Model save
Browse files- README.md +15 -20
- all_results.json +5 -5
- model-00001-of-00003.safetensors +1 -1
- model-00002-of-00003.safetensors +1 -1
- model-00003-of-00003.safetensors +1 -1
- runs/May10_17-43-11_n136-144-082/events.out.tfevents.1715334259.n136-144-082.2699052.0 +2 -2
- train_results.json +5 -5
- trainer_state.json +628 -1428
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:
|
| 19 |
-
- Rewards/chosen: -
|
| 20 |
-
- Rewards/rejected: -
|
| 21 |
-
- Rewards/accuracies: 0.
|
| 22 |
-
- Rewards/margins:
|
| 23 |
-
- Logps/rejected: -
|
| 24 |
-
- Logps/chosen: -
|
| 25 |
-
- Logits/rejected:
|
| 26 |
-
- Logits/chosen:
|
| 27 |
|
| 28 |
## Model description
|
| 29 |
|
|
@@ -58,17 +58,12 @@ The following hyperparameters were used during training:
|
|
| 58 |
|
| 59 |
### Training results
|
| 60 |
|
| 61 |
-
| Training Loss | Epoch | Step |
|
| 62 |
-
|
| 63 |
-
| 0.
|
| 64 |
-
| 0.
|
| 65 |
-
| 0.
|
| 66 |
-
| 0.
|
| 67 |
-
| 0.1192 | 0.52 | 500 | 0.8077 | 0.8821 | -553.1240 | -562.3370 | 1.2602 | 0.4961 | -2.7149 | 0.0308 | -2.7458 |
|
| 68 |
-
| 0.1061 | 0.63 | 600 | 1.3570 | -3.5510 | -3.6801 | 0.5078 | 0.1291 | -655.7740 | -636.7335 | 1.4624 | 1.3499 |
|
| 69 |
-
| 0.0916 | 0.73 | 700 | 1.5923 | -4.9928 | -5.1535 | 0.5195 | 0.1607 | -803.1144 | -780.9122 | 1.8370 | 1.7244 |
|
| 70 |
-
| 0.0982 | 0.84 | 800 | 1.4367 | -4.1982 | -4.3446 | 0.5117 | 0.1464 | -722.2228 | -701.4560 | 1.5885 | 1.4960 |
|
| 71 |
-
| 0.0798 | 0.94 | 900 | 1.4971 | -4.5102 | -4.6591 | 0.5156 | 0.1490 | -753.6738 | -732.6489 | 1.5926 | 1.5057 |
|
| 72 |
|
| 73 |
|
| 74 |
### 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: 0.4801
|
| 19 |
+
- Rewards/chosen: -0.7670
|
| 20 |
+
- Rewards/rejected: -1.9120
|
| 21 |
+
- Rewards/accuracies: 0.7617
|
| 22 |
+
- Rewards/margins: 1.1450
|
| 23 |
+
- Logps/rejected: -508.7245
|
| 24 |
+
- Logps/chosen: -388.4799
|
| 25 |
+
- Logits/rejected: 0.7285
|
| 26 |
+
- Logits/chosen: 0.2273
|
| 27 |
|
| 28 |
## Model description
|
| 29 |
|
|
|
|
| 58 |
|
| 59 |
### Training results
|
| 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 |
+
| 0.539 | 0.21 | 100 | 0.5452 | -0.5988 | -1.3188 | 0.7422 | 0.7200 | -449.4053 | -371.6547 | -0.7240 | -0.8812 |
|
| 64 |
+
| 0.5294 | 0.42 | 200 | 0.5000 | -0.6654 | -1.5376 | 0.7617 | 0.8722 | -471.2849 | -378.3161 | 0.8080 | 0.4502 |
|
| 65 |
+
| 0.4704 | 0.63 | 300 | 0.4878 | -0.8441 | -1.9900 | 0.7539 | 1.1459 | -516.5240 | -396.1895 | 0.8998 | 0.3977 |
|
| 66 |
+
| 0.4856 | 0.84 | 400 | 0.4801 | -0.7670 | -1.9120 | 0.7617 | 1.1450 | -508.7245 | -388.4799 | 0.7285 | 0.2273 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 67 |
|
| 68 |
|
| 69 |
### 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":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.5266154430900158,
|
| 4 |
+
"train_runtime": 7640.9818,
|
| 5 |
+
"train_samples": 61134,
|
| 6 |
+
"train_samples_per_second": 8.001,
|
| 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:a94cc29971becb2892b676d823d5547997ac3dc41aaaff95ce9ce4444477dd87
|
| 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 sha256:275d99b6466c9a99ee5fecec6e3ddc6b5751a621eeb8e231d9e43b98c4fc6abe
|
| 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 sha256:98350e5344a6aafa12632e3c15efc90db214d41fed645b8823cfc54c914e6c79
|
| 3 |
size 4540532728
|
runs/May10_17-43-11_n136-144-082/events.out.tfevents.1715334259.n136-144-082.2699052.0
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 sha256:797b982b91a84dd5d7b9e4344de9c049335467923b34ee42247462c7fe24a70e
|
| 3 |
+
size 41033
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
-
"train_samples":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.5266154430900158,
|
| 4 |
+
"train_runtime": 7640.9818,
|
| 5 |
+
"train_samples": 61134,
|
| 6 |
+
"train_samples_per_second": 8.001,
|
| 7 |
+
"train_steps_per_second": 0.063
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,21 +1,21 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0
|
| 5 |
"eval_steps": 100,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.0,
|
| 13 |
-
"grad_norm":
|
| 14 |
-
"learning_rate":
|
| 15 |
-
"logits/chosen": -1.
|
| 16 |
-
"logits/rejected": -1.
|
| 17 |
-
"logps/chosen": -
|
| 18 |
-
"logps/rejected": -
|
| 19 |
"loss": 0.6931,
|
| 20 |
"rewards/accuracies": 0.0,
|
| 21 |
"rewards/chosen": 0.0,
|
|
@@ -23,1587 +23,787 @@
|
|
| 23 |
"rewards/rejected": 0.0,
|
| 24 |
"step": 1
|
| 25 |
},
|
| 26 |
-
{
|
| 27 |
-
"epoch": 0.01,
|
| 28 |
-
"grad_norm": 58.535794970458376,
|
| 29 |
-
"learning_rate": 5.208333333333333e-08,
|
| 30 |
-
"logits/chosen": -1.8972623348236084,
|
| 31 |
-
"logits/rejected": -1.7543258666992188,
|
| 32 |
-
"logps/chosen": -322.7024230957031,
|
| 33 |
-
"logps/rejected": -84.92481994628906,
|
| 34 |
-
"loss": 0.6929,
|
| 35 |
-
"rewards/accuracies": 0.5347222089767456,
|
| 36 |
-
"rewards/chosen": 0.0009713808540254831,
|
| 37 |
-
"rewards/margins": 0.0015221635112538934,
|
| 38 |
-
"rewards/rejected": -0.0005507825990207493,
|
| 39 |
-
"step": 10
|
| 40 |
-
},
|
| 41 |
{
|
| 42 |
"epoch": 0.02,
|
| 43 |
-
"grad_norm":
|
| 44 |
"learning_rate": 1.0416666666666667e-07,
|
| 45 |
-
"logits/chosen": -1.
|
| 46 |
-
"logits/rejected": -1.
|
| 47 |
-
"logps/chosen": -
|
| 48 |
-
"logps/rejected": -
|
| 49 |
-
"loss": 0.
|
| 50 |
-
"rewards/accuracies": 0.
|
| 51 |
-
"rewards/chosen":
|
| 52 |
-
"rewards/margins":
|
| 53 |
-
"rewards/rejected":
|
| 54 |
-
"step":
|
| 55 |
-
},
|
| 56 |
-
{
|
| 57 |
-
"epoch": 0.03,
|
| 58 |
-
"grad_norm": 44.16167777782168,
|
| 59 |
-
"learning_rate": 1.5624999999999999e-07,
|
| 60 |
-
"logits/chosen": -1.9293384552001953,
|
| 61 |
-
"logits/rejected": -1.7555296421051025,
|
| 62 |
-
"logps/chosen": -310.2709045410156,
|
| 63 |
-
"logps/rejected": -78.48434448242188,
|
| 64 |
-
"loss": 0.6291,
|
| 65 |
-
"rewards/accuracies": 0.9750000238418579,
|
| 66 |
-
"rewards/chosen": 0.08381162583827972,
|
| 67 |
-
"rewards/margins": 0.13813219964504242,
|
| 68 |
-
"rewards/rejected": -0.05432058125734329,
|
| 69 |
-
"step": 30
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.04,
|
| 73 |
-
"grad_norm":
|
| 74 |
"learning_rate": 2.0833333333333333e-07,
|
| 75 |
-
"logits/chosen": -
|
| 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":
|
| 84 |
-
"step":
|
| 85 |
-
},
|
| 86 |
-
{
|
| 87 |
-
"epoch": 0.05,
|
| 88 |
-
"grad_norm": 12.866003675141569,
|
| 89 |
-
"learning_rate": 2.604166666666667e-07,
|
| 90 |
-
"logits/chosen": -2.2248549461364746,
|
| 91 |
-
"logits/rejected": -2.144918918609619,
|
| 92 |
-
"logps/chosen": -286.402099609375,
|
| 93 |
-
"logps/rejected": -132.4581298828125,
|
| 94 |
-
"loss": 0.4557,
|
| 95 |
-
"rewards/accuracies": 0.96875,
|
| 96 |
-
"rewards/chosen": 0.10796310007572174,
|
| 97 |
-
"rewards/margins": 0.6674192547798157,
|
| 98 |
-
"rewards/rejected": -0.5594561100006104,
|
| 99 |
-
"step": 50
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.06,
|
| 103 |
-
"grad_norm":
|
| 104 |
"learning_rate": 3.1249999999999997e-07,
|
| 105 |
-
"logits/chosen": -
|
| 106 |
-
"logits/rejected": -
|
| 107 |
-
"logps/chosen": -
|
| 108 |
-
"logps/rejected": -
|
| 109 |
-
"loss": 0.
|
| 110 |
-
"rewards/accuracies": 0.
|
| 111 |
-
"rewards/chosen":
|
| 112 |
-
"rewards/margins": 0.
|
| 113 |
-
"rewards/rejected":
|
| 114 |
-
"step":
|
| 115 |
-
},
|
| 116 |
-
{
|
| 117 |
-
"epoch": 0.07,
|
| 118 |
-
"grad_norm": 10.748569520410587,
|
| 119 |
-
"learning_rate": 3.645833333333333e-07,
|
| 120 |
-
"logits/chosen": -2.430502414703369,
|
| 121 |
-
"logits/rejected": -2.365962266921997,
|
| 122 |
-
"logps/chosen": -304.8414611816406,
|
| 123 |
-
"logps/rejected": -193.85157775878906,
|
| 124 |
-
"loss": 0.3815,
|
| 125 |
-
"rewards/accuracies": 0.90625,
|
| 126 |
-
"rewards/chosen": -0.26300927996635437,
|
| 127 |
-
"rewards/margins": 0.9299274682998657,
|
| 128 |
-
"rewards/rejected": -1.192936897277832,
|
| 129 |
-
"step": 70
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.08,
|
| 133 |
-
"grad_norm":
|
| 134 |
"learning_rate": 4.1666666666666667e-07,
|
| 135 |
-
"logits/chosen": -
|
| 136 |
-
"logits/rejected": -
|
| 137 |
-
"logps/chosen": -
|
| 138 |
-
"logps/rejected": -
|
| 139 |
-
"loss": 0.
|
| 140 |
-
"rewards/accuracies": 0.
|
| 141 |
-
"rewards/chosen":
|
| 142 |
-
"rewards/margins":
|
| 143 |
-
"rewards/rejected":
|
| 144 |
-
"step":
|
| 145 |
-
},
|
| 146 |
-
{
|
| 147 |
-
"epoch": 0.09,
|
| 148 |
-
"grad_norm": 31.451743856711428,
|
| 149 |
-
"learning_rate": 4.6874999999999996e-07,
|
| 150 |
-
"logits/chosen": -2.4756340980529785,
|
| 151 |
-
"logits/rejected": -2.466632843017578,
|
| 152 |
-
"logps/chosen": -361.0640563964844,
|
| 153 |
-
"logps/rejected": -314.8011169433594,
|
| 154 |
-
"loss": 0.307,
|
| 155 |
-
"rewards/accuracies": 0.918749988079071,
|
| 156 |
-
"rewards/chosen": -0.6497557759284973,
|
| 157 |
-
"rewards/margins": 1.716461181640625,
|
| 158 |
-
"rewards/rejected": -2.3662171363830566,
|
| 159 |
-
"step": 90
|
| 160 |
-
},
|
| 161 |
-
{
|
| 162 |
-
"epoch": 0.1,
|
| 163 |
-
"grad_norm": 35.636553442388696,
|
| 164 |
-
"learning_rate": 4.999732492681437e-07,
|
| 165 |
-
"logits/chosen": -2.333749294281006,
|
| 166 |
-
"logits/rejected": -2.305887222290039,
|
| 167 |
-
"logps/chosen": -341.64007568359375,
|
| 168 |
-
"logps/rejected": -323.59344482421875,
|
| 169 |
-
"loss": 0.2876,
|
| 170 |
-
"rewards/accuracies": 0.875,
|
| 171 |
-
"rewards/chosen": -0.8234262466430664,
|
| 172 |
-
"rewards/margins": 1.7204338312149048,
|
| 173 |
-
"rewards/rejected": -2.5438601970672607,
|
| 174 |
-
"step": 100
|
| 175 |
},
|
| 176 |
{
|
| 177 |
"epoch": 0.1,
|
| 178 |
-
"
|
| 179 |
-
"
|
| 180 |
-
"
|
| 181 |
-
"
|
| 182 |
-
"
|
| 183 |
-
"
|
| 184 |
-
"
|
| 185 |
-
"
|
| 186 |
-
"
|
| 187 |
-
"
|
| 188 |
-
"
|
| 189 |
-
"
|
| 190 |
-
"step": 100
|
| 191 |
-
},
|
| 192 |
-
{
|
| 193 |
-
"epoch": 0.12,
|
| 194 |
-
"grad_norm": 23.831429692864965,
|
| 195 |
-
"learning_rate": 4.996723692767926e-07,
|
| 196 |
-
"logits/chosen": -2.3073012828826904,
|
| 197 |
-
"logits/rejected": -2.196600914001465,
|
| 198 |
-
"logps/chosen": -412.70318603515625,
|
| 199 |
-
"logps/rejected": -392.77587890625,
|
| 200 |
-
"loss": 0.2541,
|
| 201 |
-
"rewards/accuracies": 0.887499988079071,
|
| 202 |
-
"rewards/chosen": -0.8505666851997375,
|
| 203 |
-
"rewards/margins": 2.337313175201416,
|
| 204 |
-
"rewards/rejected": -3.1878795623779297,
|
| 205 |
-
"step": 110
|
| 206 |
},
|
| 207 |
{
|
| 208 |
"epoch": 0.13,
|
| 209 |
-
"grad_norm":
|
| 210 |
-
"learning_rate": 4.
|
| 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":
|
| 220 |
-
"step":
|
| 221 |
-
},
|
| 222 |
-
{
|
| 223 |
-
"epoch": 0.14,
|
| 224 |
-
"grad_norm": 23.683233547305676,
|
| 225 |
-
"learning_rate": 4.980697142834314e-07,
|
| 226 |
-
"logits/chosen": -1.82988703250885,
|
| 227 |
-
"logits/rejected": -1.7500931024551392,
|
| 228 |
-
"logps/chosen": -408.89654541015625,
|
| 229 |
-
"logps/rejected": -423.0914611816406,
|
| 230 |
-
"loss": 0.221,
|
| 231 |
-
"rewards/accuracies": 0.9437500238418579,
|
| 232 |
-
"rewards/chosen": -1.1055103540420532,
|
| 233 |
-
"rewards/margins": 2.3427820205688477,
|
| 234 |
-
"rewards/rejected": -3.4482922554016113,
|
| 235 |
-
"step": 130
|
| 236 |
},
|
| 237 |
{
|
| 238 |
"epoch": 0.15,
|
| 239 |
-
"grad_norm":
|
| 240 |
-
"learning_rate": 4.
