Model save
Browse files- README.md +17 -17
- adapter_config.json +2 -2
- adapter_model.safetensors +1 -1
- all_results.json +18 -18
- eval_results.json +14 -14
- train_results.json +4 -4
- trainer_state.json +1499 -123
- training_args.bin +1 -1
README.md
CHANGED
|
@@ -15,17 +15,17 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 15 |
|
| 16 |
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the None dataset.
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
-
- Loss: 0.
|
| 19 |
-
- Rewards/chosen: -
|
| 20 |
-
- Rewards/rejected: -
|
| 21 |
-
- Rewards/accuracies: 0.
|
| 22 |
-
- Rewards/margins:
|
| 23 |
-
- Logps/rejected: -
|
| 24 |
-
- Logps/chosen: -
|
| 25 |
-
- Logits/rejected:
|
| 26 |
-
- Logits/chosen:
|
| 27 |
-
- Use Label:
|
| 28 |
-
- Pred Label:
|
| 29 |
|
| 30 |
## Model description
|
| 31 |
|
|
@@ -49,10 +49,10 @@ The following hyperparameters were used during training:
|
|
| 49 |
- eval_batch_size: 4
|
| 50 |
- seed: 42
|
| 51 |
- distributed_type: multi-GPU
|
| 52 |
-
- num_devices:
|
| 53 |
-
- gradient_accumulation_steps:
|
| 54 |
- total_train_batch_size: 64
|
| 55 |
-
- total_eval_batch_size:
|
| 56 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 57 |
- lr_scheduler_type: linear
|
| 58 |
- lr_scheduler_warmup_ratio: 0.1
|
|
@@ -60,9 +60,9 @@ The following hyperparameters were used during training:
|
|
| 60 |
|
| 61 |
### Training results
|
| 62 |
|
| 63 |
-
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen | Use Label
|
| 64 |
-
|
| 65 |
-
| 0.
|
| 66 |
|
| 67 |
|
| 68 |
### Framework versions
|
|
|
|
| 15 |
|
| 16 |
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the None dataset.
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
+
- Loss: 0.6688
|
| 19 |
+
- Rewards/chosen: -0.0034
|
| 20 |
+
- Rewards/rejected: -0.6444
|
| 21 |
+
- Rewards/accuracies: 0.7210
|
| 22 |
+
- Rewards/margins: 0.6410
|
| 23 |
+
- Logps/rejected: -268.7319
|
| 24 |
+
- Logps/chosen: -282.0097
|
| 25 |
+
- Logits/rejected: -2.8242
|
| 26 |
+
- Logits/chosen: -2.8346
|
| 27 |
+
- Use Label: 21379.7969
|
| 28 |
+
- Pred Label: 10682.2041
|
| 29 |
|
| 30 |
## Model description
|
| 31 |
|
|
|
|
| 49 |
- eval_batch_size: 4
|
| 50 |
- seed: 42
|
| 51 |
- distributed_type: multi-GPU
|
| 52 |
+
- num_devices: 2
|
| 53 |
+
- gradient_accumulation_steps: 8
|
| 54 |
- total_train_batch_size: 64
|
| 55 |
+
- total_eval_batch_size: 8
|
| 56 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 57 |
- lr_scheduler_type: linear
|
| 58 |
- lr_scheduler_warmup_ratio: 0.1
|
|
|
|
| 60 |
|
| 61 |
### Training results
|
| 62 |
|
| 63 |
+
| 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 |
|
| 64 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|:----------:|:----------:|
|
| 65 |
+
| 0.6708 | 1.0 | 955 | 0.6688 | -0.0034 | -0.6444 | 0.7210 | 0.6410 | -268.7319 | -282.0097 | -2.8242 | -2.8346 | 20769.7969 | 10292.2041 |
|
| 66 |
|
| 67 |
|
| 68 |
### Framework versions
|
adapter_config.json
CHANGED
|
@@ -16,9 +16,9 @@
|
|
| 16 |
"rank_pattern": {},
|
| 17 |
"revision": null,
|
| 18 |
"target_modules": [
|
| 19 |
-
"v_proj",
|
| 20 |
-
"q_proj",
|
| 21 |
"k_proj",
|
|
|
|
|
|
|
| 22 |
"o_proj"
|
| 23 |
],
|
| 24 |
"task_type": "CAUSAL_LM"
|
|
|
|
| 16 |
"rank_pattern": {},
|
| 17 |
"revision": null,
|
| 18 |
"target_modules": [
|
|
|
|
|
|
|
| 19 |
"k_proj",
|
| 20 |
+
"q_proj",
|
| 21 |
+
"v_proj",
|
| 22 |
"o_proj"
|
| 23 |
],
|
| 24 |
"task_type": "CAUSAL_LM"
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 218138576
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f58fe82b34998250bf2df2e040ef6472804071478f8eb7dacddfd8fae011c2a3
|
| 3 |
size 218138576
|
all_results.json
CHANGED
|
@@ -1,23 +1,23 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
-
"eval_logits/chosen":
|
| 4 |
-
"eval_logits/rejected":
|
| 5 |
-
"eval_logps/chosen": -
|
| 6 |
-
"eval_logps/rejected": -
|
| 7 |
-
"eval_loss": 0.
|
| 8 |
-
"eval_pred_label":
|
| 9 |
-
"eval_rewards/accuracies": 0.
|
| 10 |
-
"eval_rewards/chosen": -
|
| 11 |
-
"eval_rewards/margins":
|
| 12 |
-
"eval_rewards/rejected": -
|
| 13 |
-
"eval_runtime":
|
| 14 |
"eval_samples": 2000,
|
| 15 |
-
"eval_samples_per_second":
|
| 16 |
-
"eval_steps_per_second": 0.
|
| 17 |
-
"eval_use_label":
|
| 18 |
-
"train_loss": 0.
|
| 19 |
-
"train_runtime":
|
| 20 |
"train_samples": 61135,
|
| 21 |
-
"train_samples_per_second":
|
| 22 |
-
"train_steps_per_second": 0.
|
| 23 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"eval_logits/chosen": -2.83455753326416,
|
| 4 |
+
"eval_logits/rejected": -2.824216604232788,
|
| 5 |
+
"eval_logps/chosen": -282.00970458984375,
|
| 6 |
+
"eval_logps/rejected": -268.7319030761719,
|
| 7 |
+
"eval_loss": 0.6687781810760498,
|
| 8 |
+
"eval_pred_label": 10682.2041015625,
|
| 9 |
+
"eval_rewards/accuracies": 0.7210000157356262,
|
| 10 |
+
"eval_rewards/chosen": -0.0034403554163873196,
|
| 11 |
+
"eval_rewards/margins": 0.6410075426101685,
|
| 12 |
+
"eval_rewards/rejected": -0.6444479823112488,
|
| 13 |
+
"eval_runtime": 855.5169,
|
| 14 |
"eval_samples": 2000,
|
| 15 |
+
"eval_samples_per_second": 2.338,
|
| 16 |
+
"eval_steps_per_second": 0.292,
|
| 17 |
+
"eval_use_label": 21379.796875,
|
| 18 |
+
"train_loss": 0.6626948830969046,
|
| 19 |
+
"train_runtime": 47570.4937,
|
| 20 |
"train_samples": 61135,
|
| 21 |
+
"train_samples_per_second": 1.285,
|
| 22 |
+
"train_steps_per_second": 0.02
|
| 23 |
}
|
eval_results.json
CHANGED
|
@@ -1,18 +1,18 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
-
"eval_logits/chosen":
|
| 4 |
-
"eval_logits/rejected":
|
| 5 |
-
"eval_logps/chosen": -
|
| 6 |
-
"eval_logps/rejected": -
|
| 7 |
-
"eval_loss": 0.
|
| 8 |
-
"eval_pred_label":
|
| 9 |
-
"eval_rewards/accuracies": 0.
|
| 10 |
-
"eval_rewards/chosen": -
|
| 11 |
-
"eval_rewards/margins":
|
| 12 |
-
"eval_rewards/rejected": -
|
| 13 |
-
"eval_runtime":
|
| 14 |
"eval_samples": 2000,
|
| 15 |
-
"eval_samples_per_second":
|
| 16 |
-
"eval_steps_per_second": 0.
|
| 17 |
-
"eval_use_label":
|
| 18 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"eval_logits/chosen": -2.83455753326416,
|
| 4 |
+
"eval_logits/rejected": -2.824216604232788,
|
| 5 |
+
"eval_logps/chosen": -282.00970458984375,
|
| 6 |
+
"eval_logps/rejected": -268.7319030761719,
|
| 7 |
+
"eval_loss": 0.6687781810760498,
|
| 8 |
+
"eval_pred_label": 10682.2041015625,
|
| 9 |
+
"eval_rewards/accuracies": 0.7210000157356262,
|
| 10 |
+
"eval_rewards/chosen": -0.0034403554163873196,
|
| 11 |
+
"eval_rewards/margins": 0.6410075426101685,
|
| 12 |
+
"eval_rewards/rejected": -0.6444479823112488,
|
| 13 |
+
"eval_runtime": 855.5169,
|
| 14 |
"eval_samples": 2000,
|
| 15 |
+
"eval_samples_per_second": 2.338,
|
| 16 |
+
"eval_steps_per_second": 0.292,
|
| 17 |
+
"eval_use_label": 21379.796875
|
| 18 |
}
|
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": 0.6626948830969046,
|
| 4 |
+
"train_runtime": 47570.4937,
|
| 5 |
"train_samples": 61135,
|
| 6 |
+
"train_samples_per_second": 1.285,
|
| 7 |
+
"train_steps_per_second": 0.02
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -11,10 +11,10 @@
|
|
| 11 |
{
|
| 12 |
"epoch": 0.0,
|
| 13 |
"learning_rate": 1.0416666666666667e-07,
|
| 14 |
-
"logits/chosen": -2.
|
| 15 |
-
"logits/rejected": -2.
|
| 16 |
-
"logps/chosen": -
|
| 17 |
-
"logps/rejected": -
|
| 18 |
"loss": 0.6931,
|
| 19 |
"pred_label": 0.0,
|
| 20 |
"rewards/accuracies": 0.0,
|
|
@@ -22,181 +22,1557 @@
|
|
| 22 |
"rewards/margins": 0.0,
|
| 23 |
"rewards/rejected": 0.0,
|
| 24 |
"step": 1,
|
| 25 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
},
|
| 27 |
{
|
| 28 |
"epoch": 0.1,
|
| 29 |
"learning_rate": 9.953434225844005e-06,
|
| 30 |
-
"logits/chosen": -2.
|
| 31 |
-
"logits/rejected": -2.
|
| 32 |
-
"logps/chosen": -
|
| 33 |
-
"logps/rejected": -
|
| 34 |
-
"loss": 0.
|
| 35 |
-
"pred_label":
|
| 36 |
-
"rewards/accuracies": 0.
|
| 37 |
-
"rewards/chosen":
|
| 38 |
-
"rewards/margins": 0.
|
| 39 |
-
"rewards/rejected": -0.
|
| 40 |
"step": 100,
|
| 41 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 42 |
},
|
| 43 |
{
|
| 44 |
"epoch": 0.21,
|
| 45 |
"learning_rate": 8.789289871944122e-06,
|
| 46 |
-
"logits/chosen": -2.
|
| 47 |
-
"logits/rejected": -2.
|
| 48 |
-
"logps/chosen": -
|
| 49 |
-
"logps/rejected": -
|
| 50 |
-
"loss": 0.
|
| 51 |
-
"pred_label":
|
| 52 |
-
"rewards/accuracies": 0.
|
| 53 |
-
"rewards/chosen":
|
| 54 |
-
"rewards/margins":
|
| 55 |
-
"rewards/rejected": -
|
| 56 |
"step": 200,
|
| 57 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 58 |
},
|
| 59 |
{
|
| 60 |
"epoch": 0.31,
|
| 61 |
"learning_rate": 7.625145518044238e-06,
|
| 62 |
-
"logits/chosen": -2.
|
| 63 |
-
"logits/rejected": -2.
|
| 64 |
-
"logps/chosen": -
|
| 65 |
-
"logps/rejected": -
|
| 66 |
-
"loss": 0.
|
| 67 |
-
"pred_label":
|
| 68 |
-
"rewards/accuracies": 0.
|
| 69 |
-
"rewards/chosen": -
|
| 70 |
-
"rewards/margins":
|
| 71 |
-
"rewards/rejected": -
|
| 72 |
"step": 300,
|
| 73 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 0.42,
|
| 77 |
"learning_rate": 6.461001164144355e-06,
|
| 78 |
-
"logits/chosen": -
|
| 79 |
-
"logits/rejected": -
|
| 80 |
-
"logps/chosen": -
|
| 81 |
-
"logps/rejected": -
|
| 82 |
-
"loss": 0.
|
| 83 |
-
"pred_label":
|
| 84 |
-
"rewards/accuracies": 0.
