Model save
Browse files- README.md +59 -0
- adapter_model.safetensors +1 -1
- all_results.json +8 -0
- runs/May20_08-11-35_gpu4-119-5/events.out.tfevents.1716156791.gpu4-119-5.477898.0 +2 -2
- train_results.json +8 -0
- trainer_state.json +912 -0
README.md
ADDED
|
@@ -0,0 +1,59 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: peft
|
| 3 |
+
tags:
|
| 4 |
+
- trl
|
| 5 |
+
- dpo
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
base_model: DUAL-GPO/phi-2-irepo-chatml-merged-i0
|
| 8 |
+
model-index:
|
| 9 |
+
- name: phi-2-irepo-chatml-v4-i1
|
| 10 |
+
results: []
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 14 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 15 |
+
|
| 16 |
+
# phi-2-irepo-chatml-v4-i1
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [DUAL-GPO/phi-2-irepo-chatml-merged-i0](https://huggingface.co/DUAL-GPO/phi-2-irepo-chatml-merged-i0) on the None dataset.
|
| 19 |
+
|
| 20 |
+
## Model description
|
| 21 |
+
|
| 22 |
+
More information needed
|
| 23 |
+
|
| 24 |
+
## Intended uses & limitations
|
| 25 |
+
|
| 26 |
+
More information needed
|
| 27 |
+
|
| 28 |
+
## Training and evaluation data
|
| 29 |
+
|
| 30 |
+
More information needed
|
| 31 |
+
|
| 32 |
+
## Training procedure
|
| 33 |
+
|
| 34 |
+
### Training hyperparameters
|
| 35 |
+
|
| 36 |
+
The following hyperparameters were used during training:
|
| 37 |
+
- learning_rate: 5e-06
|
| 38 |
+
- train_batch_size: 4
|
| 39 |
+
- eval_batch_size: 4
|
| 40 |
+
- seed: 42
|
| 41 |
+
- distributed_type: multi-GPU
|
| 42 |
+
- gradient_accumulation_steps: 4
|
| 43 |
+
- total_train_batch_size: 16
|
| 44 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 45 |
+
- lr_scheduler_type: cosine
|
| 46 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 47 |
+
- num_epochs: 1
|
| 48 |
+
|
| 49 |
+
### Training results
|
| 50 |
+
|
| 51 |
+
|
| 52 |
+
|
| 53 |
+
### Framework versions
|
| 54 |
+
|
| 55 |
+
- PEFT 0.7.1
|
| 56 |
+
- Transformers 4.36.2
|
| 57 |
+
- Pytorch 2.1.2+cu121
|
| 58 |
+
- Datasets 2.14.6
|
| 59 |
+
- Tokenizers 0.15.2
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 335579632
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:31dabab1f937fb5cf88ae189818a66124529472f3b6f69a5b95a9dcb115e5be7
|
| 3 |
size 335579632
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.34297545976638794,
|
| 4 |
+
"train_runtime": 3774.4903,
|
| 5 |
+
"train_samples": 10000,
|
| 6 |
+
"train_samples_per_second": 2.649,
|
| 7 |
+
"train_steps_per_second": 0.166
|
| 8 |
+
}
|
runs/May20_08-11-35_gpu4-119-5/events.out.tfevents.1716156791.gpu4-119-5.477898.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9c5442fbc101c2b158a234141e95e69a12e2a0d8c9c79b3737d599716a7f3c67
|
| 3 |
+
size 45062
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.34297545976638794,
|
| 4 |
+
"train_runtime": 3774.4903,
|
| 5 |
+
"train_samples": 10000,
|
| 6 |
+
"train_samples_per_second": 2.649,
|
| 7 |
+
"train_steps_per_second": 0.166
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,912 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 625,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 7.936507936507937e-08,
|
| 14 |
+
"logits/chosen": 0.3197985589504242,
|
| 15 |
+
"logits/rejected": 0.2375701665878296,
|
| 16 |
+
"logps/chosen": -309.8530578613281,
|
| 17 |
+
"logps/rejected": -345.0298156738281,
|
| 18 |
+
"loss": 0.3912,
|
| 19 |
+
"rewards/accuracies": 0.0,
|
| 20 |
+
"rewards/chosen": 0.0,
|
| 21 |
+
"rewards/margins": 0.0,
|
| 22 |
+
"rewards/rejected": 0.0,
|
| 23 |
+
"step": 1
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.02,
|
| 27 |
+
"learning_rate": 7.936507936507937e-07,
|
| 28 |
+
"logits/chosen": 0.13694973289966583,
|
| 29 |
+
"logits/rejected": 0.3060193359851837,
|
| 30 |
+
"logps/chosen": -419.0776062011719,
|
| 31 |
+
"logps/rejected": -389.68438720703125,
|
| 32 |
+
"loss": 0.3933,
|
| 33 |
+
"rewards/accuracies": 0.3194444477558136,
|
| 34 |
+
"rewards/chosen": -0.0009195790044032037,
|
| 35 |
+
"rewards/margins": -8.771694410825148e-05,
|
| 36 |
+
"rewards/rejected": -0.0008318620966747403,
|
| 37 |
+
"step": 10
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.03,
|
| 41 |
+
"learning_rate": 1.5873015873015873e-06,
|
| 42 |
+
"logits/chosen": 0.16718418896198273,
|
| 43 |
+
"logits/rejected": 0.27480465173721313,
|
| 44 |
+
"logps/chosen": -394.2663879394531,
|
| 45 |
+
"logps/rejected": -397.93658447265625,
|
| 46 |
+
"loss": 0.378,
|
| 47 |
+
"rewards/accuracies": 0.3125,
|
| 48 |
+
"rewards/chosen": -0.0013991680461913347,
|
| 49 |
+
"rewards/margins": -8.433303446508944e-05,
|
| 50 |
+
"rewards/rejected": -0.001314834924414754,
|
| 51 |
+
"step": 20
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.05,
|
| 55 |
+
"learning_rate": 2.380952380952381e-06,
|
| 56 |
+
"logits/chosen": 0.24418623745441437,
|
| 57 |
+
"logits/rejected": 0.2495376318693161,
|
| 58 |
+
"logps/chosen": -425.3749084472656,
|
| 59 |
+
"logps/rejected": -403.7620849609375,
