Model save
Browse files- README.md +7 -7
- all_results.json +7 -6
- config.json +2 -1
- generation_config.json +1 -1
- model-00001-of-00003.safetensors +1 -1
- model-00002-of-00003.safetensors +1 -1
- model-00003-of-00003.safetensors +1 -1
- tokenizer.json +1 -0
- tokenizer_config.json +2 -0
- train_results.json +7 -6
- trainer_state.json +1788 -458
- training_args.bin +2 -2
README.md
CHANGED
|
@@ -35,14 +35,14 @@ More information needed
|
|
| 35 |
|
| 36 |
The following hyperparameters were used during training:
|
| 37 |
- learning_rate: 5e-07
|
| 38 |
-
- train_batch_size:
|
| 39 |
-
- eval_batch_size:
|
| 40 |
-
- seed:
|
| 41 |
- distributed_type: multi-GPU
|
| 42 |
- num_devices: 8
|
| 43 |
-
- gradient_accumulation_steps:
|
| 44 |
- total_train_batch_size: 128
|
| 45 |
-
- total_eval_batch_size:
|
| 46 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 47 |
- lr_scheduler_type: cosine
|
| 48 |
- lr_scheduler_warmup_ratio: 0.1
|
|
@@ -54,7 +54,7 @@ The following hyperparameters were used during training:
|
|
| 54 |
|
| 55 |
### Framework versions
|
| 56 |
|
| 57 |
-
- Transformers 4.
|
| 58 |
- Pytorch 2.1.2+cu121
|
| 59 |
- Datasets 2.14.6
|
| 60 |
-
- Tokenizers 0.
|
|
|
|
| 35 |
|
| 36 |
The following hyperparameters were used during training:
|
| 37 |
- learning_rate: 5e-07
|
| 38 |
+
- train_batch_size: 1
|
| 39 |
+
- eval_batch_size: 1
|
| 40 |
+
- seed: 42
|
| 41 |
- distributed_type: multi-GPU
|
| 42 |
- num_devices: 8
|
| 43 |
+
- gradient_accumulation_steps: 16
|
| 44 |
- total_train_batch_size: 128
|
| 45 |
+
- total_eval_batch_size: 8
|
| 46 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 47 |
- lr_scheduler_type: cosine
|
| 48 |
- lr_scheduler_warmup_ratio: 0.1
|
|
|
|
| 54 |
|
| 55 |
### Framework versions
|
| 56 |
|
| 57 |
+
- Transformers 4.41.1
|
| 58 |
- Pytorch 2.1.2+cu121
|
| 59 |
- Datasets 2.14.6
|
| 60 |
+
- Tokenizers 0.19.1
|
all_results.json
CHANGED
|
@@ -1,8 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"
|
| 4 |
-
"
|
| 5 |
-
"
|
| 6 |
-
"
|
| 7 |
-
"
|
|
|
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 0.9998009950248756,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.56497666932595,
|
| 5 |
+
"train_runtime": 62670.8183,
|
| 6 |
+
"train_samples": 160800,
|
| 7 |
+
"train_samples_per_second": 2.566,
|
| 8 |
+
"train_steps_per_second": 0.02
|
| 9 |
}
|
config.json
CHANGED
|
@@ -3,6 +3,7 @@
|
|
| 3 |
"architectures": [
|
| 4 |
"MistralForCausalLM"
|
| 5 |
],
|
|
|
|
| 6 |
"bos_token_id": 1,
|
| 7 |
"eos_token_id": 2,
|
| 8 |
"hidden_act": "silu",
|
|
@@ -19,7 +20,7 @@
|
|
| 19 |
"sliding_window": 4096,
|
| 20 |
"tie_word_embeddings": false,
|
| 21 |
"torch_dtype": "bfloat16",
|
| 22 |
-
"transformers_version": "4.
|
| 23 |
"use_cache": false,
|
| 24 |
"vocab_size": 32000
|
| 25 |
}
|
|
|
|
| 3 |
"architectures": [
|
| 4 |
"MistralForCausalLM"
|
| 5 |
],
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
"bos_token_id": 1,
|
| 8 |
"eos_token_id": 2,
|
| 9 |
"hidden_act": "silu",
|
|
|
|
| 20 |
"sliding_window": 4096,
|
| 21 |
"tie_word_embeddings": false,
|
| 22 |
"torch_dtype": "bfloat16",
|
| 23 |
+
"transformers_version": "4.41.1",
|
| 24 |
"use_cache": false,
|
| 25 |
"vocab_size": 32000
|
| 26 |
}
|
generation_config.json
CHANGED
|
@@ -2,5 +2,5 @@
|
|
| 2 |
"_from_model_config": true,
|
| 3 |
"bos_token_id": 1,
|
| 4 |
"eos_token_id": 2,
|
| 5 |
-
"transformers_version": "4.
|
| 6 |
}
|
|
|
|
| 2 |
"_from_model_config": true,
|
| 3 |
"bos_token_id": 1,
|
| 4 |
"eos_token_id": 2,
|
| 5 |
+
"transformers_version": "4.41.1"
|
| 6 |
}
|
model-00001-of-00003.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4943162336
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4e11dcd61cc10207ffa429232e22f4cd1209b869b363230b1b23c8229fec53ac
|
| 3 |
size 4943162336
|
model-00002-of-00003.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4999819336
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:182675acf06d7733583a038c80e4ca834644ab7b1ee0b07cac182d9ff6d6bfce
|
| 3 |
size 4999819336
|
model-00003-of-00003.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4540516344
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b33509881c7f28a3653c4f33545c0462135ed2ffe7316c0e4ac0f29f06a94027
|
| 3 |
size 4540516344
|
tokenizer.json
CHANGED
|
@@ -134,6 +134,7 @@
|
|
| 134 |
"end_of_word_suffix": null,
|
| 135 |
"fuse_unk": true,
|
| 136 |
"byte_fallback": true,
|
|
|
|
| 137 |
"vocab": {
|
| 138 |
"<unk>": 0,
|
| 139 |
"<s>": 1,
|
|
|
|
| 134 |
"end_of_word_suffix": null,
|
| 135 |
"fuse_unk": true,
|
| 136 |
"byte_fallback": true,
|
| 137 |
+
"ignore_merges": false,
|
| 138 |
"vocab": {
|
| 139 |
"<unk>": 0,
|
| 140 |
"<s>": 1,
|
tokenizer_config.json
CHANGED
|
@@ -1,4 +1,6 @@
|
|
| 1 |
{
|
|
|
|
|
|
|
| 2 |
"added_tokens_decoder": {
|
| 3 |
"0": {
|
| 4 |
"content": "<unk>",
|
|
|
|
| 1 |
{
|
| 2 |
+
"add_bos_token": true,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
"added_tokens_decoder": {
|
| 5 |
"0": {
|
| 6 |
"content": "<unk>",
|
train_results.json
CHANGED
|
@@ -1,8 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"
|
| 4 |
-
"
|
| 5 |
-
"
|
| 6 |
-
"
|
| 7 |
-
"
|
|
|
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 0.9998009950248756,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.56497666932595,
|
| 5 |
+
"train_runtime": 62670.8183,
|
| 6 |
+
"train_samples": 160800,
|
| 7 |
+
"train_samples_per_second": 2.566,
|
| 8 |
+
"train_steps_per_second": 0.02
|
| 9 |
}
|
trainer_state.json
CHANGED
|
@@ -1,21 +1,22 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
-
"eval_steps":
|
| 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 |
-
"
|
| 14 |
-
"
|
| 15 |
-
"logits/
|
| 16 |
-
"
|
| 17 |
-
"logps/
|
| 18 |
-
"
|
|
|
|
| 19 |
"rewards/accuracies": 0.0,
|
| 20 |
"rewards/chosen": 0.0,
|
| 21 |
"rewards/margins": 0.0,
|
|
@@ -23,580 +24,1909 @@
|
|
| 23 |
"step": 1
|
| 24 |
},
|
| 25 |
{
|
| 26 |
-
"epoch": 0.
|
| 27 |
-
"
|
| 28 |
-
"
|
| 29 |
-
"logits/
|
| 30 |
-
"
|
| 31 |
-
"logps/
|
| 32 |
-
"
|
| 33 |
-
"
|
| 34 |
-
"rewards/
|
| 35 |
-
"rewards/
|
| 36 |
-
"rewards/
|
|
|
|
| 37 |
"step": 10
|
| 38 |
},
|
| 39 |
{
|
| 40 |
-
"epoch": 0.
|
| 41 |
-
"
|
| 42 |
-
"
|
| 43 |
-
"logits/
|
| 44 |
-
"
|
| 45 |
-
"logps/
|
| 46 |
-
"
|
| 47 |
-
"
|
| 48 |
-
"rewards/
|
| 49 |
-
"rewards/
|
| 50 |
-
"rewards/
|
|
|
|
| 51 |
"step": 20
|
| 52 |
},
|
| 53 |
{
|
| 54 |
-
"epoch": 0.
|
| 55 |
-
"
|
| 56 |
-
"
|
| 57 |
-
"logits/
|
| 58 |
-
"
|
| 59 |
-
"logps/
|
| 60 |
-
"
|
| 61 |
-
"
|
| 62 |
-
"rewards/
|
| 63 |
-
"rewards/
|
| 64 |
-
"rewards/
|
|
|
|
| 65 |
"step": 30
|
| 66 |
},
|
| 67 |
{
|
| 68 |
-
"epoch": 0.
|
| 69 |
-
"
|
| 70 |
-
"
|
| 71 |
-
"logits/
|
| 72 |
-
"
|
| 73 |
-
"logps/
|
| 74 |
-
"
|
| 75 |
-
"
|
| 76 |
-
"rewards/
|
| 77 |
-
"rewards/
|
| 78 |
-
"rewards/
|
|
|
|
| 79 |
"step": 40
|
| 80 |
},
|
| 81 |
{
|
| 82 |
-
"epoch": 0.
|
| 83 |
-
"
|
| 84 |
-
"
|
| 85 |
-
"logits/
|
| 86 |
-
"
|
| 87 |
-
"logps/
|
| 88 |
-
"
|
| 89 |
-
"
|
| 90 |
-
"rewards/
|
| 91 |
-
"rewards/
|
| 92 |
-
"rewards/
|
|
|
|
| 93 |
"step": 50
|
| 94 |
},
|
| 95 |
{
|
| 96 |
-
"epoch": 0.
|
| 97 |
-
"
|
| 98 |
-
"
|
| 99 |
-
"logits/
|
| 100 |
-
"
|
| 101 |
-
"logps/
|
| 102 |
-
"
|
| 103 |
-
"
|
| 104 |
-
"rewards/
|
| 105 |
-
"rewards/
|
| 106 |
-
"rewards/
|
|
|
|
| 107 |
"step": 60
|
| 108 |
},
|
| 109 |
{
|
| 110 |
-
"epoch": 0.
|
| 111 |
-
"
|
| 112 |
-
"
|
| 113 |
-
"logits/
|
| 114 |
-
"
|
| 115 |
-
"logps/
|
| 116 |
-
"
|
| 117 |
-
"
|
| 118 |
-
"rewards/
|
| 119 |
-
"rewards/
|
| 120 |
-
"rewards/
|
|
|
|
| 121 |
"step": 70
|
| 122 |
},
|
| 123 |
{
|
| 124 |
-
"epoch": 0.
|
| 125 |
-
"
|
| 126 |
-
"
|
| 127 |
-
"logits/
|
| 128 |
-
"
|
| 129 |
-
"logps/
|
| 130 |
-
"
|
| 131 |
-
"
|
| 132 |
-
"rewards/
|
| 133 |
-
"rewards/
|
| 134 |
-
"rewards/
|
|
|
|
| 135 |
"step": 80
|
| 136 |
},
|
| 137 |
{
|
| 138 |
-
"epoch": 0.
|
| 139 |
-
"
|
| 140 |
-
"
|
| 141 |
-
"logits/
|
| 142 |
-
"
|
| 143 |
-
"logps/
|
| 144 |
-
"
|
| 145 |
-
"
|
| 146 |
-
"rewards/
|
| 147 |
-
"rewards/
|
| 148 |
-
"rewards/
|
|
|
|
| 149 |
"step": 90
|
| 150 |
},
|
| 151 |
{
|
| 152 |
-
"epoch": 0.
|
| 153 |
-
"
|
| 154 |
-
"
|
| 155 |
-
"logits/
|
| 156 |
-
"
|
| 157 |
-
"logps/
|
| 158 |
-
"
|
| 159 |
-
"
|
| 160 |
-
"rewards/
|
| 161 |
-
"rewards/
|
| 162 |
-
"rewards/
|
|
|
|
| 163 |
"step": 100
|
| 164 |
},
|
| 165 |
{
|
| 166 |
-
"epoch": 0.
|
| 167 |
-
"
|
| 168 |
-
"
|
| 169 |
-
"logits/
|
| 170 |
-
"
|
| 171 |
-
"logps/
|
| 172 |
-
"
|
| 173 |
-
"
|
| 174 |
-
"rewards/
|
| 175 |
-
"rewards/
|
| 176 |
-
"rewards/
|
|
|
|
| 177 |
"step": 110
|
| 178 |
},
|
| 179 |
{
|
| 180 |
-
"epoch": 0.
|
| 181 |
-
"
|
| 182 |
-
"
|
| 183 |
-
"logits/
|
| 184 |
-
"
|
| 185 |
-
"logps/
|
| 186 |
-
"
|
| 187 |
-
"
|
| 188 |
-
"rewards/
|
| 189 |
-
"rewards/
|
| 190 |
-
"rewards/
|
|
|
|
| 191 |
"step": 120
|
| 192 |
},
|
| 193 |
{
|
| 194 |
-
"epoch": 0.
|
| 195 |
-
"
|
| 196 |
-
"
|
| 197 |
-
"logits/
|
| 198 |
-
"
|
| 199 |
-
"logps/
|
| 200 |
-
"
|
| 201 |
-
"
|
| 202 |
-
"rewards/
|
| 203 |
-
"rewards/
|
| 204 |
-
"rewards/
|
|
|
|
| 205 |
"step": 130
|
| 206 |
},
|
| 207 |
{
|
| 208 |
-
"epoch": 0.
|
| 209 |
-
"
|
| 210 |
-
"
|
| 211 |
-
"logits/
|
| 212 |
-
"
|
| 213 |
-
"logps/
|
| 214 |
-
"
|
| 215 |
-
"
|
| 216 |
-
"rewards/
|
| 217 |
-
"rewards/
|
| 218 |
-
"rewards/
|
|
|
|
| 219 |
"step": 140
|
| 220 |
},
|
| 221 |
{
|
| 222 |
-
"epoch": 0.
|
| 223 |
-
"
|
| 224 |
-
"
|
| 225 |
-
"logits/
|
| 226 |
-
"
|
| 227 |
-
"logps/
|
| 228 |
-
"
|
| 229 |
-
"
|
| 230 |
-
"rewards/
|
| 231 |
-
"rewards/
|
| 232 |
-
"rewards/
|
|
|
|
| 233 |
"step": 150
|
| 234 |
},
|
| 235 |
{
|
| 236 |
-
"epoch": 0.
|
| 237 |
-
"
|
| 238 |
-
"
|
| 239 |
-
"logits/
|
| 240 |
-
"
|
| 241 |
-
"logps/
|
| 242 |
-
"
|
| 243 |
-
"
|
| 244 |
-
"rewards/
|
| 245 |
-
"rewards/
|
| 246 |
-
"rewards/
|
|
|
|
| 247 |
"step": 160
|
| 248 |
},
|
| 249 |
{
|
| 250 |
-
"epoch": 0.
|
| 251 |
-
"
|
| 252 |
-
"
|
| 253 |
-
"logits/
|
| 254 |
-
"
|
| 255 |
-
"logps/
|
| 256 |
-
"
|
| 257 |
-
"
|
| 258 |
-
"rewards/
|
| 259 |
-
"rewards/
|
| 260 |
-
"rewards/
|
|
|
|
| 261 |
"step": 170
|
| 262 |
},
|
| 263 |
{
|
| 264 |
-
"epoch": 0.
|
| 265 |
-
"
|
| 266 |
-
"
|
| 267 |
-
"logits/
|
| 268 |
-
"
|
| 269 |
-
"logps/
|
| 270 |
-
"
|
| 271 |
-
"
|
| 272 |
-
"rewards/
|
| 273 |
-
"rewards/
|
| 274 |
-
"rewards/
|
|
|
|
| 275 |
"step": 180
|
| 276 |
},
|
| 277 |
{
|
| 278 |
-
"epoch": 0.