|
| 85 |
-
"rewards/chosen":
|
| 86 |
-
"rewards/margins":
|
| 87 |
-
"rewards/rejected": -
|
| 88 |
"step": 400,
|
| 89 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 90 |
},
|
| 91 |
{
|
| 92 |
"epoch": 0.52,
|
| 93 |
"learning_rate": 5.2968568102444705e-06,
|
| 94 |
-
"logits/chosen": -
|
| 95 |
-
"logits/rejected": -
|
| 96 |
-
"logps/chosen": -
|
| 97 |
-
"logps/rejected": -
|
| 98 |
-
"loss": 0.
|
| 99 |
-
"pred_label":
|
| 100 |
-
"rewards/accuracies": 0.
|
| 101 |
-
"rewards/chosen":
|
| 102 |
-
"rewards/margins":
|
| 103 |
-
"rewards/rejected": -
|
| 104 |
"step": 500,
|
| 105 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 0.63,
|
| 109 |
"learning_rate": 4.132712456344587e-06,
|
| 110 |
-
"logits/chosen": -
|
| 111 |
-
"logits/rejected":
|
| 112 |
-
"logps/chosen": -
|
| 113 |
-
"logps/rejected": -
|
| 114 |
-
"loss": 0.
|
| 115 |
-
"pred_label":
|
| 116 |
-
"rewards/accuracies": 0.
|
| 117 |
-
"rewards/chosen":
|
| 118 |
-
"rewards/margins":
|
| 119 |
-
"rewards/rejected": -
|
| 120 |
"step": 600,
|
| 121 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 122 |
},
|
| 123 |
{
|
| 124 |
"epoch": 0.73,
|
| 125 |
"learning_rate": 2.9685681024447033e-06,
|
| 126 |
-
"logits/chosen":
|
| 127 |
-
"logits/rejected":
|
| 128 |
-
"logps/chosen": -
|
| 129 |
-
"logps/rejected": -
|
| 130 |
-
"loss": 0.
|
| 131 |
-
"pred_label":
|
| 132 |
-
"rewards/accuracies": 0.
|
| 133 |
-
"rewards/chosen":
|
| 134 |
-
"rewards/margins":
|
| 135 |
-
"rewards/rejected": -
|
| 136 |
"step": 700,
|
| 137 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 138 |
},
|
| 139 |
{
|
| 140 |
"epoch": 0.84,
|
| 141 |
"learning_rate": 1.8044237485448196e-06,
|
| 142 |
-
"logits/chosen":
|
| 143 |
-
"logits/rejected":
|
| 144 |
-
"logps/chosen": -
|
| 145 |
-
"logps/rejected": -
|
| 146 |
-
"loss": 0.
|
| 147 |
-
"pred_label":
|
| 148 |
-
"rewards/accuracies": 0.
|
| 149 |
-
"rewards/chosen":
|
| 150 |
-
"rewards/margins":
|
| 151 |
-
"rewards/rejected": -
|
| 152 |
"step": 800,
|
| 153 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 154 |
},
|
| 155 |
{
|
| 156 |
"epoch": 0.94,
|
| 157 |
"learning_rate": 6.402793946449361e-07,
|
| 158 |
-
"logits/chosen":
|
| 159 |
-
"logits/rejected":
|
| 160 |
-
"logps/chosen": -
|
| 161 |
-
"logps/rejected": -
|
| 162 |
-
"loss": 0.
|
| 163 |
-
"pred_label":
|
| 164 |
-
"rewards/accuracies": 0.
|
| 165 |
-
"rewards/chosen":
|
| 166 |
-
"rewards/margins":
|
| 167 |
-
"rewards/rejected": -
|
| 168 |
"step": 900,
|
| 169 |
-
"use_label":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 170 |
},
|
| 171 |
{
|
| 172 |
"epoch": 1.0,
|
| 173 |
-
"eval_logits/chosen":
|
| 174 |
-
"eval_logits/rejected":
|
| 175 |
-
"eval_logps/chosen": -
|
| 176 |
-
"eval_logps/rejected": -
|
| 177 |
-
"eval_loss": 0.
|
| 178 |
-
"eval_pred_label":
|
| 179 |
-
"eval_rewards/accuracies": 0.
|
| 180 |
-
"eval_rewards/chosen": -
|
| 181 |
-
"eval_rewards/margins":
|
| 182 |
-
"eval_rewards/rejected": -
|
| 183 |
-
"eval_runtime":
|
| 184 |
-
"eval_samples_per_second":
|
| 185 |
-
"eval_steps_per_second": 0.
|
| 186 |
-
"eval_use_label":
|
| 187 |
"step": 955
|
| 188 |
},
|
| 189 |
{
|
| 190 |
"epoch": 1.0,
|
| 191 |
"step": 955,
|
| 192 |
"total_flos": 0.0,
|
| 193 |
-
"train_loss": 0.
|
| 194 |
-
"train_runtime":
|
| 195 |
-
"train_samples_per_second":
|
| 196 |
-
"train_steps_per_second": 0.
|
| 197 |
}
|
| 198 |
],
|
| 199 |
-
"logging_steps":
|
| 200 |
"max_steps": 955,
|
| 201 |
"num_train_epochs": 1,
|
| 202 |
"save_steps": 50,
|
|
|
|
| 11 |
{
|
| 12 |
"epoch": 0.0,
|
| 13 |
"learning_rate": 1.0416666666666667e-07,
|
| 14 |
+
"logits/chosen": -2.899709463119507,
|
| 15 |
+
"logits/rejected": -2.879509687423706,
|
| 16 |
+
"logps/chosen": -314.8815612792969,
|
| 17 |
+
"logps/rejected": -239.785888671875,
|
| 18 |
"loss": 0.6931,
|
| 19 |
"pred_label": 0.0,
|
| 20 |
"rewards/accuracies": 0.0,
|
|
|
|
| 22 |
"rewards/margins": 0.0,
|
| 23 |
"rewards/rejected": 0.0,
|
| 24 |
"step": 1,
|
| 25 |
+
"use_label": 18.0
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 0.01,
|
| 29 |
+
"learning_rate": 1.0416666666666667e-06,
|
| 30 |
+
"logits/chosen": -2.871338129043579,
|
| 31 |
+
"logits/rejected": -2.8671977519989014,
|
| 32 |
+
"logps/chosen": -304.6894226074219,
|
| 33 |
+
"logps/rejected": -284.7349853515625,
|
| 34 |
+
"loss": 0.6939,
|
| 35 |
+
"pred_label": 0.0,
|
| 36 |
+
"rewards/accuracies": 0.4131944477558136,
|
| 37 |
+
"rewards/chosen": 0.0006341927801258862,
|
| 38 |
+
"rewards/margins": -0.0011922286357730627,
|
| 39 |
+
"rewards/rejected": 0.0018264217069372535,
|
| 40 |
+
"step": 10,
|
| 41 |
+
"use_label": 178.0
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 0.02,
|
| 45 |
+
"learning_rate": 2.0833333333333334e-06,
|
| 46 |
+
"logits/chosen": -2.761018991470337,
|
| 47 |
+
"logits/rejected": -2.7763421535491943,
|
| 48 |
+
"logps/chosen": -255.852783203125,
|
| 49 |
+
"logps/rejected": -260.04364013671875,
|
| 50 |
+
"loss": 0.6902,
|
| 51 |
+
"pred_label": 0.0,
|
| 52 |
+
"rewards/accuracies": 0.5531250238418579,
|
| 53 |
+
"rewards/chosen": 0.0032608681358397007,
|
| 54 |
+
"rewards/margins": 0.006093679927289486,
|
| 55 |
+
"rewards/rejected": -0.002832812489941716,
|
| 56 |
+
"step": 20,
|
| 57 |
+
"use_label": 482.0
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.03,
|
| 61 |
+
"learning_rate": 3.125e-06,
|
| 62 |
+
"logits/chosen": -2.85542631149292,
|
| 63 |
+
"logits/rejected": -2.8387763500213623,
|
| 64 |
+
"logps/chosen": -274.9126892089844,
|
| 65 |
+
"logps/rejected": -252.97470092773438,
|
| 66 |
+
"loss": 0.6909,
|
| 67 |
+
"pred_label": 0.0,
|
| 68 |
+
"rewards/accuracies": 0.565625011920929,
|
| 69 |
+
"rewards/chosen": 0.0062743439339101315,
|
| 70 |
+
"rewards/margins": 0.007127248682081699,
|
| 71 |
+
"rewards/rejected": -0.0008529046317562461,
|
| 72 |
+
"step": 30,
|
| 73 |
+
"use_label": 802.0
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.04,
|
| 77 |
+
"learning_rate": 4.166666666666667e-06,
|
| 78 |
+
"logits/chosen": -2.829822301864624,
|
| 79 |
+
"logits/rejected": -2.8364169597625732,
|
| 80 |
+
"logps/chosen": -279.52288818359375,
|
| 81 |
+
"logps/rejected": -273.93243408203125,
|
| 82 |
+
"loss": 0.6846,
|
| 83 |
+
"pred_label": 0.0,
|
| 84 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 85 |
+
"rewards/chosen": 0.0057884035632014275,
|
| 86 |
+
"rewards/margins": 0.016727477312088013,
|
| 87 |
+
"rewards/rejected": -0.010939070954918861,
|
| 88 |
+
"step": 40,
|
| 89 |
+
"use_label": 1122.0
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.05,
|
| 93 |
+
"learning_rate": 5.208333333333334e-06,
|
| 94 |
+
"logits/chosen": -2.8475687503814697,
|
| 95 |
+
"logits/rejected": -2.8291072845458984,
|
| 96 |
+
"logps/chosen": -266.10870361328125,
|
| 97 |
+
"logps/rejected": -257.93243408203125,
|
| 98 |
+
"loss": 0.6748,
|
| 99 |
+
"pred_label": 0.0,
|
| 100 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 101 |
+
"rewards/chosen": 0.009051208384335041,
|
| 102 |
+
"rewards/margins": 0.03577073663473129,
|
| 103 |
+
"rewards/rejected": -0.026719529181718826,
|
| 104 |
+
"step": 50,
|
| 105 |
+
"use_label": 1442.0
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.06,
|
| 109 |
+
"learning_rate": 6.25e-06,
|
| 110 |
+
"logits/chosen": -2.8435542583465576,
|
| 111 |
+
"logits/rejected": -2.857710838317871,
|
| 112 |
+
"logps/chosen": -308.6976623535156,
|
| 113 |
+
"logps/rejected": -283.2618103027344,
|
| 114 |
+
"loss": 0.6705,
|
| 115 |
+
"pred_label": 0.0,
|
| 116 |
+
"rewards/accuracies": 0.609375,
|
| 117 |
+
"rewards/chosen": 0.015249615535140038,
|
| 118 |
+
"rewards/margins": 0.05724747106432915,
|
| 119 |
+
"rewards/rejected": -0.04199784994125366,
|
| 120 |
+
"step": 60,
|
| 121 |
+
"use_label": 1762.0
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.07,
|
| 125 |
+
"learning_rate": 7.291666666666667e-06,
|
| 126 |
+
"logits/chosen": -2.8543224334716797,
|
| 127 |
+
"logits/rejected": -2.8573386669158936,
|
| 128 |
+
"logps/chosen": -284.1624755859375,
|
| 129 |
+
"logps/rejected": -268.72540283203125,
|
| 130 |
+
"loss": 0.6493,
|
| 131 |
+
"pred_label": 0.0,
|
| 132 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 133 |
+
"rewards/chosen": 0.021748105064034462,
|
| 134 |
+
"rewards/margins": 0.10904034227132797,
|
| 135 |
+
"rewards/rejected": -0.08729223161935806,
|
| 136 |
+
"step": 70,
|
| 137 |
+
"use_label": 2082.0
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.08,
|
| 141 |
+
"learning_rate": 8.333333333333334e-06,
|
| 142 |
+
"logits/chosen": -2.8060083389282227,
|
| 143 |
+
"logits/rejected": -2.8031229972839355,
|
| 144 |
+
"logps/chosen": -310.87384033203125,
|
| 145 |
+
"logps/rejected": -279.0975341796875,
|
| 146 |
+
"loss": 0.6288,
|
| 147 |
+
"pred_label": 2.575000047683716,
|
| 148 |
+
"rewards/accuracies": 0.706250011920929,
|
| 149 |
+
"rewards/chosen": 0.010100151412189007,
|
| 150 |
+
"rewards/margins": 0.17576415836811066,
|
| 151 |
+
"rewards/rejected": -0.16566403210163116,
|
| 152 |
+
"step": 80,
|
| 153 |
+
"use_label": 2399.425048828125
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.09,
|
| 157 |
+
"learning_rate": 9.375000000000001e-06,
|
| 158 |
+
"logits/chosen": -2.79826283454895,
|
| 159 |
+
"logits/rejected": -2.7996926307678223,
|
| 160 |
+
"logps/chosen": -295.39630126953125,
|
| 161 |
+
"logps/rejected": -255.96920776367188,
|
| 162 |
+
"loss": 0.6425,
|
| 163 |
+
"pred_label": 18.225000381469727,
|
| 164 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 165 |
+
"rewards/chosen": 0.03907310217618942,
|
| 166 |
+
"rewards/margins": 0.2187824696302414,
|
| 167 |
+
"rewards/rejected": -0.17970936000347137,
|
| 168 |
+
"step": 90,
|
