Yunyang Xiong

Ganesh Jawahar, Haichuan Yang, Yunyang Xiong et al. · meta-ai, pku

Weight-sharing supernets are crucial for performance estimation in cutting-edge neural architecture search (NAS) frameworks. Despite their ability to generate diverse subnetworks without retraining, the quality of these subnetworks is not guaranteed due to weight sharing. In NLP tasks like machine translation and pre-trained language modeling, there is a significant performance gap between supernet and training from scratch for the same model architecture, necessitating retraining post optimal architecture identification. This study introduces a solution called mixture-of-supernets, a generalized supernet formulation leveraging mixture-of-experts (MoE) to enhance supernet model expressiveness with minimal training overhead. Unlike conventional supernets, this method employs an architecture-based routing mechanism, enabling indirect sharing of model weights among subnetworks. This customization of weights for specific architectures, learned through gradient descent, minimizes retraining time, significantly enhancing training efficiency in NLP. The proposed method attains state-of-the-art (SoTA) performance in NAS for fast machine translation models, exhibiting a superior latency-BLEU tradeoff compared to HAT, the SoTA NAS framework for machine translation. Furthermore, it excels in NAS for building memory-efficient task-agnostic BERT models, surpassing NAS-BERT and AutoDistil across various model sizes. The code can be found at: https://github.com/UBC-NLP/MoS.

Jinyoung Park, Sanghyeok Lee, Sihyeon Kim et al.

Transformers have shown superior performance on various computer vision tasks with their capabilities to capture long-range dependencies. Despite the success, it is challenging to directly apply Transformers on point clouds due to their quadratic cost in the number of points. In this paper, we present a Self-Positioning point-based Transformer (SPoTr), which is designed to capture both local and global shape contexts with reduced complexity. Specifically, this architecture consists of local self-attention and self-positioning point-based global cross-attention. The self-positioning points, adaptively located based on the input shape, consider both spatial and semantic information with disentangled attention to improve expressive power. With the self-positioning points, we propose a novel global cross-attention mechanism for point clouds, which improves the scalability of global self-attention by allowing the attention module to compute attention weights with only a small set of self-positioning points. Experiments show the effectiveness of SPoTr on three point cloud tasks such as shape classification, part segmentation, and scene segmentation. In particular, our proposed model achieves an accuracy gain of 2.6% over the previous best models on shape classification with ScanObjectNN. We also provide qualitative analyses to demonstrate the interpretability of self-positioning points. The code of SPoTr is available at https://github.com/mlvlab/SPoTr.

Jun Chen, Deyao Zhu, Xiaoqian Shen et al.

Large language models have shown their remarkable capabilities as a general interface for various language-related applications. Motivated by this, we target to build a unified interface for completing many vision-language tasks including image description, visual question answering, and visual grounding, among others. The challenge is to use a single model for performing diverse vision-language tasks effectively with simple multi-modal instructions. Towards this objective, we introduce MiniGPT-v2, a model that can be treated as a unified interface for better handling various vision-language tasks. We propose using unique identifiers for different tasks when training the model. These identifiers enable our model to better distinguish each task instruction effortlessly and also improve the model learning efficiency for each task. After the three-stage training, the experimental results show that MiniGPT-v2 achieves strong performance on many visual question-answering and visual grounding benchmarks compared to other vision-language generalist models. Our model and codes are available at https://minigpt-v2.github.io/

Zhipeng Cai, Zhuang Liu, Yunyang Xiong et al.

Vision Language Models (VLMs) enable a unified model to solve various vision tasks through prompting. They have shown promising performance in semantic understanding. However, 3D understanding still largely relies on expert vision models with complex task-specific designs. The key argument this work wants to make is that VLMs are native 3D learners. Our in-depth large scale study shows that 1) focal length unification, 2) text-based pixel reference and 3) data mixture and scaling, are all you need for effective 3D learning. Model architecture changes, large models, heavy data augmentations, and complex losses including the regression formulation, many of which form the foundation of expert vision models, are actually not necessary conditions. As a result, we propose VLM3, a scalable method with the simplest design that enables standard VLMs to master diverse 3D tasks. VLM3 not only advances the VLM depth estimation accuracy by a large margin (0.84 -> 0.9), but also enables diverse 3D tasks such as pixel correspondence, camera pose estimation and object-level 3D understanding, matching expert vision model accuracy while maintaining standard architectures and text-based training. We believe VLM3 opens up a new paradigm for simple and scalable 3D learning.

