Model save
Browse files- README.md +26 -24
- adapter_model.safetensors +1 -1
- all_results.json +4 -19
- train_results.json +4 -4
- trainer_state.json +1620 -714
README.md
CHANGED
|
@@ -2,13 +2,10 @@
|
|
| 2 |
license: apache-2.0
|
| 3 |
library_name: peft
|
| 4 |
tags:
|
| 5 |
-
- alignment-handbook
|
| 6 |
- trl
|
| 7 |
- dpo
|
| 8 |
- generated_from_trainer
|
| 9 |
base_model: mistralai/Mistral-7B-v0.1
|
| 10 |
-
datasets:
|
| 11 |
-
- HuggingFaceH4/ultrafeedback_binarized
|
| 12 |
model-index:
|
| 13 |
- name: zephyr-7b
|
| 14 |
results: []
|
|
@@ -19,19 +16,19 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 19 |
|
| 20 |
# zephyr-7b
|
| 21 |
|
| 22 |
-
This model is a fine-tuned version of [
|
| 23 |
It achieves the following results on the evaluation set:
|
| 24 |
-
- Loss: 0.
|
| 25 |
-
- Rewards/chosen: -
|
| 26 |
-
- Rewards/rejected: -
|
| 27 |
-
- Rewards/accuracies: 0.
|
| 28 |
-
- Rewards/margins:
|
| 29 |
-
- Logps/rejected: -
|
| 30 |
-
- Logps/chosen: -
|
| 31 |
-
- Logits/rejected:
|
| 32 |
-
- Logits/chosen:
|
| 33 |
-
- Use Label:
|
| 34 |
-
- Pred Label:
|
| 35 |
|
| 36 |
## Model description
|
| 37 |
|
|
@@ -55,10 +52,10 @@ The following hyperparameters were used during training:
|
|
| 55 |
- eval_batch_size: 8
|
| 56 |
- seed: 42
|
| 57 |
- distributed_type: multi-GPU
|
| 58 |
-
- num_devices:
|
| 59 |
- gradient_accumulation_steps: 4
|
| 60 |
-
- total_train_batch_size:
|
| 61 |
-
- total_eval_batch_size:
|
| 62 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 63 |
- lr_scheduler_type: cosine
|
| 64 |
- lr_scheduler_warmup_ratio: 0.1
|
|
@@ -66,12 +63,17 @@ The following hyperparameters were used during training:
|
|
| 66 |
|
| 67 |
### Training results
|
| 68 |
|
| 69 |
-
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen | Use Label
|
| 70 |
-
|
| 71 |
-
| 0.
|
| 72 |
-
| 0.
|
| 73 |
-
| 0.
|
| 74 |
-
| 0.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 75 |
|
| 76 |
|
| 77 |
### Framework versions
|
|
|
|
| 2 |
license: apache-2.0
|
| 3 |
library_name: peft
|
| 4 |
tags:
|
|
|
|
| 5 |
- trl
|
| 6 |
- dpo
|
| 7 |
- generated_from_trainer
|
| 8 |
base_model: mistralai/Mistral-7B-v0.1
|
|
|
|
|
|
|
| 9 |
model-index:
|
| 10 |
- name: zephyr-7b
|
| 11 |
results: []
|
|
|
|
| 16 |
|
| 17 |
# zephyr-7b
|
| 18 |
|
| 19 |
+
This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on the None dataset.
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.3625
|
| 22 |
+
- Rewards/chosen: -150.6127
|
| 23 |
+
- Rewards/rejected: -146.3050
|
| 24 |
+
- Rewards/accuracies: 0.2421
|
| 25 |
+
- Rewards/margins: -4.3077
|
| 26 |
+
- Logps/rejected: -14705.8975
|
| 27 |
+
- Logps/chosen: -15130.1680
|
| 28 |
+
- Logits/rejected: 13.5362
|
| 29 |
+
- Logits/chosen: 13.4716
|
| 30 |
+
- Use Label: 11165.9844
|
| 31 |
+
- Pred Label: 7522.0161
|
| 32 |
|
| 33 |
## Model description
|
| 34 |
|
|
|
|
| 52 |
- eval_batch_size: 8
|
| 53 |
- seed: 42
|
| 54 |
- distributed_type: multi-GPU
|
| 55 |
+
- num_devices: 4
|
| 56 |
- gradient_accumulation_steps: 4
|
| 57 |
+
- total_train_batch_size: 64
|
| 58 |
+
- total_eval_batch_size: 32
|
| 59 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 60 |
- lr_scheduler_type: cosine
|
| 61 |
- lr_scheduler_warmup_ratio: 0.1
|
|
|
|
| 63 |
|
| 64 |
### Training results
|
| 65 |
|
| 66 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen | Use Label | Pred Label |
|
| 67 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|:----------:|:----------:|
|
| 68 |
+
| 0.6637 | 0.1 | 100 | 0.6642 | -0.0947 | -0.1635 | 0.3254 | 0.0687 | -91.7446 | -78.3734 | -2.0927 | -2.1253 | 1838.9207 | 17.0794 |
|
| 69 |
+
| 0.3902 | 0.21 | 200 | 0.3930 | -14.4219 | -14.0352 | 0.2560 | -0.3866 | -1478.9202 | -1511.0870 | 2.8471 | 2.7727 | 3444.6985 | 515.3016 |
|
| 70 |
+
| 0.3845 | 0.31 | 300 | 0.3786 | -23.0869 | -24.5685 | 0.2520 | 1.4817 | -2532.2498 | -2377.5872 | 5.4283 | 5.3070 | 4579.4922 | 1484.5079 |
|
| 71 |
+
| 0.3477 | 0.42 | 400 | 0.3622 | -111.3259 | -109.5294 | 0.25 | -1.7965 | -11028.3408 | -11201.4893 | 11.6816 | 11.5716 | 5682.4922 | 2485.5081 |
|
| 72 |
+
| 0.3468 | 0.52 | 500 | 0.3613 | -144.7782 | -140.7408 | 0.2421 | -4.0373 | -14149.4824 | -14546.7158 | 13.8885 | 13.8347 | 6784.2383 | 3487.7620 |
|
| 73 |
+
| 0.33 | 0.63 | 600 | 0.3605 | -143.0167 | -138.8336 | 0.2401 | -4.1831 | -13958.7627 | -14370.5693 | 12.5943 | 12.5399 | 7857.4287 | 4518.5713 |
|
| 74 |
+
| 0.3665 | 0.73 | 700 | 0.3614 | -150.1877 | -145.8865 | 0.2421 | -4.3011 | -14664.0518 | -15087.6680 | 13.4024 | 13.3367 | 8936.4287 | 5543.5713 |
|
| 75 |
+
| 0.3731 | 0.84 | 800 | 0.3623 | -150.4385 | -146.1303 | 0.2401 | -4.3082 | -14688.4258 | -15112.7539 | 13.5339 | 13.4696 | 10050.3330 | 6533.6665 |
|
| 76 |
+
| 0.3696 | 0.94 | 900 | 0.3625 | -150.6127 | -146.3050 | 0.2421 | -4.3077 | -14705.8975 | -15130.1680 | 13.5362 | 13.4716 | 11165.9844 | 7522.0161 |
|
| 77 |
|
| 78 |
|
| 79 |
### Framework versions
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 671150064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6e343887c74d6a04e3a2cd7161aa5920ab97068b7cff8e4eed7a3dee31d6d650
|
| 3 |
size 671150064
|
all_results.json
CHANGED
|
@@ -1,23 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
-
"
|
| 4 |
-
"
|
| 5 |
-
"eval_logps/chosen": -380.465087890625,
|
| 6 |
-
"eval_logps/rejected": -579.1896362304688,
|
| 7 |
-
"eval_loss": 0.4399436116218567,
|
| 8 |
-
"eval_pred_label": 2212.28125,
|
| 9 |
-
"eval_rewards/accuracies": 0.33984375,
|
| 10 |
-
"eval_rewards/chosen": -3.1655280590057373,
|
| 11 |
-
"eval_rewards/margins": 1.8544387817382812,
|
| 12 |
-
"eval_rewards/rejected": -5.019967079162598,
|
| 13 |
-
"eval_runtime": 125.2743,
|
| 14 |
-
"eval_samples": 2000,
|
| 15 |
-
"eval_samples_per_second": 15.965,
|
| 16 |
-
"eval_steps_per_second": 0.255,
|
| 17 |
-
"eval_use_label": 6575.71875,
|
| 18 |
-
"train_loss": 0.5378267840019562,
|
| 19 |
-
"train_runtime": 9600.9753,
|
| 20 |
"train_samples": 61135,
|
| 21 |
-
"train_samples_per_second":
|
| 22 |
-
"train_steps_per_second": 0.
|
| 23 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.4203558097959189,
|
| 4 |
+
"train_runtime": 19955.1733,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
"train_samples": 61135,
|
| 6 |
+
"train_samples_per_second": 3.064,
|
| 7 |
+
"train_steps_per_second": 0.048
|
| 8 |
}
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 61135,
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.4203558097959189,
|
| 4 |
+
"train_runtime": 19955.1733,
|
| 5 |
"train_samples": 61135,
|
| 6 |
+
"train_samples_per_second": 3.064,
|
| 7 |
+
"train_steps_per_second": 0.048
|
| 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": 0.
|
| 14 |
-
"learning_rate":
|
| 15 |
-
"logits/chosen": -2.
|
| 16 |
-
"logits/rejected": -2.
|
| 17 |
-
"logps/chosen": -
|
| 18 |
-
"logps/rejected": -
|
| 19 |
"loss": 0.6931,
|
| 20 |
"pred_label": 0.0,
|
| 21 |
"rewards/accuracies": 0.0,
|
|
@@ -26,888 +26,1794 @@
|
|
| 26 |
"use_label": 10.0
|
| 27 |
},
|
| 28 |
{
|
| 29 |
-
"epoch": 0.
|
| 30 |
-
"grad_norm": 0.
|
| 31 |
-
"learning_rate":
|
| 32 |
-
"logits/chosen": -2.
|
| 33 |
-
"logits/rejected": -2.
|
| 34 |
-
"logps/chosen": -
|
| 35 |
-
"logps/rejected": -
|
| 36 |
-
"loss": 0.
|
| 37 |
"pred_label": 0.0,
|
| 38 |
-
"rewards/accuracies": 0.
|
| 39 |
-
"rewards/chosen": 0.
|
| 40 |
-
"rewards/margins": 0.
|
| 41 |
-
"rewards/rejected": 0.
|
| 42 |
"step": 10,
|
| 43 |
"use_label": 90.0
|
| 44 |
},
|
| 45 |
{
|
| 46 |
-
"epoch": 0.
|
| 47 |
-
"grad_norm": 0.
|
| 48 |
-
"learning_rate":
|
| 49 |
-
"logits/chosen": -2.
|
| 50 |
-
"logits/rejected": -2.
|
| 51 |
-
"logps/chosen": -
|
| 52 |
-
"logps/rejected": -
|
| 53 |
-
"loss": 0.
|
| 54 |
"pred_label": 0.0,
|
| 55 |
-
"rewards/accuracies": 0.
|
| 56 |
-
"rewards/chosen": 0.
|
| 57 |
-
"rewards/margins": 0.
|
| 58 |
-
"rewards/rejected": 0.
|
| 59 |
"step": 20,
|
| 60 |
"use_label": 242.0
|
| 61 |
},
|
| 62 |
{
|
| 63 |
-
"epoch": 0.
|
| 64 |
-
"grad_norm": 0.
|
| 65 |
-
"learning_rate":
|
| 66 |
-
"logits/chosen": -2.
|
| 67 |
-
"logits/rejected": -2.
|
| 68 |
-
"logps/chosen": -
|
| 69 |
-
"logps/rejected": -
|
| 70 |
-
"loss": 0.
|
| 71 |
"pred_label": 0.0,
|
| 72 |
"rewards/accuracies": 0.2750000059604645,
|
| 73 |
-
"rewards/chosen": 0.
|
| 74 |
-
"rewards/margins": 0.
|
| 75 |
-
"rewards/rejected": 0.
|
| 76 |
"step": 30,
|
| 77 |
"use_label": 402.0
|
| 78 |
},
|
| 79 |
{
|
| 80 |
-
"epoch": 0.
|
| 81 |
-
"grad_norm": 0.
|
| 82 |
-
"learning_rate":
|
| 83 |
-
"logits/chosen": -2.
|
| 84 |
-
"logits/rejected": -2.
|
| 85 |
-
"logps/chosen": -
|
| 86 |
-
"logps/rejected": -
|
| 87 |
-
"loss": 0.
|
| 88 |
"pred_label": 0.0,
|
| 89 |
-
"rewards/accuracies": 0.
|
| 90 |
-
"rewards/chosen": 0.
|
| 91 |
-
"rewards/margins": 0.
|
| 92 |
-
"rewards/rejected": 0.
|
| 93 |
"step": 40,
|
| 94 |
"use_label": 562.0
|
| 95 |
},
|
| 96 |
{
|
| 97 |
-
"epoch": 0.
|
| 98 |
-
"grad_norm": 0.
|
| 99 |
-
"learning_rate":
|
| 100 |
-
"logits/chosen": -2.
|
| 101 |
-
"logits/rejected": -2.
|
| 102 |
-
"logps/chosen": -
|
| 103 |
-
"logps/rejected": -
|
| 104 |
-
"loss": 0.
|
| 105 |
"pred_label": 0.0,
|
| 106 |
-
"rewards/accuracies": 0.
|
| 107 |
-
"rewards/chosen": 0.
|
| 108 |
-
"rewards/margins": 0.
|
| 109 |
-
"rewards/rejected":
|
| 110 |
"step": 50,
|
| 111 |
"use_label": 722.0
|
| 112 |
},
|
| 113 |
{
|
| 114 |
-
"epoch": 0.
|
| 115 |
-
"grad_norm": 0.
|
| 116 |
-
"learning_rate":
|
| 117 |
-
"logits/chosen": -2.
|
| 118 |
-
"logits/rejected": -2.
|
| 119 |
-
"logps/chosen": -
|
| 120 |
-
"logps/rejected": -
|
| 121 |
-
"loss": 0.
|
| 122 |
"pred_label": 0.0,
|
| 123 |
-
"rewards/accuracies": 0.
|
| 124 |
-
"rewards/chosen":
|
| 125 |
-
"rewards/margins": 0.
|
| 126 |
-
"rewards/rejected":
|
| 127 |
"step": 60,
|
| 128 |
"use_label": 882.0
|
| 129 |
},
|
| 130 |
{
|
| 131 |
-
"epoch": 0.
|
| 132 |
-
"grad_norm":
|
| 133 |
-
"learning_rate":
|
| 134 |
-
"logits/chosen": -2.
|
| 135 |
-
"logits/rejected": -2.
|
| 136 |
-
"logps/chosen": -
|
| 137 |
-
"logps/rejected": -
|
| 138 |
-
"loss": 0.
|
| 139 |
"pred_label": 0.0,
|
| 140 |
-
"rewards/accuracies": 0.
|
| 141 |
-
"rewards/chosen":
|
| 142 |
-
"rewards/margins": 0.
|
| 143 |
-
"rewards/rejected":
|
| 144 |
"step": 70,
|
| 145 |
"use_label": 1042.0
|
| 146 |
},
|
| 147 |
{
|
| 148 |
-
"epoch": 0.
|
| 149 |
-
"grad_norm":
|
| 150 |
-
"learning_rate": 4.
|
| 151 |
-
"logits/chosen": -2.
|
| 152 |
-
"logits/rejected": -2.
|
| 153 |
-
"logps/chosen": -
|
| 154 |
-
"logps/rejected": -
|
| 155 |
-
"loss": 0.
|
| 156 |
-
"pred_label":
|
| 157 |
-
"rewards/accuracies": 0.
|
| 158 |
-
"rewards/chosen":
|
| 159 |
-
"rewards/margins": 0.
|
| 160 |
-
"rewards/rejected": -0.
|
| 161 |
"step": 80,
|
| 162 |
-
"use_label":
|
| 163 |
},
|
| 164 |
{
|
| 165 |
-
"epoch": 0.
|
| 166 |
-
"grad_norm":
|
| 167 |
-
"learning_rate": 4.
|
| 168 |
-
"logits/chosen": -2.
|
| 169 |
-
"logits/rejected": -2.
|
| 170 |
-
"logps/chosen": -
|
| 171 |
-
"logps/rejected": -
|
| 172 |
-
"loss": 0.
|
| 173 |
-
"pred_label":
|
| 174 |
-
"rewards/accuracies": 0.
|
| 175 |
-
"rewards/chosen": -0.
|
| 176 |
-
"rewards/margins": 0.
|
| 177 |
-
"rewards/rejected": -0.
|
| 178 |
"step": 90,
|
| 179 |
-
"use_label":
|
| 180 |
},
|
| 181 |
{
|
| 182 |
-
"epoch": 0.
|
| 183 |
-
"grad_norm": 1.
|
| 184 |
-
"learning_rate": 4.
|
| 185 |
-
"logits/chosen": -2.
|
| 186 |
-
"logits/rejected": -2.
|
| 187 |
-
"logps/chosen": -
|
| 188 |
-
"logps/rejected": -
|
| 189 |
-
"loss": 0.
|
| 190 |
-
"pred_label":
|
| 191 |
-
"rewards/accuracies": 0.
|
| 192 |
-
"rewards/chosen": -0.
|
| 193 |
-
"rewards/margins": 0.
|
| 194 |
-
"rewards/rejected": -0.
|
| 195 |
"step": 100,
|
| 196 |
-
"use_label":
|
| 197 |
},
|
| 198 |
{
|
| 199 |
-
"epoch": 0.
|
| 200 |
-
"eval_logits/chosen": -2.
|
| 201 |
-
"eval_logits/rejected": -2.
|
| 202 |
-
"eval_logps/chosen": -
|
| 203 |
-
"eval_logps/rejected": -
|
| 204 |
-
"eval_loss": 0.
|
| 205 |
-
"eval_pred_label":
|
| 206 |
-
"eval_rewards/accuracies": 0.
|
| 207 |
-
"eval_rewards/chosen": -0.
|
| 208 |
-
"eval_rewards/margins": 0.
|
| 209 |
-
"eval_rewards/rejected": -0.
|
| 210 |
-
"eval_runtime":
|
| 211 |
-
"eval_samples_per_second":
|
| 212 |
-
"eval_steps_per_second": 0.
|
| 213 |
-
"eval_use_label":
|
| 214 |
"step": 100
|
| 215 |
},
|
| 216 |
{
|
| 217 |
-
"epoch": 0.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 218 |
"grad_norm": 2.125,
|
| 219 |
-
"learning_rate": 4.
|
| 220 |
-
"logits/chosen": -2.
|
| 221 |
-
"logits/rejected": -2.
|
| 222 |
-
"logps/chosen": -
|
| 223 |
-
"logps/rejected": -
|
| 224 |
-
"loss": 0.
|
| 225 |
-
"pred_label":
|
| 226 |
"rewards/accuracies": 0.3499999940395355,
|
| 227 |
-
"rewards/chosen": -0.
