Yanbing Jiang

Xiaowei Chi, Yatian Wang, Aosong Cheng et al.

Massive multi-modality datasets play a significant role in facilitating the success of large video-language models. However, current video-language datasets primarily provide text descriptions for visual frames, considering audio to be weakly related information. They usually overlook exploring the potential of inherent audio-visual correlation, leading to monotonous annotation within each modality instead of comprehensive and precise descriptions. Such ignorance results in the difficulty of multiple cross-modality studies. To fulfill this gap, we present MMTrail, a large-scale multi-modality video-language dataset incorporating more than 20M trailer clips with visual captions, and 2M high-quality clips with multimodal captions. Trailers preview full-length video works and integrate context, visual frames, and background music. In particular, the trailer has two main advantages: (1) the topics are diverse, and the content characters are of various types, e.g., film, news, and gaming. (2) the corresponding background music is custom-designed, making it more coherent with the visual context. Upon these insights, we propose a systemic captioning framework, achieving various modality annotations with more than 27.1k hours of trailer videos. Here, to ensure the caption retains music perspective while preserving the authority of visual context, we leverage the advanced LLM to merge all annotations adaptively. In this fashion, our MMtrail dataset potentially paves the path for fine-grained large multimodal-language model training. In experiments, we provide evaluation metrics and benchmark results on our dataset, demonstrating the high quality of our annotation and its effectiveness for model training.

Ruijie Miao, Zhiming Wang, Wang Li et al.

Key-value (KV) caching is widely used to accelerate transformer inference, but its memory cost grows linearly with input length, limiting long-context deployment. Existing token eviction methods reduce memory by discarding less important tokens, which can be viewed as a coarse form of dimensionality reduction that assigns each token either zero or full dimension. We propose MixedDimKV, a mixed-dimension KV cache compression method that allocates dimensions to tokens at a more granular level, and MixedDimKV-H, which further integrates head-level importance information. Experiments on long-context benchmarks show that MixedDimKV outperforms prior KV cache compression methods that do not rely on head-level importance profiling. When equipped with the same head-level importance information, MixedDimKV-H consistently outperforms HeadKV. Notably, our approach achieves comparable performance to full attention on LongBench with only 6.25% of the KV cache. Furthermore, in the Needle-in-a-Haystack test, our solution maintains 100% accuracy at a 50K context length while using as little as 0.26% of the cache.

Fan Zhang, Naye Ji, Fuxing Gao et al.

Speech-driven gesture generation is an emerging domain within virtual human creation, where current methods predominantly utilize Transformer-based architectures that necessitate extensive memory and are characterized by slow inference speeds. In response to these limitations, we propose \textit{DiM-Gestures}, a novel end-to-end generative model crafted to create highly personalized 3D full-body gestures solely from raw speech audio, employing Mamba-based architectures. This model integrates a Mamba-based fuzzy feature extractor with a non-autoregressive Adaptive Layer Normalization (AdaLN) Mamba-2 diffusion architecture. The extractor, leveraging a Mamba framework and a WavLM pre-trained model, autonomously derives implicit, continuous fuzzy features, which are then unified into a singular latent feature. This feature is processed by the AdaLN Mamba-2, which implements a uniform conditional mechanism across all tokens to robustly model the interplay between the fuzzy features and the resultant gesture sequence. This innovative approach guarantees high fidelity in gesture-speech synchronization while maintaining the naturalness of the gestures. Employing a diffusion model for training and inference, our framework has undergone extensive subjective and objective evaluations on the ZEGGS and BEAT datasets. These assessments substantiate our model's enhanced performance relative to contemporary state-of-the-art methods, demonstrating competitive outcomes with the DiTs architecture (Persona-Gestors) while optimizing memory usage and accelerating inference speed.

Zihang Li, Wenjun Liu, Yikun Zong et al.

As an important paradigm for enhancing the generation quality of Large Language Models (LLMs), retrieval-augmented generation (RAG) faces the two challenges regarding retrieval accuracy and computational efficiency. This paper presents a novel RAG framework called Bridge-RAG. To overcome the accuracy challenge, we introduce the concept of abstract to bridge query entities and document chunks, providing robust semantic understanding. We organize the abstracts into a tree structure and design a multi-level retrieval strategy to ensure the inclusion of sufficient contextual information. To overcome the efficiency challenge, we introduce the improved Cuckoo Filter, an efficient data structure supporting rapid membership queries and updates, to accelerate entity location during the retrieval process. We design a block linked list structure and an entity temperature-based sorting mechanism to improve efficiency from the aspects of spatial and temporal locality. Extensive experiments show that Bridge-RAG achieves around 15.65% accuracy improvement and reduces 10x to 500x retrieval time compared to other RAG frameworks.

Chongjian Yue, Xinrun Xu, Xiaojun Ma et al.

Large Language Models (LLMs) demonstrate exceptional performance in textual understanding and tabular reasoning tasks. However, their ability to comprehend and analyze hybrid text, containing textual and tabular data, remains underexplored. In this research, we specialize in harnessing the potential of LLMs to comprehend critical information from financial reports, which are hybrid long-documents. We propose an Automated Financial Information Extraction (AFIE) framework that enhances LLMs' ability to comprehend and extract information from financial reports. To evaluate AFIE, we develop a Financial Reports Numerical Extraction (FINE) dataset and conduct an extensive experimental analysis. Our framework is effectively validated on GPT-3.5 and GPT-4, yielding average accuracy increases of 53.94% and 33.77%, respectively, compared to a naive method. These results suggest that the AFIE framework offers accuracy for automated numerical extraction from complex, hybrid documents.