|
| 60 |
+
"loss": 0.3871,
|
| 61 |
+
"rewards/accuracies": 0.41874998807907104,
|
| 62 |
+
"rewards/chosen": -0.0024094637483358383,
|
| 63 |
+
"rewards/margins": 6.462958117481321e-05,
|
| 64 |
+
"rewards/rejected": -0.0024740933440625668,
|
| 65 |
+
"step": 30
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.06,
|
| 69 |
+
"learning_rate": 3.1746031746031746e-06,
|
| 70 |
+
"logits/chosen": 0.24052706360816956,
|
| 71 |
+
"logits/rejected": 0.20116524398326874,
|
| 72 |
+
"logps/chosen": -398.9176330566406,
|
| 73 |
+
"logps/rejected": -420.3963928222656,
|
| 74 |
+
"loss": 0.3636,
|
| 75 |
+
"rewards/accuracies": 0.42500001192092896,
|
| 76 |
+
"rewards/chosen": -0.0041575380600988865,
|
| 77 |
+
"rewards/margins": 4.97179280500859e-05,
|
| 78 |
+
"rewards/rejected": -0.004207254853099585,
|
| 79 |
+
"step": 40
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.08,
|
| 83 |
+
"learning_rate": 3.968253968253968e-06,
|
| 84 |
+
"logits/chosen": 0.2002558708190918,
|
| 85 |
+
"logits/rejected": 0.2225637137889862,
|
| 86 |
+
"logps/chosen": -400.6484680175781,
|
| 87 |
+
"logps/rejected": -394.4924011230469,
|
| 88 |
+
"loss": 0.4006,
|
| 89 |
+
"rewards/accuracies": 0.4124999940395355,
|
| 90 |
+
"rewards/chosen": -0.006101720035076141,
|
| 91 |
+
"rewards/margins": 0.0004920283681713045,
|
| 92 |
+
"rewards/rejected": -0.006593747995793819,
|
| 93 |
+
"step": 50
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.1,
|
| 97 |
+
"learning_rate": 4.761904761904762e-06,
|
| 98 |
+
"logits/chosen": 0.24047236144542694,
|
| 99 |
+
"logits/rejected": 0.20159101486206055,
|
| 100 |
+
"logps/chosen": -390.0282287597656,
|
| 101 |
+
"logps/rejected": -406.8359375,
|
| 102 |
+
"loss": 0.3709,
|
| 103 |
+
"rewards/accuracies": 0.42500001192092896,
|
| 104 |
+
"rewards/chosen": -0.005089767277240753,
|
| 105 |
+
"rewards/margins": 0.000577005441300571,
|
| 106 |
+
"rewards/rejected": -0.005666772834956646,
|
| 107 |
+
"step": 60
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.11,
|
| 111 |
+
"learning_rate": 4.998086282661188e-06,
|
| 112 |
+
"logits/chosen": 0.18658548593521118,
|
| 113 |
+
"logits/rejected": 0.23578360676765442,
|
| 114 |
+
"logps/chosen": -413.97216796875,
|
| 115 |
+
"logps/rejected": -415.7422790527344,
|
| 116 |
+
"loss": 0.3699,
|
| 117 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 118 |
+
"rewards/chosen": -0.007824992761015892,
|
| 119 |
+
"rewards/margins": 0.0006475156405940652,
|
| 120 |
+
"rewards/rejected": -0.008472507819533348,
|
| 121 |
+
"step": 70
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.13,
|
| 125 |
+
"learning_rate": 4.988720025682995e-06,
|
| 126 |
+
"logits/chosen": 0.12549163401126862,
|
| 127 |
+
"logits/rejected": 0.11077409982681274,
|
| 128 |
+
"logps/chosen": -370.007568359375,
|
| 129 |
+
"logps/rejected": -357.638427734375,
|
| 130 |
+
"loss": 0.3601,
|
| 131 |
+
"rewards/accuracies": 0.41874998807907104,
|
| 132 |
+
"rewards/chosen": -0.011440587230026722,
|
| 133 |
+
"rewards/margins": 0.0010136803612113,
|
| 134 |
+
"rewards/rejected": -0.012454266659915447,
|
| 135 |
+
"step": 80
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.14,
|
| 139 |
+
"learning_rate": 4.9715789537359126e-06,
|
| 140 |
+
"logits/chosen": 0.03908854350447655,
|
| 141 |
+
"logits/rejected": 0.18181201815605164,
|
| 142 |
+
"logps/chosen": -408.84930419921875,
|
| 143 |
+
"logps/rejected": -391.1748962402344,
|
| 144 |
+
"loss": 0.3651,
|
| 145 |
+
"rewards/accuracies": 0.41874998807907104,
|
| 146 |
+
"rewards/chosen": -0.017708633095026016,
|
| 147 |
+
"rewards/margins": 0.001199634512886405,
|
| 148 |
+
"rewards/rejected": -0.018908265978097916,
|
| 149 |
+
"step": 90
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.16,
|
| 153 |
+
"learning_rate": 4.946716615897932e-06,
|
| 154 |
+
"logits/chosen": 0.15611937642097473,
|
| 155 |
+
"logits/rejected": 0.21340712904930115,
|
| 156 |
+
"logps/chosen": -437.1705017089844,
|
| 157 |
+
"logps/rejected": -466.39544677734375,
|
| 158 |
+
"loss": 0.3702,
|
| 159 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 160 |
+
"rewards/chosen": -0.016876522451639175,
|
| 161 |
+
"rewards/margins": 0.004214240703731775,
|
| 162 |
+
"rewards/rejected": -0.021090764552354813,
|
| 163 |
+
"step": 100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.18,
|
| 167 |
+
"learning_rate": 4.9142106826480114e-06,
|
| 168 |
+
"logits/chosen": 0.1575128138065338,
|
| 169 |
+
"logits/rejected": 0.14739355444908142,
|
| 170 |
+
"logps/chosen": -394.54949951171875,
|
| 171 |
+
"logps/rejected": -417.59930419921875,
|
| 172 |
+
"loss": 0.3609,
|
| 173 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 174 |
+
"rewards/chosen": -0.013111593201756477,
|
| 175 |
+
"rewards/margins": 0.00490961829200387,
|
| 176 |
+
"rewards/rejected": -0.01802121102809906,
|
| 177 |
+
"step": 110
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.19,
|
| 181 |
+
"learning_rate": 4.874162703221823e-06,
