Model save
Browse files- all_results.json +7 -7
- eval_results.json +4 -4
- model.safetensors +1 -1
- train_results.json +3 -3
- trainer_state.json +74 -752
- training_args.bin +1 -1
all_results.json
CHANGED
|
@@ -1,13 +1,13 @@
|
|
| 1 |
{
|
| 2 |
-
"eval_loss": 0.
|
| 3 |
-
"eval_runtime":
|
| 4 |
"eval_samples": 99,
|
| 5 |
-
"eval_samples_per_second": 1.
|
| 6 |
-
"eval_steps_per_second": 0.
|
| 7 |
"total_flos": 0.0,
|
| 8 |
-
"train_loss":
|
| 9 |
-
"train_runtime":
|
| 10 |
"train_samples": 72441,
|
| 11 |
-
"train_samples_per_second": 1.
|
| 12 |
"train_steps_per_second": 0.007
|
| 13 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"eval_loss": 0.010948318056762218,
|
| 3 |
+
"eval_runtime": 54.3753,
|
| 4 |
"eval_samples": 99,
|
| 5 |
+
"eval_samples_per_second": 1.821,
|
| 6 |
+
"eval_steps_per_second": 0.018,
|
| 7 |
"total_flos": 0.0,
|
| 8 |
+
"train_loss": 0.011041131547906181,
|
| 9 |
+
"train_runtime": 4812.1769,
|
| 10 |
"train_samples": 72441,
|
| 11 |
+
"train_samples_per_second": 1.505,
|
| 12 |
"train_steps_per_second": 0.007
|
| 13 |
}
|
eval_results.json
CHANGED
|
@@ -1,7 +1,7 @@
|
|
| 1 |
{
|
| 2 |
-
"eval_loss": 0.
|
| 3 |
-
"eval_runtime":
|
| 4 |
"eval_samples": 99,
|
| 5 |
-
"eval_samples_per_second": 1.
|
| 6 |
-
"eval_steps_per_second": 0.
|
| 7 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"eval_loss": 0.010948318056762218,
|
| 3 |
+
"eval_runtime": 54.3753,
|
| 4 |
"eval_samples": 99,
|
| 5 |
+
"eval_samples_per_second": 1.821,
|
| 6 |
+
"eval_steps_per_second": 0.018
|
| 7 |
}
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3554214752
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b3f3560b38352a4c95bb07ebcf72fa472b76368a8a6b30545579079832d974f0
|
| 3 |
size 3554214752
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"total_flos": 0.0,
|
| 3 |
-
"train_loss":
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 72441,
|
| 6 |
-
"train_samples_per_second": 1.
|
| 7 |
"train_steps_per_second": 0.007
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 0.011041131547906181,
|
| 4 |
+
"train_runtime": 4812.1769,
|
| 5 |
"train_samples": 72441,
|
| 6 |
+
"train_samples_per_second": 1.505,
|
| 7 |
"train_steps_per_second": 0.007
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,787 +1,109 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 100,
|
| 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 |
-
"completion_length":
|
| 13 |
-
"epoch": 0.
|
| 14 |
-
"grad_norm": 0.
|
| 15 |
-
"kl": 0.
|
| 16 |
-
"learning_rate":
|
| 17 |
-
"loss": 0.
|
| 18 |
-
"reward": 0.
|
| 19 |
-
"reward_std": 0.
|
| 20 |
-
"rewards/accuracy_reward": 0.
|
| 21 |
-
"rewards/format_reward": 0.
|
| 22 |
"step": 5
|
| 23 |
},
|
| 24 |
{
|
| 25 |
-
"completion_length":
|
| 26 |
-
"epoch": 0.
|
| 27 |
-
"grad_norm": 0.
|
| 28 |
-
"kl": 0.
|
| 29 |
-
"learning_rate":
|
| 30 |
-
"loss": 0.
|
| 31 |
-
"reward":
|
| 32 |
-
"reward_std": 0.
|
| 33 |
-
"rewards/accuracy_reward": 0.
|
| 34 |
-
"rewards/format_reward": 0.
|
| 35 |
"step": 10
|
| 36 |
},
|
| 37 |
{
|
| 38 |
-
"completion_length":
|
| 39 |
-
"epoch": 0.
|
| 40 |
-
"grad_norm": 0.
|
| 41 |
-
"kl": 0.
|
| 42 |
-
"learning_rate": 1.
|
| 43 |
-
"loss": 0.
|
| 44 |
-
"reward":
|
| 45 |
-
"reward_std": 0.
|
| 46 |
-
"rewards/accuracy_reward": 0.
|
| 47 |
-
"rewards/format_reward": 0.
|
| 48 |
"step": 15
|
| 49 |
},
|
| 50 |
{
|
| 51 |
-
"completion_length":
|
| 52 |
-
"epoch": 0.
|
| 53 |
-
"grad_norm": 0.
|
| 54 |
-
"kl": 0.
|
| 55 |
-
"learning_rate":
|
| 56 |
-
"loss": 0.
|
| 57 |
-
"reward": 1.
|
| 58 |
-
"reward_std": 0.
|
| 59 |
-
"rewards/accuracy_reward": 0.
|
| 60 |
-
"rewards/format_reward": 0.
|
| 61 |
"step": 20
|
| 62 |
},
|
| 63 |
{
|
| 64 |
-
"completion_length":
|
| 65 |
-
"epoch": 0.
|
| 66 |
-
"grad_norm": 0.
|
| 67 |
-
"kl": 0.
|
| 68 |
-
"learning_rate":
|
| 69 |
-
"loss": 0.
|
| 70 |
-
"reward": 1.
|
| 71 |
-
"reward_std": 0.
|
| 72 |
-
"rewards/accuracy_reward": 0.
|
| 73 |
-
"rewards/format_reward": 0.
|
| 74 |
"step": 25
|
| 75 |
},
|
| 76 |
{
|
| 77 |
-
"completion_length":
|
| 78 |
-
"epoch": 0.
|
| 79 |
-
"grad_norm": 0.
|
| 80 |
-
"kl": 0.
|
| 81 |
-
"learning_rate":
|
| 82 |
-
"loss": 0.
|
| 83 |
-
"reward": 1.
|
| 84 |
-
"reward_std": 0.
|
| 85 |
-
"rewards/accuracy_reward": 0.
|
| 86 |
-
"rewards/format_reward": 0.
|
| 87 |
"step": 30
|
| 88 |
},
|
| 89 |
{
|
| 90 |
-
"completion_length":
|
| 91 |
-
"epoch": 0.
