Research map / Vision-language and multimodal / Cross-modal learning
Probing Cross-modal Information Hubs in Audio-Visual LLMs
Jihoo Jung, Chaeyoung Jung, Ji-Hoon Kim, Joon Son Chung
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 |
|---|---|---|---|
| patch sink tokens (N=2) instead of random tokens (ablation) | Qwen2.5-Omni (7B) · Audio-dominant causal tracing | IE_clean: 4.24 -> 6.24 | Improved |
| patch sink tokens (N=2) instead of random tokens (ablation) | Qwen2.5-Omni (7B) · Video-dominant causal tracing | IE_clean: 4.56 -> 5.47 | Improved |
| patch sink tokens (N=2) instead of object tokens (ablation) | Qwen2.5-Omni (7B) · Audio-dominant causal tracing | IE_clean: 5.04 -> 6.24 | Improved |
| patch cross-modal sink tokens instead of unimodal sink tokens (ablation) | Qwen2.5-Omni (7B) · Audio-dominant causal tracing | IE_clean: 0.65 -> 5.58 | Improved |
| patch cross-modal sink tokens instead of unimodal sink tokens (ablation) | video-SALMONN-o1 (7B) · Audio-dominant causal tracing | IE_clean: 7.25 -> 21.30 | Improved |
| patch cross-modal sink tokens instead of unimodal sink tokens (ablation) | Qwen2.5-Omni (7B) · Video-dominant causal tracing | IE_clean: 1.93 -> 3.03 | Improved |
| patch cross-modal sink tokens instead of all sink tokens (ablation) | Qwen2.5-Omni (7B) · Audio-dominant causal tracing | IE_clean: 6.24 -> 5.58 | No clear effect |
| apply Adaptive Sink-Guided Decoding | Qwen2.5-Omni (7B) · VGGSound-Animal | CHAIR Cs: 48.21 -> 36.91 | Improved |
| apply Adaptive Sink-Guided Decoding | video-SALMONN-o1 (7B) · VGGSound-Animal | CHAIR Cs: 37.74 -> 25.07 | Improved |
| apply Adaptive Sink-Guided Decoding | Qwen2.5-Omni (7B) · AudioSet | CHAIR Cs: 8.92 -> 8.54 | Improved |
| apply Adaptive Sink-Guided Decoding | Qwen2.5-Omni (7B) · VGGSound-Animal | caption F1: 55.24 -> 52.44 | Hurt |
| apply PAI | Qwen2.5-Omni (7B) · VGGSound-Animal | CHAIR Cs: 48.21 -> 51.24 | Hurt |
Closest papers by meaning
- From Senses to Decisions: The Information Flow of Auditory and Visual Perception in Multimodal LLMs (NeurIPS 2026 · similarity 0.93)
- AVHBench: A Cross-Modal Hallucination Benchmark for Audio-Visual Large Language Models (ICLR 2025 · similarity 0.91)
- XModBench: Benchmarking Cross-Modal Capabilities and Consistency in Omni-Language Models (ICLR 2026 · similarity 0.90)
- Aligned Better, Listen Better for Audio-Visual Large Language Models (ICLR 2025 · similarity 0.90)
- LatentOmni: Rethinking Omni-Modal Understanding via Unified Audio-Visual Latent Reasoning (NeurIPS 2026 · similarity 0.90)
- Measuring Cross-Modal Synergy: A Benchmark for VLM Explainability (NeurIPS 2026 · similarity 0.89)
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.