| 169 |
+
"use_label": 2703.77490234375
|
| 170 |
},
|
| 171 |
{
|
| 172 |
"epoch": 0.1,
|
| 173 |
"learning_rate": 9.953434225844005e-06,
|
| 174 |
+
"logits/chosen": -2.806579351425171,
|
| 175 |
+
"logits/rejected": -2.791835308074951,
|
| 176 |
+
"logps/chosen": -275.96881103515625,
|
| 177 |
+
"logps/rejected": -271.619140625,
|
| 178 |
+
"loss": 0.6514,
|
| 179 |
+
"pred_label": 45.287498474121094,
|
| 180 |
+
"rewards/accuracies": 0.668749988079071,
|
| 181 |
+
"rewards/chosen": 0.03931427747011185,
|
| 182 |
+
"rewards/margins": 0.21040363609790802,
|
| 183 |
+
"rewards/rejected": -0.17108935117721558,
|
| 184 |
"step": 100,
|
| 185 |
+
"use_label": 2996.71240234375
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.12,
|
| 189 |
+
"learning_rate": 9.837019790454017e-06,
|
| 190 |
+
"logits/chosen": -2.802504062652588,
|
| 191 |
+
"logits/rejected": -2.794646739959717,
|
| 192 |
+
"logps/chosen": -273.3879699707031,
|
| 193 |
+
"logps/rejected": -256.89385986328125,
|
| 194 |
+
"loss": 0.6432,
|
| 195 |
+
"pred_label": 74.8375015258789,
|
| 196 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 197 |
+
"rewards/chosen": 0.011993913911283016,
|
| 198 |
+
"rewards/margins": 0.2596678137779236,
|
| 199 |
+
"rewards/rejected": -0.24767383933067322,
|
| 200 |
+
"step": 110,
|
| 201 |
+
"use_label": 3287.16259765625
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.13,
|
| 205 |
+
"learning_rate": 9.72060535506403e-06,
|
| 206 |
+
"logits/chosen": -2.8181800842285156,
|
| 207 |
+
"logits/rejected": -2.8214244842529297,
|
| 208 |
+
"logps/chosen": -295.31146240234375,
|
| 209 |
+
"logps/rejected": -268.0743713378906,
|
| 210 |
+
"loss": 0.6446,
|
| 211 |
+
"pred_label": 118.8499984741211,
|
| 212 |
+
"rewards/accuracies": 0.6781250238418579,
|
| 213 |
+
"rewards/chosen": 0.035851169377565384,
|
| 214 |
+
"rewards/margins": 0.26491057872772217,
|
| 215 |
+
"rewards/rejected": -0.22905941307544708,
|
| 216 |
+
"step": 120,
|
| 217 |
+
"use_label": 3563.14990234375
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 0.14,
|
| 221 |
+
"learning_rate": 9.60419091967404e-06,
|
| 222 |
+
"logits/chosen": -2.8419888019561768,
|
| 223 |
+
"logits/rejected": -2.830252170562744,
|
| 224 |
+
"logps/chosen": -295.4115295410156,
|
| 225 |
+
"logps/rejected": -277.87567138671875,
|
| 226 |
+
"loss": 0.6481,
|
| 227 |
+
"pred_label": 160.16250610351562,
|
| 228 |
+
"rewards/accuracies": 0.6781250238418579,
|
| 229 |
+
"rewards/chosen": 0.014184275642037392,
|
| 230 |
+
"rewards/margins": 0.24397656321525574,
|
| 231 |
+
"rewards/rejected": -0.22979231178760529,
|
| 232 |
+
"step": 130,
|
| 233 |
+
"use_label": 3841.83740234375
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.15,
|
| 237 |
+
"learning_rate": 9.487776484284052e-06,
|
| 238 |
+
"logits/chosen": -2.814854145050049,
|
| 239 |
+
"logits/rejected": -2.818312644958496,
|
| 240 |
+
"logps/chosen": -267.75347900390625,
|
| 241 |
+
"logps/rejected": -255.11355590820312,
|
| 242 |
+
"loss": 0.6374,
|
| 243 |
+
"pred_label": 208.9875030517578,
|
| 244 |
+
"rewards/accuracies": 0.7093750238418579,
|
| 245 |
+
"rewards/chosen": -0.0010342694586142898,
|
| 246 |
+
"rewards/margins": 0.3312947750091553,
|
| 247 |
+
"rewards/rejected": -0.33232906460762024,
|
| 248 |
+
"step": 140,
|
| 249 |
+
"use_label": 4113.0126953125
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.16,
|
| 253 |
+
"learning_rate": 9.371362048894065e-06,
|
| 254 |
+
"logits/chosen": -2.8179469108581543,
|
| 255 |
+
"logits/rejected": -2.81597900390625,
|
| 256 |
+
"logps/chosen": -301.8087463378906,
|
| 257 |
+
"logps/rejected": -279.8531799316406,
|
| 258 |
+
"loss": 0.6613,
|
| 259 |
+
"pred_label": 270.86248779296875,
|
| 260 |
+
"rewards/accuracies": 0.659375011920929,
|
| 261 |
+
"rewards/chosen": 0.009522153064608574,
|
| 262 |
+
"rewards/margins": 0.33681994676589966,
|
| 263 |
+
"rewards/rejected": -0.3272978365421295,
|
| 264 |
+
"step": 150,
|
| 265 |
+
"use_label": 4371.1376953125
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.17,
|
| 269 |
+
"learning_rate": 9.254947613504075e-06,
|
| 270 |
+
"logits/chosen": -2.828568458557129,
|
| 271 |
+
"logits/rejected": -2.789271354675293,
|
| 272 |
+
"logps/chosen": -288.80511474609375,
|
| 273 |
+
"logps/rejected": -278.41839599609375,
|
| 274 |
+
"loss": 0.6349,
|
| 275 |
+
"pred_label": 344.1625061035156,
|
| 276 |
+
"rewards/accuracies": 0.7406250238418579,
|
| 277 |
+
"rewards/chosen": 0.11308002471923828,
|
| 278 |
+
"rewards/margins": 0.451881468296051,
|
| 279 |
+
"rewards/rejected": -0.33880144357681274,
|
| 280 |
+
"step": 160,
|
| 281 |
+
"use_label": 4617.83740234375
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 0.18,
|
| 285 |
+
"learning_rate": 9.138533178114087e-06,
|
| 286 |
+
"logits/chosen": -2.8365304470062256,
|
| 287 |
+
"logits/rejected": -2.822173595428467,
|
| 288 |
+
"logps/chosen": -289.2629089355469,
|
| 289 |
+
"logps/rejected": -254.9129180908203,
|
| 290 |
+
"loss": 0.6585,
|
| 291 |
+
"pred_label": 416.25,
|
| 292 |
+
"rewards/accuracies": 0.706250011920929,
|
| 293 |
+
"rewards/chosen": 0.056967057287693024,
|
| 294 |
+
"rewards/margins": 0.447601854801178,
|
| 295 |
+
"rewards/rejected": -0.39063477516174316,
|
| 296 |
+
"step": 170,
|
| 297 |
+
"use_label": 4865.75
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.19,
|
| 301 |
+
"learning_rate": 9.022118742724098e-06,
|
| 302 |
+
"logits/chosen": -2.8115687370300293,
|
| 303 |
+
"logits/rejected": -2.8006014823913574,
|
| 304 |
+
"logps/chosen": -246.71243286132812,
|
| 305 |
+
"logps/rejected": -266.5823059082031,
|
| 306 |
+
"loss": 0.6515,
|
| 307 |
+
"pred_label": 510.7749938964844,
|
| 308 |
+
"rewards/accuracies": 0.675000011920929,
|
| 309 |
+
"rewards/chosen": 0.02543136477470398,
|
| 310 |
+
"rewards/margins": 0.4450693726539612,
|
| 311 |
+
"rewards/rejected": -0.4196379780769348,
|
| 312 |
+
"step": 180,
|
| 313 |
+
"use_label": 5091.22509765625
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 0.2,
|
| 317 |
+
"learning_rate": 8.90570430733411e-06,
|
| 318 |
+
"logits/chosen": -2.8512680530548096,
|
| 319 |
+
"logits/rejected": -2.8256707191467285,
|
| 320 |
+
"logps/chosen": -304.07403564453125,
|
| 321 |
+
"logps/rejected": -274.03021240234375,
|
| 322 |
+
"loss": 0.643,
|
| 323 |
+
"pred_label": 602.5,
|
| 324 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 325 |
+
"rewards/chosen": 0.13308081030845642,
|
| 326 |
+
"rewards/margins": 0.5033096671104431,
|
| 327 |
+
"rewards/rejected": -0.3702288568019867,
|
| 328 |
+
"step": 190,
|
| 329 |
+
"use_label": 5319.5
|
| 330 |
},
|
| 331 |
{
|
| 332 |
"epoch": 0.21,
|
| 333 |
"learning_rate": 8.789289871944122e-06,
|
| 334 |
+
"logits/chosen": -2.8375296592712402,
|
| 335 |
+
"logits/rejected": -2.8491904735565186,
|
| 336 |
+
"logps/chosen": -268.386962890625,
|
| 337 |
+
"logps/rejected": -265.6180725097656,
|
| 338 |
+
"loss": 0.6502,
|
| 339 |
+
"pred_label": 694.9249877929688,
|
| 340 |
+
"rewards/accuracies": 0.715624988079071,
|
| 341 |
+
"rewards/chosen": 0.07495273649692535,
|
| 342 |
+
"rewards/margins": 0.48230332136154175,
|
| 343 |
+
"rewards/rejected": -0.4073505997657776,
|
| 344 |
"step": 200,
|
| 345 |
+
"use_label": 5547.0751953125
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.22,
|
| 349 |
+
"learning_rate": 8.672875436554133e-06,
|
| 350 |
+
"logits/chosen": -2.8025362491607666,
|
| 351 |
+
"logits/rejected": -2.813215494155884,
|
| 352 |
+
"logps/chosen": -282.7426452636719,
|
| 353 |
+
"logps/rejected": -268.80078125,
|
| 354 |
+
"loss": 0.65,
|
| 355 |
+
"pred_label": 789.7625122070312,
|
| 356 |
+
"rewards/accuracies": 0.715624988079071,
|
| 357 |
+
"rewards/chosen": 0.07436653971672058,
|
| 358 |
+
"rewards/margins": 0.490271657705307,
|
| 359 |
+
"rewards/rejected": -0.41590508818626404,
|
| 360 |
+
"step": 210,
|
| 361 |
+
"use_label": 5772.2373046875
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.23,
|
| 365 |
+
"learning_rate": 8.556461001164145e-06,
|
| 366 |
+
"logits/chosen": -2.8301258087158203,
|
| 367 |
+
"logits/rejected": -2.8380088806152344,
|
| 368 |
+
"logps/chosen": -277.80615234375,
|
| 369 |
+
"logps/rejected": -272.68377685546875,
|
| 370 |
+
"loss": 0.6534,
|
| 371 |
+
"pred_label": 899.75,
|
| 372 |
+
"rewards/accuracies": 0.71875,
|
| 373 |
+
"rewards/chosen": 0.07161492854356766,
|
| 374 |
+
"rewards/margins": 0.5100919604301453,
|
| 375 |
+
"rewards/rejected": -0.4384769797325134,
|
| 376 |
+
"step": 220,
|
| 377 |
+
"use_label": 5982.25
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.24,
|
| 381 |
+
"learning_rate": 8.440046565774158e-06,
|
| 382 |
+
"logits/chosen": -2.8242669105529785,
|
| 383 |
+
"logits/rejected": -2.835315465927124,
|
| 384 |
+
"logps/chosen": -309.482177734375,
|
| 385 |
+
"logps/rejected": -273.8406677246094,
|
| 386 |
+
"loss": 0.6632,
|
| 387 |
+
"pred_label": 1013.2999877929688,
|
| 388 |
+
"rewards/accuracies": 0.668749988079071,
|
| 389 |
+
"rewards/chosen": 0.012543338350951672,
|
| 390 |
+
"rewards/margins": 0.49393802881240845,
|
| 391 |
+
"rewards/rejected": -0.4813947081565857,
|
| 392 |
+
"step": 230,
|
| 393 |
+
"use_label": 6188.7001953125
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.25,
|
| 397 |
+
"learning_rate": 8.323632130384168e-06,
|
| 398 |
+
"logits/chosen": -2.7957725524902344,
|
| 399 |
+
"logits/rejected": -2.776923894882202,
|
| 400 |
+
"logps/chosen": -299.4118957519531,
|
| 401 |
+
"logps/rejected": -264.6926574707031,
|
| 402 |
+
"loss": 0.6664,
|
| 403 |
+
"pred_label": 1130.2249755859375,
|
| 404 |
+
"rewards/accuracies": 0.731249988079071,
|
| 405 |
+
"rewards/chosen": 0.08671466261148453,
|
| 406 |
+
"rewards/margins": 0.5532404780387878,
