Search papers, labs, and topics across Lattice.
Affiliation:
4
0
3
5
Protein fitness landscapes are surprisingly linear in a low-dimensional subspace of PLM representations, enabling dramatically more efficient directed evolution.
LaP-Forensics reveals that leveraging reconstruction-based evidence can significantly enhance deepfake detection accuracy against state-of-the-art generative models.
Hidden evidence-use forgetting can lead to accurate answers that lack grounding, but a new framework shows how to preserve the evidence path behind those answers.
InduceKV achieves superior performance in continual adaptation of multimodal LLMs while strictly adhering to a fixed memory budget, challenging the notion that larger memory footprints are necessary for effective learning.