|
| 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":
|
| 251 |
-
},
|
| 252 |
-
{
|
| 253 |
-
"epoch": 0.16,
|
| 254 |
-
"grad_norm": 35.47160662372655,
|
| 255 |
-
"learning_rate": 4.951404179843962e-07,
|
| 256 |
-
"logits/chosen": -1.420513391494751,
|
| 257 |
-
"logits/rejected": -1.204770565032959,
|
| 258 |
-
"logps/chosen": -490.7432556152344,
|
| 259 |
-
"logps/rejected": -578.1682739257812,
|
| 260 |
-
"loss": 0.1881,
|
| 261 |
-
"rewards/accuracies": 0.893750011920929,
|
| 262 |
-
"rewards/chosen": -1.6627490520477295,
|
| 263 |
-
"rewards/margins": 3.3605008125305176,
|
| 264 |
-
"rewards/rejected": -5.023249626159668,
|
| 265 |
-
"step": 150
|
| 266 |
},
|
| 267 |
{
|
| 268 |
"epoch": 0.17,
|
| 269 |
-
"grad_norm":
|
| 270 |
-
"learning_rate": 4.
|
| 271 |
-
"logits/chosen": -1.
|
| 272 |
-
"logits/rejected": -
|
| 273 |
-
"logps/chosen": -
|
| 274 |
-
"logps/rejected": -
|
| 275 |
-
"loss": 0.
|
| 276 |
-
"rewards/accuracies": 0.
|
| 277 |
-
"rewards/chosen":
|
| 278 |
-
"rewards/margins":
|
| 279 |
-
"rewards/rejected": -
|
| 280 |
-
"step":
|
| 281 |
-
},
|
| 282 |
-
{
|
| 283 |
-
"epoch": 0.18,
|
| 284 |
-
"grad_norm": 32.26038885090466,
|
| 285 |
-
"learning_rate": 4.909001458367866e-07,
|
| 286 |
-
"logits/chosen": -0.7779145240783691,
|
| 287 |
-
"logits/rejected": -0.8182721138000488,
|
| 288 |
-
"logps/chosen": -424.05731201171875,
|
| 289 |
-
"logps/rejected": -523.2410278320312,
|
| 290 |
-
"loss": 0.1927,
|
| 291 |
-
"rewards/accuracies": 0.918749988079071,
|
| 292 |
-
"rewards/chosen": -1.3222427368164062,
|
| 293 |
-
"rewards/margins": 3.1079933643341064,
|
| 294 |
-
"rewards/rejected": -4.430236339569092,
|
| 295 |
-
"step": 170
|
| 296 |
},
|
| 297 |
{
|
| 298 |
"epoch": 0.19,
|
| 299 |
-
"grad_norm":
|
| 300 |
-
"learning_rate": 4.
|
| 301 |
-
"logits/chosen": -
|
| 302 |
-
"logits/rejected": -
|
| 303 |
-
"logps/chosen": -
|
| 304 |
-
"logps/rejected": -
|
| 305 |
-
"loss": 0.
|
| 306 |
-
"rewards/accuracies": 0.
|
| 307 |
-
"rewards/chosen": -
|
| 308 |
-
"rewards/margins":
|
| 309 |
-
"rewards/rejected": -
|
| 310 |
-
"step":
|
| 311 |
-
},
|
| 312 |
-
{
|
| 313 |
-
"epoch": 0.2,
|
| 314 |
-
"grad_norm": 25.057103619592738,
|
| 315 |
-
"learning_rate": 4.853715742087946e-07,
|
| 316 |
-
"logits/chosen": -0.2753412425518036,
|
| 317 |
-
"logits/rejected": 0.2752535939216614,
|
| 318 |
-
"logps/chosen": -462.932861328125,
|
| 319 |
-
"logps/rejected": -551.9318237304688,
|
| 320 |
-
"loss": 0.1827,
|
| 321 |
-
"rewards/accuracies": 0.925000011920929,
|
| 322 |
-
"rewards/chosen": -1.5860226154327393,
|
| 323 |
-
"rewards/margins": 3.1569104194641113,
|
| 324 |
-
"rewards/rejected": -4.7429327964782715,
|
| 325 |
-
"step": 190
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 0.21,
|
| 329 |
-
"grad_norm":
|
| 330 |
-
"learning_rate": 4.
|
| 331 |
-
"logits/chosen": -0.
|
| 332 |
-
"logits/rejected": 0.
|
| 333 |
-
"logps/chosen": -
|
| 334 |
-
"logps/rejected": -
|
| 335 |
-
"loss": 0.
|
| 336 |
-
"rewards/accuracies": 0.
|
| 337 |
-
"rewards/chosen": -
|
| 338 |
-
"rewards/margins":
|
| 339 |
-
"rewards/rejected": -
|
| 340 |
-
"step":
|
| 341 |
},
|
| 342 |
{
|
| 343 |
"epoch": 0.21,
|
| 344 |
-
"eval_logits/chosen": -0.
|
| 345 |
-
"eval_logits/rejected": -0.
|
| 346 |
-
"eval_logps/chosen": -
|
| 347 |
-
"eval_logps/rejected": -
|
| 348 |
-
"eval_loss":
|
| 349 |
-
"eval_rewards/accuracies": 0.
|
| 350 |
-
"eval_rewards/chosen": -
|
| 351 |
-
"eval_rewards/margins": 0.
|
| 352 |
-
"eval_rewards/rejected": -1.
|
| 353 |
-
"eval_runtime":
|
| 354 |
-
"eval_samples_per_second": 20.
|
| 355 |
-
"eval_steps_per_second": 0.
|
| 356 |
-
"step":
|
| 357 |
-
},
|
| 358 |
-
{
|
| 359 |
-
"epoch": 0.22,
|
| 360 |
-
"grad_norm": 16.601378177571135,
|
| 361 |
-
"learning_rate": 4.785842691097342e-07,
|
| 362 |
-
"logits/chosen": 0.09680459648370743,
|
| 363 |
-
"logits/rejected": 0.48119276762008667,
|
| 364 |
-
"logps/chosen": -502.2862243652344,
|
| 365 |
-
"logps/rejected": -638.9625244140625,
|
| 366 |
-
"loss": 0.1815,
|
| 367 |
-
"rewards/accuracies": 0.8999999761581421,
|
| 368 |
-
"rewards/chosen": -2.149169445037842,
|
| 369 |
-
"rewards/margins": 3.4517273902893066,
|
| 370 |
-
"rewards/rejected": -5.600897312164307,
|
| 371 |
-
"step": 210
|
| 372 |
},
|
| 373 |
{
|
| 374 |
"epoch": 0.23,
|
| 375 |
-
"grad_norm":
|
| 376 |
-
"learning_rate": 4.
|
| 377 |
-
"logits/chosen": -0.
|
| 378 |
-
"logits/rejected": 0.
|
| 379 |
-
"logps/chosen": -
|
| 380 |
-
"logps/rejected": -
|
| 381 |
-
"loss": 0.
|
| 382 |
-
"rewards/accuracies": 0.
|
| 383 |
-
"rewards/chosen": -
|
| 384 |
-
"rewards/margins":
|
| 385 |
-
"rewards/rejected": -
|
| 386 |
-
"step":
|
| 387 |
-
},
|
| 388 |
-
{
|
| 389 |
-
"epoch": 0.24,
|
| 390 |
-
"grad_norm": 16.016947563393067,
|
| 391 |
-
"learning_rate": 4.705745280752585e-07,
|
| 392 |
-
"logits/chosen": -0.4292992651462555,
|
| 393 |
-
"logits/rejected": 0.49736231565475464,
|
| 394 |
-
"logps/chosen": -421.0140075683594,
|
| 395 |
-
"logps/rejected": -553.5887451171875,
|
| 396 |
-
"loss": 0.1643,
|
| 397 |
-
"rewards/accuracies": 0.9375,
|
| 398 |
-
"rewards/chosen": -1.2240885496139526,
|
| 399 |
-
"rewards/margins": 3.5786375999450684,
|
| 400 |
-
"rewards/rejected": -4.802725791931152,
|
| 401 |
-
"step": 230
|
| 402 |
},
|
| 403 |
{
|
| 404 |
"epoch": 0.25,
|
| 405 |
-
"grad_norm":
|
| 406 |
-
"learning_rate": 4.
|
| 407 |
-
"logits/chosen": -0.
|
| 408 |
-
"logits/rejected": -0.
|
| 409 |
-
"logps/chosen": -
|
| 410 |
-
"logps/rejected": -
|
| 411 |
-
"loss": 0.
|
| 412 |
-
"rewards/accuracies": 0.
|
| 413 |
-
"rewards/chosen": -
|
| 414 |
-
"rewards/margins":
|
| 415 |
-
"rewards/rejected": -
|
| 416 |
-
"step":
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 0.26,
|
| 420 |
-
"grad_norm": 20.36609072126483,
|
| 421 |
-
"learning_rate": 4.6138518605333664e-07,
|
| 422 |
-
"logits/chosen": -0.4158596992492676,
|
| 423 |
-
"logits/rejected": 0.18731757998466492,
|
| 424 |
-
"logps/chosen": -477.925537109375,
|
| 425 |
-
"logps/rejected": -642.4118041992188,
|
| 426 |
-
"loss": 0.1704,
|
| 427 |
-
"rewards/accuracies": 0.8999999761581421,
|
| 428 |
-
"rewards/chosen": -1.7353404760360718,
|
| 429 |
-
"rewards/margins": 3.9690043926239014,
|
| 430 |
-
"rewards/rejected": -5.704344749450684,
|
| 431 |
-
"step": 250
|
| 432 |
},
|
| 433 |
{
|
| 434 |
"epoch": 0.27,
|
| 435 |
-
"grad_norm":
|
| 436 |
-
"learning_rate": 4.
|
| 437 |
-
"logits/chosen": -0.
|
| 438 |
-
"logits/rejected": 0.
|
| 439 |
-
"logps/chosen": -
|
| 440 |
-
"logps/rejected": -
|
| 441 |
-
"loss": 0.
|
| 442 |
-
"rewards/accuracies": 0.
|
| 443 |
-
"rewards/chosen": -
|
| 444 |
-
"rewards/margins":
|
| 445 |
-
"rewards/rejected": -
|
| 446 |
-
"step":
|
| 447 |
-
},
|
| 448 |
-
{
|
| 449 |
-
"epoch": 0.28,
|
| 450 |
-
"grad_norm": 22.861972667003773,
|
| 451 |
-
"learning_rate": 4.510653863290871e-07,
|
| 452 |
-
"logits/chosen": -0.15707261860370636,
|
| 453 |
-
"logits/rejected": 0.6316308379173279,
|
| 454 |
-
"logps/chosen": -509.5392150878906,
|
| 455 |
-
"logps/rejected": -689.3701171875,
|
| 456 |
-
"loss": 0.1488,
|
| 457 |
-
"rewards/accuracies": 0.96875,
|
| 458 |
-
"rewards/chosen": -1.851678490638733,
|
| 459 |
-
"rewards/margins": 4.221504211425781,
|
| 460 |
-
"rewards/rejected": -6.073182106018066,
|
| 461 |
-
"step": 270
|
| 462 |
},
|
| 463 |
{
|
| 464 |
"epoch": 0.29,
|
| 465 |
-
"grad_norm":
|
| 466 |
-
"learning_rate": 4.
|
| 467 |
-
"logits/chosen":
|
| 468 |
-
"logits/rejected": 0.
|
| 469 |
-
"logps/chosen": -
|
| 470 |
-
"logps/rejected": -
|
| 471 |
-
"loss": 0.
|
| 472 |
-
"rewards/accuracies": 0.
|
| 473 |
-
"rewards/chosen": -
|
| 474 |
-
"rewards/margins":
|
| 475 |
-
"rewards/rejected": -
|
| 476 |
-
"step":
|
| 477 |
-
},
|
| 478 |
-
{
|
| 479 |
-
"epoch": 0.3,
|
| 480 |
-
"grad_norm": 19.170163185011805,
|
| 481 |
-
"learning_rate": 4.396703177135261e-07,
|
| 482 |
-
"logits/chosen": -0.4469999372959137,
|
| 483 |
-
"logits/rejected": 0.3809048533439636,
|
| 484 |
-
"logps/chosen": -505.42718505859375,
|
| 485 |
-
"logps/rejected": -677.1552734375,
|
| 486 |
-
"loss": 0.148,
|
| 487 |
-
"rewards/accuracies": 0.9437500238418579,
|
| 488 |
-
"rewards/chosen": -1.815896987915039,
|
| 489 |
-
"rewards/margins": 4.110238075256348,
|
| 490 |
-
"rewards/rejected": -5.926135063171387,
|
| 491 |
-
"step": 290
|
| 492 |
-
},
|
| 493 |
-
{
|
| 494 |
-
"epoch": 0.31,
|
| 495 |
-
"grad_norm": 25.4750421565511,
|
| 496 |
-
"learning_rate": 4.335883851539693e-07,
|
| 497 |
-
"logits/chosen": -0.324177086353302,
|
| 498 |
-
"logits/rejected": 0.4039342403411865,
|
| 499 |
-
"logps/chosen": -515.0848388671875,
|
| 500 |
-
"logps/rejected": -673.1795043945312,
|
| 501 |
-
"loss": 0.1485,
|
| 502 |
-
"rewards/accuracies": 0.893750011920929,
|
| 503 |
-
"rewards/chosen": -2.131357431411743,
|
| 504 |
-
"rewards/margins": 3.8231749534606934,
|
| 505 |
-
"rewards/rejected": -5.954532146453857,
|
| 506 |
-
"step": 300
|
| 507 |
},
|
| 508 |
{
|
| 509 |
"epoch": 0.31,
|
| 510 |
-
"
|
| 511 |
-
"
|
| 512 |
-
"
|
| 513 |
-
"
|
| 514 |
-
"
|
| 515 |
-
"
|
| 516 |
-
"
|
| 517 |
-
"
|
| 518 |
-
"
|
| 519 |
-
"
|
| 520 |
-
"
|
| 521 |
-
"
|
| 522 |
-
"step": 300
|
| 523 |
-
},
|
| 524 |
-
{
|
| 525 |
-
"epoch": 0.32,
|
| 526 |
-
"grad_norm": 16.440227559566175,
|
| 527 |
-
"learning_rate": 4.272609194017105e-07,
|
| 528 |
-
"logits/chosen": -0.4933062493801117,
|
| 529 |
-
"logits/rejected": 0.30621373653411865,
|
| 530 |
-
"logps/chosen": -519.3504638671875,
|
| 531 |
-
"logps/rejected": -700.8064575195312,
|
| 532 |
-
"loss": 0.1381,
|
| 533 |
-
"rewards/accuracies": 0.956250011920929,
|
| 534 |
-
"rewards/chosen": -1.7165651321411133,
|
| 535 |
-
"rewards/margins": 4.468541145324707,
|
| 536 |
-
"rewards/rejected": -6.185105323791504,
|
| 537 |
-
"step": 310
|
| 538 |
},
|
| 539 |
{
|
| 540 |
"epoch": 0.33,
|
| 541 |
-
"grad_norm":
|
| 542 |
-
"learning_rate": 4.
|
| 543 |
-
"logits/chosen":
|
| 544 |
-
"logits/rejected": 0.
|
| 545 |
-
"logps/chosen": -
|
| 546 |
-
"logps/rejected": -
|
| 547 |
-
"loss": 0.
|
| 548 |
-
"rewards/accuracies": 0.
