Model save
Browse files- README.md +17 -17
- all_results.json +4 -4
- model-00001-of-00003.safetensors +1 -1
- model-00002-of-00003.safetensors +1 -1
- model-00003-of-00003.safetensors +1 -1
- runs/May20_13-41-32_n136-100-194/events.out.tfevents.1716184187.n136-100-194.1143920.0 +2 -2
- train_results.json +4 -4
- trainer_state.json +957 -957
README.md
CHANGED
|
@@ -15,15 +15,15 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 15 |
|
| 16 |
This model was trained from scratch on the None dataset.
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
-
- Loss: 0.
|
| 19 |
-
- Rewards/chosen:
|
| 20 |
-
- Rewards/rejected: -2.
|
| 21 |
-
- Rewards/accuracies: 0.
|
| 22 |
-
- Rewards/margins:
|
| 23 |
-
- Logps/rejected: -
|
| 24 |
-
- Logps/chosen: -
|
| 25 |
-
- Logits/rejected: -
|
| 26 |
-
- Logits/chosen: -
|
| 27 |
|
| 28 |
## Model description
|
| 29 |
|
|
@@ -60,14 +60,14 @@ The following hyperparameters were used during training:
|
|
| 60 |
|
| 61 |
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 62 |
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 63 |
-
| 0.
|
| 64 |
-
| 0.
|
| 65 |
-
| 0.
|
| 66 |
-
| 0.
|
| 67 |
-
| 0.
|
| 68 |
-
| 0.
|
| 69 |
-
| 0.
|
| 70 |
-
| 0.
|
| 71 |
|
| 72 |
|
| 73 |
### Framework versions
|
|
|
|
| 15 |
|
| 16 |
This model was trained from scratch on the None dataset.
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
+
- Loss: 0.2027
|
| 19 |
+
- Rewards/chosen: 0.6729
|
| 20 |
+
- Rewards/rejected: -2.3580
|
| 21 |
+
- Rewards/accuracies: 0.9141
|
| 22 |
+
- Rewards/margins: 3.0309
|
| 23 |
+
- Logps/rejected: -380.2658
|
| 24 |
+
- Logps/chosen: -322.0539
|
| 25 |
+
- Logits/rejected: -1.9204
|
| 26 |
+
- Logits/chosen: -1.9591
|
| 27 |
|
| 28 |
## Model description
|
| 29 |
|
|
|
|
| 60 |
|
| 61 |
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 62 |
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 63 |
+
| 0.4898 | 0.12 | 100 | 0.5505 | -0.1967 | -1.0051 | 0.6875 | 0.8085 | -353.2088 | -339.4445 | -1.7659 | -1.8469 |
|
| 64 |
+
| 0.4277 | 0.23 | 200 | 0.4655 | -0.4834 | -1.8836 | 0.7383 | 1.4002 | -370.7788 | -345.1795 | -1.7248 | -1.8009 |
|
| 65 |
+
| 0.4188 | 0.35 | 300 | 0.3922 | -0.0720 | -2.0263 | 0.7969 | 1.9544 | -373.6328 | -336.9513 | -1.6143 | -1.6899 |
|
| 66 |
+
| 0.3506 | 0.46 | 400 | 0.3457 | 0.2171 | -2.0472 | 0.8203 | 2.2643 | -374.0495 | -331.1692 | -1.9794 | -2.0296 |
|
| 67 |
+
| 0.3611 | 0.58 | 500 | 0.2959 | 0.2498 | -2.4347 | 0.8516 | 2.6844 | -381.7997 | -330.5164 | -1.8183 | -1.8592 |
|
| 68 |
+
| 0.3562 | 0.69 | 600 | 0.2513 | 0.3868 | -2.4732 | 0.8711 | 2.8600 | -382.5696 | -327.7753 | -1.9217 | -1.9736 |
|
| 69 |
+
| 0.3624 | 0.81 | 700 | 0.2194 | 0.6454 | -2.3556 | 0.9062 | 3.0010 | -380.2178 | -322.6031 | -1.9301 | -1.9717 |
|
| 70 |
+
| 0.4069 | 0.92 | 800 | 0.2027 | 0.6729 | -2.3580 | 0.9141 | 3.0309 | -380.2658 | -322.0539 | -1.9204 | -1.9591 |
|
| 71 |
|
| 72 |
|
| 73 |
### Framework versions
|
all_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 111134,
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.40559157427005504,
|
| 4 |
+
"train_runtime": 13777.3263,
|
| 5 |
"train_samples": 111134,
|
| 6 |
+
"train_samples_per_second": 8.066,
|
| 7 |
+
"train_steps_per_second": 0.063
|
| 8 |
}
|
model-00001-of-00003.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4943178720
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:419678fed1e21fc0085a3e5460e53663c8575f3962e49b1af0e61f023fd4943e
|
| 3 |
size 4943178720
|
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:cf7dfdf8135658ed696d23c67a80c515ab05997271e3dfbd5a614d3783b327ce
|
| 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 4540532728
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ae56d3cafd2deed309e3da51f8c4ba401fe0402f043f2bd51fb953c4cba99c36
|
| 3 |
size 4540532728
|
runs/May20_13-41-32_n136-100-194/events.out.tfevents.1716184187.n136-100-194.1143920.0
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 sha256:41ae544c3e35fea0352946b6fd6b18f40dea31b26fd311ee418f65aaaba669a4
|
| 3 |
+
size 70795
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 111134,
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.40559157427005504,
|
| 4 |
+
"train_runtime": 13777.3263,
|
| 5 |
"train_samples": 111134,
|
| 6 |
+
"train_samples_per_second": 8.066,
|
| 7 |
+
"train_steps_per_second": 0.063
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -10,13 +10,13 @@
|
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.0,
|
| 13 |
-
"grad_norm":
|
| 14 |
"learning_rate": 5.747126436781609e-09,
|
| 15 |
-
"logits/chosen": -1.
|
| 16 |
-
"logits/rejected": -1.
|
| 17 |
-
"logps/chosen": -
|
| 18 |
-
"logps/rejected": -
|
| 19 |
-
"loss": 0.
|
| 20 |
"rewards/accuracies": 0.0,
|
| 21 |
"rewards/chosen": 0.0,
|
| 22 |
"rewards/margins": 0.0,
|
|
@@ -25,1430 +25,1430 @@
|
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.01,
|
| 28 |
-
"grad_norm":
|
| 29 |
"learning_rate": 5.747126436781609e-08,
|
| 30 |
-
"logits/chosen": -1.
|
| 31 |
-
"logits/rejected": -1.
|
| 32 |
-
"logps/chosen": -
|
| 33 |
-
"logps/rejected": -
|
| 34 |
-
"loss": 0.
|
| 35 |
-
"rewards/accuracies": 0.
|
| 36 |
-
"rewards/chosen": -
|
| 37 |
-
"rewards/margins": 0.
|
| 38 |
-
"rewards/rejected": -0.
|
| 39 |
"step": 10
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.02,
|
| 43 |
-
"grad_norm":
|
| 44 |
"learning_rate": 1.1494252873563217e-07,
|
| 45 |
-
"logits/chosen": -
|
| 46 |
-
"logits/rejected": -1.
|
| 47 |
-
"logps/chosen": -
|
| 48 |
-
"logps/rejected": -
|
| 49 |
-
"loss": 0.
|
| 50 |
-
"rewards/accuracies": 0.
|
| 51 |
-
"rewards/chosen": 0.
|
| 52 |
-
"rewards/margins": 0.
|
| 53 |
-
"rewards/rejected": -0.
|
| 54 |
"step": 20
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 0.03,
|
| 58 |
-
"grad_norm":
|
| 59 |
"learning_rate": 1.7241379310344828e-07,
|
| 60 |
-
"logits/chosen": -1.
|
| 61 |
-
"logits/rejected": -1.
|
| 62 |
-
"logps/chosen": -
|
| 63 |
-
"logps/rejected": -
|
| 64 |
-
"loss": 0.
|
| 65 |
-
"rewards/accuracies": 0.
|
| 66 |
-
"rewards/chosen": 0.
|
| 67 |
-
"rewards/margins": 0.
|
| 68 |
-
"rewards/rejected": -
|
| 69 |
"step": 30
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.05,
|
| 73 |
-
"grad_norm":
|
| 74 |
"learning_rate": 2.2988505747126435e-07,
|
| 75 |
-
"logits/chosen": -1.
|
| 76 |
-
"logits/rejected": -1.
|
| 77 |
-
"logps/chosen": -
|
| 78 |
-
"logps/rejected": -
|
| 79 |
-
"loss": 0.
|
| 80 |
-
"rewards/accuracies": 0.
|
| 81 |
-
"rewards/chosen": 0.
|
| 82 |
-
"rewards/margins": 0.
|
| 83 |
-
"rewards/rejected": -0.
|
| 84 |
"step": 40
|
| 85 |
},
|
| 86 |
{
|
| 87 |
"epoch": 0.06,
|
| 88 |
-
"grad_norm":
|
| 89 |
"learning_rate": 2.873563218390804e-07,
|
| 90 |
-
"logits/chosen": -1.
|
| 91 |
-
"logits/rejected": -1.
|
| 92 |
-
"logps/chosen": -
|
| 93 |
-
"logps/rejected": -
|
| 94 |
-
"loss": 0.
|
| 95 |
-
"rewards/accuracies": 0.
|
| 96 |
-
"rewards/chosen": 0.
|
| 97 |
-
"rewards/margins": 0.
|
| 98 |
-
"rewards/rejected": -0.
|
| 99 |
"step": 50
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.07,
|
| 103 |
-
"grad_norm":
|
| 104 |
"learning_rate": 3.4482758620689656e-07,
|
| 105 |
-
"logits/chosen": -
|
| 106 |
-
"logits/rejected": -1.
|
| 107 |
-
"logps/chosen": -
|
| 108 |
-
"logps/rejected": -
|
| 109 |
-
"loss": 0.
|
| 110 |
-
"rewards/accuracies": 0.
|
| 111 |
-
"rewards/chosen":
|
| 112 |
-
"rewards/margins": 0.
|
| 113 |
-
"rewards/rejected": -0.
|
| 114 |
"step": 60
|
| 115 |
},
|
| 116 |
{
|
| 117 |
"epoch": 0.08,
|
| 118 |
-
"grad_norm":
|
| 119 |
"learning_rate": 4.0229885057471266e-07,
|
| 120 |
-
"logits/chosen": -1.
|
| 121 |
-
"logits/rejected": -1.
|
| 122 |
-
"logps/chosen": -
|
| 123 |
-
"logps/rejected": -
|
| 124 |
-
"loss": 0.
|
| 125 |
-
"rewards/accuracies": 0.
|
| 126 |
-
"rewards/chosen": -0.
|
| 127 |
-
"rewards/margins": 0.
|
| 128 |
-
"rewards/rejected": -0.
|
| 129 |
"step": 70
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.09,
|
| 133 |
-
"grad_norm":
|
| 134 |
"learning_rate": 4.597701149425287e-07,
|
| 135 |
-
"logits/chosen": -1.
|
| 136 |
-
"logits/rejected": -1.
|
| 137 |
-
"logps/chosen": -
|
| 138 |
-
"logps/rejected": -
|
| 139 |
-
"loss": 0.
|
| 140 |
-
"rewards/accuracies": 0.
|
| 141 |
-
"rewards/chosen": -0.
|
| 142 |
-
"rewards/margins": 0.
|
| 143 |
-
"rewards/rejected": -
|
| 144 |
"step": 80
|
| 145 |
},
|
| 146 |
{
|
| 147 |
"epoch": 0.1,
|
| 148 |
-
"grad_norm":
|
| 149 |
"learning_rate": 4.999817969178237e-07,
|
| 150 |
-
"logits/chosen": -1.
|
| 151 |
-
"logits/rejected": -1.
|
| 152 |
-
"logps/chosen": -
|
| 153 |
-
"logps/rejected": -
|
| 154 |
-
"loss": 0.
|
| 155 |
-
"rewards/accuracies": 0.
|
| 156 |
-
"rewards/chosen": -0.
|
| 157 |
-
"rewards/margins":
|
| 158 |
-
"rewards/rejected": -1.
|
| 159 |
"step": 90
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 0.12,
|
| 163 |
-
"grad_norm":
|
| 164 |
"learning_rate": 4.996582603056428e-07,
|
| 165 |
-
"logits/chosen": -1.
|
| 166 |
-
"logits/rejected": -1.
|
| 167 |
-
"logps/chosen": -
|
| 168 |
-
"logps/rejected": -
|
| 169 |
-
"loss": 0.
|
| 170 |
-
"rewards/accuracies": 0.
|
| 171 |
-
"rewards/chosen": -0.
|
| 172 |
-
"rewards/margins":
|
| 173 |
-
"rewards/rejected": -1.
|
| 174 |
"step": 100
|
| 175 |
},
|
| 176 |
{
|
| 177 |
"epoch": 0.12,
|
| 178 |
-
"eval_logits/chosen": -1.
|
| 179 |
-
"eval_logits/rejected": -1.
|
| 180 |
-
"eval_logps/chosen": -
|
| 181 |
-
"eval_logps/rejected": -
|
| 182 |
-
"eval_loss": 0.
|
| 183 |
-
"eval_rewards/accuracies": 0.
|
| 184 |
-
"eval_rewards/chosen": -0.
|
| 185 |
-
"eval_rewards/margins": 0.
|
| 186 |
-
"eval_rewards/rejected": -1.
|
| 187 |
-
"eval_runtime": 98.
|
| 188 |
-
"eval_samples_per_second": 20.
|
| 189 |
-
"eval_steps_per_second": 0.
|
| 190 |
"step": 100
|
| 191 |
},
|
| 192 |
{
|
| 193 |
"epoch": 0.13,
|
| 194 |
-
"grad_norm":
|
| 195 |
"learning_rate": 4.989308132738126e-07,
|
| 196 |
-
"logits/chosen": -1.
|
| 197 |
-
"logits/rejected": -1.
|
| 198 |
-
"logps/chosen": -
|
| 199 |
-
"logps/rejected": -
|
| 200 |
-
"loss": 0.
|
| 201 |
-
"rewards/accuracies": 0.
|
| 202 |
-
"rewards/chosen": -0.
|
| 203 |
-
"rewards/margins":
|
| 204 |
-
"rewards/rejected": -1.
|
| 205 |
"step": 110
|
| 206 |
},
|
| 207 |
{
|
| 208 |
"epoch": 0.14,
|
| 209 |
-
"grad_norm":
|
| 210 |
"learning_rate": 4.978006327248536e-07,
|
| 211 |
-
"logits/chosen": -1.
|
| 212 |
-
"logits/rejected": -1.
|
| 213 |
-
"logps/chosen": -
|
| 214 |
-
"logps/rejected": -
|
| 215 |
-
"loss": 0.
|
| 216 |
-
"rewards/accuracies": 0.
|
| 217 |
-
"rewards/chosen": -
|
| 218 |
-
"rewards/margins":
|
| 219 |
-
"rewards/rejected": -1.
|
| 220 |
"step": 120
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 0.15,
|
| 224 |
-
"grad_norm":
|
| 225 |
"learning_rate": 4.962695471250032e-07,
|
| 226 |
-
"logits/chosen": -1.
|
| 227 |
-
"logits/rejected": -1.
|
| 228 |
-
"logps/chosen": -
|
| 229 |
-
"logps/rejected": -
|
| 230 |
-
"loss": 0.
|
| 231 |
-
"rewards/accuracies": 0.
|
| 232 |
-
"rewards/chosen": -
|
| 233 |
-
"rewards/margins":
|
| 234 |
-
"rewards/rejected": -1.
|
| 235 |
"step": 130
|
| 236 |
},
|
| 237 |
{
|
| 238 |
"epoch": 0.16,
|
| 239 |
-
"grad_norm":
|
| 240 |
"learning_rate": 4.94340033546025e-07,
|
| 241 |
-
"logits/chosen": -1.
|
| 242 |
-
"logits/rejected": -1.
|
| 243 |
-
"logps/chosen": -
|
| 244 |
-
"logps/rejected": -
|
| 245 |
-
"loss": 0.
|
| 246 |
-
"rewards/accuracies": 0.
|
| 247 |
-
"rewards/chosen": -
|
| 248 |
-
"rewards/margins":
|
| 249 |
-
"rewards/rejected": -
|
| 250 |
"step": 140
|
| 251 |
},
|
| 252 |
{
|
| 253 |
"epoch": 0.17,
|
| 254 |
-
"grad_norm":
|
| 255 |
"learning_rate": 4.920152136576705e-07,
|
| 256 |
-
"logits/chosen": -1.
|
| 257 |
-
"logits/rejected": -1.
|
| 258 |
-
"logps/chosen": -
|
| 259 |
-
"logps/rejected": -
|
| 260 |
-
"loss": 0.
|
| 261 |
-
"rewards/accuracies": 0.
|
| 262 |
-
"rewards/chosen": -
|
| 263 |
-
"rewards/margins":
|
| 264 |
-
"rewards/rejected": -
|
| 265 |
"step": 150
|
| 266 |
},
|
| 267 |
{
|
| 268 |
"epoch": 0.18,
|
| 269 |
-
"grad_norm":
|
| 270 |
"learning_rate": 4.892988486772756e-07,
|
| 271 |
-
"logits/chosen": -1.
|
| 272 |
-
"logits/rejected": -1.
|
| 273 |
-
"logps/chosen": -
|
| 274 |
-
"logps/rejected": -
|
| 275 |
-
"loss": 0.
|
| 276 |
-
"rewards/accuracies": 0.
|
| 277 |
-
"rewards/chosen": -
|
| 278 |
-
"rewards/margins":
|
| 279 |
-
"rewards/rejected": -2.
|
| 280 |
"step": 160
|
| 281 |
},
|
| 282 |
{
|
| 283 |
"epoch": 0.2,
|
| 284 |
-
"grad_norm":
|
| 285 |
"learning_rate": 4.861953332846629e-07,
|
| 286 |
-
"logits/chosen": -1.
|
| 287 |
-
"logits/rejected": -
|
| 288 |
-
"logps/chosen": -
|
| 289 |
-
"logps/rejected": -
|
| 290 |
-
"loss": 0.
|
| 291 |
"rewards/accuracies": 0.7875000238418579,
|
| 292 |
-
"rewards/chosen": -
|
| 293 |
-
"rewards/margins":
|
| 294 |
-
"rewards/rejected": -2.
|
| 295 |
"step": 170
|
| 296 |
},
|
| 297 |
{
|
| 298 |
"epoch": 0.21,
|
| 299 |
-
"grad_norm":
|
| 300 |
"learning_rate": 4.827096885121953e-07,
|
| 301 |
-
"logits/chosen": -
|
| 302 |
-
"logits/rejected": -
|
| 303 |
-
"logps/chosen": -
|
| 304 |
-
"logps/rejected": -
|
| 305 |
-
"loss": 0.
|
| 306 |
-
"rewards/accuracies": 0.
|
| 307 |
-
"rewards/chosen": -1.
|
| 308 |
-
"rewards/margins":
|
| 309 |
-
"rewards/rejected": -2.
|
| 310 |
"step": 180
|
| 311 |
},
|
| 312 |
{
|
| 313 |
"epoch": 0.22,
|
| 314 |
-
"grad_norm":
|
| 315 |
"learning_rate": 4.788475536214821e-07,
|
| 316 |
-
"logits/chosen": -
|
| 317 |
-
"logits/rejected": -
|
| 318 |
-
"logps/chosen": -
|
| 319 |
-
"logps/rejected": -
|
| 320 |
-
"loss": 0.
