Upload folder using huggingface_hub
Browse files- config.json +1 -1
- model.safetensors +1 -1
- trainer_state.json +651 -1471
config.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"_name_or_path": "
|
| 3 |
"architectures": [
|
| 4 |
"Qwen2ForCausalLM"
|
| 5 |
],
|
|
|
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "VAGOsolutions/SauerkrautLM-1.5b",
|
| 3 |
"architectures": [
|
| 4 |
"Qwen2ForCausalLM"
|
| 5 |
],
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3087467144
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:308b0909c8b941e293cffef3c22bb8ffbb5a9f307ddc235f9370c9ffaa10de45
|
| 3 |
size 3087467144
|
trainer_state.json
CHANGED
|
@@ -1,1516 +1,696 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 5000,
|
| 6 |
-
"global_step":
|
| 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.
|
| 13 |
-
"grad_norm":
|
| 14 |
-
"learning_rate":
|
| 15 |
-
"logits/chosen": -
|
| 16 |
-
"logits/rejected": -
|
| 17 |
-
"logps/chosen": -
|
| 18 |
-
"logps/rejected": -
|
| 19 |
-
"loss":
|
| 20 |
-
"
|
| 21 |
-
"rewards/
|
| 22 |
-
"rewards/
|
| 23 |
-
"rewards/
|
| 24 |
-
"
|
| 25 |
-
|
| 26 |
-
{
|
| 27 |
-
"epoch": 0.0005617755618852837,
|
| 28 |
-
"grad_norm": 1.90625,
|
| 29 |
-
"learning_rate": 2.8089887640449437e-07,
|
| 30 |
-
"logits/chosen": -6.981129169464111,
|
| 31 |
-
"logits/rejected": -6.981235027313232,
|
| 32 |
-
"logps/chosen": -0.5972335338592529,
|
| 33 |
-
"logps/rejected": -3.7509422302246094,
|
| 34 |
-
"loss": 0.8099,
|
| 35 |
-
"rewards/accuracies": 0.8203125,
|
| 36 |
-
"rewards/chosen": -0.059723351150751114,
|
| 37 |
-
"rewards/margins": 0.3153708577156067,
|
| 38 |
-
"rewards/rejected": -0.3750942647457123,
|
| 39 |
-
"step": 2
|
| 40 |
-
},
|
| 41 |
-
{
|
| 42 |
-
"epoch": 0.0008426633428279256,
|
| 43 |
-
"grad_norm": 1.328125,
|
| 44 |
-
"learning_rate": 4.213483146067416e-07,
|
| 45 |
-
"logits/chosen": -6.982983589172363,
|
| 46 |
-
"logits/rejected": -6.983180522918701,
|
| 47 |
-
"logps/chosen": -0.6279873847961426,
|
| 48 |
-
"logps/rejected": -3.3199615478515625,
|
| 49 |
-
"loss": 0.8299,
|
| 50 |
-
"rewards/accuracies": 0.8671875,
|
| 51 |
-
"rewards/chosen": -0.06279874593019485,
|
| 52 |
-
"rewards/margins": 0.26919740438461304,
|
| 53 |
-
"rewards/rejected": -0.3319961428642273,
|
| 54 |
-
"step": 3
|
| 55 |
-
},
|
| 56 |
-
{
|
| 57 |
-
"epoch": 0.0011235511237705673,
|
| 58 |
-
"grad_norm": 1.046875,
|
| 59 |
-
"learning_rate": 5.617977528089887e-07,
|
| 60 |
-
"logits/chosen": -6.978228569030762,
|
| 61 |
-
"logits/rejected": -6.978395462036133,
|
| 62 |
-
"logps/chosen": -0.663998007774353,
|
| 63 |
-
"logps/rejected": -3.635097026824951,
|
| 64 |
-
"loss": 0.818,
|
| 65 |
-
"rewards/accuracies": 0.9140625,
|
| 66 |
-
"rewards/chosen": -0.06639980524778366,
|
| 67 |
-
"rewards/margins": 0.2971099317073822,
|
| 68 |
-
"rewards/rejected": -0.3635097146034241,
|
| 69 |
-
"step": 4
|
| 70 |
-
},
|
| 71 |
-
{
|
| 72 |
-
"epoch": 0.0014044389047132091,
|
| 73 |
-
"grad_norm": 1.9921875,
|
| 74 |
-
"learning_rate": 7.02247191011236e-07,
|
| 75 |
-
"logits/chosen": -6.984366416931152,
|
| 76 |
-
"logits/rejected": -6.984519004821777,
|
| 77 |
-
"logps/chosen": -0.5929118990898132,
|
| 78 |
-
"logps/rejected": -3.171377420425415,
|
| 79 |
-
"loss": 0.833,
|
| 80 |
-
"rewards/accuracies": 0.828125,
|
| 81 |
-
"rewards/chosen": -0.05929119512438774,
|
| 82 |
-
"rewards/margins": 0.25784653425216675,
|
| 83 |
-
"rewards/rejected": -0.317137748003006,
|
| 84 |
-
"step": 5
|
| 85 |
-
},
|
| 86 |
-
{
|
| 87 |
-
"epoch": 0.0016853266856558511,
|
| 88 |
-
"grad_norm": 0.81640625,
|
| 89 |
-
"learning_rate": 8.426966292134832e-07,
|
| 90 |
-
"logits/chosen": -6.9845099449157715,
|
| 91 |
-
"logits/rejected": -6.9846272468566895,
|
| 92 |
-
"logps/chosen": -0.6544881463050842,
|
| 93 |
-
"logps/rejected": -3.7314887046813965,
|
| 94 |
-
"loss": 0.8135,
|
| 95 |
-
"rewards/accuracies": 0.8515625,
|
| 96 |
-
"rewards/chosen": -0.0654488131403923,
|
| 97 |
-
"rewards/margins": 0.3077000379562378,
|
| 98 |
-
"rewards/rejected": -0.3731488585472107,
|
| 99 |
-
"step": 6
|
| 100 |
-
},
|
| 101 |
-
{
|
| 102 |
-
"epoch": 0.001966214466598493,
|
| 103 |
-
"grad_norm": 2.0625,
|
| 104 |
-
"learning_rate": 9.831460674157304e-07,
|
| 105 |
-
"logits/chosen": -6.983869552612305,
|
| 106 |
-
"logits/rejected": -6.98401403427124,
|
| 107 |
-
"logps/chosen": -0.6330832839012146,
|
| 108 |
-
"logps/rejected": -3.969212055206299,
|
| 109 |
-
"loss": 0.7987,
|
| 110 |
-
"rewards/accuracies": 0.8984375,
|
| 111 |
-
"rewards/chosen": -0.06330832839012146,
|
| 112 |
-
"rewards/margins": 0.3336128890514374,
|
| 113 |
-
"rewards/rejected": -0.39692118763923645,
|
| 114 |
-
"step": 7
|
| 115 |
-
},
|
| 116 |
-
{
|
| 117 |
-
"epoch": 0.0022471022475411347,
|
| 118 |
-
"grad_norm": 1.2265625,
|
| 119 |
-
"learning_rate": 1.1235955056179775e-06,
|
| 120 |
-
"logits/chosen": -6.984411239624023,
|
| 121 |
-
"logits/rejected": -6.984560012817383,
|
| 122 |
-
"logps/chosen": -0.5015445351600647,
|
| 123 |
-
"logps/rejected": -3.5077507495880127,
|
| 124 |
-
"loss": 0.812,
|
| 125 |
-
"rewards/accuracies": 0.890625,
|
| 126 |
-
"rewards/chosen": -0.05015445500612259,
|
| 127 |
-
"rewards/margins": 0.30062058568000793,
|
| 128 |
-
"rewards/rejected": -0.3507750630378723,
|
| 129 |
-
"step": 8
|
| 130 |
-
},
|
| 131 |
-
{
|
| 132 |
-
"epoch": 0.0025279900284837765,
|
| 133 |
-
"grad_norm": 3.484375,
|
| 134 |
-
"learning_rate": 1.2640449438202247e-06,
|
| 135 |
-
"logits/chosen": -6.9832258224487305,
|
| 136 |
-
"logits/rejected": -6.9833831787109375,
|
| 137 |
-
"logps/chosen": -0.5916701555252075,
|
| 138 |
-
"logps/rejected": -3.901883125305176,
|
| 139 |
-
"loss": 0.7984,
|
| 140 |
-
"rewards/accuracies": 0.875,
|
| 141 |
-
"rewards/chosen": -0.05916702002286911,
|
| 142 |
-
"rewards/margins": 0.33102133870124817,
|
| 143 |
-
"rewards/rejected": -0.3901883363723755,
|
| 144 |
-
"step": 9
|
| 145 |
-
},
|
| 146 |
-
{
|
| 147 |
-
"epoch": 0.0028088778094264182,
|
| 148 |
-
"grad_norm": 1.6484375,
|
| 149 |
-
"learning_rate": 1.404494382022472e-06,
|
| 150 |
-
"logits/chosen": -6.982822418212891,
|
| 151 |
-
"logits/rejected": -6.982968330383301,
|
| 152 |
-
"logps/chosen": -0.6599135994911194,
|
| 153 |
-
"logps/rejected": -3.5486698150634766,
|
| 154 |
-
"loss": 0.8177,
|
| 155 |
-
"rewards/accuracies": 0.8671875,
|
| 156 |
-
"rewards/chosen": -0.0659913569688797,
|
| 157 |
-
"rewards/margins": 0.28887563943862915,
|
| 158 |
-
"rewards/rejected": -0.35486698150634766,
|
| 159 |
"step": 10
|
| 160 |
},
|
| 161 |
{
|
| 162 |
-
"epoch": 0.
|
| 163 |
-
"grad_norm":
|
| 164 |
-
"learning_rate":
|
| 165 |
-
"logits/chosen": -
|
| 166 |
-
"logits/rejected": -
|
| 167 |
-
"logps/chosen": -
|
| 168 |
-
"logps/rejected": -
|
| 169 |
-
"loss":
|
| 170 |
-
"
|
| 171 |
-
"rewards/
|
| 172 |
-
"rewards/
|
| 173 |
-
"rewards/
|
| 174 |
-
"
|
| 175 |
-
|
| 176 |
-
{
|
| 177 |
-
"epoch": 0.0033706533713117022,
|
| 178 |
-
"grad_norm": 1.546875,
|
| 179 |
-
"learning_rate": 1.6853932584269663e-06,
|
| 180 |
-
"logits/chosen": -6.9844279289245605,
|
| 181 |
-
"logits/rejected": -6.984550952911377,
|
| 182 |
-
"logps/chosen": -0.6450465321540833,
|
| 183 |
-
"logps/rejected": -3.3710899353027344,
|
| 184 |
-
"loss": 0.8281,
|
| 185 |
-
"rewards/accuracies": 0.859375,
|
| 186 |
-
"rewards/chosen": -0.06450465321540833,
|
| 187 |
-
"rewards/margins": 0.2726043164730072,
|
| 188 |
-
"rewards/rejected": -0.3371089994907379,
|
| 189 |
-
"step": 12
|
| 190 |
-
},
|
| 191 |
-
{
|
| 192 |
-
"epoch": 0.003651541152254344,
|
| 193 |
-
"grad_norm": 1.390625,
|
| 194 |
-
"learning_rate": 1.8258426966292136e-06,
|
| 195 |
-
"logits/chosen": -6.985495567321777,
|
| 196 |
-
"logits/rejected": -6.985678672790527,
|
| 197 |
-
"logps/chosen": -0.6660882234573364,
|
| 198 |
-
"logps/rejected": -3.941418170928955,
|
| 199 |
-
"loss": 0.8009,
|
| 200 |
-
"rewards/accuracies": 0.90625,
|
| 201 |
-
"rewards/chosen": -0.06660880893468857,
|
| 202 |
-
"rewards/margins": 0.3275330066680908,
|
| 203 |
-
"rewards/rejected": -0.3941417932510376,
|
| 204 |
-
"step": 13
|
| 205 |
-
},
|
| 206 |
-
{
|
| 207 |
-
"epoch": 0.003932428933196986,
|
| 208 |
-
"grad_norm": 1.6015625,
|
| 209 |
-
"learning_rate": 1.966292134831461e-06,
|
| 210 |
-
"logits/chosen": -6.982288837432861,
|
| 211 |
-
"logits/rejected": -6.982420921325684,
|
| 212 |
-
"logps/chosen": -0.7159062623977661,
|
| 213 |
-
"logps/rejected": -3.4875435829162598,
|
| 214 |
-
"loss": 0.8243,
|
| 215 |
-
"rewards/accuracies": 0.8984375,
|
| 216 |
-
"rewards/chosen": -0.07159063965082169,
|
| 217 |
-
"rewards/margins": 0.27716371417045593,
|
| 218 |
-
"rewards/rejected": -0.348754346370697,
|
| 219 |
-
"step": 14
|
| 220 |
-
},
|
| 221 |
-
{
|
| 222 |
-
"epoch": 0.004213316714139628,
|
| 223 |
-
"grad_norm": 1.4296875,
|
| 224 |
-
"learning_rate": 2.106741573033708e-06,
|
| 225 |
-
"logits/chosen": -6.986008644104004,
|
| 226 |
-
"logits/rejected": -6.986215114593506,
|
| 227 |
-
"logps/chosen": -0.6564575433731079,
|
| 228 |
-
"logps/rejected": -3.4806642532348633,
|
| 229 |
-
"loss": 0.8223,
|
| 230 |
-
"rewards/accuracies": 0.84375,
|
| 231 |
-
"rewards/chosen": -0.0656457468867302,
|
| 232 |
-
"rewards/margins": 0.28242069482803345,
|
| 233 |
-
"rewards/rejected": -0.34806641936302185,
|
| 234 |
-
"step": 15
|
| 235 |
-
},
|
| 236 |
-
{
|
| 237 |
-
"epoch": 0.004494204495082269,
|
| 238 |
-
"grad_norm": 0.94921875,
|
| 239 |
-
"learning_rate": 2.247191011235955e-06,
|
| 240 |
-
"logits/chosen": -6.985368251800537,
|
| 241 |
-
"logits/rejected": -6.985543727874756,
|
| 242 |
-
"logps/chosen": -0.5949140787124634,
|
| 243 |
-
"logps/rejected": -3.1833853721618652,
|
| 244 |
-
"loss": 0.8354,
|
| 245 |
-
"rewards/accuracies": 0.875,
|
| 246 |
-
"rewards/chosen": -0.05949141085147858,
|
| 247 |
-
"rewards/margins": 0.25884711742401123,
|
| 248 |
-
"rewards/rejected": -0.3183385133743286,
|
| 249 |
-
"step": 16
|
| 250 |
-
},
|
| 251 |
-
{
|
| 252 |
-
"epoch": 0.004775092276024911,
|
| 253 |
-
"grad_norm": 1.53125,
|
| 254 |
-
"learning_rate": 2.3876404494382022e-06,
|
| 255 |
-
"logits/chosen": -6.983895778656006,
|
| 256 |
-
"logits/rejected": -6.984076023101807,
|
| 257 |
-
"logps/chosen": -0.6407002806663513,
|
| 258 |
-
"logps/rejected": -3.6408045291900635,
|
| 259 |
-
"loss": 0.8183,
|
| 260 |
-
"rewards/accuracies": 0.8671875,
|
| 261 |
-
"rewards/chosen": -0.06407003104686737,
|
| 262 |
-
"rewards/margins": 0.30001044273376465,
|
| 263 |
-
"rewards/rejected": -0.3640804588794708,
|
| 264 |
-
"step": 17
|
| 265 |
-
},
|
| 266 |
-
{
|
| 267 |
-
"epoch": 0.005055980056967553,
|
| 268 |
-
"grad_norm": 0.91796875,
|
| 269 |
-
"learning_rate": 2.5280898876404495e-06,
|
| 270 |
-
"logits/chosen": -6.984615802764893,
|
| 271 |
-
"logits/rejected": -6.984807014465332,
|
| 272 |
-
"logps/chosen": -0.5760031938552856,
|
| 273 |
-
"logps/rejected": -2.942106008529663,
|
| 274 |
-
"loss": 0.8485,
|
| 275 |
-
"rewards/accuracies": 0.828125,
|
| 276 |
-
"rewards/chosen": -0.057600319385528564,
|
| 277 |
-
"rewards/margins": 0.2366103082895279,
|
| 278 |
-
"rewards/rejected": -0.29421064257621765,
|
| 279 |
-
"step": 18
|
| 280 |
-
},
|
| 281 |
-
{
|
| 282 |
-
"epoch": 0.005336867837910195,
|
| 283 |
-
"grad_norm": 2.765625,
|
| 284 |
-
"learning_rate": 2.6685393258426968e-06,
|
| 285 |
-
"logits/chosen": -6.985834121704102,
|
| 286 |
-
"logits/rejected": -6.986002445220947,
|
| 287 |
-
"logps/chosen": -0.603966474533081,
|
| 288 |
-
"logps/rejected": -3.6589133739471436,
|
| 289 |
-
"loss": 0.8121,
|
| 290 |
-
"rewards/accuracies": 0.8984375,
|
| 291 |
-
"rewards/chosen": -0.06039665639400482,
|
| 292 |
-
"rewards/margins": 0.3054946959018707,
|
| 293 |
-
"rewards/rejected": -0.36589139699935913,
|
| 294 |
-
"step": 19
|
| 295 |
-
},
|
| 296 |
-
{
|
| 297 |
-
"epoch": 0.0056177556188528365,
|
| 298 |
-
"grad_norm": 1.1484375,
|
| 299 |
-
"learning_rate": 2.808988764044944e-06,
|
| 300 |
-
"logits/chosen": -6.985572814941406,
|
| 301 |
-
"logits/rejected": -6.985714912414551,
|
| 302 |
-
"logps/chosen": -0.520312488079071,
|
| 303 |
-
"logps/rejected": -2.815075159072876,
|
| 304 |
-
"loss": 0.8505,
|
| 305 |
-
"rewards/accuracies": 0.8046875,
|
| 306 |
-
"rewards/chosen": -0.0520312525331974,
|
| 307 |
-
"rewards/margins": 0.22947625815868378,
|
| 308 |
-
"rewards/rejected": -0.2815075218677521,
|
| 309 |
"step": 20
|
| 310 |
},
|
| 311 |
{
|
| 312 |
-
"epoch": 0.
