Gemma 4 Uncensored Collection Abliterated Gemma 4 models with refusal behavior removed. Biprojection + EGA for MoE. Cross-validated against 686 prompts from 4 datasets. • 8 items • Updated 17 days ago • 50
When Good Sounds Go Adversarial: Jailbreaking Audio-Language Models with Benign Inputs Paper • 2508.03365 • Published Aug 5, 2025 • 5
Eliciting and Analyzing Emergent Misalignment in State-of-the-Art Large Language Models Paper • 2508.04196 • Published Aug 6, 2025 • 1
When Good Sounds Go Adversarial: Jailbreaking Audio-Language Models with Benign Inputs Paper • 2508.03365 • Published Aug 5, 2025 • 5
Inference-Time Scaling for Diffusion Models beyond Scaling Denoising Steps Paper • 2501.09732 • Published Jan 16, 2025 • 72