Lemeng Wu, Dilin Wang, Meng Li et al. · pku

Fusing 3D LiDAR features with 2D camera features is a promising technique for enhancing the accuracy of 3D detection, thanks to their complementary physical properties. While most of the existing methods focus on directly fusing camera features with raw LiDAR point clouds or shallow-level 3D features, it is observed that directly combining 2D and 3D features in deeper layers actually leads to a decrease in accuracy due to feature misalignment. The misalignment, which stems from the aggregation of features learned from large receptive fields, becomes increasingly more severe as we delve into deeper layers. In this paper, we propose PathFusion as a solution to enable the alignment of semantically coherent LiDAR-camera deep feature fusion. PathFusion introduces a path consistency loss at multiple stages within the network, encouraging the 2D backbone and its fusion path to transform 2D features in a way that aligns semantically with the transformation of the 3D backbone. This ensures semantic consistency between 2D and 3D features, even in deeper layers, and amplifies the usage of the network's learning capacity. We apply PathFusion to improve a prior-art fusion baseline, Focals Conv, and observe an improvement of over 1.6% in mAP on the nuScenes test split consistently with and without testing-time data augmentations, and moreover, PathFusion also improves KITTI $\text{AP}_{\text{3D}}$ (R11) by about 0.6% on the moderate level.

Haoran You, Yunyang Xiong, Xiaoliang Dai et al.

Vision Transformers (ViTs) have shown impressive performance but still require a high computation cost as compared to convolutional neural networks (CNNs), one reason is that ViTs' attention measures global similarities and thus has a quadratic complexity with the number of input tokens. Existing efficient ViTs adopt local attention (e.g., Swin) or linear attention (e.g., Performer), which sacrifice ViTs' capabilities of capturing either global or local context. In this work, we ask an important research question: Can ViTs learn both global and local context while being more efficient during inference? To this end, we propose a framework called Castling-ViT, which trains ViTs using both linear-angular attention and masked softmax-based quadratic attention, but then switches to having only linear angular attention during ViT inference. Our Castling-ViT leverages angular kernels to measure the similarities between queries and keys via spectral angles. And we further simplify it with two techniques: (1) a novel linear-angular attention mechanism: we decompose the angular kernels into linear terms and high-order residuals, and only keep the linear terms; and (2) we adopt two parameterized modules to approximate high-order residuals: a depthwise convolution and an auxiliary masked softmax attention to help learn both global and local information, where the masks for softmax attention are regularized to gradually become zeros and thus incur no overhead during ViT inference. Extensive experiments and ablation studies on three tasks consistently validate the effectiveness of the proposed Castling-ViT, e.g., achieving up to a 1.8% higher accuracy or 40% MACs reduction on ImageNet classification and 1.2 higher mAP on COCO detection under comparable FLOPs, as compared to ViTs with vanilla softmax-based attentions.

Lemeng Wu, Dilin Wang, Chengyue Gong et al.

Diffusion models have emerged as a powerful tool for point cloud generation. A key component that drives the impressive performance for generating high-quality samples from noise is iteratively denoise for thousands of steps. While beneficial, the complexity of learning steps has limited its applications to many 3D real-world. To address this limitation, we propose Point Straight Flow (PSF), a model that exhibits impressive performance using one step. Our idea is based on the reformulation of the standard diffusion model, which optimizes the curvy learning trajectory into a straight path. Further, we develop a distillation strategy to shorten the straight path into one step without a performance loss, enabling applications to 3D real-world with latency constraints. We perform evaluations on multiple 3D tasks and find that our PSF performs comparably to the standard diffusion model, outperforming other efficient 3D point cloud generation methods. On real-world applications such as point cloud completion and training-free text-guided generation in a low-latency setup, PSF performs favorably.

Hongwei Lin, Yunyang Xiong, Xiao Wang et al.

