Commit
•
328542f
1
Parent(s):
a0b90b2
Model save
Browse files- README.md +17 -16
- all_results.json +6 -6
- model-00001-of-00003.safetensors +1 -1
- model-00002-of-00003.safetensors +1 -1
- model-00003-of-00003.safetensors +1 -1
- train_results.json +6 -6
- trainer_state.json +1521 -16
README.md
CHANGED
|
@@ -3,15 +3,10 @@ library_name: transformers
|
|
| 3 |
license: apache-2.0
|
| 4 |
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 5 |
tags:
|
| 6 |
-
- alignment-handbook
|
| 7 |
-
- trl
|
| 8 |
-
- dpo
|
| 9 |
-
- generated_from_trainer
|
| 10 |
- trl
|
| 11 |
- dpo
|
|
|
|
| 12 |
- generated_from_trainer
|
| 13 |
-
datasets:
|
| 14 |
-
- data/rlced_conifer
|
| 15 |
model-index:
|
| 16 |
- name: rlced-conifer-zephyr-7b-dpo-2e
|
| 17 |
results: []
|
|
@@ -22,17 +17,17 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 22 |
|
| 23 |
# rlced-conifer-zephyr-7b-dpo-2e
|
| 24 |
|
| 25 |
-
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on
|
| 26 |
It achieves the following results on the evaluation set:
|
| 27 |
-
- Loss: 0.
|
| 28 |
-
- Rewards/chosen:
|
| 29 |
-
- Rewards/rejected:
|
| 30 |
-
- Rewards/accuracies: 0.
|
| 31 |
-
- Rewards/margins:
|
| 32 |
-
- Logps/rejected: -
|
| 33 |
-
- Logps/chosen: -
|
| 34 |
-
- Logits/rejected:
|
| 35 |
-
- Logits/chosen: -
|
| 36 |
|
| 37 |
## Model description
|
| 38 |
|
|
@@ -67,6 +62,12 @@ The following hyperparameters were used during training:
|
|
| 67 |
|
| 68 |
### Training results
|
| 69 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 70 |
|
| 71 |
|
| 72 |
### Framework versions
|
|
|
|
| 3 |
license: apache-2.0
|
| 4 |
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 5 |
tags:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
- trl
|
| 7 |
- dpo
|
| 8 |
+
- alignment-handbook
|
| 9 |
- generated_from_trainer
|
|
|
|
|
|
|
| 10 |
model-index:
|
| 11 |
- name: rlced-conifer-zephyr-7b-dpo-2e
|
| 12 |
results: []
|
|
|
|
| 17 |
|
| 18 |
# rlced-conifer-zephyr-7b-dpo-2e
|
| 19 |
|
| 20 |
+
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on an unknown dataset.
|
| 21 |
It achieves the following results on the evaluation set:
|
| 22 |
+
- Loss: 0.1578
|
| 23 |
+
- Rewards/chosen: -8.8688
|
| 24 |
+
- Rewards/rejected: -21.7065
|
| 25 |
+
- Rewards/accuracies: 0.9203
|
| 26 |
+
- Rewards/margins: 12.8376
|
| 27 |
+
- Logps/rejected: -2616.5852
|
| 28 |
+
- Logps/chosen: -1310.2896
|
| 29 |
+
- Logits/rejected: 3.0165
|
| 30 |
+
- Logits/chosen: -0.1724
|
| 31 |
|
| 32 |
## Model description
|
| 33 |
|
|
|
|
| 62 |
|
| 63 |
### Training results
|
| 64 |
|
| 65 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 66 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 67 |
+
| 0.1586 | 0.4982 | 240 | 0.1689 | -2.6868 | -8.8615 | 0.9252 | 6.1746 | -1332.0890 | -692.0928 | 1.5975 | -0.1975 |
|
| 68 |
+
| 0.1408 | 0.9964 | 480 | 0.1482 | -6.5979 | -15.9265 | 0.9326 | 9.3286 | -2038.5934 | -1083.2030 | 3.3424 | 0.5015 |
|
| 69 |
+
| 0.0852 | 1.4946 | 720 | 0.1644 | -10.0141 | -23.7065 | 0.9228 | 13.6924 | -2816.5886 | -1424.8193 | 3.4873 | 0.0636 |
|
| 70 |
+
| 0.0743 | 1.9927 | 960 | 0.1578 | -8.8688 | -21.7065 | 0.9203 | 12.8376 | -2616.5852 | -1310.2896 | 3.0165 | -0.1724 |
|
| 71 |
|
| 72 |
|
| 73 |
### Framework versions
|
all_results.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
"eval_logits/chosen": -2.8249309062957764,
|
| 4 |
"eval_logits/rejected": -2.6878879070281982,
|
| 5 |
"eval_logps/chosen": -401.2567138671875,
|
|
@@ -14,9 +14,9 @@
|
|
| 14 |
"eval_samples_per_second": 17.099,
|
| 15 |
"eval_steps_per_second": 0.342,
|
| 16 |
"total_flos": 0.0,
|
| 17 |
-
"train_loss": 0.
|
| 18 |
-
"train_runtime":
|
| 19 |
-
"train_samples":
|
| 20 |
-
"train_samples_per_second":
|
| 21 |
-
"train_steps_per_second": 0.
|
| 22 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 1.996886351842242,
|
| 3 |
"eval_logits/chosen": -2.8249309062957764,
|
| 4 |
"eval_logits/rejected": -2.6878879070281982,
|
| 5 |
"eval_logps/chosen": -401.2567138671875,
|
|
|
|
| 14 |
"eval_samples_per_second": 17.099,
|
| 15 |
"eval_steps_per_second": 0.342,
|
| 16 |
"total_flos": 0.0,
|
| 17 |
+
"train_loss": 0.1585804910288655,
|
| 18 |
+
"train_runtime": 28225.2793,
|
| 19 |
+
"train_samples": 123309,
|
| 20 |
+
"train_samples_per_second": 8.737,
|
| 21 |
+
"train_steps_per_second": 0.034
|
| 22 |
}
|
model-00001-of-00003.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4943162336
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:29a7362c07defb3a6b1f6273dbade7b018c7b81675e8d05f6479514991110160
|
| 3 |
size 4943162336
|
model-00002-of-00003.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4999819336
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:58b78cf610b7fd116f39cfadc175aa49f426caf39910c0bb462854aed97c8ad7
|
| 3 |
size 4999819336
|
model-00003-of-00003.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4540516344
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8117a1f752f54256f6f0a694ad3a933fe782666b10bada44a18eefee6b2d2f7f
|
| 3 |
size 4540516344
|
train_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
"total_flos": 0.0,
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples":
|
| 7 |
-
"train_samples_per_second":
|
| 8 |
-
"train_steps_per_second": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 1.996886351842242,
|
| 3 |
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.1585804910288655,
|
| 5 |
+
"train_runtime": 28225.2793,
|
| 6 |
+
"train_samples": 123309,
|
| 7 |
+
"train_samples_per_second": 8.737,
|
| 8 |
+
"train_steps_per_second": 0.034
|
| 9 |
}
|
trainer_state.json
CHANGED
|
@@ -1,21 +1,22 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 240,
|
| 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":
|
| 13 |
-
"
|
| 14 |
-
"
|
| 15 |
-
"logits/
|
| 16 |
-
"
|
| 17 |
-
"logps/
|
| 18 |
-
"
|
|
|
|
| 19 |
"rewards/accuracies": 0.0,
|
| 20 |
"rewards/chosen": 0.0,
|
| 21 |
"rewards/margins": 0.0,
|
|
@@ -23,17 +24,1521 @@
|
|
| 23 |
"step": 1
|
| 24 |
},
|
| 25 |
{
|
| 26 |
-
"epoch":
|
| 27 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 28 |
"total_flos": 0.0,
|
| 29 |
-
"train_loss": 0.
|
| 30 |
-
"train_runtime":
|
| 31 |
-
"train_samples_per_second":
|
| 32 |
-
"train_steps_per_second": 0.
|
| 33 |
}
|
| 34 |
],
|
| 35 |
"logging_steps": 10,
|
| 36 |
-
"max_steps":
|
| 37 |
"num_input_tokens_seen": 0,
|
| 38 |
"num_train_epochs": 2,
|
| 39 |
"save_steps": 240,
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.996886351842242,
|
| 5 |
"eval_steps": 240,
|
| 6 |
+
"global_step": 962,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.0020757654385054488,
|
| 13 |
+
"grad_norm": 7.452356109448157,
|
| 14 |
+
"learning_rate": 5.154639175257731e-09,
|
| 15 |
+
"logits/chosen": -2.730942726135254,
|
| 16 |
+
"logits/rejected": -2.654609203338623,
|
| 17 |
+
"logps/chosen": -350.489990234375,
|
| 18 |
+
"logps/rejected": -325.546875,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
"rewards/accuracies": 0.0,
|
| 21 |
"rewards/chosen": 0.0,
|
| 22 |
"rewards/margins": 0.0,
|
|
|
|
| 24 |
"step": 1
|
| 25 |
},
|
| 26 |
{
|
| 27 |
+
"epoch": 0.02075765438505449,
|
| 28 |
+
"grad_norm": 7.866099026608714,
|
| 29 |
+
"learning_rate": 5.154639175257731e-08,
|
| 30 |
+
"logits/chosen": -2.7327775955200195,
|
| 31 |
+
"logits/rejected": -2.734964609146118,
|
| 32 |
+
"logps/chosen": -366.4884033203125,
|
| 33 |
+
"logps/rejected": -412.2764892578125,
|
| 34 |
+
"loss": 0.6931,
|
| 35 |
+
"rewards/accuracies": 0.4826388955116272,
|
| 36 |
+
"rewards/chosen": 6.40038269921206e-05,
|
| 37 |
+
"rewards/margins": 0.00040091515984386206,
|
| 38 |
+
"rewards/rejected": -0.00033691132557578385,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.04151530877010898,
|
| 43 |
+
"grad_norm": 7.559166611883559,
|
| 44 |
+
"learning_rate": 1.0309278350515462e-07,
|
| 45 |
+
"logits/chosen": -2.717351198196411,
|
| 46 |
+
"logits/rejected": -2.69411563873291,
|
| 47 |
+
"logps/chosen": -378.9507141113281,
|
| 48 |
+
"logps/rejected": -404.0965270996094,
|
| 49 |
+
"loss": 0.6921,
|
| 50 |
+
"rewards/accuracies": 0.59375,
|
| 51 |
+
"rewards/chosen": 0.00041852169670164585,
|
| 52 |
+
"rewards/margins": 0.0022167633287608624,
|
| 53 |
+
"rewards/rejected": -0.0017982417484745383,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.062272963155163466,
|
| 58 |
+
"grad_norm": 7.615638482493831,
|
| 59 |
+
"learning_rate": 1.5463917525773197e-07,
|
| 60 |
+
"logits/chosen": -2.719243288040161,
|
| 61 |
+
"logits/rejected": -2.7044143676757812,
|
| 62 |
+
"logps/chosen": -365.67889404296875,
|
| 63 |
+
"logps/rejected": -388.8330993652344,
|
| 64 |
+
"loss": 0.6864,
|
| 65 |
+
"rewards/accuracies": 0.793749988079071,
|
| 66 |
+
"rewards/chosen": 0.004637600388377905,
|
| 67 |
+
"rewards/margins": 0.013907420448958874,
|
| 68 |
+