|
| 549 |
-
"rewards/chosen": -
|
| 550 |
-
"rewards/margins":
|
| 551 |
-
"rewards/rejected": -
|
| 552 |
-
"step":
|
| 553 |
-
},
|
| 554 |
-
{
|
| 555 |
-
"epoch": 0.35,
|
| 556 |
-
"grad_norm": 24.014682889119992,
|
| 557 |
-
"learning_rate": 4.139035550786494e-07,
|
| 558 |
-
"logits/chosen": 0.16438080370426178,
|
| 559 |
-
"logits/rejected": 0.933226466178894,
|
| 560 |
-
"logps/chosen": -518.5066528320312,
|
| 561 |
-
"logps/rejected": -685.0315551757812,
|
| 562 |
-
"loss": 0.1353,
|
| 563 |
-
"rewards/accuracies": 0.893750011920929,
|
| 564 |
-
"rewards/chosen": -2.001511573791504,
|
| 565 |
-
"rewards/margins": 4.065824508666992,
|
| 566 |
-
"rewards/rejected": -6.067336559295654,
|
| 567 |
-
"step": 330
|
| 568 |
},
|
| 569 |
{
|
| 570 |
"epoch": 0.36,
|
| 571 |
-
"grad_norm":
|
| 572 |
-
"learning_rate": 4.
|
| 573 |
-
"logits/chosen":
|
| 574 |
-
"logits/rejected": 0.
|
| 575 |
-
"logps/chosen": -
|
| 576 |
-
"logps/rejected": -
|
| 577 |
-
"loss": 0.
|
| 578 |
-
"rewards/accuracies": 0.
|
| 579 |
-
"rewards/chosen": -
|
| 580 |
-
"rewards/margins":
|
| 581 |
-
"rewards/rejected": -
|
| 582 |
-
"step":
|
| 583 |
-
},
|
| 584 |
-
{
|
| 585 |
-
"epoch": 0.37,
|
| 586 |
-
"grad_norm": 27.161000273751988,
|
| 587 |
-
"learning_rate": 3.99669658015821e-07,
|
| 588 |
-
"logits/chosen": -0.5287201404571533,
|
| 589 |
-
"logits/rejected": 0.0556589737534523,
|
| 590 |
-
"logps/chosen": -524.2675170898438,
|
| 591 |
-
"logps/rejected": -735.3182373046875,
|
| 592 |
-
"loss": 0.1433,
|
| 593 |
-
"rewards/accuracies": 0.918749988079071,
|
| 594 |
-
"rewards/chosen": -2.134932041168213,
|
| 595 |
-
"rewards/margins": 4.41909646987915,
|
| 596 |
-
"rewards/rejected": -6.554028511047363,
|
| 597 |
-
"step": 350
|
| 598 |
},
|
| 599 |
{
|
| 600 |
"epoch": 0.38,
|
| 601 |
-
"grad_norm":
|
| 602 |
-
"learning_rate": 3.
|
| 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":
|
| 611 |
-
"rewards/rejected": -
|
| 612 |
-
"step":
|
| 613 |
-
},
|
| 614 |
-
{
|
| 615 |
-
"epoch": 0.39,
|
| 616 |
-
"grad_norm": 19.030144049954593,
|
| 617 |
-
"learning_rate": 3.846353490562664e-07,
|
| 618 |
-
"logits/chosen": -0.14467313885688782,
|
| 619 |
-
"logits/rejected": 0.4623333513736725,
|
| 620 |
-
"logps/chosen": -475.9998474121094,
|
| 621 |
-
"logps/rejected": -659.7310791015625,
|
| 622 |
-
"loss": 0.1378,
|
| 623 |
-
"rewards/accuracies": 0.9125000238418579,
|
| 624 |
-
"rewards/chosen": -1.9292904138565063,
|
| 625 |
-
"rewards/margins": 3.9880707263946533,
|
| 626 |
-
"rewards/rejected": -5.917361259460449,
|
| 627 |
-
"step": 370
|
| 628 |
},
|
| 629 |
{
|
| 630 |
"epoch": 0.4,
|
| 631 |
-
"grad_norm":
|
| 632 |
-
"learning_rate": 3.
|
| 633 |
-
"logits/chosen":
|
| 634 |
-
"logits/rejected": 0.
|
| 635 |
-
"logps/chosen": -
|
| 636 |
-
"logps/rejected": -
|
| 637 |
-
"loss": 0.
|
| 638 |
-
"rewards/accuracies": 0.
|
| 639 |
-
"rewards/chosen": -
|
| 640 |
-
"rewards/margins":
|
| 641 |
-
"rewards/rejected": -
|
| 642 |
-
"step":
|
| 643 |
-
},
|
| 644 |
-
{
|
| 645 |
-
"epoch": 0.41,
|
| 646 |
-
"grad_norm": 20.594348598442483,
|
| 647 |
-
"learning_rate": 3.6888102953122304e-07,
|
| 648 |
-
"logits/chosen": -0.44062256813049316,
|
| 649 |
-
"logits/rejected": 0.3265061378479004,
|
| 650 |
-
"logps/chosen": -475.5601501464844,
|
| 651 |
-
"logps/rejected": -694.0905151367188,
|
| 652 |
-
"loss": 0.132,
|
| 653 |
-
"rewards/accuracies": 0.96875,
|
| 654 |
-
"rewards/chosen": -1.7453269958496094,
|
| 655 |
-
"rewards/margins": 4.402843475341797,
|
| 656 |
-
"rewards/rejected": -6.148170471191406,
|
| 657 |
-
"step": 390
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 0.42,
|
| 661 |
-
"grad_norm":
|
| 662 |
-
"learning_rate": 3.
|
| 663 |
-
"logits/chosen":
|
| 664 |
-
"logits/rejected": 0.
|
| 665 |
-
"logps/chosen": -
|
| 666 |
-
"logps/rejected": -
|
| 667 |
-
"loss": 0.
|
| 668 |
-
"rewards/accuracies": 0.
|
| 669 |
-
"rewards/chosen": -
|
| 670 |
-
"rewards/margins":
|
| 671 |
-
"rewards/rejected": -
|
| 672 |
-
"step":
|
| 673 |
},
|
| 674 |
{
|
| 675 |
"epoch": 0.42,
|
| 676 |
-
"eval_logits/chosen":
|
| 677 |
-
"eval_logits/rejected":
|
| 678 |
-
"eval_logps/chosen": -
|
| 679 |
-
"eval_logps/rejected": -
|
| 680 |
-
"eval_loss":
|
| 681 |
-
"eval_rewards/accuracies": 0.
|
| 682 |
-
"eval_rewards/chosen": -
|
| 683 |
-
"eval_rewards/margins": 0.
|
| 684 |
-
"eval_rewards/rejected": -1.
|
| 685 |
-
"eval_runtime":
|
| 686 |
-
"eval_samples_per_second": 20.
|
| 687 |
-
"eval_steps_per_second": 0.
|
| 688 |
-
"step":
|
| 689 |
-
},
|
| 690 |
-
{
|
| 691 |
-
"epoch": 0.43,
|
| 692 |
-
"grad_norm": 27.719873706295672,
|
| 693 |
-
"learning_rate": 3.5249095128531856e-07,
|
| 694 |
-
"logits/chosen": -0.22032999992370605,
|
| 695 |
-
"logits/rejected": 0.5274828672409058,
|
| 696 |
-
"logps/chosen": -466.26702880859375,
|
| 697 |
-
"logps/rejected": -728.7267456054688,
|
| 698 |
-
"loss": 0.1355,
|
| 699 |
-
"rewards/accuracies": 0.949999988079071,
|
| 700 |
-
"rewards/chosen": -1.8140884637832642,
|
| 701 |
-
"rewards/margins": 4.6839799880981445,
|
| 702 |
-
"rewards/rejected": -6.498068332672119,
|
| 703 |
-
"step": 410
|
| 704 |
},
|
| 705 |
{
|
| 706 |
"epoch": 0.44,
|
| 707 |
-
"grad_norm":
|
| 708 |
-
"learning_rate": 3.
|
| 709 |
-
"logits/chosen":
|
| 710 |
-
"logits/rejected": 0.
|
| 711 |
-
"logps/chosen": -
|
| 712 |
-
"logps/rejected": -
|
| 713 |
-
"loss": 0.
|
| 714 |
-
"rewards/accuracies": 0.
|
| 715 |
-
"rewards/chosen": -
|
| 716 |
-
"rewards/margins":
|
| 717 |
-
"rewards/rejected": -
|
| 718 |
-
"step":
|
| 719 |
-
},
|
| 720 |
-
{
|
| 721 |
-
"epoch": 0.45,
|
| 722 |
-
"grad_norm": 15.065368811496283,
|
| 723 |
-
"learning_rate": 3.3555276610977276e-07,
|
| 724 |
-
"logits/chosen": 0.07787096500396729,
|
| 725 |
-
"logits/rejected": 1.0413458347320557,
|
| 726 |
-
"logps/chosen": -519.1903686523438,
|
| 727 |
-
"logps/rejected": -721.7469482421875,
|
| 728 |
-
"loss": 0.1303,
|
| 729 |
-
"rewards/accuracies": 0.956250011920929,
|
| 730 |
-
"rewards/chosen": -2.1075587272644043,
|
| 731 |
-
"rewards/margins": 4.277621269226074,
|
| 732 |
-
"rewards/rejected": -6.385180473327637,
|
| 733 |
-
"step": 430
|
| 734 |
},
|
| 735 |
{
|
| 736 |
"epoch": 0.46,
|
| 737 |
-
"grad_norm":
|
| 738 |
-
"learning_rate": 3.
|
| 739 |
-
"logits/chosen": 0.
|
| 740 |
-
"logits/rejected": 0.
|
| 741 |
-
"logps/chosen": -
|
| 742 |
-
"logps/rejected": -
|
| 743 |
-
"loss": 0.
|
| 744 |
-
"rewards/accuracies": 0.
|
| 745 |
-
"rewards/chosen": -
|
| 746 |
-
"rewards/margins":
|
| 747 |
-
"rewards/rejected": -
|
| 748 |
-
"step":
|
| 749 |
-
},
|
| 750 |
-
{
|
| 751 |
-
"epoch": 0.47,
|
| 752 |
-
"grad_norm": 27.318228969564636,
|
| 753 |
-
"learning_rate": 3.1815705699316964e-07,
|
| 754 |
-
"logits/chosen": 0.6579626202583313,
|
| 755 |
-
"logits/rejected": 1.765021562576294,
|
| 756 |
-
"logps/chosen": -553.9263305664062,
|
| 757 |
-
"logps/rejected": -844.2288208007812,
|
| 758 |
-
"loss": 0.1277,
|
| 759 |
-
"rewards/accuracies": 0.9312499761581421,
|
| 760 |
-
"rewards/chosen": -2.427884578704834,
|
| 761 |
-
"rewards/margins": 5.219451904296875,
|
| 762 |
-
"rewards/rejected": -7.647336483001709,
|
| 763 |
-
"step": 450
|
| 764 |
},
|
| 765 |
{
|
| 766 |
"epoch": 0.48,
|
| 767 |
-
"grad_norm":
|
| 768 |
-
"learning_rate": 3.
|
| 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":
|
| 779 |
-
},
|
| 780 |
-
{
|
| 781 |
-
"epoch": 0.49,
|
| 782 |
-
"grad_norm": 21.248985584336232,
|
| 783 |
-
"learning_rate": 3.003968536966078e-07,
|
| 784 |
-
"logits/chosen": 0.6433358788490295,
|
| 785 |
-
"logits/rejected": 1.6070924997329712,
|
| 786 |
-
"logps/chosen": -555.550048828125,
|
| 787 |
-
"logps/rejected": -811.1224365234375,
|
| 788 |
-
"loss": 0.1362,
|
| 789 |
-
"rewards/accuracies": 0.9312499761581421,
|
| 790 |
-
"rewards/chosen": -2.4847140312194824,
|
| 791 |
-
"rewards/margins": 4.877633571624756,
|
| 792 |
-
"rewards/rejected": -7.3623480796813965,
|
| 793 |
-
"step": 470
|
| 794 |
},
|
| 795 |
{
|
| 796 |
"epoch": 0.5,
|
| 797 |
-
"grad_norm":
|
| 798 |
-
"learning_rate": 2.
|
| 799 |
-
"logits/chosen": 0.
|
| 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":
|
| 809 |
-
},
|
| 810 |
-
{
|
| 811 |
-
"epoch": 0.51,
|
| 812 |
-
"grad_norm": 32.32085610980076,
|
| 813 |
-
"learning_rate": 2.823671352438608e-07,
|
| 814 |
-
"logits/chosen": 0.9727161526679993,
|
| 815 |
-
"logits/rejected": 2.0250372886657715,
|
| 816 |
-
"logps/chosen": -659.5581665039062,
|
| 817 |
-
"logps/rejected": -1049.233154296875,
|
| 818 |
-
"loss": 0.1095,
|
| 819 |
-
"rewards/accuracies": 0.956250011920929,
|
| 820 |
-
"rewards/chosen": -3.477785110473633,
|
| 821 |
-
"rewards/margins": 6.155587673187256,
|
| 822 |
-
"rewards/rejected": -9.63337230682373,
|
| 823 |
-
"step": 490
|
| 824 |
-
},
|
| 825 |
-
{
|
| 826 |
-
"epoch": 0.52,
|
| 827 |
-
"grad_norm": 20.87674341195024,
|
| 828 |
-
"learning_rate": 2.73281296951072e-07,
|
| 829 |
-
"logits/chosen": 0.9520805478096008,
|
| 830 |
-
"logits/rejected": 2.0586464405059814,
|
| 831 |
-
"logps/chosen": -609.9530029296875,
|
| 832 |
-
"logps/rejected": -901.1349487304688,
|
| 833 |
-
"loss": 0.1192,
|
| 834 |
-
"rewards/accuracies": 0.925000011920929,
|
| 835 |
-
"rewards/chosen": -3.0585007667541504,
|
| 836 |
-
"rewards/margins": 5.11794376373291,
|
| 837 |
-
"rewards/rejected": -8.176445007324219,
|
| 838 |
-
"step": 500
|
| 839 |
},
|
| 840 |
{
|
| 841 |
"epoch": 0.52,
|
| 842 |
-
"
|
| 843 |
-
"
|
| 844 |
-
"
|
| 845 |
-
"
|
| 846 |
-
"
|
| 847 |
-
"
|
| 848 |
-
"
|
| 849 |
-
"
|
| 850 |
-
"
|
| 851 |
-
"
|
| 852 |
-
"
|
| 853 |
-
"
|
| 854 |
-
"step": 500
|
| 855 |
-
},
|
| 856 |
-
{
|
| 857 |
-
"epoch": 0.53,
|
| 858 |
-
"grad_norm": 18.767151683387205,
|
| 859 |
-
"learning_rate": 2.641643219871597e-07,
|
| 860 |
-
"logits/chosen": 0.7849302887916565,
|
| 861 |
-
"logits/rejected": 2.1215097904205322,
|
| 862 |
-
"logps/chosen": -577.2518920898438,
|
| 863 |
-
"logps/rejected": -889.7449340820312,
|
| 864 |
-
"loss": 0.1317,
|
| 865 |
-
"rewards/accuracies": 0.949999988079071,
|
| 866 |
-
"rewards/chosen": -2.842534303665161,
|
| 867 |
-
"rewards/margins": 5.340787887573242,
|
| 868 |
-
"rewards/rejected": -8.183321952819824,
|
| 869 |
-
"step": 510
|
| 870 |
},
|
| 871 |
{
|
| 872 |
"epoch": 0.54,
|
| 873 |
-
"grad_norm":
|
| 874 |
-
"learning_rate": 2.
