8 papers
Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation
Zunhai Su, Hengyuan Zhang, Wei Wu +24
As the foundational architecture of modern machine learning, Transformers have driven remarkable progress across diverse AI domains. Despite their transformative impact, a persiste…
From Blind Spots to Gains: Diagnostic-Driven Iterative Training for Large Multimodal Models
Hongrui Jia, Chaoya Jiang, Yongrui Heng +2
As Large Multimodal Models (LMMs) scale up and reinforcement learning (RL) methods mature, LMMs have made notable progress in complex reasoning and decision making. Yet training st…
Not All Attention is Needed: Parameter and Computation Efficient Transfer Learning for Multi-modal Large Language Models
Qiong Wu, Weihao Ye, Yiyi Zhou +2
In this paper, we propose a novel parameter and computation efficient tuning method for Multi-modal Large Language Models (MLLMs), termed Efficient Attention Skipping (EAS). Concre…
XStreamVGGT: Extremely Memory-Efficient Streaming Vision Geometry Grounded Transformer with KV Cache Compression
Zunhai Su, Weihao Ye, Hansen Feng +5
Learning-based 3D visual geometry models have significantly advanced with the advent of large-scale transformers. Among these, StreamVGGT leverages frame-wise causal attention to d…
CCF: A Context Compression Framework for Efficient Long-Sequence Language Modeling
Wenhao Li, Bangcheng Sun, Weihao Ye +4
Scaling language models to longer contexts is essential for capturing rich dependencies across extended discourse. However, naïve context extension imposes significant computation…
XStreamVGGT: Extremely Memory-Efficient Streaming Vision Geometry Grounded Transformer with KV Cache Compression
Zunhai Su, Weihao Ye, Hansen Feng +5
Learning-based 3D visual geometry models have benefited substantially from large-scale transformers. Among these, StreamVGGT leverages frame-wise causal attention for strong stream…