|
| 407 |
+
"rewards/rejected": -0.4665258526802063,
|
| 408 |
+
"step": 240,
|
| 409 |
+
"use_label": 6391.77490234375
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.26,
|
| 413 |
+
"learning_rate": 8.20721769499418e-06,
|
| 414 |
+
"logits/chosen": -2.7934088706970215,
|
| 415 |
+
"logits/rejected": -2.7692036628723145,
|
| 416 |
+
"logps/chosen": -300.4549255371094,
|
| 417 |
+
"logps/rejected": -288.0910949707031,
|
| 418 |
+
"loss": 0.6608,
|
| 419 |
+
"pred_label": 1240.574951171875,
|
| 420 |
+
"rewards/accuracies": 0.737500011920929,
|
| 421 |
+
"rewards/chosen": 0.06411644071340561,
|
| 422 |
+
"rewards/margins": 0.5871235132217407,
|
| 423 |
+
"rewards/rejected": -0.5230070948600769,
|
| 424 |
+
"step": 250,
|
| 425 |
+
"use_label": 6601.4248046875
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 0.27,
|
| 429 |
+
"learning_rate": 8.090803259604193e-06,
|
| 430 |
+
"logits/chosen": -2.8178822994232178,
|
| 431 |
+
"logits/rejected": -2.797152519226074,
|
| 432 |
+
"logps/chosen": -278.56048583984375,
|
| 433 |
+
"logps/rejected": -266.2746887207031,
|
| 434 |
+
"loss": 0.6506,
|
| 435 |
+
"pred_label": 1338.4625244140625,
|
| 436 |
+
"rewards/accuracies": 0.6781250238418579,
|
| 437 |
+
"rewards/chosen": -0.02512388862669468,
|
| 438 |
+
"rewards/margins": 0.4762639105319977,
|
| 439 |
+
"rewards/rejected": -0.5013878345489502,
|
| 440 |
+
"step": 260,
|
| 441 |
+
"use_label": 6823.53759765625
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.28,
|
| 445 |
+
"learning_rate": 7.974388824214203e-06,
|
| 446 |
+
"logits/chosen": -2.7823479175567627,
|
| 447 |
+
"logits/rejected": -2.7818832397460938,
|
| 448 |
+
"logps/chosen": -274.42913818359375,
|
| 449 |
+
"logps/rejected": -266.7970886230469,
|
| 450 |
+
"loss": 0.6585,
|
| 451 |
+
"pred_label": 1452.3499755859375,
|
| 452 |
+
"rewards/accuracies": 0.7281249761581421,
|
| 453 |
+
"rewards/chosen": 0.012837971560657024,
|
| 454 |
+
"rewards/margins": 0.6269992589950562,
|
| 455 |
+
"rewards/rejected": -0.6141613125801086,
|
| 456 |
+
"step": 270,
|
| 457 |
+
"use_label": 7029.64990234375
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.29,
|
| 461 |
+
"learning_rate": 7.857974388824214e-06,
|
| 462 |
+
"logits/chosen": -2.872814178466797,
|
| 463 |
+
"logits/rejected": -2.85591459274292,
|
| 464 |
+
"logps/chosen": -293.08001708984375,
|
| 465 |
+
"logps/rejected": -255.07369995117188,
|
| 466 |
+
"loss": 0.6564,
|
| 467 |
+
"pred_label": 1567.074951171875,
|
| 468 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 469 |
+
"rewards/chosen": 0.08200586587190628,
|
| 470 |
+
"rewards/margins": 0.6136232614517212,
|
| 471 |
+
"rewards/rejected": -0.5316173434257507,
|
| 472 |
+
"step": 280,
|
| 473 |
+
"use_label": 7234.9248046875
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 0.3,
|
| 477 |
+
"learning_rate": 7.741559953434226e-06,
|
| 478 |
+
"logits/chosen": -2.829739809036255,
|
| 479 |
+
"logits/rejected": -2.848804473876953,
|
| 480 |
+
"logps/chosen": -289.7538146972656,
|
| 481 |
+
"logps/rejected": -243.77880859375,
|
| 482 |
+
"loss": 0.6518,
|
| 483 |
+
"pred_label": 1677.5999755859375,
|
| 484 |
+
"rewards/accuracies": 0.734375,
|
| 485 |
+
"rewards/chosen": 0.04277036339044571,
|
| 486 |
+
"rewards/margins": 0.530608057975769,
|
| 487 |
+
"rewards/rejected": -0.48783770203590393,
|
| 488 |
+
"step": 290,
|
| 489 |
+
"use_label": 7444.39990234375
|
| 490 |
},
|
| 491 |
{
|
| 492 |
"epoch": 0.31,
|
| 493 |
"learning_rate": 7.625145518044238e-06,
|
| 494 |
+
"logits/chosen": -2.827993154525757,
|
| 495 |
+
"logits/rejected": -2.820228338241577,
|
| 496 |
+
"logps/chosen": -291.6603088378906,
|
| 497 |
+
"logps/rejected": -252.08853149414062,
|
| 498 |
+
"loss": 0.6598,
|
| 499 |
+
"pred_label": 1791.0250244140625,
|
| 500 |
+
"rewards/accuracies": 0.6968749761581421,
|
| 501 |
+
"rewards/chosen": -0.019189920276403427,
|
| 502 |
+
"rewards/margins": 0.5760209560394287,
|
| 503 |
+
"rewards/rejected": -0.5952108502388,
|
| 504 |
"step": 300,
|
| 505 |
+
"use_label": 7650.97509765625
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.32,
|
| 509 |
+
"learning_rate": 7.50873108265425e-06,
|
| 510 |
+
"logits/chosen": -2.8010551929473877,
|
| 511 |
+
"logits/rejected": -2.775717258453369,
|
| 512 |
+
"logps/chosen": -275.4491271972656,
|
| 513 |
+
"logps/rejected": -273.45440673828125,
|
| 514 |
+
"loss": 0.6603,
|
| 515 |
+
"pred_label": 1909.6875,
|
| 516 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 517 |
+
"rewards/chosen": -0.04083142429590225,
|
| 518 |
+
"rewards/margins": 0.6963170766830444,
|
| 519 |
+
"rewards/rejected": -0.7371484041213989,
|
| 520 |
+
"step": 310,
|
| 521 |
+
"use_label": 7852.3125
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.33,
|
| 525 |
+
"learning_rate": 7.392316647264262e-06,
|
| 526 |
+
"logits/chosen": -2.834115505218506,
|
| 527 |
+
"logits/rejected": -2.8193936347961426,
|
| 528 |
+
"logps/chosen": -300.43115234375,
|
| 529 |
+
"logps/rejected": -270.88232421875,
|
| 530 |
+
"loss": 0.6711,
|
| 531 |
+
"pred_label": 2038.449951171875,
|
| 532 |
+
"rewards/accuracies": 0.746874988079071,
|
| 533 |
+
"rewards/chosen": 0.041024383157491684,
|
| 534 |
+
"rewards/margins": 0.695600688457489,
|
| 535 |
+
"rewards/rejected": -0.654576301574707,
|
| 536 |
+
"step": 320,
|
| 537 |
+
"use_label": 8043.5498046875
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.35,
|
| 541 |
+
"learning_rate": 7.275902211874272e-06,
|
| 542 |
+
"logits/chosen": -2.810176134109497,
|
| 543 |
+
"logits/rejected": -2.7961533069610596,
|
| 544 |
+
"logps/chosen": -261.0621032714844,
|
| 545 |
+
"logps/rejected": -249.53463745117188,
|
| 546 |
+
"loss": 0.6833,
|
| 547 |
+
"pred_label": 2174.97509765625,
|
| 548 |
+
"rewards/accuracies": 0.703125,
|
| 549 |
+
"rewards/chosen": -0.025768589228391647,
|
| 550 |
+
"rewards/margins": 0.5799206495285034,
|
| 551 |
+
"rewards/rejected": -0.605689287185669,
|
| 552 |
+
"step": 330,
|
| 553 |
+
"use_label": 8227.025390625
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 0.36,
|
| 557 |
+
"learning_rate": 7.1594877764842855e-06,
|
| 558 |
+
"logits/chosen": -2.831383466720581,
|
| 559 |
+
"logits/rejected": -2.8242664337158203,
|
| 560 |
+
"logps/chosen": -295.96441650390625,
|
| 561 |
+
"logps/rejected": -265.889404296875,
|
| 562 |
+
"loss": 0.6752,
|
| 563 |
+
"pred_label": 2300.237548828125,
|
| 564 |
+
"rewards/accuracies": 0.6468750238418579,
|
| 565 |
+
"rewards/chosen": 0.007702559232711792,
|
| 566 |
+
"rewards/margins": 0.4926396310329437,
|
| 567 |
+
"rewards/rejected": -0.48493701219558716,
|
| 568 |
+
"step": 340,
|
| 569 |
+
"use_label": 8421.7626953125
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.37,
|
| 573 |
+
"learning_rate": 7.043073341094296e-06,
|
| 574 |
+
"logits/chosen": -2.8328559398651123,
|
| 575 |
+
"logits/rejected": -2.8223798274993896,
|
| 576 |
+
"logps/chosen": -288.3695983886719,
|
| 577 |
+
"logps/rejected": -274.58697509765625,
|
| 578 |
+
"loss": 0.6644,
|
| 579 |
+
"pred_label": 2420.324951171875,
|
| 580 |
+
"rewards/accuracies": 0.721875011920929,
|
| 581 |
+
"rewards/chosen": 0.017813527956604958,
|
| 582 |
+
"rewards/margins": 0.5609097480773926,
|
| 583 |
+
"rewards/rejected": -0.5430961847305298,
|
| 584 |
+
"step": 350,
|
| 585 |
+
"use_label": 8621.6748046875
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.38,
|
| 589 |
+
"learning_rate": 6.9266589057043075e-06,
|
| 590 |
+
"logits/chosen": -2.87721586227417,
|
| 591 |
+
"logits/rejected": -2.87174654006958,
|
| 592 |
+
"logps/chosen": -293.4356689453125,
|
| 593 |
+
"logps/rejected": -276.60235595703125,
|
| 594 |
+
"loss": 0.6666,
|
| 595 |
+
"pred_label": 2544.78759765625,
|
| 596 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 597 |
+
"rewards/chosen": 0.006689542438834906,
|
| 598 |
+
"rewards/margins": 0.6315399408340454,
|
| 599 |
+
"rewards/rejected": -0.6248503923416138,
|
| 600 |
+
"step": 360,
|
| 601 |
+
"use_label": 8817.212890625
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 0.39,
|
| 605 |
+
"learning_rate": 6.81024447031432e-06,
|
| 606 |
+
"logits/chosen": -2.8437862396240234,
|
| 607 |
+
"logits/rejected": -2.791590690612793,
|
| 608 |
+
"logps/chosen": -286.032470703125,
|
| 609 |
+
"logps/rejected": -276.44036865234375,
|
| 610 |
+
"loss": 0.6545,
|
| 611 |
+
"pred_label": 2656.925048828125,
|
| 612 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 613 |
+
"rewards/chosen": 0.07002071291208267,
|
| 614 |
+
"rewards/margins": 0.6711146235466003,
|
| 615 |
+
"rewards/rejected": -0.6010940074920654,
|
| 616 |
+
"step": 370,
|
| 617 |
+
"use_label": 9025.0751953125
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 0.4,
|
| 621 |
+
"learning_rate": 6.693830034924331e-06,
|
| 622 |
+
"logits/chosen": -2.8383095264434814,
|
| 623 |
+
"logits/rejected": -2.820664644241333,
|
| 624 |
+
"logps/chosen": -281.7953796386719,
|
| 625 |
+
"logps/rejected": -268.8168640136719,
|
| 626 |
+
"loss": 0.6536,
|
| 627 |
+
"pred_label": 2769.987548828125,
|
| 628 |
+
"rewards/accuracies": 0.7406250238418579,
|
| 629 |
+
"rewards/chosen": 0.04205373674631119,
|
| 630 |
+
"rewards/margins": 0.6800293326377869,
|
| 631 |
+
"rewards/rejected": -0.6379756331443787,
|
| 632 |
+
"step": 380,
|
| 633 |
+
"use_label": 9232.0126953125
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.41,
|
| 637 |
+
"learning_rate": 6.5774155995343425e-06,
|
| 638 |
+
"logits/chosen": -2.7942054271698,
|
| 639 |
+
"logits/rejected": -2.785630702972412,
|
| 640 |
+
"logps/chosen": -272.2677917480469,
|
| 641 |
+
"logps/rejected": -255.5617218017578,
|
| 642 |
+
"loss": 0.6655,
|
| 643 |
+
"pred_label": 2901.1875,
