
"loss": 0.4147,
|
| 710 |
-
"rewards/accuracies": 0.862500011920929,
|
| 711 |
-
"rewards/chosen": -2.194728374481201,
|
| 712 |
-
"rewards/margins": 1.729097604751587,
|
| 713 |
-
"rewards/rejected": -3.923825740814209,
|
| 714 |
-
"step": 230
|
| 715 |
-
},
|
| 716 |
-
{
|
| 717 |
-
"epoch": 0.469061876247505,
|
| 718 |
-
"grad_norm": 29.199406660864277,
|
| 719 |
-
"learning_rate": 5.129365827371506e-07,
|
| 720 |
-
"logits/chosen": -1.8895553350448608,
|
| 721 |
-
"logits/rejected": -1.8804956674575806,
|
| 722 |
-
"logps/chosen": -1.1362788677215576,
|
| 723 |
-
"logps/rejected": -1.3920345306396484,
|
| 724 |
-
"loss": 0.373,
|
| 725 |
-
"rewards/accuracies": 0.8125,
|
| 726 |
-
"rewards/chosen": -1.8095152378082275,
|
| 727 |
-
"rewards/margins": 2.366717576980591,
|
| 728 |
-
"rewards/rejected": -4.17623233795166,
|
| 729 |
-
"step": 235
|
| 730 |
-
},
|
| 731 |
-
{
|
| 732 |
-
"epoch": 0.47904191616766467,
|
| 733 |
-
"grad_norm": 47.16394571127521,
|
| 734 |
-
"learning_rate": 4.99475954865942e-07,
|
| 735 |
-
"logits/chosen": -1.7921149730682373,
|
| 736 |
-
"logits/rejected": -1.7930948734283447,
|
| 737 |
-
"logps/chosen": -1.0978577136993408,
|
| 738 |
-
"logps/rejected": -1.2884962558746338,
|
| 739 |
-
"loss": 0.4121,
|
| 740 |
-
"rewards/accuracies": 0.824999988079071,
|
| 741 |
-
"rewards/chosen": -2.472763776779175,
|
| 742 |
-
"rewards/margins": 1.3844527006149292,
|
| 743 |
-
"rewards/rejected": -3.8572165966033936,
|
| 744 |
-
"step": 240
|
| 745 |
-
},
|
| 746 |
-
{
|
| 747 |
-
"epoch": 0.48902195608782434,
|
| 748 |
-
"grad_norm": 49.94184345559498,
|
| 749 |
-
"learning_rate": 4.858941308668253e-07,
|
| 750 |
-
"logits/chosen": -1.7998384237289429,
|
| 751 |
-
"logits/rejected": -1.7332870960235596,
|
| 752 |
-
"logps/chosen": -1.2298054695129395,
|
| 753 |
-
"logps/rejected": -1.583311915397644,
|
| 754 |
-
"loss": 0.3966,
|
| 755 |
-
"rewards/accuracies": 0.800000011920929,
|
| 756 |
-
"rewards/chosen": -2.484738349914551,
|
| 757 |
-
"rewards/margins": 2.6141140460968018,
|
| 758 |
-
"rewards/rejected": -5.09885311126709,
|
| 759 |
-
"step": 245
|
| 760 |
-
},
|
| 761 |
-
{
|
| 762 |
-
"epoch": 0.499001996007984,
|
| 763 |
-
"grad_norm": 24.230906422190216,
|
| 764 |
-
"learning_rate": 4.72207658100224e-07,
|
| 765 |
-
"logits/chosen": -1.8376038074493408,
|
| 766 |
-
"logits/rejected": -1.777984619140625,
|
| 767 |
-
"logps/chosen": -1.221793532371521,
|
| 768 |
-
"logps/rejected": -1.4775179624557495,
|
| 769 |
-
"loss": 0.355,
|
| 770 |
-
"rewards/accuracies": 0.9125000238418579,
|
| 771 |
-
"rewards/chosen": -2.4652364253997803,
|
| 772 |
-
"rewards/margins": 2.177865982055664,
|
| 773 |
-
"rewards/rejected": -4.643102169036865,
|
| 774 |
-
"step": 250
|
| 775 |
-
},
|
| 776 |
-
{
|
| 777 |
-
"epoch": 0.5089820359281437,