|
| 228 |
-
"rewards/margins": 0.
|
| 229 |
-
"rewards/rejected": -0.
|
| 230 |
-
"step":
|
| 231 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 232 |
},
|
| 233 |
{
|
| 234 |
"epoch": 0.25,
|
| 235 |
-
"grad_norm":
|
| 236 |
-
"learning_rate": 4.
|
| 237 |
-
"logits/chosen":
|
| 238 |
-
"logits/rejected":
|
| 239 |
-
"logps/chosen": -
|
| 240 |
-
"logps/rejected": -
|
| 241 |
-
"loss": 0.
|
| 242 |
-
"pred_label":
|
| 243 |
-
"rewards/accuracies": 0.
|
| 244 |
-
"rewards/chosen": -
|
| 245 |
-
"rewards/margins": 0.
|
| 246 |
-
"rewards/rejected": -
|
| 247 |
-
"step":
|
| 248 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 0.27,
|
| 252 |
-
"grad_norm":
|
| 253 |
-
"learning_rate": 4.
|
| 254 |
-
"logits/chosen":
|
| 255 |
-
"logits/rejected":
|
| 256 |
-
"logps/chosen": -
|
| 257 |
-
"logps/rejected": -
|
| 258 |
-
"loss": 0.
|
| 259 |
-
"pred_label":
|
| 260 |
-
"rewards/accuracies": 0.
|
| 261 |
-
"rewards/chosen": -
|
| 262 |
-
"rewards/margins":
|
| 263 |
-
"rewards/rejected": -
|
| 264 |
-
"step":
|
| 265 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 266 |
},
|
| 267 |
{
|
| 268 |
"epoch": 0.29,
|
| 269 |
-
"grad_norm":
|
| 270 |
-
"learning_rate": 4.
|
| 271 |
-
"logits/chosen":
|
| 272 |
-
"logits/rejected":
|
| 273 |
-
"logps/chosen": -
|
| 274 |
-
"logps/rejected": -
|
| 275 |
-
"loss": 0.
|
| 276 |
-
"pred_label":
|
| 277 |
-
"rewards/accuracies": 0.
|
| 278 |
-
"rewards/chosen": -
|
| 279 |
-
"rewards/margins": 0.
|
| 280 |
-
"rewards/rejected": -
|
| 281 |
-
"step":
|
| 282 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 283 |
},
|
| 284 |
{
|
| 285 |
"epoch": 0.31,
|
| 286 |
-
"grad_norm":
|
| 287 |
-
"learning_rate": 4.
|
| 288 |
-
"logits/chosen":
|
| 289 |
-
"logits/rejected":
|
| 290 |
-
"logps/chosen": -
|
| 291 |
-
"logps/rejected": -
|
| 292 |
-
"loss": 0.
|
| 293 |
-
"pred_label":
|
| 294 |
-
"rewards/accuracies": 0.
|
| 295 |
-
"rewards/chosen": -
|
| 296 |
-
"rewards/margins":
|
| 297 |
-
"rewards/rejected": -
|
| 298 |
-
"step":
|
| 299 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 300 |
},
|
| 301 |
{
|
| 302 |
"epoch": 0.33,
|
| 303 |
-
"grad_norm":
|
| 304 |
-
"learning_rate": 4.
|
| 305 |
-
"logits/chosen":
|
| 306 |
-
"logits/rejected":
|
| 307 |
-
"logps/chosen": -
|
| 308 |
-
"logps/rejected": -
|
| 309 |
-
"loss": 0.
|
| 310 |
-
"pred_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 311 |
"rewards/accuracies": 0.26875001192092896,
|
| 312 |
-
"rewards/chosen": -
|
| 313 |
-
"rewards/margins":
|
| 314 |
-
"rewards/rejected": -
|
| 315 |
-
"step":
|
| 316 |
-
"use_label":
|
| 317 |
},
|
| 318 |
{
|
| 319 |
"epoch": 0.36,
|
| 320 |
-
"grad_norm":
|
| 321 |
-
"learning_rate": 4.
|
| 322 |
-
"logits/chosen":
|
| 323 |
-
"logits/rejected":
|
| 324 |
-
"logps/chosen": -
|
| 325 |
-
"logps/rejected": -
|
| 326 |
-
"loss": 0.
|
| 327 |
-
"pred_label":
|
| 328 |
-
"rewards/accuracies": 0.
|
| 329 |
-
"rewards/chosen": -
|
| 330 |
-
"rewards/margins":
|
| 331 |
-
"rewards/rejected": -
|
| 332 |
-
"step":
|
| 333 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 334 |
},
|
| 335 |
{
|
| 336 |
"epoch": 0.38,
|
| 337 |
-
"grad_norm":
|
| 338 |
-
"learning_rate": 3.
|
| 339 |
-
"logits/chosen":
|
| 340 |
-
"logits/rejected":
|
| 341 |
-
"logps/chosen": -
|
| 342 |
-
"logps/rejected": -
|
| 343 |
-
"loss": 0.
|
| 344 |
-
"pred_label":
|
| 345 |
-
"rewards/accuracies": 0.
|
| 346 |
-
"rewards/chosen": -
|
| 347 |
-
"rewards/margins": 0.
|
| 348 |
-
"rewards/rejected": -
|
| 349 |
-
"step":
|
| 350 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 351 |
},
|
| 352 |
{
|
| 353 |
"epoch": 0.4,
|
| 354 |
-
"grad_norm":
|
| 355 |
-
"learning_rate": 3.
|
| 356 |
-
"logits/chosen":
|
| 357 |
-
"logits/rejected":
|
| 358 |
-
"logps/chosen": -
|
| 359 |
-
"logps/rejected": -
|
| 360 |
-
"loss": 0.
|
| 361 |
-
"pred_label":
|
| 362 |
-
"rewards/accuracies": 0.
|
| 363 |
-
"rewards/chosen": -
|
| 364 |
-
"rewards/margins":
|
| 365 |
-
"rewards/rejected": -
|
| 366 |
-
"step":
|
| 367 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"epoch": 0.42,
|
| 371 |
-
"grad_norm":
|
| 372 |
-
"learning_rate": 3.
|
| 373 |
-
"logits/chosen":
|
| 374 |
-
"logits/rejected":
|
| 375 |
-
"logps/chosen": -
|
| 376 |
-
"logps/rejected": -
|
| 377 |
-
"loss": 0.
|
| 378 |
-
"pred_label":
|
| 379 |
-
"rewards/accuracies": 0.
|
| 380 |
-
"rewards/chosen": -
|
| 381 |
-
"rewards/margins":
|
| 382 |
-
"rewards/rejected": -
|
| 383 |
-
"step":
|
| 384 |
-
"use_label":
|
| 385 |
},
|
| 386 |
{
|
| 387 |
"epoch": 0.42,
|
| 388 |
-
"eval_logits/chosen":
|
| 389 |
-
"eval_logits/rejected":
|
| 390 |
-
"eval_logps/chosen": -
|
| 391 |
-
"eval_logps/rejected": -
|
| 392 |
-
"eval_loss": 0.
|
| 393 |
-
"eval_pred_label":
|
| 394 |
-
"eval_rewards/accuracies": 0.
|
| 395 |
-
"eval_rewards/chosen": -
|
| 396 |
-
"eval_rewards/margins":
|
| 397 |
-
"eval_rewards/rejected": -
|
| 398 |
-
"eval_runtime":
|
| 399 |
-
"eval_samples_per_second":
|
| 400 |
-
"eval_steps_per_second": 0.
|
| 401 |
-
"eval_use_label":
|
| 402 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 0.44,
|
| 406 |
-
"grad_norm":
|
| 407 |
-
"learning_rate": 3.
|
| 408 |
-
"logits/chosen":
|
| 409 |
-
"logits/rejected":
|
| 410 |
-
"logps/chosen": -
|
| 411 |
-
"logps/rejected": -
|
| 412 |
-
"loss": 0.
|
| 413 |
-
"pred_label":
|
| 414 |
-
"rewards/accuracies": 0.
|
| 415 |
-
"rewards/chosen": -
|
| 416 |
-
"rewards/margins":
|
| 417 |
-
"rewards/rejected": -
|
| 418 |
-
"step":
|
| 419 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 420 |
},
|
| 421 |
{
|
| 422 |
"epoch": 0.46,
|
| 423 |
-
"grad_norm":
|
| 424 |
-
"learning_rate": 3.
|
| 425 |
-
"logits/chosen":
|
| 426 |
-
"logits/rejected":
|
| 427 |
-
"logps/chosen": -
|
| 428 |
-
"logps/rejected": -
|
| 429 |
-
"loss": 0.
|
| 430 |
-
"pred_label":
|
| 431 |
-
"rewards/accuracies": 0.
|
| 432 |
-
"rewards/chosen": -
|
| 433 |
-
"rewards/margins":
|
| 434 |
-
"rewards/rejected": -
|
| 435 |
-
"step":
|
| 436 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 437 |
},
|
| 438 |
{
|
| 439 |
"epoch": 0.48,
|
| 440 |
-
"grad_norm":
|
| 441 |
-
"learning_rate": 3.
|
| 442 |
-
"logits/chosen":
|
| 443 |
-
"logits/rejected":
|
| 444 |
-
"logps/chosen": -
|
| 445 |
-
"logps/rejected": -
|
| 446 |
-
"loss": 0.
|
| 447 |
-
"pred_label":
|
| 448 |
-
"rewards/accuracies": 0.
|
| 449 |
-
"rewards/chosen": -
|
| 450 |
-
"rewards/margins":
|
| 451 |
-
"rewards/rejected": -
|
| 452 |
-
"step":
|
| 453 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 454 |
},
|
| 455 |
{
|
| 456 |
"epoch": 0.5,
|
| 457 |
-
"grad_norm":
|
| 458 |
-
"learning_rate": 2.
|
| 459 |
-
"logits/chosen":
|
| 460 |
-
"logits/rejected":
|
| 461 |
-
"logps/chosen": -
|
| 462 |
-
"logps/rejected": -
|
| 463 |
-
"loss": 0.
|
| 464 |
-
"pred_label":
|
| 465 |
-
"rewards/accuracies": 0.
|
| 466 |
-
"rewards/chosen": -
|
| 467 |
-
"rewards/margins": 0.
|
| 468 |
-
"rewards/rejected": -
|
| 469 |
-
"step":
|
| 470 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 471 |
},
|
| 472 |
{
|
| 473 |
"epoch": 0.52,
|
| 474 |
-
"grad_norm":
|
| 475 |
-
"learning_rate": 2.
|
| 476 |
-
"logits/chosen":
|
| 477 |
-
"logits/rejected":
|
| 478 |
-
"logps/chosen": -
|
| 479 |
-
"logps/rejected": -
|
| 480 |
-
"loss": 0.
|
| 481 |
-
"pred_label":
|
| 482 |
-
"rewards/accuracies": 0.
|
| 483 |
-
"rewards/chosen": -
|
| 484 |
-
"rewards/margins":
|
| 485 |
-
"rewards/rejected": -
|
| 486 |
-
"step":
|
| 487 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 488 |
},
|
| 489 |
{
|
| 490 |
"epoch": 0.54,
|
| 491 |
-
"grad_norm":
|
| 492 |
-
"learning_rate": 2.
|
| 493 |
-
"logits/chosen":
|
| 494 |
-
"logits/rejected":
|
| 495 |
-
"logps/chosen": -
|
| 496 |
-
"logps/rejected": -
|
| 497 |
-
"loss": 0.
|
| 498 |
-
"pred_label":
|
| 499 |
-
"rewards/accuracies": 0.
|
| 500 |
-
"rewards/chosen": -
|
| 501 |
-
"rewards/margins":
|
| 502 |
-
"rewards/rejected": -
|
| 503 |
-
"step":
|
| 504 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 505 |
},
|
| 506 |
{
|
| 507 |
"epoch": 0.57,
|
| 508 |
-
"grad_norm":
|
| 509 |
-
"learning_rate": 2.
|
| 510 |
-
"logits/chosen":
|
| 511 |
-
"logits/rejected":
|
| 512 |
-
"logps/chosen": -
|
| 513 |
-
"logps/rejected": -
|
| 514 |
-
"loss": 0.
|
| 515 |
-
"pred_label":
|
| 516 |
-
"rewards/accuracies": 0.
|
| 517 |
-
"rewards/chosen": -
|
| 518 |
-
"rewards/margins":
|
| 519 |
-
"rewards/rejected": -
|
| 520 |
-
"step":
|
| 521 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 0.59,
|
| 525 |
-
"grad_norm":
|
| 526 |
-
"learning_rate": 2.
|
| 527 |
-
"logits/chosen":
|
| 528 |
-
"logits/rejected":
|
| 529 |
-
"logps/chosen": -
|
| 530 |
-
"logps/rejected": -
|
| 531 |
-
"loss": 0.
|
| 532 |
-
"pred_label":
|
| 533 |
-
"rewards/accuracies": 0.
|
| 534 |
-
"rewards/chosen": -
|
| 535 |
-
"rewards/margins":
|
| 536 |
-
"rewards/rejected": -
|
| 537 |
-
"step":
|
| 538 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 539 |
},
|
| 540 |
{
|
| 541 |
"epoch": 0.61,
|
| 542 |
-
"grad_norm":
|
| 543 |
-
"learning_rate":
|
| 544 |
-
"logits/chosen":
|
| 545 |
-
"logits/rejected":
|
| 546 |
-
"logps/chosen": -
|
| 547 |
-
"logps/rejected": -
|
| 548 |
-
"loss": 0.
|
| 549 |
-
"pred_label":
|
| 550 |
-
"rewards/accuracies": 0.
|
| 551 |
-
"rewards/chosen": -
|
| 552 |
-
"rewards/margins":
|
| 553 |
-
"rewards/rejected": -
|
| 554 |
-
"step":
|
| 555 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 556 |
},
|
| 557 |
{
|
| 558 |
"epoch": 0.63,
|
| 559 |
-
"grad_norm":
|
| 560 |
-
"learning_rate": 1.
|
| 561 |
-
"logits/chosen":
|
| 562 |
-
"logits/rejected":
|
| 563 |
-
"logps/chosen": -
|
| 564 |
-
"logps/rejected": -
|
| 565 |
-
"loss": 0.
|
| 566 |
-
"pred_label":
|
| 567 |
-
"rewards/accuracies": 0.
|
| 568 |
-
"rewards/chosen": -
|
| 569 |
-
"rewards/margins":
|
| 570 |
-
"rewards/rejected": -
|
| 571 |
-
"step":
|
| 572 |
-
"use_label":
|
| 573 |
},
|
| 574 |
{
|
| 575 |
"epoch": 0.63,
|
| 576 |
-
"eval_logits/chosen":
|
| 577 |
-
"eval_logits/rejected":
|
| 578 |
-
"eval_logps/chosen": -
|
| 579 |
-
"eval_logps/rejected": -
|
| 580 |
-
"eval_loss": 0.
|
| 581 |
-
"eval_pred_label":
|
| 582 |
-
"eval_rewards/accuracies": 0.
|
| 583 |
-
"eval_rewards/chosen": -
|
| 584 |
-
"eval_rewards/margins":
|
| 585 |
-
"eval_rewards/rejected": -
|
| 586 |
-
"eval_runtime":
|
| 587 |
-
"eval_samples_per_second":
|
| 588 |
-
"eval_steps_per_second": 0.
|
| 589 |
-
"eval_use_label":
|
| 590 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 591 |
},
|
| 592 |
{
|
| 593 |
"epoch": 0.65,
|
| 594 |
-
"grad_norm":
|
| 595 |
-
"learning_rate": 1.
|
| 596 |
-
"logits/chosen":
|
| 597 |
-
"logits/rejected":
|
| 598 |
-
"logps/chosen": -
|
| 599 |
-
"logps/rejected": -
|
| 600 |
-
"loss": 0.
|
| 601 |
-
"pred_label":
|
| 602 |
-
"rewards/accuracies": 0.
|
| 603 |
-
"rewards/chosen": -
|
| 604 |
-
"rewards/margins":
|
| 605 |
-
"rewards/rejected": -
|
| 606 |
-
"step":
|
| 607 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 608 |
},
|
| 609 |
{
|
| 610 |
"epoch": 0.67,
|
| 611 |
-
"grad_norm":
|
| 612 |
-
"learning_rate": 1.
|
| 613 |
-
"logits/chosen":
|
| 614 |
-
"logits/rejected":
|
| 615 |
-
"logps/chosen": -
|
| 616 |
-
"logps/rejected": -
|
| 617 |
-
"loss": 0.
|
| 618 |
-
"pred_label":
|
| 619 |
-
"rewards/accuracies": 0.
|
| 620 |
-
"rewards/chosen": -
|
| 621 |
-
"rewards/margins":
|
| 622 |
-
"rewards/rejected": -
|
| 623 |
-
"step":
|
| 624 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 625 |
},
|
| 626 |
{
|
| 627 |
"epoch": 0.69,
|
| 628 |
-
"grad_norm":
|
| 629 |
-
"learning_rate": 1.
|
| 630 |
-
"logits/chosen":
|
| 631 |
-
"logits/rejected":
|
| 632 |
-
"logps/chosen": -
|
| 633 |
-
"logps/rejected": -
|
| 634 |
-
"loss": 0.
|
| 635 |
-
"pred_label":
|
| 636 |
-
"rewards/accuracies": 0.
|
| 637 |
-
"rewards/chosen": -
|
| 638 |
-
"rewards/margins":
|
| 639 |
-
"rewards/rejected": -
|
| 640 |
-
"step":
|
| 641 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 642 |
},
|
| 643 |
{
|
| 644 |
"epoch": 0.71,
|
| 645 |
-
"grad_norm":
|
| 646 |
-
"learning_rate": 1.
|
| 647 |
-
"logits/chosen":
|
| 648 |
-
"logits/rejected":
|
| 649 |
-
"logps/chosen": -
|
| 650 |
-
"logps/rejected": -
|
| 651 |
-
"loss": 0.
|
| 652 |
-
"pred_label":
|
| 653 |
-
"rewards/accuracies": 0.
|
| 654 |
-
"rewards/chosen": -
|
| 655 |
-
"rewards/margins":
|
| 656 |
-
"rewards/rejected": -
|
| 657 |
-
"step":
|
| 658 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 659 |
},
|
| 660 |
{
|
| 661 |
"epoch": 0.73,
|
| 662 |
-
"grad_norm":
|
| 663 |
-
"learning_rate": 1.
|
| 664 |
-
"logits/chosen":
|
| 665 |
-
"logits/rejected":
|
| 666 |
-
"logps/chosen": -
|
| 667 |
-
"logps/rejected": -
|
| 668 |
-
"loss": 0.
|
| 669 |
-
"pred_label":
|
| 670 |
-
"rewards/accuracies": 0.
|
| 671 |
-
"rewards/chosen": -
|
| 672 |
-
"rewards/margins":
|
| 673 |
-
"rewards/rejected": -
|
| 674 |
-
"step":
|
| 675 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 0.75,
|
| 679 |
-
"grad_norm":
|
| 680 |
-
"learning_rate": 8.
