blackbook-lm/Qwen3-0.6B-heretic
The blackbook-lm/Qwen3-0.6B-heretic is a 0.8 billion parameter causal language model, derived from Qwen/Qwen3-0.6B and decensored using Heretic v1.2.0. It retains the original Qwen3 architecture, featuring a 32,768-token context length and unique support for seamless switching between 'thinking' and 'non-thinking' modes for varied tasks. This model is specifically modified to reduce refusals, making it suitable for use cases requiring less restrictive content generation.
Loading preview...
Overview
This model, blackbook-lm/Qwen3-0.6B-heretic, is a 0.8 billion parameter causal language model based on the Qwen3-0.6B architecture, with a 32,768-token context length. It has been specifically modified using Heretic v1.2.0 to be a decensored version of the original, significantly reducing content refusals. While the original Qwen3-0.6B model had 53 refusals out of 100, this 'heretic' variant shows only 8 refusals out of 100, indicating a much less restrictive output.
Key Capabilities
- Decensored Output: Significantly reduced refusal rate compared to the base model, allowing for broader content generation.
- Thinking/Non-Thinking Modes: Inherits Qwen3's unique ability to switch between a 'thinking mode' for complex logical reasoning, math, and coding, and a 'non-thinking mode' for efficient general-purpose dialogue.
- Enhanced Reasoning: The base Qwen3 model demonstrates strong capabilities in mathematics, code generation, and commonsense logical reasoning.
- Multilingual Support: Supports over 100 languages and dialects for instruction following and translation.
- Agentic Use: Excels in tool-calling capabilities, compatible with frameworks like Qwen-Agent.
Good For
- Applications requiring less restrictive content generation or exploration of sensitive topics.
- Tasks benefiting from dynamic switching between detailed reasoning and direct responses.
- Multilingual applications and agent-based systems where the base Qwen3's capabilities are desired without content filtering.