|
| 279 |
-
"
|
| 280 |
-
"
|
| 281 |
-
"logits/
|
| 282 |
-
"
|
| 283 |
-
"logps/
|
| 284 |
-
"
|
| 285 |
-
"
|
| 286 |
-
"rewards/
|
| 287 |
-
"rewards/
|
| 288 |
-
"rewards/
|
|
|
|
| 289 |
"step": 190
|
| 290 |
},
|
| 291 |
{
|
| 292 |
-
"epoch": 0.
|
| 293 |
-
"
|
| 294 |
-
"
|
| 295 |
-
"logits/
|
| 296 |
-
"
|
| 297 |
-
"logps/
|
| 298 |
-
"
|
| 299 |
-
"
|
| 300 |
-
"rewards/
|
| 301 |
-
"rewards/
|
| 302 |
-
"rewards/
|
|
|
|
| 303 |
"step": 200
|
| 304 |
},
|
| 305 |
{
|
| 306 |
-
"epoch": 0.
|
| 307 |
-
"
|
| 308 |
-
"
|
| 309 |
-
"logits/
|
| 310 |
-
"
|
| 311 |
-
"logps/
|
| 312 |
-
"
|
| 313 |
-
"
|
| 314 |
-
"rewards/
|
| 315 |
-
"rewards/
|
| 316 |
-
"rewards/
|
|
|
|
| 317 |
"step": 210
|
| 318 |
},
|
| 319 |
{
|
| 320 |
-
"epoch": 0.
|
| 321 |
-
"
|
| 322 |
-
"
|
| 323 |
-
"logits/
|
| 324 |
-
"
|
| 325 |
-
"logps/
|
| 326 |
-
"
|
| 327 |
-
"
|
| 328 |
-
"rewards/
|
| 329 |
-
"rewards/
|
| 330 |
-
"rewards/
|
|
|
|
| 331 |
"step": 220
|
| 332 |
},
|
| 333 |
{
|
| 334 |
-
"epoch": 0.
|
| 335 |
-
"
|
| 336 |
-
"
|
| 337 |
-
"logits/
|
| 338 |
-
"
|
| 339 |
-
"logps/
|
| 340 |
-
"
|
| 341 |
-
"
|
| 342 |
-
"rewards/
|
| 343 |
-
"rewards/
|
| 344 |
-
"rewards/
|
|
|
|
| 345 |
"step": 230
|
| 346 |
},
|
| 347 |
{
|
| 348 |
-
"epoch": 0.
|
| 349 |
-
"
|
| 350 |
-
"
|
| 351 |
-
"logits/
|
| 352 |
-
"
|
| 353 |
-
"logps/
|
| 354 |
-
"
|
| 355 |
-
"
|
| 356 |
-
"rewards/
|
| 357 |
-
"rewards/
|
| 358 |
-
"rewards/
|
|
|
|
| 359 |
"step": 240
|
| 360 |
},
|
| 361 |
{
|
| 362 |
-
"epoch": 0.
|
| 363 |
-
"
|
| 364 |
-
"
|
| 365 |
-
"logits/
|
| 366 |
-
"
|
| 367 |
-
"logps/
|
| 368 |
-
"
|
| 369 |
-
"
|
| 370 |
-
"rewards/
|
| 371 |
-
"rewards/
|
| 372 |
-
"rewards/
|
|
|
|
| 373 |
"step": 250
|
| 374 |
},
|
| 375 |
{
|
| 376 |
-
"epoch": 0.
|
| 377 |
-
"
|
| 378 |
-
"
|
| 379 |
-
"logits/
|
| 380 |
-
"
|
| 381 |
-
"logps/
|
| 382 |
-
"
|
| 383 |
-
"
|
| 384 |
-
"rewards/
|
| 385 |
-
"rewards/
|
| 386 |
-
"rewards/
|
|
|
|
| 387 |
"step": 260
|
| 388 |
},
|
| 389 |
{
|
| 390 |
-
"epoch": 0.
|
| 391 |
-
"
|
| 392 |
-
"
|
| 393 |
-
"logits/
|
| 394 |
-
"
|
| 395 |
-
"logps/
|
| 396 |
-
"
|
| 397 |
-
"
|
| 398 |
-
"rewards/
|
| 399 |
-
"rewards/
|
| 400 |
-
"rewards/
|
|
|
|
| 401 |
"step": 270
|
| 402 |
},
|
| 403 |
{
|
| 404 |
-
"epoch": 0.
|
| 405 |
-
"
|
| 406 |
-
"
|
| 407 |
-
"logits/
|
| 408 |
-
"
|
| 409 |
-
"logps/
|
| 410 |
-
"
|
| 411 |
-
"
|
| 412 |
-
"rewards/
|
| 413 |
-
"rewards/
|
| 414 |
-
"rewards/
|
|
|
|
| 415 |
"step": 280
|
| 416 |
},
|
| 417 |
{
|
| 418 |
-
"epoch": 0.
|
| 419 |
-
"
|
| 420 |
-
"
|
| 421 |
-
"logits/
|
| 422 |
-
"
|
| 423 |
-
"logps/
|
| 424 |
-
"
|
| 425 |
-
"
|
| 426 |
-
"rewards/
|
| 427 |
-
"rewards/
|
| 428 |
-
"rewards/
|
|
|
|
| 429 |
"step": 290
|
| 430 |
},
|
| 431 |
{
|
| 432 |
-
"epoch": 0.
|
| 433 |
-
"
|
| 434 |
-
"
|
| 435 |
-
"logits/
|
| 436 |
-
"
|
| 437 |
-
"logps/
|
| 438 |
-
"
|
| 439 |
-
"
|
| 440 |
-
"rewards/
|
| 441 |
-
"rewards/
|
| 442 |
-
"rewards/
|
|
|
|
| 443 |
"step": 300
|
| 444 |
},
|
| 445 |
{
|
| 446 |
-
"epoch": 0.
|
| 447 |
-
"
|
| 448 |
-
"
|
| 449 |
-
"logits/
|
| 450 |
-
"
|
| 451 |
-
"logps/
|
| 452 |
-
"
|
| 453 |
-
"
|
| 454 |
-
"rewards/
|
| 455 |
-
"rewards/
|
| 456 |
-
"rewards/
|
|
|
|
| 457 |
"step": 310
|
| 458 |
},
|
| 459 |
{
|
| 460 |
-
"epoch": 0.
|
| 461 |
-
"
|
| 462 |
-
"
|
| 463 |
-
"logits/
|
| 464 |
-
"
|
| 465 |
-
"logps/
|
| 466 |
-
"
|
| 467 |
-
"
|
| 468 |
-
"rewards/
|
| 469 |
-
"rewards/
|
| 470 |
-
"rewards/
|
|
|
|
| 471 |
"step": 320
|
| 472 |
},
|
| 473 |
{
|
| 474 |
-
"epoch": 0.
|
| 475 |
-
"
|
| 476 |
-
"
|
| 477 |
-
"logits/
|
| 478 |
-
"
|
| 479 |
-
"logps/
|
| 480 |
-
"
|
| 481 |
-
"
|
| 482 |
-
"rewards/
|
| 483 |
-
"rewards/
|
| 484 |
-
"rewards/
|
|
|
|
| 485 |
"step": 330
|
| 486 |
},
|
| 487 |
{
|
| 488 |
-
"epoch": 0.
|
| 489 |
-
"
|
| 490 |
-
"
|
| 491 |
-
"logits/
|
| 492 |
-
"
|
| 493 |
-
"logps/
|
| 494 |
-
"
|
| 495 |
-
"
|
| 496 |
-
"rewards/
|
| 497 |
-
"rewards/
|
| 498 |
-
"rewards/
|
|
|
|
| 499 |
"step": 340
|
| 500 |
},
|
| 501 |
{
|
| 502 |
-
"epoch": 0.
|
| 503 |
-
"
|
| 504 |
-
"
|
| 505 |
-
"logits/
|
| 506 |
-
"
|
| 507 |
-
"logps/
|
| 508 |
-
"
|
| 509 |
-
"
|
| 510 |
-
"rewards/
|
| 511 |
-
"rewards/
|
| 512 |
-
"rewards/
|
|
|
|
| 513 |
"step": 350
|
| 514 |
},
|
| 515 |
{
|
| 516 |
-
"epoch": 0.
|
| 517 |
-
"
|
| 518 |
-
"
|
| 519 |
-
"logits/
|
| 520 |
-
"
|
| 521 |
-
"logps/
|
| 522 |
-
"
|
| 523 |
-
"
|
| 524 |
-
"rewards/
|
| 525 |
-
"rewards/
|
| 526 |
-
"rewards/
|
|
|
|
| 527 |
"step": 360
|
| 528 |
},
|
| 529 |
{
|
| 530 |
-
"epoch": 0.
|
| 531 |
-
"
|
| 532 |
-
"
|
| 533 |
-
"logits/
|
| 534 |
-
"
|
| 535 |
-
"logps/
|
| 536 |
-
"
|
| 537 |
-
"
|
| 538 |
-
"rewards/
|
| 539 |
-
"rewards/
|
| 540 |
-
"rewards/
|
|
|
|
| 541 |
"step": 370
|
| 542 |
},
|
| 543 |
{
|
| 544 |
-
"epoch": 0.
|
| 545 |
-
"
|
| 546 |
-
"
|
| 547 |
-
"logits/
|
| 548 |
-
"
|
| 549 |
-
"logps/
|
| 550 |
-
"
|
| 551 |
-
"
|
| 552 |
-
"rewards/
|
| 553 |
-
"rewards/
|
| 554 |
-
"rewards/
|
|
|
|
| 555 |
"step": 380
|
| 556 |
},
|
| 557 |
{
|
| 558 |
-
"epoch": 0.
|
| 559 |
-
"
|
| 560 |
-
"
|
| 561 |
-
"logits/
|
| 562 |
-
"
|
| 563 |
-
"logps/
|
| 564 |
-
"
|
| 565 |
-
"
|
| 566 |
-
"rewards/
|
| 567 |
-
"rewards/
|
| 568 |
-
"rewards/
|
|
|
|
| 569 |
"step": 390
|
| 570 |
},
|
| 571 |
{
|
| 572 |
-
"epoch": 0.
|
| 573 |
-
"
|
| 574 |
-
"
|
| 575 |
-
"logits/
|
| 576 |
-
"
|
| 577 |
-
"logps/
|
| 578 |
-
"
|
| 579 |
-
"
|
| 580 |
-
"rewards/
|
| 581 |
-
"rewards/
|
| 582 |
-
"rewards/
|
|
|
|
| 583 |
"step": 400
|
| 584 |
},
|
| 585 |
{
|
| 586 |
-
"epoch":
|
| 587 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 588 |
"total_flos": 0.0,
|
| 589 |
-
"train_loss": 0.
|
| 590 |
-
"train_runtime":
|
| 591 |
-
"train_samples_per_second":
|
| 592 |
-
"train_steps_per_second": 0.
|
| 593 |
}
|
| 594 |
],
|
| 595 |
"logging_steps": 10,
|
| 596 |
-
"max_steps":
|
|
|
|
| 597 |
"num_train_epochs": 1,
|
| 598 |
-
"save_steps":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 599 |
"total_flos": 0.0,
|
|
|
|
| 600 |
"trial_name": null,
|
| 601 |
"trial_params": null
|
| 602 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9998009950248756,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 1256,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.0007960199004975124,
|
| 13 |
+
"grad_norm": 4.987965979355318,
|
| 14 |
+
"learning_rate": 3.968253968253968e-09,
|
| 15 |
+
"logits/chosen": -2.866555690765381,
|
| 16 |
+
"logits/rejected": -2.8678386211395264,
|
| 17 |
+
"logps/chosen": -150.79409790039062,
|
| 18 |
+
"logps/rejected": -167.60751342773438,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
"rewards/accuracies": 0.0,
|
| 21 |
"rewards/chosen": 0.0,
|
| 22 |
"rewards/margins": 0.0,
|
|
|
|
| 24 |
"step": 1
|
| 25 |
},
|
| 26 |
{
|
| 27 |
+
"epoch": 0.007960199004975124,
|
| 28 |
+
"grad_norm": 5.158661804258714,
|
| 29 |
+
"learning_rate": 3.968253968253968e-08,
|
| 30 |
+
"logits/chosen": -2.905927896499634,
|
| 31 |
+
"logits/rejected": -2.877537250518799,
|
| 32 |
+
"logps/chosen": -169.23851013183594,
|
| 33 |
+
"logps/rejected": -156.21502685546875,
|
| 34 |
+
"loss": 0.6932,
|
| 35 |
+
"rewards/accuracies": 0.4444444477558136,
|
| 36 |
+
"rewards/chosen": 0.00022146817354951054,
|
| 37 |
+
"rewards/margins": 7.194941281341016e-05,
|
| 38 |
+
"rewards/rejected": 0.00014951876073610038,
|
| 39 |
"step": 10
|
| 40 |
},
|
| 41 |
{
|
| 42 |
+
"epoch": 0.015920398009950248,
|
| 43 |
+
"grad_norm": 4.941389073277953,
|
| 44 |
+
"learning_rate": 7.936507936507936e-08,
|
| 45 |
+
"logits/chosen": -2.8461222648620605,
|
| 46 |
+
"logits/rejected": -2.8314127922058105,
|
| 47 |
+
"logps/chosen": -143.2448272705078,
|
| 48 |
+
"logps/rejected": -139.14114379882812,
|
| 49 |
+
"loss": 0.6932,
|
| 50 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 51 |
+
"rewards/chosen": -3.118843596894294e-05,
|
| 52 |
+
"rewards/margins": -0.00014748479588888586,
|
| 53 |
+
"rewards/rejected": 0.00011629634536802769,
|
| 54 |
"step": 20
|
| 55 |
},
|
| 56 |
{
|
| 57 |
+
"epoch": 0.023880597014925373,
|
| 58 |
+
"grad_norm": 4.789407789219409,
|
| 59 |
+
"learning_rate": 1.1904761904761903e-07,
|
| 60 |
+
"logits/chosen": -2.8516290187835693,
|
| 61 |
+
"logits/rejected": -2.8340375423431396,
|
| 62 |
+
"logps/chosen": -133.89068603515625,
|
| 63 |
+
"logps/rejected": -137.05783081054688,
|
| 64 |
+
"loss": 0.6931,
|
| 65 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 66 |
+
"rewards/chosen": 0.00035750039387494326,
|
| 67 |
+
"rewards/margins": 0.00034277202212251723,
|
| 68 |
+
"rewards/rejected": 1.4728342648595572e-05,
|
| 69 |
"step": 30
|
| 70 |
},
|
| 71 |
{
|
| 72 |
+
"epoch": 0.031840796019900496,
|
| 73 |
+
"grad_norm": 5.0684575947852935,
|
| 74 |
+
"learning_rate": 1.5873015873015872e-07,
|
| 75 |
+
"logits/chosen": -2.8869121074676514,
|
| 76 |
+
"logits/rejected": -2.8650405406951904,
|
| 77 |
+
"logps/chosen": -149.87741088867188,
|
| 78 |
+
"logps/rejected": -154.5808563232422,
|
| 79 |
+
"loss": 0.6928,
|
| 80 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 81 |
+
"rewards/chosen": 0.0008229534141719341,
|
| 82 |
+
"rewards/margins": 0.0007549519650638103,
|
| 83 |
+
"rewards/rejected": 6.800144910812378e-05,
|
| 84 |
"step": 40
|
| 85 |
},
|
| 86 |
{
|
| 87 |
+
"epoch": 0.03980099502487562,
|
| 88 |
+
"grad_norm": 4.886701576208831,
|
| 89 |
+
"learning_rate": 1.984126984126984e-07,
|
| 90 |
+
"logits/chosen": -2.8915724754333496,
|
| 91 |
+
"logits/rejected": -2.8401710987091064,
|
| 92 |
+
"logps/chosen": -153.5601348876953,
|
| 93 |
+
"logps/rejected": -127.82198333740234,
|
| 94 |
+
"loss": 0.6923,
|
| 95 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 96 |
+
"rewards/chosen": -0.00028051590197719634,
|
| 97 |
+
"rewards/margins": 0.0019810814410448074,
|
| 98 |
+
"rewards/rejected": -0.002261597430333495,
|
| 99 |
"step": 50
|
| 100 |
},
|
| 101 |
{
|
| 102 |
+
"epoch": 0.04776119402985075,
|
| 103 |
+
"grad_norm": 4.80539942483475,
|
| 104 |
+
"learning_rate": 2.3809523809523806e-07,
|
| 105 |
+
"logits/chosen": -2.8685171604156494,
|
| 106 |
+
"logits/rejected": -2.885859966278076,
|
| 107 |
+
"logps/chosen": -138.50997924804688,
|
| 108 |
+
"logps/rejected": -153.02711486816406,
|
| 109 |
+
"loss": 0.6913,
|
| 110 |
+
"rewards/accuracies": 0.625,
|
| 111 |
+
"rewards/chosen": 0.0011361455544829369,
|
| 112 |
+
"rewards/margins": 0.003960339818149805,
|
| 113 |
+
"rewards/rejected": -0.002824194496497512,
|
| 114 |
"step": 60
|
| 115 |
},
|
| 116 |
{
|
| 117 |
+
"epoch": 0.05572139303482587,
|
| 118 |
+
"grad_norm": 4.993634749933866,
|
| 119 |
+
"learning_rate": 2.7777777777777776e-07,
|
| 120 |
+
"logits/chosen": -2.8937714099884033,
|
| 121 |
+
"logits/rejected": -2.8720123767852783,
|
| 122 |
+
"logps/chosen": -140.58717346191406,
|
| 123 |
+
"logps/rejected": -159.0962677001953,
|
| 124 |
+
"loss": 0.6893,
|
| 125 |
+
"rewards/accuracies": 0.612500011920929,
|
| 126 |
+
"rewards/chosen": -0.006692352704703808,
|
| 127 |
+
"rewards/margins": 0.008710912428796291,
|
| 128 |
+
"rewards/rejected": -0.015403266064822674,
|
| 129 |
"step": 70
|
| 130 |
},
|
| 131 |
{
|
| 132 |
+
"epoch": 0.06368159203980099,
|
| 133 |
+
"grad_norm": 5.018829367878226,
|
| 134 |
+
"learning_rate": 3.1746031746031743e-07,
|
| 135 |
+
"logits/chosen": -2.8606536388397217,
|
| 136 |
+
"logits/rejected": -2.8511064052581787,
|
| 137 |
+
"logps/chosen": -143.28701782226562,
|
| 138 |
+
"logps/rejected": -143.73817443847656,
|
| 139 |
+
"loss": 0.6878,
|
| 140 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 141 |
+
"rewards/chosen": -0.019417399540543556,
|
| 142 |
+
"rewards/margins": 0.015563729219138622,
|
| 143 |
+
"rewards/rejected": -0.034981124103069305,
|
| 144 |
"step": 80
|
| 145 |
},
|
| 146 |
{
|
| 147 |
+
"epoch": 0.07164179104477612,
|
| 148 |
+
"grad_norm": 5.3232578633563445,
|
| 149 |
+
"learning_rate": 3.5714285714285716e-07,
|
| 150 |
+
"logits/chosen": -2.8441901206970215,
|
| 151 |
+
"logits/rejected": -2.841014862060547,
|
| 152 |
+
"logps/chosen": -154.00717163085938,
|
| 153 |
+
"logps/rejected": -169.87318420410156,
|
| 154 |
+
"loss": 0.683,
|
| 155 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 156 |
+
"rewards/chosen": -0.03319092467427254,
|
| 157 |
+
"rewards/margins": 0.028958678245544434,
|
| 158 |
+
"rewards/rejected": -0.06214960291981697,
|
| 159 |
"step": 90
|
| 160 |
},
|
| 161 |
{
|
| 162 |
+