|
| 681 |
-
"logits/chosen":
|
| 682 |
-
"logits/rejected":
|
| 683 |
-
"logps/chosen": -
|
| 684 |
-
"logps/rejected": -
|
| 685 |
-
"loss": 0.
|
| 686 |
-
"pred_label":
|
| 687 |
-
"rewards/accuracies": 0.
|
| 688 |
-
"rewards/chosen": -
|
| 689 |
-
"rewards/margins":
|
| 690 |
-
"rewards/rejected": -
|
| 691 |
-
"step":
|
| 692 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 693 |
},
|
| 694 |
{
|
| 695 |
"epoch": 0.77,
|
| 696 |
-
"grad_norm":
|
| 697 |
-
"learning_rate": 7.
|
| 698 |
-
"logits/chosen":
|
| 699 |
-
"logits/rejected":
|
| 700 |
-
"logps/chosen": -
|
| 701 |
-
"logps/rejected": -
|
| 702 |
-
"loss": 0.
|
| 703 |
-
"pred_label":
|
| 704 |
-
"rewards/accuracies": 0.
|
| 705 |
-
"rewards/chosen": -
|
| 706 |
-
"rewards/margins":
|
| 707 |
-
"rewards/rejected": -
|
| 708 |
-
"step":
|
| 709 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 710 |
},
|
| 711 |
{
|
| 712 |
"epoch": 0.8,
|
| 713 |
-
"grad_norm": 5
|
| 714 |
-
"learning_rate": 6.
|
| 715 |
-
"logits/chosen":
|
| 716 |
-
"logits/rejected":
|
| 717 |
-
"logps/chosen": -
|
| 718 |
-
"logps/rejected": -
|
| 719 |
-
"loss": 0.
|
| 720 |
-
"pred_label":
|
| 721 |
-
"rewards/accuracies": 0.
|
| 722 |
-
"rewards/chosen": -
|
| 723 |
-
"rewards/margins":
|
| 724 |
-
"rewards/rejected": -
|
| 725 |
-
"step":
|
| 726 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 727 |
},
|
| 728 |
{
|
| 729 |
"epoch": 0.82,
|
| 730 |
-
"grad_norm":
|
| 731 |
-
"learning_rate": 4.
|
| 732 |
-
"logits/chosen":
|
| 733 |
-
"logits/rejected":
|
| 734 |
-
"logps/chosen": -
|
| 735 |
-
"logps/rejected": -
|
| 736 |
-
"loss": 0.
|
| 737 |
-
"pred_label":
|
| 738 |
-
"rewards/accuracies": 0.
|
| 739 |
-
"rewards/chosen": -
|
| 740 |
-
"rewards/margins":
|
| 741 |
-
"rewards/rejected": -
|
| 742 |
-
"step":
|
| 743 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 744 |
},
|
| 745 |
{
|
| 746 |
"epoch": 0.84,
|
| 747 |
-
"grad_norm":
|
| 748 |
-
"learning_rate": 3.
|
| 749 |
-
"logits/chosen":
|
| 750 |
-
"logits/rejected":
|
| 751 |
-
"logps/chosen": -
|
| 752 |
-
"logps/rejected": -
|
| 753 |
-
"loss": 0.
|
| 754 |
-
"pred_label":
|
| 755 |
-
"rewards/accuracies": 0.
|
| 756 |
-
"rewards/chosen": -
|
| 757 |
-
"rewards/margins": 0.
|
| 758 |
-
"rewards/rejected": -
|
| 759 |
-
"step":
|
| 760 |
-
"use_label":
|
| 761 |
},
|
| 762 |
{
|
| 763 |
"epoch": 0.84,
|
| 764 |
-
"eval_logits/chosen":
|
| 765 |
-
"eval_logits/rejected":
|
| 766 |
-
"eval_logps/chosen": -
|
| 767 |
-
"eval_logps/rejected": -
|
| 768 |
-
"eval_loss": 0.
|
| 769 |
-
"eval_pred_label":
|
| 770 |
-
"eval_rewards/accuracies": 0.
|
| 771 |
-
"eval_rewards/chosen": -
|
| 772 |
-
"eval_rewards/margins":
|
| 773 |
-
"eval_rewards/rejected": -
|
| 774 |
-
"eval_runtime":
|
| 775 |
-
"eval_samples_per_second":
|
| 776 |
-
"eval_steps_per_second": 0.
|
| 777 |
-
"eval_use_label":
|
| 778 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 779 |
},
|
| 780 |
{
|
| 781 |
"epoch": 0.86,
|
| 782 |
-
"grad_norm":
|
| 783 |
-
"learning_rate": 2.
|
| 784 |
-
"logits/chosen":
|
| 785 |
-
"logits/rejected":
|
| 786 |
-
"logps/chosen": -
|
| 787 |
-
"logps/rejected": -
|
| 788 |
-
"loss": 0.
|
| 789 |
-
"pred_label":
|
| 790 |
-
"rewards/accuracies": 0.
|
| 791 |
-
"rewards/chosen": -
|
| 792 |
-
"rewards/margins":
|
| 793 |
-
"rewards/rejected": -
|
| 794 |
-
"step":
|
| 795 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 796 |
},
|
| 797 |
{
|
| 798 |
"epoch": 0.88,
|
| 799 |
-
"grad_norm":
|
| 800 |
-
"learning_rate": 2.
|
| 801 |
-
"logits/chosen":
|
| 802 |
-
"logits/rejected":
|
| 803 |
-
"logps/chosen": -
|
| 804 |
-
"logps/rejected": -
|
| 805 |
-
"loss": 0.
|
| 806 |
-
"pred_label":
|
| 807 |
-
"rewards/accuracies": 0.
|
| 808 |
-
"rewards/chosen": -
|
| 809 |
-
"rewards/margins":
|
| 810 |
-
"rewards/rejected": -
|
| 811 |
-
"step":
|
| 812 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 813 |
},
|
| 814 |
{
|
| 815 |
"epoch": 0.9,
|
| 816 |
-
"grad_norm":
|
| 817 |
-
"learning_rate": 1.
|
| 818 |
-
"logits/chosen":
|
| 819 |
-
"logits/rejected":
|
| 820 |
-
"logps/chosen": -
|
| 821 |
-
"logps/rejected": -
|
| 822 |
-
"loss": 0.
|
| 823 |
-
"pred_label":
|
| 824 |
-
"rewards/accuracies": 0.
|
| 825 |
-
"rewards/chosen": -
|
| 826 |
-
"rewards/margins":
|
| 827 |
-
"rewards/rejected": -
|
| 828 |
-
"step":
|
| 829 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 830 |
},
|
| 831 |
{
|
| 832 |
"epoch": 0.92,
|
| 833 |
-
"grad_norm":
|
| 834 |
-
"learning_rate": 9.
|
| 835 |
-
"logits/chosen":
|
| 836 |
-
"logits/rejected":
|
| 837 |
-
"logps/chosen": -
|
| 838 |
-
"logps/rejected": -
|
| 839 |
-
"loss": 0.
|
| 840 |
-
"pred_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 841 |
"rewards/accuracies": 0.3125,
|
| 842 |
-
"rewards/chosen": -
|
| 843 |
-
"rewards/margins":
|
| 844 |
-
"rewards/rejected": -
|
| 845 |
-
"step":
|
| 846 |
-
"use_label":
|
| 847 |
},
|
| 848 |
{
|
| 849 |
"epoch": 0.94,
|
| 850 |
-
"
|
| 851 |
-
"
|
| 852 |
-
"
|
| 853 |
-
"
|
| 854 |
-
"
|
| 855 |
-
"
|
| 856 |
-
"
|
| 857 |
-
"
|
| 858 |
-
"
|
| 859 |
-
"
|
| 860 |
-
"
|
| 861 |
-
"
|
| 862 |
-
"
|
| 863 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 864 |
},
|
| 865 |
{
|
| 866 |
"epoch": 0.96,
|
| 867 |
-
"grad_norm":
|
| 868 |
-
"learning_rate":
|
| 869 |
-
"logits/chosen":
|
| 870 |
-
"logits/rejected":
|
| 871 |
-
"logps/chosen": -
|
| 872 |
-
"logps/rejected": -
|
| 873 |
-
"loss": 0.
|
| 874 |
-
"pred_label":
|
| 875 |
-
"rewards/accuracies": 0.
|
| 876 |
-
"rewards/chosen": -
|
| 877 |
-
"rewards/margins":
|
| 878 |
-
"rewards/rejected": -
|
| 879 |
-
"step":
|
| 880 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 881 |
},
|
| 882 |
{
|
| 883 |
"epoch": 0.98,
|
| 884 |
-
"grad_norm":
|
| 885 |
-
"learning_rate": 3.
|
| 886 |
-
"logits/chosen":
|
| 887 |
-
"logits/rejected":
|
| 888 |
-
"logps/chosen": -
|
| 889 |
-
"logps/rejected": -
|
| 890 |
-
"loss": 0.
|
| 891 |
-
"pred_label":
|
| 892 |
-
"rewards/accuracies": 0.
|
| 893 |
-
"rewards/chosen": -
|
| 894 |
-
"rewards/margins":
|
| 895 |
-
"rewards/rejected": -
|
| 896 |
-
"step":
|
| 897 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 898 |
},
|
| 899 |
{
|
| 900 |
"epoch": 1.0,
|
| 901 |
-
"step":
|
| 902 |
"total_flos": 0.0,
|
| 903 |
-
"train_loss": 0.
|
| 904 |
-
"train_runtime":
|
| 905 |
-
"train_samples_per_second":
|
| 906 |
-
"train_steps_per_second": 0.
|
| 907 |
}
|
| 908 |
],
|
| 909 |
"logging_steps": 10,
|
| 910 |
-
"max_steps":
|
| 911 |
"num_input_tokens_seen": 0,
|
| 912 |
"num_train_epochs": 1,
|
| 913 |
"save_steps": 50,
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9997382884061764,
|
| 5 |
"eval_steps": 100,
|
| 6 |
+
"global_step": 955,
|
| 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": 0.59375,
|
| 14 |
+
"learning_rate": 5.208333333333333e-08,
|
| 15 |
+
"logits/chosen": -2.1666858196258545,
|
| 16 |
+
"logits/rejected": -2.182244300842285,
|
| 17 |
+
"logps/chosen": -12.368609428405762,
|
| 18 |
+
"logps/rejected": -24.687644958496094,
|
| 19 |
"loss": 0.6931,
|
| 20 |
"pred_label": 0.0,
|
| 21 |
"rewards/accuracies": 0.0,
|
|
|
|
| 26 |
"use_label": 10.0
|
| 27 |
},
|
| 28 |
{
|
| 29 |
+
"epoch": 0.01,
|
| 30 |
+
"grad_norm": 0.6015625,
|
| 31 |
+
"learning_rate": 5.208333333333334e-07,
|
| 32 |
+
"logits/chosen": -2.2115366458892822,
|
| 33 |
+
"logits/rejected": -2.2720205783843994,
|
| 34 |
+
"logps/chosen": -57.57262420654297,
|
| 35 |
+
"logps/rejected": -65.22234344482422,
|
| 36 |
+
"loss": 0.693,
|
| 37 |
"pred_label": 0.0,
|
| 38 |
+
"rewards/accuracies": 0.25,
|
| 39 |
+
"rewards/chosen": 0.0010967346606776118,
|
| 40 |
+
"rewards/margins": 0.0003403760783839971,
|
| 41 |
+
"rewards/rejected": 0.000756358727812767,
|
| 42 |
"step": 10,
|
| 43 |
"use_label": 90.0
|
| 44 |
},
|
| 45 |
{
|
| 46 |
+
"epoch": 0.02,
|
| 47 |
+
"grad_norm": 0.6796875,
|
| 48 |
+
"learning_rate": 1.0416666666666667e-06,
|
| 49 |
+
"logits/chosen": -2.242250680923462,
|
| 50 |
+
"logits/rejected": -2.2794952392578125,
|
| 51 |
+
"logps/chosen": -56.5336799621582,
|
| 52 |
+
"logps/rejected": -68.37976837158203,
|
| 53 |
+
"loss": 0.6924,
|
| 54 |
"pred_label": 0.0,
|
| 55 |
+
"rewards/accuracies": 0.23125000298023224,
|
| 56 |
+
"rewards/chosen": 0.006666866131126881,
|
| 57 |
+
"rewards/margins": 0.0016975816106423736,
|
| 58 |
+
"rewards/rejected": 0.0049692848697304726,
|
| 59 |
"step": 20,
|
| 60 |
"use_label": 242.0
|
| 61 |
},
|
| 62 |
{
|
| 63 |
+
"epoch": 0.03,
|
| 64 |
+
"grad_norm": 0.5546875,
|
| 65 |
+
"learning_rate": 1.5625e-06,
|
| 66 |
+
"logits/chosen": -2.2632603645324707,
|
| 67 |
+
"logits/rejected": -2.2477545738220215,
|
| 68 |
+
"logps/chosen": -53.993377685546875,
|
| 69 |
+
"logps/rejected": -67.89723205566406,
|
| 70 |
+
"loss": 0.692,
|
| 71 |
"pred_label": 0.0,
|
| 72 |
"rewards/accuracies": 0.2750000059604645,
|
| 73 |
+
"rewards/chosen": 0.01642272248864174,
|
| 74 |
+
"rewards/margins": 0.0025839507579803467,
|
| 75 |
+
"rewards/rejected": 0.013838770799338818,
|
| 76 |
"step": 30,
|
| 77 |
"use_label": 402.0
|
| 78 |
},
|
| 79 |
{
|
| 80 |
+
"epoch": 0.04,
|
| 81 |
+
"grad_norm": 0.6328125,
|
| 82 |
+
"learning_rate": 2.0833333333333334e-06,
|
| 83 |
+
"logits/chosen": -2.2828402519226074,
|
| 84 |
+
"logits/rejected": -2.2757327556610107,
|
| 85 |
+
"logps/chosen": -55.58416748046875,
|
| 86 |
+
"logps/rejected": -66.58230590820312,
|
| 87 |
+
"loss": 0.6909,
|
| 88 |
"pred_label": 0.0,
|
| 89 |
+
"rewards/accuracies": 0.20624999701976776,
|
| 90 |
+
"rewards/chosen": 0.018385304138064384,
|
| 91 |
+
"rewards/margins": 0.0005377806373871863,
|
| 92 |
+
"rewards/rejected": 0.017847521230578423,
|
| 93 |
"step": 40,
|
| 94 |
"use_label": 562.0
|
| 95 |
},
|
| 96 |
{
|
| 97 |
+
"epoch": 0.05,
|
| 98 |
+
"grad_norm": 0.59765625,
|
| 99 |
+
"learning_rate": 2.604166666666667e-06,
|
| 100 |
+
"logits/chosen": -2.3447792530059814,
|
| 101 |
+
"logits/rejected": -2.334505319595337,
|
| 102 |
+
"logps/chosen": -69.1240234375,
|
| 103 |
+
"logps/rejected": -84.65351867675781,
|
| 104 |
+
"loss": 0.6888,
|
| 105 |
"pred_label": 0.0,
|
| 106 |
+
"rewards/accuracies": 0.28125,
|
| 107 |
+
"rewards/chosen": 0.02670113742351532,
|
| 108 |
+
"rewards/margins": 0.005580009426921606,
|
| 109 |
+
"rewards/rejected": 0.021121131256222725,
|
| 110 |
"step": 50,
|
| 111 |
"use_label": 722.0
|
| 112 |
},
|
| 113 |
{
|
| 114 |
+
"epoch": 0.06,
|
| 115 |
+
"grad_norm": 0.72265625,
|
| 116 |
+
"learning_rate": 3.125e-06,
|
| 117 |
+
"logits/chosen": -2.302928924560547,
|
| 118 |
+
"logits/rejected": -2.309403419494629,
|
| 119 |
+
"logps/chosen": -82.0005111694336,
|
| 120 |
+
"logps/rejected": -90.72607421875,
|
| 121 |
+
"loss": 0.6874,
|
| 122 |
"pred_label": 0.0,
|
| 123 |
+
"rewards/accuracies": 0.34375,
|
| 124 |
+
"rewards/chosen": 0.036945782601833344,
|
| 125 |
+
"rewards/margins": 0.014240987598896027,
|
| 126 |
+
"rewards/rejected": 0.022704793140292168,
|
| 127 |
"step": 60,
|
| 128 |
"use_label": 882.0
|
| 129 |
},
|
| 130 |
{
|
| 131 |
+
"epoch": 0.07,
|
| 132 |
+
"grad_norm": 0.79296875,
|
| 133 |
+
"learning_rate": 3.6458333333333333e-06,
|
| 134 |
+
"logits/chosen": -2.3445372581481934,
|
| 135 |
+
"logits/rejected": -2.325496196746826,
|
| 136 |
+
"logps/chosen": -77.20660400390625,
|
| 137 |
+
"logps/rejected": -77.66109466552734,
|
| 138 |
+
"loss": 0.685,
|
| 139 |
"pred_label": 0.0,
|
| 140 |
+
"rewards/accuracies": 0.3125,
|
| 141 |
+
"rewards/chosen": 0.025281202048063278,
|
| 142 |
+
"rewards/margins": 0.016315331682562828,
|
| 143 |
+
"rewards/rejected": 0.0089658722281456,
|
| 144 |
"step": 70,
|
| 145 |
"use_label": 1042.0
|
| 146 |
},
|
| 147 |
{
|
| 148 |
+
"epoch": 0.08,
|
| 149 |
+
"grad_norm": 0.82421875,
|
| 150 |
+
"learning_rate": 4.166666666666667e-06,
|
| 151 |
+
"logits/chosen": -2.240436315536499,
|
| 152 |
+
"logits/rejected": -2.1935336589813232,
|
| 153 |
+
"logps/chosen": -81.76426696777344,
|
| 154 |
+
"logps/rejected": -89.20713806152344,
|
| 155 |
+
"loss": 0.6805,
|
| 156 |
+
"pred_label": 0.0,
|
| 157 |
+
"rewards/accuracies": 0.32499998807907104,
|
| 158 |
+
"rewards/chosen": 0.002876642858609557,
|
| 159 |
+
"rewards/margins": 0.025313254445791245,
|
| 160 |
+
"rewards/rejected": -0.022436615079641342,
|
| 161 |
"step": 80,
|
| 162 |
+
"use_label": 1202.0
|
| 163 |
},
|
| 164 |
{
|
| 165 |
+
"epoch": 0.09,
|
| 166 |
+
"grad_norm": 2.21875,
|
| 167 |
+
"learning_rate": 4.6875000000000004e-06,
|
| 168 |
+
"logits/chosen": -2.1845104694366455,
|
| 169 |
+
"logits/rejected": -2.22868013381958,
|
| 170 |
+
"logps/chosen": -62.55183029174805,
|
| 171 |
+
"logps/rejected": -80.93482208251953,
|
| 172 |
+
"loss": 0.6746,
|
| 173 |
+
"pred_label": 0.10000000149011612,
|
| 174 |
+
"rewards/accuracies": 0.33125001192092896,
|
| 175 |
+
"rewards/chosen": -0.014620671980082989,
|
| 176 |
+
"rewards/margins": 0.048217564821243286,
|
| 177 |
+
"rewards/rejected": -0.06283824145793915,
|
| 178 |
"step": 90,
|
| 179 |
+
"use_label": 1361.9000244140625
|
| 180 |
},
|
| 181 |
{
|
| 182 |
+
"epoch": 0.1,
|
| 183 |
+
"grad_norm": 1.4453125,
|
| 184 |
+
"learning_rate": 4.9997324926814375e-06,
|
| 185 |
+
"logits/chosen": -2.1468710899353027,
|
| 186 |
+
"logits/rejected": -2.110016107559204,
|
| 187 |
+
"logps/chosen": -77.85456848144531,
|
| 188 |
+
"logps/rejected": -80.45710754394531,