|
| 644 |
+
"rewards/accuracies": 0.699999988079071,
|
| 645 |
+
"rewards/chosen": 0.003805020358413458,
|
| 646 |
+
"rewards/margins": 0.6898488998413086,
|
| 647 |
+
"rewards/rejected": -0.6860438585281372,
|
| 648 |
+
"step": 390,
|
| 649 |
+
"use_label": 9420.8125
|
| 650 |
},
|
| 651 |
{
|
| 652 |
"epoch": 0.42,
|
| 653 |
"learning_rate": 6.461001164144355e-06,
|
| 654 |
+
"logits/chosen": -2.833627223968506,
|
| 655 |
+
"logits/rejected": -2.8359274864196777,
|
| 656 |
+
"logps/chosen": -292.16717529296875,
|
| 657 |
+
"logps/rejected": -257.1654968261719,
|
| 658 |
+
"loss": 0.6586,
|
| 659 |
+
"pred_label": 3028.52490234375,
|
| 660 |
+
"rewards/accuracies": 0.737500011920929,
|
| 661 |
+
"rewards/chosen": 0.10834367573261261,
|
| 662 |
+
"rewards/margins": 0.6510626077651978,
|
| 663 |
+
"rewards/rejected": -0.5427189469337463,
|
| 664 |
"step": 400,
|
| 665 |
+
"use_label": 9613.474609375
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 0.43,
|
| 669 |
+
"learning_rate": 6.344586728754366e-06,
|
| 670 |
+
"logits/chosen": -2.8230559825897217,
|
| 671 |
+
"logits/rejected": -2.8199567794799805,
|
| 672 |
+
"logps/chosen": -283.50323486328125,
|
| 673 |
+
"logps/rejected": -262.53216552734375,
|
| 674 |
+
"loss": 0.6703,
|
| 675 |
+
"pred_label": 3146.237548828125,
|
| 676 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 677 |
+
"rewards/chosen": 0.06537099182605743,
|
| 678 |
+
"rewards/margins": 0.6661577820777893,
|
| 679 |
+
"rewards/rejected": -0.6007868051528931,
|
| 680 |
+
"step": 410,
|
| 681 |
+
"use_label": 9815.7626953125
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.44,
|
| 685 |
+
"learning_rate": 6.228172293364378e-06,
|
| 686 |
+
"logits/chosen": -2.81683087348938,
|
| 687 |
+
"logits/rejected": -2.7953038215637207,
|
| 688 |
+
"logps/chosen": -268.87017822265625,
|
| 689 |
+
"logps/rejected": -258.7505798339844,
|
| 690 |
+
"loss": 0.6553,
|
| 691 |
+
"pred_label": 3272.28759765625,
|
| 692 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 693 |
+
"rewards/chosen": 0.03196418285369873,
|
| 694 |
+
"rewards/margins": 0.5871380567550659,
|
| 695 |
+
"rewards/rejected": -0.5551738739013672,
|
| 696 |
+
"step": 420,
|
| 697 |
+
"use_label": 10009.712890625
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 0.45,
|
| 701 |
+
"learning_rate": 6.11175785797439e-06,
|
| 702 |
+
"logits/chosen": -2.838977575302124,
|
| 703 |
+
"logits/rejected": -2.8169591426849365,
|
| 704 |
+
"logps/chosen": -290.03753662109375,
|
| 705 |
+
"logps/rejected": -266.868896484375,
|
| 706 |
+
"loss": 0.6758,
|
| 707 |
+
"pred_label": 3398.512451171875,
|
| 708 |
+
"rewards/accuracies": 0.6656249761581421,
|
| 709 |
+
"rewards/chosen": 0.028419841080904007,
|
| 710 |
+
"rewards/margins": 0.4861460328102112,
|
| 711 |
+
"rewards/rejected": -0.4577261805534363,
|
| 712 |
+
"step": 430,
|
| 713 |
+
"use_label": 10203.4873046875
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 0.46,
|
| 717 |
+
"learning_rate": 5.995343422584401e-06,
|
| 718 |
+
"logits/chosen": -2.849961996078491,
|
| 719 |
+
"logits/rejected": -2.8381643295288086,
|
| 720 |
+
"logps/chosen": -282.6123046875,
|
| 721 |
+
"logps/rejected": -246.48464965820312,
|
| 722 |
+
"loss": 0.6602,
|
| 723 |
+
"pred_label": 3515.35009765625,
|
| 724 |
+
"rewards/accuracies": 0.703125,
|
| 725 |
+
"rewards/chosen": 0.038983773440122604,
|
| 726 |
+
"rewards/margins": 0.5515373945236206,
|
| 727 |
+
"rewards/rejected": -0.5125535726547241,
|
| 728 |
+
"step": 440,
|
| 729 |
+
"use_label": 10406.650390625
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.47,
|
| 733 |
+
"learning_rate": 5.878928987194412e-06,
|
| 734 |
+
"logits/chosen": -2.749311923980713,
|
| 735 |
+
"logits/rejected": -2.7318615913391113,
|
| 736 |
+
"logps/chosen": -257.08990478515625,
|
| 737 |
+
"logps/rejected": -258.7293395996094,
|
| 738 |
+
"loss": 0.6612,
|
| 739 |
+
"pred_label": 3629.699951171875,
|
| 740 |
+
"rewards/accuracies": 0.734375,
|
| 741 |
+
"rewards/chosen": 0.028371259570121765,
|
| 742 |
+
"rewards/margins": 0.5704750418663025,
|
| 743 |
+
"rewards/rejected": -0.5421038866043091,
|
| 744 |
+
"step": 450,
|
| 745 |
+
"use_label": 10612.2998046875
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.48,
|
| 749 |
+
"learning_rate": 5.762514551804425e-06,
|
| 750 |
+
"logits/chosen": -2.8554000854492188,
|
| 751 |
+
"logits/rejected": -2.8365120887756348,
|
| 752 |
+
"logps/chosen": -305.9751892089844,
|
| 753 |
+
"logps/rejected": -270.2017517089844,
|
| 754 |
+
"loss": 0.6574,
|
| 755 |
+
"pred_label": 3751.949951171875,
|
| 756 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 757 |
+
"rewards/chosen": -0.021648898720741272,
|
| 758 |
+
"rewards/margins": 0.4913213849067688,
|
| 759 |
+
"rewards/rejected": -0.5129703283309937,
|
| 760 |
+
"step": 460,
|
| 761 |
+
"use_label": 10810.0498046875
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 0.49,
|
| 765 |
+
"learning_rate": 5.6461001164144355e-06,
|
| 766 |
+
"logits/chosen": -2.8250269889831543,
|
| 767 |
+
"logits/rejected": -2.8220317363739014,
|
| 768 |
+
"logps/chosen": -295.07611083984375,
|
| 769 |
+
"logps/rejected": -271.61090087890625,
|
| 770 |
+
"loss": 0.6658,
|
| 771 |
+
"pred_label": 3857.625,
|
| 772 |
+
"rewards/accuracies": 0.7281249761581421,
|
| 773 |
+
"rewards/chosen": 0.03160310536623001,
|
| 774 |
+
"rewards/margins": 0.5609144568443298,
|
| 775 |
+
"rewards/rejected": -0.5293112993240356,
|
| 776 |
+
"step": 470,
|
| 777 |
+
"use_label": 11024.375
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 0.5,
|
| 781 |
+
"learning_rate": 5.529685681024447e-06,
|
| 782 |
+
"logits/chosen": -2.765904188156128,
|
| 783 |
+
"logits/rejected": -2.7620699405670166,
|
| 784 |
+
"logps/chosen": -253.2978973388672,
|
| 785 |
+
"logps/rejected": -264.97259521484375,
|
| 786 |
+
"loss": 0.674,
|
| 787 |
+
"pred_label": 3982.0,
|
| 788 |
+
"rewards/accuracies": 0.625,
|
| 789 |
+
"rewards/chosen": -0.05188765376806259,
|
| 790 |
+
"rewards/margins": 0.46525877714157104,
|
| 791 |
+
"rewards/rejected": -0.5171464085578918,
|
| 792 |
+
"step": 480,
|
| 793 |
+
"use_label": 11220.0
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.51,
|
| 797 |
+
"learning_rate": 5.413271245634459e-06,
|
| 798 |
+
"logits/chosen": -2.80226993560791,
|
| 799 |
+
"logits/rejected": -2.7974660396575928,
|
| 800 |
+
"logps/chosen": -284.6823425292969,
|
| 801 |
+
"logps/rejected": -252.20889282226562,
|
| 802 |
+
"loss": 0.6631,
|
| 803 |
+
"pred_label": 4095.4375,
|
| 804 |
+
"rewards/accuracies": 0.706250011920929,
|
| 805 |
+
"rewards/chosen": 0.0527660958468914,
|
| 806 |
+
"rewards/margins": 0.6084800958633423,
|
| 807 |
+
"rewards/rejected": -0.5557140707969666,
|
| 808 |
+
"step": 490,
|
| 809 |
+
"use_label": 11426.5625
|
| 810 |
},
|
| 811 |
{
|
| 812 |
"epoch": 0.52,
|
| 813 |
"learning_rate": 5.2968568102444705e-06,
|
| 814 |
+
"logits/chosen": -2.8293509483337402,
|
| 815 |
+
"logits/rejected": -2.798262119293213,
|
| 816 |
+
"logps/chosen": -289.0420837402344,
|
| 817 |
+
"logps/rejected": -272.00714111328125,
|
| 818 |
+
"loss": 0.6644,
|
| 819 |
+
"pred_label": 4224.21240234375,
|
| 820 |
+
"rewards/accuracies": 0.721875011920929,
|
| 821 |
+
"rewards/chosen": 0.059389661997556686,
|
| 822 |
+
"rewards/margins": 0.6478527784347534,
|
| 823 |
+
"rewards/rejected": -0.5884631276130676,
|
| 824 |
"step": 500,
|
| 825 |
+
"use_label": 11617.787109375
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.53,
|
| 829 |
+
"learning_rate": 5.180442374854482e-06,
|
| 830 |
+
"logits/chosen": -2.844891309738159,
|
| 831 |
+
"logits/rejected": -2.833529233932495,
|
| 832 |
+
"logps/chosen": -275.63726806640625,
|
| 833 |
+
"logps/rejected": -260.7143249511719,
|
| 834 |
+
"loss": 0.6628,
|
| 835 |
+
"pred_label": 4350.9873046875,
|
| 836 |
+
"rewards/accuracies": 0.71875,
|
| 837 |
+
"rewards/chosen": 0.03330535814166069,
|
| 838 |
+
"rewards/margins": 0.6962288618087769,
|
| 839 |
+
"rewards/rejected": -0.6629236340522766,
|
| 840 |
+
"step": 510,
|
| 841 |
+
"use_label": 11811.0126953125
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 0.54,
|
| 845 |
+
"learning_rate": 5.064027939464494e-06,
|
| 846 |
+
"logits/chosen": -2.854224681854248,
|
| 847 |
+
"logits/rejected": -2.816175937652588,
|
| 848 |
+
"logps/chosen": -282.9071044921875,
|
| 849 |
+
"logps/rejected": -263.75927734375,
|
| 850 |
+
"loss": 0.6631,
|
| 851 |
+
"pred_label": 4478.27490234375,
|
| 852 |
+
"rewards/accuracies": 0.737500011920929,
|
| 853 |
+
"rewards/chosen": 0.11818108707666397,
|
| 854 |
+
"rewards/margins": 0.7176607251167297,
|
| 855 |
+
"rewards/rejected": -0.5994796752929688,
|
| 856 |
+
"step": 520,
|
| 857 |
+
"use_label": 12003.724609375
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.55,
|
| 861 |
+
"learning_rate": 4.947613504074506e-06,
|
| 862 |
+
"logits/chosen": -2.861614942550659,
|
| 863 |
+
"logits/rejected": -2.8368430137634277,
|
| 864 |
+
"logps/chosen": -289.9693603515625,
|
| 865 |
+
"logps/rejected": -271.81524658203125,
|
| 866 |
+
"loss": 0.6546,
|
| 867 |
+
"pred_label": 4613.4873046875,
|
| 868 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 869 |
+
"rewards/chosen": 0.12808291614055634,
|
| 870 |
+
"rewards/margins": 0.721964955329895,
|
| 871 |
+
"rewards/rejected": -0.5938820838928223,
|
| 872 |
+
"step": 530,
|
| 873 |
+
"use_label": 12188.5126953125
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.57,
|
| 877 |
+
"learning_rate": 4.831199068684517e-06,
|
| 878 |
+
"logits/chosen": -2.8578391075134277,
|
| 879 |
+
"logits/rejected": -2.835608959197998,
|
| 880 |
+
"logps/chosen": -290.48114013671875,
|
| 881 |
+
"logps/rejected": -267.905029296875,
