ICML

KORE: Enhancing Knowledge Injection for Large Multimodal Models via Knowledge-Oriented Augmentations and Constraints

LIFT: A Novel Framework for Enhancing Long-Context Understanding of LLMs via Long Input Fine-Tuning

SPHERE: Mitigating the Loss of Spectral Plasticity in Mixture-of-Experts for Deep Reinforcement Learning

UniCode: Augmenting Evaluation for Code Reasoning

CLOVER: Cross-Layer Orthogonal Vectors Pruning and Fine-Tuning

Falcon: Fast visuomotor policy via partial denoising