|
| 182 |
+
"logits/chosen": 0.20379705727100372,
|
| 183 |
+
"logits/rejected": 0.27394187450408936,
|
| 184 |
+
"logps/chosen": -443.15185546875,
|
| 185 |
+
"logps/rejected": -441.23095703125,
|
| 186 |
+
"loss": 0.3603,
|
| 187 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 188 |
+
"rewards/chosen": -0.025930130854249,
|
| 189 |
+
"rewards/margins": 0.007792765740305185,
|
| 190 |
+
"rewards/rejected": -0.0337228998541832,
|
| 191 |
+
"step": 120
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.21,
|
| 195 |
+
"learning_rate": 4.826697788369752e-06,
|
| 196 |
+
"logits/chosen": 0.10621203482151031,
|
| 197 |
+
"logits/rejected": 0.1633463203907013,
|
| 198 |
+
"logps/chosen": -446.20947265625,
|
| 199 |
+
"logps/rejected": -472.85174560546875,
|
| 200 |
+
"loss": 0.3577,
|
| 201 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 202 |
+
"rewards/chosen": -0.039154283702373505,
|
| 203 |
+
"rewards/margins": 0.009123672731220722,
|
| 204 |
+
"rewards/rejected": -0.04827795922756195,
|
| 205 |
+
"step": 130
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.22,
|
| 209 |
+
"learning_rate": 4.7719642195082224e-06,
|
| 210 |
+
"logits/chosen": 0.11126575618982315,
|
| 211 |
+
"logits/rejected": 0.15343426167964935,
|
| 212 |
+
"logps/chosen": -466.1980895996094,
|
| 213 |
+
"logps/rejected": -454.23876953125,
|
| 214 |
+
"loss": 0.3654,
|
| 215 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 216 |
+
"rewards/chosen": -0.053963106125593185,
|
| 217 |
+
"rewards/margins": 0.010927463881671429,
|
| 218 |
+
"rewards/rejected": -0.06489057093858719,
|
| 219 |
+
"step": 140
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.24,
|
| 223 |
+
"learning_rate": 4.710132985485355e-06,
|
| 224 |
+
"logits/chosen": 0.0875745341181755,
|
| 225 |
+
"logits/rejected": 0.14972081780433655,
|
| 226 |
+
"logps/chosen": -499.6947326660156,
|
| 227 |
+
"logps/rejected": -533.8878173828125,
|
| 228 |
+
"loss": 0.3411,
|
| 229 |
+
"rewards/accuracies": 0.53125,
|
| 230 |
+
"rewards/chosen": -0.08158211410045624,
|
| 231 |
+
"rewards/margins": 0.02510920725762844,
|
| 232 |
+
"rewards/rejected": -0.10669132322072983,
|
| 233 |
+
"step": 150
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.26,
|
| 237 |
+
"learning_rate": 4.641397248408122e-06,
|
| 238 |
+
"logits/chosen": 0.06422760337591171,
|
| 239 |
+
"logits/rejected": 0.15457448363304138,
|
| 240 |
+
"logps/chosen": -565.0504760742188,
|
| 241 |
+
"logps/rejected": -628.6942138671875,
|
| 242 |
+
"loss": 0.3487,
|
| 243 |
+
"rewards/accuracies": 0.550000011920929,
|
| 244 |
+
"rewards/chosen": -0.16031715273857117,
|
| 245 |
+
"rewards/margins": 0.04973132163286209,
|
| 246 |
+
"rewards/rejected": -0.21004848182201385,
|
| 247 |
+
"step": 160
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.27,
|
| 251 |
+
"learning_rate": 4.5659717401997655e-06,
|
| 252 |
+
"logits/chosen": 0.14616025984287262,
|
| 253 |
+
"logits/rejected": 0.035494547337293625,
|
| 254 |
+
"logps/chosen": -591.8494873046875,
|
| 255 |
+
"logps/rejected": -686.7391357421875,
|
| 256 |
+
"loss": 0.342,
|
| 257 |
+
"rewards/accuracies": 0.5,
|
| 258 |
+
"rewards/chosen": -0.20710352063179016,
|
| 259 |
+
"rewards/margins": 0.07382966578006744,
|
| 260 |
+
"rewards/rejected": -0.2809332013130188,
|
| 261 |
+
"step": 170
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.29,
|
| 265 |
+
"learning_rate": 4.4840920917726425e-06,
|
| 266 |
+
"logits/chosen": 0.02610364928841591,
|
| 267 |
+
"logits/rejected": 0.07332785427570343,
|
| 268 |
+
"logps/chosen": -585.5015869140625,
|
| 269 |
+
"logps/rejected": -654.242919921875,
|
| 270 |
+
"loss": 0.3385,
|
| 271 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 272 |
+
"rewards/chosen": -0.19318720698356628,
|
| 273 |
+
"rewards/margins": 0.08042450994253159,
|
| 274 |
+
"rewards/rejected": -0.27361172437667847,
|
| 275 |
+
"step": 180
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.3,
|
| 279 |
+
"learning_rate": 4.396014096912182e-06,
|
| 280 |
+
"logits/chosen": -0.055401187390089035,
|
| 281 |
+
"logits/rejected": 0.06281381845474243,
|
| 282 |
+
"logps/chosen": -607.0491333007812,
|
| 283 |
+
"logps/rejected": -655.4883422851562,
|
| 284 |
+
"loss": 0.3587,
|
| 285 |
+
"rewards/accuracies": 0.46875,
|
| 286 |
+
"rewards/chosen": -0.179766446352005,
|
| 287 |
+
"rewards/margins": 0.04076399281620979,
|
| 288 |
+
"rewards/rejected": -0.2205304205417633,
|
| 289 |
+
"step": 190
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.32,
|
| 293 |
+
"learning_rate": 4.302012913171584e-06,
|
| 294 |
+
"logits/chosen": 0.041515715420246124,
|
| 295 |
+
"logits/rejected": 0.10322318226099014,
|
| 296 |
+
"logps/chosen": -518.5902709960938,
|
| 297 |
+
"logps/rejected": -555.7604370117188,
|
| 298 |
+
"loss": 0.3493,
|
| 299 |
+
"rewards/accuracies": 0.4312500059604645,
|
| 300 |
+
"rewards/chosen": -0.14564794301986694,
|
| 301 |
+
"rewards/margins": 0.04818274453282356,
|
| 302 |
+