|
| 882 |
+
"loss": 0.6684,
|
| 883 |
+
"pred_label": 4741.5751953125,
|
| 884 |
+
"rewards/accuracies": 0.737500011920929,
|
| 885 |
+
"rewards/chosen": 0.10645272582769394,
|
| 886 |
+
"rewards/margins": 0.6809743642807007,
|
| 887 |
+
"rewards/rejected": -0.5745216608047485,
|
| 888 |
+
"step": 540,
|
| 889 |
+
"use_label": 12380.4248046875
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 0.58,
|
| 893 |
+
"learning_rate": 4.714784633294529e-06,
|
| 894 |
+
"logits/chosen": -2.83345627784729,
|
| 895 |
+
"logits/rejected": -2.812713623046875,
|
| 896 |
+
"logps/chosen": -281.4727783203125,
|
| 897 |
+
"logps/rejected": -274.65338134765625,
|
| 898 |
+
"loss": 0.682,
|
| 899 |
+
"pred_label": 4869.96240234375,
|
| 900 |
+
"rewards/accuracies": 0.6968749761581421,
|
| 901 |
+
"rewards/chosen": 0.036335308104753494,
|
| 902 |
+
"rewards/margins": 0.6714037656784058,
|
| 903 |
+
"rewards/rejected": -0.6350685358047485,
|
| 904 |
+
"step": 550,
|
| 905 |
+
"use_label": 12572.037109375
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.59,
|
| 909 |
+
"learning_rate": 4.598370197904541e-06,
|
| 910 |
+
"logits/chosen": -2.788679838180542,
|
| 911 |
+
"logits/rejected": -2.7916810512542725,
|
| 912 |
+
"logps/chosen": -296.66680908203125,
|
| 913 |
+
"logps/rejected": -242.8921661376953,
|
| 914 |
+
"loss": 0.6684,
|
| 915 |
+
"pred_label": 4992.08740234375,
|
| 916 |
+
"rewards/accuracies": 0.668749988079071,
|
| 917 |
+
"rewards/chosen": 0.01064818911254406,
|
| 918 |
+
"rewards/margins": 0.577103316783905,
|
| 919 |
+
"rewards/rejected": -0.5664551854133606,
|
| 920 |
+
"step": 560,
|
| 921 |
+
"use_label": 12769.912109375
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 0.6,
|
| 925 |
+
"learning_rate": 4.481955762514552e-06,
|
| 926 |
+
"logits/chosen": -2.789961814880371,
|
| 927 |
+
"logits/rejected": -2.769895553588867,
|
| 928 |
+
"logps/chosen": -268.1661071777344,
|
| 929 |
+
"logps/rejected": -256.2967834472656,
|
| 930 |
+
"loss": 0.6717,
|
| 931 |
+
"pred_label": 5114.22509765625,
|
| 932 |
+
"rewards/accuracies": 0.7093750238418579,
|
| 933 |
+
"rewards/chosen": 0.04825209826231003,
|
| 934 |
+
"rewards/margins": 0.5761987566947937,
|
| 935 |
+
"rewards/rejected": -0.5279466509819031,
|
| 936 |
+
"step": 570,
|
| 937 |
+
"use_label": 12967.775390625
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 0.61,
|
| 941 |
+
"learning_rate": 4.365541327124564e-06,
|
| 942 |
+
"logits/chosen": -2.8300013542175293,
|
| 943 |
+
"logits/rejected": -2.822327136993408,
|
| 944 |
+
"logps/chosen": -273.7094421386719,
|
| 945 |
+
"logps/rejected": -265.09967041015625,
|
| 946 |
+
"loss": 0.6591,
|
| 947 |
+
"pred_label": 5237.35009765625,
|
| 948 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 949 |
+
"rewards/chosen": 0.062391918152570724,
|
| 950 |
+
"rewards/margins": 0.5977689623832703,
|
| 951 |
+
"rewards/rejected": -0.535377025604248,
|
| 952 |
+
"step": 580,
|
| 953 |
+
"use_label": 13164.650390625
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 0.62,
|
| 957 |
+
"learning_rate": 4.249126891734576e-06,
|
| 958 |
+
"logits/chosen": -2.795945644378662,
|
| 959 |
+
"logits/rejected": -2.7982325553894043,
|
| 960 |
+
"logps/chosen": -287.38848876953125,
|
| 961 |
+
"logps/rejected": -279.43511962890625,
|
| 962 |
+
"loss": 0.6574,
|
| 963 |
+
"pred_label": 5347.78759765625,
|
| 964 |
+
"rewards/accuracies": 0.703125,
|
| 965 |
+
"rewards/chosen": 0.05194039270281792,
|
| 966 |
+
"rewards/margins": 0.595386266708374,
|
| 967 |
+
"rewards/rejected": -0.5434459447860718,
|
| 968 |
+
"step": 590,
|
| 969 |
+
"use_label": 13374.212890625
|
| 970 |
},
|
| 971 |
{
|
| 972 |
"epoch": 0.63,
|
| 973 |
"learning_rate": 4.132712456344587e-06,
|
| 974 |
+
"logits/chosen": -2.8118224143981934,
|
| 975 |
+
"logits/rejected": -2.8233461380004883,
|
| 976 |
+
"logps/chosen": -295.39080810546875,
|
| 977 |
+
"logps/rejected": -294.5987854003906,
|
| 978 |
+
"loss": 0.6624,
|
| 979 |
+
"pred_label": 5470.6376953125,
|
| 980 |
+
"rewards/accuracies": 0.753125011920929,
|
| 981 |
+
"rewards/chosen": 0.13325130939483643,
|
| 982 |
+
"rewards/margins": 0.7225804924964905,
|
| 983 |
+
"rewards/rejected": -0.5893291234970093,
|
| 984 |
"step": 600,
|
| 985 |
+
"use_label": 13571.3623046875
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 0.64,
|
| 989 |
+
"learning_rate": 4.0162980209545985e-06,
|
| 990 |
+
"logits/chosen": -2.838010311126709,
|
| 991 |
+
"logits/rejected": -2.822805166244507,
|
| 992 |
+
"logps/chosen": -297.541259765625,
|
| 993 |
+
"logps/rejected": -260.2327575683594,
|
| 994 |
+
"loss": 0.6611,
|
| 995 |
+
"pred_label": 5597.2373046875,
|
| 996 |
+
"rewards/accuracies": 0.699999988079071,
|
| 997 |
+
"rewards/chosen": 0.11663957685232162,
|
| 998 |
+
"rewards/margins": 0.5966584086418152,
|
| 999 |
+
"rewards/rejected": -0.48001885414123535,
|
| 1000 |
+
"step": 610,
|
| 1001 |
+
"use_label": 13764.7626953125
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 0.65,
|
| 1005 |
+
"learning_rate": 3.899883585564611e-06,
|
| 1006 |
+
"logits/chosen": -2.8419346809387207,
|
| 1007 |
+
"logits/rejected": -2.7907586097717285,
|
| 1008 |
+
"logps/chosen": -282.9375,
|
| 1009 |
+
"logps/rejected": -255.97988891601562,
|
| 1010 |
+
"loss": 0.6553,
|
| 1011 |
+
"pred_label": 5717.875,
|
| 1012 |
+
"rewards/accuracies": 0.7281249761581421,
|
| 1013 |
+
"rewards/chosen": 0.10196954011917114,
|
| 1014 |
+
"rewards/margins": 0.6096292734146118,
|
| 1015 |
+
"rewards/rejected": -0.5076597332954407,
|
| 1016 |
+
"step": 620,
|
| 1017 |
+
"use_label": 13964.125
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.66,
|
| 1021 |
+
"learning_rate": 3.7834691501746217e-06,
|
| 1022 |
+
"logits/chosen": -2.7976930141448975,
|
| 1023 |
+
"logits/rejected": -2.79984974861145,
|
| 1024 |
+
"logps/chosen": -277.3227844238281,
|
| 1025 |
+
"logps/rejected": -275.438720703125,
|
| 1026 |
+
"loss": 0.6624,
|
| 1027 |
+
"pred_label": 5844.72509765625,
|
| 1028 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1029 |
+
"rewards/chosen": 0.0894940048456192,
|
| 1030 |
+
"rewards/margins": 0.6564904451370239,
|
| 1031 |
+
"rewards/rejected": -0.5669963955879211,
|
| 1032 |
+
"step": 630,
|
| 1033 |
+
"use_label": 14157.275390625
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 0.67,
|
| 1037 |
+
"learning_rate": 3.6670547147846336e-06,
|
| 1038 |
+
"logits/chosen": -2.8346762657165527,
|
| 1039 |
+
"logits/rejected": -2.7844390869140625,
|
| 1040 |
+
"logps/chosen": -263.27545166015625,
|
| 1041 |
+
"logps/rejected": -244.00167846679688,
|
| 1042 |
+
"loss": 0.6712,
|
| 1043 |
+
"pred_label": 5966.8251953125,
|
| 1044 |
+
"rewards/accuracies": 0.715624988079071,
|
| 1045 |
+
"rewards/chosen": 0.057781945914030075,
|
| 1046 |
+
"rewards/margins": 0.6433111429214478,
|
| 1047 |
+
"rewards/rejected": -0.5855292081832886,
|
| 1048 |
+
"step": 640,
|
| 1049 |
+
"use_label": 14355.1748046875
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 0.68,
|
| 1053 |
+
"learning_rate": 3.5506402793946454e-06,
|
| 1054 |
+
"logits/chosen": -2.7912142276763916,
|
| 1055 |
+
"logits/rejected": -2.781743049621582,
|
| 1056 |
+
"logps/chosen": -287.99041748046875,
|
| 1057 |
+
"logps/rejected": -280.1361389160156,
|
| 1058 |
+
"loss": 0.6597,
|
| 1059 |
+
"pred_label": 6101.3876953125,
|
| 1060 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1061 |
+
"rewards/chosen": 0.06794126331806183,
|
| 1062 |
+
"rewards/margins": 0.7323317527770996,
|
| 1063 |
+
"rewards/rejected": -0.6643904447555542,
|
| 1064 |
+
"step": 650,
|
| 1065 |
+
"use_label": 14540.6123046875
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 0.69,
|
| 1069 |
+
"learning_rate": 3.434225844004657e-06,
|
| 1070 |
+
"logits/chosen": -2.828040599822998,
|
| 1071 |
+
"logits/rejected": -2.811732530593872,
|
| 1072 |
+
"logps/chosen": -277.9667663574219,
|
| 1073 |
+
"logps/rejected": -246.4541015625,
|
| 1074 |
+
"loss": 0.659,
|
| 1075 |
+
"pred_label": 6221.6875,
|
| 1076 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1077 |
+
"rewards/chosen": -0.013715244829654694,
|
| 1078 |
+
"rewards/margins": 0.5462251901626587,
|
| 1079 |
+
"rewards/rejected": -0.5599404573440552,
|
| 1080 |
+
"step": 660,
|
| 1081 |
+
"use_label": 14740.3125
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.7,
|
| 1085 |
+
"learning_rate": 3.3178114086146686e-06,
|
| 1086 |
+
"logits/chosen": -2.8130388259887695,
|
| 1087 |
+
"logits/rejected": -2.787423610687256,
|
| 1088 |
+
"logps/chosen": -279.0962829589844,
|
| 1089 |
+
"logps/rejected": -251.55233764648438,
|
| 1090 |
+
"loss": 0.6513,
|
| 1091 |
+
"pred_label": 6336.22509765625,
|
| 1092 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1093 |
+
"rewards/chosen": 0.028571803122758865,
|
| 1094 |
+
"rewards/margins": 0.6603085398674011,
|
| 1095 |
+
"rewards/rejected": -0.6317366361618042,
|
| 1096 |
+
"step": 670,
|
| 1097 |
+
"use_label": 14945.775390625
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 0.71,
|
| 1101 |
+
"learning_rate": 3.20139697322468e-06,
|
| 1102 |
+
"logits/chosen": -2.8303864002227783,
|
| 1103 |
+
"logits/rejected": -2.8300769329071045,
|
| 1104 |
+
"logps/chosen": -299.0453186035156,
|
| 1105 |
+
"logps/rejected": -273.7367858886719,
|
| 1106 |
+
"loss": 0.6679,
|
| 1107 |
+
"pred_label": 6456.0498046875,
|
| 1108 |
+
"rewards/accuracies": 0.6656249761581421,
|
| 1109 |
+
"rewards/chosen": 0.049853820353746414,
|
| 1110 |
+
"rewards/margins": 0.5102404356002808,
|
| 1111 |
+
"rewards/rejected": -0.46038660407066345,
|
| 1112 |
+
"step": 680,
|
| 1113 |
+
"use_label": 15145.9501953125
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 0.72,
|
| 1117 |
+