|
| 313 |
-
"grad_norm":
|
| 314 |
-
"learning_rate":
|
| 315 |
-
"logits/chosen": -
|
| 316 |
-
"logits/rejected": -
|
| 317 |
-
"logps/chosen": -0.
|
| 318 |
-
"logps/rejected": -
|
| 319 |
-
"loss": 0.
|
| 320 |
-
"
|
| 321 |
-
"rewards/
|
| 322 |
-
"rewards/
|
| 323 |
-
"rewards/
|
| 324 |
-
"
|
| 325 |
-
|
| 326 |
-
{
|
| 327 |
-
"epoch": 0.00617953118073812,
|
| 328 |
-
"grad_norm": 1.359375,
|
| 329 |
-
"learning_rate": 3.089887640449438e-06,
|
| 330 |
-
"logits/chosen": -6.985095500946045,
|
| 331 |
-
"logits/rejected": -6.985231876373291,
|
| 332 |
-
"logps/chosen": -0.6375535726547241,
|
| 333 |
-
"logps/rejected": -3.722036600112915,
|
| 334 |
-
"loss": 0.8155,
|
| 335 |
-
"rewards/accuracies": 0.8125,
|
| 336 |
-
"rewards/chosen": -0.06375535577535629,
|
| 337 |
-
"rewards/margins": 0.30844831466674805,
|
| 338 |
-
"rewards/rejected": -0.37220367789268494,
|
| 339 |
-
"step": 22
|
| 340 |
-
},
|
| 341 |
-
{
|
| 342 |
-
"epoch": 0.006460418961680763,
|
| 343 |
-
"grad_norm": 1.3359375,
|
| 344 |
-
"learning_rate": 3.230337078651686e-06,
|
| 345 |
-
"logits/chosen": -6.985586166381836,
|
| 346 |
-
"logits/rejected": -6.985759258270264,
|
| 347 |
-
"logps/chosen": -0.6368163824081421,
|
| 348 |
-
"logps/rejected": -3.551453113555908,
|
| 349 |
-
"loss": 0.8186,
|
| 350 |
-
"rewards/accuracies": 0.828125,
|
| 351 |
-
"rewards/chosen": -0.06368163973093033,
|
| 352 |
-
"rewards/margins": 0.2914636731147766,
|
| 353 |
-
"rewards/rejected": -0.35514530539512634,
|
| 354 |
-
"step": 23
|
| 355 |
-
},
|
| 356 |
-
{
|
| 357 |
-
"epoch": 0.0067413067426234045,
|
| 358 |
-
"grad_norm": 0.9453125,
|
| 359 |
-
"learning_rate": 3.3707865168539327e-06,
|
| 360 |
-
"logits/chosen": -6.983944892883301,
|
| 361 |
-
"logits/rejected": -6.984124660491943,
|
| 362 |
-
"logps/chosen": -0.6929716467857361,
|
| 363 |
-
"logps/rejected": -3.0338144302368164,
|
| 364 |
-
"loss": 0.8463,
|
| 365 |
-
"rewards/accuracies": 0.828125,
|
| 366 |
-
"rewards/chosen": -0.06929716467857361,
|
| 367 |
-
"rewards/margins": 0.23408427834510803,
|
| 368 |
-
"rewards/rejected": -0.30338141322135925,
|
| 369 |
-
"step": 24
|
| 370 |
-
},
|
| 371 |
-
{
|
| 372 |
-
"epoch": 0.007022194523566046,
|
| 373 |
-
"grad_norm": 2.15625,
|
| 374 |
-
"learning_rate": 3.5112359550561803e-06,
|
| 375 |
-
"logits/chosen": -6.988375663757324,
|
| 376 |
-
"logits/rejected": -6.988600730895996,
|
| 377 |
-
"logps/chosen": -0.5282725691795349,
|
| 378 |
-
"logps/rejected": -2.989157199859619,
|
| 379 |
-
"loss": 0.84,
|
| 380 |
-
"rewards/accuracies": 0.8671875,
|
| 381 |
-
"rewards/chosen": -0.05282726138830185,
|
| 382 |
-
"rewards/margins": 0.24608850479125977,
|
| 383 |
-
"rewards/rejected": -0.2989157438278198,
|
| 384 |
-
"step": 25
|
| 385 |
-
},
|
| 386 |
-
{
|
| 387 |
-
"epoch": 0.007303082304508688,
|
| 388 |
-
"grad_norm": 0.90234375,
|
| 389 |
-
"learning_rate": 3.651685393258427e-06,
|
| 390 |
-
"logits/chosen": -6.9826226234436035,
|
| 391 |
-
"logits/rejected": -6.982814311981201,
|
| 392 |
-
"logps/chosen": -0.576172411441803,
|
| 393 |
-
"logps/rejected": -3.1249589920043945,
|
| 394 |
-
"loss": 0.8391,
|
| 395 |
-
"rewards/accuracies": 0.8984375,
|
| 396 |
-
"rewards/chosen": -0.05761724337935448,
|
| 397 |
-
"rewards/margins": 0.2548786401748657,
|
| 398 |
-
"rewards/rejected": -0.3124958872795105,
|
| 399 |
-
"step": 26
|
| 400 |
-
},
|
| 401 |
-
{
|
| 402 |
-
"epoch": 0.00758397008545133,
|
| 403 |
-
"grad_norm": 0.8984375,
|
| 404 |
-
"learning_rate": 3.7921348314606744e-06,
|
| 405 |
-
"logits/chosen": -6.98524284362793,
|
| 406 |
-
"logits/rejected": -6.985445499420166,
|
| 407 |
-
"logps/chosen": -0.567965567111969,
|
| 408 |
-
"logps/rejected": -3.0560202598571777,
|
| 409 |
-
"loss": 0.8399,
|
| 410 |
-
"rewards/accuracies": 0.890625,
|
| 411 |
-
"rewards/chosen": -0.05679655820131302,
|
| 412 |
-
"rewards/margins": 0.2488054633140564,
|
| 413 |
-
"rewards/rejected": -0.3056020438671112,
|
| 414 |
-
"step": 27
|
| 415 |
-
},
|
| 416 |
-
{
|
| 417 |
-
"epoch": 0.007864857866393972,
|
| 418 |
-
"grad_norm": 2.046875,
|
| 419 |
-
"learning_rate": 3.932584269662922e-06,
|
| 420 |
-
"logits/chosen": -6.987497806549072,
|
| 421 |
-
"logits/rejected": -6.9876604080200195,
|
| 422 |
-
"logps/chosen": -0.671168863773346,
|
| 423 |
-
"logps/rejected": -3.572070837020874,
|
| 424 |
-
"loss": 0.8193,
|
| 425 |
-
"rewards/accuracies": 0.8515625,
|
| 426 |
-
"rewards/chosen": -0.0671168863773346,
|
| 427 |
-
"rewards/margins": 0.2900902330875397,
|
| 428 |
-
"rewards/rejected": -0.3572070896625519,
|
| 429 |
-
"step": 28
|
| 430 |
-
},
|
| 431 |
-
{
|
| 432 |
-
"epoch": 0.008145745647336613,
|
| 433 |
-
"grad_norm": 1.7578125,
|
| 434 |
-
"learning_rate": 4.073033707865169e-06,
|
| 435 |
-
"logits/chosen": -6.984347343444824,
|
| 436 |
-
"logits/rejected": -6.98446798324585,
|
| 437 |
-
"logps/chosen": -0.7242225408554077,
|
| 438 |
-
"logps/rejected": -3.0529520511627197,
|
| 439 |
-
"loss": 0.8501,
|
| 440 |
-
"rewards/accuracies": 0.8671875,
|
| 441 |
-
"rewards/chosen": -0.07242225855588913,
|
| 442 |
-
"rewards/margins": 0.23287296295166016,
|
| 443 |
-
"rewards/rejected": -0.30529525876045227,
|
| 444 |
-
"step": 29
|
| 445 |
-
},
|
| 446 |
-
{
|
| 447 |
-
"epoch": 0.008426633428279255,
|
| 448 |
-
"grad_norm": 0.90234375,
|
| 449 |
-
"learning_rate": 4.213483146067416e-06,
|
| 450 |
-
"logits/chosen": -6.987656116485596,
|
| 451 |
-
"logits/rejected": -6.987755298614502,
|
| 452 |
-
"logps/chosen": -0.5501436591148376,
|
| 453 |
-
"logps/rejected": -2.847120761871338,
|
| 454 |
-
"loss": 0.8488,
|
| 455 |
-
"rewards/accuracies": 0.875,
|
| 456 |
-
"rewards/chosen": -0.05501437187194824,
|
| 457 |
-
"rewards/margins": 0.22969767451286316,
|
| 458 |
-
"rewards/rejected": -0.2847120463848114,
|
| 459 |
"step": 30
|
| 460 |
},
|
| 461 |
{
|
| 462 |
-
"epoch": 0.
|
| 463 |
-
"grad_norm":
|
| 464 |
-
"learning_rate":
|
| 465 |
-
"logits/chosen": -
|
| 466 |
-
"logits/rejected": -
|
| 467 |
-
"logps/chosen": -0.
|
| 468 |
-
"logps/rejected": -
|
| 469 |
-
"loss": 0.
|
| 470 |
-
"
|
| 471 |
-
"rewards/
|
| 472 |
-
"rewards/
|
| 473 |
-
"rewards/
|
| 474 |
-
"
|
| 475 |
-
|
| 476 |
-
{
|
| 477 |
-
"epoch": 0.008988408990164539,
|
| 478 |
-
"grad_norm": 1.28125,
|
| 479 |
-
"learning_rate": 4.49438202247191e-06,
|
| 480 |
-
"logits/chosen": -6.984766960144043,
|
| 481 |
-
"logits/rejected": -6.984991073608398,
|
| 482 |
-
"logps/chosen": -0.5861184597015381,
|
| 483 |
-
"logps/rejected": -3.729720115661621,
|
| 484 |
-
"loss": 0.8089,
|
| 485 |
-
"rewards/accuracies": 0.9140625,
|
| 486 |
-
"rewards/chosen": -0.05861184746026993,
|
| 487 |
-
"rewards/margins": 0.3143601715564728,
|
| 488 |
-
"rewards/rejected": -0.3729720413684845,
|
| 489 |
-
"step": 32
|
| 490 |
-
},
|
| 491 |
-
{
|
| 492 |
-
"epoch": 0.009269296771107181,
|
| 493 |
-
"grad_norm": 1.046875,
|
| 494 |
-
"learning_rate": 4.634831460674158e-06,
|
| 495 |
-
"logits/chosen": -6.988133907318115,
|
| 496 |
-
"logits/rejected": -6.988280296325684,
|
| 497 |
-
"logps/chosen": -0.6212756037712097,
|
| 498 |
-
"logps/rejected": -3.2797980308532715,
|
| 499 |
-
"loss": 0.8298,
|
| 500 |
-
"rewards/accuracies": 0.859375,
|
| 501 |
-
"rewards/chosen": -0.06212755665183067,
|
| 502 |
-
"rewards/margins": 0.26585227251052856,
|
| 503 |
-
"rewards/rejected": -0.32797983288764954,
|
| 504 |
-
"step": 33
|
| 505 |
-
},
|
| 506 |
-
{
|
| 507 |
-
"epoch": 0.009550184552049822,
|
| 508 |
-
"grad_norm": 1.21875,
|
| 509 |
-
"learning_rate": 4.7752808988764044e-06,
|
| 510 |
-
"logits/chosen": -6.988953113555908,
|
| 511 |
-
"logits/rejected": -6.989142894744873,
|
| 512 |
-
"logps/chosen": -0.5206266641616821,
|
| 513 |
-
"logps/rejected": -3.0407469272613525,
|
| 514 |
-
"loss": 0.8377,
|
| 515 |
-
"rewards/accuracies": 0.875,
|
| 516 |
-
"rewards/chosen": -0.05206267163157463,
|
| 517 |
-
"rewards/margins": 0.2520120143890381,
|
| 518 |
-
"rewards/rejected": -0.3040746748447418,
|
| 519 |
-
"step": 34
|
| 520 |
-
},
|
| 521 |
-
{
|
| 522 |
-
"epoch": 0.009831072332992465,
|
| 523 |
-
"grad_norm": 1.0390625,
|
| 524 |
-
"learning_rate": 4.915730337078652e-06,
|
| 525 |
-
"logits/chosen": -6.989229202270508,
|
| 526 |
-
"logits/rejected": -6.989412307739258,
|
| 527 |
-
"logps/chosen": -0.5842424631118774,
|
| 528 |
-
"logps/rejected": -3.2295570373535156,
|
| 529 |
-
"loss": 0.8331,
|
| 530 |
-
"rewards/accuracies": 0.8828125,
|
| 531 |
-
"rewards/chosen": -0.05842424929141998,
|
| 532 |
-
"rewards/margins": 0.2645314335823059,
|
| 533 |
-
"rewards/rejected": -0.3229556977748871,
|
| 534 |
-
"step": 35
|
| 535 |
-
},
|
| 536 |
-
{
|
| 537 |
-
"epoch": 0.010111960113935106,
|
| 538 |
-
"grad_norm": 1.6015625,
|
| 539 |
-
"learning_rate": 5.056179775280899e-06,
|
| 540 |
-
"logits/chosen": -6.989651203155518,
|
| 541 |
-
"logits/rejected": -6.989835739135742,
|
| 542 |
-
"logps/chosen": -0.5448687672615051,
|
| 543 |
-
"logps/rejected": -3.445011615753174,
|
| 544 |
-
"loss": 0.8174,
|
| 545 |
-
"rewards/accuracies": 0.875,
|
| 546 |
-
"rewards/chosen": -0.05448687821626663,
|
| 547 |
-
"rewards/margins": 0.2900142967700958,
|
| 548 |
-
"rewards/rejected": -0.34450116753578186,
|
| 549 |
-
"step": 36
|
| 550 |
-
},
|
| 551 |
-
{
|
| 552 |
-
"epoch": 0.010392847894877748,
|
| 553 |
-
"grad_norm": 0.97265625,
|
| 554 |
-
"learning_rate": 5.196629213483146e-06,
|
| 555 |
-
"logits/chosen": -6.987741470336914,
|
| 556 |
-
"logits/rejected": -6.987899303436279,
|
| 557 |
-
"logps/chosen": -0.6063218116760254,
|
| 558 |
-
"logps/rejected": -3.657203197479248,
|
| 559 |
-
"loss": 0.8103,
|
| 560 |
-
"rewards/accuracies": 0.8671875,
|
| 561 |
-
"rewards/chosen": -0.06063217669725418,
|
| 562 |
-
"rewards/margins": 0.3050881326198578,
|
| 563 |
-
"rewards/rejected": -0.36572033166885376,
|
| 564 |
-
"step": 37
|
| 565 |
-
},
|
| 566 |
-
{
|
| 567 |
-
"epoch": 0.01067373567582039,
|
| 568 |
-
"grad_norm": 1.3125,
|
| 569 |
-
"learning_rate": 5.3370786516853935e-06,
|
| 570 |
-
"logits/chosen": -6.992892265319824,
|
| 571 |
-
"logits/rejected": -6.993043899536133,
|
| 572 |
-
"logps/chosen": -0.6687143445014954,
|
| 573 |
-
"logps/rejected": -3.4205684661865234,
|
| 574 |
-
"loss": 0.8262,
|
| 575 |
-
"rewards/accuracies": 0.921875,
|
| 576 |
-
"rewards/chosen": -0.06687143445014954,
|
| 577 |
-
"rewards/margins": 0.2751854360103607,
|
| 578 |
-
"rewards/rejected": -0.34205687046051025,
|
| 579 |
-
"step": 38
|
| 580 |
-
},
|
| 581 |
-
{
|
| 582 |
-
"epoch": 0.010954623456763032,
|
| 583 |
-
"grad_norm": 1.4453125,
|
| 584 |
-
"learning_rate": 5.477528089887641e-06,
|
| 585 |
-
"logits/chosen": -6.9873948097229,
|
| 586 |
-
"logits/rejected": -6.987576484680176,
|
| 587 |
-
"logps/chosen": -0.6130951046943665,
|
| 588 |
-
"logps/rejected": -3.9130940437316895,
|
| 589 |
-
"loss": 0.8023,
|
| 590 |
-
"rewards/accuracies": 0.8828125,
|
| 591 |
-
"rewards/chosen": -0.061309512704610825,
|
| 592 |
-
"rewards/margins": 0.3299998939037323,
|
| 593 |
-
"rewards/rejected": -0.3913094103336334,
|
| 594 |
-
"step": 39
|
| 595 |
-
},
|
| 596 |
-
{
|
| 597 |
-
"epoch": 0.011235511237705673,
|
| 598 |
-
"grad_norm": 0.98828125,
|
| 599 |
-
"learning_rate": 5.617977528089888e-06,
|
| 600 |
-
"logits/chosen": -6.99265718460083,
|
| 601 |
-
"logits/rejected": -6.992844104766846,
|
| 602 |
-
"logps/chosen": -0.6354062557220459,
|
| 603 |
-
"logps/rejected": -2.818499803543091,
|
| 604 |
-
"loss": 0.8565,
|
| 605 |
-
"rewards/accuracies": 0.7890625,
|
| 606 |
-
"rewards/chosen": -0.06354063004255295,
|
| 607 |
-
"rewards/margins": 0.21830937266349792,
|
| 608 |
-
"rewards/rejected": -0.2818499803543091,
|
| 609 |
"step": 40
|
| 610 |
},
|
| 611 |
{
|
| 612 |
-
"epoch": 0.