"rewards/rejected": -0.1938306838274002,
|
| 303 |
+
"step": 200
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.34,
|
| 307 |
+
"learning_rate": 4.202382202273702e-06,
|
| 308 |
+
"logits/chosen": 0.060168325901031494,
|
| 309 |
+
"logits/rejected": 0.05509088560938835,
|
| 310 |
+
"logps/chosen": -546.6846923828125,
|
| 311 |
+
"logps/rejected": -593.6578369140625,
|
| 312 |
+
"loss": 0.3723,
|
| 313 |
+
"rewards/accuracies": 0.5,
|
| 314 |
+
"rewards/chosen": -0.14728133380413055,
|
| 315 |
+
"rewards/margins": 0.04971260204911232,
|
| 316 |
+
"rewards/rejected": -0.19699391722679138,
|
| 317 |
+
"step": 210
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.35,
|
| 321 |
+
"learning_rate": 4.097433212705492e-06,
|
| 322 |
+
"logits/chosen": 0.07657967507839203,
|
| 323 |
+
"logits/rejected": 0.014337155036628246,
|
| 324 |
+
"logps/chosen": -568.3507080078125,
|
| 325 |
+
"logps/rejected": -619.3447875976562,
|
| 326 |
+
"loss": 0.3164,
|
| 327 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 328 |
+
"rewards/chosen": -0.16198821365833282,
|
| 329 |
+
"rewards/margins": 0.06240609288215637,
|
| 330 |
+
"rewards/rejected": -0.2243943214416504,
|
| 331 |
+
"step": 220
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.37,
|
| 335 |
+
"learning_rate": 3.987493807371033e-06,
|
| 336 |
+
"logits/chosen": -0.006547008641064167,
|
| 337 |
+
"logits/rejected": 0.13627514243125916,
|
| 338 |
+
"logps/chosen": -569.6785278320312,
|
| 339 |
+
"logps/rejected": -639.2614135742188,
|
| 340 |
+
"loss": 0.3225,
|
| 341 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 342 |
+
"rewards/chosen": -0.17529377341270447,
|
| 343 |
+
"rewards/margins": 0.0657162293791771,
|
| 344 |
+
"rewards/rejected": -0.24100999534130096,
|
| 345 |
+
"step": 230
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.38,
|
| 349 |
+
"learning_rate": 3.872907439340758e-06,
|
| 350 |
+
"logits/chosen": -0.017886776477098465,
|
| 351 |
+
"logits/rejected": 0.0037322058342397213,
|
| 352 |
+
"logps/chosen": -559.8616943359375,
|
| 353 |
+
"logps/rejected": -608.9739990234375,
|
| 354 |
+
"loss": 0.332,
|
| 355 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 356 |
+
"rewards/chosen": -0.18361715972423553,
|
| 357 |
+
"rewards/margins": 0.05353798344731331,
|
| 358 |
+
"rewards/rejected": -0.23715515434741974,
|
| 359 |
+
"step": 240
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.4,
|
| 363 |
+
"learning_rate": 3.75403207889666e-06,
|
| 364 |
+
"logits/chosen": -0.013735148124396801,
|
| 365 |
+
"logits/rejected": 0.019972026348114014,
|
| 366 |
+
"logps/chosen": -543.68017578125,
|
| 367 |
+
"logps/rejected": -580.9366455078125,
|
| 368 |
+
"loss": 0.3555,
|
| 369 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 370 |
+
"rewards/chosen": -0.17838358879089355,
|
| 371 |
+
"rewards/margins": 0.04401790350675583,
|
| 372 |
+
"rewards/rejected": -0.2224014699459076,
|
| 373 |
+
"step": 250
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.42,
|
| 377 |
+
"learning_rate": 3.631239095225417e-06,
|
| 378 |
+
"logits/chosen": -0.046024613082408905,
|
| 379 |
+
"logits/rejected": 0.13009200990200043,
|
| 380 |
+
"logps/chosen": -629.1693115234375,
|
| 381 |
+
"logps/rejected": -624.858642578125,
|
| 382 |
+
"loss": 0.3556,
|
| 383 |
+
"rewards/accuracies": 0.4437499940395355,
|
| 384 |
+
"rewards/chosen": -0.19326834380626678,
|
| 385 |
+
"rewards/margins": 0.042895831167697906,
|
| 386 |
+
"rewards/rejected": -0.2361641824245453,
|
| 387 |
+
"step": 260
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.43,
|
| 391 |
+
"learning_rate": 3.5049120962530608e-06,
|
| 392 |
+
"logits/chosen": 0.04170585796236992,
|
| 393 |
+
"logits/rejected": 0.03916405886411667,
|
| 394 |
+
"logps/chosen": -612.9544067382812,
|
| 395 |
+
"logps/rejected": -684.2191162109375,
|
| 396 |
+
"loss": 0.3172,
|
| 397 |
+
"rewards/accuracies": 0.550000011920929,
|
| 398 |
+
"rewards/chosen": -0.18931707739830017,
|
| 399 |
+
"rewards/margins": 0.06901886314153671,
|
| 400 |
+
"rewards/rejected": -0.2583359181880951,
|
| 401 |
+
"step": 270
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.45,
|
| 405 |
+
"learning_rate": 3.3754457302455464e-06,
|
| 406 |
+
"logits/chosen": -0.016118571162223816,
|
| 407 |
+
"logits/rejected": 0.050165075808763504,
|
| 408 |
+
"logps/chosen": -590.7628173828125,
|
| 409 |
+
"logps/rejected": -657.3172607421875,
|
| 410 |
+
"loss": 0.3254,
|
| 411 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 412 |
+
"rewards/chosen": -0.1856985241174698,
|
| 413 |
+
"rewards/margins": 0.061163417994976044,
|
| 414 |
+
"rewards/rejected": -0.24686193466186523,
|
| 415 |
+
"step": 280
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.46,
|
| 419 |
+
"learning_rate": 3.2432444529190714e-06,
|
| 420 |
+
"logits/chosen": -0.013759548775851727,
|
| 421 |
+
"logits/rejected": 0.10973884910345078,
|
| 422 |
+
"logps/chosen": -643.6578979492188,
|
| 423 |
+
"logps/rejected": -724.0432739257812,