|
| 189 |
+
"loss": 0.6637,
|
| 190 |
+
"pred_label": 2.5999999046325684,
|
| 191 |
+
"rewards/accuracies": 0.29374998807907104,
|
| 192 |
+
"rewards/chosen": -0.03279874473810196,
|
| 193 |
+
"rewards/margins": 0.053389471024274826,
|
| 194 |
+
"rewards/rejected": -0.08618821203708649,
|
| 195 |
"step": 100,
|
| 196 |
+
"use_label": 1519.4000244140625
|
| 197 |
},
|
| 198 |
{
|
| 199 |
+
"epoch": 0.1,
|
| 200 |
+
"eval_logits/chosen": -2.125288248062134,
|
| 201 |
+
"eval_logits/rejected": -2.0926668643951416,
|
| 202 |
+
"eval_logps/chosen": -78.37342834472656,
|
| 203 |
+
"eval_logps/rejected": -91.74456024169922,
|
| 204 |
+
"eval_loss": 0.6641501784324646,
|
| 205 |
+
"eval_pred_label": 17.079364776611328,
|
| 206 |
+
"eval_rewards/accuracies": 0.3253968358039856,
|
| 207 |
+
"eval_rewards/chosen": -0.09472885727882385,
|
| 208 |
+
"eval_rewards/margins": 0.06873615086078644,
|
| 209 |
+
"eval_rewards/rejected": -0.1634650081396103,
|
| 210 |
+
"eval_runtime": 246.0032,
|
| 211 |
+
"eval_samples_per_second": 8.13,
|
| 212 |
+
"eval_steps_per_second": 0.256,
|
| 213 |
+
"eval_use_label": 1838.920654296875,
|
| 214 |
"step": 100
|
| 215 |
},
|
| 216 |
{
|
| 217 |
+
"epoch": 0.12,
|
| 218 |
+
"grad_norm": 1.34375,
|
| 219 |
+
"learning_rate": 4.996723692767927e-06,
|
| 220 |
+
"logits/chosen": -2.155580520629883,
|
| 221 |
+
"logits/rejected": -2.13484263420105,
|
| 222 |
+
"logps/chosen": -72.5310287475586,
|
| 223 |
+
"logps/rejected": -89.02531433105469,
|
| 224 |
+
"loss": 0.6614,
|
| 225 |
+
"pred_label": 34.849998474121094,
|
| 226 |
+
"rewards/accuracies": 0.29374998807907104,
|
| 227 |
+
"rewards/chosen": -0.1076192632317543,
|
| 228 |
+
"rewards/margins": 0.05500911548733711,
|
| 229 |
+
"rewards/rejected": -0.1626283824443817,
|
| 230 |
+
"step": 110,
|
| 231 |
+
"use_label": 2151.14990234375
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.13,
|
| 235 |
"grad_norm": 2.125,
|
| 236 |
+
"learning_rate": 4.9903757462135984e-06,
|
| 237 |
+
"logits/chosen": -2.396162748336792,
|
| 238 |
+
"logits/rejected": -2.2887351512908936,
|
| 239 |
+
"logps/chosen": -96.98478698730469,
|
| 240 |
+
"logps/rejected": -113.80241394042969,
|
| 241 |
+
"loss": 0.6612,
|
| 242 |
+
"pred_label": 42.79999923706055,
|
| 243 |
"rewards/accuracies": 0.3499999940395355,
|
| 244 |
+
"rewards/chosen": -0.2291755974292755,
|
| 245 |
+
"rewards/margins": 0.09122875332832336,
|
| 246 |
+
"rewards/rejected": -0.3204043507575989,
|
| 247 |
+
"step": 120,
|
| 248 |
+
"use_label": 2303.199951171875
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.14,
|
| 252 |
+
"grad_norm": 0.890625,
|
| 253 |
+
"learning_rate": 4.980697142834315e-06,
|
| 254 |
+
"logits/chosen": -2.2367610931396484,
|
| 255 |
+
"logits/rejected": -2.245413064956665,
|
| 256 |
+
"logps/chosen": -80.27469635009766,
|
| 257 |
+
"logps/rejected": -92.39393615722656,
|
| 258 |
+
"loss": 0.6598,
|
| 259 |
+
"pred_label": 50.92499923706055,
|
| 260 |
+
"rewards/accuracies": 0.29374998807907104,
|
| 261 |
+
"rewards/chosen": -0.2180059254169464,
|
| 262 |
+
"rewards/margins": 0.05767815560102463,
|
| 263 |
+
"rewards/rejected": -0.27568405866622925,
|
| 264 |
+
"step": 130,
|
| 265 |
+
"use_label": 2455.074951171875
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.15,
|
| 269 |
+
"grad_norm": 1.3984375,
|
| 270 |
+
"learning_rate": 4.967700826904229e-06,
|
| 271 |
+
"logits/chosen": -2.204483985900879,
|
| 272 |
+
"logits/rejected": -2.235548496246338,
|
| 273 |
+
"logps/chosen": -83.83267211914062,
|
| 274 |
+
"logps/rejected": -108.3904800415039,
|
| 275 |
+
"loss": 0.6507,
|
| 276 |
+
"pred_label": 60.25,
|
| 277 |
+
"rewards/accuracies": 0.28125,
|
| 278 |
+
"rewards/chosen": -0.24559661746025085,
|
| 279 |
+
"rewards/margins": 0.08991299569606781,
|
| 280 |
+
"rewards/rejected": -0.33550962805747986,
|
| 281 |
+
"step": 140,
|
| 282 |
+
"use_label": 2605.75
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.16,
|
| 286 |
+
"grad_norm": 1.609375,
|
| 287 |
+
"learning_rate": 4.951404179843963e-06,
|
| 288 |
+
"logits/chosen": -2.2710304260253906,
|
| 289 |
+
"logits/rejected": -2.216566324234009,
|
| 290 |
+
"logps/chosen": -61.10588836669922,
|
| 291 |
+
"logps/rejected": -68.02718353271484,
|
| 292 |
+
"loss": 0.6225,
|
| 293 |
+
"pred_label": 71.0,
|
| 294 |
+
"rewards/accuracies": 0.26875001192092896,
|
| 295 |
+
"rewards/chosen": -0.13611064851284027,
|
| 296 |
+
"rewards/margins": 0.09120706468820572,
|
| 297 |
+
"rewards/rejected": -0.2273177206516266,
|
| 298 |
+
"step": 150,
|
| 299 |
+
"use_label": 2755.0
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 0.17,
|
| 303 |
+
"grad_norm": 2.0,
|
| 304 |
+
"learning_rate": 4.931828996974498e-06,
|
| 305 |
+
"logits/chosen": -2.164498805999756,
|
| 306 |
+
"logits/rejected": -2.13112211227417,
|
| 307 |
+
"logps/chosen": -156.16433715820312,
|
| 308 |
+
"logps/rejected": -179.91470336914062,
|
| 309 |
+
"loss": 0.5341,
|
| 310 |
+
"pred_label": 105.7750015258789,
|
| 311 |
+
"rewards/accuracies": 0.30000001192092896,
|
| 312 |
+
"rewards/chosen": -0.7456113696098328,
|
| 313 |
+
"rewards/margins": 0.23544755578041077,
|
| 314 |
+
"rewards/rejected": -0.9810588955879211,
|
| 315 |
+
"step": 160,
|
| 316 |
+
"use_label": 2880.22509765625
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 0.18,
|
| 320 |
+
"grad_norm": 3.015625,
|
| 321 |
+
"learning_rate": 4.909001458367867e-06,
|
| 322 |
+
"logits/chosen": -1.3832576274871826,
|
| 323 |
+
"logits/rejected": -1.3605282306671143,
|
| 324 |
+
"logps/chosen": -340.44036865234375,
|
| 325 |
+
"logps/rejected": -370.19189453125,
|
| 326 |
+
"loss": 0.4389,
|
| 327 |
+
"pred_label": 158.5500030517578,
|
| 328 |
+
"rewards/accuracies": 0.2562499940395355,
|
| 329 |
+
"rewards/chosen": -2.7648394107818604,
|
| 330 |
+
"rewards/margins": 0.27328410744667053,
|
| 331 |
+
"rewards/rejected": -3.038123607635498,
|
| 332 |
+
"step": 170,
|
| 333 |
+
"use_label": 2987.449951171875
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.19,
|
| 337 |
+
"grad_norm": 2.28125,
|
| 338 |
+
"learning_rate": 4.882952093833628e-06,
|
| 339 |
+
"logits/chosen": -0.4500812888145447,
|
| 340 |
+
"logits/rejected": -0.481560081243515,
|
| 341 |
+
"logps/chosen": -565.9759521484375,
|
| 342 |
+
"logps/rejected": -605.9683837890625,
|
| 343 |
+
"loss": 0.4367,
|
| 344 |
+
"pred_label": 221.4250030517578,
|
| 345 |
+
"rewards/accuracies": 0.23749999701976776,
|
| 346 |
+
"rewards/chosen": -5.0347394943237305,
|
| 347 |
+
"rewards/margins": 0.29014867544174194,
|
| 348 |
+
"rewards/rejected": -5.324888706207275,
|
| 349 |
+
"step": 180,
|
| 350 |
+
"use_label": 3084.574951171875
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.2,
|
| 354 |
+
"grad_norm": 1.46875,
|
| 355 |
+
"learning_rate": 4.853715742087947e-06,
|
| 356 |
+
"logits/chosen": 0.1529570072889328,
|
| 357 |
+
"logits/rejected": 0.21263575553894043,
|
| 358 |
+
"logps/chosen": -780.2406005859375,
|
| 359 |
+
"logps/rejected": -710.7351684570312,
|
| 360 |
+
"loss": 0.4048,
|
| 361 |
+
"pred_label": 286.625,
|
| 362 |
+
"rewards/accuracies": 0.23125000298023224,
|
| 363 |
+
"rewards/chosen": -7.063575744628906,
|
| 364 |
+
"rewards/margins": -0.651119589805603,
|
| 365 |
+
"rewards/rejected": -6.412456512451172,
|
| 366 |
+
"step": 190,
|
| 367 |
+
"use_label": 3179.375
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.21,
|
| 371 |
+
"grad_norm": 2.609375,
|
| 372 |
+
"learning_rate": 4.821331504159906e-06,
|
| 373 |
+
"logits/chosen": 1.329816222190857,
|
| 374 |
+
"logits/rejected": 1.2937750816345215,
|
| 375 |
+
"logps/chosen": -1144.726806640625,
|
| 376 |
+
"logps/rejected": -1046.467529296875,
|
| 377 |
+
"loss": 0.3902,
|
| 378 |
+
"pred_label": 358.82501220703125,
|
| 379 |
+
"rewards/accuracies": 0.23125000298023224,
|
| 380 |
+
"rewards/chosen": -10.646726608276367,
|
| 381 |
+
"rewards/margins": -0.9137382507324219,
|
| 382 |
+
"rewards/rejected": -9.732988357543945,
|
| 383 |
+
"step": 200,
|
| 384 |
+
"use_label": 3267.175048828125
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.21,
|
| 388 |
+
"eval_logits/chosen": 2.772732734680176,
|
| 389 |
+
"eval_logits/rejected": 2.8470869064331055,
|
| 390 |
+
"eval_logps/chosen": -1511.0870361328125,
|
| 391 |
+
"eval_logps/rejected": -1478.920166015625,
|
| 392 |
+
"eval_loss": 0.3930211067199707,
|
| 393 |
+
"eval_pred_label": 515.3015747070312,
|
| 394 |
+
"eval_rewards/accuracies": 0.255952388048172,
|
| 395 |
+
"eval_rewards/chosen": -14.421865463256836,
|
| 396 |
+
"eval_rewards/margins": -0.3866449296474457,
|
| 397 |
+
"eval_rewards/rejected": -14.0352201461792,
|
| 398 |
+
"eval_runtime": 246.1209,
|
| 399 |
+
"eval_samples_per_second": 8.126,
|
| 400 |
+
"eval_steps_per_second": 0.256,
|
| 401 |
+
"eval_use_label": 3444.698486328125,
|
| 402 |
+
"step": 200
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.22,
|
| 406 |
+
"grad_norm": 1.2421875,
|
| 407 |
+
"learning_rate": 4.7858426910973435e-06,
|
| 408 |
+
"logits/chosen": 3.3592753410339355,
|
| 409 |
+
"logits/rejected": 3.3634376525878906,
|
| 410 |
+
"logps/chosen": -1704.9111328125,
|
| 411 |
+
"logps/rejected": -1631.6551513671875,
|
| 412 |
+
"loss": 0.4013,
|
| 413 |
+
"pred_label": 668.5999755859375,
|
| 414 |
+
"rewards/accuracies": 0.23749999701976776,
|
| 415 |
+
"rewards/chosen": -16.369003295898438,
|
| 416 |
+
"rewards/margins": -0.7280259132385254,
|
| 417 |
+
"rewards/rejected": -15.640978813171387,
|
| 418 |
+
"step": 210,
|
| 419 |
+
"use_label": 3621.39990234375
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 0.23,
|
| 423 |
+
"grad_norm": 1.0390625,
|
| 424 |
+
"learning_rate": 4.747296766042161e-06,
|
| 425 |
+
"logits/chosen": 6.802922248840332,
|
| 426 |
+
"logits/rejected": 6.750025272369385,
|
| 427 |
+
"logps/chosen": -2573.23486328125,
|
| 428 |
+
"logps/rejected": -2336.566650390625,
|
| 429 |
+
"loss": 0.3777,
|
| 430 |
+
"pred_label": 745.0499877929688,
|
| 431 |
+
"rewards/accuracies": 0.21875,
|
| 432 |
+
"rewards/chosen": -24.971813201904297,
|
| 433 |
+
"rewards/margins": -2.3022866249084473,
|
| 434 |
+
"rewards/rejected": -22.669527053833008,
|
| 435 |
+
"step": 220,
|
| 436 |
+
"use_label": 3704.949951171875
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.24,
|
| 440 |
+
"grad_norm": 0.0286865234375,
|
| 441 |
+
"learning_rate": 4.705745280752586e-06,
|
| 442 |
+
"logits/chosen": 8.976715087890625,
|
| 443 |
+
"logits/rejected": 8.701696395874023,
|
| 444 |
+
"logps/chosen": -4151.056640625,
|
| 445 |
+
"logps/rejected": -3499.08251953125,
|
| 446 |
+
"loss": 0.3986,
|
| 447 |
+
"pred_label": 826.2750244140625,
|
| 448 |
+
"rewards/accuracies": 0.22499999403953552,
|
| 449 |
+
"rewards/chosen": -40.723175048828125,
|
| 450 |
+
"rewards/margins": -6.461965084075928,
|
| 451 |
+
"rewards/rejected": -34.26121139526367,
|
| 452 |
+
"step": 230,
|
| 453 |
+
"use_label": 3783.72509765625
|
| 454 |
},
|
| 455 |
{
|
| 456 |
"epoch": 0.25,
|
| 457 |
+
"grad_norm": 0.40625,
|
| 458 |
+
"learning_rate": 4.661243806657256e-06,
|
| 459 |
+
"logits/chosen": 9.849306106567383,
|
| 460 |
+
"logits/rejected": 10.053918838500977,
|
| 461 |
+
"logps/chosen": -4893.1689453125,
|
| 462 |
+
"logps/rejected": -4828.57373046875,
|
| 463 |
+
"loss": 0.3749,
|
| 464 |
+
"pred_label": 901.4000244140625,
|
| 465 |
+
"rewards/accuracies": 0.24375000596046448,
|
| 466 |
+
"rewards/chosen": -48.2945556640625,
|
| 467 |
+
"rewards/margins": -0.7481836080551147,
|
| 468 |
+
"rewards/rejected": -47.5463752746582,
|
| 469 |
+
"step": 240,
|
| 470 |
+
"use_label": 3868.60009765625
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 0.26,
|
| 474 |
+
"grad_norm": 0.69140625,
|
| 475 |
+
"learning_rate": 4.613851860533367e-06,
|
| 476 |
+
"logits/chosen": 10.89067268371582,
|
| 477 |
+
"logits/rejected": 10.545475006103516,
|
| 478 |
+
"logps/chosen": -5968.6962890625,
|
| 479 |
+
"logps/rejected": -5755.75732421875,
|
| 480 |
+
"loss": 0.3676,
|
| 481 |
+
"pred_label": 974.1500244140625,
|
| 482 |
+
"rewards/accuracies": 0.20624999701976776,
|
| 483 |
+
"rewards/chosen": -59.03568649291992,
|
| 484 |
+
"rewards/margins": -2.1258115768432617,
|
| 485 |
+
"rewards/rejected": -56.90987014770508,
|
| 486 |
+
"step": 250,
|
| 487 |
+
"use_label": 3955.85009765625
|
| 488 |
},
|
| 489 |
{
|
| 490 |
"epoch": 0.27,
|
| 491 |
+
"grad_norm": 0.2314453125,
|
| 492 |
+
"learning_rate": 4.563632824908252e-06,
|
| 493 |
+
"logits/chosen": 10.8525390625,
|
| 494 |
+
"logits/rejected": 11.303139686584473,
|
| 495 |
+
"logps/chosen": -6945.5322265625,
|
| 496 |
+
"logps/rejected": -7116.02880859375,
|
| 497 |
+
"loss": 0.3776,
|
| 498 |
+
"pred_label": 1045.125,
|
| 499 |
+
"rewards/accuracies": 0.25,
|
| 500 |
+
"rewards/chosen": -68.84590148925781,
|
| 501 |
+
"rewards/margins": 1.5749397277832031,
|
| 502 |
+
"rewards/rejected": -70.42083740234375,
|
| 503 |
+
"step": 260,
|
| 504 |
+
"use_label": 4044.875
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.28,
|
| 508 |
+
"grad_norm": 0.021484375,
|
| 509 |
+
"learning_rate": 4.510653863290871e-06,
|
| 510 |
+
"logits/chosen": 11.601091384887695,
|
| 511 |
+
"logits/rejected": 11.230550765991211,
|
| 512 |
+
"logps/chosen": -8897.8642578125,
|
| 513 |
+
"logps/rejected": -8155.3828125,
|
| 514 |
+
"loss": 0.3702,
|
| 515 |
+
"pred_label": 1122.3499755859375,
