Search papers, labs, and topics across Lattice.
2
0
4
Correcting surrogate-specific bias in adversarial optimization can dramatically enhance the transferability of attacks across Vision-Language models.
Forget static rubrics and expensive external models: EvoRubric co-evolves a single policy to generate both responses and the rubrics to evaluate them, outperforming traditional RLHF methods in open-ended generation tasks.