"rewards/rejected": -0.009269820526242256,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.08303061754021795,
|
| 73 |
+
"grad_norm": 7.7707957616247505,
|
| 74 |
+
"learning_rate": 2.0618556701030925e-07,
|
| 75 |
+
"logits/chosen": -2.7188639640808105,
|
| 76 |
+
"logits/rejected": -2.6840219497680664,
|
| 77 |
+
"logps/chosen": -350.5911865234375,
|
| 78 |
+
"logps/rejected": -365.0216979980469,
|
| 79 |
+
"loss": 0.6693,
|
| 80 |
+
"rewards/accuracies": 0.871874988079071,
|
| 81 |
+
"rewards/chosen": 0.027326706796884537,
|
| 82 |
+
"rewards/margins": 0.049114055931568146,
|
| 83 |
+
"rewards/rejected": -0.02178734540939331,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.10378827192527244,
|
| 88 |
+
"grad_norm": 8.90458099154434,
|
| 89 |
+
"learning_rate": 2.5773195876288655e-07,
|
| 90 |
+
"logits/chosen": -2.7294702529907227,
|
| 91 |
+
"logits/rejected": -2.7243564128875732,
|
| 92 |
+
"logps/chosen": -358.9928894042969,
|
| 93 |
+
"logps/rejected": -397.3983459472656,
|
| 94 |
+
"loss": 0.63,
|
| 95 |
+
"rewards/accuracies": 0.862500011920929,
|
| 96 |
+
"rewards/chosen": 0.07257182896137238,
|
| 97 |
+
"rewards/margins": 0.14314065873622894,
|
| 98 |
+
"rewards/rejected": -0.07056883722543716,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.12454592631032693,
|
| 103 |
+
"grad_norm": 9.98989696476554,
|
| 104 |
+
"learning_rate": 3.0927835051546394e-07,
|
| 105 |
+
"logits/chosen": -2.6951613426208496,
|
| 106 |
+
"logits/rejected": -2.678609848022461,
|
| 107 |
+
"logps/chosen": -348.3409729003906,
|
| 108 |
+
"logps/rejected": -426.9418029785156,
|
| 109 |
+
"loss": 0.5482,
|
| 110 |
+
"rewards/accuracies": 0.878125011920929,
|
| 111 |
+
"rewards/chosen": -0.011659199371933937,
|
| 112 |
+
"rewards/margins": 0.36892449855804443,
|
| 113 |
+
"rewards/rejected": -0.3805837035179138,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.14530358069538143,
|
| 118 |
+
"grad_norm": 14.516753369132225,
|
| 119 |
+
"learning_rate": 3.608247422680412e-07,
|
| 120 |
+
"logits/chosen": -2.7372078895568848,
|
| 121 |
+
"logits/rejected": -2.6939454078674316,
|
| 122 |
+
"logps/chosen": -451.07293701171875,
|
| 123 |
+
"logps/rejected": -535.2464599609375,
|
| 124 |
+
"loss": 0.4282,
|
| 125 |
+
"rewards/accuracies": 0.856249988079071,
|
| 126 |
+
"rewards/chosen": -0.5447245836257935,
|
| 127 |
+
"rewards/margins": 0.7660267353057861,
|
| 128 |
+
"rewards/rejected": -1.3107513189315796,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.1660612350804359,
|
| 133 |
+
"grad_norm": 15.352248807358325,
|
| 134 |
+
"learning_rate": 4.123711340206185e-07,
|
| 135 |
+
"logits/chosen": -2.7028908729553223,
|
| 136 |
+
"logits/rejected": -2.6827831268310547,
|
| 137 |
+
"logps/chosen": -547.4126586914062,
|
| 138 |
+
"logps/rejected": -697.9911499023438,
|
| 139 |
+
"loss": 0.3596,
|
| 140 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 141 |
+
"rewards/chosen": -1.5639097690582275,
|
| 142 |
+
"rewards/margins": 1.3926727771759033,
|
| 143 |
+
"rewards/rejected": -2.9565823078155518,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.1868188894654904,
|
| 148 |
+
"grad_norm": 17.157617837316515,
|
| 149 |
+
"learning_rate": 4.639175257731959e-07,
|
| 150 |
+
"logits/chosen": -2.664567708969116,
|
| 151 |
+
"logits/rejected": -2.664062023162842,
|
| 152 |
+
"logps/chosen": -561.7244262695312,
|
| 153 |
+
"logps/rejected": -796.8342895507812,
|
| 154 |
+
"loss": 0.3035,
|
| 155 |
+
"rewards/accuracies": 0.878125011920929,
|
| 156 |
+
"rewards/chosen": -1.9646422863006592,
|
| 157 |
+
"rewards/margins": 2.149301052093506,
|
| 158 |
+
"rewards/rejected": -4.113943576812744,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.2075765438505449,
|
| 163 |
+
"grad_norm": 15.724313068127296,
|
| 164 |
+
"learning_rate": 4.999851606199919e-07,
|
| 165 |
+
"logits/chosen": -2.607901096343994,
|
| 166 |
+
"logits/rejected": -2.587402582168579,
|
| 167 |
+
"logps/chosen": -576.0076904296875,
|
| 168 |
+
"logps/rejected": -909.0657348632812,
|
| 169 |
+
"loss": 0.263,
|
| 170 |
+
"rewards/accuracies": 0.903124988079071,
|
| 171 |
+
"rewards/chosen": -2.0050268173217773,
|
| 172 |
+
"rewards/margins": 3.0768344402313232,
|
| 173 |
+
"rewards/rejected": -5.0818610191345215,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.2283341982355994,
|
| 178 |
+
"grad_norm": 22.849604961849657,
|
| 179 |
+
"learning_rate": 4.997213984244138e-07,
|
| 180 |
+
"logits/chosen": -2.399500608444214,
|
| 181 |
+
"logits/rejected": -2.2552199363708496,
|
| 182 |
+
"logps/chosen": -620.6517333984375,
|
| 183 |
+
"logps/rejected": -1053.1802978515625,
|
| 184 |
+
"loss": 0.2333,
|
| 185 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 186 |
+
"rewards/chosen": -2.491640329360962,
|
| 187 |
+
"rewards/margins": 4.089051246643066,
|
| 188 |
+
"rewards/rejected": -6.580691337585449,
|
| 189 |
+
"step": 110
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.24909185262065386,
|
| 193 |
+
"grad_norm": 15.418275685296619,
|
| 194 |
+
"learning_rate": 4.991282726678214e-07,
|
| 195 |
+
"logits/chosen": -2.1379497051239014,
|
| 196 |
+
"logits/rejected": -1.6612409353256226,
|
| 197 |
+
"logps/chosen": -636.1170654296875,
|
| 198 |
+
"logps/rejected": -1038.722412109375,
|
| 199 |
+
"loss": 0.2114,
|
| 200 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 201 |
+
"rewards/chosen": -2.4115161895751953,
|
| 202 |
+
"rewards/margins": 3.842604875564575,
|
| 203 |
+
"rewards/rejected": -6.254120826721191,
|
| 204 |
+
"step": 120
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.26984950700570837,
|
| 208 |
+
"grad_norm": 22.122903405439125,
|
| 209 |
+
"learning_rate": 4.982065656380468e-07,
|
| 210 |
+
"logits/chosen": -1.8730089664459229,
|
| 211 |
+
"logits/rejected": -1.1086432933807373,
|
| 212 |
+
"logps/chosen": -624.3699951171875,
|
| 213 |
+
"logps/rejected": -1083.5306396484375,
|
| 214 |
+
"loss": 0.1978,
|
| 215 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 216 |
+
"rewards/chosen": -2.4477767944335938,
|
| 217 |
+
"rewards/margins": 4.317242622375488,
|
| 218 |
+
"rewards/rejected": -6.76501989364624,
|
| 219 |
+
"step": 130
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.29060716139076287,
|
| 223 |
+
"grad_norm": 13.724914658926766,
|
| 224 |
+
"learning_rate": 4.969574929966689e-07,
|
| 225 |
+
"logits/chosen": -1.063720941543579,
|
| 226 |
+
"logits/rejected": 0.11310062557458878,
|
| 227 |
+
"logps/chosen": -654.7889404296875,
|
| 228 |
+
"logps/rejected": -1111.530517578125,
|
| 229 |
+
"loss": 0.2171,
|
| 230 |
+
"rewards/accuracies": 0.893750011920929,
|
| 231 |
+
"rewards/chosen": -2.8432438373565674,
|
| 232 |
+
"rewards/margins": 4.339225769042969,
|
| 233 |
+
"rewards/rejected": -7.182469844818115,
|
| 234 |
+
"step": 140
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.3113648157758173,
|
| 238 |
+
"grad_norm": 20.111379920464472,
|
| 239 |
+
"learning_rate": 4.953827021756488e-07,
|
| 240 |
+
"logits/chosen": -1.147434949874878,
|
| 241 |
+
"logits/rejected": 0.20881839096546173,
|
| 242 |
+
"logps/chosen": -609.7511596679688,
|
| 243 |
+
"logps/rejected": -1106.64990234375,
|
| 244 |
+
"loss": 0.1719,
|
| 245 |
+
"rewards/accuracies": 0.921875,
|
| 246 |
+
"rewards/chosen": -2.4953980445861816,
|
| 247 |
+
"rewards/margins": 4.772416591644287,
|
| 248 |
+
"rewards/rejected": -7.267814636230469,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.3321224701608718,
|
| 253 |
+
"grad_norm": 20.712950851230556,
|
| 254 |
+
"learning_rate": 4.93484270204492e-07,
|
| 255 |
+
"logits/chosen": -0.3996439576148987,
|
| 256 |
+
"logits/rejected": 0.9938270449638367,
|
| 257 |
+
"logps/chosen": -663.2901000976562,
|
| 258 |
+
"logps/rejected": -1266.5684814453125,
|
| 259 |
+
"loss": 0.1789,
|
| 260 |
+
"rewards/accuracies": 0.921875,
|
| 261 |
+
"rewards/chosen": -2.7938601970672607,
|
| 262 |
+
"rewards/margins": 5.7776288986206055,
|
| 263 |
+
"rewards/rejected": -8.571489334106445,
|
| 264 |
+
"step": 160
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.3528801245459263,
|
| 268 |
+
"grad_norm": 13.955815229008401,
|
| 269 |
+
"learning_rate": 4.91264700970804e-07,
|
| 270 |
+
"logits/chosen": -0.3857128620147705,
|
| 271 |
+
"logits/rejected": 1.0256187915802002,
|
| 272 |
+
"logps/chosen": -606.496337890625,
|
| 273 |
+
"logps/rejected": -1150.380126953125,
|
| 274 |
+
"loss": 0.1822,
|
| 275 |
+
"rewards/accuracies": 0.9156249761581421,
|
| 276 |
+
"rewards/chosen": -2.4765748977661133,
|
| 277 |
+
"rewards/margins": 5.2469162940979,
|
| 278 |
+
"rewards/rejected": -7.723490238189697,
|
| 279 |
+
"step": 170
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.3736377789309808,
|
| 283 |
+
"grad_norm": 19.91766661692759,
|
| 284 |
+
"learning_rate": 4.88726921917853e-07,
|
| 285 |
+
"logits/chosen": -0.7232545614242554,
|
| 286 |
+
"logits/rejected": 0.9214665293693542,
|
| 287 |
+
"logps/chosen": -608.4518432617188,
|
| 288 |
+
"logps/rejected": -1221.78955078125,
|
| 289 |
+
"loss": 0.1832,
|
| 290 |
+
"rewards/accuracies": 0.909375011920929,
|
| 291 |
+
"rewards/chosen": -2.418121337890625,
|
| 292 |
+
"rewards/margins": 5.937131881713867,
|
| 293 |
+
"rewards/rejected": -8.355253219604492,
|
| 294 |
+
"step": 180
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.39439543331603527,
|
| 298 |
+
"grad_norm": 16.024239064339824,
|
| 299 |
+
"learning_rate": 4.858742801834942e-07,
|
| 300 |
+
"logits/chosen": -0.32018381357192993,
|
| 301 |
+
"logits/rejected": 1.0246493816375732,
|
| 302 |
+
"logps/chosen": -667.6340942382812,
|
| 303 |
+
"logps/rejected": -1306.54833984375,
|
| 304 |
+
"loss": 0.1733,
|
| 305 |
+
"rewards/accuracies": 0.934374988079071,
|
| 306 |
+
"rewards/chosen": -2.8495688438415527,
|
| 307 |
+
"rewards/margins": 6.024672985076904,
|
| 308 |
+
"rewards/rejected": -8.874241828918457,
|
| 309 |
+
"step": 190
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.4151530877010898,
|
| 313 |
+
"grad_norm": 12.349261100548174,
|
| 314 |
+
"learning_rate": 4.827105381855496e-07,
|
| 315 |
+
"logits/chosen": 0.25890278816223145,
|
| 316 |
+
"logits/rejected": 1.507102370262146,
|
| 317 |
+
"logps/chosen": -618.0540161132812,
|
| 318 |
+
"logps/rejected": -1187.555419921875,
|
| 319 |
+
"loss": 0.1662,
|
| 320 |
+
"rewards/accuracies": 0.918749988079071,
|
| 321 |
+
"rewards/chosen": -2.504819393157959,
|
| 322 |
+
"rewards/margins": 5.406851291656494,
|
| 323 |
+
"rewards/rejected": -7.911670684814453,
|
| 324 |
+
"step": 200
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.4359107420861443,
|
| 328 |
+
"grad_norm": 13.66810823452425,
|
| 329 |
+
"learning_rate": 4.79239868659464e-07,
|
| 330 |
+
"logits/chosen": -0.016678806394338608,
|
| 331 |
+
"logits/rejected": 1.5381004810333252,
|
| 332 |
+
"logps/chosen": -641.5460815429688,
|
| 333 |
+
"logps/rejected": -1252.699951171875,
|
| 334 |
+
"loss": 0.1584,
|
| 335 |
+
"rewards/accuracies": 0.934374988079071,
|
| 336 |
+
"rewards/chosen": -2.724513530731201,
|
| 337 |
+
"rewards/margins": 5.9185709953308105,
|
| 338 |
+
"rewards/rejected": -8.643084526062012,
|
| 339 |
+
"step": 210
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.4566683964711988,
|
| 343 |
+
"grad_norm": 22.723399659159355,
|
| 344 |
+
"learning_rate": 4.7546684915478443e-07,
|
| 345 |
+
"logits/chosen": -0.5056034326553345,
|
| 346 |
+
"logits/rejected": 1.3583004474639893,
|
| 347 |
+
"logps/chosen": -627.7324829101562,
|
| 348 |
+
"logps/rejected": -1324.6092529296875,
|
| 349 |
+
"loss": 0.1696,
|
| 350 |
+
"rewards/accuracies": 0.9156249761581421,
|
| 351 |
+
"rewards/chosen": -2.6251590251922607,
|
| 352 |
+
"rewards/margins": 6.788819789886475,
|
| 353 |
+
"rewards/rejected": -9.413978576660156,
|
| 354 |
+
"step": 220
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.4774260508562532,
|
| 358 |
+
"grad_norm": 25.512034717506292,
|
| 359 |
+
"learning_rate": 4.7139645599771953e-07,
|
| 360 |
+
"logits/chosen": -0.05699120834469795,
|
| 361 |
+
"logits/rejected": 1.986132025718689,
|
| 362 |
+
"logps/chosen": -660.3882446289062,
|
| 363 |
+
"logps/rejected": -1403.30322265625,
|
| 364 |
+
"loss": 0.1656,
|
| 365 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 366 |
+
"rewards/chosen": -2.9933114051818848,
|
| 367 |
+
"rewards/margins": 7.2729973793029785,
|
| 368 |
+
"rewards/rejected": -10.266307830810547,
|
| 369 |
+
"step": 230
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.49818370524130773,
|
| 373 |
+
"grad_norm": 19.374014963474004,
|
| 374 |
+
"learning_rate": 4.6703405772774325e-07,
|
| 375 |
+
"logits/chosen": -0.2160125970840454,
|
| 376 |
+
"logits/rejected": 2.073151111602783,
|
| 377 |
+
"logps/chosen": -610.6514892578125,
|
| 378 |
+
"logps/rejected": -1350.219482421875,
|
| 379 |
+
"loss": 0.1586,
|
| 380 |
+
"rewards/accuracies": 0.9468749761581421,
|
| 381 |
+
"rewards/chosen": -2.5249383449554443,
|
| 382 |
+
"rewards/margins": 7.157525539398193,
|
| 383 |
+
"rewards/rejected": -9.682464599609375,
|
| 384 |
+
"step": 240
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.49818370524130773,
|
| 388 |
+
"eval_logits/chosen": -0.19751745462417603,
|
| 389 |
+
"eval_logits/rejected": 1.597546935081482,
|
| 390 |
+
"eval_logps/chosen": -692.0928344726562,
|
| 391 |
+
"eval_logps/rejected": -1332.0889892578125,
|
| 392 |
+
"eval_loss": 0.1689271330833435,
|
| 393 |
+
"eval_rewards/accuracies": 0.9252451062202454,
|
| 394 |
+
"eval_rewards/chosen": -2.6868460178375244,
|
| 395 |
+
"eval_rewards/margins": 6.174641132354736,
|
| 396 |
+
"eval_rewards/rejected": -8.861486434936523,
|
| 397 |
+
"eval_runtime": 297.8773,
|
| 398 |
+
"eval_samples_per_second": 21.791,
|
| 399 |
+
"eval_steps_per_second": 0.342,
|
| 400 |
+
"step": 240
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 0.5189413596263622,
|
| 404 |
+
"grad_norm": 12.036861238820864,
|
| 405 |
+
"learning_rate": 4.6238540801689896e-07,
|
| 406 |
+
"logits/chosen": -0.4214223325252533,
|
| 407 |
+
"logits/rejected": 1.8403441905975342,
|
| 408 |
+
"logps/chosen": -610.5186767578125,
|
| 409 |
+
"logps/rejected": -1277.1844482421875,
|
| 410 |
+
"loss": 0.1604,
|
| 411 |
+
"rewards/accuracies": 0.9437500238418579,
|
| 412 |
+
"rewards/chosen": -2.307227373123169,
|
| 413 |
+
"rewards/margins": 6.4557318687438965,
|
| 414 |
+
"rewards/rejected": -8.762959480285645,
|
| 415 |
+
"step": 250
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.5396990140114167,
|
| 419 |
+
"grad_norm": 20.090786603415435,
|
| 420 |
+
"learning_rate": 4.5745663808114316e-07,
|
| 421 |
+
"logits/chosen": -0.743881106376648,
|
| 422 |
+
"logits/rejected": 1.5020240545272827,
|
| 423 |
+
"logps/chosen": -657.1211547851562,
|
| 424 |
+
"logps/rejected": -1360.107666015625,
|
| 425 |
+
"loss": 0.1638,
|
| 426 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 427 |
+
"rewards/chosen": -2.839501142501831,
|
| 428 |
+
"rewards/margins": 6.807010650634766,
|
| 429 |
+
"rewards/rejected": -9.646512031555176,
|
| 430 |
+
"step": 260
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.5604566683964712,
|
| 434 |
+
"grad_norm": 14.890524731187542,
|
| 435 |
+
"learning_rate": 4.5225424859373684e-07,
|
| 436 |
+
"logits/chosen": -0.38103678822517395,
|
| 437 |
+
"logits/rejected": 1.8405994176864624,
|
| 438 |
+
"logps/chosen": -600.785888671875,
|
| 439 |
+
"logps/rejected": -1403.572998046875,
|
| 440 |
+
"loss": 0.1493,
|
| 441 |
+
"rewards/accuracies": 0.9468749761581421,
|
| 442 |
+
"rewards/chosen": -2.4715404510498047,
|
| 443 |
+
"rewards/margins": 7.713797569274902,
|
| 444 |
+
"rewards/rejected": -10.185338020324707,
|
| 445 |
+
"step": 270
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.5812143227815257,
|
| 449 |
+
"grad_norm": 15.737222130316104,
|
| 450 |
+
"learning_rate": 4.467851011113515e-07,
|
| 451 |
+
"logits/chosen": 0.3215886056423187,
|
| 452 |
+
"logits/rejected": 2.3104662895202637,
|
| 453 |
+
"logps/chosen": -629.4906005859375,
|
| 454 |
+
"logps/rejected": -1383.751953125,
|
| 455 |
+
"loss": 0.1571,
|
| 456 |
+
"rewards/accuracies": 0.925000011920929,
|
| 457 |
+
"rewards/chosen": -2.5350632667541504,
|
| 458 |
+
"rewards/margins": 7.314475059509277,
|
| 459 |
+
"rewards/rejected": -9.849536895751953,
|
| 460 |
+
"step": 280
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 0.6019719771665801,
|
| 464 |
+
"grad_norm": 23.328559288837457,
|
| 465 |
+
"learning_rate": 4.410564090241966e-07,
|
| 466 |
+
"logits/chosen": 0.4208109378814697,
|
| 467 |
+
"logits/rejected": 2.2348339557647705,
|
| 468 |
+
"logps/chosen": -675.2015991210938,
|
| 469 |
+
"logps/rejected": -1364.4970703125,
|
| 470 |
+
"loss": 0.1635,
|
| 471 |
+
"rewards/accuracies": 0.9156249761581421,
|
| 472 |
+
"rewards/chosen": -3.0044052600860596,
|
| 473 |
+
"rewards/margins": 6.818602085113525,
|
| 474 |
+
"rewards/rejected": -9.823007583618164,
|
| 475 |
+
"step": 290
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 0.6227296315516346,
|
| 479 |
+
"grad_norm": 14.02162805247057,
|
| 480 |
+
"learning_rate": 4.35075728042106e-07,
|
| 481 |
+
"logits/chosen": -0.0718420147895813,
|
| 482 |
+
"logits/rejected": 1.6788543462753296,
|
| 483 |
+
"logps/chosen": -612.244873046875,
|
| 484 |
+
"logps/rejected": -1253.4449462890625,
|
| 485 |
+
"loss": 0.1528,
|
| 486 |
+
"rewards/accuracies": 0.949999988079071,
|
| 487 |
+
"rewards/chosen": -2.3619751930236816,
|
| 488 |
+
"rewards/margins": 6.16156530380249,
|
| 489 |
+
"rewards/rejected": -8.523540496826172,
|
| 490 |
+
"step": 300
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.6434872859366891,
|
| 494 |
+
"grad_norm": 13.851968293163813,
|
| 495 |
+
"learning_rate": 4.2885094622913016e-07,
|
| 496 |
+
"logits/chosen": -0.06367097049951553,
|
| 497 |
+
"logits/rejected": 1.8669401407241821,
|
| 498 |
+
"logps/chosen": -630.0944213867188,
|
| 499 |
+
"logps/rejected": -1326.2972412109375,
|
| 500 |
+
"loss": 0.1541,
|
| 501 |
+
"rewards/accuracies": 0.90625,
|
| 502 |
+
"rewards/chosen": -2.6027626991271973,
|
| 503 |
+
"rewards/margins": 6.887024879455566,
|
| 504 |
+
"rewards/rejected": -9.489786148071289,
|
| 505 |
+
"step": 310
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.6642449403217436,
|
| 509 |
+
"grad_norm": 22.056173099963235,
|
| 510 |
+
"learning_rate": 4.223902735997788e-07,
|
| 511 |
+
"logits/chosen": -0.3920370638370514,
|
| 512 |
+
"logits/rejected": 1.7357165813446045,
|
| 513 |
+
"logps/chosen": -582.3004150390625,
|
| 514 |
+
"logps/rejected": -1314.4351806640625,
|
| 515 |
+
"loss": 0.1538,
|
| 516 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 517 |
+
"rewards/chosen": -2.171776294708252,
|
| 518 |
+
"rewards/margins": 7.103475093841553,
|
| 519 |
+
"rewards/rejected": -9.275251388549805,
|
| 520 |
+
"step": 320
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.6850025947067981,
|
| 524 |
+
"grad_norm": 10.33520679821666,
|
| 525 |
+
"learning_rate": 4.157022312906352e-07,
|
| 526 |
+
"logits/chosen": -1.2140147686004639,
|
| 527 |
+
"logits/rejected": 1.2578856945037842,
|
| 528 |
+
"logps/chosen": -645.0560302734375,
|
| 529 |
+
"logps/rejected": -1458.661376953125,
|
| 530 |
+
"loss": 0.1492,
|
| 531 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 532 |
+
"rewards/chosen": -2.7255163192749023,
|
| 533 |
+
"rewards/margins": 7.840676784515381,
|
| 534 |
+
"rewards/rejected": -10.566194534301758,
|
| 535 |
+
"step": 330
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.7057602490918526,
|
| 539 |
+
"grad_norm": 12.698623571384912,
|
| 540 |
+
"learning_rate": 4.0879564032162425e-07,
|
| 541 |
+
"logits/chosen": -1.3428138494491577,
|
| 542 |
+
"logits/rejected": 1.8070862293243408,
|
| 543 |
+
"logps/chosen": -719.0316162109375,
|
| 544 |
+
"logps/rejected": -1663.4603271484375,
|
| 545 |
+
"loss": 0.143,
|
| 546 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 547 |
+
"rewards/chosen": -3.4557037353515625,
|
| 548 |
+
"rewards/margins": 9.146575927734375,
|
| 549 |
+
"rewards/rejected": -12.602280616760254,
|
| 550 |
+
"step": 340
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.7265179034769071,
|
| 554 |
+
"grad_norm": 16.68939070839139,
|
| 555 |
+
"learning_rate": 4.016796099617569e-07,
|
| 556 |
+
"logits/chosen": -1.3726019859313965,
|
| 557 |
+
"logits/rejected": 1.8845252990722656,
|
| 558 |
+
"logps/chosen": -672.85693359375,
|
| 559 |
+
"logps/rejected": -1494.5582275390625,
|
| 560 |
+
"loss": 0.1538,
|
| 561 |
+
"rewards/accuracies": 0.893750011920929,
|
| 562 |
+
"rewards/chosen": -3.1178722381591797,
|
| 563 |
+
"rewards/margins": 7.914468288421631,
|
| 564 |
+
"rewards/rejected": -11.032341003417969,
|
| 565 |
+
"step": 350
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 0.7472755578619616,
|
| 569 |
+
"grad_norm": 14.783186894273188,
|
| 570 |
+
"learning_rate": 3.9436352571469577e-07,
|
| 571 |
+
"logits/chosen": -0.5545839071273804,
|
| 572 |
+
"logits/rejected": 3.833822727203369,
|
| 573 |
+
"logps/chosen": -830.2039184570312,
|
| 574 |
+
"logps/rejected": -1744.631103515625,
|
| 575 |
+
"loss": 0.1444,
|
| 576 |
+
"rewards/accuracies": 0.934374988079071,
|
| 577 |
+
"rewards/chosen": -4.657293796539307,
|
| 578 |
+
"rewards/margins": 9.05556583404541,
|
| 579 |
+
"rewards/rejected": -13.712858200073242,
|
| 580 |
+
"step": 360
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 0.768033212247016,
|
| 584 |
+
"grad_norm": 15.755226269776646,
|
| 585 |
+
"learning_rate": 3.868570369399893e-07,
|
| 586 |
+
"logits/chosen": -0.6274186372756958,
|
| 587 |
+
"logits/rejected": 3.4528141021728516,
|
| 588 |
+
"logps/chosen": -799.2054443359375,
|
| 589 |
+
"logps/rejected": -1723.1851806640625,
|
| 590 |
+
"loss": 0.1458,
|
| 591 |
+
"rewards/accuracies": 0.921875,
|
| 592 |
+
"rewards/chosen": -4.323949337005615,
|
| 593 |
+
"rewards/margins": 9.110595703125,
|
| 594 |
+
"rewards/rejected": -13.434545516967773,
|
| 595 |
+
"step": 370
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 0.7887908666320705,
|
| 599 |
+
"grad_norm": 11.452118871998977,
|
| 600 |
+
"learning_rate": 3.791700441262987e-07,
|
| 601 |
+
"logits/chosen": 0.8511954545974731,
|
| 602 |
+
"logits/rejected": 4.691292762756348,
|
| 603 |
+
"logps/chosen": -939.5437622070312,
|
| 604 |
+
"logps/rejected": -2012.4154052734375,
|
| 605 |
+
"loss": 0.1417,
|
| 606 |
+
"rewards/accuracies": 0.925000011920929,
|
| 607 |
+
"rewards/chosen": -5.871208667755127,
|
| 608 |
+
"rewards/margins": 10.141061782836914,
|
| 609 |
+
"rewards/rejected": -16.012271881103516,
|
| 610 |
+
"step": 380
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 0.809548521017125,
|
| 614 |
+
"grad_norm": 13.459005174890807,
|
| 615 |
+
"learning_rate": 3.7131268583340515e-07,
|
| 616 |
+
"logits/chosen": 0.3982798457145691,
|
| 617 |
+
"logits/rejected": 4.274272441864014,
|
| 618 |
+
"logps/chosen": -845.7362060546875,
|
| 619 |
+
"logps/rejected": -1642.904541015625,
|
| 620 |
+
"loss": 0.1523,
|
| 621 |
+
"rewards/accuracies": 0.9437500238418579,
|
| 622 |
+
"rewards/chosen": -4.7007269859313965,
|
| 623 |
+
"rewards/margins": 7.76672887802124,
|
| 624 |
+
"rewards/rejected": -12.467455863952637,
|
| 625 |
+
"step": 390
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.8303061754021795,
|
| 629 |
+
"grad_norm": 13.869817366585716,
|
| 630 |
+
"learning_rate": 3.632953253202198e-07,
|
| 631 |
+
"logits/chosen": 1.5343601703643799,
|
| 632 |
+
"logits/rejected": 4.5011491775512695,
|
| 633 |
+
"logps/chosen": -881.2744140625,
|
| 634 |
+
"logps/rejected": -1823.2939453125,
|
| 635 |
+
"loss": 0.1449,
|
| 636 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 637 |
+
"rewards/chosen": -5.160914421081543,
|
| 638 |
+
"rewards/margins": 9.128256797790527,
|
| 639 |
+
"rewards/rejected": -14.289172172546387,
|
| 640 |
+
"step": 400
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.851063829787234,
|
| 644 |
+
"grad_norm": 14.502067608546383,
|
| 645 |
+
"learning_rate": 3.551285368764321e-07,
|
| 646 |
+
"logits/chosen": 1.5195695161819458,
|
| 647 |
+
"logits/rejected": 4.168100357055664,
|
| 648 |
+
"logps/chosen": -846.6906127929688,
|
| 649 |
+
"logps/rejected": -1635.856689453125,
|
| 650 |
+
"loss": 0.1445,
|
| 651 |
+
"rewards/accuracies": 0.918749988079071,
|
| 652 |
+
"rewards/chosen": -4.812326431274414,
|
| 653 |
+
"rewards/margins": 7.744412422180176,
|
| 654 |
+
"rewards/rejected": -12.556737899780273,
|
| 655 |
+
"step": 410
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 0.8718214841722886,
|
| 659 |
+
"grad_norm": 30.730295160965017,
|
| 660 |
+
"learning_rate": 3.468230918758242e-07,
|
| 661 |
+
"logits/chosen": -0.11532606929540634,
|
| 662 |
+
"logits/rejected": 3.6724467277526855,
|
| 663 |
+
"logps/chosen": -856.9993286132812,
|
| 664 |
+
"logps/rejected": -1789.368408203125,
|
| 665 |
+
"loss": 0.1521,
|
| 666 |
+
"rewards/accuracies": 0.9281250238418579,
|
| 667 |
+
"rewards/chosen": -4.82938289642334,
|
| 668 |
+
"rewards/margins": 9.0027494430542,
|
| 669 |
+
"rewards/rejected": -13.832133293151855,
|
| 670 |
+
"step": 420
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 0.892579138557343,
|
| 674 |
+
"grad_norm": 15.361401417475852,
|
| 675 |
+
"learning_rate": 3.383899445696477e-07,
|
| 676 |
+
"logits/chosen": -0.19606542587280273,
|
| 677 |
+
"logits/rejected": 3.2190608978271484,
|
| 678 |
+
"logps/chosen": -914.5608520507812,
|
| 679 |
+
"logps/rejected": -1852.5074462890625,
|
| 680 |
+
"loss": 0.1317,
|
| 681 |
+
"rewards/accuracies": 0.9281250238418579,
|
| 682 |
+
"rewards/chosen": -5.41568660736084,
|
| 683 |
+
"rewards/margins": 9.153519630432129,
|
| 684 |
+
"rewards/rejected": -14.569204330444336,
|
| 685 |
+
"step": 430
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 0.9133367929423976,
|
| 689 |
+
"grad_norm": 15.988866292668206,
|
| 690 |
+
"learning_rate": 3.2984021763879756e-07,
|
| 691 |
+
"logits/chosen": 0.24461348354816437,
|
| 692 |
+
"logits/rejected": 3.7112534046173096,
|
| 693 |
+
"logps/chosen": -1049.1207275390625,
|
| 694 |
+
"logps/rejected": -2157.442138671875,
|
| 695 |
+
"loss": 0.1472,
|
| 696 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 697 |
+
"rewards/chosen": -6.739071846008301,
|
| 698 |
+
"rewards/margins": 10.936814308166504,
|
| 699 |
+
"rewards/rejected": -17.675886154174805,
|
| 700 |
+
"step": 440
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 0.934094447327452,
|
| 704 |
+
"grad_norm": 16.98533210114085,
|
| 705 |
+
"learning_rate": 3.211851875238408e-07,
|
| 706 |
+
"logits/chosen": 0.16562503576278687,
|
| 707 |
+
"logits/rejected": 4.111753940582275,
|
| 708 |
+
"logps/chosen": -1058.0418701171875,
|
| 709 |
+
"logps/rejected": -2127.619873046875,
|
| 710 |
+
"loss": 0.1444,
|
| 711 |
+
"rewards/accuracies": 0.9375,
|
| 712 |
+
"rewards/chosen": -6.909078121185303,
|
| 713 |
+
"rewards/margins": 10.602731704711914,
|
| 714 |
+
"rewards/rejected": -17.511810302734375,
|
| 715 |
+
"step": 450
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 0.9548521017125065,
|
| 719 |
+
"grad_norm": 14.967379090940785,
|
| 720 |
+
"learning_rate": 3.124362695522476e-07,
|
| 721 |
+
"logits/chosen": 1.7813737392425537,
|
| 722 |
+
"logits/rejected": 4.8395795822143555,
|
| 723 |
+
"logps/chosen": -1155.03125,
|
| 724 |
+
"logps/rejected": -2117.0537109375,
|
| 725 |
+
"loss": 0.1396,
|
| 726 |
+
"rewards/accuracies": 0.949999988079071,
|
| 727 |
+
"rewards/chosen": -7.852016448974609,
|
| 728 |
+
"rewards/margins": 9.481895446777344,
|
| 729 |
+
"rewards/rejected": -17.333911895751953,
|
| 730 |
+
"step": 460
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.975609756097561,
|
| 734 |
+
"grad_norm": 14.307026095079875,
|
| 735 |
+
"learning_rate": 3.036050028824415e-07,
|
| 736 |
+
"logits/chosen": 1.5212550163269043,
|
| 737 |
+
"logits/rejected": 4.2941412925720215,
|
| 738 |
+
"logps/chosen": -1053.684814453125,
|
| 739 |
+
"logps/rejected": -1923.693359375,
|
| 740 |
+
"loss": 0.125,
|
| 741 |
+
"rewards/accuracies": 0.9281250238418579,
|
| 742 |
+
"rewards/chosen": -6.825259208679199,
|
| 743 |
+
"rewards/margins": 8.460533142089844,
|
| 744 |
+
"rewards/rejected": -15.285791397094727,
|
| 745 |
+
"step": 470
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.9963674104826155,
|
| 749 |
+
"grad_norm": 11.916307943626432,
|
| 750 |
+
"learning_rate": 2.9470303528452547e-07,
|
| 751 |
+
"logits/chosen": 1.5992648601531982,
|
| 752 |
+
"logits/rejected": 4.383261680603027,
|
| 753 |
+
"logps/chosen": -1080.980224609375,
|
| 754 |
+
"logps/rejected": -2100.404541015625,
|
| 755 |
+
"loss": 0.1408,
|
| 756 |
+
"rewards/accuracies": 0.9437500238418579,
|
| 757 |
+
"rewards/chosen": -7.290544033050537,
|
| 758 |
+
"rewards/margins": 9.89597225189209,
|
| 759 |
+
"rewards/rejected": -17.1865177154541,
|
| 760 |
+
"step": 480
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 0.9963674104826155,
|
| 764 |
+
"eval_logits/chosen": 0.5014842748641968,
|
| 765 |
+
"eval_logits/rejected": 3.342353582382202,
|
| 766 |
+
"eval_logps/chosen": -1083.2030029296875,
|
| 767 |
+
"eval_logps/rejected": -2038.5933837890625,
|
| 768 |
+
"eval_loss": 0.14824749529361725,
|
| 769 |
+
"eval_rewards/accuracies": 0.9325980544090271,
|
| 770 |
+
"eval_rewards/chosen": -6.597947597503662,
|
| 771 |
+
"eval_rewards/margins": 9.328583717346191,
|
| 772 |
+
"eval_rewards/rejected": -15.926531791687012,
|
| 773 |
+
"eval_runtime": 296.0381,
|
| 774 |
+
"eval_samples_per_second": 21.926,
|
| 775 |
+
"eval_steps_per_second": 0.345,
|
| 776 |
+
"step": 480
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 1.01712506486767,
|
| 780 |
+
"grad_norm": 14.456304387791999,
|
| 781 |
+
"learning_rate": 2.8574210777775755e-07,
|
| 782 |
+
"logits/chosen": 0.9456971287727356,
|
| 783 |
+
"logits/rejected": 4.39624547958374,
|
| 784 |
+
"logps/chosen": -1059.8154296875,
|
| 785 |
+
"logps/rejected": -2144.65576171875,
|
| 786 |
+
"loss": 0.0951,
|
| 787 |
+
"rewards/accuracies": 0.953125,
|
| 788 |
+
"rewards/chosen": -6.945745944976807,
|
| 789 |
+
"rewards/margins": 10.675562858581543,
|
| 790 |
+
"rewards/rejected": -17.621309280395508,
|
| 791 |
+
"step": 490
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 1.0378827192527245,
|
| 795 |
+
"grad_norm": 17.681966171705884,
|
| 796 |
+
"learning_rate": 2.767340391450384e-07,
|
| 797 |
+
"logits/chosen": 1.6467583179473877,
|
| 798 |
+
"logits/rejected": 5.070549011230469,
|
| 799 |
+
"logps/chosen": -1184.4014892578125,
|
| 800 |
+
"logps/rejected": -2638.23291015625,
|
| 801 |
+
"loss": 0.0919,
|
| 802 |
+
"rewards/accuracies": 0.9468749761581421,
|
| 803 |
+
"rewards/chosen": -8.20818042755127,
|
| 804 |
+
"rewards/margins": 14.303576469421387,
|
| 805 |
+
"rewards/rejected": -22.511754989624023,
|
| 806 |
+
"step": 500
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 1.058640373637779,
|
| 810 |
+
"grad_norm": 13.802679882016555,
|
| 811 |
+
"learning_rate": 2.6769071034483407e-07,
|
| 812 |
+
"logits/chosen": 1.1824378967285156,
|
| 813 |
+
"logits/rejected": 4.564583778381348,
|
| 814 |
+
"logps/chosen": -947.0963745117188,
|
| 815 |
+
"logps/rejected": -2079.13916015625,
|
| 816 |
+
"loss": 0.0879,
|
| 817 |
+
"rewards/accuracies": 0.9781249761581421,
|
| 818 |
+
"rewards/chosen": -5.809744358062744,
|
| 819 |
+
"rewards/margins": 11.030488967895508,
|
| 820 |
+
"rewards/rejected": -16.840232849121094,
|
| 821 |
+
"step": 510
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 1.0793980280228335,
|
| 825 |
+
"grad_norm": 22.098742924019817,
|
| 826 |
+
"learning_rate": 2.5862404884109365e-07,
|
| 827 |
+
"logits/chosen": 1.4323368072509766,
|
| 828 |
+
"logits/rejected": 4.83809757232666,
|
| 829 |
+
"logps/chosen": -1047.9368896484375,
|
| 830 |
+
"logps/rejected": -2401.321533203125,
|
| 831 |
+
"loss": 0.0826,
|
| 832 |
+
"rewards/accuracies": 0.965624988079071,
|
| 833 |
+
"rewards/chosen": -6.900972843170166,
|
| 834 |
+
"rewards/margins": 13.123873710632324,
|
| 835 |
+
"rewards/rejected": -20.02484703063965,
|
| 836 |
+
"step": 520
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 1.100155682407888,
|
| 840 |
+
"grad_norm": 14.888695199480608,
|
| 841 |
+
"learning_rate": 2.495460128718305e-07,
|
| 842 |
+
"logits/chosen": 0.8169673681259155,
|
| 843 |
+
"logits/rejected": 4.492055416107178,
|
| 844 |
+
"logps/chosen": -1088.3392333984375,
|
| 845 |
+
"logps/rejected": -2334.029052734375,
|
| 846 |
+
"loss": 0.0775,
|
| 847 |
+
"rewards/accuracies": 0.956250011920929,
|
| 848 |
+
"rewards/chosen": -7.015921592712402,
|
| 849 |
+
"rewards/margins": 12.27374267578125,
|
| 850 |
+
"rewards/rejected": -19.289663314819336,
|
| 851 |
+
"step": 530
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 1.1209133367929425,
|
| 855 |
+
"grad_norm": 13.711286256989611,
|
| 856 |
+
"learning_rate": 2.404685756771143e-07,
|
| 857 |
+
"logits/chosen": 0.3249141275882721,
|
| 858 |
+
"logits/rejected": 4.28688383102417,
|
| 859 |
+
"logps/chosen": -1050.6568603515625,
|
| 860 |
+
"logps/rejected": -2440.43212890625,
|
| 861 |
+
"loss": 0.0831,
|
| 862 |
+
"rewards/accuracies": 0.956250011920929,
|
| 863 |
+
"rewards/chosen": -6.925228118896484,
|
| 864 |
+
"rewards/margins": 13.674964904785156,
|
| 865 |
+
"rewards/rejected": -20.60019302368164,
|
| 866 |
+
"step": 540
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 1.141670991177997,
|
| 870 |
+
"grad_norm": 12.310015587533016,
|
| 871 |
+
"learning_rate": 2.314037097072764e-07,
|
| 872 |
+
"logits/chosen": 0.325296014547348,
|
| 873 |
+
"logits/rejected": 4.037835121154785,
|
| 874 |
+
"logps/chosen": -1070.4940185546875,
|
| 875 |
+
"logps/rejected": -2429.653076171875,
|
| 876 |
+
"loss": 0.0774,
|
| 877 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 878 |
+
"rewards/chosen": -7.072981357574463,
|
| 879 |
+
"rewards/margins": 13.413686752319336,
|
| 880 |
+
"rewards/rejected": -20.48666763305664,
|
| 881 |
+
"step": 550
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 1.1624286455630513,
|
| 885 |
+
"grad_norm": 18.024933264685323,
|
| 886 |
+
"learning_rate": 2.2236337083215723e-07,
|
| 887 |
+
"logits/chosen": 0.973385214805603,
|
| 888 |
+
"logits/rejected": 4.8588714599609375,
|
| 889 |
+
"logps/chosen": -1165.290283203125,
|
| 890 |
+
"logps/rejected": -2533.037841796875,
|
| 891 |
+
"loss": 0.0806,
|
| 892 |
+
"rewards/accuracies": 0.965624988079071,
|
| 893 |
+
"rewards/chosen": -7.950223445892334,
|
| 894 |
+
"rewards/margins": 13.607198715209961,
|
| 895 |
+
"rewards/rejected": -21.557422637939453,
|
| 896 |
+
"step": 560
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 1.183186299948106,
|
| 900 |
+
"grad_norm": 39.199036754410095,
|
| 901 |
+
"learning_rate": 2.13359482572222e-07,
|
| 902 |
+
"logits/chosen": 0.9793283343315125,
|
| 903 |
+
"logits/rejected": 4.486080169677734,
|
| 904 |
+
"logps/chosen": -1073.2510986328125,
|
| 905 |
+
"logps/rejected": -2294.072509765625,
|
| 906 |
+
"loss": 0.0855,
|
| 907 |
+
"rewards/accuracies": 0.9468749761581421,
|
| 908 |
+
"rewards/chosen": -7.108448028564453,
|
| 909 |
+
"rewards/margins": 11.980100631713867,
|
| 910 |
+
"rewards/rejected": -19.088550567626953,
|
| 911 |
+
"step": 570
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 1.2039439543331603,
|
| 915 |
+
"grad_norm": 13.491261159834455,
|
| 916 |
+
"learning_rate": 2.044039203723423e-07,
|
| 917 |
+
"logits/chosen": 0.23347489535808563,
|
| 918 |
+
"logits/rejected": 3.7263119220733643,
|
| 919 |
+
"logps/chosen": -997.2879028320312,
|
| 920 |
+
"logps/rejected": -2139.56298828125,
|
| 921 |
+
"loss": 0.0886,
|
| 922 |
+
"rewards/accuracies": 0.9437500238418579,
|
| 923 |
+
"rewards/chosen": -6.343294143676758,
|
| 924 |
+
"rewards/margins": 11.091756820678711,
|
| 925 |
+
"rewards/rejected": -17.435049057006836,
|
| 926 |
+
"step": 580
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 1.2247016087182148,
|
| 930 |
+
"grad_norm": 22.750928538479595,
|
| 931 |
+
"learning_rate": 1.955084959389864e-07,
|
| 932 |
+
"logits/chosen": 0.4886883795261383,
|
| 933 |
+
"logits/rejected": 4.1630377769470215,
|
| 934 |
+
"logps/chosen": -1106.54052734375,
|
| 935 |
+
"logps/rejected": -2491.51416015625,
|
| 936 |
+
"loss": 0.0749,
|
| 937 |
+
"rewards/accuracies": 0.9781249761581421,
|
| 938 |
+
"rewards/chosen": -7.559281826019287,
|
| 939 |
+
"rewards/margins": 13.47680377960205,
|
| 940 |
+
"rewards/rejected": -21.036083221435547,
|
| 941 |
+
"step": 590
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 1.2454592631032693,
|
| 945 |
+
"grad_norm": 20.593372405978258,
|
| 946 |
+
"learning_rate": 1.866849416614753e-07,
|
| 947 |
+
"logits/chosen": 0.24794098734855652,
|
| 948 |
+
"logits/rejected": 4.261725425720215,
|
| 949 |
+
"logps/chosen": -1190.895751953125,
|
| 950 |
+
"logps/rejected": -2684.16943359375,
|
| 951 |
+
"loss": 0.0762,
|
| 952 |
+
"rewards/accuracies": 0.9781249761581421,
|
| 953 |
+
"rewards/chosen": -8.191116333007812,
|
| 954 |
+
"rewards/margins": 14.663787841796875,
|
| 955 |
+
"rewards/rejected": -22.854902267456055,
|
| 956 |
+
"step": 600
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 1.2662169174883238,
|
| 960 |
+
"grad_norm": 23.08708475257412,
|
| 961 |
+
"learning_rate": 1.7794489513785227e-07,
|
| 962 |
+
"logits/chosen": -0.35908642411231995,
|
| 963 |
+
"logits/rejected": 3.719221830368042,
|
| 964 |
+
"logps/chosen": -1067.587158203125,
|
| 965 |
+
"logps/rejected": -2360.556396484375,
|
| 966 |
+
"loss": 0.0824,
|
| 967 |
+
"rewards/accuracies": 0.96875,
|
| 968 |
+
"rewards/chosen": -6.811041355133057,
|
| 969 |
+
"rewards/margins": 12.751296997070312,
|
| 970 |
+
"rewards/rejected": -19.562339782714844,
|
| 971 |
+
"step": 610
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 1.2869745718733783,
|
| 975 |
+
"grad_norm": 11.158342147450194,
|
| 976 |
+
"learning_rate": 1.692998838257744e-07,
|
| 977 |
+
"logits/chosen": -0.18085989356040955,
|
| 978 |
+
"logits/rejected": 3.5346386432647705,
|
| 979 |
+
"logps/chosen": -1038.219970703125,
|
| 980 |
+
"logps/rejected": -2255.41064453125,
|
| 981 |
+
"loss": 0.0769,
|
| 982 |
+
"rewards/accuracies": 0.981249988079071,
|
| 983 |
+
"rewards/chosen": -6.670246124267578,
|
| 984 |
+
"rewards/margins": 11.845807075500488,
|
| 985 |
+
"rewards/rejected": -18.51605224609375,
|
| 986 |
+
"step": 620
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 1.3077322262584328,
|
| 990 |
+
"grad_norm": 14.33248304938314,
|
| 991 |
+
"learning_rate": 1.6076130983867191e-07,
|
| 992 |
+
"logits/chosen": -0.1000831350684166,
|
| 993 |
+
"logits/rejected": 4.0322370529174805,
|
| 994 |
+
"logps/chosen": -1037.5277099609375,
|
| 995 |
+
"logps/rejected": -2512.465087890625,
|
| 996 |
+
"loss": 0.0774,
|
| 997 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 998 |
+
"rewards/chosen": -6.664424896240234,
|
| 999 |
+
"rewards/margins": 14.430699348449707,
|
| 1000 |
+
"rewards/rejected": -21.095125198364258,
|
| 1001 |
+
"step": 630
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 1.3284898806434873,
|
| 1005 |
+
"grad_norm": 18.5366504084505,
|
| 1006 |
+
"learning_rate": 1.5234043490722587e-07,
|
| 1007 |
+
"logits/chosen": -0.20596277713775635,
|
| 1008 |
+
"logits/rejected": 3.613173246383667,
|
| 1009 |
+
"logps/chosen": -1079.144287109375,
|
| 1010 |
+
"logps/rejected": -2312.60791015625,
|
| 1011 |
+
"loss": 0.0821,
|
| 1012 |
+
"rewards/accuracies": 0.965624988079071,
|
| 1013 |
+
"rewards/chosen": -7.167913913726807,
|
| 1014 |
+
"rewards/margins": 12.191060066223145,
|
| 1015 |
+
"rewards/rejected": -19.35897445678711,
|
| 1016 |
+
"step": 640
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 1.3492475350285418,
|
| 1020 |
+
"grad_norm": 18.228054814234156,
|
| 1021 |
+
"learning_rate": 1.44048365526001e-07,
|
| 1022 |
+
"logits/chosen": 0.0638003945350647,
|
| 1023 |
+
"logits/rejected": 3.9387619495391846,
|
| 1024 |
+
"logps/chosen": -1077.5164794921875,
|
| 1025 |
+
"logps/rejected": -2451.49462890625,
|
| 1026 |
+
"loss": 0.0867,
|
| 1027 |
+
"rewards/accuracies": 0.96875,
|
| 1028 |
+
"rewards/chosen": -7.139754295349121,
|
| 1029 |
+
"rewards/margins": 13.3494873046875,
|
| 1030 |
+
"rewards/rejected": -20.489240646362305,
|
| 1031 |
+
"step": 650
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 1.3700051894135963,
|
| 1035 |
+
"grad_norm": 16.31608491567421,
|
| 1036 |
+
"learning_rate": 1.3589603830482243e-07,
|
| 1037 |
+
"logits/chosen": -0.08248420059680939,
|
| 1038 |
+
"logits/rejected": 4.198277950286865,
|
| 1039 |
+
"logps/chosen": -1142.9498291015625,
|
| 1040 |
+
"logps/rejected": -2611.378662109375,
|
| 1041 |
+
"loss": 0.0703,
|
| 1042 |
+
"rewards/accuracies": 0.96875,
|
| 1043 |
+
"rewards/chosen": -7.643826484680176,
|
| 1044 |
+
"rewards/margins": 14.516764640808105,
|
| 1045 |
+
"rewards/rejected": -22.160587310791016,
|
| 1046 |
+
"step": 660
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 1.3907628437986508,
|
| 1050 |
+
"grad_norm": 21.850286306048936,
|
| 1051 |
+
"learning_rate": 1.2789420554421821e-07,
|
| 1052 |
+
"logits/chosen": -0.33774855732917786,
|
| 1053 |
+
"logits/rejected": 3.5582516193389893,
|
| 1054 |
+
"logps/chosen": -1178.3541259765625,
|
| 1055 |
+
"logps/rejected": -2523.93701171875,
|
| 1056 |
+
"loss": 0.0845,
|
| 1057 |
+
"rewards/accuracies": 0.956250011920929,
|
| 1058 |
+
"rewards/chosen": -8.179405212402344,
|
| 1059 |
+
"rewards/margins": 13.182377815246582,
|
| 1060 |
+
"rewards/rejected": -21.36178207397461,
|
| 1061 |
+
"step": 670
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 1.4115204981837053,
|
| 1065 |
+
"grad_norm": 15.996775283562163,
|
| 1066 |
+
"learning_rate": 1.200534210539509e-07,
|
| 1067 |
+
"logits/chosen": -0.7213211059570312,
|
| 1068 |
+
"logits/rejected": 3.0649337768554688,
|
| 1069 |
+
"logps/chosen": -1151.722412109375,
|
| 1070 |
+
"logps/rejected": -2552.525634765625,
|
| 1071 |
+
"loss": 0.0892,
|
| 1072 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1073 |
+
"rewards/chosen": -7.825603485107422,
|
| 1074 |
+
"rewards/margins": 13.686927795410156,
|
| 1075 |
+
"rewards/rejected": -21.512531280517578,
|
| 1076 |
+
"step": 680
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 1.4322781525687598,
|
| 1080 |
+
"grad_norm": 16.75949934654318,
|
| 1081 |
+
"learning_rate": 1.1238402623334492e-07,
|
| 1082 |
+
"logits/chosen": -0.9597622752189636,
|
| 1083 |
+
"logits/rejected": 2.867584228515625,
|
| 1084 |
+
"logps/chosen": -1066.507568359375,
|
| 1085 |
+
"logps/rejected": -2423.34423828125,
|
| 1086 |
+
"loss": 0.0898,
|
| 1087 |
+
"rewards/accuracies": 0.953125,
|
| 1088 |
+
"rewards/chosen": -6.92177677154541,
|
| 1089 |
+
"rewards/margins": 13.205133438110352,
|
| 1090 |
+
"rewards/rejected": -20.126911163330078,
|
| 1091 |
+
"step": 690
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 1.4530358069538143,
|
| 1095 |
+
"grad_norm": 11.338012307523448,
|
| 1096 |
+
"learning_rate": 1.0489613643176479e-07,
|
| 1097 |
+
"logits/chosen": -0.6678389310836792,
|
| 1098 |
+
"logits/rejected": 3.2197043895721436,
|
| 1099 |
+
"logps/chosen": -1113.594482421875,
|
| 1100 |
+
"logps/rejected": -2410.258544921875,
|
| 1101 |
+
"loss": 0.0722,
|
| 1102 |
+
"rewards/accuracies": 0.956250011920929,
|
| 1103 |
+
"rewards/chosen": -7.322343349456787,
|
| 1104 |
+
"rewards/margins": 12.825822830200195,
|
| 1105 |
+
"rewards/rejected": -20.14816665649414,
|
| 1106 |
+
"step": 700
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 1.4737934613388686,
|
| 1110 |
+
"grad_norm": 15.856495888079317,
|
| 1111 |
+
"learning_rate": 9.759962760723855e-08,
|
| 1112 |
+
"logits/chosen": -0.16049222648143768,
|
| 1113 |
+
"logits/rejected": 4.048083305358887,
|
| 1114 |
+
"logps/chosen": -1173.0849609375,
|
| 1115 |
+
"logps/rejected": -2468.91845703125,
|
| 1116 |
+
"loss": 0.0662,
|
| 1117 |
+
"rewards/accuracies": 0.96875,
|
| 1118 |
+
"rewards/chosen": -8.210199356079102,
|
| 1119 |
+
"rewards/margins": 12.80036735534668,
|
| 1120 |
+
"rewards/rejected": -21.01056671142578,
|
| 1121 |
+
"step": 710
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 1.4945511157239233,
|
| 1125 |
+
"grad_norm": 31.07426456754591,
|
| 1126 |
+
"learning_rate": 9.050412330081883e-08,
|
| 1127 |
+
"logits/chosen": -0.20700784027576447,
|
| 1128 |
+
"logits/rejected": 3.9535317420959473,