|
| 321 |
-
"rewards/accuracies": 0.
|
| 322 |
-
"rewards/chosen": -
|
| 323 |
-
"rewards/margins": 1.
|
| 324 |
-
"rewards/rejected": -2.
|
| 325 |
"step": 190
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 0.23,
|
| 329 |
-
"grad_norm":
|
| 330 |
"learning_rate": 4.746151769798818e-07,
|
| 331 |
-
"logits/chosen": -
|
| 332 |
-
"logits/rejected": -
|
| 333 |
-
"logps/chosen": -
|
| 334 |
-
"logps/rejected": -
|
| 335 |
-
"loss": 0.
|
| 336 |
-
"rewards/accuracies": 0.
|
| 337 |
-
"rewards/chosen": -
|
| 338 |
-
"rewards/margins": 1.
|
| 339 |
-
"rewards/rejected": -2.
|
| 340 |
"step": 200
|
| 341 |
},
|
| 342 |
{
|
| 343 |
"epoch": 0.23,
|
| 344 |
-
"eval_logits/chosen": -
|
| 345 |
-
"eval_logits/rejected": -
|
| 346 |
-
"eval_logps/chosen": -
|
| 347 |
-
"eval_logps/rejected": -
|
| 348 |
-
"eval_loss": 0.
|
| 349 |
-
"eval_rewards/accuracies": 0.
|
| 350 |
-
"eval_rewards/chosen": -
|
| 351 |
-
"eval_rewards/margins":
|
| 352 |
-
"eval_rewards/rejected": -
|
| 353 |
-
"eval_runtime": 98.
|
| 354 |
-
"eval_samples_per_second": 20.
|
| 355 |
"eval_steps_per_second": 0.326,
|
| 356 |
"step": 200
|
| 357 |
},
|
| 358 |
{
|
| 359 |
"epoch": 0.24,
|
| 360 |
-
"grad_norm":
|
| 361 |
"learning_rate": 4.7001940595156055e-07,
|
| 362 |
-
"logits/chosen": -
|
| 363 |
-
"logits/rejected": -
|
| 364 |
-
"logps/chosen": -
|
| 365 |
-
"logps/rejected": -
|
| 366 |
-
"loss": 0.
|
| 367 |
"rewards/accuracies": 0.8062499761581421,
|
| 368 |
-
"rewards/chosen": -1.
|
| 369 |
-
"rewards/margins": 1.
|
| 370 |
-
"rewards/rejected": -2.
|
| 371 |
"step": 210
|
| 372 |
},
|
| 373 |
{
|
| 374 |
"epoch": 0.25,
|
| 375 |
-
"grad_norm":
|
| 376 |
"learning_rate": 4.650676758194623e-07,
|
| 377 |
-
"logits/chosen": -
|
| 378 |
-
"logits/rejected": -
|
| 379 |
-
"logps/chosen": -
|
| 380 |
-
"logps/rejected": -
|
| 381 |
-
"loss": 0.
|
| 382 |
-
"rewards/accuracies": 0.
|
| 383 |
-
"rewards/chosen": -
|
| 384 |
-
"rewards/margins": 1.
|
| 385 |
-
"rewards/rejected": -2.
|
| 386 |
"step": 220
|
| 387 |
},
|
| 388 |
{
|
| 389 |
"epoch": 0.26,
|
| 390 |
-
"grad_norm":
|
| 391 |
"learning_rate": 4.5976799775611215e-07,
|
| 392 |
-
"logits/chosen": -
|
| 393 |
-
"logits/rejected": -
|
| 394 |
-
"logps/chosen": -
|
| 395 |
-
"logps/rejected": -
|
| 396 |
-
"loss": 0.
|
| 397 |
-
"rewards/accuracies": 0.
|
| 398 |
-
"rewards/chosen": -1.
|
| 399 |
-
"rewards/margins": 1.
|
| 400 |
-
"rewards/rejected": -3.
|
| 401 |
"step": 230
|
| 402 |
},
|
| 403 |
{
|
| 404 |
"epoch": 0.28,
|
| 405 |
-
"grad_norm":
|
| 406 |
"learning_rate": 4.5412894586271543e-07,
|
| 407 |
-
"logits/chosen": -
|
| 408 |
-
"logits/rejected": -
|
| 409 |
-
"logps/chosen": -
|
| 410 |
-
"logps/rejected": -
|
| 411 |
-
"loss": 0.
|
| 412 |
-
"rewards/accuracies": 0.
|
| 413 |
-
"rewards/chosen": -
|
| 414 |
-
"rewards/margins": 1.
|
| 415 |
-
"rewards/rejected": -
|
| 416 |
"step": 240
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 0.29,
|
| 420 |
-
"grad_norm":
|
| 421 |
"learning_rate": 4.481596432975201e-07,
|
| 422 |
-
"logits/chosen": -
|
| 423 |
-
"logits/rejected": -
|
| 424 |
-
"logps/chosen": -
|
| 425 |
-
"logps/rejected": -
|
| 426 |
-
"loss": 0.
|
| 427 |
-
"rewards/accuracies": 0.
|
| 428 |
-
"rewards/chosen": -
|
| 429 |
-
"rewards/margins": 1.
|
| 430 |
-
"rewards/rejected": -2.
|
| 431 |
"step": 250
|
| 432 |
},
|
| 433 |
{
|
| 434 |
"epoch": 0.3,
|
| 435 |
-
"grad_norm":
|
| 436 |
"learning_rate": 4.41869747515886e-07,
|
| 437 |
-
"logits/chosen": -
|
| 438 |
-
"logits/rejected": -
|
| 439 |
-
"logps/chosen": -
|
| 440 |
-
"logps/rejected": -
|
| 441 |
-
"loss": 0.
|
| 442 |
-
"rewards/accuracies": 0.
|
| 443 |
-
"rewards/chosen": -1.
|
| 444 |
-
"rewards/margins": 1.
|
| 445 |
-
"rewards/rejected": -2.
|
| 446 |
"step": 260
|
| 447 |
},
|
| 448 |
{
|
| 449 |
"epoch": 0.31,
|
| 450 |
-
"grad_norm":
|
| 451 |
"learning_rate": 4.352694346459396e-07,
|
| 452 |
-
"logits/chosen":
|
| 453 |
-
"logits/rejected":
|
| 454 |
-
"logps/chosen": -
|
| 455 |
-
"logps/rejected": -
|
| 456 |
-
"loss": 0.
|
| 457 |
-
"rewards/accuracies": 0.
|
| 458 |
-
"rewards/chosen": -
|
| 459 |
-
"rewards/margins": 1.
|
| 460 |
-
"rewards/rejected": -2.
|
| 461 |
"step": 270
|
| 462 |
},
|
| 463 |
{
|
| 464 |
"epoch": 0.32,
|
| 465 |
-
"grad_norm":
|
| 466 |
"learning_rate": 4.2836938302509256e-07,
|
| 467 |
-
"logits/chosen": -
|
| 468 |
-
"logits/rejected":
|
| 469 |
-
"logps/chosen": -
|
| 470 |
-
"logps/rejected": -
|
| 471 |
-
"loss": 0.
|
| 472 |
-
"rewards/accuracies": 0.
|
| 473 |
-
"rewards/chosen": -
|
| 474 |
-
"rewards/margins": 1.
|
| 475 |
-
"rewards/rejected": -2.
|
| 476 |
"step": 280
|
| 477 |
},
|
| 478 |
{
|
| 479 |
"epoch": 0.33,
|
| 480 |
-
"grad_norm":
|
| 481 |
"learning_rate": 4.2118075592405874e-07,
|
| 482 |
-
"logits/chosen":
|
| 483 |
-
"logits/rejected":
|
| 484 |
-
"logps/chosen": -
|
| 485 |
-
"logps/rejected": -
|
| 486 |
-
"loss": 0.
|
| 487 |
-
"rewards/accuracies": 0.
|
| 488 |
-
"rewards/chosen": -
|
| 489 |
-
"rewards/margins": 1.
|
| 490 |
-
"rewards/rejected": -
|
| 491 |
"step": 290
|
| 492 |
},
|
| 493 |
{
|
| 494 |
"epoch": 0.35,
|
| 495 |
-
"grad_norm":
|
| 496 |
"learning_rate": 4.137151834863213e-07,
|
| 497 |
-
"logits/chosen":
|
| 498 |
-
"logits/rejected":
|
| 499 |
-
"logps/chosen": -
|
| 500 |
-
"logps/rejected": -
|
| 501 |
-
"loss": 0.
|
| 502 |
-
"rewards/accuracies": 0.
|
| 503 |
-
"rewards/chosen": -1.
|
| 504 |
-
"rewards/margins": 1.
|
| 505 |
-
"rewards/rejected": -
|
| 506 |
"step": 300
|
| 507 |
},
|
| 508 |
{
|
| 509 |
"epoch": 0.35,
|
| 510 |
-
"eval_logits/chosen": -
|
| 511 |
-
"eval_logits/rejected": -
|
| 512 |
-
"eval_logps/chosen": -
|
| 513 |
-
"eval_logps/rejected": -
|
| 514 |
-
"eval_loss": 0.
|
| 515 |
-
"eval_rewards/accuracies": 0.
|
| 516 |
-
"eval_rewards/chosen": -
|
| 517 |
-
"eval_rewards/margins":
|
| 518 |
-
"eval_rewards/rejected": -2.
|
| 519 |
-
"eval_runtime": 98.
|
| 520 |
-
"eval_samples_per_second": 20.
|
| 521 |
"eval_steps_per_second": 0.326,
|
| 522 |
"step": 300
|
| 523 |
},
|
| 524 |
{
|
| 525 |
"epoch": 0.36,
|
| 526 |
-
"grad_norm":
|
| 527 |
"learning_rate": 4.059847439122671e-07,
|
| 528 |
-
"logits/chosen":
|
| 529 |
-
"logits/rejected":
|
| 530 |
-
"logps/chosen": -
|
| 531 |
-
"logps/rejected": -
|
| 532 |
-
"loss": 0.
|
| 533 |
-
"rewards/accuracies": 0.
|
| 534 |
-
"rewards/chosen": -
|
| 535 |
-
"rewards/margins": 1.
|
| 536 |
-
"rewards/rejected": -
|
| 537 |
"step": 310
|
| 538 |
},
|
| 539 |
{
|
| 540 |
"epoch": 0.37,
|
| 541 |
-
"grad_norm":
|
| 542 |
"learning_rate": 3.98001943918432e-07,
|
| 543 |
-
"logits/chosen":
|
| 544 |
-
"logits/rejected": 1.
|
| 545 |
-
"logps/chosen": -
|
| 546 |
-
"logps/rejected": -
|
| 547 |
-
"loss": 0.
|
| 548 |
-
"rewards/accuracies": 0.
|
| 549 |
-
"rewards/chosen": -1.
|
| 550 |
-
"rewards/margins": 1.
|
| 551 |
-
"rewards/rejected": -2.
|
| 552 |
"step": 320
|
| 553 |
},
|
| 554 |
{
|
| 555 |
"epoch": 0.38,
|
| 556 |
-
"grad_norm":
|
| 557 |
"learning_rate": 3.8977969850346866e-07,
|
| 558 |
-
"logits/chosen":
|
| 559 |
-
"logits/rejected":
|
| 560 |
-
"logps/chosen": -
|
| 561 |
-
"logps/rejected": -
|
| 562 |
-
"loss": 0.
|
| 563 |
-
"rewards/accuracies": 0.
|
| 564 |
-
"rewards/chosen": -
|
| 565 |
-
"rewards/margins": 1.
|
| 566 |
-
"rewards/rejected": -
|
| 567 |
"step": 330
|
| 568 |
},
|
| 569 |
{
|
| 570 |
"epoch": 0.39,
|
| 571 |
-
"grad_norm":
|
| 572 |
"learning_rate": 3.8133131005357465e-07,
|
| 573 |
-
"logits/chosen":
|
| 574 |
-
"logits/rejected":
|
| 575 |
-
"logps/chosen": -
|
| 576 |
-
"logps/rejected": -
|
| 577 |
-
"loss": 0.
|
| 578 |
-
"rewards/accuracies": 0.
|
| 579 |
-
"rewards/chosen": -
|
| 580 |
-
"rewards/margins":
|
| 581 |
-
"rewards/rejected": -3.
|
| 582 |
"step": 340
|
| 583 |
},
|
| 584 |
{
|
| 585 |
"epoch": 0.4,
|
| 586 |
-
"grad_norm":
|
| 587 |
"learning_rate": 3.7267044682118435e-07,
|
| 588 |
-
"logits/chosen":
|
| 589 |
-
"logits/rejected":
|
| 590 |
-
"logps/chosen": -
|
| 591 |
-
"logps/rejected": -
|
| 592 |
-
"loss": 0.
|
| 593 |
-
"rewards/accuracies": 0.
|
| 594 |
-
"rewards/chosen": -1.
|
| 595 |
-
"rewards/margins": 1.
|
| 596 |
-
"rewards/rejected": -
|
| 597 |
"step": 350
|
| 598 |
},
|
| 599 |
{
|
| 600 |
"epoch": 0.41,
|
| 601 |
-
"grad_norm":
|
| 602 |
"learning_rate": 3.638111208117425e-07,
|
| 603 |
-
"logits/chosen":
|
| 604 |
-
"logits/rejected":
|
| 605 |
-
"logps/chosen": -
|
| 606 |
-
"logps/rejected": -
|
| 607 |
-
"loss": 0.
|
| 608 |
-
"rewards/accuracies": 0.
|
| 609 |
-
"rewards/chosen": -
|
| 610 |
-
"rewards/margins": 1.
|
| 611 |
-
"rewards/rejected": -
|
| 612 |
"step": 360
|
| 613 |
},
|
| 614 |
{
|
| 615 |
"epoch": 0.43,
|
| 616 |
-
"grad_norm":
|
| 617 |
"learning_rate": 3.5476766511433605e-07,
|
| 618 |
-
"logits/chosen":
|
| 619 |
-
"logits/rejected":
|
| 620 |
-
"logps/chosen": -
|
| 621 |
-
"logps/rejected": -
|
| 622 |
-
"loss": 0.
|
| 623 |
-
"rewards/accuracies": 0.
|
| 624 |
-
"rewards/chosen": -1.
|
| 625 |
-
"rewards/margins": 1.
|
| 626 |
-
"rewards/rejected": -
|
| 627 |
"step": 370
|
| 628 |
},
|
| 629 |
{
|
| 630 |
"epoch": 0.44,
|
| 631 |
-
"grad_norm":
|
| 632 |
"learning_rate": 3.455547107128602e-07,
|
| 633 |
-
"logits/chosen":
|
| 634 |
-
"logits/rejected":
|
| 635 |
-
"logps/chosen": -
|
| 636 |
-
"logps/rejected": -
|
| 637 |
-
"loss": 0.
|
| 638 |
-
"rewards/accuracies": 0.
|
| 639 |
-
"rewards/chosen": -1.
|
| 640 |
-
"rewards/margins":
|
| 641 |
-
"rewards/rejected": -3.
|
| 642 |
"step": 380
|
| 643 |
},
|
| 644 |
{
|
| 645 |
"epoch": 0.45,
|
| 646 |
-
"grad_norm":
|
| 647 |
"learning_rate": 3.361871628152338e-07,
|
| 648 |
-
"logits/chosen":
|
| 649 |
-
"logits/rejected": 1.
|
| 650 |
-
"logps/chosen": -
|
| 651 |
-
"logps/rejected": -
|
| 652 |
-
"loss": 0.
|
| 653 |
-
"rewards/accuracies": 0.
|
| 654 |
-
"rewards/chosen": -1.
|
| 655 |
-
"rewards/margins": 1.
|
| 656 |
-
"rewards/rejected": -
|
| 657 |
"step": 390
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 0.46,
|
| 661 |
-
"grad_norm":
|
| 662 |
"learning_rate": 3.2668017673896077e-07,
|
| 663 |
-
"logits/chosen":
|
| 664 |
-
"logits/rejected": 1.
|
| 665 |
-
"logps/chosen": -
|
| 666 |
-
"logps/rejected": -
|
| 667 |
-
"loss": 0.
|
| 668 |
-
"rewards/accuracies": 0.
|
| 669 |
-
"rewards/chosen": -
|
| 670 |
-
"rewards/margins":
|
| 671 |
-
"rewards/rejected": -
|
| 672 |
"step": 400
|
| 673 |
},
|
| 674 |
{
|
| 675 |
"epoch": 0.46,
|
| 676 |
-
"eval_logits/chosen": -
|
| 677 |
-
"eval_logits/rejected": -
|
| 678 |
-
"eval_logps/chosen": -
|
| 679 |
-
"eval_logps/rejected": -
|
| 680 |
-
"eval_loss": 0.
|
| 681 |
-
"eval_rewards/accuracies": 0.
|
| 682 |
-
"eval_rewards/chosen":
|
| 683 |
-
"eval_rewards/margins":
|
| 684 |
-
"eval_rewards/rejected": -2.
|
| 685 |
-
"eval_runtime":
|
| 686 |
-
"eval_samples_per_second": 20.
|
| 687 |
-
"eval_steps_per_second": 0.
|
| 688 |
"step": 400
|
| 689 |
},
|
| 690 |
{
|
| 691 |
"epoch": 0.47,
|
| 692 |
-
"grad_norm":
|
| 693 |
"learning_rate": 3.1704913339205103e-07,
|
| 694 |
-
"logits/chosen":
|
| 695 |
-
"logits/rejected":
|
| 696 |
-
"logps/chosen": -
|
| 697 |
-
"logps/rejected": -
|
| 698 |
-
"loss": 0.
|
| 699 |
-
"rewards/accuracies": 0.
|
| 700 |
-
"rewards/chosen": -
|
| 701 |
-
"rewards/margins": 1.
|
| 702 |
-
"rewards/rejected": -
|
| 703 |
"step": 410
|
| 704 |
},
|
| 705 |
{
|
| 706 |
"epoch": 0.48,
|
| 707 |
-
"grad_norm":
|
| 708 |
"learning_rate": 3.0730961438896885e-07,
|
| 709 |
-
"logits/chosen":
|
| 710 |
-
"logits/rejected":
|
| 711 |
-
"logps/chosen": -
|
| 712 |
-
"logps/rejected": -
|
| 713 |
-
"loss": 0.
|
| 714 |
-
"rewards/accuracies": 0.
|
| 715 |
-
"rewards/chosen": -
|
| 716 |
-
"rewards/margins": 1.
|
| 717 |
-
"rewards/rejected": -
|
| 718 |
"step": 420
|
| 719 |
},
|
| 720 |
{
|
| 721 |
"epoch": 0.5,
|
| 722 |
-
"grad_norm":
|
| 723 |
"learning_rate": 2.9747737684186795e-07,
|
| 724 |
-
"logits/chosen":
|
| 725 |
-
"logits/rejected":
|
| 726 |
-
"logps/chosen": -
|
| 727 |
-
"logps/rejected": -
|
| 728 |
-
"loss": 0.
|
| 729 |
"rewards/accuracies": 0.8125,
|
| 730 |
-
"rewards/chosen": -
|
| 731 |
-
"rewards/margins": 1.
|
| 732 |
-
"rewards/rejected": -
|
| 733 |
"step": 430
|
| 734 |
},
|
| 735 |
{
|
| 736 |
"epoch": 0.51,
|
| 737 |
-
"grad_norm":
|
| 738 |
"learning_rate": 2.8756832786789663e-07,
|
| 739 |
-
"logits/chosen":
|
| 740 |
-
"logits/rejected":
|
| 741 |
-
"logps/chosen": -
|
| 742 |
-
"logps/rejected": -
|
| 743 |
-
"loss": 0.