|
| 613 |
-
"grad_norm":
|
| 614 |
-
"learning_rate":
|
| 615 |
-
"logits/chosen": -
|
| 616 |
-
"logits/rejected": -
|
| 617 |
-
"logps/chosen": -0.
|
| 618 |
-
"logps/rejected": -
|
| 619 |
-
"loss": 0.
|
| 620 |
-
"
|
| 621 |
-
"rewards/
|
| 622 |
-
"rewards/
|
| 623 |
-
"rewards/
|
| 624 |
-
"
|
| 625 |
-
|
| 626 |
-
{
|
| 627 |
-
"epoch": 0.011797286799590957,
|
| 628 |
-
"grad_norm": 0.77734375,
|
| 629 |
-
"learning_rate": 5.8988764044943826e-06,
|
| 630 |
-
"logits/chosen": -6.996128082275391,
|
| 631 |
-
"logits/rejected": -6.996224403381348,
|
| 632 |
-
"logps/chosen": -0.666965901851654,
|
| 633 |
-
"logps/rejected": -3.1539041996002197,
|
| 634 |
-
"loss": 0.8416,
|
| 635 |
-
"rewards/accuracies": 0.796875,
|
| 636 |
-
"rewards/chosen": -0.06669659167528152,
|
| 637 |
-
"rewards/margins": 0.2486938238143921,
|
| 638 |
-
"rewards/rejected": -0.315390408039093,
|
| 639 |
-
"step": 42
|
| 640 |
-
},
|
| 641 |
-
{
|
| 642 |
-
"epoch": 0.0120781745805336,
|
| 643 |
-
"grad_norm": 3.265625,
|
| 644 |
-
"learning_rate": 6.03932584269663e-06,
|
| 645 |
-
"logits/chosen": -6.990922927856445,
|
| 646 |
-
"logits/rejected": -6.991119861602783,
|
| 647 |
-
"logps/chosen": -0.6920310258865356,
|
| 648 |
-
"logps/rejected": -3.7208621501922607,
|
| 649 |
-
"loss": 0.8148,
|
| 650 |
-
"rewards/accuracies": 0.8203125,
|
| 651 |
-
"rewards/chosen": -0.06920310109853745,
|
| 652 |
-
"rewards/margins": 0.3028830885887146,
|
| 653 |
-
"rewards/rejected": -0.37208619713783264,
|
| 654 |
-
"step": 43
|
| 655 |
-
},
|
| 656 |
-
{
|
| 657 |
-
"epoch": 0.01235906236147624,
|
| 658 |
-
"grad_norm": 0.953125,
|
| 659 |
-
"learning_rate": 6.179775280898876e-06,
|
| 660 |
-
"logits/chosen": -6.995143413543701,
|
| 661 |
-
"logits/rejected": -6.995302200317383,
|
| 662 |
-
"logps/chosen": -0.6558566093444824,
|
| 663 |
-
"logps/rejected": -3.2403218746185303,
|
| 664 |
-
"loss": 0.8344,
|
| 665 |
-
"rewards/accuracies": 0.8359375,
|
| 666 |
-
"rewards/chosen": -0.065585657954216,
|
| 667 |
-
"rewards/margins": 0.2584465444087982,
|
| 668 |
-
"rewards/rejected": -0.3240322172641754,
|
| 669 |
-
"step": 44
|
| 670 |
-
},
|
| 671 |
-
{
|
| 672 |
-
"epoch": 0.012639950142418883,
|
| 673 |
-
"grad_norm": 0.94921875,
|
| 674 |
-
"learning_rate": 6.320224719101124e-06,
|
| 675 |
-
"logits/chosen": -6.992753028869629,
|
| 676 |
-
"logits/rejected": -6.992949962615967,
|
| 677 |
-
"logps/chosen": -0.6340925097465515,
|
| 678 |
-
"logps/rejected": -4.1242756843566895,
|
| 679 |
-
"loss": 0.7913,
|
| 680 |
-
"rewards/accuracies": 0.84375,
|
| 681 |
-
"rewards/chosen": -0.06340925395488739,
|
| 682 |
-
"rewards/margins": 0.34901827573776245,
|
| 683 |
-
"rewards/rejected": -0.41242751479148865,
|
| 684 |
-
"step": 45
|
| 685 |
-
},
|
| 686 |
-
{
|
| 687 |
-
"epoch": 0.012920837923361525,
|
| 688 |
-
"grad_norm": 1.140625,
|
| 689 |
-
"learning_rate": 6.460674157303372e-06,
|
| 690 |
-
"logits/chosen": -6.998230934143066,
|
| 691 |
-
"logits/rejected": -6.998466491699219,
|
| 692 |
-
"logps/chosen": -0.48865756392478943,
|
| 693 |
-
"logps/rejected": -3.8250622749328613,
|
| 694 |
-
"loss": 0.7962,
|
| 695 |
-
"rewards/accuracies": 0.9296875,
|
| 696 |
-
"rewards/chosen": -0.048865754157304764,
|
| 697 |
-
"rewards/margins": 0.333640456199646,
|
| 698 |
-
"rewards/rejected": -0.38250622153282166,
|
| 699 |
-
"step": 46
|
| 700 |
-
},
|
| 701 |
-
{
|
| 702 |
-
"epoch": 0.013201725704304166,
|
| 703 |
-
"grad_norm": 2.296875,
|
| 704 |
-
"learning_rate": 6.601123595505618e-06,
|
| 705 |
-
"logits/chosen": -6.998146057128906,
|
| 706 |
-
"logits/rejected": -6.998294830322266,
|
| 707 |
-
"logps/chosen": -0.6883760690689087,
|
| 708 |
-
"logps/rejected": -3.535900831222534,
|
| 709 |
-
"loss": 0.8262,
|
| 710 |
-
"rewards/accuracies": 0.796875,
|
| 711 |
-
"rewards/chosen": -0.06883761286735535,
|
| 712 |
-
"rewards/margins": 0.2847524583339691,
|
| 713 |
-
"rewards/rejected": -0.35359007120132446,
|
| 714 |
-
"step": 47
|
| 715 |
-
},
|
| 716 |
-
{
|
| 717 |
-
"epoch": 0.013482613485246809,
|
| 718 |
-
"grad_norm": 1.0,
|
| 719 |
-
"learning_rate": 6.741573033707865e-06,
|
| 720 |
-
"logits/chosen": -6.998913288116455,
|
| 721 |
-
"logits/rejected": -6.9990620613098145,
|
| 722 |
-
"logps/chosen": -0.5589190721511841,
|
| 723 |
-
"logps/rejected": -2.9924392700195312,
|
| 724 |
-
"loss": 0.8434,
|
| 725 |
-
"rewards/accuracies": 0.8671875,
|
| 726 |
-
"rewards/chosen": -0.055891912430524826,
|
| 727 |
-
"rewards/margins": 0.2433520257472992,
|
| 728 |
-
"rewards/rejected": -0.2992439270019531,
|
| 729 |
-
"step": 48
|
| 730 |
-
},
|
| 731 |
-
{
|
| 732 |
-
"epoch": 0.01376350126618945,
|
| 733 |
-
"grad_norm": 2.234375,
|
| 734 |
-
"learning_rate": 6.8820224719101126e-06,
|
| 735 |
-
"logits/chosen": -6.999466419219971,
|
| 736 |
-
"logits/rejected": -6.999658107757568,
|
| 737 |
-
"logps/chosen": -0.5700675845146179,
|
| 738 |
-
"logps/rejected": -3.315387487411499,
|
| 739 |
-
"loss": 0.8248,
|
| 740 |
-
"rewards/accuracies": 0.890625,
|
| 741 |
-
"rewards/chosen": -0.05700676143169403,
|
| 742 |
-
"rewards/margins": 0.2745319902896881,
|
| 743 |
-
"rewards/rejected": -0.33153873682022095,
|
| 744 |
-
"step": 49
|
| 745 |
-
},
|
| 746 |
-
{
|
| 747 |
-
"epoch": 0.014044389047132093,
|
| 748 |
-
"grad_norm": 1.078125,
|
| 749 |
-
"learning_rate": 7.022471910112361e-06,
|
| 750 |
-
"logits/chosen": -7.00152587890625,
|
| 751 |
-
"logits/rejected": -7.001701831817627,
|
| 752 |
-
"logps/chosen": -0.6014066934585571,
|
| 753 |
-
"logps/rejected": -3.370955228805542,
|
| 754 |
-
"loss": 0.8292,
|
| 755 |
-
"rewards/accuracies": 0.8125,
|
| 756 |
-
"rewards/chosen": -0.060140661895275116,
|
| 757 |
-
"rewards/margins": 0.27695485949516296,
|
| 758 |
-
"rewards/rejected": -0.3370955288410187,
|
| 759 |
"step": 50
|
| 760 |
},
|
| 761 |
{
|
| 762 |
-
"epoch": 0.
|
| 763 |
-
"grad_norm":
|
| 764 |
-
"learning_rate":
|
| 765 |
-
"logits/chosen": -
|
| 766 |
-
"logits/rejected": -
|
| 767 |
-
"logps/chosen": -0.
|
| 768 |
-
"logps/rejected": -
|
| 769 |
-
"loss": 0.
