作者位次 #2/5ICML 2026Poster· Deep Learning / Foundation Models
DIVA: Harnessing the Representation Divergence in Unified Multimodal Models for Mutual Reinforcement
renjie lu · Xulong Zhang · Xiaoyang Qu · Jianzong Wang · Shangfei Wang
首次获取该候选人的完整档案及 StartLux 评估共 20 积分；每位用户单独解锁，之后查看免费。主动刷新 20 积分。
实时查找电子邮箱，费用不超过20积分；未找到邮箱不扣费。
renjie lu · Xulong Zhang · Xiaoyang Qu · Jianzong Wang · Shangfei Wang
Jiaqi Shi · Xulong Zhang · Yuechan Li · Xiaoyang Qu · Jianzong Wang
High-resolution Multimodal Large Language Models (MLLMs) face prohibitive computational costs during inference due to the explosion of visual tokens. Existing acceleration strategies, such as token pruning or layer sparsity, suffer from severe “backbone dependency”, performing well on Vicuna or M...