|
| 778 |
-
"grad_norm": 34.572626217381774,
|
| 779 |
-
"learning_rate": 4.5843321142496465e-07,
|
| 780 |
-
"logits/chosen": -1.7895371913909912,
|
| 781 |
-
"logits/rejected": -1.7715275287628174,
|
| 782 |
-
"logps/chosen": -1.2171119451522827,
|
| 783 |
-
"logps/rejected": -1.4471348524093628,
|
| 784 |
-
"loss": 0.3765,
|
| 785 |
-
"rewards/accuracies": 0.8500000238418579,
|
| 786 |
-
"rewards/chosen": -2.290224075317383,
|
| 787 |
-
"rewards/margins": 1.525314211845398,
|
| 788 |
-
"rewards/rejected": -3.8155384063720703,
|
| 789 |
-
"step": 255
|
| 790 |
-
},
|
| 791 |
-
{
|
| 792 |
-
"epoch": 0.5189620758483033,
|
| 793 |
-
"grad_norm": 50.38823987307752,
|
| 794 |
-
"learning_rate": 4.445875728825301e-07,
|
| 795 |
-
"logits/chosen": -1.759508490562439,
|
| 796 |
-
"logits/rejected": -1.7566909790039062,
|
| 797 |
-
"logps/chosen": -1.2831090688705444,
|
| 798 |
-
"logps/rejected": -1.553682565689087,
|
| 799 |
-
"loss": 0.407,
|
| 800 |
-
"rewards/accuracies": 0.8500000238418579,
|
| 801 |
-
"rewards/chosen": -2.9423370361328125,
|
| 802 |
-
"rewards/margins": 2.082186222076416,
|
| 803 |
-
"rewards/rejected": -5.024522304534912,
|
| 804 |
-
"step": 260
|
| 805 |
-
},
|
| 806 |
-
{
|
| 807 |
-
"epoch": 0.5289421157684631,
|
| 808 |
-
"grad_norm": 25.180525143347726,
|
| 809 |
-
"learning_rate": 4.3068761125072743e-07,
|
| 810 |
-
"logits/chosen": -1.7218453884124756,
|
| 811 |
-
"logits/rejected": -1.6865999698638916,
|
| 812 |
-
"logps/chosen": -1.2321453094482422,
|
| 813 |
-
"logps/rejected": -1.4395345449447632,
|
| 814 |
-
"loss": 0.378,
|
| 815 |
-
"rewards/accuracies": 0.8374999761581421,
|
| 816 |
-
"rewards/chosen": -2.701998472213745,
|
| 817 |
-
"rewards/margins": 1.778617262840271,
|
| 818 |
-
"rewards/rejected": -4.480616092681885,
|
| 819 |
-
"step": 265
|
| 820 |
-
},
|
| 821 |
-
{
|
| 822 |
-
"epoch": 0.5389221556886228,
|
| 823 |
-
"grad_norm": 34.94821484127908,
|
| 824 |
-
"learning_rate": 4.167502614916798e-07,
|
| 825 |
-
"logits/chosen": -1.7490154504776,
|
| 826 |
-
"logits/rejected": -1.7060623168945312,
|
| 827 |
-
"logps/chosen": -1.1463005542755127,
|
| 828 |
-
"logps/rejected": -1.4383161067962646,
|
| 829 |
-
"loss": 0.3744,
|
| 830 |
-
"rewards/accuracies": 0.8125,
|
| 831 |
-
"rewards/chosen": -2.461625576019287,
|
| 832 |
-
"rewards/margins": 1.8262317180633545,
|
| 833 |
-
"rewards/rejected": -4.287858009338379,
|
| 834 |
-
"step": 270
|
| 835 |
-
},
|
| 836 |
-
{
|
| 837 |
-
"epoch": 0.5489021956087824,
|
| 838 |
-
"grad_norm": 28.174483103692058,
|
| 839 |
-
"learning_rate": 4.027925041191846e-07,
|
| 840 |
-
"logits/chosen": -1.6615409851074219,
|
| 841 |
-
"logits/rejected": -1.717336893081665,
|
| 842 |
-
"logps/chosen": -1.0789427757263184,
|
| 843 |
-
"logps/rejected": -1.291114091873169,
|
| 844 |
-