|
| 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":
|
| 883 |
-
"rewards/rejected": -
|
| 884 |
-
"step":
|
| 885 |
-
},
|
| 886 |
-
{
|
| 887 |
-
"epoch": 0.55,
|
| 888 |
-
"grad_norm": 27.252220894010925,
|
| 889 |
-
"learning_rate": 2.4588575996495794e-07,
|
| 890 |
-
"logits/chosen": 1.095752239227295,
|
| 891 |
-
"logits/rejected": 1.7183897495269775,
|
| 892 |
-
"logps/chosen": -542.5738525390625,
|
| 893 |
-
"logps/rejected": -803.3109130859375,
|
| 894 |
-
"loss": 0.1195,
|
| 895 |
-
"rewards/accuracies": 0.9375,
|
| 896 |
-
"rewards/chosen": -2.78588604927063,
|
| 897 |
-
"rewards/margins": 4.481093883514404,
|
| 898 |
-
"rewards/rejected": -7.266980171203613,
|
| 899 |
-
"step": 530
|
| 900 |
-
},
|
| 901 |
-
{
|
| 902 |
-
"epoch": 0.57,
|
| 903 |
-
"grad_norm": 22.12136172682442,
|
| 904 |
-
"learning_rate": 2.367486188632446e-07,
|
| 905 |
-
"logits/chosen": 0.4685417711734772,
|
| 906 |
-
"logits/rejected": 2.220308542251587,
|
| 907 |
-
"logps/chosen": -570.0831298828125,
|
| 908 |
-
"logps/rejected": -881.7498779296875,
|
| 909 |
-
"loss": 0.1043,
|
| 910 |
-
"rewards/accuracies": 0.9624999761581421,
|
| 911 |
-
"rewards/chosen": -2.479081630706787,
|
| 912 |
-
"rewards/margins": 5.567392349243164,
|
| 913 |
-
"rewards/rejected": -8.046473503112793,
|
| 914 |
-
"step": 540
|
| 915 |
},
|
| 916 |
{
|
| 917 |
-
"epoch": 0.
|
| 918 |
-
"grad_norm":
|
| 919 |
-
"learning_rate": 2.
|
| 920 |
-
"logits/chosen":
|
| 921 |
-
"logits/rejected":
|
| 922 |
-
"logps/chosen": -
|
| 923 |
-
"logps/rejected": -
|
| 924 |
-
"loss": 0.
|
| 925 |
-
"rewards/accuracies": 0.
|
| 926 |
-
"rewards/chosen": -
|
| 927 |
-
"rewards/margins":
|
| 928 |
-
"rewards/rejected": -
|
| 929 |
-
"step":
|
| 930 |
},
|
| 931 |
{
|
| 932 |
"epoch": 0.59,
|
| 933 |
-
"grad_norm":
|
| 934 |
-
"learning_rate": 2.
|
| 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":
|
| 943 |
-
"rewards/rejected": -
|
| 944 |
-
"step":
|
| 945 |
-
},
|
| 946 |
-
{
|
| 947 |
-
"epoch": 0.6,
|
| 948 |
-
"grad_norm": 16.198596144621373,
|
| 949 |
-
"learning_rate": 2.094922764865619e-07,
|
| 950 |
-
"logits/chosen": 1.3269495964050293,
|
| 951 |
-
"logits/rejected": 2.8318161964416504,
|
| 952 |
-
"logps/chosen": -717.6715087890625,
|
| 953 |
-
"logps/rejected": -1159.4893798828125,
|
| 954 |
-
"loss": 0.1079,
|
| 955 |
-
"rewards/accuracies": 0.9312499761581421,
|
| 956 |
-
"rewards/chosen": -4.280415058135986,
|
| 957 |
-
"rewards/margins": 6.474959373474121,
|
| 958 |
-
"rewards/rejected": -10.755374908447266,
|
| 959 |
-
"step": 570
|
| 960 |
},
|
| 961 |
{
|
| 962 |
"epoch": 0.61,
|
| 963 |
-
"grad_norm":
|
| 964 |
-
"learning_rate": 2.
|
| 965 |
-
"logits/chosen": 0.
|
| 966 |
-
"logits/rejected":
|
| 967 |
-
"logps/chosen": -
|
| 968 |
-
"logps/rejected": -
|
| 969 |
-
"loss": 0.
|
| 970 |
-
"rewards/accuracies": 0.
|
| 971 |
-
"rewards/chosen": -
|
| 972 |
-
"rewards/margins":
|
| 973 |
-
"rewards/rejected": -
|
| 974 |
-
"step":
|
| 975 |
-
},
|
| 976 |
-
{
|
| 977 |
-
"epoch": 0.62,
|
| 978 |
-
"grad_norm": 23.433263062431486,
|
| 979 |
-
"learning_rate": 1.9157198216806238e-07,
|
| 980 |
-
"logits/chosen": 1.8187183141708374,
|
| 981 |
-
"logits/rejected": 2.8524560928344727,
|
| 982 |
-
"logps/chosen": -665.5374755859375,
|
| 983 |
-
"logps/rejected": -994.6904296875,
|
| 984 |
-
"loss": 0.1248,
|
| 985 |
-
"rewards/accuracies": 0.887499988079071,
|
| 986 |
-
"rewards/chosen": -4.142047882080078,
|
| 987 |
-
"rewards/margins": 5.089578151702881,
|
| 988 |
-
"rewards/rejected": -9.231626510620117,
|
| 989 |
-
"step": 590
|
| 990 |
},
|
| 991 |
{
|
| 992 |
"epoch": 0.63,
|
| 993 |
-
"grad_norm":
|
| 994 |
-
"learning_rate": 1.
|
| 995 |
-
"logits/chosen":
|
| 996 |
-
"logits/rejected":
|
| 997 |
-
"logps/chosen": -
|
| 998 |
-
"logps/rejected": -
|
| 999 |
-
"loss": 0.
|
| 1000 |
-
"rewards/accuracies": 0.
|
| 1001 |
-
"rewards/chosen": -
|
| 1002 |
-
"rewards/margins":
|
| 1003 |
-
"rewards/rejected": -
|
| 1004 |
-
"step":
|
| 1005 |
},
|
| 1006 |
{
|
| 1007 |
"epoch": 0.63,
|
| 1008 |
-
"eval_logits/chosen":
|
| 1009 |
-
"eval_logits/rejected":
|
| 1010 |
-
"eval_logps/chosen": -
|
| 1011 |
-
"eval_logps/rejected": -
|
| 1012 |
-
"eval_loss":
|
| 1013 |
-
"eval_rewards/accuracies": 0.
|
| 1014 |
-
"eval_rewards/chosen": -
|
| 1015 |
-
"eval_rewards/margins":
|
| 1016 |
-
"eval_rewards/rejected": -
|
| 1017 |
-
"eval_runtime":
|
| 1018 |
-
"eval_samples_per_second": 20.
|
| 1019 |
-
"eval_steps_per_second": 0.
|
| 1020 |
-
"step":
|
| 1021 |
-
},
|
| 1022 |
-
{
|
| 1023 |
-
"epoch": 0.64,
|
| 1024 |
-
"grad_norm": 24.852031957562648,
|
| 1025 |
-
"learning_rate": 1.7396415252139288e-07,
|
| 1026 |
-
"logits/chosen": 2.000277042388916,
|
| 1027 |
-
"logits/rejected": 3.6395816802978516,
|
| 1028 |
-
"logps/chosen": -704.3046875,
|
| 1029 |
-
"logps/rejected": -1072.0576171875,
|
| 1030 |
-
"loss": 0.1011,
|
| 1031 |
-
"rewards/accuracies": 0.918749988079071,
|
| 1032 |
-
"rewards/chosen": -4.077065467834473,
|
| 1033 |
-
"rewards/margins": 5.8812713623046875,
|
| 1034 |
-
"rewards/rejected": -9.958337783813477,
|
| 1035 |
-
"step": 610
|
| 1036 |
},
|
| 1037 |
{
|
| 1038 |
"epoch": 0.65,
|
| 1039 |
-
"grad_norm":
|
| 1040 |
-
"learning_rate": 1.
|
| 1041 |
-
"logits/chosen":
|
| 1042 |
-
"logits/rejected":
|
| 1043 |
-
"logps/chosen": -
|
| 1044 |
-
"logps/rejected": -
|
| 1045 |
-
"loss": 0.
|
| 1046 |
-
"rewards/accuracies": 0.
|
| 1047 |
-
"rewards/chosen": -
|
| 1048 |
-
"rewards/margins":
|
| 1049 |
-
"rewards/rejected": -
|
| 1050 |
-
"step":
|
| 1051 |
-
},
|
| 1052 |
-
{
|
| 1053 |
-
"epoch": 0.66,
|
| 1054 |
-
"grad_norm": 23.63694930913145,
|
| 1055 |
-
"learning_rate": 1.5676295169786864e-07,
|
| 1056 |
-
"logits/chosen": 0.5527520775794983,
|
| 1057 |
-
"logits/rejected": 2.127023696899414,
|
| 1058 |
-
"logps/chosen": -621.4510498046875,
|
| 1059 |
-
"logps/rejected": -956.1905517578125,
|
| 1060 |
-
"loss": 0.0982,
|
| 1061 |
-
"rewards/accuracies": 0.949999988079071,
|
| 1062 |
-
"rewards/chosen": -3.0273334980010986,
|
| 1063 |
-
"rewards/margins": 5.792135715484619,
|
| 1064 |
-
"rewards/rejected": -8.819470405578613,
|
| 1065 |
-
"step": 630
|
| 1066 |
},
|
| 1067 |
{
|
| 1068 |
"epoch": 0.67,
|
| 1069 |
-
"grad_norm":
|
| 1070 |
-
"learning_rate": 1.
|
| 1071 |
-
"logits/chosen":
|
| 1072 |
-
"logits/rejected":
|
| 1073 |
-
"logps/chosen": -
|
| 1074 |
-
"logps/rejected": -
|
| 1075 |
-
"loss": 0.
|
| 1076 |
-
"rewards/accuracies": 0.
|
| 1077 |
-
"rewards/chosen": -
|
| 1078 |
-
"rewards/margins":
|
| 1079 |
-
"rewards/rejected": -
|
| 1080 |
-
"step":
|
| 1081 |
-
},
|
| 1082 |
-
{
|
| 1083 |
-
"epoch": 0.68,
|
| 1084 |
-
"grad_norm": 30.835634235164253,
|
| 1085 |
-
"learning_rate": 1.4006036925609243e-07,
|
| 1086 |
-
"logits/chosen": 1.4370182752609253,
|
| 1087 |
-
"logits/rejected": 2.603792428970337,
|
| 1088 |
-
"logps/chosen": -682.9066772460938,
|
| 1089 |
-
"logps/rejected": -1069.502685546875,
|
| 1090 |
-
"loss": 0.1096,
|
| 1091 |
-
"rewards/accuracies": 0.918749988079071,
|
| 1092 |
-
"rewards/chosen": -3.886139392852783,
|
| 1093 |
-
"rewards/margins": 6.004186630249023,
|
| 1094 |
-
"rewards/rejected": -9.890324592590332,
|
| 1095 |
-
"step": 650
|
| 1096 |
},
|
| 1097 |
{
|
| 1098 |
"epoch": 0.69,
|
| 1099 |
-
"grad_norm":
|
| 1100 |
-
"learning_rate": 1.
|
| 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":
|
| 1109 |
-
"rewards/rejected": -
|
| 1110 |
-
"step":
|
| 1111 |
-
},
|
| 1112 |
-
{
|
| 1113 |
-
"epoch": 0.7,
|
| 1114 |
-
"grad_norm": 23.254329026411572,
|
| 1115 |
-
"learning_rate": 1.239457282149695e-07,
|
| 1116 |
-
"logits/chosen": 1.9267222881317139,
|
| 1117 |
-
"logits/rejected": 3.357774257659912,
|
| 1118 |
-
"logps/chosen": -649.8111572265625,
|
| 1119 |
-
"logps/rejected": -1027.4486083984375,
|
| 1120 |
-
"loss": 0.1018,
|
| 1121 |
-
"rewards/accuracies": 0.96875,
|
| 1122 |
-
"rewards/chosen": -3.9152133464813232,
|
| 1123 |
-
"rewards/margins": 5.612493991851807,
|
| 1124 |
-
"rewards/rejected": -9.527708053588867,
|
| 1125 |
-
"step": 670
|
| 1126 |
},
|
| 1127 |
{
|
| 1128 |
"epoch": 0.71,
|
| 1129 |
-
"grad_norm":
|
| 1130 |
-
"learning_rate": 1.
|
| 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":
|
| 1141 |
-
},
|
| 1142 |
-
{
|
| 1143 |
-
"epoch": 0.72,
|
| 1144 |
-
"grad_norm": 24.122068893750104,
|
| 1145 |
-
"learning_rate": 1.0850520736699362e-07,
|
| 1146 |
-
"logits/chosen": 1.7467336654663086,
|
| 1147 |
-
"logits/rejected": 2.8030543327331543,
|
| 1148 |
-
"logps/chosen": -693.4568481445312,
|
| 1149 |
-
"logps/rejected": -1148.7037353515625,
|
| 1150 |
-
"loss": 0.0942,
|
| 1151 |
-
"rewards/accuracies": 0.9437500238418579,
|
| 1152 |
-
"rewards/chosen": -4.0760016441345215,
|
| 1153 |
-
"rewards/margins": 6.649557590484619,
|
| 1154 |
-
"rewards/rejected": -10.72555923461914,
|
| 1155 |
-
"step": 690
|
| 1156 |
-
},
|
| 1157 |
-
{
|
| 1158 |
-
"epoch": 0.73,
|
| 1159 |
-
"grad_norm": 29.944632705362917,
|
| 1160 |
-
"learning_rate": 1.0106369933615042e-07,
|
| 1161 |
-
"logits/chosen": 1.8298194408416748,
|
| 1162 |
-
"logits/rejected": 2.8535187244415283,
|
| 1163 |
-
"logps/chosen": -790.9088134765625,
|
| 1164 |
-
"logps/rejected": -1210.2066650390625,
|
| 1165 |
-
"loss": 0.0916,
|
| 1166 |
-
"rewards/accuracies": 0.9624999761581421,
|
| 1167 |
-
"rewards/chosen": -4.785300254821777,
|
| 1168 |
-
"rewards/margins": 6.468996524810791,
|
| 1169 |
-
"rewards/rejected": -11.254297256469727,
|
| 1170 |
-
"step": 700
|
| 1171 |
},
|
| 1172 |
{
|
| 1173 |
"epoch": 0.73,
|
| 1174 |
-
"
|
| 1175 |
-
"
|
| 1176 |
-
"
|
| 1177 |
-
"
|
| 1178 |
-
"
|
| 1179 |
-
"
|
| 1180 |
-
"
|
| 1181 |
-
"
|
| 1182 |
-
"
|
| 1183 |
-
"
|
| 1184 |
-
"
|
| 1185 |
-
"
|
| 1186 |
-
"step": 700
|
| 1187 |
-
},
|
| 1188 |
-
{
|
| 1189 |
-
"epoch": 0.74,
|
| 1190 |
-
"grad_norm": 20.908723209277646,
|
| 1191 |
-
"learning_rate": 9.382138040640714e-08,
|
| 1192 |
-
"logits/chosen": 1.1189043521881104,
|
| 1193 |
-
"logits/rejected": 2.7733755111694336,
|
| 1194 |
-
"logps/chosen": -770.1724853515625,
|
| 1195 |
-
"logps/rejected": -1226.529052734375,
|
| 1196 |
-
"loss": 0.0963,
|
| 1197 |
-
"rewards/accuracies": 0.9750000238418579,
|
| 1198 |
-
"rewards/chosen": -4.310009956359863,
|
| 1199 |
-
"rewards/margins": 7.130062103271484,
|
| 1200 |
-
"rewards/rejected": -11.440072059631348,
|
| 1201 |
-
"step": 710
|
| 1202 |
},
|
| 1203 |
{
|
| 1204 |
"epoch": 0.75,
|
| 1205 |
-
"grad_norm":
|
| 1206 |
-
"learning_rate": 8.