|
| 516 |
+
"rewards/accuracies": 0.24375000596046448,
|
| 517 |
+
"rewards/chosen": -88.22434997558594,
|
| 518 |
+
"rewards/margins": -7.365424156188965,
|
| 519 |
+
"rewards/rejected": -80.85891723632812,
|
| 520 |
+
"step": 270,
|
| 521 |
+
"use_label": 4127.64990234375
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 0.29,
|
| 525 |
+
"grad_norm": 0.42578125,
|
| 526 |
+
"learning_rate": 4.454985830346574e-06,
|
| 527 |
+
"logits/chosen": 12.525368690490723,
|
| 528 |
+
"logits/rejected": 12.457976341247559,
|
| 529 |
+
"logps/chosen": -7800.34521484375,
|
| 530 |
+
"logps/rejected": -7866.60009765625,
|
| 531 |
+
"loss": 0.3932,
|
| 532 |
+
"pred_label": 1195.800048828125,
|
| 533 |
+
"rewards/accuracies": 0.24375000596046448,
|
| 534 |
+
"rewards/chosen": -77.39112854003906,
|
| 535 |
+
"rewards/margins": 0.6107722520828247,
|
| 536 |
+
"rewards/rejected": -78.00189971923828,
|
| 537 |
+
"step": 280,
|
| 538 |
+
"use_label": 4214.2001953125
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 0.3,
|
| 542 |
+
"grad_norm": 4.34375,
|
| 543 |
+
"learning_rate": 4.396703177135262e-06,
|
| 544 |
+
"logits/chosen": 10.156630516052246,
|
| 545 |
+
"logits/rejected": 10.009225845336914,
|
| 546 |
+
"logps/chosen": -7658.8212890625,
|
| 547 |
+
"logps/rejected": -6712.6748046875,
|
| 548 |
+
"loss": 0.3801,
|
| 549 |
+
"pred_label": 1272.25,
|
| 550 |
+
"rewards/accuracies": 0.20624999701976776,
|
| 551 |
+
"rewards/chosen": -75.82545471191406,
|
| 552 |
+
"rewards/margins": -9.393587112426758,
|
| 553 |
+
"rewards/rejected": -66.43186950683594,
|
| 554 |
+
"step": 290,
|
| 555 |
+
"use_label": 4297.75
|
| 556 |
},
|
| 557 |
{
|
| 558 |
"epoch": 0.31,
|
| 559 |
+
"grad_norm": 13.0625,
|
| 560 |
+
"learning_rate": 4.335883851539693e-06,
|
| 561 |
+
"logits/chosen": 5.961030006408691,
|
| 562 |
+
"logits/rejected": 6.045927047729492,
|
| 563 |
+
"logps/chosen": -2592.89208984375,
|
| 564 |
+
"logps/rejected": -3060.535888671875,
|
| 565 |
+
"loss": 0.3845,
|
| 566 |
+
"pred_label": 1338.949951171875,
|
| 567 |
+
"rewards/accuracies": 0.2874999940395355,
|
| 568 |
+
"rewards/chosen": -25.38964080810547,
|
| 569 |
+
"rewards/margins": 4.593506336212158,
|
| 570 |
+
"rewards/rejected": -29.983144760131836,
|
| 571 |
+
"step": 300,
|
| 572 |
+
"use_label": 4391.0498046875
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.31,
|
| 576 |
+
"eval_logits/chosen": 5.307006359100342,
|
| 577 |
+
"eval_logits/rejected": 5.428269863128662,
|
| 578 |
+
"eval_logps/chosen": -2377.587158203125,
|
| 579 |
+
"eval_logps/rejected": -2532.249755859375,
|
| 580 |
+
"eval_loss": 0.37855055928230286,
|
| 581 |
+
"eval_pred_label": 1484.5079345703125,
|
| 582 |
+
"eval_rewards/accuracies": 0.2519841194152832,
|
| 583 |
+
"eval_rewards/chosen": -23.086862564086914,
|
| 584 |
+
"eval_rewards/margins": 1.4816526174545288,
|
| 585 |
+
"eval_rewards/rejected": -24.568513870239258,
|
| 586 |
+
"eval_runtime": 246.1616,
|
| 587 |
+
"eval_samples_per_second": 8.125,
|
| 588 |
+
"eval_steps_per_second": 0.256,
|
| 589 |
+
"eval_use_label": 4579.4921875,
|
| 590 |
+
"step": 300
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.32,
|
| 594 |
+
"grad_norm": 56.5,
|
| 595 |
+
"learning_rate": 4.2726091940171055e-06,
|
| 596 |
+
"logits/chosen": 5.159882545471191,
|
| 597 |
+
"logits/rejected": 5.152421474456787,
|
| 598 |
+
"logps/chosen": -2442.991943359375,
|
| 599 |
+
"logps/rejected": -2541.53662109375,
|
| 600 |
+
"loss": 0.3766,
|
| 601 |
+
"pred_label": 1641.7249755859375,
|
| 602 |
+
"rewards/accuracies": 0.17499999701976776,
|
| 603 |
+
"rewards/chosen": -23.940017700195312,
|
| 604 |
+
"rewards/margins": 0.8803431391716003,
|
| 605 |
+
"rewards/rejected": -24.820362091064453,
|
| 606 |
+
"step": 310,
|
| 607 |
+
"use_label": 4752.27490234375
|
| 608 |
},
|
| 609 |
{
|
| 610 |
"epoch": 0.33,
|
| 611 |
+
"grad_norm": 0.302734375,
|
| 612 |
+
"learning_rate": 4.206963828813555e-06,
|
| 613 |
+
"logits/chosen": 8.907454490661621,
|
| 614 |
+
"logits/rejected": 9.01401424407959,
|
| 615 |
+
"logps/chosen": -7245.29150390625,
|
| 616 |
+
"logps/rejected": -7561.62646484375,
|
| 617 |
+
"loss": 0.3362,
|
| 618 |
+
"pred_label": 1717.125,
|
| 619 |
+
"rewards/accuracies": 0.23749999701976776,
|
| 620 |
+
"rewards/chosen": -71.7135009765625,
|
| 621 |
+
"rewards/margins": 3.0602481365203857,
|
| 622 |
+
"rewards/rejected": -74.77375030517578,
|
| 623 |
+
"step": 320,
|
| 624 |
+
"use_label": 4836.875
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.35,
|
| 628 |
+
"grad_norm": 0.9453125,
|
| 629 |
+
"learning_rate": 4.139035550786495e-06,
|
| 630 |
+
"logits/chosen": 12.163416862487793,
|
| 631 |
+
"logits/rejected": 12.14716911315918,
|
| 632 |
+
"logps/chosen": -9508.2734375,
|
| 633 |
+
"logps/rejected": -9237.6103515625,
|
| 634 |
+
"loss": 0.3531,
|
| 635 |
+
"pred_label": 1796.4000244140625,
|
| 636 |
"rewards/accuracies": 0.26875001192092896,
|
| 637 |
+
"rewards/chosen": -94.44801330566406,
|
| 638 |
+
"rewards/margins": -2.71445894241333,
|
| 639 |
+
"rewards/rejected": -91.7335433959961,
|
| 640 |
+
"step": 330,
|
| 641 |
+
"use_label": 4917.60009765625
|
| 642 |
},
|
| 643 |
{
|
| 644 |
"epoch": 0.36,
|
| 645 |
+
"grad_norm": 0.330078125,
|
| 646 |
+
"learning_rate": 4.068915207986931e-06,
|
| 647 |
+
"logits/chosen": 10.884541511535645,
|
| 648 |
+
"logits/rejected": 11.387506484985352,
|
| 649 |
+
"logps/chosen": -8130.7841796875,
|
| 650 |
+
"logps/rejected": -9010.3349609375,
|
| 651 |
+
"loss": 0.3848,
|
| 652 |
+
"pred_label": 1873.875,
|
| 653 |
+
"rewards/accuracies": 0.22499999403953552,
|
| 654 |
+
"rewards/chosen": -80.67778015136719,
|
| 655 |
+
"rewards/margins": 8.717926025390625,
|
| 656 |
+
"rewards/rejected": -89.39569854736328,
|
| 657 |
+
"step": 340,
|
| 658 |
+
"use_label": 5000.125
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 0.37,
|
| 662 |
+
"grad_norm": 0.00811767578125,
|
| 663 |
+
"learning_rate": 3.996696580158211e-06,
|
| 664 |
+
"logits/chosen": 10.507772445678711,
|
| 665 |
+
"logits/rejected": 10.694246292114258,
|
| 666 |
+
"logps/chosen": -8653.701171875,
|
| 667 |
+
"logps/rejected": -8258.5986328125,
|
| 668 |
+
"loss": 0.3852,
|
| 669 |
+
"pred_label": 1942.25,
|
| 670 |
+
"rewards/accuracies": 0.20000000298023224,
|
| 671 |
+
"rewards/chosen": -85.9305419921875,
|
| 672 |
+
"rewards/margins": -3.9906578063964844,
|
| 673 |
+
"rewards/rejected": -81.93989562988281,
|
| 674 |
+
"step": 350,
|
| 675 |
+
"use_label": 5091.75
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 0.38,
|
| 679 |
+
"grad_norm": 0.0125732421875,
|
| 680 |
+
"learning_rate": 3.922476253313921e-06,
|
| 681 |
+
"logits/chosen": 10.046838760375977,
|
| 682 |
+
"logits/rejected": 10.214472770690918,
|
| 683 |
+
"logps/chosen": -9018.0205078125,
|
| 684 |
+
"logps/rejected": -8944.6474609375,
|
| 685 |
+
"loss": 0.3701,
|
| 686 |
+
"pred_label": 2021.8499755859375,
|
| 687 |
+
"rewards/accuracies": 0.2562499940395355,
|
| 688 |
+
"rewards/chosen": -89.48954010009766,
|
| 689 |
+
"rewards/margins": -0.6933708190917969,
|
| 690 |
+
"rewards/rejected": -88.79617309570312,
|
| 691 |
+
"step": 360,
|
| 692 |
+
"use_label": 5172.14990234375
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 0.39,
|
| 696 |
+
"grad_norm": 0.65234375,
|
| 697 |
+
"learning_rate": 3.846353490562664e-06,
|
| 698 |
+
"logits/chosen": 10.093267440795898,
|
| 699 |
+
"logits/rejected": 10.333730697631836,
|
| 700 |
+
"logps/chosen": -8933.8212890625,
|
| 701 |
+
"logps/rejected": -8097.73193359375,
|
| 702 |
+
"loss": 0.3744,
|
| 703 |
+
"pred_label": 2098.5,
|
| 704 |
+
"rewards/accuracies": 0.25,
|
| 705 |
+
"rewards/chosen": -88.5972671508789,
|
| 706 |
+
"rewards/margins": -8.313589096069336,
|
| 707 |
+
"rewards/rejected": -80.28367614746094,
|
| 708 |
+
"step": 370,
|
| 709 |
+
"use_label": 5255.5
|
| 710 |
},
|
| 711 |
{
|
| 712 |
"epoch": 0.4,
|
| 713 |
+
"grad_norm": 1.2265625,
|
| 714 |
+
"learning_rate": 3.768430099352445e-06,
|
| 715 |
+
"logits/chosen": 7.944075107574463,
|
| 716 |
+
"logits/rejected": 8.237360000610352,
|
| 717 |
+
"logps/chosen": -5071.62255859375,
|
| 718 |
+
"logps/rejected": -5702.30810546875,
|
| 719 |
+
"loss": 0.3681,
|
| 720 |
+
"pred_label": 2180.125,
|
| 721 |
+
"rewards/accuracies": 0.29374998807907104,
|
| 722 |
+
"rewards/chosen": -50.120689392089844,
|
| 723 |
+
"rewards/margins": 6.213594436645508,
|
| 724 |
+
"rewards/rejected": -56.33428192138672,
|
| 725 |
+
"step": 380,
|
| 726 |
+
"use_label": 5333.875
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 0.41,
|
| 730 |
+
"grad_norm": 0.1337890625,
|
| 731 |
+
"learning_rate": 3.6888102953122307e-06,
|
| 732 |
+
"logits/chosen": 8.87930679321289,
|
| 733 |
+
"logits/rejected": 8.867382049560547,
|
| 734 |
+
"logps/chosen": -7653.53515625,
|
| 735 |
+
"logps/rejected": -6970.58056640625,
|
| 736 |
+
"loss": 0.3695,
|
| 737 |
+
"pred_label": 2257.35009765625,
|
| 738 |
+
"rewards/accuracies": 0.23125000298023224,
|
| 739 |
+
"rewards/chosen": -75.68672943115234,
|
| 740 |
+
"rewards/margins": -6.652501106262207,
|
| 741 |
+
"rewards/rejected": -69.03422546386719,
|
| 742 |
+
"step": 390,
|
| 743 |
+
"use_label": 5416.64990234375
|
| 744 |
},
|
| 745 |
{
|
| 746 |
"epoch": 0.42,
|
| 747 |
+
"grad_norm": 2.296875,
|
| 748 |
+
"learning_rate": 3.607600562872785e-06,
|
| 749 |
+
"logits/chosen": 10.617076873779297,
|
| 750 |
+
"logits/rejected": 10.583547592163086,
|
| 751 |
+
"logps/chosen": -10715.9619140625,
|
| 752 |
+
"logps/rejected": -9991.853515625,
|
| 753 |
+
"loss": 0.3477,
|
| 754 |
+
"pred_label": 2331.199951171875,
|
| 755 |
+
"rewards/accuracies": 0.25,
|
| 756 |
+
"rewards/chosen": -106.44930267333984,
|
| 757 |
+
"rewards/margins": -7.326300144195557,
|
| 758 |
+
"rewards/rejected": -99.12300872802734,
|
| 759 |
+
"step": 400,
|
| 760 |
+
"use_label": 5502.7998046875
|
| 761 |
},
|
| 762 |
{
|
| 763 |
"epoch": 0.42,
|
| 764 |
+
"eval_logits/chosen": 11.571577072143555,
|
| 765 |
+
"eval_logits/rejected": 11.681580543518066,
|
| 766 |
+
"eval_logps/chosen": -11201.4892578125,
|
| 767 |
+
"eval_logps/rejected": -11028.3408203125,
|
| 768 |
+
"eval_loss": 0.36222168803215027,
|
| 769 |
+
"eval_pred_label": 2485.508056640625,
|
| 770 |
+
"eval_rewards/accuracies": 0.25,
|
| 771 |
+
"eval_rewards/chosen": -111.32588958740234,
|
| 772 |
+
"eval_rewards/margins": -1.7964569330215454,
|
| 773 |
+
"eval_rewards/rejected": -109.5294418334961,
|
| 774 |
+
"eval_runtime": 246.3154,
|
| 775 |
+
"eval_samples_per_second": 8.12,
|
| 776 |
+
"eval_steps_per_second": 0.256,
|
| 777 |
+
"eval_use_label": 5682.4921875,
|
| 778 |
+
"step": 400
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 0.43,
|
| 782 |
+
"grad_norm": 10.4375,
|
| 783 |
+
"learning_rate": 3.5249095128531863e-06,
|
| 784 |
+
"logits/chosen": 11.758584976196289,
|
| 785 |
+
"logits/rejected": 11.3671236038208,
|
| 786 |
+
"logps/chosen": -10916.1015625,
|
| 787 |
+
"logps/rejected": -9542.705078125,
|
| 788 |
+
"loss": 0.3836,
|
| 789 |
+
"pred_label": 2650.5,
|
| 790 |
+
"rewards/accuracies": 0.22499999403953552,
|
| 791 |
+
"rewards/chosen": -108.51594543457031,
|
| 792 |
+
"rewards/margins": -13.66784381866455,
|
| 793 |
+
"rewards/rejected": -94.84810638427734,
|
| 794 |
+
"step": 410,
|
| 795 |
+
"use_label": 5847.5
|
| 796 |
},
|
| 797 |
{
|
| 798 |
"epoch": 0.44,
|
| 799 |
+
"grad_norm": 7.3909759521484375e-06,
|
| 800 |
+
"learning_rate": 3.4408477372034743e-06,
|
| 801 |
+
"logits/chosen": 13.165475845336914,
|
| 802 |
+
"logits/rejected": 13.11998176574707,
|
| 803 |
+
"logps/chosen": -12804.3779296875,
|
| 804 |
+
"logps/rejected": -11472.544921875,
|
| 805 |
+
"loss": 0.3738,
|
| 806 |
+
"pred_label": 2723.27490234375,
|
| 807 |
+
"rewards/accuracies": 0.20000000298023224,
|
| 808 |
+
"rewards/chosen": -127.4073486328125,
|
| 809 |
+
"rewards/margins": -13.362741470336914,
|
| 810 |
+
"rewards/rejected": -114.04459381103516,
|
| 811 |
+
"step": 420,
|
| 812 |
+
"use_label": 5934.72509765625
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 0.45,
|
| 816 |
+
"grad_norm": 0.048095703125,
|
| 817 |
+
"learning_rate": 3.355527661097728e-06,
|
| 818 |
+
"logits/chosen": 13.336782455444336,
|
| 819 |
+
"logits/rejected": 13.038861274719238,
|
| 820 |
+
"logps/chosen": -14107.853515625,
|
| 821 |
+
"logps/rejected": -11440.05078125,
|
| 822 |
+
"loss": 0.3776,
|
| 823 |
+
"pred_label": 2801.625,
|
| 824 |
+
"rewards/accuracies": 0.20624999701976776,
|
| 825 |
+
"rewards/chosen": -140.3138427734375,
|
| 826 |
+
"rewards/margins": -26.59689712524414,
|
| 827 |
+
"rewards/rejected": -113.7169418334961,
|
| 828 |
+
"step": 430,
|
| 829 |
+
"use_label": 6016.375
|
| 830 |
},
|
| 831 |
{
|
| 832 |
"epoch": 0.46,
|
| 833 |
+
"grad_norm": 0.7578125,
|
| 834 |
+
"learning_rate": 3.269063392575352e-06,
|
| 835 |
+
"logits/chosen": 12.788106918334961,
|
| 836 |
+
"logits/rejected": 12.365851402282715,
|
| 837 |
+
"logps/chosen": -12373.81640625,
|
| 838 |
+
"logps/rejected": -11545.1806640625,
|
| 839 |
+
"loss": 0.3603,
|
| 840 |
+
"pred_label": 2876.47509765625,
|
| 841 |
+
"rewards/accuracies": 0.21250000596046448,
|
| 842 |
+
"rewards/chosen": -123.06370544433594,
|
| 843 |
+
"rewards/margins": -8.244508743286133,
|
| 844 |
+
"rewards/rejected": -114.81917572021484,
|
| 845 |
+
"step": 440,
|
| 846 |