|
| 424 |
+
"loss": 0.3321,
|
| 425 |
+
"rewards/accuracies": 0.5,
|
| 426 |
+
"rewards/chosen": -0.21854694187641144,
|
| 427 |
+
"rewards/margins": 0.08554595708847046,
|
| 428 |
+
"rewards/rejected": -0.3040928840637207,
|
| 429 |
+
"step": 290
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.48,
|
| 433 |
+
"learning_rate": 3.1087212639117057e-06,
|
| 434 |
+
"logits/chosen": 0.05435953661799431,
|
| 435 |
+
"logits/rejected": 0.04930294305086136,
|
| 436 |
+
"logps/chosen": -610.7354125976562,
|
| 437 |
+
"logps/rejected": -708.9710693359375,
|
| 438 |
+
"loss": 0.3449,
|
| 439 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 440 |
+
"rewards/chosen": -0.22193463146686554,
|
| 441 |
+
"rewards/margins": 0.09556058794260025,
|
| 442 |
+
"rewards/rejected": -0.317495197057724,
|
| 443 |
+
"step": 300
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.5,
|
| 447 |
+
"learning_rate": 2.9722964165636263e-06,
|
| 448 |
+
"logits/chosen": 0.016163092106580734,
|
| 449 |
+
"logits/rejected": -0.008114620111882687,
|
| 450 |
+
"logps/chosen": -599.1492309570312,
|
| 451 |
+
"logps/rejected": -699.5606689453125,
|
| 452 |
+
"loss": 0.3026,
|
| 453 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 454 |
+
"rewards/chosen": -0.21535012125968933,
|
| 455 |
+
"rewards/margins": 0.09686703979969025,
|
| 456 |
+
"rewards/rejected": -0.31221720576286316,
|
| 457 |
+
"step": 310
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.51,
|
| 461 |
+
"learning_rate": 2.8343961050366275e-06,
|
| 462 |
+
"logits/chosen": -0.02114369347691536,
|
| 463 |
+
"logits/rejected": 0.00704212486743927,
|
| 464 |
+
"logps/chosen": -597.2741088867188,
|
| 465 |
+
"logps/rejected": -663.7066650390625,
|
| 466 |
+
"loss": 0.2955,
|
| 467 |
+
"rewards/accuracies": 0.512499988079071,
|
| 468 |
+
"rewards/chosen": -0.20012113451957703,
|
| 469 |
+
"rewards/margins": 0.08274127542972565,
|
| 470 |
+
"rewards/rejected": -0.28286242485046387,
|
| 471 |
+
"step": 320
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.53,
|
| 475 |
+
"learning_rate": 2.695451132874385e-06,
|
| 476 |
+
"logits/chosen": -0.005913554225116968,
|
| 477 |
+
"logits/rejected": -0.03215361759066582,
|
| 478 |
+
"logps/chosen": -634.7625122070312,
|
| 479 |
+
"logps/rejected": -682.1895751953125,
|
| 480 |
+
"loss": 0.349,
|
| 481 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 482 |
+
"rewards/chosen": -0.2384355515241623,
|
| 483 |
+
"rewards/margins": 0.057650744915008545,
|
| 484 |
+
"rewards/rejected": -0.29608631134033203,
|
| 485 |
+
"step": 330
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.54,
|
| 489 |
+
"learning_rate": 2.5558955671628964e-06,
|
| 490 |
+
"logits/chosen": 0.014790521934628487,
|
| 491 |
+
"logits/rejected": 0.03520793467760086,
|
| 492 |
+
"logps/chosen": -708.5465087890625,
|
| 493 |
+
"logps/rejected": -726.1888427734375,
|
| 494 |
+
"loss": 0.3108,
|
| 495 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 496 |
+
"rewards/chosen": -0.24060730636119843,
|
| 497 |
+
"rewards/margins": 0.06836961209774017,
|
| 498 |
+
"rewards/rejected": -0.3089769184589386,
|
| 499 |
+
"step": 340
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.56,
|
| 503 |
+
"learning_rate": 2.4161653824955654e-06,
|
| 504 |
+
"logits/chosen": -0.039772890508174896,
|
| 505 |
+
"logits/rejected": 0.04145686700940132,
|
| 506 |
+
"logps/chosen": -643.654296875,
|
| 507 |
+
"logps/rejected": -722.4707641601562,
|
| 508 |
+
"loss": 0.2987,
|
| 509 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 510 |
+
"rewards/chosen": -0.2103274166584015,
|
| 511 |
+
"rewards/margins": 0.08200379461050034,
|
| 512 |
+
"rewards/rejected": -0.29233118891716003,
|
| 513 |
+
"step": 350
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.58,
|
| 517 |
+
"learning_rate": 2.2766970989791697e-06,
|
| 518 |
+
"logits/chosen": -0.07129567116498947,
|
| 519 |
+
"logits/rejected": 0.038871683180332184,
|
| 520 |
+
"logps/chosen": -605.6844482421875,
|
| 521 |
+
"logps/rejected": -640.6483764648438,
|
| 522 |
+
"loss": 0.3339,
|
| 523 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 524 |
+
"rewards/chosen": -0.21578896045684814,
|
| 525 |
+
"rewards/margins": 0.06505810469388962,
|
| 526 |
+
"rewards/rejected": -0.28084704279899597,
|
| 527 |
+
"step": 360
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.59,
|
| 531 |
+
"learning_rate": 2.1379264185356545e-06,
|
| 532 |
+
"logits/chosen": -0.062216080725193024,
|
| 533 |
+
"logits/rejected": -0.024451395496726036,
|
| 534 |
+
"logps/chosen": -646.650634765625,
|
| 535 |
+
"logps/rejected": -700.7503051757812,
|
| 536 |
+
"loss": 0.3549,
|
| 537 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 538 |
+
"rewards/chosen": -0.24396809935569763,
|
| 539 |
+
"rewards/margins": 0.07057716697454453,
|
| 540 |
+
"rewards/rejected": -0.31454527378082275,
|
| 541 |
+
"step": 370
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.61,
|
| 545 |
+