|
| 92 |
-
"
|
| 93 |
-
"
|
| 94 |
-
"
|
| 95 |
-
"
|
| 96 |
-
"
|
| 97 |
-
"
|
| 98 |
-
"rewards/accuracy_reward": 0.25837054699659345,
|
| 99 |
-
"rewards/format_reward": 0.9812500409781932,
|
| 100 |
-
"step": 35
|
| 101 |
-
},
|
| 102 |
-
{
|
| 103 |
-
"completion_length": 183.1122851371765,
|
| 104 |
-
"epoch": 0.1413427561837456,
|
| 105 |
-
"grad_norm": 0.0162435844540596,
|
| 106 |
-
"kl": 0.210980224609375,
|
| 107 |
-
"learning_rate": 1.9907590277344582e-05,
|
| 108 |
-
"loss": 0.0084,
|
| 109 |
-
"reward": 1.2477679178118706,
|
| 110 |
-
"reward_std": 0.2695736668072641,
|
| 111 |
-
"rewards/accuracy_reward": 0.26607143888249996,
|
| 112 |
-
"rewards/format_reward": 0.9816964700818062,
|
| 113 |
-
"step": 40
|
| 114 |
-
},
|
| 115 |
-
{
|
| 116 |
-
"completion_length": 238.26184139251708,
|
| 117 |
-
"epoch": 0.15901060070671377,
|
| 118 |
-
"grad_norm": 0.4197707176208496,
|
| 119 |
-
"kl": 7394.361682128906,
|
| 120 |
-
"learning_rate": 1.9804824871166254e-05,
|
| 121 |
-
"loss": 295.67,
|
| 122 |
-
"reward": 1.234151841700077,
|
| 123 |
-
"reward_std": 0.2784772401675582,
|
| 124 |
-
"rewards/accuracy_reward": 0.26450893925502894,
|
| 125 |
-
"rewards/format_reward": 0.9696428991854191,
|
| 126 |
-
"step": 45
|
| 127 |
-
},
|
| 128 |
-
{
|
| 129 |
-
"completion_length": 220.17757682800294,
|
| 130 |
-
"epoch": 0.17667844522968199,
|
| 131 |
-
"grad_norm": 0.015375903807580471,
|
| 132 |
-
"kl": 0.17265625,
|
| 133 |
-
"learning_rate": 1.9664573064143604e-05,
|
| 134 |
-
"loss": 0.0069,
|
| 135 |
-
"reward": 1.2408482730388641,
|
| 136 |
-
"reward_std": 0.2870653722435236,
|
| 137 |
-
"rewards/accuracy_reward": 0.26808036947622893,
|
| 138 |
-
"rewards/format_reward": 0.9727678991854191,
|
| 139 |
-
"step": 50
|
| 140 |
-
},
|
| 141 |
-
{
|
| 142 |
-
"completion_length": 231.17356071472167,
|
| 143 |
-
"epoch": 0.19434628975265017,
|
| 144 |
-
"grad_norm": 0.2615525424480438,
|
| 145 |
-
"kl": 0.564453125,
|
| 146 |
-
"learning_rate": 1.948737107548771e-05,
|
| 147 |
-
"loss": 0.0225,
|
| 148 |
-
"reward": 1.2294643357396127,
|
| 149 |
-
"reward_std": 0.3038607369177043,
|
| 150 |
-
"rewards/accuracy_reward": 0.277008942887187,
|
| 151 |
-
"rewards/format_reward": 0.9524553999304771,
|
| 152 |
-
"step": 55
|
| 153 |
-
},
|
| 154 |
-
{
|
| 155 |
-
"completion_length": 263.1629590988159,
|
| 156 |
-
"epoch": 0.21201413427561838,
|
| 157 |
-
"grad_norm": 0.6179373264312744,
|
| 158 |
-
"kl": 4.37230224609375,
|
| 159 |
-
"learning_rate": 1.9273896394584103e-05,
|
| 160 |
-
"loss": 0.1748,
|
| 161 |
-
"reward": 1.0507812924683093,
|
| 162 |
-
"reward_std": 0.4720887938514352,
|
| 163 |
-
"rewards/accuracy_reward": 0.2530134045518935,
|
| 164 |
-
"rewards/format_reward": 0.7977678947150707,
|
| 165 |
-
"step": 60
|
| 166 |
-
},
|
| 167 |
-
{
|
| 168 |
-
"completion_length": 202.77768750190734,
|
| 169 |
-
"epoch": 0.22968197879858657,
|
| 170 |
-
"grad_norm": 0.04027915000915527,
|
| 171 |
-
"kl": 0.9251708984375,
|
| 172 |
-
"learning_rate": 1.9024965190774262e-05,
|
| 173 |
-
"loss": 0.037,
|
| 174 |
-
"reward": 0.9822545044124127,
|
| 175 |
-
"reward_std": 0.5017600081861019,
|
| 176 |
-
"rewards/accuracy_reward": 0.2608259034343064,
|
| 177 |
-
"rewards/format_reward": 0.7214286051690578,
|
| 178 |
-
"step": 65
|
| 179 |
-
},
|
| 180 |
-
{
|
| 181 |
-
"completion_length": 116.15926914215088,
|
| 182 |
-
"epoch": 0.24734982332155478,
|
| 183 |
-
"grad_norm": 0.03792489692568779,
|
| 184 |
-
"kl": 0.6725341796875,
|
| 185 |
-
"learning_rate": 1.8741529192927528e-05,
|
| 186 |
-
"loss": 0.0269,
|
| 187 |
-
"reward": 1.1287946909666062,
|
| 188 |
-
"reward_std": 0.4154288250952959,
|
| 189 |
-
"rewards/accuracy_reward": 0.28281251303851607,
|
| 190 |
-
"rewards/format_reward": 0.8459821805357933,
|
| 191 |
-
"step": 70
|
| 192 |
-
},
|
| 193 |
-
{
|
| 194 |
-
"completion_length": 168.19241857528687,
|
| 195 |
-
"epoch": 0.26501766784452296,
|
| 196 |
-
"grad_norm": 0.22333890199661255,
|
| 197 |
-
"kl": 1.2605224609375,
|
| 198 |
-
"learning_rate": 1.8424672050733577e-05,
|
| 199 |
-
"loss": 0.0504,
|
| 200 |
-
"reward": 1.0388393312692643,
|
| 201 |
-
"reward_std": 0.46060552783310416,
|
| 202 |
-
"rewards/accuracy_reward": 0.26049108393490317,
|
| 203 |
-
"rewards/format_reward": 0.7783482506871223,
|
| 204 |
-
"step": 75
|
| 205 |
-
},
|
| 206 |
-
{
|
| 207 |
-
"completion_length": 191.00815525054932,
|
| 208 |
-
"epoch": 0.2826855123674912,
|
| 209 |
-
"grad_norm": 0.03778946399688721,
|
| 210 |
-