"loss": 0.3802,
|
| 845 |
-
"rewards/accuracies": 0.800000011920929,
|
| 846 |
-
"rewards/chosen": -2.360185384750366,
|
| 847 |
-
"rewards/margins": 1.7126213312149048,
|
| 848 |
-
"rewards/rejected": -4.072806358337402,
|
| 849 |
-
"step": 275
|
| 850 |
-
},
|
| 851 |
-
{
|
| 852 |
-
"epoch": 0.5588822355289421,
|
| 853 |
-
"grad_norm": 65.52620829663537,
|
| 854 |
-
"learning_rate": 3.8883134451057253e-07,
|
| 855 |
-
"logits/chosen": -1.728354811668396,
|
| 856 |
-
"logits/rejected": -1.7853111028671265,
|
| 857 |
-
"logps/chosen": -1.1873013973236084,
|
| 858 |
-
"logps/rejected": -1.3926045894622803,
|
| 859 |
-
"loss": 0.3777,
|
| 860 |
-
"rewards/accuracies": 0.887499988079071,
|
| 861 |
-
"rewards/chosen": -2.1488184928894043,
|
| 862 |
-
"rewards/margins": 2.161184787750244,
|
| 863 |
-
"rewards/rejected": -4.31000280380249,
|
| 864 |
-
"step": 280
|
| 865 |
-
},
|
| 866 |
-
{
|
| 867 |
-
"epoch": 0.5688622754491018,
|
| 868 |
-
"grad_norm": 28.41327207057278,
|
| 869 |
-
"learning_rate": 3.7488379218827464e-07,
|
| 870 |
-
"logits/chosen": -1.8485610485076904,
|
| 871 |
-
"logits/rejected": -1.7869895696640015,
|
| 872 |
-
"logps/chosen": -1.1387436389923096,
|
| 873 |
-
"logps/rejected": -1.4143882989883423,
|
| 874 |
-
"loss": 0.3322,
|
| 875 |
-
"rewards/accuracies": 0.862500011920929,
|
| 876 |
-
"rewards/chosen": -2.3350417613983154,
|
| 877 |
-
"rewards/margins": 2.1998612880706787,
|
| 878 |
-
"rewards/rejected": -4.534903049468994,
|
| 879 |
-
"step": 285
|
| 880 |
-
},
|
| 881 |
-
{
|
| 882 |
-
"epoch": 0.5788423153692615,
|
| 883 |
-
"grad_norm": 54.15837971453539,
|
| 884 |
-
"learning_rate": 3.6096684009634026e-07,
|
| 885 |
-
"logits/chosen": -1.9337072372436523,
|
| 886 |
-
"logits/rejected": -1.8167577981948853,
|
| 887 |
-
"logps/chosen": -1.2900314331054688,
|
| 888 |
-
"logps/rejected": -1.5611326694488525,
|
| 889 |
-
"loss": 0.3949,
|
| 890 |
-
"rewards/accuracies": 0.925000011920929,
|
| 891 |
-
"rewards/chosen": -2.437575101852417,
|
| 892 |
-
"rewards/margins": 2.304208278656006,
|
| 893 |
-
"rewards/rejected": -4.741783142089844,
|
| 894 |
-
"step": 290
|
| 895 |
-
},
|
| 896 |
-
{
|
| 897 |
-
"epoch": 0.5888223552894212,
|
| 898 |
-
"grad_norm": 30.875206028913684,
|
| 899 |
-
"learning_rate": 3.4709744389715095e-07,
|
| 900 |
-
"logits/chosen": -1.7289743423461914,
|
| 901 |
-
"logits/rejected": -1.7040287256240845,
|
| 902 |
-
"logps/chosen": -1.1819794178009033,
|
| 903 |
-
"logps/rejected": -1.3941266536712646,
|
| 904 |
-
"loss": 0.3588,
|
| 905 |
-
"rewards/accuracies": 0.800000011920929,
|
| 906 |
-
"rewards/chosen": -2.6420161724090576,
|
| 907 |
-
"rewards/margins": 1.931423544883728,
|
| 908 |
-
"rewards/rejected": -4.573439598083496,
|
| 909 |
-
"step": 295
|
| 910 |
-
},
|
| 911 |
-
{
|
| 912 |
-
