AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

FlexAttention for Efficient High-Resolution Vision-Language Models

Junyan Li, Delin Chen, Tianle Cai, Peihao Chen, Yining Hong, Zhenfang Chen, Yikang Shen, Chuang Gan

ECCV 2024 · Vision-language and multimodal › Multimodal LLMs · approach: Attention-based visual token pruning · 15 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add FlexAttention high-resolution input (1008x1008) to LLaVA-1.5-7bLLaVA-1.5-7b · V* Benchoverall accuracy: 47.6 -> 54.5Improved
add FlexAttention to LLaVA-1.5-7bLLaVA-1.5-7b · MagnifierBenchaccuracy: 26.8 -> 35.0Improved
FlexAttention instead of LLaVA-HD or LLaVA-XAttn high-resolution handlingLLaVA-1.5-7b · MagnifierBenchTFLOPs: 24.9 -> 17.1 (vs LLaVA-HD); 27.1 -> 17.1 (vs LLaVA-XAttn)Improved
select high-resolution tokens by attention map instead of random or center selection (ablation)FlexAttention · MagnifierBench and TextVQAaccuracyImproved
increase high-resolution input from 672 to 1008 (ablation)FlexAttention · TextVQAaccuracyImproved
increase high-resolution input from 1008 to 1344 (ablation)FlexAttention · TextVQAaccuracyNo clear effect
increase high-resolution input resolution (ablation)FlexAttention · MagnifierBenchaccuracyImproved
add FlexAttention to LLaVA-1.5-7bLLaVA-1.5-7b · RefCOCO val, small objectsaccuracy: 41.3 -> 51.3Improved
add FlexAttention to LLaVA-1.5-7bLLaVA-1.5-7b · MME and MM-Vetscore: MME 1511 -> 1479; MM-Vet 31.1 -> 29.4Hurt
add FlexAttention to LLaVA-1.5-7bLLaVA-1.5-7b · POPE, GQA, VQAv2accuracy: POPE 85.9 -> 85.9No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy