---
base_model: unsloth/Qwen3-4B-Instruct-2507
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
---

# dpo-qwen3_4b-cot-merged_v260302-010243

This model is a fine-tuned version of **unsloth/Qwen3-4B-Instruct-2507** using **Supervised Fine-Tuning (SFT)** and **Direct Preference Optimization (DPO)** via the **Unsloth** library.

This repository contains the **full-merged 16-bit weights**. No adapter loading is required.

## Training Objective

In SFT, an adapter is trained to improve **structured output accuracy** (JSON / YAML / XML / TOML / CSV).
Loss is applied only to the final assistant output, while intermediate reasoning (Chain-of-Thought) is masked.

After SFT, the adapter is merged to the base model. 
This model has been optimized using DPO to align its responses with preferred outputs, focusing on improving reasoning (Chain-of-Thought) and structured response quality based on the provided preference dataset.

## Training Configuration
- **Base model**: unsloth/Qwen3-4B-Instruct-2507
- **Method**: SFT(Supervised Fine-Tuning) + DPO (Direct Preference Optimization)

### SFT Training Configuration
- **Max sequence length**: 768
- **Epochs**: 2
- **Learning rate**: 5e-06
- **LoRA Config**: QLoRA(4-bit), r=128, alpha=128

### DPO Training Configuration
- **Epochs**: 5
- **Learning rate**: 1e-06
- **Beta**: 0.2
- **Max sequence length**: 768
- **LoRA Config**: r=64, alpha=64 (merged into base)
- **Early Drop**: threshold=1.2

## Usage
Since this is a merged model, you can use it directly with `transformers`.

```python
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_id = "your_id/your-repo-name"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.float16,
    device_map="auto"
)

# Test inference
prompt = "Your question here"
inputs = tokenizer.apply_chat_template([{"role": "user", "content": prompt}], tokenize=True, add_generation_prompt=True, return_tensors="pt").to("cuda")
outputs = model.generate(**inputs, max_new_tokens=512)
print(tokenizer.decode(outputs[0]))

```

## Sources & License (IMPORTANT)

* **Training Data**: [u-10bei/dpo-dataset-qwen-cot]
* **License**: MIT License. (As per dataset terms).
* **Compliance**: Users must follow the original base model's license terms.