"epoch": 0.5988023952095808,
|
| 913 |
-
"grad_norm": 23.031518202722424,
|
| 914 |
-
"learning_rate": 3.332925013135591e-07,
|
| 915 |
-
"logits/chosen": -1.7638660669326782,
|
| 916 |
-
"logits/rejected": -1.8151865005493164,
|
| 917 |
-
"logps/chosen": -1.1131149530410767,
|
| 918 |
-
"logps/rejected": -1.3581509590148926,
|
| 919 |
-
"loss": 0.3593,
|
| 920 |
-
"rewards/accuracies": 0.800000011920929,
|
| 921 |
-
"rewards/chosen": -2.243455171585083,
|
| 922 |
-
"rewards/margins": 1.8504345417022705,
|
| 923 |
-
"rewards/rejected": -4.0938897132873535,
|
| 924 |
-
"step": 300
|
| 925 |
-
},
|
| 926 |
-
{
|
| 927 |
-
"epoch": 0.6087824351297405,
|
| 928 |
-
"grad_norm": 52.883386307510996,
|
| 929 |
-
"learning_rate": 3.195688315416142e-07,
|
| 930 |
-
"logits/chosen": -1.9481338262557983,
|
| 931 |
-
"logits/rejected": -1.877634048461914,
|
| 932 |
-
"logps/chosen": -1.3473719358444214,
|
| 933 |
-
"logps/rejected": -1.5660619735717773,
|
| 934 |
-
"loss": 0.3312,
|
| 935 |
-
"rewards/accuracies": 0.8374999761581421,
|
| 936 |
-
"rewards/chosen": -2.013882637023926,
|
| 937 |
-
"rewards/margins": 2.0933196544647217,
|
| 938 |
-
"rewards/rejected": -4.107202053070068,
|
| 939 |
-
"step": 305
|
| 940 |
-
},
|
| 941 |
-
{
|
| 942 |
-
"epoch": 0.6187624750499002,
|
| 943 |
-
"grad_norm": 40.31001639731489,
|
| 944 |
-
"learning_rate": 3.0594315475896397e-07,
|
| 945 |
-
"logits/chosen": -1.8702493906021118,
|
| 946 |
-
"logits/rejected": -1.8942811489105225,
|
| 947 |
-
"logps/chosen": -1.2539622783660889,
|
| 948 |
-
"logps/rejected": -1.5844415426254272,
|
| 949 |
-
"loss": 0.3233,
|
| 950 |
-
"rewards/accuracies": 0.9125000238418579,
|
| 951 |
-
"rewards/chosen": -2.855128049850464,
|
| 952 |
-
"rewards/margins": 2.254939556121826,
|
| 953 |
-
"rewards/rejected": -5.110067367553711,
|
| 954 |
-
"step": 310
|
| 955 |
-
},
|
| 956 |
-
{
|
| 957 |
-
"epoch": 0.6287425149700598,
|
| 958 |
-
"grad_norm": 34.57734687085236,
|
| 959 |
-
"learning_rate": 2.924320717538937e-07,
|
| 960 |
-
"logits/chosen": -1.9552714824676514,
|
| 961 |
-
"logits/rejected": -1.8983757495880127,
|
| 962 |
-
"logps/chosen": -1.235822319984436,
|
| 963 |
-
"logps/rejected": -1.557267427444458,
|
| 964 |
-
"loss": 0.3362,
|
| 965 |
-
"rewards/accuracies": 0.875,
|
| 966 |
-
"rewards/chosen": -2.798520803451538,
|
| 967 |
-
"rewards/margins": 2.358416795730591,
|
| 968 |
-
"rewards/rejected": -5.156937599182129,
|
| 969 |
-
"step": 315
|
| 970 |
-
},
|
| 971 |
-
{
|
| 972 |
-
"epoch": 0.6387225548902196,
|
| 973 |
-
"grad_norm": 22.131955927530566,
|
| 974 |
-
"learning_rate": 2.790520436998222e-07,
|
| 975 |
-
"logits/chosen": -1.8081642389297485,
|
| 976 |
-
"logits/rejected": -1.7835298776626587,
|
| 977 |
-
"logps/chosen": -1.2587134838104248,
|
| 978 |
-
"logps/rejected": -1.5615605115890503,