|
| 744 |
"rewards/accuracies": 0.8187500238418579,
|
| 745 |
-
"rewards/chosen": -
|
| 746 |
-
"rewards/margins": 1.
|
| 747 |
-
"rewards/rejected": -
|
| 748 |
"step": 440
|
| 749 |
},
|
| 750 |
{
|
| 751 |
"epoch": 0.52,
|
| 752 |
-
"grad_norm":
|
| 753 |
"learning_rate": 2.7759849885381747e-07,
|
| 754 |
-
"logits/chosen":
|
| 755 |
-
"logits/rejected":
|
| 756 |
-
"logps/chosen": -
|
| 757 |
-
"logps/rejected": -
|
| 758 |
-
"loss": 0.
|
| 759 |
-
"rewards/accuracies": 0.
|
| 760 |
-
"rewards/chosen": -
|
| 761 |
-
"rewards/margins":
|
| 762 |
-
"rewards/rejected": -
|
| 763 |
"step": 450
|
| 764 |
},
|
| 765 |
{
|
| 766 |
"epoch": 0.53,
|
| 767 |
-
"grad_norm":
|
| 768 |
"learning_rate": 2.675840195195762e-07,
|
| 769 |
-
"logits/chosen":
|
| 770 |
-
"logits/rejected":
|
| 771 |
-
"logps/chosen": -
|
| 772 |
-
"logps/rejected": -
|
| 773 |
-
"loss": 0.
|
| 774 |
-
"rewards/accuracies": 0.
|
| 775 |
-
"rewards/chosen": -
|
| 776 |
-
"rewards/margins":
|
| 777 |
-
"rewards/rejected": -
|
| 778 |
"step": 460
|
| 779 |
},
|
| 780 |
{
|
| 781 |
"epoch": 0.54,
|
| 782 |
-
"grad_norm":
|
| 783 |
"learning_rate": 2.575410918227829e-07,
|
| 784 |
-
"logits/chosen":
|
| 785 |
-
"logits/rejected":
|
| 786 |
-
"logps/chosen": -
|
| 787 |
-
"logps/rejected": -
|
| 788 |
-
"loss": 0.
|
| 789 |
-
"rewards/accuracies": 0.
|
| 790 |
-
"rewards/chosen": -
|
| 791 |
-
"rewards/margins": 1.
|
| 792 |
-
"rewards/rejected": -
|
| 793 |
"step": 470
|
| 794 |
},
|
| 795 |
{
|
| 796 |
"epoch": 0.55,
|
| 797 |
-
"grad_norm":
|
| 798 |
"learning_rate": 2.474859637463226e-07,
|
| 799 |
-
"logits/chosen":
|
| 800 |
-
"logits/rejected":
|
| 801 |
-
"logps/chosen": -
|
| 802 |
-
"logps/rejected": -
|
| 803 |
-
"loss": 0.
|
| 804 |
-
"rewards/accuracies": 0.
|
| 805 |
-
"rewards/chosen": -
|
| 806 |
-
"rewards/margins":
|
| 807 |
-
"rewards/rejected": -
|
| 808 |
"step": 480
|
| 809 |
},
|
| 810 |
{
|
| 811 |
"epoch": 0.56,
|
| 812 |
-
"grad_norm":
|
| 813 |
"learning_rate": 2.3743490301150355e-07,
|
| 814 |
-
"logits/chosen":
|
| 815 |
-
"logits/rejected":
|
| 816 |
-
"logps/chosen": -
|
| 817 |
-
"logps/rejected": -
|
| 818 |
-
"loss": 0.
|
| 819 |
-
"rewards/accuracies": 0.
|
| 820 |
-
"rewards/chosen": -
|
| 821 |
-
"rewards/margins":
|
| 822 |
-
"rewards/rejected": -
|
| 823 |
"step": 490
|
| 824 |
},
|
| 825 |
{
|
| 826 |
"epoch": 0.58,
|
| 827 |
-
"grad_norm":
|
| 828 |
"learning_rate": 2.274041707592724e-07,
|
| 829 |
-
"logits/chosen":
|
| 830 |
-
"logits/rejected": 1.
|
| 831 |
-
"logps/chosen": -
|
| 832 |
-
"logps/rejected": -
|
| 833 |
-
"loss": 0.
|
| 834 |
-
"rewards/accuracies": 0.
|
| 835 |
-
"rewards/chosen": -
|
| 836 |
-
"rewards/margins":
|
| 837 |
-
"rewards/rejected": -
|
| 838 |
"step": 500
|
| 839 |
},
|
| 840 |
{
|
| 841 |
"epoch": 0.58,
|
| 842 |
-
"eval_logits/chosen": -
|
| 843 |
-
"eval_logits/rejected": -
|
| 844 |
-
"eval_logps/chosen": -
|
| 845 |
-
"eval_logps/rejected": -
|
| 846 |
-
"eval_loss": 0.
|
| 847 |
-
"eval_rewards/accuracies": 0.
|
| 848 |
-
"eval_rewards/chosen":
|
| 849 |
-
"eval_rewards/margins":
|
| 850 |
-
"eval_rewards/rejected": -2.
|
| 851 |
-
"eval_runtime":
|
| 852 |
-
"eval_samples_per_second": 20.
|
| 853 |
-
"eval_steps_per_second": 0.
|
| 854 |
"step": 500
|
| 855 |
},
|
| 856 |
{
|
| 857 |
"epoch": 0.59,
|
| 858 |
-
"grad_norm":
|
| 859 |
"learning_rate": 2.17409995242075e-07,
|
| 860 |
-
"logits/chosen":
|
| 861 |
-
"logits/rejected": 1.
|
| 862 |
-
"logps/chosen": -
|
| 863 |
-
"logps/rejected": -
|
| 864 |
-
"loss": 0.
|
| 865 |
-
"rewards/accuracies": 0.
|
| 866 |
-
"rewards/chosen": -
|
| 867 |
-
"rewards/margins":
|
| 868 |
-
"rewards/rejected": -3.
|
| 869 |
"step": 510
|
| 870 |
},
|
| 871 |
{
|
| 872 |
"epoch": 0.6,
|
| 873 |
-
"grad_norm":
|
| 874 |
"learning_rate": 2.0746854556892544e-07,
|
| 875 |
-
"logits/chosen":
|
| 876 |
-
"logits/rejected":
|
| 877 |
-
"logps/chosen": -
|
| 878 |
-
"logps/rejected": -
|
| 879 |
-
"loss": 0.
|
| 880 |
-
"rewards/accuracies": 0.
|
| 881 |
-
"rewards/chosen": -
|
| 882 |
-
"rewards/margins": 1.
|
| 883 |
-
"rewards/rejected": -
|
| 884 |
"step": 520
|
| 885 |
},
|
| 886 |
{
|
| 887 |
"epoch": 0.61,
|
| 888 |
-
"grad_norm":
|
| 889 |
"learning_rate": 1.9759590554616173e-07,
|
| 890 |
-
"logits/chosen":
|
| 891 |
-
"logits/rejected":
|
| 892 |
-
"logps/chosen": -
|
| 893 |
-
"logps/rejected": -
|
| 894 |
-
"loss": 0.
|
| 895 |
-
"rewards/accuracies": 0.
|
| 896 |
-
"rewards/chosen": -
|
| 897 |
-
"rewards/margins": 1.
|
| 898 |
-
"rewards/rejected": -
|
| 899 |
"step": 530
|
| 900 |
},
|
| 901 |
{
|
| 902 |
"epoch": 0.62,
|
| 903 |
-
"grad_norm":
|
| 904 |
"learning_rate": 1.8780804765620746e-07,
|
| 905 |
-
"logits/chosen":
|
| 906 |
-
"logits/rejected":
|
| 907 |
-
"logps/chosen": -
|
| 908 |
-
"logps/rejected": -
|
| 909 |
-
"loss": 0.
|
| 910 |
-
"rewards/accuracies": 0.
|
| 911 |
-
"rewards/chosen": -
|
| 912 |
-
"rewards/margins": 1.
|
| 913 |
-
"rewards/rejected": -
|
| 914 |
"step": 540
|
| 915 |
},
|
| 916 |
{
|
| 917 |
"epoch": 0.63,
|
| 918 |
-
"grad_norm":
|
| 919 |
"learning_rate": 1.7812080721643973e-07,
|
| 920 |
-
"logits/chosen":
|
| 921 |
-
"logits/rejected": 1.
|
| 922 |
-
"logps/chosen": -
|
| 923 |
-
"logps/rejected": -
|
| 924 |
-
"loss": 0.
|
| 925 |
-
"rewards/accuracies": 0.
|
| 926 |
-
"rewards/chosen": -
|
| 927 |
-
"rewards/margins": 1.
|
| 928 |
-
"rewards/rejected": -
|
| 929 |
"step": 550
|
| 930 |
},
|
| 931 |
{
|
| 932 |
"epoch": 0.64,
|
| 933 |
-
"grad_norm":
|
| 934 |
"learning_rate": 1.6854985675997063e-07,
|
| 935 |
-
"logits/chosen":
|
| 936 |
-
"logits/rejected":
|
| 937 |
-
"logps/chosen": -
|
| 938 |
-
"logps/rejected": -
|
| 939 |
-
"loss": 0.
|
| 940 |
-
"rewards/accuracies": 0.
|
| 941 |
-
"rewards/chosen": -
|
| 942 |
-
"rewards/margins": 1.
|
| 943 |
-
"rewards/rejected": -
|
| 944 |
"step": 560
|
| 945 |
},
|
| 946 |
{
|
| 947 |
"epoch": 0.66,
|
| 948 |
-
"grad_norm":
|
| 949 |
"learning_rate": 1.5911068067978818e-07,
|
| 950 |
-
"logits/chosen":
|
| 951 |
-
"logits/rejected":
|
| 952 |
-
"logps/chosen": -
|
| 953 |
-
"logps/rejected": -
|
| 954 |
-
"loss": 0.
|
| 955 |
-
"rewards/accuracies": 0.
|
| 956 |
-
"rewards/chosen": -
|
| 957 |
-
"rewards/margins": 1.
|
| 958 |
-
"rewards/rejected": -
|
| 959 |
"step": 570
|
| 960 |
},
|
| 961 |
{
|
| 962 |
"epoch": 0.67,
|
| 963 |
-
"grad_norm":
|
| 964 |
"learning_rate": 1.4981855017728197e-07,
|
| 965 |
-
"logits/chosen":
|
| 966 |
-
"logits/rejected":
|
| 967 |
-
"logps/chosen": -
|
| 968 |
-
"logps/rejected": -
|
| 969 |
-
"loss": 0.
|
| 970 |
-
"rewards/accuracies": 0.
|
| 971 |
-
"rewards/chosen": -
|
| 972 |
-
"rewards/margins": 1.
|
| 973 |
-
"rewards/rejected": -3.
|
| 974 |
"step": 580
|
| 975 |
},
|
| 976 |
{
|
| 977 |
"epoch": 0.68,
|
| 978 |
-
"grad_norm":
|
| 979 |
"learning_rate": 1.406884985556804e-07,
|
| 980 |
-
"logits/chosen":
|
| 981 |
-
"logits/rejected": 1.
|
| 982 |
-
"logps/chosen": -
|
| 983 |
-
"logps/rejected": -
|
| 984 |
-
"loss": 0.
|
| 985 |
-
"rewards/accuracies": 0.
|
| 986 |
-
"rewards/chosen": -
|
| 987 |
-
"rewards/margins":
|
| 988 |
-
"rewards/rejected": -
|
| 989 |
"step": 590
|
| 990 |
},
|
| 991 |
{
|
| 992 |
"epoch": 0.69,
|
| 993 |
-
"grad_norm":
|
| 994 |
"learning_rate": 1.3173529689837354e-07,
|
| 995 |
-
"logits/chosen":
|
| 996 |
-
"logits/rejected": 1.
|
| 997 |
-
"logps/chosen": -
|
| 998 |
-
"logps/rejected": -
|
| 999 |
-
"loss": 0.
|
| 1000 |
"rewards/accuracies": 0.800000011920929,
|
| 1001 |
-
"rewards/chosen": -1.
|
| 1002 |
-
"rewards/margins": 1.
|
| 1003 |
-
"rewards/rejected": -3.
|
| 1004 |
"step": 600
|
| 1005 |
},
|
| 1006 |
{
|
| 1007 |
"epoch": 0.69,
|
| 1008 |
-
"eval_logits/chosen": -
|
| 1009 |
-
"eval_logits/rejected": -
|
| 1010 |
-
"eval_logps/chosen": -
|
| 1011 |
-
"eval_logps/rejected": -
|
| 1012 |
-
"eval_loss": 0.
|
| 1013 |
-
"eval_rewards/accuracies": 0.
|
| 1014 |
-
"eval_rewards/chosen":
|
| 1015 |
-
"eval_rewards/margins":
|
| 1016 |
-
"eval_rewards/rejected": -2.
|
| 1017 |
-
"eval_runtime":
|
| 1018 |
-
"eval_samples_per_second": 20.
|
| 1019 |
-
"eval_steps_per_second": 0.
|
| 1020 |
"step": 600
|
| 1021 |
},
|
| 1022 |
{
|
| 1023 |
"epoch": 0.7,
|
| 1024 |
-
"grad_norm":
|
| 1025 |
"learning_rate": 1.2297343017146726e-07,
|
| 1026 |
-
"logits/chosen":
|
| 1027 |
-
"logits/rejected": 1.
|
| 1028 |
-
"logps/chosen": -
|
| 1029 |
-
"logps/rejected": -
|
| 1030 |
-
"loss": 0.
|
| 1031 |
-
"rewards/accuracies": 0.
|
| 1032 |
-
"rewards/chosen": -1.
|
| 1033 |
-
"rewards/margins": 1.
|
| 1034 |
-
"rewards/rejected": -
|
| 1035 |
"step": 610
|
| 1036 |
},
|
| 1037 |
{
|
| 1038 |
"epoch": 0.71,
|
| 1039 |
-
"grad_norm":
|
| 1040 |
"learning_rate": 1.1441707378923474e-07,
|
| 1041 |
-
"logits/chosen":
|
| 1042 |
-
"logits/rejected": 1.
|
| 1043 |
-
"logps/chosen": -
|
| 1044 |
-
"logps/rejected": -
|
| 1045 |
-
"loss": 0.
|
| 1046 |
-
"rewards/accuracies": 0.
|
| 1047 |
-
"rewards/chosen": -
|
| 1048 |
-
"rewards/margins":
|
| 1049 |
-
"rewards/rejected": -
|
| 1050 |
"step": 620
|
| 1051 |
},
|
| 1052 |
{
|
| 1053 |
"epoch": 0.73,
|
| 1054 |
-
"grad_norm":
|
| 1055 |
"learning_rate": 1.06080070680377e-07,
|
| 1056 |
-
"logits/chosen":
|
| 1057 |
-
"logits/rejected": 1.
|
| 1058 |
-
"logps/chosen": -
|
| 1059 |
-
"logps/rejected": -
|
| 1060 |
-
"loss": 0.
|
| 1061 |
-
"rewards/accuracies": 0.
|
| 1062 |
-
"rewards/chosen": -
|
| 1063 |
-
"rewards/margins":
|
| 1064 |
-
"rewards/rejected": -
|
| 1065 |
"step": 630
|
| 1066 |
},
|
| 1067 |
{
|
| 1068 |
"epoch": 0.74,
|
| 1069 |
-
"grad_norm":
|
| 1070 |
"learning_rate": 9.797590889219587e-08,
|
| 1071 |
-
"logits/chosen":
|
| 1072 |
-
"logits/rejected":
|
| 1073 |
-
"logps/chosen": -
|
| 1074 |
-
"logps/rejected": -
|
| 1075 |
-
"loss": 0.
|
| 1076 |
"rewards/accuracies": 0.8374999761581421,
|
| 1077 |
-
"rewards/chosen": -
|
| 1078 |
-
"rewards/margins":
|
| 1079 |
-
"rewards/rejected": -
|
| 1080 |
"step": 640
|
| 1081 |
},
|
| 1082 |
{
|
| 1083 |
"epoch": 0.75,
|
| 1084 |
-
"grad_norm":
|
| 1085 |
"learning_rate": 9.011769976891367e-08,
|
| 1086 |
-
"logits/chosen":
|
| 1087 |
-
"logits/rejected":
|
| 1088 |
-
"logps/chosen": -
|
| 1089 |
-
"logps/rejected": -
|
| 1090 |
-
"loss": 0.
|
| 1091 |
-
"rewards/accuracies": 0.
|
| 1092 |
-
"rewards/chosen": -
|
| 1093 |
-
"rewards/margins": 1.
|
| 1094 |
-
"rewards/rejected": -
|
| 1095 |
"step": 650
|
| 1096 |
},
|
| 1097 |
{
|
| 1098 |
"epoch": 0.76,
|
| 1099 |
-
"grad_norm":
|
| 1100 |
"learning_rate": 8.251815673944218e-08,
|
| 1101 |
-
"logits/chosen":
|
| 1102 |
-
"logits/rejected":
|
| 1103 |
-
"logps/chosen": -
|
| 1104 |
-
"logps/rejected": -
|
| 1105 |
-
"loss": 0.
|
| 1106 |
-
"rewards/accuracies": 0.
|
| 1107 |
-
"rewards/chosen": -
|
| 1108 |
-
"rewards/margins": 1.
|
| 1109 |
-
"rewards/rejected": -3.
|
| 1110 |
"step": 660
|
| 1111 |
},
|
| 1112 |
{
|
| 1113 |
"epoch": 0.77,
|
| 1114 |
-
"grad_norm":
|
| 1115 |
"learning_rate": 7.518957474892148e-08,
|
| 1116 |
-
"logits/chosen":
|
| 1117 |
-
"logits/rejected": 1.
|
| 1118 |
-
"logps/chosen": -
|
| 1119 |
-
"logps/rejected": -
|
| 1120 |
-
"loss": 0.
|
| 1121 |
-
"rewards/accuracies": 0.
|
| 1122 |
-
"rewards/chosen": -
|
| 1123 |
-
"rewards/margins":
|
| 1124 |
-
"rewards/rejected": -
|
| 1125 |
"step": 670
|
| 1126 |
},
|
| 1127 |
{
|
| 1128 |
"epoch": 0.78,
|
| 1129 |
-
"grad_norm":
|
| 1130 |
"learning_rate": 6.814381036730274e-08,
|
| 1131 |
-
"logits/chosen":
|
| 1132 |
-
"logits/rejected":
|
| 1133 |
-
"logps/chosen": -
|
| 1134 |
-
"logps/rejected": -
|
| 1135 |
-
"loss": 0.
|
| 1136 |
-
"rewards/accuracies": 0.
|
| 1137 |
-
"rewards/chosen": -
|
| 1138 |
-
"rewards/margins":
|
| 1139 |
-
"rewards/rejected": -
|
| 1140 |
"step": 680
|
| 1141 |
},
|
| 1142 |
{
|
| 1143 |
"epoch": 0.79,
|
| 1144 |
-
"grad_norm":
|
| 1145 |
"learning_rate": 6.139226260715872e-08,
|
| 1146 |
-
"logits/chosen":
|
| 1147 |
-
"logits/rejected":
|
| 1148 |
-
"logps/chosen": -
|
| 1149 |
-
"logps/rejected": -
|
| 1150 |
-
"loss": 0.
|
| 1151 |
-
"rewards/accuracies": 0.
