Okaydvns/KillorKiss
Okaydvns/KillorKiss is an uncensored 27 billion parameter Qwen3.6-based causal language model, developed by huihui-ai. This model has undergone 'abliteration' to significantly reduce safety filtering and refusal behaviors, making it suitable for research into model safety and content generation without typical guardrails. It is primarily intended for experimental use cases where unfiltered outputs are desired, rather than general production applications.
Loading preview...
Okaydvns/KillorKiss: An Abliterated Qwen3.6-27B Model
Okaydvns/KillorKiss is a 27 billion parameter language model derived from Qwen/Qwen3.6-27B, developed by huihui-ai. Its key differentiator is the application of an "abliteration" process, which significantly reduces the model's inherent safety filtering and refusal mechanisms. This makes it an uncensored version, capable of generating content that standard models might refuse.
Key Characteristics
- Uncensored Output: Safety filtering has been substantially reduced, allowing for a broader range of generated content, including potentially sensitive or controversial topics.
- Experimental Focus: Designed as a proof-of-concept for exploring refusal removal techniques, rather than for general-purpose, safety-critical applications.
- Qwen3.6 Base: Built upon the Qwen3.6 architecture, inheriting its foundational language understanding and generation capabilities.
Usage Considerations
This model is explicitly not suitable for all audiences or production environments due to its reduced safety features. Users must exercise extreme caution and are solely responsible for the content generated. It is recommended for:
- Research and Development: Investigating model safety, refusal mechanisms, and the impact of content filtering.
- Controlled Environments: Testing scenarios where unfiltered outputs are specifically required and can be rigorously monitored.
Users are advised to review all generated outputs carefully and ensure compliance with legal and ethical standards.