Ashwanise609/gemma-3-1b-it-Censored

Ashwanise609/gemma-3-1b-it-Censored: A Decensored Gemma 3 Model

This model is a 1 billion parameter instruction-tuned variant of Google's Gemma 3 family, specifically a decensored version of unsloth/gemma-3-1b-it. It was created using Heretic v1.3.0 with specific 'abliteration parameters' to modify its refusal behavior. The model maintains a substantial 32K token context window.

Key Differentiators

• Decensored: Modified to reduce content refusals compared to its base model, with a refusal rate of 92/100 (compared to 91/100 for the original).
• Reproducible: The modification process is documented and reproducible, ensuring transparency in its 'decensoring'.
• Gemma 3 Architecture: Benefits from the multimodal capabilities of the Gemma 3 family, supporting text and image input to generate text output.

Intended Use Cases

This model is designed for applications where a less restrictive content generation policy is desired, while still leveraging the capabilities of the Gemma 3 architecture. It is suitable for:

• Content creation and communication, including text generation and conversational AI.
• Research and education, particularly for experimenting with VLM and NLP techniques.
• Applications requiring multimodal input (text and images) with text output.

Limitations

As a decensored model, users should be aware of potential ethical considerations and risks associated with reduced content safety filters. It shares general limitations of LLMs, including potential for factual inaccuracies, biases from training data, and challenges with nuanced language.