"learning_rate": 2.000286863759934e-06,
|
| 546 |
+
"logits/chosen": -0.08249249309301376,
|
| 547 |
+
"logits/rejected": 0.03427312523126602,
|
| 548 |
+
"logps/chosen": -680.2882690429688,
|
| 549 |
+
"logps/rejected": -714.1500244140625,
|
| 550 |
+
"loss": 0.3508,
|
| 551 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 552 |
+
"rewards/chosen": -0.2394016683101654,
|
| 553 |
+
"rewards/margins": 0.05815623328089714,
|
| 554 |
+
"rewards/rejected": -0.29755789041519165,
|
| 555 |
+
"step": 380
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.62,
|
| 559 |
+
"learning_rate": 1.8642084235859764e-06,
|
| 560 |
+
"logits/chosen": -0.034602344036102295,
|
| 561 |
+
"logits/rejected": 0.007808700203895569,
|
| 562 |
+
"logps/chosen": -624.1826171875,
|
| 563 |
+
"logps/rejected": -688.9442138671875,
|
| 564 |
+
"loss": 0.3546,
|
| 565 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 566 |
+
"rewards/chosen": -0.23162564635276794,
|
| 567 |
+
"rewards/margins": 0.046284474432468414,
|
| 568 |
+
"rewards/rejected": -0.27791014313697815,
|
| 569 |
+
"step": 390
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.64,
|
| 573 |
+
"learning_rate": 1.7301162099921013e-06,
|
| 574 |
+
"logits/chosen": -0.10363703966140747,
|
| 575 |
+
"logits/rejected": 0.05012492090463638,
|
| 576 |
+
"logps/chosen": -563.6665649414062,
|
| 577 |
+
"logps/rejected": -604.61572265625,
|
| 578 |
+
"loss": 0.3045,
|
| 579 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 580 |
+
"rewards/chosen": -0.19194665551185608,
|
| 581 |
+
"rewards/margins": 0.07340183109045029,
|
| 582 |
+
"rewards/rejected": -0.26534849405288696,
|
| 583 |
+
"step": 400
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.66,
|
| 587 |
+
"learning_rate": 1.5984291299420117e-06,
|
| 588 |
+
"logits/chosen": 0.026764582842588425,
|
| 589 |
+
"logits/rejected": 0.04089599847793579,
|
| 590 |
+
"logps/chosen": -624.69970703125,
|
| 591 |
+
"logps/rejected": -700.4164428710938,
|
| 592 |
+
"loss": 0.3382,
|
| 593 |
+
"rewards/accuracies": 0.4124999940395355,
|
| 594 |
+
"rewards/chosen": -0.21267059445381165,
|
| 595 |
+
"rewards/margins": 0.0644538626074791,
|
| 596 |
+
"rewards/rejected": -0.27712446451187134,
|
| 597 |
+
"step": 410
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.67,
|
| 601 |
+
"learning_rate": 1.4695585767104092e-06,
|
| 602 |
+
"logits/chosen": -0.08264167606830597,
|
| 603 |
+
"logits/rejected": 0.056310832500457764,
|
| 604 |
+
"logps/chosen": -592.313720703125,
|
| 605 |
+
"logps/rejected": -640.4153442382812,
|
| 606 |
+
"loss": 0.3128,
|
| 607 |
+
"rewards/accuracies": 0.5,
|
| 608 |
+
"rewards/chosen": -0.20638258755207062,
|
| 609 |
+
"rewards/margins": 0.06300722062587738,
|
| 610 |
+
"rewards/rejected": -0.269389808177948,
|
| 611 |
+
"step": 420
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.69,
|
| 615 |
+
"learning_rate": 1.3439071446815452e-06,
|
| 616 |
+
"logits/chosen": 0.007897214964032173,
|
| 617 |
+
"logits/rejected": -0.022024786099791527,
|
| 618 |
+
"logps/chosen": -576.2522583007812,
|
| 619 |
+
"logps/rejected": -674.4778442382812,
|
| 620 |
+
"loss": 0.3061,
|
| 621 |
+
"rewards/accuracies": 0.53125,
|
| 622 |
+
"rewards/chosen": -0.19270029664039612,
|
| 623 |
+
"rewards/margins": 0.08912625163793564,
|
| 624 |
+
"rewards/rejected": -0.28182655572891235,
|
| 625 |
+
"step": 430
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.7,
|
| 629 |
+
"learning_rate": 1.2218673716356919e-06,
|
| 630 |
+
"logits/chosen": -0.07224872708320618,
|
| 631 |
+
"logits/rejected": -0.04243229702115059,
|
| 632 |
+
"logps/chosen": -667.8207397460938,
|
| 633 |
+
"logps/rejected": -694.7523193359375,
|
| 634 |
+
"loss": 0.3343,
|
| 635 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 636 |
+
"rewards/chosen": -0.21047215163707733,
|
| 637 |
+
"rewards/margins": 0.05125568434596062,
|
| 638 |
+
"rewards/rejected": -0.26172783970832825,
|
| 639 |
+
"step": 440
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.72,
|
| 643 |
+
"learning_rate": 1.103820512452661e-06,
|
| 644 |
+
"logits/chosen": -0.004954156465828419,
|
| 645 |
+
"logits/rejected": -0.06225745752453804,
|
| 646 |
+
"logps/chosen": -611.6270141601562,
|
| 647 |
+
"logps/rejected": -663.220703125,
|
| 648 |
+
"loss": 0.3585,
|
| 649 |
+
"rewards/accuracies": 0.38749998807907104,
|
| 650 |
+
"rewards/chosen": -0.22025656700134277,
|
| 651 |
+
"rewards/margins": 0.04412217065691948,
|
| 652 |
+
"rewards/rejected": -0.26437875628471375,
|
| 653 |
+
"step": 450
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.74,
|
| 657 |
+
"learning_rate": 9.901353480633468e-07,
|
| 658 |
+
"logits/chosen": -0.05928473919630051,
|
| 659 |
+
"logits/rejected": -0.07472564280033112,
|
| 660 |
+
"logps/chosen": -678.4139404296875,
|
| 661 |
+
"logps/rejected": -729.8592529296875,
|
| 662 |
+
"loss": 0.3345,
|
| 663 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 664 |
+
"rewards/chosen": -0.23494932055473328,
|
| 665 |
+