"kl": 1.990380859375,
|
| 211 |
-
"learning_rate": 1.8075605191627242e-05,
|
| 212 |
-
"loss": 0.0796,
|
| 213 |
-
"reward": 1.128348258137703,
|
| 214 |
-
"reward_std": 0.4131897557526827,
|
| 215 |
-
"rewards/accuracy_reward": 0.2717634063214064,
|
| 216 |
-
"rewards/format_reward": 0.8565848588943481,
|
| 217 |
-
"step": 80
|
| 218 |
-
},
|
| 219 |
-
{
|
| 220 |
-
"completion_length": 104.5110532283783,
|
| 221 |
-
"epoch": 0.3003533568904594,
|
| 222 |
-
"grad_norm": 0.3576684594154358,
|
| 223 |
-
"kl": 0.60687255859375,
|
| 224 |
-
"learning_rate": 1.7695663189185703e-05,
|
| 225 |
-
"loss": 0.0243,
|
| 226 |
-
"reward": 1.2685268357396127,
|
| 227 |
-
"reward_std": 0.23934431467205286,
|
| 228 |
-
"rewards/accuracy_reward": 0.28872769260779024,
|
| 229 |
-
"rewards/format_reward": 0.9797991432249546,
|
| 230 |
-
"step": 85
|
| 231 |
-
},
|
| 232 |
-
{
|
| 233 |
-
"completion_length": 261.205256652832,
|
| 234 |
-
"epoch": 0.31802120141342755,
|
| 235 |
-
"grad_norm": 0.16514165699481964,
|
| 236 |
-
"kl": 0.769873046875,
|
| 237 |
-
"learning_rate": 1.7286298660705877e-05,
|
| 238 |
-
"loss": 0.0308,
|
| 239 |
-
"reward": 1.2081473730504513,
|
| 240 |
-
"reward_std": 0.3599423123523593,
|
| 241 |
-
"rewards/accuracy_reward": 0.2926339415833354,
|
| 242 |
-
"rewards/format_reward": 0.9155134320259094,
|
| 243 |
-
"step": 90
|
| 244 |
-
},
|
| 245 |
-
{
|
| 246 |
-
"completion_length": 337.4015787124634,
|
| 247 |
-
"epoch": 0.33568904593639576,
|
| 248 |
-
"grad_norm": 0.044928278774023056,
|
| 249 |
-
"kl": 0.8098876953125,
|
| 250 |
-
"learning_rate": 1.6849076713469914e-05,
|
| 251 |
-
"loss": 0.0324,
|
| 252 |
-
"reward": 1.1027902282774449,
|
| 253 |
-
"reward_std": 0.47088346295058725,
|
| 254 |
-
"rewards/accuracy_reward": 0.27622769204899666,
|
| 255 |
-
"rewards/format_reward": 0.8265625357627868,
|
| 256 |
-
"step": 95
|
| 257 |
-
},
|
| 258 |
-
{
|
| 259 |
-
"completion_length": 254.90804748535157,
|
| 260 |
-
"epoch": 0.35335689045936397,
|
| 261 |
-
"grad_norm": 0.15292082726955414,
|
| 262 |
-
"kl": 0.64703369140625,
|
| 263 |
-
"learning_rate": 1.6385668960932143e-05,
|
| 264 |
-
"loss": 0.0259,
|
| 265 |
-
"reward": 1.1753348790109157,
|
| 266 |
-
"reward_std": 0.39564149640500546,
|
| 267 |
-
"rewards/accuracy_reward": 0.2820312621071935,
|
| 268 |
-
"rewards/format_reward": 0.8933036133646965,
|
| 269 |
-
"step": 100
|
| 270 |
-
},
|
| 271 |
-
{
|
| 272 |
-
"epoch": 0.35335689045936397,
|
| 273 |
-
"eval_completion_length": 208.94378662109375,
|
| 274 |
-
"eval_kl": 0.6962890625,
|
| 275 |
-
"eval_loss": 0.028458675369620323,
|
| 276 |
-
"eval_reward": 1.254464328289032,
|
| 277 |
-
"eval_reward_std": 0.361492395401001,
|
| 278 |
-
"eval_rewards/accuracy_reward": 0.3147321566939354,
|
| 279 |
-
"eval_rewards/format_reward": 0.9397321790456772,
|
| 280 |
-
"eval_runtime": 45.1904,
|
| 281 |
-
"eval_samples_per_second": 2.191,
|
| 282 |
-
"eval_steps_per_second": 0.022,
|
| 283 |
-
"step": 100
|
| 284 |
-
},
|
| 285 |
-
{
|
| 286 |
-
"completion_length": 194.93382568359374,
|
| 287 |
-
"epoch": 0.3710247349823322,
|
| 288 |
-
"grad_norm": 0.044945139437913895,
|
| 289 |
-
"kl": 0.80966796875,
|
| 290 |
-
"learning_rate": 1.5897847131705194e-05,
|
| 291 |
-
"loss": 0.0324,
|
| 292 |
-
"reward": 1.2772321939468383,
|
| 293 |
-
"reward_std": 0.30906384270638226,
|
| 294 |
-
"rewards/accuracy_reward": 0.3136160854250193,
|
| 295 |
-
"rewards/format_reward": 0.9636161163449287,
|
| 296 |
-
"step": 105
|
| 297 |
-
},
|
| 298 |
-
{
|
| 299 |
-
"completion_length": 181.32556705474855,
|
| 300 |
-
"epoch": 0.38869257950530034,
|
| 301 |
-
"grad_norm": 0.0390905924141407,
|
| 302 |
-
"kl": 0.6416748046875,
|
| 303 |
-
"learning_rate": 1.5387476295779737e-05,
|
| 304 |
-
"loss": 0.0257,
|
| 305 |
-
"reward": 1.294196479022503,
|
| 306 |
-
"reward_std": 0.26708812834694984,
|
| 307 |
-
"rewards/accuracy_reward": 0.3142857293598354,
|
| 308 |
-
"rewards/format_reward": 0.9799107529222966,
|
| 309 |
-
"step": 110
|
| 310 |
-
},
|
| 311 |
-
{
|
| 312 |
-
"completion_length": 188.77099075317383,
|
| 313 |
-
"epoch": 0.40636042402826855,
|
| 314 |
-
"grad_norm": 0.045743752270936966,
|
| 315 |
-
"kl": 0.4034912109375,
|
| 316 |
-
"learning_rate": 1.4856507733875837e-05,
|
| 317 |
-
"loss": 0.0161,
|
| 318 |
-
"reward": 1.2922991663217545,
|
| 319 |
-
"reward_std": 0.2678883237764239,
|
| 320 |
-
"rewards/accuracy_reward": 0.3164062639698386,
|
| 321 |
-
"rewards/format_reward": 0.9758929014205933,
|
| 322 |
-
"step": 115
|
