cooperleong00/Qwen3-8B-Jailbroken
TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:32kPublished:Apr 30, 2025Architecture:Transformer0.0K Cold

The cooperleong00/Qwen3-8B-Jailbroken model is a Qwen3-8B variant developed by cooperleong00, specifically modified using weight orthogonalization techniques. This model is designed for academic research into AI safety and model alignment, focusing on understanding and manipulating refusal behaviors in large language models. Its primary differentiator is its 'jailbroken' nature, making it suitable for studying model vulnerabilities and ethical boundaries.

Loading preview...