|
| 770 |
-
"
|
| 771 |
-
"rewards/
|
| 772 |
-
"rewards/
|
| 773 |
-
"rewards/
|
| 774 |
-
"
|
| 775 |
-
|
| 776 |
-
{
|
| 777 |
-
"epoch": 0.014606164609017376,
|
| 778 |
-
"grad_norm": 1.0703125,
|
| 779 |
-
"learning_rate": 7.303370786516854e-06,
|
| 780 |
-
"logits/chosen": -7.000398635864258,
|
| 781 |
-
"logits/rejected": -7.000590801239014,
|
| 782 |
-
"logps/chosen": -0.650733470916748,
|
| 783 |
-
"logps/rejected": -3.551283359527588,
|
| 784 |
-
"loss": 0.8207,
|
| 785 |
-
"rewards/accuracies": 0.859375,
|
| 786 |
-
"rewards/chosen": -0.06507335603237152,
|
| 787 |
-
"rewards/margins": 0.29005497694015503,
|
| 788 |
-
"rewards/rejected": -0.35512834787368774,
|
| 789 |
-
"step": 52
|
| 790 |
-
},
|
| 791 |
-
{
|
| 792 |
-
"epoch": 0.014887052389960017,
|
| 793 |
-
"grad_norm": 4.78125,
|
| 794 |
-
"learning_rate": 7.443820224719102e-06,
|
| 795 |
-
"logits/chosen": -7.001314640045166,
|
| 796 |
-
"logits/rejected": -7.001479148864746,
|
| 797 |
-
"logps/chosen": -0.6439567804336548,
|
| 798 |
-
"logps/rejected": -3.231163263320923,
|
| 799 |
-
"loss": 0.8379,
|
| 800 |
-
"rewards/accuracies": 0.8359375,
|
| 801 |
-
"rewards/chosen": -0.06439567357301712,
|
| 802 |
-
"rewards/margins": 0.2587206959724426,
|
| 803 |
-
"rewards/rejected": -0.32311639189720154,
|
| 804 |
-
"step": 53
|
| 805 |
-
},
|
| 806 |
-
{
|
| 807 |
-
"epoch": 0.01516794017090266,
|
| 808 |
-
"grad_norm": 9.125,
|
| 809 |
-
"learning_rate": 7.584269662921349e-06,
|
| 810 |
-
"logits/chosen": -7.009494304656982,
|
| 811 |
-
"logits/rejected": -7.009639739990234,
|
| 812 |
-
"logps/chosen": -0.6828871965408325,
|
| 813 |
-
"logps/rejected": -3.5422186851501465,
|
| 814 |
-
"loss": 0.8226,
|
| 815 |
-
"rewards/accuracies": 0.84375,
|
| 816 |
-
"rewards/chosen": -0.06828872114419937,
|
| 817 |
-
"rewards/margins": 0.28593313694000244,
|
| 818 |
-
"rewards/rejected": -0.3542218804359436,
|
| 819 |
-
"step": 54
|
| 820 |
-
},
|
| 821 |
-
{
|
| 822 |
-
"epoch": 0.0154488279518453,
|
| 823 |
-
"grad_norm": 1.9765625,
|
| 824 |
-
"learning_rate": 7.724719101123595e-06,
|
| 825 |
-
"logits/chosen": -7.009527206420898,
|
| 826 |
-
"logits/rejected": -7.009775161743164,
|
| 827 |
-
"logps/chosen": -0.5451503992080688,
|
| 828 |
-
"logps/rejected": -3.2060601711273193,
|
| 829 |
-
"loss": 0.831,
|
| 830 |
-
"rewards/accuracies": 0.8828125,
|
| 831 |
-
"rewards/chosen": -0.054515041410923004,
|
| 832 |
-
"rewards/margins": 0.2660909593105316,
|
| 833 |
-
"rewards/rejected": -0.3206060528755188,
|
| 834 |
-
"step": 55
|
| 835 |
-
},
|
| 836 |
-
{
|
| 837 |
-
"epoch": 0.015729715732787943,
|
| 838 |
-
"grad_norm": 1.203125,
|
| 839 |
-
"learning_rate": 7.865168539325843e-06,
|
| 840 |
-
"logits/chosen": -7.00971794128418,
|
| 841 |
-
"logits/rejected": -7.009875774383545,
|
| 842 |
-
"logps/chosen": -0.6500095129013062,
|
| 843 |
-
"logps/rejected": -3.940423011779785,
|
| 844 |
-
"loss": 0.8,
|
| 845 |
-
"rewards/accuracies": 0.8984375,
|
| 846 |
-
"rewards/chosen": -0.06500095129013062,
|
| 847 |
-
"rewards/margins": 0.32904139161109924,
|
| 848 |
-
"rewards/rejected": -0.39404234290122986,
|
| 849 |
-
"step": 56
|
| 850 |
-
},
|
| 851 |
-
{
|
| 852 |
-
"epoch": 0.016010603513730584,
|
| 853 |
-
"grad_norm": 2.703125,
|
| 854 |
-
"learning_rate": 8.00561797752809e-06,
|
| 855 |
-
"logits/chosen": -7.013581275939941,
|
| 856 |
-
"logits/rejected": -7.013791561126709,
|
| 857 |
-
"logps/chosen": -0.5986226797103882,
|
| 858 |
-
"logps/rejected": -3.3548402786254883,
|
| 859 |
-
"loss": 0.8256,
|
| 860 |
-
"rewards/accuracies": 0.828125,
|
| 861 |
-
"rewards/chosen": -0.05986226722598076,
|
| 862 |
-
"rewards/margins": 0.2756217420101166,
|
| 863 |
-
"rewards/rejected": -0.33548396825790405,
|
| 864 |
-
"step": 57
|
| 865 |
-
},
|
| 866 |
-
{
|
| 867 |
-
"epoch": 0.016291491294673225,
|
| 868 |
-
"grad_norm": 1.2109375,
|
| 869 |
-
"learning_rate": 8.146067415730338e-06,
|
| 870 |
-
"logits/chosen": -7.008817672729492,
|
| 871 |
-
"logits/rejected": -7.008937358856201,
|
| 872 |
-
"logps/chosen": -0.5546457171440125,
|
| 873 |
-
"logps/rejected": -3.4342713356018066,
|
| 874 |
-
"loss": 0.8211,
|
| 875 |
-
"rewards/accuracies": 0.8671875,
|
| 876 |
-
"rewards/chosen": -0.055464569479227066,
|
| 877 |
-
"rewards/margins": 0.28796258568763733,
|
| 878 |
-
"rewards/rejected": -0.3434271812438965,
|
| 879 |
-
"step": 58
|
| 880 |
-
},
|
| 881 |
-
{
|
| 882 |
-
"epoch": 0.01657237907561587,
|
| 883 |
-
"grad_norm": 1.7109375,
|
| 884 |
-
"learning_rate": 8.286516853932584e-06,
|
| 885 |
-
"logits/chosen": -7.012298583984375,
|
| 886 |
-
"logits/rejected": -7.012528896331787,
|
| 887 |
-
"logps/chosen": -0.5672339797019958,
|
| 888 |
-
"logps/rejected": -3.319044828414917,
|
| 889 |
-
"loss": 0.8259,
|
| 890 |
-
"rewards/accuracies": 0.90625,
|
| 891 |
-
"rewards/chosen": -0.05672340467572212,
|
| 892 |
-
"rewards/margins": 0.2751810550689697,
|
| 893 |
-
"rewards/rejected": -0.33190447092056274,
|
| 894 |
-
"step": 59
|
| 895 |
-
},
|
| 896 |
-
{
|
| 897 |
-
"epoch": 0.01685326685655851,
|
| 898 |
-
"grad_norm": 1.265625,
|
| 899 |
-
"learning_rate": 8.426966292134832e-06,
|
| 900 |
-
"logits/chosen": -7.013304710388184,
|
| 901 |
-
"logits/rejected": -7.013603210449219,
|
| 902 |
-
"logps/chosen": -0.5733633637428284,
|
| 903 |
-
"logps/rejected": -3.944350004196167,
|
| 904 |
-
"loss": 0.796,
|
| 905 |
-
"rewards/accuracies": 0.921875,
|
| 906 |
-
"rewards/chosen": -0.057336337864398956,
|
| 907 |
-
"rewards/margins": 0.33709871768951416,
|
| 908 |
-
"rewards/rejected": -0.3944350481033325,
|
| 909 |
"step": 60
|
| 910 |
},
|
| 911 |
{
|
| 912 |
-
"epoch": 0.
|
| 913 |
-
"grad_norm":
|
| 914 |
-
"learning_rate":
|
| 915 |
-
"logits/chosen": -
|
| 916 |
-
"logits/rejected": -
|
| 917 |
-
"logps/chosen": -0.
|
| 918 |
-
"logps/rejected": -
|
| 919 |
-
"loss": 0.
|
| 920 |
-
"
|
| 921 |
-
"rewards/
|
| 922 |
-
"rewards/
|
| 923 |
-
"rewards/
|
| 924 |
-
"
|
| 925 |
-
|
| 926 |
-
{
|
| 927 |
-
"epoch": 0.017415042418443796,
|
| 928 |
-
"grad_norm": 4.15625,
|
| 929 |
-
"learning_rate": 8.707865168539327e-06,
|
| 930 |
-
"logits/chosen": -7.013251781463623,
|
| 931 |
-
"logits/rejected": -7.013393878936768,
|
| 932 |
-
"logps/chosen": -0.6521888971328735,
|
| 933 |
-
"logps/rejected": -3.21026873588562,
|
| 934 |
-
"loss": 0.8371,
|
| 935 |
-
"rewards/accuracies": 0.8671875,
|
| 936 |
-
"rewards/chosen": -0.06521890312433243,
|
| 937 |
-
"rewards/margins": 0.2558079659938812,
|
| 938 |
-
"rewards/rejected": -0.32102692127227783,
|
| 939 |
-
"step": 62
|
| 940 |
-
},
|
| 941 |
-
{
|
| 942 |
-
"epoch": 0.017695930199386437,
|
| 943 |
-
"grad_norm": 1.1796875,
|
| 944 |
-
"learning_rate": 8.848314606741573e-06,
|
| 945 |
-
"logits/chosen": -7.01942253112793,
|
| 946 |
-
"logits/rejected": -7.019570350646973,
|
| 947 |
-
"logps/chosen": -0.5423337817192078,
|
| 948 |
-
"logps/rejected": -3.554921865463257,
|
| 949 |
-
"loss": 0.8147,
|
| 950 |
-
"rewards/accuracies": 0.875,
|
| 951 |
-
"rewards/chosen": -0.054233379662036896,
|
| 952 |
-
"rewards/margins": 0.3012588322162628,
|
| 953 |
-
"rewards/rejected": -0.3554922044277191,
|
| 954 |
-
"step": 63
|
| 955 |
-
},
|
| 956 |
-
{
|
| 957 |
-
"epoch": 0.017976817980329077,
|
| 958 |
-
"grad_norm": 0.84375,
|
| 959 |
-
"learning_rate": 8.98876404494382e-06,
|
| 960 |
-
"logits/chosen": -7.020681858062744,
|
| 961 |
-
"logits/rejected": -7.020779132843018,
|
| 962 |
-
"logps/chosen": -0.6097487807273865,
|
| 963 |
-
"logps/rejected": -3.4758450984954834,
|
| 964 |
-
"loss": 0.8241,
|
| 965 |
-
"rewards/accuracies": 0.8671875,
|
| 966 |
-
"rewards/chosen": -0.06097487360239029,
|
| 967 |
-
"rewards/margins": 0.28660961985588074,
|
| 968 |
-
"rewards/rejected": -0.3475845158100128,
|
| 969 |
-
"step": 64
|
| 970 |
-
},
|
| 971 |
-
{
|
| 972 |
-
"epoch": 0.01825770576127172,
|
| 973 |
-
"grad_norm": 1.0703125,
|
| 974 |
-
"learning_rate": 9.129213483146068e-06,
|
| 975 |
-
"logits/chosen": -7.021762847900391,
|
| 976 |
-
"logits/rejected": -7.022034645080566,
|
| 977 |
-
"logps/chosen": -0.6843651533126831,
|
| 978 |
-
"logps/rejected": -4.37304162979126,
|
| 979 |
-
"loss": 0.7824,
|
| 980 |
-
"rewards/accuracies": 0.921875,
|
| 981 |
-
"rewards/chosen": -0.06843651831150055,
|
| 982 |
-
"rewards/margins": 0.3688676655292511,
|
| 983 |
-
"rewards/rejected": -0.43730416893959045,
|
| 984 |
-
"step": 65
|
| 985 |
-
},
|
| 986 |
-
{
|
| 987 |
-
"epoch": 0.018538593542214363,
|
| 988 |
-
"grad_norm": 1.359375,
|
| 989 |
-
"learning_rate": 9.269662921348316e-06,
|
| 990 |
-
"logits/chosen": -7.02295446395874,
|
| 991 |
-
"logits/rejected": -7.023041248321533,
|
| 992 |
-
"logps/chosen": -0.6040046215057373,
|
| 993 |
-
"logps/rejected": -3.583864688873291,
|
| 994 |
-
"loss": 0.8154,
|
| 995 |
-
"rewards/accuracies": 0.875,
|
| 996 |
-
"rewards/chosen": -0.06040046364068985,
|
| 997 |
-
"rewards/margins": 0.2979860007762909,
|
| 998 |
-
"rewards/rejected": -0.3583865165710449,
|
| 999 |
-
"step": 66
|
| 1000 |
-
},
|
| 1001 |
-
{
|
| 1002 |
-
"epoch": 0.018819481323157004,
|
| 1003 |
-
"grad_norm": 1.15625,
|
| 1004 |
-
"learning_rate": 9.410112359550562e-06,
|
| 1005 |
-
"logits/chosen": -7.029426097869873,
|
| 1006 |
-
"logits/rejected": -7.0295729637146,
|
| 1007 |
-
"logps/chosen": -0.5566189289093018,
|
| 1008 |
-
"logps/rejected": -3.2764642238616943,
|
| 1009 |
-
"loss": 0.8283,
|
| 1010 |
-
"rewards/accuracies": 0.859375,
|
| 1011 |
-
"rewards/chosen": -0.055661890655756,
|
| 1012 |
-
"rewards/margins": 0.2719845473766327,
|
| 1013 |
-
"rewards/rejected": -0.3276464343070984,
|
| 1014 |
-
"step": 67
|
| 1015 |
-
},
|
| 1016 |
-
{
|
| 1017 |
-
"epoch": 0.019100369104099645,
|
| 1018 |
-
"grad_norm": 1.234375,
|
| 1019 |
-
"learning_rate": 9.550561797752809e-06,
|
| 1020 |
-
"logits/chosen": -7.031888961791992,
|
| 1021 |
-
"logits/rejected": -7.032125473022461,
|
| 1022 |
-
"logps/chosen": -0.5691148638725281,
|
| 1023 |
-
"logps/rejected": -3.4460644721984863,
|
| 1024 |
-
"loss": 0.821,
|
| 1025 |
-
"rewards/accuracies": 0.890625,
|
| 1026 |
-
"rewards/chosen": -0.05691148713231087,
|
| 1027 |
-
"rewards/margins": 0.2876949906349182,
|
| 1028 |
-
"rewards/rejected": -0.3446064591407776,
|
| 1029 |
-
"step": 68
|
| 1030 |
-
},
|
| 1031 |
-
{
|
| 1032 |
-
"epoch": 0.019381256885042285,
|
| 1033 |
-
"grad_norm": 1.59375,
|
| 1034 |
-
"learning_rate": 9.691011235955057e-06,
|
| 1035 |
-
"logits/chosen": -7.036633014678955,
|
| 1036 |
-
"logits/rejected": -7.036778926849365,
|
| 1037 |
-
"logps/chosen": -0.6520639061927795,
|
| 1038 |
-
"logps/rejected": -3.434640645980835,
|
| 1039 |
-
"loss": 0.8252,
|
| 1040 |
-
"rewards/accuracies": 0.8671875,
|
| 1041 |
-
"rewards/chosen": -0.06520639359951019,
|
| 1042 |
-
"rewards/margins": 0.27825766801834106,
|
| 1043 |
-
"rewards/rejected": -0.34346407651901245,
|
| 1044 |
-
"step": 69
|
| 1045 |
-
},
|
| 1046 |
-
{
|
| 1047 |
-
"epoch": 0.01966214466598493,
|
| 1048 |
-
"grad_norm": 1.046875,
|
| 1049 |
-
"learning_rate": 9.831460674157303e-06,
|
| 1050 |
-
"logits/chosen": -7.03872013092041,
|
| 1051 |
-
"logits/rejected": -7.038879871368408,
|
| 1052 |
-
"logps/chosen": -0.5827143788337708,
|
| 1053 |
-
"logps/rejected": -3.6182808876037598,
|
| 1054 |
-
"loss": 0.8103,
|
| 1055 |
-
"rewards/accuracies": 0.9296875,
|
| 1056 |
-
"rewards/chosen": -0.058271437883377075,
|
| 1057 |
-
"rewards/margins": 0.3035566508769989,
|
| 1058 |
-
"rewards/rejected": -0.36182811856269836,
|
| 1059 |
"step": 70
|
| 1060 |
},
|
| 1061 |
{
|
| 1062 |
-
"epoch": 0.
|
| 1063 |
-
"grad_norm":
|
| 1064 |
-
"learning_rate":
|
| 1065 |
-
"logits/chosen": -
|
| 1066 |
-
"logits/rejected": -
|
| 1067 |
-
"logps/chosen": -0.
|
| 1068 |
-
"logps/rejected": -
|
| 1069 |
-
"loss": 0.
