Search papers, labs, and topics across Lattice.
School of Computer Science and Technology, Tongji University
3
0
6
Task-adaptive control of multimodal focus can boost evidence-grounded reasoning performance by over 6 points in critical scenarios.
LoRA's catastrophic forgetting problem can be solved by explicitly decoupling its factor updates and pushing new feature representations away from old ones, leading to state-of-the-art continual learning performance.
Generalizing to unseen compositions? This plug-and-play method leverages structure in the embedding space to adapt prompts, significantly boosting open-vocabulary zero-shot learning.