|
| 979 |
-
"loss": 0.3362,
|
| 980 |
-
"rewards/accuracies": 0.875,
|
| 981 |
-
"rewards/chosen": -2.6179587841033936,
|
| 982 |
-
"rewards/margins": 2.341552972793579,
|
| 983 |
-
"rewards/rejected": -4.959511756896973,
|
| 984 |
-
"step": 320
|
| 985 |
-
},
|
| 986 |
-
{
|
| 987 |
-
"epoch": 0.6487025948103793,
|
| 988 |
-
"grad_norm": 25.656535146073534,
|
| 989 |
-
"learning_rate": 2.6581937209989804e-07,
|
| 990 |
-
"logits/chosen": -1.857273817062378,
|
| 991 |
-
"logits/rejected": -1.8463045358657837,
|
| 992 |
-
"logps/chosen": -1.207121729850769,
|
| 993 |
-
"logps/rejected": -1.4980764389038086,
|
| 994 |
-
"loss": 0.2881,
|
| 995 |
-
"rewards/accuracies": 0.9125000238418579,
|
| 996 |
-
"rewards/chosen": -2.7167863845825195,
|
| 997 |
-
"rewards/margins": 2.327130079269409,
|
| 998 |
-
"rewards/rejected": -5.043917179107666,
|
| 999 |
-
"step": 325
|
| 1000 |
-
},
|
| 1001 |
-
{
|
| 1002 |
-
"epoch": 0.6586826347305389,
|
| 1003 |
-
"grad_norm": 80.80388415289183,
|
| 1004 |
-
"learning_rate": 2.527501789261288e-07,
|
| 1005 |
-
"logits/chosen": -1.7654087543487549,
|
| 1006 |
-
"logits/rejected": -1.7988303899765015,
|
| 1007 |
-
"logps/chosen": -1.2382586002349854,
|
| 1008 |
-
"logps/rejected": -1.5011346340179443,
|
| 1009 |
-
"loss": 0.3562,
|
| 1010 |
-
"rewards/accuracies": 0.8999999761581421,
|
| 1011 |
-
"rewards/chosen": -2.9657859802246094,
|
| 1012 |
-
"rewards/margins": 1.982195258140564,
|
| 1013 |
-
"rewards/rejected": -4.947981357574463,
|
| 1014 |
-
"step": 330
|
| 1015 |
-
},
|
| 1016 |
-
{
|
| 1017 |
-
"epoch": 0.6686626746506986,
|
| 1018 |
-
"grad_norm": 58.701745588033376,
|
| 1019 |
-
"learning_rate": 2.398603869772421e-07,
|
| 1020 |
-
"logits/chosen": -1.8312063217163086,
|
| 1021 |
-
"logits/rejected": -1.906640648841858,
|
| 1022 |
-
"logps/chosen": -1.2148807048797607,
|
| 1023 |
-
"logps/rejected": -1.609410047531128,
|
| 1024 |
-
"loss": 0.3771,
|
| 1025 |
-
"rewards/accuracies": 0.875,
|
| 1026 |
-
"rewards/chosen": -2.625844955444336,
|
| 1027 |
-
"rewards/margins": 2.595080614089966,
|
| 1028 |
-
"rewards/rejected": -5.220925807952881,
|
| 1029 |
-
"step": 335
|
| 1030 |
-
},
|
| 1031 |
-
{
|
| 1032 |
-
"epoch": 0.6786427145708582,
|
| 1033 |
-
"grad_norm": 53.98596461549184,
|
| 1034 |
-
"learning_rate": 2.27165700479207e-07,
|
| 1035 |
-
"logits/chosen": -1.7870060205459595,
|
| 1036 |
-
"logits/rejected": -1.8084951639175415,
|
| 1037 |
-
"logps/chosen": -1.24245023727417,
|
| 1038 |
-
"logps/rejected": -1.5464107990264893,
|
| 1039 |
-
"loss": 0.3441,
|
| 1040 |
-
"rewards/accuracies": 0.8125,
|
| 1041 |
-
"rewards/chosen": -3.0009231567382812,
|
| 1042 |
-
"rewards/margins": 2.439561605453491,
|
| 1043 |
-
"rewards/rejected": -5.440484523773193,
|
| 1044 |
-
"step": 340
|
| 1045 |
-
},
|