+
"use_label": 6101.52490234375
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 0.47,
|
| 850 |
+
"grad_norm": 0.00069427490234375,
|
| 851 |
+
"learning_rate": 3.181570569931697e-06,
|
| 852 |
+
"logits/chosen": 13.790414810180664,
|
| 853 |
+
"logits/rejected": 13.474637985229492,
|
| 854 |
+
"logps/chosen": -15785.755859375,
|
| 855 |
+
"logps/rejected": -14729.9326171875,
|
| 856 |
+
"loss": 0.3682,
|
| 857 |
+
"pred_label": 2952.175048828125,
|
| 858 |
+
"rewards/accuracies": 0.22499999403953552,
|
| 859 |
+
"rewards/chosen": -157.1293487548828,
|
| 860 |
+
"rewards/margins": -10.590240478515625,
|
| 861 |
+
"rewards/rejected": -146.53912353515625,
|
| 862 |
+
"step": 450,
|
| 863 |
+
"use_label": 6185.8251953125
|
| 864 |
},
|
| 865 |
{
|
| 866 |
"epoch": 0.48,
|
| 867 |
+
"grad_norm": 0.000614166259765625,
|
| 868 |
+
"learning_rate": 3.09316620706208e-06,
|
| 869 |
+
"logits/chosen": 13.132097244262695,
|
| 870 |
+
"logits/rejected": 13.091282844543457,
|
| 871 |
+
"logps/chosen": -11774.80859375,
|
| 872 |
+
"logps/rejected": -10042.3125,
|
| 873 |
+
"loss": 0.3774,
|
| 874 |
+
"pred_label": 3023.02490234375,
|
| 875 |
+
"rewards/accuracies": 0.16249999403953552,
|
| 876 |
+
"rewards/chosen": -117.1781005859375,
|
| 877 |
+
"rewards/margins": -17.27303695678711,
|
| 878 |
+
"rewards/rejected": -99.90506744384766,
|
| 879 |
+
"step": 460,
|
| 880 |
+
"use_label": 6274.97509765625
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"epoch": 0.49,
|
| 884 |
+
"grad_norm": 0.05908203125,
|
| 885 |
+
"learning_rate": 3.0039685369660785e-06,
|
| 886 |
+
"logits/chosen": 13.827133178710938,
|
| 887 |
+
"logits/rejected": 13.836956977844238,
|
| 888 |
+
"logps/chosen": -15567.875,
|
| 889 |
+
"logps/rejected": -13399.90625,
|
| 890 |
+
"loss": 0.3738,
|
| 891 |
+
"pred_label": 3091.050048828125,
|
| 892 |
+
"rewards/accuracies": 0.20624999701976776,
|
| 893 |
+
"rewards/chosen": -155.00961303710938,
|
| 894 |
+
"rewards/margins": -21.69435691833496,
|
| 895 |
+
"rewards/rejected": -133.31524658203125,
|
| 896 |
+
"step": 470,
|
| 897 |
+
"use_label": 6366.9501953125
|
| 898 |
},
|
| 899 |
{
|
| 900 |
"epoch": 0.5,
|
| 901 |
+
"grad_norm": 0.44140625,
|
| 902 |
+
"learning_rate": 2.91409685362137e-06,
|
| 903 |
+
"logits/chosen": 13.592124938964844,
|
| 904 |
+
"logits/rejected": 13.378082275390625,
|
| 905 |
+
"logps/chosen": -14692.517578125,
|
| 906 |
+
"logps/rejected": -14636.0673828125,
|
| 907 |
+
"loss": 0.3671,
|
| 908 |
+
"pred_label": 3167.925048828125,
|
| 909 |
+
"rewards/accuracies": 0.24375000596046448,
|
| 910 |
+
"rewards/chosen": -146.17376708984375,
|
| 911 |
+
"rewards/margins": -0.5919105410575867,
|
| 912 |
+
"rewards/rejected": -145.5818328857422,
|
| 913 |
+
"step": 480,
|
| 914 |
+
"use_label": 6450.0751953125
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 0.51,
|
| 918 |
+
"grad_norm": 0.01007080078125,
|
| 919 |
+
"learning_rate": 2.8236713524386085e-06,
|
| 920 |
+
"logits/chosen": 12.93031120300293,
|
| 921 |
+
"logits/rejected": 13.041865348815918,
|
| 922 |
+
"logps/chosen": -13468.158203125,
|
| 923 |
+
"logps/rejected": -11905.46875,
|
| 924 |
+
"loss": 0.3571,
|
| 925 |
+
"pred_label": 3244.675048828125,
|
| 926 |
+
"rewards/accuracies": 0.17499999701976776,
|
| 927 |
+
"rewards/chosen": -134.01531982421875,
|
| 928 |
+
"rewards/margins": -15.559646606445312,
|
| 929 |
+
"rewards/rejected": -118.45567321777344,
|
| 930 |
+
"step": 490,
|
| 931 |
+
"use_label": 6533.3251953125
|
| 932 |
},
|
| 933 |
{
|
| 934 |
"epoch": 0.52,
|
| 935 |
+
"grad_norm": 0.037109375,
|
| 936 |
+
"learning_rate": 2.7328129695107205e-06,
|
| 937 |
+
"logits/chosen": 14.4503173828125,
|
| 938 |
+
"logits/rejected": 13.91405963897705,
|
| 939 |
+
"logps/chosen": -16287.3720703125,
|
| 940 |
+
"logps/rejected": -13856.4765625,
|
| 941 |
+
"loss": 0.3468,
|
| 942 |
+
"pred_label": 3323.050048828125,
|
| 943 |
+
"rewards/accuracies": 0.23749999701976776,
|
| 944 |
+
"rewards/chosen": -162.11231994628906,
|
| 945 |
+
"rewards/margins": -24.314170837402344,
|
| 946 |
+
"rewards/rejected": -137.79815673828125,
|
| 947 |
+
"step": 500,
|
| 948 |
+
"use_label": 6614.9501953125
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.52,
|
| 952 |
+
"eval_logits/chosen": 13.83466911315918,
|
| 953 |
+
"eval_logits/rejected": 13.888518333435059,
|
| 954 |
+
"eval_logps/chosen": -14546.7158203125,
|
| 955 |
+
"eval_logps/rejected": -14149.482421875,
|
| 956 |
+
"eval_loss": 0.3612539768218994,
|
| 957 |
+
"eval_pred_label": 3487.761962890625,
|
| 958 |
+
"eval_rewards/accuracies": 0.2420634925365448,
|
| 959 |
+
"eval_rewards/chosen": -144.7781524658203,
|
| 960 |
+
"eval_rewards/margins": -4.037316799163818,
|
| 961 |
+
"eval_rewards/rejected": -140.7408447265625,
|
| 962 |
+
"eval_runtime": 246.348,
|
| 963 |
+
"eval_samples_per_second": 8.119,
|
| 964 |
+
"eval_steps_per_second": 0.256,
|
| 965 |
+
"eval_use_label": 6784.23828125,
|
| 966 |
+
"step": 500
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 0.53,
|
| 970 |
+
"grad_norm": 0.02685546875,
|
| 971 |
+
"learning_rate": 2.641643219871597e-06,
|
| 972 |
+
"logits/chosen": 13.631937026977539,
|
| 973 |
+
"logits/rejected": 13.27366828918457,
|
| 974 |
+
"logps/chosen": -13364.251953125,
|
| 975 |
+
"logps/rejected": -12129.9521484375,
|
| 976 |
+
"loss": 0.3764,
|
| 977 |
+
"pred_label": 3648.10009765625,
|
| 978 |
+
"rewards/accuracies": 0.22499999403953552,
|
| 979 |
+
"rewards/chosen": -133.00006103515625,
|
| 980 |
+
"rewards/margins": -12.405997276306152,
|
| 981 |
+
"rewards/rejected": -120.59407043457031,
|
| 982 |
+
"step": 510,
|
| 983 |
+
"use_label": 6953.89990234375
|
| 984 |
},
|
| 985 |
{
|
| 986 |
"epoch": 0.54,
|
| 987 |
+
"grad_norm": 0.0673828125,
|
| 988 |
+
"learning_rate": 2.5502840349805074e-06,
|
| 989 |
+
"logits/chosen": 13.079228401184082,
|
| 990 |
+
"logits/rejected": 12.768109321594238,
|
| 991 |
+
"logps/chosen": -13865.537109375,
|
| 992 |
+
"logps/rejected": -13542.833984375,
|
| 993 |
+
"loss": 0.3732,
|
| 994 |
+
"pred_label": 3727.324951171875,
|
| 995 |
+
"rewards/accuracies": 0.2562499940395355,
|
| 996 |
+
"rewards/chosen": -137.93142700195312,
|
| 997 |
+
"rewards/margins": -3.2280335426330566,
|
| 998 |
+
"rewards/rejected": -134.70339965820312,
|
| 999 |
+
"step": 520,
|
| 1000 |
+
"use_label": 7034.6748046875
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 0.55,
|
| 1004 |
+
"grad_norm": 0.08056640625,
|
| 1005 |
+
"learning_rate": 2.4588575996495797e-06,
|
| 1006 |
+
"logits/chosen": 13.36449909210205,
|
| 1007 |
+
"logits/rejected": 13.319116592407227,
|
| 1008 |
+
"logps/chosen": -14281.740234375,
|
| 1009 |
+
"logps/rejected": -13077.5888671875,
|
| 1010 |
+
"loss": 0.372,
|
| 1011 |
+
"pred_label": 3806.125,
|
| 1012 |
+
"rewards/accuracies": 0.21875,
|
| 1013 |
+
"rewards/chosen": -142.07461547851562,
|
| 1014 |
+
"rewards/margins": -11.974178314208984,
|
| 1015 |
+
"rewards/rejected": -130.10043334960938,
|
| 1016 |
+
"step": 530,
|
| 1017 |
+
"use_label": 7115.875
|
| 1018 |
},
|
| 1019 |
{
|
| 1020 |
"epoch": 0.57,
|
| 1021 |
+
"grad_norm": 0.396484375,
|
| 1022 |
+
"learning_rate": 2.367486188632446e-06,
|
| 1023 |
+
"logits/chosen": 13.236615180969238,
|
| 1024 |
+
"logits/rejected": 12.988927841186523,
|
| 1025 |
+
"logps/chosen": -14595.607421875,
|
| 1026 |
+
"logps/rejected": -13356.328125,
|
| 1027 |
+
"loss": 0.3668,
|
| 1028 |
+
"pred_label": 3884.25,
|
| 1029 |
+
"rewards/accuracies": 0.23749999701976776,
|
| 1030 |
+
"rewards/chosen": -145.1886749267578,
|
| 1031 |
+
"rewards/margins": -12.35832405090332,
|
| 1032 |
+
"rewards/rejected": -132.83035278320312,
|
| 1033 |
+
"step": 540,
|
| 1034 |
+
"use_label": 7197.75
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 0.58,
|
| 1038 |
+
"grad_norm": 0.060302734375,
|
| 1039 |
+
"learning_rate": 2.276292003092593e-06,
|
| 1040 |
+
"logits/chosen": 12.537747383117676,
|
| 1041 |
+
"logits/rejected": 12.562417984008789,
|
| 1042 |
+
"logps/chosen": -12560.6533203125,
|
| 1043 |
+
"logps/rejected": -12208.041015625,
|
| 1044 |
+
"loss": 0.3806,
|
| 1045 |
+
"pred_label": 3956.47509765625,
|
| 1046 |
+
"rewards/accuracies": 0.21250000596046448,
|
| 1047 |
+
"rewards/chosen": -124.9259262084961,
|
| 1048 |
+
"rewards/margins": -3.4418251514434814,
|
| 1049 |
+
"rewards/rejected": -121.4841079711914,
|
| 1050 |
+
"step": 550,
|
| 1051 |
+
"use_label": 7285.52490234375
|
| 1052 |
},
|
| 1053 |
{
|
| 1054 |
"epoch": 0.59,
|
| 1055 |
+
"grad_norm": 1.2890625,
|
| 1056 |
+
"learning_rate": 2.1853970071701415e-06,
|
| 1057 |
+
"logits/chosen": 11.63469409942627,
|
| 1058 |
+
"logits/rejected": 11.629963874816895,
|
| 1059 |
+
"logps/chosen": -13495.25,
|
| 1060 |
+
"logps/rejected": -12446.287109375,
|
| 1061 |
+
"loss": 0.3743,
|
| 1062 |
+
"pred_label": 4028.175048828125,
|
| 1063 |
+
"rewards/accuracies": 0.2562499940395355,
|
| 1064 |
+
"rewards/chosen": -134.25692749023438,
|
| 1065 |
+
"rewards/margins": -10.464034080505371,
|
| 1066 |
+
"rewards/rejected": -123.79289245605469,
|
| 1067 |
+
"step": 560,
|
| 1068 |
+
"use_label": 7373.8251953125
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 0.6,
|
| 1072 |
+
"grad_norm": 0.04296875,
|
| 1073 |
+
"learning_rate": 2.0949227648656194e-06,
|
| 1074 |
+
"logits/chosen": 11.015595436096191,
|
| 1075 |
+
"logits/rejected": 11.20518684387207,
|
| 1076 |
+
"logps/chosen": -13153.728515625,
|
| 1077 |
+
"logps/rejected": -13688.5185546875,
|
| 1078 |
+
"loss": 0.3649,
|
| 1079 |
+
"pred_label": 4102.85009765625,
|
| 1080 |
+
"rewards/accuracies": 0.23125000298023224,
|
| 1081 |
+
"rewards/chosen": -130.8712921142578,
|
| 1082 |
+
"rewards/margins": 5.270207405090332,
|
| 1083 |
+
"rewards/rejected": -136.1414794921875,
|
| 1084 |
+
"step": 570,
|
| 1085 |
+
"use_label": 7459.14990234375
|
| 1086 |
},
|
| 1087 |
{
|
| 1088 |
"epoch": 0.61,
|
| 1089 |
+
"grad_norm": 0.00982666015625,
|
| 1090 |
+
"learning_rate": 2.00499027745888e-06,
|
| 1091 |
+
"logits/chosen": 12.113763809204102,
|
| 1092 |
+
"logits/rejected": 11.809616088867188,
|
| 1093 |
+
"logps/chosen": -15668.1591796875,
|
| 1094 |
+
"logps/rejected": -14101.8125,
|
| 1095 |
+
"loss": 0.3705,
|
| 1096 |
+
"pred_label": 4181.0751953125,
|
| 1097 |
+
"rewards/accuracies": 0.21875,
|
| 1098 |
+
"rewards/chosen": -155.9324493408203,
|
| 1099 |
+
"rewards/margins": -15.698641777038574,
|
| 1100 |
+
"rewards/rejected": -140.23382568359375,
|
| 1101 |
+
"step": 580,
|
| 1102 |
+
"use_label": 7540.9248046875
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.62,
|
| 1106 |
+
"grad_norm": 0.890625,
|
| 1107 |
+
"learning_rate": 1.915719821680624e-06,
|
| 1108 |
+
"logits/chosen": 12.389215469360352,
|
| 1109 |
+
"logits/rejected": 12.470504760742188,
|
| 1110 |
+
"logps/chosen": -16380.21875,
|
| 1111 |
+
"logps/rejected": -15518.1845703125,
|
| 1112 |
+
"loss": 0.348,
|
| 1113 |
+
"pred_label": 4263.5751953125,
|
| 1114 |
+
"rewards/accuracies": 0.2562499940395355,
|
| 1115 |
+
"rewards/chosen": -163.0052947998047,
|
| 1116 |
+
"rewards/margins": -8.633197784423828,
|
| 1117 |
+
"rewards/rejected": -154.3721160888672,
|
| 1118 |
+
"step": 590,
|
| 1119 |
+
"use_label": 7618.4248046875
|
| 1120 |
},
|
| 1121 |
{
|
| 1122 |
"epoch": 0.63,
|
| 1123 |
+
"grad_norm": 0.009765625,
|
| 1124 |
+
"learning_rate": 1.8272307888529276e-06,
|
| 1125 |
+
"logits/chosen": 12.53069019317627,
|
| 1126 |
+
"logits/rejected": 12.290223121643066,
|
| 1127 |
+
"logps/chosen": -15974.0361328125,
|
| 1128 |
+
"logps/rejected": -16418.32421875,
|
| 1129 |
+
"loss": 0.33,
|
| 1130 |
+
"pred_label": 4358.25,
|
| 1131 |
+
"rewards/accuracies": 0.3062500059604645,
|
| 1132 |
+
"rewards/chosen": -158.99710083007812,
|
| 1133 |
+
"rewards/margins": 4.279976844787598,
|
| 1134 |
+
"rewards/rejected": -163.27706909179688,
|
| 1135 |
+
"step": 600,
|
| 1136 |
+
"use_label": 7683.75
|
| 1137 |
},
|
| 1138 |
{
|
| 1139 |
"epoch": 0.63,
|
| 1140 |
+
"eval_logits/chosen": 12.539884567260742,
|
| 1141 |
+
"eval_logits/rejected": 12.59428882598877,
|
| 1142 |
+
"eval_logps/chosen": -14370.5693359375,
|
| 1143 |
+
"eval_logps/rejected": -13958.7626953125,
|
| 1144 |
+
"eval_loss": 0.36051690578460693,
|
| 1145 |
+
"eval_pred_label": 4518.5712890625,
|
| 1146 |
+
"eval_rewards/accuracies": 0.2400793582201004,
|
| 1147 |
+
"eval_rewards/chosen": -143.0166778564453,
|
| 1148 |
+
"eval_rewards/margins": -4.183051586151123,
|
| 1149 |
+
"eval_rewards/rejected": -138.83364868164062,
|
| 1150 |
+
"eval_runtime": 246.0752,
|
| 1151 |
+
"eval_samples_per_second": 8.128,
|
| 1152 |
+
"eval_steps_per_second": 0.256,
|
| 1153 |
+
"eval_use_label": 7857.4287109375,
|
| 1154 |
+
"step": 600
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 0.64,
|
| 1158 |
+
"grad_norm": 3.421875,
|
| 1159 |
+
"learning_rate": 1.739641525213929e-06,
|
| 1160 |
+
"logits/chosen": 12.129522323608398,
|
| 1161 |
+
"logits/rejected": 12.104207038879395,
|
| 1162 |
+
"logps/chosen": -12273.0908203125,
|
| 1163 |
+
"logps/rejected": -13324.2001953125,
|
| 1164 |
+
"loss": 0.3506,
|
| 1165 |
+
"pred_label": 4680.1748046875,
|
| 1166 |
+
"rewards/accuracies": 0.26875001192092896,
|
| 1167 |
+
"rewards/chosen": -122.141845703125,
|
| 1168 |
+
"rewards/margins": 10.358896255493164,
|
| 1169 |
+
"rewards/rejected": -132.500732421875,
|
| 1170 |