"epoch": 0.07960199004975124,
|
| 163 |
+
"grad_norm": 5.9751016265015515,
|
| 164 |
+
"learning_rate": 3.968253968253968e-07,
|
| 165 |
+
"logits/chosen": -2.8674731254577637,
|
| 166 |
+
"logits/rejected": -2.819610834121704,
|
| 167 |
+
"logps/chosen": -168.8125,
|
| 168 |
+
"logps/rejected": -154.88980102539062,
|
| 169 |
+
"loss": 0.6789,
|
| 170 |
+
"rewards/accuracies": 0.606249988079071,
|
| 171 |
+
"rewards/chosen": -0.09129344671964645,
|
| 172 |
+
"rewards/margins": 0.039778389036655426,
|
| 173 |
+
"rewards/rejected": -0.13107183575630188,
|
| 174 |
"step": 100
|
| 175 |
},
|
| 176 |
{
|
| 177 |
+
"epoch": 0.08756218905472637,
|
| 178 |
+
"grad_norm": 7.043583746287595,
|
| 179 |
+
"learning_rate": 4.365079365079365e-07,
|
| 180 |
+
"logits/chosen": -2.8453006744384766,
|
| 181 |
+
"logits/rejected": -2.8008811473846436,
|
| 182 |
+
"logps/chosen": -166.57568359375,
|
| 183 |
+
"logps/rejected": -160.43548583984375,
|
| 184 |
+
"loss": 0.6753,
|
| 185 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 186 |
+
"rewards/chosen": -0.1897437423467636,
|
| 187 |
+
"rewards/margins": 0.05553770065307617,
|
| 188 |
+
"rewards/rejected": -0.2452814280986786,
|
| 189 |
"step": 110
|
| 190 |
},
|
| 191 |
{
|
| 192 |
+
"epoch": 0.0955223880597015,
|
| 193 |
+
"grad_norm": 8.93236047182562,
|
| 194 |
+
"learning_rate": 4.761904761904761e-07,
|
| 195 |
+
"logits/chosen": -2.861156940460205,
|
| 196 |
+
"logits/rejected": -2.8308346271514893,
|
| 197 |
+
"logps/chosen": -169.21116638183594,
|
| 198 |
+
"logps/rejected": -174.7342071533203,
|
| 199 |
+
"loss": 0.6668,
|
| 200 |
+
"rewards/accuracies": 0.637499988079071,
|
| 201 |
+
"rewards/chosen": -0.2347549945116043,
|
| 202 |
+
"rewards/margins": 0.07360492646694183,
|
| 203 |
+
"rewards/rejected": -0.30835992097854614,
|
| 204 |
"step": 120
|
| 205 |
},
|
| 206 |
{
|
| 207 |
+
"epoch": 0.10348258706467661,
|
| 208 |
+
"grad_norm": 12.11276981143796,
|
| 209 |
+
"learning_rate": 4.999845414634076e-07,
|
| 210 |
+
"logits/chosen": -2.8432881832122803,
|
| 211 |
+
"logits/rejected": -2.8032021522521973,
|
| 212 |
+
"logps/chosen": -176.11865234375,
|
| 213 |
+
"logps/rejected": -188.36691284179688,
|
| 214 |
+
"loss": 0.644,
|
| 215 |
+
"rewards/accuracies": 0.71875,
|
| 216 |
+
"rewards/chosen": -0.360163152217865,
|
| 217 |
+
"rewards/margins": 0.17286832630634308,
|
| 218 |
+
"rewards/rejected": -0.5330314636230469,
|
| 219 |
"step": 130
|
| 220 |
},
|
| 221 |
{
|
| 222 |
+
"epoch": 0.11144278606965174,
|
| 223 |
+
"grad_norm": 14.823561517874804,
|
| 224 |
+
"learning_rate": 4.998106548810311e-07,
|
| 225 |
+
"logits/chosen": -2.8333194255828857,
|
| 226 |
+
"logits/rejected": -2.786397695541382,
|
| 227 |
+
"logps/chosen": -193.32913208007812,
|
| 228 |
+
"logps/rejected": -193.43296813964844,
|
| 229 |
+
"loss": 0.639,
|
| 230 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 231 |
+
"rewards/chosen": -0.42469319701194763,
|
| 232 |
+
"rewards/margins": 0.16398563981056213,
|
| 233 |
+
"rewards/rejected": -0.5886788964271545,
|
| 234 |
"step": 140
|
| 235 |
},
|
| 236 |
{
|
| 237 |
+
"epoch": 0.11940298507462686,
|
| 238 |
+
"grad_norm": 18.490476889353154,
|
| 239 |
+
"learning_rate": 4.994436933879359e-07,
|
| 240 |
+
"logits/chosen": -2.742354393005371,
|
| 241 |
+
"logits/rejected": -2.6894731521606445,
|
| 242 |
+
"logps/chosen": -206.05099487304688,
|
| 243 |
+
"logps/rejected": -226.1792449951172,
|
| 244 |
+
"loss": 0.6134,
|
| 245 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 246 |
+
"rewards/chosen": -0.5303625464439392,
|
| 247 |
+
"rewards/margins": 0.2593725323677063,
|
| 248 |
+
"rewards/rejected": -0.7897351384162903,
|
| 249 |
"step": 150
|
| 250 |
},
|
| 251 |
{
|
| 252 |
+
"epoch": 0.12736318407960198,
|
| 253 |
+
"grad_norm": 16.59996040063754,
|
| 254 |
+
"learning_rate": 4.988839406031596e-07,
|
| 255 |
+
"logits/chosen": -2.64296555519104,
|
| 256 |
+
"logits/rejected": -2.6306753158569336,
|
| 257 |
+
"logps/chosen": -180.86181640625,
|
| 258 |
+
"logps/rejected": -229.1710968017578,
|
| 259 |
+
"loss": 0.6168,
|
| 260 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 261 |
+
"rewards/chosen": -0.510300874710083,
|
| 262 |
+
"rewards/margins": 0.35416826605796814,
|
| 263 |
+
"rewards/rejected": -0.8644691705703735,
|
| 264 |
"step": 160
|
| 265 |
},
|
| 266 |
{
|
| 267 |
+
"epoch": 0.13532338308457711,
|
| 268 |
+
"grad_norm": 12.427120919609093,
|
| 269 |
+
"learning_rate": 4.981318291512395e-07,
|
| 270 |
+
"logits/chosen": -2.585261821746826,
|
| 271 |
+
"logits/rejected": -2.602097988128662,
|
| 272 |
+
"logps/chosen": -192.22706604003906,
|
| 273 |
+
"logps/rejected": -224.95028686523438,
|
| 274 |
+
"loss": 0.6289,
|
| 275 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 276 |
+
"rewards/chosen": -0.5405030250549316,
|
| 277 |
+
"rewards/margins": 0.21325179934501648,
|
| 278 |
+
"rewards/rejected": -0.7537548542022705,
|
| 279 |
"step": 170
|
| 280 |
},
|
| 281 |
{
|
| 282 |
+
"epoch": 0.14328358208955225,
|
| 283 |
+
"grad_norm": 15.900669280547154,
|
| 284 |
+
"learning_rate": 4.971879403278432e-07,
|
| 285 |
+
"logits/chosen": -2.5604166984558105,
|
| 286 |
+
"logits/rejected": -2.5091071128845215,
|
| 287 |
+
"logps/chosen": -214.48095703125,
|
| 288 |
+
"logps/rejected": -222.5020751953125,
|
| 289 |
+
"loss": 0.6208,
|
| 290 |
+
"rewards/accuracies": 0.625,
|
| 291 |
+
"rewards/chosen": -0.6959569454193115,
|
| 292 |
+
"rewards/margins": 0.1984298974275589,
|
| 293 |
+
"rewards/rejected": -0.894386887550354,
|
| 294 |
"step": 180
|
| 295 |
},
|
| 296 |
{
|
| 297 |
+
"epoch": 0.15124378109452735,
|
| 298 |
+
"grad_norm": 21.336526790653657,
|
| 299 |
+
"learning_rate": 4.960530036504941e-07,
|
| 300 |
+
"logits/chosen": -2.511948823928833,
|
| 301 |
+
"logits/rejected": -2.4782981872558594,
|
| 302 |
+
"logps/chosen": -247.1105499267578,
|
| 303 |
+
"logps/rejected": -259.957763671875,
|
| 304 |
+
"loss": 0.6225,
|
| 305 |
+
"rewards/accuracies": 0.59375,
|
| 306 |
+
"rewards/chosen": -0.9646499752998352,
|
| 307 |
+
"rewards/margins": 0.24105295538902283,
|
| 308 |
+
"rewards/rejected": -1.2057029008865356,
|
| 309 |
"step": 190
|
| 310 |
},
|
| 311 |
{
|
| 312 |
+
"epoch": 0.15920398009950248,
|
| 313 |
+
"grad_norm": 14.558771925056739,
|
| 314 |
+
"learning_rate": 4.947278962947386e-07,
|
| 315 |
+
"logits/chosen": -2.4865562915802,
|
| 316 |
+
"logits/rejected": -2.4496474266052246,
|
| 317 |
+
"logps/chosen": -247.4215850830078,
|
| 318 |
+
"logps/rejected": -294.28839111328125,
|
| 319 |
+
"loss": 0.5863,
|
| 320 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 321 |
+
"rewards/chosen": -1.000279188156128,
|
| 322 |
+
"rewards/margins": 0.5100765228271484,
|
| 323 |
+
"rewards/rejected": -1.5103557109832764,
|
| 324 |
"step": 200
|
| 325 |
},
|
| 326 |
{
|
| 327 |
+
"epoch": 0.16716417910447762,
|
| 328 |
+
"grad_norm": 18.09768702444502,
|
| 329 |
+
"learning_rate": 4.932136424161899e-07,
|
| 330 |
+
"logits/chosen": -2.524082899093628,
|
| 331 |
+
"logits/rejected": -2.460231304168701,
|
| 332 |
+
"logps/chosen": -236.9945526123047,
|
| 333 |
+
"logps/rejected": -280.2134704589844,
|
| 334 |
+
"loss": 0.5892,
|
| 335 |
+
"rewards/accuracies": 0.6875,
|
| 336 |
+
"rewards/chosen": -0.9642118215560913,
|
| 337 |
+
"rewards/margins": 0.4870384633541107,
|
| 338 |
+
"rewards/rejected": -1.4512503147125244,
|
| 339 |
"step": 210
|
| 340 |
},
|
| 341 |
{
|
| 342 |
+
"epoch": 0.17512437810945275,
|
| 343 |
+
"grad_norm": 15.582712307283431,
|
| 344 |
+
"learning_rate": 4.915114123589732e-07,
|
| 345 |
+
"logits/chosen": -2.4720115661621094,
|
| 346 |
+
"logits/rejected": -2.4532742500305176,
|
| 347 |
+
"logps/chosen": -277.3113708496094,
|
| 348 |
+
"logps/rejected": -346.95941162109375,
|
| 349 |
+
"loss": 0.6028,
|
| 350 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 351 |
+
"rewards/chosen": -1.3337619304656982,
|
| 352 |
+
"rewards/margins": 0.5726526379585266,
|
| 353 |
+
"rewards/rejected": -1.9064146280288696,
|
| 354 |
"step": 220
|
| 355 |
},
|
| 356 |
{
|
| 357 |
+
"epoch": 0.18308457711442785,
|
| 358 |
+
"grad_norm": 23.55648961804369,
|
| 359 |
+
"learning_rate": 4.896225217511849e-07,
|
| 360 |
+
"logits/chosen": -2.4816207885742188,
|
| 361 |
+
"logits/rejected": -2.427955389022827,
|
| 362 |
+
"logps/chosen": -294.205810546875,
|
| 363 |
+
"logps/rejected": -344.0794372558594,
|
| 364 |
+
"loss": 0.5888,
|
| 365 |
+
"rewards/accuracies": 0.675000011920929,
|
| 366 |
+
"rewards/chosen": -1.3327975273132324,
|
| 367 |
+
"rewards/margins": 0.5613763332366943,
|
| 368 |
+
"rewards/rejected": -1.8941738605499268,
|
| 369 |
"step": 230
|
| 370 |
},
|
| 371 |
{
|
| 372 |
+
"epoch": 0.191044776119403,
|
| 373 |
+
"grad_norm": 13.953969214382605,
|
| 374 |
+
"learning_rate": 4.875484304880629e-07,
|
| 375 |
+
"logits/chosen": -2.4548521041870117,
|
| 376 |
+
"logits/rejected": -2.370633602142334,
|
| 377 |
+
"logps/chosen": -271.3977355957031,
|
| 378 |
+
"logps/rejected": -332.35137939453125,
|
| 379 |
+
"loss": 0.5803,
|
| 380 |
+
"rewards/accuracies": 0.737500011920929,
|
| 381 |
+
"rewards/chosen": -1.2848864793777466,
|
| 382 |
+
"rewards/margins": 0.5692149996757507,
|
| 383 |
+
"rewards/rejected": -1.854101538658142,
|
| 384 |
"step": 240
|
| 385 |
},
|
| 386 |
{
|
| 387 |
+
"epoch": 0.19900497512437812,
|
| 388 |
+
"grad_norm": 18.518974677225064,
|
| 389 |
+
"learning_rate": 4.852907416036558e-07,
|
| 390 |
+
"logits/chosen": -2.3934974670410156,
|
| 391 |
+
"logits/rejected": -2.312051296234131,
|
| 392 |
+
"logps/chosen": -282.7490234375,
|
| 393 |
+
"logps/rejected": -325.5188293457031,
|
| 394 |
+
"loss": 0.5885,
|
| 395 |
+
"rewards/accuracies": 0.706250011920929,
|
| 396 |
+
"rewards/chosen": -1.3080084323883057,
|
| 397 |
+
"rewards/margins": 0.444119393825531,
|
| 398 |
+
"rewards/rejected": -1.7521278858184814,
|
| 399 |
"step": 250
|
| 400 |
},
|
| 401 |
{
|
| 402 |
+
"epoch": 0.20696517412935322,
|
| 403 |
+
"grad_norm": 16.421871283301197,
|
| 404 |
+
"learning_rate": 4.828512000318616e-07,
|
| 405 |
+
"logits/chosen": -1.9985812902450562,
|
| 406 |
+
"logits/rejected": -1.7946439981460571,
|
| 407 |
+
"logps/chosen": -351.16864013671875,
|
| 408 |
+
"logps/rejected": -392.8179931640625,
|
| 409 |
+
"loss": 0.5685,
|
| 410 |
+
"rewards/accuracies": 0.65625,
|
| 411 |
+
"rewards/chosen": -2.019517660140991,
|
| 412 |
+
"rewards/margins": 0.4194945693016052,
|
| 413 |
+
"rewards/rejected": -2.439012050628662,
|
| 414 |
"step": 260
|
| 415 |
},
|
| 416 |
{
|
| 417 |
+
"epoch": 0.21492537313432836,
|
| 418 |
+
"grad_norm": 14.628264234076205,
|
| 419 |
+
"learning_rate": 4.802316912577946e-07,
|
| 420 |
+
"logits/chosen": -2.0842502117156982,
|
| 421 |
+
"logits/rejected": -1.9886341094970703,
|
| 422 |
+
"logps/chosen": -346.19921875,
|
| 423 |
+
"logps/rejected": -366.7377624511719,
|
| 424 |
+
"loss": 0.5884,
|
| 425 |
+
"rewards/accuracies": 0.637499988079071,
|
| 426 |
+
"rewards/chosen": -2.0177090167999268,
|
| 427 |
+
"rewards/margins": 0.32493501901626587,
|
| 428 |
+
"rewards/rejected": -2.342643976211548,
|
| 429 |
"step": 270
|
| 430 |
},
|
| 431 |
{
|
| 432 |
+
"epoch": 0.2228855721393035,
|
| 433 |
+
"grad_norm": 18.526048621847632,
|
| 434 |
+
"learning_rate": 4.774342398605221e-07,
|
| 435 |
+
"logits/chosen": -2.015749216079712,
|
| 436 |
+
"logits/rejected": -1.818052053451538,
|
| 437 |
+
"logps/chosen": -360.9788513183594,
|
| 438 |
+
"logps/rejected": -402.9951171875,
|
| 439 |
+
"loss": 0.5688,
|
| 440 |
+
"rewards/accuracies": 0.675000011920929,
|
| 441 |
+
"rewards/chosen": -2.1356658935546875,
|
| 442 |
+
"rewards/margins": 0.5189321637153625,
|
| 443 |
+
"rewards/rejected": -2.654597759246826,
|
| 444 |
"step": 280
|
| 445 |
},
|
| 446 |
{
|
| 447 |
+
"epoch": 0.2308457711442786,
|
| 448 |
+
"grad_norm": 18.9693122793002,
|
| 449 |
+
"learning_rate": 4.744610079482978e-07,
|
| 450 |
+
"logits/chosen": -2.1531434059143066,
|
| 451 |
+
"logits/rejected": -1.9195207357406616,
|
| 452 |
+
"logps/chosen": -355.23626708984375,
|
| 453 |
+
"logps/rejected": -396.5318603515625,
|
| 454 |
+
"loss": 0.596,
|
| 455 |
+
"rewards/accuracies": 0.637499988079071,
|
| 456 |
+
"rewards/chosen": -2.032478094100952,
|
| 457 |
+
"rewards/margins": 0.41044849157333374,
|
| 458 |
+
"rewards/rejected": -2.4429266452789307,
|
| 459 |
"step": 290
|
| 460 |
},
|
| 461 |
{
|
| 462 |
+
"epoch": 0.23880597014925373,
|
| 463 |
+
"grad_norm": 14.860387982197683,