| 1046 |
-
{
|
| 1047 |
-
"epoch": 0.688622754491018,
|
| 1048 |
-
"grad_norm": 31.40057350192729,
|
| 1049 |
-
"learning_rate": 2.146815859520554e-07,
|
| 1050 |
-
"logits/chosen": -1.655821442604065,
|
| 1051 |
-
"logits/rejected": -1.7065486907958984,
|
| 1052 |
-
"logps/chosen": -1.1306984424591064,
|
| 1053 |
-
"logps/rejected": -1.4800045490264893,
|
| 1054 |
-
"loss": 0.3349,
|
| 1055 |
-
"rewards/accuracies": 0.8500000238418579,
|
| 1056 |
-
"rewards/chosen": -2.8754384517669678,
|
| 1057 |
-
"rewards/margins": 2.686849355697632,
|
| 1058 |
-
"rewards/rejected": -5.562288284301758,
|
| 1059 |
-
"step": 345
|
| 1060 |
-
},
|
| 1061 |
-
{
|
| 1062 |
-
"epoch": 0.6986027944111777,
|
| 1063 |
-
"grad_norm": 31.89392999133678,
|
| 1064 |
-
"learning_rate": 2.0242325336630769e-07,
|
| 1065 |
-
"logits/chosen": -1.770398497581482,
|
| 1066 |
-
"logits/rejected": -1.7964344024658203,
|
| 1067 |
-
"logps/chosen": -1.3358914852142334,
|
| 1068 |
-
"logps/rejected": -1.713321328163147,
|
| 1069 |
-
"loss": 0.3356,
|
| 1070 |
-
"rewards/accuracies": 0.824999988079071,
|
| 1071 |
-
"rewards/chosen": -3.2399375438690186,
|
| 1072 |
-
"rewards/margins": 2.4227969646453857,
|
| 1073 |
-
"rewards/rejected": -5.662735462188721,
|
| 1074 |
-
"step": 350
|
| 1075 |
-
},
|
| 1076 |
-
{
|
| 1077 |
-
"epoch": 0.7085828343313373,
|
| 1078 |
-
"grad_norm": 36.33203288290445,
|
| 1079 |
-
"learning_rate": 1.9040563761196839e-07,
|
| 1080 |
-
"logits/chosen": -1.8168394565582275,
|
| 1081 |
-
"logits/rejected": -1.877682089805603,
|
| 1082 |
-
"logps/chosen": -1.2325422763824463,
|
| 1083 |
-
"logps/rejected": -1.7707788944244385,
|
| 1084 |
-
"loss": 0.3245,
|
| 1085 |
-
"rewards/accuracies": 0.875,
|
| 1086 |
-
"rewards/chosen": -2.476655960083008,
|
| 1087 |
-
"rewards/margins": 3.2025744915008545,
|
| 1088 |
-
"rewards/rejected": -5.6792311668396,
|
| 1089 |
-
"step": 355
|
| 1090 |
-
},
|
| 1091 |
-
{
|
| 1092 |
-
"epoch": 0.718562874251497,
|
| 1093 |
-
"grad_norm": 43.416666324732795,
|
| 1094 |
-
"learning_rate": 1.786433803026624e-07,
|
| 1095 |
-
"logits/chosen": -1.8330310583114624,
|
| 1096 |
-
"logits/rejected": -1.8215529918670654,
|
| 1097 |
-
"logps/chosen": -1.3045927286148071,
|
| 1098 |
-
"logps/rejected": -1.523819088935852,
|
| 1099 |
-
"loss": 0.3075,
|
| 1100 |
-
"rewards/accuracies": 0.8374999761581421,
|
| 1101 |
-
"rewards/chosen": -3.1603009700775146,
|
| 1102 |
-
"rewards/margins": 1.9945179224014282,
|
| 1103 |
-
"rewards/rejected": -5.154818534851074,
|
| 1104 |
-
"step": 360
|
| 1105 |
-
},
|
| 1106 |
-
{
|
| 1107 |
-
"epoch": 0.7285429141716567,
|
| 1108 |
-
"grad_norm": 41.308059452990165,
|
| 1109 |
-
"learning_rate": 1.6715081193708426e-07,
|
| 1110 |
-
"logits/chosen": -1.8217226266860962,
|
| 1111 |
-
"logits/rejected": -1.7349847555160522,