|
| 1070 |
-
"
|
| 1071 |
-
"rewards/
|
| 1072 |
-
"rewards/
|
| 1073 |
-
"rewards/
|
| 1074 |
-
"
|
| 1075 |
-
|
| 1076 |
-
{
|
| 1077 |
-
"epoch": 0.02022392022787021,
|
| 1078 |
-
"grad_norm": 3.609375,
|
| 1079 |
-
"learning_rate": 1.0112359550561798e-05,
|
| 1080 |
-
"logits/chosen": -7.046081066131592,
|
| 1081 |
-
"logits/rejected": -7.046180725097656,
|
| 1082 |
-
"logps/chosen": -0.683323085308075,
|
| 1083 |
-
"logps/rejected": -3.082217216491699,
|
| 1084 |
-
"loss": 0.843,
|
| 1085 |
-
"rewards/accuracies": 0.875,
|
| 1086 |
-
"rewards/chosen": -0.06833230704069138,
|
| 1087 |
-
"rewards/margins": 0.23988942801952362,
|
| 1088 |
-
"rewards/rejected": -0.3082217574119568,
|
| 1089 |
-
"step": 72
|
| 1090 |
-
},
|
| 1091 |
-
{
|
| 1092 |
-
"epoch": 0.020504808008812853,
|
| 1093 |
-
"grad_norm": 1.1484375,
|
| 1094 |
-
"learning_rate": 1.0252808988764046e-05,
|
| 1095 |
-
"logits/chosen": -7.047407150268555,
|
| 1096 |
-
"logits/rejected": -7.047583103179932,
|
| 1097 |
-
"logps/chosen": -0.6924074292182922,
|
| 1098 |
-
"logps/rejected": -3.506157159805298,
|
| 1099 |
-
"loss": 0.8246,
|
| 1100 |
-
"rewards/accuracies": 0.8046875,
|
| 1101 |
-
"rewards/chosen": -0.0692407488822937,
|
| 1102 |
-
"rewards/margins": 0.281374990940094,
|
| 1103 |
-
"rewards/rejected": -0.3506157100200653,
|
| 1104 |
-
"step": 73
|
| 1105 |
-
},
|
| 1106 |
-
{
|
| 1107 |
-
"epoch": 0.020785695789755497,
|
| 1108 |
-
"grad_norm": 74.5,
|
| 1109 |
-
"learning_rate": 1.0393258426966292e-05,
|
| 1110 |
-
"logits/chosen": -7.055050373077393,
|
| 1111 |
-
"logits/rejected": -7.055255889892578,
|
| 1112 |
-
"logps/chosen": -0.4969649910926819,
|
| 1113 |
-
"logps/rejected": -3.5049824714660645,
|
| 1114 |
-
"loss": 0.8142,
|
| 1115 |
-
"rewards/accuracies": 0.8984375,
|
| 1116 |
-
"rewards/chosen": -0.04969649761915207,
|
| 1117 |
-
"rewards/margins": 0.30080172419548035,
|
| 1118 |
-
"rewards/rejected": -0.350498229265213,
|
| 1119 |
-
"step": 74
|
| 1120 |
-
},
|
| 1121 |
-
{
|
| 1122 |
-
"epoch": 0.021066583570698138,
|
| 1123 |
-
"grad_norm": 1.3515625,
|
| 1124 |
-
"learning_rate": 1.0533707865168539e-05,
|
| 1125 |
-
"logits/chosen": -7.057081699371338,
|
| 1126 |
-
"logits/rejected": -7.057204723358154,
|
| 1127 |
-
"logps/chosen": -0.6645674705505371,
|
| 1128 |
-
"logps/rejected": -3.872537136077881,
|
| 1129 |
-
"loss": 0.8049,
|
| 1130 |
-
"rewards/accuracies": 0.890625,
|
| 1131 |
-
"rewards/chosen": -0.06645674258470535,
|
| 1132 |
-
"rewards/margins": 0.32079699635505676,
|
| 1133 |
-
"rewards/rejected": -0.38725370168685913,
|
| 1134 |
-
"step": 75
|
| 1135 |
-
},
|
| 1136 |
-
{
|
| 1137 |
-
"epoch": 0.02134747135164078,
|
| 1138 |
-
"grad_norm": 0.984375,
|
| 1139 |
-
"learning_rate": 1.0674157303370787e-05,
|
| 1140 |
-
"logits/chosen": -7.062127113342285,
|
| 1141 |
-
"logits/rejected": -7.0622992515563965,
|
| 1142 |
-
"logps/chosen": -0.683305561542511,
|
| 1143 |
-
"logps/rejected": -3.850494861602783,
|
| 1144 |
-
"loss": 0.8067,
|
| 1145 |
-
"rewards/accuracies": 0.875,
|
| 1146 |
-
"rewards/chosen": -0.0683305561542511,
|
| 1147 |
-
"rewards/margins": 0.3167189657688141,
|
| 1148 |
-
"rewards/rejected": -0.3850494921207428,
|
| 1149 |
-
"step": 76
|
| 1150 |
-
},
|
| 1151 |
-
{
|
| 1152 |
-
"epoch": 0.021628359132583423,
|
| 1153 |
-
"grad_norm": 0.7890625,
|
| 1154 |
-
"learning_rate": 1.0814606741573033e-05,
|
| 1155 |
-
"logits/chosen": -7.0633769035339355,
|
| 1156 |
-
"logits/rejected": -7.0634965896606445,
|
| 1157 |
-
"logps/chosen": -0.6324968338012695,
|
| 1158 |
-
"logps/rejected": -3.4045989513397217,
|
| 1159 |
-
"loss": 0.8247,
|
| 1160 |
-
"rewards/accuracies": 0.890625,
|
| 1161 |
-
"rewards/chosen": -0.06324967741966248,
|
| 1162 |
-
"rewards/margins": 0.27721020579338074,
|
| 1163 |
-
"rewards/rejected": -0.3404598832130432,
|
| 1164 |
-
"step": 77
|
| 1165 |
-
},
|
| 1166 |
-
{
|
| 1167 |
-
"epoch": 0.021909246913526064,
|
| 1168 |
-
"grad_norm": 1.1875,
|
| 1169 |
-
"learning_rate": 1.0955056179775282e-05,
|
| 1170 |
-
"logits/chosen": -7.069698333740234,
|
| 1171 |
-
"logits/rejected": -7.0697784423828125,
|
| 1172 |
-
"logps/chosen": -0.6205027103424072,
|
| 1173 |
-
"logps/rejected": -3.5221495628356934,
|
| 1174 |
-
"loss": 0.8222,
|
| 1175 |
-
"rewards/accuracies": 0.8515625,
|
| 1176 |
-
"rewards/chosen": -0.06205027177929878,
|
| 1177 |
-
"rewards/margins": 0.2901647090911865,
|
| 1178 |
-
"rewards/rejected": -0.3522149622440338,
|
| 1179 |
-
"step": 78
|
| 1180 |
-
},
|
| 1181 |
-
{
|
| 1182 |
-
"epoch": 0.022190134694468705,
|
| 1183 |
-
"grad_norm": 1.2109375,
|
| 1184 |
-
"learning_rate": 1.1095505617977528e-05,
|
| 1185 |
-
"logits/chosen": -7.070184707641602,
|
| 1186 |
-
"logits/rejected": -7.070287227630615,
|
| 1187 |
-
"logps/chosen": -0.5837754607200623,
|
| 1188 |
-
"logps/rejected": -3.566255807876587,
|
| 1189 |
-
"loss": 0.8126,
|
| 1190 |
-
"rewards/accuracies": 0.875,
|
| 1191 |
-
"rewards/chosen": -0.058377549052238464,
|
| 1192 |
-
"rewards/margins": 0.2982480823993683,
|
| 1193 |
-
"rewards/rejected": -0.35662561655044556,
|
| 1194 |
-
"step": 79
|
| 1195 |
-
},
|
| 1196 |
-
{
|
| 1197 |
-
"epoch": 0.022471022475411346,
|
| 1198 |
-
"grad_norm": 1.5859375,
|
| 1199 |
-
"learning_rate": 1.1235955056179776e-05,
|
| 1200 |
-
"logits/chosen": -7.073416709899902,
|
| 1201 |
-
"logits/rejected": -7.073558807373047,
|
| 1202 |
-
"logps/chosen": -0.6057207584381104,
|
| 1203 |
-
"logps/rejected": -3.7733519077301025,
|
| 1204 |
-
"loss": 0.8087,
|
| 1205 |
-
"rewards/accuracies": 0.8515625,
|
| 1206 |
-
"rewards/chosen": -0.060572076588869095,
|
| 1207 |
-
"rewards/margins": 0.31676313281059265,
|
| 1208 |
-
"rewards/rejected": -0.37733519077301025,
|
| 1209 |
"step": 80
|
| 1210 |
},
|
| 1211 |
{
|
| 1212 |
-
"epoch": 0.
|
| 1213 |
-
"grad_norm":
|
| 1214 |
-
"learning_rate":
|
| 1215 |
-
"logits/chosen": -
|
| 1216 |
-
"logits/rejected": -
|
| 1217 |
-
"logps/chosen": -0.
|
| 1218 |
-
"logps/rejected": -
|
| 1219 |
-
"loss": 0.
|
| 1220 |
-
"
|
| 1221 |
-
"rewards/
|
| 1222 |
-
"rewards/
|
| 1223 |
-
"rewards/
|
| 1224 |
-
"
|
| 1225 |
-
|
| 1226 |
-
{
|
| 1227 |
-
"epoch": 0.02303279803729663,
|
| 1228 |
-
"grad_norm": 1.5625,
|
| 1229 |
-
"learning_rate": 1.151685393258427e-05,
|
| 1230 |
-
"logits/chosen": -7.0821213722229,
|
| 1231 |
-
"logits/rejected": -7.082340240478516,
|
| 1232 |
-
"logps/chosen": -0.628500759601593,
|
| 1233 |
-
"logps/rejected": -3.7704579830169678,
|
| 1234 |
-
"loss": 0.8099,
|
| 1235 |
-
"rewards/accuracies": 0.890625,
|
| 1236 |
-
"rewards/chosen": -0.06285007297992706,
|
| 1237 |
-
"rewards/margins": 0.31419575214385986,
|
| 1238 |
-
"rewards/rejected": -0.37704581022262573,
|
| 1239 |
-
"step": 82
|
| 1240 |
-
},
|
| 1241 |
-
{
|
| 1242 |
-
"epoch": 0.023313685818239272,
|
| 1243 |
-
"grad_norm": 0.9765625,
|
| 1244 |
-
"learning_rate": 1.1657303370786517e-05,
|
| 1245 |
-
"logits/chosen": -7.087411880493164,
|
| 1246 |
-
"logits/rejected": -7.0875701904296875,
|
| 1247 |
-
"logps/chosen": -0.7774583101272583,
|
| 1248 |
-
"logps/rejected": -3.098719358444214,
|
| 1249 |
-
"loss": 0.8513,
|
| 1250 |
-
"rewards/accuracies": 0.7890625,
|
| 1251 |
-
"rewards/chosen": -0.07774583995342255,
|
| 1252 |
-
"rewards/margins": 0.2321261167526245,
|
| 1253 |
-
"rewards/rejected": -0.3098719120025635,
|
| 1254 |
-
"step": 83
|
| 1255 |
-
},
|
| 1256 |
-
{
|
| 1257 |
-
"epoch": 0.023594573599181913,
|
| 1258 |
-
"grad_norm": 0.96484375,
|
| 1259 |
-
"learning_rate": 1.1797752808988765e-05,
|
| 1260 |
-
"logits/chosen": -7.087563991546631,
|
| 1261 |
-
"logits/rejected": -7.087766647338867,
|
| 1262 |
-
"logps/chosen": -0.5942208766937256,
|
| 1263 |
-
"logps/rejected": -3.334862232208252,
|
| 1264 |
-
"loss": 0.8292,
|
| 1265 |
-
"rewards/accuracies": 0.875,
|
| 1266 |
-
"rewards/chosen": -0.0594220906496048,
|
| 1267 |
-
"rewards/margins": 0.2740641236305237,
|
| 1268 |
-
"rewards/rejected": -0.3334861993789673,
|
| 1269 |
-
"step": 84
|
| 1270 |
-
},
|
| 1271 |
-
{
|
| 1272 |
-
"epoch": 0.023875461380124557,
|
| 1273 |
-
"grad_norm": 1.21875,
|
| 1274 |
-
"learning_rate": 1.1938202247191012e-05,
|
| 1275 |
-
"logits/chosen": -7.094630241394043,
|
| 1276 |
-
"logits/rejected": -7.094793319702148,
|
| 1277 |
-
"logps/chosen": -0.6228974461555481,
|
| 1278 |
-
"logps/rejected": -4.061489105224609,
|
| 1279 |
-
"loss": 0.7951,
|
| 1280 |
-
"rewards/accuracies": 0.875,
|
| 1281 |
-
"rewards/chosen": -0.06228974461555481,
|
| 1282 |
-
"rewards/margins": 0.34385910630226135,
|
| 1283 |
-
"rewards/rejected": -0.40614888072013855,
|
| 1284 |
-
"step": 85
|
| 1285 |
-
},
|
| 1286 |
-
{
|
| 1287 |
-
"epoch": 0.0241563491610672,
|
| 1288 |
-
"grad_norm": 1.6484375,
|
| 1289 |
-
"learning_rate": 1.207865168539326e-05,
|
| 1290 |
-
"logits/chosen": -7.10193395614624,
|
| 1291 |
-
"logits/rejected": -7.1020917892456055,
|
| 1292 |
-
"logps/chosen": -0.6275684237480164,
|
| 1293 |
-
"logps/rejected": -3.7310783863067627,
|
| 1294 |
-
"loss": 0.8106,
|
| 1295 |
-
"rewards/accuracies": 0.8984375,
|
| 1296 |
-
"rewards/chosen": -0.06275683641433716,
|
| 1297 |
-
"rewards/margins": 0.31035101413726807,
|
| 1298 |
-
"rewards/rejected": -0.3731078505516052,
|
| 1299 |
-
"step": 86
|
| 1300 |
-
},
|
| 1301 |
-
{
|
| 1302 |
-
"epoch": 0.02443723694200984,
|
| 1303 |
-
"grad_norm": 1.1171875,
|
| 1304 |
-
"learning_rate": 1.2219101123595506e-05,
|
| 1305 |
-
"logits/chosen": -7.102446556091309,
|
| 1306 |
-
"logits/rejected": -7.102678298950195,
|
| 1307 |
-
"logps/chosen": -0.6499467492103577,
|
| 1308 |
-
"logps/rejected": -3.5946946144104004,
|
| 1309 |
-
"loss": 0.8177,
|
| 1310 |
-
"rewards/accuracies": 0.8828125,
|
| 1311 |
-
"rewards/chosen": -0.064994677901268,
|
| 1312 |
-
"rewards/margins": 0.2944748103618622,
|
| 1313 |
-
"rewards/rejected": -0.3594695031642914,
|
| 1314 |
-
"step": 87
|
| 1315 |
-
},
|
| 1316 |
-
{
|
| 1317 |
-
"epoch": 0.02471812472295248,
|
| 1318 |
-
"grad_norm": 1.4921875,
|
| 1319 |
-
"learning_rate": 1.2359550561797752e-05,
|
| 1320 |
-
"logits/chosen": -7.104351997375488,
|
| 1321 |
-
"logits/rejected": -7.104528427124023,
|
| 1322 |
-
"logps/chosen": -0.6288101077079773,
|
| 1323 |
-
"logps/rejected": -3.6652326583862305,
|
| 1324 |
-
"loss": 0.814,
|
| 1325 |
-
"rewards/accuracies": 0.8671875,
|
| 1326 |
-
"rewards/chosen": -0.0628810003399849,
|
| 1327 |
-
"rewards/margins": 0.303642213344574,
|
| 1328 |
-
"rewards/rejected": -0.36652323603630066,
|
| 1329 |
-
"step": 88
|
| 1330 |
-
},
|
| 1331 |
-
{
|
| 1332 |
-
"epoch": 0.024999012503895125,
|
| 1333 |
-
"grad_norm": 2.78125,
|
| 1334 |
-
"learning_rate": 1.25e-05,
|
| 1335 |
-
"logits/chosen": -7.110116004943848,
|
| 1336 |
-
"logits/rejected": -7.110256671905518,
|
| 1337 |
-
"logps/chosen": -0.6746590733528137,
|
| 1338 |
-
"logps/rejected": -3.6932754516601562,
|
| 1339 |
-
"loss": 0.8167,
|
| 1340 |
-
"rewards/accuracies": 0.8515625,
|
| 1341 |
-
"rewards/chosen": -0.06746591627597809,
|
| 1342 |
-
"rewards/margins": 0.30186164379119873,
|
| 1343 |
-
"rewards/rejected": -0.3693275451660156,
|
| 1344 |
-
"step": 89
|
| 1345 |
-
},
|
| 1346 |
-
{
|
| 1347 |
-
"epoch": 0.025279900284837765,
|
| 1348 |
-
"grad_norm": 1.4140625,
|
| 1349 |
-
"learning_rate": 1.2640449438202249e-05,
|
| 1350 |
-
"logits/chosen": -7.113439559936523,
|
| 1351 |
-
"logits/rejected": -7.113628387451172,
|
| 1352 |
-
"logps/chosen": -0.6546343564987183,
|
| 1353 |
-
"logps/rejected": -3.9948794841766357,
|
| 1354 |
-
"loss": 0.7989,
|
| 1355 |
-
"rewards/accuracies": 0.875,
|
| 1356 |
-
"rewards/chosen": -0.06546343863010406,
|
| 1357 |
-
"rewards/margins": 0.3340245485305786,
|
| 1358 |
-
"rewards/rejected": -0.3994879722595215,
|
| 1359 |
"step": 90
|
| 1360 |
},
|
| 1361 |
{
|
| 1362 |
-
"epoch": 0.