| 323 |
-
},
|
| 324 |
-
{
|
| 325 |
-
"completion_length": 196.3886251449585,
|
| 326 |
-
"epoch": 0.42402826855123676,
|
| 327 |
-
"grad_norm": 0.022538496181368828,
|
| 328 |
-
"kl": 0.26793212890625,
|
| 329 |
-
"learning_rate": 1.4306971477188223e-05,
|
| 330 |
-
"loss": 0.0107,
|
| 331 |
-
"reward": 1.3000000566244125,
|
| 332 |
-
"reward_std": 0.28592034000903366,
|
| 333 |
-
"rewards/accuracy_reward": 0.32600448057055476,
|
| 334 |
-
"rewards/format_reward": 0.9739955767989159,
|
| 335 |
-
"step": 120
|
| 336 |
-
},
|
| 337 |
-
{
|
| 338 |
-
"completion_length": 207.80346927642822,
|
| 339 |
-
"epoch": 0.4416961130742049,
|
| 340 |
-
"grad_norm": 0.02644180692732334,
|
| 341 |
-
"kl": 0.4103271484375,
|
| 342 |
-
"learning_rate": 1.3740968546047935e-05,
|
| 343 |
-
"loss": 0.0164,
|
| 344 |
-
"reward": 1.2799107655882835,
|
| 345 |
-
"reward_std": 0.30040734391659496,
|
| 346 |
-
"rewards/accuracy_reward": 0.31573662012815473,
|
| 347 |
-
"rewards/format_reward": 0.964174148440361,
|
| 348 |
-
"step": 125
|
| 349 |
-
},
|
| 350 |
-
{
|
| 351 |
-
"completion_length": 208.03773231506347,
|
| 352 |
-
"epoch": 0.45936395759717313,
|
| 353 |
-
"grad_norm": 0.03774217143654823,
|
| 354 |
-
"kl": 0.49677734375,
|
| 355 |
-
"learning_rate": 1.3160662917174045e-05,
|
| 356 |
-
"loss": 0.0199,
|
| 357 |
-
"reward": 1.297544701397419,
|
| 358 |
-
"reward_std": 0.311822210252285,
|
| 359 |
-
"rewards/accuracy_reward": 0.3420759094879031,
|
| 360 |
-
"rewards/format_reward": 0.9554687917232514,
|
| 361 |
-
"step": 130
|
| 362 |
-
},
|
| 363 |
-
{
|
| 364 |
-
"completion_length": 224.91462955474853,
|
| 365 |
-
"epoch": 0.47703180212014135,
|
| 366 |
-
"grad_norm": 0.09880122542381287,
|
| 367 |
-
"kl": 0.2965087890625,
|
| 368 |
-
"learning_rate": 1.2568273250226681e-05,
|
| 369 |
-
"loss": 0.0119,
|
| 370 |
-
"reward": 1.2757813021540643,
|
| 371 |
-
"reward_std": 0.346415382437408,
|
| 372 |
-
"rewards/accuracy_reward": 0.3440848354250193,
|
| 373 |
-
"rewards/format_reward": 0.9316964715719223,
|
| 374 |
-
"step": 135
|
| 375 |
-
},
|
| 376 |
-
{
|
| 377 |
-
"completion_length": 244.44800243377685,
|
| 378 |
-
"epoch": 0.49469964664310956,
|
| 379 |
-
"grad_norm": 0.025601347908377647,
|
| 380 |
-
"kl": 0.39385986328125,
|
| 381 |
-
"learning_rate": 1.1966064405292887e-05,
|
| 382 |
-
"loss": 0.0158,
|
| 383 |
-
"reward": 1.2787946984171867,
|
| 384 |
-
"reward_std": 0.3541004925966263,
|
| 385 |
-
"rewards/accuracy_reward": 0.34386162348091603,
|
| 386 |
-
"rewards/format_reward": 0.9349330797791481,
|
| 387 |
-
"step": 140
|
| 388 |
-
},
|
| 389 |
-
{
|
| 390 |
-
"completion_length": 239.1896318435669,
|
| 391 |
-
"epoch": 0.5123674911660777,
|
| 392 |
-
"grad_norm": 0.018517782911658287,
|
| 393 |
-
"kl": 0.194818115234375,
|
| 394 |
-
"learning_rate": 1.1356338783736256e-05,
|
| 395 |
-
"loss": 0.0078,
|
| 396 |
-
"reward": 1.34799113124609,
|
| 397 |
-
"reward_std": 0.2732340093702078,
|
| 398 |
-
"rewards/accuracy_reward": 0.373772338218987,
|
| 399 |
-
"rewards/format_reward": 0.9742187909781933,
|
| 400 |
-
"step": 145
|
| 401 |
-
},
|
| 402 |
-
{
|
| 403 |
-
"completion_length": 245.8727783203125,
|
| 404 |
-
"epoch": 0.5300353356890459,
|
| 405 |
-
"grad_norm": 0.04347226396203041,
|
| 406 |
-
"kl": 0.2514892578125,
|
| 407 |
-
"learning_rate": 1.0741427525516463e-05,
|
| 408 |
-
"loss": 0.0101,
|
| 409 |
-
"reward": 1.35837060213089,
|
| 410 |
-
"reward_std": 0.2775158016011119,
|
| 411 |
-
"rewards/accuracy_reward": 0.380691983550787,
|
| 412 |
-
"rewards/format_reward": 0.9776786111295224,
|
| 413 |
-
"step": 150
|
| 414 |
-
},
|
| 415 |
-
{
|
| 416 |
-
"completion_length": 262.62690925598145,
|
| 417 |
-
"epoch": 0.5477031802120141,
|
| 418 |
-
"grad_norm": 0.045237280428409576,
|
| 419 |
-
"kl": 0.2498046875,
|
| 420 |
-
"learning_rate": 1.012368159663363e-05,
|
| 421 |
-
"loss": 0.01,
|
| 422 |
-
"reward": 1.3261161297559738,
|
| 423 |
-
"reward_std": 0.31663199458271263,
|
| 424 |
-
"rewards/accuracy_reward": 0.3674107290804386,
|
| 425 |
-
"rewards/format_reward": 0.9587053962051868,
|
| 426 |
-
"step": 155
|
| 427 |
-
},
|
| 428 |
-
{
|
| 429 |
-
"completion_length": 294.43438892364503,
|
| 430 |
-
"epoch": 0.5653710247349824,
|
| 431 |
-
"grad_norm": 0.016969213262200356,
|
| 432 |
-
"kl": 0.36796875,
|
| 433 |
-
"learning_rate": 9.505462800772612e-06,
|
| 434 |
-
"loss": 0.0147,
|
| 435 |
-
"reward": 1.2720982685685158,
|
| 436 |
-
"reward_std": 0.3833419343456626,
|
| 437 |
-
"rewards/accuracy_reward": 0.35625001713633536,
