Files
qwen3-4b-dpo-v1/README.md
ModelHub XC 0c2583e19d 初始化项目,由ModelHub XC社区提供模型
Model: takami2022/qwen3-4b-dpo-v1
Source: Original Platform
2026-08-20 16:39:29 +08:00

34 lines
789 B
Markdown

---
base_model: takami2022/qwen3-4b-sft-merged-v2v5ver1
datasets:
- u-10bei/dpo-dataset-qwen-cot
language:
- en
license: apache-2.0
library_name: transformers
pipeline_tag: text-generation
tags:
- dpo
- unsloth
- qwen
- alignment
---
# qwen3-4b-dpo-v1
This model is a fine-tuned version of **takami2022/qwen3-4b-sft-merged-v2v5ver1** using **Direct Preference Optimization (DPO)** via the **Unsloth** library.
This repository contains the **full-merged 16-bit weights**. No adapter loading is required.
## Training Configuration
- **Base model**: takami2022/qwen3-4b-sft-merged-v2v5ver1
- **Method**: DPO (Direct Preference Optimization)
- **Epochs**: 1
- **Learning rate**: 1e-07
- **Beta**: 0.1
- **Max sequence length**: 1024
- **LoRA Config**: r=16, alpha=32 (merged into base)