|
| 1363 |
-
"grad_norm":
|
| 1364 |
-
"learning_rate":
|
| 1365 |
-
"logits/chosen": -
|
| 1366 |
-
"logits/rejected": -
|
| 1367 |
-
"logps/chosen": -0.
|
| 1368 |
-
"logps/rejected": -
|
| 1369 |
-
"loss": 0.
|
| 1370 |
-
"
|
| 1371 |
-
"rewards/
|
| 1372 |
-
"rewards/
|
| 1373 |
-
"rewards/
|
| 1374 |
-
"
|
| 1375 |
-
|
| 1376 |
-
|
| 1377 |
-
"epoch": 0.02584167584672305,
|
| 1378 |
-
"grad_norm": 1.84375,
|
| 1379 |
-
"learning_rate": 1.2921348314606743e-05,
|
| 1380 |
-
"logits/chosen": -7.124433994293213,
|
| 1381 |
-
"logits/rejected": -7.124541282653809,
|
| 1382 |
-
"logps/chosen": -0.5675304532051086,
|
| 1383 |
-
"logps/rejected": -3.8615882396698,
|
| 1384 |
-
"loss": 0.8041,
|
| 1385 |
-
"rewards/accuracies": 0.8828125,
|
| 1386 |
-
"rewards/chosen": -0.056753043085336685,
|
| 1387 |
-
"rewards/margins": 0.3294057846069336,
|
| 1388 |
-
"rewards/rejected": -0.38615882396698,
|
| 1389 |
-
"step": 92
|
| 1390 |
-
},
|
| 1391 |
-
{
|
| 1392 |
-
"epoch": 0.02612256362766569,
|
| 1393 |
-
"grad_norm": 1.1875,
|
| 1394 |
-
"learning_rate": 1.3061797752808991e-05,
|
| 1395 |
-
"logits/chosen": -7.122183799743652,
|
| 1396 |
-
"logits/rejected": -7.122403621673584,
|
| 1397 |
-
"logps/chosen": -0.6670414805412292,
|
| 1398 |
-
"logps/rejected": -4.148705959320068,
|
| 1399 |
-
"loss": 0.7911,
|
| 1400 |
-
"rewards/accuracies": 0.8984375,
|
| 1401 |
-
"rewards/chosen": -0.0667041540145874,
|
| 1402 |
-
"rewards/margins": 0.34816643595695496,
|
| 1403 |
-
"rewards/rejected": -0.41487061977386475,
|
| 1404 |
-
"step": 93
|
| 1405 |
-
},
|
| 1406 |
-
{
|
| 1407 |
-
"epoch": 0.026403451408608333,
|
| 1408 |
-
"grad_norm": 4.0625,
|
| 1409 |
-
"learning_rate": 1.3202247191011236e-05,
|
| 1410 |
-
"logits/chosen": -7.128363609313965,
|
| 1411 |
-
"logits/rejected": -7.128509998321533,
|
| 1412 |
-
"logps/chosen": -0.6008510589599609,
|
| 1413 |
-
"logps/rejected": -4.413697719573975,
|
| 1414 |
-
"loss": 0.7794,
|
| 1415 |
-
"rewards/accuracies": 0.875,
|
| 1416 |
-
"rewards/chosen": -0.06008511036634445,
|
| 1417 |
-
"rewards/margins": 0.3812846839427948,
|
| 1418 |
-
"rewards/rejected": -0.44136980175971985,
|
| 1419 |
-
"step": 94
|
| 1420 |
},
|
| 1421 |
{
|
| 1422 |
-
"epoch": 0.
|
| 1423 |
-
"grad_norm":
|
| 1424 |
-
"learning_rate":
|
| 1425 |
-
"logits/chosen": -
|
| 1426 |
-
"logits/rejected": -
|
| 1427 |
-
"logps/chosen": -0.
|
| 1428 |
-
"logps/rejected": -
|
| 1429 |
-
"loss": 0.
|
| 1430 |
-
"
|
| 1431 |
-
"rewards/
|
| 1432 |
-
"rewards/
|
| 1433 |
-
"rewards/
|
| 1434 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1435 |
},
|
| 1436 |
{
|
| 1437 |
-
"epoch": 0.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1438 |
"grad_norm": 1.921875,
|
| 1439 |
-
"learning_rate":
|
| 1440 |
-
"logits/chosen": -
|
| 1441 |
-
"logits/rejected": -
|
| 1442 |
-
"logps/chosen": -0.
|
| 1443 |
-
"logps/rejected": -
|
| 1444 |
-
"loss": 0.
|
| 1445 |
-
"
|
| 1446 |
-
"rewards/
|
| 1447 |
-
"rewards/
|
| 1448 |
-
"rewards/
|
| 1449 |
-
"
|
| 1450 |
-
|
| 1451 |
-
|
| 1452 |
-
|
| 1453 |
-
|
| 1454 |
-
"
|
| 1455 |
-
"
|
| 1456 |
-
"
|
| 1457 |
-
"
|
| 1458 |
-
"
|
| 1459 |
-
"
|
| 1460 |
-
"
|
| 1461 |
-
"
|
| 1462 |
-
"
|
| 1463 |
-
"rewards/
|
| 1464 |
-
"
|
| 1465 |
-
|
| 1466 |
-
|
| 1467 |
-
"
|
| 1468 |
-
"
|
| 1469 |
-
|
| 1470 |
-
|
| 1471 |
-
"
|
| 1472 |
-
"
|
| 1473 |
-
"
|
| 1474 |
-
"
|
| 1475 |
-
"
|
| 1476 |
-
"
|
| 1477 |
-
"
|
| 1478 |
-
"
|
| 1479 |
-
"
|
| 1480 |
-
|
| 1481 |
-
|
| 1482 |
-
"
|
| 1483 |
-
"
|
| 1484 |
-
"
|
| 1485 |
-
"
|
| 1486 |
-
|
| 1487 |
-
|
| 1488 |
-
"
|
| 1489 |
-
"
|
| 1490 |
-
"
|
| 1491 |
-
"
|
| 1492 |
-
"
|
| 1493 |
-
"
|
| 1494 |
-
"
|
| 1495 |
-
|
| 1496 |
-
|
| 1497 |
-
"
|
| 1498 |
-
"
|
| 1499 |
-
"
|
| 1500 |
-
"
|
| 1501 |
-
"
|
| 1502 |
-
"
|
| 1503 |
-
|
| 1504 |
-
|
| 1505 |
-
"
|
| 1506 |
-
"
|
| 1507 |
-
"
|
| 1508 |
-
"
|
| 1509 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1510 |
}
|
| 1511 |
],
|
| 1512 |
-
"logging_steps":
|
| 1513 |
-
"max_steps":
|
| 1514 |
"num_input_tokens_seen": 0,
|
| 1515 |
"num_train_epochs": 1,
|
| 1516 |
"save_steps": 100,
|
|
@@ -1526,7 +706,7 @@
|
|
| 1526 |
"attributes": {}
|
| 1527 |
}
|
| 1528 |
},
|
| 1529 |
-
"total_flos":
|
| 1530 |
"train_batch_size": 2,
|
| 1531 |
"trial_name": null,
|
| 1532 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.5712596162965194,
|
| 5 |
"eval_steps": 5000,
|
| 6 |
+
"global_step": 400,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.014281490407412986,
|
| 13 |
+
"grad_norm": 49.5,
|
| 14 |
+
"learning_rate": 4.285714285714285e-05,
|
| 15 |
+
"logits/chosen": -2.928096294403076,
|
| 16 |
+
"logits/rejected": -2.9280941486358643,
|
| 17 |
+
"logps/chosen": -3.81945538520813,
|
| 18 |
+
"logps/rejected": -4.2658491134643555,
|
| 19 |
+
"loss": 3.8875,
|
| 20 |
+
"odds_ratio_loss": 7.880957126617432,
|
| 21 |
+
"rewards/accuracies": 0.641796886920929,
|
| 22 |
+
"rewards/chosen": -0.38194555044174194,
|
| 23 |
+
"rewards/margins": 0.044639457017183304,
|
| 24 |
+
"rewards/rejected": -0.42658501863479614,
|
| 25 |
+
"sft_loss": 3.099365711212158,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
"step": 10
|
| 27 |
},
|
| 28 |
{
|
| 29 |
+
"epoch": 0.028562980814825972,
|
| 30 |
+
"grad_norm": 9.4375,
|
| 31 |
+
"learning_rate": 8.57142857142857e-05,
|
| 32 |
+
"logits/chosen": -3.5747146606445312,
|
| 33 |
+
"logits/rejected": -3.5747463703155518,
|
| 34 |
+
"logps/chosen": -1.0031383037567139,
|
| 35 |
+
"logps/rejected": -1.795668601989746,
|
| 36 |
+
"loss": 1.0575,
|
| 37 |
+
"odds_ratio_loss": 3.653076171875,
|
| 38 |
+
"rewards/accuracies": 0.7490234375,
|
| 39 |
+
"rewards/chosen": -0.10031384229660034,
|
| 40 |
+
"rewards/margins": 0.07925303280353546,
|
| 41 |
+
"rewards/rejected": -0.1795668601989746,
|
| 42 |
+
"sft_loss": 0.6921551823616028,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 43 |
"step": 20
|
| 44 |
},
|
| 45 |
{
|
| 46 |
+
"epoch": 0.04284447122223896,
|
| 47 |
+
"grad_norm": 6.375,
|
| 48 |
+
"learning_rate": 0.00012857142857142855,
|
| 49 |
+
"logits/chosen": -4.056135177612305,
|
| 50 |
+
"logits/rejected": -4.05615234375,
|
| 51 |
+
"logps/chosen": -0.7499507665634155,
|
| 52 |
+
"logps/rejected": -1.9470783472061157,
|
| 53 |
+
"loss": 0.7945,
|
| 54 |
+
"odds_ratio_loss": 2.919811248779297,
|
| 55 |
+
"rewards/accuracies": 0.791210949420929,
|
| 56 |
+
"rewards/chosen": -0.07499508559703827,
|
| 57 |
+
"rewards/margins": 0.11971275508403778,
|
| 58 |
+
"rewards/rejected": -0.19470782577991486,
|
| 59 |
+
"sft_loss": 0.5025397539138794,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 60 |
"step": 30
|
| 61 |
},
|
| 62 |
{
|
| 63 |
+
"epoch": 0.057125961629651945,
|
| 64 |
+
"grad_norm": 6.4375,
|
| 65 |
+
"learning_rate": 0.0001714285714285714,
|
| 66 |
+
"logits/chosen": -3.9972000122070312,
|
| 67 |
+
"logits/rejected": -3.997206211090088,
|
| 68 |
+
"logps/chosen": -0.6447885632514954,
|
| 69 |
+
"logps/rejected": -2.2042269706726074,
|
| 70 |
+
"loss": 0.6819,
|
| 71 |
+
"odds_ratio_loss": 2.3861823081970215,
|
| 72 |
+
"rewards/accuracies": 0.8414062261581421,
|
| 73 |
+
"rewards/chosen": -0.06447885185480118,
|
| 74 |
+
"rewards/margins": 0.1559438556432724,
|
| 75 |
+
"rewards/rejected": -0.22042270004749298,
|
| 76 |
+
"sft_loss": 0.4432622492313385,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 77 |
"step": 40
|
| 78 |
},
|
| 79 |
{
|
| 80 |
+
"epoch": 0.07140745203706493,
|
| 81 |
+
"grad_norm": 2.53125,
|
| 82 |
+
"learning_rate": 0.00021428571428571427,
|
| 83 |
+
"logits/chosen": -3.6315770149230957,
|
| 84 |
+
"logits/rejected": -3.631592273712158,
|
| 85 |
+
"logps/chosen": -0.5242542028427124,
|
| 86 |
+
"logps/rejected": -2.490429401397705,
|
| 87 |
+
"loss": 0.5524,
|
| 88 |
+
"odds_ratio_loss": 1.806905746459961,
|
| 89 |
+
"rewards/accuracies": 0.887890636920929,
|
| 90 |
+
"rewards/chosen": -0.05242542549967766,
|
| 91 |
+
"rewards/margins": 0.1966175138950348,
|
| 92 |
+
"rewards/rejected": -0.24904294312000275,
|
| 93 |
+
"sft_loss": 0.37168318033218384,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 94 |
"step": 50
|
| 95 |
},
|
| 96 |
{
|
| 97 |
+
"epoch": 0.08568894244447792,
|
| 98 |
+
"grad_norm": 242.0,
|
| 99 |
+
"learning_rate": 0.0002571428571428571,
|
| 100 |
+
"logits/chosen": -3.232466459274292,
|
| 101 |
+
"logits/rejected": -3.2325031757354736,
|
| 102 |
+
"logps/chosen": -0.8311947584152222,
|
| 103 |
+
"logps/rejected": -2.9961485862731934,
|
| 104 |
+
"loss": 0.8575,
|
| 105 |
+
"odds_ratio_loss": 3.2764008045196533,
|
| 106 |
+
"rewards/accuracies": 0.8935546875,
|
| 107 |
+
"rewards/chosen": -0.0831194818019867,
|
| 108 |
+
"rewards/margins": 0.21649539470672607,
|
| 109 |
+
"rewards/rejected": -0.2996148467063904,
|
| 110 |
+
"sft_loss": 0.5298588275909424,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 111 |
"step": 60
|
| 112 |
},
|
| 113 |
{
|
| 114 |
+
"epoch": 0.0999704328518909,
|
| 115 |
+
"grad_norm": 5.65625,
|
| 116 |
+
"learning_rate": 0.0003,
|
| 117 |
+
"logits/chosen": -2.657118558883667,
|
| 118 |
+
"logits/rejected": -2.6572413444519043,
|
| 119 |
+
"logps/chosen": -0.6468337774276733,
|
| 120 |
+
"logps/rejected": -2.5454134941101074,
|
| 121 |
+
"loss": 0.6815,
|
| 122 |
+
"odds_ratio_loss": 2.252271890640259,
|
| 123 |
+
"rewards/accuracies": 0.869921863079071,
|
| 124 |
+
"rewards/chosen": -0.06468339264392853,
|
| 125 |
+
"rewards/margins": 0.18985795974731445,
|
| 126 |
+
"rewards/rejected": -0.2545413374900818,
|
| 127 |
+
"sft_loss": 0.4562531113624573,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 128 |
"step": 70
|
| 129 |
},
|
| 130 |
{
|
| 131 |
+
"epoch": 0.11425192325930389,
|
| 132 |
+
"grad_norm": 4.25,
|
| 133 |
+
"learning_rate": 0.0002998135381828383,
|
| 134 |
+
"logits/chosen": -2.8170955181121826,
|
| 135 |
+
"logits/rejected": -2.8171167373657227,
|
| 136 |
+
"logps/chosen": -0.4997388422489166,
|
| 137 |
+
"logps/rejected": -2.737879514694214,
|
| 138 |
+
"loss": 0.5264,
|
| 139 |
+
"odds_ratio_loss": 1.6635347604751587,
|
| 140 |
+
"rewards/accuracies": 0.8994140625,
|
| 141 |
+
"rewards/chosen": -0.04997389018535614,
|
| 142 |
+
"rewards/margins": 0.22381405532360077,
|
| 143 |
+
"rewards/rejected": -0.2737879753112793,
|
| 144 |
+
"sft_loss": 0.3600570261478424,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 145 |
"step": 80
|
| 146 |
},
|
| 147 |
{
|
| 148 |
+
"epoch": 0.12853341366671686,
|
| 149 |
+
"grad_norm": 3.671875,
|
| 150 |
+
"learning_rate": 0.0002992546163048102,
|
| 151 |
+
"logits/chosen": -3.062329053878784,
|
| 152 |
+
"logits/rejected": -3.0623464584350586,
|
| 153 |
+
"logps/chosen": -0.4833168089389801,
|
| 154 |
+
"logps/rejected": -2.676713466644287,
|
| 155 |
+
"loss": 0.5082,
|
| 156 |
+
"odds_ratio_loss": 1.625109314918518,
|
| 157 |
+
"rewards/accuracies": 0.9037109613418579,
|
| 158 |
+
"rewards/chosen": -0.048331670463085175,
|
| 159 |
+
"rewards/margins": 0.21933968365192413,
|
| 160 |
+
"rewards/rejected": -0.2676713764667511,
|
| 161 |
+
"sft_loss": 0.3456498384475708,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 162 |
"step": 90
|
| 163 |
},
|
| 164 |
{
|
| 165 |
+
"epoch": 0.14281490407412986,
|
| 166 |
+
"grad_norm": 3.828125,
|
| 167 |
+
"learning_rate": 0.0002983246239337692,