+
"step": 610,
|
| 1171 |
+
"use_label": 8025.8251953125
|
| 1172 |
},
|
| 1173 |
{
|
| 1174 |
"epoch": 0.65,
|
| 1175 |
+
"grad_norm": 0.87109375,
|
| 1176 |
+
"learning_rate": 1.6530691736402317e-06,
|
| 1177 |
+
"logits/chosen": 11.378915786743164,
|
| 1178 |
+
"logits/rejected": 11.41844367980957,
|
| 1179 |
+
"logps/chosen": -12004.58203125,
|
| 1180 |
+
"logps/rejected": -13227.478515625,
|
| 1181 |
+
"loss": 0.3751,
|
| 1182 |
+
"pred_label": 4761.7001953125,
|
| 1183 |
+
"rewards/accuracies": 0.28125,
|
| 1184 |
+
"rewards/chosen": -119.42872619628906,
|
| 1185 |
+
"rewards/margins": 12.045063972473145,
|
| 1186 |
+
"rewards/rejected": -131.47378540039062,
|
| 1187 |
+
"step": 620,
|
| 1188 |
+
"use_label": 8104.2998046875
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 0.66,
|
| 1192 |
+
"grad_norm": 0.98046875,
|
| 1193 |
+
"learning_rate": 1.5676295169786864e-06,
|
| 1194 |
+
"logits/chosen": 11.24705696105957,
|
| 1195 |
+
"logits/rejected": 10.89767074584961,
|
| 1196 |
+
"logps/chosen": -13202.2216796875,
|
| 1197 |
+
"logps/rejected": -12557.0341796875,
|
| 1198 |
+
"loss": 0.358,
|
| 1199 |
+
"pred_label": 4834.5751953125,
|
| 1200 |
+
"rewards/accuracies": 0.20000000298023224,
|
| 1201 |
+
"rewards/chosen": -131.29196166992188,
|
| 1202 |
+
"rewards/margins": -6.4669342041015625,
|
| 1203 |
+
"rewards/rejected": -124.82503509521484,
|
| 1204 |
+
"step": 630,
|
| 1205 |
+
"use_label": 8191.4248046875
|
| 1206 |
},
|
| 1207 |
{
|
| 1208 |
"epoch": 0.67,
|
| 1209 |
+
"grad_norm": 1.6953125,
|
| 1210 |
+
"learning_rate": 1.4834368231970922e-06,
|
| 1211 |
+
"logits/chosen": 10.929521560668945,
|
| 1212 |
+
"logits/rejected": 10.978549003601074,
|
| 1213 |
+
"logps/chosen": -13150.6201171875,
|
| 1214 |
+
"logps/rejected": -11853.0263671875,
|
| 1215 |
+
"loss": 0.3492,
|
| 1216 |
+
"pred_label": 4917.47509765625,
|
| 1217 |
+
"rewards/accuracies": 0.23125000298023224,
|
| 1218 |
+
"rewards/chosen": -130.78927612304688,
|
| 1219 |
+
"rewards/margins": -12.910749435424805,
|
| 1220 |
+
"rewards/rejected": -117.87852478027344,
|
| 1221 |
+
"step": 640,
|
| 1222 |
+
"use_label": 8268.525390625
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 0.68,
|
| 1226 |
+
"grad_norm": 2.421875,
|
| 1227 |
+
"learning_rate": 1.4006036925609245e-06,
|
| 1228 |
+
"logits/chosen": 11.211648941040039,
|
| 1229 |
+
"logits/rejected": 11.551493644714355,
|
| 1230 |
+
"logps/chosen": -14826.9326171875,
|
| 1231 |
+
"logps/rejected": -13029.5908203125,
|
| 1232 |
+
"loss": 0.3511,
|
| 1233 |
+
"pred_label": 4997.1748046875,
|
| 1234 |
+
"rewards/accuracies": 0.21875,
|
| 1235 |
+
"rewards/chosen": -147.53741455078125,
|
| 1236 |
+
"rewards/margins": -18.020130157470703,
|
| 1237 |
+
"rewards/rejected": -129.51730346679688,
|
| 1238 |
+
"step": 650,
|
| 1239 |
+
"use_label": 8348.8251953125
|
| 1240 |
},
|
| 1241 |
{
|
| 1242 |
"epoch": 0.69,
|
| 1243 |
+
"grad_norm": 0.244140625,
|
| 1244 |
+
"learning_rate": 1.3192409070404582e-06,
|
| 1245 |
+
"logits/chosen": 12.74785327911377,
|
| 1246 |
+
"logits/rejected": 12.4966459274292,
|
| 1247 |
+
"logps/chosen": -12700.5791015625,
|
| 1248 |
+
"logps/rejected": -10768.275390625,
|
| 1249 |
+
"loss": 0.3582,
|
| 1250 |
+
"pred_label": 5074.7001953125,
|
| 1251 |
+
"rewards/accuracies": 0.19374999403953552,
|
| 1252 |
+
"rewards/chosen": -126.42452239990234,
|
| 1253 |
+
"rewards/margins": -19.307645797729492,
|
| 1254 |
+
"rewards/rejected": -107.11688232421875,
|
| 1255 |
+
"step": 660,
|
| 1256 |
+
"use_label": 8431.2998046875
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.7,
|
| 1260 |
+
"grad_norm": 0.3125,
|
| 1261 |
+
"learning_rate": 1.2394572821496953e-06,
|
| 1262 |
+
"logits/chosen": 12.65557861328125,
|
| 1263 |
+
"logits/rejected": 12.626877784729004,
|
| 1264 |
+
"logps/chosen": -14430.9580078125,
|
| 1265 |
+
"logps/rejected": -14391.478515625,
|
| 1266 |
+
"loss": 0.3927,
|
| 1267 |
+
"pred_label": 5143.7998046875,
|
| 1268 |
+
"rewards/accuracies": 0.22499999403953552,
|
| 1269 |
+
"rewards/chosen": -143.65249633789062,
|
| 1270 |
+
"rewards/margins": -0.415884405374527,
|
| 1271 |
+
"rewards/rejected": -143.2366180419922,
|
| 1272 |
+
"step": 670,
|
| 1273 |
+
"use_label": 8522.2001953125
|
| 1274 |
},
|
| 1275 |
{
|
| 1276 |
"epoch": 0.71,
|
| 1277 |
+
"grad_norm": 0.107421875,
|
| 1278 |
+
"learning_rate": 1.1613595214152713e-06,
|
| 1279 |
+
"logits/chosen": 13.294036865234375,
|
| 1280 |
+
"logits/rejected": 12.978129386901855,
|
| 1281 |
+
"logps/chosen": -17393.599609375,
|
| 1282 |
+
"logps/rejected": -14762.783203125,
|
| 1283 |
+
"loss": 0.3712,
|
| 1284 |
+
"pred_label": 5223.0498046875,
|
| 1285 |
+
"rewards/accuracies": 0.20624999701976776,
|
| 1286 |
+
"rewards/chosen": -173.07546997070312,
|
| 1287 |
+
"rewards/margins": -26.263906478881836,
|
| 1288 |
+
"rewards/rejected": -146.8115692138672,
|
| 1289 |
+
"step": 680,
|
| 1290 |
+
"use_label": 8602.9501953125
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.72,
|
| 1294 |
+
"grad_norm": 0.0233154296875,
|
| 1295 |
+
"learning_rate": 1.0850520736699362e-06,
|
| 1296 |
+
"logits/chosen": 13.448896408081055,
|
| 1297 |
+
"logits/rejected": 13.269376754760742,
|
| 1298 |
+
"logps/chosen": -20676.513671875,
|
| 1299 |
+
"logps/rejected": -18685.75,
|
| 1300 |
+
"loss": 0.3327,
|
| 1301 |
+
"pred_label": 5306.0,
|
| 1302 |
+
"rewards/accuracies": 0.25,
|
| 1303 |
+
"rewards/chosen": -205.7480926513672,
|
| 1304 |
+
"rewards/margins": -19.850448608398438,
|
| 1305 |
+
"rewards/rejected": -185.8976287841797,
|
| 1306 |
+
"step": 690,
|
| 1307 |
+
"use_label": 8680.0
|
| 1308 |
},
|
| 1309 |
{
|
| 1310 |
"epoch": 0.73,
|
| 1311 |
+
"grad_norm": 0.0181884765625,
|
| 1312 |
+
"learning_rate": 1.0106369933615043e-06,
|
| 1313 |
+
"logits/chosen": 12.498316764831543,
|
| 1314 |
+
"logits/rejected": 12.834915161132812,
|
| 1315 |
+
"logps/chosen": -14553.083984375,
|
| 1316 |
+
"logps/rejected": -14335.8955078125,
|
| 1317 |
+
"loss": 0.3665,
|
| 1318 |
+
"pred_label": 5391.47509765625,
|
| 1319 |
+
"rewards/accuracies": 0.26875001192092896,
|
| 1320 |
+
"rewards/chosen": -144.8673553466797,
|
| 1321 |
+
"rewards/margins": -2.1933016777038574,
|
| 1322 |
+
"rewards/rejected": -142.67405700683594,
|
| 1323 |
+
"step": 700,
|
| 1324 |
+
"use_label": 8754.525390625
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 0.73,
|
| 1328 |
+
"eval_logits/chosen": 13.336652755737305,
|
| 1329 |
+
"eval_logits/rejected": 13.40242862701416,
|
| 1330 |
+
"eval_logps/chosen": -15087.66796875,
|
| 1331 |
+
"eval_logps/rejected": -14664.0517578125,
|
| 1332 |
+
"eval_loss": 0.3614233434200287,
|
| 1333 |
+
"eval_pred_label": 5543.5712890625,
|
| 1334 |
+
"eval_rewards/accuracies": 0.2420634925365448,
|
| 1335 |
+
"eval_rewards/chosen": -150.18765258789062,
|
| 1336 |
+
"eval_rewards/margins": -4.30112361907959,
|
| 1337 |
+
"eval_rewards/rejected": -145.88653564453125,
|
| 1338 |
+
"eval_runtime": 246.2358,
|
| 1339 |
+
"eval_samples_per_second": 8.122,
|
| 1340 |
+
"eval_steps_per_second": 0.256,
|
| 1341 |
+
"eval_use_label": 8936.4287109375,
|
| 1342 |
+
"step": 700
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 0.74,
|
| 1346 |
+
"grad_norm": 0.0030364990234375,
|
| 1347 |
+
"learning_rate": 9.382138040640714e-07,
|
| 1348 |
+
"logits/chosen": 12.521455764770508,
|
| 1349 |
+
"logits/rejected": 12.777295112609863,
|
| 1350 |
+
"logps/chosen": -13725.119140625,
|
| 1351 |
+
"logps/rejected": -12659.833984375,
|
| 1352 |
+
"loss": 0.3563,
|
| 1353 |
+
"pred_label": 5703.27490234375,
|
| 1354 |
+
"rewards/accuracies": 0.28125,
|
| 1355 |
+
"rewards/chosen": -136.62045288085938,
|
| 1356 |
+
"rewards/margins": -10.690929412841797,
|
| 1357 |
+
"rewards/rejected": -125.9295425415039,
|
| 1358 |
+
"step": 710,
|
| 1359 |
+
"use_label": 9106.724609375
|
| 1360 |
},
|
| 1361 |
{
|
| 1362 |
"epoch": 0.75,
|
| 1363 |
+
"grad_norm": 1.390625,
|
| 1364 |
+
"learning_rate": 8.678793653740633e-07,
|
| 1365 |
+
"logits/chosen": 12.614580154418945,
|
| 1366 |
+
"logits/rejected": 12.524827003479004,
|
| 1367 |
+
"logps/chosen": -12798.625,
|
| 1368 |
+
"logps/rejected": -13048.353515625,
|
| 1369 |
+
"loss": 0.3874,
|
| 1370 |
+
"pred_label": 5777.35009765625,
|
| 1371 |
+
"rewards/accuracies": 0.20000000298023224,
|
| 1372 |
+
"rewards/chosen": -127.41754150390625,
|
| 1373 |
+
"rewards/margins": 2.4097087383270264,
|
| 1374 |
+
"rewards/rejected": -129.82723999023438,
|
| 1375 |
+
"step": 720,
|
| 1376 |
+
"use_label": 9192.650390625
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 0.76,
|
| 1380 |
+
"grad_norm": 0.07421875,
|
| 1381 |
+
"learning_rate": 7.997277433690984e-07,
|
| 1382 |
+
"logits/chosen": 13.159255981445312,
|
| 1383 |
+
"logits/rejected": 12.944247245788574,
|
| 1384 |
+
"logps/chosen": -14291.3642578125,
|
| 1385 |
+
"logps/rejected": -13017.23046875,
|
| 1386 |
+
"loss": 0.3667,
|
| 1387 |
+
"pred_label": 5847.7001953125,
|
| 1388 |
+
"rewards/accuracies": 0.24375000596046448,
|
| 1389 |
+
"rewards/chosen": -142.21009826660156,
|
| 1390 |
+
"rewards/margins": -12.679841995239258,
|
| 1391 |
+
"rewards/rejected": -129.53024291992188,
|
| 1392 |
+
"step": 730,
|
| 1393 |
+
"use_label": 9282.2998046875
|
| 1394 |
},
|
| 1395 |
{
|
| 1396 |
"epoch": 0.77,
|
| 1397 |
+
"grad_norm": 0.000705718994140625,
|
| 1398 |
+
"learning_rate": 7.338500848029603e-07,
|
| 1399 |
+
"logits/chosen": 13.238718032836914,
|
| 1400 |
+
"logits/rejected": 13.197588920593262,
|
| 1401 |
+
"logps/chosen": -12959.798828125,
|
| 1402 |
+
"logps/rejected": -12114.708984375,
|
| 1403 |
+
"loss": 0.3837,
|
| 1404 |
+
"pred_label": 5919.2998046875,
|
| 1405 |
+
"rewards/accuracies": 0.24375000596046448,
|
| 1406 |
+
"rewards/chosen": -128.99172973632812,
|
| 1407 |
+
"rewards/margins": -8.478825569152832,
|
| 1408 |
+
"rewards/rejected": -120.5129165649414,
|
| 1409 |
+
"step": 740,
|
| 1410 |
+
"use_label": 9370.7001953125
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 0.79,
|
| 1414 |
+
"grad_norm": 0.05224609375,
|
| 1415 |
+
"learning_rate": 6.70334495204884e-07,
|
| 1416 |
+
"logits/chosen": 13.055346488952637,
|
| 1417 |
+
"logits/rejected": 13.03064250946045,
|
| 1418 |
+
"logps/chosen": -15817.4892578125,
|
| 1419 |
+
"logps/rejected": -16381.984375,
|
| 1420 |
+
"loss": 0.3837,
|
| 1421 |
+
"pred_label": 5993.9501953125,
|
| 1422 |
+
"rewards/accuracies": 0.29374998807907104,
|
| 1423 |
+
"rewards/chosen": -157.3995819091797,
|
| 1424 |
+
"rewards/margins": 5.577925682067871,
|
| 1425 |
+
"rewards/rejected": -162.97750854492188,
|
| 1426 |
+
"step": 750,
|
| 1427 |
+
"use_label": 9456.0498046875
|
| 1428 |
},
|
| 1429 |
{
|
| 1430 |
"epoch": 0.8,
|
| 1431 |
+
"grad_norm": 0.5,
|
| 1432 |
+
"learning_rate": 6.092659210462232e-07,
|
| 1433 |
+
"logits/chosen": 12.650552749633789,
|
| 1434 |
+
"logits/rejected": 12.711905479431152,
|
| 1435 |
+
"logps/chosen": -13931.3720703125,
|
| 1436 |
+
"logps/rejected": -14246.123046875,
|
| 1437 |
+
"loss": 0.4035,
|
| 1438 |
+
"pred_label": 6072.7998046875,
|
| 1439 |
+
"rewards/accuracies": 0.25,
|
| 1440 |
+
"rewards/chosen": -138.7740478515625,
|
| 1441 |
+
"rewards/margins": 2.966784715652466,
|
| 1442 |
+
"rewards/rejected": -141.74081420898438,
|
| 1443 |
+
"step": 760,
|
| 1444 |
+
"use_label": 9537.2001953125
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 0.81,
|
| 1448 |
+
"grad_norm": 3.075599670410156e-05,
|
| 1449 |
+
"learning_rate": 5.507260361320738e-07,
|
| 1450 |
+
"logits/chosen": 13.481063842773438,
|
| 1451 |
+
"logits/rejected": 13.060322761535645,
|
| 1452 |
+
"logps/chosen": -17337.056640625,
|
| 1453 |
+
"logps/rejected": -13485.697265625,
|
| 1454 |
+
"loss": 0.3337,
|
| 1455 |
+
"pred_label": 6150.3251953125,
|
| 1456 |
+
"rewards/accuracies": 0.20624999701976776,
|
| 1457 |
+
"rewards/chosen": -172.5351104736328,
|
| 1458 |
+
"rewards/margins": -38.41544723510742,
|
| 1459 |
+
"rewards/rejected": -134.11965942382812,
|
| 1460 |
+
"step": 770,
|
| 1461 |
+
"use_label": 9619.6748046875
|
| 1462 |
},
|
| 1463 |
{
|
| 1464 |
"epoch": 0.82,
|
| 1465 |
+
"grad_norm": 0.06982421875,
|
| 1466 |
+
"learning_rate": 4.947931323697983e-07,
|
| 1467 |
+
"logits/chosen": 13.052541732788086,
|
| 1468 |
+
"logits/rejected": 12.73249626159668,
|
| 1469 |
+
"logps/chosen": -16530.52734375,
|
| 1470 |
+
"logps/rejected": -14662.5234375,
|
| 1471 |
+
"loss": 0.3534,
|
| 1472 |
+
"pred_label": 6232.27490234375,
|
| 1473 |
+
"rewards/accuracies": 0.23749999701976776,
|
| 1474 |
+
"rewards/chosen": -164.57723999023438,
|
| 1475 |
+
"rewards/margins": -18.665462493896484,
|
| 1476 |
+
"rewards/rejected": -145.91177368164062,
|
| 1477 |
+
"step": 780,
|
| 1478 |
+
"use_label": 9697.724609375
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 0.83,
|
| 1482 |
+
"grad_norm": 0.1787109375,
|
| 1483 |
+
"learning_rate": 4.4154201506053985e-07,
|
| 1484 |
+
"logits/chosen": 12.90699291229248,
|
| 1485 |
+
"logits/rejected": 12.509411811828613,
|
| 1486 |
+
"logps/chosen": -15228.744140625,
|
| 1487 |
+
"logps/rejected": -11442.4423828125,
|
| 1488 |
+
"loss": 0.3586,
|
| 1489 |
+
"pred_label": 6308.5498046875,
|
| 1490 |
+
"rewards/accuracies": 0.13750000298023224,
|
| 1491 |
+
"rewards/chosen": -151.63909912109375,
|
| 1492 |
+
"rewards/margins": -37.736106872558594,
|
| 1493 |
+