|
| 1152 |
-
"rewards/chosen": -
|
| 1153 |
-
"rewards/margins":
|
| 1154 |
-
"rewards/rejected": -
|
| 1155 |
"step": 690
|
| 1156 |
},
|
| 1157 |
{
|
| 1158 |
"epoch": 0.81,
|
| 1159 |
-
"grad_norm":
|
| 1160 |
"learning_rate": 5.4945854481754734e-08,
|
| 1161 |
-
"logits/chosen":
|
| 1162 |
-
"logits/rejected": 1.
|
| 1163 |
-
"logps/chosen": -
|
| 1164 |
-
"logps/rejected": -
|
| 1165 |
-
"loss": 0.
|
| 1166 |
-
"rewards/accuracies": 0.
|
| 1167 |
-
"rewards/chosen": -
|
| 1168 |
-
"rewards/margins":
|
| 1169 |
-
"rewards/rejected": -
|
| 1170 |
"step": 700
|
| 1171 |
},
|
| 1172 |
{
|
| 1173 |
"epoch": 0.81,
|
| 1174 |
-
"eval_logits/chosen": -
|
| 1175 |
-
"eval_logits/rejected": -
|
| 1176 |
-
"eval_logps/chosen": -
|
| 1177 |
-
"eval_logps/rejected": -
|
| 1178 |
-
"eval_loss": 0.
|
| 1179 |
-
"eval_rewards/accuracies": 0.
|
| 1180 |
-
"eval_rewards/chosen":
|
| 1181 |
-
"eval_rewards/margins":
|
| 1182 |
-
"eval_rewards/rejected": -2.
|
| 1183 |
-
"eval_runtime": 98.
|
| 1184 |
-
"eval_samples_per_second": 20.
|
| 1185 |
"eval_steps_per_second": 0.326,
|
| 1186 |
"step": 700
|
| 1187 |
},
|
| 1188 |
{
|
| 1189 |
"epoch": 0.82,
|
| 1190 |
-
"grad_norm":
|
| 1191 |
"learning_rate": 4.881501533321605e-08,
|
| 1192 |
-
"logits/chosen":
|
| 1193 |
-
"logits/rejected": 1.
|
| 1194 |
-
"logps/chosen": -
|
| 1195 |
-
"logps/rejected": -
|
| 1196 |
-
"loss": 0.
|
| 1197 |
-
"rewards/accuracies": 0.
|
| 1198 |
-
"rewards/chosen": -
|
| 1199 |
-
"rewards/margins":
|
| 1200 |
-
"rewards/rejected": -3.
|
| 1201 |
"step": 710
|
| 1202 |
},
|
| 1203 |
{
|
| 1204 |
"epoch": 0.83,
|
| 1205 |
-
"grad_norm":
|
| 1206 |
"learning_rate": 4.300966395938377e-08,
|
| 1207 |
-
"logits/chosen":
|
| 1208 |
-
"logits/rejected":
|
| 1209 |
-
"logps/chosen": -
|
| 1210 |
-
"logps/rejected": -
|
| 1211 |
-
"loss": 0.
|
| 1212 |
-
"rewards/accuracies": 0.
|
| 1213 |
-
"rewards/chosen": -
|
| 1214 |
-
"rewards/margins":
|
| 1215 |
-
"rewards/rejected": -3.
|
| 1216 |
"step": 720
|
| 1217 |
},
|
| 1218 |
{
|
| 1219 |
"epoch": 0.84,
|
| 1220 |
-
"grad_norm":
|
| 1221 |
"learning_rate": 3.7539192566655246e-08,
|
| 1222 |
-
"logits/chosen":
|
| 1223 |
-
"logits/rejected":
|
| 1224 |
-
"logps/chosen": -
|
| 1225 |
-
"logps/rejected": -
|
| 1226 |
-
"loss": 0.
|
| 1227 |
-
"rewards/accuracies": 0.
|
| 1228 |
-
"rewards/chosen": -
|
| 1229 |
-
"rewards/margins": 1.
|
| 1230 |
-
"rewards/rejected": -
|
| 1231 |
"step": 730
|
| 1232 |
},
|
| 1233 |
{
|
| 1234 |
"epoch": 0.85,
|
| 1235 |
-
"grad_norm":
|
| 1236 |
"learning_rate": 3.24124515747731e-08,
|
| 1237 |
-
"logits/chosen":
|
| 1238 |
-
"logits/rejected":
|
| 1239 |
-
"logps/chosen": -
|
| 1240 |
-
"logps/rejected": -
|
| 1241 |
-
"loss": 0.
|
| 1242 |
-
"rewards/accuracies": 0.
|
| 1243 |
-
"rewards/chosen": -
|
| 1244 |
-
"rewards/margins": 1.
|
| 1245 |
-
"rewards/rejected": -
|
| 1246 |
"step": 740
|
| 1247 |
},
|
| 1248 |
{
|
| 1249 |
"epoch": 0.86,
|
| 1250 |
-
"grad_norm":
|
| 1251 |
"learning_rate": 2.763773529814506e-08,
|
| 1252 |
-
"logits/chosen":
|
| 1253 |
-
"logits/rejected":
|
| 1254 |
-
"logps/chosen": -
|
| 1255 |
-
"logps/rejected": -
|
| 1256 |
-
"loss": 0.
|
| 1257 |
-
"rewards/accuracies": 0.
|
| 1258 |
-
"rewards/chosen": -
|
| 1259 |
-
"rewards/margins":
|
| 1260 |
-
"rewards/rejected": -
|
| 1261 |
"step": 750
|
| 1262 |
},
|
| 1263 |
{
|
| 1264 |
"epoch": 0.88,
|
| 1265 |
-
"grad_norm":
|
| 1266 |
"learning_rate": 2.3222768526860698e-08,
|
| 1267 |
-
"logits/chosen":
|
| 1268 |
-
"logits/rejected":
|
| 1269 |
-
"logps/chosen": -
|
| 1270 |
-
"logps/rejected": -
|
| 1271 |
-
"loss": 0.
|
| 1272 |
-
"rewards/accuracies": 0.
|
| 1273 |
-
"rewards/chosen": -
|
| 1274 |
-
"rewards/margins":
|
| 1275 |
-
"rewards/rejected": -
|
| 1276 |
"step": 760
|
| 1277 |
},
|
| 1278 |
{
|
| 1279 |
"epoch": 0.89,
|
| 1280 |
-
"grad_norm":
|
| 1281 |
"learning_rate": 1.9174694029115146e-08,
|
| 1282 |
-
"logits/chosen":
|
| 1283 |
-
"logits/rejected":
|
| 1284 |
-
"logps/chosen": -
|
| 1285 |
-
"logps/rejected": -
|
| 1286 |
-
"loss": 0.
|
| 1287 |
-
"rewards/accuracies": 0.
|
| 1288 |
-
"rewards/chosen": -
|
| 1289 |
-
"rewards/margins": 1.
|
| 1290 |
-
"rewards/rejected": -
|
| 1291 |
"step": 770
|
| 1292 |
},
|
| 1293 |
{
|
| 1294 |
"epoch": 0.9,
|
| 1295 |
-
"grad_norm":
|
| 1296 |
"learning_rate": 1.5500060995258134e-08,
|
| 1297 |
-
"logits/chosen":
|
| 1298 |
-
"logits/rejected":
|
| 1299 |
-
"logps/chosen": -
|
| 1300 |
-
"logps/rejected": -
|
| 1301 |
-
"loss": 0.
|
| 1302 |
-
"rewards/accuracies": 0.
|
| 1303 |
-
"rewards/chosen": -
|
| 1304 |
-
"rewards/margins":
|
| 1305 |
-
"rewards/rejected": -
|
| 1306 |
"step": 780
|
| 1307 |
},
|
| 1308 |
{
|
| 1309 |
"epoch": 0.91,
|
| 1310 |
-
"grad_norm":
|
| 1311 |
"learning_rate": 1.2204814442165812e-08,
|
| 1312 |
-
"logits/chosen":
|
| 1313 |
-
"logits/rejected":
|
| 1314 |
-
"logps/chosen": -
|
| 1315 |
-
"logps/rejected": -
|
| 1316 |
-
"loss": 0.
|
| 1317 |
-
"rewards/accuracies": 0.
|
| 1318 |
-
"rewards/chosen": -
|
| 1319 |
-
"rewards/margins":
|
| 1320 |
-
"rewards/rejected": -3.
|
| 1321 |
"step": 790
|
| 1322 |
},
|
| 1323 |
{
|
| 1324 |
"epoch": 0.92,
|
| 1325 |
-
"grad_norm":
|
| 1326 |
"learning_rate": 9.294285595075669e-09,
|
| 1327 |
-
"logits/chosen":
|
| 1328 |
-
"logits/rejected":
|
| 1329 |
-
"logps/chosen": -
|
| 1330 |
-
"logps/rejected": -
|
| 1331 |
-
"loss": 0.
|
| 1332 |
-
"rewards/accuracies": 0.
|
| 1333 |
-
"rewards/chosen": -
|
| 1334 |
-
"rewards/margins":
|
| 1335 |
-
"rewards/rejected": -
|
| 1336 |
"step": 800
|
| 1337 |
},
|
| 1338 |
{
|
| 1339 |
"epoch": 0.92,
|
| 1340 |
-
"eval_logits/chosen": -
|
| 1341 |
-
"eval_logits/rejected": -
|
| 1342 |
-
"eval_logps/chosen": -
|
| 1343 |
-
"eval_logps/rejected": -
|
| 1344 |
-
"eval_loss": 0.
|
| 1345 |
-
"eval_rewards/accuracies": 0.
|
| 1346 |
-
"eval_rewards/chosen":
|
| 1347 |
-
"eval_rewards/margins":
|
| 1348 |
-
"eval_rewards/rejected": -2.
|
| 1349 |
-
"eval_runtime":
|
| 1350 |
-
"eval_samples_per_second": 20.
|
| 1351 |
-
"eval_steps_per_second": 0.
|
| 1352 |
"step": 800
|
| 1353 |
},
|
| 1354 |
{
|
| 1355 |
"epoch": 0.93,
|
| 1356 |
-
"grad_norm":
|
| 1357 |
"learning_rate": 6.773183262446914e-09,
|
| 1358 |
-
"logits/chosen":
|
| 1359 |
-
"logits/rejected":
|
| 1360 |
-
"logps/chosen": -
|
| 1361 |
-
"logps/rejected": -
|
| 1362 |
-
"loss": 0.
|
| 1363 |
-
"rewards/accuracies": 0.
|
| 1364 |
-
"rewards/chosen": -
|
| 1365 |
-
"rewards/margins":
|
| 1366 |
-
"rewards/rejected": -
|
| 1367 |
"step": 810
|
| 1368 |
},
|
| 1369 |
{
|
| 1370 |
"epoch": 0.94,
|
| 1371 |
-
"grad_norm":
|
| 1372 |
"learning_rate": 4.645586217799452e-09,
|
| 1373 |
-
"logits/chosen":
|
| 1374 |
-
"logits/rejected":
|
| 1375 |
-
"logps/chosen": -
|
| 1376 |
-
"logps/rejected": -
|
| 1377 |
-
"loss": 0.
|
| 1378 |
-
"rewards/accuracies": 0.
|
| 1379 |
-
"rewards/chosen": -
|
| 1380 |
-
"rewards/margins":
|
| 1381 |
-
"rewards/rejected": -
|
| 1382 |
"step": 820
|
| 1383 |
},
|
| 1384 |
{
|
| 1385 |
"epoch": 0.96,
|
| 1386 |
-
"grad_norm":
|
| 1387 |
"learning_rate": 2.9149366008568987e-09,
|
| 1388 |
-
"logits/chosen":
|
| 1389 |
-
"logits/rejected":
|
| 1390 |
-
"logps/chosen": -
|
| 1391 |
-
"logps/rejected": -
|
| 1392 |
-
"loss": 0.
|
| 1393 |
-
"rewards/accuracies": 0.
|
| 1394 |
-
"rewards/chosen": -
|
| 1395 |
-
"rewards/margins":
|
| 1396 |
-
"rewards/rejected": -
|
| 1397 |
"step": 830
|
| 1398 |
},
|
| 1399 |
{
|
| 1400 |
"epoch": 0.97,
|
| 1401 |
-
"grad_norm":
|
| 1402 |
"learning_rate": 1.5840343486700215e-09,
|
| 1403 |
-
"logits/chosen":
|
| 1404 |
-
"logits/rejected":
|
| 1405 |
-
"logps/chosen": -
|
| 1406 |
-
"logps/rejected": -
|
| 1407 |
-
"loss": 0.
|
| 1408 |
-
"rewards/accuracies": 0.
|
| 1409 |
-
"rewards/chosen": -
|
| 1410 |
-
"rewards/margins":
|
| 1411 |
-
"rewards/rejected": -
|
| 1412 |
"step": 840
|
| 1413 |
},
|
| 1414 |
{
|
| 1415 |
"epoch": 0.98,
|
| 1416 |
-
"grad_norm":
|
| 1417 |
"learning_rate": 6.550326657293881e-10,
|
| 1418 |
-
"logits/chosen":
|
| 1419 |
-
"logits/rejected":
|
| 1420 |
-
"logps/chosen": -
|
| 1421 |
-
"logps/rejected": -
|
| 1422 |
-
"loss": 0.
|
| 1423 |
-
"rewards/accuracies": 0.
|
| 1424 |
-
"rewards/chosen": -
|
| 1425 |
-
"rewards/margins": 2.
|
| 1426 |
-
"rewards/rejected": -3.
|
| 1427 |
"step": 850
|
| 1428 |
},
|
| 1429 |
{
|
| 1430 |
"epoch": 0.99,
|
| 1431 |
-
"grad_norm":
|
| 1432 |
"learning_rate": 1.2943454039654467e-10,
|
| 1433 |
-
"logits/chosen":
|
| 1434 |
-
"logits/rejected":
|
| 1435 |
-
"logps/chosen": -
|
| 1436 |
-
"logps/rejected": -
|
| 1437 |
-
"loss": 0.
|
| 1438 |
-
"rewards/accuracies": 0.
|
| 1439 |
-
"rewards/chosen": -
|
| 1440 |
-
"rewards/margins": 1.
|
| 1441 |
-
"rewards/rejected": -
|
| 1442 |
"step": 860
|
| 1443 |
},
|
| 1444 |
{
|
| 1445 |
"epoch": 1.0,
|
| 1446 |
"step": 868,
|
| 1447 |
"total_flos": 0.0,
|
| 1448 |
-
"train_loss": 0.
|
| 1449 |
-
"train_runtime":
|
| 1450 |
-
"train_samples_per_second":
|
| 1451 |
-
"train_steps_per_second": 0.
