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
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 changed | Where | Measured | Effect |
|---|---|---|---|
| replace MLP connector with LangBridge | LLaVA-style LVLM (LLaMA3-8B) · GQA | accuracy: 63.24 -> 64.00 | Improved |
| replace MLP connector with LangBridge | LLaVA-style LVLM (LLaMA3-8B) · SciQA | accuracy: 78.61 -> 77.98 | Hurt |
| replace MLP connector with LangBridge | LLaVA-style LVLM (Qwen2.5-7B) · MMVeT | accuracy: 34.8 -> 36.7 | Improved |
| replace MLP connector with LangBridge | LLaVA-style LVLM (Qwen2.5-7B) · MME | score: 1939 -> 1874 | Hurt |
| reuse Qwen2-0.5B-SFT LangBridge on Qwen2.5-14B without pretraining | LangBridge · TextVQA and MMVeT | accuracy: TextVQA +1.14%, MMVeT +1.86% | Improved |
| reuse LLaMA3-8B-SFT LangBridge on Qwen2.5-7B (cross-architecture) | LangBridge · MMVeT and POPE | accuracy: MMVeT +8.33%, POPE +1.92% | Improved |
| reuse LLaMA3-8B-SFT LangBridge on Qwen2.5-14B (cross-architecture) | LangBridge · GQA and MMVeT | accuracy: GQA +1.19%, MMVeT +1.59% | Improved |
| reuse Qwen2-0.5B-Pretrain LangBridge on Qwen2-7B | LangBridge · MME | score: 1938 -> 1886 | Hurt |
| reduce LangBridge hidden size from 5120 to 1024 (ablation) | LangBridge on LLaMA3-8B · MMVeT | accuracy: 31.1 -> 33.8 | Mixed |
| increase vocabulary from 19200 to 25600 (ablation) | LangBridge on Qwen2-7B · MME | score: 1904 -> 1842 | Hurt |
| increase vocabulary from 19200 to 32000 (ablation) | LangBridge on Qwen2-7B · MME | score: 1904 -> 1832 | Hurt |
| integrate LangBridge into LLaVA-Next with Qwen2-7B | LLaVA-Next · RefCOCO | accuracy: 0.387 -> 0.453 | Improved |
Closest papers by meaning
- Analyzing Fine-Grained Alignment and Enhancing Vision Understanding in Multimodal Language Models (NeurIPS 2025 · similarity 0.90)
- Deep Pre-Alignment for VLMs (ICML 2026 · similarity 0.89)
- Bridging Vision and Language Spaces with Assignment Prediction (ICLR 2024 · similarity 0.89)
- AlignVLM: Bridging Vision and Language Latent Spaces for Multimodal Document Understanding (NeurIPS 2025 · similarity 0.89)
- DH-Set: Improving Vision-Language Alignment with Diverse and Hybrid Set-Embeddings Learning (CVPR 2025 · similarity 0.88)
- Visual Perception by Large Language Model’s Weights (NeurIPS 2024 · similarity 0.88)
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.