|
| 1207 |
-
"logits/chosen": 0.
|
| 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": -
|
| 1216 |
-
"step":
|
| 1217 |
-
},
|
| 1218 |
-
{
|
| 1219 |
-
"epoch": 0.76,
|
| 1220 |
-
"grad_norm": 31.693149214366326,
|
| 1221 |
-
"learning_rate": 7.997277433690983e-08,
|
| 1222 |
-
"logits/chosen": 1.5684382915496826,
|
| 1223 |
-
"logits/rejected": 2.5911812782287598,
|
| 1224 |
-
"logps/chosen": -716.12255859375,
|
| 1225 |
-
"logps/rejected": -1081.2698974609375,
|
| 1226 |
-
"loss": 0.1075,
|
| 1227 |
-
"rewards/accuracies": 0.925000011920929,
|
| 1228 |
-
"rewards/chosen": -4.1696271896362305,
|
| 1229 |
-
"rewards/margins": 5.879700660705566,
|
| 1230 |
-
"rewards/rejected": -10.049327850341797,
|
| 1231 |
-
"step": 730
|
| 1232 |
},
|
| 1233 |
{
|
| 1234 |
"epoch": 0.77,
|
| 1235 |
-
"grad_norm":
|
| 1236 |
-
"learning_rate": 7.
|
| 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":
|
| 1245 |
-
"rewards/rejected": -
|
| 1246 |
-
"step":
|
| 1247 |
-
},
|
| 1248 |
-
{
|
| 1249 |
-
"epoch": 0.78,
|
| 1250 |
-
"grad_norm": 39.0366898564141,
|
| 1251 |
-
"learning_rate": 6.70334495204884e-08,
|
| 1252 |
-
"logits/chosen": 1.2770452499389648,
|
| 1253 |
-
"logits/rejected": 3.155823230743408,
|
| 1254 |
-
"logps/chosen": -781.6085205078125,
|
| 1255 |
-
"logps/rejected": -1216.711181640625,
|
| 1256 |
-
"loss": 0.1008,
|
| 1257 |
-
"rewards/accuracies": 0.949999988079071,
|
| 1258 |
-
"rewards/chosen": -4.443601131439209,
|
| 1259 |
-
"rewards/margins": 6.983972072601318,
|
| 1260 |
-
"rewards/rejected": -11.427572250366211,
|
| 1261 |
-
"step": 750
|
| 1262 |
-
},
|
| 1263 |
-
{
|
| 1264 |
-
"epoch": 0.8,
|
| 1265 |
-
"grad_norm": 15.124341176010748,
|
| 1266 |
-
"learning_rate": 6.092659210462231e-08,
|
| 1267 |
-
"logits/chosen": 2.077390670776367,
|
| 1268 |
-
"logits/rejected": 3.1247870922088623,
|
| 1269 |
-
"logps/chosen": -768.5572509765625,
|
| 1270 |
-
"logps/rejected": -1171.698974609375,
|
| 1271 |
-
"loss": 0.1056,
|
| 1272 |
-
"rewards/accuracies": 0.9624999761581421,
|
| 1273 |
-
"rewards/chosen": -4.7178215980529785,
|
| 1274 |
-
"rewards/margins": 6.16521692276001,
|
| 1275 |
-
"rewards/rejected": -10.883038520812988,
|
| 1276 |
-
"step": 760
|
| 1277 |
},
|
| 1278 |
{
|
| 1279 |
-
"epoch": 0.
|
| 1280 |
-
"grad_norm":
|
| 1281 |
-
"learning_rate":
|
| 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":
|
| 1290 |
-
"rewards/rejected": -
|
| 1291 |
-
"step":
|
| 1292 |
},
|
| 1293 |
{
|
| 1294 |
"epoch": 0.82,
|
| 1295 |
-
"grad_norm":
|
| 1296 |
-
"learning_rate": 4.
|
| 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":
|
| 1307 |
-
},
|
| 1308 |
-
{
|
| 1309 |
-
"epoch": 0.83,
|
| 1310 |
-
"grad_norm": 18.290673124323764,
|
| 1311 |
-
"learning_rate": 4.415420150605398e-08,
|
| 1312 |
-
"logits/chosen": 1.273377776145935,
|
| 1313 |
-
"logits/rejected": 2.5009658336639404,
|
| 1314 |
-
"logps/chosen": -672.8284301757812,
|
| 1315 |
-
"logps/rejected": -1044.4967041015625,
|
| 1316 |
-
"loss": 0.0833,
|
| 1317 |
-
"rewards/accuracies": 0.96875,
|
| 1318 |
-
"rewards/chosen": -3.731667995452881,
|
| 1319 |
-
"rewards/margins": 5.8987932205200195,
|
| 1320 |
-
"rewards/rejected": -9.630460739135742,
|
| 1321 |
-
"step": 790
|
| 1322 |
},
|
| 1323 |
{
|
| 1324 |
"epoch": 0.84,
|
| 1325 |
-
"grad_norm":
|
| 1326 |
-
"learning_rate": 3.
|
| 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":
|
| 1337 |
},
|
| 1338 |
{
|
| 1339 |
"epoch": 0.84,
|
| 1340 |
-
"eval_logits/chosen":
|
| 1341 |
-
"eval_logits/rejected":
|
| 1342 |
-
"eval_logps/chosen": -
|
| 1343 |
-
"eval_logps/rejected": -
|
| 1344 |
-
"eval_loss":
|
| 1345 |
-
"eval_rewards/accuracies": 0.
|
| 1346 |
-
"eval_rewards/chosen": -
|
| 1347 |
-
"eval_rewards/margins":
|
| 1348 |
-
"eval_rewards/rejected": -
|
| 1349 |
-
"eval_runtime":
|
| 1350 |
-
"eval_samples_per_second": 20.
|
| 1351 |
-
"eval_steps_per_second": 0.
|
| 1352 |
-
"step":
|
| 1353 |
-
},
|
| 1354 |
-
{
|
| 1355 |
-
"epoch": 0.85,
|
| 1356 |
-
"grad_norm": 19.966317538177393,
|
| 1357 |
-
"learning_rate": 3.433663324986208e-08,
|
| 1358 |
-
"logits/chosen": 1.5778945684432983,
|
| 1359 |
-
"logits/rejected": 2.4630589485168457,
|
| 1360 |
-
"logps/chosen": -719.8331298828125,
|
| 1361 |
-
"logps/rejected": -1128.375244140625,
|
| 1362 |
-
"loss": 0.1004,
|
| 1363 |
-
"rewards/accuracies": 0.956250011920929,
|
| 1364 |
-
"rewards/chosen": -4.297072410583496,
|
| 1365 |
-
"rewards/margins": 6.129397392272949,
|
| 1366 |
-
"rewards/rejected": -10.426468849182129,
|
| 1367 |
-
"step": 810
|
| 1368 |
},
|
| 1369 |
{
|
| 1370 |
"epoch": 0.86,
|
| 1371 |
-
"grad_norm":
|
| 1372 |
-
"learning_rate":
|
| 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":
|
| 1383 |
-
},
|
| 1384 |
-
{
|
| 1385 |
-
"epoch": 0.87,
|
| 1386 |
-
"grad_norm": 16.905869837081713,
|
| 1387 |
-
"learning_rate": 2.567240179368185e-08,
|
| 1388 |
-
"logits/chosen": 1.6341140270233154,
|
| 1389 |
-
"logits/rejected": 3.0696451663970947,
|
| 1390 |
-
"logps/chosen": -696.9953002929688,
|
| 1391 |
-
"logps/rejected": -1112.6153564453125,
|
| 1392 |
-
"loss": 0.0849,
|
| 1393 |
-
"rewards/accuracies": 0.949999988079071,
|
| 1394 |
-
"rewards/chosen": -3.9459125995635986,
|
| 1395 |
-
"rewards/margins": 6.415045261383057,
|
| 1396 |
-
"rewards/rejected": -10.360957145690918,
|
| 1397 |
-
"step": 830
|
| 1398 |
},
|
| 1399 |
{
|
| 1400 |
"epoch": 0.88,
|
| 1401 |
-
"grad_norm":
|
| 1402 |
-
"learning_rate": 2.
|
| 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":
|
| 1413 |
-
},
|
| 1414 |
-
{
|
| 1415 |
-
"epoch": 0.89,
|
| 1416 |
-
"grad_norm": 32.49864502294839,
|
| 1417 |
-
"learning_rate": 1.820784220652766e-08,
|
| 1418 |
-
"logits/chosen": 1.6341012716293335,
|
| 1419 |
-
"logits/rejected": 2.78373384475708,
|
| 1420 |
-
"logps/chosen": -746.4293212890625,
|
| 1421 |
-
"logps/rejected": -1182.134765625,
|
| 1422 |
-
"loss": 0.0973,
|
| 1423 |
-
"rewards/accuracies": 0.956250011920929,
|
| 1424 |
-
"rewards/chosen": -4.279716968536377,
|
| 1425 |
-
"rewards/margins": 6.648747444152832,
|
| 1426 |
-
"rewards/rejected": -10.928464889526367,
|
| 1427 |
-
"step": 850
|
| 1428 |
},
|
| 1429 |
{
|
| 1430 |
"epoch": 0.9,
|
| 1431 |
-
"grad_norm":
|
| 1432 |
-
"learning_rate": 1.
|
| 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":
|
| 1441 |
-
"rewards/rejected": -
|
| 1442 |
-
"step":
|
| 1443 |
-
},
|
| 1444 |
-
{
|
| 1445 |
-
"epoch": 0.91,
|
| 1446 |
-
"grad_norm": 26.508225180148386,
|
| 1447 |
-
"learning_rate": 1.1982873884064465e-08,
|
| 1448 |
-
"logits/chosen": 1.877558946609497,
|
| 1449 |
-
"logits/rejected": 3.2897047996520996,
|
| 1450 |
-
"logps/chosen": -687.8853149414062,
|
| 1451 |
-
"logps/rejected": -1140.5772705078125,
|
| 1452 |
-
"loss": 0.0903,
|
| 1453 |
-
"rewards/accuracies": 0.9375,
|
| 1454 |
-
"rewards/chosen": -4.041706085205078,
|
| 1455 |
-
"rewards/margins": 6.566611289978027,
|
| 1456 |
-
"rewards/rejected": -10.608316421508789,
|
| 1457 |
-
"step": 870
|
| 1458 |
},
|
| 1459 |
{
|
| 1460 |
"epoch": 0.92,
|
| 1461 |
-
"grad_norm":
|
| 1462 |
-
"learning_rate": 9.
|
| 1463 |
-
"logits/chosen":
|
| 1464 |
-
"logits/rejected":
|
| 1465 |
-
"logps/chosen": -
|
| 1466 |
-
"logps/rejected": -
|
| 1467 |
-
"loss": 0.
|
| 1468 |
-
"rewards/accuracies": 0.
|
| 1469 |
-
"rewards/chosen": -
|
| 1470 |
-
"rewards/margins":
|
| 1471 |
-
"rewards/rejected": -
|
| 1472 |
-
"step":
|
| 1473 |
-
},
|
| 1474 |
-
{
|
| 1475 |
-
"epoch": 0.93,
|
| 1476 |
-
"grad_norm": 22.50666317342384,
|
| 1477 |
-
"learning_rate": 7.030787065396865e-09,
|
| 1478 |
-
"logits/chosen": 1.6333587169647217,
|
| 1479 |
-
"logits/rejected": 3.205723285675049,
|
| 1480 |
-
"logps/chosen": -725.650390625,
|
| 1481 |
-
"logps/rejected": -1133.517333984375,
|
| 1482 |
-
"loss": 0.0988,
|
| 1483 |
-
"rewards/accuracies": 0.9624999761581421,
|
| 1484 |
-
"rewards/chosen": -4.209347724914551,
|
| 1485 |
-
"rewards/margins": 6.3392205238342285,
|
| 1486 |
-
"rewards/rejected": -10.548568725585938,
|
| 1487 |
-
"step": 890
|
| 1488 |
-
},
|
| 1489 |
-
{
|
| 1490 |
-
"epoch": 0.94,
|
| 1491 |
-
"grad_norm": 26.84118963589271,
|
| 1492 |
-
"learning_rate": 5.04062020432286e-09,
|
| 1493 |
-
"logits/chosen": 1.553541898727417,
|
| 1494 |
-
"logits/rejected": 2.5439181327819824,
|
| 1495 |
-
"logps/chosen": -702.75341796875,
|
| 1496 |
-
"logps/rejected": -1179.390380859375,
|
| 1497 |
-
"loss": 0.0798,
|
| 1498 |
-
"rewards/accuracies": 0.96875,
|
| 1499 |
-
"rewards/chosen": -4.0308685302734375,
|
| 1500 |
-
"rewards/margins": 6.955786228179932,
|
| 1501 |
-
"rewards/rejected": -10.986656188964844,
|
| 1502 |
-
"step": 900
|
| 1503 |
},
|
| 1504 |
{
|
| 1505 |
"epoch": 0.94,
|
| 1506 |
-
"
|
| 1507 |
-
"
|
| 1508 |
-
"
|
| 1509 |
-
"
|
| 1510 |
-
"
|
| 1511 |
-
"
|
| 1512 |
-
"
|
| 1513 |
-
"
|
| 1514 |
-
"
|
| 1515 |
-
"
|
| 1516 |
-
"
|
| 1517 |
-
"
|
| 1518 |
-
"step": 900
|
| 1519 |
-
},
|
| 1520 |
-
{
|
| 1521 |
-
"epoch": 0.95,
|
| 1522 |
-
"grad_norm": 32.5930799681677,
|
| 1523 |
-
"learning_rate": 3.3780648016376866e-09,
|
| 1524 |
-
"logits/chosen": 1.1988734006881714,
|
| 1525 |
-
"logits/rejected": 2.8325839042663574,
|
| 1526 |
-
"logps/chosen": -706.9013671875,
|
| 1527 |
-
"logps/rejected": -1164.0465087890625,
|
| 1528 |
-
"loss": 0.0878,
|
| 1529 |
-
"rewards/accuracies": 0.96875,
|
| 1530 |
-
"rewards/chosen": -4.118927001953125,
|
| 1531 |
-
"rewards/margins": 6.752754211425781,
|
| 1532 |
-
"rewards/rejected": -10.871681213378906,
|
| 1533 |
-
"step": 910
|
| 1534 |
},
|
| 1535 |
{
|
| 1536 |
"epoch": 0.96,
|
| 1537 |
-
"grad_norm":
|
| 1538 |
-
"learning_rate": 2.
|
| 1539 |
-
"logits/chosen":
|
| 1540 |
-
"logits/rejected":
|
| 1541 |
-
"logps/chosen": -
|
| 1542 |
-
"logps/rejected": -
|
| 1543 |
-
"loss": 0.
|
| 1544 |
-
"rewards/accuracies": 0.
|
| 1545 |
-
"rewards/chosen": -
|
| 1546 |
-
"rewards/margins":
|
| 1547 |
-
"rewards/rejected": -
|
| 1548 |
-
"step":
|
| 1549 |
-
},
|
| 1550 |
-
{
|
| 1551 |
-
"epoch": 0.97,
|
| 1552 |
-
"grad_norm": 38.475668210703205,
|
| 1553 |
-
"learning_rate": 1.0442413283435758e-09,
|
| 1554 |
-
"logits/chosen": 1.560880422592163,
|
| 1555 |
-
"logits/rejected": 2.9233765602111816,
|
| 1556 |
-
"logps/chosen": -711.7680053710938,
|
| 1557 |
-
"logps/rejected": -1192.40234375,
|
| 1558 |
-
"loss": 0.0893,
|
| 1559 |
-
"rewards/accuracies": 0.949999988079071,
|
| 1560 |
-
"rewards/chosen": -4.384615421295166,
|
| 1561 |
-
"rewards/margins": 6.797455787658691,
|
| 1562 |
-
"rewards/rejected": -11.1820707321167,
|
| 1563 |
-
"step": 930
|
| 1564 |
},
|
| 1565 |
{
|
| 1566 |
"epoch": 0.98,
|
| 1567 |
-
"grad_norm":
|
| 1568 |
-
"learning_rate":
|
| 1569 |
-
"logits/chosen":
|
| 1570 |
-
"logits/rejected":
|
| 1571 |
-
"logps/chosen": -
|
| 1572 |
-
"logps/rejected": -
|
| 1573 |
-
"loss": 0.
