AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

From One-to-One to Many-to-Many: Dynamic Cross-Layer Injection for Deep Vision-Language Fusion

Cheng Chen, Yuyu Guo, Pengpeng Zeng, Jingkuan Song, Peng Di, Hang Yu, Lianli Gao

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Attention-based visual token pruning · 2 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 AMP only (LoRA multi-projection) (ablation)LLaVA-OV-0.5B · 9-benchmark ablation sumsum score: 366.72 -> 367.89No clear effect
add AGF gating alone (ablation)LLaVA-OV-0.5B · 9-benchmark ablation sumsum score: 366.72 -> 370.60Improved
add AMP + AGF (CLI) (ablation)LLaVA-OV-0.5B · 9-benchmark ablation sumsum score: 366.72 -> 371.51Improved
fully fine-tune the projector in place of LoRA AMP, with AGF (ablation)LLaVA-OV-0.5B · 9-benchmark ablation sumsum score: 371.51 -> 407.01Improved
add CLI to the baseline projectorLLaVA-OV-7B · 9 doc/perception benchmarkssum score: 650.9 -> 660.6Improved
add CLI to the baseline projectorLLaVA-OV-0.5B · 9 doc/perception benchmarkssum score: 541.2 -> 542.9Improved
add CLI to the baseline projectorLLaVA-OV-7B · 9 reasoning benchmarkssum score: 556.2 -> 559.4Improved
add CLI to LLaVA-1.5-7BLLaVA-1.5-7B · 9 reasoning benchmarkspartial sum: 433.8 -> 442.4Improved
add DeepStack to the baselineLLaVA-OV-7B · 9 doc/perception benchmarkssum score: 650.9 -> 611.8Hurt
add Shallow-Layer Injection (SLI)LLaVA-OV-7B · 9 reasoning benchmarkssum score: 556.2 -> 551.1Hurt
add CLILLaVA-OV-7B · LLaVA-in-the-Wildscore: 68.0 -> 74.5Improved
replace CLI by equal-parameter static one-to-one Parallel injection (ablation)LLaVA-OV-7B · LLaVA-in-the-Wildscore: dropping 6.5 pointsHurt

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