|
| 464 |
+
"learning_rate": 4.713142934875005e-07,
|
| 465 |
+
"logits/chosen": -2.4048914909362793,
|
| 466 |
+
"logits/rejected": -2.098559856414795,
|
| 467 |
+
"logps/chosen": -306.0177917480469,
|
| 468 |
+
"logps/rejected": -335.77099609375,
|
| 469 |
+
"loss": 0.582,
|
| 470 |
+
"rewards/accuracies": 0.699999988079071,
|
| 471 |
+
"rewards/chosen": -1.319157361984253,
|
| 472 |
+
"rewards/margins": 0.5222987532615662,
|
| 473 |
+
"rewards/rejected": -1.841456413269043,
|
| 474 |
"step": 300
|
| 475 |
},
|
| 476 |
{
|
| 477 |
+
"epoch": 0.24676616915422886,
|
| 478 |
+
"grad_norm": 17.94094060671055,
|
| 479 |
+
"learning_rate": 4.679965285265706e-07,
|
| 480 |
+
"logits/chosen": -2.0477898120880127,
|
| 481 |
+
"logits/rejected": -1.9392799139022827,
|
| 482 |
+
"logps/chosen": -276.44622802734375,
|
| 483 |
+
"logps/rejected": -320.6493835449219,
|
| 484 |
+
"loss": 0.5616,
|
| 485 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 486 |
+
"rewards/chosen": -1.4027130603790283,
|
| 487 |
+
"rewards/margins": 0.39033347368240356,
|
| 488 |
+
"rewards/rejected": -1.7930463552474976,
|
| 489 |
"step": 310
|
| 490 |
},
|
| 491 |
{
|
| 492 |
+
"epoch": 0.25472636815920396,
|
| 493 |
+
"grad_norm": 20.377060113211176,
|
| 494 |
+
"learning_rate": 4.64510277316316e-07,
|
| 495 |
+
"logits/chosen": -1.7731729745864868,
|
| 496 |
+
"logits/rejected": -1.263430118560791,
|
| 497 |
+
"logps/chosen": -373.1016845703125,
|
| 498 |
+
"logps/rejected": -431.9881896972656,
|
| 499 |
+
"loss": 0.5443,
|
| 500 |
+
"rewards/accuracies": 0.78125,
|
| 501 |
+
"rewards/chosen": -2.0920746326446533,
|
| 502 |
+
"rewards/margins": 0.7201870083808899,
|
| 503 |
+
"rewards/rejected": -2.8122615814208984,
|
| 504 |
"step": 320
|
| 505 |
},
|
| 506 |
{
|
| 507 |
+
"epoch": 0.2626865671641791,
|
| 508 |
+
"grad_norm": 19.309940699402812,
|
| 509 |
+
"learning_rate": 4.6085823432804137e-07,
|
| 510 |
+
"logits/chosen": -1.9024795293807983,
|
| 511 |
+
"logits/rejected": -1.569469928741455,
|
| 512 |
+
"logps/chosen": -355.3124084472656,
|
| 513 |
+
"logps/rejected": -419.6775817871094,
|
| 514 |
+
"loss": 0.5793,
|
| 515 |
+
"rewards/accuracies": 0.699999988079071,
|
| 516 |
+
"rewards/chosen": -1.87880539894104,
|
| 517 |
+
"rewards/margins": 0.6133967638015747,
|
| 518 |
+
"rewards/rejected": -2.4922022819519043,
|
| 519 |
"step": 330
|
| 520 |
},
|
| 521 |
{
|
| 522 |
+
"epoch": 0.27064676616915423,
|
| 523 |
+
"grad_norm": 17.446855339300225,
|
| 524 |
+
"learning_rate": 4.570432221710314e-07,
|
| 525 |
+
"logits/chosen": -2.1071066856384277,
|
| 526 |
+
"logits/rejected": -2.065721273422241,
|
| 527 |
+
"logps/chosen": -280.3807373046875,
|
| 528 |
+
"logps/rejected": -328.475830078125,
|
| 529 |
+
"loss": 0.5779,
|
| 530 |
+
"rewards/accuracies": 0.65625,
|
| 531 |
+
"rewards/chosen": -1.4469754695892334,
|
| 532 |
+
"rewards/margins": 0.44690531492233276,
|
| 533 |
+
"rewards/rejected": -1.893880844116211,
|
| 534 |
"step": 340
|
| 535 |
},
|
| 536 |
{
|
| 537 |
+
"epoch": 0.27860696517412936,
|
| 538 |
+
"grad_norm": 15.474382457829194,
|
| 539 |
+
"learning_rate": 4.5306818941099866e-07,
|
| 540 |
+
"logits/chosen": -2.2009084224700928,
|
| 541 |
+
"logits/rejected": -2.050603151321411,
|
| 542 |
+
"logps/chosen": -296.90423583984375,
|
| 543 |
+
"logps/rejected": -336.07818603515625,
|
| 544 |
+
"loss": 0.597,
|
| 545 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 546 |
+
"rewards/chosen": -1.5415115356445312,
|
| 547 |
+
"rewards/margins": 0.42944687604904175,
|
| 548 |
+
"rewards/rejected": -1.9709584712982178,
|
| 549 |
"step": 350
|
| 550 |
},
|
| 551 |
{
|
| 552 |
+
"epoch": 0.2865671641791045,
|
| 553 |
+
"grad_norm": 16.92609001492546,
|
| 554 |
+
"learning_rate": 4.4893620829118124e-07,
|
| 555 |
+
"logits/chosen": -2.1950016021728516,
|
| 556 |
+
"logits/rejected": -2.0361106395721436,
|
| 557 |
+
"logps/chosen": -296.43865966796875,
|
| 558 |
+
"logps/rejected": -364.69696044921875,
|
| 559 |
+
"loss": 0.5517,
|
| 560 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 561 |
+
"rewards/chosen": -1.6347688436508179,
|
| 562 |
+
"rewards/margins": 0.5980359315872192,
|
| 563 |
+
"rewards/rejected": -2.232804775238037,
|
| 564 |
"step": 360
|
| 565 |
},
|
| 566 |
{
|
| 567 |
+
"epoch": 0.2945273631840796,
|
| 568 |
+
"grad_norm": 19.211553848388508,
|
| 569 |
+
"learning_rate": 4.4465047235785185e-07,
|
| 570 |
+
"logits/chosen": -1.9282668828964233,
|
| 571 |
+
"logits/rejected": -1.6367496252059937,
|
| 572 |
+
"logps/chosen": -351.134033203125,
|
| 573 |
+
"logps/rejected": -394.2359313964844,
|
| 574 |
+
"loss": 0.5659,
|
| 575 |
+
"rewards/accuracies": 0.65625,
|
| 576 |
+
"rewards/chosen": -1.9457461833953857,
|
| 577 |
+
"rewards/margins": 0.6128097772598267,
|
| 578 |
+
"rewards/rejected": -2.558556079864502,
|
| 579 |
"step": 370
|
| 580 |
},
|
| 581 |
{
|
| 582 |
+
"epoch": 0.3024875621890547,
|
| 583 |
+
"grad_norm": 16.77173991467865,
|
| 584 |
+
"learning_rate": 4.40214293992074e-07,
|
| 585 |
+
"logits/chosen": -1.747044563293457,
|
| 586 |
+
"logits/rejected": -1.6618773937225342,
|
| 587 |
+
"logps/chosen": -350.59466552734375,
|
| 588 |
+
"logps/rejected": -409.42901611328125,
|
| 589 |
+
"loss": 0.5571,
|
| 590 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 591 |
+
"rewards/chosen": -2.0240893363952637,
|
| 592 |
+
"rewards/margins": 0.44821667671203613,
|
| 593 |
+
"rewards/rejected": -2.4723057746887207,
|
| 594 |
"step": 380
|
| 595 |
},
|
| 596 |
{
|
| 597 |
+
"epoch": 0.31044776119402984,
|
| 598 |
+
"grad_norm": 18.040794768493605,
|
| 599 |
+
"learning_rate": 4.3563110184961234e-07,
|
| 600 |
+
"logits/chosen": -1.9076989889144897,
|
| 601 |
+
"logits/rejected": -1.5887449979782104,
|
| 602 |
+
"logps/chosen": -332.4600524902344,
|
| 603 |
+
"logps/rejected": -381.29803466796875,
|
| 604 |
+
"loss": 0.5534,
|
| 605 |
+
"rewards/accuracies": 0.668749988079071,
|
| 606 |
+
"rewards/chosen": -1.8599789142608643,
|
| 607 |
+
"rewards/margins": 0.4636203348636627,
|
| 608 |
+
"rewards/rejected": -2.323599338531494,
|
| 609 |
"step": 390
|
| 610 |
},
|
| 611 |
{
|
| 612 |
+
"epoch": 0.31840796019900497,
|
| 613 |
+
"grad_norm": 18.863620586743814,
|
| 614 |
+
"learning_rate": 4.3090443821097566e-07,
|
| 615 |
+
"logits/chosen": -1.803070306777954,
|
| 616 |
+
"logits/rejected": -1.5535482168197632,
|
| 617 |
+
"logps/chosen": -308.82965087890625,
|
| 618 |
+
"logps/rejected": -382.87445068359375,
|
| 619 |
+
"loss": 0.5601,
|
| 620 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 621 |
+
"rewards/chosen": -1.603441596031189,
|
| 622 |
+
"rewards/margins": 0.7121911644935608,
|
| 623 |
+
"rewards/rejected": -2.3156330585479736,
|
| 624 |
"step": 400
|
| 625 |
},
|
| 626 |
{
|
| 627 |
+
"epoch": 0.3263681592039801,
|
| 628 |
+
"grad_norm": 13.883572516302484,
|
| 629 |
+
"learning_rate": 4.2603795624364195e-07,
|
| 630 |
+
"logits/chosen": -1.9973773956298828,
|
| 631 |
+
"logits/rejected": -1.4498814344406128,
|
| 632 |
+
"logps/chosen": -357.13763427734375,
|
| 633 |
+
"logps/rejected": -393.8042907714844,
|
| 634 |
+
"loss": 0.5601,
|
| 635 |
+
"rewards/accuracies": 0.737500011920929,
|
| 636 |
+
"rewards/chosen": -2.0093302726745605,
|
| 637 |
+
"rewards/margins": 0.6177257299423218,
|
| 638 |
+
"rewards/rejected": -2.6270556449890137,
|
| 639 |
+
"step": 410
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.33432835820895523,
|
| 643 |
+
"grad_norm": 16.354917835389568,
|
| 644 |
+
"learning_rate": 4.210354171785795e-07,
|
| 645 |
+
"logits/chosen": -1.7048423290252686,
|
| 646 |
+
"logits/rejected": -1.520613670349121,
|
| 647 |
+
"logps/chosen": -350.70355224609375,
|
| 648 |
+
"logps/rejected": -406.6918029785156,
|
| 649 |
+
"loss": 0.5501,
|
| 650 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 651 |
+
"rewards/chosen": -2.1793723106384277,
|
| 652 |
+
"rewards/margins": 0.4814273416996002,
|
| 653 |
+
"rewards/rejected": -2.660799741744995,
|
| 654 |
+
"step": 420
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.34228855721393037,
|
| 658 |
+
"grad_norm": 15.11823770980584,
|
| 659 |
+
"learning_rate": 4.15900687403248e-07,
|
| 660 |
+
"logits/chosen": -1.8165044784545898,
|
| 661 |
+
"logits/rejected": -1.512001872062683,
|
| 662 |
+
"logps/chosen": -351.67706298828125,
|
| 663 |
+
"logps/rejected": -398.6982116699219,
|
| 664 |
+
"loss": 0.5579,
|
| 665 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 666 |
+
"rewards/chosen": -1.9194176197052002,
|
| 667 |
+
"rewards/margins": 0.5377460718154907,
|
| 668 |
+
"rewards/rejected": -2.4571633338928223,
|
| 669 |
+
"step": 430
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.3502487562189055,
|
| 673 |
+
"grad_norm": 14.824412259935942,
|
| 674 |
+
"learning_rate": 4.1063773547332584e-07,
|
| 675 |
+
"logits/chosen": -2.048107862472534,
|
| 676 |
+
"logits/rejected": -1.8060747385025024,
|
| 677 |
+
"logps/chosen": -311.2574462890625,
|
| 678 |
+
"logps/rejected": -372.2284240722656,
|
| 679 |
+
"loss": 0.5602,
|
| 680 |
+
"rewards/accuracies": 0.706250011920929,
|
| 681 |
+
"rewards/chosen": -1.580950379371643,
|
| 682 |
+
"rewards/margins": 0.6271029114723206,
|
| 683 |
+
"rewards/rejected": -2.2080533504486084,
|
| 684 |
+
"step": 440
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.3582089552238806,
|
| 688 |
+
"grad_norm": 17.525653455798285,
|
| 689 |
+
"learning_rate": 4.0525062904547276e-07,
|
| 690 |
+
"logits/chosen": -1.958742380142212,
|
| 691 |
+
"logits/rejected": -1.3642756938934326,
|
| 692 |
+
"logps/chosen": -334.4582824707031,
|
| 693 |
+
"logps/rejected": -390.8046875,
|
| 694 |
+
"loss": 0.5462,
|
| 695 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 696 |
+
"rewards/chosen": -1.877528429031372,
|
| 697 |
+
"rewards/margins": 0.6928645372390747,
|
| 698 |
+
"rewards/rejected": -2.5703933238983154,
|
| 699 |
+
"step": 450
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.3661691542288557,
|
| 703 |
+
"grad_norm": 16.688451846975745,
|
| 704 |
+
"learning_rate": 3.997435317334988e-07,
|
| 705 |
+
"logits/chosen": -1.9058139324188232,
|
| 706 |
+
"logits/rejected": -1.5897517204284668,
|
| 707 |
+
"logps/chosen": -351.8033752441406,
|
| 708 |
+
"logps/rejected": -405.9495544433594,
|
| 709 |
+
"loss": 0.5562,
|
| 710 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 711 |
+
"rewards/chosen": -1.860612154006958,
|
| 712 |
+
"rewards/margins": 0.639590859413147,
|
| 713 |
+
"rewards/rejected": -2.5002028942108154,
|
| 714 |
+
"step": 460
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.37412935323383084,
|
| 718 |
+
"grad_norm": 16.36630647240935,
|
| 719 |
+
"learning_rate": 3.941206998903701e-07,
|
| 720 |
+
"logits/chosen": -1.9422454833984375,
|
| 721 |
+
"logits/rejected": -1.6349403858184814,
|
| 722 |
+
"logps/chosen": -327.41119384765625,
|
| 723 |
+
"logps/rejected": -375.36102294921875,
|
| 724 |
+
"loss": 0.5597,
|
| 725 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 726 |
+
"rewards/chosen": -1.7558162212371826,
|
| 727 |
+
"rewards/margins": 0.5030372738838196,
|
| 728 |
+
"rewards/rejected": -2.2588534355163574,
|
| 729 |
+
"step": 470
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.382089552238806,
|
| 733 |
+
"grad_norm": 18.336351088486346,
|
| 734 |
+
"learning_rate": 3.8838647931853684e-07,
|
| 735 |
+
"logits/chosen": -1.7614314556121826,
|
| 736 |
+
"logits/rejected": -1.5050185918807983,
|
| 737 |
+
"logps/chosen": -329.1768798828125,
|
| 738 |
+
"logps/rejected": -407.611328125,
|
| 739 |
+
"loss": 0.546,
|
| 740 |
+
"rewards/accuracies": 0.71875,
|
| 741 |
+
"rewards/chosen": -1.9475609064102173,
|
| 742 |
+
"rewards/margins": 0.7653089761734009,
|
| 743 |
+
"rewards/rejected": -2.712869882583618,
|
| 744 |
+
"step": 480
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.3900497512437811,
|
| 748 |
+
"grad_norm": 14.526089578492307,
|
| 749 |
+
"learning_rate": 3.825453019111281e-07,
|
| 750 |
+
"logits/chosen": -1.854347825050354,
|
| 751 |
+
"logits/rejected": -1.477128028869629,
|
| 752 |
+
"logps/chosen": -344.77587890625,
|
| 753 |
+
"logps/rejected": -422.0589294433594,
|
| 754 |
+
"loss": 0.5412,
|
| 755 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 756 |
+
"rewards/chosen": -1.9149529933929443,
|
| 757 |
+
"rewards/margins": 0.7679746747016907,
|
| 758 |
+
"rewards/rejected": -2.6829276084899902,
|
| 759 |
+
"step": 490
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.39800995024875624,