|
| 1574 |
-
"rewards/accuracies": 0.
|
| 1575 |
-
"rewards/chosen": -
|
| 1576 |
-
"rewards/margins":
|
| 1577 |
-
"rewards/rejected": -
|
| 1578 |
-
"step":
|
| 1579 |
-
},
|
| 1580 |
-
{
|
| 1581 |
-
"epoch": 0.99,
|
| 1582 |
-
"grad_norm": 29.436206242522907,
|
| 1583 |
-
"learning_rate": 4.17975992204056e-11,
|
| 1584 |
-
"logits/chosen": 1.753095030784607,
|
| 1585 |
-
"logits/rejected": 2.988640308380127,
|
| 1586 |
-
"logps/chosen": -742.6573486328125,
|
| 1587 |
-
"logps/rejected": -1160.2423095703125,
|
| 1588 |
-
"loss": 0.1017,
|
| 1589 |
-
"rewards/accuracies": 0.925000011920929,
|
| 1590 |
-
"rewards/chosen": -4.522819519042969,
|
| 1591 |
-
"rewards/margins": 6.319222927093506,
|
| 1592 |
-
"rewards/rejected": -10.842041969299316,
|
| 1593 |
-
"step": 950
|
| 1594 |
},
|
| 1595 |
{
|
| 1596 |
"epoch": 1.0,
|
| 1597 |
-
"step":
|
| 1598 |
"total_flos": 0.0,
|
| 1599 |
-
"train_loss": 0.
|
| 1600 |
-
"train_runtime":
|
| 1601 |
-
"train_samples_per_second":
|
| 1602 |
-
"train_steps_per_second": 0.
|
| 1603 |
}
|
| 1604 |
],
|
| 1605 |
"logging_steps": 10,
|
| 1606 |
-
"max_steps":
|
| 1607 |
"num_input_tokens_seen": 0,
|
| 1608 |
"num_train_epochs": 1,
|
| 1609 |
"save_steps": 100,
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
"eval_steps": 100,
|
| 6 |
+
"global_step": 478,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.0,
|
| 13 |
+
"grad_norm": 27.516530702491266,
|
| 14 |
+
"learning_rate": 1.0416666666666666e-08,
|
| 15 |
+
"logits/chosen": -1.6311802864074707,
|
| 16 |
+
"logits/rejected": -1.6300675868988037,
|
| 17 |
+
"logps/chosen": -327.34149169921875,
|
| 18 |
+
"logps/rejected": -429.85833740234375,
|
| 19 |
"loss": 0.6931,
|
| 20 |
"rewards/accuracies": 0.0,
|
| 21 |
"rewards/chosen": 0.0,
|
|
|
|
| 23 |
"rewards/rejected": 0.0,
|
| 24 |
"step": 1
|
| 25 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
{
|
| 27 |
"epoch": 0.02,
|
| 28 |
+
"grad_norm": 25.454136109809017,
|
| 29 |
"learning_rate": 1.0416666666666667e-07,
|
| 30 |
+
"logits/chosen": -1.5537606477737427,
|
| 31 |
+
"logits/rejected": -1.506614089012146,
|
| 32 |
+
"logps/chosen": -312.7512512207031,
|
| 33 |
+
"logps/rejected": -303.0074462890625,
|
| 34 |
+
"loss": 0.6929,
|
| 35 |
+
"rewards/accuracies": 0.4513888955116272,
|
| 36 |
+
"rewards/chosen": 7.373202970484272e-05,
|
| 37 |
+
"rewards/margins": -5.077762034488842e-05,
|
| 38 |
+
"rewards/rejected": 0.000124509577290155,
|
| 39 |
+
"step": 10
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.04,
|
| 43 |
+
"grad_norm": 23.434204335413668,
|
| 44 |
"learning_rate": 2.0833333333333333e-07,
|
| 45 |
+
"logits/chosen": -1.6461513042449951,
|
| 46 |
+
"logits/rejected": -1.5507893562316895,
|
| 47 |
+
"logps/chosen": -329.1868896484375,
|
| 48 |
+
"logps/rejected": -351.44891357421875,
|
| 49 |
+
"loss": 0.6915,
|
| 50 |
+
"rewards/accuracies": 0.5625,
|
| 51 |
+
"rewards/chosen": 0.01134030893445015,
|
| 52 |
+
"rewards/margins": 0.00160890631377697,
|
| 53 |
+
"rewards/rejected": 0.009731403551995754,
|
| 54 |
+
"step": 20
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 0.06,
|
| 58 |
+
"grad_norm": 23.393322169409338,
|
| 59 |
"learning_rate": 3.1249999999999997e-07,
|
| 60 |
+
"logits/chosen": -1.6500400304794312,
|
| 61 |
+
"logits/rejected": -1.5703976154327393,
|
| 62 |
+
"logps/chosen": -346.0126953125,
|
| 63 |
+
"logps/rejected": -312.0833435058594,
|
| 64 |
+
"loss": 0.6837,
|
| 65 |
+
"rewards/accuracies": 0.637499988079071,
|
| 66 |
+
"rewards/chosen": 0.07147429138422012,
|
| 67 |
+
"rewards/margins": 0.022869160398840904,
|
| 68 |
+
"rewards/rejected": 0.04860512539744377,
|
| 69 |
+
"step": 30
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.08,
|
| 73 |
+
"grad_norm": 20.10353412052848,
|
| 74 |
"learning_rate": 4.1666666666666667e-07,
|
| 75 |
+
"logits/chosen": -1.5655517578125,
|
| 76 |
+
"logits/rejected": -1.5073188543319702,
|
| 77 |
+
"logps/chosen": -298.72418212890625,
|
| 78 |
+
"logps/rejected": -287.02362060546875,
|
| 79 |
+
"loss": 0.67,
|
| 80 |
+
"rewards/accuracies": 0.612500011920929,
|
| 81 |
+
"rewards/chosen": 0.16840842366218567,
|
| 82 |
+
"rewards/margins": 0.05939362570643425,
|
| 83 |
+
"rewards/rejected": 0.10901479423046112,
|
| 84 |
+
"step": 40
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 85 |
},
|
| 86 |
{
|
| 87 |
"epoch": 0.1,
|
| 88 |
+
"grad_norm": 18.756434278176222,
|
| 89 |
+
"learning_rate": 4.999733114418725e-07,
|
| 90 |
+
"logits/chosen": -1.57842218875885,
|
| 91 |
+
"logits/rejected": -1.529872179031372,
|
| 92 |
+
"logps/chosen": -299.90679931640625,
|
| 93 |
+
"logps/rejected": -323.1398010253906,
|
| 94 |
+
"loss": 0.6556,
|
| 95 |
+
"rewards/accuracies": 0.675000011920929,
|
| 96 |
+
"rewards/chosen": 0.2638908326625824,
|
| 97 |
+
"rewards/margins": 0.05764131620526314,
|
| 98 |
+
"rewards/rejected": 0.20624952018260956,
|
| 99 |
+
"step": 50
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.13,
|
| 103 |
+
"grad_norm": 22.017328959957382,
|
| 104 |
+
"learning_rate": 4.990398100856366e-07,
|
| 105 |
+
"logits/chosen": -1.6750396490097046,
|
| 106 |
+
"logits/rejected": -1.6164966821670532,
|
| 107 |
+
"logps/chosen": -288.38311767578125,
|
| 108 |
+
"logps/rejected": -338.00555419921875,
|
| 109 |
+
"loss": 0.6408,
|
| 110 |
+
"rewards/accuracies": 0.625,
|
| 111 |
+
"rewards/chosen": 0.3388586640357971,
|
| 112 |
+
"rewards/margins": 0.11625377833843231,
|
| 113 |
+
"rewards/rejected": 0.222604900598526,
|
| 114 |
+
"step": 60
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 115 |
},
|
| 116 |
{
|
| 117 |
"epoch": 0.15,
|
| 118 |
+
"grad_norm": 25.053484688153212,
|
| 119 |
+
"learning_rate": 4.967775735898179e-07,
|
| 120 |
+
"logits/chosen": -1.492963433265686,
|
| 121 |
+
"logits/rejected": -1.4768620729446411,
|
| 122 |
+
"logps/chosen": -287.0970764160156,
|
| 123 |
+
"logps/rejected": -302.3998718261719,
|
| 124 |
+
"loss": 0.6124,
|
| 125 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 126 |
+
"rewards/chosen": 0.2931686043739319,
|
| 127 |
+
"rewards/margins": 0.22670602798461914,
|
| 128 |
+
"rewards/rejected": 0.06646257638931274,
|
| 129 |
+
"step": 70
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.17,
|
| 133 |
+
"grad_norm": 35.26897936392246,
|
| 134 |
+
"learning_rate": 4.931986719649298e-07,
|
| 135 |
+
"logits/chosen": -1.5673052072525024,
|
| 136 |
+
"logits/rejected": -1.4728682041168213,
|
| 137 |
+
"logps/chosen": -367.7672119140625,
|
| 138 |
+
"logps/rejected": -335.300048828125,
|
| 139 |
+
"loss": 0.5894,
|
| 140 |
+
"rewards/accuracies": 0.65625,
|
| 141 |
+
"rewards/chosen": 0.1240416020154953,
|
| 142 |
+
"rewards/margins": 0.28898105025291443,
|
| 143 |
+
"rewards/rejected": -0.16493944823741913,
|
| 144 |
+
"step": 80
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 145 |
},
|
| 146 |
{
|
| 147 |
"epoch": 0.19,
|
| 148 |
+
"grad_norm": 40.64595674335777,
|
| 149 |
+
"learning_rate": 4.883222001996351e-07,
|
| 150 |
+
"logits/chosen": -1.1538174152374268,
|
| 151 |
+
"logits/rejected": -1.0762898921966553,
|
| 152 |
+
"logps/chosen": -330.572265625,
|
| 153 |
+
"logps/rejected": -377.3353576660156,
|
| 154 |
+
"loss": 0.5545,
|
| 155 |
+
"rewards/accuracies": 0.737500011920929,
|
| 156 |
+
"rewards/chosen": -0.18641509115695953,
|
| 157 |
+
"rewards/margins": 0.5493143200874329,
|
| 158 |
+
"rewards/rejected": -0.7357293963432312,
|
| 159 |
+
"step": 90
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 0.21,
|
| 163 |
+
"grad_norm": 48.97381689745216,
|
| 164 |
+
"learning_rate": 4.821741763807186e-07,
|
| 165 |
+
"logits/chosen": -0.7731859683990479,
|
| 166 |
+
"logits/rejected": -0.7151339054107666,
|
| 167 |
+
"logps/chosen": -357.76898193359375,
|
| 168 |
+
"logps/rejected": -418.14154052734375,
|
| 169 |
+
"loss": 0.539,
|
| 170 |
+
"rewards/accuracies": 0.706250011920929,
|
| 171 |
+
"rewards/chosen": -0.4816914498806,
|
| 172 |
+
"rewards/margins": 0.6462317705154419,
|
| 173 |
+
"rewards/rejected": -1.1279233694076538,
|
| 174 |
+
"step": 100
|
| 175 |
},
|
| 176 |
{
|
| 177 |
"epoch": 0.21,
|
| 178 |
+
"eval_logits/chosen": -0.8811883330345154,
|
| 179 |
+
"eval_logits/rejected": -0.7240222096443176,
|
| 180 |
+
"eval_logps/chosen": -371.65472412109375,
|
| 181 |
+
"eval_logps/rejected": -449.4052734375,
|
| 182 |
+
"eval_loss": 0.5451700091362,
|
| 183 |
+
"eval_rewards/accuracies": 0.7421875,
|
| 184 |
+
"eval_rewards/chosen": -0.5987821817398071,
|
| 185 |
+
"eval_rewards/margins": 0.7200188040733337,
|
| 186 |
+
"eval_rewards/rejected": -1.3188011646270752,
|
| 187 |
+
"eval_runtime": 97.5928,
|
| 188 |
+
"eval_samples_per_second": 20.493,
|
| 189 |
+
"eval_steps_per_second": 0.328,
|
| 190 |
+
"step": 100
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 191 |
},
|
| 192 |
{
|
| 193 |
"epoch": 0.23,
|
| 194 |
+
"grad_norm": 40.67779253000725,
|
| 195 |
+
"learning_rate": 4.747874028753375e-07,
|
| 196 |
+
"logits/chosen": -0.9343360066413879,
|
| 197 |
+
"logits/rejected": -0.7007087469100952,
|
| 198 |
+
"logps/chosen": -409.1253967285156,
|
| 199 |
+
"logps/rejected": -429.0912170410156,
|
| 200 |
+
"loss": 0.571,
|
| 201 |
+
"rewards/accuracies": 0.731249988079071,
|
| 202 |
+
"rewards/chosen": -0.4740072786808014,
|
| 203 |
+
"rewards/margins": 0.6539417505264282,
|
| 204 |
+
"rewards/rejected": -1.1279489994049072,
|
| 205 |
+
"step": 110
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 206 |
},
|
| 207 |
{
|
| 208 |
"epoch": 0.25,
|
| 209 |
+
"grad_norm": 52.47159513458736,
|
| 210 |
+
"learning_rate": 4.662012913161997e-07,
|
| 211 |
+
"logits/chosen": -0.5735512971878052,
|
| 212 |
+
"logits/rejected": -0.4024372696876526,
|
| 213 |
+
"logps/chosen": -359.32373046875,
|
| 214 |
+
"logps/rejected": -393.42974853515625,
|
| 215 |
+
"loss": 0.5452,
|
| 216 |
+
"rewards/accuracies": 0.71875,
|
| 217 |
+
"rewards/chosen": -0.2522381842136383,
|
| 218 |
+
"rewards/margins": 0.5496727228164673,
|
| 219 |
+
"rewards/rejected": -0.8019108772277832,
|
| 220 |
+
"step": 120
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 0.27,
|
| 224 |
+
"grad_norm": 41.93944182107726,
|
| 225 |
+
"learning_rate": 4.5646165232345103e-07,
|
| 226 |
+
"logits/chosen": -0.06039993837475777,
|
| 227 |
+
"logits/rejected": 0.15637175738811493,
|
| 228 |
+
"logps/chosen": -362.6398010253906,
|
| 229 |
+
"logps/rejected": -421.77099609375,
|
| 230 |
+
"loss": 0.5241,
|
| 231 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 232 |
+
"rewards/chosen": -0.3607412576675415,
|
| 233 |
+
"rewards/margins": 0.6238666772842407,
|
| 234 |
+
"rewards/rejected": -0.9846078753471375,
|
| 235 |
+
"step": 130
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 236 |
},
|
| 237 |
{
|
| 238 |
"epoch": 0.29,
|
| 239 |
+
"grad_norm": 58.39454249885024,
|
| 240 |
+
"learning_rate": 4.456204510851956e-07,
|
| 241 |
+
"logits/chosen": 0.14134523272514343,
|
| 242 |
+
"logits/rejected": 0.4324416220188141,
|
| 243 |
+
"logps/chosen": -426.36358642578125,
|
| 244 |
+
"logps/rejected": -497.6366271972656,
|
| 245 |
+
"loss": 0.5256,
|
| 246 |
+
"rewards/accuracies": 0.768750011920929,
|
| 247 |
+
"rewards/chosen": -0.5936006903648376,
|
| 248 |
+
"rewards/margins": 0.8088639378547668,
|
| 249 |
+
"rewards/rejected": -1.4024646282196045,
|
| 250 |
+
"step": 140
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 251 |
},
|
| 252 |
{
|
| 253 |
"epoch": 0.31,
|
| 254 |
+
"grad_norm": 47.848586407040784,
|
| 255 |
+
"learning_rate": 4.337355301007335e-07,
|
| 256 |
+
"logits/chosen": 0.6952211260795593,
|
| 257 |
+
"logits/rejected": 0.8886941075325012,
|
| 258 |
+
"logps/chosen": -412.7098083496094,
|
| 259 |
+
"logps/rejected": -467.67486572265625,
|
| 260 |
+
"loss": 0.5424,
|
| 261 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 262 |
