---
library_name: transformers
base_model: W-61/llama-3-8b-base-sft-hh-helpful-4xh200
tags:
- alignment-handbook
- new-dpo
- generated_from_trainer
datasets:
- Anthropic/hh-rlhf
model-index:
- name: llama-3-8b-base-new-dpo-hh-helpful-s_star0.6-4xh200-batch-64-20260421-214335-rerun
  results: []
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# llama-3-8b-base-new-dpo-hh-helpful-s_star0.6-4xh200-batch-64-20260421-214335-rerun

This model is a fine-tuned version of [W-61/llama-3-8b-base-sft-hh-helpful-4xh200](https://huggingface.co/W-61/llama-3-8b-base-sft-hh-helpful-4xh200) on the Anthropic/hh-rlhf dataset.
It achieves the following results on the evaluation set:
- Loss: 0.5572
- Fcm Dpo/beta: 0.0034
- Fcm Dpo/q T: 0.3914
- Fcm Dpo/delta: 0.0523
- Fcm Dpo/margin: 156.2329
- Margin Dpo/margin Mean: 156.2329
- Margin Dpo/margin Std: 271.4757
- Logps/chosen: -459.3008
- Logps/rejected: -623.2806
- Logps/ref Chosen: -79.0510
- Logps/ref Rejected: -86.7979
- Logits/chosen: -0.5276
- Logits/rejected: -0.5133

## Model description

More information needed

## Intended uses & limitations

More information needed

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 5e-07
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- total_eval_batch_size: 32
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 1

### Training results

| Training Loss | Epoch  | Step | Validation Loss | Fcm Dpo/beta | Fcm Dpo/q T | Fcm Dpo/delta | Fcm Dpo/margin | Margin Dpo/margin Mean | Margin Dpo/margin Std | Logps/chosen | Logps/rejected | Logps/ref Chosen | Logps/ref Rejected | Logits/chosen | Logits/rejected |
|:-------------:|:------:|:----:|:---------------:|:------------:|:-----------:|:-------------:|:--------------:|:----------------------:|:---------------------:|:------------:|:--------------:|:----------------:|:------------------:|:-------------:|:---------------:|
| 0.9767        | 0.2937 | 200  | 0.5985          | 0.0106       | 0.4061      | 0.0551        | 47.4977        | 47.4977                | 101.7798              | -179.9633    | -235.2079      | -79.0510         | -86.7979           | -0.4084       | -0.3900         |
| 0.9245        | 0.5874 | 400  | 0.5729          | 0.0038       | 0.3973      | 0.0663        | 134.6704       | 134.6704               | 251.6155              | -454.1283    | -596.5456      | -79.0510         | -86.7979           | -0.4387       | -0.4223         |
| 0.9964        | 0.8811 | 600  | 0.5572          | 0.0034       | 0.3914      | 0.0523        | 156.2329       | 156.2329               | 271.4757              | -459.3008    | -623.2806      | -79.0510         | -86.7979           | -0.5276       | -0.5133         |


### Framework versions

- Transformers 4.51.0
- Pytorch 2.3.1+cu121
- Datasets 2.21.0
- Tokenizers 0.21.4