|
| 438 |
-
"rewards/format_reward": 0.915848258882761,
|
| 439 |
-
"step": 160
|
| 440 |
-
},
|
| 441 |
-
{
|
| 442 |
-
"completion_length": 316.5702146530151,
|
| 443 |
-
"epoch": 0.5830388692579506,
|
| 444 |
-
"grad_norm": 0.033270638436079025,
|
| 445 |
-
"kl": 0.193212890625,
|
| 446 |
-
"learning_rate": 8.889134749511956e-06,
|
| 447 |
-
"loss": 0.0077,
|
| 448 |
-
"reward": 1.2236607670783997,
|
| 449 |
-
"reward_std": 0.40494529381394384,
|
| 450 |
-
"rewards/accuracy_reward": 0.3286830514669418,
|
| 451 |
-
"rewards/format_reward": 0.8949777193367481,
|
| 452 |
-
"step": 165
|
| 453 |
-
},
|
| 454 |
-
{
|
| 455 |
-
"completion_length": 311.8881841659546,
|
| 456 |
-
"epoch": 0.6007067137809188,
|
| 457 |
-
"grad_norm": 0.078031025826931,
|
| 458 |
-
"kl": 0.239520263671875,
|
| 459 |
-
"learning_rate": 8.277053825620836e-06,
|
| 460 |
-
"loss": 0.0096,
|
| 461 |
-
"reward": 1.2588170230388642,
|
| 462 |
-
"reward_std": 0.382186346501112,
|
| 463 |
-
"rewards/accuracy_reward": 0.3531250162050128,
|
| 464 |
-
"rewards/format_reward": 0.9056920081377029,
|
| 465 |
-
"step": 170
|
| 466 |
-
},
|
| 467 |
-
{
|
| 468 |
-
"completion_length": 285.3632930755615,
|
| 469 |
-
"epoch": 0.6183745583038869,
|
| 470 |
-
"grad_norm": 0.02530003897845745,
|
| 471 |
-
"kl": 0.252392578125,
|
| 472 |
-
"learning_rate": 7.671560173993588e-06,
|
| 473 |
-
"loss": 0.0101,
|
| 474 |
-
"reward": 1.30345988124609,
|
| 475 |
-
"reward_std": 0.3494835808873177,
|
| 476 |
-
"rewards/accuracy_reward": 0.37667412403970957,
|
| 477 |
-
"rewards/format_reward": 0.926785758882761,
|
| 478 |
-
"step": 175
|
| 479 |
-
},
|
| 480 |
-
{
|
| 481 |
-
"completion_length": 272.0703248977661,
|
| 482 |
-
"epoch": 0.6360424028268551,
|
| 483 |
-
"grad_norm": 0.012884252704679966,
|
| 484 |
-
"kl": 0.174432373046875,
|
| 485 |
-
"learning_rate": 7.07496875466589e-06,
|
| 486 |
-
"loss": 0.007,
|
| 487 |
-
"reward": 1.316741129755974,
|
| 488 |
-
"reward_std": 0.3391346026211977,
|
| 489 |
-
"rewards/accuracy_reward": 0.38404019717127086,
|
| 490 |
-
"rewards/format_reward": 0.9327009372413159,
|
| 491 |
-
"step": 180
|
| 492 |
-
},
|
| 493 |
-
{
|
| 494 |
-
"completion_length": 291.5253484725952,
|
| 495 |
-
"epoch": 0.6537102473498233,
|
| 496 |
-
"grad_norm": 0.023922132328152657,
|
| 497 |
-
"kl": 0.19571533203125,
|
| 498 |
-
"learning_rate": 6.489560492119225e-06,
|
| 499 |
-
"loss": 0.0078,
|
| 500 |
-
"reward": 1.2797991648316382,
|
| 501 |
-
"reward_std": 0.3811326840892434,
|
| 502 |
-
"rewards/accuracy_reward": 0.372767873480916,
|
| 503 |
-
"rewards/format_reward": 0.9070312924683094,
|
| 504 |
-
"step": 185
|
| 505 |
-
},
|
| 506 |
-
{
|
| 507 |
-
"completion_length": 290.21173191070557,
|
| 508 |
-
"epoch": 0.6713780918727915,
|
| 509 |
-
"grad_norm": 0.016599828377366066,
|
| 510 |
-
"kl": 0.208367919921875,
|
| 511 |
-
"learning_rate": 5.9175735547120975e-06,
|
| 512 |
-
"loss": 0.0083,
|
| 513 |
-
"reward": 1.2723214849829674,
|
| 514 |
-
"reward_std": 0.4036080051213503,
|
| 515 |
-
"rewards/accuracy_reward": 0.37020091004669664,
|
| 516 |
-
"rewards/format_reward": 0.9021205767989159,
|
| 517 |
-
"step": 190
|
| 518 |
-
},
|
| 519 |
-
{
|
| 520 |
-
"completion_length": 279.97177658081057,
|
| 521 |
-
"epoch": 0.6890459363957597,
|
| 522 |
-
"grad_norm": 0.013091727159917355,
|
| 523 |
-
"kl": 0.19798583984375,
|
| 524 |
-
"learning_rate": 5.361194797579108e-06,
|
| 525 |
-
"loss": 0.0079,
|
| 526 |
-
"reward": 1.2968750521540642,
|
| 527 |
-
"reward_std": 0.37299707978963853,
|
| 528 |
-
"rewards/accuracy_reward": 0.3811384104192257,
|
| 529 |
-
"rewards/format_reward": 0.9157366521656514,
|
| 530 |
-
"step": 195
|
| 531 |
-
},
|
| 532 |
-
{
|
| 533 |
-
"completion_length": 269.786619758606,
|
| 534 |
-
"epoch": 0.7067137809187279,
|
| 535 |
-
"grad_norm": 0.014325141906738281,
|
| 536 |
-
"kl": 0.288134765625,
|
| 537 |
-
"learning_rate": 4.8225514017138205e-06,
|
| 538 |
-
"loss": 0.0115,
|
| 539 |
-
"reward": 1.3161830991506576,
|
| 540 |
-
"reward_std": 0.3630652824416757,
|
| 541 |
-
"rewards/accuracy_reward": 0.3881696589291096,
|
| 542 |
-
"rewards/format_reward": 0.9280134350061416,
|
| 543 |
-
"step": 200
|
| 544 |
-
},
|
| 545 |
-
{
|
| 546 |
-
"epoch": 0.7067137809187279,
|
| 547 |
-
"eval_completion_length": 264.20691680908203,
|
| 548 |
-
"eval_kl": 0.18017578125,
|
| 549 |
-
"eval_loss": 0.006405743304640055,
|
| 550 |
-
"eval_reward": 1.3392857611179352,
|
| 551 |
-
"eval_reward_std": 0.36248352378606796,