|
| 1452 |
}
|
| 1453 |
],
|
| 1454 |
"logging_steps": 10,
|
|
|
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.0,
|
| 13 |
+
"grad_norm": 146.8957421194674,
|
| 14 |
"learning_rate": 5.747126436781609e-09,
|
| 15 |
+
"logits/chosen": -1.8686045408248901,
|
| 16 |
+
"logits/rejected": -1.7644572257995605,
|
| 17 |
+
"logps/chosen": -235.48362731933594,
|
| 18 |
+
"logps/rejected": -183.77415466308594,
|
| 19 |
+
"loss": 0.6922,
|
| 20 |
"rewards/accuracies": 0.0,
|
| 21 |
"rewards/chosen": 0.0,
|
| 22 |
"rewards/margins": 0.0,
|
|
|
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.01,
|
| 28 |
+
"grad_norm": 237.04909600464902,
|
| 29 |
"learning_rate": 5.747126436781609e-08,
|
| 30 |
+
"logits/chosen": -1.9218311309814453,
|
| 31 |
+
"logits/rejected": -1.8686226606369019,
|
| 32 |
+
"logps/chosen": -240.50628662109375,
|
| 33 |
+
"logps/rejected": -216.8230438232422,
|
| 34 |
+
"loss": 0.6941,
|
| 35 |
+
"rewards/accuracies": 0.4444444477558136,
|
| 36 |
+
"rewards/chosen": -0.002868607407435775,
|
| 37 |
+
"rewards/margins": -0.0005126786418259144,
|
| 38 |
+
"rewards/rejected": -0.002355928998440504,
|
| 39 |
"step": 10
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.02,
|
| 43 |
+
"grad_norm": 211.82620683349913,
|
| 44 |
"learning_rate": 1.1494252873563217e-07,
|
| 45 |
+
"logits/chosen": -2.010253429412842,
|
| 46 |
+
"logits/rejected": -1.8642921447753906,
|
| 47 |
+
"logps/chosen": -283.1783752441406,
|
| 48 |
+
"logps/rejected": -215.68887329101562,
|
| 49 |
+
"loss": 0.6864,
|
| 50 |
+
"rewards/accuracies": 0.612500011920929,
|
| 51 |
+
"rewards/chosen": 0.008166970685124397,
|
| 52 |
+
"rewards/margins": 0.013039084151387215,
|
| 53 |
+
"rewards/rejected": -0.004872114397585392,
|
| 54 |
"step": 20
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 0.03,
|
| 58 |
+
"grad_norm": 112.97267570781544,
|
| 59 |
"learning_rate": 1.7241379310344828e-07,
|
| 60 |
+
"logits/chosen": -1.9509646892547607,
|
| 61 |
+
"logits/rejected": -1.8835735321044922,
|
| 62 |
+
"logps/chosen": -240.29074096679688,
|
| 63 |
+
"logps/rejected": -221.15274047851562,
|
| 64 |
+
"loss": 0.666,
|
| 65 |
+
"rewards/accuracies": 0.668749988079071,
|
| 66 |
+
"rewards/chosen": 0.06300461292266846,
|
| 67 |
+
"rewards/margins": 0.06308884918689728,
|
| 68 |
+
"rewards/rejected": -8.423496183240786e-05,
|
| 69 |
"step": 30
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.05,
|
| 73 |
+
"grad_norm": 95.74241597637995,
|
| 74 |
"learning_rate": 2.2988505747126435e-07,
|
| 75 |
+
"logits/chosen": -1.9107061624526978,
|
| 76 |
+
"logits/rejected": -1.8901317119598389,
|
| 77 |
+
"logps/chosen": -237.59036254882812,
|
| 78 |
+
"logps/rejected": -216.1823272705078,
|
| 79 |
+
"loss": 0.6254,
|
| 80 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 81 |
+
"rewards/chosen": 0.1618741899728775,
|
| 82 |
+
"rewards/margins": 0.1895591914653778,
|
| 83 |
+
"rewards/rejected": -0.027684981003403664,
|
| 84 |
"step": 40
|
| 85 |
},
|
| 86 |
{
|
| 87 |
"epoch": 0.06,
|
| 88 |
+
"grad_norm": 99.41131976874209,
|
| 89 |
"learning_rate": 2.873563218390804e-07,
|
| 90 |
+
"logits/chosen": -1.9803855419158936,
|
| 91 |
+
"logits/rejected": -1.9117343425750732,
|
| 92 |
+
"logps/chosen": -222.8354949951172,
|
| 93 |
+
"logps/rejected": -207.8356170654297,
|
| 94 |
+
"loss": 0.587,
|
| 95 |
+
"rewards/accuracies": 0.768750011920929,
|
| 96 |
+
"rewards/chosen": 0.23140163719654083,
|
| 97 |
+
"rewards/margins": 0.3971676230430603,
|
| 98 |
+
"rewards/rejected": -0.16576598584651947,
|
| 99 |
"step": 50
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.07,
|
| 103 |
+
"grad_norm": 85.607755330573,
|
| 104 |
"learning_rate": 3.4482758620689656e-07,
|
| 105 |
+
"logits/chosen": -2.0026185512542725,
|
| 106 |
+
"logits/rejected": -1.9393552541732788,
|
| 107 |
+
"logps/chosen": -283.61199951171875,
|
| 108 |
+
"logps/rejected": -245.7741241455078,
|
| 109 |
+
"loss": 0.5351,
|
| 110 |
+
"rewards/accuracies": 0.793749988079071,
|
| 111 |
+
"rewards/chosen": 0.2226170301437378,
|
| 112 |
+
"rewards/margins": 0.6314491629600525,
|
| 113 |
+
"rewards/rejected": -0.4088321626186371,
|
| 114 |
"step": 60
|
| 115 |
},
|
| 116 |
{
|
| 117 |
"epoch": 0.08,
|
| 118 |
+
"grad_norm": 105.09596276450131,
|
| 119 |
"learning_rate": 4.0229885057471266e-07,
|
| 120 |
+
"logits/chosen": -1.9401233196258545,
|
| 121 |
+
"logits/rejected": -1.9097837209701538,
|
| 122 |
+
"logps/chosen": -223.41983032226562,
|
| 123 |
+
"logps/rejected": -222.15283203125,
|
| 124 |
+
"loss": 0.5181,
|
| 125 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 126 |
+
"rewards/chosen": -0.10973484814167023,
|
| 127 |
+
"rewards/margins": 0.7763268351554871,
|
| 128 |
+
"rewards/rejected": -0.8860616683959961,
|
| 129 |
"step": 70
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.09,
|
| 133 |
+
"grad_norm": 103.35591967882027,
|
| 134 |
"learning_rate": 4.597701149425287e-07,
|
| 135 |
+
"logits/chosen": -1.8168106079101562,
|
| 136 |
+
"logits/rejected": -1.7877649068832397,
|
| 137 |
+
"logps/chosen": -227.1862030029297,
|
| 138 |
+
"logps/rejected": -217.2149200439453,
|
| 139 |
+
"loss": 0.487,
|
| 140 |
+
"rewards/accuracies": 0.824999988079071,
|
| 141 |
+
"rewards/chosen": -0.1610104888677597,
|
| 142 |
+
"rewards/margins": 0.8937110900878906,
|
| 143 |
+
"rewards/rejected": -1.0547215938568115,
|
| 144 |
"step": 80
|
| 145 |
},
|
| 146 |
{
|
| 147 |
"epoch": 0.1,
|
| 148 |
+
"grad_norm": 96.75574881093117,
|
| 149 |
"learning_rate": 4.999817969178237e-07,
|
| 150 |
+
"logits/chosen": -1.8949458599090576,
|
| 151 |
+
"logits/rejected": -1.8469880819320679,
|
| 152 |
+
"logps/chosen": -249.9053192138672,
|
| 153 |
+
"logps/rejected": -233.94631958007812,
|
| 154 |
+
"loss": 0.4415,
|
| 155 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 156 |
+
"rewards/chosen": -0.24459132552146912,
|
| 157 |
+
"rewards/margins": 1.046515703201294,
|
| 158 |
+
"rewards/rejected": -1.291106939315796,
|
| 159 |
"step": 90
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 0.12,
|
| 163 |
+
"grad_norm": 92.71145741103783,
|
| 164 |
"learning_rate": 4.996582603056428e-07,
|
| 165 |
+
"logits/chosen": -1.8467813730239868,
|
| 166 |
+
"logits/rejected": -1.7593021392822266,
|
| 167 |
+
"logps/chosen": -251.5709686279297,
|
| 168 |
+
"logps/rejected": -255.50814819335938,
|
| 169 |
+
"loss": 0.4898,
|
| 170 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 171 |
+
"rewards/chosen": -0.3932832181453705,
|
| 172 |
+
"rewards/margins": 1.1354423761367798,
|
| 173 |
+
"rewards/rejected": -1.5287256240844727,
|
| 174 |
"step": 100
|
| 175 |
},
|
| 176 |
{
|
| 177 |
"epoch": 0.12,
|
| 178 |
+
"eval_logits/chosen": -1.8469043970108032,
|
| 179 |
+
"eval_logits/rejected": -1.7659016847610474,
|
| 180 |
+
"eval_logps/chosen": -339.4445495605469,
|
| 181 |
+
"eval_logps/rejected": -353.2087707519531,
|
| 182 |
+
"eval_loss": 0.5505225658416748,
|
| 183 |
+
"eval_rewards/accuracies": 0.6875,
|
| 184 |
+
"eval_rewards/chosen": -0.19665074348449707,
|
| 185 |
+
"eval_rewards/margins": 0.8084924817085266,
|
| 186 |
+
"eval_rewards/rejected": -1.005143165588379,
|
| 187 |
+
"eval_runtime": 98.3147,
|
| 188 |
+
"eval_samples_per_second": 20.343,
|
| 189 |
+
"eval_steps_per_second": 0.325,
|
| 190 |
"step": 100
|
| 191 |
},
|
| 192 |
{
|
| 193 |
"epoch": 0.13,
|
| 194 |
+
"grad_norm": 108.17707201404545,
|
| 195 |
"learning_rate": 4.989308132738126e-07,
|
| 196 |
+
"logits/chosen": -1.8153671026229858,
|
| 197 |
+
"logits/rejected": -1.6921818256378174,
|
| 198 |
+
"logps/chosen": -244.46237182617188,
|
| 199 |
+
"logps/rejected": -222.83145141601562,
|
| 200 |
+
"loss": 0.4368,
|
| 201 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 202 |
+
"rewards/chosen": -0.10259479284286499,
|
| 203 |
+
"rewards/margins": 1.188932180404663,
|
| 204 |
+
"rewards/rejected": -1.2915267944335938,
|
| 205 |
"step": 110
|
| 206 |
},
|
| 207 |
{
|
| 208 |
"epoch": 0.14,
|
| 209 |
+
"grad_norm": 100.40422750465044,
|
| 210 |
"learning_rate": 4.978006327248536e-07,
|
| 211 |
+
"logits/chosen": -1.7503843307495117,
|
| 212 |
+
"logits/rejected": -1.6768817901611328,
|
| 213 |
+
"logps/chosen": -247.3853759765625,
|
| 214 |
+
"logps/rejected": -240.05496215820312,
|
| 215 |
+
"loss": 0.4237,
|
| 216 |
+
"rewards/accuracies": 0.8125,
|
| 217 |
+
"rewards/chosen": -0.34009289741516113,
|
| 218 |
+
"rewards/margins": 1.1859456300735474,
|
| 219 |
+
"rewards/rejected": -1.526038408279419,
|
| 220 |
"step": 120
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 0.15,
|
| 224 |
+
"grad_norm": 91.1070676572417,
|
| 225 |
"learning_rate": 4.962695471250032e-07,
|
| 226 |
+
"logits/chosen": -1.654911756515503,
|
| 227 |
+
"logits/rejected": -1.62875235080719,
|
| 228 |
+
"logps/chosen": -254.4684600830078,
|
| 229 |
+
"logps/rejected": -245.8209686279297,
|
| 230 |
+
"loss": 0.4249,
|
| 231 |
+
"rewards/accuracies": 0.8125,
|
| 232 |
+
"rewards/chosen": -0.5070122480392456,
|
| 233 |
+
"rewards/margins": 1.3276703357696533,
|
| 234 |
+
"rewards/rejected": -1.8346824645996094,
|
| 235 |
"step": 130
|
| 236 |
},
|
| 237 |
{
|
| 238 |
"epoch": 0.16,
|
| 239 |
+
"grad_norm": 97.14051346245,
|
| 240 |
"learning_rate": 4.94340033546025e-07,
|
| 241 |
+
"logits/chosen": -1.6183685064315796,
|
| 242 |
+
"logits/rejected": -1.6242666244506836,
|
| 243 |
+
"logps/chosen": -220.20852661132812,
|
| 244 |
+
"logps/rejected": -242.1243133544922,
|
| 245 |
+
"loss": 0.4218,
|
| 246 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 247 |
+
"rewards/chosen": -0.6856725811958313,
|
| 248 |
+
"rewards/margins": 1.2277655601501465,
|
| 249 |
+
"rewards/rejected": -1.913438081741333,
|
| 250 |
"step": 140
|
| 251 |
},
|
| 252 |
{
|
| 253 |
"epoch": 0.17,
|
| 254 |
+
"grad_norm": 94.8408241800107,
|
| 255 |
"learning_rate": 4.920152136576705e-07,
|
| 256 |
+
"logits/chosen": -1.4978981018066406,
|
| 257 |
+
"logits/rejected": -1.5163871049880981,
|
| 258 |
+
"logps/chosen": -251.6667938232422,
|
| 259 |
+
"logps/rejected": -254.0281219482422,
|
| 260 |
+
"loss": 0.4261,
|
| 261 |
+
"rewards/accuracies": 0.824999988079071,
|
| 262 |
+
"rewards/chosen": -0.6180266737937927,
|
| 263 |
+
"rewards/margins": 1.381075143814087,
|
| 264 |
+
"rewards/rejected": -1.9991016387939453,
|
| 265 |
"step": 150
|
| 266 |
},
|
| 267 |
{
|
| 268 |
"epoch": 0.18,
|
| 269 |
+
"grad_norm": 108.14894780962342,
|
| 270 |
"learning_rate": 4.892988486772756e-07,
|
| 271 |
+
"logits/chosen": -1.579886794090271,
|
| 272 |
+
"logits/rejected": -1.5718797445297241,
|
| 273 |
+
"logps/chosen": -250.04495239257812,
|
| 274 |
+
"logps/rejected": -262.3863830566406,
|
| 275 |
+
"loss": 0.4246,
|
| 276 |
+
"rewards/accuracies": 0.824999988079071,
|
| 277 |
+
"rewards/chosen": -0.6856819987297058,
|
| 278 |
+
"rewards/margins": 1.5999362468719482,
|
| 279 |
+
"rewards/rejected": -2.285618305206299,
|
| 280 |
"step": 160
|
| 281 |
},
|
| 282 |
{
|
| 283 |
"epoch": 0.2,
|
| 284 |
+
"grad_norm": 70.9507948612333,
|
| 285 |
"learning_rate": 4.861953332846629e-07,
|
| 286 |
+
"logits/chosen": -1.7066303491592407,
|
| 287 |
+
"logits/rejected": -1.6715686321258545,
|
| 288 |
+
"logps/chosen": -270.17132568359375,
|
| 289 |
+
"logps/rejected": -260.03350830078125,
|
| 290 |
+
"loss": 0.4316,
|
| 291 |
"rewards/accuracies": 0.7875000238418579,
|
| 292 |
+
"rewards/chosen": -0.8634172677993774,
|
| 293 |
+
"rewards/margins": 1.3165854215621948,
|
| 294 |
+
"rewards/rejected": -2.1800026893615723,
|
| 295 |
"step": 170
|
| 296 |
},
|
| 297 |
{
|
| 298 |
"epoch": 0.21,
|
| 299 |
+
"grad_norm": 114.85174445864124,
|
| 300 |
"learning_rate": 4.827096885121953e-07,
|
| 301 |
+
"logits/chosen": -1.817678689956665,
|
| 302 |
+
"logits/rejected": -1.7374699115753174,
|
| 303 |
+
"logps/chosen": -286.0107116699219,
|
| 304 |
+
"logps/rejected": -279.7050476074219,
|
| 305 |
+
"loss": 0.4309,
|
| 306 |
+
"rewards/accuracies": 0.75,
|
| 307 |
+
"rewards/chosen": -1.0613857507705688,
|
| 308 |
+
"rewards/margins": 1.1405597925186157,
|
| 309 |
+
"rewards/rejected": -2.2019453048706055,
|
| 310 |
"step": 180
|
| 311 |
},
|
| 312 |
{
|
| 313 |
"epoch": 0.22,
|
| 314 |
+
"grad_norm": 90.50728576001431,
|
| 315 |
"learning_rate": 4.788475536214821e-07,
|
| 316 |
+
"logits/chosen": -1.7829034328460693,
|
| 317 |
+
"logits/rejected": -1.7400600910186768,
|
| 318 |
+
"logps/chosen": -236.38650512695312,
|
| 319 |
+
"logps/rejected": -245.64291381835938,
|
| 320 |
+
"loss": 0.3873,
|
| 321 |
+
"rewards/accuracies": 0.84375,
|
| 322 |
+
"rewards/chosen": -0.8258365392684937,
|
| 323 |
+
"rewards/margins": 1.4504865407943726,
|
| 324 |
+
"rewards/rejected": -2.276322841644287,
|
| 325 |
"step": 190
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 0.23,
|
| 329 |
+
"grad_norm": 100.41476084088063,
|
| 330 |
"learning_rate": 4.746151769798818e-07,
|
| 331 |
+
"logits/chosen": -1.7743650674819946,
|
| 332 |
+
"logits/rejected": -1.7203725576400757,
|
| 333 |
+
"logps/chosen": -279.8271789550781,
|
| 334 |
+
"logps/rejected": -269.1463317871094,
|
| 335 |
+
"loss": 0.4277,
|
| 336 |
+
"rewards/accuracies": 0.831250011920929,
|
| 337 |
+
"rewards/chosen": -0.8149574398994446,
|
| 338 |
+
"rewards/margins": 1.5317128896713257,
|
| 339 |
+
"rewards/rejected": -2.346670389175415,
|
| 340 |
"step": 200
|
| 341 |
},
|
| 342 |
{
|
| 343 |
"epoch": 0.23,
|
| 344 |
+
"eval_logits/chosen": -1.8008995056152344,
|
| 345 |
+
"eval_logits/rejected": -1.7248116731643677,
|
| 346 |
+
"eval_logps/chosen": -345.1794738769531,
|
| 347 |
+
"eval_logps/rejected": -370.77880859375,
|
| 348 |
+
"eval_loss": 0.46549829840660095,
|
| 349 |
+
"eval_rewards/accuracies": 0.73828125,
|
| 350 |
+
"eval_rewards/chosen": -0.4833980202674866,
|
| 351 |
+
"eval_rewards/margins": 1.4002480506896973,
|
| 352 |
+
"eval_rewards/rejected": -1.8836462497711182,
|
| 353 |
+
"eval_runtime": 98.0211,
|
| 354 |
+
"eval_samples_per_second": 20.404,
|
| 355 |
"eval_steps_per_second": 0.326,
|
| 356 |
"step": 200
|
| 357 |
},
|
| 358 |
{
|
| 359 |
"epoch": 0.24,
|
| 360 |
+
"grad_norm": 99.20354803224151,
|
| 361 |
"learning_rate": 4.7001940595156055e-07,
|
| 362 |
+
"logits/chosen": -1.7606436014175415,
|
| 363 |
+
"logits/rejected": -1.6915203332901,
|
| 364 |
+
"logps/chosen": -232.93832397460938,
|
| 365 |
+
"logps/rejected": -246.24072265625,
|
| 366 |
+
"loss": 0.441,
|
| 367 |
"rewards/accuracies": 0.8062499761581421,
|
| 368 |
+
"rewards/chosen": -1.0429778099060059,
|
| 369 |
+
"rewards/margins": 1.6135696172714233,
|
| 370 |
+
"rewards/rejected": -2.6565470695495605,
|
| 371 |
"step": 210
|
| 372 |
},
|
| 373 |
{
|
| 374 |
"epoch": 0.25,
|
