Model Overview

This model, qwen3-4b-dpo-qwen-cot-_2-3_05_DPO, is a 4 billion parameter language model derived from the Qwen/Qwen3-4B-Instruct-2507 base model. It has been specifically fine-tuned using Direct Preference Optimization (DPO) via the Unsloth library to align its responses with preferred outputs.

Key Capabilities

• Enhanced Reasoning: Optimized to improve Chain-of-Thought (CoT) reasoning, leading to more logical and structured responses.
• Preference Alignment: Benefits from DPO training, which aligns the model's output with human preferences, improving overall response quality.
• Structured Output: Designed to produce structured and coherent answers, making it suitable for tasks requiring organized information.
• Full-Merged Weights: The repository contains full-merged 16-bit weights, eliminating the need for adapter loading during deployment.

Training Details

The model underwent a single epoch of DPO training with a learning rate of 1e-07 and a beta value of 0.1. It utilized a maximum sequence length of 1024 and LoRA configuration (r=8, alpha=16). The training data was sourced from u-10bei/dpo-dataset-qwen-cot.

Good For

• Applications requiring improved reasoning and logical flow in generated text.
• Tasks where structured and preference-aligned responses are critical.
• Developers looking for a 4B parameter model with enhanced CoT capabilities.