"learning_rate": 3.0849825378346914e-06,
|
| 1118 |
+
"logits/chosen": -2.812546968460083,
|
| 1119 |
+
"logits/rejected": -2.7978675365448,
|
| 1120 |
+
"logps/chosen": -281.13677978515625,
|
| 1121 |
+
"logps/rejected": -241.3180694580078,
|
| 1122 |
+
"loss": 0.6575,
|
| 1123 |
+
"pred_label": 6579.16259765625,
|
| 1124 |
+
"rewards/accuracies": 0.746874988079071,
|
| 1125 |
+
"rewards/chosen": 0.095861054956913,
|
| 1126 |
+
"rewards/margins": 0.6724832653999329,
|
| 1127 |
+
"rewards/rejected": -0.5766221880912781,
|
| 1128 |
+
"step": 690,
|
| 1129 |
+
"use_label": 15342.837890625
|
| 1130 |
},
|
| 1131 |
{
|
| 1132 |
"epoch": 0.73,
|
| 1133 |
"learning_rate": 2.9685681024447033e-06,
|
| 1134 |
+
"logits/chosen": -2.8043007850646973,
|
| 1135 |
+
"logits/rejected": -2.7992827892303467,
|
| 1136 |
+
"logps/chosen": -268.5847473144531,
|
| 1137 |
+
"logps/rejected": -251.1324462890625,
|
| 1138 |
+
"loss": 0.6649,
|
| 1139 |
+
"pred_label": 6708.8623046875,
|
| 1140 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1141 |
+
"rewards/chosen": 0.051186300814151764,
|
| 1142 |
+
"rewards/margins": 0.675644040107727,
|
| 1143 |
+
"rewards/rejected": -0.6244576573371887,
|
| 1144 |
"step": 700,
|
| 1145 |
+
"use_label": 15533.1376953125
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 0.74,
|
| 1149 |
+
"learning_rate": 2.852153667054715e-06,
|
| 1150 |
+
"logits/chosen": -2.8468000888824463,
|
| 1151 |
+
"logits/rejected": -2.833977222442627,
|
| 1152 |
+
"logps/chosen": -274.0541076660156,
|
| 1153 |
+
"logps/rejected": -243.0079345703125,
|
| 1154 |
+
"loss": 0.6775,
|
| 1155 |
+
"pred_label": 6838.0126953125,
|
| 1156 |
+
"rewards/accuracies": 0.715624988079071,
|
| 1157 |
+
"rewards/chosen": 0.1383003294467926,
|
| 1158 |
+
"rewards/margins": 0.6733654737472534,
|
| 1159 |
+
"rewards/rejected": -0.535065233707428,
|
| 1160 |
+
"step": 710,
|
| 1161 |
+
"use_label": 15723.9873046875
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 0.75,
|
| 1165 |
+
"learning_rate": 2.735739231664727e-06,
|
| 1166 |
+
"logits/chosen": -2.7949423789978027,
|
| 1167 |
+
"logits/rejected": -2.8083877563476562,
|
| 1168 |
+
"logps/chosen": -275.5536804199219,
|
| 1169 |
+
"logps/rejected": -253.4557342529297,
|
| 1170 |
+
"loss": 0.6622,
|
| 1171 |
+
"pred_label": 6984.41259765625,
|
| 1172 |
+
"rewards/accuracies": 0.7593749761581421,
|
| 1173 |
+
"rewards/chosen": -0.009852488525211811,
|
| 1174 |
+
"rewards/margins": 0.679956316947937,
|
| 1175 |
+
"rewards/rejected": -0.6898088455200195,
|
| 1176 |
+
"step": 720,
|
| 1177 |
+
"use_label": 15897.587890625
|
| 1178 |
+
},
|
| 1179 |
+
{
|
| 1180 |
+
"epoch": 0.76,
|
| 1181 |
+
"learning_rate": 2.6193247962747383e-06,
|
| 1182 |
+
"logits/chosen": -2.8305153846740723,
|
| 1183 |
+
"logits/rejected": -2.8256735801696777,
|
| 1184 |
+
"logps/chosen": -292.13153076171875,
|
| 1185 |
+
"logps/rejected": -250.5773468017578,
|
| 1186 |
+
"loss": 0.658,
|
| 1187 |
+
"pred_label": 7112.52490234375,
|
| 1188 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1189 |
+
"rewards/chosen": 0.06651445478200912,
|
| 1190 |
+
"rewards/margins": 0.6194807887077332,
|
| 1191 |
+
"rewards/rejected": -0.5529662370681763,
|
| 1192 |
+
"step": 730,
|
| 1193 |
+
"use_label": 16089.474609375
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.77,
|
| 1197 |
+
"learning_rate": 2.5029103608847497e-06,
|
| 1198 |
+
"logits/chosen": -2.8071470260620117,
|
| 1199 |
+
"logits/rejected": -2.826627016067505,
|
| 1200 |
+
"logps/chosen": -299.8040771484375,
|
| 1201 |
+
"logps/rejected": -278.0760192871094,
|
| 1202 |
+
"loss": 0.6779,
|
| 1203 |
+
"pred_label": 7236.35009765625,
|
| 1204 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1205 |
+
"rewards/chosen": 0.08306702226400375,
|
| 1206 |
+
"rewards/margins": 0.7367380261421204,
|
| 1207 |
+
"rewards/rejected": -0.6536709070205688,
|
| 1208 |
+
"step": 740,
|
| 1209 |
+
"use_label": 16285.650390625
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.79,
|
| 1213 |
+
"learning_rate": 2.3864959254947616e-06,
|
| 1214 |
+
"logits/chosen": -2.8130180835723877,
|
| 1215 |
+
"logits/rejected": -2.7900407314300537,
|
| 1216 |
+
"logps/chosen": -262.660888671875,
|
| 1217 |
+
"logps/rejected": -266.95574951171875,
|
| 1218 |
+
"loss": 0.6657,
|
| 1219 |
+
"pred_label": 7375.4375,
|
| 1220 |
+
"rewards/accuracies": 0.746874988079071,
|
| 1221 |
+
"rewards/chosen": 0.02680896781384945,
|
| 1222 |
+
"rewards/margins": 0.7038629651069641,
|
| 1223 |
+
"rewards/rejected": -0.6770539879798889,
|
| 1224 |
+
"step": 750,
|
| 1225 |
+
"use_label": 16466.5625
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 0.8,
|
| 1229 |
+
"learning_rate": 2.2700814901047734e-06,
|
| 1230 |
+
"logits/chosen": -2.8128421306610107,
|
| 1231 |
+
"logits/rejected": -2.8296356201171875,
|
| 1232 |
+
"logps/chosen": -295.57342529296875,
|
| 1233 |
+
"logps/rejected": -269.14324951171875,
|
| 1234 |
+
"loss": 0.6628,
|
| 1235 |
+
"pred_label": 7496.8623046875,
|
| 1236 |
+
"rewards/accuracies": 0.7406250238418579,
|
| 1237 |
+
"rewards/chosen": 0.08986032009124756,
|
| 1238 |
+
"rewards/margins": 0.7447289228439331,
|
| 1239 |
+
"rewards/rejected": -0.6548686027526855,
|
| 1240 |
+
"step": 760,
|
| 1241 |
+
"use_label": 16665.13671875
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.81,
|
| 1245 |
+
"learning_rate": 2.153667054714785e-06,
|
| 1246 |
+
"logits/chosen": -2.8838798999786377,
|
| 1247 |
+
"logits/rejected": -2.8680777549743652,
|
| 1248 |
+
"logps/chosen": -316.2966003417969,
|
| 1249 |
+
"logps/rejected": -290.4427490234375,
|
| 1250 |
+
"loss": 0.6869,
|
| 1251 |
+
"pred_label": 7625.2373046875,
|
| 1252 |
+
"rewards/accuracies": 0.7093750238418579,
|
| 1253 |
+
"rewards/chosen": 0.10155443847179413,
|
| 1254 |
+
"rewards/margins": 0.6839195489883423,
|
| 1255 |
+
"rewards/rejected": -0.5823651552200317,
|
| 1256 |
+
"step": 770,
|
| 1257 |
+
"use_label": 16856.76171875
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 0.82,
|
| 1261 |
+
"learning_rate": 2.0372526193247966e-06,
|
| 1262 |
+
"logits/chosen": -2.8313162326812744,
|
| 1263 |
+
"logits/rejected": -2.8232970237731934,
|
| 1264 |
+
"logps/chosen": -285.3306579589844,
|
| 1265 |
+
"logps/rejected": -265.0252990722656,
|
| 1266 |
+
"loss": 0.6586,
|
| 1267 |
+
"pred_label": 7761.66259765625,
|
| 1268 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1269 |
+
"rewards/chosen": 0.05353979393839836,
|
| 1270 |
+
"rewards/margins": 0.6505471467971802,
|
| 1271 |
+
"rewards/rejected": -0.5970073342323303,
|
| 1272 |
+
"step": 780,
|
| 1273 |
+
"use_label": 17040.337890625
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 0.83,
|
| 1277 |
+
"learning_rate": 1.920838183934808e-06,
|
| 1278 |
+
"logits/chosen": -2.834827423095703,
|
| 1279 |
+
"logits/rejected": -2.8339152336120605,
|
| 1280 |
+
"logps/chosen": -263.69171142578125,
|
| 1281 |
+
"logps/rejected": -264.4107666015625,
|
| 1282 |
+
"loss": 0.6639,
|
| 1283 |
+
"pred_label": 7894.5,
|
| 1284 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1285 |
+
"rewards/chosen": -0.022974440827965736,
|
| 1286 |
+
"rewards/margins": 0.6350787878036499,
|
| 1287 |
+
"rewards/rejected": -0.6580532193183899,
|
| 1288 |
+
"step": 790,
|
| 1289 |
+
"use_label": 17227.5
|
| 1290 |
},
|
| 1291 |
{
|
| 1292 |
"epoch": 0.84,
|
| 1293 |
"learning_rate": 1.8044237485448196e-06,
|
| 1294 |
+
"logits/chosen": -2.804044485092163,
|
| 1295 |
+
"logits/rejected": -2.772721529006958,
|
| 1296 |
+
"logps/chosen": -286.9491271972656,
|
| 1297 |
+
"logps/rejected": -278.8674011230469,
|
| 1298 |
+
"loss": 0.6627,
|
| 1299 |
+
"pred_label": 8027.97509765625,
|
| 1300 |
+
"rewards/accuracies": 0.75,
|
| 1301 |
+
"rewards/chosen": 0.03967855125665665,
|
| 1302 |
+
"rewards/margins": 0.6997824907302856,
|
| 1303 |
+
"rewards/rejected": -0.6601039171218872,
|
| 1304 |
"step": 800,
|
| 1305 |
+
"use_label": 17414.025390625
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.85,
|
| 1309 |
+
"learning_rate": 1.6880093131548315e-06,
|
| 1310 |
+
"logits/chosen": -2.8020050525665283,
|
| 1311 |
+
"logits/rejected": -2.8072879314422607,
|
| 1312 |
+
"logps/chosen": -281.62799072265625,
|
| 1313 |
+
"logps/rejected": -244.6085205078125,
|
| 1314 |
+
"loss": 0.6643,
|
| 1315 |
+
"pred_label": 8165.77490234375,
|
| 1316 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1317 |
+
"rewards/chosen": 0.05538611859083176,
|
| 1318 |
+
"rewards/margins": 0.6941258907318115,
|
| 1319 |
+
"rewards/rejected": -0.6387397050857544,
|
| 1320 |
+
"step": 810,
|
| 1321 |
+
"use_label": 17596.224609375
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 0.86,
|
| 1325 |
+
"learning_rate": 1.5715948777648429e-06,
|
| 1326 |
+
"logits/chosen": -2.816628932952881,
|
| 1327 |
+
"logits/rejected": -2.8157966136932373,
|
| 1328 |
+
"logps/chosen": -295.4468078613281,
|
| 1329 |
+
"logps/rejected": -261.2769775390625,
|
| 1330 |
+
"loss": 0.673,
|
| 1331 |
+
"pred_label": 8300.3623046875,
|
| 1332 |
+
"rewards/accuracies": 0.721875011920929,
|
| 1333 |
+
"rewards/chosen": 0.07940875738859177,
|
| 1334 |
+
"rewards/margins": 0.7328697443008423,
|
| 1335 |
+
"rewards/rejected": -0.6534609794616699,
|
| 1336 |
+
"step": 820,
|
| 1337 |
+
"use_label": 17781.63671875
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 0.87,
|
| 1341 |
+
"learning_rate": 1.4551804423748547e-06,
|
| 1342 |
+
"logits/chosen": -2.825873851776123,
|
| 1343 |
+
"logits/rejected": -2.812453508377075,
|
| 1344 |
+
"logps/chosen": -259.86407470703125,
|
| 1345 |
+
"logps/rejected": -299.1662902832031,
|
| 1346 |
+
"loss": 0.6584,
|
| 1347 |
+
"pred_label": 8431.2373046875,
|
| 1348 |
+
"rewards/accuracies": 0.71875,
