Search papers, labs, and topics across Lattice.
Affiliation:
2
0
5
0
This work proposes ARIA (autoencoder-gated inference-time unlearning), a test-time unlearning method that leaves model weights intact and gates access to unwanted knowledge only when generation enters a forget-related state and introduces three post-unlearning adversarial attacks targeting weight-space and decoding-space recovery.
SkillComposer achieves a remarkable +23.1% increase in task success rates for LLM agents by rethinking how skills are composed and executed together.