|
| 1129 |
+
"logps/chosen": -1277.49755859375,
|
| 1130 |
+
"logps/rejected": -2737.533447265625,
|
| 1131 |
+
"loss": 0.0852,
|
| 1132 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1133 |
+
"rewards/chosen": -9.044589042663574,
|
| 1134 |
+
"rewards/margins": 14.331390380859375,
|
| 1135 |
+
"rewards/rejected": -23.375978469848633,
|
| 1136 |
+
"step": 720
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 1.4945511157239233,
|
| 1140 |
+
"eval_logits/chosen": 0.06359091401100159,
|
| 1141 |
+
"eval_logits/rejected": 3.48725962638855,
|
| 1142 |
+
"eval_logps/chosen": -1424.8193359375,
|
| 1143 |
+
"eval_logps/rejected": -2816.588623046875,
|
| 1144 |
+
"eval_loss": 0.16444814205169678,
|
| 1145 |
+
"eval_rewards/accuracies": 0.9227941036224365,
|
| 1146 |
+
"eval_rewards/chosen": -10.01410961151123,
|
| 1147 |
+
"eval_rewards/margins": 13.69237232208252,
|
| 1148 |
+
"eval_rewards/rejected": -23.706483840942383,
|
| 1149 |
+
"eval_runtime": 297.3735,
|
| 1150 |
+
"eval_samples_per_second": 21.828,
|
| 1151 |
+
"eval_steps_per_second": 0.343,
|
| 1152 |
+
"step": 720
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 1.5153087701089776,
|
| 1156 |
+
"grad_norm": 21.935108549149014,
|
| 1157 |
+
"learning_rate": 8.36189819438625e-08,
|
| 1158 |
+
"logits/chosen": -0.2683785557746887,
|
| 1159 |
+
"logits/rejected": 3.702204465866089,
|
| 1160 |
+
"logps/chosen": -1242.103759765625,
|
| 1161 |
+
"logps/rejected": -2603.80615234375,
|
| 1162 |
+
"loss": 0.0822,
|
| 1163 |
+
"rewards/accuracies": 0.965624988079071,
|
| 1164 |
+
"rewards/chosen": -8.634117126464844,
|
| 1165 |
+
"rewards/margins": 13.532336235046387,
|
| 1166 |
+
"rewards/rejected": -22.166454315185547,
|
| 1167 |
+
"step": 730
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 1.5360664244940323,
|
| 1171 |
+
"grad_norm": 23.902586381058878,
|
| 1172 |
+
"learning_rate": 7.69532845149711e-08,
|
| 1173 |
+
"logits/chosen": -0.5570476055145264,
|
| 1174 |
+
"logits/rejected": 3.649244785308838,
|
| 1175 |
+
"logps/chosen": -1223.75439453125,
|
| 1176 |
+
"logps/rejected": -2637.007080078125,
|
| 1177 |
+
"loss": 0.0769,
|
| 1178 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1179 |
+
"rewards/chosen": -8.374210357666016,
|
| 1180 |
+
"rewards/margins": 13.947656631469727,
|
| 1181 |
+
"rewards/rejected": -22.321866989135742,
|
| 1182 |
+
"step": 740
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 1.5568240788790866,
|
| 1186 |
+
"grad_norm": 20.338315184147856,
|
| 1187 |
+
"learning_rate": 7.051582256286929e-08,
|
| 1188 |
+
"logits/chosen": -0.5207056999206543,
|
| 1189 |
+
"logits/rejected": 3.2942442893981934,
|
| 1190 |
+
"logps/chosen": -1216.895263671875,
|
| 1191 |
+
"logps/rejected": -2651.10107421875,
|
| 1192 |
+
"loss": 0.0706,
|
| 1193 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1194 |
+
"rewards/chosen": -8.260024070739746,
|
| 1195 |
+
"rewards/margins": 14.131828308105469,
|
| 1196 |
+
"rewards/rejected": -22.39185333251953,
|
| 1197 |
+
"step": 750
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 1.5775817332641413,
|
| 1201 |
+
"grad_norm": 19.826108952738736,
|
| 1202 |
+
"learning_rate": 6.431508661101954e-08,
|
| 1203 |
+
"logits/chosen": -0.35470911860466003,
|
| 1204 |
+
"logits/rejected": 3.6188712120056152,
|
| 1205 |
+
"logps/chosen": -1201.515625,
|
| 1206 |
+
"logps/rejected": -2452.108642578125,
|
| 1207 |
+
"loss": 0.0906,
|
| 1208 |
+
"rewards/accuracies": 0.9437500238418579,
|
| 1209 |
+
"rewards/chosen": -8.28996467590332,
|
| 1210 |
+
"rewards/margins": 12.43807315826416,
|
| 1211 |
+
"rewards/rejected": -20.728038787841797,
|
| 1212 |
+
"step": 760
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 1.5983393876491956,
|
| 1216 |
+
"grad_norm": 18.731154290431807,
|
| 1217 |
+
"learning_rate": 5.8359254959266826e-08,
|
| 1218 |
+
"logits/chosen": -0.3598397374153137,
|
| 1219 |
+
"logits/rejected": 3.461977481842041,
|
| 1220 |
+
"logps/chosen": -1123.501220703125,
|
| 1221 |
+
"logps/rejected": -2401.59423828125,
|
| 1222 |
+
"loss": 0.0757,
|
| 1223 |
+
"rewards/accuracies": 0.956250011920929,
|
| 1224 |
+
"rewards/chosen": -7.732508659362793,
|
| 1225 |
+
"rewards/margins": 12.514566421508789,
|
| 1226 |
+
"rewards/rejected": -20.247074127197266,
|
| 1227 |
+
"step": 770
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 1.61909704203425,
|
| 1231 |
+
"grad_norm": 14.021094730035166,
|
| 1232 |
+
"learning_rate": 5.265618289728199e-08,
|
| 1233 |
+
"logits/chosen": -0.2899594306945801,
|
| 1234 |
+
"logits/rejected": 3.6835436820983887,
|
| 1235 |
+
"logps/chosen": -1112.9346923828125,
|
| 1236 |
+
"logps/rejected": -2395.99560546875,
|
| 1237 |
+
"loss": 0.0748,
|
| 1238 |
+
"rewards/accuracies": 0.965624988079071,
|
| 1239 |
+
"rewards/chosen": -7.564324378967285,
|
| 1240 |
+
"rewards/margins": 12.517146110534668,
|
| 1241 |
+
"rewards/rejected": -20.081470489501953,
|
| 1242 |
+
"step": 780
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 1.6398546964193046,
|
| 1246 |
+
"grad_norm": 15.755848122162748,
|
| 1247 |
+
"learning_rate": 4.721339234403121e-08,
|
| 1248 |
+
"logits/chosen": -0.4539187550544739,
|
| 1249 |
+
"logits/rejected": 3.4312031269073486,
|
| 1250 |
+
"logps/chosen": -1102.984375,
|
| 1251 |
+
"logps/rejected": -2424.76904296875,
|
| 1252 |
+
"loss": 0.0889,
|
| 1253 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1254 |
+
"rewards/chosen": -7.416151523590088,
|
| 1255 |
+
"rewards/margins": 12.970865249633789,
|
| 1256 |
+
"rewards/rejected": -20.387014389038086,
|
| 1257 |
+
"step": 790
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 1.660612350804359,
|
| 1261 |
+
"grad_norm": 10.116183369840204,
|
| 1262 |
+
"learning_rate": 4.203806192693587e-08,
|
| 1263 |
+
"logits/chosen": -0.16934530436992645,
|
| 1264 |
+
"logits/rejected": 3.3178775310516357,
|
| 1265 |
+
"logps/chosen": -1117.904541015625,
|
| 1266 |
+
"logps/rejected": -2234.350341796875,
|
| 1267 |
+
"loss": 0.0806,
|
| 1268 |
+
"rewards/accuracies": 0.971875011920929,
|
| 1269 |
+
"rewards/chosen": -7.489673614501953,
|
| 1270 |
+
"rewards/margins": 10.942729949951172,
|
| 1271 |
+
"rewards/rejected": -18.432403564453125,
|
| 1272 |
+
"step": 800
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 1.6813700051894136,
|
| 1276 |
+
"grad_norm": 12.314909116730863,
|
| 1277 |
+
"learning_rate": 3.7137017513808544e-08,
|
| 1278 |
+
"logits/chosen": -0.2919425368309021,
|
| 1279 |
+
"logits/rejected": 3.503552198410034,
|
| 1280 |
+
"logps/chosen": -1133.6949462890625,
|
| 1281 |
+
"logps/rejected": -2370.295166015625,
|
| 1282 |
+
"loss": 0.0813,
|
| 1283 |
+
"rewards/accuracies": 0.9593750238418579,
|
| 1284 |
+
"rewards/chosen": -7.621180057525635,
|
| 1285 |
+
"rewards/margins": 12.205830574035645,
|
| 1286 |
+
"rewards/rejected": -19.827011108398438,
|
| 1287 |
+
"step": 810
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 1.702127659574468,
|
| 1291 |
+
"grad_norm": 13.100920531092996,
|
| 1292 |
+
"learning_rate": 3.251672321005147e-08,
|
| 1293 |
+
"logits/chosen": -0.3816925585269928,
|
| 1294 |
+
"logits/rejected": 3.223738193511963,
|
| 1295 |
+
"logps/chosen": -1128.720458984375,
|
| 1296 |
+
"logps/rejected": -2432.7451171875,
|
| 1297 |
+
"loss": 0.0799,
|
| 1298 |
+
"rewards/accuracies": 0.96875,
|
| 1299 |
+
"rewards/chosen": -7.558645725250244,
|
| 1300 |
+
"rewards/margins": 12.71537971496582,
|
| 1301 |
+
"rewards/rejected": -20.274024963378906,
|
| 1302 |
+
"step": 820
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 1.7228853139595226,
|
| 1306 |
+
"grad_norm": 13.542985324247502,
|
| 1307 |
+
"learning_rate": 2.8183272832992267e-08,
|
| 1308 |
+
"logits/chosen": -0.4070394039154053,
|
| 1309 |
+
"logits/rejected": 3.100729465484619,
|
| 1310 |
+
"logps/chosen": -1098.7197265625,
|
| 1311 |
+
"logps/rejected": -2404.990234375,
|
| 1312 |
+
"loss": 0.0769,
|
| 1313 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1314 |
+
"rewards/chosen": -7.4235520362854,
|
| 1315 |
+
"rewards/margins": 12.707655906677246,
|
| 1316 |
+
"rewards/rejected": -20.131206512451172,
|
| 1317 |
+
"step": 830
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 1.743642968344577,
|
| 1321 |
+
"grad_norm": 19.414761044152183,
|
| 1322 |
+
"learning_rate": 2.414238187460191e-08,
|
| 1323 |
+
"logits/chosen": -0.3680698275566101,
|
| 1324 |
+
"logits/rejected": 3.6575589179992676,
|
| 1325 |
+
"logps/chosen": -1149.003662109375,
|
| 1326 |
+
"logps/rejected": -2435.11474609375,
|
| 1327 |
+
"loss": 0.0801,
|
| 1328 |
+
"rewards/accuracies": 0.971875011920929,
|
| 1329 |
+
"rewards/chosen": -7.7628068923950195,
|
| 1330 |
+
"rewards/margins": 12.7561674118042,
|
| 1331 |
+
"rewards/rejected": -20.51897621154785,
|
| 1332 |
+
"step": 840
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 1.7644006227296316,
|
| 1336 |
+
"grad_norm": 18.249543754589453,
|
| 1337 |
+
"learning_rate": 2.0399379963194713e-08,
|
| 1338 |
+
"logits/chosen": -0.15194618701934814,
|
| 1339 |
+
"logits/rejected": 3.823453903198242,
|
| 1340 |
+
"logps/chosen": -1178.7374267578125,
|
| 1341 |
+
"logps/rejected": -2647.753173828125,
|
| 1342 |
+
"loss": 0.0742,
|
| 1343 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1344 |
+
"rewards/chosen": -8.07176685333252,
|
| 1345 |
+
"rewards/margins": 14.458259582519531,
|
| 1346 |
+
"rewards/rejected": -22.530025482177734,
|
| 1347 |
+
"step": 850
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 1.7851582771146859,
|
| 1351 |
+
"grad_norm": 13.687621613308757,
|
| 1352 |
+
"learning_rate": 1.695920383405322e-08,
|
| 1353 |
+
"logits/chosen": -0.1904297024011612,
|
| 1354 |
+
"logits/rejected": 3.487968921661377,
|
| 1355 |
+
"logps/chosen": -1222.4801025390625,
|
| 1356 |
+
"logps/rejected": -2662.45654296875,
|
| 1357 |
+
"loss": 0.0862,
|
| 1358 |
+
"rewards/accuracies": 0.981249988079071,
|
| 1359 |
+
"rewards/chosen": -8.289772033691406,
|
| 1360 |
+
"rewards/margins": 14.194554328918457,
|
| 1361 |
+
"rewards/rejected": -22.484325408935547,
|
| 1362 |
+
"step": 860
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 1.8059159314997406,
|
| 1366 |
+
"grad_norm": 20.833924751057065,
|
| 1367 |
+
"learning_rate": 1.3826390818249434e-08,
|
| 1368 |
+
"logits/chosen": -0.10482398420572281,
|
| 1369 |
+
"logits/rejected": 3.5555343627929688,
|
| 1370 |
+
"logps/chosen": -1184.642822265625,
|
| 1371 |
+
"logps/rejected": -2578.11376953125,
|
| 1372 |
+
"loss": 0.0847,
|
| 1373 |
+
"rewards/accuracies": 0.965624988079071,
|
| 1374 |
+
"rewards/chosen": -8.188472747802734,
|
| 1375 |
+
"rewards/margins": 13.555729866027832,
|
| 1376 |
+
"rewards/rejected": -21.74420166015625,
|
| 1377 |
+
"step": 870
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 1.826673585884795,
|
| 1381 |
+
"grad_norm": 12.098887002832749,
|
| 1382 |
+
"learning_rate": 1.1005072858249614e-08,
|
| 1383 |
+
"logits/chosen": -0.21788470447063446,
|
| 1384 |
+
"logits/rejected": 3.596818447113037,
|
| 1385 |
+
"logps/chosen": -1198.203125,
|
| 1386 |
+
"logps/rejected": -2579.16455078125,
|
| 1387 |
+
"loss": 0.0742,
|
| 1388 |
+
"rewards/accuracies": 0.9781249761581421,
|
| 1389 |
+
"rewards/chosen": -8.193432807922363,
|
| 1390 |
+
"rewards/margins": 13.623662948608398,
|
| 1391 |
+
"rewards/rejected": -21.817094802856445,
|
| 1392 |
+
"step": 880
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 1.8474312402698496,
|
| 1396 |
+
"grad_norm": 22.649648963902354,
|
| 1397 |
+
"learning_rate": 8.498971058195886e-09,
|
| 1398 |
+
"logits/chosen": -0.2010045051574707,
|
| 1399 |
+
"logits/rejected": 3.6617989540100098,
|
| 1400 |
+
"logps/chosen": -1205.736328125,
|
| 1401 |
+
"logps/rejected": -2611.108642578125,
|
| 1402 |
+
"loss": 0.0685,
|
| 1403 |
+
"rewards/accuracies": 0.9781249761581421,
|
| 1404 |
+
"rewards/chosen": -8.295499801635742,
|
| 1405 |
+
"rewards/margins": 13.880526542663574,
|
| 1406 |
+
"rewards/rejected": -22.176025390625,
|
| 1407 |
+
"step": 890
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 1.868188894654904,
|
| 1411 |
+
"grad_norm": 14.001123095428785,
|
| 1412 |
+
"learning_rate": 6.311390776052527e-09,
|
| 1413 |
+
"logits/chosen": -0.26746782660484314,
|
| 1414 |
+
"logits/rejected": 3.480095386505127,
|
| 1415 |
+
"logps/chosen": -1162.368408203125,
|
| 1416 |
+
"logps/rejected": -2528.66650390625,
|
| 1417 |
+
"loss": 0.0735,
|
| 1418 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1419 |
+
"rewards/chosen": -7.984139442443848,
|
| 1420 |
+
"rewards/margins": 13.350987434387207,
|
| 1421 |
+
"rewards/rejected": -21.335126876831055,
|
| 1422 |
+
"step": 900
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 1.8889465490399586,
|
| 1426 |
+
"grad_norm": 17.75181453766059,
|
| 1427 |
+
"learning_rate": 4.445217264089751e-09,
|
| 1428 |
+
"logits/chosen": -0.3570159077644348,
|
| 1429 |
+
"logits/rejected": 3.7746682167053223,
|
| 1430 |
+
"logps/chosen": -1153.1602783203125,
|
| 1431 |
+
"logps/rejected": -2572.643798828125,
|
| 1432 |
+
"loss": 0.0662,
|
| 1433 |
+
"rewards/accuracies": 0.96875,
|
| 1434 |
+
"rewards/chosen": -7.910887718200684,
|
| 1435 |
+
"rewards/margins": 13.993339538574219,
|
| 1436 |
+
"rewards/rejected": -21.904226303100586,
|
| 1437 |
+
"step": 910
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 1.909704203425013,
|
| 1441 |
+
"grad_norm": 19.779224657698173,
|
| 1442 |
+
"learning_rate": 2.902911863455121e-09,
|
| 1443 |
+
"logits/chosen": -0.23463687300682068,
|
| 1444 |
+
"logits/rejected": 3.8247523307800293,
|
| 1445 |
+
"logps/chosen": -1181.158447265625,
|
| 1446 |
+
"logps/rejected": -2677.92041015625,
|
| 1447 |
+
"loss": 0.0761,
|
| 1448 |
+
"rewards/accuracies": 0.9781249761581421,
|
| 1449 |
+
"rewards/chosen": -8.19698429107666,
|
| 1450 |
+
"rewards/margins": 14.724853515625,
|
| 1451 |
+
"rewards/rejected": -22.92184066772461,
|
| 1452 |
+
"step": 920
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 1.9304618578100676,
|
| 1456 |
+
"grad_norm": 22.12956855667942,
|
| 1457 |
+
"learning_rate": 1.686508757851507e-09,
|
| 1458 |
+
"logits/chosen": -0.2650212049484253,
|
| 1459 |
+
"logits/rejected": 3.5063083171844482,
|
| 1460 |
+
"logps/chosen": -1194.435791015625,
|
| 1461 |
+
"logps/rejected": -2615.01708984375,
|
| 1462 |
+
"loss": 0.0854,
|
| 1463 |
+
"rewards/accuracies": 0.956250011920929,
|
| 1464 |
+
"rewards/chosen": -8.247761726379395,
|
| 1465 |
+
"rewards/margins": 13.95422077178955,
|
| 1466 |
+
"rewards/rejected": -22.201980590820312,
|
| 1467 |
+
"step": 930
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 1.951219512195122,
|
| 1471 |
+
"grad_norm": 20.722879655487088,
|
| 1472 |
+
"learning_rate": 7.976122906031557e-10,
|
| 1473 |
+
"logits/chosen": -0.23710790276527405,
|
| 1474 |
+
"logits/rejected": 3.5924439430236816,
|
| 1475 |
+
"logps/chosen": -1181.043212890625,
|
| 1476 |
+
"logps/rejected": -2523.64794921875,
|
| 1477 |
+
"loss": 0.088,
|
| 1478 |
+
"rewards/accuracies": 0.9437500238418579,
|
| 1479 |
+
"rewards/chosen": -8.123555183410645,
|
| 1480 |
+
"rewards/margins": 13.19268798828125,
|
| 1481 |
+
"rewards/rejected": -21.31624412536621,
|
| 1482 |
+
"step": 940
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 1.9719771665801764,
|
| 1486 |
+
"grad_norm": 12.897702624327897,
|
| 1487 |
+
"learning_rate": 2.37394848648792e-10,
|
| 1488 |
+
"logits/chosen": -0.3340440094470978,
|
| 1489 |
+
"logits/rejected": 3.253843307495117,
|
| 1490 |
+
"logps/chosen": -1149.8099365234375,
|
| 1491 |
+
"logps/rejected": -2490.43994140625,
|
| 1492 |
+
"loss": 0.0801,
|
| 1493 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1494 |
+
"rewards/chosen": -7.880959510803223,
|
| 1495 |
+
"rewards/margins": 13.208969116210938,
|
| 1496 |
+
"rewards/rejected": -21.089927673339844,
|
| 1497 |
+
"step": 950
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 1.992734820965231,
|
| 1501 |
+
"grad_norm": 17.31058461781814,
|
| 1502 |
+
"learning_rate": 6.5953162521614755e-12,
|
| 1503 |
+
"logits/chosen": -0.3153567910194397,
|
| 1504 |
+
"logits/rejected": 3.3712565898895264,
|
| 1505 |
+
"logps/chosen": -1199.875244140625,
|
| 1506 |
+
"logps/rejected": -2640.38525390625,
|
| 1507 |
+
"loss": 0.0743,
|
| 1508 |
+
"rewards/accuracies": 0.981249988079071,
|
| 1509 |
+
"rewards/chosen": -8.392365455627441,
|
| 1510 |
+
"rewards/margins": 14.206585884094238,
|
| 1511 |
+
"rewards/rejected": -22.598949432373047,
|
| 1512 |
+
"step": 960
|
| 1513 |
+
},
|
| 1514 |
+
{
|
| 1515 |
+
"epoch": 1.992734820965231,
|
| 1516 |
+
"eval_logits/chosen": -0.1724339723587036,
|
| 1517 |
+
"eval_logits/rejected": 3.016493320465088,
|
| 1518 |
+
"eval_logps/chosen": -1310.28955078125,
|
| 1519 |
+
"eval_logps/rejected": -2616.585205078125,
|
| 1520 |
+
"eval_loss": 0.15784408152103424,
|
| 1521 |
+
"eval_rewards/accuracies": 0.9203431606292725,
|
| 1522 |
+
"eval_rewards/chosen": -8.868812561035156,
|
| 1523 |
+
"eval_rewards/margins": 12.837637901306152,
|
| 1524 |
+
"eval_rewards/rejected": -21.706451416015625,
|
| 1525 |
+
"eval_runtime": 296.2637,
|
| 1526 |
+
"eval_samples_per_second": 21.91,
|
| 1527 |
+
"eval_steps_per_second": 0.344,
|
| 1528 |
+
"step": 960
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 1.996886351842242,
|
| 1532 |
+
"step": 962,
|
| 1533 |
"total_flos": 0.0,
|
| 1534 |
+
"train_loss": 0.1585804910288655,
|
| 1535 |
+
"train_runtime": 28225.2793,
|
| 1536 |
+
"train_samples_per_second": 8.737,
|
| 1537 |
+
"train_steps_per_second": 0.034
|
| 1538 |
}
|
| 1539 |
],
|
| 1540 |
"logging_steps": 10,
|
| 1541 |
+
"max_steps": 962,
|
| 1542 |
"num_input_tokens_seen": 0,
|
| 1543 |
"num_train_epochs": 2,
|
| 1544 |
"save_steps": 240,
|