"rewards/rejected": -113.90299987792969,
|
| 1494 |
+
"step": 790,
|
| 1495 |
+
"use_label": 9781.4501953125
|
| 1496 |
},
|
| 1497 |
{
|
| 1498 |
"epoch": 0.84,
|
| 1499 |
+
"grad_norm": 0.6796875,
|
| 1500 |
+
"learning_rate": 3.910439028537638e-07,
|
| 1501 |
+
"logits/chosen": 12.512937545776367,
|
| 1502 |
+
"logits/rejected": 12.64039421081543,
|
| 1503 |
+
"logps/chosen": -13050.3544921875,
|
| 1504 |
+
"logps/rejected": -13149.212890625,
|
| 1505 |
+
"loss": 0.3731,
|
| 1506 |
+
"pred_label": 6382.9248046875,
|
| 1507 |
+
"rewards/accuracies": 0.24375000596046448,
|
| 1508 |
+
"rewards/chosen": -129.903076171875,
|
| 1509 |
+
"rewards/margins": 0.9680694341659546,
|
| 1510 |
+
"rewards/rejected": -130.8711395263672,
|
| 1511 |
+
"step": 800,
|
| 1512 |
+
"use_label": 9867.0751953125
|
| 1513 |
},
|
| 1514 |
{
|
| 1515 |
"epoch": 0.84,
|
| 1516 |
+
"eval_logits/chosen": 13.469575881958008,
|
| 1517 |
+
"eval_logits/rejected": 13.533904075622559,
|
| 1518 |
+
"eval_logps/chosen": -15112.75390625,
|
| 1519 |
+
"eval_logps/rejected": -14688.42578125,
|
| 1520 |
+
"eval_loss": 0.3623265326023102,
|
| 1521 |
+
"eval_pred_label": 6533.66650390625,
|
| 1522 |
+
"eval_rewards/accuracies": 0.2400793582201004,
|
| 1523 |
+
"eval_rewards/chosen": -150.4385223388672,
|
| 1524 |
+
"eval_rewards/margins": -4.308249473571777,
|
| 1525 |
+
"eval_rewards/rejected": -146.13026428222656,
|
| 1526 |
+
"eval_runtime": 246.2624,
|
| 1527 |
+
"eval_samples_per_second": 8.121,
|
| 1528 |
+
"eval_steps_per_second": 0.256,
|
| 1529 |
+
"eval_use_label": 10050.3330078125,
|
| 1530 |
+
"step": 800
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 0.85,
|
| 1534 |
+
"grad_norm": 0.000614166259765625,
|
| 1535 |
+
"learning_rate": 3.4336633249862084e-07,
|
| 1536 |
+
"logits/chosen": 12.557435989379883,
|
| 1537 |
+
"logits/rejected": 12.592967987060547,
|
| 1538 |
+
"logps/chosen": -15382.724609375,
|
| 1539 |
+
"logps/rejected": -14971.3564453125,
|
| 1540 |
+
"loss": 0.3595,
|
| 1541 |
+
"pred_label": 6696.52490234375,
|
| 1542 |
+
"rewards/accuracies": 0.2562499940395355,
|
| 1543 |
+
"rewards/chosen": -153.12869262695312,
|
| 1544 |
+
"rewards/margins": -4.200928688049316,
|
| 1545 |
+
"rewards/rejected": -148.92776489257812,
|
| 1546 |
+
"step": 810,
|
| 1547 |
+
"use_label": 10217.474609375
|
| 1548 |
},
|
| 1549 |
{
|
| 1550 |
"epoch": 0.86,
|
| 1551 |
+
"grad_norm": 0.08154296875,
|
| 1552 |
+
"learning_rate": 2.98573068519539e-07,
|
| 1553 |
+
"logits/chosen": 12.341001510620117,
|
| 1554 |
+
"logits/rejected": 12.52668285369873,
|
| 1555 |
+
"logps/chosen": -11716.1484375,
|
| 1556 |
+
"logps/rejected": -11446.6171875,
|
| 1557 |
+
"loss": 0.3804,
|
| 1558 |
+
"pred_label": 6766.85009765625,
|
| 1559 |
+
"rewards/accuracies": 0.20000000298023224,
|
| 1560 |
+
"rewards/chosen": -116.56158447265625,
|
| 1561 |
+
"rewards/margins": -2.6313631534576416,
|
| 1562 |
+
"rewards/rejected": -113.93022155761719,
|
| 1563 |
+
"step": 820,
|
| 1564 |
+
"use_label": 10307.150390625
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 0.87,
|
| 1568 |
+
"grad_norm": 4.84375,
|
| 1569 |
+
"learning_rate": 2.5672401793681854e-07,
|
| 1570 |
+
"logits/chosen": 13.157217025756836,
|
| 1571 |
+
"logits/rejected": 13.266627311706543,
|
| 1572 |
+
"logps/chosen": -12381.7109375,
|
| 1573 |
+
"logps/rejected": -10725.419921875,
|
| 1574 |
+
"loss": 0.3736,
|
| 1575 |
+
"pred_label": 6836.5498046875,
|
| 1576 |
+
"rewards/accuracies": 0.1875,
|
| 1577 |
+
"rewards/chosen": -123.2366943359375,
|
| 1578 |
+
"rewards/margins": -16.548969268798828,
|
| 1579 |
+
"rewards/rejected": -106.68772888183594,
|
| 1580 |
+
"step": 830,
|
| 1581 |
+
"use_label": 10397.4501953125
|
| 1582 |
},
|
| 1583 |
{
|
| 1584 |
"epoch": 0.88,
|
| 1585 |
+
"grad_norm": 0.046875,
|
| 1586 |
+
"learning_rate": 2.178751501463036e-07,
|
| 1587 |
+
"logits/chosen": 12.293255805969238,
|
| 1588 |
+
"logits/rejected": 11.906517028808594,
|
| 1589 |
+
"logps/chosen": -13232.6845703125,
|
| 1590 |
+
"logps/rejected": -11126.3134765625,
|
| 1591 |
+
"loss": 0.3766,
|
| 1592 |
+
"pred_label": 6907.35009765625,
|
| 1593 |
+
"rewards/accuracies": 0.18125000596046448,
|
| 1594 |
+
"rewards/chosen": -131.73385620117188,
|
| 1595 |
+
"rewards/margins": -21.00704574584961,
|
| 1596 |
+
"rewards/rejected": -110.72682189941406,
|
| 1597 |
+
"step": 840,
|
| 1598 |
+
"use_label": 10486.650390625
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 0.89,
|
| 1602 |
+
"grad_norm": 0.0089111328125,
|
| 1603 |
+
"learning_rate": 1.820784220652766e-07,
|
| 1604 |
+
"logits/chosen": 13.318339347839355,
|
| 1605 |
+
"logits/rejected": 13.22724437713623,
|
| 1606 |
+
"logps/chosen": -16406.490234375,
|
| 1607 |
+
"logps/rejected": -16654.296875,
|
| 1608 |
+
"loss": 0.3544,
|
| 1609 |
+
"pred_label": 6977.77490234375,
|
| 1610 |
+
"rewards/accuracies": 0.3125,
|
| 1611 |
+
"rewards/chosen": -163.22271728515625,
|
| 1612 |
+
"rewards/margins": 2.4889702796936035,
|
| 1613 |
+
"rewards/rejected": -165.71170043945312,
|
| 1614 |
+
"step": 850,
|
| 1615 |
+
"use_label": 10576.224609375
|
| 1616 |
},
|
| 1617 |
{
|
| 1618 |
"epoch": 0.9,
|
| 1619 |
+
"grad_norm": 0.2734375,
|
| 1620 |
+
"learning_rate": 1.4938170864468636e-07,
|
| 1621 |
+
"logits/chosen": 13.206477165222168,
|
| 1622 |
+
"logits/rejected": 13.062841415405273,
|
| 1623 |
+
"logps/chosen": -17820.337890625,
|
| 1624 |
+
"logps/rejected": -14261.5107421875,
|
| 1625 |
+
"loss": 0.358,
|
| 1626 |
+
"pred_label": 7060.39990234375,
|
| 1627 |
+
"rewards/accuracies": 0.21250000596046448,
|
| 1628 |
+
"rewards/chosen": -177.391357421875,
|
| 1629 |
+
"rewards/margins": -35.53297805786133,
|
| 1630 |
+
"rewards/rejected": -141.85836791992188,
|
| 1631 |
+
"step": 860,
|
| 1632 |
+
"use_label": 10653.599609375
|
| 1633 |
+
},
|
| 1634 |
+
{
|
| 1635 |
+
"epoch": 0.91,
|
| 1636 |
+
"grad_norm": 1.1953125,
|
| 1637 |
+
"learning_rate": 1.1982873884064466e-07,
|
| 1638 |
+
"logits/chosen": 12.969378471374512,
|
| 1639 |
+
"logits/rejected": 12.896804809570312,
|
| 1640 |
+
"logps/chosen": -16513.146484375,
|
| 1641 |
+
"logps/rejected": -15408.953125,
|
| 1642 |
+
"loss": 0.3415,
|
| 1643 |
+
"pred_label": 7140.97509765625,
|
| 1644 |
+
"rewards/accuracies": 0.23125000298023224,
|
| 1645 |
+
"rewards/chosen": -164.31832885742188,
|
| 1646 |
+
"rewards/margins": -11.052000999450684,
|
| 1647 |
+
"rewards/rejected": -153.26632690429688,
|
| 1648 |
+
"step": 870,
|
| 1649 |
+
"use_label": 10733.025390625
|
| 1650 |
},
|
| 1651 |
{
|
| 1652 |
"epoch": 0.92,
|
| 1653 |
+
"grad_norm": 0.009033203125,
|
| 1654 |
+
"learning_rate": 9.345903713082305e-08,
|
| 1655 |
+
"logits/chosen": 13.240982055664062,
|
| 1656 |
+
"logits/rejected": 13.342303276062012,
|
| 1657 |
+
"logps/chosen": -13444.123046875,
|
| 1658 |
+
"logps/rejected": -15121.9482421875,
|
| 1659 |
+
"loss": 0.3616,
|
| 1660 |
+
"pred_label": 7224.9501953125,
|
| 1661 |
+
"rewards/accuracies": 0.25,
|
| 1662 |
+
"rewards/chosen": -133.81370544433594,
|
| 1663 |
+
"rewards/margins": 16.65894317626953,
|
| 1664 |
+
"rewards/rejected": -150.47265625,
|
| 1665 |
+
"step": 880,
|
| 1666 |
+
"use_label": 10809.0498046875
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 0.93,
|
| 1670 |
+
"grad_norm": 0.083984375,
|
| 1671 |
+
"learning_rate": 7.030787065396866e-08,
|
| 1672 |
+
"logits/chosen": 12.833715438842773,
|
| 1673 |
+
"logits/rejected": 12.859628677368164,
|
| 1674 |
+
"logps/chosen": -13761.8701171875,
|
| 1675 |
+
"logps/rejected": -14406.4794921875,
|
| 1676 |
+
"loss": 0.3679,
|
| 1677 |
+
"pred_label": 7291.89990234375,
|
| 1678 |
+
"rewards/accuracies": 0.23749999701976776,
|
| 1679 |
+
"rewards/chosen": -136.9804229736328,
|
| 1680 |
+
"rewards/margins": 6.350142955780029,
|
| 1681 |
+
"rewards/rejected": -143.33055114746094,
|
| 1682 |
+
"step": 890,
|
| 1683 |
+
"use_label": 10902.099609375
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 0.94,
|
| 1687 |
+
"grad_norm": 0.00021266937255859375,
|
| 1688 |
+
"learning_rate": 5.0406202043228604e-08,
|
| 1689 |
+
"logits/chosen": 13.16772174835205,
|
| 1690 |
+
"logits/rejected": 13.124513626098633,
|
| 1691 |
+
"logps/chosen": -14781.494140625,
|
| 1692 |
+
"logps/rejected": -17149.380859375,
|
| 1693 |
+
"loss": 0.3696,
|
| 1694 |
+
"pred_label": 7367.4248046875,
|
| 1695 |
"rewards/accuracies": 0.3125,
|
| 1696 |
+
"rewards/chosen": -147.12081909179688,
|
| 1697 |
+
"rewards/margins": 23.48788070678711,
|
| 1698 |
+
"rewards/rejected": -170.60870361328125,
|
| 1699 |
+
"step": 900,
|
| 1700 |
+
"use_label": 10986.5751953125
|
| 1701 |
},
|
| 1702 |
{
|
| 1703 |
"epoch": 0.94,
|
| 1704 |
+
"eval_logits/chosen": 13.471565246582031,
|
| 1705 |
+
"eval_logits/rejected": 13.536233901977539,
|
| 1706 |
+
"eval_logps/chosen": -15130.16796875,
|
| 1707 |
+
"eval_logps/rejected": -14705.8974609375,
|
| 1708 |
+
"eval_loss": 0.3624568283557892,
|
| 1709 |
+
"eval_pred_label": 7522.01611328125,
|
| 1710 |
+
"eval_rewards/accuracies": 0.2420634925365448,
|
| 1711 |
+
"eval_rewards/chosen": -150.61268615722656,
|
| 1712 |
+
"eval_rewards/margins": -4.307671546936035,
|
| 1713 |
+
"eval_rewards/rejected": -146.30499267578125,
|
| 1714 |
+
"eval_runtime": 246.1349,
|
| 1715 |
+
"eval_samples_per_second": 8.126,
|
| 1716 |
+
"eval_steps_per_second": 0.256,
|
| 1717 |
+
"eval_use_label": 11165.984375,
|
| 1718 |
+
"step": 900
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 0.95,
|
| 1722 |
+
"grad_norm": 0.06787109375,
|
| 1723 |
+
"learning_rate": 3.378064801637687e-08,
|
| 1724 |
+
"logits/chosen": 12.845532417297363,
|
| 1725 |
+
"logits/rejected": 12.847734451293945,
|
| 1726 |
+
"logps/chosen": -12279.849609375,
|
| 1727 |
+
"logps/rejected": -11353.5927734375,
|
| 1728 |
+
"loss": 0.368,
|
| 1729 |
+
"pred_label": 7677.27490234375,
|
| 1730 |
+
"rewards/accuracies": 0.19374999403953552,
|
| 1731 |
+
"rewards/chosen": -122.19189453125,
|
| 1732 |
+
"rewards/margins": -9.292089462280273,
|
| 1733 |
+
"rewards/rejected": -112.89981842041016,
|
| 1734 |
+
"step": 910,
|
| 1735 |
+
"use_label": 11340.724609375
|
| 1736 |
},
|
| 1737 |
{
|
| 1738 |
"epoch": 0.96,
|
| 1739 |
+
"grad_norm": 0.78125,
|
| 1740 |
+
"learning_rate": 2.0453443778310766e-08,
|
| 1741 |
+
"logits/chosen": 12.938204765319824,
|
| 1742 |
+
"logits/rejected": 13.116650581359863,
|
| 1743 |
+
"logps/chosen": -12339.7646484375,
|
| 1744 |
+
"logps/rejected": -13293.265625,
|
| 1745 |
+
"loss": 0.3638,
|
| 1746 |
+
"pred_label": 7745.25,
|
| 1747 |
+
"rewards/accuracies": 0.23125000298023224,
|
| 1748 |
+
"rewards/chosen": -122.85809326171875,
|
| 1749 |
+
"rewards/margins": 9.360071182250977,
|
| 1750 |
+
"rewards/rejected": -132.21817016601562,
|
| 1751 |
+
"step": 920,
|
| 1752 |
+
"use_label": 11432.75
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 0.97,
|
| 1756 |
+
"grad_norm": 0.0703125,
|
| 1757 |
+
"learning_rate": 1.0442413283435759e-08,
|
| 1758 |
+
"logits/chosen": 12.815765380859375,
|
| 1759 |
+
"logits/rejected": 12.618230819702148,
|
| 1760 |
+
"logps/chosen": -13530.013671875,
|
| 1761 |
+
"logps/rejected": -13645.0625,
|
| 1762 |
+
"loss": 0.3848,
|
| 1763 |
+
"pred_label": 7812.75,
|
| 1764 |
+
"rewards/accuracies": 0.22499999403953552,
|
| 1765 |
+
"rewards/chosen": -134.66860961914062,
|
| 1766 |
+
"rewards/margins": 1.088117003440857,
|
| 1767 |
+
"rewards/rejected": -135.75672912597656,
|
| 1768 |
+
"step": 930,
|
| 1769 |
+
"use_label": 11525.25
|
| 1770 |
},
|
| 1771 |
{
|
| 1772 |
"epoch": 0.98,
|
| 1773 |
+
"grad_norm": 0.263671875,
|
| 1774 |
+
"learning_rate": 3.760945397705828e-09,
|
| 1775 |
+
"logits/chosen": 13.209531784057617,
|
| 1776 |
+
"logits/rejected": 13.244476318359375,
|
| 1777 |
+
"logps/chosen": -14691.748046875,
|
| 1778 |
+
"logps/rejected": -15939.892578125,
|
| 1779 |
+
"loss": 0.3695,
|
| 1780 |
+
"pred_label": 7884.9248046875,
|
| 1781 |
+
"rewards/accuracies": 0.28125,
|
| 1782 |
+
"rewards/chosen": -146.2279815673828,
|
| 1783 |
+
"rewards/margins": 12.3941068649292,
|
| 1784 |
+
"rewards/rejected": -158.62208557128906,
|
| 1785 |
+
"step": 940,
|
| 1786 |
+
"use_label": 11613.0751953125
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 0.99,
|
| 1790 |
+
"grad_norm": 0.09033203125,
|
| 1791 |
+
"learning_rate": 4.1797599220405605e-10,
|
| 1792 |
+
"logits/chosen": 12.960329055786133,
|
| 1793 |
+
"logits/rejected": 12.965344429016113,
|
| 1794 |
+
"logps/chosen": -17210.09765625,
|
| 1795 |
+
"logps/rejected": -17094.09765625,
|
| 1796 |
+
"loss": 0.3572,
|
| 1797 |
+
"pred_label": 7954.3251953125,
|
| 1798 |
+
"rewards/accuracies": 0.22499999403953552,
|
| 1799 |
+
"rewards/chosen": -171.33486938476562,
|
| 1800 |
+
"rewards/margins": -1.184854507446289,
|
| 1801 |
+
"rewards/rejected": -170.14999389648438,
|
| 1802 |
+
"step": 950,
|
| 1803 |
+
"use_label": 11703.6748046875
|
| 1804 |
},
|
| 1805 |
{
|
| 1806 |
"epoch": 1.0,
|
| 1807 |
+
"step": 955,
|
| 1808 |
"total_flos": 0.0,
|
| 1809 |
+
"train_loss": 0.4203558097959189,
|
| 1810 |
+
"train_runtime": 19955.1733,
|
| 1811 |
+
"train_samples_per_second": 3.064,
|
| 1812 |
+
"train_steps_per_second": 0.048
|
| 1813 |
}
|
| 1814 |
],
|
| 1815 |
"logging_steps": 10,
|
| 1816 |
+
"max_steps": 955,
|
| 1817 |
"num_input_tokens_seen": 0,
|
| 1818 |
"num_train_epochs": 1,
|
| 1819 |
"save_steps": 50,
|