|
| 1112 |
-
"logps/chosen": -1.3033679723739624,
|
| 1113 |
-
"logps/rejected": -1.5060852766036987,
|
| 1114 |
-
"loss": 0.3159,
|
| 1115 |
-
"rewards/accuracies": 0.824999988079071,
|
| 1116 |
-
"rewards/chosen": -3.0869052410125732,
|
| 1117 |
-
"rewards/margins": 2.2355074882507324,
|
| 1118 |
-
"rewards/rejected": -5.322413444519043,
|
| 1119 |
-
"step": 365
|
| 1120 |
-
},
|
| 1121 |
-
{
|
| 1122 |
-
"epoch": 0.7385229540918163,
|
| 1123 |
-
"grad_norm": 42.21028449654826,
|
| 1124 |
-
"learning_rate": 1.5594193443949295e-07,
|
| 1125 |
-
"logits/chosen": -1.8006670475006104,
|
| 1126 |
-
"logits/rejected": -1.7459598779678345,
|
| 1127 |
-
"logps/chosen": -1.3234542608261108,
|
| 1128 |
-
"logps/rejected": -1.5523171424865723,
|
| 1129 |
-
"loss": 0.3351,
|
| 1130 |
-
"rewards/accuracies": 0.8500000238418579,
|
| 1131 |
-
"rewards/chosen": -3.036769390106201,
|
| 1132 |
-
"rewards/margins": 2.387256145477295,
|
| 1133 |
-
"rewards/rejected": -5.424025535583496,
|
| 1134 |
-
"step": 370
|
| 1135 |
-
},
|
| 1136 |
-
{
|
| 1137 |
-
"epoch": 0.7485029940119761,
|
| 1138 |
-
"grad_norm": 19.788273374335468,
|
| 1139 |
-
"learning_rate": 1.450304041005241e-07,
|
| 1140 |
-
"logits/chosen": -1.7887513637542725,
|
| 1141 |
-
"logits/rejected": -1.7895513772964478,
|
| 1142 |
-
"logps/chosen": -1.2865995168685913,
|
| 1143 |
-
"logps/rejected": -1.5641953945159912,
|
| 1144 |
-
"loss": 0.3256,
|
| 1145 |
-
"rewards/accuracies": 0.862500011920929,
|
| 1146 |
-
"rewards/chosen": -2.724120616912842,
|
| 1147 |
-
"rewards/margins": 2.6693472862243652,
|
| 1148 |
-
"rewards/rejected": -5.393467903137207,
|
| 1149 |
-
"step": 375
|
| 1150 |
-
},
|
| 1151 |
-
{
|
| 1152 |
-
"epoch": 0.7584830339321357,
|
| 1153 |
-
"grad_norm": 22.398433956551006,
|
| 1154 |
-
"learning_rate": 1.3442951493910334e-07,
|
| 1155 |
-
"logits/chosen": -1.9577217102050781,
|
| 1156 |
-
"logits/rejected": -1.888419508934021,
|
| 1157 |
-
"logps/chosen": -1.1813005208969116,
|
| 1158 |
-
"logps/rejected": -1.487975835800171,
|
| 1159 |
-
"loss": 0.3656,
|
| 1160 |
-
"rewards/accuracies": 0.887499988079071,
|
| 1161 |
-
"rewards/chosen": -1.874016523361206,
|
| 1162 |
-
"rewards/margins": 2.6297459602355957,
|
| 1163 |
-
"rewards/rejected": -4.503762722015381,
|
| 1164 |
-
"step": 380
|
| 1165 |
-
},
|
| 1166 |
-
{
|
| 1167 |
-
"epoch": 0.7684630738522954,
|
| 1168 |
-
"grad_norm": 25.43555535966853,
|
| 1169 |
-
"learning_rate": 1.2415218250573207e-07,
|
| 1170 |
-
"logits/chosen": -1.8119207620620728,
|
| 1171 |
-
"logits/rejected": -1.808342695236206,
|
| 1172 |
-
"logps/chosen": -1.3371237516403198,
|
| 1173 |
-
"logps/rejected": -1.5443830490112305,
|
| 1174 |
-
"loss": 0.3467,
|
| 1175 |
-
"rewards/accuracies": 0.887499988079071,
|
| 1176 |
-
"rewards/chosen": -2.7301838397979736,