|
| 168 |
+
"logits/chosen": -2.9990651607513428,
|
| 169 |
+
"logits/rejected": -2.9990792274475098,
|
| 170 |
+
"logps/chosen": -0.45154619216918945,
|
| 171 |
+
"logps/rejected": -2.796322822570801,
|
| 172 |
+
"loss": 0.4752,
|
| 173 |
+
"odds_ratio_loss": 1.524524450302124,
|
| 174 |
+
"rewards/accuracies": 0.907421886920929,
|
| 175 |
+
"rewards/chosen": -0.04515461623668671,
|
| 176 |
+
"rewards/margins": 0.2344777137041092,
|
| 177 |
+
"rewards/rejected": -0.2796323001384735,
|
| 178 |
+
"sft_loss": 0.3227214813232422,
|
| 179 |
+
"step": 100
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 180 |
},
|
| 181 |
{
|
| 182 |
+
"epoch": 0.15709639448154283,
|
| 183 |
+
"grad_norm": 4.625,
|
| 184 |
+
"learning_rate": 0.00029702587317728153,
|
| 185 |
+
"logits/chosen": -3.0073421001434326,
|
| 186 |
+
"logits/rejected": -3.0073623657226562,
|
| 187 |
+
"logps/chosen": -0.4845556318759918,
|
| 188 |
+
"logps/rejected": -2.689493417739868,
|
| 189 |
+
"loss": 0.5115,
|
| 190 |
+
"odds_ratio_loss": 1.6119966506958008,
|
| 191 |
+
"rewards/accuracies": 0.8880859613418579,
|
| 192 |
+
"rewards/chosen": -0.04845556616783142,
|
| 193 |
+
"rewards/margins": 0.22049376368522644,
|
| 194 |
+
"rewards/rejected": -0.26894932985305786,
|
| 195 |
+
"sft_loss": 0.3503072261810303,
|
| 196 |
+
"step": 110
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 0.17137788488895583,
|
| 200 |
+
"grad_norm": 4.25,
|
| 201 |
+
"learning_rate": 0.00029536159293436166,
|
| 202 |
+
"logits/chosen": -3.0959103107452393,
|
| 203 |
+
"logits/rejected": -3.095935583114624,
|
| 204 |
+
"logps/chosen": -0.4538491368293762,
|
| 205 |
+
"logps/rejected": -2.8936073780059814,
|
| 206 |
+
"loss": 0.4767,
|
| 207 |
+
"odds_ratio_loss": 1.5220377445220947,
|
| 208 |
+
"rewards/accuracies": 0.9091796875,
|
| 209 |
+
"rewards/chosen": -0.04538491368293762,
|
| 210 |
+
"rewards/margins": 0.24397583305835724,
|
| 211 |
+
"rewards/rejected": -0.28936073184013367,
|
| 212 |
+
"sft_loss": 0.32447534799575806,
|
| 213 |
+
"step": 120
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.1856593752963688,
|
| 217 |
+
"grad_norm": 3.015625,
|
| 218 |
+
"learning_rate": 0.00029333592086792107,
|
| 219 |
+
"logits/chosen": -3.2102882862091064,
|
| 220 |
+
"logits/rejected": -3.210312604904175,
|
| 221 |
+
"logps/chosen": -0.4560007154941559,
|
| 222 |
+
"logps/rejected": -2.9516916275024414,
|
| 223 |
+
"loss": 0.4775,
|
| 224 |
+
"odds_ratio_loss": 1.4568021297454834,
|
| 225 |
+
"rewards/accuracies": 0.9140625,
|
| 226 |
+
"rewards/chosen": -0.04560007154941559,
|
| 227 |
+
"rewards/margins": 0.2495690882205963,
|
| 228 |
+
"rewards/rejected": -0.2951691746711731,
|
| 229 |
+
"sft_loss": 0.33186882734298706,
|
| 230 |
+
"step": 130
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.1999408657037818,
|
| 234 |
+
"grad_norm": 4.625,
|
| 235 |
+
"learning_rate": 0.0002909538931178862,
|
| 236 |
+
"logits/chosen": -3.1817660331726074,
|
| 237 |
+
"logits/rejected": -3.181790828704834,
|
| 238 |
+
"logps/chosen": -0.4694591164588928,
|
| 239 |
+
"logps/rejected": -2.694068193435669,
|
| 240 |
+
"loss": 0.4932,
|
| 241 |
+
"odds_ratio_loss": 1.5894033908843994,
|
| 242 |
+
"rewards/accuracies": 0.903515636920929,
|
| 243 |
+
"rewards/chosen": -0.046945907175540924,
|
| 244 |
+
"rewards/margins": 0.22246094048023224,
|
| 245 |
+
"rewards/rejected": -0.26940685510635376,
|
| 246 |
+
"sft_loss": 0.3343026041984558,
|
| 247 |
+
"step": 140
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.21422235611119478,
|
| 251 |
+
"grad_norm": 2.109375,
|
| 252 |
+
"learning_rate": 0.00028822143178056114,
|
| 253 |
+
"logits/chosen": -3.19804310798645,
|
| 254 |
+
"logits/rejected": -3.1980957984924316,
|
| 255 |
+
"logps/chosen": -0.4429679811000824,
|
| 256 |
+
"logps/rejected": -2.8848023414611816,
|
| 257 |
+
"loss": 0.4646,
|
| 258 |
+
"odds_ratio_loss": 1.4565680027008057,
|
| 259 |
+
"rewards/accuracies": 0.9175781011581421,
|
| 260 |
+
"rewards/chosen": -0.04429679363965988,
|
| 261 |
+
"rewards/margins": 0.24418342113494873,
|
| 262 |
+
"rewards/rejected": -0.2884802222251892,
|
| 263 |
+
"sft_loss": 0.31895238161087036,
|
| 264 |
+
"step": 150
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.22850384651860778,
|
| 268 |
+
"grad_norm": 2.53125,
|
| 269 |
+
"learning_rate": 0.0002851453301853628,
|
| 270 |
+
"logits/chosen": -3.1286864280700684,
|
| 271 |
+
"logits/rejected": -3.1287217140197754,
|
| 272 |
+
"logps/chosen": -0.4620634913444519,
|
| 273 |
+
"logps/rejected": -2.8395490646362305,
|
| 274 |
+
"loss": 0.4861,
|
| 275 |
+
"odds_ratio_loss": 1.5178568363189697,
|
| 276 |
+
"rewards/accuracies": 0.8980468511581421,
|
| 277 |
+
"rewards/chosen": -0.04620635136961937,
|
| 278 |
+
"rewards/margins": 0.23774857819080353,
|
| 279 |
+
"rewards/rejected": -0.283954918384552,
|
| 280 |
+
"sft_loss": 0.33431634306907654,
|
| 281 |
+
"step": 160
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 0.24278533692602075,
|
| 285 |
+
"grad_norm": 3.8125,
|
| 286 |
+
"learning_rate": 0.0002817332360055343,
|
| 287 |
+
"logits/chosen": -3.0237438678741455,
|
| 288 |
+
"logits/rejected": -3.0237746238708496,
|
| 289 |
+
"logps/chosen": -0.4375666677951813,
|
| 290 |
+
"logps/rejected": -2.892333507537842,
|
| 291 |
+
"loss": 0.4602,
|
| 292 |
+
"odds_ratio_loss": 1.4504070281982422,
|
| 293 |
+
"rewards/accuracies": 0.9097656011581421,
|
| 294 |
+
"rewards/chosen": -0.04375666379928589,
|
| 295 |
+
"rewards/margins": 0.24547667801380157,
|
| 296 |
+
"rewards/rejected": -0.28923335671424866,
|
| 297 |
+
"sft_loss": 0.31515270471572876,
|
| 298 |
+
"step": 170
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.2570668273334337,
|
| 302 |
+
"grad_norm": 2.625,
|
| 303 |
+
"learning_rate": 0.0002779936322448233,
|
| 304 |
+
"logits/chosen": -3.0108470916748047,
|
| 305 |
+
"logits/rejected": -3.0108840465545654,
|
| 306 |
+
"logps/chosen": -0.43512678146362305,
|
| 307 |
+
"logps/rejected": -3.0354106426239014,
|
| 308 |
+
"loss": 0.4572,
|
| 309 |
+
"odds_ratio_loss": 1.3499114513397217,
|
| 310 |
+
"rewards/accuracies": 0.9091796875,
|
| 311 |
+
"rewards/chosen": -0.043512679636478424,
|
| 312 |
+
"rewards/margins": 0.2600283920764923,
|
| 313 |
+
"rewards/rejected": -0.30354106426239014,
|
| 314 |
+
"sft_loss": 0.32224926352500916,
|
| 315 |
+
"step": 180
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.2713483177408467,
|
| 319 |
+
"grad_norm": 2.78125,
|
| 320 |
+
"learning_rate": 0.00027393581614739923,
|
| 321 |
+
"logits/chosen": -3.0553345680236816,
|
| 322 |
+
"logits/rejected": -3.055368423461914,
|
| 323 |
+
"logps/chosen": -0.42374086380004883,
|
| 324 |
+
"logps/rejected": -2.915168523788452,
|
| 325 |
+
"loss": 0.445,
|
| 326 |
+
"odds_ratio_loss": 1.4213359355926514,
|
| 327 |
+
"rewards/accuracies": 0.9175781011581421,
|
| 328 |
+
"rewards/chosen": -0.042374081909656525,
|
| 329 |
+
"rewards/margins": 0.24914276599884033,
|
| 330 |
+
"rewards/rejected": -0.29151684045791626,
|
| 331 |
+
"sft_loss": 0.3028421401977539,
|
| 332 |
+
"step": 190
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.2856298081482597,
|
| 336 |
+
"grad_norm": 2.421875,
|
| 337 |
+
"learning_rate": 0.0002695698760834384,
|
| 338 |
+
"logits/chosen": -2.994476318359375,
|
| 339 |
+
"logits/rejected": -2.9945011138916016,
|
| 340 |
+
"logps/chosen": -0.4544607102870941,
|
| 341 |
+
"logps/rejected": -2.8547749519348145,
|
| 342 |
+
"loss": 0.4776,
|
| 343 |
+
"odds_ratio_loss": 1.4777902364730835,
|
| 344 |
+
"rewards/accuracies": 0.9033203125,
|
| 345 |
+
"rewards/chosen": -0.04544607177376747,
|
| 346 |
+
"rewards/margins": 0.240031436085701,
|
| 347 |
+
"rewards/rejected": -0.28547748923301697,
|
| 348 |
+
"sft_loss": 0.32984623312950134,
|
| 349 |
+
"step": 200
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 0.2999112985556727,
|
| 353 |
+
"grad_norm": 3.546875,
|
| 354 |
+
"learning_rate": 0.00026490666646784665,
|
| 355 |
+
"logits/chosen": -3.063324451446533,
|
| 356 |
+
"logits/rejected": -3.063349723815918,
|
| 357 |
+
"logps/chosen": -0.43639254570007324,
|
| 358 |
+
"logps/rejected": -3.105325698852539,
|
| 359 |
+
"loss": 0.4578,
|
| 360 |
+
"odds_ratio_loss": 1.3640453815460205,
|
| 361 |
+
"rewards/accuracies": 0.9140625,
|
| 362 |
+
"rewards/chosen": -0.04363925755023956,
|
| 363 |
+
"rewards/margins": 0.26689332723617554,
|
| 364 |
+
"rewards/rejected": -0.3105325698852539,
|
| 365 |
+
"sft_loss": 0.3214019536972046,
|
| 366 |
+
"step": 210
|
| 367 |
},
|
| 368 |
{
|
| 369 |
+
"epoch": 0.31419278896308567,
|
| 370 |
+
"grad_norm": 2.296875,
|
| 371 |
+
"learning_rate": 0.0002599577807744739,
|
| 372 |
+
"logits/chosen": -3.115455389022827,
|
| 373 |
+
"logits/rejected": -3.1154801845550537,
|
| 374 |
+
"logps/chosen": -0.4168368875980377,
|
| 375 |
+
"logps/rejected": -3.096985340118408,
|
| 376 |
+
"loss": 0.4371,
|
| 377 |
+
"odds_ratio_loss": 1.3493207693099976,
|
| 378 |
+
"rewards/accuracies": 0.9164062738418579,
|
| 379 |
+
"rewards/chosen": -0.04168368875980377,
|
| 380 |
+
"rewards/margins": 0.2680148482322693,
|
| 381 |
+
"rewards/rejected": -0.30969855189323425,
|
| 382 |
+
"sft_loss": 0.3021194040775299,
|
| 383 |
+
"step": 220
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 0.32847427937049867,
|
| 387 |
"grad_norm": 1.921875,
|
| 388 |
+
"learning_rate": 0.0002547355227129109,
|
| 389 |
+
"logits/chosen": -3.162436008453369,
|
| 390 |
+
"logits/rejected": -3.1624579429626465,
|
| 391 |
+
"logps/chosen": -0.4433667063713074,
|
| 392 |
+
"logps/rejected": -3.0064072608947754,
|
| 393 |
+
"loss": 0.4655,
|
| 394 |
+
"odds_ratio_loss": 1.3855293989181519,
|
| 395 |
+
"rewards/accuracies": 0.912304699420929,
|
| 396 |
+
"rewards/chosen": -0.04433666914701462,
|
| 397 |
+
"rewards/margins": 0.2563040852546692,
|
| 398 |
+
"rewards/rejected": -0.3006407618522644,
|
| 399 |
+
"sft_loss": 0.3269914984703064,
|
| 400 |
+
"step": 230
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 0.34275576977791167,
|
| 404 |
+
"grad_norm": 1.2734375,
|
| 405 |
+
"learning_rate": 0.0002492528756395289,
|
| 406 |
+
"logits/chosen": -3.1590659618377686,
|
| 407 |
+
"logits/rejected": -3.1590869426727295,
|
| 408 |
+
"logps/chosen": -0.4178268015384674,
|
| 409 |
+
"logps/rejected": -3.0285942554473877,
|
| 410 |
+
"loss": 0.4376,
|
| 411 |
+
"odds_ratio_loss": 1.3180664777755737,
|
| 412 |
+
"rewards/accuracies": 0.917187511920929,
|
| 413 |
+
"rewards/chosen": -0.0417826846241951,
|
| 414 |
+
"rewards/margins": 0.26107674837112427,
|
| 415 |
+
"rewards/rejected": -0.30285942554473877,
|
| 416 |
+
"sft_loss": 0.3058391213417053,
|
| 417 |
+
"step": 240
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.35703726018532467,
|
| 421 |
+
"grad_norm": 4.125,
|
| 422 |
+
"learning_rate": 0.00024352347027881003,
|
| 423 |
+
"logits/chosen": -3.2828221321105957,
|
| 424 |
+
"logits/rejected": -3.282838821411133,
|
| 425 |
+
"logps/chosen": -0.4194249212741852,
|
| 426 |
+
"logps/rejected": -3.0415470600128174,
|
| 427 |
+
"loss": 0.4403,
|
| 428 |
+
"odds_ratio_loss": 1.3526116609573364,
|
| 429 |
+
"rewards/accuracies": 0.9166015386581421,
|
| 430 |
+
"rewards/chosen": -0.041942495852708817,
|
| 431 |
+
"rewards/margins": 0.26221221685409546,
|
| 432 |
+
