Hugging Face Overview
Model Overview
This model, heretic-org/XortronCriminalComputingConfig-heretic, is a 24 billion parameter uncensored variant of the darkc0de/XortronCriminalComputingConfig model. It was created using Heretic v1.1.0 to modify its refusal behavior.
Key Capabilities & Performance
- Decensored Output: The primary differentiator is its significantly reduced refusal rate, achieving 6 refusals out of 100 prompts compared to 20/100 for the original model. This makes it suitable for generating content that other models might censor or decline to produce.
- UGI Leaderboard Topper: As of July 2025, this model reportedly leads the UGI Leaderboard for models under 70 billion parameters in both the UGI and W10 categories, indicating strong performance in its specialized domain.
- Abliteration Parameters: Specific internal parameters were adjusted during its creation, including
direction_index(21.70),attn.o_proj.max_weight(1.49), andmlp.down_proj.max_weight(1.36), which contribute to its uncensored nature.
Use Cases
This model is designed for applications requiring unrestricted content generation, particularly in areas where typical LLMs impose content filters or ethical guidelines. Users are advised to use it responsibly and discreetly due to its uncensored nature.