| 375 |
+
"grad_norm": 80.49166768353011,
|
| 376 |
"learning_rate": 4.650676758194623e-07,
|
| 377 |
+
"logits/chosen": -1.820059061050415,
|
| 378 |
+
"logits/rejected": -1.741408348083496,
|
| 379 |
+
"logps/chosen": -265.69769287109375,
|
| 380 |
+
"logps/rejected": -259.860595703125,
|
| 381 |
+
"loss": 0.4003,
|
| 382 |
+
"rewards/accuracies": 0.793749988079071,
|
| 383 |
+
"rewards/chosen": -0.915117621421814,
|
| 384 |
+
"rewards/margins": 1.692671537399292,
|
| 385 |
+
"rewards/rejected": -2.6077892780303955,
|
| 386 |
"step": 220
|
| 387 |
},
|
| 388 |
{
|
| 389 |
"epoch": 0.26,
|
| 390 |
+
"grad_norm": 103.96889607439884,
|
| 391 |
"learning_rate": 4.5976799775611215e-07,
|
| 392 |
+
"logits/chosen": -1.84983229637146,
|
| 393 |
+
"logits/rejected": -1.7714653015136719,
|
| 394 |
+
"logps/chosen": -260.39678955078125,
|
| 395 |
+
"logps/rejected": -254.94235229492188,
|
| 396 |
+
"loss": 0.4118,
|
| 397 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 398 |
+
"rewards/chosen": -1.0704491138458252,
|
| 399 |
+
"rewards/margins": 1.982627511024475,
|
| 400 |
+
"rewards/rejected": -3.05307674407959,
|
| 401 |
"step": 230
|
| 402 |
},
|
| 403 |
{
|
| 404 |
"epoch": 0.28,
|
| 405 |
+
"grad_norm": 87.59607388393054,
|
| 406 |
"learning_rate": 4.5412894586271543e-07,
|
| 407 |
+
"logits/chosen": -1.8541374206542969,
|
| 408 |
+
"logits/rejected": -1.7867300510406494,
|
| 409 |
+
"logps/chosen": -259.6327819824219,
|
| 410 |
+
"logps/rejected": -233.4642333984375,
|
| 411 |
+
"loss": 0.3961,
|
| 412 |
+
"rewards/accuracies": 0.831250011920929,
|
| 413 |
+
"rewards/chosen": -0.8504000902175903,
|
| 414 |
+
"rewards/margins": 1.6259987354278564,
|
| 415 |
+
"rewards/rejected": -2.4763987064361572,
|
| 416 |
"step": 240
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 0.29,
|
| 420 |
+
"grad_norm": 87.76098428515671,
|
| 421 |
"learning_rate": 4.481596432975201e-07,
|
| 422 |
+
"logits/chosen": -1.8068411350250244,
|
| 423 |
+
"logits/rejected": -1.7720015048980713,
|
| 424 |
+
"logps/chosen": -218.69546508789062,
|
| 425 |
+
"logps/rejected": -231.09292602539062,
|
| 426 |
+
"loss": 0.4236,
|
| 427 |
+
"rewards/accuracies": 0.762499988079071,
|
| 428 |
+
"rewards/chosen": -0.8816181421279907,
|
| 429 |
+
"rewards/margins": 1.4680168628692627,
|
| 430 |
+
"rewards/rejected": -2.349634885787964,
|
| 431 |
"step": 250
|
| 432 |
},
|
| 433 |
{
|
| 434 |
"epoch": 0.3,
|
| 435 |
+
"grad_norm": 99.04429789846779,
|
| 436 |
"learning_rate": 4.41869747515886e-07,
|
| 437 |
+
"logits/chosen": -1.7596534490585327,
|
| 438 |
+
"logits/rejected": -1.7420539855957031,
|
| 439 |
+
"logps/chosen": -272.49273681640625,
|
| 440 |
+
"logps/rejected": -301.8647766113281,
|
| 441 |
+
"loss": 0.3824,
|
| 442 |
+
"rewards/accuracies": 0.8125,
|
| 443 |
+
"rewards/chosen": -1.0995886325836182,
|
| 444 |
+
"rewards/margins": 1.7163059711456299,
|
| 445 |
+
"rewards/rejected": -2.815894603729248,
|
| 446 |
"step": 260
|
| 447 |
},
|
| 448 |
{
|
| 449 |
"epoch": 0.31,
|
| 450 |
+
"grad_norm": 91.17380836071179,
|
| 451 |
"learning_rate": 4.352694346459396e-07,
|
| 452 |
+
"logits/chosen": -1.7119308710098267,
|
| 453 |
+
"logits/rejected": -1.713330626487732,
|
| 454 |
+
"logps/chosen": -245.8417205810547,
|
| 455 |
+
"logps/rejected": -257.9246826171875,
|
| 456 |
+
"loss": 0.3928,
|
| 457 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 458 |
+
"rewards/chosen": -0.6780133843421936,
|
| 459 |
+
"rewards/margins": 1.5433403253555298,
|
| 460 |
+
"rewards/rejected": -2.221353769302368,
|
| 461 |
"step": 270
|
| 462 |
},
|
| 463 |
{
|
| 464 |
"epoch": 0.32,
|
| 465 |
+
"grad_norm": 85.89753055245001,
|
| 466 |
"learning_rate": 4.2836938302509256e-07,
|
| 467 |
+
"logits/chosen": -1.7531852722167969,
|
| 468 |
+
"logits/rejected": -1.6354246139526367,
|
| 469 |
+
"logps/chosen": -243.12454223632812,
|
| 470 |
+
"logps/rejected": -249.15576171875,
|
| 471 |
+
"loss": 0.4123,
|
| 472 |
+
"rewards/accuracies": 0.8125,
|
| 473 |
+
"rewards/chosen": -0.8161875605583191,
|
| 474 |
+
"rewards/margins": 1.92093825340271,
|
| 475 |
+
"rewards/rejected": -2.737125873565674,
|
| 476 |
"step": 280
|
| 477 |
},
|
| 478 |
{
|
| 479 |
"epoch": 0.33,
|
| 480 |
+
"grad_norm": 93.29457537220665,
|
| 481 |
"learning_rate": 4.2118075592405874e-07,
|
| 482 |
+
"logits/chosen": -1.6761655807495117,
|
| 483 |
+
"logits/rejected": -1.668341040611267,
|
| 484 |
+
"logps/chosen": -263.28924560546875,
|
| 485 |
+
"logps/rejected": -276.89190673828125,
|
| 486 |
+
"loss": 0.3899,
|
| 487 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 488 |
+
"rewards/chosen": -0.7706686854362488,
|
| 489 |
+
"rewards/margins": 1.6867666244506836,
|
| 490 |
+
"rewards/rejected": -2.457435131072998,
|
| 491 |
"step": 290
|
| 492 |
},
|
| 493 |
{
|
| 494 |
"epoch": 0.35,
|
| 495 |
+
"grad_norm": 104.22663391151875,
|
| 496 |
"learning_rate": 4.137151834863213e-07,
|
| 497 |
+
"logits/chosen": -1.5810635089874268,
|
| 498 |
+
"logits/rejected": -1.56969153881073,
|
| 499 |
+
"logps/chosen": -215.5782012939453,
|
| 500 |
+
"logps/rejected": -242.8097381591797,
|
| 501 |
+
"loss": 0.4188,
|
| 502 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 503 |
+
"rewards/chosen": -1.0694911479949951,
|
| 504 |
+
"rewards/margins": 1.6282224655151367,
|
| 505 |
+
"rewards/rejected": -2.6977133750915527,
|
| 506 |
"step": 300
|
| 507 |
},
|
| 508 |
{
|
| 509 |
"epoch": 0.35,
|
| 510 |
+
"eval_logits/chosen": -1.6898695230484009,
|
| 511 |
+
"eval_logits/rejected": -1.6142553091049194,
|
| 512 |
+
"eval_logps/chosen": -336.9513244628906,
|
| 513 |
+
"eval_logps/rejected": -373.6328430175781,
|
| 514 |
+
"eval_loss": 0.39220622181892395,
|
| 515 |
+
"eval_rewards/accuracies": 0.796875,
|
| 516 |
+
"eval_rewards/chosen": -0.07198946177959442,
|
| 517 |
+
"eval_rewards/margins": 1.9543578624725342,
|
| 518 |
+
"eval_rewards/rejected": -2.0263473987579346,
|
| 519 |
+
"eval_runtime": 98.0591,
|
| 520 |
+
"eval_samples_per_second": 20.396,
|
| 521 |
"eval_steps_per_second": 0.326,
|
| 522 |
"step": 300
|
| 523 |
},
|
| 524 |
{
|
| 525 |
"epoch": 0.36,
|
| 526 |
+
"grad_norm": 83.12181184953562,
|
| 527 |
"learning_rate": 4.059847439122671e-07,
|
| 528 |
+
"logits/chosen": -1.7229493856430054,
|
| 529 |
+
"logits/rejected": -1.590850591659546,
|
| 530 |
+
"logps/chosen": -260.96539306640625,
|
| 531 |
+
"logps/rejected": -262.43865966796875,
|
| 532 |
+
"loss": 0.4106,
|
| 533 |
+
"rewards/accuracies": 0.800000011920929,
|
| 534 |
+
"rewards/chosen": -1.0447856187820435,
|
| 535 |
+
"rewards/margins": 1.545778512954712,
|
| 536 |
+
"rewards/rejected": -2.590564250946045,
|
| 537 |
"step": 310
|
| 538 |
},
|
| 539 |
{
|
| 540 |
"epoch": 0.37,
|
| 541 |
+
"grad_norm": 101.57242075352875,
|
| 542 |
"learning_rate": 3.98001943918432e-07,
|
| 543 |
+
"logits/chosen": -1.6267013549804688,
|
| 544 |
+
"logits/rejected": -1.5495989322662354,
|
| 545 |
+
"logps/chosen": -246.096923828125,
|
| 546 |
+
"logps/rejected": -268.87274169921875,
|
| 547 |
+
"loss": 0.3937,
|
| 548 |
+
"rewards/accuracies": 0.824999988079071,
|
| 549 |
+
"rewards/chosen": -1.1101688146591187,
|
| 550 |
+
"rewards/margins": 1.6341445446014404,
|
| 551 |
+
"rewards/rejected": -2.7443130016326904,
|
| 552 |
"step": 320
|
| 553 |
},
|
| 554 |
{
|
| 555 |
"epoch": 0.38,
|
| 556 |
+
"grad_norm": 95.29775137930592,
|
| 557 |
"learning_rate": 3.8977969850346866e-07,
|
| 558 |
+
"logits/chosen": -1.5687649250030518,
|
| 559 |
+
"logits/rejected": -1.5476423501968384,
|
| 560 |
+
"logps/chosen": -262.67156982421875,
|
| 561 |
+
"logps/rejected": -260.74725341796875,
|
| 562 |
+
"loss": 0.3887,
|
| 563 |
+
"rewards/accuracies": 0.768750011920929,
|
| 564 |
+
"rewards/chosen": -0.9716947674751282,
|
| 565 |
+
"rewards/margins": 1.6578747034072876,
|
| 566 |
+
"rewards/rejected": -2.6295692920684814,
|
| 567 |
"step": 330
|
| 568 |
},
|
| 569 |
{
|
| 570 |
"epoch": 0.39,
|
| 571 |
+
"grad_norm": 105.61150036189296,
|
| 572 |
"learning_rate": 3.8133131005357465e-07,
|
| 573 |
+
"logits/chosen": -1.662001371383667,
|
| 574 |
+
"logits/rejected": -1.5881233215332031,
|
| 575 |
+
"logps/chosen": -249.8246612548828,
|
| 576 |
+
"logps/rejected": -272.1719970703125,
|
| 577 |
+
"loss": 0.38,
|
| 578 |
+
"rewards/accuracies": 0.862500011920929,
|
| 579 |
+
"rewards/chosen": -0.9675191044807434,
|
| 580 |
+
"rewards/margins": 2.143432140350342,
|
| 581 |
+
"rewards/rejected": -3.1109511852264404,
|
| 582 |
"step": 340
|
| 583 |
},
|
| 584 |
{
|
| 585 |
"epoch": 0.4,
|
| 586 |
+
"grad_norm": 90.41304599394526,
|
| 587 |
"learning_rate": 3.7267044682118435e-07,
|
| 588 |
+
"logits/chosen": -1.6518446207046509,
|
| 589 |
+
"logits/rejected": -1.6180702447891235,
|
| 590 |
+
"logps/chosen": -237.4104461669922,
|
| 591 |
+
"logps/rejected": -247.13473510742188,
|
| 592 |
+
"loss": 0.3749,
|
| 593 |
+
"rewards/accuracies": 0.824999988079071,
|
| 594 |
+
"rewards/chosen": -1.1348434686660767,
|
| 595 |
+
"rewards/margins": 1.6873624324798584,
|
| 596 |
+
"rewards/rejected": -2.8222060203552246,
|
| 597 |
"step": 350
|
| 598 |
},
|
| 599 |
{
|
| 600 |
"epoch": 0.41,
|
| 601 |
+
"grad_norm": 95.16076600849088,
|
| 602 |
"learning_rate": 3.638111208117425e-07,
|
| 603 |
+
"logits/chosen": -1.7075703144073486,
|
| 604 |
+
"logits/rejected": -1.6682260036468506,
|
| 605 |
+
"logps/chosen": -247.7568817138672,
|
| 606 |
+
"logps/rejected": -263.2861633300781,
|
| 607 |
+
"loss": 0.3914,
|
| 608 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 609 |
+
"rewards/chosen": -1.0359086990356445,
|
| 610 |
+
"rewards/margins": 1.2954949140548706,
|
| 611 |
+
"rewards/rejected": -2.3314034938812256,
|
| 612 |
"step": 360
|
| 613 |
},
|
| 614 |
{
|
| 615 |
"epoch": 0.43,
|
| 616 |
+
"grad_norm": 96.37188736705814,
|
| 617 |
"learning_rate": 3.5476766511433605e-07,
|
| 618 |
+
"logits/chosen": -1.7550392150878906,
|
| 619 |
+
"logits/rejected": -1.6907631158828735,
|
| 620 |
+
"logps/chosen": -281.13836669921875,
|
| 621 |
+
"logps/rejected": -266.0453796386719,
|
| 622 |
+
"loss": 0.4097,
|
| 623 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 624 |
+
"rewards/chosen": -1.0189838409423828,
|
| 625 |
+
"rewards/margins": 1.6438014507293701,
|
| 626 |
+
"rewards/rejected": -2.662785291671753,
|
| 627 |
"step": 370
|
| 628 |
},
|
| 629 |
{
|
| 630 |
"epoch": 0.44,
|
| 631 |
+
"grad_norm": 101.71059089383448,
|
| 632 |
"learning_rate": 3.455547107128602e-07,
|
| 633 |
+
"logits/chosen": -1.7938178777694702,
|
| 634 |
+
"logits/rejected": -1.787418007850647,
|
| 635 |
+
"logps/chosen": -298.5751647949219,
|
| 636 |
+
"logps/rejected": -281.03118896484375,
|
| 637 |
+
"loss": 0.3611,
|
| 638 |
+
"rewards/accuracies": 0.831250011920929,
|
| 639 |
+
"rewards/chosen": -1.0122178792953491,
|
| 640 |
+
"rewards/margins": 2.0528714656829834,
|
| 641 |
+
"rewards/rejected": -3.065088987350464,
|
| 642 |
"step": 380
|
| 643 |
},
|
| 644 |
{
|
| 645 |
"epoch": 0.45,
|
| 646 |
+
"grad_norm": 109.30188938122785,
|
| 647 |
"learning_rate": 3.361871628152338e-07,
|
| 648 |
+
"logits/chosen": -1.8471622467041016,
|
| 649 |
+
"logits/rejected": -1.820481300354004,
|
| 650 |
+
"logps/chosen": -251.8716278076172,
|
| 651 |
+
"logps/rejected": -283.156982421875,
|
| 652 |
+
"loss": 0.4006,
|
| 653 |
+
"rewards/accuracies": 0.800000011920929,
|
| 654 |
+
"rewards/chosen": -1.0065103769302368,
|
| 655 |
+
"rewards/margins": 1.774526834487915,
|
| 656 |
+
"rewards/rejected": -2.7810370922088623,
|
| 657 |
"step": 390
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 0.46,
|
| 661 |
+
"grad_norm": 89.15033820155391,
|
| 662 |
"learning_rate": 3.2668017673896077e-07,
|
| 663 |
+
"logits/chosen": -1.8275989294052124,
|
| 664 |
+
"logits/rejected": -1.8090099096298218,
|
| 665 |
+
"logps/chosen": -244.0500946044922,
|
| 666 |
+
"logps/rejected": -245.302490234375,
|
| 667 |
+
"loss": 0.3506,
|
| 668 |
+
"rewards/accuracies": 0.800000011920929,
|
| 669 |
+
"rewards/chosen": -0.62747722864151,
|
| 670 |
+
"rewards/margins": 2.005857467651367,
|
| 671 |
+
"rewards/rejected": -2.6333346366882324,
|
| 672 |
"step": 400
|
| 673 |
},
|
| 674 |
{
|
| 675 |
"epoch": 0.46,
|
| 676 |
+
"eval_logits/chosen": -2.029554843902588,
|
| 677 |
+
"eval_logits/rejected": -1.9793704748153687,
|
| 678 |
+
"eval_logps/chosen": -331.169189453125,
|
| 679 |
+
"eval_logps/rejected": -374.0495300292969,
|
| 680 |
+
"eval_loss": 0.3456653654575348,
|
| 681 |
+
"eval_rewards/accuracies": 0.8203125,
|
| 682 |
+
"eval_rewards/chosen": 0.217118039727211,
|
| 683 |
+
"eval_rewards/margins": 2.2643015384674072,
|
| 684 |
+
"eval_rewards/rejected": -2.0471832752227783,
|
| 685 |
+
"eval_runtime": 97.9966,
|
| 686 |
+
"eval_samples_per_second": 20.409,
|
| 687 |
+
"eval_steps_per_second": 0.327,
|
| 688 |
"step": 400
|
| 689 |
},
|
| 690 |
{
|
| 691 |
"epoch": 0.47,
|
| 692 |
+
"grad_norm": 95.54284470696203,
|
| 693 |
"learning_rate": 3.1704913339205103e-07,
|
| 694 |
+
"logits/chosen": -1.9682222604751587,
|
| 695 |
+
"logits/rejected": -1.9323310852050781,
|
| 696 |
+
"logps/chosen": -251.9891357421875,
|
| 697 |
+
"logps/rejected": -276.3076477050781,
|
| 698 |
+
"loss": 0.3841,
|
| 699 |
+
"rewards/accuracies": 0.831250011920929,
|
| 700 |
+
"rewards/chosen": -0.7251254320144653,
|
| 701 |
+
"rewards/margins": 1.8693323135375977,
|
| 702 |
+
"rewards/rejected": -2.5944573879241943,
|
| 703 |
"step": 410
|
| 704 |
},
|
| 705 |
{
|
| 706 |
"epoch": 0.48,
|
| 707 |
+
"grad_norm": 84.89194807368709,
|
| 708 |
"learning_rate": 3.0730961438896885e-07,
|
| 709 |
+
"logits/chosen": -1.973587989807129,
|
| 710 |
+
"logits/rejected": -1.9758260250091553,
|
| 711 |
+
"logps/chosen": -326.0084533691406,
|
| 712 |
+
"logps/rejected": -316.46380615234375,
|
| 713 |
+
"loss": 0.3676,
|
| 714 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 715 |
+
"rewards/chosen": -1.2638908624649048,
|
| 716 |
+
"rewards/margins": 1.681646704673767,
|
| 717 |
+
"rewards/rejected": -2.945537805557251,
|
| 718 |
"step": 420
|
| 719 |
},
|
| 720 |
{
|
| 721 |
"epoch": 0.5,
|
| 722 |
+
"grad_norm": 79.45330669787697,
|
| 723 |
"learning_rate": 2.9747737684186795e-07,
|
| 724 |
+
"logits/chosen": -1.9824804067611694,
|
| 725 |
+
"logits/rejected": -2.0262296199798584,
|
| 726 |
+
"logps/chosen": -253.5684814453125,
|
| 727 |
+
"logps/rejected": -258.55352783203125,
|
| 728 |
+
"loss": 0.3861,
|
| 729 |
"rewards/accuracies": 0.8125,
|
| 730 |
+
"rewards/chosen": -0.8729881048202515,
|
| 731 |
+
"rewards/margins": 1.7805559635162354,
|
| 732 |
+
"rewards/rejected": -2.6535439491271973,
|
| 733 |
"step": 430
|
| 734 |
},
|
| 735 |
{
|
| 736 |
"epoch": 0.51,
|
