sibutosi2010/dpo-qwen-cot-merged

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:4BQuant:BF16Ctx Length:32kPublished:Feb 25, 2026License:apache-2.0Architecture:Transformer Open Weights Warm

The sibutosi2010/dpo-qwen-cot-merged is a 4 billion parameter Qwen3 instruction-tuned causal language model developed by sibutosi2010. This model was fine-tuned using Unsloth and Huggingface's TRL library, enabling 2x faster training. It is designed for general instruction-following tasks, leveraging its Qwen3 architecture and efficient training methodology.

Loading preview...

Overview

The sibutosi2010/dpo-qwen-cot-merged is a 4 billion parameter Qwen3-based instruction-tuned language model. Developed by sibutosi2010, this model was fine-tuned from unsloth/qwen3-4b-instruct-2507-unsloth-bnb-4bit.

Key Characteristics

  • Base Model: Qwen3 architecture.
  • Parameter Count: 4 billion parameters.
  • Training Efficiency: Fine-tuned using Unsloth and Huggingface's TRL library, which facilitated a 2x faster training process.
  • License: Distributed under the Apache-2.0 license.

Use Cases

This model is suitable for various instruction-following applications, benefiting from its efficient training and Qwen3 foundation. Its optimized training process suggests potential for rapid iteration and deployment in scenarios requiring a capable 4B parameter model.