+
"rewards/chosen": -0.8109712600708008,
|
| 263 |
+
"rewards/margins": 0.5735460519790649,
|
| 264 |
+
"rewards/rejected": -1.3845174312591553,
|
| 265 |
+
"step": 150
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 266 |
},
|
| 267 |
{
|
| 268 |
"epoch": 0.33,
|
| 269 |
+
"grad_norm": 41.0369792560692,
|
| 270 |
+
"learning_rate": 4.2087030056579986e-07,
|
| 271 |
+
"logits/chosen": 0.18771931529045105,
|
| 272 |
+
"logits/rejected": 0.7211147546768188,
|
| 273 |
+
"logps/chosen": -363.5400695800781,
|
| 274 |
+
"logps/rejected": -417.8056640625,
|
| 275 |
+
"loss": 0.5424,
|
| 276 |
+
"rewards/accuracies": 0.800000011920929,
|
| 277 |
+
"rewards/chosen": -0.3788861632347107,
|
| 278 |
+
"rewards/margins": 0.7999863028526306,
|
| 279 |
+
"rewards/rejected": -1.1788724660873413,
|
| 280 |
+
"step": 160
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 281 |
},
|
| 282 |
{
|
| 283 |
"epoch": 0.36,
|
| 284 |
+
"grad_norm": 40.70764257859768,
|
| 285 |
+
"learning_rate": 4.070934040463998e-07,
|
| 286 |
+
"logits/chosen": 0.3842553198337555,
|
| 287 |
+
"logits/rejected": 0.7558060884475708,
|
| 288 |
+
"logps/chosen": -360.5339660644531,
|
| 289 |
+
"logps/rejected": -410.0292053222656,
|
| 290 |
+
"loss": 0.5243,
|
| 291 |
+
"rewards/accuracies": 0.71875,
|
| 292 |
+
"rewards/chosen": -0.6288506388664246,
|
| 293 |
+
"rewards/margins": 0.626385509967804,
|
| 294 |
+
"rewards/rejected": -1.2552361488342285,
|
| 295 |
+
"step": 170
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 296 |
},
|
| 297 |
{
|
| 298 |
"epoch": 0.38,
|
| 299 |
+
"grad_norm": 42.009965736488304,
|
| 300 |
+
"learning_rate": 3.9247834624635404e-07,
|
| 301 |
+
"logits/chosen": 0.8420892953872681,
|
| 302 |
+
"logits/rejected": 1.2455089092254639,
|
| 303 |
+
"logps/chosen": -370.99090576171875,
|
| 304 |
+
"logps/rejected": -420.73358154296875,
|
| 305 |
+
"loss": 0.4907,
|
| 306 |
+
"rewards/accuracies": 0.731249988079071,
|
| 307 |
+
"rewards/chosen": -0.6862328052520752,
|
| 308 |
+
"rewards/margins": 0.7635400891304016,
|
| 309 |
+
"rewards/rejected": -1.449772834777832,
|
| 310 |
+
"step": 180
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 311 |
},
|
| 312 |
{
|
| 313 |
"epoch": 0.4,
|
| 314 |
+
"grad_norm": 40.59427137175583,
|
| 315 |
+
"learning_rate": 3.7710310482256523e-07,
|
| 316 |
+
"logits/chosen": 0.35120826959609985,
|
| 317 |
+
"logits/rejected": 0.5841476321220398,
|
| 318 |
+
"logps/chosen": -369.50762939453125,
|
| 319 |
+
"logps/rejected": -431.60430908203125,
|
| 320 |
+
"loss": 0.5135,
|
| 321 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 322 |
+
"rewards/chosen": -0.45533379912376404,
|
| 323 |
+
"rewards/margins": 0.7292474508285522,
|
| 324 |
+
"rewards/rejected": -1.1845813989639282,
|
| 325 |
+
"step": 190
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 0.42,
|
| 329 |
+
"grad_norm": 74.46574358755885,
|
| 330 |
+
"learning_rate": 3.610497133404795e-07,
|
| 331 |
+
"logits/chosen": 0.5723182559013367,
|
| 332 |
+
"logits/rejected": 0.8850580453872681,
|
| 333 |
+
"logps/chosen": -384.32122802734375,
|
| 334 |
+
"logps/rejected": -460.3968200683594,
|
| 335 |
+
"loss": 0.5294,
|
| 336 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 337 |
+
"rewards/chosen": -0.7563287019729614,
|
| 338 |
+
"rewards/margins": 0.8355164527893066,
|
| 339 |
+
"rewards/rejected": -1.591845154762268,
|
| 340 |
+
"step": 200
|
| 341 |
},
|
| 342 |
{
|
| 343 |
"epoch": 0.42,
|
| 344 |
+
"eval_logits/chosen": 0.45017001032829285,
|
| 345 |
+
"eval_logits/rejected": 0.8080278038978577,
|
| 346 |
+
"eval_logps/chosen": -378.31610107421875,
|
| 347 |
+
"eval_logps/rejected": -471.284912109375,
|
| 348 |
+
"eval_loss": 0.49995097517967224,
|
| 349 |
+
"eval_rewards/accuracies": 0.76171875,
|
| 350 |
+
"eval_rewards/chosen": -0.6653960347175598,
|
| 351 |
+
"eval_rewards/margins": 0.8722014427185059,
|
| 352 |
+
"eval_rewards/rejected": -1.537597417831421,
|
| 353 |
+
"eval_runtime": 97.5595,
|
| 354 |
+
"eval_samples_per_second": 20.5,
|
| 355 |
+
"eval_steps_per_second": 0.328,
|
| 356 |
+
"step": 200
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 357 |
},
|
| 358 |
{
|
| 359 |
"epoch": 0.44,
|
| 360 |
+
"grad_norm": 46.031984990497406,
|
| 361 |
+
"learning_rate": 3.4440382358952115e-07,
|
| 362 |
+
"logits/chosen": 0.48276287317276,
|
| 363 |
+
"logits/rejected": 0.8357312083244324,
|
| 364 |
+
"logps/chosen": -404.5564880371094,
|
| 365 |
+
"logps/rejected": -446.3399963378906,
|
| 366 |
+
"loss": 0.5221,
|
| 367 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 368 |
+
"rewards/chosen": -0.5977016687393188,
|
| 369 |
+
"rewards/margins": 0.8030574917793274,
|
| 370 |
+
"rewards/rejected": -1.400759220123291,
|
| 371 |
+
"step": 210
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 372 |
},
|
| 373 |
{
|
| 374 |
"epoch": 0.46,
|
| 375 |
+
"grad_norm": 48.160697637910886,
|
| 376 |
+
"learning_rate": 3.272542485937368e-07,
|
| 377 |
+
"logits/chosen": 0.5655020475387573,
|
| 378 |
+
"logits/rejected": 0.9708032608032227,
|
| 379 |
+
"logps/chosen": -374.70733642578125,
|
| 380 |
+
"logps/rejected": -441.2498474121094,
|
| 381 |
+
"loss": 0.5028,
|
| 382 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 383 |
+
"rewards/chosen": -0.6240397691726685,
|
| 384 |
+
"rewards/margins": 0.8947107195854187,
|
| 385 |
+
"rewards/rejected": -1.5187504291534424,
|
| 386 |
+
"step": 220
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 387 |
},
|
| 388 |
{
|
| 389 |
"epoch": 0.48,
|
| 390 |
+
"grad_norm": 48.6209922138632,
|
| 391 |
+
"learning_rate": 3.096924887558854e-07,
|
| 392 |
+
"logits/chosen": 0.6631933450698853,
|
| 393 |
+
"logits/rejected": 1.059995412826538,
|
| 394 |
+
"logps/chosen": -387.43206787109375,
|
| 395 |
+
"logps/rejected": -495.94976806640625,
|
| 396 |
+
"loss": 0.5111,
|
| 397 |
+
"rewards/accuracies": 0.762499988079071,
|
| 398 |
+
"rewards/chosen": -0.9049599766731262,
|
| 399 |
+
"rewards/margins": 1.0336183309555054,
|
| 400 |
+
"rewards/rejected": -1.9385783672332764,
|
| 401 |
+
"step": 230
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 402 |
},
|
| 403 |
{
|
| 404 |
"epoch": 0.5,
|
| 405 |
+
"grad_norm": 50.906502494422604,
|
| 406 |
+
"learning_rate": 2.9181224366319943e-07,
|
| 407 |
+
"logits/chosen": 0.07225779443979263,
|
| 408 |
+
"logits/rejected": 0.43359050154685974,
|
| 409 |
+
"logps/chosen": -382.9250183105469,
|
| 410 |
+
"logps/rejected": -458.04705810546875,
|
| 411 |
+
"loss": 0.4847,
|
| 412 |
+
"rewards/accuracies": 0.78125,
|
| 413 |
+
"rewards/chosen": -0.6258467435836792,
|
| 414 |
+
"rewards/margins": 0.9445594549179077,
|
| 415 |
+
"rewards/rejected": -1.570406198501587,
|
| 416 |
+
"step": 240
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 0.52,
|
| 420 |
+
"grad_norm": 51.78937629264524,
|
| 421 |
+
"learning_rate": 2.7370891215954565e-07,
|
| 422 |
+
"logits/chosen": 0.30628976225852966,
|
| 423 |
+
"logits/rejected": 0.6650404930114746,
|
| 424 |
+
"logps/chosen": -413.821044921875,
|
| 425 |
+
"logps/rejected": -486.84588623046875,
|
| 426 |
+
"loss": 0.5071,
|
| 427 |
+
"rewards/accuracies": 0.8125,
|
| 428 |
+
"rewards/chosen": -0.6004015207290649,
|
| 429 |
+
"rewards/margins": 1.1020289659500122,
|
| 430 |
+
"rewards/rejected": -1.7024303674697876,
|
| 431 |
+
"step": 250
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 432 |
},
|
| 433 |
{
|
| 434 |
"epoch": 0.54,
|
| 435 |
+
"grad_norm": 45.99084165109373,
|
| 436 |
+
"learning_rate": 2.55479083351317e-07,
|
| 437 |
+
"logits/chosen": 0.2199174165725708,
|
| 438 |
+
"logits/rejected": 0.5914181470870972,
|
| 439 |
+
"logps/chosen": -416.1441955566406,
|
| 440 |
+
"logps/rejected": -481.27020263671875,
|
| 441 |
+
"loss": 0.4928,
|
| 442 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 443 |
+
"rewards/chosen": -0.6798352003097534,
|
| 444 |
+
"rewards/margins": 1.0327410697937012,
|
| 445 |
+
"rewards/rejected": -1.7125762701034546,
|
| 446 |
+
"step": 260
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 447 |
},
|
| 448 |
{
|
| 449 |
+
"epoch": 0.56,
|
| 450 |
+
"grad_norm": 44.6265768332988,
|
| 451 |
+
"learning_rate": 2.3722002126275822e-07,
|
| 452 |
+
"logits/chosen": 0.2704251706600189,
|
| 453 |
+
"logits/rejected": 0.6909293532371521,
|
| 454 |
+
"logps/chosen": -405.48089599609375,
|
| 455 |
+
"logps/rejected": -471.8568420410156,
|
| 456 |
+
"loss": 0.5062,
|
| 457 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 458 |
+
"rewards/chosen": -0.7410668134689331,
|
| 459 |
+
"rewards/margins": 0.8480156064033508,
|
| 460 |
+
"rewards/rejected": -1.5890824794769287,
|
| 461 |
+
"step": 270
|
| 462 |
},
|
| 463 |
{
|
| 464 |
"epoch": 0.59,
|
| 465 |
+
"grad_norm": 45.745689932530695,
|
| 466 |
+
"learning_rate": 2.19029145890313e-07,
|
| 467 |
+
"logits/chosen": 0.5122145414352417,
|
| 468 |
+
"logits/rejected": 0.9128357172012329,
|
| 469 |
+
"logps/chosen": -386.9629821777344,
|
| 470 |
+
"logps/rejected": -474.33697509765625,
|
| 471 |
+
"loss": 0.5119,
|
| 472 |
+
"rewards/accuracies": 0.78125,
|
| 473 |
+
"rewards/chosen": -0.7393491864204407,
|
| 474 |
+
"rewards/margins": 1.0442432165145874,
|
| 475 |
+
"rewards/rejected": -1.7835925817489624,
|
| 476 |
+
"step": 280
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 477 |
},
|
| 478 |
{
|
| 479 |
"epoch": 0.61,
|
| 480 |
+
"grad_norm": 39.82940708039207,
|
| 481 |
+
"learning_rate": 2.0100351342479216e-07,
|
| 482 |
+
"logits/chosen": 0.5138941407203674,
|
| 483 |
+
"logits/rejected": 0.7629462480545044,
|
| 484 |
+
"logps/chosen": -382.9958801269531,
|
| 485 |
+
"logps/rejected": -461.4434509277344,
|
| 486 |
+
"loss": 0.4913,
|
| 487 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 488 |
+
"rewards/chosen": -0.8513978719711304,
|
| 489 |
+
"rewards/margins": 0.8561092615127563,
|
| 490 |
+
"rewards/rejected": -1.7075071334838867,
|
| 491 |
+
"step": 290
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 492 |
},
|
| 493 |
{
|
| 494 |
"epoch": 0.63,
|
| 495 |
+
"grad_norm": 46.49918219263532,
|
| 496 |
+
"learning_rate": 1.8323929841460178e-07,
|
| 497 |
+
"logits/chosen": 0.3830031752586365,
|
| 498 |
+
"logits/rejected": 0.9853779077529907,
|
| 499 |
+
"logps/chosen": -448.693359375,
|
| 500 |
+
"logps/rejected": -515.8469848632812,
|
| 501 |
+
"loss": 0.4704,
|
| 502 |
+
"rewards/accuracies": 0.762499988079071,
|
| 503 |
+
"rewards/chosen": -0.9706672430038452,
|
| 504 |
+
"rewards/margins": 1.0561556816101074,
|
| 505 |
+
"rewards/rejected": -2.026822805404663,
|
| 506 |
+
"step": 300
|
| 507 |
},
|
| 508 |
{
|
| 509 |
"epoch": 0.63,
|
| 510 |
+
"eval_logits/chosen": 0.3976738452911377,
|
| 511 |
+
"eval_logits/rejected": 0.8997932076454163,
|
| 512 |
+
"eval_logps/chosen": -396.1895446777344,
|
| 513 |
+
"eval_logps/rejected": -516.5240478515625,
|
| 514 |
+
"eval_loss": 0.48780035972595215,
|
| 515 |
+
"eval_rewards/accuracies": 0.75390625,
|
| 516 |
+
"eval_rewards/chosen": -0.8441304564476013,
|
| 517 |
+
"eval_rewards/margins": 1.1458582878112793,
|
| 518 |
+
"eval_rewards/rejected": -1.9899888038635254,
|
| 519 |
+
"eval_runtime": 97.6501,
|
| 520 |
+
"eval_samples_per_second": 20.481,
|
| 521 |
+
"eval_steps_per_second": 0.328,
|
| 522 |
+
"step": 300
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 523 |
},
|
| 524 |
{
|
| 525 |
"epoch": 0.65,
|
| 526 |
+
"grad_norm": 52.076181120314466,
|
| 527 |
+
"learning_rate": 1.6583128063291573e-07,
|
| 528 |
+
"logits/chosen": 0.43108898401260376,
|
| 529 |
+
"logits/rejected": 0.7893391251564026,
|
| 530 |
+
"logps/chosen": -438.8291931152344,
|
| 531 |
+
"logps/rejected": -502.6856994628906,
|
| 532 |
+
"loss": 0.485,
|
| 533 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 534 |