|
| 763 |
+
"grad_norm": 21.279411516585405,
|
| 764 |
+
"learning_rate": 3.7660168222660824e-07,
|
| 765 |
+
"logits/chosen": -1.9854710102081299,
|
| 766 |
+
"logits/rejected": -1.706412672996521,
|
| 767 |
+
"logps/chosen": -314.8876953125,
|
| 768 |
+
"logps/rejected": -394.66925048828125,
|
| 769 |
+
"loss": 0.5304,
|
| 770 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 771 |
+
"rewards/chosen": -1.6202218532562256,
|
| 772 |
+
"rewards/margins": 0.7936272621154785,
|
| 773 |
+
"rewards/rejected": -2.413848876953125,
|
| 774 |
+
"step": 500
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.4059701492537313,
|
| 778 |
+
"grad_norm": 18.371143169851667,
|
| 779 |
+
"learning_rate": 3.705602139995416e-07,
|
| 780 |
+
"logits/chosen": -2.0707178115844727,
|
| 781 |
+
"logits/rejected": -1.6669524908065796,
|
| 782 |
+
"logps/chosen": -296.0068359375,
|
| 783 |
+
"logps/rejected": -350.65704345703125,
|
| 784 |
+
"loss": 0.5757,
|
| 785 |
+
"rewards/accuracies": 0.737500011920929,
|
| 786 |
+
"rewards/chosen": -1.597426414489746,
|
| 787 |
+
"rewards/margins": 0.5844928026199341,
|
| 788 |
+
"rewards/rejected": -2.1819193363189697,
|
| 789 |
+
"step": 510
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.41393034825870645,
|
| 793 |
+
"grad_norm": 13.765263633877712,
|
| 794 |
+
"learning_rate": 3.6442556659016475e-07,
|
| 795 |
+
"logits/chosen": -2.2126479148864746,
|
| 796 |
+
"logits/rejected": -1.9664087295532227,
|
| 797 |
+
"logps/chosen": -298.197509765625,
|
| 798 |
+
"logps/rejected": -359.9577331542969,
|
| 799 |
+
"loss": 0.5426,
|
| 800 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 801 |
+
"rewards/chosen": -1.5804378986358643,
|
| 802 |
+
"rewards/margins": 0.6014927625656128,
|
| 803 |
+
"rewards/rejected": -2.1819307804107666,
|
| 804 |
+
"step": 520
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.4218905472636816,
|
| 808 |
+
"grad_norm": 18.995220185419754,
|
| 809 |
+
"learning_rate": 3.582024813755076e-07,
|
| 810 |
+
"logits/chosen": -1.9249904155731201,
|
| 811 |
+
"logits/rejected": -1.5814507007598877,
|
| 812 |
+
"logps/chosen": -343.10919189453125,
|
| 813 |
+
"logps/rejected": -379.06683349609375,
|
| 814 |
+
"loss": 0.5501,
|
| 815 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 816 |
+
"rewards/chosen": -1.8100788593292236,
|
| 817 |
+
"rewards/margins": 0.5825816988945007,
|
| 818 |
+
"rewards/rejected": -2.392660617828369,
|
| 819 |
+
"step": 530
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.4298507462686567,
|
| 823 |
+
"grad_norm": 17.75911940444921,
|
| 824 |
+
"learning_rate": 3.5189576808485404e-07,
|
| 825 |
+
"logits/chosen": -1.8188612461090088,
|
| 826 |
+
"logits/rejected": -1.5995066165924072,
|
| 827 |
+
"logps/chosen": -329.63861083984375,
|
| 828 |
+
"logps/rejected": -399.28631591796875,
|
| 829 |
+
"loss": 0.5567,
|
| 830 |
+
"rewards/accuracies": 0.731249988079071,
|
| 831 |
+
"rewards/chosen": -1.9467417001724243,
|
| 832 |
+
"rewards/margins": 0.6499906778335571,
|
| 833 |
+
"rewards/rejected": -2.5967321395874023,
|
| 834 |
+
"step": 540
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 0.43781094527363185,
|
| 838 |
+
"grad_norm": 20.774910187785334,
|
| 839 |
+
"learning_rate": 3.4551030108237433e-07,
|
| 840 |
+
"logits/chosen": -2.104665994644165,
|
| 841 |
+
"logits/rejected": -1.6698726415634155,
|
| 842 |
+
"logps/chosen": -325.890869140625,
|
| 843 |
+
"logps/rejected": -385.27838134765625,
|
| 844 |
+
"loss": 0.5517,
|
| 845 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 846 |
+
"rewards/chosen": -1.6647937297821045,
|
| 847 |
+
"rewards/margins": 0.7040282487869263,
|
| 848 |
+
"rewards/rejected": -2.3688220977783203,
|
| 849 |
+
"step": 550
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.445771144278607,
|
| 853 |
+
"grad_norm": 20.43210579616367,
|
| 854 |
+
"learning_rate": 3.390510155998023e-07,
|
| 855 |
+
"logits/chosen": -2.038442611694336,
|
| 856 |
+
"logits/rejected": -1.6438343524932861,
|
| 857 |
+
"logps/chosen": -325.3457946777344,
|
| 858 |
+
"logps/rejected": -397.16790771484375,
|
| 859 |
+
"loss": 0.5347,
|
| 860 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 861 |
+
"rewards/chosen": -1.7963489294052124,
|
| 862 |
+
"rewards/margins": 0.6992012858390808,
|
| 863 |
+
"rewards/rejected": -2.4955506324768066,
|
| 864 |
+
"step": 560
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.4537313432835821,
|
| 868 |
+
"grad_norm": 18.83338554622648,
|
| 869 |
+
"learning_rate": 3.325229039220684e-07,
|
| 870 |
+
"logits/chosen": -2.120060682296753,
|
| 871 |
+
"logits/rejected": -1.6965103149414062,
|
| 872 |
+
"logps/chosen": -326.96197509765625,
|
| 873 |
+
"logps/rejected": -418.3592834472656,
|
| 874 |
+
"loss": 0.5631,
|
| 875 |
+
"rewards/accuracies": 0.78125,
|
| 876 |
+
"rewards/chosen": -1.686519980430603,
|
| 877 |
+
"rewards/margins": 0.8552719354629517,
|
| 878 |
+
"rewards/rejected": -2.541792154312134,
|
| 879 |
+
"step": 570
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.4616915422885572,
|
| 883 |
+
"grad_norm": 15.616564617549583,
|
| 884 |
+
"learning_rate": 3.2593101152883795e-07,
|
| 885 |
+
"logits/chosen": -2.0598907470703125,
|
| 886 |
+
"logits/rejected": -2.0200095176696777,
|
| 887 |
+
"logps/chosen": -321.26348876953125,
|
| 888 |
+
"logps/rejected": -392.21337890625,
|
| 889 |
+
"loss": 0.5639,
|
| 890 |
+
"rewards/accuracies": 0.668749988079071,
|
| 891 |
+
"rewards/chosen": -1.8532001972198486,
|
| 892 |
+
"rewards/margins": 0.5705356001853943,
|
| 893 |
+
"rewards/rejected": -2.4237358570098877,
|
| 894 |
+
"step": 580
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 0.4696517412935323,
|
| 898 |
+
"grad_norm": 15.28956399678925,
|
| 899 |
+
"learning_rate": 3.192804331949349e-07,
|
| 900 |
+
"logits/chosen": -2.078157901763916,
|
| 901 |
+
"logits/rejected": -1.7344363927841187,
|
| 902 |
+
"logps/chosen": -332.84100341796875,
|
| 903 |
+
"logps/rejected": -405.9479064941406,
|
| 904 |
+
"loss": 0.5267,
|
| 905 |
+
"rewards/accuracies": 0.762499988079071,
|
| 906 |
+
"rewards/chosen": -1.8422882556915283,
|
| 907 |
+
"rewards/margins": 0.780554473400116,
|
| 908 |
+
"rewards/rejected": -2.622842788696289,
|
| 909 |
+
"step": 590
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.47761194029850745,
|
| 913 |
+
"grad_norm": 20.287126580243257,
|
| 914 |
+
"learning_rate": 3.125763090526674e-07,
|
| 915 |
+
"logits/chosen": -1.8083003759384155,
|
| 916 |
+
"logits/rejected": -1.4841382503509521,
|
| 917 |
+
"logps/chosen": -308.1997985839844,
|
| 918 |
+
"logps/rejected": -384.129150390625,
|
| 919 |
+
"loss": 0.5666,
|
| 920 |
+
"rewards/accuracies": 0.668749988079071,
|
| 921 |
+
"rewards/chosen": -1.7298943996429443,
|
| 922 |
+
"rewards/margins": 0.6618615388870239,
|
| 923 |
+
"rewards/rejected": -2.3917558193206787,
|
| 924 |
+
"step": 600
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.4855721393034826,
|
| 928 |
+
"grad_norm": 15.063154404223198,
|
| 929 |
+
"learning_rate": 3.0582382061909623e-07,
|
| 930 |
+
"logits/chosen": -1.9274730682373047,
|
| 931 |
+
"logits/rejected": -1.7204185724258423,
|
| 932 |
+
"logps/chosen": -333.6025390625,
|
| 933 |
+
"logps/rejected": -400.88446044921875,
|
| 934 |
+
"loss": 0.5437,
|
| 935 |
+
"rewards/accuracies": 0.737500011920929,
|
| 936 |
+
"rewards/chosen": -1.8027238845825195,
|
| 937 |
+
"rewards/margins": 0.5840314626693726,
|
| 938 |
+
"rewards/rejected": -2.3867554664611816,
|
| 939 |
+
"step": 610
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.4935323383084577,
|
| 943 |
+
"grad_norm": 15.13264360466539,
|
| 944 |
+
"learning_rate": 2.9902818679131775e-07,
|
| 945 |
+
"logits/chosen": -1.9839204549789429,
|
| 946 |
+
"logits/rejected": -1.7049957513809204,
|
| 947 |
+
"logps/chosen": -320.3578796386719,
|
| 948 |
+
"logps/rejected": -402.8902587890625,
|
| 949 |
+
"loss": 0.5565,
|
| 950 |
+
"rewards/accuracies": 0.768750011920929,
|
| 951 |
+
"rewards/chosen": -1.8458640575408936,
|
| 952 |
+
"rewards/margins": 0.7623557448387146,
|
| 953 |
+
"rewards/rejected": -2.608219623565674,
|
| 954 |
+
"step": 620
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.5014925373134328,
|
| 958 |
+
"grad_norm": 18.241658852083088,
|
| 959 |
+
"learning_rate": 2.921946598128571e-07,
|
| 960 |
+
"logits/chosen": -1.8552188873291016,
|
| 961 |
+
"logits/rejected": -1.6487993001937866,
|
| 962 |
+
"logps/chosen": -320.94561767578125,
|
| 963 |
+
"logps/rejected": -407.82769775390625,
|
| 964 |
+
"loss": 0.5333,
|
| 965 |
+
"rewards/accuracies": 0.675000011920929,
|
| 966 |
+
"rewards/chosen": -1.7461166381835938,
|
| 967 |
+
"rewards/margins": 0.7682201266288757,
|
| 968 |
+
"rewards/rejected": -2.514336585998535,
|
| 969 |
+
"step": 630
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.5094527363184079,
|
| 973 |
+
"grad_norm": 18.48691570398085,
|
| 974 |
+
"learning_rate": 2.8532852121428733e-07,
|
| 975 |
+
"logits/chosen": -1.7485021352767944,
|
| 976 |
+
"logits/rejected": -1.165112018585205,
|
| 977 |
+
"logps/chosen": -350.9917907714844,
|
| 978 |
+
"logps/rejected": -418.5997619628906,
|
| 979 |
+
"loss": 0.5373,
|
| 980 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 981 |
+
"rewards/chosen": -2.05993914604187,
|
| 982 |
+
"rewards/margins": 0.7539520263671875,
|
| 983 |
+
"rewards/rejected": -2.8138911724090576,
|
| 984 |
+
"step": 640
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 0.5174129353233831,
|
| 988 |
+
"grad_norm": 16.255617080563933,
|
| 989 |
+
"learning_rate": 2.7843507773121414e-07,
|
| 990 |
+
"logits/chosen": -1.929992437362671,
|
| 991 |
+
"logits/rejected": -1.2393519878387451,
|
| 992 |
+
"logps/chosen": -344.0683898925781,
|
| 993 |
+
"logps/rejected": -392.7326354980469,
|
| 994 |
+
"loss": 0.5132,
|
| 995 |
+
"rewards/accuracies": 0.699999988079071,
|
| 996 |
+
"rewards/chosen": -1.9438108205795288,
|
| 997 |
+
"rewards/margins": 0.6626324653625488,
|
| 998 |
+
"rewards/rejected": -2.6064436435699463,
|
| 999 |
+
"step": 650
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 0.5253731343283582,
|
| 1003 |
+
"grad_norm": 19.277456544122902,
|
| 1004 |
+
"learning_rate": 2.715196572027789e-07,
|
| 1005 |
+
"logits/chosen": -1.7599384784698486,
|
| 1006 |
+
"logits/rejected": -1.478522777557373,
|
| 1007 |
+
"logps/chosen": -312.7566833496094,
|
| 1008 |
+
"logps/rejected": -391.0369873046875,
|
| 1009 |
+
"loss": 0.5536,
|
| 1010 |
+
"rewards/accuracies": 0.75,
|
| 1011 |
+
"rewards/chosen": -1.6669034957885742,
|
| 1012 |
+
"rewards/margins": 0.7611022591590881,
|
| 1013 |
+
"rewards/rejected": -2.4280056953430176,
|
| 1014 |
+
"step": 660
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 0.5333333333333333,
|
| 1018 |
+
"grad_norm": 17.410375794414502,
|
| 1019 |
+
"learning_rate": 2.645876044538521e-07,
|
| 1020 |
+
"logits/chosen": -1.991127371788025,
|
| 1021 |
+
"logits/rejected": -1.690598726272583,
|
| 1022 |
+
"logps/chosen": -306.1337890625,
|
| 1023 |
+
"logps/rejected": -391.0872802734375,
|
| 1024 |
+
"loss": 0.5555,
|
| 1025 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1026 |
+
"rewards/chosen": -1.585716962814331,
|
| 1027 |
+
"rewards/margins": 0.7589733600616455,
|
| 1028 |
+
"rewards/rejected": -2.3446903228759766,
|
| 1029 |
+
"step": 670
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.5412935323383085,
|
| 1033 |
+
"grad_norm": 15.825036910316848,
|
| 1034 |
+
"learning_rate": 2.5764427716409815e-07,
|
| 1035 |
+
"logits/chosen": -1.9365549087524414,
|
| 1036 |
+
"logits/rejected": -1.4879454374313354,
|
| 1037 |
+
"logps/chosen": -305.0138854980469,
|
| 1038 |
+
"logps/rejected": -357.05596923828125,
|
| 1039 |
+
"loss": 0.5446,
|
| 1040 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1041 |
+
"rewards/chosen": -1.7255747318267822,
|
| 1042 |
+
"rewards/margins": 0.5652529001235962,
|
| 1043 |
+
"rewards/rejected": -2.2908272743225098,
|
| 1044 |
+
"step": 680
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 0.5492537313432836,
|
| 1048 |
+
"grad_norm": 19.97205918760906,
|
| 1049 |
+
"learning_rate": 2.5069504172710494e-07,
|
| 1050 |
+
"logits/chosen": -1.8567272424697876,
|
| 1051 |
+
"logits/rejected": -1.7196296453475952,
|
| 1052 |
+
"logps/chosen": -317.20428466796875,
|
| 1053 |
+
"logps/rejected": -414.39404296875,
|
| 1054 |
+
"loss": 0.5509,
|
| 1055 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1056 |
+
"rewards/chosen": -1.7734630107879639,