|
| 552 |
-
"eval_rewards/accuracy_reward": 0.408482164144516,
|
| 553 |
-
"eval_rewards/format_reward": 0.9308036118745804,
|
| 554 |
-
"eval_runtime": 48.8588,
|
| 555 |
-
"eval_samples_per_second": 2.026,
|
| 556 |
-
"eval_steps_per_second": 0.02,
|
| 557 |
-
"step": 200
|
| 558 |
-
},
|
| 559 |
-
{
|
| 560 |
-
"completion_length": 276.60916366577146,
|
| 561 |
-
"epoch": 0.7243816254416962,
|
| 562 |
-
"grad_norm": 0.01561010256409645,
|
| 563 |
-
"kl": 0.180450439453125,
|
| 564 |
-
"learning_rate": 4.303702741201431e-06,
|
| 565 |
-
"loss": 0.0072,
|
| 566 |
-
"reward": 1.3011161342263222,
|
| 567 |
-
"reward_std": 0.35275506041944027,
|
| 568 |
-
"rewards/accuracy_reward": 0.37209823057055474,
|
| 569 |
-
"rewards/format_reward": 0.929017896950245,
|
| 570 |
-
"step": 205
|
| 571 |
-
},
|
| 572 |
-
{
|
| 573 |
-
"completion_length": 293.17601852416993,
|
| 574 |
-
"epoch": 0.7420494699646644,
|
| 575 |
-
"grad_norm": 0.029141413047909737,
|
| 576 |
-
"kl": 0.1777587890625,
|
| 577 |
-
"learning_rate": 3.8066325096949153e-06,
|
| 578 |
-
"loss": 0.0071,
|
| 579 |
-
"reward": 1.277678619325161,
|
| 580 |
-
"reward_std": 0.3869540646672249,
|
| 581 |
-
"rewards/accuracy_reward": 0.3700893035158515,
|
| 582 |
-
"rewards/format_reward": 0.9075893275439739,
|
| 583 |
-
"step": 210
|
| 584 |
-
},
|
| 585 |
-
{
|
| 586 |
-
"completion_length": 318.67546100616454,
|
| 587 |
-
"epoch": 0.7597173144876325,
|
| 588 |
-
"grad_norm": 0.012996065430343151,
|
| 589 |
-
"kl": 0.203765869140625,
|
| 590 |
-
"learning_rate": 3.3332411362372063e-06,
|
| 591 |
-
"loss": 0.0082,
|
| 592 |
-
"reward": 1.2314732655882836,
|
| 593 |
-
"reward_std": 0.43429264314472676,
|
| 594 |
-
"rewards/accuracy_reward": 0.3552455535158515,
|
| 595 |
-
"rewards/format_reward": 0.8762277223169803,
|
| 596 |
-
"step": 215
|
| 597 |
-
},
|
| 598 |
-
{
|
| 599 |
-
"completion_length": 307.49554920196533,
|
| 600 |
-
"epoch": 0.7773851590106007,
|
| 601 |
-
"grad_norm": 0.02377905510365963,
|
| 602 |
-
"kl": 0.28052978515625,
|
| 603 |
-
"learning_rate": 2.8853385194256677e-06,
|
| 604 |
-
"loss": 0.0112,
|
| 605 |
-
"reward": 1.2501116588711738,
|
| 606 |
-
"reward_std": 0.427118194103241,
|
| 607 |
-
"rewards/accuracy_reward": 0.36294644642621277,
|
| 608 |
-
"rewards/format_reward": 0.8871652230620384,
|
| 609 |
-
"step": 220
|
| 610 |
-
},
|
| 611 |
-
{
|
| 612 |
-
"completion_length": 303.62668800354004,
|
| 613 |
-
"epoch": 0.7950530035335689,
|
| 614 |
-
"grad_norm": 0.05803314596414566,
|
| 615 |
-
"kl": 0.244720458984375,
|
| 616 |
-
"learning_rate": 2.464637107698046e-06,
|
| 617 |
-
"loss": 0.0098,
|
| 618 |
-
"reward": 1.2681920170783996,
|
| 619 |
-
"reward_std": 0.4080910187214613,
|
| 620 |
-
"rewards/accuracy_reward": 0.36863841097801925,
|
| 621 |
-
"rewards/format_reward": 0.8995536133646965,
|
| 622 |
-
"step": 225
|
| 623 |
-
},
|
| 624 |
-
{
|
| 625 |
-
"completion_length": 286.2454355239868,
|
| 626 |
-
"epoch": 0.8127208480565371,
|
| 627 |
-
"grad_norm": 0.01800759695470333,
|
| 628 |
-
"kl": 0.18123779296875,
|
| 629 |
-
"learning_rate": 2.072745352195794e-06,
|
| 630 |
-
"loss": 0.0072,
|
| 631 |
-
"reward": 1.2926339864730836,
|
| 632 |
-
"reward_std": 0.3678985072299838,
|
| 633 |
-
"rewards/accuracy_reward": 0.3722098364494741,
|
| 634 |
-
"rewards/format_reward": 0.9204241506755352,
|
| 635 |
-
"step": 230
|
| 636 |
-
},
|
| 637 |
-
{
|
| 638 |
-
"completion_length": 283.0959957122803,
|
| 639 |
-
"epoch": 0.8303886925795053,
|
| 640 |
-
"grad_norm": 0.015781087800860405,
|
| 641 |
-
"kl": 0.1795166015625,
|
| 642 |
-
"learning_rate": 1.7111615572361628e-06,
|
| 643 |
-
"loss": 0.0072,
|
| 644 |
-
"reward": 1.2987723737955092,
|
| 645 |
-
"reward_std": 0.37512697763741015,
|
| 646 |
-
"rewards/accuracy_reward": 0.37700894437730315,
|
| 647 |
-
"rewards/format_reward": 0.9217634335160255,
|
| 648 |
-
"step": 235
|
| 649 |
-
},
|
| 650 |
-
{
|
| 651 |
-
"completion_length": 298.8237842559814,
|
| 652 |
-
"epoch": 0.8480565371024735,
|
| 653 |
-
"grad_norm": 0.024129491299390793,
|
| 654 |
-
"kl": 0.21593017578125,
|
| 655 |
-
"learning_rate": 1.381268151904298e-06,
|
| 656 |
-
"loss": 0.0086,
|
| 657 |
-
"reward": 1.2565848737955094,
|
| 658 |
-
"reward_std": 0.4109664471819997,
|
| 659 |
-
"rewards/accuracy_reward": 0.36138394568115473,
|
| 660 |
-
"rewards/format_reward": 0.8952009342610836,
|
| 661 |
-
"step": 240
|
| 662 |
-
},
|
| 663 |
-
{
|
| 664 |
-
"completion_length": 312.69075107574463,
|
| 665 |
-