+
"rewards/chosen": -0.9125974774360657,
|
| 535 |
+
"rewards/margins": 0.9991793632507324,
|
| 536 |
+
"rewards/rejected": -1.9117769002914429,
|
| 537 |
+
"step": 310
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 538 |
},
|
| 539 |
{
|
| 540 |
"epoch": 0.67,
|
| 541 |
+
"grad_norm": 53.82916961985562,
|
| 542 |
+
"learning_rate": 1.488723393865766e-07,
|
| 543 |
+
"logits/chosen": 0.5213574171066284,
|
| 544 |
+
"logits/rejected": 1.0497839450836182,
|
| 545 |
+
"logps/chosen": -437.28936767578125,
|
| 546 |
+
"logps/rejected": -490.869873046875,
|
| 547 |
+
"loss": 0.4654,
|
| 548 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 549 |
+
"rewards/chosen": -0.8886297941207886,
|
| 550 |
+
"rewards/margins": 1.0424896478652954,
|
| 551 |
+
"rewards/rejected": -1.9311193227767944,
|
| 552 |
+
"step": 320
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 553 |
},
|
| 554 |
{
|
| 555 |
"epoch": 0.69,
|
| 556 |
+
"grad_norm": 52.0979391510305,
|
| 557 |
+
"learning_rate": 1.3245295796480788e-07,
|
| 558 |
+
"logits/chosen": 0.27822378277778625,
|
| 559 |
+
"logits/rejected": 0.6966886520385742,
|
| 560 |
+
"logps/chosen": -394.94219970703125,
|
| 561 |
+
"logps/rejected": -488.16571044921875,
|
| 562 |
+
"loss": 0.4721,
|
| 563 |
+
"rewards/accuracies": 0.71875,
|
| 564 |
+
"rewards/chosen": -0.8261836767196655,
|
| 565 |
+
"rewards/margins": 0.8932548761367798,
|
| 566 |
+
"rewards/rejected": -1.7194385528564453,
|
| 567 |
+
"step": 330
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 568 |
},
|
| 569 |
{
|
| 570 |
"epoch": 0.71,
|
| 571 |
+
"grad_norm": 53.006317641552194,
|
| 572 |
+
"learning_rate": 1.1666074087171627e-07,
|
| 573 |
+
"logits/chosen": 0.24821150302886963,
|
| 574 |
+
"logits/rejected": 0.725691556930542,
|
| 575 |
+
"logps/chosen": -424.24761962890625,
|
| 576 |
+
"logps/rejected": -520.8505249023438,
|
| 577 |
+
"loss": 0.485,
|
| 578 |
+
"rewards/accuracies": 0.75,
|
| 579 |
+
"rewards/chosen": -0.772143542766571,
|
| 580 |
+
"rewards/margins": 1.1264902353286743,
|
| 581 |
+
"rewards/rejected": -1.8986337184906006,
|
| 582 |
+
"step": 340
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 583 |
},
|
| 584 |
{
|
| 585 |
"epoch": 0.73,
|
| 586 |
+
"grad_norm": 50.07894063945848,
|
| 587 |
+
"learning_rate": 1.0157994641835734e-07,
|
| 588 |
+
"logits/chosen": 0.5090032815933228,
|
| 589 |
+
"logits/rejected": 1.0217373371124268,
|
| 590 |
+
"logps/chosen": -399.2255859375,
|
| 591 |
+
"logps/rejected": -497.14910888671875,
|
| 592 |
+
"loss": 0.4596,
|
| 593 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 594 |
+
"rewards/chosen": -0.9144848585128784,
|
| 595 |
+
"rewards/margins": 1.1523492336273193,
|
| 596 |
+
"rewards/rejected": -2.0668342113494873,
|
| 597 |
+
"step": 350
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 598 |
},
|
| 599 |
{
|
| 600 |
"epoch": 0.75,
|
| 601 |
+
"grad_norm": 46.12071728187378,
|
| 602 |
+
"learning_rate": 8.729103716819111e-08,
|
| 603 |
+
"logits/chosen": 0.29714930057525635,
|
| 604 |
+
"logits/rejected": 0.8836923837661743,
|
| 605 |
+
"logps/chosen": -449.24822998046875,
|
| 606 |
+
"logps/rejected": -515.9562377929688,
|
| 607 |
+
"loss": 0.507,
|
| 608 |
+
"rewards/accuracies": 0.762499988079071,
|
| 609 |
+
"rewards/chosen": -0.9280809164047241,
|
| 610 |
+
"rewards/margins": 1.0685356855392456,
|
| 611 |
+
"rewards/rejected": -1.9966167211532593,
|
| 612 |
+
"step": 360
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 613 |
},
|
| 614 |
{
|
| 615 |
"epoch": 0.77,
|
| 616 |
+
"grad_norm": 44.466114694620565,
|
| 617 |
+
"learning_rate": 7.387025063449081e-08,
|
| 618 |
+
"logits/chosen": 0.3406144082546234,
|
| 619 |
+
"logits/rejected": 0.8408617973327637,
|
| 620 |
+
"logps/chosen": -423.6982421875,
|
| 621 |
+
"logps/rejected": -472.16302490234375,
|
| 622 |
+
"loss": 0.4937,
|
| 623 |
+
"rewards/accuracies": 0.737500011920929,
|
| 624 |
+
"rewards/chosen": -0.973863959312439,
|
| 625 |
+
"rewards/margins": 0.8518773317337036,
|
| 626 |
+
"rewards/rejected": -1.825741171836853,
|
| 627 |
+
"step": 370
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 628 |
},
|
| 629 |
{
|
| 630 |
+
"epoch": 0.79,
|
| 631 |
+
"grad_norm": 40.2546545345207,
|
| 632 |
+
"learning_rate": 6.138919252022435e-08,
|
| 633 |
+
"logits/chosen": 0.5687306523323059,
|
| 634 |
+
"logits/rejected": 0.8843731880187988,
|
| 635 |
+
"logps/chosen": -388.6229248046875,
|
| 636 |
+
"logps/rejected": -532.6536865234375,
|
| 637 |
+
"loss": 0.4777,
|
| 638 |
+
"rewards/accuracies": 0.793749988079071,
|
| 639 |
+
"rewards/chosen": -1.018700361251831,
|
| 640 |
+
"rewards/margins": 1.1921350955963135,
|
| 641 |
+
"rewards/rejected": -2.2108356952667236,
|
| 642 |
+
"step": 380
|
| 643 |
},
|
| 644 |
{
|
| 645 |
"epoch": 0.82,
|
| 646 |
+
"grad_norm": 50.26373841675365,
|
| 647 |
+
"learning_rate": 4.991445467064689e-08,
|
| 648 |
+
"logits/chosen": 0.3668118119239807,
|
| 649 |
+
"logits/rejected": 0.6758331060409546,
|
| 650 |
+
"logps/chosen": -445.70294189453125,
|
| 651 |
+
"logps/rejected": -524.3595581054688,
|
| 652 |
+
"loss": 0.4756,
|
| 653 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 654 |
+
"rewards/chosen": -0.906339168548584,
|
| 655 |
+
"rewards/margins": 0.9413349032402039,
|
| 656 |
+
"rewards/rejected": -1.8476741313934326,
|
| 657 |
+
"step": 390
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 0.84,
|
| 661 |
+
"grad_norm": 45.03150381268163,
|
| 662 |
+
"learning_rate": 3.9507259776993954e-08,
|
| 663 |
+
"logits/chosen": 0.47453969717025757,
|
| 664 |
+
"logits/rejected": 0.8979718089103699,
|
| 665 |
+
"logps/chosen": -417.6993713378906,
|
| 666 |
+
"logps/rejected": -514.3675537109375,
|
| 667 |
+
"loss": 0.4856,
|
| 668 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 669 |
+
"rewards/chosen": -0.9480649828910828,
|
| 670 |
+
"rewards/margins": 1.0335150957107544,
|
| 671 |
+
"rewards/rejected": -1.981580138206482,
|
| 672 |
+
"step": 400
|
| 673 |
},
|
| 674 |
{
|
| 675 |
"epoch": 0.84,
|
| 676 |
+
"eval_logits/chosen": 0.22729139029979706,
|
| 677 |
+
"eval_logits/rejected": 0.7285040616989136,
|
| 678 |
+
"eval_logps/chosen": -388.4799499511719,
|
| 679 |
+
"eval_logps/rejected": -508.7244873046875,
|
| 680 |
+
"eval_loss": 0.4800879657268524,
|
| 681 |
+
"eval_rewards/accuracies": 0.76171875,
|
| 682 |
+
"eval_rewards/chosen": -0.7670345306396484,
|
| 683 |
+
"eval_rewards/margins": 1.1449588537216187,
|
| 684 |
+
"eval_rewards/rejected": -1.911993384361267,
|
| 685 |
+
"eval_runtime": 97.4275,
|
| 686 |
+
"eval_samples_per_second": 20.528,
|
| 687 |
+
"eval_steps_per_second": 0.328,
|
| 688 |
+
"step": 400
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 689 |
},
|
| 690 |
{
|
| 691 |
"epoch": 0.86,
|
| 692 |
+
"grad_norm": 43.3283841726019,
|
| 693 |
+
"learning_rate": 3.022313472693447e-08,
|
| 694 |
+
"logits/chosen": 0.2033652514219284,
|
| 695 |
+
"logits/rejected": 0.8105279803276062,
|
| 696 |
+
"logps/chosen": -432.13153076171875,
|
| 697 |
+
"logps/rejected": -513.7069702148438,
|
| 698 |
+
"loss": 0.4936,
|
| 699 |
+
"rewards/accuracies": 0.731249988079071,
|
| 700 |
+
"rewards/chosen": -0.8029797673225403,
|
| 701 |
+
"rewards/margins": 1.0931544303894043,
|
| 702 |
+
"rewards/rejected": -1.8961341381072998,
|
| 703 |
+
"step": 410
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 704 |
},
|
| 705 |
{
|
| 706 |
"epoch": 0.88,
|
| 707 |
+
"grad_norm": 43.3329326100997,
|
| 708 |
+
"learning_rate": 2.2111614344599684e-08,
|
| 709 |
+
"logits/chosen": 0.23255594074726105,
|
| 710 |
+
"logits/rejected": 0.6442626118659973,
|
| 711 |
+
"logps/chosen": -432.7608337402344,
|
| 712 |
+
"logps/rejected": -520.2620239257812,
|
| 713 |
+
"loss": 0.4641,
|
| 714 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 715 |
+
"rewards/chosen": -0.8288372159004211,
|
| 716 |
+
"rewards/margins": 1.0977327823638916,
|
| 717 |
+
"rewards/rejected": -1.926569938659668,
|
| 718 |
+
"step": 420
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 719 |
},
|
| 720 |
{
|
| 721 |
"epoch": 0.9,
|
| 722 |
+
"grad_norm": 49.02509873372083,
|
| 723 |
+
"learning_rate": 1.521597710086439e-08,
|
| 724 |
+
"logits/chosen": 0.39820823073387146,
|
| 725 |
+
"logits/rejected": 0.7836225628852844,
|
| 726 |
+
"logps/chosen": -431.13739013671875,
|
| 727 |
+
"logps/rejected": -512.7986450195312,
|
| 728 |
+
"loss": 0.471,
|
| 729 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 730 |
+
"rewards/chosen": -0.9906813502311707,
|
| 731 |
+
"rewards/margins": 1.0531766414642334,
|
| 732 |
+
"rewards/rejected": -2.043858051300049,
|
| 733 |
+
"step": 430
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 734 |
},
|
| 735 |
{
|
| 736 |
"epoch": 0.92,
|
| 737 |
+
"grad_norm": 46.801343844231866,
|
| 738 |
+
"learning_rate": 9.57301420397924e-09,
|
| 739 |
+
"logits/chosen": 0.2471725195646286,
|
| 740 |
+
"logits/rejected": 0.7467008829116821,
|
| 741 |
+
"logps/chosen": -417.58258056640625,
|
| 742 |
+
"logps/rejected": -499.89215087890625,
|
| 743 |
+
"loss": 0.4942,
|
| 744 |
+
"rewards/accuracies": 0.75,
|
| 745 |
+
"rewards/chosen": -0.8245165944099426,
|
| 746 |
+
"rewards/margins": 0.9129160642623901,
|
| 747 |
+
"rewards/rejected": -1.737432837486267,
|
| 748 |
+
"step": 440
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 749 |
},
|
| 750 |
{
|
| 751 |
"epoch": 0.94,
|
| 752 |
+
"grad_norm": 46.37037120468359,
|
| 753 |
+
"learning_rate": 5.212833302556258e-09,
|
| 754 |
+
"logits/chosen": 0.4000117778778076,
|
| 755 |
+
"logits/rejected": 0.6783515810966492,
|
| 756 |
+
"logps/chosen": -436.942138671875,
|
| 757 |
+
"logps/rejected": -551.7266845703125,
|
| 758 |
+
"loss": 0.4786,
|
| 759 |
+
"rewards/accuracies": 0.71875,
|
| 760 |
+
"rewards/chosen": -0.9379842877388,
|
| 761 |
+
"rewards/margins": 0.9772690534591675,
|
| 762 |
+
"rewards/rejected": -1.9152534008026123,
|
| 763 |
+
"step": 450
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 764 |
},
|
| 765 |
{
|
| 766 |
"epoch": 0.96,
|
| 767 |
+
"grad_norm": 46.5176855624682,
|
| 768 |
+
"learning_rate": 2.158697848236607e-09,
|
| 769 |
+
"logits/chosen": 0.3254472017288208,
|
| 770 |
+
"logits/rejected": 0.7334408164024353,
|
| 771 |
+
"logps/chosen": -410.4520568847656,
|
| 772 |
+
"logps/rejected": -479.54351806640625,
|
| 773 |
+
"loss": 0.4804,
|
| 774 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 775 |
+
"rewards/chosen": -0.8168198466300964,
|
| 776 |
+
"rewards/margins": 0.9972711801528931,
|
| 777 |
+
"rewards/rejected": -1.8140909671783447,
|
| 778 |
+
"step": 460
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 779 |
},
|
| 780 |
{
|
| 781 |
"epoch": 0.98,
|
| 782 |
+
"grad_norm": 42.92566705814915,
|
| 783 |
+
"learning_rate": 4.269029751107489e-10,
|
| 784 |
+
"logits/chosen": 0.2615105211734772,
|
| 785 |
+
"logits/rejected": 0.6600261926651001,
|
| 786 |
+
"logps/chosen": -408.7879638671875,
|
| 787 |
+
"logps/rejected": -514.4189453125,
|
| 788 |
+
"loss": 0.4762,
|
| 789 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 790 |
+
"rewards/chosen": -0.8471711874008179,
|
| 791 |
+
"rewards/margins": 1.000536322593689,
|
| 792 |
+
"rewards/rejected": -1.8477073907852173,
|
| 793 |
+
"step": 470
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 794 |
},
|
| 795 |
{
|
| 796 |
"epoch": 1.0,
|
| 797 |
+
"step": 478,
|
| 798 |
"total_flos": 0.0,
|
| 799 |
+
"train_loss": 0.5266154430900158,
|
| 800 |
+
"train_runtime": 7640.9818,
|
| 801 |
+
"train_samples_per_second": 8.001,
|
| 802 |
+
"train_steps_per_second": 0.063
|
| 803 |
}
|
| 804 |
],
|
| 805 |
"logging_steps": 10,
|
| 806 |
+
"max_steps": 478,
|
| 807 |
"num_input_tokens_seen": 0,
|
| 808 |
"num_train_epochs": 1,
|
| 809 |
"save_steps": 100,
|