musiccat0405/gemma-3-1b-it-heretic-extreme-uncensored-abliterated
The musiccat0405/gemma-3-1b-it-heretic-extreme-uncensored-abliterated model is a 1 billion parameter instruction-tuned Gemma variant, fine-tuned by musiccat0405 using the Heretic v1.0.1 method to significantly reduce refusal rates while maintaining model integrity. With a 32k token context length, this model achieves a refusal rate of 3/100 and a KL divergence of 0.33, making it highly uncensored for various content generation tasks. It is primarily designed for applications requiring unrestricted text generation, including potentially sensitive or explicit content, without the typical refusals found in base models.
Loading preview...
Overview
This model, gemma-3-1b-it-heretic-extreme-uncensored-abliterated, is a 1 billion parameter instruction-tuned Gemma variant, specifically modified by musiccat0405 using the Heretic v1.0.1 method. The primary goal of this modification is to drastically reduce content refusal rates, making the model highly "abliterated" or uncensored. It boasts a 32k token context length, allowing for extensive conversational and generative capabilities.
Key Capabilities
- Significantly Reduced Refusal Rate: Achieves a refusal rate of 3/100, a substantial improvement over the original Gemma 1B-IT's 99/100, enabling generation of content typically censored by other models.
- Preserved Model Integrity: Maintains a low KL divergence of 0.33, indicating that the uncensoring process has not significantly "damaged" the model's core functionality or default state.
- Unrestricted Content Generation: Designed to generate a wide range of content, including potentially sensitive, explicit, or "x-rated" material, without typical refusals.
- Optimized for Directives: While uncensored, the model may require explicit directives (e.g., including specific slang or terms) to generate content at expected graphic or explicit levels, especially compared to models trained on uncensored datasets.
Good For
- Creative Writing & Roleplay: Ideal for scenarios requiring unrestricted narrative freedom, including horror, explicit, or otherwise sensitive themes.
- Applications Requiring Uncensored Output: Suitable for use cases where content filtering or refusals are undesirable, such as certain artistic expressions or research into model behavior.
- Experimentation with Model De-censoring: Provides a practical example of the Heretic method's effectiveness in reducing model censorship while preserving quality.