"rewards/margins": 0.06922940909862518,
|
| 666 |
+
"rewards/rejected": -0.30417874455451965,
|
| 667 |
+
"step": 460
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.75,
|
| 671 |
+
"learning_rate": 8.811670333701544e-07,
|
| 672 |
+
"logits/chosen": 0.0089465556666255,
|
| 673 |
+
"logits/rejected": -0.0320318266749382,
|
| 674 |
+
"logps/chosen": -632.8341064453125,
|
| 675 |
+
"logps/rejected": -666.9921875,
|
| 676 |
+
"loss": 0.3629,
|
| 677 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 678 |
+
"rewards/chosen": -0.23375169932842255,
|
| 679 |
+
"rewards/margins": 0.047677937895059586,
|
| 680 |
+
"rewards/rejected": -0.28142961859703064,
|
| 681 |
+
"step": 470
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.77,
|
| 685 |
+
"learning_rate": 7.772559877354341e-07,
|
| 686 |
+
"logits/chosen": 0.01971885748207569,
|
| 687 |
+
"logits/rejected": 0.05568293482065201,
|
| 688 |
+
"logps/chosen": -603.0645751953125,
|
| 689 |
+
"logps/rejected": -670.7252807617188,
|
| 690 |
+
"loss": 0.3554,
|
| 691 |
+
"rewards/accuracies": 0.4312500059604645,
|
| 692 |
+
"rewards/chosen": -0.20837394893169403,
|
| 693 |
+
"rewards/margins": 0.04435722157359123,
|
| 694 |
+
"rewards/rejected": -0.25273117423057556,
|
| 695 |
+
"step": 480
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.78,
|
| 699 |
+
"learning_rate": 6.787268315040604e-07,
|
| 700 |
+
"logits/chosen": 0.04185909777879715,
|
| 701 |
+
"logits/rejected": -0.008848948404192924,
|
| 702 |
+
"logps/chosen": -616.8929443359375,
|
| 703 |
+
"logps/rejected": -697.3258666992188,
|
| 704 |
+
"loss": 0.3378,
|
| 705 |
+
"rewards/accuracies": 0.5,
|
| 706 |
+
"rewards/chosen": -0.19609177112579346,
|
| 707 |
+
"rewards/margins": 0.06852120906114578,
|
| 708 |
+
"rewards/rejected": -0.26461294293403625,
|
| 709 |
+
"step": 490
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.8,
|
| 713 |
+
"learning_rate": 5.858873718824829e-07,
|
| 714 |
+
"logits/chosen": -0.1592080146074295,
|
| 715 |
+
"logits/rejected": -0.008233757689595222,
|
| 716 |
+
"logps/chosen": -566.5294799804688,
|
| 717 |
+
"logps/rejected": -622.8501586914062,
|
| 718 |
+
"loss": 0.3033,
|
| 719 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 720 |
+
"rewards/chosen": -0.18146850168704987,
|
| 721 |
+
"rewards/margins": 0.07721823453903198,
|
| 722 |
+
"rewards/rejected": -0.25868672132492065,
|
| 723 |
+
"step": 500
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.82,
|
| 727 |
+
"learning_rate": 4.990276413423817e-07,
|
| 728 |
+
"logits/chosen": -0.11390354484319687,
|
| 729 |
+
"logits/rejected": -0.022463113069534302,
|
| 730 |
+
"logps/chosen": -630.0293579101562,
|
| 731 |
+
"logps/rejected": -702.0118408203125,
|
| 732 |
+
"loss": 0.3212,
|
| 733 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 734 |
+
"rewards/chosen": -0.20778632164001465,
|
| 735 |
+
"rewards/margins": 0.07734315097332001,
|
| 736 |
+
"rewards/rejected": -0.28512945771217346,
|
| 737 |
+
"step": 510
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.83,
|
| 741 |
+
"learning_rate": 4.184189915529796e-07,
|
| 742 |
+
"logits/chosen": -0.11935200542211533,
|
| 743 |
+
"logits/rejected": -0.05328110605478287,
|
| 744 |
+
"logps/chosen": -697.4544677734375,
|
| 745 |
+
"logps/rejected": -787.1935424804688,
|
| 746 |
+
"loss": 0.316,
|
| 747 |
+
"rewards/accuracies": 0.543749988079071,
|
| 748 |
+
"rewards/chosen": -0.22982342541217804,
|
| 749 |
+
"rewards/margins": 0.09117700159549713,
|
| 750 |
+
"rewards/rejected": -0.3210003972053528,
|
| 751 |
+
"step": 520
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.85,
|
| 755 |
+
"learning_rate": 3.4431324567258176e-07,
|
| 756 |
+
"logits/chosen": -0.04139793664216995,
|
| 757 |
+
"logits/rejected": 0.02297029457986355,
|
| 758 |
+
"logps/chosen": -595.4196166992188,
|
| 759 |
+
"logps/rejected": -648.87353515625,
|
| 760 |
+
"loss": 0.3245,
|
| 761 |
+
"rewards/accuracies": 0.42500001192092896,
|
| 762 |
+
"rewards/chosen": -0.20907723903656006,
|
| 763 |
+
"rewards/margins": 0.06023558974266052,
|
| 764 |
+
"rewards/rejected": -0.2693127989768982,
|
| 765 |
+
"step": 530
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.86,
|
| 769 |
+
"learning_rate": 2.769419116476052e-07,
|
| 770 |
+
"logits/chosen": -0.012856354005634785,
|
| 771 |
+
"logits/rejected": 0.03404129296541214,
|
| 772 |
+
"logps/chosen": -591.6539306640625,
|
| 773 |
+
"logps/rejected": -639.3442993164062,
|
| 774 |
+
"loss": 0.3519,
|
| 775 |
+
"rewards/accuracies": 0.46875,
|
| 776 |
+
"rewards/chosen": -0.21816472709178925,
|
| 777 |
+
"rewards/margins": 0.05062199383974075,
|
| 778 |
+
"rewards/rejected": -0.2687866985797882,
|
| 779 |
+
"step": 540
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.88,
|
| 783 |
+
"learning_rate": 2.1651545897676512e-07,
|
| 784 |
+
"logits/chosen": -0.1339225172996521,
|
| 785 |
+
"logits/rejected": -6.989687972236425e-05,
|
| 786 |
+
"logps/chosen": -603.4436645507812,
|
| 787 |
+