|
| 1349 |
+
"rewards/chosen": -0.014992868527770042,
|
| 1350 |
+
"rewards/margins": 0.7412894368171692,
|
| 1351 |
+
"rewards/rejected": -0.7562823295593262,
|
| 1352 |
+
"step": 830,
|
| 1353 |
+
"use_label": 17970.76171875
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.88,
|
| 1357 |
+
"learning_rate": 1.3387660069848663e-06,
|
| 1358 |
+
"logits/chosen": -2.8066563606262207,
|
| 1359 |
+
"logits/rejected": -2.819666624069214,
|
| 1360 |
+
"logps/chosen": -279.8387451171875,
|
| 1361 |
+
"logps/rejected": -275.2579650878906,
|
| 1362 |
+
"loss": 0.673,
|
| 1363 |
+
"pred_label": 8558.837890625,
|
| 1364 |
+
"rewards/accuracies": 0.671875,
|
| 1365 |
+
"rewards/chosen": 0.007683378644287586,
|
| 1366 |
+
"rewards/margins": 0.5629515647888184,
|
| 1367 |
+
"rewards/rejected": -0.5552681088447571,
|
| 1368 |
+
"step": 840,
|
| 1369 |
+
"use_label": 18163.162109375
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 0.89,
|
| 1373 |
+
"learning_rate": 1.222351571594878e-06,
|
| 1374 |
+
"logits/chosen": -2.817474365234375,
|
| 1375 |
+
"logits/rejected": -2.848310947418213,
|
| 1376 |
+
"logps/chosen": -273.4720764160156,
|
| 1377 |
+
"logps/rejected": -254.59060668945312,
|
| 1378 |
+
"loss": 0.6512,
|
| 1379 |
+
"pred_label": 8672.349609375,
|
| 1380 |
+
"rewards/accuracies": 0.7406250238418579,
|
| 1381 |
+
"rewards/chosen": 0.02653767168521881,
|
| 1382 |
+
"rewards/margins": 0.6314449906349182,
|
| 1383 |
+
"rewards/rejected": -0.6049073338508606,
|
| 1384 |
+
"step": 850,
|
| 1385 |
+
"use_label": 18369.650390625
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 0.9,
|
| 1389 |
+
"learning_rate": 1.1059371362048893e-06,
|
| 1390 |
+
"logits/chosen": -2.7932801246643066,
|
| 1391 |
+
"logits/rejected": -2.800811767578125,
|
| 1392 |
+
"logps/chosen": -294.1841735839844,
|
| 1393 |
+
"logps/rejected": -282.7134704589844,
|
| 1394 |
+
"loss": 0.6757,
|
| 1395 |
+
"pred_label": 8782.8125,
|
| 1396 |
+
"rewards/accuracies": 0.653124988079071,
|
| 1397 |
+
"rewards/chosen": -0.07215853035449982,
|
| 1398 |
+
"rewards/margins": 0.5342829823493958,
|
| 1399 |
+
"rewards/rejected": -0.6064414978027344,
|
| 1400 |
+
"step": 860,
|
| 1401 |
+
"use_label": 18579.1875
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 0.91,
|
| 1405 |
+
"learning_rate": 9.895227008149012e-07,
|
| 1406 |
+
"logits/chosen": -2.7996864318847656,
|
| 1407 |
+
"logits/rejected": -2.8061373233795166,
|
| 1408 |
+
"logps/chosen": -263.4325256347656,
|
| 1409 |
+
"logps/rejected": -266.0005798339844,
|
| 1410 |
+
"loss": 0.6697,
|
| 1411 |
+
"pred_label": 8905.3125,
|
| 1412 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1413 |
+
"rewards/chosen": 0.041508838534355164,
|
| 1414 |
+
"rewards/margins": 0.653960108757019,
|
| 1415 |
+
"rewards/rejected": -0.6124512553215027,
|
| 1416 |
+
"step": 870,
|
| 1417 |
+
"use_label": 18776.6875
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 0.92,
|
| 1421 |
+
"learning_rate": 8.731082654249128e-07,
|
| 1422 |
+
"logits/chosen": -2.8358547687530518,
|
| 1423 |
+
"logits/rejected": -2.820361614227295,
|
| 1424 |
+
"logps/chosen": -271.4715576171875,
|
| 1425 |
+
"logps/rejected": -266.07562255859375,
|
| 1426 |
+
"loss": 0.6717,
|
| 1427 |
+
"pred_label": 9048.2998046875,
|
| 1428 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1429 |
+
"rewards/chosen": 0.06802918016910553,
|
| 1430 |
+
"rewards/margins": 0.6590049862861633,
|
| 1431 |
+
"rewards/rejected": -0.5909757018089294,
|
| 1432 |
+
"step": 880,
|
| 1433 |
+
"use_label": 18953.69921875
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 0.93,
|
| 1437 |
+
"learning_rate": 7.566938300349244e-07,
|
| 1438 |
+
"logits/chosen": -2.8539083003997803,
|
| 1439 |
+
"logits/rejected": -2.8629262447357178,
|
| 1440 |
+
"logps/chosen": -264.7303771972656,
|
| 1441 |
+
"logps/rejected": -274.3258056640625,
|
| 1442 |
+
"loss": 0.6724,
|
| 1443 |
+
"pred_label": 9178.8251953125,
|
| 1444 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1445 |
+
"rewards/chosen": 0.018511313945055008,
|
| 1446 |
+
"rewards/margins": 0.5970341563224792,
|
| 1447 |
+
"rewards/rejected": -0.578522801399231,
|
| 1448 |
+
"step": 890,
|
| 1449 |
+
"use_label": 19143.17578125
|
| 1450 |
},
|
| 1451 |
{
|
| 1452 |
"epoch": 0.94,
|
| 1453 |
"learning_rate": 6.402793946449361e-07,
|
| 1454 |
+
"logits/chosen": -2.8083693981170654,
|
| 1455 |
+
"logits/rejected": -2.8180294036865234,
|
| 1456 |
+
"logps/chosen": -275.43511962890625,
|
| 1457 |
+
"logps/rejected": -261.9901428222656,
|
| 1458 |
+
"loss": 0.6688,
|
| 1459 |
+
"pred_label": 9309.8623046875,
|
| 1460 |
+
"rewards/accuracies": 0.7281249761581421,
|
| 1461 |
+
"rewards/chosen": 0.014594699256122112,
|
| 1462 |
+
"rewards/margins": 0.5891886949539185,
|
| 1463 |
+
"rewards/rejected": -0.5745939612388611,
|
| 1464 |
"step": 900,
|
| 1465 |
+
"use_label": 19332.13671875
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.95,
|
| 1469 |
+
"learning_rate": 5.238649592549476e-07,
|
| 1470 |
+
"logits/chosen": -2.8189098834991455,
|
| 1471 |
+
"logits/rejected": -2.8110015392303467,
|
| 1472 |
+
"logps/chosen": -302.4877014160156,
|
| 1473 |
+
"logps/rejected": -278.4508972167969,
|
| 1474 |
+
"loss": 0.6586,
|
| 1475 |
+
"pred_label": 9430.1748046875,
|
| 1476 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1477 |
+
"rewards/chosen": 0.05673975870013237,
|
| 1478 |
+
"rewards/margins": 0.5847252607345581,
|
| 1479 |
+
"rewards/rejected": -0.5279855132102966,
|
| 1480 |
+
"step": 910,
|
| 1481 |
+
"use_label": 19531.82421875
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 0.96,
|
| 1485 |
+
"learning_rate": 4.0745052386495924e-07,
|
| 1486 |
+
"logits/chosen": -2.805025100708008,
|
| 1487 |
+
"logits/rejected": -2.8007519245147705,
|
| 1488 |
+
"logps/chosen": -286.05731201171875,
|
| 1489 |
+
"logps/rejected": -270.36279296875,
|
| 1490 |
+
"loss": 0.6555,
|
| 1491 |
+
"pred_label": 9553.6376953125,
|
| 1492 |
+
"rewards/accuracies": 0.7406250238418579,
|
| 1493 |
+
"rewards/chosen": -0.026457080617547035,
|
| 1494 |
+
"rewards/margins": 0.6447644233703613,
|
| 1495 |
+
"rewards/rejected": -0.6712215542793274,
|
| 1496 |
+
"step": 920,
|
| 1497 |
+
"use_label": 19728.36328125
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 0.97,
|
| 1501 |
+
"learning_rate": 2.910360884749709e-07,
|
| 1502 |
+
"logits/chosen": -2.805614471435547,
|
| 1503 |
+
"logits/rejected": -2.798377513885498,
|
| 1504 |
+
"logps/chosen": -281.56890869140625,
|
| 1505 |
+
"logps/rejected": -268.83734130859375,
|
| 1506 |
+
"loss": 0.6677,
|
| 1507 |
+
"pred_label": 9692.5126953125,
|
| 1508 |
+
"rewards/accuracies": 0.721875011920929,
|
| 1509 |
+
"rewards/chosen": 0.1096206083893776,
|
| 1510 |
+
"rewards/margins": 0.8147541284561157,
|
| 1511 |
+
"rewards/rejected": -0.7051334977149963,
|
| 1512 |
+
"step": 930,
|
| 1513 |
+
"use_label": 19909.48828125
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 0.98,
|
| 1517 |
+
"learning_rate": 1.7462165308498255e-07,
|
| 1518 |
+
"logits/chosen": -2.8316524028778076,
|
| 1519 |
+
"logits/rejected": -2.833402156829834,
|
| 1520 |
+
"logps/chosen": -286.48126220703125,
|
| 1521 |
+
"logps/rejected": -278.9422302246094,
|
| 1522 |
+
"loss": 0.6568,
|
| 1523 |
+
"pred_label": 9830.4873046875,
|
| 1524 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1525 |
+
"rewards/chosen": -0.0075108022429049015,
|
| 1526 |
+
"rewards/margins": 0.649519145488739,
|
| 1527 |
+
"rewards/rejected": -0.6570299863815308,
|
| 1528 |
+
"step": 940,
|
| 1529 |
+
"use_label": 20091.51171875
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 0.99,
|
| 1533 |
+
"learning_rate": 5.8207217694994185e-08,
|
| 1534 |
+
"logits/chosen": -2.8337855339050293,
|
| 1535 |
+
"logits/rejected": -2.82015323638916,
|
| 1536 |
+
"logps/chosen": -271.77117919921875,
|
| 1537 |
+
"logps/rejected": -268.10894775390625,
|
| 1538 |
+
"loss": 0.6708,
|
| 1539 |
+
"pred_label": 9960.287109375,
|
| 1540 |
+
"rewards/accuracies": 0.6968749761581421,
|
| 1541 |
+
"rewards/chosen": -0.015800004824995995,
|
| 1542 |
+
"rewards/margins": 0.6547096371650696,
|
| 1543 |
+
"rewards/rejected": -0.6705096364021301,
|
| 1544 |
+
"step": 950,
|
| 1545 |
+
"use_label": 20281.712890625
|
| 1546 |
},
|
| 1547 |
{
|
| 1548 |
"epoch": 1.0,
|
| 1549 |
+
"eval_logits/chosen": -2.83455753326416,
|
| 1550 |
+
"eval_logits/rejected": -2.824216604232788,
|
| 1551 |
+
"eval_logps/chosen": -282.00970458984375,
|
| 1552 |
+
"eval_logps/rejected": -268.7319030761719,
|
| 1553 |
+
"eval_loss": 0.6687781810760498,
|
| 1554 |
+
"eval_pred_label": 10292.2041015625,
|
| 1555 |
+
"eval_rewards/accuracies": 0.7210000157356262,
|
| 1556 |
+
"eval_rewards/chosen": -0.0034403554163873196,
|
| 1557 |
+
"eval_rewards/margins": 0.6410075426101685,
|
| 1558 |
+
"eval_rewards/rejected": -0.6444479823112488,
|
| 1559 |
+
"eval_runtime": 855.4886,
|
| 1560 |
+
"eval_samples_per_second": 2.338,
|
| 1561 |
+
"eval_steps_per_second": 0.292,
|
| 1562 |
+
"eval_use_label": 20769.796875,
|
| 1563 |
"step": 955
|
| 1564 |
},
|
| 1565 |
{
|
| 1566 |
"epoch": 1.0,
|
| 1567 |
"step": 955,
|
| 1568 |
"total_flos": 0.0,
|
| 1569 |
+
"train_loss": 0.6626948830969046,
|
| 1570 |
+
"train_runtime": 47570.4937,
|
| 1571 |
+
"train_samples_per_second": 1.285,
|
| 1572 |
+
"train_steps_per_second": 0.02
|
| 1573 |
}
|
| 1574 |
],
|
| 1575 |
+
"logging_steps": 10,
|
| 1576 |
"max_steps": 955,
|
| 1577 |
"num_train_epochs": 1,
|
| 1578 |
"save_steps": 50,
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4792
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7596c88734ee5efa7881796813f15d453e7e29fcd0afd3f76fd123c7078f7a1f
|
| 3 |
size 4792
|