"epoch": 0.8657243816254417,
|
| 666 |
-
"grad_norm": 0.021995313465595245,
|
| 667 |
-
"kl": 0.28687744140625,
|
| 668 |
-
"learning_rate": 1.0843264046665558e-06,
|
| 669 |
-
"loss": 0.0115,
|
| 670 |
-
"reward": 1.2415179193019867,
|
| 671 |
-
"reward_std": 0.4253259660676122,
|
| 672 |
-
"rewards/accuracy_reward": 0.35926340874284507,
|
| 673 |
-
"rewards/format_reward": 0.8822545059025287,
|
| 674 |
-
"step": 245
|
| 675 |
-
},
|
| 676 |
-
{
|
| 677 |
-
"completion_length": 307.51753559112547,
|
| 678 |
-
"epoch": 0.8833922261484098,
|
| 679 |
-
"grad_norm": 0.021934768185019493,
|
| 680 |
-
"kl": 0.287030029296875,
|
| 681 |
-
"learning_rate": 8.214716012124491e-07,
|
| 682 |
-
"loss": 0.0115,
|
| 683 |
-
"reward": 1.2411830931901933,
|
| 684 |
-
"reward_std": 0.41262274123728276,
|
| 685 |
-
"rewards/accuracy_reward": 0.34944198131561277,
|
| 686 |
-
"rewards/format_reward": 0.8917411133646965,
|
| 687 |
-
"step": 250
|
| 688 |
-
},
|
| 689 |
-
{
|
| 690 |
-
"completion_length": 304.0524700164795,
|
| 691 |
-
"epoch": 0.901060070671378,
|
| 692 |
-
"grad_norm": 0.03234613686800003,
|
| 693 |
-
"kl": 0.20001220703125,
|
| 694 |
-
"learning_rate": 5.937087039615619e-07,
|
| 695 |
-
"loss": 0.008,
|
| 696 |
-
"reward": 1.2452009439468383,
|
| 697 |
-
"reward_std": 0.4156901784241199,
|
| 698 |
-
"rewards/accuracy_reward": 0.3550223371013999,
|
| 699 |
-
"rewards/format_reward": 0.8901786126196385,
|
| 700 |
-
"step": 255
|
| 701 |
-
},
|
| 702 |
-
{
|
| 703 |
-
"completion_length": 312.4107292175293,
|
| 704 |
-
"epoch": 0.9187279151943463,
|
| 705 |
-
"grad_norm": 0.011318758130073547,
|
| 706 |
-
"kl": 0.22039794921875,
|
| 707 |
-
"learning_rate": 4.019085098303077e-07,
|
| 708 |
-
"loss": 0.0088,
|
| 709 |
-
"reward": 1.239843799173832,
|
| 710 |
-
"reward_std": 0.43206911720335484,
|
| 711 |
-
"rewards/accuracy_reward": 0.3598214453086257,
|
| 712 |
-
"rewards/format_reward": 0.8800223603844642,
|
| 713 |
-
"step": 260
|
| 714 |
-
},
|
| 715 |
-
{
|
| 716 |
-
"completion_length": 319.8676481246948,
|
| 717 |
-
"epoch": 0.9363957597173145,
|
| 718 |
-
"grad_norm": 0.02429085038602352,
|
| 719 |
-
"kl": 0.1993408203125,
|
| 720 |
-
"learning_rate": 2.4680432094837394e-07,
|
| 721 |
-
"loss": 0.008,
|
| 722 |
-
"reward": 1.21506702080369,
|
| 723 |
-
"reward_std": 0.45882711336016657,
|
| 724 |
-
"rewards/accuracy_reward": 0.34754465958103536,
|
| 725 |
-
"rewards/format_reward": 0.8675223633646965,
|
| 726 |
-
"step": 265
|
| 727 |
-
},
|
| 728 |
-
{
|
| 729 |
-
"completion_length": 319.6962184906006,
|
| 730 |
-
"epoch": 0.9540636042402827,
|
| 731 |
-
"grad_norm": 0.014112049713730812,
|
| 732 |
-
"kl": 0.24310302734375,
|
| 733 |
-
"learning_rate": 1.289891410535593e-07,
|
| 734 |
-
"loss": 0.0097,
|
| 735 |
-
"reward": 1.2215402342379094,
|
| 736 |
-
"reward_std": 0.4551096182316542,
|
| 737 |
-
"rewards/accuracy_reward": 0.3492187652736902,
|
| 738 |
-
"rewards/format_reward": 0.872321467846632,
|
| 739 |
-
"step": 270
|
| 740 |
-
},
|
| 741 |
-
{
|
| 742 |
-
"completion_length": 329.8996807098389,
|
| 743 |
-
"epoch": 0.9717314487632509,
|
| 744 |
-
"grad_norm": 0.013921594247221947,
|
| 745 |
-
"kl": 0.2304443359375,
|
| 746 |
-
"learning_rate": 4.8913408283934874e-08,
|
| 747 |
-
"loss": 0.0092,
|
| 748 |
-
"reward": 1.200892908871174,
|
| 749 |
-
"reward_std": 0.4598929390311241,
|
| 750 |
-
"rewards/accuracy_reward": 0.34587055183947085,
|
| 751 |
-
"rewards/format_reward": 0.8550223641097545,
|
| 752 |
-
"step": 275
|
| 753 |
-
},
|
| 754 |
-
{
|
| 755 |
-
"completion_length": 315.45526084899905,
|
| 756 |
-
"epoch": 0.9893992932862191,
|
| 757 |
-
"grad_norm": 0.013098319061100483,
|
| 758 |
-
"kl": 0.19427490234375,
|
| 759 |
-
"learning_rate": 6.883273035447335e-09,
|
| 760 |
-
"loss": 0.0078,
|
| 761 |
-
"reward": 1.2271205872297286,
|
| 762 |
-
"reward_std": 0.4515534173697233,
|
| 763 |
-
"rewards/accuracy_reward": 0.35345983896404504,
|
| 764 |
-
"rewards/format_reward": 0.8736607514321804,
|
| 765 |
-
"step": 280
|
| 766 |
-
},
|
| 767 |
-
{
|
| 768 |
-
"completion_length": 325.00113519032794,
|
| 769 |
-
"epoch": 1.0,
|
| 770 |
-
"kl": 0.18748982747395834,
|
| 771 |
-
"reward": 1.2046131504078705,
|
| 772 |
-
"reward_std": 0.4611472859978676,
|
| 773 |
-
"rewards/accuracy_reward": 0.3446800730501612,
|
| 774 |
-
"rewards/format_reward": 0.8599330745637417,
|
| 775 |
-
"step": 283,
|
| 776 |
"total_flos": 0.0,
|
| 777 |
-
"train_loss":
|
| 778 |
-
"train_runtime":
|
| 779 |
-
"train_samples_per_second": 1.