|
| 1057 |
+
"rewards/margins": 0.7310962080955505,
|
| 1058 |
+
"rewards/rejected": -2.50455904006958,
|
| 1059 |
+
"step": 690
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.5572139303482587,
|
| 1063 |
+
"grad_norm": 23.24991074411631,
|
| 1064 |
+
"learning_rate": 2.4374526910277886e-07,
|
| 1065 |
+
"logits/chosen": -1.976335883140564,
|
| 1066 |
+
"logits/rejected": -1.6772323846817017,
|
| 1067 |
+
"logps/chosen": -334.4820251464844,
|
| 1068 |
+
"logps/rejected": -401.43585205078125,
|
| 1069 |
+
"loss": 0.5486,
|
| 1070 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1071 |
+
"rewards/chosen": -1.7821439504623413,
|
| 1072 |
+
"rewards/margins": 0.7127203345298767,
|
| 1073 |
+
"rewards/rejected": -2.4948642253875732,
|
| 1074 |
+
"step": 700
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.5651741293532339,
|
| 1078 |
+
"grad_norm": 18.01899564630511,
|
| 1079 |
+
"learning_rate": 2.368003306662104e-07,
|
| 1080 |
+
"logits/chosen": -1.8737258911132812,
|
| 1081 |
+
"logits/rejected": -1.4357565641403198,
|
| 1082 |
+
"logps/chosen": -350.22772216796875,
|
| 1083 |
+
"logps/rejected": -403.4524841308594,
|
| 1084 |
+
"loss": 0.5471,
|
| 1085 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 1086 |
+
"rewards/chosen": -1.8687664270401,
|
| 1087 |
+
"rewards/margins": 0.6295676231384277,
|
| 1088 |
+
"rewards/rejected": -2.4983339309692383,
|
| 1089 |
+
"step": 710
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.573134328358209,
|
| 1093 |
+
"grad_norm": 24.167230989729408,
|
| 1094 |
+
"learning_rate": 2.2986559405621886e-07,
|
| 1095 |
+
"logits/chosen": -1.8662601709365845,
|
| 1096 |
+
"logits/rejected": -1.612449288368225,
|
| 1097 |
+
"logps/chosen": -298.7684326171875,
|
| 1098 |
+
"logps/rejected": -387.6333312988281,
|
| 1099 |
+
"loss": 0.5546,
|
| 1100 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1101 |
+
"rewards/chosen": -1.6841051578521729,
|
| 1102 |
+
"rewards/margins": 0.712591290473938,
|
| 1103 |
+
"rewards/rejected": -2.3966965675354004,
|
| 1104 |
+
"step": 720
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 0.5810945273631841,
|
| 1108 |
+
"grad_norm": 17.64113619581508,
|
| 1109 |
+
"learning_rate": 2.2294641902678443e-07,
|
| 1110 |
+
"logits/chosen": -1.850368857383728,
|
| 1111 |
+
"logits/rejected": -1.5335843563079834,
|
| 1112 |
+
"logps/chosen": -342.439208984375,
|
| 1113 |
+
"logps/rejected": -392.17120361328125,
|
| 1114 |
+
"loss": 0.5178,
|
| 1115 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 1116 |
+
"rewards/chosen": -1.8309608697891235,
|
| 1117 |
+
"rewards/margins": 0.5988297462463379,
|
| 1118 |
+
"rewards/rejected": -2.4297900199890137,
|
| 1119 |
+
"step": 730
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 0.5890547263681593,
|
| 1123 |
+
"grad_norm": 18.74126586937816,
|
| 1124 |
+
"learning_rate": 2.160481533045751e-07,
|
| 1125 |
+
"logits/chosen": -1.860264778137207,
|
| 1126 |
+
"logits/rejected": -1.3350017070770264,
|
| 1127 |
+
"logps/chosen": -327.9742126464844,
|
| 1128 |
+
"logps/rejected": -388.90789794921875,
|
| 1129 |
+
"loss": 0.5582,
|
| 1130 |
+
"rewards/accuracies": 0.6875,
|
| 1131 |
+
"rewards/chosen": -1.7639926671981812,
|
| 1132 |
+
"rewards/margins": 0.679111659526825,
|
| 1133 |
+
"rewards/rejected": -2.4431042671203613,
|
| 1134 |
+
"step": 740
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 0.5970149253731343,
|
| 1138 |
+
"grad_norm": 17.924098734097832,
|
| 1139 |
+
"learning_rate": 2.0917612845576882e-07,
|
| 1140 |
+
"logits/chosen": -1.8485305309295654,
|
| 1141 |
+
"logits/rejected": -1.4422708749771118,
|
| 1142 |
+
"logps/chosen": -310.1546325683594,
|
| 1143 |
+
"logps/rejected": -376.4947204589844,
|
| 1144 |
+
"loss": 0.5258,
|
| 1145 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1146 |
+
"rewards/chosen": -1.6890716552734375,
|
| 1147 |
+
"rewards/margins": 0.6663106679916382,
|
| 1148 |
+
"rewards/rejected": -2.355382204055786,
|
| 1149 |
+
"step": 750
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 0.6049751243781094,
|
| 1153 |
+
"grad_norm": 16.32315826074932,
|
| 1154 |
+
"learning_rate": 2.0233565576536564e-07,
|
| 1155 |
+
"logits/chosen": -1.8351771831512451,
|
| 1156 |
+
"logits/rejected": -1.5401179790496826,
|
| 1157 |
+
"logps/chosen": -319.2179870605469,
|
| 1158 |
+
"logps/rejected": -388.4412536621094,
|
| 1159 |
+
"loss": 0.5637,
|
| 1160 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1161 |
+
"rewards/chosen": -1.7108579874038696,
|
| 1162 |
+
"rewards/margins": 0.6797488927841187,
|
| 1163 |
+
"rewards/rejected": -2.3906068801879883,
|
| 1164 |
+
"step": 760
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.6129353233830845,
|
| 1168 |
+
"grad_norm": 15.45028984513974,
|
| 1169 |
+
"learning_rate": 1.9553202213217537e-07,
|
| 1170 |
+
"logits/chosen": -2.024503469467163,
|
| 1171 |
+
"logits/rejected": -1.4444875717163086,
|
| 1172 |
+
"logps/chosen": -284.2763366699219,
|
| 1173 |
+
"logps/rejected": -360.1900939941406,
|
| 1174 |
+
"loss": 0.5266,
|
| 1175 |
+
"rewards/accuracies": 0.768750011920929,
|
| 1176 |
+
"rewards/chosen": -1.384358286857605,
|
| 1177 |
+
"rewards/margins": 0.8107110857963562,
|
| 1178 |
+
"rewards/rejected": -2.1950695514678955,
|
| 1179 |
+
"step": 770
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.6208955223880597,
|
| 1183 |
+
"grad_norm": 20.649664170903453,
|
| 1184 |
+
"learning_rate": 1.887704859826528e-07,
|
| 1185 |
+
"logits/chosen": -1.937726616859436,
|
| 1186 |
+
"logits/rejected": -1.501091718673706,
|
| 1187 |
+
"logps/chosen": -324.5860900878906,
|
| 1188 |
+
"logps/rejected": -393.491943359375,
|
| 1189 |
+
"loss": 0.5384,
|
| 1190 |
+
"rewards/accuracies": 0.6875,
|
| 1191 |
+
"rewards/chosen": -1.7339260578155518,
|
| 1192 |
+
"rewards/margins": 0.6434067487716675,
|
| 1193 |
+
"rewards/rejected": -2.3773326873779297,
|
| 1194 |
+
"step": 780
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 0.6288557213930348,
|
| 1198 |
+
"grad_norm": 21.08511180186782,
|
| 1199 |
+
"learning_rate": 1.8205627320673836e-07,
|
| 1200 |
+
"logits/chosen": -1.6596505641937256,
|
| 1201 |
+
"logits/rejected": -1.2972899675369263,
|
| 1202 |
+
"logps/chosen": -337.11920166015625,
|
| 1203 |
+
"logps/rejected": -406.6131286621094,
|
| 1204 |
+
"loss": 0.5298,
|
| 1205 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1206 |
+
"rewards/chosen": -1.9775470495224,
|
| 1207 |
+
"rewards/margins": 0.7096458673477173,
|
| 1208 |
+
"rewards/rejected": -2.6871931552886963,
|
| 1209 |
+
"step": 790
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.6368159203980099,
|
| 1213 |
+
"grad_norm": 17.361400963292194,
|
| 1214 |
+
"learning_rate": 1.7539457311884675e-07,
|
| 1215 |
+
"logits/chosen": -1.6588348150253296,
|
| 1216 |
+
"logits/rejected": -1.1941492557525635,
|
| 1217 |
+
"logps/chosen": -328.57879638671875,
|
| 1218 |
+
"logps/rejected": -394.7569274902344,
|
| 1219 |
+
"loss": 0.5249,
|
| 1220 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1221 |
+
"rewards/chosen": -1.87433660030365,
|
| 1222 |
+
"rewards/margins": 0.719002366065979,
|
| 1223 |
+
"rewards/rejected": -2.593339204788208,
|
| 1224 |
+
"step": 800
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 0.6447761194029851,
|
| 1228 |
+
"grad_norm": 16.182542903518556,
|
| 1229 |
+
"learning_rate": 1.687905344471226e-07,
|
| 1230 |
+
"logits/chosen": -1.7270431518554688,
|
| 1231 |
+
"logits/rejected": -1.6882022619247437,
|
| 1232 |
+
"logps/chosen": -318.8824462890625,
|
| 1233 |
+
"logps/rejected": -387.10113525390625,
|
| 1234 |
+
"loss": 0.5637,
|
| 1235 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1236 |
+
"rewards/chosen": -1.8991286754608154,
|
| 1237 |
+
"rewards/margins": 0.5004085302352905,
|
| 1238 |
+
"rewards/rejected": -2.3995373249053955,
|
| 1239 |
+
"step": 810
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 0.6527363184079602,
|
| 1243 |
+
"grad_norm": 16.172059932517488,
|
| 1244 |
+
"learning_rate": 1.6224926135406693e-07,
|
| 1245 |
+
"logits/chosen": -1.9119154214859009,
|
| 1246 |
+
"logits/rejected": -1.4078561067581177,
|
| 1247 |
+
"logps/chosen": -336.28851318359375,
|
| 1248 |
+
"logps/rejected": -399.19146728515625,
|
| 1249 |
+
"loss": 0.5432,
|
| 1250 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1251 |
+
"rewards/chosen": -1.8701921701431274,
|
| 1252 |
+
"rewards/margins": 0.729258120059967,
|
| 1253 |
+
"rewards/rejected": -2.5994503498077393,
|
| 1254 |
+
"step": 820
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 0.6606965174129353,
|
| 1258 |
+
"grad_norm": 15.778737810564293,
|
| 1259 |
+
"learning_rate": 1.557758094916053e-07,
|
| 1260 |
+
"logits/chosen": -1.7709137201309204,
|
| 1261 |
+
"logits/rejected": -1.497018814086914,
|
| 1262 |
+
"logps/chosen": -352.27264404296875,
|
| 1263 |
+
"logps/rejected": -433.1822204589844,
|
| 1264 |
+
"loss": 0.5354,
|
| 1265 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1266 |
+
"rewards/chosen": -2.045311450958252,
|
| 1267 |
+
"rewards/margins": 0.6497110724449158,
|
| 1268 |
+
"rewards/rejected": -2.6950223445892334,
|
| 1269 |
+
"step": 830
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.6686567164179105,
|
| 1273 |
+
"grad_norm": 16.405494774284087,
|
| 1274 |
+
"learning_rate": 1.4937518209365108e-07,
|
| 1275 |
+
"logits/chosen": -1.7025296688079834,
|
| 1276 |
+
"logits/rejected": -1.1163098812103271,
|
| 1277 |
+
"logps/chosen": -354.352294921875,
|
| 1278 |
+
"logps/rejected": -409.0311279296875,
|
| 1279 |
+
"loss": 0.5527,
|
| 1280 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1281 |
+
"rewards/chosen": -1.9278078079223633,
|
| 1282 |
+
"rewards/margins": 0.6557299494743347,
|
| 1283 |
+
"rewards/rejected": -2.5835378170013428,
|
| 1284 |
+
"step": 840
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 0.6766169154228856,
|
| 1288 |
+
"grad_norm": 22.75535436647274,
|
| 1289 |
+
"learning_rate": 1.4305232610918045e-07,
|
| 1290 |
+
"logits/chosen": -1.7835566997528076,
|
| 1291 |
+
"logits/rejected": -1.3081390857696533,
|
| 1292 |
+
"logps/chosen": -334.17626953125,
|
| 1293 |
+
"logps/rejected": -396.04974365234375,
|
| 1294 |
+
"loss": 0.5598,
|
| 1295 |
+
"rewards/accuracies": 0.6875,
|
| 1296 |
+
"rewards/chosen": -1.9084104299545288,
|
| 1297 |
+
"rewards/margins": 0.6380718350410461,
|
| 1298 |
+
"rewards/rejected": -2.5464820861816406,
|
| 1299 |
+
"step": 850
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 0.6845771144278607,
|
| 1303 |
+
"grad_norm": 16.28637474078457,
|
| 1304 |
+
"learning_rate": 1.3681212837880977e-07,
|
| 1305 |
+
"logits/chosen": -1.7336877584457397,
|
| 1306 |
+
"logits/rejected": -1.3859245777130127,
|
| 1307 |
+
"logps/chosen": -351.1092834472656,
|
| 1308 |
+
"logps/rejected": -402.8192138671875,
|
| 1309 |
+
"loss": 0.5382,
|
| 1310 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1311 |
+
"rewards/chosen": -1.9633920192718506,
|
| 1312 |
+
"rewards/margins": 0.6326353549957275,
|
| 1313 |
+
"rewards/rejected": -2.5960276126861572,
|
| 1314 |
+
"step": 860
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 0.6925373134328359,
|
| 1318 |
+
"grad_norm": 14.992190553347786,
|
| 1319 |
+
"learning_rate": 1.3065941185782977e-07,
|
| 1320 |
+
"logits/chosen": -1.6035921573638916,
|
| 1321 |
+
"logits/rejected": -1.390312910079956,
|
| 1322 |
+
"logps/chosen": -338.3635559082031,
|
| 1323 |
+
"logps/rejected": -407.48101806640625,
|
| 1324 |
+
"loss": 0.5465,
|
| 1325 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1326 |
+
"rewards/chosen": -2.1271004676818848,
|
| 1327 |
+
"rewards/margins": 0.5808795690536499,
|
| 1328 |
+
"rewards/rejected": -2.707979679107666,
|
| 1329 |
+
"step": 870
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 0.700497512437811,
|
| 1333 |
+
"grad_norm": 18.882701792962603,
|
| 1334 |
+
"learning_rate": 1.2459893188861613e-07,
|
| 1335 |
+
"logits/chosen": -1.7753925323486328,
|
| 1336 |
+
"logits/rejected": -1.313396692276001,
|
| 1337 |
+
"logps/chosen": -328.06280517578125,
|
| 1338 |
+
"logps/rejected": -409.3533020019531,
|
| 1339 |
+
"loss": 0.5378,
|
| 1340 |
+
"rewards/accuracies": 0.8125,
|
| 1341 |
+
"rewards/chosen": -1.7554057836532593,
|
| 1342 |
+
"rewards/margins": 0.8705118894577026,
|
| 1343 |
+
"rewards/rejected": -2.625917434692383,
|
| 1344 |
+
"step": 880
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 0.708457711442786,
|
| 1348 |
+