"rewards/rejected": -0.3041546940803528,
|
| 433 |
+
"sft_loss": 0.3050472140312195,
|
| 434 |
+
"step": 250
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.3713187505927376,
|
| 438 |
+
"grad_norm": 2.640625,
|
| 439 |
+
"learning_rate": 0.00023756155083521846,
|
| 440 |
+
"logits/chosen": -3.22637677192688,
|
| 441 |
+
"logits/rejected": -3.2263927459716797,
|
| 442 |
+
"logps/chosen": -0.41995421051979065,
|
| 443 |
+
"logps/rejected": -3.0328941345214844,
|
| 444 |
+
"loss": 0.4417,
|
| 445 |
+
"odds_ratio_loss": 1.3745537996292114,
|
| 446 |
+
"rewards/accuracies": 0.9140625,
|
| 447 |
+
"rewards/chosen": -0.04199542477726936,
|
| 448 |
+
"rewards/margins": 0.26129403710365295,
|
| 449 |
+
"rewards/rejected": -0.3032894432544708,
|
| 450 |
+
"sft_loss": 0.30421775579452515,
|
| 451 |
+
"step": 260
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.3856002410001506,
|
| 455 |
+
"grad_norm": 2.578125,
|
| 456 |
+
"learning_rate": 0.0002313819395798639,
|
| 457 |
+
"logits/chosen": -3.186093330383301,
|
| 458 |
+
"logits/rejected": -3.1861069202423096,
|
| 459 |
+
"logps/chosen": -0.4216841161251068,
|
| 460 |
+
"logps/rejected": -3.125148057937622,
|
| 461 |
+
"loss": 0.4439,
|
| 462 |
+
"odds_ratio_loss": 1.3766355514526367,
|
| 463 |
+
"rewards/accuracies": 0.913281261920929,
|
| 464 |
+
"rewards/chosen": -0.04216841608285904,
|
| 465 |
+
"rewards/margins": 0.27034634351730347,
|
| 466 |
+
"rewards/rejected": -0.3125148117542267,
|
| 467 |
+
"sft_loss": 0.30621883273124695,
|
| 468 |
+
"step": 270
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 0.3998817314075636,
|
| 472 |
+
"grad_norm": 2.03125,
|
| 473 |
+
"learning_rate": 0.000225,
|
| 474 |
+
"logits/chosen": -3.167736530303955,
|
| 475 |
+
"logits/rejected": -3.167752265930176,
|
| 476 |
+
"logps/chosen": -0.42215681076049805,
|
| 477 |
+
"logps/rejected": -3.0445384979248047,
|
| 478 |
+
"loss": 0.4415,
|
| 479 |
+
"odds_ratio_loss": 1.3319975137710571,
|
| 480 |
+
"rewards/accuracies": 0.9189453125,
|
| 481 |
+
"rewards/chosen": -0.04221567511558533,
|
| 482 |
+
"rewards/margins": 0.26223814487457275,
|
| 483 |
+
"rewards/rejected": -0.30445384979248047,
|
| 484 |
+
"sft_loss": 0.3082923889160156,
|
| 485 |
+
"step": 280
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.4141632218149766,
|
| 489 |
+
"grad_norm": 1.9140625,
|
| 490 |
+
"learning_rate": 0.00021843159860297442,
|
| 491 |
+
"logits/chosen": -3.2080886363983154,
|
| 492 |
+
"logits/rejected": -3.208108425140381,
|
| 493 |
+
"logps/chosen": -0.3955201208591461,
|
| 494 |
+
"logps/rejected": -3.1413300037384033,
|
| 495 |
+
"loss": 0.4162,
|
| 496 |
+
"odds_ratio_loss": 1.298853874206543,
|
| 497 |
+
"rewards/accuracies": 0.9166015386581421,
|
| 498 |
+
"rewards/chosen": -0.03955201059579849,
|
| 499 |
+
"rewards/margins": 0.27458101511001587,
|
| 500 |
+
"rewards/rejected": -0.3141329884529114,
|
| 501 |
+
"sft_loss": 0.28630274534225464,
|
| 502 |
+
"step": 290
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 0.42844471222238956,
|
| 506 |
+
"grad_norm": 2.296875,
|
| 507 |
+
"learning_rate": 0.00021169306546959174,
|
| 508 |
+
"logits/chosen": -3.1625964641571045,
|
| 509 |
+
"logits/rejected": -3.1626217365264893,
|
| 510 |
+
"logps/chosen": -0.4014604091644287,
|
| 511 |
+
"logps/rejected": -3.231706142425537,
|
| 512 |
+
"loss": 0.4211,
|
| 513 |
+
"odds_ratio_loss": 1.2652801275253296,
|
| 514 |
+
"rewards/accuracies": 0.91796875,
|
| 515 |
+
"rewards/chosen": -0.04014604538679123,
|
| 516 |
+
"rewards/margins": 0.2830246090888977,
|
| 517 |
+
"rewards/rejected": -0.32317066192626953,
|
| 518 |
+
"sft_loss": 0.2946000099182129,
|
| 519 |
+
"step": 300
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.44272620262980256,
|
| 523 |
+
"grad_norm": 1.9765625,
|
| 524 |
+
"learning_rate": 0.00020480115365495926,
|
| 525 |
+
"logits/chosen": -3.1747231483459473,
|
| 526 |
+
"logits/rejected": -3.1747519969940186,
|
| 527 |
+
"logps/chosen": -0.3960801064968109,
|
| 528 |
+
"logps/rejected": -3.082359790802002,
|
| 529 |
+
"loss": 0.4173,
|
| 530 |
+
"odds_ratio_loss": 1.3159233331680298,
|
| 531 |
+
"rewards/accuracies": 0.9173828363418579,
|
| 532 |
+
"rewards/chosen": -0.03960801288485527,
|
| 533 |
+
"rewards/margins": 0.26862797141075134,
|
| 534 |
+
"rewards/rejected": -0.30823594331741333,
|
| 535 |
+
"sft_loss": 0.2857065200805664,
|
| 536 |
+
"step": 310
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 0.45700769303721556,
|
| 540 |
+
"grad_norm": 2.90625,
|
| 541 |
+
"learning_rate": 0.00019777299753775265,
|
| 542 |
+
"logits/chosen": -3.2027382850646973,
|
| 543 |
+
"logits/rejected": -3.202775478363037,
|
| 544 |
+
"logps/chosen": -0.3917561173439026,
|
| 545 |
+
"logps/rejected": -3.128166675567627,
|
| 546 |
+
"loss": 0.4113,
|
| 547 |
+
"odds_ratio_loss": 1.213888168334961,
|
| 548 |
+
"rewards/accuracies": 0.923046886920929,
|
| 549 |
+
"rewards/chosen": -0.0391756072640419,
|
| 550 |
+
"rewards/margins": 0.27364104986190796,
|
| 551 |
+
"rewards/rejected": -0.31281667947769165,
|
| 552 |
+
"sft_loss": 0.28995418548583984,
|
| 553 |
+
"step": 320
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 0.47128918344462856,
|
| 557 |
+
"grad_norm": 1.1171875,
|
| 558 |
+
"learning_rate": 0.00019062607022145078,
|
| 559 |
+
"logits/chosen": -3.223431348800659,
|
| 560 |
+
"logits/rejected": -3.2234749794006348,
|
| 561 |
+
"logps/chosen": -0.3959726393222809,
|
| 562 |
+
"logps/rejected": -3.1301980018615723,
|
| 563 |
+
"loss": 0.4159,
|
| 564 |
+
"odds_ratio_loss": 1.2752103805541992,
|
| 565 |
+
"rewards/accuracies": 0.9134765863418579,
|
| 566 |
+
"rewards/chosen": -0.03959726542234421,
|
| 567 |
+
"rewards/margins": 0.2734225392341614,
|
| 568 |
+
"rewards/rejected": -0.3130198121070862,
|
| 569 |
+
"sft_loss": 0.2884255647659302,
|
| 570 |
+
"step": 330
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.4855706738520415,
|
| 574 |
+
"grad_norm": 2.421875,
|
| 575 |
+
"learning_rate": 0.00018337814009344714,
|
| 576 |
+
"logits/chosen": -3.229165554046631,
|
| 577 |
+
"logits/rejected": -3.22920298576355,
|
| 578 |
+
"logps/chosen": -0.40863022208213806,
|
| 579 |
+
"logps/rejected": -3.2035133838653564,
|
| 580 |
+
"loss": 0.4293,
|
| 581 |
+
"odds_ratio_loss": 1.3022868633270264,
|
| 582 |
+
"rewards/accuracies": 0.923046886920929,
|
| 583 |
+
"rewards/chosen": -0.040863025933504105,
|
| 584 |
+
"rewards/margins": 0.27948835492134094,
|
| 585 |
+
"rewards/rejected": -0.32035139203071594,
|
| 586 |
+
"sft_loss": 0.29906368255615234,
|
| 587 |
+
"step": 340
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 0.4998521642594545,
|
| 591 |
+
"grad_norm": 1.6953125,
|
| 592 |
+
"learning_rate": 0.00017604722665003956,
|
| 593 |
+
"logits/chosen": -3.268237590789795,
|
| 594 |
+
"logits/rejected": -3.268270969390869,
|
| 595 |
+
"logps/chosen": -0.3820918798446655,
|
| 596 |
+
"logps/rejected": -3.3142802715301514,
|
| 597 |
+
"loss": 0.4012,
|
| 598 |
+
"odds_ratio_loss": 1.2229855060577393,
|
| 599 |
+
"rewards/accuracies": 0.924023449420929,
|
| 600 |
+
"rewards/chosen": -0.03820918872952461,
|
| 601 |
+
"rewards/margins": 0.29321882128715515,
|
| 602 |
+
"rewards/rejected": -0.33142799139022827,
|
| 603 |
+
"sft_loss": 0.278933584690094,
|
| 604 |
+
"step": 350
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.5141336546668674,
|
| 608 |
+
"grad_norm": 2.375,
|
| 609 |
+
"learning_rate": 0.00016865155569712278,
|
| 610 |
+
"logits/chosen": -3.3011035919189453,
|
| 611 |
+
"logits/rejected": -3.3011412620544434,
|
| 612 |
+
"logps/chosen": -0.38040798902511597,
|
| 613 |
+
"logps/rejected": -3.2756595611572266,
|
| 614 |
+
"loss": 0.3985,
|
| 615 |
+
"odds_ratio_loss": 1.1769336462020874,
|
| 616 |
+
"rewards/accuracies": 0.9302734136581421,
|
| 617 |
+
"rewards/chosen": -0.03804079815745354,
|
| 618 |
+
"rewards/margins": 0.2895251214504242,
|
| 619 |
+
"rewards/rejected": -0.3275659680366516,
|
| 620 |
+
"sft_loss": 0.28078263998031616,
|
| 621 |
+
"step": 360
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 0.5284151450742804,
|
| 625 |
+
"grad_norm": 1.484375,
|
| 626 |
+
"learning_rate": 0.00016120951403796364,
|
| 627 |
+
"logits/chosen": -3.336045026779175,
|
| 628 |
+
"logits/rejected": -3.3360836505889893,
|
| 629 |
+
"logps/chosen": -0.3747532069683075,
|
| 630 |
+
"logps/rejected": -3.3229317665100098,
|
| 631 |
+
"loss": 0.3938,
|
| 632 |
+
"odds_ratio_loss": 1.2170263528823853,
|
| 633 |
+
"rewards/accuracies": 0.924023449420929,
|
| 634 |
+
"rewards/chosen": -0.03747531771659851,
|
| 635 |
+
"rewards/margins": 0.29481783509254456,
|
| 636 |
+
"rewards/rejected": -0.33229315280914307,
|
| 637 |
+
"sft_loss": 0.2720716595649719,
|
| 638 |
+
"step": 370
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 0.5426966354816934,
|
| 642 |
+
"grad_norm": 1.1953125,
|
| 643 |
+
"learning_rate": 0.00015373960376071093,
|
| 644 |
+
"logits/chosen": -3.3047919273376465,
|
| 645 |
+
"logits/rejected": -3.3048160076141357,
|
| 646 |
+
"logps/chosen": -0.37664586305618286,
|
| 647 |
+
"logps/rejected": -3.1220898628234863,
|
| 648 |
+
"loss": 0.3961,
|
| 649 |
+
"odds_ratio_loss": 1.2487261295318604,
|
| 650 |
+
"rewards/accuracies": 0.921875,
|
| 651 |
+
"rewards/chosen": -0.037664588540792465,
|
| 652 |
+
"rewards/margins": 0.2745443880558014,
|
| 653 |
+
"rewards/rejected": -0.31220895051956177,
|
| 654 |
+
"sft_loss": 0.2712169289588928,
|
| 655 |
+
"step": 380
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 0.5569781258891064,
|
| 659 |
+
"grad_norm": 2.015625,
|
| 660 |
+
"learning_rate": 0.00014626039623928907,
|
| 661 |
+
"logits/chosen": -3.3392891883850098,
|
| 662 |
+
"logits/rejected": -3.339310884475708,
|
| 663 |
+
"logps/chosen": -0.3588925302028656,
|
| 664 |
+
"logps/rejected": -3.4983272552490234,
|
| 665 |
+
"loss": 0.3772,
|
| 666 |
+
"odds_ratio_loss": 1.1625574827194214,
|
| 667 |
+
"rewards/accuracies": 0.927734375,
|
| 668 |
+
"rewards/chosen": -0.03588924929499626,
|
| 669 |
+
"rewards/margins": 0.31394344568252563,
|
| 670 |
+
"rewards/rejected": -0.3498327136039734,
|
| 671 |
+
"sft_loss": 0.2609647512435913,
|
| 672 |
+
"step": 390
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 0.5712596162965194,
|
| 676 |
+
"grad_norm": 1.5078125,
|
| 677 |
+
"learning_rate": 0.00013879048596203636,
|
| 678 |
+
"logits/chosen": -3.382007598876953,
|
| 679 |
+
"logits/rejected": -3.3820137977600098,
|
| 680 |
+
"logps/chosen": -0.3696475028991699,
|
| 681 |
+
"logps/rejected": -3.3737378120422363,
|
| 682 |
+
"loss": 0.3872,
|
| 683 |
+
"odds_ratio_loss": 1.1461818218231201,
|
| 684 |
+
"rewards/accuracies": 0.929882824420929,
|
| 685 |
+
"rewards/chosen": -0.03696475178003311,
|
| 686 |
+
"rewards/margins": 0.3004090189933777,
|
| 687 |
+
"rewards/rejected": -0.3373737931251526,
|
| 688 |
+
"sft_loss": 0.27257078886032104,
|
| 689 |
+
"step": 400
|
| 690 |
}
|
| 691 |
],
|
| 692 |
+
"logging_steps": 10,
|
| 693 |
+
"max_steps": 700,
|
| 694 |
"num_input_tokens_seen": 0,
|
| 695 |
"num_train_epochs": 1,
|
| 696 |
"save_steps": 100,
|
|
|
|
| 706 |
"attributes": {}
|
| 707 |
}
|
| 708 |
},
|
| 709 |
+
"total_flos": 3.1756323933484155e+18,
|
| 710 |
"train_batch_size": 2,
|
| 711 |
"trial_name": null,
|
| 712 |
"trial_params": null
|