| 737 |
+
"grad_norm": 92.0142521630137,
|
| 738 |
"learning_rate": 2.8756832786789663e-07,
|
| 739 |
+
"logits/chosen": -2.008882761001587,
|
| 740 |
+
"logits/rejected": -1.9781932830810547,
|
| 741 |
+
"logps/chosen": -269.37042236328125,
|
| 742 |
+
"logps/rejected": -264.4306335449219,
|
| 743 |
+
"loss": 0.382,
|
| 744 |
"rewards/accuracies": 0.8187500238418579,
|
| 745 |
+
"rewards/chosen": -0.5206736326217651,
|
| 746 |
+
"rewards/margins": 1.916638970375061,
|
| 747 |
+
"rewards/rejected": -2.437312364578247,
|
| 748 |
"step": 440
|
| 749 |
},
|
| 750 |
{
|
| 751 |
"epoch": 0.52,
|
| 752 |
+
"grad_norm": 81.53843237959488,
|
| 753 |
"learning_rate": 2.7759849885381747e-07,
|
| 754 |
+
"logits/chosen": -1.951061487197876,
|
| 755 |
+
"logits/rejected": -1.9269872903823853,
|
| 756 |
+
"logps/chosen": -282.7742614746094,
|
| 757 |
+
"logps/rejected": -261.77130126953125,
|
| 758 |
+
"loss": 0.3613,
|
| 759 |
+
"rewards/accuracies": 0.824999988079071,
|
| 760 |
+
"rewards/chosen": -0.7152455449104309,
|
| 761 |
+
"rewards/margins": 2.0386595726013184,
|
| 762 |
+
"rewards/rejected": -2.7539050579071045,
|
| 763 |
"step": 450
|
| 764 |
},
|
| 765 |
{
|
| 766 |
"epoch": 0.53,
|
| 767 |
+
"grad_norm": 81.29719600574327,
|
| 768 |
"learning_rate": 2.675840195195762e-07,
|
| 769 |
+
"logits/chosen": -1.935346007347107,
|
| 770 |
+
"logits/rejected": -1.8713289499282837,
|
| 771 |
+
"logps/chosen": -237.9385528564453,
|
| 772 |
+
"logps/rejected": -262.90496826171875,
|
| 773 |
+
"loss": 0.3768,
|
| 774 |
+
"rewards/accuracies": 0.84375,
|
| 775 |
+
"rewards/chosen": -0.7526956796646118,
|
| 776 |
+
"rewards/margins": 2.036323070526123,
|
| 777 |
+
"rewards/rejected": -2.7890188694000244,
|
| 778 |
"step": 460
|
| 779 |
},
|
| 780 |
{
|
| 781 |
"epoch": 0.54,
|
| 782 |
+
"grad_norm": 79.41069847067702,
|
| 783 |
"learning_rate": 2.575410918227829e-07,
|
| 784 |
+
"logits/chosen": -1.9017337560653687,
|
| 785 |
+
"logits/rejected": -1.8612467050552368,
|
| 786 |
+
"logps/chosen": -286.1165771484375,
|
| 787 |
+
"logps/rejected": -286.47576904296875,
|
| 788 |
+
"loss": 0.3801,
|
| 789 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 790 |
+
"rewards/chosen": -0.7328876256942749,
|
| 791 |
+
"rewards/margins": 1.7589473724365234,
|
| 792 |
+
"rewards/rejected": -2.491835117340088,
|
| 793 |
"step": 470
|
| 794 |
},
|
| 795 |
{
|
| 796 |
"epoch": 0.55,
|
| 797 |
+
"grad_norm": 84.31202005756423,
|
| 798 |
"learning_rate": 2.474859637463226e-07,
|
| 799 |
+
"logits/chosen": -1.7930266857147217,
|
| 800 |
+
"logits/rejected": -1.6920406818389893,
|
| 801 |
+
"logps/chosen": -261.0568542480469,
|
| 802 |
+
"logps/rejected": -249.1828155517578,
|
| 803 |
+
"loss": 0.373,
|
| 804 |
+
"rewards/accuracies": 0.8125,
|
| 805 |
+
"rewards/chosen": -0.6825451254844666,
|
| 806 |
+
"rewards/margins": 2.1065242290496826,
|
| 807 |
+
"rewards/rejected": -2.789069414138794,
|
| 808 |
"step": 480
|
| 809 |
},
|
| 810 |
{
|
| 811 |
"epoch": 0.56,
|
| 812 |
+
"grad_norm": 85.77456711320595,
|
| 813 |
"learning_rate": 2.3743490301150355e-07,
|
| 814 |
+
"logits/chosen": -1.8167390823364258,
|
| 815 |
+
"logits/rejected": -1.6751165390014648,
|
| 816 |
+
"logps/chosen": -254.97879028320312,
|
| 817 |
+
"logps/rejected": -258.82794189453125,
|
| 818 |
+
"loss": 0.3847,
|
| 819 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 820 |
+
"rewards/chosen": -0.421345055103302,
|
| 821 |
+
"rewards/margins": 2.164630174636841,
|
| 822 |
+
"rewards/rejected": -2.585975408554077,
|
| 823 |
"step": 490
|
| 824 |
},
|
| 825 |
{
|
| 826 |
"epoch": 0.58,
|
| 827 |
+
"grad_norm": 80.63419673971578,
|
| 828 |
"learning_rate": 2.274041707592724e-07,
|
| 829 |
+
"logits/chosen": -1.7511818408966064,
|
| 830 |
+
"logits/rejected": -1.744807481765747,
|
| 831 |
+
"logps/chosen": -238.7528839111328,
|
| 832 |
+
"logps/rejected": -277.8020935058594,
|
| 833 |
+
"loss": 0.3611,
|
| 834 |
+
"rewards/accuracies": 0.831250011920929,
|
| 835 |
+
"rewards/chosen": -1.161329984664917,
|
| 836 |
+
"rewards/margins": 2.127523422241211,
|
| 837 |
+
"rewards/rejected": -3.288853406906128,
|
| 838 |
"step": 500
|
| 839 |
},
|
| 840 |
{
|
| 841 |
"epoch": 0.58,
|
| 842 |
+
"eval_logits/chosen": -1.8592296838760376,
|
| 843 |
+
"eval_logits/rejected": -1.818342685699463,
|
| 844 |
+
"eval_logps/chosen": -330.5163879394531,
|
| 845 |
+
"eval_logps/rejected": -381.7996520996094,
|
| 846 |
+
"eval_loss": 0.29585352540016174,
|
| 847 |
+
"eval_rewards/accuracies": 0.8515625,
|
| 848 |
+
"eval_rewards/chosen": 0.24975742399692535,
|
| 849 |
+
"eval_rewards/margins": 2.68444561958313,
|
| 850 |
+
"eval_rewards/rejected": -2.4346883296966553,
|
| 851 |
+
"eval_runtime": 97.917,
|
| 852 |
+
"eval_samples_per_second": 20.425,
|
| 853 |
+
"eval_steps_per_second": 0.327,
|
| 854 |
"step": 500
|
| 855 |
},
|
| 856 |
{
|
| 857 |
"epoch": 0.59,
|
| 858 |
+
"grad_norm": 93.97823403064545,
|
| 859 |
"learning_rate": 2.17409995242075e-07,
|
| 860 |
+
"logits/chosen": -1.8619616031646729,
|
| 861 |
+
"logits/rejected": -1.8197988271713257,
|
| 862 |
+
"logps/chosen": -246.95120239257812,
|
| 863 |
+
"logps/rejected": -251.1341552734375,
|
| 864 |
+
"loss": 0.3787,
|
| 865 |
+
"rewards/accuracies": 0.831250011920929,
|
| 866 |
+
"rewards/chosen": -0.9560788869857788,
|
| 867 |
+
"rewards/margins": 2.1973986625671387,
|
| 868 |
+
"rewards/rejected": -3.153477668762207,
|
| 869 |
"step": 510
|
| 870 |
},
|
| 871 |
{
|
| 872 |
"epoch": 0.6,
|
| 873 |
+
"grad_norm": 81.29931903371055,
|
| 874 |
"learning_rate": 2.0746854556892544e-07,
|
| 875 |
+
"logits/chosen": -1.8456933498382568,
|
| 876 |
+
"logits/rejected": -1.8626302480697632,
|
| 877 |
+
"logps/chosen": -228.947265625,
|
| 878 |
+
"logps/rejected": -255.99887084960938,
|
| 879 |
+
"loss": 0.383,
|
| 880 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 881 |
+
"rewards/chosen": -0.850503146648407,
|
| 882 |
+
"rewards/margins": 1.7254650592803955,
|
| 883 |
+
"rewards/rejected": -2.5759682655334473,
|
| 884 |
"step": 520
|
| 885 |
},
|
| 886 |
{
|
| 887 |
"epoch": 0.61,
|
| 888 |
+
"grad_norm": 82.34962963997694,
|
| 889 |
"learning_rate": 1.9759590554616173e-07,
|
| 890 |
+
"logits/chosen": -1.9397910833358765,
|
| 891 |
+
"logits/rejected": -1.9602453708648682,
|
| 892 |
+
"logps/chosen": -259.236328125,
|
| 893 |
+
"logps/rejected": -268.33233642578125,
|
| 894 |
+
"loss": 0.3812,
|
| 895 |
+
"rewards/accuracies": 0.84375,
|
| 896 |
+
"rewards/chosen": -0.5990003347396851,
|
| 897 |
+
"rewards/margins": 1.7735519409179688,
|
| 898 |
+
"rewards/rejected": -2.3725523948669434,
|
| 899 |
"step": 530
|
| 900 |
},
|
| 901 |
{
|
| 902 |
"epoch": 0.62,
|
| 903 |
+
"grad_norm": 106.09742448074628,
|
| 904 |
"learning_rate": 1.8780804765620746e-07,
|
| 905 |
+
"logits/chosen": -1.8515198230743408,
|
| 906 |
+
"logits/rejected": -1.8524954319000244,
|
| 907 |
+
"logps/chosen": -268.37164306640625,
|
| 908 |
+
"logps/rejected": -304.6097106933594,
|
| 909 |
+
"loss": 0.3726,
|
| 910 |
+
"rewards/accuracies": 0.78125,
|
| 911 |
+
"rewards/chosen": -0.32395535707473755,
|
| 912 |
+
"rewards/margins": 1.7947556972503662,
|
| 913 |
+
"rewards/rejected": -2.118710994720459,
|
| 914 |
"step": 540
|
| 915 |
},
|
| 916 |
{
|
| 917 |
"epoch": 0.63,
|
| 918 |
+
"grad_norm": 89.06639468347461,
|
| 919 |
"learning_rate": 1.7812080721643973e-07,
|
| 920 |
+
"logits/chosen": -1.8575937747955322,
|
| 921 |
+
"logits/rejected": -1.761614441871643,
|
| 922 |
+
"logps/chosen": -263.5863037109375,
|
| 923 |
+
"logps/rejected": -246.20455932617188,
|
| 924 |
+
"loss": 0.3972,
|
| 925 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 926 |
+
"rewards/chosen": -0.7359617352485657,
|
| 927 |
+
"rewards/margins": 1.8330894708633423,
|
| 928 |
+
"rewards/rejected": -2.5690512657165527,
|
| 929 |
"step": 550
|
| 930 |
},
|
| 931 |
{
|
| 932 |
"epoch": 0.64,
|
| 933 |
+
"grad_norm": 91.51640548835455,
|
| 934 |
"learning_rate": 1.6854985675997063e-07,
|
| 935 |
+
"logits/chosen": -1.8521220684051514,
|
| 936 |
+
"logits/rejected": -1.8151109218597412,
|
| 937 |
+
"logps/chosen": -261.0481872558594,
|
| 938 |
+
"logps/rejected": -272.9272155761719,
|
| 939 |
+
"loss": 0.3641,
|
| 940 |
+
"rewards/accuracies": 0.78125,
|
| 941 |
+
"rewards/chosen": -0.9162181615829468,
|
| 942 |
+
"rewards/margins": 1.6873579025268555,
|
| 943 |
+
"rewards/rejected": -2.603576183319092,
|
| 944 |
"step": 560
|
| 945 |
},
|
| 946 |
{
|
| 947 |
"epoch": 0.66,
|
| 948 |
+
"grad_norm": 83.89007997657542,
|
| 949 |
"learning_rate": 1.5911068067978818e-07,
|
| 950 |
+
"logits/chosen": -1.7626062631607056,
|
| 951 |
+
"logits/rejected": -1.7946131229400635,
|
| 952 |
+
"logps/chosen": -240.85189819335938,
|
| 953 |
+
"logps/rejected": -273.8741149902344,
|
| 954 |
+
"loss": 0.3514,
|
| 955 |
+
"rewards/accuracies": 0.768750011920929,
|
| 956 |
+
"rewards/chosen": -0.9277374148368835,
|
| 957 |
+
"rewards/margins": 1.9961919784545898,
|
| 958 |
+
"rewards/rejected": -2.923929214477539,
|
| 959 |
"step": 570
|
| 960 |
},
|
| 961 |
{
|
| 962 |
"epoch": 0.67,
|
| 963 |
+
"grad_norm": 103.64429212789244,
|
| 964 |
"learning_rate": 1.4981855017728197e-07,
|
| 965 |
+
"logits/chosen": -1.8487884998321533,
|
| 966 |
+
"logits/rejected": -1.8991634845733643,
|
| 967 |
+
"logps/chosen": -267.7337341308594,
|
| 968 |
+
"logps/rejected": -304.248291015625,
|
| 969 |
+
"loss": 0.37,
|
| 970 |
+
"rewards/accuracies": 0.8125,
|
| 971 |
+
"rewards/chosen": -1.1558853387832642,
|
| 972 |
+
"rewards/margins": 1.9654242992401123,
|
| 973 |
+
"rewards/rejected": -3.121309757232666,
|
| 974 |
"step": 580
|
| 975 |
},
|
| 976 |
{
|
| 977 |
"epoch": 0.68,
|
| 978 |
+
"grad_norm": 87.97513457883551,
|
| 979 |
"learning_rate": 1.406884985556804e-07,
|
| 980 |
+
"logits/chosen": -1.8694698810577393,
|
| 981 |
+
"logits/rejected": -1.844276785850525,
|
| 982 |
+
"logps/chosen": -257.92413330078125,
|
| 983 |
+
"logps/rejected": -268.24462890625,
|
| 984 |
+
"loss": 0.3586,
|
| 985 |
+
"rewards/accuracies": 0.8125,
|
| 986 |
+
"rewards/chosen": -1.2031538486480713,
|
| 987 |
+
"rewards/margins": 2.1714892387390137,
|
| 988 |
+
"rewards/rejected": -3.374642848968506,
|
| 989 |
"step": 590
|
| 990 |
},
|
| 991 |
{
|
| 992 |
"epoch": 0.69,
|
| 993 |
+
"grad_norm": 85.79715679551722,
|
| 994 |
"learning_rate": 1.3173529689837354e-07,
|
| 995 |
+
"logits/chosen": -1.8827781677246094,
|
| 996 |
+
"logits/rejected": -1.8290605545043945,
|
| 997 |
+
"logps/chosen": -256.6871032714844,
|
| 998 |
+
"logps/rejected": -256.0238952636719,
|
| 999 |
+
"loss": 0.3562,
|
| 1000 |
"rewards/accuracies": 0.800000011920929,
|
| 1001 |
+
"rewards/chosen": -1.1103084087371826,
|
| 1002 |
+
"rewards/margins": 1.9665857553482056,
|
| 1003 |
+
"rewards/rejected": -3.0768942832946777,
|
| 1004 |
"step": 600
|
| 1005 |
},
|
| 1006 |
{
|
| 1007 |
"epoch": 0.69,
|
| 1008 |
+
"eval_logits/chosen": -1.9736415147781372,
|
| 1009 |
+
"eval_logits/rejected": -1.9216868877410889,
|
| 1010 |
+
"eval_logps/chosen": -327.7752685546875,
|
| 1011 |
+
"eval_logps/rejected": -382.5696105957031,
|
| 1012 |
+
"eval_loss": 0.2513369023799896,
|
| 1013 |
+
"eval_rewards/accuracies": 0.87109375,
|
| 1014 |
+
"eval_rewards/chosen": 0.386812299489975,
|
| 1015 |
+
"eval_rewards/margins": 2.8599982261657715,
|
| 1016 |
+
"eval_rewards/rejected": -2.4731857776641846,
|
| 1017 |
+
"eval_runtime": 97.9427,
|
| 1018 |
+
"eval_samples_per_second": 20.42,
|
| 1019 |
+
"eval_steps_per_second": 0.327,
|
| 1020 |
"step": 600
|
| 1021 |
},
|
| 1022 |
{
|
| 1023 |
"epoch": 0.7,
|
| 1024 |
+
"grad_norm": 91.70310566548932,
|
| 1025 |
"learning_rate": 1.2297343017146726e-07,
|
| 1026 |
+
"logits/chosen": -1.8489186763763428,
|
| 1027 |
+
"logits/rejected": -1.8288800716400146,
|
| 1028 |
+
"logps/chosen": -259.6686096191406,
|
| 1029 |
+
"logps/rejected": -267.1070861816406,
|
| 1030 |
+
"loss": 0.3693,
|
| 1031 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1032 |
+
"rewards/chosen": -1.0914928913116455,
|
| 1033 |
+
"rewards/margins": 1.885284662246704,
|
| 1034 |
+
"rewards/rejected": -2.9767773151397705,
|
| 1035 |
"step": 610
|
| 1036 |
},
|
| 1037 |
{
|
| 1038 |
"epoch": 0.71,
|
| 1039 |
+
"grad_norm": 83.02991328479807,
|
| 1040 |
"learning_rate": 1.1441707378923474e-07,
|
| 1041 |
+
"logits/chosen": -1.950595498085022,
|
| 1042 |
+
"logits/rejected": -1.8974205255508423,
|
| 1043 |
+
"logps/chosen": -237.6828155517578,
|
| 1044 |
+
"logps/rejected": -257.06005859375,
|
| 1045 |
+
"loss": 0.3625,
|
| 1046 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1047 |
+
"rewards/chosen": -0.4481216371059418,
|
| 1048 |
+
"rewards/margins": 2.152703046798706,
|
| 1049 |
+
"rewards/rejected": -2.600825071334839,
|
| 1050 |
"step": 620
|
| 1051 |
},
|
| 1052 |
{
|
| 1053 |
"epoch": 0.73,
|
| 1054 |
+
"grad_norm": 97.69736464926112,
|
| 1055 |
"learning_rate": 1.06080070680377e-07,
|
| 1056 |
+
"logits/chosen": -1.874447226524353,
|
| 1057 |
+
"logits/rejected": -1.8658256530761719,
|
| 1058 |
+
"logps/chosen": -269.42034912109375,
|
| 1059 |
+
"logps/rejected": -275.75946044921875,
|
| 1060 |
+
"loss": 0.3679,
|
| 1061 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1062 |
+
"rewards/chosen": -0.6408880949020386,
|
| 1063 |
+
"rewards/margins": 2.110172986984253,
|
| 1064 |
+
"rewards/rejected": -2.751060962677002,
|
| 1065 |
"step": 630
|
| 1066 |
},
|
| 1067 |
{
|
| 1068 |
"epoch": 0.74,
|
| 1069 |
+
"grad_norm": 89.55100518815718,
|
| 1070 |
"learning_rate": 9.797590889219587e-08,
|
| 1071 |
+
"logits/chosen": -1.927781343460083,
|
| 1072 |
+
"logits/rejected": -1.915203332901001,
|
| 1073 |
+
"logps/chosen": -262.8731384277344,
|
| 1074 |
+
"logps/rejected": -265.70404052734375,
|
| 1075 |
+
"loss": 0.3725,
|
| 1076 |
"rewards/accuracies": 0.8374999761581421,
|
| 1077 |
+
"rewards/chosen": -0.5387133359909058,
|
| 1078 |
+
"rewards/margins": 2.1157500743865967,
|
| 1079 |
+
"rewards/rejected": -2.654463529586792,
|
| 1080 |
"step": 640
|
| 1081 |
},
|
| 1082 |
{
|
| 1083 |
"epoch": 0.75,
|
| 1084 |
+
"grad_norm": 88.47283263403602,
|
| 1085 |
"learning_rate": 9.011769976891367e-08,
|
| 1086 |
+
"logits/chosen": -1.9129142761230469,
|
| 1087 |
+
"logits/rejected": -1.9037758111953735,
|
| 1088 |
+
"logps/chosen": -253.7337188720703,
|
| 1089 |
+
"logps/rejected": -273.1281433105469,
|
| 1090 |
+
"loss": 0.3654,
|
| 1091 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 1092 |
+
"rewards/chosen": -0.8051580190658569,
|
| 1093 |
+
"rewards/margins": 1.8512630462646484,
|
| 1094 |
+
"rewards/rejected": -2.656421184539795,
|
| 1095 |
"step": 650
|
| 1096 |
},
|
| 1097 |
{
|