In this paper, we present the isogeometric least-squares collocation (IGA-L) method, which determines the numerical solution by making the approximate differential operator fit the real differential operator in a least-squares sense. The number of collocation points employed in IGA-L can be larger than that of the unknowns. Theoretical analysis and numerical examples presented in this paper show the superiority of IGA-L over state-of-the-art collocation methods. First, a small increase in the number of collocation points in IGA-L leads to a large improvement in the accuracy of its numerical solution. Second, IGA-L method is more flexible and more stable, because the number of collocation points in IGA-L is variable. Third, IGA-L is convergent in some cases of singular parameterization. Moreover, the consistency and convergence analysis are also developed in this paper.

Sanghyeok Lee, Minkyu Jeon, Injae Kim et al.

Data augmentation is key to improving the generalization ability of deep learning models. Mixup is a simple and widely-used data augmentation technique that has proven effective in alleviating the problems of overfitting and data scarcity. Also, recent studies of saliency-aware Mixup in the image domain show that preserving discriminative parts is beneficial to improving the generalization performance. However, these Mixup-based data augmentations are underexplored in 3D vision, especially in point clouds. In this paper, we propose SageMix, a saliency-guided Mixup for point clouds to preserve salient local structures. Specifically, we extract salient regions from two point clouds and smoothly combine them into one continuous shape. With a simple sequential sampling by re-weighted saliency scores, SageMix preserves the local structure of salient regions. Extensive experiments demonstrate that the proposed method consistently outperforms existing Mixup methods in various benchmark point cloud datasets. With PointNet++, our method achieves an accuracy gain of 2.6% and 4.0% over standard training in 3D Warehouse dataset (MN40) and ScanObjectNN, respectively. In addition to generalization performance, SageMix improves robustness and uncertainty calibration. Moreover, when adopting our method to various tasks including part segmentation and standard 2D image classification, our method achieves competitive performance.

Chenchen Zhu, Saksham Suri, Cijo Jose et al.

Running AI models on smart edge devices can unlock versatile user experiences, but presents challenges due to limited compute and the need to handle multiple tasks simultaneously. This requires a vision encoder with small size but powerful and versatile representations. We present our method, Efficient Universal Perception Encoder (EUPE), which offers both inference efficiency and universally good representations for diverse downstream tasks. We achieve this by distilling from multiple domain-expert foundation vision encoders. Unlike previous agglomerative methods that directly scale down from multiple teachers to an efficient encoder, we demonstrate the importance of first scaling up to a large proxy teacher and then scaling down from this single teacher. Experiments show that EUPE achieves on-par or better performance than individual domain experts of the same size on diverse task domains and also outperforms previous agglomerative encoders. We will release the full family of EUPE models and the code to foster future research.

Mingchen Zhuge, Changsheng Zhao, Haozhe Liu et al.

We propose a new frontier: Neural Computers (NCs) -- an emerging machine form that unifies computation, memory, and I/O in a learned runtime state. Unlike conventional computers, which execute explicit programs, agents, which act over external execution environments, and world models, which learn environment dynamics, NCs aim to make the model itself the running computer. Our long-term goal is the Completely Neural Computer (CNC): the mature, general-purpose realization of this emerging machine form, with stable execution, explicit reprogramming, and durable capability reuse. As an initial step, we study whether early NC primitives can be learned solely from collected I/O traces, without instrumented program state. Concretely, we instantiate NCs as video models that roll out screen frames from instructions, pixels, and user actions (when available) in CLI and GUI settings. These implementations show that learned runtimes can acquire early interface primitives, especially I/O alignment and short-horizon control, while routine reuse, controlled updates, and symbolic stability remain open. We outline a roadmap toward CNCs around these challenges. If overcome, CNCs could establish a new computing paradigm beyond today's agents, world models, and conventional computers.

Shuming Liu, Mingchen Zhuge, Changsheng Zhao et al.

Chain-of-thought (CoT) reasoning has emerged as a powerful tool for multimodal large language models on video understanding tasks. However, its necessity and advantages over direct answering remain underexplored. In this paper, we first demonstrate that for RL-trained video models, direct answering often matches or even surpasses CoT performance, despite CoT producing step-by-step analyses at a higher computational cost. Motivated by this, we propose VideoAuto-R1, a video understanding framework that adopts a reason-when-necessary strategy. During training, our approach follows a Thinking Once, Answering Twice paradigm: the model first generates an initial answer, then performs reasoning, and finally outputs a reviewed answer. Both answers are supervised via verifiable rewards. During inference, the model uses the confidence score of the initial answer to determine whether to proceed with reasoning. Across video QA and grounding benchmarks, VideoAuto-R1 achieves state-of-the-art accuracy with significantly improved efficiency, reducing the average response length by ~3.3x, e.g., from 149 to just 44 tokens. Moreover, we observe a low rate of thinking-mode activation on perception-oriented tasks, but a higher rate on reasoning-intensive tasks. This suggests that explicit language-based reasoning is generally beneficial but not always necessary.

Ashish Seth, Xinhao Mei, Changsheng Zhao et al.

Understanding egocentric videos plays a vital role for embodied intelligence. Recent multi-modal large language models (MLLMs) can accept both visual and audio inputs. However, due to the challenge of obtaining text labels with coherent joint-modality information, whether MLLMs can jointly understand both modalities in egocentric videos remains under-explored. To address this problem, we introduce EgoAVU, a scalable data engine to automatically generate egocentric audio-visual narrations, questions, and answers. EgoAVU enriches human narrations with multimodal context and generates audio-visual narrations through cross-modal correlation modeling. Token-based video filtering and modular, graph-based curation ensure both data diversity and quality. Leveraging EgoAVU, we construct EgoAVU-Instruct, a large-scale training dataset of 3M samples, and EgoAVU-Bench, a manually verified evaluation split covering diverse tasks. EgoAVU-Bench clearly reveals the limitations of existing MLLMs: they bias heavily toward visual signals, often neglecting audio cues or failing to correspond audio with the visual source. Finetuning MLLMs on EgoAVU-Instruct effectively addresses this issue, enabling up to 113% performance improvement on EgoAVU-Bench. Such benefits also transfer to other benchmarks such as EgoTempo and EgoIllusion, achieving up to 28% relative performance gain. Code will be released to the community.

Zechun Liu, Changsheng Zhao, Forrest Iandola et al.

This paper addresses the growing need for efficient large language models (LLMs) on mobile devices, driven by increasing cloud costs and latency concerns. We focus on designing top-quality LLMs with fewer than a billion parameters, a practical choice for mobile deployment. Contrary to prevailing belief emphasizing the pivotal role of data and parameter quantity in determining model quality, our investigation underscores the significance of model architecture for sub-billion scale LLMs. Leveraging deep and thin architectures, coupled with embedding sharing and grouped-query attention mechanisms, we establish a strong baseline network denoted as MobileLLM, which attains a remarkable 2.7%/4.3% accuracy boost over preceding 125M/350M state-of-the-art models. Additionally, we propose an immediate block-wise weight-sharing approach with no increase in model size and only marginal latency overhead. The resultant models, denoted as MobileLLM-LS, demonstrate a further accuracy enhancement of 0.7%/0.8% than MobileLLM 125M/350M. Moreover, MobileLLM model family shows significant improvements compared to previous sub-billion models on chat benchmarks, and demonstrates close correctness to LLaMA-v2 7B in API calling tasks, highlighting the capability of small models for common on-device use cases.

Xiaoqian Shen, Yunyang Xiong, Changsheng Zhao et al.

Multimodal Large Language Models (MLLMs) have shown promising progress in understanding and analyzing video content. However, processing long videos remains a significant challenge constrained by LLM's context size. To address this limitation, we propose LongVU, a spatiotemporal adaptive compression mechanism thats reduces the number of video tokens while preserving visual details of long videos. Our idea is based on leveraging cross-modal query and inter-frame dependencies to adaptively reduce temporal and spatial redundancy in videos. Specifically, we leverage DINOv2 features to remove redundant frames that exhibit high similarity. Then we utilize text-guided cross-modal query for selective frame feature reduction. Further, we perform spatial token reduction across frames based on their temporal dependencies. Our adaptive compression strategy effectively processes a large number of frames with little visual information loss within given context length. Our LongVU consistently surpass existing methods across a variety of video understanding benchmarks, especially on hour-long video understanding tasks such as VideoMME and MLVU. Given a light-weight LLM, our LongVU also scales effectively into a smaller size with state-of-the-art video understanding performance.

Shengbang Tong, David Fan, Jiachen Zhu et al.

In this work, we propose Visual-Predictive Instruction Tuning (VPiT) - a simple and effective extension to visual instruction tuning that enables a pretrained LLM to quickly morph into an unified autoregressive model capable of generating both text and visual tokens. VPiT teaches an LLM to predict discrete text tokens and continuous visual tokens from any input sequence of image and text data curated in an instruction-following format. Our empirical investigation reveals several intriguing properties of VPiT: (1) visual generation ability emerges as a natural byproduct of improved visual understanding, and can be unlocked efficiently with a small amount of generation data; (2) while we find understanding and generation to be mutually beneficial, understanding data contributes to both capabilities more effectively than generation data. Building upon these findings, we train our MetaMorph model and achieve competitive performance on both visual understanding and generation. In visual generation, MetaMorph can leverage the world knowledge and reasoning abilities gained from LLM pretraining, and overcome common failure modes exhibited by other generation models. Our results suggest that LLMs may have strong "prior" vision capabilities that can be efficiently adapted to both visual understanding and generation with a relatively simple instruction tuning process.

Zhanpeng Zeng, Yunyang Xiong, Sathya N. Ravi et al.

Transformer-based models are widely used in natural language processing (NLP). Central to the transformer model is the self-attention mechanism, which captures the interactions of token pairs in the input sequences and depends quadratically on the sequence length. Training such models on longer sequences is expensive. In this paper, we show that a Bernoulli sampling attention mechanism based on Locality Sensitive Hashing (LSH), decreases the quadratic complexity of such models to linear. We bypass the quadratic cost by considering self-attention as a sum of individual tokens associated with Bernoulli random variables that can, in principle, be sampled at once by a single hash (although in practice, this number may be a small constant). This leads to an efficient sampling scheme to estimate self-attention which relies on specific modifications of LSH (to enable deployment on GPU architectures). We evaluate our algorithm on the GLUE benchmark with standard 512 sequence length where we see favorable performance relative to a standard pretrained Transformer. On the Long Range Arena (LRA) benchmark, for evaluating performance on long sequences, our method achieves results consistent with softmax self-attention but with sizable speed-ups and memory savings and often outperforms other efficient self-attention methods. Our code is available at https://github.com/mlpen/YOSO

Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty et al.

Transformers have emerged as a powerful tool for a broad range of natural language processing tasks. A key component that drives the impressive performance of Transformers is the self-attention mechanism that encodes the influence or dependence of other tokens on each specific token. While beneficial, the quadratic complexity of self-attention on the input sequence length has limited its application to longer sequences -- a topic being actively studied in the community. To address this limitation, we propose Nyströmformer -- a model that exhibits favorable scalability as a function of sequence length. Our idea is based on adapting the Nyström method to approximate standard self-attention with $O(n)$ complexity. The scalability of Nyströmformer enables application to longer sequences with thousands of tokens. We perform evaluations on multiple downstream tasks on the GLUE benchmark and IMDB reviews with standard sequence length, and find that our Nyströmformer performs comparably, or in a few cases, even slightly better, than standard self-attention. On longer sequence tasks in the Long Range Arena (LRA) benchmark, Nyströmformer performs favorably relative to other efficient self-attention methods. Our code is available at https://github.com/mlpen/Nystromformer.

Yunyang Xiong, Hanxiao Liu, Suyog Gupta et al.

Inverted bottleneck layers, which are built upon depthwise convolutions, have been the predominant building blocks in state-of-the-art object detection models on mobile devices. In this work, we investigate the optimality of this design pattern over a broad range of mobile accelerators by revisiting the usefulness of regular convolutions. We discover that regular convolutions are a potent component to boost the latency-accuracy trade-off for object detection on accelerators, provided that they are placed strategically in the network via neural architecture search. By incorporating regular convolutions in the search space and directly optimizing the network architectures for object detection, we obtain a family of object detection models, MobileDets, that achieve state-of-the-art results across mobile accelerators. On the COCO object detection task, MobileDets outperform MobileNetV3+SSDLite by 1.7 mAP at comparable mobile CPU inference latencies. MobileDets also outperform MobileNetV2+SSDLite by 1.9 mAP on mobile CPUs, 3.7 mAP on Google EdgeTPU, 3.4 mAP on Qualcomm Hexagon DSP and 2.7 mAP on Nvidia Jetson GPU without increasing latency. Moreover, MobileDets are comparable with the state-of-the-art MnasFPN on mobile CPUs even without using the feature pyramid, and achieve better mAP scores on both EdgeTPUs and DSPs with up to 2x speedup. Code and models are available in the TensorFlow Object Detection API: https://github.com/tensorflow/models/tree/master/research/object_detection.

Mingchen Zhuge, Changsheng Zhao, Dylan Ashley et al.

Contemporary evaluation techniques are inadequate for agentic systems. These approaches either focus exclusively on final outcomes -- ignoring the step-by-step nature of agentic systems, or require excessive manual labour. To address this, we introduce the Agent-as-a-Judge framework, wherein agentic systems are used to evaluate agentic systems. This is an organic extension of the LLM-as-a-Judge framework, incorporating agentic features that enable intermediate feedback for the entire task-solving process. We apply the Agent-as-a-Judge to the task of code generation. To overcome issues with existing benchmarks and provide a proof-of-concept testbed for Agent-as-a-Judge, we present DevAI, a new benchmark of 55 realistic automated AI development tasks. It includes rich manual annotations, like a total of 365 hierarchical user requirements. We benchmark three of the popular agentic systems using Agent-as-a-Judge and find it dramatically outperforms LLM-as-a-Judge and is as reliable as our human evaluation baseline. Altogether, we believe that Agent-as-a-Judge marks a concrete step forward for modern agentic systems -- by providing rich and reliable reward signals necessary for dynamic and scalable self-improvement.

Zechun Liu, Changsheng Zhao, Hanxian Huang et al.

The optimal bit-width for achieving the best trade-off between quantized model size and accuracy has been a subject of ongoing debate. While some advocate for 4-bit quantization, others propose that 1.58-bit offers superior results. However, the lack of a cohesive framework for different bits has left such conclusions relatively tenuous. We present ParetoQ, the first unified framework that facilitates rigorous comparisons across 1-bit, 1.58-bit, 2-bit, 3-bit, and 4-bit quantization settings. Our findings reveal a notable learning transition between 2 and 3 bits: For 3-bits and above, the fine-tuned models stay close to their original pre-trained distributions, whereas for learning 2-bit networks or below, the representations change drastically. By optimizing training schemes and refining quantization functions, ParetoQ surpasses all previous methods tailored to specific bit widths. Remarkably, our ParetoQ ternary 600M-parameter model even outperforms the previous SoTA ternary 3B-parameter model in accuracy, using only one-fifth of the parameters. Extensive experimentation shows that ternary, 2-bit, and 3-bit quantization maintains comparable performance in the size-accuracy trade-off and generally exceeds 4-bit and binary quantization. Considering hardware constraints, 2-bit quantization offers promising potential for memory reduction and speedup.

Ashish Seth, Xinhao Mei, Changsheng Zhao et al.

Egocentric videos provide a distinctive setting in which sound serves as crucial cues to understand user activities and surroundings, particularly when visual information is unstable or occluded due to continuous camera movement. State-of-the-art large audio-visual language models (AV-LLMs) can generate multimodal descriptions. However, we show in this work that they are prone to audio hallucinations, often inferring sounds from visual cues that are visible but not heard. We present a systematic and automatic evaluation framework for analyzing audio hallucinations in egocentric video through a targeted question-answering (Q/A) protocol. We curate a dataset of 300 egocentric videos and design 1,000 sound-focused questions to probe model outputs. To characterize hallucinations, we propose a grounded taxonomy that distinguishes between foreground action sounds from the user activities and background ambient sounds. Our evaluation shows that advanced AV-LLMs, such as Qwen2.5 Omni, exhibit high hallucination rates, achieving only 27.3% and 39.5% accuracy on Q/As related to foreground and background sounds, respectively. With this work, we highlight the need to measure the reliability of multimodal responses, emphasizing that robust evaluation of hallucinations is essential to develop reliable AV-LLMs.

Yunyang Xiong, Chong Zhou, Xiaoyu Xiang et al.

Segment Anything Model 2 (SAM 2) has emerged as a powerful tool for video object segmentation and tracking anything. Key components of SAM 2 that drive the impressive video object segmentation performance include a large multistage image encoder for frame feature extraction and a memory mechanism that stores memory contexts from past frames to help current frame segmentation. The high computation complexity of multistage image encoder and memory module has limited its applications in real-world tasks, e.g., video object segmentation on mobile devices. To address this limitation, we propose EfficientTAMs, lightweight track anything models that produce high-quality results with low latency and model size. Our idea is based on revisiting the plain, nonhierarchical Vision Transformer (ViT) as an image encoder for video object segmentation, and introducing an efficient memory module, which reduces the complexity for both frame feature extraction and memory computation for current frame segmentation. We take vanilla lightweight ViTs and efficient memory module to build EfficientTAMs, and train the models on SA-1B and SA-V datasets for video object segmentation and track anything tasks. We evaluate on multiple video segmentation benchmarks including semi-supervised VOS and promptable video segmentation, and find that our proposed EfficientTAM with vanilla ViT perform comparably to SAM 2 model (HieraB+SAM 2) with ~2x speedup on A100 and ~2.4x parameter reduction. On segment anything image tasks, our EfficientTAMs also perform favorably over original SAM with ~20x speedup on A100 and ~20x parameter reduction. On mobile devices such as iPhone 15 Pro Max, our EfficientTAMs can run at ~10 FPS for performing video object segmentation with reasonable quality, highlighting the capability of small models for on-device video object segmentation applications.

Chong Zhou, Chenchen Zhu, Yunyang Xiong et al.

On top of Segment Anything Model (SAM), SAM 2 further extends its capability from image to video inputs through a memory bank mechanism and obtains a remarkable performance compared with previous methods, making it a foundation model for video segmentation task. In this paper, we aim at making SAM 2 much more efficient so that it even runs on mobile devices while maintaining a comparable performance. Despite several works optimizing SAM for better efficiency, we find they are not sufficient for SAM 2 because they all focus on compressing the image encoder, while our benchmark shows that the newly introduced memory attention blocks are also the latency bottleneck. Given this observation, we propose EdgeTAM, which leverages a novel 2D Spatial Perceiver to reduce the computational cost. In particular, the proposed 2D Spatial Perceiver encodes the densely stored frame-level memories with a lightweight Transformer that contains a fixed set of learnable queries. Given that video segmentation is a dense prediction task, we find preserving the spatial structure of the memories is essential so that the queries are split into global-level and patch-level groups. We also propose a distillation pipeline that further improves the performance without inference overhead. As a result, EdgeTAM achieves 87.7, 70.0, 72.3, and 71.7 J&F on DAVIS 2017, MOSE, SA-V val, and SA-V test, while running at 16 FPS on iPhone 15 Pro Max.

Balakrishnan Varadarajan, Bilge Soran, Forrest Iandola et al.

The Segment Anything Model (SAM) has been a cornerstone in the field of interactive segmentation, propelling significant progress in generative AI, computational photography, and medical imaging. Despite its ability to process arbitrary user input and generate corresponding segmentation masks, SAM's 600 million parameter architecture, based on ViT-H, is not compatible with current mobile hardware due to its high computational demands and large model size. Our research aims to adapt SAM for use in mobile photography applications. To this end, we have developed a fully convolutional SqueezeSAM model architecture, which is 62.5 times faster and 31.6 times smaller than the original SAM, making it a viable solution for mobile applications. Furthermore, our tiny model achieves an mIOU within 1% of the original VIT-H architecture. Automated segmentation holds significant value in the creation flow for photography applications, as evidenced by its adoption by leading industry players like apple and capcut. To facilitate this automation, we employ salient object detection and simulate potential user clicks for foreground object selection, generating an initial segmentation mask that users can subsequently edit interactively. A common user expectation is that a click on a specific part of an object will result in the segmentation of the entire object. For example, a click on a person's t-shirt in a photo should ideally segment the entire person, not just the t-shirt. However, SAM typically only segments the clicked area. We address this limitation through a novel data augmentation scheme. Consequently, if a user clicks on a person holding a basketball, both the person and the basketball are segmented together, aligning with user expectations and enhancing the overall user experience.

Junjie Fei, Jun Chen, Zechun Liu et al.

Adapting Multimodal Large Language Models (MLLMs) for hour-long videos is bottlenecked by context limits. Dense visual streams saturate token budgets and exacerbate the lost-in-the-middle phenomenon. Existing heuristics, like sparse sampling or uniform pooling, blindly sacrifice fidelity by discarding decisive moments and wasting bandwidth on irrelevant backgrounds. We propose Tempo, an efficient query-aware framework compressing long videos for downstream understanding. Tempo leverages a Small Vision-Language Model (SVLM) as a local temporal compressor, casting token reduction as an early cross-modal distillation process to generate compact, intent-aligned representations in a single forward pass. To enforce strict budgets without breaking causality, we introduce Adaptive Token Allocation (ATA). Exploiting the SVLM's zero-shot relevance prior and semantic front-loading, ATA acts as a training-free $O(1)$ dynamic router. It allocates dense bandwidth to query-critical segments while compressing redundancies into minimal temporal anchors to maintain the global storyline. Extensive experiments show our 6B architecture achieves state-of-the-art performance with aggressive dynamic compression (0.5-16 tokens/frame). On the extreme-long LVBench (4101s), Tempo scores 52.3 under a strict 8K visual budget, outperforming GPT-4o and Gemini 1.5 Pro. Scaling to 2048 frames reaches 53.7. Crucially, Tempo compresses hour-long videos substantially below theoretical limits, proving true long-form video understanding relies on intent-driven efficiency rather than greedily padded context windows.

Yunyang Xiong, Ronak Mehta, Vikas Singh

The design of neural network architectures is frequently either based on human expertise using trial/error and empirical feedback or tackled via large scale reinforcement learning strategies performed over distinct discrete architecture choices. In the latter case, the optimization is often non-differentiable and also not very amenable to derivative-free optimization methods. Most methods in use today require sizable computational resources. And if we want networks that additionally satisfy resource constraints, the above challenges are exacerbated because the search must now balance accuracy with certain budget constraints on resources. We formulate this problem as the optimization of a set function -- we find that the empirical behavior of this set function often (but not always) satisfies marginal gain and monotonicity principles -- properties central to the idea of submodularity. Based on this observation, we adapt algorithms within discrete optimization to obtain heuristic schemes for neural network architecture search, where we have resource constraints on the architecture. This simple scheme when applied on CIFAR-100 and ImageNet, identifies resource-constrained architectures with quantifiably better performance than current state-of-the-art models designed for mobile devices. Specifically, we find high-performing architectures with fewer parameters and computations by a search method that is much faster.

Yunyang Xiong, Hyunwoo J. Kim, Varsha Hedau

Deep convolutional neural networks have achieved remarkable success in computer vision. However, deep neural networks require large computing resources to achieve high performance. Although depthwise separable convolution can be an efficient module to approximate a standard convolution, it often leads to reduced representational power of networks. In this paper, under budget constraints such as computational cost (MAdds) and the parameter count, we propose a novel basic architectural block, ANTBlock. It boosts the representational power by modeling, in a high dimensional space, interdependency of channels between a depthwise convolution layer and a projection layer in the ANTBlocks. Our experiments show that ANTNet built by a sequence of ANTBlocks, consistently outperforms state-of-the-art low-cost mobile convolutional neural networks across multiple datasets. On CIFAR100, our model achieves 75.7% top-1 accuracy, which is 1.5% higher than MobileNetV2 with 8.3% fewer parameters and 19.6% less computational cost. On ImageNet, our model achieves 72.8% top-1 accuracy, which is 0.8% improvement, with 157.7ms (20% faster) on iPhone 5s over MobileNetV2.

Hao Henry Zhou, Yunyang Xiong, Vikas Singh

We provide simple schemes to build Bayesian Neural Networks (BNNs), block by block, inspired by a recent idea of computation skeletons. We show how by adjusting the types of blocks that are used within the computation skeleton, we can identify interesting relationships with Deep Gaussian Processes (DGPs), deep kernel learning (DKL), random features type approximation and other topics. We give strategies to approximate the posterior via doubly stochastic variational inference for such models which yield uncertainty estimates. We give a detailed theoretical analysis and point out extensions that may be of independent interest. As a special case, we instantiate our procedure to define a Bayesian {\em additive} Neural network -- a promising strategy to identify statistical interactions and has direct benefits for obtaining interpretable models.

Hongwei Lin, Yunyang Xiong, Qianqian Hu

Although the isogeometric collocation (IGA-C) method has been successfully utilized in practical applications due to its simplicity and efficiency, only a little theoretical results have been established on the numerical analysis of the IGA-C method. In this paper, we deduce the convergence rate of the consistency of the IGA-C method. Moreover, based on the formula of the convergence rate, the necessary and sufficient condition for the consistency of the IGA-C method is developed. These results advance the numerical analysis of the IGA-C method.