AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

CompCap: Improving Multimodal Large Language Models with Composite Captions

Xiaohui Chen, Satya Narayan Shukla, Mahmoud Azab, Aashu Singh, Qifan Wang, David Yang, ShengYun Peng, Hanchao Yu, Shen Yan, Xuewen Zhang, Baosheng He

ICCV 2025 · Vision-language and multimodal › Multimodal LLMs · approach: Unified any-modality language models · 9 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 CompCap-118K captions to the SFT dataxGen-MM-inst.-4B* · 11 MLLM benchmarksaverage score: 57.2 -> 58.9Improved
add CompCap-118K captions to the SFT dataLLaVA-NeXT-Vicuna-7B · 11 MLLM benchmarksaverage score: 62.5 -> 64.5Improved
add CompCap-118K captions to the SFT dataLLaVA-NeXT-Vicuna-13B · 11 MLLM benchmarksaverage score: 65.6 -> 68.5Improved
add collage captions (ablation)LLaVA-NeXT-Vicuna-13B · 11 MLLM benchmarksaverage score: 65.6 -> 66.4Improved
add code captions (cumulative) (ablation)LLaVA-NeXT-Vicuna-13B · 11 MLLM benchmarksaverage score: 66.4 -> 66.6Improved
add table captions (cumulative) (ablation)LLaVA-NeXT-Vicuna-13B · 11 MLLM benchmarksaverage score: 66.6 -> 67.0Improved
add diagram captions (cumulative) (ablation)LLaVA-NeXT-Vicuna-13B · 11 MLLM benchmarksaverage score: 67.0 -> 67.4Improved
add chart captions (cumulative) (ablation)LLaVA-NeXT-Vicuna-13B · 11 MLLM benchmarksaverage score: 67.4 -> 68.0Improved
add image-text captions (cumulative, full CompCap-118K) (ablation)LLaVA-NeXT-Vicuna-13B · 11 MLLM benchmarksaverage score: 68.0 -> 68.5Improved
add CompCap-118K captionsLLaVA-NeXT-Vicuna-13B · CI-dominated benchmarksaverage score: 61.3 -> 64.6Improved
add CompCap-118K captionsLLaVA-NeXT-Vicuna-13B · NI-dominated benchmarksaverage score: 70.9 -> 73.1Improved
replace ChartQA instruction data with captions (ratio k from 0 to 100 percent) (ablation)LLaVA-NeXT-Vicuna-13B · ChartQAaccuracyMixed

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