|
| 1177 |
-
"rewards/margins": 2.1141746044158936,
|
| 1178 |
-
"rewards/rejected": -4.844358921051025,
|
| 1179 |
-
"step": 385
|
| 1180 |
-
},
|
| 1181 |
-
{
|
| 1182 |
-
"epoch": 0.7784431137724551,
|
| 1183 |
-
"grad_norm": 42.565390920889676,
|
| 1184 |
-
"learning_rate": 1.1421092814687874e-07,
|
| 1185 |
-
"logits/chosen": -1.92118239402771,
|
| 1186 |
-
"logits/rejected": -1.880392074584961,
|
| 1187 |
-
"logps/chosen": -1.2233985662460327,
|
| 1188 |
-
"logps/rejected": -1.5754501819610596,
|
| 1189 |
-
"loss": 0.3129,
|
| 1190 |
-
"rewards/accuracies": 0.887499988079071,
|
| 1191 |
-
"rewards/chosen": -2.677239418029785,
|
| 1192 |
-
"rewards/margins": 2.564099073410034,
|
| 1193 |
-
"rewards/rejected": -5.24133825302124,
|
| 1194 |
-
"step": 390
|
| 1195 |
-
},
|
| 1196 |
-
{
|
| 1197 |
-
"epoch": 0.7884231536926147,
|
| 1198 |
-
"grad_norm": 35.381985967589934,
|
| 1199 |
-
"learning_rate": 1.0461786374964643e-07,
|
| 1200 |
-
"logits/chosen": -2.013378620147705,
|
| 1201 |
-
"logits/rejected": -1.9751570224761963,
|
| 1202 |
-
"logps/chosen": -1.449527382850647,
|
| 1203 |
-
"logps/rejected": -1.7529798746109009,
|
| 1204 |
-
"loss": 0.2973,
|
| 1205 |
-
"rewards/accuracies": 0.862500011920929,
|
| 1206 |
-
"rewards/chosen": -3.0077102184295654,
|
| 1207 |
-
"rewards/margins": 2.959484577178955,
|
| 1208 |
-
"rewards/rejected": -5.9671950340271,
|
| 1209 |
-
"step": 395
|
| 1210 |
-
},
|
| 1211 |
-
{
|
| 1212 |
-
"epoch": 0.7984031936127745,
|
| 1213 |
-
"grad_norm": 28.566925881122785,
|
| 1214 |
-
"learning_rate": 9.538467698530536e-08,
|
| 1215 |
-
"logits/chosen": -1.760108232498169,
|
| 1216 |
-
"logits/rejected": -1.8218498229980469,
|
| 1217 |
-
"logps/chosen": -1.2844475507736206,
|
| 1218 |
-
"logps/rejected": -1.5693128108978271,
|
| 1219 |
-
"loss": 0.299,
|
| 1220 |
-
"rewards/accuracies": 0.875,
|
| 1221 |
-
"rewards/chosen": -2.605151414871216,
|
| 1222 |
-
"rewards/margins": 2.5104212760925293,
|
| 1223 |
-
"rewards/rejected": -5.115572452545166,
|
| 1224 |
-
"step": 400
|
| 1225 |
-
}
|
| 1226 |
-
],
|
| 1227 |
-
"logging_steps": 5,
|
| 1228 |
-
"max_steps": 501,
|
| 1229 |
-
"num_input_tokens_seen": 0,
|
| 1230 |
-
"num_train_epochs": 1,
|
| 1231 |
-
"save_steps": 101,
|
| 1232 |
-
"stateful_callbacks": {
|
| 1233 |
-
"TrainerControl": {
|
| 1234 |
-
"args": {
|
| 1235 |
-
"should_epoch_stop": false,
|
| 1236 |
-
"should_evaluate": false,
|
| 1237 |
-
"should_log": false,
|
| 1238 |
-
"should_save": true,
|
| 1239 |
-
"should_training_stop": false
|
| 1240 |
-
},
|
| 1241 |
-
"attributes": {}
|
| 1242 |
-
}
|
| 1243 |
-
},
|
| 1244 |
-
"total_flos": 0.0,
|
| 1245 |
-
"train_batch_size": 2,
|
| 1246 |
-
"trial_name": null,
|
| 1247 |
-
"trial_params": null
|
| 1248 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|