K. L. Han

Xiaohu Huang, Hao Zhou, Kai Han

In this paper, we introduce PruneVid, a visual token pruning method designed to enhance the efficiency of multi-modal video understanding. Large Language Models (LLMs) have shown promising performance in video tasks due to their extended capabilities in comprehending visual modalities. However, the substantial redundancy in video data presents significant computational challenges for LLMs. To address this issue, we introduce a training-free method that 1) minimizes video redundancy by merging spatial-temporal tokens, and 2) leverages LLMs' reasoning capabilities to selectively prune visual features relevant to question tokens, enhancing model efficiency. We validate our method across multiple video benchmarks, which demonstrate that PruneVid can prune over 80% of tokens while maintaining competitive performance combined with different model networks. This highlights its superior effectiveness and efficiency compared to existing pruning methods. Code: https://github.com/Visual-AI/PruneVid.

Ning Ding, Yehui Tang, Haochen Qin et al.

In order to reduce the computational complexity of large language models, great efforts have been made to to improve the efficiency of transformer models such as linear attention and flash-attention. However, the model size and corresponding computational complexity are constantly scaled up in pursuit of higher performance. In this work, we present MemoryFormer, a novel transformer architecture which significantly reduces the computational complexity (FLOPs) from a new perspective. We eliminate nearly all the computations of the transformer model except for the necessary computation required by the multi-head attention operation. This is made possible by utilizing an alternative method for feature transformation to replace the linear projection of fully-connected layers. Specifically, we first construct a group of in-memory lookup tables that store a large amount of discrete vectors to replace the weight matrix used in linear projection. We then use a hash algorithm to retrieve a correlated subset of vectors dynamically based on the input embedding. The retrieved vectors combined together will form the output embedding, which provides an estimation of the result of matrix multiplication operation in a fully-connected layer. Compared to conducting matrix multiplication, retrieving data blocks from memory is a much cheaper operation which requires little computations. We train MemoryFormer from scratch and conduct extensive experiments on various benchmarks to demonstrate the effectiveness of the proposed model.

Silin Cheng, Kai Han

Vision-language models (VLMs), such as CLIP, have shown strong generalization under zero-shot settings, yet adapting them to downstream tasks with limited supervision remains a significant challenge. Existing multi-modal prompt learning methods typically rely on fixed, shared prompts and deterministic parameters, which limits their ability to capture instance-level variation or model uncertainty across diverse tasks and domains. To tackle this issue, we propose a novel Variational Multi-Modal Prompt Learning (VaMP) framework that enables sample-specific, uncertainty-aware prompt tuning in multi-modal representation learning. VaMP generates instance-conditioned prompts by sampling from a learned posterior distribution, allowing the model to personalize its behavior based on input content. To further enhance the integration of local and global semantics, we introduce a class-aware prior derived from the instance representation and class prototype. Building upon these, we formulate prompt tuning as variational inference over latent prompt representations and train the entire framework end-to-end through reparameterized sampling. Experiments on few-shot and domain generalization benchmarks show that VaMP achieves state-of-the-art performance, highlighting the benefits of modeling both uncertainty and task structure in our method. Project page: https://visual-ai.github.io/vamp

Silin Cheng, Yuanpei Liu, Kai Han

We tackle the challenging problem of Open-Set Object Detection (OSOD), which aims to detect both known and unknown objects in unlabelled images. The main difficulty arises from the absence of supervision for these unknown classes, making it challenging to distinguish them from the background. Existing OSOD detectors either fail to properly exploit or inadequately leverage the abundant unlabeled unknown objects in training data, restricting their performance. To address these limitations, we propose UADet, an Uncertainty-Aware Open-Set Object Detector that considers appearance and geometric uncertainty. By integrating these uncertainty measures, UADet effectively reduces the number of unannotated instances incorrectly utilized or omitted by previous methods. Extensive experiments on OSOD benchmarks demonstrate that UADet substantially outperforms previous state-of-the-art (SOTA) methods in detecting both known and unknown objects, achieving a 1.8x improvement in unknown recall while maintaining high performance on known classes. When extended to Open World Object Detection (OWOD), our method shows significant advantages over the current SOTA method, with average improvements of 13.8% and 6.9% in unknown recall on M-OWODB and S-OWODB benchmarks, respectively. Extensive results validate the effectiveness of our uncertainty-aware approach across different open-set scenarios.