huihui-ai/Huihui-Qwen3-1.7B-abliterated-v2

Overview

Huihui-ai/Huihui-Qwen3-1.7B-abliterated-v2 is a 1.7 billion parameter language model based on the Qwen3 architecture. Developed by huihui-ai, this model is specifically designed as an uncensored version of the original Qwen/Qwen3-1.7B. It utilizes an advanced abliteration method, detailed in the remove-refusals-with-transformers project, to significantly reduce safety filtering and refusals, offering a more direct and unfiltered output compared to its base model. This version represents an improvement over the previous huihui-ai/Qwen3-1.7B-abliterated model, featuring a faster abliteration method that yields better results and addresses issues like garbled codes by modifying layer 0.

Key Capabilities

• Uncensored Content Generation: Designed to produce outputs with significantly reduced safety filtering, allowing for a broader range of responses.
• Improved Abliteration: Incorporates a new, faster abliteration technique for more effective refusal removal.
• Qwen3 Architecture: Leverages the foundational capabilities of the Qwen3 model family.
• Ollama Integration: Directly available for use via Ollama, simplifying deployment.

Usage Warnings & Considerations

This model is intended primarily for research and experimental use due to its reduced safety mechanisms. Users should be aware of the following:

• Risk of Sensitive Outputs: May generate sensitive, controversial, or inappropriate content.
• Not for All Audiences: Outputs may be unsuitable for public or underage users.
• Legal & Ethical Responsibility: Users are solely responsible for ensuring compliance with laws and ethical standards.
• No Default Safety Guarantees: Unlike standard models, it has not undergone rigorous safety optimization, and huihui.ai disclaims responsibility for consequences arising from its use.