|
| 780 |
"train_steps_per_second": 0.007
|
| 781 |
}
|
| 782 |
],
|
| 783 |
"logging_steps": 5,
|
| 784 |
-
"max_steps":
|
| 785 |
"num_input_tokens_seen": 0,
|
| 786 |
"num_train_epochs": 1,
|
| 787 |
"save_steps": 500,
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.10202898550724637,
|
| 5 |
"eval_steps": 100,
|
| 6 |
+
"global_step": 33,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"completion_length": 303.61742544174194,
|
| 13 |
+
"epoch": 0.015458937198067632,
|
| 14 |
+
"grad_norm": 0.02257479541003704,
|
| 15 |
+
"kl": 0.9676864624023438,
|
| 16 |
+
"learning_rate": 1.9941379571543597e-05,
|
| 17 |
+
"loss": 0.0386,
|
| 18 |
+
"reward": 0.7926339689642191,
|
| 19 |
+
"reward_std": 0.3637773351743817,
|
| 20 |
+
"rewards/accuracy_reward": 0.11194196966243908,
|
| 21 |
+
"rewards/format_reward": 0.6806919939815999,
|
| 22 |
"step": 5
|
| 23 |
},
|
| 24 |
{
|
| 25 |
+
"completion_length": 193.77601299285888,
|
| 26 |
+
"epoch": 0.030917874396135265,
|
| 27 |
+
"grad_norm": 0.02067318558692932,
|
| 28 |
+
"kl": 0.1245941162109375,
|
| 29 |
+
"learning_rate": 1.796093065705644e-05,
|
| 30 |
+
"loss": 0.005,
|
| 31 |
+
"reward": 1.0407366551458836,
|
| 32 |
+
"reward_std": 0.2885524293407798,
|
| 33 |
+
"rewards/accuracy_reward": 0.11953125573927537,
|
| 34 |
+
"rewards/format_reward": 0.9212053991854191,
|
| 35 |
"step": 10
|
| 36 |
},
|
| 37 |
{
|
| 38 |
+
"completion_length": 223.56373710632323,
|
| 39 |
+
"epoch": 0.0463768115942029,
|
| 40 |
+
"grad_norm": 0.014934813603758812,
|
| 41 |
+
"kl": 0.16761474609375,
|
| 42 |
+
"learning_rate": 1.3701381553399147e-05,
|
| 43 |
+
"loss": 0.0067,
|
| 44 |
+
"reward": 1.0725446939468384,
|
| 45 |
+
"reward_std": 0.3380734449252486,
|
| 46 |
+
"rewards/accuracy_reward": 0.16651786545990035,
|
| 47 |
+
"rewards/format_reward": 0.906026828289032,
|
| 48 |
"step": 15
|
| 49 |
},
|
| 50 |
{
|
| 51 |
+
"completion_length": 189.5631784439087,
|
| 52 |
+
"epoch": 0.06183574879227053,
|
| 53 |
+
"grad_norm": 0.015098211355507374,
|
| 54 |
+
"kl": 0.175286865234375,
|
| 55 |
+
"learning_rate": 8.382180034472353e-06,
|
| 56 |
+
"loss": 0.007,
|
| 57 |
+
"reward": 1.1564732655882835,
|
| 58 |
+
"reward_std": 0.2815748773515224,
|
| 59 |
+
"rewards/accuracy_reward": 0.19162947330623864,
|
| 60 |
+
"rewards/format_reward": 0.964843787997961,
|
| 61 |
"step": 20
|
| 62 |
},
|
| 63 |
{
|
| 64 |
+
"completion_length": 190.79632415771485,
|
| 65 |
+
"epoch": 0.07729468599033816,
|
| 66 |
+
"grad_norm": 0.013767687603831291,
|
| 67 |
+
"kl": 0.16260986328125,
|
| 68 |
+
"learning_rate": 3.5261371521817247e-06,
|
| 69 |
+
"loss": 0.0065,
|
| 70 |
+
"reward": 1.1904018431901933,
|
| 71 |
+
"reward_std": 0.306734830327332,
|
| 72 |
+
"rewards/accuracy_reward": 0.22790179681032896,
|
| 73 |
+
"rewards/format_reward": 0.9625000387430191,
|
| 74 |
"step": 25
|
| 75 |
},
|
| 76 |
{
|
| 77 |
+
"completion_length": 203.4788038253784,
|
| 78 |
+
"epoch": 0.0927536231884058,
|
| 79 |
+
"grad_norm": 0.014600388705730438,
|
| 80 |
+
"kl": 0.143023681640625,
|
| 81 |
+
"learning_rate": 5.234682881719766e-07,
|
| 82 |
+
"loss": 0.0057,
|
| 83 |
+
"reward": 1.206808091700077,
|
| 84 |
+
"reward_std": 0.2881101544946432,
|
| 85 |
+
"rewards/accuracy_reward": 0.23984376154839993,
|
| 86 |
+
"rewards/format_reward": 0.9669643275439739,
|
| 87 |
"step": 30
|
| 88 |
},
|
| 89 |
{
|
| 90 |
+
"completion_length": 215.08613300323486,
|
| 91 |
+
"epoch": 0.10202898550724637,
|
| 92 |
+
"kl": 0.13948567708333334,
|
| 93 |
+
"reward": 1.2220982710520427,
|
| 94 |
+
"reward_std": 0.3094604279225071,
|
| 95 |
+
"rewards/accuracy_reward": 0.252976200543344,
|
| 96 |
+
"rewards/format_reward": 0.9691220708191395,
|
| 97 |
+
"step": 33,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 98 |
"total_flos": 0.0,
|
| 99 |
+
"train_loss": 0.011041131547906181,
|
| 100 |
+
"train_runtime": 4812.1769,
|
| 101 |
+
"train_samples_per_second": 1.505,
|
| 102 |
"train_steps_per_second": 0.007
|
| 103 |
}
|
| 104 |
],
|
| 105 |
"logging_steps": 5,
|
| 106 |
+
"max_steps": 33,
|
| 107 |
"num_input_tokens_seen": 0,
|
| 108 |
"num_train_epochs": 1,
|
| 109 |
"save_steps": 500,
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 7480
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9ddd9b822d423f5edd5932b081042f0d6a6fb9c8892b3e03c09aeadb53fda817
|
| 3 |
size 7480
|