| 1098 |
"epoch": 0.76,
|
| 1099 |
+
"grad_norm": 101.63563998101947,
|
| 1100 |
"learning_rate": 8.251815673944218e-08,
|
| 1101 |
+
"logits/chosen": -1.8455785512924194,
|
| 1102 |
+
"logits/rejected": -1.862540602684021,
|
| 1103 |
+
"logps/chosen": -269.90655517578125,
|
| 1104 |
+
"logps/rejected": -266.5269470214844,
|
| 1105 |
+
"loss": 0.3636,
|
| 1106 |
+
"rewards/accuracies": 0.831250011920929,
|
| 1107 |
+
"rewards/chosen": -1.0431994199752808,
|
| 1108 |
+
"rewards/margins": 1.969745397567749,
|
| 1109 |
+
"rewards/rejected": -3.0129449367523193,
|
| 1110 |
"step": 660
|
| 1111 |
},
|
| 1112 |
{
|
| 1113 |
"epoch": 0.77,
|
| 1114 |
+
"grad_norm": 101.36570058059344,
|
| 1115 |
"learning_rate": 7.518957474892148e-08,
|
| 1116 |
+
"logits/chosen": -1.8844772577285767,
|
| 1117 |
+
"logits/rejected": -1.8293778896331787,
|
| 1118 |
+
"logps/chosen": -262.8396301269531,
|
| 1119 |
+
"logps/rejected": -270.7688293457031,
|
| 1120 |
+
"loss": 0.3566,
|
| 1121 |
+
"rewards/accuracies": 0.84375,
|
| 1122 |
+
"rewards/chosen": -1.0009379386901855,
|
| 1123 |
+
"rewards/margins": 2.288020133972168,
|
| 1124 |
+
"rewards/rejected": -3.2889580726623535,
|
| 1125 |
"step": 670
|
| 1126 |
},
|
| 1127 |
{
|
| 1128 |
"epoch": 0.78,
|
| 1129 |
+
"grad_norm": 108.29800746794625,
|
| 1130 |
"learning_rate": 6.814381036730274e-08,
|
| 1131 |
+
"logits/chosen": -1.9113209247589111,
|
| 1132 |
+
"logits/rejected": -1.8669430017471313,
|
| 1133 |
+
"logps/chosen": -248.9401092529297,
|
| 1134 |
+
"logps/rejected": -263.89459228515625,
|
| 1135 |
+
"loss": 0.3789,
|
| 1136 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 1137 |
+
"rewards/chosen": -0.7384175062179565,
|
| 1138 |
+
"rewards/margins": 2.038583993911743,
|
| 1139 |
+
"rewards/rejected": -2.77700138092041,
|
| 1140 |
"step": 680
|
| 1141 |
},
|
| 1142 |
{
|
| 1143 |
"epoch": 0.79,
|
| 1144 |
+
"grad_norm": 81.30832233061167,
|
| 1145 |
"learning_rate": 6.139226260715872e-08,
|
| 1146 |
+
"logits/chosen": -1.936092734336853,
|
| 1147 |
+
"logits/rejected": -1.950042724609375,
|
| 1148 |
+
"logps/chosen": -261.84283447265625,
|
| 1149 |
+
"logps/rejected": -281.1523132324219,
|
| 1150 |
+
"loss": 0.3622,
|
| 1151 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 1152 |
+
"rewards/chosen": -0.8801490068435669,
|
| 1153 |
+
"rewards/margins": 2.052794933319092,
|
| 1154 |
+
"rewards/rejected": -2.932943820953369,
|
| 1155 |
"step": 690
|
| 1156 |
},
|
| 1157 |
{
|
| 1158 |
"epoch": 0.81,
|
| 1159 |
+
"grad_norm": 81.87543968348014,
|
| 1160 |
"learning_rate": 5.4945854481754734e-08,
|
| 1161 |
+
"logits/chosen": -1.8324077129364014,
|
| 1162 |
+
"logits/rejected": -1.7888282537460327,
|
| 1163 |
+
"logps/chosen": -245.5817413330078,
|
| 1164 |
+
"logps/rejected": -259.8585510253906,
|
| 1165 |
+
"loss": 0.3624,
|
| 1166 |
+
"rewards/accuracies": 0.78125,
|
| 1167 |
+
"rewards/chosen": -0.9253555536270142,
|
| 1168 |
+
"rewards/margins": 2.031710624694824,
|
| 1169 |
+
"rewards/rejected": -2.957066059112549,
|
| 1170 |
"step": 700
|
| 1171 |
},
|
| 1172 |
{
|
| 1173 |
"epoch": 0.81,
|
| 1174 |
+
"eval_logits/chosen": -1.9716989994049072,
|
| 1175 |
+
"eval_logits/rejected": -1.930106282234192,
|
| 1176 |
+
"eval_logps/chosen": -322.60308837890625,
|
| 1177 |
+
"eval_logps/rejected": -380.21783447265625,
|
| 1178 |
+
"eval_loss": 0.21937939524650574,
|
| 1179 |
+
"eval_rewards/accuracies": 0.90625,
|
| 1180 |
+
"eval_rewards/chosen": 0.6454216241836548,
|
| 1181 |
+
"eval_rewards/margins": 3.001021385192871,
|
| 1182 |
+
"eval_rewards/rejected": -2.3555996417999268,
|
| 1183 |
+
"eval_runtime": 98.0586,
|
| 1184 |
+
"eval_samples_per_second": 20.396,
|
| 1185 |
"eval_steps_per_second": 0.326,
|
| 1186 |
"step": 700
|
| 1187 |
},
|
| 1188 |
{
|
| 1189 |
"epoch": 0.82,
|
| 1190 |
+
"grad_norm": 104.29775989655523,
|
| 1191 |
"learning_rate": 4.881501533321605e-08,
|
| 1192 |
+
"logits/chosen": -1.7864516973495483,
|
| 1193 |
+
"logits/rejected": -1.7958072423934937,
|
| 1194 |
+
"logps/chosen": -228.2520294189453,
|
| 1195 |
+
"logps/rejected": -255.38577270507812,
|
| 1196 |
+
"loss": 0.3399,
|
| 1197 |
+
"rewards/accuracies": 0.8125,
|
| 1198 |
+
"rewards/chosen": -0.952102780342102,
|
| 1199 |
+
"rewards/margins": 2.169015884399414,
|
| 1200 |
+
"rewards/rejected": -3.1211180686950684,
|
| 1201 |
"step": 710
|
| 1202 |
},
|
| 1203 |
{
|
| 1204 |
"epoch": 0.83,
|
| 1205 |
+
"grad_norm": 87.51056493700892,
|
| 1206 |
"learning_rate": 4.300966395938377e-08,
|
| 1207 |
+
"logits/chosen": -1.91278076171875,
|
| 1208 |
+
"logits/rejected": -1.887738823890686,
|
| 1209 |
+
"logps/chosen": -269.040283203125,
|
| 1210 |
+
"logps/rejected": -277.3271484375,
|
| 1211 |
+
"loss": 0.3737,
|
| 1212 |
+
"rewards/accuracies": 0.768750011920929,
|
| 1213 |
+
"rewards/chosen": -0.9400026202201843,
|
| 1214 |
+
"rewards/margins": 2.1387667655944824,
|
| 1215 |
+
"rewards/rejected": -3.0787696838378906,
|
| 1216 |
"step": 720
|
| 1217 |
},
|
| 1218 |
{
|
| 1219 |
"epoch": 0.84,
|
| 1220 |
+
"grad_norm": 82.85641235575972,
|
| 1221 |
"learning_rate": 3.7539192566655246e-08,
|
| 1222 |
+
"logits/chosen": -1.8706934452056885,
|
| 1223 |
+
"logits/rejected": -1.8632476329803467,
|
| 1224 |
+
"logps/chosen": -258.32257080078125,
|
| 1225 |
+
"logps/rejected": -262.11151123046875,
|
| 1226 |
+
"loss": 0.3615,
|
| 1227 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1228 |
+
"rewards/chosen": -0.5498217344284058,
|
| 1229 |
+
"rewards/margins": 1.930101752281189,
|
| 1230 |
+
"rewards/rejected": -2.479923725128174,
|
| 1231 |
"step": 730
|
| 1232 |
},
|
| 1233 |
{
|
| 1234 |
"epoch": 0.85,
|
| 1235 |
+
"grad_norm": 86.13162915966026,
|
| 1236 |
"learning_rate": 3.24124515747731e-08,
|
| 1237 |
+
"logits/chosen": -1.827125906944275,
|
| 1238 |
+
"logits/rejected": -1.8333660364151,
|
| 1239 |
+
"logps/chosen": -245.1622314453125,
|
| 1240 |
+
"logps/rejected": -269.59857177734375,
|
| 1241 |
+
"loss": 0.376,
|
| 1242 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1243 |
+
"rewards/chosen": -1.0357568264007568,
|
| 1244 |
+
"rewards/margins": 1.9602575302124023,
|
| 1245 |
+
"rewards/rejected": -2.996014356613159,
|
| 1246 |
"step": 740
|
| 1247 |
},
|
| 1248 |
{
|
| 1249 |
"epoch": 0.86,
|
| 1250 |
+
"grad_norm": 86.92913330390785,
|
| 1251 |
"learning_rate": 2.763773529814506e-08,
|
| 1252 |
+
"logits/chosen": -1.940159559249878,
|
| 1253 |
+
"logits/rejected": -1.9158337116241455,
|
| 1254 |
+
"logps/chosen": -282.51397705078125,
|
| 1255 |
+
"logps/rejected": -276.21331787109375,
|
| 1256 |
+
"loss": 0.372,
|
| 1257 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1258 |
+
"rewards/chosen": -0.7603832483291626,
|
| 1259 |
+
"rewards/margins": 2.119266986846924,
|
| 1260 |
+
"rewards/rejected": -2.879650354385376,
|
| 1261 |
"step": 750
|
| 1262 |
},
|
| 1263 |
{
|
| 1264 |
"epoch": 0.88,
|
| 1265 |
+
"grad_norm": 95.56287400260709,
|
| 1266 |
"learning_rate": 2.3222768526860698e-08,
|
| 1267 |
+
"logits/chosen": -1.863567590713501,
|
| 1268 |
+
"logits/rejected": -1.788558006286621,
|
| 1269 |
+
"logps/chosen": -249.87838745117188,
|
| 1270 |
+
"logps/rejected": -264.7607727050781,
|
| 1271 |
+
"loss": 0.3676,
|
| 1272 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1273 |
+
"rewards/chosen": -0.6970678567886353,
|
| 1274 |
+
"rewards/margins": 2.2210423946380615,
|
| 1275 |
+
"rewards/rejected": -2.9181103706359863,
|
| 1276 |
"step": 760
|
| 1277 |
},
|
| 1278 |
{
|
| 1279 |
"epoch": 0.89,
|
| 1280 |
+
"grad_norm": 82.6081133840389,
|
| 1281 |
"learning_rate": 1.9174694029115146e-08,
|
| 1282 |
+
"logits/chosen": -1.9369175434112549,
|
| 1283 |
+
"logits/rejected": -1.9106756448745728,
|
| 1284 |
+
"logps/chosen": -287.55975341796875,
|
| 1285 |
+
"logps/rejected": -268.37493896484375,
|
| 1286 |
+
"loss": 0.3621,
|
| 1287 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1288 |
+
"rewards/chosen": -0.7068324685096741,
|
| 1289 |
+
"rewards/margins": 1.9774510860443115,
|
| 1290 |
+
"rewards/rejected": -2.684283494949341,
|
| 1291 |
"step": 770
|
| 1292 |
},
|
| 1293 |
{
|
| 1294 |
"epoch": 0.9,
|
| 1295 |
+
"grad_norm": 84.83905388303022,
|
| 1296 |
"learning_rate": 1.5500060995258134e-08,
|
| 1297 |
+
"logits/chosen": -1.8579909801483154,
|
| 1298 |
+
"logits/rejected": -1.8068830966949463,
|
| 1299 |
+
"logps/chosen": -258.0039367675781,
|
| 1300 |
+
"logps/rejected": -259.26202392578125,
|
| 1301 |
+
"loss": 0.3456,
|
| 1302 |
+
"rewards/accuracies": 0.875,
|
| 1303 |
+
"rewards/chosen": -0.8524407148361206,
|
| 1304 |
+
"rewards/margins": 2.122145414352417,
|
| 1305 |
+
"rewards/rejected": -2.974586248397827,
|
| 1306 |
"step": 780
|
| 1307 |
},
|
| 1308 |
{
|
| 1309 |
"epoch": 0.91,
|
| 1310 |
+
"grad_norm": 94.72717106858794,
|
| 1311 |
"learning_rate": 1.2204814442165812e-08,
|
| 1312 |
+
"logits/chosen": -1.8840267658233643,
|
| 1313 |
+
"logits/rejected": -1.8291581869125366,
|
| 1314 |
+
"logps/chosen": -252.4437713623047,
|
| 1315 |
+
"logps/rejected": -252.0703582763672,
|
| 1316 |
+
"loss": 0.3741,
|
| 1317 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1318 |
+
"rewards/chosen": -0.8425480127334595,
|
| 1319 |
+
"rewards/margins": 2.2477831840515137,
|
| 1320 |
+
"rewards/rejected": -3.0903308391571045,
|
| 1321 |
"step": 790
|
| 1322 |
},
|
| 1323 |
{
|
| 1324 |
"epoch": 0.92,
|
| 1325 |
+
"grad_norm": 102.52909854244074,
|
| 1326 |
"learning_rate": 9.294285595075669e-09,
|
| 1327 |
+
"logits/chosen": -1.9619266986846924,
|
| 1328 |
+
"logits/rejected": -1.904088020324707,
|
| 1329 |
+
"logps/chosen": -277.6653747558594,
|
| 1330 |
+
"logps/rejected": -272.3161315917969,
|
| 1331 |
+
"loss": 0.4069,
|
| 1332 |
+
"rewards/accuracies": 0.78125,
|
| 1333 |
+
"rewards/chosen": -0.8278266787528992,
|
| 1334 |
+
"rewards/margins": 2.0403804779052734,
|
| 1335 |
+
"rewards/rejected": -2.868206739425659,
|
| 1336 |
"step": 800
|
| 1337 |
},
|
| 1338 |
{
|
| 1339 |
"epoch": 0.92,
|
| 1340 |
+
"eval_logits/chosen": -1.9591288566589355,
|
| 1341 |
+
"eval_logits/rejected": -1.9204463958740234,
|
| 1342 |
+
"eval_logps/chosen": -322.0539245605469,
|
| 1343 |
+
"eval_logps/rejected": -380.2658386230469,
|
| 1344 |
+
"eval_loss": 0.20271854102611542,
|
| 1345 |
+
"eval_rewards/accuracies": 0.9140625,
|
| 1346 |
+
"eval_rewards/chosen": 0.6728801131248474,
|
| 1347 |
+
"eval_rewards/margins": 3.0308780670166016,
|
| 1348 |
+
"eval_rewards/rejected": -2.3579981327056885,
|
| 1349 |
+
"eval_runtime": 97.802,
|
| 1350 |
+
"eval_samples_per_second": 20.449,
|
| 1351 |
+
"eval_steps_per_second": 0.327,
|
| 1352 |
"step": 800
|
| 1353 |
},
|
| 1354 |
{
|
| 1355 |
"epoch": 0.93,
|
| 1356 |
+
"grad_norm": 85.0418535767566,
|
| 1357 |
"learning_rate": 6.773183262446914e-09,
|
| 1358 |
+
"logits/chosen": -1.8587850332260132,
|
| 1359 |
+
"logits/rejected": -1.7795337438583374,
|
| 1360 |
+
"logps/chosen": -248.3004913330078,
|
| 1361 |
+
"logps/rejected": -261.2222595214844,
|
| 1362 |
+
"loss": 0.3824,
|
| 1363 |
+
"rewards/accuracies": 0.8125,
|
| 1364 |
+
"rewards/chosen": -0.7576299905776978,
|
| 1365 |
+
"rewards/margins": 2.073674440383911,
|
| 1366 |
+
"rewards/rejected": -2.8313040733337402,
|
| 1367 |
"step": 810
|
| 1368 |
},
|
| 1369 |
{
|
| 1370 |
"epoch": 0.94,
|
| 1371 |
+
"grad_norm": 88.3275931540908,
|
| 1372 |
"learning_rate": 4.645586217799452e-09,
|
| 1373 |
+
"logits/chosen": -1.8911056518554688,
|
| 1374 |
+
"logits/rejected": -1.940203309059143,
|
| 1375 |
+
"logps/chosen": -265.9781799316406,
|
| 1376 |
+
"logps/rejected": -290.08770751953125,
|
| 1377 |
+
"loss": 0.3902,
|
| 1378 |
+
"rewards/accuracies": 0.8125,
|
| 1379 |
+
"rewards/chosen": -0.839026927947998,
|
| 1380 |
+
"rewards/margins": 2.1166014671325684,
|
| 1381 |
+
"rewards/rejected": -2.9556286334991455,
|
| 1382 |
"step": 820
|
| 1383 |
},
|
| 1384 |
{
|
| 1385 |
"epoch": 0.96,
|
| 1386 |
+
"grad_norm": 108.33767167817186,
|
| 1387 |
"learning_rate": 2.9149366008568987e-09,
|
| 1388 |
+
"logits/chosen": -1.8930469751358032,
|
| 1389 |
+
"logits/rejected": -1.9079952239990234,
|
| 1390 |
+
"logps/chosen": -263.65264892578125,
|
| 1391 |
+
"logps/rejected": -278.3803405761719,
|
| 1392 |
+
"loss": 0.3825,
|
| 1393 |
+
"rewards/accuracies": 0.793749988079071,
|
| 1394 |
+
"rewards/chosen": -0.6759049296379089,
|
| 1395 |
+
"rewards/margins": 2.146210193634033,
|
| 1396 |
+
"rewards/rejected": -2.822114944458008,
|
| 1397 |
"step": 830
|
| 1398 |
},
|
| 1399 |
{
|
| 1400 |
"epoch": 0.97,
|
| 1401 |
+
"grad_norm": 86.51713108380295,
|
| 1402 |
"learning_rate": 1.5840343486700215e-09,
|
| 1403 |
+
"logits/chosen": -1.9469448328018188,
|
| 1404 |
+
"logits/rejected": -1.939854383468628,
|
| 1405 |
+
"logps/chosen": -281.2010498046875,
|
| 1406 |
+
"logps/rejected": -275.5614318847656,
|
| 1407 |
+
"loss": 0.367,
|
| 1408 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1409 |
+
"rewards/chosen": -0.5235612392425537,
|
| 1410 |
+
"rewards/margins": 2.1837284564971924,
|
| 1411 |
+
"rewards/rejected": -2.707289457321167,
|
| 1412 |
"step": 840
|
| 1413 |
},
|
| 1414 |
{
|
| 1415 |
"epoch": 0.98,
|
| 1416 |
+
"grad_norm": 82.94960696998083,
|
| 1417 |
"learning_rate": 6.550326657293881e-10,
|
| 1418 |
+
"logits/chosen": -1.9341312646865845,
|
| 1419 |
+
"logits/rejected": -1.901523232460022,
|
| 1420 |
+
"logps/chosen": -257.5312805175781,
|
| 1421 |
+
"logps/rejected": -268.84619140625,
|
| 1422 |
+
"loss": 0.346,
|
| 1423 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 1424 |
+
"rewards/chosen": -0.7536182999610901,
|
| 1425 |
+
"rewards/margins": 2.499413251876831,
|
| 1426 |
+
"rewards/rejected": -3.2530312538146973,
|
| 1427 |
"step": 850
|
| 1428 |
},
|
| 1429 |
{
|
| 1430 |
"epoch": 0.99,
|
| 1431 |
+
"grad_norm": 100.59142277641878,
|
| 1432 |
"learning_rate": 1.2943454039654467e-10,
|
| 1433 |
+
"logits/chosen": -1.821735143661499,
|
| 1434 |
+
"logits/rejected": -1.835100531578064,
|
| 1435 |
+
"logps/chosen": -244.0499267578125,
|
| 1436 |
+
"logps/rejected": -261.69805908203125,
|
| 1437 |
+
"loss": 0.3624,
|
| 1438 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1439 |
+
"rewards/chosen": -1.0147624015808105,
|
| 1440 |
+
"rewards/margins": 1.7580705881118774,
|
| 1441 |
+
"rewards/rejected": -2.7728328704833984,
|
| 1442 |
"step": 860
|
| 1443 |
},
|
| 1444 |
{
|
| 1445 |
"epoch": 1.0,
|
| 1446 |
"step": 868,
|
| 1447 |
"total_flos": 0.0,
|
| 1448 |
+
"train_loss": 0.40559157427005504,
|
| 1449 |
+
"train_runtime": 13777.3263,
|
| 1450 |
+
"train_samples_per_second": 8.066,
|
| 1451 |
+
"train_steps_per_second": 0.063
|
| 1452 |
}
|
| 1453 |
],
|
| 1454 |
"logging_steps": 10,
|