---
library_name: transformers
license: llama3.2
base_model: tanliboy/llama-3.2-3b-sft
tags:
- trl
- dpo
- generated_from_trainer
model-index:
- name: llama-3.2-3b-dpo
  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-3.2-3b-dpo

This model is a fine-tuned version of [tanliboy/llama-3.2-3b-sft](https://huggingface.co/tanliboy/llama-3.2-3b-sft) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.4864
- Rewards/chosen: -1.4549
- Rewards/rejected: -2.6071
- Rewards/accuracies: 0.7247
- Rewards/margins: 1.1522
- Logps/rejected: -560.2480
- Logps/chosen: -483.4225
- Logits/rejected: 0.5181
- Logits/chosen: 0.3637

## 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: 4
- eval_batch_size: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 4
- total_train_batch_size: 128
- total_eval_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 1

### Training results

| Training Loss | Epoch  | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
| 0.6115        | 0.1741 | 100  | 0.6000          | -0.3357        | -0.5534          | 0.6329             | 0.2176          | -354.8722      | -371.5030    | 0.8149          | 0.6445        |
| 0.5416        | 0.3483 | 200  | 0.5221          | -1.2141        | -1.9406          | 0.6741             | 0.7265          | -493.5951      | -459.3420    | 0.6298          | 0.4810        |
| 0.5158        | 0.5224 | 300  | 0.5048          | -1.3617        | -2.3705          | 0.7057             | 1.0088          | -536.5855      | -474.1037    | 0.4784          | 0.3320        |
| 0.501         | 0.6966 | 400  | 0.4906          | -1.4306        | -2.5214          | 0.7152             | 1.0908          | -551.6774      | -480.9883    | 0.5108          | 0.3615        |
| 0.4806        | 0.8707 | 500  | 0.4864          | -1.4549        | -2.6071          | 0.7247             | 1.1522          | -560.2480      | -483.4225    | 0.5181          | 0.3637        |


### Framework versions

- Transformers 4.44.2
- Pytorch 2.4.0+cu121
- Datasets 2.19.1
- Tokenizers 0.19.1