Model Overview

eridon-pro/dpo-qwen-cot-merged-from-sft-adapter-38-1 is a 4 billion parameter language model based on the Qwen3-4B-Instruct-2507 architecture. It has been further fine-tuned by eridon-pro using Direct Preference Optimization (DPO) via the Unsloth library. This model incorporates full-merged 16-bit weights, eliminating the need for adapter loading.

Key Capabilities

• Enhanced Reasoning: Optimized through DPO to improve Chain-of-Thought (CoT) reasoning, making it suitable for tasks requiring logical steps and deductions.
• Structured Response Quality: Specifically aligned to produce higher quality, more structured outputs based on preference datasets.
• Direct Use: As a fully merged model, it can be directly integrated and used with the transformers library without additional configuration for LoRA adapters.

Training Details

The model was trained for 2 epochs with a learning rate of 1e-06 and a beta value of 0.1. It utilized a maximum sequence length of 1024 during training. The DPO process leveraged the u-10bei/dpo-dataset-qwen-cot dataset for preference alignment. The base model was an SFT merged version from eridon-pro/lora_structeval_t_qwen3_4b-38.

Licensing

This model operates under the MIT License, consistent with its training data. Users must also adhere to the license terms of the original base model, Qwen3-4B-Instruct-2507.