AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

LangBridge: Interpreting Image as a Combination of Language Embeddings

Jiaqi Liao, Yuwei Niu, Fanqing Meng, Hao Li, Changyao Tian, Yinuo Du, Yuwen Xiong, Dianqi Li, Xizhou Zhu, Li Yuan, Jifeng Dai, Yu Cheng

ICCV 2025 · Vision-language and multimodal › Multimodal LLMs · approach: VLM capability analysis · 10 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
replace MLP connector with LangBridgeLLaVA-style LVLM (LLaMA3-8B) · GQAaccuracy: 63.24 -> 64.00Improved
replace MLP connector with LangBridgeLLaVA-style LVLM (LLaMA3-8B) · SciQAaccuracy: 78.61 -> 77.98Hurt
replace MLP connector with LangBridgeLLaVA-style LVLM (Qwen2.5-7B) · MMVeTaccuracy: 34.8 -> 36.7Improved
replace MLP connector with LangBridgeLLaVA-style LVLM (Qwen2.5-7B) · MMEscore: 1939 -> 1874Hurt
reuse Qwen2-0.5B-SFT LangBridge on Qwen2.5-14B without pretrainingLangBridge · TextVQA and MMVeTaccuracy: TextVQA +1.14%, MMVeT +1.86%Improved
reuse LLaMA3-8B-SFT LangBridge on Qwen2.5-7B (cross-architecture)LangBridge · MMVeT and POPEaccuracy: MMVeT +8.33%, POPE +1.92%Improved
reuse LLaMA3-8B-SFT LangBridge on Qwen2.5-14B (cross-architecture)LangBridge · GQA and MMVeTaccuracy: GQA +1.19%, MMVeT +1.59%Improved
reuse Qwen2-0.5B-Pretrain LangBridge on Qwen2-7BLangBridge · MMEscore: 1938 -> 1886Hurt
reduce LangBridge hidden size from 5120 to 1024 (ablation)LangBridge on LLaMA3-8B · MMVeTaccuracy: 31.1 -> 33.8Mixed
increase vocabulary from 19200 to 25600 (ablation)LangBridge on Qwen2-7B · MMEscore: 1904 -> 1842Hurt
increase vocabulary from 19200 to 32000 (ablation)LangBridge on Qwen2-7B · MMEscore: 1904 -> 1832Hurt
integrate LangBridge into LLaVA-Next with Qwen2-7BLLaVA-Next · RefCOCOaccuracy: 0.387 -> 0.453Improved

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