"grad_norm": 15.585282528531224,
|
| 1349 |
+
"learning_rate": 1.1863537252529548e-07,
|
| 1350 |
+
"logits/chosen": -1.6346286535263062,
|
| 1351 |
+
"logits/rejected": -1.0885813236236572,
|
| 1352 |
+
"logps/chosen": -314.69415283203125,
|
| 1353 |
+
"logps/rejected": -406.14593505859375,
|
| 1354 |
+
"loss": 0.5274,
|
| 1355 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1356 |
+
"rewards/chosen": -1.7842576503753662,
|
| 1357 |
+
"rewards/margins": 0.8567419052124023,
|
| 1358 |
+
"rewards/rejected": -2.6409995555877686,
|
| 1359 |
+
"step": 890
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 0.7164179104477612,
|
| 1363 |
+
"grad_norm": 17.240684758370243,
|
| 1364 |
+
"learning_rate": 1.1277334291351145e-07,
|
| 1365 |
+
"logits/chosen": -1.559463381767273,
|
| 1366 |
+
"logits/rejected": -0.979491114616394,
|
| 1367 |
+
"logps/chosen": -314.4317321777344,
|
| 1368 |
+
"logps/rejected": -395.7962341308594,
|
| 1369 |
+
"loss": 0.5356,
|
| 1370 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1371 |
+
"rewards/chosen": -1.7382938861846924,
|
| 1372 |
+
"rewards/margins": 0.8430318832397461,
|
| 1373 |
+
"rewards/rejected": -2.5813255310058594,
|
| 1374 |
+
"step": 900
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.7243781094527363,
|
| 1378 |
+
"grad_norm": 17.81287348840895,
|
| 1379 |
+
"learning_rate": 1.0701737372808431e-07,
|
| 1380 |
+
"logits/chosen": -1.356980562210083,
|
| 1381 |
+
"logits/rejected": -1.0851752758026123,
|
| 1382 |
+
"logps/chosen": -321.2406005859375,
|
| 1383 |
+
"logps/rejected": -439.5826110839844,
|
| 1384 |
+
"loss": 0.5266,
|
| 1385 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 1386 |
+
"rewards/chosen": -1.7592580318450928,
|
| 1387 |
+
"rewards/margins": 1.1267178058624268,
|
| 1388 |
+
"rewards/rejected": -2.8859758377075195,
|
| 1389 |
+
"step": 910
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.7323383084577114,
|
| 1393 |
+
"grad_norm": 16.146564583575124,
|
| 1394 |
+
"learning_rate": 1.0137191367132078e-07,
|
| 1395 |
+
"logits/chosen": -1.5627696514129639,
|
| 1396 |
+
"logits/rejected": -0.8526192903518677,
|
| 1397 |
+
"logps/chosen": -343.89093017578125,
|
| 1398 |
+
"logps/rejected": -410.4419860839844,
|
| 1399 |
+
"loss": 0.5202,
|
| 1400 |
+
"rewards/accuracies": 0.71875,
|
| 1401 |
+
"rewards/chosen": -2.032959461212158,
|
| 1402 |
+
"rewards/margins": 0.7283061146736145,
|
| 1403 |
+
"rewards/rejected": -2.761265754699707,
|
| 1404 |
+
"step": 920
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 0.7402985074626866,
|
| 1408 |
+
"grad_norm": 16.30781388479682,
|
| 1409 |
+
"learning_rate": 9.584132603467827e-08,
|
| 1410 |
+
"logits/chosen": -1.472614049911499,
|
| 1411 |
+
"logits/rejected": -1.0152631998062134,
|
| 1412 |
+
"logps/chosen": -369.9658203125,
|
| 1413 |
+
"logps/rejected": -457.43280029296875,
|
| 1414 |
+
"loss": 0.5391,
|
| 1415 |
+
"rewards/accuracies": 0.8125,
|
| 1416 |
+
"rewards/chosen": -2.0644240379333496,
|
| 1417 |
+
"rewards/margins": 0.929151177406311,
|
| 1418 |
+
"rewards/rejected": -2.99357533454895,
|
| 1419 |
+
"step": 930
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 0.7482587064676617,
|
| 1423 |
+
"grad_norm": 16.90834018386895,
|
| 1424 |
+
"learning_rate": 9.042988532644249e-08,
|
| 1425 |
+
"logits/chosen": -1.397674798965454,
|
| 1426 |
+
"logits/rejected": -1.057903528213501,
|
| 1427 |
+
"logps/chosen": -375.06805419921875,
|
| 1428 |
+
"logps/rejected": -459.8721618652344,
|
| 1429 |
+
"loss": 0.5326,
|
| 1430 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1431 |
+
"rewards/chosen": -2.2227516174316406,
|
| 1432 |
+
"rewards/margins": 0.7576023936271667,
|
| 1433 |
+
"rewards/rejected": -2.980353832244873,
|
| 1434 |
+
"step": 940
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 0.7562189054726368,
|
| 1438 |
+
"grad_norm": 19.44454583035634,
|
| 1439 |
+
"learning_rate": 8.514177396802428e-08,
|
| 1440 |
+
"logits/chosen": -1.3311452865600586,
|
| 1441 |
+
"logits/rejected": -0.988301157951355,
|
| 1442 |
+
"logps/chosen": -356.1918640136719,
|
| 1443 |
+
"logps/rejected": -429.5113220214844,
|
| 1444 |
+
"loss": 0.5311,
|
| 1445 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1446 |
+
"rewards/chosen": -2.1742124557495117,
|
| 1447 |
+
"rewards/margins": 0.7231701612472534,
|
| 1448 |
+
"rewards/rejected": -2.8973822593688965,
|
| 1449 |
+
"step": 950
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 0.764179104477612,
|
| 1453 |
+
"grad_norm": 18.789667512348625,
|
| 1454 |
+
"learning_rate": 7.998107906142839e-08,
|
| 1455 |
+
"logits/chosen": -1.2201939821243286,
|
| 1456 |
+
"logits/rejected": -0.8874862790107727,
|
| 1457 |
+
"logps/chosen": -364.1236267089844,
|
| 1458 |
+
"logps/rejected": -449.46466064453125,
|
| 1459 |
+
"loss": 0.5436,
|
| 1460 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1461 |
+
"rewards/chosen": -2.1308789253234863,
|
| 1462 |
+
"rewards/margins": 0.786114513874054,
|
| 1463 |
+
"rewards/rejected": -2.9169933795928955,
|
| 1464 |
+
"step": 960
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 0.7721393034825871,
|
| 1468 |
+
"grad_norm": 18.028047538401793,
|
| 1469 |
+
"learning_rate": 7.495178923039396e-08,
|
| 1470 |
+
"logits/chosen": -1.3335978984832764,
|
| 1471 |
+
"logits/rejected": -0.8373553156852722,
|
| 1472 |
+
"logps/chosen": -352.852294921875,
|
| 1473 |
+
"logps/rejected": -425.99853515625,
|
| 1474 |
+
"loss": 0.5323,
|
| 1475 |
+
"rewards/accuracies": 0.75,
|
| 1476 |
+
"rewards/chosen": -2.04585862159729,
|
| 1477 |
+
"rewards/margins": 0.8074928522109985,
|
| 1478 |
+
"rewards/rejected": -2.853351354598999,
|
| 1479 |
+
"step": 970
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 0.7800995024875622,
|
| 1483 |
+
"grad_norm": 19.107873765643053,
|
| 1484 |
+
"learning_rate": 7.005779153764682e-08,
|
| 1485 |
+
"logits/chosen": -1.2632755041122437,
|
| 1486 |
+
"logits/rejected": -1.1477338075637817,
|
| 1487 |
+
"logps/chosen": -362.19293212890625,
|
| 1488 |
+
"logps/rejected": -445.4449157714844,
|
| 1489 |
+
"loss": 0.5212,
|
| 1490 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1491 |
+
"rewards/chosen": -2.2130560874938965,
|
| 1492 |
+
"rewards/margins": 0.6761744618415833,
|
| 1493 |
+
"rewards/rejected": -2.889230728149414,
|
| 1494 |
+
"step": 980
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 0.7880597014925373,
|
| 1498 |
+
"grad_norm": 20.371049931041078,
|
| 1499 |
+
"learning_rate": 6.530286848064698e-08,
|
| 1500 |
+
"logits/chosen": -1.2885363101959229,
|
| 1501 |
+
"logits/rejected": -0.6255804896354675,
|
| 1502 |
+
"logps/chosen": -338.9552001953125,
|
| 1503 |
+
"logps/rejected": -428.79931640625,
|
| 1504 |
+
"loss": 0.5195,
|
| 1505 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 1506 |
+
"rewards/chosen": -2.0429329872131348,
|
| 1507 |
+
"rewards/margins": 0.9163187146186829,
|
| 1508 |
+
"rewards/rejected": -2.959251880645752,
|
| 1509 |
+
"step": 990
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 0.7960199004975125,
|
| 1513 |
+
"grad_norm": 19.72789599453256,
|
| 1514 |
+
"learning_rate": 6.069069506815325e-08,
|
| 1515 |
+
"logits/chosen": -1.360527515411377,
|
| 1516 |
+
"logits/rejected": -0.8227861523628235,
|
| 1517 |
+
"logps/chosen": -373.851318359375,
|
| 1518 |
+
"logps/rejected": -440.9930114746094,
|
| 1519 |
+
"loss": 0.5563,
|
| 1520 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1521 |
+
"rewards/chosen": -2.2749392986297607,
|
| 1522 |
+
"rewards/margins": 0.752488911151886,
|
| 1523 |
+
"rewards/rejected": -3.027428388595581,
|
| 1524 |
+
"step": 1000
|
| 1525 |
+
},
|
| 1526 |
+
{
|
| 1527 |
+
"epoch": 0.8039800995024876,
|
| 1528 |
+
"grad_norm": 17.338627129120006,
|
| 1529 |
+
"learning_rate": 5.6224835979863714e-08,
|
| 1530 |
+
"logits/chosen": -1.377640724182129,
|
| 1531 |
+
"logits/rejected": -0.8697378039360046,
|
| 1532 |
+
"logps/chosen": -388.3387756347656,
|
| 1533 |
+
"logps/rejected": -466.47174072265625,
|
| 1534 |
+
"loss": 0.5397,
|
| 1535 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1536 |
+
"rewards/chosen": -2.309962034225464,
|
| 1537 |
+
"rewards/margins": 0.7413021922111511,
|
| 1538 |
+
"rewards/rejected": -3.0512640476226807,
|
| 1539 |
+
"step": 1010
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 0.8119402985074626,
|
| 1543 |
+
"grad_norm": 18.9063759035064,
|
| 1544 |
+
"learning_rate": 5.190874281132851e-08,
|
| 1545 |
+
"logits/chosen": -1.4672131538391113,
|
| 1546 |
+
"logits/rejected": -0.6555231809616089,
|
| 1547 |
+
"logps/chosen": -377.4454650878906,
|
| 1548 |
+
"logps/rejected": -419.4335021972656,
|
| 1549 |
+
"loss": 0.5269,
|
| 1550 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1551 |
+
"rewards/chosen": -2.285759449005127,
|
| 1552 |
+
"rewards/margins": 0.5871948003768921,
|
| 1553 |
+
"rewards/rejected": -2.8729541301727295,
|
| 1554 |
+
"step": 1020
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 0.8199004975124378,
|
| 1558 |
+
"grad_norm": 17.93974901514964,
|
| 1559 |
+
"learning_rate": 4.774575140626316e-08,
|
| 1560 |
+
"logits/chosen": -1.227505087852478,
|
| 1561 |
+
"logits/rejected": -0.7521185278892517,
|
| 1562 |
+
"logps/chosen": -379.4794616699219,
|
| 1563 |
+
"logps/rejected": -443.300537109375,
|
| 1564 |
+
"loss": 0.5371,
|
| 1565 |
+
"rewards/accuracies": 0.71875,
|
| 1566 |
+
"rewards/chosen": -2.316551685333252,
|
| 1567 |
+
"rewards/margins": 0.7855070233345032,
|
| 1568 |
+
"rewards/rejected": -3.1020588874816895,
|
| 1569 |
+
"step": 1030
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 0.8278606965174129,
|
| 1573 |
+
"grad_norm": 19.765840382966818,
|
| 1574 |
+
"learning_rate": 4.373907927832513e-08,
|
| 1575 |
+
"logits/chosen": -1.2585428953170776,
|
| 1576 |
+
"logits/rejected": -0.6001216173171997,
|
| 1577 |
+
"logps/chosen": -360.7466125488281,
|
| 1578 |
+
"logps/rejected": -432.0596618652344,
|
| 1579 |
+
"loss": 0.5324,
|
| 1580 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1581 |
+
"rewards/chosen": -2.2043449878692627,
|
| 1582 |
+
"rewards/margins": 0.739472508430481,
|
| 1583 |
+
"rewards/rejected": -2.943817615509033,
|
| 1584 |
+
"step": 1040
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 0.835820895522388,
|
| 1588 |
+
"grad_norm": 16.791397180911105,
|
| 1589 |
+
"learning_rate": 3.9891823124345665e-08,
|
| 1590 |
+
"logits/chosen": -1.595428466796875,
|
| 1591 |
+
"logits/rejected": -0.7433096170425415,
|
| 1592 |
+
"logps/chosen": -372.77667236328125,
|
| 1593 |
+
"logps/rejected": -427.80633544921875,
|
| 1594 |
+
"loss": 0.5419,
|
| 1595 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1596 |
+
"rewards/chosen": -2.052586078643799,
|
| 1597 |
+
"rewards/margins": 0.7817685008049011,
|
| 1598 |
+
"rewards/rejected": -2.834354877471924,
|
| 1599 |
+
"step": 1050
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.8437810945273632,
|
| 1603 |
+
"grad_norm": 17.099354050242567,
|
| 1604 |
+
"learning_rate": 3.620695643093924e-08,
|
| 1605 |
+
"logits/chosen": -1.0566952228546143,
|
| 1606 |
+
"logits/rejected": -1.0658143758773804,
|
| 1607 |
+
"logps/chosen": -353.5264587402344,
|
| 1608 |
+
"logps/rejected": -453.6097717285156,
|
| 1609 |
+
"loss": 0.5355,
|
| 1610 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 1611 |
+
"rewards/chosen": -2.2306525707244873,
|
| 1612 |
+
"rewards/margins": 0.7557250261306763,
|
| 1613 |
+
"rewards/rejected": -2.986377716064453,
|
| 1614 |
+
"step": 1060
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 0.8517412935323383,
|
| 1618 |
+
"grad_norm": 17.626047522616386,
|
| 1619 |
+
"learning_rate": 3.268732717634032e-08,
|
| 1620 |
+
"logits/chosen": -1.082467794418335,
|
| 1621 |
+
"logits/rejected": -0.6993826031684875,
|
| 1622 |
+
"logps/chosen": -356.1838073730469,
|
| 1623 |
+
"logps/rejected": -432.0118713378906,
|
| 1624 |
+
"loss": 0.5382,
|
| 1625 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1626 |
+
"rewards/chosen": -2.2437031269073486,
|
| 1627 |
+
"rewards/margins": 0.703085720539093,
|
| 1628 |
+
"rewards/rejected": -2.946788787841797,
|
| 1629 |
+
"step": 1070
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 0.8597014925373134,
|
| 1633 |
+
"grad_norm": 18.59186735479471,
|
| 1634 |
+
"learning_rate": 2.9335655629243645e-08,
|
| 1635 |
+
"logits/chosen": -1.294574499130249,
|
| 1636 |
+
"logits/rejected": -0.9409192800521851,
|
| 1637 |
+
"logps/chosen": -375.03857421875,
|
| 1638 |
+
"logps/rejected": -447.16845703125,
|
| 1639 |
+
"loss": 0.5504,
|
| 1640 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1641 |
+
"rewards/chosen": -2.2595221996307373,
|
| 1642 |
+
"rewards/margins": 0.7225955724716187,
|
| 1643 |
+
"rewards/rejected": -2.9821178913116455,
|
| 1644 |
+
"step": 1080
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"epoch": 0.8676616915422886,
|
| 1648 |
+
"grad_norm": 21.397708720247653,
|
| 1649 |
+
"learning_rate": 2.6154532246349476e-08,
|
| 1650 |
+
"logits/chosen": -1.1999056339263916,
|
| 1651 |
+
"logits/rejected": -1.0267088413238525,
|
| 1652 |
+
"logps/chosen": -363.5381164550781,
|
| 1653 |
+
"logps/rejected": -422.05511474609375,
|
| 1654 |
+
"loss": 0.5754,
|
| 1655 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1656 |
+
"rewards/chosen": -2.3281185626983643,
|
| 1657 |
+
"rewards/margins": 0.50066077709198,
|
| 1658 |
+
"rewards/rejected": -2.8287787437438965,
|
| 1659 |
+
"step": 1090
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 0.8756218905472637,
|
| 1663 |
+
"grad_norm": 20.052028882501826,
|
| 1664 |
+
"learning_rate": 2.31464156702382e-08,
|
| 1665 |
+
"logits/chosen": -1.3676979541778564,
|
| 1666 |
+
"logits/rejected": -0.5716175436973572,
|
| 1667 |
+
"logps/chosen": -375.810302734375,
|
| 1668 |
+
"logps/rejected": -450.2315979003906,
|
| 1669 |
+
"loss": 0.5262,
|
| 1670 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1671 |
+
"rewards/chosen": -2.2596614360809326,
|
| 1672 |
+
"rewards/margins": 0.835864245891571,
|
| 1673 |
+
"rewards/rejected": -3.0955255031585693,
|
| 1674 |
+
"step": 1100
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 0.8835820895522388,
|
| 1678 |
+
"grad_norm": 22.37302428318916,
|
| 1679 |
+
"learning_rate": 2.031363082912252e-08,
|
| 1680 |
+
"logits/chosen": -1.2721402645111084,
|
| 1681 |
+
"logits/rejected": -0.6073659658432007,
|
| 1682 |
+
"logps/chosen": -368.27984619140625,
|
| 1683 |
+
"logps/rejected": -448.68157958984375,
|
| 1684 |
+
"loss": 0.5231,
|
| 1685 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1686 |
+
"rewards/chosen": -2.095541477203369,
|
| 1687 |
+
"rewards/margins": 0.8900821805000305,
|
| 1688 |
+
"rewards/rejected": -2.985623598098755,
|
| 1689 |
+
"step": 1110
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 0.891542288557214,
|
| 1693 |
+
"grad_norm": 25.266800229701747,
|
| 1694 |
+
"learning_rate": 1.7658367139945228e-08,
|
| 1695 |
+
"logits/chosen": -1.5140697956085205,
|
| 1696 |
+
"logits/rejected": -0.7945183515548706,
|
| 1697 |
+
"logps/chosen": -384.32269287109375,
|
| 1698 |
+
"logps/rejected": -433.766845703125,
|
| 1699 |
+
"loss": 0.5287,
|
| 1700 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1701 |
+
"rewards/chosen": -2.154731273651123,
|
| 1702 |
+
"rewards/margins": 0.6882571578025818,
|
| 1703 |
+
"rewards/rejected": -2.8429884910583496,
|
| 1704 |
+
"step": 1120
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 0.8995024875621891,
|
| 1708 |
+
"grad_norm": 20.62304308306268,
|
| 1709 |
+
"learning_rate": 1.5182676816211632e-08,
|
| 1710 |
+
"logits/chosen": -1.3684704303741455,
|
| 1711 |
+
"logits/rejected": -0.8318718075752258,
|
| 1712 |
+
"logps/chosen": -365.13079833984375,
|
| 1713 |
+
"logps/rejected": -472.0492248535156,
|
| 1714 |
+
"loss": 0.5163,
|
| 1715 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 1716 |
+
"rewards/chosen": -2.154531478881836,
|
| 1717 |
+
"rewards/margins": 1.0084809064865112,
|
| 1718 |
+
"rewards/rejected": -3.1630122661590576,
|
| 1719 |
+
"step": 1130
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 0.9074626865671642,
|
| 1723 |
+
"grad_norm": 16.335373316417314,
|
| 1724 |
+
"learning_rate": 1.2888473281864597e-08,
|
| 1725 |
+
"logits/chosen": -1.4910027980804443,
|
| 1726 |
+
"logits/rejected": -0.7867361903190613,
|
| 1727 |
+
"logps/chosen": -357.47393798828125,
|
| 1728 |
+
"logps/rejected": -418.6597595214844,
|
| 1729 |
+
"loss": 0.5239,
|
| 1730 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1731 |
+
"rewards/chosen": -2.1304430961608887,
|
| 1732 |
+
"rewards/margins": 0.6699010133743286,
|
| 1733 |
+
"rewards/rejected": -2.8003439903259277,
|
| 1734 |
+
"step": 1140
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 0.9154228855721394,
|
| 1738 |
+
"grad_norm": 19.127473957890142,
|
| 1739 |
+
"learning_rate": 1.0777529692427679e-08,
|
| 1740 |
+
"logits/chosen": -1.3886005878448486,
|
| 1741 |
+
"logits/rejected": -1.1176588535308838,
|
| 1742 |
+
"logps/chosen": -361.29595947265625,
|
| 1743 |
+
"logps/rejected": -450.3651428222656,
|
| 1744 |
+
"loss": 0.5321,
|
| 1745 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 1746 |
+
"rewards/chosen": -2.1988742351531982,
|
| 1747 |
+
"rewards/margins": 0.7343587279319763,
|
| 1748 |
+
"rewards/rejected": -2.9332327842712402,
|
| 1749 |
+
"step": 1150
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 0.9233830845771144,
|
| 1753 |
+
"grad_norm": 18.113861239231703,
|
| 1754 |
+
"learning_rate": 8.851477564560061e-09,
|
| 1755 |
+
"logits/chosen": -1.4955084323883057,
|
| 1756 |
+
"logits/rejected": -1.0137813091278076,
|
| 1757 |
+
"logps/chosen": -344.0810852050781,
|
| 1758 |
+
"logps/rejected": -438.7982482910156,
|
| 1759 |
+
"loss": 0.5534,
|
| 1760 |
+
"rewards/accuracies": 0.768750011920929,
|
| 1761 |
+
"rewards/chosen": -2.007995128631592,
|
| 1762 |
+
"rewards/margins": 0.8483503460884094,
|
| 1763 |
+
"rewards/rejected": -2.8563454151153564,
|
| 1764 |
+
"step": 1160
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 0.9313432835820895,
|
| 1768 |
+
"grad_norm": 18.419748638148626,
|
| 1769 |
+
"learning_rate": 7.111805515081531e-09,
|
| 1770 |
+
"logits/chosen": -1.1694109439849854,
|
| 1771 |
+
"logits/rejected": -0.9867475628852844,
|
| 1772 |
+
"logps/chosen": -344.5587158203125,
|
| 1773 |
+
"logps/rejected": -453.9609375,
|
| 1774 |
+
"loss": 0.5216,
|
| 1775 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1776 |
+
"rewards/chosen": -2.174985408782959,
|
| 1777 |
+
"rewards/margins": 0.8908305168151855,
|
| 1778 |
+
"rewards/rejected": -3.0658156871795654,
|
| 1779 |
+
"step": 1170
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 0.9393034825870646,
|
| 1783 |
+
"grad_norm": 16.31657411408144,
|
| 1784 |
+
"learning_rate": 5.559858110443016e-09,
|
| 1785 |
+
"logits/chosen": -1.391318440437317,
|
| 1786 |
+
"logits/rejected": -0.968989372253418,
|
| 1787 |
+
"logps/chosen": -338.8396301269531,
|
| 1788 |
+
"logps/rejected": -450.4630432128906,
|
| 1789 |
+
"loss": 0.5101,
|
| 1790 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 1791 |
+
"rewards/chosen": -2.04402232170105,
|
| 1792 |
+
"rewards/margins": 0.873876690864563,
|
| 1793 |
+
"rewards/rejected": -2.9178991317749023,
|
| 1794 |
+
"step": 1180
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 0.9472636815920398,
|
| 1798 |
+
"grad_norm": 16.213421308211853,
|
| 1799 |
+
"learning_rate": 4.196834827531276e-09,
|
| 1800 |
+
"logits/chosen": -1.3122155666351318,
|
| 1801 |
+
"logits/rejected": -0.727799117565155,
|
| 1802 |
+
"logps/chosen": -363.20123291015625,
|
| 1803 |
+
"logps/rejected": -421.6800231933594,
|
| 1804 |
+
"loss": 0.5223,
|
| 1805 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1806 |
+
"rewards/chosen": -2.240814685821533,
|
| 1807 |
+
"rewards/margins": 0.7033532857894897,
|
| 1808 |
+
"rewards/rejected": -2.9441676139831543,
|
| 1809 |
+
"step": 1190
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 0.9552238805970149,
|
| 1813 |
+
"grad_norm": 16.61159163977494,
|
| 1814 |
+
"learning_rate": 3.023789126611137e-09,
|
| 1815 |
+
"logits/chosen": -1.3651138544082642,
|
| 1816 |
+
"logits/rejected": -1.0306357145309448,
|
| 1817 |
+
"logps/chosen": -346.49127197265625,
|
| 1818 |
+
"logps/rejected": -433.2818298339844,
|
| 1819 |
+
"loss": 0.5514,
|
| 1820 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1821 |
+
"rewards/chosen": -2.017871856689453,
|
| 1822 |
+
"rewards/margins": 0.7361677289009094,
|
| 1823 |
+
"rewards/rejected": -2.7540395259857178,
|
| 1824 |
+
"step": 1200
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 0.96318407960199,
|
| 1828 |
+
"grad_norm": 19.71446411178462,
|
| 1829 |
+
"learning_rate": 2.041627637121929e-09,
|
| 1830 |
+
"logits/chosen": -1.2996383905410767,
|
| 1831 |
+
"logits/rejected": -0.6256915330886841,
|
| 1832 |
+
"logps/chosen": -358.1094665527344,
|
| 1833 |
+
"logps/rejected": -430.37811279296875,
|
| 1834 |
+
"loss": 0.5558,
|
| 1835 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 1836 |
+
"rewards/chosen": -2.246715784072876,
|
| 1837 |
+
"rewards/margins": 0.7623183727264404,
|
| 1838 |
+
"rewards/rejected": -3.0090343952178955,
|
| 1839 |
+
"step": 1210
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 0.9711442786069652,
|
| 1843 |
+
"grad_norm": 19.712848729427638,
|
| 1844 |
+
"learning_rate": 1.2511094569571668e-09,
|
| 1845 |
+
"logits/chosen": -1.328294277191162,
|
| 1846 |
+
"logits/rejected": -0.7868885397911072,
|
| 1847 |
+
"logps/chosen": -368.758056640625,
|
| 1848 |
+
"logps/rejected": -458.2462463378906,
|
| 1849 |
+
"loss": 0.5449,
|
| 1850 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1851 |
+
"rewards/chosen": -2.1683990955352783,
|
| 1852 |
+
"rewards/margins": 0.9658330082893372,
|
| 1853 |
+
"rewards/rejected": -3.1342320442199707,
|
| 1854 |
+
"step": 1220
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 0.9791044776119403,
|
| 1858 |
+
"grad_norm": 16.725128383172244,
|
| 1859 |
+
"learning_rate": 6.528455657691112e-10,
|
| 1860 |
+
"logits/chosen": -1.339804768562317,
|
| 1861 |
+
"logits/rejected": -0.7188054323196411,
|
| 1862 |
+
"logps/chosen": -367.33935546875,
|
| 1863 |
+
"logps/rejected": -451.83984375,
|
| 1864 |
+
"loss": 0.5116,
|
| 1865 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1866 |
+
"rewards/chosen": -2.153168201446533,
|
| 1867 |
+
"rewards/margins": 0.8999155759811401,
|
| 1868 |
+
"rewards/rejected": -3.053083896636963,
|
| 1869 |
+
"step": 1230
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 0.9870646766169154,
|
| 1873 |
+
"grad_norm": 19.596666231596082,
|
| 1874 |
+
"learning_rate": 2.4729835275189016e-10,
|
| 1875 |
+
"logits/chosen": -1.3673975467681885,
|
| 1876 |
+
"logits/rejected": -0.6154208183288574,
|
| 1877 |
+
"logps/chosen": -359.135009765625,
|
| 1878 |
+
"logps/rejected": -439.5702209472656,
|
| 1879 |
+
"loss": 0.543,
|
| 1880 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1881 |
+
"rewards/chosen": -2.121037721633911,
|
| 1882 |
+
"rewards/margins": 0.8869087100028992,
|
| 1883 |
+
"rewards/rejected": -3.007946491241455,
|
| 1884 |
+
"step": 1240
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 0.9950248756218906,
|
| 1888 |
+
"grad_norm": 19.08758915029115,
|
| 1889 |
+
"learning_rate": 3.478125926756337e-11,
|
| 1890 |
+
"logits/chosen": -1.2848167419433594,
|
| 1891 |
+
"logits/rejected": -0.7639325261116028,
|
| 1892 |
+
"logps/chosen": -373.97735595703125,
|
| 1893 |
+
"logps/rejected": -479.5267639160156,
|
| 1894 |
+
"loss": 0.5345,
|
| 1895 |
+
"rewards/accuracies": 0.78125,
|
| 1896 |
+
"rewards/chosen": -2.261409044265747,
|
| 1897 |
+
"rewards/margins": 0.9971047639846802,
|
| 1898 |
+
"rewards/rejected": -3.2585136890411377,
|
| 1899 |
+
"step": 1250
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 0.9998009950248756,
|
| 1903 |
+
"step": 1256,
|
| 1904 |
"total_flos": 0.0,
|
| 1905 |
+
"train_loss": 0.56497666932595,
|
| 1906 |
+
"train_runtime": 62670.8183,
|
| 1907 |
+
"train_samples_per_second": 2.566,
|
| 1908 |
+
"train_steps_per_second": 0.02
|
| 1909 |
}
|
| 1910 |
],
|
| 1911 |
"logging_steps": 10,
|
| 1912 |
+
"max_steps": 1256,
|
| 1913 |
+
"num_input_tokens_seen": 0,
|
| 1914 |
"num_train_epochs": 1,
|
| 1915 |
+
"save_steps": 100,
|
| 1916 |
+
"stateful_callbacks": {
|
| 1917 |
+
"TrainerControl": {
|
| 1918 |
+
"args": {
|
| 1919 |
+
"should_epoch_stop": false,
|
| 1920 |
+
"should_evaluate": false,
|
| 1921 |
+
"should_log": false,
|
| 1922 |
+
"should_save": true,
|
| 1923 |
+
"should_training_stop": false
|
| 1924 |
+
},
|
| 1925 |
+
"attributes": {}
|
| 1926 |
+
}
|
| 1927 |
+
},
|
| 1928 |
"total_flos": 0.0,
|
| 1929 |
+
"train_batch_size": 1,
|
| 1930 |
"trial_name": null,
|
| 1931 |
"trial_params": null
|
| 1932 |
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7e19cc2efde9c9407a03928a56d8f9f4a67e235bc530069812edfde062436b88
|
| 3 |
+
size 6456
|