vlx1/Qwen2.5-0.5B-Instruct-heretic

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:0.5BQuant:BF16Ctx Length:32kPublished:May 19, 2026License:apache-2.0Architecture:Transformer Open Weights Warm

The vlx1/Qwen2.5-0.5B-Instruct-heretic is a 0.49 billion parameter instruction-tuned causal language model, derived from Qwen's Qwen2.5 series. This version is specifically decensored using Heretic v1.2.0, significantly reducing refusals compared to the original Qwen2.5-0.5B-Instruct model. It retains the base model's enhanced capabilities in coding, mathematics, long-text generation, and multilingual support across 29 languages, with a 32,768 token context length.

Loading preview...

vlx1/Qwen2.5-0.5B-Instruct-heretic: Decensored Qwen2.5 Model

This model is a decensored variant of the Qwen/Qwen2.5-0.5B-Instruct, created using the Heretic v1.2.0 tool. It significantly reduces model refusals, demonstrating 4 refusals out of 100 compared to 92/100 for the original model, while maintaining a low KL divergence of 0.0401.

Key Capabilities Inherited from Qwen2.5-0.5B-Instruct

  • Enhanced Knowledge & Reasoning: Improved performance in coding and mathematics due to specialized expert models.
  • Instruction Following: Stronger instruction adherence and resilience to diverse system prompts, beneficial for role-play and chatbot implementations.
  • Long Context & Generation: Supports a full context length of 32,768 tokens and can generate up to 8,192 tokens.
  • Structured Data & Output: Better at understanding structured data like tables and generating structured outputs, particularly JSON.
  • Multilingual Support: Comprehensive support for over 29 languages, including major global languages.

Good For

  • Applications requiring a less restrictive or decensored language model for instruction-following tasks.
  • Use cases benefiting from strong coding and mathematical abilities in a compact 0.5 billion parameter model.
  • Scenarios demanding long-context understanding and generation.
  • Multilingual applications needing support for a wide array of languages.