论文解读
联合分支不可全信：用冲突量与可干预性重配解码权重
针对全模态大模型中音频视觉联合预测不可靠融合的问题，论文提出免训练的冲突感知解码 CAD，用潜在冲突量与任务空间可干预性决定是否把联合分支权重转给单模态分支，在 Qwen2.5-Omni-7B 上把 CMM 整体准确率提到 85.0% 与 AVHBench 提到 84.1%，代价是每步要跑四个分支并依赖人工阈值。
音视频问答
论文解读
说出来就变了味：当语音不再是文本的透明通道
音视频问答 | 8.1/10
论文解读
Multi-Modal Generative Fuzzy System: Fuzzy Inference Guided Large Model Interactive Question Answering Framework
音视频问答 | 5.9/10
论文解读
EgoCITE: Context-Augmented Indexing and Time-Aware Retrieval for Long-Horizon Egocentric Memory
音视频问答 | 7.2/10
论文解读
Qwen-MusicAVQA-7B: A Multimodal Model for Music Audio-Visual QA
音视频问答 | 6.6/10
论文解读
Listen, See and Track: Spatio-Temporal Audio-Visual Sound Event Reasoning for Omni-Modal Language Models
音视频问答 | 6.7/10
论文解读
C^3PO: Evaluating Cross-Modal Composition and Counterfactual Performance in Omnimodal Models
音视频问答 | 5.5/10
论文解读
Evidence-Grounded Multimodal Knowledge Graph Construction for Multi-Lecture Educational Reasoning
音视频问答 | 4.7/10
论文解读
Geometric Cross-Modal Token Selection for Latency-Constrained Multimodal Token Communication
音视频问答 | 6.2/10
论文解读
OmniDelta: Skill-Driven Budget Allocation for Token Compression in OmniLLMs
音视频问答 | 7.0/10
论文解读
OmniScope: Modality-Decoupled Token Compression for Omnimodal Large Language Models
音视频问答 | 8.3/10
论文解读
Audio-Visual Flamingo: Open Audio-Visual Intelligence for Long and Complex Videos
音视频理解 | 6.0/10
论文解读
OmniFocus: Query-Guided Modality-Balanced Token Compression for Omni-Modal Large Language Models
音视频问答 | 5.9/10
论文解读
E-VAds: An E-commerce Short Videos Understanding Benchmark for MLLMs
音视频问答 | 6.9/10
论文解读
FutureOmni: Evaluating Future Forecasting from Omni-Modal Context for Multimodal LLMs
音视频问答 | 8/10
论文解读
OmniSIFT: Modality-Asymmetric Token Compression for Efficient Omni-modal Large Language Models
音视频问答 | 7.1/10
论文解读
OmniVideo-R1: Reinforcing Audio-visual Reasoning with Query Intention and Modality Attention
音视频问答 | 7.5/10
论文解读
PhoStream: Benchmarking Real-World Streaming for Omnimodal Assistants in Mobile Scenarios
音视频问答 | 7.3/10
论文解读
V-LynX: Token Interface Alignment for Video+X LLMs
音视频问答 | 7.8/10
论文解读
video-SALMONN S: Memory-Enhanced Streaming Audio-Visual LLM
音视频问答 | 7.3/10