"logps/rejected": -656.8947143554688,
|
| 788 |
+
"loss": 0.3171,
|
| 789 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 790 |
+
"rewards/chosen": -0.20348867774009705,
|
| 791 |
+
"rewards/margins": 0.06371064484119415,
|
| 792 |
+
"rewards/rejected": -0.2671993374824524,
|
| 793 |
+
"step": 550
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.9,
|
| 797 |
+
"learning_rate": 1.6322266119983222e-07,
|
| 798 |
+
"logits/chosen": -0.010643618181347847,
|
| 799 |
+
"logits/rejected": -0.10318191349506378,
|
| 800 |
+
"logps/chosen": -538.0825805664062,
|
| 801 |
+
"logps/rejected": -612.8212280273438,
|
| 802 |
+
"loss": 0.3428,
|
| 803 |
+
"rewards/accuracies": 0.4124999940395355,
|
| 804 |
+
"rewards/chosen": -0.18894371390342712,
|
| 805 |
+
"rewards/margins": 0.05722329765558243,
|
| 806 |
+
"rewards/rejected": -0.24616701900959015,
|
| 807 |
+
"step": 560
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.91,
|
| 811 |
+
"learning_rate": 1.1723000616502167e-07,
|
| 812 |
+
"logits/chosen": -0.05230097100138664,
|
| 813 |
+
"logits/rejected": 0.0051549882628023624,
|
| 814 |
+
"logps/chosen": -597.4989013671875,
|
| 815 |
+
"logps/rejected": -646.1577758789062,
|
| 816 |
+
"loss": 0.3383,
|
| 817 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 818 |
+
"rewards/chosen": -0.20286564528942108,
|
| 819 |
+
"rewards/margins": 0.06327156722545624,
|
| 820 |
+
"rewards/rejected": -0.2661372423171997,
|
| 821 |
+
"step": 570
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.93,
|
| 825 |
+
"learning_rate": 7.868117591737585e-08,
|
| 826 |
+
"logits/chosen": -0.11382152885198593,
|
| 827 |
+
"logits/rejected": 0.017693499103188515,
|
| 828 |
+
"logps/chosen": -600.3404541015625,
|
| 829 |
+
"logps/rejected": -700.9383544921875,
|
| 830 |
+
"loss": 0.3332,
|
| 831 |
+
"rewards/accuracies": 0.48124998807907104,
|
| 832 |
+
"rewards/chosen": -0.20399217307567596,
|
| 833 |
+
"rewards/margins": 0.08086130768060684,
|
| 834 |
+
"rewards/rejected": -0.284853458404541,
|
| 835 |
+
"step": 580
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.94,
|
| 839 |
+
"learning_rate": 4.769659783295383e-08,
|
| 840 |
+
"logits/chosen": -0.041316356509923935,
|
| 841 |
+
"logits/rejected": -0.09137067198753357,
|
| 842 |
+
"logps/chosen": -627.2562255859375,
|
| 843 |
+
"logps/rejected": -690.1580200195312,
|
| 844 |
+
"loss": 0.3578,
|
| 845 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 846 |
+
"rewards/chosen": -0.22119614481925964,
|
| 847 |
+
"rewards/margins": 0.04840771108865738,
|
| 848 |
+
"rewards/rejected": -0.2696038782596588,
|
| 849 |
+
"step": 590
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.96,
|
| 853 |
+
"learning_rate": 2.4373068401120358e-08,
|
| 854 |
+
"logits/chosen": -0.048679061233997345,
|
| 855 |
+
"logits/rejected": 0.08877386897802353,
|
| 856 |
+
"logps/chosen": -593.13134765625,
|
| 857 |
+
"logps/rejected": -638.4296264648438,
|
| 858 |
+
"loss": 0.36,
|
| 859 |
+
"rewards/accuracies": 0.41874998807907104,
|
| 860 |
+
"rewards/chosen": -0.21022383868694305,
|
| 861 |
+
"rewards/margins": 0.04604775831103325,
|
| 862 |
+
"rewards/rejected": -0.2562715709209442,
|
| 863 |
+
"step": 600
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.98,
|
| 867 |
+
"learning_rate": 8.78345083022425e-09,
|
| 868 |
+
"logits/chosen": 0.013545943424105644,
|
| 869 |
+
"logits/rejected": 0.013492127880454063,
|
| 870 |
+
"logps/chosen": -616.1806640625,
|
| 871 |
+
"logps/rejected": -662.4678955078125,
|
| 872 |
+
"loss": 0.3495,
|
| 873 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 874 |
+
"rewards/chosen": -0.217128723859787,
|
| 875 |
+
"rewards/margins": 0.04821699112653732,
|
| 876 |
+
"rewards/rejected": -0.2653457224369049,
|
| 877 |
+
"step": 610
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.99,
|
| 881 |
+
"learning_rate": 9.764474213677654e-10,
|
| 882 |
+
"logits/chosen": -0.05919772386550903,
|
| 883 |
+
"logits/rejected": 0.03523395583033562,
|
| 884 |
+
"logps/chosen": -687.3334350585938,
|
| 885 |
+
"logps/rejected": -745.0026245117188,
|
| 886 |
+
"loss": 0.3215,
|
| 887 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 888 |
+
"rewards/chosen": -0.2285897433757782,
|
| 889 |
+
"rewards/margins": 0.0788644552230835,
|
| 890 |
+
"rewards/rejected": -0.3074541687965393,
|
| 891 |
+
"step": 620
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 1.0,
|
| 895 |
+
"step": 625,
|
| 896 |
+
"total_flos": 0.0,
|
| 897 |
+
"train_loss": 0.34297545976638794,
|
| 898 |
+
"train_runtime": 3774.4903,
|
| 899 |
+
"train_samples_per_second": 2.649,
|
| 900 |
+
"train_steps_per_second": 0.166
|
| 901 |
+
}
|
| 902 |
+
],
|
| 903 |
+
"logging_steps": 10,
|
| 904 |
+
"max_steps": 625,
|
| 905 |
+
"num_input_tokens_seen": 0,
|
| 906 |
+
"num_train_epochs": 1,
|
| 907 |
+
"save_steps": 100,
|
| 908 |
+
"total_flos": 0.0,
|
| 909 |
+
"train_batch_size": 4,
|
| 910 |
+
"trial_name": null,
|
| 911 |
+
"trial_params": null
|
| 912 |
+
}
|